Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Process intensification in hydrogen production via membrane-based reactive separations
(USC Thesis Other)
Process intensification in hydrogen production via membrane-based reactive separations
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Process Intensification in Hydrogen Production via Membrane-
Based Reactive Separations
By
Ashkan Garshasbi
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)
August 2017
2
Acknowledgments
I dedicate this work to my parents: my mother, Bakhtawar, and my father, Ardeshir, who
offered unconditional love and support and have always been there for me. Without them, I may
never have gotten to where I am today.
I would also like to use this opportunity to express my sincere gratitude toward my
advisor, Professor Dr. Theodore T Tsotsis , for his guidance, motivational talks, and enthusiasm
through the years of my graduate studies and research.
3
Contents
Acknowledgments........................................................................................................................... 2
1. Chapter 1: Introduction and Motivation ................................................................................. 5
1.1 Energy Usage ................................................................................................................... 5
1.2 Hydrogen as an Alternative Fuel ...................................................................................... 7
1.3 Global Climate Change .................................................................................................... 9
1.4 Combined-Cycle Processes ............................................................................................ 10
2. Chapter 2: Reaction Kinetics ................................................................................................ 14
2.1 Water Gas Shift Reaction ............................................................................................... 14
2.2 Experimental Set-up ....................................................................................................... 16
2.3 The Study of WGS Reaction Kinetics............................................................................ 19
2.4 1-D Heterogeneous PBR Model ..................................................................................... 21
2.5 Intraparticle Mass Transfer Limitation .......................................................................... 31
2.6 Impact of Model Tar and Organic Vapor on the Reaction ............................................. 35
2.7 Conclusions .................................................................................................................... 36
3. Chapter 3: Membrane Study ................................................................................................. 38
3.1 Membranes and Membrane Properties ........................................................................... 38
3.2 Mathematical Model for Membrane Transport .............................................................. 41
3.3 Carbon Membrane Transport Properties ........................................................................ 42
3.3.1 The Performance of the CMS Membranes at the Targeted Operating Conditions . 42
3.4 Conclusion ...................................................................................................................... 56
4 Chapter 4: Membrane Reactor Study ................................................................................... 57
4.1 Membrane Reactors........................................................................................................ 57
4.2 Modeling ........................................................................................................................ 59
4.3 Experimental Procedure ................................................................................................. 60
4.4 Membrane Reactor Results ............................................................................................ 61
4.5 Impact of Tar and Organic Vapors on Membrane Reactor Performance ...................... 75
4.6 Conclusion ...................................................................................................................... 78
5 Chapter 5: Modeling .............................................................................................................. 80
5.1 Introduction .................................................................................................................... 80
5.2 1-D Adiabatic Packed-Bed Reactor ............................................................................... 80
5.3 1-D Adiabatic Membrane Reactor ................................................................................. 84
5.4 2-D Isothermal Membrane Reactor ................................................................................ 90
5.5 Multistage Reactor ......................................................................................................... 93
4
5.6 Conclusion .................................................................................................................... 100
6. Summary and Future Study ................................................................................................. 101
6.1 Summary ...................................................................................................................... 101
6.2 Suggestions for Future Work ....................................................................................... 103
References ................................................................................................................................... 104
Appendix ..................................................................................................................................... 114
5
1. Chapter 1: Introduction and Motivation
1.1 Energy Usage
In 2006, world energy consumption was equivalent to the energy released by burning 3.14
cubic miles of oil (CMO, which is a unit of energy consumption often used to aid the public’s
understanding of global-scale energy needs and resources). Fig.1 displays annual global energy
consumption in 2006 corresponding to different common sources of energy, all expressed in terms
of CMO.
Fig. 1: Proportion of world energy use in 2006 from different sources [1]
Between the years 2006 and 2013, a 17% increase in world energy consumption was
reported [1]. During this time period, most of the increase in energy consumption was attributed
to the use of fossil fuels, and among these fuels, over the aforementioned seven-year period, coal
6
had the largest increase in usage. A breakdown of global energy use (in CMO terms) in 2013 is
shown in Fig. 2.
Fig. 2: Proportion of world energy use in 2013 from different sources [1]
Given the current growth rate of world energy consumption, it is projected [1] that humans
will likely consume between 6 to 9 CMO/yr of energy by the year 2050. In order to generate 1
CMO of energy annually from various renewable energy sources alone using currently available
technologies, an employment of the following additional facilities would be necessary [1]: for
hydroelectric power, 200 dams operating at 50% availability and outputting 18 GW each; for
nuclear power, 2500 nuclear plants operating at 90% availability and outputting 900 MW each;
for photovoltaic power generation, 4.2 billion solar panels operating at 20% availability and
outputting 2.1 kW. From this analysis it is clear that for renewable energy sources to be able to
provide a substantial fraction of the world’s future energy needs very significant new investments
7
must be made. The global demand for fossil energy resources over the next several years is,
therefore, likely to remain quite robust [1].
1.2 Hydrogen as an Alternative Fuel
As an alternative carbon-free energy carrier that burns cleanly (i.e., no unburned
hydrocarbons [UHCs] and CO2 are produced during its combustion), hydrogen (H2) is attracting
attention today for its potential use in transportation in urban areas and for power generation in
order to reduce air pollution. In addition to its use as a fuel, H2 is also used extensively as a raw
material in the production of many chemical products.
Hydrogen is quite abundant in the universe, being its most common element. As an
example, our own sun burns 600 million tons of hydrogen each second. But unlike fossil fuels,
large reservoirs of hydrogen are not found on Earth. Instead, the hydrogen atoms are bound
together with other elements in molecules, and it takes energy, therefore, to produce hydrogen so
that it can be used as a fuel. Hydrogen can be produced from sources such as natural gas, coal, and
oil via thermochemical processing. It can also be produced from renewable sources like biomass
and water (via electrolysis), with the maximum environmental benefit realized when the required
energy comes from nuclear, wind, and solar sources [2]. The cost of producing H2 varies
significantly with the raw source type, the production technology, and the method of distribution.
Table 1 below shows the total cost of hydrogen (production and distribution) based on an analysis
in the year 2004 [3]. For comparison purposes, the energy content of 1 kg of hydrogen is
approximately the same as that of one gallon of gasoline [3].
8
Primary Energy Source Total Costs
$/kg H2
Natural gas reforming 1.99
Natural gas +CO2 capture 2.17
Coal gasification 1.91
Coal +CO2 capture 1.99
Wind electrolysis 7.60
Biomass gasification 7.04
Biomass pyrolysis 6.22
Nuclear thermal splitting of water 2.33
Gasoline (for reference) $1.12/gal
Table 1: Estimated cost of hydrogen production, transportation, and distribution for different energy
sources (2004 data, see [3]).
Because of favorable economics and resource availability, conventional fossil fuels are
currently the most common source of H2, which is commonly produced either via steam or
autothermal reforming of natural gas or via coal gasification (Fig. 3 shows the breakdown of H2
production from different primary sources). Both processes, however, are also associated with the
co-production of large quantities of CO2, a powerful greenhouse gas [4, 5]; see further discussion
in Sec. 1.3 below.
9
Fig. 3: World hydrogen production structure (2004 data).
1.3 Global Climate Change
Mounting evidence in recent years seems to indicate that the Earth’s climate is changing
due likely to both natural cycles and human activity. Greenhouse gases, such as carbon dioxide,
methane and nitrogen oxide, are reported to be major contributors to global warming. Although
such greenhouse gases are also produced naturally, studies indicate that human economic activities
are likely to be the major reason for the recent increase in atmospheric greenhouse gas
concentrations. Apparently, the rate of greenhouse gas production by human activities is outpacing
the rate by which the Earth’s systems can presently absorb or use these gases.
Greenhouse gases act by absorbing light, which helps them achieve an excited energetic
state, and then by emitting the absorbed energy in the form of heat, a large fraction of which
reaches the Earth’s surface [6]. Among all the greenhouse gases, carbon dioxide is the primary one
produced by humans. Currently, the principal source of CO2 emissions is from energy production,
such as the burning of natural gas or coal for heat and power generation. Currently, the atmospheric
Natural gas
48%
Petroleum
30%
Electrolysis
4%
Coal
18%
10
carbon dioxide concentration is the highest ever recorded, exceeding pre-industrial revolution
levels by almost 100 ppm and exceeding concentrations measured in gas pockets of ice cores,
reaching back 400,000 years [7]. There is sufficient evidence that climatic change is a driving force
for shifts in the state of various natural systems on a global scale [8]. For example, one of the first
biological indicators of global warming is the bleaching of coral reefs, which have been
particularly impacted because they are so sensitive to small changes in water temperature [9].
Global warming has also caused a shift in spring events, such as leaf unfolding, blooming dates,
species migration, and time of reproduction [10]. A recent study shows that increasing atmospheric
CO2 concentrations may even affect public health in areas where 50% of Americans live [11].
Because global CO2 emissions are on the rise and are causing climatic changes, it is
necessary to take prompt action in order to reduce these emissions. Research focusing on the
reduction of CO2 concentrations is vital. Particularly, it is worth further investigating current
approaches for electricity production, in terms of maximizing both process efficiency and CO2
capturing effectiveness.
1.4 Combined-Cycle Processes
As discussed above, in order to meet global energy demand it is undeniable that we will
continue to rely in the foreseeable future on fossil fuels until new, sustainable and clean
technologies have evolved enough to be able to replace the conventional sources. Attention in
recent years has focused, therefore, on trying to generate power from coal and natural gas in a
more environmentally-benign manner. As the most abundant (and relatively cheap) fossil fuel on
Earth, coal is a particularly strong candidate for environmentally-benign power generation both in
the United States and worldwide (particularly in China). Current estimates indicate that there are
more than 200 years of recoverable coal reserves [12], based on current rates of coal consumption.
11
A cost-effective method for curbing emissions so that coal can be deployed more extensively is,
therefore, highly desirable.
Integrated gasification combined cycle (IGCC) is a power-generation technology that uses
coal which is attracting particular attention today because it allows for the convenient separation
of carbon dioxide for further sequestration and storage, and which also leads to significantly lower
emissions [13] of other toxic pollutants. In the IGCC technology, coal is not combusted directly to
produce electricity but it is gasified instead to produce syngas, a gaseous mixture of CO and
hydrogen. Upon further processing (via the water gas shift (WGS) reaction) one generates pure
H2, to be used for clean electricity production (or for other chemical uses) [1], and a separate CO2
stream ready for further sequestration and storage. A key advantage, furthermore, of the IGCC
technology is that it can be conveniently retrofitted within existing power plants. In summary, the
IGCC technology fits well into the hydrogen economy of the future [14].
Fig. 4 shows a typical flow-diagram for the IGCC process [15, 16]. The current IGCC
process involves first the production of pure O2 (via the separation of oxygen from air in an air
separation unit (ASU)), and second the reaction of coal with steam and/or oxygen in a gasifier to
produce syngas. The syngas is, typically, cooled down in order to remove its various contaminants
such as H2S, tars, and various organic vapors and is then reheated in order to be reacted further
with steam in a WGS reactor to maximize its hydrogen content by converting CO into hydrogen
and CO2. The WGS reaction is exothermic and its equilibrium conversion decreases with an
increase in temperature. Therefore, typically the WGS reaction is carried out in two reactors in
series: a high-temperature shift (HTS) reactor first, followed by a low-temperature shift (LTS)
reactor, in order to overcome equilibrium limitations and to increase CO conversion at practical
space velocities [17,18]. The IGCC technology was first developed ~30 years ago, and the energy
12
efficiency of the IGCC power plant can be as high as 60%, although IGCC is still not practiced for
commercial-scale power production. The total process, as it now stands, is complex and energy-
intensive and, thus, not very attractive, particularly in the context of carbon capture and storage
(CCS).
Fig. 4: Flow diagram of a conventional integrated gasification combined cycle (IGCC) process for coal
[19].
In recent years, in order to improve the IGCC process, a number of novel technologies have
been studied in order to increase the efficiency of the WGS reaction step, including adsorptive
reactors (AR), membrane reactors (MR), and chemical-looping reactors [12, 20]. Among these
technologies, membrane reactors are attracting particular interest because of their ability to attain
WGS conversions above the equilibrium limit while operating at low H2O/CO ratios (1–2
compared to ~9 or higher for the conventional WGS reactors), and also because of their ability to
generate high-purity H2. WGS-MRs have the potential to dramatically increase the efficiency of
the IGCC process, which is very important, as recent studies show that the IGCC process cost is
very sensitively tied to its efficiency, e.g., a 10% decrease in efficiency typically results in an
increase in costs of around 50% [12, 21].
13
Despite the fact that WGS-MRs show excellent promise for application in IGCC for coal,
they have yet to find commercial use. The reason for this lack of commercial use, is that there are
a number of persistent technical challenges that still need to be overcome. These are then the focus
of the efforts described in this Thesis, and will be discussed further below. The rest of this Thesis
is organized as follows: Chapter 2 focuses on the investigation of the kinetics of the water-gas shift
reaction over a sulfided CoMo/Al2O3 catalyst and includes a comparison between three Langmuir-
Hinshelwood-Hougen-Watson-type rate expressions and a power-law rate expression. Chapter 3
focuses on the permeability of carbon molecular sieve membranes through an analysis of pure-
and mixed-gas permeances of gases commonly found in coal syngas using operating pressures of
up to 200 psig and operating temperatures of up to 300 °C. Chapter 4 focuses on enhanced
hydrogen production via the use of a carbon molecular sieve membrane reactor at operating
pressures of up to 200 psig and operating temperatures of up to 300 ⁰C. Chapter 5 focuses on
expanding the modeling for our lab-scale packed-bed and membrane reactor for adiabatic 1-D and
isothermal 2-D cases. Chapter 5 also presents a discussion of alternative configurations for
membrane reactors for enhanced hydrogen production and purification.
14
2. Chapter 2: Reaction Kinetics
2.1 Water Gas Shift Reaction
The water gas shift reaction (equation R.1) has been known for more than a century, and
is widely applied in the chemical industry for H2 production [22, 23].
𝐶𝑂 +𝐻 2
𝑂 ↔𝐶 𝑂 2
+𝐻 2
∆𝐻 298
𝐾 = −41.2 𝑘𝐽 /𝑚𝑜𝑙 (R.1)
The WGS reaction is moderately exothermic and is a typical example of a reaction
controlled by equilibrium, especially at higher temperatures. The WGS reaction does not result in
a change in the number of moles, and reactor pressure does not have, therefore, any effect on its
equilibrium conversion. Fig.5 shows the WGS equilibrium constant KP (in terms of pressures) as
a function of temperature for pressures between 10 and 50 bar [24].
Fig. 5: Variation of WGS equilibrium constant with temperature [24].
As shown in Fig. 5, lower reaction temperatures lead to higher CO equilibrium
conversions, as expected. On the other hand, higher temperatures favor faster kinetics. As a result,
as noted above, in the industry the WGS reaction is, typically, carried out in two reactors placed
15
in series: A HTS reactor first, followed by a LTS reactor. A Fe2O3/Cr2O3 catalyst is, usually, used
in the HTS reactor, which operates adiabatically in the temperature range of 320–550 °C
(depending on the feed composition) and a pressure range of 20-30 bar. In the HTS reactor the CO
concentration is, typically, reduced from 10-13 % to 2-3 % (dry basis). The LTS reactor operates
as a second stage (to the HTS reactor) in the temperature range of 200–250 °C and pressure range
of 10-30 bar, and usually employs a Cu/ZnO/Al2O3 catalyst. In the LTS reactor, the CO
concentration is reduced to 0.2-0.4%. Because water must be in the vapor phase in order for the
WGS reaction to occur, it is important that the temperature of the LTS reactor always stays above
its dew-point corresponding to the reactor’s operating pressure. This, then, determines the lowest
allowable inlet reactor temperature. The LTS reactor is also operated adiabatically, with the
temperature rise along the reactor length staying typically below ~15 °C.
The HTS process and catalyst were first introduced by the German company BASF around
1915. The catalyst’s structure has not changed much from its original formulation. It consists of
Fe2O3 promoted with Cr2O3 to prevent sintering, at normal operating conditions, with the life
expectancy of the catalyst being around 1-3 years. The LTS catalyst was first used industrially in
the 1960’s, and was initially derived from CuO, ZnO, and Al2O3 precursors. This catalyst is active
at temperatures as low as 200 °C and is likely to be affected by mass transfer limitations at higher
temperatures. The usual life expectancy of this catalyst is 1-2 years, when potential poisons (e.g.,
H2S) are strictly controlled.
The importance of the WGS reaction is that it reduces the CO concentration (CO is a side
product in the generation of hydrogen in coal gasifiers) in the gasifier off-gas while increasing its
H2 concentration. Carrying out the WGS reaction is principally dictated by the need to produce
additional H2. Another motivating factor, however, derives from the fact that unreacted CO in the
16
hydrogen product can poison the catalyst in downstream processes. For example, carbon monoxide
is a poison for the iron catalyst used in ammonia synthesis, and it could also damage fuel cells
based on proton exchange membrane (PEM), alkaline, and phosphoric acid technologies [25].
Numerous experimental and computational studies have been conducted to date [26-40] to
investigate different metallic catalysts (e.g., Fe, Co, Mo, Ni, Cu, Ru, Rh, Pd, Ag, Re, Os, Ir, Pt, U,
and Au) supported on various metal oxides (e.g., TiO2, ZnO, CeO2, La2O3, ZrO2, Al2O3, MgO,
Fe2O3, Cr2O3, and SiO2), and to illuminate the WGS reaction mechanism [41, 42]. However,
despite the fact that the WGS reaction is catalyzed by many different types of catalysts, on the
commercial scale (when treating syngas mixtures from methane steam reforming), it is primarily
carried out by using either Fe–Cr2O3-type HTS or Cu–ZnO-type LTS catalysts [43, 44]. However,
coal- and/or biomass-derived syngas usually contains numerous contaminants, including tar, H2S,
heavier hydrocarbons, and NH3, and these compounds (particularly the sulfur-containing species)
are strong poisons for the conventional (based on copper, nickel, and iron) catalysts used in the
processing of syngas. For this reason, Co–Mo sulfur-tolerant shift catalysts (known as sour-shift
catalysts) have also been developed. These sulfur-tolerant Co-Mo catalysts are excellent
candidates for use in the aforementioned IGCC process. The use of such sour-shift catalysts
eliminates the need for an acid gas removal (AGR) unit, since the catalysts can be used directly in
the WGS reactor without any prior gas clean-up processes [45, 46].
2.2 Experimental Set-up
For our research, a lab-scale MR (containing a single-membrane tube) experimental system
was designed and constructed, which is capable of operating over a broad range of conditions
relevant to the accompanying field tests that our industrial collaborators are conducting.
Specifically, for the research reported in this Thesis this lab-scale MR will be operated with
17
simulated oxygen-blown coal-gasifier syngas, for temperatures of up to 300
o
C and total pressures
of up to 200 psig.
A schematic of the WGS-MR system is shown in Fig. 6. The tubular CMS membrane is
sealed inside the tubular stainless steel (SS) reactor (Fig. 7) using graphite O-rings and
compression fittings. The reactor has a length of 25.4 cm and an inside diameter of 3.2 cm. The
sour-shift WGS catalyst particles are first ground to the selected size, thoroughly mixed with
ground quartz particles (of the same size), and are then loaded into the annular space in between
the membrane and the reactor body prior to the initiation of the experiments. We dilute the catalyst
with inert quartz particles in order to completely fill the annular reactor volume, and to be able to
operate the reactor bed under isothermal conditions. Additionally, with the aid of a thermocouple
sliding inside a thermo-well imbedded in the reactor module, we are able to measure temperatures
at multiple points inside of the reactor in order to ensure isothermal reactor operation.
The experimental WGS-MR system consists of three different sections: (i) The feed
section, which consists of gas cylinders, gas regulators, mass-flow controllers (MFCs), pumps, and
the steam-generating units; (ii) the reactor section, which consists of the MR, a furnace for heating
the reactor, pressure gauges for measuring the pressure, a back-pressure regulator to control the
reactor side pressure, two condensers and two moisture-traps to remove the water from the reject-
and permeate-side streams of the reactor, and two traps to remove the H2S from the same streams;
(iii) the analysis section, which consists of a gas chromatograph (GC) capable of analyzing the
concentrations of the exit gas streams, two bubble flow-meters for measuring the total exit flow
rates, and Drager tubes to measure the H2S concentration. (Drager tubes are graduated tubes that
contain a Cu compound that reacts with the H2S and produces CuS, which results in a color change
18
from blue to black. The degree of color change is read on a linear scale on the colorimetric
detection tube.)
Fig. 6: Experimental set-up used in the membrane reactor experiments.
19
Fig. 7: a) Photograph of the membrane reactor module. b) Schematic of the membrane reactor module.
The WGS-MR system can be used as a lab-scale packed-bed reactor (PBR) with the aid of
which one can study the WGS kinetics by closing the permeation side of the MR module, which
sets the flux of gases through the membrane to zero, with the MR in this case effectively
functioning as a PBR.
