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 33 (2008) 2679 – 2694

Available at www.sciencedirect.com

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

Some issues in modelling methane catalytic

decomposition in fluidized bed reactors

P. Ammendolaa, R. Chironea, G. Ruoppoloa, G. Russoa,b, R. Solimenea,

a
Istituto di Ricerche sulla Combustione, CNR, Piazzale V. Tecchio, 80, 80125 Napoli, Italy
b
Dipartimento di Ingegneria Chimica, Università di Napoli Federico II, Piazzale V. Tecchio, 80, 80125 Napoli, Italy

art i cle info ab st rac t

Article history: This paper reports a model of fluidized bed thermo-catalytic decomposition (TCD) of
Received 29 March 2007 methane. The novelty of the model consists of taking into account the occurrence of
Received in revised form different competitive phenomena: methane catalytic decomposition, catalyst deactivation
4 January 2008 due to carbon deposition on the catalyst particles and their reactivation by means of carbon
Accepted 23 March 2008 attrition. Comparison between theoretical and experimental data shows the capability of
Available online 12 May 2008 the present model to predict methane conversion and deactivation time during the process.
The model demonstrates to be also a useful tool to investigate the role played by operative
Keywords:
parameters such as fluidizing gas velocity, temperature, size and type of the catalyst.
Methane catalytic decomposition
In particular, the model results have been finalized to characterize the attrition phenomena
Fluidized bed reactor
as a novel strategy in catalyst regeneration.
Hydrogen production
& 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights
Attrition phenomena
reserved.

1. Introduction not be obtained using nickel-based catalysts [6]. On the

contrary, Fe-based catalysts are more stable at higher
The thermo-catalytic decomposition (TCD) of methane is temperatures in the order of 700–1000 1C, but deactivation
an attractive process towards the production of hydrogen occurs upon repeated cycles, resulting in a short lifetime [6].
with reduced CO2 emissions [1–3]. The use of a catalyst is In addition, whatever the catalyst, deposited carbon generally
extremely advantageous since the non-catalytic thermal has a filamentous shape, i.e. nanofibres or nanotubes, with
decomposition would require elevated process temperatures, metal particles on their tips. This is an undesirable feature
i.e. above 1200 1C [1,2]. Fluidized beds have been indicated in with reference to applications in fluidized bed reactors where
the last years as an efficient reactor solution for TCD process attrition is likely to remove the active metal phase away from
compared to fixed bed reactors [1–5]. Consequently, new the support, resulting in a decrease in hydrogen production
criteria have to be taken into account for catalyst design. In and in an increase in consumption of metals [2].
particular, a suitable catalyst should be characterized by a low Muradov et al. [1–3] have investigated the feasibility of
propensity to attrition in addition to the requirements of high using both activated carbon and carbon black as catalysts for
thermal stability and conversion efficiency [6]. methane decomposition in fluidized beds. They pointed out
Catalytic systems containing Ni [7,8] and Fe [9,10] have been that the process is more advantageous than that involving a
largely tested in the past. The Ni-based catalysts have a metal catalyst: low cost, high temperature resistance and
maximum operative temperature of 600 1C. As a consequence, tolerance to potentially harmful compounds. On the other
methane conversion being thermodynamically limited at this hand, the drawbacks are: (i) the poor catalytic activity of
temperature, concentrated hydrogen streams (H2460%) can- carbon in comparison with metal catalyst; (ii) the carbon

Corresponding author. Tel.: +39 0817682237; fax: +39 0815936936.

E-mail address: [email protected] (R. Solimene).
0360-3199/$ - see front matter & 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2008.03.033
 ARTICLE IN PRESS
2680 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 33 (2008) 2679 – 2694

Nomenclature Si2 internal catalyst surface occupied by deposited

carbon, m2
a catalyst specific surface area, m2/g t time, s
co inlet methane concentration, mol/m3 td deactivation time, s
cb methane concentration in the bubble phase, T temperature, K
mol/m3 U fluidizing gas velocity, m/s
ce methane concentration in the dense phase, Ub rise velocity of bubbles through the bed, m/s
mol/m3 Ubr rise velocity of isolated bubbles, m/s
db0 bubble diameter just above the distributor, m Umf minimum fluidization velocity, m/s
dbm maximum bubble diameter, m v generic input variable, a.u.
db effective bubble diameter at x*, m Vc volume of a catalyst particle, m3
dc catalyst particle diameter, m Wc amount of carbon deposited on the external
dp mean pore diameter of a catalyst particle, nm surface of all bed particles, g
D mean methane diffusivity, m2/s wec amount of carbon deposited on the external
DCH4 methane diffusivity, m2/s surface of a catalyst particle, g
Deff methane bulk diffusivity, m2/s wic amount of carbon deposited on the internal
Dk methane Knudsen diffusivity, m2/s surface of a catalyst particle, g
Dr bed diameter, m x distance above the distributor, m
E1 catalyst activation energy, J/mol x* distance above the distributor corresponding to
E2 deposited carbon activation energy, J/mol the calculation of db , m
Ec carbon elutriation rate, g/s yCu copper mass fraction of catalyst particles, g/g
hcd thickness of the carbon deposits, nm z generic output variable, a.u.
H bubbling fluidized bed height, m
Hmf static bed height, m Greek symbols
k1 surface intrinsic kinetic constant of the catalyst,
m3/(m2s) e fractional change in volume of the system be-
k01 catalyst frequency factor, m3/(m2s) tween no conversion and complete conversion of
k2 surface intrinsic kinetic constant of the deposited methane, –
carbon, m3/(m2s) ec porosity factor of a catalyst particle, –
k02 deposited carbon frequency factor, m3/(m2s) emf void fraction in the bed at minimum fluidization
ka carbon attrition constant, m1 condition, –
kg mass transfer coefficient outside the particle, m/s rb bulk density of bed of catalyst particles, g/m3
kp apparent reaction constant per volume of catalyst rc apparent catalyst density, g/m3
particle, s1 t tortuosity factor of a catalyst particle, –
kep apparent external reaction constant per volume tg mean gas residence time, s
of catalyst particle, s1
kip apparent internal reaction constant per volume of Dimensionless numbers
catalyst particle, s1
Kbe overall coefficient of gas interchange between Da Damköhler number
bubble and dense phase, s1 etd model relative error of deactivation time
Kbc coefficient of gas interchange between bubble and eXi model relative error of initial methane conversion
cloud-wake region, s1 eXr model relative error of residual methane conver-
Kce coefficient of gas interchange between cloud- sion
wake region and dense phase, s1 Remf Reynolds number at minimum fluidization con-
MC carbon molecular weight, g/mol dition
MCu copper molecular weight, g/mol Sc Schimdt number
MCH4 methane molecular weight, g/mol Sh Sherwood number
MH2 hydrogen molecular weight, g/mol Sz sensitivity of a generic output variable
Nc number of catalyst particles, – X methane conversion
PC carbon production rate, g/s Xbe index of gas exchange between bubbles and
PH2 hydrogen production rate, g/s dense phase
Scd surface of catalyst particle occupied by deposited Xi initial methane conversion
carbon per carbon mass, m2/g Xr residual methane conversion
Se total external surface of a catalyst particle, m2 b fluidization velocity excess with respect to the
Si total internal surface of a catalyst particle, m2 fluidizing gas velocity
Se1 external active surface of a catalyst particle, m2 Y Thiele number
Si1 internal active surface of a catalyst particle, m2 Z effectiveness factor of a catalyst particle
Se2 external catalyst surface occupied by deposited
carbon, m2
 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 33 (2008) 2679 – 2694 2681

