DEGREE PROJECT IN THE FIELD OF TECHNOLOGY

ENGINEERING CHEMISTRY
AND THE MAIN FIELD OF STUDY
CHEMICAL SCIENCE AND ENGINEERING,
SECOND CYCLE, 30 CREDITS
STOCKHOLM, SWEDEN 2020

Stripper Modification of a Standard

MEA Process for Heat Integration
with a Pulp Mill

PATY ARANGO MUÑOZ

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ENGINEERING SCIENCES IN CHEMISTRY,
BIOTECHNOLOGY AND HEALTH
 Abstract

The 20 largest pulp mills in Sweden emit around 20 million tonnes of CO2 per year. These
emissions are considered carbon-neutral since they originate from biogenic sources. The
pulp and paper industry is therefore a good candidate for the application of BECCS
(Bioenergy with Carbon Capture and Storage) and has the potential to play a significant role
for reaching the long-term mitigation target set by the Swedish government that Sweden
should be climate-neutral by year 2045. In this thesis, a MEA-based chemical absorption and
desorption process was rigorously modelled in Aspen Plus using the rate-based method.

Validation of the absorber and stripper model was conducted before the standard process
was modified to a configuration that enables heat integration of a significant amount of excess
heat from the capture process in, for example, a Kraft pulp mill. CO2 removal rate and rich
solvent loading were used as performance indicators to validate the absorber columns. The
reboiler duty and lean solvent loading served as performance indicators in the stripper
validation. The columns were dimensioned considering 90 wt% capture rate. Efficient use of
the entire packing in the absorber and stripper columns was ensured by testing different
solvent flow rates.

Suitable temperature levels for heat integration, within and across the capture plant, were
obtained through an assessment of different versions of a stripper overhead compression
configuration. The evaluation of the modified MEA processes took into account the steam
conservation potential and energy efficiency potential. The simulation results indicate that the
modified stripper may lead to savings of up to 11% in steam consumption. Heat integration
between the capture plant and a specific process in a reference Kraft pulp mill resulted in
energy savings of the same order of magnitude. Thereby, making the BECCS concept a more
attractive solution for the Swedish pulp and paper industry to mitigate climate change.

Keywords
Carbon dioxide capture, pulp mill, chemical absorption, modelling, heat integration
 Sammanfattning

De 20 största massabruken i Sverige släpper tillsammans ut ungefär 20 miljoner ton CO2 per
år. Dessa utsläpp har biogent ursprung och anses därför vara klimatneutrala. Massa- och
pappersindustrin är därmed en lämplig kandidat för implementeringen av BECCS (eng.
Bioenergy with Carbon Capture and Storage) och har en betydande potential att nå de, av
den svenska regeringen, uppsatta klimatmålen som säger att Sverige inte ska några
nettoutsläpp av växthusgaser till atmosfären senast år 2045. I detta examensarbete
simulerades kemiska absorptions- och desorptionsprocesser med MEA som lösningsmedel
genom att tillämpa den hastighetsbaserade metoden i en rigorös modell i Aspen Plus.

Stripper- och absorptionsmodellerna validerades innan standardprocessen modifierades till

en konfiguration som möjliggör värmeintegration av koldioxidinfångningens överskottsvärme
med, exempelvis, ett sulfatmassabruk. Avskiljningsgraden och laddning hos den mättade
lösningen användes som prestandaindikatorer för att validera absorptionskolonnerna.
Återkokarens energiåtgång och laddning hos den omättade lösningen användes som
prestandaindikatorer för att validera stripperkolonnerna. Samtliga kolonner dimensionerades
för att erhålla 90 vikt% avskiljningsgrad. Olika flödeshastigheter av lösningsmedlet testades
för att säkerställa effektivt nyttjande av packningen i absorptions- och stripperkolonnerna.

Lämpliga temperaturnivåer för värmeintegration, inom och utanför, koldioxidinfångningen

erhölls genom att utvärdera olika varianter av en stripper-overhead-kompression
konfiguration. Utvärderingen av den modifierade MEA processen tog hänsyn till potentialen
för ångbesparing och energieffektivisering. Resultat från simuleringarna tyder på att den
modifierade strippern skulle kunna ge besparingar på upp emot 11 % i ånganvändning.
Energibesparingar i samma storleksordning kunde även erhållas genom värmeintegration
mellan koldioxidinfångningen och en särskild process i ett referensbruk. Implementering av
BECCS-konceptet på det här sättet skulle därmed kunna bli ett mer attraktivt alternativ för
den svenska massa- och pappersindustrin att bekämpa klimatförändringarna.

Nyckelord
Koldioxidinfångning, massabruk, kemisk absorption, modellering, värmeintegration
 Acknowledgments

I would like to express my sincere gratitude to my supervisor Jens Wolf from RISE
Bioeconomy for entrusting me with this project and for his unconditional encouragement,
guidance and support during the course of this thesis. Jens and his colleagues at RISE went
out of their way to make me feel welcome in the organization.

I am also particularly grateful for the assistance and support provided by my supervisor and
examiner Matthäus Bäbler. He made sure to steer me in the right direction regarding the
content of the thesis and always provided valuable input.

Lastly, my gratitude towards my parents, Rosa and Javier, and my brother Alberto is beyond
measure. They have constantly believed in me and made sacrifices to ensure that I had the
best opportunities possible throughout my entire life. Special thanks to the rest of my family
and friends that are always supporting me.
 Contents

1. Introduction 1
1.1. Aim ............................................................................................................................... 1
2. Carbon capture 2
2.1. Bioenergy with CCS .................................................................................................... 2
2.2. Carbon capture technologies ...................................................................................... 3
2.3. Amine absorption ......................................................................................................... 4
3. Swedish pulp mills 6
3.1. Key processes ............................................................................................................. 6
3.1. Carbon emissions ........................................................................................................ 8
3.2. Process integration ...................................................................................................... 9
4. MEA process modelling 10
4.1. Standard process configuration................................................................................. 10
4.2. Equilibrium and rate-based models........................................................................... 11
4.3. Modelling tools ........................................................................................................... 12
4.4. Thermodynamics and kinetics ................................................................................... 13
4.5. Material transport ....................................................................................................... 14
5. Methodology 15
5.1. Validation ................................................................................................................... 16
5.1.1. Absorber ............................................................................................................................... 16
5.1.2. Stripper ................................................................................................................................. 18
5.2. RadFrac model .......................................................................................................... 19
5.3. Dimensioning ............................................................................................................. 20
5.3.1. Absorber ............................................................................................................................... 21
5.3.2. Stripper ................................................................................................................................. 22
5.4. Standard MEA process.............................................................................................. 23
5.5. Modified MEA process............................................................................................... 25
6. Results and discussion 28
6.1. Standard MEA process.............................................................................................. 28
6.2. Modified MEA process............................................................................................... 28
7. Conclusion 32
8. Future work 33
8.1. Aspen Plus model ...................................................................................................... 33
 8.2. Carbon capture design .............................................................................................. 33
References 34

Appendix A 40
Absorption validation: parameters and results
Appendix B 44
Stripper validation: parameters and results
Appendix C 45
Absorber dimensioning
Appendix D 46
Stripper dimensioning
Appendix E 47
Validation of external partial condenser
Appendix F 48
Steam conservation potential
 Nomenclature

Abbreviations

ADt Air dry tonne

BECCS Bioenergy with Carbon Capture and Storage
CCS Carbon Capture and Storage
CSTR Continuous Stirred-Tank Reactor
DCC Direct Contact Cooler
dP Pressure drop
DS Dry Solid
ENRTL-RK Electrolyte Non-Random Two Liquid with Redlich-Kwong thermodynamic model
GMENCC Aspen Plus ENRTL interaction parameters for molecule-ion and ion-ion pairs
GtCO2e Gigatonnes of equivalent carbon dioxide
H/D Height to diameter ratio
IEA International Energy Agency
IPCC Intergovernmental Panel on Climate Change
L/G Liquid to gas ratio
Lmin Minimum lean solvent flow rate
NCG Non-Condensable Gases
Post-CC Post-Combustion Capture
PSRK Predictive Soave-Redlich-Kwong equation of state model
RISE Research Institutes of Sweden
SINTEF Foundation for Scientific and Industrial Research at the Norwegian Institute of
Technology
TRL Technical Readiness Level
TRS Total Reduced Sulfur
tCO2 Tonne CO2 captured
UTA University of Texas at Arlington

Chemical compounds

AMP 2-amino-2-methyl-1-propanol
CO2 Carbon dioxide
H2 O Water
K2CO3 Potassium carbonate
MDEA Methyldiethanolamine
MEA Monoethanolamine
NH3 Ammonia
NO2 Nitrogen dioxide
O2 Oxygen
PZ Piperazine
SOx Sulfur oxides
SO2 Sulfur dioxide
SO2 Sulfur trioxide
 1. Introduction
One of the most frequently used processes for CO2 capture from flue gases is alkanoamine
scrubbing as it has been known since the beginning of the 20th century (Hammer et al., 2006)
and is currently used in several industry processes, such as natural gas sweetening,
ammonia production and syngas upgrading and synthesis. However, the principal aim of the
CO2 capture in the early years was the separation of the CO 2 to upgrade the natural gas
found in reservoirs for commercial purposes. The use of this technology for emission
reductions in the process industry with the purpose of tackling the greenhouse effect was first
ed i he 1970 . A h gh, i a i ece ea ha he ca b ca ea d
storage (CCS) concept started to attract an increasing number of large project initiatives.

