Korean J. Chem. Eng.

, 28(12), 2242-2249 (2011)

DOI: 10.1007/s11814-011-0122-5

INVITED REVIEW PAPER

Modeling and simulations of a reformer used in direct reduction of iron
Abdelhamid Ajbar , Khalid Alhumaizi, and Mustafa Soliman
†

Department of Chemical Engineering, King Saud University, P. O. Box 800, Riyadh 11421, Saudi Arabia
( )
Received 26 February 2011 • accepted 5 May 2011

Abstract−This paper presents a detailed modeling and simulations of a reformer unit used in the direct reduction of
iron (DRI) process. A one-dimensional heterogeneous model for the catalyst tubes which takes into account the intraparti-
cle mass transfer resistance was developed, while the furnace was modeled with bottom firing configuration. Validation
against data from a local iron/steel plant showed satisfactory results. The performance variables of the unit were the
process gas temperature, wall temperature and conversions of hydrogen, methane and carbon dioxide. The profiles of
these output variables along the distance were calculated. The effect of operating parameters such as inlet temperature,
natural gas flow rate and gas composition was also determined.
Key words: Direct Reduction of Iron, Reformer, Modeling, Simulation, Performance

INTRODUCTION [20], and Selcuk et al. [22], for instance, used the flux method for
radiative transfer [18] to develop models for reformer furnaces. One
The direct reduction process has gained growing importance in notable contribution is the work of Farhadi et al. [2] who carried
the last decades as a source of metallic units for electric arc furnaces out a detailed one-dimensional heterogeneous model for the Midrex
used for the production of steel. Midrex technology is the most im- reformer. In a recent study, Shayegan et al. [5] developed a two-
portant DRI process and consists of around 58% of the world total dimensional heterogeneous model for the reformer. The authors
DRI production [1]. In a Midrex process, natural gas is used to manu- presented arguments for the need to account for temperature gradi-
facture the reducing gas in a reforming process. The Midrex reformer ents in radial dimension for their studied unit. These include low
is a bottom-fired box-type furnace with the structure of a straight- Reynolds number and low tube length to diameter ratio.
flow, co-current type heat exchanger. Conventional steam reform- In this paper, a one-dimensional heterogeneous model for the
ers have been in use in various industries such as ammonia pro- catalyst tubes which takes into account the intraparticle mass trans-
duction and methanol plants. However, the production of syngas fer resistance is developed for the reformer. The furnace is mod-
for reducing iron ores in a Midrex process differs from conventional eled with bottom firing configuration. Besides its relative simplicity,
steam reformers in a number of ways [2-5]: (1) The steam in a Midrex the choice of one-dimensional model is based on the recent work
plant is not supplied by a steam generation unit but is provided by of Shayegan et al. [5], who concluded that the quality of predic-
the steam content of recycled off gas stream from the furnace; (2) tions of the two dimensional model depended strongly on fitted cor-
the Midrex reformer operates at lower pressures of 2-3 atm com- relations for wall to fluid heat transfer. They also reported that one-
pared to pressures of 20-40 atm in a conventional steam reformer; (3) dimensional models had better predictions for flue gas temperatures.
the Midrex reformer operates at low stoichiometric ratio of oxidant Moreover, the lack of any radial measurements for the unit under
to carbon compared to conventional steam reformers; (4) the con- study limits the usefulness of any two dimensional model. Another
tent of carbon dioxide in the process feed gas in a Midrex reformer worthy difference in this work is the choice of kinetics. Previous
is higher than most conventional steam reformers which increases studies [2,5] used the simple first-order kinetic model of Akers and
the risk of carbon deposit. Camp [23] for which an analytical solution is available for the effec-
The mathematical modeling of conventional steam reformers tiveness factor, whereas this work uses a more rigorous Langmuir-
used in chemical and petrochemical industries has been studied well Hinshelwood type of kinetics [7] for which a numerical solution
in the literature [6-17]. Various models with different degrees of for the effectiveness factors is needed.
complexity were used for the design and simulation of these units.
These models range from one-dimensional heterogeneous models REACTION KINETICS
to two dimensional models [9,15,16]. It is also known that the heat
transfer from the furnace to catalyst tubes highly affects the perfor- The kinetic rate expressions considered in this project are those
mance of the reactor. Different furnace designs are also available in developed by Xu and Froment [7], based on Langmuir-Hinshelwood
the literature [18-20]. (Hougen-Watson) approach using a commercial catalyst (Holder
The modeling of reformers used in direct reduced iron plants has Topsoe Ni/Mg Al2O4 spinel) in an integral flow reactor. Xu and Fro-
not received similar attention in the literature. Murty and Murthy ment [7] have taken into consideration the main steam reforming
reactions together with the water-gas shift reaction:
To whom correspondence should be addressed.
†

