CFD Simulation of Process-driven Particle Fragmentation in a Coal Bed

Gasifier
Franz Holzleithnera, Roland Eisla, Markus Haidera, Georg Aichingerb
a
Institute for Energy Systems and Thermodynamics, Vienna University of Technology,
Getreidemarkt 9 / E302, A-1060 Vienna, Austria.
b
Siemens VAI Metals Technologies GmbH & Co, Turmstrasse 44, Ironmaking Technology, A-4031 Linz,
Austria

Abstract:

Good gas-solid contact is essential in a coal bed gasifier such as in a COREX® melter gasifier. The charged
particle size distribution and the particle fragmentation behavior inside the slowly moving fixed bed
strongly influence the local counter current gas flow and therefore also the rate of heat transfer between
gas and coal, the rate of drying, devolatilisation and gasification.

COREX® is the first commercially operating smelting reduction process, based on coal instead of coke, as
alternative for industrial ironmaking route via the blast furnace. Besides coal the melter gasifier also
contains reduced iron ore and additives, which are not subject of this paper.

General multiphase models in commercial CFD codes are not directly applicable to the simulation of
moving reactive beds considering changes in particle size distribution. A customized approach based on a
combination of Eulerian and Lagrangian formulation is used to describe the flow of gas and solids as well
as the physical and chemical processes across the moving bed reactor.

The solids flow and the gas flow are represented by a set of Eulerian equations. So the flow of solids
respectively the flow of a granular material is treated as a continuum with appropriate material
properties. The balances for solids and gas flow are interconnected via source terms. The energy balance
of the solids flow and the models for fragmentation and devolatilization are implemented by means of a
Lagrangian formulation. The solids flow consists of a set of particle sizes including dust. This set of particle
sizes changes according to local process parameters which are: solids pressure, shear stress, rate of water
evaporation (coal drying), rate of devolatilization, rate of gasification and rate of temperature change. To
take this into account a fragmentation model has been developed which solves a conservation equation
for each particle size. The local source terms within these equations are connected to the above
mentioned local process parameters. Due to the fact that the considered moving bed consists of non-
uniformly sized particles the temperature of small particles will be different from the temperature of
larger particles. However, the temperature of the particles is important for the rate of drying,
devolatilization and gasification. Therefore an energy balance for each particle size is implemented within
the presented model.

In a first step the model has been used to study the impact of a changing particle size distribution on the
gas flow and heat transfer between gas und solids. The effect of fragmentation on the devolatilization
process has been simulated too. Next development steps are the integration of models for coal drying and
gasification as well as a gas phase reaction model.

1/13
 1. Introduction
The main components of a typical COREX®-plant for the production of liquid iron are shown in figure 1.
The two main components of the COREX® process are the reduction shaft and the melter-gasifier. In the
reduction shaft iron ore is reduced up to a reduction ratio of 80 to 95% [Carpenter, 2004] by the reduction
gas which consists mainly of CO and H2. The reduced iron ore coming from the bottom of the reduction
shaft as well as coal and additives are charged at the top of the melter gasifier.

The melter gasifier itself could be divided into four sections: the dome at the top, the slowly moving fixed
bed, the hearth, and the sump at the bottom. In the dome the remaining hydrocarbons evolving from the
packed bed are cracked to ensure that the reduction gas consists mainly of gases which are usable for iron
ore reduction. The fixed bed consists of a mixture of pre-reduced iron ore, coal and additives. In this
section the remaining iron ore is reduced completely and the drying and devolatilization processes of the
coal particles take place. Due to the heating, drying and devolatilization processes, shear stress in the bed
and solids pressure the coal particle size distribution is subject to change. Pure oxygen is injected at the
bottom of the fixed bed in order to gasify the remaining char, melt the iron and dissolve the additives. A
sufficiently coarse particle size distribution of the char matrix is important to allow counter-current flow
of gas and liquids. This is also valid for the hearth underneath to ensure smooth tapping of liquid iron and
slag at periodic intervals. The sump at the bottom serves as equilibration basin for liquid iron.

The purpose of this work is to present a possible strategy for modeling heating-, pyrolysis- and
fragmentation processes of the coal in slowly moving fixed beds similar to that of the COREX® melter-
gasifier.

