International Journal of Greenhouse Gas Control 111 (2021) 103447

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control

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

Full Length Article

Assessment of CO2 capture efficiency in packed bed versus 3D-printed

monolith reactors for SEWGS using CFD modeling
Vlad-Cristian Sandu a, Ana-Maria Cormos a, *, Ionela-Dorina Dumbrava a, Arpad Imre-Lucaci a,
Calin-Cristian Cormos a, Robert de Boer b, Jurriaan Boon b, Soraya Sluijter b
a
Faculty of Chemistry and Chemical Engineering, Babeș-Bolyai University, Arany Janos 11, Cluj-Napoca 400028, Romania
b
TNO Energy Transition, PO Box 15, Petten 1755 ZG, the Netherlands

A R T I C L E I N F O A B S T R A C T

Keywords: Sorption Enhanced Water-Gas Shift (SEWGS) couples the water gas-shift reaction with CO2 adsorption on
3D-printing potassium-promoted hydrotalcite (K-HTC) sorbent material for hydrogen production and CO2 capture. Here,
CO2 capture computational fluid dynamics (CFD) models are developed to simulate the adsorption step in SEWGS considering
CFD
both packed bed and monolith reactors. The goal of this research is increasing SEWGS productivity by using 3D-
Monolith
Packed bed
printed structured bed reactors, as opposed to conventional packed bed reactors, due to the well-known ad­
SEWGS vantages of monoliths, such as less mass transfer limitations and lower pressure drops. The paper presents nu­
merical modeling and simulation work supported by experimental validation in order to compare packed bed
and monolith reactors with respect to SEWGS performance. The packed bed multiscale CFD model is validated
using breakthrough experimental data. A bench-scale CFD structured bed model is developed and validated
based on breakthrough measurements performed by TNO using 3D-printed K-HTC monolith structures.
Furthermore, geometry effects on mass transfer efficiency are investigated through CFD modeling for the bench-
scale reactor. A scale-up of the monolith model to pilot-scale allows for a proper comparison with the packed bed
technology. Monolith reactor model breakthrough predictions show there is a considerable increase in mass
transfer rate over the packed bed reactor for the adsorption step in SEWGS, demonstrating promising potential
towards enhancing the carbon capture technology.

advancement towards reducing emissions of CO2. The captured CO2 can

1. Introduction be stored or reused in order to produce other chemicals (IPCC, 2005).
A technology coupling the pre-combustion capture of CO2 with the
As the energy demand worldwide is increasing, fossil fuels still production of H2 from the Water Gas-Shift reaction is the Sorption
remain the most important energy source associated with greenhouse Enhanced Water Gas-Shift (SEWGS) technology. SEWGS is similar to a
gas emissions. In recent years, hydrogen has become one of the most pressure swing adsorption (PSA) process, where adsorption of CO2 on
studied energy carriers with the potential to replace fossil fuels and solid material takes place at pressures of 10–40 bar and temperatures
reduce toxic combustion emissions (IPCC, 2014). However, 48% of between 300 and 500 ◦ C, during the WGS reaction (Lee et al., 2010;
hydrogen is currently produced by Steam Methane Reforming (SMR), Boon et al., 2014).
due to its technological maturity and economic viability, followed by CO + H2 O⇌CO2 + H2 (1)
oil/naphtha reforming accounting for 30%. Other methods of produc­
tion include gasification of coal, more economically favorable, and by CO2 + ⋅ ⇌CO2 ⋅ (2)
using renewable sources such as biomass and bio-derived sources (da
Depending on process conditions, steam can also be adsorbed:
Silva Veras et al., 2017). Carbon capture technology combined with a
fossil fuel-based hydrogen production unit could become a promising

Abbreviations: CFD, computational fluid dynamics; IGCC, integrated gasification combined cycle; K-HTC, potassium promoted hydrotalcite; LDF, linear driving
force; PSA, pressure swing adsorption; SERP, sorption-enhanced reaction process; SEWGS, sorption-enhanced water-gas shift; SMR, steam methane reforming; TNO,
Netherlands Organization for Applied Scientific Research; TSA, temperature swing adsorption; WGS, water-gas shift.
* Corresponding author.
E-mail address: [email protected] (A.-M. Cormos).

https://doi.org/10.1016/j.ijggc.2021.103447
Received 24 February 2021; Received in revised form 16 July 2021; Accepted 28 August 2021
Available online 7 September 2021
1750-5836/© 2021 Elsevier Ltd. All rights reserved.
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Nomenclature p0 saturation pressure Pa

pc critical pressure Pa
A nanopore-sorbate interaction parameter - pi component partial pressure Pa
Ac microchannel area m2 q* equilibrium amount adsorbed mol•kg− 1
ci component concentration mol•m− 3 qi component adsorbent loading mol•kg− 1
cpe,i component concentration inside pellet mol•m− 3 qs maximum (monolayer) amount adsorbed mol•kg− 1
dc column diameter m Q inlet flowrate NL•min− 1
dpe pellet diameter m r radial coordinate from center to pellet surface -
dH channel hydraulic diameter m rpe pellet radius m
De,i component effective diffusion coefficient inside porous RADS,i component adsorption rate mol•m− 3•s− 1
domain m2•s− 1 t time s
Di component diffusion coefficient m2•s− 1 T fluid temperature K
DL,i component axial dispersion coefficient m2•s− 1 Tc critical temperature K
Dpe,i component effective diffusion coefficient inside the pellet u fluid velocity m•s− 1
m2•s− 1 V0 limiting nanopore volume per mass of sorbent cm3•kg− 1

E nanopore-sorbate interaction energy J•mol− 1

i species index - Greek
Ip∗ specific polar moment of inertia of microchannel cross εb bed porosity (void fraction) -
εm material porosity (void fraction) -
section -
μ fluid dynamic viscosity Pa•s
K surface-sorbate interaction parameter Pa− 1
K adsorbent permeability m2
L reactor or channel length m
ρ fluid density kg•m− 3
m pore-size distribution parameter -
ρb bed density kg•m− 3
n Freundlich isotherm parameter -
τ viscous stress tensor Pa
N number of pellets per unit volume of bed -
p fluid pressure Pa