E-mail: aajbar@ksu.edu.sa ↔
CH4+H2O CO+3H2 (1)
2242
 Modeling and simulations of DRI reformer 2243
↔
CO+H O CO +H (2) nH =nH f +(3xCH +xCO )nCH f (16)
CH +2H O↔CO +4H
2 2 2 2 2, 4 2 4,

2 2 2 2
(3) nT=nT f +2xCH nCH f
, 4 4,
(17)
The kinetics of the dry reforming reaction It is therefore possible to express the number of moles ni of any com-
CH +CO 4 2
↔2CO+2H 2
(4) ponent i in the reacting mixture, as well as the total number of moles
nT, in terms of molar feed rates and the conversions of methane and
are negligible with respect to the steam reforming reactions (reac- carbon dioxide.
tions 1 and 3). CO2 mainly hinders the forward reactions (2) & (3).
The authors [7,24] also used an adsorption-desorption mechanism MODEL DEVELOPMENT
model consisting of 13 steps and reached the following rate equa-
tions: 1. Model Assumptions
P P P P 0.5 The following are the main assumptions used for the develop-
r = k ⎛⎝------------------ − ---------------⎞⎠ /DEN
CH4 H2 O H2 CO 2
(5) ment of the reformer model:
1

P
1
K 2.5

• The reactor is at steady state.

1
H2

P P
r = k ⎛⎝----------------- − P--------⎞⎠ /DEN
CO H2 O CO 2
(6) • Radial distribution of the temperature and the concentration
2
P 2
K H2 2
of the different components inside the reactor are uniform (i.e., one-
P P P P 0.5 dimensional model).
r = k ⎛⎝------------------ − ----------------⎞⎠ /DEN
CH4 H2 O H2 CO 2 2
(7) • Heat and mass diffusions in the longitudinal directions are negli-
3

P
3
KK 3.5

gible considering the very high gas velocity at which the reactor is
1 2
H2

with operated (i.e., axial dispersion is negligible).

k-----------------
P - • Ideal behavior is assumed for the gases.
DEN =1+ k P + k P + k P +
H2 O H2 O
(8) • Mass and heat transfer resistances between the fluid and the
CO CO H2 H2 CH4 CH 4
P
particle surface are negligible.
H2

where Pi is the partial pressure of component i. The expressions for 2. Model Equations
reaction rate constants k1, k2 and k3, absorption constants kCO, kH O, 2 A one-dimensional heterogeneous model is used to describe the
kCH and kH , and the equilibrium constants K1 and K2 are summa-
4 2 catalyst tube, which takes into account only the intraparticle mass
rized in Table 1. transfer resistance. One reformer tube performance is assumed to
The rate of disappearance and formation of CH4 and CO2 (taken be representative of any other tube, by assuming the feed gas to be
as key components) is given by: distributed equally among the reformer tubes, and that heat flux pass-
RCH =r +r , RCO =r +r (9) ing through each tube wall is the same regardless of the location of
the tube inside the furnace with respect to the location of the burners.
4 1 3 2 2 3

