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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

Available online at www.sciencedirect.com

ScienceDirect

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

Parametric study of pressure swing adsorption

cycle for hydrogen purification using Cu-BTC

Jinsheng Xiao a,b,*, Liang Fang a, Pierre B
enard b, Richard Chahine b

a
Hubei Key Laboratory of Advanced Technology for Automotive Components and Hubei Collaborative Innovation
Center for Automotive Components Technology, School of Automotive Engineering, Wuhan University of Technology,
Hubei, 430070, China
b
Hydrogen Research Institute, Universit
e Du Qu
ebec a
 Trois-Rivieres, QC, G9A 5H7, Canada

article info abstract

Article history: Metal organic framework (MOF), for example Cu-BTC, has the characteristics of structure
Received 13 February 2018 adversity, high pore volume, large surface area and strong selectivity. It is being considered
Received in revised form as a new adsorbent in the field of pressure swing adsorption (PSA). A model describing
5 May 2018 hydrogen mixture flow, heat and mass transfer with multi-component adsorption is
Accepted 9 May 2018 developed for predicting breakthrough curves and performance of PSA cycles in the
Available online 1 June 2018 hydrogen purification system using Cu-BTC as adsorbent. The model is implemented on
Aspen platform and validated by experiments. Hydrogen purification performances (purity,
Keywords: recovery, productivity) were evaluated, and parametric study on the performance of
Hydrogen purification hydrogen purification has been performed. The results show that the simulated mole
Pressure swing adsorption fractions, temperature and pressure in the PSA cycles agree with the experiments very
Metal organic framework well. In general, the variation trend of hydrogen purity is opposite to that of recovery and
Heat transfer productivity. As the parametric study shows, within a certain range, higher adsorption
Simulation pressure, shorter feeding time and lower feeding flow rate lead to higher hydrogen purity,
Optimization then lower recovery and productivity. Parametric studies help to effectively improve
hydrogen purification performance in the Cu-BTC adsorption bed. Furthermore, a multi-
objective algorithm is needed to optimize the PSA process.
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

pervasive application in the separation and purification of

Introduction hydrogen. Some new materials like MOF are valuable to be
studied as adsorbents in PSA cycles. Breakthrough experiment
With the heavy use of fossil fuels, environmental pollution is is a very important step of PSA cycle, so the study on PSA cycle
increasing and environmental problems are getting more and should start from breakthrough curves. Casas N and his col-
more serious. At this time, hydrogen becomes the focus of leagues used activated carbon to do breakthrough experi-
research because of its many advantages, such as high heat ments of two-component mixture H2/CO2 [2]. More
value, no pollution and so on [1]. But now a main reason why experiments were carried out for different parameters such as
hydrogen energy is not widely available is that it is difficult to pressure, temperature and species concentration, then sim-
purify. As pressure swing adsorption technology has been ulations of breakthrough curves were performed using a

du Que
* Corresponding author. Hydrogen Research Institute, Universite
bec a
 Trois-Rivie
res, QC, G9A 5H7, Canada.
E-mail address: [email protected] (J. Xiao).
https://doi.org/10.1016/j.ijhydene.2018.05.054
0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13963

