Simulation of a Moving-Bed Reactor

and a Fluidized-Bed Reactor by DPM

and MPPIC in OpenFOAM®

Kwonwoo Jang, Woojoo Han and Kang Y. Huh

Abstract Simulations are performed for a moving-bed reactor in a rotary kiln and
a fluidized-bed reactor in a FINEX plant. The DEM (Discrete Element Method) and
the MPPIC (Multiphase Particle-In-Cell) methods are combined with a compress-
ible reacting flow in OpenFOAM® 2.3.x. The computational load is reduced by the
DPM (Discrete Particle Method), in which a computational parcel represents a fixed
number of identical particles in the DEM. The slumping and rolling modes are repro-
duced by adjusting particle–particle and particle–wall friction coefficients to match
the regime map in Henein et al. [1]. Validation is performed in a pilot-scale rotary
kiln for reduction of iron ore with heat input from LPG (Liquefied Petroleum Gas).
Simulation results in a lab-scale reactor are validated against those by commercial
software and experimental data for the fluidized-bed reactor. Simulation results show
good agreement with actual operating data for an industrial-scale fluidized-bed reac-
tor in the FINEX process. Reasonable trends are reproduced for the bed burners and
the collective motion of particles of different diameters in the FINEX plant.

1 Introduction

Moving-bed reactors and fluidized-bed reactors are widely used in the industries
of petroleum processing, coal gasification, nuclear plants, steel manufacturing and
water and waste treatment [2]. Both bed systems usually involve high particle concen-
trations, intense chemical reaction, and strong convective and radiative heat transfer.
Many experiments have been carried out in laboratories and pilot-scale reactors to
propose and validate the analytical models for gas flow and particle behavior in the
reactor. Computational fluid dynamics (CFD) is a promising tool for understanding
complicated multiphase phenomena with rapidly increasing low-cost computational
power. There are two modeling approaches for multiphase flow, Eulerian–Eulerian

K. Jang · W. Han · K. Y. Huh (B)

Department of Mechanical Engineering, Pohang University
of Science and Technology, Pohang, Gyungbuk 790-784, Republic of Korea
e-mail: [email protected]

© Springer Nature Switzerland AG 2019 419

J. M. Nóbrega and H. Jasak (eds.), OpenFOAM® ,
https://doi.org/10.1007/978-3-319-60846-4_30
420 K. Jang et al.

and Eulerian–Lagrangian, for continuous and discrete particle phases [3–5]. Two
Eulerian–Lagrangian approaches of interest are the Discrete Element Method (DEM)
and the Multiphase Particle-in-Cell (MPPIC) method. The DEM solves the transport
and collision of all particles going through transitional and rotational motion. The
computational load for DEM tends to increase exponentially with the total number of
particles for a large-scale industrial system [6]. There have been two approaches to
reducing the computation time of DEM; one is the DPM employing computational
parcels with each representing a fixed number of particles sharing the same charac-
teristics and the other is the MPPIC model [7], which considers the stress gradient
of the solid phase through an interaction force in the Eulerian field. The MPPIC is
more efficient than the DEM for a large-scale problem, although a sufficient number
of parcels are required to guarantee stability in the corresponding Eulerian grid and
flow field [8]. In this work, we applied the open source code, OpenFOAM® , to sim-
ulate particles and gas flow in a moving-bed and a fluidized-bed, coupled with 3D
turbulent flow, combustion, and heat transfer.

2 Physical Models

2.1 Discrete Particle Method (DPM)

The solid phase is represented by individual Lagrangian parcels, whereas the gas
phase is described by the following Eulerian mass, momentum, and energy conser-
vation equations:
Mass:
 
∂ θfρf  
+ ∇ · θ f ρ f u f  δ ṁ pf ; (2.1)
∂t
Momentum:
 
∂ θfρfuf    
+ ∇ · θ f ρ f u f u f  −θ f ∇ p + θ f ρ f g + ∇ · θ f τ f + F; (2.2)
∂t
Species:
 
