Академический Документы
Профессиональный Документы
Культура Документы
The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Available at www.sciencedirect.com
Article history: We have developed 3-D model equations for a cryo-adsorption hydrogen storage tank,
Received 1 December 2009 where the energy balance accommodates the temperature and pressure variation of all the
Received in revised form thermodynamic properties. We then reduce the 3-D model to the 1-D isobaric system and
7 January 2010 study the isobaric refueling period, for simplified geometry and charging conditions. The
Accepted 8 January 2010 hydrogen capacity evolution predicted by the 1-D axial bed model is significantly different
Available online 13 February 2010 than that predicted by the lumped-parameter model because of the presence of sharp
temperature gradients during refueling. The 1-D model predicts a higher hydrogen capacity
Keywords: than the lumped-parameter model. This observation can be rationalized by the fact that
Hydrogen storage modeling a bed with temperature gradients on equilibration should desorb gas, whenever the
Cryogenic adsorption adsorbed phase entropy is lower than the gas phase entropy. The 1-D analysis of the
Metal-organic frameworks isobaric refueling period does not show any significant difference in hydrogen capacity
Refueling evolution among the axial, single and multicartridge annular bed designs. Hence, a multi-
cartridge annular design, though giving a slightly lower pressure drop, does not offer any
heat and mass transfer enhancement over the single cartridge design. And, the single
cartridge annular design appears to be optimal.
ª 2010 Professor T. Nejat Veziroglu. Published by Elsevier Ltd. All rights reserved.
Dormancy & Venting – While the vehicle is parked, due to non-ideal, and its specific heat capacity varies significantly
continuous heat leak into the fuel-tank, its pressure builds with temperature. Hence, in this work we develop a generic
up to vent pressure, say 25 bar, and gas is vented to the fuel form of the 3-D model, which allows variation in all the
cell (or atmosphere). The process of venting is also called thermodynamic properties. We then reduce the 3-D model to
boil-off, in analogy with liquid hydrogen storage systems. the 1-D isobaric system and study the isobaric refueling
The time period till the start of venting is called dormancy. period, for simplified geometry and charging conditions. We
study only the isobaric period of refueling herein, since the
Interestingly these fuel-tank processes occur over different isobaric period is much longer than the initial adsorption
time scales: refueling over a few minutes, discharge over a few period. We have solved these 1-D equations using finite
hours, dormancy over a few days, and venting over a few element method, using the commercial package COMSOLª.
weeks. In [3] we had shown that even the fastest process, i.e.
refueling, is quasi-static i.e. local equilibrium conditions
prevail. Hence, the slower processes are obviously quasi-
static. When the molecular processes are fast, slow processes
are expected to have negligible internal gradients and are
generally amenable for a lumped-parameter analysis. Hence,
we had developed a quasi-static lumped-parameter model for
the cryo-adsorber fuel-tank in [3]. However, one might antic-
ipate that higher dimensional models might be required,
especially for the fastest process i.e. refueling. The refueling
period could be divided into two periods:
2. Packed-bed designs
Exit
A fixed mass of adsorbent can be packed within a pressure Fig. 2 – Longitudinal and cross-sectional views of a single
vessel in various ways. The classical style is the axial bed, annular packed bed.
shown in Fig. 1, where the gas flows axially through a cylin-
drical bed. An alternative design, giving lower pressure drop,
is the annular backed bed, shown in Fig. 2, where the gas is in its pores. The gaseous hydrogen stored in the dead-spaces
ideally expected to flow radially. Hence, an annular bed design around the packed bed but within the pressure vessel (for
is often used in industrial adsorption or catalytic systems; see example, space between the cartridges, header and collector
for example the metal hydride beds [6,7], adsorption refrig- spaces etc.) are not considered.
