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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9

Available online at www.sciencedirect.com

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

CFD model for charge and discharge cycle of adsorptive

hydrogen storage on activated carbon

Jinsheng Xiao a,b,*, Rong Peng c, Daniel Cossement b, Pierre Bénard b, Richard Chahine b
a
Hubei Key Laboratory of Advanced Technology for Automotive Components, School of Automotive Engineering, Wuhan University of
Technology, Hubei 430070, China
b
Hydrogen Research Institute, Université du Québec à Trois-Rivières, QC G9A 5H7, Canada
c
School of Materials Science and Engineering, Wuhan University of Technology, Hubei 430070, China

article info abstract

Article history: High surface area activated carbons and other microporous adsorbents have generated
Received 20 March 2012 a significant amount of interest over the past decade as storage media for hydrogen and
Received in revised form natural gas, due to their high storage capacity at low temperatures and their use in gas
13 October 2012 purification processes. This paper uses computational fluid dynamics (CFD) to simulate the
Accepted 31 October 2012 charging and discharging of a sorption-based hydrogen storage system. The CFD model is
Available online 5 December 2012 based on the mass, momentum and energy conservation equations of a system formed of
gaseous and adsorbed hydrogen, an activated carbon bed and steel tank walls. The
Keywords: adsorption process is modeled using the DubinineAstakov adsorption isotherms extended
Hydrogen storage to the supercritical regime. The model is implemented using Fluent. In our study, we can
Activated carbon obtain accuracy peak temperature of simulation due to a non-constant isosteric heat of
Adsorption adsorption is used, derived from the model isotherms. We adopt piecewise heat capacity to
Charge consider the heat capacity of the adsorbed phase of hydrogen. We can make a conclusion
Discharge that the simulated temperatures without consideration of heat capacity for hydrogen in
Computational fluid dynamics adsorbed phase (cpa), rise faster and reach higher peaks than the simulated temperatures
with consideration of cpa, and diverge more from experimental results. Also, we study the
changes of temperature, pressure and adsorption during the charging and discharging
processes as well as when the system is idle (which we define as dormancy) in the case of
room temperature water cooling. The results are compared with experimental data from
a storage unit cooled with room temperature water. The simulated pressure is in a good
agreement with the experimental values. The simulated temperature profiles are also
generally in good agreement with the experimental values, except close to the inlet and the
wall. In addition, we have studied the effect of quality of the mesh on the accuracy and
stability of the numerical computation and the influence of the mass flow rates on
temperature and adsorption capacity.
Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

* Corresponding author. Hydrogen Research Institute, Université du Québec à Trois-Rivières, QC G9A 5H7, Canada. Tel.: þ1 819 376
5011x4478; fax: þ1 819 376 5164.
E-mail addresses: Jinsheng.Xiao@uqtr.ca (J. Xiao), Pierre.Benard@uqtr.ca (P. Bénard), Richard.Chahine@uqtr.ca (R. Chahine).
0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2012.10.119
 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9 1451

Nomenclature p0 saturation pressure of the vapor at temperature T

or limit pressure, Pa
cpa heat capacity of hydrogen in adsorbed phase, J/
R universal gas constant, 8.314 J/mol K
kg K
Q energy source term, W/m3
cpg heat capacity of hydrogen in gas phase, J/kg K !
F momentum source term, N/m3
C2 inertial resistance coefficient, m1
Fi component of momentum source term in i
Dp average diameter of particles, m
direction, N/m3
Eg energy of hydrogen gas, J/kg
Sm mass source term, kg/m3 s
Es energy of solid, J/kg !u physical velocity, m/s
DH isosteric adsorption heat, J/mol !v superficial velocity vector, m/s
hi species enthalpy along i direction, J/kg
! j!
vj magnitude of velocity vector, m/s
Ji diffusion flux of species i, kg/m2 s
vi component of velocity in i direction, m/s
keff effective thermal conductivity of the medium, W/
mK Greek symbols
kg hydrogen gas thermal conductivity, W/m K 1/k viscous resistance factor, m2
ks solid medium conductivity, W/m K a enthalpic factor, J/mol
m distribution parameter of isotherms b entropic factor, J/mol K
MH2 molecular mass of hydrogen gas, kg/mol εb bed porosity
na absolute adsorption amount per mass of m dynamic viscosity of hydrogen gas, Pa s
adsorbent, mol/kg rg hydrogen gas density, kg/m3
n0 limit adsorption, mol/kg rp particle density of activated carbon, kg/m3
p static pressure or equilibrium pressure of gas, Pa rs solid density, kg/m3
s stress tensor, Pa

