Advanced Process Modeling and Optimization of Amine-Based Carbon
Uploaded by
Ottors kebin philipAdvanced Process Modeling and Optimization of Amine-Based Carbon
Uploaded by
Ottors kebin philipGraduate Theses, Dissertations, and Problem Reports
2022
Advanced Process Modeling and Optimization of Amine-Based
Carbon Capture Process
Paul Jide Terhemba Akula
[email protected]
Follow this and additional works at: https://researchrepository.wvu.edu/etd
Part of the Other Chemical Engineering Commons, Thermodynamics Commons, and the Transport
Phenomena Commons
Recommended Citation
Akula, Paul Jide Terhemba, "Advanced Process Modeling and Optimization of Amine-Based Carbon
Capture Process" (2022). Graduate Theses, Dissertations, and Problem Reports. 11370.
https://researchrepository.wvu.edu/etd/11370
This Dissertation is protected by copyright and/or related rights. It has been brought to you by the The Research
Repository @ WVU with permission from the rights-holder(s). You are free to use this Dissertation in any way that is
permitted by the copyright and related rights legislation that applies to your use. For other uses you must obtain
permission from the rights-holder(s) directly, unless additional rights are indicated by a Creative Commons license
in the record and/ or on the work itself. This Dissertation has been accepted for inclusion in WVU Graduate Theses,
Dissertations, and Problem Reports collection by an authorized administrator of The Research Repository @ WVU.
For more information, please contact [email protected].
Advanced Process Modeling and Optimization of Amine-Based Carbon Capture Process
Paul Terhemba Jide Akula
Dissertation submitted to the
Benjamin M. Statler College of Engineering and Mineral Resources
at West Virginia University
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
in
Chemical Engineering
Debangsu Bhattacharyya, Ph.D., Chair
Richard Turton, Ph.D.
Stephen E. Zitney, Ph.D.
David S. Mebane, Ph.D.
John C. Eslick, Ph.D.
David C. Miller, Ph.D.
Department of Chemical and Biomedical Engineering
Morgantown, West Virginia
2022
Keywords: CO2 capture, parameter estimation, optimization, IDAES, symmetric electrolyte
NRTL model, excess enthalpy, monoethanolamine, reactive absorption, plate heat exchanger
Copyright 2022 Paul T. J. Akula
Abstract
Advanced Process Modeling and Optimization of Amine-Based Carbon Capture Process
Paul T. J. Akula
With the rise of carbondioxide (CO2) concentration in the atmosphere to more than 400 parts per
million (ppm), research efforts have been focused on achieving net-zero carbon emission
technologies. Post-combustion CO2 capture (PCC) is a key strategy in reducing CO2 emissions.
Amine-based CO2 capture is the baseline technology for retrofitting existing power stations.
However, the integration of amine-based PCC technology with power plants to reduce greenhouse
gas emissions incurs a high energy penalty, decreasing a powerplant’s efficiency by about 23
percentage points. Understanding the capture plant dynamics plays an important role in its
technical and economic performance. Rigorous models are critical to understanding how to address
the emission of CO2 from large point sources such as power plants and the techno-economic
performance of capture units, as the power plants increasingly cycle due to increased penetration
of renewables into the grid. These issues are very important because it addresses the costs of
achieving net-zero carbon systems in carbon capture-fitted plants. Modeling and simulation of
advanced multiscale process units have great potential in accelerating the commercialization of
carbon capture technologies.
The objective of this study is to develop high-fidelity process units that are computationally
tractable for amine-based CO2 capture process in an equation-oriented platform suitable for
optimization. These process units include gas-liquid contactors for absorption/stripping operation,
rich-lean heat exchanger, reboiler, and condenser. Process models of these units involve
development of rigorous sub-models for the thermodynamic and transport properties, and the
hydrodynamic and mass transfer sub-models of specific unit operations. Rigorous thermophysical
property models for challenging solvent systems in PCC such as mixed solvent electrolyte systems
and advanced mass transfer models are developed. A comprehensive description of the
thermodynamic framework for multi-electrolyte mixed solvent systems is presented, where the
parameter structure of the symmetric electrolyte-Non-Random Two Liquid (e-NRTL) model is
reformulated and a thermodynamically consistent and analytically derived formulation for the
excess enthalpy is developed from the e-NRTL model. The refined parameter structure of the e-
NRTL model removes a numerical singularity in the absence of ionic species and extends the
derived excess enthalpy expressions to non-electrolyte systems. The thermodynamic framework
is demonstrated for the MEA-H2O-CO2 case study using experimental data on thermodynamic
quantities for the binary MEA-H2O and the ternary MEA-H2O-CO2 systems. Also, multi-
component transport of molecular and ionic species is modeled using Maxwell-Stefan (MS)
transport equation and Nernst-Planck equation that includes the effect of electrostatic forces in the
two-film model for interface transfer in gas-liquid contactors. The process units are validated with
available data from the National Carbon Capture Center (NCCC) in Alabama, USA, and Norway’s
Technology Centre Mongstad (TCM) test facility for carbon capture technologies, along with
Wetted Wall Column data from the literature.
A plant-wide model of a reference monoethanolamine (MEA)-based PCC unit is developed and
used for steady-state optimization under part-load operations and variable capture rates using flue
gas similar to pulverized coal and natural gas-combined cycle power plants to minimize the energy
penalty and identify optimal operating conditions under part-load and variable capture operations.
The process unit models are implemented in the Institute for the Design of Advanced Energy
System (IDAES) computational platform. This provides carbon capture technology in an
integrated platform with access to advanced solver algorithms for the design and operation of
complex and interacting systems, such as the upstream power generation and the downstream post-
combustion carbon capture. Several approaches are developed for generating good initial guesses
for future state and algebraic variables. In addition, capabilities in IDAES for activating and
deactivating constraints are exploited to develop a sequential initialization strategy. IDAES
software platform is open-source and all codes developed as part of this work are publicly
available. Hence, researchers can use and/or modify the codes for designing and optimizing their
respective technologies. Models, algorithms, and codes developed as part this research can also be
used in academic institutions for teaching and research.
Table of Contents
Abstract ........................................................................................................................................... ii
Table of Contents ........................................................................................................................... iv
Dedication ...................................................................................................................................... ix
Acknowledgement .......................................................................................................................... x
Disclaimer ...................................................................................................................................... xi
List of Figures ............................................................................................................................... xii
List of Tables ................................................................................................................................ xv
List of Appendices ...................................................................................................................... xvii
Chapter 1 Introduction ............................................................................................................... 1
1.1 Research Motivation ........................................................................................................ 1
1.2 Rate-based models for PCC process ................................................................................ 2
1.3 Scope of Research ............................................................................................................ 4
Chapter 2 Thermodynamic Framework Development .............................................................. 6
2.1 Introduction ...................................................................................................................... 6
2.2 Equations of state ........................................................................................................... 10
2.2.1 Virial ............................................................................................................................ 10
2.2.2 E-NRTL ....................................................................................................................... 14
2.3 Analytical Excess Enthalpy ............................................................................................ 19
2.4 Chemical Equilibrium : Speciation calculation .............................................................. 22
iv
2.5 Vapor-Liquid Equilibrium............................................................................................. 23
2.6 Enthalpy Model .............................................................................................................. 25
2.7 Application to MEA-H2O-CO2 ..................................................................................... 25
2.7.1 MEA Chemical and Physical Equilibrium................................................................... 25
2.7.2 MEA Heat of Absorption ............................................................................................. 26
2.8 Parameter Estimation with Parmest ............................................................................... 27
2.9 Results and Discussions ................................................................................................. 29
2.9.1 H2O-MEA Binary System ........................................................................................... 29
2.9.2 MEA-H2O-CO2 Ternary system. ................................................................................. 33
2.10 Conclusion .................................................................................................................. 41
Chapter 3 Transport, Mass Transfer and Hydraulic Models ................................................... 43
3.1 Introduction .................................................................................................................... 43
3.2 Transport Properties ....................................................................................................... 45
3.2.1 Maxwell-Stefan(MS) Diffusivity and Binary Mass transfer Coefficient .................... 46
3.3 Hydraulic Models ........................................................................................................... 47
3.3.1 Liquid Holdup .............................................................................................................. 48
3.3.2 Flooding Velocity ........................................................................................................ 52
3.3.3 Pressure Drop Correlation............................................................................................ 53
3.4 Mass Transfer Correlations ............................................................................................ 54
3.4.1 Mass Transfer Coefficient............................................................................................ 54
v
3.4.2 Interfacial Area ............................................................................................................ 55
3.5 Analysis of packed column hydraulics and mass transfer models. ................................ 58
3.6 Conclusion...................................................................................................................... 59
Chapter 4 Interface Transfer Models ....................................................................................... 60
4.1 Introduction .................................................................................................................... 60
4.2 Enhancement Factor Model ........................................................................................... 62
4.3 Rigorous Two-Film Model............................................................................................. 64
4.3.1 Vapor film mass balance .............................................................................................. 65
4.3.2 Vapor film energy balance ........................................................................................... 66
4.3.3 Liquid film mass balance ............................................................................................. 67
4.3.4 Liquid film energy balance .......................................................................................... 69
4.3.5 Degrees of freedom analysis ........................................................................................ 70
4.3.6 Liquid film discretization ............................................................................................. 72
4.4 Conclusion...................................................................................................................... 74
Chapter 5 Unit Operation Models Development and Validation for MEA-Based CO2 capture
75
5.1 Packed Column Model ................................................................................................... 75
5.1.1 Gen 1- continuous differential contactor (CDC) packed column model ..................... 77
5.1.2 Gen 2- stage-wise contactor (SWC) packed column model ........................................ 84
5.2 Plate Heat Exchanger Model .......................................................................................... 89
vi
5.2.1 PHE Background ......................................................................................................... 89
5.2.2 PHE Mathematical model ............................................................................................ 91
5.2.3 PHE parameter estimation and validation................................................................... 94
5.3 Conclusion...................................................................................................................... 98
Chapter 6 Analysis and Optimization of CO2 Capture Process under Part-Load and Variable
Capture Operations ....................................................................................................................... 99
6.1 Introduction .................................................................................................................... 99
6.2 Configuration and operating range of the referenced MEA capture process .............. 102
6.3 Analysis of RL-Heat Exchanger temperature approach on the reboiler duty ............ 104
6.4 Effect of part load operation on the RL- Heat Exchanger temperature approach ........ 108
6.5 Optimal operating conditions at base load ................................................................... 109
6.6 Coupled part load and variable capture operation ........................................................ 112
6.7 Conclusion.................................................................................................................... 114
Chapter 7 Final Remarks and Future Work ........................................................................... 115
7.1 Final Remarks .............................................................................................................. 115
7.2 Recommendations for Future Work ............................................................................. 116
Appendix A : Virial equation of state ......................................................................................... 118
Appendix B : Excess Gibbs energy model ................................................................................. 119
Appendix C : Activity coefficient model .................................................................................... 122
Appendix D : Excess enthalpy derivation ................................................................................... 130
vii
Appendix E : MEA-H2O-CO2 Correlations for Phase and Chemical Equilibria ........................ 136
Appendix F : Enthalpy model .................................................................................................... 146
Appendix G: Transport properties .............................................................................................. 149
Appendix H: Experimental Data ................................................................................................. 153
References ................................................................................................................................... 157
viii
Dedication
To my wife, Chi, who stood strongly with me all through the years and to my two sons: Jayden and
Jesse and my family.
ix
Acknowledgement
I wish to express my profound gratitude to my advisor, Dr. Debangsu Bhattacharyya, for the
opportunity to be part of the IDAES project, and for all his advice, support, and guidance during
the course of this work. I would also like to thank Dr. John C. Eslick, Dr. David Mebane, Dr.
David Miller, Dr. Richard Turton, and Dr. Stephen Zitney for serving on my Advisory and
Examining Committee. I would also like express my appreciation to Linda Rogers who always sent
me reminders and ensured I met all the requirement for a graduate research assistant in the department.
I had the honor of working with an awesome group both past and present (Samson, Chirag, Anca,
Yifan, Paul, Ryan, Goutham, Pushpitha, Christian, Katie, Elijah, Parikshit Zhien, Katherine,
Chandra, Vivek, Ana, Angan, Opeyemi, Matthew, Pavitra, Samuel, Daniel, Sung, Harish, Ijiwole,
Nishant, Stephen, Poulomi, Quang,) during these years in Prof. Bhattacharyya’s Advanced Process
and Energy Systems Engineering Group. It was an unusual time in the pandemic, but we had time
to discuss and share ideas about our works, seminar presentations and life.
I also had the opportunity to work with many other competent professionals who assisted me in
the fulfillment of this dissertation. I would like to all the NETL team and the whole of IDAES
technical team.
Finally, I would like to acknowledge the financial support to this research. This work was
conducted as part of the Institute for the Design of Advanced Energy Systems (IDAES) with
support through the Simulation-Based Engineering, Crosscutting Research Program within the
U.S. Department of Energy’s Office of Fossil Energy and Carbon Management. I want to also
thank the management of the Petroleum Technology Development Fund (PTDF) for their
scholarship support.
x
Disclaimer
This paper was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor any agency thereof, nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for
the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed,
or represents that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does
not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States
Government or any agency thereof. The views and opinions of authors expressed herein do not
necessarily state or reflect those of the United States Government or any agency thereof.
xi
List of Figures
Figure 1: MEA Process flowsheet of some major process units .................................................... 3
Figure 2 : E-NRTL local composition using three types of cells ................................................. 14
Figure 3: Comparison of excess enthalpy prediction from e-NRTL model (mole fractions of solutes
set to zero in the MEA-H2O-CO2 system) with MEA-H2O excess enthalpy data at 25°C from
Touhara et al.(1981)63, using various parameters from the literature. ......................................... 31
Figure 4: Comparison of MEA-H2O phase envelope predictions to: (A) P-xy data at 60 °C from
Kim et al. (2008)58 (B) P-xy data at 90 °C from Tochigi et al. (1999)47 and (C) T-xy data at 101.325
kPa from Cai et al. (1996).49 ......................................................................................................... 32
Figure 5: Comparison of the natural logarithm of estimated and experimental values of CO2 partial
pressure as function of CO2 loading in MEA solution ................................................................. 36
Figure 6:Comparison of model predictions with experimental data for CO2 equilibrium pressure
over MEA-H2O-CO2 system ....................................................................................................... 36
Figure 7:Comparison of estimated and experimental values of differential heat of CO2 absorption
as function of CO2 loading in MEA solution ............................................................................... 38
Figure 8:Comparison of model predictions with differential CO2 heat of absorption data for 30 wt
% MEA solution. .......................................................................................................................... 39
Figure 9: Comparison of model results at 40 °C with NMR data from Jakobsen et al. (2005) for 30
wt% MEA solution and that of Hilliard(2008)39 and Böttinger (2008)64. .................................... 39
Figure 10:Comparison of experimental data (Weiland et al., 1997), CCSI44 and Plaza’s77 Aspen
model predictions to model predictions of specific heat capacity at 25 °C for 30 wt % and 40 wt
% MEA solution as function CO2 loading. .................................................................................. 40
xii
Figure 11: Comparison of excess enthalpy for tertiary MEA-H2O-CO2 at binary limiting condition
: (A) Aspen model35 at 80 C (B) This work. ................................................................................ 40
Figure 12 : Structured packing: (a) Mellapak 250Y (b) Simplified diagram of structured packing
....................................................................................................................................................... 44
Figure 13: Structured Packing Channel Geometry ....................................................................... 51
Figure 14: Comparison of hydraulic and mass transfer models in the pre-loading region ........... 58
Figure 15: Liquid film discretization options : (A) algorithm (B) plots ....................................... 73
Figure 16: Implementation of packed column models in IDAES: (A) generation 1 CDC model
(B) generation 2 SWC model. ....................................................................................................... 76
Figure 17: Packed bed initialization strategy ............................................................................... 79
Figure 18: Comparison of absorber temperature profiles between the model and the NCCC data
for (A) Case 3 (B) Case 2 with Hatta number and GM model for enhancement factor calculation.
....................................................................................................................................................... 81
Figure 19: Comparison of model predictions with (A) NCCC Stripper data (B) TCM 2015 baseline
data. ............................................................................................................................................... 83
Figure 20: Incidence matrix of the stage gas-liquid contactor (nfe=10, ncp=1)........................... 86
Figure 21 : Initialization strategy for gas-liquid contactor model: (A) Sequential solution of sub-
blocks (B) sequential solution of blocks ....................................................................................... 87
Figure 22: Stage model results : (A) stage height = 0.25 m (B) stage height = 0.5 m ................. 88
Figure 23: (a) Typical PHE layout (b) Chevron Plate (c) 4-pass-4-pass PHE showing sub-heat
exchangers in pass 2..................................................................................................................... 93
Figure 24: Parity plot for the PHE heat transferred ...................................................................... 95
xiii
Figure 25 : NCCC PHE predictions (A) Exit temperature (B) Channel temperature profile for Case
....................................................................................................................................................... 96
Figure 26: Fixed temperature approach analysis showing: (A) PHE and stripper with operating
conditions for two specific cases, (B) PHE area and reboiler duty with respect to hot end
temperature approach for Case I: low lean load scenario and Case II: high lean loading scenario.
..................................................................................................................................................... 107
Figure 27: RL-HX temperature approach under part load operation (60 -100% baseload) at 90%
CO2 capture. ............................................................................................................................... 109
Figure 28: Utility cost at 90% capture for varying L/G ratio at the base load of flue gas with low
CO2 inlet concentrations............................................................................................................. 111
Figure 29:Utility cost at 90% capture for varying L/G ratio at the base load of flue gas with high
CO2 inlet concentrations............................................................................................................. 111
Figure 30: Optimal operating cost for variable flue gas load and percentage capture for (A) low
CO2 inlet composition (B) high CO2 inlet composition.............................................................. 113
xiv
List of Tables
Table 1: Generic Second Virial Coefficient and its Temperature Derivatives ............................. 12
Table 2: Vapor phase thermodynamic properties from virial equation of state ........................... 13
Table 3: Symmetric e-NRTL parameter structure ........................................................................ 16
Table 4 :Reformulated Symmetric e-NRTL parameters ............................................................... 18
Table 5: Experimental data of MEA-H2O-CO2 ternary system used in the regression .............. 28
Table 6: MEA(1) + H2O(2) binary interaction parameters from selected publications ............... 29
Table 7:Selection criteria of MEA(1) + H2O(2) binary parameters ............................................. 30
Table 8: Estimated parameters for MEA-H2O-CO2 ternary system ............................................ 34
Table 9: Correlation coefficients of estimated parameters ........................................................... 35
Table 10: Transport properties ...................................................................................................... 45
Table 11: MS Diffusivities for MEA-CO2-H2O system ............................................................... 46
Table 12: Correlations for MS diffusivities for MEA-H2O-CO2 system ...................................... 47
Table 13 : Ionic conductivities 92 .................................................................................................. 47
Table 14: Packing Channel Dimensions ....................................................................................... 51
Table 15: Summary statistics of Billet and Schultes (1999)73 packing constants......................... 57
Table 16 : Two- film model variables........................................................................................... 71
Table 17: Two film model equations ............................................................................................ 72
Table 18:Comparison of absorber model predictions for CO2 capture percentage with NCCC data
....................................................................................................................................................... 80
Table 19: PHE Parameter estimation result .................................................................................. 94
Table 20: Plate heat exchanger thermal model ............................................................................. 97
Table 21: Typical Composition (mol%) of flue gas stream at TCM .......................................... 103
xv
Table 22: Design and operating parameters of the reference plant ............................................ 103
Table 23:Ranges of key process parameters for optimization studies ........................................ 104
Table 24 : Utility cost data (Obtained from Turton et al.133) ...................................................... 104
Table 25: Optimal L/G (mol/mol) ratios under variable flue gas load and CO2 removal target 112
Table 26 :PDH parameters .......................................................................................................... 121
Table 27: Dielectric parameters for solvent species ................................................................... 127
Table 28 :Symmetric E-NRTL parameter structure................................................................... 128
Table 29 : Default parameters35 settings for MEA-H2O-CO2 system ....................................... 129
Table 30: Fitted NRTL parameters ............................................................................................. 129
Table 31 : Extended Antoine Vapor pressure Parameters .......................................................... 136
Table 32 : Parameters of CO2 Henry's constant in water, Pa ...................................................... 138
Table 33 : Molar volume parameters74, m3/mol.......................................................................... 138
Table 34: Characteristic Volume Parameters (m3/kmol) ............................................................ 139
Table 35 : Physical Constants39 .................................................................................................. 142
Table 36 : Standard state properties and heat capacity parameters for MEA-H2O-CO2 system 144
Table 37: Watson heat of for vaporization parameters39 ............................................................ 147
Table 38 :Sutherland's parameters .............................................................................................. 149
Table 39: Liquid viscosity parameters 74 .................................................................................... 150
Table 40 : Parameters for pure component surface tension ........................................................ 150
Table 41: Surface tension parameters74 ...................................................................................... 151
Table 42: Diffusion Volumes in Fuller-Schettler-Giddings Correlation .................................... 151
Table 43: Ying and Eimer (2012) diffusivity parameters for CO2 in aqueous MEA mixture ... 152
Table 44 : Absorber flow streams -NCCC Steady-State data ..................................................... 153
xvi
Table 45: Absorber intercoolers between the packed beds -NCCC Steady-State data............... 154
Table 46: Stripper, reboiler and heat exchanger (HX) NCCC steady state data......................... 154
Table 47 : TCM absorber 2015 baseline data from Faramarzi et al. (2017). .............................. 155
Table 48: TCM Stripper 2015 baseline data from Faramarzi et al. (2017). ................................ 156
List of Appendices
Appendix A : Virial equation of state ......................................................................................... 118
Appendix B : Excess Gibbs energy model ................................................................................. 119
Appendix C : Activity coefficient model .................................................................................... 122
Appendix D : Excess enthalpy derivation ................................................................................... 130
Appendix E : MEA-H2O-CO2 Correlations for Phase and Chemical Equilibria ........................ 136
Appendix F : Enthalpy model .................................................................................................... 146
Appendix G: Transport properties .............................................................................................. 149
Appendix H: Experimental Data ................................................................................................. 153
xvii
Chapter 1 Introduction
Mitigating carbon emission is one of the greatest challenges for the human society now. While
decarbonization of the power sector using renewables is rapidly being deployed all over the world,
it is difficult to completely eliminate fossil-energy systems for power generation in the foreseeable
future due to intermittency of the renewables and lack of sufficient regional availability of these
renewable resources. Furthermore, there are other important sectors like petroleum,
petrochemicals, cement, fine chemical that release large amount of CO2 to the atmosphere. Thus,
post-combustion CO2 capture is anticipated to be one important technology among the mix of
technologies that can help to achieve large-scale decarbonization.
1.1 Research Motivation
With the rise of carbondioxide (CO2) concentration in the atmosphere to more than 400 parts per
million (ppm), research efforts have been focused on achieving net-zero carbon emission
technologies. Post-combustion CO2 capture (PCC) is a key strategy in reducing CO2 emissions
and amine-based PCC is the baseline technology for retrofitting existing power stations. However,
the integration of amine-based PCC technology with power plants to reduce greenhouse gas
emissions incurs a high energy penalty, decreasing a powerplant’s efficiency by about 23
percentage points1. Understanding the capture plant dynamics plays an important role in its
technical and economic performance. Rigorous models are critical to understanding how to address
the emission of CO2 from large point sources such as power plants and the techno-economic
performance of capture units, as the power plants increasingly cycle due to increased penetration
of renewables into the grid. These issues are very important because it addresses the costs of
achieving net-zero carbon systems in carbon capture-fitted plants. Modeling and simulation of
1
advanced multiscale process units have great potential in accelerating the commercialization of
carbon capture technologies2.
1.2 Rate-based models for PCC process
A typical amine-based CO2 capture plant is shown Figure 1. In the absorber, CO2 is absorbed from
the flue gas into the lean amine solution. The rich amine solution leaving the bottom of the absorber
is preheated in a rich/lean amine heat exchanger and sent to the stripper to desorb the dissolved
CO2. Low pressure (LP) steam extracted from the steam turbine of the power plant is typically
used in the reboiler of the stripper. The hot lean solvent leaving the stripper is recycled back to the
absorber after being cooled in the rich/lean crossover heat exchanger followed by a water cooler.
Reactions takes places in the liquid as CO2 is absorbed. As reaction kinetics of MEA-based systems
are very fast, these systems are often limited by mass transfer. Therefore, a rigorous model of mass
transfer is very important especially for the absorber that operates at considerably lower
temperature than the stripper thus leading to mass transfer limitations. As opposed to assuming
vapor-liquid equilibrium in the bulk, rate-based models account for mass transfer and heat transfer
limitations while assuming physical equilibrium to exist only at the vapor-liquid interface.
2
Figure 1: MEA Process flowsheet of some major process units
Many research studies have focused on developing detailed process models of CO2 capture plants,
with emphasis on the absorber and stripper, to capture the limiting mechanisms in the gas-liquid
contactors. In addition to differences in the properties and reaction kinetic models in these studies,
these models greatly differ in how mass transfer and heat transfer resistances are represented. Mass
and heat transfer resistances are described rigorously by the two-film model. The two-film model
assumes the presence of liquid and vapor films close to the liquid and vapor bulks while physical
equilibrium is assumed to exist at the interface of both films. A simplified model to describe the
mass transfer resistance in the gas and liquid films is by considering a linear driving force across
the gas-liquid interface, where the mass transfer enhancement due to chemical reactions is
quantified by an enhancement factor.3 The use of enhancement factor model transforms the
description of the gas-liquid interface phenomena from a set of differential equations to a set of
3
algebraic equations. This transforms the overall 2D column model ( radial and axial directions) to
a 1D model in the axial direction, thereby reducing the computation cost.
1.3 Scope of Research
The objective of this study is to develop high-fidelity process units that are computationally
tractable for amine-based CO2 capture process in an equation-oriented platform suitable for
optimization. The contributions of these works may be summarized as follows:
• reformulation of the parameter structure of the symmetric e-NRTL model to avoid numerical
singularities when electrolytes are absent,
• presentation of new analytical expressions of excess enthalpy applicable to both aqueous
electrolyte and non-electrolyte systems,
• a new mixing rule for Henry’s constant in the mixed solvent based on the reduction of the
Brelvi-O’Connell model,
• regressed standard-state parameters for MEA-CO2-H2O systems species along with binary
‘ion pair’-water parameters using a simplified solution chemistry
• a generic rate-based packed column model is developed and validated using the WWC data,
and data from NCCC and TCM pilot plants. The contactor model uses an enhancement factor
model that is superior to the Hatta number approximation and more accurate for a wide
range of operating conditions.
• a generic rate-based two-film model is developed for explicit computation of interphase
fluxes based on Maxwell-Stefan and Nernst-Planck diffusion equations.
• a generic single staged gas-liquid contactor model is developed using the two-film model
and forms the building block for a stage-wise packed column model.
4
• a novel algebraic plate heat exchanger model is developed and validated with the pilot plant
data and used for simulation and optimization of the plant-wide model
• the column models are implemented in the IDAES framework with a four-level initialization
strategy that aids convergence for different operating conditions.
• the impact of temperature approach in the rich-lean heat exchanger on reboiler duty under
steady-state part-load operation is accounted for by using a rigorous plate heat exchanger
model.
• Steady-state optimization under part-load condition and flexible capture is performed
considering high and low flue gas compositions that are similar to those from the pulverized-
coal and NGCC power plants, respectively.
The remaining portion of this dissertation is organized as follows: Chapter 2 describes the
thermodynamic framework development, while chapter 3 presents the model development for
transport properties, mass transfer and hydraulic correlations for the design of a two-phase
countercurrent-flow packed column for reactive absorption or desorption. In Chapter 4, the
development of a rigorous two-film model for mass and energy fluxes across gas-liquid interfaces
for reactive absorption or desorption process is presented along with a generic enhancement factor
model. Chapter 5 describes the development and validation of high-fidelity process units for
amine-based CO2 capture process and chapter 6 presents the optimization of the capture unit. Final
remarks and future works are discussed in chapter 7.
5
Chapter 2 Thermodynamic Framework Development
Accurate thermodynamic properties of electrolyte systems are critical for the design and operation
of many chemical processes. In this chapter, a comprehensive description of the thermodynamic
framework for multi-electrolyte mixed solvent systems is presented, where the parameter structure
of the symmetric electrolyte-Non-Random Two Liquid (e-NRTL) model is reformulated and a
thermodynamically consistent and analytically derived formulation for the excess enthalpy is
developed from the e-NRTL model. The vapor phase properties using the virial equation of states
are also presented. The thermodynamic framework is demonstrated for the MEA-H2O-CO2 case
study using experimental data on thermodynamic quantities for the binary MEA-H2O system and
the ternary MEA-H2O-CO2 system. The contributions of this work include: (i) reformulation of
the parameter structure of the symmetric e-NRTL model to avoid numerical singularities when
electrolytes are absent, (ii) presentation of new analytical expressions of excess enthalpy
applicable to both aqueous electrolyte and non-electrolyte systems, (iii) a new mixing rule for
Henry’s constant in the mixed solvent based on the reduction of the Brelvi-O’Connell model, and
(iv) regressed standard-state parameters for protonated MEA ( MEAH + ) and MEA carbamate (
MEACOO− ) species along with their binary ‘ion pair’-water parameters using a simplified
solution chemistry.
2.1 Introduction
The effect of composition and temperature changes on the thermodynamic properties of
electrolyte systems is of prime importance for the design and operation of many separation tasks
utilizing phase-contacting operations such as absorption, distillation, extraction, and stripping.4, 5
The incorporation of a rigorous thermodynamic framework for electrolyte systems in the IDAES6
6
computational platform to support the design and optimization of aqueous electrolytes systems is
the motivation for this work. One such application is the post-combustion CO2 capture using
electrolyte systems that includes various type of aqueous amines.7-9 The enthalpy of CO2
absorption affects the temperature profile in the towers that affects chemical speciation in the liquid
phase, vapor-liquid equilibrium (VLE) as well as the mass and heat transfer rates, all of which, in
turn, affect the enthalpy of the system.10, 11, 12 Therefore, VLE and enthalpy of CO2 absorption play
a key role in determining the capture rate and energy requirement for CO2 capture.13 Furthermore,
operating conditions of these electrolyte systems can vary widely as the tower flue gas flowrate,
and the solvent conditions (loading, temperature, flowrate) change due to load variation in the
host power plant.14, 15
For optimal design and operation of these systems, a comprehensive
treatment of chemical equilibrium to account for speciation, phase equilibrium for molecular
solutes, activity coefficients for the non-ideal solvent mixture, and calorimetric properties is
desired.
The departure of electrolyte solutions from ideal behavior can be represented by excess Gibbs
free energy models.16 Among the several semi-empirical excess Gibbs energy models for
electrolyte systems (Pitzer model17, extended UNIQUAC model18, OLI-MSE model19, e-NRTL
model20, 21-23 etc.), the e-NRTL model proposed by Chen and Evans (1986)23 and extended by
Mock et al. (1986)21 for mixed solvents has gained wide industrial recognition and is applied to
almost all types of electrolyte systems.24 The e-NRTL model is based on the like-ion repulsion and
local electroneutrality assumptions and represents the sum of two terms with additional terms as
needed: the first term accounts for the long-range electrostatic forces between ions, and the second
term for short-range forces between all species. Song and Chen (2009)22 presented a new
symmetric electrolyte NRTL model which differs from the original e-NRTL in the choice of
7
reference state for ions and the use of the symmetric Pitzer-Debye-Hückel term. This led to the
elimination of the ionic charge fraction quantities from the excess Gibbs energy and activity
coefficient expressions. Hence, the assumptions of constant ionic charge fractions quantities were
no longer considered while calculating activity coefficients as done previously in the original e-
NRTL. This new symmetric e-NRTL model is quite similar to the original e-NRTL for mixed
solvents when the reference state for the ions is transformed back to the infinite-dilution aqueous
solution.
The e-NRTL model provides analytical expressions for calculating excess Gibbs free energy as
well as expressions for calculating activity coefficients obtained via the Gibbs-Duhem equation
which enables interpolation/extrapolation of experimental data with respect to composition.