eration systems [8], catalytic reactors [9], chromatographic
columns [10] etc. However, it is well-known, both through
simulations and experiments that annular packed beds are
3. Pressure drops for three packed-bed
more prone to flow maldistribution than axial packed beds
designs
[11,12]. The presence and effect of flow maldistribution can be
systematically studied only in a two-dimensional model. The
We compute the steady-state pressure drop for a non-
CFD studies [13,14] show that having the inlet and outlet ports
adsorbing hydrogen gas flow of 120 g/s through these three
on the same side, such that the inner and outer flows are anti-
beds at 20 bar and 80 K. To compute the pressure drop across
parallel gives low flow maldistribution. Note that this design is
different from that given in Fig. 2. Such design variations will
be considered in our future two-dimensional studies.
Instead of arranging the whole bed as a single annular bed,
one could arrange the adsorbent bed into multiple cartridges,
as in Fig. 3. In this work, we study the relative advantages of
these three different designs, especially in the context of
isobaric refueling period. In particular, we seek to answer the
following questions: for a fixed mass of adsorbent, is there any
process advantage (in heat and mass transfer sense) in
increasing the number of cartridges or is there an optimum
number of cartridges? These questions cannot be addressed
within the premise of a lumped-parameter model, since all
the designs are equivalent at that level, and we need at least
a one dimensional model.
In our earlier work [3] we had considered a MOF of bulk
density 385.05 kg/m3 and a bed of 90.37 kg mass and 0.2347 m3
volume. We consider packed beds of the same mass and
volume in this study. For all the designs we fix the bed length
at 1 m and back calculate the bed diameter from the bed
volume. For single and 7-cartridge annular beds, assuming the
distribution diameter to be 1 cm, the outer bed diameters are
respectively 0.5467 m and 0.2068 m. While comparing
different bed designs, we consider only the hydrogen capacity Fig. 3 – Cross-sectional view of a 7-cartridge annular
in the bed, both adsorbed hydrogen and the gaseous hydrogen packed bed.
Author's personal copy
a bed for an unidirectional axial flow the widely used form of values for bed density and total porosity, we have the general
Ergun equation is [15, p. 191], form of mass balance as:
DP vq vrg / !
¼ amg U þ brg U2 ; (1) rb þ 3t þ V $ U rg ¼ 0: (9)
L vt vt
where Expanding the gradient term we have
150ð1 3b Þ2 rb
vq
þ 3t
vrg ! / / !
þ U $ V rg þ rg ðV $ U Þ ¼ 0: (10)
a¼ (2) vt vt
33b d2p
Q r ; tÞzq ½Tð!
qð! r ; tÞ; Pð!
r ; tÞ; (14)
Ur ¼ (6)
2prL
and
is the radial superficial velocity. Note that in a radial flow the
superficial velocity varies with radial location. Integrating this vq vq
z : (15)
equation for an annular bed from the inner radius ri to the vt vt
outer radius ro gives The Langmuir adsorption isotherm is
2
Q ro Q 1 1 q bP
Po ¼ Pi amg ln brg : (7) ¼ : (16)
2pL ri 2pL ri ro qm 1 þ bP
Using this equation, the pressure drop across a single and Temperature dependence of the two Langmuir parameters are
seven cartridge annular beds are 337.72 Pa and 12.05 Pa, represented as in [3] as
respectively. These calculations show that there is significant
B
reduction in pressure drop as we move from axial bed design b ¼ b0 exp (17)
T
to single cartridge annular bed design to seven cartridge
annular bed design. and
qm0
qm ¼ ; (18)
f ðTÞ
4. Development of 3-D model equations with
Then, we have
The basic form of the transient mass balance (derivation
details are in the supporting material) is vq vq vT vq vP
¼ þ
vt vT P vt vP T vt
v / !
