Académique Documents
Professionnel Documents
Culture Documents
For an endothermic reaction such as steam methane reforming (SMR), temperature gradients in the packed
bed can play a significant role in process performance. A reactor design strategy that considers the combination
of tubular packed-bed microreactor (TPBM) operation and catalyst dilution by adsorbent is proposed for
enhancement of the SMR process. Simulation studies have been performed with the proposed strategy to
quantitatively study the effect of heat transfer limitation and process performance in terms of methane conversion
and hydrogen purity. This is carried out using a dynamic 2D pseudohomogeneous model incorporating kinetic
and reactor/reaction mechanisms describing the process. The results of these studies show that the temperature
gradient in TPBM is better controlled with less catalyst loading in comparison with traditionally used larger
diameter reactors. It is seen that improved heat transfer characteristics of the TPBM combined with the effects
of catalyst dilution and adsorption complement each other, and enhanced performance in terms of methane
conversion and hydrogen purity is observed. The findings based on this study give an insight into the performance
improvement of the SMR process and show that cost reduction and process miniaturization of fuel processors
for fuel cell applications are possible.
required. Recently, a new generalized strategy using the principle not only as a diluent for the catalyst but also at the same time
of temperature-induced equilibrium shift by controlling the as an adsorbent for CO2. Steam methane reforming (SMR)
subsection wall temperature has been developed by Xiu et al.23,24 reactions are equilibrium-controlled, and selective removal of
This strategy is an efficient way for controlling the temperature CO2 from the reaction zone increases methane conversion
along the reactor. However, other subsection-controlling pa- according to Le Chatelier’s principle. Hydrotalcite adsorbent,
rameters such as number of subsections, the subsection packing which selectively removes CO2 from the reaction zone and is
ratio of adsorbent and catalyst, and temperature in each stable under wet gas and high temperature, is particularly known
subsection need to be well-optimized. to be suitable for the SMR process.36,37
The heat transfer rates in the packed-bed reactor are known Temperature gradients in SMR in a packed bed can adversely
to depend weakly on internal field intraparticle temperature affect process performance, and this article focuses on the use
gradients in comparison with external field interparticle gradi- of tubular packed-bed microreactors (TPBM) and employs a
ents.25 The external field interparticle temperature gradients can catalyst dilution strategy for performance enhancement of SMR
be minimized by decreasing the reactor diameter, by diluting for controlling the temperature gradient in the bed. As discussed
the solids, or by diluting the reactant. Due to volume constraints, above, these two aspects of reactor design (viz., microreactor
reactant dilution may not be feasible in commercial reformers. operation and catalyst dilution by solids) have generally been
Therefore, it has been suggested that controlling external field studied separately. In the present work, we consider a strategy
interparticle heat transfer in a packed-bed reactor by decreasing where the above two features can take place simultaneously.
the reactor diameter and by solids dilution may prove to be For the above objective, we have studied the SMR process
advantageous.26,27 performance in TPBM for diluted and undiluted conditions using
Tubular packed-bed reactors with small diameters have the a dynamic 2D pseudohomogeneous model. The model incor-
advantage of providing higher surface-to-volume ratios. This porates mass and energy balances with nonlinear adsorption
results in decreasing mass transport and heat transport resistances isotherm together with a linear-driving force (LDF) model for
and makes them attractive for achieving efficient thermal and adsorption rate and this is coupled with the SMR reactions. The
mass transfer rates. Karim et al.27 applied this concept for results of simulation are presented and discussed in detail in
methanol steam reforming on a commercial CuO/ZnO/Al2O3 section 3 and show the improvement in SMR performance
catalyst and showed its potential in the area of fuel processing pertaining to effects of varying reactor diameter, extent of
for fuel cell applications. Their study points out that there exist catalyst dilution by adsorbent, and solids packing density.
heat transfer limitations in the bed and that the majority of
activity measurements that have been reported for methanol 2. 2D Mathematical Model for the SMR Process
steam reforming could be influenced by temperature gradients
The key chemical reactions of the SMR process are given
within the reactor.