∂ θ f ρ f Y f,i    
+ ∇ · θ f ρ f Y f u f  ∇ · ρ f Dθ f ∇Y f,i + δ ṁ i,chem ; (2.3)
∂t
Energy:
∂    Dp
θfρf h f + ∇ · θfρf h fuf  θf − ∇q + Q̇ + Sh . (2.4)
∂t Dt
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 421

Each particle is tracked for its translational and rotational motion by the following
equations of motion [9]:

d2 ri
mi  f i + m i g, (2.5)
dt 2
d2 ωi
Ii 2  ti , (2.6)
dt
for the mass of the ith particle, m i , its position, ri , and the total force, f i , due to
drag and contact with wall and other particles, except the gravitational force. Ii and
ti are the moment of inertia and the total torque of the ith particle. f i is given as a
sum of the drag force, f D , the contact force, f C , and the net pressure force by the
surrounding gas flow. f D is given by the Ergun-Wen and Yu model [10] to consider
the effect of gas volume fraction in a dense particle bed. f C is the sum of the normal
and tangential contact forces, f norm and f tang , estimated as

f norm  kδ + λ0 vn (2.7)
f tang  −kt ξ − λt vt (2.8)

where f norm is given by the linear contact model, whereas f tang is given by the slid-
ing/sticking friction model. k, λ0 , δ and vn are, respectively, the spring stiffness,
damping coefficient, overlap in the soft sphere model and relative velocity in the
normal direction. kt and λt are the tangential spring stiffness and tangential dissipa-
tion parameter. ξ and vt are the tangential overlap and tangential relative velocity.
The friction force is modeled as f friction  μf norm for the friction coefficient, μ. The
tangential friction force, f tang , is limited as
  
f tang  min f C ,  f tan g  . (2.9)

In the DPM, each computational parcel represents n identical particles to reduce

excessive computational load of the DEM to a manageable level. The force, f in ,
excluding the contact force on a parcel, is modeled as [11]

f in  n f i1 , (2.10)

where f i1 is the force on each single particle according to an experimental correlation.

Heat and mass transfer for a parcel are similarly modeled as

den  n de1 , (2.11)

dm n  n dm 1 , (2.12)

where de1 and dm 1 are the estimations for each single particle by experimental
correlations. The normal and tangential contact forces are represented by the com-
422 K. Jang et al.

putational parcels contacting each other in the same way as the individual particles
in Eqs. (2.7) and (2.8).

2.2 Multiphase Particle-In-Cell (MPPIC)

The gas phase equations in the MPPIC are the same as those in the DEM. The proba-
bility distribution function, φ x, u p , m p , t , is defined to describe particle dynamics
in terms of particle position, x, particle velocity, u p , and particle mass, m p . The
particle phase is governed by the Liouville equation [12],
∂φ
+ ∇ · (φu p ) + ∇u p · (φA)  0, (2.13)
∂t
where the particle acceleration is given by
1 1
A  D p (u f − u p ) − ∇p + g − ∇τ. (2.14)
ρp θs ρ p

The four terms on the right represent, respectively, drag force, pressure gradient,
gravity, and gradient in the interparticle stress, τ [7]. The drag coefficient, D p , is given
by the Ergun-Wen and Yu model [10]. u f and u p are the fluid and particle velocities.
ρ p is the density of particles and θs is the particle volume fraction. Equation (2.14)
corresponds to Eq. (2.5) with a different expression for the particle contact force, f C ,
given in terms of the interparticle stress by the Harris and Crighton model [13] as
β
Ps θs
∇τ  , (2.15)
max[(θCP − θs ), ε(1 − θs )]

where θCP is the maximum volume fraction for the packed particle bed and Ps is the
corresponding particle pressure. Aureais et al. [14] recommended a constant value
between 2 and 5 for β. ε is a small number on the order of 10−7 [15].