0
r q þ 3t rg þ V $ U rg ¼ 0: (8) f ðTÞ B vT q vP
vt b ¼ q þ þ ; (20)
2
f ðTÞ ð1 þ bPÞT vt ð1 þ bPÞP vt
The adsorbate diffusion in and out of the element is generally
negligible compared to the convective component, and hence see [3] for the intermediate steps. Using these expressions in
it is typically neglected in the literature [4]. Assuming constant the mass balance, we have
Author's personal copy
vT vT 1 v a
a11 rg aPg Ur þ rg ðrUr Þ ¼ 0: (51) 140
vt vr r vr B
t = 0.5 min
Expanding the last term gives
1.0 min
vT vT vUr rg Ur 120
a11 rg aPg Ur þ rg þ ¼ 0: (52)
vt vr vr r
1.5 min
T (K)
vT vT 1 v vT
a21 þ rg CPg Ur leff r ¼ 0: (55)
vt vr r vr vr 110 2.0 min
Expanding the last term and rearranging we have 1.5 min
vT leff vT v2 T
a21 þ rg CPg Ur leff 2 ¼ 0: (56)
100
1.0 min
vt r vr vr
We solve the coupled mass and energy balance equations in t = 0.5 min
COMSOLª, with the initial conditions: U ¼ 0 and T ¼ T0, all over B
the domain at t ¼ 0, and the boundary conditions U ¼ Uf and T ¼ Tf
90
at the feed side, and vT=vx ¼ 0 or vT=vr ¼ 0 at the exit side. This 0 0.5 1
Neumann-type exit boundary condition for flow systems is x (m)
popularly known as the Danckwerts’ boundary condition [20].
Theoretically, this boundary condition can be applied only for Fig. 4 – (a) Temperature and (b) mass flow rate evolution in an
a steady system. For an unsteady system, this boundary condition axial flow bed. At t [ 0.5 min, the region 0–A is the saturated
is believed to suppress the gradients near the exit or induce region at feed temperature (80 K), the region A–B is the
numerical instability. However, in Section 6, we observe that adsorption zone where the bed is simultaneously cooling
there is no perceptible gradient suppression near exit in our (from 140 K to 80 K) and adsorbing, and the region B–1 is the
results, even when the adsorption front is moving out of the bed. non-adsorbing region at initial bed temperature (140 K).
Author's personal copy
the initial bed temperature. Hence, beyond point B, the gas Once we have the temperature and pressure evolution of the
just passes through the bed without cooling the bed any bed, integrating this expression over the entire length of the
further. Thus, in the region B–1 the bed is saturated at its bed gives the total hydrogen content of the bed at any time.
initial temperature. As time progresses region A–B, the We define the refueling time as the time at which the total
adsorption zone, moves downstream, leaving the bed hydrogen content of the bed is 5 kg (as mentioned in Section 2,
upstream fully saturated at 80 K. In the flow rate evolution we do not consider the gaseous hydrogen storage in the dead-
within the bed in Fig. 4, the same inferences could be seen. In spaces). For the isobaric axial bed, with a 120 g/s feed at 80 K
the adsorption zone A–B flow rate decreases due to adsorp- and 20 bar, we found the refueling time to be 135.8 s or
tion, and in the saturated zones 0–A and B–1 the flow rate 2.26 min. From Fig. 4, we can see that even by the end of
remains constant. Once the bed is fully saturated, gas merely refueling significant temperature gradients (about 40 K) exist
passes through the bed without change in flow rate. Fig. 5 within the bed. We will return to this observation later in this
shows the gas density and velocity evolution within the bed. section.
