A

tamaa©

K

1

igmtmi

fabrrppro

f

1d

Chemical Engineering Journal 139 (2008) 136–146

Simulation of effect of catalytic particle clustering on methanesteam reforming in a circulating fluidized bed reformer

Wang Shuyan a, Yin Lijie a, Lu Huilin a,∗, He Yurong a,Jianmian Ding b, Liu Guodong a, Li Xiang a

a School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, Chinab MSC Software, 2 MacArthur Place, Santa Ana, CA 92707, USA

Received 9 April 2007; received in revised form 26 July 2007; accepted 27 July 2007

bstract

Effect of catalytic particle clustering on methane steam reforming in a circulating fluidized bed (CFB) reformer is analyzed numerically using awo-dimensional hydrodynamic model coupling with methane steam reforming reaction kinetics. Gas flow fields and distributions of temperature

nd gas species in a Haldor Topsoe Ni/Mg Al2O4 spinel catalytic particle cluster are predicted. The computed results indicate that the rates ofethane steam reaction of a catalyst particle in the cluster are less than that of an isolated catalyst particle out side the cluster. The yields of H2, CO

nd H2O in the cluster increase with the increase of operating temperature and inlet gas velocity but decrease with the increase of reactor pressurend feed steam.

2007 Elsevier B.V. All rights reserved.

dized

otcrloahTdograta

eywords: Methane steam reforming; Catalyst particle cluster; Circulating flui

. Introduction

Steam reforming of light hydrocarbons is an industriallymportant chemical reaction and a key step for producing hydro-en and syngas [1,2]. About 50% of hydrogen comes fromethane steam reforming [3,4]. Recent demanding for hydrogen

hat utilized by many processes such as oil refining, methanol,etallurgy, ammonia, etc., have imposed a strong economic

ncentive to improve the hydrogen production technology.The methane steam reforming are endothermic reactions

avored at high temperature. The conventional steam reformersre fixed-bed catalytic reactors with catalyst loaded in a num-er of tubes placed in a furnace. The conventional fixed-bedeformers have some disadvantages such as low heat transferates, diffusion resistance in the catalyst pores, and large tem-erature gradients [5–9]. The endothermic reforming reaction

rocesses are inefficient because of less heat transferred into theeformer. Moreover, the reactions produce significant quantitiesf carbon dioxide. The fluidized bed reformer is an attractive

∗ Corresponding author. Tel.: +86 10 451 8641 2258;ax: +86 10 451 8622 1048.

E-mail address: huilin@hit.edu.cn (L. Huilin).

tty

ftsa

385-8947/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.cej.2007.07.097

bed reformer; Numerical simulations

ption in the syngas production because of high rate of heatransfer to maintain the isothermal operation. The overall pro-ess is generally integrated with the fluidized-bed combustor andeformer [10–14] to obtain advantages of thermodynamic equi-ibrium, more methane conversion and high hydrogen yield. Inrder to improve further the performance of methane reformingnd catalyst regeneration, the circulating fluidized bed (CFB)as been widely utilized [15–18] as chemical processing unit.he CFB reformer can also be a major unit for hydrogen pro-uction. In a CFB reformer, the deactivated catalyst flowing outf the reformer with the exit gas stream can be separated in aas–solid separator [19]. The regenerated catalyst is can then beecycled to the CFB riser. The carbon is completely burned inn integrated combustor and the catalyst is fully regenerated inhe overall process. The fluidized solid catalyst particles acteds an internal heat carrier provide uniform reacting tempera-ure in the riser. More steam may be provided to the reformero avoid formation of coke and to achieve the high hydrogenield.

It is generally agreed that the suspended particles may partly

orm clusters moving upward in the center, and downward nearhe riser walls in the risers [20]. Experiments have demon-trated the flow behavior of clusters in circulating fluidized bedsnd obtained the cluster motion, cluster formation time, time-

W. Shuyan et al. / Chemical Engineeri

Nomenclature

Cg gas specific heat (kJ/(m3 K))C1, C2 empirically determined constantsdp particle diameter (m)D molecular diffusion coefficient (m2/s)Dc diameter of cluster (m)DCH4 diffusion coefficient of CH4 (m2/s)g gravity (m/s2)Gk turbulent kinetic energy production (kg/(m s3))GR production due to reaction (kg/(m s3))h convective heat transfer coefficient (kJ/(m2 s K))�Hj heat of reaction for reaction j (kJ/kmol)i, j, k coordinate directionk turbulent kinetic energy (m2/s2)k1 rate constant for (A) (mol Pa0.5/(kg s))k2 rate constant for (B) (mol/(Pa kg s))KCO2 adsorption equilibrium constant of CO2 (Pa−1)KCO adsorption equilibrium constant of CO (Pa−1)KCH4 adsorption equilibrium constant of CH4 (Pa−1)KH2O adsorption equilibrium constant of H2O (Pa−1)K1 equilibrium constant (Pa2)K2 equilibrium constant, dimensionlessMi molecular weight of species i (kg/kmol)n particle numberp gas pressure (Pa)Pi partial pressure of component i (Pa)r radial direction (m)ri rate of reaction i (kmol/(kg s))R radius of particle (m)Rc radius of cluster (m)T temperature (K)ug inlet gas velocity (m/s)um mean axial velocity of gas (m/s)x vertical coordinate (m)Yi mass fraction of species i

