A Transient Study of Double-Jacketed Membrane Reactor Via Methanol Steam Reforming

A transient study of double-jacketed membrane reactor via
methanol steam reforming

Chi-Hua Fu, Jeffrey C.S. Wu*
Department of Chemical Engineering, National Taiwan University, No. 1, Section 4, Roosevelt Road, Taipei, Taiwan 10617, ROC
a r t i c l e i n f o
Article history:
Received 20 August 2008
Received in revised form
2 October 2008
Accepted 2 October 2008
Published online -
Keywords:
Double-jacketed membrane reactor
Methanol steam reforming
Unsteady state
Transient
Simulation
a b s t r a c t
A non-isothermal unsteady-state model was established to simulate methanol steam
reforming using a double-jacketed Pd membrane reactor. At steady state, a self-sustained
membrane reactor was achieved by the oxidation of residual methanol and hydrogen from
reformer for endothermic steam reforming. The molar fractions of species and reformer
temperature were analyzed under co-current operation between oxidation and reformer
sides. The start-up of reformer was simulated under two conditions: (1) The catalyst
temperature was lower than the inuent temperature and (2) The catalyst temperature
was higher than inuent temperature. Condition 1 yielded higher methanol conversion
and reformer temperature than condition 2 at steady state. Moreover, the instability of
species can be minimized on condition 1 during start-up. The uctuation of membrane
reactor at steady state was also studied. Two strategies were compared to analyze the
reformer response when temporary extra hydrogen was required. The results showed that
increasing inlet methanol outperformed increasing reformer temperature.
2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights
reserved.
1. Introduction
Research in hydrogen generation has drawn much attention
in recent years [15]. Due to the difculties of storage and
transportation, the integration of hydrogen formation and
purication shows the competitive advantage over traditional
processes such as pressure swing adsorption or cryogenic
technology. Simultaneous hydrogen formation and purica-
tion can accelerate the reaction rate, leading to a higher
conversion or lower reaction temperature than that of tradi-
tional processes. Of all the studies of hydrogen production
processes, steam reforming of hydrogen-rich fuel combined
with the permeation of a supported membrane reactor shows
competitiveness in this eld [13,6,7]. However, the steam
reforming of fuel is highly endothermic, and therefore
external heat is needed to drive the reactions. Autothermal
reforming is to add a small amount of oxygen in the feedstock,
thus the necessary heat for the endothermic steam reforming
can be supplied by the exothermic oxidation of fuel [811]. A
mixture of fuel, steam, and oxygen (or air) is fed into the
reactor, where steamreforming and catalytic partial oxidation
take place in one system. The advantages of autothermal
reforming are simple reactor design, less heat transfer resis-
tances and fast start-up time. However, it is hard to adjust the
oxygen to fuel ratio to meet the thermal-neutral condition.
The partial oxidation of fuel yields high carbon monoxide
concentration, which is an undesirable product [12].
Using the external oxidation reactor is an alternate option
that can prevent several drawbacks to the autothermal
reactor, though external heat transfer causes heat resistance
problems. The additional catalytic combustor has higher costs
than that of an all-in-one system. However, separating the
catalytic oxidation and steam reforming to different systems
can simplify reactions, leading to more hydrogen permeation
* Corresponding author. Tel.: 886 223631994; fax: 886 223623040.
E-mail address: [email protected] (J.C.S. Wu).
Avai l abl e at www. sci encedi r ect . com
j our nal homepage: www. el sevi er . com/ l ocat e/ he
ARTICLE IN PRESS
0360-3199/$ see front matter 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2008.10.005
i nt e r na t i ona l j o ur na l o f hy d r o g e n e ne r gy x x x ( 2 0 0 8 ) 1 9
Please cite this article in press as: Fu C-H, Jeffrey C.S. Wu, A transient study of double-jacketed membrane reactor via methanol
steam reforming, International Journal of Hydrogen Energy (2008), doi:10.1016/j.ijhydene.2008.10.005
and lower carbon monoxide concentration. Using competitive
oxidative catalysts, such as Pt/Al
2
O
3
or Pd/Al
2
O
3
in the
external oxidation reactor, have higher heat generation ef-
ciency of fuel than that of using reforming catalysts. Han et al.
studied an on-site hydrogen generator with a steam reformer,
a hydrogen permeator and a catalytic combustor [13]. When
the system reaches steady state, methanol is almost con-
verted completely, and results in 65% of hydrogen. After the
hydrogen purication by a PdCu alloy membrane, hydrogen
recovery can reach 75%. The remaining 25% hydrogen can be
oxidized to supply the heat for steam reforming.
Previous researchers have done some detailed studies on
the hydrogen generation using membrane reactors by simu-
lation. Harold et al. used an one-dimensional isothermal
model of palladium supported membrane reactor to study
different kinds of hydrogen generation for fuel cell applica-
tions. Later, a comparison of conventional and autothermal
membrane reactors was studied. They concluded that
a smaller size autothermal membrane reactor can give the
same hydrogen generation as that of a conventional one
[1416]. Ji et al. used an one-dimensional non-isothermal
heterogeneous model to simulate the methane membrane
reactor. An optimal operation point is presented to balance
the equipment investment, hydrogen production rate and
overall energy efciency [17]. Chans group used a two-
dimensional non-isothermal unsteady-state model to simu-
late the autothermal reforming of methane and the results
were validated with experimental data. The prediction
showed that the reformer could reach steady state within
5 min after the preheating treatment of reformer. Further
studies found the optimal air to methane, steam to methane
ratio, and the space velocity [1820].
So far, although reforming in membrane reactors has been
studied extensively, most of the studies focused on nding the
Nomenclature
A
c
cross-section area of reformer, m
2
A
m
permeation area per unit volume, m
2
/m
3
A
o
heat transfer area between reformer and catalytic
combustor, m
2
C concentration, mol/m
3
Cp average heat capacity, J/mK
De
i
effective gas diffusivity, m
2
/s
De
i,p
effective gas diffusivity inside the catalyst, m
2
/s
d
p
catalyst particle diameter, m
d
m
membrane thickness, m
d
o
wall thickness of outer jacket, m
d
s
thickness of Pd membrane support, m
G
0
mass ow ux, kg/m
2
s
h
r
effective heat transfer coefcient in reformer,
W/m
2
K
h
o
effective heat transfer coefcient in catalytic
combustor, W/m
2
K
h
p
effective heat transfer coefcient in permeator,
W/m
2
K
j
H
ChiltonColburn factor
J
H2
mass ux of hydrogen from reformer to
permeator, mol/(m
2
s)
k
r
catalyst heat conductivity, W/mK
k
m
heat conductivity of membrane, W/mK
k
p
