Page | 1
Chemical Engineering Study Project 4
Abstract: The separation of CO2 from the flue gas of a 600 MWe Coal fired power station
was investigated with the aim of achieving 90% recovery of CO2 and a permeate purity of
90% CO2. The separation was to be performed using a dual stage membrane process. A range
of variables were investigated including membrane area, membrane configurations, sweep
flowrates and pressure ratios. The dual stage process was able to achieve up to 89.57%
recovery and 85.87% purity at a substantial energy cost. The dual stage process was deemed
to be able to achieve the targets but not within acceptable limits.
Paul Connaghan s0820010
Submission December 2011
University of Edinburgh, School of Engineering (and Electronics)
Chemical Engineering
Study Project 4 (U04634)
Membrane Separations for CO2 Capture: Dual
Stage Process simulations
Paul Connaghan
S0820010
Page | 2
Table of Contents
1 Introduction ........................................................................................................................ 3
2 Literature Review ............................................................................................................... 4
2.1 Current Technologies .................................................................................................. 4
2.2 Theory of membrane separation units ......................................................................... 6
2.2.1 Membrane types ................................................................................................... 7
2.2.2 Models used ......................................................................................................... 8
2.2.3 Membrane parameters ........................................................................................ 11
2.3 Conclusion ................................................................................................................. 13
3 Modelling Process ............................................................................................................ 14
3.1 Aims .......................................................................................................................... 14
3.2 Simulator Modelling Strategy ................................................................................... 14
3.2.1 Manipulated Variables ....................................................................................... 14
3.2.2 Monitored Variables .......................................................................................... 16
3.2.3 Key Performance Parameters ............................................................................. 17
3.2.4 Simulator Interface............................................................................................. 18
3.3 Results ....................................................................................................................... 19
3.4 Critical Review .......................................................................................................... 27
3.5 Conclusion ................................................................................................................. 32
4 Appendices ....................................................................................................................... 33
Page | 3
1 Introduction
A substantial amount of CO2 is produced through anthropogenic activities; In particular the
amount produced through power generation. Power generation is estimated to account for up
to 80% of worldwide greenhouse gas production[1]. Furthermore up to 25% of the energy
supply and 40% of the emissions produced are generated by coal fired power stations and the
rest largely through gas and oil [2]. With power demands increasing, especially within
developing countries which have a plentiful supply of coal: the emissions from coal fired
energy are set to increase by around 350% from 2000 to 2050 [3]. This represents an increase
from 40% to 55% of total global power emissions [4].
The effects of CO2 on the mechanisms of global warming has not been understood fully as of
yet but many climate models have shown that the continued increasing trends of CO2
atmospheric concentration will have a dramatic effect on the earth’s climate by 2100[5].
More and more governments and industries are looking for ways to reduce and limit their
CO2 emissions due to increasing numbers of directives and protocols. As a result of these
issues there is mounting pressure to find a way to sequester CO2 at an affordable rate. This
report looks into sequestration of coal fired emissions and how to achieve a satisfactory
degree of separation through membrane technologies.
There are a number of options available for capturing the CO2 produced by coal fired power
plants. The three main options as outlined by DOE [6] and the IEA [7] are:
1- Post-combustion – CO2 capture from the flue gas produced
2- Pre-combustion – CO2 capture from gasified coal synthesis gas
3- Oxy-combustion – Fuel is burnt with almost pure oxygen to produce a high purity
CO2 effluent
Both pre-combustion and oxy-combustion can capture 90% of CO2 produced with a high
purity [8] but are unable to sequester the CO2 produced by a direct fired coal power plant
without further modifications [9]: pre-combustion requires gasified coal syngas to be
prepared; oxy-combustion requires special equipment for N2 separation[10]; both require a
turbine system for combustion. The vast majority of coal fired power plants are directly fired
air combustion coal burning and this type will be continued to be constructed for the
foreseeable future. With this in mind researching possible sequestration techniques for these
Page | 4
standard air fired combustion systems is an important area if we are to try to reduce our CO2
emissions.
A number of techniques are available for post combustion CO2 sequestration which must
deal with two main considerations:
1- Large flowrates of flue gas [4, 11]
2- The low partial pressures of CO2 in the flue gas
The main sequestration method that is promising is the use of amine absorption. This is a
proven technology which has been used for many years for industrial applications; however,
when used for power plant sequestrations it will be costly and energy intensive. Use of this
technology could result in a rise in cost of energy anywhere between 50-90% which is
undesirable[4].
Membranes are a developing area with respect to CO2 separations. They are a promising
candidate for this process due to a number of advantages: the ability to deal with large flows;
lack of moving parts means lower maintenance costs; a very small footprint. There is much
doubt as to whether membranes can be used to achieve the separation realistically. Although
they have many advantages, the use of a single membrane unit is insufficient for achieving
the necessary separation and we must look to dual stage processes for the feasibility of
achieving DOE targets [12]. There are a number of key factors that can affect a dual stage
process and all of these must be investigated if the process is to be competitive.
2 Literature Review
2.1 Current Technologies
There is a lot of effort being put into developing gas separation techniques as these are used
in so many different industries and processes. Furthermore to this, a lot of effort is being put
towards developing techniques for the separation of CO2 due to increasing pressure on
industries and the apparent criticality of the CO2 emissions.
As mentioned before the increasing pressure on reducing CO2 emissions to the atmosphere
has bolstered the research performed in the area of CO2 separation. The three main options
for reducing CO2 emissions as produced through power generation are as follows[8]:
Page | 5
- Reduction of energy intensity
- Reduction of carbon intensity
- Enhancing CO2 sequestration
As power generation from fossil fuels is a mature technology there is little in the way of
substantial improvements that can be made in the efficiency of existing power stations and
therefore we must look to points two and three for improvements[8]. Reduction in carbon
intensity is not feasible for fossil fuel schemes; however, improvements could be made
through partial or complete use of renewable sources such as biofuels or wind energy. Again
this has little potential when looking into existing power stations as it would require costly
modifications to existing plants[9]. In order to have a sizeable impact on current CO2
emissions, sequestration needs to be investigated.
There are a number of technologies currently being researched for post combustion CO2
sequestration. The most mature is chemical absorption and the most extensively studied
within chemical absorption is monoethanolamine (MEA) absorption[8], which has been used
for many years in industries for a range of separations including CO2 recovery and natural
gas sweetening [9]. Amine absorption has been a prime candidate for CO2 sequestration for
some time. Some of the advantages associated are that it is a well developed technology
which is highly selective and can produce high purity CO2. However, it carries numerous
disadvantages such as large energy penalties, large plant footprint, degradation of material
and equipment and a low CO2 loading capacity [8]. The energy penalty and running costs
incurred through amine absorption is severe and questions the overall feasibility of the
process.
Membranes themselves are a somewhat mature area of technology with 20 years of
commercial experience and 50 years of development [13]. Loeb and Sourirajan first created
high flux anisotropic membranes with large surface area capabilities in 1960 which attracted
considerable interest due to applications to reverse osmosis[13]. Permea adapted the Loeb
Sourirajan membrane to produce the world’s first commercially available membrane, Prism,
in 1980. This was successfully applied to H2 recovery processes and showed the potential of
membranes for performing low cost separations. There has been substantial research into
membranes over the past 10-15 years as the market share has grown rapidly to an estimated
£2-3 billion per year[14]. The majority of the research has been focused on advancing the
Page | 6
permeability and selectivity of membranes; two key properties for any separation.
