Fuel Processing Technology 91 (2010) 68–79

Contents lists available at ScienceDirect

Fuel Processing Technology

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / f u p r o c

Computational modelling of the impact of particle size to the heat transfer coefficient
between biomass particles and a fluidised bed
K. Papadikis a, S. Gu b,⁎, A.V. Bridgwater a
a
School of Engineering and Applied Science, Aston University, Aston Triangle, Birmingham B4 7ET, United Kingdom
b
School of Engineering Sciences, University of Southampton, Highfield, Southampton, SO17 1BJ, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: The fluid–particle interaction and the impact of different heat transfer conditions on pyrolysis of biomass
Received 12 June 2009 inside a 150 g/h fluidised bed reactor are modelled. Two different size biomass particles (350 μm and 550 μm
Received in revised form 24 July 2009 in diameter) are injected into the fluidised bed. The different biomass particle sizes result in different heat
Accepted 11 August 2009
transfer conditions. This is due to the fact that the 350 μm diameter particle is smaller than the sand particles
of the reactor (440 μm), while the 550 μm one is larger. The bed-to-particle heat transfer for both cases is
Keywords:
CFD
calculated according to the literature. Conductive heat transfer is assumed for the larger biomass particle
Fluidised bed (550 μm) inside the bed, while biomass–sand contacts for the smaller biomass particle (350 μm) were
Fast pyrolysis considered unimportant. The Eulerian approach is used to model the bubbling behaviour of the sand, which
Heat transfer is treated as a continuum. Biomass reaction kinetics is modelled according to the literature using a two-stage,
Multiphase flow semi-global model which takes into account secondary reactions. The particle motion inside the reactor is
computed using drag laws, dependent on the local volume fraction of each phase. FLUENT 6.2 has been used
as the modelling framework of the simulations with the whole pyrolysis model incorporated in the form of
User Defined Function (UDF).
© 2009 Elsevier B.V. All rights reserved.

1. Introduction 2. Background

Extensive research has been conducted recently in the renewable
energy sources. The understanding of the major factors that affect
biomass fast pyrolysis is of great importance. The mechanism of heat
transfer greatly affects fast pyrolysis in various ways. Rapid heating of
biomass and low residence time is necessary for high vapour yields
and low char and gas formation. Studies have shown that the type of
heat transfer mechanism is greatly affected by the particle size
injected on the fluidised bed and modified Nusselt numbers were
derived to account for this fact.
In the present study, the heat transfer from the bubbling bed to the
different size biomass particles is modelled according to the findings
of Collier et al. [1] for the 350 μm diameter particle and the
penetration theory of Mickley and Fairbanks [2] for the 550 μm
diameter particle, assuming that conductive heat transfer is dominant
in the near-particle region [3]. This is done by extending the already
developed CFD model by the authors [4–6] for the simulation of
biomass fast pyrolysis in fluidised beds. The kinetics mechanism used
to model biomass pyrolysis is a two-stage semi-global mechanism
with kinetic constants by Chan et al. [7], Liden et al. [8] and Di Blasi [9].
⁎ Corresponding author. Tel.: +44 23 8059 8520; fax: +44 23 8059 3230.
E-mail address: s.gu@soton.ac.uk (S. Gu).
Pyrolysis is thermal degradation of organic materials in the
absence of oxygen. The primary pyrolysis products of biomass
particles are usually referred to as condensable (tars) and noncon-
densable volatiles and char. The condensable volatiles (tars) are also
often classified as liquids, and the noncondensable volatiles as gases
mainly CO, CO2, H2 and C1–C2 hydrocarbons [10,11]. The liquids are
frequently subdivided into water and organics. Secondary reactions of
the tars produced result in the formation of secondary degradation
products, and generally reduced organics yields.
Several models have been developed and applied to biomass
pyrolysis [9–22]. Most of the models apply numerical analysis to
determine heat, mass and momentum transport effects in a single
biomass particle varying the pyrolysis conditions. Usually, the
influence of parameters such as, size, shape, moisture, reaction
mechanisms, heat transfer rates and particle shrinkage is the main
objective of study. The numerical investigations of Di Blasi [9–14] are
typical examples of single particle models, that study the heat, mass
and momentum transport through biomass particles by varying the
pyrolysis conditions, feedstock properties (implementing different
chemical kinetics mechanisms), particle sizes, effect of shrinkage etc.
The models provide very useful information about different condi-
tions of pyrolysis and product yields and can be an excellent guide for
every researcher on the field. Saastamoinen [15], and Saastamoinen
and Richard [16] studied the effect of drying to devolatilisation stating

0378-3820/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.fuproc.2009.08.016
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79 69

