Renewable and Sustainable Energy Reviews 50 (2015) 547566

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

Modeling of CO2 capture via chemical absorption processes

An extensive literature review
I.P. Koronaki n, L. Prentza, V. Papaefthimiou
Laboratory of Applied Thermodynamics, School of Mechanical Engineering, Thermal Engineering Section National Technical University of Athens, Heroon
Polytechniou 9, Zografou Campus 15780, Athens, Greece

art ic l e i nf o a b s t r a c t

Article history: Climate change mainly due to the release of greenhouse gases into the atmosphere is getting alarming
Received 17 October 2014 dimensions. CO2 capture from point source emissions is a promising solution, lately receiving signicant
Received in revised form attention. In particular, chemical absorption of CO2 from ue gases using aqueous solvents (mainly
11 March 2015
alkanolamines) is a well-known process, studied in detail. Modern research aims to optimize this
Accepted 23 April 2015
process, maximizing the absorption rates and minimizing the parasitic but not negligible energy
requirements for solvent regeneration. This type of analysis requires considering the coupling of the
Keywords: absorption with the power plant operation or other source of CO2. The operation uctuations and
Chemical absorption disturbances, such as load variations or start-up mode have to be reected in the process modeling,
CCS
justifying the emerging need for dynamic modeling. However, dynamic analysis is not always realizable
Rate-based model
as dynamic experimental data are scarce in order to enable accurate model validation. Thus, steady state
Dynamic modeling
CO2 models are still convenient for certain cases. The current work provides a short description of the main
modeling approaches followed and enlists representative steady state and dynamic models found in
literature. Finally, a primary comparison is performed for some comparable models that used the same
set of experimental data for model validation.
& 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
2. Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
3. Chemical absorption process and solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
4. Chemical absorption modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
4.1. Modeling requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
4.2. Modeling approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
4.3. Mass transfer approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
4.4. Thermodynamic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
4.4.1. Kent
Eisenberg model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
4.4.2. Electrolyte non-random two-liquid (e-NTRL) model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
4.4.3. EoS/GE models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 550
4.4.4. SAFT (statistical association uid theory) model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
4.5. Kinetic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
4.6. Absorption column model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
4.7. Dynamic modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
5. Models review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
5.1. Rational for this review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551

n
Corresponding author. Tel.: 30 2107721581; fax: 30 2107723670.
E-mail addresses: koronaki@central.ntua.gr (I.P. Koronaki), prentzaloukia@central.ntua.gr (L. Prentza), vpapaeft@central.ntua.gr (V. Papaefthimiou).

http://dx.doi.org/10.1016/j.rser.2015.04.124
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
548 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

5.2. Chemical absorption models literature review. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551

5.2.1. Steady state models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551
5.2.2. Dynamic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
5.2.3. Model comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
6. Conclusions and scope for future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Appendix A. List with recent models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565

1. Introduction
Facing climatic change as a detrimental result of the increasing
global energy demand, various policies have been adopted aiming to
mitigate the consequences, mainly to the reduction of greenhouse
emissions. During the last years, the penetration of renewable sources
of energy has been mainly promoted as a promising solution.
However, it seems that fossil fuels will still be a predominant
component of the global energy mix, posing the need for more direct
ways of response. At the same time, EU and its member states are
committed to an emission reduction target of 20% by 2020, compared
with 1990 levels. This target could be increased to 30% under
conditions set out by the European Council [1]
Carbon Capture and Storage (CCS) is a promising technique
gaining increasing interest among researchers and policymakers.
In specic, CCS is a process where CO2 is separated from industrial
production or energy conversion, is transported to the storage site
under high pressure and nally geologically stored, aiming to
realize zero emissions in the process of fossil energy extraction,
conversion and usage [2]. CCS refers to the capture of CO2
emissions either before the combustion of the fuel or directly
from the ue gas. Post combustion methods are more applicable as
they have no retrotting requirements for the existing equipment.
The most applied and efcient post combustion method is
chemical absorption. In comparison with other post-combustion
CO2 absorption processes, chemical absorbents ensure higher
absorption efciency and selectivity, and lower energy [3].
In order to enable a wider scale implementation of CCS, detailed
and efcient modeling of the process is a prerequisite. This is because
the study of reactive absorption is mainly based on simulations, as
experimental data are limited and not always reliable. Chemical
absorption of CO2 by aqueous solvents, mainly amines, has been
studied for decades by numerous researchers with satisfying results.
The developed models include both kinetic and thermodynamic
aspects and are aiming to optimize the absorption. It is only recently
though, that these attempts include dynamic characteristics and more
realistic representations of the process.
The current work objective is to outline briey the common
modeling approaches in the eld of post-combustion CO2 absorp-
tion and also attempt to list a wide spectrum of specic models,
steady and dynamic, encompassing their assumptions, modeling
and validations tools and main outcome. Finally, the paper aims to
conclude on the prediction efciency of several models, compar-
ing, to the point it is possible, results of different researchers
which are based on the same experimental data.
2. Structure
The article is structured in 5 Sections. Section 1 is the
introductory section providing information about the CO2 emis-
sions problem and the necessity to confront this. In following,
Section 2 outlines the chemical absorption process and certain
commercially used solvents. Section 3 provides the fundamentals,
assumptions and requirements of the reactive absorption
modeling. Different mass transfer models, thermodynamic and
kinetic models are listed. At the end an introduction to dynamic
modeling is included. Section 4 includes the analysis of represen-
tative models from literature. Different sub-sections for steady
state and dynamic models are encompassed, providing informa-
tion for the assumptions made, equations applied and main
outcome. In following in the same Section, three absorption units
are presented that have provided very useful and widely used
experimental results. Based on these data, certain rough compar-
isons are performed between models. Summarizing, Section 5
outlines the conclusions made by this literature research and
suggest elds for future work.
3. Chemical absorption process and solvents
Chemical absorption (or reactive absorption) is the process
where a gas is absorbed by a liquid phase with combination of
reaction and absorptive mass transport. Amine-based systems,
carbonate-based systems, aqueous ammonia and ionic liquids
based systems are typical systems for chemical absorption.
In specic for CO2 capture, chemical absorption involves the
reaction of CO2 with a chemical solvent forming a weakly bonded
intermediate compound, a process that can be reversed applying heat
and resulting to the original solvent and a CO2 stream [3]. The
regeneration process is the most demanding by means of energy, as it
may cost up to 14% of the power plant efciency [4]. During chemical
absorption the ue gas enters usually an absorption tower where it
contacts countercurrent the solvent under low temperature (40
60 1C). The rich in CO2 solvent is compressed, heated in an exchanger
and inserted in the desorption column (stripper) where it is regen-
erated under low pressure (close to ambient) and high temperatures
(100140 1C). The regenerated (lean) solvent exits the column and re-
enters the absorber after dissipating its heat in the rich solvent
through the above mentioned exchanger.
The chemical absorption process was initially performed using
amine aqueous solvents. The most mature and commercially
applied amine is Monoethanolamine (MEA) due to its high
absorptive capacity. MEA is considered an attractive solvent at
low partial pressures of CO2 in the ue gas due to its fast reaction
rates compared to that of secondary and tertiary amines [5]. Other
common amines are Diethanolamine (DEA), Methyl diethanola-
mine (MDEA), Diglycolamine (DGA), and Diisopropanolamine
(DIPA) and Piperazine (PZ) or mixtures of amines that are
characterized by satisfactory reaction rates and lower regeneration
energy than MEA. Piperazine is usually added as a promoter as it
has been found to have faster reaction rates than MEA. Due to its
high volatility though, its application in CO2 absorption is more
expensive and is still under development [6]. A special category of
amines are sterically hindered amines (SHA) that e.g. 2-Amino-2-
methyl-1-propanol (AMP). SHA form more unstable carbamates in
comparison with other amines. Thus, they are more efcient
during regeneration, as the unstable carbamates are easier to be
reversed, leading therefore to quicker desorption and less energy
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
requirements [7]. Alternative solvents for chemical absorption are
aqueous ammonia, ionic liquids and potassium carbonate. In
specic, aqueous ammonia is a strongly promising solvent char-
acterized by many assets, as high absorption efciency, high
absorption capacity, low energy requirement for absorbent regen-
eration and wide distribution of resources [8].
4. Chemical absorption modeling
Within this section the main modeling processses and tools are
provided. The required steps in order to simulate accurately the
process of CO2 absorption are described as well as the dominating
techniques.
4.1. Modeling requirements
Chemical absorption process involves complex mechanisms both of
thermodynamic and chemical nature. This process is described as
reactive absorption considering that reactions and absorptive mass
transport phenomena occur simultaneously. An important character-
istic of reactive absorption processes is the variation of reaction
velocities, from very low (slow reactions) up to innitely high
(instantaneous reactions).Thus, the modeling of these mechanisms is
challenging and requires a multi-factor design. The ideal model has to
be rigorous enough to encompass all complex aspects of the problem
and manageable enough in order to enable an easy implementation of
the modeling process [9].
A thermodynamic model has to describe through its differential
equations certain fundamentals: the chemical equilibrium, the
vapour
liquid equilibrium, the liquid phase constitution, the
regeneration energy, and the two phases mass [10].
4.2. Modeling approaches
There are two modeling approaches corresponding reactive
processes: the equilibrium models and the rate-based models (or
non-equilibrium models). The equilibrium models assume
vapour
liquid equilibrium at each stage [11]. Non-equilibrium
models consider the actual reaction rates and are more accurate as
physical equilibrium is hard to attain.
Equilibrium models assume a quick and cursive mass transfer.
Thus, they are more applicable for solvents with high reaction rates,
e.g. primary amines. Equilibrium models apply MESH (Material
balances, Equilibrium equations, Mole Fractions Sum conditions and
Heat balance) equations. These models are usually based on theore-
tical stages in which liquid and vapour phases reach equilibrium.
Stages use height equivalent to a theoretical plate (HETP) and assume
the equilibrium state between the streams leaving each stage [12].
These stages correspond either to real column stages, for disc columns,
or equivalent stages, for columns with ller materials.
Non-equilibrium models, as mentioned, estimate the real
reaction rates and mass transfer resistances. These phenomena
are often approached via enhancement factors that amend the
equilibrium computations. The enhancement factor is actually the
chemical mass transfer coefcient to the physical mass transfer
coefcient. It is either obtained by tting experimental data or
theoretically based on simplied model assumptions [13]. This is a
very common approach of modeling reaction kinetics. However,
this approach is often simplied as binary systems are not suitable
to estimate the enhancement factors and theoretical approaches
are based on rough simplications. More complex simulations are
instructed that consider directly the reaction rates.
Rate based models describe the interfacial molar uxes directly.
The mass transfer is usually described through Maxwell
Stefan
equations from the kinetic theory of gases modied for binary
549
systems with gas and liquids [9]. The modied equation giving the
generalized driving force is the following:
m xlf nlf
xlf nlf
X i j j i
di
j1 C lft Dij
where D is the Maxwell
Stefan diffusion coefcient, xlfi and xlfj are
the liquid mole fractions in the liquid lm for i and j components,
nlfj and nlfi are the molar uxes in the liquid lm for i and j
components and C lft is the total molar concentration in the
liquid lm.
The electrolytes effect should be included in this analysis and
for dilute electrolyte systems, the diffusional interactions can be
neglected giving the generalized Maxwell
Stefan equations
reduced to the following Nernst
Planck equation:
 
