i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/he

A numerical performance study of a fixed-bed

reactor for methanol synthesis by CO2
hydrogenation

ri a, Bulut Tekgül a, Erkki Laurila a,

Daulet Izbassarov a,*, Judit Nya
Tanja Kallio , Annukka Santasalo-Aarnio a, Ossi Kaario a, Ville Vuorinen a
b

a
Department of Mechanical Engineering, Aalto University, FI-00076, Espoo, Finland
b
Department of Chemistry and Materials Science, Aalto University, FI-00076, Espoo, Finland

highlights

A novel computational tool is presented for CO2 and H2-based methanol synthesis.

Aspen Plus (1D) and OpenFOAM (1D/2D) fixed-bed reactor simulations are performed.

Maximum methanol yield is found to be limited by 1D solution.

The isothermal random and staggered configurations outperformed the in-line system.

article info abstract

Article history: Synthetic fuels are needed to replace their fossil counterparts for clean transport. Presently,
Received 12 September 2020 their production is still inefficient and costly. To enhance the process of methanol production
Received in revised form from CO2 and H2 and reduce its cost, a particle-resolved numerical simulation tool is pre-
5 January 2021 sented. A global surface reaction model based on the Langmuir-Hinshelwood-Hougen-
Accepted 4 February 2021 Watson kinetics is utilized. The approach is first validated against standard benchmark
Available online 6 March 2021 problems for non-reacting and reacting cases. Next, the method is applied to study the
performance of methanol production in a 2D fixed-bed reactor under a range of parameters. It
Keywords: is found that methanol yield enhances with pressure, catalyst loading, reactant ratio, and
CFD packing density. The yield diminishes with temperature at adiabatic conditions, while it
OpenFOAM shows non-monotonic change for the studied isothermal cases. Overall, the staggered and
Methanol synthesis the random catalyst configurations are found to outperform the in-line system.
CO2 hydrogenation © 2021 The Author(s). Published by Elsevier Ltd on behalf of Hydrogen Energy Publications
Catalyst particles LLC. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

electricity storage potential [8]. CO2 can be captured and uti-

Introduction lized from industrial emissions [38] or directly from air [7].
Such a technology is currently considered crucial to achieve
Methanol production via CO2 hydrogenation is a highly CO2 neutrality or even CO2 negativity [15]. CO2 can be used to
contemporary topic due to its CO2 mitigation and excess produce building materials, fuels and chemicals. Most of the

* Corresponding author.
E-mail address: [email protected] (D. Izbassarov).
https://doi.org/10.1016/j.ijhydene.2021.02.031
0360-3199/© 2021 The Author(s). Published by Elsevier Ltd on behalf of Hydrogen Energy Publications LLC. This is an open access article under the CC BY-NC-
ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
15636 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

CO2-based synthetic fuels are produced by hydrogenation obtained by both models are similar when the operating
which provides their energy content [16]. However, to achieve conditions are in the conventional region. Note that most of
a minimum carbon neutrality of these compounds, hydrogen the previous contributions address steady-state and pseudo-
should be of non-fossil origin. One of the most established H2 homogeneous models while only few works considered dy-
production processes is biomass gasification [6,9]. Moreover, namic modeling [29]. Comprehensive summary of the 1D
water electrolysis with renewable electricity is another models can be found in the review by Bozzano and Manenti
emergent technology for H2 production [4]. Hydrogen is a [8]. Several studies extended the 1D models to 2D [3,36,43].
valuable product for the chemical industry, and a long-term Solsvik and Jakobsen [43] studied the effect of various closures
storage material for renewable electricity [2]. However, utili- for mass diffusion on the pellet and the reactor levels. They
zation of hydrogen as a transportation fuel [14,44] is also found that the difference between various closures were
known to pose certain challenges including unfavorable minor on both levels. Petera et al. [36] simulated different
storage [17,33], transportation and fueling conditions [25]. phenomena occurring in the catalyst and the reactor. They
Meanwhile, methanol is a versatile compound that can serve studied the mechanism of hot spot formation due to inci-
as a raw material for the chemical and petrochemical in- dental reduction in the size of the catalyst. Arab et al. [3]
dustries [34]. Furthermore, its role as pure fuel or fuel additive compared the performance of a conventional fixed-bed
is increasing [21], and its growing recognition in the marine reactor to a monolithic one. It was found that both reactors
industry makes methanol production from CO2 and hydrogen performed similarly at low space velocity (SV), while the latter
an intriguing topic [40]. one outperformed at high velocities.
Currently, methanol is almost solely produced from natu- Another approach is to use computational fluid dynamics
ral gas, except for China where inexpensive coal is the main (CFD) simulations. Within a CFD framework, a fixed-bed can
raw material [8]. Methanol synthesis from CO2 and H2 works be modeled by the effective continuum approach or by an
similarly to natural gas-based synthesis meaning that the actual particle-resolved simulation. In the first method, the
same reactions take place at similar conditions, and is usually full-size reactor can be taken into account, while the latter one
carried out in a fixed-bed reactor [37]. A summary of experi- enables only reduced-length reactor simulations. However,
ments studying the rate equations of kinetic models under the latter approach provides more detailed information
different conditions is presented by Bozzano and Manenti [8]. regarding local transport phenomena inside the bed by taking
The developed Langmuir-Hinshelwood-Hougen-Watson into account the actual geometry of the catalyst particles.
(LHHW) type kinetic models often differ in the source of car- Moreover, it is found that at small tube-to-particle diameter
bon provided for the formation of methanol. This is due to the ratios, the effects of the inhomogeneous structure become
uncertainty in the origin of the carbon molecule in the important and the plug flow model predictions become
methanol compound as some consider either only carbon questionable. Thus, there is a growing interest in the particle-
monoxide [45,49] or carbon dioxide [41,48] providing the car- resolved simulations [19].
bon molecule, while others consider both of them [13,26]. The Despite advances in coupled CFD simulations with reac-
most frequently used kinetic models were proposed by Graaf tion kinetics [10,19], CFD studies on methanol synthesis in
et al. [13] and Vanden Bussche and Froment [48]. fixed-bed reactors are scarce. Mirvakili et al. [30] investigated
The performance of the methanol synthesis reactor the effect of flow mal-distribution on the performance of an
strongly depends on the fluid flow and its interaction with the industrial reactor. Redondo et al. [39] studied the performance
catalyst material as shown in Fig. 1. Several approaches have of various tubular and tube-cooled reactors. Both of the
been proposed to study the methanol synthesis in a fixed-bed studies utilized the effective continuum approach. Recently,
reactor. One common numerical model is the one- the particle-resolved approach was used by Karthik and Buwa
dimensional approach, i.e. the plug flow reactor model. 1D [22] where they studied the selection of optimal geometry of
models can be classified into two types as pseudo- the catalyst. In the aforementioned studies, the conventional
homogeneous and heterogeneous models. In the first model, LHHW kinetics was coupled with CFD. It is evident that there
the fixed-bed is treated as a pseudo-continuum. On the other is a gap in understanding the effects of flow dynamics on
hand the heterogeneous model treats both the gas and pellet methanol synthesis reactor. Thus, in the present study,
phases explicitly. Manenti et al. [28] concluded that results particle-resolved CFD simulations coupled with LHHW