However, analytical expressions of excess enthalpy from the e-NRTL model to enable
interpolation/extrapolation of experimental data with respect to temperature are not available in
the public domain. Current approaches for calculating the excess enthalpy from the e-NRTL model
are either proprietary (i.e., the implementation in a specific software is not clearly mentioned in
the public domain) or done numerically.25, 26 Hossain et al. (2016)25 calculated the excess enthalpy
for single electrolyte systems at 298.15 K temperature in Aspen Plus by numerically evaluating
the long-range contribution while for the short-term contribution, the complete analytical
expressions were not provided. Hessen et. al. (2010)27 implemented the refined e-NRTL model in
the RGrad28 software platform and applied it to regress the e-NRTL model parameters for
alkanolmines-H2O-CO2 systems. The authors noted that the novel symbolic gradient calculations
in RGrad were preferred to numerical differentiation that resulted in a less accurate and unstable
code with increased computational effort. Løvefall (2008)28 identified potential pitfalls in symbolic
gradient calculations that can lead to inaccuracies when applied to thermodynamic models.
8
Numerical evaluation of the excess enthalpy introduces numerical approximations that can lead to
inaccuracies and increased computation expense when applied to parameter estimation or
optimization problems.
Theoretically, the e-NRTL model reduces to the NRTL model in the absence of electrolytes. In
practice, this would lead to numerical singularities unless the parameter structure of the e-NRTL
model is reformulated. Reformulating the e-NRTL parameter structure allows the derived excess
enthalpy expressions from the e-NRTL model to reduce to that of the NRTL model of Renon and
Prausnitz (1968)29 in the absence of electrolytes. Since the excess enthalpy expressions are derived
from the e-NRTL model considering true species, computing “apparent” excess enthalpy when the
experimental thermodynamic data are reported in terms of apparent components requires
transformation.30, 31
Experimental enthalpy and VLE data provide valuable information about the temperature
dependency of the activity coefficients.32, 33In this work, the symmetric e-NRTL model is applied
to the monoethanolamine (MEA)-H2O-CO2 system- mainly because there is a large amount of data
in the public domain for this system that helps to validate the proposed approach. In addition, MEA
is one of the most widely studied systems thus making it easier to compare the results from this
work with the published literature. Most of the work that exists in the public domain where the e-
NRTL model was applied to thermodynamic modeling of amine-based capture, MEA-based
system in particular, used Aspen Plus software that offers readily available capabilities for
modeling electrolyte systems along with the implementation of the e-NRTL model under its
thermodynamic model library.27, 34-39 Hilliard (2008)39 used the e-NRTL model in Aspen Plus for
aqueous alkanolamine solutions including MEA and estimated several parameters using the VLE,
enthalpy of absorption, heat capacity and NMR speciation data. Zhang et. al. (2011)35 regressed
9
MEA-H2O binary parameters in Aspen Plus using VLE, excess enthalpy and heat capacity data
for the MEA-H2O binary system while evaluating excess enthalpy for the MEA-H2O-CO2 system
numerically. Composition and temperature dependence of activity coefficients can be estimated
by simultaneous regression of binary interaction parameters considering data for different
thermodynamic quantities.33, 40
2.2 Equations of state
2.2.1 Virial
In this work, a suite of thermodynamic property package for the vapor phase is developed based
on the B-truncated (second virial coefficient) volume explicit virial equation of state (VEOS). The
B-truncated VEOS has the numerical advantage of having a single root and adequately represents
the vapor phase at low to moderate pressure or where the vapor phase reduced density is less than
one-half.32 In addition, the thermodynamic properties from the VEOS form the bases for
computing the thermodynamic properties for the molecular species (Henry components/solvents)
in the liquid phase. The VEOS representation for pure species or gas mixtures is given by the Eq.
(2.1),
BP
Z = 1+ (2.1)
RT
where Z is the compressibility factor, B is the generalized second virial coefficient which is
substance dependent and a function of temperature. B is defined for both pure species and
mixtures as given by Eq. (2.2)
Bii : Pure Species
B=
yi y j Bij : Mixtures
i j
(2.2)
10
The generalized second virial coefficient is adequately represented by the Abbort16 equations as
shown in Table 1 along with its temperature derivatives. The like molecular terms ( i = j ) in Table
1 indicate pure components terms, and the unlike molecular terms ( i j ) indicate interaction
between different molecules for extension to mixtures. The cross coefficients are evaluated by
using the combining rules shown in the second column of Table 1. Alternatively, a gas mixture
may be treated as a pure component (pseudocritical component) by using the pseudocritical rules
given in the last column of Table 1. The complete suite of vapor phase thermodynamic properties
is based on heat capacity data of an ideal gas and the virial equation of state as summarized in
Table 2. The detailed expressions for the temperature derivatives of the mixture virial coefficient
term and of the residual terms in Table 2 are provided Eqs A1-A4 of Appendix A.
The isobaric heat capacity ( C igp,i ), molar enthalpy ( H iig ), and molar entropy( Siig ) of an ideal gas
are described in Eqs. (2.3) - (2.5),
T
C igp,i (T , P) = ( A + BT + CT + DT 3 )dT (2.3)
2
Tref
T
H iig (T , P) = H i ,ref ( Pref , Tref ) + C (2.4)
ig
p ,i dT
Tref
T
C igp ,i P
S (T , P) = Si ,ref ( Pref , Tref ) + dT − R ln (2.5)
ig
i
T P
Tref ref
11
Table 1: Generic Second Virial Coefficient and its Temperature Derivatives
Definitions Combining Rules for Cross Pseudocritical Parameters
Coefficients
ij = 0.5 (i + j ) p = yii
(B + ij Bij1 )
RTc ,ij
Bij = 0
ij i
Pc ,ij
dBij R dBij0 dBij1 Z c ,ij = 0.5 ( Z c ,i + Z c , j ) Tc , p = yiTc ,i
= + i
dT Pc ,ij dTr ,ij dTr ,ij
Vc ,ij = 0.125 (Vc1/3
,i + Vc , j )
Pp = yi Pc ,i
3
0.422 1/3
Bij0 = 0.083 −
Tr1.6
,ij
i
Tc ,ij = (1 − ij )(Tc ,iTc , j )
1/2
0.172 T
Bij1 = 0.139 − Tr , p =
Tr4.2
,ij Tc , p
dBij0 0.6752 Z c ,ij RTc ,ij P
= Pc ,ij = Pr , p =
dTr ,ij Tr2.6 Vc ,ij Pc , p
,ij
dBij1 0.7224
=
dTr ,ij Tr5.2
,ij
d dBij0 −1.7552
=
dTr ,ij dTr ,ij Tr3.6
,ij
d dBij1 −3.75648
=
dTr ,ij dTr ,ij Tr6.2
,ij
12
Table 2: Vapor phase thermodynamic properties from virial equation of state
Property Description Residual Term
Pure Property
Log Fugacity ln i = ( Pr Tr ) ( B0 + i B1 )
Fugacity 𝑓𝑖 = 𝜙𝑖 𝑃
Molar Volume Vi = Z i RT / P
Molar Enthalpy H i = H iig + H iR H iR = RTc Pc B 0 − Tr dB 0 dTr + ( B1 − Tr dB1 dTr )
Molar Entropy Si = Siig + SiR SiR = RPr ( dB 0 dTr + dB1 dTr )
Molar Gibbs Energy Gi = H i − TSi
Molar Internal Energy U i = H i − PVi
Mixture properties
Molar volume V = ZRT / P
Molar Enthalpy H = i yi H iig + H R H R = PT ( B T − dB dT )
Molar Entropy H = i yi Siig + S R S R = − P dB dT
Molar Gibbs Energy G = H − TS
Molar Internal Energy U = H − PV
Isobaric molar heat C p = i C pig,i + C pR C pR = − PT Tc2 ( d 2 B dTr2 )
capacity
Isochoric molar heat CV = C p − C* C * = 1 R ( R + P dB dT )
2
capacity
Partial Molar Properties
Log fugacity coefficient (
ln ˆj = P RT 2 i yi Bij − B )
Partial molar Gibbs G j = H igj − TS igj + RT ln y j + G Rj G Rj = RT ln ˆj
energy
Partial molar entropy S j = − ( dG dT )
P, x
Partial molar enthalpy H j = G j + TS j
Partial molar volume (
V j = RT P ln ˆj + 1 )
Partial molar internal U j = H j − PV j
energy
13
2.2.2 E-NRTL
The e-NRTL model for a multicomponent electrolyte system of mixed solvents is described using
the following sets:
S s = set of solvent species; s = solvents
Sc = set of cationic species; c = cations,
S a = set of anionic species ; a = anions,
S m = set of molecular species; m = molecules
S ma = S m S a
S mc = S m Sc
Sca = Sc S a
S mac = S m Sca
The e-NRTL model is based on the local composition concept using three types of cells as shown
in Figure 2. One cell type consists of a solvent molecule at the center with the local
electroneutrality assumption determining the distribution of solvent molecules, anions, and cations
in the immediate neighborhood.23
Figure 2 : E-NRTL local composition using three types of cells
The other two cell types have either an anion or cation at the center and an immediate
neighborhood consisting of solvent molecules and oppositely charged ions as determined by the
like-ion repulsion assumption.
14
The symmetric e-NRTL model22 is written as a sum of two contributions combining the symmetric
Pitzer-Debye-Huckel (PDH) model for the long-range (LR) ion-ion electrostatic interactions with
the symmetric NRTL model for short-range (SR) interactions based on the local composition
framework is defined by Eq. (2.6) ,
G**,ex G**,ex, SR G*,ex, LR G ex, Born
= + + (2.6)
RT RT RT RT
where G*,ex , LR / RT is the unsymmetric long-range contribution, G**,ex , SR / RT is the unsymmetric
short-range contribution based on the infinite dilution aqueous solution reference state. The Born
term is added to convert reference conditions for the long-range contribution from infinite dilution
in mixed solvent system to aqueous solution by capturing the difference between the dielectric
constant of water and the solvent mixture. For brevity, details of each term on the right hand-side
of Eq. (2.6) are presented in Appendix B (Eq B1-B10).
Activity coefficients are used to represent liquid-phase non-ideality for predicting phase
equilibrium behavior of molecular species, liquid phase speciation and liquid enthalpy. Aside the
canonical variables (T, P, x) of Gibbs free energy, the activity coefficients also depend on the
choice of reference state41 as described in Eqs. (2.7) and (2.8) (see Eqs and ),
1 nG ex
ln ˆi = (2.7)
RT ni T , P ,n j i
ln i* = ln i*, SR + ln i, PDH , i = any solute (ref : mixed-solvent)
**
ln ˆi = ln i = ln i **, SR
+ ln i , PDH
+ ln i , i = any solute (ref : aqueous solution) (2.8)
Born
ln = ln + ln
SR , PDH
, i = any solvent
i i i
Where ˆi is the activity coefficient of any species, i* is the activity coefficients of solutes based
on infinite-dilution mixed-solvent reference state, i** is the activity coefficients of solutes based
15
on infinite-dilution aqueous solution reference state, and i is the activity coefficients of solvent
species. For conciseness, the expressions for the activity coefficients are presented in Appendix
C (Eqs C1 -C13 )
The parameter structure for the symmetric e-NRTL model is as shown in Table 3. Rows 1-6 of
Table 3 show the three types of binary interaction parameters: molecule-molecule ( mm, mm ),
molecule-electrolyte ( m, ca or ca, m ), and electrolyte-electrolyte ( ca, ca or ca, ca ) parameters.
The non-randomness ( ij ) factors in rows 1-6 are commonly fixed to 0.2. The energy interaction
parameters are weak but well-behaved functions of temperature where the molecule-molecule (
mm , mm ), and molecule-electrolyte ( m,ca , ca ,m ) energy interaction parameters are defined as
functions of temperature as shown in rows 1-4 and the electrolyte-electrolyte interaction
parameters are set to zero23 ( ca ,ca = ca,ca = Eca,ca = 0 ). The Boltzmann kind factors in rows 1-6
of Table 3 are readily calculated from the relation, Gij = exp(−ij ij ) .
Table 3: Symmetric e-NRTL parameter structure
, , G,
Row Interactions
Nonrandomness Energy interaction Boltzmann kind factors
1 m-m' mm mm = A mm + Bmm /T Gmm = exp( − mm mm )
2 m'-m mm mm = A mm + Bmm /T Gmm = exp( − mm mm )
3 ca-m ca ,m ca ,m = Cca ,m + Dca ,m /T Gca ,m = exp(− ca ,m ca ,m )
4 m-ca m ,ca m,ca = Cm,ca + Dm,ca /T Gm,ca = exp(− m,ca m,ca )
5 ca-ca' ca ,ca ca ,ca = Eca ,ca Gca ,ca = exp(− ca ,ca ca ,ca )
16
6 ca-c'a ca ,ca ca ,ca = Eca ,ca Gca ,ca = exp(− ca ,ca ca ,ca )
7 cm cm = Ya ca ,m cm = − ln ( Gcm ) /cm Gcm = Y G a ca , m
a aSa
8 am am = Yc ca ,m am = − ln ( Gam ) / am Gam = Y G c ca , m
c cSc
9 mc mc = Ya ca ,m mc = − ln ( Gmc ) /mc Gmc = Y G
aS a
a m , ca
a
10 ma ma = Yc ca ,m ma = − ln ( Gma ) /ma Gma = YcGm ,ca
c cSc
11 ca ca = Yc ca ,ca ca = − ln ( Gca ) /ca Gca = YcGca ,ca
c c
12 ac ac = Ya ca ,ca ac = − ln ( Gac ) / ac Gac = YaGca ,ca
a a
Parameters in rows 7-12 of Table 1 arise from the like-ion and electroneutrality assumptions which
create three types of cells depending on the central species: a central molecule, a central anion,
and a central cation. The non-randomness and Boltzmann kind factors parameters in rows 7-12
of Table 1 are calculated via simple mixing rules using the cationic charge composition fraction,
Yc and the anionic charge composition fraction, Ya defined in (2.9)
Xc Xa
Yc = , Ya = (2.9)
X c
cSc
X a
aSa
where X i is the effective mole fraction of species i as define in Eq. C5 of Appendix C.
In the absence of electrolytes, the ionic charge fractions ( Yc , Ya ) and the Boltzmann kind factors
( Gcm , Gam , Gmc , Gma , Gca , Gac ) become zero (technically, the ionic charge fractions also become
undefined). Hence, the energy interaction terms ( cm , am , mc , ma , ca , ac ) cannot be evaluated. In
this work, (2.9) is deactivated in the absence of electrolytes and the ionic charge fractions are set
17
to zero, while the expressions for the Boltzmann kind factors and non-randomness parameters are
reformulated using the ionic charge fractions as shown in Table 4 to remove these singularities.
Thus, in the absence of electrolytes, parameters in rows 7-12 of Table 4 are calculated as follows:
ij = Gij = 1, ij = 0 while in the presence of electrolyte, (2.9) is reactivated and the original
expressions are obtained as Y = Y
aSa
a
cSc
c = 1 and electroneutrality is adhered to in the electrolyte.
Table 4 :Reformulated Symmetric e-NRTL parameters
, G,
Row Interactions
Nonrandomness Boltzmann kind factors
7 cm cm = Ya ca ,m + 1 − Ya Gcm = Y G a ca , m + 1 − Ya
a aS a aSa aSa
8 am am = Yc ca ,m + 1 − Yc Gam = Y G c ca , m + 1 − Yc
c cSc cSc cSc
9 mc mc = Ya ca ,m + 1 − Ya Gmc = Y G a m , ca + 1 − Ya
a aS a aSa aSa
10 ma ma = Yc ca ,m + 1 − Yc Gma = Yc Gm ,ca + 1 − Yc
c cSc cSc cSc
11 ca ca = Yc ca ,ca + 1 − Yc Gca = YcGca ,ca + 1 − Yc
c c c c
12 ac ac = Ya ca ,ca + 1 − Ya Gac = YaGca ,ca + 1 − Ya
a a a a
18
2.3 Analytical Excess Enthalpy
For electrolyte systems, the compositions of the true species are connected to the apparent
components via material balances and chemical equilibrium.30, 31 In this work, the excess enthalpy
expressions are first developed in the true-species space and then projected to the model space
spanned by apparent-components following the procedure of Copeman et. al. (1982).30 This
projection is equivalent to capturing the change in true species composition that occurs due to the
temperature dependence of equilibrium constants. In the true species space, analytical expressions
for excess enthalpy are derived via the Gibbs-Helmholtz relationship by taking the partial
derivative of the excess Gibbs energy expression based on the un-symmetric aqueous phase
infinite dilution reference state (see Eq. (2.6)) as shown in Eq.(2.10) ,
(G**,ex / RT )
H = − RT
ex 2
(2.10)
T P, x
This implies taking the temperature derivative of the excess Gibbs energy expressions at constant
pressure and composition for the short-range (SR), long-range (LR), and born term contribution
are shown in Eq. (2.11),
(G**,ex , SR / RT ) (G (*,ex , LR / RT ) (G ex , Born / RT )
H ex = − RT 2 + + (2.11)
T T T P, x
P , x P , x
The contribution of the Born term to the excess enthalpy is given in Eq. (2.12) ,
n0,i H i + H apparent − n0,i H i + H apparent − n0,CO2 H CO
L hex L hex V
iSm Final iSm Initial
2
H abs = (2.12)
n0,CO2
Where rB is the born radius, e is the elementary electron charge, o is the vacuum permittivity,
k is the Boltzmann constant, Z i is the absolute charge of ionic species, s is the dielectric
19
constant of mixed solvent, H2O is the dielectric constant of pure water, Aj , Cj are the parameters
for the dielectric constant of solvent j as defined in Eq C14, wj is the solute-free mass fraction
as defined in Eq. B8.
The long-range contribution to the excess enthalpy is given by Eq. (2.13),
(G*,ex , LR / RT ) Ax
= − x ln (1 + I 1/2
x )
4I
(2.13)
T P, x T P , x
where I x is the ionic strength as defined in Eq B5, is the closest approach parameter and the
temperature derivative of the Debye-Huckel parameter on mole fraction basis, A / T P , x is
defined by Eqs D8-D12 in Appendix D.
The short-range contribution to the excess enthalpy is given by Eq. (2.14) ,
(G **,ex , SR / RT ) (G ex , SR / RT ) ln cSR,H,2O
= − xc
T P,x T P,x c Sc T P , x
(2.14)
ln aSR,H,2O ln hSR,H,2O
− xa − xh
a Sa T P , x h ( Sm − Ss ) T P , x
Where the temperature derivative of the infinite dilution activity coefficient of the solutes,
ln iSR ,
,H 2 O / T P , x , based on Eq C9 in Appendix C is given by Eqs D24-D26 in Appendix D. The
contribution to the excess enthalpy from the local interactions of the symmetric excess Gibbs
energy comes from three terms : m where a molecular component is at the center, c where a
cationic species is at the center, and a where an anionic species is at the center as shown in Eq.
(2.15)
(G ex , SR / RT )
T = m + c + a (2.15)
P, x
20
These terms ( m , c , a ) are defined in Eqs (2.16) -(2.18),
G
X i Gim im + im Gim X i Gim im X i im
iS T T iSmac iSmac T
m = X m mac − (2.16)
m Sm X i Gim
2
iSmac X i Gim
iSmac
G
X i Gic ic + ic Gic X i Gic ic X i ic
iS T T iSma iSma T
c = X c ma − (2.17)
c Sc X i Gic
2
iSma X i Gic
iSma
Gia
X i Gia ia + ia Gia
i ia ia i
X G X
iS T T T
iSmc iSmc
a = X a mc − (2.18)
a Sa X iGia
2
iSmc X i Gia
iSmc
The temperature derivative of the Boltzmann kind factors( Gim / T , Gia / T , Gic / T ) are
defined in Eq. (2.19) .The explicit expressions are given in Eq. D14-D16 of Appendix D,
Gij ij
= − ij Gij , j = m, a, c (2.19)
T T
The temperature derivative of the energy interaction terms ( im / T , ia / T , ic / T ) are
defined in Eqs. (2.20) - (2.22),
B
− im2 if i = m
T
im 1
2 a ia , m ia , m ia , m
T = − YD G if i = c (2.20)
P, x imGimT aSa
1
2 c ci , m ci , m ci , m
− YD G if i = a
imGimT cSc
21
1
ic − Y D i ,ca Gi ,ca if i = m
T = ic GicT
2 a i , ca
aSa (2.21)
P, x
0 if i = a
1
ia − Y D i ,ca Gi ,ca if i = m
T = ia GiaT
2 c i , ca
cSc (2.22)
P,x
0 if i = c
In the absence of electrolytes, the excess enthalpy expressions reduce to the m term. The excess
enthalpy can be calculated from the true-species plane to the apparent-component plane30 as given
by Eq. (2.23),
ex
H apparent = H ex n − RN ε ( ln K / (1/ T ) )
iSmac
i
T T
0 (2.23)
where ni is number of moles for true species i , N0 n0
is a vector of the apparent components’
moles before they react, K m
is a vector of the natural logarithm of the chemical equilibrium
mn0
constants, ε is a matrix of the m reactions with elements as defined in Eq. (2.24) with the
restriction of a non-zero row,
1 if j is the reference component in reaction i
ij = (2.24)
0 otherwise
The expressions for the partial derivative of the equilibrium constant of reaction j ,
ln K j / (1/ T ) are given in Eqs E33-E42 of Appendix E.
P, x
2.4 Chemical Equilibrium : Speciation calculation
The aqueous phase speciation model is determined from Eq. (2.25),
N = AN 0
A = J + vε (2.25)
ln K = vT ln a
22
nm
Where N n
is a vector of the moles of the true species present at equilibrium, v is the
matrix of stoichiometric coefficients, a n
is a vector of the natural logarithm of activities of all
T
the true species, J = I n0
0n0 ( n −n0 ) where I n0 is the identity matrix30. The equilibrium constants
are computed from Eq. (2.26),
− RT ln K j = G j =
iSmac
ij Gi , j 1, 2,..., m (2.26)
where the standard state Gibbs free energy of the species are denoted by Gi .
2.5 Vapor-Liquid Equilibrium
Phase equilibrium describing the distribution of molecular species between the vapor and liquid
phases is given by Eq.(2.27) ,
fˆi V = fˆi L (2.27)
where fˆi V and fˆi L are the fugacity coefficient of species i in the vapor and liquid phase mixture
respectively. Using the activity coefficient model, the liquid phase fugacity coefficient of species
i is defined by Eq. (2.28),
fˆi L = fˆi ˆi xi (2.28)
where fˆi is the standard state fugacity species i and ˆi is the activity coefficient of species i
related to the chosen standard state fugacity and mole fraction of species i . For solvents, the pure
component standard fugacity is commonly used as shown in Eq. (2.29),
fˆi = fi ˆi = i ( as xi → 1, i → 1) (2.29)
where i is the symmetric activity coefficient, and f i is the pure component fugacity coefficient
as defined in Eq. E24. For supercritical components, the infinite dilution standard state fugacity
23
in a hypothetical state (*) at unit mole fraction is chosen. This corresponds to the extrapolation of
Henry’s law from an infinite dilute state as shown in Eq. (2.30),
fˆi = H i ,mix ˆi = i*,mix ( as xi → 0, i*,mix → 1) (2.30)
where i*,mix is the unsymmetric activity coefficient of solute i in the mixed solvent, H i ,mix is the
Henry’s constant in the mixed solvent. The Henry’s constant in the mixed solvent is calculated
based on those in the pure solvents as given in Eq. (2.31) ,
H H
ln i ,mix = ws ln is , (2.31)
i ,mix sSs is
where H is is the Henry’s constant of molecular solute i in pure solvent s as given in eq D6, is
is the infinite dilution activity coefficient of molecular solute i in solvent s , i,mix is infinite
dilution activity coefficient of molecular solute i in the mixed solvent. The weighting factor,
ws is defined as given in Eq. (2.32),
2/3
xs V
ws = s S s , sj = i, j (2.32)
xssj
jS s
V
i,s
where xs is the mole fraction of the solvent on solute free- basis, Vi, s is the partial molar volume
of molecular solute i at infinite dilution in solvent s and is obtained from the Brelvi-O’Connell
model42 (see Eq. E8).
24
2.6 Enthalpy Model
The enthalpy model for the electrolyte mixture in the true-species plane is given by Eq. (2.33)
H L = xi H iL + xi H j ,aq + H ex (2.33)
iS s j{ solutes }
where H iL is the molar enthalpy of solvent i as defined in Eqs F2-F5 of Appendix E, H j ,aq is the
molar enthalpy of solute j (molecular or ionic) in aqueous-phase infinite dilution and are defined
in Eqs F6 -F7.
2.7 Application to MEA-H2O-CO2
This section demonstrates the application of the thermodynamic framework to MEA-H2O-CO2
system for the regression of standard state and e-NRTL binary interaction parameters using vapor-
liquid equilibrium and heat of absorption data.
2.7.1 MEA Chemical and Physical Equilibrium
The aqueous phase reversible reactions for the MEA-H2O- CO2 system are modeled with similar
assumptions in Morgan et al. (2017)36:
R1: 2MEA + CO2 ⎯
K1
→ MEAH + + MEACOO−
R 2 : MEA + CO2 + H 2O ⎯→
K2
MEAH + + HCO3−
Morgan et al (2017) have shown that these two reactions represent the chemistry of the CO2 loaded
mono-ethanolamine solution adequately. Using CO2 as the reference component in this reduced
set of reactions, the projector term (second term in Eq. (2.23)) can be written as shown in Eq. after
some algebraic manipulation,
RNT0 εT ( ln K / (1/ T ) ) = R nMEACOO− ln K1 / (1/ T ) + nHCO− ln K 2 / (1/ T ) (2.34)
3
25
The matrices N, N 0 , J , v, ε considering CO2 as the reference component in the reduced solution
chemistry for the MEA-H2O- CO2 system are given as follows:
nCO2 1 0 0 −1 −1
0 −2 −1
nMEA 1 0
n0,CO2 nH O
, J = 0 0 1 0 −1 1 0 0
N 0 = n0,MEA , N = , ν = , ε =
2
nHCO− 0 0 0 0 1 2 0 0
n0,H O
2
3
0 0 0 1 1
nMEAH+
n 0 0 0 1 0
MEACOO−
For brevity, the standard state Gibbs free energy of the species, Gi are given in Eqs E24- E33 in
Appendix E. For the vapor-liquid equilibrium model, the correlation for jk (R2=0.9997) over the
temperature range of 301.15 K - 393.15 K is obtained from the Brelvi-O’Connell model as given
as given by Eq. (2.35)
1
MEA,H O = 8.7105 − 1.324ln T (K), H O,MEA = , MEA,MEA = H2O,H2O = 1 (2.35)
2 2
MEA,H O 2
2.7.2 MEA Heat of Absorption
The heat of CO2 absorption, H abs in aqueous MEA solution is derived from enthalpy balance of
an absorption process (or mixing process) at constant temperature during which n0,CO2 mols of
CO2 gas dissolves into an initial MEA-H2O-CO2 system with n0,i number of moles to
iSm Initial
form a final solution with n0,i number of moles as given in Eq. (2.36)
iSm Final
n0,i H i + H apparent − n0,i H i + H apparent − n0,CO2 H CO
L hex L hex V
iSm Final iSm Initial
2
H abs = (2.36)
n0,CO2
26
2.8 Parameter Estimation with Parmest
The standard state thermodynamic parameters and binary “ion pair-water” interaction parameters
for the major ionic species ( MEAH + , MEACOO − ) in the simplified solution chemistry of the
MEA-H2O-CO2 ternary system are regressed from selected experimental data using Parmest. All
other NRTL parameters are set to literature values as shown in Table 29 and Table 30 in Appendix
C, unless otherwise specified. Parmest is an open source Python package distributed as part of
Pyomo for model-based parameter estimation and determination of uncertainty associated with
the point estimates.43 The estimator in Parmest requires: (i) a model with the objective function,
(ii) the list of parameters (thetas: θ ) to estimate, and (iii) the experimental data.
The least square objective function supplied to the estimator in Parmest is given by Eq. (2.37),
( )
NP Nh
Fobj = p ln pCO + h ( H abs , j − H abs , j )
2 2
exp
2 ,i
− ln pCO
cal
2 ,i
exp cal
(2.37)
i =1 j =1
where N p and N h are the total number of data points for CO2 partial pressure and heat of
absorption experiments respectively, p = 1, and h = 0.2 . This sets the partial pressure of CO2 (
kPa ) and the heat of CO2 absorption ( kJ / mol CO 2 ) on the same scale. The partial pressure of
CO2 is transformed as shown in Eq. (2.37) to aid convergence44 of the parameter estimation
problem. The objective function is minimized by the adjustment of model parameters listed in
Table 8 subject to the thermodynamic model described in the previous sections and parameter
bounds (obtained via preliminary sensitivity studies) as given by Eq. (2.38),
lb ub (2.38)
27
Mathias and O’Connell (2012)45 evaluated the thermodynamic consistency between the VLE and
heat of absorption data of CO2 in aqueous MEA solution using the Gibbs-Helmholtz (G-H)
equation. The authors45 identified inconsistencies between the VLE and calorimetric data and
recommended the use of new measurements of VLE data at 393 K and above and new heat of
absorption data at 373 K and above to remove the inconsistencies. Kim et al. (2014)46 presented
additional measurements of calorimetric data and partial pressure of CO2 in aqueous MEA
solution where the measured CO2 heat of absorption data were lower compared to those reported
earlier by Kim and Svendsen (2007) at 80 °C and 120 °C. The authors46 concluded that the
temperature effect on heat of absorption remains uncertain and the G-H method cannot be used to
validate the temperature dependency of heat of CO2 absorption. In this work, the VLE and heat of
absorption data of the ternary MEA-H2O-CO2 system used in the parameter regression are
summarized in Table 5,
Table 5: Experimental data of MEA-H2O-CO2 ternary system used in the regression
Data Type Data Temperature Loading MEA CO2 Reference
Points (°C) (mol weight partial
CO2/MEA) percent pressure
(kPa)
VLE 138 40 – 80 0.017 – 15 – 45 0.0007 – Aronu et al.
0.565 19 (2011)
VLE 55 40 – 60 0.114 – 17 – 40 0.005 – Hilliard (2008)
0.591 50
VLE 38 25 – 100 0.003 – 30 0.0015 – Jou et al. (1995)
0.589 822
VLE 25 100 – 120 0.314 – 29.5-39.3 12 – 703 Xu (2011)
0.516
VLE 7 120 0.07 – 0.48 30 1.26 – Kim et al.(2014)
430.36
Heat of 27 40 – 120 0.06 – 0.69 30 Kim et al.(2014)
absorption
Heat of 66 40-80 0.041 – 30 Kim and
absorption 0.678 Svendsen (2007)
28
2.9 Results and Discussions
2.9.1 H2O-MEA Binary System
Extensive experimental data on VLE,47, 48, 49 excess enthalpy,50, 51 heat capacity,52, 53 and activity
coefficients44 have been used by various authors27, 35, 36, 54, 55
to determine the binary NRTL
interaction parameters of the MEA-H2O binary system. In this work, it is desired to select the
binary NRTL interaction parameters of the MEA-H2O system from literature and use them to
evaluate the derived analytical expressions of excess enthalpy from the e-NRTL model for
comparison with experimental data. The following studies that have used/reported NRTL binary
interaction parameters of the MEA-H2O binary system for predicting CO2 solubility in MEA-H2O-
CO2 ternary system are considered: Austgen et al.(1989),54 Hessen et al.(2010),27 Liu et al.
(1999),56 and Zhang et al.(2011)35 as shown in Table 6. The binary interaction parameters from
these studies are evaluated based on Renon and Prausnitz’s probable error29 test and the Redlich-
Kister area test at 40°C and 80°C as shown in Table 6 .