Advancement on this front has been limited with the exception of some membranes which are
in early stages [15], and will most likely not be ready for commercialisation for some time.
Some success has come from development of hybrid membranes which can combine optimal
features from two or more membranes with impressive properties. It has been suggested that
for significant improvement a deeper understanding of the transport on a molecular level
should be sought after[16], which suggests that our current understanding and models of the
transport within membranes is insufficient.
As mentioned before, membranes are already in widespread use in industry; one of the
greatest benefits from membranes is the process intensification scale that can be achieved due
to their small footprint and efficiency. The use of membranes in water desalination can yield
a process 10 times more energetically efficient than thermal options [16] and Baker [17]
states that 5000-10000 N2 separation units are in operation worldwide and account for 1/3 of
N2 production.
The application of membranes in the sequestration of CO2 is a relatively undeveloped area
compared to industrial applications. Much research has been done to try and ascertain the
feasibility and optimum processes for it but little has been done in the way of larger scale
investigations. A pilot plant was constructed in 2008 in Arizona, Cholla, to test a membrane
separation process on the flue gas stream of a 995MW power plant [18]. The project is
looking to investigate a 6 month operational period with emphasis on the cost efficient
fabrication of a membrane skid and assessing the performance of the membrane. A map of
cost reductions for feasibility will also be generated. The initial results are promising with
respect to contaminants such as SOx within the process but the overall results are limited as of
yet. Another pilot plant has been constructed and reported on in Australia but there is limited
data as to the success of it [19]: CHEMECA [20] stated that it was producing purities in the
range of 20-50% but with no specifications on the recovery achieved, it can be assumed that
the process is achieving 90% recovery.
2.2 Theory of membrane separation units
Membranes have been used for many years to perform separations on a variety of fluids
[21].As discussed in section 2.1 some processes are well defined and have been optimised
over many years and some are emerging technologies which require much research. There is
Page | 7
a substantial amount of theory surrounding membrane separation which must be understood
if any process is to be correctly modelled.
2.2.1 Membrane types
The basis for all membrane separations is that there are two process streams divided by a
permeable wall, or membrane. The structure of the membrane can vary greatly depending on
what material it is constructed from but the membrane types largely fall into two categories:
porous and dense.
Porous membranes are those which have many pores permeating through the membrane,
these are analogous to a conventional sieve or filter; it is by these pores that the membrane is
able to separate different gases. Figure 1a shows a porous membrane; the separation takes
place chiefly by physical size separation where smaller molecules will be able to pass through
the pores and larger ones will be unable to fit through, the driving force is primarily the
pressure gradient and molecular motion of the gaseous species. Due to the pore sizes used (20
- 1000+ Å [13]) these membranes are used for separation of large molecules such as colloidal
mixtures or bio molecules such as proteins[22].
Figure 1 [13] - (a) porous membrane (b) dense membrane
Dense or polymeric membranes consist of a dense polymeric material which separates
gaseous molecules by allowing certain species to preferentially adsorb and diffuse through
the membrane; this can be seen in figure 1b. The separation will therefore not only work on
the basis of size but will favour different molecules; this is advantageous when separating
molecules of similar size such as in flue gases. The driving force for dense separations relies
on the chemical potential gradient of the species in question. Dense membranes will be
Page | 8
focussed on in this paper as these are the most promising membranes for a commercially
viable gas separation process[21].
2.2.2 Models used
There have been a number of models used to describe the diffusion of species through
membranes. The modelling process carried out in this simulator must be understood if an
investigation is to be successfully carried out. The relationships used to model the process are
laid out in this section.
The models used to describe porous membrane transport are problematic; as of yet no unified
theory has been accepted for widespread use[13]. This is largely due to the heterogeneous
nature of porous membranes, as seen in figure 1a. Another difficulty is the lack of a
satisfactory set of parameters that can be used to correctly characterise porous membranes.
As discussed in 2.2.1 porous membranes are unable to separate gaseous mixtures due to the
limiting size of the pores and as such will not be discussed any further.
The model used to describe transport through polymeric membranes is known as the solution-
diffusion model. This has been developed through a series of improvements over a period of
20 years from 1960-1980 [13]. The transport of gaseous species through a polymeric film can
be seen to happen in 3 main steps:
1- The molecule is absorbed into the membrane feed surface and dissolved into the
polymer
2- The molecule diffuses through the polymeric membrane
3- The molecule is desorbed from the permeate side of the membrane
The absorption and desorption of the molecule is relatively simple to model and can be
represented by the conditions (Temperature, Pressure and Concentration) of the membrane
feed and a characteristic known as the sorption coefficient. This will be covered in more
detail later.
The diffusion of a gaseous molecule through a polymeric membrane entails the molecule
moving through the empty spaces created by the thermal movement of polymer chains,
known as the free volume [13]. The relations used to describe the diffusion through a
membrane start with the proposition that the driving force for the gaseous species is
dependent on the chemical potential. The use of chemical potential effectively links the
Page | 9
temperature, pressure, concentration and electrical potential together as can be seen in
equation 2 and 3.
By using the observation that the movement of a species from a high to low concentration is
governed by Ficks law we obtain equation 1[13] to describe species flux across the system:
(1)
Where Ji is the flux of component i, Li is the coefficient of proportionality and µi is the
chemical potential of component i in the membrane, R is the molar gas constant, T the
temperature and ni the number of moles of component i.
It can be seen that an expression describing the component concentration gradient within the
membrane is required to model the flux of species i. This is acquired using the chemical
potential of the species throughout the membrane. The chemical potential of the species is
equated at the membrane feed interface: the feed gas phase is taken as a compressible fluid;
the membrane phase as an incompressible medium. The chemical potential of each phase can
be seen in equations 2 and 3 [13]:
– Incompressible medium - (2)
– Compressible fluid - (3)
Here, γi represents the activity coefficient of component, vi the molar volume and p the
pressure. A superscript o indicates a reference value and a subscripted o indicates the feed
interface with subscripted m indicating the membrane phase. Equations 2 and 3 can be
rearranged to provide an expression for the concentration of the species in the membrane as
shown in equation 4 which is simplified due to the exponential term being very close to one:
(4)
By calculating the concentration of component i from equation 4 and combining the terms for
the feed-membrane and membrane-permeate interfaces and then combining this with
equation 1 we can arrive at equation 6 which fully describes the flux across the membrane. A
sorption coefficient has been defined as in equation 5:
Page | 10
(5)
(6)
Here, ρm denotes the molar density of component i and Di the diffusion coefficient. l is the
thickness of the membrane. The subscripted l indicates the permeate side of the membrane. In
equation 6, KiGDi are combined to give the permeability coefficient of the gaseous species.
Similarly, an expression can be derived describing the permeate-membrane interface. In
order to model the flux across the membrane system a mass balance is used; a mass balance
is constructed over both interfaces of the membrane. The mass balance for the feed side can
be seen in equation 7, the mass balance for the permeate side has a positive flux as
components are flowing into it. The system is treated as a 1D plug flow scenario; other
assumptions taken are that the system is isothermal, isobaric, the gases behave ideally and
that there is constant permeability.