Fig. 1 e Schematic of the methanol synthesis reactor used in the present work. The contours represent velocity magnitude
with the scale ranging from 0 (white) to 7.4Ui (black) where Ui is the average inlet velocity.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15637

kinetics are performed to study the performance of the previous works [20,46], the flux limited Gamma scheme of
methanol synthesis reactor under various conditions. The Jasak et al. [18] is used to discretize the convective terms,
main objectives of the present work are to 1) present a novel while the diffusion terms are approximated using central
computational tool for methanol synthesis, 2) validate and differences. Finally, a second order implicit scheme is utilized
verify the tool with available numerical and experimental for temporal discretization.
data, 3) study the performance of the fixed-bed reactor for a
range of parameters via the developed computational tool. Kinetic model

In this paper, formation of methanol is considered both from

Mathematical formulation CO2 and CO. The popular model of Graaf et al. [13] is updated
with equilibrium constants from Lim et al. [26] and experi-
Governing equations and discretization mental data of An et al. [1]. Methanol production occurs by
hydrogenation of CO2 and CO with a Cu/Zn/Al/Zr catalyst [8].
The reacting flow is described by the compressible Navier- The mechanism is based on three main reactions:
Stokes equations. The continuity, momentum, species and
ðR1Þ CO þ 2H2 #CH3 OH; DH298K ¼ 90:2kJ=mol (7)
energy equations are given by:

vr ðR2Þ CO2 þ 3H2 #CH3 OH þ H2 O; DH298K ¼ 48:8kJ=mol

þ V,ðruÞ ¼ 0; (1)
vt (8)


vru   2
þ V,ðruuÞ ¼ Vp þ V, m Vu þ VuT  mV,uI ; (2) ðR3Þ CO2 þ H2 #CO þ H2 O; DH298K ¼ þ41:3kJ=mol (9)
vt 3
Formation of by-products, e.g. dimethyl ether, is not
vrYk considered here due to their negligible concentration [47].
þ V,ðruYk Þ ¼ V,ðrDVYk Þ þ u_ k ; k ¼ 1; …; ns ; (3)
vt Following Kiss et al. [23], the rate expression can be written
in a generic form.
 
vrht vp l
þ V,ðruht Þ ¼ þ V, Vhs þ u_ h ; (4) AB
vt vt cp r¼ ; (10)
C
where u, p, Yk, hs, r, m, l, cp, u_ k , u_ h are the velocity, the pressure,
 
the mass fraction of specie k, the sensible enthalpy, the den- Ea
A ¼ aTn exp ; (11)
sity, the viscosity, the thermal conductivity, the heat capacity, RT
the reaction rate of species k and the heat release rate,
Y Y b
respectively. Above, ht is the total enthalpy defined as ht ¼ hs þ B ¼ B1
b
f i i  B2 fj j ; (12)
0.5|u|2, I is the identity tensor and ns is the number of species.
For all species, the diffusion coefficient is D ¼ l/(rcp) as Lewis  hY c i m
number is considered to be equal to unity. Note that the spe- C ¼ Ci f jj (13)
cies formation/consumption rate (u_ k ) and the heat source (u_ h )
where A, B, C are the kinetic term, the driving force and the
are defined only on the catalytic walls and can be written as:
adsorption term, respectively. The rate equations for R1, R2
Xnf
Xnr
Ai ajk rij and R3 are:
u_ k ¼ rcat MWk ; (5)
i¼1 j¼1
Vi h . i
1:5 0:5
k1 KCO fCO f H2  fCH3 OH f H2 Keq1
r1 ¼  h  . i; (14)
0:5
X
ns 1 þ KCO fCO þ KCO2 fCO2 f H2 þ KH2 O K0:5 H2 fH2 O
0
u_ h ¼ Dhf ;k u_ k ; (6)
k¼1
h . i
1:5 1:5
where rcat is the catalyst density, MWk is the molecular weight k2 KCO2 fCO2 f H2  fCH3 OH fH2 O f H2 Keq2
r2 ¼  h  . i;
0:5
0
and hf ;k is the enthalpy of formation of species k, Ai is the area 1 þ KCO fCO þ KCO2 fCO2 f H2 þ KH2 O K0:5 H2 fH2 O

of face i of the catalytic wall, Vi is the volume of the cell (15)

adjacent to face i, ajk is the stoichiometric coefficient of
component k and rij is the reaction rate of reaction j. The h . i
k3 KCO2 fCO2 fH2  fH2 O fCO Keq3
catalyst density is defined as rcat ¼ mcat/Ageometric, where mcat r3 ¼  h  . i;
0:5
and Ageometric are mass and geometric area of the catalyst, 1 þ KCO fCO þ KCO2 fCO2 f H2 þ KH2 O K0:5 H2 fH2 O

respectively. Note that rcat is used to redefine the reaction rate (16)
in terms of a molar flux, which was proposed earlier by Kuroki
et al. [24]. Above, nr and nf represent the number of reactions where f corresponds to fugacity [Pa] and the reaction rate r is
and faces of a catalytic wall. Finally, the density of the mixture in [mol/(gcat s)]. All coefficients are in the form of Eq. (11) with
is calculated using ideal gas law. n ¼ 0. All the constants are given in Table 1. Note that the ideal
The flow equations (Eqs. (1)e(4)) are solved by the pressure gas assumption is used in the present work, thus partial
implicit splitting of operators (PISO) algorithm within the pressure is used instead of fugacity. To facilitate compre-
finite-volume OpenFOAM 6 [51] framework. Similar to our hensive validation of the method, the LHHW kinetic model is
15638 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

Table 1 e Constants for the reactions R1, R2 and R3.

