Anda di halaman 1dari 13

See

discussions, stats, and author profiles for this publication at:


https://www.researchgate.net/publication/222530810

Adsorption-Enhanced Steam-Methane
Reforming

Article in Chemical Engineering Science September 2000


DOI: 10.1016/S0009-2509(99)00597-7

CITATIONS READS

191 228

2 authors:

Yulong Ding Esat Alpay


University of Birmingham University of Surrey
293 PUBLICATIONS 11,227 CITATIONS 70 PUBLICATIONS 1,159 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Nanouptake_ COST action View project

All content following this page was uploaded by Esat Alpay on 21 June 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Chemical Engineering Science 55 (2000) 3929}3940

Adsorption-enhanced steam}methane reforming


Y. Ding, E. Alpay*
Department of Chemical Engineering and Chemical Technology, Imperial College of Science, Technology and Medicine, Prince Consort Road,
London SW7 2BY, UK
Received 8 September 1999; received in revised form 6 December 1999; accepted 13 December 1999

Abstract

Experimental and theoretical studies of steam}methane reforming in the presence of a hydrotalcite-based CO adsorbent are

presented. Attention is given to the analysis of the transient behaviour of a tubular (integral) reactor when an Ni-based catalyst is
admixed with the adsorbent. Considerable enhancement of the methane conversion is experimentally demonstrated. Enhancement
arises from the favourable shifts in the reaction equilibria of the reforming and water}gas shift reactions towards further CO

production. As predicted, the potential for conversion enhancement is shown to increase under the conditions of a high reactor space
time, high operating pressure, or a low steam-to-methane feed ratio, i.e. when reaction equilibrium limitations are important.
A mathematical model, accounting for mass transfer limited adsorption kinetics, non-linear (Langmuirian) adsorption equilibria and
a general reaction kinetic model, is shown to accurately predict the observed elution pro"les from the reactor, and thus the degree of
conversion enhancement.  2000 Elsevier Science Ltd. All rights reserved.

Keywords: SMR; Sorption enhancement; Transient kinetics; Ni-catalyst; Hydrotalcite; CO adsorbent; Mathematical modelling


1. Introduction hydrogen can permeate) and, more recently, polymeric,


ceramic and zeolitic membranes. The membranes may
The advantages of coupling reaction systems with act as permselective barriers, or as an integral part of the
some form of in situ separation have been widely re- catalytically active surface. Practical issues such as mem-
ported in the literature. Such hybrid con"gurations may brane pore blockage, thermal and mechanical stability,
substantially improve reactant conversion or product and the dilution caused by the need for sweep (i.e. per-
selectivity and, for reversible reactions, establish a more meate purge) gases, have limited the usefulness of the
favourable reaction equilibrium than that which could be membrane reactor systems. Nevertheless, the bene"ts of
achieved under conventional reactor operation. Reaction the membrane systems have been demonstrated though
enhancement may enable a lower temperature of opera- a wide number of experimental reaction studies, exam-
tion, which in turn may alleviate the problems associated ples of which include the dehydrogenation of ethane
with catalyst fouling, high process energy requirements (Tsotsis, Champagnie, Vasileiadis, Zraka & Minet, 1992),
and poor energy integration within the plant environ- cyclohexane (Sun & Khang, 1988), ethylbenzene (Wu,
ment. For gas-phase catalytic reactions, the separation Gerdes, Pszczolowski, Bhave & Liu, 1990), and acetylene
can be based on adsorption, selective permeation (Itoh, Xu & Sathe, 1993), CO production via the
through a membrane, or through simultaneous reaction water}gas shift reaction (Uemiya, Sato, Ando & Kikuchi,
of the targeted molecule (e.g. the reaction inhibitor) with 1991), and steam}methane reforming (Adris, Lim
a chemical acceptor. & Grace, 1994, 1997).
A comprehensive review on membrane-based reaction In comparison to the membrane reactors, a relatively
systems has been given by Armor (1995). Advances have small amount of work has been carried out on systems
been made in the use of metallic membranes (often combining reaction with adsorption or chemical accep-
Group VIII metals which only small molecules like tor-based separation processes. Even so, such processes
o!er distinct advantages to the membrane-based systems
* Corresponding author. Tel.: 0044-171-594-5625; fax: 0044-171-594- in terms of the material tolerance to high temperatures
5604. and pressures, and the wide choice and availability of
E-mail address: e.alpay@ic.ac.uk (E. Alpay). adsorbents for achieving the desired separations under

0009-2509/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 9 9 ) 0 0 5 9 7 - 7
3930 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

reaction conditions. Furthermore, even through use of Reforming reactions (1) and (2) are strongly endothermic,
a purge gas for regeneration, the e!ective separation of so the forward reaction is favoured by high temperatures,
the primary adsorbate from other non- or weakly adsor- while the water}gas shift reaction (3) is moderately
bing species can be achieved. Some examples of recent exothermic and is therefore favoured by low temper-
works employing chemical acceptors or adsorbents for atures. The reforming reactions will also be favoured at
reaction enhancement are now summarised. low pressures, whereas the water}gas shift reaction is
Han and Harrison (1994) studied hydrogen production largely una!ected by changes in pressure. In the presence
via the water}gas shift reaction using CaO as a CO of a selective CO adsorbent, the conversion of CH to
  
acceptor in a tubular reactor. CO conversions were CO though reaction (2) is favoured, as is the production

reported which exceeded that of the thermodynamic of CO through CO intermediate. For a reaction which

equilibrium conversion under the speci"ed operating is not kinetically limited, the use of an adsorbent will thus
conditions. Brun-Tsekhovoi, Zadorin, Katsobashvili and enable a lower operating temperature for a desired con-
Kourdyumov (1986) showed a very signi"cant enhance- version. However, on equilibration of the adsorbent, the
ment of CH conversion to H in a #uidised bed reactor separation e!ect is, of course, lost. This necessitates the
 
containing Ni-based catalyst balls, and a specially treated periodic regeneration of the adsorbent and thus, for
form of dolomite as adsorbent. Typical industrial operat- example, the pressure and concentration swing type op-
ing conditions were considered in this work, i.e. pressure erations mentioned above. In other words, sorption-en-
levels of 10}10 kPa, and an operating temperature of hanced reaction processes are inherently dynamic in
6273C. Goto, Tagawa and Oomiya (1993) studied the operation. Adequate design and scale-up of such pro-
dehydrogenation of cyclohexane over a Pt}alumina cata- cesses will thus require information on the kinetics of
lyst and CaNi alloy as a hydrogen acceptor. The adsorption and desorption, as well as reaction kinetic

