Article

pubs.acs.org/IECR

Design and Assessment of a Membrane and Absorption Based

Carbon Dioxide Removal Process for Oxidative Coupling of Methane
Alberto Penteado,* Erik Esche, Daniel Salerno, Hamid Reza Godini, and Gunter Wozny
Process Dynamics and Operations Group, Technische Universtitat Berlin, Sekretariat KWT-9, Strae des 17, Juni 135, D-10623
Berlin, Germany

ABSTRACT: The oxidative coupling of methane (OCM) is a direct path for converting methane into ethene (ethylene), which
is one of the most important building blocks for the chemical industry. Carbon dioxide is generated as a byproduct in the reactor
and must be separated in order to produce the pure olen. This step is commonly achieved by amine scrubbing, in which a
signicant amount of energy is consumed for regenerating the amine. In this contribution, a hybrid system employing gas
separation membranes and absorption is modeled and simulated in order to carry out preliminary engineering design and
economic evaluations to assess the feasibility of applying this process on industrial scale. It is demonstrated that the hybrid
process oers economic advantage when compared to the standalone absorption process when carbon dioxide is used as diluent
in the OCM reactor.

INTRODUCTION
The oxidative coupling of methane (OCM) is the catalytic
and the products ethane (C2H6) and ethylene (C2H4) into
carbon monoxide (CO) and carbon dioxide (CO2). Besides
conversion of methane into ethene (ethylene), which is one of that, the reaction network is highly exothermic and higher
the most valuable building blocks for the chemical industry with temperatures tend to favor the combustion reactions, reducing
a global production of over 141 million tones in 2011.1 Thus, selectivity toward products. This highlights the importance of
OCM allows for a direct path from feedstock such as natural novel active and selective catalyst development combined with
gas, shale gas, or biogas into value-added chemicals, avoiding ecient reactor design. Su et al. suggests an upper bound of
the intermediate production of syngas.2 As a result of ethylene 28% for the C2 yield (combined C2H4 and C2H6 yield),
being normally produced through naphtha or ethane cracking, considering a continuous single-pass process with CH4 and O2
which are very energy demanding and generate a signicant cofeed.12 This means that even with enhanced reaction
amount of carbon dioxide emissions, OCM also presents an performance the downstream separation task is far from trivial.
opportunity for emission reduction in the chemical industry.3 Hence, an integrated approach considering both the reactor
In 1982 Keller and Bhasin published the rst determination and the separation steps is needed in order to synthesize an
of active catalysts for OCM.4 To date, several catalysts have ecient OCM process. The OCM process structure commonly
been tested for OCM such as La2O3/CaO,5 Li/MgO,6 and consists of a reaction section, a compression section, in which
MnNa2WO4/SiO2.7 Also, dierent reactor concepts, such as the reactor outlet is cooled, compressed, and the condensed
membrane reactors8 and udized bed reactors,9 as well as
dierent feeding policies have been studied experimentally and Received: January 11, 2016
through simulations10 and optimizations.11 The high temper- Revised: May 31, 2016
ature required for the OCM reaction, which is typically above Accepted: June 14, 2016
973 K, also leads to the oxidation of the educt methane (CH4) Published: June 14, 2016

2016 American Chemical Society 7473 DOI: 10.1021/acs.iecr.5b04910

Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

Figure 1. Block diagram of the OCM process.

Figure 2. Process ow diagram for an exemplary hybrid CO2 removal process employing a single membrane and absorption (process conguration
b).

water is removed, a carbon dioxide removal section, and nally (MDEA) and piperazine (PZ). Using a mixture of 37 wt %
a distillation section, consisting of two cryogenic distillation MDEA and 3 wt % PZ, the energy consumption is reduced to
columns to remove the light components and to separate the 3.7 MJ/kgCO2.14 The synthesis of new energy-ecient uids for
ethylene from the ethane. The classical OCM process structure CO2 absorption is currently an active research eld; however,
is reproduced in the block diagram in Figure 1. Besides the industrial applicability is still limited due to the fact that these
presented structure, other congurations are currently under uids are not yet produced in large scale and at commercial
investigation. Notably, the use of CO2 as diluent gas instead of prices.
N2 tends to avoid the formation of more CO2 in the reactor, Gas separation membranes can alternatively be employed for
thus enhancing the C2 yield. Furthermore, CO2 helps to handle CO2 separation. Glassy polymeric membranes, such as cellulose
the reactions exothermicity, avoiding the formation of acetate and polyimides, can be applied for natural gas
hotspots. This is of special interest when combining OCM sweetening, especially for small to medium capacities (6000
with the dry reforming of methane,13 given that the latter could 50 000 N m3/h) or even for larger capacities in remote or
use both the heat and the carbon dioxide produced in the oshore applications15 and for biogas upgrading.16,17 Applic-
OCM reactor. ability for ue gas treatment is limited because separation in
The carbon dioxide removal is one of the most critical steps. membranes is driven by the pressure dierential. This is not a
Specially, if CO2 dilution is used, then this step accounts for a drawback for the OCM process because compression is already
reasonable share of the ethylene production cost. The CO2 required to carry out the cryogenic distillation step. Gas
removal is commonly achieved by reactive absorption with separation membranes oer many advantages such as
amines (process conguration a) as shown in the right side straightforward installation, small spatial footprint, quick start-
(absorption section) of Figure 2. In the absorption process, a up, simple operation, and reduced environmental impact.15
signicant amount of heat is required in order to regenerate the Scale-up is also relatively easy because it happens mainly in
amine solution in a stripping column. Experiments carried out terms of adding more modules operating in parallel. This is
on mini-plant scale at Technische Universtitat Berlin showed however an economic disadvantage if compared to traditional
that by employing the benchmark absorber monoethanolamine separation techniques, such as distillation and absorption,
(MEA) 30 wt % aqueous solution, as much as 5 MJ per which tend to scale up very well. A further concern for the
kilogram of removed CO2 is required.14 This can be reduced by OCM process is the ethylene loss because it is a high-value
the use of other amines, such as N-methyl-diethanolamine product. Current membrane technology does not oer a
7474 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

signicant CO2 selectivity toward C2H4, so a standalone Hence, a proper balance between the energy savings and the
membrane-based separation would hardly make any sense. The product loss is missing to eectively assess whether the hybrid
combination of membrane and absorption in a hybrid process membrane-absorption process is of advantage. In this
could, however, exploit the advantages of both techniques and contribution, the feasibility of applying the hybrid process
provide a cost-ecient solution for OCM. An exemplary industrially is investigated through rigorous simulations and
process ow diagram for a hybrid CO2 separation system using preliminary economic evaluation taking into consideration the
a one-stage membrane separation and absorption/desorption eect of product loss. Dierent feed scenarios are considered so
(process conguration b) is presented in Figure 2. as to reect alternative OCM reactor concepts to produce 100
A hybrid process concept employing a single at sheet kt per year of ethylene. Dierent process congurations for the
envelope polyimide membrane module and absorption (such as CO2 removal step are thoroughly analyzed aiming for the
in process conguration b) has been investigated on mini-plant removal of 97% of the inlet CO2.