catalyst must be periodically unloaded because the carbon methane catalytic decomposition, catalyst deactivation due
produced by the TCD process has a lower activity than that of to carbon deposition on the catalyst and catalyst reactivation
the initial catalyst. However, the authors pointed out the by means of carbon attrition. A main feature of the model is
feasibility of using fluidized bed reactors and carbon as related to taking into account the positive effect of attrition
catalyst on the basis of a scale-up model [3]. phenomena, commonly considered as a disadvantage of the
Ammendola et al. [11] have proposed a copper-based catalytic fluidized bed processes, on the regeneration of the
catalyst. Advantages of this catalytic system are: (i) relatively active external surface of catalyst particles. The hypotheses
high catalytic activity over that reported for carbon- and made in the model formulation have been validated by
nickel-based catalysts; (ii) high operative temperature up to comparing theoretical results and experimental data ob-
1000 1C; (iii) relatively high mechanical resistance to attrition. tained by Ammendola et al. [12]. A sensitivity analysis of
This catalyst is a good candidate to be used in fluidized bed the model has been carried out with reference to operative
applications, considering that, in addition to a low attrition parameters (properties of catalyst particles and of carbon
rate of the catalyst, it is also characterized by carbon deposits) which have been considered relevant for the
deposition without formation of carbon fibres with the metal uncertainties on their value. The model has been also used
particles on their tips. This catalytic system has been to investigate the role played by operating conditions such as
extensively studied in a laboratory-scale fluidized bed reactor fluidizing gas velocity, size and type of the catalyst and to
[12] in terms of methane to hydrogen conversion, amount of highlight the operating conditions which yield the attrition
carbon accumulated on the catalyst, effectiveness of attrition phenomena useful for the catalyst regeneration.
in removing deposited carbon produced by methane TCD
process and deactivation time.
In the present paper a model of methane TCD carried out in 2. Mathematical model
bubbling fluidized bed reactors is presented. The model has
been formulated taking into account the fate of catalytic A block diagram of the model is shown in Fig. 1. The model is
particles subjected to different competitive phenomena: structured in three sub-models, which account for the main

catalyst
temperature properties
CH4 inlet kinetic of CH deposited carbon
FB 4
hydrodynamicsconcentration decomposition properties

methane
methane methane conversion

catalytic
catalytic hydrogen production
decomposition
decomposition deactivation time

total active amount of

catalyst surface produced carbon

internal active
catalyst surface carbon
carbon
catalyst pore distribution
deposition

amount of carbon
external active
deposited on external
catalyst surface
catalyst surface

carbon attrition constant

carbon
carbon
removal
removal
carbon elutriation rate

Fig. 1 – Block diagram of the model.

 ARTICLE IN PRESS
2682 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 33 (2008) 2679 – 2694

concurring phenomena: catalytic decomposition of methane, sions as carbon fines transported in the exit gasses as well as
catalyst deactivation due to carbon deposition on catalyst for renewal of a part of external active surface of the catalyst
active surface and removal of deposited carbon by attrition. particle. According to the scheme it has been considered that
The input variables of the first block are the operating constant values of H2 production and carbon elutriation rate,
conditions, the fluidized bed hydrodynamics and the catalyst Ec, can be obtained. This means that carbon attrition rate is
and deposited carbon properties. Time by time the first block able to balance carbon deposition rate. Under this condition
needs the values of the actual active surface of the catalyst there is only a partial carbon recovering of the external
particle to calculate the reaction rate and, in turn, the surface of the catalyst particle. Fig. 2A reduces to Fig. 2B when
methane conversion and the hydrogen production as a attrition is not present. In this case a full recovering of the
function of time. Accordingly, it is also possible to calculate catalyst is obtained and, due to a deposited carbon activity
if the deactivation occurs and, in this case, the deactivation significantly lower than metal activity, a relatively low value
time of catalyst particles. The second block requires as input of H2 production is expected.
variable the catalyst pore size distribution and, considering
the actual active surface of the catalyst, it calculates the 2.1. Model assumptions
carbon deposited on the internal and the external active
surfaces of the catalyst particles. The third block determines The main assumptions concerning hydrodynamics, methane
the amount of carbon removed by attrition from the external decomposition, carbon deposition and carbon removal by
surface of the catalyst particles which is the elutriated attrition are, separately, detailed below.
amount of carbon in the flue gas.
The model has been developed by considering the beha- 2.1.1. Hydrodynamics
viour of a single catalytic particle during the TCD of methane The well-sound two-phase theory of fluidization [13] has been
in a bubbling fluidized bed and by integrating this contribu- considered to model the hydrodynamics of the fluidized bed
tion to the total number of particles forming the bed to obtain reactor. Accordingly, two different phases are distinguished
the performance of the reactor. A schematic representation of (Fig. 3): the dense phase which contains all the bed material
the single particle phenomena is shown in Fig. 2A. At t ¼ 0, and the bubble phase which consists of the swarm of bubbles
the fresh catalyst is represented as uniformly covered by effectively present inside the bed. The two-phase theory of
metallic active sites. At t40, methane decomposes producing fluidization establishes that a gas flow rate corresponding to
hydrogen and carbon. The carbon deposits on the particle the incipient fluidization condition percolates through the
active surface, determining a decrease in particle activity and, dense phase whereas the remaining part of the total gas flow
in turn, reducing methane decomposition rate. In parallel rate rises along the bed as bubbles. The volume changes due
with these phenomena, attrition of carbon deposited on the to the methane conversion along the bed axis are considered
external surface of particle is responsible for carbon emis- proportional to the methane conversion by means of a factor,

Fig. 2 – A conceptual representation of thermo-catalytic decomposition of methane on a single bed particle. (A) Presence of
attrition phenomena; (B) absence of attrition phenomena.
 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 33 (2008) 2679 – 2694 2683

X The actual methane decomposition rate is calculated by

accounting for intrinsic kinetics of metal catalyst and
deposited carbon and their relative active surfaces.
x cb (t,H) ce(t)
The methane conversion has been assumed as a quasi-
Kbe steady state process with respect to variations of decomposi-
tion kinetics due to the catalyst deactivation.
Bubble Dense
phase phase
2.1.3. Carbon deposition
The carbon produced by methane decomposition is deposited
on the internal and external surfaces of the catalyst. Methane
cb (t, x) ce (t) decomposes on an active site whether it is carbon or metallic
without the formation of filamentous carbon. It has been also
U-Umf Umf assumed that the carbon deposits along the pores of catalyst
particle reducing the pore diameter, without reducing its
specific surface area. This assumption limits the validity
U, c0 of the model to cases far from conditions of pore occlusion.
The model has been used verifying that the reduction of pore
Fig. 3 – Scheme of the fluidized bed according to the two-
size was smaller than 1 nm with an initial average pore size of
phase theory of fluidization.
10 nm. The relative amount of carbon deposited on the
external and internal surfaces of the catalyst is calculated
on the basis of the respective methane decomposition rates,
e, defined as the fractional change in volume of the system which, in turn, are proportional to the actual CH4 concentra-
between no conversion and complete conversion of methane tion. CH4 profile inside particles has been calculated through
[14]. Mass transfer between the dense phase and the bubble Thiele number [14], which, on the basis of the previous
phase has been described by an overall mass transfer discussion on the limited variation in pore diameter, mainly
coefficient, Kbe. Solids and gases are perfectly mixed in the varies during the decomposition process due to the reduction
dense phase [15]. Bubbles rise in plug flow with a constant of catalyst activity caused by carbon deposition.
size along the bed axis [15]. The adopted bubble size is an
average value calculated by a semi-empirical correlation
2.1.4. Carbon removal
which takes into account the variation of the bubble size
The mechanism assumed for carbon removal is the attrition
due to bubble coalescence and to volume changes owing to
of the carbon accumulated on the external surface of
the occurrence of the methane conversion along the bed axis
fluidized particles. In particular, it has been assumed that
as a function of the operating conditions [16]. The tempera-
the mechanical abrasion generated by the motion of the bed
ture is uniform in the reactor [17] and constant with time.
particles enables one to remove the carbon deposited on the
Temperatures between 800 and 1000 1C are considered.
external surface of the catalyst particles restoring the
metallic active sites and producing carbon fines. These latter
2.1.2. Methane decomposition
are elutriated according to the correlation, proposed by Miccio
Decomposition of the methane (CH4-C+2H2) is the only
and Massimilla [23], valid for the production of carbon fines
occurring reaction taking place on the catalyst surface. The
due to the mechanical abrasion of carbon-spotted bed
reaction is irreversible, valid assumption at the considered
material:
temperatures, as determined by thermodynamical evalua-
tions [18], and first order with respect to methane concentra- Ec ¼ ka ðU  Umf ÞWc (1)
tion as reported in literature for metallic catalysts [19].
Carbon elutriation rate, Ec, is proportional to an attrition
Considering that the model is applied to operative tempera-
constant, ka, to the fluidization excess velocity, (U–Umf), and to
tures lower than 1000 1C, the homogenous gas phase methane
the carbon amount, Wc, actually present on the external
decomposition has not been taken into account. This
surface of the catalyst particles forming the whole bed. It
hypothesis, reasonable on the basis of literature indications
must be noted that the choice of Eq. (1) has been done on the
[1,2], has been also confirmed by calculations under the tested
basis of the nature of the interaction between copper-based
operative conditions.
catalyst and produced carbon, often described by van der
The carbon produced by methane decomposition covers the
Waals forces [24,25].
catalyst active sites reducing catalyst activity. However,
the deposited carbon also presents a catalytic activity to the
methane decomposition process even if much lower than
2.2. Model equations
that of metallic catalysts [20]. The reaction order with respect
to methane concentration for this chemical process is
On the basis of model assumptions, the model can be reduced
unknown. However, literature data show that activated
to the following set of equations:
carbons and carbon blacks are characterized by a reaction
order of 0.5 [21] and 1 [22], respectively. For the sake of inlet CH4 moles  outlet CH4 moles
Methane conversion ¼
simplicity, in this work a first reaction order is considered. inlet CH4 moles
 ARTICLE IN PRESS
2684 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 33 (2008) 2679 – 2694

