Sei sulla pagina 1di 11

Chemical Engineering Science 56 (2001) 57}67

Simulation of a membrane reactor for oxidative dehydrogenation


of propane, incorporating radial concentration and
temperature pro"les
K. Hou , R. Hughes *, R. Ramos, M. MeneH ndez, J. SantamarmH a
Department of Chemical and Environmental Engineering, Faculty of Science, University of Zaragoza, 50009 Zaragoza, Spain
Department of Chemical Engineering, University of Salford, Salford M54WT, UK
Received 1 March 2000; received in revised form 7 June 2000; accepted 30 August 2000

Abstract
A model for membrane reactors is proposed, taking into account the radial concentration pro"les for each reactant, as well as the
temperature pro"les. This model is applicable for a membrane reactor with a "xed bed of catalyst inside the membrane and with
distribution of one of the reactants along the reactor through the membrane. The model predictions are compared with the results
obtained in an experimental reactor, and the e!ect of an increase in the reactor diameter and other changes in the operation
conditions are studied by means of the model.  2001 Elsevier Science Ltd. All rights reserved.

1. Introduction
Membrane reactors have been widely studied during
the last decade, as may be seen in several reviews (Hsieh,
1991, 1996; Shu, Grandjean, Van Seste, & Kaliaguine,
1991; Zaman & Chakma, 1994; Saracco & Specchia,
1994; Falconer, Noble, & Sperry, 1995; Coronas &
SantamarmH a, 1999). Although many possibilities exist, two
main kinds of membrane reactor may be identi"ed:
(a) the membrane is employed to separate selectively one
of the reaction products. This is used to improve the yield
in an equilibrium limited reaction, and has been demonstrated for several dehydrogenation reactions, using
hydrogen-selective membranes; (b) the membrane is employed to distribute one of the reactants along a catalyst
bed. This operation mode has been employed in many
oxidation reactions, such as oxidative coupling of
methane (Coronas, Menendez, & SantamarmH a, 1994a;
Tonkovich, Secker, Reed, Roberts, & Cox, 1995;
Ramachandra, Lu, Ma, Moxer, & Dixon, 1996), oxidative dehydrogenation of ethane (Coronas, Menendez,
& SantamarmH a, 1995a,b; Tonkovich, Zike, Jimenez,
Roberts, & Cox, 1996), propane (Pantazidis, Dalmon, &

* Corresponding author. Tel.: #44-161-2955081; fax: #44-1612955380.


E-mail address: r.hughes@chemistry.salford.ac.uk (R. Hughes).

Mirodatos, 1995; Ramos, Menendez, & SantamarmH a,


2000) and butane (TeH llez, Menendez, & SantamarmH a,
1997); butane oxidation to maleic anhydride (Qin,
Daiqui, & Zhongtao, 1995; TeH llez, Mallada, Menendez,
SantamarmH a, & Lombardo, 1996) and has also been proposed for methane oxidation to methanol (Chellappa,
Fuangfoo, & Viswanath, 1997). In many cases an
improvement in yield to the desired product has been
obtained, due to the lower oxygen partial pressure, that
often improves the selectivity to the desired products and
minimises production of CO and CO .

Simultaneously with the experimental studies, several
authors have developed mathematical models to predict
the behaviour of the reactor by means of simulation.
In this way, SantamarmH a, Miro, and Wolf (1991),
SantamarmH a, Menendez, Pena, and Barahona (1992) and
Reyes et al. (1993a,b) suggested that the distribution of
oxygen along several points along the reactor can improve the yield to ethane and ethylene in oxidative coupling of methane (OCM). Several authors employed
a simulation model to show how this system provides
improved selectivity in phthalic anhydride production
(Papageorgiu & Froment, 1996) and some safety and
operational advantages in oxidative dehydrogenation of
ethane (Al-Sherehy, Adris, Soliman, & Hughes, 1998).
For the latter with a large number of distribution points
behaviour similar to that of the membrane reactor is
obtained. Cheng and Xhuai (1995) and Kao, Lei, and Lin

0009-2509/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 0 0 ) 0 0 4 2 2 - X

