University of Southern California Dissertations and Theses

Methanol synthesis in a contactor-type membrane reactor

(USC Thesis Other)

Methanol synthesis in a contactor-type membrane reactor

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Methanol Synthesis in a Contactor-Type
Membrane Reactor
By
Zhongtang Li

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFRONIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)

December 2017

i

ACKNOWLEDGMENTS
Firstly, I would like to express my sincere gratitude to my advisor Professor Theodore T.
Tsotsis for the continuous support of my Ph.D. study and related research, for his patience,
motivation, and immense knowledge. He is not only my advisor, but also a mentor in my life who
helped me get over all the challenges I faced during my graduate studies and research.
Besides my advisor, I would like to thank all the Mork Family Department of Chemical
Engineering and Material Sciences staff, my friends and colleagues during the years that I spend
at USC.
Last but not the least, I would like to thank my family: my parents and my relatives for
their endless love and support throughout the years of my graduate study.

ii

Table of Contents
1 Introduction ................................................................................................................. 1
1.1 Background and Motivation .................................................................................................1
1.2 Methanol Synthesis .............................................................................................................3
1.2.1 Methanol ............................................................................................................................. 3
1.2.2 Methanol Synthesis ............................................................................................................. 7
1.2.3 Review of Industrial MeS Processes .................................................................................... 9
1.3 Membrane Reactors .......................................................................................................... 16
1.3.1 Selective Membranes ........................................................................................................ 16
1.3.2 Membrane Reactors .......................................................................................................... 23
1.4 Objectives and Scope of the Work ...................................................................................... 29
1.5 References ......................................................................................................................... 30
2 Kinetic Studies ............................................................................................................ 36
2.1 Literature Survey ............................................................................................................... 36
2.2 Experimental Studies ......................................................................................................... 46
2.2.1 Experimental Set-Up .......................................................................................................... 46
2.2.2 Experimental Procedure .................................................................................................... 48
2.3 Parameter Estimation ........................................................................................................ 52
2.4 Results and Discussion ....................................................................................................... 53
iii

2.5 Conclusion ......................................................................................................................... 58
2.6 References ......................................................................................................................... 58
3 Methanol Synthesis in a Membrane Reactor ............................................................... 60
3.1 Membrane Reactors for Methanol Synthesis ...................................................................... 60
3.2 Membrane Reactor Experiment .......................................................................................... 67
3.2.1 Membrane Reactor Configuration..................................................................................... 67
3.2.2 Membrane Modification ................................................................................................... 68
3.2.3 Experimental Set-Up .......................................................................................................... 72
3.2.4 Experimental Procedure .................................................................................................... 75
3.3 Results and Discussion ....................................................................................................... 77
3.3.1 Effect of Temperature ....................................................................................................... 78
3.3.2 Effect of Pressure ............................................................................................................... 80
3.3.3 Effect of W/F ...................................................................................................................... 81
3.3.4 Effect of Sweep Liquid Flow Rate ...................................................................................... 82
3.4 Conclusion ......................................................................................................................... 83
3.5 References ......................................................................................................................... 83
4 Membrane Reactor Modeling ..................................................................................... 85
4.1 Scope of the Simulations .................................................................................................... 85
4.2 Model Equations ................................................................................................................ 86
iv

4.3 Results and Discussion ....................................................................................................... 99
4.4 Scale-up Studies ............................................................................................................... 110
4.5 Conclusion ....................................................................................................................... 112
4.6 Reference ........................................................................................................................ 112
5 Thesis Conclusion and Suggestions for Future work ................................................... 114

1

1 Introduction
1.1 Background and Motivation
With the constantly growing world population along with continued industrialization, one common
consequence is the increased global demand for energy. Up to now, fossil raw materials such as
natural gas, crude oil, coal and lignite (brown coal) have dominated the world’s energy supply
with a share of up to 80 %, mainly due to their high energy density, along with their good
transportability via pipelines.
These traditional energy resources are, however, limited. The fast-growing world population and
the booming economies of countries such as China and India also intensify the concerns regarding
the future availability of fossil raw materials. As an example, if the rate of the crude oil
consumption per person in China gradually reaches that in Germany (a total amount about 93
million tons as reported in 2010), half of the world’s crude oil production will be consumed by
China alone.
Another major concern in burning fossil fuels is the challenge of Carbon Dioxide (CO2) generation
as the by-product of the combustion reaction. CO2 is thought to be the key contributor to global
warming, and its emissions have to be reduced considerably in the future to avoid the potential
catastrophic consequences. Improving the efficiency of the current technologies used in coal-
burning power plants in order to minimize CO2 emissions is the conventional approach today
towards greener power generation processes. Many studies, however, report that this may not be
sufficient to mitigate global warming, and point out the need to transition from producing energy
using coal to utilizing biofuels and other renewable energy sources for that purpose.
2

As noted above, Green Energy, also known as Renewable Energy (RE), is one of the proposed
solutions to the dwindling fossil fuel resources and the global warming concerns. RE’s share of
the overall power production is projected to increases constantly over the next few decades; as an
example, a share of about 35 %, 50 % and 80 % of the market is predicted for Germany in 2020,
2030 and 2050, respectively [1]. Solar energy, biomass, wind and ocean wave power, geothermal
energy, nuclear energy as well as carbon dioxide conversion into fossil fuels can be considered as
the major potential sources for green energy production. A general overview on these new sources
of energy can be found, for example, in [2].
Converting carbon dioxide into various organic compounds provides a potential way to utilize a
waste material for the production of both fuels and valuable chemicals while simultaneously
decreasing the amount of global warming emissions to the environment. Using CO2 as a feedstock
for the production of chemicals, however, can only reduce the overall emissions up to 1 %, even
if a major part of all the chemicals used today in the world were to come from CO2 as a feedstock.
Using carbon dioxide as a feedstock for the production of fuels, on the other hand, is a much more
effective way to reduce its emissions, of up to around 10 %.
There are various potential ways of converting CO2, the by-product of the combustion of all fossil
materials, into fuels and/or chemicals. Hydrogenation to methanol, hydrogenation to synthesis gas,
reforming with methane, hydrogenation to methane and photocatalytic/electrocatalytic reduction
are a few of them, with the corresponding reactions 1-1 to 1-6 listed below:

𝐶𝑂
2
+ 3𝐻 2
↔ 𝐶 𝐻 3
𝑂𝐻 + 𝐻 2
(1-1)
𝐶𝑂
2
+ 𝐻 2
↔ 𝐶𝑂 + 𝐻 2
𝑂 (1-2)
3

𝐶𝑂
2
+ 𝐶 𝐻 4
↔ 2𝐶𝑂 + 2𝐻 2
(1-3)
𝐶 𝑂 2
+ 4𝐻 2
↔ 𝐶 𝐻 4
+ 2𝐻 2
𝑂 (1-4)
𝐶 𝑂 2
+ 𝐻 2
𝑂 ↔ 𝐶𝑂 + 𝐻 2
+ 𝑂 2
(1-5)
𝐶 𝑂 2
+ 𝐻 2
𝑂 ↔ −𝐶 𝐻 2
− +1.5𝑂 2
(1-6)
Each of the above pathways has its own challenges. For example, dry reforming (equation 1-3)
requires high temperatures and expensive catalysts. The electrocatalytic pathway (equation 1-5)
attracts many researchers today, but it is still in the fundamental stage and faces significant hurdles
towards commercialization. This work focuses specifically on the conversion of carbon dioxide
into methanol, via reaction (1-1).
1.2 Methanol Synthesis
1.2.1 Methanol
Methanol, with the chemical formula CH3OH (often abbreviated as MeOH) is among the most
important materials of the chemical industry. The global annual consumption of methanol is
around 53 million tons in 2011 [4]. In a recent report it was stated that, from 2009 to 2016, the
global methanol consumption had a 7.7 % compounded annual growth rate (CAGR), with China
currently sharing nearly half of the total global demand [5]. Figure. 1-1 shows, for example, the
methanol consumption in the production of fuel products in China from 2000 – 2016. Methanol
was first discovered as a by-product of the destructive distillation of wood, and hence is also
commonly known as wood alcohol, methyl alcohol, wood spirits or wood naphtha. Methanol with
the main physical properties listed in Table 1-1 [6], is a light, colorless, volatile and flammable
4

liquid with a smell very similar to ethanol, but unlike ethanol, methanol is highly toxic for human
consumption. It has numerous applications such as a feedstock for the production of a variety of
chemicals, as fuel for vehicles, as an energy carrier, etc.

Figure 1-1 China’s use of methanol in liquid fuels has grown rapidly since 2000

5

Table 1-1 Physical properties of methanol

There are already several techniques available in the marketplace to synthesize C2-C4 olefins,
aromatics, and other basic chemicals from methanol [8-10]. For example, in China, a plant with a
capacity of 500,000 ton has been built to convert methanol into olefins. Methanol is also an
important source of power as a fuel for compression-ignition (CI) engines, internal combustion
(IC) engines and fuel cells. It is also used as the basic raw material to produce derivatives, like Di-
Methyl Ether (DME), Methyl-Tert-Butyl Ether (MTBE) and Di-Methyl Carbonate (DMC) which
find applications as fuels and fuel additives [11-13]. Methanol can also be used as a carrier of
hydrogen in energy storage applications, as anti-freeze agent, solvent, etc.
6

Converting carbon dioxide into methanol is a key route towards greener energy production and as
a result, has attracted a lot of researchers’ attention. The concept of methanol economy shown in
Figure1-2 emphasizes the fact that MeOH fulfills almost all the requirements of an optimal raw
material.

Figure 1-2 Concept of methanol economy [2]
Methanol has an energy density of 22.7 MJ/kg, which is acceptable as an energy storage material.
[2]. This value is compared to the density of other common fuel sources in Figure 1-3. When used
as an energy carrier, being a liquid provides methanol with the advantage of much easier
transportation and storage, comparable to that of the conventional energy carriers such as gasoline,
diesel or kerosene. By comparison, methane and hydrogen, which are also considered as potential
candidates for energy storage, need to be stored at high pressure to improve their hydrogen storage
per unit volume. Although they have higher energy density than methanol, the cost of storage and
transportation is significant. Another advantage of methanol is it can be produced from carbon
dioxide, which can help solve the problems caused by the green-house effect. Thus, methanol has
7

the potential to serve as an alternative replacement for the limited fossil energy resources but with
less negative impacts on the environment.

Figure 1-3 Comparison of the energy densities of different chemical energy carriers on a volumetric and gravimetric scale [2]
1.2.2 Methanol Synthesis
Though methanol can be synthesized via several methods including from direct oxidation of
methane [6], the most common approach to generate methanol is catalytically from synthesis gas
(a mixture of CO, CO2 and hydrogen) produced from fossil resources, such as coal (via
gasification) and natural gas (whose primary component is methane) either via steam reforming
(equation 1-7) or partial oxidation (equation 1-8).
CH4 + H2O → CO + 3H2 ΔH
o
= 206 (kJ/mol) (1-7)
CH4 + ½O2 → CO + 2H2 ΔH
o
= -36 (kJ/mol) (1-8)
The reactions taking place during methanol synthesis are equation 1-1 and the additional two
reactions listed below (however, only two out of these three reactions are linearly independent).
8

CO + 2H2 → CH3OH ΔH
o
= -90.6 (kJ/mol) (1-9)
CO + H2O → CO2 + H2 ΔH
o
= -41.2 (kJ/mol) (1-10)
All these reactions are exothermic, as indicated by their standard heats of reaction ΔH
o
, and the
overall methanol synthesis reaction is associated with a volume decrease. Thus, a high pressure
and a low temperature are desired for the process to attain the highest conversion possible.
With the commercially available Cu-ZnO-Al2O3 catalyst, the selectivity to methanol is very high,
typically above 99 %. [6]. However, the production of by-products, most commonly higher
alcohols [14-16], esters [16,17,18], hydrocarbons [15,17,19], ethers (mainly dimethyl ether)
[17,20] and ketones [21] can be promoted by catalyst impurities such as iron, cobalt, nickel, alkali,
and the catalyst itself (Al2O3 also catalyzes DME formation). More details about the various
catalysts and their impact on the methanol synthesis are discussed in [22].
There are two different methanol synthesis (MeS) processes based on the catalyst used. The first
type is the so-called high-pressure MeS process that uses a ZnO/Cr2O3 catalyst. The conditions
applied during this process are, typically, a temperature range from 300-450
o
C and a pressure
range from 25-35 MPa. Due to the severe operating conditions required, this process was quickly
replaced by the so-called low-pressure MeS process following the development of a Cu-ZnO
catalyst. The Cu-ZnO catalyst was first used by BASF in the early 1920’s [23,24]. However, due
to its significant sensitivity towards sulfur (H2S concentrations lower than 0.1 ppm were required)
and halide impurities, the catalyst was first used industrially by Imperial Chemical Industries (ICI)
in 1966, around 40 years after the catalyst was first discovered. Since then, methanol production
underwent a revolution, leading to a rapid grow in the size of the commercial methanol synthesis
9

plants. A brief overview of the various catalysts used in today’s MeS industrial field is found in
Table 1-2:
Table 1-2 Summary of typical copper-containing catalysts for the low-pressure methanol synthesis

1.2.3 Review of Industrial MeS Processes
The commercial process scheme for methanol synthesis mainly includes three parts: synthesis gas
generation, methanol synthesis, and methanol separation. Currently, several companies produce
methanol at a large scale, with the market share of each depicted in Figure 1-4. The following
section briefly describes the main commercial processes for MeS named in Figure 1-4.

10

Figure 1-4 Market share of major providers of methanol
A schematic of the Johnson Matthey/Davy Process Technology (JM/DPT) process is shown in
Figure 1-5. In this process, the syngas is generated through a reaction between steam and oxygen
with a hydrocarbon; this step may involve the reforming of natural gas/naphtha (there are two
approaches available for the reforming step, Autothermal Reforming (ATR) using oxygen, and
conventional Steam Methane Reforming (SMR)) or the gasification of coal, biomass or coke. In
the next step, the methanol synthesis reaction takes place. To increase the yield of the reactor, a
recycle loop is utilized. Finally, a conventional distillation column is used for the purification step
[2].
11

Figure 1-5 A schematic of the Johnson Matthey/Davy Process Technology for low-pressure methanol synthesis [2]
Lurgi developed a quasi-isothermal process to generate syngas in the 1970’s. For their methanol
synthesis process, the basic steps are the same as the original ICI LPM process. The differences
are mainly in the reactor design and the heat removal method. The schematic of this process is
shown in Figure 1-6. As indicated in the schematic, a multi-tubular water-cooled reactor is used in
this process, in which the catalyst is placed in the tube side while the steam flows in the shell-side
of the reactor. In this design, the heat generated in the reaction side can be removed by the steam
in the shell side and then be utilized to generate high-pressure steam. The high-pressure steam can
either be used directly or be recycled back to the compressor. In this design, the catalyst is
maintained at lower temperatures, and thus its lifetime is increased while the amount of by-product
12

formation is decreased. Additionally, higher space-time yields are achieved in this configuration,
which reduces the amount of required catalyst.

Figure 1-6 Lurgi conventional methanol synthesis process. (Courtesy of Air Liquide Global E&C Solutions) [2]
The Haldor-Topsoe process also uses natural gas or associated gas as feedstock. In this
configuration, three adiabatic reactors with heat exchangers in between are implemented for the
methanol synthesis section. In this way, the heat generated by the exothermic reaction can be used
to heat-up the water saturators. The arrangement is shown in Figure 1-7.
13

Figure 1-7 A simplified process flow diagram of Haldor Topsoe A/S methanol synthesis process via two-step reforming. BFW
boiling feed water, NG natural gas, HP high pressure. [31]
Recently, new approaches to synthesize methanol have been studied by several research groups.
Selectivity problems, the extreme reaction conditions required or associated environmental issues,
however, have hindered these approaches from being scaled-up and commercialized. For example,
in one study, Zhang et al. [32] discussed an alternative process to generate methanol from the direct
oxidation of methane (equation 1-11 below) either in the liquid or the gas phase, using both
heterogeneous and homogeneous catalysts.
𝐶𝐻
4
+ 1/2𝑂 2
→ 𝐶𝐻
3
𝑂𝐻 (1-11)
Another proposed approach [33-36] is the hydrogenation of CO2 (equation 1-12) with the hydrogen
produced from methane in high-temperature pyrolysis (equation 1-13). The overall reaction is
expressed in equation 1-14.
𝐶𝑂
2
+ 6𝐻 2
→ 2𝐶𝐻
3
𝑂𝐻 + 2𝐻 2
𝑂 (1-12)
14

3𝐶𝐻
4
→ 3𝐶 + 6𝐻 2
(1-13)
3𝐶𝐻
4
+ 2𝐶𝑂
2
→ 2𝐶𝐻
3
𝑂𝐻 + 2𝐻 2
𝑂 + 3𝐶 (1-14)
Methanol can also be generated enzymatically from methane with the reaction 1-15 [37,38]:
𝑁𝐴𝐷𝐻 + 𝐶𝐻
4
+ 𝑂 2
→ 𝑁𝐴𝐷 +
+ 𝐶𝐻
3
𝑂𝐻 + 𝐻 2
𝑂
(1-15)
Lurgi has developed a process consisting of one adiabatic reactor and one isothermal reactor as
depicted in Figure 1-8. The adiabatic reactor is loaded with 200 kg of catalyst, which is operated
at 80 bar and 240 – 280
o
C. The water cooled isothermal reactor contains 800 kg of the same
catalyst and operates at approximately 78 bar and 270
o
C. The rate of synthesis gas supplied to the
adiabatic reactor is 11,000 Nm
3
/(h*m
3
catalyst) while the rate for the isothermal reactor is 12,000
Nm
3
/(h*m
3
catalyst). By doing this, the volume of the water-cooled reactor is reduced by 20 % as
compared to the systems where the make-up gas (MUG) is introduced to the methanol synthesis
loop directly, leading to a considerable saving in costs of reactor construction as well as of piping
[39].
15

Figure 1-8 Schematic of the Lurgi process for methanol synthesis based on CO2 and H2 [40]
The CAMERE process is another example of a reactor designed for carbon dioxide hydrogenation
to form methanol through the reverse-water-gas-shift reaction. This process is still at the stage of
a pilot-plant, funded by the Korean government. The process consists of a methanol formation
reactor followed by a RWGS reactor, connected with a water-removal unit, as shown in Figure 1-
9 [41]. In this process, the first reactor converts CO2 into CO and H2O via the RWGS reaction
(equation 1-10). The product stream containing CO, CO2, H2 and H2O is then freed from water,
and serves as the reactant stream in the methanol reactor. Recycled stream is used in both of the
reactors to improve the conversion rate. The pilot plant based on this design, has a capacity of 100
kg per day and the methanol selectivity is more than two times higher than the one in the methanol
synthesis by direct hydrogenation of CO2 in a single reactor.
16

Figure 1-9 schematic of the CAMERE process for methanol production [42]
1.3 Membrane Reactors
1.3.1 Selective Membranes
According to IUPAC [43], membranes are structures with lateral dimensions much greater than
their thickness in which mass transfer is driven by a gradient of pressure, concentration, electrical
potential, temperature, etc. The membrane is a structure that serves as a barrier that can selectively
remove some species while preventing other species from passing through. There are two main
factors that characterize the performance of a membrane, namely its permeation rate and its
selectivity. The permeation rate of a given species through a membrane is determined by its
permeance, which is defined as the amount (mass or molar) of the species passing through the
membrane per unit area per unit of time. The selectivity is defined as the ratio of the permeance of
a given species to that of the permeance of other species in the mixture to be separated. Generally,
a membrane with low permeance will have high selectivity and a membrane with high permeability
17

often shows not sufficient selectivity. Ideally, a membrane with both high selectivity and
permeability is preferred.
Membranes are widely used today for a variety of separations: the separation of mixtures of fluids
(vapor, gas, and miscible liquid like organic mixtures), solid/liquid and liquid/liquid dispersions,
and dissolved solids and solutes from liquids [44]. Membrane processes have attracted a lot of
research interest in the last couple of decades or so, as manifested from the number of scientific
publications on the topic, see Figure 1-10.