that the surface temperature of the particles after drying can exceed been performed by Collier et al. [1] and a Nusselt number based on the
those of the initiation of devolatilisation, meaning that drying and thermal conductivity of the gas was derived for (U ≤ Umf), while for
pyrolysis may overlap. The models of Babu and Chaurasia [17–22] (U > Umf) the Nusselt number appeared to have a constant value.
treat the pyrolysis process in a similar way as the one of Di Blasi, and Numerical investigations have also been performed using the Eulerian
focus more on the heat transfer effects and estimation of optimum approach [40,41] and the results showed a strong coupling between
parameters for biomass pyrolysis. They also study the effect of the the local solid volume distribution and the heat transfer coefficients.
enthalpy of reaction on product yields and vary the shrinkage param- They also showed a good agreement with the penetration theory [2,3]
eters of biomass close to the complete degradation of the particles. as well as the influence of the rising bubbles.
They found that their approach gave better validation with experi- The present study focuses on the impact of the biomass particle
mental data compared to those of Di Blasi. size on fast pyrolysis, due to the different size of particles and different
Sub-models for CFD applications have also been developed [23,24] heating conditions. The scope of the study is to further extend the
to make simulations of biomass pyrolysis possible and less compu- model developed by Papadikis et al. [4–6] and investigate the factors
tationally intensive. Saastamoinen [24] has developed a mathematical that will be affected regarding the heat and momentum transport
sub-model for the devolatilisation of large particles for CFD applica- from the bubbling bed to the biomass particles as well as the product
tions. The model is based on solving two differential equations for the yields.
locations of two temperature isotherms, which are the drying and the
devolatilisation isotherm. The model is also able to predict effects of 3. Model description
particle size, shape, moisture, density and reactivity. Rostami et al.
[23] have developed a CFD sub-model for biomass pyrolysis based on The 150 g/h fast pyrolysis lab scale reactor of Aston University is
the Distributed Activation Energy Model (DAEM). In their work, the illustrated in Fig. 1. Nitrogen flows through a porous plate at the
integrals that are present in DAEM and express the complex reactions bottom of the reactor at a velocity of U0 = 0.3 m/s. The superficial
of biomass pyrolysis and the evolution of different volatile species, velocity is approximately 4 times greater than the minimum fluidising
have been mathematically expressed in closed forms so that DAEM velocity Umf of the reactor, which is typically around 0.08 m/s using a
can be incorporated more efficiently in a CFD code. Their predictions sand bed with average particle diameter of 440 μm Geldart B Group
were generally in good agreement with the experimental data and the [42].
accuracy can be improved by slight changes in the kinetic parameters. The two biomass particles are injected at the centre of the sand
Fluidised beds are the most widely used type of reactor for fast bed, i.e. the middle of the sand bed height and the centre of its
pyrolysis, as they offer a number of advantages, such as high heat diameter, which has been previously fluidised for 1 s. The reactor is
transfer rates and good temperature control. The hydrodynamics of heated by a furnace to the required temperature for pyrolysis of
fluidised beds have been widely investigated, both experimentally 500 °C. The wall thermal boundary condition refers to the tempera-
and numerically, to allow validation of the model results. To date most ture of 773 K that the reactor has reached due to furnace heating. The
of the computational research interest has been focused on the simu- wall temperature is assumed to remain constant during the simu-
lation of the fluidised bed hydrodynamics, using either the Eulerian lation. The particles are injected at the same time, one next to each
(continuum) [25] or the Lagrangian (discrete element) model [26,27], other with no initial velocity and affected only by gravity, in order to
and models like the one developed by Bokkers et al. [28] which is achieve same initial conditions as much as possible. Momentum is
based on the modelling of the larger bubbles as discrete elements that transferred from the bubbling bed to the biomass particle as well as
are tracked individually during their rise through the emulsion phase, from the formed bubbles inside the bed. According to Bridgwater [43],
which is considered as a continuum. Due to the significant increase in the most appropriate biomass particles sizes for liquid fuel production
computing power of recent years, these models have now made
computational modelling of multiphase granular flows possible,
though it is still very challenging, particularly so for industrial scale
reactor units.
Compared to two-phase flows that have been extensively studied
[29–32], the information for bubble three-phase flows is relatively
limited [32,33]. The calculation of drag forces on particles that are part
of a solid/liquid/gas mixture is a more complicated case and certain
assumptions have to be made. Kolev [32] analyses the bubble three-
phase flow by making the assumption that the solid particles are carried
by the liquid or a gas/liquid mixture or the gas alone, depending on the
local volume fraction of each one of the continuous phases. A detailed
analysis of the momentum transport phenomena in fluidised bed
reactors has been reported by the authors in [4].
The bed to surface heat transfer has been widely studied through
the years for different cases, using different approaches [34,35].
Boterill [36] and Yates [37] studied the heat transfer between a
bubbling fluidised bed of smaller particles to a fixed cooling tube or a
stationary wall. Agarwal [38] studied the heat transfer to a large freely
moving particle in gas fluidised bed of smaller particles and Parmar
and Hayhurst [39] tried to measure the heat transfer coefficient for
freely moving phosphor bronze spheres (diam. 2–8 mm) around a
bed of hot sand fluidised by air. Mickley and Fairbanks [2] first pointed
out that bed-to-surface heat transfer is an unsteady process in which
“packets” of the emulsion phase carry heat to or from the surface
residing there for a short period of time, before moving back to the
bulk of the bed and being replaced by new material. Also, studies on
the influence of bed particle size to the heat transfer coefficient have Fig. 1. Fluidised bed reactor geometry and boundary conditions.
70 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79

lie in the range of 0.01–0.6 × 10− 2 m with temperature between 700 4.1.2. Momentum conservation
and 800 K. The studied biomass particles are chosen to be 0.035 × Newton's second law of motion states that the change in momen-
10− 2 m and 0.055 × 10− 2 m diameter, which is more or less the size of tum equals the sum of forces on the domain. In gas–solid fluidised
the particles, due to feeding problems, for a small rig like the one beds the sum of forces consists of the viscous force ∇ · τ̿s, the solids
studied in this paper. Bigger rigs and commercial plants use larger pressure force ∇ps, the body force εsρsg, the static pressure force
particles in the range of 2–5 mm. εs ⋅ ∇p and the interphase force Kgs(ug − us) for the coupling of gas
Depending on the regime of interest and the size of the particle the and solid momentum equations by drag forces.
heat transfer on its surface will be computed according to the litera- Gas phase:
ture. When the particles are carried only by the fluidising gas the well-
known Ranz–Marshall [44,45] correlation is used. If the particle is ∂ðεg ρg vg Þ
+ ∇⋅ðεg ρg vg ⊗vg Þ ð3Þ
found inside the bubbling bed and its diameter is bigger than the sand ∂t
particles (dp > db) the heat transfer coefficient is calculated using the = −εg ⋅∇p + ∇⋅ τg̿ + εg ρg g + Kgs ðug −us Þ;
penetration theory [2,3]. If the diameter of the particle is smaller than
that of the sand bed (dp < db), then the correlation found by Collier Solid phase:
et al. [1] is used.
The reaction kinetics is based on a two-stage, semi-global mecha- ∂ðεs ρs vs Þ
nism, with kinetic constants suitable for wood pyrolysis according to + ∇⋅ðεs ρs vs ⊗vs Þ ð4Þ
∂t
Chan et al. [7] for the primary stage and Liden et al. [8] and Di Blasi [9] for = −εs ⋅∇p−∇ps + ∇⋅ τs̿ + εs ρs g + Kgs ðug −us Þ;
the secondary reactions. This scheme has been chosen for this study
because it can predict the correct behaviour of wood pyrolysis including
where the solid-phase stress tensor is given by,
the dependence of the product yields on temperature [7,10,14].
The scope of the simulation is to quantify the heat and momentum