Important to mention is that only the material coal has been considered in this work. The behavior of
other materials like iron, iron ore and additives as well as the interaction of these materials with the coal
are not subject of this work.

Figure 1: Simplified flow sheet of a typical COREX® plant for iron making (Siemens VAI).

2/13
 2. Description of the Model
Multiphase models in commercially available CFD Codes are not directly capable of simulating a reacting
slowly moving fixed bed with changes in particle size distribution.

A lot of work has been done on the simulation of slowly moving fixed beds in blast furnaces which are to
some extent similar to the slowly moving fixed bed of a COREX-Reactor. In most of the work concerning
solids flow in blast furnaces three (or more) coupled Eulerian sets of equations are used for determining
the flow characteristics of the solids (and fines), liquid and gas phase flow. The particle size distribution is
thereby considered in the form of bed voidage and a constant mean particle diameter (e.g.: Sauter´s
diameter). This is used as basis for the calculation of the relevant fluid-mechanical and chemical processes
occurring in the slowly moving bed (for example [2] or [11]).

The modeling approach adopted in this work is a customized combination of a Eulerian and a Lagrangian
formulation. The gas phase flow is represented by using mass, species, momentum and energy balances in
Eulerian notation. The solid phase (coal in this work) is depicted by means of a momentum balance in the
Eulerian frame of reference. The particle specific information like mass of the particles, particle size
distribution, temperature of the different particle sizes (particle energy balance), devolatilization and
material properties of the particles are calculated on the streamlines of the solids flow field.

The interaction between gas and solids is represented via source terms in the relevant balances.

The commercial CFD-software ANSYS FLUENT 12.0.16 has been used to implement the necessary
extensions to the code via user-defined functions (UDF).

In the following section the considered balances and implemented models are briefly discussed.

2.1 Gas phase

2.1.1 Governing Equations:

The governing fluid dynamic equations for the gas phase are the well known equations for describing a
flow of a reacting multi component mixture with diffusion in a Eulerian frame of reference:

• Overall Mass Balance:

∂
dt
( ε g ρ g ) + ∇ ⋅ ( ε g ρ g wg ) = S m


(1)

• Species transport:

∂
( ε g ρ g Yi ) + ∇ ⋅ ( ε g ρ g wg Yi ) = −∇ ⋅ Ji + Si



(2)
dt

• Momentum balance:

∂
dt
( ε g ρ g wg ) + ∇ ⋅ ( ε g ρ g wg wg ) = −∇pg + ∇τ + ε g ρ g g − Fg , s






(3)

The interphase momentum exchange term F is evaluated by using the classical equation of Ergun [4]:

3/13
 ε s2µg

ε s ρg




Fg , s = 150 ( − ) + wg − ws ( wg − ws )


2
wg ws 1.75 (4)
εg d d

Thereby d represents the so called “Sauter´s diameter“.

The gas phase energy balance in Eulerian notation is given by:

∂  
( ε g ρ g hg ) + ∇ ⋅ ( ε g ρ g wg hg ) = ∇ ⋅  λ∇T − ∑ hi J i  + Sh .


(5)
dt  i 

The source term S h in Eq. (5) consists of the convective heat transfer between solids and gas phase as
well as the Heats of Reactions of the devolatilzation process.

The gas phase species properties like viscosity, specific heat and thermal conductivity have been included
as polynomial functions of temperature and composition according to [6].

For the determination of the bulk porosity of the coal bed the model of Ouchiyama and Tanaka [12] has
been implemented.

2.2 Solid phase

As stated above the momentum balance is solved in an Eulerian reference frame whereas the particle
specific information is calculated on the streamlines of the solids flow. In the following paragraphs these
models are briefly presented.

2.2.1 Momentum Balance:

The momentum balance for the solids flow is given by Eq. (6)

∂
( ε s ρ s ws ) + ∇ ⋅ ( ε s ρ s ws ws ) = −∇ps + ∇τ s + ε s ρ s g + Fg ,s .






(6)
dt

In the present work the emphasis lies on the influence of fragmentation on the inter phase heat transfer
and the pyrolysis process. Thus the shear stresses are modeled by means of a Newtonian fluid with high
viscosity. If a more proper description of the solids flow is desired, a Bingham medium or the kinetic
theory of granular flow may be considered.

