Steam reforming behavior of methanol using paper-structured catalysts: Experimental and computational fluid dynamic analysis

ARTICLE IN PRESS

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ijhydene

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 0

0360-3199/$ - see frodoi:10.1016/j.ijhyde

�Corresponding auE-mail address:

Steam reforming behavior of methanol usingpaper-structured catalysts: Experimental andcomputational fluid dynamic analysis

Shuji Fukahoria, Hirotaka Kogaa, Takuya Kitaokaa,�,Mitsuyoshi Nakamurab, Hiroyuki Wariishia

aDepartment of Forest and Forest Products Sciences, Graduate School of Bioresource and Bioenvironmental Sciences,

Kyushu University, Fukuoka 812-8581, JapanbR&D Division, F.C.C. Co. Ltd., Shizuoka 431-1304, Japan

a r t i c l e i n f o

Article history:

Received 9 November 2007

Received in revised form

25 December 2007

Accepted 27 December 2007

Available online 14 February 2008

Keywords:

Paper-structured catalyst

Methanol steam reforming

Heat transfer

Computational fluid dynamic

analysis

nt matter & 2008 Internane.2007.12.063

thor. Tel./fax: [email protected]

a b s t r a c t

Copper–zinc oxide (Cu/ZnO) catalyst powders were impregnated into paper-structured

composites (catalyst paper) using a papermaking process. The paper-structured catalyst

was subjected to the methanol steam reforming (MSR) process and exhibited excellent

performance compared with those achieved by pellet-type or powdered catalyst. The

catalyst paper demonstrated a relatively stable gas flow as compared to catalyst pellets.

Furthermore, the MSR process was simulated by computational fluid dynamic (CFD)

analysis, and the heat conductivity influence of the catalyst layer was investigated. Higher

heat conductivity contributed to both higher methanol conversion and lower carbon

monoxide concentration; localization of heat and chemical species such as hydrogen and

carbon dioxide were improved, resulting in suppression of reverse water–gas shift reaction.

The CFD analysis was applied to the design of a catalyst layer in which a suitable shape was

suggested, where carbon monoxide formation was further suppressed without a decrease

in the methanol conversion.

& 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights

reserved.

1. Introduction

With the recent trend of making industrial processes more

environmentally friendly, there has been much attention

focused on the preparation of catalysts having high activity

and selectivity. In doing so, invariably it requires certain

modifications to be made to the catalyst composition and

nanometer scale morphology [1,2]. In addition to these

factors, a number of other physical aspects including the

transfer of mass and heat on the catalyst surface are also

significant for realizing practical catalytic processes. With

respect to both mass and heat transfer, novel catalysts with

tional Association for Hy

2993..jp (T. Kitaoka).

controlled regular architecture (so-called structured catalysts)

have been developed; the tailoring of the micrometer-order

pores inside the catalyst and the selection of supporting

materials are academic and practical areas of ongoing

research and development to improve the diffusion and heat

transfer of the products [3–6]. In a recent paper, Groppi et al.

reported on the thermal behavior of the structured

Pd=g–Al2O3 catalyst, which was wash-coated on a variety of

plate-type metal supports for studying gas/solid exothermic

reactions, and it was revealed that the type of metal support

significantly affected the formation of heat spots [5]. Takaha-

shi et al. developed a silica–alumina catalyst with continuous

drogen Energy. Published by Elsevier Ltd. All rights reserved.

ARTICLE IN PRESS

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 01662

micrometer-scale pores for the dehydration of alcohol, and

the contribution of the macropores on catalytic performance

was quantitatively evaluated [4].

In our previous studies [7–10], porous paper-structured

catalysts (so-called catalyst paper) was successfully prepared

using a papermaking technique; Copper–zinc oxide (Cu/ZnO)

catalyst particles were homogeneously scattered on glass and

silicon carbide (SiC) fiber networks having micrometer-scale

pores. The catalyst paper showed excellent methanol steam

reforming (MSR) performance, and the micrometer-scale

pores inside the catalyst paper presumably provide the

appropriate flow path for the reactants and modified gas,

resulting in higher methanol conversion and lower carbon

monoxide concentration, as compared to those achieved

using catalyst powder and catalyst pellets. Moreover, the

presence of SiC fibers in the catalyst paper improved the heat

transfer and heat distribution, contributing to higher metha-

nol conversion and lower carbon monoxide concentration

compared to those of catalyst paper prepared with glass

fibers. Thus, the paper-structured catalyst provides an

appropriate environment for the reforming catalyst by

optimizing the gas flow and heat distribution around the

catalyst, hence the catalyst paper prepared by a simple

papermaking technique is proposed for desirable use as

practical catalyst-based materials.