2.3 The Study of WGS Reaction Kinetics
Sulfur-resistant Co/Mo/Al2O3 catalysts have been available commercially since the early
1970’s and have been used when the WGS reaction is applied to treat feed streams containing
significant amounts of sulfur contaminants (like the coal-derived syngas streams of interest in this
study). The alumina support provides hydrothermal stability, making this class of catalysts
desirable for both low-and intermediate-temperature WGS (200 – 450 °C). The typical catalyst life
of the Co/Mo catalyst is reportedly ~2 years [47].
membrane
membrane
Shell side inlet
Tube side
(sweep side)
inlet
Tube side
(sweep side)
outlet
Thermo-well
Shell side outlet
Thermo-well catalyst or quartz
a)
b)
20
Catalyst form Extrudates
Catalyst size 0.003 m
Chemical Composition
CoO: 3-4 wt%
MoO3: 13-15 wt%
Al2O3: 80-85 wt%
Bulk Density 592.68×10
3
g/m
3
Surface Area 160-220
m
2
/g
Pore Volume 0.55-0.65×10
-6
m
3
/g
Table 2: Physical and chemical properties of the Co-Mo/Al 2O 3 sour-shift catalyst.
In our studies, the water gas shift reaction was investigated using a commercial
Co/Mo/Al2O3 sour-shift catalyst (the physical and chemical properties of which are shown in Table
2) in the lab-scale PBR described in Section 2.2 under isothermal conditions. We investigated a
range of temperatures (220-300 °C) and a range of pressures of up to 200 psig (14.6 atm) (Note
that the lower temperature limit is dictated by the need to maintain a certain level of catalyst
activity, while the upper temperature limit is dictated by concerns about membrane stability).
For the experiments reported here, 10 g of catalyst particles are crushed into smaller
particles and their sizes are sorted with the aid of mesh-screens in the range of 600-800 𝝁 m. Prior
to loading them into the reactor, as noted above, the catalyst particles are mixed and diluted with
quartz particles of the same size in order to completely fill the reactor space (to prevent by-pass)
and to be able to operate the reactor bed under isothermal conditions. Since the Co and Mo metal
components of the fresh catalyst, as received, are in their oxidized form, they must be sulfided
prior to the initiation of the reaction. The activation procedure involves the in-situ reduction of the
21
metals using a gas mixture containing H2 and H2S (5 mol% of H2S) using a temperature and
pressure protocol as specified by the catalyst manufacturer (as noted above, this catalyst is quite
resistant to H2S, and in fact, a minimum ratio of H2S/H2O is required in order to preserve its active
sulfide form). For all experiments (a total of 80 experiments were carried out, with the raw data
being shown in Appendix) we used a single feed syngas composition (namely, H2: CO: CO2: CH4:
H2S: H2O =2.6:1:2.1:0.8:0.05:1.12), typical of the composition of a coal/oxygen-blown gasifier
off-gas [16]. We utilize in these experiments WC/FCO values in the range of 55-170, where WC (10
g) is the weight of undiluted catalyst, and FCO is the molar flow rate of CO (mol/h). For each
experimental point, the reactor was allowed to run for 1 hr at steady-state conditions before the
WGS conversion was measured. The simulated oxygen-blown gasifier syngas was provided pre-
mixed in a gas tank (Praxair, Specialty Gases & Equipment).
2.4 1-D Heterogeneous PBR Model
To analyze the PBR data, we have utilized an isothermal 1-D heterogeneous model (for
justification for the use of this model, see further discussion in Sec. 2.5 below) in which the mass
balance for each component in the reactor is described by equation 2.1.
𝑑 𝑛 𝑗 𝐹 𝑑𝑉
= 𝜈 𝑗 (1−𝜀 𝑣 ) 𝛽 𝑐 𝜌 𝑐 𝑅 𝐹 (2.1)
where is the molar flow rate (mol/h) for component j in the reactor, V the reactor volume
variable (m
3
), υj is the stoichiometric coefficient for component j (negative for reactants, positive
for products, and zero for inserts), the bed porosity in the feed side, the fraction of the solid
volume occupied by the catalysts, the catalyst density (g/m
3
), and R
F
the WGS reaction rate
(mol/g. h). The pressure drop in the packed-bed is calculated using the Ergun equation (equations
2.2 - 2.4 below).
F
j
n
v
c
c
22
−
𝑑 𝑃 𝐹 𝑑𝑉
=10
−6
𝑓 (𝐺 𝐹 )
2
𝐴 𝐹 𝑔 𝐶 𝑑 𝑃 𝜌 𝐹 (2.2)
𝑓 = ⌊
1−𝜀 𝜐 𝜀 𝜐 3
⌋ ⌊1.75+
150 (1−𝜀 𝜐 )
𝑁 𝑅𝑒
𝐹 ⌋ (2.3)
𝑁 𝑅𝑒
𝐹 =
𝑑 𝑃 𝐺 𝐹 𝜇 𝐹 (2.4)
where is the feed-side pressure (bar), the cross-sectional area available to flow for the
reactor (m
2
), the viscosity (g/m.h), the particle diameter in the feed side (m),
the superficial mass flow velocity in the feed side (g/m
2
·h), the average velocity of the fluid
(m/h), the average fluid density (g/m
3
), and the gravity conversion factor. The above set of
equations are solved numerically together with the following boundary conditions:
at V = 0: nj = nj0 and P
F
= (2.5)
where is the inlet feed-side pressure (bar) and nj0 is the inlet molar flow rate for component j
(mol/h).
Our group previously utilized an empirical power-law rate equation to describe the kinetics
of the WGS reaction measured at temperatures of 220-300 ⁰C and at pressures up to 5 atm [48]
over the same commercial sour-shift catalyst. This power-law rate expression is a relatively simple
form of the rate equation that is useful for the design of reactors, and is given by equations 2.6 and
2.7 below (It should be noted that the methane present in the syngas mixture acts here as an inert
gas and does not participate in the WGS reaction):
−𝑅 𝐶𝑂
=𝐴 exp (
𝐸 𝑅𝑇
)𝑃 𝐶𝑂
𝑎 𝑃 𝐻 2
𝑂 𝑏 𝑃 𝐶 𝑂 2
𝑐 𝑃 𝐻 2
𝑑 (1−𝛽 ) (2.6)
𝛽 =
1
𝐾 𝑒𝑞
𝑃 𝐶𝑂
2
𝑃 𝐻 2
𝑃 𝐶𝑂
𝑃 𝐻 2
𝑂 (2.7)
F
P
F
A
F
P
d
F F F
Gu
F
u
F
c
g
0
F
P
0
F
P
23
In the above equations, Pi is the partial pressure for component i (atm), E the activation energy
(J/mol), T the temperature (K), R the gas constant (J/mol K), and K eq is the overall reaction
equilibrium constant in terms of partial pressures, which is given by the following equation 2.8
[49]:
ln(𝐾 𝑒𝑞
)=
5693.5
𝑇 +1.077ln𝑇 +5.44 ×10
−4
𝑇 −1.125×10
−7
𝑇 2
−
49170
𝑇 2
−13.148 (2.8)
Most authors, including ourselves in this Thesis, use the simpler equation 2.9 to calculate the
equilibrium constant that provides values which are fairly close to those obtained from equation
2.8 [50, 51]:
𝐾 𝑒𝑞
=exp
4577.8
𝑇 − 4.33 (2.9)
We have fitted the experimental conversion data in the extended range of pressures and
temperatures to the above empirical power law rate expression and have derived the rate
parameters (A, E, a, b, c and d). For the parameter fitting, we have utilized three built-in MATLAB
functions: Genetic Algorithm (GA), “nlinfit”, and “lsqcurvefit”. The GA is used to give an initial
guess, and “nlinfit” narrows down the GA results for the “lsqcurvefit”. The standard optimization
algorithms generate a single point at each iteration, and the sequence of points approaches an
optimal solution, whereas GA generates a population of points at each iteration, and that
population approaches the optimum solution. At each step, then, the GA selects individuals from
this current population randomly to approach the optimum in the next step. The results from GA
then are used in “nlinfit” and then in “lsqcurvefit” in order to estimate the unknown reaction rate
parameters (A, E, a, b, c and d). During parameter estimation, the reaction order parameters a, b,
c, and d are forced to lie in the range of (-4 and 4), based on prior studies of WGS reaction kinetics
[41, 53]. Table 3 lists the reaction rate parameters for the power-law rate model calculated using
24
the fitting method described above; Fig. 8 shows the iso-conversion graph of experimental data
vs. the calculated data generated with the power-law rate model employing the parameters in Table
3.
In comparison to the WGS reaction employing other catalysts, less mechanistic
information is available for the kinetics of the WGS reaction with the Co/Mo/Al2O3 sour-shift
catalyst. A key challenge that one faces with this catalyst is that its activity depends strongly on
its sulfidation state, which is, in turn, determined by the prevailing reaction environment, in
general, and the H2S concentration, in particular [52]. In addition to the empirical power-law rate
models [48], like the one discussed above, there are also several microkinetic models, which are
available for the WGS reaction on a Co/Mo/Al2O3 catalyst. Osa and coworkers [53], for example,
studied the applicability of three different WGS mechanisms from the technical literature,
specifically, a formate intermediate (WGS FI), an associative (WGS A), and a direct oxidation
(WGS DO) mechanism to data generated over a commercial sulfided Co/Mo catalyst.
A[mol/(atm
(a+b+c+d)
· h · g)] 18957
E [J/mol] 58074
a 4
b -1.46
c 0.13
d -1.44
Table 3: Fitted kinetic parameters for the power-law rate model (equation 2.6).
25
Fig. 8: The iso-conversion graph for the power-law rate model.
The sets of elementary reactions describing each one of these three mechanisms, and the
corresponding global rate expressions for the WGS reaction, as obtained by Osa et al. [53], are as
follows:
Formate intermediate model:
(1) 𝐶𝑂 + ∗
𝑘 −1
←
𝑘 1
→ 𝐶𝑂 ∗
𝑘 1
𝑘 −1
=𝐾 1
(2) 𝐶 𝑂 2
+ ∗
𝑘 −2
←
𝑘 2
→ 𝐶 𝑂 2
∗
𝑘 2
𝑘 −2
=𝐾 2
(3) 𝐻 2
𝑂 + ∗
𝑘 −3
←
𝑘 3
→ 𝐻 2
𝑂 ∗
𝑘 3
𝑘 −3
=𝐾 3
(4) 𝐻 2
+ 2∗
𝑘 −4
←
𝑘 4
→ 2 𝐻 ∗
𝑘 4
𝑘 −4
=𝐾 4
(5) 𝐶𝑂 ∗+𝐻 2
𝑂 ∗
𝑘 −5
←
𝑘 5
→ 𝐻𝐶𝑂𝑂 ∗+ 𝐻 ∗
𝑘 5
𝑘 −5
=𝐾 5
Rate determining step
(6) 𝐻𝐶𝑂𝑂 ∗+ ∗
𝑘 −6
←
𝑘 6
→ 𝐻 ∗+𝐶 𝑂 2
∗
𝑘 6
𝑘 −6
=𝐾 6
∗
T
= ∗ +CO∗+H
2
𝑂 ∗
Rate equation :
𝑅 𝐶 𝑂 2
=
𝐾 𝑤 (𝑝 𝐶𝑂
𝑃 𝐻 2
𝑂 −
𝑝 𝐶 𝑂 2
𝑝 𝐻 2
𝐾 𝑒𝑞
)
(1+𝐾 1
𝑝 𝐶𝑂
+𝐾 3
𝑝 𝐻 2
𝑂 )
2
Where kw= k5 K1 K3 ( mol/g.s.atm
2
)
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80
Simulated CO conversion
Measured Co conversion
26
(2.10)
Associative mechanism:
(1) 𝑂 + ∗
𝑘 −1
←
𝑘 1
→ 𝐶𝑂 ∗
𝑘 1
𝑘 −1
=𝐾 1
(2) 𝐶 𝑂 2
+ ∗
𝑘 −2
←
𝑘 2
→ 𝐶 𝑂 2
∗
𝑘 2
𝑘 −2
=𝐾 2
(3) 𝐻 2
𝑂 + ∗
𝑘 −3
←
𝑘 3
→ 𝐻 2
𝑂 ∗
𝑘 3
𝑘 −3
=𝐾 3
(4) H
2
O∗+ ∗
𝑘 −4
←
𝑘 4
→ OH∗+H∗
𝑘 4
𝑘 −4
=𝐾 4
(5) 𝐻 2
+ 2∗
𝑘 −5
←
𝑘 5
→ 2 𝐻 ∗
𝑘 5
𝑘 −5
=𝐾 5
(6) 𝐶𝑂 ∗+𝑂𝐻 ∗
𝑘 −6
←
𝑘 6
→ 𝐻𝐶𝑂𝑂 ∗+ ∗
𝑘 6
𝑘 −6
=𝐾 6
Rate determining step
(7) 𝐻𝐶𝑂𝑂 ∗+ ∗
𝑘 −7
←
𝑘 7
→ 𝐻 ∗+𝐶 𝑂 2
∗
𝑘 7
𝑘 −7
=𝐾 7
∗
𝑇 = ∗ +𝐶𝑂 ∗+𝐻 2
𝑂 ∗
Rate equation:
𝑅 𝐶 𝑂 2
=
𝐾 𝑤 (
𝑝 𝐶𝑂
𝑃 𝐻 2
𝑂 (𝑝 𝐻 2
)
1
2
−
𝑝 𝐶 𝑂 2
(𝑝 𝐻 2
)
1
2
𝐾 𝑒𝑞
)
(
1+𝐾 1
𝑝 𝐶𝑂
+(𝐾 𝑝 𝐻 2
𝑂 (𝑝 𝐻 2
)
1
2
)
)
2
kw= k6 K1K3K4K5
-0.5
( mol/ g. S.atm
3/2
),
K=K4K3/K5
0.5
(2.11)
Direct oxidation mechanism:
(1) 𝐶𝑂 + ∗
𝑘 −1
←
𝑘 1
→ 𝐶𝑂 ∗
𝑘 1
𝑘 −1
=𝐾 1
(2) 𝐻 2
+ 2∗
𝑘 −2
←
𝑘 2
→ 2 𝐻 ∗
𝑘 2
𝑘 −2
=𝐾 2
(3) 𝐻 2
𝑂 +2
∗
𝑘 −3
←
𝑘 3
→ 𝑂𝐻 ∗+𝐻 ∗
𝑘 3
𝑘 −3
=𝐾 3
(4) 𝑂𝐻 ∗+ ∗
𝑘 −4
←
𝑘 4
→ 𝑂 ∗
+𝐻 ∗
𝑘 4
𝑘 −4
=𝐾 4
(5) 𝐶𝑂 ∗+𝑂 ∗
𝑘 −5
←
𝑘 5
→ 𝐶 𝑂 2
∗+ ∗
𝑘 5
𝑘 −5
=𝐾 5
(6) 𝐶 𝑂 2
∗
𝑘 −6
←
𝑘 6
→ 𝐶 𝑂 2
+ ∗
𝑘 6
𝑘 −6
=𝐾 6
Rate determining step
∗
𝑇 = ∗ +𝐶 𝑂 2
∗
27
Rate equation:
𝑅 𝐶 𝑂 2
=
𝐾 𝑤 (
𝑝 𝐶𝑂
𝑃 𝐻 2
𝑂 𝑝 𝐻 2
−
𝑝 𝐶 𝑂 2
𝐾 𝑒𝑞
)
(1+
𝐾 𝑝 𝐶𝑂
𝑝 𝐻 2
𝑂 𝑝 𝐻 2
)
kw= k6K1K2
-1
K3K4K5 ( mol/g.s.atm), K= K1K2
-1
K3K4K5
(2.12)
where * T is total number of catalyst active sites, and * is the number of free catalyst active sites.
The rate expressions corresponding to these three microkinetic models have been used to
fit the experimental data as well. According to our calculations, the K values in the numerator of
equations 2.10-2.12 are approximately equal to zero and do not have much impact on the results
of the fitted model. Table 4 shows the values for Kw for each of the three microkinetic mechanisms.
Temperature
[°C]
Kw Direct
oxidation
[mol/g.s.atm]
Kw Associative
[mol/g.s.atm
3/2
]
Kw Formate
intermediate
[mol/g.s.atm
2
]
300 1.07×10
-5
8.00×10
-6
7.00×10
-6
280 7.49×10
-6
3.81×10
-6
2.41×10
-6
250 3.82×10
-6
2.07×10
-6
9.99×10
-7
220 1.66×10
-6
1.00×10
-6
5.24×0
-7
Table 4: Reaction rate constant for the three WGS microkinetic models.
The activation energy and the corresponding pre-exponential factor were also computed
using the Arrhenius equation:
𝐿𝑛 (𝐾 𝑤 )=−𝐸 𝑎 (
1
𝑅𝑇
)+𝐿𝑛 (𝐴 ) (2.13)
where Kw is the rate constant ( mol/g.s.atm), T is the temperature (K), A is a pre-exponential factor,
R is the ideal gas constant (J K
−1
mol
−1
), and Ea is the activation energy (J/mol). Fig. 9 shows the
results from plotting the data presented in Table 4. The calculated values for Ea and A are shown
in Table 5.
28
Fig. 9: –Ln(k w) vs. 1/RT for the direct oxidation, associative, and formate intermediate models.
A
[mol/g.s.atm]
Ea
[J/mol]
Direct oxidation 1.11 54871
Associative 1.56 58717
Formate
intermediate
25.91 73360
Table 5. Activation energies and pre-exponential factors calculated from linear regression for the different
microkinetic models.
Figs. 10 -12 show the iso-conversion graphs of experimental data vs. the calculated fit for
the direct oxidation, associative, and formate intermediate models.
0
2
4
6
8
10
12
14
16
2 2.1 2.2 2.3 2.4 2.5
- Ln(k
w
)
1/RT*10
4
(mol/J)
Direct oxidation
Associative
Formate intermediate
29
Fig. 10: Iso-conversion plot of the direct oxidation model.
Fig. 11: Iso-conversion plot of the associative model.
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80
Simulated CO conversion
Measured Co conversion
Direct oxidation
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80
Simulated CO conversion
Measured Co conversion
Associative
30
Fig. 12: Iso-conversion plot of the formate intermediate model.
Discrimination among the rival models (the empirical power-law and the three
microkinetic models) was done by applying the root-mean-square deviation (RMSD) test. The
results of these tests for the different models are shown in Table 6. Clearly, based on the RMSD
test, the empirical power-law model and the direct oxidation microkinetic model perform the best.
RMSD
Direct oxidation 3.38
Associative 5.12
Formate intermediate 8.04
Empirical model 3.32
Table 6: Results for RMSD for each model.
In the above tests, the RMSD has been calculated by equation 2.14.
𝑅𝑀𝑆𝐷 = √
1
𝑛 ∑ (𝑋 𝑒 −𝑋 𝑠 )
2 𝑛 1
(2.14)
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80
Simulated CO conversion
Measured Co conversion
Formate intermediate
31
where Xe is the measured experimental WGS conversion, and Xs is the WGS conversion calculated
by the different models. The CO conversion is defined by equation 2.15.
𝑋 𝐶𝑂
=
𝑛 𝐶𝑂𝑜 𝐹 − 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝐹
𝑛 𝐶𝑂𝑜 𝐹 (2.15)
where 𝑛 𝐶𝑂𝑜 𝐹 is the CO molar flow rate at the inlet of the reactor (mol/h) and 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝐹 is the CO
molar flow rate at the exit of the reactor (mol/h).
Fig. 13 shows the experimental CO conversions at 300 ⁰C and at a pressure 195 psig for
different WC/FCO values as well as the theoretical CO conversions calculated from the various rate
models. Consistent with the RMSD tests, Fig. 13 also clearly indicates that the Direct Oxidation
and the empirical power-law model perform the best.
Fig. 13: Experimental CO conversions at 300 ⁰C and a pressure of 195 psig for different W C/F CO values,
and calculated CO conversions from the various rate models.
2.5 Intraparticle Mass Transfer Limitation
A pseudo-homogeneous 1-D PBR model has been utilized to model the reaction kinetics
experiments. The key assumption for the use of this model is that the catalyst effectiveness factor
20
25
30
35
40
45
50
55
60
65
0 20 40 60 80 100 120 140 160 180
CO Conversion
Weight of catalyst / Molar flow rate of CO
Direct oxidation
Associative
Formate intermediate
Power-Law
32
is independent of conversion, which implies that no significant intraparticle diffusional limitations
are present. To investigate the validity of this assumption, a heterogeneous packed-bed reactor
model was also developed in order to validate the applicability of the homogeneous reactor model
to describe the laboratory data. For such a model the mass balance equation for each component
along the length of the reactor is as follows:
𝑑 𝑛 𝑗 𝐹 𝑑𝑉
= ɳ𝜈 𝑗 (1−𝜀 𝑣 ) 𝛽 𝑐 𝜌 𝑐 R
CO
(2.16)
where ɳ is the effectiveness factor. The effectiveness factor is calculated using the following
equation for spherical particles.
ɳ=
3 𝐷 𝑒𝐶𝑂
𝑑 𝐶 𝐶𝑂
𝑑𝑟
⁞
𝑅
𝑅 𝑝 (−𝑅 𝐶𝑂
)𝜌 𝑐 (2.17)
where D
e
,
CO
is the effective diffusivity of CO ( m
2
/ s) and Rp is the catalyst particle radius (m).
The pressure drop along the length of the reactor is again calculated by the Ergun equation 2.2-
2.4.