Greek symbolsε energy dissipation rate (m2/s3)εg porosityλg thermal conductivity of gas (kJ/(m s K))μg gas viscosity (kg/(m s))μl laminar gas viscosity (kg/(m s))νt kinematic viscosity (m2/s)ρc catalytic density (kg/m3)ρg gas density (kg/m3)σ empirically determined constantφj catalyst activity function

SubscriptsA, B, C reaction routesg gas phaseR at particle surfaces solid phase

fHrcr

pchfitgp

2

mrwidtsnwnmf

2

f

ng Journal 139 (2008) 136–146 137

raction of existence and occurrence in CFB risers [21–27].ence, in-depth knowledge about particle clustering effect on

eactions in CFB riser is useful in predicting performance ofirculating fluidized bed reactors and hence the design the CFBeactor.

In this work, we numerically investigate the effect of catalystarticle clustering on methane steam reforming with a Ni/Al2O3atalyst particle cluster using a simplified two-dimensionalydrodynamic model with catalytic chemical reactions. Gas flowelds and distributions of gas temperature and gas species in

he cluster are predicted. The effects of gas temperature, inletas velocity and reactor pressure on methane reforming areresented.

. Mathematical models

Fig. 1 shows the geometry of the cluster. A cluster isodeled as a round two-dimensional disk in the riser. The

adius of the cluster disk is Rc. Each particle in the clusteras labeled. Particles in the cluster are uniformly arranged

n the cluster, and remain stationary. Particle diameter andensity are assumed to be constant during reforming reac-ions. Radiation heat transfer between gas and particle is muchmall compared with convective heat transfer, and therefore,eglected. Radiation heat transfer between particles and thealls, and heat transfer between gas and catalyst particle areot considered in present study. The conservation equations ofass, species and energy in Eulerian coordinates are given as

ollows.

.1. Gas phase continuity equation

The mass conservation equation of gas can be written as

ollows [28,29]:

∂ρg

∂t+ ∂

∂xj

(ρguj) = 0 (1)

Fig. 1. Scheme of catalyst particle cluster.

1 ineeri

2

a

wit

2

mte

a

w

G

T

2

[

wa(

2

C

wu(s

2

tirpF

TP

ITIIIGGDDDDD

38 W. Shuyan et al. / Chemical Eng

.2. Gas phase momentum equation

The conservation of momentum for gas phase is expresseds follows [28,29]:

∂

∂t(ρgui) + ∂

∂xj

(ρgujui)

= ∂p

∂xi

+ ∂

∂xj

(μg

∂uj

∂xi

)+ ∂

∂xj

(μg

∂ui

∂xj

)+ ρgg (2)

here μg = μl + μt is the effective viscosity and μt = Cμρgk2/εs the gas phase turbulent viscosity, which was modeled by k-�urbulent model given below.

.3. Turbulence model for gas phase

In present study, we employ the standard k-ε turbulenceodel for simulating turbulent flow in the riser. The standard k-ε

urbulence model gives gas phase equations of turbulent kineticnergy [29]

∂

∂t(ρgk) + ∂

∂xj

(ρgujk) = ∂

∂xj

(μt

σk

∂k

∂xj

)+ Gk − ρgε (3)

nd turbulent kinetic energy dissipation rate

∂

∂t(ρgε) + ∂

∂xj

(ρgujε) = ∂

∂xj

(μt

σε

∂ε

∂xj

)+ ε

k[C1Gk − C2ρgε]

(4)

here Gk is the production term,

k = μg

(∂ui

∂xj

+ ∂uj

∂xi

)∂ui

∂xj

− 2

3δij

∂uk

∂xk

(5)

The empirical constants, Cμ, σε, C1 and C2, are listed inable 1.

.4. Energy conservation of gas phase

The energy balance equation of gas phase is written as follows28,29]:

ris

C

r

able 1arameters used in the simulations

nlet gas velocity 1.0 m/sotal pressure 0.1 MPanlet gas temperature 750 Knlet mass fraction of CH4 0.2286nlet mass fraction of H2O 0.7714as density 0.2847 kg/m3

as viscosity 1.546 × 10−5 kg/m-siffusion coefficient of CH4 2.064 × 10−5 m2/siffusion coefficient of H2O 2.178 × 10−5 m2/siffusion coefficient of CO2 1.381 × 10−5 m2/siffusion coefficient of CO 1.92 × 10−5 m2/siffusion coefficient of H2 6.34 × 10−5 m2/s

ng Journal 139 (2008) 136–146

∂

∂t(CgTg) + ∂

∂xj

(ujCgTg)

= ∂

∂xj

[(λg + Cgνt

σh

)∂Tg

∂xj

]+∑

i

ρc�Hiri (6)

here ri is the reaction rates of routes A, B and C. As in previousssumption, radiation heat transfer process is not included in Eq.6).