1. Introduction a numerical model based on the solution of the 2D transport

equations for mass, momentum and energy in porous media
Hydrogen is a renewable and potentially low emission energy to investigate hydrogen charging in packed bed storage tank.
vector currently being considered as a replacement fuel for Results show that the capacity limitation due to thermal
fossil sources. Its efficient and safe storage is a challenge that is heating can be avoided by using high thermal conductivity
being addressed through four approaches: liquefaction, packed bed. Xiao et al. [9] used a finite element model to
compression, absorption in metal hydrides and adsorption in simulate the heat and mass transfer in activated carbon
high surface area nanoporous materials [1]. Each method has hydrogen storage tank. The heat capacity of adsorbed phase,
its own advantages and limitations, and probably their own the contact thermal resistance between the AC bed and the
niches. For example, liquid hydrogen has the advantages of steel wall, and the inertial resistance of the high speed
light weight, but it consumes a lot of energy during liquefac- charging hydrogen gas are considered in the model.
tion and lost some hydrogen mass during dormancy. The computational fluid dynamics (CFD) is performed
Hydrogen physisorption has been considered as one of the based on the finite control volume (FCV) method. Recently, the
most promising storage methods for meeting the goals of the CFD was widely used to simulate the hydrogen storage system
US Department of Energy (DOE) hydrogen plan for fuel cell based on activated carbon [1,6,10e12]. Among these works,
powered vehicles [2,3]. Hirscher and Panella [4] reported that Herosilla-Lara et al. [1] studied the hydrogen storage in tank
the maximum storage capacity for an activated carbon with packed with activated carbon at room temperature by
a BET (BrunauereEmmetteTeller) surface area of 2560 m2/g at carrying out experiments and simulations. They built an
77.4 K was 4.5 wt%. Kojima et al. [5] studied hydrogen axisymmetrical geometry used for simulations of the charging
adsorption and desorption in different carbon materials at 296 of the tank. However they only considered the charging
and 77 K. They claimed that superactivated carbon having process. Xiao et al. [13] studied the heat and mass transfer in
specific surface area (SSA) of 3220 m2/g, adsorbed and desorbed a hydrogen storage system packed with activated carbon at
1.3wt% of hydrogen at 296 K and 5 wt% of hydrogen at 77 K. room temperature (295 K) and medium storage pressure
The porous high surface area activated carbon (AC) has (10 MPa) by using the Fluent, overall good agreement is found
been proposed as storage media for hydrogen as it is an between computation and experimental data published by
effective and low-cost adsorption material [6]. A number of Hermosilla-Lara et al [1]. However，they only considered the
research studies about adsorptive hydrogen storage on acti- charging process and the thermal capacity of gas phase
vated carbon have been carried out in recent years. Paggiaro hydrogen, and used the constant isosteric heat of adsorption.
et al. [7] formulated a two-dimensional model to study the Xiao et al. [14] developed a finite element model for simulating
thermal effects during high pressure filling of a cryo- the whole cycle of charge and discharge processes in an
adsorptive hydrogen storage tank. The adsorbent is activated carbon hydrogen storage tank by using the Comsol.
powdered activated carbon. They obtained a good agreement In their work, they only considered the thermal capacity of gas
between experiments and simulations. Momen et al. [8] used phase hydrogen.
1452 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9