by:
Catalyst dilution by solids increases the wall area to catalyst
volume for heat transfer. It establishes specific temperature CH4 + H2O S CO + 3H2, ∆H298 ) 206.2 kJ/mol (1)
profiles (both in the radial and axial directions) in the packed-
bed reactor and thereby improves the isothermicity while CH4 + 2H2O S CO2 + 4H2, ∆H298 ) 164.9 kJ/mol (2)
simultaneously suppressing other effects that adversely affect
performance (e.g., nature of axial dispersion, channeling loss). CO + H2O S CO2 + H2, ∆H298 ) -41.1 kJ/mol (3)
The strategy of dilution by solids has been studied to improve
packed-bed reactor performance and is now being increasingly The mathematical model used to describe the SMR reactor
applied.17-19,28-31 The cost of catalyst forms a major component with adsorbent dilution is a dynamic 2D pseudohomogeneous
in reactor operation, and dilution can help in this aspect. In model that considers the nonisothermal, nonadiabatic, and
addition to packed-bed catalytic reactors, catalyst dilution by nonisobaric nature of operation. The model assumptions used
solids has been applied to fluidized-bed reactors32,33 and trickle- are:
bed reactors.34,35 Care should be taken, however, to prevent 1. Axial dispersed plug flow prevails in the bed.
bypass of flow inside a reactor while employing the dilution 2. Mass dispersion in the axial direction is considered.
strategy.29-31 Catalyst dilution studies reported in the literature 3. Mass dispersion in the radial direction is assumed to be
use inert solid-catalyst mixtures. In the present study, we negligible.
assume that the catalyst is diluted with an adsorbent that acts 4. The system is nonisothermal. Thermal dispersion in both
axial and radial directions is considered.
(25) Kirillov, V. A.; Kuzin, N. A.; Mescheryakov, V. D.; Droboshevich, 5. The reaction kinetic model employed is that proposed by
V. I. Chem. Eng. Sci. 2001, 56, 381-386. Xu and Froment.38 This kinetic model has been shown to be
(26) Mears, D. E. J. Catal. 1971, 20, 127-131. useful to study the SMR process.39-41
(27) Karim, A.; Bravo, J.; Datye, A. Appl. Catal., A 2005, 282, 101-
109.
6. Volumetric change of flow due to adsorption and reaction
(28) Taniewski, M.; Lacjowicz, A.; Skutil, K.; Czechowicz, D.; Ferreria, is taken into account in the overall material balance.
R. M. Q. Chem. Eng. Sci. 1996, 55, 4271-4278. 7. The gas is assumed to be an ideal gas.
(29) Hwang, S.; Smith, R. Chem. Eng. Sci. 2004, 59, 4229-4243. 8. The adsorbent and catalyst particles are the same size and
(30) Berger, R. J.; Pérez-Ramı́rez, J.; Kapteijn, F.; Moulijn, J. A. Chem.
Eng. J. 2002, 90, 173-183. spherical in shape.
(31) van den Bleek, C. M.; van der Wiele, K.; van de Berg, P. J. Chem.
Eng. Sci. 1969, 24, 681-694. (36) Zou, Y.; Mata, V.; Rodrigues, A. E. Ind. Eng. Chem. Res. 2001,
(32) Irani, R. K.; Kulkarni, B. D.; Doraiswamy, L. K. Ind. Eng. Chem. 40, 204-209.
Process. Des. DeV. 1979, 18, 648-655. (37) Ding, Y.; Alpay, E. Chem. Eng. Sci. 2000, 55, 3461-3474.
(33) Irani, R. K.; Kulkarni, B. D.; Doraiswamy, L. K.; Hussain, S. Z. (38) Xu, J.; Froment, G. F. AIChE J. 1989, 35, 88-96.
