15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 598

Fatemeh Ghodoosi
Mathematical Modeling of Reverse
Mohammad Reza Khosravi-
Nikou* Water-Gas Shift Reaction in a Fixed-Bed
Ahmad Shariati
Reactor
A two-dimensional, pseudo-homogeneous model is used to study the reverse
water-gas shift (RWGS) reaction in a fixed-bed reactor under steady-state condi-
tions. A redox kinetic model is applied in the reactor model for a description of
the reaction kinetics. Results show that the model predictions have an average
error of 2.7 % and are in good agreement with experimental data for nonequi-
librium conditions. With increasing space velocity, the CO2 conversion decreases,
but the temperature drop in the reactor increases. Furthermore, a H2/CO2 ratio
equal to one in the feed is the optimum ratio. Finally, the temperature, concentra-
tion, and pressure profiles inside the reactor are predicted.

Keywords: Fixed-bed reactor, Kinetics, Mathematical model, Reverse water-gas shift

Received: April 10, 2016; revised: August 01, 2016; accepted: January 04, 2017
DOI: 10.1002/ceat.201600220

1 Introduction CO2 þ H2 Ð CO þ H2 O; DH298 K ¼ 41:2 kJ mol1 (1)

The concentration of CO2 has constantly risen from about The endothermic RWGS reaction can be used at high tem-
280 ppm in the atmosphere before the industrial revolution to perature over Fe-, FeSi-, Fe3O4-, Fe-Cu-, ZnO-, Pt-, Pt-Ca-,
about 390 ppm in 2010, and is predicted to reach about and Pt-Mg-based catalysts [22–26], and this reaction is used at
570 ppm by the end of the century [1]. The increase in CO2 low temperature over Cu-, Cu-ZnO-, and Pd-based catalysts
emissions arguably contributes to the increase in global tem- [25, 27, 28].
perature and climate changes due to the greenhouse effect. The products of the RWGS reaction are CO and H2O. How-
Hence, there has been increasing pressure on countries and sci- ever, at some level of CO, it may react with hydrogen, which
entists to curb CO2 emissions and develop efficient CO2 cap- exists in reaction medium to form methane according to the
ture and utilization systems [2–5]. side reaction given in Eq. (2):
Reducing CO2 emission is an extensive and long-term task.
In principle, there are three possible strategies in this regard: CO þ 3H2 fi CH4 þ 3H2 O (2)
reduction of the amount of CO2 produced, storage of CO2, and
By using the impregnation method, instead of a precipitation
usage of CO2 [1, 6–8]. The first strategy requires improvements
method, in catalyst preparation the CO selectivity can be
in energy efficiency and a switch toward less carbon intensive
enhanced in the RWGS reaction [29]. Also, the addition of
energy sources, such as hydrogen and renewable energy
metals, such as Fe, K, and Pd, supported on Al, Si, and Ce can
[7, 9–11].
increase CO selectivity in the RWGS reaction [28–30].
Storage of CO2, as the second strategy, involves the develop-
The most-used reactor for the RWGS reaction is a catalytic
ment of new technologies for capturing [4, 12–14] and CO2
fixed-bed reactor [31]. Several mathematical models for this
sequestration, which is a relatively well-established process
type of reactor have been developed to date. Pereira Duarte
[1, 7, 15–17]. As an economical, safe, and renewable carbon
et al. [32] in 1984 introduced three types of models: (i) pseudo-
source, CO2 turns out to be an attractive C1 building block for
homogeneous; (ii) heterogeneous, written in the most usual,
making organic chemicals, materials, and carbohydrates
but incorrect way; and (iii) heterogeneous, written in the cor-
[18, 19].
rect way, namely, heat transfer through the solid phase is
Conversion of CO2 into CO by catalytic hydrogenation is
recognized as one of the most promising processes for CO2 uti- –
lization. In industry, synthesis gas, H2 and CO, can be used to
Fatemeh Ghodoosi, Dr. Mohammad Reza Khosravi-Nikou, Dr. Ahmad
produce long-chain hydrocarbons through the Fischer-Tropsch Shariati
reaction [20, 21]. The reverse water-gas shift (RWGS) reaction [email protected]
is one available method for the production of CO. The RWGS Gas Engineering Department, Petroleum University of Technology,
reaction is presented as Eq. (1): The End of Ayatollah Behbahani Highway, Ahwaz 63431, Iran.