½ð1 þ Þð1  b eXbe Þ þ DaðtÞð1 þ  eXbe Þ

XðtÞ ¼
ð1 þ Þð1  eXbe Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½ð1 þ Þð1  b eXbe Þ þ DaðtÞð1 þ  eXbe Þ2 þ 2 DaðtÞð1 þ Þð1  eXbe Þð1  b eXbe Þ
þ (2)
ð1 þ Þð1  eXbe Þ

Mass rate of hydrogen production 0 1 0 1

  Variation rate Carbon accumulation
inlet CH4 moles B C B C
¼ 2MH2 XðtÞ B of the internal C B rate on the internal C
time B C B C
B C ¼ Scd B C
pD2r B surface of a C B surface of a C
PH2 ðtÞ ¼ 2MH2 Uc0 XðtÞ (3) @ A @ A
4 catalyst particle catalyst particle
Carbon elutriation rate, dSi1 ðtÞ dwic ðtÞ
¼ Scd (8)
Ec ðtÞ ¼ ka ðUð1 þ 0:5XðtÞÞ  Umf Þwec ðtÞNc (4) dt dt

0 1 0 1 Methane conversion is time dependent through Damköhler

Carbon accumulation Carbon deposition
B C B C number dependence. In fact, according to its definition,
B rate on the external C B rate on the external C Da(t) ¼ f(kp(t)) [26], Damköhler number varies with time due
B C B C
B C¼B C
B surface of a C B surface of a C to catalyst deactivation as a result of the reduction of the
@ A @ A
catalyst particle catalyst particle apparent kinetic constant of catalyst, kp(t), owing to carbon
0 1 deposition on both external and internal surface of the
Carbon removal
B C particles. The apparent kinetics of the catalyst is evaluated
B rate by attrition C time by time during the deactivation process taking into
B C
B C
B from a C account: (i) the actual amount of carbon deposited on the
@ A
catalyst particle internal, wic ðtÞ, and external, wec ðtÞ, active surfaces of a single
catalyst particle; (ii) the actual active surfaces, either internal
dwec ðtÞ Ec ðtÞ or external, of the catalyst, Si1 ðtÞ and Se1 ðtÞ.
¼ kep ðtÞce ðtÞVc MC  (5)
dt Nc The full details of model equations and their derivation are
0 1 0 1 reported in Appendices A and B.
Carbon accumulation Carbon deposition
B C B C
B rate on the internal C B rate on the internal C
B C B C
B C¼B C
B
@ surface of a C B
A @ surface of a C
A
3. Results and discussion
catalyst particle catalyst particle
The input variables, assumed as the base case, are listed
dwic ðtÞ
¼ kip ðtÞce ðtÞVc MC (6) in Table 1 in terms of operating conditions, properties of
dt
catalytic bed material and properties of deposited carbon. The
0 1 0 1
Variation rate Carbon accumulation operating conditions are those of a typical experiment carried
B C B C out by Ammendola et al. [12]. The physical properties of the
B of the external C B rate on the external C
B C B C catalyst particles and their fluidization behaviour have been
B C ¼ Scd B C
B active surface of a C B surface of a C
@ A @ A obtained from previous characterization tests reported else-
catalyst particle catalyst particle where [11,12,18]. The evaluation of kinetic parameters of the
dSe1 ðtÞ dwec ðtÞ catalyst has been experimentally obtained in a fixed bed
¼ Scd (7)
dt dt micro-reactor at a pre-set methane concentration and varying

Table 1 – Input model variables (base case)

Operating conditions Catalyst properties Deposited carbon properties

U (m/s) 0.058 rc (g/m3) 20  105 Scd (m2/g) 5500

Hmf (m) 0.12 rb (g/m3) 8.6  105 hcd (nm) 0.335
Dr (m) 0.027 a (m2/g) 100 ka (m1) 102
c0 (mol/m3) 0.568 dc (m) 3.5  104 k02 (m3/(m2s)) 7
T (K) 1073 ec (–) 0.6 E2 (J/mol) 2.0  105
DCH4 (m2/s) 2  104 t (–) 3
dp (nm) 10
Umf (m/s) 0.029
emf (–) 0.57
yCu (g/g) 0.084
k01 (m3/(m2s)) 2.1
E1 (J/mol) 1.55  105
 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 33 (2008) 2679 – 2694 2685

