Fluidic effects on kinetic parameter estimation in lab-scale catalysis

testing A critical evaluation based on computational uid dynamics

Gregor D. Wehinger
n
, Thomas Eppinger, Matthias Kraume
Chemical and Process Engineering, Technische Universitt Berlin, Fraunhoferstr. 33-36, 10587 Berlin, Germany
H I G H L I G H T S

Two different lab-scale catalytic reactors were simulated.

Fluidic effects on kinetic parameter estimation were specied.

Stagnation ow reactors are suitable to suppress uidic effects.

Neglecting radial proles leads to erroneous kinetic parameters for xed-beds.

The prediction of temperature proles is of major importance.

a r t i c l e i n f o
Article history:
Received 18 December 2013
Received in revised form
10 February 2014
Accepted 17 February 2014
Available online 6 March 2014
Keywords:
Catalysis
Chemical reactors
Computational uid dynamics
Kinetics
Oxidative coupling of methane
Dry reforming of methane
a b s t r a c t
The inuence of uidic effects on two different kinetic parameter identications in lab-scale catalysis
testing was investigated using computational uid dynamics. Firstly, the dry reforming of methane in a
stagnation ow reactor with a detailed surface mechanismwas simulated fully in three-dimensional. It is
shown that the 3D simulations are not advantageous over the commonly used stagnation-ow
boundary-layer problem description. This reactor setting is a valuable example of how uidic effects
on kinetic parameter estimation can be suppressed. Secondly, the oxidative coupling of methane
in a xed-bed reactor with a 10-step kinetic mechanism was simulated with a porous-media model.
The experimental results could not be reproduced. The underlying plug-ow model for kinetic para-
meter identication fails in this highly exothermic reactor, because of signicant radial temperature
proles and resulting radial concentration proles. The correct prediction of temperature proles is of
major signicance. This investigation highlights the importance of well dened reactor congurations in
combination with spatially resolved temperature and concentration proles for the determination of
reliable kinetic parameters for highly exothermic or endothermic reactions.
& 2014 Elsevier Ltd. All rights reserved.
1. Introduction
Over the last decades the utilization of computers in the eld of
chemical reactor description and design has increased dramati-
cally. This is not only due to faster and cheaper technical devices
but also due to the usability of CFD-codes either commercial or
non-commercial. One of the preconditions in reactor modeling is
the knowledge of the reaction kinetics. Especially when catalysts
are involved in chemical reactions, i.e., in more than 8090%
of industrial processes today (Marcilly, 2003), kinetic description
can be critical. Many of the kinetics used for reactor design are
LangmuirHinshelwoodHougenWatson (LHHW) types. However,
parameters of these models can be questionable and the fundamental
mechanisms elusive, although the specic LHHW model is capable
to reproduce experiments accurately, cf. Salciccioli et al. (2011).
Consequently, a reactor scale-up with such a model would lead to
erroneous predictions.
In all chemical reactors, there is an interplay between chemical
kinetics and transport of momentum, heat and mass. Different
time and length scales appear at different facets of a reactor
process. The so-called multiscale methodology gives a quantitative
correlation between various measures of performance and oper-
ating variables (Dudukovic, 2009). At different levels of knowledge
descriptions can be carried out. At the reactor scale, the depth
ranges from ideal reactors to computational uid dynamics (CFD)
(Dudukovic, 2010).
Not only in extremely exothermic and endothermic reactions,
the chemistry and transport phenomena are intrinsically con-
nected. However, in those reactions it is more complicated to
distinguish between the different contributions. Dautzenberg and
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/ces
Chemical Engineering Science
http://dx.doi.org/10.1016/j.ces.2014.02.025
0009-2509 & 2014 Elsevier Ltd. All rights reserved.
n
Corresponding author. Tel.: 49 30 314 28733; fax: 49 30 314 21134.
E-mail address: gregor.wehinger@tu-berlin.de (G.D. Wehinger).
Chemical Engineering Science 111 (2014) 220230
colleagues have highlighted how to minimize the effects of
transport phenomena while estimating kinetic parameters in their
Ten Commandments for Testing Catalysts (Dautzenberg, 1989).
Nevertheless, the impact of uidic effects in the reactor toward the
estimated kinetics is often not further specied. Simplied che-
mical models fail when chemical coupling with gas phase reac-
tions is important or where a large proportion of the heat-release
occurs in the boundary-layer (Pfefferle, 1995). Plug ow reactor
models, for example, are unsatisfactory for catalytic combustion
applications, because they simplify drastically the heat and mass
transport (Dalle Nogare et al., 2008). It is therefore recommended
to model such reactors with full CFD or at least lumped uid
dynamic models, e.g., pseudo-heterogeneous models (Korup et al.,
2013). Several authors have highlighted the importance to deter-
mine mass transport limitations while measuring catalytic reac-
tions (Horn et al., 2007; Kapteijn and Moulijn, 2008; Schuurman,
2008).
One of the critical points in the identication of kinetic
parameters is the temperature prole. It is inuenced by and
likewise inuences the reaction path. However, the modeling is
complex, since temperature is inuenced by several physical
transport processes, e.g., heat released by reaction can be trans-
ferred by conduction and convection through the uid, conduction
through the particles and radiation (Wolf et al., 1997; Dixon, 2012).
It was shown that the kinetics will be more inuenced by the
temperature prole than vice versa (Dalle Nogare et al., 2011). In
other words a model that is capable to predict the temperature
proles is able to predict the species proles correctly.
Being aware of these considerations, several scientists investi-
gated spatially resolved experimental data in combination with
spatially resolved numerical data, in recent years. Other authors
have emphasized the testing under realistic conditions besides
spatially resolution (Horn et al., 2006; Urakawa and Baiker, 2009;
Korup et al., 2013). Some of these congurations are: 3D simulated
catalytic gauze, e.g., Rinnemo et al. (1997) and Quiceno et al.
(2006), 1D simulated catalytic foam, e.g., Horn et al. (2007), Dalle
Nogare et al. (2011) and Korup et al. (2013), 3D simulated channel
with catalytic walls, e.g., Sa et al. (2010) and Hettel et al. (2013),
1D or 2D simulations of stagnation ow reactors, e.g., McGuire
et al. (2009, 2011), Karakaya and Deutschmann (2013) and Yuan
et al. (2008), and 1D simulated annular reactors, e.g., Maestri et al.
(2008).
In this paper, we investigated two current catalytic reactions,
i.e., the dry reforming of methane (DRM), an overview given by
Bradford and Vannice (1999), and the oxidative coupling of
methane (OCM), a review is presented by Arndt et al. (2011). The
focus lies on two kinetic descriptions: a detailed surface mechan-
ism for DRM, cf. McGuire et al. (2011), and a homogeneous-
heterogeneous OCM-kinetics by Stansch et al. (1997). The reactor
congurations of each experiment were reproduced by 3D che-
mically reacting ow simulations. The results of the experiments
and simulations are presented, compared and discussed with
special attention to the temperature contribution. The inuences
of the chosen reactor type toward uid dynamics and therefore
kinetic parameter estimation were investigated.
2. Simulating chemically reacting ow
In this study, full three-dimensional governing equations form
the basis for the calculations. They represent the most precise
description of reacting ow in an arbitrary geometry (Mladenov
et al., 2010). The conservation of total mass, momentum in x; y; z
directions, mass of species and energy provides the solution for
velocity, pressure, temperature and species concentration in the
calculation domain. The equations for a laminar problem are with
Einstein convention.
Conservation of mass:

t

v
i

x
i
0 1
where is the mass density, t is the time, x
i
are the Cartesian
coordinates and v
i
are the velocity components.
Conservation of momentum:
v
i

t

v
i
v
j

x
j

p
x
i

ij
x
j
g
i
2
The stress tensor
ij
is given by

ij

v
i
x
j

v
j
x
i

2
3

ij
v
k
x
k
3
where is the mixture viscosity and
ij
is the Kronecker delta,
which is unity for i j, else zero.
Conservation of species i:
Y
i

t

v
j
Y
i

x
j

j
i;j

x
j
R
hom
i
for i 1; ; N
g
4
with Y
i
the mass fraction of species i in the mixture Y
i
m
i
=m with
m as total mass. R
i
hom
is the net production rate by homogeneous
reactions, whereas N
g
represents the number of gas phase species.
The components j
i;j
of the diffusion mass ux are modeled by the
mixture-average formulation:
j
i;j