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 599

included in the heat balance of phase. In that comparison, the 2.2 RWGS Reactor System
importance of the two-dimensional type (iii) model is empha-
sized. Although it has seldom been used, it represents correctly, Catalytic tests were conducted in a fixed-bed reactor with an
for this kind of reactor, the heat and mass transfer phenomena inner diameter of 20 mm and a length of 150 mm at atmo-
in and between phases. spheric pressure. A schematic diagram of the experimental
In a subsequent study, Pereira Duarte et al. developed a het- apparatus and configuration of the reactor are shown in Fig. 1.
erogeneous one-dimensional model for nonadiabatic fixed-bed When Fe-Mo/Al2O3 powder catalyst was loaded into the reac-
catalytic reactors, and analyzed the results obtained in compar- tor, prior to the reaction, it was reduced in situ at 923 K for 4 h
ison with the two-dimensional model and other one-dimen- in a 200 mL min–1 flow of a mixture of H2/N2 (1/5). After being
sional models. The choice between a pseudo-homogeneous or reduced, H2 and CO2 were continuously fed into the reactor,
heterogeneous model depends on the importance of the inter- and reaction was performed over the catalyst to produce CO
facial gradients, which depend mainly on the ratio of Dp/Dt1). If and H2O. A cold trap at the outlet of the reactor was used to
the reactor operates under mild conditions, namely, those con- condense any water from the gas product stream.
ditions that generate almost parabolic radial temperature pro- All products were analyzed by using a gas chromatograph
files, a one-dimensional model can be used. On the other hand, (Young Lin) equipped with Q and MS capillary columns and a
for more extreme conditions, a higher hot spot temperature or HID detector. CO, H2, CO2, and CH4 were detected by GC and
higher apparent activation energies, a two-dimensional model their respective mole fractions were calculated from the peak
is suggested [32]. area with a third-order calibration function. Tests were
In a study by Lue et al. [33], the modeling processes of fixed- repeated five times to find the CO2 conversion for each catalyst
bed catalytic reactors for Fisher-Tropsch synthesis were dis- loading. The results led to an estimated accuracy of ± 3 % in
cussed. The proposed model was two-dimensional heterogene- experimental measurements.
ous and considered both fluid and solid phases as a continuum.
This model accounts for axial and radial dispersion of heat in
the fluid and solid phases, and axial and radial dispersion of 3 Development of the Reactor Model
mass in the fluid phase.
A fixed-bed reactor model for the methanation of CO2 was The upmost principle of the reactor model establishing proce-
investigated by Schlereth and Hinrichsen [34]. In their study, dures is to represent the actual reactor as accurately as possible
the model predictions of differently detailed pseudo-homo- under the conditions of necessary assumptions and simplifica-
geneous reactor models and a heterogeneous reactor model, in tions, which, as a result, also helps the simulation to converge
which intraparticle transport of mass was described according efficiently and steadily to a reliable result. Therefore, the
to a dusty-gas approach, were compared to assess the required following assumptions are made based on the above descrip-
level of detail in terms of modeling the heat transfer, fluid flow tions:
characteristics, and transport resistances on the pellet scale. – In practice, in the studied system, the gas velocity is great
Under specific conditions, intraparticle mass transfer and (0.4–0.9 m s–1) and the catalyst particles are small
external heat transfer need to be considered to describe the (55–60 mm), heat and mass transfer between catalyst and
temperature and concentration profiles adequately. fluid are strong enough. In other words, the difference in
Herein, the Fe-Mo/Al2O3 catalyst was synthesized by using temperature and concentration between the catalyst phase
an impregnation method to prevent side reactions. RWGS and fluid phase can be ignored for a catalytic reaction [36].
reaction experiments over the Fe-Mo/Al2O3 catalyst were per- This implies that a pseudo-homogeneous reactor model can
formed in a fixed-bed reactor. Then a two-dimensional pseu- be applied in this study.
do-homogeneous reactor model based on two-dimensional – In the mass balance equation, a new redox kinetic model,
material and energy balance equations and incorporating the which was developed in a previous study [37], was used to
redox reaction kinetic model for RWGS reaction was devel- describe the RWGS reaction rate.
oped. This model describes the performance of fixed-bed reac- – In the energy balance equation, it was assumed that heat
tor with a catalyst bed that behaves as a porous medium. transfer at the wall was equal to constant heat flux tempera-
ture of the reactor, which is more realistic than the assump-
tions made in previously in the literature.
2 Experimental Section – In the momentum balance equation, it was assumed that the
pressure drop in the radial direction was negligible. Also, the
2.1 Catalyst Preparation pressure drop in the axial direction is dominated by friction.

The Fe-Mo/Al2O3 catalyst was synthesized through two-step

impregnation method from (NH4)6Mo7O244H2O (Merck), 3.1 Mathematical Modeling of Reactor
g-Al2O3 (Merck Co., 170 m2g–1, 99 % pure, porosity = 1.3 cm3g–1,
pore diameter = 18 nm) and Fe(NO3)26H2O (Merck Co., 99 % A steady-state, two-dimensional, pseudo-homogeneous model
pure), as described in [35]. for the fixed-bed reactor in which the catalyst bed is assumed
to be a porous medium is considered. The mass, energy, and
– momentum balance equations for the differential element vol-
1) List of symbols at the end of the paper. ume in the selected reactor are given below.

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 600

in which Cp,f is the fluid heat capacity, Ct

is the total gas concentration, and leff is the
effective thermal conductivity. The enthal-
py change of reaction is defined by Eq. (8):

ZT
DHr;T ¼ DHr;298 þ DCp ðT ÞdT (8)
298

in which DHr,T is the enthalpy change of

the reaction at temperature T, DHr,298 is the
standard enthalpy change of the reaction
(at 298 K), and DCp is defined by Eq. (9):
X X
DCp ðT Þ ¼ Cp; products  Cp; reactants
(9)

The boundary conditions of mass and

energy balances are given by Eqs. (10)–
(13):
Figure 1. A schematic diagram of the fixed-bed reactor. z ¼ 0 : Ci ¼ Ci;in ; T ¼ Tin (10)