Figure 1-10 Number of publications as determined by using the keyword ‘membranes’ (Scopus database: www.scopus.com)
Membranes processes can be classified as Ultrafiltration (UF), Pervaporation (PV), Nanofiltration
(NF), Microfiltration (MF), Vapor Permeation (VP), Gas Permeation (GP) and Reverse Osmosis
(RO) processes, each process being summarized briefly below [44]:
1. UF: removal of macromolecules, colloids and ionic species from liquids
2. NF: the separation of miscible liquids
0
10000
20000
30000
40000
50000
60000
70000
80000
number of publications
year
18

3. MF: the filtration of micron and submicron size particulates from liquids and gases
4. VP: the selective separation of vapor mixtures
5. GP: the selective separation of gas mixtures
6. RO: the (virtually) complete removal of all dissolved and suspended material from water
or other solvents.
The processes described above are distinguished from each other mainly based on the molecular
size of the species involved, as is shown in Figure 1-11, which also determines the corresponding
pore size of the membranes employed for those processes, as described in Figure 1-12.

Figure 1-11 Molecular size involved in pressure driven membrane separation processes [45]
19

Figure 1-12 Membrane’s pore size for different separation process [45]
Membranes can also be classified based on their nature and geometry, as briefly summarized in
Figure 1-13.

Figure 1-13 Classification of membranes. Adapted from [46]
Based on their nature, membrane can be classified as biological and synthetic, differing completely
by structure and functionality [47]. Biological membranes are easy to manufacture but their
20

operation is limited within certain temperature (often below 100
o
C) and pH range. Further,
biological membranes are difficult to clean and are amenable to microbial attacks. Synthetic
membranes are divided into organic and inorganic types, differentiated by the degree of tolerance
to temperature of operation (organic membrane’s operation temperature is limited to less than 250
o
C, whereas inorganic membrane can stand temperature of up to 1000
o
C [48]). Inorganic
membranes can be either porous or dense. For porous membranes, based on their pore size, they
can be further classified as microporous (dp < 2 nm), mesoporous (2 nm < dp < 50 nm),
macroporous (dp > 50 nm)) and dense. Microporous membranes with pore sizes of molecular
dimensions (dp ≈ 0.5 nm) [49] are known as molecular sieve membranes. Dense membranes are
made of metals, e.g., Pd and its alloys and Ag and its alloys.
Among inorganic membranes, ceramic membranes have been around the longest. Their first
modern industrial application was in the separation of U-238 and U-235 isotopes for making
nuclear weapons and fuels in the 1940’s and 1950’s [50]. Due to the separation method applied
(forcing highly corrosive UF6 through semipermeable membranes), only oxides such as Al2O3,
TiO2 and ZrO2 could withstand the harsh environments. In 1962, Loeb and Sourirajan [51] came
up with the idea of dividing a membrane into a “skin” and a porous support layer. This innovation
motivated also the development of a new generation of ceramic membranes. In a multi-layer
ceramic membrane, the skin layer is the key part that determines the separation properties and the
support layer provides mechanical support and uninterrupted flux. Intermediated membrane layers
are often utilized to make it possible to deposit the top layers on the underlying supports [52].
Since the initial development of ceramic membranes, their application of ceramic has quickly
expanded to a wide range of fields such as the food and beverage industries [53,54], biotechnology
21

[55] and gas separation [56,57], although their use is typically limited to smaller installations. For
example, after the start of full-scale installations for water purification and wastewater treatment
in Japan, in 1998, Europe and the United State have adapted the idea recently. The application of
ceramic membranes water treatment applications has been performed by companies such as Orelis,
Veolia Water, Hyflux, Kubota, Atech Innovations, TAMI Industries, Metawater, Inoceramic
GmbH, Mitsuim, Meidensha, Pervatech, Jiangsu Jiuwu and Ceraver [58,59]. With their clear
advantages in chemical and thermal stability, higher flux, longer lifetime and higher recoveries,
ceramic membranes will be employed in more applications in the future (a full discussion on the
potential of ceramic membrane can be found in the book [60]). Recently, ceramic membranes have
also become an important component for the development of fuel cells and in the hydrogen
economy.
The various mechanisms via which molecules transport through membranes are shown in Figure
1-14.
22

Figure 1-14 Representation of some mass transport mechanism through membranes
Surface diffusion takes place when the molecules are adsorbed and transport on the pore surface.
Knudsen diffusion takes place when the average pore diameter of the membrane is smaller than
the mean free path of the diffusing molecules. Under this condition, collisions of the molecules
with the pore walls dominate transport. Molecular sieving takes place when membranes with very
small pores exclude some of the molecules in the mixture from passing through. Capillary
condensation occurs when a vapor in a mixture condenses in the membrane structure and thus
excludes other less volatile species from transporting through. Capillary condensation is more
23

important under high pressure and low temperature conditions for membranes with relatively small
pores and is due to capillary forces.
1.3.2 Membrane Reactors
A Membrane Reactor (MR) is a device that combines a chemical reaction with the separation of a
product or reactant. By having reaction and separation occurring simultaneously in the same unit
the MR typically attains higher efficiency, which results in increased productivity and easier
downstream processing.
Figure 1-15 shows the number of publications over the past 17 years on membrane reactors,
indicating that the topic is attracting growing research interest in recent years.

Figure 1-15 Number of publications by using the keyword ‘membrane reactors’ (Scopus database: www.scopus.com)
MR’s can be classified mainly into three different types: extractor-type MR, distributor-type MR,
and contactor-type MR. In the extractor-type membrane reactors, the function of the membrane is
to remove one or more products from the reactor zone. By doing so, the reaction is promoted and
0
200
400
600
800
1000
1200
Number of Papers
Years
24

a conversion beyond thermodynamic equilibrium may be reached. In the distributor-type MR, the
membranes are used to distribute more uniformly along the reactor length one or more reactants.
The key reason for doing so is to try to improve the selectivity, and also to potentially avoid the
presence of cold or hot spots. In a contactor-type MR, the membrane is typically catalytic, and
feeding the reactants from the opposite sides of the membrane helps to ensure effective contact
and to improve catalyst activity. Figure 1-16 shows the schematics of these three types of
membrane reactors.

Figure 1-16 Brief schematic of extractor, distributor, and contactor membrane reactor types for a general
reaction A + B ↔ C + D [45]
25

In conventional MR, the membrane and catalyst are both installed inside the same reactor (these
MR are also known as Integrated Membrane Reactor (IMR)), with the catalyst being placed either
in the shell side (Figure 1-17) or the tube side (Figure 1-18) of the membrane.

Figure 1-17 IMR conceptual layout with the catalyst in the shell side

Figure 1-18 IMR conceptual layout with the catalyst in the tube side
In some instances, the optimal conditions for the catalytic reaction are not compatible with the
membrane function. In such cases, one may employ a Staged Membrane Reactor (SMR), which
differs from the IMR in that the membrane is assembled externally. The SMR usually consists by
a series of modules. As an example, Figure 1-20 shows a reactor system that consists of two
reactors plus two separators. As an example, for the specific reaction A + B  C, the reactants
are fed into the first reactor and are partially converted into C. Then the exit mixture is fed into a
separation unit that has a selective membrane that removes C. The exit stream from the separator
is fed into the second reactor, then to the second separation module, etc. By employing the SMR,
26

the streams of pure product are collected from the permeate flow in the separation modules and a
higher conversion can be reached. In the SMR configuration one has the advantage of
independently setting and adjusting the operation conditions in each module to attain the optimal
conditions. On the other hand, the synergy between reaction and separation is much stronger in
the IMR.

Figure 1-19 Schematic of SMR configuration that has two reaction plus separation steps. [45]
Table 1-3 shows some of the reactions that have been studied to date in MR. Table 1-4 reports
some of the applications that are already implemented worldwide at the pre-commercial or
commercial stage. It should be noted that bioreactors for wastewater treatment are already
industrial, whereas other applications such as enzyme membrane reactors and MR for hydrogen
production are still at a pre-industrial phase.

27

Table 1-3 Selected publications on membrane reactor applications

28

Table 1-4 Selected applications of worldwide implemented membrane reactors

29

1.4 Objectives and Scope of the Work
The performance of the methanol synthesis reactor can be improved either by enhancing the
activity of the catalyst or by developing novel methods to carry out the reaction [83-87]. In our
work, we propose the utilization of a high-temperature membrane reactor with modified membrane
characteristics for CO2-based methanol synthesis. As a novel idea, we utilize a liquid sweep flow
in the permeate side of the membrane. The modified membrane serves as an interface contactor
between liquid in the tube side and the gas phase in the shell side.
To improve the performance of the MR, an inert sweep agent has to be selected which shows high
solubility towards MeOH while the solubility of the reactants in is relatively low. Additionally,
the boiling point of this sweep agent has to be relatively high for it to remain in the liquid phase
despite the high temperatures employed in the reactor. In this work, Tetraethylene Glycol Dimethyl
Ether (TGDE) has been chosen as the sweep agent that flows in the tube side of the reactor. When
the liquid swept is pumped through the membrane with the aid of a HPLC pump, it removes a
considerable amount of MeOH generated inside the reactor, thus allowing a conversion beyond
the equilibrium limitation.
This work starts with the study on the kinetics of the methanol synthesis from synthesis gas
(mixture of H2, CO and CO2) in a packed-bed reactor to determine the corresponding kinetic
parameter values. After generating the kinetic parameters, membrane reactor experiments are
performed which show improved performance compared to the PBR. In addition, a model of the
MR is developed and validated with the experimental data, and a process design based on the
model is carried out.
30

1.5 References
1. Pieper C, Rubel H (2012) Electricity storage: Making large-scale adoption of wind and solar
energies a reality:163-181
2. Bertau M, Offermanns H, Plass L, Schmidt F, Wernicke H (2014) Methanol: The Basic
Chemical and Energy Feedstock of the Future: Asinger's Vision Today. Springer Science &
Business Media,
3. Olah GA (2005) Beyond oil and gas: the methanol economy 44:2636-2639
4. Olah GA, Goeppert A, Prakash GS (2011) Beyond oil and gas: the methanol economy. John
Wiley & Sons,
5. http://www.prnewswire.com/news-releases/global-methanol-consumption-witnesses-growth-
of-77-cagr-from-2009-to-2016-300458026.html
6. English A, Rovner J, Brown J, Davies S, Updated by Staff (2000) Methanol. In: Anonymous
Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley & Sons, Inc.,
7. (1991) NFPA 325M: Fire Hazard Properties of Flammable Liquids, Gases, and Volatile Solids.
The Association,
8. Unneberg E, Kolboe S (1988) Formation of p-Xylene From Methanol Over H-ZSM-5 36:195-
199
9. Conte M, Lopez-Sanchez JA, He Q, Morgan DJ, Ryabenkova Y, Bartley JK, Carley AF, Taylor
SH, Kiely CJ, Khalid K (2012) Modified zeolite ZSM-5 for the methanol to aromatics reaction
2:105-112
10. Zeng D, Yang J, Wang J, Xu J, Yang Y, Ye C, Deng F (2007) Solid-state NMR studies of
methanol-to-aromatics reaction over silver exchanged HZSM-5 zeolite 98:214-219
11. Elvers B (2008) Handbook of fuels
12. (2010) World Butylenes Analysis
13. (2012) World Methanol Analysis
14. Smith KJ, Anderson RB (1984) A chain growth scheme for the higher alcohols synthesis
85:428-436
31

15. K. Klier in S. Kaliaguine, A. Mahay (1984) Catalysis on the Energy Scene, Studies in Surface
Science and Catalysis 19:439
16. Denise B, Sneeden R, Hamon C (1982) Hydrocondensation of carbon dioxide: IV 17:359-366
17. Büchele W, Roos H, Wanjek H, Müller H (1996) Catalyst research—One of the cornerstones
of modern chemical production 30:33-39
18. Denise B (1985) RPA Sneeden C 1 Mol 1:307
19. R. Bardet, J. Thivolle-Cazat, Y. Trambouze (1985) C1 Mol. Chem. 1:201
20. Matulewicz E (1984)
21. Elliott D, Pennella F (1989) The formation of ketones in the presence of carbon monoxide over
CuO/ZnO/Al2O3 119:359-367
22. Chinchen G, Spencer M (1988) A comparison of the water-gas shift reaction on chromia-
promoted magnetite and on supported copper catalysts 112:325-327
23. BASF (1923) DE 109495
24. BASF (1925) GB 237030
25. P. Courty et al. (1987) EP 0152314
26. M. Schneider, K. Kochloefl, J. Ladebeck (1987) Su¨d Chemie EP 0125689
27. E.F. Magoon (1973) US 3709919
28. J. Gallagher YHK (1965) GB 1159035
29. F.J. Bro¨cker et al. (1978) DE 2846614
30. Höppener R, Doesburg E, Scholten J (1986) Preparation and characterization of stable
copper/zinc oxide/alumina catalysts for methanol systhesis 25:109-119
31. Aasberg-Petersen K, Nielsen CS, Dybkjær I, Perregaard J (2008) Large scale methanol
production from natural gas 22
32. Zhang Q, He D, Zhu Q (2003) Recent progress in direct partial oxidation of methane to
methanol 12:81-89
33. T.V. Choudary, D.W. Goodman, Catalysis 19, 164–183 (2006)
32

34. H.F. Abbas, W.M.A. Wan Daud, Int. J. Hydrogen Energy 35, 1160–1190 (2010)

35. N. Muradov, Int. J. Hydrogen Energy 26, 1165–1175 (2001)
36. N. Muradov, T.N. Veziroglu, Int. J. Hydrogen Energy 30, 225–237 (2005
37. Merkx M, Kopp DA, Sazinsky MH, Blazyk JL, Müller J, Lippard SJ (2001) Aktivierung von
Disauerstoff und Hydroxylierung von Methan durch lösliche Methan‐Monooxygenase: eine
Geschichte von zwei Eisenatomen und drei Proteinen 113:2860-2888
38. Chan SI, Lu Y, Nagababu P, Maji S, Hung M, Lee MM, Hsu I, Minh PD, Lai JC, Ng KY
(2013) Efficient oxidation of methane to methanol by dioxygen mediated by tricopper clusters
52:3731-3735
39. Pruschek R, Oeljeklaus G, Haupt G, Zimmermann G, Jansen D, Ribberink J (1997) The role
of IGCC in CO 2 abatement 38:S153-S158
40. Gronemann V, Liebner W, di Zanno P, Pontzen F, Rothaemel M (2010) CO2 based methanol
ready for industrial scale 36:1-4
41. Park S, Chang J, Lee K (2004) Carbon Dioxide Utilization for Global Sustainability:
Proceedings of the 7th International Conference on Carbon Dioxide Utilization, Seoul, Korea,
October 12-16, 2003. Elsevier,
42. Yamaguchi T, Fukui H, Kobayashi M, Nakada N, Arai T (1998)
43. Koros W, Ma Y, Shimidzu T (1996) Terminology for membranes and membrane processes
(IUPAC Recommendations 1996) 68:1479-1489
44. Scott K (1995) Handbook of industrial membranes. Elsevier,
45. Basile A, De Falco M, Centi G, Iaquaniello G (2016) Membrane Reactor Engineering:
Applications for a Greener Process Industry. John Wiley & Sons,
46. Basile A, Iulianelli A, Tong J (2015) Single-stage hydrogen production and separation from
fossil fuels using micro-and macromembrane reactors:445-468
47. Xia Y, Lu Y, Kamata K, Gates B, Yin Y (2002) Macroporous materials containing three-
dimensionally periodic structures:69-100
48. van Veen HM, Bracht M, Hamoen E, Alderliesten PT (1996) Feasibility of the application of
porous inorganic gas separation membranes in some large-scale chemical processes 4:641-680
33

49. Verweij H (2012) Inorganic membranes 1:156-162
50. Gillot J (1991) The developing use of inorganic membranes: A historical perspective. In:
Anonymous Inorganic Membranes Synthesis, Characteristics and Applications, Springer, , pp 1-
9.
51. Loeb S, Sourirajan S (1962) Sea water demineralization by means of an osmotic membrane.
In: Anonymous ACS Publications,
52. Burggraaf AJaC,L (1996) Fundamentals of Inorganic Membrane Science and Technology
53. Van der Horst H, Hanemaaijer J (1990) Cross-flow microfiltration in the food industry. State
of the art. Desalination 77:235-258
54. Bolduan P, Latz M (2000) Ceramic membranes and their application in food ard beverage
processing. Filtration Sep 37:36-38
55. Shackleton R (1987) The application of ceramic membranes to the biological industries 37:67-
69
56. de Vos RM, Verweij H (1998) High-selectivity, high-flux silica membranes for gas separation.
Science 279:1710-1711
57. Aaron D, Tsouris C (2005) Separation of CO2 from flue gas: a review. Sep Sci Technol 40:321-
348
58. Mulder M (1996) Polarisation phenomena and membrane fouling. In: Anonymous Basic
Principles of Membrane Technology, Springer, , pp 416-464.
59. Baker RW (2007) Membrane technology and applications. 2004
60. Gitis V, Rothenberg G (2016) Ceramic membranes: new opportunities and practical
applications. John Wiley & Sons,
61. Tanaka K, Yoshikawa R, Ying C, Kita H, Okamoto K (2001) Application of zeolite membranes
to esterification reactions 67:121-125
62. Zhang W, Qing W, Chen N, Ren Z, Chen J, Sun W (2014) Enhancement of esterification
conversion using novel composite catalytically active pervaporation membranes. J Membr Sci
451:285-292
63. Molinari R, Caruso A, Poerio T (2009) Direct benzene conversion to phenol in a hybrid
photocatalytic membrane reactor 144:81-86
34

64. De Falco M, Piemonte V, Di Paola L, Basile A (2014) Methane membrane steam reforming:
heat duty assessment. Int J Hydrogen Energy 39:4761-4770
65. De Falco M, Iaquaniello G, Salladini A (2011) Experimental tests on steam reforming of
natural gas in a reformer and membrane modules (RMM) plant. J Membr Sci 368:264-274
66. Kikuchi E (2000) Membrane reactor application to hydrogen production 56:97-101
67. ITOH N, SHINDO Y, HARAYA K, HAKUTA T (1988) A membrane reactor using
microporous glass for shifting equilibrium of cyclohexane dehydrogenation. J Chem Eng Japan
21:399-404
68. Wu J, Gerdes T, Pszczolkowski J, Bhave R, Liu P, Martin E (1990) Dehydrogenation of
ethylbenzene to styrene using commercial ceramic membranes as reactors. Sep Sci Technol
25:1489-1510
69. Becker YL, Dixon AG, Moser WR, Ma YH (1993) Modelling of ethylbenzene
dehydrogenation in a catalytic membrane reactor. J Membr Sci 77:233-244
70. Sundmacher K, Rihko-Struckmann L, Galvita V (2005) Solid electrolyte membrane reactors:
Status and trends 104:185-199
71. Basile A, Drioli E, Santell F, Violante V, Capannelli G, Vitulli G (1996) A study on catalytic
membrane reactors for water gas shift reaction 10:53-61
72. Iyoha O, Enick R, Killmeyer R, Howard B, Morreale B, Ciocco M (2007) Wall-catalyzed
water-gas shift reaction in multi-tubular Pd and 80wt% Pd–20wt% Cu membrane reactors at
1173K. J Membr Sci 298:14-23
73. Battersby S, Duke MC, Liu S, Rudolph V, da Costa, Joao C Diniz (2008) Metal doped silica
membrane reactor: operational effects of reaction and permeation for the water gas shift reaction.
J Membr Sci 316:46-52
74. Akamatsu K, Nakane M, Sugawara T, Hattori T, Nakao S (2008) Development of a membrane
reactor for decomposing hydrogen sulfide into hydrogen using a high-performance amorphous
silica membrane. J Membr Sci 325:16-19
75. Iojoiu EE, Landrivon E, Raeder H, Torp EG, Miachon S, Dalmon J (2006) The “Watercatox”
process: Wet air oxidation of industrial effluents in a catalytic membrane reactor: First report on
contactor CMR up-scaling to pilot unit 118:246-252
76. Espro C, Arena F, Tasselli F, Regina A, Drioli E, Parmaliana A (2006) Selective oxidation of
propane on Nafion/PEEK-WC catalytic membranes in a multifunctional reaction system 118:253-
258
35

