Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Reaction Kinetics, Mechanisms and Catalysis

https://doi.org/10.1007/s11144-019-01567-z

Mathematical modeling of a catalytic membrane


reactor: dehydrogenation of methanol over copper
on silica‑montmorillonite composite

Ekaterina V. Shelepova1 · Lyudmila Y. Ilina1 · Aleksey A. Vedyagin1,2 

Received: 15 December 2018 / Accepted: 15 March 2019


© Akadémiai Kiadó, Budapest, Hungary 2019

Abstract
The process of methanol dehydrogenation over nanostructured copper-containing
catalyst was simulated using a two-dimensional non-isothermal stationary math-
ematical model of the catalytic membrane reactor. The model considers mass and
heat transfer in both axial and radial directions. Additionally, it takes into account
the change of the reaction mixture volume occurs as a result of chemical reactions
and selective removal of hydrogen through the membrane. The reaction of metha-
nol dehydrogenation realized within the inner side of tubular membrane reactor was
thermodynamically conjugated with hydrogen oxidation reaction taking place in
the outer (shell) side. The effects of various parameters on the process performance
have been calculated and discussed. The most effective way to realize the process
of methanol dehydrogenation in a catalytic membrane reactor was found to use the
small values of residence times along with the temperature of the reactor outer wall
of about 125–150  °C. The heat generated by the exothermic oxidation reaction in
the shell side can be efficiently utilized to warm the reaction zone up to the desired
temperature.

Keywords  Mathematical modeling · Catalytic membrane reactor · Methanol


dehydrogenation · Hydrogen removal · Thermodynamically conjugated processes

Abbreviations

List of symbols
Am Area of membrane, ­m2
Ct,s,c
i Concentrations, kmol ­m
−3

cp Heat capacity coefficient, kJ g−1 ­K−1

* Aleksey A. Vedyagin
vedyagin@catalysis.ru
1
Boreskov Institute of Catalysis SB RAS, pr. Ac. Lavrentieva, 5, Novosibirsk, Russia 630090
2
National Research Tomsk Polytechnic University, Lenin av., 30, Tomsk, Russia 634050

13
Vol.:(0123456789)
Reaction Kinetics, Mechanisms and Catalysis

Dt,c
ei Effective coefficient of radial diffusion of component i, ­m ­s
2 −1

Dij Molecular diffusivity for component i in a binary mixture of i and


j, ­m2 ­s−1
t,c
Dm Coefficient of molecular diffusion, ­m2 ­s−1
Dkn Knudsen diffusion coefficient, ­m2 ­s−1
de Equivalent pore channel diameter, m
de1, ­de2 Equivalent diameter, m
dk Diameter of catalyst, m
dr Diameter of membrane reactor, cm
Gt,s Gas flow rate, ml min−1
–∆Hj Heat effect of reaction j, kJ mol−1
l Length of reactor, m
Mi Molecular weight of ith compound, g ­mole−1
Nt,s Number of components in reaction mix
NR Number of reactions within the tube side of reactor
Perm Permeability
Pw Perimeter of wall, m
PH2 t,s,c Partial pressure of ­H2, atm
P0 Pressure at normal conditions, atm
Q0 Permeability constant, kmol ­m−1 ­s−1 ­atm−1/2, defined in
[1]: Q 0 = 1.0061 × 10−12 exp (−767.38∕T)
QH2 Hydrogen permeation rate through the membrane, kmol ­s−1
r1,2 Radial coordinate into the fixed bed catalyst, in the ceramic sup-
port, m
R Universal gas constant, J mol−1 ­K−1
Scr.s. Cross section area of shell side, ­m2
Ssp1,sp2 Specific surface area, ­m−1
Tt,s,c,w Temperature, K
Tcr Critical temperature of substance, K
T0 Temperature at normal conditions, K
ut0 Linear velocity, m/s, ut0 = Gt/πR21
ut,s
l Axial velocity, m s
−1

Vcr Critical volume of substance, ­cm3 ­mol−1


wj Rate of reaction, kmol kg−1 cat ­s
−1

w = 𝜂 ROx, H2 Rate of the hydrogen oxidation reaction, kmol kg−1


s s
cat ­s
−1

yi Mole fraction of ith component

Greek letters
αw Coefficient of heat-transfer at wall, kJ m−2 ­s−1 ­K−1,
𝛼w = Nuwe 𝜆g ∕de
α1,2 Coefficient of heat-transfer between the membrane/exterior wall
and fixed bed catalyst (shell side), kJ m−2 ­s−1 ­K−1
δ Membrane thickness, m
δc Ceramic support thickness, m
εt,s,c Porosity of catalyst layer (tube, shell side), ceramic support
γij Stoichiometric coefficient for i-component into j-reaction

13
Reaction Kinetics, Mechanisms and Catalysis

𝜆t ef Effective coefficient of radial thermal conductivity, J m−1 ­s−1 ­K−1


λc Thermal conductivity of the ceramic support, J m−1 ­s−1 ­K−1
λg Thermal conductivity of argon gas, reference value, J m−1 ­s−1 ­K−1
μ Dynamic viscosity of a gas mix, kg m−1 ­s−1
ρt,s
G Gas density, kg m
−3
t,s
ρk Density of catalyst, kg m−3

Indexes
c Ceramic support
in Inlet
s Shell side
surf Surface
t Tube side
w Outer wall of reactor