Density changes almost by a factor of two within the bed,
showing that the temperature dependence of thermodynamic
properties should be properly accounted. In the adsorption 6.1. Comparison of 1-D axial bed model and 0-D
zone, it is seen that the gas velocity increases steeply due to model results
sharp rise in temperature and consequent fall in density,
occurring within the adsorption zone. Fig. 6 shows the hydrogen capacity evolution predicted by the
Consider a circular slice of the axial bed at x, of elemental 1-D axial bed model and by the lumped-parameter model for
thickness Dx. Its total hydrogen content is obtained by summing the isobaric refueling period. The respective isobaric refueling
up the adsorbed phase and gas phase contributions as: times are 5.86 and 2.26 min. (Note that this 0-D result is
without the steel thermal mass and heat leak into the system,
pR2 Dxrb q ½Tðx; tÞ; Pðx; tÞ þ pR2 Dx3t rg ½Tðx; tÞ; Pðx; tÞ: (57)
so that the 0-D and 1-D results can be compared on a common
basis.) Such a large discrepancy in the hydrogen capacity
evolution and hence the refueling time is due to the significant
temperature gradients existing within the bed during the
isobaric refueling period, as seen in Fig. 4. We expect lesser
disparity between the 0-D and the 1-D models for slower
processes like discharge, dormancy and venting, since the
temperature gradients are expected to be milder.
In order to check this large discrepancy between the 0-D
and 1-D models during the isobaric refueling period, we
implemented a tanks-in-series model (not reported here).
A 0-D model is a single tank model, while the 1-D model is
equivalent to a sufficiently large number of tanks-in-series. In
fact, serially coupled 0-D models lead to a tanks-in-series
model. A tanks-in-series model showed that the two curves
shown in Fig. 6 can be smoothly interpolated as the number of
tanks increases from 1 to about 25.
4
H load (kg)
3
2
2 0−D
1
0 2 4 6
time (mins)
Interestingly, from Fig. 6, we see that a 0-D model always from the bed and back into the gas phase, resulting in a slight
predicts lower hydrogen capacity than a 1-D model. The basic increase in the gas pressure.
distinction between the 0-D and 1-D models is the presence of Earlier in this section, we noted that even at the end of
temperature gradients in the latter. The presence of tempera- refueling significant temperature gradients, up to 40 K temper-
ture gradient in a 1-D model facilitates higher hydrogen ature difference exists within the bed. If the tank is isolated after
capacity than the 0-D model. This can be demonstrated by the refueling, with the temperature gradient slowly equilibrating, as
heuristic model shown in Fig. 7. Consider a kilogram of MOF at we note in this section, due to desorption back into the gas phase
80 K and another kilogram of MOF at 120 K being brought in there can be a slight overshoot in the tank pressure.
thermal contact. Assuming that their specific heat capacities
do not change significantly, we could say that the equilibrium
temperature of the composite block (neglecting for the 7. Results for annular bed designs and
moment, the energetics of adsorption or desorption) will be comparison of the three bed designs
about 100 K. From the adsorption data in the Appendix, one
could compute the excess adsorbed hydrogen at 20 bar in 1 kg Figs. 8 and 9 show the simulation results for the single cartridge
MOF at 80 K as 49.73 g, 1 kg of MOF at 120 K as 19.15 g and 2 kg of annular bed. Similar result hold seven cartridge annular bed
MOF at 100 K as 61.69 g. This calculation indicates that 7.19 g of (figures not shown), except that one seventh of the total flow is
hydrogen is returned to the gas phase when the temperature apportioned to each cartridge. The phenomenon of simulta-
gradients equilibrate within the composite block. Note that neous cooling and adsorption in annular beds is similar to that
this heuristic model does not capture the ‘‘secondary’’ effects in the axial bed, except for the geometric effects. From Fig. 8, it
which occur in reality, but illustrates the fact that a bed with can be seen that the adsorption zone moves rapidly near the
gradients has more adsorbed hydrogen than a bed at its inner diameter and slowly near the periphery. In Fig. 9, at long
equilibrated temperature, and thus it explains qualitatively the times the superficial gas velocity approaches a rectangular
discrepancy between the lumped-parameter and 1-D models. hyperbola, since for the non-adsorbing case the radial velocity
We can understand the above fact from a thermodynamic at any location is given by Eq. (6).