Coefficient Expression Unit
k1 4.0638  106 exp(-11695/RT) kmol/kgcat, s , Pa
k2 1.5188  1033 exp(-266010/RT) kmol/kgcat, s , Pa
k3 9.0421  108exp(-112860/RT) kmol/kgcat, s , Pa0.5
KCO 8.3965  1011 exp(118270/RT) Pa1
KCO2 1.7214  1010exp(81287/RT) Pa1
KCO =Keq1 3.5408  1012exp(19832/RT) Pa
KCO2 =Keq2 2.5813  1010exp(26788/RT) Pa
KCO2 =Keq3 6.1221  1013exp(125226/RT) Pa1
KH2 O =K0:5
H2 4.3676  1012exp(115080/RT) Pa0.5

implemented both in the Aspen Plus® and the OpenFOAM. Flow over single cylinder
The details of implementation in the Aspen Plus® can be Next, the flow solver is validated for a non-reacting heat
found in Ref. [32] based on the work of Kiss et al. [23] and An transfer problem. Accordingly, 2D simulations of laminar flow
et al. [1]. Here, the following commonly used simplifications over a stationary isothermal cylinder are performed. The
are assumed: (1) only surface reaction are considered, (2) mesh is generated using snappyHexMesh utility, a built-in
catalyst deactivation is insignificant, (3) side reactions are utility within the OpenFOAM. The base mesh consists of
negligible and (4) diffusion inside the pellets is instantaneous. hexahedral cells, while near the cylinder surface a body con-
Moreover, the catalytically active surface area is not taken forming boundary layer mesh is used. The size of the
into account, but instead the geometric one is used. computational domain is 45d  20d, where d is the diameter of
the cylinder. The cylinder is positioned 15d away from the
inlet. Air flow (N2:O2:Ar ¼ 78:21:1%, v/v) with uniform velocity
Results and discussion (U) and constant temperature (Ti ¼ 293.15 K) is imposed at the
inlet. The outlet pressure is fixed at p ¼ 101,325 Pa, while the
Validation cylinder surface temperature is kept constant at Tw ¼ 303.15 K.
Symmetry boundary conditions are applied on top and bottom
Plug flow reactor boundaries.
Firstly, the LHHW kinetic model implementation is validated. First, a grid independence study is performed at Re ¼ rUD/
The validation is done with a plug flow reactor against data m ¼ 100 on three different meshes: M0, M1, M2 with the grid
available in the literature. Within the plug flow reactor model, size in the boundary layer 0.01d, 0.0068d, 0.0046d, respec-
the flow is assumed to be steady, frictionless, and one- tively. A summary of the results (drag and lift coefficients,
dimensional. Diffusion in the radial and axial directions is Strouhal number and average Nusselt number) are presented
neglected. Basically, uniform distribution of pellets and con- in Table 3. The drag coefficient is defined as Cd ¼ 2Fd/rU2d
stant velocity inside the reactor is assumed. Kiss et al. [23] and where Fd is the drag force, the lift coefficient as Cl ¼ 2Fl/rU2d
An et al. [1] studied the problem numerically and experimen- where Fl is the lift force, the Strouhal number as St ¼ fd/U,
tally, respectively. The reactor is assumed to be isothermal and where f is vortex shedding frequency, the Nusselt number as
characterized by radius r ¼ 6 mm and length L ¼ 500 mm. The Nu ¼ qd/k(Tw  Ti), where q is the heat flux on the surface of
reactor is fed with a composition of inlet mole fractions cylinder. Note that CNuD denotes surface- and time-averaged
R ¼ H2:CO2 ¼ 3. Simulations are performed for temperature and Nusselt number. The M1 is chosen for the simulations.
space velocity sweeps at fixed pressure p ¼ 5 MPa. In the T Next, simulations are performed for different Reynolds
sweep, the space velocity is kept constant at SV ¼ 6 m3/kgcath, numbers, i.e. Re ¼ 10, 20, 40 and 100. Comparison of the local
while the temperature is varied in the range T ¼ 483e543 K. In pressure coefficient Cp ¼ 2(p  pi)/rU2 and Nu with the data
the SV sweep, the temperature is fixed at T ¼ 523 K, while the from the literature are shown in Fig. 2. As can be seen, there is
space velocity is changed between SV ¼ 1e10 m3/kgcath. CO2 a good agreement with the data in the literature.
conversion (XCO2 ) and CH3OH yield (YCH3OH ) are defined as
Flow over in-line and staggered tube banks
YCO2;i  YCO2;o YCH3 OH;o MWCO2
XCO2 ¼ ; YCH3 OH ¼ ; (17) Finally, the method is validated on a more challenging non-
YCO2;i YCO2 ;i MWCH3 OH
reactive heat transfer problem. Two different configurations
where subscripts i and o denote inlet and outlet of the reactor, are considered, i.e. in-line and staggered tube banks. The in-
respectively. Table 2 presents CO2 conversion and CH3OH line and the staggered cases have 81 and 77 cylinders,
yield for the space velocity and temperature sweeps. As can be respectively. The size of the computational domain for both
seen in the table, there is a good agreement between the cases is 18d  22d, where d is the cylinder diameter. Moreover,
Aspen Plus® data of Kiss et al. [23] and the present results the center-to-center distance between two cylinders in lon-
obtained by both OpenFOAM and Aspen Plus®, thus indicating gitudinal and transverse directions is 2d for both cases. The
the accurate implementation of the kinetic model. In general, mesh used for the in-line configuration has 697,536 cells while
the maximum absolute deviation is found to be approximately for the staggered case the mesh has 675,312 cells, with the grid
5% for the present data, which is consistent with findings of size near the cylinder taken as 0.005d for both cases. The
An et al. [1] and Kiss et al. [23]. boundary conditions are the same as in the previous section.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15639

Table 2 e Comparison of current simulations with data of An et al. [1] and Kiss et al. [23]: CO2 conversion and CH3OH yield.
SV and T sweeps are done at fixed T ¼ 523 K and SV ¼ 6 m3/kgcath, respectively.
Conversion/Yield An et al. [1] Kiss et al. [23] Aspen Plus OpenFOAM
3
SV [m /kgcath] 1 XCO2 0.262 0.2428 0.2417 0.2428
YCH3 OH 0.193 0.1548 0.1532 0.1547
2 XCO2 0.26 0.2428 0.2417 0.2428
YCH3 OH 0.191 0.1548 0.1532 0.1547
4 XCO2 0.256 0.2421 0.2409 0.2424
YCH3 OH 0.18 0.1544 0.1527 0.1545
6 XCO2 0.25 0.2398 0.2381 0.2406
YCH3 OH 0.166 0.153 0.1511 0.1534
8 XCO2 0.243 0.2362 0.2341 0.2376
YCH3 OH 0.153 0.1509 0.1488 0.1517
10 XCO2 0.23 0.2321 0.2297 0.2340
YCH3 OH 0.134 0.1485 0.1462 0.1496
T [K] 483 XCO2 0.17 0.1244 0.1184 0.1252
YCH3 OH 0.11 0.1152 0.1096 0.1159
503 XCO2 0.225 0.1957 0.1902 0.1984
YCH3 OH 0.155 0.1597 0.1549 0.1618
523 XCO2 0.255 0.2398 0.2381 0.2406
YCH3 OH 0.178 0.153 0.1511 0.1534
543 XCO2 0.25 0.2404 0.2396 0.2404
YCH3 OH 0.14 0.1058 0.1045 0.1058

respectively. We note that the observed deviations are within

Table 3 e Grid convergence for non-reacting flow over
the accuracy of the correlation which is approximately 25%.
cylinder at Re ¼ 100.
Finally, as expected the heat transfer is enhanced in staggered
Mesh Number of cells Cd Cl St CNuD
configuration.
M0 50,440 1.3557 ±0.3224 1.6865 5.1452
M1 102,132 1.3585 ±0.3289 1.6865 5.1298 Methanol production in 2D fixed-bed
M2 211,946 1.3610 ±0.3356 1.7133 5.1130