gas heat conductivity in permeator, W/mK
k
o
heat conductivity of outer jacket, W/mK
k
s
heat conductivity of Pd membrane support, W/mK
L reactor length, m
M molecular weight, g/mol
P
i
pressure, atm
P
i
partial pressure of i species, Pa
P
H2
hydrogen partial pressure, atm
Pr Prandtl number
R ideal gas constant, atmL/(mol K)
Re Reynolds number
R
W/M
molar ratio between water and methanol
r catalyst particle radius, m
r
i
reaction rate of i species, mol/kg cat s
r
s
i
reaction rate of i species on catalyst surface, mol/
kg cat s
r
j
reaction rate of j species, mol/kg cat s
r
p
permeator radius, m
S
a
heat transfer area per volume of catalyst, m
2
/m
3
T temperature, K
t time, s
U
1
overall heat transfer coefcient between reformer
and catalytic combustor, W/m
2
K
U
2
overall heat transfer coefcient between reformer
and permeator, W/m
2
K
u supercial velocity, m
V volume, m
3
WHSV weight hourly space velocity, h
1
x molar fraction (Eq. (18))
X methanol conversion (Eq. (3))
z axial coordinate of reactor, m
DH reaction enthalpy, J/mol
Greek letters
b permeation coefcient of Pd membrane, mol/
ms atm
1/2
3 void fraction
h effectiveness factor
r density, kg/m
3
s tortuosity factor
m viscosity, kg/ms
y molar volume, m
3
/mol
Subscripts
0 initial
g gas
i component number in reformer
j component number in catalytic combustor
H
2
hydrogen
methanol methanol
o oxidation
p permeation
r reforming
s catalyst
i nt e r na t i o na l j o ur na l o f hyd r o g e n e ne r g y x x x ( 2 0 0 8 ) 1 9 2
ARTICLE IN PRESS
optimal operation at steady-state by adjusting the inlet
compositions, reformer pressure, temperature, and reactor
design. On the other hand, only a few researchers studied the
autothermal reforming process under unsteady-state condi-
tions [1921]. The research work on membrane reformers with
external heat supplement under unsteady state is still unclear.
In our previous study, we used a 1-D non-isothermal model to
simulate the hydrogen production using a self-sustained
double-jacketed membrane reactor under steady state [22].
However, during the transient period, species variations and
reformer temperature uctuations are difcult to measure by
experimental study, and thus the optimumoperating strategy
is hard to acquire. It is important to understand the reformer
response during the transient period. The optimal control as
well as the corresponding strategy should be studied thor-
oughly when species and/or temperature variations occur at
the start-up. Through a simulation study, the results allow us
to choose an optimal start-up strategy and can be used as a
guideline for experiments. When double-jacketed membrane
reactors are running under unsteady state, the responses of
mass and temperature variations in the permeator, reformer
and combustor should be considered simultaneously. The
main concern is how fast the reactor can reach steady state
from start-up. Howlong it would take to return to steady state
when a mass or temperature disturbance occurs.
The objective of this study is to simulate the non-
isothermal unsteady state of a double-jacketed membrane
reactor in methanol steam reforming. In particular, during
start-up, the consumption and production of species in the
steam reformer and the temperatures of gases and catalysts
are investigated under two conditions: (1) the temperature of
inlet feed is higher than that of catalyst; (2) the temperature of
catalyst bed is higher than that of inlet feed. A special situa-
tion is also considered when a temporary extra hydrogen
demand occurs under steady-state operation.
2. Kinetics and mathematical model
2.1. Reaction kinetics
Methanol is one of the hydrogen-rich fuels used in membrane
reactors. Using methanol as the reforming gas has been
studied extensively in recent years [911,14,16,23,24]. Meth-
anol is in liquid state at room temperature and atmospheric
pressure, and can be easily stored and transported. Methanol
has high hydrogen to carbon ratio and the absence of carbon
to carbon bonds can produce high hydrogen yield and low
carbon dioxide. Methanol steam reforming can be operated
under lower temperature than that of other hydrocarbons
such as methane or gasoline. The overall reaction of methanol
steam reforming can be expressed in Eq. (1)
CH
3
OHH
2
O !
CuO=ZnO=Al
2
O
3
CO
2
3H
2
DH
298K
49 kJ=mol (1)
A comprehensive kinetic model was developed by Peppley
et al. [23,24] over a commercial CuO/ZnO/Al
2
O
3
catalyst. The
results showed that methanol decomposition is the slowest,
as compared to the steam reforming and watergas shift,
because it takes place on different active sites. Therefore
carbon monoxide is one of the byproducts and is depended on
the reformer operating conditions. Besides steam reforming,
methanol decomposition and water gas shift are also incor-
porated in this model. The details of the kinetic model are
shown in Appendix.
The reaction expression of methanol catalytic oxidation on
Pt/g-Al
2
O
3
is depicted in Eq. (2), with a power law kinetic
equation shown in Eq. (3) [25].
CH
3
OH1:5O
2 !
Pt=gAl
2
O
3
CO
2
2H
2
O DH
298K
676kJ=mol (2)
r
o
4:69110
9
e
48;500=RT
C
1:4
methanol
X<12%
3:15310
6
e
18;100=RT
C
1:4
methanol
X>12%
(3)
There are few researches which studied the reaction
kinetics of hydrogen catalytic oxidation on Pt/g-Al
2
O
3
. We
assume that the reaction rate of hydrogen catalytic oxidation
is the same as methanol catalytic oxidation. The variation of
oxidation rate wont appreciably affect the simulation
because oxidation is much faster than reforming. The CO
oxidation is neglected because the CO amount is much fewer
than methanol and hydrogen.
2.2. Mathematical model
Methanol steamreforming ina membrane reactor is simulated
using an one-dimensional non-isothermal unsteady-state
model. Fig. 1 shows the schematic diagram of a double-jack-
eted membrane reactor. Fromthe outer layer to the inner layer
are the catalytic combustor, reformer and permeator, respec-
tively. The catalytic combustor is lled with oxidative catalysts
such as Pt/Al
2
O
3
or Pd/Al
2
O
3
. The reformer is lled with
reforming catalyst CuO/ZnO/Al
2
O
3
. A palladium membrane is
coated on the surface of the permeator due to the 100%
hydrogen permselectivity of palladium. Therefore pure
hydrogen can be obtained from the permeator. The mass and
energy balances are evaluated simultaneously in three parts.
Some assumptions are made to simplify calculations. The
concentration and temperature gradient in the radial axis is
neglected based on plug ow reactor assumption. The mass
and thermal diffusion is negligible, when compared with the
convectionand conductionof mass and enthalpy. The physical
and chemical properties of gases and catalysts are evaluated
based onthe average values of theoretical or experimental data
which are shown in Table 1. The governing equations and
boundary conditions are listed below, with their symbols and
denitions shown in nomenclature.
Mass balances
1. Reformer
3
vC
i
vt
u
r
vC
i
vz
h
i
r
s;r
r
i
; i CH
3
OH; H
2
O; CO
2
; CO (4)
3
vC
H
2
vt
u
r
vC
H
2
vz
h
H
2
r
s;r
_
r
H
2
_
b
d
m
A
m
_