F
nlfi C lft Dli;ef f xlfi xlfi zi xlfi nlfm
RT
where Dli;ef f is the effective diffusion coefcient, zi the ionic charge
of component i, F Faradays constant, R the gas constant, T the
temperature and the electrical potential.
This approach requires also the assumption of electroneutrality
in each region of the liquid phase expressed by the following
formula:
X
m
xi z i 0
i1
Another method to describe the interfacial mass transfer is
KarhunenLoeve Galerkin (KLG). The KLG method is a Galerkin
method employing the empirical eigenfunctions of the Karhunen
Loeve decomposition as basis functions. [14].
Both equilibrium and rate-based models may consider the reaction
kinetics or not. If the reaction rate is slower than the mass transfer rate,
kinetics should be considered as a dominating factor [13]. Fig. 1 is
widely cited as it clusters reactive absorption models through a very
efcient approach [15]. The lowest level of complexity begins from the
bottom left of Fig. 1 with the simple equilibrium stage model. This
model assumes physical equilibrium and no reactions within theore-
tical stages. The next level moving rightwards is the equilibrium model
including also reaction kinetics. The model can consider the bulk phase
reaction kinetics or both the bulk and lm reaction kinetics, depending
on its complexity and accuracy. The equilibrium stage extended by the
consideration of kinetically controlled reactions is physically incon-
sistent; however, this model has often been applied in the literature.
Upwards and beginning from the left, the simplest rate based approach
assumes mass transfer rates and could take into account moving
rightwards, additional factors such as reaction kinetics and electrolytes
inuence. The 5 types of models are numbered from 1 to 5 beginning
from the less complex (equilibrium stage, reaction equilibrium) to the
most complicated (rate based with reaction kinetics, lm reactions and
electrolytes), in order to be referenced later.
4.3. Mass transfer approaches
Different approaches were followed in order to describe the
mass transfer phenomena between the liquid and the gas phase.
The dominant theories are the two lm theory and the penetration
theory.
The two-lm theory [16] suggest that there are two thin areas
(lms) that are adjacent to the two phases interface separating it
from the bulk phases, which are ideally mixed. It is assumed that
550 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
Fig. 1. Different modeling approaches [Kenig et al., [9]].
overall mass and heat transfer resistance is concentrated in these
two lms considering that bulk phases are in equilibrium. The two
lm theory is prevailing among rate-based models. An important
factor is the lm thickness that is usually calculated using
empirical correlations. Two-lm theory is described by Ficks rst
law, using mass transfer coefcient correlations for a binary
system A
B.
dxA dC
N A
CDA;B
DA;B A ; 0 o x o n
dx dx
where N is the ux of each component, is the lm thickness and
D the diffusion coefcient.
These correlations reect the mass transport dependence on
physical properties and process hydrodynamics and are available
from the literature [15]. Models using two-lm theory, often
assume that the reactions take place only within the liquid lm,
when the reaction is characterized as fast [17].
The penetration theory was described by Higbie on 1935 [18]. It
states that diffusion is unsteady state process and the molecules of
the solute are in constant random motion, where clusters of these
molecules arrive at the interface, remaining there for a xed
period of time, and some of them penetrate while the rest mixes
back into the bulk of the phase. The process is described by Ficks
second law with boundary conditions, which is the following for a
binary system A
B.
2 C A C A
DA;B
x2 t
At t 0; C A C Ao 0
At x 0; C A C i
At x 1; C A C Ao 0
When systems with electrolytes are studied (e.g. water, amines),
their inuence cannot be neglected. Electrolyte systems are often
approached by the determination of free (or excess) Gibbs energy (GE
models). Gibbs energy is a result of two contribution terms; long-
range interaction contributions attributed to the electrostatic forces
between ions and short-range forces between all species (ion
ion,
molecule
molecule, molecule
ion) [19]. These models usually com-
pute the ionic activity coefcients through Debye
Hunkel equations.
4.4. Thermodynamic models
A fundamental step towards chemical absorption modeling is the
development of an efcient thermodynamic model to describe the
vapor
liquid phase equilibrium. In general, equations of vaporliquid
equilibrium describe the principle of chemical absorption and inves-
tigate the effects on gas solubility caused by temperature and pressure
[2022]. In specic for CO2 absorption, the main task of such models is
to determine the partial pressure of CO2 over an aqueous solution of
alkalonamine (or other solvent) and also quantify the energy required
for regeneration. In following, some widely used thermodynamic
models are provided, applied in their initial form or modied by
many researchers.
4.4.1. Kent
Eisenberg model
Kent
Eisenberg model was introduced on 1976 and it was the
rst model describing CO2 absorption by MEA solvents. This model
is an equilibrium model but is still advanced due to its simplicity.
Kent and Eisenberg described the two phase equilibrium through
Henry constants. A basic bottleneck for applying this model is that
it cannot predict the liquid phase composition [10]. The model was
enhanced by Krirpiphat and Tontiwachwuthikul on 1995, predict-
ing the carbon dioxide solubility in aqueous 2-amino-2-methyl-l-
propanol (AMP) solutions.
4.4.2. Electrolyte non-random two-liquid (e-NTRL) model
A widely accepted and applied Gibbs energy model is e-NTRL,
introduced by Chen et al. [23,24]. It is noteworthy that e-NTRL has
been included in the commercial ASPEN Plus package and has been
modied by various researchers. The eNRTL model is used to correlate
mean ionic activity coefcients of mixed solvent electrolyte systems
[19]. e-NTRL encompasses two main assumptions. It considers local
electro-neutrality and high repulsive forces between ions so that each
ion has an immediate neighborhood consisting of molecular species
and oppositely charged ionic species [19].
4.4.3. EoS/GE models
A variation of GE models is EoS/GE (equation of state with Gibbs
free energy) models. These models determine the Gibbs energy
through equations of state. The rst EoS/GE model was introduced
by Chunxi and Furst [25] and includes a specially adapted equation of
state considering the electrostatic interactions in the liquid phase.
UNIQUAC is also one of the classical models for EoS modeling of
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
activity coefcients. Through UNIQUAC, activity coefcients are calcu-
lated in relation to the excess Gibbs free energy. UNIFAC is another
model that can be described as a group contribution method. While
being based on essentially the same model as UNIQUAC it considers
molecules as a set of groups. By obtaining data for segments of
molecules it is possible to predict the activity of any molecule
composed of groups of which parameters are known. This gives
UNIFAC predictive properties that UNIQUAC does not have [26].
4.4.4. SAFT (statistical association uid theory) model
SAFT model also uses an equation of state (SAFT equation) but it
determines Helmholtz excess energy instead of Gibbs. SAFT model
does not require the equilibrium constants like other models but is
based on a number of factors. SAFT equations of state are clustered
in the mean eld theories, leading to low accuracy when local
properties of the uid have big uctuations [27].
4.5. Kinetic models
In order to approach the kinetics of chemical absorption of
CO2, the rst step of the researcher is to determine which
reactions are taking place in the liquid bulk. The type of reactions
considered depends on the assumptions for the CO2-solvent
reaction mechanism. There are three main approaches on the
mechanism of CO2 reaction with amine solvents, the zwitterion
mechanism, the termolecular and the based-catalyzed hydration
mechanism.
 The zwitterion mechanism
This reaction mechanism was rst proposed by Caplow and
reintroduced by Danckwerts. Based on this mechanism, the
solvent and CO2 form zwitterions rstly, and then zwitterion is
further deprotonated by a base to form carbamate [8].
 An extensive literature review is included in this section,
The termolecular mechanism
Termolecular mechanism suggests that the formation of the
zwitterions and the deprotonization occur simultaneously.
 Base-catalyzed hydration mechanism
Tertiary alkanolamines cannot react directly with CO2. Such
amines have a base-catalytic effect on the hydration of CO2 and
amine dissociation reaction may also occur.
Generally, the reaction of CO2 with primary, secondary and
sterically hindered amines is governed by the zwitterion mechan-
ism, whereas the reaction with tertiary amines is described by the
base-catalyzed hydration of CO2 [28].
The above mentioned reactions occurring within the liquid
phase are clustered in two categories; equilibrium reactions and
kinetically controlled reactions. Equilibrium reactions are fast
enough reactions to consider that the species are in chemical
equilibrium. Kinetically determined reactions should consider the
reaction kinetics. Equilibrium and rate constants (usually given by
temperature dependent correlations) are fundamental parameters
of a reactive absorption model, as they lead to the determina-
tion of the liquid phase composition. These constants usually
derive from tting experimental data or other researchers corre-
lations. In literature there are various equations for different
solvents, correlating equilibrium and rate constants with
temperature.
4.6. Absorption column model
Reactive absorption is a complex process making its modeling
challenging. Within a model, a number of assumptions for the
absorption column should be performed in order to simplify the
551
process. The main common assumptions that most simulations adopt
are the following:
 All reactions are performed within the liquid lm. The liquid
bulk is at chemical equilibrium.
 The column is adiabatic and pressure is constant.
 There is plug ow of gas and liquid
 The gas and liquid ow rates are constant throughout
the column
 The diffusion in the axial direction and the solvent evapora-
tions are negligible.
 The interface temperature is equal to the temperature of the
bulk phase.
 Ideal phases
4.7. Dynamic modeling
The majority of models refer to steady state conditions. However,
the CO2 absorption process should be coupled with the power plant
(or other industry) real time operation. In case of biofuels and coal-
based power plants, the parameters of the fuel might vary during
operation leading to uctuations in ue gas composition [29]. The
effect of these uctuations is crucial to be reected by the model. A
dynamic simulation tool can include three factors [30]; optimal
design of the absorption/desorption column often interferes with
each other, the load variation and the emerging technologies that
make the process more complex. Dynamic simulation will play a
pivotal role in identifying any operational bottlenecks at transient
conditions for the integrated power and CO2 capture plants. How-
ever, as mentioned, the validation of such models is not always
possible, as dynamic experimental data are very scarce.
5. Models review
providing information for a wide range of models examined.
5.1. Rational for this review
During the last decade, an increment of interest on modeling
post-combustion CO2 capture is observed. Numerous researchers
develop models covering all simulation approaches, different
solvents, evaluating different parameters aiming to the optimiza-
tion of the process and the effective junction with the power plant.
Searching literature there are only few attempts to list these
projects [29,31]. Existing reviews address mainly dynamic model-
ing and are providing brief descriptions of the studied models. The
current work attempts to describe in detail and compare in a
quantitative approach a number of fundamental models while also
providing tables presenting in short more models.
5.2. Chemical absorption models literature review
An extensive literature survey has been carried out, listing a
number of representative prototype models developed by
researchers based on previous work. The current work aims to
cover the spectrum of different solvents and modeling approaches.
5.2.1. Steady state models
At rst, Table 1 encompasses 7 steady state models that are
going to be further described in following. Following, the models
are further described and tables with main correlations applied are
presented (Tables 214).
Equilibrium models, as mentioned before, are not very applic-
able as equilibrium is very unlike to be reached. However, a
representative equilibrium stage model is presented.
552 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

Table 1
Overview of the studied steady state models.

Researcher Model Solvent Simulated Evaluation parameter Experimental data for Sensitivity analysis
type equipment validation

[Mores et al. 2 MEA Absorber/ Total heat duty/CO2 Pilot Plant Study at the Not performed
[32]] Desorber recovery and CO2 recovery/ University of Texas, Austin
amine solution ow-rate [Dugas, [38]]
optimization
[Khan et al., 4 MEA Absorber CO2 and MEA Two pilot [Tontiwachwuthikul Different correlations for mass transfer coefcients
[41]] concentrations along the et al. [43], Aroonwilas and one
column industrial plant [Pintola et al.
[44]]
[Kucka et al., 5 Amines Absorber CO2 concentration and Pilot [Tontiwachwuthikul et al. Evaluating different correlations for reaction kinetics
[49]] temperature prole [43]], and industrial plant
[Pintola et al. [44]]
[Bolhr- 4 MDEA Absorber CO2 and H2S A commercially operated Not performed
Norden- /Desorber concentrations along the selective MDEA plant
kampf et al., column
[52]]
[Gabrielsen 4 AMP Absorber CO2 concentration, liquid Pilot plant [Tontiwachwuthikul Performed for different parameters, model sensitive
et al. [5]] temperature along the et al. [43] to surface tension of the liquid solution, the Henrys
column Law constant for CO2, and the diffusivity of CO2 in
the liquid solution
[Puxty et al., 5 Amines/ Absorber Calculated versus measured A wetted wall apparatus Not performed
[57]] Ammmonia CO2 absorption ux
mixtures
[Al-Baghli 4 MEA and Absorber Column stages in order to Not available Not performed
et al., [58]] DEA achieve a specic level of
separation

Table 2
Mores model correlations.

