Você está na página 1de 22

Efficient Dynamic Charge and Discharge of an

Adsorbed Natural Gas Storage System


V. GOETZ AND S. BILOE

IMP-CNRS UPR 8521, Institut de Science et Genie des Materiaux


et Procedes, Perpignan, France
Dynamic methane charges and discharges of a 2 L adsorbed natural gas (ANG) ves-
sel, filled with two different kinds of adsorbent composite blocks (ACBs), were
studied. The two ACBs show the same isothermal stored methane capacity, that
is, 100 V=V. The first one is highly conductive but has low permeability, and the
second one is highly permeable but has low thermal conductivity. From experimental
results, it is shown that the heat transfer phenomena mainly govern the ACBs per-
formance for dynamic methane storage. On the other hand, the coupling between
heat and mass transfer within the ACB becomes actual for the dynamic methane dis-
charge process. The simulation model, which was validated here for both the charge
and discharge steps and for the two ACBs, was used for a preliminary dimensioning
of a real-size ANG vessel.
Keywords Adsorption; Gas storage; Heat transfer; Mass transfer
Introduction
Natural gas (NG) is a promising alternative as a transportation fuel (Parkyns and
Quinn, 1995) that may make it possible to significantly reduce the level of air pol-
lution. However, the disadvantage of NG is its low volumetric energy density com-
pared to that of conventional liquid fuels. Current natural gas vehicles (NGVs)
employ storage vessels at high pressure (P 20 MPa). This implies high manufactur-
ing and filling costs. Adsorbed natural gas (ANG) on suitable microporous adsor-
bents offers interesting opportunities for developing NGV technology. ANG
technology can provide adequate energy density at a low pressure, that is, 3.5 MPa
and at room temperature (Wegrzyn and Gurevich, 1996). To be really competitive
with conventional fuels, a 150-cycled volume of natural gas per unit of adsorbent
volume, measured at normal conditions (P 0:1 MPa, T 273 K), has to be
reached. This value is lower than that obtained by compressed natural gas (CNG):
220 V=V at 20 MPa and at room temperature. However, an ANG storage vessel
can be filled with NG from a domestic pipeline using a single-stage compressor only,
compared to the expensive multistage compressor used for CNG technology. Conse-
quently, a value of 150 V=V represents a compromise.
The development of suitable microporous adsorbents for ANG storage is a cur-
rently active area of research (Alca~ nniz-Monge et al., 1997; Baker, 1998; Biloe et al.,
Received 10 October 2002; in final form 22 November 2003.
Address correspondence to V. Goetz, IMP-CNRS UPR 8521, Institut de Science et Genie
des Materiaux et Procedes, Rambla de la Thermodynamique, 66100 Perpignan, France.
E-mail: goetz@univ-perp.fr, biloe@univ-perp.fr
Chem. Eng. Comm., 192:876896, 2005
Copyright # Taylor & Francis Inc.
ISSN: 0098-6445 print/1563-5201 online
DOI: 10.1080/009864490510761
876
2001a; Burchell and Rogers, 2000; Inomata et al., 2002; MacDonald and Quinn,
1996, 1998; Menon and Komarneni, 1998). In the design of a successful ANG sto-
rage system, different scales have to be taken into account.
First, is the scale of the microporous solid. The microporous structure allows
defining the theoretical performance. The definition of the optimal pore size distri-
bution is often made with the general assumption of an isothermal charge and dis-
charge process (Bojan et al., 1992; Cracknell and Gubbins, 1992; Cracknell et al.,
1993; Matranga et al., 1992; McEnaney et al., 1998; Nicholson, 1998; Quirke and
Tennison, 1996; Chen et al., 1997). A common performance criterion is the available
methane capacity, defined as:
Q
av
Q
s
T
o
; P
c
Q
s
T
o
; P
d
1
where Q
s
(T
o
, P
c
) and Q
s
(T
o
, P
d
) are the total stored gas capacity at room tempera-
ture and charge pressure (P
c
3:5 MPa) and at room temperature and depletion
pressure (P
d
0:1 MPa), respectively. Several authors often refer to it as the iso-
thermal gas capacity. From molecular simulation, Matranga et al. (1992) have
shown, for example, that an optimal pore size of 1.14 nm maximizes the available
methane capacity, measured over an operating cycle (3.41.4 MPa, T 288 K).
As a result, the available methane capacity is close to 200 V=V for a monolithic
adsorbent made of activated carbon. Nevertheless, charging and discharging an
ANG storage system generate large heat- and mass-transfer limitations. For this pur-
pose, we have recently used the well-known Dubinin-Astakhov equation (Dubinin
and Astakhov, 1971) to quantify, for a given microstructure of an activated carbon,
the effects of both the heat- and mass-transfer limitations during the charge and the
discharge processes (Biloe et al., 2002).
Second is the packing adsorbent scale. Elaboration of packing adsorbent is com-
monly carried out by either the use of microporous solid only or the simultaneous
use of microporous adsorbent and a binder (phenolic resin, polymer, etc.). The heat-
and mass-transfer properties of the packing adsorbent will directly influence the per-
formance of an ANG storage system. Consequently, the performance obtained in