perspective too. If an adsorbent bed with internal temperature
gradients is isolated from its environment, the second law of
thermodynamics dictates that, on equilibration the system
a
140
entropy should increase. In most cases of gas physisorption,
the entropy of the adsorbed phase is lesser than that of the gas t = 0.5 min
phase. (If the adsorbed molecule dissociates on the surface, it
could favor an increase in entropy. But compared to the gas 120 1.0 min
phase, typically an adsorbed phase has much lower degrees of
T (K)
b
49.73 g H2 19.15 g H2 3.0 min
120
1 kg MOF 1 kg MOF
2.5 min
+
mass flow rate (g/s)
1.0 min
100
2 kg MOF Returned
at 100 K & to
t = 0.5 min
20 bar gas phase
61.69 g H2 7.19 g H2 90
0.05 0.1 0.15 0.2 0.25
r (m)
Fig. 7 – Heuristic model to demonstrate that a bed with
temperature gradients contains more adsorbed hydrogen Fig. 8 – (a) Temperature and (b) mass flow rate evolution in
than the bed at its average temperature. a single cartridge annular bed.
Author's personal copy
a
3.0 min
5
6
2.5 min
2.0 min 4
H load (kg)
(kg/m )
3
5
1.5 min 3
2
g
Axial
1.0 min Annular−1
4 2
Annular−7
t = 0.5 min
1
3 0 1 2 3
0.05 0.1 0.15 0.2 0.25 time (mins)
r (m)
Fig. 10 – Hydrogen capacity evolution among the different
packed-bed designs.
b 0.6
0.035
t = 0.5 min
0.03
flow maldistribution. The material cost increases with the
0.025 number of cartridges due to an increase in the mass of
0.4 1.0 min structural steel (increased diameter of pressure vessel and
U (m/s)
Fig. 10 compares the hydrogen capacity among the three The results presented in above sections are for a constant feed
different packed-bed designs under consideration. Surpris- flow rate of 120 g/s. Now we present the refueling time at
ingly, there is negligible difference between the three designs different feed flow rates, for the three designs under consid-
in terms of hydrogen capacity evolution. This observation eration, in Table 1. Especially, in the axial bed results, we see
may be explained as follows: since the internal surface area that doubling the feed flow rate approximately halves the
within the pellets is very high, the adsorbent is already in refueling time. This scaling can be reasoned as follows: in this
intimate contact with the gas. Hence, rearranging the bed isobaric model, effective thermal conductivity of the bed is the
geometry does not make much difference in the rate at which only transport property which contains the dimension of
adsorption happens.
The primary motivation of this study was to answer the
following questions: for a fixed mass of adsorbent, is there any
process advantage in increasing the number of cartridges in Table 1 – Refueling time for the three bed designs at
an annular bed design and is there an optimum number of various feed flow rates.
cartridges during the isobaric refueling period? Since we do Feed flow Refueling time (s)
not find significant difference in the hydrogen capacity rate (g/s)
evolution, we infer that there is no significant heat and mass Axial bed Single cartridge Seven cartridge
annular bed annular bed
transfer advantage in moving from axial bed to single
cartridge annular bed or to seven cartridge annular bed 240 68.0 68.2 69.5
designs. Hence, the only difference in these design enhance- 120 136.1 136.9 141.8
ments is the decrease in pressure drop, as we noted in Section 60 272.4 275.7 294.0
30 545.6 557.8 625.0
3. Increase in the number of cartridges in an annular design
15 1093.6 1140.1 1380.5
results in higher material cost, fabrication complexity and
Author's personal copy
time, and all the remaining properties are thermodynamic or For all the bed designs, we find that the product of refueling
material properties which do not contain the dimension of time and feed flow rate is approximately constant for a given
time. When the effective thermal conductivity is low, as in feed temperature and pressure, i.e. doubling the feed flow rate
MOF, it does not significantly contribute to the thermal halves the refueling time. This scaling, though approximate,
diffusion of the adsorption front through the bed, leaving helps us in estimating the refueling time at arbitrary flow rates
superficial gas velocity or flow rate as the main variable given the refueling time at a flow rate.
scaling the refueling time. This scaling, though approximate,
helps us in estimating the refueling time at arbitrary feed flow
rates given the refueling time at a flow rate, since the product
Acknowledgments
of refueling time and feed flow rate is approximately
a constant, for a given feed temperature and pressure.