Introduction

In recent years, the application of inorganic membranes in catalytic reactors causes


an increasing interest, since they possess high thermal and chemical stability [2, 3].
Due to the high specific surface, high selectivity and permeability, and the ability to
control the permeation rate, the membranes can be used in many industrially impor-
tant processes like food industry, environmental and biotechnological technologies,
biomedicine, etc. [4].
The membrane in a chemical reactor can provide different functions such as
selective removal of one or few products from reaction zone, controllable supply of
the reagents, or an interface for the contact of explosive reagents. The main advan-
tage of the catalytic membrane reactor is that the target reaction and membrane
separation occur simultaneously, which diminishes significantly the process cost if
compare with conventional reactors [5]. In addition, the selective removal of certain
products from the reaction zone shifts the thermodynamic equilibrium of reversible
reactions towards the product side, thus increasing the conversion [6]. It allows one
to decrease the temperature of the process and enhance the selectivity, since the high
temperatures are known to facilitate the acceleration of side reactions and deacti-
vation of the catalysts. During the last 50  years, the advantageous application of
membrane reactor concept was successfully demonstrated for a variety of industrial
processes such as hydrogenation, direct and oxidative dehydrogenation, catalytic
decomposition, etc. [7–13].
In present work, a process of methanol dehydrogenation into methyl formate has
been considered. This process attracts a growing attention in recent years, since
methyl formate was proclaimed as a key compound of single-carbon chemistry, and
a wide range of chemicals can be produced from it [14, 15]. Methyl formate, in its
turn, can be obtained via different ways, but the methanol dehydrogenation process
is characterized by relatively mild conditions and lower sensitivity to the feedstock
impurities if compare with conventionally used process of methanol carbonylation
[16].

13
Reaction Kinetics, Mechanisms and Catalysis

On the other hand, the process of methanol dehydrogenation to methyl formate is


complicated by a number of side reactions. Thus, the methyl formate formed can be
decomposed into carbon monoxide and hydrogen over the same catalyst and within
the same temperature range. Moreover, the thermodynamics of the reaction of meth-
anol dehydrogenation to methyl formate is unfavorable. An increased temperature
is required to achieve the high methyl formate yield. Unfortunately, at such tem-
peratures, the side reaction of the methyl formate decomposition accelerates as well.
However, such methyl formate production process could be suitably carried out in a
methanol production plant because the decomposed products are the starting materi-
als for methanol production. As it was already mentioned, the conversion of reagents
can be improved by removing at least one of the products from the reaction zone
using a selective hydrogen permeable membrane. For the case of methanol dehydro-
genation, it was reported by Prof. Rozovskii and co-authors [17, 18].
Despite the successful examples of membrane reactor application for the metha-
nol dehydrogenation processes, the use of this approach in practice is limited. The
process parameters are still need to be optimize, which can be done by means of a
mathematical modeling and simulation. There are a huge number of developed mod-
els for catalytic membrane reactors of different configurations. It is worth noting that
the most part of existing mathematical models consider the mass transfer processes
in a reactor, support and membrane, but neglect the heat transfer processes that play
a significant role. Some of the models are one-dimensional and take into account
only convective mass transport with flow and flux through the membrane, while
the diffusion along the reactor length and radius is not taken into account. Thus, a
one-dimensional stationary mathematical model was used to simulate the process of
methyl formate production in the endothermic part of the reactor conjugated with
the methanol synthesis in its exothermic part [19]. A similar one-dimensional iso-
thermal model was used to study the conjugated processes of methanol conversion
to dimethyl ether, methyl formate and hydrogen in a membrane reactor with two
membranes (water permeable membrane in the exothermic part and hydrogen per-
meable one in the endothermic part) [20]. However, the isothermal models give no
possibility to investigate any thermal effects on the process performance, which are
of great importance, especially in the case of conjugated processes. An example of
such conjugation was reported by Bakhtyari et  al. for a thermally coupled mem-
brane reactor, when the convective heat transfer along the reactor length was taken
into account [21]. The disadvantage of one-dimensional mathematical models is that
they limit the analysis of a number of parameters, which influence the process per-
formance. A non-isothermal two-dimensional model is reported in [22] to be used
for modelling of the methanol oxidative dehydrogenation. This model considers a
mass-transfer process through the pores of a composite membrane, but radial heat
and mass distributions in the inner and outer parts of the tube were neglected. At the
same time, these distributions affecting the performance of the whole process are of
great importance to be taken into account.
Among the significant number of mathematical models for catalytic membrane
reactors there is no one taking into account simultaneously both the heat and mass
transfer processes in both the axial and radial directions of the membrane reactor,
over the support and membrane, and also the change of the reaction mixture volume

13
Reaction Kinetics, Mechanisms and Catalysis

taking place due to chemical reactions and selective removal of products. Recently,
we have reported the model developed for dehydrogenation processes of hydrocar-
bons [23]. In the next step, the model was adapted for methanol dehydrogenation
process, when target reaction of methanol dehydrogenation to methyl formate and
consecutive side reaction of methyl formate decomposition into carbon monoxide
and hydrogen [24, 25]. The simulation of the process was performed for the cop-
per-containing catalysts based on carbon and silica supports [26, 27]. In the present
paper, the possibilities of the developed mathematical model are demonstrated for
the copper catalyst supported on the nanostructured silica-montmorillonite com-
posite material. It should be noted that the kinetics of the methanol dehydrogena-
tion over this catalyst differs significantly from the kinetics for previously studied
catalytic systems [28]. The effect of such parameters as the inlet temperatures of gas
flows in the tube and shell sides and residence time on the process performance is
analyzed.

Mathematical model of a membrane reactor

The scheme of catalytic membrane reactor is presented in Fig.  1. The membrane


reactor consists of two concentric tubes. The interior ceramic tube is filled with the
fixed bed catalyst active in a methanol dehydrogenation reaction. A very thin palla-
dium-silver alloy is deposited on the outer surface of thermostable ceramic support.
The space between the tube and shell side of the reactor is filled by an oxidation
catalyst.
Several simplifying assumptions for two-dimensional non-isothermal stationary
mathematical model were used:

1. Steady-state conditions;
2. Negligible convective radial transfer and axial dispersion;
3. High heat conduction of the membrane material;
4. Negligible internal mass- and energy-transport limitations inside the catalyst pel-
lets (tube side) and negligible external mass and heat transfer resistances at the
surface of the pellets (tube, shell side).

Fig. 1  The scheme of the membrane reactor

13
Reaction Kinetics, Mechanisms and Catalysis

The developed two-dimensional non-isothermal stationary reactor model takes into


account the mass and energy balance equations with the appropriate boundary condi-
tions for both tube and shell sides, and for ceramic support layer. The permeation rate
of hydrogen ­(QH2) through the dense membrane layer permeable only for hydrogen was
determined by the Sieverts law. The mass transfer equations were taken from [29]. The
equations of mass and energy balances are given below.