(7)
In equation 7, the permeability coefficient and the thickness are combined in order to give the
overall permeance of the membrane with respect to species i, π; the permeance is analogous
to the permeability of a membrane. Furthermore, to allow consideration of other components
the permeance of component i is taken as the key permeance and a dimensionless ratio
known as the selectivity is defined, α. This relates the permeance of each component relative
to component i. A more thorough explanation of the modelling of membrane processes can
be seen in [23].
The simulator uses the backwards finite difference method to solve the mass balance over the
interval. This method uses two calculated points and a solution interval to solve to varying
levels of accuracy. By calculating F at each point and taking the difference between these
points the method can approximate the gradient of the function (in this case dF/dA). This is
solved iteratively until the values are in close agreement. One step of the method can be seen
in equation (8):
(8)
Page | 11
Where h is the step size and relates to the interval between j and j-1. The smaller the step size
the greater the accuracy that is achieved.
2.2.3 Membrane parameters
It is obvious the flux of any species will depend heavily on many factors. Significant research
has been done into the effect these parameters have on membrane separations. This section
looks at reported trends and the reasons for any relationships present.
Membrane - Arguably the most important feature of any membrane process is what
membrane and therefore what selectivity and permeance is being used. If a process is to meet
the targets set by the DOE then the optimal values for permeability and selectivity will need
to be chosen. It has been shown that an increase in permeance will give an increase in
recovery [4, 12]. This can be explained through equation 7 which shows the permeance being
directly linked to the flux and therefore the recovery of CO2. Merkel also shows that an
increase in permeance of 400% will lead to a lower capture cost by approximately 50% [4]. It
can be seen that increasing the permeance will slightly lower the purity of the captured CO2
but to an almost negligible degree; conversely by altering the selectivity we can change the
purity but with negligible effect on the recovery [12]. As discussed in 2.1 these values are set
by the membrane used and are therefore relatively rigid with respect to membrane processes.
Membrane Area - The area of the membrane used has been mainly used as a benchmark for
the success of a process. This largely holds true for processes as the capital cost of
membranes is prohibitive; the polaris membrane can cost upwards of $50/m2 [4], this can be
very expensive when membrane areas reach into millions of m2. A well reported trend is that
with an increase in area the recovery of the process will increase and the purity of the
permeate will decrease [4, 12].
Membrane Configuration - One parameter with countless different options is the
configuration of the membrane units along with the use of either compression or vacuum to
drive the flow. Zhaos paper looks into the use of different configurations of compressors and
vacuums used to achieve 70% recovery [24]. The variation of configuration leads to large
differences between the energy and the area required for the (Area varying from 19m2 to
404m2, energy use varying from 113kWh/te to 312kWh/te): the key result being that there is
a trade off between area and energy use. Using a vacuum pump is attractive as although
vacuums are usually more energy intensive to produce the equipment will have to deal with a
Page | 12
lower flowrate as it is being place on the permeate stream which should lead to lower energy
consumption, however [14] shows that with vacuum suction a larger area will be required
leading to another trade off between favourable values.
A key area with less research done is the actual configurations and how they can be varied
with respect to recycle streams and what effect this has on the efficiency of separation [25].
This is to be looked into in depth in this report and the results analysed.[12] shows that with a
single membrane we cannot expect to achieve both high recovery and high permeate purity
within an acceptable cost. This well researched finding has meant that more and more
research is being done into dual and triple stage separations. With this a much higher
separation can be achieved but with the key result of more energy being used as compressors
and vacuums are needed for each stage.
Sweep Configuration - The use of a sweep stream and the recycling of streams has had some
investigation but the effect of differing compositions of sweep stream on the separation has
not been fully researched yet. The typical sweep stream used is a set amount of the retentate
stream. The flowrate of the sweep stream can also have an effect on the separation as this will
directly affect the PPDF of the species and therefore warrants investigation.
Temperature - The temperature and pressure of operation are also extremely important
parameters as they can have significant effects on the results of the separation. The
temperature can be seen to have a direct affect on the flux achieved through equation 7; this
is closely linked to the idea of thermal motion and the fact that a higher temperature will lead
to larger molecular motions. The actual operating temperature is fairly limited by the
membrane being used; most membranes are not able to withstand elevated temperatures and
could experience adverse performance if used out with close to ambient temperatures. Due to
this, there has been very little work done on the effect of temperature on the separation
achieved and it is unlikely that there is much to be done in this area. The development of
new, more robust, membranes able to withstand higher temperatures whilst still possessing
desirable permeance and selectivity would warrant further investigation into this area;
inorganic membranes have been tested at temperatures as high as 600°C while polymeric
membranes area usually limited to around 200°C [13, 16].
Pressure - The pressure operated at is a major factor for separations as this not only has a
direct effect on the separation achieved but heavily influences the power consumed. Much
Page | 13
work has been done into operation at different pressure ratios[4, 12]. The pressure ratio can
be seen to limit the concentration enrichment attainable through equations 9 and 10[4]. These
show that the concentration enrichment cannot exceed the pressure ratio between the feed and
permeate. Merkel has shown that with an increase in pressure ratio the energy required to
achieve 90% recovery increases but the area also decreases [4]; the optimum ratios were
identified at values between 5 – 10.
(9)
(10)
2.3 Conclusion
As shown in this section, there has been a lot of work done in the area of CO2 sequestration
with many different technologies being investigated. This creates a very competitive
environment for achieving satisfactory separations. The options of pre and oxy combustion
are becoming more attractive with respect to building a new power plant but the end of pipe
techniques for post combustion should not be overlooked.
The theory behind membrane transport has been heavily researched over 50 years and can
describe membrane processes well but little knowledge is possessed of detailed molecular
movements during separations. With the introduction of supercomputers this is being
developed and may yet yield results which are able to be applied to new membranes.
Membranes have been developed commercially over the past 30 years for separations of
many kinds with some becoming industry preferred choices. The separation of CO2, however,
is relatively undeveloped and although much work has been done in the area, still requires
progress if it is to compete with the alternatives.
The successful trial of pilot plants will demonstrate the actual effectiveness of membranes for
separation of flue gases and will be key to determining whether further research could yield
results. From the work done so far on membrane separations of CO2 from flue gases it seems
unlikely that the targets required by the DOE will be able to be met by membranes within the
near future; a significant breakthrough is required if membranes are to become competitive
with other technologies.
Page | 14
3 Modelling Process
3.1 Aims
The aim of this process was to model dual stage membrane separation simulations using the
Unisim design R400 package. The separation was to be performed on the flue gas stream of a
600 MWe plant as specified by the DOE report [11]. The feasibility of achieving 90%
recovery and 90% permeate purity of CO2 using two membrane units was investigated. The
parameters investigated ranged from membrane configuration to pressure ratio and can be
seen summarised in section 3.2.3. A number of key aspects of the process are to be monitored
including CO2 recovery, permeate purity and energy use.
3.2 Simulator Modelling Strategy
As shown in section 2.2.3 there are a number of key variables to be investigated in order to
find the optimum configuration of a dual stage membrane process. The method of
investigation of these variables is set out in this section.
As many papers have investigated obtaining high recovery (70/90%) and economic energy
use with little focus on the permeate purity this report will chiefly aim to achieve the DOE
targets of 90% recovery and 90% permeate purity and then look into the power consumed.