77. Gröschel L, Haidar R, Beyer A, Cölfen H, Frank B, Schomäcker R (2005) Hydrogenation of
propyne in palladium-containing polyacrylic acid membranes and its characterization. Ind Eng
Chem Res 44:9064-9070
78. Sznejer G, Sheintuch M (2004) Application of a carbon membrane reactor for dehydrogenation
reactions 59:2013-2021
79. Sakamoto Y, Chen F, Furukawa M, Noguchi M (1992) Permeability and diffusivity of
hydrogen in palladium-rich Pd-Y (Gd)-Ag ternary alloys. J Alloys Compounds 185:191-205
80. Veolia Water.
http://www.veoliawatertechnologies.co.za/medias/news/2011%E2%80%9306%E2%80%9330,B
ellville_WWTW_upgrade062011.htm
81. (2016) Citic Envirotech. www.unitedenvirotech.com/article/membraneproducts/ submerged‐
membrane‐modules/
82. (2016) Degremont/Suez (n.d.). www.degremont.com/en/know‐how/municipalwater‐
treatment/reuse/hollow‐fibre‐membrane‐bioreactor‐ultrafor‐ps/
83. Struis, Rudolf Paul Wilhelm Jozef, Stucki S, Wiedorn M (1996) A membrane reactor for
methanol synthesis. J Membr Sci 113:93-100
84. Westerterp KR, Kuczynski M, Bodewes TN, Vrijland MS (1989) Neue Konvertersysteme für
die Methanol‐Synthese 61:193-199
85. Wu J, Saito M, Takeuchi M, Watanabe T (2001) The stability of Cu/ZnO-based catalysts in
methanol synthesis from a CO 2-rich feed and from a CO-rich feed 218:235-240
86. Rahimpour M, Ghader S (2004) Enhancement of CO conversion in a novel Pd–Ag membrane
reactor for methanol synthesis 43:1181-1188
87. Naqvi M, Yan J, Dahlquist E (2012) Bio-refinery system in a pulp mill for methanol production
with comparison of pressurized black liquor gasification and dry gasification using direct
causticization. Appl Energy 90:24-31

36

2 Kinetic Studies
In order to develop and utilize a membrane reactor model to analyze the experimental data, one
must know the values of the kinetic parameters for the specific catalyst and conditions utilized. To
accomplish this, in this research a series of kinetic experiments were performed, which are
discussed in detail in this Chapter. Before presenting the experimental data and their analysis, first,
a comprehensive review of technical literature on the MeS kinetics is presented.
2.1 Literature Survey
The reactions involved in methanol synthesis have already been described in Chapter 1, and are
repeated for easy reference below. It should be noted that out of the three reactions, only two of
them are independent, and need to be considered for the design of the reactor.
CO + 2H2 → CH3OH ΔH
o
= -90.6 (kJ/mol) (1-9)
CO2 + 3H2 → CH3OH + H2O ΔH
o
= -49.5 (kJ/mol
(1-10)
CO + H2O → CO2 + H2 ΔH
o
= -41.2 (kJ/mol) (1-11)
As mentioned in Chapter 1, the history of the methanol synthesis can mainly be divided into two
stages: high-pressure processes and low-pressure processes. In the beginning of the field, the lower
activities of the ZnO/Cr2O3 catalyst employed required the reaction to be performed under high-
pressure conditions. With the development of the higher activity Cu/ZnO/Al 2O3 catalyst, the high-
pressure process was replaced by the low-pressure one. Therefore, the kinetic models described in
this section of the Thesis are primarily derived for the low-pressure methanol synthesis process.
37

The first kinetic model developed for the low-temperature MeS with a Cu/ZnO/Al2O3 catalyst was
the one developed by Lenov et al. [1], with the global rate expression given by equation 2-1 below.
In order to derive the rate equation, they assumed that the CO is the main source of carbon for
methanol production, thus neglecting the influence of CO2 in the feed.
𝒓 𝑪 𝑯 𝟑 𝑶𝑯
= 𝒌 (
𝑷 𝑪𝑶
𝟎 .𝟓 𝑷 𝑯 𝟐 𝑷 𝑪 𝑯 𝟑 𝑶𝑯
𝟎 .𝟔𝟔
−
𝑷 𝑪 𝑯 𝟑 𝑶𝑯
𝟎 .𝟑𝟒
𝑷 𝑪𝑶
𝟎 .𝟓 𝑷 𝑯 𝟐 𝑲 𝟐 ∗
) (2-1)
Klier et al. [2] studied the impact of CO2 in the feed gas on the rate of methanol synthesis. They
still treated CO as the major source of carbon and determined that there is an optimal ratio of
CO2/CO that maximizes the methanol synthesis rate. They concluded that in a system with a CO2
concentration between 0 and 10 %, the methanol is the only product, while at higher concentrations
methane is also produced as a by-product. They reported that with a syngas not containing CO2,
there is no surface formate formed and the methanol synthesis proceeds at a low rate, with
simultaneous irreversible catalyst deactivation. On the other hand, when the syngas contains a high
CO2 concentration, CO2 will block the active sites during formate generation, leading to a decrease
in the amount of produced methanol. For the best result, CO2 concentration should, therefore, be
controlled at an optimal level. Their rate equation is as follows:
𝑟 = 𝑘 𝐴 0
3
𝐾 ′3
(
𝑃 𝐶𝑂
2
𝑃 𝐶𝑂
)
3
[1+𝐾 ′
(
𝑃 𝐶𝑂
2
𝑃 𝐶𝑂
)]
3

𝐾 𝐶𝑂
𝐾 𝐻 2
2
(𝑃 𝐶𝑂
𝑃 𝐻 2
2
−
𝑃 𝑀𝑒𝑂𝐻 𝐾 𝑒𝑞
)
(1+𝐾 𝐶𝑂
𝑃 𝐶𝑂
+𝐾 𝐶𝑂
2
𝑃 𝐶𝑂
2
+𝐾 𝐻 2
𝑃 𝐻 2
)
3
+ 𝑘 ′
(𝑃 𝐶𝑂
2
−
1
𝐾 𝑒𝑞
′
𝑃 𝑀𝑒𝑂𝐻 𝑃 𝐻 2
𝑂 𝑃 𝐻 2
3
)
(2-2)
with the equilibrium constants 𝐾 𝑒𝑞
and 𝐾 𝑒𝑞
′
defined by the following equations 2-3 and 2-4.
𝐾 𝑒𝑞
=
𝑃 𝑀𝑒𝑂𝐻 ,𝑒𝑞
𝑃 𝐻 2 ,𝑒𝑞
2
𝑃 𝐶𝑂 ,𝑒𝑞
(2-3)
𝐾 𝑒𝑞
′
=
𝑃 𝑀𝑒𝑂𝐻 ,𝑒𝑞
𝑃 𝐻 2
𝑂 ,𝑒𝑞
𝑃 𝐻 2
𝑂 ,𝑒𝑞
3
𝑃 𝐶 𝑂 2
,𝑒𝑞
(2-4)
38

Villa et al. [3] also assumed CO as the only source of carbon in MeS, but they considered both the
methanol synthesis from H2 and CO and the reverse water gas shift reaction of H2 and CO2. Their
rate expressions are reported below:
𝑟 𝐶 𝐻 3
𝑂𝐻
=
𝑓 𝐶𝑂
𝑓 𝐻 2
2
− 𝑓 𝐶 𝐻 3
𝑂𝐻
/𝐾 2
∗
(𝐴 +𝐵 𝑓 𝐶𝑂
+𝐶 𝑓 𝐻 2
+𝐺 𝑓 𝐶 𝐻 3
𝑂𝐻
)
3

(2-5)
𝑟 𝑅𝑊𝐺𝑆 =
𝑓 𝐶𝑂
2
𝑓 𝐻 2
−𝑓 𝐶𝑂
𝑓 𝐻 2
𝐾 3
∗

(𝑀 )
2

(2-6)
Graaf et al. [4] performed a series of kinetic experiments using a mixture of CO, CO 2 and H2 as
feed over a commercial Cu-Zn-Al catalyst in a temperature range from 210 to 245 ℃. They
determined that methanol can be formed from both CO and CO2 simultaneously. A dual-site
Langmuir-Hinshelwood mechanism was used to describe their results as in equations 2-7 to 2-9,
which fitted their experimental data better than the model proposed by Villa et al. [3].
𝑟 𝐶𝐻
3
𝑂𝐻 ,1
=
𝑘 𝑝𝑠 ,1𝑐 ′
𝐾 𝐶 𝑂 2
[𝑓 𝐶 𝑂 2
𝑓 𝐻 2
3/2
−𝑓 𝐶𝐻
3
𝑂𝐻
𝑓 𝐻 2
𝑂 /(𝑓 𝐻 2
3/2
𝐾 2
∗
)]
(1+𝐾 𝐶𝑂
𝑓 𝐶𝑂
+𝐾 𝐶 𝑂 2
𝑓 𝐶 𝑂 2
)[𝑓 𝐻 2
1/2
+(𝐾 𝐻 2
𝑂 /𝐾 𝐻 2
1/2
)𝑓 𝐻 2
𝑂
(2-7)
𝑟 𝐶𝐻
3
𝑂𝐻 ,2
=
𝑘 𝑝𝑠 ,2𝑐 ′
𝐾 𝐶𝑂
[𝑓 𝐶𝑂
𝑓 𝐻 2
3/2
−𝑓 𝐶𝐻
3
𝑂𝐻
/(𝑓 𝐻 2
1/2
𝐾 2
∗
)]
(1+𝐾 𝐶𝑂
𝑓 𝐶𝑂
+𝐾 𝐶 𝑂 2
𝑓 𝐶 𝑂 2
)[𝑓 𝐻 2
1/2
+(𝐾 𝐻 2
𝑂 /𝐾 𝐻 2
1/2
)𝑓 𝐻 2
𝑂 (2-8)
𝑟 𝑅𝑊𝐺𝑆 =
𝑘 𝑝𝑠 ,1𝑏 ′
𝐾 𝐶 𝑂 2
[𝑓 𝐶 𝑂 2
𝑓 𝐻 2
− 𝑓 𝐶𝑂
𝑓 𝐻 2
𝑂 𝐾 3
∗
)]
(1 + 𝐾 𝐶𝑂
𝑓 𝐶𝑂
+ 𝐾 𝐶 𝑂 2
𝑓 𝐶 𝑂 2
)[𝑓 𝐻 2
1/2
+ (𝐾 𝐻 2
𝑂 /𝐾 𝐻 2
1/2
)𝑓 𝐻 2
𝑂
(2-9)
Rozovskii et al. [5] performed kinetic experiments at relatively short contact times as well as tracer
investigations to study the fundamentals of methanol synthesis and decomposition over Cu-based
catalysts. They reported an alternative pathway of CO→CO2→CH3OH when the reaction is
performed in a temperature range of 180 - 260
o
C, and at pressures from 0.1 - 20 MPa. They
provided a mechanism for the reaction which is shown in Figure 2-1. Based on this mechanism,
they proposed the reaction rate in equation 2-10 in which ki and Ki are rate and surface equilibrium
39

constants, Kp(m) is the methanol synthesis equilibrium constant, and Pi the partial pressures of each
component.
𝑟 =
𝑘 1
𝑃 𝐻 2
(1−
𝑃 𝑚 𝑃 𝐻 2
𝑂 𝐾 𝑝 (𝑚 )
𝑃 𝐻 2
3
𝑃 𝐶𝑂
2
)
1+𝐾 2
𝑃 𝐻 2
𝑂 +
𝐾 2
𝑃 𝐻 2
𝑂 (𝐾 1
𝑃 𝐶𝑂
2
)
⁄

(2-10)

Figure 2-1 . Mechanism for methanol synthesis proposed by Rozovskii et al. [5]
Park et al. [6] developed a kinetic model for methanol synthesis based on a three-site adsorption
theory, taking both the hydrogenation of CO and CO2 into consideration. They used 118
experimental data points to fit the parameters in the rate equation. This work considered DME as
a by-product of the methanol synthesis, and consequently the reactions involved are modified as
shown below:
CO + 2H 2 → CH 3OH (1-9)
CO 2 + 3H 2 → CH 3OH + H 2O (1-10)
CO + H 2O → CO 2 + H 2 (1-11)
2CH 3OH → CH 3OCH 3+ H 2O (2-11)
The reported rate expressions in this work are presented in equations 2-12 to 2-15.
40

𝑟 𝑊𝐺𝑆 =
𝑘 𝐵 ′
𝐾 𝐶𝑂
2
(𝑓 𝐶𝑂
2
𝑓 𝐻 2
−
𝑓 𝐶𝑂
𝑓 𝐻 2
𝑂 𝑓 𝐻 2
𝑂 )
(1+𝐾 𝐶𝑂
2
𝑓 𝐶𝑂
2
)(1+𝐾 𝐻 2
0.5
𝑓 𝐻 2
0.5
+𝐾 𝐻 2
𝑂 𝑓 𝐻 2
𝑂
(2-12)

(2-13)
(2-14)

(2-15)
where Ci (mol/m
3
) stands for the concentration of component i (The reaction rate expression of
DME was directly taken from [2]). The estimated kinetic parameters for the above rate equations
are listed in Table 2-1.
Table 2-1 Values of the kinetic parameters [6]

Table 2-2 briefly summarizes some of the key studies on the kinetics of the methanol production
and the reported rate expressions.
Table 2-2 Reported rate expressions for Methanol Synthesis
Catalyst T ( ◦C)
P
(atm)
Kinetic Equation Reference
41

Cu/ZnO/Cr 2O 3
300-
330
200-
315
Natta (1955)
Cu/ZnO/Al 2O 3
300-
330
200-
315

Pasquon
(1960)
Cu/ZnO/Al 2O 3
220-
260
40-
55

Leonov
(1973)
Cu-Zn
225-
250
75

Klier et al
(1982)
Cu-Zn-Al
215-
245
30-
95

Villa et al.
(1985)
Cu-Zn
235-
265
80-
140

Seyfert -
Luft (1985)
Cu-Zn-Al
Cu-Zn-Cr
N/A N/A
Dybkjaer
(1985)
Cu-Zn-Al

210-
245
15-
50

Graaf et al
(1988)
23
3
23
2*
2
3
/
()
CO H CH OH
CH OH
CO H CH OH
f f f K
r
A Bf Cf Df


  
23
2
23
2
1/2 3
( / )
(1 )
()
H CO CH OH eq
H
CO H CH OH
f f f K
r Kf
A Bf Cf Df


  
3 2
2
3
1
0.5
0.5 *
2
()
CH OH
CH OH
b
CO H
b
CO H
P
PP
rk
P P P K


2
23
3
22
2 2 2
3
3 3 2 *
2
3
2
*3
1
( / ) ( / )
[1 ( / )] ( )
'( (1/ )( / ))
H CH OH
CH OH
redox CO CO CO
CH OH n
redox CO CO CO CO
CO H O H
K P P P P P K
r const
K P P F K P
k P K P P P




23
3
23
2 2 2
2*
2
3
*
3
2
/
()
CO H CH OH
CH OH
CO H CH OH
CO H CO H O
RWGS
f f f K
r
A Bf Cf Gf
f f f f K
r
M


  


23
2 3 2 2
2
1
2
1 2 3 4 5 6
/
()
o
CO H CH OH P
CO H CH OH H CO CO
f f f K
r
A A f A f A f A f f A f


    
2 2 3 2 2 3
2 2 2 2
1/2 2
1 2 3
1/2 1/2 1/2
2 3 4 3
[ / ( )]
(1 )[1 / ( ) ]
o
CO H CH OH H O H P
CO H H O H
A A A f f f f f K
r
A f A f A f A f


  
2 2 2 3 2 2
3
2 2 2 2 2 2
2 3 2
3
2 2 2 2 2 2
' 3/2 3/2 *
,1 2
,1 1/2 1/2
' 3/2 1/2 *
,2 2
,2 1/2 1/2
[ / ( )]
(1 )[ ( / ) ]
[ / ( )]
(1 )[ ( / ) ]
ps c CO CO H CH OH H O H
CH OH
CO CO CO CO H H O H H O
ps c CO CO H CH OH H
CH OH
CO CO CO CO H H O H H O
RWGS
k K f f f f f K
r
K f K f f K K f
k K f f f f K
r
K f K f f K K f
r


  


  

2 2 2 2
2 2 2 2 2 2
'*
,1 3
1/2 1/2
[]
(1 )[ ( / ) ]
ps b CO CO H CO H O
CO CO CO CO H H O H H O
k K f f f f K
K f K f f K K f

  
42

ZnO-Al 2O 3
200-
240
50-
70

Skrzypek
(1991)
Cu-ZnO-Al 2O 3
180-
280
15-
51

Van den
Bussche et
al. (1996)
Cu/ZnO/ZrO 2/
Al 2O 3/SiO 2
200-
275

r M=k M[P(CO 2)P(H 2)-P(CH 3OH)P(H 2O)/[K MP
2
(H 2)]}/A
2

r R = k R{P(CO 2)-P(CO)P(H 2O)/[K RP(H 2)]}/A
A= 1+K H2P H21/2+K CO2P H2O+K H2CO2P H2(1/2)P CO2+K H2OHP H2O/P H2+K H2OP H2O
Kubota
(2001)
Cu-based
Zn-Croxide
240 52

Rozovskii
(2003)
Cu/Zn/Al/Zr
Fibrous
230-
260
25-
48

Xin et al.
(2009)
Cu/ZnO/Al2O3
473.15–
573.15
50-
100

M. Peter et
al. (2012)
2 2 2 2
3
22
2 2 2 2
33
2 2 2 2
3
22
2 2 2 2 2 2
33
2
1
2
11 3
2
1
2 2
( ) (1/ )( / )
[]
(1 )
( ) (1/ )( / )
2 [ ]
(1 )
CH OH
CH OH CH OH
CH OH
CH OH CH OH
CO H p H O H
H CO
H CO H O H O CO CO
CO H p H O H
H CO
H H H O H O CO CO
P P K P P P
r k K K
K P K P K P K P
P P K P P P
r k K K
K P K P K P K P


   


   
2 2 2 2 2
3
2 2 2 2 2 2
22
2 2 2 2
2 2 2 2 2 2
22
' ' * 3
5 2 3 4
3
34
'
13
34
[1 (1/ )( / )]
(1 ( / )( / ) )
[1 ( )( / )]
(1 ( / )( / ) )
CH OH
H O H O
H O H O
a CO H H O H CO
MeOH
H H H H H O H O
CO CO H O H CO
RWGS
H H H H H O H O
k K K K P P K P P P P
r
K K K K P P K P K P
k P K P P P P
r
K K K K P P K P K P


  


  
2
2
22
2 2 2
3 3
()
2 2 1
(1 )
1 / ( )
m H O
H
p m H CO
H O H O CO
PP
kP
K P P
r
K P K P K P




2 3 2
2 2 2 2 2 2
2 2 2 2
2 2 2 2 2 2
2 2 3 2 2
3/2 1/2
11
1 1/2 1/2
22
2 1/2 1/2
3/2 3/2
2
3
[ / ( )]
(1 )[ ( / ) ]
[ / )]
(1 )[ ( / ) ]
3 [ / (
CO CO H CH OH H f
CO CO CO CO H H O H H O
CO CO H H O CO f
CO CO CO CO H H O H H O
CO CO H CH OH H O H f
k K f f f f K
r
K f K f f K K f
k K f f f f K
r
K f K f f K K f
k K f f f f f K
r


  


  


2 2 2 2 2 2
3
1/2 1/2
)]
(1 )[ ( / ) ]
CO CO CO CO H H O H H O
K f K f f K K f   
43

Cu/ZnO/Al2O3
503-
613
50-
90

N. Park et al.
(2014)

In our work, the kinetic model developed by Van den Bussche et al. [7] has been used for the
analysis of the experimental data. Van den Bussche et al. [7] developed a steady-state kinetic
model for both the methanol synthesis and the water gas shift reaction on a commercial
Cu/ZnO/Al2O3 catalyst. They performed their experiments in a temperature range from 180 to 280
o
C and pressures of up to 51 bar. The authors proposed the rate expressions in equations 2-16 and
2-17 which are derived based on the assumption that CO2 is the main source of carbon in the
methanol synthesis, based on the prior studies of Rozovskii [5] and Chinchen et al. [8]:

(2-16)

(2-17)

The values for the thermodynamic equilibrium constants K1
*
and K3
*
were taken from Graaf et al.
[4], as in equations 2-18 and 2-19 below.