The Swedish government has set the long-term climate goal that Sweden should be climate-
neutral by year 2045 (Regeringskansliet, 2017). Some interesting large point sources of CO2
emissions in Sweden with respect to CCS are industries such as pulp and paper, integrated
steel mills, oil refineries, cement and chemical production plants. Most of the attention for
CO2 abatement has nonetheless been towards the fossil-fueled industries and much less on
the pulp and paper industry. The CO2 emissions from pulp and integrated pulp and paper
mills in this heavily industrialized region originate from biogenic sources, thus considered
carbon-neutral (Ga a d i , 2017). This industry is therefore a good candidate for the
application of BECSS, a concept that combines bioenergy with CCS, especially since a
conventional pulp mill emits millions of tonnes of CO2 per year (Global CCS Institute, 2011).

1.1. Aim
The main objective of the thesis is to build a rate-based model of an amine-based
absorption/desorption process for CO2 capture in Aspen Plus. Another objective is to
evaluate the possibility to integrate excess heat from a modified capture process in, for
example, a Kraft pulp mill. However, due to patent pending by RISE, the content of this report
will neither disclose which of the reference pulp mill processes is considered for such an
integration scenario nor present any quantitative information relating to the needed
customization of the processes involved. The thesis objectives were accomplished through:

Validation of the absorber and stripper model by comparing simulations results with
data from four different experimental trials that were found in literature.
Modification of the capture process by adding one vapor compression step in the
stripper reflux.
Calculation of performance values, e.g. heat duty, and resulting heat temperature
levels in the standard and modified MEA processes.
Evaluation of the energy saving potential in a heat integration scenario between an
energy intensive pulp mill process and the carbon capture plant.

1
 2. Carbon capture
By the end of 2019, over 260 million tonnes (Mt) of CO 2 emissions from anthropogenic
sources have already been captured and permanently stored. Most of the injected CO 2 was
used to increase oil production through enhanced oil recovery. 40 Mt per year is the
approximate global capture and storage capacity of large-scale CCS facilities that are
currently operating or under construction. Although, if including the facilities at earlier stages
of development the total number of large-scale facilities would add up to 51; accordingly, their
capture and storage capacity is around 98 Mt per year. The largest share of CO2 captured
from these plants all together originates from either coal fired power plants, natural gas
processing or coal gasification plants. In addition to these larger projects, there are 39 pilot
and demonstration scale CCS facilities (Global CCS Institute, 2019). In the general case,
BECCS has been recognized at an international level, e.g. in reports published by the IEA
and IPCC respectively, for being the only large-scale technology capable of removing CO 2
from the atmosphere (ZEP & EBTP, 2012).

2.1. Bioenergy with CCS

The e c ed ca b c ce he ie f bi a a a ca b -neutral energy
source by assuming that the equivalent amount of CO2 released during the energy use of
biomass is captured from the atmosphere during its growth. Concerns that the likelihood of
achieving the required safe atmospheric CO 2 stabilization level is diminishing with every
passing year has driven the attention towards carbon-negative energy systems. Bioenergy
with CCS falls under the definition of such systems, which enable a greater amount of
atmospheric CO2 to be absorbed and sequestered by the system than the CO 2 amount that
would ultimately be released back to the atmosphere if not removed from the natural cycle
(Muradov, 2014). The prerequisite for this notion to be valid is that the biomass is managed
in a sustainable way (Onarheim et al., 2017). More specifically, the concept of BECCS refers
to the capture of CO2 released, through combustion or processing, at biomass point emission
sources as well as the transport and injection of it into deep underground in geological
formations for permanent storage. Not only can the recapturing and permanent storage of
CO2 with this approach lead to negative CO 2 emissions (Figure 1) but it also enables
mitigation of carbon emissions that have already occurred (Global CCS Institute, 2011).

Figure 1 Illustration of how emissions that have already occurred can be mitigated by biomass
fueled industrial plants employing the BECCS concept (Global CCS Institute, 2011).

2
The use of BECCS in several model studies was reported in the Fifth Assessment Report of
IPCC (2014) in which several scenarios of pathways are taken into consideration to achieve
climate stabilization at 2ºC above pre-industrial levels. This report revealed the full range of
potential climate impact of BECCS systems as 0-22 Gt of negative CO2 emissions per year
in 2100, where the largest potential corresponds to scenarios with the highest mitigation
ambition consistent with a 2ºC target (Smith & Porter, 2018). The mean level of the overall
potential for BECCS was estimated at around 12.1 GtCO 2e/year (Fuss et al., 2016). Although,
more realistic estimates through cautious assumptions in regard to a sustainable supply of
biomass suggest a much lower removal capacity of BECCS: 3.3-7.5 GtCO2e/year (Mclaren,
2012). Furthermore, the emission consequences between different BECCS technologies can
differ as the amount of sequestered CO2 may vary along the supply chain and the bioenergy
may substitute different technologies (Fuss et al., 2016); for example, co-firing fossil fuels
with 30% biomass requires a larger geological storage capacity when compared to the
corresponding co-firing fossil fuels with 50% biomass (Mclaren, 2012).

2.2. Carbon capture technologies

Aqueous amine solutions, the most common among chemical absorption technologies, have
been used to remove CO2 from natural gas for several decades (Bottoms, 1931). Two large-
scale coal fired plants that employ an amine-based system for post-combustion capture
(Post-CC) are the Boundary Dam in Canada and the Petra Nova in the United States.
Therefore, amine-based solvents are considered to have a technical readiness level (TRL)
of 9 when implemented in power plants with post-combustion capture (Bui et al., 2018). With
just three years in between their respective commissioning, the cost per tCO2 (tonne CO2
captured) reduced from over $100 at the Boundary Dam to below $65 at the Petra Nova.
Furthermore, most recent studies estimate the corresponding cost for facilities planning to
start their operation in 2024-2028 at around $43 (Global CCS Institute, 2019). All this
indicates that the more frequent the commissioning of mature carbon capture systems for
industrial facilities becomes the more cost-feasible capture systems can be expected in the
near future.

Among the most recent developments in polymeric membranes, the commercially available
Polaris membrane has achieved TRL 7 and if its projected implementation goes as planned
it may reach TRL 8 next year already (Batoon et al., 2019). Moreover, it has proven
successful when used in post-combustion systems for CO2 separation from syngas (MTR,
n.d.). The suitability and selection of a capture technology will depend on the specific stream
properties of any industrial process, such as moisture content and CO2 concentration. This
matter is specially addressed in a study (Hasan et al., 2012) that evaluated both an absorption
process and a membrane process over a range of feed compositions and flow rates. Capture
processes that include either membrane or adsorption require a moisture content of 0.1% or
less to circumvent reduced CO2 recovery and corrosion problems, respectively. On the other
hand, absorption-based processes can tolerate a feed saturated with H2O and will therefore
handle a high moisture feed much better. The lower concentration of CO 2 the more capital
and energy intensive separation process of the gas. Operation under oxidizing atmosphere
may shorten the life of chemical solvents used in gas separation units (Muradov, 2014).