3 Implementation Strategy in OpenFOAM®

The basic libraries, CollidingCloud and MPPICCloud, for DEM and MPPIC were
included in OpenFOAM® version 2.3.x. The multiphase flow solver with these
libraries was provided for a single species in an incompressible form. We developed
a new solver and the associated libraries in a compressible form in three sequen-
tial stages for gas–solid reaction and heat transfer in a moving-bed reactor and a
fluidized-bed reactor. First, the basic CollidingCloud and MPPICCloud are com-
bined together with the physical models for heat transfer and gas–solid reaction. The
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 423

Fig. 1 Implementation strategy in OpenFOAM®

developed cloud libraries are respectively called ReactingMultiphaseCollidingCloud

and ReactingMultiphaseMPPICCloud. Second, we modified the basic incompress-
ible solver as a new compressible solver. The developed libraries are implemented
into the new solver for heat transfer, reaction, and particle motion, as depicted in
Fig. 1. In the final stage, we added the sub-models for chemical reaction and heat
transfer. The eddy dissipation model was implemented for diffusion-controlled com-
bustion in a compressible turbulent flow with multiple species transport in the gas
flow solver [16]. The random pore model is implemented to consider the surface reac-
tion on DPM particles [17]. The WSGGM (Weighted Sum of Gray Gases Model)
was implemented for radiation in the gas phase [18].

4 Results for the Moving-Bed Reactor

4.1 Case Setup

Henein et al. [1] developed a regime map for particle motion in a rotary kiln in terms
of bed depth versus rotational speed or percent fill versus Froude number. It includes
particle motion in the slipping, slumping, rolling, cascading, and cataracting modes.
The experimental conditions are listed in Table 1 [19]. The kiln is a cylinder with a
diameter of 0.4 m and an axial length of 0.46 m. Particles are composed of limestone
with a mean diameter of 4.3 mm. The friction coefficients are specified arbitrarily
424 K. Jang et al.

Table 1 Experimental Property Range

conditions in Henein et al.
[19] Material Limestone
Kiln axial length (m) 0.46
Kiln diameter (m) 0.4
RPM range (rpm) 0.5–10
Bed depth range (m) 0.03–0.07
Particle diameter (mm) 4.3

Fig. 2 Computational mesh

for the pilot-scale rotary kiln
in Tsuji [20]

as 0.1 between particles and 0.2 between particles and wall, with no reliable models
available in the given conditions.
Tsuji [20] investigated experimentally the reduction behavior of iron ore in a pilot-
scale rotary kiln. Simulation is performed for particles and gas flow in the 3D domain
in Fig. 2 in this work. Table 2 lists the operation condition and ore compositions.
The kiln rotates at 0.33 rpm with a diameter of 1.8 m and an axial length of 1.0 m.
The bed is composed of briguettes of a homogeneous mixture at a uniform size of
30 mm × 25 mm × 15 mm. They are represented as spherical particles of the mean
diameter of 27.8 mm of the homogeneous mixture treated as a single component in
simulation. The initial particle temperature is 1273 K [20]. The reduction ratio is
set initially at 5%, as in experiment. The friction coefficients are the same as those
employed for the regime map in Henein et al. The adiabatic condition is specified on
all wall boundaries.
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 425

Table 2 Experimental conditions and ore chemical compositions in Tsuji [20]

Property Value Chemical Mass%
RPM 0.33 SiO2 44.36
Ore (kg) 560 Fe 9.05
Coal (Anthracite) (kg) 101 Al2 O3 0.41
Limestone (kg) 44.0 Ni 2.45
Particle size (mm) 30 × 25 × 15 CaO 0.08
Fuel LPG MgO 26.68
Fuel mass flow rate 30
(m3 /h)
Air mass flow rate 709.5
(m3 /h)

Table 3 Simulation cases Case # rpm h (m) θ (°) Mode

and modes of particle motion
1 10 0.035 23.2 Rolling
2 10 0.045 23.0 Rolling
3 10 0.053 23.3 Rolling
4 10 0.07 23.1 Rolling
5 0.5 0.035 21.3–27.2 Slumping
6 0.5 0.045 19.4–24.4 Slumping
7 0.5 0.053 21.3–25.8 Slumping
8 0.5 0.07 30.3 Rolling