H2 O + ⋅⇌H2 O⋅ (3) the separation of gas mixtures. Verougstraete et al. (2020), evaluated
CO2 capture using a honeycomb monolith in a TSA process, in terms of
Syngas is converted and separated into streams of CO2 and H2 at high
regeneration time, regeneration methods and steam presence during
purity, making SEWGS exceptionally well suited for pre-combustion
adsorption and desorption. The results demonstrated that the sorbent
capture, and thus, mitigation of greenhouse gas emissions. The sor­
enabled CO2 capture at very high flow rates and under humid condi­
bent is periodically regenerated by purging with steam at low pressure.
tions. It was found that very high fluid velocities lead to mass transfer
Unlike conventional WGS technology, requiring two consecutive re­
limitations and that, due to the exothermic nature of adsorption, heat
actors for high CO conversion, SEWGS combines the second reactor and
effects have an important influence when cycling at low temperatures.
PSA into a single SEWGS unit, reducing the number of process steps and
The monolith structure is considered the archetype of structured
necessary equipment (Boon et al., 2015).
sorbents. An efficient way of understanding the various interactions
The advantages of SEWGS have been extensively studied and
between physical and chemical processes within structured sorbents
experimentally demonstrated (Boon et al., 2014; 2015; van Selow et al.,
would be through detailed modeling and simulation of monolith re­
2009). When considering IGCC power decarbonization, the SEWGS
actors, consequently, comprehensive research was done on the matter.
concept showed a significantly lower Specific Energy Consumption for
Kolaczkowski (1999) went over some challenges encountered when
CO2 Avoided of 2.5 MJ/kgCO2 in comparison to using Selexol in
modeling monolith reactors, such as selection of kinetics, correlations of
pre-combustion mode which yielded 3.7 MJ/kgCO2 for similar CO2
mass and heat transfer, etc. Heck et al. (2001) presented an introductory
avoidance rates (Gazzani et al., 2013), or 4.2 MJ/kgCO2 in the case of
review based on the principles of monoliths when used as support for
post-combustion technologies applied in pulverized-coal power plant
catalytic gas phase reactions. Models of various scales (washcoat, single
layouts (Manzolini et al., 2020). The most used sorbent in the SEWGS
or multi-channel) and dimension levels (1D, 2D, 3D) were developed
process is a potassium-promoted hydrotalcite-based material (K-HTC),
and simulated in multiphase processes (Chen et al., 2008). Pahinkar
due to its ability to quickly adsorb CO2 reversibly at high temperature
et al. (2015) studied the use of adsorbent microchannels in a typical PSA
(200–600 ◦ C), high thermal and mechanical stability and high selec­
process for cycle time reduction, in which the adsorbent layer was
tivity to CO2 (Ebner et al., 2006; Lee et al., 2007; Bhatta et al., 2015;
attached to the monolith wall.
Coenen et al., 2017).
This work focuses on modeling adsorption in the SEWGS process.
Most industrial adsorption processes are carried out in columns
Thus far, few papers have been published on SEWGS modeling (Boon
packed with adsorbent in pellet form. A potential limitation of using
et al., 2014; Wright et al., 2011) and even less on CFD models developed
such a configuration is the high pressure drop per unit length. Monolith
to simulate SEWGS using monolith reactors for hydrogen production. A
reactors (structured beds) have the potential to overcome such opera­
1D model was developed and subsequently validated by Reijers et al.
tional challenges, because of their high void fraction and large surface
(2009) to investigate the performance of sorption-enhanced reaction
and contact areas, resulting in a low pressure drop when operated at
process (SERP) for CO2 capture considering the SMR and WGS reactions
high flow rates (Palma et al., 2016; Akri et al., 2018; Hosseini et al.,
in a fixed bed reactor. Wright et al. (2011) developed a SEWGS model for
2020). There is ample research highlighting the advantages of mono­
integrated gasification combined cycle (IGCC) using fixed bed reactors
lithic structures in CO2 capture processes involving high flow rates
to study effects of various feedstock and concluded that a similar amount
(Hosseini et al., 2020). An investigation into the adsorption equilibria of
of steam was required to operate SEWGS units, regardless of the syngas
CO2, CH4 and N2 on activated carbon monoliths was undertaken by
source being natural gas or coal. Lee et al. (2010) presented a multi­
Ribeiro et al. (2008). The results were used in modeling pressure swing
component adsorption isotherm for SEWGS using K-HTC and the reactor
adsorption (PSA) and temperature swing adsorption (TSA) processes for
model developed based on the adsorption isotherm and linear driving

2
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 1. (a) Packed bed 3D representation, (b) 3D-printed structures, (c) monolith 3D representation, (d) quarter monolith 3D representation, (e) single channel 3D
representation and (f) single channel 2D-axisymmetric representation; blue is the K-HTC sorbent material, while grey is the fluid domain.

force (LDF) kinetics was in agreement with breakthrough experiments. A height (1D), while microscale was described using the pellet radius (1D).
CFD monolith reactor model considering the SMR and WGS reactions for The model was validated based on experimental data and input from
hydrogen production was developed by Irani et al. (2011) and was run in TNO, the Netherlands.
steady state to investigate whether surface or volume approach was Subsequently, a 3D structured bed CFD model was developed to
more accurate for implementing reaction rates in monolith reactors. study adsorption dynamics for CO2 adsorption for SEWGS. The model
Abouelella et al. (2018) developed a 1D CFD multiscale model, using the geometry was based on a structure that was 3D-printed by TNO. The
LDF approximation to represent the adsorption rate, to study adsorption model was validated using experimental breakthrough data from TNO.
of CO2 using different capture materials and the simulation results As the method of fabrication for the monolith reactor was 3D-printing,
matched experimental data well. the entire dividing wall consists of sorbent K-HTC material, instead of
The goal of this research is to increase the productivity of SEWGS the traditional adsorbent layer coating and monolith wall as support
using 3D-printed structured bed reactors, as opposed to the conventional combination. Unlike published models, the effects of 3D-printing the
packed bed reactors, by taking advantage of the reduced mass transfer structure entirely out of adsorbent material are captured by the 3D CFD
limitations and pressure drops provided by structuring. Monoliths are model developed in this work, as CO2 loading takes place due to the
the starting point for this work, as they provide many benefits as contribution of multiple channels. Recently, successful methods have
explained above. However, the novelty of this work lies in the fact that been developed for the 3D-printing of microporous ceramic materials
3D-printing allows for producing more complex structures that cannot (Abouelella et al., 2018). Most models published on SEWGS use the LDF
be obtained by traditional manufacturing methods once the structure- approximation to estimate the adsorption rate; however, this work uses
productivity relations are well understood and can be modelled. This the equilibrium theory to calculate adsorption rate as internal and
paper presents numerical modeling and simulation work backed by external mass transfer resistances are accurately described by the CFD
experimental validation with the goal of investigating packed bed and model through diffusion with a Fickian type of approach.
monolith reactors used in SEWGS in regards to performance and provide Finally, the model was upscaled to a size comparable to the packed
insight into the enhanced mass transfer efficiency provided by the bed technology and a 2D-axisymmetric monolith reactor CFD model was
structured bed reactor. To this end, a multiscale packed bed reactor CFD developed to study the fluid dynamics and mass transfer occurring in the
model was developed to simulate competitive adsorption on K-HTC in adsorption stage of SEWGS. Adsorption rate of CO2 and H2O was
the SEWGS process, where macroscale was represented by the bed calculated using the equilibrium theory expression based on an isotherm