After having covered the necessary technological requirements for the capture process of
CCS projects, their success will also depend on the availability of safe geological storage for
the captured CO2. Other factors that can help bring such projects into the operation phase

3
comprise supportive policy, legislative frameworks and secure financial funding (Bui et al.,
2018). Unfortunately, negative carbon emissions are currently not taken into account in the
European Union Emissions Trading System (Onarheim et al., 2017). Despite the non-existent
incentives for implementation of CCS technologies in the pulp and paper industry, several
studies have been published on this research topic in recent years. The extra energy demand
needed for the capture process was assessed in a study (Hektor, 2008) in which a
comparison of two different absorbents, monoethanolamine (MEA) and chilled ammonia, also
was done. As different configurations for the energy supply were considered, it was
concluded that an increasing degree of heat integration can be profitable for pulp and paper
mills when choosing chilled ammonia over MEA as the solvent. The study also denoted that
post-combustion capture of CO2 can be economically feasible for the pulp and paper industry
under certain favorable market conditions.

2.3. Amine absorption

Amine-based absorption is the most commonly used process in Post-CC projects for gas
cleanup. A series of advanced amines with improved stability and properties have been
developed with the aim of lowering stripping steam requirement and enabling their use in
power plants with thermal integration of carbon capture systems: sterically hindered amines
(KS-1, KS-2 and KS-3), 2-amino-2-methyl-1-propanol (AMP), Cansolv and HTC Purenergy.
The main challenges to the amine-based carbon capture systems when applied to Post-CC
are associated with large parasitic loads due to: heating required for the absorbent
regeneration, pumping of solutions and compressing of the gas to pipeline pressure
(Muradov, 2014). The capital and operating costs of CO2 capture processes with sorbents
are largely dictated by their kinetic and thermophysical properties. The chemical absorbent
MEA has become the benchmark amine for CO2 capture from electricity generation due to its
particular suitability for low CO2 partial pressure applications (Bui et al., 2018).

Table 1 Energy consumption for various carbon capture absorbents, including single amine and
amine blends, based on a standard absorber/stripper process configuration. The duty marked with
asterisk (*) is calculated through simulation (Bui et al., 2018).

Solvent Reboiler duty (GJ/tCO2)

30 wt% MEA 3.6 4.0
40 wt% MEA 3.1 3.3
40 wt% piperazine (PZ) 2.9
Cansolv 2.3
28 wt% AMP + 17 wt% PZ 3.0 3.2
MEA + MDEA (variable mix ratio) 2.0 3.7
Aqueous ammonia (NH3) 2.0 2.9*
Aqueous potassium carbonate (K2CO3) 2.0 2.5

Using 30 weight percent (wt%) MEA for 90% CO2 removal from flue gas (10 15 kPa CO2) in
a standard separation process usually requires stripper reboiler duties of at least 3.6 4.0 GJ
per tCO2. Reducing this value has become the primary goal in the chemical absorbent

4
research, even if the reboiler energy requirement is not the only metric that defines absorbent
performance. Consequently, new absorbents are benchmarked against the value for 30 wt%
MEA by default. Table 1 shows this value among the corresponding energy consumption
values of other absorbents. The widespread use of MEA in industry is due to its favorable
characteristics in terms of CO2 mass transfer rate, biodegradability and cost. The most
prominent downsides with this solvent are moderate levels of toxicity, moderate rates of
oxidative and thermal degradation. Moreover, it has corrosive tendencies when used at
higher concentrations. In the case of MEA and methyl-diethanolamine (MDEA) blends, the
reboiler duty increases with higher ratios of MEA (Bui et al., 2018). Recent process
development indicates that the thermal degradation and corrosiveness of MEA, when used
at higher concentrations than 30 wt%, can be circumvented via additives. Introducing high
performance oxidative inhibitors would enable the feasibility of 40 wt% MEA, thereby
reducing the regeneration energy demand to below 3.3 GJ per tCO2 (Lemaire et al., 2014).

The MEA solvent is sensitive to impurities such as nitrogen dioxide (NO 2), oxygen (O2), sulfur
oxides (SOx) and dust. The quality of the flue gases to be treated should therefore be
considered when integrating a Post-CC process to an industrial process. For example, amine
degradation and solvent foaming is triggered by dust and particulate matters. These
impurities could also lead to plugging and scaling of process equipment. Sulfur oxides of the
type SO2 and SO3 are formed during combustion of fuel containing sulfur and react with
amines to form heat stable salts. The SO2 concentration in flue gases from the recovery boiler
and the power boiler of a pulp mill is typically lower than in the flue gas from coal-fired power
plants. However, if the sulfur content would be high, the addition of a flue gas desulfurization
plant should be implemented prior to the capture unit. Alternatively, the direct contact cooler
(DCC) could be adjusted to scrub out sulfur components by operating with an appropriate
dosage of alkali solution (IEAGHG, 2016).

Amine degradation alters the viscosity, conductivity and pH of the absorbent as well as the
CO2 absorbing capacity. Plant operation parameters affected by excessive amine-
degradation rates include susceptibility of the absorbent to foam, pressure drop across
packed column and the liquid to gas ratio (L/G) required to maintain CO2 capture rates.
Atmospheric emissions, amine reclamation and corrosion are some other aspects of the
design and operation of a Post-CC plant that are closely related to amine degradation
(Reynolds et al., 2016).

5
 3. Swedish pulp mills
Producing pulp and paper requires large amounts of heat and power input. The pulp, paper
and printing industry together stand for 5.7% of global industrial final energy use, of which
printing is a very small share (IEA, 2007). The total emissions from large Swedish pulp mills,
with annual emissions exceeding 500 kt, add up to around 20 Mt CO2 per year. These
emissions arise from combustion of residual biomass streams that cannot be made into pulp.
The power and steam produced through combustion is primarily intended for the internal
energy use in the pulp and paper mill. However, these by-product streams may in some cases
undergo conversion to electricity for the purpose of external energy use (Normann et al.,
2019). The following text summarizes the main processes (Figure 2) in a modern Swedish
pulp mill and explains which carbon emission sources are the most relevant to consider in
capture scenarios of this industry.

Figure 2 Overview of the key processes in a Kraft pulp mill (Normann et al., 2019)

3.1. Key processes

Chemical pulping based on the Kraft process dominates in Europe with respect to its whole
production of market pulp, i.e. produced from non-integrated pulp mills. According to gathered
data from 2008, Sweden was the leading pulp producer in Europe that year since all of its
Kraft pulp mills combined together produced 12.1 Mt pulp per year (Suhr et al., 2015). The
Kraft chemical process is also the most commonly used pulping method in Sweden. This
process begins with the separation of the cellulose from the wood by cooking the raw material
in a chemical mixture, the so called white liquor. While the pulp material, i.e. separated
cellulose, is being processed into the desired pulp or paper product, the spent cooking
chemicals will be recovered in the recovery boiler by combusting it together with the
remainder of the wood, e.g. dissolved lignin. The liquid process stream entering the recovery
boiler is usually referred to as black liquor (Garðarsdóttir et al., 2018). Normally, the power
plant configuration in a non-integrated kraft pulp mill constitutes a recovery boiler and a power

6
boiler feeding a back-pressure turbine. The heat generated upon burning the strong
(concentrated) black liquor in the recovery boiler is used to produce high-pressure (HP),
superheated steam. Part of this HP steam will be converted to electrical power as it passes
the turbine. Medium-pressure (MP) and low-pressure (LP) steam are needed to cover the
heat energy demand in other parts of the kraft pulp process (Suhr et al., 2015).

Figure 3 Illustration of the main process steps of the calcium and alkali recovery circuits in a Kraft
pulp mill (BMU Austria, 1995)

The recovered cooking chemicals, i.e. weak black liquor, are processed in the chemical
recovery system (Figure 3) before re-used in other parts of the pulping process (ZEP & EBTP,
2012). The main processes in the chemical recovery system are black liquor evaporation,
incineration of evaporated liquors and causticizing. Key functions of these process steps
include:

Recovery of inorganic pulping chemicals

Prevention and control of pollution through significant reduction of the wastewater
load discharged and extensive reduction of emissions to air
Recovery of the energy content as process steam and electrical power
Incineration of dissolved organic material (Suhr et al., 2015)

The black liquor collected from pulp washing usually contains around 14 18% dissolved
solids and needs to be concentrated before it can be burnt. The dry solid (DS) content in the
liquor is increased in a multi-effect evaporation plant to 70 85%. The upper limit for the dry
solid increase is usually 72 74% at atmospheric pressure. Modern plants pressurize the
liquor to concentrate it up to 85% DS. Some of the combustible material will separate as non-
condensable gases (NCG) during evaporation. These gases contain malodorous gas
compounds that vary in their degree of contamination according to their origin (Suhr et al.,
2015). Swedish recovery boilers usually combust black liquor with a dry solid content of
around 70% (FRBC, 2010). Usually, black liquor evaporation is the highest steam consumer
in energy-efficient market bleached kraft pulp mills (Table 2). However, heat and electricity

7
consumption data should not be interpreted without specific energy balances of the mill, since
methods to monitor, calculate and report the energy used differ between mills. The
geographical location of the mill may also affect the heat consumption to some degree (Suhr
et al., 2015).