4.2 Results and Discussion

4.2.1 Transverse Particle Motion

The cases in Table 3 involve bed depths between 0.035 and 0.07 m at 10 and 0.5 rpm.
Simulation was performed with 4000 DPM parcels to represent about 105 particles
for 14.8 kg of limestone [19]. Figure 3 shows simulation results for transverse particle
motion in Henein’s regime map. The particles were in the rolling mode for all bed
depths at 10 rpm with an inclination angle, θ , of about 23°. The inclination angle
did not show any explicit dependence on the bed depth. The particles showed the
slumping mode at 0.5 rpm with particle movement of θ between about 20° and 25°,
except for Case 8. The particle bed repeated the sequence of moving up with the wall
and then sliding down against the wall periodically in the slumping mode. In the
rolling mode, the external bed shape remains stationary as the particles slide down,
forming an active layer on the top surface and an inactive region in the interior region
[19].
426 K. Jang et al.

Fig. 3 Transverse particle

motion in the regime map

4.2.2 Reduction of Iron Ore

Figure 4a shows the mean temperature distribution on the cross-sectional plane in the
quasi-equilibrium state. Note the lower temperature in the bed due to the heat sink as
a result of endothermic coal gasification reaction. There is a significant temperature
gradient in the axial direction in the bed, as well as in the freeboard, as reduction
proceeds with combustion gas flowing toward the outlet. The peak temperature is
about 1800 K at the entrance of the kiln, while the minimum temperature of about
1000 K occurs at the lower right corner in the bed in Fig. 4a.
Figure 4b, c shows the temperature and the degree of reduction of the particles at
57 min since the initial state. Particles show higher temperatures and higher reduction
ratios near the burner as a result of enhanced heat transfer from the freeboard gas at
its maximum temperature. The reduction ratio is relatively uniform due to efficient
mixing in the radial direction in Fig. 4c, while particles show higher temperatures
on the bed surface as a result of heat transfer from the freeboard gas. This results
from a heat transfer faster than particle mixing and reduction chemistry to maintain
a non-negligible temperature gradient in the bed. Results show inefficient mixing
of particles in the axial direction as compared with mixing in the radial direction.
All computations were performed on a cluster with 60 cores of the 2nd Intel Xeon
Processor E5-2650 at 2.3 GHz. It took about seven days for the simulation of one hour
of real time with one thousand DPM parcels and about 55,000 fluid meshes. This
could be evidence of the current DPM implementation not being fully optimized
for parallel processing, since CFDEM [21] showed much faster performance for
comparable simple problems in an incompressible flow.
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 427

Fig. 4 Temperature distribution in the pilot-scale rotary kiln (a) and temperature (b) and reduction
ratio (c) of the particles at 57 min in the pilot-scale rotary kiln

5 Results for the Fluidized-Bed Reactor

5.1 Lab-Scale Fluidized-Bed Reactor

5.1.1 Comparison with the Commercial Software

We compared the results with those by the commercial software, ANSYS-FLUENT

[22], for validation of the newly developed solver and libraries. We selected the
case of a simple uniformly fluidized-bed reactor in Goldschmidt et al. [23]. The
computational domain is rectangular with the width of 0.15 m, the depth of 0.015 m
and the height of 0.45 m as shown in Fig. 5. Glass beads of a uniform diameter of
2.5 mm are filled at the initial bed height of 0.15 m. Details of the operation conditions
and the numerical models are summarized in Table 4. Figure 5 shows instantaneous
contours of the gas volume fraction by ANSYS-FLUENT and OpenFOAM® . A large
bubble at the bed surface shows similar shapes with no significant difference in the
other aspects of the results by the two programs.
We also compared the results by OpenFOAM® with those by Barracuda [15] which
is another commercial software widely used in fluidized-bed reactors. We performed
simulation of Lin et al. [24] involving a cylinder filled with glass beads with the
diameters ranging between 0.42 and 0.6 mm. The initial bed height is 11 cm and the
mean fluidization velocity is 0.648 m/s. A 2D axisymmetric computational domain
is shown in Fig. 6. Details of the operation conditions and the numerical models are
summarized in Table 5. Figure 6 shows comparison of the measured particle volume
fractions with the calculation results by OpenFOAM® and Barracuda. There is no
428 K. Jang et al.