The authors would like to thank Raghunathan K, Michael
Compared to the axial bed, the single cartridge annular bed
Herrmann, Scott Jorgensen, Ulrich Eberle, Dieter Hasenauer,
design has a lower flow path length. As the number of
Rainer Immel, Gregory Meisner, Anne Dailly and Eric Poirier
cartridges increases, the radial thickness of the bed and hence
for technical data and/or valuable suggestions during the
the flow path length decreases. As flow path length decreases,
course of model development. Senthil Kumar would like to
the effect of thermal conductivity of the adsorbent on the
acknowledge constructive comments from his group
thermal diffusion of the adsorption front increases. Hence, we
members at India Science Lab during his presentations while
can see that the above scaling is less accurate as we move
the model development was in progress.
from the axial bed to single cartridge annular bed to seven
cartridge annular bed designs.
Appendix A. Simulation input data
annular bed design without increasing the material cost, As in [3], the remaining adsorbent parameters are
fabrication complexity and flow maldistribution. DHa ¼ 4:0 kJ=mol ¼ 2 106 J=kg, leff ¼ 0:32 W=m=K, and
Author's personal copy
X
8
n [4] Mota JPB, Rodrigues AE, Saatdjian E, Tondeur D. Dynamics of
log10 CPs ¼ cn log10 T ; (A.4) natural gas adsorption storage systems employing activated
n¼0
carbon. Carbon 1997;35:1259–70.
where (c8, ., c0) ¼ (0.447169361516, 5.888038301049, [5] Arumugam BK, Banks JF, Wankat PC. Pressure effects in
32.914887850468, 101.724106321425, 189.552581009475, adsorption systems. Adsorption 1999;5:261–78.
[6] Gopal MR, Murthy SS. Prediction of heat and mass transfer in
217.283997486133, 148.061506476350, 51.560061018285,
annular cylindrical metal hydride beds. Int J Hydrogen
6.714743948231), CPs is in J kg1 K1 and T in K.
Energy 1992;17:795–805.
The bed density and porosity values are from [3]: [7] Fukada S, Morimitsu S, Shimoozaki N. Heat and mass
rb ¼ 385:05 kg=m3 ; 3b ¼ 0:36; 3t ¼ 0:7837. transfer in concentric-annular-tube bed packed with ZrV1.
9Fe0.1 particles. J Alloys Compd 2004;375:305–12.
A.3. Initial and boundary conditions [8] Liu Y, Leong KC. The effect of operating conditions on
performance of zeolite/water adsorption cooling systems.
Appl Therm Eng 2005;25:1403–18.
Using the lumped-parameter model reported in [3], for a feed
[9] Yoo CS, Dixon AG. Modeling and simulation of a mixed-flow
at 20 bar pressure and 80 K temperature, and a flow rate of
reactor for ammonia and methanol synthesis. Chem Eng Sci
120 g/s, we get a temperature 134.25 K at the end of the initial 1988;43:2859–65.
adsorption period. In a 1-D model we can neglect the thermal [10] Gu T, Tsai GJ, Tsao GT. A theoretical study of
mass of the structural components and the heat leak into the multicomponent radial flow chromatography. Chem Eng Sci
tank through them, since the focus is only on obtaining the 1991;46:1279–88.
data for bed dynamics. So we recalculated the lumped- [11] Ponzi PR, Kaye LA. Effect of flow maldistribution on
conversion and selectivity in radial flow fixed-bed reactors.