the reaction temperature in the range 600–800 1C [18]. The decreases with time from an initial value Xi (ffi0.9) to a residual
data have been elaborated by means of an Arrhenius plot to value Xr (ffi0.2) at tdffi6500 s, followed by a second stage, in
estimate k01 (2.1 m3/(m2s)) and E1 (1.55  105 J/mol). It is worth which it is constant. The first stage is that expected considering
noting that the obtained value of the activation energy is that methane conversion decreases with time due to the
between that of metallic (0.6  105 J/mol) [27] and of carbon- carbon deposition and the corresponding decrease in catalyst
based catalysts (2.0  105 J/mol) [22] according to the effective activity. The second stage is due to two different and constant
intermediate activity of copper. The value of proportionality contributions: (i) the internal decomposition rate is constant
constant of Eqs. (7) and (8), Scd, represents the surface of because the internal catalyst surface, completely recovered by
catalyst particle occupied by deposited carbon per carbon carbon, does not vary during the carbon deposition according
mass. Assuming that one mole of carbon deactivates one to model hypothesis; (ii) the external decomposition rate due to
mole of copper, this parameter has been estimated by the activity of the accumulated carbon and to the activity of the
metallic catalyst sites renewed by attrition is constant because
catalyst particle surface carbon mole
Scd ¼ carbon attrition rate is able to balance external carbon
catalyst particle copper moles carbon mass
deposition rate yielding constant the external surface occupied
arc Vc MCu
¼ (9) by the carbon and by metallic catalyst.
rc Vc yCu MC
Carbon elutriation rate (Fig. 4) monotonically increases up
where a, rc, Vc, yCu, MC and MCu are the catalyst specific to a stationary value at tffi8000 s. Again this steady state
surface area, the apparent catalyst density, the volume of condition corresponds to the balance between the carbon
catalyst particle, the copper mass fraction of catalyst, the deposition rate on the external catalyst surface and the
carbon and copper molecular weight, respectively. Using the carbon attrition rate. The model has been used to calculate
catalyst data reported in Ammendola et al. [12], the estimated the relative contribution of external metal activity renewed by
value is in the order of 5000–6000 m2/g. The thickness of the attrition and activity of total deposited carbon on actual CH4
carbon deposits on Cu-based catalyst is unknown in litera- decomposition rate. Under the tested conditions, attrition
ture. However, NiCuAl catalysts have shown the formation of contributes to an amount lower than 1% of the residual value
high-order deposited carbon structurally close to a perfect of methane conversion.
graphite [28]. As a consequence, in the model it has been The predicted profiles, presented in Fig. 4, are in agreement
assumed equal to the thickness of an elementary cell of with the conceptual representation of methane TCD in a
graphite. The adopted value of ka has been determined bubbling fluidized bed presented in Fig. 2A.
working out experimental data [12,18] obtained collecting A better understanding of the mechanisms active during
the elutriated material during methane decomposition tests the fluidized bed TCD process can be obtained considering
and analysing their carbon content. The kinetic parameters the ratio between the active and the total external surface of a
of the carbon deposits have been chosen from literature data catalyst particle, Se1 =Se , the ratio between the active and the
concerning the activity of carbon blacks [22]. total internal surface of a catalyst particle, Si1 =Si , the carbon
Fig. 4 shows typical model output variables: methane amount deposited on the external, wec , and internal, wic ,
conversion and carbon production rate (solid line) and carbon surface of a single bed particle. Fig. 5 shows these variables as
elutriation rate (dashed line) as a function of time. Methane functions of time.
conversion profile, i.e. carbon production rate profile, is The active external surface of catalyst particle decreases
characterized by a first stage, in which it monotonically with time until, at tffi8000 s, a steady state condition is

1.0 25
20
carbon production rate, PC (105), g/s

carbon elutriation rate, EC (.109), g/s

0.8 20
methane conversion, X, -

16

0.6 15
12

8 0.4 10

4 0.2 5

0 0.0 0
0 2000 4000 6000 8000 10000
time, s

Fig. 4 – Model results (base case). Methane conversion X and carbon production rate Pc (solid line) and carbon elutriation rate
Ec (dashed line) as functions of time during methane TCD in a bubbling fluidized bed reactor.
 ARTICLE IN PRESS
2686 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 33 (2008) 2679 – 2694

approached corresponding to a value of about 13% of the total activity of deposited carbon. It has been verified that in the
external surface. Correspondingly, the amount of carbon considered time interval the accumulated carbon does not
accumulated on the external surface, wec , increases with time significantly modify the total internal surface of the catalyst
until a stationary value is reached at tffi8000 s corresponding particle.
to a value of about 6  1011 g. These results can be analysed
by taking into account that the carbon deposition rate is 3.1. Comparison between experimental and model results
initially much larger than the carbon attrition rate, propor-
tional to the amount of the deposited carbon, thus the active The developed model has been validated by comparing the
external surface of the catalyst particle decreases and the time series of methane conversion obtained during TCD tests
amount of deposited carbon increases until a stationary state in a lab-scale bubbling fluidized bed reactor [12,18] and the
is achieved. experimental values of the initial, Xi, and the residual, Xr,
The active internal surface of catalyst particle decreases methane conversion and of the deactivation time, td, with the
with time up to a complete covering of the surface at model computations in the same operating conditions.
tffi6500 s, corresponding to td, which can be considered the The experimental set-up, procedure and measurement tech-
‘‘primary’’ deactivation time. Contrarily to the external sur- niques of the experiments have been described elsewhere
face, being in this case absent carbon attrition, the amount of [12,18].
carbon accumulated on the internal surface, wic , increases Fig. 6 shows the comparison between the model and
even at t4td, but at a lower rate due to the low catalytic the experimental time-resolved methane conversion [12,18].

1.0 10
and total particle surface, S1e/Se, -

carbon deposited on external

particle surface, wCe (.1011), g
ratio between external active

0.8 8

0.6 6

0.4 4

0.2 2

0.0 0

1.0 10
and total particle surface, S1i/Si, -

carbon deposited on internal

ratio between internal active

particle surface, wCi (.107), g

0.8 8

0.6 6

0.4 4

0.2 2

0.0 0
0 2000 4000 6000 8000 10000
time, s

Fig. 5 – Model results (base case). (A) Ratio of the active to the total external surface of a catalyst particle Se1 =Se and carbon
amount deposited on the external surface of a catalyst particle wec as functions of time; (B) ratio of the active to the total
internal surface of a catalyst particle Si1 =Si and carbon amount deposited on the internal surface of a catalyst particle wic as
functions of time.
 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 33 (2008) 2679 – 2694 2687

1.0 In particular, the conditions are the same as the base case
experimental data Hmf = 0.07 m except for the fluidizing gas velocity which is 0.087 m/s.
model computations
The influence of bubbling fluidized bed hydrodynamics is also
0.8
investigated considering three different bed heights: 0.07, 0.10
and 0.12 m. The agreement between the model computations
0.6 and the experimental data is quite good during both the
initial and the final step of the deactivation process.
0.4 Table 2 and Fig. 7 present the comparison between model
predictions and experimental results, obtained with different
superficial gas velocities and bed heights [12,18], in terms
0.2
of Xi, Xr, td and their relative errors. The model predictions are
in good agreement with the experimental data whatever
0.0 operative conditions are considered. As a matter of fact, the
relative error is lower than 710% in most of the investigated
1.0
conditions.
Hmf = 0.10 m
methane conversion, X, -

0.8
3.2. Sensitivity analysis
0.6
The model has been also used to investigate the influence of
properties of catalyst and of deposited carbon on the
0.4
performance of the fluidized bed reactor (Fig. 8). In particular,
the specific surface area, a, and the intrinsic reaction
0.2 constant, k01 , of the catalyst, as well as the surface of catalyst
particle occupied by deposited carbon per carbon mass, Scd,
have been increased and decreased with respect to the
0.0
base case. The study has been further pursued in order to
1.0 determine the sensitivity of relevant output variables to
Hmf = 0.12 m changes in these input variables. To this end, the standard
0.8 procedure for linearized sensitivity [29] has been used. Each
input variable has been changed by some fraction (720%) of
their values assumed as base case. This variation has been
0.6 fixed on the basis of either uncertainties in their experimental
determination or intentional changes with respect to base
0.4 case due to operative requirements. The sensitivity of the
generic output variable z has been evaluated as

0.2 ðz  zþ Þ=zb

Sz ¼ (10)
ðv  vþ Þ=vb

0.0 where subscript b indicates the base case value. Superscripts

0 1000 2000 3000 4000 5000 6000
and + indicate, for the generic input variable v, the left and
time, s
the right extremes of the assumed range of variation,
Fig. 6 – Model computations: comparison with experimental whereas for the output variable z they indicate the values
data obtained under different bed heights (U ¼ 0.087 m/s).