3
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

published by Boon et al. (2014). simulated in time dependent studies at high pressure. For the structured
bed bench-scale model, three different flow rate values were simulated
2. Model development for three different feed temperature values at ambient pressure for
validation purposes. For the pilot-scale monolith model, all three flow
To compare the performances of both technology types, packed bed rate cases were initially simulated in stationary studies for pressure drop
reactor and monolith reactor, it was necessary to develop detailed estimations, followed by a time dependent study for the first flow rate
models of the considered systems. The CFD models were developed case, where species transfer and fluid flow were simulated dynamically
using COMSOL Multiphysics. Schematic representations of each tech­ at high pressure.
nology are represented in Fig. 1. The packed bed reactor dimensions and The packed bed used in the experiments was equipped with 7 ther­
operating conditions are based on an experimental setup used by TNO mocouples. The reactor wall was traced at 400 ◦ C, so no adiabatic
for breakthrough measurements. The packed bed model has been vali­ operation. Over the entire operation time and reactor length, the tem­
dated using experimental data provided by TNO (IPCC, 2021). When perature range measured by the thermocouples was between
building a packed bed model, the main challenge is that species trans­ 400–415 ◦ C, with 415 ◦ C being the result of the exothermic effect caused
port and adsorption are occurring in dimensions of different orders of by the front of adsorption. The equilibrium CO2 loading on the K-HTC
magnitude: macropores between the pellets and micropores inside the adsorbent was calculated using the double-component isotherm pre­
pellets. A multiscale approach was chosen for the packed bed reactor sented in the paper and showed negligible differences at 400 ◦ C and
model where macroscale was represented by a 1D space geometry, more 415 ◦ C; therefore, the assumption of operating at isothermal conditions
specifically the bed height, while microscale was also described by a 1D would not interfere with breakthrough predictions. Similarly, the tem­
space geometry, the pellet radius. The packed bed was considered a perature increase during the bench-scale monolith reactor experiments
homogenous domain and was modelled as porous media. was not significant enough to affect CO2 loading; thus, operating con­
The bench-scale model for the monolith reactor was developed by ditions were assumed to be isothermal during modeling. For the sake of
taking into account the 3D-printed structure with hexagonal channels consistency, since both validated models (packed bed and bench-scale
(Fig. 1b) designed by TNO to undergo breakthrough experiments. structured bed) assumed isothermal conditions, the same was also
Instead of the traditional washcoat-substrate arrangement, the dividing considered for the pilot-scale monolith model.
wall between channels completely consists of sorbent K-HTC material SEWGS performance increases with a better utilization of the
built with 3D-printing technology. As experimental data was available adsorbent bed and is strongly correlated with adsorption mass transfer
for the lower-scale model, we decided on a more complex approach to efficiency, therefore, the focal point of the work was investigating CO2
capture as much of the phenomena occurring in order to properly vali­ adsorption mass transfer rate, thus the WGS reaction itself became of
date the model. A 3D approach was chosen for the geometry of the secondary importance to adsorption and was neglected in both experi­
structured bed model, in order to properly highlight the complex nature mental and modeling work. While there was no CO or H2 in the feed
of the process and phenomena involved. However, instead of simulating composition, the presence of steam was mandatory since the mechanism
the entire geometry (Fig. 1c) and solving the full 3D equations, plane of adsorption on K-HTC is competitive adsorption between steam and
symmetry was used to split the initial geometry into smaller parts. CO2.
Symmetry was assumed by using a vertical and a horizontal plane, There was no symmetry considered for the packed bed and pilot-
leading to a quarter of the entire geometry (Fig. 1d). By keeping the scale model; therefore, the equations had to be fully solved. For all
same element size when building the mesh, this approach used a fourth reactor simulations, fluid flow was solved initially at steady state, fol­
of the initial number of degrees of freedom, and thus reduced the overall lowed by using the steady state solution as an initial value in the sub­
computation time by at least 4 times. Two computational domains were sequent time dependent study of simultaneous momentum and mass
considered: the fluid domain and sorbent material domain modelled as transfer. To avoid inconsistent boundary conditions, a small inlet section
porous media. To further evaluate the mass transfer efficiency of the with assumed symmetry at the walls was added to the monolith channel
structured bed, three different channel geometry configurations were models. The mesh was refined until mesh convergence was achieved for
simulated. all models. In the case of the pilot-scale monolith model, as the length to
In regards to the pilot-scale monolith reactor model, the comparison width ratio was around 1200, the main geometry was split into smaller
between packed bed and monolith reactors was enabled by considering domains reducing the time needed for the meshing process.
the same mass of adsorbent for both technologies.
When considering model upscaling, an important consideration was
the need to find an acceptable balance between model complexity and 2.1. Assumptions
computational requirements, since CFD models tend to be on the higher
side of computational costs. The domain volume for the pilot-scale Common and reactor specific assumptions are presented below:
model was 386 times larger than the lower scale model, requiring an
immense amount of computational power when considering the same • Uniform distribution of sorbent activity and uniform gas distribution
level of model complexity. Therefore, a single channel with a 2D- among the sorbent were considered;
axisymmetric geometry (Fig. 1f) was chosen to model the monolith • Isothermal conditions were considered;
structure. By assuming evenly distributed flow and identical channels, it • Diffusion (Fick’s law) was assumed to occur across the entire porous
is possible to represent the entire monolith structure with a single matrix and was governed by an effective diffusion coefficient;
channel, as the governing equations are identical regardless of multi- or • Axial dispersion was considered;
single channel approach. Moreover, a single channel model requires • Water-gas shift reaction was not considered (focus on was adsorp­
much less computational effort, while being able to provide information tion, no CO in gas composition);
on the performance of the entire reactor. Mei et al. (2006) analyzed • Instantaneous competitive adsorption mechanism of CO2 and H2O
simulation results from models considering an entire reactor and a single onto sorbent surface was considered;
channel and discovered that simulation predictions were valid in both • Compressible fluid flow was considered;
cases. The main differences between the monolith models were scale,
operating conditions, channel shape (hexagonal vs. circular for the Packed bed reactor model
pilot-scale model) and geometry system (3D vs. 2D-axisymmetric),
while the equations describing the phenomena remained identical. • Sorbent pellets were considered to have a cylindrical shape with
For the packed bed configuration, three different flow rates were uniform size;