2
τ̿s = εs μ s ð∇us + ∇us Þ + εs λs − μ s ∇⋅us Is̿ ;
T
transport inside the reactor and the impact of biomass size on these ð5Þ
3
factors. When the particles are injected inside the reactor, they can
either be inside a bubble or inside the packed bed. The code will be
and the Gidaspow interphase exchange coefficient,
able to identify the regime of interest, depending on the local volume
fraction of the two continuous phases, and calculate the drag,
3 εs εg ρg jus −ug j −2:65
buoyancy and virtual mass forces according to the state, as well as Kgs = C εg for εg > 0:8; ð6Þ
4 d ds
the bed-to-particle heat transfer coefficient. The different size of the
particles will result in different particle volumes as well as different
heat and momentum transfer conditions. It will be investigated ε2s μ g εs ρg jus −ug j
Kgs = 150 + 1:75 for εg ≤0:8; ð7Þ
whether the different size of the particles will have a major impact on εg d2s ds
the heat transfer conditions, product yields and momentum transport
as well as the determination of an optimum particle size for this type where the drag coefficient is given by
of reactor. It has to be noted that the paper presents purely a com-
putational simulation and that no experiments were carried out for 24 h 0:687
i
this specific aspect. Cd = 1 + 0:15ðεg Res Þ ; ð8Þ
εg Res

4. Mathematical model
and

4.1. Multiphase flow governing equations ds ρg jus −ug j

Res = : ð9Þ
μg
The simulations of the bubbling behaviour of the fluidised bed
were performed by solving the equations of motion of a multifluid
system. An Eulerian model for the mass and momentum for the gas The bulk viscosity λs is a measure of the resistance of a fluid to
(nitrogen) and fluid phases, was applied, while the kinetic theory of compression which is described with the help of the kinetic theory of
granular flow, was applied for the conservation of the solid's fluctua- granular flows
tion energy. The Eulerian model is already incorporated in the main rffiffiffiffiffiffi
code of FLUENT and its governing equations are expressed in the 4 Θs
λs = εs ρs ds g0;ss ð1 + ess Þ : ð10Þ
following form. 3 π

4.1.1. Mass conservation The tangential forces due to particle interactions are summarised
Eulerian–Eulerian continuum modelling is the most commonly in the term called solids shear viscosity, and it is defined as
used approach for fluidised bed simulations. The accumulation of
mass in each phase is balanced by the convective mass fluxes. The μ s = μ s;col + μ s;kin + μ s;fr ; ð11Þ
phases are able to interpenetrate and the sum of all volume fractions
in each computational cell is unity. where the collision viscosity of the solids μ s,col is
Gas phase:
rffiffiffiffiffiffi
4 Θs
∂ðεg ρg Þ μ s;col = εs ρs ds g0;ss ð1 + ess Þ ; ð12Þ
+ ∇⋅ðεg ρg vg Þ = 0; ð1Þ 5 π
∂t
the frictional viscosity
Solid phase:

ps sinðϕgs Þ
∂ðεs ρs Þ μ s;fr = pffiffiffiffiffiffiffi ð13Þ
+ ∇⋅ðεs ρs vs Þ = 0: ð2Þ 2 I2D
∂t
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79 71

and the Gidaspow [46] kinetic viscosity There are several correlations for the drag factor f in the litera-
pffiffiffiffiffiffiffiffiffi  2 ture [49–51]. The one used in this study is the correlation of Putnam
10ρs ds Θs π 4 [51]
μ s;kin = × 1 + εs g0;ss ð1 + ess Þ : ð14Þ
96εs g0;ss ð1 + ess Þ 5
Rer ð2 = 3Þ
f =1+ for Rer < 1000 ð23Þ
The solids pressure ps, which represents the normal force due to 6
particle interactions, and the transfer of kinetic energy ϕgs are given
by f = 0:0183Rer for 1000≤Rer < 3 × 10 :
5
ð24Þ
2
ps = εs ρs Θs + 2ρs ð1 + ess Þεs g0;ss Θs ð15Þ
The second term on the right hand side of the equation represents
the gravity and buoyancy force, while the third term represents the
and
unsteady force of virtual mass force which is expressed as
ϕgs = −3Kgs Θs : ð16Þ

ρcon Vd ducon dud
Fvm = − ð25Þ
2 dt dt
4.1.3. Fluctuation energy conservation of solid particles
The solid-phase models discussed above are based on two crucial According to Kolev [32], if bubble three-phase flow (i.e. solid
properties, namely the radial distribution function g0,ss and granular particles in bubbly flow) is defined, two sub-cases are distinguished. If
temperature Θs. The radial distribution function is a measure for the volume fraction of the space among the solid particles, if they
the probability of interparticle contact. The granular temperature were closely packed is smaller than the liquid fraction (in this case the
represents the energy associated with the fluctuating velocity of Eulerian sand)
particles.
ε⁎s < εs ; ð26Þ
 
3 ∂
ðεs ρs Θs Þ + ∇⋅ðεs ρs us Θs Þ = ð−ps Is̿ + τ̿s Þ : ∇⋅us ð17Þ
2 ∂t where
+ ∇⋅ðkΘs ⋅∇⋅Θs Þ−γΘs :
1−εdm
where τ̿s is defined in Eq. (5). The diffusion coefficient of granular ε⁎s = ε ð27Þ
εdm d
temperature kΘs according to [46] is given by:
pffiffiffiffiffiffiffiffiffi then the theoretical possibility exists that the particles are carried
150ρs ds Θs π
kΘs = only by the liquid. The hypothesis is supported if we consider the ratio
384ð1 + ess Þg0;ss
of the free setting velocity in gas and liquid
 2 rffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
6 2 Θs wdg ρd −ρg ρs
× 1 + εs g0;ss ð1 + ess Þ + 2ρs ds εs g0;ss ð1 + ess Þ : ð18Þ
5 π = >> 1: ð28Þ
wds ρd −ρs ρg

The radial distribution function g0,ss is defined as

Due to great differences between gas and liquid densities, the
" !1 = 3 #−1
εs particles sink much faster in gas than in a liquid. Therefore, the drag
g0;ss = 1− ð19Þ force between gas and solid particle is zero and the drag force
εs;max
between solid and liquid is computed for a modified particle volume
fraction εp
and the collision dissipation energy as
εd
12ð1−e2ss Þg0;ss εp = ð29Þ
γΘs = pffiffiffi
2 3=2
ρs εs Θs : ð20Þ εs + εd
ds π
and an effective continuum viscosity μ eff,con
An analytical discussion of the solid-phase properties can be found
on Boemer et al. [47].