Mores model

Parameter Correlation Nomenclature

Species mass balance in Lz 1xizLzxiz Vz
1yjz
1Vzyjz 0 aiz: activity

stage z
Energy balance in stage z
Chemical equilibrium
constants
Stage efciency
Enhancement factor
D: diffusivity of CO2 in MEA
: enhancement factor
Lz 1hz
Lzhz Vz
1Hz
1
VzHz HR
HH2O 0 G and L:gas and liquid velocities
 v
K m aiz vi xiz iz i kL,G: mass transfer coefcient for liquid and gas phase respectively
i i kr: forward reaction constant
L: liquid molar ow
z V z yiz
V z
1 yiz
1
or V: vapour molar ow
V z yniz
V z
1 yiz
1
h    i ! vi: estequiometric coefcient
h=a G' =RTkG G L' =kL L E yniz : equilibrium composition of molecular specie considered leaving the
z 1
e
z stage z
p
DCO2 kr;CO2
MEA MEA kr;CO2
OH CO2  iz : coefcient activity
Ez kL : stripping factor
: gas and liquid densities

Mores et al. [32] introduced an equilibrium stage model, in
order to simulate CO2 absorption by MEA solvents and optimize
the operating conditions to remove CO2 from ue-gases in a stage
column. The simulation tool determines the optimal operation
through two factors, given the ue gas composition; the ratio
between the total absorbed CO2 and the total heating and cooling
utilities (exchanger, condenser and reboiler heat duties) and the
ratio between total absorbed CO2 and the total amine ow-rate.
The model assumptions include ideal mixtures, equilibrium stages,
neglected solvent vaporization, and two-lm model. Mass and
energy balances are applied for each theoretical stage z. Chemical
equilibrium constants are calculated through activity coefcients
taken from Aboudheir et al. [33]. Electrolytes inuence is con-
sidered applying charge and ionic mass balance in each theoretical
stage. Moreover, thermodynamic non-idealities are expressed via
stage efciencies correlations as a function of liquid physical
properties and an enhancement factor. In order to optimize the
operations of the absorber the indicator of CO2 removal efciency
is introduced comparing the CO2 input and output concentrations.
In addition, the effective area of mass transfer is determined, using
equations developed by Onda et al. [34], Wilson [35] and Bravo
and Fair [36]. A similar model for the desorber is applied assuming
the same constraints. The stripper model also includes mass and
energy balances corresponding to the condensers, heaters and
reboiler. Finally, it considers constraints to compute the Henrys
law constant, enthalpies, reaction heats, viscosity, vapour pres-
sures, fugacity coefcients and surface tension among others.
The developed VLE (vapourliquid equilibrium) model was
validated against simulation results by Austgen and Rochelle [37]
and Aboudheir et al. [33]. The CO2 concentration is gured in
relation to the individual species concentrations and partial CO2
pressure. The agreement between the model and literature is
satisfactory. The absorber column is validated compared with
experimental measurements from pilot plant [38] and simulation
results using HYSYS process simulator. Results are presented for
the three approaches mentioned before (Onda et al. [34], Wilson
[35] and Bravo and Fair [36]) and compared with the experimental
data and simulation. Temperature, CO2 and vapour ow rate
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566 553

Table 3
Khan model correlations.

Khan model

Parameter Correlation Nomenclature

Enhancement factor E 1 h 1:35 1 i1=1:35 a: gasliquid interfacial area per unit volume of packing
1 1 1:35
E1
1 E1
1

where: CAi: CO2 concentration at the gasliquid interface

p
C Bl DB
E1 1 bD ; E1 Mp; M 2 DAl k02C2Bl CBl: MEA concentration in bulk liquid
C Ai Al tanh M
kl

Rate constant log k2 10:99
2152

T
CMEA: MEA concentration
p
Overall rate of absorption of CO2 r CO2   2  CCO2,g: molar concentration of CO2 in the gas phase
CO

H CO2
1

kg a k0 aE
1

Henrys constant H CO2 105:3
0:035C MEA
1140=T DAl and DBl: diffusivities of CO2 and MEA in liquid, respectively
 
Mass balance equation d uz C CO2;g
G: molar ow rate of gas per unit cross-sectional area of the column
dN CO2;g
dz
dz
SCO2
Energy balance equations dT g   Hg: gas phase heat transfer coefcient
GY CO2 C pg dz
hg a T g
T l
LC pl dT l dT
GY CO2 C pg dzg GHR
dY CO
H S dz 2 HR: heat of reaction
dz
Hs: heat of solution
k2: second-order rate constant
E: enhancement factor
HCO2: is the Henrys law constant
0
kg and kl : the gas- and liquid-lm physical mass transfer coefcients
NCO2,g( uzCCO2,g GYCO2): molar ux of CO2
pCO2: partial pressure of CO2 in bulk gas phase
SCO2: source term representing the overall rate of absorption of CO2
Tg: gas temperature
Tl: liquid temperature
uz: gas velocity in the axial direction
YCO2: mole ratio of CO2 in the gas phase

Table 4
Kucka model correlations.

Kucka model

Parameter Correlation Nomenclature

Mass action law p xvp;r vpp;r a: specic area

K r e xpv v
ep;r ep;r AC: cross-section area of column
Deff: effective diffusion coefcient
Bulk mass balance equations Lxlb
nlb lb l F: Faradays constant
i a Ac Ri Ac
i i
z
h: molar enthalpy
ni molar ux of component i
Gygb p: product index
z
i
ngb i
i a Ac
q: heat ux
RG: gas constant
Bulk enthalpy balance equations Lhlb r: reaction index
z qlf ai Ac
R: reaction rate
t: total
Ghgb
qgf ai Ac : activity coefcients
z
: lm thickness
: dimensionless lm co-ordinate
 
Film balance equations clft Dlef f ;i xlfi : thermal conductivity
nlft
xlfi zi RGF T xlfi nlft
1 l l
: volumetric hold-up
Pressure drop in the gas bulk p
z f L; G; l; g;
Heat ux through lms P
nl
T l lf
nlfi hi
lf
q
l
lf
l

i1
gf
T g
P
ng
gf
qgf
g g ngf
i hi
i1

proles are illustrated along the column height. All approaches Following the model validation, a scenario is tested where the
provide good predictions for both experimental data and simula- optimal operating conditions in absorber to maximize the CO2
tion results. In specic, for temperature and vapour ow rate the recovery are investigated. The optimization is oriented towards
correlation of Bravo is in a better agreement with the simulation two factors: the minimization of total heat duty/CO2 recovery
results. On the other hand, experimental measurements are better (OF1) and the maximization of CO2 recovery/amine solution ow-
predicted by Ondas equation. The validation of the stripper is rate (OF2). Proles of these factors optimal values, heating and
performed using experimental data by Tobienssen et al. [39]; cooling total utilities, amount of CO2 recovered, CO2 lean loading
Freguia and Rochelle, [40] and Dugas [38]. The paper does not and amine solution ow-rates are presented along the column
present analytical results for the regenerator but the predictions height. From these gures it can be seen that the optimal OF1
are in satisfactory agreement with the model simulations. value decreases as the absorber height increases. The same
554 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

Table 5
Bolhr
Nordenkampf model correlations.

Bolhr
Nordenkampf model

Parameter Correlation Nomenclature

Overall mass and energy transfer rates through the interfacial area on N i;k yi;k 1 V k 1
yi;k V k aint;k N Vi;k aint:the interfacial area
stage k of the column
Ek H k 1 V k 1
H k V k H: specic enthalpy
Mole fractions and at the vapourliquid interface yl Hi: Henry-coefcient of component i
Hi;k xli;k
i;k

Enhancement factor ECO2  1

1
2=3 E1 :enhancement factor for innite fast reaction
1

E1
1
Ha3=2 E 1 3=2
q
Film thickness 3wL Ha: Hatta number
2 , w  aint  u  sin
3
aint g sin
L
Ni: molar ux of component i
u: the average liquid lm velocity
V: vapour ow
the angel of the inclined plane in reference to the
horizontal plane

Table 6
Gabrielsen model correlations.

Gabrielsen model

Parameter Correlation Nomenclature

Molar ux of CO2 NCO2 aAc

1L dx B
L dxH2O L dX CO2 a: specic wetted area for mass transfer
1
 dz 1 dz
1 dz
Stoichiometric relations for the chemical reaction dG

N CO2 N H2 O aAc AC: cross-sectional area of the column
dz
 
dyCO N CO2 aAc yCO
1 N H2 O yCO aAc C 0AMP : concentration of AMP in the bulk liquid
2
2
G
2
dz
 
dyH O NH O aAc yH O
1 N CO2 yH O aAc C inf
CO2 : concentration of molecular CO2 at the liquid interphase
2
dz
2 2
G
2

dL

N CO2 aA c k2: rate constant
dz
 
dxCO2 N H2 O X CO2
N CO2 aAc K CO2 :combined Henrys Law and chemical equilibrium constant for
L
dz
   
dxH2 O N H 2 O xH2 O
1 N CO2 aAc CO2 partial pressure
dz
L Di,L: diffusivity of component i in the liquid
dT G

GCqaAc NCO2: molar ux of CO2
dz
 p;G

dT L N CO2 C p;CO2 N H2 O C p;H2 O aAc T L
T G q: heat ux
dz
LC p;L
qaAc N CO2 H CO2 N H2 O HH2 O


LC p;L LC p;L

q
Enhancement factor E1
E
M E1 XCO2: mole fraction of chemically bound CO2 in the bulk of the liquid phase

E
E q
1


E
tanh M EE1
1
E 1

where: HH2O: heat of condensation of H2O

 
DAMP;L C 0AMP k2 DCO2 ;L C 0AMP HCO2: heat of absorption of CO2
E1 1 , M 2
2DCO2 ;L C inf
CO
kL;CO
2 2

Equilibrium partial pressure pnCO2 K CO2 X CO2 1
: loading, mole CO2/mole initial amine
 
Heat of absorption CO2 R
8161 47652

Table 7
Puxty model correlations.

Puxty model

Parameter Correlation Nomenclature

Differential equations ci
Di xc2i
r i
2
ci: the concentration of species i
t
Averaged CO2 ux across the gas-liquid interface DCO R t cCO DCO2;W ; DN2O;W :respective diffusion coefcients of CO2 and N2O in water
N
te 2 0e x 2 dt DCO2; Am , DN2O;Am : respective diffusion coefcients of CO2 and N2O in an aqueous amine
solution
Ratio of diffusion coefcients between N2O and CO2in DCO2; Am DN2O;Am DCO2; W N: the time averaged ux
DN2O;W
water ri: the rate of formation or destruction of i by chemical reaction
x: the distance from the gas-liquid interface
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566 555

Table 8
Al-Baghli model correlations.

Al-Baghli model

Parameter Correlation Nomenclature

Enhancement factor
Flux of the solute gas

DidCi=dzj z 0 ri j z 0 Ci,int and Ci,bulk the interfacial and bulk molar concentration of the solute gas
E kl Ci; int
Ci;blk
  Di is the diffusivity of solute gas i in the liquid
N i;z V in yi;in
V out yi;out kg;i a P i;out
P i;int
0   kg,i and kl,i: gas and liquid phase physical mass transfer coefcients of the solute gas i
kl;i aEi C i;int
C i;bulk

Overall rate of absorption of CO2 For MEA

r CO2 kCO2 C CO2 C RR0 NH

kCO2 C RR0 NH2 C RR0 NCO2
Pi,int the interfacial partial pressure
K Am C RR0 NH
For DEA ri |z 0 the rate of surface reaction at the edge of the liquid lm

r CO2 kH2O C H2O kRR0 NH C RR0 NH R C2H4OH, R H for MEA

( )
C ' C RR0 NCO2
and
C CO2 C RR0 NH
RR NH2 R R C2H4OH for DEA
K Am C RR0 NH

Table 9
Overview of the studied dynamic models.

Researcher Model Solvent Simulated Evaluation parameter Data for validation Sensitivity
type equipment analysis

[Kvamsdal 4 MEA Absorber Temperature prole along the column, effect of Pilot Plant from Separation Research Not
et al., disturbances on temperature and absorption rate Program (SRP) at the University of Texas at performed
[30]] Austin [Dugas, [38]]
[Ghaemi 4 Partially Absorber Temperature prole along the column, CO2 concentration Absorption column pilot plant Not
et al. carbonated across the lm performed
[61]] ammonia
[Kenig et al. 5 Aqueous Absorber Dynamic and steady concentration proles in the Pilot plant of the coke plant August Not
[15]] ammonia discretized liquid lm, axial proles of the mass transfer Thyssena in Duisburg, Germany. performed
and reaction rates
[Harun 4 MEA Absorber/ L/G ratio, CO2 absorption, CO2 loading, gas ow rate in the Simulation results from Aspen Plus Not
et al. Desorber stripper performed
[63]]
[Jayarathna 4 MEA Absorber Time dependency of temperature proles, CO2 loading, CO2 Pilot Plant Study at the University of Texas, Not
et al., removal efciency, re-boiler heat duty Austin [Dugas, [38]] performed
2013]
[Lawal et al. 5 MEA Absorber/ Temperature proles along the column height, water mass Pilot plant of Separations Research Not
[70]] Desorber fraction, lean loading, CO2 capture level, reboiler duty Program at the University of Texas at performed
through time Austin

approach is followed aiming to OF2 optimization, concluding that
optimal OF2 and amine ow-rate increase with the increasing of
the absorber height. Finally, higher total heating and cooling
utilities are needed to reach the maximum CO2 recovery when
the absorber height is increased.
Khan et al. [41] developed a typical, simple steady state rate-
based model simulating CO2 absorption by MEA solvent, using
theoretical column stages. The developed model was validated
compared with two pilot columns [42,43] and one industrial
column [44]. The model applies the two lm theory and uses an
enhancement factor to consider the reaction kinetics. The overall
second order reaction rate is calculated through CO2 and MEA
concentrations as this approach is valid [45] and the rate constant
correlation was taken from [46]. The overall CO2 absorption rate is
calculated through Levenspiel [47] correlation and it is corrected
through the enhancement factor. Gas and liquidlm mass trans-
fer coefcients and the interfacial area for random packings are
obtained from the widely used correlations provided by Onda et al.
[34]. A number of semi-empirical correlations that relate CO2
diffusivity to temperature and viscosity are used. Henrys constant
was used in order to calculate the solubility of CO2 in MEA
solutions as a function of MEA concentration, CMEA, and solution
temperature. Column assumptions include steady-state and adia-
batic operations, plug-ow of gas and liquid, negligible diffusion in
the axial direction, constant gas and liquid ow rates throughout
the column, negligible evaporation of solvent and constant pres-
sure. Mass and energy balance equations were solved for different
heights of the absorption column to calculate CO2 concentration
prole along the column.
The comparisons of simulation predictions with the experi-
mental data showed the following: for Tontiwachwuthikul pilot
plant (random packing), overall agreement between the prediction
and measurement is satisfactory for CO2 and MEA concentration
along the column. The CO2 concentration is slightly underpre-
dicted for the bottom part of the column and for MEA over-
predicted respectively. However, this part is crucial as prediction
results showed that the major absorption process takes place in
the bottom part of the column as the concentrations decrease
rapidly. In addition, in these areas, the predicted solvent tempera-
ture deviates from experimental data. This deviation is according
to the authors due to the neglection of the solvent evaporation in
the energy balance. The temperature prole conrms that the
main exothermic absorption occurs in the same column heights, as
a rapid temperature drop is predicted. The predictions are also in a
very good agreement with experimental data for Aroonwilas pilot
plant (structured packing). For the industrial plant experimental
data were not available. Predictions were only validated against
the CO2 exit concentration with very good results.
A sensitivity analysis is also performed using different correlations
for mass transfer coefcients. The correlations for the gas and
556 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

Table 10
Kvamsdal model correlations.

Kvamsdal model

Parameter Correlation Nomenclature

Mass balance equations g

dC gi

C gi
ug C i g
ag=l :gasliquid interfacial area
dt z
ug z
ag=l N i
dC li C li
l dt
ul z ag=l N i C nMEA :liquid concentration of free MEA
T a     g ; l :gas and liquid holdup
Heat balance equations g dtg
ug zg P
dT g=l
C i Cpi g hg=l T l
T g
ag=l  
hg=l : interfacial heat transfer coefcient
l dt
ul z P C i Cpi l
T l
dT g

   
hg=l T l
T g
Hr N CO2
H vap N H2 O
H out T l
T amb

 
Species molar ux N i K tot;i P eq;
i
n

P gi K tot;i : total mass transfer coefcient
p
Enhancement factor kr;CO2 C nMEA DCO2 kr;CO2 : reaction rate coefcient for the reaction of CO2 with the aqueous MEA
ECO2 l
kCO
2

CO2 equilibrium pressure P eq; n l f ree N i : ux of component i

Equilibrium pressure of MEA
and H2O
CO2 H E;CO2 C tot xCO2 CO2
n n
P eq;
i xi i P i P eq;
i : Equilibrium pressure
P gi : partial pressure of
each component

ug ; ul : gas and liquid velocities

H r : heat of absorption of CO2
H vap : Heat of vapourization of H2O
H out : heat transfer coefcient for heat transferred from the absorber to the
surroundings
xfCO
ree
2
: free CO2 mole fraction at equilibrium coefcient
: activity coefcient

Table 11
Ghaemi model correlations.

Ghaemi model

Parameter Correlation Nomenclature

Dynamic mass balance equations    

a: specic packing surface
d
U z
dt i;j
1
Lj
1 xi;j
1
Lj xi;j nBi;j al RBLi;j L A
 
U Li;j xi;j U Lt;j xi;j L cLt;j A A: cross sectional area
h i  
0 z Gj
1 ys;j
1
Gj ys;j nBs;j al RBGi;j G A
1 E: specic energy holdup

h: molar enthalpy
h i  
Dynamic energy balance equations dEL

z
1 B B
Lj
1 hL;j
1
Lj hL;j qBL al RBL L oRL A
dt
h i  
0
z
1 B B
Gj
1 hG;j
1
Lj hG;j qBG al RBG G oRG A Ha: Hatta number

Heat transfer rate   P m N: interfacial molar ux

gf gf
qgf
gf T i
T gb ngf
i hi
i1
  Pm
lf
nlfi hi
lf
qlf
lf T lb
T i
i1
Molar concentration across the lm P
1 P1 hR i q: heat ux
t
Q n t

C ; t e
Q n t sinnn 0e f n d
n1 hn 1 i
sinn C l t C B t
C l t

p p
Enhancement factor k2 DCO2 NH3  R: reaction rate
E 1 Ha2 ; Ha kL
U: spedic molar holdup

: lm thickness
: volumetric holdup

liquidlm mass transfer coefcients given by Onda et al. [34]
provide better predictions than the penetration theory of Higbie
[18] and the correlation of Bravo et al. [48].
Kucka et al. [49] introduced a rigorous rate-based model to
simulate CO2 absorption by aqueous MEA solvents. Two types of
reactions are considered, namely the kinetically controlled reac-
tions and the instantaneous. In order to consider instantaneous
reactions, the mass action law based on species activities is
applied. Chemical equilibrium constants and reaction constants
are obtained from literature via activity based correlations. For-
ward and reverse reaction constants lead to the reaction rate
calculation. Contrary to the majority of models using an enhance-
ment factor, this model considers mass and heat transfer resis-
tances according to the lm theory by explicit calculation of the
interfacial uxes. The inuence of electrolytes is also included,
correlating the lm equations with reaction and diffusion kinetics.
The reaction rates are implemented as terms into the mass
transfer equations describing lm phenomena and into the bal-
ances describing liquid and bulk behavior. Bulk and liquid balance
equations are solved as well as correlations for enthalpy balance
and pressure drop in the gas bulk phase. In addition, mass transfer
phenomena across the interface are described by MaxwellStefan
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566 557

Table 12
Harun model correlations.

Harun model

Parameter Correlation Nomenclature

Molar balance for the two dC i;g C i;g ug ag=l : specic gas-liquid interfacial area
dt

z
ag=l N i
phases
dC i;l C i;l A: cross sectional area
dt

ul z
ag=l N i
Energy balance for the two T g T h a T
T C i;g ; C i;l : the molar concentrations of component i in gas and liquid phases
t
ug zg gl glPl g C i;g C p;i
phases
T l
t
ul Tz

l P
agl
C i;l C p;i C nCO2 : molar concentration of CO2 at the liquid side interface
   
hgl T l
T g
Hrxn N CO2
H vap N H2 O hout T l
T amb

Rate of mass transfer 1

k1g;i kHiE C p;metal : tube metal specic heat capacity
K g;i
p
l;i

Enhancement factors k2 C nMEA DCO2 C nMEA :liquid molar concentration of free MEA
Eabs k l;CO2
DMEACOO
p
K eq C MEA D: diffusivity
Estr 1  p 
DCO
p
2
D
n n
p
1 2 MEACOO
DCO K eq C CO C CO C CO2
2 2 2

Reboiler balance equations dM i;reb

F in xi;in
F v yi;out
F l xi;out Dc: diameter of the column
dt
dEreb
F in H in
F v H v
F l H l Q reb Eh;x : volumetric specic internal energy
dt
2
M i;reb 4 dc LB m xi Ereb : energy holdup
2
Ereb 4 dc LB H l m F: molar owrate
Heat exchanger equations For tube and shell: hgl :interfacial heat transfer coefcient
dEh;x 1 Q kg;i kl;i :mass transfer coefcients
dt

AL z Q f lux r
2

Q Tube;f lux U Tube T metal
T tube k2: second order reaction order
Q Shell;f lux ntube U Shell T metal
T Shell L: tube length
For tube metal: LB : liquid level of reboiler drum
 
mmetal C p;metal dTdt metal
Q
L
Q Tube;f lux
Shell;f lux M i;reb : molar holup
ntube
 
2
mmetal Lmetal 4 do;tube
di;tube
2 N i : molar ux of the component i
q: interfacial heat transfer
Q: energy ow rate
Q f lux : heat ux
Q Tube;f lux ; Q Shell;f lux :heat ux from tube wall to tube and from tube wall to shell
T amb : ambient temperature
T metal ; T tube ; T Shell : tube metal temperature, tube uid temperature and shell
uid temperature
ug ; ul : gas and liquid velocities
U Tube ; U Shell : overall mass transfer coefcient of the tube and shell
m : molar density
metal : tube metal density

Table 13
Kenig model correlations.

Kenig model

Parameter Correlation Nomenclature

   
Mass balance equations for liquid and gas phases lb

z Lxi nlb lb ai: specic gas-liquid interfacial area
i a Ri liq Ac
i lb
t U i
   
0 gb
gb i gb
Gyi ni a Ri gas Ac Ac : column cross section
z
Driving force xlfi ni molar ux of component i
di RT 1lf di
d
xlfi zi RT
F 1 d
lf d
 
Film balance equations clft Dlef f ;i xlfi Ri: total component reaction rate
nlft
xlfi zi RGF T
1
xlfi nlft
l
l

 
Reaction kinetics RG: gas constant
r 3 k3 T cCO2 cNH2
K1eq cH2NCOO
cH3 O
3 U: specic molar holdup
: lm thickness
: dimensionless lm co-ordinate
: volumetric hold-up

equations, assuming local electroneutrality and neglecting diffu- model was validated using two experimental data sets by Tonti-
sion interactions. Through these equations the two lm compo- wachwuthikul pilot plant and one data set from an industrial
nent uxes and the heat ux are determined. e-NTRL model is plant. For the rst occasion, the inuence of the number of axial
used to describe thermodynamic non-idealities in the liquid phase segments on the gas composition is calculated, showing that the
and for the gas phase the SoaveRedlichKwong equation of state minimum number of axial segments which provides results of
[50]. In order to simplify calculations, discretization of the lm sufcient accuracy for both data sets is 15. Gas phase composition
region is performed, dividing it to several balance regions. The and liquid phase temperature prole are presented along the
558 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

Table 14
Jayarathna model correlations.

Jayarathna model

Parameter Correlation Nomenclature

U
Columns balance equations clj:Col ulCol cj:Col
l
1 0 C: specie concentration
t zCol nj:Col;trans nj0 :Col;gen0
C vj:Col uvCol cj:Col
v

1
Ej: enhancement factor
zCol 1
nj0 :Col;trans0
1
t
 
T lCol ulCol T lCol h Aw
ov
T vCol
T lCol Hj : Henrys law coefcient
t zCol 3 l Pn
C p:Col clj;Col
j 1
 
3 U Pn 0
3
n0CO2;Col; trans
H abs j 1 nj;Col; trans
H vap:j
P P
C lp;Col nj 1 clj;Col C lp;Col nj 1 clj;Col
ov
  
T vCol uvCol T vCol h Aw T vCol
T lCol
P cvj:Col C vp:j:Col
t
1
zCol
1

 
Condenser equations dZ lRD
A 1 l m _ lRD;In
l V_ RD;Out
l K vj : local gas side mass transfer coefcient
dt RD
 
dclj:RD cl mlRD;In :mass ow rate of liquid into the reux drum
dt
1 l n_ lj:RD:In
j:RD l
mlRD;In
ARD Z RD

Reboiler equations l P v 
QRB: reboiler heat duty
Q_ RB mlRB;In C^ p RB H r n_ j:RB H~ vap:j
 
Buffer tank equations dZ lBT
1 l m _l
l V_
l rr: vector with overall reaction rates as elements
dt ABT BT;In BT;Out
 
dclj:BT clj:BT u: velocity
dt
1
n_ lj:BT;In ABT Z lBT n_ 0j:BT;gen
m
_ lBT;In
l
ABT Z lBT
P  l    P    
n_ lj:BT;In C~ j:BT;In
l l r

H~ r;m r m
l
v: is the stoichiometric matrix
P T BT ;In
T BT cl ABT Z lBT C~ p;j
m 1 P clj:BT C~ p;j
dT lBT l l
dt
j:BT

Overall mass transfer coefcient K ov

CO2
  1  HAbs, HVap, Hr heat of absorption, vapourization,
1=K vCO HCO2 =E CO2 K lCO reaction
2 2

Rate of generation of specie j in n_ 0gen vT r r : liquid hold up c, h cold, hot l,v: liquid, vapour
the liquid phase RD: reux drum ov: overall value : per volume gen,
trans: generation, transfer

column. The agreement was satisfactory for both parameters values (Non-equilibrium: 0.22 vol.%; TSWEET: 0.28 vol.% average
compared with the experimental data. Validating the model deviation).
against industrial experimental measurements, results are also in A rate based model studying aqueous solution of 2-amino-2-
a very good agreement compared with methods with enhance- methyl-1-propanol (AMP) which is a sterically hindered amine
ment factors. Finally, a sensitivity test was conducted, applying was developed by Gabrielsen et al. [5]. The model aims to predict
3 approaches for reaction kinetics from literature (12, 46 and 51). the solubility of CO2 in AMP aqueous solutions as well as the
The rst two approaches provide almost identical results for the absorption heat a function of loading and temperature. The
gas phase composition prole, in comparison with Danckwerts following assumptions were considered in order to model the
that leads to an underestimation of the column height of 15%. packed column: two lm theory, reactions only within the liquid
Bolhr-Nordenkampf et al. [52] developed a predictive tool for lm, constant temperature between the interface and the liquid
MDEA gas scrubbing processes, modifying Aspen plus RATEFRAC bulk, neglected axial dispersion and solvent evapouration, adia-
algorithm. The model is validated compared with a commercially batic conditions and ideal phases.
operated selective MDEA plant in Germany. The approach of stages Mass and energy balances are calculated through mass transfer
is adopted as calculations are made tray by tray and two-lm coefcients and an enhancement factor in order to take into
theory is considered for the mass transfer phenomena. The overall account the reaction inuence. Mass-transfer coefcients and the
mass and energy transfer coefcient rates are calculated through specic gasliquid interfacial area are calculated using correlations
Khrishnamurthy equations. Henrys law is used to calculate the from Billet and Schultes [53]. The enhancement factor is calculated
mole fraction at the vapourliquid interface. Mass transfer coef- using the expression proposed by van Krevelen and Hoftijzer [54].
cients for the two phases are determined through MaxwellStefan The model assumes a liquid phase ideality and uses one chemical
equations. Within the same work a prediction model for deso- equilibrium reaction to consider nonidealities combined Henrys
rption was developed too. Desorption was simulated through an Law and equilibrium constant. Using chemical equilibrium con-
equilibrium stage model. Equilibrium models are applicable for the stant, partial pressure of CO2 can be determined through the
stripper as there are high temperatures and therefore high vapourliquid equilibrium (VLE) model for aqueous solutions of
equilibrium like reaction rates. Considering the reaction kinetics, alkanolamines, presented in Gabrielsen et al. 2005 [55], as a
certain reactions are considered as very fast and are described by function of CO2 loading. This correlation gives satisfactory results
equilibrium constants and others reaction rates are taken from for the partial pressure over the loading range and temperatures
literature provided in a table in the paper. In order to describe the considered, compared with experimental data from Tontiwach-
reaction kinetics, an enhancement factor by Danckwerts equations wuthikul et al. [1992] including concentration and temperature
was used in order to correlate the physical mass transfer coef- proles along the column. In addition, the heat of absorption of
cients with the actual reactive transfer coefcients. Predictions for CO2 in the AMP solution (HCO2) was calculated, using the Gibbs
absorption were obtained for two packing heights using the Helmholtz equation [55]. However, there were no experimental
developed model and an equilibrium commercial tool. The devel- data available to verify the model predictions.
oped model predicts the measured H2S values far better (21 ppm Reaction kinetics is modeled assuming the zwitterion mechan-
average deviation) than the equilibrium approach (88 ppm aver- ism and using a correlation for the rate constant by Saha [56].
age deviation) simulation and gives comparable results for the CO2 Finally, prediction data are compared with experimental, for
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
concentration proles, liquid temperature proles and liquid
loading in different column heights. The results derived were very
good for the CO2 concentration in both gas and liquid phases,
although the liquid temperature is slightly overpredicted at the
bottom of the column. Sensitivity analysis was also performed,
concluding that the model is most sensitive to the surface tension
of the liquid solution, Henrys Law constant for CO2, and the
diffusivity of CO2 in the liquid solution.
Another steady state model was introduced by Puxty et al. [57]
describing CO2 mass transfer in amine/ammonia single systems or
their mixtures. This model was developed from scratch in Matlab and
was validated against experimental measurements from a wetted-wall
apparatus. Model assumptions include neglected axial dispersion and
heating effects, constant physical properties, two-lm theory. The
system of partial differential equations to be solved is dened by
second Ficks law considering also reaction rates. The reaction kinetics
vary as described in the previous model from fast to kinetically
controlled reactions. The rate and equilibrium constants were taken
from published data and corrected through activity coefcients. The
diffusion ux was calculated by the rst Ficks law for different heights
of the column.
The results from single systems are combined in order to
predict the mixtures results. Thus, complex reactions are not
involved. Single systems are solvents of monoethanolamine
(MEA) piperazine (PZ) and ammonia (NH3). Mixtures studies are
PZ
AMP and PZ
NH3 where PZ is used as a promoter. For single
solvent systems, agreement between predictions and experimen-
tal data for calculated versus measured CO2 absorption ux, is very
satisfactory. In particular for aqueous ammonia the data scattering
increases with temperature due to the high volatility that
increases the uncertainty of ammonia concentration in the liquid.
For mixtures studied, the excellent agreement of predicted and
measured values proves that behaviour of these mixtures can be
represented by individual component systems.
Al-Baghli et al. [58] introduced a steady state, rate-based model
applicable for MEA and DEA solvents. Solubility of both CO2 and H2S is
studied. Two-lm theory and stages approach is applied. The model
considers that reactions take place entirely within the liquid lm and
the phases are ideal. The liquid and the gas leaving a tray have the
same temperature and composition as the liquid and the gas on the
tray. Physical equilibrium of the interface is expressed through Henrys
law. This model also uses the enhancement factor to describe the
reaction kinetics. The enhancement factor was calculated through a
system of ve nonlinear differential equations governing diffusion and
reaction in the liquid lm. Moreover, the Kent
Eisenberg equations
are used to calculate the concentrations of the species in the liquid
bulk. The reaction rates are determined through the Danckwerts
approach. In addition, the heat of absorption is predicted for the two
solvents, using correlations for enthalpy and different temperatures.
Tray hydraulics and to obtain the tower diameter, tray spacing, weir
height, and pressure drop across the trays were obtained by Econo-
mopoulos algorithm. It was observed that MEA, as expected, requires
fewer stages than DEA in order to achieve a specic level of separation.
In addition, the prole of the enhancement factor for both MEA and
DEA is presented showing that for CO2 the factor increases mildly as
we approach the top of the tower, while for H2S the factor increases
sharply until its concentration is almost zero and then starts decreas-
ing until the top stages of the tower. Finally temperature prole of the
gas or liquid along the tower is presented reecting as expected the
absorption heat transfer phenomena.
5.2.2. Dynamic models
A respective analysis follows for the dynamic models studied
within this work. At this point, should be noted that sensitivity
analysis mentioned below, refers to a specic analysis taking into
559
account the impact of many parameters. The following dynamic
models test different disturbances for the system, but do not
perform a sensitivity analysis as it is dened.
Kvamsdal et al. [59] developed one of the rst dynamic
approaches simulating CO2 absorption by MEA solvents, using
gPROMS tool for model development. The initial tool was rstly
introduced on 2008 and was validated only against steady state
data from a MEA campaign at the University of Texas at Austin
(UT). In a subsequent work [30], the simulation tool was validated
compared with experimental results both for steady (from the
pilot plant at UT) and transient conditions (from a pilot plant at
NTNU/SINTEF). In the context of the same work different
parameter-correlations are studied and compared. The developed
simulation tool tests two scenarios for dynamic operation: start-
up and load reduction. The model applies the following assump-
tions for the wetted column: plug ow, axial differential mass and
energy balances for both gas and liquid phases, linear pressure
drop, ideal gas phase, two lm theory, no accumulation in gas and
liquid lms, thermal equilibrium between the liquid and packing
material, condensation of water vapour occurs at gas-liquid inter-
face as well as at the walls and effective contact area is equal to the
specic area of the packing material. Heat and mass balances are
solved for the bulk phases. All the mass transfer coefcients are
determined by correlations given by Onda et al. [34]. An enhance-
ment factor is used to describe the reaction inuence on mass
transfer. The reaction rate coefcient is based on data from Freguia
and Rochelle [40] and is corrected through the activity coefcients.
An equilibrium speciation [60] model is used to determine the free
CO2 mole fraction at equilibrium, the activity coefcient and the
concentration of free MEA and Wilson equation for the activity
coefcients of MEA and H2O. The equilibrium pressure of species is
calculated through activity coefcients. The physical properties
requires, are obtained by literature.
Firstly, steady state validation of the tool was performed,
against experimental measurements by Separation Research Pro-
gram (SRP) at the University of Texas at Austin. The temperature
prole deriving from the model agrees with the measurements,
with more satisfactory agreement with specic tests.
In following, a dynamic analysis is performed, testing two
scenarios: the change in liquid and ue gas ow rate during the
start-up and the load reduction cases. For the rst case, the liquid
temperature prole during start-up is provided, showing that the
T bulge location changed from the bottom of the column to the top
as the ue gas ow rate increased and the L/G thereby decreased.
Moreover, a decrease of the absorption rate values is observed. For
the second scenario, the upstream cooling tower and the blower
were also considered in a very simplied approach. In addition,
load reduction considered two cases; a reduction in load or
exhaust gas ow rate, without changing the liquid ow rate and
a decrease of the liquid ow rate such that the nal removal rate
was the same as before the gas ow rate reduction.
Ghaemi et al. [61] developed a rate-based model to simulate
carbon dioxide absorption by partially carbonated ammonia solu-
tions. The tool applies the two lm theory and assumes 100 stages
and ideal mixtures. Mass and energy balances equations are solved
for each stage. To describe axial temperature proles, energy
balances are modied including the product of the liquid molar
holdup and the specic enthalpy as energy capacity. In addition,
heat transfer rate is determined. The species concentration is
calculated through an initial boundary value problem considering
unsteady-state mass and heat transfer. Diffusion in the interface is
described by Maxwell-Stefan equations. Within this work the
interfacial compositions are determined through equations based
on the semi-empirical Pitzer equation. Mass transfer coefcients
are taken from Onda [34] and physical properties are obtained by
literature. Reaction kinetics considers reaction between CO2 and
560 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
water, reaction between CO2 and hydroxyl ions and reaction
between CO2 and NH3. The overall rate of reaction is the summa-
tion of the aforementioned reaction rates. Rate constants as a
function of temperature, are taken from Danckwerts and
Sharma [51].
The simulation tool was validated against pilot plant experi-
mental data deriving from steady state measurements. Dynamic
model equations are solved both with analytical and numerical
methods. The CO2 concentration was calculated in relation to the
lm coordinates from 0.0001 to 0.1 s. The results vary with time
and justify the need for dynamic simulation. At rst some steady
state results are presented. The CO2 concentration in relation to
the lm coordinates is presented for steady state conditions,
calculated both through the dynamic model and by using an
enhancement factor from literature [62]. The use of the enhance-
ment factor leads to an important relative error compared with
the numerical solution. In addition gas mole fractions over the
column height are presented. CO2 mole fraction decreases along
the column height due to its absorption, whereas NH3 and H2O
concentrations increase as a result of mass transfer from liquid
phase to the gas. Moreover, temperature prole along the column
is given, showing that temperature increases along the column, as
expected for an exothermal process. In order to determine dyna-
mically the overall CO2 concentration in the liquid as a function of
time and stage, both dynamic and enhancement factor approaches
were applied. The same approach is followed for the temperature
proles. The researchers use the relative deviation to compare the
two models. As proved, the enhancement factor leads to higher
deviations almost for all parameters measured.
Another dynamic approach to the CO2 absorption by MEA was
proposed by Harun et al. [63], based on the work by Kvamsdal
et al. [59]. The absorption column was discretized in a number of
stages and the two lm theory is applied. Mass and energy balance
equations are applied for each stage and each component for the
two phases. A correlation for the rate of mass transfer is used,
including an enhancement factor to consider reaction kinetics.
Different equations for the enhancement factor of the absorber
and the stripper are provided.
Moreover, the reboiler is modeled using a simple equilibrium
stage model. P-T ash calculations were used to determine the
compositions of vapour and liquid phases, at given pressure,
temperature and overall compositions. In following, the heat
exchanger is simulated developing a mathematical model, con-
sidering it as a counter-current shell and tube heat exchanger. This
model describes the heat exchanged between the hot lean amine
solution and the cold rich amine solution coming from stripper
and absorber, respectively. The simulation tool was validated
against steady state data. For the absorber, experimental measure-
ments by Lawal et al. [64] were used. By the results it is shown
that the model slightly underpredicts the temperature prole
along the column. The stripper model was validated comparing
the results with simulation data from Aspen plus Radfrac model. In
this case, temperature for different column heights is overpre-
dicted by the developed model. The simulation results for the
reboiler and heat exchanger present satisfactory agreement with
those obtained with Aspen Plus.
Dynamic simulations were conducted using a partial load
reduction scenario. The case study assumes a reduction of 50% in
the ue gas ow rate in relation to the initial steady state value.
The liquid to gas ratio (L/G) was calculated through time, con-
cluding that increased as the ue gas ow rate decreased linearly.
This is justied as the lean ow rate remains constant since the
stripper heat duty is the same. The CO2 loading decreased with
time. In addition, the percentage of CO2 absorbed signicantly
increased during the test. This is due to the corresponding
reduction in CO2 available to react with free amine in the absorber.
Finally, it is observed that gas ow rate of CO2 in the stripper was
signicantly reduced during the ramp test.
Kenig et al. [15] based on the steady state model described in the
previous section [49] developed a dynamic model simulating CO2
absorption by aqueous ammonia solvents. Mass balance equations
for the bulk phases include molar holdup terms due to dynamic
nature of the model. However, gas hold-up is neglected. For the lm
region, mass transfer phenomena are reduced to a steady state
problem. The effect of electrolytes is described via the electrical
driving force and the diffusion phenomena using Nernst Planck
equations and assuming local electroneutrality. The thickness of the
lm region is determined by empirical mass transfer coefcient
correlations. The model is validated simulating an innovative pur-
ication processes called ammonia hydrogen sulphide circulation
scrubbing (ASCS). The rate of the direct reaction between carbon
dioxide and hydroxyl ions (the most important reaction) is given by
Astarita, Savage and Bisio [65]. In addition, the reaction kinetics of
NH3 and CO2 has been considered in the model equations. The
developed model was compared with experimental data from ASCS
sidestream of the coke plant August Thyssena in Duisburg, Germany.
Concentration proles in the discretized liquid lm are presented.
The interfacial molar uxes are provided both using Mazwell
Stefan
equations without considering the electrical force and using Nernst
Planck equations. Fluxes calculated with the effective diffusion
approach differ only slightly from those obtained by the Maxwell
-
Stefan equations without the consideration of generalized driving
forces. The optimal height of the absorption column was investigated
testing several discretizations of the packing section of the absorber
with different numbers of segments in relation with the mass
transfer and reaction rates. Finally, axial concentration proles are
compared with experimental measurements.
Dynamic simulations include a sudden ammonia concentration
growth in the solvent feed case. Dynamic concentration proles
for the gas and liquid products are given as a function of time. The
results show that increasing solvent concentration reduces the
selectivity of the chemisorption and much of the ammonia gets
lost by the CO2 consumption.
Jayarathna et al. [66] developed a general model for amine
based solvents. Separate sub-models are developed for absorber,
stripper, condenser and reboiler. The main balance equations are
considered the same for the main columns, altering the boundary
conditions of the differential equations deriving from the reux
stream enters the top most section of the stripping tower The
model assumption include constant ow rate, ideal phases,
neglected interfacial mass transfer of N2 and O2, reactions only
in the liquid lm, linear pressure drop, adiabatic system, each
phase in a control volume of the spatially discretized model
behaves as a continuous stirred tank.
The condenser of the stripping tower is modeled as a con-
denser block followed by a reux drum. The re-boiler is modeled
using a xed vapour-to-feed fraction and the reboiler heat duty is
calculated. The heat exchanger model consists of energy balance
equations for the rich (cold) and lean (hot) amine streams in the
cross heat exchanger and cooling water (cold) and lean amine
(hot) streams in the lean cooler. Additional correlations are applied
for the buffer tank model.
Mass transfer is considered as a concentration driven process.
The equilibrium concentrations are determined by Henrys law
coefcient which is taken from Kent
Eisenberg model. For CO2
the overall mass transfer coefcient is given as a function of the
local mass transfer coefcients, Henrys law coefcient and the
enhancement factor. The forward reaction rate coefcients are
calculated by correlations by Jamal et al. [67] and Freguia and
Rochelle [40], while the equilibrium coefcients for the reactions
are found using the KentEisenberg model [68]. Using these
coefcients the overall reaction rate is calculated.
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
The model was validated compared with experimental data
[38,59,69]. CO2 loadings, CO2 removal efciency, re-boiler duty,
temperature prole and ow Rates are compared with the measure-
ments and are found to be in good agreement. For dynamic simula-
tions, a number of cases of operation disturbances are validated
against steady state data. These scenarios are namely, increase of
molar ow rate of the inlet gas stream into the absorption tower,
increase of the volume ow rate of the inlet MEA stream into the
absorption tower and increase if V/F fraction of the re-boiler.
Lawal et al. [70] introduced a model investigating the operation
of coal-red power plants with post-combustion capture using
dynamic modeling and simulation. The model is a result of
numerous amendments of a dynamic model, simulating both the
absorber and the desorber linked together with recycle. Assump-
tions for the absorber include adiabatic system, plug ow, linear
pressure drop, negligible holdup in the vapour bulk, solvent
degradation and reactions are assumed to reach equilibrium. The
regenerator model consists of the GasLiquid Contactor model (for
the packed column section) linked with two ash drum models
which represent the condenser and reboiler of the unit. Correla-
tions for the heat of absorption were obtained from the literature.
Mass and energy balance equations are solved for the two
columns. In specic, for the vapour bulk accumulation of mass
and energy holdups are negligible. In following, correlations for
the heat loss and desorption heat are applied as a function of
temperature and other parameters. Mass transfer coefcients in
the liquid and vapour lms were determined by correlations given
by Onda et al. [34]. Phase equilibrium between liquid and vapour
phases is assumed at the interface. The Maxwell
Stefan equations
are solved, considering the two lm theory. The equilibrium molar
compositions of the components in the vapour and liquid phases,
are estimated based on the vapour and liquid fugacity coefcients.
Physical property estimations for viscosities, enthalpies and den-
sities were determined using the physical property package Multi-
ash, while the fugacity coefcients and ash calculations in the
reboiler and condenser were determined using the Electrolyte
Non-random-two-liquid (NRTL) model in Aspen Properties and
accessed through the CAPE-OPEN Thermo interface. The steady-
state validation of the chemical absorption plant model was
carried out using data from the Separations Research Program at
the University of Texas at Austin [38]. Validation was performed
for stand-alone columns and for the integrated system. Simulation
results were validated using the temperature prole of the
absorber column measured in the pilot plant, with good agree-
ment and higher deviations for the desorber predictions. The
predictions for the integrated system presented higher deviation
from the experimental case for the temperature prole especially
in the regenerator because of certain model limitations. Finally, a
dynamic analysis is performed, testing the behavior of the system
under the following scenarios; switching off water balance control,
increasing ue gas ow rate to the absorber, reducing reboiler heat
duty and increasing CO2 concentration of ue gas to the absorber
column. Dynamic analyses showed that the absorber performance
is more sensitive to the molar L/G ratio than the actual ow rates
of the solvent and ue gas. The performance of the regenerator is
affected by the reboiler duty.
5.2.3. Model comparisons
The current section includes comparisons of key simulated
parameters by models that used the same experimental measure-
ments as reference for validation. As mentioned before, there are
no dynamic experimental data available. Therefore, both for steady
state and dynamic models, validation was performed against
steady state data. Thus, the comparisons consider the models as
steady state. The comparison was made using information
561
included in the cited articles, thus in some cases the comparison
is graphical and includes a wide range of values and in other cases
individual values are compared. The models to be compared are
either described in detail in the previous section or described in
short in this section.
In following, a short description of the pilot plants where the
experiments were conducted is provided.
5.2.3.1. Tontiwachwuthikul pilot plant. Tontiwachwuthikul et al.
[43] carried out a series of experiments in a larger counter
current pilot-plant absorption column consisting of six packed
beds with a total packing height of 6.6 m and 0.1 m in internal
diameter. The column was packed with 12.7 mm ceramic Berl
saddles. Steady-state gas phase CO2 and MEA concentrations and
solution temperatures were measured along the column height for
different gas/liquid ratio, inlet CO2 and MEA concentrations.
Three models were selected for comparison, all of them using a
specic run tests (T22) measurements of Tontiwachwuthikul
plant. Khan model was analyzed in the dynamic models section.
The other models are work of McDowell et al. [71] and Liu et al.
[72]. McDowell uses a rate based approach and applies SAFT-VR
equation of state to describe the thermophysical properties and
uid-phase behaviour of this process instead of using an enhance-
ment factor. The dynamic analysis of the model includes evalua-
tion of the effect of changing the lean solvent ow rate and
thermodynamic state on column behavior. Liu proposes a complex
computational mass transfer model (CMT). The model consists of
the differential mass transfer equation with its auxiliary closing
equations and the accompanied formulations of computational
uid dynamics (CFD) and computational heat transfer (CHT).
Two parameters are presented along the column height, the
CO2 concentrations % (Fig. 2) and the liquid phase temperature
(Fig. 3). As seen by the gures, Khan model slightly underpredicts
experimental measurements while McDowell and Liu predictions
overpredict Tontiwachwuthikuls data especially in the area
between 1 and 3 m height column. At the same area Liu provides
more accurate predictions from McDowell for CO2 concentration.
In Fig. 3, the temperature prole along the column is illustrated
for the three models and the respective experimental data. The
same remarks are observed for the liquid phase temperature
predictions as for the CO2 concentration. However, Mc Dowell
model seems to be the only one predicting the expected tempera-
ture bulge. It is known that a temperature bulge can be located in
the bottom of the absorber column [73].
5.2.3.2. Munmorah pilot plant. This pilot plant was jointly constructed
by Australias Common wealth Scientic and Industrial Research
Organization (CSIRO) and Delta Electricity at Munmorah power
station. The pilot plant mainly includes two absorbers in parallel or
series, one heat exchanger and one stripper. The inner diameter and
Fig. 2. Comparison of the CO2 concentration predictions of Khan and McDowell
models and experimental data from Tontiwachwuthikul [T22].
562 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

Fig. 3. Comparison of the temperature prole predictions of Khan and McDowell models and experimental data from Tontiwachwuthikul [T22].

Table 15
Lawal model correlations.

Lawal model

Parameter Correlation Nomenclature

Mass balance for bulk phases dMi

L U1A
F Li A: cross-sectional area
dt y N i U Sp U MW i U

0
F

L U1A yi
V
N i U Sp U MW i U ct: total molar concentration
 
Energy balance for bulk phases
L

1 F H
Sp U U H cond conv f: fugacity coefcient
L U A y liq H liq H abs HL
dU
dt
 
F V
0 L
U 1A yH Sp U U H cond conv Fi: component mass ow rate
vap H vap

Heat Loss (HL) H: heat ux

     
Heat of absorption habs R U
14281
1092554
2

6800882 32670:01 U HL: heat loss to surroundings

Film equations nc xM N
xM N R
P  L: length of column section
1 xi
M
i
z ct
1 i k k T
i;k 298:15
k1
1 GT 1 H f lux M: mass holdup
z
z 0
Interface model L V MW:molecular weight
f i xM;L
i f i xM;V
i

N: molar ux nc: number of components

R: universal gas constant
Sp: specic area
U: energy holdup
: CO2 loading
: lm thickness
: thermal conductivity
: viscosity
: wetted area ratio
Superscripts
Conv:convection
I: interface
L:liquid
Lb:liquid bulk
Lf:liquid lm
M:molar
R:reference
V:vapour
Vb:vapour bulk
Vf:vapour lm
Subscripts
Abs: absorption
H: enthalpy
i: component number
Liq: liquid
Vap: vapour

the packed height of each absorber are 0.6 m and 3.9 m, respectively. Two models are compared using the same experimental
The inner diameter and the packed height of the stripper are 0.4 m and measurements by Munmorah pilot plant. In this case, individual
3.5 m, respectively. The absorbers and stripper are both packed with values of parameters were provided so the comparison is per-
25 mm Pall rings. formed via Table 15. In order to compare the two values the
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566 563

relative deviation is calculated via the following equation: models for the NH3CO2 H2O system to predict the performance of
CO2 capture. The analysis includes the species concentration prole,
X exp:
X pred:
RD%  100 temperature prole, mass transfer rate and coefcient in the gas and
X exp:
liquid bulk phase along the packing height. Observing the table, it is
not clear which model is more accurate as they are both in good
The rst researcher [74] used the commercial RateFrac model of agreement with the experimental data (Table 16).
Aspen plus to simulate the CO2 absorption and regeneration processes
of a research scale pilot plant. The effects of lean solvent ow rate, NH3
5.2.3.3. Pilot plant study at the university of Texas at Austin. The
concentration, CO2 loading, and temperature on the absorption
absorber column of the pilot plant is a packed column with a
process were analyzed. In addition, optimization analysis for the CO2
diameter of 0.427 m and total packing height of 6.1 m. This column
regeneration process was carried out. The second model was devel-
consists of two 3.05 m packed bed sections with a collector plate
oped by Qi et al. [75] and also aims to simulate CO2 absorption using
and redistributors between the beds. The models compared are by
aqueous ammonia. The model is based on the RateSep module, a rate-
Lawal [76] and Jayarathna [66] described in detail before. As seen
based absorption and stripping unit operation model in Aspen Plus,
by Table 17, in most parameters, Jayarathna model predicts more
and uses the available thermodynamic, kinetic and transport property
efciently the experimental measurements.

Table 16
Comparison of Munmorah simulations. 6. Conclusions and scope for future work

Munmorah pilot plant simulations

It is unquestionable that the need for techniques targeting to
Parameter Model polluting emissions reduction, becomes imperative. CCS is as described
a satisfactory solution, especially post-combustion technologies. Che-
Case 31 mical absorption is extensively studied, especially amine and ammonia
CO2 absorption rate (kg/h) Qi Zhang systems. Various researchers have studied the thermodynamic and
Model prediction 77.4 83.49
Experimental 80 80
chemical phenomena aiming to the optimization of the system.
Relative deviation (%) 3.25 4.36 Literature includes mainly steady state analysis examining parameters
NH3 concentration in exit gas (ppm) as species concentration and temperature prole along the column.
Model prediction 18,589 14,415 Most researchers have mainly dealt with absorption column, while
Experimental 19,463 19,463
few models exist for desorption. The papers that include both columns
Relative deviation (%) 4.491 25.936
CO2 loading of rich solvent usually study them separately, not considering their interaction.
Model prediction 0,338 0.36 Supplementary equipment as reboiler, condenser etc. are usually
Experimental 0.345 0.345 simulated with rough simplications. By this spectrum, the columns
Relative deviation (%) 2.029 4.348 operations are well understood.
Case 31B
CO2 absorption rate (kg/h)
However, there is a necessity to interconnect the system with
Model prediction 61.8 66.2 real conditions of the power plant. Disturbances that may occur in
Experimental 68.2 68.2 actual operation conditions are usually not reected in the existing
Relative deviation (%) 9.384 2.933 models. In cases that such an analysis is attempted, results cannot
NH3 concentration in exit gas (ppm)
be validated. The target of the developed researches is the
Model prediction 8959 7936
Experimental 9173 9173 commercialization of the process and this requires a very careful
Relative deviation (%) 2.333 13.485 investigation of its real impact. Thus, during the last years
CO2 loading of rich solvent increasing modeling attempts are observed, that develop dynamic
Model prediction 0.25 0.33 simulation tools. Dynamic CO2 capture model simulations have
Experimental 0.34 0.34
only analysed restricted disturbances as start-up or load uctua-
Relative deviation (%) 26.471 2.941
Case 38 tions. In addition, dynamic models are validated only in a steady
CO2 absorption rate (kg/h) state approach as dynamic experimental measurements are not
Model prediction 80.8 75
Experimental 75.44 75.44
Table 17
Relative deviation (%) 7.105 0.583
Comparisons of Austin pilot plant simulations.
NH3 concentration in exit gas (ppm)
Model prediction 3849 5132
Parameter Model
Experimental 3756 3756
Relative deviation (%) 2.476 36.635
CO2 removal efciency (%) Jayarathna Lawal
CO2 loading of rich solvent
Model prediction 96 99.5
Model prediction 0.22 0.4
Experimental 95 95
Experimental 0.42 0.42
Relative deviation (%) 1.053 4.737
Relative deviation (%) 47.619 4.762
Case 39 Rich solvent loading (mol/mol) Jayarathna Lawal
CO2 absorption rate (kg/h) Model prediction 0.448 0.456
Model prediction 87.8 87.33 Experimental 0.428 0.428
Experimental 85.6 85.6 Relative deviation (%) 4.673 6.542
Relative deviation (%) 2.570 2.021
Lean solvent loading (mol/mol) Jayarathna Lawal
NH3 concentration in exit gas (ppm)
Model prediction 0.285 0.279
Model prediction 8611 7740
Experimental 0.279 0.279
Experimental 10,071 10,071
Relative deviation (%) 2.151 0
Relative deviation (%) 14.497 23.146
CO2 loading of rich solvent Absorber inlet liquid temperature (K) Jayarathna Lawal
Model prediction 0.284 0.38 Model prediction 314 331
Experimental 0.38 0.38 Experimental 314 314
Relative deviation (%) 25.263 0 Relative deviation (%) 0 5.414
564 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566

currently available. Dynamic models mainly deal with MEA as the
most applied solvent. Concluding, the future works should:
 Develop dynamic models, including all equipment in an inte-
grated way, testing more realistic and complex power plant
operation scenarios.
 Collect dynamic experimental data in order to be able to
validate simulation tools as verication is necessary to increase
the model reliability and accuracy.
 Investigate alternative solvents or blends that increase
efciency.
 Perform sensitivity analysis that will examine more parameters
and their effect on the optimization process.
The utmost objective is the optimization of the process,
providing the optimal solution for each case.
The current review could be extended by encompassing latest
future models and also by performing more thorough comparisons
between comparable simulations. More models could be evaluated
in relation to their efciency with different solvents or models
validated against the same experimental data could be compared
using more efciency indicators.
Acknowledgements
The author acknowledge with appreciation the Laboratory of
Applied Thermodynamics, Thermal Engineering Section, School of
Mechanical Engineering, National Technical University of Athens,
for the support of the work on which this paper is based.

Appendix A. List with recent models

Researcher Model Dynamic/ Solvent Simulated Evaluation parameter Data for validation
type Steady equipment

