See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277140571

Modeling of oxidative dehydrogenation of ethane to ethylene on a

MoVTeNbO/TiO2 catalyst in an industrial-scale packed bed catalytic reactor

Article  in  The Chemical Engineering Journal · November 2015

DOI: 10.1016/j.cej.2015.05.128

CITATIONS READS

17 481

4 authors:

Gamaliel Che-Galicia R.s. Ruiz

Benemérita Universidad Autónoma de Puebla Metropolitan Autonomous University
10 PUBLICATIONS   65 CITATIONS    28 PUBLICATIONS   217 CITATIONS   

SEE PROFILE SEE PROFILE

Felipe Lopez-Isunza Carlos O. Castillo-Araiza

Metropolitan Autonomous University Metropolitan Autonomous University- Campus Iztapalapa
57 PUBLICATIONS   670 CITATIONS    50 PUBLICATIONS   245 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

On the Understanding and Scaling up of Catalytic Selective Oxidations. View project

Elucidating Kinetic, Adsorption and Partitioning Phenomena from a Single Well Tracer Method: Laboratory and Bench Scale Studies View project

All content following this page was uploaded by Carlos O. Castillo-Araiza on 23 April 2016.

The user has requested enhancement of the downloaded file.

 Chemical Engineering Journal 280 (2015) 682–694

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Modeling of oxidative dehydrogenation of ethane to ethylene

on a MoVTeNbO/TiO2 catalyst in an industrial-scale packed
bed catalytic reactor
Gamaliel Che-Galicia, Richard S. Ruiz-Martínez, Felipe López-Isunza, Carlos O. Castillo-Araiza ⇑
Grupo de Procesos de Transporte y Reacción en Sistemas Multifásicos, Depto. de Ingeniería de Procesos e Hidráulica, Universidad Autónoma Metropolitana – Iztapalapa, Av. San
Rafael Atlixco No. 186, C.P. 09340 México D.F., Mexico

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 This work simulates the industrial

catalytic behavior of ODH-Et on
MoVTeNbO.
 Simulations are carried out using a 2D
pseudo-heterogeneous model.
 A kinetic model is developed and it
describes successfully laboratory
experiments.
 Industrial reactor model accounts for
the effect of hydrodynamics on heat
transfer.
 Wall-cooled packed-bed reactor is an
ideal design for the ODH-Et on
MoVTeNbO.

a r t i c l e i n f o a b s t r a c t

Article history: The oxidative dehydrogenation of ethane (ODH-Et) seems the most promising alternative to produce ethy-
Received 5 March 2015 lene compared to conventional processes. Nevertheless, there is not even a pilot plant for ODH-Et world-
Received in revised form 20 May 2015 wide nowadays. This work presents the simulation of the catalytic behavior of a highly active and selective
Accepted 22 May 2015
MoVTeNbO catalyst for the ODH-Et in a wall-cooled industrial-scale packed-bed reactor with a tube to
Available online 6 June 2015
particle diameter ratio equal to 3.12. The feasibility of using this complex yet necessary reactor design,
as well as the influence of operating conditions on conversion and yield along the reactor are analyzed.
Keywords:
The simulations are carried out using a two-dimensional pseudo-heterogeneous model, which makes
Oxidative dehydrogenation reaction
Low dt/dp packed bed reactor
use of both a reliable kinetic model and reliable transport parameters. Specifically, the kinetic model,
Hydrodynamics obtained from lab-scale experimental data, is coupled to the reactor model accounting for transport phe-
Ethylene nomena wherein the effect of hydrodynamics on heat transfer is assessed from independent experiments
Reaction kinetics in absence of reaction in an industrial-scale reactor. The developed kinetics successfully accounts for both
MoVTeNbO catalyst the inhibiting effect of adsorbed water on oxidative dehydrogenation and total oxidations and the effect of
the inlet partial pressure of oxygen and ethane on conversion and selectivity. Besides, the reactor model
elucidates the importance of accounting for the role of the hydrodynamics on the heat transport in order to
have reliable conversion and yield predictions. From the parametric sensitivity study, the temperature of
the coolant fluid and the inlet ethane concentration are variables for tuning the reactor performance.
Ó 2015 Elsevier B.V. All rights reserved.

Abbreviations: BIC, Bayesian information criterion; BLA, boundary layer approximation; CA, classical approach; dt/dp, tube-to-particle diameter ratio; ER, Eley–Rideal; GC,
gas chromatograph; HTP, heat transport parameters; LHHW, Langmuir–Hinshelwood–Hougen–Watson; MvK, Mars-van Krevelen; ODEs, ordinary differential equations;
ODH-Et, oxidative dehydrogenation of ethane; PL, power law.
⇑ Corresponding author. Tel.: +52 55 5804 4648.
E-mail addresses: gamaliel.che@gmail.com (G. Che-Galicia), coca@xanum.uam.mx (C.O. Castillo-Araiza).

http://dx.doi.org/10.1016/j.cej.2015.05.128
1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 683

Nomenclature

Roman letters wn objective function weight factor of each response

as external surface to particle volume ratio, m1 Wcat mass of catalyst, gcat
s
Ai0
natural logarithm of pre-exponential factor, mmol (gcat - Xn conversion of component n
h)1 Yn yield of component n
Cn molar concentration of component n in the gas phase, z axial position, mr
kmol m3f
Cn,ss molar concentration of component n in the gas phase at Greek letters
steady state, kmol m3 f a vector of parameters accounted for in the objective
Cno feed molar concentration of component n, kmol m3 f function
Cns molar concentration of component n in the solid phase, b vector of parameters accounted for in the objective
kmol m3s function
Cns,ss molar concentration of component n in the solid phase DHn standard enthalpy of adsorption for component n, kJ
at steady state, kmol m3s (mol K)1
Cpf heat capacity of the fluid, kJ (kgf K)1 DHi reaction enthalpy, kJ mol1
Cps heat capacity of the solid, kJ (kgs K)1 DSn Standard entropy of adsorption for component n used in
dp particle diameter, ms Eq. (17), kJ (mol K)1
dt reactor diameter, mr DSn Standard entropy of adsorption for component n, J (-
Deffr radial mass dispersion coefficient, m2r h1 mol K)1
Deffz axial mass dispersion coefficient, m2r h1 cC2 H6 ratio of oxygen and ethane partial pressure,
1
EA activation energy, kJ mol1 molO2 molC2 H6
Fn molar flow rate of the component n, mmol h1 cC2 H4 ratio of oxygen and ethene partial pressure,
1
hg interfacial heat transport coefficient, kJ m2
s (h K)1 molO2 molC2 H4
hw 2
wall heat transfer coefficient, kJ mr (h K) 1 e void fraction, m3f m3
r
ki reaction rate coefficient, mmol (gcat h)1 mi stoichiometric number
keffr radial effective thermal conductivity, kJ (mr h K)1 h fractional surface coverage
keffz axial effective thermal conductivity, kJ (mr h K)1 qb bed density, kgcat m3r
kg interfacial mass transfer coefficient, m3f m2
s h
1 qf fluid density, kgf m3
f
Kn adsorption equilibrium coefficient for component n, rj Horiuti number
Pa1
L reactor length, mr Subscripts
m partial reaction order for oxygen cat catalyst
pi partial pressure of component n, Pa eff effective
r radial position, mr exp experiment
ri specific reaction rate of reaction i, mmol (gcat h)1 f fluid
Rn net reaction rate of the component n, mmol (gcat h)1 n component n
R universal gas constant, kJ (mol K)1 o inlet, superficial
RSS objective function (residual sum of squares) p particle
S active sites for kinetic model mechanism r reactor
Sg standard entropy of gas phase molecule, J (mol K)1 s solid
t time, h tab tabulated
T temperature, K ⁄ vacant sites density
To feed temperature, K
Tb salt bath temperature, K Superscripts
Ts solid temperature, K ^ calculated
Ts,ss steady state solid temperature, K o inlet, standard
Tss steady state fluid temperature, K ⁄ reference
uo superficial velocity, m3f m2
r h
1

1. Introduction instance: it overcomes equilibrium limitations; permits lower tem-

perature operation (<500 °C) and, hence, energy saving; it
Ethylene is one of the principal building blocks in petrochemical decreases the number of side reactions; and, also, avoids catalyst
industry as it is used as intermediate for the production of many deactivation by coke formation due to the presence of molecular
valuable products, namely polyethylene, ethylene oxide, ethylene oxygen contained in the reaction mixture [4,5,7,8].
dichloride and ethylene benzene, among other important chemical Notwithstanding these advantages there are two industrial chal-
compounds [1,2]. Industry and academy have focused their lenges for producing ethylene via ODH-Et. One of them is related
research on finding competitive technologies to produce ethylene to the design of an active and selective catalyst to produce ethy-
[2–5] so as to face its increasing demand and also, as an effort of lene, while the other one has to do with the design of the reactor
reducing the environmental impact and energy consumption of technology. In fact today, to the best of our knowledge, there is
the conventional commercial processes for producing this olefin, not even a pilot plant for ODH-Et worldwide.
i.e., pyrolysis, fluid catalytic cracking and catalytic dehydrogena- A broad variety of catalytic formulations to produce ethylene out
tion [5,6]. The oxidative dehydrogenation of ethane (ODH-Et), in of ethane in presence of oxygen have been investigated [9–19].
this regard, seems to be the most promising alternative since it From these formulations, the multimetallic mixed oxide catalyst
offers many advantages compared to conventional processes, for based on Mo, V, Te and Nb is, undoubtedly, the most efficient and
684 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