When using dual stage membrane processes it is important to note that one membrane will
typically be deemed the high recovery membrane and one the high purity membrane. The
high recovery membrane will be the membrane which has the retentate exiting the process;
this membrane requires high recovery because as much CO2 needs to be removed as possible
before discharging the retentate to the atmosphere. The high purity membrane is the
membrane which has the permeate stream going to sequestration: high purity is required here
if the DOE targets are to be met [11]. Examples of high purity and high recovery membranes
can be seen in figure 2c; the membrane with streams 1 and 2 entering it is the high recovery
membrane and the membrane with stream 5 entering is the high purity membrane.
3.2.1 Manipulated Variables
There are a number of variables that should be manipulated in order to ascertain their effect
on the success of the separation process and find the optimal setup for a dual stage process.
Page | 15
Membrane Configurations - The first area to be investigated was the use of different
configurations for the membrane unit. As each unit has four streams associated with it this
gives rise to a large number of possible configurations that could be used. Not all of these
configurations will give useful or indeed meaningful benefits to the separation. A total of 5
configurations have been chosen for investigation as these have been specified as promising
arrangements in Agrawals paper [25]. These can be seen in figure 2:
Figure 2 [25]- Membrane configurations
Investigating these configurations will allow the optimal arrangement to be determined with
respect to the key performance parameters. Each configuration will have the sweep fraction,
flowrate, membrane parameters and pressure ratio held constant and the areas varied until
90% recovery has been achieved. This should allow an effective comparison of the ability of
each configuration to achieve the separation. For the first membrane unit in figure 2c stream
1 is the feed, stream 2 the sweep, stream 3 the retentate and stream 4 the permeate.
Membrane Area – As discussed in section 2.2.3 the area of the membrane used in the
separation has a direct effect on the separation achieved. Membranes can generally achieve a
high recovery through utilising a large area which is able to recover most of the CO2 passing
over the membrane; a consequence of equation 7. The area of each membrane in the
configuration will be altered from the lowest possible (some membrane areas are unstable
and cannot be solved with this simulator, this is discussed further in section 3.4) to a
sufficiently high value. The effect this has on both the recovery and permeate purity will be
determined for both high purity and high recovery membranes.
Page | 16
Sweep Flowrate Fraction – The sweep flowrate fraction is an important variable. The
fraction is as defined in equation (11) and directly affects the process. The sweep flowrate
fraction is the molar flowrate percentage based on the feed.
(11)
Where n dot represents the molar flowrate of the stream. The effect on the process is derived
from the PPDF governing the transport. By either changing the flowrate of the sweep gas or
its composition we can directly affect the overall PPDF and therefore the flux through the
membrane. By using a high sweep flowrate fraction (typically 10-20%) the recovery of a
membrane can be increased. A lower sweep fraction is utilised in high purity membranes:
typically 0-100 mol/s or less than 1% is used. These will be varied from 0-2% for the high
purity membrane and 0-20% for the high recovery membrane.
Pressure ratios – The use of pressure ratios is an important aspect of membrane separations
as this can greatly affect the requirements of the process. A larger pressure ratio will mean
that a lower area is required for the same recovery. This is a direct result of the pressure ratio
issue discussed in 2.2.3, equation 10. Using a larger pressure ratio will lead to increased
running costs as the energy consumed by the compressors accounts for the majority of the
overall energy use. The pressure ratio will be varied from 5 to 10 (corresponding to a change
in inlet pressure from 110 kPa to 220 kPa where the outlet is kept constant at 22 kPa), this has
been deemed as the suitable limit for pressure ratios with respect to power consumption[4].
Membrane Selectivity and Permeance – The selectivity and Permeance of membranes are
determined by which membrane is being used and as such the actual values feasible in reality
are limited. The values used for the majority of the report are those of the Polaris Membrane
which has a CO2/N2 selectivity of 50 and a CO2 permeance of 1000 GPU [4]. A theoretical
membrane is also investigated which has a selectivity of 100 and a permeance of 12000 GPU,
this has been named PIM + + and utilises the permeance of a membrane known as PIM [15]
and the selectivity which has been deemed the optimal limit attainable.
3.2.2 Monitored Variables
In order to successfully analyse this simulation a number of variables need to be monitored
and analysed to provide key performance parameters as discussed in section 3.2.
Page | 17
The first and foremost variable to be monitored is the flowrates into and out of the
membranes. The molar flowrate of each component should be recorded for the feed stream,
sweep stream, permeate stream and retentate stream. By monitoring all of these flowrates we
can determine the necessary key performance indicators.
The simulator used had a built in spreadsheet function that allowed all molar flowrates to be
measured and dynamically displayed. This ensured all variables could be recorded and key
performance indicators calculated efficiently and swiftly.
The power consumption of all units was monitored. All compressors were taken as a positive
power consumption owing to the energy required to raise the pressure of components. The
power consumption of turbo expanders was taken as negative owing to the energy released
when a gases pressure is reduced.
3.2.3 Key Performance Parameters
Certain key performance parameters were to be calculated in order to ascertain the success
and feasibility of the separation achieved. The key performance indicators for this process are
defined as the recovery, the permeate purity, the retentate purity and the overall energy use.
The recovery is defined how much of the CO2 entering the process is recovered in the
outgoing permeate stream (stream 7 in figure 2c). The recovery was calculated as in equation
(12):
(12)
The permeate and retentate purities were taken as the mole fractions of the corresponding
streams with respect to CO2 (streams 3 and 7 respectively in figure 2c). This could be easily
calculated using unisims spreadsheet function and equation (13):
(13)
The overall power consumption of the process was determined simply by measuring the
power consumption of individual compressors and expanders and combining them as laid out
in 3.2.2 and equation (14):
(14)
Page | 18
Another useful indicator for comparison between processes is the energy required to capture
one mole of CO2, or the specific power consumption. This is easily worked out from the data
gathered and is defined as in equation 15:
(15)
Table 1 - Summary of Variables
Manipulated Variables Monitored Variables Key Performance indicators
Membrane area
Membrane Configurations
As seen in figure 2
Sweep Flowrate Fraction
Permeate and Feed side pressures
Membrane Selectivity and Permeance
Molar Flowrates
Feed Stream
Sweep Stream
Permeate Stream
Retentate Stream
Power Consumption
CO2 Recovery
Permeate purity (CO2 content)
Retentate purity (CO2 content)
Overall power consumption
Specific Power consumption
3.2.4 Simulator Interface
The simulator used was an addition to the design program, Unisim Design R400. It was
developed by the University of Edinburgh to model membrane separations with a focus on
CO2 from flue gas streams.
The interface consisted of 4 panels. The first panel allowed the user to connect feed,
permeate, retentate and sweep streams to the unit operation. The second panel allowed the
user to enter the parameters of the membrane: area, key permeance, selectivity of each
component. The third panel allowed the user to set the number of grid points used int he
solving mechanism. The fourth panel was the standard Unisim worksheet panel and allowed
the user to monitor all streams entering and leaving the membrane. The interface of the
membrane unit was well integrated into the Unisim environment and was able to be
understood quickly with a basic knowledge of Unisim. The interface panels can be seen in
appendix IIIa.
As mentioned in 3.2.2 the spreadsheet function was used to aid data gathering. The
spreadsheet layout can be seen in appendix IIIb. All relevant details are easily observed and
key performance parameters are calculated automatically.