Table 2 General steam consumption levels of a market bleached kraft pulp mill with well-designed
and operated processes, expressed as an annual average. Steam for electrical power production
and the primary thermal energy necessary for lime reburning are not included (Suhr et al., 2015).
Process Cooking O2 delignification Bleaching Drying Evaporation Other Total
Steam
consumption 1.6 2.0 0.2 0.4 1.5 2.0 2.2 2.6 4.0 4.5 1.5 2.0 11-12
(GJ/ADt)

3.1. Carbon emissions

Pulp and integrated pulp and paper mills represent the majority of the large point sources of
CO2 in this heavily industrialized region. Introducing the BECCS concept in the pulp and
paper industry would therefore enable a significant potential for creating CO 2 -negative
facilities in Sweden. The recovery boiler accounts for around 75% of the total plant emissions.
Almost all Swedish pulp and paper mills use bio-oil rather than fossil-based oil in their lime
kilns, which are responsible for 10-15% of the emissions. The remainder emissions originate
from the power boiler, also around 10 15%, in which the bark and other biofuels from the
wood that did not get used in the process are combusted. Because of the variable operation
of the power boiler, it is considered to be the least feasible source for CO2 capture
(Garðarsdóttir et al., 2018). Besides black liquor, waste wood is sometimes also used in the
recovery boiler and power boiler. These types of fuel classify as biomass derived fuels.
Therefore, 75-100% of CO2 emitted from a modern pulp mill or integrated pulp and paper mill
is commonly assumed to be carbon neutral as both fuels are considered biogenic if they are
sourced sustainably (Onarheim et al., 2017).

Some features of modern pulp and paper mills that have been identified as challenges in
applying CCS retrofit are: technical restrictions in the plant layout, impurities in the flue gas,
the size of single sources, as well as heavily integrated processes (ZEP & EBTP, 2012).
Energy supply can also be a limiting factor for carbon capture implementation when paper
production is integrated in Kraft pulp mills (Kuparinen et al., 2019). Thermal integration of the
capture process is usually considered with the intention of improving energy efficiency upon
implementation of carbon capture at different mills. Most importantly, it is the plant-specific
process layout that reflects on the potential for such integration. True steam surplus is
commonly available at Kraft pulp mills and pure thermomechanical pulp mills. On the other
hand, integrated Kraft pulp and paper mills and paper mills without virgin pulp production lack
steam surplus and would therefore need to import external fuel to cover the extra heat
demand that the capture process implies (Jönsson & Berntsson, 2012). New opportunities
for BECCS may emerge if black liquor gasification is introduced or when the Swedish pulp
and paper industry start including biofuels and/or specialty chemical production, i.e.
biorefining (Rootzén et al., 2018), in their product and process portfolios.

8
 3.2. Process integration
System solutions are as important as new technologies for reducing energy use in industry.
Process integration measures vary from case to case in terms of technical solutions and
energy saving. Identifying where and to what extent process integrations tools have been
used and how much they have contributed to energy saving is more difficult than with new
technology solutions. The three main features of process integration methods are the use of:
rules of thumb (heuristics), thermodynamics and optimization techniques. There is usually a
significant overlap between various methods. Nowadays, the trend is strongly towards
methods that incorporate all three features. Exergy analysis and pinch analysis are methods
with particular focus on thermodynamics. The latter has proven to be powerful when
developing new mill processes and concepts. Large potentials for energy savings in the pulp
and paper industry have been identified through studies in the United States, Canada,
Finland and Sweden. Results from these process integration studies, mainly pinch analyses,
have been implemented in more than 100 mills worldwide. New process integration tools and
methods, e.g. more efficient heat exchanger networks, can lead to energy savings in the
order of 10 40% for chemical pulp mills with relatively high energy consumption. Novel
system solutions, such as integration of the secondary heat system and the evaporation
plant, can lead to energy saving in the order of 15-30% for energy efficient mills (IEA, 2007).

Even though the emission levels from the pulp and paper industry are comparable to those
of fossil-fueled industries, no pilot or demonstration initiatives focusing on CCS have been
established for this industry yet (Ga a d i , 2017). Instead, several conceptual studies
have been conducted in order to assess the viability of implementing BECCS in the pulp and
paper industry. Onarheim et al. (2017) concluded in one of these studies that favorable
opportunities for heat integration may exist for this industry by implementing CO 2 capture
technologies. Their evaluation considered two hypothetical reference mills situated in Finland
and the capture cases assessed involved the CO 2 emissions of the flue gases from the
recovery boiler, the power boiler and the lime kiln and various combinations of these. The
results indicated that the technical feasibility of retrofitting Post-CC to an existing pulp mill or
pulp and board mill is very dependent on the existing power and steam production onsite.
For example, in the case of an integrated pulp and board mill that aims for 90% capture rate,
excess steam produced onsite will not be sufficient and will therefore require an auxiliary
boiler to assist with supplement steam. Whether the extra steam demand of the CO 2 capture
plant is sufficient or not also depends on the flue gas volume and its partial pressure of CO 2.

9
 4. MEA process modelling
This chapter begins with the description of a process flow diagram consisting of two major
parts, the absorption and desorption processes. The absorber is the most expensive and
largest unit (Øi, 2010), whereas the reboiler connected to the desorption column is the main
contributor to the total energy demand of this capture process (Li et al., 2016). Then, a brief
comparison between equilibrium-based and rate-based models will be provided prior to
explaining the most relevant chemical reactions in the MEA process.

4.1. Standard process configuration

A schematic representation of the amine-based Post-CC plant is given in Figure 4. The flue
gas will enter the bottom the absorber (2) after passing through the flue gas conditioning unit
(1) that usually includes a DCC, where it is quenched and most of the residual particular
matter is removed (IEAGHG, 2016). The absorber operates at a lower temperature than the
stripper to ensure a higher affinity for CO2 absorption. The lean amine solvent will selectively
absorb CO2 as it comes in contact with the flue gas throughout the absorber packing. CO2
dissolves into the absorbent after it has diffused from the bulk gas to the gas-liquid interface.
As CO2 reacts with the amine present in solution, its concentration will be depleted at the
gas-liquid interface. This decrease is what maintains the driving force for CO2 to move from
the gas to the liquid phase (Puxty & Maeder, 2016). While the solvent flows downwards
through the absorber, the CO2-lean gas moves countercurrent with it and passes through the
washing section at the top of the column before it is emitted to the atmosphere. Any traces
of amine components and degraded by-products will be removed from the treated flue gas in
the water wash section, in the top part of the absorber column (Onarheim et al., 2017).

Figure 4 Conventional CO2 absorption-desorption process configuration (Ga a d i , 2017).

The CO2-rich solvent is withdrawn from the bottom of the absorber and is then directed to the
top of the stripper (5) through the solvent heat exchanger (3), in which residual heat from the
hot CO2-lean solvent is recovered. Hot steam is produced in the stripper reboiler (6) and flows
upwards counter-currently to the rich solvent to assist in the thermal regeneration of the
absorbent. Steam is used in the desorption column to maintain the absorbents temperature
since the process is endothermic, as well as to maintain the desorption driving force by
diluting the CO2 being released below the equilibrium partial pressure (Puxty & Maeder,

10
2016). The regenerated solvent discharged from the bottom of the stripper is cooled in two
steps before re-circulating it to the top of the absorber column. First, it is pre-cooled in the
solvent heat-exchanger (3) and then in the solvent cooler (4) where water is used as cooling
medium. The wet CO2-rich gas exiting the stripper will pass through a condenser (7) for
moisture removal before sending it to the CO2 compression train (8) (Onarheim et al., 2017).