In the present study, the MSR behavior of the catalytic

materials, SiC fiber-containing catalyst paper, catalyst powder

and catalyst pellets, under different space velocities (SVs) was

investigated. In addition, the products and heat distributions

were simulated by computational fluid dynamic (CFD) analy-

sis, and the effect of both heat conductivity and the shape of

the catalyst layer on MSR performance were also discussed.

2. Materials and methods

2.1. Materials

Commercial Cu/ZnO catalyst pellets (MDC-3; cylindrical

shape; 3 mm diameter and 3 mm height; SUD-CHEMIE, Ltd.)

were used; the pellets were pulverized in part into 100-mesh

pass powders using a bowl mill. Pulp fibers, used as a

tentative supporting matrix in the wet papermaking process,

were obtained by beating commercial bleached hardwood

kraft pulp to 300 ml of Canadian Standard Freeness [11] with a

Technical Association of the Pulp and Paper Industry (TAPPI)

standard beater [12]. Glass fibers (CMLF208; ca. 0:8mm

diameter; heat conductivity; ca. 1:0 W=m K; Nippon Sheet

Glass, Ltd.) were cut into ca. 0.5 mm length using a four-flute

end mill. Fibrous SiC whiskers (size; ca. 0:5mm diameter and

30mm length; heat conductivity; ca. 25.5 W/m K) were pur-

chased from Tateho Chemical Industry, Ltd. Two types of

flocculants: polydiallyldimethylammonium chloride (PDAD-

MAC, molecular weight ðMwÞ: ca. 3� 105; charge density (CD):

5.5 meq/g; Aldrich, Ltd.) and anionic polyacrylamide (A-PAM,

HH-351; Mw: ca. 4� 106; CD: 0.64 meq/g; Kurita, Ltd.) were

used as retention aids. An alumina sol (Snowtex 520, Nissan

Chemicals, Ltd.) was used as a binder for enhancing the

physical strength of the catalyst paper.

2.2. Preparation of catalyst paper

The essential catalyst paper preparation details were reported

in the previous studies [7–10]. Briefly, glass and SiC fibers were

used as fiber matrix components in the catalyst paper. A fiber

suspension containing catalyst powders was mixed with

PDADMAC (0.5% total solids), alumina sol and A-PAM (0.5%

total solids), in that order. The mixture was poured into a pulp

fiber suspension, and then various types of catalyst paper

with SiC fibers of 20% w/w per total fiber content were

prepared according to the TAPPI test methods [13]. Following

pressing at 350 kPa for 5 min, the wet sheets were dried in an

oven at 105 �C for 30 min. One paper composite ð2� 104 mm2Þ

before calcination consisted of catalyst powders (4.5 g), glass

fibers (4.0 g), SiC fibers (1.0 g), alumina sol binder (0.1 g) and

pulp (0.25 g). Details of the SiC fibers and pulp content have

already been investigated [9]. The paper composites obtained

were thermally treated at 350 �C for 24 h to remove the organic

components, and then thermowelded to improve the physical

strength.

2.3. Reforming process and performance test

Ten circular catalyst paper samples, each with an area of 8�

102 mm2 and thickness of 1 mm, were stacked on top of each

other ð8� 103 mm3Þ and placed in a stainless steel thermo-

regulatory reaction cylinder (Fig. 1). Similarly, catalyst powder

or pellets with total volumes of 8� 103 mm3 were placed

inside the reaction cylinder. When either catalyst powder or

pellets were used, inert silica–alumina powders with the

same particle sizes were also mixed in to adjust the occupied

volume to 8� 103 mm3. A schematic diagram of the gas flow

system for the MSR process has already been reported [7]. The

MSR reactants were introduced into the reactor at a 2:1 molar

ratio of water to methanol (steam/carbon, S/C ratio ¼ 2) at a

constant flow rate (SV ¼ 537:5, 1075 and 2150 h�1Þ. In the case

of the autothermal reforming (ATR) process, a mixed gas of

methanol/water/oxygen (molar ratio: 1.00:2.00:0.125) was fed

into the reactor [8,10].