effective material since it presents an outstanding activity and method is used. From XRD pattern using Rietveld refinement, the
selectivity to ethylene, apparently attributed to the presence of crystalline phase composition (wt%) is analyzed to be the M1 major
the M1 crystalline phase. Particularly, V species are the active sites crystalline phase present in the catalyst (92 wt% M1 and 8 wt% M2
in ethane activation, Mo species enhance the catalytic activity of V crystalline phase) [18]. A specific surface area of 7 m2 g1 cat is
atoms, Te species are associated with the formation of the active obtained from N2 physisorption in an AUTOSORB-I instrument.
and selective crystalline phase M1, and Nb species leads to enhance Details concerning the preparation as well as the physicochemical
the selectivity to ethylene [11,13,20–23]. From this end, envisaging characteristics of this catalyst are given elsewhere [18].
a possible future application of this material in ODH-Et to a com-
mercial scale requires, as a first step, of the conceptual design of 2.2. Experimental data
the catalytic reactor technology. Nevertheless, owing to the
exothermicity of the ODH-Et reaction, mainly related to the The data used in the kinetic modeling is obtained from experi-
unavoidably heat released during the formation of COX (CO and ments carried out in a laboratory isothermal fixed-bed reactor at
CO2) out of ethane and ethylene, the reactor design is as yet a com- atmospheric pressure. The fixed-bed reactor, made of quartz, has
plex chemical engineering assignment, as it is analyzed in the an internal diameter of 1.0  102 m and length of 4.0  102 m.
following. Experiments were performed using a bed consisting of 0.60 g of
The wall-cooled catalytic reactor with low tube-to-particle catalyst with average particle size of 150 lm. The reaction feed-
diameter ratio (dt/dp < 5) seems a suitable option to commercially stock was composed of a mixture of ethane (or ethylene), oxygen
achieve the ODH-Et over the MoVTeNbO formulation [24,25], since and nitrogen with a varying composition depending on the exper-
this configuration and catalyst design promote the fast removal of iment. The experimental conditions and conversion ranges are
the heat generated, maximize the desirable product yield, and listed in Table 1. The reactor effluent was analyzed online period-
avoids strong catalyst deactivation by a sintering process [26–34], ically by gas chromatography (GC). A detailed description of the
as observed for other well-known studied exothermic partial oxida- equipment has been reported in a previous study [57].
tion reactions. In fact, several researches [35–41] have considered
the wall-cooled packed bed catalytic reactor for simulating the pos- 2.3. Reaction rates
sible performance of the ODH-Et over catalytic systems different to
the MoVTeNbO formulation. Nevertheless, these studies are ques- The reaction network for ODH-Et over MoVTeNbO catalysts is
tionable since they do not make use of reliable kinetic models and proposed taking into account the experimental observations given
neglect the effect of the low dt/dp on the void fraction profile and, in [57]. Both parallel and consecutive reactions are considered, in
hence, on velocity field and its role on the heat transport [42–46]. which the species detected by GC (ethane, ethylene, carbon
The conceptual design of the industrial reactor for the ODH-Et monoxide and carbon dioxide) are only included. In this regard:
over a MoVTeNbO catalyst requires of both a reliable kinetic model Ethylene is produced from oxidative dehydrogenation of ethane
and reliable transport parameters. A few number of kinetic models (r1):
have been developed to describe the macroscopic mechanism of
ODH-Et, namely power law (PL), Langmuir–Hinshelwood–Houge
C2 H6 þ 0:5O2 ! C2 H4 þ H2 O ð1Þ
n–Watson (LHHW), Eley Rideal (ER), Mars-van Krevelen (MvK) or Ethane can also be oxidized to form CO2 (r2) and/or CO (r3):
a combination of MvK and LHHW (MvK–LHHW) [47–58], neverthe-
less, these models are specific for every catalyst formulation.
C2 H6 þ 3:5O2 ! 2CO2 þ 3H2 O ð2Þ
Specifically for the MoVTeNbO catalyst, the LHHW model, devel-
oped by our research group, is one of the most suitable one nowa-
C2 H6 þ 2:5O2 ! 2CO þ 3H2 O ð3Þ
days [57] since other kinetic models present a lack of statistical Ethylene can also react to form CO2 (r4) and/or CO (r5).
and phenomenological significance. Notwithstanding LHHW is a
potential kinetic model to overcome the conceptual design of the
C2 H4 þ 3O2 ! 2CO2 þ 2H2 O ð4Þ
industrial ODH-Et reactor using the MoVTeNbO catalytic system,
C2 H4 þ 2O2 ! 2CO þ 2H2 O ð5Þ
some aspects need to be accounted for, i.e., the relatively weak effect
of oxygen partial pressure on reaction rates, the high affinity of Table 2 shows the reaction mechanism considered for building
water to adsorb on active sites affecting negatively reaction rates the kinetic model. The mechanism accounts for some of the partic-
and the significative effect of inlet partial pressure of ethane on ular aspects of the MoVTeNbO formulation reported in the litera-
reaction rates. ture [57], namely a relatively weak effect of oxygen partial
The aim of this study is to model the industrial performance of a pressure on reaction rates, a high tendency of water to adsorb on
highly active/selective MoVTeNbO catalyst for the production of active sites and the significative effect of inlet partial pressure of
ethylene out of ethane via ODH-Et in a wall-cooled packed bed cat-
alytic reactor presenting a dt/dp equal to 3.12. This industrial-scale
reaction system is simulated using a two-dimensional Table 1
Range of experimental conditions during the steady state in the ODH-Et experiments.
pseudo-heterogeneous model coupled to a kinetic model devel-
oped herein for MoVTeNbO formulation and accounts, in some Operational condition Unit Range
detail, for the effect of hydrodynamics on heat transfer through pC2 H6 ;o kPa 5–22
the proper estimation of heat transport parameters from pilot pC2 H4 ;o kPa 5–22
plant experiments in absence of reaction. pO2 ;o kPa 5–22
cC2 H6 ¼ yO2 ;o y1
C2 H6 ;o
1
molO2 ;o molC2 H6 ;o 0.5–2.0
cC2 H4 ¼ yO2 ;o y1
C2 H4 ;o
1
molO2 ;o molC2 H4 ;o 0.5–2.0
2. Reaction kinetics W cat F 1
C2 H6 ;o
1
g cat h molC2 H6 ;o 10–140

W cat F 1 1
g cat h molC2 H4 ;o 10–130
C2 H4 ;o
2.1. Catalyst T °C 400–480
X C2 H 6 1
molC2 H6 molC2 H6 ;o 0.17–0.85
A MoVTeNbO catalyst with a nominal atomic ratio of X C2 H 4 1
molC2 H4 molC2 H4 ;o 0.03–0.14
Mo:V:Te:Nb equal to 1:0.24:0.24:0.18 prepared by precipitation
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 685

Table 2 represented by pn, and the fraction coverage of active sites is

Reaction steps and catalytic cycles considered for the kinetic model to describe the denoted by h . The fraction of free active sites is obtained by com-
ODH-Et.
bining Eqs. (11) to (13) resulting in Eq. (14):
Step Elementary reaction step rI rII rIII rIV rV
A O2 þ 2S ¢ 2OAS 1 7 5 3 2
1
1 C2 H6 þ OAS ! C2 H4 þ H2 OAS 2 0 0 0 0 h ¼ 1
ð14Þ
2 C2 H6 þ 7OAS ! 2CO2 þ 3H2 OAS þ 4S 0 2 0 0 0 1 þ ðK O2 pO2 Þ2 þ K H2 O pH2 O
3 C2 H6 þ 5OAS ! 2CO þ 3H2 O  S þ 2S 0 0 2 0 0
4 C2 H4 þ 6OAS ! 2CO2 þ 2H2 OAS þ 4S 0 0 0 1 0 The combination of specific reaction rates gives the net reaction
5 C2 H4 þ 4OAS ! 2CO þ 2H2 OAS þ 2S 0 0 0 0 1 rate of formation for the n-th component, which is expressed by
B H2 OAS ¢ H2 O þ S 2 6 6 2 2
Eq. (15):

X
5
ethane on conversion and selectivity. Thus, the model accounts for Rn ¼ mn;i ri ð15Þ
the following assumptions: (i) there is a single type of active sites i¼1
(S) on the catalyst surface, (ii) the only compounds adsorbed on the
where mn;i is the stoichiometric coefficient of the component n in the
active sites are oxygen and water, (iii) oxygen adsorption is disso-
i-th reaction, vide Eqs. (1)–(5). In order to reduce the correlation
ciative (step A) while water adsorption is associative (step B), (iv)
between activation energies and pre-exponential factors as well
surface reaction steps (steps 1–5) are considered to be fast taking
as standard adsorption enthalpies and entropies during the param-
place over a finite number of active sites, (v) since the construction
eters estimation procedure, both Arrhenius and van’t Hoff equations
of the model is based on the pseudo equilibrium approach, surface
are used in their reparametrized form.
reactions are the rate-determining steps and adsorption and des-
  
orption steps are quasi-equilibrated. In Table 2, rj is the so-called EA;i 1 1
ki ¼ exp A0i    ð16Þ
Horiuti number, that describes the times that every adsorption– R T T
desorption and/or reaction steps need to occur in order to com-
plete a single catalytic cycle comprising the overall reactions  o  
DSn DHon 1 1
(Eqs. (1)–(5)). Notice that the adsorption of molecular O2 (step A) K n ¼ exp    ð17Þ
R R T T
is a crucial stage for the surface reactions to take place (steps 1–5).
The equations that describe the rate of reactions for ODH-Et where A0i is the natural logarithm of the pre-exponential factor for
over MoVTeNbO are given through Eqs. (6)–(10), and are written the i-th reaction, EA,i is the activation energy factor for the i-th reac-
as a function of the partial pressures of the gas phase components tion, T is the reaction temperature, T* is the averaged reaction tem-
according to the reaction mechanism. The calculation of ri involves perature, DSon is the standard adsorption entropy of component n,
a rate coefficient denoted by ki, an adsorption equilibrium coeffi- DHon is the standard adsorption enthalpy of component n and R is
cient for the n-th component represented by Kn, the partial pres- the universal gas constant.
sure of the reactant n represented by pn, the fraction coverage of
oxygen denoted by hO and the reaction order related to the partial 2.3.1. Laboratory reactor model
pressure of the gas phase denoted by mi: The continuity equation for component n at the steady state
assumes that the flow pattern in the laboratory reactor is of plug
r 1 ¼ k1 pC2 H6 hO ð6Þ flow type and that there is no radial temperature or concentration
gradient as corroborated elsewhere [57]:
r 2 ¼ k2 pC2 H6 hm
O
2
ð7Þ dF n
¼ Rn ð18Þ
dW cat
r3 ¼ k3 pC2 H6 hm
O
3
ð8Þ
with the following initial conditions:

r 4 ¼ k4 pC2 H4 hm
O
4
ð9Þ F n ¼ F no ; for W cat ¼ 0 ð19Þ
where Fn is the molar flow rate of component n, Fno is the inlet
r 5 ¼ k5 pC2 H4 hm
O
5
ð10Þ molar flow rate of component n and Wcat is the mass of the catalyst
Since the fraction coverage of different species on the catalytic loaded into the reactor.
surface is conserved through time, the global balance of the frac-
tion sites corresponds to Eq. (11): 2.3.2. Parameter estimation
The development of a kinetic model must account for a proper
assessment of the model parameters from both physicochemical
h þ hO þ hH2 O ¼ 1 ð11Þ and statistical perspectives since these evaluations are essential
The fraction coverage of a specific component is defined as the for carrying out a reliable chemical reactor design. The kinetic
number of sites occupied by such a species relative to the total parameters are estimated by minimization of the following
number of available sites corresponding to Eq. (12) for oxygen weighted objective function:
and Eq. (13) for water: nresp X
X nexp
2 b1 ;b2 ;...;bn
RSSðbÞ ¼ w ðF k;n  F^k;n Þ ! min ð20Þ
n¼1 k¼1
1
hO ¼ ðK O2 pO2 Þ h2 ð12Þ
where b is an optimal parameter vector, nexp is the number of inde-
hH2 O ¼ K H2 O pH2 O h ð13Þ pendent experiments, nresp is the number of responses, Fk,n and F^k;n
are the n-th experimental and predicted responses for the k-th
where the adsorption equilibrium coefficient for the n-th compo- observation, respectively; and wn is the weight factor assigned to
nent is represented by Kn, the partial pressure of the reactant n is the n-th response.
686 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

The subroutine VODE is used to solve the corresponding set of Table 3

ODEs [59]. The initial minimization of the objective function, vide Reactor dimensions, catalyst dimensions, parameters and operational conditions used
in the reactor model.
Eq. (20), in the model regression is carried out using the
Rosenbrock method [60] and then, the ODRPACK subroutines are Value Refs.
called for fitting the corresponding experimental data points [61]. Reactor and catalyst dimensions
These subroutines can perform either weighted orthogonal distance L, mr 2.5 –
regression or nonlinear least square problems for explicit and impli- dt, mr 0.0256 [45]
dp, ms 0.0082 [45]
cit models using multi-response data with an implementation of the e, m3f m3 0.48 [68]
r
Levenberg–Marquardt method, as documented elsewhere [62]. qb, kgcat m3
r 75 –
Three criteria are used to evaluate the kinetic model. Firstly, the Operational conditions
agreement between the calculated and experimental molar flows P, atm 1 –
is compared using the so-called parity plot, thus, a careful assess- Tb, °C 400–480 –
ment of both individual and global statistical significance is carried To, °C 200 –
Rep 1400a –
out. The individual statistical significance of the parameters is
Flow rate, N m3 h1 4 [68]
assessed by a t-test while the global significance of the regression Ethane feed concentration, % mole 1–40 –
is assessed by the F-test. Aside, the parameter correlation between Oxygen feed concentration, % mole 7 –
pairs of estimated parameters is accounted for by computing the Transport parameters
so-called binary linear correlation coefficients (qij). When the value Deffr, m2r h1 32 [68,69]
of qij is close to ±1 a strong linear relationship between the esti- Deffz, m2r h1 53 [68,69]
mated parameters i and j occurs. kg, m3f m2 s h
1
576 [68,69]
hg, kJ m2s (h K)1 928.8 [68,69]
Finally, kinetic parameters have to be contrasted with a series of keffr, kJ (mr h K)1 8.35b [45]
thermodynamic criteria in order to determine whether they are keffr, kJ (mr h K)1 9.72c [45]
physically meaningful or not. In relation to the Arrhenius equation hw, kJ m2 r (h K)1 468b [45]
parameters, the activation energy should be lower to 210 kJ mol1 hw, kJ m2 r (h K)1 1051.2c [45]
[63] since larger values indicate the presence of catalyst deactiva- a
Calculated at standard conditions.
tion, for instance, by sintering [64]; whereas for the van’t Hoff equa- b
Estimated by classical approach.
c
tion parameters, estimated values of standard adsorption enthalpy Estimated by boundary layer approximation.
and entropy need to be confronted with a set of criteria stipulated
by Boudart [65], which are, often, overlooked in literature. catalytic reactor with low dt/dp. This methodology accounts for
Since adsorption is an exothermic process, the adsorption the role of void fraction and hydrodynamics on heat transport in
enthalpy has to satisfy the inequality: the reactor through the proper determination, using experiments
DHon > 0 ð21Þ in absence of reaction, of the effective heat transport parameters
(HTP) such as the radial and axial effective thermal conductivities,
the negative value of the standard entropy of adsorption must keffr and keffz, respectively, and the wall heat transfer coefficient, hw.
be higher than zero and lower than the corresponding standard Namely, these parameters are estimated using the boundary layer
entropy of the gas phase specie (Sn;g ), namely: approximation that makes use of a two-dimensional heat transport
model coupled to the Navier–Stokes–Darcy–Forchheimer equation
0 < DSn < Sn;g ð22Þ
and the existing void fraction profile [45].
where DSn is the standard entropy of adsorption. Table 3 presents the main variables used for all simulations dis-
The entropy in fact decreases when a gaseous molecule is trans- played in this study for the industrial-scale reactor modeling.
ferred from a three-dimensional phase, i.e, the gas phase, to a Reactor and catalyst dimensions are the same as those used to
two-dimensional phase, i.e, the catalyst surface. The gas phase obtain observations in absence of reaction from an industrial scale
standard entropy values of ethane, oxygen, ethylene, CO, CO2 and wall-cooled packed bed reactor [68]. These pure heat transport
water are computed at 440 °C amounting to 275, 231, 257, 223, experiments were used in a previous research [45] to estimate
246 and 218 J (mol K)1, respectively, based on reference [57]. the heat transport parameters (keffr, keffz, and hw), which are crucial
An additional criterion to be satisfied during the parameter esti- for modeling the studied reactor with low dt/dp. Besides, the calcu-
mation for the nonisothermal situation related to adsorption reac- lated active catalyst density (qb) is ca. 75 kg m3, this value corre-
tion, relates to the change in volume that occurs when a gaseous sponds to the active MoVTeNbO formulation coated over the
molecule is adsorbed over the surface of a solid. This is specifically external surface of TiO2 pellets packed in the reactor. The
expressed in the following form: MoVTeNbO active layer is considered to be 0.15 mm since mass
and heat transport limitations are, properly, minimized as
41:8 < DSon < 51:04  1:4DHon ð23Þ observed from kinetic experiments. The feedstock is a mixture of
ethane, oxygen and an inert with composition defined from kinetic
experiments. Thus, the feed inlet O2/C2H6 ratios are strategically
3. Industrial-scale reactor model selected to be out of the explosive region.
The wall-cooled packed bed reactor model is based on averaged
The performance of a multi-tubular wall-cooled packed bed cat- general conservation relations for mass and energy but considering
alytic reactor is, normally, predicted by the modeling of a single effective transport parameters. The two-dimensional pseudo-
tube, assuming, in a reliable manner, that all tubes behave almost heterogeneous model used is given as follows:
identically [66,67]. Due to this, this work simulates the perfor- Gas phase:
mance of the MoVTeNbO catalytic formulation during the
ODH-Et in a single tube industrial-scale packed bed reactor operat- !
ing in a non-isothermal and non-adiabatic mode. The reliability of @C n @C n @ 2 C n 1 @C n @2Cn
e þ uo ¼ eDeffr þ þ eDeffz
the simulations is based on the modeling approach proposed by @t @z @r 2 r @r @z2
Castillo-Araiza and López-Isunza [33,34,45], which allows the pre-
þ ð1  eÞkg as ðC ns  C n Þ ð24Þ
diction of the observed behavior of a wall-cooled packed bed
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 687

!
@T @T @ 2 T 1 @T @2T is carried out in order to support the main kinetic assumptions.
eqf C pf þ uo qf C pf ¼ keffr þ þ keffz On the other hand, Section 4.2 presents the industrial-scale cat-
@t @z @r 2 r @r @z2
alytic reactor simulations, which are aimed at elucidating the per-
þ ð1  eÞhg as ðT s  TÞ ð25Þ formance of MoVTeNbO catalyst during the ODH-Et in the
Solid phase: wall-cooled packed bed reactor. Firstly, the catalytic reactor is
modeled either considering those HTP that are estimated assuming
@C ns X5 plug flow, as treated nowadays in literature simulations [68,72–
ð1  eÞ ¼ ð1  eÞkg as ðC n  C ns Þ þ qb mni r i ð26Þ 74], or considering HTP that accounts for the effect of hydrody-
@t i¼1
namics on heat transport phenomena [33,34,45]. The comparison
of these simulations allows elucidate the need for considering
@T s X5
qb C ps ¼ ð1  eÞhg as ðT  T s Þ þ qb ðDHi Þri ð27Þ the role of velocity field on heat transport via the corresponding
@t i¼1 HTP when a packed bed reactor presenting a low dt/dp is modeled.
Secondly, a parametric sensibility analysis is presented with the
The corresponding initial and boundary conditions are:
aim to unfold the influence of coolant temperature, ethane and
t ¼ 0; C n ¼ C n;ss and C ns ¼ C ns;ss ð28Þ oxygen inlet concentrations on the catalytic performance of
MoVTeNbO catalyst. Finally, an analysis of the influence of the
T ¼ T ss and T s ¼ T s;ss ð29Þ adsorption of both water and oxygen on industrial reactor perfor-
mance is presented.
@C n
z ¼ 0; uo C no ¼ uo C n  eDeffz ð30Þ
@z 4.1. Kinetics