In this work, the computational fluid dynamics (CFD) 2.2. Momentum conservation equation
method is used to simulate charge and discharge cycle of
adsorptive hydrogen storage on activated carbon with room The momentum equation can be expressed as [15]:
temperature water cooling. In our study, non-constant isos-
v
!
 !
teric heat of adsorption is used, derived from the model rg v þ V$ rg !
v!v ¼ Vp þ V$s þrg !
g þ F (4)
vt
isotherms. The range of isothermal heat of adsorption is huge
due to the great variation of hydrogen adsorbed by the acti- where s is the stress tensor and the momentum source term
!
vated carbon, during the charge and discharge cycle of F (N/m3) is based on the Ergun equation and includes viscous
adsorptive hydrogen storage on activated carbon. So that and inertial losses. Its component in i direction, “Fi”, for simple
a non-constant isosteric heat of adsorption is used. When the homogeneous porous media, can be written as [15]:
charging mass flow rate at entrance is large, we have to
m 1
consider the heat capacity of the hydrogen adsorbed by the Fi ¼  vi  C2 rg j!
v jvi (5)
k 2
activated carbon because the great variation of the adsorbed
where m (Pa s) is the dynamic viscosity of hydrogen gas, 1/k (1/
phase of hydrogen in a short time. But we can not use the user
m2) is the viscous resistance coefficient, C2 (1/m) is the inertial
defined function (UDF) in Fluent to define the heat capacity. In
resistance coefficient, j!v j is the magnitude of the velocity
order to solve this problem, we adopt piecewise heat capacity
vector and vi is the velocity component along the i direction.
to consider the heat capacity for hydrogen in adsorbed phase
(cpa). The quality of the mesh plays a significant role in the
accuracy and stability of the numerical computation. We 2.3. Energy conservation equation
attempt to search for the optimum cell number to obtain the
accurate and stable results of numerical computation. The energy equation expresses the balance between the
On that basis, we study the changes of temperature, energy amount accumulated in the tank and the energy
pressure and adsorption during the charging and discharging changes due to convective flow, pressure work, conductive
processes as well as when the system is idle (which we define and thermal dispersion fluxes and heat release due to
as dormancy) in the case of room temperature water cooling. adsorption. It can be expressed as [15]:
Also, the effects of the isosteric heat of adsorption, the heat
vh i h
i
capacity and the mass flow rates on the hydrogen storage εb rg Eg þ ð1  εb Þrp Es þ V$ !
v rg Eg þ p
properties (pressure, temperature and adsorbed density vt
 

distributions) are studied during the charging, dormancy and ¼ V$ keff VT þ V$ s $! v þQ (6)
discharging processes.
where E is the sum of internal energy and kinetic energy and
expressed by enthalpy h as:
2. CFD model for hydrogen storage system
p v2
E¼h þ (7)
2.1. Mass conservation equation r 2

So the pressure work is considered in the terms of left hand

In porous media, the mass conservation equation can be side of the energy conservation equation. The potential
written as [15]: energy by gravity may be neglected here. It is obvious that the

 viscous dissipation F is included in the term V$ðs $!
v Þfrom the
v εb rg

þ V$ rg !
following relations:
v ¼ Sm (1)
vt



where Sm (kg/m3 s) is the mass source term. The source term v ¼!
V$ s $! v $ V$ s þ s : V!
v ¼!
v $ V$s þ F (8)
models the mass transfer between the gas phase and the
absorbed phase per unit volume per second. It can be written The energy source term Q in the energy conservation
as: equation can be expressed as [13]:

v 
Sm ¼  rb qa (2) Q ¼ DHSm =MH2 (9)
vt
Note that the bed density can be expressed with bed where Sm (kg/m3 s) is the mass source term, DH (J/mol) is the
porosity and particle density: rb ¼ ð1  εb Þrp ðkg AC=m3 Þ, and isosteric heat of adsorption which relate to the absolute
the mass ratio of adsorbate to adsorbent can be expressed as: adsorption. Q (W/m3) is the energy source term, which models
qa ¼ MH2 na ðkg H2 =kg ACÞ, so that the mass source term can the generated heat per unit volume.
also be written as [13]:

vna vna 2.4. Adsorption model

Sm ¼ rb MH2 ¼ ð1  εb Þrp MH2 (3)
vt vt
where εb is the bed porosity, rb (kg/m3) is the bed density and The adsorption sub-models used in this work include models
rp (kg/m3) is the particle density of activated carbon, na (mol/ for the adsorption isotherms and the isosteric heat of
kg) is the absolute adsorbed amount per unit adsorbent and adsorption. A modified DubinineAstakhov (DeA) adsorption
MH2 (kg/mol) is the molecular mass of hydrogen. model is used to describe the adsorption isotherms in
 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9 1453

Fig. 1 e Geometric model (a) and grid (b) of hydrogen storage tank.

