CHEMISTRY & CHEMICAL TECHNOLOGY

Chem. Chem. Technol., 2019, Chemical

Vol. 13, No. 2, pp. 205–211 Technology

MATHEMATICAL MODELING AND SIMULATION

OF DIRECT REDUCTION OF IRON ORE IN A MOVING BED
REACTOR BY THE SINGLE PARTICLE MODEL
Vahab Ghalandari1, *, Hassan Hashemipour Rafsanjani1

https://doi.org/10.23939/chcht13.02.205
Abstract.1 In this work, a mathematical model is plants. Both models satisfactorily fitted with the
developed for simulating the behavior of a counter-current experimental data, but it should be noted these models do
moving bed reactor, in which the reduction of porous iron not consider the variation of solid porosity. Melchiori and
ore pellets to sponge iron is simulated. Simultaneous mass Cunu [6], Ahn and Choi [7] carried out the numerical
and energy balances within both the solid particles and the comparison between the shrinking core model and the
reactor, will lead to a set of coupled ordinary differential grain model. They found that a grain model has a better
equations. The iron ore reduction kinetics was modeled performance than a shrinking core model to predict
with a single particle model. The model was able to experimental and industrial results.
satisfactorily reproduce the data of Gilmore Steel Moreover, some researchers assumed the direct
Corporation (USA). Eventually, the effects of reducing reduction reactor model including only one reacting gas.
gas parameters and pellet characteristics such as porosity Most of them have used pure hydrogen [8, 9], pure carbon
on the reduction extent have been investigated. monoxide [10] or a mixture of hydrogen and carbon
monoxide as a reducing gas [11, 12], but in fact the
Keywords: simulation, direct reduction, moving bed reducing gas is a mixture of hydrogen, carbon monoxide,
reactor, single particle model, porosity. carbon dioxide, steam and methane as a synthesis gas.
In this work, the moving bed direct reduction
1. Introduction reactor has been modeled by a single particle concept.
This model considers the variation of solid porosity. The
The direct reduction is a significant process to model equations were derived from mass and heat balance
produce metallic iron from iron ore using reducing gases. of the solid particle and bed of the reactor. The modeling
In the moving bed solid-gas reactor the gas flow upward results have been compared with Gilmore planet data [3]
and the solids flow downward by gravity in a counter to indicate the model validation. This simulation is used to
current. Mass and heat transfer between the phases, as study the effect of operating parameters on the reactor
well as chemical reaction occur simultaneously, which performance.
reduce the hematite pellets. In the last few years, the direct
reduction of iron ore process has been investigated by 2. Mathematical Model
using some mathematical models.
Szekely and El-Tawil [1], Towhidi and Szekely [2],
Parisi and Laborde [3] used the unreacted shrinking core 2.1. Chemical Reaction
model to model hematite pellets. This model was in The reaction of hematite to iron is carried out
accordance with the experimental data, although it should through the series of reactions: hematite → magnetite →
be noted that the variation of solid porosity as a main wustite → iron. The overall reaction which is considered
parameter is not considered in the model. Recently, Arabi in the several simulation works [3-5] is a direct reduction
and Hashemipour [4] modeled the moving bed direct of hematite to iron with the reducing gases hydrogen and
reduction reactor by the three shrinking core model, Nouri carbon monoxide as R1 and R2:
et al. [5] modeled it by the grain model for industrial
R1: Fe2O3 + 3H2→ 2Fe + 3H2O (1)
1
R2: Fe2O3+ 3CO → 2Fe + 3CO2 (2)
Chemical Engineering Department, Shahid Bahonar University of The reaction of hematite with other reducing gases
Kerman, Iran
*
[email protected] such as CO2, H2O and CH4 is not considered in this study
 Ghalandari V., Rafsanjani H., 2019 because of low content and negligible reactivity. The
206 Vahab Ghalandari et al.

chemical reaction rates of the reactions R1 and R2 are changes within the particle [13]. In this simulation, a
introduced as elementary with the following reaction rate second order functionality of a surface area versus particle
equations [3]: porosity is considered. This relation is derived using the
RR1 = k H 2 C 'H 2 ' mathematical points (εp = 0, Sg = 0), (εp = 1, Sg = 0) and
k H 2 = 0.114exp( −179.13/ T ) , cm/s (3) ( ε p = ε 0p , S g = S g0 ) as the following equation:

RR 2 = kCO C 'CO ' (ε p − ε p 2 )

S g = S g0 (11)
kCO = 0.283exp( −342.43/ T ) , cm/s (4) (ε 0p − ε 0p 2 )
The particle density is also variable as the reaction
2.2. Particle Modeling progresses and the particle porosity change. This
functionality is defined as ρp = ρt(1–εp). The value of ρt is
In this work, a modified volume reaction model a true density of the solid. The pore diffusion of gas
named a single particle model is used as a new approach component within the solid particle (Dei) is a function of
for consideration of pores structural changes within the the particle porosity. This relation can be evaluated using
solid phase. Molar volumes of hematite and iron are expression [14]:
different and this causes structural changes of the solid
during the reaction. In this model, the reaction of spherical Dei = Diε p 2 (12)
solid particle with the reactant gases is considered The gas diffusion coefficients as a function of the
isothermal. In addition, it is assumed that the particle solid temperature can be evaluated using expressions [3]:
diameter remains constant during the reactions. The mole DH2 = 1.467·10 -6·Ts1.75, cm2/s (13)
balances of H2 and CO components within the solid DCO = 1.276·10 -7·Ts1.75, cm2/s (14)
particle base on the model are as follows: The above mathematical modeling of the particle
1 ∂ 2 ∂C 'H 2  leads to a set of coupled nonlinear ordinary differential
⋅  r DeH 2  − ( RR1 S g ) = 0 (5) equations.
r ∂r 
2
∂r 
1 ∂ 2 ∂C 'CO  2.3. Reactor Modeling
⋅  r DeCO − ( RR 2 S g ) = 0
∂r 
(6)
r ∂r 
2
In this study a non-catalytic moving bed reactor is
The boundary conditions for the above equations modeled and simulated in the non-isothermal condition. In
are as follows: the moving bed reactor, the gas flows upward and
dC 'H 2 dC 'CO countercurrent to the downward flow of solids. Some of
At r = 0 = =0 (7) the general assumptions such as steady state operation and
dr dr
plug flow of the gas and solid streams in the reactor are
At r = Rp C 'H 2 = C H 2 , C 'CO = CCO (8)
considered to derive the reactor models. The mass balance
where CH2 and CCO are the reactant gases concentration in of gaseous reactants within the reactor can be written as
the bulk phase (in the reactor). Mole balance on the follows:
hematite within the solid particle gives the following dCH 2 1 − ε b
equation: ug + J H2 = 0 (15)
dz Vp
dCC
= RC = 3RR1 + 3RR 2 (9) dCCO 1 − ε b
dt ug + J CO = 0 (16)
where CC is the hematite concentration in the particle dz Vp
Vε ρ The mass balance on the reactant solid can be
which is CC = p p p .
Mw written as follows:
The Eq. (9) can be rearranged as: dC 1 − ε b
us C + ( J H 2 + J CO ) = 0 (17)
d ε p 3M w dz 3Vp
= ( R 1 + RR 2 ) (10)
dt V p ρ p R JH2 and JCO are described as diffusion rates of the
gas components into the solid particle. These factors can
With the initial condition ε p = ε 0
p in Eq. (10), the be evaluated using expressions:
time variable is defined as a distance of solid travel from
dC 'H 2
its inlet (t = Z/us). Therefore the initial condition of solid J H 2 = − DeH 2 ε p Ap ( r = rp ) (18)
porosity is the inlet solid porosity of the reactor. dr
In the single particle model, the solid particle dC 'CO
J CO = − DeCOε p Ap ( r = rp ) (19)
properties (εp, ρp and Sg) are variable because of structural dr
 Mathematical Modeling and Simulation of Direct Reduction of Iron Ore in a Moving Bed… 207