The amounts of Cu/ZnO catalyst were adjusted to 0.6 g in

each case. Prior to the MSR reaction, the catalyst samples

were reduced in situ with hydrogen at 250 �C for 1.5 h, followed

by a complete purge of all flow lines with nitrogen for 30 min.

The gas composition during the catalytic reaction was

monitored at reaction temperatures ranging from 200 to

300 �C using two online gas chromatographs (GCs) and one

offline GC. The exhaust gases generated in the MSR reaction

were passed through a cold trap in an ice bath. The unreacted

methanol and water residues were separated from the

gaseous components and then quantified using an offline

GC-FID (GC-17A Shimadzu, Ltd.) fitted with a Supel-Q Plot

column (0:53mm� 30 m, Shimadzu, Ltd.) at a furnace tem-

perature of 60 �C. The major gaseous products of hydrogen

and carbon dioxide were determined online using GC-TCD,

fitted with a Porapak-Q column (3 mm� 2 m, Shinwa Chemi-

cal Industries, Ltd.), while carbon monoxide, which was

isolated as a minor by-product, was measured by first

converting it to CH4 with an online methanizer (MTN-1,

Shimadzu, Ltd.), and then measuring the quantity of CH4

produced using a GC-FID fitted with a Porapak-Q column. All

ARTICLE IN PRESS

20.0 mm (100 meshes)

33.6 mm(170 meshes)

Methanol and steam

Modified gas

330 mm

ThermocoupleThermocouple

10.0 mm (50 meshes)

Stainless tube Catalyst layer

Fig. 1 – Schematic illustration of the reactor model used for the CFD simulation.

Table 1 – Specific heat conductivity ðCpÞ thermal con-ductivity ðgÞ and viscosity ðZÞ of chemical species

Species Cp (J/kg K) g (W/m K) Z (kg/m s)

CO2 Piecewise-linear 0.0323 2:61� 10�5

CO Piecewise-linear 0.0386 2:84� 10�5

H2 Polynomial Polynomial Polynomial

CH3OH Piecewise-linear 0.0351 1:88� 10�5

H2O Polynomial Polynomial Polynomial

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 0 1663

measurements were repeated at least four times and

standard deviations were obtained.

2.4. CFD analysis

CFD software, FLUENT ver. 6.2, was used to computationally

simulate the MSR process [14,15]. A steady state, segregated

solver was used to solve the heat, momentum, mass and

species transport equations in two-dimensional (2-D) reactor

configurations. The semi-implicit method of pressure–liquid

equation method and first-order upwind discretization

scheme were selected for quantifying the pressure–velocity

coupling and density interpolation, respectively.

The reactor model was constructed as follows. A full-scale

reactor model was built by GAMBIT (330 mm length, 33.6 mm

diameter). The reaction was considered to occur axisymme-

trically in a steady state, and a corresponding 2-D model was

constructed (shown in Fig. 1). The catalyst layer and

surrounding region were divided into 17 000 meshes with

dimensions of ca. 200mm� 200mm. A porous zone model

(including chemical reactions) was adopted for the catalyst

layer. The viscous resistance of the catalyst layer in all

directions was set at 1:2� 105=m2, while the fluid porosity

was set at 50%, in accordance with the values employed in the

prepared catalyst paper [9].

Five chemical species: methanol, water, hydrogen, carbon

dioxide and carbon monoxide, were used as ideal gases. The

specific heat conductivity (CpÞ, thermal conductivity (gÞ and

viscosity (ZÞ of the chemical species were set and shown in

Table 1. Cp of chemical species, g and Z of hydrogen and water

were temperature-dependent parameters compiled in the

FLUENT database (e.g. thermal conductivity of hydrogen at a

temperature T(K) is calculated by using the following equa-

tion: gH2¼ 8:27� 10�2

þ 3:56� 10�4 Tþ 1:07� 10�8 T2Þ. The

density of the mixed gas was dealt with as a non-compressive

ideal gas. The Cp, g, and Z values of the mixed gas obeyed the

mass-weighted mixing law.

The g values of catalyst layer meshes on the CFD analysis

were virtually set according to the constituent proportion of

the catalyst paper composed of glass fiber (58%), SiC fiber

(14%), Cu (12%), ZnO (13.5%) and Al2O3 (2.5%) with the

corresponding g values of 1.0, 25.5, 401, 54.0 and 29.0 W/m K,

respectively. The heterogeneous heat circumstance simulated

in this study seems to be partially consistent with that of the

real catalyst paper. Gravity was taken into account only for

the axial direction and was set at g ¼ 9:81 m=s.