Model equations

Mass balances

Tube side:
0 < r1 < R1
To describe the mass transfer processes in the tube side, the convective mass transfer
in the axial direction, diffusion in the radial direction and chemical reactions were taken
into account.
( )
𝜕( u tl C ti ) 1 𝜕 𝜕 C ti ∑NR

(1)
t t t t
=𝜀 r D + 𝜌k (1 − 𝜀 ) × 𝛾ij w j , ∀i
𝜕l r 1 𝜕 r 1 1 ei 𝜕 r 1 j=1

Boundary conditions:

𝜕 C ti
l = 0 ∶ C ti = C tin ; r 1 = 0 ∶ = 0 (2)
𝜕r1

At the boundary tube/ceramic support:

𝜕 C ti || 𝜕 C ci ||
r 1 = R 1 ∶ C ti = C ci ; D te 𝜀 t | = D ce 𝜀 c | , ∀i (3)
i 𝜕 r 1 ||r =R i 𝜕 r 2 ||r =R
1 1 2 1

Ceramic support:
R1 < r2 < R2
The diffusion of all substances in the radial direction was considered for ceramic
support layer.
( )
𝜀c 𝜕 𝜕 C ci
r 2 D ce = 0, ∀i (4)
r2 𝜕r2 i 𝜕r
2

Boundary conditions:
The boundary conditions for ceramic support/tube are the identical to those applied for
tube/ceramic support.
At the boundary ceramic support/shell:

13
Reaction Kinetics, Mechanisms and Catalysis

𝜕 C ci ||
r2 = R2 ∶ | = 0, ∀i ≠ H2 (5)
𝜕 r 2 ||r =R
2 2

Hydrogen flux through the membrane has been proportional to the difference of
the square roots of the hydrogen partial pressures across the membrane:

𝜕 C cH | ( )[√ √ ]
2 |
Q0
D ce 𝜀 c | = P cH − P sH (6)
H2 𝜕 r 2 ||r =R 𝛿 2 2
2 2

Shell side:
The convective mass transfer in the axial direction, flux of the hydrogen across the
membrane and the catalytic reaction of hydrogen oxidation were taken into consid-
eration with appropriate boundary conditions.
𝜕 (u sl C sH ) Q� H2 Pw
2
= + 𝜌sk (1 − 𝜀s ) × 𝛾i ws , Q�H
𝜕l Scr.s. 2
(7)
( )[√ c √ ]
= QH2 ∕A m , Q H 2 = Q 0 A m ∕𝛿 P H − P sH
2 2

𝜕 (u sl C si )
= 𝜌sk (1 − 𝜀s ) × 𝛾i ws , i = O2 , H2 O (8)
𝜕l
Boundary conditions:
l = 0∶ C sH = C sH O = 0, COs = COs (9)
2 2 2 2 ,in

Energy balances

Tube side:
0 < r1 < R1
To describe the heat transfer processes in the tube side, the convective heat transfer
in the axial direction, thermal conductivity in the radial direction and heat effect of
the reactions were taken into account.
( )
) ∑
N
𝜕Tt 1 𝜕 𝜕Tt ( R

𝜌tG c tp u tl = 𝜆 tef r 1 + 𝜌tk 1 − 𝜀t × w j (−ΔH j ) (10)


𝜕l r1 𝜕r1 𝜕r1 j=1

Boundary conditions:
𝜕Tt
l = 0 ∶ T t = T tin ; r 1 = 0 ∶ = 0 (11)
𝜕r1
At the boundary tube/ceramic support:

13
Reaction Kinetics, Mechanisms and Catalysis

𝜕 T t || 𝜕 T c ||
r 1 = R 1 ∶ T t = T c , 𝜆 tef | = 𝜆c (12)
𝜕 r 1 |r 1 =R 1 𝜕 r 2 ||r 2 =R 1

Ceramic support:
R1 < r2 < R2
The thermal conductivity in the radial direction was considered for ceramic support
layer.
( )
1 𝜕 𝜕Tc
(13)
c
𝜆 r2 =0
r2 𝜕r2 𝜕r2
Boundary conditions:
The boundary conditions for ceramic support/tube are the identical to those applied
for tube/ceramic support.
At the boundary ceramic support/shell:

𝜕 T c ||
r2 = R2 ∶ 𝜆c = 𝛼 1 (T s − T c ) (14)
𝜕 r 2 ||r 2 =R 2

Shell side:
In the shell side the convective heat transfer in the axial direction, heat of hydrogen
oxidation, heat exchange (between shell side and the ceramic support; between shell
side and the reactor outer wall) were taken into consideration.
𝜕T s
= Ssp1 𝛼 1 (T c − T s ) + Ssp2 𝛼 2 (T w − T s ) + 𝜌sk (1 − 𝜀s ) × ws (−ΔH)
𝜌sG csp u sl
𝜕l
(15)
Boundary conditions:
l = 0 ∶ T s = T sin (16)
Account of volume change

The developed model enables considering the volume change in the tube and shell
sides of the reactor due to reaction stoichiometry and hydrogen diffusion through a
membrane. Change of mole amount is considered in the equation for calculation of
axial velocity. The gas mixture velocities in the tube and shell sides were determined
from the mass conservation equations.

(N )
𝜕 y ci ||
R1
𝜕 u tl2𝜀c T0 ∑ t
2RT0 t ∑ ∑
Nt NR

∫ i=1 j=1 ij j 1 1
= D ce | + 𝜌 (1 − 𝜀 t
) 𝛾 w r dr
𝜕l R1 Tav 𝜕r2 | | 2 k
i
R1 P0
i=1 |r 2 =R 1 0
(17)

13
Reaction Kinetics, Mechanisms and Catalysis

Q 0 [√ c √ ] ∑ N
𝜕 u sl 2R2 RT0 s

= P H2
− P s
H2
+ 𝜌s
k (1 − 𝜀s
) 𝛾i ws (18)
𝜕l P0 (R23 − R22 ) 𝛿 i=1

Heating the gas flows

To heat up the reaction mixture in the dehydrogenation zone, the heat released due
to the exothermic reaction of hydrogen oxidation in the shell side of the membrane
reactor can be used. In this case, the inlet temperature of the gas flows in both the
tube and shell sides of the reactor is preset, and the heat transfer coefficient α2 in
Eq. 5 is equal to zero, in order to exclude the heat exchange with the reactor outer
wall.