The gas component concentration within the solid

3. Results and Discussion
particle is predicted using the single particle model. The
energy balance for gas and solid phases can be stated as
follows: 3.1. Simulation Validation
dT A In this section, a comparison between the results of
Gmg C pg g + p (1 − ε b ) h(Tg − Ts ) = 0 (20)
dz V p the simulation model with the values of existing gas
composition and metallization of an industrial plant is
dTs (1 − ε b )
GmsC ps + [ Ap h(Tg − Ts ) + performed. The model is validated using experimental
dz Vp data from Gilmore plant [3]. The operating conditions of
+∆H H 2 J H 2 + ∆H CO J CO = 0 (21) the plant reactor and the inlet gas composition are
presented in Table 2. The reactor simulation was run with
Boundary conditions for the above equations are as
these parameter and the results are shown in Fig. 1.
follows:
The outlet gas composition (Fig. 1, length 0 cm)
At z = 0 CH 2 = CHin2 , CCO = CCO
in
, CC = CCin (22) and outlet solid conversion (Fig. 1, length 975 cm)
At z = 0 Tg = Tgin , Ts = Tsin (23) predicted from the simulation are compared with the
experimental data. The comparison is presented in Table
The heat capacity of the gas and solid components
3. Results of the simulation model are in good agreement
are considered variable as the temperature variation. The
with the plant data with an average error of 3.2 %.
heat transfer coefficient and heat capacity of the gas and
In the next sections, the effects of some operating
solid components used in the simulation are shown in
conditions of inlet streams on the solid conversion are
Table 1.
investigated. To study the effect of these parameters, the
amounts are selected based on the Gilmore operating
Table 1
conditions.
The heat transfer coefficient
and heat capacity of the components 3.2. Temperature Profiles of Gas
Symbols, units Values and Solid within the Reactor
C pFe 2O 3 , cal/mol·K 24.72+0.01604T–423400/T2 Basical temperature profile along the reactor
C pH 2 , cal/mol·K 6.62+0.00081T (especially solid temperature) is the main parameter in the
C pCO , cal/mol·K 5.12+0.00333T reactor operation because it affects the reduction progress
2 directly. Fig. 2 illustrates the temperature profiles of gas
h, cal/cm ·s·K 1·10-4
and solid phases along the reactor. The temperature of gas
stream decreases from bottom (gas inlet) toward the top of
The solid conversion can be evaluated using the
the reactor, while the temperature profile of solid stream
expression:
has an ascending trend with a higher slope. This is
C
X = 1 − Cin (24) because of heat transfer from the gas to the solid in
CC addition to the endothermic reactions within the solid
particles.
2.4. The Numerical Solution The effects of temperature of the gas and solid inlet
The nonlinear ordinary differential equations of the on the solid conversion along the reactor are shown in
moving bed reactor and solid particle are solved Figs. 3 and 4. Fig. 3 illustrates that increasing gas inlet
simultaneously. These equations describe variation of the temperature improves the solid conversion but there is a
gas component in the reactor and solid particle. The gas negligible effect of the higher gas inlet temperature. This
concentrations in the reactor are related to the gas can be for higher diffusion limitation at the higher
concentrations profile within the solid particle with the temperature. Therefore the optimum temperature of the
pore changes. These nonlinear differential equations are gas inlet is about 1000 K. Fig. 4 illustrates that increasing
solved using numerical algorithm programmed with solid inlet temperature has less effect on the solid
Matlab software. The software code is programmed using conversion improvement (in the temperature range of
ODE solution algorithm. 325–525 K). The reason is that this temperature range is
too low to affect the chemical reaction.
208 Vahab Ghalandari et al.

Table 2
Operating conditions of Gilmore plant [3]
Parameters Values
Gas flow rate, Nm3/h 53863
Inlet gas composition, wt % (at z = L)
H2 52.58
CO 29.97
H2O 4.65
CO2 4.80
CH4+N2 8.10
Pressure, kPa 141.8
Gas inlet temperature, K 1230
Solid flow rate (Fe), t/h 26.40
Iron ore particle density (ρt), g/cm3 4.70
Sponge iron particle density (ρp), g/cm3 3.20
Particle radius (Rp), cm 0.55
Inlet particle porosity (εp) 0.30
Solid inlet temperature, K 320
Reaction zone length (L), cm 975
Reactor diameter (R), cm 426
Bed porosity (εb) 0.55
Table 3
Comparison of Gilmore data with the developed model prediction
Gilmore data, wt % Developed model data, wt % Error, %
Outlet gas composition
H2 37.0 37.2 0.53
CO 18.9 18.7 1.06
H2O 21.2 20.5 3.41
CO2 14.3 15.5 7.74
Inert 8.6 8.1 6.17
Solid conversion 93.0 92.7 0.32