2.2.2 Particle Specific Information:

2.2.2.1 Heat Balance of a single particle:

The considered heat balance of a single Particle is given by Eq. (7):

= α corr Ap (Tg − Tp ) − ∆H Pyrolysis .

dH p
(7)
dt

The solid phase energy balance is solved for a representative particle of each particle size class. The
cumulative change in enthalpy of all particles in all size classes is determined by summing up the

4/13
individual enthalpy changes of each particle size class. The resulting value represents the source term in
the energy balance of the gas phase flow.

Two influences have an important effect on the particle temperature. Because of the heat conduction
inside the particle, especially the larger particles will have high Biot numbers. Thus the temperature
distribution inside the particle needs to be taken into account. The second aspect is that heat transfer
coefficients in packed beds are observed ([1], [13]) to be much lower than the heat transfer coefficient
predicted by correlations like the correlation of Gnielinski [5].

Therefore the heat transfer coefficient α calculated by the correlation of Gnielinski [5] is corrected for
these effects.

The “corrected” heat transfer coefficient α corr is evaluated by means of Eq. (8) [15]

1
α corr = t f . (8)
1 1 dp
+
α κ 2λs

The factor t f takes into account that the area available for heat transfer in the packed bed is smaller than

1 dp
the sum of the surface area of the single particles. The term is used to estimate the influence of
κ 2λs
heat conduction inside the particle. So the particle temperatures calculated by this model are mean
particle temperatures. In the current study values of t f = 0.1 [2] and κ = 3 [15] are used.

2.2.2.2 Pyrolysis Model:

The pyrolysis model is based on the model of [8] and [9]. This model is capable of predicting the
composition of Volatile matter, kinetics of evolution and heats of reaction of the coal pyrolysis process.
The original model [8] deploys a parallel first order reaction model with distributed activation energies
(DAE) for determining the evolution of the different volatile species.

Due to the fact that the numerical integration required by the DAE approach is computationally very
expensive since frequent numerical integration is necessary for all size classes in every time step over the
whole range of the activation energy distribution, a simplified model has been realized. The evolution of
the different volatile matter species is described by a set of parallel first order reactions given by Eq. (9):

dmi  E 
= k0 exp  − i  mi . (9)
dt  RT 

The considered species of volatile matter i are CH4, CO, CO2, Tar, H2, H2O, NH3 and H2S. Only the primary
coal pyrolysis has been covered so the tar cracking reactions in the gas phase have not yet been included.

2.2.2.3 Fragmentation model:

The fragmentation of coal particles in slowly moving fixed beds is a governing factor for the gas flow and
subsequently for the heat transfer and for the chemical processes like drying, devolatilization and
gasification. The model is based on previous models for pulverized coal combustion of [14] and [10]. The

5/13
six major factors which are supposed to be the driving forces for the fragmentation of coal particles inside
the slowly moving fixed bed of a COREX® melter gasifier are:

• solids pressure (k=0),

• shear stress (k=1),
• rate of heating (k=2),
• rate of drying (k=3),
• rate of devolatilization (k=4) and
• rate of gasification (k=5).

The fragmentation model is formulated in terms of the number of particles in a certain size class following
a distinct stream line.

 dN 0   −C C +  d1  
3 3
d 
  ∑ b0,1Sk ,1 K k ∑k 0,n k ,n k 
k k
 dt   0 1 ... ...  n  b S K
d
 0 k  d0  N 
    0
dN
 1  3
d   N
 dt   0 −C1 − ∑ b1,1 k
Sk ,1 K k ... ...  n  ∑ b1,kn Sk ,n K k   1 
 = k  d1  k  ...  .
 
(10)
 ...     ... 
 ...   ... 0 ... ... ...
 
  ... ... ... ... ...   Nn 
 dN n  
 dt   0 0 0 0 −Cn − ∑ bn ,n Sk ,n K k 
k

 k 

The rate of change of the particle size i due to fragmentation is generally assumed to be proportional to
the number of particles in each size class at the time t on a certain position on a distinct stream line. The
numerical distribution of a particle size class i into smaller particle size classes j due to the process k is
k
given by the elements of the fragmentation matrix bi , j . The term S k , j is the fragmentation rate constant

for each process k and each size class i. The variable K k contains the driving force of the process k . For
example K k for the process “heating rate” is the rate of temperature change of the particle in K/s. The
variable C j accounts for the reduction of the particle size due to gasification. The fragmentation matrices

bik, j and the fragmentation rate constants Sk , j are determined by suitable experiments.