@T MoVTeNbO material is an attractive catalytic formulation since

uo qf C pf T o ¼ uo qf C pf T  keffz ð31Þ
@z it leads to ethane conversion and selectivity to ethylene relevant
for the industry [8]. A previous publication [57] was dedicated to
@C n @T
z ¼ L; ¼0 and ¼0 ð32Þ study the effect of feedstock composition, temperature and
@z @z space-time on a set of catalytic responses over the MoVTeNbO cat-
@C n @T alyst formulation. Fig. 1 summarizes main ODH-Et observations at
r ¼ 0; ¼ 0 and ¼0 ð33Þ different ethane conversions and selectivity to ethylene, CO and
@r @r
CO2. Specifically, the selectivity to the reaction products decreases
@C n @T in the following order: ethylene >> CO > CO2, ranging from 76% to
r ¼ Rt ; ¼ 0 and keffr ¼ hw ðT  T b Þ ð34Þ 96% for ethylene, 2.5–16.5% for CO and 1.5–7.5% for CO2. The region
@r @r
of high ethane conversion demands operating at large values of
where Cn is the molar concentration of component n in the gas both: temperature (480 °C) and space-time (70 gcat h mol1 ethane), a
phase, Deffr is the radial mass dispersion coefficient, Deffz is the axial scenario at which the selectivity to ethylene decreases and, hence,
mass dispersion coefficient, qb is the fixed-bed density (active cat- that of COx is favored. Besides, Fig. 1 also presents how the pro-
alyst density), qf is the fluid density, Cpf is the specific heat capacity posed kinetic model (Eqs. (6)–(11)) fits adequately the conversion
of the fluid, Cps is the specific heat capacity of the solid, keffr is the of ethane and the main reaction products in the ODH-Et.
radial effective thermal conductivity, keffz is the axial effective ther- The model adequacy is also evaluated by means of both statis-
mal conductivity and hw is the wall heat transfer coefficient. tical and phenomenological tests. Fig. 2 shows the parity plot for
The performance of the reactor is evaluated by the conversion the molar flow rates of all observed components (C2H6, C2H4, CO,
of reactants as well as by the yield of products. The conversion is CO2, O2 and H2O). Model regression presents statistical signifi-
defined as: cance, yielding adequate agreement with the experimental data
number of moles of n reacted over a wide range of experimental conditions, vide Table 1.
Xn ¼ n ¼ C2 H6 or O2 ð35Þ Namely, deviations from observations are random without pre-
number of moles of n fed
senting any undesired statistical trend. In this result, the random
while the yield of the n-th product is defined as:
error is constant, without the presence of a systematic association
number of moles of product n formed
Yn ¼
number of moles of ethane fed 100
n ¼ C2 H4 ; CO; CO2 ; and H2 O ð36Þ
The resulting model is given in terms of a set of parabolic partial 80
differential equations, which is solved numerically by the method
Selectivity [%]

of orthogonal collocation using 5 and 50 interior collocation points 60

C2H4 Kinetic model
at the radial and axial coordinates, respectively, employing shifted CO
CO2
Legendre polynomials [70]. The resulting set of ordinary differen-
tial equations is, then, solved by a Runge–Kutta method [71]. 40

4. Results and discussion 20

The results are divided in two sections. Section 4.1 discusses the
0
kinetics of ODH-Et on MoVTeNbO catalytic system. The model reli-
0 20 40 60 80 100
ability is firstly discussed, in terms of the statistical significance of
Ethane conversion [%]
both the regression and the parameters; thus, the parameter values
are analyzed to evaluate whether they are physically meaningful or Fig. 1. Selectivity to ethylene, CO2, and CO versus ethane conversion. Full lines
not. An analysis on the adsorption phenomena of both water and are calculated with the kinetic model (T = 400–480 °C; space-time = 23–70 gcat
oxygen on the active sites from MoVTeNbO catalytic formulation h mol1
ethane; nominal inlet molar ratio C2/O2/N2 = 9/7/84).
688 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

15 [79], in order to discard that model resulting in overfitting because

50 of the addition of unnecessary phenomena/parameters. From this
10
end, the developed kinetics is the most suitable one to describe
Fn calculated [mmol/h]

40 5
ODH-Et observations since it leads to a lower value of BIC (110.7)
than the LHHW model (159.6), obtaining an absolute difference
0 (DBIC) of ca. 49 [79]. Experimental data reported elsewhere [57]
30 0 5 10 15
elucidated how the inlet partial pressure of ethane gives rise to a
C2H6
O2
positive effect on ethane conversion due to its over oxidation to
20 carbon oxides rather than its adsorption on catalytic active sites.
C2H4
CO In this respect, the developed kinetics accounts for the effect of
10 CO2
H2O
inlet partial pressure of ethane on reaction rates; a situation that
is not possible when the LHHW model since it was insensitive to
0 the inlet partial pressure of ethane [57]. Therefore, statistical and
0 10 20 30 40 50 experimental evidences do indicate that the kinetic model devel-
Fn observed [mmol/h] oped in this study is the most suitable to be evaluated during
industrial-scale reactor simulations.
Fig. 2. Parity plot comparing experimental with calculated reactor outlet molar The estimated values for the adsorption equilibrium constants
flow rates for the kinetic model. The full lines are the first bisector and the dashed for oxygen and water, vide Table 4, indicate that standard
lines represent a deviation of 10%.
adsorption enthalpies are negative with water being the compo-
nent with the smallest standard adsorption enthalpy
(DHn = 128.2 kJ mol1). This suggests that water binds more
Table 4
Kinetic parameters values and corresponding 95% probability confidence intervals of strongly to MoVTeNbO catalyst formulation than oxygen
the kinetic model used to describe the ODH-Et over the MoVTeNbO catalyst. (DHn = 45.6 kJ mol1). Furthermore, Boudart0 s criteria are totally
satisfied since the values of standard adsorption entropies are
Parameter Estimated Lower Upper t-Value
value limit limit between 41.8 J (mol K)1 and the corresponding gas phase molec-
ular standard entropy. The magnitude of the estimated standard
A01 ; mmol (gcat h)1 5.50E+00 5.49E+00 5.52E+00 7.59E+02
adsorption enthalpies is similar to those values reported in litera-
A02 ; mmol (gcat h)1 6.86E-01 6.84E-01 6.88E-01 6.79E+02
A03 ; mmol (gcat h)1 1.58E+00 1.57E+00 1.58E+00 6.61E+02 ture [57,75–77]. For instance, the standard adsorption enthalpy of
A04 ; mmol (gcat h)1 2.60E+00 2.59E+00 2.61E+00 4.80E+02 oxygen is reported from 19 to 98 kJ mol1 [75] and the adsorp-
A05 ; mmol (gcat h)1 7.87E-01 7.84E-01 7.90E-01 5.14E+02 tion heat of water, measured on various metal oxides surfaces,
EA;1 ; kJ mol1 9.05E+01 9.02E+01 9.07E+01 7.90E+02 ranges between 36 and 113 kJ mol1 [76,77]. According to the
EA;2 ; kJ mol1 1.65E+02 1.64E+02 1.66E+02 5.43E+02 estimated activation energies, the rate of formation of ethylene
EA;3 ; kJ mol1 1.50E+02 1.49E+02 1.50E+02 7.91E+02 from ethane, vide Eq. (1), is the reaction that presents the lowest
EA;4 ; kJ mol1 1.39E+02 1.38E+02 1.39E+02 5.31E+02
activation energy i.e., 90.5 kJ mol1. In contrast, the formation of
EA;5 ; kJ mol1 1.32E+02 1.31E+02 1.32E+02 4.96E+02
DSoO2 ; J (mol K)1 2.15E+02 2.14E+02 2.16E+02 8.33E+02 COX from both ethane and ethylene, vide Eqs. (2)–(5), require acti-
DSoH2 O ; J (mol K)1 4.20E+01 4.19E+01 4.21E+01 6.75E+02 vation energies ranging from 131.8 kJ mol1 to 165.0 kJ mol1.
DHoO2 ; kJ mol1 4.56E+01 4.55E+01 4.57E+01 8.56E+02 Specifically, ethylene is the principal contributor to COX because
DHoH2 O ; kJ mol1 1.28E+02 1.28E+02 1.28E+02 7.48E+02 total oxidations from ethane (ca. 150 kJ mol1) require even more
m2 9.22E-01 9.19E-01 9.24E-01 7.37E+02 energy than the total oxidations from ethylene (ca. 130 kJ mol1).
m3 9.06E-01 9.03E-01 9.08E-01 7.23E+02 Our estimated values of activation energies, ranging from 90.5 to
m4 1.23E+00 1.22E+00 1.23E+00 5.79E+02 165.0 kJ mol1, are similar to the values reported in the literature
m5 9.05E-01 9.02E-01 9.08E-01 6.37E+02
for different catalytic systems for the ODH-Et (from 50 to
Note: F value = 7642, Ftab = 2.79, ttab = 1.97 at 1-a = 0.95 and 396° of freedom. 150 kJ mol1) [53,57,78]. The reaction orders associated to the par-
tial pressure of oxygen are found to be below 1.0, indicating that
reaction rates are weakly affected by changes in oxygen partial
between experimental points at any of the different operating con- pressure.
ditions studied.
Table 4 presents the F-value of the regression, and the esti-
mated parameters along with the 95% confidence interval and
t-value. The calculated F-value amounts to 7642, a value three 4.1.1. Fractional surface coverage
orders of magnitude larger than the tabulated one (2.79), corrobo- Adsorption phenomenon on MoVTeNbO formulation seems a
rating the satisfactory global significance of the model regression. key aspect to be analyzed during a proper kinetic study since reac-
Besides, all parameters are statistically significant since their con- tant and product adsorptions affect significantly reaction rates and,
fidence interval is significantly narrow and their t-value (ca. 240– hence, catalyst activity and selectivity. Fig. 3 displays the predicted
435) is higher than the t-value tabulated (1.97). All parameters surface coverage of adsorbed oxygen and adsorbed water as a func-
are not statistically correlated since their binary correlations coef- tion of temperature and space-time, which are calculated from Eqs.
ficients (qij) are always lower than 0.5. (12) and (13), respectively. The MoVTeNbO active sites are mostly
As commented earlier the LHHW model has been the most suit- covered with adsorbed water (ca. 0.95) rather than with adsorbed
able one nowadays for describing the ODH-Et on the MoVTeNbO oxygen (ca. 0.0). Thus, the present kinetics accounts properly for
catalytic formulation [57], since other kinetic models present a lack the negative effect of adsorbed water on reaction rates since the
of statistical significance. In this regard, a statistical analysis is car- high levels of water concentration and its affinity on active sites,
ried out to discriminate between this previously reported LHHW based on the highest water adsorption enthalpy, do indeed inhibit
model and the kinetic model developed in this study. oxidative dehydrogenation and total oxidations. Besides, the low
Notwithstanding the proposed kinetic model presents a slightly fraction coverage of oxygen on catalyst active sites and the low
better F-value (7642) than the LHHW model (7567), both models reaction orders associated to the partial pressure of oxygen, indi-
are evaluated through the Bayesian information criterion (BIC) cate that the kinetics accounts satisfactorily for the weak effect
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 689