P
H
2
;r
_

P
H
2
;p
_ _
(5)
2. Permeator
vC
H
2
vt
u
p
vC
H
2
vz

b
d
m
A
m
_

P
H
2
;r
_

P
H
2
;p
_ _
(6)
i nt e r na t i ona l j o ur na l o f hy d r o g e n e ne r gy x x x ( 2 0 0 8 ) 1 9 3
ARTICLE IN PRESS
3. Oxidation side
3
vC
j
vt
u
o
vC
j
vz
h
j
r
s;o
_
r
j
_
; j CH
3
OH; O
2
; H
2
; H
2
O; CO
2
(7)
Energy balance
1. Reformer
r
g;r
Cp
g;r
_
3
vT
g;r
vt
u
r
vT
g;r
vz
_
U
1
A
o
_
T
o
T
g;r
_
U
2
A
m
_
T
g;r
T
p
_
S
a
h
r
_
T
g;r
T
s;r
_
(8)
r
s;r
Cp
s;r
vT
s;r
vt
k
r
v
2
T
s;r
vz
2
h
i
r
s;r
r
i
DH
i
S
a
h
r
_
T
g;r
T
s;r
_
(9)
2. Permeator
r
H
2
Cp
H
2
_
vT
p
vt
u
p
vT
p
vz
_
_
J
H
2
Cp
H
2
U
2
A
m
_
_
T
g;r
T
p
_
(10)
3 . Oxidation side
r
g;o
Cp
g;o
_
3
vT
o
vt
u
o
vT
o
vz
_
h
j
r
s;o
_
r
j
_
DH
j
U
1
A
o
_
T
o
T
g;r
_
(11)
2.3. Boundary conditions
z 0; C
i
C
i;0
; C
j
C
j;0
; C
H
2
0 permeator
T
g;r
T
g;r;0
; k
r
vTs;r
vz
h
r
A
c
_
T
g;r;0
T
s;r;0
_
; T
p
T
p;0
; T
o
T
o;0
z 1;
vTs;r
vz
0
(12)
t 0
; C
i
C
i;0
; C
j
C
j;0
; C
H
2
0 permeator
T
g;r
T
g;r;0
; T
s;r
T
s;r;0
; T
p
T
p;0
; T
o
T
o;0
(13)
The hydrogen permeability through Pd membrane is taken
from Uemiya et al. [1]:
b 1:567 10
5
e
8410=RT
(14)
The effectiveness factor, h
i
, is a function of temperature, the
effective diffusivity and the species partial pressure in cata-
lyst. A continuity equation is used to calculate h
i
.
De
i;p
_
d
2
P
i
dr
2

1
r
dP
i
dr
_
r
s
r
i
(15)
h
i

_
V
0
r
s
r
i
dV
0
V
r
s
r
s
i
(16)
where the effective diffusivities of species inside the catalyst
are calculated by [26].
De
i;p