Y
i
X
i
D
M
i
X
i
x
j

D
T
i
T
T
x
j
5
with the effective diffusivity D
i
M
between species i and the remain-
ing mixture. X
i
is the molar fraction of species i. M
i
represents the
molecular weight of species i and T the temperature. The binary
diffusion coefcients D
i
are obtained through polynomial ts.
The molar fraction X
i
is dened as
X
i

1

Ng
j 1
Y
j
M
j
Y
i
M
i
6
Conservation of energy in terms of specic enthalpy h:
h
t

v
j
h
x
j

j
q;j
x
j

p
t
v
j
p
x
j

jk
v
j
x
k
S
h
7
where S
h
is the heat source. Diffusive heat transport j
q;j
is dened as
j
q;j

T
x
j

Ng
i 1
h
i
j
i; j
8
with thermal conductivity of the mixture and mixture specic
enthalpy h:
h
Ng
i 1
Y
i
h
i
T 9
with the specic enthalpy as a function of temperature h
i
h
i
T.
Ideal gas was assumed connecting pressure, temperature and
density to close the governing equations:
p
RT

Ng
i 1
X
i
M
i
10
In addition, NASA polynomial functions were used to derive heat
capacity c
p;i
. For more information see Deutschmann (2008) and
Kee et al. (2003).
All simulations were carried out with the simulation software
STAR-CCM version 8.04.010 of CD-adapco (2013).
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 221
2.1. Spatially resolved simulations
In the case of DRM only reactions on the surface are taken into
account. The reaction mechanism is presented in a detailed way,
i.e., adsorption, surface reaction and desorption are described.
In the following, the essential equations are given.
2.1.1. Computational uid dynamics and heterogeneous chemical
reactions
The chemical process on the catalytic surface is coupled via
boundary conditions with the species distribution equation (4). Under
steady-state conditions gas-phase molecules of species i, which are
produced/consumed at the catalytic surface by desorption/adsorption,
have to diffuse from/to the catalyst (Deutschmann, 2008):
n
!
j
i
!
R
het
i
11
The heterogeneous reaction term R
i
het
can be written as
R
het
i
F
cat=geo
M
i
_ s
i
12
where is the effectiveness factor, M
i
is the molar weight, _ s
i
is the
molar net production rate of gas-phase species i and F
cat=geo
is the
ratio of catalytic active area A
catalytic
to geometric area A
geometric
:
F
cat=geo
A
catalytic
=A
geometric
13
The mean-eld approximation is applied for modeling the sur-
face reactions. It assumes that adsorbates are randomly distributed
on the surface, which is viewed as being uniform, cf. Deutschmann
(2008) and Kee et al. (2003). The molar net production _ s
i
is
expressed as
_ s
i

Ks
k 1

ik
k
f
k

Ng Ns N
b
j 1
c
'
jk
j
14
where K
s
is the number of surface reactions and c
j
are the species
concentrations, in mol/m
2
for the adsorbed species N
s
and in mol/
m
3
for the gas phase species N
g
and bulk species N
b
. Additionally,
the surface coverage takes into account the surface site density
(mol/m
2
), which represents the maximum number of species that
can adsorb on a unit surface area, and a coordinate number s
i
expressing the number of surface sites which are covered by this
species, as well as the concentration c
i
:

i
c
i
s
i

1
15
the time dependent variation of
i
can be written as

i
t

_ s
i
s
i

16
Under steady state conditions the left hand side of Eq. (16) becomes
zero. One can modify the reaction rate expression by the concen-
tration, or coverage, of some surface species:
k
f
k
A
k
T

k
exp
E
a
k
RT


Ns
i 1

i
k
i
exp

i
k

i
RT

17
with two additional coverage parameters,
ik
and
ik
. The term
including
ik
indicates the modication of the surface rate expres-
sion proportional to any arbitrary power of a surface species
concentration.
ik
represents a modication of the activation energy
as a function of coverage.
The occurrence of adsorption reactions leads to a modication of
the conventional rate coefcient by referencing sticking coefcients S
i
:
k
ads
f
k

S
0
i

RT
2M
i
s
18
with S
i
0
as the initial (uncovered surface) sticking coefcient and

Ns
j 1
'
jk
the sumof all the surface reactant's stoichiometric coef-
cients, cf. Kee et al. (2003) and Deutschmann (2008).
In addition, the operator splitting algorithm was applied. This
algorithm enables the decoupling of the general species transport
equation, due to the different time scales of the chemical reactions
and the ow eld. The time integration of the chemical state
(species mass fractions and enthalpy) was performed in two steps.
First, only the chemical source terms were taken into account for
the integration of a time interval. Second, the ow eld was
integrated over a time interval without the chemical source terms,
cf. Ren and Pope (2008).
2.1.2. Detailed reaction mechanism: DRM
Dry reforming of methane (DRM) is a catalytic process in which
CH
4
and CO
2
react to synthesis gas, i.e., CO H
2
. It is called dry
since steam is not used in this reforming process. In a global
formulation, one can write
CH
4
CO
2
2H
2
2CO; H 260 kJ=mol 19
This highly endothermic catalytic process is performed at tem-
peratures of 7001000 1C.
In this study a microkinetics scheme, see Table 1, was used
from McGuire et al. (2011) for the dry reforming of methane by Rh
supported strontium-substituted hexaaluminate. The mean-eld
approximation underlies this kinetics. It was adopted from a
surface mechanism of the catalytic partial oxidation of iso-
octane on rhodium published by Hartmann et al. (2010). The
surface reaction mechanism contains 42 irreversible elementary
reactions including 12 surface-adsorbed species and 6 gas-phase
species, as well as reactions involving HCO
n
, cf. McGuire et al.
(2011). In Table 1 the asterisk indicates a surface site or surface
species.
2.2. Pseudo-homogeneous simulations
The OCM kinetics from Stansch et al. (1997) is based on a
pseudo-homogeneous plug-ow model. In this model no distinc-
tion is made between the solid phase, i.e., the catalytic spheres or
pellets, and the uid phase, consisting of the voids between the
particles. The reaction rates are related to the entire reactor
volume. Therefore, the porous-media model was used in the CFD
simulations. In the following the essential equations of this model
are given.
2.3. Computational uid dynamics and porous-media model
Due to the large tube-to-particle-diameter, i.e., N20, the
porous-media model was used to describe the ow eld inside
the xed-bed reactor. As mostly, in a porous region the shapes of
the catalytic particles are not considered explicitly. Instead, an
empirically determined ow resistance is added as a source term
S
i;porous
to the right-hand side of Eq. (2), which accounts for the
pressure drop. We used the Ergun equation to model the viscous
resistance by terms of permeability k
p
and the inertial resistance :
S
i;porous


k
p
v
i
j v
!
jv
i

20
1
k
p

1501
2

3
d
2
p
21

1:751

3
d
p
22
where is the mean porosity of the bed, d
p
is the particle diameter
and is the solid density.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 222
In the porous-media model the chemical reaction description is
related to the entire reactor volume in Eq. (4):
R
hom
i
M
i

KV
k 1

ik
R
k
kg=m
3
s 23
where K
V
is the number of reactions and R
k
is the k-th
reaction rate.
Additionally, the enthalpy formulation equations (7) and (9) are
modied. The total enthalpy is now a sum of the enthalpies of
the uid and the solid, whereas their contribution is expressed by
the porosity . In general, mass and heat transfer resistance as well
as the detailed xed bed geometry are neglected. These simplica-
tions are compensated by introducing effective thermal conduc-
tivity
eff
and diffusion coefcients D
eff
(Froment and Bischoff,
1990).
2.3.1. Lumped reaction mechanism: OCM
The oxidative coupling of methane (OCM) represents an exother-
mic heterogeneous-homogeneous reaction system. Although many
publications neglect homogeneous gas phase reactions, they play an
important role (Arndt et al., 2012). The overall pathway of the OCM
can be written as
2CH
4
O
2
C
2
H
4
2H
2
O
H
r
800 1C 139 kJ=mol 24
However, due to thermodynamic reasons, the partial oxidation
equation (25) or total oxidation equation (26) is favored in the OCM:
CH
4