scale at Technische Universitat Berlin, showing a further
reduction in the energy requirement down to 2.8 MJ/kgCO2.14 MODELING
However, for capturing 90% of the inlet CO2 amount, around The overall model is divided into a membrane section and a
13% of the inlet C2H4 is lost. This is hardly acceptable for absorption section. The model for the membrane module is
industrial purposes. A two-stage stripping membrane cascade, initially implemented in the free modeling environment
in which the retentate outow is fed to a second membrane MOSAIC.19 MOSAIC is a convenient equation oriented
module as shown in Figure 3 (process conguration d) has modeling, simulation, and optimization environment, providing
been proposed to reduce this loss.18 The system is optimized in a LATEX-based equation input for creating a model. The
order to minimize the energy applied for the CO2 removal, model is thus written together with its description at a
which led to an energy requirement of 2.4 MJ/kgCO2, but documentation level, and a problem specication can be
ethylene loss is still above 8%. When imposing an upper bound created by selecting the input variables. Equation systems can
to the ethylene loss, the required energy is again increased. be solved internally using the available BzzMath numerical
library.20 MOSAICs code generation tool can also automati-
cally generate code for the solution in dierent programming
languages and environments, such as Matlab, Scilab, gPROMs,
Modelica, C++, Python, AMPL, GAMS, or even a user-dened
language. A code for the solution of the membrane model in
Aspen Custom Modeler (ACM) is generated, and the ACM
model is subsequently exported as a unit operation into Aspen
Plus for owsheet simulations. The absorption section is
modeled in Aspen Plus using the thermodynamic and unit
operation models contained therein and using an example le
provided in the softwares library as a starting point. The full
absorption/desorption model is detailed and validated against
experimental data elsewhere.21
Absorption Section. Thermodynamics. A nonsymmetric
electrolyte model already implemented in Aspen Plus, namely,
ENRTL-RK, is applied for the liquid phase. The set of
electrolyte equilibrium reactions is introduced, and the
equilibrium constants are calculated by minimizing the systems
Gibbs free energy. A PC-SAFT equation of state is used for the
vapor phase because it is the default in the example le used.
Parameters for MEA and water are regressed or retrieved by
Aspen Technology21 from various sources, and the other
parameters are taken from the softwares databank. The
RedlichKwong equation of state is tested as a numerically
cheaper alternative providing similar results. The binary
parameters for Henry constants implemented in the example
le are kept. Additionally, parameters for the Henry constant of
ethylene in MEA are estimated on the basis of data from
Carroll and Mather22 in order not to underestimate ethylene
loss in the absorption process.
Reactions. The full set of liquid-phase reactions can be seen
below. Minimization of the Gibbs free energy is used for the
equilibrium reactions (eqs 13) and kinetic parameters from
Pinsent et al.23 and Hikita et al.24 are used for the kinetically
controlled reactions (eqs 47).

Figure 3. Process ow diagrams for the membrane sections of the MEAH+ + H 2O MEA + H3O+ (1)
dierent process congurations: (c) single membrane with permeate
recycle, (d) two-stage stripping cascade, and (e) two-stage rectication
cascade. 2H 2O H3O+ + OH (2)