Fig. 5 Schematic configuration and instantaneous contours of the gas volume fraction a by
ANSYS-FLUENT and b by OpenFOAM®

Table 4 Simulation conditions for the reactor in Goldschmidt et al. [23]

Operating conditions Value Numerical model Value
Inlet gas Air Number of grid 675/2D
Fluidization velocity 1.875 Time step size (s) 2 × 10−4
(m/s)
Gas density (kg/m3 ) 1.2 Drag model Ergun-Wen and Yu
Particle density 2526
(kg/m3 )
Number of particle 24,750 Particle–particle Harris-Crighton
interaction model
Gas temperature (K) 293

significant difference between the two programs with the predicted results showing a
similar trend with experimental data but with significant deviation at some locations.
Simulations show the particle volume fractions larger than measurements at the
bottom, as the voidage due to high gas jet velocities could not be reproduced by the
MPPIC model. Deviations in the peak volume fraction are caused by the particle
behaviors with strong collision and circulation not properly taken into account in the
MPPIC model and might be improved by modification of the model constants, Ps
and β, in Table 5.
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 429

Fig. 6 Schematic configuration and comparison of average particle volume fractions

Table 5 Simulation conditions for the reactor in Lin et al. [24]

Operating conditions Value Numerical model Value
Fluid density (kg/m3 ) 1.093 Number of grid 360/2D axisymmetric
Fluid viscosity (Pa s) 1.95 × 10−5
Particle density 2500 Time step size (s) 1 × 10−4
(kg/m3 )
Number of particle 20,736 Ps (Pa)/β 75/5

5.1.2 Comparison with DEM Results

We compared the results by OpenFOAM® with experimental data and the results by
DEM for another lab-scale reactor in Muller et al. [25]. It has a rectangular shape of
the width of 44 mm, the depth of 10 mm and the height of 1200 mm, similar with
the shape in Fig. 5. Poppy seeds were used as fluidizing particles at the initial bed
height of 30 mm. Details of the operation conditions and the numerical models are
summarized in Table 6.
Figure 7 shows comparison of the measured gas volume fractions with the simula-
tion results at two different heights. Note good agreement of the gas volume fractions
calculated by DEM and MP-PIC approaches with experimental data in Fig. 7. The
MP-PIC considers only linear collision of particles, whereas the DEM takes into
account both linear and rotational collisions. Better agreement is observed in the
430 K. Jang et al.

Table 6 Simulation conditions for the reactor in Muller et al. [25]

Operating conditions Value Numerical model Value
Air temperature (K) 298.15 Number of grid 3000/3D
Fluid viscosity (Pa s) 1.8 × 10−5 Time step size (s) 1 × 10−4
Particle density 1000 Operating time (s) 23
(kg/m3 )
Number of particle 9240 Ps (Pa)/β 10/2

Fig. 7 Comparison of the mean gas volume fractions by experiment, DEM and MP-PIC at the
heights of a 16.4 mm and b 31.2 mm

central region at the mid-axial location of y  31.2 mm, whereas there is larger
deviation near the walls on both sides. No significant difference was confirmed in
the characteristics of the fluidized bed with constant gas density between the original
MPPICFoam and the newly developed ThermoMPPICFoam.

5.2 Industrial-Scale Fluidized-Bed Reactor

5.2.1 Case Setup

Figure 8 shows the FINEX process, a new iron-making technology under develop-
ment by POSCO. It produces less pollutants and is more economical than the conven-
tional blast furnace since it uses lower quality ore and coal without any pretreatment
processes such as cokes and sinter plants [26]. The major role of fluidized-bed reac-
tors is in reducing the ore fine and increasing the particle temperature to produce
hot compacted iron (HCI). The R2 reactor is about 17 m high and the radius of the
distributor is about 4.5 m. A mixture of carbon monoxide, methane, and hydrogen
gas is injected through 960 holes in the distributor on the bottom. The iron ore has
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 431

Fig. 8 A schematic diagram of the FINEX process [26]

Fig. 9 Simplified geometry of the fluidized-bed R2 reactor; a full reactor, b 1/10 sector

diameters between 0.063 and 8 mm and ten burners are installed as heat sources
around the bottom of the bed in the R2 reactor. The oxygen burner has two small
nozzle holes, with the two injections meeting at an angle of 45°. A 1/10 sector of
the reactor is modeled to reduce the computational cost in this work. The 1/10 sector
consists of approximately 0.8 million hexahedral cells and 2 million computational
parcels in MPPIC. The simplified reactor geometry is shown in Fig. 9.
432 K. Jang et al.