1.00 1.0
(a) (b)

Fractional coverage [-]

Fractional coverage [-]

0.95 0.8

θ∗ 0.6 θ∗
0.90
θ Ο2 θΟ2
0.10 θΗ2Ο θΗ2Ο
0.4

0.05 0.2

0.00 0.0
400 420 440 460 480 0 10 20 30 40 50 60 70
Temperature [ºC] Wcat/Fethane,o [gcat h/molethane,o]

Fig. 3. (a) Fractional surface coverage versus temperature at inlet molar ratio C2H6/O2/N2 = 9/7/84 and space-time of 70 gcat h mol1
ethane,o; and (b) Fractional coverage versus
space-time at inlet molar ratio C2H6/O2/N2 = 9/7/84 and T = 440 °C.

of the inlet partial pressure of oxygen on ethane conversion and during the modeling of the ODH-Et in a wall-cooled catalytic reac-
selectivity to ethylene [57]. tor packed with MoVTeNbO catalyst externally deposited on a non-
The presence of oxygen is, as observed, crucial for the oxidative porous support of TiO2. Particularly, the boundary layer
dehydrogenation of ethane and total oxidations to occur. Although approximation is used to estimate keff and hw from heat transfer
catalytic studies [80,81] over V materials for other reactions sug- observations in absence of reaction, thus, these parameters con-
gest that bulk lattice oxygen participates in the reaction and it is taining the hydrodynamic information are used to model the
replaced by catalyst reoxidation with oxygen from the gas phase, industrial scale reactor by means of the classical model, vide Eqs.
for the current catalyst the superficial lattice oxygen seems to be (24)–(34). This approach uses a less expensive numerical scheme
the main species participating on several reactions [14,57,81–84], since it only contains heat and mass balances instead of momen-
which explains the high selectivity to ethylene. The activity of V tum, heat and mass balances which computationally speaking is
catalysts mainly depends on the oxygen’s lability, nevertheless this very time-consuming.
lability is a function of catalyst chemical composition and crys- In order to be judicious on the modeling approach applied to
talline structure [14,57,82,83]. In this respect, MoVTeNbO formula- elucidate the role of MoVTeNbO externally deposited on TiO2 for
tion presents a relatively large capacity for oxygen recombination the ODH-Et in a wall-cooled catalytic reactor with low dt/dp, we
in its active M1 phase [14,57,82,83], i.e. the oxygen species are present the simulations of an industrial-scale reactor for the cat-
immediately used for oxidation reactions from the first lattice lay- alytic o-xylene partial oxidation using the aforementioned
ers and are overtaken from the gas phase. This might be the reason approach. The o-xylene partial oxidation is selected since observa-
why the proposed mechanism, vide Table 2, adequately represents tions at industrial-scale with and without reaction are available
the observed kinetic phenomenon for MoVTeNbO rather the from studies carried out by the research group [33,34,68].
well-known Mars and van-Krevelen redox mechanism [85]. Kubo Moreover, this reaction system presents the same geometric (L
et al., [84], during the evaluation of propane ammoxidation with and dt/dp) and operational conditions (Rep) that are used for the
lattice oxygen from Mo–V–O-based complex metal oxide catalysts, ODH-Et reaction in this work, vide Table 3. From this end, Fig. 4
observed that the migration rate of the oxygen from the bulk to displays temperature profiles for two oxidation reactions, let0 s
superficial lattice is too slow in comparison with the consumption say, the o-xylene partial oxidation used herein as a reference
rate of oxygen from the surface lattice. In this respect, a relatively (Fig. 4a), and the ODH-Et (Fig. 4b). Both reaction cases compare
small amount of oxygen is demanded to reoxidize the metallic temperature profiles along the reactor using distinct effective
sites that are reduced after the propane oxidation reactions, thus, HTP estimated from the same set of experiments in absence of
explaining the high selectivity to ethylene of this type of mixed reaction: (i) parameters estimated via the classical approach (CA)
oxide catalysts. Namely, a low amount of oxygen on surface lattice assuming an uniform velocity profile with constant void fraction
along with a low reaction temperature give rise to ethylene forma- [68,72–74] and (ii) parameter estimated via the boundary layer
tion out of ethane rather than total oxidations. approximation (BLA) in which the role of void fraction profile
and velocity field are assessed [33,34,45]. Table 3 presents the
4.2. Industrial-scale reactor predictions numerical values of hw and keff estimated via the BLA and CA from
the same experimental dataset. These HTP are larger for BLA,
The adequate consideration of heat transport phenomena in the reflecting differences in the estimated rates of radial heat transport
model of a wall-cooled packed bed catalytic reactor with low dt/dp through the core of the packed bed and close to the wall. To this
is, undoubtedly, one of the most essential aspects to be accounted end, the predictions of measured temperature profiles in Fig. 4a
for during the conceptual design of this type of reaction systems. for the o-xylene partial oxidation case elucidate the suitability of
There are several approximations that have been used to model the HTP values from the BLA, giving better predictions than when
heat transport in a wall-cooled catalytic reactor with low dt/dp. HTP values using the CA approach are employed in the reactor
Some studies [33,34,45] suggest that the inclusion of velocity pro- model, vide, i.e. Eqs. (24)–(34). Therefore, hw and keff are effective
files improves heat transport description; however, others studies parameters that through BLA lump differently the hydrodynamic
[68,72–74] have neglected the influence of the hydrodynamics in and heat-transport phenomena in the packed bed leading to better
the modeling of this type of reactors. In this regard, a practical predictions. Fig. 4b presents temperature profiles for the ODH-Et.
but reliable approach, developed by our research group, is herein As expected, when the reactor model makes use of HTP estimated
applied to account for the role of hydrodynamics on heat transport considering plug flow, there is the presence of a higher hot spot
690 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

460
550 (a) (b)
500 450
Temperature [ºC]

Temperature [ºC]
450
440

400

Experimental observation
430
350 HTP estimated with BLA
HTP estimated with BLA
HTP estimated with CA
HTP estimated with CA
300 420
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m] Reactor length [m]

Fig. 4. Axial temperature profiles in the reactor using different HTP: (a) Comparison between prediction and observation for the partial oxidation of o-xylene on a V2O5/TiO2
catalyst at: Tb = 376 °C, inlet molar ratio o-xylene/air = 1/99 and To = 190 °C [33]; and (b) Predictions for the oxidative dehydrogenation of ethane on MoVTeNbO/TiO2 catalyst
at: Tb = 440 °C, inlet molar ratio C2H6/O2/N2 = 9/7/84 and To = 200 °C.