3
2
O
2
CO2H
2
O
H
r
800 1C 519 kJ=mol 25
CH
4
2O
2
CO
2
2H
2
O
H
r
800 1C 801 kJ=mol 26
Catalysts are used to achieve a reasonable C
2
yield (Lunsford, 2000).
The OCM process is carried out at temperatures between 600 and
900 1C, the so-called OCM-window (Sinev et al., 2009). Above
these temperatures the CO
X
formation and severe catalyst deactiva-
tion increase, which leads to a lower C
2
yield. These facts and the
cost intensive gas separation are the reasons why the OCM is still
not used in industry.
The 10-step kinetic-scheme from Stansch et al. (1997) is the
most comprehensive intrinsic kinetics up to date (Kondratenko
and Baerns, 2008). The kinetics was determined for a La
2
O
3
/CaO
catalyst. It includes nine surface reactions and one gas phase
reaction. Eight species participate in the reaction network, see
below: C
2
H
4
, C
2
H
6
, CH
4
, CO
2
, CO, O
2
, H
2
O, H
2
. Methane is
consumed either non-selectively to carbon dioxide via step
1 and carbon monoxide via step 3 or selectively to ethane via
step 2. Ethylene is produced subsequently. Carbon monoxide
represents an intermediate product in this reaction scheme.
Step 1: CH
4
2O
2
-CO
2
2H
2
O
Step 2: 2CH
4
0:5O
2
-C
2
H
6
H
2
O
Step 3: CH
4
O
2
-COH
2
OH
2
Step 4: CO0:5O
2
-CO
2
Step 5: C
2
H
6
0:5O
2
-C
2
H
4
H
2
O
Step 6: C
2
H
4
2O
2
-2CO2H
2
O
Step 7: C
2
H
6
-C
2
H
4
H
2
Step 8: C
2
H
4
2H
2
O-2CO4H
2
Step 9: COH
2
O-CO
2
H
2
Step 10: CO
2
H
2
-COH
2
O
The reaction rate for the i-th step is described by Langmuir-
Hinshelwood or Arrhenius approaches, Eqs. (27)(32), with the
kinetic parameters given in Table 2.
r
j

k
0;j
e
E
a;j
=RT
p
n
j
O
2
p
m
j
C
1K
j;CO
2
e
H
ad;j
CO
2
=RT
p
CO
2

2
; j 1; 36 27
r
2

k
0;2
e
E
a;2
=RT
K
O
2
e
H
ad;2
O
2
=RT
p
O
2

n
2
p
m
2
CH
4
1K
O
2
e
H
ad;2
O
2
=RT
p
O
2

n
2
K
2;CO
2
e
H
ad;2
CO
2
=RT
p
CO
2

2
28
r
7
k
0;7
e
E
a;7
=RT
p
m7
C
2
H
6
29
r
8
k
0;8
e
E
a;8
=RT
p
m
8
C
2
H
4
p
n
8
H
2
O
30
Table 1
Dry reforming of methane surface reaction mechanism from McGuire et al. (2011).
Reaction A
a
E (kJ/mol)
1 H
2
n n -H
n
H
n
1:0 10
2b
0.0
2 O
2
n n -O
n
O
n
1:0 10
2b
0.0
3 CH
4
n -CH
4
n
8:0 10
3b
0.0
4 H
2
O n -H
2
O
n
1:0 10
3b
0.0
5 CO
2
n -CO
2
n
4:8 10
2b
0.0
6 CO n -CO
n
5:0 10
1b
0.0
7 H
n
H
n
-H
2
n n
3:0 10
21
77.8
8 O
n
O
n
-O
2
n n
1:3 10
22
355.2

n
O
280
c
9 H
2
O
n
-H
2
O n
3:0 10
13
45.0
10 CO
n
-CO n
3:5 10
13
133.4

n
CO
15
c
11 CO
2
n
-CO
2
n
4:1 10
11
18.0
12 CH
4
n
-CH
4
n
1:9 10
14
25.1
13 H
n
O
n
-OH
n
n
5:0 10
22
83.7
14 OH
n
n -H
n
O
n
3:0 10
20
37.7
15 H
n
OH
n
-H
2
O
n
n
3:0 10
20
33.5
16 H
2
O
n
n -H
n
OH
n
5:0 10
22
106.4
17 OH
n
OH
n
-H
2
O
n
O
n
3:0 10
21
100.8
18 H
2
O
n
O
n
-OH
n
OH
n
3:0 10
21
171.8
19 C
n
O -CO
n
n
5:2 10
23
97.9
20 CO
n
n -C
n
O
n
2:5 10
21
169.0
21 CO
n
O
n
-CO
2
n
n
5:5 10
18
121.6
22 CO
2
n
n-CO
n
O
n
3:0 10
21
171.8
23 CO
n
H
n
-HCO
n
n
5:0 10
19
108.9
24 HCO
n
n -CO
n
H
n
3:7 10
21
0.0

n
CO
50
c
25 HCO
n
n -CH
n
O
n
3:7 10
24
59.5
26 CH
n
O
n
-HCO
n
n
3:7 10
21
167.5
27 CH
4
n
n -CH
3
n
H
n
3:7 10
21
61.0
28 CH
3
n
H
n
-CH
4
n
n
3:7 10
21
51.0
29 CH
3
n
n -CH
2
n
H
n
3:7 10
24
103.0
30 CH
2
n
H
n
-CH
3
n
n
3:7 10
23
44.1
31 CH
2
n
n -CH
n
H
n
3:7 10
24
100.0
32 CH
n
H
n
-CH
2
n
n
3:7 10
21
68.0
33 CH
n
n -C
n
H
n
3:7 10
21
21.0
34 C
n
H
n
-CH
n
n
3:7 10
21
172.8
35 CH
4
n
O
n
-CH
3
n
OH
n
1:7 10
24
80.34
36 CH
3
n
OH
n
-CH
4
n
O
n
3:7 10
21
24.27
37 CH
3
n
O
n
-CH
2
n
OH
n
3:7 10
24
120.31
38 CH
2
n
OH
n
-CH
3
n
O
n
3:7 10
21
15.06
39 CH
2
n
O
n
-CH
n
OH
n
3:7 10
24
114.5
40 CH
n
OH
n
-CH
2
n
O
n
3:7 10
21
36.82
41 CH
n
O
n
-C
n
OH
n
3:7 10
21
30.13
42 C
n
OH
n
-CH
n
O
n
3:7 10
21
136.0
a
Arrhenius parameters for rate constants. Units: pre-exponential factor A for
unimolecular reactions (s
1
), for bimolecular reactions (cm
2
mol
1
s
1
).
b
Initial Sticking coefcient S
i
0
().
c
Coverage dependent activation energy in Eq. (17). For more information see
e.g. Kee et al. (2003).
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 223
r
9
k
0;9
e
E
a;9
=RT
p
m
9
CO
p
n
9
H
2
O
31
r
10
k
0;10
e
E
a;10
=RT
p
m
10
CO
2
p
n
10
H
2
32
The kinetic parameters were estimated from experimental
results based on a one-dimensional, pseudo-homogeneous plug-
ow model. For each experiment the axial temperature prole, as
well as inlet and outlet molar concentrations were measured. The
reactor was then treated as a cascade of perfectly stirred reactors
accounting for the axial temperature. Finally, the kinetic para-
meters were estimated by an optimization routine, cf. Stansch
(1995) and Stansch et al. (1997).
3. DRM stagnation ow reactor simulations
3.1. Numerical setup
The rst set of numerical simulations is orientated toward the
stagnation-ow reactor experiments of McGuire et al. (2011).
Stagnation ow reactors are used in practical applications, such
as chemical-vapor-deposition reactors for electronic thin-lm
growth or in research on heterogeneous chemistry and reactive
ow (Warnatz et al., 1994). A gas ow is guided to a catalytic disk
which generates a viscous boundary layer. In this layer the
temperature and species compositions depend only on one inde-
pendent variable (i.e., the axial coordinate). However, the velocity
distribution can be two-dimensional or three-dimensional. Scaling
of the primary velocity leads to a simplied 1D stagnation-ow
eld approximation (Kee et al., 2003). Several groups have studied
catalytic stagnation ow to identify detailed chemistry (Warnatz
et al., 1994; Deutschmann et al., 1996; Taylor et al., 2003; McGuire
et al., 2009, 2011; Karakaya and Deutschmann, 2013).
In this paper the calculation domain of the stagnation-ow
reactor was abstracted as a at cylinder, see Fig. 1. The diameter
represents the catalytic surface in the experiments, i.e., 5.0 cm. The
height of the domain is 1.6 cm, which is in accordance with the
separation distance between the reactant inlet and the catalyst in
the experiments. A ner mesh is chosen close to the catalytic
surface where the steepest gradients of the transport quantities
appear. Toward the inlet the mesh size gets larger, see Fig. 1. After
a mesh independence study, the total number of cells amounts to
approx. 800,000. The boundary layer, approx. 3.5 mm in height, is
resolved with E20 cells in the axial coordinate z.
The conditions at the ow inlet are: velocity in axial direction
u
in
0:9 m=s, pressure p
in
1 bar, temperature T
in
27 1C, mole
fractions x
in;CO
2
=x
in;CH
4
=x
in;Ar
0:15=0:10=0:75. The reactor in the
experiments operated under steady state conditions at 40 kPa,
therefore at the outlet the pressure is set to p
out
40 kPa. At the
catalytic surface the temperature is constant at T
cat
800 1C. Only
heterogeneous reactions are taken into account. Consequently, at
the catalytic wall surface reactions occur with the microkinetics
described in Section 2.1.2 and Table 1. The catalyst is dispersed in
the porous support. Therefore, the catalytic surface is much
greater than the geometric surface. The F
cat=geo
value in Eq. (13)
is set to 90 as in McGuire et al. (2011). The total available surface
site density is 2:72 10
9
mol=cm
2
. The catalyst layer amounts
to approx. 20 m in the experiment, which leads to negligible
intraporal diffusion (Seyed-Reihani and Jackson, 2004). Conse-
quently, no additional pore model is used.
Table 2
Kinetic parameters for OCM by Stansch et al. (1997).
Step k
0;j
(mol/(g s Pa
mn
)) E
a;j
(kJ/mol) K
j;CO2
(Pa
1
) H
ad;CO2
(kJ/mol) K
O2
(Pa
1
) H
ad;O2
(kJ/mol) m
j
n
j
1 0.210
5
48 0.25 10
12
175 0.24 0.76
2 23.2 182 0.83 10
13
186 0.2310
11
124 1.0 0.40
3 0.5210
6
68 0.36 10
13
187 0.57 0.85
4 0.1110
3
104 0.40 10
12
168 1.0 0.55
5 0.17 157 0.45 10
12
166 0.95 0.37
6 0.06 166 0.16 10
12
211 1.0 0.96
7 1.210
7
226 1.0
8 9.310
3
300 0.97 0
9 0.1910
3
173 1.0 1.0
10 0.2610
1
220 1.0 1.0
Inlet
Outlet
Catalytic Surface
D = 50 mm
H