Boundary conditions:
At V = 0: 𝑛 𝑗 𝐹 =𝑛 𝑗 0
𝐹 , 𝑃 𝐹 =𝑃 0
𝐹
In order to calculate the effectiveness factor in equation 2.17, the intraparticle CO concentration
profile is calculated by the following equation:
2
𝑟
𝑑 𝐶 𝑟 ,𝐶𝑂
𝑑𝑟
+
𝑑 2
𝐶 𝑟 ,𝐶𝑂
𝑑 𝑟 2
=
(−𝑅 𝐶𝑂
) 𝜌 𝑐 𝐷 𝑒𝐶𝑂
(2.18)
−𝑅 𝑟 ,𝐶𝑂
=𝐴 exp (
𝐸 𝑅𝑇
)𝑃 𝑟 ,𝐶𝑂
𝑎 𝑃 𝑟 ,𝐻 2
𝑂 𝑏 𝑃 𝑟 ,𝐶 𝑂 2
𝑐 𝑃 𝑟 ,𝐻 2
𝑑 (1−𝛽 𝑟 ) (2.19)
𝛽 𝑟 =
1
𝐾 𝑒𝑞
𝑃 𝑟 .𝐶𝑂
2
𝑃 𝑟 ,𝐻 2
𝑃 𝑟 ,𝐶𝑂
𝑃 𝑟 ,𝐻 2
𝑂
33
𝐶 𝑟 𝐻 2
𝑂 =
𝐷 𝑒 ,
𝐶𝑂
( 𝐶 𝑟 ,
𝐶𝑂
−𝐶 𝑠 ,
𝐶𝑂
)+𝐶 𝑠 ,
𝐻 2
𝑂 𝐷 𝑒 ,
𝐻 2
𝑂 𝐷 𝑒 ,
𝐻 2
𝑂 (2.20)
𝐶 𝑟 𝐶 𝑂 2
=
𝐷 𝑒 ,
𝐶𝑂
( 𝐶 𝑠 ,
𝐶𝑂
−𝐶 𝑟 ,
𝐶𝑂
)+𝐶 𝑠 ,
𝐶 𝑂 2
𝐷 𝑒 ,
𝐶 𝑂 2
𝐷 𝑒 ,
𝐶 𝑂 2
(2.21)
𝐶 𝑟 𝐻 2
=
𝐷 𝑒 ,
𝐶𝑂
( 𝐶 𝑠 ,
𝐶𝑂
−𝐶 𝑟 ,
𝐶𝑂
)+𝐶 𝑠 ,
𝐻 2
𝐷 𝑒 ,
𝐻 2
𝐷 𝑒 ,
𝐻 2
(2.22)
where D
e
,
j
is the effective diffusivity of component j (m
2
/s), Cr,j is the concentration of component
j inside the catalyst particle (mol/m
3
), Pr,j is the partial pressure for component j inside the catalyst
particle(atm), Cs,j is the concentration of component j at the surface of catalyst particle (mol/m
3
),
and Rr,CO is reaction rate inside the catalyst particle (mol/ h g). Equations 2.20-2.22 are derived
from the prevailing mass balances within the catalyst particle.
The effective diffusivity is given by the following equation:
𝐷 𝑒 ,
𝑗 =
𝜀 𝑐 𝜏 𝐷 𝑘 ,
𝑖 (2.23)
where D
k
,
i
is the Knudsen diffusivity of component j (m
2
/s), εc is catalyst porosity, and τ is the
catalyst tortuosity factor. The catalyst porosity and tortuosity [69] are calculated by the following
equations:
𝜀 =𝑉 𝑉 𝜌 𝐶 (2.24)
𝜏 =
1
𝜀 (2.25)
where Vv is catalyst pore volume (m
3
/g), see Table 2. The dominant type of gas transport inside
the catalyst particles is assumed to be Knudsen diffusion (our project collaborators from UCLA
have repeated these calculations employing the dusty gas model and their conclusions and results
are pretty similar with those presented here). This assumption is generally valid (depending on the
pressure) when the average pore size is 5-100 nm.
34
Fig. 14 displays the CO concentration profile inside a single catalyst particle at the entrance
of the packed-bed reactor, where the composition is equal to the feed composition, and Fig. 15
shows the effectiveness factor along the length of the reactor. Fig. 14 indicates that the species
concentration does not change significantly inside the catalyst particle compared to the
concentration at the surface of the particle. Fig. 15 indicates that the effectiveness factor is close
to 1 along the length of the reactor. These results justify then the use of the pseudo-homogenous
reactor model (equation 2.1) to estimate the parameters of the reaction rate models.
Fig. 14: The CO concentration profile inside the catalyst particle when the surface composition is the same
as that in the bulk for the packed-bed reactor (feed composition H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12; reactor temperature 300 °C and reactor pressure 200 psig).
38
38.1
38.2
38.3
38.4
38.5
38.6
0 0.1 0.2 0.3 0.4
CO concentration ( mol/m
3
)
Radius of catalyst particle (mm)
35
Fig. 15: The effectiveness factor along the length of the BPR. Other conditions feed pressure 190 psig,
temperature =300 ºC, W C / F CO (weight of catalyst /molar flow rate of CO) = 165 and feed composition (H 2:
CO: CO 2: CH 4: H 2S:H 2O=2.6:1:2.1:0.8:0.05:1.12).
2.6 Impact of Model Tar and Organic Vapor on the Reaction
The composition of the syngas depends on the type of coal used, the oxygen concentration
of the oxidant utilized during the gasification step, and the reaction conditions in the gasifier. The
impurities in syngas include gases such as NH3, H2S, and organic vapors, as well as high-molecular
weight compounds known collectively as tars. Tar-like species present in syngas comprise a
potential challenge to catalyst performance [77, 78].
Several experiments have been carried-out here, in order to investigate the possible effects
of tar and organic vapors on the WGS reaction kinetics with the Co/Mo/Al2O3 catalyst. We have
used the same model coal syngas utilized for the packed-bed kinetic model evaluations (H 2: CO:
CO 2: CH 4: H 2S:H 2O=2.6:1:2.1:0.8:0.05:1.12) along with 0.8 vol % naphthalene and 6.4 vol % toluene
added to the syngas feed. The naphthalene was dissolved in toluene in an appropriate amount, and
it was added to the feed flow as a gas with the aid of an HPLC pump and a boiler.
0.9
0.92
0.94
0.96
0.98
1
1.02
0 0.2 0.4 0.6 0.8 1 1.2
Effectiveness factor
Dimensionless length of reactor
36
In this set of experiments, the packed-bed reactor was first run at 250 ⁰C and 200 psig without
adding the simulated tar for one hour, and the conversion was measured. After one hour, the
simulated tar was added to the feed stream, and the conversion was measured continuously during
a 3-hour time period. Fig. 16 shows the packed-bed reactor conversion vs. time for WC/FCO values
equal to 55 and 82, respectively. The results demonstrate that the conversion for the packed-bed
reactor in the presence of the model tar and organic vapor did not change much compared to the
conversion obtained in their absence.
Fig. 16: Packed-bed CO conversion vs. time in present of a model tar compound (250 ⁰C and at 200 psig).
2.7 Conclusions
We have generated experimental data in a bench-scale PBR for the WGS reaction over a
commercial sulfided Co/Mo catalyst for temperatures up to 300 ⁰C and pressures up to 200 psig.
The data have been fitted using an empirical power-law model as well as three different
microkinetic models. RMSD statistical tests demonstrate that the empirical power-law model and
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0 0.5 1 1.5 2 2.5 3 3.5
Co conversion
Time [h]
W/F=55 with tar model W/F=55 without tar model
W/F=82 with tar model W/F=82 without tar model
37
the direct oxidation microkinetic model provide the most adequate fit to our experimental data for
the WGS reaction.
We have also investigated the likelihood of internal mass transfer limitations being present
inside the catalyst particle by calculating the effectiveness factor along the reactor length. The
results of these calculations show that the pseudo-homogeneous PBR model provides a good
description of the experimental data for the lab-scale reactor and that it can be used for the kinetic
rate data analysis. That the reaction is not affected by intraparticle mass transfer limitations is an
adequate assumption for this study mostly because of the small size of the 600-800 µm particles
utilized in our kinetic experiments. It should be noted, however, that using catalyst particles with
sizes ~700 µm is not practical for large-scale reactors, and for such reactors intraparticle
diffusional limitations may, therefore, be present. We have also shown that the presence of a model
tar and organic vapor does not affect the WGS reaction rate.
38
3. Chapter 3: Membrane Study
3.1 Membranes and Membrane Properties
There has been a renewed interest in recent years in gas separations, driven to a large extent
by energy applications like the production of hydrogen as a vehicular fuel or for power generation,
e.g., the aforementioned IGCC technology, particularly if their performance could be increased
and their costs reduced. Membrane-based hydrogen purification could be a good candidate for this
purpose. Membrane separation processes are easier to control than pressure-swing adsorption
(PSA) processes. Membrane purification processes are usually operated continuously and at steady
state, assuming that the feed stream composition does not change much. The control of membrane
processes may be as simple as controlling the operation temperature and feed and permeate
pressures.
Membranes can be classified into two generic classes: biological, like those encountered
in nature in plants, animals, and humans, and synthetic or man-made. Synthetic membranes can
also be divided into various categories, such as porous and dense, polymeric/organic and inorganic,
and symmetric and asymmetric, among other classifications. The distinction between porous and
dense membranes relates to their structure. Porous membranes have pores that can be either
randomly distributed (for amorphous materials like polymers or silica and alumina) or well ordered
(for crystalline materials like zeolites). The pore size can be as small as a few angstroms (3-4,
typical for gas separation membranes) to a few μm (typical of microfiltration (MF) membranes for
liquid-phase separations). Transport of gas molecules through the pores of membranes can occur
by four different mechanisms (depending on the size and physical structure of the pores) either by
themselves or in combination: Knudsen and/or molecular diffusion, partial condensation/capillary
condensation, selective adsorption and surface diffusion, and molecular sieving.
39
Dense membranes have no pores, and instead consist of a dense thin film (polymer or
metal). Transport through these membranes is thought to occur by a solution/diffusion mechanism,
in which transport results because of a driving force due to a pressure, concentration, or electrical
potential gradient. Dense membranes are particularly well-suited for gas-phase separations. For
example, in cases where high hydrogen purity is needed, dense metal (e.g., Pd or Pd-alloy)
membranes are strongly preferred over porous membranes. A reason for that is because dense
metal membranes can generate high-purity hydrogen via a single-step process. A downside with
these membranes is their high cost and their sensitivity to various impurities found in hydrogen-
containing streams, particularly H2S.
Gas separation membranes, whether dense or microporous (average pore diameter <2 nm,
but typically significantly less than that), generally have an asymmetric structure. They consist of
a top, very thin layer that performs the separation function which sits on the top of a much thicker
macroporous (average pore diameter >50 nm) support that is simply there to provide mechanical
strength. Metal (e.g., porous stainless steel) and ceramics are common materials used as membrane
supports. Metal support tubes are more robust and are easier to install into the membrane modules
and housing. On other hand, metal supports are prone to the problem of thermal expansion
mismatch between the support and the top membrane layer, and are also more sensitive to the
presence of corrosive gas impurities.
Synthetic membranes, at present, are mostly polymer-based. However, in recent years,
interest in membranes made of inorganic materials has increased. The latter materials are of
particular interest for high-temperature, reactive applications (like the WGS reaction of interest in
this Thesis), since polymeric membranes are not robust under such conditions. Fig. 17 shows the
common materials used for inorganic porous and dense membranes [54].
40
Fig. 17: Different types of inorganic membranes [54].
Carbon molecular sieve (CMS) membranes, of principal interest in this research, are among
the most popular microporous inorganic membranes today, with the fabrication of crack-free CMS
membrane films on underlying supports attracting significant research attention [55]. High
selectivity over polymer membranes, the capability of operating at higher pressures, permeance
stability over time, smaller diffusion activation energies, improved thermal stabilities, the
flexibility of fabrication to achieve different separation properties from the same precursor, and
uniform pore size and distribution are some advantages of carbon membranes over polymeric
membranes [55]. Many thermosetting polymers such as poly(vinylidene chloride) (PVDC),
poly(furfuryl alcohol) (PFA), cellulose, cellulose triacetate, saran copolymer, polyacrylonitrile
(PAN), phenol formaldehyde, and various polyimide polymers are common CMS membrane
precursors [56]. There are three stages involved in carbon membrane fabrication: (I) the selection
Inorganic
membrane
Dense
•Metals(pallladium,
silver and their
alloys)
•Solid electrolytes
(Zirconia)
•Nickel
porous
•Oxides(Alumina,
Titania,Zirconia)
•Glass(Silica)
•Metal
•Zeolite
Asymmetric
Symmetric
41
of polymeric precursor materials, (II) a stabilization or pre-oxidation process, and (III) a pyrolytic
process [ 57].
CMS membranes are very suitable for separating gas mixtures with only slightly different
molecular sizes such as O2 (3.46 Å), N2 (3.64 Å), CO2 (3.3 Å), CH4 (3.8 Å), and H2 (2.89 Å) [ 58,
70]. CMS membranes show good potential for application in H2 production, where molecular
sieving is the dominant separation mechanism of H2 from other larger molecules when using such
membranes. These membranes have pores with sizes less than 5 𝐴 ̇ , and that makes them an
excellent material to use to separate hydrogen from coal-derived and biomass-derived syngas [18].
3.2 Mathematical Model for Membrane Transport
A 1-D membrane separator operating under isothermal conditions in the absence of
concentration polarization is described by the following equations:
dn
j
F
dV
=−α
m
U
j
(P
j
F
−P
j
P
) (3.1)
dn
j
P
dV
=α
m
U
j
(P
j
F
−P
j
P
) (3.2)
Where is the molar flow rate (mol/h) for component j in the feed side, the corresponding
molar flow rate (mol/h) in the permeate side, V the reactor volume variable (m
3
), the surface
area of the membrane per unit reactor volume (m
2
/m
3
) and Uj (mol/m
2
. h. bar) the permeance for
component j, described by the following equation.
F
j
=U
j
(P
j
F
−P
j
P
) (3.3)
F
j
n
P
j
n
m
42
where Fj is the molar flux (mol/m
2
. h), through the membrane, Pj
F
(bar) the partial pressure for
component j in the feed side, and Pj
P
(bar) the partial pressure for component j in the permeate
side. (When the membrane is installed inside a MR, the total pressure in the feed-side may
decrease, and in such instance the pressure drop is calculated using the Ergun equations 2.2 – 2.4).
To analyze the data in this chapter, we assumed that the membrane module operates
isothermally (this has been validated experimentally) under ideal gas law conditions, and unless
otherwise noted, the external mass-transfer resistances are negligible (i.e., that transport through
the membrane dominates). The mixed-gas permeances reported in the remainder of this chapter
(e.g., Fig. 22, 23, Tables 8, 9, and 10) were calculated by fitting the experimental data, via the use
of the above equations, and by assuming that the permeance does not vary along the length of the
separator.
3.3 Carbon Membrane Transport Properties
In all our experiments, we have tested tubular membranes fabricated by Media & Process
Technology Inc., of Pittsburgh, PA (MPT). These membranes have an inner diameter of 3.6 mm,
outer diameter of 5.6 mm, and a total length of 254 mm. In these membranes, the carbon separation
layer is placed on the outside surface of a macroporous alumina ceramic tube.
3.3.1 The Performance of the CMS Membranes at the Targeted Operating Conditions
Before their use in the water gas shift membrane reactor (WGS-MR), the transport
properties of the CMS membranes were investigated in both pure-gas and mixed-gas experiments.
43
3.3.1.1 Pure-Gas Permeation Studies
The permeances of pure gases, typical of the major components in coal-derived syngas,
were measured at different temperatures (up to 300 °C) and a fixed feed pressure (200 psig). The
results for one of the MPT membranes tested (CMS#1) are shown in Fig. 18. The experiments in
Fig. 18 were carried-out with the exit in the high-pressure (reject) side being closed to flow. The
feed pressure was adjusted and set by using a forward-pressure valve to an appropriate value. After
the system pressure became stable, the permeate-side flow rate was measured with a bubble flow-
meter. The lab temperature and pressure were measured with a digital thermometer and an analog
barometer, respectively, in order to convert the flows to STP conditions (m
3
/hr) according to the
following equation.
𝑄 𝑆𝑇𝑃 =𝑄 𝑎𝑐𝑡𝑢𝑎𝑙
𝑃 𝑎𝑐𝑡𝑢𝑎𝑙 ×𝑇 𝑆𝑇𝑃
𝑃 𝑆𝑇𝑃 ×𝑇 𝑎𝑐𝑡𝑢𝑎𝑙 (3.4)
where QSTP is the standard volumetric flow rate of gas (m
3
/h), Qactual is the actual volumetric flow
rate of gas (m
3
/h), PSTP is standard pressure (101.325 kPa), Pactual is the lab pressure (kPa –
measured by a lab barometer), TSTP is standard temperature (273.15 K), and Tactual is the lab
temperature (K).
In order to measure water permeation, as reported in Fig. 18, a pure steam stream was
generated with the aid of an HPLC pump and a boiler, and the pressure of the feed-side was
maintained at an appropriate value with the aid of a needle valve. The flow rate of water
transporting into the permeate side was measured with the aid of a water adsorbent bed, which is
a glass vessel loaded with mesh-size 6 anhydrous desiccant (Drierite.co), and was installed at the
exit of the permeate side beyond the heated zone. The permeate steam was passed through the
adsorbent bed for 10 min, and then the adsorbent bed was capped immediately after each
measurement. The adsorbent bed was weighed before and after each experiment to determine any
44
weight change, which was then used to calculate the water permeance. A total of three such tests
were conducted for each permeance measurement at a given temperature.
As can be seen in Fig. 18, membrane CMS#1 is quite permselective towards H2 with
respect to other syngas components such as CO, N2, CO2, and CH4. On the other hand, the H2O
permeance through this membrane is also quite high as well, most likely because the CMS pore
surface is quite hydrophilic. This is a challenge for the use of these membranes in the WGS-MR,
since water is a reactant for the WGS reaction and high permeation rates will result in reactant
loss. One way to potentially overcome this hurdle, is via the use of a steam sweep, as will be
discussed further in this document.
Hydrogen permeation is activated, as shown in Fig. 19 where the hydrogen permeation
data are shown to obey an Arrhenius-like relationship
𝐿𝑛 (𝑈 𝐻 2
)=−𝐸 𝑎 (
1
𝑅𝑇
)+𝐿𝑛 (𝑈 𝐻 2
,0
) (3.5)
where UH2 is the H2 permeance (m
3
/m
2
. h. bar), UH2,0 (m
3
/m
2
. h. bar) is a pre-exponential factor, R
(kJ/K mol) is the gas constant, Ea (kJ/mol) the activation energy, and T (K) the temperature.
45
Fig.18: Pure-gas permeance at different temperatures up to 300
o
C and 200 psig (CMS#1)
Fig. 19: -Ln(U H2) vs. 1/RT for H 2 for temperatures up to 300 ⁰C and at 200 psig (CMS#1).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 100 200 300 400
Permeance [m
3
/m
2
*hr*bar ]
Temperature ⁰C
H
2
CO
CO
2
CH
4
N
2
H
2
O
0
0.5
1
1.5
2
2.5
0 0.5 1 1.5 2 2.5 3 3.5 4
-Ln(U
H 2
)
1/RT × 10
4
(mol/J)
46
Fig. 20: Pure slow-gas permeance vs. 1/T
0.5
for temperatures up to 300 ⁰C and at 200 psig (CMS#1).
The calculated activation energy value from Fig. 19 is Ea = 9.54 (kJ/mol), while the value of the
pre-exponential factor is UH2,0=1.57 m
3
/m
2
. h. bar. The permeance of most of the slow gases (CO,
N2, CH4), on the other hand, decreases with temperature. Specifically, as Fig. 20 shows, the
permeances of these slow gases is proportional to 1/T
0.5
, which is indicative of the fact that these
gases permeate via a Knudsen-type mechanism via cracks and imperfections of the CMS
membrane top layer. The CO2 permeance in Fig. 20, on the other hand, displays a non-linear
behavior, first decreasing with increasing temperature and then increasing, indicative of a more
complex transport mechanism that may also involve adsorption and capillary condensation.
Activated transport is typical of the permeation of fast gases through these membranes.
Fig. 21, for example, shows the permeance of He (2.89 Å) measured at various temperatures (up
to 300 °C) and at 200 psig for a different membrane (CMS#4). This is again a permselective
membrane with a (He:N2) ideal separation factor (at 250 ⁰C) equal to 95. He permeation through
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.035 0.04 0.045 0.05 0.055 0.06
Permeance [m
3
/m
2
*hr*bar ]
1/T
0.5
[1/K
0.5
]
CO CO CH N
2 2 4
47
CMS#4 again obeys an Arrhenius-like behavior with the activation energy calculated to be 5.58
kJ/mol, with a pre-exponential factor equal to 1.41 m
3
/m
2
.h.bar.
Fig. 21: Ln(U He) vs. 1/RT for He for temperatures up to 300 ⁰C and at 200 psig (CMS#4).