workers showed that at 150}1903C and ambient pres- models under transient conditions in the presence of an
sure, the overall conversion of cyclohexane to benzene adsorbent.
could be exceeded by three-fold when compared to the Research work on the kinetics of the SMR process
catalyst-only case. Most recently, Carvill, Hufton and dates back to the 18th century (see Sabatier, 1922; Marek
Sircar (1996) and Hufton, Mayorga and Sircar (1999) & Hahn, 1932), with the "rst extensive study by Akers
describe the general concept of the Sorption Enhanced and Camp in 1955. A good review of the work up until
Reaction Process (SERP), which utilises pressure and 1970 is given by Van Hook (1980), which covers kinetic
concentration swing adsorption principles for reaction studies over porous nickel catalysts and nickel foil
enhancement; see also Vaporciyan and Kadlec (1989) and over large temperature (260}10003C) and pressure
Alpay, Chatsiriwech and Kershenbaum (1995). CO pro- (100}5000 kPa) ranges. A considerable amount of work
duction via the reverse water}gas shift reaction was spe- on the kinetic aspects of the SMR process has been
ci"cally considered by Carvill et al. (1996), in which NaX carried out since 1970; see, for example, Schnell (1970),
zeolite was used as an adsorbent for water. The authors Ross and Steel (1973), Allen (1975), Phung Quach and
showed that at 2503C and 480 kPa, a CO conversion of Rouleau (1975), Munsted and Grabke (1981), De Deken,

36% could be achieved; a conventional plug #ow reactor Deves and Froment (1982), and Xu and Froment (1989a,
required an operating temperature of 5653C for the same b). To date, the rate models proposed by Xu and Fro-
conversion. Furthermore, the process generated a high ment (1989a) are considered to be the most general in
purity (#99% (v/v)) CO stream as product. Hufton et al. form, and have been extensively tested under typical
(1999) applied the SERP concept to H production via industrial operating conditions (see, also, Elnashaie,

the steam}methane reforming (SMR) reactions. In speci- Adris, Al-Ubaid & Soliman, 1990). However, like most
"c, using a hydrotalcite-based CO adsorbent, and previous work, the models are applicable to steady-state

a commercial Ni-based catalyst, the authors showed that kinetics, and untested for forced-dynamic operation. It is
at 4503C and 480 kPa, #95% (v/v) H could be produc- interesting to note, however, that where some attention

ed directly from reactor. The CH to H conversion was has been given to the transient kinetics, ideal surface
 
82%, which could only be achieved at approximately conditions were maintained though vacuum operations;
6503C with a conventional SMR reactor. see, for example, Ross and Steel (1973). This, of course,
The present work also considers the sorption-en- limits the applicability of such models for process design
hanced steam}methane reforming (SE-SMR) process. applications.
The key reactions of the SMR process are given as: In this work, attention is given to the analysis of SMR
reaction kinetics under transient conditions depicting
CH #H O 0 CO#3H , *H "206 kJ/mol, (1)
    SERP-type operation, both in the presence and absence
CH #2H O 0 CO #4H , *H "164.9 kJ/mol, of a selective absorbent for CO . Particular attention is
     
given to the transient analysis of the Xu and Froment
(2)
(1989a) kinetic model. The work then considers the in#u-
CO#H O 0 CO #H , *H "!41 kJ/mol. (3) ence of operating parameters on the degree of separation
   
Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940 3931

enhancement. In doing so, mathematical models of the a convection oven. The outlet #ow rate from the reactor
process are veri"ed, which can ultimately be used in the was monitored with an Aalborg GFM-17 mass #ow
design, analysis or scale-up of pressure or concentration meter and a soap-bubble #owmeter. The catalyst bed
swing based adsorptive reactors. temperature was detected with a type K thermocouple
(positioned at the centre of the reactor along the central
axis), and a back pressure regulator used to maintain
2. Experimental a constant reactor pressure. The reactor e%uent #ow was
split into sample and vent lines, each equipped with
A commercial Ni-based catalyst (United Catalyst Inc.) a water condenser. For the former, time delays in the
containing 25}35% Ni, 25}35% NiO, 5}15% MgO and sample analysis were minimised by use of 1/16 transfer
15}25% sodium silicate, was used in this work. The lines, and a low-volume condenser unit. The sample line
original catalyst (1/8 cylindrical pellets; BET area of was connected to a Valco 16-loop valve, the operational
137.6 m/g) was crushed and sieved into two particle size schedule of which was computer controlled. The sample
groups: 0.11}0.25 and 0.25}0.5 mm. The CO adsorbent line and the sample valve were heat traced, and the

consisted of an industrially supplied potassium promoted temperature controlled at 1103C by a PID controller.
hydrotalcite, which was previously measured for its capa- A Shimadzu gas chromatograph (GC-14B), equipped
city and stability under wet gas conditions (see Ding with a TCD detector and a Porapak-Q column, was used
& Alpay, 2000). The adsorbent was also crushed and for sample analysis. In addition, two on-line Telegan
sieved into the particle size groups mentioned above. As CO analysers (0}30 and 0}5%FSD) were used to moni-

a catalyst diluent, e.g. in the absence of adsorbent, silicon tor the reactor e%uent CO concentration.

carbide particles were employed. High-purity methane For the reaction studies in the absence of adsorbent,
(99.95% (v/v)) and hydrogen (99.995% (v/v)) gases were approximately 7.2 g of catalyst was admixed with dense
supplied from gas cylinders; steam supply to the SMR silicon carbide particles (&1 : 3 mass ratio), and packed
reactor was generated from distilled water. into the reactor. For the sorption-enhanced reaction
A schematic diagram of the experimental apparatus is studies, approximately 7.2 g of catalyst was admixed with
given in Fig. 1. The reactor consisted of a stainless-steel 14.8 g of CO adsorbent. Reactor operating conditions

tubular column of internal diameter 12.4 mm and length for both the adsorbent and adsorbent-free systems are
220 mm, packed with a mixture of catalyst and adsorbent summarised in Table 1.
(or silicon carbide) particles. The reactor was "tted with As mentioned above, conversion enhancement arises
inlet and outlet lines for introducing feed gas (CH and prior to the equilibration of the adsorbent. In this work,

H O), purge as (H /He and H O) and reducing gas (H ). transient operation was imposed by means of step cha-
   
The inlet CH #ow rate was controlled by a Brooks nges in feed concentration. Due to the negligible resi-

5850E mass#ow controller. An HPLC pump was used to dence time of methane in the reactor, even in the presence
supply water to the reactor via a vaporiser (i.e. a heated of adsorbent (i.e. typically less than 0.2 s), the conversion
tubular column packed with silicon carbide particles); enhancement factor at any given time (E(t)) can be quan-
both the reactor and water vaporiser were mounted in ti"ed by the normalised conversion of methane (X  ) in
!&
the presence (ad) and absence (nad) of adsorbent, i.e.