58

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

(1997) presented models for a membrane reactor employed for OCM, showing that an improved yield is
obtained, when compared with a conventional "xed-bed
reactor. A similar membrane reactor model, with oxygen
distribution and product removal, also for OCM, was
presented by Lu, Dixon, Moser, and Ma (1997). A comparative simulation between a conventional "xed-bed
reactor and a ceramic dense membrane reactor, i.e. made
with oxygen-conducting materials, has been presented by
Kao et al. (1997). Papavassiliou, Lee, Nestlerode, and
Harold (1997) studied in detail the transport through
porous inorganic membranes, aimed at their use in
membrane reactors. Coronas, Gonzalo, Lafarga, and
Menendez (1997) showed how the e!ect of the catalytic
activity of the membrane may a!ect the yield improvements achieved by the oxygen distribution. This model
may explain some experimental results better than models that do not consider the catalytic activity of the
membrane. A simulation using a detailed kinetic model
for oxidative dehydrogenation of butane has been presented by Tellez et al. (1999).
A common hypothesis in all the above simulation
studies was that the radial concentration and temperature pro"les are #at. This approach is probably correct
in most laboratory-scale reactors, where the thickness of
the bed is small (only a few millimeters); however, if
a commercial-scale reactor is envisaged, it is necessary to
have information on the e!ect of the catalyst bed width
on the reactor behaviour. Obviously, in a wide membrane both concentration and temperature pro"les will
appear. From an economic point of view, it is desirable to
employ large diameter tubes, since this will decrease the
required membrane area. Therefore, since oxygen is supplied only from the membrane side, it is probable that
oxygen concentration pro"les would appear in a wide
membrane reactor. Also temperature gradients are more
probable in a wide reactor. Therefore, it was considered
important to have a model of a membrane reactor, taking
into consideration any concentration and temperature
pro"les. In the present work a model has been developed
for the oxidative dehydrogenation of propane in a membrane reactor using kinetics obtained over a V/MgO in
a small-scale reactor. The membrane used was a silicamodi"ed alumina membrane through which oxygen was
allowed to permeate into the enclosed catalyst bed.
Although the present results were obtained for the particular oxidative dehydrogenation of propane the model
may be easily extended to other catalytic selective oxidations. The experimental system is composed of a ceramic
membrane, in which a "xed bed of V/MgO catalyst is
enclosed. The experimental reaction system has been
described previously (e.g. Tellez et al., 1997). Dimensions
of the experimental reaction system are 0.7 cm internal
diameter, 14 cm length, with approximately 2 g of catalysts contained within the tube. A particle size of between
0.25 and 0.50 mm ensured the absence of inner concen-

tration gradients and this was con"rmed experimentally.


The membrane was obtained by impregnation of ceramic
micro"ltration membranes (supplied by SCT-US "lters)
with silica sols. Details on this modi"cation and comprehensive study on the performance of this reactor have
been presented by Ramos et al. (2000).

2. Reactor model
2.1. Reaction kinetics
From the results of an experimental study using a differential reactor of diameter 0.7 cm, in which propane/oxygen or propene/oxygen mixtures were fed to
a bed with a V/MgO catalyst, a Mars}van Krevelen
kinetic equation was deduced. These kinetic expressions
were employed to predict the results for several integral
reactors. For the case of a plug #ow "xed-bed reactor, the
kinetic equations were able to predict propane conversion in close agreement with those obtained experimentally (Ramos et al., 2000). These reaction kinetics
were therefore considered appropriate and were employed in the following 2-D membrane reactor model.
The kinetic data were obtained in the temperature range
450}5503C, at propane partial pressures between 2 and
12 kPa and with oxygen partial pressures from 2 to 12
kPa. Experiments feeding reaction products (propene,
water) were also carried out. An inhibition e!ect by water
was found in experiments in which water was fed with the
reactants, and such an e!ect was also included in the
kinetic equations. Experimental data were tested against
several reaction models by a non-linear least-squares
method and the best "t selected. A detailed discussion of
the kinetic modelling may be found in Ramos et al.
(2000). The reaction set describing the process of oxidative dehydrogenation of propane is shown in Table 1,
together with the values of the kinetic constants.
2.2. 2-D model for an isothermal catalytic membrane
reactor
The main di!erence between the membrane reactor
model and conventional models for "xed-bed reactors
lies in the existence of a radial component of gas velocity,
that cannot be neglected, since one of the reactants is fed
through the reactor wall (Fig. 1). In this way, together
with the di!usive transport in the radial direction, a convective radial mass transport term is added.
The 2-D model of a catalytic membrane reactor was
developed with the following assumptions:
(a) Isobaric reactor conditions.
(b) Cylindrical symmetry.
(c) The gas velocity at any point has two components:
axial and radial. The axial component is independent