activated carbons. The absolute adsorption isotherm is of Maxsorb and submerged in a Dewar flask. The tempera-
modeled using [16]: ture evolution inside the tank is measured using eight ther-
  m   mocouples distributed along the vertical axis and the radius.
RT m p0 The hydrogen inflow and outflow rates are quantified with
na ¼ n0 exp  ln (10)
a þ bT p
flow meters. Before starting the tests, the Dewar flask
where n0 (mol/kg) is the limiting adsorption, the values of n0 is enclosing the storage tank was filled with room temperature
71.6 mol/kg, m is a number generally close to a small integer water. Then, the tank was filled with hydrogen at the specified
and equal to 2 for most activated carbons. R is the universal flow rate. The dynamic pressures, temperatures, and flow-
gas constant, p is the equilibrium pressure, p0 is the saturation rates of the system were monitored during the charging and
pressure. The value of p0 is 1.47e9 Pa. The parameters a and b discharging processes as well as when the system is idle. The
are the enthalpic and the entropic factors, equal to 3080 J/mol detail description and some illustrations of the experimental
and 18.9 J/(mol K), respectively. system were shown in [17].
The adsorption amount has significant effect on isosteric
heat of adsorption, which decreases during adsorption while 3.2. Geometry model
increasing during desorption. Based on the model isotherms
of Equation (10), the heat of adsorption can be obtained using The Commercial CFD software Fluent is used to simulate the
the following equation [14]: hydrogen adsorption process during charging, dormancy and
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi discharging. Fig. 1(a) shows the geometric model of the
DH ¼ a lnðn0 =na Þ (11) hydrogen storage tank. The gray area (except the pipe) corre-
sponds to the tank packed with activated carbon. The
Note that the low coverage limit exhibits a weak singularity
temperature monitor points are shown as red dots, distributed
associated with the absence of a proper Henry’s law limit for
along the axial and radial directions. Fig. 1(b) shows the grid of
the Dubinin adsorption isotherms. This weak singularity has,
the hydrogen storage tank, which has 5377 nodes. The
however, little effect in the high pressure behavior.
dimensions of the geometry model and the location of the
monitoring points are given in Table 1.

3. Model parameters of hydrogen storage 3.3. Material properties

system
The material properties of hydrogen storage tank are given in
3.1. Experimental system for model validation
Table 2. The effective thermal conductivity is computed as the

The experiments were performed at the Hydrogen Research

Institute of the University of Quebec at Trois-Rivieres [17]. The
Table 2 e Material properties.
experiments were performed with a 2.5 L tank filled with 671 g
Activated carbon Density (kg/m3) 517.6
Specific heat capacity (J/kg K) 825
Thermal conductivity (W/m K) 0.646
Table 1 e Geometric parameters and coordinates. Porosity 0.49
Hydrogen Density Ideal gas
Parameter Value (mm) Parameter Value (mm)
Specific heat capacity (J/kg K) 14,700
Inner radius of tank 46.9 Point C2 (143.75, 0) Thermal conductivity (W/m K) 0.206
Thickness of tank wall 3.9 Point C3 (188.75, 0) Viscosity (Pa s) 8.411e-6
Length of tank 400 Point C4 (233.75, 0) Viscous resistance coefficient (m2) 8.29055e7
Inner radius of pipe 4 Point C5 (278.75, 0) Inertial resistance coefficient (m1) 7586
Length of pipe 53.75 Point C6 (323.75, 0) Steel wall Density (kg/m3) 7830
Thickness of pipe wall 1 Point Cr (248.75, 23.5) Specific heat capacity (J/kg K) 468
Point C1 (98.75, 0) Point Cw (233.75, 46.9) Thermal conductivity (W/m K) 13
1454 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9

Table 3 e Values of qaand c0ps .

Time (s) 0e953 953e3822 3822e4694 4694e6000 Table 5 e Initial and environmental conditions.
qa 0.0168027 0.0168027 0.0090476 0.0012925 Test No. pi (MPa) Ti (K) Tf (K) h (W/(m2 K)
c0ps (J/kg K) 1072 1072 958 844
18 0.032 301.5 301.7 36
19 0.049 302 302.5 36
20 0.032 302.4 302.5 36
volume average of the fluid (hydrogen) conductivity and the
solid (activated carbon) conductivity, and it may be expressed
as [15]:

keff ¼ εb kg þ ð1  εb Þks (12) 4. Results and discussions

where kg and ks (W/m K) are, respectively, the thermal The computational fluid dynamics (CFD) method is used to
conductivities of hydrogen and activated carbon. simulate the charging and discharging of a sorption-based
The CFD software Fluent normally considers only two hydrogen storage system. In our study, we adopt piecewise
components for the mixture properties of porous media. The heat capacity to consider the heat capacity of the adsorbed
specific heat is written as a volumetric average of the gas phase of hydrogen and non-constant isosteric heat of
phase and the solid phase: adsorption. On that basis, we study the changes of tempera-
  ture, pressure and adsorption during the charging, dormancy
rcp m ¼ εb rg cpg þ ð1  εb Þrp c0ps (13)
and discharging processes. Also, the effects of the isosteric
heat of adsorption, the heat capacity and the mass flow rate
where the modified specific heat for the solid phase includes on the hydrogen storage properties (pressure, temperature
the contribution of the adsorbed phase of hydrogen, and is and adsorbed density distributions) are studied during the
given by: charging, dormancy and discharging processes.
c0ps ¼ cps þ qa cpa zcps þ qa cpg (14)
4.1. Mesh sensitive analysis
where qa ¼ na MH2 , we set c0ps as the specific heat for the solid
phase (adsorbent and adsorbate) in the Fluent. We calculate The quality of the mesh plays a significant role in the accuracy
the average values of na at the end of the charging and dis- and stability of the numerical computation. The cell shape
charging according to the values of experimental temperature and cell size have an important impact on the accuracy and
and pressure. Then, we calculate the different values of c0ps stability of the numerical solution. The finer meshes can
during the processes of charging, discharging and dormancy. result higher accuracy of solution but will cost more computer
The values of c0ps at the end of the charging and discharging are memory and more CPU (Central Processing Unit) time. We
1072 J/(kg K) and 844 J/(kg K). The different values of c0ps during attempt to find the optimum mesh size by which we can get
the charging, discharging and dormancy processes are given more accurate numerical solution in shorter CPU time. In
in Table 3. order to study the effect of mesh quality on solution, we
consider three cases: (1) finer mesh: 19,671 cells and 20,654
3.4. Boundary and initial conditions nodes, (2) intermediate mesh: 4885 cells and 5377 nodes, (3)
coarser mesh: 3126 cells and 3520 nodes. We exam the effects
Boundary and initial conditions of the three tests are given in of the cell size on the accuracy of simulation pressure and
Table 4 and Table 5. The boundary and initial conditions used simulation temperature at C4 respectively. When the finer
in this section are based on Test No. 20, which was performed mesh is adopted, the accuracy almost has no more increase
at the Hydrogen Research Institute of the University of Quebec but it doubles the run time. If we use coarser mesh, the
at Trois-Rivieres. The pressure is easier to measure accurately stability of the numerical computation will decrease. So that
than the mass flow rate. Therefore, we correct the intake and the intermediate mesh is adopted in our study.
output hydrogen flow rates based on the measured pressure
profile.
4.2. Mass balance analysis

Hydrogen exists in the storage tank in both adsorbed and

gaseous phases. Fig. 2 shows the mass balance of hydrogen in
the system. The ma represents the mass of hydrogen adsor-
Table 4 e Inlet boundary conditions.
bed by the activated carbon. The amount of hydrogen is
Test No. Charge period Discharge period calculated by use of the modified DubinineAstakhov (DeA)
_ (kg/s) Tin (K) Time (s) m
Time (s) m _ (kg/s) Tin (K) adsorption model. The mg represents the mass in the gas
phase. The mt is the total mass of hydrogen present in tank.
18 0e196 9.556e-5 294.9 2828e3616 2.171e-5 297.7
Fig. 2 shows the total amount of hydrogen stored equals the
19 0e442 4.355e-5 296.1 3046e3907 2.186e-5 298.0
20 0e953 2.048e-5 297.6 3822e4694 2.186e-5 297.7
sum of hydrogen in adsorbed phase and gaseous phase in the
adsorbent bed.
 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9 1455

Fig. 2 e Mass balance analysis of gaseous (mg), adsorbed (ma), and total (mt) hydrogen masses.

4.3. Comparison of pressures is higher in the center of the tank and lower close to the
entrance and the walls. At 4694 s, the situation is reversed: the
Fig. 3 shows a comparison of the pressure obtained from the temperature is lower in the center of the tank and higher close
simulations with the experimental data obtained for the to the system boundaries.
system modeled. In general Fig. 3 shows good agreement with The absolute adsorption density at different locations
the experimental results, although the equilibrium pressures remains almost uniform during charging and discharging.
after charge are slightly higher than the experimental result. Fig. 4 (c) and (d) show contours of absolute adsorption in the
tank at 953 s and 4694 s respectively. At 953 s, the largest
absolute adsorption is observed close to the wall of the tank,
4.4. Distributions of temperature and adsorption whereas the smallest absolute adsorption is located at the
density center of the tank. At 4694 s, the situation is reversed and the
largest absolute adsorption is observed near the center of the
Fig. 4 (a) and (b) show the temperature distributions in the tank and the smallest close to the wall of the tank. This is
tank at 953 s and 4694 s respectively. At 953 s, the temperature because the absolute adsorption depends on the local pres-
sure and temperature. As the pressure distribution is uniform
throughout the tank while the temperature distribution var-
ies at different places, the absolute adsorption amount is
largest where the temperature is lowest and the absolute
adsorption amount is lowest where the temperature is
largest.