εp −1:55
μ eff;con = 1− : ð30Þ
εdm
4.2. Forces on discrete particles

If the volume fraction of the space among the solid particles, if they
The coupling between the continuous and discrete phases has
were closely packed is larger than the liquid fraction
been developed in a UDF to take into account the bubbling behaviour
of the bed. For an analytical discussion of this section the reader is
ε⁎s > εs ; ð31Þ
referred on the previous work done by the authors in this aspect [4].
Assuming a spherical droplet with material density of ρd inside a fluid,
then only
the rate of change of its velocity can be expressed as [48]

 εdg = εd ð1−εs = ε⁎Þ ð32Þ
dud f ρ s
= ðucon −ud Þ + g 1− con + Fvm ; ð21Þ
dt τu ρd
are surrounded by gas and the drag force can be calculated between
where f is the drag factor and τu the velocity response time one single solid particle and gas as for a mixture

ρd D2 εdg
τu = : ð22Þ εp = : ð33Þ
18μ con εg + εdg
72 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79

4.3. Reaction kinetics However, assuming that conductive heat transfer is dominant in
the near-particle-region, when the particle is inside the bed and its
The reaction kinetics of biomass pyrolysis is modelled using a two- diameter is larger than the sand particles (dp > db), the penetration
stage, semi-global model. The mechanism is illustrated in Fig. 2. theory can be applied with the following mixture properties [3]:
The mechanism utilises the Arrhenius equation which is defined as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
km ðρCp Þm
Ki = Ai expð−Ei = RTÞ ð34Þ hpen = ð41Þ
π ⋅t

The values of the kinetic parameters were obtained by Chan et al. with
[7] for the primary pyrolysis products, while the fourth and fifth
reaction from Liden et al. [8] and Di Blasi [9] respectively. The model km = εg kg + ð1−εg Þks ð42Þ
associates via a UDF, the reaction kinetics mechanism with the
discrete biomass particle injected in the fluidised bed. The particle's and
properties change according to the reaction mechanism due to the
phase transition phenomena. Momentum and heat transfer on the ðρCp Þm = εg ρg Cp;g + ð1−εg Þρs Cp;s : ð43Þ
particle are calculated according to the new condition in the UDF, and
the variables associated with it are updated in each time-step. Intra- If the diameter of the biomass particle is smaller than the diameter
particle secondary reactions due to the catalytic effect of char are of the sand particles then a modified Nusselt number is used to
taken into account, resulting on secondary vapour cracking. calculate the heat transfer coefficient according to the findings of
Collier et al. [1].
4.4. Heat transfer
0:62 0:2
Nu = 2 + 0:9Re ðdp =db Þ ; ð44Þ
The heat conduction along the radius of the particle is calculated
by solving the heat diffusion equation for an isotropic particle. The where
particle is discretised in the UDF for the complete pyrolysis model and
the reader is referred to the work previously done by the authors in ρg Umf dp
Re = : ð45Þ
this aspect [5]. μg



∂ 1 ∂ 2 ∂T ∂ρ The density of the particle is calculated as the sum of all the
ðρCpeff TÞ = 2 keff r + ð−ΔHÞ − ð35Þ
∂t r ∂r ∂r ∂t discrete densities that have been produced from the discretisation of
the particle as it is illustrated in Fig. 3. The radius of the particle is
The boundary condition at the surface of the particle is given by discretised to N number of grid points numbered from k = 0 to k = N,
where 0 is the centre of the particle, generating N discrete volumes.
hðT∞ −Ts Þ = −keff
∂T
∂r j r=R
ð36Þ
The density distribution along the radius of the particle is calculated
according to the discrete masses of the solid phases (wood and char)
that correspond to the specific discrete volume. The discrete masses of
and at the centre of the particle the solid phases are calculated using a linear approximation between
two neighbouring points that form a discrete volume, according to the
∂T
∂r j r=0
= 0: ð37Þ mass fraction in time of the specific phase.
The discrete volumes are given by

The effective thermal conductivity keff and effective specific heat 4 3 3

dVk = πðrk −rk−1 Þ for k = 1…N ð46Þ
capacity Cpeff are given by 3

keff = kc + jkw −kc jψw ð38Þ where

Cpeff = Cpc + jCpw −Cpc jψw : ð39Þ rk−1 = rk −dr: ð47Þ

The discrete mass dmk is calculated by

The heat transfer coefficient is evaluated from the well-known
Ranz–Marshall [44] correlation, when the particle is carried only by ðψkw + ψk−1w Þ + ðψkc + ψk−1c Þ mp
the fluidising gas dmk = for k = 1…N ð48Þ
2 N
hdp 1 = 2 1=3
Nu = = 2:0 + 0:6Red Pr : ð40Þ
kg

Fig. 3. (Left) Particle discretisation and discrete volume generation. (Right) Char
Fig. 2. Two-stage, semi-global model [52]. formation during pyrolysis.
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79 73

Therefore, the discrete particle densities dρpk along its radius are Table 1
given by Simulation Parameters.