(2-18)
44

(2-19)
A total of 276 experiments were performed by Van den Bussche et al. [7] to determine the kinetic
parameters, which are described by equation 2-20. In this equation, Tav has the value of 501.57 K.
Then the values are calculated for the equation form k(i)= A(i) exp(B(i)/RT) and the values for
A(i) and B(i) are listed in Table 2-3.
K
(i)
= A(i)∗ exp(−
B(i)
R
(
1
T
AV
−
1
T
)) (2-20)
Table 2-3 Parameter values used in the rate equations of Van den Bussche et al. [7]. The groupings of parameters were
expressed in the form k(i)= A(i)exp(B(i)/RT).
√𝑲 𝑯 𝟐
A
B
0.499
17,197
𝑲 𝑯 𝟐 𝑶
A
B
6.62×10
-11

124,119
𝑲 𝑯 𝟐 𝑶 𝑲 𝟖 𝑲 𝟗 𝑲 𝑯 𝟐
A
B
3,453.38
—
𝒌 𝟓𝒂
′
𝑲 𝟐 ′
𝑲 𝟑
𝑲 𝟒 𝑲 𝑯 𝟐
A
B
1.07
36,696
𝒌 𝟏 ′

A
B
1.22×10
10

-94,765
45

Figure 2-2 Reaction mechanism
Figure 2-2 shows the reaction mechanism used for methanol synthesis by Van den Bussche et al.
[7]. As is shown on the left side of the figure, both CO2 and H2 adsorb on the copper surface
dissociatively. The surface oxygen or alkaline species promote the oxidizing adsorption of CO 2.
Carbonate structures are then formed on the oxidized copper surface by a further adsorption of
CO2 [9-11]. These carbonates are quickly hydrogenated, first to bicarbonate structures and
subsequently to Cu formate, formaldehyde, methoxy species, and finally methanol [7]. The rate
determining step is considered to be the hydrogenation of the formate species, which is generally
accepted to be the longest living intermediate in methanol synthesis on copper [12-14]. The
hydrogenation of CO2 to methanol will release surface oxygen, which is later hydrogenated by the
available hydrogen atoms, leading to the generation of hydroxyl groups and subsequently water.
This process is shown on the right hand side of Figure 2-2, and it is actually the reverse water gas
shift reaction. The rate-limiting step is the adsorption of CO2 onto the copper surface [15-17].
46

2.2 Experimental Studies
2.2.1 Experimental Set-Up
The schematic of our lab-scale experimental set-up used for the kinetic experiments is shown in
Figure 2-3.

Figure 2-3 Schematic of the experimental set-up used for the kinetic experiments
This set-up consists of three key parts: the gas delivery system, the reactor module, and the gas
analysis system. The inlet feed-stream consists of CO, CO2 (in our case a pre-mixed CO/CO2
cylinder is used. The cylinder is ordered from Matheson), H2 and N2, and is delivered to the reactor
from gas cylinders, the flow rates being controlled through individual Mass Flow Controllers
(MFC, Brooks 5400). The gases, after being mixed, they are first pre-heated using heating tapes
and then fed into the MeS reactor, which operates in a pressure range of 20-30 bar and a
temperature range of 200-240 °C. Three ceramic band-heaters are used to provide the reactor with
the required heat and to maintain the reactor at a constant temperature (further details shown in
Figure 2-4). Each heater is controlled by an individual PID controller. In order to monitor the
47

temperatures, a three-point thermocouple from OMEGA is installed in a thermowell inside the
reactor.
The reactor pressure is controlled by a back-pressure regulator (TESCOM 26-1765-24-161). The
BPR also reduces the pressure of the outlet gas/vapor stream to atmospheric. A condenser is
installed in the reactor outlet line after the BPR. Its role is to condense the methanol and the other
liquid products for further analysis. The condenser, is filled with glass beads and is immersed in a
cooling bath that, typically, contains a mixture of dry-ice and acetone, and whose temperature is -
78 °C, ensuring the complete condensation of methanol in the gas outlet. This is verified by the
fact that no MeOH is detected (via GC analysis) in the gas phase. The gas stream exiting the
condenser that contains CO, CO2 and H2 is directed to a Gas Chromatograph equipped with a
Thermal Conductivity Detector (GC-TCD, HP5800,column# Carboxe 1000) in order to analyze
its composition. The composition of the liquid phase that is collected in the condenser, which
contains the reaction products, i.e., H2O, MeOH, and potential by-products (e.g., Dimethyl Ether
(DME)) is analyzed with a Gas Chromatograph equipped with a Flame Ionization Detector (GC-
FID, HP5800, column # DB-1301). The tubing in between the reactor and the BPR are heated with
heating tapes at 170 °C in order to prevent methanol from condensing on the tube wall. In order to
verify that mass balance is closed, after the reactor reaches steady-state, the mass of the material
collected in the condenser is compared with the value calculated by the inlet/outlet gas flow
measurements. More details will be provided later.
48

Figure 2-4 Schematic of the heating system. A three-point thermocouple is installed into a well in the reactor, the measured
points are noted as 1, 2 and 3. Three PID controllers, each connected to a SSR (Solid State Relay) are used to control the three
heaters.

2.2.2 Experimental Procedure
In our experiments, we have used a Cu-based methanol synthesis catalyst (MK-121) from Haldor-
Topsoe with the composition listed in Table 2-4. Before the start of the experiments, 30 g of the
powdered catalyst with a size from 600 µm to 850 µm was diluted with quartz of the same size,
and the mixture was then loaded into the reactor. The same reactor employed for the kinetic
experiments is also utilized for the MR experiments. In order to fill the reactor, its bottom part
was first filled with pure quartz (with smaller particle size from 150 µm to 600 µm). Then the
49

membrane was installed, and the catalyst intermixed with quartz was added to create the MR
section. Above the MR section, a bed of coarser quartz particles (size from 850 µm to 1000 µm)
were added. When the reactor operates as a PBR for the kinetic studies, the inlet and outlet of the
membrane are kept closed (the pressure drop in the reactor is negligible, less than 0.1 bar . With
this configuration, the PBR (and the MR as well) operate quite isothermally, with the temperature
difference across the reactor length being less than 2
o
C.
Table 2-4 MK-121 catalyst properties, (catalyst cylinders with domed ends, 6x4 mm)
Property Value
Chemical Composition
Cu %
Zn %
Al %
Graphite, Oxygen in the metallic oxides,
Carbonates, moisture, %

>43
20±3
5±1

Balance
Axial crush strength, kg/cm
2

Expected filling density, kg/l
>220
1
Each experiment starts with the calibration of the various MFC’s. Then a leak test is performed,
as follows:
1. The reactor is pressurized at 30 bar with N2.
2. The valves at the reactor inlet and outlet are closed.
50

3. The pressure of the reactor is measured after 24 h (1 d).
4. If the rate of the pressure drop is determined to be less than 0.2 bar/d, then the system is
considered leak-free and ready to begin experiments with.
The catalyst as received is in an oxidized state because of its exposure to atmospheric conditions.
Before beginning the experiments, therefore, the catalyst must be activated via hydrogenation, a
step that converts the copper oxide into the catalytically-active elemental copper form according
to the following reaction:
𝐶𝑢𝑂
+ 𝐻 2
→ 𝐶𝑢

+ 𝐻 2
𝑂 ∆𝐻 450𝐾 = −89.00 𝑘𝑗 /𝑚𝑜𝑙 (2-21)
To activate the catalyst, the reactor is heated to 180
o
C, at a rate of 50
◦
C /h under a pressure of 18
bar while purging with N2 (flow rate of 600 cc/min). In the next step, H 2 is introduced into the
reactor with an initial concentration of 12 vol. % balanced with N2, while keeping the total flow
rate constant at 600 cc/min. At this point, an initial increase of about 10
◦
C in the temperature of
the catalyst bed is observed, after which the temperature returns back to its original value. After 3
h from the time we started exposing the catalyst to the H2 mixture, the H2 concentration is increased
to 14 vol. % and the same procedure (i.e., increments in concentration of 2 vol. %, waiting for 3 h,
etc.) is repeated till the H2 concentration reaches to 18 vol. %. After the hydrogen concentration
reached to 18 %, after waiting for another 3 h, the catalyst bed temperature is increased to 200
o
C
with a rate of 10
o
C/h. In the next step, the hydrogen concentration is increased to 22 vol. % over
a period of 18 h, during which period the total flow rate is kept constant at 600 cc/min. When a H2
concentration of 22 vol. % is reached, the reactor is left there for an additional 3 h to complete the
activation process. During the activation process, the pressure is kept constant at ~15 bar. (It should
51

be noted that the activation procedure in this work, due to experimental limitations with our set-
up, has been modified compared to the procedure recommended by the vendor.).
After the catalyst is activated, the reactor is purged with N2 at a flow rate of 600 cc/min for ~ 1 h
before starting the kinetic experiments. In these experiments, we study the impact on the
performance of the reactor of five key parameters, namely temperature (T), pressure (P), catalyst
weight to inlet flow-rate ratio (W/F), the carbon factor in the feed (CF = mol CO/( mol CO + mol
CO2)), and the feed stoichiometric number (SN = (mol H2- mol CO2)/(mol CO + mol CO2) ). A
total of 53 experiments have been carried out by varying these 5 parameters according to the values
listed in Table 2-5:
Table 2-5. Parameters and their values considered in the design of the kinetic experiments
Parameter Level 1 Level 2 Level 3 Level 4 Level 5
Carbon Factor (CF)
=CO/(CO+CO 2)
0.625 0.714 - - -
Stoichiometric Number
(SN)=(H 2-CO 2)/(CO+CO 2)
2 3 - - -
Pressure (bar) 20 25 30 - -
Catalyst weight/Inlet flow-
rate (W/F)
15 20 25 30 40
Temperature (
◦
C) 200 210 220 230 240

52

For each separate experiment, the experimental conditions are adjusted to the target values through
the PID temperature controllers, the BPR, and the MFC’s. The temperature and pressure are
continuously monitored and adjusted. At the same time, the outlet flow rate as well as the outlet
composition are measured by a bubble flow meter and GC-TCD. The steady state is assumed to
be reached when a steady outlet composition is observed (no significant change of the GC analysis
as well as of the outlet flow rate). After the experiment, the liquid condensed in the condenser is
analyzed by GC-MS.
2.3 Parameter Estimation
To estimate the corresponding parameters in the kinetic model, as indicated by equations 2-16 to
2-19, we minimize the following objective function:
𝐹 𝑜𝑏𝑗 = ∑ (𝑋 𝑒𝑥𝑝 − 𝑋 𝑐𝑎𝑙 )
2
(2-22)
where 𝑋 𝑒𝑥𝑝 stands for the experimentally observed carbon conversion and 𝑋 𝑐𝑎𝑙 is the calculated
carbon conversion. The carbon conversion is defined as
𝑋 𝑐𝑎𝑟𝑏𝑜𝑛 =
(𝐹 𝑖𝑛𝑙𝑒𝑡 𝐶𝑂
+𝐹 𝑖𝑛𝑙𝑒𝑡 𝐶 𝑂 2
)−(𝐹 𝑜𝑢𝑡𝑙𝑒𝑡 𝐶𝑂
+𝐹 𝑜𝑢𝑡𝑙𝑒𝑡 𝐶 𝑂 2
)
(𝐹 𝑖𝑛𝑙𝑒𝑡 𝐶𝑂
+𝐹 𝑖𝑛𝑙𝑒𝑡 𝐶 𝑂 2
)
∗ 100 (2-23)
In order to reduce the temperature dependence, the following equation is used when estimating
the parameters:
K
(i)
= A(i)∗ exp(−
B(i)
R
(
1
T
0
−
1
T
)) (2-24)
Where T
0
= 498K.
53

As a first step, Genetic Algorithm (GA) was used to provide an initial guess for the fminsearch
function built in Matlab that uses the Nelder-Mead simplex algorithm. The use of the GA ensures
finding the global minimum for the objective function. After that, the fminsearch function is used
to further refine the value of the minimum starting with the initial guess generated by GA. The
Matlab ode15s ODE solver is applied.

2.4 Results and Discussion
As noted above, a total number of 53 experiments were carried out, and the corresponding
experimental parameters utilized and the conversions attained are listed in Table 2-6. To make
sure the mass balance is closed, in select experiments, the amount of methanol collected in the
condenser was measured and compared to the amount calculated based on the gas-phase
measurements. In these experiments, after the reactor reaches steady-state (as indicated by the gas-
phase measurements), we let the experiment run for an additional time to collect an enough amount
of liquid in the condenser. By measuring the inlet and outlet gas composition, we calculate how
much CO/CO2 is converted into MeOH. Then the liquid collected in the condenser is analyzed by
the GC-MS and the amount of MeOH in the liquid phase is calculated. The amount of methanol
collected in the condenser was ~96-98 % of that predicted.
Table 2-6. Experimental results
run# T--C P--bar
W/F
g*h/mol SN CF Exp. Conv.
1 220 30 15 2 0.625 29.84±1.67
2 220 30 20 2 0.625 36.03±1.39
3 220 30 30 2 0.625 35.95±0.93
4 220 30 25 2 0.625 36.71±0.32
5 220 30 40 2 0.625 45.03±2.3
6 220 25 20 2 0.625 31.91±0.77
54

7 220 25 15 2 0.625 21.95±1.05
8 220 25 30 2 0.625 38.49±1.7
9 220 25 25 2 0.625 34.43±1.89
10 220 20 20 2 0.625 26.29±1.69
11 220 20 20 2 0.625 27.56±1.74
12 220 20 30 2 0.625 32.07±1.48
13 220 20 25 2 0.625 29.08±1.56
14 220 20 15 2 0.625 20.28±1.06
15 220 20 40 2 0.625 33.61±1.10
16 210 20 30 2 0.625 25±1.52
17 210 20 20 2 0.625 20±0.5
18 210 20 25 2 0.625 22.5±1.38
19 210 25 20 2 0.625 22.5±0.88
20 210 25 30 2 0.625 27.5±1.21
21 210 25 25 2 0.625 25±0.43
22 210 30 30 2 0.625 30±2.2
23 210 30 20 2 0.625 27.2±0.25
24 210 30 25 2 0.625 27.5±0.7
25 200 20 20 2 0.625 12.89±0.53
26 200 20 30 2 0.625 22.59±1.02
27 200 20 25 2 0.625 19.3±0.12
28 200 25 20 2 0.625 15.5±0.71
29 200 25 30 2 0.625 26.05±0.82
30 200 25 25 2 0.625 21.2±0.1
31 200 30 20 2 0.625 16.13±0.22
32 200 30 30 2 0.625 25.09±0.69
33 200 30 25 2 0.625 21.41±1.41
34 220 30 20 2 0.625 33.7±1.63
35 230 30 20 2 0.625 32.46±0.26
36 230 30 25 2 0.625 35.42±1.72
37 230 30 15 2 0.625 30.73±1.47
38 220 30 20 3 0.714 42.64±1.03
39 230 30 20 3 0.714 38.99±1.56
40 240 30 15 2 0.625 26.19±0.19
41 230 25 25 2 0.625 27.58±0.58
42 230 25 20 3 0.714 35.08±1.6
43 230 25 25 3 0.714 39.95±1.58
44 220 25 25 2 0.625 32.09±1.54
45 220 25 25 3 0.714 36.33±1.19
46 240 25 15 3 0.714 32.75±0.58
47 240 25 15 2 0.625 23.69±1.01
48 240 20 15 3 0.714 29.24±1.37
49 240 20 15 2 0.625 24.2±0.82
50 230 20 15 3 0.714 30.22±1.14
51 230 20 25 3 0.714 31.47±1.17
52 220 20 25 3 0.714 32.38±0.64
53 220 20 15 3 0.714 27.28±0.89

55

Our experiments indicate that the MeOH selectivity is always higher than 98 %, so for the analysis
of the experimental data we utilize the rate expressions developed by Van den Bussche et al. [7b],
equations 2-16 to 2-19, in which:
𝑘 1
= 𝑘 5𝑎 ′
𝐾 2
′
𝐾 3
𝐾 4
𝐾 𝐻 2
𝑘 3
= 𝐾 𝐻 2

The compound k(i) values are then expressed in the form of an Arrhenius-type equation, k(i)=
A(i)exp (−
B(i)
R
(
1
T
) ) . (which is justified from the fact that each of the rate constants constituting
these terms also obeys a similar Arrhenius-type expression).