Fig. 10 Instantaneous images of a gas volume fraction and b particles motion

5.2.2 Simulation Under Actual Operating Conditions

The standard k−ε model is employed with the EDM (Eddy Dissipation Model) as
the turbulent combustion model [16]. The WSGGM [18] is employed to estimate the
gas phase absorption and emission coefficients for radiative heat transfer. Figure 10
shows instantaneous images of the gas volume fraction and the particle motion of
different particle diameters. It shows a reasonable trend of the internal field, including
the gas volume fractions. It is confirmed that bubbles form near the bottom by the
jet flow through the distributor. Most small particles tend to move and accumulate
along the reactor wall in the upper reactor region.
The axial pressure difference is important for confirming the validity of the predic-
tions of particle motion. It largely depends on the total mass of particles and remains
approximately constant beyond the minimum fluidization velocity [27]. Figure 11
shows good agreement with the measured pressure differences, with minor deviation
due to inaccurate locations of the pressure transducers away from the distributor.
Particles are recirculated continuously by the gas flow through the distributor, while
the gas temperature remains relatively uniform due to violent gas phase mixing and
heat transfer with particles by turbulence.

6 Conclusion

(1) Simulations are performed for a coupled solution of particle bed, turbulent
flow, combustion and heat transfer in a pilot-scale rotary kiln and a full-scale
fluidized-bed reactor in the FINEX plant. New libraries are developed for DPM
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 433

Fig. 11 Measurements and predictions of a pressure and b mean temperature

and MPPIC to reduce the computational load and to be combined with the
compressible reacting flow solver in OpenFOAM® version 2.3.x.
(2) Particle motions are reproduced for the rolling and slumping modes of a moving
bed in Henein’s regime map. Particles flow continuously forming an active layer
on the top surface of the bed, while particles move at a slower rate in the inactive
core region in the bed in the rolling mode. More work is required for proper
specification of the friction coefficients in different regimes.
(3) Validation is performed in a pilot-scale rotary kiln for reduction of iron ore with
heat input from LPG fuel. Results show non-negligible variation of particle
temperature and reduction ratio in the radial direction as well as in the axial
direction in the bed. This is due to heat transfer with freeboard gas occurring
faster than the coal gasification reaction and radial mixing of particles under the
given experimental conditions.
(4) The simulation results were validated against those by ANSYS-FLUENT, Bar-
racuda and experimental data in the lab-scale fluidized-bed reactors in literature.
There was no significant difference between the results by OpenFOAM® and the
commercial software, with good agreement with available experimental data.
(5) Particle motions including recirculation, bubble formation and destruction are
successfully reproduced in the fluidized-bed R2 reactor of the FINEX plant. The
gas temperature is relatively uniform due to efficient mixing and heat transfer
with particles in the bed. Good agreement is shown for the axial pressure dif-
ference, although with minor deviation due to inaccurate pressure measurement
locations.
434 K. Jang et al.