CO2, but also H2/CO/CO2 and H2/CO/CO2/CH4/N2 multicompo-

Nomenclature nent gases [6]. Gomez L F et al. used MIL-53(Al) to do some
breakthrough experiments of binary CO2/CH4 mixture [7]. Yang
ap particle external specific area, m
1
J et al. set experiments to study the dynamics of PSA on a zeolite
bi extended Langmuir isotherm parameter, bar
1
5A bed using H2/CO2 and H2/CO, and got the best pressure
ci molar concentrations of ith component, mol/
condition in the process of adsorption and desorption [8]. Lee J J
m3
et al. improved the structure of adsorption bed and used a dual
c molar concentrations of mixture, mol/m3
activated carbon bed to simulate the breakthrough curves of
cps specific heat capacity of solid phase, J/kg/K
four-component mixture H2/CO/CO2/CH4, they found that the
cpw specific heat capacity of wall, J/kg/K
dual bed could not only reduce the spatial occupancy, but also
Cpg heat capacity of gas phase, J/mol/K
effectively reduce the influence of thermal effect, and improve
Cpi heat capacity of adsorbed phase, J/mol/K
the purity of hydrogen [9]. Jee J G et al. used activated carbon
db internal bed diameter, m
and zeolite as the layer of adsorbent and studied the influence
dp particle diameter, m
of adsorption pressure, carbon ratio on breakthrough curves
DL axial dispersion coefficient, s/m2
and bed temperatures [10]. There is a lot of heat exchange in
hgs heat transfer coefficient between gas and solid
PSA process, so Ahn S et al. used activated carbon and zeolite
phase, W/m2/K
5A as adsorption beds to set some experiments to make
hin heat transfer coefficient between gas phase and
hydrogen recovery from coal gas and study the thermal effects
bed wall, W/m2/K
[11]. More breakthrough experiments were performed on a
hout heat transfer coefficient between wall and
zeolite 5A bed [12], and the layered bed composed with zeolite
environment, W/m2/K
5A and activated carbon [13].
DHi heat of adsorption of ith component, J/mol
On the basis of the breakthrough curves, it is necessary to
ki mass transfer coefficient of ith component, 1/s
study the process of PSA cycles to do purification. In order to
KLg thermal conductivity of gas phase, W/m/K
do pre-combustion CO2 capture, Casas N et al. also did some
KLs thermal conductivity of solid phase, W/m/K
PSA experiments and simulations, but it was a little different
Kw thermal conductivity of wall, W/m/K
from PSA experiments for purification because he added two
madsorbent the mass of adsorbent, kg
steps before and after pressure equalization step [14,15]. You
Mi molar weight of ith component, kg/mol
Y W and his colleagues developed two-bed experiments and
n*i equilibrium adsorption amount, mol/kg
did some study on P/F ratio and adsorption pressure in order
ni dynamic adsorption amount for each
to purify hydrogen, but simulated temperatures did not fit
component, mol/kg
very well with experiment data [16]. Lopes F V S et al. set five-
nsi saturation adsorption amount for each
step fast-cycling VPSA experiments on an activated carbon
component, mol/kg
bed, and got high purity hydrogen with high recovery [17],
p pressure, bar
they also did simulation on step time and flow rates and
t time, s
studied its effects on the purity of hydrogen [18]. And J Y et al.
T temperature of adsorption bed, K
set experiments on two zeolite 5A beds using two components
Tg gas temperature, K
(H2/CO and H2/CH4) and built a mathematical model to do
Ts solid phase temperature, K
research on the step time of PSA cycles [19]. Cruz P et al. did
Tw wall temperature, K
some research on producing oxygen from air by PSA and VSA
vz Darcy's velocity, m/s
(vacuum swing adsorption), moreover, they also did some
yi gas molar fraction of ith component
parametric study [20]. Yavary M et al. built two-bed and six-
z axial direction, m
bed PSA models using zeolite5A, some parametric study
Greek symbols were also been done to study the performance of hydrogen
εb bed porosity purification, but the simulation lacked some experiments for
εp particle porosity validation [21]. Lee J J et al. developed a spatial occupancy PSA
dw wall thickness, m process which can be used in hydrogen station using heat
m dynamic viscosity, m/s exchange [9]. Experiments and simulations on PSA cycles
rb bed density, kg/m3 were also done on layered bed which included zeolite 5A,
rp particle density, kg/m3 activated carbon and other adsorbent [22,23]. In addition to
rs skeletal density, kg/m3 activated carbon and zeolite, some new materials are applied
rw wall density, kg/m3 to PSA as adsorbents. Chowdhury P et al. studied the
adsorption properties of MIL-101 and Cu-BTC [24]. Silva B et al.
used the Cu-BTC as adsorbents and carried out a series of
model with a piping system to describe the distance between experiements of breakthrough curves and PSA cycles, they
the outlet of the adsorption bed and detector [3,4]. They also also evaluated the performance of hydrogen purification
did some breakthrough curve experiments on MOF and UiO-67/ through their simulations [25]. Agueda V I and his colleagues
MCM-41 adsorbents to learn about the material adsorption did some study on MOF, and they found that multicomponent
properties [5]. Lopes F V S et al. also did some breakthrough gas has a longer breakthrough time on MOF compared to
experiments and VPSA (Vacuum Pressure Swing Adsorption) traditional adsorption material such as zeolite, and it is more
experiments on activated carbon bed but not only just using H2/ suitable for the application of PSA for hydrogen purification
13964 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