7475 DOI: 10.1021/acs.iecr.5b04910

Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

HCO3 + H 2O CO32 + H3O+ (3) carbons, and the second one is a poly(ethylene oxide)
poly(butylene terephtalate) thin lm composite membrane
CO2 + OH HCO3 (4) (PEO/PolyActive), which oers lower CO2 selectivity but
higher permeances. The membrane materials and modules are
HCO3 CO2 + OH (5) designed and produced by co-workers at the Helmholtz-
Zentrum Geesthacht Center for Materials and Coastal
MEA + CO2 + H 2O MEACOO + H3O+ (6) Research, Germany.17
A 1D solution-diusion model is applied on the basis of
MEACOO + H3O+ MEA + CO2 + H 2O (7) balance and ux equations.31 The PengRobinson equation of
Mass and Heat Transfer. The absorption and stripping state32 is applied on the retentate side, whereas ideal gas is
columns are designed through rate-based simulations using the assumed for the low pressure permeate side. The ow is
built-in correlations in Aspen Plus. Sulzer structured packings assumed to be isothermal on both sides with a 5 K cooling on
Mellapak 350X and 350Y are selected as the internals of the the permeate side to account for the JouleThomson eect.
absorption and stripping columns, respectively, as they are Negligible concentration polarization and pressure drop are
recommended by the vendor and commonly employed for CO2 observed experimentally, so these eects are not considered in
capture. The study of the performance of dierent packing the model. An Arrhenius-type equation is used to calculate the
types is outside of the scope of this contribution. Plug ow is permances in the PI membrane, whereas the free-volume-
assumed for the vapor, whereas the liquid is assumed to be well- theory equation is used to calculate the permeances in the PEO
mixed. Liquid and vapor lms are nonideal with reactions membrane. The parameters for these equations are obtained by
taking place in the liquid phase. Correlations by Bravo et al. Brinkmann et al.17 The resulting dierential algebraic equation
(1985)25 and Hanley and Chen (2012)26 are applied for the system is discretized using orthogonal collocation on nite
mass transfer coecients and interfacial area. The Chilton and elements with third-order Lagrangian polynomials on Radau
Colburn method is used for the heat transfer coecient.27 roots. The number of nite elements for the simulation and
Packing hold-ups are calculated as proposed by Bravo et al. optimization runs is xed to 10, which provides reasonable
(1992)28 Pressure drop is calculated according to vendors computation time and results identical to simulations carried
correlation. These correlations have been previously assessed in out with 20 nite elements. Basic model equations such as the
literature showing good agreement with experimental data for component mole balance in the permeate side (eq 8) and
the given application.29,30 Both authors highlight the retentate side (eq 9) and the permeance calculations for the PI
importance of tting parameters such as the interfacial area membrane (eq 10) and PEO membrane (eq 11) are presented.
factor, the heat transfer factor, and the liquid and vapor mass Further details on the implementation, such as numerical
transfer coecient factors to precisely mimic plant performance reformulations on the PengRobinson equation of state for
and behavior. The diameters of the absorber (4 cm) and the improved numerical stability, are given by Esche et al.18
stripper (10 cm) available for experiments are rather small, dniP
which greatly increases wall eects that are negligible in = Li(fiR fiP )h
industrial-sized equipment. The purpose of this work is to dz (8)
provide basic engineering design to allow for a preliminary cost
estimation on industrial scale, so no attempt to match dniR
= Li(fiR fiP )h
experimental data from the mini-plant is made, and these dz (9)
factors are left to unity.
Membrane Section. A at sheet envelope type membrane E PI
LiPI = Li0,PI exp R
a, i
module is considered for this application. It consists of a
cylindrical metal casting containing round membrane envelopes RT (10)
placed in series around an inner tube for the permeate outlet.
The retentate ows through the shell side, being directed by the E PEO
LiPEO = Li0,PEO exp
a, i
ow diverters. The module is schematically represented in R
+ i2
Figure 4. The membrane envelopes are produced in two RT
dierent versions. The rst one has a polyimide active layer
NC
fiR + fiP
(PI/Matrimid), oering high CO2 selectivity toward hydro- 12 mi0 exp(miT T R )
i=1 i 2 (11)
The model has been validated using data from mini-plant
experiments showing a good agreement in pressures up to 20
bar, but the accuracy decreases for higher pressures.33 The
model is later applied for the optimal design of the two-stage
stripping cascades.18 The model is herein further extended in
MOSAIC to allow for integration into Aspen Plus simulations.
For the model to be able to simulate industrial ow rates, a
scale-up by numbering-up approach is applied. This consists of
dividing the inlet ow by the design ow rate for the membrane
module to obtain the necessary number of parallel modules. A
single module is then calculated, and the resulting ows for the
Figure 4. Schematic representation of the at sheet envelope type permeate and retentate streams are multiplied to the number of
membrane module. modules to obtain the total outlet ows. An overall energy
7476 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

balance is formulated by introducing ACM enthalpy function Table 2. Utility and Ethylene Loss Costs
calls in MOSAIC. Ports are introduced to allow for the model
description cost unit
to be connected to streams in Aspen. MOSAICs code
generation is used to automatically generate a code in ACM steam 2.2 109 USD/J
modeling language, which is compiled in Aspen Custom cooling water 2.12 1010 USD/J
Modeler and exported as a custom unit operation into Aspen electricity 2.15 108 USD/J
Plus. ethylene loss 1.00 USD/kgC2H4
The compressors are modeled as isentropic with an eciency
of 0.72 and two stages with interstage and outlet coolers down applicable for a wide range of ethylene prices without the need
to 308 K. The intermediate pressure is the geometric average for estimating the costs for the other process steps that are
between the suction and discharge pressures.

outside of the scope of this contribution.

Absorption Section. An initial analysis is performed in
SIMULATION AND OPTIMIZATION order to determine the best operating pressure for the
The models are solved using the available numerical methods absorption column. The upstream water removal/compression
inside Aspen Plus v8.6. The Sparse solver is applied for the section is typically operated at 10 bar, and the downstream
membrane modules. The Broyden method is applied for distillation section is operated at 32 bar.35 These pressures are,
converging tear streams, providing signicantly better con- therefore taken as the lower and upper bounds to be considered
vergence than the default Wegstein method. The SQP method for the absorption section. The question is whether to perform
is applied for sequential optimization in the membrane section. the compression from 10 to 32 bar upstream or downstream of
No proper variable translation could be achieved when the absorber column. A higher pressure is benecial, for it
switching from sequential mode into equation-oriented mode, facilitates the CO2 absorption. This reduces both the necessary
so simultaneous optimization could not be performed. Optimal liquid to gas ratio (L/G), which is the ratio between the total
design of the absorption section is thus carried out only in the mass ow rate of amine solution and the total mass ow rate of
form of extensive simulation studies and sensitivity analyses. gas, and the reboiler duty. Therefore, the pressure of 32 bar has
The simulation and design of the CO2 capture processes are been previously adopted for the operation of the absorption
performed for an OCM plant producing 100 kt of ethylene per section.14 By operating the absorber at a higher pressure,
year in the reactor. The overall CO2 capture target of 97% is however, more ethylene is lost. This behavior is demonstrated
adopted for both the standalone absorption and the hybrid in Figure 5 for feed scenario II.
process congurations. To reect the dierent reactor design
and operating policies currently under investigation, four feed
scenarios with dierent compositions are considered in this
study. It is assumed that the reactor outlet passes through a
compression and cooling section, where all the water is
removed and the gas enters the CO2 capture section
compressed up to 10 bar and cooled down to 308 K. The
dierent compositions are presented in Table 1. Composition I