3.1.1 Mass Balance ¶Ci ¶T

z¼L : ¼ ¼0 (11)
¶z ¶z
The mass balance in the reactor is given by Eq. (3):
 2  ¶Ci ¶T
¶ðus Ci Þ ¶ Ci 1 ¶Ci ¶2 Ci r¼0: ¼ ¼0 (12)
¼ Deff ;i þ þ þ rs ð1  eb Þni rðC; TÞ ¶r ¶r
¶z ¶r2 r ¶r ¶z 2
(3) ¶Ci ¶T
r ¼ Dt =2 : ¼ 0 ; leff ¼ q ¢¢ (13)
¶r ¶r
in which Ci is the concentration of species i, z and r are the
axial and radial positions, Deff,i is the effective diffusivity, rs is in which L is the bed length and q¢¢ is the constant wall heat
the catalyst density, eb is the porosity of the bed, ni is the stoi- flux.
chiometric coefficient, r(C,T) is the rate of the reaction, and us
is the superficial velocity. With an ideal gas behavior approxi-
mation, the linear velocity can be calculated from Eq. (4) [38]: 3.1.3 Momentum Balance
T
us ¼ a (4) The momentum balance is considered for pseudo-steady state.
P
The pressure drop correlation with the Ergun friction factor for
in which a is given by Eqs. (5) and (6): gas flow through a packed-bed reactor is written as Eq. (14)
[39]:
a ¼ GRg (5) ! !
 
dP 1  eb ru2s 150
4Ft  ¼ þ 1:75 (14)
G¼ (6) dz e3b Dp Rep
pD2t

in which T is the temperature, P is the total pressure of gas, G in which r the is density of the gas mixture, Dp is the catalyst
is the total molar flux, Rg is the universal gas constant, Ft is the particle diameter, Rep is the particle Reynolds number defined
total molar flow rate, and Dt is the reactor tube diameter. by Eq. (15):

rus;0 Dp
Rep ¼ (15)
3.1.2 Energy Balance ð1  eb Þm

The energy balance equation in the reactor is Eq. (7): in which us,0 is the initial superficial velocity and m is the vis-
 2  cosity of the gas mixture. The boundary condition (Eq. (16)), at
¶T ¶ T 1 ¶T ¶2 T the bed exit, is applied to maintain the effluent stream leaving
Cp;f Ct us ¼ leff þ þ
¶z ¶r 2 r ¶r ¶z 2 the reactor at atmospheric conditions:
 
þ DHr;T rs ð1  eb ÞrðC; TÞ (7) z ¼ L : P ¼ Patm (16)

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 601

3.2 Heat and Mass Transfer Correlations 3.3 Physical Properties and Mixing Rules

3.2.1 Bed Porosity The properties of Fe-Mo/Al2O3 catalyst used in the RWGS
reaction are presented in Tab. 1. It has a particle size of
The porosity of the catalyst bed can be calculated from Eq. (17) 55–60 mm and a solid density of 4580 kg m–3. The thermal con-
[40]: ductivity of the catalyst (solid phase) is calculated from the vol-
" # ume fraction of the compounds in the particle by using
ðDt =Dp  2Þ2 Eq. (24) [39]:
eb ¼ 0:38 þ 0:073 1  (17)
ðDt =Dp Þ2 1
ls ¼ P f (24)
v;i

ls;i
3.2.2 Effective Thermal Conductivity
in which fv,i and ls,i are the volume fraction and thermal con-
The effective thermal conductivity can be obtained from ductivity of compound i in the particle, respectively.
Eq. (18), in which the arrangements of the solid matrix and
pores are in parallel and in series [41]: Table 1. Physical properties of the catalyst.

1 2 Property Value
leff ¼ feb lf þ ð1  eb Þls g þ feb =lf þ ð1  eb Þ=ls g1
3 3
Fe [mass %] 2.8
(18)
Mo [mass %] 8.6
in which lf and ls are thermal conductivities of fluid and solid
phases, respectively. Al2O3 [mass %] 88.6

Particle size [mm] 55–60

2 –1
3.2.3 Effective Diffusion Coefficient Specific surface area [m g ] 102
–3
Density [kg m ] 4580
The effective mass diffusion coefficient is expressed by Eq. (19) –1 –1
Thermal conductivity [W m K ] 7.35
[42]:
e
Deff ;i ¼ b Dm;i (19)
t 3.3.1 Gas Diffusivity
in which Dm,i is the molecular diffusion coefficient of species i
and t is the tortuosity factor, which is related to the porosity Gas-phase diffusion (Dm,i) can be approximated by using
through Eq. (20) [43]: Eq. (25) [45]:
tðeb Þ ¼ 1  p lneb (20) 1  yi
Dm;i ¼ P (25)
j „ i yj =Dij
in which p is the tortuosity parameter and, for porous systems
composed of freely overlapping spheres, is 1/2. If the gas density in which Dij is the binary gas-phase diffusivity and yi is the
is low, or if the pores are quite small, or both, the molecules mole fraction of species i. Binary gas-phase diffusivities for a
collide with the pore wall much more frequently than with each wide range of applicable temperatures and pressures are esti-
other. This is known as Knudsen diffusion. For gases in a mated through published correlations detailed in Tab. 2.
straight round pore, it can be represented by Eq. (21) [44]:
sffiffiffiffiffiffi
eb T
DK;i ¼ 19 400 (21) 3.3.2 Gas-Phase Heat Capacity
Sg rb Mi