3.3. Variation of Gas and Solid Flow Rates 3.4. Feed Gas Composition
Fig. 5 illustrates the effect of inlet reducing gas Generally the inlet gas stream contains H2, CO,
flow rate on the conversion of iron ore particles in the CO2, H2O, CH4 and N2 which is a product of the syngas
range of ±10 %. When the gaseous flow rate is decreased, production unit where H2 and CO components are more
a lower solid conversion is obtained. This is due to
effective than the others in the reduction of hematite. In
decreasing of reducing gaseous content in the reactor as
the gas flow rate is decreased. In this condition, the this study these two components are considered to reduce
overall reaction rate decreases and therefore the solid the solid. Therefore the feed gas composition named as a
conversion gets worse. molar ratio of H2/CO is a main parameter in the prediction
Fig. 6 illustrates the effect of variation of the inlet of solid reduction but it is highly restricted. This is due to
solid flow rate in the range of ±15 % on the conversion of decreasing the low mass transfer of the reducing gas into
iron ore particle. As the solid flow rate is decreased, the the solid.
conversion of outlet particles is increased because the Fig. 7 illustrates the effect of this molar ratio on the
residence time of iron ore particles in the reactor is solid conversion. The solid conversion is decreased with
increased. Since the reaction is happened in the solid the decrease in CO concentration (with increasing H2/CO
phase, so the solid particles have more time to react with ratio) in the feed gas. Since the rate constant for CO is
the gaseous reactants.
In other words, it can predict that in this system the higher than that of hydrogen at the same temperature,
gas content should be an excess and in this condition the therefore the higher CO content is preferred from the
effect of this parameter on the solid conversion is low, reaction viewpoint. But in the real condition, removing of
while the solid is consumed along the reactor considerably H2 from the gas inlet causes the coke formation on the
and so a solid flow rate is more important than a gas flow sponge iron due to decomposition of CO component and
rate on the reactor performance. therefore this parameter cannot vary in a wide range.
 Mathematical Modeling and Simulation of Direct Reduction of Iron Ore in a Moving Bed… 209

Fig. 1. Profile of gas composition along the reactor Fig. 2. Variation of solid and gas temperature along the reactor

Fig. 3. Effect of gas inlet temperature Fig. 4. Effect of solid inlet temperature
on the conversion of iron ore on the conversion of iron ore

Fig. 5. Effect of gas flow rate on the conversion Fig. 6. Effect of solid flow rate
of iron ore (Q = 53863 Nm3/h) on the conversion of iron ore

Fig. 7. Effect of feed gas composition Fig. 8. Effect of particle porosity

on the conversion of iron ore on the solid conversion
210 Vahab Ghalandari et al.

Fig. 9. Gas composition (CO) profile inside Fig. 10. Gas composition (H2) profile inside
the particle in some different reactor length the particle in some different reactor length

3.5. Effect of Inlet Particle Porosity of a reactor length on the gas concentration profile within
the solid particle. Because of counter current flow of gas
The particle porosity increases as the reduction and solid stream in the reactor, the effect of length on the
progress due to the difference in molar volume of gas profile within the particle is not vigorous. This weak
hematite (as reactant) and iron (as a product). This effect is observed in both figures.
variation of porosity is very important because the
reaction surface and pore diffusivity, as well, are related to
porosity directly. The initial porosity should be in the pre- 4. Conclusions
determined amount and it can not vary in the wide range.
The effect of inlet solid porosity on the conversion of In this work, reduction of iron ores in the moving
solid is shown in Fig. 8. It can be seen that when the inlet bed reactor was simulated by the single particle model. In
particle porosity is increased, an increasing in solid order to simulate the system, the equations of mass and
conversion is observed. This is because of higher surface energy balances into the reactor and within the particle are
area and lower gas diffusion resistance in the higher derived for the solid and gases components. Then the
porosity of solid. This condition causes improvement of
equations were solved in the countercurrent gas-solid flow.
overall reaction rate.
The results of the simulation were in a good agreement with
3.6. Gas Composition Profile inside the experimental data of Gilmore plant data (average error
3.2 %). The effects of reactor and solid particle parameters
the Particle were investigated by the simulation. The results of the
Figs. 9 and illustrate gases compositions (CO and simulation show that the inlet reducing gas parameters
H2) profile within the solid particle in several points in the (composition, temperature and flow rate) and the porosity
reactor length. It can be seen that the slope of composition of the inlet solid are the main and effective parameters on
profile for both gases is decreased from an external the solid conversion. In addition, the optimum amount of
surface to the center of the particle. These slopes are the reactor length was 900 cm and the optimum gas inlet
related strongly to the gases compositions in the bulk flow temperature was 1000 K. Comparison of the gas
within the reactor. In addition, solid porosity affects the concentration profile within the solid particles in different
gas diffusivity and thus the gas concentration profile reactor length showed that the gas diffusion within the solid
within the solid particle. These factors explain the effect particles had the main effect on the reaction progress.