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

59

Table 1
Kinetics of oxidative dehydrogenation of propane
Reaction
C H #SoPC H #H O#S
 
 

C H #7SoP3CO#4H O#7S
 

C H #10SoP3CO #4H O#10S
 


C H #6SoP3CO#3H O#6S
 

C H #9SoP3CO #3H O#9S
 


0.5O #SPSo


(1)
(2)
(3)
(4)
(5)
(6)

Reaction rate
r "k P   h f 

 !& &r "k P   h f 

 !& &r "k P   h f 

 !& &r "k P   h f 

 !& &r "k P   h f 

 !& &r "k P  h

 - 1

where
k P 
 h "
 k P #(k #7k #10k )P
#(6k #9k )P  
 -


 ! &

 !&
h "1!h
1

1
f  "
& - 1#K P
F & Kinetic parameters (k "k exp(!E /R))
G
G
G
i
1
2
k
kmol/kg s bar
1.756;10
18.62
G
E
kJ/mol
154.5
97.36
G

3
0.0196
48.41

4
6.32;10
208.5

5
2.274;10
182.3

K
F
2.765;10
139.9 (*H )
F

6
1.986;10
205.1

with
r
u "u .
P
UR
Fig. 1. Scheme of the membrane reactor employed for selective oxidations, with distribution of oxygen along a catalyst bed.

(d)
(e)

(f )
(g)

of the radial position. This approximation is equivalent to neglecting the pressure drop due to the wall
friction, compared with that due to the particles.
Neglecting the change in volume due to reaction, this
implies that the radial component of gas velocity
varies linearly with the radius (for detailed derivation,
see the appendix).
Radial transport of mass has two components: diffusive and convective.
Mass transport by di!usion in the axial component is negligible compared with the convective
transport.
The ideal gas law is applicable, given the high temperatures and low pressures used.
A pseudo-homogeneous model is adopted for the
catalytic section.

Writing a mass balance for a compound i, results in the


following equation at steady state:
*F
De
G" G
*z
R
%

 



1 * *(P /)
r G
r *r
*r

1 *(u r) P
P G #o l r
#
GH H
r *r P
(1)

(2)

A heat balance yields

  


*
1
*(u )
P
j * *

"
r
# P Cp G
G
*z
F Cp r *r *r
*r
P
G G
#o r (!*H )
H
H

(3)

With the following initial and boundary conditions:


at z"0 F "F , " ,
G
G


(4)

at r"R P "0, except P  "1 " -,


G
0

*P 
*P
G "0, except b
- "u ,
U
*r
*r
0
0
q"!j

*
.
K *r

(5)

The above equations were solved by the method of lines,


i.e. the derivatives with respect to the radius were approached by "nite di!erences and the resulting set of
ordinary di!erential equations were solved by a Runge}
Kutta method.
Neglecting the static contribution to the radial
heat transfer, the e!ective radial thermal conductivity
for the catalytic bed was computed using the following

60

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

correlation (Li, 1986):


0.14RePr
j "j
.
(6)

D 1#46(d /d )
N R
Gas viscosities were computed using Chung's equation
(Reid, Prausnitz, & Poling, 1987) and the viscosity of the
mixture by using Wilke's equation (Reid et al., 1987).
Thermal conductivities for each component and gas mixture were also calculated following a method taken from
Reid et al. (1987). The molecular di!usivities were calculated from Fuller's equation (Fuller, Schettler, &
Giddings, 1966) and Wilke's equation (Wilke, 1950).