The method is finally applied to study methanol production in

a two-dimensional fixed-bed as depicted in Fig. 4. The
Temperature ((Tw  Ti)/Tw) distribution for the in-line and the
computational domain is chosen with size of 42 mm  12 mm,
staggered configurations are shown in Fig. 3. Comparison of
while the morphology of the bed is described by three
the average Nusselt number with an empirical correlation [52]
different configurations, i.e. in-line, staggered and random.
and numerical simulations [27,50] is presented in Table 4. The
Loose and dense packings are considered in this work as
results are in good agreement especially with numerical so-
shown in Fig. 4. Data used for different configurations are
lution of Wang et al. [50] and Lo and Su [27]. The difference
shown in Table 5. For dense packing, one more random case
between the empirical correlation and current results are 13%
(polydispersed) is considered where the diameter of cylinders
and 17% for the in-line and staggered configurations,

Fig. 2 e Validation of non-reacting single cylinder case. Comparison of local pressure coefficient and Nusselt number for
non-reacting flow over cylinder at Re ¼ 10, 20, 40 and 100.
15640 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

Fig. 3 e Validation of non-reacting in-line and staggered tube banks. Temperature fields for (left) the in-line and (right) the
staggered arrangements at Re ¼ 20.

Table 4 e Comparison of the CNuD for flow past a tube in a bank at Re ¼ 20.
Configuration Zhukauskas [52] Wang et al. [50] Lo and Su [27] Present
Staggered 3.45 4.08 4.07 4.03
In-line 3.07 … … 3.46

vary in the range of 0.72 mm  d  1.68 mm. Note that the total At the inlet, fully developed velocity profile is imposed,
area of the cylinders is same in both random cases. while no-slip conditions are assumed at the walls. Mean inlet
Following previous simulations [23] and experimental data velocity (Ui) is chosen so that its value is equivalent to a cy-
[1], the base line parameters are defined as T ¼ 523 K, lindrical tube (D ¼ 12 mm) with same SV resulting in Reynolds
p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6 and SV ¼ 6 m3/kgcath. Note that, Kiss number Re ¼ rUiD/m ¼ 1000. For pressure, zero-gradient
et al. [23] performed only isothermal 1D cases whereas here boundary condition is imposed at the inlet and the walls
both Aspen Plus® (1D) and OpenFOAM (1D/2D) are used. Also, and fixed pressure condition is used for the outlet. Fixed
both adiabatic and isothermal reactors are considered in the values of Ti and R for temperature and species are imposed at
current work. the inlet, respectively. For all variables, except pressure, zero-

Fig. 4 e Schematic illustration of a catalytic fixed-bed reactor with various packed configurations. The CO2 and H2 enter from
the left while they exit from the right end of the domain. The loose and dense packings consist of 42 and 64 cylinders,
respectively.

Table 5 e Dimensions of loose and dense packing configurations. Below, S stands for center-to-center distance between
cylinders, which is same in the longitudinal and the transverse directions for the in-line and staggered cases, R represents
randomly generated distance, Ncyl is total number of cylinders (d) and F is the solid volume fraction.
Configuration Lu [mm] Lb [mm] S [mm] d [mm] Ncyl [] F [%] Mesh
Loose In-line 12 12 1.8 1.2 42 33 437,577
Staggered 12 12 1.8 1.2 42 33 437,778
Random 12 12 R 1.2 42 33 460,162
Dense In-line 12 11.1 1.4 1.2 64 54 503,998
Staggered 12 11.1 1.4 1.2 64 54 507,071
Random 12 14.4 R 1.2 64 42 501,472
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15641

gradient conditions are imposed at the outlet. Adiabatic and configurations, high local methanol mass fraction is observed
isothermal boundary conditions are applied for the tempera- in the wake regions, while it is low in the channel regions. It is
ture on the reactor walls. Note that ideal isothermal condition clear that the in-line configuration shows the poorest perfor-
is assumed, i.e., the temperature inside the reactor is mance, due to relatively small interaction of the reactants
constant. with the catalytic surface. There are large regions of unreacted
flow in the channels especially in the vicinity of the upper and
Velocity and methanol distribution the lower walls. Moreover, the regions with reacted flow in the
First, velocity and methanol distributions are studied for the wake regions have only minor contribution to the overall
in-line, the random and the staggered configurations performance. As can be seen both staggered and random
assuming adiabatic boundary conditions. Distribution of non- configurations have stronger interaction of catalyst with
dimensional streamwise velocity U/Ui and methanol mass reactants.
fraction fields are depicted in Fig. 5. The flow direction is from
left to right. Detailed inspection reveals different zones in the Bed configuration effects
fixed-bed region. In the downstream of cylinders, stagnant/ Next, the effect of bed configuration on the mean mass frac-
reverse flow zones are noticed. Moreover, it is clear that the tion and temperature is studied. Fig. 6 shows mean mass
flow has preference towards zones with less resistance fraction profiles of reactant (CO2), products (CH3OH, CO) and
creating channels in interstitial regions and in the vicinity of temperature along the adiabatic and isothermal reactors. Note
the walls. Flow accelerates in the channels reaching that, due to space consideration other components are not
maximum up to a factor of U/Ui ¼ 5 for random configuration. shown. Moreover, all the mass fraction profiles provided in
Note that these features are consistent with previous obser- the paper are averaged in time and space. As can be seen the
vations [11]. For the in-line configuration, the main flow shows solution for random and staggered cases are similar while the
preference in horizontal direction thus is similar to straight in-line case shows the poorest performance. The CO2 starts
channel flow. As a result, the main flow does not reach some reacting in the inlet of porous media and its value is
parts of the surface of cylinders. On the other hand, the decreasing along the reactor. On the other hand, mass fraction
staggered case is more similar to a periodical converging- of products (CH3OH and CO) is increasing along the reactor.
diverging channel flow. Flow physics results in larger effec- Note that production of CO is unwanted and would be recy-
tive surface area (compared to the in-line case) exposed to the cled in the experiment. It is clear that the isothermal reactor
flow. In the random case, the flow moves in more tortuous outperforms the adiabatic one. For adiabatic case, mass frac-
channels due to non-uniform local porosity. tion profiles of CH3OH and CO are comparable, while for
Significant interplay between instantaneous velocity and isothermal one methanol is produced almost 3 times more
methanol fields can be observed as shown Fig. 5. Aforemen- than carbon monoxide. Temperature increases along the
tioned regions inside the bed significantly affect the reaction reactor due to exothermicity of the methanol synthesis re-
rates by influencing residence time distribution, i.e., the resi- actions. Similarly, the temperature profile follows closely for
dence time is high in the zero/back flow regions, while it is random and staggered cases, while the values are smaller for
short in the high speed channels. As a result, in all the the in-line case.