Page | 19
3.3 Results
Initially all 5 configurations were investigated to try and ascertain which was the most
efficient at achieving 90% recovery of CO2. The permeate purity was to be investigated
afterwards. The results of attaining 90% recovery can be seen summarised in table 2:
Table 2 - Configuration Investigation
Configuration Area 1 Area 2 Total
Area
Recovery Permeate
purity
Retentate
Purity
Power
consumption
Specific Power
consumption
x105m
2 x10
5m
2 x10
5m
2 % % % x100MW kJ/mol CO2
1 50 6.1 56.1 83.53 51.95 .5/8.39 .9714 39.62
2 22 4.2 26.2 90.76 33.57/9.236 1.75 0.62 20.41
3 43 2.75 45.75 90.12 63.76 1.128 1.335 50.45
4 5.5 42 47.5 90.17 63.28 1.142 1.264 47.76
5 2.7 43 45.7 90.09 63.69 1.221 1.295 48.9
It should be noted that the pressure ratio for each configuration was kept at 5 (inlet pressure
of 110 kPa; outlet pressure of 22 kPa) unless stated otherwise.
It can be seen from table 2 that configurations 3,4 and 5 are all promising configurations as
they are able to achieve 90% recovery with a substantial permeate purity. Configuration 1 is
by far the weakest. This required the largest area of all configurations, was unable to achieve
90% recovery (partially due to simulator stability issues discussed later) and gave a retentate
stream with too high a CO2 content. Configuration 2 was promising with respect to achieving
90% recovery at a low energy cost and membrane area but was unable to produce a
satisfactory permeate purity. Configurations 1 and 2 are unable to achieve satisfactory
retentate or permeate streams due to the well known issue mentioned in Zhao [12] stating that
1 membrane unit is unable to provide both recovery and purity, showing that both a high
recovery and high purity membrane area required. Configuration 1 can achieve a low
retentate purity at first but is unable to recover much in the second membrane if purity is also
desired here. Configuration 2 can recover a large amount in both but again but will not be
able to provide high purity at the same time.
Configurations 3, 4 and 5 are all comparable as the areas are similar and specific power
consumption is also similar. Configuration 4 would seem to be the most attractive at this
Page | 20
point due to the lower specific power consumption (~4% lower than 3 and 5) which would
most likely be the deciding factor due to the overall similarity in membrane areas. All
retentate purities are acceptable as they are below 2%. The permeate purities here appeared to
reach a practical limit approaching 65% and surpassing this value seemed problematic.
It should be noted that the pressure ratio issue discussed in equation 10 does not limit this
process. Although the input concentration of CO2 is 14.1% which would suggest a limit of
70.5% at a pressure ratio of 5, the use of two membrane units and a recycle loop changes this
as the feed to the high purity membrane will be significantly higher, around 40% in most
cases, which theoretically allows the process to surpass 90% purity.
From the results for achieving 90% recovery of CO2 with each configuration it can be seen
that configurations 1 and 2 are not able to fulfil the DOE goals adequately due to the
insufficient retentate or permeate purity issues as discussed above. From these results 3,4 and
5 are the most promising and these will be taken forward for further investigation. The next
stage was to investigate the effect of changing the area of the membrane. Initially the area
was varied and all relevant key performance indicators calculated. The results of varying the
area can be seen in figures 3 a,b and c; it should be noted that the numbers next to each point
correspond to the area of the membrane measured in 105m. All parameters were kept at the
values in table 2 unless being altered.
It should be noted that not all values recorded are included in this section and certain values
have been omitted for the sake of clarity. All recorded values can be seen in appendix IV
which is on a data CD. For more detailed information on the position of compressors and
expanders appendix IIIc should be examined, this is also on the data CD.
Page | 21
Figure 3 - (a)(b)(c) effect of changing membrane areas
The variables changed in figure 3 a,b and c were:
The area of the purity membrane
The area of the recovery membrane
52
57
62
67
72
82 84 86 88 90 92 94
Pe
rme
ate
pu
rity
(%
CO
2)
Recovery (%)
Figure 3 a - Configuration 3 area change
purity membrane area
recovery membrane area
56
58
60
62
64
66
68
82 84 86 88 90 92 94
Pe
rme
ate
Pu
rity
(%
CO
2)
Recovery (%)
Figure 3 b - Configuration 4 area change
purity membrane area
recovery membrane area
56
57
58
59
60
61
62
63
64
65
66
82 84 86 88 90 92 94
Pe
rme
ate
Pu
rity
(%
CO
2)
Recovery (%)
Figure 3 c - Configuration 5 area change
purity membrane area
recovery membrane area
Page | 22
All 3 configurations can be seen to obey similar relationships; with an increase in the purity
membrane area the recovery will increase but the permeate purity will decrease. This has
been widely reported as a standard trend and can be linked to equation 7 [4, 12]. By
increasing the area of the membrane, the flux of CO2 is increased leading to a higher
recovery. However, by increasing the membrane area, the flux of other molecules (in this
case H2O, N2 and O2) also increases, leading to a decrease in permeate purity as the CO2
concentration is diluted. The specific power consumption falls with an increase in the purity
membrane area, this is due to the increased flux across the membrane. This will not only lead
to more power recovered from the expander operating on the permeate stream but the larger
amount of gases removed from the process will mean that less compressing power is required
in other streams such as the recycle.
The configurations can also be seen to increase in recovery and permeate purity if the area of
the recovery membrane is increased. This is due to the fact that a larger area will allow a
larger CO2 flux as shown above, again leading to an increase in recovery. This sends more
CO2 to the high purity membrane and as the purity membrane area is constant causes the
recycle stream and therefore the feed stream to be richer in CO2. The increased CO2 content
passing over the high purity membrane causes an increase in CO2 purity in the permeate
stream. This trend would most likely have been different if configurations 1 and 2 had been
investigated. The lack of recycle would have caused the permeate purity to decrease slightly
as the feed would not have been as rich in CO2. The expected result here would have been for
the recovery to increase and the purity to decrease as has been shown in literature [4], this
highlights the importance of the recycle stream. An increase in the size of the recovery
membrane causes an increase in specific power consumption as the larger membrane flux
creates a much larger recycle stream which needs to be recompressed before entering another
membrane.
The trends do not behave ideally as a straight line would be expected. This can largely be
attributed to the simulation solver, equation 8 in particular, and is discussed further in 3.4
The next step was to investigate the effect that changing the flowrate of the sweep stream had
on the separation process. The investigation into the sweep stream can be seen summarised in
figure 4 a,b,c where the numbers on the graph correspond to the sweep flowrate fraction:
Page | 23
Figure 4 - (a)(b)(c) Effect of changing the sweep fraction
The variables changed in figure 4 a,b and c were:
The sweep flowrate fraction of the purity membrane
The sweep flowrate fraction of the recovery membrane
56
57
58
59
60
61
62
63
64
65
66
67
76 78 80 82 84 86 88 90 92
Pe
rme
ate
pu
rity
(%
CO
2)
Recovery (%)
Figure 4 a - Configuration 3 Sweep change
recovery membrane sweep fraction
purity membrane sweep fraction
56
58
60
62
64
66
68
75 77 79 81 83 85 87 89 91 93
Pe
rme
ate
pu
rity
(%
CO
2)
Recovery (%)
Figure 4 b - Configuration 4 Sweep change
recovery membrane sweep fraction
purity membrane sweep fraction
54
56
58
60
62
64
66
68
76 78 80 82 84 86 88 90 92
Pe
rme
ate
pu
rity
(%
CO
2)
Recovery (%)
Figure 4 c - Configuration 5 Sweep change
recovery membrane sweep fraction
purity membrane sweep fraction
Page | 24
Again all three configurations can be seen to obey similar relationships. When the sweep
flowrate fraction of the recovery membrane is increased, the recovery and permeate purity
can be seen to increase. This is due to the CO2 concentration being decreased in the permeate
side due to the increased air flowrate; this causes a larger PPDF and increases the CO2 flux
across the membrane leading to a higher recovery. The increased CO2 feed to the permeate
membrane causes a higher CO2 flux relative to other components leading to a higher
permeate purity. The increased CO2 content also passes on to the recycle leading to a higher
feed content of CO2, these two effects act together to increase both the recovery and purity.