=
1
6

m
m
Heated
catalytic surface
D = 50 mm
Reactant
Feed
Microprobe
r
z
Fig. 1. (a) Scheme of stagnation ow reactor. (b) Calculation grid for the 3D stagnation ow reactor with closer view of local mesh renement.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 224
3.2. Results and discussion
Fig. 2 shows the temperature and axial velocity proles as a
function of the axial distance from the surface at a radial distance
of 10 mm from the center of the domain. The one-dimensional
simulations are based on an axisymmetric stagnation-ow bound-
ary-layer problem, cf. Kee et al. (2003). It can be clearly seen, that
the one-dimensional prediction of McGuire et al. (2011) and the
three-dimensional simulation are in excellent agreement.
In Figs. 3 and 4 mole fractions over distance from surface are
illustrated. Only the near-surface boundary layer is shown
z 03 mm. The crosses indicate the experimentally measured
mole fractions from McGuire. The lines represent the simulations.
Again, the one-dimensional and three-dimensional proles are in
excellent agreement. The microkinetic 1D and 3D models predict
the experiments accurately, except for the measurement of CO
2
and CO closest to the surface. This is due to the inuence of the
microprobe to the uid eld, cf. McGuire et al. (2011).
Fig. 4 shows the mole fractions over distance for the DRM
products, i.e., CO and H
2
. As a result of excess CO
2
, the reverse
watergas-shift process (H
2
CO
2
COH
2
O) causes an
increased CO:H
2
ratio. The boundary layer of H
2
is larger than
the boundary layer of CO, due to the higher mass diffusivity of H
2
at these temperatures. The simulations predict the product con-
centration proles in good accuracy to the experiments. And again,
one-dimensional and three-dimensional simulations t very well.
Comparing the concentration proles between one-dimensional
and three-dimensional simulations, it is evident that a higher spatial
resolution of the domain does not lead to more information. Fig. 5
illustrates this nding. The temperature and concentration proles
show a one-dimensional orientation toward the catalytic surface
except for the edges of the domain. It can be concluded that the
stagnation-ow reactor can be simulated adequately by a one-
dimensional model. It has to be kept in mind that a measure device,
e.g., a microprobe, has always an inuence on the ow and concen-
tration eld. In the case of the stagnation-ow reactor measurements
close to the surface are affected. In other cases, e.g., honeycomb
channels, the in situ measurement devices have a stronger inuence of
the ow eld, as Hettel et al. (2013) showed recently by CFD
simulations. Nonetheless, the stagnation-ow conguration is well
suited to investigate reaction kinetics with a low inuence of uidic
effects. In addition, modeling the pore diffusion would enrich such
calculations.
4. OCM xed-bed reactor simulations
4.1. Numerical setup
The second set of numerical simulations reproduces the xed-
bed reactor experiments performed by Stansch et al. (1997). In
these experiments a small microreactor with an inner diameter of
6 mm was used made up of quartz to suppress the inuence of the
reactor material. A La
2
O
3
=CaO catalyst was applied with particle
diameters of approx. 0.3 mm. The catalytic bed was diluted with
quartz particles to minimize hot spot formation. The dilution ratio,
i.e., mass of quartz to mass of catalyst, accounted to six. The
reactor itself was placed in a thermostated uidized bed. Inside
the xed bed a thermocouple moving in axial direction measured
temperature proles. In the experiments no radial temperature
proles were measured. The reactor operated under steady-state
conditions. In the experiments, the feed gas, containing O
2
and
CH
4
, was diluted with N
2
to suppress high temperature formation.
For different modied contact times ( m
cat
=
_
V in kg s/m
3
) the
reactor was heated until a specic hot spot temperature, i.e., the
0
100
200
300
400
500
600
700
800
0 2 4 6 8 10
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
T
e
m
p
e
r
a
t
u
r
e

[

C
]
A
x
i
a
l

v
e
l
o
c
i
t
y

[
m
/
s
]
Distance from surface z [mm]
T
v
z
3D
1D
Fig. 2. Temperature and velocity prole of the stagnation-ow simulation.
One-dimensional (from McGuire et al., 2011) and three-dimensional results.
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0 0.5 1 1.5 2 2.5 3
M
o
l
e

f
r
a
c
t
i
o
n

[
-
]
Distance from surface z [mm]
Inlet
3D
1D
Exp
CO
2
CH
4
Fig. 3. Mole fractions for CH
4
and CO
2
over distance from surface. Comparison
between experiments, 1D simulations from McGuire et al. (2011) and 3D
simulations.
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0 0.5 1 1.5 2 2.5 3
M
o
l
e

f
r
a
c
t
i
o
n

[
-
]
Distance from surface z [mm]
3D
1D
Exp
CO
H
2
Fig. 4. Mole fractions of H
2
and CO over distance from surface. Comparison
between experiments, 1D simulations from McGuire et al. (2011) and 3D
simulations.
D = 50 mm
H