Let the conversions of methane and carbon dioxide be defined by: 3. Reactor Equations
xCH =(nCH f − nCH )/nCH f (10) The material, energy and momentum balance equations can be
written as follows:
4 4, 4 4,

xCO =(nCO − nCO f)/nCH f (11)

2 2 2, 4,
dxCH /dl=AρθηCH RCH /nCH f (18)
Stoichiometric relations in terms of xCH , xCO can be written as fol-
4 4 4 4,

dxCO /dl=AρθηCO RCO /nCH f (19)

4 2

lows: 2 2 2 4,

nCH =nCH f − xCH nCH f

4 4, 4 4,
(12) 1 -⎧ ρ ∆H η
dT/dl = ---------- r r)
( +
C U ⎨⎩ θ 1 CH4 1 2

nH O=nH O f − (xCH +xCO )nCH f (13)

p s

U (T − T) ⎫
2 2 , 4 2 4,

nCO=nCO f +(xCH − xCO )nCH (14) + ρθ ∆ Hη ( r + r ) + 4----- (20)

, 4 2 4,
2 CO2 2 3
dt i
⎬
⎭
S

nCO =nCO f +xCO nCH f

2 2, 2 4,
(15)
dP
-------- = f(G /ρ d )
T 2
(21)
dl g P

Table 1. Kinetic parameters rate constants [7]

Parameter Value T is the temperature of the reacting mixture, Ts the temperature of
k1 9.49×1015 exp(− 240.1/RT) inner tube skin, PT the total pressure along the reformer, A the re-
k2 4.39×106 exp(− 67.13/RT) former tube cross sectional area, ρθ the bulk density of the catalyst,
ηCH and ηCO are the effectiveness factors associated with the reac-
k3 2.29×1015 exp(− 243.9/RT) 4 2

tions of CH4 and CO2, respectively, dti the internal diameter of the
kCO 8.23×10−5 exp(− 70.65/RT) reformer tubes, us the superficial gas velocity and G is the gas mass
kCH 4
6.65×10−4 exp(38.28/RT) velocity. The heat transfer coefficient U needed in the energy bal-
kH O 2
1.77×105 exp(− 88.68/RT) ance equation (Eq. (20)) is calculated using the correlation devel-
kH 2
6.12×10−9 exp(82.9/RT) oped in [8,24]. The Ergun correlation is used to calculate the friction
K1 exp(− 26830.0/T+30.114) coefficient f in the momentum balance Eq. (21) for large Reynolds
K2 exp(4400/T − 4.036) number:
Korean J. Chem. Eng.(Vol. 28, No. 12)
2244 A. Ajbar et al.
f =1.751---------
−ε
3
(22) where L− is the beam length and
ε
α =β+a +a , α =a +a
1 t r 2 t (32)
r

where ε is the bed void fraction. where a and a are, respectively, the refractory furnace area and the
4. Catalyst Particle Equations r t

tube side furnace area to furnace free volume half ratio. The refrac-
The catalyst pellets are assumed to be a slab with a characteris- tory temperature T is given by:
tic length . The material balance equations for the catalyst pellet
lc
r

take the following form: T = ⎛E--------------

+E ⎞
1/4

⎝
- 1 2
⎠
(33)
σ
r

d P /dω = R RTl /D
2 2
(23) 2

The differential heat balance on the flue gas stream takes the follow-
CH4, p CH4 c CH4, e

d P /dω = − R RTl /D
2
CO2, p
2
(24)
CO 2
2
c CO2, e
ing form:
with the boundary conditions,
G C d-------------------------------------------
[T − F(T − T )]
*
g
- = 2 β[E − ψ E ] g, 0
(34)
P = P & P = P at ω =1.0
CH 4, P CH 4, s CO 2, P
(25) CO2, s
g g
dl 1 g

dP where G is the process flue gas mass velocity, T* the adiabatic flame
-------------- = dP
-------------- = 0 at ω = 0
g

dω
CH 4, P

dω
CO 2, P
(26) temperature, F the fraction of the fuel burned along the reformer
and E is the gas emissive powers in side fired furnaces. The flue
g