3
s
De
i
(17)
where 3 is the void fraction and s read as the tortuosity factor.
The effective diffusivities De
i
are given by [27]
De
i

1
1 x
i
n
j1;jsi
De
i;j
x
j
(18)
where the binary diffusivities are estimated by FullerSchet-
tlerGiddings correlation [28].
De
i;j

0:00101 T
1:75
M
1=2
i;j
P
_
y
1=3
i

y
1=3
j
_
2
(19)
U
1
and U
2
are the overall heat transfer coefcients of reformer/
combustor and permeator/reformer side, respectively. They
are given by the following expressions
U
1

_
1
h
r
d
o
k
o
1
h
o
_
1
(20)
Fig. 1 Schematic diagram of a double-jacketed Pd membrane reactor.
Table 1 Parameters of reactor used in the simulation
Permeator radius r
p
(m) 6.35 10
3
Reformer radius r
r
(m) 1.27 10
2
Combustor radius r
o
(m) 1.91 10
2
Reformer length L (m) 1.0 10
1
Membrane thickness d (m) 1.0 10
5
Catalyst density r
s
(kg m
3
) 1300
Surface area
of catalyst S
a
(m
2
kg
1
)
56,000
Catalyst particle
diameter d
p
(m)
2.0 10
3
Hydrogen permeability
over Pd membrane b (mol m
1
s
1
atm
0.5
)
1.567 10
5
e
1011.5/T
ARTICLE IN PRESS
U
2

_
1
h
p
d
m
k
m
d
s
k
s
1
h
r
_
1
(21)
The effective heat transfer coefcients, h
r
or h
o
, are evalu-
ated by ChiltonColburn factor j
H
[27].
The ChiltonColburn factor of gas through packed beds can
be described as follows.
j
H
2:19Re
2=3
0:78Re
0:381
h
r
CpG
0
Pr
2=3
(22)
The gas heat transfer coefcient in permeator is calculated
by the Colburn correlation [28].
h
p
0:023
k
p
r
p
Re
0:8
Pr
1=3
(23)
To simplify the physical interpretation, some dimension-
less groups are used to generalize and analyze the results. The
dimensionless groups are presented in Eq. (24)
z