Table 6: MEA(1) + H2O(2) binary interaction parameters from selected publications
Source ID A12 B12 A21 B21
Austgen et al.(1989) W1 0 -649.75 1.674 0
Hessen et al.(2010) W2 1.998 -1030 -0.5897 250.6
Liu et al. (1999) W3 -1.6097 -265.20 1.6489 125.1
Zhang et al.(2011) W4 1.5201 -910.3 0.1559 110.8
Renon & Prausitz (1968)29, 57 defined the probable error in a point estimate as the maximum
variation in the parameter that keeps the root mean square (rms) deviation less than 1.5 times its
optimum rms deviation. The authors29, 57 studied several systems based on this definition and
concluded that, parameters representing the VLE data poorly have larger probable errors.29 The
probable error in the binary parameters is calculated as the absolute error in the binary parameter
29
difference ( i.e g11 − g 22 = RT 12 − 21 ) . Table 7 shows that the probable error in the parameters
of studies W2 and W4 are one order of magnitude smaller than those in studies W1 and W3.
Redlich-Kister area test measures the thermodynamic consistency of VLE data where the
logarithmic value of the ratio of the activity coefficients as a function of mole fraction sums to
zero. As the reduced data is represented by the fitted parameters in the thermodynamic model,
applying the area test will reveal fitted parameters that result in thermodynamically consistent
models. The parameters of studies W2 and W4 satisfy the area test with the smallest area.
Table 7:Selection criteria of MEA(1) + H2O(2) binary parameters
Test W1 W2 W3 W4
40°C
Probable Error (kJ/mol) 9.76 3.91 11.73 4.94
Area 0.00240 0.00047 0.00479 0.00062
80°C
Probable Error (kJ/mol) 10.32 3.05 12.81 4.48
Area 0.00106 0.00012 0.00388 0.00020
A plot of the excess enthalpy predictions from the e-NRTL model for the MEA-H2O-CO2 ternary
system when the mole fractions of the solutes(ionic and molecular) are set to zero is shown in
Figure 3. With the reformulated parameter structure of the symmetric e-NRTL model, the excess
enthalpy expression for the MEA-H2O binary system was evaluated without numerical
singularities or the addition of small tolerance terms. Practical application of this reformulation
includes the simultaneous regression of thermodynamic properties parameters from binary and
higher order systems and mixed solvent systems such as mixed amines using the same excess
Gibbs energy model. The binary interaction parameters from studies W2 and W4 show satisfactory
30
representation of the excess enthalpy data from Touhara et al (1982)51 at 25°C. As excess enthalpy
data provides reliable information on the temperature dependency of activity coefficients, binary
parameters from study W2 are chosen in this work since they have the minimum probable error
and area test values, whilst providing satisfactory predictions of the phase envelopes of the MEA-
H2O binary system as shown in Figure 4.
Figure 3: Comparison of excess enthalpy prediction from e-NRTL model (mole fractions of solutes
set to zero in the MEA-H2O-CO2 system) with MEA-H2O excess enthalpy data at 25°C from
Touhara et al.(1981)63, using various parameters from the literature.
31
A
Figure 4: Comparison of MEA-H2O phase envelope predictions to: (A) P-xy data at 60 °C from
Kim et al. (2008)58 (B) P-xy data at 90 °C from Tochigi et al. (1999)47 and (C) T-xy data at 101.325
kPa from Cai et al. (1996).49
32
2.9.2 MEA-H2O-CO2 Ternary system.
The parameter estimation problem with 14250 variables and 14240 equality constraints was solved
on a desktop computer featuring six intel core i7-8700K 3.7 GHz processor and 16 GB RAM using
IPOPT59 paired with HSL MA27 linear solver60 in 3511 CPU seconds. Estimated values of the
, aq
reference parameters ( f GMEACOO −
,298.15
, aq
, f H MEACOO −
,298.15
, C p,,iaq ) for MEACOO− and MEAH +
ions and their water-ion pair interaction parameters are shown in Table 8 along with the
corresponding standard deviations. All the regressed parameters have low standard deviation
except the value of C p,,iaq for MEAH + ion and the ion pair-water interaction parameter,
C(MEAH+ , MEACOO− ),H O . Zhang et al. (2011)35 also observed that the value of C p,,iaq for MEACOO−
2
ion has higher variance while considering the five basic aqueous phase reactions for MEA-H2O-
CO2 system. This may be due to scatter in the experimental data both within and between different
sources of data and/or strong correlation among these parameters as shown in Table 9 that reports
the correlation coefficient of the estimated parameters. The ‘ion-pair’-water interaction parameter,
C(MEAH+ , MEACOO− ),H O shows a strong negative correlation with the infinite dilute heat capacity (
2
C p,,iaq ) of the MEAH + ion.
33
Table 8: Estimated parameters for MEA-H2O-CO2 ternary system
Parameter ID Parameter Value Standard Deviation
Standard State Formation Parameters (kJ/mol)
, aq
S1 f GMEACOO −
,298.15
-481.9 0.382
, aq
S2 f GMEAH +
,298.15
-179.3 0.110
, aq
S3 f H MEACOO −
,298.15
-712.0 0.030
, aq
S4 f H MEAH +
,298.15
-324.0 0.038
Standard State C p,,iaq Parameters (kJ/mol K)
Cp1 AMEACOO− 0.1135 0.615
Cp2 AMEAH+ 0.2584 0.12
Electrolyte-Molecule Binary Parameters
E1 C + −
18.588 0.661
H 2 O,(MEAH , MEACOO )
E3 D H O,(MEAH+ , MEACOO− ) -1533.6 0.844
2
E4 D(MEAH+ , MEACOO− ),H O 1224.1 0.0376
2
34
Table 9: Correlation coefficients of estimated parameters
ID Cp1 Cp2 E1 E3 E4 S1 S2 S3 S4
Cp1 1
Cp2 -0.1397 1
E1 -0.1680 0.8563 1
E3 0.2066 0.7504 0.5760 1
E4 0.0096 -0.0001 -0.0178 0.0045 1
S1 0.3413 0.5862 0.3281 0.3768 0.0211 1
S2 -0.0359 0.0391 -0.0003 0.0146 0.0001 0.0420 1
S3 0.4247 -0.5456 -0.7399 -0.2809 0.0044 0.1034 0.0097 1
S4 -0.5670 0.3278 0.1955 0.1579 -0.0013 -0.1582 -0.0135 -0.6218 1
35
Figure 5: Comparison of the natural logarithm of estimated and experimental values of CO2 partial
pressure as function of CO2 loading in MEA solution
Figure 6:Comparison of model predictions with experimental data for CO2 equilibrium pressure
over MEA-H2O-CO2 system
36
The ratio of model predictions to measured CO2 partial pressure (i.e. ln PCO2 ,cal / ln PCO2 ,exp ) of the
MEA-H2O-CO2 system is shown in Figure 5. The deviations are random around the unit line
showing no bias. Figure 6 shows the model comparison with the data specifically in the range of
industrial interest- i.e., for 30 % wt. MEA concentration over the CO2 lean loading of 0.003 -0.5
and temperature of 40 °C -120 °C. The mean absolute percentage error (MAPE) obtained as
defined in Eq. (2.39) for the CO2 partial pressure prediction is 40.5 %.
100% n estimatei − experiment i
MAPE =
n i =1 experiment i
(2.39)
The enthalpy of CO2 absorption affects the temperature profile in the absorber or stripper column
which in turn affects the CO2 capture rate and the energy requirement for solvent regeneration.12,
34, 61
Figure 7 and Figure 8 show the satisfactory reduction of the experimental data of Kim et al.,
201446 (labeled as dataset A) and Kim and Svendsen, 200762 (labelled as dataset B) for the
differential heat of CO2 absorption in aqueous MEA solution at 40 °C, 80 °C, and 120 °C. The
model predictions are more consistent with the data of Kim and Svendsen (2007) at 80 °C (label
as dataset 80B). The mean absolute percentage error (MAPE) obtained for the differential heat of
absorption is 5.9 %.
Mass transfer rate-based simulation of absorber/stripper columns requires the bulk liquid
equilibrium composition resulting from the aqueous phase reversible reactions7, 11, 63. Figure 9
shows the comparison of speciation plot as a function of CO2 loading for 30 wt % MEA solution
at 40 °C with experimental data from Bottinger (2008)64, Jakobsen et al.(2005)65, and Aspen model
predictions from Hilliard (2008)39. The experimental NMR data from Jakobsen et al.(2005)65 are
well represented by the model even though speciation data were not included in the parameter
regression. Also, specific heat capacity data were not considered in the parameter regression. For
37
validation, the heat capacity of the liquid mixture is calculated using the central difference
approximation as shown in Eq. (2.40)
H L (T + T ) − H L (T − T )
C p ,mix (2.40)
2T
where H L is the enthalpy of the liquid mixture calculated from in Eq. F1 in Appendix F. The
calculated specific heat capacity for 30 and 40 wt % MEA solution as function of CO2 loading at
25 °C is compared to the data from Weiland et al. (1997) and the predictions from the CCSI36 and
Plaza66 Aspen models as shown in Figure 10. The calculated specific heat capacity from this work
is consistent with the data from Weiland et al. (1997) and the CCSI Aspen model predictions.
Figure 7:Comparison of estimated and experimental values of differential heat of CO2 absorption
as function of CO2 loading in MEA solution
38
Figure 8:Comparison of model predictions with differential CO2 heat of absorption data for 30
wt % MEA solution.
Figure 9: Comparison of model results at 40 °C with NMR data from Jakobsen et al. (2005) for 30
wt% MEA solution and that of Hilliard(2008)39 and Böttinger (2008)64.
39
Figure 10:Comparison of experimental data (Weiland et al., 1997), CCSI44 and Plaza’s77 Aspen
model predictions to model predictions of specific heat capacity at 25 °C for 30 wt % and 40 wt
% MEA solution as function CO2 loading.
Figure 11: Comparison of excess enthalpy for tertiary MEA-H2O-CO2 at binary limiting condition
: (A) Aspen model35 at 80 C (B) This work.
40
Figure 11A shows the breakdown of the enthalpy model for tertiary MEA-H2O-CO2 along with
the excess enthalpy. It is observed that the excess enthalpy obtained by numerical derivation does
not have the curvature typical for systems with non-randomness factor set to 0.2, and its limiting
value at zero CO2 loading as would be expected for a binary MEA-H2O system is inconsistent with
experimental data. However, as Aspen model is propriety, it is unknown if this excess enthalpy is
true species based or apparent species based. Nonetheless, the excess enthalpy computation
obtained Figure 11B follows the curvature expected for systems with randomness factor set to
0.2, and its limiting value is consistent with binary MEA-H2O system as CO2 loading tends to
zero.
The comprehensive thermodynamic framework presented in this work reduces available
experimental data for aqueous electrolyte and non-electrolyte systems to a predictive model with
parameters over a wide range of conditions. In addition to the thermodynamic model, the accuracy
of the predictions strongly depends on the quality of the experimental data and the method of
parameter regression. Baker’s method67 of regression is adopted in this work which gives excellent
results when using high quality data.5 However, other approaches can also be considered like the
maximum-likelihood method68 which considers measurement errors. Parmest has other methods
such as the Bootstrap method which does not rely on the assumption of symmetry or normality.
2.10 Conclusion
This chapter presents a comprehensive description of the thermodynamic framework applicable to
both aqueous electrolyte and non-electrolyte systems and develops a reformulated e-NRTL model
in Pyomo. The reformulated e-NRTL model avoids numerical singularities due to the absence of
ionic species. Analytical expressions for excess enthalpy applicable to both aqueous electrolyte
and non-electrolyte systems are presented. The thermodynamic model describes the liquid phase
41
nonideality of the multicomponent MEA-H2O-CO2 system satisfactorily over a wide range of
operating conditions of practical importance. The contribution of excess enthalpy to the enthalpy
of the non-ideal solvent mixture is crucial for reliable determination of the temperature dependence
of the e-NRTL binary interaction parameters for consistent thermodynamic models. The ‘probable
error’ test is a quick and rough estimate for selecting valid binary NRTL interaction parameters.
The analytical excess enthalpy expressions can be readily extended to other mixed-solvent
systems, especially at high concentration where both liquid-liquid and vapor-liquid equilibrium
are important. This will provide better understanding of mixed solvent systems by adding reliable
information on the temperature dependence of e-NRTL binary interaction parameters. The
reformulated thermodynamic model with analytical expression for excess enthalpy will be used in
future work to develop a rigorous, dynamic rate-based model for flexible CO2 capture. For a typical
absorber model, multiple calls of the thermodynamic properties based on the analytical excess
enthalpy expressions can provide more accurate and consistent enthalpy and VLE model
predictions. Finally, the model will be available for public download at IDAES_GitHub_Page.
42
Chapter 3 Transport, Mass Transfer and Hydraulic Models
This chapter describes the model development for transport properties, mass transfer and hydraulic
correlations for the design of a two-phase countercurrent-flow packed column for reactive
absorption or desorption. Transport properties of interest include Maxwell-Stefan and Fick’s
diffusivities, dynamic viscosity, thermal conductivity, and surface tension. The Mass transfer and
hydraulic correlations include the liquid holdup, pressure drop, flooding velocity, mass transfer
coefficients and interfacial area. Billet and Schultes correlations69, 70-73 for mass transfer and
hydraulics are widely used in most process simulators which rely on packing-specific constants
and dimensions. Based on the database provided by Billet and Schultes73, data for valid ranges for
the constants in Billet and Schultes correlations are available to be used as a guide for packings
that are not available in the database using the packing material type and specific area. Viscosity
and surface tension models developed by Morgan et. al (2015)74 and the integrated mass transfer
models developed by Soares et. al. (2018)38 for MEA-CO2-H2O system are adopted in this work.
Lastly, an analysis of the integrated mass transfer models for admissible flows in a packed column
for stable operation is also presented.
3.1 Introduction
Application of the structured packings in two-phase countercurrent-flow packed column for
reactive absorption or other separation columns is very common due to their low pressure drop
and high mass transfer rates in both phases. Structured packings are made of corrugated sheets
arranged in parallel with flow channels inclined, typically at an angle of 45 or 60 deg to the column
axis as shown in Figure 12.
43
(b)
(a)
Figure 12 : Structured packing: (a) Mellapak 250Y (b) Simplified diagram of structured packing
Structured packings offer large specific surface area (80 to 1000 m2/m3 ) with void fractions of
about 98%. Various mass transfer and hydraulics correlations38, 73, 75, 76-78 have been developed to
describe the relationship between the two-phase countercurrent flow and interphase interactions in
the packed column. Recently, Chinen et. al (2018) developed an integrated mass transfer model
where the model selection and parameter estimation for diffusivity, interfacial area, liquid-side and
gas-side mass transfer coefficients, and the kinetic model were done simultaneously using the
wetted wall column (WWC) data along with the data from the packed towers for the MEA-H2O-
CO2 system. Holdup and interfacial area models from the integrated mass transfer approach
developed in Chinen et (2018) are adopted in this work. Morgan et. al (2015)74 updated the
viscosity model from Weiland et al. (1998)52, and the surface tension from Jayarathna et al.
(2013)79 for MEA-CO2-H2O, which are also adopted in this work. The liquid-side and gas-side
mass transfer coefficients from Chinen et al (2018)38 were computed from Billet and Schultes
correlations.69, 70-73 Hence, a study of the integrated mass transfer models for admissible flows in
a packed column for stable operation is also presented. For electrolyte systems, transport
44
properties for ionic species are required for the application of Maxwell-Stefan diffusion equations
which include binary diffusivities of true species in the liquid phase. Methods for computing
thermal conductivity, binary Maxwell-Stefan diffusivities, and Fick’s diffusivities are obtained
from the literature.
3.2 Transport Properties
Transport properties for the rate-based model include dynamic viscosity, Fick’s diffusivity, and
thermal conductivity for both liquid and vapor phases along with the surface tension as shown in
Table 10. Detailed equations are prersented in Appendix G. Modified correlations for the viscosity,
surface tension and density models by Morgan et al. are adopted in this work.74 Fick’s diffusivity
of CO2 in the liquid phase is calculated from correlations by Ying and Eimer.80 Fick’s diffusivity
of MEA in the liquid phase is estimated from the correlation by Snijder et al.,81 while Fick’s
difusivity of carbamate and potonated amine are assumed to be equal and obtained from Hoff et
al.82 Deatiled discussions on Maxwell-Stefan diffusivities are provided in the next section as they
apply to the rigorous two-film model to be developed in the next chapter.
Table 10: Transport properties
property/correlation Reference/Remark
Viscosity of the liquid solution Morgan et al.74
Viscosity of the vapor components Sutherland83
Vapor Viscosity Sutherland,83 Wilke84
Liquid thermal conductivity Sato-Riedel correlation,32
Li’s mixing rule85
vapor thermal conductivity Reid et al.32
Binary diffusivity of vapor components Seader and Henly,86 Fuller correlation
Diffusivity of CO2 in aqueous MEA Ying and Eimer87
Diffusivity of MEA in solution Snijder et al.81
Diffusivity of MEAH+ and MEACOO- Hoff et al.82
Liquid surface tension Asprion,88 Morgan et al.74
45
3.2.1 Maxwell-Stefan(MS) Diffusivity and Binary Mass transfer Coefficient
For n number of species in the liquid phase, n(n − 1) 2 MS binary diffusivities are required to
describe the MS transport equations. For MEA-CO2-H2O system with six species, binary
diffusivities are listed in Table 11 and the methods for their calculations are shown in Table 12.
Table 11: MS Diffusivities for MEA-CO2-H2O system
H 2O MEA MEACOO − MEAH + HCO3−
L L
CO2 ÐM,W ÐM,M ÐL−,M ÐL+ ,M ÐL−,M
HCO3− ÐL−,W ÐL−,M ÐL−,− Ð+L,−
MEAH + ÐL+ ,W ÐL+ ,M ÐL+ ,−
MEACOO− ÐL−,W Ð−L,M
MEA ÐLM,W
Ion-water pair binary diffusivities are calculated from the infinite dilution diffusivities using
Nernst-Hartley equation and extended to finite concentrations using Stokes-Einstein viscosity
relation. Ionic conductivities required in the Nernst-Hartley relation for MEA-CO2-H2O system
are given in Table 13, where H2O,T=298 is the viscosity of water at 298 K and soln is the viscosity
of the solution at temperature T in Kelvin. CO2-H2O binary diffusivity is estimated based on
89
correlations presented by Versteeg and Van Swaalj . MEA-H2O and CO2-MEA binary
diffusivity are estimated from the infinite dilution diffusion coefficient using the relation of Wilke
and Chang 90and extended to finite concentration using the modified Stokes-Einstein relation as
proposed by Versteeg and Van Swaalj 89. The binary diffusion coefficient of the ions with respect
to a molecular species is set equal to the effective diffusivity of the ion in the liquid mixture. The
91
counter ion pair binary diffusivities are computed as described by Wesselingh and Krishna
while the binary diffusion coefficient of the like-ion is set equal to the mean of the effective
diffusivities of the two ions.
46
Table 12: Correlations for MS diffusivities for MEA-H2O-CO2 system
Binary pair Correlation
Ion-water RT i H2 O,T=298
ÐiL,H2 O ( cm 2 / s ) = , i = MEAH +
, MEACOO − , HCO3−
zi F 2 soln,T
Molecule-solvent H2 O
0.8
L
ÐCO 2 ,H 2 O
( m2 / s ) = 2.35 10−6 exp − 2119
T soln
0.6
( cm / s ) = 7.4 10 (2.26 18.02) T V H2O
0.5
−8
Ð L 2
MEA,H 2 O
H2 O H2 O soln
( cm2 / s ) = 7.4 10−8 (1 61.08) T (VMEA )−0.6
0.5
L
ÐCO 2 ,MEA
soln
Ion-molecular i MEAH + , MEACOO − , HCO3−
ÐiL,m ( cm 2 / s ) =ÐiL,H2 O xk , m CO 2 , MEA
k CO 2 , MEA,H 2 O
k
Counter-ion ÐL+ ,H2 O + ÐL− ,H2 O i 0.55
ÐL+ , − , i = 0.5 z 2j x j
2 z z 2.3
+ − jions
like-ion ÐiL,m + ÐiL,m
Ð L
−, − ( cm / s ) =
2 i = HCO3−
2
i = MEACOO−
Table 13 : Ionic conductivities 92
ion ionic conductivity( i ,Scm 2 mol −1 )
MEAH + 73.6
MEACOO − 16.4
−
HCO 3 44.5
The corresponding binary mass transfer coefficients to the MS binary diffusivities are determined
using the properties of the mixture(density ,viscosity etc.) and the MS binary diffusivities in the
appropriate mass transfer coefficient model such as the Billet and Schultes correlations.
3.3 Hydraulic Models
Hydraulic correlations include liquid holdup, pressure drop, and flooding velocity correlations.
Liquid holdup in a packed column is an important variable for absorption with chemical reactions
as it affects the extent of reaction. Many authors70, 77, 93 have used liquid holdup as a key parameter
47
in their pressure drop and mass transfer coefficient models. Hence, the liquid holdup model affects
the performance of other hydraulic and mass transfer models. Although structured packed
columns are built to have low pressure drops, modelling the pressure drop by using appropriate
pressure drop correlation is also important for pressure-driven flow across the columns during
dynamic simulations.
3.3.1 Liquid Holdup
Liquid holdup, hL is the volume of liquid in the void spaces of the packing, per volume of packing
and it is the sum of the static and the operating (dynamic) holdup. The operating holdup refers to
the volume of liquid that drains out of the bed when the gas and liquid flows are stopped per
volume of packing while the static holdup represents the volume of liquid remaining in the bed
after a long drainage time per volume of packing. In a countercurrent gas and liquid flow, the
hydrodynamic behavior of the column is classified into three regions of operation using the loading
and flooding lines: preloading, loading, and flooding region. In the preloading region, the liquid
holdup is constant for a given superficial liquid velocity and independent of the gas flow velocity
as the liquid flows freely onto the void spaces negligible impact from the counter current gas flow.
Above the loading point, the liquid holdup increases as the frictional forces between the gas and
liquid increases and support some of the descending liquid, thus slowing down the liquid flow.71
At the flood point, most of the liquid is held up71 as the shear force in the gas flow is strong enough
to retain the liquid such that liquid holdup is very large and the liquid filles the entire cross-
sectional area.
Billet and Schultes (1993)71, 73 developed a mathematical model for determining the liquid hold-
up in gas (vapor)-liquid countercurrent columns filled with random or structured packings up to
the flooding point using data from 56 different column packings and 16 gas-liquid systems. Their
mathematical model is based on a physical model that approximates the void volume of packed
48
columns with vertical, parallel flow channels, in which the liquid flows counter-currently
downwards along the wall forming a thin film. The liquid hold-up model considers the flow
characteristics in a packed column governed by the geometry, surface structure, and arrangement
of the packing, and the physical properties and velocities of the two phases as given by Eqs. (3.1)
to (3.4),
13
u
hL = hL,s + ( hL, Fl − hL,s ) V (3.1)
u
V, Fl
0.05
w,20
= 2.2hL ,s L
w,20 L
hL , Fl (3.2)
2/3
ah
1/3
1 L
hL , s = 12 uL a 2p , uV uV,s (3.3)
g L ap
u 0.15 u 2 a 0.1
Ch L L L p if Re 5
ah a p L g uL L
= , Re = (3.4)
ap 0.25
uL L uL2 a p
0.1
a p L
0.85Ch if Re 5
a p L g
where is hL, s , hL, Fl are the liquid hold-ups at the loading and flooding point respectively, ah / a p
is the ratio of hydraulic to geometric surface areas that captures the deviation of the physical model
from the actual flow pattern in the column, Ch is a constant to be obtained experimentally, a p is
the packing specific area, u L , uV are the liquid and vapor loads respectively, while uV,s , uV, Fl are
the vapor load at the loading and flooding point respectively, w,20 is the dynamic viscosity of
water at 20°C, w,20 is the density of water at 20°C. Eq. (3.3) is only valid up to the loading point,
and for a given liquid load, the liquid holdup is constant71 in the preloading region. Hence, the
49
liquid holdup in the preloading region can be computed once the loading velocity is known as
given by Eqs. (3.5) to (3.8)
g p 1 L
1/2 1/3 1/6 1/2
1/2 1 L L
uV, s = 1/6 − a p 12 uL , s 12 uL , s (3.5)
s a p g L g L V
−2 ns
g L
0.4
L V
s = 2 FLV , FLV = (3.6)
Cs
V V L
if FLV 0.4 : ns = −0.326, Cs = Cs ,table
0.1588
(3.7)
if FLV 0.4 : ns = −0.723, Cs = Cs ,table 0.695 L
V
V L
uL,s = u (3.8)
L V V, s
where s is the resistance coefficient at the loading point, FLV is the flow parameter and is
dependent on the L / V mass flow ratio, Cs ,table is the packing-specific constant. For stable
operation of the column, the preloading region is preferred which is typically about 70% of the of
the superficial velocity at the flooding point. Jammula et. al (2021)76 developed a new model for
calculating the superficial velocity at the loading point as given by Eq. (3.9)
hL0.3, pre (1 − hL , pre ) ( sin )
0.5
0.6
0.5
d
uV, s = 64.6 L e (3.9)
L ( L L )
0.5
u 0.5 0.5
G
where hL, pre is the Okstate76, 94 liquid holdup model in the preloading region as given by Eq. (3.10)
1/3
modified
a1.23 Re0.41 L2 LuL
hL , pre = 0.114 p 0.28 , Remodified = (3.10)
(1 − p ) L g sin a p L
2
where is the corrugation angle i.e the angle of the flow channel (in degrees) with respect to the
horizonal as shown in Figure 13. The geometric area of packing is defined as given by Eq. (3.11)
50
4S
ap = (3.11)
Bh
Figure 13: Structured Packing Channel Geometry
where S is the channel side (corrugation spacing) , B is the channel base, h is the crimp height as
shown in Figure 13 . The corrugation angle for Sulzer Mellapak packing is 45 deg for the Y-type
and is often measured with respect to the column axis (vertical). Table 14 shows a summary of
the geometry for some common packings.
Table 14: Packing Channel Dimensions
Packing Corrugatio Channel Channel Crimp Void Specific
n angle, side, S base, B height, h fraction, area, a p
(deg) (mm) (mm) (mm) ε (m2/m3)
Mellapak 250Y (M250Y) 45 17 24.1 11.9 0.95 250
Mellapak 500Y (M500Y) 45 8.1 9.6 6.53 0.92 500
MellapakPlus 252Y 45 17 24.1 11.9 0.98 250
(MP252Y)
Flexipac 1Y 45 9 12.7 6.4 0.91 410
(F1Y)
51
The liquid holdup model developed by Soares et. al. (2018)38 for MEA-CO2-H2O system from the
integrated mass transfer approach is given Eq. (3.12)
0.6471
1 1/3 u
L = 11.45 2 2/3 L L (3.12)
S g L a p
3.3.2 Flooding Velocity
Flooding condition governs the maximum flowrates of phases that can pass through a packed
column. Mass transfer conditions are optimal near the flooding point, however, for safe and stable
operation of the column, the column diameter is often designed for gas velocities that are in the
vicinity of the flooding velocity, usually of about 0.6 to 0.8 of the flooding velocity. The channel
model by Billet and Schultes (1999)73 predicts the flooding velocity as given the Eqs. (3.13) to
(3.16),
2g
1/2 ( − h )3/2 h 1/2
L
1/2
uV, Fl = p L , Fl L , Fl (3.13)
Fl p
1/2
a p V
where s is the resistance coefficient at the flooding point as given by Eq. (3.14)
−2 nFl
g L
0.2
Fl = 2 FLV (3.14)
CFl V
if FLV 0.4 : nFl = −0.194, CFl = CFl ,table
0.1028
(3.15)
if FLV 0.4 : nFl = −0.708, CFl = CFl ,table 0.6244 L
V
V L
uL , Fl = u (3.16)
L V V, Fl
while the column diameter can be calculated for stable operation at/below the loading point can
be obtained from a fraction of the flooding velocity, it is more convenient to design the column
52
using the loading velocity when using Billets and Schultes channel model. Recently, Jammula et
al (2021)76 presented a new flooding velocity model for structured packings as given by Eq. (3.17)
to (3.18)
0.5
d 0.36 0.25 sin 3 −
= 0.695 h L0.25 (1 − hL , Fl ,oks ) g (3.17)
L V
uV, Fl
L V
1/3
a 0.88 modified
Re0.41 L2
hL , Fl ,oks =2 p
0.15 (3.18)
(1 − p ) L g sin
2
where hL, Fl ,oks is the OkState76 liquid holdup flooding model for predicting the critical liquid
holdup at any given liquid load, d h is the hydraulic diameter as given by Eq. (3.19),
p
dh = 4 (3.19)
ap
3.3.3 Pressure Drop Correlation
Pressure drop in the gas flow is an important criterion for the choice of packing. Pressure drop also
affects flowrates in dynamic simulations. Billet and Schultes (1999) correlation for pressure drop
P
per unit length for irrigated packing, is given by Eq. (3.20),
H
P ap FV2 1
= L (3.20)
( p − hL ) 2 K
3
H
where the gas load factor, FV = uV V , K is the wall factor as given by Eq. (3.21),
1 2 1 dp 1− p
= 1+ , d p = 6 (3.21)
K 3 1 − p dc ap
where d p is the particle diameter, d c is the column diameter, L is the resistance coefficient for
irrigated packing as given by Eq. (3.22),
53
1.5 0.3
64 1.8 − h hL 13000
C p ,0 + 0.08 p L exp 1/2 L
u (3.22)
Re V Re V p hL , s ap g
where for normal operation of the column (i.e., at or below the loading point), the ratio hL / hL , s = 1
3.4 Mass Transfer Correlations
The mass transfer correlations include the mass transfer coefficients for the liquid and vapor phases
and the interfacial area model. Billet and Schultes’ mass transfer empirical models for vapor and
liquid phases are widely used in most process simulators, and they depend on how the liquid
holdup is computed along with packing-specific constants. The interfacial area for mass transfer
is dependent on the material type and dimensions of the packing along with the physicochemical
properties of the flowing liquid. Hence, an analysis of the interfacial area models from selected
authors|(Tsai,78 Billet and Schultes73) and the effect of liquid holdup models from selected authors
(Tsai78 , Jammula94, Billet and Schultes73 ) on the mass transfer coefficient for admissible flows
in a packed column for stable operation is presented .
3.4.1 Mass Transfer Coefficient
The mass transfer coefficient models for the vapor and liquid phase by Billet and Schultes (1999)
are given by Eq. (3.23) and Eq. (3.24) respectively,
3/4
u V ae
1/3
1 a 3/2
kV ae = CV p
DV V V ; (3.23)
a
( p − hL ) V DV a p
1/2
d h1/2 p V
a
1/2 1/2
u D
k L ae = CL 12 L L a p e
1/6
(3.24)
a
hL d h p
where CV , CL are constants, characterizing the shape and structure of the packing.
54
3.4.2 Interfacial Area
The effective interfacial area, ae model for mass transfer provided by Billet and Schultes (1999)72,
73
up to the flood point is given by Eqs. (3.25) to (3.27)
13
ae ae,s ae,Fl ae ,s uV
= + − (3.25)
a p a p a p a p uV,Fl
0.56
ae, Fl ae,s
= 7 L (3.26)
ap ap
w,20
−0.2 0.75 −0.45
−0.5 u d uL2 L d h uL2
= 1.5 ( a p d h ) L L h
ae,s
(3.27)
ap L L gd h
where ae,s , ae,Fl are the effective interfacial area at the loading point and flooding points
respectively, w,20 is the surface tension of water at 20°C. Billet and Schultes (1995)72 noted that
for systems in which the surface tension decreases along the downward path of the liquid in the
column (referred as negative system), the effective interfacial area will be reduced due to the
formation of vortices. This is accounted for by the Marangoni number as given by Eq. (3.28) to
ae,s a
= e ,s
a p negative system a p Eq. (3.7)
(
1 − 2.4 10−4 Ma L
0.5
) (3.28)
d L x
Ma L = (3.29)
dx DL L aP
where d L / dx describes the differential change in surface tension with the liquid concentration
x of the volatile component along the column, and x is the driving concentration difference from
the liquid bulk to the surface in contact with the gas.
55
Tsai et. al (2010)78 developed an effective interfacial area model based on effective area data of a
variety of packings as a function of liquid load, liquid viscosity and surface tension covering
property ranges of amine solvents as given by Eq. (3.30),
0.116
ae
4/3
1/3 u L
= 1.34 g
L
(3.30)
ap L a
p
while Chinen et. al (2018)38 updated the constant 1.34 to 1.42 based on their results from the
simultaneous parameter estimation technique.