4
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

• At the pellet-fluid interface, film resistance was considered ( )

20 1
negligible; DLi = + ⋅2⋅uinterstitial ⋅rpe (10)
Re⋅Sc 2
Monolith reactor models Mass transfer inside the pellet was represented by the diffusion of
fluid components due to a concentration gradient and adsorption of CO2
• Behavior across all monolith channels was considered identical, and steam. Thus, a diffusion-adsorption equation (Eq. (11)) was solved
enabling the simulation of the entire structure through a single along the pellet radius for each of the components:
channel (pilot-scale); { ( ) }
• At the sorbent-fluid interface, mass transfer resistances were negli­ ∂c ∂ ∂cpe,i
4π N r2 rpe
2
εpe pe,i + − r2 Dpe,i = r2 rpe
2
RADS,i (11)
gible (film resistance ignored). ∂t ∂r ∂r
Eq. (11) took into consideration a dimensionless radial coordinate
2.2. Model equations going from the center to the pellet surface r, the pellet radius rpe , the
number of pellets per unit volume of bed N, component adsorption
Continuity Equation RADS,i . The equation was designed with spherical pellets in mind, but
The equations were solved in unsteady state for all models, therefore through the use of correlations (Eq. (12)), it was modified to describe
fluid flow was considered compressible with a dynamic density. The mass transfer in cylindrical pellets:
expression for the continuity equation (Eq. (4)) was: 3
rpe, sphere = ⋅rpe, cylinder (12)
∂ 2
(ρ) + ∇⋅(ρu) = 0 (4)
∂t Adsorption was calculated with the equilibrium theory (Eq. (13))
Ergun Equation – packed bed reactor model (Kumar, 1989) assuming instantaneous mass transfer, where component
The packed bed was considered uniform porous media due to being loading was described by the double component isotherm seen in Eq.
filled with adsorbent hydrotalcite pellets. Pressure distribution along the (19) for CO2 and Eq. (20) for H2O:
reactor was solved using the differential form of Ergun’s equation (Eq. dqi dqi dpi dqi dpj dqi dT
(5)) (Ergun, 1952), where εb was bed voidage, Dpe was pellet diameter, μ RADS,i = = ⋅ + ⋅ + ⋅ (13)
dt dpi dt dpj dt dT dt
was fluid viscosity, ρ was fluid density and u was velocity:
where q was loading amount, p was partial pressure and T was tem­
∂P μu (1 − εb )2 1.75ρu2 (1 − εb )
− = 150 2 ⋅ + ⋅ (5) perature.
∂x Dpe ε3b Dpe ε3b At the pellet-fluid interface, film resistance was considered negligible
Navier-Stokes and Brinkman Equations – monolith reactor models (Eq. (14)), where c was bulk concentration and cpe was pellet
The flow regimes inside the monolith channels were established to be concentration:
laminar for both models, as indicated by the Reynolds number between ci = cpe,i (14)
33 and 198, calculated for the pilot-scale model accounting for a fluid
density ρ = 10.036 kg/m3 and viscosity μ = 3.09⋅10− 5 Pa⋅s, while for Species transport – monolith reactor models
the bench-scale model, the Reynolds number was between 0.5 and 1.5, Mass transfer inside the monolith channels was described by a con­
calculated for a fluid density ρ = 0.306 kg/m3 and viscosity μ = vection diffusion expression (Eq. (15)):
2.47⋅10− 5 Pa⋅s. The laminar fluid flow was solved using the Navier- ∂ci ( )
+ ∇⋅ − DL,i ∇ci + u⋅∇ci = 0 (15)
Stokes equations for laminar flow (Eq. (6)) (the conservation of mass ∂t
(Eq. (4)) and the conservation of the momentum):
The axial dispersion coefficient DLi was estimated using the Aris-
∂u Taylor correlation for laminar flow, taking into account average veloc­
ρ + ρ(u⋅∇)u = ∇⋅(pI + τ) (6)
∂t ity, hydraulic diameter and molecular diffusion (Eq. (16)) (Baharudin
and Watson, 2017):
where τ was the viscous stress tensor (Pa), which was calculated using 〈 2〉 2
Eq. (7): u ⋅d
DLi = Di + z H (16)
192⋅Di
( ) 2
τ = μ ∇u + (∇u)T − μ(∇⋅u)I (7) Transport of species in the sorbent was described by a porous media
3
transport equation (Eq. 17) and accounted for convection, diffusion and
Flow inside the monolith walls was described by the Brinkman
adsorption:
equations (Eq. (8)) accounting for sorbent porosity εm and permeability
κ: ∂ci ( )
εm ⋅ + ∇⋅ − De,i ∇ci + u⋅∇ci = RADS,i (17)
( ) ∂t
1 ∂u 1 1 1 μ
ρ + ρ(u⋅∇)u = ∇⋅ pI + τ − (8) The component effective diffusion coefficient Dei was estimated by
εm ∂t εm εm εm κ
using the sorbent material porosity εm and tortuosity τ. Diffusion was
Species transport – packed bed reactor model assumed to be isotropic.
Mass transfer in the bed was represented by the convection and (ε )
diffusion of fluid components through the macropores and was solved De,i =
m
Di (18)
with a porous media convection-diffusion equation (Eq. (9)) accounting τ
for bed porosity εb , component molecular diffusion in the gas stream Di Adsorption took place in the porous domain and was calculated with
and component concentration ci : the equilibrium theory (Eq. (13)).

∂ci ∂2 ci ∂ci 2.3. Adsorption isotherm

εb ⋅ − DL i 2 + u⋅ = 0 (9)
∂t ∂x ∂x
The axial dispersion coefficient DLi was estimated using the Wakao A model based on two double isotherms (Boon et al., 2014) was
and Funazkri correlation in packed beds, based on the pellet Peclet, derived to take into account surface and nanopores sites contributions
Reynolds and Schmidt numbers (Eq. (10)) (Wakao and Funazkri, 1978): for CO2 and H2O adsorption at high pressures. The surface adsorption of

5
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Table 1 volume filling theory, where the assumption of no lateral interactions