Ind. Eng. Chem. Process Des. DeV. 1982, 21, 192-195. (39) Ding, Y.; Alpay, E. Chem. Eng. Sci. 2000, 55, 3929-3940.
(34) Van Klinken, J.; Van Dongen, R. H. Chem. Eng. Sci. 1980, 35, (40) Xiu, G. H.; Soarse, J. L.; Rodrigues, A. E. AIChE J. 2002, 48, 2817-
59-66. 2832.
(35) Al-Dahnan, M. H.; Dudukovic, M. P. AIChE J. 1996, 42, 2594- (41) Xiu, G. H.; Li, P.; Rodrigues, A. E. Chem. Eng. J. 2003, 95, 83-
2606. 93.
Performance Enhancement of SMR Energy & Fuels, Vol. 20, No. 2, 2006 465
( )
10. The effect of changing bed voidage with variation ∂t ∂z
n ∂qji n III
∑ ∑ ∑υijηjRj∆HRj )
of reactor-to-particle diameter ratio43 on methane conver-
sion, hydrogen purity, and in the Ergun equation has been
Fad -∆Hadi - Fcat
∂t
( ) ( )
i)1 i)1 j)I
considered.
11. The gas phase and the catalyst/adsorbent particles are ∂ ∂T 1 ∂ ∂T
kz + krr (8)
assumed to be in local mass/thermal equilibrium at all times. ∂z ∂z r ∂r ∂r
12. Five chemical species are considered, viz., CH4, H2O,
CO, H2, and CO2. Because the effective axial and radial thermal conductivities
13. The nonlinear Langmuir model is used to describe the do not vary by orders of magnitude, we choose for purpose of
multicomponent adsorption equilibrium isotherm for CO2. simulation studies kr ) kz and use an estimated value obtained
14. The LDF model is used for the adsorption rate mecha- from the correlation given by Yagi et al.45 For the simulation
nism. In the present study, hydrotalcite37 is used as adsorbent of the undiluted bed, the adsorbent parameters (mi, bi, kfi,
for CO2 and the Langmuir isotherm model, and the LDF rate ∆Hadi,Fad) are zero-valued.
model was found to give an adequate description of CO2 6. Initial conditions:
adsorption and desorption behavior. The initial conditions (at t ) 0) used in the present study are
15. The catalyst is assumed to be uniform and small in size, as follows:
so that the effectiveness factor is considered to be unity and is
PH2
the same for all reactions.
T ) Tf, qji ) 0, PH2 ) Pf, CH2 ) , Pi ) 0,
16. External field interparticle heat transfer effects (both RTf
radial and axial temperature gradients in the framework of Ci ) 0 (i ) CH4, H2O, CO, CO2) (9)
packed-bed reactor taken as a whole) in the reactor are
considered. where Tf and Pf are feed gas temperature and pressure,
17. There is complete mixing of the adsorbent and catalyst respectively.
particles with negligible catalyst deactivation. 7. Boundary conditions:
18. The feed stream and column wall are maintained at the The following boundary conditions are used in the simulations
same constant temperature. (i) Reactor inlet (z ) 0, 0 e r e R0)
For the above assumptions, the governing equations with
initial and boundary conditions are summarized below.