3.3.1.2 Binary/Mixed-Gas Separation Experiments
3.3.1.2.1 Experimental Procedure
Mixed-gas permeation experiments have also been carried out with these membranes, in
order to investigate their ability to separate mixtures relevant to the WGS reaction. In all the mixed-
gas experiments reported here, the following experimental conditions were utilized:
Membrane temperature: 300±5
o
C
Trans-membrane pressure gradient (Δ P): 13.1±0.1 bar
Permeate-side pressure atmospheric
For the mixed-gas experiments we either used certified gas mixtures, or we prepared the
feed compositions by mixing different pure gases whose flow rates were set via MFCs to the
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Ln(U
He
)
1/RT × 10
4
(mol/J)
48
appropriate values. For some of the mixtures studied, water vapor was added to the mixed-gas
flow generated via the use of a HPLC pump and a boiler. In the experiments, the feed-side pressure
was adjusted and set by using a back-pressure regulator to an appropriate value. After the system
became stable and the feed flow rate was verified via measurement with a bubble flow meter, the
permeate and reject side flow rates were then measured with a bubble flow-meter as well. For
comparative purposes, single-gas experiments were also carried out with the same membranes,
with both the reject (feed) and permeate sides open to flow.
The lab temperature and pressure were measured with a digital thermometer and an analog
barometer, respectively, in order to convert the flows to STP conditions (m
3
/hr) via Equation (3.4).
The concentration (dry-basis) of each component (other than H2S) in the reject and permeate sides
was measured using an on-line GC. Drager tubes were used to measure the H2S concentration. The
molar flow rate of each component at the reject and permeate sides was calculated based on the
measured overall flow rates for each side and the measured concentration of each component. The
water molar flow rate at the permeate side was measured according to the technique described in
Sec. 3.3.1.1 with the aid of a water adsorbent bed loaded with anhydrous desiccant (Drierite.co).
The mathematical model used to calculate the permeance of each gas in the mixture is
described in Section 3.2. In the model, we have assumed that the main resistance to transport is
due to the membrane, unless otherwise noted.
3.3.1.2.2 Binary-Gas Permeation Measurements
Experiments with binary H2/CO2 and He/N2 gas mixtures with different compositions have
been carried out. The results for two different MPT membranes (CMS#1 and CMS#2) are shown
in Figs. 22 and 23. (In this set of binary gas experiments, after each work day, the flowing gas was
switched to atmospheric pressure argon at 5 sccs, while the permeate side valve was kept closed.)
49
For the H2/CO2 binary gas mixtures (Fig. 22), the module was packed with quartz particles
in order to improve external mass transfer. The results demonstrate an initial drop in the hydrogen
permeance (with respect to its single-gas value). However, subsequent to this initial drop the
hydrogen permeance remained unaffected by the presence of CO2. On the other hand, the
permeance of CO2 remained unaffected by the presence of hydrogen and very close to the single-
gas value. A potential explanation of this behavior is that the hydrogen because of its smaller size
accesses a larger fraction of the pore space than CO2, i.e., a large fraction of the pores exist where
H2 can enter but CO2 cannot, thus the significantly larger permeance of H2 when compared to that
of CO2. Hydrogen, in addition, has access to all pores CO2 has access to and transports through.
CO2 has a higher affinity for the CMS surface than hydrogen, however, and when present in the
mixture adsorbs on the surface of some of these pores, thus blocking access to hydrogen, which
then explains the initial drop in the permeance of hydrogen. Hydrogen, on the other hand, is not
able to block the transport of CO2.
50
Fig. 22: Experimental binary-gas permeances of a CO 2/H 2 mixture as a function of mixture composition
and comparison with the single-gas permeances, CMS#2 (The module in this case was packed with quartz
particles to improve external mass transfer). Numbers in the figure correspond to the chronological order
of the permeance measurement experiments.
Fig.23: Experimental He/N 2 binary-gas permeances as a function of mixture composition and comparison
with the single-gas permeances. CMS#1 (the module in this case was empty).
1
2
3 4
5 6 7
8
9
1 2 3 4
5
6 7
8
9
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8 1
Permeance CO
2
[ m
3
/m
2
*hr*bar]
Permeance H
2
[ m
3
/m
2
*hr*bar ]
H
2
Feed molar fraction
CO pure gas permeance H pure gas permeance H CO 2 2 2
2
0
0.005
0.01
0.015
0.02
0.025
0.03
0
0.2
0.4
0.6
0.8
1
1.2
0 20 40 60 80 100 N
2
permeance[ m
3
/m
2
*hr*bar ]
He permeance [m
3
/m
2
*hr*bar]
N
2
feed molar percentage fraction
He pure gas permeance He N N pure gas permeance
2
2
51
Fig. 24: Experimental He/N 2 binary-gas permeances as a function of mixture composition (taking into
account external mass transfer limitations) and comparison with the single-gas permeances. CMS#1 (the
module in this case was empty).
Fig. 23 shows the results for the He/N2 mixture. The results in Fig. 23 are a bit different
than the behavior shown in Fig. 22. The N2 permeance is lower than its single-gas value, only
approaching that value for the higher concentration mixtures. The permeance of He is lower than
its single-gas value (like in the case of H2 in Fig. 22), but unlike the behavior in Fig. 22 where the
H2 permeance after its initial drop remained unaffected by an increase in the CO 2 content, in Fig.
23 the He permeance continued to drift downwards with N2 content. A potential explanation for
the differences in behavior may be due to the fact that the module in the He/N2 binary gas mixture
experiment was empty and because these CMS membranes are highly permeable, it is possible
that external mass transfer limitations may begin to come into play.
In order to verify this fact, in fitting the experiment data in Fig. 23, the assumption that
external mass transport limitations are negligible was relaxed, and a revised permeation model was
used for the membrane separator operating under such conditions. In the new model, instead of
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0
0.2
0.4
0.6
0.8
1
1.2
0 20 40 60 80 100
N
2
permeance[ m
3
/m
2
*hr*bar ]
He permeance [m
3
/m
2
*hr*bar]
N
2
feed molar percentage fraction
He He pure gas permeance N N pure gas permeance
2 2
52
using the Uj in equations 3.1 and 3.2 we have used Uj, overall (mol/m
2
. h. bar), which is calculated
using the following equations:
1
U
j
overall ⁄
=
1
U
j
⁄
+
1
k
j
p
⁄
(3.6)
k
j
p
=k
j
∗V
mol,j
/𝑃 𝑗 𝐹 (3.7)
k
j
=
0.0732d
e
0.6
D
0.67
u
0.6
d
o
o.4
ϑ
0.27
(3.8)
𝐷 =
1.858 ×10
−3
𝑇 3/2
√
∑1/𝑀 𝑗 𝑃 𝐹 Ω𝜎 12
2
(3.9)
In the above equations, kj is the mass transfer coefficient for component j (m/s), V
mol,j
is the molar
volume of component j (mol/m
3
), 𝑑 𝑒 is the hydraulic diameter, which is four times the cross-
sectional area divided by the wetted perimeter (m), D is the diffusion coefficient of the feed gas
(m
2
/s), u is the velocity of the feed gas (m/s), d0 is the outside diameter of the membrane (m), ϑ
kinematic viscosity of feed gas (m
2
/s), the reactor volume (m
3
), Mj is the molar mass of component
j (kg/ mol), Ω is a temperature-dependent collision integral, , σi is collision diameter of molecule i
(A⁰), and 𝜎 12
=
(𝜎 1
+ 𝜎 2
)
2
is the average collision diameter (A⁰) .
Fig. 24 shows the He and N2 mixed-gas permeances calculated by taking into account the
external mass transfer resistances. The behavior in Fig. 24 pretty much now resembles the behavior
in Fig. 22, The permeance of N2, as with CO2, remains unaffected by the presence of helium and
is very close to single-gas value. The permeance of He continues to be below its single-gas value
but it no longer drifts downwards with N2 content. A potential explanation for this behavior is
very much in line with that given to explain the behavior in Fig. 22. N2 has a higher affinity for
the CMS surface than does He, however, and when present in the mixture adsorbs on the surface
of some of the pores, thus blocking access to helium, which then explains the initial drop in the
53
permeance of helium. Helium, on the other hand, is not able to block the transport of N2 as it is
totally unabsorbing on the surface of pores where N2 has access to..
3.3.1.2.3 Mixed-Gas Permeation Measurements
We have also carried out several mixed-gas (i.e., more than two gases in the mixture)
permeation measurements with the CMS membranes and compared them with single-gas
permeation experiments with the same membranes. These experiments were carried as part of our
efforts to investigate the long-term stability of these membranes. In Table 7 below, we report two
sets of single-gas permeation data (permeances and ideal separation factors with respect to H2) of
MPT membrane CMS#3 (the module in these experiments was empty, but very high feed flow
rates were utilized so that the reported permeance values are free of external mass transport
effects). The first set of single-gas permeances were measured on 9-20-2015 in the morning (a
number of mixed-gas permeation experiments took place after that, as described further below).
The second set of single-gas permeation measurements took place on 9-21-2015 at night, right
after the last mixed-gas permeation experiment was completed. Changes in the permeation
characteristics of these membranes are observed, albeit small for the fast gases (an increase in the
H2 permeance by 5%, and that of CO2 by 6.6%) but significant more so for the slow gases (and
increase in the CO permeance by 21.7%, and that of CH4 by 46.6%). As a result, the (H2/CO2)
ideal separation factor remains virtually unaffected, the (H2/CO) ideal separation factor reduces
by ~9%, and the (H2/CH4) separation factor reduces by ~ 29%.
54
Gas H2 SF CO SF CO2 SF CH4 SF
9-20-15
[m
3
/m
2
*hr*bar ]
1.21 1 0.023 53 0.045 27 0.015 80
9-21-15
[m
3
/m
2
*hr*bar ]
1.27 1 0.028 45 0.048 26 0.022 57
Table 7. Single-gas permeances and ideal separation factors at 200 psig transmembrane pressure and 300
°C (CMS#3).
3.3.1.2.4 CO2-CO-CH4 Mixed-Gas Permeation Measurements:
Table 8 below shows the mixed-gas permeation data generated in an experiment in the
evening of 9-20-15. After this experiment was completed, the flowing gas was switched to Argon
overnight at 2 sccs and 5 psig pressure at 300
o
C while the permeate side valve was kept closed.
The mixed-gas permeances are relatively close to the single-gas values measured on 9-21-2015
(the mixed-gas permeance of CO2 is ~2% smaller, the mixed-gas permeance of CO is ~7% smaller,
and the CH4 permeance is ~18% smaller).
Table 8. Mixed-gas permeances and separation factors, Exp.#1 (CMS#3).
3.3.1.2.5 CO2-CO-CH4-H2-H2S Mixed-Gas Permeation Measurements
This experiment was carried out on the morning of 9-21-15. Table 9 below reports the
experimental data. Note that the mixed-gas permeances for the slow gases CO, CH4, and CO2 are
Gas Molar fraction
(%) in the feed
Feed flow
rate cc/s
Permeate side
flow rate cc/s
Mixed-gas
permeances
m
3
/m
2
*hr*bar
Single-gas
permeances
on 9-21-15,
m
3
/m
2
*hr*bar
CO2 50
9.97 0.37 0.047 0.048
CH4 20 3.98 0.048 0.018 0.022
CO 30 6.32 0.12 0.026 0.028
Total 100 20.27 0.53
55
virtually identical between experiments #1 (Table 8) and #2 (Table 9), which indicates that the
overnight treatment in Ar at 300
o
C and the mixed gas experiments do not appear to affect the
properties of the membrane. The mixed-gas H2 permeance is ~ 5% less than the single-gas value
measured on 9-20-15 and 9.5% less than the single-gas value measured on 9-21-15.
Gas Molar
fraction (%)
in the feed
Feed flow
rate cc/s
Permeate
side flow
rate cc/s
Mixed-gas
permeances
m
3
/m
2
*hr*bar
Single-gas permeances
on 9-21-15,
m
3
/m
2
*hr*bar
H2 40.87 12.5 5.53 1.15 1.27
CO 14.79 4.52 0.079 0.027 0.028
CH4 11.8 3.61 0.042 0.018 0.022
CO2 31.84 9.74 0.30 0.048 0.048
H2S 0.7 0.21 0.0021 0.015
Total 100 30.58 5.96
Table 9. Mixed-gas permeances and separation factors, Exp.#2 (CMS#3).
3.3.1.2.6 CO2-CO-CH4-H2-H2S-H2O Mixed-gas Permeation Measurements
This experiment was carried out on the evening of 9-21-15. Table 10 below reports the
experimental data. Note that the mixed-gas permeances for the slow gases CO, CH4, and CO2 are
virtually identical between experiments # 2 (data shown in Table 9) and # 3 (table 10), which
indicates that the wet mixed-gas experiments did not appear to affect the properties of the
membrane. Also, the presence of water had no effect on the permeances of the slow gases. The
mixed-gas H2 permeance is virtually identical to the single-gas value measured on 9-20-15 and
~4% less than the single-gas value measured on 9-21-15.
56
Gas Molar
fraction (%)
in the feed
Feed flow
rate cc/s
Permeate
side flow
rate cc/s
Mixed-gas
permeances
m
3
/m
2
*hr*bar
Single-gas permeances
on 9-21-15,
m
3
/m
2
*hr*bar
H2 28.24 12.5 4.63 1.24 1.27
CO 10.22 4.52 0.056 0.028 0.028
CH4 8.15 3.61 0.03 0.019 0.022
CO2 22.00 9.74 0.2 0.046 0.048
H2S 0.48 0.214 0.0021 0.022
H2O 30.89 13.67 4.21 0.94
Total 100 44.24 9.12
Table 10. Mixed-gas permeances and separation factors, Exp.#3 (CMS#3).
3.4 Conclusion
The permeances of pure N2, H2, CH4, CO2, CO, and H2O gases have been measured at
temperatures of up to 300 ⁰C and 200 psig of feed pressure. The selectivity between hydrogen and
the other slower gases, that are typically present in coal syngas, is high at temperatures higher than
250 ⁰C.
Several permeation experiments of binary and multi-gas mixtures in the presence of water
and H2S have also been carried out and are reported here. The mixed-gas permeance values are
fairly close to the measured pure-gas permeances (measured at the end of the 2-day series of
experiments), and remain quite stable (compare the data in Tables 8-10). The results demonstrate
the stability in the hydrogen permeance and the selectivity between hydrogen and other gases in
the presence of CO, H2O, H2S, and under our experimental conditions. These results demonstrate
that the CMS membrane is a strong candidate for use in hydrogen separation and for WGS reaction
enhancement in the processing of coal syngas via the IGCC process.
57
4 Chapter 4: Membrane Reactor Study
4.1 Membrane Reactors
Studies of dehydrogenation reactions using membrane reactors (MR) first appeared in the
scientific literature in the late 1960s [59], and interest in the topic has continued since. Such
reactors combine the catalytic reaction and hydrogen separation functions in a single unit [61]
(there are also various other types of MR where the membrane provides a means to deliver, in a
controlled way, a certain reagent or to remove a reaction intermediate to favorably impact
selectivity, but these MR are not of interest in this Thesis, and have been reviewed elsewhere [62]).
With an increasing interest in recent years in hydrogen as a potential clean energy carrier, scientific
interest in high-temperature MR, utilizing inorganic membranes, as an alternative to traditional
reactors for carrying out the WGS and the steam reforming reactions has substantially increased
as well [60].
Our focus here, in particular, is in MR for which the membrane is not catalytic and this
function, instead, is provided by a bed of catalysts packed in proximity to the membrane (such
reactors are known as packed-bed membrane reactors or PBMR). The key membrane function is
to separate in situ hydrogen, the desired product, from the reaction mixture. By doing so, however,
the membrane also provides the added benefit of increasing the reactor’s yield, making it possible
to achieve higher conversions than the traditional steam reforming or WGS reactors at the same
operating conditions..
Our key focus in this Thesis is WGS-MR, particularly their application in the IGCC
environment. There are a number of scientific studies in the current technical literature on using
MRs to carry out the WGS reaction [23, 59, 60, 71-75], in general, and under conditions relevant
to the IGCC application, in particular [65, 66 ]. A challenge with using H2-selective membranes
58
in the WGS reaction in coal-based gasification plants is membrane stability under high-pressure
and high-temperature conditions and in the presence of steam and possibly other trace components,
such as H2S and various tars and organic hydrocarbons. Typical membrane materials used for H2
separation [63, 64] in WGS-MR are: (i) dense metals, typically Pd and its alloys, which have nearly
infinite H2 selectivity, but are expensive, are poisoned by H2S, even at low concentrations, and are
subject to H2 embrittlement; (ii) dense polymers, which are inexpensive, but have low H2
selectivity and are subject to thermal degradation; (iii) amorphous silica, which has high cost and
is hydrothermally unstable; and (iv) porous carbons that have relatively lower selectivity than Pd
membranes, but are less costly and more robust to the WGS reaction conditions [65]. They are
also the primary focus of this Thesis.
A recent DOE report [76] describes various IGCC processes with and without CO2 capture.
The report concludes that to reach the goal of 90 % CO2 capture in a new IGCC plant using the
conventional Selexol™ process for carbon capture necessitates an increase in the levelized cost of
electricity (LCOE) by ~30%, which is higher than the <10% DOE target set for CO2 capture
processes integrated with IGCC [66]. The high cost of using conventional processes in IGCC with
CCS motivates, therefore, this study where we investigate the application of CMS membranes in
a WGS-MR operating with simulated coal-derived syngas in combination with an industrial
Co/Mo sour-shift catalyst. On the top of the aforementioned benefits of using such a reactor (in
situ H2 separation, enhanced yield), an additional important benefit for the proposed WGS-MR
technology for IGCC is that CO2 can be removed at high pressures and high concentrations (in the
MR reject stream), making it significantly less costly to implement CCS when compared to direct
CO2 removal from the flue-gas of a conventional power-plant [67].
59
4.2 Modeling
To analyze the MR experiential data generated in this Thesis, we have utilized an
isothermal co-current flow (feed to permeate) 1-D MR model previously utilized by our group for
describing such reactors [48]. Mass balances for each component in the feed and permeate sides
are described by equations 4.1 and 4.2, respectively.
𝑑 𝑛 𝑗 𝐹 𝑑𝑉
= −𝛼 𝑚
𝑈 𝑗 (𝑃 𝑗 𝐹 −𝑃 𝑗 𝑃 )+ 𝜈 𝑗 (1−𝜀 𝑣 ) 𝛽 𝑐 𝜌 𝑐 𝑟 𝐹 (4.1)
𝑑 𝑛 𝑗 𝑃 𝑑𝑉
= 𝛼 𝑚
𝑈 𝑗 (𝑃 𝑗 𝐹 −𝑃 𝑗 𝑃 ) (4.2)
In the above equations, is the molar flow rate (mol/h) for component j in the feed side, the
corresponding molar flow rate (mol/h) in the permeate side, V the reactor volume variable (m
3
),
the surface area of the membrane per unit reactor volume (m
2
/m
3
), υj the stoichiometric
coefficient for component j (negative for reactants and positive for products), the bed porosity
in the feed side, the fraction of the solid volume occupied by the catalysts, the catalyst
density (g/m
3
), and the WGS reaction rate (mol/g. h). The pressure drop in the packed-bed is
calculated using the Ergun equation (equations 2.2-2.4).
The CO conversion is defined by Eq. 4.3.
𝑋 𝐶𝑂
=
𝑛 𝐶𝑂𝑜 𝐹 − ( 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝐹 + 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝑃 )
𝑛 𝐶𝑂𝑜 𝐹 (4.3)
F
j
n
P
j
n
m
v
c
c
F
r
60
where 𝑛 𝐶𝑂𝑜 𝐹 is the CO molar flow rate at the inlet (mol/h), 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝐹 is the CO molar flow rate at the
exit of the reactor’s feed side (mol/h), and 𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡 𝑃 is the CO molar flow rates at the exit of the
reactor’s permeate side (mol/h).
The hydrogen recovery ( ) is given by equation 4.4.
𝑅𝑒
𝐻 2
=
𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 (𝑛 𝐻 2,𝑒𝑥𝑖𝑡 𝐹 + 𝑛 𝐻 2,𝑒𝑥𝑖𝑡 𝑃 )
(4.4)
Where 𝑛 𝐻 2,𝑒𝑥𝑖𝑡 𝐹 is the hydrogen molar flow rate at the exit of the reactor’s feed side (mol/h) and
𝑛 𝐻 2,𝑒𝑥𝑖𝑡 𝑃 is the hydrogen molar flow rate at the exit of the reactor’s permeate side (mol/h).
The hydrogen purity (dry-basis, ) is given by equation 4.5.
𝑃 𝑢 𝐻 2
=
𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 +𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝑃 +𝑛 𝐶𝑂 2,𝑒𝑥𝑖𝑡
𝑃 +𝑛 𝐶𝐻 4,𝑒𝑥𝑖𝑡
𝑃 (4.5)
where 𝑛 𝐶𝑂
2,𝑒𝑥𝑖𝑡 𝑃 is the carbon dioxide molar flow rate at the exit of the reactor’s permeate side
(mol/h) and 𝑛 𝐶𝐻
4,𝑒𝑥𝑖𝑡 𝑃 is the methane molar flow rate at the exit of the reactor’s permeate side
(mol/h).