4.5. Comparison of temperatures

Fig. 5 compares the temperature obtained numerically with

the experimental data. In general, all the simulated temper-
ature curves agree well with the experimental curves.
However, the temperatures at C1 and Cw are affected by the
inlet boundary condition and the thermal exchanges with the
wall, respectively. We used a more realistic geometric model
(Fig. 1) to replace the simplified wide-open geometric model
for obtaining good agreement temperature at point C1 with
Fig. 3 e Pressure comparison in hydrogen storage tank. the experimental.
1456 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9

Fig. 4 e Contours of temperature (K) at (a) 953 s and (b) 4694 s, and absolute adsorption (mol/kg) at (c) 953 s and (d) 4694 s.

4.6. Velocity field analysis hydrogen in adsorbed phase. Three cases related to Test No. 18,
19 and 20 which have nominal charging flow rates of 60, 30 and
Fig. 6 shows the velocity magnitude distributions along axis at 15 SLPM (standard liters per minute) are adopted for the study.
different time during charge and discharge respectively. Fig. 7 Fig. 8 shows the effect of the effective heat capacity on
shows the velocity distributions at 50 s and 4644 s respec- temperature at C4 when the charging mass flow rate at entrance
tively. Generally speaking, the maximum magnitude of the is (a) 9.556  105 kg/s (Test No. 18), (b) 4.355  105 kg/s (Test
velocity is observed near the entrance. During charge, the No. 19) and (c) 2.048  105 kg/s (Test No. 20), respectively Fig. 9.
maximum magnitude of the velocity at C1, C2 and C3 are The simulated temperatures without consideration of heat
0.09 m/s, 0.056 m/s and 0.047 m/s respectively. During capacity for hydrogen in adsorbed phase (cpa), rise faster and
discharge, the maximum magnitude of the velocity at C1, C2 reach higher peaks than the simulated temperatures with
and C3 are 0.089 m/s, 0.055 m/s and 0.046 m/s respectively. consideration of cpa, and diverge more from experimental
The maximum magnitude of the velocity at C1, C2 and C3 results. These differences are more significant for higher
during charge and discharge are almost the same, but in charging mass flow rates, for example, 4.355  105 kg/s or
opposite directions. 9.556  105 kg/s. In general, the inclusion of the heat capacity
of hydrogen in adsorbed phase leads to better experimental
agreement as it pertains to temperature.
5. Parametric study
5.2. Isosteric heat of adsorption on temperature
The parameters used for this study are based on the operating
conditions of the experiments (namely Test No. 18, 19 and 20), The value of the isosteric heat of adsorption is determined by
which were performed at the Hydrogen Research Institute of a large number of experiments. The value of the isosteric heat
the University of Quebec at Trois-Rivieres. The boundary and of adsorption is different in different literature due to different
initial conditions used in previous section are based on Test experimental conditions and structure of activated carbon.
No. 20. The pressure is easier to measure accurately than the The range of isosteric adsorption heat is from 3000 J/mol to
mass flow rate. Therefore, we correct the intake and output 10,000 J/mol in different literature. The value of the isosteric
hydrogen flow rates based on the measured pressure profile. heat of adsorption is 3300 J/mol in [10], 10,500 J/mol in [11].
In our study, we have considered the dependence of the
5.1. Effective heat capacity on temperature isosteric heat of adsorption on the adsorbed density. In order
to validate this approach we compare four cases: (1) the
In order to study the effect of the effective heat capacity of isosteric heat of adsorption is set to a constant value of 3185 J/
adsorptive system on temperature variations, we simulated the mol; (2) the isosteric heat of adsorption is set to a constant value
system without and with consideration of the heat capacity of of 7500 J/mol; (3) a constant value of 5250 J/mol is used, obtained
 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9 1457

Fig. 5 e Temperature comparison at different monitoring

points along (a) axial and (b) radial positions. Fig. 6 e Velocity magnitude distributions along axis at
different time during (a) charge and (b) discharge.