Table 2 – Comparison between experimental and model results obtained under different operating conditions

Operating conditions Experimental values Model values Relative errors

U (m/s) Hmf (m) Xi (–) Xr (–) td (s) Xi (–) Xr (–) td (s) eXi (%) eXr (%) etd (%)

0.058 0.07 0.87 0.11 4780 0.84 0.11 4364 3.5 0 8.7
0.12 0.91 0.16 6760 0.91 0.18 6424 0 12.5 5.0

0.087 0.07 0.79 0.08 2980 0.78 0.07 3504 1.3 8.7 17.6
0.10 0.82 0.11 4280 0.84 0.11 4398 2.4 0 2.7
0.12 0.84 0.09 4350 0.86 0.12 4880 2.4 33.3 12.1

0.116 0.07 0.75 0.07 2850 0.72 0.06 3076 4.0 14.3 7.9
 ARTICLE IN PRESS
2688 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 33 (2008) 2679 – 2694

td experimental, s 1.0
0 2000 4000 6000 8000 10000 k10 = 1.68 m3/m2s
1.0 10000 k10 = 2.10 m3/m2s*
model methane conversion, -

0.8 k10 = 2.52 m3/m2s

Xi
0.8 Xf 8000
td
0.6

td model, s
0.6 6000

0.4
0.4 4000

0.2 2000 0.2

0.0 0 0.0
0.0 0.2 0.4 0.6 0.8 1.0
experimental methane conversion, - 1.0
a = 80 m2/g
a = 100 m2/g*
Fig. 7 – Model computations: comparison with experimental a = 120 m2/g
0.8

methane conversion, -
data obtained under different operating conditions (dashed
lines 710%).
0.6

that it assumes for these extremes. The sensitivity values 0.4

obtained are reported in Table 3.
The catalyst intrinsic kinetic constant plays a twofold role:
on one hand, increasing the kinetics increases the initial 0.2
conversion of the methane; on the other, the deactivation
time is smaller and the steady state condition is achieved in a 0.0
step-like manner.
1.0
The influence of the specific surface area of the catalyst Scd = 4400 m2/g
particle on the methane conversion is rather trivial: the initial Scd = 5500 m2/g*
conversion as well the deactivation time increases with 0.8 Scd = 6600 m2/g
the total internal surface of the particle. The surface of
catalyst particle occupied by deposited carbon per carbon
0.6
mass, Scd, is a characteristic parameter of the physical and
chemical structure of the carbon deposits and of its interac-
tion with metallic catalyst sites. It has a crucial effect on 0.4
the deactivation process enabling possibly a large variation
of the deactivation time. These considerations are confirmed 0.2
by the parametric analysis reported in Table 3. In particular,
it is evident that: (i) the initial methane conversion is
0.0
barely influenced positively by the intrinsic kinetics and
0 2000 4000 6000 8000 10000
by the specific area of the catalyst, k01 and a, respectively;
time, s
(ii) the residual methane conversion depends almost linearly
on the specific area of the catalyst, a; (iii) the deactivation Fig. 8 – Model computations: influence of the properties of
time is subject to the variations of the three investigated catalyst and of deposited carbon. (*) continuous line base
input variables. On the whole, it can be observed that case.
the deactivation process is strongly influenced by the
specific area of the catalyst, a, and by the surface of
catalyst particle occupied by deposited carbon per carbon
mass, Scd.
TCD process in a large-scale operation (bed diameter about
4 m). In particular, the present investigation has been
3.3. Relevance of attrition phenomena carried out at different catalyst particle sizes 3.5  104
and 0.8  104 m, fluidizing gas velocities 0.1 and 0.5 m/s,
The model has been also used to highlight the feasibility that mean gas residence times tg ¼ 5 and 10 s and temperatures
attrition can operate as a useful process in catalyst regenera- 800 and 1000 1C. It is worth noting that at higher considered
tion for industrial applications. The model computations temperature the contribution of the homogeneous gas
have been obtained disregarding the activity of deposited phase reaction to methane decomposition calculated by
carbon in order to emphasize the role of carbon attrition Chemkins simulations is less than 3% of methane conversion
and choosing the values of the operating conditions on the introducing, as a consequence, a negligible error in the model
basis of results achieved by Muradov et al. [3] for methane results.
 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 33 (2008) 2679 – 2694 2689

Table 3 – Sensitivity of output variablesa to changes of individual input variables

Input variable Base case Variation Xi SXi Xr SXr td (s) Std

1.68 0.884 0.176 6754

k01 (m3/(m2 s)) 2.10
2.52 0.919
0.096
0.176
0
6318
0.170

80 0.884 0.146 5496

a (m2/g) 100 0.096 0.806 0.722
120 0.919 0.204 7350

4400 0.905 0.176 8030

Scd (m2/g) 5500 0 0 1.041
6600 0.905 0.176 5354

a
Values of output variables in the base case are: Xi ¼ 0.91, Xr ¼ 0.18, and td ¼ 6424 s.

1.0 1
Fig. 9A shows the final ratio of the active to the total
external surface of a catalyst particle, hSe1 =Se iss , and the 0.8
0.1
residual methane conversion, Xr, as function of the carbon T = 800°C
0.6 τg = 5s

final ratio between external active and total particle surface, <S1e/Se>ss, -
attrition constant, ka, obtained with a catalyst particle size
of 3.5  104 m, at temperature of 800 1C and with a mean gas 0.01
residence time of 5 s for two different fluidization velocities, 0.4
0.1 and 0.5 m/s, continuous and dashed line, respectively. 0.001
hSe1 =Se iss presents an increasing monotone sigmoid curve 0.2
0.1 m/s
which highlights a strong dependence of the catalyst 0.5 m/s
regeneration on carbon attrition constant. When the fluidiz- 0.0 0.0001

residual methane conversion, Xr, -

ing gas velocity is 0.1 m/s, a completed regeneration of 1.0 1
external surface is possible for values of ka higher than
1 m1. On the other hand, the residual methane conversion 0.8
0.1
increases with carbon attrition constant reaching a final very T = 800°C
low value in the order of 0.2% corresponding to the complete
0.6 τg = 10s
0.01
regeneration of the external surface of the catalyst particle.
0.4
This evidence highlights that under the investigated condi-
tions the contribution of the external surface of the particles 0.001
0.2
to the reaction rate is negligible also for completely regener- 0.1 m/s
ated particles. When the fluidization velocity is increased up 0.5 m/s
0.0 0.0001
to 0.5 m/s, the curves of hSe1 =Se iss and of Xr as a function of
1.0 1
carbon attrition constant maintain the same shape but they
are simply shifted to lower values of ka. This result confirms
0.8
that attrition rate—proportional to the fluidization excess 0.1
U = 0.5m/s
velocity—is more effective towards the external regeneration
0.6 τg = 10s
of catalyst particle without any modification of the maximum
0.01
residual methane conversion. 0.4
Fig. 9B shows the final ratio of the active to the
total external surface of a catalyst particle, hSe1 =Se iss , and 0.001
0.2
the residual methane conversion, Xr, as a function of 1000°C
the carbon attrition constant, ka, obtained in the same 800°C
0.0 0.0001
operating conditions as those of Fig. 9A but with a larger 0.0001 0.0010 0.0100 0.1000 1.0000
mean gas residence time of 10 s. The model results under- carbon attrition rate, ka, m-1
line that: (i) the mean gas residence time does not substan-
tially influence the regeneration of the external surface Fig. 9 – Model computations: influence of carbon attrition
of the catalyst particle; (ii) an increase of mean gas residence rate on the final active external surface of catalyst particle
time determines a higher residual methane conversion and on the residual methane conversion (dc ¼ 3.5  104 m).
(E0.4%).
Fig. 9C shows the final ratio of the active to the total
external surface of a catalyst particle, hSe1 =Se iss , and the gas residence time of 10 s and at two different temperatures,
residual methane conversion, Xr, as a function of the carbon 800 and 1000 1C, continuous and dashed line, respectively. A
attrition constant, ka, obtained with a catalyst particle size of higher temperature mainly determines an increase in the
3.5  104 m, at a fluidization velocity of 0.5 m/s, with a mean intrinsic kinetic constant of the catalyst k1 by about one order
 ARTICLE IN PRESS
2690 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 33 (2008) 2679 – 2694