Fig. 5 e Loosely packed catalytic fixed-bed reactor. Instantaneous (left) non-dimensional streamwise velocity (U/Ui) and
(right) CH3OH mass fraction field distributions in an adiabatic reactor for (top) the in-line, (middle) the random and (bottom)
the staggered configurations. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).
15642 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

Fig. 6 e Loosely packed catalytic fixed-bed reactor. Effects of the configuration on the mean mass fraction profiles of CH3OH,
CO, CO2 and T for (top row) adiabatic and (middle row) isothermal reactors. For bottom row, red and black colors represent
results for adiabatic and isothermal reactors, respectively. The axial position of the bed is between 0 and 0.012 m.
(SV ¼ 6 m3/kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K). (For interpretation of the references to colour in this
figure legend, the reader is referred to the Web version of this article.)

Fig. 7 e Loosely packed catalytic fixed-bed reactor. Effects of the configuration on the CH3OH (,) and CO (B) yield for (left)
adiabatic and (right) isothermal reactors. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).

Next, CH3OH and CO yield values are compared for the 2D data. By definition the 1D case is insensitive to configura-
three configurations as shown in Fig. 7. As expected, 1D re- tion as local flow inhomogeneities are neglected, i.e., perfect
sults of OpenFOAM and Aspen Plus® are in good agreement. mixing. On the other hand, the 2D case has regions with
However, 1D results show higher values than corresponding unreacted flow. As a result, the in-line configuration clearly
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15643

experimental study of An et al. [1]. Profiles of mean temper-

ature along the adiabatic and the isothermal reactors are
plotted in Fig. 8. Moreover, CH3OH and CO yield values are
presented in Fig. 9. For the adiabatic case, methanol produc-
tion decreases with temperature, while it shows non-
monotonic trend for isothermal one. On the other hand,
amount of CO increases with temperature for both reactors. It
can be observed that within the range studied in here, the
most favorable temperature is 473 K for the adiabatic case,
while it is 523 K for the isothermal one. Carefully inspecting
the temperature profiles could explain the deviation, cf. Fig. 8.
Temperature increases in the adiabatic reactor at Ti ¼ 473 K
and 523 K while it decreases at 573 K. The overall trend could
be explained by the fact that methanol synthesis is mainly
exothermic reaction (increases temperature), i.e., increase in
the temperature deteriorates the equilibrium (decreases
Fig. 8 e Loosely packed catalytic fixed-bed reactor. Effects methanol production). While the reverse water-gas shift re-
of the inlet temperature on the mean profile of T for action is endothermic (decreases temperature), thus having
adiabatic and isothermal reactors. Red and black colors the opposite effect. It is worth noting that in the isothermal
represent results for adiabatic and isothermal reactors, case the temperature inside the reactor is constant. On the
respectively. The axial position of the bed is between 0 and other hand, for adiabatic case the temperature is a balance of
0.012 m. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, exothermic and endothermic reactions. At Ti ¼ 473 K, the
Re ¼ 1000). (For interpretation of the references to colour in methanol synthesis dominates over the reverse water-gas
this figure legend, the reader is referred to the Web version shift reaction thus having high values of methanol and tem-
of this article.) perature. The temperature along the adiabatic reactor ap-
proaches optimal value for the isothermal case (Ti ¼ 523 K) but
reaches only 512.8 K. On the other hand, at Ti ¼ 573 K the
shows the poorest performance. Note that there is a signifi-
reverse water-gas shift reaction is more dominant resulting in
cant difference between adiabatic and isothermal reactors.
higher CO production and decrease in the T along the adia-
For adiabatic reactor, CH3OH and CO yields are comparable,
batic reactor. Moreover, it is observed that both isothermal
whereas for isothermal reactor, CH3OH yield is approximately
and adiabatic reactors show similar performance at Ti ¼ 573 K.
three times larger than corresponding CO one. It proves that
This could be attributed to only a small change in temperature
isothermal condition is important for best performance.
along the bed. Note that increasing temperature beyond Ti ¼
Hereafter, the random configuration is used unless otherwise
573 K may lead to catalyst deactivation such as catalyst sin-
stated.
tering or obstruction of active sites [42].

Temperature effects
Pressure effects
Next, the effect of inlet temperature is studied for the random
Then we examine the effects of pressure by varying p ¼ 2.5, 5
configuration. For this purpose, computations are conducted
and 10 MPa for the random configuration while keeping the
by varying the inlet temperature in the range of
other parameters same as in the base case. The results are
473 K  Ti  573 K while the other parameters are same as in
shown in Figs. 10 and 11. The CH3OH yield is favored by in-
the base case. The temperature range is in agreement with
crease of pressure for both reactors which could be attributed

Fig. 9 e Loosely packed catalytic fixed-bed reactor. Effects of the inlet temperature on the CH3OH (,) and CO (B) yield for
(left) adiabatic and (right) isothermal reactors. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000).
15644 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

high pressure enhances methanol production, it poses a

safety issue for the reactor.