P and P designate, respectively, the partial pressure of component

i, p i, s gas inlet temperature T is given by g, 0

i in catalyst particle and on catalyst particle surface, ω the dimen- E 1/4

sionless coordinate of the catalyst pellet and D is the effective molec- i, e T = -------
g, 0
σψ
1.0
(35)
ular diffusivity of component i. The slab is assumed to be isother-
mal while external mass and heat transfer resistances are assumed where E is the value of E at the gas inlet. The heat transfer by
negligible. The physico-chemical parameters such as viscosity, ther-
1, 0 1

radiation to the reformer tubes is given by:

mal conductivity, molecular diffusivities, are obtained using stan- Q =2ε a (E − E )V (36)
dard correlations [24,25]. r t t r t

5. Modeling of the Furnace where E is the tube emissive powers in side fired furnaces, E =σT
t r r
4

One type of firing will be considered, the bottom fired furnace. and V is the free volume of the reformer.
Roesler [18] introduced the two flux model for the one-dimensional
furnaces. The model is capable of taking into account the radiative NUMERICAL SOLUTION STRATEGY
heat along and normal to the gas flow. We will adopt in this work
the Roesler model that was modified by Filla [26] to take the fol- The heat transfer equations describing the furnace can be decou-
lowing form: pled from those describing the reaction in the reformer tubes by
assuming values of the outer wall temperature profile of the reformer
d E- = α {β(E − σψT ) + ε a (E − σψT
2
--------- 1 4 4
tubes, then carrying out consecutive iterations between the furnace
0)
dl 2 1 1 g t t 1 t,

equations and the tube equations until the two iterations match. For
+ ε a ( ( 1− ψ ) E − ψ E )}
r r 1 2 (27) the bottom-fired furnace, the differential equations describing the
d---------
2
E- = α {ε a (E − σ(1− ψ)T ) + ε a (ψE − (1− ψ)E )} heat transfer in the furnace are discretized by the orthogonal collo-
dl 2
2
2 t t 2
4
t, 0 r r 2 1 (28) cation method using a cubic polynomial leading to simultaneous
non-linear algebraic equations, which are solved by the subroutine
with the following boundary conditions: ZSPOW of the IMSL (International Mathematical and Statistical
At =0, Library) which is based on discretized Newton method. On the other
l
hand, using the assumed wall temperature profile, the four differen-
----1-dE 1 dE- = -----------
-------- = − ----1- ------- ε
- {( 1− ψ ) E − ψ E }
2 r
(29) tial equations describing the reformer tubes are solved by a fourth-
α dl α dl 2 − ε
1 2
1 2 r

order Runge-Kutta routine with variable step size, and then itera-
At =L, l tions are performed to obtain the correct wall temperature. At every
1 dE 1 dE ε
− ----- -------- = ----- -------- = ------------ {( 1− ψ ) E − ψ E }
1 2 r
(30) location along the reactor the effectiveness factors for the CH and
α dl α dl 2 − ε
4

1 2 r
1 2
CO have to be calculated. For this purpose, the modified colloca-
2

E and E are, respectively, the grey and clear gas component heat tion method developed by Soliman [27] is used. The effectiveness
1 2

gas flux, T the temperature of furnace gas, T the outer tube sur- factors are calculated using the following formulas:
g t, o

face temperature, σ the Stefan-Boltzmann constant, ψ the band radi- PCO

dP
1/2

2.0 × R (P P -------------- /R (37)

2, S

ation fraction, β the absorption coefficient and ε , ε are, respectively, g t

η CO 2
= ∫ CO 2
,
CH 4, p CO 2, p
)
CO2, p

C CO2, S

the emissivities of the flue gas, refractory slab and tube wall. The PCO
2, C
1

other parameters appearing in Eqs. (27)-(30) are defined in the no- 1/2

menclature. In this model, it is assumed that the gas emissivity can η = 2.0 ×
PCH
4, S