[Pintola 4 S MEA Absorber MEA concentration, enhancement factor, Pilot plant

et al., CO2 concentration, temperature along
[44]] the column
[Aboudheir 4 S AMP Absorber Concentration and temperature proles Two pilot plants (University of
et al., along the packed column British Columbia and
[77]] University of Regina)
[Nittaya 4 D MEA Absorber/ The performance of different control Pilot plant [Dugas, [38]]
et al., Desorber schemes was evaluated under different
[78]] scenarios, e.g. changes in ue gas ow
rates, set point tracking, and valve
stiction
[Dugas and 5 S MEA and PZ Absorber Mass transfer coefcients in relation to Pilot plant [Dugas, [38]]
Rochelle, CO2 loading, temperature and amine
[79]] concentrations
[Saimpert 4 S MEA Absorber CO2 and temperature proles, CO2 CSIROs PCC Process
et al., /Desorber loading, outlet liquid ow rate, water Development Facility and a
[80]] absorbed, CO2 desorbed, reux ow rate, pilot plant (Tarong Power
loading for different operating points Station)
[Gaspar 4 D Alkanolamines Absorber CO2 molar fraction, liquid phases Pilot plants Aroonwilas and
et al., temperature, L/V ratio along the column Veawab [42], Tobiesen and
[81]] height. CO2 molar fraction through time Svendsen [39] and
Tontiwachwuthikul et al. [43]
[Moioli 5 S PZ Absorber Temperature and CO2 vapour molar Pilot plant [Plaza, [83]]
et al., fraction along the column
[82]]
[Sema et al., 4 S DEAB Absorber Liquid phases speciation and Pilot plant [Shi et al. [85]]
[84]] concentration in aqueous DEAB
solutions as a function of CO2 loading,
absorption rate, second order reaction
rate constant k2
 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
References
[1] Huaman RNE, Jun TX. Energy related CO2 emissions and the progress on CCS
projects: a review. Renew Sustain Energy Rev 2013;31:36885.
[2] Ming Z, Shaojie O, Yingjie Z. CCS technology development in China: status,
problems and countermeasures-based on SWOT analysis. Renew Sustain
Energy Rev 2014;39:60416.
[3] Intergovernmental Panel on Climate Change (IPCC). Special report on carbon
dioxide capture and storage. Cambridge, UK: Cambridge University press;
2005.
[4] Koroneos C, Grigoriadis Th, Kosmidou M. Energy analysis of a lignite power
plant with carbon sequestration. In: Proceedings of the 23rd international
conference on efciency, cost, optimization, simulation and environmental
impact of energy systems. Lausanne: ECOS; 2010.
[5] Gabrielsen J, Michelsen M, Stenby E, Kontogeorgis G. Modeling of CO2 absorber
using an AMP solution. AIChE J 2006;52:344351.
[6] Leung DYC, Caramanna G, Maroto-Valer MM. An overview of current status of
carbon dioxide and storage technologies. Renew Sustain Energy Rev
2014;39:42643.
[7] Sakwattanapong R, Aroonwilas A, Veawab A. Reaction rate of CO2 in aqueous
MEA-AMP solution: experiment and modeling. Energy Proc 2009;1:21724.
[8] Zhao B, Su Y, Tao W, Li L, Peng Y. Post-combustion CO2 capture by aqueous
ammonia: a state-of-the-art review. Int J Greenh Gas Control 2012;9:35571.
[9] Kenig EY, Schneider R, Gorak A. Reactive absorption: optimal process design
via optimal modelling. Chem Eng Sci 2001;56:34350.
[10] Vrachnos A. Absorption of acid gases into aqueous alkanolamine solutions:
experimental investigation and thermodynamic modeling. Athens: National
Technical University of Athens; 2007 PhD Thesis.
[11] Peng J, Lextrait S, Edgar T, Eldridge RB. A comparison of steady-state
equilibrium and rate-based models for packed reactive distillation columns.
Ind Eng Chem Res 2002;41:273544.
[12] Kucka L, Richter J, Kenig EY, Gorak A. Determination of gasliquid reaction
kinetics with a stirred cell reactor. Sep Purif Technol 2003;31:16375.
[13] Kenig EY, Seferlis P. Modeling reactive absorption. Chem Eng Prog
2009:6571.
[14] Park HM. Reduced-order modeling of carbon dioxide absorption and deso-
rption with potassium carbonate promoted by piperazine. Int J Heat Mass
Transf 2014;73:60015.
[15] Kenig EY, Schneider R, Gorak A. Rigorous dynamic modelling of complex
reactive absorption processes. Chem Eng Sci 1999;54:5195203.
[16] Whitman WG. The two-lm theory of absorption. Chem Met Eng 1923;29:147.
[17] Astarita G. Mass transfer with chemical reaction. New York: Elsevier; 1967.
[18] Higbie R. The rate of absorption of pure gas into a still liquid during short
periods of exposure. Trans Am Inst Chem Eng 1935;31:365.
[19] Chen C, Song Y. Generalized electrolyte-NRTL model for mixed-solvent
electrolyte systems. www.interscience.wiley.com; 2004.
[20] Koronaki IP, Christodoulaki RI, Papaefthimiou VD, Rogdakis ED. Critical review
of coupled heat and mass transfer models for a liquid desiccant adiabatic
dehumidier and regenerator. Adv Build Energy Res 2014;8(2):11736.
[21] Koronaki IP, Christodoulaki RI, Papaefthimiou VD, Rogdakis ED. Thermody-
namic analysis of a counter ow adiabatic dehumidier with different liquid
desiccant materials. Appl Therm Eng 2013;50(1):36173.
[22] Koronaki IP, Christodoulaki RI, Papaefthimiou VD, Rogdakis ED. Thermody-
namic analysis of an internally heated regenerator using three liquid desiccant
solutions. In: Proceedings of the ASME 2012 Heat transfer summer conference
collocated with the ASME 2012 uids engineering division. Summer meeting
and the ASME 2012 10th international conference on nanochannels, micro-
channels and minichannels. HT; 2012.
[23] Chen C, Britt HI, Boston JF, Evans LB. Local composition model for excess gibbs
energy of electrolyte systems. Part I: single solvent, Single Completely
Dissociated Electrolyte Systems. AIChE J 1982;28:588.
[24] Chen C, Evans LB. A local composition model for the excess Gibbs energy of
aqueous electrolyte systems. AIChE J 1986;32:444.
[25] Chunxi L, Furst W. Representations of CO2 solubility in aqueous MDEA
solutions using an electrolyte equation of state. Chem Eng Sci 2000;55:2975.
[26] Falck da Silva E. Modelling of activity coefcients using computational
chemistry. Report in DIK; 2002.
[27] Diamantonis N. Evaluation of statistical associating uid theory (SAFT) and
perturbed chain-SAFT equations of state for the calculation of thermodynamic
derivative properties of uids related to carbon capture and sequestration.
Energy Fuels 2011;25(7):333443.
[28] Vaidya PD, Kenig EY. CO2-alkanolamine reaction kinetics: a review of recent
studies. Chem Eng Technol 2007;30:146774.
[29] Chikukwa A, Enaasen N, Kvamsdal HM, Hillestad M. Dynamic modeling of
post-combustion CO2 capture using amines a review. Energy Proc
2012;23:8291.
[30] Kvamsdal HM, Jakobsen JP, Hoff KA. Dynamic modeling and simulation of a
CO2 absorber column for post-combustion CO2 capture. Chem Eng Process:
Process Intensif 2009;48(1):13544.
565
[31] Bui M, Gunawan I, Verheyen V, Feron P, Meuleman E, Adeloju S. Dynamic
modelling and optimisation of exible operation in post-combustion CO2
capture plants a review. Comput Chem Eng 2014;61:24565.
[32] Mores P, Scenna N, Mussati S. Post-combustion CO2 capture process: equili-
brium stage mathematical model of the chemical absorption of CO2 into
monoethanolamine (MEA) aqueous solution. Chem Eng Res Des
2011;89:158799.
[33] Aboudheir A, Tontiwachwuthikul P, Chakma A, Idem R. Kinetic of reactive
absorption of carbon dioxide in high CO2-loaded, concentrated aqueous
monoethanolamine solutions. Chem Eng Sci 2003;58:5195210.
[34] Onda K, Takeuchi H, Okumoto Y. Mass transfer coefcients between gas and
liquid phases in packed column. J Chem Eng Jpn 1968;1(1):5661.
[35] Wilson I. Gas liquid contact area of random and structured packing. Texas:
University of Texas at Austin; 2004 M.S. Thesis.
[36] Bravo JL, Fair JR. Generalized correlation for mass transfer in packed distilla-
tion columns. Ind Eng Chem Process Des Dev 1982;21:162.
[37] Austgen DM, Rochelle GT. Model of vaporliquid equilibria for aqueous acid
gasalkanolamine systems using the electrolyte-NRTL equation. Ind Eng Chem
Res 1989;28:106073.
[38] Dugas ER. Pilot plant study of carbon dioxide capture by aqueous mono-
ethanolamine. Texas: University of Texas at Austin; 2006 M.S. Thesis.
[39] Tobienssen FA, Juliussen O, Svendsend H. Experimental validation of a
rigorous desorber model for CO2 post-combustion capture. Chem Eng Sci
2008;63:264156.
[40] Freguia S, Rochelle GT. Modeling of CO2 capture by aqueous monoethanola-
mine. AIChE J 2003;49:167686.
[41] Khan FM, Krishnamoorthi V, Mahmud T. Modelling reactive absorption of CO2
in packed columns for post-combustion carbon capture applications. Chem
Eng Res Des 2011;89:16008.
[42] Aroonwilas A, Chakma A, Tontiwachwuthikul P, Veawab A. Mathematical
modelling of mass transfer and hydrodynamics in CO2 absorbers packed with
structure packings. Chem Eng Sci 2003;58:403953.
[43] Tontiwachwuthikul P, Meisen A, Lim J. CO2 absorption by NaOH, monoetha-
nolamine and 2-amino-2-methyl-1-propanol solutions in a packed column.
Chem Eng Sci 1992;47(2):38190.
[44] Pintola T, Tontiwachwuthikul P, Meisen A. Simulation of pilot plant and
industrial CO2MEA absorbers. Gas Sep Purif 1993;7(1):4752.
[45] Danckwerts PV. Gasliquid reactions. New York: McGraw-Hill; 1970.
[46] Hikita H, Asal S, Ishikawa H, Honda M. The kinetics of reactions of CO2 with
MEA, DEA and TEA by a rapid mixing method. Chem Eng J 1977;13:712.
[47] Levenspiel O. Chemical reaction engineering. 3rd ed. New York: John Wiley &
Sons; 1999.
[48] Bravo JL, Rocha JA, Fair JR. Mass transfer in gauze packings. Hydrocarb Process
1985;64:915.
[49] Kucka L, Muller I, Kenig EY, Gorak A. On the modeling and simulation of sour
gas absorption by aqueous amine solutions. Chem Eng Sci 2003;58:35718.
[50] Reid RC, Prausnitz JM, Poling BE. The properties of gases and liquids. 5th ed.
New York: McGraw-Hill; 2000.
[51] Danckwerts PV, Sharma MM. The absorption of carbon dioxide into solutions
of alkalis and amines. Chem Eng 1987:CE24480.
[52] Bolhr-Nordenkampf M, Friedl A, Koss U, Tork T. Modelling selective H2S
absorption and desorption in an aqueous MDEA-solution using a rate-based
non-equilibrium approach. Chem Eng Process 2004;43:70115.
[53] Billet R, Schultes M. Predicting mass transfer in packed columns. Chem Eng
Technol 1993;16:19.
[54] Krevelen DW, Hoftijzer PJ. Kinetics of gasliquid reactions. Part I. General
theory. Rec Trav Chim 1948;67:56386.
[55] Gabrielsen J, Michelsen ML, Stenby EH, Kontogeorgis GM. A model for
estimating CO2 solubility in aqueous alkanolamines. Ind Eng Chem Res
2005;44:334854.
[56] Saha AK, Bandyopadhyay SS, Biswas AK. Solubility and diffusivity of N2O and
CO2 in aqueous solutions of 2-amino-2-methyl-1-propanol. J Chem Eng Data
1993;38:7882.
[57] Puxty G, Rowland R, Attalla M. Describing CO2 mass transfer in amine/
ammonia mixtures no shuttle mechanism required. Energy Proc
2011;4:136976.
[58] Al-Baghli NA, Pruess SA, Yesavage VF, Selim MS. A rate-based model for the
design of gas absorbers for the removal of CO2 and H2S using aqueous
solutions of MEA and DEA. Fluid Phase Equilibria 2001;185:3143.
[59] Kvamsdal HM, Zakeri A, Einbu A, Chikukwa A, Hillestad M. A comparison of
different parameter correlation models and the validation of an MEA-based
absorber model. Energy Proc 2011;4:152633.
[60] Hoff KA, Juliussen O, Falk-Pedersen O, Svendsen HF. Modeling and experi-
mental study of carbon dioxide absorption in aqueous alkanolamine solutions
using a membrane contactor. Ind Eng Chem Res 2004;43:490821.
[61] Ghaemi A, Shahhosseini S, Maragheh MG. Nonequilibrium dynamic modeling
of carbon dioxide absorption by partially carbonated ammonia solutions.
Chem Eng J 2009;149:1107.
[62] Last W, Stichlmair J. Determination of mass transfer parameters by means of
chemical absorption. Chem Eng Technol 2002;25:38591.
566 I.P. Koronaki et al. / Renewable and Sustainable Energy Reviews 50 (2015) 547566
[63] Harun N, Douglas PL, Ricardez-Sandoval L, Croiset E. Dynamic simulation of
MEA absorption processes for CO2 capture from fossil fuel power plant. Energy
Proc 2011;1(4):147885.
[64] Lawal A, Wang M, Stephenson P, Koumpouras G, Yeung H. Dynamic modelling
and analysis of post-combustion CO2 chemical absorption process for coal-
red power plants. Fuel 2010;89:2791801.
[65] Astarita G, Savage DW, Bisio A. Gas treating with chemical solvents. New York:
Wiley; 1983.
[66] Jayarathna SA, Lie B, Melaaen MC. Amine based CO2 capture plant: dynamic
modeling and simulations. Int J Greenh Gas Control 2013;4:28290.
[67] Jamal A, Meisen A, Lim CJ. Kinetics of carbon dioxide absorption and
desorption in aqueous alkanolamine solutions using a novel hemispherical
contactor. Experimental results and parameter estimation. Chem Eng Sci
2006;61:6590603.
[68] Kent RL, Eisenberg B. Better data for amine treating. Hydrocarb Process
1976:8790.
[69] Zhang Y, Chen H, Chen C, Plaza JM, Dugas R, Rochelle GT. Rate-based process
modeling study of CO2 capture with aqueous monoethanolamine solution. Ind
Eng Chem Res 2009;48:923346.
[70] Lawal A, Wang M, Stephenson P, Koumpouras G, Yeung H. Dynamic modelling
and analysis of post-combustion CO2 chemical absorption process for coal-
red power plants. Fuel 2010;89:2791801.
[71] Mac Dowell N, Samsatli NJ, Shah N. Dynamic modelling and analysis of an
amine-based post-combustion CO2 capture absorption column. Int J Greenh
Gas Control 2013;12:24758.
[72] Liu GB, Yu KT, Yuan XG, Liu CJ, Guo QC. Simulations of chemical absorption in
pilot-scale and industrial-scale packed columns by computational mass
transfer. Chem Eng Sci 2006;61:651129.
[73] Kvamsdal HM, Rochelle GT. Effects of the temperature bulge in CO2 absorption
from ue gas by aqueous monoethanolamine. Ind Eng Chem Res
2008;47:86775.
[74] Zhang M, Guo Y. Rate based modeling of absorption and regeneration for CO2
capture by aqueous ammonia solution. Appl Energy 2013;111:14252.
[75] Qi G, Wang S, Yu H, Wardhaugh L, Feron P, Chen C. Development of a rate-
based model for CO2 absorption using aqueous NH3 in a packed column. Int J
Greenh Gas Control 2013;17:45061.
[76] Lawal A, Wang M, Stephenson P, Yeung H. Dynamic modelling of CO2
absorption for post combustion capture in coal-red power plants. Fuel
2009;88:245562.
[77] Aboudheir A, Tontiwachwuthikul P, Idem R. Rigorous model for predicting the
behavior of CO2 absorption into AMP in packed-bed absorption columns. Ind
Eng Chem Res 2006;45:25537.
[78] Nittaya T, Douglas PL, Croiset E, Ricardez-Sandoval LA. Dynamic modelling and
control of MEA absorption processes for CO2 capture from power plants. Fuel
2014;116:67291.
[79] Dugas RE, Rochelle GT. Modeling CO2 absorption into concentrated aqueous
monoethanolamine and piperazine. Chem Eng Sci 2011;66:52128.
[80] Saimpert M, Puxty G, Qureshi S, Wardhaugh L, Cousins A. A new rate based
absorber and desorber modelling tool. Chem Eng Sci 2013;96:1025.
[81] Gaspar J, Cormos AM. Dynamic modeling and absorption capacity assessment
of CO2 capture process. Int J Greenh Gas Control 2012;8:4555.
[82] Moioli S, Pellegrini LA. Improved rate-based modeling of the process of CO2
capture with PZ solution. Chem Eng Res Des; 2014.
[83] Plaza JM. Modeling of carbon dioxide absorption using aqueous monoetha-
nolamine. piperazine and promoted potassium carbonate [Ph.D. Thesis].
Austin: University of Texas; 2012.
[84] Sema T, Naami A, Liang Z, Idem R, Tontiwachwuthikul P, Shi H, et al. Analysis
of reaction kinetics of CO2 absorption into a novel reactive 4-diethylamino-2-
butanol solvent. Chem Eng Sci 2012;81:2519.
[85] Shi H, Liang Z, Sema T, Naami A, Usubharatana P, Idem R, Saiwan C,
Tontiwachwuthikul P. Solvent chemistry: NMR analysis and studies for
amine-CO2-H2O systems with VLE modeling for CO2 capture process. Carbon
Manage 2012;3:185200.