.5. Species conservation equation of gas phase

The mass balance for a gas species k (k = CO2, CO, H2O andH4) is [28]

∂

∂t(ρgYk) + ∂

∂xj

(ρgujYk)

= ∂

∂xj

[(ρkDk + μt

σY

)∂Yk

∂xj

]+∑

i

ρcMiri,k (7)

here Yk is the local mass fraction of species k, Dk the molec-lar diffusion coefficient, σY is the turbulent Schmidt numberσY = 0.7). The last term at right hand side of Eq. (7) is massource from methane reforming reactions.

.6. Reaction kinetics

Methane steam reforming reaction models associated withhe different rate expressions were proposed [30–34]. Abasharnvestigated the mathematical models coupling with methaneeforming with Haldor Topsoe Ni/Mg Al2O4 spinel catalystarticles in fluidized bed [4]. Based on the work of Xu androment [30], the following reaction rates consisting of threeepresentative reactions are used for methane steam reform-

ng and water-shift reactions with Haldor Topsoe Ni/Mg Al2O4pinel catalyst particles:

H4 + H2O ↔ CO + 3H2 �H298 = 206.0 kJ/mol (A)

1 = k1(PCH4PH2O/P2.5

H2) − (PCOP0.5

H2/K1)

DEN2 (8)

Diameter of particle 100 �mDiameter of cluster 975 �mNumber of particles 19Temperature of particle 750 KParticle density 3900 kg/m3

Porosity of cluster 0.8Empirically determined constant C1 1.44Empirically determined constant Cμ 0.09Empirically determined constant σε 1.3Empirically determined constant C2 1.92Empirically determined constant σk 1.0Empirically determined constant σh 1.0

neeri

C

r

C

r

D

wa

k

k

k

K

K

K

K

K

K

cTltra

r

wtc

2

safl

u

i

w

mpacatcdn[cwteFm(rthrough out all present computations except otherwise stated.A fixed time step of 1 × 10−5 s is used. The simulation runs0.8 s of actual flow times. Fig. 2 shows the instantaneous massfluxes of H2, CO and CO2 gas species and gas temperature at

W. Shuyan et al. / Chemical Engi

O + H2O ↔ CO2 + H2 ΔH298 = −41.0 kJ/mol (B)

2 = k2(PCOPH2O/PH2 ) − (PCO2/K2)

DEN2 (9)

H4 + 2H2O ↔ CO2 + 4H2 ΔH298 = 164.9 kJ/mol (C)

3 = k3(PCH4P

2H2O/P3.5

H2) − (PCO2P

0.5H2

/K1K2)

DEN2 (10)

EN=1+KCH4PCH4 + KH2PH2+KCOPCO+KH2OPH2O/PH2

(11)

here the reaction rate constants of reaction (A), (B) and (C)re, respectively

1 = 8.336×1017 exp

(−28879.0

Tg

)(mol Pa0.5/kg s) (12)

2 = 12.19 exp

(−8074.3

Tg

)(mol Pa−1.0/kg s)

(13)

3 = 2.012 × 1017 exp

(−29336.0

Tg

)(mol Pa0.5/kg s)

(14)

1 = 10266.76 × 106exp(−26830.0/Tg + 30.11

)(Pa2)

(15)

2 = exp(4400.0/Tg − 4.063

)(16)

CH4 = 6.65 × 10−9 exp

(4604.28

Tg

)(Pa−1) (17)

H2 = 6.12 × 10−14(

9971.13

Tg

)(Pa−1) (18)

CO = 8.23 × 10−10 exp

(8497.71

Tg

)(Pa−1) (19)

H2O = 1.77 × 105 exp

(−10666.35

Tg

)(20)

Carbon may be generated by the decompositions of hydro-arbons and carbon monoxide over reforming catalyst particles.he reaction rate will be affected by coke formation. The cata-

yst activity is defined as the ratio of the catalytic reaction rateo the initial reaction rate with fresh catalyst. Thus, the reaction

ate rj [34] for the deactivation of the catalyst can be expresseds

c,j = φjrj (21)

here φj is catalyst activity function. In present simulations,he value of catalyst activity function is taken to be 1.0 and noarbon deposition is assumed.

ng Journal 139 (2008) 136–146 139

.7. Boundary conditions and simulation procedures

The inlet gas compositions, temperature and velocity werepecified. At the outlet, the continuous gas and heat flow bound-ry conditions were used. No-slip boundary condition for gasow at the particle surface is applied

= k = ε = 0 (22)

The boundary condition for gas species at the particle surfaces

(ρkDk + μt

σY

)∂Yk

∂xj

∣∣∣∣Rc

=∑

ρcMkrk (23)

The boundary condition of temperature at both particle andall surfaces is

∂Tg

∂xj

∣∣∣∣Rc

= 0 (24)