Space velocity effects

Here, the effect of the space velocity on the reactor efficiency
is studied. The space velocity is varied between 6 and 600 m3/
kgcath while keeping the other parameters as in the base case.
The results are plotted in Fig. 12. Increase of SV leads to
decrease in the yield of both CH3OH and CO. Increase in SV
could be attained by either increasing the feed speed or
decreasing the mass of catalyst. Both methods were tested for
the 1D case and identical solutions were obtained. In case of
increased feed, the residence time decreases which leads to
decrease in the yield. Similarly, decreasing catalytic activity by
decreasing mass of catalyst results in diminished methanol
production. Although not shown here, it is observed that the
Fig. 10 e Loosely packed catalytic fixed-bed reactor. Effects mean temperature decreased with SV for the adiabatic case,
of the pressure on the mean profile of T for adiabatic and which could be attributed to the reduction in the methanol
isothermal reactors. Red and black colors represent results yield.
for adiabatic and isothermal reactors, respectively. The
axial position of the bed is between 0 and 0.012 m. Reactants ratio effects
(SV ¼ 6 m3/kgcath, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K). Another important parameter the inlet reactants ratio R is
(For interpretation of the references to colour in this figure studied at R ¼ 3, 6, 9 and 12. The results are plotted in Fig. 13.
legend, the reader is referred to the Web version of this Note that stoichiometric number is optimal at R ¼ 3. Thus,
article.) increasing R leads to excess H2 in the feed and increase of CO
and CH3OH yield. Moreover, it is found that the temperature
increases with the reactants ratio. As expected, the isothermal
to rise in reactants concentration. At p ¼ 2.5 MPa, solution is case outperforms by producing more methanol and less CO.
similar for both adiabatic and isothermal conditions due to Overall, the current results for the isothermal case are in
the almost constant temperature inside the reactor, cf. Fig. 10. agreement with data in literature [1,23,35].
At higher pressures, the differences in CH3OH and CO yield
between adiabatic and isothermal conditions increases due to Packing density effects
higher bed temperature in the former case [5]. While in Finally, simulations are performed to investigate the effects of
isothermal conditions the bed temperature stays at constant packing density. For this purpose, dense packing with in-line
Ti ¼ 523 K regardless the change in pressure, for the adiabatic (F ¼ 54%), staggered (F ¼ 54%) and random (F ¼ 42%) config-
case it increases with pressure, cf. Fig. 10. As it has been dis- urations are considered. The details of the configurations are
cussed in the Temperature effects subsection higher tempera- mentioned above and can be seen in Table 5.
tures in adiabatic conditions lead to a decrease in CH3OH Instantaneous contours of methanol mass fraction for or-
yield, cf. Fig. 9. Note that, the CO formation is insensitive to dered and random packings in adiabatic regime are shown in
pressure variation in the adiabatic case, while it diminishes in Fig. 14. As mentioned earlier, in addition to regular random
the isothermal one. This could be attributed to enhanced CO case one more polydispersed case is considered with varying
conversion to methanol at constant temperature. Even though size of diameters. Moreover, CH3OH and CO yield values for

Fig. 11 e Loosely packed catalytic fixed-bed reactor. Effects of the pressure on the CH3OH (,) and CO (B) yield for (left)
adiabatic and (right) isothermal reactors. (SV ¼ 6 m3/kgcath, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15645

Fig. 12 e Loosely packed catalytic fixed-bed reactor. Effects of the space velocity on the CH3OH (,) and CO (B) yield for (left)
adiabatic and (right) isothermal reactors. (p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).

Fig. 13 e Loosely packed catalytic fixed-bed reactor. Effects of the reactants ratio R ¼ H2:CO2 on the CH3OH (,) and CO (B)
yield for (left) adiabatic and (right) isothermal reactors. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa, Re ¼ 1000, T ¼ 523 K).

both adiabatic and isothermal cases are plotted in Fig. 15. densely packed system leads to more uniform distribution of
Overall performance is similar to the loose packing cases the species in the bed, thus approaching 1D case. Freund et al.
observed in Figs. 5 and 7, i.e., the in-line case shows lowest [12] mentioned that the critical aspect ratio of tube-to-particle
yield values. It is especially evident in Fig. 14, where unreacted diameter, below which the plug flow model becomes inade-
flow is observed near the top and bottom walls for the in-line quate, is around 10. In the current work, the average aspect
case. Moreover, the yield values of densely packed systems ratio is 10 for all the cases, thus approaching the plug flow
are higher than corresponding loose packings, reaching model results. Moreover, Mousazadeh et al. [31] validated
values of the 1D approach except the in-line case. Indeed, their 2D simulations with same aspect ratio against plug flow

Fig. 14 e Densely packed catalytic fixed-bed reactor. Instantenous methanol mass fraction fields for random (uniform and
polydispersed), in-line and staggered configurations in the adiabatic reactor. The flow goes from left to right. (SV ¼ 6 m3/
kgcath, p ¼ 5 MPa, R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).
15646 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

Fig. 15 e Densely packed catalytic fixed-bed reactor. Effects of the configuration on the CH3OH (,) and CO (B) yield for (left)
adiabatic and (right) isothermal reactors. Poly stands for the polydispersed case. (SV ¼ 6 m3/kgcath, p ¼ 5 MPa,
R ¼ H2:CO2 ¼ 6, Re ¼ 1000, T ¼ 523 K).

model. Overall, it is observed that the yield values increases arrangement led to the poorest performance due to the
with packing density approaching its upper limit set by 1D limited interplay between reactants and the catalytic sur-
approach. face. Moreover, the staggered and the random configura-
tions showed similar performance.
4. Both isothermal and the adiabatic reactors were consid-
Conclusions ered. The isothermal reactor generally outperformed the
corresponding adiabatic one due to exothermicity of the
The LHHW kinetic model was coupled with OpenFOAM solver methanol synthesis reaction resulting in higher CH3OH
and Aspen Plus® for CO2 and hydrogen-based methanol syn- and lower CO yield.
thesis in a fixed-bed reactor. The implementation of the ki- 5. Overall, it was observed that the methanol yield increased
netic model was validated for the plug flow reactor and the with the pressure, the catalyst loading and the reactant
results were found to be in a good agreement with experi- ratio. The yield decreased with temperature for the adia-
mental [1] and numerical [23] data. Next, the flow solver was batic case, while it showed non-monotonic trend for the
validated against heat transfer over single and tube bank isothermal case.
problems. Finally, the method was applied to a more chal-
lenging problem for two-dimensional fixed-bed reactor with
cylindrical pellets. The effects of packing density, bed config-
uration, temperature, pressure, reactant ratio, and space ve- Declaration of competing interest
locity on adiabatic and isothermal fixed-bed reactors were
examined. The overall findings can be summarised as: The authors declare that they have no known competing
financial interests or personal relationships that could have
1. The OpenFOAM (1D/2D) and the Aspen Plus® (1D) results appeared to influence the work reported in this paper.
were compared. In overall, results obtained from both 1D
and 2D simulations showed similar trend for the effects of
temperature, pressure, reactant ratio, and space velocity. Acknowledgments
However, due to perfect mixing assumption, the 1D model
was insensitive to the variation of bed configuration and The authors acknowledge to the Academy of Finland for the
packing density. support of this research through Grant Nos. 332835, 333069
2. The difference between the 1D and the 2D cases were and profiling funding 5 for Energy Storage with grant number
pronounced at loose packing conditions. The methanol 326346. This work was supported by the Aalto University
yield values obtained from 2D simulations were generally under grant 97410923. Business Finland is acknowledged for
smaller than the ones in the 1D case. However, the yield the main financial support (grant 8797/31/2019) of the
values converged to the corresponding 1D values with P2XEnable project (p2xenable.fi). The project partners are also
packing density. It was observed that the maximum thanked for their financial contributions. The authors are
methanol yield values were limited by quantities obtained grateful for the use of the computer facilities within the Aalto
from 1D simulations. University School of Science “Science-IT” project. We thank
3. The in-line, the staggered, and the random configurations Atmadeep Bhattacharya for fruitful discussions during
were compared for 2D case. It was found that the in-line writing the manuscript.
 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8 15647