Table 1. Feed Gas Composition Scenarios

composition (mol %)
feed scenario CO2 C2H4 N2 CH4
I 11.0 6.0 33.0 50.0
II 9.0 9.0 16.7 65.3
III 24.5 4.5 8.0 63.0
IV 38.0 5.0 10.0 47.0
Figure 5. Reboiler duty (black diamonds) and ethylene loss (white
emulates the outlet of an OCM reactor using N2 dilution. squares) for dierent absorption operating pressures.
Composition II mimics a reactor using a really high CH4 to O2
ratio instead. Finally, compositions III and IV consider the use To properly assess the merit of operating the absorber at a
of CO2 dilution in the reactor. higher pressure, the total operating cost, comprising the utility
Utilities are introduced to the simulation in order to cost rate and ethylene loss cost rate, for dierent pressures
determine their consumption and cost rates. The default utility within this range is calculated. It is assumed that the gas is
costs from Aspen Plus are used because they reect typical received at 10 bar, and a rst compressor can raise the pressure
chemical plant values. Additionally, a value is attributed to the up to an intermediate value for the operation of the absorber.
ethylene loss based on current ethylene market prices.34 These The outlet of the absorber is then further compressed up to 32
values are displayed in Table 2. Adopting the ethylene market bar by a second compressor. Figure 6 exhibits the result of this
price for the ethylene loss is a conservative approach. At this analysis for feed scenario II with the total operating cost and its
stage of the process, the ethylene price is certainly lower most relevant components: steam for the reboiler, electricity for
because it still needs to be further processed to achieve the compressors, and the ethylene loss. Electricity for the
chemical or polymer grade. The membranes are less selective to pumps and the cooling water are accounted for but are left out
ethylene than the absorption, so this assumption tends to favor of the plot for sake of clarity. Despite the fact that the reboiler
the process conguration with standalone absorption. It duty is lower at higher operating pressures, the ethylene loss
ensures, however, that the feasible scenarios are industrially and the total compression duty are higher, resulting in a higher
7477 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

The next step is to achieve the desired CO2 capture target of

97%. This can be done with the use of a design specication,
which varies the amine solution ow rate in order to achieve a
specied CO2 capture. A sensitivity analysis is then performed
by changing the packed height and calculating the necessary
amine solution ow rate in each case. Increasing the absorber
packed height decreases the required amount of amine for a
given CO2 capture target, thus reducing the L/G ratio. The
sensitivity analysis results for feed scenario II can be seen in
Figure 7a. After arriving at 18 m of packed height, the reduction
in the L/G ratio is below 0.1%; therefore, this packed height is
selected for the absorber.

Figure 6. Operating cost for dierent absorber operating pressures.

Total operating cost rate (crosses), electricity for compression cost
rate (squares), steam cost rate (diamonds), and ethylene loss cost rate
(triangles).

operating cost. The ethylene loss cost rate is the variable most
aected by the operating pressure. The main cost driver is the
compression, and performing the compression in the down-
stream of the absorber is preferred for the total gas ow rate to
be compressed is smaller given that the CO2 is removed. This is
even more critical for the CO2 diluted scenarios, as the
reduction in the total gas ow rate is more signicant. Thus, the
pressure of 10 bar is selected for the absorption column. By
analyzing Figure 6, it also possible to picture that there may be
a minimum total cost which lies below the lower bound of 10
bar imposed by the upstream water removal section. This shall
be further investigated when synthesizing the full OCM
process.
The workow developed to simulate the absorption section
is very similar to the one previously described by Agbonghae et
al.;36 however, the applications described therein comprise CO2
capture for natural-gas-red or coal-red power plants as well as
natural gas combined cycle power plants. In this contribution,
further details are given on the specic tools inside Aspen Plus
to provide a guideline on the simulation and optimal design of
such systems using the software. This task is described by
Agbonghae et al. as being ...a combination of science and art
based on experience....36
The owsheet starts with an equilibrium calculation of the
absorption column. A proper starting value for the amine Figure 7. Sensitivity analysis for the packed height of the (a) absorber
solution ow rate is specied based on previously reported L/G and (b) stripper. Mass transfer correlations by Bravo et al. (1985)25
ratios to achieve convergence. The inlet lean amine solution is (white squares) and by Hanley and Chen (2012)26 (black diamonds).
preloaded with approximately 0.1 molCO2/molMEA because this
will be a recycle stream from the stripping column once the Once the absorber design is nished, the ash, the rich-amine
owsheet is complete. This strategy helps to close the recycle pump, and the leanrich exchanger are simulated. The pressure
later. The absorber is operated at 10 bar, and pressure drop is in the ash is nearly atmospheric to allow for a reasonable
neglected at this point. With the column initialized in amount of CO2 to be removed from the rich-amine stream. The
equilibrium mode, a packing type is selected to run a packing rich pump increases the pressure to 2.5 bar. The leanrich
sizing, which calculates the column diameter for a given exchanger is initially simulated as a heater, given that no recycle
fractional capacity. The default fractional capacity of 0.62 is stream is yet available. The procedure to design the stripping
used, and a foaming factor of 0.85 is applied. The calculation column is similar to that of the absorber; however,
mode is then set to rate-based, and a packing rating is specications for the partial condenser and for the reboiler
performed by applying the previously mentioned correlations are necessary. The stripper is operated at 2.5 bar. The
and specifying the packing type and height and the calculated previously reported required duty of around 4 MJ/kgCO214 is
diameter. Once the column has converged in rate-based mode, employed to provide reasonable starting values for the reboiler
a rst estimation of the packing hold-ups is obtained. Then, the duty, whereas the distillate ow is set to be large enough
reaction kinetics can be introduced in the specication sheet. compared to the inlet CO2 and H2O ows. Once the stripper is
7478 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