To simulate the experimental data, we assume the reactor to be plug-flow (PBR) and that the
methanol synthesis is described by two independent reactions: the reaction between CO2 and
hydrogen to produce methanol (with r
MeOH
representing its rate equation), and the reverse water
gas shift reaction between CO2 and H2 (with r
RWGS
representing its rate equation). The reactor is
then described by the following equations 2-25 to 2-29:
dF
CO
dW
= r
RWGS
(2-25)
dF
CO
2
dW
= −r
MeOH
− r
RWGS
(2-26)
dF
H
2
O
dW
= r
MeOH
+ r
RWGS
(2-27)
dF
H
2
dW
= −3r
MeOH
− r
RWGS
(2-28)
dF
CH
3
OH
dW
= r
MeOH
(2-29)

56

In the above equations, Fi is the molar flow rate (mol/s) of each component, and W (kg) is the
weight of catalyst variable. The carbon conversion is defined by equation 2-29 above, which
assumes that no methanol is present in the feed as is the case with all the experiments reported
here.
Table 2-7 shows the values of the estimated parameters together with their 95 % confidence limits
(estimated using “nlparci”). It also compares the corresponding values from the original study Van
den Bussche et al. [7].
Table 2-7 Parameter values from different literatures
Parameter values for our experiments
Parameter A B
k1
1.28 (±0.09) -63,088 (±124)
k2
157,982 (±688) -554 (±88)
k3
2.37 (±0.31) -1481 (±168)
k4
1240 (±53) -9,495 (±391)
k5
2.14×10
17
(±1.26×10
16
) 162,354 (±240)
Parameter values from Van den Bussche et al. [7]
Parameter A B
k1
1.07

-36,696
k2
3,453.38

--
k3
0.249 -34394
k4
6.62×10
-11

-124,119
k5
1.22×10
10

94,765

57

There are differences among the values calculated in these two studies, likely reflecting the
empirical nature of the global rate expression utilized and differences among the catalysts and
conditions employed by the two studies.
The agreement between the predicted and experimental conversions can be seen from Figure 2-5:

Figure 2-5 Agreement between predicted and experimental conversions (*: experimental conversions)
As we can see, acceptable agreement between predicted conversion and experimental conversion
can be reached, which indicates the satisfactory goodness of fit. Therefore, the kinetic parameters
in Table 2-7 will be used in the membrane reactor modeling in Chapter 4 to simulate the
experimental behavior of the membrane reactor.
58

2.5 Conclusion
The kinetics of the methanol synthesis reaction over a commercial catalyst have been investigated
in our experimental set-up. The rate expressions of Vanden Bussche et al. [7] were used to fit the
experimental data. The values of parameters in the rate expression were determined and acceptable
agreement between the predicted and experimental conversions was observed. The estimated
parameter values are used in the modeling of the behavior of the membrane reactor in Chapter 4.
2.6 References
1. Leonov V, Karavaev M, Tsybina E, Petrishcheva G (1973) Kinetics of methanol synthesis on a
low-temperature catalyst 14:970-975
2. Klier K, Chatikavanij V, Herman R, Simmons G (1982) Catalytic synthesis of methanol from
COH2: IV. The effects of carbon dioxide 74:343-360
3. Villa P, Forzatti P, Buzzi-Ferraris G, Garone G, Pasquon I (1985) Synthesis of alcohols from
carbon oxides and hydrogen. 1. Kinetics of the low-pressure methanol synthesis 24:12-19
4. Graaf G, Stamhuis E, Beenackers A (1988) Kinetics of low-pressure methanol synthesis
43:3185-3195
5. Rozovskii AY, Lin GI (2003) Fundamentals of methanol synthesis and decomposition 22:137-
150
6. Park N, Park M, Lee Y, Ha K, Jun K (2014) Kinetic modeling of methanol synthesis over
commercial catalysts based on three-site adsorption. Fuel Process Technol 125:139-147
7. Bussche KV, Froment G (1996) A steady-state kinetic model for methanol synthesis and the
water gas shift reaction on a commercial Cu/ZnO/Al 2 O 3 catalyst 161:1-10
8. Chinchen G, Denny P, Parker D, Spencer M, Whan D (1987) Mechanism of methanol synthesis
from CO 2/CO/H 2 mixtures over copper/zinc oxide/alumina catalysts: use of 14 C-labelled
reactants 30:333-338
9. Ng KL, Chadwick D, Toseland B (1999) Kinetics and modelling of dimethyl ether synthesis
from synthesis gas 54:3587-3592
59

10. Hadden R, Vandervell H, Waugh K, Webb G (1988) The adsorption and decomposition of
carbon dioxide on polycrystalline copper 1:27-33
11. Lim H, Park M, Kang S, Chae H, Bae JW, Jun K (2009) Modeling of the kinetics for methanol
synthesis using Cu/ZnO/Al2O3/ZrO2 catalyst: influence of carbon dioxide during hydrogenation.
Ind Eng Chem Res 48:10448-10455
12. Xin A, Yizan Z, Zhang Q, Jinfu W (2009) Methanol synthesis from CO2 hydrogenation with
a Cu/Zn/Al/Zr fibrous catalyst. Chin J Chem Eng 17:88-94
13. Anderson J (1989) Methane to higher hydrocarbons 47:177-196
14. Neophytides SG, Marchi AJ, Froment GF (1992) Methanol synthesis by means of diffuse
reflectance infrared Fourier transform and temperature-programmed reaction spectroscopy 86:45-
64
15. Neophytides S, Froment G (1992) A bench scale study of reversed flow methanol synthesis.
Ind Eng Chem Res 31:1583-1589
16. Nakamura J, Campbell JM, Campbell CT (1990) Kinetics and mechanism of the water-gas
shift reaction catalysed by the clean and Cs-promoted Cu (110) surface: a comparison with Cu
(111) 86:2725-2734
17. Ernst K, Campbell CT, Moretti G (1992) Kinetics of the reverse water-gas shift reaction over
Cu (110) 134:66-74

60

3 Methanol Synthesis in a Membrane Reactor
3.1 Membrane Reactors for Methanol Synthesis
As discussed in Chapter 1, a membrane reactor is a system that combines chemical reaction with
separation using membranes [1]. By being able to selectively remove in situ various chemicals,
membrane reactors may provide advantages such as higher process synergy, greater energy
efficiency, and the ability to reach conversion beyond equilibrium-limitation. Mainly inorganic
membranes, such as ceramic membranes [2,3] and Pd membranes are utilized due to their good
tolerance under high-temperature and high-pressure conditions. With the development of various
good property materials, organic membranes are also gaining more attention for use in membrane
reactor applications [4,5].
Membrane reactors have also been suggested previously in the literature for MeOH synthesis and
for the Fischer-Tropsch reaction to overcome both low conversion and thermal management
hurdles [6]. This section will briefly summarize the research status on membrane reactors for
methanol synthesis. A more detailed summary of all the pre-2013 membrane reactors studies can
be found in [7]. Rahimpour [8] carried out a numerical simulation for a Pd-membrane reactor for
methanol synthesis. In their research, they proposed a Pd-membrane reactor as depicted in Figure
3-1. In this configuration, pure hydrogen is fed into the shell side of the reactor while the catalyst
is placed in the tube side. The synthesis gas (containing CO, CO2 and H2) is fed in the membrane
tube-side, where the catalyst is placed as well and the reaction takes place, while additional H 2 is
fed in the membrane shell-side and permeates through hydrogen permselective Pd membrane to
reach the reactor side. By doing this, the hydrogen concentration in the reaction side is always kept
61

at a high level, shifting the equilibrium reaction. They studied the effects of different MR operating
parameters, such as reaction-side flow rate and H2/CO2 ratio in the feed, reaction-side pressure, as
well as membrane thickness. As a conclusion, they reported that the MR performance is favored
by the following conditions: increasing the shell-side pressure, increasing the hydrogen flow rate
in the shell side, and decreasing the flow rate of the feed, and decreasing the membrane thickness.

Figure 3-1 A Pd-membrane reactor system
Struis et al. [9] carried out an experimental study of methanol synthesis in a MR employing a
Nafion® membrane that was permselective towards methanol and water with the goal to improve
the MeS selectivity. The configuration of the membrane reactor is shown in Figure 3-2, with the
feed gas (syngas) flowing in the membrane tube-side and a sweep gas flowing counter-currently
in the shell-side. They studied the performance of various membranes with different counter-ions
such as Li
+
, H
+
and K
+
at temperature of up to 200
o
C, with better long-term stability achieved by
using Li
+
as the counter-ion. They demonstrated improvements in reactor performance by
increasing pressure, and optimizing membrane structure and the modules. In their follow-up work,
a model was developed for the membrane reactor and the prior experimental results were verified.
[10].
62

Figure 3-2 A tubular membrane reactor that is operated in counter-current flow mode. The reaction takes place in the tube side
of the membrane that has a wall thickness of δ.
A modeling study of a membrane reactor for methanol synthesis that utilizes a water-permselective
membrane was carried out by Farsi et al. [11]. They proposed using an alumina-silica composite
membrane for in situ water vapor removal in order to achieve a higher rate of methanol production.
As is shown in Figure 3-3, the MR is surrounded and cooled by flowing water that removes the
heat generated. Inside the MR, the catalyst is placed in the membrane shell-side, while the tube-
side is swept by a sweep gas. During reaction, the water that is generated transports through the
membrane (which is presumed to prevent the other components from passing through) and gets
quickly removed by the sweep gas, leading to a low water concentration along the reactor length.
Thus, the reaction equilibrium is shifted, with the MR achieving higher CO2 conversions. The
model predicts ~4 % enhancement in conversion. The authors also note a potential added benefit
from removing the water vapor, which is an enhancement in the catalyst lifetime, since the
presence of water is also thought to promote catalyst deactivation via sintering.
63

Figure 3-3 Schematic of a water permselective membrane reactor
Rahimpour et al. [12] also developed a model for a reactor system shown in Figure 3-4, that
employs a water-cooled fluidized-bed reactor (FBR) and a fluidized-bed membrane reactor
(FBMR) that recycle the heat generated by the reaction. A schematic of the water-cooled FBR is
shown in Figure 3-5, while a schematic of the gas-cooled FBMR is shown in Figure 3-6. Initially
the syngas is fed through the tube side of the membrane reactor. While the syngas passes through
the reactor, the hydrogen in the syngas is permeated through the membrane (a 1.1 µm Pd-Ag
membrane) to the reaction zone (shell side), shifting the equilibrium. Besides, the heat generated
by the reaction is used to preheat the syngas. Then the pre-heated syngas is fed into the first reactor
for a first-step reaction after which it is fed into the shell side of the membrane reactor for further
reaction. They concluded that by applying this configuration, lower pressure drop as well as longer
catalyst lifetime can be achieved compared to a conventional dual-type reactor system. Moreover,
the model predicted enhanced methanol selectivity.
64

Figure 3-4 A cascade fluidized-bed membrane reactor for methanol synthesis

65

Figure 3-5 Schematic of the water-cooled fluidized bed reactor

Figure 3-6 Schematic of the gas-cooled fluidized bed reactor that contains a hydrogen per-selective membrane
66

In summary, all the membrane reactors s to date for methanol synthesis use a sweep gas and
employ membranes that provide the desired selectivity (e.g., toward H2 or MeOH) in addition to
being able to withstand the high-pressure and high-temperature conditions. Though a lot of effort
has been devoted to date to the development of such materials, hydrogen or MeOH permselective
membranes which are robust in the MeS environment are presently lacking. Further, if such
membranes are ever developed their initial and replacement costs are likely to turn out to be
prohibitive for the proposed application. In this work we investigate instead a different concept,
in which the membrane functions as an interphase contactors between a liquid sweep flowing in
the tube side with for methanol synthesis environment present in its shell side. In our proposed
process, the separation selectivity is provided by the sweep liquid used and not the membrane.
Thus, commercial off-the-self inorganic membranes can be utilized, which require no further
major development effort beyond modifying the hydrophobicity of their surface. One can modify
the membrane’s selective properties by choosing different sweep liquids, which means that
membrane selectivity no longer fully depends on the membrane’s material itself, thus saving on
the cost of materials investment. In addition, one is no longer constrained by the limited availability
of permselective membrane materials, and can instead depend on a greater range of liquid solvents
to attain desired selective properties, which provides us with a greater freedom for modifying the
membrane’s selective property. This chapter will describe the experimental design and the
corresponding membrane reactor experimental results.
67

3.2 Membrane Reactor Experiment
3.2.1 Membrane Reactor Configuration
The configuration or the membrane reactor used in this study is shown as Figure 3-7. A ceramic
membrane (see further details below) is installed in a packed-bed membrane reactor (PBMR) that
contains a MeS catalyst. The membrane divides the reactor into two zones, the shell-side where
the catalyst is packed and where the reaction takes place, and the permeate side (membrane tube-
side) where an organic solvent (in this case Teraethylene Glycol Dimethyl Ether (TGDE)) flows
and serves as a sweep stream. The solvent (TGDE) is chosen because methanol has high solubility
in it but gases like hydrogen and CO do not. TGDE was also previously selected as sweep solvent
by Westerterp at al. [13] in their studies of MeS in a trickle-bed reactor.
To prevent the solvent for completely wetting the membrane, its surface is rendered hydrophobic
(see further details below). To prevent the gases in the reacting side from penetrating into the
permeate side, the pressure of the permeate side (TGDE side) is maintained ~1 bar higher than that
of the reaction side with the aid of a back-pressure regulator installed in the liquid line. By
employing a liquid sweep in the tube-side, the methanol generated in the shell-side is removed in
situ from the reactor side, resulting in a shift of equilibrium towards methanol. The methanol is
then removed from the resulting TGDE/MeOH stream, with the solvent being recycled back into
the MR. One potential other advantage of the solvent sweep, from a process design standpoint, is
for carrying away the exothermic heat of reaction. The membrane plays an important role as an
interphase contactor between the reactive mixture and the product removing solvent without
forcing the solvent to come directly in contact with the catalyst (as is the case with employing
68

trickle-bed reactors), thus reducing the impact that such catalyst may have on solvent stability and
as a result prolonging its life-time.

Figure 3-7 Schematic of the membrane reactor designed for methanol synthesis
3.2.2 Membrane Modification
We utilize in this project a multilayer ceramic membrane, provided to us by Media and Process
Technology, Inc. (M&PT) whose properties, as reported by the manufacturer, are shown in Table
3-2. As explained above, the role of the membrane is to remove in situ the methanol that is
generated in the shell-side while preventing the reactant gases from passing through. In order for
the membrane to function that way, we require the pressure in the tube side to be higher than that
in the shell side so that the liquid will penetrate into the pores of the membrane, blocking the
pathway for gas transport, as shown in Figure 3-8. In addition, the membrane must have a fairly
hydrophobic surface that will prevent the solvent from completely penetrating through the
membrane because that creates a very large resistance for the MeOH molecules to permeate
through.
69

Figure 3-8 Gas-liquid phase contact in a hydrophobic membrane

Table 3-1 Properties of the ceramic membrane installed in the membrane reactor
Layer Material Thickness (µm) Average pore size (Å)
Support α- Alumina 1100 2000- 4000
First layer α- Alumina 10-20 500
Second layer γ- Alumina 2-3 100
Outer Diameter (mm) 5.7
Inner Diameter (mm) 4.7

The desired membrane configuration is shown in Figure 3-8. However, the fresh ceramic
membranes are naturally hydrophilic. In order to create then membranes with a hydrophobic
character their surface must be appropriately modified. For that, we have adapted a modification
method first developed by Lu et al. [14] which uses a fluoroalkylsilane (FAS) compound as a
surface modifier. FAS compounds are fluorinated organosilanes that have hydrolysable groups and
hydrophobic ends [15]. It is thought that FAS compounds attach to the membrane surface through
70

a reaction between their hydrolysable groups with the surface hydroxyl groups of the metal oxide
in the membrane surface, as shown schematically in Figure 3-9 [14,15].

Figure 3-9 FAS attachment mechanism onto the membrane surface
Prior to the surface modification step, the membrane was glazed on both ends, for ~1 cm long, to
ensure complete sealing when it is installed in the reactor. The glazing of the membrane is
performed by following the procedure described below:
1. Cut the original membrane provided by M&PT into 9 cm long tubes.
2. Clean the membrane tube with acetone and let it dry for 1 h.
3. Glaze both ends to about 1cm long from each end, and let the membrane dry in air for
about 3 h.
4. Place the membrane into an oven and follow the following thermal treatment:
a. Increase the temperature to 590 °C at a rate of 5 °C/min and let it stay there for 1 h.
b. Continue increasing the temperature to 850 °C at a rate of 1 °C/min and let it stay there
for an additional 1 h.
c. Let the membrane cool-down to room temperature by shutting the heater off.
71

After the glazing step is completed, the membrane is taken out of the oven and the surface
modification process is carried out as follows:
1. Clean the membrane tube via ultra-sonication in ethanol and in DI-water for 30 min each.
2. Soak the membrane in a solution of ethanol and DI-water with a 2:1 volume ratio for 24 h.
3. Dry the membrane at 333 K for 24 h.
4. Prepare the surface modification solution (0.1 mol/l) according to the following procedure:
Dissolve FAS into hexane under vigorous stirring for 12 h at room temperature (Add
0.76554 g FAS into a 15 ml volumetric flask, and then add hexane into it to prepare the 15
ml solution).
5. Immerse the dry membrane into the FAS/hexane solution prepared in step 4 at room
temperature, ultra-sonicate it for 30 min and let it in the solution for an additional 24 h to
allow the surface coupling reaction to complete.
6. Thoroughly rinse the membrane with hexane (at least 5 times) to remove any unreacted
FAS from the membrane surface.
7. Dry the membrane at 373 K for 12 h.
8. Repeat the steps 5, 6, and 7 several times (4 or more times recommended).
9. Place the modified membrane in a furnace, heat it in an argon flow at a temperature of 200
°C for 6 h.
After the modification step, the membrane now has a hydrophobic surface (Figure 3-10) and is
ready to be installed into the reactor. In our preliminary work (in collaboration with my colleague
Sahar Soltani), several studies have been performed to determine the characteristics of the
modified membrane, such as its break-through pressure, contact angle measurements, membrane
72

morphology, FTIR-DRIFTS characterization and Thermogravimetric Analysis (TGA). The results
of these studies are reported in greater detail elsewhere [16].

Figure 3-10 DI-water drops on the surface of the modified membrane

3.2.3 Experimental Set-Up
The set-up used for the experimental studies on the performance of the MR is the same one
described in Chapter 2. The key difference is that in the set-up utilized in this section (for a
schematic see Figure 3-11), the packed-bed reactor is replaced with a membrane reactor, and a
liquid pumping system has also been integrated into the setup which is used to pump the sweep
liquid through the membrane. (The pumping system consists basically of the solvent reservoir and
a HPLC pump and the associated tubing and valves). A schematic of the membrane reactor itself
is shown as Figure 3-12. A tubular membrane divides the reactor into two zones, the shell-side and
tube-side. The catalyst is packed in the shell-side, while the liquid sweep flows through the tube-
side. Thus the methanol generated in the shell-side is continuously removed by the liquid sweep
along the membrane. After exiting the membrane reactor, the liquid passes through a back-pressure
73

regulator, which is used to control the tube-side pressure. After the BPR, the pressure of the liquid
is reduced to atmosphere pressure. A condenser that is operating under room temperature is
installed after the BPR in order to release the dissolved gases in the TGDE. This step is necessary
because the solubility of CO, CO2 and H2 in the solvent, though relatively low, it is still finite under
the elevated MR pressure conditions, and it is important, therefore, to account for that, in order to
properly close mass balances. The gas stream from the liquid condenser is then joined with the gas
stream exiting the MR shell-side and constitutes the total gas stream exiting the MR. This gas
stream is then sent into a different condenser immersed in a mixture of acetone and dry-ice in order
to collect the condensable vapors (e.g., MeOH). The gas phase exiting the condenser is analyzed
by an online GC equipped with a TCD, while the liquid-phase samples are analyzed by a GC
equipped with a FID. The way that the liquid sample is analyzed is described as follow:
1. After the experiment is finished, the condenser is removed from the system
2. Open the valve on the condenser and take ~1 ml sample into a sample bottle
3. The liquid in the sample bottle is then analyzed by the GC-FID.
74

Figure 3-11 Schematic of the experimental set-up used for the MR studies

Figure 3-12 A detailed schematic of the membrane reactor

75

3.2.4 Experimental Procedure
The catalyst preparation and loading as well as the method of activation follows the same
procedure as discussed in detail in Chapter 2. After the catalyst is activated, the system is ready
for the MR experiments. The experimental procedure followed is similar to that in the kinetic
experiments, except that in the membrane reactor experiments, we also need to analyze the stream
exiting the reactor tube-side as described above.
Before the start of the MR experiments, the following preparatory work is carried out:
1. Calibration of all the MFCs for different gases under the different pressures that will be
used in the experiments
2. HPLC pump calibration
3. GC calibration
4. Leakage test
After the work described above is finished, one is then ready to carry out the experiments:
1. Set the reactor temperature to its pre-determined value using the temperature controllers
2. Turn on the GC-TCD
3. Open the gas cylinders (H2, CO/CO2)
4. Set the MFC for the different gases at their pre-determined values
5. Turn on the heating elements for all preheaters
6. Open the valves to allow the gases to pass into the vent line
7. Use the BPR to set the reactor’s shell-side pressure
8. Verify the gas flow through the vent line with a bubble-flow meter
76