4.3 Experimental Procedure
Several membrane reactor experiments have been carried-out in order to investigate how
both the membrane and catalyst perform under the WGS reaction environment. The experimental
procedure followed in these experiments is the same as that described in Section 2.1. In all the MR
experiments reported here, we again used 10 g of catalyst and a certified mixed-gas as feed with
composition (on a dry basis) of H2:CO:CO2:CH4:H2S =2.6:1:2.13:0.8:0.05 (intended to simulate a
coal gasifier’s exit composition), and a nearly stoichiometric H2O/CO ratio in the feed of 1.1. We
have used two different membrane configurations in the WGS-MR experiments reported in this
2
Re
H
2
H
Pu
61
Thesis. In the first series of MR experiments, the membrane had both ends open, and steam was
used as a sweep stream in the permeance side of the membrane. In the second set of MR
experiments, the MR had a so-called candle-filter configuration. Specifically, in this MR
configuration use was made of a CMS membrane with one end closed. The membrane was attached
to the reactor module only on the open-end, with the other end free-standing. The reason for
employing the candle-filter MR configuration is because it allows to install the membrane in the
module, with its dead-end side not being affixed to the module, so as to minimize damage to the
membrane due to potential thermal expansion mismatch between the membrane and the SS
module.
Throughout the experiments we have utilized single-gas permeation measurements to
monitor the state of the membrane through the membrane reactor experiments. As noted
previously, for these membranes single-gas and mixed permeances are quite close to each other,
and single-gas permeation experiments are quite less time consuming to carry out.
4.4 Membrane Reactor Results
Table 11 shows the results of the single-gas permeation measurements before the initiation
of the first series of MR experiments (carried out on 10-20-2015. In this series of experiments, we
utilized membrane CMS#3, whose single-gas permeation properties were previously measured on
9-21-2015 and are shown in Table 7. During the ensuing period from 9-21-2015 to the date that
gas permeation measurements were carried out prior to the initiation of the MR experiments (10-
20-2015), the membrane module was first cooled down to room temperature overnight under an
He atmosphere for which the flow rate was less than 5 sccs and the pressure was 5 psig. Then, the
module was loaded with 10 g of WGS catalyst thoroughly mixed with ground quartz particles (of
the same size, 600-800 𝝁 m) and then catalyst activation took place between 9-23-15 to 10-14-15,
62
the same way as explained in section 2.3. Note that the membrane has undergone changes during
that period. Specifically, the permeance of the fast gas (hydrogen) increased by ~6%, while the
permeance of CO increased by ~21%, the permeance of CO2 by ~9%, and the permeance of CH4
by 41%.
Table 11 shows the single-gas permeation measurements with same membrane after the
MR experiments had been completed. There are minor changes in the membrane permeation
properties observed which are, however, significantly less pronounced that those observed
between the set of measurements on 9-21-2015 and 10-20-2015. Specifically, for the set of single-
gas permeation experiments immediately preceding and following the MR experiments the H 2
permeance increased by ~3%, the CO permeance increased by ~6%, the permeance of CO2 by
~4%, and the permeance of CH4 by ~6%.
Gas H 2 SF CO SF CO 2 SF CH 4 SF
Permeance measured
on 10-20-15
[m
3
/m
2
*hr*bar ]
1.35 1 0.034 39 0.054 25 0.031 44
Permeance measured
on 10-29-15
[m
3
/m
2
*hr*bar ]
1.39 1 0.036 38 0.056 25 0.033 42
Table 11. Single-gas permeances and ideal separation factors at 300 °C and 200 psig (CMS#3).
After the completion of the single-gas permeation experiments, the MR experiments were
carried-out. The first set of MR experiments was carried out at 300 ⁰C and feed pressure of 200
psig. At each WC/FCO, the MR permeate side was first closed, so that the reactor functions as a
PBR. The feed pressure was then adjusted and set to an appropriate value by using a back-pressure
valve. After the system pressure became stable (typically, after ~1 hr), the PBR conversion was
measured. After the PBR experiment was completed, the permeate side was opened in order to
63
conduct the MR experiments. Steam was added as a sweep to the permeate side, when chosen to
do so. The MR feed pressure was adjusted and set using a back-pressure valve to an appropriate
value. After the system pressure became stable (after approximately 1 hr), the permeate side and
reject side flow rates were measured with a bubble flow-meter. The temperature of the membrane
and reactor bed was checked with the aid of a sliding thermocouple inside a thermo-well in order
to ensure that the reactor operates under isothermal conditions. The lab temperature and pressure
were measured with a digital thermometer and an analog barometer, respectively, in order to
convert the flows to STP conditions (m
3
/hr). The compositions of the permeate side and reject side
flows were also measured with a GC. The PBR and MR experiments for each WC/FCO lasted for
about 11 hr in total. In between experiments, the system was kept under flowing (5 sccs) argon at
15 psig pressure while the permeate side valve was kept closed.
Fig. 25 shows the measured CO conversion vs. (weight of catalyst /molar flow rate of CO
or W/FCO) for the membrane reactor for three different steam sweep ratios employed at the
permeate side. Shown on the same Figure are the experimentally measured conversions in the
PBR. The conversions in the MR is higher than those of the conventional packed-bed reactor, and
the conversion increases with W/FCO. The membrane conversion increases with the sweep ratio.
The use of steam as a sweep gas is advantageous for the WGS-MR in that it dilutes the H2
concentration in the permeate stream, thus increasing the permeation rate. In addition, the use of
steam as a sweep helps to reduce the loss of reactant H2O from the reactor side. The latter
beneficial effect is apparent in the conversion data with no sweep, where the conversion drops
after WC/FCO=110, as a result of H2O reactant losses to the permeate side.
Fig. 26 shows the hydrogen purity (dry-basis) vs. (weight of catalyst /molar flow rate of
CO) for the membrane reactor for three different sweep ratios. The purity slightly increases with
64
an increase in the sweep ratio, and also declines with W/FCO. Fig. 27 shows the hydrogen recovery
vs. (weight of catalyst /molar flow rate of CO) for the membrane reactor for same three sweep
ratios. The hydrogen recovery increases slightly with increasing sweep ratios as well as W/FCO
increases. Because of the high hydrogen permeance of these membranes, high recoveries are
attained.
Fig. 25: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (MR) (Weight
of catalyst 10 g (CMS#3))
Fig. 26: Hydrogen purity vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (MR).
(Weight of catalyst 10 g (CMS#3)).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 50 100 150 200
CO conversion
Weight of catalyst / Molar flow rate of CO
SW=0 SW=0.1 SW=0.3 Packed-bed reactor
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 20 40 60 80 100 120 140 160 180
H
2
purity
Weight of catalyst / Molar flow rate of CO
SW=0 SW=0.1 SW=0.3
65
Fig. 27: Hydrogen recovery vs. (weight of catalyst/molar flow rate of CO) for membrane reactor (Weight
of catalyst 10 g (CMS#3)).
In section 4.2 we present an isothermal model 1-D model to describe the behavior of the
WGS-MR. We employ this model here to fit the experimental data. For the catalytic kinetics we
use the empirical reaction rate expression (equation 2.6), as well as the rate expression
corresponding to the direct oxidation microkinetic model (equation 2.12). For the membrane
properties we used the pure gas permeance measured on 10-29-15 ( Table 11). Additionally, water
permeance was set as an adjustable parameter estimated with the aid of the “lsqcurvefit” function
in MATLAB using the model explained in section 4.2. The separation factor between hydrogen
and water was calculated to be ~ 5, which lies in the range of values (1-10) we typically measure
with such membranes. Figs. 28, 29, and 30 show the conversion vs. (weight of catalyst /molar flow
rate of CO) for the membrane reactor for different sweep ratios in the permeate side and the
corresponding fits of the model using both the empirical and microkinetic models. The figures
predict the behavior of the MR generally well, particularly for finite sweep ratios, but less so for
the case with no sweep (e.g., Fig. 28). Figs. 31, 32, and 33 show the hydrogen purity (dry-basis)
0.84
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
0 50 100 150 200
H
2
recovery
Weight of catalyst / Molar flow rate of CO
SW=0 SW=0.1 SW=0.3
66
vs. (weight of catalyst /molar flow rate of CO) for the membrane reactor for different sweep ratios
in the permeate side and the corresponding fits using both the empirical and microkinetic models.
The model predicts the H2 purity quite well at different sweep ratios. Figs. 34, 35, and 36 show
the hydrogen recovery (dry-basis) vs. (weight of catalyst /molar flow rate of CO) for the membrane
reactor for different sweep ratios in the permeate side and the fitted results using both the empirical
and microkinetic models. The model again predicts the H2 recovery quite well for different sweep
ratios.
Fig. 28: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and no sweep (CMS#3)).
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80 100 120 140 160 180
CO conversion
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation Equilibrium Conversion
67
Fig. 29: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and 0.1 sweep ratio (CMS#3)).
Fig. 30: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and 0.3 sweep ratio (CMS#3)).
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80 100 120 140 160 180
CO conversion
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation Equilibrium Conversion
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
20 40 60 80 100 120 140 160 180
CO conversion
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation Equilibrium Conversion
68
Fig. 31: Hydrogen purity vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and no sweep (CMS#3)).
Fig. 32: Hydrogen purity vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and 0.1 sweep ratio (CMS#3)).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80 100 120 140 160 180
H
2
purity
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80 100 120 140 160 180
H
2
purity
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
69
Fig. 33: Hydrogen purity vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g and 0.3 sweep ratio (CMS#3)).
Fig. 34: Hydrogen recovery vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight
of catalyst 10 g and no sweep (CMS#3)).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80 100 120 140 160 180
H
2
purity
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
20 40 60 80 100 120 140 160 180
H
2
recovery
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
70
Fig. 35: Hydrogen recovery vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight
of catalyst 10 g and 0.1 sweep ratio (CMS#3)).
Fig. 36: Hydrogen recovery vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight
of catalyst 10 g and 0.3 sweep ratio (CMS#3)).
The second set of MR experiments was carried out with the reactor operating in a candle-
filter configuration employing a CMS membrane (CMS#5) with one end closed. (The catalyst in
these experiments was first partially pre-sulfided in another module, and then transferred to the
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
0 20 40 60 80 100 120 140 160 180
H
2
recovery
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
0 20 40 60 80 100 120 140 160 180
H
2
recovery
Weight of catalyst / Molar flow rate of CO
Empirical model Direct oxidation
71
module containing CMS#5, when the membrane contained in the prior module got accidentally
damaged. The catalyst activation was completed in the CMS#5 module using the activation
procedure previously described). Select measurements of the catalyst’s activity (with the reactor
operating as a PBR with the permeate side of the MR closed) indicated that the catalyst followed
the empirical rate expression described by equation 2.6. The candle-filter-type membrane CMS#5
had a length of 25.4 cm, out of which 19 cm represent the active membrane length, with glazed
impermeable regions on either side of the active membrane. The catalyst was thoroughly mixed
with ~ 180 g of quartz particles of equal size (600-800 𝝁 m), and then loaded in the annual space
in between the membrane and the reactor wall. In modeling the experimental data, because of the
membrane consisting of two inactive regions surrounding the core active region, the reactor was
modeled as a three reactors in series: a packed-bed reactor followed by a membrane reactor
followed by another packed-bed reactor. The weight of catalyst for each reactor was taken
proportional to its length. To model the PBR’s we used equation 2.1, and for modeling the MR we
used equations 4.1 and 4.2. The pressure drop in the reactor reject side was calculated using the
Ergun equation (equations 2.2-2.4).
Table 12 shows the results of the single-gas permeation measurements before and after the
MR experiments in order to investigate the selectivity stability of the candle filter membrane
(CMS#5) under the WGS-MR environment. (Note that the CO permeance in Table 12 is measured
in a H2, CO, and CO2 mixture following the method described in chapter 3. Total flow rate of mix
gas was 30 sccs with 13% of CO2, 22% of CO, and 65 % of H2).
72
Gas H 2
SF (H 2/
H 2)
CO
*
SF(H 2/CO) CO 2
SF(H 2/
CO 2)
CH 4
SF(H 2/
CH 4)
[m
3
/m
2
*hr*bar ]
1-18-17
1.3 1 0.018 70 0.039 33 0.011 123
[m
3
/m
2
*hr*bar ]
2-1-17
1.31 1 0.024 54 0.042 30 0.013 98
Table 12. Single-gas permeances and ideal separation factors at 250 °C and 200 psig (CMS#5).
* The CO permeance reported in the Table is measured using a mixed-gas.
The results of the second set of MR experiments with the candle filter membrane are shown
in Fig. 37, Fig. 38, and Fig. 39. Fig. 37 shows the conversion vs. (weight of catalyst /molar flow
rate of CO) for the membrane reactor, the packed-bed reactor, and the results of the model fits. In
the simulations, we used the empirical global rate expression of equation 2.6. For the membrane
permeation, we used the average of gas permeances measured on 1-18-17 and 2-1-17 (Table 12).
In these simulations the water permeance was considered as an adjustable parameter and its value
was estimated with the aid of the “lsqcurvefit” function in MATLAB using the model explained
in section 4.2. The separation factor between hydrogen and water was calculated to be ~ 1.1 which
is in the typical range of values experimentally measured with these membranes. The conversion
of the MR is higher than that of the conventional packed-bed reactor. Fig. 38 shows the
experimental hydrogen purity (dry-basis) vs. (weight of catalyst /molar flow rate of CO) for the
membrane reactor and the model fit. The model fits the experimental H2 purity quite well. Fig. 39
shows the experimental hydrogen recovery vs. (weight of catalyst /molar flow rate of CO) for the
membrane reactor and model fit. The model fits the experimental H2 recovery equally well.
In order to monitor the state of the membrane throughout the series of experiments, we
have measured its He and N2 permeability. The data are shown in Tables 17 and 18 in the
Appendix. In Table 17 we indicate the permeance data measured with the membrane module being
73
empty. Prior to sending to USC, the permeation characteristics were measured at M&P
laboratories. M&P provides us, unfortunately, only with raw data (which are also shown in Table
17) so we convert those first to permeation values at STP conditions (by assuming that the
laboratory temperature at M&P was 15
o
C and the atmospheric pressure 1 atm). When comparing
the M&P permeance data measured at similar conditions (temperature of 250
o
C, pressure of 20
psig) the He permeance is slightly lower (~3%) while the N2 permeance was somewhat larger
(~7%). Such differences are, however, quite common in measurements among various
laboratories, and involve uncertainties in the experimental procedures and equipment used. Table
17 also indicates the permeance for both He and N2 (measured at USC) at 250
o
C and two different
pressure (20 psig and 200 psig). The values are experimentally indistinguishable, indicating a very
high quality membrane.
Subsequent to the measurement of the transport properties of the membrane in the empty
module, the module was filled with catalyst and quartz, as noted above and its N 2 permeance ws
measured at 250
o
C and 200 psig and its value (see Table 18) remained close (0.013 m
3
/m
2
*hr*bar)
to the value measured prior to the loading with catalyst, indicative that no damage was done to the
membrane during the loading procedure. Subsequently the catalyst activation procedure was
initiated involving to exposure to high concentrations of H2S. The He and N2 permeance of the
membrane were then re-measured (at 250
o
C and 200 psig) after ~ 7 days of activation, on 1/18/17.
The He permeance had declined to 1.35 m
3
/m
2
*hr*bar and the N2 permeance (equal in value to
the CH4 permeance measured on the same day—see Table 12) had declined to 0.011
m
3
/m
2
*hr*bar. We are not entirely sure what causes these changes in the membranes permeation
properties during the activation procedure, but most likely relates to H2S adsorption in the
membrane pore structure. Subsequently the MR experiments were initiated which lasted for 10
74
days, for which data are shown in Figs 37-39. During that period we monitored the state of the
membrane via measurements of its He and N2 permeances. The N2 permeance slowly recovered
and stabilized to its initial value of ~ 0.014 m
3
/m
2
*hr*bar (and so did the CH4 permeance, see
Table 12). On the other hand, the He permeance (and the H2 permeance as well, see table 12)
never recovered. We attribute this behavior to the fact that the changes in N2 (CH4) permeation are
due to reversible adsorption of H2S in the mesopores (pinholes/cracks) through which gases
permeate through the membranes. The changes in He and H2 permeation rates, on the other hand,
are due to irreversible adsorption of H2S in the micropores through which these fast gases transport
through.
Fig. 37: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g, reactor temperature 250 °C, and feed pressure 195 psig (CMS#5)).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 20 40 60 80 100 120 140 160 180
CO conversion
Weight of catalyst / Molar flow rate of CO
Empirical model Packed-bed reactor
75
Fig. 38: Hydrogen purity vs. (weight of catalyst /molar flow rate of CO) for membrane reactor (Weight of
catalyst 10 g, reactor temperature 250 °C, and feed pressure 195 psig (CMS#5)).
Fig. 39: Hydrogen recovery vs. (weight of catalyst/molar flow rate of CO) for membrane reactor (Weight
of catalyst 10 g, reactor temperature 250 °C, and feed pressure 195 psig (CMS#5)).
4.5 Impact of Tar and Organic Vapors on Membrane Reactor Performance
The impact of the model tar and organic vapors on the kinetics of the water-gas shift reaction
was discussed in section 2.6. The results from these experiments show that the presence of
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 20 40 60 80 100 120 140 160 180
H
2
purity
Weight of catalyst / Molar flow rate of CO
Empirical model
0
0.2
0.4
0.6
0.8
1
1.2
0 20 40 60 80 100 120 140 160 180
H
2
recovery
Weight of catalyst / Molar flow rate of CO
Empirical model
76
naphthalene (model tar) and toluene (model organic vapor) do not have an impact on the
performance of the catalyst at a temperature of 250 ⁰C and a pressure of 200 psig over a three-hour
reaction time period. We have also conducted the membrane reactor experiment in the presence of
tar. We have used the same simulated coal syngas used for the evaluation of the reaction kinetics
in the PBR together with 0.8 vol. % naphthalene and 6.4 vol. % toluene added to the syngas feed.
To generate the tar/organic vapor/syngas mixture, the naphthalene was dissolved in toluene in an
appropriate amount and was added to the feed flow with the aid of an HPLC pump and a boiler.
During idle times, in between experiments, the membrane was kept under an N2 flow (5 sccs) at
15 psig while the permeate side was closed.
The first set of experiments carried out in the presence of the simulated tar took place for
each WC/FCO directly after the reaction test in the presence of tar as explained in section 2.6. Table
13 shows the permeance measurements before and after the first set of membrane reactor
experiments were carried out in the presence of tar. The permeances after the first set of
experiments at 250 ⁰C are less than the permeances before the simulated tar experiments. A
possible explanation is that the tar component condensed on the membrane pores. Deposits of tar
contaminants on the module and membrane were also observed by our industrial partner at
temperatures below 270 ⁰C [80].
Gas H 2
SF(H 2/
H 2)
CO
*
SF(H 2/CO) CO 2
SF(H 2/
CO 2)
CH 4
SF(H 2/
CH 4)
He N 2 SF(He/N 2)
[m
3
/m
2
*hr*bar ]
3-8-17
1.35 1 0.029 47 0.046 30 0.014 100
1.34 0.015 92
[m
3
/m
2
*hr*bar ]
3-14-17
1.05 1 - - 0.03 35 0.011 95
1.05 0.011 95
Table 13: Single-gas permeances and ideal separation factors at 250 °C and 200 psig (CMS#5).
* The CO permeances reported in the Table were measured in mixed-gas experiments (Total flow rate of
mix gas was 30 sccs with 13% of CO 2, 22% of CO, and 65 % of H 2).
77
Fig. 40 shows the membrane reactor conversions in the presence of simulated tar and
compares these results with the MR results in Fig. 37. The conversions are significantly less than
previous measurements for which the simulated tar was not present in feed stream. A possible
reason could be due to membrane fouling in the presence of the simulated tar, which may have
inhibited the ability for hydrogen molecules to permeate at adequate fluxes through the membrane
because of blockages of the membrane’s surface pores by the naphthalene molecules.
Fig. 40: Conversion vs. (weight of catalyst /molar flow rate of CO) for membrane reactor in the presence
of tar and conversion results from Fig. 37 (Weight of catalyst 10 g, reactor temperature 250 °C, and feed
pressure 195 psig (CMS#5)).
In order to investigate the impact of temperature on the membrane fouling observed in Fig.
40, a second set of MR experiments was conducted at the higher temperature of 300 ⁰C. In addition,
the single-gas permeances were measured before the MR experiments in the presence of simulated
tar at 300 ⁰C were initiated and after the MR experiments were completed, and the data are shown
in Table 14. The single-gas permeances before and after the MR experiments remain virtually
unaffected, which is in line with our prior experience with these membranes that shows that
exposure to model tar and organic compounds leave these membranes completely unaffected [16].
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 10 20 30 40 50 60 70 80 90
CO conversion
Weight of catalyst / Molar flow rate of CO
Tar model Without tar model data from Fig. 37
78
Gas H 2
SF(H 2
/ H 2)
CO
*
SF(H 2
/CO)
CO 2
SF(H 2
/ CO 2)
CH 4
SF(
H 2/
CH
4)
He N 2 SF(He/N 2)
[m
3
/m
2
*hr*bar ]
3-21-17
1.54
1 0.028 55 0.044 35 0.014 111
1.56 0.016 101
[m
3
/m
2
*hr*bar ]
3-24-17
1.54
1 0.028 54 0.045 33 0.014 110
1.56 0.016 100
Table 14: Single-gas permeances and ideal separation factors at 300 °C and 200 psig (CMS#5).
* The CO permeances reported in the Table were measured in mixed-gas experiments (Total flow rate of
mixed-gas was 30 sccs with 13% of CO 2, 22% of CO, and 65 % of H 2).