converged in rate-based mode including the reaction kinetics, LEAN) or a lesser amount of a leaner amine (lower LEAN).
design specications are introduced to x a lean-amine loading Using a leaner amine solution is desired in the absorber, for its
(LEAN) of 0.1 molCO2/molMEA at the bottom by varying the increased CO2 absorption capacity reduces the L/G ratio.
reboiler duty and the CO2 mass fraction at the top at 0.9 by However, more heat must be introduced in the stripping to
varying the distillate ow rate. A sensitivity analysis is again regenerate it. Thus, using a higher LEAN can decrease the
performed by changing the stripper packed height and reboiler duty up to a certain point. However, if LEAN is too
calculating the necessary reboiler duty on each case. The result high, then more amine is recirculated through the system,
of this sensitivity analysis for feed scenario II is presented in which causes additional heating, cooling, and pumping costs
Figure 7b. The reduction in the reboiler duty by adding 1 m of and increases equipment size. To nd the optimal operating
packing drops below 0.1% for a packed height of 12 m; point, the specication for the lean-amine solution is varied
therefore, this packed height is selected for the stripping from 0.1 to 0.2 molCO2/molMEA, and the utility costs for each
column. case are reported. The sensitivity analysis results for feed
Figure 7 also presents a comparison between the two scenario II are presented in Figure 8. The minimum utility
employed mass transfer correlations in both the absorption and
stripping columns. While they dier in the lower range of
packed heights, the results are identical in the actual operating
range of both columns. In this case, both correlations provide
identical results, and the one by Hanley and Chen26 is preferred
because convergence is achieved more quickly and easily when
changing the packed height during the sensitivity analyses.
The calculated pressure drop per unit height in the 2.7 m
diameter absorber and 2.9 m diameter stripper are lower than
the value of 2 mbar/m informed by the vendor and that
typically recommended in literature37 as the maximum allowed
pressure drop per unit height. The pressure drop are 1.2 mbar/
m in the absorber and 0.4 mbar/m in the stripper. This
indicates that the default fractional capacity of 0.62 adopted by
the software is rather low for this type of packing, and a higher
fractional capacity, between 0.7 and 0.8, could be applied
resulting in smaller column diameters. Figure 8. Sensitivity analysis for the operating point (lean loading and
The strippers bottom stream is pumped up to 10 bar in the L/G ratio). Utility cost rate (black diamonds) and L/G ratio (white
lean pump and is now available for closing the energy recycle in squares).
the leanrich exchanger. The heater block is replaced by a heat
exchanger block, which preheats the rich-amine stream by
cooling the lean-amine stream. The dierence between the hot consumption can be achieved at around 0.14 molCO2/molMEA,
stream outlet temperature and the cold stream inlet temper- and the respective L/G ratio in the absorber is 2.12 kg/kg. The
ature is set to 10 K. The cooled amine solution enters a mixer/ specic reboiler duty in this case is 3.85 MJ/kgCO2. This
tank, in which a makeup stream is added to replace the water
and amine losses in the gaseous streams. The makeup is methodology is repeated to all four feed scenarios because the
standalone absorption is to be used as reference to be
calculated by a balance block. The outlet of the mixer is cooled
compared with the hybrid processes. The total utility cost
down to 308 K and is recycled back to the absorber by the use
divided by the amount of captured CO2 is calculated as the
of a transfer block, which copies this stream into the inlet lean
specic absorption cost and assumed to be constant for each
amine stream, virtually closing the recycle. The design
scenario during the optimization of the upstream membrane
specication xes the LEAN in the bottom of the stripper to
section.
the same value as the lean amine entering the absorber, so these
Membrane Section. The design of the membrane section
streams are nearly identical. Therefore, only few iterations are
is achieved through sequential optimization. This section is
necessary to converge the owsheet once the transfer block is
simulated in a separate owsheet and dierent process
activated and the recycle closed. Attempts to close the recycle congurations are considered. The standalone absorption
by actually connecting the streams are useless and would hardly (process conguration a) is taken as the base case for the
permit any change in the operating point in order to perform comparison. For the hybrid processes, the following cong-
any further analysis of the system. Thus, the presented strategy urations are considered: (b) single membrane unit and
greatly improves the stability and usability of the model. absorption as in Figure 2, (c) single membrane unit with
A change in operating point can be easily achieved by the permeate recycle and absorption, (d) two-stage stripping
following steps. The transfer block is deactivated, leading the cascade and absorption, and (e) two-stage rectication cascade
amine recycle stream to be purged. The design variables are and absorption. The owsheets for the membrane sections of
changed. The simulation is run once and then once again after process congurations ce are shown in Figure 3.
reactivating the transfer block. This procedure is used to The optimization problem can be stated as the minimization
conduct a sensitivity analysis for the lean-amine loading. The of an objective function, consisting of a simplied operational
rich-amine loading at the bottom of the absorber (RICH) is cost, subject to the model equations described in the Modeling
bound by the amine absorption capacity to around 0.5 molCO2/ section. The simplied operational cost consists of the utility
molMEA. It is possible to achieve the same CO2 capture duty by cost and the cost applied to the ethylene lost in the outlet
recirculating a higher amount of a more loaded amine (higher streams, as given in Table 2, as well as the cost to remove the
7479 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

remaining CO2 on the retentate stream through absorption (to IR(IR + 1)N )
reach the total of 97%) at the xed cost per kg of CO2 Total Cost = OPEX + CAPEX
(1 + IR)N 1) (12)
calculated in the base case (process conguration a). The