=

1
6

m
m
Fig. 5. Temperature distribution (left) and CH
4
molar fraction (right) of the
stagnation-ow reactor.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 225
maximum temperature in the xed-bed, was reached. Then, the
inlet and outlet feed composition was measured. For more
information see Stansch et al. (1997).
A schematic view of the calculation domain with detail on
mesh renement in the catalytic zone gives Fig. 6. The total
number of cells accounts for approx. 140,000.
Parameters utilized for the CFD simulations of the OCM are
listed in Table 3. In the upstream part a non-reacting porous
medium was assumed with the same properties than in the
catalytic bed. The mass and therefore the height of the catalytic
bed varied with modied contact time m
cat
=
_
V. Inlet velocities
were calculated from these modied contact times. In the catalytic
zone the reaction mechanism was implemented, see Chapter 2.2.
Two temperature settings were applied. For the rst setting the
reactor was simulated with a measured one-dimensional axial
temperature prole from Stansch (1995), see Fig. 7. For the second
setting heat transfer q
s
(W/m
2
) was applied from inside of the
reactor to the ambiance, which is dened by T
ambient
:
q
s
hT
S
T
ambient
33
where h is the heat transfer coefcient, which is set to 2070 W/m
2
K,
cf. Stansch (1995), and T
S
is the surface temperature. The inlet
temperature as well as the ambient temperature were applied in
such a way that the maximum temperature in the reactor met the
specic hot-spot temperature of the experiments, i.e., 700 or 830 1C.
A radiation model was not used.
4.2. Results and discussion
In the rst set of simulations the axial temperature prole from
Stansch (1995) was applied. This led to one-dimensional concen-
tration proles. In Fig. 7 the mole fractions of O
2
, CO
2
, C
2
H
4
and
C
2
H
6
over the reactor length are shown. As it can be seen, the
temperature of the xed bed already increases before the gas ow
enters the catalytic zone (starting at z0). This is due to conduc-
tion in the solid and through the uid. After the temperature
reaches its hot-spot at z 1 mm, it decreases due to the absence of
oxygen and heat transfer downstream. Diffusion of several species
upstream of the catalytic bed can be noticed. Due to the high hot-
spot temperature of 850 1C and a high modied contact time,
oxygen is already totally consumed after 1.5 mm. After that
distance the mole fraction proles of ethane and ethylene change
only slightly. The concentration of carbon dioxide increases until
the end of the reactor. Without oxygen only reaction steps 710
are active.
Fig. 8 compares reactor quantities between experiments and
plug ow simulations of Stansch et al. (1997) and CFD simulations
with the applied axial temperature prole for inlet conditions:
x
in;CH
4
=x
in;O
2
=x
in;N
2
0:689=0:097=0:214, modied contact time
(37 kg s/m
3
) and hot-spot temperature (850 1C). Although the
reactor was modeled three-dimensional, the applied axial tem-
perature prole leads to an almost one-dimensional species
distribution (not shown). As it can be seen, the simulations
performed with the applied axial temperature prole are in very
good agreement with the simulated results reported by Stansch
(1995). In that case, a 3D or 2D simulation does not provide more
information than the plug ow model. However, for up-scale
modeling temperature proles are not known but have to be
calculated. Therefore, not only the uid dynamic model, e.g., plug
ow or full three-dimensional, but also the applied kinetics
inuence the temperature prole.
The plug ow model is one of the easiest assumptions simulat-
ing chemical reactors. Schuurman (2008) and Kapteijn and
Moulijn (2008) give overviews of laboratory catalytic reactors
and modeling. But for catalytic combustion these models fail due
to excessive simplications in heat and mass transfer (Dalle
Nogare et al., 2011), e.g., resulting in radial proles. The inuence
of heat released by reaction is of great importance for the kinetic
description. Consequently, the kinetics will be more inuenced by
the temperature prole than vice versa (Dalle Nogare et al., 2008,
2011). In the second set of simulations not the entire temperature
Inlet
D = 6.0 mm
Upstream
Catalytic
Fixed-Bed
H = 2.0 - 51 mm
Downstream
Outlet
r
z

Fig. 6. Schematic xed-bed reactor of the OCM simulations with detail on local
mesh renement in the catalytic zone.
Table 3
Parameter setting for CFD simulations.
Symbol Value
Particle diameter d
P
(mm) 0.3
Inner tube diameter D (mm) 6.0
Bed porosity () 0.5
Thermal conductivity solid (W/m K) 1.5 (VDI, 2013)
Volume gas ow _
Vgas (m
3
/s)
4 10
6
Dilution ratio DR () 6.0
Mass of catalyst m
cat
(mg) 6.8/14.3/30.3/59.5/99.7/199.7
Height of the catalytic
packed bed
H
R
(cm) 0.2/0.4/0.8/1.9/2.7/5.1
Contact time
mcat =
_
V (kg s/m
3
)
1.7/3/7/15/25/50
Kinetic parameters See 2
700
750
800
850
900
-2 0 2 4 6 8
0
0.025
0.05
0.075
0.1
T
e
m
p
e
r
a
t
u
r
e

[

C
]
M
o
l
e

f
r
a
c
t
i
o
n

x
k

[
-
]

Axial position z [mm]
Exp.
Sim.
O
2
CO
2
C
2
H
6
C
2
H
4
Temperature
Reacting zone
Fig. 7. Axial temperature prole from Stansch (1995) and simulated mole fraction
proles of species O
2
, CO
2
, C
2
H
6
and C
2
H
4
. Inlet conditions: x
in;CH4
=x
in;O2
=
x
in;N2
0:689=0:097=0:214; modied contact time 37 kg s/m
3
; hot-spot tempera-
ture 850 1C.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 226
prole was applied but only the ambient temperature and inlet
temperature, to reach a specic hot-spot temperature. In the
following, different contact times and hot-spot temperature con-
gurations are compared with experimental results.
Fig. 9 shows the calculated temperature prole and mole
fraction proles over the axial position for a hot-spot temperature
of 830 1C and a modied contact time of 15 kg s/m
3
. Species
proles look similar to Fig. 7. Only the axial temperature prole
differs in terms of spreading.
However, more information give two-dimensional views of the
reactor, see Figs. 10 and 11. As mentioned before, temperature
distributions are very important for the reactor characteristics.
Fig. 10 shows the two-dimensional temperature prole and
methane prole for the same setting as Fig. 9. Signicant radial
temperature proles appear with up to 50 K temperature differ-
ence. This inuences dramatically the species production rates and
hence the species concentrations. Radial proles can be noticed for
concentrations, too. Due to the temperature prole, the lowest
methane concentration is reached rst in the center of the reactor.
Similar proles appear for O
2
. The species CO
2
, C
2
H
4
and C
2
H6 are
produced earlier in the axis of the xed bed due to the tempera-
ture distribution.
Recently, axial and radial temperature proles were measured
for a exothermic catalytic foam and xed-bed for the hydrogena-
tion of benzene, cf. Grf et al. (2014). They clearly showed radial
temperature distributions. Temperature differences between the
axis and close to the wall regions reached up to 30 K. This indicates
that temperature distributions, both axial and radial, inside
tubular reactors play an important role.
Finally, Figs. 12 and 13 compare different reactor quantities
over modied contact time between experimental results and
simulations. The larger the contact time, the greater the oxygen
and methane conversion and consequently the larger the yield of
CO
2
and C
2
products, respectively. In Fig. 12 the simulated results
at 830 1C t well with the experimental results for the conversion
of O
2
and CH
4
. But in Fig. 13 the simulated C
2
yields are lower than
the experimental values. Consequently, the simulated CO
2
yield is
larger than the experimental. It has to be noticed that the methane
conversion is limited due to full oxygen conversion. Nevertheless,
it seems that the reactivity is too high. This trend can be also
observed for the settings with a hot-spot temperature of 700 1C.
For modied contact times larger than 10 kg s/m
3
the simulated
O
2
conversion exceeds the experimental. Because of the stoichio-
metry, the methane conversion follows. Similar ndings give the
yields of C
2
and CO
2
.
0
20
40
60
80
100
X
CH4
X
O2
S
C2
Y
C2H4
Y
C2H6
Y
CO
Y
CO2
C
o
n
v
e
r
s
i
o
n

X
,

s
e
l
e
c
t
i
v
i
t
y

S
,

y
i
e
l
d

Y

[
%
]
Stansch-experiment
Stansch-simulation
1D-Temp-profile-simulation
Fig. 8. Comparison of conversion, selectivity and yield for different species
between experiments and simulations of Stansch et al. (1997) and simulations
with axial temperature prole.
750
775
800
825
850
0 5 10 15 20 25
0
0.025
0.05
0.075
0.1
T
e
m
p
e
r
a
t
u
r
e