Property Value Comment

dmk
dρpk = ð49Þ Biomass density, ρw 700 kg/m3 Wood
dVk Biomass particle diameter, dp 0.055 × 10− 2 m Fixed
0.035 × 10− 2 m Fixed
and the total average particle density is Biomass specific heat capacity, Cpw 1500 J/kg K Wood
Char specific heat capacity, Cpc 1100 J/kg K Char
Biomass thermal conductivity, kw 0.105 W/m K Wood
1 N
ρpav = ∑ dρ : ð50Þ Char thermal conductivity, kc 0.071 W/m K Char
N k = 1 pk Superficial velocity, U0 0.3 m/s ≈4Umf
Gas density, ρg 0.456 kg/m3 Nitrogen (773 K)
Gas viscosity, μ g 3.44 × 10− 5 kg/ms Nitrogen (773 K)
Gas specific heat capacity, Cp,g 1091.6 J/kg K Nitrogen (773 K)
5. Model parameters
Gas thermal conductivity, kg 0.0563 W/m K Nitrogen (773 K)
Solids particle density, ρs 2500 kg/m3 Sand
For the implementation of the model certain parameters have Sand specific heat capacity, Cp,s 835 J/kg K Fixed
been quantified and assumptions made, in order to provide, as much Sand thermal conductivity, ks 0.35 W/m K Fixed
Mean solids particle diameter, ds 440 μm Uniform distribution
as possible, an insight to the fast pyrolysis process in bubbling beds.
Restitution coefficient, ess 0.9 Value in literature
• The particles used in the simulation were assumed to be totally Initial solids packing, εs 0.63 Fixed value
Static bed height 0.08 m Fixed value
spherical, whereas the particles used in experiments can be found
Bed width 0.04 m Fixed value
on all sorts of shapes. The actual sphericity of the particles greatly Heat of reaction ΔH = − 255 kJ/kg Koufopanos et al. [53]
differs from 1. This would have an impact on the drag and virtual
mass forces and consequently on the trajectory of the particle inside
the reactor. the biomass particles. For the first second of the simulation the bed is
• The particles were injected very close to each other, to achieve fluidised without the biomass particles injected in it. Since the sand
similar initial heating and momentum transport conditions for both has gained some velocity, the biomass particles are injected and
of them. momentum is transferred from the fluidised sand to the particles.
• The model does not take into account the vapour evolution from the Examining Fig. 4, we can see that at the time that the particles are
discrete phase, as this would slow down the simulation significant- injected, a small bubble has been formed just underneath the injection
ly. The mass sources though are calculated by the code, however point. The motion of this bubble eventually carries the particles to
they are not loaded in the simulation and not released in the the top of the bed. At the time that the particles reach the unstable
computational domain. The inclusion of the tiny amount of mass splash zone of the fluidised bed, they cannot be entrained since their
source of vapours that are produced in each time-step has a major terminal velocity (Ut,350 = 0.94 m/s and Ut,550 = 1.72 m/s at a density
impact in the computational time of a 3-D simulation like the one of 700 kg/m3) is much higher than the velocity of nitrogen at the
performed in this study. For a complete analysis of the vapour freeboard of the reactor. The unstable splash zone of the fluidised bed
evolution the reader is referred to the study of the authors in this is the zone where bubbles erupt and the solids are ejected from the
aspect [5]. fluidised bed to the freeboard of the reactor. Consequently, high
• The model assumes a plug flow profile at the inlet of the reactor. nitrogen velocities appear in this region due to the eruption of bubbles,
• The geometry of the reactor has been discretised using a structured which effectively entrain the lighter char particles formed during
grid. The average side length of the computational cells is about pyrolysis. The velocity of nitrogen at the freeboard ranges from 0.47 to
1 mm resulting to a total number of 216,635 cells for the 3- 0.71 m/s at the centre of the reactor, which gradually tends to 0 m/s
Dimensional case with minimum cell volume of 9.25 × 10− 10 m3, approaching the wall. Consequently the heavy biomass particles fall
maximum cell volume of 3.3 × 10− 9 m3, and an equisize skewness of back to the bed. The particles stay close to the splash zone of the bed,
the worst element of 0.35. while the formed bubbles at the central axis of the reactor push them
• The time-step used for the simulation was of the order of 10− 5 s. The towards the wall of the reactor on the following seconds. As pyrolysis
reason for this small time advancements was that the computational of the particles takes place and density drops (Fig. 5), the smaller
algorithms in the UDF had to converge simultaneously with the particle (350 μm) is entrained at approximately 6 s of simulation
algorithms of FLUENT. The explicit central difference scheme used (5 s residence time). The entrainment of the particle is also a strong
for the discretisation of the heat diffusion equation demands very function of the radial position of the particle inside the reactor, since
small time-steps due to the small radial discretisation resulting from nitrogen velocities close to the wall tend to be much lower than at
particles of 350 μm and 550 μm in diameter. For further analysis of the centre. At the time of entrainment, the density of both particles
the CFD code structure together with the flow-chart and advantages has dropped to approximately 190 kg/m3. The respective terminal
and drawbacks of the model, the reader is referred to the previously velocities for this density are Ut,350 = 0.31 m/s and Ut,550 = 0.62 m/s,
published work of the authors in this aspect [5]. something that makes the entrainment of the smaller particle easier
The simulation parameters are shown in Table 1. even from the near wall area. The larger particle cannot be entrained if
it does not approach the centre of the reactor, where larger nitrogen
6. Results and discussions velocities are observed at the freeboard.

6.1. Bed hydrodynamics and particle positions 6.2. Heat transfer and product yields

Fig. 4 illustrates the hydrodynamics of the fluidised bed at different
simulation times together with the positions of the particles.
Whenever the particles are not visible, they are covered by the
formed bubbles. Also, the velocity magnitude vectors of nitrogen are
plotted to illustrate the motion and velocity of nitrogen inside the bed,
since it plays a significant role in the heat and momentum transport to
Fig. 6 shows the heat transfer coefficient for both particles at
different times during the simulation. We can immediately observe
that the heat transfer coefficient is much larger for the smaller particle,
having a constant value of approximately ≈310 W/m2 K when the
particle finds itself among the fluidised sand particles. This is what the
correlation of Collier et al. [1] suggests in Eq. (44). When the particle is
74 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79

carried by the bubbles or by nitrogen in the freeboard of the reactor,
the heat transfer coefficient increases up to approximately ≈500 W/
m2 K due to convection effects from the higher velocity nitrogen in that
area.
However, the heat transfer coefficient is smaller for the larger
particle when finds itself among the fluidised sand (approximately
≈150 W/m2 K) and conductive effects are assumed to be dominant,
with a maximum value of approximately ≈310 W/m2 K at the time of
injection. The heat transfer coefficient becomes greater (approxi-
mately ≈300 W/m2 K) when the particle is carried by the bubbles or
transferred to the splash zone of the bubbling bed where convective
effects by the high velocity nitrogen are dominant. This phenomenon
comes in perfect agreement with the results of Collier et al. [1]. In their
study, it was shown that the heat transfer coefficient appears to reach
a maximum value when dp > db and U/Umf slightly exceeds unity and
when the bed was fluidised well above the minimum fluidisation

Fig. 4. Bed hydrodynamics with particle positions and vectors of nitrogen velocity magnitude. Blue particle: 350 µm, Red particle: 550 µm. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79 75

Fig. 4 (continued).