Additionally, the membrane reactor conversion at WC/FCO= 55 at 300 ⁰C and at 200 psig
was measured. The experimental conversion at this condition in the presence of simulated tar and
organic vapor was measured to be 55%, while the simulation result shows a conversion of 54%.
In our model, we used the values of permeance measured on 3-24-17 (Table 14) and assumed the
selectivity between hydrogen and water to be 1.1. From the measured and model results it can be
concluded that the presence of tar does not have an impact on the MR performance at this
temperature and pressure.
4.6 Conclusion
The experimental results to date show the ability of the MR reactor to operate under the
desired conditions and to improve the efficiency of the WGS reaction. Additionally, the
experimental results demonstrate the potential for the MR to carry out the in-situ separation of
hydrogen using a CMS membrane and a commercial Co/Mo sour-shift catalyst. We conclude from
the results that the CMS-MR is a good candidate technology for incorporation into IGCC power
plants for environmentally-benign power generation. Fitting the experimental data with an
isothermal 1-D MR model indicates that the model is adequate to fit the MR experimental data
79
generated in the lab-scale MR system.
80
5 Chapter 5: Modeling
5.1 Introduction
Up to now, we have used an 1-D isothermal model to describe the experimental data
generated in our lab-scale packed-bed and membrane reactors. In this Chapter we develop models
which may be more appropriate to describe the behavior of other types of reactors, which may find
potential application during the field-scale testing and evaluation of the technology. Specifically,
we study: (i) non-isothermal, adiabatic 1-D membrane reactors, (ii) 2-D membrane reactors where
we relax the assumption of no-radial dependence, and (iii) multistage configurations combining
PBRs followed by membrane separators. In this chapter, these different types of membrane
reactors are analyzed and discussed and their advantages/disadvantages over isothermal membrane
reactors for the WGS reaction are discussed.
5.2 1-D Adiabatic Packed-Bed Reactor
For the 1-D non-isothermal model for the PBR the following assumptions were made: plug-
flow conditions with negligible radial temperature/concentration gradients, constant bed porosity
in the axial and radial directions, and ideal gas behavior. In addition to equations 2.1-2.4, for the
catalytic kinetics we use the rate expression corresponding to the direct oxidation microkinetic
model (equation 2.12). The following energy balance equation is used in order to describe the
adiabatic packed-bed reactor:
(∑𝑛 𝑗 𝐶 𝑝 𝑗 )
𝑑𝑇
𝑑𝑉
=(−∆𝐻 𝑅𝑥
𝑇 ((1−𝜀 𝑣 )𝛽 𝑐 𝜌 𝑐 𝑅 𝐶 𝑂 )) (5.1)
𝐶 𝑝 𝑗 =𝐴 𝑗 +𝐵 𝑗 .𝑇 +𝐶 𝑗 .𝑇 2
+𝐷 𝑗 .𝑇 −2
(5.2)
81
∆𝐻 𝑅𝑥 ,𝑇 = ∆𝐻 298
𝐾 +∫ ∆(∑𝐶 𝑝𝑗
) 𝑑𝑇 𝑇 298
(5.3)
where CPj [J/mol∙K] is the heat capacity of component j where in coefficients are taken from
Elliott and Lira [79] , ΔHRxT [J/mol] the heat of the WGS reaction at temperature T, ∆𝐻 298
𝐾 the
heat of the WGS reaction at reference temperature, and T
[K] the feed side temperature.
COMSOL was used to model the adiabatic packed-bed reactor using equations 2.1-4 and
5.1, with the Ergun equation being used to calculate the pressure drop (in the simulations we
assumed, for simplicity, that the viscosity of gas is constant with change of the temperature. For
laboratory reactors where the pressure drops are normally small this a reasonable assumption. For
industrial scale reactors, one must employ the appropriate equation for the mixture viscosity). Fig.
41 shows the CO conversion along the length of the reactor, and Fig. 42 shows the temperature of
the bed along the length of the reactor for an inlet temperature of 300 °C (other conditions are
indicated in the captions of the Figures ).
Fig. 41: CO conversion vs. dimensionless reactor length for adiabatic and isothermal operation (Weight of
catalyst 10 g, W C/F CO is 55, pressure is 200 psig, inlet temperature is 300 ⁰C, and feed composition is H 2:
CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12).
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
CO conversion
Dimensionless reactor length
Isothermal Adiabatic
Equilibrium conversion at 573.15 K Equilibrium conversion at 629.83 K
82
Fig. 42: Temperature along the length of the reactor for adiabatic operation (Weight of catalyst 10 g,
W C/F CO is 55, pressure is 200 psig, inlet temperature is 300 ⁰C, and feed composition is H 2: CO: CO 2:
CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12).
Because the WGS is an exothermic reaction, as Fig. 42 indicates the bed temperature rises
along the length of the reactor and as a result the reaction rate accelerates the net outcome being
that the reactor reaches equilibrium. However, for the corresponding isothermal reactor that is not
the case with the reactor not yet reaching equilibrium. As a result, the conversion reached in the
adiabatic reactor is lower than that of the isothermal reactor, because increasing the temperature
of the bed reduces the equilibrium conversion of the WGS reaction. Fig. 41 also demonstrates this
decrease in equilibrium conversion for the adiabatic reactor (for which the outlet temperature is
357 °C) compared to that for the isothermal reactor (for which the outlet temperature is 300 °C).
Figure 43 compares the exit conversions (left axis) for the isothermal and adiabatic PBR’s
for a range of W/FCO values. Shown on the same Figure (right axis) is the exit temperatures of the
adiabatic PBR for the same range of W/FCO values. According to Fig. 43, for the range of WC/FCO,
570
580
590
600
610
620
630
640
0 0.2 0.4 0.6 0.8 1
Temperature [K]
Dimensionless reactor length
83
studied the conversion of the adiabatic reactor is higher than that for the isothermal reactor at lower
WC/FCO values. As the WC/FCO increases, the exit temperature of the adiabatic PBR reactor also
increases eventually leveling-off at the higher WC/FCO values. For this range of high WC/FCO values
both the isothermal and adiabatic PBR track the equilibrium conversion and because the exit
temperature for the adiabatic PBR is higher than that for the isothermal PBR the conversion of the
adiabatic PBR is lower than that of the isothermal PBR. For the lower WC/FCO values the
conversions in the isothermal PBR are substantially lower than the equilibrium conversions and
the exit adiabatic PBR conversions are higher, as shown in Fig. 43. For these conditions it is, of
course, more beneficial to operate the reactor under adiabatic conditions, as long as the exit
temperatures do not exceed the safe temperature of operation for the sour-shift catalyst, which is
510
o
C according to the catalyst manufacturer.
Fig. 43: CO conversion vs. (weight of catalyst /molar flow rate of CO) for adiabatic, isothermal operation
and corresponding adiabatic PBR exit temperature (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200
psig, inlet temperature is 300 ⁰C, and feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12).
620
625
630
635
640
645
650
0
10
20
30
40
50
60
70
0 50 100 150 200
Temperature [K]
CO conversion
Weight of catalyst / Molar flow rate of CO
Isothermal Adiabatic
Equilibrium conversion at 300 ⁰C Equilibrium conversion at 357 ⁰C
Adiabatic exit temperature
84
5.3 1-D Adiabatic Membrane Reactor
For the 1-D adiabatic MR model the same assumptions as the ones used in Section 5.2 are
utilized. In addition to these assumptions, it was also assumed that the membrane is catalytically
inactive. In our modeling we have used the permeances listed in Fig. 18 at 300 ⁰C for slow gases.
For H2, permeance changes with respect to temperature were taken into account by using equation
3.5 and the values calculated from Fig. 19.
Equations 4.1 and 4.2 that were used for the isothermal model still apply for adiabatic
operation. However, for the adiabatic reactor operations, energy balance equations are necessary
to describe the temperature variation in the feed and permeate sides of the membrane reactor,
which are expressed as:
Feed-side:
∑(𝑛 𝑗 𝐹 𝐶 𝑝 𝑗 )
𝑑 𝑇 𝐹 𝑑𝑉
=(−∆𝐻 𝑅𝑥
((1−𝜀 𝑣 )𝛽 𝑐 𝜌 𝑐 𝑅 𝐶𝑂
)− 𝛼 𝑚 𝑈 𝑚 ( 𝑇 𝐹 −𝑇 𝑃 )+( 𝑇 𝐹 −𝑇 𝑃 )∑(𝛾 𝛼 𝑚 𝐹 𝑗 𝐶 𝑝 𝑗 ) (5.4)
Permeate-side:
∑(𝑛 𝑗 𝑃 𝐶 𝑝 𝑗 )
𝑑 𝑇 𝑃 𝑑𝑉
= 𝛼 𝑚 𝑈 𝑚 ( 𝑇 𝐹 −𝑇 𝑃 )+( 𝑇 𝐹 −𝑇 𝑃 )∑((1−𝛾 )𝛼 𝑚 𝐹 𝑗 𝐶 𝑝 𝑗 ) (5.5)
T
F
= T
F
0, T
P
= T
P
0 at V=0 (γ = 1 for Fj < 0 and γ = 0 for Fj > 0)
where Um [J/s∙m
2
∙K] is the overall heat transfer coefficient through the membrane,
[K] is the
feed-side temperature, [K] is the permeate-side temperature, and Fj is the permeation flux of
component j (mol/m
2
·s). Equations 5.2 and 5.3 account for the fact the temperatures of the fluid
flowing on the feed side and that flowing on the permeate side may be different. This is an added
feature of the membrane reactor operation, in that the sweep stream in addition to enhancing the
permeation flux through the membrane may also play the dual role of a cooling stream.
The overall heat-transfer coefficient U for the tubular membrane is given as:
F
T
P
T
85
𝑈 𝑚 =
1
1
ℎ
𝑝 +(
𝑟 𝑖 𝑘 𝑚 )ln(
𝑟 𝑜 𝑟 𝑖 )+(
𝑟 𝑖 𝑟 𝑜 )/ℎ
𝐹 (5.6)
where km [J/s∙m∙K] is the thermal conductivity of the porous ceramic membrane, hF and hP
[J/s∙m
2
∙K] are the heat transfer coefficients of the feed-side and the permeate-side, respectively,
and ri [m] and ro [m] are the inner and outer diameters of the membrane, respectively.
The theoretical model proposed by Aivazov and Domashnev (Aivazov and Domashnev,
1968) well correlates the thermal conductivity of porous ceramics with the porosity according to
equation 5.5:
𝑘 𝑚 𝑘 𝑚 0
=
1− Ꜫ
𝑚 1+𝑚 Ꜫ
𝑚 2
(5.7)
where km,o [J/s∙m∙K] is the thermal conductivity of the nonporous ceramic material, εm is the
porosity of membrane support, and m is a constant that is taken to be equal to 3.0 as explained by
Sugawara and Yoshizawa (Sugawara and Yoshizawa, 1962).
COMSOL was used to model the adiabatic membrane reactor by using equations 4.1-4.2 ,
2.2-2.4 and 5.2-5.5 with the Ergun equation being used to calculate the pressure drop, with the
viscosity of gas taken to be constant with temperature. Fig. 44 compares the conversion (left axis)
in the adiabatic membrane reactor and compares it to the operation of the isothermal membrane
reactor under similar operating conditions (the inlet feed temperature in both cases is 300
o
C, and
no sweep is used, with other conditions indicated in the Figure caption). Shown on the same Figure
(right axis) is the temperature profile in the adiabatic reactor. The isothermal membrane reactor
shows a higher exit conversion than that of the adiabatic MR despite the fact that it never manages
to exceed the PBR equilibrium conversion while the adiabatic MR does exceed the corresponding
PBR equilibrium conversion. This is because the equilibrium WGS conversion is a strong function
of temperature, decreasing with increasing temperature, as Fig. 44 also indicates.
86
Fig. 44: Membrane reactor conversion and temperature profile vs. length of reactor for isothermal and
adiabatic MR model (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200 psig, and inlet temperature is
300 °C, and feed composition is H 2: CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12. Note that the
temperature of reactor and permeate side are the same along the length of the reactor).
Fig.45 compares the exit conversions (left axis) for the isothermal and adiabatic MR for a
range of WC/FCO values. Shown on the same Figure (right axis) is the exit temperature for both the
reject and permeate sides (equal for the 1-D MR model with no sweep). The isothermal membrane
reactor shows a higher exit conversion than that of the adiabatic MR beyond a certain W/FCO value,
attributed again to the strong dependence of equilibrium conversion on temperature. As Fig. 45
shows, MR beyond a certain WC/FCO value the conversion in the isothermal MR exceeds the PBR
equilibrium value as well.
560
570
580
590
600
610
620
630
640
650
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Temperature [K]
CO conversion
Dimensionless reactor length
Isothermal
Adiabatic
Equilibrium conversion at 300 ⁰C for PBR
Equilibrium conversion at 366 ⁰C for PBR
Temperature
87
Fig. 45: CO conversion vs. (weight of catalyst /molar flow rate of CO) for adiabatic, isothermal operation
and corresponding adiabatic MR exit temperature (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200
psig, inlet temperature is 300 ⁰C, and feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12).
It is clear from Fig. 45 that for a region of WC/FCO values the adiabatic temperatures exceed
the safe operating temperature range of the CMS membranes (~350
o
C). One effective way to
control that is to employ steam in its dual role as sweep stream as well as a coolant stream. In
Fig. 46, 47, and 48 we show the conversion (left axis) for the adiabatic membrane reactor (for
W/FCO=55) for three different inlet sweep stream temperatures (i.e., 200
o
C, 250
o
C, and 300
o
C)
along the membrane reactor length (the feed-side inlet temperature is 300 ⁰C and the other
conditions employed are shown in the Figure caption). Shown on the same Figure (right axis) are
the corresponding feed-side and permeate side temperatures. It should be noted that the exit
620
630
640
650
660
670
0
10
20
30
40
50
60
70
80
90
0 20 40 60 80 100 120 140 160 180
Temperature [K]
CO conversion
Weight of catalyst / Molar flow rate of CO
Isothermal Adiabatic
Equilibrium conversion at 300 ⁰C for PBR Equilibrium conversion at 355 ⁰C for PBR
Equilibrium conversion at 366 ⁰C for PBR Equilibrium conversion at 376 ⁰C for PBR
Equilibrium conversion at 385 ⁰C for PBR Adiabatic exit temperature at reactor side
88
temperature of the reactor side decreases with decreasing inlet sweep inlet temperature . The
maximum temperature of the MR in Fig. 47 is ~335 °C, which is within the safe operating
temperature range for the CMS membranes (~350
o
C). Employing a colder sweep also has a
beneficial effect on the MR exit conversion as it tracks the equilibrium conversion, which for the
exothermic WGS reaction is higher for lower temperatures as it has been previously noted.
Fig. 46: Membrane reactor conversion and temperature profile vs. length of the reactor for the adiabatic
MR model with 0.3 sweep ratio of steam at 300 °C (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200
psig, and feed inlet temperature is 300 °C, and feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12).
450
470
490
510
530
550
570
590
610
630
650
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Temperature [K]
CO conversion
Dimensionless reactor length
Adiabatic Reactor side temperature Permeate side temperature
89
Fig. 47: Membrane reactor conversion and temperature profile vs. length of reactor for adiabatic MR model
with 0.3 sweep ratio of steam at 250 °C (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200 psig, and
feed inlet temperature is 300 °C , and feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12).
Fig. 48: Membrane reactor conversion and temperature profile vs. length of the reactor for the adiabatic
MR model with 0.3 sweep ratio of steam at 200 °C (Weight of catalyst 10 g, W C/F CO is 55, pressure is 200
psig, and feed inlet temperature is 300 °C, and feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.12).
450
470
490
510
530
550
570
590
610
630
650
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Temperature [K]
CO conversion
Dimensionless reactor length
Adiabatic Reactor side temperature Permeate side temperature
450
470
490
510
530
550
570
590
610
630
650
0
10
20
30
40
50
60
70
0 0.2 0.4 0.6 0.8 1
Temperature [K]
CO conversion
Dimensionless reactor length
Adiabatic Reactor side temperature Permeate side temperature
90
5.4 2-D Isothermal Membrane Reactor
As it was mentioned previously, for the 1-D MR model, radial differences in concentration
at the reactor side were neglected. The presence of a membrane and the flow of the various species
through the reactor in the radial direction may render this assumption questionable. By
constructing a 2-D pseudo-homogeneous model, it is possible to account for these radial
differences in concentration. An isothermal 2-D pseudo-homogeneous MR model is, therefore,
utilized here that assumes no radial velocity component (Ur = 0) and neglects dispersion in the
axial direction for reactor side. The equations for the membrane reactor are as follows:
Reactor side:
𝐷 𝑒 ,𝑅 𝜕 2
𝐶 𝑖 ,𝑅 𝜕 𝑟 2
+
𝐷 𝑒 ,𝑅 𝑟 𝜕 𝐶 𝑖 ,𝑅 𝜕𝑟
−𝑈 𝑧 ,𝑅 𝜕 𝐶 𝑖 ,𝑅 𝜕𝑧
−𝐶 𝑖 ,𝑅 𝜕 𝑈 𝑧 ,𝑅 𝜕𝑍
=𝑣 𝑖 (1−𝜀 𝑣 )𝛽 𝑐 𝜌 𝑐 𝑅 𝐶𝑂
(5.8)
Boundary conditions for equation 5.8:
𝑧 =0: 𝐶 𝑖 ,𝑅 =𝐶 0
𝑖 ,𝑅 ,𝑈 𝑧 ,𝑅
=𝑈 𝑧 ,𝑅 0
(5.9)
𝑟 =𝑅 𝑟𝑒𝑎𝑐𝑡𝑜𝑟 :
𝜕 𝐶 𝑖 ,𝑅 𝜕𝑟
=0, (5.10)
𝑟 =𝑟 𝑚𝑒𝑚𝑏𝑟𝑎𝑛𝑒 ∶ −𝐷 𝑒 ,𝑅 𝜕 𝐶 𝑖 ,𝑅 𝜕𝑟
=𝐽 𝑖 (5.11)
Permeate side:
𝜕 (𝐶 𝑖 ,𝑃 𝑈 𝑧 ,𝑃
)
𝜕𝑍
=α
PE
𝐽 𝑖
𝑟 𝑜 𝑟 𝑖 (5.12)
Boundary conditions for equations 5.9:
𝑧 =0: 𝐶 𝑖 ,𝑃 =𝐶 0
𝑖 ,𝑃 ,𝑈 𝑧 ,𝑃
=𝑈 𝑧 ,𝑃 0
(5.13)
Where De,R is the radial dispersion coefficient in the reactive side (m
2
/s), Uz,R is the axial average
velocity on the reactive side (m/s), Uz,P is the axial velocity on the permeate side (m/s), ri is inside
radius of membrane support (m), ro is outside radius of membrane support (m), and Ji is the molar
91
flux through the membrane with respect to the outside radius of the membrane support (mol/ m
2
.
s) and calculated by: J
i
=U
j
(P
j
F
−P
j
P
) .
The effective radial dispersivity (D er) can be calculated by the following formula[81]:
𝐷 𝑒𝑟
=
𝑈 𝑠 𝑋 𝑓 8[2−(2−
2𝑑 𝑝𝑣
𝑑 𝑡 )
2
(5.14)
where Xf is the effective mixing length, XF = F dpv, where F for spherical particle is 1.15, US is
superficial velocity (m/s), dt is reactor tube diameter (m), dpv is the diameter of a sphere with
equivalent external surface area to that of the catalyst particle (m), and αPE is inside perimeter of
membrane stupor divided per inside cross-sectional area of membrane support (1/m) [68].
Fig. 49 shows the axial concentration of hydrogen along the length of the reactor by using
equations 5.8-5.14 and was generated using COMSOL. The figure shows there is a hydrogen
concentration gradient in the radial direction at the beginning of the reactor, but for most of the
reactor length, there is no noticeable radial profile in hydrogen concentration. We can conclude
that in our lab-scale membrane reactor the radial concentration profiles are negligible. Fig 50
shows the final conversion of MR vs. various weight of catalyst /molar flow rate of CO values for
the MR using 2-D model equations (5.7-8) and 1-D model equations (4.1-2). We also used equation
2.12 for the reaction rate and equations 2.2-2.4 (Ergun equation) to calculate the pressure drop
along the length of the reactor. The results of the 1-D and 2-D MR models are fairly close,
demonstrating again that the radial concentration profiles for our lab-scale MR are negligible. The
declining concentration profiles for the larger W/FCO values are due to the negative impact of
water loss.
92
Fig. 49: Hydrogen axial concentartion profile vs. length of membrane reactor and radius of reactor. (Weight
of catalyst 10 g, W c/F CO is 55, pressure is 200 psig, and inlet temperature is 300 ⁰C, and feed composition
is H 2: CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12).
lenght of reactor [m]
Radius of reactor [m]
The color indicating the value of
hydrogen concentration [ mol/m
3
]
93
Fig. 50: CO conversion vs. (weight of catalyst/molar flow rate of CO) for 1-D and 2-D MR models (Weight
of catalyst 10 g, W C/F CO is 55, pressure is 200 psig, inlet temperature is 300 ⁰C, and feed composition is H 2:
CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12).