3. Isothermal simulations
In a "rst approach, the membrane reactor was
simulated assuming constant temperature. This allows
the e!ect of oxygen distribution along the reactor to be
observed, as well as the e!ect of operating conditions
(heterogeneous distribution of oxygen, reactor radius).
Basic input data for the isothermal simulations are given
in Table 2.
The results for the 2-D model were "rst compared with
conventional 1-D models for catalytic methane reactions
(Cheng & Xhuai, 1995; Kao et al., 1997). From a comparison of the 2-D model with the results using a 1-D
model, for the same operating conditions (Fig. 2) it is
clear that for the conditions of the experiments both
a 1-D or a 2-D model can predict equally well the
experimental results. This is due to the small diameter of
the experimental system (i.d. 0.7 cm of catalytic bed used).
Oxygen partial pressure pro"les obtained with the 2-D
model at four reactor lengths for a temperature of 823 K
are shown in Fig. 3. An oxygen partial pressure gradient
is apparent albeit small, with a higher concentration near
the reactor wall, even for conditions that result in similar
behaviour as that predicted by the 1-D model. Values of
the oxygen partial pressure are lower near the inlet
(z/"0.1), increasing progressively with distance along
the reactor.
For larger radius of the reactors simulation was done
at two temperatures with di!erent feed compositions.
For each temperature all other reaction conditions, including reaction temperature and pressure, feed composition and contact time were kept constant. Reactor size
was changed by keeping constant the length/radius ratio.
This allows the e!ect of reactor radius on propane conversion, propylene selectivity and radial pro"les of oxygen partial pressure to be examined (Figs. 4a, b and 5). It
may be seen that both the propane conversion and the
propylene selectivity decrease as the reactor radius increases. It seems that in a narrow reactor it is possible to
have a relatively #at oxygen concentration pro"le,
whereas in a wide reactor large oxygen concentration
di!erences occur at certain values of r/R, whereas the

Table 2
Basic input data
Reactor radius, cm
Catalyst weight, g
Reaction pressure, bar
Reaction temperature, K
= /F   , kg h/kmol
 !&
Feed composition, C H /N
  

0.35
2.8

0.70
44.8

1.12
183.5

1.40
358.4

1.12, 120
798, 823
65.3, 130.7
4/42, 4/88

oxygen concentration at points, near to the centre of the


reactor is so small that almost no reaction is present
there. The decreased propane conversion in the wider
reactors is related to the lower selectivity, since the nonselective reactions (CO and CO formation) consume

more oxygen per mole of propane reacted than the selective reactions. In conclusion, a decreased selectivity is the
drawback to be paid for a lower cost of membrane when
a wide membrane is employed. Probably an optimum
membrane size exists, that will depend on the relative
costs of membranes and reactants.
The oxygen partial pressure pro"les change, as may be
expected, with the reactor radius, giving larger gradients
for the widest reactors (Fig. 5). In this "gure the oxygen
partial pressures are estimated at a "xed z/ value of 0.4
for all reactor tube radii. Also the length/radius is "xed at
80 for those longer tube lengths, decreasing to 40 for the
shortest tube. Steep gradients near the reactor wall occur
for R'0.7 cm, whereas the pro"les are almost #at in the
central region. When Fig. 5 pro"les are compared with
those in Fig. 3 (note the di!erent oxygen partial pressure
scales) it can be seen that for the 0.35 cm radius 14 cm
length reactor, simulated in Fig. 3, the oxygen partial
pressure gradients are not very signi"cant. This is con"rmed by the plot of the gradient for this size reactor
shown in Fig. 5 by the lowest curve.
Simulation is a useful tool to optimize the reactor. The
developed model has been employed to check the e!ect of
the propane/oxygen ratio. The results are shown in
Fig. 6. There is a continuous increase in propane conversion when the oxygen #ow feed is increased, which combined with the decrease in selectivity (not shown) results
in a maximum in yield for a propane to oxygen ratio of
around 1. This simulation result is in agreement with
previous experimental "ndings (Ramos et al., 2000). It
has also been suggested (Coronas, Menendez, &
SantamarmH a, 1994b) that in order to dose the oxygen with
a tailored #ux at each point of the bed, use of a variable
permeation membrane should be employed. The e!ect of
a variable permeability along the membrane is considered by making the radial component of gas #ux at the
wall (u ) vary linearly with the position. Therefore, u is
U
U
given by
u (z)"(1!a)u #2au z/
U
U
U

(7)

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

61

Fig. 2. Comparison of simulated results with experimental results for 1-D and 2-D models (experimental results by Ramos et al. (2000)).

Fig. 3. Oxygen radial partial pressure pro"les and their variation with reactor length, P"1.2 bar, catalyst weight: 2.8 g, total #ow: 200 cm/min,
C H /O /N ratio"4/8/88.
   

when the permeation increases linearly along the reactor,


and by
1!az/
u
u (z)"
U
1!a/2 U

non-uniform modes with the uniform mode, we de"ne


the percentage change in propane conversion and
propene yield as

(8)

when the permeation decreases along the reactor. Here


a is the non-uniform factor and u is the average perU
meation rate of oxygen along the membrane.
Thus, when a"0, an uniform distribution is obtained along the membrane, and the most non-uniform
distribution is obtained with a"1. To compare the

X
(a)!X   (a"0)
!&
dX   " ! &
;100,
!&
X   (a"0)
!&
>
(a)!>   (a"0)
!&
d>   " ! &
;100.
!&
>   (a"0)
!&

62

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

Fig. 4. E!ect of reactor radius on: (a) propane conversion; and (b) propene selectivity and yield.

The percentage variations of propane conversion and


propene yield, resulting from the non-uniform modes
with varying a, are shown in Fig. 7a and b where the
di!erence from the uniform mode for the conversion of
propane and yield of propene are represented by dX  
!&
and d>   , respectively. It can be seen that a linear
!&
increase mode has a negative e!ect on the propane
conversion, and positive e!ects on the propene yield,
whereas the opposite happens with the linear decrease in
permeation. In the former case, the propane concentration in the "rst part of the reactor is high compared to
that in the remainder of the reactor and this coupled with
the low oxygen concentration decreases the oxidative
dehydrogenation of propane. Also the relatively low oxygen level reduces the extent of further oxidation of the

product, leading to an increase in propene selectivity.


The opposite e!ect occurs with a linear decrease in oxygen permeation along the reactor length. If the propane/oxygen ratio is equal to or greater than the value
that provides the maximum yield, there is very little e!ect
on conversion and yield, particularly on the yield. However, for the case with a small propane/oxygen ratio (i.e.
the amount of oxygen fed is greater than the optimum
value), the e!ect of the oxygen distribution should be
considered. In summary, an uniform mode should be
adopted, provided that an optimum amount of oxygen
feed is employed. Other distribution methods, proposed
by Coronas et al. (1994b) but not studied here, may provide other advantages (e.g. higher oxygen conversion if
a non-permeable zone is located at the end of the reactors).

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

63

Fig. 5. E!ect of reactor dimensions on radial pro"les of oxygen partial pressure, =/F   "130.7 kg h/kmol, "823 K, P"1.2 bar,
!&
C H /O /N "4/8/88.
   

Fig. 6. E!ect of the oxygen #ow rate on propene yield, other conditions as in Fig. 5.

4. Non-isothermal simulation
This simulation has focused on the e!ects of inlet
temperature of the feed and the thermal conductivity on
propane conversion and selectivity, as well as the temperature pro"les along axis and radius. Fig. 8 shows the
radial average temperature along the reactor axis, for two
values of the membrane thermal conductivity (j ) obK
tained from the thermal properties of alumina quoted in
Kuzman (1976) at two values of the initial temperature

( ). It may be seen that the temperature tends to stabil 


ize at a value depending on the membrane thermal conductivity, and independent of the initial temperature. The
minimum obtained for a j value of 6.69;10\ at an
K
inlet feed temperature of 798 K is caused by the relatively
high thermal conductivity of the membrane in this case,
together with the lower shell side oxygen feed temperature of 773 K. Under these conditions, the reaction
temperature near the inlet decreases due to heat transfer
to the exterior being faster than the heat generated by

64

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

Fig. 7. Percentage variation of propane conversion and propene yield, by changing from an uniform distribution of oxygen along the membrane to
a linear variation of permeation, other conditions as the same in Fig. 5: (a) the linear increase mode; and (b) the linear decrease mode

reaction. Ultimately the pro"le levels out as shown in


Fig. 8. In the membrane reactor the reaction proceeds at
a similar rate along the reactor length due to the low
oxygen partial pressure, limited by permeation through
the membrane, and its small variation along the reactor
axis. Additionally, simulated propane partial pressures
were e!ectively independent of temperature and thermal
conductivity in the region investigated. Because of this
near uniform reaction rate, the heat generation will be
similar at all points, thus explaining the #at axial temperature pro"le towards the end of the reactor. The safety
improvement and the decreased hot spot formation have
been claimed as advantages for this kind of reactor
(Coronas et al., 1995a), and is con"rmed by the above
results.
The radial temperature pro"les at several points of the
reactor are shown in Fig. 9, for two di!erent values of the

membrane thermal conductivity. It is surprising that


even in a narrow reactor as the one simulated in this
"gure (0.7 cm diameter), some temperature gradients can
appear; for example, in one of the curves (corresponding
to j "1.67;10\ kJ/m s K and z/"0.15) the radial
K
variation of temperature is close to 15 K. Obviously, if
a wider reactor is preferred, the radial temperature pro"les must be taken into account. Some similarities occur,
between a membrane reactor and a catalyst pellet in
terms of radial pro"les of concentration and temperature,
but this will be the subject of future work.

5. Conclusions
A mathematical model of a membrane reactor has
been employed to predict radial concentration and

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

65

Fig. 8. Radial average temperature pro"le along reactor axis as a function of j , =/F   "130.7 kg h/kmol, "773 K, P"1.2 bar,
K
!&
Q
C H /O /N "4/5/88.
   

Fig. 9. Radial temperature pro"les along reactor length for two values of j , =/F   "130.7 kg h/kmol, "773 K, P"1.2 bar,
K
!&
Q
C H /O /N "4/5/88.
   

temperature radial pro"les in a membrane reactor. The


model, which to the authors knowledge, considers for the
"rst time, the convective #ow in the radial direction, has
shown that the radial concentration pro"les a!ect the
selectivity and yield attainable in the reactor. Wide membrane tubes, that are suitable from an economical point
of view, result in lower selectivity. The model is useful to
estimate the changes in selectivity, allowing this to be
considered in the "nal reactor design. For the studied
reaction (oxidative dehydrogenation of propane), a propane/oxygen ratio of around 1 provides the highest yield
to propene. With this feed a variation in permeability
along the reactor length does not provide any signi"cant

improvement in yield. The model presented herein is also


useful to predict the temperature gradients that may
appear in the catalyst bed, both radially and axially. In
general, this model may be useful for the scale-up of
membrane reactors, from laboratory to pilot-plant scale
and for preliminary feasibility studies on their industrial
application.

Notations
CG
N
D
G

heat capacity of component i, kJ/mol K


di!usivity of component i, m/s.

66

d
N
d
R
E
G
f 
&F
G
*H
F
*H
H
K
F
k
H
k
H

P
P
G
P
P
q
R
%
R
C
R
r
r
H
S

Q
u , u
U U
u,
P
z

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67

diameter of catalyst particle, m


diameter of catalyst bed, m
activation energy of reaction i, kJ/mol
de"ned in Table 1
molar #ux of component i, kmol/m s.
adsorption heat of water vapour, kJ/mol
reaction enthalpy of reaction j, kJ/mol
adsorption constant of water vapour, 1/bar
reaction rate constant of reaction j, kmol/kg s bar.
pre-exponential factor of reaction j, kmol/kg s bar.
length of reactor, m
reaction pressure, bar
partial pressure of component i, bar
Prandtl number, !NI
HD
heat #ux, kJ/m s
gas universal constant, bar m/kmol K
Reynolds number, BN SM
I
radius of catalyst bed, m
radial coordinate of catalyst bed, m
reaction rate of reaction j, kmol/kg s
active site of catalyst
reaction temperature, K
inlet temperature of feed, K
temperature in shell-side, K
local and average permeation rate of oxygen
along the membrane,
respectively, kmol/m s.
rate of oxygen inside of catalyst bed, kmol/m s.
axial coordinate, m

Greek symbols
a
factor of non-uniform distribution
b
permeability of oxygen, kmol/m s bar
e
porosity of catalyst bed
j
e!ective thermal conductivity of catalyst bed,

kJ/m s K
j
thermal conductivity of #uid phase in catalyst
D
bed, kJ/m s K
j
thermal conductivity of membrane, kJ/m s K
K
v
stoichiometric coe$cient of component in reacGH
tion j
o
density of catalyst bed, kg/m
h
fractional surface coverage of oxygen.

h
fractional vacant surface
Q
Subscripts
i
component i
j
reaction j

Appendix
It can be considered that the gas velocity at any point
has two components: an axial (u ) and a radial (u )
X
P
component. In a "rst approach the change in molar
number due to reaction is neglected. This is reasonable

since for the experimental system modelled here, given


the small change in total molar number (e.g. for a 50%
conversion, the change in molar number is only around
0.5%). If the catalyst bed enclosed in the membrane is of
cylindrical shape, a mass balance over a volume element
with radius equal to the bed radius, and a height equal to
dz, gives
(nR)(du /dz)"u (2nR)
X
U
and therefore

(A.1)

du /dz"2u /R.
(A.2)
X
U
A mass balance over a cylindrical volume element with
radius equal to r gives
(nr)(du /dz)"u (2nr).
(A.3)
X
P
Combining Eqs. (A.2) and (A.3), the following relationship is obtained:
u "u (r/R).
(A.4)
P
U
Therefore, in this system the radial component of the gas
velocity varies linearly with the radius.

References
Al-Sherehy, F., Adris, A. M., Soliman, M. A., & Hughes, R. (1998).
Avoidance of #ammability and temperature runaway during oxidative dehydrogenation using a distributed feed. Chemical Engineering
Science, 53, 3965.
Cheng, S., & Shuai, X. (1995). Simulation of a catalytic membrane
reactor for oxidative coupling of methane. A.I.Ch.E. Journal, 41,
1598.
Chellappa, A. S., Fuangfoo, S., & Viswanath, D. S. (1997). Homogeneous oxydation of methane to methanol: E!ect of CO , N and


H at high oxygen conversions. Industrial and Engineering Chem
istry Research, 36, 1401.
Coronas, J., Gonzalo, A., Lafarga, D., & MeneH ndez, M. (1997). E!ect of
the membrane activity on the performance of a catalytic membrane
reactor. A.I.Ch.E. Journal, 43, 3095.
Coronas, J., MeneH ndez, M., & SantamarmH a, J. (1994a). Methane oxidative coupling using porous ceramic membrane reactors: II. Reaction
studies. Chemical Engineering Science, 49, 2015.
Coronas, J., MeneH ndez, M., & SantamarmH a, J. (1994b). Development of
ceramic membrane reactors with a non-uniform permeation pattern. Application to methane oxidative coupling. Chemical Engineering Science, 49, 4749.
Coronas, J., MeneH ndez, M., & SantamarmH a, J. (1995a). The porous-wall
ceramic membrane reactor: An inherently safer contacting device
for gas-phase oxidation of hydrocarbons. Journal of oss Prevention
in the Process Industries 8, 97.
Coronas, J., MeneH ndez, M., & SantamarmH a, J. (1995b). Use of a ceramic
membrane reactor for the oxidative dehydrogenation of ethane to
ethylene and higher hydrocarbons. Industrial and Engineering Chemistry Research, 34, 4229.
Coronas, J., & SantamarmH a, J. (1999). Catalytic reactors based on porous
ceramic membranes. Catalysis Today, 51, 377.
Falconer, J. L., Noble, R. D., & Sperry, D. P. (1995). Catalytic
membrane reactors, In Stern, S. A., & Noble, R. D. (Eds.), Membrane
separations technology: Principles and applications (p. 669).
Amsterdam: Elsevier.