real working conditions differs from that in theoretical conditions. The main experi-
mental (Chang and Talu, 1996; Jasionowski et al., 1989) or simulated results (Mota,
1995; Mota et al., 1997a,b) in the literature deal with the heat-transfer limitation
only. Nevertheless, to reach a high gas storage capacity, expressed in terms of stored
volume of gas per adsorbent volume, it is necessary to realize adsorbents with a very
high density. They are traditionally obtained by mechanical compression of the
microporous solid (Greebank, 1992; Manzi et al., 1997; Rubel and Stencel, 2000).
Such materials could involve important mass-transfer limitations (Cook et al.,
1999). Thus, it is obvious that the elaboration of adsorbent blocks is a crucial step
in designing an ANG storage system.
Third is the ANG vessel scale. Good adequacy is necessary between the ANG
vessel and the heat- and mass-transfer properties of the packing adsorbent, as well
as the real working conditions. Both the charge and the discharge processes are gen-
erally performed at a constant flow rate. Nevertheless, an interesting practical objec-
tive is to minimize the load time, which happens with an uncontrolled inlet flow rate.
Thus, at t 0, a constant pressure of 3.5 MPa is imposed instantaneously at the inlet
of the ANG vessel. To minimize the temperature increase related to exothermic
adsorption, the enhancement of the heat exchange from the vessel to the
surroundings is needed, for example.
Efficient Dynamic Charge and Discharge. 877
A study has been reported already on the dynamic methane discharges on a 2 L
vessel filled with a new adsorbent composite block (ACB) (Biloe et al., 2001b). The
first objective was to demonstrate experimentally, through a laboratory device, the
functioning of the proposed ACB under various working conditions. As a result, a
delivered methane capacity of 100 V=V was reached for a large range of methane
flow rates (1 < v < 10 L
NTP
min
1
). The second objective was to validate a
dynamic simulation model coupling heat and mass transfer within the ACB. Finally,
from the analysis of the thermodynamic path of the adsorbent, a qualitative descrip-
tion of the coupling phenomena was proposed.
The purpose of the present work is to continue this study. First, the model vali-
dation is extended to the charge process by comparison with new experimental
results. To our knowledge, no experimental results on the charge process of an
ANG vessel have been reported in previous articles. Then, two different ACBs are
tested in order to illustrate the relevance of the transfer characteristics of each
ACB during both the charge and the discharge processes. The first one, named
ACB
k
, is highly conductive but has a relatively low permeability. On the other hand,
the second one, named ACB
k
, has a very high permeability but a low thermal con-
ductivity. These adsorbents were tested in the same ANG vessel under the same
working conditions. From experimental results, we show that different transfer
properties are needed for both the charge and the discharge processes.
Characteristics of Adsorbent Composite Blocks
The adsorbent composite blocks are a mixture of a superactived carbon, labeled
Maxsorb
1
, expanded natural graphite used as a thermal binder, and a thermoplastic
polymer used as a mechanical binder [Biloe et al., 2001a]. First, Maxsorb
1
was
coated with the thermoplastic binder according to Boses patent (Bose et al.,
1991). Second, it was mixed with expanded natural graphite, and the obtained mix-
ture was compressed directly in the ANG cylindrical vessel and then heated up to the
melting temperature of the polymer. Thereafter, the ACB was air cooled under
applied stress. The polymers solidification ensures mechanical cohesion as well as
high level of density of the composite blocks. In addition, a gas diffuser was inserted
along the symmetry axis of each ACB. ACBs are characterized by both their
apparent density (q
ACB
) and their weight ratio of expanded natural graphite (w
ENG
).
The relevant characteristics of each ACB are listed in Table I.
Table I. Characteristics of the adsorbent composite blocks
ACB
k
ACB
k
q
ACB
kg m
3
448 330
q
ac
kg m
3
331.5 327
w
ENG
(%) 26 1
kW m
1
K
1
7.3 0.37
k (m
2
) 1 10
14
2:4 10
12
Q
av
V V
1
103 104
Activated carbon Maxsorb
1
Maxsorb
1
Dubinin parameters W
o
1:17 cm
3
g
1
; E
o
15; 884 J mol
1
; n 1:39
878 V. Goetz and S. Biloe
As the two binders, that is, expanded natural graphite (ENG) and thermoplastic
polymer, are inert according to methane adsorption, the available methane capacity
depends only on the activated carbon adsorption isotherm as well as its apparent
density in the ACB. For both ACBs, the available capacity is close to 100 V=V
(Table I). The heat- and mass-transfer properties of the ACBs depend on both the
apparent density of the ACB (q
ACB
) and the weight ratio of ENG in the ACB
(w
ENG
). The transfer properties of ACB
k
have been determined previously (Biloe
et al., 2001b). For ACB
k
, the permeability has been estimated by the Carman-
Kozeny correlation (Carman, 1956):
k
e
3
m
d
2
180 1 e
m