z
L
; t

t
L=u
0
; C
i

C
i
C
methanol;0
; T

T
T
0
(24)
Using dimensionless groups is helpful for scaling-up. For
instance, dimensionless time is the real time divided by space
time of reactants in the reactor. Therefore, a large reactor with
long space time has the same dimensionless time as a small
reactor with short space time.
The species concentration and temperature are solved by
nite difference method with 100 50 matrix using the
Mathematica.
3. Results and discussions
3.1. Variation of temperatures and concentrations
during start-up period
Before the reactor starts up, the catalyst bed and inlet feed
should be preheated to the working temperature. However, it
takes a long time and a lot of enthalpy to heat up the reformer
and inlet feed. One may think that it is possible to heat up the
reformer or inlet gases toatemperature lower thantheworking
temperature to save preheating time or cost, and thus an
interesting considerationis proposed. Atransient simulation is
carried out from the start of inlet feed till the reformer reaches
steady state. The main concern is how fast the reformer can
fully reach steady state, therefore, the prediction is estimated
based on two conditions, (1) the hot inlet feed is set to be 543 K,
and the cold catalyst bed is 433 K, which is the minimum
temperature of the kinetic data proposed by Peppley et al.
[23,24], (2) the cold inlet feed is set to be 433 K, and the hot
catalyst bed is 543 K. The reason for choosing 433543 K is
because methanol steam reforming is usually operated in this
region. In addition, we expect a fast start-up time or less pre-
heating costs when either gas stream or catalyst starts at low
temperature. Theinitial temperatureof thecatalyticcombustor
is set to be 543 K, that is, methanol mixedwithair at 543 Kis fed
into the catalytic combustor with the oxidative catalyst Pt/g-
Al
2
O
3
of 543 K. Therefore, steam reforming can take place
simultaneously when the inlet feed is fed into the reformer
because oxidation provides the extra energy supplement.
For a self-sustained double-jacketed membrane reactor,
the systemcan be run without additional fuel feeding because
the un-converted methanol and un-permeated hydrogen is
oxidized to provide the endothermic reforming. However, an
extra methanol/air ow is needed in the oxidation side
because no hydrogen is produced in the beginning when the
reformer is starting up. The extra methanol feeding can be
stopped after the reformer reaches steady state, and then the
double-jacketed reactor is operated under self-sustained
condition. During the start-up period, the ratio of methanol
mass ow rate between oxidation side and reformer is set to
be 0.1. The reason of choosing 0.1 is because the enthalpy
generated by methanol oxidation in oxidation side can meet
the requirement of methanol steam reforming according to
the theoretical energy balance estimation. An assumption of
excess oxygen is proposed to ensure the total oxidation of
methanol, thus the ratio of mass owrate of air and methanol
in oxidation side is set to be a little larger than 5.
Fig. 2 illustrates the temperature distributions as a function
of reformer length at different times under condition (1)
operation. The reformer starts when the catalyst bed is cold,
the hot streamand external fuel combustion is used to heat up
the catalyst bed, indicating the increasing temperature prole
near the entrance of reformer. Then the steam reforming
reaction adsorbs heat, cools down the temperature in the rear
part of reformer. Furthermore, it can also be found that during
the transient period, the temperature difference between
gases and catalyst diminishes with reformer position, and
maintains a difference of approximately 5 K when the
reformer reaches steady state. This means that the heat
transfer is from gases to catalyst bed, where the endothermic
reforming reaction cools down the temperature on the cata-
lyst surface.
Fig. 3 shows the molar fractions of species at the exit of
reformer as a function of dimensionless time under condition
(1) operation. The molar fractions of methanol and steam
decreases monotonically with time during the transient
period, leading to an increase in hydrogen, carbon dioxide and
very little carbon monoxide production. The molar fractions
of each species reach steady state when the dimensionless
Dimensionless reactor length
0.0 0.2 0.4 0.6 0.8 1.0
T
e
m
p
e
r
a
t
u
r
e
(
K
)
400
420
440
460
480
500
520
540
8
15
75
150
solid line : gases;
dotted line : catalyst
dimenionless time 3
Fig. 2 Gases and catalysts temperature proles as
a function of dimensionless length under condition 1 (hot
stream (543 K) is fed in to cold reformer (433 K)). WHSV of
methanol is 20 h
L1
, R
W/M
is 1.2, reformer pressure is 5 bar.
ARTICLE IN PRESS
time is about 150. This indicates that, inthe beginning, most of
the enthalpy carried by the hot stream is consumed to raise
the temperature of catalyst bed instead of supplying to the
reforming reaction, resulting in a low conversion of methanol
and steam. When the temperature of the catalyst bed is high
enough, the reforming reaction takes over. The catalyst bed is
continuously heated by the hot stream and external catalytic
oxidation of methanol, but it is also simultaneously cooled
down by the endothermic reaction. An increasing tendency of
the temperature in Fig. 2 indicates the increasing consump-
tion rate of methanol and steam because the reforming rate is
proportional to the catalyst surface temperature. In addition,
the theoretical hydrogen production is three times higher
than the production of carbon dioxide; however, the molar
fraction of hydrogen in the reformer is less than that of carbon
dioxide because part of the hydrogen permeates into the
permeator. Nearly 0.1%of carbon monoxide is observed under
condition (1) operation because the temperature of catalyst
bed is always lower than 510 K, favoring water gas shifting but
not methanol decomposition.
Fig. 4 depicts the temperature distributions as a function of
reformer length in time sequence under condition (2) opera-
tion. The reformer is preheated to the working temperature
(543 K), and a cold stream (433 K) is fed into the hot reformer.
The results show a signicantly increasing trend of gaseous
temperature proles at the entrance of reformer, and then
maintaina slightly increasing trendintherear part of reformer.
During the transient period, the temperature curves of catalyst
bed keep monotonically increasing in the reformer. Then the
catalyst temperature gradually decreases when it reaches
steady state. This indicates that, initially, the cold stream is
heated by the hot catalyst bed and external combustor. Then
the steam reforming absorbs enthalpy from the catalyst
surface, indicating the decreasing trend of the catalyst bed
temperature as time increases. During the transient period, the
reformer temperature increases slightly in the rear part of
reformer because methanol is almost completely converted,
and the heat from the catalytic combustor is provided to heat
up the reformer temperature. Whenit reaches steady state, the
catalyst temperature is lower than that in the beginning
because the cold gas streamand endothermic reforming keeps
absorbing heat from catalyst. The slightly decreasing trend in
the rear part of reformer temperature under steady state
(dimensionless time 150) is because the heat consumed by
reforming is slightly higher than that supplied by external heat
transfer.
Fig. 5 shows the molar fractions of species at the exit of
reformer as a function of dimensionless time under condition
(2) operation. Initially, methanol and steam is consumed
rapidly when the feedstock is fed in, accompanied with large
production of hydrogen, carbon dioxide, and carbon
monoxide. The methanol and steamconversion decrease with
time after 20 dimensionless time, and reach steady state after
about 100 dimensionless time. This means that, initially, the
reforming reaction occurs simultaneously when the cold
Dimensionless reactor length
0.0 0.2 0.4 0.6 0.8 1.0
T
e
m
p
e
r
a
t
u
r
e
(
K
)
400
450
500
550
600
8
75
15
3
solid line : gases
dotted line : catalyst
150 dimensionless time
Fig. 4 Gases and catalysts temperature proles as
a function of dimensionless length under condition 2 (cold
stream (433 K) is fed in to hot reformer (543 K)). WHSV of
methanol is 20 h
L1
, R
W/M
Dimensionless time
0 50 100 150 200
M
o
l
a
r

f
r
a
c
t
i
o
n
0.0
0.1
0.2
0.3
0.4
0.5
0.6
CH
3
OH
H
2
O
H
2
CO
2
CO
Fig. 5 The molar fractions of species in reformer as
a function of dimensionless time under condition 2(cold
stream (433 K) is fed in to hot reformer(543 K)). WHSV of
methanol is 20 h
L1
, R
W/M
Dimensionless time
0 50 100 150 200 250 300
M
o
l
a
r