references Technol 2016;7. http://ijtech.eng.ui.ac.id/old/index.php/

journal/article/view/6890.
[18] Jasak H, Weller H, Gosman A. High resolution NVD
differencing scheme for arbitrarily unstructured meshes. Int
[1] An X, Zuo Y, Zhang Q, Wang J. Methanol synthesis from CO2
J Numer Methods Fluid 1999;31:431e49.
hydrogenation with a Cu/Zn/Al/Zr fibrous catalyst. Chin J
[19] Jurtz N, Kraume M, Wehinger GD. Advances in fixed-bed
Chem Eng 2009;17:88e94.
reactor modeling using particle-resolved computational fluid
[2] Andersson J, Gro€ nkvist S. Large-scale storage of hydrogen.
dynamics (CFD). Rev Chem Eng 2019;35:139e90.
Int J Hydrogen Energy 2019;44:11901e19. https://doi.org/
[20] Kahila H, Wehrfritz A, Kaario O, Masouleh] MG, Maes N,
10.1016/j.ijhydene.2019.03.063. http://www.sciencedirect.
Somers B, et al. Large-eddy simulation on the influence of
com/science/article/pii/S0360319919310195.
injection pressure in reacting Spray A. Combust Flame
[3] Arab S, Commenge JM, Portha JF, Falk L. Methanol synthesis
2018;191:142e59.
from CO2 and H2 in multi-tubular fixed-bed reactor and
[21] Karimkashi S, Kahila H, Kaario O, Larmi M, Vuorinen V.
multi-tubular reactor filled with monoliths. Chem Eng Res
Numerical study on tri-fuel combustion: ignition properties
Des 2014;92:2598e608.
of hydrogen-enriched methane-diesel and methanol-diesel
[4] Baykara SZ. Hydrogen: a brief overview on its sources,
mixtures. Int J Hydrogen Energy 2020;45:4946e62.
production and environmental impact. Int J Hydrogen Energy
[22] Karthik GM, Buwa VV. A computational approach for the
2018;43:10605e14. https://doi.org/10.1016/
selection of optimal catalyst shape for solid-catalysed gas-
j.ijhydene.2018.02.022. http://www.sciencedirect.com/
phase reactions. React Chem Eng 2020;5:163e82.
science/article/pii/S0360319918304002.
[23] Kiss AA, Pragt J, Vos H, Bargeman G, De Groot M. Novel
[5] Bertau M. Methanol: The basic chemical and energy
efficient process for methanol synthesis by CO2
feedstock of the future. 1st ed. Heidelberg: Springer; 2014.
hydrogenation. Chem Eng J 2016;284:260e9.
[6] Bhattacharya A, Bhattacharya A, Datta A. Modeling of
[24] Kuroki M, Ookawara S, Ogawa K. A high-fidelity CFD model
hydrogen production process from biomass using oxygen
of methane steam reforming in a packed bed reactor. J Chem
blown gasification. Int J Hydrogen Energy 2012;37:18782e90.
Eng Jpn 2009;42:s73e8. https://doi.org/10.1252/jcej.08we256.
[7] Boot-Handford ME, Abanades JC, Anthony EJ, Blunt MJ,
[25] Kurtz J, Sprik S, Bradley TH. Review of transportation
Brandani S, Mac Dowell N, Ferna
ndez JR, Ferrari MC, Gross R,
hydrogen infrastructure performance and reliability. Int J
Hallett JP, Haszeldine RS, Heptonstall P, Lyngfelt A,
Hydrogen Energy 2019;44:12010e23. https://doi.org/10.1016/
Makuch Z, Mangano E, Porter RTJ, Pourkashanian M,
j.ijhydene.2019.03.027. http://www.sciencedirect.com/
Rochelle GT, Shah N, Yao JG, Fennell PS. Carbon capture and
science/article/pii/S0360319919309516.
storage update. Energy Environ Sci 2014;7:130e89. https://
[26] Lim HW, Park MJ, Kang SH, Chae HJ, Bae JW, Jun KW.
doi.org/10.1039/C3EE42350F.
Modeling of the kinetics for methanol synthesis using Cu/
[8] Bozzano G, Manenti F. Efficient methanol synthesis:
ZnO/Al2O3/ZrO2 catalyst: influence of carbon dioxide during
perspectives, technologies and optimization strategies. Prog
hydrogenation. Ind Eng Chem Res 2009;48:10448e55.
Energy Combust Sci 2016;56:71e105.
[27] Lo D, Su DT. An embedding finite element analysis of heat
[9] Das B, Bhattacharya A, Datta A. Kinetic modeling of biomass
transfer on the surface of circular cylinders in flow. Int J Heat
gasification and tar formation in a fluidized bed gasifier using
Mass Tran 2012;55:6916e26.
equivalent reactor network (ERN). Fuel 2020;280:118582.
[28] Manenti F, Cieri S, Restelli M. Considerations on the steady-
[10] Dixon AG, Partopour B. Computational fluid dynamics for
state modeling of methanol synthesis fixed-bed reactor.
fixed bed reactor design. Ann Rev Chem Biomolec Eng
Chem Eng Sci 2011;66:152e62.
2020;11.
[29] Manenti F, Cieri S, Restelli M, Bozzano G. Dynamic modeling
[11] Dixon AG, Walls G, Stanness H, Nijemeisland M, Stitt EH.
of the methanol synthesis fixed-bed reactor. Comput Chem
Experimental validation of high Reynolds number CFD
Eng 2013;48:325e34.
simulations of heat transfer in a pilot-scale fixed bed tube.
[30] Mirvakili A, Bakhtyari A, Rahimpour MR. A CFD modeling to
Chem Eng J 2012;200e202:344e56.
investigate the impact of flow mal-distribution on the
[12] Freund H, Zeiser T, Huber F, Klemm E, Brenner G, Durst F,
performance of industrial methanol synthesis reactor. Appl
Emig G. Numerical simulations of single phase reacting flows
Therm Eng 2018;128:64e78.
in randomly packed fixed-bed reactors and experimental
[31] Mousazadeh F, van Den Akker H, Mudde RF. Direct
validation. Chem Eng Sci 2003;58:903e10. 17th International
numerical simulation of an exothermic gas-phase reaction
Symposium of Chemical Reaction Engineering (IS CRE 17).
in a packed bed with random particle distribution. Chem Eng
[13] Graaf G, Stamhuis E, Beenackers A. Kinetics of low-pressure
Sci 2013;100:259e65. https://doi.org/10.1016/j.ces.2013.02.019.
methanol synthesis. Chem Eng Sci 1988;43:3185e95.
11th International Conference on Gas-Liquid and Gas-Liquid-
[14] Gumber S, Gurumoorthy AVP. Chapter 25 - Methanol
Solid Reactor Engineering. http://www.sciencedirect.com/
economy versus hydrogen economy. In: Basile A, Dalena F,
science/article/pii/S0009250913001206.
editors. Methanol. Elsevier; 2018. p. 661e74. http://www.
ri J, Magdeldin M, Larmi M, Ja
€ rvinen M, Santasalo-
[32] Nya
sciencedirect.com/science/article/pii/
Aarnio A. Techno-economic barriers of an industrial-scale
B978044463903500025X.
methanol CCU-plant. J CO2 Utilization 2020;39:101166.
[15] International Energy Agency. Exploring Clean Energy
[33] Olabi AG, saleh Bahri A, Abdelghafar AA, Baroutaji A,
Pathways: The role of CO2 storage. Tech Rep Int Energy
Sayed ET, Alami AH, et al. Large-scale hydrogen production
Agency 2019a;(July):3e5. https://webstore.iea.org/download/
and storage technologies: current status and future
direct/2824?fileName¼Exploring_Cle an_Energy_Pathways.
directions. Int J Hydrogen Energy 2020. https://doi.org/
pdf.
10.1016/j.ijhydene.2020.10.110.
[16] International Energy Agency. Putting CO2 to Use: Creating
[34] Olah GA, Goeppert A, Prakash GKS. Beyond oil and gas: The
Value from Emissions. Tech Rep Int Energy Agency
methanol economy. 2nd ed. Weinheim: Wiley-VCH; 2009.
2019b;(September):3e4. https://webstore.iea.org/download/
[35] Park N, Park MJ, Lee YJ, Ha KS, Jun KW. Kinetic modeling of
direct/2830?fileName¼Putting_CO2_to_Use.pdf.
methanol synthesis over commercial catalysts based on
[17] Jalil Z, Rahwanto A, Mulana F, Mustanir M. Desorption
three-site adsorption. Fuel Process Technol
temperature characteristic of Mg-based hydrides catalyzed
2014;125:139e47.
by nano-SiO2 prepared by high energy ball milling. Int J
15648 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 6 ( 2 0 2 1 ) 1 5 6 3 5 e1 5 6 4 8