1. The overall mass balance equation: -bDL ( )
∂Ci
∂z
) uf(Cfi - Ci) (10a)
∂C
t +
∂(uC)
+ Fad ∑
n ∂qji n III
- Fcat∑∑υijηjRj ) 0 (4)
-kz(∂T∂z ) ) u CC f pg(Tf - T) (10b)
∂t ∂z i)1 ∂t i)1 j)I u ) u f, P ) P f (10c)
2. The component mass balance for component i in the gas (ii) Reactor outlet (z ) L, 0 e r e R0)
phase:
∂Ci ∂T
( )
) 0, )0 (10d)
∂Ci ∂(uCi) ∂qji III
∂ ∂Ci ∂z ∂z
t + + Fad - Fcat ∑ υijηjRj ) b DL (5)
∂u ∂P
∂t ∂z ∂t j)I ∂z ∂z ) 0, )0 (10e)
∂z ∂z
where DL is the axial dispersion coefficient and is evaluated by (iii) At the center of reactor (r ) 0, z g 0)
the correlation given by Edwards and Richardson.44
3. Mass transfer rates: ∂Ci ∂T
) 0, )0 (10f)
∂r ∂r
∂qji
) kfi(q/i - qji) (6) ∂u ∂P
∂t ) 0, )0 (10g)
∂r ∂r
where kfi is the LDF mass transfer coefficient and may be (iv) At the reactor wall (r ) R0, z g 0)
evaluated by the correlation reported by Ding and Alpay.37
4. Adsorption equilibrium: ∂Ci
)0 (10h)
∂r
mibiPi
q/i ) (7) -kr
∂T
) hw(Tw - T) (10i)
n ∂r
1+ ∑ bjPj
∂u ∂P
j)1 ) 0, )0 (10j)
∂r ∂r
5. The energy balance for the bed-volume element: where Tw is the wall temperature (assumed to be equal to Tf for
purposes of simulation) and hw is the wall heat transfer
(42) Ergun, S. Chem. Eng. Sci. 1952, 48, 89-94.
(43) de Klerk, A. AIChE J. 2003, 49, 2022-2029. coefficient.
(44) Edwards, M. H.; Richardson, J. F. Chem. Eng. Sci. 1968, 23, 109-
123. (45) Yagi, S.; Kunii, D.; Wakao, N. AIChE J. 1960, 6, 543-546.
466 Energy & Fuels, Vol. 20, No. 2, 2006 Rajasree et al.
(
b ) i + 0.35 exp -0.39
D
dp ) (11)
parameter values and operating conditions.
Equations 4-8 with initial and boundary conditions (eqs
9 and 10) are numerically solved by a finite differ-
where i is defined as the average bed voidage of an infinite ence formulation, and the solutions obtained predict the SMR
diameter column. de Klerk43 showed that the effect of D/dp on process performance for undiluted and diluted bed conditions.
b is more pronounced for small values of D/dp (<4) and the For simulation purposes the finite difference formula-
bed voidage becomes constant at higher values of D/dp (>15). tion employed a sufficient number of grid points along
In the present study, the value of average voidage for the infinite the reactor length and radius so that numerically stable and
diameter column is assumed to be the situation when D/dp ) accurate solutions that describe the system dynamics in time
5040 (i.e., i ) 0.48). The average bed voidage b can then be are obtained.
calculated by using eq 11. The effect of change of bed voidage The overall mass balance equation (eq 6) and Ergun equa-
with D/dp has therefore been incorporated in the SMR model tion42 are solved to obtain the velocity and pressure profiles in
formulation. the reactor along with eqs 5 and 8 for the effluent mole fractions
and reactor temperature profiles. The process performance is
3. Results and Discussion
then quantitatively compared by studying the conversion of
It may be noted that the model considered is sufficiently methane XCH4 and hydrogen purity yH2(dry basis) defined as,
rigorous for analysis of SMR system behavior and is based on respectively,40
Performance Enhancement of SMR Energy & Fuels, Vol. 20, No. 2, 2006 467
Feed of CH4 (mol/s) - Effluent of CH4 (mol/s) state temperature profiles were obtained by solving the dynamic
XCH4 ) model for a sufficiently long time. The steady-state axial
Feed of CH4 (mol/s)
temperature profiles at the center (at r ) 0.0 m) and the radial
[ ( )( ) ]
RTf uPyCH4 temperature profiles where the temperature is the highest at z
) 1- (12) ) 0.03 m are plotted in Figures 1 and 2, respectively. For
ufPfyCH4 feed
RT outlet
varying Tf, Figure 1 shows that steep axial temperature gradients
exist that move further downstream for higher Tf. With increase
y H2
in Tf, higher values of conversion of methane (XCH4) are possible.