4.2. Equilibrium and rate-based models

The historical approach to simulate separation columns has for long been based on the
equilibrium-stage concept, in which the vapor and liquid phases are assumed to be at a state
of thermodynamic and chemical equilibrium at each theoretical stage. Although, streams
leaving a real tray or section of a packed column are not in equilibrium. The actual separation
achieved in real columns depends on the rates of mass transfer from the vapor to the liquid
phase. The magnitude of these rates in turn depends on the extent to which the vapor and
liquid streams are not in equilibrium with each other. Furthermore, neither reaction kinetics
or film diffusion are considered within this approach, causing inconsistency in simulation
results for chemical absorption and desorption processes. The rate-based model should
therefore be used in reactive systems (Taylor et al., 2003) as it considers the material transfer
and the kinetics of the chemical reactions present along the column packing while the
equilibrium is only assumed to occur at the gas-liquid interface (Neveux et al., 2017).

As flow scheme modifications shift the kinetics and thermodynamics in columns, the use of
equilibrium stages in such studies is particularly suboptimal. For example, the addition of
an intercooler in the absorber will in theory enhance the solubility of CO2, thus maximizing
the driving force for absorption, but at the cost of a lower reaction rate which in turn
minimizes the transfer rate. Such a modification entails withdrawal of the solvent passing at
an intermediate point in the column and cooling it down before it is sent back into the
absorber. Now, when adding an intercooler to an absorption process using 30 wt% MEA as
solvent, an equilibrium-stage model predicts around 12% reduction of the reboiler heat duty
(Ahn et al., 2013), while both pilot plant experiments (Knudsen et al., 2011) and rate-based
simulations (Le Moullec & Kanniche, 2011) with the same process modification demonstrate
a much lower reduction (1 to 2%). Consequently, rate-based models should be regarded as
mandatory for accurate predictions in process design studies (Neveux et al., 2017).

The simulation results of Zhang and Chen (2013) from the equilibrium model of a stripper
indicate a clear tendency toward underestimation of the reboiler heat duty when compared
to experimental data and simulations results from a rate-based approach. Moreover, the
same equilibrium-stage model in their study overpredicted the CO2 removal percentage and
provided poor predictions of the temperature and CO2 concentration profiles in the absorber.
A way to produce similar results as those from a rate-based approach is by tuning the
Murphree efficiency values with a trial and error technique until the simulation results show
a closer agreement to experimental data (Marik Singh et al., 2017).

Considering that the main energy consumer is the reboiler, the use of either a rate-based
method or an equilibrium setup for the desorption column adjusted to match results of a
rate-based one should be emphasized when economical aspects will be part of the
optimization analysis in a steady-state design. There are cases though, in where the
equilibrium approach is preferred, such as when other data systems (e.g. electricity,
weather and carbon markets) are to be incorporated in dynamic data-driven models (Abdul

11
Manaf et al., 2016) and upon evaluation of the dynamic performance of different scenarios
expected to occur during operation (Aspen Technology, Inc., 2014) or in emergency
situations (Øi, 2010).

4.3. Modelling tools

The evaluation of process performance of various plant configurations by building and
operating infrastructure can be costly. Using simulation software is a much less costly
approach to assess process plant performance, in regard to time and capital (Bui et al., 2014).
Therefore, computer-aided process simulation is nowadays recognized as an essential tool
in the chemical process industries. It plays a key role in the evaluation of technical and
feasibility studies, investigation of feed flexibility, flow-sheet optimization and interpretation of
pilot plant data. A lot of mathematical models are usually involved when performing process
simulation calculations. They can be thermodynamic, non-equilibrium, physical property or
fluid mechanics models (Solbraa, 2002). Commercially available software that have been
used to model amine-based Post-CC processes at steady-state conditions are: Aspen
HYSIS, Aspen Plus, Pro/II and ProMax (Øi, 2010). Moreover, the absorption part of the
capture process has been simulated dynamically, i.e. the system changes over time, using
Matlab (Conference on Mathematical Modelling & Troch, 2009) and gPROMS (Kvamsdal et
al., 2009).

Each of these process modelling tools have their own strengths, weaknesses and special
features. For example, both Aspen HYSIS and Aspen Plus use a sequential modular flow-
sheeting framework by default. This extremely effective and commonly used approach works
with an algorithm that allows the simulation to go in one direction only: downstream. In the
case of Aspen Plus, where the modules can only be executed one at a time, the algorithm
will consider the order in which they should be computed as it analyzes the flowsheet by
following the flow of material, energy or information. The output streams and other
performance information of each block will only be calculated when all the input streams
details and necessary model parameters have been specified. In contrast, HYSIS has an
inbuilt ability for information to move upstream as it allows for the user to specify a desired
outlet stream temperature into the form for the outlet stream rather than into the form of the
heat exchanger. Entering user information into the stream would be ignored in Aspen Plus.

The rate-based model makes use of correlations to predict the actual performance of small
packing sections and does not involve height equivalent of a theoretical plate. Instead, the
attainment of equilibrium is assumed to occur at the gas liquid interface only (Solbraa, 2002).
The RadFrac model in Aspen Plus uses rigorous inside-out algorithms by default to solve a
system of nonlinear algebraic equations consisting of phase equilibrium, energy balance,
material balance and summation equations for each theoretical stage. These algorithms
consist of two nested iteration loops. An approximate set of thermodynamic parameters are
used in the inner loop to solve the system of equations. Exact thermodynamic models are
employed in the outer loop to update the parameters of empirical equations used in the inner
loop (Haydary, 2018).

HYSIS will automatically compute information for some blocks by default without having to
run the full simulation as soon as enough data are available. On the other hand, Aspen Plus
has the advantage of automatically selecting tear streams based on flow-sheet structure. This

12
is an attractive feature to some users as it relieves you from the habit of always having to
think about adding a Recycle block whenever the solver needs an abstract point to generate
new guesses at each iteration, which is the case for HYSIS and ProMax. Pro/II is similar to
Aspen Plus in this aspect, as well as when it comes to their general form-based model
construction and sequential modular flowsheets. Although some of their libraries of chemicals
and physical property models are different from each other, both tools are suitable for
integration with other software for the purposes of process control, optimization and dynamic
modelling.

ProMax on the other hand, besides being similar to HYSIS in functionality, is a software
specifically created for CO2 removal and gas sweeting applications and therefore includes
optimized proprietary convergence algorithms for absorber and stripper models. Another
practical feature is that it exists as an add-on module to Microsoft Visio. Besides being a
chemical process tool, gProms is also an advanced ordinary differential equation integrator
that operates in an equation-oriented environment. The model created in the graphical user
interphase of this software can be either steady-state or dynamic and will ultimately consist
of one large system of equations. It is also possible to build the models from scratch using
custom equations. Aspen Plus has a significant advantage over gProms for steady-state
simulations when operating in equation-oriented mode since the sequential modular mode
can be used to initialize and solve the flowsheet with less degree of difficulty (Adams, 2018).

4.4. Thermodynamics and kinetics

The electrolyte-non-random-two-liquid-ba ed he d a ic ac age ENRTL-RK a

selected to describe the thermodynamics in the MEA process since it considers the strong
non-ideality of the liquid electrolyte solution. The coupling with the Redlich-Kwong (RK)
equation of state in this model enables the computation of the vapor properties (Madeddu
et al., 2019). The chemistry of the MEA- CO2 - H2O system can be represented by a set of
equilibrium and reversible kinetic reactions. The reactions shown in Table 3 are widely used
for modelling purposes, including by (Freguia & Rochelle, 2003), and will therefore also be
used in this study.

Table 3 Chemical reactions considered in the MEA process.

No. Type Reaction

1 Equilibrium 2H2 O ⇌ H O+ OH
2 Equilibrium MEAH + H2 O ⇌ MEA H O+
3 Equilibrium HCO H2 O ⇌ H O+ CO 2
4 Kinetic MEA CO2 H2 O MEACOO H O+
5 Kinetic MEACOO H O+ MEA CO2 H2 O
6 Kinetic CO2 OH HCO
7 Kinetic HCO CO2 OH

Reaction 1 is the water dissociation, reaction 2 the protonation of MEA and reaction 3 the
bicarbonate dissociation. Reactions 4 and 5 are the forward and reverse reactions for
carbamate formation, and reaction 6 and 7 are the forward and reverse reactions for
bicarbonate formation. The necessary parameters for the calculation of the chemical
equilibrium constants for reactions 1 7, using the standard Gibbs free energy change, were

13
obtained from the databank of Aspen Plus. Power law expressions (Eq. 1) were used to
calculate the reaction rates, r, of reactions 4 7.