5.5 Multistage Reactor
Multistage reactors, that combine PBR with membrane separators are potential alternatives
to using membrane reactors. Fig. 51, for example, shows a schematic of a multistage reactor that
combines two PBRs and two membrane separators in series. The potential advantage of such a
multistage reactor, is that the membrane and the catalyst bed are not in contact with each other,
and the operation of one is not limited by the presence of the other. It also facilitates easier
maintenance of the separators and the packed-bed reactors during operation. So, one can operate
the PBR at its optimum temperature without concerns about damaging the membrane. On the other
hand, operating the PBR and membrane separators separately has the disadvantage that the synergy
that is created by incorporating them both in the same unit is lost.
40
45
50
55
60
65
70
75
80
0 50 100 150 200
CO conversion
Weight of catalyst / Molar flow rate of CO
1-D model
2-D model
94
Fig. 51: Schematic of a multistage reactor with two packed-bed reactors and two membrane separators.
In the modeling of the multistage reactor/separator system of Figure 51, equations 2.1-2.4
were used for describing the packed-bed reactors, and equations 3.2-3.4 were used for describing
the separators. For the multistage reactor/separator system of Figure 51 consisting of two packed-
bed reactors and two membrane separators in series, conversion can be described by equation 5.11:
𝑋 𝐶𝑂
=
𝑛 𝐶𝑂 0
𝐹 ,𝑃𝐵 1
−(𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
+𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝐹 ,𝑀 2
)
𝑛 𝐶𝑂 0
𝐹 ∗100 (5. 15)
The hydrogen recovery for such a multistage reactor with two packed-bed reactors and two
membrane separators can be calculated by equation 5.12:
𝑅 𝑒 𝐻 2
=
𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
𝑛 𝐻 2,𝑒𝑥 𝑖 𝑡
𝑃 ,𝑀 1
+𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝐹 ,𝑀 2
+𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
∗100 (5.16)
The hydrogen purity (dry-basis) of this multistage reactor/separator system can be calculated by
equation 5.13:
𝑃 𝑢 𝐻 2
=
𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐶𝑂 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐶𝐻 4,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 1
+𝑛 𝐻 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
+𝑛 𝐶𝑂 ,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
+𝑛 𝐶𝑂 2,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
+𝑛 𝐶𝐻 4,𝑒𝑥𝑖𝑡
𝑃 ,𝑀 2
(5.17)
95
In the above equations, 𝑛 𝑖 ,𝑒𝑥𝑖𝑡
𝐹 is the molar flow rate (mol/ h) of component i at the exit of the PBR
or at the exit of the feed-side (reject-side) of the membrane separator, and 𝑛 𝑖 ,𝑒𝑥𝑖𝑡
𝑃 is the molar flux
(mol/m
2
. h) of component i at the exit of the permeate side of the membrane separator. The
superscript PB1 corresponds to the first PBR in series, PB2 corresponds to the second PBR, M1
correspond to the first membrane separator, and M2 corresponds to the second membrane
separator, and 𝑛 𝐶𝑂 0
𝐹
is the inlet molar flowrate of CO at reactor side (mol/h).
Fig. 52 compares the conversion of a multistage reactor (containing two packed-bed
reactors and two membrane separators, as described above) with a single membrane reactor. In
this comparison, the total catalyst weight, membrane surface areas, and initial feed conditions are
the same for both the multistage reactor and the membrane reactor. Fig. 52 shows that at lower
WC/FCO, the conversion of the membrane reactor is higher than that of the multistage reactor, and
that at higher WC/FCO the multistage reactor shows higher conversion. This is likely due to the loss
of water as a reactant through the membrane, which would affect the conversion at higher WC/FCO
for the membrane reactor. In the multistage reactor, on the other hand, water losses as a reactant
only occur in the first membrane separator. As a consequence, the negative impact of water loss
in the membrane reactor eventually overtakes the hydrogen separation enhancement effect in the
MR, and eventually the multistage reactor gives higher conversion at higher WC/FCO.
Fig. 53 shows the H2 recovery (dry-basis) for the multistage reactor compared to that for
the membrane reactor. According to Fig.53, with increasing values of WC/FCO, the H2 recovery for
the multistage reactor increases compared to that for the membrane reactor. This difference in
H2 recovery can be due to differences in the configuration of these two reactors. In membrane
reactors, the concentration of H2 increases in the permeate side along the length of the reactor, and
consequently, the transmembrane partial pressure of H2 decreases. On the other hand, in our
96
multistage reactor configuration, the concentration of H2 should be zero at the beginning of the
second membrane separator, and consequently, the transmembrane partial pressure of H2 is higher
than that for the same length of membrane reactor. Due to this difference in H2 transmembrane
partial pressure, the multistage reactor H2 recovery is higher than that for the membrane reactor
with the same initial conditions and surface area of membrane. Fig. 54 shows the H2 purity (dry-
basis) for both the multistage reactor and the membrane reactor. The H2 purity mainly depends on
membrane selectivity. The H2 purity is, therefore, almost the same for the multistage reactor and
the membrane reactor.
Fig. 52: Conversion vs. (weight of catalyst /molar flow rate of CO) for (i) Membrane reactor; (ii) Multistage
reactor, as indicated (Weight of catalyst is 50 g, pressure is 200 psig, and inlet temperature is 300 ⁰C, feed
composition is H 2: CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12, and for the MR the total length of
the membrane is 25.4 cm, while for the multistage reactor the length of each membrane is assumed to be
12.7 cm).
0
10
20
30
40
50
60
70
0 200 400 600 800
CO conversion
weight of catalyst / molar flow rate of CO
Membrane Reactor(SR=0)
Multistage Reactor (SR=0)
97
Fig. 53: H 2 recovery vs. (weight of catalyst /molar flow rate of CO) for (i) Membrane reactor; (ii) Multistage
reactor, as indicated (Weight of catalyst is 50 g, pressure is 200 psig, and inlet temperature is 300 ⁰C, feed
composition is H 2: CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12, and for the MR the total length of
the membrane is 25.4 cm, while for the multistage reactor the length of each membrane is assumed to be
12.7 cm).
Fig. 54: H 2 purity vs. (weight of catalyst/molar flow rate of CO) for (i) Membrane reactor; (ii) Multistage
reactor, as indicated (Weight of catalyst is 50 g, pressure is 200 psig, and inlet temperature is 300 ⁰C, feed
composition is H 2: CO: CO 2: CH 4: H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12, and for the MR the total length of
the membrane is 25.4 cm, while for the multistage reactor the length of each membrane is assumed to be
12.7 cm).
0
20
40
60
80
100
120
0 100 200 300 400 500 600 700
H
2
recovery
weight of catalyst / Molar flow rate of CO
Multistage reactor Membrane reactor
0
20
40
60
80
100
120
0 100 200 300 400 500 600 700
H
2
purity
weight of catalyst / Molar flow rate of CO
Multistage reactor Membrane reactor
98
Fig. 55 shows the molar flow rate profile of species along the multistage reactor for
different steps. Fig. 55 a) shows the molar flow rate for the first packed-bed in the multistage
reactor. In this conventional WGS-PBR, and there is no reaction enhancement due to hydrogen
separation or water loss. However, the rate of CO conversion in this PBR (even in the absence of
the membrane) is quite acceptable because the feed concentration of CO is high. Fig. 55 b) shows
the molar flow rate at the feed and permeate sides of the first separator of the multistage reactor.
In this step, most of the hydrogen separates from the feed, and some water is also lost. Fig. 55 c)
shows the second packed-bed in our multistage reactor system. In this step, the remaining CO
reacts and the rate of conversion is accelerated due to hydrogen separation from the feed mixture
in the previous step. Fig. 55d) shows the second separator, which separates the remaining hydrogen
produced by the second PBR from the feed mixture as a product.
99
Fig. 55: Species concentartion profiles along the length of the reactor/separtaor a) first PBR, b) first
membrane separator c) second PBR, and d) second separator (weight of catalyst = 25 g, weight of
catalyst/molar flow rate of CO = 200, weight of catalyst in each packed-bed reactor = 12.5 g, SR = 0,
pressure is 200 psig, and inlet temperature is 300 ⁰C, feed composition is H 2: CO: CO 2: CH 4: H 2S:
H 2O=2.6:1:2.1:0.8:0.05:1.1, and the total length of membrane is assumed to be 25.4 cm (for the multistage
reactor, the length of each membrane is assumed to be 12.7 cm, and half of total feed molar flow rate of
water is added to the second PBR feed).
100
5.6 Conclusion
We have compared the 1-D isothermal model to the 1-D adiabatic model for the PBR and MR.
The results show that for the adiabatic PBR the bed temperature rises quickly, and due to a
corresponding reduction in equilibrium conversion, the final conversion in that reactor may be less
than that for the isothermal PBR. We also have demonstrated that a reduction in conversion due to
rising bed temperatures can also be an issue for the adiabatic MR. Temperature rises in the MR
not only affect the MR final conversion, but they can also affect the membrane lifetime, if they
exceed the safe operating temperature range for the membrane. We have also shown that we can
use cool steam as a sweep gas to increase the CO conversion and to keep the temperature along
the length of the reactor within the CMS temperature operating range. We have also developed an
isothermal 2-D MR model. According to our model, for the lab-scale MR the concentration profile
is not significant in the radial direction of the reactor feed-side of the MR. We have also compared
the isothermal 1-D with the isothermal 2-D MR final conversion and have shown that these two
conversions are almost identical.
We have also introduced the multistage reactor concept, which is an alternative
configuration to the MR with some potential advantages over the MR. We have shown, for
example, that above a certain WC/FCO value, the multistage system, while employing the same
amount of catalyst and membrane active surface area, can result in higher final conversions than
the MR.
101
6. Summary and Future Study
6.1 Summary
Global climate change due to economic-related human activity on Earth is an undeniable
problem, and present renewable energy technologies are insufficient to cover the current growth
rate in global energy demand. Due to this reality, the development of more efficient and
environmentally-friendly processes involving fossil fuels, especially coal, is an attractive topic of
ongoing research.
Among the many processes currently available, the IGCC process is particularly promising
as an environmentally-friendly technique for large-scale coal utilization. The IGCC process has
not., as yet, been used on the industrial scale for energy production due to its energy intensity (i.e.,
energy penalty losses) as well as to process complexity. In order to overcome the energy
intensiveness and process complexity issues of the IGCC process, our group has proposed the use
of a WGS-MR technology. In order to improve the efficiency and cost-effectiveness of this WGS-
MR process, we have used a Co/Mo/Al2O3 sour-shift catalyst and a CMS membrane.
The sour-shift catalyst is more robust than the conventional WGS catalysts (iron-, zinc-,
copper-, and chrome-based catalysts) against the syngas contaminants, especially H2S. We have
studied the WGS reaction kinetics over the sour-shift catalyst in a PBR for temperatures up to 300
⁰C and pressures up to 200 psig with our simulated coal syngas composition (H2: CO: CO2: CH4:
H2S: H2O=2.6:1:2.1:0.8:0.05:1.1) and have fitted the data to an empirical reaction rate model and
to three microkinetic models taken from the literature. Our results show that the empirical model
and the so-called direct oxidation microkinetic model display the best fits to our experimental data.
We chose the CMS membrane for our MR process due to its relatively low fabrication cost
and its robustness under our reaction conditions. In our research we have investigated the pure and
102
binary/multi-mixed gas permeabilities of the CMS membrane for the common gases present in
coal-derived and biomass-derived syngas. The results show that the CMS membranes show
significant selectivity between H2 and other slower gases, thus making the CMS membrane an
excellent candidate for hydrogen separation in the WGS-MR process The transport of fast gases
like H2 and He obeys an Arrhenius-type equation, and we have calculated the activation energies
and pre-exponential factors by fitting the experimental data. For slow gases we have shown that
they transport via Knudsen diffusion, with their permeances showing a 1/T
0.5
dependence. During
the transport of mixtures, the permeances of the fast gases initially drop in the presence of slower
gases in the mixture, whereas the permeances of the slow gases do not decrease notably in the
presence of faster gases.
We have used the sour-shift catalyst and CMS membrane in a WGS-MR system and have
shown that the MR displays higher conversion than a conventional PBR setup for temperatures up
to 300⁰ and pressures up to 200 psig. In some of the MR experiments, steam has been used as a
sweep stream in the permeate side in order to increase the conversion and to improve the hydrogen
purity and hydrogen recovery. We have also developed an isothermal 1-D model and have used
the kinetic model generated (see Chapter 2) to describe the behavior of the MR with different
sweep ratios and temperatures up to 300 ⁰C and pressures up to 200 psig.
In Chapter 5, we developed an adiabatic 1-D model for both the PBR and MR, and we
showed how the temperature rise along the length of the PBR and MR can reduce the conversions
obtained from these reactors. Additionally, our adiabatic MR models have demonstrated how the
use of cool steam as a sweep can help keep the membrane temperature within the safe operational
temperature range.
103
We have also developed an isothermal 2-D model for the MR in order to investigate the
radial concentration profiles and possible mass transfer limitations of gases (especially hydrogen)
to the surface of membrane. Our 2-D model shows that the radial concentration profiles are
negligible for the lab-scale reactor. We also have shown that the final conversion of the MR
obtained from the 1-D model is virtually the same as that from the 2-D model.
Additionally, we have introduced the multistage reactor as an alternative option to the MR
and have explained some of the advantages of the multistage reactor over the MR. We have shown
that at certain flow rates (with the same initial operating conditions, the same weight of catalyst,
and the same membrane surface area), the multistage reactor can give higher conversions than the
MR.
6.2 Suggestions for Future Work
In order for the WGS-MR technology to be applied on a larger scale, further research should be
conducted. The WGS kinetic studies should be carried out under a broader range of conditions
such as higher pressures, different feed compositions, and additional coal syngas contaminants
such as ammonia. Additionally, the development of a more robust membrane that has more stable
properties under the experimental conditions is important. The performance of the MR needs to be
investigated at higher pressures, and the impact of tar on the WGS reactor and on the CMS
permeability needs to be investigated over longer time periods and at higher pressures.
104
References
[1] Crane, Hewitt, Edwin Kinderman, and Ripudaman Malhotra. A cubic mile of oil: realities and
options for averting the looming global energy crisis. Oxford University Press, 2010.
[2] Gupta, Ram B., ed. Hydrogen fuel: production, transport, and storage. Crc Press, 2008.
[3] U.S. Department of Energy, Office of Basic Energy Sciences, Committee on Alternatives and
Strategies for Future Hydrogen Production and Use, The Hydrogen Economy: Opportunities,
Costs, Barriers, and R&D Needs, National Research Council, National Academies Press,
Washington, available at http://www.nap.edu/catalog/10922.html (accessed May 2007), 2004.
[4] Scholz, Walter H. "Processes for industrial production of hydrogen and associated
environmental effects." Gas separation & purification 7, no. 3 (1993): 131-139.
[5] Board on Energy and Environmental Systems National Research Council, Division on
Engineering and Physical Sciences National Research Council, and Washington National
Academy of Engineering. The hydrogen economy: Opportunities, costs, barriers, and R&D needs.
National Academies Press, 2004
[6] Rand, David Anthony James, and Ronald Dell. Hydrogen energy: challenges and prospects.
Vol. 1. Royal Society of Chemistry, 2008.
[7] Petit, Jean-Robert, Jean Jouzel, Dominique Raynaud, Narcisse I. Barkov, J-M. Barnola,
Isabelle Basile, Michael Bender et al. "Climate and atmospheric history of the past 420,000 years
from the Vostok ice core, Antarctica." Nature 399, no. 6735 (1999): 429-436.
[8] Parmesan, Camille, and Gary Yohe. "A globally coherent fingerprint of climate change impacts
across natural systems." Nature 421, no. 6918 (2003): 37-42.
[9] Glynn, Peter W. "Coral reef bleaching in the 1980s and possible connections with global
warming." Trends in Ecology & Evolution 6, no. 6 (1991): 175-179.
105
[10] Rosenzweig, Cynthia, David Karoly, Marta Vicarelli, Peter Neofotis, Qigang Wu, Gino
Casassa, Annette Menzel et al. "Attributing physical and biological impacts to anthropogenic
climate change." Nature 453, no. 7193 (2008): 353-357.
[11] Ögel, Z. B., Y. Yüzügüllü, S. Mete, U. Bakir, Y. Kaptan, D. Sutay, and A. S. Demir.
"Production, properties and application to biocatalysis of a novel extracellular alkaline phenol
oxidase from the thermophilic fungus Scytalidium thermophilum." Applied microbiology and
biotechnology 71, no. 6 (2006): 853-862.
[12] Liu, Paul KT, Muhammad Sahimi, and Theodore T. Tsotsis. "Process intensification in
hydrogen production from coal and biomass via the use of membrane-based reactive
separations." Current Opinion in Chemical Engineering 1, no. 3 (2012): 342-351.
[13] Descamps, C., Chakib Bouallou, and Mohamed Kanniche. "Efficiency of an integrated
gasification combined cycle (IGCC) power plant including CO 2 removal." Energy 33, no. 6
(2008): 874-881.
[14] Christou, Costas, Ioannis Hadjipaschalis, and Andreas Poullikkas. "Assessment of integrated
gasification combined cycle technology competitiveness." Renewable and Sustainable Energy
Reviews 12, no. 9 (2008): 2459-2471.
[15] Stiegel, Gary J., and Massood Ramezan. "Hydrogen from coal gasification: an economical
pathway to a sustainable energy future." International Journal of Coal Geology 65, no. 3 (2006):
173-190.
[16] Yu, Jiang, Mingyang Tan, Paul KT Liu, Muhammad Sahimi, and Theodore T. Tsotsis.
"Hydrogen production from biomass-derived syngas using a membrane reactor based
process." Industrial & Engineering Chemistry Research 53, no. 2 (2013): 819-827.
106
[17] Chiesa, Paolo, and Stefano Consonni. "Shift reactors and physical absorption for low-CO2
emission IGCCs." Journal of Engineering for Gas Turbines and Power 121, no. 2 (1999): 295-
305.
[18] Jones, Cheryl W., and William J. Koros. "Characterization of ultramicroporous carbon
membranes with humidified feeds." Industrial & engineering chemistry research 34, no. 1 (1995):
158-163.
[19] Hla, San Shwe, G. J. Duffy, L. D. Morpeth, A. Cousins, D. G. Roberts, J. H. Edwards, and D.
Park. "Catalysts for water–gas shift processing of coal‐derived syngases." Asia‐Pacific Journal of
Chemical Engineering 5, no. 4 (2010): 585-592.
[20] Rezvani, Sina, Ye Huang, David McIlveen-Wright, Neil Hewitt, and Jayanta Deb Mondol.
"Comparative assessment of coal fired IGCC systems with CO 2 capture using physical absorption,
membrane reactors and chemical looping." Fuel 88, no. 12 (2009): 2463-2472.
[21] Lyngfelt, A., and C. Azar. "Proceedings of minisymposium on carbon dioxide capture and
storage." (1999).
[22] Twing,M. V (ED.),’’ Catalyst Handbook””(2
nd
ed.), Wolfe publishing, London, (1989).
[23] Babita, K., S. Sridhar, and K. V. Raghavan. "Membrane reactors for fuel cell quality hydrogen
through WGSR–review of their status, challenges and opportunities." international journal of
hydrogen energy 36, no. 11 (2011): 6671-6688.
[24] Farrauto, R. J., and C. H. Bartholomew. "Chapter 9 of “Fundamentals of Industrial Catalytic
Processes”." pub.: Blackie Academia & Professional (1997): 523-535.
[25] Song, Chunshan. "Fuel processing for low-temperature and high-temperature fuel cells:
Challenges, and opportunities for sustainable development in the 21st century." Catalysis
today 77, no. 1 (2002): 17-49.
107
[26] Gorte, R. J., and S. Zhao. "Studies of the water-gas-shift reaction with ceria-supported
precious metals." Catalysis Today 104, no. 1 (2005): 18-24.
[27] Grenoble, D. C., M. M. Estadt, and D. F. Ollis. "The chemistry and catalysis of the water
gas shift reaction: 1. The kinetics over supported metal catalysts." Journal of Catalysis 67, no. 1
(1981): 90-102.
[28] Hilaire, S., X. Wang, T. Luo, R. J. Gorte, and J. Wagner. "A comparative study of water-
gas-shift reaction over ceria-supported metallic catalysts." Applied Catalysis A: General 258, no.
2 (2004): 271-276.
[29] Jacobs, Gary, Emilie Chenu, Patricia M. Patterson, Leann Williams, Dennis Sparks, Gerald
Thomas, and Burtron H. Davis. "Water-gas shift: comparative screening of metal promoters for
metal/ceria systems and role of the metal." Applied Catalysis A: General 258, no. 2 (2004): 203-
214.
[30] Jacobs, Gary, Patricia M. Patterson, Uschi M. Graham, Dennis E. Sparks, and Burtron H.
Davis. "Low temperature water-gas shift: kinetic isotope effect observed for decomposition of
surface formates for Pt/ceria catalysts." Applied Catalysis A: General 269, no. 1 (2004): 63-73.
[31] Lei, Yun, Noel W. Cant, and David L. Trimm. "Activity Patterns for the ‘‘Water Gas Shift
Reaction Over Supported Precious Metal Catalysts’’." Catalysis letters 103, no. 1 (2005): 133-
136.
[32] Panagiotopoulou, Paraskevi, and Dimitris I. Kondarides. "Effect of morphological
characteristics of TiO 2-supported noble metal catalysts on their activity for the water–gas shift
reaction." Journal of catalysis 225, no. 2 (2004): 327-336.