(X ) !(X  )
E(t)" !&  !&   100. (4)
(X  )
!&  
Thus, values of E(t)'0 indicate conversion enhance-
ment in the presence of adsorbent. At steady state, and
for similarly packed reactors (i.e. catalyst mass loading
and distribution), E(t) should approach 0. A typical ex-
perimental cycle involved the following steps: (i) heat-up
of the reactor at atmospheric pressure under a hydrogen
environment to the preset temperature, (ii) water supply
to the system so that the molar ratio of H O to H is
 
approximately the same as the desired H O-to-CH
 
ratio in the reaction step, (iii) pressurisation of the system
to the preset pressure, (iv) switch from H to CH to
 
initiate the reaction step, (v) depressurisation of the unit,
(vi) low-pressure purge of the unit with H and steam,

and (vii) reduction of the catalyst with H at 4803C for
Fig. 1. Schematic representation of the experimental system, MFC 
* mass#ow controller, PC * personal computer. 3 h. Steps (vi) and (vii) were carried out as precautions
3932 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

Table 1
List of experimental

Run type Space time H O/CH Pressure Temperature Particle size


 
(g-cat h/mol-CH ) (kPa) (3C) (mm)

WA/NA 5.37}17.9 3 445.72 455 0.25}0.5
WA/NA 10.7 2}6 445.72 455 0.25}0.5
WA/NA 10.7 3 308}721.5 457.5 0.25}0.5
WA/NA 10.7 3 445.72 428}467.5 0.25}0.5
WA 10.7 3 445.72 459 0.11}0.25

Initial reactor environment: H O : H (He)"H O : CH


   
WA * with adsorbent, NA * no adsorbent.

towards any carbonaceous deposits on the catalyst, and where i denotes CH , H O, H , CO CO, and He. the
   
thus to ensure a reproducible catalyst activity from one semi-empirical correlation proposed by Edwards and
experiment to the next. Reproducibility was subsequently Richardson (1968) was used to estimate the axial disper-
con"rmed through repetition of the experiments. The sion coe$cient D . The catalyst e!ectiveness factor, g ,
X G
measurement of the reactor e%uent concentration pro- was set as unity for the crushed catalyst used in this work;
"les in the reaction step (step (iv)) enabled both transient see Section 4 for details. The reaction kinetic model of Xu
and steady-state reaction kinetics to be tested. Note that and Froment (1989a) can be summarised as
due to non-isothermal nature of the SMR process, the

 
measured wall temperature at the middle point of the k P P
R "  P  P  ! & !- (DEN), (6a)
reactor is taken as the characteristic temperature in this ' P  !& & - K
& '
work.

 
k P P
R "  P  P  ! & !- (DEN), (6b)
'' P  !& & - K
& ''
3. Mathematical modelling

 
k P P
R "  P P  ! & !- (DEN), (6c)
3.1. Governing equations ''' P !- & - K
& '''
DEN"1#K P #K  P  #K  P 
A dynamic model accounting for non-isothermal, !- !- & & !& !&
non-adiabatic, and non-isobaric operation, was de-
#K P /P , (6d)
veloped to describe both the SMR and SE-SMR & - & - &
processes. For the SE-SMR process, the reactions and where R ( j"I!III) denotes the reaction rate of the
adsorption were assumed to take place on the surfaces of H
SMR reactions (1) and (2) and the water}gas shift reac-
the catalyst and adsorbent, respectively. A Langmuir tion (3). The formation rate of component i, r , was
model was used to describe the adsorption equilibria of G
then calculated by using Eqs. (1)}(3) and (6a)}(6d);
CO , and a linear driving force (LDF) model for the for example, r  "!(R #R ), r  "3(R #R )#
 !& ' '' & ' '''
intraparticle mass transfer of the adsorbent; further de- (R #R ). The rate constants k (i"1}3) and the ad-
tails of equilibria and kinetic measurements, and math- '' ''' G
sorption constants K ( j"I}III) are function of temper-
ematical model development under non-reacting H
ature, details of which are given by Xu and Froment
conditions, are given by Ding and Alpay (2000). As men- (1989a).
tioned above, the general reaction kinetic model pro- Pressure distribution in the packed-bed reactor was
posed by Xu and Froment (1989a) was considered in this described by the Ergun equation (Ergun, 1952)
work. Other principal model assumptions can be sum-
marised as: axially dispersed plug #ow, perfect gas behav- *P
"!K u!K u, (7)
iour, no radial concentration or temperature gradients, *z " 4
and a catalyst/adsorbent packing of uniform voidage and
particle size. where K and K are parameters corresponding to the
" 4
Based on the above assumptions, component mass viscous and kinetic pressure loss terms, respectively.
balances for the packed-bed reactor can be written as Semi-empirical relationships for K and K have been
" 4
derived by Ergun (1952) (see also MacDonald, El-Sayed,

 
* * *C *(uC )
(e C #o q )" D G ! G #g o r, Mow & Dullien, 1979).
*t R G @  G *z X *z *z G @  G
For compressible #ow, the energy balance for the
(5) reactor is given by (see, for example, Bird, Stewart &
Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940 3933

Lightfoot, 1960) 3.2. Boundary and initial conditions


*
(o C e #o C ) For the simulation of the reaction step, initial condi-
*t E NE R @ NQ
tions (t"0, z3(0, )) were set to depict a clean and

 
*P * * * isothermal bed, but with a pre-imposed pressure pro"le:
"e # K ! (o uC )
R *t *z X *z *z E NE q "0, (14a)
G