by fitting the simulation data to the experimental results; and (4)

the isosteric heat of adsorption is calculated using Equation (9).
adsorbed by the activated carbon in three cases (Test No. 18,
The isosteric heat of adsorption is set to a constant value of
Test No. 19 and Test No. 20). The charging mass flow rates at
3185 J/mol in [13], the simulated temperature curves agree with
entrance in three cases are 9.556  105 kg/s, 4.355  105 kg/s
the experimental curves during charge. When the constant
and (c) 2.048  105 kg/s. Pressure up to the same value (about
value of 3185 J/mol is used in this paper, the simulated
9 MPa) at the end of charging in three cases.
temperature curves agree with the experimental curves during
Fig. 10(a) shows the effect of mass flow rates on the mass of
charge. However, the peak temperature of simulation is higher
hydrogen adsorbed by the activated carbon. The mass of
than the experimental results during discharge. When the
hydrogen adsorbed by the activated carbon in the Test No. 18
constant value of 5250 J/mol is used, obtained by fitting the
is lower than in the Test No. 20. This is because the absolute
simulation data to the experimental results, we can obtain
adsorption depends on the local pressure and temperature.
accurate peak temperature of simulation. However, we can
The temperature in the Test No. 18 is extremely higher than
obtain more accurately peak temperature of simulation, when
the Test No. 20.
non-constant isosteric heat of adsorption is used. When the
Fig. 10(b) shows the effect of mass flow rates on the
variational isosteric heat of adsorption which relate to the
temperature at C4. For higher charging mass flow rates, the
absolute adsorption is used, the value of isosteric heat of
temperature will reach the peak temperature in a short time.
adsorption varies from 9500 J/mol to 4500 J/mol during charge,
When the mass flow rates are 9.556  105 kg/s,
and from 4500 J/mol to 7500 J/mol during discharge.
4.355  105 kg/s and 2.048  105 kg/s, respectively, the
temperature will increase 48 K, 38 K and 15 K at the end of the
5.3. Effect of mass flow rates on adsorption filling process. The peak temperature in the Test No. 18 is
extremely higher than in the Test No. 20. When the mass flow
In order to study the effect of mass flow rates on adsorption, rates are large, the heat generated due to the adsorption and
we compare the temperature at C4 and the mass of hydrogen compression is not able to dissipate in time.
1458 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9

Fig. 9 e Effect of isosteric heat of adsorption on

temperature at C4 based on Test No. 20.

Fig. 7 e Velocity magnitude distributions along axial

positions with different radii at (a) 50 s and (b) 4644 s.

Fig. 8 e Effect of heat capacity of adsorbed hydrogen (Cpa) Fig. 10 e Effect of mass flow rate on (a) the absolute
on temperature at C4 under different charging flow rates. adsorption and (b) temperature at C4.
 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 3 8 ( 2 0 1 3 ) 1 4 5 0 e1 4 5 9 1459