The specific energy consumption of process rises with an increase in recovery membrane
flowrate as this allows more gas into the system to be recompressed for the second membrane
and also leads to a larger recycle stream.
An increase in the sweep flowrate fraction of the purity membrane can be seen to cause a
slight increase in recovery and a decrease in purity. The dilution of the CO2 concentration due
to the increased air flowrate leads to a lower permeate purity and this causes a larger PPDF
which allows a larger CO2 flux which causes the increase in recovery. This is compounded by
the lower CO2 content in the recycle further leading to a decrease in permeate purity. The
specific power consumption for an increase in purity membrane sweep flowrate fraction
decreases as the larger flow exiting in the permeate stream allows more energy to be
recovered from the turbo expander downstream.
An interesting result in changing the recovery sweep fraction is that the recovery can be seen
to decrease after a certain value (may not be completely clear here due to accuracy issues).
This is most likely due to the fact that increasing the sweep flowrate will eventually lead to
the PPDF being decreased due to the CO2 content in the air sweep stream. The trends
observed here are largely the same when repeated at a pressure ratio of 10, however the point
at which a configuration reaches a maximum with respect to recovery can be seen to change
significantly (at a pressure ratio of 10 the maximum was 6.7% as compared to a maximum at
13% for a pressure ratio of 5).
Configuration 4 would appear to be different in that the recovery does not tail off as it does
with configurations 3 and 5. Configurations 3 and 5 can be seen to reach a limit of recovery
at different sweep flowrate fractions (13% for configuration 3 and 17.7% for configuration 5)
and therefore it can be assumed that the range of investigation was not large enough to allow
the maximum to be reached for configuration 4.
Page | 25
The key result from these investigations is the knock-on effect of the recycle and the fact that
this can change reported trends.
The next step was to investigate the effect that changing the pressure ratio had on the
processes, this can be seen summarised in table 3:
Table 3 - Pressure ratio investigation
Config. area 1
area
2
Total
Area
Recover
y
Permeate
mol frac
Retentate
mol frac.
energy
use
pressure
in
pressure
out
press
ure
ratio
x105m2 x105m
2 x105m
2 % %CO2 %CO2 x100MW kPa kPA
3 43 2.75 45.75 90.02 63.76 1.128 1.335 110 22 5
3 13.5 1 14.5 90.22 76.6 1.198 1.995 220 22 10
4 5.5 42 47.5 90.7 63.39 1.147 1.268 110 22 5
4 1.25 16.5 17.75 89.99 78.15 1.146 2.595 220 22 10
5 2.7 43 45.7 90.09 63.69 1.221 1.295 110 22 5
5 1 15 16 90.82 75.55 0.8409 2 220 22 10
The area required for each configuration to achieve 90% can be seen to drop drastically; by
around 65% for each configuration, this represents substantial capital cost savings in a
membrane process. This is due to equation 7 and the PPDF increasing meaning that a smaller
area is required to provide the same CO2 flux relating to 90% recovery. The downside to
increasing the pressure ratio is the substantial increase in energy use due to the additional
compression; an increase of 49%, 104% and 54% for configurations 3, 4 and 5 respectively.
The permeate purity is increased which is most likely a result of the pressure ratio discussed
in equation 10. By doubling the pressure ratio the concentration enrichment possible is
effectively doubled as well which leads to a higher permeate purity. The retentate can be seen
to remain relatively constant. A decrease in retentate purity is seen in configuration 5 and can
be explained due to the increased recovery leaving a lower CO2 flow in the retentate.
The effect of changing the sweep stream from an air stream to the recycled retentate stream
was then investigated. This is effectively splitting stream 6 in figure 2c and recycling this to
streams 2 and 8 to be used as the sweep. The results can be seen summarised in table 4:
Page | 26
Table 4 - Sweep composition investigation
Setup area 1 area 2 Total Area Recovery
Permeate
mol frac
Retentate
mol frac.
energy
use
pressure
ratio
x105m
2 x10
5m
2 x10
5m
2 % %CO2 %CO2 x100MW
C3 – air sweep 43 2.75 45.75 90.02 63.76 1.128 1.335 5
C3 – double ret.
sweep 40 3 43 90.16 62.41 1.419 1.263 5
C3 – double ret.
sweep 13 1.05 14.05 90.17 75.75 1.349 1.939 10
C4 – air sweep 5.5 42 47.5 90.7 63.39 1.147 1.268 5
C4 – double ret.
sweep 6 38 44 90.38 59.5 1.281 1.285 5
C4 – double ret.
sweep 1.15 18 19.15 90.7 79.35 1.162 2.967 10
C5 – air sweep 2.7 43 45.7 90.09 63.69 1.221 1.295 5
C5 – double ret.
sweep 2.85 41 43.85 90.02 62.48 1.447 1.247 5
The recycling of the retentate actually leads to an increase in the CO2 content of the sweep;
this seems counterintuitive as by changing the sweep to the retentate the PPDF is decreased
as the retentate has a higher CO2 content than the air being used. By recycling the otherwise
discarded retentate stream it allows the process to capture some more CO2 from the stream
and therefore increase the recovery (or allow a smaller area to provide the same recovery).
The decrease in purity is a result of the lower PPDF allowing less CO2 to permeate. A small
decrease in power is seen when the sweep stream is recycled, this can be attributed to the use
of an already pressurised retentate stream being put through an expander to recover some
energy. The air stream previously used was being expanded from 101.3 kPa to 22 kPa and the
retentate stream is being expanded from 110 kPa to 22 kPa therefore allowing more energy to
be recovered than before.
With the major parameters investigated it can be seen that achieving 90% recovery and 90%
permeate purity is unrealistic without consuming a significant amount of energy. As
mentioned above, a theoretical membrane known as PIM + + was tested. This membrane is a
hybrid between the PIM membrane[15] and a theoretical limit for membrane selectivity. PIM
+ + would have the selectivity of PIM and a selectivity of 100 which has been deemed as an
Page | 27
attainable limit for membranes. If this membrane can be produced commercially membrane
processes stand a much better chance at achieving DOE targets.
Table 5 - Testing of PIM + + membrane
Total Area Recovery
Permeate
mol frac
Retentate
mol frac.
energy
use
pressure
ratio
x105m2 % %CO2 %CO2 x100MW
config. 3 3.76 87.24 71.6 1.615 1.405 5
config 3. ret.
Recycle. PR 10 1.406 89.28 84.9 1.409 2.558 10
config. 4 3.7 83.9 72.67 2.12 1.358 5
config 4. ret.