R (P , P dP
) --------------
CH4, p
/R (38)
be approximated by
CH 4

PCH
∫
4, C
CH 4 CH 4, p CO 2, p
C 2
CH4, S

ε =ψ {(1− exp(− BL− /2)}

g (31) With
December, 2011
 Modeling and simulations of DRI reformer 2245
2 2
C1= RT-----------
l
-, C2 = RT-----------
c l
- c
(39)
D D
CO 2, e CH 4, e

Some model parameters were needed to be tuned in order to fit the

plant data. These include the effectiveness factors for carbon diox-
ide (Eq. (37)) and methane (Eq. (38)) as well as the heat transfer
coefficient. After obtaining the center pressures, it is required to
have a relation between the pressures of CH4 and CO2 at any position
inside the catalyst particle in order to obtain the integrals in equa-
tions (Eq. (37)) and (Eq. (38)). This relation is obtained by linear
interpolation between the center and surface pressures to get:
P −P
---------------------------- P −P
CO2, p CO 2, S
= ----------------------------
CO 2, c CO2, S
(40)
P −P
CH4, p CH 4, S
P −PCH 4, c CH4, S

RESULTS AND DISCUSSION

The simulations were carried out using the reformer design and
operating parameters presented in Tables 2 and 3. First, a valida-

Table 2. Nominal values of design parameters of reformer Fig. 1. Variations of process gas temperature and effect of some
operating parameters.
Parameter Value
Number of tubes 521 tion study was carried out using plant data from a local iron/steel
Inside tube diameter (m) 0.2 plant. Table 4 summarizes the comparison between model predic-
Tube thickness (m) 0.024 tions and plant data for exit conditions in the reformer. The avail-
Tube length (m) 7.9 able plant data are the process temperature and the dry composition
Catalyst particles characteristic length (m) 0.0038 of the process gas. Process gas temperature shows excellent fit. As
Catalyst pellets bulk density (kg/m3) 250 for dry composition, it can be seen that while the compositions of
Catalyst pore radius (A) 80 H2 and CO show very good fit, those of CO2 and CH4 show some
Tortosity 2.74 deviations. The model predicts that too much methane is steam re-
formed, while the mole fraction of carbon dioxide is larger. These
Table 3. Nominal values of operating parameters of reformer discrepancies could be due to the chosen kinetics and also to the
tuning of the effectiveness factors. Moreover, given the small values
Parameter Value of the composition themselves, measurement errors could also con-
Natural gas flow rate to reformer (103Nm3/hr) 22.37 tribute to these deviations.
Fraction of methane in natural gas (%) 95 Fig. 1 shows the profiles of process gas temperature. An increase
Total pressure (atm) 2.46 in the gas temperature along the distance is expected since the heat
Temperature (K) 822.2 from fuel is enough to sustain the increase in the temperature. The
H2O/CH4 0.44 profiles of both the wall temperature and the flue gas temperature
H2/CH4 2.46 show, on the other hand, a non monotonic behavior (Figs. 2(a)-(b)).
CO2/CH4 0.54 The profiles are plotted for the four collocation points only (explain-
N2/CH4 0.17 ing the non smoothness of the profiles). Initially, the reaction rates
CO/CH4 1.26
are small, but they keep increasing along the distance causing a rise
in both wall and flue gas temperatures. However, since the reaction
is endothermic, the heat removal increases along the distance, which
Table 4. Comparison between plant data and model predictions causes a decrease in the temperatures.
at the reformer exit conditions Figs. 1 and 3 show some sensitivity analysis for the effect of key
Model Plant operating parameters. A decrease in the inlet temperature (from 822.21
Parameter predictions data to 750 K) causes evidently a decrease in the process gas tempera-
Process gas exit temperature (K)
ture (Fig. 1). A decrease in the ratio of CO2/CH4 (from 0.54 to 0.35)
948 945
causes a less extent of dry reforming, which leads to a decrease in
Composition of dry process gas (%) the heat needed for reforming and therefore causing an increase in
H2 57.0 57.3 the process gas temperature (Fig. 1). A decrease in the ratio of CO/
CO 36.1 36.3 CH4 (from 1.26 to 1.00) pushes the reactions to occur in the forward
CO2 05.2 03.5 direction, and since the reactions are endothermic, this will cause a
CH4 00.3 01.2 decrease in the gas temperature.
N2 01.4 01.7 Fig. 3 shows that an increase in the flow rate of methane (from
Korean J. Chem. Eng.(Vol. 28, No. 12)
2246 A. Ajbar et al.

Fig. 3. Variations of process gas temperature and effect of some

operating parameters.

Fig. 2. (b) Variations of: (a) outside wall temperature; (b) flue gas
temperature with dimensionless distance at the collocation
points.
Fig. 4. Variations of hydrogen conversion and effect of some oper-
ating parameters.
2.07 kmol/h to 2.33 kmol/h) evidently decreases the gas tempera-
ture rise for a given constant heat. The increase in the ratio of H2O/ will cause the hydrogen conversion to be smaller (Fig. 4). A decrease
CH4 (from 0.43 to 0.70) causes more steam reforming and there- in the ratio of CO2/CH4 causes the water gas shift reaction to in-
fore more heat required for the reaction, and this will ultimately re- crease resulting in an increase in the conversion of hydrogen.
duce the process gas temperature. On the other hand, the effect of a Fig. 5 shows that an increase in the ratio of H2O/CH4 causes an
decrease in the ratio of H2/CH4 (from 2.45 to 1.8) causes more re- increase in reforming and this causes an increase in hydrogen con-
forming and this will cause a reduction in the gas temperature. version. On the other hand, a decrease in the ratio of H2/CH4 increases
The variations of conversions of H2, CO2 and CH4 along the dis- reforming and this causes hydrogen conversion to increase. A de-
tance are shown in Figs. 4-9. It can be seen that hydrogen conver- crease in the ratio of CO/CH4 increases the reverse water gas shift
sion increases with distance as a natural result of reforming. Fig. 4 reaction, therefore causing a decrease in hydrogen.
shows that a decrease in the inlet temperature causes less reform- As for the conversion of CH4, Fig. 6 shows that the conversion
ing and therefore leads to a decrease in the conversion compared of methane increases with the distance as it is consumed along the
to the nominal case. An increase in the methane flow rate results in tubes. A decrease in the inlet temperature causes less reforming and
smaller residence time and therefore small extent of reaction that therefore a decrease in conversion of methane relative to the base
December, 2011
 Modeling and simulations of DRI reformer 2247

Fig. 5. Variations of hydrogen conversion and effect of some oper- Fig. 7. Variations of methane conversion and effect of some oper-
ating parameters. ating parameters.

Fig. 8. Variations of carbon monoxide conversion and effect of some

operating parameters.
Fig. 6. Variations of methane conversion and effect of some oper-
ating parameters. perature causes less reforming and thus less conversion of CO. An
increase in methane flow rate results in smaller residence time, and
case. The increase in the methane flow rate causes less reforming therefore less conversion of CO. Unlike hydrogen, a decrease in
and leads to a decrease in methane conversion compared with the the ratio of CO2/CH4 decreases the CO conversion because of less
base case. The decrease in the ratio of CO2/CH4 results in less dry reforming. Fig. 9 shows that the increase in H2O/CH4 increases the
reforming, and therefore the value of methane conversion decreases conversion of CO because of the increase in reforming. A decrease
relative to the base case. Fig. 7 shows that the increase in the ratio in H2/CH4 causes a decrease in the conversion of CO because some
of H2O/CH4 increases the reforming and therefore increases the con- of the CO produced in the reforming is consumed in the water gas
version of CH4, while the decrease in the ratio of H2/CH4 causes shift reaction. Finally, a decrease in the ratio CO/CH4 causes evi-
more reforming and also leads to an increase in CH4 conversion. dently an increase in CO conversion because of more reforming.
Finally, a decrease in the ratio of CO/CH4 results in more reform-
ing and this causes an increase in methane conversion. CONCLUSIONS
Similar to hydrogen, Figs. 8-9 show an increase in the conver-
sion of carbon monoxide with distance. A decrease in the inlet tem- A first principle mathematical model was developed for a reform-
Korean J. Chem. Eng.(Vol. 28, No. 12)
2248 A. Ajbar et al.
f : friction factor
F : fraction of fuel burned along the reformer
ni f : molar flow rate of gas i in the feed (i=H , CH , CO, CO ,
, 2 4 2

H O) [Kmol/hr]
2

G : process gas mass velocity [Kg/hr·m ] 2

Gg : process flue gas mass velocity [Kg/hr·m ] 2

K , K : equilibrium constant for reaction 1, 2 [bar ]

1 2
2

k , k : rate coefficient of reaction 1, 3 respectively [Kmol·bar/ Kgcat·

1 3

hr]
k : rate coefficient of reaction 2 [Kmol/Kg-cat·hr·bar]
2

ki : adsorption constants for (i=CH , CO and H ) [bar− ]

4 2
1

kH O : dissociative adsorption constant for H O

2 2

L : reformer tube heated length [m]

l : coordinate [m]
lc : characteristic length of pellets [m]
ni : number of moles of component i in the reacting mixture
[Kmol/hr]
Pi : partial pressure of component i in the gas bulk [bar]
Fig. 9. Variations of carbon monoxide conversion and effect of some Pi c : partial pressure of component i at the center of the particle
,

operating parameters. [bar]

Pi p : partial pressure of component i at any position inside the parti-
,

cle [bar]
er used in the Midrex direct reduction plant. A steady state, one- Pi s : partial pressure of component i on catalyst particle surface
,

dimensional model with negligible axial dispersion was adopted [bar]

for the catalyst tubes. The furnace was, on the other hand, modeled Qr : net heat release rate [Kcal/hr]
with bottom firing configuration. Satisfactory results were obtained ri : rates of reactions (i=1, 2 and 3) [Kmol/Kgcat·hr]
when the model was validated against temperature and dry gas mole R : ideal gas constant [KJ/Kmol·K]
fractions, the only data available from a local iron/steel plant. The T : temperature of the reacting mixture [K]
model was solved using the orthogonal collocation method for the Tg : temperature of furnace gas [K]
base case to obtain local profiles of temperature and conversions, Ts : temperature of inner tube skin [K]
and was subsequently used for a sensitivity analysis of the effect of Tr : refractory temperature [K]
reactor operating parameters. Simulations were carried out to pre- V : free volume of reformer [m ] 3

dict the reformer performance for variations in the inlet temperature,

methane flow rate and ratios of H O, H , CO , CO to methane. The
2 2 2
Greek Letters
results obtained from the sensitivity analysis may be helpful in further ∆Hi : enthalpy change of reaction i [Kcal/mol]
optimization studies of the unit. εg : emissivity of the flue gas
εr : emissivity of refractory slab
ACKNOWLEDGEMENT εt : emissivity of tube wall
ηi : effectiveness factors associated with component i
This work was made possible by a generous grant from the Na- ρg : process gas density [Kg/m ] 3

tional Plan for Science and Technology (Project # 08-ENE337-2) ρθ : catalyst bed bulk density [Kg/m ] 3

of Saudi Arabia. σ : Stefan-Boltzmann constant

ω : dimensionless coordinate of the catalyst pellet
NOMENCLATURE ψ : band radiation fraction
A : reformer tube cross sectional area [m ] 2
REFERENCES
ar : refractory furnace area to furnace free volume half ratio [m− ] 1

at : tube side furnace area to furnace free volume half ratio [m− ] 1
1. Midrex, Inc., http://www.midrex.com/.
Cp : heat capacity of process gas [KJ/Kg K] 2. F. Farhadi, M.Y.M. Hashemi and M.B. Babaheidari, Ironmak. Steel-
Cg : heat capacity of flue gas [KJ/Kg K] mak., 30, 18 (2003).
dp : center to center distance between catalyst particle diameter 3. F. Farhadi, M. B. Babaheidari and M. M. Y. Hashemi, Appl. Therm.
[m] Eng., 25, 2398 (2005).
dti : internal diameters of the reformer tubes [m] 4. M. Sadri, K. Vakhshouri and M. M. Y. M. Hashemi, Ironmak. Steel-
E : grey gas component heat gas flux [W/m ]
1
2
mak., 34, 115 (2007).
E : clear gas component heat gas flux [W/m ]
2
2
5. J. Shayegan, M. M. Y. Hashemi and K. Vakhshouri, Can. J. Chem.
Eg : gas emissive powers in side fired furnaces [J/m ·hr] 2
Eng., 86, 757 (2008).
Et : tube emissive powers in side fired furnaces [J/m ·hr] 2
6. M. H. Hyman, Hydrocarb. Process., 49, 131 (1968).
December, 2011
 Modeling and simulations of DRI reformer 2249
7. J. Xu and G. F. Froment, AIChE J., , (1969).
35 44 , 6342 (2005).
8. J. C. De Deken, E. F. Devos and G. F. Froment, Steam reforming of 17. S.-K. Jeon, C.-S. Park, S.-D. Kim, B.-H Song and J. M. Norbeck,
natural gas: Intrinsic kinetics, diffusional influences, and reactor Korean J. Chem. Eng., , 1279 (2008).
25

design, Chemical Reaction Engineering, ACS Symp. Ser., 196, Bos- 18. F. C. Roesler, Chem. Eng. Sci., , 1325 (1967).
22

ton (1982). 19. C. McGreavy and M. Newman, IEEE Conf. on the Ind. Appl. of
9. J. R. Rostrup-Nielsen, J. Catal., , 31 (1984).
85 Dynamic Modelling, Durham., Sep. (1964).
10. M. A. Soliman, S. El-Nashaie, A. Al-Obaid and I. Idriss, Chem. Eng. 20. C. P. P. Singh and D. N. Saraf, Ind. Eng. Chem. Proc. Des. Dev., ,
18

Sci., , 1801 (1988).

43 1 (1979).
11. K. Ravi, Y. K. Joshi and B. K. Guha, Chem. Eng. Technol., , 358
12 21. C. V. S. Murty and M. V. K. Murthy, Ind. Eng. Chem. Res., , 1832
27

(1989). (1988).
12. S. El-Nashaie, A. Idriss, M. A. Soliman and A. Al-Ubaid, Can. J. 22. N. Selcuk, R. G. Siddal and J. M. Beer, J. Inst. Fuel., , 89 (1975).
48

Chem. Eng., , 786 (1992).

70 23. W. W. Akers and D. P. Camp, AIChE J., , 471 (1955).
4

13. H. M. Kvamsdal, H. F. Svendsen and O. Olsvik, Chem. Eng. Sci., 24. G. F. Froment and L. B. Bischoff, Chemical reactor: Analysis and
54 , 2697 (1999). design, John Wiley & Sons (1990).
14. J. K. Rajesh, S. K. Gupta and A. K. Ray, Ind. Eng. Chem. Res., ,39 25. R. Reid and T. Sherwood. The Properties of gases and liquids,
706 (2000). McGraw Hill (1958).
15. M. N. Pedenera, J. Pina, D. O. Borio and V. Bucala, Chem. Eng. J., 26. M. Filla, Chem. Eng. Sci., , 159 (1984).
39

94 , 29 (2003). 27. M. A. Soliman, Chem. Eng. Sci., , 1198 (1988).

43

16. A. G. Dixon, M. Nijemeisland and E. H. Stitt, Ind. Eng. Chem. Res.,

Korean J. Chem. Eng.(Vol. 28, No. 12)