2
2
where e
m
and d are the macroporosity between the activated carbon particles and the
mean diameter of the activated carbon particles, respectively. The thermal conduc-
tivity of ACB
k
has been identified with the help of the experimental thermal gradi-
ents within the ACB during a charge step at a constant methane flow rate.
Experimental Study and Model
The experimental laboratory device (Figure 1) has been detailed previously (Biloe
et al., 2001b). Here, we review the main points:
. A 2 L ANG vessel cylindrical in shape (diameter length: 110:3 215 mm);
. The gas used is pure methane (99.99%);
. A controlled methane flow rate during the charge and discharge processes, via two
mass flow controllers. The values measured by these mass flow controllers (Brooks
5850 S) are expressed in Nl mn
1
with accuracy equal to 1%of the full scale, that
is, 10 Nl mn
1
. They are directly used in the calculation of the quantities of meth-
ane involved during the charge and discharge phases;
. A measurement of both the inlet and outlet pressures of the ANG vessel during the
charge and the discharge processes, respectively;
. A measurement of the thermal gradients with the help of four thermocouples dis-
tributed radially throughout the ANG vessel and one placed on the reactor wall.
. A controlled temperature at the reactor external surface by flowing water by
forced convection in a double casing. The heat exchange coefficient between water
and the external surface area (h
tf
) is equal to 700 W m
2
K
1
. Thanks to the in
situ consolidation, the heat exchange coefficient between the ACB and the internal
surface of the vessel can be considered as not limiting.
At the start of each charge step, the vessel was at room temperature
(T
o
298 K) and depletion pressure (P
d
0:1 MPa). For the current working con-
ditions, the experiments were carried out until the vessel returned to a uniform tem-
perature (T T
o
) and at charge pressure (P
c
3:5 MPa). At the start of each
discharge step, the vessel was at room temperature and charge pressure. Whatever
the working conditions, the experiments were carried out until the vessel returned
to a uniform temperature (T T
o
) and at depletion pressure (P
d
0:1 MPa).
The system is modeled in the form of three distinct media (Figure 2): the ACB,
the external fluid, and the vessel tube separating them. Charge and discharge are
Efficient Dynamic Charge and Discharge. 879
simulated solving the mass and heat balances in the ACB with the assumption of a
local equilibrium among the adsorbed gas phase, the solid phase, and the gaseous
phase. The geometry ratio (diameter=length) of the vessel and the presence of a
gas diffuser allow us to consider a one-dimensional problem according the radial
direction. The set of non linear differential equations is solved by the control-volume
method (Patankar, 1980).
Mass balance is given by:
@
@t
e
o
qq
g
q
ac
q
_
r q
g
u
g
_ _
0 3
where e
o
is the accessible porosity for the gas, varying with the amount adsorbed q.
The gas density q
g
and the amount adsorbed q are related to both the local pressure
P and the local temperature T according to the ideal gas equation and the well-
known Dubinin-Astakhov (DA) equation (Dubinin and Astakhov, 1971) respectively:
q
g

M
g
P
RT
4
qP; T q
a
W
o
exp
A
bE
o
_ _
n
_ _
5
Figure 1. Experimental laboratory device; 0, gas reservoir; 1, pressure regulator; 2, vacuum
pump; 3, filter; 4, mass flow controller; 5, pressure transducer; 6, filter 2 mm; 7, filter 7 mm;
8, safety valve; 9, type K thermocouple; 10, reactor; 11, balance; 12, closed room; 13, venti-
lator; 14, data acquisition system.
880 V. Goetz and S. Biloe
Polanyis adsorption potential A is given by:
A RT ln
P
sat
P
_ _
6
where P
sat
is the saturated vapor pressure. The adsorbed gas density q
a
and the satu-
rated vapor pressure P
sat
were calculated according to Dubinins (1975) and Osawas
(Osawa et al., 1976) expressions respectively. The superficial velocity of the gaseous
phase, u
g
, in the ACB is determined by Darcys law (Dullien, 1992):
u
g

k
l
g
rP 7
The heat balance in the ACB is given by:
@
@t
C
t
T r krT q
ac
Dh j j
M
g
@q
@t
0 8
where C
t
is the overall volumetric heat capacity:
C
t
qC
p