6. Conclusions Acknowledgments

A CFD model was developed to simulate whole char- We wish to acknowledge the supports from the NSERC
geedischarge cycle of adsorptive hydrogen storage on acti- Hydrogen Canada (H2CAN) Strategic Research Network and
vated carbon. We used the CFD software Fluent to simulate the Natural Resources Canada (NRCAN).
the hydrogen adsorption process in a hydrogen storage unit
during charging and discharging, and when the system is idle
in the case of room temperature water cooling. In our study,
references
we adopt piecewise heat capacity to consider the heat
capacity of the adsorbed phase of hydrogen and non-constant
[1] Hermosilla-Laraa G, Momena G, Marty PH, Neindrea BL,
isosteric heat of adsorption. We study the changes of
Hassounia K. Hydrogen storage by adsorption on activated
temperature, pressure and adsorption during the charging
carbon: investigation of the thermal effects during the
and discharging processes as well as when the system is idle. charging process. Int J Hydrogen Energ 2007;32:1542e53.
Also, the effects of the isosteric heat of adsorption, the heat [2] Hynek S, Fuller W, Bentley J. Hydrogen storage by carbon
capacity and the mass flow rates on the hydrogen storage sorption. Int J Hydrogen Energ 1997;22:601e10.
properties are studied during the charging, dormancy and [3] Dillion AC, Jones KM, Bekkedahl TA, Kiang CH, Bethune DS,
discharging processes. Heben MJ. Storage of hydrogen in single-walled carbon
nanotubes. Nature 1997;386:377.
The mesh sensitive analysis is made to obtain accuracy
[4] Hirscher M, Panella B. Nanostructures with high surface area
and stable numerical solution in shorter run times. We have for hydrogen storage. J Alloy Compd 2005;404e406:399e401.
tried to search for the optimum cell number, so that we can [5] Kojima Y, Kawai Y, Koiwai A, Suzuki N, Haga T, Hioki T.
get more accurate and stable numerical solution in shorter Hydrogen adsorption and desorption by carbon materials. J
run times. Alloy Compd 2006;421:204e8.
The non-constant heat of adsorption which relate to the [6] Momen G, Hermosilla G, Michau A, Pons M, Firdaous M,
absolute adsorption was adopted to achieve validity for the Marty PH, et al. Experimental and numerical investigation of
the thermal effects during hydrogen charging in packed bed
whole cycle. We replaced the constant isosteric heat of
storage tank. Int J Heat Mass Tran 2009;52:1495e503.
adsorption by the various heat of adsorption. We can obtain [7] Paggiaro P, Michl F, Bénard P, Polifke W. Cryo-adsorptive
the accuracy peak temperature of simulation when the hydrogen storage on activated carbon. II: investigation of the
various heat of adsorption is used. thermal effects during filling at cryogenic temperatures. Int J
The heat capacity for hydrogen in adsorbed phase was Hydrogen Energ 2010;35:648e59.
considered to improve accuracy in the whole cycle. We have [8] Momen G, Jafari R, Hassouni K. On the effect of process
temperature on the performance of activated carbon bed
considered the effect of the heat capacity of the adsorbed
hydrogen storage tank. Int J Therm Sci 2010;49:1468e76.
phase on the results. The simulated temperatures without
[9] Xiao JS, Liu Y, Bénard P, Chahine R. Finite element simulation
consideration of heat capacity for hydrogen in adsorbed phase of heat and mass transfer in activated carbon hydrogen
(cpa), rise faster and reach higher peaks than the simulated storage tank. Int J Heat Mass Tran 2012;55:6864e72.
temperatures with consideration of (cpa), and diverge more [10] Delahaye A, Aoufi A, Gicquel A. Improvement of hydrogen
from experimental results. storage by adsorption using 2-D modeling of heat effects.
The model, implanted in Fluent, is well validated by AIChE J 2002;48:2061e73.
[11] Lamari M, Aoufi A, Malbrunot P. Thermal effects in dynamic
experiments and applied for parametric study. The simulation
storage of hydrogen by adsorption. AIChE J 2000;46:632e46.
results were compared with experimental measurements. [12] Momen G, Hermosilla G, Michau A, Pons M, Firdaouss M,
The simulated pressures showed good agreement with the Hassouni K. Hydrogen storage in an activated carbon bed:
experimental values. Except for the monitoring points at C1 effect of energy release on storage capacity of the tank. Int J
and Cw, the simulated temperatures were also in agreement Hydrogen Energ 2009;34:3799e809.
with the experimental data. [13] Xiao JS, Tong L, Deng CH. Simulation of heat and mass
We have studied mass flow rates on the mass of transfer in activated carbon tank for hydrogen storage. Int J
Hydrogen Energ 2010;35:8106e16.
hydrogen adsorbed by the activated carbon. The mass of
[14] Xiao JS, Wang JJ, Cossement D, Bénard P, Chahine R. Finite
hydrogen adsorbed by the activated carbon in the Test element model for charge and discharge cycle of activated
No. 18 is lower than in the Test No. 20. This is because carbon hydrogen storage. Int J Hydrogen Energ 2012;37:802e10.
the absolute adsorption depends on the local pressure [15] ANSYS, Inc. FLUENT 6.3 user’s guide; 2006.
and temperature. The temperature in the Test No. 18 is [16] Richard MA, Bénard P, Chahine R. Gas adsorption process in
extremely higher than the Test No. 20. When the mass activated carbon over a wide temperature range above the
critical point. Part1: modified Dubinin-Astakhov model.
flow rates are 9.556  105 kg/s, 4.355  105 kg/s and
Adsorption 2009;15:43e51.
2.048  105 kg/s, respectively, the temperature at C4 will
[17] Richard MA, Cossement D, Chandonia PA, Chahine R, Mori D,
increase 48 K, 38 K and 15 K at the end of the filling Hirose K. Preliminary evaluation of the performance of an
process. The larger flow rates result in less amount of adsorption-based hydrogen storage system. AIChE 2009;55:
adsorption due to thermal effects. 2985e96.