The above equations are solved using the finite volumeethod [35] with appropriate boundary conditions. The cluster

orosity can be calculated from εg = (D2c − nd2

p)/D2c , where Dc

nd n are the diameter of cluster and number of particles in theluster, respectively. The parameters for the porosity calculationre listed in Table 1. Nineteen catalyst particles form the clus-er and give a cluster porosity of 0.8. The physical domain ofylindrical boundary of each particle was mapped onto a squareomain in the computation domain using the body-fitted coordi-ate to make the rectangular grids for the boundary of particles36]. Fresh gases flow into the cluster, and react on the surface ofatalyst particles. Sensitivity of the computed results to grid sizeas tested by increasing the number of computational grids until

he grid size has no significant effect on the results. The grid sizeffect on the computed H2/CO with various grids is shown inig. 13. It is noticed that the difference between the results fromedium grid sizes (162 × 328) and those from finer grid sizes

196 × 374) is minor, but from coarse grid sizes (122 × 260). Toeduce the computation times, the medium grid sizes were used

Fig. 2. Instantaneous mass flux of gases and temperature at exit.

1 ineering Journal 139 (2008) 136–146

ts(

3

3

p1e

ttmcRvF

Ff

F

40 W. Shuyan et al. / Chemical Eng

he exit. Time-averaged values were calculated from last 0.5 s. Aet of simulation needs about 24 h of CPU time on PC computer80 GB hard disk, 128 Mb Ram and of 600 MHz CPU).

. Simulation and analysis

.1. Base case simulations

In the base simulations, the inlet gas velocity, inlet gas tem-erature, porosity and the reformer pressure were set to be.0 m/s, 750 K, 0.8 and 0.1 MPa, respectively, as listed in Table 1xcept otherwise specified.

Fig. 3 shows the axial velocity of gas phase passing throughhe centerline of the cluster. The negative axial velocity meanshat gas backing flow in the down-stream, seeing Fig. 8a. Theean axial gas velocity through the cluster (y = 0), um, is cal-

ulated from computed axial velocities in the cluster domains.

atio of mean axial velocity through the cluster to inlet gaselocity, um/ug, is then determined. The results are shown inig. 4 [37]. Simulated results show that the ratio of gas velocity

Fig. 3. Gas velocity distribution along lateral direction.

ig. 4. Ratio of mean axial velocity through cluster to inlet gas velocity as aunction of cluster porosity.

ugt

lthttrd

arprWicso

ig. 5. Distributions of gas temperature along axial and lateral directions.

m/ug is 0.035 for the base case. This means only about 3.5% ofases flow through the cluster, while most gas bypasses it dueo the flow resistance of the cluster.

Fig. 5 shows the gas temperature distribution along axial andateral directions of the cluster. The gas temperature decreaseshrough the cluster due to endothermic reforming reaction. Theigh gas temperatures are located at the front of the cluster facinghe incoming gas. The low gas temperature is at the backside ofhe cluster. In the radial direction, the gas temperature increasesadially away from the cluster. The maximum of gas temperatureifference in the cluster is about 40 K.

The variations of temperature and gas flux in the cluster canffect methane steam reforming of the cluster. Fig. 6 shows theeaction rates for some particles in the cluster and an insolatedarticle. Particles located in the front of the cluster have the higheaction rate r1 and r3, and low r2 due to high gas temperature.

e found that the reaction rate r1 and r3 of the isolated particle

n the flowing stream was larger than that of particles in theluster. Hence, particle clustering reduces the rates of methaneteam reaction. Fig. 7 shows the axial and radial distributionsf H2, CO and CO2 along the centerline of the cluster. In the

Fig. 6. Reaction rates of particle in the cluster.

W. Shuyan et al. / Chemical Engineering Journal 139 (2008) 136–146 141

cies a

asdbordofi

3

dttec

fWowisa

ddtaii

Fig. 7. Molar fractions of gas spe

xial direction, the molar concentrations of H2, CO and CO2pecies increase, reach a maximum at the cluster center, thenecrease. The variation distributions of H2O and CH4 are causedy reforming reactions of (A) and (C). The high concentrationf CO species at the tail of cluster is due to reforming shifteactions of (A) and (B). From the gases distribution in the lateralirection, we noticed that the non-uniformity of produced gasesf CH4 and H2O in the cluster is affected by the uneven floweld in the cluster.

.2. Effect of cluster porosity

The cluster porosity value can be changed by increase orecrease of the cluster size or number of particles occupied in

he cluster. Fig. 8 shows the contour of gas axial velocity throughhe cluster at three different cluster porosities. With cluster diam-ters of 975, 1378 and 1949 �m, respectively, the correspondingluster porosities are 0.8, 0.9 and 0.95, respectively. A vortex was

ltoH

Fig. 8. Axial velocity of gas as a function of cluster porosities. (

long axial and lateral directions.

ormed at the back of the cluster at the cluster porosity of 0.8.ith the increase of cluster porosity, the backing flow at the back

f the cluster is disappeared. More gas can flow into the clusterith the increase of cluster porosity, and hence more methane

s reformed in the cluster and the flow rates of H2, CO and CO2pecies increased as well. Less CO and more H2 selectivity arelso observed.