ACB
qqC
p

a
9
where (qC
p
)
ACB
and (qC
p
)
a
are the volumetric heat capacity of ACB and adsorbed
gas phase respectively. Here, the volumetric heat capacity of the gaseous phase is
considered negligible. The heat capacity per unit mass of the adsorbed gas phase,
(C
p
)
a
, has been taken equal to that of the liquid phase.
The initial conditions are:
Tr; t 0 T
o
10
Pr; t 0 1 MPa; for the charge 11
Figure 2. Schematic representation of the experimental vessel with the three distinct media
(r
o
0:5 mm, r
1
53:15 mm).
Efficient Dynamic Charge and Discharge. 881
P r; t 0 3:5 MPa; for the discharge 12
The boundary conditions are expressed as follows:
@T
@r

rr
0
0 13
k
@T
@r

rr
1
h
st
T
t
Tj
rr
1
_ _
14
and
@P
@r

rr
1
0 15
where T
t
is the temperature of the tube, r
o
is the gas diffuser border, and r
1
is the
vessel inner radius. At r r
o
, a constant discharge flow rate, v, was applied accord-
ing a calculation procedure described in detail previously (Biloe et al., 2001b).
The tube exchanges heat with the ACB and the external fluid through a heat
exchange coefficient h
st
and h
tf
respectively. The heat equation for the tube is:
qC
p

t
@T
t
@t

h
st
S
st
V
t
Tj
rr
1
T
t

h
tf
S
tf
V
t
T
f
T
t
16
where S
st
and S
tf
are the different exchange surfaces, and V
t
is the volume of the tube
wall. As regard to the heat exchange coefficient h
st
, one assumes that the transfer of
heat is not limiting, compared with h
tf
(700 W m
2
K
1
). As a matter of fact, the
procedure of consolidation ensures a very good contact between the ACB and the
surfaces reactor. The value selected is 2500 W m
2
K
1
.
The initial conditions for the tube and water, the external fluid, are:
T
t
T
f
T
o
298 K 17
Results and Discussion
Dynamic Charge
Figures 3(a) and 3(b) show the thermal behavior of each ACB during the charge step.
At t 0, an inlet constant methane flow rate (v 7:5 L
NTP
min
1
) is applied. This
places the ACB in adsorption conditions. Due to the heat of adsorption (exothermic
process), the temperature inside the ACB increases. As a result, the temperature
gradients mainly depend on the heat transfer properties of the ACBs.
For ACB
k
(Figure 3(a)), the thermal conductivity is high enough to perform
quasi-isothermal adsorption. After a small increase of all temperatures, the charge
step goes on at a constant temperature. The temperature drop between the external
heat transfer fluid (water) and the ACB corresponds to the temperature gradient,
which is necessary in order to extract the required adsorption heat flux. At t s,
the charge step at constant flow rate is no more respected. Adsorption heat flux is less
important. As a result, the temperatures decrease and return quickly (10 min) to room
temperature T
o
. This state corresponds to a complete charge of the vessel. During all
the charge process, the maximum temperature gradient inside ACB
k
is lower than
4 K. This indicates clearly that the heat transfer by conduction is not limiting.
882 V. Goetz and S. Biloe
In addition, the simulated values are in good agreement with those experimental
values for both the vessel wall temperature and the temperature profiles within the
ACB
k
.
For ACB
k
(Figure 3(b)), because of its low conductivity, the adsorption heat flux
generates an important temperature gradient. This gradient indicates that for ACB
k
,
the coupling between heat transfer by conduction and adsorption heat production
becomes effective. During the charge step at a constant methane flow rate (t < s),
the temperatures within the ACB increase continuously. As a result, the temperature
gradient inside ACB
k
reaches a maximum value of 35 K when t s. Then, as for the
previous case, the process at a constant flow rate is no more respected. Nevertheless,
for ACB
k
, the decrease in the temperatures that allows reaching a complete filling
happens slowly, that is, 35 min. In addition, an interesting phenomenon is brought
to light by the evolution of T
1
as a function of time. According to its position inside
the tank (r 0:011 m), that is, near the gas diffuser and so far from the heat exchange
surface, this thermocouple should be at the highest temperature. Nevertheless, as the
methane flows inside the vessel at room temperature, it provides a cooling effect that
locally minimizes the temperature rise. The simulation model does not take into
account the heat transfer through the gaseous phase. Indeed, the gaseous phase
and the solid phase are thought to be in equilibrium. This observation explains the
Figure 3. Experimental (symbols) and simulated (lines) temperatures at different radial posi-
tions for (a) ACB
k
and (b) ACB
k
, for a methane flow rate of v 7:5 L
NTP
min
1
. T
1
, T
2
, T
3
,
T
4
are the temperatures at radius r 11, 22, 33, and 44 mm, respectively.T
t
is the temperature
at the vessel wall. T
1s
, T
2s
, T
3s
, T
4s
, and T
ts
are the corresponding simulated temperatures.
Efficient Dynamic Charge and Discharge. 883
largest discrepancy between the experimental temperature T
1
and the simulated one
(T
1s
). If the methane convection has a local impact near the gas diffuser, as demon-
strated later, it does not affect the global performance of the ANG vessel.
For both the ACBs, Figures 4 and 5 show the experimental and simulated press-
ure profiles in the gas diffuser and the charge flow rate profiles, respectively. As illu-
strated by the two figures, a constant charge flow rate cannot be ensured any longer
when the pressure in the gas diffuser is close to the charge pressure. For ACB
k
,
the slower increase of the pressure inside the diffuser indicates clearly a better
potentiality to methane adsorption. This is confirmed by the records and the com-
parisons of the inlet charge flow rates for both ACBs.
Performance
The performance of the ACBs was evaluated by dynamic efficiency, which is defined
as follows for the charge step (g
C
) and the discharge step (g
D
), respectively:
g
C