The MSR process occurs in two steps, methanol reforming

and reverse water–gas shift reactions:

CH3OHþH2O!a1

3H2 þ CO2, (1)

H2 þ CO2 $a2

a02

COþH2O. (2)

The reaction rate of the two steps (a1, a2, a02, mol=m3 s) were

defined using the reaction rate constant (k1, k2, k02) and molar

concentration ðCi; mol=m3Þ; k1 and k2 were calculated accord-

ing to Arrhenius equation [16,17] using gaseous constant R ¼

8:31 J=K mol and temperature ðT;KÞ:

a1 ¼ k1 � C0:6CH3OH � C

0:4H2O,

a2 ¼ k2 � CH2� CCO2

,

a02 ¼ k02 � CCO � CH2O,

k1 ¼ f1 expð�E1=RTÞ,

k2 ¼ f2 expð�E2=RTÞ.

ARTICLE IN PRESS

0

50

100

0CO concentration (ppm)

Met

hano

l con

vers

ion

(%)

0

50

100

0

Met

hano

l con

vers

ion

(%)

CO concentration (ppm)

0

50

100

0

Met

hano

l con

vers

ion

(%)

CO concentration (ppm)

15000100005000 15000100005000

15000100005000

Fig. 2 – MSR performance: (a) catalyst paper, (b) catalyst powder and (c) catalyst pellets; with SV ¼ 537:5 h�1 (diamonds),

1075 h�1 (squares) and 2150 h�1 (triangles).

Reaction time (h)

Gas

flo

w (

SCC

M)

0

50

100

0 321

Fig. 3 – Modified gas flow rate: catalyst paper (red line) and

catalyst pellet (blue line). (For interpretation of the

references to color in this figure legend, the reader is

referred to the web version of the article).

6000

4000

2000

0

100

60

0

Met

hano

l con

vers

ion

(%)

CO

con

cent

ratio

n (p

pm)

21501075

40

80

20

Space velocity (h-1)4300

Fig. 4 – Comparison of methanol conversion (bars) with

carbon monoxide concentration (lines): experimental

(closed) and computational results (open).

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 01664

f r and Er corresponded to the frequency factor and activation

energy of the rth reaction, respectively, and were fitted to the

experimental results using catalyst powder; f1 ¼ 8:0� 108 s�1,

f2 ¼ 4:0� 108 m3=mol s, E1 ¼ 7:0� 104 J=mol and E2 ¼ 1:0�

105 J=mol. The reverse reaction of (2) (water–gas shift reaction)

occurred simultaneously and reaction rate constant k02 was

calculated using equilibrium constant of rth reaction Kr,

k02 ¼ k2=K2.

ARTICLE IN PRESS

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 0 1665

The boundary conditions were set as follows: inlet gas

temperature was 300 �C, the inlet gas velocity and temperature

of the reactor wall (TwÞ were altered in the appropriate ranges

of 1075–4300 h�1 and 265–380 �C, respectively. The outlet

pressure was ambient.

3. Results and discussion

3.1. Effect of SV on MSR performance

The paper-structured catalyst had the external appearance of

cardboard, and was convenient for handling and processing.

The essential catalyst paper preparation details were ob-

tained from the previous related studies [7–10]. The MSR

24.0

0.0

12.0

18.0

6.0

60.0

0.0

30.0

45.0

15.0

300

200

250

275

225

20.0 mm

33.6

Thermo

Catalyst layer

Fig. 5 – CFD simulation results; distribution of: (a) temperat

concentration (%), (d) hydrogen concentration (%), (e) carbon diox

(ppm).