optimization variables in the membrane module are the
membrane areas and the permeate pressure. In this version of
the software, one single optimization variable can be selected RESULTS AND DISCUSSION
inside a custom unit operation for optimization. Therefore, the All simulated owsheets are able to achieve the CO2 removal
membrane area is selected as the optimization variable, and the target of 97%. None of the hybrid process congurations
optimization is run repeatedly for dierent manually introduced present an advantage over the standalone absorption for feed
values of the permeate pressure. Splitters are added to allow for scenarios I and II. For these scenarios, the optimizer bypasses
the membrane modules to be bypassed and for streams to be the membranes, brings the membrane areas to the lower bound,
purged. These are not shown in the process ow diagrams for or runs into local solutions worse than the base case. In these
sake of clarity. The split ratios are also selected as optimization cases, the ethylene loss is costlier than the energy savings
variables. achieved by introducing an upstream membrane process. The
By increasing the membrane areas, more CO2 can be low CO2 concentration in the gas streams does not allow for
captured, thus reducing the necessary amine ow rate and the sucient driving force for an eective separation through gas
reboiler duty in the absorption section. However, by increasing permeation. When considering gas ows diluted with N2 or
the membrane areas, more ethylene is lost and utility costs, CH4 in the OCM reaction step, standalone absorption oers a
such as electricity for compression, are also added in process better performance for the downstream CO2 capture step. Most
congurations ce. This optimization intends to size the membranes for carbon dioxide separation, including the ones
membrane section for each process conguration by providing considered in this publication, have been developed and applied
a simple balance (simplied operational cost) between the costs for natural gas sweetening and/or biogas upgrading,17 to which
on each section. For the owsheets presenting a simplied product loss is not as critical given the lower value of the
operational cost lower than the operational cost of the base case produced pipeline quality natural gas. The PEO membrane has
(process conguration a), a rigorous simulation and design of a CO2 selectivity toward C2H4 of around 3 at room
the downstream absorption section is carried out as previously temperature, which is found to be insucient for the given
described in order to obtain a detailed operational cost and to industrial application. No economically feasible conguration
design the equipment to allow for the equipment cost applying the PEO membrane is found. For the hybrid process
estimation. to become interesting with these feed scenarios, higher
Economic Analysis. The operational cost (OPEX) is performance membranes, in terms of permeances and CO2
calculated as the sum of the cost rates for the utilities and for selectivity toward C2H4, are required.
the ethylene loss. This is enough to allow for a comparison For the CO2 diluted gas streams (feed scenarios III and IV),
between the dierent process congurations studied and to the higher inlet CO2 concentration leads to a more ecient
separation in the membranes. Furthermore, the higher CO2
analyze the CO2 capture step of the OCM process individually.
amount to be removed requires a higher amine ow rate in the
An yearly operational factor of 0.95 is assumed, resulting in
standalone absorption, which increases the ethylene loss in this
8322 operating hours per year.
section. By applying the PI membrane in both single and
The capital investment cost (CAPEX) is calculated as the
rectication cascade congurations (b and e, respectively), the
sum of all the installed equipment cost. The cost of equipment
operational cost for the CO2 separation can be reduced. For
is estimated using Aspen Economic Analyzer, which allows for congurations c and d, the optimizer led the membrane
the on-the-y sizing and cost estimation of equipment based on modules to be bypassed, brought the membrane areas to their
an up-to-date data bank. The absorber and stripper are lower bounds, or the simplied operational cost is still higher
structured packed columns as previously described. The than that of the base case a, so these congurations are left out
centrifugal pumps are installed with a reserve in parallel. The of the detailed economic analysis. The single membrane
leanrich exchanger, the amine solution cooler, the condenser, conguration b oers the simplest alternative for the hybrid
and the gas coolers are classic TEMA BEM type shell and tubes process because only the installation of the membrane modules
exchangers, whereas the reboiler is a kettle-type TEMA BKU is necessary. For the rectication cascade e, the permeate gas
exchanger. The mixer/tank and the condenser accumulator are compressor and the coolers are also required. However, the
horizontal pressure vessels. The compressor is a multistage rectication cascade allows for the recovery of plenty of the
centrifugal compressor. Costing for the membrane modules is ethylene permeating through the rst membrane module.
carried out separately based on the pressure vessel, given that Hence, it allows for increased membrane areas, drastically
this is the major cost driver for the module. To keep reducing the removal duty of the absorption section. For the
calculations conservative, no degression factor for the serial case with the highest inlet CO2 concentration (feed scenario
production of the modules is applied, and an installation cost IV) the rectication cascade removes more than 60% of the
factor of 1.2 is used. The base material for the equipment in the inlet CO2 while losing only 2% of the inlet C2H4. Besides that,
absorption section is stainless steel AISI 304, whereas carbon the permeate of the second membrane module is a nearly pure
steel is adopted in the membrane section. carbon dioxide stream that can be recycled directly as diluent to
The total cost of each process conguration is calculated as the reactor or employed in another process without any further
the sum of the operational cost (OPEX) with the annualized processing. The main simulation results are presented in Table
capital investment cost (CAPEX) as given in eq 12. The IR is 3. The presented values are the membrane areas per module
the interest rate of 15%, and N is the operational life of the and the CO2 capture and C2H4 loss in each section for the
plant, which is assumed to be 20 years. feasible combinations of feed scenario and process cong-
7480 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

Table 3. Main Simulation Results for the Feasible Process current ethylene market price and that the CO2 capture step
Congurations represents a large share in the nal ethylene production cost in
these scenarios.

membrane section absorption section
feed scenario CO2 C2H4 CO2 C2H4
and membrane capture loss capture loss CONCLUSIONS
conguration areas (m2) (%) (%) (%) (%)
A model has been successfully developed in order to simulate
IIIa 97.0 1.5 an absorption and membrane based hybrid process for
IIIb 1.0 17.9 1.3 79.1 1.3 removing carbon dioxide from an outlet gas stream of the
IIIe 3.6 and 6.2 38.9 0.8 58.1 1.0 oxidative coupling of methane reactor. The workow allows for
IVa 97.0 2.3 the documentation-based modeling in MOSAIC with the
IVb 2.5 39.5 3.7 57.5 1.6 subsequent integration of models into Aspen owsheets. This is
IVe 5.5 and 7.2 63.5 2.0 33.5 1.1 of particular interest for the implementation of customized unit
operation models for novel process equipment. While
uration. The optimal permeate pressures are 1.1 bar for these MOSAIC is a convenient modeling platform, the custom unit
cases. operation model, in this case a gas separation membrane
From the capital investment point of view, the owsheet for module, can be later integrated into a process owsheet in
the hybrid process tends to be costlier because it has more combination with well-established unit operations for the
equipment. However, the membrane section takes up part of process synthesis and cost estimation steps.
the CO2 removal duty, so the equipment in the absorption The hybrid CO2 removal process for OCM is thoroughly
section is smaller for the hybrid process. Notably, by adding the analyzed, and its economical feasibility is assessed taking into
membrane section, the diameters of the columns can be consideration the product (ethylene) loss. Dierent cong-
reduced because the total gas ow rate is lower. The membrane urations of the hybrid process are compared with the state-of-
modules are relatively cheap compared to the absorption and the-art technology, which is the stand-alone absorption process
stripping columns. The construction materials that can be
with a 30 wt % monoethanolamine (MEA) aqueous solution. It
applied in the membrane section are also cheaper than those in
is demonstrated that ethylene loss plays a major role in the
the absorption section. Consequently, a trade-o between
economic feasibility of the CO2 capture step of the OCM
adding more equipment and reducing the size of current
process, because ethylene is a high-value product.
expensive equipment is observed. For feed scenario III, the
For OCM reaction concepts employing N2 or CH4 dilution,
equipment cost of both hybrid processes are slightly higher
the hybrid CO2 removal process is hardly of any economic
than that of the standalone absorption. For feed scenario IV,
advantage compared to the standalone absorption. While
however, the equipment cost for both hybrid processes are
energy savings could be achieved, the product loss does not
lower. As expected, the CAPEX for the rectication cascade is
compensate for it. The use of the PEO membrane for the given
higher than the single membrane module, given that the
compressor and the coolers are necessary, but because this application is also found to be infeasible, given its low CO2
conguration reduces even further the removal duty of the selectivity toward C2H4.
absorption section, the actual increase in CAPEX is not that However, when considering carbon dioxide diluted ows in
high. the OCM reactor, the hybrid process is economically attractive.
The calculated operational cost, the capital investment cost, The concept of combining OCM with other CO2 intensive and
and the total cost are presented in Table 4. The process specially endothermic processes, such as the dry reforming of
methane, has a lot of potential and shall be further investigated.
Gas permeation can bring a signicant advantage for the CO2
Table 4. Economic Analysis Results
removal step in this case. For feed scenario III, the best
scenario and OPEX CAPEX total cost conguration is a single PI membrane, whereas for feed
conguration (kUSD/year) (kUSD) (kUSD/year) scenario IV, a two-stage PI membrane rectication cascade is
IIIa 8142 15653 10643 the most economical one.
IIIb 7696 17024 10416 Outlook. It is important to model and simulate in detail
IIIe 7518 18706 10507 each step of the OCM process individually. This helps to
IVa 12122 23922 15944 enhance the understanding and provide input for the design of
IVb 11262 18542 14224 the integrated owsheet. However, for an optimal process
IVe 9726 19764 12884 synthesis, the whole process superstructure has to be
considered simultaneously. To achieve that, surrogate models
that are more suitable for optimization have to be applied. It is
conguration b (with a single PI membrane) oers the lowest necessary to include economic terms, notably, the operational
total cost when considering the feed scenario III. However, the cost, to the objective function to ensure that the solutions are
dierence in the total cost is relatively low and most likely economically feasible. Furthermore, the environmental impact
much lower than the margin of error contained in the cost of the OCM process must also be assessed within the process
estimates. For feed scenario IV, the cheapest process synthesis stage. Alternative separation techniques, such as
conguration is the rectication cascade e. The dierence in pressure and temperature swing adsorption, are still to be
this case is more signicant, showing that the gas permeation is investigated. Finally, the OCM is also to be combined with
certainly a cost-eective alternative when considering highly other chemical processes, such as ethane cracking, considering
CO2 diluted gas streams for the OCM process. The annualized that many shale gas compositions are also rich in ethane, dry
total cost for these CO2 removal steps are 104 USD/tC2H4 and reforming of methane because it could use the heat and the
128 USD/tC2H4. These costs are relatively low considering the carbon dioxide produced in the OCM reactor, and methanol
7481 DOI: 10.1021/acs.iecr.5b04910
Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