Q
dc
Q
av
18
g
D

Q
dd
Q
av
19
Figure 3. Continued.
884 V. Goetz and S. Biloe
where Q
dc
and Q
dd
are the stored and the delivered methane capacity, respectively,
until the desired flow rate is respected (t s). As described previously, Q
av
is the
available methane capacity, which is close to 100 V=V for both ACBs (Table I).
Figure 6(a) shows the experimental and simulated dynamic efficiency (g
C
) as a
function of the charge flow rate. For both ACBs, the model is in good agreement
with the experimental results for the overall range of variation of the methane flow
rate. The largest discrepancy is not more than 5%. For ACB
k
, that is, the highly con-
ductive activated carbon blocks, the efficiency stays at a high level. Indeed, the
efficiency is higher than 0.95 whatever the flow rate. In spite of its low permeability,
the mass-transfer limitation within the ACB is negligible and does not influence the
overall performance at the scale of the vessel. On the other hand, for ACB
k
, which is
characterized by low thermal conductivity, the stored methane capacity depends
highly on the methane flow rate. As a result, for v 10 L
NTP
min
1
, the dynamic
efficiency is about 0.8, reducing the available methane capacity from 100 to 80 V=V.
Thus, the poor heat transfer by conduction within ACB
k
does not make it possible to
extract the adsorption heat flux. The temperatures stay high and thus limit the
adsorption capacity of the activated carbon. Finally, for the charge step, the thermal
conductivity is the key factor for the design of an ANG storage system.
Figure 6(b) shows the experimental and simulated dynamic efficiency (g
D
) versus
the discharge flow rate. Similarly, good agreement can be seen between simulated
Figure 4. Experimental (symbols) and simulated (lines) pressure profiles in the gas diffuser for
ACB
k
and ACB
k
, for a methane flow rate of v 7:5 L
NTP
min
1
.
Efficient Dynamic Charge and Discharge. 885
and experimental results. As one can see, the dynamic efficiency of both ACBs is
quite similar. For example, the delivered methane capacity is close to 83 V=V and
87 V=V for ACB
k
and ACB
k
, respectively, for v 10 L
NTP
min
1
. Consequently,
the heat-transfer limitation within ACB
k
penalizes the overall performance in the
same order of magnitude as the mass-transfer limitation within ACB
k
. If the physical
causes are different, the consequences are quite similar. In conclusion, for the dis-
charge, optimization of the ACB will result in compromising the heat- and mass-
transfer properties.
The thermodynamic paths of the ACBs in the Clausius-Clapeyron represen-
tation (Figure 7) provide a useful tool to quantify the influence of the heat- and
mass-transfer limitations. In the Clausius-Clapeyron representation, each line corre-
sponds to an isostere where the amount of adsorbed gas remains constant. The lines
are separated from each other by a constant step equal to 0:012 gCH
4
g
1
(activated carbon). Simulated thermodynamic paths were calculated from knowledge
of the average volumetric pressure and temperature of the ACB at each time. The
end of both the charge and the discharge at constant methane flow rate (t s)
was denoted by the symbol O. These points, C
k
and D
k
, for example, characterize
the main limitation either from heat transfer or mass transfer within the ACB.
The thermodynamic path of ACB
k
indicates clearly that the only limitation is
the heat transfer. Indeed, at the end of the charge at a constant methane flow rate
Figure 5. Experimental (symbols) and simulated (lines) charge flow rate profiles for ACB
k
and
ACB
k
.
886 V. Goetz and S. Biloe
(Figure 7, Point C
k
), the average volumetric pressure within the ACB is close to the
charge pressure (P
c
3:5 MPa). Similarly, at the end of the discharge at t s (Fig-
ure 7, Point D
k
), the average volumetric pressure within the ACB is close to the
depletion pressure. On the other hand, the average volumetric temperature is far
from room temperature (T
o
298 K). The temperature drop is about 20 and 30 K
for the charge (Figure 7, Point C
k
) and the discharge (Figure 7, Point D
k
), respect-
ively. Although the temperature rise is more important for the charge than the dis-
charge, the dynamic efficiencies are close to each other. This is caused by the position
of the activated carbons isosteres.
For ACB
k
, isothermal charge and discharge processes are well illustrated.
Although the pressure gradient within ACB
k
is small (DP 0:06 MPa) at the end
of the discharge at a constant flow rate of 7:5 L
NTP
min
1
, it leads to the character-
istic point D
k
that is situated on the same isostere as D
k
. On the other hand, there is
no pressure gradient at the end of the charge at the same constant flow rate charge
(Figure 7, Point C
k
). Consequently, as described previously, the limitation from the
mass transfer does not influence the overall performance. This phenomenon is
attributed to the closed dependence of the level of pressure on the viscous flux N
V
according to:
Figure 6. Dynamic efficiencies vs. various constant flow rates during (a) charge and (b) dis-
charge process, respectively. The symbols and the curves denote the experimental results
and model, respectively.
Efficient Dynamic Charge and Discharge. 887
N
V