f
r
a
c
t
i
o
n
0.0
0.2
0.4
0.6
0.8
1.0
CH
3
OH
H
2
O
H
2
CO
2
CO
Fig. 3 The molar fractions of species in reformer as
a function of dimensionless time under condition 1 (hot
stream (543 K) is fed in to cold reformer (433 K)). WHSV of
methanol is 20 h
L1
, R
W/M
ARTICLE IN PRESS
stream ows to the hot catalyst surface, leading to the high
methanol conversion. Then the cold inlet feed keeps cooling
down the catalyst bed, resulting in a lower reforming rate and
methanol conversion, as compared with those in the begin-
ning. In addition, endothermic reforming also uptakes part of
thermal energy from the catalyst. When it reaches steady
state, the methanol conversion under condition (2) operation
is lower than that under condition (1) operation because the
temperature on the catalyst surface is lower than that under
condition (1) operation.
The maximummolar fractionof hydrogenhas atime lagover
other species. This indicates that the maximumconsumptionof
methanol results in the maximum production of hydrogen;
meanwhile, the high hydrogen concentration in the reformer
drives part of the hydrogenpermeating into the permeator, thus
competition between hydrogen production and permeation
causes the lag of maximum hydrogen molar fraction. The
production rate of carbon dioxide is proportional to the
consumption rate of methanol, and it is found that carbon
dioxideyieldsmorethan30%intheefuentsif apart of hydrogen
permeates into the permeator. In addition, nearly 5% of carbon
monoxide is observedinthe rst 20 dimensionless time because
thehot catalyst surfacedecomposes apart of methanol resulting
in a high concentration of undesired carbon monoxide.
A summary can be made fromthe results above. When it is
operated under condition (1), the reformer has low methanol
conversion and temperature of reformer in the beginning,
while high methanol conversion and temperature of reformer
is exhibited when it reaches steady state. Meanwhile, condi-
tion (1) operation also ensures a low concentration variation
due to its low reforming rate during the transient period.
Therefore, choosing condition (1) operation can result in more
steady species concentrations than under condition (2) oper-
ation. On the contrary, operating under condition (2) results in
high methanol conversion and reformer temperature in the
beginning, but methanol conversion and temperature of
reformer is not high when it reaches steady state. In addition,
operating under condition (2) can have a short start-up time as
compared to that under condition (1) operation. Therefore we
can conclude that the catalyst should be preheated to a high
temperature if a short start-up time and fast hydrogen yield is
desired. However, it causes a uctuation of species concen-
trations and low methanol conversion when it reaches steady
state. Moreover, more CO is formed during the rst 20
dimensionless time. On the other hand, the gas streamshould
be preheated to a high temperature if high methanol conver-
sion and hydrogen yield is required at steady state.
3.2. Fluctuation during steady-state operation
When the reformer is operated under steady state, one might
need to adjust the settings if some special requirements were
made. For instance, the methanol or steam supplement is
suddenly changed, an unusual reformer temperature drop
occurs, or an extra hydrogen demand is required. Chan et al.
used a 2-D heterogeneous model to simulate the transient
performance of autothermal methane steam reforming [29].
Three step-up and step-down reactor uctuations were
analyzed. The overshoot/undershoot of species and temper-
ature is also discussed. However the transient study of
membrane reactors is not be explored. In this study, we
thought of an extra hydrogen demand as a simulation
example and analyzed the results via methanol conversion,
temperature variation, and relative hydrogen production
uctuation. The extra hydrogen demand is set to be 1.5 times
the ordinary production. Two scenarios are considered, rst,
the methanol feed rate in the reformer is increased to meet
this demand. The extra hydrogen demand lasts for 300
dimensionless time. Fig. 6 shows the responses of the meth-
anol conversion, reformer temperature, and relative hydrogen
production as a function of dimensionless time. The response
time of methanol conversion is about 40 dimensionless time,
while the response time of the reformer temperature is larger
than that of methanol conversion. The time lag of reformer
temperature from methanol conversion is because when
methanol conversion decreases as the mass ow rate
increases, it takes time to transfer heat from the oxidation
side to the reformer, leading to the time lag of temperature.
Fig. 6 also shows that an offset of relative hydrogen
production occurs in the beginning of the pulse. Relative
hydrogen production is the ratio of transient-period rate to
ordinary rate of hydrogen. Although increasing methanol and
steam mass ow rate allows for extra hydrogen production,
such operation sacrices reforming efciency by lowering
methanol conversion and reforming temperature. In other
words, more than1.5 times of methanol is required to meet the
extra hydrogen demand. The original reactor is optimized for
maximummethanol conversionandhydrogenyield. Therefore,
if long-termextrahydrogendemandisrequired, thereactor size
should be enlarged instead of increasing methanol feed.
Another way to increase the temporary hydrogen produc-
tion is to raise the reformer temperature by feeding extra
methanol to the oxidation side. Increasing fuels in the oxida-
tion side can increase the temperature of the oxidation side,
leading to a higher thermal driving force fromoxidationside to
reformer. Fig. 7 illustrates the responses of methanol conver-
sion, reformer temperature, and relative hydrogen production
against dimensionless time. The extra methanol required in
the oxidation side is estimated to increase 15% of methanol
consumption. The methanol conversion of the reformer can
only increase slightly because prior steady-state conversion is
already high. This suggests that most of the extra enthalpy
Dimensionless time
0 200 400 600 800
M
e
t
h
a
n
o
l