Common and reactor specific parameters. between adsorbed species enabled the amount of gas adsorbed to be
Parameter Packed bed Monolith column Monolith reactor determined by the limiting nanopore volume V0, pore-size distribution
column (pilot-scale) (bench-scale) parameter m and the nanopore-sorbate interaction energy E (Ozawa
Column et al., 1976; Doong and Yang, 1988; Do, 1998). As such, in the case of
specifications CO2 (Eq. (19)) (Boon et al., 2014):
Column height [m] 2 1.8 0.036
Column diameter, 0.038 0.014 qsCO2 KCO2 pCO2 ACO2 (V0 − V0 AH2 O )
q∗CO2 = + (19)
dc [m] 1 + KCO2 pCO2 vm,CO2 (1 − ACO2 AH2 O )
Bed specifications
Bed height [m] 1.8 1.746 0.036 where qs was maximum amount adsorbed, K was surface-sorbate
Bed density, ρb [kg 980 540 interaction parameter, p was partial pressure, A was nanopore-sorbate
m− 3 ]
interaction parameter and vm was molar volume; and for H2O (Eq.
Material porosity, 0.3
(20)) (Boon et al., 2014):
εm [-]
Sorbent mass [kg] 2 0.0022 AH2 O (V0 − V0 ACO2 )
(20)
1/n
Pellet height [m] 0.005 - - q∗H2 O = KH2 O pH2 O +
vm,H2 O (1 − ACO2 AH2 O )
Pellet diameter, dpe 0.005 - -
[m]
3
Channel diameter - 0.866•10− 0.001
[m]
2.4. Model parameters
3 3
Channel wall - 0.575•10− 0.5•10−
thickness [m] Process, operating and geometry related parameters are presented in
Number of channels - 695 64 Table 1. For the packed bed model, three different flow rates were
[-]
simulated. Based on optimal channel dimensions, the number of chan­
Work pressure [Pa] 2•106 1•105
Feed temperature 678 648, 683, 723 nels for the pilot-scale monolith reactor was 695, resulting in a slightly
[K] shorter length and identical bed porosity. All other operating conditions
Inlet Flow rate [NL 25, 50, 150 0.05, 0.1, 0.15 were identical between the two configurations. The bench-scale mono­
min− 1] lith model was simulated at three different flow rates and temperatures.
Feed composition
N2 (diluent) [mol 90 -
%] 3. Results and discussion
CO2 [mol%] 4 30
H2O [mol%] 5 10
3.1. Packed bed model – CO2 breakthrough curves – all flow rate cases
Ar (inert) [mol%] 1 -
He (inert) [mol%] - 60
Fig. 2 shows the predicted breakthrough curves and times for all
three flow rate cases for the packed bed. CO2 concentration was plotted
CO2 was described by a Langmuir isotherm. For the surface contribution as a function of time. The markers were the experimental data, while the
of H2O, a Freundlich isotherm was used, as to reflect the heterogenous straight lines were the predicted data.
character of the sites for H2O. Nanopore contributions of both CO2 and The predicted curves were similar in shape and time of initial
H2O were described through a Dubinin-Astakhov equation based on breakthrough with the experimental curves. A relative concentration

Fig. 2. CO2 concentration plotted against time at the reactor outlet. Comparison of experimental data with model predicted data for all flow rate cases.

6
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 3. CO2 concentration plotted against normalized time at the reactor outlet. Comparison of experimental data with model predicted data for all flow rate cases.

Fig. 4. (a) CO2 concentration distribution at simulation times 500, 1000, 1500 s, (b) CO2 concentration in an arbitrarily chosen pellet at a bed height of 1 m at
simulation time 1000 s.

7
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 5. Parity plot showing the distribution of experimental versus predicted values of CO2 breakthrough times.

breakthrough point of 0.5 was considered. The breakthrough times were cases for the packed bed as a function of normalized time to enable a
almost identical across all flow rates when comparing the predicted and proper comparison between the shapes of the breakthrough curves. The
experimental data. Although the top sections of the curves did not normalized time was calculated by dividing the simulation time with the
overlap, this effect is seen and accepted in literature (Abouelella et al., time corresponding to a concentration of 2 mol%. The markers were the
2018). In this work, the reason for the deviation was assumed to be due experimental data, while the straight lines were the modelled data.
to the mass transfer limitations unaccounted for by the isotherm. For the first flow rate value of 25 NL/min, the curves were over­
Overall, the comparison between breakthrough curves suggests a good lapping completely until the upper section of the curve, followed by a
fitting of the data sets. deviation attributed to the adsorption mechanism. For the higher flow
Fig. 3 shows the predicted breakthrough curves for all three flow rate rates, an acceptable fit of the curves was seen, while the slight deviation

Fig. 6. Model predicted normalized CO2 concentration plotted against time at the reactor outlet.

8
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 7. (a), (b), (c) zigzag segments used in simulations, (d) normalized CO2 concentration plotted against normalized time predicted by the monolith models with
different channel configurations at atmospheric pressure.

in the bottom section could be attributed to axial dispersion. Fig. 3 also 3.3. Bench-scale monolith model – CO2 breakthrough curves
highlighted a trend that with increasing flow rate, the breakthrough
curves lost their sharpness, which was expected behavior and in accor­ The experimental and simulated breakthrough times for all tem­
dance with experimental data. perature and flow rate cases considered at a relative concentration
(fraction of outlet concentration over inflow concentration) of 0.5 for
3.2. Packed bed model - concentration distribution – 25 NL/min both experimental and simulated results were compared in Fig. 5 by
plotting the experimental average breakthrough times versus the model
Fig. 4 presents CO2 concentrations plotted against bed height at predicted breakthrough times. The coefficient of determination (R2)
various simulation times in the bulk, between the macropores, as well as resulted in a value of 0.89, indicating a good fit between experimental
inside the pellets. data and model predictions.
The bulk concentration of CO2 was overlapped by the concentration Fig. 6 shows the model predicted breakthrough curves for the
in the gas phase at the surface of the pellet. These values were identical monoliths at several temperatures and flow rates (nine cases) at low
because of the fluid-pellet interface boundary condition specified, where pressure conditions. Normalized CO2 concentration was plotted as a
film resistance was negligible. Fig. 4a also shows the concentration function of time. The normalized concentration was calculated by
gradient of CO2 in the gas phase at the center of the pellet. Considering a dividing the concentration at the simulation time by the inlet
randomly selected pellet at a random bed height of 1 m, Fig. 4a indicates concentration.
that the concentration values would range between 0 and 2.5 mol%. It can be noticed how breakthrough was achieved sooner at a higher
Fig. 4b shows a 2D surface graph highlighting CO2 concentration inside flowrate, as the sorbent reaches saturation capacity faster due to more
a cylinder pellet, with values along the pellet radius between center and CO2 coming into the system. Furthermore, temperature did not have a
surface ranging from 0 to 2.47 mol%. significant impact over breakthrough at any of the flow rate values, as
the curves in each of the 3 flow rate groups were almost overlapping.

9
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

which can be attributed to the induced mixing effects.

3.5. Pilot-scale monolith model - velocity profiles

The stationary flow study used component concentrations based on

the inflow composition values found in Table 1. Laminar velocity pro­
files in the axial and radial directions can be seen in Fig. 8, highlighting
the laminar nature of the fluid flow inside of the channel, as well as the
decreasing velocity inside of the porous domain, where gradients are
non-negligible.
The point of contact between the inlet section and channel can be
seen in Fig. 9. The inlet section assumed a slip boundary condition at the
walls, necessary in order to avoid inconsistent boundary conditions. The
resulting velocity profile shows a uniform average velocity inside of the
inlet section. The streamlines show details of the flow more clearly. After
the transition zone, the development of the flow can be noticed, as well
as the fully developed laminar flow inside of the channel.

3.6. Packed bed & monolith models – pressure drop predictions

Fig. 8. Velocity profiles calculated for the monolith reactor in the (a) axial One of the main advantages of a structured bed over a packed bed
direction and (b) radial direction. configuration is the lower pressure drop inside of the channels. There­
fore, correctly calculating the pressure drop is of the utmost importance
3.4. Bench-scale monolith model – zigzag channel configurations to properly describe the process, while also ensuring an accurate pre­
diction of the flow.
The effects of mass transfer efficiency in a structured bed were Predicted pressure distributions by the bench-scale monolith model
examined by simulating three different zigzag channel configurations at for 50 NmL/min and by the pilot-scale monolith model for 25 NL/min
100 NmL/min and 683 K. The three different segments considered are are shown in Fig. 10a and 10b, respectively. While pressure drop ex­
seen in Fig. 7a–c. These segments were multiplied until the desired periments were not done, the values were analytically validated
channel length was reached, after which all 64 channels were added (Table 2) using literature, with estimations calculated with the well-
together to represent the full monolith structure. known and proven Ergun equation (Pesic et al., 2015) for the packed
To compare mass transfer efficiency, normalized CO2 concentration bed reactor and Eq. (21) (Akbari et al., 2009), applicable for laminar
was plotted versus normalized time in Fig. 7d. A sharper profile suggests flow in microchannels, for the monolith reactor.
better mass transfer. It can be noticed that the least sharp breakthrough L
curve was the one predicted by the monolith model with straight ΔP = 16π2 μQIp∗ (21)
A2c
channels, while all monolith models with zigzag configurations pre­
dicted sharper breakthrough curves. The sharpest curve was the result of where Ip∗ was the specific polar momentum of inertia of the channel cross
the monolith model with the zigzag configuration using configuration a, section.

Fig. 9. Velocity profile with plotted streamlines at the monolith reactor entrance.

10
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 10. (a) Pressure drop contour calculated by the bench-scale monolith reactor model for a flow rate of 50 NmL/min; (b) pressure drop contour calculated by the
pilot-scale monolith model for a flow rate of 25 NL/min.

Table 2
Analytical and predicted pressure drop values for 25, 50, 150 NL/min for the packed bed and pilot-scale model, and 50, 100, 150 NmL/min for the bench-scale
monolith model.
Flowrate case PBΔP [Pa] PB SIMΔP [Pa] Pilot-scaleΔP [Pa] Pilot-scaleSIM ΔP [Pa] Bench-scaleΔP [Pa] Bench-scaleSIM ΔP [Pa]

#1 229.89 230 271 274 1.52 1.59

#2 714.95 715 542 554 3.03 2.99
#3 5241.19 5241 1628 1660 4.54 4.49

Both the analytical and predicted pressure drop values increased was expected as the sorbent mass was identical for all cases. It can be
with increasing velocity, as expected. Furthermore, the calculated and noticed that mass transfer was slower for the packed bed configuration,
predicted values corresponded well. Comparing the predicted pressure since the breakthrough curves were not as sharp as the monolith model
drop values for both reactor types, we can see a much lower drop in predicted breakthrough curve.
pressure for the monolith reactor, especially at higher flowrates. To properly investigate mass transfer efficiency, CO2 concentration is
plotted against normalized time (Fig. 13). A sharper breakthrough curve
3.7. Monolith reactor models – species transfer profile translates into less mass transfer resistance, although profile
changes are also an effect of axial dispersion. Comparing the break­
Mass transfer of CO2 predicted by both monolith models are shown in through curve profiles, the sharpest curve was by far the curve predicted
Fig. 11. Fig. 11a shows the concentration distribution at an arbitrarily by the monolith model, with an almost ideal breakthrough curve with
chosen simulation time of 80 s predicted by the bench-scale model. The instantaneous mass transfer, meaning the most efficient mass transfer
3D model captured the non-uniformity of the CO2 loading, where the was seen in the monolith configuration.
sorbent material from the inner region took up CO2 first as an effect of Fig. 14 compares mass transfer zones predicted by the packed bed
the contribution of multiple surrounding channels, followed by the and monolith models at an arbitrarily chosen simulation time of 900 s.
sorbent material from the outer region. Fig. 11b highlights the transport The length of the mass transfer zone depends on the mass transfer
mechanisms accounted for by the pilot-scale model at various times. rate. A higher mass transfer rate creates a shorter mass transfer zone.
Convection-dominated mass transfer inside the channel, diffusion inside The mass transfer zone calculated by the monolith model is much
the porous domain, as well as dispersion effects of the species inside the shorter than the packed bed one and the increase in CO2 concentration
channel, described by the Taylor model for laminar flow can be noticed. after breakthrough is going to be much steeper for the monolith reactor,
indicating that the bed is being utilized more efficiently.
3.8. Packed bed & monolith models – mass transfer efficiency
4. Conclusions
Fig. 12 shows the experimental and predicted breakthrough curves
for the packed bed configuration, as well as the predicted breakthrough A mathematical model of a packed bed reactor based on the equa­
curve by the monolith model for the flow rate of 25 NL/min. CO2 con­ tions describing mass and momentum transfer was developed to
centration was plotted as a function of time. describe the adsorption step of SEWGS. The developed mathematical
For the packed bed data sets, the profiles of the curves were very model was implemented in COMSOL Multiphysics and the simulation
similar, with breakthrough starting at around the point in both cases. A results were compared with the values determined experimentally by
breakthrough point was considered at a relative concentration (fraction TNO. Following the numerical simulations performed, a very good
of outlet concentration over inflow concentration) of 0.5 and the times concordance of the values resulting from the simulations was observed
predicted by the packed bed and monolith models were in agreement with the values obtained by experimental measurements. The compar­
with the experimental time (seen in the annotation in Fig. 12), which ison between the experimentally measured data and the simulation-

11
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 11. (a) 3D surface graph of normalized CO2 concentration predicted by the bench-scale model, 25 NmL/min at 80 s; (b) normalized CO2 concentration profiles
predicted by the 2D-axisymmetric pilot-scale model for 25 NL/min at 450 s and 900 s, respectively.

Fig. 12. Experimental (blue diamonds) and simulated (green) CO2 breakthrough curves for the packed bed reactor case and simulated (red) CO2 breakthrough curve
for the monolith reactor case. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).

12
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Fig. 13. CO2 concentration at the bed outlet plotted against dimensionless time defined relative to the respective breakthrough time for each case (seen in Fig. 12).

Fig. 14. Mass transfer zones calculated for packed bed (red) and monolith (blue) reactors at 900 s simulation time. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)

based ones was performed considering: the pressure drops down the to describe the adsorption step of SEWGS. The models were imple­
length of the reactor, the breakthrough time and the profile of the mented in COMSOL Multiphysics. A bench-scale model was developed
breakthrough curves. Mathematical models of a monolith reactors based and validated using breakthrough experiments performed by TNO using
on equations describing mass and momentum transfer were developed 3D-printed K-HTC monolith structures at atmospheric pressure. The

13
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

simulations were run at different flow rate values and temperatures, Gazzani, M., MacChi, E., Manzolini, G., 2013. CO2 capture in integrated gasification
combined cycle with SEWGS - part A: thermodynamic performances. Fuel 105,
leading to a good fit between experimental data and predictions, with
206–219. https://doi.org/10.1016/j.fuel.2012.07.048.
special focus on breakthrough times. Following a scaleup of the bench- Manzolini, G., Giuffrida, A., Cobden, P.D., van Dijk, H.A.J., Ruggeri, F., Consonni, F.,
scale model, a comparison between packed bed and monolith reactors 2020. Techno-economic assessment of SEWGS technology when applied to
took place with focus on mass transfer efficiency. The profile of the integrated steel-plant for CO2 emission mitigation. Int. J. Greenh. Gas. Control 94,
102935. https://doi.org/10.1016/j.ijggc.2019.102935.
breakthrough curve predicted by the monolith reactor model was much Ebner, A.D., Reynolds, S.P., Ritter, J.A., 2006. Understanding the adsorption and
sharper than the one for the packed bed, suggesting less mass transfer desorption behavior of CO2 on a K-promoted hydrotalcite-like compound (HTlc)
resistance and higher mass transfer rate. Furthermore, the mass transfer through nonequilibrium dynamic isotherms. Ind. Eng. Chem. Res. 45, 6387–6392.
https://doi.org/10.1021/ie060389k.
zone calculated by the monolith model was shorter when compared to Lee, K.B., Verdooren, A., Caram, H.S., Sircar, S., 2007. Chemisorption of carbon dioxide
the mass transfer zone corresponding to the packed bed model, indi­ on potassium-carbonate-promoted hydrotalcite. J. Colloid Interface Sci. 308, 30–39.
cating a more efficient utilization of the bed. These validated models can https://doi.org/10.1016/J.JCIS.2006.11.011.
Bhatta, L.K.G., Subramanyam, S., Chengala, M.D., Olivera, S., Venkatesh, K., 2015.
aid in the design of more complex 3D-structured sorbent materials to Progress in hydrotalcite like compounds and metal-based oxides for CO2 capture: a
further improve the productivity in the SEWGS process, by optimizing review. J. Clean. Prod. 103, 171–196. https://doi.org/10.1016/j.
the channel geometry for greater mixing effect, studying the desorption jclepro.2014.12.059.
Coenen, K., Gallucci, F., Pio, G., Cobden, P., van Dijk, E., Hensen, E., et al., 2017. On the
step of SEWGS in order to investigate mass transfer rate or extracting influence of steam on the CO2 chemisorption capacity of a hydrotalcite-based
certain coefficients in order to build simpler, less computationally adsorbent for SEWGS applications. Chem. Eng. J. 314, 554–569. https://doi.org/
intensive, but useful models in SEWGS development. 10.1016/J.CEJ.2016.12.013.
Palma, V., Ricca, A., Meloni, E., Martino, M., Miccio, M., Ciambelli, P., 2016.
Experimental and numerical investigations on structured catalysts for methane
CRediT authorship contribution statement steam reforming intensification. J. Clean. Prod. 111, 217–230. https://doi.org/
10.1016/j.jclepro.2015.09.004.
Akri, M., Achak, O., Granger, P., Wang, S., Batiot-Dupeyrat, C., Chafik, T., 2018.
Vlad-Cristian Sandu: Conceptualization, Methodology, Software,
Autothermal reforming of model purified biogas using an extruded honeycomb
Validation, Writing – original draft. Ana-Maria Cormos: Conceptuali­ monolith: a new catalyst based on nickel incorporated illite clay promoted with
zation, Funding acquisition, Supervision, Writing – review & editing. MgO. J. Clean. Prod. 171, 377–389. https://doi.org/10.1016/j.jclepro.2017.09.251.
Hosseini, S., Moghaddas, H., Masoudi Soltani, S., Kheawhom, S., 2020. Technological
Ionela-Dorina Dumbrava: Conceptualization, Methodology, Writing –
applications of honeycomb monoliths in environmental processes: a review. Process
original draft. Arpad Imre-Lucaci: Conceptualization, Resources, Su­ Saf. Environ. Prot. 133, 286–300. https://doi.org/10.1016/j.psep.2019.11.020.
pervision, Writing – review & editing. Calin-Cristian Cormos: Funding Ribeiro, R.P., Sauer, T.P., Lopes, F.V., Moreira, R.F., Grande, C.A., Rodrigues, AE., 2008.
acquisition, Supervision, Writing – review & editing. Robert de Boer: Adsorption of CO2, CH4, and N2 in activated carbon honeycomb monolith. J. Chem.
Eng. Data 53, 2311–2317.
Conceptualization, Project administration, Writing – review & editing. Verougstraete, B., Martín-Calvo, A., Van der Perre, S., Baron, G., Finsy, V., Denayer, J.F.
Jurriaan Boon: Conceptualization, Investigation, Writing – review & M., 2020. A new honeycomb carbon monolith for CO2 capture by rapid temperature
editing. Soraya Sluijter: Conceptualization, Investigation, Writing – swing adsorption using steam regeneration. Chem. Eng. J. 383, 123075. https://doi.