parameter model result for the initial adsorption period
AIChE J 1979;25:100–8.
neglecting the steel thermal mass and heat leak into the tank. [12] Bolton GT, Hooper CW, Mann R, Stitt EH. Flow distribution
A tank initially at 1.1 bar and 120 K had a temperature of and velocity measurement in a radial flow fixed bed reactor
140.87 K at the end of initial adsorption period i.e. when the using electrical resistance tomography. Chem Eng Sci 2004;
tank pressure reached 20 bar. Since we have not yet per- 59:1989–97.
formed the 1-D analysis for the initial adsorption period, we [13] Heggs PJ, Ellis DI, Ismail MdS. The modeling of fluid-flow
arbitrarily assume the initial condition for the isobaric refuel- distributions in annular packed beds. Gas Separ Purif 1994;8:
257–64.
ing period as P ¼ 20 105 Pa, T0 ¼ 140 K, since it is represen-
[14] Kareeri AA, Zughbi HD, Al-Ali HH. Simulation of flow
tative of typical operating conditions. For the axial, single distribution in radial flow reactors. Ind Eng Chem Res 2006;
cartridge annular and seven cartridge annular designs the 48:2862–74.
flow velocities at the feed side are 0.08237 m/s, 0.6154 m/s and [15] Bird RB, Stewart WE, Lightfoot EN. Transport phenomena.
0.0879 m/s. We observed the following simulation artifact 2nd ed. NY: John Wiley & Sons; 2002.
with the Lagrange elements: upstream of the adsorption front [16] NIST web book. http://webbook.nist.gov/chemistry/fluid/.
[17] Kaviany M. Principles of heat transfer in porous media. 2nd
there was a small dip in the temperature even below the feed
ed. NY: Springer-Verlag; 1995.
temperature, which is surely unphysical. However, with the
[18] Ahluwalia RK, Peng JK. Dynamics of cryogenic hydrogen
use of Hermite cubic elements, this artifact vanished. storage in insulated pressure vessels for automotive
applications. Int J Hydrogen Energy 2008;33:4622–33.
[19] Sereno C, Rodrigues A. Can steady-state momentum
Appendix B. Supplementary data equations be used in modelling pressurization of adsorption
beds? Gas Separ Purif 1993;7:167–74.
[20] Danckwerts PV. Continuous flow systems: distribution of
Supplementary data associated with this article can be found,
residence times. Chem Eng Sci 1953;2:1.
in the online version, at doi:10.1016/j.ijhydene.2010.01.025. [21] Parulekar SJ, Ramakrishna D. Analysis of axially dispersed
systems with general boundary conditions – I. Chem Eng Sci
1984;39:1571–9.
references [22] Lee TT, Wang FY, Newell RB. On the evaluation of the exit
boundary condition for the axial dispersion bioreactor
system. Chem Eng Technol 1998;21:901–10.
[1] Zhou L. Progress and problems in hydrogen storage methods. [23] Huang WC, Chou CT. Comparison of radial and axial-flow
Renew Sustain Energy Rev 2005;9:395–408. rapid pressure swing adsorption processes. Ind Eng Chem
[2] Dailly A, Vajo JJ, Ahn CC. Saturation of hydrogen sorption in Res 2003;42:1998–2006.
Zn benzenedicarboxylate and Zn naphthalenedicarboxylate. [24] Sankararao B, Gupta SK. Modeling and simulation of fixed
J Phys Chem B Lett 2006;110:1099–101. bed adsorbers (FBAs) for multi-component gaseous
[3] Kumar VS, Raghunathan K, Kumar S. A lumped-parameter separations. Comput Chem Eng 2007;31:1282–95.
model for a cryo-adsorber hydrogen storage system. Int J [25] Thomas JM. The existence of endothermic adsorption.
Hydrogen Energy 2009;34:5466–75. J Chem Educ 1961;38:138–9.