Based on the database provided by Billet and Schultes73, a summary data of valid ranges for the
constants in Billet and Schultes correlations is provided as guide for new packing designs or those
not accounted for in the database using the packing material type and specific area as shown in
Table 15.
56
Table 15: Summary statistics of Billet and Schultes (1999)73 packing constants
Packing type Cs CFl Ch Cpo CL CV
min avg max min avg max min avg max min avg max min avg max min avg max
Random M 2.521 2.849 3.560 1.579 1.968 2.340 0.620 0.829 1.369 0.327 0.770 1.056 0.850 1.474 2.038 0.300 0.396 0.495
P 2.654 2.953 3.612 1.522 2.049 2.472 0.383 0.666 1.173 0.311 0.578 0.976 0.410 1.260 1.913 0.257 0.365 0.459
C 2.454 2.712 2.875 1.547 2.026 2.410 0.577 0.826 1.167 0.538 0.811 1.329 1.246 1.438 1.744 0.210 0.345 0.465
Structured M 2.610 3.025 3.178 1.996 2.332 2.558 0.431 0.551 0.690 0.191 0.305 0.460 0.971 1.165 1.405 0.270 0.370 0.422
P 2.653 2.803 3.075 1.973 1.974 1.975 0.511 0.511 0.511 0.250 0.420 0.481 0.739 0.894 1.006 0.167 0.325 0.412
C 2.664 3.375 3.793 1.655 2.517 3.024 1.066 1.375 1.900 0.233 0.301 0.417 1.278 1.292 1.317 0.327 0.331 0.333
Random M 2.521 2.892 3.560 1.579 2.058 2.558 0.431 0.759 1.369 0.191 0.645 1.056 0.850 1.402 2.038 0.270 0.390 0.495
+
Structured P 2.653 2.941 3.612 1.522 2.043 2.472 0.383 0.661 1.173 0.250 0.558 0.976 0.410 1.212 1.913 0.167 0.361 0.459
C 2.454 2.872 3.793 1.547 2.144 3.024 0.577 0.915 1.900 0.233 0.675 1.329 1.246 1.418 1.744 0.210 0.343 0.465
M = Metal, P = Plastic, C = Ceramic, avg =
a C p
a p
57
3.5 Analysis of packed column hydraulics and mass transfer models.
Since the mass transfer coefficient are dependent on the liquid holdup correlation, an analysis of
the interfacial area models from selected authors (Tsai,78 Billet and Schultes73) and the effect of
liquid holdup models from selected authors (Tsai78 , Jammula94, Billet and Schultes73 ) on the
mass transfer coefficient for admissible flows in a packed column for stable operation(pre-loading
region) is presented in Figure 14.
(a) (b)
(c) (d)
Figure 14: Comparison of hydraulic and mass transfer models in the pre-loading region
58
In the preloading region in Figure 14(b) (where the ratio of the liquid velocity to the liquid load
is less than 1 as shown in Figure 14(d)) , the holdup prediction from Tsai’s model78 is less than
that from the models of Jammula76 and Billet and Schultes73. The impact of the liquid-holdup
prediction is seen in the predicted mass transfer coefficient in Figure 14(a) where predicted mass
transfer coefficient is greater than the predictions from the other models. It should be noted that
while the other models predicts the holdup to increase asymptotically as the liquid load increases,
the Tsai liquid holdup model predicts the holdup to increase monotonically. Figure 14(c) shows
that the maximum wetted area of the packing from Tsai’s interfacial area model78 in the range
studied is higher than the dry specific area (250 m2/m3). Opposed to that, the model from Billet
and Schultes73 predicts the maximum wetted area to be 150 m2/m3 .
3.6 Conclusion
Various leading models for transport properties, and mass transfer and hydraulic correlations for
the design of a two-phase countercurrent-flow packed column for reactive absorption or desorption
are presented in this chapter. Based on the analysis of the effect of holdup on the mass transfer
coefficients, the Billet and Schulte (1999) holdup model will be adopted in this work along with
the interfacial area model from Tsai (2010) using the updated parameters from Chinen et al. (2018).
59
Chapter 4 Interface Transfer Models
This chapter describes the development of a rigorous two-film model for mass and energy fluxes
across gas-liquid interfaces for reactive absorption or desorption process that considers the
acceleration of mass transfer due to reactions in the liquid phase explicitly without the use of
enhancement factors. Various algorithms are developed for spatial discretization to obtain accurate
results with lower number of finite elements in the liquid film which helps reduce the
computational cost and improve convergence. In addition, a simplified two-film model is adopted,
where a general method (GM) for calculating enhancement factor over a wide range of reaction
regimes that can adequately account for the instantaneous, fast, and intermediate reaction regimes
over wide operating conditions. These two approaches for describing mass and heat transport
across the interphase are used in the next chapter for developing the continuous differential
contactor packed column (1st generation, packed column model, denoted as Gen 1 that uses
enhancement factor approach) and the stagewise gas-liquid contactor packed column (2nd
generation packed column model, denoted as Gen 2 that uses the rigorous two-film model).
4.1 Introduction
Reactive absorption or desorption is the selective dissolution of gaseous components in a liquid
solvent that is accompanied by chemical reactions95, 96
. Traditionally, reactive absorption
processes are developed based on the equilibrium stage model which assumes that the streams
leaving each stage reaches equilibrium. However, since absorption is a rate-controlled
phenomenon, equilibrium is rarely attained11. Hence, resistance to mass and heat transfer are
accurately captured using non-equilibrium rate-based approaches where separate conservation
equations are written for the vapor and liquid phases under consideration of mass and heat fluxes
60
across the interface.11, 63, 96 In addition, the effect of reactions on the fluxes are also captured. For
simple reactions, this effect is often captured by enhancement factors derived analytically which
are only valid for a few reaction types or the transfer of only one solute component across the
interface97. The application of enhancement factors for complex reaction scheme involving several
parallel or consecutive reversible reactions may not be accurate as in the case of amine-based CO2
capture. Putta et. al (2017)3 has demonstrated the applicability of various enhancement factor
models for CO2 absorption into aqueous amine where the reversible reactions were re-defined to
suit the enhancement factor models. In this work, a rigorous interfacial model is developed that
considers mass and heat transfer resistances using the film theory through explicit computation of
the interfacial fluxes.
While the non-equilibrium stage model96, 98 that considers spatial variation in mass transfer and
heat transfer fluxes in the films along with chemical reactions95, 99 can be very accurate, these
models are computationally challenging for plant-wide optimization. The enhancement factor
method provides a compromise between computational tractability and accuracy for modeling the
mass and heat transfer across gas-liquid interfaces accompanied by chemical reactions.100 Many
authors have used the enhancement factor based model for optimization and control
applications.100-103 Patron and Ricardez-Sandoval100 used an enhancement factor based model to
design a robust nonlinear model predictive controller. The authors noted that an accurate model of
the plant is necessary for obtaining good control performance. Typically pseudo first order and
instantaneous reversible reactions are assumed for computing the enhancement factors for
absorption and stripping, respectively.104, 105, 106 However, these assumptions may not adequately
represent all the reaction regimes occurring along the packing height during widely variable
absorption and stripping conditions107 expected for flexible operation of the CO2 capture unit.
61
Dang and Rochelle108 have shown that a more accurate enhancement factor model is required at
high loading. Commonly used enhancement factor models are those developed by van Krevelen
and Hoftijzer 109 and Savage et al.110 The enhancement factor model developed by van Krevelen
and Hoftijzer is given by an implicit equation based on the assumption that the concentration of
the dissolved unreacted gas in the liquid bulk is zero. Under these assumptions, the enhancement
factor can be explicitly calculated.111 Savage et al.110 improved the model of van Krevelen and
Hoftijzer to account for reversible and instantaneous reactions regimes. Gaspar et al.107 have
presented a general method (GM) for calculating the enhancement factor that is superior to the
aforementioned methods and can adequately account for the instantaneous, fast, and intermediate
reaction regimes over wide temperature and loading ranges; however, the enhancement factor can
no longer be calculated explicitly, instead requiring the solution of a system of non-linear
equations. Gaspar et al.107 have recommended the secant method for obtaining the enhancement
factor using the GM. In this work, we show that, with appropriate bounds on the enhancement
factor and a step-by-step initialization routine developed using the Institute for Design of
Advanced Energy Systems (IDAES) Process Systems Engineering (PSE) computational platform,
112
the GM can be solved reliably for tower design and optimization under widely varying process
conditions.
4.2 Enhancement Factor Model
A computationally efficient approximate method to capture the effect of reactions on the mass
transfer rate is to the enhancement factor. While use of this approach helps to reduce the
110, 113
computational cost, typical enhancement factor approximations are restricted to a given
reaction regime. Gaspar et al.107 have presented a general method (GM) for calculating the
enhancement factor for all reaction regimes for reversible reactions under absorption and
62
desorption conditions. In their approach,107 the set of differential equations describing mass
transport in the liquid film is transformed into a set of algebraic equations. The GM method applied
to the solution chemistry for MEA-H2O-CO2 system as described in section 2.4 are shown in Eqs.
(4.1) to (4.7),
E = 1 + ( E − 1)
(1 − ) i
MEA
(4.1)
(1 − ) b
CO2
1 − *CO2
E = Ha i
(4.2)
1 − bCO2
MEA
*CO2 = bCO2 iMEAH+ iMEACOO− ( iMEA )
−2
(4.3)
t
DMEAH+ ,LCMEA
i
MEAH+
= 1+
2 DMEAH+ ,LC t (1 − )
i
MEA (4.4)
MEAH+
(1 − )
DMEA,LCMEA
i
MEACOO−
= 1+ t
i
MEA (4.5)
2 DMEACOO− ,LC MEACOO−
t
DMEA,LCMEA
E = 1 + *
(4.6)
2 DCO2 ,LCCO 2
(k )
0.5
t t
CCO rx CMEA DCO2 ,L
b
CO2 = *
2
, Ha = (4.7)
CCO 2
kL,CO2
E and MEA are the two unknowns in Eqs (4.1) and (4.2). iMEAH+ , iMEACOO− , *CO2 are
i
dimensionless concentrations of the reaction products, which depend on MEA (dimensionless
i
concentration of MEA at the interface) as given in Eqs. (4.3) -(4.5) . E is the instantaneous
enhancement factor. CO2 is a dimensionless concentration driving force where for CO2 1
b b
63
absorption occurs and for CO2 1 desorption occurs. krx is the overall reaction rate constant of
b
CO2, while Ha is the Hatta number. Di , L is the Fick’s diffusivities of the species in the solvent.
4.3 Rigorous Two-Film Model
A rigorous interfacial model comprised of the vapor and liquid films is developed that considers
mass and heat transfer resistances using the film theory through explicit computation of the
interfacial fluxes in both films. In the liquid film, simultaneous mass transfer and chemical
reactions are modeled by considering both the effect of chemical potential gradient as well as
electric potential gradient if electrolytes are present. Multi-component transport of molecular and
ionic species is modeled by an extended Maxwell-Stefan (MS) transport equation that can capture
thermodynamic non-idealities and effect of electrostatic forces. Furthermore, for fast reaction
kinetics in the liquid film, there is significant steepness and nonlinearity of the transport profile
especially close to the liquid film boundaries. These lead to considerable convergence issues
especially when good initial guesses are difficult to generate. Hence, various grid distribution
algorithms are developed to obtain accurate results with lower number of finite elements in the
liquid film which helps reduce the computational cost and improve convergence. The difference
equation is applied in the vapor film to reduce the computational expense while due to the reactions
in the liquid film, differential equations are used in the liquid film. To felicitate coupling of the
two-film model with the bulk phases, the end nodes for discretization of the liquid film are
explicitly defined as given below:
64
Var. Descrip. Interface Liq. Bulk
sL = 0 sL = 1
Molar flux N i
L,I
N iL , B i ncL
Mole fraction x i
I
xiB i ncL
Temperature T I
TLB i ncL
4.3.1 Vapor film mass balance
For a generic absorption/desorption process, following sets are defined ,
V
SIN = set stagnant components in vapor phase
SINL = set non-volatile components in liquid phase
SDF = set diffusing components crossing the interface
S N = set{nth component}
SA = set{all components}
SAN = S A − S N
ncV = n ( S IN
V
+ SDF )
ncL = n ( S INL + S DF )
where ncV is the number of components in the vapor phase, and ncL is the number of components
in the liquid phase. For reactive absorption/desorption where no reaction takes placing in the vapor
film, the fluxes of the diffusing components are constant. Hence finite difference equations are
used to approximate the Maxwell-Stefan (MS) transport equation as given by Eq. (4.8), where the
right-hand side of Eq. (4.8) has been reformulated using the summation of mole fraction to unity
to eliminate the mole fraction of the nth species,
Niv ncV
N kv v
1
ncV −1
1 Niv
y −y =
i
I
CTv kinv
i
B
+ v v
y + Ni
i CTv v − k
kikv
y −
CTv kinv
, i = 1,.., n
V
c − 1 (4.8)
k =1 CT kik k =1 kin
k i k i
65
where Niv is the molar flux of component i in the vapor film, CTv is the average total
concentration in the vapor film, kikv is the vapor phase mass transfer coefficient, yiI , yiB are mole
fractions of component i at the vapor phase interface and bulk respectively, yi is the average
mole fraction as defined by Eq. (4.9)
yi = 0.5 ( yiI + yiB ) (4.9)
Since only ncv − 1 of the MS equations are independent, the summation equation is applied to
determine the composition of the nth component at the vapor interface as given by Eq. (4.10)
ncv
y
i =1
i
I
=1 (4.10)
The N iV integration constants are determined by (i) equilibrium relationships, (ii) zero fluxes of
stagnant components, and continuity of fluxes at the interphase as given by Eqs. (4.11) to (4.13)
respectively,
fˆiV = fˆi L , i SDF (4.11)
NiV = 0, i SIN
V
(4.12)
NiV = NiL, I , i SDF (4.13)
( )
Where fˆiV = f yiI , ˆiV , P , fˆi L = f ( xiI , i , P sat , He, P ) as described in the thermodynamic
framework in section 2.5.
4.3.2 Vapor film energy balance
The interfacial energy flux in the vapor film , V is the integration constant for the vapor film
energy balance and is defined as given in Eq. (4.14)
ncv
= h (T − T
V
V
B I
)+N V
i H iV (4.14)
i =1
66
where H iV is the partial molar enthalpy of the species computed at vapor bulk conditions, h is the
heat transfer coefficient computed from the ratio of the thermal conductivity of the vapor film to
the thickness of the vapor film, T I is the interface temperature which is same for both vapor and
liquid sides, TVB is the temperature at the vapor bulk.
4.3.3 Liquid film mass balance
The conservation of species in the liquid film is given by Eq. (4.15) where N iL is the molar flux
of component i in the liquid film , l is the liquid film thickness and Rk is the reaction rate of
reaction k , vik is the stoichiometric coefficient of species i in reaction k , r is the total number
of reactions. Rk is defined as the net rate of formation of the reference component whose
stoichiometric coefficient is unity in reaction k per unit time per unit volume of fluid.
dNiL r
= L vik Rk , i = 1,......, ncL (4.15)
dsL k =1
For the electrolyte system, the generalized Maxwell-Stefan equation which accounts for both the
chemical and electrical potential gradient (neglecting thermo-diffusion, pressure gradients and
external forces) is given by Eq. (4.16)
dxi Fxi zi d NiL ncL
N kL x
+ = +
dsL RT L dsL CTL kinL k =1 CT kikL i
k i (4.16)
NL ncL −1
1 1 N L
+ i
L k L
−
kikL
xk − L L , i = 1,.., nc − 1
i L
C
T k =1
k i
in C k
T in
where F is the Faraday constant, is the electrical potential, zi is the charge of species i , CTL is
the molarity of the mixture. Boundary conditions for Eqs. (4.15) are given by Eq (4.17) and are
67
taken at the interface ( sL = 0 ) .Boundary conditions for Eq (4.16) are given by Eq. (4.18) and are
taken at the liquid bulk ( sL = 1)
sL = 0 sL = 1
(4.17)
N = NiL , I , i = 1,......, ncL
i
L
xi n = xiB , i = 1,.., ncL − 1
where the values of the mole fractions at the liquid bulk, xiB are obtained from the conservation
equations in the liquid bulk.
xi ( sL = 0) = xiI , i SDF
(4.18)
NiL, I = 0, i SIN L
The values of the interfacial composition, xiI are determined from Eq. (4.11) and Eq. (4.13). The
mole fraction of the nth component not considered in Eq. (4.16) is computed from Eq. (4.19)
ncL −1
xn + xi = 1 (4.19)
i =1
in
For electrochemical systems, the electro-neutrality condition in the liquid film is satisfied by Eq.
(4.20) which also helps to calculate the electric potential gradient. The differential form of Eq.
(4.20) is better for numerical convergence allowing the possibility of variable and or adaptive
discretization.
ncL −1 ncL −1 ncL −1
dx
xn zn + xi zi = 0 or x (z i i − zn ) = − zn or ds ( z i
i − zn ) = 0 (4.20)
i =1 i =1 i =1 L
in in in
For dilute electrolyte systems characterized by the presence of excess solvent (e.g, n = H 2 O ), the
Maxwell-Stefan , Eq. (4.16) reduces11 to the Nernst-Planck equation as shown in Eq. (4.21)
dx Fx z d
NiL = −CTL kinL i + i Li + xi N n , i = 1,.., nc − 1
L L
(4.21)
dsL RT dsL
68
, where the first, second and third terms on the right-hand side of Eq. (4.21) correspond to the
diffusion , migration and convection terms respectively. As seen in Eq. (4.21), the interacting
diffusion coefficients which are not well characterized in the literature are eliminated. In this work,
both the generalized Maxwell-Stefan and Nernst -Planck equations are used to study the effect of
excess solvent assumption and access the risk in using the uncertain interacting diffusion
coefficients. When using the Nernst-Planck assumption, Eq. (4.16) is replaced with Eq.(4.22),
dxi Fxi zi d − NiL + xi N nL
+ L
= L L
, i = 1,.., ncL − 1 (4.22)
dsL RT dsL CT kin
where water must be chosen as the nth component. Reaction rate expressions are defined as given
in Eq. (4.23)
L
nc
v
ai i ,k
ncL E A, k
Rk = kk ai i ,k 1 − i =1 , k
k , fw = k k exp − , k = 1,..., r
v o
(4.23)
i =1 K eq ,k RT
fw
, where ( kk ) fw is the forward rate constant of reaction k , k ko is the frequency factor, E A,k is the
activation energy, Keq ,k is the equilibrium constant for reaction k , ai is the activity of the species
that is computed from the E-NRTL model.
4.3.4 Liquid film energy balance
The liquid film energy balance is given by Eq. (4.24)
dT L L nL L L
c
= Ni H i − L (4.24)
dsL L i =1
where L is the integration constant of the liquid film energy balance i.e the energy flux in the
liquid film, L is the thermal conductivity of the liquid . The heat transfer coefficient is not
69
adopted because the film thickness for mass and heat transfer are assumed to be the same114. The
boundary condition corresponding to Eq. (4.24) is given by Eq. (4.25)
T L ( sL = 1) = TLB (4.25)
where the bulk liquid temperature, TLB is obtained from the bulk phase energy balance. Additional
constraint is required to evaluate the integration constant, L such as the interface temperature as
given by Eq. (4.26)
T L ( sL = 0) = T I (4.26)
The interface temperature, T I is unknown (for phase equilibrium at the interface, TVI = TLI = T I )
and is computed using the continuity of energy fluxes as shown in Eq. (4.27)
L =v (4.27)
Constraints for the remaining end-node variables ( xi , NiL , T L ) not used as boundary conditions
that facilitate the connection of the two-film model to the bulk phases are defined as given in Eq.
(4.28)
xi ( sL = 0) = xiI , i S IN
L
xi ( sL = 1) = xiB i=n (4.28)
N ( sL = 1) = N , i = 1,..., n
i
L
i
B L
c
4.3.5 Degrees of freedom analysis
The degrees of freedom considering the number of variables as shown in Table 16 and the number
of equations as shown in Table 17 for the two-film model is :
(3n
V
c ) ( )
+ 4ncL + 2 ( ncL nd ) + 2nd + 4 − 2ncV + 3ncL + 2 ( ncL nd ) + 2nd + 2 = ncV + ncL + 2
70
Table 16 : Two- film model variables
Balance region Variable No. of Variables
Vapor Film yiI , yiB 2ncV
N iV ncV
v , TVB , T I 3
Liquid Film xi (n L
c − 1) nd
xn 1 nd
xiI , xiB 2ncL
NiL ncL nd
NiL, I , NiL, B 2ncL
d nd − 1
dsL
L , TLB 2
T L
1 nd
Total No. of Var. 3n + 4n + 2 ( ncL nd ) + 2nd + 4
V
c
L
c
where nd is the number of discrete points in the liquid film and is given by nd = 2 ncp + nfe + 1 .
The degrees of freedom are satisfied by the following specifications:
Vapor film : yiB (i = 1,..., ncV ), TVB
Liquid film : xiB (i = 1,..., ncL ), TLB
71
Table 17: Two film model equations
Equation No. of Equation No. of Equations
Equations
(4.8) ncv − 1 (4.18) n ( SDF ) + n ( SIN
L
) = ncL
(4.10) 1 (4.19) nd − 1
(4.11) n ( SDF ) (4.20) nd − 1
(4.12) n ( SIN
V
) (4.24) nd − 1
(4.13) n ( SDF ) (4.25) 1
(4.14) 1 (4.26) 1
(4.15) ncL ( nd − 1) (4.27) 1
(4.16) (n L
c − 1) ( nd − 1) (4.28) ncL + n ( SIN
L
) +1
(4.17) 2ncL − 1 Total : 2ncV + 3ncL + 2 ( ncL nd ) + 2nd + 2
4.3.6 Liquid film discretization
Fast reactions in the liquid film leads to convergence issues when using equidistant discretization
schemes . In this work, four different discretization options are developed to tackle the stiffness at
the boundaries of the liquid film due to reactions. Figure 15 shows the four discretization options:
(i) Chebyshev grid distribution, (ii) symmetric grid distribution, (i) left asymmetric grid
distribution and (iv) right asymmetric grid distribution. The symmetric and asymmetric options
have a density parameter that produces equidistant spacing when set to unity, but non-equidistant
spacing is obtained for values of the density parameter greater than 1 as shown in Figure 15B.
Asprion (2006)114 used the symmetric and asymmetric options for various regimes of reaction
kinetics, but in this work, the Chebyshev option led to more successful convergence of the models
and is used throughout.
72
(A)
(B)
Figure 15: Liquid film discretization options : (A) algorithm (B) plots
73
4.4 Conclusion
A rigorous two-film model for mass and energy fluxes across gas-liquid interfaces for reactive
absorption or desorption process that can accurately account for the acceleration of mass transfer
due to reactions in the liquid phase explicitly without the use of enhancement factors has been
developed. Various grid distribution algorithms are developed to obtain accurate results with
lower number of finite elements in the liquid film which helps reduce the computational cost and
improve convergence. Furthermore, a general method (GM) for calculating enhancement factor
over a wide range of reaction regimes is adopted . These two approaches for describing mass and
heat transport across the interphase are used in the next chapter for developing the packed column
model.
74
Chapter 5 Unit Operation Models Development and Validation for MEA-
Based CO2 capture
This chapter describes the development and validation of high-fidelity process units that are
computationally tractable for amine-based CO2 capture process in an equation-oriented platform
suitable for optimization. These process units include the absorber, stripper, rich-lean heat
exchanger, reboiler, and condenser unit models which are validated with multi-scale data from
multiple sources using the data from the wetted wall column (WWC) at the University of Texas,
Austin, data from the National Carbon Capture center (NCCC), and Technology Center Mongstad
(TCM).
5.1 Packed Column Model
The packed columns for absorption or striping are modeled as (i) the first-generation continuous
differential contactor model (Gen I) , (ii) the second-generation stage-wise contactor model (Gen
II) as shown in Figure 16. The Gen I model is based on the enhancement factor approach while
the Gen II model computes the interface fluxes explicitly based on the film theory. Details of each
model is presented next.
75
(A)
(B)
Figure 16: Implementation of packed column models in IDAES: (A) generation 1 CDC model
(B) generation 2 SWC model.
76
5.1.1 Gen 1- continuous differential contactor (CDC) packed column model
The Gen I packed model is built based on the IDAES control volume 1 dimension (CV1D). The
conservations equations for the material flows is given by Eq. (5.1)
M t , x , p , j Ft , x , p , j
L = fd + L Ntransfer ,t , x , p , j (5.1)
t x
where f d is a flow direction term, which allows for material flow to be defined in either
direction. The vapor stream is set to the forward direction while the liquid stream is set to the
backward direction. M t , x , p , j is the material holdup and is computed form Eq. (5.2)
M t , x , p , j = t , x , p , j A t , x , p (5.2)
where t , x , p , j is the density of component j in phase p at time t and location x . Ft , x , p , j is the
stream flow terms , Ntransfer ,t , x , p , j is the mass transfer term, t , x , p is the phase fraction which is equal
to unity as the control volume is set up for the phases separately, L is the length of the control
volume.
The energy conservation is given by Eq. (5.3),
Et , x, p H t , x, p
L = fd + L H transfer ,t , x + L Qtransfer ,t , x (5.3)
p t p x
where Et , x , p isteh energy holdup, H t , x , p is the enthalpy flow term, H transfer ,t , x is the enthalpy
transfer due to mass transfer and Qtransfer ,t , x is the heat transfer by conduction.
The pressure balance is given Eq. (5.4), where Pt , x is the pressure drop
Pt , x
0 = fd + L Pt , x (5.4)
x
77
Initialization strategy
The general column model is a 1D rate based model consisting of differential and algebraic
equations (DAE) and is solved in IDAES using the interior point algorithm, IPOPT 59. The packed
bed model was discretized using the Pyomo DAE finite difference method with 20 finite elements
resulting to 939 constraints and 7053 Expressions. For aiding to convergence especially in absence
of a ‘good’ initial guess, an initialization approach is developed. Two homotopy/continuation
parameters ( 1 , 2 ) are employed for the initialization procedure as shown in Figure 17. The
functions, ( g1 (x), g 2 (x), f(x) ) ,describe the model equations when the homotopy parameters
become zero or unity. The initialization routine for each packed bed in the column first solves only
mass balance equations by turning off the heat and mass transfer rate equations. Then, the
isothermal chemical absorption continuation parameter, 1 , is used to turn on the mass transfer
equations gradually with values ranging from 0 to 1. Subsequently, the adiabatic chemical
absorption continuation parameter 2 is used to turn on the heat transfer equations gradually with
values ranging from 0 to 1 to finish initializing each packed bed. The 4-stage initialization/solution
strategy takes approximately 65 CPU-seconds on an intel CoreTM i7-6500U processor to solve the
discretized packed bed model.
78
Figure 17: Packed bed initialization strategy
Model Validation
Considering data from NCCC, the prediction of CO2 percentage in the absorber as shown in
Table 18 compares well with the experimental data, given that the largest deviation is 4% of CO2
capture when compared with the data for the gas side. Details of these test runs can be found in
the Appendix H. It can be observed in Table 18 that Case 3 and 5 have the highest discrepancies.
Morgan et al (2018)37 used the same data for model validation with an Aspen Plus based model,
and found that when the uncertainty in composition measurement were taken into consideration,
model prediction for Case 5 (Case#K6 in that paper) became very close to the plant data. Future
works should incorporate and evaluate the effects of the measurement and model uncertainties on
the predicted results.
79
Table 18:Comparison of absorber model predictions for CO2 capture percentage with NCCC data
Case Model Gas-side Liquid -side L/G LLD
No. Prediction NCCC data NCCC data (mol/mol) (mol CO2/mol MEA)
1 99.98 99.91 89.45 3.75 0.14
2 99.88 99.49 93.26 6.25 0.25
3 73.89 78.57 70.86 1.75 0.08
4 99.99 99.53 90.69 3.72 0.11
5 67.60 59.03 58.89 3.54 0.35
6 99.97 98.07 93.92 6.67 0.15
7 53.50 55.48 52.30 1.68 0.24
Figure 18 shows the predicted liquid phase temperature profiles of the absorber compared with
NCCC pilot plant temperature measurements. The predicted temperature profile of case 3 (Figure
18A) where the L/G ratio is relatively low (1.75) agrees well with the measured data. In case 2,
the L/G ratio is relatively high (6.25) resulting in the temperature bulge at the bottom of the column
as shown in Figure 18B, which suggests an increased absorption rate at this packing height .12
Consequently, reactions in the liquid phase can deviate from the fast pseudo-first-order zone
(where E = Ha) as the concentration of the free MEA at the interface differs from the bulk
concentration. Hence, setting the enhancement factor to the Hatta number for absorption operation
in this region can lead to errors as seen in Bed1 of Figure 18B where it is observed that the GM
for the enhancement factor calculation differs significantly from the Hatta number. This behavior
has also been reported by other authors.107, 115 Overall, the model prediction shows good agreement
with the pilot plant data.
80
(A)
(B)
Figure 18: Comparison of absorber temperature profiles between the model and the NCCC data
for (A) Case 3 (B) Case 2 with Hatta number and GM model for enhancement factor calculation.
81
Similarly, the temperature profile of the stripper compares well with the measured NCCC data
for desorption of CO2 as shown in Figure 19A. Furthermore, the 2015 baseline data116 of the TCM
amine plant shown in Appendix H are also used to validate the absorber and stripper columns as
shown in Figure 19B. The packing type in TCM is Koch Glitsch Flexipac 2X. A specific surface
area of 225 m2/m3 and void fraction of 0.97 are used for the TCM packing. The temperature profile
predicted by the model for this specific test run compares well with the plant data as shown in
Figure 19B. For the corresponding CO2 loading of the rich solvent leaving the absorber, the model
has an absolute percentage deviation (APD) of 4.1 %. Comparison with the corresponding stripper
temperature profile is shown in Figure 19B. Once again, a good agreement with the experimental
data can be observed.
82
(A)
(B)
Figure 19: Comparison of model predictions with (A) NCCC Stripper data (B) TCM 2015
baseline data.
83
5.1.2 Gen 2- stage-wise contactor (SWC) packed column model
A modular approach is adopted for the gen 2 model where a staged gas-liquid contactor model is
first developed. The gen 2 model is a coupling of the staged gas-liquid contactor model to form a
column. The stage gas-liquid contactor model has three blocks: the vapor phase block which
contains the vapor bulk and vapor film, the liquid phase block which contains the liquid bulk and
liquid film and the hydraulic block which contains the hydrodynamics for the stage. The staged
gas-liquid contactor model uses the IDAES control volume 0 dimension (CV0D) for the vapor and
liquid bulk phases. The equations for the vapor and liquid films are presented in the previous
chapter. The material conservation equations for the bulk phases are given by Eq. (5.5)
M t , p , j
= Fin,t , p , j − Fout ,t , p , j + N kinectic ,t , p , j + Nreversible ,t , p , j + Ntransfer ,t , p, j (5.5)
t
where the material holdup , M t , p , j is given by Eq. (5.6)
M t , p , j = t , p , jVtt , p (5.6)
The rate of species generation by irreversible reactions, N kinectic ,t , p , j is given by Eq. (5.7)
N kinectic ,t , p , j = r , p , j X kinectic ,t ,r (5.7)
where r , p , j is the stoichiometric coefficient of component j in phase p for reaction r and
X kinectic,t ,r is the extent of the irreversible reaction. For reversible reactions, the net rate of species
generation , N reversible,t , p , j is given by Eq. (5.8)
N reversible,t , p , j = r , p , j Rr ,tVt (5.8)
r
84
where Rr ,t is the rate of the reversible reaction based on the reference component being referred
to for the forward rate constant and whose stoichiometric coefficient must be unity or adjusted to
unity.
The enthalpy balance for the bulk phases is given by Eq. (5.9)
Et , p
= H in ,t , p − H out ,t , p + Qtransfer ,t + H transfer ,t (5.9)
t p p
where is the Et , p is the energy holdup and is given by Eq. (5.10)
Et , p = U t , pVtt , p (5.10)
U t , p is the specific internal energy of phase p at time t .