in which DK,i is the Knudsen diffusion coefficient of species i Useful gas-phase heat capacity (Cp,i) values are shown in Tab. 3
(cm2s–1), Sg is the total surface of the porous material (cm2g–1), [38]. The heat capacity of the mixture (Cp,f) is taken as the mo-
rb is the average bulk density (g cm–3), T is the temperature lar average of the gas component heat capacity [45].
(K), and Mi is the molecular weight of species i. The Knudsen
diffusion coefficient for a porous solid (DK eff ;i Þ becomes
Ns
X
Cp;f ¼ yi Cp;i (26)
Eq. (22): i¼1
e
DK eff ;i ¼ b DK;i (22)
t
So, the effective diffusivity is calculated from molecular diffu- 3.3.3 Gas-Phase Viscosity
sivity and Knudsen diffusivity by using Eq. (23) [43]:
The gas-phase viscosity of species can be estimated from the
1 t 1 1 DIPPR correlation used in Aspen Plus, shown in Tab. 4. [38].
¼ ð þ Þ (23)
Deff ;i eb DK;i Dm;i The viscosity of the gas mixture is estimated from Eq. (27) [46]:

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 602

Table 2. Binary gas diffusivity for component pairs.

Pair A B C D E F Eq.*

H2-CO 15.39 ·10–3 1.548 0.316 ·108 1 –2.8 1067 a

–5
H2-CO2 3.14 ·10 1.75 – 0 11.7 0 a

H2-H2O – 1.020 – – – – b
–5
CO-CO2 3.15 ·10 1.57 – 0 113.6 0 a
–5
CO-H2O 0.187 ·10 2.072 – 0 0 0 a
–5
CO2-H2O 9.24 ·10 1.5 – 0 307.9 0 a
B –2D 2 2 –1
* (a) Dij = (AT /P)[ln(C/T)] exp(–E/T–F/T ), (b) Dij = B/P; T in K, P in atm, and Dij in cm s [38].

Table 3. Gas-phase heat capacity constants for selected species Table 5. Gas-phase thermal conductivity constants for selected
(Cp,i = A + BT + CT2 + D/T2; T in K and Cp,i in cal mol–1K–1) [38]. species (l = (ATB)/(1 + C/T + D/T2); T in K and l in kcal h–1m–1K–1)
[38].
Species A B C D
Species A B C D
CO 6.60 1.2 ·10–3 0 0
–4
CO 5.1489 ·10 0.6863 57.13 501.92
CO2 10.34 2.74 ·10–3 0 –1.955 ·105
CO2 3.1728 –0.3838 964 1.86 ·106
H2 6.62 0.81 ·10–3 0 0
H2 2.2811 ·10–3 0.7452 12 0
H2O 8.22 0.15 ·10–3 1.34 ·10–6 0
–6
H2O 5.3345 ·10 1.3973 0 0

Table 4. Gas-phase viscosity constants for selected species

(m = AT B)/(1 + C/T + D/T2); T in K and m in N s m–2) [38].  
2
m 0:5 Mj 0:25
Species A B C D 1 þ ð iÞ
mj Mi
Aij ¼ pffiffiffi 0:5 (29)
CO 1.1127 ·10 –6
0.5338 94.7 0 8 1 þ ðMi =Mj Þ
CO2 2.148 ·10–6 0.46 290 0
Note that this equation reduces to Aij = 1.
–7
H2 1.797 ·10 0.685 –0.59 140

H2O 1.7096 ·10–8 1.1146 0 0

3.4 Reaction Kinetics and Equilibrium

In previous work, a suitable kinetic equation for the RWGS

Ns
X yi mi reaction on Fe-Mo/Al2O3 catalyst was determined (Eq. (30))
m¼ PNs qffiffiffiffiffiffiffiffiffiffiffiffiffiffi (27)
[37].
i¼1 j¼1 yj Mj =Mi
PCO PH2 O
in which m is the viscosity of the gas-phase mixture and mi is kðPCO2  Þ
K e PH2
the gas-phase viscosity of species i. rCO2 ¼ P
(30)
1þ KH2 O H2 O
P H2

3.3.4 Gas Thermal Conductivity in which k is the rate constant, which according to the Arrhe-
nius equation is given by Eq. (31):
The gas-phase thermal conductivity of species is estimated with   117 432 J mol1
the DIPPR equation used in Aspen Plus (Aspen, 2007), shown k ¼ 11 101:2 mol s1 g1
cat: bar
1
expð Þ
in Tab. 5 [38]. Thermal conductivity of the mixture is calcu- Rg T
lated by the Wilke’s approach (Eq. (28)) [46]. (31)
Ns
X yi li KH2 O is the adsorption coefficient of H2O, which is related to
lf ¼ PNs (28) the temperature through Eq. (32):
i¼1 j¼1 yj Aij

The binary interaction parameter (Aij) can be found based 51 979 J mol1
KH2 O ¼ ð96 808Þexpð Þ (32)
on the Chapman-Enskog model (Eq. (29)) [47]. Rg T

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 603

Ke is the equilibrium constant, which for the 1

RWGS reaction is given by Eq. (33): 5g
0.9
  0.8g
ln 1=Ke ¼ 12:11 þ 5319 T 1 þ 1:012 lnT 0.8
0.7 10g
þ 1:144 · 104T
0.6

V-rao
(33)
0.5
0.4
3.5 Integration Strategy 0.3
The model was implemented and solved in 0.2
the equation-based process modeling platform 0.1
gPROMS 4.0. The partial differential equations were 0
discretized along the fixed-bed axial direction by us- 30 40 50 60 70 80
ing backward finite differences and along the bed Mol% of H2 in feed
radial direction by using centered finite differences.
Figure 2. The V-ratio distribution versus temperature.