k
l
g
P
RT
rP 20
Thus, there is a dominant influence of the mass transfer at low pressure and
especially with an ACB of low permeability. Although these remarks apply to the
ANG vessel design considered here, they clearly demonstrate that the influence of
the heat- and mass-transfer properties on the overall performance is closely linked
to the isosteric network. The isosteric network depends closely on the activated car-
bon microporous characteristics. Thus, the microporous scale as well as the macro-
porous scale should be considered to optimize globally the performance levels under
dynamic conditions.
Charge Process Without a Controlled Methane Flow Rate
In practice, the discharge step operates at a given methane flow rate, which is in
relation to the consumption of an NGV. For bus fleets, the charge can be performed
over a long time, such as overnight. Nevertheless, for the special case of private
Figure 6. Continued.
888 V. Goetz and S. Biloe
vehicles, a very fast filling time is one of the main objectives. This working mode cor-
responds to a charge without a controlled methane flow rate. This could be
accomplished by instantaneously applying a constant charge pressure (P
c
3:5
MPa). For this experimental procedure, flow rate measurements were not possible
as the values exceed the overall range of the mass flow controllers. For this purpose,
the results are presented in terms of average volumetric temperature drop versus time
(Figure 8). In addition, we consider that the vessel is completely filled when the aver-
age volumetric temperature of the ACB is close to room temperature with an accu-
racy of 1%.
As expected, for the charge, the thermal conductivity of the ACB is the key factor
for the design of an efficient ANG storage vessel. For ACB
k
, the maximum tempera-
ture rise is close to 75 K. It is of the same order of magnitude as the result obtained by
Jasionowski et al. (1989) when rapidly pressurizing a vessel filled with AX-21, an acti-
vated carbon similar to Maxsorb
1
. For ACB
k
, at the start of charge, a very fast meth-
ane adsorption also leads to a large increase in temperature. Nevertheless, because of
its high thermal conductivity, the time it takes to return to room temperature is much
more rapid. As a result, the complete filling is performed in less than 15 min while
1.5 h are necessary for ACB
k
. More generally, the difference between a complete fill-
ing of the ANG vessels versus the methane flow rate is shown in Table II.
Figure 7. Average thermodynamic paths of ACB
k
and ACB
k
in the Clausius-Clapeyron rep-
resentation for a methane flow rate of v 7:5 L
NTP
min
1
. The symbols, O, denote the end of
both the discharge and the charge processes at constant flow rate (t s).
Efficient Dynamic Charge and Discharge. 889
Dimensioning of a Real Size ANG Vessel
Based on a delivered methane capacity of 100 V=V, a natural-gas vehicle needs 52 L
of ACBs for a driving range of 300 km, consuming 24 L
NTP
min
1
at 90 km h
1
(Talu, 1992). The dimensioning is made for the charge process only, because of its large
detrimental effect on dynamic performance. If one desired a complete filling of the ves-
sel in 300 s, a reasonable practical objective, the corresponding methane flowrate must
Table II. Total filling time of a 2 L ANG vessel filled with adsorbent composite
blocks as a function of the charge flow rates
vL
NTP
min
1
) t
ACBk
(h) t
ACBk
(h)
1 4.4 4.84
3 1.44 2.28
5 0.98 1.76
7.5 0.62 1.6
10 0.43 1.46
no control 0.17 1.4
Figure 8. Average volumetric temperature drop (mean volumetric ACB temperature minus T
o
)
vs. time for a rapid charge.
890 V. Goetz and S. Biloe
be close to 1040 L
NTP
min
1
. This objective can be reached only with an ideal ANG
vessel dimensioning, that is, without heat- and mass-transfer limitations.
The performance of ANG systems versus time was simulated for several vessel
radii (Figure 9). For each radius, there is a corresponding vessel length in order to
give the desired overall volume of 52 L. Before carrying on the discussion, we define
h, the ANG vessel dynamic load level:
ht
Q
s
T; P
Q
av
21
where Q
s
(T, P) is the stored methane capacity at loading time t. Thus, when the ves-
sel load is performed at a constant methane flow rate, h is a linear function of time.
In addition, at the end of the constant flow rate we get:
ht s g
C
22
For the smallest radius, that is r 0:017 m, the performance of the ACBs is close
to each other. A load level of 0.9 needs almost the same duration, that is 300 and
280 s for ACB
k
and ACB
k