of magnitude, from 6  108 to 9  107 m3/(m2s), producing, a more effective catalyst regeneration by attrition (i.e.
as a consequence, the following effects: (i) the complete at the same ka a slightly higher hSe1 =Se iss is observed)
regeneration of external surface and, in turn, the maximum and, moreover, a higher residual methane conversion.
value of Xr are reached at higher values of ka with respect to In particular, it can be observed that the maximum values
800 1C; (ii) the maximum value of Xr increases by about one of Xr obtained in correspondence of the smaller particles is
order of magnitude, up to about 6%, with respect to 800 1C. It higher: about 2% and more than 10% at 800 and 1000 1C,
can be concluded that for the coarser bed particles and the respectively.
operating conditions investigated the role played by carbon The model results suggest that the attrition could play
attrition is moderate. an important role in the performance of the reactor when
Figs. 10A–C show the final ratio of the active to the the contribution of external surface of the bed particles
total external surface of a catalyst particle, hSe1 =Se iss , and to the reaction rate is significant, i.e. in the case of completely
the residual methane conversion, Xr, as a function of the regenerated catalyst particles and of high surface intrinsic
carbon attrition constant, ka, obtained with a lower catalyst kinetics. The comparison of Figs. 9 and 10 underlines that for
particle size of 0.8  104 m, at the same operative conditions the finer bed particles investigated, this contribution is
as those of Figs. 9A–C, respectively. The observed phenom- obviously larger as the total external surface of the bed
enology is very similar to that obtained with bed particles particles per catalyst mass is higher.
with larger size as it regards both the catalyst regeneration
by carbon attrition and the residual methane conversion.
It is worth noting that a decrease in particle size determines 4. Conclusions

The following conclusions can be made:

1.0 1 1. A simple and novel model of a bubbling fluidized bed TCD

of methane has been developed taking into account the
0.8 fate of catalytic particles subjected to different competitive
0.1 phenomena: methane catalytic decomposition, catalyst
T = 800°C
0.6 τg = 5s deactivation due to carbon deposition on the catalyst and
final ratio between external active and total particle surface, <S1e/Se>ss, -

0.01 catalyst reactivation by means of carbon attrition. The

0.4 novelty of the model is related to taking into account the
effect of attrition phenomena, commonly considered as a
0.001
0.2 disadvantage of the catalytic fluidized bed processes, on
0.1 m/s the regeneration of the external surface of the catalyst
0.5 m/s
0.0 0.0001 particles.
residual methane conversion, Xr, -

1.0 1 2. The model enables one to predict the performance of the

fluidized bed reactor in terms of methane conversion,
0.8 hydrogen production, elutriated carbon and deactivation
0.1
T = 800°C time. On the whole, the methane TCD process can be
0.6 τg = 10s characterized by an initial, Xi, and residual, Xr, methane
0.01 conversion and a deactivation time, td.
0.4 3. The model results and experimental data have been
0.001 compared in terms of time-resolved methane conversion
0.2 curves, initial, Xi, and residual, Xr, methane conversion
0.1 m/s
0.5 m/s and deactivation time, td, for different operating condi-
0.0 0.0001 tions. Good agreement within 710% has been obtained in
1.0 1 most of the investigated cases.
4. The sensitivity analysis of the model to changes of the
0.8
0.1 properties of catalyst and of deposited carbon on the
U = 0.5m/s performance of the fluidized bed reactor has demon-
0.6 τg = 10s
strated that the deactivation process is strongly influenced
0.01
by the specific area of the catalyst, a, and by the surface of
0.4
catalyst particle occupied by deposited carbon per carbon
0.001 mass, Scd.
0.2
1000°C 5. The model has shown that carbon attrition plays a key role
800°C in the regeneration of the external catalyst surface in all
0.0 0.0001
0.0001 0.0010 0.0100 0.1000 1.0000 the investigated conditions. This phenomenon can also
carbon attrition rate, ka, m-1 affect the residual methane conversion when the con-
tribution of the external surface of the particle to the
Fig. 10 – Model computations: influence of carbon attrition reaction rate is high enough. This condition is emphasized
rate on the final active external surface of catalyst particle by a high intrinsic kinetic rate and by a small size of
and on the residual methane conversion (dc ¼ 0.8  104 m). catalyst particles.
 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 33 (2008) 2679 – 2694 2691

Appendix A. Hydrodynamics and methane decomposition

Following a procedure analogous to that developed by Harrison and Davidson [15], the actual methane concentration in the
dense phase can be expressed as

1  b eXbe þ 0:5XðtÞð1  eXbe Þ

ce ðtÞ ¼ co (A.1)
1  beXbe þ 0:5XðtÞð1  eXbe Þ þ DaðtÞ
where b, Xbe, Da(t), ce(t), co and e are the fluidization velocity excess with respect to the fluidizing gas velocity, the index of gas
exchange between bubbles and dense phase, the Damköhler number, the concentration of methane in the dense phase and in
the inlet stream, and the fractional change in volume of the system between no conversion and complete conversion of
methane, respectively. The actual methane concentration in the bubble phase cb(t,x) can be expressed as

cb ðt; xÞ ¼ ce ðtÞ þ ðco  ce ðtÞÞ eXbe x=H (A.2)

where x is the distance above the distributor. As a consequence, the exit conversion of methane, X, taking into account the exit
bubble phase methane concentration and the dense phase methane concentration, can be easily obtained [15]

½ð1 þ Þð1  b eXbe Þ þ DaðtÞð1 þ  eXbe Þ

XðtÞ ¼
ð1 þ Þð1  eXbe Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
½ð1 þ Þð1  beXbe Þ þ DaðtÞð1 þ  eXbe Þ2 þ 2 DaðtÞð1 þ Þð1  eXbe Þð1  b eXbe Þ
þ . (A.3)
ð1 þ Þð1  eXbe Þ
The fluidization velocity excess with respect to the fluidizing gas velocity, b, is calculated by [15]

U  Umf
b¼ (A.4)
U
where U and Umf are the fluidizing gas velocity and the minimum fluidization velocity.
The index of gas exchange between bubbles and dense phase, Xbe, may be expressed as a dimensionless cross-flow ratio,
defined as [15]

Kbe  H
Xbe ¼ (A.5)
Ub
where Kbe is the overall coefficient of gas interchange between bubble phase and dense phase, H is the bubbling fluidized bed
height and Ub is the rise velocity of bubbles through the bed.
H and Ub can be calculated by the following equation system [16]:
8
> U þ ðUð1 þ 0:5XðtÞÞ  Umf Þ
>
>
> H ¼ Hmf br
>
> Ubr
>
>
> x ¼ ð0:4C0:5Þ  H
>
>
>  
<  0:3  x
db ¼ dbm  ðdbm  db0 Þ  exp (A.6)
>
> Dr
>
>  
>
>
>  0:5 d
> Ubr ¼ 0:711  ðg  db Þ  1:2  exp 1:45 b
>
>
> Dr
>
:
Ub ¼ Uð1 þ 0:5XðtÞÞ  Umf þ Ubr

where Hmf, Ubr, x*, db ; dbm, db0 and Dr are the static bed height, the rise velocity of isolated bubbles, a pre-set distance above the
distributor, the effective bubble diameter at x*, the maximum bubble diameter, the bubble diameter just above the distributor
and the bed diameter, respectively. db is the mean value of the bubble size along the bed axis and it is assumed as the bubble
diameter in all the fluidized bed.
Kbe can be calculated from the following relation [17]:

1 1 1
¼ þ (A.7)
Kbe Kbc Kce
where Kbc and Kce are the coefficient of gas interchange between bubble and cloud-wake region and the coefficient of gas
interchange between cloud-wake region and dense phase, respectively.
Kbc can be evaluated from the expression derived by Davidson and Harrison [15]

! 0 1
1=2
U DCH  g1=4
Kbc ¼ 4:5 mf þ 5:85@ 4
n5=4
A (A.8)
dnb db

where DCH4 is the methane diffusion.