The flow rate can be computed by summing the product ofensity and mass fraction of gas species and velocity with theot product of the facet area vector and the facet velocity vec-or. Fig. 9a shows the flow rates of H2, CO and CO2 speciess a function of cluster porosity. With the increase of poros-ty, more gas passes through the cluster. This will result in thencrease of reaction rates. Hence with the increase of porosity

ess CO and more H2 selectivity are observed. Fig. 9b showshe flow rate of CH4 as a function of porosity. The molar ratiof H2/CO was increased since more CH4 are conversed into2 with the increase of porosity. Production of synthesis gas

a) porosity = 0.8, (b) porosity = 0.9 and (c) porosity = 0.95.

142 W. Shuyan et al. / Chemical Engineering Journal 139 (2008) 136–146

ough

f(tvttmr

3

oimp

Hwioarfo

3

taoaiw

bWbpdttcn

a

Fig. 9. Flow rates of gas species thr

rom methane and steam is influenced by the water-shift reactionB). Steam reforming of methane is reduced due to the produc-ion of CO. Performance of methane steam reforming will bearied by the formation of the cluster. Computed results showhat the produced gas species of an isolated particle is higherhan any individual particle in the cluster. This indicates that the

ethane reforming was restrained by particle clustering in CFBeformers.

.3. Effect of gas temperature

Increasing the operating temperature has a positive effectn the reforming reaction of methane since methane reform-ng is highly endothermic. Results of methane conversion andass fraction of gases as a function of operating temperature are

resented in Fig. 10a.The effect of operating gas temperature on molar ratio of

2/CO is shown in Fig. 10b. The molar ratio of H2/CO decreasesith the increase of operating temperature. The methane reform-

ng reaction is dominant by the inlet gas temperature. On thether hand, the reverse reaction of water shifting is promoted

t high temperature. From simulations, the decrease of H2/COatios with increasing inlet gas temperature was observed. There-ore, the CFB can help to externally heat to raise the temperaturef the reformer for achieving better reaction selectivity.

mt

i

Fig. 10. Produced mass fluxes of gas species through th

the cluster as a function of porosity.

.4. Effect of inlet gas velocity

With the increase of inlet gas velocity, more gases can flowhrough the cluster. Fig. 11a shows the flow rates of H2, COnd CO2 as a function of inlet gas velocities. The flow ratesf gas species increase with the increase of inlet gas velocitys expected. More CH4 is found in the cluster with more CH4ntroduced from the inlet. Yields of H2, CO and CO2 are alteredith the inlet gas velocity.The reformed methane flux as a function of inlet gas velocity

y the cluster and the isolated particle are shown in Fig. 11b.ith the increase of inlet gas velocity, the reformed methane

y the cluster is monotonically increased but for an isolatedarticle the reformed methane flux increases to a point, and thenecreases as the inlet gas velocity increases. This may be dueo the more flow passing the particle if more flow is introducedhe riser. Note that if the inlet gas velocity is high enough theluster breakage may occur. However, the cluster breakage isot considered in present investigation.

The molar ratios of H2/CO as a function of inlet gas velocityre shown in Fig. 11b. As the inlet gas velocity increases, the

olar ratio is increased monotonously for both the cluster and

he isolated particle.Experiments and simulations have shown that the gas velocity

s high in the riser center and low near the riser walls. With

e cluster as a function of operating temperature.

W. Shuyan et al. / Chemical Engineering Journal 139 (2008) 136–146 143

mass fluxes of gas species and ratio of H2/CO.

uiFltufttrs

3

rtrat

ta

Ft

Fig. 13. Computed ratio of H2/CO at two different inlet gas velocity boundaryconditions.

Fig. 11. Influence of inlet gas velocity on

niform inlet gas velocity, the gas flow in the cluster locatedn the center of the riser is concavely distributed, as shown inig. 12. If the inlet gas velocity is linearly increased from the

eft to right of the riser and with same amount inlet flow rate,he shape of the flow distribution is still close to the case withniform inlet velocity, but the H2/CO ratio is slightly reducedor the case of non-uniform inlet gas velocity compared withhe case of uniform inlet velocity, as shown in Fig. 13. Thus,he effect of inlet gas velocity distribution on methane steameforming reaction of the cluster will not be important in theimulations.

.5. Effect of reactor pressure

The rates of methane reforming reaction were related witheactor pressure. Fig. 14 shows the reaction rates r1, r2 and r3 ofhe cluster as a function of reactor pressure. With the increase ofeactor pressure, the reaction rates of reaction (A), (B) and (C)re decreased. Thus, the reaction will be rapidly decreased with

he increase of reactor pressure.

The effect of reactor pressure on the concentration distribu-ions of H2, CO and CO2 are shown in Fig. 15a. The reactionsre thermodynamically favored at low pressure according to Le

ig. 12. Influence of inlet gas velocity boundary conditions on axial gas velocityhrough the centerline of cluster. Fig. 14. Reaction rates of the cluster as a function of reactor pressure.