velocity the heat transfer coefficient decreased. The low values of heat
transfer coefficient of the larger particle are due to the fact, that at the
time of injection the bubbles in the bed have become the dominant
phase. Therefore, the properties of the emulsion phase (Eqs. (42) and
(43)) due to high voidage variations at the heat transfer surface have
been gas-dominated. It is also easy to observe that great variations on
the heat transfer coefficient occur almost instantly during the motion
of the particle inside the reactor. This actually happens due to the
transition of the particle from dense packed regions to more dilute
zones especially when the particle floats at the splash zone of the
reactor. The bubble eruption causes the particles to be ejected at the
freeboard where high nitrogen velocities appear, giving rise to more
violent convective effects. If the particle is heavy enough and cannot
be entrained by the gas flow then it falls back and sinks on the bed
76 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79
Fig. 5. Density drop of the particles.
where heat transfer is governed by the collisions with sand particles.
It is also important to mention that the heat transfer conditions are
significantly affected by the path that the particle will follow inside
the reactor. The size and shape of the particle as well as the point and
time of injection are crucial factors. The former ones affect the amount
of drag and virtual mass exerted on the particles, while the latter ones
concern the bubble formation and generally the gas flow character-
istics inside the fluidised bed due to the unsteady nature of the flow.
Fig. 7 shows the temperature at the surface and the centre of the
particles until they reach the temperature of the reactor. Due to the
small Biot number of the particles (Bi350 ≈ 0.22, Bi550 ≈ 0.17, using the
thermal conductivity of biomass), something that was shown at a
previous study of the authors [5], the small temperature gradient
inside the particle does not affect pyrolysis products, resulting in a
uniform radial distribution of product yields. As Fig. 7 shows the
temperature difference at the surface and the centre of the particles
are not significant especially at pyrolysis elevated temperatures
(>400 °C). Thus the product yields shown at Fig. 8 are given averaged
in the volume of the particle and not radially distributed. The case of
course would not be the same if a larger diameter particle was
examined, where the Biot number is greater and the temperature
gradient more significant.
Fig. 6. Heat transfer coefficient for both particles.
Fig. 7. Temperature rise at the surface and the centre of the particles.
The biomass particles were almost totally pyrolysed in the 6 s of
the simulation (5 s of pyrolysis). As it is shown on Fig. 8 the different
heat transfer rates resulted in different biomass degradation rates
which tend to converge towards the end of the simulation. However,
the final product yields seem to be similar in this case, since the
smaller particle was not quickly entrained due to its movement close
to the wall. It is obvious though that the degradation of the smaller
particle started faster and its rate was reduced towards the end of
the simulation due to the low unreacted biomass content left on the
particle. A that point the larger particle was reacting faster and
the final yields are almost identical. Another factor that influences the
product yields is the small size of the biomass particles, resulting in a
relatively uniform temperature distribution inside the particles. Thus
there is not a thick moving layer of char formed at the surface of the
particle that would significantly affect the secondary reactions.
Therefore, the products for both particles are expected to be identical
at the complete degradation of the particle. In this case, the pyrolysis
of both particles resulted to approximately ≈5% of wood left, ≈61% of
tars, ≈21% char and ≈13% of gas. According to Di Blasi [14] the same
reaction kinetics scheme, produces yields of ≈23% char, ≈13% gas
and ≈64% tar, for complete (100%) biomass degradation at 773 K.
Therefore, the yields produced in this simulation are pretty similar
Fig. 8. Product yields for both particles.
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79
Fig. 9. Velocity components of the particles.
with those produced by the single particle models in the literature,
since complete degradation of the particle would result to almost
identical yields percentages. This indicates the correct implementa-
tion of the reaction kinetics in the UDF associated with the discrete
particle.
6.3. Particle dynamics
The velocity of the particles is calculated by integrating in time the
equation of motion for discrete particles (Eq. (21)), and their new
position is illustrated by the red and blue spheres inside the reactor
(Fig. 4). The position of the particles in the reactor is a result of the
heat transfer and phase change due to reaction effects. Different heat
transfer rates will result in different biomass degradation rates and
consequently different particle properties in time. The density drop of
the particle will differ and the drag and virtual mass forces exerted on
the particle will significantly change. The model can predict the
particle position inside the reactor, as it is subjected to pyrolysis,
taking all of these effects into account.
Fig. 9 illustrates the velocity components of the particles as they
move inside the reactor. In Section 4.2, the various parameters that
Fig. 10. Drag force per unit mass.
77
Fig. 11. Virtual mass force per unit mass.
affect the motion of the discrete particles were discussed and it is
really obvious that the most important physical ones are the diameter
of the particles and their density, since they highly define the drag,
virtual mass, gravitational and buoyant forces. Fig. 9 shows that the
velocities of the two particles are almost identical most of the
simulation time except in two cases. Close to 1.8 s of simulation there
is a sudden negative y-velocity increase for the larger particle due to a
large amount of drag force shown on Fig. 10. This is possibly due to the
bubble eruption shown on Fig. 4 at 1.5 s, which created a downward
flow of solids that tried to sink the large particle into the bed. The high
volume fraction of sand created this big amount of drag force to the
particle. Also, at the end of the simulation the smaller particle appears
to have an increase on its x and y-velocity components as it is being
ejected from the bed and accelerates towards the outlet of the reactor.
This is caused due to its large density drop and its small diameter
which made the particle easy to be entrained from the fluidising gas
even from a region close to the wall, where nitrogen velocities tend to
be lower.
Figs. 10 and 11 show the drag and virtual mass force exerted on
each particle at different times. As we can see the drag force is the
dominant momentum transport mechanism, while the virtual mass
plays a relatively insignificant role only when the particles are carried
by the sand, due to high density differences between them. An ana-
lytical description of the effect of each force on momentum transport
in fluidised bed can be found on a previous study by the authors [4].
7. Conclusions
The effect of different size particles on the heat transfer coefficient
between biomass particles and a fluidised bed was modelled.
The results showed that the different size particles which result in
different heat transfer mechanism affect the heat transfer coefficient
significantly. This results in different temperature profiles at the
surfaces and centres of the biomass particles. The temperature
gradients inside the particles can be neglected due to their small
Biot numbers, which result to uniform radial product distribution at
elevated pyrolysis temperatures (>400 °C). The final product yields
for different size particles and different heat transfer rates depend on
the residence time of the particle in the reactor. Tending towards the
complete degradation of the particles, as pyrolysis continues, the final
products tend to be similar, due to the small size of the particles and
the limited catalytic action of the char layer, due to insignificant
temperature gradients. If the particles were entrained close to 2 s of
pyrolysis (3 s of simulation) then a considerable difference of about
4% tar would be noticeable. According to the study smaller particles
78 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79