2.2.2.4 Material Properties of the coal particles:

• Specific Heat of coal:

Because of the heating, drying, devolatilization and gasification processes the temperature and the
composition of the coal particles are changing during the residence time in the reactor. As a consequence
the specific heat of the coal particles is not constant during these processes. To take these effects into
account the specific heat model of [9] has been implemented.

The reason for the selection of this model is that the specific heat could be calculated on the basis of the
actual composition and temperature of the particle. The specific heat model is assumed to be valid for the
whole temperature range considered.

6/13
 • Thermal conductivity:

Both, the changes in temperature and composition of the particle have a significant effect on the thermal
conductivity of the coal particles. Therefore the model of [2] has been chosen to determine the thermal
conductivity of coal, char and intermediate products. The thermal conductivity is estimated in terms of
the “true density” of the particle and the particle temperature. The term “true denstity” in this case
means the density of the daf-coal excluding the pore volumes.

3. Simulation
In order to test the models presented above four steady state simulations of a slowly moving coal bed
reactor have been performed. In a first step the influence of fragmentation on gas flow and heat transfer
between coal and gas has been studied. In a second step the effects of fragmentation on the
devolatilization process has been simulated.

3.1 Geometry

The geometry which has been used for conducting the simulations is depicted in Figure 1. The cylinder has
a radius of R=4,5m and a height of H=9m. This geometry is to some extent similar to the fixed bed region
of a COREX® melter gasifier.

Figure 1: Geometry of the reactor.

3.2 Boundary Conditions

The coal (daf) mass flow entering the cylinder on the top is set to 50.5 kg/s with an inlet temperature of
300 K. The analysis of the coal used for both simulations is given in Table 1.

C H O N S Volatile Matter
89.4 4.9 3.7 1.2 0.8 25
Table 1: Analysis of the coal (wt%, daf)

7/13
The particle size distribution of the coal at the inlet is summarized in Table 2.

dp [mm] 0.0001 0.25 0.5 5 8 16 20 25 31.5 50

X [%] 0.1 0.5 1 10 20 30 15 11.5 7 4.9
Table 2: Particle size distribution at the top of the reactor (coal inlet).

For the simulation runs with coal devolatilization the kinetic parameters specified in Table 3 have been
set.

Species CH4 C2H6 CO CO2 Tar H2 H2O NH3 H2S

Ea [MJ/kmol] 280.5 238.3 267.3 253.7 230.3 332.6 230.3 279.1 286.3
k0 [1/s] 1.x1013 1.x1013 1.x1013 1.x1013 1.x1013 1.x1013 1.x1013 1.x1013 1.x1013

Table 3: Kinetic Parameters of volatile species release.

It should be noted that for the purpose of demonstration the frequency factors are assumed to be equal
and constant for all species of volatile matter. The Activation Energies of the different volatile species
were set to the mean values of the activation energy distribution used by [8]. Of course for the application
of the model for design or evaluation purposes the parameter setting must be further detailed. For the
sake of presenting the basic characteristics of the model the current approach is taken to be sufficient.

The inlet mass flow rate of gas (100% CO) at the bottom (gas inlet) of the cylinder is assumed to be 65
kg/s. The temperature of the gas at the gas inlet is set to 1500 K.

4. Results

4.1 Heat Transfer without Pyrolysis

The first two simulation runs have been conducted without considering the pyrolysis process in order to
study the influence of the fragmentation process on the temperature distribution inside the reactor.

The distribution of the bed voidage as fraction of volume gas per m³ reactor volume is depicted in Figure
Figure 2. Due to the increasing solids pressure and the heating rate the coal particles are subject to
fragmentation on the way through the reactor. The bed voidage decreases from a value of 0.407 to a
value of 0.38. The reason for the decrease in voidage is that larger particles are breaking into smaller
ones, which occupy more void space. As a consequence the bulk density of the coal bed increases.