[26]. Further research on PSA should be performance optimi-

Table 2 e Extended Langmuir moldel parameters, mass
zation by using new adsorbent such as MOF.
transfer cofficient and specific heat capacity of adsorbed
The authors of current paper have built and validated a phase.
mathematical model of PSA process for hydrogen purification
Component nsi bi0 bi1 ki Cpi
[27], and simulated breakthrough curves in simple beds of
(mol/kg) (10
4 bare1) (J/mol) (s
1) (J/mol/K)
zeolite and activated carbon [28] and the layered bed
CO 4.49 4.71 21550 26.87 36.7
composited with these two materials [29]. Through previous
CO2 15.0 0.372 23004 23.27 37.937
research, study on the performance and optimization of PSA H2 1.46 12.3 10620 72.00 28.878
cycle are valuable to carry out for hydrogen purification. N2 7.45 2.69 12514 24.47 29.282
The objective of this work is to build a model of break- CH4 7.25 1.96 16584 25.8 36.205
through and PSA process in Cu-BTC bed for hydrogen purifi-
cation on Aspen Adsorption. This model was applied for
Table 3 e Operating conditions used in breakthrough
simulation after its validation by breakthrough and PSA ex-
experiments.
periments [25]. This work not only evaluated the performance
but also did parametric study on the performance of hydrogen Feed molar fraction Feeding rate Starting time
(10
6m3,s
1) of desorption (s)
purification. The influence of adsorption pressure and flow
rate in breakthrough process, and adsorption pressure, H2: CO2:CH4 ¼ 0.355:0.47:0.175 9.9 6550
H2: CO2:N2 ¼ 0.453:0.339:0.208 9.53 6700
feeding time and feeding rate in four-step PSA cycles for
H2: CO2:CO ¼ 0.559:0.253:0.188 9.42 9700
hydrogen purification performance are studied.
Adsorption pressure: 2 bar, Feeding temperature: 303 K, Purging
gas: H2.

Model description
where KLg is the thermal conductivity of gas phase and Cpg is
Mass conservation equations heat capacity of gas phase, cg is the gas molar fraction, p is the
bed pressure, hgs is the heat transfer coefficient between gas
Each component obeys the mass conservation equations, phase and solid phase, ap means particle external specific
which can be known as area, hin is heat transfer coefficient between gas and bed wall,
db stands for bed inner diameter. Tg , Ts and Tw are gas phase
vci vðvz ci Þ v2 ci vni
εb þ ¼ εb DL 2
rb ; i ¼ 1; :::; N (1) temperature, solid phase temperature and wall temperature
vt vz vz vt
separately. The term pvvz =vz in the gas phase energy equation
where ci is the molar concentration of each component. εb Eq. (2) is used to expressed the effect of compression, which is
means bed porosity. DL stands for the axial dispersion coeffi- the reversible rate of internal energy increase per unit volume
cient. vz means the Darcy's velocity, ni represents the by compression.
adsorption amount of species i. rb is the bed density. z is the The energy conservation equation for the solid phase is
bed axial position and t is time.
X X N  
vTs N

 vTs v2 Ts vni
rs cps þ rs Cpi ni ¼ KLs 2 þ rs DHi
Energy conservation equation vt i¼1
vt vz i¼1
vt


þ ap hgs Tg
Ts (3)
In the process of PSA, temperature is changed and energy
conservation equation is observed along the adsorption bed. where KLs is the thermal conductivity of solid phase, rs is
The energy conservation equation includes gas phase, solid skeletal density, cps stands for the heat capacity of solid phase,
phase and wall. The energy conservation equation for gas can Cpi means the heat capacity of adsorbed phase, DHi is the
be known as adsorption heat of component i in Cu-BTC bed.
The energy conservation equation for the column wall is
vTg vTg v2 Tg vvz

εb cg Cpg þ cg Cpg vz ¼ εb KLg 2
p þ ap hgs Ts
Tg


vt vz vz vz vTw v2 Tw 4db hin Tg
Tw
4ðdb þ dw Þhout Tw
Tf
rw cpw ¼ Kw 2 þ
4hin
 vt vz 2
ðdb þ dw Þ
d2b
þ Tw
Tg (2)
db (4)

Table 1 e Cu-BTC adsorbent properties parameters and adsorption bed characteristics.

Parameter Value Parameter Value
Internal bed diameter, db (m) 0.021 Bed porosity, εb 0.52
Bed length, L (m) 0.31 Wall specific heat, cpw (J/kg/K) 500
Particle specific heat, cps (J/kg/K) 1457 Gas heat conductivity, KLg (W/m/K) 0.4
Particle radius, Rp (m) 0.0015 Solid heat conductivity, KLs (W/m/K) 0.7
Particle density, rp (kg/m3) 823 Wall heat conductivity, Kw (W/m/K) 16
Skeletal density, rs (kg/m3) 1379 Gas/solid heat transfer coefficient, hgs (W/m2/K) 75
Wall density, rw (kg/m3) 8238 Inner wall heat transfer coefficient, hin (W/m2/K) 50
Particle porosity, εp 0.78 Outer wall heat transfer coefficient, hout (W/m2/K) 20
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13965