, respectively (Figures 9(a) and 9(b)). Based on classical
heat exchanger technology, a real ANG vessel design consists of a bundle of cylin-
drical tubes, containing the ACB, where the heat external fluid (water) flows in a
Figure 9. Simulated dynamic load level of a 52 L ANG vessel filled with (a) ACB
k
and (b)
ACB
k
, vs. load time and for various ANG vessel radii.
Efficient Dynamic Charge and Discharge. 891
double casing equipped with baffles (Figure 10(a)). According to a standard design,
for an ANG vessel 1.15 m in length, this requires a bundle of 50 tubes, arranged in a
square mesh (a 0:047 m), and a double casing diameter of 0.49 m. The overall vol-
ume, including the heat transfer fluid volume, is close to 215 L. The biggest radius,
that is r 0:121 m, was determined in order to obtain an effective ACB volume that
was close to 52 L for an ANG vessel 1.15 m in length. In that case, the difference
between the overall vessel volume and the vessel volume excluding the double casing
is negligible (Figure 10(b)). For this more compact design, the only possibility for
reaching a complete filling of the vessel in a short duration is provided by ACB
k
.
As a result, a dynamic load level of 0.9 takes only about 700 s for ACB
k
. On the
other hand, more than two hours are required for ACB
k
. Even if this example cor-
responds to the particular case of an ANG vessel in cylindrical shape and for water
as a heat transfer external fluid, it shows the possibilities for highly conductive
ACBs. Basically, this kind of material should allow combining two criteria for the
development of ANG technology: compactness of the overall system (including
the heat exchanger) and a rapid and efficient charge process. Moreover, as suggested
previously, this study demonstrates that the design of an efficient ANG storage sys-
tem is closely connected to the heat- and mass-transfer properties of the adsorbent,
as well as to the real dynamic working conditions.
Figure 9. Continued.
892 V. Goetz and S. Biloe
Conclusion
Experimental studies on both the charge and the discharge of an ANG vessel filled
with two different ACBs were performed. The results demonstrate two major effects:
. First, for the charge process, the heat transfer by conduction within the ACB
mainly conditions the performance levels. Obviously, a high heat transfer property
Figure 10. Dimensioning of an ANGvessel: (a) standard design derived fromthe heat exchanger
technology and (b) more compact design.
Efficient Dynamic Charge and Discharge. 893
within the ACB must be combined with efficient heat transfer at the vessel wall.
The use of expanded natural graphite allows meeting these criteria. It was used
as a very highly conductive matrix and as a support for activated carbons. In
addition, it provides good thermal contact, which means a high value of the heat
exchange coefficient. For the external surface, the use of water as a heat external
fluid, flowing in a double casing, also provides a sufficient value for the heat
exchange coefficient. From a practical point of view, such heat exchange con-
ditions could easily be accomplished with the engine cooling circuit already
existing in vehicles.
. Second, for the discharge, the mass-transfer properties of the ACB, as well as the
heat-transfer properties, should be considered. Even if these conclusions have a
general sense, they are partially connected to the activated carbon tested as well
as the vessel geometry and vessel dimension.
The mathematical model, which was previously validated for the highly conduc-
tive ACB
k
in the case of discharge (Biloe et al., 2001b), was extended and validated
to the charge process and for another kind of ACB that presents high permeability
but low thermal conductivity (ACB
k
). Now, the main objective is to use the model as
a dimensioning tool and to develop an optimization procedure. A first dimensioning
example was given in the last part of the article, but it must be generalized. The opti-
mization procedure will be performed through the data acquisition of the heat- and
mass-transfer properties of the ACBs as a function of the implementation factors:
the activated carbon density and the weight ratio of expanded natural graphite in
the ACB. A complete optimization procedure should take into account the influence
of the microporous characteristics of the activated carbon, which defined the pos-
ition of the isosteric network in the Clausius-Clapeyron diagram and consequently
the impact of the heat- and mass-transfer limitations on the dynamic performance.