behavior was investigated with respect to SV. Fig. 2 displays

the SV-dependent correlation between the degree of metha-

nol conversion and carbon monoxide concentration obtained

by changing reaction temperatures when modified with SiC

fiber-containing catalyst paper, catalyst powder and catalyst

pellets. Under relatively low SV conditions ðSV ¼ 537:5 h�1Þ,

small differences were observed between the MSR perfor-

mances obtained for the catalyst paper, powder and pellets;

the flow rate of the reactant became slow and the effect of the

intra-particle diffusion limitation decreased. The MSR perfor-

mance deteriorated with increasing SV conditions in all

cases, especially in catalyst pellet. In relatively high SV

conditions ðSV ¼ 2150 h�1Þ, a higher temperature is required

to achieve a similar methanol conversion compared to that

obtained under low SV conditions, causing carbon monoxide

5000

0

2500

3750

1250

40.0

0.0

20.0

30.0

15.0

70.0

0.0

35.0

52.5

17.5

gas flow

mm

couple

Stainless tube

ure ðunits ¼ �CÞ, (b) methanol concentration (%), (c) water

ide concentration (%) and (f) carbon monoxide concentration

ARTICLE IN PRESS

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 01666

formation through endothermic reverse water–gas shift

reaction [16,18,19]. It is well known that such side reaction

progresses with increasing catalyst size, thus carbon mon-

oxide concentration drastically increased when the catalyst

pellet was used. On the other hand, the MSR performances of

catalyst paper were slightly affected by SV, which were in

contrast to those obtained by using either catalyst powder

or catalyst pellets. In the paper-structured catalyst, micro-

meter scale catalysts were scattered on the fiber–mix

matrix consisting of ceramic fibers [7–10], therefore shift

reaction was suppressed, and consequently the excellent

MSR performance was achieved even under the high SV

conditions Such unique characteristics are advantageous

for practical applications under the required higher SV

condition.

Fig. 3 profiles the flow rate (standard cubic centimeter per

minute, SCCM) of the generated gas monitored with time,

obtained using catalyst paper and catalyst pellets at a

reaction temperature of 205 �C. A large oscillation of gas flow

rate with time was observed when catalyst pellets were

employed, resulting in unstable hydrogen production. On the

other hand, the catalyst paper demonstrated a relatively

stable gas flow. The average gas flow rates obtained when

catalyst paper and catalyst pellets were used corresponded to

52:2� 2:5 and 34:0� 8:4 SCCM, respectively. Such reforming

stability with regard to the ATR of methanol was discussed in

our previous study [8]. For practical applications, the constant

supply of hydrogen fuel is a striking factor for the generation

0

50

100

200

External wall temperature (°C)

Met

hano

l con

vers

ion

(%)

0

50

100

0CO conc

Met

hano

l con

vers

ion

(%)

350300250

Fig. 6 – CFD simulation results; heat conductivity of catalyst la

conversion, (b) carbon monoxide concentration and (c) relations

concentration.

of electricity by hydrogen fuel cells. The unique porous

structure of the catalyst paper is expected to contribute to a

uniform flow of gaseous components, and to be supplied

either to the catalyst surface or to quickly pass through the

catalyst layer.

3.2. CFD analysis of MSR process

First, the validity of the constructed simulation model was

checked. Fig. 4 displays a comparison of the methanol

conversion and carbon monoxide concentration under differ-

ent SV conditions, obtained either computationally or experi-

mentally. The order of methanol conversion efficiency and

carbon monoxide concentration, and their corresponding SV-

dependent behavior, were analogous between computational

and experimental results, providing certain validity to the

constructed model.

The distribution of temperature and the chemical species

(methanol, water, hydrogen, carbon dioxide and carbon

monoxide) are displayed in Fig. 5. There are low temperature

regions near the centerline of the reactor where hydrogen

concentration is low, indicating that the heat transfer from

the reactor wall is insufficient, and that the inward heat

transfer is an important factor to achieving high hydrogen

production. The hydrogen relatively generated at the periph-

eral part of the catalyst paper mounted in the reactor due to

the greater, more sufficient heat supply from the outside of

the reactor. Carbon monoxide was mainly generated at the

0

4000

8000

200External wall temperature (°C)

CO

con

cent

ratio

n (p

pm)

4000entration (ppm)

350300250

8000

yer is 1.0 (triangles) and 25.5 W/m K (squares); (a) methanol

hip between methanol conversion and carbon monoxide

ARTICLE IN PRESS

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 0 1667

peripheral part of the catalyst paper, as was hydrogen, espe-

cially at the bottom part of the catalyst layer. It was assumed

that the high hydrogen concentration and sufficient quantity

of heat at the peripheral and bottom part of the catalyst layer

stimulated the reverse water–gas shift reaction, resulting in a

high carbon monoxide concentration.