[

C
]
M
o
l
e

f
r
a
c
t
i
o
n

x
k

[
-
]
Axial position z [mm]
O
2
CO
2
C
2
H
6
C
2
H
4
Temperature
Reacting zone
Fig. 9. Simulated axial temperature prole and mole fraction proles of species O
2
,
CO
2
, C
2
H
6
and C
2
H
4
. Inlet conditions: x
in;CH4
=x
in;O2
=x
in;N2
0:702=0:098=0:2; mod-
ied contact time 15 kg/m
3
; hot-spot temperature 830 1C.
0 3 mm 3 mm
1
9

m
m
Fig. 10. Temperature (left) and mole fraction prole of CH
4
(right) for hot-spot
temperature of 830 1C and modied contact time of 15 kg s/m
3
. Height of catalytic
bed 19 mm.
0 3 mm 3 mm
1
9

m
m
Fig. 11. Mole fraction proles of C
2
H
4
(left) and C
2
H
6
(right) for hot-spot
temperature of 830 1C and modied contact time of 15 kg s/m
3
. Height of catalytic
bed 19 mm.
0
20
40
60
80
100
0 10 20 30 40 50
0
10
20
30
40
50
C
o
n
v
e
r
s
i
o
n

O
2

[
%
]
C
o
n
v
e
r
s
i
o
n

C
H
4

[
%
]
Modified contact time [kg s/m
3
]
Exp. O
2
Sim. O
2
Exp. CH
4
Sim. CH
4
700 C
830 C
Fig. 12. Conversion of O
2
and CH
4
over modied contact time. Experimental results
(points) from Stansch et al. (1997) and simulation results (lines) for different hot-
spot temperatures of 700 1C (solid and crosses) and 830 1C (dashed and circles).
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 227
As already mentioned, the temperature prole highly inu-
ences the reactivity. One explanation of the differences between
3D simulations and the experiments of Stansch and co-workers
give the overall temperature distribution, see Fig. 10. In the highly
exothermic OCM process, radial temperature proles appear.
Stansch and co-workers did not take the energetic deviations
between plug ow and real reactor into account. That leads to an
overestimation of reactivity, because higher temperatures are
assumed in near-wall regions, where in reality lower temperatures
occur. As demonstrated in Figs. 12 and 13 for hot-spot temperature
of 700 1C oxygen conversion is overestimated for similar max-
imum temperatures in the reactor.
Fig. 14 helps to summarize the limitations of the Stansch OCM
kinetics. It shows the measured temperature prole in the axial
position in the center of the xed-bed reactor, as well as two
proles of simulations, i.e., variation 1 and variation 2. In variation
1 the same temperature boundary conditions as in the experi-
ments were applied, i.e., T
inlet
T
ambient
680 1C. In variation 2 the
ambient temperature and inlet temperature were modied until
the hot spot of 850 1C is reached. The lled curves represent the
highest and lowest temperature at an axial position, i.e., radial
temperature proles. As it can be seen, both simulated tempera-
ture proles differ a lot from the measured prole. An accurate
temperature prediction fails due to the applied kinetics. Therefore,
species proles cannot be predicted correctly.
It can be concluded that the Stansch kinetic model of the OCM,
besides thermodynamically inconsistency and suppression of gas
phase reactions, cf. Arndt et al. (2012), underlies a wrong assump-
tion for temperature distribution on one side. On the other side,
the physical signicance of kinetic parameters is questionable.
Therefore, multi-dimensional simulations with this kinetic model
might lead in specic cases to uncertain results.
The applied porous media model neglects certain physical
phenomena, e.g., mass and heat transfer resistances due to
boundary layers, intraparticle diffusion, etc. Due to the high
tube-to-particle-ratio of 20, the porous-media model sounds
appropriate for the Stansch reactor. An extensive study on the
inuence of different models for the ow and chemistry in
catalytic channels was presented by Raja et al. (2000). In specic
cases simplied models are appropriate. In others they are not
leading to incorrect predictions. 2D porous-media models can
simulate certain xed-bed reactor settings, especially for large
tube-to-particle-ratios. Several additions, e.g., dispersion terms or
radial porosity proles, enrich the simplied model. However, for
low tube-to-particle-ratios and highly exothermic or endothermic
reactions the xed bed should be resolved by means of particles,
cf. Freund et al. (2005) and Dixon and Nijemeisland (2001).
Backow, asymmetric ow and wall effects are dominant, which
inuences species concentrations in reactive ows signicantly.
For such systems spatially resolved uid dynamics with detailed
reaction mechanisms that distinguish between adsorption, surface
reaction and desorption, as well as gas phase reactions are
required. Coupling detailed uid dynamics with lumped reaction
descriptions can lead to erroneous predictions.
On the one hand, the parameters of the Stansch model could be
adjusted with the consideration of radial temperature proles
when combined with porous-media models. However, the nal
kinetic model has to reproduce accurately the temperature prole.
But still, LHHW models involve certain limitations (Salciccioli
et al., 2011). On the other hand, an OCM kinetics should be
developed by using micro-kinetic models and well dened reactor
congurations, as it is standard in combustion research. An
example of a kinetics that follows this direction was presented
by Sun et al. (2008) and recently by Alexiadis et al. (2014).
Furthermore, multiscale considerations have to be taken into
account, cf. Vlachos (2005) and Mhadeshwar and Vlachos (2005).
5. Conclusion
The dramatically increased computer aided chemical engineer-
ing design calls for reliable kinetics. Especially in fast highly
endothermic or exothermic catalytic reactions the chemistry and
transport phenomena are strongly connected. This fact compli-
cates the nding of intrinsic kinetics. In this paper, two examples
are presented how uid dynamics inuence lab scale kinetic
parameter identication in different ways.
In the rst case the dry reforming of methane in a stagnation-
ow reactor was simulated fully three-dimensional with a detailed
surface reaction mechanism. It is shown that the three-dimensional
formulation gives no more information than the stagnation-ow
boundary-layer problem, which is one-dimensional. In the applied
well dened reactor a boundary layer is formed toward the heated
catalytic surface. Consequently, surface as well as gas phase con-
centration proles can be measured promptly to estimate kinetic
data. Furthermore, the boundary-layer problem describes the reac-
tor in an adequate manner. No time-consuming three-dimensional
CFD simulations have to be carried out. This reactor setting is a
valuable example of how uidic effects on kinetic parameter
estimation can be suppressed.
In the second case a xed-bed reactor setting of the oxidative
coupling of methane was investigated. The reactor was simulated
three-dimensionally with the porous-media model and a 10-step
reaction mechanism. It is shown that the assumed plug ow
model, underlying the experimental reaction mechanism determi-
nation, fails in this highly exothermic reactor. Temperature
0
2
4
6
8
10
12
14
0 10 20 30 40 50
0
1
2
3
4
5
6
7
C
o
n
v
e
r
s
i
o
n

C
2

[
%
]
C
o
n
v
e
r
s
i
o
n

C
O
2

[
%
]
Modified contact time [kg s/m
3
]
Exp. C
2
Sim. C
2
Exp. CO
2
Sim. CO
2
700 C
830 C
Fig. 13. Yield of C
2
and CO
2
over modied contact time. Experimental results
(points) from Stansch et al. (1997) and simulation results (lines) for different hot-
spot temperatures of 700 1C (solid and crosses) and 830 1C (dashed and circles).
550
600
650
700
750
800
850
900
-8 -6 -4 -2 0 2 4 6 8
T
e
m
p
e
r
a
t
u
r
e