𝑟 𝑘𝑇 𝑛 𝑒𝑥𝑝 ∏𝑖=1 𝜒𝑖 𝛾𝑖 𝑎
(1)
𝑇

In order for the simulation to proceed one needs to provide the software with the activity
basis rate constants, k and E, of each forward and reverse reaction considered in the
thermodynamic model. The pre-exponential factor k and the activation energy E are
tabulated in Appendix A Absorption validation: parameters and results. T is the absolute
temperature, R is the universal gas constant, N is the number of components in the reaction,
𝜒𝑖 is the mole fraction of component i, 𝛾𝑖 is the activity coefficient of component i and 𝑎𝑖 is
the stochiometric coefficient of component i (Aspen Technology, Inc., 2014). In this study,
the factor n is zero.

4.5. Material transport

Three main theories have been suggested for the quantification of the material transport
across the gas-liquid interface in the rate-based model: the two-film theory by Lewis and
Whitman, the penetration theory by Higbie and the surface renewal model by Danckwerts.
The most used theory for CO2 absorption is the two-film theory since a significant number of
correlations for the parameters evaluation are easily accessible in literature. In this theory, all
the resistance to material and energy transfer is assumed to take place in two thin films close
to the gas-liquid interface. As showed in Figure 5, the spatial domain represented by the rate-
based segment can be divided in four parts: gas bulk, gas film, liquid film and liquid bulk. The
absorption process involves the mass transfer of CO 2 from the gas bulk to the interface,
through the gas film, followed by absorption into the liquid where it will react with the solvent.
Reactive columns in the RadFrac model are considered as a finite number of continuous
stirred-tank reactors (CSTRs) connected in series. Five different flow models are available in
the RadFrac model for evaluation of the bulk conditions in each segment: mixed flow (bulk
properties=outlet conditions which approximates an ideal CSTR), countercurrent flow (bulk
properties=average between the inlet and the outlet conditions), and VPlug/VPlugP/LPlug
flow (average conditions for one phase and outlet conditions for the other one) (Massimiliano
et al., 2019).

Figure 5 Heat (q), and mass (N) transfer fluxes on a differential element of packing, dz. In this
schematic representation of the two-film theory, the mole fractions in the gas and liquid phases are x
and y, respectively. T denotes the temperatures (Neveux et al., 2017).

14
 5. Methodology
In the validation of the absorber, the focus was on the liquid temperature profile, CO2 removal
efficiency and the rich loading (i.e. CO2 loading in the rich solvent at the absorber outlet).
Next, the amount of amine solvent and the diameter and packing height of the absorber
(5.3.1. Absorber) and stripper (5.3.2. Stripper) are determined in order to satisfy the intended
performance of the capture plant according to the design criteria as presented in the next
section (Table 8). The sizing of the absorber and stripper also took into consideration the
characteristic composition, flowrate, temperature and pressure of a flue gas stream
originating from the recovery boiler of a stand-alone softwood market pulp mill. The existing
low-pressure (LP) steam at 4.5 bar may be extracted from the steam cycle to provide the
required steam in the stripper reboiler (Normann et al., 2019). A schematic of the workflow is
shown in Figure 6.
Methodology

Setting up standard
Validation
process

- absorber (lab and pilot scale) - connection between absorber and stripper
- stripper (two different reflux configurations) - figuring out tear streams
- adjustments in Property environment - reconstruction of stripper reflux
- adjustments in Simulation environment

Stripper
Sizing
modification

- discretization (no. stages) - adding a compressor before condenser

- solvent flow rate - varied operation pressure (1, 1.3 and 1.8 bar)
- 90% capture rate - heat curves from condenser evaluated for different
- pressure drop (△P) compressor discharge pressures (4-10 bar)
28
9/4/20 28

Figure 6 Workflow of the modelling and simulation steps needed to reach the thesis objectives.

The CO2 loading gives an indication of the absorption capacity of the solvent (Madeddu et
al., 2018) and is defined as the ratio between the CO2 and MEA apparent molar fractions
according to Eq. (2). This performance indicator was calculated for every simulation in Aspen
Plus by creating the Property Set, ML-LOAD, in the Properties Environment and then adding
it to the report page through two steps in the Simulation Environment: first by selecting the
corresponding ID in the Property Sets window from the Setup | Report options | Stream tab
and then selecting it again in the Analysis | Report | Properties tab of each RadFrac block.

𝑥 𝑥 2+ 𝑥 +𝑥 2 +𝑥
𝐿𝑜𝑎𝑑𝑖𝑛𝑔 2
(2)
𝑥 𝑥 +𝑥 ++ 𝑥

15
 5.1. Validation
The model used in this study was validated by following the main steps of a systematic
procedure developed by Madeddu et al. (2019). One of the most highlighted steps in their
validation process is the evaluation of number of segments, i.e. discretization points in axial
direction of the packing section. The main reason for such analysis being that a sufficiently
high number of segments is needed from a mathematical point of view to obtain a correct
numerical solution of the process model. The appropriate number of segments is identified
by varying this parameter until the difference between two consecutive sets of profiles, e.g.
temperature profile, becomes negligible. Another positive effect of the special consideration
to the discretization of the axial domain is the fact that the temperature bulge also gets
correctly described. For example, it takes 30 segments for the temperature bulge to start
showing in the computed temperature profile of the lab-scale absorber.

The appropriate amount of discretization points depends from case to case. Usually, the
larger the column the greater the number of segments is needed to correctly describe the
absorption/desorption process with the RadFrac model in Aspen Plus. The adequate number
of segments was evaluated for all absorption/desorption cases studied in this work.

5.1.1. Absorber
The absorption model was validated by comparing model predictions and literature data. The
literature data are taken from two pilot-plant facilities that differ in operating conditions, size
and packing type (random vs. structured): a lab-scale and a large-scale absorption plant.
Experimental data from only one run from each plant was selected for the model validation:
T22 (Tontiwachwuthikul et al., 1992) from the lab-scale and 1-A2 (Razi et al., 2013) from the
pilot-scale plant. After setting up the Properties and Simulation environment in Aspen Plus
some other values needed to be provided to enable the simulation run: the kinetics
parameters for the reversible reactions (4 7 in Table 3) a d he He a c a
coefficients for CO2-H2O and CO2-MEA. The He a c a i a i a e
property parameter to consider in the absorption process as it represents the solubility of CO2
in a given solvent (Bui et al., 2018).

Neither of these input parameters were provided by the Aspen Plus databank by default upon
choosing the ENRTL-RK thermodynamic package for the MEA-CO2-H2O system. These
values therefore had to be found in literature. Six different sets of kinetic parameters were
tested and only two of these resulted in adequate fitting with experimental data when
c bi ed i h a i ab e e f He a c a c efficie . H e e , he i lation
results were still not matching those from the experimental data (see case D0 for the pilot-
scale absorber in Table 4) which is why different combinations of electrolyte pair parameters
(GMENCC) were evaluated together with the most promising sets of kinetic parameters.
These parameters relate to molecule-ion and ion-ion interactions in electrolyte solutions and
are also required for the calculation of major thermodynamic properties when using the
ENRTL-RK method. To keep better track of these changes, the combined kinetic and
GMENCC parameters were referred to as G0 G4 and D0 D4. See Appendix A Absorption
validation: parameters and results for more details about the results presented and to better
understand the notations used in this section.

16
Table 4 Comparison of experimental data with simulation results of performance indicators of some
of the absorption cases tested in the validation procedure.

Performance D0 D1 D3 Experimental
Pilot- CO2 removal (%) 66.2 63.6 89.5 90
scale Rich loading 0.436 0.427 0.510 -
Lab- CO2 removal (%) 98.2 94.8 98.1 100
scale Rich loading 0.461 0.450 0.461 0.443

While the same set of chemical reactions is used to model both the absorber and stripper in
most cases, some studies have proposed different reaction kinetics in regard to reaction 5
(Table 3), which dictates the rate of CO2 desorption. In that case, the reaction rate constants
are calculated through linear regression (Aspen Technology, Inc., 2014). Depending on the
chosen temperature range for the absorber and stripper respectively, the output values will
differ. This and other measures to adjust the kinetic parameters depending on the operation
conditions of the process could explain why the values found in literature differed so much
from each other.