c
o
n
v
e
r
s
i
o
n
(
%
)
40
50
60
70
80
90
100
R
e
f
o
r
m
e
r

t
e
m
p
e
r
a
t
u
r
e
(
K
)
470
475
480
485
490
R
e
l
a
t
i
v
e

H
2

p
r
o
d
u
c
t
i
o
n
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Methanol conversion
Reformer temperature
Relative H
2
production
Fig. 6 The responses of methanol conversion, reformer
temperature, and relative hydrogen production as
a function of dimensionless time with an extra hydrogen
demand.
ARTICLE IN PRESS
from the oxidation side is used to raise the reformer temper-
ature, which is also observed in Fig. 7. The response time of
methanol conversion is almost the same as reformer
temperature because methanol is completely consumedwhen
reformer temperature increases. In addition, only near 30%
increase of hydrogen production is observed in Fig. 7 because
near all of the hydrogen has permeated into the permeator.
This suggests a maximum extra hydrogen production with an
increasing reformer temperature. However, increasing
reformer temperature is not a good idea because such opera-
tion needs extra methanol for heating fromthe oxidation side,
and cannot meet 1.5 times hydrogen demand. Furthermore,
high temperature favors CO production.
In summary, increasing the inlet mass ow rate is more
appropriate than increasing reformer temperature for
a temporary extra hydrogen demand. But, neither one ts the
high efciency of the reformer for long-term usage.
4. Conclusion
Asimulation of an unsteady-state double-jacketed membrane
reactor is presented. The start-up of a double-jacketed
membrane reactor is evaluated under two conditions. A short
start-up time is found in either condition. We recommend
that initial temperature of catalyst bed is higher than inlet gas
because of high methanol conversion and temperature at
steady state. In addition, the instability of species can be
minimized because of low reaction rate during the transient
period. When an extra hydrogen demand is required,
increasing the inlet methanol can meet the requirement while
sacricing with a low methanol conversion. Alternatively,
extra hydrogen can also be produced by increasing the
reformer temperature from additional methanol oxidation.
However such operation can only provide limited extra
hydrogen since the prior steady-state conversion is already at
the optimal level. Both adjustments are only suitable for
a short-term extra hydrogen demand.
Acknowledgment
Financial support by the Ministry of Economic Affairs, Taiwan,
under grant 95-EC-17-A-09-S1-019 is gratefully acknowledged.
Appendix.
Kinetics of methanol steam reforming over CuO/ZnO/Al
2
O
3
Methanol steam reforming
CH
3
OHH
2
O4CO
2
3H
2
; DH
298 K
49:4 kJ mol
1
(A.1)
Methanol decomposition
CH
3
OH4CO2H
2
; DH
298 K
90:5 kJ mol
1
(A.3)
r
D

k
D
K
CH
3
O2
_
P
m
=P
0:5
h
_ _
1 P
2
h
P
co
=k
D
P
m
_
C
T
S2
C
T
S2a
S
a
_
1 K
CH
3
O2
_
P
m
=P
0:5
h
_
K
OH2
_
P
w
=P
0:5
h
_
_ _
1 K
0:5
H2a
P
0:5
h
_
(A.4)
Water gas shift reaction
COH
2
O4CO
2
H
2
; DH
298 K
41:1 kJ mol
1
(A.5)
Dimensionless time
0 200 400 600 800
M
e
t
h
a
n
o
l

c
o
n
v
e
r
s
i
o
n
(
%
)
90
92
94
96
98
100
R
e
f
o
r
m
e
r

t
e
m
p
e
r
a
t
u
r
e
(
K
)
460
470
480
490
500
R
e
l
a
t
i
v
e

H
2

p
r
o
d
u
c
t
i
o
n
0.6
0.8
1.0
1.2
1.4
Methanol conversion
Reformer temperature
Relative H
2
production
Fig. 7 The responses of methanol conversion, reformer
temperature, and relative hydrogen production as
a function of dimensionless time with an extra hydrogen
demand.
r
R

k
R
K
CH
3
O1
_
P
m
=P
0:5
h
_ _
1 P
3
h
P
c
=k
R
P
m
P
w
_
C
T
S1
C
T
S1a
S
a
_
1 K
CH
3
O1
_
P
m
=P
0:5
h
_
K
HCOO1
P
c
P
0:5
h
K
OH1
_
P
w
=P
0:5
h
_
_ _
1 K
0:5
H1a
P
0:5
h
_ (A.2)
Table A.1 The parameters relate to the kinetics of
methanol steam reforming, methanol decomposition,
and water gas shift reaction.
k
R
(m
2
mol
1
s
1
) 7.4 10
14
e
102,800/RT
K
CH3O1
(bar
0.5
) 6.55 10
3
e
20,000/RT
K
HCOO1
(bar
1.5
) 2.3 10
9
e
100,000/RT
K
OH1a
(bar
0.5
) 4.74 10
3
e
20,000/RT
K
H(1a)
(bar
0.5
) 5.43 10
6
e
50,000/RT
k
D
(m
2
mol
1
s
1
) 3.8 10
20
e
170,000/RT
K
CH3O1
(bar
0.5
) 3.69 10
1
e
20,000/RT
K
OH2
(bar
0.5
) 3.69 10
1
e
20,000/RT
K
H(2a)
(bar
0.5
) 3.86 10
3
e
50,000/RT
k
W
(m
2
mol
1
s
1
) 5.9 10
13
e
87,600/RT
C
T
S1
C
T
S2
m
2
mol 7.5 10
6
C
T
S1a
C
T
S2a
m
2
mol 1.5 10
5
ARTICLE IN PRESS
r
W