Recycle. PR 10 1.56 83.89 79.24 2.77 4.146 10
config. 5 3.98 89.33 69.38 1.242 1.354 5
config 5. ret.
Recycle. PR 10 1.5525 89.87 85.87 1.14 2.739 10
From these results it can be seen that the increase in membrane parameters gives very
promising results. The configurations running at a pressure ratio of 5 give good results,
although still not achieving DOE targets. The configurations can come within 20% of the
targets at a power consumption of around 22% of the plants output. The increase in energy
consumption relative to the polaris membrane is due to the selectivity allowing less gas to
permeate and causing a larger recycle which has to be compressed. By running at a pressure
ratio of 10 the permeate purity can be increased significantly but at a great cost with respect
to energy consumption. Configuration 5 can almost achieve the DOE targets when used with
PIM + + and a pressure ratio of 10; at an energy use of 45.65% of the power plants output.
The area required decreases massively with PIM + + but will also be much more expensive
as this is not a physically constructed
3.4 Critical Review
This section looks into the simulation and assesses the viability of the modelling process and
how accurate the results may be.
Assumptions
There are a number of assumptions made in this modelling process which could have a
significant effect on the viability of the model produced. The membrane units are assumed to
be isobaric and isothermal. In reality there will be a slight pressure and temperature drop over
Page | 28
the membrane. The temperature drop is a largely safe assumption as with the proper design
any possible drop could be minimised and rendered negligible. The assumption that the
membrane is isobaric is justified as the membrane is a dense polymer and treating it as an
incompressible medium is valid; again, a slight drop may be witnessed in reality but would be
negligible with respect to effects on the process.
The assumption of constant permeability can have more of an effect on the viability of the
process compared to others. In reality the permeability of the membrane would change with
numerous factors: pressure, temperature, time, concentration. For the purposes of the
simulation this assumption is valid as including varying permeability would greatly
complicate the model. Much research has been done into the variance of permeability in areas
such as plasticisation or time dependence and if the simulation was to be carried out
experimentally this could be accounted for. The Cholla pilot plant has reported that the
membranes have been performing well with respect to contaminants which could degrade
performance over time, this would seem to reinforce this assumption.
The assumption of 1D plug flow allows the flux to be easily approximated. It is unlikely that
the flow would be close to 1D plug flow in reality as there would be flow in multiple
directions due to local pressure gradients that would most likely be present. The flux of the
multidirectional flows can be assumed to be at least an order of magnitude smaller than the
assumed flow direction. Also, as all CO2 is entering one plane and leaving the other parallel
plane the overall flow can easily be modelled as 1D plug flow with respect to overall flow
direction.
Lastly, the permeate side flow was assumed to have perfect mixing. This allows the permeate
side concentration to be approximated and easily calculated. In reality the concentration at the
permeate interface would be significantly higher which would act to decrease the PPDF and
lower the flux. Using a sweep flow will add to the mixing on the permeate side but a
concentration gradient would still be present. The extent that this assumption has on the
process cannot be properly determined and would warrant further investigation before
designing any potential membrane processes.
Modelling process
The modelling process aimed to be carried out in the most efficient way which would allow
all necessary variables to be calculated. By looking into the requirements of each
Page | 29
configuration initially, the optimum configurations were able to be ascertained and these
brought forward for further investigation, reducing the time required for simulations.
The parameters investigated in this report seem to be sufficient to give a good understanding
of the possibilities for a membrane separation process. Some additional areas of investigation
which would have been advantageous but were unable to be carried out due to time
constraints are:
- More than two pressure ratios could be investigated to determine any possible
trending effects
- More than two membranes could be investigated so that a deeper understanding of the
membrane parameters could be gained
- The investigations could be carried out with 80 grid points to provide greater accuracy
and more distinct trends
- Carrying out an investigation into power consumption without any sweep streams
may allow a better understanding of how parameters affect this
- An investigation into altering the sweep fraction with a recycled sweep stream could
yield interesting results but would be time consuming due to the use of two recycle
loops
- Investigation into the energy recoverable from heat sources such as compressed gases
could yield considerable savings
Carrying out these additional points would be very time consuming and as such only the main
areas have been investigated this paper.
As the energy consumption is such a key parameter in this process it would be advisable to
carry out further investigation into this to determine the actual accuracy of the values
observed. Unisim takes into account factors that would affect the power consumption of
compressors and vacuums such as adiabatic and isentropic efficiencies and idealities. The
power consumption in reality would likely be different to those calculated but cannot
accurately be judged whether they would be higher or lower.
The key performance indicators used are mostly adequate for analysis of the overall
performance and only one additional performance indicator would be included if the
simulation was repeated, the process selectivity. The process selectivity relates the permeate
CO2/N2 ratio to that of the feed side and is used as a variable in [12].
Page | 30
Notes on the simulator
The simulator was able to model the majority of the process with ease and performed well.
The biggest issue would be the fact that the simulator experienced some stability issues with
solving around certain values.
This was first noticed when loading a previous simulation; the simulator was unable to solve
if the area of each membrane was increased straight to the expected value initially. When
trying to solve, the simulator experienced negative flowrates in some streams and would
eventually reach an inconsistency which prevented it from being solved. Instead the area of
each membrane had to be gradually increased until the required value was reached; this added
some time to the simulations but was not overly time consuming. This was also experienced
with the sweep flowrates. A useful method was to take note of the last values run in the
simulation before the file was saved, if these were put into the simulation it would be able to
run at the required values straight away and would save time.
This stability issue was also observed when altering variables such as the flowrate and area. If
a large change was made to the variable (eg. sweep flowrate of 3000 mol/s to 100 mol/s) the
simulator would be unable to solve. Instead the values should be changed gradually (in steps
of 250-500 mol/s for example) and the simulator would be able to solve successfully.
The simulator solved to acceptable accuracy levels with respect to comparisons of runs and
observing trends; if the process was to be carried out in real life it is unlikely that any higher
accuracy would be desired or even needed. The accuracy is largely determined by the number
of grid points used which relate to equation 8. 50 grid points were used for the majority of
simulations as this gave an acceptable trade off between solving time and accuracy. For final
results of individual runs such as the comparison between all 5 configurations or the pressure
ratio investigation, 80 points were used in order to provide boosted accuracy allowing better
comparisons between results.
Even when 80 grid points were used some discrepancy was noticed when the same values
were simulated twice; the discrepancy was less than with 50 grid points (~0.2% as opposed to
~0.8%). The simulation also seemed relatively sensitive to starting values. When changing
variables the results would often be the same if the values were changed from 0 to the final
value in steps but would change if the values were changed from the final value to 0. This is
Page | 31
due to the fact that the simulator would have the flowrates and compositions from the last run
and would solve iteratively using these as a starting point, leading to small differences in final
values.
Process viability/Results
The main aim of this report was to determine the viability and effect of parameters on a dual
stage membrane separation process.
As discussed above, the results are relatively accurate and can be deemed accurate enough for
the purposes of comparing technologies and determining the viability of the process.
It can be seen from the results that a membrane process will be unable to attain a permeate
purity of 90% at a pressure ratio of 5 within acceptable conditions. Achieving 90% recovery
can be done relatively easy with a membrane area of around 4.5 million m2 and a power
consumption of around 20% of the plant output. With a pressure ratio of 5 the permeate
purity seems to reach a maximum at around 70% which is a high permeate purity but not high
enough to meet the DOE targets.