org/10.1016/j.cej.2019.123075.
review & editing. Kolaczkowski, S.T., 1999. Modeling catalytic combustion in monolith reactors –
challenges faced. Catal. Today 47, 209–218.
Heck, R.M., Gulati, S., Farruto, R.J., 2001. The application of monoliths for gas phase
Declaration of Competing Interest catalytic reactions. Chem. Eng. J. 82, 149–156.
Chen, J., Yang, H., Wang, N., Ring, Z., Dabros, T., 2008. Mathematical modeling of
The authors declare that they have no known competing financial monolith catalysts and reactors for gas phase reactions. Appl. Catal. A Gen. 345,
1–11. https://doi.org/10.1016/J.APCATA.2008.04.010.
interests or personal relationships that could have appeared to influence Pahinkar, D.G., Garimella, S., Robbins, T.R., 2015. Feasibility of using adsorbent-coated
the work reported in this paper. microchannels for pressure swing adsorption: parametric studies on
depressurization. Ind. Eng. Chem. Res. 54, 10103–10114.
Wright A.D., White V., Hufton J.R., Quinn R., Cobden P.D., van Selow E.R.. CAESAR:
Acknowledgments development of a SEWGS model for IGCC. Energy Procedia 2011;4:1147–54.
10.1016/J.EGYPRO.2011.01.167.
This work was supported by a grant of the Romanian National Au­ Reijers, H.T.J., Boon, J., Elzinga, G.D., Cobden, P.D., Haije, W.G., van den Brink, RW.,
2009. Modeling study of the sorption-enhanced reaction process for CO2 capture. I.
thority for Scientific Research and Innovation, CCCDI - UEFISCDI, Model development and validation. Ind. Eng. Chem. Res. 48, 6966–6974. https://
project number COFUND-ACT ERANET–3D-CAPS (contract number: doi.org/10.1021/ie801319q.
87/2017): "3D Printed Capture Materials for Productivity Step-Change", Irani, M., Alizadehdakhel, A., Pour, A.N., Hoseini, N., Adinehnia, M., 2011. CFD
modeling of hydrogen production using steam reforming of methane in monolith
within PNCDI III.
reactors: Surface or volume-base reaction model? Int. J. Hydrog. Energy 36,
15602–15610. https://doi.org/10.1016/J.IJHYDENE.2011.09.030.
References Abouelella, D.M., Fateen, S.E.K., Fouad, M.M.K., 2018. Multiscale modeling study of the
adsorption of CO2 using different capture materials. Evergreen 5, 43–51. https://doi.
org/10.5109/1929729.
IPCC, 2014. Climate Change 2014: Synthesis Report. Intergovernmental Panel on
ACT 3D-CAPS project. Final technical report [to be published] 2021.
Climate Change, Geneva, Switzerland. Report, Contribution of Working Groups I, II
Mei, H., Li, C., Liu, H., Ji, S., 2006. Simulation of catalytic combustion of methane in a
and III to the Fifth Assessment Report of the.
monolith honeycomb reactor. Chin. J. Chem. Eng. 14, 56–64.
da Silva Veras, T., Mozer, T.S., da Costa Rubim Messeder dos Santos, D., da Silva
Ergun, S., 1952. Fluid flow through packed columns. Chem. Eng. Prog. 48, 89–94.
César, A., 2017. Hydrogen: trends, production and characterization of the main
Wakao, N., Funazkri, T., 1978. Effect of fluid dispersion coefficients on particle-to-fluid
process worldwide. Int. J. Hydrog. Energy 42, 2018–2033. https://doi.org/10.1016/
mass transfer coefficients in packed beds. Correlation of sherwood numbers. Chem.
j.ijhydene.2016.08.219.
Eng. Sci. 33, 1375–1384. https://doi.org/10.1016/0009-2509(78)85120-3.
IPCC. Carbon Dioxide Capture and Storage. Report, Prepared by Working Group III of the
Kumar, R., 1989. Adsorption column blowdown: adiabatic equilibrium model for bulk
Intergovernmental Panel on Climate Change, Cambridge, England, 2005: 2005.
binary gas mixtures. Ind. Eng. Chem. Res. 28, 1677–1683. https://doi.org/10.1021/
Lee, J.M., Min, Y.J., Lee, K.B., Jeon, S.G., Na, J.G., Ryu, H.J., 2010. Enhancement of CO2
ie00095a017.
sorption uptake on hydrotalcite by impregnation with K2CO3. Langmuir 26,
Baharudin, L., Watson, M.J., 2017. Monolithic substrate support catalyst design
18788–18797. https://doi.org/10.1021/la102974s.
considerations for steam methane reforming operation. Rev. Chem. Eng. 34,
Boon, J., Cobden, P.D., van Dijk, H.A.J., Hoogland, C., van Selow, E.R., van Sint
445–563.
Annaland, M., 2014. Isotherm model for high-temperature, high-pressure adsorption
Ozawa, S., Kusumi, S., Ogino, Y., 1976. Physical adsorption of gases at high pressure. IV.
of CO2 and H2O on K-promoted hydrotalcite. Chem. Eng. J. 248, 406–414. https://
An improvement of the Dubinin—Astakhov adsorption equation. J. Colloid Interface
doi.org/10.1016/J.CEJ.2014.03.056.
Sci. 56, 83–91. https://doi.org/10.1016/0021-9797(76)90149-1.
Boon, J., Cobden, P.D., van Dijk, H.A.J., van Sint Annaland, M., 2015. High-temperature
Doong, S.J., Yang, R.T., 1988. A simple potential-theory model for predicting mixed-gas
pressure swing adsorption cycle design for sorption-enhanced water-gas shift. Chem.
adsorption. Ind. Eng. Chem. Res. 27, 630–635. https://doi.org/10.1021/
Eng. Sci. 122, 219–231. https://doi.org/10.1016/J.CES.2014.09.034.
ie00076a017.
van Selow, E.R., Cobden, P.D., van den Brink, R.W., Hufton, J.R., Wright, A., 2009.
Performance of sorption-enhanced water-gas shift as a pre-combustion CO2 capture
technology. Energy Procedia 1, 689–696. https://doi.org/10.1016/J.
EGYPRO.2009.01.091.

14
V.-C. Sandu et al. International Journal of Greenhouse Gas Control 111 (2021) 103447

Do, D.D., 1998. Adsorption Analysis: Equilibria and Kinetics. Imperial College Press, temperatures. Chem. Ind. Chem. Eng. Q. 21, 419–427. https://doi.org/10.2298/
London, UK. https://doi.org/10.1142/p111 vol. 2vol. 2. CICEQ140618044P.
Pesic, R., Radoicic, T.K., Boskovic-Vragolovic, N., Arsenijevic, Z., Grbavcic, Z., 2015. Akbari, M., Sinton, D., Bahrami, M., 2009. Pressure drop in rectangular microchannels as
Pressure drop in packed beds of spherical particles at ambient and elevated air compared with theory based on arbitrary cross section. J. Fluids Eng. Trans. ASME
131, 0412021–0412028. https://doi.org/10.1115/1.3077143.

15