where Kw is thermal conductivity, rw is density, cpw is heat where ni and nsi are the equilibrium and saturation adsorption
capacity, dw is thickness of the wall, hout is heat transfer co- amount. pi is the pressure of each component, bi is a param-
efficient between the wall and external environment. eter in extended Langmuir isotherm, which can be known as.
pi ¼ yi p; bi ¼ bi0 expðbi1 =TÞ; i ¼ 1; :::; N
Momentum conservation equation The adsorption kinetics of gas adsorption in Cu-BTC bed
can be expressed by the linear driving force (LDF) model:
Ergun equation can be used to describe the given pressure
vni

gradient and Darcy's velocity [30]: ¼ ki n*i
ni ; i ¼ 1; :::; N (7)
vt
dp 150mð1
εb Þ2 rð1
εb Þ ! where ki is the mass transfer coefficient of each component.

¼ vz þ 1:75 vz j v j (5)
dz d2p ε3b dp ε3b
Equation of state
where m is the dynamic viscosity, dp is the particle diameter.

We can know that the concentration of the mixture c ¼ n=V ¼

Adsorption isotherms and kinetics
p=ðRTÞ from the ideal gas equation of state pV ¼ nRT. Then the
molar concentration of component i can be written as
To describe the adsorption isotherms, the extended Langmuir
model is given as ni ni =V ci ci RT
yi ¼ ¼ ¼ ¼ ; i ¼ 1; :::; N (8)
n n=V c p
bi pi
n*i ¼ nsi PN ; i ¼ 1; :::; N (6)
1þ j¼1 bj pj
The density of the mixture can be calculated by

Table 4 e Operating conditions used in four-step PSA cycles.

Feeding mixture and composition Feeding rate (10
4 mol s
1) Times in PSA cycle (s)
Pressurization Feed Depressurization Purge
H2:CO2:CH4 ¼ 0.78:0.18:0.04 4.46 50 212 62 192
H2:CO2:N2 ¼ 0.67: 0.28:0.05 3.95 50 282 62 192
H2:CO2:CO ¼ 0.68: 0.27:0.05 3.88 50 212 62 192

Adsorption pressure: 3.5 bar, Purge pressure: 1.0 bar, Feeding temperature: 303 K

Fig. 1 e PSA cycle flow chart in Aspen Adsorption.

13966 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

Model parameters
X
N
r ¼ cM ¼ c yi Mi (9)
i¼1 Material properties and operation conditions
where M is the molar weight of the mixture, Mi is the molar
weight of each component. The Cu-BTC adsorbent properties parameters and adsorption
The specific heat capacity of the mixture is calculated by bed characteristics are shown as Table 1, and other parame-
molar fraction weighted average form as ters like LDF coefficient, extended Langmuir model parame-
ters in Cu-BTC are given in Table 2. The operating parameters
X
n
for breakthrough experiment including adsorption and
Cpg ¼ yi Cpi (10)
i¼1 desorption process are show in Table 3. And Table 4 shows the
operating conditions used in four-step (pressurization, feed,
depressurization and purge) PSA cycles.

Fig. 2 e Simulated (line) and experimental (symbol) [25] results of breakthrough curves (a) and temperature profiles at three
thermocouple positions (b) for the mixture H2/CO2/CH4.
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13967

PSA cycle flow chart feed pure H2 through U1 to W1 to make the bed can be
reused.
Simulation was conducted based on Aspen Adsorption. Fig. 1 The finite difference method is adopted Aspen Adsorption
shows the PSA cycle flow chart in Aspen Adsorption. In for the PSA simulation. In this simulation process, after
breakthrough experiment, just use the inlet (F1), bed and sensitivity analysis, the adsorption bed is divided into 20 nodes
outlet (P1), what needs illustration is that, after adsorption, along length of the bed and the time step size is set as 10s.
pure H2 will feed in the bed in order to desorption. The four-
step PSA cycle for hydrogen purification includes pressuri-
zation, feed, depressurization and purge. In pressurization Model validation
step, just feed through F1 to make the pressure higher. After
that, feed the mixture gas through F1 to P1 and the pure For model validation, the simulation results are compared
hydrogen will be got out of P1. In depressurization step, use with experimental values which come from Bruna Silva et al.'s
W1 to make the adsorbates get out of the bed. In purge step, breakthrough and PSA experiments [25].