560 1.0

(a)
1- Conversion and Yield
(b)
520 0.9
Temperature [ºC]

Tb=480 ºC
480
0.8 - - - - Initial steady state
Tb=440 ºC 0.2 - - - - Transient state
440

Tb=400 ºC 0.1
400
Initial steady state Second steady state
Final steady state Transient state
360 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m] Reactor length [m]

1.0 1.0
- - - - Final steady state
0.9 (c) (d) - - - - Transient state
1- Conversion and Yield

1- Conversion and Yield

0.8
0.8

0.7 0.6
- - - - Second steady state
- - - - Transient state
0.3 0.4
0.2
0.2
0.1

0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m] Reactor length [m]

Fig. 5. Reactor model predictions when Tb is increased 40 °C in a ramp change with an inlet molar ratio C2H6/O2/N2 = 9/7/84: (a) Temperature profiles; (b) Conversion and
yield profiles at Tb = 400 °C; (c) Conversion and yield profiles at Tb = 440 °C; (d) Conversion and yield profiles at Tb = 480 °C (For (b)–(d): (d) C2H6 conversion; (h) O2
conversion; (s) C2H4 yield; (j) H2O yield; (N) CO yield and (4) CO2 yield).

since the role of hydrodynamics on heat transfer mechanisms is parametric sensitivity study for the performance of MoVTeNbO
neglected. Owing to the above, the following sections present the system during the ODH-Et in an industrial scale wall-cooled
simulations of the industrial-scale reactor for ODH-Et considering packed bed reactor, this section is aimed at simulating the reactor
those HTP estimated from BLA. when perturbations of cooling bath temperature and inlet ethane
concentration are considered. The first case, illustrated in Fig. 5,
4.2.1. Parametric sensitivity study shows the predicted dynamic response when reactor starts up at
The behavior of a reactor is a function of the initial process con- Tb of 400 °C with an inlet molar ratio C2H6/O2/N2 = 9/7/84, once
ditions, reactor disturbances and the trajectories of the manipu- the steady state is reached, it is followed by perturbations of the
lated reactor variables. In order to analyze an open-loop cooling bath temperature, namely 440 °C and 480 °C. The second
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 691

480 1.0
Initial steady state
(a) Second steady state
Third steady state
(b)
470

1- Conversion and Yield

Final steady state 0.8

Temperature [ºC]
Transient state
460
40 mole % of ethane 0.6
18 mole % of ethane - - - - Initial steady state
450 - - - - Transient state
9 mole % of ethane
0.4
440
1 mole % of ethane
0.2
430

420 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m] Reactor length [m]

1.0 1.0
(c) (d)
1- Conversion and Yield

1- Conversion and Yield

0.8 0.8

0.6 0.6
- - - - Second steady state - - - - Third steady state
- - - - Transient state - - - - Transient state
0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m] Reactor length [m]

1.0
(e)
1- Conversion and Yield

0.8

0.6
- - - - Final steady state
- - - - Transient state
0.4

0.2

0.0
0.0 0.5 1.0 1.5 2.0 2.5
Reactor length [m]

Fig. 6. Reactor model predictions when ethane feed concentration is increased at different% mole in a ramp change with a Tb = 440 °C: (a) Temperature profiles; (b)
Conversion and yield profiles at 1% mole of ethane; (c) Conversion and yield profiles at 9% mole of ethane; (d) Conversion and yield profiles at 18% mole of ethane; (e)
Conversion and yield profiles at 40% mole of ethane (For (b)–(e): (d) C2H6 conversion; (h) O2 conversion; (s) C2H4 yield; (j) H2O yield; (N) CO yield and (4) CO2 yield).

case, illustrated in Fig. 6, displays the predicted dynamic response 400 °C or 440 °C leads to mild hot spots of ca. 1 °C or 7 °C respec-
when reactor starts up at Tb of 440 °C with an inlet molar ratio of tively whereas a Tb equal to 480 °C provokes a pronounced hot spot
C2H6/O2/N2 = 1/7/92, once the steady state is reached, it is followed with a temperature rise of ca. 56 °C. In the latter, the rate of heat
by perturbations of the inlet molar ratio of ethane, specifically removal is minor in comparison with the rate of heat generated.
C2H6/O2/N2 = 9/7/84, C2H6/O2/N2 = 18/7/75 and Besides Figs. 5b–d display the effect of coolant temperature on
C2H6/O2/N2 = 40/7/53. the overall consumption rates of the reactants. The increment of
Fig. 5 displays transient and steady state temperature, conver- Tb favors both ethane and oxygen conversions. Analyzing the pro-
sion (ethane and oxygen) and yield (ethylene, COX and water) pro- duct distribution, an increase in Tb brings out a positive effect on
files along the reactor length at different coolant temperatures the yield of ethylene out of ethane. However, a higher Tb leads to
maintaining a fixed inlet molar ratio C2H6/O2/N2 = 9/7/84. So far, higher carbon oxides yields, which are the responsible of both
an increment in coolant temperature causes an increase in the the formation of hot spots and the decrease in the selectivity to
magnitude of the hot spot, vide Fig. 5a. Specifically a Tb equal to ethylene. In this regard, a lower Tb is recommended for a
692 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

1.0 coverage profiles. The relatively low fraction coverage of oxygen

determines the product distribution along the reactor, revealing
the selective role of the MoVTeNbO catalyst to produce ethylene
Fractional coverage [-]

0.8
out of ethane. Aside, the high fraction coverage of water leads to
a negative effect on both type of reactions, those selective to pro-
0.6 θΗ2Ο Tb=400 ºC duce ethylene and those selective to obtain COx (vide Fig. 5).
θ∗ Tb=440 ºC Therefore, water inhibits the oxidative dehydrogenation of ethane
θΟ2 Tb=480 ºC and total oxidations through its adsorption on MoVTeNbO sites.
0.4
Thus, water seems to play an important role on this type of cat-
alytic reaction, as documented for other catalytic partial oxidations
0.2 of different hydrocarbons [87].

0.0
0.0 0.5 1.0 1.5 2.0 2.5 5. Conclusions
Reactor length [m]
The implementation of a new catalytic material at industrial
Fig. 7. Fractional surface coverage along the reactor at differ bath temperatures level requires the knowledge from the conceptual design of the
with an inlet molar ratio C2H6/O2/N2 = 9/7/84. reactor technology. Modeling has been the most reliable engineer-
ing tool to carry out the preliminary reactor design. In this regard,
this work shows through modeling the catalytic performance of
satisfactory ethane conversion along with ethylene selectivity MoVTeNbO catalyst for ODH-Et in a wall-cooled industrial-scale
avoiding catalyst damage due to high temperatures [23]. The catalytic packed bed reactor presenting a low dt/dp ratio. To have
behavior of the industrial scale reactor presented in Fig. 5 shows a confidence on model simulations, a proper determination of both
similar tendencies as those of the typical industrial-scale packed kinetic parameters and heat transport parameters from indepen-
bed reactors used to perform highly exothermic reactions dent experiments should be carefully carried out. In terms of the
[24,30,33,34,67,68]. kinetics, an adequate agreement between model and experimental
In the design of this type of complex reactors it is necessary to values is observed, wherein the obtained parameters are physically
operate without the presence of drastic hot spots in order to meaningful and are found to be statistically significant. The pro-
decrease the irreversible deactivation of the catalysts by a sintering posed kinetics addresses several aspects observed experimentally,
phenomenon [86]. Particularly, the stability of the MoVTeNbO cat- namely it accounts for both the negative effect of adsorbed water
alyst is restricted to operate below 500 °C, due to energy saving on reaction rates and the weak effect of the inlet partial pressure
but, mainly, because operations above 500 °C leads to the irre- of oxygen on ethane conversion and selectivity to ethylene. In this
versible loss of catalytic activity as a consequence of the removal respect, even the presence of oxygen is crucial for the oxidative
of tellurium from the MoVTeNbO catalytic formulation [23], i.e. dehydrogenation of ethane and total oxidations to occur, the
specifically, the lost of Te from MoVTeNbO material occurs prefer- superficial lattice oxygen from the MoVTeNbO catalyst seems to
entially from the end section of the M1 crystal phase, across the be the main species participating on reactions, explaining, thus,
[0 0 1] plane. The removal of Te species modifies the crystal phase its high selectivity to ethylene.
composition of the M1 phase causing the partial destruction of In terms of the industrial-scale reactor, the simulations suggest
the MoVTeNbO with the simultaneous formation of the MoO2 that hydrodynamics is an essential aspect to be considered during
phase and, in consequence, the activity and selectivity to ethylene the modeling of this type of reaction systems presenting a low dt/-
are affected as elucidated elsewhere [23]. dp. The coolant temperature is the main variable affecting ethane
Fig. 6 compares transient and steady state temperature, conver- conversions and yield to ethylene. For instance, coolant tempera-
sion (ethane and oxygen) and yield (ethylene, COX and water) pro- tures larger than 440 °C lead to hot spot magnitudes that could
files along the reactor length at different ethane inlet cause the damage to the MoVTeNbO structure. As observed from
concentrations at Tb equals to 440 °C. The position of the hot spot laboratory experiments, water formation affects conversion and
is located toward the first meter of reactor length, a zone with selectivity, and catalyst is weakly oxidized along the reactor
higher concentration of ethane. An increment in the inlet ethane length, giving insights on the highly selectivity of the MoVTeNbO
concentration causes a slight increment of the magnitude of the catalytic formulations.
hot spot, vide Fig. 6a. Particularly, 1% or 9% of ethane in the feed- On the basis of the results obtained herein, the wall-cooled
stock leads to mild hot spots of ca. 1 °C or 7 °C, respectively, packed-bed reactor represents a promising alternative for petro-
whereas 18% or 40% of ethane causes hot spots with a temperature chemical industry to carry out the ODH-Et over a MoVTeNbO cat-
rise of ca. 12 °C or 34 °C, respectively. Figs. 6b–e show the effect of alyst, which is highly selective to ethylene. Besides, the model
inlet ethane concentration on ethane conversion and product developed in this work, in general, and the modeling strategy
yields. An increase in ethane inlet concentration causes a rapid oxy- based on laboratory and pilot plant experiments in absence of reac-
dehydrogenation (r1), but mainly of the total oxidations (r2–r5) tion, in particular, are potential tools to be used in future studies
reactions, thus, provoking large amounts of heat generation. for the conceptual design and scale-up of industrial reactor system
Therefore, the increase in ethane inlet concentration favors reactant for the ODH-Et on MoVTeNbO material.
conversions and product yields rather than ethylene selectivity
(vide Figs. 6b–e). Acknowledgments
Finally, a brief analysis of the influence of the adsorption of both
water and oxygen on catalytic reactor performance is presented. As This study was supported by El Consejo Nacional de Ciencia y
observed in Fig. 7, the catalyst surface is mainly occupied by water Tecnología (CONACYT) under Project No. 181104. Gamaliel
and vacancy sites, since the surface coverage of oxygen is found to Che-Galicia also thanks CONACYT for providing a postgraduate fel-
be negligibly small along the reactor length, as elucidated from lab- lowship. We also acknowledge to Eng. J.C. Castillo-Rodriguez, Eng.
oratory experiments in Section 4.1.1. The reactions occurring on C. Tzompantzi-Flores and Eng. O. Romero-Pérez for the technical
the MoVTeNbO catalytic formulation are a function of the surface support and discussion on the developed kinetic model. Authors
 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694 693