[

C
]
Axial position z [mm]
Stansch
Var. 1
Var. 2
Hot-spot temperature
Reacting zone
Fig. 14. Axial temperature prole from Stansch (1995) and 3D simulated proles.
Inlet conditions: x
in;CH4
=x
in;O2
=x
in;N2
0:689=0:097=0:214; modied contact time
37 kg s m
3
. Variation 1: hot-spot-temperature 850 1C, T
inlet
735 1C. Variation 2:
T
inlet
680 1C.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 228
distribution inuences the kinetics in a serious way. Radial
temperature proles and therefore radial concentration proles
were observed. The negligence of these radial temperature proles
led to an erroneous kinetic parameter estimation. The reactivity of
the OCM reaction was underestimated in the model. In addition,
measured temperature proles could not be reproduced. Hence,
the usage of this particular OCM kinetics might lead to incorrect
predictions for reactor scale-up.
As a conclusion, it can be said that kinetics of highly exothermic
or endothermic reactions should be carried out in well dened
reactor congurations in combination with spatially resolved
concentration and temperature measurements.
Acknowledgments
This study is part of the Cluster of Excellence Unifying Concepts
in Catalysis (Unicat), which is coordinated by the Technische
Universitt Berlin. The authors would like to thank the Deutsche
Forschungsgemeinschaft DFG within the framework of the German
Initiative of Excellence for nancial support.
References
Alexiadis, V., Thybaut, J., Kechagiopoulos, P., Chaar, M., Veen, A.V., Muhler, M., Marin, G.,
2014. Oxidative coupling of methane: catalytic behaviour assessment via compre-
hensive microkinetic modelling. Appl. Catal. B: Environ. 150151, 496505, http://
www.sciencedirect.com/science/article/pii/S0926337313007881.
Arndt, S., Laugel, G., Levchenko, S., Horn, R., Baerns, M., Schefer, M., Schlgl, R.,
Schomcker, R., 2011. A critical assessment of Li/MgO-based catalysts for the
oxidative coupling of methane. Catal. Rev. 53 (4), 424514, http://www.
tandfonline.com/doi/abs/10.1080/01614940.2011.613330.
Arndt, S., Otremba, T., Simon, U., Yildiz, M., Schubert, H., Schomcker, R., 2012. Mn
Na
2
WO
4
/SiO
2
as catalyst for the oxidative coupling of methane. What is really
known? Appl. Catal. A: Gen. 425426, 5361, http://www.sciencedirect.com/
science/article/pii/S0926860X12001330.
Bradford, M.C.J., Vannice, M.A., 1999. CO
2
reforming of CH
4
. Catal. Rev. 41 (1), 142,
http://www.tandfonline.com/doi/abs/10.1081/CR-100101948.
CD-adapco, 2013. STAR-CCM 8.04. www.cd-adapco.com.
Dalle Nogare, D., Degenstein, N., Horn, R., Canu, P., Schmidt, L., 2008. Modeling
spatially resolved proles of methane partial oxidation on a rh foam catalyst
with detailed chemistry. J. Catal. 258 (1), 131142, http://www.sciencedirect.
com/science/article/pii/S0021951708002182.
Dalle Nogare, D., Degenstein, N., Horn, R., Canu, P., Schmidt, L., 2011. Modeling
spatially resolved data of methane catalytic partial oxidation on rh foam
catalyst at different inlet compositions and owrates. J. Catal. 277 (2), 134148,
http://www.sciencedirect.com/science/article/pii/S0021951710003775.
Dautzenberg, M.F., 1989. Ten Guidelines for Catalyst Testing. ACS Symposium Series,
vol. 411, pp. 99119. http://pubs.acs.org/doi/abs/10.1021/bk-1989-0411.ch011
(Chapter 12).
Deutschmann, O., 2008. Computational uid dynamics simulation of catalytic reactors.
In: Handbook of Heterogeneous Catalysis. Wiley-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, pp. 18111828, http://dx.doi.org/10.1002/9783527610044.
hetcat0097 (Chapter 6.6).
Deutschmann, O., Schmidt, R., Behrendt, F. Warnatz, J., 1996. Numerical modeling of
catalytic ignition. In: Twenty-Sixth International Symposium on Combustion,
The Combustion Institute, Pittsburgh, pp. 17471754.
Dixon, A.G., 2012. Fixed bed catalytic reactor modelling the radial heat transfer
problem. Can. J. Chem. Eng. 90 (3), 507527, http://dx.doi.org/10.1002/cjce.
21630.
Dixon, A.G., Nijemeisland, M., 2001. CFD as a design tool for xed-bed reactors. Ind.
Eng. Chem. Res. 40 (23), 52465254, http://pubs.acs.org/doi/abs/10.1021/
ie001035a.
Dudukovic, M.P., 2009. Frontiers in reactor engineering. Science 325 (5941), 698701,
http://www.sciencemag.org/content/325/5941/698.abstract.
Dudukovic, M.P., 2010. Reaction engineering: status and future challenges.
Chem. Eng. Sci. 65 (1), 311, http://www.sciencedirect.com/science/article/
pii/S0009250909006125.
Freund, H., Bauer, J., Zeiser, T., Emig, G., 2005. Detailed simulation of transport
processes in xed-beds. Ind. Eng. Chem. Res. 44 (16), 64236434, http://pubs.
acs.org/doi/abs/10.1021/ie0489453.
Froment, G., Bischoff, K., 1990. Chemical Reactor Analysis and Design. Wiley,
London
Grf, I., Rhl, A.K., Kraushaar-Czarnetzki, B., 2014. Experimental study of heat transport
in catalytic sponge packings by monitoring spatial temperature proles in a cooled-
wall reactor. Chem. Eng. J. 244, 234242, http://dx.doi.org/10.1016/j.cej.2014.01.060
(0). http://www.sciencedirect.com/science/article/pii/S1385894714000850.
Hartmann, M., Maier, L., Minh, H., Deutschmann, O., 2010. Catalytic partial
oxidation of iso-octane over rhodium catalysts: an experimental, modeling,
and simulation study. Combust. Flame 157 (9), 17711782, http://www.
sciencedirect.com/science/article/pii/S0010218010000830.
Hettel, M., Diehm, C., Torkashvand, B., Deutschmann, O., 2013. Critical evaluation of
in situ probe techniques for catalytic honeycomb monoliths. Catal. Today 216, 210,
http://www.sciencedirect.com/science/article/pii/S0920586113002289.
Horn, R., Degenstein, N., Williams, K., Schmidt, L., 2006. Spatial and temporal
proles in millisecond partial oxidation processes. Catal. Lett. 110 (34),
169178, http://dx.doi.org/10.1007/s10562-006-0117-8.
Horn, R., Williams, K., Degenstein, N., Bitsch-Larsen, A., Nogare, D.D., Tupy, S.,
Schmidt, L., 2007. Methane catalytic partial oxidation on autothermal rh and
pt foam catalysts: oxidation and reforming zones, transport effects, and
approach to thermodynamic equilibrium. J. Catal. 249 (2), 380393, http://
www.sciencedirect.com/science/article/pii/S0021951707001753.
Kapteijn, F., Moulijn, J.A., 2008. Laboratory catalytic reactors: aspects of catalyst testing.
In: Handbook of Heterogeneous Catalysis. Wiley-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, pp. 20192045, http://dx.doi.org/10.1002/9783527610044.
hetcat0108 (Chapter 9.1).
Karakaya, C., Deutschmann, O., 2013. Kinetics of hydrogen oxidation on Rh/Al
2
O
3
catalysts studied in a stagnation-ow reactor. Chem. Eng. Sci. 89, 171184,
http://www.sciencedirect.com/science/article/pii/S0009250912006501.
Kee, R.J., Colin, M.E., Glarborg, P., 2003. Chemically Reacting Flow, Theory and
Practice. Wiley, Hoboken, NJ
Kondratenko, E.V., Baerns, M., 2008. Oxidative coupling of methane. In: Handbook of
Heterogeneous Catalysis. Wiley-VCHVerlag GmbH& Co. KGaA, Weinheim, Germany,
pp. 30103023, http://dx.doi.org/10.1002/9783527610044.hetcat0152 (Chapter 13.17).
Korup, O., Goldsmith, C.F., Weinberg, G., Geske, M., Kandemir, T., Schlgl, R., Horn, R.,
2013. Catalytic partial oxidation of methane on platinum investigated by
spatial reactor proles, spatially resolved spectroscopy, and microkinetic
modeling. J. Catal. 297 (0), 116, http://www.sciencedirect.com/science/