Parameters of mathematical model

The following dependences are used for determination of mass and heat transfer
coefficients.
The effective coefficient of radial diffusion [30]: Dte = ADtm + BDtm Ree Sc ,
Ree = ve de 𝜌g ∕𝜇—Reynolds number; Sc = 𝜇∕(𝜌g Dm )—Schmidt number (diffusion
Prandtl’s criterion). The effective�
coefficient of molecular diffusion is determined by
∑n � �
Wilkes formula: D m = (1 − y i )
t, c
j=1, j≠i
y j ∕D i j  . Molecular diffusivity for com-
ponent i in a( binary )n mixture of i and j was calculated according to equation:
D i j = D i j (T0 ) T∕T0 (P0 ∕P) , the coefficients D ij (T0 ) were calculated by the for-
mula [31]:
( ) ( )0.5 / ( ) [( ) ( ) ]2
T0 1.81 1 1 Tcr, i Tcr,j 0.1406 Vcr,i 0.4 Vcr,j 0.4
Dij (T0 ) = 0.43 + P0 + .
102 Mi Mj 104 102 102

The coefficient of radial diffusion √ for ceramic support layer: Dce = ((1∕Dcm )+
(1∕Dkn )) × perm , Dkn = 3 rcap ui , ui = 𝜋M8RT
−1 2
3 − average thermal velocity of molecule.

The entropy: Si (T) = Si (298) + ∫298


i ∕10
T
cpi (T)d ln T .
The substance enthalpy: Hi (T) = Hi (298) + ∫298 T
cpi (T)dT  . The heat capacity of
∑ P ∑
gas mixture: ct,s
p
= c y
pi i . The bulk density of gas mixture: 𝜌G = RT yi Mi . The
i i
effective coefficient of radial thermal conductivity [30]: 𝜆ef ,g ∕𝜆g = 𝜆0 ∕𝜆g + BRee Pr ,
Pr = 𝜇 cp ∕𝜆g is the Prandtl number. The heat transfer coefficient between the outer
wall of membrane reactor and catalyst in the shell side [32]: 1∕𝛼2 = 1∕𝛼w +
de2 ∕4𝜆ef ,g , 𝛼w = Nuwe 𝜆g ∕de , Nuwe = 3𝜆0 + 0.09 Re0.8 Pr1∕3 is the Nusselt number;
2𝜆
g √ e

de2 = 2Req is the equivalent diameter, Req = R23 − R22  . The heat transfer coefficient
between the membrane and catalyst in the shell compartment was determined by the
same formula [33].

13
Reaction Kinetics, Mechanisms and Catalysis

Pressure change on the reactor length was calculated according to Ergun’s equa-
tion [30]:
dp 150𝜇 u(1 − 𝜀)2 1.75𝜌 u2 (1 − 𝜀)
= + , l = 0 ∶ p = pin .
dl d 2 𝜀3 d𝜀3
Both viscosity and thermal conductivity of a gas mixture were calculated depend-
ing on real composition of a mixture in each point of the reactor. Viscosity and
thermal conductivity( for individual
)m substance
( were
)n determined by the following for-
mulas: 𝜇(T) = 𝜇0 T∕T0 , 𝜆(T) = 𝜆0 T∕T0  , where 𝜆0 , 𝜇0 are the thermal con-
ductivity and dynamic viscosity of substance under normal conditions ­(T0 = 273 K,
­P0 = 1  atm.); n and m—the exponents to be defined experimentally. Viscosity and
thermal conductivity of a gas mixture were defined by the following Eq. [31]:
−1
⎡ � �⎤ � � �0.5 � �0.25 �2 � √ �� � �

n
⎢ �n
𝜇
yj ⎥ 𝜇 𝜇i Mi Mi 0.5
𝜇 i ⎢1 +
yi ⎥⎥
𝜇= Φij , Φij = 1 + 2 2 1+ ,
i=1 ⎢ j=1
𝜇j Mj Mj
⎣ j≠i ⎦
−1
⎡ � �⎤ � � ��−1∕2 � � �0.5 � �0.25 �2

n
⎢ �n
yj ⎥ 1.065 Mi 𝜆i Mi 𝜆 𝜇i Mj
𝜆= 𝜆 i ⎢1 + 𝜆
Φij 𝜆
⎥ , Φij = √ 1 + M 1+ , i = .
i=1 ⎢ j=1
y i ⎥ 2 2 j 𝜆j M j 𝜆 j 𝜇 j Mi
⎣ j≠i ⎦

The mathematical model of catalytic membrane reactor was verified earlier for
the case of ethane dehydrogenation process [34].

Reaction kinetics

Tube side
The methanol dehydrogenation process was thermodynamically conjugated with
the hydrogen oxidation process. The dehydrogenation of methanol to methyl formate
was simulated over nanostructured Cu/SMC catalyst, where SMC is silica-montmo-
rillonite composite [35]. The reaction was considered along with the side reaction of
methyl formate decomposition:
[1] ­2CH3OH ⇔ C­ H3OCHO + 2H2
[2] ­CH3OCHO ⇒ 2CO + 2H2
The kinetic parameters for these reactions on Cu/SMC catalysts were used for
calculations [28]: w1 = k+1 (CCH
2
OH
− CCH3 OCHO CH2 ∕Keq ), k+1 = k10 e−Ea1 ∕RT   , ­k10 =
3 2
−ΔG
 7.28·108 ­M−1s−1, ­Ea1 = 48.9 kJ/mol, Keq = e RT  , ΔG = ΔHR − TΔSR.
The reaction of methyl formate decomposition into CO and hydrogen
is practically irreversible [36]: w2 = k2 CCH3 OCHO , k2 = k20 e−Ea2 ∕RT , k20 = 1.035⋅
106 s−1 , Ea2 = 48.6 kJ∕mol.
Reaction of methanol dehydrogenation to formaldehyde was not taken into
account, while it occurs at 320 °C and above, and the methanol dehydrogenation to
methyl formate takes place in a membrane reactor at lower temperatures if compare
with a conventional reactor.