yH2(dry basis) ) (13)
1 - yH2O In turn, this leads to higher heat removal rates by reaction, and
consequently, steeper temperature gradients are observed near
Temperature Gradients in a Tubular Packed-Bed Mi- the reactor inlet. Figure 2, however, shows that even for a 6-mm-
croreactor. To check whether the SMR performance is affected diameter reactor radial temperature gradients can exist due to
by the existence of temperature gradients, simulation studies interparticle heat transfer, and this becomes more dominant at
for a 6-mm diameter packed-bed reactor for varying feed higher Tf. Mears26 developed a criterion to assess the presence
temperature (Tf) was first carried out. The catalyst loading (Wc) of radial heat transfer resistance by evaluating R0/RP. According
was kept constant at 6.45 g. The values of other parameters to the criterion, if R0/RP > 100 then the resistance may be
used in the simulation studies are listed in Table 1. The steady neglected. In our study, R0/RP ) 12, and the results obtained in
Figure 4. Temperature profiles: (a) d ) 2 mm and (b) d ) 15 mm. Values of other parameters are given in Table 1.
468 Energy & Fuels, Vol. 20, No. 2, 2006 Rajasree et al.
Figure 5. Velocity profiles: (a) d ) 2 mm and (b) d ) 15 mm. Values of other parameters are given in Table 1.
Figures 1 and 2 showing the presence of heat transfer limitations conversion increases from 20 to 100% upon reducing d
are supported by the Mears criterion. The above results suggest from 25 to 2.8 mm. A reactor with d ) 2.8 mm with the
that the SMR reactor diameter is an important parameter that corresponding length can therefore be designed for full
affects external field interparticle heat transport mechanisms. methane conversion. It is seen that, even at lower d, process
We therefore studied the effects of decreasing reactor diameter miniaturization by appropriate design and integration is feasi-
on performance, and the results obtained are discussed next. ble for the calculated values of reactor length and chosen
SMR Process Performance in Tubular Packed-Bed Mi- diameter.
croreactors. Studies are carried out by reducing the reactor Figure 4 compares the axial and radial temperature profiles
diameter (d) systematically from 25 mm to 2 mm, keeping plotted for reactors having d ) 2 mm and (Figure 4a) and 15
a catalyst loading of 6.45 g. The reaction was assumed to be mm (Figure 4b). It is clearly seen that the temperature gradients
carried out at Tf ) 773 K. The other parameter values are chosen are controlled for d ) 2 mm when compared to d ) 15 mm,
as given in Table 1. The steady-state process performance especially in the radial direction. It has been reported that the
in terms of methane conversion (XCH4) and H2 purity (yH2) advantages resulting from isothermality by reducing the reactor
is plotted in Figure 3. As seen, there is a significant improve- diameter are significantly higher than the effects of channeling
ment in the process performance with decrease in d. For for R0/RP ratios as low as even 4.26,31 In the present study, for
decreasing d, the reactor length L needs to be altered to main- a 2-mm-diameter reactor where isothermality is approached, R0/
tain constant catalyst loading. To assess the results, the needed RP ratio is 4. This suggests that choosing an appropriate tubular
reactor length with decreasing reactor diameter is also shown microreactor diameter is possible with negligible channeling
in Figure 3. For a catalyst loading of 6.45 g, the methane effects.
Performance Enhancement of SMR Energy & Fuels, Vol. 20, No. 2, 2006 469
Figure 6. Pressure profiles: (a) d ) 2 mm and (b) d ) 15 mm. Values of other parameters are given in Table 1.