1
4 Results and Discussion
0.9 5g

4.1 Operating Conditions 0.8 0.8g

0.7
The proposed model is applied to predict three sets 10g
of experimental data. These data were collected 0.6
from the fixed bed reactor which was described in 0.5
V-rao

Sect. 2.2. All experiments were carried out at atmo- 0.4

spheric pressure and other operating conditions
0.3
are summarized in Tab. 6.
0.2
0.1
4.2 Thermodynamic Analysis
0
The potential direction of the reaction from a ther- 300 400 500 600 700 800
modynamic point of view can be determined ac- Temperature (K)
cording to the V value given by Eq. (34).
Figure 3. The V-ratio distribution versus mol % of H2 at 600 C.
Q
ð i Pivi Þ
V¼ (34)
Ke

If the V value is less than one, the reaction is expected to 4.3 Validation of the Proposed Reactor Model
proceed forward to the right; otherwise the reaction shows a
tendency to approach equilibrium or proceeds backward [48]. For each of the experimental conditions, simulations were per-
The analysis of the product gas composition in the RWGS formed by using mesh sizes of Nz = 28 and Nr = 10. The parity
experiments under all conditions (all temperature ranges, var- plot for the outlet pressure of CO2 is shown in Fig. 4. By com-
ious volumetric flow rates, and various masses of catalyst load- paring experimental data with model predictions, the reactor
ing in the reactor), presented in Figs. 2 and 3, have V values model has an average error of 2.7 % and an average absolute
smaller than one. Thus, the reaction proceeds to the right. error of 5.6 %. This result illustrates good agreement between
experimental and predicted data, which
Table 6. Operating conditions for experiments. proves that the proposed model has the capa-
bility to describe the performance of the
Data set 1 Data set 2 Data set 3 fixed-bed reactor for the catalytic RWGS
Inlet temperature [C] 400, 500, 600, 700 400, 500, 600, 700 500, 600, 700 reaction. These errors may be due to the fol-
lowing reasons:
H2 in feed [mol %] 38, 50, 58, 63, 70 38, 50, 58, 63, 70 38, 50, 58 – The model is homogeneous, so mass and
Bed length [mm] 0.9 5.6 11.2 heat-transfer resistances between the sur-
face catalyst and bulk gas phase and within
Bed porosity 0.38 0.38 0.38
the catalyst particle (external and internal
–1
GHSV [mL h gcat. ] –1
30 000 80 000 18 000 limitations) are neglected.
– The use of gas-solid mixture properties
Mass of catalyst loading [g] 0.8 5 10
(porous media properties).

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 604

80 4.4.2 Effect of Feed Composition

70
Predicted outlet pressure of CO2 (kPa)

Fig. 6 shows the effect of feed composition (in format of

60 changing H2/CO2 molar ratio in the feed) on CO2 conversion.
+10% The simulated conditions are T = 873 K, P = 1 bar, and
50 GHSV = 80 000 mL h–1gcat.–1 (22.2 cm3s–1gcat.–1). Fig. 6 illustrates
40 that the increase in H2/CO2 leads to an increase in CO2 conver-
−10% sion. This rise in conversion before H2/CO2 = 1 has a steep
30 slope and after that the slope becomes gentler. Due to less
access to H2 than to CO2, it can be concluded that the H2/CO2
20
ratio equal to one is the optimized ratio for the RWGS reac-
10 tion.
0
0 20 40 60 80
4.4.3 Effect of Feed Gas Temperature
Experimental outlet pressure of CO2 (kPa)

Figure 4. Parity plot for the outlet pressure of CO2 with an inter- Fig. 7 shows the effect of feed gas temperature on temperature
val of ±10 %. drop along the reactor. The simulated conditions are P = 1 atm,
H2/CO2 = 1, and GHSV = 80 000 mL h–1gcat.–1 (22.2 cm3s–1gcat.–1).
Fig. 7 shows that, at high feed temperature, the temperature

– The assumption of ideal gas behavior. a) 30

– The use of empirical equations to calculate phys- 20 000
ical properties of gases and solid. 40 000
25
60 000
80 000
4.4 Effect of the Main Operational 20
Parameters
XCO2%

15
The effect of parameters such as feed gas tempera-
ture, feed composition, and space velocity on the
10
fixed-bed reactor behavior are presented in the fol-
lowing sections.
5

4.4.1 Effect of Space Velocity 0

0 1 2 3 4 5
z(mm)
For the system with a fixed volume and pressure,
the reactor space velocity is equal to the volumetric
b) 874
flow rate of raw materials per unit mass of catalyst.
The simulated conditions are as follows: T = 873 K,
P = 1 atm, H2/CO2 = 1 (molar ratio). Correspond- 872
ingly, the simulated results are shown in Fig. 5.
Fig. 5 a shows that the CO2 conversion decreases 870
Temperature (K)

with increasing space velocity. This is because, with

increasing space velocity, the residence time of the 868
reactants on the catalyst becomes shorter, so there
is less opportunity for the reaction, and therefore,
conversion remains low. 866 20 000

Fig. 5 b shows that, with increasing space veloci- 40 000

ty, the temperature drop in the reactor increases. In 864 60 000
high space velocities, the conversion is low, but im- 80 000
plies a large inlet gas volume flux and a large 862
amount of raw materials, resulting in much greater 0 1 2 z(mm) 3 4 5
reaction heat consumption and a greater tempera-
ture drop. Figure 5. (a) The effect of space velocity on the conversion distribution of CO2
along the axial direction. (b) The effect of space velocity on the temperature dis-
tribution along the axial direction of the reactor (T = 6000 C, P = 1 atm, H2/CO2 = 1,
GHSV = 20 000–80 000 mL h–1gcat.–1).