production given that it could make use of the H2 from the (7) Langfeld, K.; Frank, B.; Strempel, V. E.; Berger-Karin, C.;
reformer and the available CO2. Weinberg, G.; Kondratenko, E. V.; Schomacker, R. Comparison of

oxidizing agents for the oxidative coupling of methane over state-of-

AUTHOR INFORMATION the-art catalysts. Appl. Catal., A 2012, 417418, 145152.
(8) Godini, H. R.; Trivedi, H.; de Villasante, A. G.; Gorke, O.; Jaso,
Corresponding Author S.; Simon, U.; Berthold, A.; Witt, W.; Wozny, G. Design and
*E-mail: [email protected], demonstration of an experimental membrane reactor set-up for
[email protected]. Phone: +49 30 314 29515. oxidative coupling of methane. Chem. Eng. Res. Des. 2013, 91, 2671
Notes 2681.
(9) Sadjadi, S.; Simon, U.; Godini, H.; Gorke, O.; Schomacker, R.;
The authors declare no competing nancial interest.

Wozny, G. Reactor material and gas dilution effects on the

performance of miniplant-scale fluidized-bed reactors for oxidative
ACKNOWLEDGMENTS coupling of methane. Chem. Eng. J. 2015, 281, 678687.
A.P. gratefully acknowledges the support from CAPES, (10) Kao, Y. K.; Lei, L.; Lin, Y. S. A Comparative Simulation Study
Coordination for the Improvement of Higher Education on Oxidative Coupling of Methane in Fixed-Bed and Membrane
PersonnelBrazil (grant number: BEX 11946/13-0). Financial Reactors. Ind. Eng. Chem. Res. 1997, 36, 35833593.
support from the Cluster of Excellence Unifying Concepts in (11) Esche, E.; Arellano-Garcia, H.; Wozny, G.; Biegler, L. Optimal
Catalysis by the German Research Foundation (DFG EXC Operation of a Membrane Reactor Network. Comput.-Aided Chem.
Eng. 2012, 31, 13211325.
314) is also gratefully acknowledged.

(12) Su, Y. S.; Ying, J. Y.; Green, W. H., Jr. Upper bound on the yield
for oxidative coupling of methane. J. Catal. 2003, 218, 321333.
NOMENCLATURE (13) Godini, H. R.; Xiao, S.; Kim, M.; Gorke, O.; Song, S.; Wozny, G.
Dual-membrane reactor for methane oxidative coupling and dry
Base Names
methane reforming: Reactor integration and process intensification.
Lennard-Jones molecular diameter [] Chem. Eng. Process. 2013, 74, 153164.
CAPEX Capital cost/expenditure [U.S. dollars (USD)] (14) Stunkel, S.; Illmer, D.; Drescher, A.; Schomacker, R.; Wozny, G.
Ea Activation energy [kJ/kmol] On the design, development and operation of an energy efficient CO2
f Fugacity [Pa] removal for the oxidative coupling of methane in a miniplant scale.
h Memnbrane height [m] Appl. Therm. Eng. 2012, 43, 141147.
IR Yearly interest rate [%] (15) Bernardo, P.; Drioli, E.; Golemme, G. Membrane Gas
L Permeance [kmol/(m2 s Pa)] Separation: A Review/State of the Art. Ind. Eng. Chem. Res. 2009,
m0 Parameter in free-volume theory [Pa1] 48, 46384663.
mT Parameter in free-volume theory [K1] (16) Scholz, M.; Frank, B.; Stockmeier, F.; Fal, S.; Wessling, M.
N Number of operating years [years] Techno-economic Analysis of Hybrid Processes for Biogas Upgrading.
n Molar ow [kmol/s] Ind. Eng. Chem. Res. 2013, 52, 1692916938.
OPEX Operational cost/expenditure [USD/year] (17) Brinkmann, T.; Naderipour, C.; Pohlmann, J.; Wind, J.; Wolff,
T.; Esche, E.; Muller, D.; Wozny, G.; Hoting, B. Pilot scale
R Universal gas constant [kJ/(kmol K)]
investigations of the removal of carbon dioxide from hydrocarbon
z Membrane length [m] gas streams using poly(ethylene oxide)-poly(butylene terephthalate)-
Subscripts PolyActiveTM) thin film composite membranes. J. Membr. Sci. 2015,
i Chemical component i 489, 237247.
Superscripts (18) Esche, E.; Muller, D.; Song, S.; Wozny, G. Optimization during
0 the process synthesis: enabling the oxidative coupling of methane by
Reference condition minimizing the energy required for the carbon dioxide removal. J.
P Permeate side Cleaner Prod. 2015, 91, 100108.
PEO Poly(ethylene oxide) membrane (19) Kuntsche, S.; Barz, T.; Kraus, R.; Arellano-Garcia, H.; Wozny,
PI Polyimide membrane G. MOSAIC a web-based modeling environment for code generation.
R Retentate side Comput. Chem. Eng. 2011, 35, 22572273.