 ARTICLE IN PRESS
2692 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 33 (2008) 2679 – 2694

Kce can be calculated from the following equation [17]:

!1=2
DCH4  mf  Ub
Kce ¼ 6:77 (A.9)
dbn3
where emf is the void fraction in the bed at minimum fluidization condition.
The Damköhler number is calculated by [26]

kp ðtÞrb Hmf
DaðtÞ ¼ (A.10)
rc U
where kp(t) is the apparent reaction constant per volume of catalyst particle, rc is the apparent catalyst density and rb is the bulk
density of catalyst bed.
The apparent reaction constant kp(t) varies with time as the active surface of the catalyst varies with time as a consequence of
the deposition of carbon produced by methane decomposition. Accounting for the simultaneous action of the different
resistances, i.e. gas diffusion in the external boundary layer, intra-particle diffusion, intrinsic kinetics on the surface of catalyst
and intrinsic kinetics on the surface of deposited carbon, kp(t) can be expressed as

1
kp ðtÞ ¼ (A.11)
ðVc =kg Se Þ þ ðVc =ðk1 ðSi1 ðtÞZðSi1 ðtÞ; Si2 ðtÞÞ þ Se1 ðtÞÞ þ k2 ðSi2 ðtÞZðSi1 ðtÞ; Si2 ðtÞÞ þ Se2 ðtÞÞÞÞ
where Vc, kg, Se, k1, k2, Si1 ðtÞ; Se1 ðtÞ; Si2 ðtÞ; Se2 ðtÞ and ZðSi1 ðtÞ; Si2 ðtÞÞ are the volume of a single particle of catalyst, the mass transfer
coefficient outside the particle, the total external surface of the catalyst particle, the surface intrinsic kinetic constant of the
catalyst, the surface intrinsic kinetic constant of the deposited carbon, the internal and external active surfaces of a single
catalyst particle, the internal and external catalyst surfaces occupied by the deposited carbon and the efficiency of a catalyst
particle, respectively. The surface intrinsic kinetic constants, k1 and k2, can be expressed by Arrhenius’s law as first order of
approximation as

k1 ¼ k01 eE1 =RT

k2 ¼ k02 eE2 =RT (A.12)
where k01 , k02 , E1 and E2 are the catalyst and carbon frequency factors and the activation energies of the catalyst and of the carbon
deposits, respectively. The efficiency of the catalytic particle is a function of time accounting for the effective surface active
inside the particle and the reduction of pore dimension due to the carbon deposition. The relationship between the efficiency
and the Thiele number [14], YðSi1 ðtÞ; Si2 ðtÞÞ; in this case, is

3YðSi1 ðtÞ; Si2 ðtÞÞ cothð3YðSi1 ðtÞ; Si2 ðtÞÞÞ  1

ZðSi1 ðtÞ; Si2 ðtÞÞ ¼ (A.13)
3YðSi1 ðtÞ; Si2 ðtÞÞ2

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dc k1 Si1 ðtÞ þ k2 Si2 ðtÞ
YðSi1 ðtÞ; Si2 ðtÞÞ ¼ (A.14)
6 Vc DðSi2 ðtÞÞ

8
DCH C
>
> 4
1 < Deff ¼ t
 qffiffiffiffiffiffiffiffiffiffiffi
DðSi2 ðtÞÞ ¼ 1 1 C wi ðtÞS (A.15)
þ > i
> D ðS ðtÞÞ ¼ 485 dp  2 c i cd hcd T
Deff Dk ðSi ðtÞÞ : k 2 t S M CH4
2

where dc, DðSi2 ðtÞÞ; Deff, Dk ðSi2 ðtÞÞ; ec, t, dp, wic ðtÞ; Scd, Si, hcd, T and MCH4 are the catalyst particle diameter, the mean, bulk and
Knudsen diffusivity inside the particle, the porosity, the tortuosity factor and the mean pore diameter of a catalyst particle, the
carbon deposited on the internal surface of a single bed particle, the surface of catalyst particle occupied by deposited carbon per
carbon mass, the internal surface of a catalyst particle, the thickness of the deposited carbon, temperature and molecular weight
of methane, respectively. It is noteworthy that the Thiele number varies during reaction time as a result of changes in pore
diameter and in catalyst particle activity due to carbon deposition. The actual average pore size is calculated by considering that
the carbon deposits uniformly along the pores reducing, in turn, the diameter. The temporal evolution of the deactivation
process has been taken into account considering the variation of the active internal and external surfaces of the catalyst particle
with time.
The mass transfer coefficient outside the particle kg can be evaluated from the expression derived by Hayhurst and Parmar for
fluidized beds [30]

 
dc kg Remf 1=2
Sh ¼ ¼ 2mf þ 0:68  Sc1=3 (A.16)
DCH4 mf

where Sh, Remf and Sc are the Sherwood number, the Reynolds number at minimum fluidizing conditions and the Schimdt
number [31], respectively.
 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 33 (2008) 2679 – 2694 2693

Appendix B. Carbon deposition and removal

The deactivation process occurring on the internal and external surfaces of the catalytic particles can be modelled taking into
account that:

(I) The transient mass balance on the carbon deposited on the external surface of a single bed particle, wec ðtÞ; is determined
by both the effective external reaction rate of carbon accumulation proportional to the methane decomposition rate and
the removal rate of carbon by the mechanical abrasion of bed particles:

dwec ðtÞ
¼ kep ðtÞce ðtÞVc MC  ka wec ðtÞðUð1 þ 0:5XðtÞÞ  Umf Þ (B.1)
dt
where kep ðtÞ
is the external surface reaction constant, MC is carbon molecular weight, ka is the carbon attrition constant.
(II) The transient mass balance on the carbon deposited on the internal surface of a single bed particle, wic ðtÞ; is determined
by the effective internal reaction rate of carbon accumulation proportional to the methane decomposition rate:

dwic ðtÞ
¼ kip ðtÞce ðtÞVc MC (B.2)
dt
where kip ðtÞ
is the internal surface reaction constant.
(III) The transient balance on the external active surface of a single catalyst particle, Se1 ðtÞ; is similar to the mass balance on
the deposited carbon, but it is referred to the external active catalyst surface:

dSe1 ðtÞ
¼ Scd ðkep ðtÞce ðtÞVc MC  ka wec ðtÞðUð1 þ 0:5XðtÞÞ  Umf ÞÞ. (B.3)
dt

(IV) The transient balance on the internal active surface of a single catalyst particle, Si1 ðtÞ; only accounts for the
accumulation of carbon deposits on the internal active surface of catalyst:

dSi1 ðtÞ
¼ Scd kip ðtÞce ðtÞVc MC (B.4)
dt

The external and internal surfaces’ reaction constants, kep ðtÞ and kip ðtÞ; can be expressed as

1
kep ðtÞ ¼ (B.5)
ðVc =kg Se Þ þ ðVc =ðk1 Se1 ðtÞ þ k2 Se2 ðtÞÞÞ þ ððk1 Si1 ðtÞ þ k2 Si2 ðtÞÞZðSi1 ðtÞ; Si2 ðtÞÞVc Þ=ððk1 Se1 ðtÞ þ k2 Se2 ðtÞÞkg Se ÞÞ

1
kip ðtÞ ¼ (B.6)
ðVc =kg Se Þ þ ðVc =ððk1 Si1 ðtÞ þ k2 Si2 ðtÞÞZðSi1 ðtÞ; Si2 ðtÞÞÞÞ þ ðððk1 Se1 ðtÞ þ k2 Se2 ðtÞÞVc Þ=ððk1 Si1 ðtÞ þ k2 Si2 ðtÞÞZðSi1 ðtÞ; Si2 ðtÞÞkg Se ÞÞ
and the active and the deactivated surfaces are related to each other by
(
Se1 ðtÞ þ Se2 ðtÞ ¼ Se
(B.7)
Si1 ðtÞ þ Si2 ðtÞ ¼ Si