Fig. 3 e Simulated (line) and experimental (symbol) [25] results of breakthrough curves (a) and temperature profiles at three
thermocouple positions (b) for the mixture H2/CO2/N2.
13968 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

Model validation by breakthrough experiments saturated gas breakthrough. Cu-BTC has stronger adsorption
capability for CO2 than other gases, so carbon dioxide is the
The comparisons of molar fractions and temperatures are last one to breakthrough. After about 1500s, molar fraction no
shown in Figs. 2e4. longer changes because the adsorption reaches equilibrium.
Breakthrough curve is an important subject to study the At the desorption step, with the feeding of H2, the molar
performance of adsorption process, it is a curve of gas molar fraction of CO2, CH4, N2 and CO drop down, while the drop rate
fraction at the outlet of adsorption bed over time. Fig. 2(a) is slower, so only using pure H2 under feeding condition to
shows the evolutions of molar fractions of each components purge the bed is unreasonable, because it will lengthen the
with ternary gases (H2, CO2, CH4) which were simulated in cycle time and decrease the system recovery.
Aspen Adsorption, so are Fig. 3(a) and Fig.4(a), but for different Fig. 2(b) shows the gas phase temperature for the ternary
gas mixture. The simulation enjoyed good agreements with gas mixture (H2, CO2, CH4), Figs. 3(b) and Fig.4(b) also show the
experimental data. At the initial time, the bed is full of gas temperatures but for other mixtures. The three tempera-
hydrogen. As the mixture gases continue to be charged, the ture detection points are placed at the entrance position (2 cm),

Fig. 4 e Simulated (line) and experimental (symbol) [25] results of breakthrough curves (a) and temperature profiles at three
thermocouple positions (b) for the mixture H2/CO2/CO.
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13969

middle position (12.5 cm) and exit position (24 cm) of the
adsorption bed. As the mixture gases continue to be charged,
the adsorbent adsorbs mixture, which releases a lot of
adsorption heat and makes the bed temperature keep rising
until getting peak. Subsequently, the bed temperatures drop
until approaching the environmental temperature because of
the heat exchange between bed wall and the outside. The bed
temperature at 2 cm height rises faster than 12.5 cm and 24 cm,
because the adsorption occurs along the inlet of the bed.
After a while, the temperature drops during desorption and
approaches the environmental temperature finally. And we
can know that the time spent in the desorption process is
obviously longer than that in the adsorption process, so it is
important to reduce the pressure before purge, otherwise it
will cost a lot of time and pure hydrogen.
The solid temperature is 1.6 K higher than gas temperature
in the current case when the heat transfer coefficient (hgs ) is
75 W/m2/K. Increasing hgs will decrease the difference be-
tween the gas or solid phase.

Model validation by PSA cycle operations

In the last chapter, pure hydrogen is used to flush the bed to

make the bed regeneration. It not only takes a long time, but
also requires a large amount of pure hydrogen, which affects
the hydrogen recovery of the system. Therefore, in order to
ensure the performance of hydrogen purification, pressure
should be changed among PSA cycle. It is very important to
simulate the four-step (pressurization, feed, depressurization
and purge) PSA cycles.
Fig 5(a) and (b) show pressure and temperature profiles in
the bed of ternary component mixtures (H2/CO2/CH4) in PSA
cycles. In the case of higher pressure, adsorbents have a larger
adsorption capacity for gas, which is conducive to the sepa-
ration of gas adsorption. So during feed step, CO2 and CH4 are
adsorbed under 3.5 bar, temperature rises. And in the case of
lower pressure, the adsorption capacity of adsorbent is
weaker, which can facilitate the regeneration of the bed. So
during purge step, CO2 and CH4 are desorbed under 1 bar,
temperature drops.
Fig. 5(c) and (d) show the molar fraction for cycles 1, 2, 3 and
cycles 14, 15, 16 of ternary component mixtures (H2/CO2/CH4)
in PSA cycles. During depressurization, CO2 and CH4 are des-
orbed, their fractions rise, but in Cu-BTC bed, there are still a
little CO2 and CH4 adsorbed, so purge step is needed to make
bed clear and can be reused. During purge, fractions of CO2
and CH4 decrease as H2 is introduced. During the whole PSA
cycle, the sum of the mole fractions of all gases equals 1. The
simulated results enjoy good agreements with experimental
data. Similar to the simulation with ternary component mix-
tures (H2/CO2/CH4), the simulations with H2/CO2/N2 and H2/
CO2/CO of molar fraction are observed in Fig. 6.
The performance of PSA cycles is evaluated by purity, re-
covery and productivity of hydrogen [25]: Fig. 5 e Comparison between simulated (line) and
Z tfeed experimental (symbol) [25] results of pressure curves (a),
cH2 Vz jz¼L dt temperature profiles near the column bottom (b), molar
Purity ¼ 0
Z ð%Þ (11) fractions during PSA cycles 1,2,3 (c) and cycles 14,15,16 (d)
P
N tfeed
ci Vz jz¼L dt for the mixture H2/CO2/CH4.
i¼1 0
13970 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

Fig. 6 e Comparison between simulated (line) and experimental (symbol) [25] results of molar fractions for the mixture H2/
CO2/N2 (a) and H2/CO2/CO (b) during PSA cycles 1,2,3.