13
Reaction Kinetics, Mechanisms and Catalysis

Shell side
In the shell compartment of the membrane reactor the catalytic reaction of hydro-
gen oxidation was considered.
2H2 + O2 → 2H2 O, ΔH298 = −242.7 kJ/mol
The kinetic model reported by Tavazzi et al. [37] was used for the computation:
pO2
ROx,H2 = kOx,H2 PH2 𝜎O2 , 𝜎O2 = −6
; kOx,H2 = k873(K) exp[−Eatt ∕R((1∕T)−(1∕873))]
10 + pO2

k873(K) = 4.196 mol g−1 s−1 atm−1; ­Eatt/R = 5000 K.
The kinetic parameters were obtained for the Rh/α-Al2O3 spherical catalyst.

Numerical algorithms

The mathematical model of catalytic membrane reactor consists of the partial derivatives
Eqs. 1–18 (PDEs). Each PDE system is transformed into the non-linear set of ordinary
differential equations (ODEs) using the method of lines. At the same time, the derivative
along the length is not approximated [38]. In this case, the system of ODEs is as follows:

⎛ ci,1 ⎞ ⎛ bi,1 ci,1 0....................0 ⎞ ⎛ di,1 ⎞


dCi ⎜c ⎟ ⎜ a b c .................0 ⎟ ⎜d ⎟
Ci = ⎜
i,2 ⎟
Ai = ⎜ ⎟ ; D = ⎜ i,2 ⎟;
i,2 i,2 i,2
= Ai Ci + Di ; ;
dl ⎜ .......⎟ ⎜ ................................... ⎟ i
⎜ .......⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ci,nt ⎠ ⎝ 0................... ..ai,nt bi,nt ⎠ ⎝ di,nt ⎠
(1′)
The obtained ODEs are treated numerically using the appropriate method for ODE
solving. A semi-implicit Rosenbrock algorithm of the second order of accuracy with an
automatic choice of the integration step was used to solve the set of ODEs in this case
[39]:


m

m−1
yn+1 = yn + pi kn i [I − ai hfy (yn + 𝛾ij kn j )] kni
i=1 j=1


i−1
= hf (yn + 𝛽ij kn j )
j=1

Here I is the singular matrix, f­ y is the Jacobian matrix, a­ i, ­pi, γij, βij are parameters.
In the case when all a­ i are equal to each other, and γij = 0, the most effective realiza-
tion of Rosenbrock-type methods arises. At the same time, at each time step, it becomes
necessary to calculate and invert only a one matrix of dimension N. The L-stable sec-
ond-order accuracy method can be classified as a particular case [39]:
√ √
yj = yj−1 + (0.5 + 0.25 2) kj−1 1
+ (0.5 − 0.25 2) kj−1 2
(2′)

13
Reaction Kinetics, Mechanisms and Catalysis


Dj−1 = I − (1 − 0.5 2)hfy (yj−1 ) (3′)

1
Dj−1 kj−1 = hf (yj−1 ) (4′)


2
Dj−1 kj−1 = hf (yj−1 + 1
2 kj−1 ) (5′)

ODEs were solved using this method.


To solve the nonlinear system (1′), an iterative process is organized. For finding
the solution in the jth layer along the length at the first iteration, the nonlinear terms
­Ai, ­Bi, ­Ci, ­Di of the ODEs are calculated through the known values in the previous
layer (j − 1) (for the first point of the length—through the known input conditions).
Then, we calculate the coefficients of the Jacobian matrix (the Jacobian matrix in
the case of organizing the iterative process described below has tridiagonal form),
the coefficients of the matrix ­Dj−1 (3′). The values of k1 j−1 are calculated solving the
linear system (4′). The values of k2 j−1 are calculated solving the linear system (5′),
using k1 j−1 counted before. The coefficients k1 j−1 , k2 j−1 are calculated by marching,
including that for the ceramic support. The values of y­ j are calculated using system
(2′).
At the second iteration, the terms A­ i, ­Bi, ­Ci, ­Di are recalculated using known solu-
tions from the previous iteration. Solving the Eqs.  (4‘, 5′, 2′), we found ­yj. As a
result, due to the iterative process, ODE system becomes linear. The Jacobian matrix
in this case is tridiagonal, therefore the coefficients k1 j−1 , k2 j−1 are calculated by
marching, including that for the ceramic support.
Thus, the resulting systems of ODE for calculating the concentrations of all
components and temperatures were solved using a semi-implicit Rosenbrock-type
method of the second order of accuracy with an automatic choice of the integra-
tion step. The integration step along the length was selected based on the specified
integration accuracy. When the integration accuracy was varied, a value was deter-
mined, further reduction of which did not lead to an increase in the accuracy of the
results obtained. The size of the difference array (grid) along the radius also was
varied. The number of points along the radius was chosen in such a way that their
further increase did not lead to changes of the obtained results.