144 W. Shuyan et al. / Chemical Engineering Journal 139 (2008) 136–146

mola

Ca

swm

3

tsgAdfsft

swTp

masTti

3

h4Dntaif

Fig. 15. Profiles of mass fluxes of gas species and

hatelier’s principle [38]. Therefore, the flow rates of H2, COnd CO2 decrease with the increase of reactor pressure.

Effect of the reactor pressure on the molar ratio of H2/CO ishown in Fig. 15b. It is shown that the ratios of H2/CO increasedith the increase of reactor pressure. The high pressure is ther-odynamically unfavorable to the methane reforming.

.6. Effect of H2O/CH4

Computed flow rates of gas species with various feed steamo methane ratios are shown in Fig. 16a. The same trends of gaspecies in the cluster and the isolated particle are observed. At aiven temperature, the catalytic activity of Haldor Topsoe Ni/Mgl2O4 spinal catalyst particles with respect to CH4 mass fluxesecreased with the increase of the inlet steam to methane ratiosrom 2.0 to 4.0. In general, with more steam fed, more methanehould be reformed. However, an optimum steam-feeding rateor the maximum methane conversion has to be evaluated sincehe steam generation is an expensive process.

As shown in Fig. 16b, the H2/CO ratios increase with the

team to methane ratios from 2.0 to 4.0. Similarly, reformingith the cluster and the isolated particle alter the H2/CO ratios.he steam may reduce the carbon deposition onto the catalystarticles [39]. The carbon deposited on the catalysts can be dra-

C

2

Fig. 16. Profiles of mass fluxes of gas species and molar ratio o

r ratio of H2/CO as a function of reactor pressure.

atically reduced with the addition of steam. Thus, steam playsn important role in the effective removal of carbon. However, astated earlier, more steam will need more energy consumption.o produce hydrogen efficiently and economically, the steam

o methane feed ratio has to keep in an economical value inndustrial applications.

.7. Effect of coke formation

Traditionally, methane steam reforming is carried out atigh temperatures (around 800 ◦C), pressures between 1.6 and.1 MPa and steam to methane ratios within a range of 2–4.eactivation of nickel catalysts by carbon formation is a sig-ificant issue in methane steam reforming caused by fouling ofhe Ni surface, blockage of the pores of the catalytic particlend disintegration of the support material [39]. Thermodynam-cally, the most probable reactions for carbon formation are theollowing equations [40–41]:

CH4 cracking

H4 ↔ C + 2H2 �H298 = 122.3 kJ/mol (D)

Boudouard

CO ↔ C + CO2 ΔH298 = −125.2 kJ/mol (E)

f H2/CO at three different feed steam and methane ratios.

neering Journal 139 (2008) 136–146 145

C

TaS

r

w

r

r

D

Crae

k

k

K

K

K

caOttAt

fcwar1

tiw

4

icmtmCcwp

isupocTC

A

oNp

R

W. Shuyan et al. / Chemical Engi

CO reduction

O + H2 ↔ C + H2O ΔH298 = −84.0 kJ/mol (F)

he net rate of carbon deposition (reactions (D)–(F)) is evalu-ted by means of the following kinetic expressions reported bynoeck et al. [42]:

C,net = rC,D + rC,(E+F)

DEN2C

(25)

here

C,D = kMKCH4

(pCH4 − p2

H2

KM,av

)(26)

C,(E+F) = kO

(KO,H2OpCO

K1

KM,av

− 1

KO,H2O

pH2O

PH2

)(27)

ENC = 1 +(

KCH4pCH4 + 1

Kr

p3/2H2

+ 1

KO,H2O

pH2O

PH2

)(28)

The term rC,D of Eq. (25) represents the carbon formation byH4 cracking and its gasification by H2 (reaction (D)), while

C,(E+F) groups the carbon formation by reactions (E) and (F),nd its gasification by CO2 and H2O. Applying the Arrheniusquation and van’t Hoff equation to these parameters

M = 55619 exp

(−65053

RTg

),

O = 1.67 × 109 exp

(−148916

RTg

)(29)

r = 251893 exp

(−95489

RTg

),

O,H2O = 17372 exp

(−60615

RTg

),

M,av = 2.47 × 107 exp

(−122325

RTg

)(30)

The values of these coefficients involved in Eqs. (26)–(29)an be found in [42]. The net rate of carbon formation has beenssumed deposition on the surface of each particle in the cluster.nce the model is solved and the gas composition and tempera-

ure profiles are known, the rc,net expression is used to calculatehe potential of carbon formation of each particle in the cluster.ccording to the kinetic criterion, if rc,net ≥ 0 at any position in

he reactor, carbon deposition occurs [42].Fig. 17 shows the H2/CO ratios with and without the coke

ormation at the cluster porosity of 0.8. Compared to thease without the coke formation, the H2/CO ratio is reduced

ith the coke formation. Since the reaction rates (A) and (C)

re decreased with the coke formation. The consumed CH4educed from 1.631 × 10−6 kg/s without the coke formation to.614 × 10−6 kg/s with the coke formation in the cluster. Due to

Fig. 17. Effect of coke formation on H2/CO ratios.

he Boudouard and the CO reduction, the CO2 concentration isncreased and more carbon produced. Thus, the coke formationill reduce the H2/CO ratio in the clusters.