3.3. Influences of heat conductivity and catalyst papershape on MSR performance

In our previous study, SiC fiber-containing catalyst paper with

a SiC fiber content of 20%, exhibited higher methanol

conversion and lower carbon monoxide concentration com-

pared to catalyst paper prepared using just glass fibers as the

fiber component [9]. The influence of the heat conductivity of

the catalyst layer on MSR performance was investigated by

CFD analysis. The heat conductivity of all catalyst layer

meshes were set at 1.0 or 25.5 W/m K, corresponding to those

of glass and SiC fibers, and their temperature-dependent MSR

behaviors were evaluated. For the heat conductivity of the

catalyst layer, Fig. 6 displays the CFD simulation results

70.0

0.0

35.0

52.5

17.5

24000

0

12000

18000

6000

380

180

280

330

230

Fig. 7 – Distribution of temperature and chemical species at meth

catalyst layer is 1.0 W/m K (left) and 25.5 W/m K (right); tempera

and carbon monoxide concentration (bottom, ppm).

corresponding to the methanol conversion and carbon

monoxide concentration. The relationship between the

methanol conversion efficiency and carbon monoxide con-

centration was improved with increasing heat conductivity.

Moreover, the distribution profiles of the heat and carbon

monoxide concentrations have been simulated when the

methanol conversion was ca. 80% (Fig. 7). In the case where

the heat conductivity of the catalyst layer meshes was

1.0 W/m K, Tw, the methanol conversion and carbon mon-

oxide concentrations were 380 �C, 79.6% and 11 500 ppm,

respectively. When the heat conductivity of the catalyst layer

meshes was 25.5 W/m K, Tw was 265 �C and the methanol

conversion and carbon monoxide concentration were 81.7%

and 1700 ppm, respectively; the reaction temperature de-

creased by 115 K at the same methanol conversion. Accord-

ingly, the equilibrium of endothermic reverse water–gas shift

reaction became unfavorable, possibly leading to the drastic

decrease of carbon monoxide concentration. Furthermore,

low heat conductivity caused a larger temperature gradient in

the radius direction, resulting in heterogeneous hydrogen

production. Such localization of heat and hydrogen possibly

anol conversion of ca. 80% when the heat conductivity of the

ture (upper, units ¼ �C), hydrogen concentration (middle, %)

ARTICLE IN PRESS

300

200

250

275

225

70.0

0.0

35.0

52.5

17.5

5000

0

2500

3750

1250

321 4

2 3 1 4

3000

2000

1000

0

100

80

60

40

0

20

Met

hano

l con

vers

ion

(%)

CO

con

cent

ratio

n (p

pm)

21 43

Fig. 8 – Effect of the catalyst layer shape on MSR performance: (a) constructed catalyst layer model; gray regions represent the

catalyst layer, (b) MSR performance at the reaction temperature of 300 �C; methanol conversion (open bars) and carbon

monoxide concentration (filled bars), (c) distribution of temperature and chemical species of each model; temperature (upper,

units ¼ �C), hydrogen concentration (middle, %) and carbon monoxide concentration (bottom, ppm).

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 01668

triggered a reverse water–gas shift reaction, resulting in a

high carbon monoxide concentration.

It was pointed out that the carbon monoxide generated at

the peripheral and bottom part of the catalyst layer, where

the reverse water–gas shift reaction rather than the MSR

reaction occurred because the hydrogen concentration hardly

increased, as shown in Fig. 5(d). Therefore, the catalyst layer

was partially cut in the peripheral and bottom part of the

catalyst layer, and its MSR performances compared (Fig. 8);

the peripheral and bottom part of the catalyst layer was cut

by 1.0, 5.0 and 8.4 mm, respectively (displayed as models 2, 3

and 4). The methanol conversion and carbon monoxide

concentration were affected by the shape of the catalyst

layer. The carbon monoxide concentration when simulated

using model 2 was lowered by 20%, as compared with that

simulated using model 1 without a reduction in the methanol

conversion efficiency. In the case of models 3 and 4, the

carbon monoxide concentration decreased still further, while

ARTICLE IN PRESS

20.0 mm

33.6 mm

31.6 mm

0

50

100

0CO concentration (ppm)

Met

hano

l con

vers

ion

(%)

300020001000

Fig. 9 – Schematic illustration of the tuned catalyst layer (a)

and reforming performance (b); MSR with a normal catalyst

layer (closed diamonds), MSR with a tuned catalyst layer

(open diamonds), ATR with a normal catalyst layer (closed

circles), ATR with a tuned catalyst layer (open circles).

