Journal of Natural Gas Science and Engineering 9 (2012) 28e38

Contents lists available at SciVerse ScienceDirect

Journal of Natural Gas Science and Engineering

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

Incorporating differential evolution (DE) optimization strategy to boost hydrogen

and DME production rate through a membrane assisted single-step DME heat
exchanger reactor
R. Vakili, M.R. Rahimpour*, R. Eslamloueyan
Department of Chemical Engineering, School of Chemical and Petroleum Engineering, Shiraz University, Shiraz 71345, Iran

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

Article history: The present contribution aims to enhance dimethyl ether (DME) production rate as well as hydrogen as
Received 15 April 2012 clean-burning fuels and versatile applications. In this regard, a thermally coupled membrane configu-
Received in revised form ration (TCMDR), which is able to produce hydrogen and DME simultaneously, is proposed. Here, direct
11 May 2012
DME synthesis from syngas and cyclohexane dehydrogenation reaction are coupled and occur in the
Accepted 16 May 2012
exothermic and endothermic compartments, respectively. The dehydrogenated product (hydrogen) is
Available online 26 June 2012
pushed through the wall of the third partition, which is a Pd/Ag membrane composite, in order to
overcome the equilibrium constraints of dehydrogenation reaction. Moreover, the optimal operating
Keywords:
Thermally coupled heat exchanger reactor
conditions are sought by aid of differential evolution (DE) algorithm as a powerful optimization tech-
Pd/Ag membrane nique. During the optimization step, the sum of carbon monoxide and cyclohexane conversions along
Direct dimethyl ether (DME) synthesis with the hydrogen mole fraction in the permeation side is considered as the objective function. Finally,
Hydrogen production the TCMDR behavior is examined based on the achievements during the optimization procedure and
Differential evolution optimization a one-dimensional steady-state heterogeneous model. The results show considerable DME enhancement
approach in the TCMDR by 10.3% and 11.4% compared with the conventional direct DME synthesis reactor (CDR)
and thermally coupled DME reactor (TCDR) arrangements and at the same time the amount of endo-
thermic raw material drops about 120.3 kmol/h.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction 1.1. Direct DME synthesis

Energy is an indispensable element in our everyday life and With ever growing concerns on environmental pollution,
most of the energy we use nowadays comes from fossil fuels. energy security, and future oil supplies, the global community is
Natural gas is one of the major fossil energy sources and main seeking non-petroleum based alternative fuels, along with more
options for the use of natural gas are power generation and advanced energy technologies to increase the efficiency of energy
conversion to petrochemicals like ammonia, dimethyl ether (DME) use. The most promising alternative fuel will be the fuel that has the
and methanol. Since, air pollution is one of the most serious envi- greatest impact on society. The major impact areas include well-to-
ronmental challenges all over the world converting natural gas to wheel greenhouse gas emissions, non-petroleum feed stocks, well-
clean burning fuels like DME can be a helpful way to decline to-wheel efficiencies, fuel versatility, infrastructure, availability,
greenhouse-gas (GHG) emissions (Economides and Wood, 2009; economics, and safety. Compared to some of the other leading
Abbas and Wan Daud, 2010). alternative fuel candidates (i.e., methane, methanol, ethanol, and
The introduction section is divided into some subsections in FischereTropsch fuels), dimethyl ether appears to have the largest
order to give crucial orientations on various aspects of the present potential impact on society. Thus, it is one of the superior fuel
work. The first section provides an economical overview of direct candidates for the next generation and should be considered to
DME production rather its customary approach as well as a short eliminate the dependency on petroleum (Semelsberger et al., 2006;
summary about hydrogen as a green fuel. Secondly, the concept of Omata et al., 2002). Therefore, several studies have been done in
process integration and its alternatives are discussed. recent years concerning to increase and improve the productivity of
DME process (Liu et al., 2011; Khaleel, 2010; Jin et al., 2007; Wang
* Corresponding author. Tel.: þ98 711 2303071; fax: þ98 711 6287294. et al., 2009; Sai-Prasad et al., 2008). Broadly speaking, DME is
E-mail address: [email protected] (M.R. Rahimpour). conventionally produced by a two stage process. In the first step,

1875-5100/$ e see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jngse.2012.05.006
 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38 29

syngas undergoes methanol synthesis followed by a dehydrogena- UOP (Dautzenberg and Mukherjee, 2001) and BP (Ramaswamy,
tion plant to DME production. This is known as the indirect method 2006) for styrene and syngas productions respectively by
of DME synthesis. Recently, a new technique called direct DME coupling with combustion reactions.
synthesis, was proposed and gained much more attention due to its
economical superiority over the indirect method (Nie et al., 2005). 1.2.2. Membrane reactor
In this new process, the methanol production and dehydration step The potential of membranes for gas separation has been known
occur simultaneously on the hybrid (bi-functional) catalysts in one for more than 30 years (Baker, 2002). During this time, significant
bed and consequently the methanol purification unit can be development in membrane science has been come to stage from
ignored. Undoubtedly, the direct method has higher economic academic and industrial viewpoints (Bottino et al., 2002; Criscuoli
feasibility in comparison with the conventional illustration (indi- et al., 2001; Brunetti et al., 2007) and studies are still in progress.
rect method) since the production costs of the single-step process Two groups of polymeric and inorganic membranes are discussed,
are 20% lower compared with the two-step one. Indeed, the initial but the major investigations have been concentrated on the inor-
investment will be smaller (Fukunaga et al., 2008; Tan et al., 2005). ganic membrane reactors thanks to their excellent thermal stability
Hydrogen offers the best energy-to-weight ratio of any fuel. It is at high reaction temperature (Choi et al., 2000). It is commonly
not only an ideal energy carrier for the future mainly due to its high accepted that using membrane technology in the conventional
conversion efficiency, recyclability and nonpolluting nature but plants drives toward greater economic and environmental effi-
also a fundamental raw material and feedstock in petroleum, ciency (Sousa et al., 2001). Membrane reactors offer the possibility
chemical engineering, chemical fertilizer and metallurgical indus- to combine reaction and product separation in a single operation.
tries (Adris et al., 1997). Although hydrogen is often referred as an Therefore, the interest in such a configuration resides in their
clean energy that its combustion produces only water, the capability to control the reaction zone confinement and their
production of hydrogen from hydrocarbons by means of current selectivity. The membrane provides a selective removal of one or
processes yields CO2, a greenhouse gas (Wang et al., 2008; Kariya more products simultaneously with the reaction, so that the reac-
et al., 2003; Yolcular and Olgun, 2008). One of the environmen- tion equilibrium is continually shifted to increase the product
tally friendly methods to produce hydrogen without CO2-emission formation (Basile et al., 2001). There are a multitude of units and
is cyclohexane dehydrogenation (Biniwale et al., 2005). various hydrogenation and dehydrogenation reactions which need
to reach optimal productivity. One apparent opportunity that
1.2. Process intensification would seem to match well with the current feature of membrane is
the dehydrogenation of cyclohexane to benzene (Yang and Chou,
Process intensification is a concept that was introduced in the 2008). Here, membrane can be used effectively to increase
late 1970’s in the chemical industry for the purpose of reducing hydrogen production and benzene selectivity.
environmental emissions, energy and materials consumptions. In
practice, it is the established convention within the industry and it 1.3. Objectives
relates to reduction of at least 3e4 fold in magnitude. As far as
chemical reactors form the heart of any chemical processes, inno- Since DME and hydrogen are optimum large-scale fuels in the
vations in catalytic reactors are often the preferred starting point. near future, even minuscule improvements in their production
Over the last two decades, innovative multifunctional heat processes, which would be tantamount to higher production
exchanger and membrane reactors have been developed to inten- capacities, are economically favorable. Considering this, a multi-
sify chemical processes by synergistically combining chemical functional auto-thermal fixed-bed reactor for recuperative
reaction with heat and mass transport in a single vessel (Harmsen, coupling of single-step DME synthesis from syngas and cyclo-
2010; Van der Bruggen et al., 2004; Stankiewicz and Moulijn, 2004). hexane dehydrogenation reaction was proposed by Vakili et al.
(2011). Now, in a continuous effort we probe the effect of incor-
1.2.1. Thermally coupled reactor (heat exchanger reactor) porating membrane concept into this configuration. Furthermore,
Multifunctional reactors integrate, in one vessel, one or more differential evolution (DE) strategy is exploited to identify the most
transport processes and a reaction system and are increasingly used favorable operating conditions which enhance the desired
in industries as process intensification tools. A multifunctional production rates (hydrogen and DME production rates). Generally,
reactor can be used for coupling exothermic and endothermic
reactions. The exothermic reaction is utilized as the energy source
to drive the endothermic one. The coupled reactors are divided into Fresh water
three various configurations.

Syngas Steam drum

 Direct coupling (directly coupled adiabatic reactor).
 Regenerative coupling (reverse-flow reactor).
 Recuperative coupling (co and counter-current heat exchanger
reactors).

The latter configuration offers several advantages over the other

ones. First, the products of the endothermic reaction always sepa-
rate from the exothermic side. In addition, here the reactors have Natural Gas Reforming unit
more flexibility for their operational parameters in both sides
separately and can adjusted independently without affecting the
other streams (Ramaswamy, 2006). The researches were carried
out by Hunter and McGuire (1980), Van Sint Annaland and Nijssen
Pure DME
(2002), Elnashaie et al. (2000) and Altimari and Bildea (2009) are Distillation unit
appreciable in this regard. Among all of these laboratories scale
experiments, two commercial scale configurations also installed by Fig. 1. A schematic diagram of CDR.
30 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38

Table 1 Table 2
Operating conditions and catalyst characteristics for conventional DME reactor (Hu The operating conditions for endothermic and permeation sides of TCDR and
et al., 2008). TCMDR.

Parameter Value Unit Parameter Value

Feed composition (mole fraction) Endothermic side
CO 0.1716 e Feed composition (mole fraction)
CO2 0.0409 e C6H12 0.2
DME 0.0018 e C6H6 0.0
CH3OH 0.003 e H2 0.0
H2O 0.0002 e Ar 0.8
H2 0.4325 e Total molar flow rate (mol s 1) 0.12
N2 0.316 e Inlet pressurea (Pa) 1.013  105
CH4 0.044 e Inlet temperature (K) 493.0
Inlet temperature 493 K Particle diameterb (m) 3.55  10 3
Inlet pressure 50 bar Bed void fraction 0.39
Number of pipes 4177 e
Permeation side
Diameter of pipes f38  2 mm
Feed composition (mole fraction)
Volumetric flow rate of raw gas 2.04  105 Nm3 h 1
Ar 1.0
Length of reactor 5.8 m
H2 0.0
Temperature of boiling water 513 K
Total molar flow rate (mol s 1) 1.0
Thermal conductivity of wall 48 Wm 1K 1
Inlet pressure (bar) 0.1
Typical properties of catalyst a
Obtained from Jeong et al. (2004).
Particle diameter 5 mm b
Obtained from Koukou et al. (1997).
Density of catalyst bed 1200 kg/m3
Porosity 0.455 e

to others. A schematic diagram of CDR suggested by Hu et al. (2008)

for one-step DME synthesis is shown in Fig. 1. Here, the DME
the main objective of this work is optimization of a thermally
synthesis occurs in tube compartments over bi-functional catalyst
coupled membrane reactor for more hydrogen and DME
packs where methanol production and dehydration occur concur-
production.
rently. The boiling water in the shell side absorbs the heat of
reactions and produces water vapor. The characteristics and oper-
2. Process description ating conditions of this configuration are presented in Table 1.

2.1. Conventional direct DME synthesis reactor (CDR) 2.2. Thermally coupled DME reactor (TCDR)

Due to more available information on the design and In TCDR (Vakili et al., 2011), the direct DME synthesis takes place
manufacturing of fixed-bed reactors in industrial scales, the direct in the tube side and a catalytic dehydrogenation reaction (endo-
DME production is generally proposed in these reactors compared thermic reaction) is supposed for the shell side instead of using

CO,CO2,DME,CH3OH,H2O,H2 CO,CO2,DME,CH3OH, C6H12,H2

,N2,CH4 H2O,H2,N2,CH4
C6H12,H2
Ar
Pd-Ag MEMBRANE LAYER

Pd-Ag MEMBRANE LAYER

INSULATED WALL

INSULATED WALL
INSULATED WALL

INSULATED WALL

SOLID WALL

SOLID WALL
PERMEATION

PERMEATION
TRANSFER
SOLID WALL

SOLID WALL
TRANSFER

HEAT
H2

H2
HEAT

Exothermic side
Permeation side
(Direct DME synthesis)
Endothermic side Exothermic side Endothermic side
(Cyclohexane dehydrogenation) (Direct DME synthesis) (Cyclohexane dehydrogenation)

Fig. 2. A schematic diagram of TCDR. Fig. 3. A schematic diagram of TCMDR.

 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38 31

Table 3 the permeation of hydrogen shifts the thermodynamic equilibrium

Reaction rate constant, the adsorption equilibrium constant, and the reaction of endothermic reaction and sequentially increases the cyclo-
equilibrium constant for cyclohexane dehydrogenation.
hexane conversion. The specifications of the permeation side were
k ¼ AexpðB=RTÞ A B(K) provided in Table 2.
kðmol m3 pa 1s 1Þ 0.221 4270
KB ðpa 1 Þ 2.03  10 10 6270
KP ðpa3 Þ 4.89  1035 3190 3. Reactions scheme and kinetics expressions

3.1. Direct DME synthesis (exothermic side)

boiling water. The generated heat in the tube side is continuously DME synthesis from syngas is consists of three exothermic
transferred to the endothermic side (shell compartment) and reversible reactions. In this study, a bi-functional catalyst has been
drives the dehydrogenation reaction. As a result, DME and used which is made of the commercial catalysts of methanol
hydrogen are produced in a recuperative reactor by achieving auto- synthesis (CuO/ZnO/Al2O3) and methanol dehydration (g-Al2O3)
thermal condition. A conceptual schematic for TCDR in co-current with a mass ratio 1:1. The following reactions are considered as
mode is provided in Fig. 2. The corresponding operating condi- independent reactions in the direct synthesis of DME from syngas.
tions are clubbed into Table 2. It worth to mention that, the oper-
ating conditions of conventional process including the catalyst CO þ 2H2 4CH3 OH DH298K ¼ 90:55 kJ=mol (1)
load, inlet molar flow rate, inlet temperature of reacting materials
and etc. are considered as the basic principles for the design of CO2 þ 3H2 4CH3 OH þ H2 O DH298K ¼ 49:43 kJ=mol (2)
TCDR and achieving the endothermic side characteristics.
2CH3 OH4CH3 OCH3 þ H2 O DH298K ¼ 21:003 kJ=mol (3)
2.3. Thermally coupled membrane DME reactor (TCMDR)
The reaction rate expressions which have been selected from
Nie and Fang (2004) together with reaction constants have been
Innovations in catalytic reactor technologies, which somehow
reported in our previous work (Vakili et al., 2011).
could be the heart of chemical processes, are often the preferred
starting point to revamp and improve the operability. Hence,
a membrane feature of thermally coupled DME reactor is suggested 3.2. Dehydrogenation of cyclohexane (endothermic side)
here (see Fig. 3). Basically, the operation of TCMDR which is
composed of three concentric tube reactors is similar to that of Today, hydrogen carriers such as cyclohaxane are considered for
TCDR except with minor changes. As can be seen from Fig. 3, the hydrogen storage and transmission because of main advantages
DME synthesis and cyclohexane dehydrogenation take place in the from the following points of view (Itoh et al., 2003; Jain et al., 2010):
inner and the middle compartments, respectively. In this novel
configuration, the wall of endothermic side is coated by a PdeAg  Higher hydrogen content (e.g., 7.1 wt.% of cyclohexane) which
membrane to extract hydrogen from endothermic side into the is very attractive compared with metal hydrides (at most
third side (permeation side). According to Lechatelier’s principle, 3 wt.%).

Table 4
Mass and energy balance equations, pressure drop equation and corresponding boundary conditions.

Definition Equation
Exothermic and endothermic sides
Mass and energy balances for solid phase


av cj kgi;j ygi;j ysi;j þ hj ri;j rb ¼ 0 (6)

 N

av hf Tjg Tjs þ rb
X
hj ri;j DHf ;i ¼ 0 (7)
i 1
Mass and energy balances for fluid phase 1 dFi;j
 pDj
þ avj cj kgi;j ysi;j ygi;j b JH2 ¼ 0 (8)
AC dz AC


g
Cpgj d Fj Tj

g
 pDj g g pDj  g g
þ avj hf Tjs

Tj  U1 2 T2 T1 b U2 3 T2 T3 ¼ 0 (9)
AC dz AC AC

Pressure drop equation 2

dP 150m ð1 3Þ Q 1:75r ð1 3 Þ Q 2
¼ 2 þ (10)
dz fs d2p 33 Ac fs dp 33 A2c

Boundary conditions z ¼ 00Fi;j ¼ Fi0;j ; Tjg ¼ T0j

g
; Pjg ¼ P0j
g
(11)

Permeation side
Mass and energy balances dFi;j
þ 4pDj JH2 ¼ 0 (12)
dz


d Fj Tjg ZT2
g g  g g
Cpj þ pDj JH2 CpH2 dT þ pDj U2 3 T2 T3 ¼ 0 (13)
dz
T3

Boundary conditions z ¼ 00Fi;3 ¼ Fi0;3 ; T3g ¼ T0;3

g
(14)
32 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38

 More convenient for storage and transportation due to high Table 6

boiling point (bp ¼ 80.7  C) Comparison between simulation results and Hu’s model.

 The dehydrogenated products, benzene and toluene, can be Parameters Hu et al. (2008) Simulation result Relative error (%)
reversibly hydrogenated and reused and those are all liquids at Output components (mole fraction)
ordinary temperatures. DME 0.0491 0.05 1.8
 Its essentially zero CO2 impact, giving a positive environmental CO 0.0877 0.0903 2.8
CO2 0.0671 0.064 4.8
contribution and also solves the troubles and problems in
Outlet temperature (K) 516.6 519.5 0.56
hydrogen storage conditions and medium preparation.

The dehydrogenation of cyclohexane to benzene reaction is

represented as follow:  Heat loss to surrounding is neglected (insulated walls).

C6 H12 4C6 H6 þ 3H2 DH298K ¼ þ206:2 kJ=mol (4) In order to obtain the mass and energy balance equations,
a differential fragment is selected along the axial direction inside
The following equation is used for cyclohexane dehydrogena-
the reactor. After writing steady state balances, the equations are
tion rate, rC (Itoh, 1987):
divided by the length of the differential element, which is then


3
 approached zero. This work leads to a set of ordinary differential
k KP PC =PH 2
P B equations (ODEs). These governing equations as well as the related
rC ¼ (5)
boundary conditions are summarized in Table 4. The pressure drop


3
1 þ KB KP PC =PH 2
in fixed-bed reactor is calculated based on Ergun’s equation as is
where k, KB, Pi, and KP are the reaction rate constant, the adsorption presented by Eq. (10) (Fogler, 1992). In Eq. (9), positive and negative
equilibrium constant for benzene, the partial pressure in Pa, and signs are used for the exothermic and endothermic sides, respec-
the reaction equilibrium constant, respectively, which are clubbed tively. Moreover, b is equal to zero for exothermic side and is one for
into Table 3. Pt/Al2O3 pellets are utilized for this reaction. endothermic side. In the mass balance equation of the permeation
side, 4 is equal to 1 for hydrogen and zero for Argon. Auxiliary
correlations related to heat transfer and physical properties should
4. Mathematical modeling be added to solve the set of differential equations (see Table 5).
The composite membranes in this study are made of a 6 mM thin
4.1. Governing equations layer of palladiumesilver alloy. The membrane is deposited as
a continuous layer on the outer surface of a thermo stable support.
The following assumptions are considered for modeling of the The flux of hydrogen permeating through the outer Pd/Ag
proposed TCDR and TCMDR: membrane is assumed to follow Sievert’s law. The accuracy of this
model increases with temperature. According to the latest studies,
 One-dimensional heterogeneous model is considered for both reasonable results achieve when P20:5 P30:5 is below 600 Pa1/2.
endothermic and exothermic sides. Sievert’s law is expressed by:
 The reactors are simulated under the steady-state condition.
 Axial diffusion of heat and mass are neglected compared with EH2 ;p
 
the convection and bulk movement. Q0 Exp
RT
qffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi
 The radial diffusion in the catalyst particles is neglected in each JH2 ¼ PH2 ;2 PH2 ;3 (15)
dH2
side.
 Packed beds have symmetry in both axial and radial directions where PH2 and dH2 are hydrogen partial pressure in Pa and the
(bed porosity is constant). membrane thickness, respectively. The pre-exponential factor Q0 is
 The gas mixture is considered as an ideal gas in both catalytic taken to be 6:33  10 7 molm 1 s 1 Pa 0:5 and the activation
reactor sides. energy EH2 ;p is 15.7 kJmol 1 (Chen, 2004; Shu et al., 1994).

Table 5
Auxiliary correlations.

Parameter Equation Ref.

Component heat capacity Cp ¼ a þ bT þ cT 2 þ dT 2
Mixture heat capacity Based on local compositions
Viscosity of reaction mixtures Based on local compositions
Mixture thermal conductivity Lindsay and Bromley (1950)
0:42 0:67 u
Mass transfer coefficient between gas and solid phases kgi ¼ 1:17Re Sci g  103 Cussler (1984)
2Rp ug
Re ¼
m
m
Sci ¼ 4
rDim  10
1 y
Dim ¼ P yi Wilke (1949)
i
i ¼ j Dij qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:43  10 7 T 3=2 1=Mi þ 1=Mj
Dij ¼ pffiffiffi 1=3 1=3
Reid et al. (1997)
2Pðvci þ vcj Þ2

1 1 Ai lnðDo =Di Þ Ai 1
Overall heat transfer coefficient ¼ þ þ
U hi 2pLKw Ao ho
2=3 0:407
h Cp m 0:458 rudp
  
Heat transfer coefficient between gas phase and reactor wall ¼ Smith (1980)
Cp rm K 3B m
 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38 33

Table 7 1500
Optimization results of TCMDR, the applied strategy and parameters in DE
algorithm.

Parameter Value 1000

The optimization results
Initial molar flow rate of exothermic side (mol/s) 0.594
Initial molar flow rate of endothermic side (mol/s) 0.112 500
Exothermic inlet temperature (K) 513.4

Heat (W/m)
Endothermic inlet temperature (K) 513.4
Inlet temperature of sweep gas (K) 454.0 0

Strategy and parameters

Strategy DE/best/1/bin
Number of population (NP) 50 -500
Scaling factor (F) 0.8 Generated heat in the exothermic side
Crossover constant (CR) 0.8 Transferred heat from solid wall
-1000 Consumed heat in the endothermic side
Transferred heat from membrane surface
4.2. Numerical solution Net transferred heat to the endothermic side

-1500
In order to solve the set of differential equations, the related 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ODEs are converted to non-linear algebraic equations by backward Dimensionless length
finite difference method. The inlet operating conditions of three
Fig. 5. Variations of generated, consumed and transferred heat in OTCMDR.
sides are obvious. Hence, the GausseeNewton method in MATLAB
programming environment is used to solve this initial value h i
problem (IVP). The reactor length is divided to 100 sections (no grid ObjFun ¼ XCO þ Xcyc þ 10yH2;3 (16)
dependency is observed) and the solution for one node is used as
the initial conditions of the next one. The hydrogen mole fraction in above equation has been multi-
plied by ten to assimilate the order of magnitudes. Since, typical
4.3. Model validation optimization parameters are equipment size, recycle flows and
operating conditions like temperature, pressure and concentration;
In order to check the validity of the mathematical model and five decision variables are manipulated during the optimization
reaction rates under the simplified assumptions, the proposed CDR process. These variables are the inlet molar flow rates and inlet
by Hu et al. (2008) is modeled and the achieved results are temperatures of the exothermic and endothermic sides as well as
compared against our numerical predictions. The successful the inlet temperature of the permeation side. The variation ranges
achievements regarding this step are depicted in Table 6. of these variables are:

5. Optimization 0:05 < F0;exo < 1:5 mol=s (17)

5.1. Objective function & decision variables 0:05 < F0;endo < 1:5 mol=s (18)

The present study seeks to identify the optimum design of the 493 < T0;exo < 533 K (19)
proposed configuration and maximizing DME and benzene
productions as well as hydrogen recovery in TCMDR. Therefore, the
423 < T0;endo < 523 K (20)
sum of carbon monoxide and cyclohexane conversions and the
hydrogen mole fraction in the permeation side is considered as
objective function: 298 < T0;3 < 533 K (21)

a 535 b 520 520

CDR
530 TCDR
OTCMDR
Temperature of permeation side (K)

525
Endothermic temperature (K)

510 500
520
Temperature (K)

515
500 480
510

505

500 490 460

495
TCDR
OTCMDR
490 480 440
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionless Length Dimensionless length

Fig. 4. Temperature profiles of (a) exothermic side, (b) endothermic and permeation sides in CDR, TCDR and OTCDMR versus reactor dimensionless length.
34 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38

a 415 b 0.06
405 CDR
TCDR
0.05 OTCMDR
DME production (ton day-1)
390

DME mole fraction

375 0.04

360
0.03

345 0.052
0.02 0.05
330 0.048
0.046
0.01
315 0.96 0.98 1

300 0
CDR TCDR OTCMDR 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionless Length
c 0.18 d 0.075
CDR CDR
0.17 TCDR TCDR
OTCMDR 0.07 OTCMDR
0.16

CO2 mole fraction

0.15 0.065
CO mole fraction

0.14 0.06
0.13
0.055
0.12

0.11 0.05
0.1
0.045
0.09

0.08 0.04
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionless Length Dimensionless Length
-3
x 10
e 11 f 0.44
CDR
10 TCDR
OTCMDR
0.42
9
Methanol mole fraction

0.0108
8
H2 mole fraction

0.0106 0.4
0.34
7 0.0104
0.335

0.0102
0.38
6 0.9 0.95 1 0.33

0.325
5 0.36 0.96 0.98 1

4
CDR
0.34
TCDR
3 OTCMDR

2 0.32
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionless Length Dimensionless Length

g 0.04
CDR
0.035 TCDR
OTCMDR

0.03
H2O mole fraction

0.025 0.038

0.036
0.02
0.034

0.015 0.032
0.9 0.95 1

0.01

0.005

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Dimensionless Length

Fig. 6. (a) DME production, mole fraction of (b) DME, (c) CO, (d) CO2, (e) Methanol, (f) H2 and (g) H2O in various reactor.
 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38 35

5.2. Constraints 6.1. Thermal behavior

The temperature of exothermic side should be higher than the Fig. 4(a) illustrates the temperature profile of the exothermic
endothermic temperature to have an appropriate heat transfer side in CDR, TCDR and OTCDMR as a function of reactor dimen-
driving force. From another perspective, higher temperatures sionless length. The trend of the temperature profile variations in
endanger the catalyst longevity and aggravate the probability of the exothermic reactions is crucial in the reaction progress. Since, at
rapid catalyst deactivation due to sintering. Therefore, the the reactor entrance the reaction kinetic is predominant, temper-
maximum temperatures of the reaction sides are set to be lower ature surge (high temperature) is in favor of reaction rate and
than 533 and 523 in both sides, respectively (Nie and Fang, 2004; conversion enhancement. By the time, the thermodynamic equi-
Itoh, 1987). The following constraints are applied to control this librium will prevail and the temperature profile declines to
optimization problem: augment production yield. Based on the above descriptions, the
OTCMDR reminds the most favorable temperature history along the
Texo >Tendo (22) reactor length since it provides the highest temperature at the
reactor inlet and reduces it to the minimum amount at the outlet
max
Texo  533 (23) (Fig. 4(a)).
The temperature profiles along the endothermic and perme-
max
Tendo  523 (24) ation sides of TCDR and OTCMDR are compared in Fig. 4(b). Fig. 5
justifies the corresponding thermal behavior of OTCMDR. As can
out out
be seen, at the reactor entrance, the generated heat in the
FDME  FDME;CDR (25) exothermic side precedes the transferred heat from solid wall
which triggers to a hot spot. As the temperature goes up, the
FCout
6 H6
 FCout
6 H6 ;TCDR
(26) exothermic reaction reaches to its equilibrium and the generated
heat of exothermic side decreases compared with the transferred
The last two terms (Eqs. (25) and (26)) confirm DME and heat. As a result, the temperature reduces gradually. On the
benzene enhancements in the optimized thermally coupled contrary, the absolute heat which is transferred to the endothermic
membrane DME reactor (OTCMDR) compared with other configu- side (i.e. transferred heat from the exothermic and permeation
rations. The above constraints are considered in the optimization sides) is less compared with the consumed heat by the endo-
procedure using penalty function. The penalty function method thermic reaction. As a result a deep is created in its temperature
eliminates the unacceptable results. This method involves penal- profile (Fig. 4(b)). Afterward, the temperature amplifies up to
izing the objective function in proportion to the extent of constraint dimensionless length ¼ 0.45 and diminishes further.
violation (i.e., the penalty function takes a finite value when
a constraint violates and a value of zero when constraint is satis-
6.2. Molar behavior
fied) (Edgar et al., 2001). In this study, the penalty parameter is 108
however it may change from problem to problem. The ultimate
Generally, flow augmentation leads to residence time reduction
objective function for minimization is as follows:
and eventually lower extent of reaction. Surprisingly, the optimi-
3 zation procedure enables us to increases the exothermic feed molar
ObjFun þ 108 G2i flow rate of OTCMDR up to 8% compared with CDR and TCDR while
X
Result ¼ (27)
i¼1 the final DME mole fraction in the outlet stream remains constant.
The DME production capacity boosts up to 38.6 ton/day (10.3%) and
where:
42 ton/day (11.4%) compared with the CDR and TCDR, respectively.
Fig. 6(a) and (b) proves these facts. Fig. 6(c)e(g) examines the molar
G1 ¼ maxf0; ðTendo Texo Þg (28)
behavior of exothermic elements.
 max  Table 8 evaluates the total value of exothermic and endothermic
G2 ¼ max 0; Texo 533 (29) reaction rates in the TCDR and OTCMDR. As can be noticed, both
 max  exothermic and endothermic reaction rates are higher in the
G3 ¼ max 0; Tendo 523 (30) OTCMDR. This can be explained by the fact that the suggested
configuration under the optimized conditions provides a superior
n

out out
o temperature profile.
G4 ¼ max 0; FDME;CDR FDME (31)
In OTCMDR the initial molar flow rate of endothermic side
declines about 120.3 kmol/h compared to the TCDR which offers
a higher residence time. Implementing lower cyclohexane feed
n
o
G5 ¼ max 0; FCout
6 H6 ;TCDR
FCout
6 H6
(32)
flow rate plus integration of membrane concept to the reactor (to
More details about DE algorithm, its strategies and also adap- overcome the equilibrium constraints and prevail the hydrogen
tation of operating parameters have been presented in previous production rate) recover the cyclohexane conversion and benzene
works (Babu and Munawar, 2007; Babu and Angira, 2006). mole fraction about 8.3% and 58% (Fig. 7(a) and (b)). Generally, the
endothermic section of OTCMDR operates with the least feed flow

6. Results and discussion

Table 8
Table 7 lists the obtained results from optimization of TCMDR, Values of reaction rates in TCDR and OTCMDR.

the applied strategy and the employed parameters. The optimized Reaction rate (mol/s) TCDR OTCMDR
thermally coupled membrane reactor (OTCMDR) is simulated R1 (CO hydrogenation) 252.74 280.68
based on the optimum identified values and the performance R2 (CO2 hydrogenation) 29.58 31.11
of this optimized configuration is compared with CDR and R3 (methanol dehydration) 104.13 116.54
R4 (cyclohexane dehydrogenation) 95 96.35
TCDR one’s.
36 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38

rate and highest cyclohexane conversion while benzene production recovery and hydrogen mole fraction in the permeation side of
does not change in comparison with TCDR. These facts recognize OTCMDR. Hydrogen recovery is defined as:
that utilizing OTCMDR is beneficial and has more economical
feasibility due to higher production rates. Fig. 7(c) depicts hydrogen FH2 ;3
Hydrogen recovery ¼ (33)
FC6 H12 ;in

0.9
7. Conclusion
0.8 a
In our theoretical investigation the effect of membrane concept
0.7 into a thermally coupled heat exchanger reactor is examined for
hydrogen and direct DME production. Besides, the best configura-
Cyclohexane conversion

0.6 tion with the optimal geometric and operating variables is followed
in order to improve the reactor performance. Even minor
0.5 improvements in production efficiency may result in significant
profit increase, energy conservation and environmental protection,
0.4 especially for such large-scale units. The simulation results show
that the temperature profile of exothermic side in OTCMDR
0.3
improves in comparison with other configurations. The DME
production increases about 10.3% and 11.4% relative to CDR and
0.2
TCDR. Besides, cyclohexane conversion raises about 8.3% compared
0.1 TCDR with that of TCDR while the endothermic feed flow rate decreases
OTCMDR 120.3 kmol/h. Therefore, utilizing OTCMDR not only enhances the
0 net profit of plant owing to more DME production also reduces
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 operational costs owing to reduction in endothermic feed flow rate.
Dimensionless length As far as, this is a relatively new concept in Chemical engineering
0.18 and Reactor design, immediate needs for theoretical studies are
tangible. Mathematical modeling in conjunction with the optimi-
0.16 b zation strategy helps policy makers to identify promising technol-
ogies as well as the advantages and disadvantages of such
0.14
a configuration. In addition, it can contribute to considerable
Benzene mole fraction

savings in money and time during the expensive stage of pilot plant
0.12
development. However, the lab scale experiment along with the
0.1
optimal investigation is considered as the subsequent step in the
road toward successful commercialization of this configuration.
0.08
Notations
0.06

0.04
av specific surface area of catalyst pellet (m2 m 3)
AC cross section area of each tube (m2)
0.02 TCDR
OTCMDR Ai inside area of inner tube (m2)
0 Ao outside area of inner tube (m2)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 C total concentration (mol m 3)
Dimensionless length Cp specific heat of the gas at constant pressure (J mol 1)
0.06 3 CpH2 specific heat of hydrogen at constant pressure (J mol 1)
dp particle diameter (m)
c Di tube inside diameter (m)
0.05 2.5
Dij binary diffusion coefficient of component i in j (m2 s 1)
Dim diffusion coefficient of component i in the mixture
(m2 s 1)
Hydrogen recovery

0.04 2
H2 mole fraction

Do tube outside diameter (m)

fi partial fugacity of component i (bar)
0.03 1.5
F total molar flow rate (mol s 1)
G mass velocity (kg m 2 s 1)
0.02 1 hf gasesolid heat transfer coefficient (W m 2 K 1)
hi heat transfer coefficient between fluid phase and reactor
wall in exothermic side (W m 2 K 1)
0.01 0.5
ho heat transfer coefficient between fluid phase and reactor
wall in endothermic side (W m 2 K 1)
0 0 DHf,i enthalpy of formation of component i (J mol 1)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
JH2 hydrogen permeation flux (mol m 2 s 1)
Dimensionless length
ki reaction rate constants
Fig. 7. (a) Cyclohexane conversion and (b) benzene mole fraction in the endothermic kg mass transfer coefficient for component i (m s 1)
side, (c) hydrogen recovery and hydrogen mole fraction in the permeation sides. K conductivity of fluid phase (W m 1 K 1)
 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38 37

KB adsorption equilibrium constant for benzene (Pa 1) Babu, B.V., Munawar, S.A., 2007. Differential evolution strategies for optimal design
of shell-and-tube heat exchangers. Chem. Eng. Sci. 62, 3720e3739.
Kfi equilibrium constant of reaction i in DME synthesis
Basile, A., Chiappetta, G., Tosti, S., Violante, V., 2001. Experimental and simulation of
Ki adsorption equilibrium constant for component i in DME both Pd and Pd/Ag for a water gas shift membrane reactor. Sep. Purif. Technol.
synthesis reaction 25, 549e571.
Kp equilibrium constant for dehydrogenation reaction (Pa3) Biniwale, R.B., Kariya, N., Ichikawa, M., 2005. Dehydrogenation of cyclohexane over
Ni based catalysts supported on activated carbon using spray-pulsed reactor
Kpi equilibrium constant based on partial pressure for and enhancement in activity by addition of a small amount of Pt. Catal. Lett.
component i in methanol synthesis reaction 105, 83e87.
Kw thermal conductivity of reactor wall (W m 1 K 1) Bottino, A., Capannelli, G., Comite, A., 2002. Catalytic membrane reactors for the
oxidehydrogenation of propane: experimental and modelling study. J. Membr.
L reactor length (m) Sci. 197, 75e88.
Mi molecular weight of component i (g mol 1) Brunetti, A., Caravella, A., Barbieri, G., Drioli, E., 2007. Simulation study of water gas
N number of components (N ¼ 8 for DME synthesis shift reaction in a membrane reactor. J. Membr. Sci. 306, 329e340.
Chen, Z., 2004. A novel circulating fluidized bed membrane reformer for efficient
reaction, N ¼ 4 for cyclohexane dehydrogenation reaction, pure hydrogen production for fuel cells from higher hydrocarbons. Ph.D. thesis,
N ¼ 2 for third side) Auburn University.
P total pressure (for exothermic side: bar; for endothermic Choi, J.S., Song, I.K., Lee, W.Y., 2000. Performance of double-pipe membrane
reactor comprising heteropolyacid catalyst and polymer membrane for the
side: Pa) MTBE (methyl tert-butyl ether) decomposition. J. Membr. Sci. 166,
Pi partial pressure of component i (Pa) 159e175.
rCO rate of reaction for hydrogenation of CO (mol kg 1 s 1) Criscuoli, A., Basile, A., Drioli, E., Loiacono, O., 2001. An economic feasibility study
for water gas shift membrane reactor. J. Membr. Sci. 181, 21e27.
rCO2 rate of reaction for hydrogenation of CO2 (mol kg 1 s 1)
Cussler, E.L., 1984. Diffusion Mass Transfer in Fluid Systems, second ed. Cambridge
rDME rate of reaction for dehydration of methanol (mol University Press.
kg 1 s 1) Dautzenberg, F.M., Mukherjee, M., 2001. Process intensification using multifunc-
rC rate of reaction for dehydrogenation of cyclohexane tional reactors. Chem. Eng. Sci. 56, 251e267.
Economides, M.J., Wood, D.A., 2009. The state of natural gas. J. Nat. Gas Sci. Eng. 1,
(mol m 3 s 1) 1e13.
R universal gas constant (J mol 1 K 1) Edgar, T.F., Himmelblau, D.M., Lasdon, L.S., 2001. Optimization of Chemical
Rp particle radius (m) Processes. McGraw-Hill Chemical Engineering Series, second ed.
Elnashaie, S.S.E.H., Moustafa, T., Alsoudani, T., Elshishini, S.S., 2000. Modeling and
Re Reynolds number basic characteristics of novel integrated dehydrogenation-hydrogenation
Sci Schmidt number of component i membrane catalytic reactors. Comput. Chem. Eng. 24, 1293e1300.
T temperature (K) Fogler, H.S., 1992. Elements of Chemical Reaction Engineering, second ed. Prentice-
Hall, Englewood Cliffs NJ.
Tj,max maximum temperature in jth side Fukunaga, T., Ryumon, N., Shimazu, S., 2008. The influence of metals and acidic
ug linear velocity of fluid phase (m s 1) oxide species on the steam reforming of dimethyl ether (DME). Appl. Catal. A
U overall heat transfer coefficient between exothermic and 348, 193e200.
Harmsen, J., 2010. Process intensification in the petrochemicals industry: drivers
endothermic sides (W m 2 K 1) and hurdles for commercial implementation. Chem. Eng. Process 49, 70e73.
vi atomic diffusion volumes Hu, Y., Nie, Zh, Fang, D., 2008. Simulation and model design of pipe-shell reactor for
yi mole fraction of component i (mol mol 1) the direct synthesis of dimethyl ether from syngas. J. Nat. Gas. Chem. 17,
195e200.
Z axial reactor coordinate (m)
Hunter, I.B., McGuire, G., 1980. Method and apparatus for catalytic heat-exchange.
h catalyst effectiveness factor US Patent 4214867.
3B void fraction of catalytic bed Itoh, N., 1987. A membrane reactor using palladium. AIChE J. 33, 1576e1578.
3 catalyst void fraction Itoh, N., et al., 2003. Hydrogen recovery from cyclohexane as a chemical hydrogen
carrier using a palladium membrane reactor. Catal. Today 82, 119e125.
rb density of catalytic bed (kg m 3) Jain, I.P., Lal, C., Jain, A., 2010. Hydrogen storage in Mg: a most promising material.
rs density of catalyst (kg m 3) Int. J. Hydrogen Energy 35, 5133e5144.
m viscosity of fluid phase (kg m 1 s 1) Jeong, B.H., Sotowa, K.I., Kusakabe, K., 2004. Modeling of an FAU type zeolite
membrane reactor for the catalytic dehydrogenation of cyclohexane. Chem.
4s sphericity factor Eng. J. 103, 69e75.
Jin, D., Zhu, B., Hou, Zh., Fei, J., Lou, H., Zheng, X., 2007. Dimethyl ether synthesis via
methanol and syngas over rare earth metals modified zeolite Y and dual
Superscripts and subscripts CueMneZn catalysts. Fuel 86, 1707e1713.
I component i Kariya, N., Fukuoka, A., Utagawa, T., Sakuramoto, M., Goto, Y., Ichikawa, M., 2003.
Efficient hydrogen production using cyclohexane and decalin by pulse-spray
0 initial conditions
mode reactor with Pt catalysts. Appl. Catalyst A 247, 247e259.
J number of side (1: exothermic side, 2: endothermic side; Khaleel, A., 2010. Methanol dehydration to dimethyl ether over highly porous
3:permeation side) xerogel alumina catalyst: flow rate effect. Fuel Process Technol. 91, 1505e1509.
Koukou, M.K., Chaloulon, G., Papayannakos, N., Markatos, N.C., 1997. Mathematical
In inlet conditions
modeling of the performance of nonisothermal membrane reactors. Int. J. Heat
Out outlet condition Mass Trans. 40, 2407e2417.
G gas phase Lindsay, A.L., Bromley, L.A., 1950. Thermal conductivity of gas mixture. Ind. Eng.
S at catalyst surface Chem. 42, 1508e1510.
Liu, D., Yao, Ch., Zhang, J., Fang, D., Chen, D., 2011. Catalytic dehydration of methanol
Ss steady state condition to dimethyl ether over modified g-Al2O3 catalyst. Fuel 90, 1738e1742.
Nie, Zh, Fang, D., 2004. Global kinetics of direct synthesis of dimethyl ether from
syngas containing N2 over bifunctional mixed catalyst. J. East China Univ. Sci.
References Technol. 30, 370e375 (in Chinese).
Nie, Zh., Liu, H., Liu, D., Ying, W., Fang, D., 2005. Intrinsic kinetics of dimethyl ether
Abbas, H.F., Wan Daud, W.M.A., 2010. Hydrogen production by methane decom- synthesis from syngas. J. Nat. Gas. Chem. 14, 22e28.
position: a review. Int. J. Hydrogen Energy 35, 1160e1190. Omata, K., Watanabe, Y., Umegaki, T., Ishiguro, G., Yamada, M., 2002. Low-pressure
Adris, A.M., Lim, C.J., Grace, J.R., 1997. The fluidized-bed membrane reactor for steam DME synthesis with cu-bases hybrid catalyst using temperature-gradient
methane reforming: model verification and parametric study. Chem. Eng. Sci. reactor. Fuel 81, 1605e1609.
52, 1609e1622. Ramaswamy, R.C., 2006. Steady-state and dynamic reactor models for coupling
Altimari, P., Bildea, C.S., 2009. Integrated design and control of plant wide systems exothermic and endothermic reactions. Washington University, Ph.D thesis.
coupling exothermic and endothermic reactions. Comput. Chem. Eng. 33, Reid, R.C., Sherwood, T.K., Prausnitz, J., 1997. The Properties of Gases and Liquids,
911e923. third ed. McGraw-Hill, New York.
Baker, R.W., 2002. Future directions of membrane gas separation technology. Ind. Sai-Prasad, P.S., Bae, J.W., Kang, S.H., Lee, Y.J., Jun, K.W., 2008. Single-step synthesis
Eng. Chem. Res. 41, 1393e1411. of DME from syngas on CueZnOeAl2O3/zeolite bifunctional catalysts: the
Babu, B.V., Angira, R., 2006. Modified differential evolution (MDE) for optimization superiority of ferrierite over the other zeolites. Fuel Process Technol. 89,
of non-linear chemical processes. Comput. Chem. Eng. 30, 989e1002. 1281e1286.
38 R. Vakili et al. / Journal of Natural Gas Science and Engineering 9 (2012) 28e38

Semelsberger, T.A., Borup, R.L., Green, H.L., 2006. Dimethyl ether (DME) as an Van der Bruggen, B., Curcio, E., Drioli, E., 2004. Process intensification in the textile
alternative fuel. J. Power Source 156, 497e511. industry: the role of membrane technology. J. Environ. Manage. 73, 267e274.
Shu, G., Grandjean, B.P.A., Kaliaguine, S., 1994. Methane steam reforming in Van Sint Annaland, M., Nijssen, R.C., 2002. A novel reverse flow reactor coupling
symmetric Pd- and Pd-Ag/porous SS membrane reactor. Appl. Catalyst A 119, endothermic and exothermic reactions: an experimental study. Chem. Eng. Sci.
305e325. 57, 4967e4985.
Smith, J.M., 1980. Chemical Engineering Kinetics. McGraw-Hill, New York. Wang, B., Goodman, W.D., Froment, G.F., 2008. Kinetic modeling of pure hydrogen
Sousa, J.M., Cruz, P., Mendes, A., 2001. Modeling a catalytic polymeric non-porous production from decalin. J. Catal. 253, 229e238.
membrane reactor. J. Membr. Sci. 181, 241e252. Wang, D., Han, Y., Tan, Y., Tsubaki, N., 2009. Effect of H2O on Cu-based catalyst in
Stankiewicz, A., Moulijn, J.A., 2004. Re-engineering the Chemical Processing Plant, one-step slurry phase dimethyl ether synthesis. Fuel Process Technol. 90,
Process Intensification. New York, USA. 446e451.
Tan, Y., Xie, H., Cui, H., Han, Y., Zhong, B., 2005. Modification of cu-based methanol Wilke, C.R., 1949. Estimation of liquid diffusion coefficients. Chem. Eng. Prog. 5, 218.
synthesis catalyst for dimethyl ether synthesis from syngas in slurry phase. Yang, H.S., Chou, C.T., 2008. Non-isothermal simulation of cyclohexane dehydro-
Catal. Today 104, 1025e1029. genation in an inert membrane reactor with catalytic pellets in the feed-side
Vakili, R., Pourazadi, E., Setoodeh, P., Eslamloueyan, R., Rahimpour, M.R., 2011. Direct chamber. J. Chin. Inst. Chem. Eng. 39, 227e235.
dimethyl ether (DME) synthesis through a thermally coupled heat exchanger Yolcular, S., Olgun, O., 2008. Ni/Al2O3 catalysts and their activity in dehydrogenation
reactor. Appl. Energy 88, 1211e1223. of methylcyclohexane for hydrogen production. Catal. Today 138, 198e202.