Pressure balance for the bulks phase is given by Eq. (5.11)
0 = Pin,t − Pout ,t + Pt (5.11)
Using Pyomo DAE to discretize the liquid film with one collocation point (ncp = 1) and 10 finite
elements(nfe = 2). The incidence matrix of the stage model with the resulting 1152 variables and
1152 constraints is as a shown in Figure 20. Since there are no empty spots on the diagonal, the
incidence matrix is non-singular, and a feasible solution is expected to exist. Figure 21 shows the
initialization strategy for the gas-liquid contactor model. The vapor and liquid properties are
initialized using the vapor and liquid stream inputs respectively. Next, the velocities are computed
and used to initialize the hydraulics and mass transfer coefficients. Next, the liquid film sub-block
is solved by fixing the boundary conditions at the interface and at the bulk. At the liquid film
interface, the fluxes of non-volatile components are set to zero while that for the equilibrium
components are set to small values. The Interface temperature is fixed to that of the liquid input.
The liquid film is solved after fixing the bulk boundary condition as the liquid input.
85
Figure 20: Incidence matrix of the stage gas-liquid contactor (nfe=10, ncp=1)
The liquid bulk sub-block is initialized by fixing the mass and energy fluxes to that from the liquid
film. Next, the liquid film and liquid bulk blocks are connected to solve the liquid phase block.
Same approach is applied to the vapor phase block. Finally, the vapor phase and liquid phase
blocks are connected to solve the gas-liquid contactor model.
86
A
Figure 21 : Initialization strategy for gas-liquid contactor model: (A) Sequential solution of sub-
blocks (B) sequential solution of blocks
Figure 22 shows the stage model results for stage height of 0.05 and 0.25 m for MEA-CO2-H2O
system. In the liquid film, 0.0 is the interface and 1.0 is the bulk. At a stage height of 0.05 m, about
0.776 mol/s of CO2 is absorbed and the liquid bulk temperature is raised to 335 K while at stage
height of 0.25 m, about 1.698 mol/s of CO2 is absorbed and the liquid bulk temperature is raised
to 336 K.
87
A B
Concentration profile
Temperature profile
Stream flows
Figure 22: Stage model results : (A) stage height = 0.05 m (B) stage height = 0.25 m
88
Reboiler and Condenser Models
The reboiler and condenser models are developed as CV0D blocks for a single equilibrium stage
and the expressions for the chemical equilibrium, material balance for the true species, and the
phase equilibrium constraints, which govern the distribution of the components between the vapor
and liquid phases, are same as those of the packed column without the interface fluxes. The reboiler
model is validated together with the stripper model.
5.2 Plate Heat Exchanger Model
A mathematical model of a plate heat exchanger (PHE) as part of an amine scrubbing process for
post-combustion CO2 capture has been developed and implemented in the Institute for the Design
of Advanced Energy Systems (IDAES) process system engineering (PSE) framework. The thermal
model is based on the effectiveness-Number of Transfer Units (ε-NTU) approach where the non-
smooth min-max capacitance rates are replaced by smooth functions to improve convergence
characteristics of the optimization problem. Parameters for the convective heat transfer coefficient
correlations are estimated using the data from National Carbon Capture Center (NCCC). The
simulation results are found to be in good agreement with the pilot plant data. The model can be
utilized for optimizing PHE designs to maximize energy efficiency.
5.2.1 PHE Background
One of the major issues of the amine scrubbing processes is high operating cost due to the energy
demand for solvent regeneration. To reduce the energy requirements, heat from the lean solvent
leaving the stripper bottom is recovered for heating the rich solvent going to the stripper. Efficient
recovery of this heat is essential for reducing the energy requirements of amine scrubbing
processes. The industry standard for achieving this heat recovery is to use plate heat exchangers
89
(PHEs) that offer high heat transfer coefficient, variable heat transfer area easily accomplished by
adding/removing plates, and ease of cleaning. In the existing literature on post-combustion CO2
capture, the lean/rich solvent heat exchanger is commonly modeled as a counter-current shell and
tube heat exchanger. Harun et al. (2012) developed the model of a lean/rich solvent heat exchanger
using the counter-current shell and tube heat exchanger model in gPROMS’ Process Model
Library. The model assumed no phase change, and constant pressure. Flø et al (2015) modeled the
heat exchanger as a distributed shell and tube heat exchanger and validated the model with data
from an in-house pilot plant. Nittaya et al. (2014) developed the model of a countercurrent shell
and tube heat exchanger with one tube pass, but they did not present any validation studies. Several
models of PHEs have been developed for non-amine scrubbing applications. Georgiadis and
Machietto developed a model of a PHE for milk pasteurization by incorporating a protein fouling
118
model with a dynamic thermal model . The model was solved using gPROMS with the finite
difference approach. Dardour et al (2009) developed a linear system of first order differential
equations representing the number of channels in a PHE which was solved using the shooting
method. Gut and Pinto (2003) developed an algorithm to study the impact of PHE configuration
on its performance. For amine scrubbing processes, Lin and Rochelle (2017)121 developed a model
of a PHE, which was used for minimizing the cost of the exchanger by considering the velocity
and log mean temperature difference (LMTD) as the decision variables. To the best of our
knowledge, no detailed model of the PHE for amine scrubbing applications is currently available
in the open literature. In this work, a one-dimensional mathematical model of a PHE is developed.
The model is validated using pilot plant data. A parameter estimation framework is used to estimate
the optimal values of the model parameters.
90
5.2.2 PHE Mathematical model
An algebraic model of a PHE with plate dividers is developed by applying the ε-NTU method. The
ε-NTU method is computationally superior to the LMTD approach to predict unknown
temperature of a fluid 122. PHE is a common type of heat exchanger which consists of many plates
stacked together in a frame as shown in Figure 23(a). The plates are corrugated with gaskets at
the corners and pressed together to form a series of channels where the fluids flow alternatively
and exchange heat through the thin plates. A PHE configuration is determined by the number of
plates, their perforations and the inlet and outlet connections which define the flow distribution
(Gut and Pinto, 2003). For a given number of passes, P , and the number of channels in a pass,
NC ,a multi-pass flow configuration of a PHE can be simplified into a series of sub-heat
exchangers. Figure 23(c) is an example of a 4-pass-4-pass PHE configuration, where the series of
sub-heat exchangers are formed from adjacent channels in pass 2. For a PHE with plate dividers,
the sub-heat exchangers are in parallel or counter-current flow configuration depending on if P is
odd or even, respectively. For a large number of plates (greater than 35 plates), the effect of end
plates can be ignored. Hence, the outlet temperature of a pass can be predicted from an energy
balance of only one sub-heat exchanger in that pass. The ε-NTU method can be used to estimate
the outlet temperature of each sub heat exchangers where assumption of constant thermal
properties inherent in the derivation of the ε-NTU method is valid. The boundary conditions are
written to make them consistent with the modeling domain that comprises sub-heat exchangers
between passes. The PHE thermal model based on the ε-NTU method for each sub heat exchanger
consists of the following assumptions: (i) Steady state and adiabatic operation, (ii) Constant
physical properties in each sub-heat exchanger, (iii) No phase changes and fouling, and (iv) No
leakage of fluids. For a generic PHE with P number of passes and NCi number of channels for
91
pass i ( i I = 1, 2,...., P ), the governing equations for this approach are developed by
applying the energy balance to each sub-heat exchanger as shown in Table 20.
92
(a)
(b)
(c)
Figure 23: (a) Typical PHE layout (b) Chevron Plate (c) 4-pass-4-pass PHE showing sub-heat
exchangers in pass 2
93
5.2.3 PHE parameter estimation and validation
The convective heat transfer coefficient parameters are estimated since the optimal values of these
parameters depend on the fluid properties as well as the specifications of the PHE (e.g.
configuration, geometry and material of construction). The IDAES framework is used to estimate
parameters and uncertainties where the sIPOPT solver generates the covariance matrix for
parameter estimates via the reduced Hessian matrix.123 Data from the NCCC (Appendix H) were
used for parameter estimation (odd numbered cases) and model validation (even numbered cases).
The parameter estimation problem formulation is given by Eq. (5.12), where n is the number of
experiments, h(x; θ) and g(x; θ) are equality and inequality constraints in the PHE model,
min f obj = ( Qhexp
,i − Qh ,i )
n 2
cal
a1 , a2
i =1
s.t : h(x; θ) = 0 (5.12)
g (x; θ) 0
The optimal parameters are presented in
Table 19. The comparison of calculated vs plant heat duty for the PHE is presented in Figure 24 .
The calculated heat loads are found to have 4% error.
Table 19: PHE Parameter estimation result
Parameter Estimate SE
a1 0.4000 0.0094
a2 0.5746 0.3724
94
Figure 24: Parity plot for the PHE heat transferred
The PHE model predictions for the lean and rich solvent exit temperature are consistent with the
NCCC pilot plant data as shown in Figure 25A (Details of these test runs are in the Appendix H) .
A hot-end approach temperature as low as 4 ◦C can be obtained as shown in Figure 25B using
PHE. The PHE thermal model based on the ε-NTU approach converges reliably and is
computationally efficient.
95
(A)
(B)
Figure 25 : NCCC PHE predictions (A) Exit temperature (B) Channel temperature profile for
Case
96
Table 20: Plate heat exchanger thermal model
Sub-HX Equations Hot side(Lean solvent) Cold side(Rich solvent)
Energy Balance T i
=T i
−
i Cmin
i
(Thi,in − Tci,in ) T i
=T i
+
i Cmin
i
(Thi,in − Tci,in )
h , out h , in c , out c , in
Chi Cci
mh C p , h mc C p ,c
Capacitance rate Chi = Cci =
NCi NCi
( ) ( )
0.5
= min Cci , Chi 0.5 Cci + Chi − Cci − Chi + 1
2
i
Minimum capacitance Cmin
( ) ( )
0.5
= max Cci , Chi 0.5 Cci + Chi + Cci − Chi + 2
2
i
Maximum capacitance Cmax
i
Cmin
Capacitance ratio CR , i = i
Cmax
1 − exp − NTU i (1 − CR ,i )
,
1 − CR ,i exp − NTU i (1 − CR ,i ) CRi 1
c ,i ( P is EVEN ) =
NTU i
Effectiveness i = , CR ,i = 1,
NTU i + 1
1 − exp − NTU i (1 + CR ,i )
p ,i ( P is ODD ) =
1 + CR , i
U i Ap
Number of Transfer Units NTU i = i
Cmin
Change of pass Thi,out = Thi,+in1 , i = 1,..., P − 1 Tci,in = Tci,+out1 , i = 1,..., P − 1
Boundary Condition
Inlet Boundary conditions Thi,in = Th, IN , i = 1 Tci,in = Tc, IN , i = P
mh mc
Channel mass flow velocities Gch,i = Gcc,i =
bwNCi bwNCi
4mh 4mc
Port mass flow velocities h
Gport = c
Gport =
d port
2
dport
2
Pressure drop applied to both 2 f ( L + d port ) PGc2 G2
P = + 1.4 P port + g ( L + d port )
dh 2
sides
h C
a1 ( Rei ) ( Pri ) a1 ( Rei ) ( Pri )
h a2 h a3 C a2 C a3
Heat transfer coefficient hih = i hiC = i
dh dh
Overall heat transfer 1 1 p 1
= h+ + ,
coefficient U i hi p hiC
97
5.3 Conclusion
Process units for absorption, stripping, heat recovery in the rich-lean heat exchanger, reboiler, and
condenser have been developed and validated with lab-scale and pilot plant data. The models show
excellent agreement with the experimental data and will be used for optimization studies.
98
Chapter 6 Analysis and Optimization of CO2 Capture Process under Part-
Load and Variable Capture Operations
Existing power plants are frequently load following due to increasing penetration of renewables
into the grid. For power plants integrated with CO2 capture, optimal operation of the capture unit
at part-load and variable capture conditions can be exploited to reduce the operating costs. This
chapter presents insights into the performance of a reference MEA-based post-combustion CO2
capture unit under steady-state part-load and variable capture operation. The referenced plant-wide
model is based on the gen 1 column model, the rich-lean plate heat exchanger model and other
balance of plant developed in this work. Flue gas similar to pulverized coal and natural gas
combined cycle power plants are used for the optimization studies. Aspen Plus models are often
used to fix the hot-end temperature of the cross-heat exchanger while performing part load
optimization for the capture unit. Hence, Aspen models from CCSI are used to conduct analyses
on the stripping section of the capture plant and the results show variation in the hot-end
temperature of the cross-heat exchanger under part-load operation which corresponds to changing
the heat transfer area during operation. In this work, a rigorous plate heat exchanger model is
adopted without fixing the hot-end temperature approach for the optimization. Results from the
optimization study shows that the deviation of the liquid-to-gas flowrate ratio in the packed
column with respect to its optimal value results in a steep rise in the energy penalty. Hence,
optimal operation of an existing capture unit is crucial for minimizing the energy penalty under
part-load and variable capture operation.
6.1 Introduction
Post-combustion CO2 capture (PCC) can considerably reduce the amount of CO2 emitted from
industrial sources. However, integration of the amine-based capture plant with power plants incurs
99
an energy penalty resulting in as high as 22.9 percentage points decrease in the power plant
1, 124
efficiency. As power plants cycle more due to increased penetration of renewables into the
grid,15, 125 the downstream capture plant will face considerable disturbances due to rapid and large
changes in the flue gas flowrate and CO2 concentration.102, 126, 127 He et al. 102
observed that the
flue gas load variation at high frequency can have large impact on the CO2 capture rate. Mac
Dowell and Shah 128 presented a multi-period optimization framework that decouples the operation
of the power plant from the efficiency penalty imposed by the capture plant using flexible
operating strategies. Six distinct operating periods due to electricity price variation in a day were
identified. It was observed that the power plant operated relatively steadily for longer hours at a
given load factor. Thus steady-state optimization can be helpful for improving the economics of
the capture plant. Capital and operating costs of capture plants strongly depend on the
configuration. 8, 126, 129 For a given configuration, operating conditions play a vital role in reducing
the operating costs. Optimal operating conditions can greatly vary under flexible operation of the
capture plant at part-load condition depending on the plant configuration and operational
strategies. Studies considering the effect of flue gas CO2 composition on amine-based CO2 capture
34, 130, 131, 132
systems have been conducted by various authors. Mores et al.133 developed a
mathematical model for amine based post combustion CO2 capture and formulated an optimization
problem in the General Algebraic Modeling Systems (GAMS) platform to minimize the total
annual cost via simultaneous optimization of the operating conditions (solvent flowrates,
temperature and pressure) and dimensions of the equipment (absorber, stripper, heat exchanger
,pumps and compressors); however, the column model was simplified to use Murphree efficiency
to capture mass transfer limitations. The optimization was repeated for several CO2 removal targets
(70% -95%) for a fixed flue gas flowrate (10 kmol/s) and CO2 composition (4.22%). Hasan et.
100
al.131 used the RADFRAC model in Aspen Plus to minimize the total annualized cost of the CO2
capture process. The optimization was performed for different flue gas flowrates (0.1-10 kmol/s)
and CO2 compositions (1-70 vol%) to capture at least 90% of CO2. Oh and Kim134 developed an
equilibrium-based model for an MEA solvent in UniSim. They minimized the energy requirement
for different part-load operation to maintain 90% capture rate; however, rate-based models for
reactive absorption are preferable for optimization studies to consider mass transfer limitations. 11,
63, 96
Choi et al.135 used Unisim interfaced with Matlab for modeling the CO2 capture unit and
optimizing it for flue gases from non-power industries representing a variation in flue gas flowrate
and CO2 content. Feron et al.132 recently presented a study on the techno-economic performance
for CO2 capture above 90% from a natural gas combined cycle plant and a coal fired power plant.
The model was developed using ProTreat considering a fixed temperature approach of 20 °C in
the rich-lean cross heat exchanger. The review of the literature on plant-wide optimization of CO2
capture plants presented above identifies the following limitations that this work seeks to address:
(i) CO2 removal rate is set at 90% or varied within a narrow range while the capture rate is expected
to vary considerably under variable capture scenario while achieving a cumulative target capture
over a period of time, (ii) simultaneous variation of the flue gas flow rate, composition and capture
is typically not considered; however, under the flexible operation conditions, it is likely that the
power plant is load-following thus both the flowrate and composition of the flue gas are likely to
vary depending on the technology of the power plant, its design, fuel composition, operational
strategy, etc., (iii) several part-load optimization studies consider a fixed temperature approach in
the lean/rich heat exchanger but the temperature approach is likely to vary under flexible operation
when using an existing heat exchanger due to the fixed heat transfer area. In this work, a steady-
state plant-wide optimization is conducted to determine the optimum conditions under part-load
101
operations and variable capture rates and analysis of the perform with analysis on the temperature
approach of the rich-lean heat exchange of a given heat transfer area.
6.2 Configuration and operating range of the referenced MEA capture process
For the reference plant, the flue gas compositions considered are similar to that at the CO2
Technology Centre Mongstad (TCM) test facility. The TCM amine plant has the capacity of
treating up to about 60,000 Sm3/h of flue gas from either a natural gas turbine combined heat and
power (CHP) plant or a residual fluidized catalytic cracker (RFCC) unit 136. The CHP flue gas has
a CO2 composition of about 3-4 mol % comparable to NGCC while the RFCC flue gas contains a
higher CO2 composition (13-14 mol %) similar to a PC power plant as shown in Table 21 137. The
packing height and diameter for the columns of the reference plant are set to the CHP configuration
of the TCM amine plant. By retaining the 4-pass-4-pass Z-configuration of the NCCC PHE, the
total heat transfer area of the RL-HX for the reference plant is set to 308 m2 same as that at the
TCM plant103. This translates to a PHE with 28 channels per pass for each fluid. The design and
operating parameters of the amine plant for the part load and variable capture optimization studies
are shown in Table 22. Hence, given the aforementioned design and operating conditions of the
reference plant, the optimization search space is defined by the ranges of operation for flue gas
flowrate, CO2 capture percentage, lean solvent CO2 loading, and CO2 composition in flue gas as
given in Table 23.
The objective of the part load optimization is to minimize the operating cost (OPC) defined in
Eq.(6.1) , which includes the cost of steam consumed for the reboiler duty, ( Qreb ), the cost of
electricity required for the pumping duty ( Q pump ) of the lean and rich solvents, and the cost of
cooling water required for condenser and cooler cooling duties ( Qc ). Costs of steam, electricity
102
and cooling water, given by c s , ce , and cw , respectively, are shown in Table 24. These prices are
obtained from Turton et al. 138
Table 21: Typical Composition (mol%) of flue gas stream at TCM
Flue gas condition O2 N2+Ar CO2 H2 O
High CO2 composition ( yCO2 ,HIGH ) 3.2 77.9 14.7 4.2
Low CO2 composition ( yCO2 ,LOW ) 14.4 80.6 3.6 1.4
Table 22: Design and operating parameters of the reference plant
Design specification Value
Absorber diameter 3m
Absorber packing height 24 m
Stripper diameter 1.3 m
Stripper packing height 8m
Absorber/Stripper packing type MellapakPlus 252Y
PHE area of single plate 1.38 m2
PHE No. of passes 4
PHE channels per pass 28
Operating parameters Value
MEA concentration 30 wt%
Lean solvent Temperature 308.15 K
Absorber/Stripper Pressure 101.3/175 kPa
Condenser Temperature 303.15 K
Flue gas Temperature 301.15 K
Flue gas Base load (100%) 0.7049 kmol/s
103
Table 23:Ranges of key process parameters for optimization studies
Process parameter low high
Flue gas flowrate 60% base load 100% base load
CO2 capture percentage 75 % 90%
lean solvent CO2 loading 0.1 0.4
CO2 composition in flue gas 3.7 mol% 14.7 mol %
( )
OPC ( M$ / yr ) = cs Qreb + cwQc + ceQpump thr/yr (6.1)
Table 24 : Utility cost data (Obtained from Turton et al.133)
Utility Cost
Steam $2.03/GJ
Electricity $0.0674/kWh
Cooling $0.378/GJ
water
Costs of steam and cooling water are based off only the operating costs under the assumption
that the capital investment required to build the generation facility has already been depreciated133.
The selected steam price is based on the steam extracted from a low-pressure turbine (5 barg, 160
°C ) with credit for electricity using a natural gas fuel source133. The price of cooling water is the
utility cost for producing cooling water using a cooling tower with a supply and return temperature
of 30 °C and 45 °C, respectively. Obviously, these prices will change depending on the cost of
electricity used, location, design, technology, and other factors. The reader is referred to Turton et
al.133 for more details about the cost basis. The plant is assumed to operate for 8000 hours per year
( thr/yr ). Optimization results presented below are generated by Gen 1 model.
6.3 Analysis of RL-Heat Exchanger temperature approach on the reboiler duty
Rich solvent leaving the absorber needs to be efficiently heated for stripping CO2 while lean
solvent leaving the stripper needs to be cooled down to favor absorption. Thus, preheating the rich
104
solvent in the RL-HX reduces the reboiler duty. Consequently, it also reduces the cooling duty
for the lean solvent before it is sent to the absorber. The hot end temperature approach of the RL-
HX is often assumed to be fixed by several studies focused on operational optimization. The
motivation behind the study presented in this subsection is to evaluate the impact of this
assumption on the reboiler duty under off-design operation scenario. First, it should be noted that
during off-design operation, the CO2 loading in the rich and lean solvent as well as the solvent
flowrate can vary significantly. However, the RL-HX has a fixed area. It may be possible to change
the heat transfer area by adding or removing plates but that would require shutdown of the plant
and therefore changing the area of RL-HX is not considered to be feasible during plant operation.
Two largely different operating conditions are evaluated. Case I considers a case with low lean
CO2 loading while Case II considers a case with high lean CO2 loading. Other conditions
corresponding to these cases are shown in Figure 26A. For this study only the desorption section
including the RL-HX is simulated as shown in Figure 26A. The simulation is done by using the
models from the U.S. DOE’s Carbon Capture Simulation Initiative (CCSI) toolset available for
free download at: https://github.com/CCSI-Toolset/MEA_ssm.
Figure 26B shows the variation of design area with the hot-end temperature approach and the
variation of the reboiler duty with the hot-end temperature approach. It can be observed that at
low lean solvent loading, a decrease in the hot-end temperature approach in the range studied here
has only minor effect on the reboiler duty. This is because at very low lean loading, the dominant
contribution to the reboiler duty is the heat required for the endothermic reaction rather than the
sensible heat. On the other hand, for high CO2 lean loading which leads to higher circulation rate
such as in Case II, the reboiler duty varies significantly with the hot-end temperature approach as
the sensible heat requirement for heating the large flow of circulating solvent constitutes a larger
105
portion of the reboiler duty. In summary, the study shows that: (i) the design area would differ
depending on the design temperature approach and design operating conditions, and (ii) the impact
of the given design area (or the hot end temperature approach) on the reboiler duty depends on the
operating conditions. Hence, a rigorous RL-HX model with a given heat transfer area is more
appropriate for optimization under off-design operation than the fixed temperature approach.
106
(A)
(B)
Figure 26: Fixed temperature approach analysis showing: (A) PHE and stripper with operating
conditions for two specific cases, (B) PHE area and reboiler duty with respect to hot end
temperature approach for Case I: low lean load scenario and Case II: high lean loading scenario.
107
6.4 Effect of part load operation on the RL- Heat Exchanger temperature approach
Next results from the optimization study using models developed in this work are presented
starting with the effect of part load operation on temperature approach in the RL-heat exchanger.
The RL-HX is developed as a PHE with the design parameters given in Table 22. Figure 27 shows
the hot end temperature approach of the RL-HX under part load operation and varying L/G ratio
at 90% capture for flue gas with high and low CO2 composition. It can be observed that the
minimum Thot end at each part-load for both the flue gas with low and high CO2 content is similar
for the optimum L/G ratio. Also, a change in the flue gas load with high CO2 content from
100% - 60% results in a change of the minimum Thot end from 19 K to 16 K while that of the flue
gas with low CO2 content changed from 16 K to 14 K. For other operating points , Thot end varies
significantly depending on the lean loading. The reduction in the energy penalty of the capture
plant as a result of preheating the rich solvent in the RL-HX has a higher impact at higher lean
loading when the solvent circulation rate is higher thus requiring a higher amount of sensible heat.
108
Figure 27: RL-HX temperature approach under part load operation (60 -100% baseload) at 90%
CO2 capture.
6.5 Optimal operating conditions at base load
The L/G ratio is optimized by varying the lean solvent loading (LLD) to determine the circulation
rate required to maintain the absorber performance at the given capture rate. Figure 28 shows the
sensitivity of OPC with respect to the L/G ratio at 90% capture of low inlet CO2 concentration
flue gas at base load. The optimal L/G ratio is observed at about 1.78 mol/mol and the OPC
breakdown into reboiler duty, pumping duty and cooling duty is also shown. For a given capture
rate, a lower L/G ratio requires a lower lean loading to satisfy the desired extent of CO2 capture in
the absorber. Figure 28 shows that as the L/G ratio becomes smaller than about 1.78 mol/mol,
there is a steeper decrease in the desired lean loading. This leads to a steeper rise in the reboiler
duty and a consequent steeper rise in the OPC as shown in Figure 28. As the L/G ratio is increased
beyond the optimal value, the reboiler duty keeps increasing mainly due to the increasing
109
requirement of sensible heat. The cooler/condenser duty follows the similar trend as the reboiler,
i.e. rising steeply when L/G is lower than optimal and a gradual rise when the L/G ratio is higher
than the optimal value. Pumping cost monotonically keeps increasing with higher L/G ratio as
expected. Due to these cumulative effects, the OPC starts rising when the L/G is higher than
optimal.
Figure 29 shows the utility costs at 90 % capture for varying L/G ratio required to treat flue gas
with high CO2 inlet concentration at the base load. A higher optimum L/G ratio (~7.4 mol/mol)
for treating flue gas with 14.7 mol% CO2 inlet concentration is obtained when compared with flue
gas with 3.6 mol% composition. One key characteristic common to both Figure 28 and Figure 29
is that the OPC increases steeply below the optimal L/G ratio showing the impact the L/G ratio
has on the energy penalty. The study shows that operating the capture plant at optimal L/G ratio
will be highly desired for minimizing the operating cost.
110
Figure 28: Utility cost at 90% capture for varying L/G ratio at the base load of flue gas with low
CO2 inlet concentrations
Figure 29:Utility cost at 90% capture for varying L/G ratio at the base load of flue gas with high
CO2 inlet concentrations.
111
6.6 Coupled part load and variable capture operation
The studies conducted here considers steady-state part-load and variable capture operation. Figure
30 shows a bar graph for the optimal OPC vs variable capture for flue gas with low and high CO2
composition at three distinct loads of 100%, 80% and 60%. The optimal OPC keep rising with
higher percentage capture for both low concentration and high concentration flue gases. Similarly,
for a given percentage capture, optimal OPC keeps rising with higher load.
Table 25 provides a summary of the optimal L/G ratios for each load and variable percentage
capture. The flue gas with a higher CO2 content corresponding to a pulverized coal power plant
requires a higher L/G ratio compared to the NGCC flue gas with a lower CO2 content. For same
absorber column size and flue gas flow rate, the optimal required solvent flow rate at a given
removal target reduces when treating flue gas with a lower CO2 composition. It should be noted
that this study has been conducted by considering the same equipment dimensions for treating both
the high and low concentration flue gas and the plant hardware design is not necessarily optimal.
The optimal OPC is likely to vary if the plant equipment items are optimally designed by
considering the CO2 concentration in the feed flue gas.
Table 25: Optimal L/G (mol/mol) ratios under variable flue gas load and CO2 removal target
CO2 percentage capture (%)
Flue gas load (%)
75 90
60 (yCO2, Low) 1.41 1.74
80 (yCO2, Low) 1.41 1.76
100 (yCO2, Low) 1.42 1.78
60 (yCO2, High) 5.94 7.16
80 (yCO2, High) 6.05 7.29
100 (yCO2, High) 6.49 7.40
112
Figure 30: Optimal operating cost for variable flue gas load and percentage capture for (A) low
CO2 inlet composition (B) high CO2 inlet composition.
113
6.7 Conclusion
Steady-state plant-wide optimization of a refence capture unit under part-load operation and
variable capture rates is conducted with analysis on the performance of the reference RL-HX.
Analysis on the performance of the reference RL-HX showed that the hot end temperature
approach can considerably vary under off-design operation. Performance evaluation of the amine
reference plant shows that the optimal operating conditions at base load are suboptimal for part-
load and variable capture operation. The study shows that if L/G keeps becoming lower than the
optimal value, the energy penalty keeps becoming steeper. If the L/G ratio is higher than the
optimal, the energy penalty also keeps rising but the rise is less steep than when the L/G ratio is
lower than optimal. The optimal L/G ratio is found to be higher for capturing CO2 from a flue gas
with higher CO2 concentration. For a given load, as the percentage capture is increased, the optimal
L/G ratio is generally found to increase especially for the flue gas with higher concentration of
CO2. This study shows that optimal operation of an existing capture unit is crucial to minimizing
the parasitic energy loss in the capture plant under part-load and variable capture conditions. The
studies conducted here consider the same equipment size for both low and high concentration flue
gas. Furthermore, the equipment designs are not necessarily optimal. For improving the plant
economics further, it will also be desired to optimally design the equipment items with due
consideration of the part-load and variable capture operation as well as the CO2 concentration in
the incoming flue gas.
114
Chapter 7 Final Remarks and Future Work
7.1 Final Remarks
The IDAES computation platform is utilized to develop a detailed model of an amine-based CO2
capture process applicable to any solvent system. The initialization routine is found to lead to
successful convergence of this highly nonlinear system starting with initial guesses that are far
from the solution. Model results compare well with the data from the NCCC pilot plant, TCM pilot
plant and WWC experiments under considerable variation of solvent flowrate, gas flow rate, CO2
concentration in the gas, variable capture rate and steam flowrate to the reboiler as would be
expected under flexible capture operation.
A novel comprehensive thermodynamic sub-model applicable to both aqueous electrolyte and
non-electrolyte systems is presented where the reformulated e-NRTL model avoids numerical
singularities due to the absence of ionic species. The thermodynamic model describes the liquid
phase nonideality of the multicomponent MEA-H2O-CO2 system satisfactorily over a wide range
of operating conditions of practical importance.
A new summary data of valid ranges for the constants in Billet and Schultes correlations is
presented to aid column design for new packing and those not accounted for in the database using
the packing material type and specific area for the design of a two-phase countercurrent-flow
packed column for reactive absorption or desorption.
A rigorous and generic two-film model with support for both Nernst-Planck and Maxwell-Stefan
transport equations is developed and a staged gas-liquid contactor model is built to be the building
block for the Gen 2 model.
The Gen 1 column model, the reboiler model, and plate heat exchanger model along with all
properties sub-models have been uploaded on the IDAES GitHub page. The staged gas-liquid
115
contactor model along with the rigorous-two film will be available for future release on IDAES
GitHub website.
Despite highly developed computing power and numerical methods, the application of Gen 2
model is still faced with numerical issues, and this has limited its applicability to industrial
problems. Therefore, Gen 2 model provided in this work will facilitate a better understanding of
multi-component reactive absorption to provide a reliable and robust model for optimization
studies.
7.2 Recommendations for Future Work
Future work will include coupling of the stage gas-liquid contactor model to build the gen 2 stage
column. For the property sub models, the determination of uncertainties in the point estimate of
thermodynamic parameters is difficulty in the presences of collinearities. Tracing the source of
these collinearity is desired, one likely area to consider in this regard is the expression for the
energy interaction parameters which has same expression as the polynomial form of the
equilibrium constant. Also, physical properties for the electrolyte system used in this work are
based on apparent species which are obtained through true to apparent species conversion. The
generic packed columns developed in this work can be used to update existing properties for
electrolytes for applicability to amine systems such as the method of Andrade equations for
calculating the viscosities of electrolyte mixtures, the Clarke equation for calculating the molar
volume of electrolyte species, etc.