References

1. H. Henein, J.K. Brimacombe, A.P. Watkinson (1983) The modeling of transverse solids motion
in rotary kilns. Metall Trans B 14 (2):207–220.
2. P. Trambouze, J.-P. Euzen (2004) Chemical Reactors: From Design to Operation. Technip,
Paris.
3. P.A. Cundall, O.D. Strack (1979) A discrete numerical model for granular assemblies. Geotech-
nique 29 (1):47–65.
4. Y. Tsuji, T. Kawaguchi, T. Tanaka (1993) Discrete particle simulation of two-dimensional
fluidized bed. Powder Technol 77 (1):79–87.
5. Y. Tsuji (2007) Multi-scale modeling of dense phase gas–particle flow. Chem Eng Sci
62(13):3410–3418.
6. Dalibor. Jajcevic, Eva Siegmann, Charles Radeke et al (2013) Large-scale CFD-DEM simula-
tions of fluidized granular systems. Chem Eng Sci 98(19):298–310.
7. M. J. Anderews, P.J. O’Rourke (1996) The multiphase particle-in-cell (MP-PIC) method for
dense particulate flows. Int J Multiphase Flow 22(2):379–402.
8. Sreekanth Pannala, Madhava Syamlal, Thomas J. O’Brien (2010) Computational gas-solids
flows and reacting system: Theory, Methods and Practice. IGI Global, New York.
9. Stefan Luding (2008) Introduction to discrete element methods: basic of contact force models
and how to perform the micro-macro transition to continuum theory. European Journal of
Environmental and Civil Engineering 12:785–826.
10. C. Y. Wen, Y. H. Yu (1966) A generalized method for predicting the minimum fluidization
velocity, Chem. Eng. AIChE J 12(3):610–612.
11. Mikio Sakai, Yoshinori Yamada, Yusuke Shigeto et al (2010) Large-scale discrete element
modeling in a fluidized bed. Int J Numer Methods Fluids 64:1319–1335.
12. F.A. Williams (1985) Combustion Theory, 2nd edition. The Benjamin/Cumming Publishing,
California.
13. S. E. Harris, D. G. Crighton (1994) Solitons, solitary waves, and voidage disturbances in
gas-fulidized beds. J Fluid Mech 266:243–276.
14. F. M. Aureais, R. Jackson, W. B. Russel (1988) The solution of shocks and the effects of
compressible sediments in transient settling. J Fluid Mech 195:437–462.
15. D. M. Snider (2001) An incompressible three-dimensional multiphase particle-in-cell method
for dense particle flows. J Comput Phys 170(2):523–549.
16. N. Peters (2000) Turbulent Combustion. Cambridge University Press, Cambridge.
17. H. Watanabe, M. Otaka (2006) Numerical simulation of coal gasification in entrained flow coal
gasifier. Fuel 85:1935–1943.
18. John R. Howell, Pinar Menguc, Robert Siegel (2015) Thermal radiation heat transfer, 6th
edition. CRC press, Florida.
19. H. Henein, J.K. Brimacombe, A.P. Watkinson (1983) Experimental study of transverse bed
motion in rotary kilns. Metall Trans B 14 (2):191–205.
20. H. Tsuji (2012) Behavior of reduction and growth of metal in Smelting of Saprolite Ni-ore in
a Rotary Kiln for Production of Ferro-nickel Alloy. ISIJ Int 52 (6):1000–1009.
21. CFDEM (2016) CFEDM Benchmarks. http://www.cfdem.com/cfdem-benchmarks.
22. ANSYS. ANSYS FLUENT User’s Guide. ANSYS Inc., Cannonsburg, PA, 2012.
23. M.J.V. Goldschmidt, R. Beetstra, J.A.M. Kuipers (2004) Hydrodynamic modelling of dense
gas-fluidised beds: comparison and validation of 3D discrete particle and continuum models.
Powder Technol 142:23–47.
24. J. S. Lin, M. M. Chen, B. T. Chao (1985) A novel radioactive particle tracking facility for
measurement of solids motion in gas fluidized beds. AIChE J 31(3):465–473.
Simulation of a Moving-Bed Reactor and a Fluidized-Bed Reactor … 435

25. C. R. Muller, S. A. Scott, D. J. Holland et al. (2009) Validation of a discrete element model
using magnetic resonance measurements. Particuology 7(4):297–306.
26. POSCO (2014) Technological Advance of the FINEX® Ironmaking Process. Paper presented
at Asian Pellets and DRI Conference, Zurich Marriott Hotel, Singapore, 8–9 July 2014.
27. D. Kunii, O. Levenspiel (1991) Fluidization Engineering, 2nd edition. Butterworth-Heinemann,
Oxford.