According to (Kvamsdal & Rochelle, 2008), the liquid-to-gas mass ratio (L/G) is one of the
main factors that influence the location and magnitude of a bulge in a temperature profile.
The bulge is usually located at the column top for values of L/G lower than 5, while for values
of L/G above 6 it is located at the bottom part. The needed L/G value to reach the required
separation performance is usually low in columns filled with structured packing, mainly
because of their higher surface area compared to random packed columns. Consequently,
bottom bulges are typical in random packed columns (Errico et al., 2016). Figure 7 shows the
temperature profiles of the absorption columns calculated with the different model cases
presented in Table 4. The model consistently predicts a temperature bulge in the top section
of the column for the pilot plant. This is expected because of a 4.1 value for the L/G in this
example. On the other hand, a bottom temperature bulge shows in the temperature profile of
the lab-scale plant, which is no surprise considering its relatively high L/G value of 7.43.

Lab-scale plant Pilot plant

60 80
exp exp
55 D0 D0
D1 D1
75
Liquid temperature [ o C]

Liquid temperature [ o C]

50 D3 D3

45 70
40
65
35

30 60
25
55
20

0 1 2 3 4 5 6 7 0 5 10 15
Distance from bottom [m] Distance from bottom [m]

Figure 7 Temperature profiles of the absorber columns in the lab-scale and pilot plant, calculated
with 50 stages with different combinations of electrolyte pair parameters. Experimental data (exp)
relates to different points throughout each packing section.

17
The results from the D3 alternative excelled in similarity with the literature values in regard to
the absorber performance: with no more than 2% error in the estimation of the CO2 removal
efficiency and 4% error in the estimation of the rich loading (Table 4). The good agreement
between the model and experimental results (independently of the operating conditions, the
column dimensions and the temperature bulge location) indicates that alternative D3 can be
used with confidence when conducting a new absorber design.

5.1.2. Stripper
The desorption model was validated by comparing model predictions and literature data. The
literature data for this validation step are taken from two pilot-plant facilities that differ in
operating conditions, size and packing type (Sulzer Mellapak 250Y vs. Flexipac 1Y). These
were, the pilot plant facility of SINTEF (a Norwegian independent research organization) and
the pilot plant facility of University of Texas at Austin (UTA). Experimental data from only one
run from each plant was selected for the model validation: no. 1 (Tobiesen et al., 2008) from
the SINTEF plant and no. 47 (Dugas, 2006) from the UTA plant. The stripping section in both
of the plants is equipped with structured packing, a reboiler and a partial condenser. The
main difference between the two cases is in how the water stream exiting the condenser is
distributed to other parts of the stripper. In the SINTEF case, the condensed water is mixed
with the lean solvent exiting the stripper bottom before entering the reboiler. The UTA case,
on the other hand, resembles the desorber configuration of a standard MEA process since
the condensed water is sent back to the stripper top as reflux. Also, the run that was selected
from the UTA plant operated under vacuum condition (0.69 bar) and that from the SINTEF
plant at higher pressure (1.96 bar).

Table 5 Error in the estimation of the reboiler duty with number of segments for the UTA plant.

Number of segments
Model Performance Experimental
10 80 90
Reboiler duty (kWth) 292 251 250 205
I
Error (%) 43 22 22 -
Reboiler duty (kWth) 249 217 217 205
II
Error (%) 21 6 6 -

To prove validity of the MEA model in the case different experimental data sets are available,
Madeddu et al. (2019) chose the degrees of freedom differently for each plant. The same
concept was also applied in this study. The UTA stripper was the first one validated, in which
the condenser temperature and the CO 2 gas molar flow rate (hence, also the lean loading)
were fixed by assigning two design specifications inside the RadFrac block. Using the same
model correlations as the ones used during the absorber validation (Model I) gave a 22%
error upon estimation of the reboiler duty, see Table 5. Instead of tweaking parameters in the
Properties environment, as was the case in the absorber validation, it was decided to adjust
other parameters directly in the Simulation environment through the sheets in Rate-Based
Modeling | Rate-based Setup. Model II resulted from this adjustment of parameters.

The parameters that differ between the two models are: reaction condition factor, interfacial
area method, interfacial area factor, heat transfer factor, liquid mass transfer coefficient factor
and vapor mass transfer coefficient factor. See Appendix B Stripper validation: parameters

18
 and results for detailed parameter specification of the models. The temperature in the stripper
column in both examples was overestimated, although quite consistently, with 1 2ºC for the
UTA plant and 2 3ºC for the SINTEF plant (Figure 8). The overestimation of the temperatures
still remains when using model II.

UTA plant SINTEF plant

94 124
exp exp
93 Model-I Model-I
Liquid temperature [ o C]

Liquid temperature [ o C]
Model-II Model-II
122
92

91
120
90

89
118
88

87 116
0 1 2 3 4 5 6 0 1 2 3 4
Distance from bottom [m] Distance from bottom [m]

Figure 8 Temperature profiles of the stripper columns for the UTA and SINTEF plant calculated with
90 stages. The simulation results include the reboiler temperature (bottom). Experimental data (exp)
relates to different points throughout each packing section.

Differently from the validation of the UTA plant, the condenser temperature and the reboiler
duty were fixed in the simulation of the SINTEF plant. Consequently, the performance value
evaluated in this example was the lean loading, i.e. CO2 loading in the lean solvent stream
exiting from the bottom stage. Similar to the overprediction of the reboiler duty, the loading
was also overpredicted in the stripper validation of the SINTEF example. Although, it was not
possible to reduce the error in the estimation of this performance value, by changing from
Model I to II, to the same extent as with the UTA stripper. The suggested model overpredicted
the lean loading from the SINTEF stripper with 14% (see Appendix B Stripper validation:
parameters and results).

5.2. RadFrac model

The Aspen Plus RadFrac distillation column, based on the two-film theory with a rate-based
approach, was used as a basis for simulating the absorber and stripper units in this work.
This method is by far one of the most popular choice among researchers for simulation of
MEA-based CO2 capture processes (Biermann et al., 2018; Errico et al., 2016; Ferrara et
al., 2017; Fosbøl et al., 2014; Freguia & Rochelle, 2003, p. 2; Garcia et al., 2017;
Garðarsdóttir et al., 2018; Hasan et al., 2012; Hwang et al., 2019; Le Moullec & Kanniche,
2011; Li et al., 2016; Luo & Wang, 2017; Madeddu et al., 2018; Marik Singh et al., 2017;
Onarheim et al., 2017; Razi et al., 2013; Zhang & Chen, 2013). Transport property models
are necessary for the determination of various physical and thermal properties such as
viscosity, density, surface tension, thermal conductivity and diffusivity. These models are
based on empirical correlations involving flow parameters and fluid properties such as heat
and mass transfer coefficients, interfacial area, pressure drop and liquid holdup (Hasan et
al., 2012). Table 6 summarizes the correlation parameters used to model the absorption
and desorption columns.

19
 Table 6 Selected correlation parameters for the RadFrac model.

Model and column properties

Packing material Sulzer Mellapak 250Y
Flow model Mixed
Film resistance Discretize film for liquid
Consider film for vapor
Discretization points for liquid film 5
Mass transfer correlation method (Bravo et al., 1985)
Heat transfer correlation method Chilton and Colburn
Liquid holdup correlation method (Bravo et al., 1992)

The mixed flow model was chosen over e.g. countercurrent (which in theory should predict
more accurate results for packing) because of convergence difficulties in this study with the
latter option. (Zhang et al., 2009) evaluated how well the different flow models in RadFrac
predict CO2 loading, CO2 removal% and temperature profiles of a pilot plant absorber. They
concluded that the different flow models have only very minor influence on the overall capture
rate performance, and that the mixed flow model predicted the most reliable predictions in
regard to the temperature profiles. To ease convergence in the absorber block even more,
the number of maximum iterations was increased to 100 and the damping level was set to
Mild on the Convergence| Convergence | Basic sheet. The number of maximum iterations
was also increased to 100 in the stripper block. Although, since the absorber block was more
difficult to converge than the stripper block, a specific convergence algorithm was selected
by choosing Absorber=Yes on the Convergence| Convergence | Advanced sheet. This
alteration of the standard algorithm for simulation of the RadFrac model could not be selected
for the stripper blocks in this work since they had an inbuilt reboiler.

5.3. Dimensioning
The flue gas characteristics (Table 7) considered in this study originate from simulation
calculations of a reference pulp mill model developed by RISE Bioeconomy in WinGEMS (a
simulation tool that is primarily used for mass and energy balances in the pulp and paper
industry). This mill has a design capacity of 700 000 air dry tonne (ADt) of pulp per year. The
recovery boiler, lime kiln and power boiler are responsible for the major CO2 emissions from
this plant. However, only the flue gas from the recovery boiler is considered for the carbon
capture in this study.