The evolution of the particle size distribution in Figure 3 clearly shows the increase of the volume fraction
of the smaller particle size classes. The volume fraction of the larger particles with a diameter of 20 mm or
bigger decreases at a height of approximately 8 m while the smaller particle size classes with a mean
diameter of or below 8mm are increasing. The particle size class with a mean diameter of 16 mm is
gaining volume fraction up to a height of approximately 3 m due to the fragmentation of the upper size
classes. Below a height of 3 m the fragmentation to smaller size classes gets dominant over the incoming
particles of the upper size classes and the volume fraction of the size class 16 mm is decreasing.

8/13
 Figure 2: Distribution of bed voidage [-] inside the reactor.

Figure 3: Particle size distribution inside the reactor

The temperature distribution of the gas and the bulk phase are shown in Figure 4. The gas temperature
with considered particle fragmentation is well below the gas temperature without particle fragmentation.
The bulk temperature of the coal in case of particle fragmentation reaches slightly higher values at the gas
inlet. This is a consequence of the increase in volume fraction of the smaller size classes. Hence, the
available surface area for heat transfer is increasing too. So the heat transfer between solids and gas is
enhanced.

9/13
 Figure 4: Comparison of Gas and Bulk Temperature distributions with and without considered Particle Fragmentation

4.2 Heat Transfer with Pyrolysis and Fragmentation

The third and fourth simulation case illustrate the influence of particle fragmentation on the pyrolysis
behavior of the coal particles. The bed voidage at the bottom of the reactor in case of fragmentation is
approximately 0.395. Compared to the simulation without pyrolysis this value is slightly higher. The
explanation for this effect is that as a consequence of the devolatilization process the gas mass flow is
increasing from the bottom to the top of the reactor. As a result the lifting force acting on the slowly
moving coal bed is increased. Thus the solids pressure and subsequently the amount of fragmenting
particles are decreasing.

The gas and bulk temperatures with and without fragmentation behave very similar compared to the
simulation runs without pyrolysis. The effect of bulk temperature increase at the bottom of the reactor in
case of particle fragmentation is somewhat smaller compared to the previous run due to the higher bed
voidage at the bottom.

The distributions of the mean volatile fraction of the coal bed for the cases with and without particle
fragmentation are shown in Figure 5. Due to the higher bulk temperature of the slowly moving coal bed in
case of particle fragmentation the pyrolysis process is slightly further progressed. Despite the fact that in
the presented simulations the decrease in solids porosity from the top of the reactor to the bottom is only
2.9 % the difference in the mean volatile fraction at the bottom of the reactor is as much as 43 %.

Hence, for accurate simulation results the particle fragmentation has to be considered in the simulations.

10/13
 Figure 5: Comparison of mean volatile fractions with and without particle fragmentation.

5. Conclusion
The particle size distribution is a governing factor for heat transfer and the chemical processes inside
slowly moving bed gasifiers. A model has been presented which addresses the process of particle
fragmentation by deploying a kinetic approach for the relation between the major forces acting on the
coal particle and the change in particle size distribution.The influence of fragmentation on heat transfer
and the devolatilization process has been determined.
The results of these simulations clearly indicate that neglecting the fragmentation of the coal particles in
gasifiers leads to inaccurate results especially in cases significant changes in the particle size distribution.

The next development steps will include implementation of models for coal drying and gasification with
respect to particle fragmentation as well as reaction mechanisms for the homogenous gas phase reactions
of the different species.

Acknowledgement:

The Competence Center for Excellent Technologies in Advanced Metallurgical and Environmental Process
Development is part of the COMET (Competence Center for Excellent Technologies) program which is funded
by the Bundesministerium für Verkehr, Innovation und Technologie, Bundesministerium für Wirtschaft, Familie
und Jugend, Land Oberösterreich, Land Steiermark, steirischen Wirtschaftsförderungsges.m.b.H., Land Tirol and
Tiroler Zukunftsstiftung.