The investigations into the effect of the area and the sweep flowrate serve to highlight the
importance of the recycle stream. The recycle stream not only allows more CO2 to be
captured and therefore recovered but can have a knock-on effect which can further increase
the permeate purity, or, in some cases decrease the permeate purity.
Merkel states that to be competitive with current absorption the membrane process should use
less than 30% of the power plants output[4]. This shows that it is unlikely that a pressure ratio
of 5 will be able to be competitive with respects to overall purity at the same time as
achieving 90% recovery. Increasing the pressure ratio must be done with attention to the
power consumption as this increases the power consumption significantly.
The final investigations show that in order to approach the 90% recovery and 90% purity
mark we need increasingly difficult conditions such as high pressure ratios and experimental
membranes.
The results obtained seem to show that at the current level of technology it is possible, but not
feasible, to obtain 90% recovery and 90% permeate purity from a dual stage membrane
process.
Page | 32
3.5 Conclusion
The modelling process carried out here effectively shows the fact that there are a large
number of factors which have a significant effect on the success of a membrane separation
process.
This report successfully investigated some of these factors and much research has been done
on a multitude of other factors from different viewpoints. The sheer number of
configurations, variables and performance indicators that can be changed and measured is
prohibitively exhausting. This report has attempted to focus on the parameters largely
relevant to achieving the DOE targets.
The results obtained show that considerable levels of purity can be achieved at relatively low
costs but the reality of achieving DOE targets at the current level of technology and within an
acceptable cost are unlikely to be met. Achieving 90% recovery is relatively easy compared
to 90% purity and can be done within acceptable energy use but at a substantial membrane
area and therefore cost.
Overall, membranes are a promising CO2 separation technology which requires key
commercially available breakthroughs in order to compete with other developing
technologies. Membranes with selectivities in excess of 100 would be beneficial as it appears
that reaching 90% permeate purity is the limiting step for this process.
Page | 33
4 Appendices
I. References
1. IEA, Emissions from Fuel Combustion 1971-2004, 2006a, International Energy
Agency: Paris, France.
2. Quadrelli, R. and S. Peterson, The energy–climate challenge: Recent trends in CO2
emissions from fuel combustion. Energy Policy, 2007. 35(11): p. 5938-5952.
3. The Future of Coal – Options for a Carbon Constrained World. MIT Interdisci-
plinary Study, 2007.
4. Merkel, T.C., et al., Power plant post-combustion carbon dioxide capture: An
opportunity for membranes. Journal of Membrane Science, 2010. 359(1-2): p. 126-
139.
5. J.T. Houghton, Y.D., D.J. Griggs, M. Noguer, P.J. van der Linden, X. Dai, K. and
C.A.J. Maskell, Climate Change 2001: The Scientific Basis. Cambridge
University Press, 2001. New York.
6. J.D. Figueroa, T.F., S. Plasynski, H. McIlvried, R.D. Srivastava Advances in CO2
capture technology – the U.S. Department of Energy’s Carbon Sequestration
Program. International Journal of Greenhouse Gas Control 2008. 2: p. 9–20.
7. Adams, D., Davison, J., Capturing CO2, IEA Greenhouse Gas R&D Programme
Report. 2007.
8. Yang, H., et al., Progress in carbon dioxide separation and capture: A review.
Journal of Environmental Sciences, 2008. 20(1): p. 14-27.
9. Wang, M., et al., Post-combustion CO2 capture with chemical absorption: A state-of-
the-art review. Chemical Engineering Research and Design, 2011. 89(9): p. 1609-
1624.
10. Scheffknecht, G., et al., Oxy-fuel coal combustion—A review of the current state-of-
the-art. International Journal of Greenhouse Gas Control, 2011. 5: p. S16-S35.
11. Cost and Performance Baseline for Fossil Energy Plants. Volume 1: Bituminous
Coal and Natural Gas to Electricity, Aug 2007. DOE/NETL.
12. Zhao, L., et al., A parametric study of CO2/N2 gas separation membrane processes
for post-combustion capture. Journal of Membrane Science, 2008. 325(1): p. 284-294.
13. Baker, R.W., Membrane Technology And Applications. 2004, Wiley: Membrane
Technology and Research, Inc.
14. Haiqing Lin, T.M., Richard Baker, The Membrane Solution to Global Warming, in
Sixth Annual Conference on Carbon Capture & Sequestration2008, Membrane
Technology and Research, Inc: Pittsburgh, Pennsylvania.
15. Budd, P., et al., Gas permeation parameters and other physicochemical properties of
a polymer of intrinsic microporosity: Polybenzodioxane PIM-1. Journal of Membrane
Science, 2008. 325(2): p. 851-860.
Page | 34
16. P. Bernardo, E.D., G. Golemme, Membrane Gas Separation: A Review/State of the
Art. Industrial and Engineering Chemistry Reasearch, 2009. 48: p. 4638-4663.
17. Baker, R.W., Future Directions of Membrane Gas Separation Technology. Industrial
and Engineering Chemistry Reasearch, 2002. 41: p. 1393-1411.
18. Jared P. Ciferno, J.D.F., Tim Merkel, Membrane Process to Capture CO2 from
Power Plant Flue Gas, N.D. The energy lab, Editor 2008.
19. C. Scholes, G.C., W. Tao, G. Stevens, S. Kentish, B. Hooper, A. Qader, Membrane
based carbon capture pilot plant trials, in AMS-5: The fifth Conference of Aseanian
Membrane Society, Kobe, Japan2009.
20. Membrane Based Pilot Plant Trials of Carbon Dioxide Capture. 2011 [cited 2011
14/12/2011]; Conference proceedings of CHEMECA]. Available from:
http://www.chemeca2011.com/abstract/244.asp.
21. Kirk-Othmer, Membrane Technology, Kirk-Othmer Encyclopedia of Chemical
Technology. 2004, Wiley.
22. S. Ripperger, G.S., Wuppertal, Microporous membranes in biotechnical application.
Bioprocess Engineering, 1986. 1: p. 43-49.
23. K. Li, D.R.A.R.H., Mathematical Modelling of Multicomponent Membrane
Parameters. Journal of Membrane Science, 1990. 52: p. 205-219.
24. Zhao, L., et al., Multi-stage gas separation membrane processes used in post-
combustion capture: Energetic and economic analyses. Journal of Membrane
Science, 2010. 359(1-2): p. 160-172.
25. Agrawal, R. and J.G. Xu, Gas-separation membrane cascades utilizing limited
numbers of compressors. Aiche Journal, 1996. 42(8): p. 2141-2154.
Page | 35
II. Nomenclature
Variable Value Units
Ji Molar flux mol/m2s
Li Coefficient of proportionality
µi Chemical potential J/mol
R Gas constant J/K mol
T Temperature K
ni number of moles mol
x distance m
γ activity coefficient mol-1
v molar volume m3/mol
P pressure Pa
Mi molar weight g/mol
ρ molar density m3/mol
Di Diffusion coefficient m2/s
π Key permeance GPU (10-6
cm3/cm
2s cmHg)
α Selectivity N/A
ṅ molar flowrate mol/s
Page | 36
III. Drawings
A – Simulator Data Entry Screens
B – Spreadsheet Data gathering function
C – Please see attached data CD
IV. Data
Please see attached CD