REFERENCES
(1) True, W. R. Global ethylene capacity continues advance in 2011.
(20) Buzzi-Ferraris, G.; Manenti, F. BzzMath: Library Overview and
Recent Advances in Numerical Methods. In Computer Aided Chemical
Engineering; Proceedings of the 22nd European Symposium on
Oil. Gas. J. 2012, 110, 7. Computer Aided Process Engineering; Lockhart Bogle, I. D.,
(2) Ashcroft, A. T.; Cheetham, A. K.; Foord, J. S.; Green, M. L. H.; Fairweather, M., Eds.; Elsevier: Amsterdam, The Netherlands, 2012;
Grey, C. P.; Murrell, A. J.; Vernon, P. D. F. Selective oxidation of Vol. 22, pp 13121316.
methane to synthesis gas using transition metal catalysts. Nature 1990, (21) Rate-Based Model of the CO2 Capture Process by MEA using
344, 319321. Aspen Plus; Aspen Technology: Bedford, MA, 2014.
(3) Gielen, D.; Moriguchi, Y.; Yagita, H. CO2 emission reduction for (22) Carroll, J. J.; Mather, A. E. A model for the solubility of light
Japanese petrochemicals. J. Cleaner Prod. 2002, 10, 589604. hydrocarbons in water and aqueous solutions of alkanolamines. Chem.
(4) Keller, G.; Bhasin, M. Synthesis of ethylene via oxidative coupling Eng. Sci. 1997, 52, 545552.
of methane: I. Determination of active catalysts. J. Catal. 1982, 73, 9 (23) Pinsent, B.; Pearson, L.; Roughton, F. The Kinetics of
19. Combination of Carbon Dioxide with Hydroxide Ions. Trans. Faraday
(5) Stansch, Z.; Mleczko, L.; Baerns, M. Comprehensive Kinetics of Soc. 1956, 52, 15121520.
Oxidative Coupling of Methane over the La2O3/CaO Catalyst. Ind. (24) Hikita, H.; Asai, S.; Ishikawa, H.; Honda, M. The kinetics of
Eng. Chem. Res. 1997, 36, 25682579. reactions of carbon dioxide with monoethanolamine, diethanolamine
(6) Simon, U.; Arndt, S.; Otremba, T.; Schlingmann, T.; Gorke, O.; and triethanolamine by a rapid mixing method. Chem. Eng. J. 1977, 13,
Dinse, K.-P.; Schomacker, R.; Schubert, H. Li/MgO with spin sensors 712.
as catalyst for the oxidative coupling of methane. Catal. Commun. (25) Bravo, J.; Rocha, J.; Fair, J. Mass transfer in gauze packings.
2012, 18, 132136. Hydrocarbon Process. 1985, 64, 9195.

7482 DOI: 10.1021/acs.iecr.5b04910

Ind. Eng. Chem. Res. 2016, 55, 74737483
Industrial & Engineering Chemistry Research Article

(26) Hanley, B.; Chen, C.-C. New Mass-Transfer Correlations for

Packed Towers. AIChE J. 2012, 58, 132152.
(27) R.Taylor, R. K. Multicomponent Mass Transfer; Wiley: New York,
1993.
(28) Bravo, J.; Rocha, J.; Fair, J. A comprehensive model for the
performance of columns containing structured packings. Inst. Chem.
Eng. Symp. Ser. 1992, 32, 641.
(29) Afkhamipour, M.; Mofarahi, M. Sensitivity analysis of the rate-
based CO2 absorber model using amine solutions (MEA, MDEA and
AMP) in packed columns. Int. J. Greenhouse Gas Control 2014, 25, 9
22.
(30) Razi, N.; Svendsen, H. F.; Bolland, O. Assessment of mass
transfer correlations in rate-based modelingof a large-scale CO2
capture with MEA. Int. J. Greenhouse Gas Control 2014, 26, 93108.
(31) Ohlrogge, K., Ebert, K., Eds. Membranen: Grundlagen, Verfahren
und industrielle Anwendungen; WILEY-VCH Verlag GmbH &
Co.KGaA: Weinheim, 2006.
(32) Peng, D.-Y.; Robinson, D. B. A New Two-Constant Equation of
State. Ind. Eng. Chem. Fundam. 1976, 15, 5964.
(33) Song, S.; Esche, E.; Stunkel, S.; Brinkmann, T.; Wind, J.;
Shishatskiy, S.; Wozny, G. Energy, Equipment and Cost Savings by
Using aMembrane Unit in an Amine-Based AbsorptionProcess for
CO2 Removal. Chem. Ing. Tech. 2013, 85, 12211227.
(34) Platts McGraw-Hill Financial. Platts Global Ethylene Price Index;
http://www.platts.com/news-feature/2014/petrochemicals/pgpi/
ethylene (accessed on July 7, 2015).
(35) Salerno-Paredes, D. Optimal Synthesis of Downstream
Processes using the Oxidative Coupling of Methane Reaction. Ph.D.
Thesis, Technische Universitat Berlin, Technische Universitat Berlin,
Germany, 2013.
(36) Agbonghae, E. O.; Hughes, K. J.; Ingham, D. B.; Ma, L.;
Pourkashanian, M. Optimal Process Design of Commercial-Scale
Amine-Based CO2 Capture Plants. Ind. Eng. Chem. Res. 2014, 53,
1481514829.
(37) Kister, Z. H. Distillation Design; McGraw-Hill: New York, 1992.

7483 DOI: 10.1021/acs.iecr.5b04910

Ind. Eng. Chem. Res. 2016, 55, 74737483