. Conclusions

A mathematical model for predicting methane steam reform-ng of Haldor Topsoe Ni/Mg Al2O4 spinel catalyst particlesluster is proposed by coupling hydrodynamics with catalyticethane steam reactions. Flow behavior of gases in the clus-

er was investigated. The distributions of gas temperature andass fraction of gas species are predicted. The yields of H2,O and methane conversions increase with the increase of theluster porosity, temperature and inlet gas velocity, but decreaseith the increase of the steam to methane ratio and operatingressure.

The catalyst particle clustering strongly affects flow behav-or and methane steam reforming in a CFB reformer. Presentimulations provided a preliminary picture for the start of sim-lations of methane steam CFB reformers. To more accuratelyredict flow behavior on methane reforming, carbonation effectn reforming reactions should be considered. Detail models forluster formation and deformation are also need to be developed.hese models will be included in our future simulations for theFB reformers.

cknowledgement

This work was supported by Natural Science Foundationf China through grant nos. 50620120442 and 20606006 andSFC-Petro China Company Limited under the cooperativeroject no. 20490200.

eferences

[1] J.N. Armor, Review: the multiple roles for catalysis in the production ofH2, Appl. Catal. A: Gen. 176 (1999) 159–176.

[2] A. Effendi, Z.G. Zhang, K. Hellgardt, K. Honda, T. Yoshida, Steam reform-ing of a clean model biogas over Ni/Al2O3 in fluidized- and fixed-bedreactors, Catal. Today 77 (2002) 181–189.

1 ineeri

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

46 W. Shuyan et al. / Chemical Eng

[3] G. Barbieri, D.M. Francesco, Simulation of the methane steam reformingprocess in a catalytic Pd-membrane reactor, Ind. Eng. Chem. Res. 36 (1997)2121–2127.

[4] M.E.E. Abashar, Coupling of steam and dry reforming of methane in cat-alytic fluidized bed membrane reactors, Int. J. Hydrogen Energy 29 (2004)799–808.

[5] J.R. Rostrup-Nielsen, Coking on nickel catalysts for steam reforming ofhydrocarbons, J. Catal. 33 (1974) 184–201.

[6] A.M. Groote, G.F. Froment, Simulation of the catalytic partial oxidationof methane to synthesis gas, Appl. Catal. A: Gen. 138 (1996) 245–264.

[7] U. Olsbye, O. Moen, U.A. Slagtern, I.M. Dahl, An investigation of the cok-ing properties of fixed and fluid bed reactors during methane-to-synthesisgas reactions, Appl. Catal. A: Gen. 228 (2002) 289–303.

[8] K. Tomishige, Syngas production from methane reforming with CO2/H2Oand O2 over NiO–MgO solid solution catalyst in fluidized bed reactors,Catal. Today 89 (2004) 405–418.

[9] K. Johnsen, H.J. Ryu, J.R. Grace, C.J. Lim, Sorption-enhanced steamreforming of methane in fluidized bed reactor with dolomite as CO2-acceptor, Chem. Eng. Sci. 61 (2006) 1195–1202.

10] A.M. Adris, C.J. Lim, J.R. Grace, The fluidized-bed membrane reactor forsteam methane reforming: model verification and parametric study, Chem.Eng. Sci. 52 (1997) 1609–1622.

11] S. Roy, B.B. Pruden, A.M. Adris, C.J. Lim, J.R. Grace, Fluidized-bedsteam methane reforming with oxygen input, Chem. Eng. Sci. 54 (1999)2095–2102.

12] T. Wurzel, S. Malcus, L. Mleczko, Reaction engineering investigationsof CO2 reforming in a fluidized bed reactor, Chem. Eng. Sci. 55 (2000)3955–3966.

13] K. Tomishige, Y. Matsuo, Y. Sekine, K. Fujimoto, Effective methanereforming with CO2 and O2 under pressurized condition using NiO-MgOand fluidized bed reactor, Catal. Commun. 2 (2001) 11–15.

14] M.A. Rakib, K.I. Alhumaizi, Modeling of a fluidized bed membranereactor for the steam reforming of methane: Advantages of oxygen addi-tion for favorable hydrogen production, Energy Fuels 19 (2005) 2129–2135.

15] P. Prasad, S.S.E.H. Elnashaie, Novel circulating fluidized bed membranereformer for the efficient production of ultraclean fuels from hydrocarbons,Ind. Eng. Chem. Res. 41 (2002) 6518–6527.

16] Z. Chen, Y. Yan, S.S.E.H. Elnashaie, Novel circulating fast fluidized bedmembrane reformer for efficient product ion of hydrogen from steamreforming of methane, Chem. Eng. Sci. 58 (2003) 4335–4349.