Z tfeed Z tpurge
cH2 Vz jz¼L dt
cH2 Vz jz¼L dt
Recovery ¼ Z tfeed
0 0
Z tpressurization ð%Þ (12) Table 5 e Performances of hydrogen purification
evaluated by simulation for three mixtures.
cH2 Vz jz¼0 dt þ cH2 Vz jz¼0 dt
0 0
Feeding mixture and Productivity
composition Purity Recovery (mol/kg/h)
Z tfeed Z tpurge
(%) (%)
cH2 Vz jz¼L dt
cH2 Vz jz¼L dt
p 2
Productivity ¼ db 0 0
ðmol=kg=hÞ H2:CO2:CH4 ¼ 0.78:0.18:0.04 99.867 46.767 8.488
4 tcycle madsorbent =3600 H2:CO2:N2 ¼ 0.67: 0.28:0.05 98.249 49.803 8.471
(13) H2:CO2:CO ¼ 0.68: 0.27:0.05 99.696 47.753 6.758
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13971

Fig. 7 e Effects of feeding flow rate (a) and adsorption pressure (b) on breakthrough features of the mixture H2/CO2/
CH4 ¼ 0.355:0.47:0.175.

Table 6 e Parametric study of hydrogen purification performances for the mixture H2/CO2/CH4 ¼ 0.78:0.18:0.04.
Factors and values Purity (%) Recovery (%) Productivity (mol/kg/h)
Adsorption pressure (bar) 2.5 99.6621 64.074 8.71
3.5 99.8672 46.767 8.49
5 99.9269 40.798 7.76
7 99.9296 33.268 7.03
Feeding flow rate (mol/s) 2.46  10
4 99.9766 34.495 5.18
4.46  10
4 99.8672 46.767 8.49
6.46  10
4 99.4729 59.187 12.3
8.46  10
4 98.9534 69.363 16.4
Feeding time (s) 192 99.9098 46.697 8.15
212 99.8672 46.767 8.49
232 99.8029 48.825 9.29
252 98.7041 49.948 9.54
13972 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

Fig. 8 e Effects of adsorption pressure (a), feeding flow rate (b) and feeding time (c) on hydrogen purity, recovery and
productivity from the mixture H2/CO2/CH4 ¼ 0.78:0.18:0.04.
 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4 13973

From definitions, larger purity will induce, inversely, lower and conversely lower purity, because faster feeding flow rates
recovery and lower productivity. The values of these perfor- mean more mixture gases in a certain time, and cause more
mance parameters for hydrogen purification from the mix- CO2 and CH4 in the mixture gases in the outlet, which can lead
tures H2/CO2/CH4, H2/CO2/N2 and H2/CO2/CO are presented in to the lower purity but high recovery and productivity.
Table 5. In order to study the influence of feeding time in PSA
process, different feeding time (192s, 212s, 232s, 252s) were set
respectively. We take ternary component mixtures (H2/CO2/
Parametric study CH4) to analyze the performance of PSA cycles. From Fig. 8(c),
we can conclude that longer feeding time induces higher re-
Parametric studies are performed for both the breakthrough covery and productivity, and conversely lower purity, because
curves and the PSA cycles. We will study the effects of feeding longer feeding time means more mixture gases in the
flow rate and adsorption pressure on the breakthrough curves adsorption process, and causes more CO2 and CH4 in the
of gas adsorption in Cu-BTC bed, and the effects of adsorption mixture gases in the outlet, which can lead to the lower purity
pressure, feeding flow rate and feeding time on the perfor- but high recovery and productivity.
mance of PSA cycles.