9. Switch the valves so that to direct the reactants flow into the reactor
10. Use the BPR connected in the liquid line to set the liquid pressure in the reactor’s tube-side
(typically ~1 bar higher than the reaction side)
11. Open the valves in the liquid line
12. Turn on the HPLC pump
13. Wait for the steady state to be reached in the reactor
14. Take measurements
Though the technique of rendering the membrane surface hydrophobic is quite successful in
preventing membrane infiltration by the solvent some minor leakage is unavoidable. To verify the
leak rate, the membrane is tested under room temperature with continuously purging TGDE in the
tube side with a pressure ~ 1.5 bar higher than shell side. The outlet is recycled back into the TGDE
reservoir and the leak rate can then be measured by the difference of the amount of TGDE in the
reservoir after a certain time period. In our experiment, the liquid lose rate is always smaller than
2 ml/d.
In order to avoid the solvent from potentially damaging the catalyst, at regular intervals (after ~16
h of continuously running the experiment) we implemented the following procedure in order to rid
the MR of any TGDE that may have accumulated (physically adsorbed) on the surface of the
catalyst:
1. Shut-down the TGDE flow
2. Shut-down the CO/CO2 flow, while setting the H2 flow rate to 150 cc/min
3. Wait for 1 h to make sure that the CO/CO2 is purged out of the reactor
77

4. Decrease the pressure of both the reaction side and permeate side to atmospheric while
keeping the temperature at 220
o
C
5. Leave the system under the above condition overnight to purge out all the TGDE out of the
reactor
3.3 Results and Discussion
A series of preliminary experiments (Table 3-3) have been carried out to test the performance of
the membrane reactor under different temperatures, pressures, liquid flow rates and W/F (catalyst
weight / total flow rate) with a constant stoichiometric number (SN = 2) as well as carbon factor
(𝐹 𝐶 𝑂 2
/𝐹 𝐶𝑂
=0.6).
Table 3-2 MR experiments
run#
T--
o
C
P
(bar)
WF--
kg*s/mol SN CF LF(ml/min)
W/F
(g*h/mol)
Flow Rate
(ml/min)
Exp.
Conversion
1 220 32 140 2 0.625 1 38.9 288.5 45.3
2 220 32 170 2 0.625 1 47.2 237.6 48.9
3 220 32 110 2 0.625 1 30.6 367.2 43.2
4 220 32 80 2 0.625 1 22.2 504.9 36.4
5 220 32 170 2 0.625 2.5 47.2 237.6 54.6
6 220 32 170 2 0.625 2 47.2 237.6 52.9
7 220 32 170 2 0.625 1.5 47.2 237.6 50.5
8 214 32 170 2 0.625 1 47.2 237.6 45.8
9 207 32 170 2 0.625 1 47.2 237.6 41.8
10 200 32 170 2 0.625 1 47.22 237.6 39.4
11 220 24 170 2 0.625 1 47.22 237.6 38.4
12 220 20 170 2 0.625 1 47.22 237.6 35.5
13 220 32 110 2 0.625 6 47.22 367.2 52.2
14 220 32 140 2 0.625 6 47.22 288.5 58.5
78

15 220 32 170 2 0.625 6 47.22 237.6 62.7

3.3.1 Effect of Temperature

Figure 3-13 Effect of temperature on MR conversion. P = 32 bar, liquid sweep rate = 1 cc/min, W/F = 170 kg*s/mol, SN = 2, CF
= 0.625
Figure 3-13 compares the experimental MR conversions with the conversion in a PBR under the
same experimental conditions. The fairly narrow range of temperatures studied is due to concerns
about damaging the coating that renders the membrane surface hydrophobic (with this particular
silylating agent we have found that a safe operating temperature is < 230
o
C [16]) and because of
concerns with catalyst selectivity (the catalyst manufacturer recommends a reaction temperature
79

higher than 190
o
C in order to prevent the production of wax). As shown in the figure, the
conversion of MR increases with increased temperature, due likely to the increased reaction rate
and methanol transport across the membrane as the temperature increases. However, for the PBR,
the conversion passes through a maximum with the reactor conversion at T = 220
o
C being lower
than that at T = 214
o
C, due likely to thermodynamic limitations coming into play under this
relatively high W/F and temperature conditions. Shown on the same figure are the calculated
equilibrium conversions. Note that at this low temperature, the equilibrium conversion is high but
the PBR conversion is low because of the low reaction rate prevailing at these conditions. At the
temperature of 220
o
C, the PBR conversion approaches the equilibrium conversion and under this
condition, the membrane reactor attains a conversion that is higher than the PBR conversion but
also the equilibrium conversion as well. Of all the experiments presented here, the MR conversions
are higher than the corresponding PBR values, though the gains are relatively small due to the size
of the membrane utilized, and the low value of liquid sweep flow rate employed.
80

3.3.2 Effect of Pressure

Figure 3-14 Effect of pressure on MR conversion. T = 220
o
C, liquid sweep rate = 1 cc/min, W/F = 170 kg*s/mol, SN = 2, CF =
0.625
Figure 3-14 shows the effect of pressure on the MR conversion as well as on the PBR conversion
under the same experimental conditions. Shown on the same figure are the calculated equilibrium
conversions. As one can see, at this relatively high temperature the BPR conversions are fairly
close to the equilibrium conversions, particularly for the lower pressures. For all conditions in this
figure, the MR exhibits conversions that are higher than equilibrium. The MR, PBR and the
calculated equilibrium conversions all increase with increasing pressure, which is to be expected
since the methanol synthesis process results in an overall reduction in the number of moles, and is
thus thermodynamically favored at higher pressures. An added reason for the increase in the
81

membrane reactor conversion with increasing pressure, in addition to more methanol being
generated is that the increase in pressure means that a higher amount of methanol transports
through the membrane and is removed by the TGDE, thus further shifting the equilibrium to the
methanol generation side. For the whole range of pressures studied, the MR attains higher
conversions than the PBR.
3.3.3 Effect of W/F

Figure 3-15 Effect of W/F on MR conversion. P = 32 bar, T = 220
o
C, liquid sweep rate = 1 cc/min, SN = 2, CF = 0.625
Figure 3-15 plots the MR conversions vs. W/F for two series of experiments employing different
liquid sweep flow rates. Shown on the same figure are the PBR conversions as wee as the
calculated equilibrium conversion. As expected, both the MR and the PBR conversions increase
82

with increasing W/F (in our case this is equivalent to a decreasing reactant flow rate since the
catalyst weight is kept constant in the experiments). The figure also indicates that a higher liquid
sweep rate promotes the performance of MR by removing more MeOH from the reaction side.
3.3.4 Effect of Sweep Liquid Flow Rate

Figure 3-16 Effect of liquid sweep flow rate on MR conversion. P = 32 bar, T = 220
o
C, W/F = 170 kg*s/mol, SN = 2, CF =
0.625
Figure 3-16 shows the effect of increasing the sweep liquid flow rate on the MR conversion. As
expected, the reactor conversion increases with increasing liquid sweep flow rate. This is because
a higher liquid sweep flow rate helps to remove a larger amount of methanol from the reaction
side, and in so doing pushes the equilibrium toward methanol generation. There is, of course, a
83

limit of how high of a flow rate one can employ, since the higher the flow rate is the more dilute
the MeOH concentration will be, which then negatively impacts downstream processing. Selecting
the most optimal liquid sweep flow rate is, therefore, an important aspect of process optimization,
see further discussion in Chapter 5. (It should be noted, that in the experiments reported here in
order to maintain reactor isothermality, the liquid sweep was preheated to the operating
temperature prior to entering the membrane tube-side).
3.4 Conclusion
In this Chapter, a novel liquid sweep MR has been studied experimentally. The liquid applied has
high a B.P. and high solubility toward methanol. With the main product methanol being removed
from the reaction zone, higher conversion is observed in the MR. A fluorosurfactant agent (FAS)
is used to modify the membrane’s surface property into being hydrophobic. The modified
membrane was characterized by different methods. The experimental result observed with the MR
reactor were compared with those from PBR. Although limited by the size of membrane as well
as the reactor itself, increased total carbon conversions were observed in the MR, indicating the
feasibility of the proposed concept.
3.5 References
1. A.S. Michaels, New separation technique for CPI, Chem. Eng. Prog. 64 (1968) 31-&.
2. F. Akin, Y. Lin, Selective oxidation of ethane to ethylene in a dense tubular membrane reactor,
J. Membr. Sci. 209 (2002) 457-467.
3. H. Wang, Y. Cong, W. Yang, Partial oxidation of ethane to syngas in an oxygen-permeable
membrane reactor, J. Membr. Sci. 209 (2002) 143-152.
84

4. W. Lee, H. Kim, T.K. Kim, H. Chang, Nafion based organic/inorganic composite membrane for
air-breathing direct methanol fuel cells, J. Membr. Sci. 292 (2007) 29-34.
5. S. Ren, G. Sun, C. Li, Z. Liang, Z. Wu, W. Jin, X. Qin, X. Yang, Organic silica/Nafion®
composite membrane for direct methanol fuel cells, J. Membr. Sci. 282 (2006) 450-455.
6. R. Espinoza, E. Du Toit, J. Santamaria, M. Menendez, J. Coronas, S. Irusta, Use of membranes
in Fischer-Tropsch reactors, Studies in Surface Science and Catalysis 130 (2000) 389-394.
7. S. Soltani, M. Sahimi, T. Tsotsis, Catalytic Membrane Reactors, Encyclopedia of Membrane
Science and Technology (2013).
8. M. Rahimpour, S. Ghader, Theoretical Investigation of a Pd‐membrane Reactor for Methanol
Synthesis, Chem. Eng. Technol. 26 (2003) 902-907.
9. Struis, Rudolf Paul Wilhelm Jozef, S. Stucki, M. Wiedorn, A membrane reactor for methanol
synthesis, J. Membr. Sci. 113 (1996) 93-100.
10. R. Struis, S. Stucki, Verification of the membrane reactor concept for the methanol synthesis,
Applied Catalysis A: General 216 (2001) 117-129.
11. M. Farsi, A. Jahanmiri, Dynamic modeling of a H 2 O-permselective membrane reactor to
enhance methanol synthesis from syngas considering catalyst deactivation, Journal of Natural Gas
Chemistry 21 (2012) 407-414.
12. M. Rahimpour, M. Bayat, F. Rahmani, Dynamic simulation of a cascade fluidized-bed
membrane reactor in the presence of long-term catalyst deactivation for methanol synthesis,
Chemical Engineering Science 65 (2010) 4239-4249.
13. D. Mendes, A. Mendes, L. Madeira, A. Iulianelli, J. Sousa, A. Basile, The water‐gas shift
reaction: from conventional catalytic systems to Pd‐based membrane reactors—a review, Asia‐
Pacific Journal of Chemical Engineering 5 (2010) 111-137.
14. J. Lu, Y. Yu, J. Zhou, L. Song, X. Hu, A. Larbot, FAS grafted superhydrophobic ceramic
membrane, Appl. Surf. Sci. 255 (2009) 9092-9099.
15. C.C. Wei, K. Li, Preparation and characterization of a robust and hydrophobic ceramic
membrane via an improved surface grafting technique, Ind Eng Chem Res 48 (2009) 3446-3452.
16. S. Soltani, Methanol synthesis in a membrane reactor (2014).

85

4 Membrane Reactor Modeling
4.1 Scope of the Simulations
As is discussed in Chapter 3, in our proposed membrane reactor concept a liquid sweep flow is
applied in the membrane tube-side in order to remove in situ the methanol that is generated in the
reactor. As such, the membrane serves as an interphase contactor to bring in intimate contact the
gas and liquid phases [1].
In order to maximize the mass transfer rate and thus to optimize the membrane’s function, the
interfacial area of contact should be made as high as possible, and the transport resistance through
the membrane should be minimized. In the two extreme cases, the membrane could either be totally
filled with gas (non-wetting case) or with liquid (complete wetting case). In the MR configuration
employed here, a membrane filled totally by gas is ideal, as it implies a minimal transport
resistance to the membrane reactor’s performance. However, in practice, this is not entirely
possible without risking the gas from bubbling through. So, the experimental configuration studied
in Chapter 3 and modeled here involves membranes in which the liquid has penetrated to some
extent into the membrane as to prevent the gas phase to bubble through and only to allow the
highly soluble components like MeOH to permeate through.
In this Chapter, we first present the model that is used to describe the membrane reactor behavior.
The numerical technique used to simulate the reactor is then presented. The model is then validated
by the experimental data. Finally, the model is employed for some preliminary process design and
scaled-up.
86

4.2 Model Equations
We assume here that the reactor operates under isothermal conditions, the reactor feed-side
(indicated by superscript F) operates under plug-flow conditions, and there is no pressure drop in
neither the feed-side nor the permeate-side (indicated by superscript P). Then the mass balance
equations for the feed-side and permeate-side are as follows:
𝑑 𝐹 𝑖 𝐹 𝑑𝑉
= 𝜌 𝑏 ∑ 𝑟 𝑖 −(𝑎 𝑚𝑣
𝐹 )𝑁 𝑖 𝑚𝑔

(4-1)
𝑑 𝐹 𝑖 𝑃 𝑑𝑉
= (𝑎 𝑚𝑣
𝑃 )𝑁 𝑖 𝑚𝑙

(4-2)
𝐹 𝑖 𝐹 (mol/s), (i = CO, CO2, H2, H2O, CH3OH, TGDE) is the molar flow rate of each component in
the feed-side and 𝐹 𝑖 𝑃 (mol/s), (I = CO, CO2, H2, H2O, CH3OH, TGDE) is the molar flow rate of
each component in the permeate-side, 𝜌 𝑏 (kg/cm
3
) is the bulk catalyst density, 𝑟 𝑖 (mol/kg.s) the
reaction rate for species i (see Chapter 3), 𝑉 (cm
3
) the reactor volume variable, 𝑁 𝑖 𝑚𝑔
(mol/cm
2
.s)
the molar flux of species i through the membrane on the feed-side, 𝑁 𝑖 𝑚𝑙
(mol/cm
2
.s) the molar
flux of species i through the membrane on the permeate-side, 𝑎 𝑚𝑣
𝐹 (cm
2
/cm
3
) the membrane
geometric area per unit of reactor volume on the feed-side, and 𝑎 𝑚𝑣
𝑃 (cm
2
/cm
3
) the membrane
geometric area per unit of reactor volume on the permeate-side. In an expanded form and in terms
of the membrane length as the independent variable the above equations become:
1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝐶𝑂
𝐹 𝑑𝑧
= 𝜌 𝑏 𝑟 𝑅𝑊𝐺𝑆 − 𝑎 𝑚 𝐹 𝑁 𝐶𝑂 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-3)
1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝐶𝑂
2
𝐹 𝑑𝑧
= (−𝜌 𝑏 𝑟 𝑀𝑒𝑡 ℎ𝑎𝑛𝑜𝑙 − 𝜌 𝑏 𝑟 𝑅𝑊𝐺𝑆 )− 𝑎 𝑚 𝐹 𝑁 𝐶𝑂
2
,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-4)
87

1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝐻 2
𝐹 𝑑𝑧
= (−3𝜌 𝑏 𝑟 𝑀𝑒𝑡 ℎ𝑎𝑛𝑜𝑙 − 𝜌 𝑏 𝑟 𝑅𝑊𝐺𝑆 )− 𝑎 𝑚 𝐹 𝑁 𝐻 2
,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-5)
1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡 𝑜𝑟
2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝐻 2𝑂 𝐹 𝑑𝑧
= (𝜌 𝑏 𝑟 𝑀𝑒𝑡 ℎ𝑎𝑛𝑜𝑙 + 𝜌 𝑏 𝑟 𝑅𝑊𝐺𝑆 )− 𝑎 𝑚 𝐹 𝑁 𝐻 2𝑂 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-6)
1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝐶𝐻
3
𝑂𝐻
𝐹 𝑑𝑧
= 𝜌 𝑏 𝑟 𝑀𝑒𝑡 ℎ𝑎𝑛𝑜𝑙 − 𝑎 𝑚 𝐹 𝑁 𝐶𝐻
3
𝑂𝐻 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-7)
1
(𝜋 𝑅 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 2
−𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ,𝑜𝑢𝑡 2
)
𝑑 𝐹 𝑇𝐺𝐷𝐸 𝐹 𝑑𝑧
= −𝑎 𝑚 𝐹 𝑁 𝑇𝐺𝐷𝐸 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑜𝑢𝑡 𝑚𝑔

(4-8)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝐶𝑂
𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝐶𝑂 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-9)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝐶𝑂
2
𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝐶𝑂
2
,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-10)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝐻 2
𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝐻 2
,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-11)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝐻 2𝑂 𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝐻 2𝑂 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-12)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝐶𝐻
3
𝑂𝐻
𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝐶𝐻
3
𝑂𝐻 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-13)
(
1
𝜋 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 2
)
𝑑 𝐹 𝑇𝐺𝐷𝐸 𝑃 𝑑𝑧
= 𝑎 𝑚 𝑃 𝑁 𝑇𝐺𝐷𝐸 ,𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑚𝑙

(4-14)
together with the following boundary conditions at z=0:
𝐹 𝐶𝑂
0,𝐹 = 𝐹 𝐶𝑂
𝑓𝑒𝑒𝑑 ,𝐹 (4-15) 𝐹 𝐶𝑂
0,𝑃 = 0 (4-21)
𝐹 𝐶𝑂
2
0,𝐹 = 𝐹 𝐶𝑂
2
𝑓𝑒𝑒𝑑 ,𝐹 (4-16) 𝐹 𝐶𝑂
2
0,𝑃 = 0 (4-22)
𝐹 𝐻 2
0,𝐹 = 𝐹 𝐻 2
𝑓𝑒𝑒𝑑 ,𝐹 (4-17) 𝐹 𝐻 2
0,𝑃 = 0 (4-23)
88

𝐹 𝐻 2
𝑂 0,𝐹 = 𝐹 𝐻 2
𝑂 𝑓𝑒𝑒𝑑 ,𝐹 = 0 (4-18) 𝐹 𝐻 2
𝑂 0,𝑃 = 0 (4-24)
𝐹 𝐶𝐻
3
𝑂𝐻
0,𝐹 = 𝐹 𝐶𝐻
3
𝑂𝐻
𝑓𝑒𝑒𝑑 ,𝐹 = 0 (4-19) 𝐹 𝐶𝐻
3
𝑂𝐻
0,𝑃 = 0 (4-25)
𝐹 𝑇 𝐺𝐷𝐸 0,𝐹 = 0 (4-20) 𝐹 𝑇𝐺𝐷𝐸 0,𝑃 = 𝐹 𝑇𝐺𝐷𝐸 𝑓𝑒𝑒𝑑 ,𝑃 (4-26)
Where rmembrane out is the outside membrane radius, rmembrane the inside membrane radius, RReactor the
reactor tube inside radius, and z the membrane length variable.
To describe the transport of each component inside the section of the membrane that is occupied
by the gas phase, the Dusty Gas Model (DGM) is used [2]. The diffusion process is due to two
regimes: molecule–wall interactions (Knudsen diffusion), which is more prevalent in small pores
and molecule-molecule interactions (molecular diffusion), with Dij and 𝐷 𝑖 𝑘 being the Knudsen and
molecular diffusion coefficients. The DGM, is described by the following equation (4-27):
∑
𝑥 𝑗 𝑁 𝑖 −𝑥 𝑖 𝑁 𝑗 𝐷 𝑖𝑗
𝑒 +
𝑁 𝑖 𝐷 𝑖 ,𝑘 𝑒 = −
1
𝑅𝑇
∇𝑃 𝑖 𝑛 𝑗 =1
𝑗 ≠𝑖 −
𝑃 𝑖 𝑅𝑇
(
𝐵 0
µ
)
∇𝑃 𝐷 𝑖 ,𝑘 𝑒
(4-27)
in witch xi is the mole fraction of component i, ∇𝑃 is the total pressure gradient, ∇𝑃 𝑖 is the
component partial pressure gradient, and R is the gas constant. B0 is the permeability coefficient,
and µ is the viscosity (µP), which can be calculated using the Corresponding States Method [3].
The effective Knudsen diffusivity of i 𝐷 𝑖 ,𝐾 𝑒 (cm
2
/s) is calculated using the following equation:
𝐷 𝑖 ,𝐾 𝑒 = 10
4
𝜀 𝜏 𝑑𝑝
3
√
8𝑅 𝑇 𝜋 𝑀 𝑖 (4-28)
89

where T (K) is the temperature, R (J/mol.K) the gas constant, Mi (kg/mol) the molecular weight of
i, 𝘀 the porosity (in our case 0.25), 𝜏 the tortuosity (in our case 4) and dp (m) the average pore size
of the porous layer.
The effective Binary diffusivity is calculated by equation 4-29:
𝐷 𝑖𝑗
𝑒 =
𝜀 𝜏 𝐷 𝑖𝑗
(4-29)
whereby the binary gas phase diffusivity of i and j is calculated by:
𝐷 𝑖𝑗
=
𝜀 𝜏 (10
−3
)𝑇 1.75
(
1
𝑀 𝑖 +
1
𝑀 𝑗 )
0.5
𝑃 [(𝜐 𝑖 )
1
3+(𝜐 𝑗 )
1
3
]
2