108
[33] Panagiotopoulou, Paraskevi, and Dimitris I. Kondarides. "Effect of the nature of the support
on the catalytic performance of noble metal catalysts for the water–gas shift reaction." Catalysis
Today 112, no. 1 (2006): 49-52.
[34] Wang, X., and Raymond J. Gorte. "The effect of Fe and other promoters on the activity of
Pd/ceria for the water-gas shift reaction." Applied Catalysis A: General 247, no. 1 (2003): 157-
162.
[35] Wang, X., Raymond J. Gorte, and J. P. Wagner. "Deactivation mechanisms for Pd/ceria during
the water–gas-shift reaction." Journal of catalysis 212, no. 2 (2002): 225-230.
[36] Wheeler, C., A. Jhalani, E. J. Klein, S. Tummala, and L. D. Schmidt. "The water–gas-shift
reaction at short contact times." Journal of catalysis 223, no. 1 (2004): 191-199.
[37] Haryanto, Agus, Sandun D. Fernando, SD Filip To, Philip H. Steele, Lester Pordesimo, and
Sushil Adhikari. "Hydrogen production through the water− gas shift reaction: thermodynamic
equilibrium versus experimental results over supported Ni catalysts." Energy & Fuels 23, no. 6
(2009): 3097-3102.
[38] Velu, S., K. Suzuki, M. P. Kapoor, F. Ohashi, and T. Osaki. "Selective production of hydrogen
for fuel cells via oxidative steam reforming of methanol over CuZnAl (Zr)-oxide
catalysts." Applied Catalysis A: General 213, no. 1 (2001): 47-63.
[39] Tibiletti, Daniele, EA Bart de Graaf, Siew Pheng Teh, Gadi Rothenberg, David Farrusseng,
and Claude Mirodatos. "Selective CO oxidation in the presence of hydrogen: fast parallel screening
and mechanistic studies on ceria-based catalysts." Journal of Catalysis 225, no. 2 (2004): 489-497.
[40] de la Osa, A. R., A. De Lucas, A. Romero, P. Casero, J. L. Valverde, and P. Sánchez. "High
pressure Water Gas Shift performance over a commercial non-sulfide CoMo catalyst using
industrial coal-derived syngas." Fuel 97 (2012): 428-434.
109
[41] Smith, R. J., Muruganandam Loganathan, and Murthy Shekhar Shantha. "A review of the
water gas shift reaction kinetics." International Journal of Chemical Reactor Engineering 8, no. 1
(2010).
[42] Mendes, D., A. Mendes, L. M. Madeira, A. Iulianelli, J. M. Sousa, and 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, no. 1 (2010): 111-137.
[43] Cormos, Calin-Cristian. "Integrated assessment of IGCC power generation technology with
carbon capture and storage (CCS)." Energy 42, no. 1 (2012): 434-445.
[44] LeValley, Trevor L., Anthony R. Richard, and Maohong Fan. "The progress in water gas shift
and steam reforming hydrogen production technologies–a review." international journal of
hydrogen energy 39, no. 30 (2014): 16983-17000.
[45] Kohl, Arthur L., and Richard Nielsen. Gas purification. Gulf Professional Publishing, 1997.
[46] Kersten, Sascha RA, Wim PM van Swaaij, Leon Lefferts, and Kulathuiyer Seshan. "Options
for catalysis in the thermochemical conversion of biomass into fuels." Catalysis for Renewables:
From feedstock to energy production (2007): 119-145.
[47] Srivatsa, Narendra R., and Sol W. Weller. "Water-gas shift kinetics over sulfided catalyst:
elevated pressure." In 9th International Congress on Catalysis. 1988.
[48] Abdollahi, Mitra, Jiang Yu, Hyun Tae Hwang, Paul KT Liu, Richard Ciora, Muhammad
Sahimi, and Theodore T. Tsotsis. "Process intensification in hydrogen production from biomass-
derived syngas." Industrial & Engineering Chemistry Research 49, no. 21 (2010): 10986-10993.
[49] Chinchen, G. C., P. J. Denny, J. R. Jennings, M. S. Spencer, and K. C. Waugh. "Synthesis of
methanol: part 1. Catalysts and kinetics." Applied Catalysis 36 (1988): 1-65.
110
[50] Moe, J. M. "Design of water-gas shift reactors." Chem. Eng. Prog.;(United States) 58, no. 3
(1962).
[51] Choi, Yongtaek, and Harvey G. Stenger. "Water gas shift reaction kinetics and reactor
modeling for fuel cell grade hydrogen." Journal of Power Sources 124, no. 2 (2003): 432-439.
[52] Liu, Ke, Chunshan Song, and Velu Subramani, eds. Hydrogen and syngas production and
purification technologies. Hoboken, NJ: wiley, 2010.
[53] De la Osa, A. R., A. De Lucas, A. Romero, J. L. Valverde, and P. Sánchez. "Kinetic models
discrimination for the high pressure WGS reaction over a commercial CoMo
catalyst." international journal of hydrogen energy 36, no. 16 (2011): 9673-9684.
[54] H.P. Hsieh, “Inorganic membranes”, Membr. Mater. Proc. 84(1990) 1.
[55] Baker, Richard W. Membrane technology. John Wiley & Sons, Inc., 2000.
[56] Koresh, Jacob, and Abraham Soffer. "Study of molecular sieve carbons. Part 1.—Pore
structure, gradual pore opening and mechanism of molecular sieving." Journal of the Chemical
Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases 76 (1980): 2457-2471.
[57] Salleh, Wan Norhayati Wan, Ahmad Fauzi Ismail, T. Matsuura, and Mohd Sohaimi Abdullah.
"Precursor selection and process conditions in the preparation of carbon membrane for gas
separation: A review." Separation & Purification Reviews 40, no. 4 (2011): 261-311.
[58] Fuertes, Antonio B. "Adsorption-selective carbon membrane for gas separation." Journal of
Membrane Science 177, no. 1 (2000): 9-16.
[59] Gallucci, Fausto, Ekain Fernandez, Pablo Corengia, and Martin van Sint Annaland. "Recent
advances on membranes and membrane reactors for hydrogen production." Chemical Engineering
Science 92 (2013): 40-66.
111
[60] Basile, Angelo, Stefano Curcio, Giuseppe Bagnato, Simona Liguori, S. M. Jokar, and Adolfo
Iulianelli. "Water gas shift reaction in membrane reactors: Theoretical investigation by artificial
neural networks model and experimental validation." International Journal of Hydrogen
Energy 40, no. 17 (2015): 5897-5906.
[61] Koros, W. J., Y. H. Ma, and T. Shimidzu. "Terminology for membranes and membrane
processes (IUPAC Recommendations 1996)." Pure and Applied Chemistry 68, no. 7 (1996): 1479-
1489.
[62] Dittmeyer, Roland, and Jürgen Caro. "Catalytic membrane reactors." Handbook of
Heterogeneous Catalysis (2008).
[63] Marano, John J., and Jared P. Ciferino. "Integration of gas separation membranes with IGCC
identifying the right membrane for the right job." Energy Procedia 1, no. 1 (2009): 361-368.
[64] Ockwig, Nathan W., and Tina M. Nenoff. "Membranes for hydrogen separation." Chemical
Reviews 107, no. 10 (2007): 4078-4110.
[65] Lima, Fernando V., Prodromos Daoutidis, Michael Tsapatsis, and John J. Marano. "Modeling
and optimization of membrane reactors for carbon capture in integrated gasification combined
cycle units." Industrial & Engineering Chemistry Research 51, no. 15 (2012): 5480-5489.
[66] Merkel, Tim C., Meijuan Zhou, and Richard W. Baker. "Carbon dioxide capture with
membranes at an IGCC power plant." Journal of membrane science 389 (2012): 441-450.
[67] Favre, Eric. "Carbon dioxide recovery from post-combustion processes: Can gas permeation
membranes compete with absorption?." Journal of Membrane Science 294, no. 1 (2007): 50-59
[68] Bauer, R., and E. U. Schlunder. "Effective Radial Thermal-Conductivity of Packings in Gas-
Flow. 2. Thermal-Conductivity of Packing Fraction without Gas-Flow." International Chemical
Engineering 18, no. 2 (1978): 189-204.
112
[69] Wakao, Noriaki, and J. M. Smith. "Diffusion in catalyst pellets." Chemical Engineering
Science 17, no. 11 (1962): 825-834.
[70] Li, Hang, Zhuonan Song, Xiaojie Zhang, Yi Huang, Shiguang Li, Yating Mao, Harry J.
Ploehn, Yu Bao, and Miao Yu. "Ultrathin, molecular-sieving graphene oxide membranes for
selective hydrogen separation." Science 342, no. 6154 (2013): 95-98.
[71] Itoh, N., and K. Haraya. "A carbon membrane reactor." Catalysis Today 56, no. 1 (2000):
103-111.
[72] Lukyanov, B. N., D. V. Andreev, and V. N. Parmon. "Catalytic reactors with hydrogen
membrane separation." Chemical Engineering Journal 154, no. 1 (2009): 258-266.
[73] Morpeth, Leigh D., and Michael D. Dolan. "Modelling and experimental studies of a water–
gas shift catalytic membrane reactor." Chemical Engineering Journal 276 (2015): 289-302.
[74] Briceño, Kelly, Adolfo Iulianelli, Daniel Montané, Ricard Garcia-Valls, and Angelo Basile.
"Carbon molecular sieve membranes supported on non-modified ceramic tubes for hydrogen
separation in membrane reactors." international journal of hydrogen energy 37, no. 18 (2012):
13536-13544.
[75] De Lorenzo, Luca, Thomas G. Kreutz, Paolo Chiesa, and Robert H. Williams. "Carbon-free
hydrogen and electricity from coal: options for syngas cooling in systems using a hydrogen
separation membrane reactor." Journal of Engineering for Gas Turbines and Power 130, no. 3
(2008): 031401.
[76] Klara, J., M. Woods, O. J. Capicotto, J. T. Haslbeck, M. Matuszewski, L. L. Pinkerton, M. D.
Rutkowski, R. L. Schoff, and V. Vaysman. "Cost and performance baseline for fossil energy
plants." National Energy Technology Laboratory, DOE/NETL-2007/1281 (2007).
113
[77] Bridgwater, Anthony V. "Renewable fuels and chemicals by thermal processing of
biomass." Chemical Engineering Journal 91, no. 2 (2003): 87-102.
[78] Schiefelbein, G. F. "Biomass thermal gasification research: Recent results from the United
States DOE's Research Program." Biomass 19, no. 1-2 (1989): 145-159.
[79] Elliott, J. Richard, and Carl T. Lira. Introductory chemical engineering thermodynamics.
Upper Saddle River, NJ: Prentice Hall PTR, 1999.
[80] Parsley, Doug, Richard J. Ciora, Diane L. Flowers, John Laukaitaus, Amy Chen, Paul KT
Liu, Jiang Yu, Muhammad Sahimi, Alex Bonsu, and Theodore T. Tsotsis. "Field evaluation of
carbon molecular sieve membranes for the separation and purification of hydrogen from coal-and
biomass-derived syngas." Journal of Membrane Science 450 (2014): 81-92.
[81] Bauer, R., and E. U. Schlunder. "EFFECTIVE RADIAL THERMAL-CONDUCTIVITY OF
PACKINGS IN GAS-FLOW. 1. CONVECTIVE TRANSPORT COEFFICIENT." International
Chemical Engineering 18, no. 2 (1978): 181-188.
114
Appendix
WGS Reaction Data
Figure 56 below summarizes all the experimental data generated with the Co-Mo WGS catalyst.
Fig. 56: Packed-bed CO conversion vs. (weight of catalyst/molar flow rate of CO) for temperatures up to
300 ⁰C and pressures up to 200 psig. (Weight of catalyst 10 g, and feed composition is H 2: CO: CO 2: CH 4:
H 2S: H 2O=2.6:1:2.1:0.8:0.05:1.12).
0
10
20
30
40
50
60
70
40 60 80 100 120 140 160 180
CO conversion
Weight of catalyst / Molar flow rate of CO
T=220 C, P= 30 psig
T=220 C, P=50 psig
T=220 C, P=150 psig
T=220 C, P=200 psig
T=250 C, P=30 psig
T=250 C, P=50 psig
T=250 C, P=150 psig
T=280 C, P=30 psig
T=280 C, P=50 psig
T=280 C, P=150 psig
T=280 C, P=200 psig
T=300 C, P=30 psig
T=300 C, P=50 psig
T=300 C, P=150 psig
T=300 C, P=200 psig
115
CMS Evaluation
CMS#5:
Part ID:
CMS-
M&P-12
T (⁰C) P (psig) He [m
3
/m
2
*hr*bar ] N 2 [m
3
/m
2
*hr*bar] selectivity
M&P 15 20 0.3127 0.0065 47.9 (He/N 2)
M&P 250 20 1.6043 0.0137 117.4 (He/N 2)
M&P-STP 15 20 0.30 0.0062 47.9 (He/N 2)
M&P-STP 250 20 1.52 0.013 117.4 (He/N 2)
USC(1/4) 20 20 0.21 0.0057 27 (He/N 2)
USC(1/4) 20 200 0.20 0.0074 37 (He/N 2)
USC(1/5) 250 20 1.47 0.014 105 (He/N 2)
USC(1/5) 250 200 1.44 0.014 105(He/N 2)
Table 17: Pur- gas permeances of N 2 and He at various temperatures and pressures of up to 200 psig and
comparison with M&P membrane data. (M&P data were converted to STP by assuming the temperature
was 15 ⁰C and the pressure was 1 atm at the M&P lab). All gases used, including the overnight gas
atmosphere, were ultra-high pure grade. N 2 and He cylinders were equipped with moisture and oxygen
adsorbents.
116
Part ID:
CMS-
M&P-12
T (⁰C) P
(psig)
He
[m
3
/m
2
*hr*bar ]
N 2
[m
3
/m
2
*hr*bar]
selectivity
USC(1/10) 250 200 0.013 The module had been filled with
catalyst without using Teflon
tube.
The membrane was exposed to a
mixture of nitrogen, hydrogen,
carbon dioxide, and hydrogen
sulfide.
USC(1/13) 250 200 0.0125 The membrane was exposed to
hydrogen sulfide up to 5%.
USC(1/16) 250 200 0.011 The membrane temperature was
raised to 300 C the day before
and was exposed to hydrogen
sulfide up to 5%.
USC(1/18) 250 200 1.35 0.011 118 The measurement took place at
the end of the activation process.
USC(1/19) 250 200 0.012
USC(1/23) 250 200 0.012 Membrane reactor experiment
took pace before this
measurement
USC(1/25) 250 200 0.013 Membrane reactor experiment
took pace before this
measurement
USC(1/30) 250 200 0.014 Membrane was idle under N 2
atmosphere at 250 ⁰C
USC(2/1) 250 200 1.35 0.014
Table 18: Permeances of He and N 2 during the activation processes. The module temperature was kept at
250 ⁰C and was under an N 2 flow atmosphere when the module was not in use for other experiments. Several
different gas permeances were measured before and after the membrane reactor experiments at 250 ⁰C and
at 200 psig in order to evaluate the membranes during the membrane reactor experiments.
Abstract (if available)
Abstract
Global climate change due to economic-related human activity on Earth is an undeniable problem, and present renewable energy technologies are insufficient to cover the current growth rate in global energy demand. Due to this reality, the development of more efficient and environmentally-friendly processes involving fossil fuels, especially coal, is an attractive topic of ongoing research. ❧ Among the many processes currently available, the IGCC process is particularly promising as an environmentally-friendly technique for large-scale coal utilization. The IGCC process has not., as yet, been used on the industrial scale for energy production due to its energy intensity (i.e., energy penalty losses) as well as to process complexity. In order to overcome the energy intensiveness and process complexity issues of the IGCC process, our group has proposed the use of a WGS-MR technology. In order to improve the efficiency and cost-effectiveness of this WGS-MR process, we have used a Co/Mo/Al₂O₃ sour-shift catalyst and a CMS membrane. ❧ The sour-shift catalyst is more robust than the conventional WGS catalysts (iron-, zinc-, copper-, and chrome-based catalysts) against the syngas contaminants, especially H₂S. We have studied the WGS reaction kinetics over the sour-shift catalyst in a PBR for temperatures up to 300 ℃ and pressures up to 200 psig with our simulated coal syngas composition (H₂: CO: CO₂: CH₄: H₂S: H₂O=2.6:1:2.1:0.8:0.05:1.1) and have fitted the data to an empirical reaction rate model and to three microkinetic models taken from the literature. Our results show that the empirical model and the so-called direct oxidation microkinetic model display the best fits to our experimental data. ❧ We chose the CMS membrane for our MR process due to its relatively low fabrication cost and its robustness under our reaction conditions. In our research we have investigated the pure and binary/multi-mixed gas permeabilities of the CMS membrane for the common gases present in coal-derived and biomass-derived syngas. The results show that the CMS membranes show significant selectivity between H₂ and other slower gases, thus making the CMS membrane an excellent candidate for hydrogen separation in the WGS-MR process The transport of fast gases like H₂ and He obeys an Arrhenius-type equation, and we have calculated the activation energies and pre-exponential factors by fitting the experimental data. For slow gases we have shown that they transport via Knudsen diffusion, with their permeances showing a 1/T0.5 dependence. During the transport of mixtures, the permeances of the fast gases initially drop in the presence of slower gases in the mixture, whereas the permeances of the slow gases do not decrease notably in the presence of faster gases. ❧ We have used the sour-shift catalyst and CMS membrane in a WGS-MR system and have shown that the MR displays higher conversion than a conventional PBR setup for temperatures up to 300° and pressures up to 200 psig. In some of the MR experiments, steam has been used as a sweep stream in the permeate side in order to increase the conversion and to improve the hydrogen purity and hydrogen recovery. We have also developed an isothermal 1-D model and have used the kinetic model generated (see Chapter 2) to describe the behavior of the MR with different sweep ratios and temperatures up to 300 ℃ and pressures up to 200 psig. ❧ In Chapter 5, we developed an adiabatic 1-D model for both the PBR and MR, and we showed how the temperature rise along the length of the PBR and MR can reduce the conversions obtained from these reactors. Additionally, our adiabatic MR models have demonstrated how the use of cool steam as a sweep can help keep the membrane temperature within the safe operational temperature range. ❧ We have also developed an isothermal 2-D model for the MR in order to investigate the radial concentration profiles and possible mass transfer limitations of gases (especially hydrogen) to the surface of membrane. Our 2-D model shows that the radial concentration profiles are negligible for the lab-scale reactor. We also have shown that the final conversion of the MR obtained from the 1-D model is virtually the same as that from the 2-D model. ❧ Additionally, we have introduced the multistage reactor as an alternative option to the MR and have explained some of the advantages of the multistage reactor over the MR. We have shown that at certain flow rates (with the same initial operating conditions, the same weight of catalyst, and the same membrane surface area), the multistage reactor can give higher conversions than the MR.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The use of carbon molecule sieve and Pd membranes for conventional and reactive applications
PDF
On the use of membrane reactors in biomass utilization
PDF
A high efficiency, ultra-compact process for pre-combustion CO₂ capture
PDF
An integrated 'one-box' process for hydrogen production
PDF
Methanol synthesis in the membrane reactor
PDF
A flow-through membrane reactor for destruction of a chemical warfare simulant
PDF
Lab-scale and field-scale study of siloxane contaminants removal from landfill gas
PDF
A study of the application of membrane-based reactive separation to the carbon dioxide methanation
PDF
Methanol synthesis in a membrane reactor
PDF
A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production
PDF
The purification of contaminated air streams via recative separation techniques
PDF
Development of carbon molecular-sieve membranes with tunable properties: modification of the pore size and surface affinity
PDF
Hydrogen storage in carbon and silicon carbide nanotubes
PDF
Mesoscale SOFC-based power generator system: modeling and experiments
PDF
A process-based molecular model of nano-porous silicon carbide membranes
PDF
Biogas reforming: conventional and reactive separation processes and the preparation and characterization of related materials
PDF
Novel methods for landfill gas and biogas clean-up
PDF
Methanol synthesis in a contactor-type membrane reactor
PDF
Preparation of polyetherimide nanoparticles by electrospray drying, and their use in the preparation of mixed-matrix carbon molecular-sieve (CMS) membranes
PDF
Studies of transport phenomena in hydrotalcite membranes, and their use in direct methanol fuel cells
Asset Metadata
Creator
Garshasbi, Ashkan
(author)
Core Title
Process intensification in hydrogen production via membrane-based reactive separations
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Chemical Engineering
Degree Conferral Date
2017-08
Publication Date
06/01/2019
Defense Date
01/27/2017
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
carbon molecular sieve membranes,membrane reactor,OAI-PMH Harvest,water gas shift reaction
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Tsotsis, Theodore T. (
committee chair
), Egolfopoulos, Fokion (
committee member
), Shing, Katherine (
committee member
)
Creator Email
ag101.hhh@gmail.com,garshasb@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC11258400
Unique identifier
UC11258400
Identifier
etd-GarshasbiA-5381-0.pdf (filename)
Legacy Identifier
etd-GarshasbiA-5381-0
Dmrecord
380512
Document Type
Dissertation
Format
theses (aat)
Rights
Garshasbi, Ashkan
Internet Media Type
application/pdf
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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
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
Repository Email
cisadmin@lib.usc.edu
Tags
carbon molecular sieve membranes
membrane reactor
water gas shift reaction