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 605

25

20

15
XCO2%

10

0
0.50 1.00 1.50 2.00 2.50
H2/CO2

Figure 6. The effect of feed composition on CO2 conversion

(T = 6000 C, P = 1 atm, GHSV = 80 000 mL h–1gcat.–1).

16
673 Figure 8. CO2 concentration profile (T = 6000 C, H2/CO2 = 1 and
14 GHSV = 80 000 mL h–1gcat.–1).
773
12
873
the contents of the reactor (gases and catalyst bed) and there is
Temperature drop (K)

10
973 not a sufficient heat transfer rate from the wall into the reactor,
8 so the temperature drops in the reactor. The maximum tem-
perature drop is about 10 C along the axial and radial direc-
6
tions.
4 The pressure profile is presented in Fig. 10. It is assumed that
2
the pressure drop in the radial direction is negligible. Also, it is
assumed that the pressure drop in the axial direction is domi-
0 nated by friction. Fig. 10 shows that the pressure drop over the
0 1 2 3 4 5 bed is about 140 Pa.
z(mm)
Figure 7. The effect of feed temperature on temperature drop
(P = 1 atm, H2/CO2 = 1, GHSV = 80 000 mL h–1gcat.–1).

drop in the reactor increases. This is because, with increasing

feed temperature, the rate of the reaction increases and, since
the RWGS reaction is an endothermic reaction, more heat is
consumed, so the temperature drop increases.

4.5 Concentration, Temperature, and Pressure

Profiles Predicted by the Model

The CO2 concentration profile in the axial and radial directions

of the bed is presented in Fig. 8. The concentration of CO2 is
reduced along the axial direction because CO2 is consumed in
the reaction. In the initial section of the bed, the decrease in
CO2 concentration is high due to a high reaction rate, but upon
Figure 9. Temperature profile (T = 6000 C, H2/CO2 = 1 and
consumption of reactants and a decrease in the reaction rate GHSV = 80 000 mL h–1gcat.–1).
along the bed, the profile becomes smoother. The concentra-
tion change in the radial direction of the bed is very small. This
can be explained because no radial velocity exists in the bed 5 Conclusions
and the mass dispersion coefficient in the radial direction is
very small. For the RWGS reaction, the Fe-Mo/Al2O3 catalyst was synthe-
The temperature profile in the axial and radial directions of sized through an impregnation method. Testing of the catalyst
the bed is presented in Fig. 9. The RWGS reaction is endother- was conducted in a fixed-bed reactor. A steady-state, two-
mic and requires heat to progress. This heat is obtained from dimensional, pseudo-homogeneous reactor model was pro-

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 606

Ke [–] equilibrium constant

Ki [–] adsorption coefficient of species i
L [m] bed length
Mi [kg kmol–1] molecular weight of species i
Ns [–] number of species
P [bar] reactor pressure
p [–] tortuosity parameter
Pi [bar] partial pressure of species i
q† [kJ m–2s–1] heat flux
r [m] reactor radial element
r(C,T) [mol s–1gcat.–1] rate of reaction
Rg [kJ kmol–1K–1] universal gas constant
Rep [–] Reynolds number based on
particle diameter
T [K] temperature
Sg [cm2g–1] total surface of porous material
us [m s–1] superficial velocity
us,0 [m s–1] initial superficial velocity
Figure 10. Pressure profile T = 6000 C, H2/CO2 = 1 and GHSV = V [–] thermodynamic ratio
80 000 mL h–1gcat.–1). yi [–] molar fraction of species i
z [m] reactor axial element
posed for the RWGS reaction in the fixed-bed reactor. In the
reactor model, a new redox kinetic model was used to describe Greek letters
the RWGS reaction rate. The model was implemented and eb [–] porosity of the bed
solved in the equation-based process modeling platform leff [kJ s–1m–1K–1] effective thermal conductivity
gPROMS 4.0. Three sets of experimental data were used to vali- lf [kJ s–1m–1K–1] thermal conductivity of the gas
date the proposed reactor model. The parity plot for CO2 outlet mixture
pressure showed that the reactor model had an average error of li [kJ s–1m–1K–1] thermal conductivity of species i
2.7 % and an average absolute error of 5.6 %. These results ls [kJ s–1m–1K–1] thermal conductivity of the
illustrate good agreement between experimental data and mod- catalyst
el-predicted data, and proves that the proposed model has the ls,i [kJ s–1m–1K–1] thermal conductivity of
capability of describing the performance of the fixed-bed reac- component i in the catalyst
tor for the catalytic RWGS reaction. m [N s m–2] viscosity of the gas mixture
mi [N s m–2] viscosity of species i
The authors have declared no conflict of interest. mi [–] stoichiometric coefficient of
species i
r [kg m–3] gas density
Symbols used rb [kg m–3] bulk density
rs [kg m–3] catalyst density
Aij [–] binary interaction parameter t [–] tortuosity
Ci [kmol m–3] concentration of gas species i
Cp,f [kJ kmol–1] heat capacity of the gas mixture Abbreviations
Cp,i [kJ kmol–1] heat capacity of gas species i
Ct [kmol m–3] total gas concentration GHSV gas hourly space velocity
Deff,i [m2s–1] effective diffusion coefficient of RWGS reverse water-gas shift
species i
Dij [m2s–1] binary gas-phase diffusivity
DK,i [m2s–1] Knudsen diffusion coefficient of References
species i
Dm,i [m2s–1] molecular diffusivity of species i [1] X. Xiaoding, J. A. Moulijn, Energy Fuels 1996, 10 (2),
in the mixture 305–325. DOI: 10.1021/ef9501511
Dp [m] diameter of the catalyst particle [2] F. N. Ridha, V. Manovic, Y. Wu, A. Macchi, E. J. Anthony,
Dt [m] diameter of the reactor tube Int. J. Greenhouse Gas Control 2013, 16, 21–28. DOI:
Ft [kmol s–1] total molar flow rate of gas 10.1016/j.ijggc.2013.02.026
fv,i [–] volume fraction of solid phase i [3] K. Volkart, C. Bauer, C. Boulet, Int. J. Greenhouse Gas
G [kmol m–2s–1] molar flux of the gas Control 2013, 16, 91–106. DOI: 10.1016/j.ijggc.2013.03.003
[4] B. Li, Y. Duan, D. Luebke, B. Morreale, Appl. Energy 2013,
DHr [kJ kmol–1] heat of reaction
102, 1439–1447. DOI: 10.1016/j.apenergy.2012.09.009
k [mol s–1gcat.–1bar–1] reaction rate constant

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
 15214125, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ceat.201600220 by <Shibboleth>[email protected], Wiley Online Library on [14/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Research Article 607

[5] M. Czaun et al., J. CO2 Util. 2013, 1, 1–7. DOI: 10.1016/ [26] C. S. Chen, W. H. Cheng, S. S. Lin, Appl. Catal., A 2004,
j.jcou.2013.03.007 257 (1), 97–106. DOI: 10.1016/S0926-860X(03)00637-9
[6] M. Mikkelsen, M. Jorgensen, F. C. Krebs, Energy Environ. [27] R. A. Koeppel, A. Baiker, A. Wokaun, Appl. Catal., A 1992,
Sci. 2010, 3 (1), 43–81. DOI: 10.1039/B912904A 84 (1), 77–79. DOI: 10.1016/0926-860X(92)80340-I
[7] H. Yang et al., J. Environ. Sci. 2008, 20 (1), 14–27. DOI: [28] D. J. Pettigrew, D. L. Trimm, N. W. Cant, Catal. Lett. 1994,
10.1016/S1001-0742(08)60002-9 28 (2), 313–315. DOI: 10.1007/bf00806061
[8] S. K. Hoekman, A. Broch, C. Robbins, R. Purcell, Int. [29] S. R. Yan, K. W. Jun, J. S. Hong, M. J. Choi, K. W. Lee, Appl.
J. Greenhouse Gas Control 2010, 4 (1), 44–50. DOI: 10.1016/ Catal., A 2000, 194, 63–70. DOI: 10.1016/S0926-
j.ijggc.2009.09.012 860X(99)00354-3
[9] M. Soltanieh, K. Mahmoodi Azar, M. Saber, Int. J. Green- [30] F. J. Perez-Alonso, M. Ojeda, T. Herranz, S. Rojas, J. M. Gon-
house Gas Control 2012, 7, 145–152. DOI: 10.1016/ zález-Carballo, P. Terreros, J. L. G. Fierro, Catal. Commun.
j.ijggc.2012.01.008 2008, 9 (9), 1945–1948. DOI: 10.1016/j.catcom.2008.03.024
[10] A. Talebian-Kiakalaieh, N. A. S. Amin, H. Mazaheri, Appl. [31] S. Saeidi, N. A. Saidina Amin, M. R. Rahimpour, J. CO2 Util.
Energy 2013, 104, 683–710. DOI: 10.1016/j.apenergy. 2014, 5, 66–81. DOI: 10.1016/j.jcou.2013.12.005
2012.11.061 [32] S. I. Pereira Duarte, G. F. Baneto, N. O. Lemcoff, Chem. Eng.
[11] M. Aresta, A. Dibenedetto, A. Angelini, J. CO2 Util. 2013, Sci. 1984, 39, 1025–1031. DOI: 10.1016/j.jcou.2013.12.005
3–4, 65–73. DOI: 10.1016/j.jcou.2013.08.001 [33] Q. S. Lue, Z. X. Zhang, J. L. Zhou, J. Nat. Gas Chem. 1999,
[12] A. Ramı́rez, S. Hagedoorn, L. Kramers, T. Wildenborg, 8 (2), 137–150. DOI: 10.1016/S0140-6701(00)87621-3
C. Hendriks, Int. J. Greenhouse Gas Control 2010, 4 (2), [34] D. Schlereth, O. Hinrichsen, Chem. Eng. Res. Des. 2014,
367–380. DOI: 10.1016/j.ijggc.2009.10.015 92 (4), 702–712. DOI: 10.1016/j.cherd.2013.11.014
[13] H. He et al., Energy Environ. Sci. 2013, 6 (2), 488–493. DOI: [35] A. Gharibi Kharaji, A. Shariati, M. A. Takassi, Chin. J. Chem.
10.1039/C2EE24139K Eng. 2013, 21 (9), 1007–1014. DOI: 10.1016/S1004-9541
[14] J. Wang, V. Manovic, Y. Wu, E. J. Anthony, Appl. Energy (13)60573-X
2010, 87 (4), 1453–1458. DOI: 10.1016/j.apenergy. [36] Y.-P. Zhu, S. Tu, Z.-H. Luo, Chem. Eng. Res. Des. 2012,
2009.08.010 90 (9), 1361–1371. DOI: 10.1016/j.cherd.2011.12.004
[15] A. J. Hunt, E. H. K. Sin, R. Marriott, J. H. Clark, [37] F. Ghodoosi, M.Sc. Thesis, Petroleum University of Technol-
ChemSusChem 2010, 3 (3), 306–322. DOI: 10.1002/cssc. ogy, Ahwaz 2015.
200900169 [38] T. A. Adams, P. Barton, Int. J. Hydrogen Energy 2009,
[16] V. Vandeginste, K. Piessens, Int. J. Greenhouse Gas Control 34 (21), 8877–8891. DOI: 10.1016/j.ijhydene.2009.08.045
2008, 2 (4), 571–581. DOI: 10.1016/j.ijggc.2008.02.001 [39] L. Han, Z. Zhou, G. M. Bollas, Chem. Eng. Sci. 2013, 104,
[17] M. C. M. Nasvi, P. G. Ranjith, J. Sanjayan, Appl. Energy 233–249. DOI: 10.1016/j.ces.2013.09.021
2013, 102, 1391–1398. DOI: 10.1016/j.apenergy.2012.09.004 [40] J. A. Francesconi, M. C. Mussati, P. A. Aguirre, J. Power
[18] K. D. Ramachandriya, D. K. Kundiyana, M. R. Wilkins, J. B. Sources 2007, 173 (1), 467–477. DOI: 10.1016/j.jpowsour.
Terrill, H. K. Atiyeh, R. L. Huhnke, Appl. Energy 2013, 112, 2007.04.048
289–299. DOI: 10.1016/j.apenergy.2013.06.017 [41] E. Donskoi, D. L. S. McElwain, Metall. Mater. Trans. B 2003,
[19] W.-H. Chen, B.-J. Lin, Appl. Energy 2013, 101, 551–559. 34 (1), 93–102. DOI: 10.1007/s11663-003-0059-0
DOI: 10.1016/j.apenergy.2012.06.030 [42] P. Grathwohl, Diffusion in Natural Porous Media: Conta-
[20] T. Riedel, M. Claeys, H. Schulz, G. Schaub, S. S. Nam, K. W. minant Transport, Sorption/Desorption and Dissolution
Jun, M. J. Choi, G. Kishan, K. W. Lee, Appl. Catal., A 1999, Kinetics, Springer, New York 1998.
186 (1–2), 201–213. DOI: 10.1016/S0926-860X(99)00173-8 [43] M. Matyka, A. Khalili, Z. Koza, Phys. Rev. E 2008, 78 (2),
[21] S. S. Nam, H. Kim, G. Kishan, M. J. Choi, K. W. Lee, Appl. 026306. DOI: 10.1103/PhysRevE.78.026306
Catal., A 1999, 179 (1–2), 155–163. DOI: 10.1016/S0926- [44] C. N. Satterfield, Mass Transfer in Heterogeneous Catalysis,
860X(98)00322-6 MIT Press, Cambridge 1970.
[22] S. W. Park, O. S. Joo, K. D. Jung, H. Kim, S. H. Han, Appl. [45] R. H. Perry, D. W. Green, Perry’s Chemical Engineers’ Hand-
Catal., A 2001, 211 (1), 81–90. DOI: 10.1016/S0926- book, 7th ed., McGraw-Hill, New York 1997.
860X(00)00840-1 [46] B. E. Poling, J. M. Prausnitz, J. P. O’Connel, The Properties of
[23] M. C. Roman-Martı́nez, D. Cazorla-Amorós, A. Linares- Gases and Liquids, 5th ed., McGraw-Hill, New York 2001.
Solano, C. Salinas-Martı́nez de Lecea, Appl. Catal., A 1994, [47] J. O. Hirschfelder, C. F. Curtiss, R. B. Bird, Molecular Theory
116 (1–2), 187–204. DOI: 10.1016/0926-860X(94)80289-0 of Gases and Liquids, Wiley, New York 1955.
[24] J. Kaspar, M. Graziani, A. H. Rahman, A. Trovarelli, E. J. S. [48] M. H. Halabi, M. H. J. M. de Croon, J. van der Schaaf, P. D.
Vichi, E. C. da Silva, Appl. Catal., A 1994, 117 (2), 125–137. Cobden, J. C. Schouten, Appl. Catal., A 2010, 389 (1–2),
DOI: 10.1016/0926-860X(94)85093-3 80–91. DOI: 10.1016/j.apcata.2010.09.005
[25] M. S. Spencer, Catal., Lett. 1995, 32 (1), 9–13. DOI: 10.1007/
BF00806097

Chem. Eng. Technol. 2017, 40, No. 3, 598–607 ª 2017 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com