(4-30)
(--equation 24-42 on page 412 of book ‘fundamentals of Momentum,Heat and Mass Transfer’)
Dij is the binary gas phase diffusivity of i and j in cm
2
/s, T the absolute temperature in K, P the
absolute pressure in atm, Mi, Mj the molecular weight of i and j in g/mol, vi, vj the molecular
volumes in cm
3
/g-mol (Table 24.3 on page 410 of book ‘Fundamentals of Momentum, Heat and
Mass Transfer).
When there is no total pressure difference over the part of the membrane that is occupied by gas,
which is the assumption we make in this study, then equation (4-27) simplifies into equation (4-
31) below:
∑
𝑥 𝑖 𝑁 𝑗 −𝑥 𝑗 𝑁 𝑖 𝐷 𝑖𝑗
𝑒 −
𝑁 𝑖 𝐷 𝑖 ,𝑘 𝑒 =
1
𝑅 𝑇 𝑑 𝑃 𝑖 𝑑𝑟
𝑛 𝑗 =1
𝑗 ≠𝑖 =
𝑑 𝐶 𝑖 𝑑𝑟

(4-31)
The concentration profiles for different components in the gas-filled membrane section are then
described by the following DGM equations:
90

∑
𝑥 𝑖 𝑁 𝑗 𝑚𝑔
−𝑥 𝑗 𝑁 𝑖 𝑚𝑔
𝐷 𝑖𝑗
𝑒 −
𝑁 𝑖 𝑚𝑔
𝐷 𝑖 ,𝑘 𝑒 =
1
𝑅𝑇
𝑑 𝑃 𝑖 𝑚𝑔
𝑑𝑟
𝑛 𝑗 =1
𝑗 ≠𝑖 =
𝑑 𝐶 𝑖 𝑚𝑔
𝑑𝑟

(4-32)
which can be re-written in a matrix form:
𝐷 𝐷𝐺𝑀 ⃑⃑⃑⃑⃑⃑⃑⃑⃑⃑
∗ 𝑁 𝑔 ⃑⃑⃑⃑⃑
=
𝑑𝑐
𝑑𝑧
⃑⃑⃑

Where 𝐷 𝐷𝐺𝑀 ⃑⃑⃑⃑⃑⃑⃑⃑⃑⃑
has the following form:
−
(

∑
𝑐 𝑗 𝐷 1𝑗 𝑒𝑓𝑓

6
𝑗 =2
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 1𝐾 𝑒𝑓𝑓
−
𝑐 1
𝐷 12
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 1
𝐷 13
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 1
𝐷 14
𝑒𝑓 𝑓 ∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 1
𝐷 15
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 1
𝐷 16
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 2
𝐷 21
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
∑
𝑐 𝑗 𝐷 2𝑗 𝑒𝑓𝑓

6
𝑗 =1
𝑗 ≠2
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 2𝐾 𝑒𝑓𝑓
−
𝑐 2
𝐷 23
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 2
𝐷 24
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 2
𝐷 25
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 2
𝐷 26
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 3
𝐷 31
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 3
𝐷 32
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
∑
𝑐 𝑗 𝐷 3𝑗 𝑒𝑓 𝑓
6
𝑗 =1
𝑗 ≠3
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 3𝐾 𝑒𝑓𝑓
−
𝑐 3
𝐷 34
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 3
𝐷 35
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 3
𝐷 36
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 4
𝐷 41
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 4
𝐷 42
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 4
𝐷 43
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
∑
𝑐 𝑗 𝐷 4𝑗 𝑒𝑓𝑓

6
𝑗 =1
𝑗 ≠4
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 4𝐾 𝑒𝑓𝑓
−
𝑐 4
𝐷 45
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 4
𝐷 46
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 5
𝐷 51
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 5
𝐷 52
𝑒 𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 5
𝐷 53
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 5
𝐷 54
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
∑
𝑐 𝑗 𝐷 5𝑗 𝑒𝑓𝑓

6
𝑗 =1
𝑗 ≠5
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 5𝐾 𝑒𝑓𝑓
−
𝑐 5
𝐷 56
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 6
𝐷 61
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 6
𝐷 62
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 6
𝐷 63
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 6
𝐷 64
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
−
𝑐 6
𝐷 65
𝑒𝑓𝑓
∑ 𝑐 𝑘 6
𝑘 =1
∑
𝑐 𝑗 𝐷 6𝑗 𝑒𝑓𝑓

6
𝑗 =1
𝑗 ≠6
∑ 𝑐 𝑘 6
𝑘 =1
+
1
𝐷 6𝐾 𝑒𝑓 𝑓 )

And
𝑑𝑐
𝑑𝑧
⃑⃑⃑
has the following form:
(

𝑑 𝑐 1
𝑑𝑧
𝑑 𝑐 2
𝑑𝑧
𝑑 𝑐 3
𝑑𝑧
𝑑 𝑐 4
𝑑𝑧
𝑑 𝑐 5
𝑑𝑧
𝑑 𝑐 6
𝑑𝑧
)

The continuity equation in cylindrical coordinates is:
91

𝑑 (𝑟 𝑁 𝑖 𝑚𝑔
)
𝑑𝑟
= 0 𝑖 = 𝐶𝑂 ,𝐶 𝑂 2
,𝐻 2
,𝐻 2
𝑂 ,𝐶𝐻
3
,𝑇𝐺𝐷𝐸
(4-33)
equations 4-32 and 4-33 in their expanded form constitute twelve independent ordinary differential
equations. The feed concentrations for all species in the entrance of the reactor on the shell side of
the membrane (see equations 4-15 to 4-20) are known; if the fluxes for all species (𝑁 𝑖 𝑚𝑔
, i=CO,
CO2, H2O, H2, CH3OH, TGDE) were known there as well, one would then be able to integrate the
equations as an initial value problem and calculate the transport of all the components in the
membrane. This then points out the feasibility of using a shooting method to calculate the values
of these fluxes, which will be described later in this section.
The equations describing the transport of each component in the liquid phase are as follows:
𝑁 𝐶𝑂 _𝑟 𝑚𝑙
=
𝐷 𝐶𝑂
𝑒 ,𝑃 r∗ ln (
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝐶𝑂
𝑒𝑞
− 𝐶 𝐶𝑂
𝑃 )
(4-34)
𝑁 𝐶𝑂
2
_𝑟 𝑚𝑙
=
𝐷 𝐶 𝑂 2
𝑒 ,𝑃 r∗ ln (
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝐶𝑂
2
𝑒𝑞
− 𝐶 𝐶𝑂
2
𝑃 )
(4-35)
𝑁 𝐻 2
_𝑟 𝑚𝑙
=
𝐷 𝐻 2
𝑒 ,𝑃 r∗ ln (
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝐻 2
𝑒𝑞
− 𝐶 𝐻 2
𝑃 )
(4-36)
𝑁 𝐻 2
𝑂 _𝑟 𝑚𝑙
=
𝐷 𝐻 2
𝑂 𝑒 ,𝑃 r∗ ln (
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝐻 2
𝑂 𝑒𝑞
− 𝐶 𝐻 2𝑂 𝑃 )
(4-37)
𝑁 𝐶𝐻
3
𝑂𝐻 _𝑟 𝑚𝑙
=
𝐷 𝐶𝐻
3
𝑂𝐻
𝑒 ,𝑃 r∗ ln (
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝐶𝐻
3
𝑂𝐻
𝑒𝑞
− 𝐶 𝐶𝐻
3
𝑂𝐻
𝑃 )
(4-38)
92

𝑁 𝑇𝐺𝐷𝐸 _𝑟 𝑚𝑙
=
𝐷 𝐶𝐻
3
𝑂𝐻
𝑒 ,𝑃 r∗ ln(
𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 + 𝛿 )
(𝐶 𝑇𝐺𝐷𝐸 𝑒𝑞
− 𝐶 𝐶𝐻
3
𝑂𝐻
𝑃 )
(4-39)
where 𝛿 the liquid penetration through a membrane thickness (cm), and 𝐷 𝑖 𝑒 ,𝑃 is the effective
diffusion coefficient (cm
2
/s) of i calculated using the following equation:
𝐷 𝑖 𝑒 ,𝑃 =
𝜀 𝜏 𝐷 𝑖 (4-40)
where ε is the porosity and τ the tortuosity of the membrane layer, and 𝐷 𝑖 is calculated using the
Wilke-Chang [4] equation (4-41)
𝐷 𝑖 = 7.4 ∗ 10
−8
√𝜑 𝑖 𝑀 𝑖 𝑇 𝜇 𝜗 𝑖 0.6
(4-41)
The above equations need to be solved with the following boundary conditions.
At r = r membrane-out
𝐶 𝑖 𝑚𝑔
= 𝐶 𝑖 𝐹 (𝑖 = 𝐶𝑂 ,𝐶𝑂
2
,𝐻 2
𝑂 ,𝐻 2
,𝐶𝐻
3
𝑂𝐻 ,𝑇𝐺𝐷𝐸 )
(4-42)
At r = r membrane-in
𝐶 𝑖 𝑚𝑙
= 𝐶 𝑖 𝑃 (𝑖 = 𝐶𝑂 ,𝐶𝑂
2
,𝐻 2
𝑂 ,𝐻 2
,𝐶𝐻
3
𝑂𝐻 ,𝑇𝐺𝐷𝐸 )
(4-43)
In the above equation 𝐶 𝑖 𝐹 is the gas phase concentration of i in the feed-side, 𝐶 𝑖 𝑃 the liquid
concentration in the permeate side. Also, negligible external mass transport limitations are
assumed.
At r=rliq-gas, i.e., the position of the gas-liquid interphase inside the membrane, a liquid-gas
equilibrium is reached, thus the fugacity of a component in the gas phase must be equal to the
fugacity of the same component in the liquid phase:
93

𝑓 𝑖 𝑔 = 𝑓 𝑖 𝑙 (𝑖 = 𝐶𝑂 ,𝐶𝑂
2
,𝐻 2
𝑂 ,𝐻 2
,𝐶𝐻
3
𝑂𝐻 ,𝑇𝐺𝐷𝐸 )
(4-44)
The Soave-Redlich-Kwong (SRK) equation of state is used to calculate the fugacity of each
component at the liquid-gas interface (described in greater detail later).
At this point, it should be clear that what we are solving is actually a combination of two boundary
value problems, as indicated by Figure 4-1

Figure 4-1 Schematic at the inlet of the reactor
a) A gas transport problem in the gas region inside the membrane as shown in the right hand
side of Figure 4-1 with the boundary conditions defined as:
𝐶 𝑖 𝑚𝑔
= 𝐶 𝑖 𝐹 @ 𝑟 = 𝑟 membrane −out

94

𝐶 𝑖 𝑚𝑔
= 𝐶 𝑖 𝑒𝑞 _𝑔𝑎𝑠 @ 𝑟 = 𝑟 liq−gas

b) A liquid transport problem in the liquid region inside the membrane as indicated by the
left hand side of Figure 4-1 with the boundary conditions defined as:
𝐶 𝑖 𝑚𝑙
= 𝐶 𝑖 𝑃 @ 𝑟 = 𝑟 membrane −in

𝐶 𝑖 𝑚𝑙
= 𝐶 𝑖 𝑒𝑞 _𝑙𝑖𝑞 @ 𝑟 = 𝑟 liq−gas

For the first iteration, 𝐶 𝑖 𝐹 equals to the feed composition is the gas phase and 𝐶 𝑖 𝑃 is the feed
composition in the liquid phase (we assume here no external mass transport limitations in the gas
and liquid phases in the membrane shell-side and tube-side). If one knows the values of the fluxes
at r = 𝑟 membrane −out
, one can then calculate the gas phase composition at the gas-liquid interphase
by solving the ordinary differential equations 4-32 and 4-33. And the liquid-phase composition at
the interphase could then be calculated by solving the ordinary differential equations 4-34 to 4-39.
To obtain the values of the fluxes, we applied here a shooting technique. The technique involves
first making initial guesses for the value of the fluxes at 𝑟 = 𝑟 membrane −out
. With the values of
concentrations and fluxes all known at one single boundary, one can then conveniently solve the
above system of equations (4-32 to 4-32 for gas phase and 4-34 to 4-39 for liquid phase) as an
initial-value problem; this then allows one to calculate the concentrations of every component,
both for the gas and liquid phases, as well as at the gas-liquid interphase. A shooting technique
requires a reasonable initial guess, however. The lower bound of the initial guess is zero cause all
the fluxes have to be larger than zero. The upper bound of the fluxes are those ones under which
at the gas-liquid inter phase the concentration of the components equal to zero. To get the upper
95

bound of the initial guess, we need to solve a so-called boundary value problem described as
follow:
Find the value of the fluxes of the components under which the concentration at the reaction side
and the gas-liquid interphase satisfy the following conditions:
a) @ 𝑟 = 𝑟 membrane −out
𝐶 𝑖 𝑚𝑔
= 𝐶 𝑖 𝐹
b) @ 𝑟 = 𝑟 liq−gas
𝐶 𝑖 𝑚𝑔
= 0
A problem like this could be solved by the built-in function in Matlab, bvp4c. The bvp4c, as
indicated by its name (bvp stands for boundary value problem), solves a boundary value problem
that contains unknown parameters (in this case the fluxes). After providing a rough initial value of
the fluxes (in our case the value provided as the initial guess is the value calculated simply by with
employing Fick’s law, with the binary diffusivities being set equal to the binary diffusivities of
each component with respect to H2 since this is only a roughly guess), the program will adjust the
fluxes until the boundary conditions on both end are satisfied.
After these initial values for the fluxes are calculated, we use the Genetic Algorithm and the
fminsearch (Matlab build-in function) to find the values of fluxes that provide the global minimum
of the following equation (which is the criterion indicating the equilibrium was reached):
𝑒𝑟𝑟 = ∑ (𝑓 𝑖 𝑔 − 𝑓 𝑖 𝑙 )
2
6
𝑖 =1
(4-45)
The role of the Genetic Algorithm is to provide an initial guess for the fminsearch which is close
to the global minimum. Then, the fminsearch continues from this initial guess to find a local
minimum near the initial guess, and in so doing ensures that the global minimum is reached. It
should be noted here, that the bvp4c and the Genetic Algorithm are only used here to provide the
96

fluxes in the first integration along the membrane reactor. After that, only the fminsearch is used
with the initial guess being the fluxes from the previous iteration. As long as the length interval in
the reactor integration is set to be small enough value, employing the previous value of the flux as
an initial guess for the fminsearch is sufficient to obtain the global minimum.
In equation 4-45, 𝑓 𝑖 𝑔 / 𝑓 𝑖 𝑔 stands for the gas/liquid fugacity of each component at the gas-liquid
interphase. As noted above, the fugacity coefficient of a component in a mixture is calculated by
the SRK equation [5]:
ln
𝑓 𝑖 𝑝 𝑥 𝑖 =
𝑏 𝑖 𝑏 (𝑍 − 1)− ln(𝑍 − 𝐵 )−
𝐴 𝐵 (2
𝑎 𝑖 0.5
𝑎 0.5
−
𝑏 𝑖 𝑏 )ln(1 +
𝐵 𝑍 ) (4-46)
where,
𝐴 =
0.42747𝑝 𝑇 2
(∑ 𝑥 𝑖 𝑇 𝑐𝑖
𝛼 𝑖 0.5
𝑝 𝑐𝑖
0.5
)
2
(4-47)
𝐵 =
0.08664𝑝 𝑇 ∑ 𝑥 𝑖 𝑇 𝑐𝑖
𝑝 𝑐𝑖
(4-48)
𝛼 𝑖 is calculated by 𝛼 𝑖 0.5
= 1 + 𝑚 𝑖 (1− 𝑇 𝑅𝑖
0.5
) , where 𝑚 𝑖 = 0.480+ 1.57𝜔 𝑖 − 0.176 𝜔 𝑖 2

𝑎 𝑖 0.5
𝑎 0.5
=
𝛼 𝑖 0.5
𝑇 𝑐𝑖
𝑝 𝑐𝑖
0.5
∑
𝑥 𝑖 𝛼 𝑖 0.5
𝑇 𝑐𝑖
𝑝 𝑐𝑖
0.5
(4-49)
𝑏 𝑖 𝑏 =
𝑇 𝑐𝑖
𝑝 𝑐𝑖
∑
𝑥 𝑖 𝑇 𝑐𝑖
𝑝 𝑐𝑖
(4-50)
Z is obtained by solving the following cubic equation:
𝑍 3
− 𝑍 2
+ 𝑍 (𝐴 − 𝐵 − 𝐵 2
)− 𝐴𝐵 = 0 (4-51)
97

As explained in [4], the smallest root must be chosen for the liquid phase and the greatest one for
a vapor phase. The value of Z is then substituted into equation 4-37 to calculate the fugacity of
each component.
In order to verify the accuracy of the SRK model to describe the thermodynamic properties of our
mixtures, it has been used to model the phase equilibrium experimental data previously reported
by Khosla et al. [6] They carried out vapor-liquid equilibrium (VLE) experiments of the same
synthesis gas components, methanol and water with TGDE in the temperature range 473-513 K
and in the pressure range 0.1-10 MPa. We also compared these results with those generated using
ProII [7]. The results are shown as Table 4-1. In this table, the Inlet column stands for the inlet
composition of the components fed into the flash unit, the Matlab column includes the simulated
results provided by our model, and the ProII column are the simulated results provided by ProII,
while the column titled Experiment reports the reported experiment results.