 
*q
!o H G # (R o gH ) *P
@  G *t G @  G 0G "!K u !K u , (14b)
G G *z "  4 
4;
#  ( !). (8) *P/*t"0, (14c)
Dr U
" , (14d)
The bed e!ective conductivity, K , can be expressed as
X 
(see Yagi & Kunii, 1957; Kunii & Smith, 1960; De Wasch C "P /(R ) (14e)
& Froment, 1972; Li & Finlayson, 1977; Wakao G G 
& Kaguei, 1982) At the onset of the reaction step (i.e. the point at which
CH is supplied to the reactor), the following boundary
K K 
X " X #a(Pr) (Re ), (9) conditions were used in the simulations:
k k N (i) Reactor entrance (z"0)
E E
where K is the static e!ective conductivity accounting
X !D (*C /*z)"u(C !C ),
X G DG G
(15a)
for the e!ects of conduction and radiation, see Kunii and
Smith (1960) and Ding and Alpay (2000). !K (*/*z)"uo C ( !), (15b)
X E NE D
For a bed packed with spherical particles, the wall}bed u"Q /A. (15c)
heat transfer coe$cient, ; , in Eq. (8) is given by Li and D

Finlayson (1977) as (ii) Reactor outlet (z")

  *C /*z"0,
; D 6d (15d)
 P "2.03Re  exp ! N G
k N D
E P */*z,0, (15e)
(Re "20}7600, d /D "0.05}0.3). (10)
N N P P"P . (15f )
*
Eq. (10) is not applicable to very low feed #ow rates, as
Q in Eq. (15c) is the feed volumetric #ow rate measured
; should approach a "xed value where Re P0. How- D
 N under the local conditions.
ever, De Wasch and Froment (1972) give the following
correlation for ; at very low Re :
 N
K
; " N "6.15 X (11) Table 2
 0C  D Parameters (constants) used in the simulations
P
which can be linearly combined with Eq. (10) to approx- Parameters (constants) Value Unit
imate ; over the entire range of Re relevant to this
 N b  SE-SMR: 2.36;10\; SMR:0 Pa\
work. !-
The Langmuir model for CO adsorption can be d (average) 1.8;10\, 3.57;10\ m
N
 D 1.27;10\ m
written as P
C 42 J/mol K
NE
m b P C 850 J/kg K
NQ
qH  " !- !- !- . (12) e 0.35
!-
*
1#b  P 
!- !- k
E
0.09 J/m K
k 0.3 J/m K
Other reaction components were considered to be non- Q
H SE-SMR: !17000, SMR: 0 J/mol
adsorbing. The LDF model is given by  !-
(reactor length) 0.22 m
m  SE-SMR: 0.65, SMR: 0 mol/kg
*q !-
G "k (qH!q ), (13) b

0.95 *
*t G G G e 0.47
@
*
e 0.64
R
*
where k is the e!ective mass transfer coe$cient. Note that U 0.2 *
*q /*t"0 for non-adsorbing species in the SE-SMR pro- c 0.667
G
*
cess, and *q /*t"0 for all species in the SMR process. As j 1.0
Q
*
G k 2.87;10\ Pa s
mentioned above, Langmuir and LDF parameters have E
o SE-SMR: 609.3, SMR:0 kg/m
been previously measured for the CO }H O-hydro- @ 
  o
@ 
139.0 kg/m
talcite system; see Ding and Alpay (2000).
3934 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

Eqs. (5)}(15) were simultaneously solved in the the SMR run (&1.5 min). However, since the experi-
gPROMS modelling environment (Process Systems En- mental conditions are nearly identical, the steady-state
terprises Ltd.). The spatial discretisation method of or- concentrations of CH , H , and CO are approximately
  
thogonal collocation on "nite elements was employed. equal, indicating no strong direct catalyst/adsorbent in-
Second-order collocation on 100 elements was found to teraction. The measured e%uent CO concentrations were
give a converged solution in which component balance found to be very low ((0.2% (v/v)) for all these runs.
errors (associated with the numerical integration) did not The longer transient time of the SE-SMR run is mainly
exceed 1%. A summary of parameters (constants) used due to the interaction between the adsorption and reac-
for the simulations is given in Table 2. tion processes, i.e. the retention of CO and its sub-

sequent in#uence on local reaction kinetics and
equilibria. Adsorption of CO not only decreases the

4. Results and discussion gas-phase CO partial pressure, but also releases some

heat, both of which are favourable for the reforming
Typical reactor e%uent concentration pro"les mea- reactions (1) and (2). On the other hand, an increase in the
sured in the SMR and SE-SMR experiments under SMR reaction rate will result in a decrease in the reactor
similar operating conditions (4503C, 445.7 kPa, H O/ temperature due to the endothermic nature of the re-

CH "6) are shown in Fig. 2(a) and (b), respectively. It forming reactions; see below for further discussions on

can be seen that in the presence of CO adsorbent, the the reactor temperature pro"les in the presence and ab-

transient period (&5 min) is much longer than that of sence of adsorbent. Also shown in Fig. 2(a) and (b) are the
simulated e%uent concentration pro"les. The excellent
agreement between the experiments and simulations for
both the SMR and SE-SMR processes indicates that the
rate expressions proposed by Xu and Froment (1989a)
are suitable for both the transient and steady-state peri-
ods of operation, even in the presence of adsorbent. This
suggests that the microkinetic dynamics of reaction are
relatively fast, and that the physically admixed nature of
catalyst and adsorbent precludes any local e!ect of ad-
sorption on reaction intermediates, and hence on mo-
lecular kinetic steps.
Fig. 3 shows typical measured temporal pro"les of
methane conversion and CO exit concentration. Corre-

sponding measured and calculated enhancement factors
are shown in Fig. 4. It can be seen that there exists a peak
in conversion enhancement, which disappears on the

Fig. 2. E%uent concentration (mole fraction) pro"les on a water-free Fig. 3. Comparison of methane conversion pro"les and CO break-

basis: comparison between experiments and simulations (reactor through curve (reactor pressure"445.7 kPa, temperature"
pressure"445.7 kPa, temperature"450$23C, feed #ow rate of 450$23C, feed #ow rate of CH "250 ml/min (STP), molar ratio

CH "250 ml/min (STP), molar ratio of H O to CH "6, cata- of H O to CH "6, catalyst/adsorbent (or carbide) particle
    
lyst/adsorbent (or carbide) particle size"0.25}0.5 mm). size"0.25}0.5 mm).
Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940 3935

strated in Fig. 5, in which calculated axial temperature


pro"les for the SMR and SE-SMR processes are shown
at di!erent times from the onset of the reaction step.
The complex interaction of adsorption and reaction
precludes any intuitive design of forced-periodic adsor-
ptive reactors. However, some insight into favourable
operating regimes can be gained by considering the speci-
"c e!ects of key operating and design parameters on the
enhancement factor and the overall yield of product. In
this work, the following parameters have been considered
in turn: space time (g-cat h/mol-CH ), H O-to-CH feed
  
molar ratio, temperature, pressure and adsorbent/cata-
lyst particle size. Fig. 6 shows the e!ect of space time on
these enhancement factor; other operating conditions are
set as 445.7 kPa, 4553C, H O : CH "3 mol/mol. Under
 
Fig. 4. Conversion enhancement: comparison between experiments
these conditions, higher conversions are achieved, and
and simulations (reactor pressure"445.7 kPa, temperature" approach the equilibrium conversion of approximately
450$23C, feed #ow rate of CH "250 ml/min (STP), molar ratio 20% at 4553C. The bene"cial e!ects of local CO adsorp-
 
of H O to CH "6, catalyst/adsorbent (or carbide) particle
  tion are realised as the reaction equilibrium conversion is
size"0.25}0.5 mm, measured steady-state conversion"23.98%). approached. For a relatively low space time, the reaction
kinetics dictate a low production rate of CO , and adsor-

bent equilibration (i.e. CO breakthrough) occurs rela-

tively rapidly. Where conversions are so low that the
onset of the breakthrough of CO . A slight negative backward reaction rates are small, the advantages of

enhancement (of 1}2%) after the sorption-enhancement CO removal from the reaction zone will be minimised.

peak is also observed in both experiments and simula- Thus, a low space time will generally lead to poor conver-
tions. This negative enhancement may be attributed to sions, for which the advantages of in situ separation will
the propagation of a weak thermal (cold) wave prior the not be realised due to the rapid propagation velocity of
approach to the steady state. Thus, whilst CO adsorp- the CO concentration front, and the diminished inhibi-
 
tion enhances overall reaction rates through favourable tion of CO on the net reaction rates. It is important to

equilibria shifts, as the adsorbent becomes equilibrated, note however, that in this case, reactor space time (i.e.
a slightly lower temperature pro"le in the reactor results feed #ow rate) has a signi"cant in#uence on heat transfer
in a small (and temporary) reduction in the conversion rates and subsequently on the reactor temperature pro-
relative to the catalyst-only reactor. This e!ect is demon- "le. In speci"c, a high space time will lead to poor heat

Fig. 5. Temperature distribution: Comparison between SMR and SE-SMR (4503C, 445.7 kPa, feed #ow rate of CH "250 ml/min (STP) H O to
 
CH "6, catalyst/adsorbent (or carbide) particle size"0.25}0.5 mm).

3936 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

tion are signi"cant, and thus greater potential for separ-


ation-enhanced conversion exists. Conversely, at high
partial pressures of water, a relatively high equilibrium
conversion exists, and thus the signi"cance of CO ad-

sorption on conversion enhancement decreases. Accord-
ing to reactions (1) and (2), the ideal ratio of H O to CH
 
is between 1 and 2, depending on the desired "nal reac-
tion products. For the production of H , this ratio

should be 2. However, a H O : CH ratio greater than
 
this stoichiometric value is often required to avoid cata-
lyst deactivation due to carbon deposition. Carbon de-
position principally occurs at high temperatures through
the decomposition of methane (CH "C#2H ,
 
*H "75 kJ/mol), and to a smaller extent by the

Boudouard reaction (2CO"C#CO , *H "
Fig. 6. E!ect of space time on the conversion enhancement (reactor  
pressure"445.7 kPa, reactor temperature"455$53C, molar raito of !173 kJ/mol) which is thermodynamically favoured at
H O to CH "3, catalyst/adsorbent size"0.25}0.5 mm, measured
  low temperatures (see Rostrup-Nielsen, 1984). Although
steady-state conversions"14.50%, 17.0%, 14.90%, 15.2%, 16.50%, sorption-enhanced steam}methane reforming takes place
15.75% and 18.10% for space times"17.89, 13.50, 10.7, 8.96, 7.68, 6.72
and 5.37 g-cat h/mol-CH correspondingly).
at a considerably lower temperature than that used in
 the conventional reactor, some carbon formation via the
Boudouard reaction cannot be avoided. Furthermore,
the adsorption of CO is likely to shift the decomposition

of CO to further carbon formation. As a consequence,
whilst it is feasible to operate the SE-SMR at lower
H O/CH ratios, this again will be dictated by the level
 
of catalyst coking. However, in this work, no catalyst
deactivation was apparent, possibly due to relatively
short times-on-stream of catalyst (i.e. &200 h), and the
very low CO partial pressures measured in all experi-
ments.
The e!ect of operating temperature on the enhance-
ment factor is shown in Fig. 8; other operating conditions
are set as 445.7 kPa, H O : CH "3 mol/mol, 10.7 g-cat
 
h/mol-CH . As temperature decreases, the steady-stage

yield of CO decreases, i.e. CH conversions decrease
 
Fig. 7. E!ect of feed molar ratio of H O to CH on the conversion from 16.3 to 12.7% as the temperature is reduced from
 
enhancement (reactor pressure"445.7 kPa, reactor temperature" 467.5 to 4283C. As for the case of low reactor space times,
455$43C, space time"10.7 g-cat h/mol-CH , catalyst/adsorbent
 the bene"ts of in situ separation are negligible, even
size"0.25}0.5 mm, measured steady-state conversions"13.0%, though the adsorption a$nity of CO is relatively high.
14.87%, 16.65%, 20.50% and 23.70% for H O : CH "2, 3, 4, 5 and 
  However, at relatively high temperatures of operation,
6 correspondingly).
a high yield of CO is achieved, but further enhancement

of conversion is di$cult due to the relatively low adsorp-
tion a$nity. Thus, as indicated in Fig. 8, an optimal
transfer, and thus a cooler reactor. Hence, the decrease in temperature exists for maximum enhancement. In order
the steady-state conversion with increasing space time to achieve an equivalent product yield at a temperature
(see "gure caption of Fig. 6) is to expected, although lower than that of an adsorbent-free (conventional) reac-
equilibrium constrains dominate for relatively high space tor, careful selection of the adsorbent and catalyst is
times. needed such that the kinetics of reaction and the capacity
The e!ect of the feed molar ratio of H O to CH on of the adsorbent are in accord. When the adsorbent needs
 
the enhancement factor is shown in Fig. 7; other operat- to be operated at a lower temperature, it may be possible
ing conditions are set as 445.7 kPa, 4553C, 10.7 g-cat to have a two-stage reactor consisting of a high-temper-
h/mol-CH . Methane conversion enhancement is fa- ature catalyst-only stage, and a lower temperature cata-

voured by low H O : CH ratios, i.e. at relatively low lyst#adsorbent stage. Further discussion on analytical
 
partial pressures of feed H O. For low H O partial criteria for favourable catalyst and adsorbent combina-
 
pressures, the equilibrium conversion is relatively low, i.e. tions is given by Sheikh, Kershenbaum and Alpay
the backward reactions for methane and water produc- (1998).
Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940 3937

Fig. 8. E!ect of reactor temperature on the conversion enhancement Fig. 10. E!ect of particle sizes of catalyst and adsorbent on the transi-
(reactor pressure"445.7 kPa, molar ratio of H O to CH "3, space
  ent kinetics (reaction temperature"459$13C, reactor pressure"
time"10.7 g-cat h/mol-CH , catalyst/adsorbent size"0.25}0.5 mm,
 445.7 kPa, space time"10.7 g-cat h/mol-CH , H O : CH "3).
  
measured steady-state conversions"12.70%, 14.15%, 14.87% and
16.25% for reactor temperatures"428, 445, 458 and 467.53C corre-
spondingly).
Fig. 10 shows the e!ect of particle size of catalyst and
adsorbent on the methane conversion; other operating
conditions are set as 4593C, 445.7 kPa, 10.7 g-cat h/mol-
CH , H O : CH "3 mol/mol. It can be seen that the
  
methane conversion for experiments with 0.11}0.25 mm
particles can be up to 20% higher than that with
0.25}0.5 mm particles in the transient period. However,
this di!erence diminishes when the reactor approaches
the steady stage. Given that the e!ect of particle size is
only apparent during the transient stage, it is concluded
that the above results are likely to be associated with the
mass transfer resistances inside the adsorbent particles,
i.e. the e!ectiveness factor of the catalyst approaches
unity. As mentioned above, a reduction in the #ow rate of
reactant, or an increase in reactor length, will reduce the
Fig. 9. E!ect of reactor pressure on the conversion enhancement (reac-
signi"cance of the adsorption and desorption kinetics.
tor temperature"445.5$13C, molar ratio of H O to CH "3, space As mentioned in Section 2, the transient experiments in
 
time"10.7 g-cat h/mol-CH , catalyst/adsorbent size"0.25}0.5 mm, this work consisted of a step change from a hydro-

measured steady-state conversions"17.8%, 14.87%, 12.13% and gen}steam feed to a methane}steam feed. Experiments
10.5% for reactor pressures"308, 445.72, 583.68 and 721.5 kPa corre- have also been performed in which H is replaced with
spondingly). 
He prior to the onset of the reaction step (H O : He

&3 mol/mol), i.e. a reactor environment free of H prior

to the introduction of methane and steam. A comparison
Fig. 9 shows the e!ect of reactor pressure on the of the CH conversion for the two cases is given in

enhancement factor; other operating conditions are Fig. 11; operating conditions for both cases are set as
set as 457.53C, 10.7 g-cat h/mol-CH , H O : CH " 4453C, 445.7 kPa, 6.27 g-cat h/mol-CH , H O : CH "
     
3 mol/mol. It can be seen that the enhancement factor 3 mol/mol. It can be seen that at early times, the methane
increases with increasing reactor pressure. Higher pres- conversion is higher under an initial He}H O environ-

sures will favour greater CO adsorption, and will result ment. This inhibitive e!ect of H on the reaction kinetics
 
in a greater potential for reaction enhancement due to is well documented; see, for example. Bodrov, Apel'baum
a reduction in the equilibrium conversion of CH . Thus, and Tempkin (1967, 1968), Phung Quach and Rouleau

although SMR reactions (1) and (2) are favoured by low (1975) and Ross and Steel (1973). For example, Phung
pressures, the above result suggests that the sorption- Quach and Rouleau (1975) studied the kinetics of SMR
enhanced SMR reactor can be operated at higher pres- over Ni/a-Alumina catalyst in a continuous stirred tank
sures to achieve the same product yield. This, of course, is reactor at 350}4503C; a Langmuir}Hinshelwood}
particularly bene"cial where the direct high-pressure de- Hougen}Watson type rate equation was found to "t their
livery of H is of economic advantage. kinetic data satisfactorily, and the equation obviously

3938 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

adsorption, was found to accurately predict the elution


pro"les of the reaction species over an admixture of Ni
catalyst and hydrotalcite adsorbent. Kinetic and equilib-
rium parameters for adsorption and reaction were em-
ployed from previous (independent) measurements; no
model "tting parameters were needed. Comparison of the
elution pro"les of methane for the adsorbent-free and
adsorbent-present cases led to a measure of the reaction
(conversion) enhancement. Thus, direct evidence for the
increase in net reaction rates due to in situ separation
was attained. Furthermore, investigations on the in#u-
ence of some key operating parameters on the degree of
enhancement led to the following conclusions: (i) a high
Fig. 11. E!ect of reactor initial environment on the transient kinetics reactor space time is favourable for minimising the e!ects
(reaction temperature"4453C, reactor pressure"445.7 kPa, space of adsorbent intraparticle mass transfer resistances and,
time"6.27 g-cat h/mol-CH , H O : CH "3, catalyst/adsorbent size
  
of course, for overcoming kinetic limitation to reaction,
"0.25}0.5 mm). (ii) for a given conversion of CH or yield of CO , the
 
SE-SMR process enables operation with lower steam-
to-methane ratios, and/or higher operating pressures,
and (iii) under the conditions of SE-SMR operation,
implied the inhibitive e!ect of the presence of H in the negligible deactivation of the catalyst occurs. Currently,

feed. Nevertheless, the in#uence of the initial presence of separation-enhanced reaction studies in a semi-technical
H on the dynamics of conversions is short-lived, i.e. less scale reactor are under way, in which the typical operat-

than 1 min. This suggests that under pressure and ing steps of a pressure swing based adsorption unit are
concentration swing type operations, and for the being imposed. The work will enable the further testing of
temperatures considered in this work, adsorbent regen- kinetic and process models under cyclic operation.
eration can be achieved under a hydrogen-free environ-
ment without adverse e!ect on the catalyst activity, and
thus the dynamics of the in situ separation and reaction Notation
step.
Having validated transient reaction kinetics in the a constant in the empirical correlation for the ef-
presence of adsorbent, future work will involve model- fective thermal conductivity, dimensionless
based optimisation methods for the selection and design A cross-sectional area of the reactor, m
of adsorptive reactor con"gurations; see, for example, b Langmuir model constant for component, i,
G
Yongsunthon and Alpay (1999). In particular, considera- Pa\
tion will be given to process design based on distinct C gas-phase concentration of component i in the
DG
objectives, such as: (i) the lowest energy utilisation for feed, mol/m
a speci"ed product yield, and (ii) the highest concentra- C molar concentration of gas-phase component i,
G
tion (bulk separation) of hydrogen for a speci"ed conver- mol/m
sion. Ultimately, the stability of the adsorbent under C gas-phase heat capacity, J/mol K
NE
cyclic reactor conditions, i.e. under constant temporal C solid-phase heat capacity, J/kg K
NQ
and spatial variations of gas composition and temper- d particle diameter, m
N
ature, will govern the commercial feasibility. Thus, fur- D inner diameter of the reactor, m
P
ther experimental work will be required, at least at D axial dispersion coe$cient, m/s
X
a semi-technical scale, to address such issues under iden- E(t) conversion enhancement, %
ti"ed process con"gurations. H adsorption heat of component i (on adsorbent
G
surface), J/mol
H reaction heat of reaction i, J/mol
0G
5. Conclusions k LDF mass transfer coe$cient, s\
k gas-phase thermal conductivity, J/m K
E
The steady-state kinetic model of Xu and Froment k rate constant of reaction i, i"1, 2; mol Pa /
G
(1989a) for steam}methane reforming has been shown to kg-cat s, i"3: mol/kg-cat s Pa
be applicable to transient reactor operation, both in the k solid-phase thermal conductivity, J/m K
Q
presence and absence of adsorbent. In particular, a reac- K Ergun equation coe$cient, N s/m
"
tor model accounting for non-isothermal, non-adiabatic K equilibrium constant of reactions (1)}(3), i"I,
G
and non-isobaric operation, and mass transfer limited II: Pa, i"III: dimensionless
Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940 3939

K adsorption constant for component j (on cata- Alpay, E., Chatsiriwech, D., & Kershenbaum, L. S. (1995). Combined
H reaction and separation in pressure swing processes. Chemical En-
lyst surface), j"CO, H , CH : Pa\, j"H O:
   gineering Science, 49, 5845}5864.
dimensionless
Armor, J. N. (1995). Membrane catalysis: What is it now, what needs to
K Ergun equation coe$cient, N s/m
4 be done? Catalysis Today, 25, 199}207.
K static e!ective thermal conductivity, J/m K
X Bird, R. B., Stewart, W. E., & Lightfoot, E. N. (1960). Transport phe-
K e!ective thermal conductivity, J/m K nomena. New York: Wiley.
X Bodrov, N. M., Apel'baum, L. O., & Tempkin, M. I. (1967). Kinetics of
reactor length, m
m Langmuir model constant for component i, the reaction of methane with water vapour catalysed by nickel on
G a porous carrier. Kinetika L Kataliz, 8, 821}828.
mol/kg Bodrov, N. M., Apel'baum, L. O., & Tempkin, M. I. (1968). Kinetics of
P pressure, Pa the reaction of methane with steam on the surface of nickel at
P partial pressure of gas-phase component i, Pa 400}6003C. Kinetika L Kataliz, 9, 1065}1071.
G Brun-Tsekhovoi, A. R., Zadorin, A. N., Katsobashvili, Ya. R., & Kou-
P outlet pressure, Pa
* rdyumov, S. S. (1986). The process of catalytic steam-reforming of
Pr Prandtl number, dimensionless
hydrocarbons in the presence of carbon dioxide acceptor. In Pro-
q solid-phase concentration (average over an ad-
G ceedings of the World hydrogen energy conference, vol. 2 (pp.
sorbent particle), mol/kg 885}900). New York: Pergamon Press.
qH equilibrium solid-phase concentration, Carvil, B. T., Hufton, J. R., & Sircar, S. (1996). Sorption-enhanced
G reaction process. A.I.Ch.E. Journal, 42, 2765}2772.
mol/kg
Q volumetric #ow rate of the feed gas (inlet), De Deken, J. C., Deven, E. F., & Froment, G. F. (1982). Steam reform-
D ing of natural gas. Chemical Reaction Engineering ACS Symposium
m/s Series, 196, 101}117.
r formation rate of component i, mol/kg-cat s
G De Wash, A. P., & Froment, G. F. (1972). Heat transfer in packed bed.
R universal gas constant, J/mol K Chemical Engineering Science, 27, 567}576.
Re particle Reynolds number, dimensionless Ding, Y., Alpay, E. (2000). Equilibria and kinetics of high temperature
N CO adsorption on hydrotalcite adsorbent, in press.
R rate of reaction j ( j"1}3), mol/kg-cat s 
H Edwards, M. F., & Richardson, J. F. (1968). Gas dispersion in packed
t time, s beds. Chemical Engineering Science, 23, 109}123.
temperature, K Elnashaie, S. S. B. H., Adris, A. M., Al-Ubaid, A. S., & Soliman, M. A.
feed gas temperature (inlet), K (1990). On the non-monotonic behaviour of methane steam reform-
D ing kinetics. Chemical Engineering Science, 45, 491}501.
initial bed temperature, reference temperature,
 Ergun, S. (1952). Fluid #ow through packed columns. Chemical Engin-
K
eering Progress, 48, 89}94.
wall temperature, K
U Goto, S., Tagawa, T., & Oomiya, T. (1993). Dehydrogen of cyclohexane
u super"cial velocity, m/s in a PSA reactor using hydrogen storage alloy. Chemical Engineer-
u initial super"cial velocity, m/s ing Essays (Japan), 19, 978}983.
 Han, C., & Harrison, D. P. (1994). Simultaneous shift reaction and
; overall bed}wall transfer coe$cient, J/m K
 carbon dioxide separation for the direct production of hydrogen.
X  conversion of CH , %
!&  Chemical Engineering Science, 49, 5875}5883.
Hufton, J. R., Mayorga, S., & Sircar, S. (1999). Sorption-enhanced reac-
Greek letters tion process for hydrogen production. A.I.Ch.E. Journal, 45, 248}256.
Itoh, N., Xu, W. C., & Sathe, A. M. (1993). Capability of permeate
e total voidage of the adsorbent bed, hydrogen through palladium-based membranes for acetylene
R hydrogenation. Industrial and Engineering Chemistry Research, 32,
dimensionless
2614}2619.
g catalyst e!ectiveness factor, dimensionless
G Kunii, D., & Smith, J. M. (1960). Heat transfer characteristics of porous
k gas-phase viscosity, Pa s
E rocks. A.I.Ch.E. Journal, 6, 71}78.
o bulk density of the adsorbent, kg/m Li, Chi-hsiung, & Finlayson, B. A. (1977). Heat transfer in packed beds
@ 
o bulk density of the catalyst, kg/m * A reevaluation. Chemical Engineering Science, 32, 1055}1066.
@ 
o gas-phase density, mol/m MacDonald, I. F., El-Sayed, M. S., Mow, K., & Dullien, F. A. L. (1979).
E Flow through porous media * the Ergun equation revisited. Indus-
trial Engineering Chemistry Fundamentals, 18, 199}207.
Marek, L. F., & Hahn, D. A. (1932). Catalytic oxidation of organic
References compounds in the vapour phase: Chemical Catalog Co.
Munsted, P., & Grabke, H. J. (1981). Kinetics of the steam reforming of
Adris, A. M., Lim, C. J., & Grace, J. R. (1994). The #uidised-bed methane with nickel, iron and iron}nickel alloys as catalysts. Jour-
membrane reactor (FBMR) system: A pilot scale experimental nal of Catalysis, 72, 279}287.
study. Chemical Engineering Science, 49, 5833}5843. Phung Quach, T. Q., & Rouleau, D. (1975). Kinetics of the methane-
Adris, A. M., Lim, C. J., & Grace, J. R. (1997). The #uidised-bed steam reaction over nickel catalyst in a continuous stirred tank
membrane reactor for steam methane reforming: Model veri"cation reactor. Journal of Applied Chemistry of Biotechnology, 25, 445}459.
and parametric study. Chemical Engineering Science, 52, 1609}1622. Ross, R. H., & Steel, M. C. F. (1973). Mechanism of the steam reforming
Akers, W. W., & Camp, D. P. (1955). Kinetics of the methane}steam of methane over a coprecipitated nickel}alumina catalyst. Journal of
reaction. A.I.Ch.E. Journal, 1, 471}475. the Chemical Society, Faraday Transactions I, 69, 11}21.
Allen, D. W. (1975). Kinetics of the methane}steam reaction. Industrial Rostrup-Nielsen, J. R. (1984). Catalytic steam reforming. In J. R. Ander-
and Engineering Chemistry Process Design and Development, 14, son, & M. Boudart, Catalysis science and technology, vol. 4. Berlin:
256}259. Springer.
3940 Y. Ding, E. Alpay / Chemical Engineering Science 55 (2000) 3929}3940

Sabatier, P. (1922). Catalysis in organic chemistry. New York: Van Vaporciyan, G. G., & Kadlec, R. H. (1989). Periodic separating reactors:
Nostrand. Experiments and theory. A.I.Ch.E. Journal, 35, 831}841.
Schnell, C. R. (1970). Reaction mechanism of steam reforming. Journal Wakao, N., & Kaguei, S. (1982). Heat and mass transfer in packed bed.
of the Chemical Society, B, 158}163. New York: Gordon and Breach.
Sheikh, J., Kershenbaum, L. S., & Alpay, E. (1998). Analytical basis for Wu, J. C. S., Gerdes, T. E., Pszczolowski, J. L., Bhave, R. R., & Liu, P. K.
separation enhanced reaction in continuous #ow processes. Chem- T. (1990). Dehydrogenation of ethylbenzene to styren using com-
ical Engineering Science, 53, 2933}2939. mercial ceramic membranes as reactors. Separation Science Techno-
Sun, Y. M., & Khang, S. J. (1988). Catalytic membrane for simultaneous logy, 25, 1489}1510.
chemical reaction and separation applied to a dehydrogenation Xu, J., & Froment, G. F. (1989a). Methane steam reforming, methana-
reaction. Industrial and Engineering Chemistry, Research, 27, tion and water}gas shift: I. Intrinsic kinetics. A.I.Ch.E. Journal, 35,
1136}1142. 88}96.
Tsotsis, T. T., Champagnie, A. M., Vasileiadis, S. P., Zraka, Z. D., Xu, J., & Froment, G. F. (1989b). Methane steam reforming: II. Diwu-
& Minet, R. G. (1992). Packed bed catalytic membrane reactor. sional limitations and reactor simulation. A.I.Ch.E. Journal, 35,
Chemical Engineering Science, 47, 2903}2908. 97}103.
Uemiya, S., Sato, N., Ando, H., & Kikuchi, E. (1991). The water gas shift Yagi, S., & Kunii, D. (1957). Studies on e!ective conductivities in
reaction assisted by a palladium membrane reactor. Industrial and packed beds. A.I.Ch.E. Journal, 3, 373}381.
Engineering Chemistry, Research, 30, 585}589. Yongsunthon, L., & Alpay, E. (1999). Design of periodic adsorptive
Van Hook, J. P. (1980). Methane}steam reforming. Catalysis Reviews of reactors of the optimal integration of reaction, separation and
Science Engineering, 21, 1}51. heat-exchange. Chemical Engineering Science, 54, 2647.

View publication stats

Anda mungkin juga menyukai