article/pii/S0021951712002795.
Lunsford, J.H., 2000. Catalytic conversion of methane to more useful chemicals and
fuels: a challenge for the 21st century. Catal. Today 63 (24), 165174, http://
www.sciencedirect.com/science/article/pii/S0920586100004569.
Maestri, M., Vlachos, D.G., Beretta, A., Groppi, G., Tronconi, E., 2008. Steam and dry
reforming of methane on Rh: microkinetic analysis and hierarchy of kinetic
models. J. Catal. 259 (2), 211222, http://www.sciencedirect.com/science/
article/pii/S0021951708003138.
Marcilly, C., 2003. Present status and future trends in catalysis for rening and
petrochemicals. J. Catal. 216 (12), 4762 40th Anniversary Commemorative Issue.
http://www.sciencedirect.com/science/article/pii/S002195170200129X.
McGuire, N.E., Sullivan, N.P., Deutschmann, O., Zhu, H., Kee, R.J., 2011. Dry reforming
of methane in a stagnation-ow reactor using Rh supported on strontium-
substituted hexaaluminate. Appl. Catal. A: Gen. 394 (12), 257265, http://
www.sciencedirect.com/science/article/pii/S0926860X11000147.
McGuire, N.E., Sullivan, N.P., Kee, R.J., Zhu, H., Nabity, J.A., Engel, J.R., Wickham, D.T.,
Kaufman, M.J., 2009. Catalytic steam reforming of methane using Rh supported
on Sr-substituted hexaaluminate. Chem. Eng. Sci. 64 (24), 52315239, http://
www.sciencedirect.com/science/article/pii/S0009250909005806.
Mhadeshwar, A.B., Vlachos, D.G., 2005. Hierarchical multiscale mechanism devel-
opment for methane partial oxidation and reforming and for thermal decom-
position of oxygenates on rh. J. Phys. Chem. B 109 (35), 1681916835, http://
pubs.acs.org/doi/abs/10.1021/jp052479t.
Mladenov, N., Koop, J., Tischer, S., Deutschmann, O., 2010. Modeling of transport
and chemistry in channel ows of automotive catalytic converters. Chem.
Eng. Sci. 65 (2), 812826, http://www.sciencedirect.com/science/article/pii/
S000925090900637X.
Pfefferle, L.D., 1995. Heterogeneous/homogeneous reactions and transport coupling for
catalytic combustion systems: a review of model alternatives. Catal. Today 26 (34),
255265, Selected papers presented at the International Workshop on Catalytic
Combustion. http://www.sciencedirect.com/science/article/pii/0920586195001477.
Quiceno, R., Prez-Ramrez, J., Warnatz, J., Deutschmann, O., 2006. Modeling the
high-temperature catalytic partial oxidation of methane over platinum gauze:
detailed gas-phase and surface chemistries coupled with 3D ow eld simula-
tions. Appl. Catal. A: Gen. 303 (2), 166176, http://www.sciencedirect.com/
science/article/B6TF5-4JJGB4N-2/2/e6bb7e0c60a2f10ede037a360e67c39f.
Raja, L.L., Kee, R.J., Deutschmann, O., Warnatz, J., Schmidt, L.D., 2000. A
critical evaluation of NavierStokes, boundary-layer, and plug-ow models of
the ow and chemistry in a catalytic-combustion monolith. Catal. Today 59
(12), 4760, http://www.sciencedirect.com/science/article/B6TFG-4066STY-5/
2/a46b7955eaee395453193ea7c4b9980d.
Ren, Z., Pope, S.B., 2008. Second-order splitting schemes for a class of reactive
systems. J. Comput. Phys. 227 (17), 81658176, http://www.sciencedirect.com/
science/article/pii/S0021999108003057.
Rinnemo, M., Deutschmann, O., Behrendt, F., Kasemo, B., 1997. Experimental and
numerical investigation of the catalytic ignition of mixtures of hydrogen and
oxygen on platinum. Combust. Flame 111 (4), 312326, http://www.sciencedir
ect.com/science/article/pii/S0010218097000023.
Sa, J., Fernandes, D.L.A., Aiouache, F., Goguet, A., Hardacre, C., Lundie, D., Naeem, W.,
Partridge, W.P., Stere, C., 2010. Spacims: spatial and temporal operando
resolution of reactions within catalytic monoliths. Analyst 135, 22602272,
http://dx.doi.org/10.1039/C0AN00303D.
Salciccioli, M., Stamatakis, M., Caratzoulas, S., Vlachos, D., 2011. A review of
multiscale modeling of metal-catalyzed reactions: mechanism development
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 229
for complexity and emergent behavior. Chem. Eng. Sci. 66 (19), 43194355,
http://www.sciencedirect.com/science/article/pii/S000925091100368X.
Schuurman, Y., 2008. Aspects of kinetic modeling of xed bed reactors. Catal.
Today 138 (12), 1520, Conference Paul Sabatier: Catalysis Contributions
to Key Societal Challenges. http://www.sciencedirect.com/science/article/pii/
S0920586108001752.
Seyed-Reihani, S.-A., Jackson, G.S., 2004. Effectiveness in catalytic washcoats
with multi-step mechanisms for catalytic combustion of hydrogen. Chem.
Eng. Sci. 59 (24), 59375948, http://www.sciencedirect.com/science/article/
pii/S0009250904004646.
Sinev, M.Y., Fattakhova, Z., Lomonosov, V., Gordienko, Y.A., 2009. Kinetics of
oxidative coupling of methane: bridging the gap between comprehension
and description. J. Nat. Gas Chem. 18 (3), 273287, http://www.sciencedirect.
com/science/article/pii/S1003995308601280.
Stansch, Z., 1995. Kinetische Untersuchung zur oxidativen Kupplung von Methan an
einem La
2
O
3
/CaO-Katalysator (Ph.D. thesis), Ruhr-Universitt Bochum (in
German).
Stansch, Z., Mleczko, L., Baerns, M., 1997. Comprehensive kinetics of oxidative
coupling of methane over the La
2
O
3
/CaO catalyst. Ind. Eng. Chem. Res. 36 (7),
25682579, http://pubs.acs.org/doi/abs/10.1021/ie960562k.
Sun, J., Thybaut, J.W., Marin, G.B., 2008. Microkinetics of methane oxidative coupling.
Catal. Today 137 (1), 90102, Recent Developments in Combinatorial Catalysis
Research and High-Throughput Technologies. http://www.sciencedirect.com/
science/article/B6TFG-4SBYYH7-2/2/83ac857026d1413038920b1d034f98ce.
Taylor, J.D., Allendorf, M.D., McDaniel, A.H., Rice, S.F., 2003. In situ diagnostics and
modeling of methane catalytic partial oxidation on pt in a stagnation-ow
reactor. Ind. Eng. Chem. Res. 42 (25), 65596566, http://pubs.acs.org/doi/abs/
10.1021/ie020934r.
Urakawa, A., Baiker, A., 2009. Space-resolved proling relevant in hetero-
geneous catalysis. Top. Catal. 52 (10), 13121322, http://dx.doi.org/10.1007/
s11244-009-9312-3.
VDI, 2013. VDI Heat Atlas. Berlin, Heidelberg: Springer. pp. 551614.
Vlachos, D.G., 2005. A review of multiscale analysis: examples from systems
biology, materials engineering, and other uidsurface interacting systems.
In: Marin, G.B. (Ed.), Advances in Chemical Engineering Multiscale Analysis.
of Advances in Chemical Engineering, vol. 30Academic Press, Amsterdam,
The Netherlands, pp. 161, http://www.sciencedirect.com/science/article/pii/
S0065237705300019
Warnatz, J., Allendorf, M.D., Kee, R.J., Coltrin, M.E., 1994. A model of elementary
chemistry and uid mechanics in the combustion of hydrogen on platinum
surfaces. Combust. Flame 96 (4), 393406, http://www.sciencedirect.com/
science/article/pii/0010218094901074.
Wolf, D., Hohenberger, M., Baerns, M., 1997. External mass and heat transfer
limitations of the partial oxidation of methane over a Pt/MgO catalyst
consequences for adiabatic reactor operation. Ind. Eng. Chem. Res. 36 (8),
33453353, http://pubs.acs.org/doi/abs/10.1021/ie960739a.
Yuan, T., Lai, Y.-H., Chang, C.-K., 2008. Numerical studies of heterogeneous reaction
in stagnation ows using one-dimensional and two-dimensional cartesian
models. Combust. Flame 154 (3), 557568, http://www.sciencedirect.com/
science/article/pii/S0010218008001958.
G.D. Wehinger et al. / Chemical Engineering Science 111 (2014) 220230 230