98

Table 4-1 Solubility comparisons of SRK model with ProII simulations and experimental results
T= 513
K
Inlet
Liquid phase outlet T= 493
K
Inlet
Liquid phase outlet
P= 52.72
bar
Matlab
ProII Experiment
P= 52.72
bar
Matlab
ProII Experiment
TGDE 0.3580
0.7817
0.7916 0.7836 TGDE 0.3580
0.7689
0.78 0.7737
CO 0.1457 0.0199 0.021 0.017 CO 0.1457 0.0189 0.0193 0.0152
CO2 0.0358 0.0115 0.0118 0.0094 CO2 0.0358 0.0120 0.0127 0.0111
H2 0.3069 0.0388 0.0401 0.0355 H2 0.3069 0.0359 0.036 0.0319
H2O 0.0255 0.0246 0.012 0.0245 H2O 0.0255 0.0271 0.0123 0.0286
MeOH 0.1278 0.1234 0.124 0.1261 MeOH 0.1278 0.1373 0.1349 0.1362
T= 473
K
Inlet
Liquid phase outlet T= 493
K
Inlet
Liquid phase-outlet
P= 52.72
bar
Matlab
ProII Experiment
P= 78.58
bar
Matlab
ProII Experiment
TGDE 0.3580 0.7537 0.7676 0.7276 TGDE 0.3580 0.7143 0.722 0.7681
CO 0.1457 0.0180 0.0184 0.0141 CO 0.1457 0.0281 0.0286 0.0203
CO2 0.0358 0.0126 0.0132 0.0086 CO2 0.0358 0.0165 0.0175 0.0132
H2 0.3069 0.0333 0.0333 0.0273 H2 0.3069 0.0539 0.0539 0.047
H2O 0.0255 0.0297 0.0125 0.0342 H2O 0.0255 0.0310 0.0169 0.0273
MeOH 0.1278 0.1527 0.1503 0.1848 MeOH 0.1278 0.1562 0.1541 0.1199
T= 473
K
Inlet
Liquid phase outlet T= 513
K
Inlet
Liquid phase outlet
P= 78.58
bar
Matlab
ProII Experiment
P= 78.58
bar
Matlab
ProII Experiment
TGDE 0.3580 0.7038 0.7137 0.7046 TGDE 0.3580 0.7230 0.7347 0.7284
CO 0.1457 0.0267 0.0272 0.0232 CO 0.1457 0.0297 0.0203 0.0303
CO2 0.0358 0.0173 0.0181 0.0132 CO2 0.0358 0.0161 0.0122 0.017
H2 0.3069 0.0499 0.0499 0.0409 H2 0.3069 0.0583 0.0524 0.0583
H2O 0.0255 0.0333 0.0171 0.0349 H2O 0.0255 0.0288 0.0306 0.0169
MeOH 0.1278 0.1691 0.1671 0.179 MeOH 0.1278 0.1441 0.146 0.142

As one can see from Table 4-1, the SRK model can predict well the experimental results, thus
demonstrating its ability to correctly predict the thermodynamic properties of our mixtures. The
99

calculated methanol concentration in TGDE is even closer to experimental value that the values
calculated by ProII. The relative differences are within 5 %, except for the cases T = 473 K, P =
52.72 bar and T = 493 K, P = 78.58 bar.
4.3 Results and Discussion
The experimental MR data are described in Chapter 3. We have used in the simulations the kinetic
rate expressions developed in Chapter 2. For the membrane characteristics, we used the properties
reported in Table 3.1. In the simulations, we used as an adjustable parameter the liquid layer
thickness, with the goal of minimizing the following objective function:
𝐸 = ∑(𝑋 𝑒𝑥𝑝 − 𝑋 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 )
2
𝑛 𝑖 =1

For that, we use the fminsearch subroutine in Matlab. In order to obtain a good initial guess for the
film thickness δ, we first calculated the value of the objective function E for a number of liquid
film thicknesses (e.g., 1.0x10
-3
cm, 1.5x10
-3
cm, 2.0x10
-3
cm and 2.5x10
-3
cm) with the results
shown in Figure 4-2. From this Figure, the minimum value in the line is used as an intimal guess
in the fminsearch subroutine in Matlab to calculate the film thickness δ that minimizes the
objective function E.

100

Figure 4-2 The effect of liquid layer thickness
Figure 4-3 shows the agreement between the experimental results and the simulated ones
employing the fitted value of δ equal to 1.87x10
-3
cm. The agreement is quite adequate, especially
considering the fact that we have only employed a single adjustable parameter.

Figure 4-3 Agreement between experimental and calculated conversions (%)

101

We have also compared the predicted exit MeOH concentration in the permeate side with the
experimentally measured values. The comparison for three different experiments is shown as
Table 4-2 (for further details about the conditions of the experiments refer to Table 3.1). The
agreement is again satisfactory:
Table 4-2 Agreement between experimental and predicted MeOH concentrations on the permeation side (mole%)
Exp # Experimental Concentration (%) Predicted Concentration (%)
13 6.22 6.28
14 6.14 6.35
15 6.75 6.4

The model has been subsequently utilized to study the impact of the various operating conditions
on the MR performance, with the results to be further discussed below. For all the cases presented,
unless otherwise noted, the liquid penetration thickness is assumed to be 1 µm , with the values
for the rest of the experimental parameters utilized indicated on the figure captions.
102

Figure 4-4 Conversion vs. W/F for different temperatures at P=30 bar, CF=0.625, SN=2,
𝑎 𝑚 𝐹 𝜌 𝑏 = 0.53 cm
2
/g, and sweep flow rate=
6 ml/min
Figure 4-4 shows the effect of temperature on the conversion of both the PBR and the MR. With
increasing W/F, the carbon conversion in both the PBR and the MR increases. For higher values
of W/F, the conversion in the PBR approaches the equilibrium value while the conversion in the
MR continues to increase. The MR conversion shows interesting non-monotonic behavior with
respect to temperature. At low W/F values, the conversion increases with increasing temperature,
but at higher W/F values the conversion line corresponding to the lower temperature (220
o
C)
overtakes those corresponding to the higher temperatures. The reason is that under these higher
103

W/F, the methanol generated in the reaction side is close to equilibrium, thus under low
temperature, the amount of methanol generated is larger than that under a high temperature, leading
as a result to higher amount of methanol removed by the liquid sweep. For low values of W/F, the
reactor operates away from equilibrium, under kinetic limitations with the reaction rates and the
quantities of MeOH produced being relatively low. Under such conditions far from equilibrium a
higher temperature will lead to a faster reaction rate, generating more methanol, increasing the
amount of methanol removed, and thus promoting the performance of the MR.

Figure 4-5 Conversion vs. W/F for different pressures at T=220
o
C, CF=0.625, SN=2,
𝑎 𝑚 𝐹 𝜌 𝑏 =0.53 cm
2
/g, and sweep flow rate=6
ml/min.
104

Figure 4-5 shows the effect of reactor pressure on conversion for both the MR and PBR for a
temperature of 220
o
C. The increase in reactor pressure is beneficial for the performance of both
the MR and the PBR, as expected for a reaction like MeS that results in a decrease in the overall
number of moles. For the MR case, in particular, the increase in pressure has the additional
beneficial effect of increasing the MeOH transport across the membrane, thus shifting the reaction
rate towards the product side. For sufficiently high pressures and W/F values, conversions close
to 90 % are attained in the MR, significantly higher than the equilibrium conversions under these
conditions.

Figure 4-6 Conversion vs. W/F for different liquid thicknesses at P=30 bar, T=220
o
C, CF=0.625, SN=2,
𝑎 𝑚 𝐹 𝜌 𝑏 =0.53 cm
2
/g, and
sweep flow rate=6 ml/min.
105

Figure 4-6 shows the effect of reactor pressure on conversion for both the MR and PBR for a
temperature of 220
o
C. The increase in reactor pressure is beneficial for the performance of both
the MR and the PBR as expected for a reaction like MeS that results in decrease in the overall
number of moles. For the MR case, in particular, the increase in pressure has the additional
beneficial effect in increasing the MeOH transport across the membrane thus shifting the reaction
rate towards the product side. For sufficiently high pressure and W/F values, conversions close to
90 % are attained in the MR, significantly higher than the equilibrium conversions under these
conditions.

Figure 4-7 Conversion vs. W/F for different liquid thicknesses at P=30 bar, T=220
o
C, CF=0.625, SN=2,
𝑎 𝑚 𝐹 𝜌 𝑏 =0.53 cm
2
/g, and
sweep flow rate=6 ml/min.
106

Figure 4-7 shows the effect of varying the liquid layer thickness within the membrane on the
performance of MR (for a pre-determined degree of membrane surface hydrophobicity,
experimentally the thickness can be varied by adjusting the pressure gradient across the membrane
from the tube-side to the shell-side). Four values of the liquid layer thickness are evaluated in the
figure, corresponding to four different levels of liquid infiltration into the membrane structure: top
membrane layer filled (2 µm), both top and intermediate layers filled (22 µm), both top and
intermediate layers completely filled with the support layer only partially filled (500 µm), and both
top layers and the underlying support layer completely filled as well (1099 µm).
As can be seen, the carbon conversion in the MR is higher than that in PBR for all four cases.
However, the carbon conversion in the MR substantially increases with decreasing the liquid layer
thickness, though the effect saturates at the smaller layer thicknesses. Even for small layer
thicknesses, however, transport resistance through the membrane is still dominated by the
resistance of the liquid-filled part of the pore structure. For example, the permeance of the
membrane with only the top layer filled with liquid (layer thickness of 2 µm) is ~20 times greater
than that of the membrane with both the top and intermediate layers filled with liquid (layer
thickness of 22 µm). That the impact of membrane thickness saturates from a point and beyond,
indicates that for the conditions selected in the figure the methanol activity gradient driving
transport across the membrane vanishes and the membrane is no longer active for MeOH transport
beyond that point. This points the need for reactor optimization in which the reaction and
separation components are appropriately balanced.
107

Figure 4-8 Conversion vs. W/F for different feed compositions at P=30 bar, T=220
o
C,
𝑎 𝑚 𝐹 𝜌 𝑏 = 0.53 cm
2
/g, and sweep flow rate=6
ml/min.
Figure 4-8 shows the effect of using different feed compositions on the MR and PBR conversion.
Four different compositions are studied characterized by different combinations of SN (SN=(FH2-
FCO2)/(FCO2+ FCO)) and CF = 0.714 (CF = FCO/(FCO2+ FCO) ). For all compositions, once more, the
carbon conversion in the MR is substantially higher than that in the PBR. When comparing the
cases with the same SN and different CF one observes that the higher CF gives higher carbon
conversion, indicating the fact that CO is the major source of methanol. When comparing the cases
with the same CF and different SN one observes the higher SN gives us higher carbon conversion,
which is consistent with the methanol generation reaction, for which for a higher concentration of
H2, the equilibrium is pushed towards methanol generation. One interesting behavior we notice
108

from the figure is that although under high W/F values, the conversion in MR reactor agrees with
that in a PBR (i.e. the higher conversion in PBR the higher conversion in MR), at low W/F value,
the conversion for a higher SN is lower. One possible reason is that under a high SN value, the
carbon concentration is low, leading to a lower reaction rate for methanol generation. As a result,
the carbon conversion is lower.

Figure 4-9 Conversion vs. W/F for different liquid sweep ratios at P=30 bar, T=220
o
C, CF=0.625, SN=2, and
𝑎 𝑚 𝐹 𝜌 𝑏 =0.53 cm
2
/g.
Figure 4-9 shows the effect of varying the liquid sweep flow rate on MR performance. As it was
also observed in the experiments discussed in Chapter 3, higher liquid sweep flow rates imply a
higher quantity of MeOH being removed, thus leading to a better performance of the MR. From a
109

process design standpoint, however, there is an optimum value of sweep liquid flow rates, as higher
flow rates imply higher potential separation costs downstream and the need to recycle a greater
quantity of the solvent.

Figure 4-10 Conversion vs. W/F for different membrane surface areas at P=30 bar, T=220
o
C, CF=0.625, SN=2, and sweep
flow rate=6 ml/min.
Figure 4-10 shows the effect on MR performance of varying the membrane area employed (four
different membrane areas are studied equivalent to 0.01, 0.1, 0.2, and 1.0 of the base case value of
a
m
F
ρ
b
=0.53). As expected, the higher the membrane area used is the higher the carbon conversion
that is achieved. The impact of increasing the membrane area saturates rather quickly for the
110

conditions investigated in the figure, pointing out the need, once more, for appropriately balancing
the reaction and separation rates.
4.4 Scale-up Studies
Based on the model developed, we have also performed some hypothetical process design studies.
The proposed simplified membrane reactor system is shown in Figure 4-11. It consists of the
membrane reactor and a flash unit. The reactor characteristics and selected operating conditions
are listed in Table 4-3, while the composition of the reactant feed-stream (corresponding to CF =
0.625 and SN = 2) is shown in Table 4-4, together with the compositions of the exit streams. Under
these operating conditions the reactor attains a 74 % conversion (the corresponding equilibrium
conversion under these conditions is 61%) while the MeOH concentration in the liquid outlet is
15.73% (the data are listed in Table 4-4).

Figure 4-11 Schematic of the membrane reactor process

111

Table 4-3 Membrane reactor characteristics and operating conditions
Diameter (m) 4.8
Length (m) 3.2
Membrane Diameter (cm) 2.5
Membrane Thickness (cm) 1
Number of Membranes 2000
Liquid Impregnation Thickness (cm) 2 * 10
-3

Catalyst Weight (metric ton) 4
T (
o
C) 200
P (bar) 55
Pressure of Permeate Side (bar) 56

Table 4-4 Composition of reactor streams

Gas Phase
Gas CO CO 2 H 2O H 2 MeOH TGDE
inlet --
mol/s 18.75 11.25 0 70 0 0
outlet--
mol/s 0.95 3.12 0.56 16.51 2.59 0.09
Liquid Phase
inlet --
mol/s 0 0 0 0 0 90
outlet--
mol/s 0.38 3.7 3.87 5.34 19.27 89.91
112

Table 4-5 Outlet composition of the flash unit
Phase Gas Phase
Gas CO CO 2 H 2O H 2 MeOH TGDE
outlet--
mole/sec 0.37 3.66 3.6 5.32 17.81 5.25
Liquid Phase
outlet--
mole/sec 0.0013 0.04 0.3 0.016 1.46 84.66

From Table 4-4 we can see that the methanol concentration in the liquid outlet is around 16 %
(15.73 %), which is high enough to ensure an effective usage of the TGDE. Table 4-5 indicates
that more than 90 % of the methanol entered into the flash unit is distilled into the gas phase. The
liquid phase that only contains very small (< 2 %) amount of methanol could be then directly
recycled into the reactor without further operation.
4.5 Conclusion
In this chapter, we develop a membrane reactor model and validate it by comparison with the
experimental results presented in Chapter 3. Then the effect of different parameters is investigated.
Based on the model, a scale-up scenario is proposed, which could be serve as guidance for the
future work. The result indicates that this novel membrane reactor concept for methanol synthesis
has a lot of attractive properties that could be further investigated.
4.6 Reference
1. K.K. Sirkar, Other new membrane processes, in: W.S.W. Ho, K.K. Sirkar (Eds.), Membrane
Handbook, Chapman & Hall, New York, 1992, pp. 899±904.
113

2. K.K. Sirkar, Other new membrane processes, in: W.S.W. Ho, K.K. Sirkar (Eds.), Membrane
Handbook, Chapman & Hall, New York, 1992, pp. 904±909.
3. B.E. Poling, J.M. Prausnitz, J.P. O’Connell, “The properties of gases and liquids” McGraw-Hill
Professional, USA (2000), ISBN: 0-07011-682-2
4. C. R. Wilke, P. Chang, AIChE Journal, 1, 264–270 (1955).
5. G. Soave, Equilibrium constants from a modified Redlich-Kwong equation of state, Chemical
Engineering Science 27 (1972) 1197-1203.
6. P. KHOSLA, C. KRISHNAN, J.R. ELLIOTT Jr, J. Berty, Binary and multicomponent vapor-
liquid equilibria of synthesis gas components, methanol and water with tetra ethylene glycol
dimethyl ether (tetraglyme), Chem. Eng. Commun. 102 (1991) 35-46.
7. S. Soltani, Methanol synthesis in a membrane reactor (2014).

114

5 Thesis Conclusion and Suggestions for Future work
In our work, so far, we have studied a MR for methanol synthesis employing a liquid sweep to
increase the conversion. We use in our study ceramic membranes supplied by our industrial
collaborator (Media and Process Technology, Inc.) with a nominal pore size of 100 Å. The ceramic
membrane was modified in order to acquire a hydrophobic pore surface. In our reactor, membrane
itch and separation performance results primarily from the ids of the liquid solvent employed rather
than the membrane itself. Thus off-the-self commercial materials can be utilized with no need to
develop new membranes, thus saving us the cost on new material development. In addition, one
can explore the use of different liquid solvents with a range of selective properties, which provides
more flexibility than having to develop new membrane materials with the required selective
properties.
Chapter 1 briefly discussed the background information on methanol properties, use and
production, membrane reactors, and ceramic membranes. From the existing technical literature,
we conclude that the proposed MR for methanol synthesis employing a liquid sweep is an attractive
process concept to investigate.
Chapter 2 discusses the kinetic study for methanol synthesis. In the literature review section, a
number of prior studies are introduced and discussed. The experimental set-up and procedure are
then discussed in detail. A widely-used kinetic model is utilized to fit the experimental results and
derive the kinetic parameters.
In Chapter 3, the literature review section briefly discusses the background of MR for methanol
synthesis. Subsequently the experimental set-up and procedure are described. Then a series of
115

membrane reactor experiments are described. A fluorosurfactant agent (FAS) is used to modify
the membrane’s surface into being hydrophobic. The experimental results show a higher
conversion in the MR as compared to the PBR.
In Chapter 4, a membrane reactor model is developed and validated by comparing with the
experimental results presented in Chapter 3. The effect of different parameters is investigated.
Based on the model, a scale-up scenario is proposed, which could serve as a guidance for the future
work. The results indicate that this novel membrane reactor concept for methanol synthesis has a
lot of attractive properties that need to be further investigated.
In the experiments described in this thesis, TGDE is used as the sweeping liquid due to its good
solubility towards methanol and relative high boiling point. But in the experiments we still
observed some TGDE loss due to the leakage and the ion of TGDE. Thus, as a suggestion for
future work, sweep liquids that have higher boiling point as well as methanol solubility such as
ironic liquids should be tried. Also, if we one identify find another membrane modification method
or membrane materials that could withstand higher temperatures, a broader range of experimental
condition could be explored for further applications. The technique of sealing the membrane
should be investigated further, ideally to completely prevent any liquid loss. Last but not the least,
one can vary the size and the number of membrane installed in the lab-scale MR to investigate
how it impacts the MR performance.

Abstract (if available)

Abstract This work starts with the study on the kinetics of the methanol synthesis from synthesis gas
(mixture of H₂, CO and CO₂) in a packed-bed reactor to determine the corresponding kinetic
parameter values. After generating the kinetic parameters, membrane reactor experiments are
performed which show improved performance compared to the PBR. In addition, a model of the
MR is developed and validated with the experimental data, and a process design based on the
model is carried out.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Methanol synthesis in the membrane reactor

PDF

Methanol synthesis in a membrane reactor

PDF

A flow-through membrane reactor for destruction of a chemical warfare simulant

PDF

A high efficiency, ultra-compact process for pre-combustion CO₂ capture

PDF

Biogas reforming: conventional and reactive separation processes and the preparation and characterization of related materials

PDF

The use of carbon molecule sieve and Pd membranes for conventional and reactive applications

PDF

Catalytic methane ignition over freely-suspended palladium nanoparticles

PDF

Spontaneous countercurrent and forced imbibition in gas shales

PDF

Lab-scale and field-scale study of siloxane contaminants removal from landfill gas

PDF

On the use of membrane reactors in biomass utilization

PDF

Process intensification in hydrogen production via membrane-based reactive separations

PDF

Studies of transport phenomena in hydrotalcite membranes, and their use in direct methanol fuel cells

PDF

Fabrication of nanoporous silicon carbide membranes for gas separation applications

PDF

Development of polystyrene sulfonic acid-polyvinylidene fluoride (PSSA-PVDF) blends for direct methanol fuel cells (DMFCS)

PDF

A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production

PDF

A study of dispersive mixing and flow based lumping/delumping in gas injection processes

PDF

The study of mass transfer in gas shales and the optimization of hydraulic stimulation processes via additives

PDF

Continuous flow synthesis of catalysts with custom made reactor with flow and temperature studies

PDF

Applications of continuous flow reactors towards screening catalytically active nanoparticle syntheses

PDF

A process-based molecular model of nano-porous silicon carbide membranes

Asset Metadata

Creator Li, Zhongtang (author)

Core Title Methanol synthesis in a contactor-type membrane reactor

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 09/27/2017

Defense Date 08/28/2017

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag gas-liquid phase equilibrium,liquid sweep flow,membrane reactor,methanol synthesis,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Tsotsis, Theodore T. (committee chair), Jessen, Kristian (committee member), Prakash, Surya G. (committee member)

Creator Email zhongtal@usc.edu,zhongtangli@alumni.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-438923

Unique identifier UC11264478

Identifier etd-LiZhongtan-5787.pdf (filename),usctheses-c40-438923 (legacy record id)

Legacy Identifier etd-LiZhongtan-5787.pdf

Dmrecord 438923

Document Type Dissertation

Rights Li, Zhongtang

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA