Fakouribaygi2015 PDF

CHERD-1650; No.
of Pages 11
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Contents lists available at ScienceDirect
Chemical Engineering Research and Design
journal homepage: www.elsevier.com/locate/cherd
Application of the perturbed chain-SAFT equation

of state for modeling CO2 solubility in aqueous
monoethanolamine solutions
Sadjad Fakouri Baygi, Hassan Pahlavanzadeh ∗

Faculty of Chemical Engineering, Tarbiat Modares University, Tehran, Iran
a b s t r a c t
CO2 removal by treatment of acid gases by aqueous alkanolamines is a very significant operation from industrial
and environmental point of view. To attain a comprehensive thermodynamic model of the CO2 –MEA–H2 O in a wide
range of temperature and CO2 partial pressures, Perturbed Chain-Statistical Associating Fluid Theory (PC-SAFT) EOS is
applied to predict the absorption of carbon dioxide by MEA (MonoEthanolAmine). In order to find the best association
scheme for MEA in PC-SAFT EOS, three pure parameter sets for MEA in the 2B, 3B and 4C association schemes are
determined in temperature range 303.15–443.15 K. Temperature independent binary interaction parameters have
been adjusted in the VLE calculation for three schemes of MEA and two schemes of water. Binary VLE calculations
show the 3B scheme for MEA and the 4C scheme for water indicate the best prediction in the MEA–H2 O system. Excess
enthalpy data for aqueous MEA are predicted by kij , which has been adjusted in VLE calculations. The 3B scheme for
MEA and the 4C scheme for water also are used to find CO2 solubility in the ternary system of CO2 –MEA–H2 O system.
Ideal Smith–Missen algorithm has been applied to find the concentration of all species in chemical equilibrium.
Results show the 3B association scheme for MEA and the 4C association scheme for water in PC-SAFT EOS have
better agreement with binary and ternary experimental data. PC-SAFT EOS is able to anticipate the CO2 solubility
in the CO2 –MEA–H2 O system without any regression in the ternary system. The CO2 solubility in ternary system
is compared to e-NRTL as a common thermodynamics model. The average absolute partial pressure deviations for
PC-SAFT and e-NRTL are calculated around 36% and 42%, respectively.
© 2014 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers.
Keywords: PC-SAFT EOS; Association schemes; Monoethanolamine; Independent binary interaction; Smith-Missen
Algorithm; CO2 capture
1. Introduction and associating systems was a primary consideration in its

selection in this study to model CO2 solubility in aqueous
Aqueous alkanolamines play a vital role in the removal of MEA solutions. On the other hand, there are some practical
acid gases in many industries such as natural gas treatment issues about application of association term in PC-SAFT EOS,
and CO2 capture from flue gases. Consequently, a rigorous because the performance of the PC-SAFT EOS depends seri-
and predictive thermodynamics model of CO2 absorption is ously on the selection of an association scheme due to its
essential to model these types of systems. PC-SAFT EOS, effect on parameter estimation. Different association schemes
developed by Gross and Sadowski (2001, 2002), is one of the result in different parameter sets that yield similar pure
most prevalent SAFT EOS versions, providing an applicable component vapor pressure and liquid density. However, dif-
and reliable thermodynamics framework to model thermo- ferent mixture results, such as heat of mixing and binary VLE
dynamics properties of many systems, especially asymmetric may be obtained. In addition, this issue is critical regarding
systems. The ability of PC-SAFT EOS to predict asymmetric alkanolamine molecules, because these molecules consist of
∗
Corresponding author. Tel.: +98 21 82883312; fax: +98 21 82883381.
E-mail address: pahlavzh@modares.ac.ir (H. Pahlavanzadeh).
http://dx.doi.org/10.1016/j.cherd.2014.07.017
0263-8762/© 2014 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers.
Please cite this article in press as: Fakouri Baygi, S., Pahlavanzadeh, H., Application of the perturbed chain-SAFT equation of state for modeling
CO2 solubility in aqueous monoethanolamine solutions. Chem. Eng. Res. Des. (2014), http://dx.doi.org/10.1016/j.cherd.2014.07.017
CHERD-1650; No. of Pages 11
ARTICLE IN PRESS
2 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Nomenclature
ai activity of species i
ãres residual Helmholtz energy
ãhc hard-chain Helmholtz energy
ãdisp dispersion interactions Helmholtz energy
ãassoc association interactions Helmholtz energy
d average segment diameter
ghs
ij
radial pair distribution function for segments of
component i in the hard sphere system
h enthalpy
kB Boltzmann constant (1.38066 × 10−23 J.K−1 )
Kx The chemical equilibrium constants in mole
fraction based
Km The chemical equilibrium constants in molality Fig. 1 – Schematic diagram of the model.
based
m pure number of segments in a molecule
and extended UNIQUAC by Faramarzi et al. (2009). Common
Mt total moles of MEA
features shared by most of the gE thermodynamics models
np number of experimental data points
that are applied to model CO2 –alkanolamine–water systems
x mole fraction in liquid phase
include a large number of complicated adjustable binary inter-
y mole fraction in gas phase
action parameters that must be correlated to multicomponent
XAi fraction of molecules of species i that are not
system experimental data. On the contrary, the models that
bound at site A
use the equations of states, particularly association EOSs,
p pressure (Pa)
do not require this amount of experimental data, except for
T temperature (K)
modifying the interaction energies. However, these modifi-
W total moles of water in liquid phase
cations do not require multicomponent (ternary, quaternary,
Greek symbols etc.) system experimental data. Some association models have
˛ CO2 liquid loading (mole CO2 /mole MEA) recently been applied in order to model these systems, such as
ε segment dispersion interaction energy CPA for CO2 –MDEA–H2 O system by Zoghi et al. (2012), SAFT-
εAi Bj energy of hydrogen bonding between site A at VR for CO2 –MEA–H2 O system and by Rodriguez et al. (2012)
molecule i and site B at molecule j and Mac Dowell et al. (2009) and PC-SAFT by Nasrifar and
Ai Bj volume of hydrogen bonding between site A at Tafazzol (2010). Nasrifar and Tafazzol (2010) applied PC-SAFT
molecule i and site B at molecule j EOS to describe solubility of CO2 and H2 S in aqueous solu-
molar density tions of MDEA, DEA, and MEA, but their model involves very
n molecular density (number of molecules/Å3 ) large errors for systems involve CO2 especially CO2 –MEA–H2 O
pure temperature independent segment diam- system.
eter (Å) Limitations in work of Nasrifar and Tafazzol (2010) are the
i stoichiometric coefficient of component i in motivating factors for this study. In this study, our previous
reaction work (Pahlavanzadeh and Fakouri Baygi, 2013) on thermody-
Ai Bj strength of hydrogen bonding between the site namics modeling of CO2 –MDEA–H2 O system by PC-SAFT EOS
A at molecule i and the site B at molecule j is extended to the CO2 –MDEA–H2 O system. PC-SAFT EOS was
employed in this research, similarly to our earlier work, to find
Subscripts and superscripts CO2 physical absorption, and an ideal Smith–Missen method
exp experimental was used to find chemical species concentration. A schematic
cal calculated diagram of the model is presented in Fig. 1.
E excess
L liquid 2. PC-SAFT EOS
min minimum
max maximum Gross and Sadowski (2001, 2002) introduced PC-SAFT EOS in
i, j, k indices terms of free Helmholtz energy.
sat saturation
ãres = ãhc + ãdisp + ãassoc (1)
distinctive functional groups (hydroxyl and amine groups). where ãres , ãhc , ãdisp , and ãassoc represent residual Helmholtz
Therefore, several association schemes should be checked in energy, the contribution of the hard-chain reference fluid
binary systems to find the suitable schemes (Avlund, 2011; consists of spherical segments Helmholtz energy, the contri-
Pahlavanzadeh and Fakouri Baygi, 2013). bution of dispersive attractions to the Helmholtz energy, and
In the past few decades, some gE thermodynamics mod- the contribution due to short-range association interactions
els have been implemented for correlation and prediction of (hydrogen bonding) Helmholtz energy, respectively. Gross and
CO2 solubility in aqueous MEA solutions and other aqueous Sadowski (2001) presented the contribution of the hard-chain
alkanolamines solutions, such as electrolyte NRTL by Zhang reference fluid Helmholtz energy and the contribution of dis-
et al. (2011), Austgen et al. (1989) and Posey and Rochelle (1997) persive attractions to the Helmholtz energy. The contribution
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 3
of the association Helmholtz energy is given by the following

Table 1 – Correlations constants for the saturated vapor
expression (Gross and Sadowski, 2002): pressure (psat ) and saturated liquid density (L,sat ) of MEA
in Eqs. (9) and (10) (from Avlund et al. (2008)).
1 Ai 1

ãassoc = xi ln XAi − X + (2) Constants psat (Pa) L,sat (mol L−1 )
2 2
i Ai A 92.624 1.0011
B −10,367 0.22523
where xi is mole fraction of molecule i, and the XAi is a fraction C −9.4699 678.2
D 1.9 × 10−18 0.21515
of molecules of species i that are not bound at site A. XAi is
E 6 0
expressed by following equation (Gross and Sadowski, 2002):
Tmin (K) 283.65
⎛ ⎞−1 Tmax (K) 678.2
Maximum deviation 8.1% 0.3%
XAi = ⎝1 + n xj XBj Ai Bj ⎠ (3) Average deviation 2.1% 0.1%
j Bj
where n and Ai Bj are molecular density and strength of

3. Modeling results
hydrogen bonding between the site A at molecule i with the
site B at molecule j, respectively. Hydrogen bond strength is
expressed by Eq. (4) (Gross and Sadowski, 2002) 3.1. Pure parameters calculation

Alkanolamines consist of hydroxyl groups and one or more
Ai Bj εAi Bj
= d3ij ghs
ij
(dij )Ai Bj exp −1 (4) amine groups, so sometimes selecting a precise association
kB T
scheme is complicated. To address their distinctive functional
groups, a variety association schemes can be proposed.
where εAi Bj and Ai Bj are energy and volume of hydrogen bond- MEA is a primary alkanolamine consisting of one hydroxyl
ing between site A at molecule i and site B at molecule j, group and one primary amine group. There is some degree of
respectively. dij = (di + dj )/2 is average temperature dependent ambiguity regarding the participation of primary amine group
segment diameter, ghs ij
is radial pair distribution function for in association. Therefore, different association schemes are
segments of component i in the hard sphere system which has considered for MEA that do not specify the role of hydroxyl and
been presented by Gross and Sadowski (2001), kB is Boltzmann primary amine groups in association; rather, these associa-
constant and T represents temperature in Kelvin. tion schemes regard MEA as a molecule with some association
Temperature dependent segment diameter, di , is given by sites that they can associate other molecules sites with same
the following expression (Gross and Sadowski, 2001): strength. For example, Nasrifar and Tafazzol (2010) consid-
ered the 4C association scheme for MEA. On the other hand,
−3ε some association schemes specify the role of hydroxyl and
i
di = i 1 − 0.12 exp (5)
kB T primary amine groups in association; for example, Mac Dowell
et al. (2009) applied an association scheme that mediated three
where i and εi are pure temperature independent seg- sites for hydroxyl groups and three sites for amine groups for
ment diameter and segment dispersion interaction energy of SAFT-VR EOS.
molecule i respectively. In this study, parameter sets of MEA for PC-SAFT EOS are
Gross and Sadowski (2002) used two mixing rules for asso- determined for 2B, 3B and 4C schemes. In addition, a parame-
ciating multicomponent systems that are expressed in Eqs. (6) ter set for the 4C association scheme are determined with less
and (7): errors for saturated pressure and saturated density respect to
results of Nasrifar and Tafazzol (2010).
3 To calculate parameter sets for MEA, the following objective
√
√ i j function is applied
Ai Bj = Ai Bi Aj Bj
1
(6)
2 i + j
np sat,exp np sat,exp
p − psat,cal − isat,cal
i i i
OF = + (8)
εAi Bi + εAj Bj pi
sat,exp sat,exp
i
εAi Bj = (7) i i
2
Specific terminology is used in the study that was where psat and sat indicate saturated vapor pressure and satu-
introduced by Huang and Radosz (1990) to denote associa- rated liquid density of pure species, respectively. Experimental
tion structure. They have classified eight different association saturated vapor pressures and saturated liquid densities
schemes. In their terminology, different molecules were char- obtained from DIPPR correlations that Avlund et al. (2008) used
acterized by different schemes and several possibilities for for determining MEA parameters set for CPA EOS. In addition,
participation in association. The molecules with one proton Avlund et al. (2008) presented these correlations have reli-
donor site and one proton acceptor site correspond to the able agreement with experimental data. The correlations are
2B association scheme, the molecules with two proton donor expressed by Eqs. (9) and (10) for the vapor pressure (psat ) and
sites and one proton acceptor site and vice versa correspond saturated liquid density (L,sat ) of MEA, respectively.
to the 3B association scheme and the molecules with two pro-
B

ton donor sites and two proton acceptor sites correspond to psat (Pa) = exp A + + C ln (T/K) + D(T/K)
E
(9)
the 4C association scheme. (T/K)
ARTICLE IN PRESS
Table 2 – Pure component parameters used in this work.

Species Association m ε/kB AB εAB /kB AAD%a Ref.
scheme
psat L,sat
4C 4.5208 2.6574 237.6864 0.187533 989.8984 0.24 0.23 This work

MEA 3B 4.5354 2.6019 204.0438 0.118488 2383.4744 1.75 0.43 This work
2B 3.0353 3.0435 277.174 0.037470 2586.3 0.62 0.12 This work
2B 1.9599 2.362 279.42 0.1750 2059.28 1.18 3.92 Diamantonis and

H2 O
Economou (2011)
4C 2.1945 2.229 141.66 0.2039 1804.17 1.98 0.83 Diamantonis and
Economou (2011)
CO2 Non- 2.0729 2.7852 169.21 – – 2.78 2.73 Gross and Sadowski
associating (2001)
a
AAD% = 100/np × | exp
− cal
|/ exp
, is psat and L,sat .
mol A considered as an associating molecule. These parameter sets

L,sat = (10)
L D are presented in Table 2. Binary VLE of CO2 –H2 O and MEA–H2 O
B[1+(1−(T/K)/C)]
systems are calculated and compared by using two association
schemes for H2 O.
where the correlations constants are presented in Table 1.
Pure parameter sets for MEA with 2B, 3B and 4C association
schemes in temperature range 303.15–443.15 K are determined
and the results are presented in Table 2, and the diagrams of 450
saturated vapor pressure versus saturated liquid density and 440

saturated liquid density versus temperature of saturated pure 430
MEA are presented in Figs. 2 and 3, respectively.
420
So far, a number of association schemes and parame-
410
ter sets have been reported in the literature for water and
carbon dioxide. Diamantonis and Economou (2011) used sev- 400
eral saturation data for finding the accurate parameter sets 390
for water in the 2B and 4C association schemes. So in this
380
study, two association schemes for H2 O from Diamantonis
T/K
370
and Economou and one parameter set for CO2 , are compared
for selection of the best parameter sets for VLE calculations 360
of multicomponent system. Meanwhile, CO2 molecule is not 350
340
110
330
MEA (2B)
MEA (3B) MEA (2B)
100 320
MEA (4C) MEA (3B)
DIPPR Correlation 310 MEA (4C)
90 DIPPR Correlation
300
880 900 920 940 960 980 1000 1020
80 sat
ρ /kg.m
-3
70 Fig. 3 – Saturated liquid density versus temperature for

pure MEA at temperature from 303.15 K to 443.15 K.
60
Psat /kPa
Comparison of the 4C, 3B and 2B association schemes to

50 DIPPR correlations.
40
Table 3 – AAD% for MEA–H2 O system in different
30 association schemes.
Association schemes AAD%a kij
20
MEA H2 O xMEA yMEA
10
2B 2B 5.66 2.38 −0.0420
0 3B 4.24 2.08 −0.0146
880 900 920 940 960 980 1000 1020 4C 5.41 2.28 −0.0362
sat -3
ρ /kg.m
2B 4C 3.52 8.63 −0.1305
Fig. 2 – Saturated vapor pressure versus saturated liquid 3B 3.54 1.54 −0.0520
density for pure MEA at temperature from 303.15 K to 4C 8.05 3.60 −0.1245
443.15 K. Comparison of the 4C, 3B and 2B association a

AAD% = 100/np × | exp
− cal
|/ exp
, is x and y.
schemes to DIPPR correlations.
ARTICLE IN PRESS
Fig. 4 – Isobaric temperature–composition T–x slices of the vapor–liquid equilibrium of MEA–H2 O with three association
scheme for MEA and 2B association scheme for H2 O. The symbols correspond to the experimental data at P = 66.66 kPa and
P = 101.33 kPa from Cai et al. (1996) and Park and Lee (1997), and the dotted and solid curves represent the PC-SAFT
predictions (kij = 0) and correlations (kij from Table 3): (a) 2B scheme for MEA at P = 66.66 kPa, (b) 2B scheme for MEA at
P = 101.33 kPa, (c) 3B scheme for MEA at P = 66.66 kPa, (d) 3B scheme for MEA at P = 101.33 kPa, (e) 4C scheme for MEA at
P = 66.66 kPa, (f) 4C scheme for MEA at P = 101.33 kPa.
ARTICLE IN PRESS
Fig. 5 – Isobaric temperature–composition T–x slices of the vapor–liquid equilibrium of MEA–H2 O with three association
scheme for MEA and 4C association scheme for H2 O. The symbols correspond to the experimental data at P = 66.66 kPa and
P = 101.33 kPa from Cai et al. (1996) and Park and Lee (1997), and the dotted and solid curves represent the PC-SAFT
predictions (kij = 0) and correlations (kij from Table 3): (a) 2B scheme for MEA at P = 66.66 kPa, (b) 2B scheme for MEA at
P = 101.33 kPa, (c) 3B scheme for MEA at P = 66.66 kPa, (d) 3B scheme for MEA at P = 101.33 kPa, (e) 4C scheme for MEA at
P = 66.66 kPa, (f) 4C scheme for MEA at P = 101.33 kPa.
ARTICLE IN PRESS
Fig. 6 – Excess enthalpy of MEA–H2 O mixture at T = 298.15 K

and T = 342.15 K. Comparison of PC-SAFT (kij = −0.0520)
calculations to experimental data from Posey (1996) and
Touhara et al. (1982). kij is adjusted to vapor–liquid Fig. 7 – Isothermal pressure–composition P–x slices of the
equilibrium data (see Fig. 5(c) and (d)). Dash lines and solid vapor–liquid equilibrium of CO2 –H2 O. The symbols
lines represent model predictions (kij = 0) and model correspond to the experimental data from Campos et al.
correlations (kij = −0.0520), respectively. (2009) at T = 298.2 K, T = 303.2 K, T = 313.2 K and T = 323.2 K
and the solid curves represent to the PC-SAFT EOS
3.2. Binary systems predictions (kij = 0).
Finding proper parameter sets and schemes is the princi- association scheme for MEA indicates the best agreement with
pal part of the modeling with association EOSs, because the experimental data for both schemes of water with respect to
pure parameter sets have to be adjusted with the saturation other schemes. Therefore, the 3B scheme for MEA is selected.
experimental data, and these parameter sets and association The binary VLE results with both the 2B and 4C association
schemes exhibit same results in pure systems. Therefore, the schemes for water, and 2B, 3B and 4C association schemes for
best association schemes can be found in binary systems. MEA, are presented in Figs 4 and 5, respectively.
Binary interaction parameters are adjusted for binary sys- Experimental mixing enthalpies at 298.15 K and 342.15 K
tems to obtain accurate VLE calculations. The conventional from Posey (1996) and Touhara et al. (1982) are predicted with
Berthelot–Lorentz combining rule is expressed by Eq. (11) is
applied in this study.

εij = εi εj (1 − kij ) (11)
where kij is binary interaction parameter (Gross and Sadowski,

2001).
3.2.1. MEA–H2 O system

Binary VLE between MEA and water are shown in Figs. 4 and 5
with different association schemes for MEA and water, and
binary interaction parameters are adjusted with the following
objective function with experimental data from Cai et al. (1996)
np exp np exp
x − xcal y − ycal
i i i i
OF = exp + exp (12)
xi yi
i=1 i=1
In order to use this objective function both Bubble-T and

Dew-T calculations must be applied. This objective function
is used to adjust general kij even in small mole fractions in
both phases. Binary interaction parameters between MEA and
water are presented in Table 3. VLE results for experimental Fig. 8 – Comparison of the experimental data for species
data from Cai et al. (1996) at pressures 66.66 kPa and 101.33 kPa concentration in CO2 –MEA–H2 O and the prediction of ideal
are shown and compared in Figs. 4 and 5. In addition, exper- Smith–Missen algorithm at T = 293.15 K. MEA concentration
imental data from Park and Lee (1997) are predicted with is 30 wt.%. Symbols represent experimental data from
PC-SAFT EOS. The AAD% and kij from Table 3 show that the 3B Jakobsen et al. (2005).
ARTICLE IN PRESS
Table 4 – Equilibrium constants correlations used for modeling in Eq. (17).

Reaction Equilibrium constant A B C D Temperature range (K) Ref.
no.
a a
H3 O+ OH−
(R1) a
K1 = 132.899 −13445.9 22.4773 0 273.15–498.15 Edwards et al. (1978)
a2
H2 O
a a
H3 O+ HCO−
(R2) a
K2 = 2
3 231.465 −12092.10 −36.7816 0 273.15–498.15 Edwards et al. (1978)
a a
H2 O CO2
a a
H3 O+ CO−2
(R3) a
K3 = a 3 216.049 −12431.70 −35.4819 0 273.15–498.15 Edwards et al. (1978)
H2 O aHCO−
3
a a
HCO− MEA
(R4) a
K4 = a 3 −1.8652 −1545.3 0 0 293.15–323.15 Tong et al. (2012)
H2 O aMEACOO−
a a
H O+ MEA
(R5) K5 = a 3 a 2.1211 −8189.38 0 −0.007484 273.15–323.15 Bates and Pinching (1951)
H2 O MEAH+
a
This equilibrium constant was converted to mole fraction based.
the 3B scheme for MEA and the 4C scheme for water, and than 10−3 ) in Fig. 7. These small errors in the range of CO2
results are shown in Fig. 6. solubility are proper to model CO2 –MEA–H2 O system.
3.2.2. CO2 –H2 O system 3.3. Ternary systems (CO2 –MEA–H2 O)

Both 4C and 2B association schemes for water were applied
in calculating the VLE CO2 –H2 O mixture in our previous work The procedure of finding CO2 solubility in this system is
(Pahlavanzadeh and Fakouri Baygi, 2013), and the results show described in Fig. 1. This ternary system includes both chemical
that the 4C association scheme for water includes better equilibrium and multicomponent system phase equilibrium.
results respect with the 2B association scheme in binary VLE of The liquid phase consists of both molecular species and ionic
CO2 –H2 O system. In addition, the prediction of CO2 solubility species, and the gas phase consists of only molecular species.
in water at temperature from 298.2 K to 313.2 K is presented in In this work, no regression is applied for VLE prediction
Fig. 7, and is compared with experimental data from Campos of the CO2 –MEA–H2 O system by PC-SAFT EOS. This means
et al. (2009) As one can see, PC-SAFT EOS shows great predic- the binary interaction parameters (kij ) between (CO2 –H2 O)
tion for CO2 –H2 O system at very low CO2 mole fraction (less and (CO2 –MEA) have been set to zero, and the binary
Table 5 – Comparison between experimental data and model predictions for CO2 partial pressure of the CO2 –MEA–H2 O
system.
Source Temperature, CO2 partial MEA mole CO2 e-NRTL by Zhang PC-SAFT EOS
T (K) pressure, p (kPa) fraction loading, ˛ et al. (2011) by this work
np b AAD% np b AAD%
Jones et al. (1959) 313–413 0.073–918 0.05 0.076–0.728 48 32.2 50 26.17

Lee et al. (1974) 313–373 1.2–6616 0.05 0.139–1.19 45 44.4 45 36.51
Lawson and 313–413 1.3–2750 0.05–0.11 0.11–0.929 24 30.7 21 34.30
Garst (1976)
Lee et al. (1976) 298–393 0.1–10,000 0.02–0.11 0.065–2.152 256 50.5 253 34.76
Isaacs et al. 353–373 0.009–1.75 0.05 0.0368–0.315 19 112 16 22.66
(1980)
Austgen et al. 313–353 0.093–228 0.05 0.266–0.698 8 42.6 8 11.97
(1991)
Shen and Li 313 1.57–2550 0.05 0.561–1.049 13 35.5 13 30.92
(1992)
Dawodu and 373 455–3863 0.09 0.541–0.723 5 43.2 5 17.05
Meisen (1994)
Song et al. (1996) 312 3.1–2359 0.11 0.49–1.061 10 74.4 10 66.87
Jane and Li (1997) 353 3.57–121.8 0.05 0.363–0.58 7 37.7 7 30.91
Park et al. (1997) 313 3.5–2092 0.05 0.512–1.046 7 28.6 7 21.25
Mathonat et al. 313–393 2000–10000 0.11 0.55–1.07 9 36.0 9 35.62
(1998)
Park et al. (2002) 313 2.6–2189 0.03–0.05 0.478–1.068 13 49.8 13 50.65
Tong et al. (2012) 313–393 3.95–408.17 0.11 0.211–0.748 – – 21 53.53
a
Hilliard (2008) 313–333 0.005–50.2 0.06–0.16 0.114–0.591 55 35.5 42 34.97
a
Jou et al. (1995) 273–423 0.0012–19954 0.11 0.002–1.324 124 33.5 100 43.16
a
Ma’mun et al. 393 7.354–191.9 0.11 0.155–0.4182 19 13.5 19 21.03
(2005)
a
Xu and 373–443 12–1626 0.11 0.303–0.52 63 28.0 52 46.88
Rochelle (2011)
Overall 725 42.29 691 36.42
a
expin e-NRTL regression
used
These data sets are by Zhang et al. (2011).
b
AAD% = 100/np × pCO − pcal
CO
/pexp
CO
.
2 2 2
ARTICLE IN PRESS
Fig. 9 – Experimental data from Jones et al. (1959) in

wMEA = 15.3 wt.% at temperatures 313.15–413.15 K (symbols) Fig. 11 – Comparison between experimental data from Jou
and CO2 partial pressures prediction by PC-SAFT EOS (solid et al. (1995), Tong et al. (2012) and Ma’mun et al. (2005) in
lines) of CO2 –MEA–H2 O system. wMEA = 30 wt.% at temperatures 313 and 393 K (symbols)
and CO2 partial pressures prediction by PC-SAFT EOS (solid
lines) of CO2 –MEA–H2 O system.
interaction parameter between (MEA–H2 O) has been set to
−0.0520. Meanwhile, interaction energies between CO2 and CO2 + 2H2 O ↔ H3 O+ + HCO3 − (R2)
MEA were considered in this study (see Table 2 and Eq. (11)),
but no regression was applied to modify their interaction ener- HCO3 − + H2 O ↔ H3 O+ + CO3 2− (R3)
gies. In the other words, if kij is set to zero between molecules i
and j in PC-SAFT EOS, it does not mean that there is no interac- MEACOO− + H2 O ↔ HCO3 − + MEA (R4)
tion between molecules i and j; rather, it means that PC-SAFT
is able to predict interaction energies between molecules i and MEAH+ + H2 O ↔ H3 O+ + MEA (R5)
j without any correction.
The chemical reactions that take place in the liquid phase Charge balance is expressed as:
for CO2 –MEA–H2 O system can be expressed as:
[H3 O+ ] + [MEAH+ ] − [MEACOO− ] − [HCO− 2− −
3 ] − 2[CO3 ] − [OH ]
2H2 O ↔ H3 O+ + OH− (R1)
=0 (13)
Fig. 10 – Experimental data from Lee et al. (1976) in Fig. 12 – Comparison between experimental data from Park
mMEA = 3.75 at temperatures 313.15–413.15 K (symbols) and et al. (1997) and Shen and Li (1992) in wMEA = 15.3 wt.% at
CO2 partial pressures prediction by PC-SAFT EOS (solid T = 313.15 K (symbols) and CO2 partial pressures prediction
lines) of CO2 –MEA–H2 O system. by PC-SAFT EOS (solid lines) of CO2 –MEA–H2 O system.
ARTICLE IN PRESS
by Edwards et al. (1978), and then these constants are con-

verted to the mole fraction based on the following equation
(Austgen, 1989).
1

ln Kx = ln Km − ln i (18)
0.01802
i=
/ H2 O
where Kx , Km and are chemical equilibrium constant in mole

fraction based, chemical equilibrium constant in molality
based and stoichiometric coefficients in reaction, respectively.
Results of PC-SAFT EOS predictions compared to different
experimental data are presented in Table 5. The results of
the e-NRTL model as implemented by Zhang et al. (2011) for
the similar systems have been illustrated in Table 5 for com-
parison reasons. It is worth it to point out that although the
PC-SAFT EOS uses no regression in VLE calculation, the model
predictions are acceptable within the average relative devia-
tion in CO2 partial pressure for many experimental data are
shown in Table 5.
Fig. 13 – Comparison between experimental data from Fig. 9 shows that the prediction of the CO2 partial pres-
Austgen et al. (1991) in mMEA = 2.5 kmol m−3 at sure in 15.3 wt.% and at wide range of temperatures between
temperatures 313.15 and 353.15 K (symbols) and CO2 partial 313.15 and 413.15 K, with experimental data from Jones et al.
pressures prediction by PC-SAFT EOS (solid lines) of (1959) is reliable. Fig. 10 exhibits that PC-SAFT is able to antici-
CO2 –MEA–H2 O system. pate the experimental CO2 partial pressure data from Lee et al.
(1976) in molality 3.75 MEA and at temperatures from 313.15 to
393.15 K acceptably. Model predictions and experimental data
Three mass balances are also expressed as: from Jou et al. (1995), Tong et al. (2012) and Ma’mun et al.
(2005) in 30 wt.% MEA at two temperatures 313 K and 393 K
are compared in Fig. 11. One can see from Fig. 11 that PC-
Mt = [MEA] + [MEAH+ ] + [MEACOO− ] (14)
SAFT EOS has acceptable prediction with experimental data at
these temperatures. There are also some deviations between
˛Mt = [CO2 ] − [HCO− 2− −
3 ] + [CO3 ] + [MEACOO ] (15)
experimental data at 393.15 K. While experimental data from
Tong et al. (2012) and Ma’mun et al. (2005) were reported in
W = [H2 O] + [HCO− 2− + −
3 ] + [CO3 ] + [H3 O ] + [OH ] (16)
similar MEA concentration and temprature ranges. System-
atic deviations can be observed at the high CO2 loading region
where ˛, Mt , and W are loading of CO2 in aqueous MEA, total
(approximately more than 0.9) under the different MEA con-
moles of MEA, and total amount of H2 O in liquid phase, respec-
centration and temperature ranges in Figs. 10–12. Comparison
tively.
of experimental data from Austgen et al. (1991) and PC-SAFT
Chemical reactions (R1)–(R5) and linear equations (13)–(16)
EOS shows good performance at the average loading region
are solved by the Lagrange method simultaneously, which
(0.266–0.698) in Fig. 13.
was introduced by Smith–Missen (Austgen, 1989). In order to
find the concentration of species in chemical equilibrium, it
is assumed their activity equal to their mole fraction (ai = xi ) 4. Conclusion
for all species, and the effects of the ion pairs are neglected
in equilibrium. It is necessary to say that this assumption In this work, PC-SAFT EOS is applied for the prediction of
(ai = xi ) has been applied in several similar works (Nasrifar and a CO2 –MEA–H2 O system devoid of using any regression in
Tafazzol, 2010; Gabrielsen et al., 2005). To check these assumed the ternary system in temperature range 298–443 K and CO2
results with experimental data, the concentration of some partial pressure range 0.001–20,000 kPa. Because of the impor-
molecular and ionic species are compared with NMR exper- tance of association schemes in parameter estimation in
imental speciation data from Jakobsen et al. (2005) in Fig. 8. associated EOSs, several association schemes are checked for
The chemical equilibrium constants for reactions (R1)–(R5) MEA, and the results indicate that the 3B association scheme
are calculated by Eq. (17): for MEA shows more conformity with binary VLE experimen-
tal data more than other association scheme. Therefore, both
B the 3B association scheme for MEA and the 4C association
ln Kx = A + + C ln (T/K) + D(T/K) (17)
(T/K) scheme for water (due to its consistency in CO2 –H2 O system)
are applied. Binary interaction parameters between MEA and
where the constants of Eq. (17) for reactions (R1)–(R5) are pre- H2 O in different association schemes are adjusted to correct
sented in Table 4. VLE prediction of this binary system. In addition, excess
In this work, the constants of Edwards et al. (1978) are enthalpy of the MEA and the water system is predicted by the
used for reactions (R1)–(R3), and for achievement the best ideal binary interaction parameter that was adjusted in VLE cal-
chemical equilibrium (because of this assumption (ai = xi )) the culation. In summary, PC-SAFT EOS performs an acceptable
constants of Tong et al. (2012) and Bates and Pinching (1951) prediction in the ternary system of CO2 –MEA–H2 O in a wide
are used for reactions (R4) and (R5), respectively. Chemical range of temperatures and CO2 partial pressure without using
equilibrium for reaction (R4) was reported in molality based any regression in multicomponent system, and its average
ARTICLE IN PRESS
absolute partial pressure deviation is about 36.42% among Jakobsen, J.P., Krane, J., Svendsen, H.F., 2005. Ind. Eng. Chem. Res.
691 data points. 44, 9894–9903.
Jane, I.S., Li, M.-H., 1997. J. Chem. Eng. Data 42, 98–105.
Jones, J.H., Froning, H.R., Claytor, E.E., 1959. J. Chem. Eng. Data 4,
Acknowledgments
85–92.
Jou, F.-Y., Mather, A.E., Otto, F.D., 1995. Can. J. Chem. Eng. 73,
Authors are thankful to Hossein Eghbali for providing VLE 140–147.
data set of CO2 –MEA–H2 O system. Lawson, J.D., Garst, A.W., 1976. J. Chem. Eng. Data 21, 20–30.
Lee, J.I., Otto, F.D., Mather, A.E., 1974. Can. J. Chem. Eng. 52,
803–805.
References Lee, J.I., Otto, F.D., Mather, A.E., 1976. J. Appl. Chem. Biotechnol.
26, 541–549.
Austgen Jr., D., 1989. A Model of Vapor–Liquid Equilibria for Acid Mac Dowell, N., Llovell, F., Adjiman, C.S., Jackson, G., Galindo, A.,
Gas–Alkanolamine–Water Systems. Texas University, Austin, 2009. Ind. Eng. Chem. Res. 49, 1883–1899.
TX, USA. Ma’mun, S., Nilsen, R., Svendsen, H.F., Juliussen, O., 2005. J. Chem.
Austgen, D.M., Rochelle, G.T., Peng, X., Chen, C.C., 1989. Ind. Eng. Eng. Data 50, 630–634.
Chem. Res. 28, 1060–1073. Mathonat, C., Majer, V., Mather, A.E., Grolier, J.P.E., 1998. Ind. Eng.
Austgen, D.M., Rochelle, G.T., Chen, C.C., 1991. Ind. Eng. Chem. Chem. Res. 37, 4136–4141.
Res. 30, 543–555. Nasrifar, K., Tafazzol, A.H., 2010. Ind. Eng. Chem. Res. 49,
Avlund, A.S., 2011. Extension of Association Models to Complex 7620–7630.
Chemicals. Department of Chemical and Biochemical Pahlavanzadeh, H., Fakouri Baygi, S., 2013. J. Chem. Thermodyn.
Engineering, Technical University of Denmark. 59, 214–221.
Avlund, A.S., Kontogeorgis, G.M., Michelsen, M.L., 2008. Ind. Eng. Park, S.-B., Lee, H., 1997. Korean J. Chem. Eng. 14, 146–148.
Chem. Res. 47, 7441–7446. Park, S.-B., Shim, C.-S., Lee, H., Lee, K.-H., 1997. Fluid Phase
Bates, R.G., Pinching, G.D., 1951. J. Res. Natl. Bur. Stand. 46, 349352. Equilib. 134, 141–149.
Cai, Z., Xie, R., Wu, Z., 1996. J. Chem. Eng. Data 41, 1101–1103. Park, J.-Y., Yoon, S.J., Lee, H., Yoon, J.-H., Shim, J.-G., Lee, J.K., Min,
Campos, C.E.P.S., Villardi, H.G.D.A., Pessoa, F.L.P., Uller, A.M.C., B.-Y., Eum, H.-M., Kang, M.C., 2002. Fluid Phase Equilib. 202,
2009. J. Chem. Eng. Data 54, 2881–2886. 359–366.
Dawodu, O.F., Meisen, A., 1994. J. Chem. Eng. Data 39, 548–552. Posey, M.L., 1996. Thermodynamic Model for Acid Gas Loaded
Diamantonis, N.I., Economou, I.G., 2011. Energy Fuels 25, Aqueous Alkanolamine Solutions. University of Texas at
3334–3343. Austin.
Edwards, T.J., Maurer, G., Newman, J., Prausnitz, J.M., 1978. AIChE Posey, M.L., Rochelle, G.T., 1997. Ind. Eng. Chem. Res. 36,
J. 24, 966–976. 3944–3953.
Faramarzi, L., Kontogeorgis, G.M., Thomsen, K., Stenby, E.H., 2009. Rodriguez, J., Mac Dowell, N., Llovell, F., Adjiman, C.S., Jackson, G.,
Fluid Phase Equilib. 282, 121–132. Galindo, A., 2012. Mol. Phys. 110, 1325–1348.
Gabrielsen, J., Michelsen, M.L., Stenby, E.H., Kontogeorgis, G.M., Shen, K.P., Li, M.H., 1992. J. Chem. Eng. Data 37, 96–100.
2005. Ind. Eng. Chem. Res. 44, 3348–3354. Song, J.-H., Yoon, J.-H., Lee, H., Lee, K.-H., 1996. J. Chem. Eng. Data
Gross, J., Sadowski, G., 2001. Ind. Eng. Chem. Res. 40, 1244–1260. 41, 497–499.
Gross, J., Sadowski, G., 2002. Ind. Eng. Chem. Res. 41, Tong, D., Trusler, J.P.M., Maitland, G.C., Gibbins, J., Fennell, P.S.,
5510–5515. 2012. Int. J. Greenh. Gas Control 6, 37–47.
Hilliard, M.D., 2008. A Predictive Thermodynamic Model for an Touhara, H., Okazaki, S., Okino, F., Tanaka, H., Ikari, K., Nakanishi,
Aqueous Blend of Potassium Carbonate, Piperazine, and K., 1982. J. Chem. Thermodyn. 14, 145–156.
Monoethanolamine for Carbon Dioxide Capture from Flue Xu, Q., Rochelle, G., 2011. Energy Proc. 4, 117–124.
Gas. University of Texas at Austin. Zhang, Y., Que, H., Chen, C.-C., 2011. Fluid Phase Equilib. 311,
Huang, S.H., Radosz, M., 1990. Ind. Eng. Chem. Res. 29, 2284–2294. 67–75.
Isaacs, E.E., Otto, F.D., Mather, A.E., 1980. J. Chem. Eng. Data 25, Zoghi, A.T., Feyzi, F., Dehghani, M.R., 2012. Ind. Eng. Chem. Res.
118–120. 51, 9875–9885.

Fakouribaygi2015 PDF

Uploaded by

Copyright:

Available Formats

Fakouribaygi2015 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fakouribaygi2015 PDF

Uploaded by

Copyright:

Available Formats

CHERD-1650; No.

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

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

Application of the perturbed chain-SAFT equation

Sadjad Fakouri Baygi, Hassan Pahlavanzadeh ∗

1. Introduction and associating systems was a primary consideration in its

of the association Helmholtz energy is given by the following

where n and Ai Bj are molecular density and strength of

Table 2 – Pure component parameters used in this work.

4C 4.5208 2.6574 237.6864 0.187533 989.8984 0.24 0.23 This work

2B 1.9599 2.362 279.42 0.1750 2059.28 1.18 3.92 Diamantonis and

mol  A considered as an associating molecule. These parameter sets

saturated vapor pressure versus saturated liquid density and 440

70 Fig. 3 – Saturated liquid density versus temperature for

Comparison of the 4C, 3B and 2B association schemes to

443.15 K. Comparison of the 4C, 3B and 2B association a

Fig. 6 – Excess enthalpy of MEA–H2 O mixture at T = 298.15 K

where kij is binary interaction parameter (Gross and Sadowski,

3.2.1. MEA–H2 O system

In order to use this objective function both Bubble-T and

Table 4 – Equilibrium constants correlations used for modeling in Eq. (17).

3.2.2. CO2 –H2 O system 3.3. Ternary systems (CO2 –MEA–H2 O)

Jones et al. (1959) 313–413 0.073–918 0.05 0.076–0.728 48 32.2 50 26.17

Overall 725 42.29 691 36.42

Fig. 9 – Experimental data from Jones et al. (1959) in

by Edwards et al. (1978), and then these constants are con-

where Kx , Km and  are chemical equilibrium constant in mole

You might also like

where n and Ai Bj are molecular density and strength of

mol A considered as an associating molecule. These parameter sets

where Kx , Km and are chemical equilibrium constant in mole