[36] Petera J, Nowicki L, Ledakowicz S. New numerical algorithm methanol, with gasolinedenergetic and economic aspects.
for solving multidimensional heterogeneous model of the Int J Hydrogen Energy 1998;23:387e96. https://doi.org/
fixed bed reactor. Chem Eng J 2013;214:237e46. 10.1016/S0360-3199(97)00077-3. https://www.sciencedirect.
[37] Pontzen F, Liebner W, Gronemann V, Rothaemel M, Ahlers B. com/science/article/pii/S0360319997000773.
CO2-based methanol and DME e efficient technologies for [45] Takagawa M, Ohsugi M. Study on reaction rates for methanol
industrial scale production. Catal Today 2011;171:242e50. synthesis from carbon monoxide, carbon dioxide, and
[38] Porter RT, Fairweather M, Kolster C, Mac Dowell N, Shah N, hydrogen. J Catal 1987;107:161e72. https://doi.org/10.1016/
Woolley RM. Cost and performance of some carbon capture 0021-9517(87)90281-8. http://www.sciencedirect.com/
technology options for producing different quality CO2 science/article/pii/0021951787902818.
product streams. Int J Greenhouse Gas Control 2017. https:// [46] Tekgül B, Kahila H, Kaario O, Vuorinen V. Large-eddy
doi.org/10.1016/j.ijggc.2016.11.020. simulation of dual-fuel spray ignition at different ambient
[39] Redondo B, Shah MT, Pareek VK, Utikar RP, Webley PA, temperatures. Combust Flame 2020;215:51e65.
Patel J, Lee WJ, Bhatelia T. Intensified isothermal reactor for [47] Ushikoshi K, Mori K, Kubota T, Watanabe T, Saito M.
methanol synthesis. Chem Eng Proces - Process Intensific Methanol synthesis from CO2 and H2 in a bench-scale test
2019;143:107606. plant. Appl Organomet Chem 2000;14:819e25.
[40] Santasalo-Aarnio A, Nyari J, Wojcieszyk M, Kaario O, [48] Vanden Bussche K, Froment G. A steady-state kinetic model
Kroyan Y, Magdeldin M, Larmi M, Ja € rvinen M. Application of for methanol synthesis and the water gas shift reaction on a
synthetic renewable methanol to power the future propulsion. commercial Cu/ZnO/Al2O3 catalyst. J Catal 1996;161:1e10.
Technical Report 2020. https://doi.org/10.4271/2020-01-2151. [49] Villa P, Forzatti P, Buzzi-Ferraris G, Garone G, Pasquon I.

[41] Setinc M, Levec J. Dynamics of a mixed slurry reactor for the Synthesis of alcohols from carbon oxides and hydrogen. 1.
three-phase methanol synthesis. Chem Eng Sci Kinetics of the low-pressure methanol synthesis. Ind Eng
2001;56:6081e7. https://doi.org/10.1016/S0009-2509(01)00212- Chem Process Des Dev 1985;24:12e9. https://doi.org/10.1021/
3. http://www.sciencedirect.com/science/article/pii/ i200028a003.
S0009250901002123. [50] Wang Z, Fan J, Luo K, Cen K. Immersed boundary method for
[42] Shahrokhi M, Baghmisheh G. Modeling, simulation and the simulation of flows with heat transfer. Int J Heat Mass
control of a methanol synthesis fixed-bed reactor. Chem Eng Tran 2009;52:4510e8.
Sci 2005;60:4275e86. [51] Weller HG, Tabor G, Jasak H, Fureby C. A tensorial approach
[43] Solsvik J, Jakobsen HA. Multicomponent mass diffusion in to computational continuum mechanics using object-
porous pellets: Effects of flux models on the pellet level and oriented techniques. Comput Phys 1998;12:620e31.
impacts on the reactor level. Application to methanol [52] Zhukauskas A. Heat transfer from tubes in cross flow. In:
synthesis. Can J Chem Eng 2013;91:66e76. Irvine TF, Hartnett JP, editors. Advances in heat transfer, vol.
[44] Specht M, Staiss F, Bandi A, Weimer T. Comparison of the 8. Academic Press; 1972. p. 93e160.
renewable transportation fuels, liquid hydrogen and