Finally, as we have tried to illustrate in this article, a complete optimization pro-
cedure should take into account the connection between the three different scales
involved in the design of an ANG storage vessel: microporous solid (adsorbent com-
posite block (ACB)), ANG vessel design, and process management.
Nomenclature
C
p
heat capacity, J kg
1
E
o
characteristic energy of adsorption, J mol
1
d mean diameter, m
h heat exchange coefficient, W m
2
K
1
k permeability, m
2
M molar mass, kg mol
1
n D.A. exponent
P pressure, Pa
P
sat
saturated vapor pressure, Pa
q adsorbed gas, kg kg
1
Q gas capacity, V=V
r radius, m
R perfect gas constant, J mol
1
K
1
S heat exchange surface, m
2
T temperature, K
894 V. Goetz and S. Biloe
T
o
room temperature, K
u
g
gas velocity, m s
1
V volume, m
3
w
ENG
expanded natural graphite mass ratio
W
o
total microporous volume, m
3
kg
1
Greek Letters
b affinity coefficient
Dh isosteric heat of adsorption, J mol
1
e
m
macroporosity
e
o
accessible porosity
g efficiency
k heat conductivity, W m
1
K
1
l viscosity, Pa s
m flow rate, L
NTP
min
1
q density, kg m
3
h vessel dynamic load level
Subscripts
a adsorbed gas
ac activated carbon
ACB activated carbon block
av available
C charge
dc dynamic charge
dd dynamic discharge
D discharge
f fluid
g gas
s solid phase
st solid tube
t tube
tf tube fluid
References
Alca~ nniz-Monge, J., De La Casa-Lillo, M. A., Cazorla-Am ooros, D., and Linares-Solano, A.
(1997). Carbon, 35, 291.
Baker, F. S. (1998). U.S. Patent No. 5,710,092.
Biloe, S., Goetz, V., and Mauran, S. (2001a). Carbon, 39, 1653.
Biloe, S., Goetz, V., and Mauran, S. (2001b). AIChE J., 47, 2819.
Biloe, S., Goetz, V., and Guillot, A. (2002). Carbon, 40, 1653.
Bojan, M. J., Van Slooten, R., and Steele, W. (1992). Sep. Sci. Technol., 27, 1837.
Bose, T. K., Chahine, R., and St-Arnaud, J. M. (1991). U.S. Patent No. 4,999,330.
Burchell, T. D. and Rogers, M. R. (2000). In Proceedings of the International Conference on
Carbon, Berlin, 133.
Carman, P. C. (1956). Flow of Gases through Porous Media, Butterworths, London.
Chang, K. J. and Talu, O. (1996). Appl. Therm. Eng., 16, 359.
Chen, X. S., McEnaney, B., Mays, B., Alca~ nniz-Monge, J., Cazorla-Am ooros, D., and Linares-
Solano, A. (1997). Carbon, 35, 1251.
Efficient Dynamic Charge and Discharge. 895
Cook, T. L., Komodromos, C., Quinn, D. F., and Ragan, S. (1999). Carbon Materials for
Advanced Technologies, ed. T. D. Burchell, Pergamon, Oxford.
Cracknell, R. F. and Gubbins, K. E. (1992). J. Mol. Liq., 54, 262.
Cracknell, R. F., Gordon, P., and Gubbins, K. E. (1993). J. Phys. Chem., 97, 494.
Dubinin, M. M. (1975). Progress in Surface and Membrane Science, D. A. Cadenhead,
New York.
Dubinin, M. M. and Astakhov, V. A. (1971). Adv. Chem. Ser., 102, 69.
Dullien, F. A. L. (1992). Porous Media: Fluid Transport and Pore Structure, Academic Press,
San Diego.
Greebank, M. (1992). U.S. Patent No. 5,094,736.
Inomata, K., Kanazawa, K., Urabe, Y., Honoso, H., and Araki, T. (2002). Carbon, 40, 87.
Jasionowski, W. J., Tiller, A. J., Fata, J. A., Arnold, J. M., Gauthier, S. W., and Shiraki, Y. A.
(1989). In Proceedings of the International Gas Research Conference, Tokyo, 1192.
MacDonald, J. A. F. and Quinn, D. F. (1996). Carbon, 34, 1103.
MacDonald, J. A. F. and Quinn, D. F. (1998). Fuel, 77, 61.
Manzi, S., Valladares, D., Marchese, J., and Zgrablich, G. (1997). Adsorp. Sci. Technol., 15,
301.
Matranga, K. R., Myers, A. L., and Glandt, E. D. (1992). Chem. Eng. Sci., 47, 1569.
McEnaney, B., Mays, B., and Chen, X. S. (1998). Fuel, 77, 557.
Menon, V. C. and Komarneni, S. (1998). J. Porous Mater., 5, 43.
Mota, J. P. B. (1995), Ph.D. diss., Institut National Polytechnique de Lorraine, Nancy,
France.
Mota, J. P. B., Rodrigues, A. E., Saatdjian, E., and Tondeur, D. (1997a). Carbon, 35, 1259.
Mota, J. P. B., Rodrigues, A. E., Saatdjian, E., and Tondeur, D. (1997b). Adsorption, 3, 117.
Nicholson, D. (1998). Carbon, 36, 1511.
Osawa, S., Kusumi, S., and Ogino, Y. (1976). J. Colloid Interface Sci., 56, 83.
Parkyns, N. D. and Quinn, D. F. (1995). Porosity in Carbons, ed. J. W. Patrick Arnold,
London.
Patankar, S. V. (1980). Numerical Heat Transfer and Fluid Flow, McGraw-Hill, New York.
Quirke, N. and Tennison, S. R. R. (1996). Carbon, 34, 1281.
Rubel, A. M. and Stencel, J. M. (2000). Fuel, 79, 1095.
Talu, O. (1992). In Fundamentals of Adsorption; Proceedings of the Fourth International Con-
ference on Fundamentals of Adsorption, Kyoto, May 1722, 1992, 655, Elsevier, New
York.
Wegrzyn, J. and Gurevich, M. (1996). Appl. Energy, 55, 71.
896 V. Goetz and S. Biloe

Você também pode gostar