17] J. Grace, A.E.M. Adris, T. Boyd, C.J. Lim, Hydrogen from an internallycirculating fluidized bed membrane reactor, Int. J. Chem. Reactor Eng. 3(2005) 1–12.

18] S.S.E.H. Einashaie, P. Prasad, Z. Chen, Static bifurcation characteristics ofan autothermal circulating fluidized bed hydrogen generator for fuel cells,Ind. Eng. Chem. Res. 44 (2005) 4871–4883.

19] P. Prasad, S.S.E.H. Elnashaie, Coupled steam and oxidative reformingfor hydrogen production in a novel membrane circulating fluidized-bed

reformer, Ind. Eng. Chem. Res. 42 (2003) 4715–4722.

20] K. Kunii, O. Levenspiel, Fluidization Engineering, second ed., ButterworthHeinemann, MA, USA, 1991.

21] C.J. Chen, Experiments that address phenomenological issues in fast flu-idization, Chem. Eng. Sci. 54 (1999) 5529–5539.

[

ng Journal 139 (2008) 136–146

22] A.K. Sharma, K. Tuzla, J. Matsen, J.C. Chen, Parametric effects of particlesize and gas velocity on cluster characteristics in fast fluidized beds, PowderTechnol. 111 (2000) 114–122.

23] S.V. Manyele, J.H. Parssinen, J.X. Zhu, Characterizing particle aggregatesin a high-density and high-flux CFB riser, Chem. Eng. J. 88 (2002) 151–161.

24] M. Horio, H. Kuroki, Three-dimensional flow visualization of dilutely dis-persed solids in bubbling and circulating fluidized beds, Chem. Eng. Sci.49 (1994) 2413–2421.

25] V.D. Moortel, E. Azario, R. Santini, L. Tadrist, Experimental analysis ofthe gas-particle flow in a circulating fluidized bed using a phase Dopplerparticle analyzer, Chem. Eng. Sci. 53 (1998) 1883–1899.

26] P.D. Noymer, L.R. Glicksman, Cluster motion and particle-convective heattransfer at the wall of a circulating fluidized bed, Int. J. Heat Mass Transfer41 (1998) 147–158.

27] C. Guenther, R. Breault, Wavelet analysis to characterize cluster dynamicsin a circulating fluidized bed, Powder Technol. 173 (2007) 163–173.

28] G.F. Froment, K.B. Bischoff, Chemical Reactor Analysis and Design, sec-ond ed., Wiley, New York, 1990.

29] V.V. Ranade, Computational Flow Modeling for Chemical Reactor Engi-neering, Plenum Press, New York, 2001.

30] J. Xu, G.F. Froment, Methane steam reforming, methanation and water-gasshift: I. Intrinsic kinetics, AIChE J. 35 (1989) 88–96.

31] S. Grevskott, T. Rusten, M. Hillestad, E. Edwin, O. Olsvik, Modelling andsimulation of a steam reforming tube with furnace, Chem. Eng. Sci. 56(2001) (2001) 597–603.

32] K. Jarosch, T.E. Solh, H.I. de Lasa, Modelling the catalytic steam reformingof methane: discrimination between kinetic expressions using sequentiallydesigned experiments, Chem. Eng. Sci. 57 (2002) 3439–3442.

33] F. Gallucci, L. Paturzo, A. Basile, A simulation study of the steam reformingof methane in a dense tubular membrane reactor, Int. J. Hydrogen Energy29 (2004) 611–617.

34] Z. Chen, S.S.E.H. Elnashaie, Steady-state modeling and bifurcation behav-ior of circulating fluidized bed membrane reformer–regenerator for theproduction of hydrogen for fuel cells from heptane, Chem. Eng. Sci. 59(2004) 3965–3979.

35] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, McGraw-Hill, NewYork, 1998.

36] Z. Yunhau, L. Huilin, H. Yurong, J. Ding, Y. Lijie, Numerical prediction ofcombustion of carbon particle clusters in a circulating fluidized bed riser,Chem. Eng. J. 118 (2006) 1–10.

37] J.C. Moran, L.R. Glicksman, Experimental and numerical studies on thegas flow surrounding a single cluster applied to a circulating fluidized bed,Chem. Eng. Sci. 58 (2003) 1879–1886.

38] I.N. Levine, Physical Chemistry, McGraw-Hill, New York, 2002.39] D.L. Trimm, Coke formation and minimization during steam reforming

reactions, Catal. Today 37 (1997) 233–238.40] J.R.R. Nielsen, Catalytic steam reforming, Catal. Sci. Technol. 1 (1984)

1–117.41] M.N. Pedernera, J. Pina, D.O. Borio, Kinetic evaluation of carbon formation

in a membrane reactor for methane reforming, Chem. Eng. J. 134 (2007)138–144.

42] J.W. Snoeck, G.F. Froment, M. Fowles, Kinetic evaluation of carbon forma-tion in steam/CO2-natural gas reformers: influence of the catalyst activityand alkalinity, Int. J. Chem. Reactor Eng. A7 (2003) 1–16.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具