K. Hou et al. / Chemical Engineering Science 56 (2001) 57}67


Fuller, E. N., Schettler, P. D., & Giddings, J. C. (1966). A new method
for prediction of binary gas phase di!usion coe$cients. Industrial
and Engineering Chemistry, 58, 19.
Hsieh, H. P. (1991). Inorganic membrane reactors. Catalysis ReviewsScience Engineering, 33(1&2).
Hsieh, H. P. (1996). Inorganic membranes for separation and reaction.
Amsterdam: Elsevier.
Kao, Y. K., Lei, L., & Lin, Y. S. (1997). A comparative simulation study
on oxidative coupling of methane in "xed-bed and membrane
reactors. Industrial and Engineering Chemistry Research, 36, 3583.
Kuzman, R. (1976). Handbook of thermodynamic tables and charts (p. 37).
London: McGraw-Hill.
Li, S. (1986). Chemistry and catalytic reaction engineering. House of
Chemical Industry, Beijing, (p. 239).
Lu, Y., Dixon, A. G., Moser, W. R., & Ma, Y. H. (1997). Analysis and
optimization of cross-#ow reactors for oxidative coupling of methane. Industrial and Engineering Chemistry Research, 36, 559.
Pantazidis, A., Dalmon, J. A., & Mirodatos, C. (1995). Oxidative dehydrogenation of propane on catalytic membrane reactors. Catalysis
Today, 25, 403.
Papageorgiu, J. N., & Froment, G. F. (1996). Phtalic anhydride synthesis-reactor optimization aspects. Chemical Engineering Science,
51, 2091.
Papavassiliou, V., Lee, D., Nestlerode, J., & Harold, M. P. (1997).
Pneumatically controlled transport and reaction in inorganic membranes. Industrial and Engineering Chemistry Research, 36, 4954.
Qin, Z., Daiqui, Y., & Zhongtao, H. (1995). Maleic anydride via membrane catalysed oxidation of n-butane. Shiyou Huagong, 24, 875 (in
Chinese).
Ramachandra, A. M., Lu, Y., Ma, Y. H., Moxer, W. R., & Dixon, A. G.
(1996). Oxidative coupling of methane in porous vycor membrane
reactors. Journal of Membrance Science, 116, 253.
Ramos, R., MeneH ndez, M., & SantamarmH a, J. (2000). Oxidative deydrogenation of propane in an inert membrane reactor. Catalysis
Today, 56, 239.
Reid, R. C., Prausnitz, J. M., & Poling, B. E. (1987). The properties of
gases & solids. New York: McGraw-Hill.
Reyes, S. C., Iglesia E., & Kelkar, C. P. (1993a). Kinetic-transport
models of bimodal reaction sequences, 1, Homogeneous and

67

heterogeneous pathways on oxidative coupling of methane. Chemical Engineering Science, 48, 2643.
Reyes, S. C., Kelkar, C. P., & Iglesia, E. (1993b). Kinetic transport
models and the design of catalyst and reactor for the oxidative
coupling of methane. Catalysis etters, 19, 167.
SantamarmH a, J. M., MiroH , E. E., & Wolf, E. E. (1991). Reactor simulation
studies of methane oxidative coupling on a Na}NiTiO3 catalyst.
Industrial and Engineering Chemistry Research, 30, 1157.
SantamarmH a, J., MeneH ndez, M., Pen a, J. A., & Barahona, J. I. (1992).
Methane oxidative coupling in "xed bed reactors with a distributed
oxygen feed: A simulation study. Catalysis Today, 13, 353.
Saracco, G., & Specchia, V. (1994). Catalytic inorganic membrane
reactors: Present experience and future opportunities. Catalysis Reviews-Science Engineering, 69, 1036.
Shu, J., Grandjean, B. P. A., Van Seste, A., & Kaliaguine, S. (1991).
Catalytic palladium-based membrane reactors: A review. Canadian
Journal of Chemical Engineering, 69, 1036.
TeH llez, C., Mallada, R., MeneH ndez, M., SantamarmH a, J., & Lombardo,
E.A. (1996). Synthesis of maleic anhydride in a membrane
reactor with high butane concentration. Proceedings of the XV
Iberoamerican symposium on catalysis, vol. 2, Cordoba, Argentine,
(p. 775).
Tellez, C., MeneH ndez, M., & SantamarmH a, J. (1997). Oxidative dehydrogenation of butane using membrane reactors. A.I.Ch.E. Journal,
43, 777.
Tellez, C., MeneH ndez, M., & SantamarmH a, J. (1999). Simulation of an
inert membrane reactor for the oxidative dehydrogenation of butane. Chemical Engineering Science, 54, 2925.
Tonkovich, A. L. Y., Zike, J. L., Jimenez, D. M., Roberts, G. L., & Cox,
J. L. (1996). Experimental investigations of inorganic membrane
reactors: A distributed feed approach for partial oxidation reactions. Chemical Engineering Science, 51, 798.
Tonkovich, A. L. Y., Secker, R. B., Reed, E. L., Roberts, &G. L., & Cox,
J. L. (1995). Membrane reactor/separator: A design for bi-molecular
reactant addition. Separation Science Technology, 30, 1609.
Wilke, C. R. (1950). Di!usional properties of multicomponents gases.
Chemical Engineering Progress, 46(2), 95.
Zaman, J., & Chakma, A. (1994). Inorganic membrane reactors. Journal
of Membrance Science, 92, 1.

Potrebbero piacerti anche