Nomenclature
Ap outer surface area of the L total reactor length X solid conversion C molar concentration within
particle the gas phase
Cpg, heat capacity for gas and np number of particle per ∆H heat of reaction Cin initial molar concentration
Cps solid phases unit volume of the bed
D molecular diffusivity Q gas flow-rate εp particle porosity Sg specific surface
De effective diffusivity of r particle radius ε 0p initial particle S g0 initial specific surface
gas porosity
Gmg, gas and solid molar Tg, Ts gas and solid phase εb reactor bed ρp initial particle density
Gms flows temperatures porosity
 Mathematical Modeling and Simulation of Direct Reduction of Iron Ore in a Moving Bed… 211

h heat transfer t time Z reactor length rp total particle

coefficient radius
k reaction rate ug, us gas and solid velocity C’ molar concentration within the solid Mw molecular
constant particle weight
Vp particle volume

[14] Dogu T.: Chem. Eng. J., 1981, 21, 213.

References https://doi.org/10.1016/0300-9467(81)80005-6
[1] Szekely J., El-Tawil Y.: Metall. Mater. Trans. B, 1976, 7, 490. Received: January 16, 2018 / Revised: February 08, 2018 /
https://doi.org/10.1007/BF02652723 Accepted: June 05, 2018
[2] Towhidi N., Szekely J.: Ironmaking Steelmaking, 1981, 6, 237.
[3] Parisi D., Laborde M.: Chem. Eng. J., 2004, 104, 35. МАТЕМАТИЧНЕ МОДЕЛЮВАННЯ
https://doi.org/10.1016/j.cej.2004.08.001
ТА СИМУЛЯЦІЯ ПРЯМОГО ВІДНОВЛЕННЯ
[4] Arabi S., Hashemipour Rafsanjani H.: Chem. Product Process
Model., 2008, 3, 40. https://doi.org/10.2202/1934-2659.1230 ЗАЛІЗНОЇ РУДИ В РЕАКТОРІ З РУХОМИМ ШАРІ
[5] Nouri S., Ale Ebrahim H., Jamshidi E.: Chem. Eng. J., 2011, НА МОДЕЛІ ОКРЕМОЇ ЧАСТИНКИ
166, 704. https://doi.org/10.1016/j.cej.2010.11.025
[6] Melchiori T., Canu P.: Ind. Eng. Chem. Res, 2014, 53, 8980. Анотація. Розроблено математичну модель для
https://doi.org/10.1021/ie403030g симуляції відновлення пористих залізорудних окатишів до
[7] Ahn H., Choi S.: Comp. Chem. Eng., 2017, 97, 13. губчастого заліза в протиточному реакторі з рухомим шаром.
https://doi.org/10.1016/j.compchemeng.2016.11.005 На основі масового та теплового балансів як в твердих
[8] McKewan W.: J. Metals, 1964, 16, 781. частинках, так і реакторах, виведено звичайні диференціальні
[9] Usui T., Ohmi M., Yamamura E.: ISIJ Int, 1990, 30, 347. рівняння. За моделлю окремої частинки змодельована кінетика
https://doi.org/10.2355/isijinternational.30.347 відновлення залізної руди. Показано, що розроблена модель
[10] Tien R., Turkdogan E.: Metall. Trans, 1972, 3, 2039. задовільно відтворює дані корпорації Gilmore Steel (США).
https://doi.org/10.1007/BF02643212 Досліджено вплив параметрів відновлювального газу та
[11] Kam E., Hughes R.: Trans. IChmE, 1981, 59, 196. характеристик окатишів, таких як пористість, на ступінь
[12] Negri E., Alfano O., Chiovetta M.: Ind. Eng. Chem. Res, 1991, відновлення.
30, 474. https://doi.org/10.1021/ie00051a007
[13] Lee S., Angus J., Edwards R., Gardner N.: AIChE J. 1984, 30, Ключові слова: симуляція, пряме відновлення, реактор з
583. https://doi.org/10.1002/aic.690300409 рухомим шаром, модель окремої частинки, пористість.