would be preferred for fast pyrolysis applications in fluidised bed dm disperse phase maximum packing
since they provide a good heat transfer mechanism according to the eff effective
literature, as well as easier entrainment during their degradation. fr frictional
Also, smaller particle sizes reduce the effect of secondary reactions g gas
resulting in higher bio-oil yields. i general index
Computational fluid dynamics models can give important infor- k radial position
mation regarding the overall process of fast pyrolysis. They can kin kinetic
efficiently used to derive important conclusions in the industrial
m mixture
sector, regarding the design and optimisation of bubbling fluidised
mf minimum fluidisation
bed reactors by deeply understanding the factors that highly influence
p particle
the process.
s solids
Nomenclature t terminal
Ai pre-exponential factor, 1/s T stress tensor
Bi Biot number, dimensionless v velocity
Cd drag coefficient, dimensionless vm virtual mass
Cp specific heat capacity, J/kg K w wood
di diameter, m 0 initial value
D droplet diameter, m 350 350 μm particle
E activation energy, J/mol 550 550 μm particle
ess restitution coefficient, dimensionless
g gravitational acceleration, m/s2
g0,ss radial distribution coefficient, dimensionless References
h convective heat transfer coefficient, W/m2 K
I̿ stress tensor, dimensionless [1] A.P. Collier, A.N. Hayhurst, J.L. Richardson, S.A. Scott, The heat transfer coefficient
between a particle and a bed (packed or fluidised) of much larger particles,
I2D second invariant of the deviatoric stress tensor, dimensionless
Chemical Engineering Science 59 (2004) 4613–4620.
f drag factor, dimensionless [2] H.S. Mickley, D.F. Fairbanks, Mechanism of heat transfer to fluidised beds,
Fi force, N/kg American Institute of Chemical Engineers Journal 1 (3) (1955) 374.
[3] J.A.M. Kuipers, W. Prins, W.P.M. Van Swaaij, Numerical calculation of wall-to-bed
k thermal conductivity, W/m K
heat-transfer coefficients in gas-fluidised beds, American Institute of Chemical
kΘs diffusion coefficient for granular energy, kg/s m Engineers Journal 38 (1992) 1079–1091.
Kgs gas/solid momentum exchange coefficient, dimensionless [4] K. Papadikis, A.V. Bridgwater, S. Gu, CFD modelling of the fast pyrolysis of biomass
m mass, kg in fluidised bed reactors, Part A: Eulerian computation of momentum transport in
bubbling fluidised beds, Chemical Engineering Science 63 (16) (August 2008)
Nu Nusselt number 4218–4227.
p pressure, Pa [5] K. Papadikis, S. Gu, A.V. Bridgwater, CFD modelling of the fast pyrolysis of biomass
Pr Prandtl number in fluidised bed reactors. Part B: heat, momentum and mass transport in bubbling
fluidised beds, Chemical Engineering Science 64 (5) (2009) 1036–1045.
r radial coordinate, m [6] K. Papadikis, S. Gu, A.V. Bridgwater, CFD modelling of the fast pyrolysis of biomass
R universal gas constant, J/mol K in fluidised bed reactors: modelling the impact of biomass shrinkage, Chemical
Re Reynolds number, dimensionless Engineering Journal 149 (1–3) (2009) 417–427.
[7] W.R. Chan, M. Kelbon, B.B. Krieger, Modelling and experimental verification of
t time, s physical and chemical processes during pyrolysis of large biomass particle, Fuel 64
T temperature, K (1985) 1505–1513.
U0 superficial gas velocity, m/s [8] A.G. Liden, F. Berruti, D.S. Scott, A kinetic model for the production of liquids from
the flash pyrolysis of biomass, Chemical Engineering Communications 65 (1988)
ui velocity, m/s
207–221.
V volume, kg/m3 [9] C. Di Blasi, Analysis of convection and secondary reaction effect effects within
wi free setting velocity, m/s porous solid fuels undergoing pyrolysis, Combustion Science Technology 90
(1993) 315–339.
[10] C. Di Blasi, Heat, momentum and mass transport through a shrinking biomass
Greek letters particle exposed to thermal radiation, Chemical Engineering Science 51 (1996)
γΘs collision dissipation of energy, kg/s3 m 1121–1132.
ΔH heat of reaction, J/kg [11] C. Di Blasi, Modelling the fast pyrolysis of cellulosic particles in fluid-bed reactors,
Chemical Engineering Science 55 (2000) 5999–6013.
εi volume fraction, dimensionless [12] C. Di Blasi, Numerical simulation of cellulose pyrolysis, Biomass and Bioenergy
Θi granular temperature, m2/s2 7 (1994) 87–98.
λi bulk viscosity, kg/s m [13] C. Di Blasi, Kinetic and heat transfer control in the slow and flash pyrolysis of
solids, Industrial and Engineering Chemistry Research 35 (1996) 37–46.
μi shear viscosity, kg/s m [14] C. Di Blasi, Comparison of semi-global mechanisms for primary pyrolysis of
ρi density, kg/m3 lignocellulosic fuels, Journal of Analytical and Applied Pyrolysis 47 (1998) 43–64.
τv velocity response time, s [15] J.J. Saastamoinen, Model for Drying and Pyrolysis in an Updraft Gasifier, in: A.V.
τ̿i
Bridgwater (Ed.), Advances in Thermochemical Biomass Conversion, Blackie,
stresses tensor, Pa London, 1993, pp. 186–200.
ϕgs transfer rate of kinetic energy, kg/s3 m [16] J. Saastamoinen, J.-R. Richard, Simultaneous drying and pyrolysis of solid fuel
ψ mass fraction particles, Combust Flame 106 (1996) 288–300.
[17] B.V. Babu, A.S. Chaurasia, Modelling and simulation of pyrolysis: influence of particle
size and temperature, Proceedings of International Conference on Multimedia and
Subscripts Design, Vol. 4, Mumbai, India, 2002, pp. 103–128.
av average [18] B.V. Babu, A.S. Chaurasia, Modelling and simulation of pyrolysis: effect of convec-