SV ¼ 537:5 h�1.

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 0 1669

the methanol conversion also decreased; the methanol con-

version was less than 70% when simulated using model 4.

These results strongly suggest this outcome, such that an

appropriate catalyst layer shape could be designed according

to the reforming conditions, making CFD analysis a powerful

tool for efficient material development.

3.4. ATR process using SiC fiber-containing catalyst paper

In our previous study, the catalyst paper prepared with

ceramic fibers was subjected to ATR process: the combined

process of endothermic MSR and exothermic partial oxida-

tion of methanol [20–22]. In the ATR process, the relationship

between the methanol conversion and carbon monoxide

concentration was superior to that of MSR; therefore, the

SiC fiber-containing paper was also subjected to the ATR

process. Moreover, it was indicated through CFD simulations

that the catalyst layer model 2 in Fig. 8(a) showed promise for

carbon monoxide reduction. Thus, a catalyst paper having

similar shape to model 2 in Fig. 8(a) was experimentally

prepared and subjected to both MSR and ATR processes. Two

types of catalyst paper with different diameter were com-

bined: five circular catalyst paper samples with a diameter of

33.6 mm were used as the first part of the catalyst, and five

circular catalyst paper samples with a diameter of 31.6 mm

were used as the latter part. A schematic illustration of the

tuned catalyst layer and its reforming performance are shown

in Fig. 9. When a modified catalyst with a tuned shape was

employed in both MSR and ATR processes, the carbon

monoxide concentration decreased by 10% without a con-

comitant decrease in the methanol conversion, resulting in a

superior relationship between the methanol conversion and

carbon monoxide concentration. The ATR rather than the

MSR process proceeded in a high methanol conversion at a

low carbon monoxide concentration. In the ATR process using

a catalyst with a tuned shape at SV ¼ 537:5 h�1, the carbon

monoxide concentration was lowered below 1000 ppm, which

is one-tenth of the value obtained using Cu/ZnO powder [19].

Various parameters, including the reaction temperature,

velocity and composition of inlet gas, heat conductivity, the

amount of catalyst in the reactor and catalyst shape, are

easily controllable in the CFD model. At this stage, the carbon

monoxide concentration of modified gas was not low yet

enough to permit direct supply to practical polymer electro-

lyte fuel cells; however, a combined use of experimental and

computational results is expected to make catalyst material

development more speedy and effective.

4. Conclusion

The MSR performance of various catalytic materials was

investigated both experimentally and computationally. The

paper-structured catalyst exhibited excellent and stable MSR

performance even under high SV conditions, as contrasted

with those obtained using catalyst powder or pellets, which is

an appealing result for practical application. Furthermore, the

effects of heat conductivity and the shape of the catalyst layer

were demonstrated by CFD analysis. The improvement in the

heat conductivity of the catalyst layer contributed to the rapid

heat transfer and resultant small temperature gradient,

suppressing undesired carbon monoxide formation. In addi-

tion, CFD analysis also revealed that the shape of the catalyst

layer significantly influenced the MSR performance, while the

relationship between the methanol conversion and carbon

monoxide concentration was improved in both MSR and ATR

processes by tuning the shape of the catalyst layer. In the ATR

process by using a paper-structured catalyst with a tuned

shape, the carbon monoxide concentration halved as com-

pared to that obtained in the MSR process. From these results,

it was confirmed that the paper-structured catalyst, which is

both easy handling and easily fabricated, is a promising

candidate for high-performance catalyst-based materials

Acknowledgments

This research was supported by Research Fellowships of the

Japan Society for the Promotion of Science for Young

ARTICLE IN PRESS

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 1 6 6 1 – 1 6 7 01670

Scientists (S.F. and H.K.) and by an Industrial Technology

Research Grant Program in 2003 from the New Energy and

Industrial Technology Development Organization (NEDO) of

Japan (T.K.). The authors wish to thank Ms. K. Hayashida and

Ms. Y. Kubara for their assistance with CFD analysis.

R E F E R E N C E S

[1] Bichon P, Asheim M, Jordal A, Sperle T, Fathi M, Holmen A,et al. Hydrogen from methanol steam-reforming overCu-based catalysts with and without Pd promotion. Int JHydrogen Energy 2007;32(12):1799–805.