also thank valuable comments from the anonymous reviewers [29] Y.S. Cheng, F. López-Isunza, T. Mongkhonsi, L.S. Kershenbaum, Estimation of
catalyst activity profiles in fixed-bed reactors with decaying catalysts, Appl.
whose careful revisions and critics decisively contributed in
Catal. A 106 (1993) 193–199.
improving the quality of this paper. [30] J.N. Papageorgiou, G.F. Froment, Simulation models accounting for radial
voidage profiles in fixed-bed reactors, Chem. Eng. Sci. 50 (1995) 3043–3056.
[31] O. Bey, G. Eigenberger, Gas flow and heat transfer through catalyst filled tubes,
References Int. J. Therm. Sci. 40 (2001) 152–164.
[32] D. Wen, Y. Ding, Heat transfer of gas flow through a packed bed, Chem. Eng.
Sci. 61 (2006) 3532–3542.
[1] A. Clements, M. Dunn, V. Firth, L. Hubbard, J. Lazonby, D. Waddington, The
[33] C.O. Castillo-Araiza, F. López-Isunza, Modeling the partial oxidation of o-
Essential Chemical Industry, Chemical Industry Education Centre, York, 2010.
xylene in an industrial packed-bed catalytic reactor: the role of
[2] M.A. Bañares, Supported metal oxide and other catalysts for ethane
hydrodynamics and catalyst activity in the heat transport, Ind. Eng. Chem.
conversion: a review, Catal. Today 51 (1999) 319–348.
Res. 49 (2010) 6845–6853.
[3] S. Albonetti, F. Cavani, F. Trifirò, Key aspects of catalyst design for the selective
[34] C.O. Castillo-Araiza, F. López-Isunza, The role of catalyst activity on the steady
oxidation of paraffins, Catal. Rev. 38 (1996) 413–438.
state and transient behavior of an industrial-scale fixed bed catalytic reactor
[4] F. Cavani, F. Trifirò, The oxidative dehydrogenation of ethane and propane as
for the partial oxidation of o-xylene on V2O5/TiO2, Chem. Eng. J. 176–177
an alternative way for the production of light olefins, Catal. Today 24 (1995)
(2011) 26–32.
307–313.
[35] M. Fattahi, M. Kazemeini, F. Khorasheh, A. Darvishi, A.M. Rashidi, Fixed-bed
[5] C.A. Gärtner, A.C. van Veen, J.A. Lercher, Oxidative dehydrogenation of ethane:
multi-tubular reactors for oxidative dehydrogenation in ethylene process,
common principles and mechanistic aspects, ChemCatChem 5 (2013) 3196–
Chem. Eng. Technol. 36 (2013) 1692–1700.
3217.
[36] Z. Skoufa, E. Heracleous, A.A. Lemonidou, Investigation of engineering aspects
[6] K. Weissermel, H.-J. Arpe, Industrial Organic Chemistry, WILEY-VCH Verlag
in ethane ODH over highly selective Ni0.85Nb0.15Ox catalyst, Chem. Eng. Sci. 84
GmbH & Co. KGaA, Weinheim, 2003.
(2012) 48–56.
[7] T. Blasco, J.M. López, Nieto, oxidative dehydrogenation of short chain alkanes
[37] M.L. Rodriguez, D.E. Ardissone, E. Heracleous, A.A. Lemonidou, E. López, M.N.
on supported vanadium oxide catalysts, Appl. Catal. A 157 (1997) 117–142.
Pedernera, D.O. Borio, Oxidative dehydrogenation of ethane to ethylene in a
[8] F. Cavani, N. Ballarini, A. Cericola, Oxidative dehydrogenation of ethane and
membrane reactor: a theoretical study, Catal. Today 157 (2010) 303–309.
propane: how far from commercial implementation?, Catal Today 127 (2007)
[38] M.L. Rodriguez, D.E. Ardissone, A.A. Lemonidou, E. Heracleous, E. López, M.N.
113–131.
Pedernera, D.O. Borio, Simulation of a membrane reactor for the catalytic
[9] E.M. Thorsteinson, T.P. Wilson, F.G. Young, P.H. Kasai, The oxidative
oxidehydrogenation of ethane, Ind. Eng. Chem. Res. 48 (2009) 1090–1095.
dehydrogenation of ethane over catalysts containing mixed oxides of
[39] C. Hamel, Á. Tóta, F. Klose, E. Tsotsas, A. Seidel-Morgenstern, Analysis of single
molybdenum and vanadium, J. Catal. 52 (1978) 116–132.
and multi-stage membrane reactors for the oxidation of short-chain alkanes-
[10] E. Morales, J.H. Lunsford, Oxidative dehydrogenation of ethane over a lithium-
Simulation study and pilot scale experiments, Chem. Eng. Res. Des. 86 (2008)
promoted magnesium oxide catalyst, J. Catal. 118 (1989) 255–265.
753–764.
[11] J.M. López Nieto, P. Botella, M.I. Vázquez, A. Dejoz, The selective oxidative
[40] E. López, E. Heracleous, A.A. Lemonidou, D.O. Borio, Study of a multitubular
dehydrogenation of ethane over hydrothermally synthesized MoVTeNb
fixed-bed reactor for ethylene production via ethane oxidative
catalysts, Chem. Commun. (2002) 1906–1907.
dehydrogenation, Chem. Eng. J. 145 (2008) 308–315.
[12] M. Panizza, C. Resini, F. Raccoli, G. Busca, R. Catani, S. Rossini, Oxidation of
[41] F.A. Al-Sherehy, A.M. Adris, M.A. Soliman, R. Hughes, Avoidance of
ethane over vanadia-alumina-based catalysts: co-feed and redox experiments,
flammability and temperature runaway during oxidative dehydrogenation
Chem. Eng. J. 93 (2003) 181–189.
using a distributed feed, Chem. Eng. Sci. 53 (1998) 3965–3976.
[13] P. Botella, E. García-González, A. Dejoz, J.M. López Nieto, M.I. Vázquez, J.
[42] R.B. Benenati, C.F. Brosilow, Void fraction distribution in beds of spheres, J. Am.
González-García, Selective oxidative dehydrogenation of ethane on
Inst. Chem. Eng. 8 (1962) 359–361.
MoVTeNbO mixed metal oxide catalysts, J. Catal. 225 (2004) 428–438.
[43] E. Tsotsas, E.U. Schlunder, Some remarks on channelling and on radial
[14] P. Botella, E. García-González, J.M. López Nieto, J.M. González-Calbet,
dispersion in packed beds, Chem. Eng. Sci. 43 (1988) 1200–1203.
MoVTeNbO multifunctional catalysts: Correlation between constituent
[44] M. Winterberg, E. Tsotsas, A. Krischke, D. Vortmeyer, A simple and coherent set
crystalline phases and catalytic performance, Solid State Sci. 7 (2005) 507–
of coeficients for modelling of heat and mass transport with and without
519.
chemical reaction in tubes filled with spheres, Chem. Eng. Sci. 55 (2000) 967–
[15] E. Heracleous, A.A. Lemonidou, Ni–Nb–O mixed oxides as highly active and
979.
selective catalysts for ethene production via ethane oxidative
[45] C.O. Castillo-Araiza, H. Jiménez-Islas, F. López-Isunza, Heat-transfer studies in
dehydrogenation. Part I: Characterization and catalytic, J. Catal. 237 (2006)
packed-bed catalytic reactors of low tube/particle diameter ratio, Ind. Eng.
162–174.
Chem. Res. 46 (2007) 7426–7435.
[16] B. Solsona, M.I. Vázquez, F. Ivars, A. Dejoz, P. Concepción, J.M. López, Nieto,
[46] C.O. Castillo-Araiza, F. Lopez-Isunza, Hydrodynamic models for packed beds
Selective oxidation of propane and ethane on diluted Mo-V-Nb-Te mixed-
with low tube-to-particle diameter ratio, Int. J. Chem. Reactor Eng. 6 (2008) A1.
oxide catalysts, J. Catal. 252 (2007) 271–280.
[47] S.T. Oyama, A.A. Middlebrook, G.A. Somorjai, Kinetics of ethane oxidation on
[17] X. Lin, C.A. Hoel, W.M.H. Sachtler, K.R. Poeppelmeier, E. Weitz, Oxidative
vanadium oxide, J. Phys. Chem. 94 (1990) 5029–5033.
dehydrogenation (ODH) of ethane with O2 as oxidant on selected transition
[48] J. Le Bars, J.C. Vedrine, A. Auroux, Role of surface acidity on vanadia/silica
metal-loaded zeolites, J. Catal. 265 (2009) 54–62.
catalysts used in the oxidative dehydrogenation of ethane, Appl. Catal. A 88
[18] J.S. Valente, R. Quintana-Solórzano, H. Armendáriz-Herrera, G. Barragán-
(1992) 179–195.
Rodríguez, J.M. López-Nieto, Kinetic study of oxidative dehydrogenation of
[49] C.-Y. Kao, K.-T. Huang, B.-Z. Wan, Ethane oxydehydrogenation over supported
ethane over MoVTeNb mixed-oxide catalyst, Ind. Eng. Chem. Res. 53 (2014)
vanadium oxides, Ind. Eng. Chem. Res. 33 (1994) 2066–2072.
1775–1786.
[50] M.D. Argyle, K. Chen, At.T. Bell, E. Iglesia, Ethane oxidative dehydrogenation
[19] J. Santander, E. López, A. Diez, M. Dennehy, M. Pedernera, G. Tonetto, Ni–Nb
pathways on vanadium oxide catalysts, J. Phys. Chem. B 106 (2002) 5421–
mixed oxides: one-pot synthesis and catalytic activity for oxidative
5427.
dehydrogenation of ethane, Chem. Eng. J. 255 (2014) 185–194.
[51] D. Linke, D. Wolf, M. Baerns, S. Zeyß, U. Dingerdissen, Catalytic partial
[20] J.M. López Nieto, P. Botella, B. Solsona, J.M. Oliver, The selective oxidation of
oxidation of ethane to acetic acid over Mo1V0.25Nb0.12Pd0.0005Ox: II. Kinetic
propane on Mo-V-Te-Nb-O catalysts: the influence of Te-precursor, Catal.
modelling, J. Catal. 205 (2002) 32–43.
Today 81 (2003) 87–94.
[52] F. Klose, M. Joshi, C. Hamel, A. Seidel-Morgenstern, Selective oxidation of
[21] P. Concepción, P. Botella, J.M. López Nieto, Catalytic and FT-IR study on the
ethane over a VOx/c-Al2O3 catalyst – investigation of the reaction network,
reaction pathway for oxidation of propane and propylene on V- or Mo-V based
Appl. Catal. A 260 (2004) 101–110.
catalysts, Appl. Catal. A 278 (2004) 45–56.
[53] R. Grabowski, J. Słoczyński, Kinetics of oxidative dehydrogenation of propane
[22] B. Solsona, F. Ivars, P. Concepción, J.M. López Nieto, Selective oxidation of n-
and ethane on VOx/SiO2 pure and with potassium additive, Chem. Eng. Process.
butane over MoV-containing oxidic bronze catalysts, J. Catal. 250 (2007) 128–
44 (2005) 1082–1093.
138.
[54] E. Heracleous, A.A. Lemonidou, Ni–Nb–O mixed oxides as highly active and
[23] J.S. Valente, H. Armendáriz-Herrera, R. Quintana-Solórzano, P. del Ángel, N.
selective catalysts for ethene production via ethane oxidative
Nava, A. Massó, J.M. López Nieto, Chemical, structural, and morphological
dehydrogenation. Part II: Mechanistic aspects and kinetic modeling, J. Catal.
changes of a MoVTeNb catalyst during oxidative dehydrogenation of ethane,
237 (2006) 175–189.
ACS Catal. 4 (2014) 1292–1301.
[55] F. Rahman, K.F. Loughlin, M.A. Al-Saleh, M.R. Saeed, N.M. Tukur, M.M. Hossain,
[24] G.F. Froment, K.B. Bischoff, Chemical reactor analysis and design, John Wiley &
K. Karim, A. Mamedov, Kinetics and mechanism of partial oxidation of ethane
Sons, New York, 1979.
to ethylene and acetic acid over MoV type catalysts, Appl. Catal. A 375 (2010)
[25] P. Arpentinier, F. Cavani, F. Trifiro, The Technology of Catalytic Oxidations,
17–25.
Technip, Paris, 2001.
[56] S.A. Al-Ghamdi, M.M. Hossain, H.I. de Lasa, Kinetic modeling of ethane
[26] J. Lerou, G.F. Froment, Velocity, temperature and conversion profiles in fixed
oxidative dehydrogenation over VOx/Al2O3 catalyst in a fluidized-bed riser
bed catalytic reactor, Chem. Eng. Sci. 32 (1977) 853–861.
simulator, Ind. Eng. Chem. Res. 52 (2013) 5235–5244.
[27] T. Daszkowski, G. Eigenberger, A reevaluation of fluid flow heat transfer and
[57] G. Che-Galicia, R. Quintana-Solórzano, J.S. Valente, R.S. Ruiz-Martínez, C.O.
chemical reaction in catalyst filled tubes, Chem. Eng. Sci. 47 (1992) 2245–
Castillo-Araiza, Kinetic modeling of the oxidative dehydrogenation of ethane
2250.
to ethylene over a MoVTeNbO catalytic system, Chem. Eng. J. 252 (2014)
[28] V. Blasco, C. Royo, A. Monzbn, J. Santamaria, Catalyst sintering in fixed-bed
75–88.
reactors: deactivation rate and thermal history, AIChE J. 38 (1992) 237–243.
694 G. Che-Galicia et al. / Chemical Engineering Journal 280 (2015) 682–694