Notation:

Ap Particle surface area, [m2]

bk Progeny matrix for process k, [-]
Cj Gasification rate of particle size j [1/s]
dp Particle diameter [m]
Ei Activation energy [J/mol]


Fg , s Interphase momentum exchange [N/m3]

11/13
 Gravity [m/s2]


g
h Specific enthalpy [J/kg]
Hp Particle enthalpy [J]
∆H Pyrolysis Enthalpy [J]


Ji Diffusive Flux of species i [kg/m2s]
k0 Frequency factor [1/s]
k Process responsible for fragmentation
K Quantity describing a process responsible for fragmentation
mi Volatile mass of species i inside the particle [kg]
n Number of size classes [-]
Nj Number of particles in size class j [-]
pg Pressure of the gas phase [N/m2]
pg Pressure of the solids phase [N/m2]
R Universal gas constant [J/kmolK]
Sm Source term in the gas mass balance [kg/m3s ]
Si Source term in the species balance [kg/m3s ]
Sh Source term in the gas energy balance [W/m3 ]
Sk , j Rate constant of fragmentation of size class j due to process k
tf Factor for the correction of the heat transfer coefficient [-]
T Temperature [K]
Tg Gas temperature [K]
Tp Particle temperature [K]
wg Velocity of the gas mixture [m/s]
ws Velocity of the solids [m/s]
Yi Species mass fraction [-]

Greek letters:

α Heat transfer coefficient [W/m2K]

α corr Corrected heat transfer coefficient [W/m2K]
εg Volume fraction of gas [-]
εs Volume fraction of solid [-]
ρg Gas density [kg/m3]
ρs Solid density [kg/m3]
κ Correction factor for thermal conductivity inside the particle [-]
λ Thermal conductivity [W/mK]
λp Thermal conductivity of the particle [W/mK]
τg Gas phase shear stress [N/m2]
τs Solid phase shear stress [N/m2]
µg Average viscosity of the gas phase mixture [Pas]

12/13
References:

[1] Atkinson, B. and Merrick, D.: Mathematical models of the thermal decomposition of coal: Heat
transfer and temperature profiles in a coke oven charge, Fuel 1983, Vol. 62, 553

[2] Austin et al.: Mathematical model of four phase motion and heat transfer in the blast furnace, ISIJ
International, Vol. 37 (1997), No. 5, pp. 458-467

[3] Carpenter, A.: Use of coal in direct ironmaking processes, IEA, London 2004

[4] Ergun, S.: Fluid Flow through Packed Columns. Chem. Eng. Prog., 48(2):89–94, 1952.

[5] Gnielinski V.: Wärmeübertragung Partikel-Fluid in durchströmten Haufenwerken, VDI-Wärmeatlas

Kap.Gj1, Springer 2006

[6] Kleiber et al.: Stoffwerte von sonstigen chemisch einheitlichen Flüssigkeiten und Gasen, VDI-
Wärmeatlas Kap.Gj1, Springer 2006

[7] Lee et al.: A development of computer model for simulating the transport phenomena in COREX
melter-gasifier, ISIJ International, Vol. 39 (1999), No. 4 pp. 319-328

[8] Merrick, D.: Mathematical models of the thermal decomposition of coal: 1. The evolution of volatile
matter, Fuel 1983, Vol. 62, 534

[9] Merrick, D.: Mathematical models of the thermal decomposition of coal: 2. Specific heats and heats
of reaction, Fuel 1983, Vol. 62, 540

[10] Mitchell, R.: An intrinsic kinetics-based,particle population balance model for char oxidation during
pulverized coal combustion, Proc. of the Combustion Institute, Vol. 28, pp. 2261-2270, 2000

[11] Nogami, H. et al.: Multi-demensional transient mathematicla simulator of blast furnace process
based on multi-fluid and kinetic theories, Computers and Chemical Engineering 29 (2005) 2438-
2448

[12] Ouchiyama, N. and Tanaka, T.: Porosity Estimation from Particle Size distribution, Ind. Eng.
Chem. Fundam. 1986, 25, 125-129

[13] Pedrag et al.: An improved model for fixed bed coal combustion and gasification, Fuel 1995,
Vol. 74 No. 4 pp. 582-594

[14] Shah KV et al. A kinetic-empirical model for particle size distribution evolution during pulverised
fuel combustion. Fuel 2010

[15] Zeisel. H: Mathematische Modellierung und numerische Simulation der Vorgänge im Hochofen,
Trauner, Linz,1995

13/13