[2] Perez-Hernandez R, Gutierrez-Martınez A, Gutierrez-WingCE. Effect of Cu loading on CeO2 for hydrogen production byoxidative steam reforming of methanol. Int J HydrogenEnergy 2007;32(14):2888–94.

[3] Nakanishi K. Pore structure control of silica gels based onphase separation. J Porous Mater 1997;4(2):67–112.

[4] Liu W, Roy S. Effect of channel shape on gas/liquid catalyticreaction performance in structured catalyst/reactor. ChemEng Sci 2004;59(22–23):4927–39.

[5] Groppi G, Airoldi G, Cristiani C, Tronconi E. Characteristics ofmetallic structured catalysts with high thermal conductivity.Catal Today 2000;60(1–2):57–62.

[6] Takahashi R, Sato S, Sodesawa T, Arai K, Yabuki M. Effect ofdiffusion in catalytic dehydration of alcohol over silica–alumina with continuous macropores. J Catal 2005;229(1):24–9.

[7] Fukahori S, Kitaoka T, Tomoda A, Suzuki R, Wariishi H.Methanol steam reforming over paper-like composites ofCu/ZnO catalyst and ceramic fiber. Appl Catal A Gen2006;300(2):155–61.

[8] Koga H, Fukahori S, Kitaoka T, Tomoda A, Suzuki R, WariishiH. Autothermal reforming of methanol using paper-likeCu/ZnO catalyst composites prepared by a papermakingtechnique. Appl Catal A Gen 2006;309(2):263–9.

[9] Fukahori S, Koga H, Kitaoka T, Tomoda A, Suzuki R, WariishiH. Hydrogen production from methanol using a SiC fiber-containing paper composite impregnated with Cu/ZnOcatalyst. Appl Catal A Gen 2006;310(17):138–44.

[10] Koga H, Fukahori S, Kitaoka T, Mitsuyoshi N, Wariishi H.Paper-structured catalyst with porous fiber-network micro-structure for autothermal hydrogen production. Chem Eng J.doi:10.1016/j.cej.2007.11.042.

[11] Technical Association of the Pulp and Paper Industry. Testmethods, T227; 1997.

[12] Technical Association of the Pulp and Paper Industry. Testmethods, T200; 1997.

[13] Technical Association of the Pulp and Paper Industry. Testmethods, T205; 1997.

[14] Kumar V, Saini S, Sharma M, Nigam KDP. Pressure drop andheat transfer study in tube-in-tube helical heat exchanger.Chem Eng Sci 2006;61(13):4403–16.

[15] Vadakkapattu VC, Lee TW. Multi-dimensional simulations ofthe chemical vapor deposition for thermal barrier coatingsusing ZrCl4–H2–CO2–Ar gas mixtures. Surf Coat Technol2006;201(3–4):1065–73.

[16] Agrell J, Birgersson H, Boutonnet M. Steam reforming ofmethanol over a Cu=ZnO=Al2O3 catalyst: a kinetic analysisand strategies for suppression of CO formation. J PowerSources 2002;106(1–2):249–57.

[17] Patel S, Pant KK. Experimental study and mechanistic kineticmodeling for selective production of hydrogen via catalyticsteam reforming of methanol. Chem Eng Sci2007;62(18–20):5425–35.

[18] Nagano S, Miyagawa H, Azegami O, Ohsawa K. Heat transferenhancement in methanol steam reforming for a fuel cell.Energy Convers Manage 2001;42(15–17):1817–29.

[19] Purnama H, Ressler T, Jentoft RE, Soerijanto H, Schlogl R,Schomacker R. CO formation/selectivity for steam reformingof methanol with a commercial CuO=ZnO=Al2O3 catalyst.Appl Catal A Gen 2004;259(1):83–94.

[20] Mizsey P, Newson E, Truong T, Hottinger P. The kinetics ofmethanol decomposition: a part of autothermal partialoxidation to produce hydrogen for fuel cells. Appl Catal AGen 2001;213(2):233–7.

[21] Cavallaro S, Chiodo V, Vita A, Freni S. Hydrogen productionby auto-thermal reforming of ethanol on Rh=Al2O3 catalyst.J Power Sources 2003;123(1):10–6.

[22] Yong ST, Hidajat K, Kawi S. Reaction study of autothermalsteam reforming of methanol to hydrogen using a novelnano CuZnAl-catalyst. J Power Sources 2004;131(1–2):91–5.