k
W
K
OH1
_
P
co
P
w
=P
0:5
h
_
1 P
h
P
c
=k
W
P
co
P
w
_
C
T
S1
_
2
S
a
_
1 K
CH
3
O1
P
m
=P
0:5
h
K
HCOO1
P
c
P
0:5
h
K
OH1
P
w
=P
0:5
h
_
2
(A.6)
where P
m
is the partial pressure of methanol, P
w
is the partial
pressure of water, P
h
is the partial pressure of hydrogen, P
c
is
the partial pressure of carbon dioxide, and P
co
is the partial
pressure of carbon monoxide. The corresponding parameters
are listed in Table A.1. [16,23,24]
r e f e r e n c e s
[1] Uemiya S, Sato N, Ando H, Matsuda T, Kikuchi E. Steam
reforming of methane in a hydrogen-permeable membrane
reactor. Appl Catal 1991;67(2):22330.
[2] Collins JP, Way JD. Preparation and characterization of
a composite palladium-ceramic membrane. Ind Eng Chem
Res 1993;32(12):300613.
[3] Lin YM, Lee GL, Rei MH. An integrated purication and
production of hydrogen with a palladium membrane-
catalytic reactor. Catal Today 1998;44(14):3439.
[4] Armor JN. The multiple roles for catalysis in the production
of H-2. Appl Catal A 1999;176(2):15976.
[5] Breen JP, Ross JRH. Methanol reforming for fuel-cell
applications: development of zirconia-containing CuZnAl
catalysts. Catal Today 1999;51(34):52133.
[6] Itoh N. A membrane reactor using palladium. AICHE J 1987;
33(9):15768.
[7] Itoh N, Govind R. Combined oxidation and dehydrogenation
in a palladium membrane reactor. Ind Eng Chem Res 1989;
28(10):15547.
[8] Jenkins JW, Shutt E. Hydrogen generation using a novel
concept. Platinum Metals Rev 1989;33:11827.
[9] Hohlein B, Boe M, Bogild-Hansen J, Brockerhoff P, Colsman G,
Emonts B, et al. Hydrogen from methanol for fuel cells in
mobile system: development of a compact reformer. J Power
Sources 1996;61(12):1437.
[10] Geissler K, Newson E, Vogel F, Truong TB, Hottinger P,
Wokaun A. Autothermal methanol reforming for hydrogen
production in fuel cell applications. Phys Chem Chem Phys
2001;3(3):28993.
[11] Mizsey P, Newson E, Truong TB, Hottinger P. The kinetics of
methanol decomposition: a part of autothermal partial
oxidation to produce hydrogen for fuel cells. Appl Catal A
2001;213(2):2337.
[12] De Goote AM, Froment GF. Simulation of the catalytic partial
oxidation of methane to synthesis gas. Appl Catal A 1996;
138(2):24564.
[13] Han J, Kim IS, Choi KS. High purity hydrogen generator for
on-site hydrogen production. Int J Hydrogen Energy 2002;
27(10):10437.
[14] Harold MP, Nair B, Kolios G. Hydrogen generation in a Pd
membrane fuel processor: assessment of methanol-based
reaction systems. Chem Eng Sci 2003;58(12):255171.
[15] Lattner JR, Harold MP. Comparison of conventional and
membrane reactor fuel processors for hydrocarbon-based
PEM fuel cell system. Int J Hydrogen Energy 2004;29(4):
393417.
[16] Lattner JR, Harold MP. Comparison of methanol-based fuel
processors for PEM fuel cell systems. Appl Catal B 2005;
56(12):14969.
[17] Ji P, van der Kooi HJ, Arons JD. Simulation and
thermodynamic analysis of an integrated process with H-2
membrane CPO reactor for pure H-2 production. Chem Eng
Sci 2003;58(17):390111.
[18] Hoang DL, Chan SH. Modeling of a catalytic autothermal
methane reformer for fuel cell applications. Appl Catal A
2004;268(12):20716.
[19] Chan SH, Wang HM. Thermodynamic and kinetic modeling
of an autothermal methanol reformer. J Power Sources 2004;
126(12):815.
[20] Hoang DL, Chan SH, Ding OL. Kinetic and modeling study of
methane steam reforming over sulde nickel catalyst on
a gamma alumina support. Chem Eng J 2005;112(13):111.
[21] Papadias D, Lee SHD, Chmielewski DJ. Autothermal
reforming of gasoline for fuel cell applications: a transient
reactor model. Ind Eng Chem Res 2006;45(17):584158.
[22] Fu CH, Wu JCS. Mathematical simulation of hydrogen
production via methanol steam reforming using double-
jacketed membrane reactor. Int J Hydrogen Energy 2007;
32(18):48309.
[23] Peppley BA, Amphlett JC, Kearns LM, Mann RF. Methanol-
steam reforming on Cu/ZnO/Al
2
O
3
. Part 1: the reaction
network. Appl Catal A 1999;179(12):219.
[24] Peppley BA, Amphlett JC, Kearns LM, Mann RF.
Methanol-steam reforming on Cu/ZnO/Al
2
O
3
catalysts. Part
2. A comprehensive kinetic model. Appl Catal A
1999;179(12):3149.
[25] Pasel J, Emonts B, Peters R, Stolten D. A structured
test reactor for the evaporation of methanol on the
basis of a catalytic combustion. Catal Today
2001;69(14):193200.
[26] Froment GF, Bischoff KB. Chemical reactor analysis and
design. 2nd ed. New York, NY: John Wiley & Sons, Inc.; 1990.
[27] Bird RB, Stewart WE, Lightfoot EN. Transport phenomena.
2nd ed. New York, NY: John Wiley & Sons, Inc.; 1990.
[28] Perry R, Green D, Maloney J. Perrys chemical engineers
handbook. 7th ed. New York, NY: McGraw Hill, Inc.; 1999.
[29] Chan SH, Hoang DL, Ding OL. Transient performance of an
autothermal reformer a 2-D modeling approach. Int J Heat
Mass Transfer 2005;48(1920):420514.
ARTICLE IN PRESS

A Transient Study of Double-Jacketed Membrane Reactor Via Methanol Steam Reforming

Uploaded by

A Transient Study of Double-Jacketed Membrane Reactor Via Methanol Steam Reforming

Uploaded by

A transient study of double-jacketed membrane reactor via

methanol steam reforming

You might also like