Parametric study on breakthrough curves Conclusions

The study of breakthrough curves of multicomponent gas is A model describing hydrogen mixture flow, heat and mass
particularly important for the performance of hydrogen pu- transfer with multi-component adsorption has been devel-
rification and designing PSA cycle, so the effects of feeding oped for predicting breakthrough curves and PSA cycles by
flow rate and adsorption pressure on breakthrough curves are using Cu-BTC as adsorbent. The model was built on Aspen
studied in this session. Adsorption and validated, the simulations of breakthrough
Fig. 7(a) shows the breakthrough curves for H2, CO2 and CH4 and PSA cycles fit well with experimental data from the
under three different feeding flow rates (6.9 cm3/s, 9.9 cm3/s, reference in both breakthrough curves and PSA cycles in Cu-
12.9 cm3/s) to analyze its influence of performance of BTC adsorption bed. Hydrogen purification performances
hydrogen purification. The different symbols stand for the (purity, recovery, productivity) were evaluated, and para-
different simulation results under three different feeding flow metric study on breakthrough curves and PSA cycles has been
rates conditions. We can conclude that faster feeding rate performed. Three different pressures and gas feeding rates
induces earlier breakthrough, because the faster feeding rate were used in simulation for breakthrough curves, and we can
means more adsorbates in a certain time, which leads the know lower pressure and faster feeding flow rate cause earlier
adsorbent reaching equilibration. breakthrough points. Parametric studies on PSA cycles show
Fig. 7(b) shows the breakthrough curves for H2, CO2 and that, within a certain range, higher adsorption pressure,
CH4 under three different adsorption pressure (1bar, 2bar, shorter feeding time and lower feeding rate lead to higher
3bar) to analyze its influence of performance of hydrogen hydrogen purity, but lower recovery and productivity.
purification. The different symbols stand for different simu- Furthermore, a multi-objective algorithm is needed to opti-
lation results of adsorption pressure. It's clear that higher mize the PSA process.
adsorption pressure induces later breakthrough due to higher
adsorption capability.

Acknowledgments
Parametric study on PSA cycles

We wish to thank the financial supports from the National

Effects of adsorption pressure, feeding flow rate and feeding
Natural Science Foundation of China (Project No. 51476120),
time on the performance of PSA cycles are studied in this
the 111 Project of China (No. B17034) and the Natural Sciences
session. Table 6 shows the results of the parametric studies.
and Engineering Research Council of Canada.
To study the influence of adsorption pressure in PSA pro-
cess, different pressures (2 bar, 3.5 bar, 5 bar, 7 bar) were set
respectively. We take ternary component mixtures (H2/CO2/ references
CH4) to analyze the performance of PSA. From Fig. 8(a), we can
conclude that higher adsorption pressures induce higher pu-
rity, and inversely lower recovery and productivity, because [1] Navarro RM, Guil R, Fierro JLG. Introduction to hydrogen
higher pressures mean more adsorption amount of H2, and production. Compend Hydrogen Energy 2015:21e62.
cause less CO2 and CH4 in the mixture gases in the outlet, [2] Schell J, Casas N, Pini R, Mazzotti M. Pure and binary
which can lead to the higher purity and lower recovery and adsorption of CO2, H2, and N2 on activated carbon. Adsorpt
Journal Int Adsorpt Soc 2012;18(1):49e65.
productivity.
[3] Casas N, Schell J, Pini R, Mazzotti M. Fixed bed adsorption of
To study the effect of feeding flow rate in PSA process, CO2/H2 mixtures on activated carbon: experiments and
different feeding flow rates were set. We take ternary modeling. Adsorpt J Int Adsorpt Soc 2012;18(2):143e61.
component mixtures (H2/CO2/CH4) to analyze the perfor- [4] Joss L, Mazzotti M. Modeling the extra-column volume in a
mance of PSA. From Fig. 8(b), we can conclude that faster small column setup for bulk gas adsorption. Adsorpt J Int
feeding flow rates induce higher recovery and productivity, Adsorpt Soc 2012;18(5e6):381e93.
13974 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 3 ( 2 0 1 8 ) 1 3 9 6 2 e1 3 9 7 4

[5] Casas N, Schell J, Blom R, Mazzotti M. MOF and UiO-67/MCM- [19] And JY, Lee CH, Chang JW. Separation of hydrogen mixtures
41 adsorbents for pre-combustion CO2, capture by PSA: by a two-bed pressure swing adsorption process using
breakthrough experiments and process design. Separ Purif zeolite 5A. Ind Eng Chem Res 1997;36(7):2789e98.
Technol 2013;112(112):34e48. [20] Cruz P, Santos JC, Magalha ~ es FD, Mendes A. Cyclic
[6] Lopes FVS, Grande CA, Rodrigues AE. Activated carbon for adsorption separation processes: analysis strategy and
hydrogen purification by pressure swing adsorption: optimization procedure. Chem Eng Sci 2003;58(14):3143e58.
multicomponent breakthrough curves and PSA performance. [21] Yavary M, Ale-Ebrahim H, Falamaki C. The effect of reliable
Chem Eng Sci 2011;66(3):303e17. prediction of final pressure during pressure equalization
[7] Gomez LF, Zacharia R, Be
nard P, Chahine R. Simulation of steps on the performance of PSA cycles. Chem Eng Sci
binary CO2/CH4 mixture breakthrough profiles in MIL-53(Al). J 2011;66(12):2587e95.
Nanomater 2015. https://doi.org/10.1155/2015/439382. [22] Rahimpour MR, Ghaemi M, Jokar SM, Dehghani O, Jafari M,
[8] Yang J, Han S, Cho C, Lee CH, Lee H. Bulk separation of Amiri S, et al. The enhancement of hydrogen recovery in PSA
hydrogen mixtures by a one-column PSA process. Sep unit of domestic petrochemical plant. Chem Eng J
Technol 1995;5(4):239e49. 2013;226(12):444e59.
[9] Lee JJ, Kim MK, Lee DG, Ahn H, Kim MJ, Lee CH. Heat- [23] Riboldi L, Bolland O. Pressure swing adsorption for
exchange pressure swing adsorption process for hydrogen coproduction of power and ultrapure H2 in an IGCC plant
separation. AIChE J 2008;54(8):2054e64. with CO2 capture. Int J Hydrogen Energy
[10] Jee JG, Kim MB, Lee CH. Adsorption Characteristics of 2016;41(25):10646e60.
hydrogen mixtures in a layered bed: binary, ternary, and five- [24] Chowdhury P, Mekala S, Dreisbach F, Gumma S. Adsorption
component mixtures. Ind Eng Chem Res 2001;40(3):809e15. of CO, CO2 and CH4 on Cu-BTC and MIL-101 metal organic
[11] Ahn S, You YW, Lee DG, Kim KH, Min O, Lee CH. Layered two- frameworks: effect of open metal sites and adsorbate
and four-bed PSA processes for H2 recovery from coal gas. polarity. Microporous Mesoporous Mater 2012;152(4):246e52.
Chem Eng Sci 2012;68(1):413e23. [25] Silva B, Solomon I, Ribeiro AM, Lee U, Hwang YK. H2
[12] Ahn H, Chun C, Park M, Ahn IS, Lee CH. Thermal effects on purification by pressure swing adsorption using Cu-BTC.
the breakthrough curve of a hydrogen ternary system at a Separ Purif Technol 2013;118(6):744e56.
fixed bed. Separ Sci Technol 2001;36(10):2121e45. [26] Agueda VI, Delgado JA, Uguina MA, Brea P, Spjelkavik AI,
[13] Park JH, Kim JN, Cho SH, Kim JD, Yang RT. Adsorber Blom R, et al. Adsorption and diffusion of H2, N2, CO, CH4,
dynamics and optimal design of layered beds for and CO2, in UTSA-16 metal-organic framework extrudates.
multicomponent gas adsorption. Chem Eng Sci Chem Eng Sci 2015;124:159e69.
1998;53(23):3951e63.
nard P, Chahine R. Heat and mass transfer
[27] Xiao JS, Li RP, Be
[14] Schell J, Casas N, Marx D, Mazzotti M. Precombustion CO2 model of multicomponent adsorption system for hydrogen
capture by pressure swing adsorption (PSA): comparison of purification. Int J Hydrogen Energy 2015;40(14):4794e803.
laboratory PSA experiments and simulations. Ind Eng Chem [28] Xiao JS, Peng YZ, Be
nard P, Chahine R. Thermal effects on
Res 2013;52(24):8311e22. breakthrough curves of pressure swing adsorption for
[15] Casas N, Schell J, Joss L, Mazzotti M. A parametric study of a hydrogen purification. Int J Hydrogen Energy
PSA process for pre-combustion CO2 capture. Separ Purif 2016;41:8236e45.
Technol 2013;104:183e92. [29] Peng YZ, Ming Y, Xiao JS, Be
nard P, Chahine R. Thermal
[16] You YW, Lee DG, Yoon KY, Moon DK, Kim SM, Lee C-H. H2 effects on hydrogen purification breakthrough curves of
PSA purifier for CO removal from hydrogen mixtures. Int J activated carbon/zeolite layered bed. J Eng Thermophys
Hydrogen Energy 2012;37(23):18175e86. 2016;37(7):1511e8 [in Chinese].
[17] Lopes FVS, Grande CA, Rodrigues AE. Fast-cycling VPSA for [30] Sereno C, Rodrigues A. Can steady-state momentum
hydrogen purification. Fuel 2012;93(1):510e23. equations be used in modelling pressurization of adsorption
[18] Lopes FVS, Grande CA, Rodrigues AE. A parametric study of beds. Gas Separ Purif 1993;7(3):167e74.
layered bed PSA for hydrogen purification. Chem Eng Sci
2008;63(21):5258e73.