The carbon elutriation rate, Ec(t), can be calculated by

Ec ðtÞ ¼ ka ðUð1 þ 0:5XðtÞÞ  Umf ÞWc ðtÞ ¼ ka ðUð1 þ 0:5XðtÞÞ  Umf Þwec ðtÞNc (B.8)
where Wc(t) and Nc are the carbon deposited on the external surface of bed particles and the number of catalyst particles,
respectively. Nc can be evaluated as

pD2r =4Hmf rb
Nc ¼ (B.9)
Vc rc
The production rates of carbon, PC, and of hydrogen, PH2 ; can be calculated by

pD2r
PC ðtÞ ¼ MC Uc0 XðtÞ (B.10)
4

pD2r Mathcad environment using a fourth order Runge–Kutta

PH2 ðtÞ ¼ 2MH2 Uc0 XðtÞ (B.11) algorithm with adaptive integration step size. The computed
4
results of the differential equations system are used to
Eqs. (A.1), (A.2), (A.4)–(A.16) and (B.5)–(B.7) have been rear- evaluate the exit methane conversion, the carbon elutriation
ranged and, then, substituted in Eqs. (B.1)–(B.4) in order to rate and the carbon and hydrogen production rates by means
obtain four independent first order differential equations. of Eqs. (A.3) and (B.8)–(B.11), respectively. All the variables
This differential equations system has been solved in
 ARTICLE IN PRESS
2694 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 33 (2008) 2679 – 2694

involved were evaluated as functions of time during the TCD [13] Toomey RD, Johnstone MF. Gaseous fluidization of solid
of methane. particles. Chem Eng Prog 1952;48:220–6.
[14] Levenspiel O. Chemical reaction engineering. 3rd ed. New York:
Wiley; 1999.
R E F E R E N C E S
[15] Davidson JK, Harrison D. Fluidised particles. Cambridge:
University Press; 1963.
[16] Clift R, Grace JR. Continuous bubbling and slugging. In:
[1] Muradov N, Smith F, Huang C, T-Raissi A. Autothermal Davidson JF, Clift R, Harrison D, editors. Fluidization. 2nd ed.
catalytic pyrolysis of methane as a new route to hydrogen London: Academic Press; 1985.
production with reduced CO2 emissions. Catal Today [17] Kunii D, Levenspiel O. Bubbling bed model—model for the
2006;116:281–8. flow of gas through a fluidized bed. Ind Eng Chem
[2] Muradov NZ, Veziroğlu TN. From hydrocarbon to hydrogen– 1968;7:446–52.
carbon to hydrogen economy. Int J Hydrogen Energy 2005; [18] Ammendola P. PhD thesis, Università degli Studi di Napoli
30:225–37. Federico II, Napoli, Italy, 2006.
[3] Muradov N, Chen Z, Smith F. Fossil hydrogen with reduced [19] Zein SHS, Mohamed AR, Sai PST. Kinetic studies on catalytic
CO2 emission: modeling thermocatalytic decomposition of decomposition of methane to hydrogen and carbon over
methane in a fluidized bed of carbon particles. Int J Hydrogen Ni/TiO2 Catalyst. Ind Eng Chem Res 2004;43:4864–70.
Energy 2005;30:1149–58. [20] Muradov NZ. CO2-free production of hydrogen by catalytic
[4] Lee KK, Han GY, Yoon KJ, Lee BK. Thermocatalytic hydrogen pyrolysis of hydrocarbon fuel. Energy Fuel 1998;12:41–8.
production from methane in a fluidized bed with activated [21] Kim MH, Lee EK, Jun JH, Kong SJ, Han GY, Lee BK, et al.
carbon catalyst. Catal Today 2004;93–95:81–6. Hydrogen production by catalytic decomposition of methane
[5] Dunker AM, Kumar S, Mulawa PA. Production of hydrogen over activated carbons: kinetic study. Int J Hydrogen Energy
by thermal decomposition of methane in a fluidized-bed 2004;29:187–93.
reactor—effects of catalyst, temperature, and residence time. [22] Lee EK, Lee SY, Han GY, Lee BK, Lee T-J, Jun JH, et al. Catalytic
Int J Hydrogen Energy 2006;31:473–84. decomposition of methane over carbon blacks for CO2-free
[6] Ogihara H, Takenaka S, Yamanaka I, Tanabe E, Genseki A, hydrogen production. Carbon 2004;42:2641–8.
Otsuka K. Formation of highly concentrated hydrogen [23] Miccio M, Massimilla L. Fragmentation and attrition of carbo-
through methane decomposition over Pd-based alloy cata- naceous particles generated from the fluidized bed combustion
lysts. J Catal 2006;238:353–60. of fuel–water slurries. Powder Technol 1991;65:335–42.
[7] Aiello R, Fiscus JE, zur Loye H-C, Amiridis MD. Hydrogen [24] Tao XY, Zhang XB, Zhang L, Cheng JP, Liu F, Luo JH, et al.
production via the direct cracking of methane over Ni/SiO2: Synthesis of multi-branched porous carbon nanofibers and
catalyst deactivation and regeneration. Appl Catal A—Gen their application in electrochemical double-layer capacitors.
2000;192:227–34. Carbon 2006;44:1425–8.
[8] Takenaka S, Kobayashi S, Ogihara H, Otsuka K. Ni/SiO2 [25] Dorfman S, Mundim KC, Fuks D, Berner A, Ellis DE, Van
catalyst effective for methane decomposition into hydrogen Humbeeck J. Atomistic study of interaction zone at copper–
and carbon nanofiber. J Catal 2003;217:79–87. carbon interfaces. Mater Sci Eng C 2001;15:191–3.
[9] Ermakova MA, Ermakov DY, Chuvilin AL, Kuvshinov GG. [26] Rosner D E. Transport processes in chemically reacting flow
Decomposition of methane over iron catalysts at the range of systems. Mineola, New York: Dover Publications; 2000.
moderate temperatures: the influence of the catalytic [27] Koerts T, Deelen M, van Santen R. Hydrocarbon formation
systems and the reaction conditions on the yield of carbon from methane by a low-temperature two-step reaction
and morphology of carbon filaments. J Catal 2001;201: sequence. J Catal 1992;138:101–14.
183–97. [28] Suelves I, Lázaro MJ, Moliner R, Echegoyen Y, Palacios JM.
[10] Takenaka S, Serizawa M, Otsuka K. Formation of filamentous Characterization of NiAl and NiCuAl catalysts prepared by
carbons over supported Fe catalysts through methane different methods for hydrogen production by thermo cataly-
decomposition. J Catal 2004;222:520–31. tic decomposition of methane. Catal Today 2006;116:271–80.
[11] Ammendola P, Chirone R, Lisi L, Ruoppolo G, Russo G. Copper [29] Rudd DF, Watson CC. Strategy of process engineering. New
catalysts for H2 production via CH4 decomposition. J Mol York: Wiley; 1968.
Catal A—Chem 2007;266:31–9. [30] Hayhurst AN, Parmar MS. Measurement of the mass transfer
[12] Ammendola P, Ruoppolo G, Chirone R, Russo G. H2 produc- coefficient and Sherwood number for carbon spheres burn-
tion by catalytic methane decomposition in fixed and ing in a bubbling fluidized bed. Combust Flame
fluidized bed reactors. In: Winter F, editor. Proceedings of 2002;130:361–75.
19th international conference on fluidized bed combustion. [31] Bird RB, Stewart WE, Lightfoot EN. Transport phenomena.
Vienna, Austria, 2006. New York: Wiley; 1960.