We also studied the effects of velocity change and pres- The dynamics of the SMR process in time at the exit of the
sure drop due to adsorption and reaction in the reactor. The reactor is illustrated in Figure 7a,b in terms of XCH4 and yH2 for
nature of the velocity profiles obtained is illustrated in Fig- d ) 2, 4, 10, and 25 mm. It is observed from Figure 7a,b that
ure 5 for d ) 2 mm (Figure 5a) and d ) 15 mm (Figure 5b). the transient period is longer as the reactor diameter becomes
The results indicate that, for the similar superficial veloci- smaller from 25-mm to 4-mm reactors. For a 2-mm-diameter
ties, smaller-diameter reactors can provide good radial flow reactor, 100% conversion at the reactor outlet is seen throughout
features in the extended reaction zone. Hence, it may now the operation of the process. This suggests that, for a TPBM
be possible to control the hydrodynamics and thereby the with lower diameter, a lower Wc and therefore of shorter length
reaction behavior in the reaction zone. Likewise, the pres- would suffice for achieving complete conversion. Our calcula-
sure drop for the lower d ) 2 mm reactor is in the range of tion showed that a 2-mm-diameter reactor and length corre-
2-3 kPa, does not seems to be significant. It may also be sponding to a Wc of 5.5 g yield 100% conversion. Hence,
seen in Figure 6 that the magnitudes and nature of pres- operating with TPBM will help in running the SMR process at
sure profiles for both d ) 2 mm and d ) 15 mm reactors low Wc than conventionally employed larger-diameter reactors
are not considerably different. This is again advantageous from for the same performance parameters.
design considerations as the effects of pressure drop may Effect of Adsorbent Dilution on the SMR Process. Simula-
not be important for the SMR process taking place in a tion for effects of catalyst dilution is carried out by assuming
TPBM. hydrotalcite as the diluent that also acts as adsorbent for CO2
470 Energy & Fuels, Vol. 20, No. 2, 2006 Rajasree et al.
Figure 7. Transient profiles of SMR performance as a function of reactor diameter at the exit: (a) Methane conversion (XCH4) and (b) hydrogen
purity (yH2). Values of other parameters are given in Table 1.
Figure 8. Effect of adsorbent dilution on SMR performance as a function of reactor diameter. Constant bulk packing density of (a) methane
conversion (XCH4) and (b) hydrogen purity (yH2). Solid line: undiluted reactor. Dashed line: diluted reactor. Values of other parameters are given
in Table 1.
in the reaction zone. The equilibrium and kinetic constants for the undiluted beds for the same catalyst loading, Wc ) 6.25 g
hydrotalcite adsorbent have been taken from Ding and Al- (with Wa ) 0.0 g) and Fb(Fb ) Fcat) equal to 249 kg/m3, are
pay.37,39 The effects of adsorbent dilution are investigated in also given in Figure 8a,b for comparison. The solid lines are
two ways: (1) by comparing the performance of dynamics of for the undiluted systems, while the dashed lines are for the
diluted and undiluted TPBM of varying diameters having a fixed diluted reactors. It is clearly seen that for smaller-diameter
catalyst loading (Wc) and (2) by studying the effects of dilution reactors (e.g., Figure 8a, 4-mm curve) the effects of dilution
as a function of adsorbent loading (Wa) in a TPBM for varying are much more pronounced than those for the larger-diameter
bulk packing density Fb with constant catalyst loading. reactor (e.g., Figure 8a, 25-mm curve). Interestingly, for smaller-
Microreactor Dynamics with Dilution. In carrying out the diameter tubular microreactors (e.g., Figure 8a, 4-mm curve)
studies, we assume Wc ) 6.25 g and Wa ) 12.5 g for the diluted the time taken to reach the steady state is far greater than the
bed. The bulk packing density Fb(Fb ) Fcat + Fad) is assumed larger-diameter reactor (e.g., Figure 8a, 25 mm). This implies
to be 747 kg/m3. The process performance in time at the outlet that higher conversions are possible for longer time duration
of the diluted beds in terms of XCH4 and yH2 is presented, for the diluted bed when compared to the undiluted bed. The
respectively, in Figure 8a,b for varying d. The performance of radial temperature drop due to the endothermic nature of SMR
Performance Enhancement of SMR Energy & Fuels, Vol. 20, No. 2, 2006 471