The two-film model increases the sensitivity of the fluxes with respect to the liquid phase binary
diffusivities which are not well characterized in the literature. Hence, parameter estimation studies
can be conducted to estimate these parameters with plant data. The liquid phase diffusivities are
very vital for accurate computation of the liquid film thickness. The two-film model is very
116
sensitive to the liquid film thickness which varies from micro scale to nano scale. In this work,
the film thickness is determined from mass transfer correlations and the properties of the mixture.
However, Sherwood number can be investigated for better prediction of the film thickness. A
dynamic model is particularly suitable for study on flexible operation especially when the plant
undergoes frequent and rapid transition of part-load and flexible capture operation. Future work
will validate the stage-wise columns with dynamic data to support flexible operation.
117
Appendix A : Virial equation of state
First and second Temperature derivatives of mixture second virial coefficient
dB Tc ,i dBij0 dBij1
= R yi y j + i (A1)
dT Pc ,i dTr ,ij dTr ,ij
d 2B Tc ,i d 2 Bij0 d 2 Bij1
= R yi y j + i (A2)
dT 2 Pc ,i dTr2,ij dTr2,ij
Temperature derivative of partial molar log fugacity coefficient
ln ˆk P B − 2 yi Bi ,k 1 dB dB
= + 2 yi ik − (A3)
T P , x R T dT dT
2
T
Temperature derivative of partial residual Gibbs
GkR ln ˆk
= R T + ln ˆk (A4)
T P , x T
118
Appendix B : Excess Gibbs energy model
The asymmetric convention is desired for ionic species in aqueous systems, hence the unsymmetric
excess Gibbs’s energy, G**,ex / RT is given by Eq. (B1)
G**,ex G**,ex, SR G*,ex, LR G ex, Born
= + + (B1)
RT RT RT RT
The un-symmetric local interaction contribution, G**,ex , SR / RT based on infinite dilution in the
aqueous solutions for ions and henry components is given by Eq. (B2)
G**,ex , SR G ex , SR
= − xc ln c − xa ln a − xh ln h (B2)
RT RT c Sc a S a h ( S m − S s )
Where ln i for i = c, a, h , represents the infinite-dilution activity coefficients of cations, anions,
and henry components respectively and the expressions are provided in Appendix C (Eq. (C9)).
The symmetric short-range contribution of the excess Gibbs energy, G ex , SR / RT , for the
multicomponent system is given by Eq. (B3)
G ex , SR
X i Gim im XG i ic ic XG i ia ia
= Xm + Xc + Xa
iSmac iSma iSmc
(B3)
RT m Sm
iSmac
X i Gim cSc XG
iSma
i ic aS a XG
iSmc
i ia
The long-range unsymmetric PDH term, G*,ex , LR / RT for the mixed solvent is given by Eq. (B4)
G*,ex , LR
= − x x ln (1 + I 1/2
x )
4A I
(B4)
RT
Where is the closest approach parameter, I x is the ionic strength that captures the composition
dependence in terms of mole fraction as given by Eq. (B5)
1
Ix =
2 iSca
xi Z i2 (B5)
119
Ax is the Debye-Huckel parameter on mole fraction basis given by Eq. (B6)
3/2
1 1/2 e2
Ax = ( 2 N A s ) (B6)
3 4 o s kT
Where s is the solvent dielectric constant(or relative permittivity of the solvent) and is extended
to mixed solvents by simple mixing rules as given by Eq. (B7) - Eq.(B8)
s = wj j (B7)
jS s
M j xj
wj = , j Ss (B8)
M
k S s
x
k k
Where j is the dielectric constant of pure solvent j which is calculated from Eq. (C14), M j is
the molecular weight of solvent j. The model density, s (mol/m3 ) is extended to mixed solvents
as given by Eq. (B9)
1
s
= xV
jS s
j j (B9)
Where V j (m3 /mol) is the molar volume of pure solvent j and is calculated from Eq, (E7)
x is the solute-free mole fraction of the solvent computed from Eq. (C11)
The values of the PDH parameters are summarized in .
120
Table 26 :PDH parameters
Parameter value
NA 6.022137 1023 mol−1
o 8.854 10−12 C2 N −1m −1
k 1.38110−23 J/K
e 1.602 10−19 C
14.9
All ions have a unit charge in the aqueous amine
Zi
system
The excess Gibbs energy Born term, G ex , Born / RT which corrects the difference between the
dielectric constant for water and the mixed solvent is given by Eq. (B10)
G ex , Born e2 1 1
= − xi Zi (B10)
2
RT 800 o rB kT s w i
121
Appendix C : Activity coefficient model
Solvents
The symmetric activity coefficients i of the solvent species in the symmetric scale is normalized
as the pure component reference frame same as the asymmetric convention. Hence, activity
coefficients for the solvent species is defined as in Eq. (C1)
ln i = ln iSR + ln i, PDH , i Ss (C1)
Solutes
For chemical equilibrium, the unsymmetric activity coefficients of ionic and molecular solutes
are based on the aqueous phase reference state and denoted as i** , While for phase equilibrium,
the unsymmetric activity coefficients of molecular solutes are based on the mixed solvent
reference state and denoted as i . These are computed from Eq (C2). The born term contribution
applies only to the ionic species (goes to zero for molecular solutes),
ln i** = ln i**, SR + ln i, PDH + ln iBorn , i Smac − S s
(C2)
ln i* = ln i*, SR + ln i, PDH , i Sm − Ss
The short-range contribution i**,SR is computed from Eq (C3),
ln i**,SR = ln iSR − ln iSR ,
,H2O , i Smac − Ss (C3)
For the henry components, the short range contribution i*,SR based on the mixed solvent
reference state as required in phase equilibrium is computed from Eq. (C4)
ln i*,SR = ln iSR − ln iSR ,
,mix , i Sm − Ss (C4)
122
The infinite dilution activity coefficients in water, iSR ,
,H2O are defined in Eq. C9 while the infinite
dilution activity coefficient in the mixed solvents iSR ,
, mix for the henry component is defined in
Eq.C10
Short Range Contribution
The short –range interactions contribution is based on local composition of three types of
neighborhood species: cations, anions, and molecular species. The effective local composition of
species in each type of neighborhood is given by Eq. (C5)
1 i Sm 0 i Sm
X i = xi zi zi = , note : in PDH model, Zi = (C5)
zi i Sca zi i Sca
where xi is the true species mole fraction and zi is the ionic charge.
The expression for the natural logarithm of the symmetric activity coefficient, ln iSR of molecular
species, cations and anions are given in Eq. C6-C8 respectively,
For m S m :
X i Gim im
X mGmm
X iGim im
+
iSmac iSmac
ln SR
= mm −
mS m X i Gim X i Gim
m
X i Gim
iSmac iSmac iSmac
X c Gmc
X iGic ic
+
iSma
mc − X i Gic
(C6)
cSc X i Gic
iS ma iS ma
X a Gma
X iGia ia
+
iSmc
ma − X i Gia
aS a X i Gia
iSmc iSmc
123
For c S c :
1
X iGic ic
iSma
ln c =
SR
zc X iGic
iSma
X mGcm
X iGim im
+
iSmac
cm − X i Gim
(C7)
mSm X i Gim
iSmac iS mac
X a Gca
X iGia ia
+
iSmc
ca − X i Gia
aSa X i Gia
iSmc iSmc
For a Sa :
1
XG
iSmc
i ia ia
ln aSR =
za XG
iSmc
i ia
X mGam
X iGim im
+
iSmac
am − (C8)
mSm X i Gim X iGim
iSmac iSmac
X c Gac
X iGic ic
+
iSma
ac − X i Gic
cSc X i Gic
iSma iSma
The infinite dilution activity coefficients of the ionic and molecular solutes based on the aqueous
phase reference state is defined in Eq. (C9)
ln mSR,H,2O = ln mSR ( xw → 1) = ( wm + Gmw mw )
ln cSR,H,2O = ln cSR ( xw → 1) = zc ( wc + Gcw cw ) (C9)
ln aSR,H,2O = ln aSR ( xw → 1) = za ( wa + Gaw aw )
The infinite dilution activity coefficients of the henry components (molecular solutes) based on
the mixed solvent reference state is defined in Eq. C10,
124
ln mSR,mix
,
= ln mSR ( xa → 0, xc → 0, xm → 0 )
xG i
im im
xG
xiGis is
(C10)
+ s ms
iS s iS s
= ms −
xG i im sS s xiGis
xiGis
iS s iS s iS s
Where xs is the mole fraction of the solvent species in the solute free solvent mixture as given
by Eq.
xi
xi = , i S s (C11)
x
j Ss
j
Long Range Contribution
The expression for the natural logarithm of the activity coefficient contribution of the un-
symmetric PDH model is given by Eq . (C12)
2Z 2 Z 2 I 1/2 − 2 I 3/2
ln i, PDH = − Ax i ln (1 + I x1/2 ) + i x 1/2x , i Smac (C12)
1+ Ix
For mixed solvent systems, the reference state of the PDH model at infinite dilution in the mixed
solvent is converted to infinite dilution in the aqueous solution by the Born approximation term as
given in Eq. (C13)
e2 Zi2 1 1
ln Born
= − , i Sac (C13)
800 o rB kT s w
i
125
Symmetric E-NRTL Parameter Structure
The adjustable parameters for the symmetric E-NRTL model include:
• Long-range parameters
o dielectric constant coefficient of the pure component solvent
o ionic radius of species in the Born term and approach parameter
• short-range interaction parameters
o molecule-molecule interaction parameters
o molecule-salt interaction parameters
o salt-salt interaction parameters
For the long-range parameters, the dielectric constant for the solvent species is calculated from Eq.
(C13) and for CO2-MEA-H2O systems, the parameters are given in Table 27. The ionic radius of
−10
species usually has values within the range 2 to 6 10 m .139 They are usually assigned a default
value of 3 Å.
1 1
j = Aj + Bj − , j Ss (C14)
T C j
126
Table 27: Dielectric parameters for solvent species
Solvent, j Aj Bj C j Source
Ikada (1969)Ikada,
Hida, Okamoto,
MEA 35.76 14836 273.15
Hagino and Koizumi
140
Hilliard(2008)Hilliard
MEA 31.07 15128 298.15
39
Que et. al(2011)Que
Water 78.51 31989 298
and Chen 141
Table 28 shows the parameter structure required for the short-range contribution which includes
the non-randomness parameters, ij , the energy interaction parameters, ij , and the Boltzmann
kind factors, Gij .
127
Table 28 :Symmetric E-NRTL parameter structure
, , G,
s/n Interactions
Nonrandomness Energy interaction Boltzmann kind factors
1 m-m' mm mm = A mm + Bmm /T Gmm = exp( − mm mm )
2 m'-m mm mm = A mm + Bmm /T Gmm = exp( − mm mm )
3 m-m mm mm = 0 Gmm = exp(− mm mm ) = 1
4 ca-m ca ,m ca ,m = Cca ,m + Dca ,m /T Gca ,m = exp(− ca ,m ca ,m )
5 m-ca m ,ca m,ca = Cm,ca + Dm,ca /T Gm,ca = exp(− m,ca m,ca )
6 ca-ca' ca ,ca ca ,ca = Eca,ca Gca ,ca = exp(− ca ,ca ca ,ca )
7 ca-c'a ca ,ca ca ,ca = Eca ,ca Gca ,ca = exp(− ca ,ca ca ,ca )
8 cm cm = Ya ca ,m cm = − ln ( Gcm ) /cm Gcm = Y G
aSa
a ca , m
a
9 am am = Yc ca ,m am = − ln ( Gam ) / am Gam = Y G c ca , m
c cSc
10 mc mc = Ya ca ,m mc = − ln ( Gmc ) /mc Gmc = Y G
aS a
a m , ca
a
11 ma ma = Yc ca ,m ma = − ln ( Gma ) /ma Gma = YcGm ,ca
c cSc
12 ca ca = Yc ca ,ca ca = − ln ( Gca ) /ca Gca = YcGca ,ca
c c
13 ac ac = Ya ca ,ca ac = − ln ( Gac ) / ac Gac = YaGca ,ca
a a
128
The default settings for the short-range interactions parameters for MEA-H2O-CO2 systems as
given by Zhang et al(2011)35 are shown in Table 29
Table 29 : Default parameters35 settings for MEA-H2O-CO2 system
, ,
s/n Interactions
Nonrandomness Energy interaction
1 m-m' mm = 0.2 mm =0
2 m'-m mm = 0.2 mm = 0
3 m-m mm = 0.2 mm = 0
4 ca-m ca ,m = 0.2 ca ,m = −4
5 m-ca m,ca = 0.2 m,ca = 8
6 ca-ca' ca ,ca = 0.2 ca ,ca = 0
7 ca-c'a ca ,ca = 0.2 ca ,ca = 0
The fitted binary interaction parameters for MEA-H2O-CO2 systems by Zhang et. Al(2011)35 are
given in Table 30
Table 30: Fitted NRTL parameters
Component i Component j 1,ij 2,ij
Molecule-Molecule A mm Bmm
H2O MEA 0.1559 110.80
MEA H2O 1.5201 -910.30
Molecule-Salt Cm,ca Dm,ca
H2O MEAH + , MEACOO − 6.732 -9.2407
+ −
H2O MEAH , HCO 3 8.5721 0.0
Salt-Molecule Cca ,m Dca ,m
MEAH + , MEACOO − H2O -3.1628 39.498
MEAH + , HCO3− H2O -4.0092 0.0
129
Appendix D : Excess enthalpy derivation
Analytical expression for the excess enthalpy is derived by taking the partial derivative of the un-
symmetric excess Gibbs energy expression given in eq. with respect to temperature at constant
pressure and composition as shown in eq. (D1),
(G**,ex / RT )
H ex = − RT 2 (D1)
T P, x
This implies taking temperature derivative of the excess Gibbs energy expressions at constant
pressure and composition for the short-range, long-range, and born term contribution are shown in
eq. (D2),
(G**,ex ,SR / RT ) (G*,ex , LR / RT ) (G ex , Born / RT )
H ex = − RT 2 + + (D2)
T T T P, x
P, x P, x
Born Term
e 2 xi Z i2
(G ex , Born / RT ) (1/ sT ) (1/ wT )
= i
−
T P , x 800 o rB k T T
(D3)
e 2 xi Z i2
1 ( wT ) 1 ( sT )
= i
−
800 o rB k ( wT ) 2
T ( sT ) 2 T
Based on eq. and eq,
T
sT = wj jT , jT = Aj T + Bj − , j Ss (D4)
jS s C j
130
Therefore,
( sT ) ( jT ) 1
T
= w j
T
= wj Aj −
C j
jS s jS s (D5)
( wT ) 1
= Aw − ,
T Cw
Substituting eq. (D5) into eq. (D3), gives,
1 1
(G ex , Born / RT )
e2 xi Z i2 AH2O − C w A − C
j j
H2O jS s
= −
j
i
(D6)
T P, x 800 o rB k ( H2OT ) 2 ( sT ) 2
Long-Range Term
(G*,ex, LR / RT ) Ax
= − x ln (1 + I 1/2
x )
4I
(D7)
T P, x T P , x
From the definition of Ax in eq.,
( ( sT )−3/2 ) ( s1/2 )
3/2
Ax 1 1/2 e2
T = s
1/2
+ ( sT ) −3/2
, = ( 2 N A ) (D8)
P, x T T
3 4 o k
Therefore,
Ax 1/2 3 −5/2 ( sT ) 1 ( s )
T = s − 2 ( sT ) T
+ ( sT ) −3/2 s−1/2
2 T
(D9)
P,x
The temperature derivative of the molar density of the solvent mixture as defined in eq. (D10) is,
s V j
= − s2 xj (D10)
T jS s T
The derivative of the pure solvent molar volume using from(E7), is as given in eq. C11,
131
V j − M j ( 2a jT + b j ) V j ( 2a jT + b j )
= =− (D11)
T (a T + b jT + c j ) a jT 2 + b jT + c j
2 2
j
Therefore,
s V j ( 2a jT + b j )
= s2 xj (D12)
T jS s a j T 2 + b jT + c j
The pure solvent molar volume parameters for MEA-H2O-CO2 system are shown in Table D3.
Short Range Term
The partial derivative of the SR term as given in Eq. with respect to temperature at constant
pressure and composition is given in eq (D13),
(G **,ex , SR / RT ) (G ex , SR / RT ) ln cSR,H,2O
= − xc
T P,x T P,x c Sc T P , x
(D13)
ln aSR,H,2O ln hSR,H,2O
− xa − xh
a Sa T P , x h ( Sm − Ss ) T P , x
The temperature dependency of the terms on the left hand side of eq (D13) comes from the binary
energy interaction parameters, ( im , ic , ia ) and the Boltzmann kind factors, (Gim , Gic ,Gia ) when a
central molecular, cationic or anionic component is surrounded by other species according to the
like-ion repulsion and the local electroneutrality assumptions. Hence, the temperature derivatives
are presented in eq. (D14) to (D19),
Boltzmann kind factors:
132
B G
−im im im im2 im if i = m
e if i = m T
G
Gim = Ya Gia ,m if i = c , im = (1 T 2 ) Ya Dia ,m ia ,mGia ,m if i = c (D14)
aSa T P , x aSa
YG
c c ci , m if i = a
(1 T ) c Yc Dci ,m ci ,mGci ,m if i = a
2
Sc
Sc
Ya Gi ,ca if i = m (1 T 2 ) Ya Di ,ca i ,ca Gi ,ca if i = m
aSa Gic
Gic = , = aSa (D15)
a ci ,ca
Y G if i = a T P, x
0 if i = a
aSa
YcGi ,ca if i = m (1 T 2 ) Yc Di ,ca i ,ca Gi ,ca if i = m
cSc Gia
Gia = , = cSc (D16)
c ia ,ca
Y G if i = c T P,x 0 if i = c
cSc
Binary energy interactions:
Aim + Bim / T if i = m
im = ,
− ln ( Gim ) / im if i = c, a
B
− im2 if i = m
T (D17)
im 1
2 a ia , m ia , m ia , m
T = − YD G if i = c
P,x im GimT aS a
1
2 c ci , m ci , m ci , m
− YD G if i = a
imGimT cSc
ic = − ln ( Gic ) / ic , i = m, a
1
ic − Y D i ,ca Gi ,ca if i = m (D18)
T = ic GicT
2 a i , ca
aSa
P, x
0 if i = a
133
ia = − ln ( Gia ) / ia , i = m, c
1
ia − Y D i ,ca Gi ,ca if i = m (D19)
T = ia GiaT
2 c i , ca
cSc
P,x
0 if i = c
The contribution to the excess enthalpy from the symmetric excess Gibbs energy comes from three
terms : m where a molecular component is at the center, c where a cationic species is at the
center, and a where an anionic species is at the center as shown in eq. (D20)
(G ex , SR / RT )
T = m + c + a (D20)
P, x
These terms ( m , c , a ) are defined in eqs. (D21) to (D23) ,
Gim
X i Gim im + im Gim
X i Gim im X i
iS T T T
m = X m mac − mac mac
iS iS
(D21)
m Sm X iGim
2
iSmac X i Gim
iSmac
Gic
X i Gic ic + ic Gic X i Gic ic X i
iSma T T iSma T
iS ma
c = Xc − (D22)
c Sc X G
2
X i Gic
i ic
iSma
iSma
Gia
X i Gia ia + ia Gia
i ia ia i
X G X
iS T T T
iSmc iSmc
a = X a mc − (D23)
a Sa X iGia
2
iSmc X i Gia
iSmc
The temperature derivative of the infinite dilution activity coefficient of the solutes based on eq.
is given as,
134
ln cSR,H,2O wc G
= zc + Gcw cw + cw cw (D24)
T P , x T T T
ln aSR,H,2O wa G
= za + Gaw aw + aw aw (D25)
T P , x T T T
ln h,H, SR2O wh G
= + Ghw hw + hw hw (D26)
T P , x T T T
135
Appendix E : MEA-H2O-CO2 Correlations for Phase and Chemical Equilibria
Saturation Pressure
The saturation pressure is calculated from the extended Antoine equation as given in eq (E1), and
the parameters are shown in Table 31,
C2
ln Pi sat = C1 + + C3 ln T + C4T 2 , i Ss (E1)
T
Table 31 : Extended Antoine Vapor pressure Parameters
Solvent C1 C2 C3 C4 Source
MEA, Hilliard(2008)Hilliard
165.87 -13492 -21.9 1.3800e-5
ln Pi sat (kPa) 39
Que and
sat
H2O, ln Pi (Pa) 73.649 -7258.2 -7.3037 4.1653e-6 Chen(2011)Que and
Chen 141
Henry constant
The Henry’s constant, H i ,mix for CO2 in the mixed solvent of water and MEA is calculated from
eq. (E2), 142
H
ln H i ,mix = ln i + ws ln is , i = CO2 (E2)
sSs is
In this work , a reduced expression for the weighting factor is presented as shown in eq. (E3),
xs
ws = s S s , (E3)
xssj
jS s
136
Where jk is defined in eq. (E4) and depends on the solvent density and the characteristic volume
of the solute and solvents as computed from Brelvi-O’Connell model42
2/3
Vi, j
jk = (E4)
Vi ,k
For the MEA-H2O-CO2 system , the reduced expression for jk (R2=0.9997) is given by eq.
(E5) . This is obtained from the Brelvi-O’Connell model(see eq. (E8)) over the temperature range
of 301.15 to 393.15 using parameters in Table D4 and the solvent density in eq. (E7).
1
MEA,H O = 8.7105 − 1.324ln T (K), MEA,H O = , MEA,MEA = H2O,H2O = 1 (E5)
2 2
H O,MEA
2
xs is the mole fraction of the solvent on solute free- basis, i = iSR , is infinite dilution activity
coefficient of molecular solute in the mixed solvent, is is the infinite dilution activity coefficient
of molecular solute in pure solvent s, H is is the Henry’s constant of the solute in pure solvent as
given in eq. (E6) for MEA and water,
bis Vis ( P − Pi sat )
ln H is = ais + + cis ln(T ) + disT + , i = CO2 , s = H 2O
T RT (E6)
896.5
H is = 6.6434 108 exp − , i = CO2 , s = MEA
T
The CO2-solvent pair parameters of Henry’s constant in eq. (E6) are shown in Table 32,
137
Table 32 : Parameters of CO2 Henry's constant in water, Pa
Solvent species ais bis cis dis source
H2O 170.713 -8477.7 -21.957 0.00578 Chen et. al (1979)143
Yan and
H2O 100.65 -6167.7 -10.191 -0.01
Chen(2010)144
Zhang and
H2O 91.344 -5876 -8.598 -0.012
Chen(2010)145
Pure Component Molar Volume
The molar density of the pure solvent is obtained from the molar volume, Vs which is calculated
from eq. (E7), where the parameters for the molar volume are given in Table 33,
Ms
Vs = (E7)
asT + bsT + cs
2
Table 33 : Molar volume parameters74, m3/mol
Solvent, s as bs cs Ms
MEA −5.35162 10−1 −4.51417 102 1.19451106 61.08
H2O −3.2484 1.65 103 7.93 105 18.02
138
Partial molar volume of molecular solute at infinite dilution 42
1 − Cis
Vis = ,
ks
0.62
v
Cis = − i exp ( −2.4467 + 2.12074 sr ) , 2.0 sr= H2O 2.785
vs
0.62
v
Cis = − i exp ( 3.02214 − 1.87085 sr + 0.71955( sr ) 2 ) , 2.785 sr= MEA (E8)
vs
ks = s exp s A + S ( B − C S ) − 1 , A = −0.42704, B = 2.089, C = 0.42367
sr = s vs , S = sr − 1
s is the molar density in mol/cm3, hence the characteristic volume should be converted to similar
units to compute the reduced density, sr . The characteristic volume, vi is computed using the
Brelvi-O’Connell characteristic volume parameters in Table 34.
Table 34: Characteristic Volume Parameters (m3/kmol)
Species v1,BO
i v2,BOi Source
Brelvi and
H2O 0.0464 0.0 O’Connell(1972)Brelvi
and O'Connell 42
Zhang et. al(2011)Zhang,
MEA 0.225 0.0
Que and Chen 35
Yan and Chen(2010)Yan
CO2 0.177 -3.42e-4
and Chen 144
vi = v1,BOi + v2,BOi T , i = CO2 , MEA, H2O
139
Chemical Equilibrium Constant
The basic equilibrium reactions9, 54 involved in the MEA-H2O-CO2 system is given by eqs (E9) to
(E13),
2H2 O ⎯⎯
Kw
→ H3O+ + OH− (E9)
MEAH+ + H2 O ⎯⎯
Kam
→ MEA + H3O+ (E10)
MEACOO− + H2 O ⎯⎯
Kcarb
→ MEA + HCO3− (E11)
(E12)
K
CO 2 + 2H 2 O ⎯⎯→
CO2
H 3O + + HCO3−
HCO3− + H2 O ⎯⎯⎯
Kbicarb
→ H3O+ + CO32− (E13)
+ − 2−
Morgan et al (2017) have shown that H3O , OH , CO3 ionic species can be neglected at the
process conditions of industrial importance. Hence, they used a reduced set of reactions that
represents the chemistry of the CO2 loaded mono-ethanolamine solution adequately as given in
eqs. (E14) to (E15),
R1: 2MEA + CO2 ⎯
K1
→ MEAH + + MEACOO− (E14)
R2: MEA + CO2 + H 2O ⎯→
K2
MEAH + + HCO3− (E15)
The matrices N, N 0 , J , v, ε considering CO2 as the reference component in the reduced solution
chemistry for the MEA-H2O- CO2 system are given as follows:
nCO2 1 0 0 −1 −1
0 −2 −1
nMEA 1 0
n0,CO2 nH O
, J = 0 0 1 0 −1 1 0 0
N 0 = n0,MEA , N = , ν = , ε =
2
nHCO− 0 0 0 0 1 2 0 0
n0,H O
2
3
0 0 0 1 1
nMEAH+
n 0 0 0 1 0
MEACOO−
140
The aqueous phase speciation model is developed using the un-symmetric aqueous phase activity
coefficient for the molecular solutes and all ionic species for consistency with the equilibrium
constants.54 The special model obtained after some algebraic manipulation to eliminate 1 and 2
is given by Eqs.(E16)- (E21),
xMEACOO− xMEAH+ MEACOO
*
−
*
MEAH +
K1 = (E16)
xCO2 ( xMEA ) ( MEA )
2 2
CO
* **
2
xMEAH+ xHCO− MEAH
**
+
**
HCO −
K2 = 3 3
(E17)
xCO2 xH2O xMEA **
CO 2 **
**
MEA H 2O
nMEA + nMEAH+ + nMEACOO- = n0,MEA (E18)
nCO2 + nHCO− + nMEACOO- = n0,CO2 (E19)
3
nH2O + nMEAH+ − nMEACOO- = n0,H2O (E20)
nMEAH+ = nHCO− + nMEACOO- (E21)
3
The equilibrium constants are calculated from the standard state Gibbs, Gi free energy of the
species as given by eq (E22),
− RT ln K j = Gj =
iSmac
ij Gi , j = 1, 2 (E22)
Standard State Gibbs Free Energy
Gi for Solvents
f s (T )
Gs (T ) = Gsig (T ) + RT ln , s = MEA, H 2O (E23)
P
Where the standard state pressure , P = 101325 Pa , the pure solvent fugacity, f s (T ) is defined in eq. (E24)
Bss Pssat + Vs ( P − Pssat )
f s (T ) = P exp
sat
(E24)
s
RT
141
The second viral coefficient, Bss is calculated from eq. (E25),
Bˆ s RTc , s
Bss = , Bˆ s = B0 + s B1 s S s
Pc , s
0.422
B0 = 0.083 − (E25)
Tr1.6
0.172
B1 = 0.139 −
Tr4.2
The properties of the solvents are given in Table 35
Table 35 : Physical Constants39
Species Tc (K) Pc (Pa)
H2O 0.344 647.096 22048320
MEA 0.83505 614.45 4468000
CO2 0.224 304.2 7383000
The ideal gas Gibbs free energy, Gsig for the solvent is computed in eq. (E27) based on the ideal
gas heat capacity correlation given in eq. (E26),
C igp,s ( J/mol.K ) = As + BsT + CsT 2 + DsT 3 (E26)
Gs (T ) = f Gsig,298.15t + f H sig,298.15 (1 − t ) + AsT0 ( t − 1 − t ln t )
BsT02 CsT03 DsT04 (E27)
− ( t − 1) − ( t − 1) ( t + 2 ) − ( t − 1) ( t 2 + 2t + 3)
2 2 2
2 6 12
T
Where t = , T0 = 298.15 and the standard state properties and heat capacity parameters are
T0
given in Table D6.
142
Gi for Henry component
The standard state for the Henry component is the aqueous phase infinite dilution, hence,
Gh (T ) = Gh ,aq (T ), h = CO 2
H i,w (E28)
Gh ,aq (T ) = Ghig (T ) + RT ln
P
Where Ghig is computed from eq. (E27) and H i , w is computed from eq. (E6).
Gi for Ions
The standard state for ionic species is chosen as the Gibbs energy at aqueous infinite dilution and
it is computed as given in eq. (E29) based on the heat capacity correlation in eq. (E30)
Gi (T ) = Gi ,aq (T ), i HCO3− , MEACOO − , MEAH +
BiT02 (E29)
Gi ,aq (T ) = f Gi,298.15 t + f H i,298.15 (1 − t ) + AT
i 0 ( t − 1 − t ln t ) − ( t − 1)
, aq , aq 2
C p,,iaq ( J/mol.K ) = Ai + BT
i (E30)
Since f Gi,298.15
, aq
depends on the concentration scale and the available data are in the molality scale,
it is converted from the molality scale where water is the solvent to mole fraction scale as given in
eq. (E31)
f Gi,298.15
, aq
= f Gi,298.15
, aq , molality
+ RT ln 55.51 (E31)
The standard state properties and heat capacity parameters for the MEA-H2O-CO2 system are
provided in Table D6, where,
143
G
ig
(J/mol) for molecular species
f G298.15 = f 298.15
f G298.15
, aq , molality
(J/mol) for ionic species
H ig (J/mol) for molecular species
f H 298.15 = f 298.15 , aq , molality
(E32)
f H 298.15 (J/mol) for ionic species
C igp,i ( J/mol.K ) for molecular species
C p ,i = ,aq
C p ,i ( J/mol.K ) for ionic species
Table 36 : Standard state properties and heat capacity parameters for MEA-H2O-CO2 system
Componen
f G298.15
f H 298.15
A B C D Source
ts
1.3207e -0.1513e-
MEA 2.8158e-1 3.1287e-8 -103300 -206700 a35
1 3
H2O 32.22 1.923e-3 10.548e-6 -3.594e-9 -228590 -241820 b33
1.9795e -5.6019e-
CO2 7.3437e-2 1.7153e-8 -394370 -393510 c145
1 5
MEAH+ 7.39e2 -1.63 0 0 -189620 -331640 a35,d36
MEACOO
- -5.98e-4 4.23e-2 0 0 -492990 -707470 a35,d36
HCO3- -29.26 0 0 0 -586770 -691990 b33
a b c d
Zhang et al.(2011), Gmehling et al(2019), Zhang and Chen(2010), Morgan et al(2017)
Partial derivative of Equilibrium constant
ln K j T2 (Gi / T )
ln K j / (1/ T ) −T
2
= vij , j = 1, 2 (E33)
T P, x R iSmac T P , x
P, x
For Ions
, aq
(Gi / T ) f H i ,298.15 T 1
=− + Ai 02 − (E34)
T T T
2
T
For Henry component
, aq
(Gi / T ) f H i ,298.15 T 1 B 1
=− + Ai 02 − − i 1 − 2
T T T 2 t
2
T
(E35)
C T0 Di 2 T02 ln H i , w
− i T − 2 − T − 2 + R
3 t 4 t T
144
ln H i , w b c
R = R − iw2 + iw + diw (E36)
T T T
For Solvents
(Gi / T ) (Giig / T ) ln fi
= + R (E37)
T T T
, aq
(Giig / T ) f H i ,298.15 T 1 B 1 C T0 Di 2 T02
=− + Ai 02 − − i 1 − 2 − i T − 2 − T − 2 (E38)
T T T 2 t 3 t 4
2
T t
B P sat P sat Bii Pi sat Bii Vs Pi sat ( P − Ps ) (V
sat
ln fi ln Pi
sat
/T)
T = T + ii i + i − − + (E39)
s
RT T RT T RT 2 RT T R T
ln Pi sat C C Pi sat ln Pi sat
= − 22i + 3i + 2C4iT , = Pi sat (E40)
T T T T T
dBii R 0.6752 i 0.7224
= + (E41)
dT Pc ,i Tr2.6 Tr5.2
(Vs / T ) V ( 3asT + 2bs + cs / T )
=− s (E42)
T asT 3 + bsT 2 + csT
145
Appendix F : Enthalpy model
The liquid phase enthalpy of the system is calculated as shown in eq. (F1)
H L = xi H iL + x j H j ,aq + H ex (F1)
iS s j{ solutes }
L
Where xi and x j represent the mole fraction of solvent i and solute j respectively, H i is the
,aq
molar enthalpy of solvent i , H j is the molar enthalpy of solute j (molecular or ionic) in
aqueous-phase infinite dilution, and H ex is the excess enthalpy based on aqueous-phase infinite
dilution and is calculated from eq. (D1).
Solvents
The molar enthalpy of the pure solvent is calculated from the reference point set under ideal gas
standard conditions (Tref = 298.15K , H ref ,i = f Hi ,298.15 ) as given in eq. (F2)
T
H iL = f H iig,298.15 +
298.15
C igp ,i dT + H idep
,T , P s
− vap H i ,T , i S s (F2)
f Hiig,298.15 is the ideal gas heat of formation at the reference temperature , C igp,i is the specific heat
capacity of the ideal gas, H idep
,T , P s
is the departure term added to obtained the enthalpy of the
saturated vapor and vap H i ,T is enthalpy of vaporization at the system temperature. The heat
capacity integral is evaluated as given in eq. (F3) where the heat capacity parameters are given in
Table. D6,
T
C igp ,i dT = Ai (T − Tref ) +
2
(
Bi 2
T − Tref2 ) + i (T 3 − Tref3 ) + i (T 4 − Tref4 ) , i Sm
C
3
D
4
(F3)
Tref = 298.15
The enthalpy departure term is calculated using a high precision equation of state (e.g PC SAFT
EOS). However, for low to moderate pressure, the Pitzer correlations for second viral coefficient
146
is adopted as given in eq. (F4) for computing the enthalpy departure term for the solvent where
the second viral coefficients , Bii is defined in eq. (E25),
dB
H idep
,T , P s
= Pi sat Bii − T ii (F4)
dT
The heat of vaporization is computed from the Watson correlations as shown in eq. (F5) using
parameters obtained from Hilliard(2008) as shown in Table 37,
ai + bi (1−t tc ,i )
1 − t tc ,i
vap H i (J/mol) = H1,i , t = T (K) − 273.15 (F5)
1− t t
1,i c ,i
Table 37: Watson heat of for vaporization parameters39
Symbol H2O MEA
H1,i 40655 54835.8
t1,i 100.0 126.67
tc , i 373.946 398.25
ai 0.3106 0.3288
bi 0.0 -0.0857
Solutes(Molecular or Henry components)
, aq
The molar enthalpy of the Henry’s component (CO2) at aqueous-phase infinite dilution, H CO2 is
computed as given in eq. (F6)
2
ln H CO2 ,H2O
, aq
H CO = H V
CO − RT
T
V
(
= H CO2 − R −bCO2 ,H2O + cCO2 ,H2OT + d CO2 ,H2OT
2
)
2 2
T
(F6)
= f H + C dT + H
v ig ig dep
H CO2 i ,298.15 p ,i i ,T , P
298.15
147
The enthalpy departure term is computed from eq.(F4), the Henry parameters
(bCO2 ,H2O , cCO2 ,H2O , dCO2 ,H2O ) are obtained from Table D2, while the heat capacity parameters given
in Table D6 are used to evaluate the integral as given in eq. (F3).
Solutes(Ionic species)
The enthalpy of the ionic species based on aqueous phase infinite dilution reference state is
computed as given in (F7) and the parameters are shown in Table D6.
H i,aq = f H i,298.15
, aq
+ Ai (T − Tref ) +
2
(T − Tref2 ) , i MEAH+ , MEACOO− , HCO3− (F7)
Bi 2
Heat of Absorption
The heat of CO2 absorption , H abs in aqueous MEA solution is derived from the enthalpy
balance of a mixing process during which nCO2 mols of CO2 gas dissolves into an initial MEA-
H2O-CO2 system of nInitial moles to form a final solution of nFinal moles as given in eq. (F8),
nFinal H L, Final − nInitial H L, Initial − nCO2 H CO
v
H abs = 2
(F8)
nCO2
Heat Capacity
The heat capacity of the liquid mixture is calculated using central difference approximation as
given in eq. (F9)
H H (T + T ) − H L (T − T )
C p (J/mol.K) = L L (F9)
T P , x 2T
148
Appendix G: Transport properties
Viscosity
Vapor phase
Pure component
Sutherland equation is used for CO2, N2 and O2.
V ,i0 0.555Ti0 + Ci
V ,i ( Pa.s ) = ( )
3/2
(T + 459.67) / T (G1)
1000 0.555(T + 459.67) + Ci
i0
Where T is temperature in kelvin.
Lucas method of corresponding states is used for water vapor32.
Table 38 :Sutherland's parameters
V ,i0 Ti0 Ci
CO 2 0.01480 527.67 240
N2 0.01781 540.99 111
O2 0.02018 526.05 127
Viscosity of vapor mixture84
V ,i
V ( Pa.s ) =
i j =n
1 + 1 yiij
yi
j =1
j i (G2)
1 + ( / ) ( MW / MW )
1/2 1/4 2
ij =
i ,V j ,V j i
( )
2 2 1 + ( MWi / MW j )
1/2
149
Liquid phase
H O ( a + b ) TL + c + d ( e + f T + g ) + 1
L ( Pa s ) = exp
2
1000 TL2
1.3272 293.15−TL − 0.001053(TL − 293.15)
2
H O ( mPa s ) = 1.002 10 TL −168.15
2
(G3)
MWMEA xMEA
= 100r , where r =
MWMEA xMEA + MWH2O xH2O
mol CO 2 xCO2
=
mol MEA xMEA
Table 39: Liquid viscosity parameters 74
a b c d e f g
-0.0838 2.8817 33.651 1817 0.00847 0.0103 -2.3890
Surface Tension
Pure component
c2,i + c3,
i (TL / Tc ,i ) + c4,i (TL / Tc ,i )
2
T
i ( N/m ) = c1,i 1 − L i MEA,H 2 O
Tc ,i (G4)
CO ( N/m ) = S1r 2 + S 2 r + S3 + T ( S 4 r 2 + S5r + S 6 )
2
Table 40 : Parameters for pure component surface tension
c1,i c3,i
88
c2,i c4,i Tc,i (K)
MEA 0.09945 1.067 0 0 614.45
H2O 0.18548 2.717 -3.554 -2.047 647.13
74
S1 S2 S3 S4 S5 S6
CO2 -5.987 3.7699 -0.43164 0.018155 -0.01207 0.002119
150
Liquid solution
L ( N/m ) = H O + ( CO − H O ) F , x
2 2 2 CO2 CO2
(
+ MEA − H 2O F ,MEA xMEA )
F ,CO = a + b + c 2 + d r + e r 2 (G5)
2
F ,MEA = f + g + h 2 + i r + j r 2
Table 41: Surface tension parameters74
a b c d e
2.4558 -1.5311 3.4994 -5.6398 10.2109
f g h i j
2.3122 4.5608 -2.3924 5.3324 -12.0494
Diffusivity
Vapor phase
(M + M j ) MiM j
Dij ( m 2 /s ) = 1.013 10−2 TV1.75
i
(G6)
( )
2
P 3 VDi + 3 VDj
Table 42: Diffusion Volumes in Fuller-Schettler-Giddings Correlation
MEA H2O CO2 N2 O2
58.62 13.1 26.7 18.5 16.3
Effective or pseudo binary diffusivity 146
1 − yi
DV ,i ( m 2 /s ) = n y
(G7)
D j =1
j
ij
j i
151
Liquid phase
CO2
b0 + b1CMEA, L
DL ,CO2 ( m 2 /s ) = ( a0 + a1CMEA, L + a2CMEA
2
, L ) exp (G8)
TL
Where CMEA, L is the concentration of MEA in the liquid and T is temperature in kelvin.
Table 43: Ying and Eimer (2012) diffusivity parameters for CO2 in aqueous MEA mixture
Parameters a0 a1 a2 b0 b1
Values 2.35 10−6 2.9837 10−8 −9.7078 10−9 −2119 −20.1320
MEA81
DL ,MEA ( m 2 /s ) = exp −13.275 −
2198.3
− 7.8142 10−5 CMEA, L (G9)
TL
MEACOO-82
1000
DL ,MEACOO− = exp −22.64 − − 0.7 ln L (G10)
TL
152
Appendix H: Experimental Data
Table 44 : Absorber flow streams -NCCC Steady-State data
CASES 1 2 3 4 5 6 7
Lean solvent absorber flowrate (kg/hr) 6804 11794 3175 6804 6804 11643 3175
Lean solvent absorber loading (mol CO2/MEA) 0.145 0.247 0.083 0.108 0.347 0.154 0.239
Lean solvent MEA composition (g MEA/MEA+H2O) 0.298 0.312 0.31 0.306 0.307 0.285 0.311
Lean solvent absorber inlet temperature (ºC) 40.97 40.52 46.72 41.57 40.87 40.57 42.66
Rich solvent absorber flowrate (kg/hr) 7242 12284 3343 7212 7063 12043 3337
Rich solvent absorber loading (mol CO2 /MEA) 0.384 0.385 0.47 0.295 0.469 0.275 0.474
Rich solvent MEA composition (g MEA/MEA+ H2O) 0.3 0.314 0.328 0.308 0.309 0.289 0.318
Flue gas flowrate (kg/hr) 2266 2261 2261 2253 2255 2255 2250
Flue gas CO2 weight fraction 0.1731 0.1718 0.1724 0.1395 0.1381 0.1393 0.1401
Flue gas H2O weight fraction 0.0462 0.053 0.0525 0.0505 0.0464 0.0466 0.0533
Flue gas N2 weight fraction 0.7116 0.7066 0.7072 0.7152 0.7183 0.7193 0.7134
Flue gas O2 weight fraction 0.069 0.0686 0.0678 0.0948 0.0971 0.0948 0.0932
Flue gas inlet temperature (ºC) 42.48 44.94 44.73 43.78 42.18 42.47 44.87
Absorber pressure (kPa) 108.82 107.06 107.65 106.94 107.1 107.26 107.49
153
Table 45: Absorber intercoolers between the packed beds -NCCC Steady-State data
CASES 1 2 3 4 5 6 7
Intercooler 1 flowrate (kg/hr) 7364.83 12340.85 3848.86 7357.23 7539.09 11634.52 3832.25
Intercooler 1 return temperature (ºC) 40.13 40.19 43.34 41.15 43.39 39.99 43.31
Intercooler 2 flowrate (kg/hr) 7421.57 12261.52 2429.55 7180.54 7179.52 11504.64 2335.07
Intercooler 2 return temperature (ºC) 43.32 43.32 43.33 43.36 43.26 40.39 40.03
Number of absorber beds/ intercoolers 3/2 3/2 3/2 3/2 3/2 3/2 3/2
Table 46: Stripper, reboiler and heat exchanger (HX) NCCC steady state data
CASES 1 2 3 4 5 6 7
Stripper pressure (kPa) 183.87 182.06 184.15 183.43 179.88 183.45 182.84
Rich solvent stripper inlet temperature (ºC) 104.8 104.8 97.6 109 95.7 110.1 98.4
Stripper solvent exit temperature (ºC) 120.18 117.43 122.53 121.68 110.21 120.35 117.69
Reboiler duty (kW) 431 430 427 677 171 677 166
Lean solvent HX flowrate (kg/hr) 6811.2 11804.4 2998.8 6811.2 6811.2 3178.8 6811.2
Lean solvent HX temperature, inlet/exit (ºC) 119.3/57.5 116.6/63.8 103.3/51.6 120.8/58.5 109.5/50.5 119.7/61.7 116.7/51.9
Rich solvent HX flowrate (kg/hr) 7246.8 12294.0 3344.4 7218.0 7070.4 3340.8 7246.8
Rich solvent HX temperature, inlet/exit (ºC) 53.4/111.8 59.3/110.1 46.6/97.4 56.1/110.5 45.8/103.0 57.7/111.8 48.4/108.9
154
Table 47 : TCM absorber 2015 baseline data from Faramarzi et al. (2017).
Flue gas supply rate (Sm3/h) 59430
Flue gas supply temperature (°C) 29.8
Flue gas supply pressure (barg) 0.01
Flue gas supply CO2 concentration (vol%) 3.7
Flue gas supply O2 concentration (vol%) 14.6
Flue gas supply water content (vol%) 3.7
Lean amine supply flow rate (kg/h) 57434
Lean MEA concentration (wt%) 31
Lean CO2 loading (mol CO2/mol MEA) 0.2
Active absorber packing height (m) 24
Lean amine supply temperature (°C) 37
Absorber Beds Height (m) Temperature (°C)
Upper Absorber Bed 23.5 47.4
22.5 51.7
21.5 51.6
20.5 50.5
19.5 49.9
18.5 48.9
Middle Absorber Bed 17.5 47.2
16.5 46.0
15.5 44.4
14.5 43.1
13.5 42.2
12.5 40.9
Lower Absorber Bed 11.5 40.6
10.5 41.6
9.5 37.4
8.5 37.1
7.5 35.9
6.5 34.3
5.5 34.1
4.5 33.8
3.5 32.9
2.5 33.2
1.5 32.5
0.5 32.4
155
Table 48: TCM Stripper 2015 baseline data from Faramarzi et al. (2017).
Stripper overhead pressure (barg) 0.91
Stripping section packing height (m) 8
Rich solution supply temperature (°C) 110.7
Stripper Bed Height (m) Temperature (°C)
7 102.7
6 103.1
5 104.5
Stripper Packing 4 107.7
3 112.1
2 114.7
1 119.4
Stripper sump temperature (°C) 121.0
156
References
(1) Supekar, S. D.; Skerlos, S. J. Reassessing the efficiency penalty from carbon capture in coal-
fired power plants. Environmental science & technology 2015, 49 (20), 12576-12584.
(2) Miller, D. C.; Agarwal, D.; Bhattacharyya, D.; Boverhof, J.; Cheah, Y.-W.; Chen, Y.; Eslick,
J.; Leek, J.; Ma, J.; Mahapatra, P. Innovative computational tools and models for the design,
optimization and control of carbon capture processes. In Computer Aided Chemical Engineering,
Vol. 38; Elsevier, 2016; pp 2391-2396.
(3) Putta, K. R.; Tobiesen, F. A.; Svendsen, H. F.; Knuutila, H. K. Applicability of enhancement
factor models for CO2 absorption into aqueous MEA solutions. Applied Energy 2017, 206, 765-
783.
(4) Turton, R.; Bhattacharyya, D.; Whiting, W. B.; Shaeiwitz, J. A. Analysis, synthesis and design
of chemical processes; Pearson Education, 2018.
(5) Prausnitz, J. M.; Anderson, T. F.; Grens, E. A.; A., E. C.; R., H.; P., O. C. J. Computer
calculations for multicomponent vapor-liquid and liquid-liquid equilibria; Prentice Hall, 1980.
(6) Lee, A.; Ghouse, J. H.; Eslick, J. C.; Laird, C. D.; Siirola, J. D.; Zamarripa, M. A.; Gunter, D.;
Shinn, J. H.; Dowling, A. W.; Bhattacharyya, D. The IDAES process modeling framework and
model library—Flexibility for process simulation and optimization. Journal of Advanced
Manufacturing and Processing 2021, 3 (3), e10095.
(7) Akula, P.; Eslick, J.; Bhattacharyya, D.; Miller, D. C. Model Development, Validation, and
Optimization of an MEA-Based Post-Combustion CO2 Capture Process under Part-Load and
Variable Capture Operations. Industrial & Engineering Chemistry Research 2021. DOI:
10.1021/acs.iecr.0c05035.
(8) Bhattacharyya, D.; Miller, D. Post-combustion CO 2 capture technologies — a review of
processes for solvent-based and sorbent-based CO 2 capture. Current Opinion in Chemical
Engineering 2017, 17, 78-92. DOI: 10.1016/j.coche.2017.06.005.
(9) Kohl, A. L.; Nielsen, R. Gas purification; Gulf Professional Publishing, 1997.
(10) Zhang, Y.; Chen, H.; Chen, C.-C.; Plaza, J. M.; Dugas, R.; Rochelle, G. T. Rate-based process
modeling study of CO2 capture with aqueous monoethanolamine solution. Industrial &
engineering chemistry research 2009, 48 (20), 9233-9246. Pandya, J. Adiabatic gas absorption
157
and stripping with chemical reaction in packed towers. Chemical Engineering Communications
1983, 19 (4-6), 343-361.
(11) Taylor, R.; Krishna, R. Multicomponent mass transfer; John Wiley & Sons, 1993.
(12) Kvamsdal, H. M.; Rochelle, G. T. Effects of the temperature bulge in CO2 absorption from
flue gas by aqueous monoethanolamine. Industrial & Engineering Chemistry Research 2008, 47
(3), 867-875.
(13) Wang, M.; Lawal, A.; Stephenson, P.; Sidders, J.; Ramshaw, C. Post-combustion CO2 capture
with chemical absorption: A state-of-the-art review. Chemical Engineering Research and Design
2011, 89 (9), 1609-1624. DOI: https://doi.org/10.1016/j.cherd.2010.11.005. Liang, Z.; Fu, K.;
Idem, R.; Tontiwachwuthikul, P. Review on current advances, future challenges and consideration
issues for post-combustion CO2 capture using amine-based absorbents. Chinese Journal of
Chemical Engineering 2016, 24 (2), 278-288.
(14) Bui, M.; Flø, N. E.; de Cazenove, T.; Mac Dowell, N. Demonstrating flexible operation of
the Technology Centre Mongstad (TCM) CO2 capture plant. International Journal of Greenhouse
Gas Control 2020, 93, 102879. DOI: https://doi.org/10.1016/j.ijggc.2019.102879. Cohen, S. M.;
Rochelle, G. T.; Webber, M. E. Optimal operation of flexible post-combustion CO2 capture in
response to volatile electricity prices. Energy Procedia 2011, 4, 2604-2611.
(15) Roeder, V.; Kather, A. Part Load Behaviour of Power Plants with a Retrofitted Post-
combustion CO2 Capture Process. Energy Procedia 2014, 51, 207-216. DOI:
https://doi.org/10.1016/j.egypro.2014.07.024.
(16) Smith, J. M. V. N., H C;Abbott,M. M;. Introduction to chemical engineering thermodynamics.
7th ed.; ACS Publications: 1950.
(17) Pitzer, K. S. Thermodynamics of electrolytes. I. Theoretical basis and general equations. The
Journal of Physical Chemistry 1973, 77 (2), 268-277.
(18) Thomsen, K. Modeling electrolyte solutions with the extended universal quasichemical
(UNIQUAC) model. Pure and applied chemistry 2005, 77 (3), 531-542. Hingerl, F. F.; Wagner,
T.; Kulik, D. A.; Thomsen, K.; Driesner, T. A new aqueous activity model for geothermal brines
in the system Na-K-Ca-Mg-H-Cl-SO4-H2O from 25 to 300° C. Chemical Geology 2014, 381, 78-
93. Thomsen, K.; Rasmussen, P.; Gani, R. Correlation and prediction of thermal properties and
phase behaviour for a class of aqueous electrolyte systems. Chemical Engineering Science 1996,
51 (14), 3675-3683.
158
(19) Wang, P.; Anderko, A.; Young, R. D. A speciation-based model for mixed-solvent electrolyte
systems. Fluid Phase Equilibria 2002, 203 (1-2), 141-176. Wang, P.; Springer, R.; Anderko, A.;
Young, R. Modeling phase equilibria and speciation in mixed-solvent electrolyte systems. Fluid
Phase Equilibria 2004, 222, 11-17. Wang, P.; Anderko, A.; Springer, R.; Young, R. Modeling
phase equilibria and speciation in mixed-solvent electrolyte systems: II. Liquid–liquid equilibria
and properties of associating electrolyte solutions. Journal of Molecular Liquids 2006, 125 (1),
37-44.
(20) Bollas, G. M.; Chen, C. C.; Barton, P. I. Refined electrolyte‐NRTL model: Activity coefficient
expressions for application to multi‐electrolyte systems. AIChE Journal 2008, 54 (6), 1608-1624.
DOI: 10.1002/aic.11485. Chen, C. C.; Britt, H. I.; Boston, J.; Evans, L. Local composition model
for excess Gibbs energy of electrolyte systems. Part I: Single solvent, single completely dissociated
electrolyte systems. AIChE Journal 1982, 28 (4), 588-596.
(21) Mock, B.; Evans, L.; Chen, C. C. Thermodynamic representation of phase equilibria of mixed‐
solvent electrolyte systems. AIChE Journal 1986, 32 (10), 1655-1664.
(22) Song, Y.; Chen, C.-C. Symmetric electrolyte nonrandom two-liquid activity coefficient
model. Industrial & engineering chemistry research 2009, 48 (16), 7788-7797.
(23) Chen, C. C.; Evans, L. B. A local composition model for the excess Gibbs energy of aqueous
electrolyte systems. AIChE Journal 1986, 32 (3), 444-454.
(24) Chen, C.-C. Toward development of activity coefficient models for process and product
design of complex chemical systems. Fluid phase equilibria 2006, 241 (1-2), 103-112. Chen, C.-
C.; Mathias, P. M. Applied thermodynamics for process modeling. American Institute of Chemical
Engineers. AIChE Journal 2002, 48 (2), 194.
(25) Hossain, N.; Bhattacharia, S. K.; Chen, C. C. Temperature dependence of interaction
parameters in electrolyte NRTL model. AIChE Journal 2016, 62 (4), 1244-1253.
(26) Aspen Plus Documentation. Aspen Physical Property System. In Physical property methods,
Aspen Technology Inc.
: Cambridge, MA, USA, 2013.
(27) Hessen, E. T.; Haug-Warberg, T.; Svendsen, H. F. The refined e-NRTL model applied to
CO2–H2O–alkanolamine systems. Chemical Engineering Science 2010, 65 (11), 3638-3648. DOI:
https://doi.org/10.1016/j.ces.2010.03.010.
159
(28) Løvfall, B. T. Computer realization of thermodynamic models using algebraic objects. Thesis,
Norwegian University of Science and Technology, 2008.
(29) Renon, H.; Prausnitz, J. M. Local compositions in thermodynamic excess functions for liquid
mixtures. AIChE journal 1968, 14 (1), 135-144.
(30) Copeman, T. W.; Stein, F. P. On the derivation and application of solution thermodynamics
for reactive components. Fluid Phase Equilibria 1982, 9 (2), 149-165.
(31) Copeman, T. W.; Stein, F. P. On the derivation and application of solution thermodynamics
for reactive components: Application to an acid-gas system. Fluid Phase Equilibria 1983, 11 (1),
49-64. Perry, R. L.; Telotte, J. C.; O'Connell, J. P. Solution thermodynamics for “reactive”
components. Fluid Phase Equilibria 1981, 5 (3-4), 245-277.
(32) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The properties of gases and liquids. 1987.
(33) Gmehling, J.; Kleiber, M.; Kolbe, B.; Rarey, J. Chemical thermodynamics for process
simulation; Wiley Online Library, 2019.
(34) Freguia, S.; Rochelle, G. T. Modeling of CO2 capture by aqueous monoethanolamine. Aiche
Journal 2003, 49 (7), 1676-1686. DOI: DOI 10.1002/aic.690490708.
(35) Zhang, Y.; Que, H.; Chen, C.-C. Thermodynamic modeling for CO2 absorption in aqueous
MEA solution with electrolyte NRTL model. Fluid Phase Equilibria 2011, 311, 67-75. DOI:
https://doi.org/10.1016/j.fluid.2011.08.025.
(36) Morgan, J. C.; Chinen, A. S.; Omell, B.; Bhattacharyya, D.; Tong, C.; Miller, D. C.
Thermodynamic modeling and uncertainty quantification of CO2-loaded aqueous MEA solutions.
Chemical Engineering Science 2017, 168, 309-324. DOI:
https://doi.org/10.1016/j.ces.2017.04.049.
(37) Morgan, J. C.; Soares Chinen, A.; Omell, B.; Bhattacharyya, D.; Tong, C.; Miller, D. C.;
Buschle, B.; Lucquiaud, M. Development of a Rigorous Modeling Framework for Solvent-Based
CO2 Capture. Part 2: Steady-State Validation and Uncertainty Quantification with Pilot Plant
Data. Industrial & Engineering Chemistry Research 2018, 57 (31), 10464-10481.
(38) Soares Chinen, A.; Morgan, J. C.; Omell, B.; Bhattacharyya, D.; Tong, C.; Miller, D. C.
Development of a Rigorous Modeling Framework for Solvent-Based CO2 Capture. 1. Hydraulic
and Mass Transfer Models and Their Uncertainty Quantification. Industrial & Engineering
Chemistry Research 2018, 57 (31), 10448-10463. DOI: 10.1021/acs.iecr.8b01471.
160
(39) Hilliard, M. D. A predictive thermodynamic model for an aqueous blend of potassium
carbonate, piperazine, and monoethanolamine for carbon dioxide capture from flue gas. PhD
Thesis., The University of Texas at Austin., 2008.
(40) Nicolaides, G. L.; Eckert, C. A. Optimal representation of binary liquid mixture nonidealities.
Industrial & Engineering Chemistry Fundamentals 1978, 17 (4), 331-340.
(41) Chen, C. C.; Song, Y. Generalized electrolyte‐NRTL model for mixed‐solvent electrolyte
systems. AIChE Journal 2004, 50 (8), 1928-1941. DOI: 10.1002/aic.10151.
(42) Brelvi, S. W.; O'Connell, J. P. Correspondling states correlations for liquid compressibility
and partial molal volumes of gases at infinite dilution in liquids. AIChE Journal 1972, 18 (6),
1239-1243. DOI: 10.1002/aic.690180622.
(43) Klise, K. A.; Nicholson, B. L.; Staid, A.; Woodruff, D. L. Parmest: Parameter Estimation Via
Pyomo. In Computer Aided Chemical Engineering, Vol. 47; Elsevier, 2019; pp 41-46. Woodruff,
D. L.; Staid, A.; Nicholson, B. L.; Klise, K. A. PARMEST: PARAMETER ESTIMATION VIA
PYOMO; Sandia National Lab.(SNL-NM), Albuquerque, NM (United States), 2018.
(44) Chang, H. T.; Posey, M.; Rochelle, G. T. Thermodynamics of alkanolamine-water solutions
from freezing point measurements. Industrial & engineering chemistry research 1993, 32 (10),
2324-2335.
(45) Mathias, P. M.; O’Connell, J. P. The Gibbs–Helmholtz equation and the thermodynamic
consistency of chemical absorption data. Industrial & Engineering Chemistry Research 2012, 51
(13), 5090-5097.
(46) Kim, I.; Hoff, K. A.; Mejdell, T. Heat of Absorption of CO2 with Aqueous Solutions of MEA:
New Experimental Data. Energy Procedia 2014, 63, 1446-1455. DOI:
https://doi.org/10.1016/j.egypro.2014.11.154.
(47) Tochigi, K.; Akimoto, K.; Ochi, K.; Liu, F.; Kawase, Y. Isothermal vapor− liquid equilibria
for water+ 2-aminoethanol+ dimethyl sulfoxide and its constituent three binary systems. Journal
of Chemical & Engineering Data 1999, 44 (3), 588-590.
(48) Park, S.-B.; Lee, H. Vapor-liquid equilibria for the binary monoethanolamine+ water and
monoethanolamine+ ethanol systems. Korean Journal of chemical engineering 1997, 14 (2), 146-
148. Nath, A.; Bender, E. Isothermal vapor-liquid equilibriums of binary and ternary mixtures
containing alcohol, alkanolamine, and water with a new static device. Journal Of Chemical And
Engineering Data 1983, 28 (4), 370-375.
161
(49) Cai, Z.; Xie, R.; Wu, Z. Binary isobaric vapor− liquid equilibria of ethanolamines+ water.
Journal of Chemical & Engineering Data 1996, 41 (5), 1101-1103.
(50) Posey, M. L. Thermodynamic model for acid gas loaded aqueous alkanolamine solutions.
1998.
(51) Touhara, H.; Okazaki, S.; Okino, F.; Tanaka, H.; Ikari, K.; Nakanishi, K. Thermodynamic
properties of aqueous mixtures of hydrophilic compounds 2. Aminoethanol and its methyl
derivatives. The Journal of Chemical Thermodynamics 1982, 14 (2), 145-156.
(52) Weiland, R. H.; Dingman, J. C.; Cronin, D. B.; Browning, G. J. Density and viscosity of some
partially carbonated aqueous alkanolamine solutions and their blends. Journal of Chemical &
Engineering Data 1998, 43 (3), 378-382.
(53) Chiu, L.-F.; Li, M.-H. Heat capacity of alkanolamine aqueous solutions. Journal of Chemical
& Engineering Data 1999, 44 (6), 1396-1401.
(54) Austgen, D. M.; Rochelle, G. T.; Peng, X.; Chen, C. C. Model of vapor-liquid equilibria for
aqueous acid gas-alkanolamine systems using the electrolyte-NRTL equation. Industrial &
Engineering Chemistry Research 1989, 28 (7), 1060-1073. DOI: 10.1021/ie00091a028.
(55) Schmidt, K.; Maham, Y.; Mather, A. Use of the NRTL equation for simultaneous correlation
of vapour-liquid equilibria and excess enthalpy: applications to aqueous alkanolamine systems.
Journal of thermal analysis and calorimetry 2007, 89 (1), 61-72.
(56) Liu, Y.; Zhang, L.; Watanasiri, S. Representing vapor− liquid equilibrium for an aqueous
MEA− CO2 system using the electrolyte nonrandom-two-liquid model. Industrial & engineering
chemistry research 1999, 38 (5), 2080-2090.
(57) Renon, H. M. Thermodynamic Properties of Nonideal Liquid Mixtures. Doctoral dissertation,
University of California, Berkeley, 1966.
(58) Kim, I.; Svendsen, H. F.; Børresen, E. Ebulliometric determination of vapor− liquid equilibria
for pure water, monoethanolamine, n-methyldiethanolamine, 3-(methylamino)-propylamine, and
their binary and ternary solutions. Journal of Chemical & Engineering Data 2008, 53 (11), 2521-
2531.
(59) Biegler, L. T.; Zavala, V. M. Large-scale nonlinear programming using IPOPT: An
integrating framework for enterprise-wide dynamic optimization. Computers & Chemical
Engineering 2009, 33 (3), 575-582. DOI: https://doi.org/10.1016/j.compchemeng.2008.08.006.
162
(60) Duff, I. S.; Reid, J. K. MA27-a set of Fortran subroutines for solving sparse symmetric sets
of linear equations; UKAEA Atomic Energy Research Establishment London, 1982.
(61) Oyenekan, B. A.; Rochelle, G. T. Energy Performance of Stripper Configurations for CO2
Capture by Aqueous Amines. Industrial & Engineering Chemistry Research 2006, 45 (8), 2457-
2464. DOI: 10.1021/ie050548k.
(62) Kim, I.; Svendsen, H. F. Heat of Absorption of Carbon Dioxide (CO2) in Monoethanolamine
(MEA) and 2-(Aminoethyl)ethanolamine (AEEA) Solutions. Industrial & Engineering Chemistry
Research 2007, 46 (17), 5803-5809. DOI: 10.1021/ie0616489.
(63) Taylor, R.; Krishna, R.; Kooijman, H. Real-world modeling of distillation. transfer 2003,
1000, 1.
(64) Böttinger, W.; Maiwald, M.; Hasse, H. Online NMR spectroscopic study of species
distribution in MEA–H2O–CO2 and DEA–H2O–CO2. Fluid Phase Equilibria 2008, 263 (2), 131-
143.
(65) Jakobsen, J. P.; Krane, J.; Svendsen, H. F. Liquid-Phase Composition Determination in
CO2−H2O−Alkanolamine Systems: An NMR Study. Industrial & Engineering Chemistry
Research 2005, 44 (26), 9894-9903. DOI: 10.1021/ie048813+.
(66) Plaza, J. M. Modeling of carbon dioxide absorption using aqueous monoethanolamine,
piperazine and promoted potassium carbonate. Doctoral dissertation, 2012.
(67) Barker, J. Determination of activity coefficients from total pressure measurements. Australian
Journal of Chemistry 1953, 6 (3), 207-210.
(68) Bard, Y. Nonlinear parameter estimation; 1974. Anderson, T.; Abrams, D.; Grens, E.
Evaluation of parameters for nonlinear thermodynamic models. AIChE Journal 1978, 24 (1), 20-
29.
(69) Billet, R. Relationship between residence time, fluid dynamics and efficiency in
countercurrent flow equipment. Chemical engineering & technology 1988, 11 (1), 139-148. Billet,
R.; Schultes, M. Predicting mass transfer in packed columns. Chemical Engineering &
Technology: Industrial Chemistry‐Plant Equipment‐Process Engineering‐Biotechnology 1993, 16
(1), 1-9.
(70) Billet, R.; Schultes, M. Modelling of pressure drop in packed columns. Chemical Engineering
& Technology: Industrial Chemistry‐Plant Equipment‐Process Engineering‐Biotechnology 1991,
14 (2), 89-95.
163
(71) Billet, R.; Schultes, M. A physical model for the prediction of liquid hold‐up in two‐phase
countercurrent columns. Chemical Engineering & Technology: Industrial Chemistry‐Plant
Equipment‐Process Engineering‐Biotechnology 1993, 16 (6), 370-375.
(72) Billet, R.; Schultes, M. Fluid dynamics and mass transfer in the total capacity range of packed
columns up to the flood point. Chemical Engineering & Technology: Industrial Chemistry‐Plant
Equipment‐Process Engineering‐Biotechnology 1995, 18 (6), 371-379.
(73) Billet, R.; Schultes, M. Prediction of mass transfer columns with dumped and arranged
packings: updated summary of the calculation method of Billet and Schultes. Chemical
Engineering Research and Design 1999, 77 (6), 498-504.
(74) Morgan, J. C.; Bhattacharyya, D.; Tong, C.; Miller, D. C. Uncertainty quantification of
property models: Methodology and its application to CO2‐loaded aqueous MEA solutions. AIChE
Journal 2015, 61 (6), 1822-1839.
(75) Aroonwilas, A.; Veawab, A.; Tontiwachwuthikul, P. Behavior of the mass-transfer coefficient
of structured packings in CO2 absorbers with chemical reactions. Industrial & engineering
chemistry research 1999, 38 (5), 2044-2050. Brunazzi, E.; Paglianti, A.; Spiegel, L.; Tolaini, F.
Hydrodynamics of a gas-liquid column equipped with MellapakPlus packing. In International
conference on Distillation and Absorption, 2002. Jammula, A. K.; Whiteley, J. R.; Resetarits, M.
R.; Cai, T. J. New Liquid Holdup and Load Point Models in the Loading Region for Sheet Metal-
Structured Packings and Performance Comparisons with Literature Models. Industrial &
Engineering Chemistry Research 2020, 59 (34), 15309-15316. Morgan, J. C.; Chinen, A. S.;
Anderson-Cook, C.; Tong, C.; Carroll, J.; Saha, C.; Omell, B.; Bhattacharyya, D.; Matuszewski,
M.; Bhat, K. S.; et al. Development of a framework for sequential Bayesian design of experiments:
Application to a pilot-scale solvent-based CO2 capture process. Applied Energy 2020, 262,
114533. DOI: https://doi.org/10.1016/j.apenergy.2020.114533.
(76) Jammula, A. K.; Whiteley, J. R.; Resetarits, M. R.; Cai, T. J. New Liquid Holdup Models in
Both Preloading and Flooding Regions and a Flood Velocity Model to Identify Flooding for Sheet
Metal Structured Packings. Industrial & Engineering Chemistry Research 2021.
(77) Rocha, J. A.; Bravo, J. L.; Fair, J. R. Distillation Columns Containing Structured Packings:
A Comprehensive Model for Their Performance. 2. Mass-Transfer Model. Industrial &
Engineering Chemistry Research 1996, 35 (5), 1660-1667. DOI: 10.1021/ie940406i.
164
(78) Tsai, R. E.; Seibert, A. F.; Eldridge, R. B.; Rochelle, G. T. A dimensionless model for
predicting the mass‐transfer area of structured packing. AIChE journal 2011, 57 (5), 1173-1184.
(79) Jayarathna, S. A.; Lie, B.; Melaaen, M. C. Amine based CO2 capture plant: Dynamic
modeling and simulations. International Journal of Greenhouse Gas Control 2013, 14, 282-290.
(80) Ying, J.; Eimer, D. A. Measurements and Correlations of Diffusivities of Nitrous Oxide and
Carbon Dioxide in Monoethanolamine + Water by Laminar Liquid Jet. Industrial & Engineering
Chemistry Research 2012, 51 (50), 16517-16524. DOI: 10.1021/ie302745d.
(81) Snijder, E. D.; te Riele, M. J.; Versteeg, G. F.; Van Swaaij, W. Diffusion coefficients of
several aqueous alkanolamine solutions. Journal of Chemical and Engineering data 1993, 38 (3),
475-480.
(82) Hoff, K. A.; Juliussen, O.; Falk-Pedersen, O.; Svendsen, H. F. Modeling and experimental
study of carbon dioxide absorption in aqueous alkanolamine solutions using a membrane
contactor. Industrial & engineering chemistry research 2004, 43 (16), 4908-4921.
(83) Sutherland, W. XXXVII. The viscosity of mixed gases. Philosophical Magazine Series 5
1895, 40 (246), 421-431. DOI: 10.1080/14786449508620789.
(84) Wilke, C. R. A Viscosity Equation for Gas Mixtures. The Journal of Chemical Physics 1950,
18 (4), 517-519. DOI: doi:http://dx.doi.org/10.1063/1.1747673.
(85) Li, C. Thermal conductivity of liquid mixtures. AIChE journal 1976, 22 (5), 927-930.
(86) Seader, J.; Henly, E. Seperation process principles, 2nd. John wiley and sons. Inc., New York:
2006.
(87) Ying, J.; Eimer, D. A. Determination and measurements of mass transfer kinetics of CO2 in
concentrated aqueous monoethanolamine solutions by a stirred cell. Industrial & Engineering
Chemistry Research 2013, 52 (7), 2548-2559.
(88) Asprion, N. Surface tension models for aqueous amine blends. Industrial & engineering
chemistry research 2005, 44 (18), 7270-7278.
(89) Versteeg, G. F.; Van Swaalj, W. Solubility and diffusivity of acid gases (carbon dioxide,
nitrous oxide) in aqueous alkanolamine solutions. Journal of Chemical & Engineering Data 1988,
33 (1), 29-34. DOI: 10.1021/je00051a011.
(90) Wilke, C.; Chang, P. Correlation of diffusion coefficients in dilute solutions. AIChE Journal
1955, 1 (2), 264-270.
165
(91) Wesselingh, J.; Krishna, R. Mass transfer in multicomponent mixtures; Delft University Press
Delft, 2000.
(92) Han, S.-J.; Wee, J.-H. Estimation of Correlation between Electrical Conductivity and CO2
Absorption in a Monoethanolamine Solvent System. Journal of Chemical & Engineering Data
2013, 58 (9), 2381-2388. DOI: 10.1021/je400358f.
(93) Stichlmair, J.; Bravo, J.; Fair, J. General model for prediction of pressure drop and capacity
of countercurrent gas/liquid packed columns. Gas Separation & Purification 1989, 3 (1), 19-28.
Onda, K.; Takeuchi, H.; Okumoto, Y. Mass transfer coefficients between gas and liquid phases in
packed columns. Journal of Chemical Engineering of Japan 1968, 1 (1), 56-62.
(94) Jammula, A. K. New liquid holdup, load point and flooding velocity models in different
regions of operations for a structured packed column. Oklahoma State University, 2014.
(95) Kenig, E. Y.; Wiesner, U.; Górak, A. Modeling of Reactive Absorption Using the
Maxwell−Stefan Equations. Industrial & Engineering Chemistry Research 1997, 36 (10), 4325-
4334. DOI: 10.1021/ie970186j.
(96) Krishnamurthy, R.; Taylor, R. A nonequilibrium stage model of multicomponent separation
processes. Part I: model description and method of solution. AIChE Journal 1985, 31 (3), 449-456.
(97) Kucka, L.; Müller, I.; Kenig, E. Y.; Górak, A. On the modelling and simulation of sour gas
absorption by aqueous amine solutions. Chemical Engineering Science 2003, 58 (16), 3571-3578.
DOI: https://doi.org/10.1016/S0009-2509(03)00255-0.
(98) Taylor, R.; Kooijman, H. A.; Hung, J. S. A second generation nonequilibrium model for
computer simulation of multicomponent separation processes. Computers & Chemical
Engineering 1994, 18 (3), 205-217. DOI: https://doi.org/10.1016/0098-1354(94)85009-7. Klöker,
M.; Kenig, E. Y.; Hoffmann, A.; Kreis, P.; Górak, A. J. C. E.; Intensification, P. P. Rate-based
modelling and simulation of reactive separations in gas/vapour–liquid systems. 2005, 44 (6), 617-
629.
(99) Kenig, E. Y.; Górak, A.; Pyhälahti, A.; Jakobsson, K.; Aittamaa, J.; Sundmacher, K.
Advanced rate‐based simulation tool for reactive distillation. AIChE Journal 2004, 50 (2), 322-
342. DOI: doi:10.1002/aic.10030. Kenig, E.; Górak, A. A film model based approach for
simulation of multicomponent reactive separation. Chemical Engineering and Processing:
Process Intensification 1995, 34 (2), 97-103.
166
(100) Patron, G. D.; Ricardez-Sandoval, L. A robust nonlinear model predictive controller for a
post-combustion CO2 capture absorber unit. Fuel 2020, 265, 116932.
(101) Zaman, M.; Lee, J. H. Optimization of the various modes of flexible operation for post-
combustion CO2 capture plant. Computers & Chemical Engineering 2015, 75, 14-27.
(102) He, Z.; Sahraei, M. H.; Ricardez-Sandoval, L. A. Flexible operation and simultaneous
scheduling and control of a CO2 capture plant using model predictive control. International
Journal of Greenhouse Gas Control 2016, 48, 300-311.
(103) Montañés, R. M.; Flø, N. E.; Nord, L. O. Dynamic process model validation and control of
the amine plant at CO2 Technology Centre Mongstad. Energies 2017, 10 (10), 1527.
(104) Gáspár, J.; Cormoş, A.-M. Dynamic modeling and validation of absorber and desorber
columns for post-combustion CO2 capture. Computers & Chemical Engineering 2011, 35 (10),
2044-2052. DOI: https://doi.org/10.1016/j.compchemeng.2010.10.001. Gabrielsen, J.; Michelsen,
M. L.; Stenby, E. H.; Kontogeorgis, G. M. Modeling of CO2 absorber using an AMP solution.
AIChE journal 2006, 52 (10), 3443-3451. Tobiesen, F. A.; Juliussen, O.; Svendsen, H. F.
Experimental validation of a rigorous desorber model for post-combustion capture. Chemical
Engineering Science 2008, 63 (10), 2641-2656. DOI: http://dx.doi.org/10.1016/j.ces.2008.02.011.
Kvamsdal, H. M.; Jakobsen, J. P.; Hoff, K. A. Dynamic modeling and simulation of a CO2
absorber column for post-combustion CO2 capture. Chemical Engineering and Processing:
Process Intensification 2009, 48 (1), 135-144. DOI: http://dx.doi.org/10.1016/j.cep.2008.03.002.
(105) Harun, N.; Nittaya, T.; Douglas, P. L.; Croiset, E.; Ricardez-Sandoval, L. A. Dynamic
simulation of MEA absorption process for CO 2 capture from power plants. International Journal
of Greenhouse Gas Control 2012, 10, 295-309.
(106) Flø, N. E.; Knuutila, H.; Kvamsdal, H. M.; Hillestad, M. Dynamic model validation of the
post-combustion CO 2 absorption process. International Journal of Greenhouse Gas Control
2015, 41, 127-141.
(107) Gaspar, J.; Fosbøl, P. L. A general enhancement factor model for absorption and desorption
systems: A CO2 capture case-study. Chemical Engineering Science 2015, 138, 203-215. DOI:
http://dx.doi.org/10.1016/j.ces.2015.08.023.
(108) Dang, H.; Rochelle, G. T. CO2 absorption rate and solubility in
monoethanolamine/piperazine/water. Separation science and technology 2003, 38 (2), 337-357.
167
(109) Van Krevelen, D.; Hoftijzer, P. Kinetics of gas‐liquid reactions part I. General theory.
Recueil des Travaux Chimiques des Pays‐Bas 1948, 67 (7), 563-586.
(110) Savage, D. W.; Astarita, G.; Joshi, S. Chemical absorption and desorption of carbon dioxide
from hot carbonate solutions. Chemical Engineering Science 1980, 35 (7), 1513-1522.
(111) Wellek, R.; Brunson, R.; Law, F. Enhancement factors for gas‐absorption with second‐order
irreversible chemical reaction. The Canadian Journal of Chemical Engineering 1978, 56 (2), 181-
186. Edwards, W. Section 14 in Perry s Chemical Engineering Handbook, Perry, RH; Green, DW;
Maloney, JO, Eds. McGraw-Hill Book Co., New York, NY: 1984. DeCoursey, W. Absorption
with chemical reaction: development of a new relation for the Danckwerts model. Chemical
Engineering Science 1974, 29 (9), 1867-1872. Porter, K. Effect of contact-time distribution on gas
absorption with chemical reaction. TRANSACTIONS OF THE INSTITUTION OF CHEMICAL
ENGINEERS AND THE CHEMICAL ENGINEER 1966, 44 (1), T25-&.
(112) Miller, D. C.; Siirola, J. D.; Agarwal, D.; Burgard, A. P.; Lee, A.; Eslick, J. C.; Nicholson,
B.; Laird, C.; Biegler, L. T.; Bhattacharyya, D. Next Generation Multi-Scale Process Systems
Engineering Framework. In Computer Aided Chemical Engineering, Vol. 44; Elsevier, 2018; pp
2209-2214.
(113) DeCoursey, W.; Thring, R. Effects of unequal diffusivities on enhancement factors for
reversible and irreversible reaction. Chemical engineering science 1989, 44 (8), 1715-1721.
(114) Asprion, N. Nonequilibrium rate-based simulation of reactive systems: simulation model,
heat transfer, and influence of film discretization. Industrial & engineering chemistry research
2006, 45 (6), 2054-2069.
(115) Tobiesen, F. A.; Svendsen, H. F.; Juliussen, O. Experimental validation of a rigorous
absorber model for CO2 postcombustion capture. AIChE Journal 2007, 53 (4), 846-865. Luo, X.;
Hartono, A.; Hussain, S.; Svendsen, H. F. Mass transfer and kinetics of carbon dioxide absorption
into loaded aqueous monoethanolamine solutions. Chemical Engineering Science 2015, 123, 57-
69.
(116) Faramarzi, L.; Thimsen, D.; Hume, S.; Maxon, A.; Watson, G.; Pedersen, S.; Gjernes, E.;
Fostås, B. F.; Lombardo, G.; Cents, T. Results from MEA testing at the CO2 Technology Centre
Mongstad: Verification of baseline results in 2015. Energy Procedia 2017, 114, 1128-1145.
168
(117) Nittaya, T.; Douglas, P. L.; Croiset, E.; Ricardez-Sandoval, L. A. Dynamic modeling and
evaluation of an industrial-scale CO2 capture plant using monoethanolamine absorption processes.
Industrial & Engineering Chemistry Research 2014, 53 (28), 11411-11426.
(118) Georgiadis, M. C.; Macchietto, S. Dynamic modelling and simulation of plate heat
exchangers under milk fouling. Chemical Engineering Science 2000, 55 (9), 1605-1619.
(119) Dardour, H.; Mazouz, S.; Bellagi, A. Numerical Analysis of Plate Heat Exchanger
Performance in Co-Current Fluid Flow Configuration. World Academy of Science, Engineering
and Technology 2009, 3, 2278-0181.
(120) Gut, J. A. W.; Pinto, J. M. Modeling of plate heat exchangers with generalized
configurations. International Journal of Heat and Mass Transfer 2003, 46 (14), 2571-2585. DOI:
http://dx.doi.org/10.1016/S0017-9310(03)00040-1.
(121) Lin, Y.-J.; Rochelle, G. T. Heat Transfer Enhancement and Optimization of Lean/Rich
Solvent Cross Exchanger for Amine Scrubbing. Energy Procedia 2017, 114, 1890-1903.
(122) Nellis, G.; Klein, S. Heat Transfer; Cambridge University Press, 2009.
(123) Pirnay, H.; López-Negrete, R.; Biegler, L. T. Optimal sensitivity based on IPOPT.
Mathematical Programming Computation 2012, 4 (4), 307-331.
(124) Smith, N.; Miller, G.; Aandi, I.; Gadsden, R.; Davison, J. Performance and costs of CO2
capture at gas fired power plants. Energy Procedia 2013, 37, 2443-2452.
(125) Sarda, P.; Hedrick, E.; Reynolds, K.; Bhattacharyya, D.; Zitney, S. E.; Omell, B.
Development of a dynamic model and control system for load-following studies of supercritical
pulverized coal power plants. Processes 2018, 6 (11), 226. Bankole, T.; Jones, D.; Bhattacharyya,
D.; Turton, R.; Zitney, S. E. Optimal scheduling and its Lyapunov stability for advanced load-
following energy plants with CO2 capture. Computers & Chemical Engineering 2018, 109, 30-47.
Heuberger, C. F.; Mac Dowell, N. Real-World Challenges with a Rapid Transition to 100%
Renewable Power Systems. Joule 2018, 2 (3), 367-370. DOI:
https://doi.org/10.1016/j.joule.2018.02.002.
(126) Hanak, D. P.; Biliyok, C.; Manovic, V. Evaluation and Modeling of Part-Load Performance
of Coal-Fired Power Plant with Postcombustion CO2 Capture. Energy & Fuels 2015, 29 (6), 3833-
3844. DOI: 10.1021/acs.energyfuels.5b00591.
(127) Bui, M.; Gunawan, I.; Verheyen, V.; Feron, P.; Meuleman, E.; Adeloju, S. Dynamic
modelling and optimisation of flexible operation in post-combustion CO2 capture plants—A
169
review. Computers & Chemical Engineering 2014, 61, 245-265. DOI:
https://doi.org/10.1016/j.compchemeng.2013.11.015.
(128) MacDowell, N.; Shah, N. The multi-period optimisation of an amine-based CO2 capture
process integrated with a super-critical coal-fired power station for flexible operation. Computers
& Chemical Engineering 2015, 74, 169-183.
(129) Cohen, S. M.; Rochelle, G. T.; Webber, M. E. Optimizing post-combustion CO2 capture in
response to volatile electricity prices. International Journal of Greenhouse Gas Control 2012, 8,
180-195. Cohen, S. M.; Rochelle, G. T.; Webber, M. E. J. J. o. E. R. T. Turning CO2 capture on
and off in response to electric grid demand: A baseline analysis of emissions and economics. 2010,
132 (2), 021003.
(130) Alie, C.; Backham, L.; Croiset, E.; Douglas, P. L. Simulation of CO2 capture using MEA
scrubbing: a flowsheet decomposition method. Energy Conversion and Management 2005, 46 (3),
475-487. DOI: http://dx.doi.org/10.1016/j.enconman.2004.03.003. Abu-Zahra, M. R.; Niederer, J.
P.; Feron, P. H.; Versteeg, G. F. CO2 capture from power plants: Part II. A parametric study of the
economical performance based on mono-ethanolamine. International journal of greenhouse gas
control 2007, 1 (2), 135-142. Rodríguez, N.; Mussati, S.; Scenna, N. Optimization of post-
combustion CO2 process using DEA–MDEA mixtures. Chemical engineering research and
design 2011, 89 (9), 1763-1773. Mores, P.; Scenna, N.; Mussati, S. A rate based model of a packed
column for CO2 absorption using aqueous monoethanolamine solution. International Journal of
Greenhouse Gas Control 2012, 6, 21-36. Li, K.; Leigh, W.; Feron, P.; Yu, H.; Tade, M. Systematic
study of aqueous monoethanolamine (MEA)-based CO2 capture process: techno-economic
assessment of the MEA process and its improvements. Applied energy 2016, 165, 648-659. Bui,
M.; Gunawan, I.; Verheyen, T.; Meuleman, E. Dynamic operation of liquid absorbent-based post-
combustion CO2 capture plants. In Absorption-Based Post-combustion Capture of Carbon
Dioxide, Elsevier, 2016; pp 589-621. Mahasenan, N.; Brown, D. R. Beyond the big picture:
Characterizaton of CO2-laden streams and implications for capture technologies. In Greenhouse
Gas Control Technologies 7, Elsevier, 2005; pp 1817-1820.
(131) Hasan, M. M. F.; Baliban, R. C.; Elia, J. A.; Floudas, C. A. Modeling, Simulation, and
Optimization of Postcombustion CO2 Capture for Variable Feed Concentration and Flow Rate. 1.
Chemical Absorption and Membrane Processes. Industrial & Engineering Chemistry Research
2012, 51 (48), 15642-15664. DOI: 10.1021/ie301571d.
170
(132) Feron, P.; Cousins, A.; Jiang, K.; Zhai, R.; Thiruvenkatachari, R.; Burnard, K. Towards zero
emissions from fossil fuel power stations. International Journal of Greenhouse Gas Control 2019,
87, 188-202.
(133) Mores, P.; Rodríguez, N.; Scenna, N.; Mussati, S. CO2 capture in power plants:
Minimization of the investment and operating cost of the post-combustion process using MEA
aqueous solution. International Journal of Greenhouse Gas Control 2012, 10, 148-163.
(134) Oh, S.-Y.; Kim, J.-K. Operational optimization for part-load performance of amine-based
post-combustion CO2 capture processes. Energy 2018, 146, 57-66. DOI:
https://doi.org/10.1016/j.energy.2017.06.179.
(135) Choi, J.; Cho, H.; Yun, S.; Jang, M.-G.; Oh, S.-Y.; Binns, M.; Kim, J.-K. Process design and
optimization of MEA-based CO2 capture processes for non-power industries. Energy 2019, 185,
971-980.
(136) Hamborg, E. S.; Smith, V.; Cents, T.; Brigman, N.; Falk-Pedersen, O.; De Cazenove, T.;
Chhaganlal, M.; Feste, J. K.; Ullestad, Ø.; Ulvatn, H. Results from MEA testing at the CO2
Technology Centre Mongstad. Part II: Verification of baseline results. Energy Procedia 2014, 63,
5994-6011.
(137) Thimsen, D.; Maxson, A.; Smith, V.; Cents, T.; Falk-Pedersen, O.; Gorset, O.; Hamborg, E.
S. Results from MEA testing at the CO2 Technology Centre Mongstad. Part I: Post-Combustion
CO2 capture testing methodology. Energy Procedia 2014, 63, 5938-5958. Shah, M. I. Co2 Capture
from Rfcc Flue Gas with 30w% Mea at Technology Centre Mongstad, Process Optimization and
Performance Comparison. In 14th Greenhouse Gas Control Technologies Conference Melbourne,
2018; pp 21-26.
(138) U.S. Energy Information Administration (EIA). Electric Power Monthly. 2020.
https://www.eia.gov/electricity/monthly/index.php (accessed 2020 September 4).
(139) Tester, J. W.; Modell, M. Thermodynamics and Its Applications, 3rd; Prentice Hall, Inc.,
New Jersey, USA, 1997.
(140) Ikada, E.; Hida, Y.; Okamoto, H.; Hagino, J.; Koizumi, N. Dielectric properties of
ethanolamines. 1969.
(141) Que, H.; Chen, C.-C. Thermodynamic Modeling of the NH3–CO2–H2O System with
Electrolyte NRTL Model. Industrial & Engineering Chemistry Research 2011, 50 (19), 11406-
11421. DOI: 10.1021/ie201276m.
171
(142) Van Ness, H.; Abbott, M. Vapor‐liquid equilibrium. Part VI. Standard state fugacities for
supercritical components. AIChE Journal 1979, 25 (4), 645-653.
(143) Chen, C. C.; Britt, H. I.; Boston, J. F.; Evans, L. B. Extension and application of the pitzer
equation for vapor‐liquid equilibrium of aqueous electrolyte systems with molecular solutes.
AIChE Journal 1979, 25 (5), 820-831.
(144) Yan, Y.; Chen, C.-C. Thermodynamic modeling of CO2 solubility in aqueous solutions of
NaCl and Na2SO4. The Journal of Supercritical Fluids 2010, 55 (2), 623-634. DOI:
https://doi.org/10.1016/j.supflu.2010.09.039.
(145) Zhang, Y.; Chen, C.-C. Thermodynamic modeling for CO2 absorption in aqueous MDEA
solution with electrolyte NRTL model. Industrial & Engineering Chemistry Research 2010, 50
(1), 163-175.
(146) Wilke, C.; Lee, C. Estimation of diffusion coefficients for gases and vapors. Industrial &
Engineering Chemistry 1955, 47 (6), 1253-1257.
172
You might also like
- Evo Morales Jon Stewart Interview TranscriptNo ratings yetEvo Morales Jon Stewart Interview Transcript2 pages
- An Integrated Model Framework For Carbon Management Technologies - Volume 1No ratings yetAn Integrated Model Framework For Carbon Management Technologies - Volume 1184 pages
- Modeling and Optimization of Adsorption-Based Separation Systems For Improved Process PerformanceNo ratings yetModeling and Optimization of Adsorption-Based Separation Systems For Improved Process Performance84 pages
- Monoethanolamine-Based CO2 Absorption Optimization in Natural Gas Plant Utility Units (1)No ratings yetMonoethanolamine-Based CO2 Absorption Optimization in Natural Gas Plant Utility Units (1)8 pages
- 525554.rev CEST2011 Matijasevic FullpaperNo ratings yet525554.rev CEST2011 Matijasevic Fullpaper8 pages
- Energy Science Engineering - 2015 - Li - Systematic Study of Aqueous Monoethanolamine%E2%80%90based CO2 Capture Process ModelNo ratings yetEnergy Science Engineering - 2015 - Li - Systematic Study of Aqueous Monoethanolamine%E2%80%90based CO2 Capture Process Model17 pages
- On The Use of Molecular-Based Thermodynamic Models To Assess The Performance of Solvents For CO Capture Processes: Monoethanolamine SolutionsNo ratings yetOn The Use of Molecular-Based Thermodynamic Models To Assess The Performance of Solvents For CO Capture Processes: Monoethanolamine Solutions54 pages
- Model validation and simulation issues for carbon dioxide reactive absorption with mixed aminesNo ratings yetModel validation and simulation issues for carbon dioxide reactive absorption with mixed amines10 pages
- Sciencedirect: Isotherm and Thermodynamic Analysis of Carbon Dioxide On Activated CarbonNo ratings yetSciencedirect: Isotherm and Thermodynamic Analysis of Carbon Dioxide On Activated Carbon8 pages
- Comparison of Different CO Recovery Processes in Their Optimum Operating Conditions FromNo ratings yetComparison of Different CO Recovery Processes in Their Optimum Operating Conditions From6 pages
- Practical Experience in Post-Combustion CO2 Capture Using Reactive Solvents in Large Pilot and Demonstration PlantsNo ratings yetPractical Experience in Post-Combustion CO2 Capture Using Reactive Solvents in Large Pilot and Demonstration Plants20 pages
- Simulation of CCS (MEA Absorption) Using Aspen Plus (Good) - 2019No ratings yetSimulation of CCS (MEA Absorption) Using Aspen Plus (Good) - 20195 pages
- A Century of Re-Exploring CO2 Capture SolventsNo ratings yetA Century of Re-Exploring CO2 Capture Solvents5 pages
- Exploring The Limits of Adsorption-Based Co Capture Using Mofs With Pvsa - From Molecular Design To Process EconomicsNo ratings yetExploring The Limits of Adsorption-Based Co Capture Using Mofs With Pvsa - From Molecular Design To Process Economics103 pages
- Demonstrating Full-Scale Post-Combustion CO2 Capture For Coal-Fired Power Plants Through Dynamic Modelling and SimulationNo ratings yetDemonstrating Full-Scale Post-Combustion CO2 Capture For Coal-Fired Power Plants Through Dynamic Modelling and Simulation14 pages
- Aqueous Ammonia Process For CO2 Capture: IOP Conference Series: Earth and Environmental ScienceNo ratings yetAqueous Ammonia Process For CO2 Capture: IOP Conference Series: Earth and Environmental Science2 pages
- 51 2021 CCST Techno-EconomicAnalysisofAmine-basedCO2CaptureTechnologyNo ratings yet51 2021 CCST Techno-EconomicAnalysisofAmine-basedCO2CaptureTechnology9 pages
- Aspen Plus Example of CO2 Capture From Syngas For IGCC Using MEA - v12No ratings yetAspen Plus Example of CO2 Capture From Syngas For IGCC Using MEA - v128 pages
- Toward Smart Carbon Capture With Machine Learning: PerspectiveNo ratings yetToward Smart Carbon Capture With Machine Learning: Perspective19 pages
- Process Analysis of Improved Process Modifications for Ammonia-basedNo ratings yetProcess Analysis of Improved Process Modifications for Ammonia-based13 pages
- Assignment 1 KVH Tharuka Egt 16 00540.editedNo ratings yetAssignment 1 KVH Tharuka Egt 16 00540.edited11 pages
- Wu - 18 - Modelling of A Post-Combustion Carbon Dioxide Capture Absorber Using Potassium Carbonate Solvent in Aspen Custom ModellerNo ratings yetWu - 18 - Modelling of A Post-Combustion Carbon Dioxide Capture Absorber Using Potassium Carbonate Solvent in Aspen Custom Modeller10 pages
- The Outlook For Improved Carbon Capture TechnologyNo ratings yetThe Outlook For Improved Carbon Capture Technology42 pages
- Formic Acid Synthesis A Case Study of CONo ratings yetFormic Acid Synthesis A Case Study of CO17 pages
- Production of Methane by Using Carbon Capturing Method for Better SustainabilityNo ratings yetProduction of Methane by Using Carbon Capturing Method for Better Sustainability18 pages
- Advanced Materials - 2017 - Adil - Valuing Metal Organic Frameworks For Postcombustion Carbon Capture A Benchmark StudyNo ratings yetAdvanced Materials - 2017 - Adil - Valuing Metal Organic Frameworks For Postcombustion Carbon Capture A Benchmark Study11 pages
- CO2 capture from syngas for IGCC using MEA with HYSYSNo ratings yetCO2 capture from syngas for IGCC using MEA with HYSYS10 pages
- Guidelines for the Determination of Standardized Semiconductor Radiation Hardness ParametersFrom EverandGuidelines for the Determination of Standardized Semiconductor Radiation Hardness ParametersNo ratings yet
- Summary Review on the Application of Computational Fluid Dynamics in Nuclear Power Plant DesignFrom EverandSummary Review on the Application of Computational Fluid Dynamics in Nuclear Power Plant DesignNo ratings yet
- 4th Quarter Module Week 3-4 (Food Processing 10) : Department of Education100% (2)4th Quarter Module Week 3-4 (Food Processing 10) : Department of Education9 pages
- Industrial Automation & Robotics Final Exam AssignmentNo ratings yetIndustrial Automation & Robotics Final Exam Assignment2 pages
- English Sample Essays SPM (I Do Not Own The Copyright)No ratings yetEnglish Sample Essays SPM (I Do Not Own The Copyright)11 pages
- Spreying and Seeding Robo".: Design and Operation of Agriculture Based Pesticide Spraying and Seeding RobotNo ratings yetSpreying and Seeding Robo".: Design and Operation of Agriculture Based Pesticide Spraying and Seeding Robot16 pages