Results and discussion

In this section, the variations of process parameters investigated and corresponding


profiles will be presented. The parameters values used for the simulation of the pro-
cess in membrane reactor are listed below.
L = 0.15  m, ­r1 = 0.39 × 10−2 m, ­dr = 0.2 × 10−1 m, δ = 6 × 10−6 μm, δc = 1.1  mm,
­dc = 1 μm, ρt,s 6 −3 t,s
k  = 1 × 10 g m , dk  = 0.15 × 10
−2
m, ε t,s = 0.5, λc = 0.1 J m−1 ­s−1 ­K−1,
t,s W t,s
Tin  = 25 °C, ­T  = 150 °C, ­P  = 1 atm, C CH OH, in = 10 vol%, C tAr, in = 90 vol%.
t
3

13
Reaction Kinetics, Mechanisms and Catalysis

Fig. 2  The concentration profiles along the reactor length (membrane reactor with hydrogen oxidation) at
­Tw = 150 °C: a τ = 0.5 s; b τ = 5.8 s

The mathematical model of a tubular reactor (without the hydrogen removal to


the shell side of the membrane reactor) was verified for the methanol dehydrogena-
tion process [28]. Verification of the mathematical model of the catalytic membrane
reactor was carried out for the ethane dehydrogenation process [34].
The mathematical model allows one to calculate different profiles along the
length and radius of the reactor. The concentration profiles of methanol, methyl
formate, hydrogen and carbon monoxide along the reactor length for two resi-
dence times are presented in Fig.  2. Here, the residence time of the reaction
mixture is defined as τ = L/u0t. As follows from Fig. 2a, along with a decrease of
methanol concentration an increase of the concentration of all reaction products,
including by-products, is observed.
It is known that the residence time is one of the key parameters affecting the
process performance. It affects both the reagent conversion and product selectiv-
ity. Thus, an increase of the residence time up to τ = 5.8 s changes cardinally the
character of the curves (Fig.  2b). As seen, the side reaction of methyl formate
decomposition significantly accelerates, thus diminishing the concentration of a
target product – methyl formate.
The radial concentration profiles obtained by using the developed model are
shown in Fig. 3. Since the membrane is permeable for hydrogen only, the concen-
tration gradient along the radius appears only for this component of the reaction
mixture. It can be seen that hydrogen removal in the final sector of the reactor
length is the most efficient, when the difference in partial pressures (which deter-
mines the hydrogen flux through the membrane according to the Sievert’s law) in
the inner and outer parts of the reactor reaches its maximum value.
Using the developed model it is possible to estimate how the hydrogen removal
through the membrane and its further oxidation in the shell side of the reactor
affects the methanol conversion. In order to compare the results with that for a
conventional tubular reactor (i.e., without hydrogen removing through the mem-
­ H2 was equated to zero.
brane to the shell side of the reactor), the hydrogen flux Q
To estimate the conversion in a membrane reactor without hydrogen oxidation,
the rate of hydrogen oxidation reaction ­ws in the shell side of the reactor was

13
Reaction Kinetics, Mechanisms and Catalysis

Fig. 3  The hydrogen concentra-


tion profiles in radial direction
of the tube side and ceramic
support (membrane reactor with
hydrogen oxidation), τ = 0.5 s,
­Tw = 150 °C

equated to zero. Fig.  4 compares the conversion profiles for these three varia-
tions considered. The methanol conversion in a tubular reactor increases along
the length of the reactor and reaches a value of 60% at the reactor outlet. The
conversion in the membrane reactor exceeds the value in the tubular reactor by
8% due to hydrogen diffusion into the shell side of the reactor and a shift of the
thermodynamic equilibrium of dehydrogenation reaction towards products. The
maximum methanol conversion of 79% was achieved in a membrane reactor
with hydrogen oxidation. In this case, the hydrogen flux through the membrane
increases, and the equilibrium of the dehydrogenation reaction is shifted more
significantly. Based on the numerical calculations performed, it can be concluded
that the methanol dehydrogenation process can be realized most efficiently in
a membrane reactor with additional hydrogen oxidation in the shell side of the
reactor.
There are a number of parameters that may affect the process performance in the
membrane reactor. The residence time and the temperature (in our case, the temper-
ature of the reactor outer wall) should be noted first of all. The mathematical mode-
ling allows one to estimate and clearly demonstrate the influence of these parameters

Fig. 4  Methanol conversion
along the reactor length for
tubular reactor (TR), membrane
reactor (MR) and membrane
reactor with hydrogen oxidation
(MRHO); τ = 5.8 s, ­Tw = 150 °C

13
Reaction Kinetics, Mechanisms and Catalysis

­w
Fig. 5  Methanol conversion (a), methyl formate selectivity (b) and yield (c) versus temperature T
(membrane reactor with hydrogen oxidation)

on the process characteristics. Fig. 5 demonstrates the temperature dependences of


methanol conversion, methyl formate selectivity and yield within a range of resi-
dence time of 0.1–10 s. As it follows from Fig. 5a, b, methanol conversion increases
along with a rise of the residence time and temperature, while methyl formate selec-
tivity decreases dramatically, thus indicating an increase of side reaction contribu-
tion. All it results in different shapes of the curves for the temperature dependences
of methyl formate yield (Fig.  5c). There are two declining curves for the case of
residence times of 3 and 10 s, two dome-shaped curves (τ = 0.5 and 1 s), and one
ascending curve (τ = 0.1 s).
Fig. 6 presents corresponding residence time dependences for different reaction
temperatures. As seen from these graphs, high enough values of methyl formate
selectivity and maximal values of methyl formate yield can be obtained only within
vary narrow range of residence times (0.2 s and less), when the conversion of meth-
anol is minimal.
Based on the numerical calculations made, it can be concluded that the most
effective way to realize the process of methanol dehydrogenation in a catalytic mem-
brane reactor is to use the small values of residence time along with the temperature
of the reactor outer wall of about 125–150 °C. Note that the short residence times
are preferable for different processes including the processes dealing with metha-
nol [40]. On the other hand, an alternative approach to hold on the desired level of

13
Reaction Kinetics, Mechanisms and Catalysis

a b
100 100

Methyl formate selectivity, %


90 ° C 90
00
Methanol conversion, %

w =2
80 °C 80
T 75
°C
=1

50
T w

70 w =1 70
C
T 5°
w =12
60 T 60
C
w =100°
50 T 50 T w
=1
00
°C
40 40
30 30 T w
=1
25
°C
20 20 T w
w =150
T w T °C
10 10 =2 =17
00 5
°C °C
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Residence time, s Residence time, s

c 35

30
Methyl formate yield, %

25
w
T =
100°C
20

15
w
T
=12
5°C
10
T w
=1
T w 50
5 T w =1 °C
=2 7 5°C
00
°C
0
0 2 4 6 8 10 12 14 16 18 20
Residence time, s

Fig. 6  Methanol conversion (a), methyl formate selectivity (b) and yield (c) versus residence time (mem-
brane reactor with hydrogen oxidation)

temperature in the reaction zone is to use the heat released due to the exothermic
reaction of hydrogen oxidation in the shell side of the reactor. In this case, the inlet
temperatures of the gas flows Tt,s
in should be preset.
As it was already mentioned, the advantage of the non-isothermal mathematical
model is the possibility to analyze the temperature profiles. First of all, it allows
obtaining an optimal value of inlet temperature of the gas mixture Tt,s in , correspond-
ing to the high values of methanol conversion, methyl formate selectivity and yield.
Second, it gives the opportunity to estimate the maximum values of temperatures
in the reaction zone, which is also important, since the process of coke formation
might take place at high temperatures.
Fig. 7 shows the temperature profiles along the reactor length in the tube and
shell sides of the reactor for methanol dehydrogenation simulated in a tubular
reactor, and in a membrane reactor with hydrogen removal by an inert gas or air
(with consideration of the hydrogen oxidation reaction).
In a tube side (Fig.  7a), the temperature is substantially reduced due to the
endothermicity of the reaction. When no additional heat is released (in the case of
tubular and membrane reactors), the temperature in the reactor outlet is minimal.
A small difference in outlet temperature observed for these two reactor types is
connected with hydrogen removal through the membrane (when both mass and
heat transfer processes take place). In the case of the conjugated process (with

13
Reaction Kinetics, Mechanisms and Catalysis

Fig. 7  Temperature profiles along the length in tube (a) and shell (b) sides of tubular reactor (TR), mem-
t
brane reactor (MR) and membrane reactor with hydrogen oxidation (MRHO), Tin  = 150 °C; Tsin = 100 °C;
τ = 0.5 s

oxidation of removed hydrogen in a shell side), the heat released compensates the
endothermic temperature fall, and the temperature starts to rise up to 145 °C. The
minimal value of temperature reached in this case was 125 °C (at reactor length
of 0.04 m).
The temperature profiles in the shell side of the reactor differ from those in
s
the tube side (Fig. 7b). The inlet temperature Tin was preset to 100 °C. Therefore,
the initial section for all three types of reactors is characterized by a temperature
increase due to heat exchange with the tube side of the reactor. Then, a characteristic
temperature fall due to the endothermicity of the dehydrogenation process can be
observed for the tubular and membrane reactors. If the hydrogen oxidation reaction
is simulated in the shell side of the reactor, the temperature rises up and reaches
150 °C at the reactor outlet.
Thus, the developed mathematical model allows one to analyze the concentration
and temperature profiles in a wide range of parameters. Numerical calculations help
to obtain the optimal parameters, when the process of methanol dehydrogenation to
methyl formate can be realized most efficiently.

Conclusions

The two-dimensional mathematical model for catalytic membrane reactor with


dense hydrogen-permeable membrane has been developed in order to simulate the
process of methanol dehydrogenation. The model allows one to study the influ-
ence of certain relevant parameters, such as temperature (including the inlet tem-
peratures of gas flows in the tube and shell sides) and residence time on the process
performance. The results of process simulation have shown the higher efficiency of
the catalytic membrane reactor in comparison with the tubular one. The maximum
value of methyl formate yield was obtained in the case of thermodynamically con-
jugated processes, when the reaction of hydrogen oxidation in the shell side of the
membrane reactor was additionally considered.

13
Reaction Kinetics, Mechanisms and Catalysis

Based on the numerical calculations made, it can be concluded that the most
effective way to realize the process of methanol dehydrogenation in the catalytic
membrane reactor is to use the small values of residence times along with the tem-
perature of the reactor’s outer wall of about 125–150 °C. When the inlet tempera-
tures of the gas flows are preset, an alternative way to keep the desired temperature
regime in the reaction zone is to recover the heat from the exothermic oxidation
reaction taking place in the shell side of the membrane reactor.

Funding  This study was supported by the Ministry of Science and High Education of Russian Federa-
tion (Project AAAA-A17-117041710086-6). The analysis of experimental results was partly carried out
within the Governmental Program “Science” of Tomsk Polytechnic University (Project No. 4.5200.2017).

Compliance with ethical standards 

Conflict of interest  The authors declare that they have no conflict of interest.

References
1. Gobina E, Hughes R (1994) Ethane dehydrogenation using a high temperature catalytic membrane
reactor. J Membr Sci 90:11–19
2. Zaman J, Chakma A (1994) Inorganic membrane reactors. J Membr Sci 92:1–28
3. Moparthi A, Uppaluri R, Gill BS (2010) Economic feasibility of silica and palladium composite
membranes for industrial dehydrogenation reactions. Chem Eng Res Design 88:1088–1101
4. Noble RD (1987) An overview of membrane separations. Sep Sci Technol 22:731–743
5. Gryaznov V (1999) Membrane catalysis. Catal Today 51:391–395
6. Rezai SAS, Traa Y (2008) Equilibrium shift in membrane reactors: a thermodynamic analysis of the
dehydrogenative conversion of alkanes. J Membr Sci 319:279–285
7. Gobina E, Hou K, Hughes R (1995) Ethane dehydrogenation in a catalytic membrane reactor cou-
pled with a reactive sweep gas. Chem Eng Sci 50:2311–2319
8. She Y, Han J, Ma YH (2001) Palladium membrane reactor for the dehydrogenation of ethylbenzene
to styrene. Catal Today 67:43–53
9. Dittmeyer R, Höllein V, Daub K (2001) Membrane reactors for hydrogenation and dehydrogenation
processes based on supported palladium. J Mol Catal A: Chem 173:135–184
10. Ahchieva D, Peglow M, Heinrich S, Mörl L, Wolff T, Klose F (2005) Oxidative dehydrogenation of
ethane in a fluidized bed membrane reactor. Appl Catal A Gen 296:176–185
11. Collins JP, Way JD (1994) Catalytic decomposition of ammonia in a membrane reactor. J Membr
Sci 96:259–274
12. Li G, Kanezashi M, Yoshioka T, Tsuru T (2013) Ammonia decomposition in catalytic membrane
reactors: simulation and experimental studies. AIChE J 59:168–179
13. Lytkina AA, Ilin AB, Yaroslavtsev AB (2016) Study of methanol steam reforming and ethanol con-
version in conventional and membrane reactors. Petrol Chem 56:1048–1055
14. Lu Zh, Gao D, Yin H, Wang A, Liu Sh (2015) Methanol dehydrogenation to methyl formate cata-
lyzed by S ­ iO2-, hydroxyapatite-, and MgO-supported copper catalysts and reaction kinetics. J Ind
Eng Chem 31:301–308
15. Gao D, Yin H, Feng Y, Wang A (2015) Coupling reaction between methanol dehydrogenation
and maleic anhydride hydrogenation over zeolite-supported copper catalysts. Can J Chem Eng
93:1107–1118
16. Rybakov AA, Bryukhanov IA, Larin AV, Zhidomirov GM (2016) Theoretical aspects of methanol
carbonylation on copper-containing zeolites. Petrol Chem 56:259–266
17. Gorshkov SV, Lin GI, Rozovskii AY, Serov YM, Uhm SJ (1999) Hydrogen diffusion through a
Pd-Ru membrane in the course of methanol dehydrogenation to methyl formate on a copper-con-
taining catalyst. Kinet Catal 40:93–99

13
Reaction Kinetics, Mechanisms and Catalysis

18. Gorshkov SV, Lin GI, Rozovskii AY (1999) The mechanism of methanol dehydrogenation with
methyl formate formation and the methods of process selectivity control. Kinet Catal 40:334–337
19. Goosheneshin A, Maleki R, Iranshahi D, Rahimpour MR, Jahanmiri A (2012) Simultaneous pro-
duction and utilization of methanol for methyl formate synthesis in a looped heat exchanger reactor
configuration. J Nat Gas Chem 21:661–672
20. Bakhtyari A, Haghbakhsh R, Rahimpour MR (2016) Investigation of thermally double coupled dou-
ble membrane heat exchanger reactor to produce dimethyl ether and methyl formate. J Nat Gas Sci
Eng 32:185–197
21. Bakhtyari A, Mohammadi M, Rahimpour MR (2015) Simultaneous production of dimethyl ether
(DME), methyl formate (MF) and hydrogen from methanol in an integrated thermally coupled
membrane reactor. J Nat Gas Sci Eng 26:595–607
22. Brinkmann T, Perera SP, Thomas WJ (2001) An experimental and theoretical investigation of
a catalytic membrane reactor for the oxidative dehydrogenation of methanol. Chem Eng Sci
56:2047–2061
23. Shelepova EV, Vedyagin AA, Mishakov IV, Noskov AS (2011) Mathematical modeling of the pro-
pane dehydrogenation process in the catalytic membrane reactor. Chem Eng J 176–177:151–157
24. Shelepova EV, Ilina LY, Vedyagin AA (2017) Theoretical predictions on dehydrogenation of metha-
nol over copper-silica catalyst in a membrane reactor. Catal Today. https​://doi.org/10.1016/j.catto​
d.2017.11.023
25. Shelepova EV, Ilina LY, Vedyagin AA (2018) Energy-efficient dehydrogenation of methanol in a
membrane reactor: a mathematical modeling. Chem Pap 72:2617–2629
26. Shelepova EV, Ilina LY, Vedyagin AA (2017) Kinetic studies of methanol dehydrogenation. Part I:
copper-silica catalysts. React Kinet Mech Catal 120:449–458
27. Shelepova EV, Ilina LY, Vedyagin AA (2017) Kinetic studies of methanol dehydrogenation. Part III:
carbon-supported copper catalysts. React Kinet Mech Catal 122:385–398
28. Shelepova EV, Ilina LY, Vedyagin AA (2017) Kinetic studies of methanol dehydrogenation. Part II:
copper on silica-montmorillonite composite. React Kinet Mech Catal 121:403–412
29. Abashar MEE, Al-Rabiah AA (2005) Production of ethylene and cyclohexane in a catalytic mem-
brane reactor. Chem Eng Proc 44:1188–1196
30. Aerov ME, Todes OM, Narinskii NA (1979) Apparatus with Stationary Granular Layer. Khimiya (in
Russian), Leningrad
31. Bretshnayder S (1966) The properties of gases and liquids. Chemistry, Moscow
32. Chumakova NA, Matros YS (1989) Mathematical modeling of fixed-bed catalytic reactors under
condition of given pressure drop. In: Mathematical modeling of catalytic reactors. Novosibirsk,
Nauka, pp 5–26 (in Russian)
33. Adrover ME, Lopes E, Borio DO, Pedernera MN (2009) Simulation of a membrane reactor for the
WGS reaction: pressure and thermal effects. Chem Eng J 154:196–202
34. Shelepova EV, Vedyagin AA, Noskov AS (2013) Effect of catalytic combustion of hydrogen on
dehydrogenation in a membrane reactor. II. Dehydrogenation of ethane. Verification of the math-
ematical model. Combust Explo Shock Waves 49(2):125–132
35. Vedyagin AA, Nizovskii AI, Golokhvast KS, Tsyrulnikov PG (2014) Nanocomposites on the

basis of layered silicates as the catalysts for the dehydrogenation of methanol. Nanotechnol Russia
9:693–699
36. Stull DR, Westrum EF, Sinke GC (1969) The chemical thermodynamics of organic compounds.
Wiley, New York
37. Tavazzi I, Beretta A, Groppi G, Forzatti P (2006) Development of a molecular kinetic scheme for
methane partial oxidation over a Rh/α-Al2O3 catalyst. J Catal 241:1–13
38. Fletcher CAJ (1988) Computational techniques for fluid dynamics 1, fundamental and general tech-
niques. Springer, Berlin
39. Novikov EA (1990) Numerical methods for solution of differential equations in chemical kinetics.
In: Mathematical methods in chemical kinetics. Novosibirsk, Nauka, pp 53–68 (in Russian)
40. Schmidt LD (2000) Millisecond chemical reactions and reactors. Stud Surf Sci Catal 130:61–81

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.

13

You might also like