[58] R. Quintana-Solórzano, G. Barragán-Rodríguez, H. Armendáriz-Herrera, J.M. [75] G. Saracco, F. Geobaldo, G. Baldi, Methane combustion on Mg-doped LaMnO3
López-Nieto, J.S. Valente, Understanding the kinetic behavior of a Mo–V–Te– perovskite catalysts, Appl. Catal. B 20 (1999) 277–288.
Nb mixed oxide in the oxydehydrogenation of ethane, Fuel 138 (2014) 15–26. [76] M. Egashira, S. Kawasumi, S. Kagawa, T. Seiyama, Temperature programmed
[59] P.N. Brown, G.D. Byrne, A.C. Hindmarsh, VODE: a variable-coefficient ODE desorption study of water adsorbed on metal oxides. I. Anatase and rutile, Bull.
solver, SIAM J. Sci. Stat. Comput. 10 (1989) 1038–1051. Chem. Soc. Jpn. 51 (1978) 3144–3149.
[60] H.H. Rosenbrock, An automatic method for finding the greatest or least value [77] M.P. Heynderickx, J.W. Thybaut, H. Poelman, D. Poelman, G.B. Marin, Kinetic
of a function, Comput. J. 3 (1960) 175–184. modeling of the total oxidation of propane over CuO–CeO2/c–Al2O3, Appl.
[61] P.T. Boggs, J.R. Donaldson, R.H. Byrd, R.B. Schnabel, Algorithm 676 ODRPACK: Catal. B 95 (2010) 26–38.
software for weighted orthogonal distance regression, ACM T. Math. Software [78] D. Wolf, N. Dropka, Q. Smejkal, O. Buyevskaya, Oxidative dehydrogenation of
15 (1989) 348–364. propane for propylene production–comparison of catalytic processes, Chem.
[62] D.W. Marquardt, An algorithm for least-squares estimation of nonlinear Eng. Sci. 56 (2001) 713–719.
parameters, J. Soc. Ind. Appl. Math. 11 (1963) 431–441. [79] G. Schwarz, Estimating the dimension of a model, Ann. Stat. 6 (1978) 461–464.
[63] E. Santacesaria, Kinetics and transport phenomena, Catal. Today 34 (1997) [80] J.C. Vedrine, J.M.M. Millet, J.-C. Volta, Molecular description of active sites in
393–400. oxidation reactions: acid–base and redox properties, and role of water, Catal.
[64] C.H. Bartholomew, Mechanisms of catalyst deactivation, Appl. Catal. A 212 Today 32 (1996) 115–123.
(2001) 17–60. [81] K. Chen, A. Khodakov, J. Yang, A.T. Bell, E. Iglesia, Isotopic tracer and kinetic
[65] M. Boudart, D.E. Mears, M.A. Vannice, Kinetics of heterogeneous catalytic studies of oxidative dehydrogenation pathways on vanadium oxide catalysts, J.
reactions, Ind. Chim. Belg. 32 (1967) 281–284. Catal. 186 (1999) 325–333.
[66] G.F. Froment, Fixed bed catalytic reactors. technological and fundamental [82] M. Aouine, J.L. Dubois, J.M.M. Millet, Crystal chemistry and phase composition
design aspects, Chem. Ing. Tech. 46 (1974) 374–386. of the MoVTeNbO catalysts for the ammoxidation of propane, Chem. Commun.
[67] A.I. Anastasov, An investigation of the kinetic parameters of the o-xylene 13 (2001) 1180–1181.
oxidation process carried out in a fixed bed of high-productive vanadia-titania [83] M. Hävecker, S. Wrabetz, J. Kröhnert, L.-I. Csepei, R. Naumann d’Alnoncourt,
catalyst, Chem. Eng. Sci. 58 (2003) 89–98. Y.V. Kolen’ko, F. Girgsdies, R. Schlögl, A. Trunschke, Surface chemistry of
[68] F. López-Isunza, Steady state and dynamic behavior of an industrial fixed bed phase-pure M1 MoVTeNb oxide during operation in selective oxidation of
catalytic reactor Ph.D. thesis, University of London (Imperial College), United propane to acrylic acid, J. Catal. 285 (2012) 48–60.
Kingdom, 1983. [84] J. Kubo, N. Watanabe, W. Ueda, Propane ammoxidation with lattice oxygen of
[69] C.N. Satterfield, Mass transfer in heterogeneous catalysis, M.I.T. Press, Mo-V-O-based complex metal oxide catalysts, Chem. Eng. Sci. 63 (2008) 1648–
Cambridge, 1970. 1653.
[70] B. Finlayson, Nonlinear Analysis in Chemical Engineering, McGraw-Hill, New [85] P. Mars, D.W. van Krevelen, Oxidations carried out by means of vanadium
York, 1980. oxide catalysts, Chem. Eng. Sci. Spec. Suppl. 3 (1954) 41–59.
[71] L. Lapidus, J. Seinfeld, Numerical Solution of Ordinary Differential Equations, [86] S.S.E.H. Elnashaie, S.S. Elshishini, Modelling, Simulation and Optimization of
Academic Press, New York, 1971. Industrial Fixed Bed Catalytic Reactors, Gordon and Breach Science Publishers,
[72] C.H. Li, B.A. Finlayson, Heat transfer in packed beds-a reevaluation, Chem. Eng. Yverdon, 1993.
Sci. 32 (1977) 1055–1066. [87] M.M. Bettahar, G. Costentin, L. Savary, J.C. Lavalley, On the partial oxidation of
[73] B.D. Kulkarni, L.K. Doraiswamy, Estimation of effective transport properties in propane and propylene on mixed metal oxide catalysts, App. Catal. A 145
packed bed reactors, Catal. Rev. 22 (1980) 431–483. (1996) 1–48.
[74] D. Wen, Y. Ding, Heat transfer of gas flow through a packed bed, Chem. Eng.
Sci. 61 (2006) 3532–3542.

View publication stats