tive heat transfer and orders of reactions, Proceedings of International Symposium
b bed material
and 55th Annual Session of IIChE (CHEMCON-2002), OU, Hyderabad, India, 2002,
c char pp. 105–106.
con continuous phase [19] B.V. Babu, A.S. Chaurasia, Modelling for pyrolysis of solid particle: kinetics and
col collision heat transfer effects, Energy Conversion and Management 44 (2003) 2251–2275.
[20] B.V. Babu, A.S. Chaurasia, Modelling, simulation, and estimation of optimum param-
d droplet eters in pyrolysis of biomass, Energy Conversion and Management 44 (2003)
D Drag 2135–2158.
 K. Papadikis et al. / Fuel Processing Technology 91 (2010) 68–79
[21] B.V. Babu, A.S. Chaurasia, Modelling and simulation of pyrolysis of biomass: effect
of heat of reaction, Proceedings of International Symposium on Process Systems
Engineering and Control (ISPSEC 03) for Productivity Enhancement Through
Design and Optimisation, IIT-Bombay, Mumbai, India, 2003, pp. 181–186.
[22] B.V. Babu, A.S. Chaurasia, Pyrolysis of biomass: improved models for simultaneous
kinetics and transport of heat, mass, and momentum, Energy Conversion and
Management 45 (2004) 1297–1327.
[23] A.A. Rostami, M.R. Hajaligol, S.E. Wrenn, A biomass pyrolysis submodel for CFD
applications, Fuel 83 (2004) 1519–1523.
[24] J.J. Saastamoinen, Simplified model for calculation of devolatilisation in fluidised
beds, Fuel 85 (2006) 2388–2395.
[25] C.C. Pain, S. Mansoorzadeh, J.L.M. Gomes, C.R.E. de Oliveira, A numerical
investigation of bubbling gas-solid fluidised bed dynamics in 2-D geometries,
Powder Technology 128 (1) (2002) 56–77.
[26] S.V. Apte, K. Mahesh, T. Lundgren, A Eulerian-Lagrangian model to simulate two-
phase/particulate flows, Centre for Turbulence Research, Annual Research Briefs
2003.
[27] K.D. Kafui, C. Thornton, M.J. Adams, Discrete particle-continuum fluid modelling of
gas–solid fluidised beds, Chemical Engineering Science 57 (2002) 2395–2410.
[28] G.A. Bokkers, J.A. Laverman, M. Van Sint Annaland, J.A.M. Kuipers, Modelling of
large-scale dense gas–solid bubbling fluidised beds using a novel discrete bubble
model, Chemical Engineering Science 61 (2006) 5590–5602.
[29] M. Ishii, N. Zuber, Relative motion and interfacial drag coefficient in dispersed
two-phase flows of bubbles, drops and particles, Paper 56 a, AIChE 71st Ann, Meet,
Miami, 1978.
[30] M. Ishii, T.C. Chawla, Local drag laws in dispersed two-phase flow, NUREG/CR-
1230, ANL-79-105, 1979.
[31] A. Tomiyama, Struggle with computational bubble dynamics, Third International
Conference on Multiphase Flow, ICMF 98, Lyon, France, 1998.
[32] N.I. Kolev, Multiphase Flow Dynamics 2, Thermal and Mechanical Interactions,
2nd editionSpringer, 2005.
[33] M. Sommerfeld, Bubbly Flows: Analysis, Modelling and Calculation, Springer,
2004.
[34] D. Kunii, O. Levenspiel, Fluidization Engineering, 2nd edition, Butterworth-
Heinemann, 1991.
[35] B.R. Andeen, L.R. Glicksman, Heat transfer to horizontal tubes in shallow fluidised
beds, ASMEpaper 76-HT-67, 1976.
[36] J.S.M. Boterill, Fluid-bed Heat Transfer, Academic Press, New York, 1975.
79
[37] J.G. Yates, Fundamentals of fluidised-bed chemical processes, Butterworths,
London, 1983.
[38] P.K. Agarwal, Transport phenomena in multi-particle systems—IV. Heat transfer to
a large freely moving particle in gas fluidised bed of smaller particles, Chemical
Engineering Science 46 (1991) 1115–1127.
[39] M.S. Parmar, A.N. Hayhurst, The heat transfer coefficient for a freely moving
sphere in a bubbling fluidised bed, Chemical Engineering Science 57 (2002)
3485–3494.
[40] A. Schmidt, U. Renz, Eulerian computation of heat transfer in fluidised beds,
Chemical Engineering Science 54 (1999) 5515–5522.
[41] A. Schmidt, U. Renz, Numerical prediction of heat transfer in fluidised beds by a
kinetic theory of granular flows, International Journal of Thermal Sciences 39
(2000) 871–885.
[42] D. Geldart, Types of gas fluidization, Powder Technology 7 (1973) 285.
[43] Bridgwater, A.V., Principles and practice of biomass fast pyrolysis processes for
liquids, Journal of Analytical and Applied Pyrolysis 51 (1999) 3–22.
[44] W.E. Ranz, W.R. Marshall, Evaporation from drops, Part I, Chemical Engineering
Progress 48 (1952) 141–146.
[45] W.E. Ranz, W.R. Marshall, Evaporation from drops, Part II, Chemical Engineering
Progress 48 (1952) 173–180.
[46] D. Gidaspow, Multiphase Flow and Fluidization: Continuum and Kinetic Theory
Descriptions, Academic Press, New York, 1994.
[47] A. Boemer, H. Qi, U. Renz, Eulerian simulation of bubble formation at a jet in a two-
dimensional fluidised beds, International Journal of Multiphase Flow 23 (5) (1997)
927–944.
[48] C.T. Crowe, M. Sommerfeld, Y. Tsuji, Multiphase Flows with Droplets and Particles,
CRC Press LLC, 1998.
[49] L. Schiller, A. Naumann, Uber die grundlegenden Berechungen bei der Schwerk-
raftaufbereitung, Ver. Deut. Ing., vol. 77, 1933, p. 318.
[50] R. Clift, W.H. Gauvin, The motion of particles in turbulent gas streams, Proc.
Chemeca '70, vol. 1, 1970, p. 14.
[51] A. Putnam, Integrable form of droplet drag coefficient, ARS JNl, vol. 31, 1961,
p. 1467.
[52] F. Shafizadeh, P.P.S. Chin, Thermal deterioration of wood, A.C.S. Symposium Series
43 (1977) 57–81.
[53] C.A. Koufopanos, N. Papayannakos, G. Maschio, A. Lucchesi, Modelling of the
pyrolysis of biomass particles. Studies on kinetics, thermal and heat transfer
effects, The Canadian Journal of Chemical Engineering 69 (1991) 907–915.