Table 7 Flow rate, temperature and composition of the wet flue gas from the recovery boiler in the
reference pulp mill.

Flue gas rate (tonne/h) 722

Temperature (ºC) 175
CO2 (wt%) 20.8
N2 (wt%) 63
O2 (wt%) 2.4
H2O (wt%) 13.8

The generated steam for the pulping process originates from the combustion of the remaining
bark (after the refining process) in the power boiler (Normann et al., 2019). It is also assumed

20
that SO2 and other impurities in the treated flue gas stream have already been removed by
optimization of the process conditions during combustion and through other flue gas cleaning
processes than those included in the CO2 capture plant (Skagestad et al., 2018).

To avoid underestimation of the column dimensions and required duty, the packing section
was modelled with 100 stages in the absorber and with 70 stages in the stripper. Since the
stripper is equipped with a reboiler and a condenser, the packed section in this column
constitutes of 68 discretization points. A column flooding limit had to bet set to determine the
absorber and stripper diameters with reference to a specific point in each column. This value
was fixed at 80% and the base stage was chosen to represent the more stressed (limiting)
part of the column. The base stage was identified as the column stage that had the highest
vapor flow rate by looking at the column hydraulic results after each simulation run.

An approximate value for the limiting stage could easily be found in the Column Internals |
INT-1 | Column Hydraulic Results report of each RadFrac block. Although, the value was only
true with ±1 precision upon comparison with the more precise number depicted in the column
hydraulic results. The base stage and flooding limit were input values in the Design tab. The
design-mode option in this tab had to be activated to enable calculation of the optimal column
diameter. The following steps in the design analysis consider each absorption/desorption
column as standalone, hence, the simulation results obtained from section 5.3.1. Absorber
are later used as initial guesses in section 5.3.2. Stripper. Following this coupled system
analysis (Madeddu et al., 2018), it was decided to introduce another constraint: that the
packing height/diameter ratio of both the absorber and stripper must be higher or equal to 1.
The pressure drop was considered by using a correlation from literature: 20.83 mmH2O per
meter of packing (Luo & Wang, 2017).

5.3.1. Absorber
The motivation behind testing different solvent flow rates was to avoid the presence of
isothermal zones in the column (plateau shape). Thereby, ensuring the use of the entire
packing. The height of each example (Figure 9) corresponds to the minimum required for
90% capture rate and was decided through design specs followed with the sensitivity analysis
block in Aspen plus. The design specs block was used to fix the constraint that not more or
less than 10 wt% of CO2 leaves with the exiting gas stream, and through that obtain an
approximate value of the minimum lean solvent flow rate (Lmin). Lmin is a term defined in the
infinite method as the required solvent rate through a theoretical column of infinite height
(taken here as 100 m) which was modelled with 100 segments in this study.

The sensitivity analysis comprised the computation of the mass flow rate of the CO 2 in the
exit gas stream of the absorber together with the corresponding lean solvent flow rate. Since
the results from the sensitivity analysis were more accurate than those from the design
specs, the former was used to select a suitable Lmin (2190 tonne/hr). The minimum numbers
of absorbers were also determined through this first step of the column dimensioning
process. According to different works in the literature (Madeddu et al., 2019), the diameter
of absorbers with this type of packing should not be higher than 12 m. Since it was found
that the absorber diameter to process the recovery boiler flue gas is ca 10.7 m (Appendix
C Absorber dimensioning for more details), the design analysis continued with the
conclusion that only one absorber unit is needed to meet the capture criteria in this study.

21
 Varying flow rates of lean amine
90

Lmin H=100 m
85
1.05Lmin H=25.4 m
1.1Lmin H=14.2 m
80
1.1Lmin H=14.0 m dP=25 mbar
1.15Lmin H=10.2 m
75
Liquid temperature [ o C]

70

65

60

55

50

45

40
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative distance from bottom [m]

Figure 9 Varying flow rates of the MEA solvent with 0.25 lean loading that enters the top of the
absorber column. L/G ratios increased from 3.4 to 3.9 when increasing Lmin to 1.15Lmin.

As shown in Figure 9, the higher the solvent flow rate the shorter column height is needed to
remove a fixed amount of carbon dioxide. The dashed line represents the simulation results
from when the absorber was modelled with pressure drop. The temperature profile with
1.15Lmin is also plotted there. However, the corresponding column height (10.2 m) with these
conditions is lower than the column diameter, thus violating one of the main design criteria,
i.e. H/D ratio should not be less than 1. It was therefore not suitable to choose these
dimensions. It was decided to have an absorber equipped with a 14 m packed structured
column with 11 m in diameter, thus a lean solvent flow rate of 2410 tonne/hr (=1.1Lmin), since
the temperature profile achieved with these process parameters does not indicate any
plateau shape and efficient use of the entire packing will be ensured (theoretically).

The higher the column height the higher the capital expenses of the capture plant, which is
why one aims to choose the column height to be as short as possible. A pressure drop was
added to the absorber model and its resulting temperature profile is represented with a
dashed line in the figure above. The temperature point maximum is located on the absorber
top, which is typical in absorbers with structured packings with L/G values lower than 5. The
L/G ratio achieved with this absorber size is 3.7 which means that the temperature profile
produced with this Aspen Plus aligns with literature inferences regarding this aspect.

5.3.2. Stripper
Appropriate dimensions of the stripper column were obtained by having the starting values
of the rich solvent with regard to flow rate, temperature, pressure and composition be almost
the same as the output values generated in 5.3.1. Absorber. The pressure and temperature
of the rich solvent were significantly altered (compared absorber outlet values) since the
solvent was passed through a heat exchanger to reach 99ºC. All the smooth lines in Figure

22
 10 represent simulations run with the condenser temperature fixed at 40ºC. The process
design criteria (Table 8) clearly says that the discharge temperature from this unit has to be
20ºC, which is why another simulation was run with the correct outlet conditions. The results
from this run are depicted with a dashed line and also takes into account a pressure drop.

The reboiler duty was not significantly affected by the changes in column heights (see
Appendix D Stripper dimensioning), which is why it was chosen to pursue with the shortest
of the column heights analyzed in this section, i.e. 9 m. The corresponding column diameter
of this column height is 6.33 m. Looking at the temperature profile trend in the picture below
indicates that isothermal zones are avoided with shorter the stripper column heights, hence,
it could have been possible to choose a column height as short as 7 m. However, since it
was not known how much the column diameter would need to be altered (reduced or
increased) due to the following stripper configurations, the shorter column heights were
discarded for the stripper column in this study. Even if it would have meant a lower capital
cost for the capture plant.

Varying packed column heights of stripper

120 H=9 m
o
H=9 m, dP=15 mbar, cond=20 C
H=11 m
H=13 m
H=15 m
115 H=17 m
Liquid temperature [ o C]

110

105

100

95
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relative distance from bottom [m]

Figure 10 Varying packing heights of the stripper column with fixed solvent flow rate.

5.4. Standard MEA process

Some of the design parameters (flue gas and solvent temperature at absorber inlet, CO2 gas
temperature at reflux outlet, MEA concentration at washer outlet) presented in Table 8 were
selected on basis of the design specifications for capture plants in the CO2stCap project
(Biermann et al., 2018; Garðarsdóttir et al., 2018). The absorber, washer and stripper
operating pressure, pressure drop, packing height and diameter were defined in the previous
chapter.

23
 Table 8 Main design parameters for the standard MEA process.

Flue gas CO2 capture rate (wt%) 90

Absorber top stage pressure (bar) 1
Absorber pressure drop (bar) 0.025
Absorber packing height (m) 14
Absorber packing diameter (m) 11
Washer top stage pressure (bar) 1
Washer pressure drop (bar) 0.004
Washer packing height (m) 5.4
Washer packing diameter (m) 11
Stripper top stage pressure (bar) 1.8
Stripper pressure drop (bar) 0.015
Stripper packing height (m) 9
Stripper packing diameter (m) 6.6
MEA concentration (wt%) 30
Flue gas temperature at absorber inlet (ºC) 40
Lean solvent temperature at absorber inlet (ºC) 40
Rich solvent temperature after Heat-X (ºC) 107
Rich solvent temperature at stripper inlet (ºC) 99
Rich solvent feed stage in stripper 2
Reboiler temperature 121
CO2 gas temperature at condenser outlet (ºC) 20
Lean CO2 loading (mol CO2/mol MEA) 0.25
MEA concentration at washer outlet (ppm) <1

G3 64

MM1 69 25
WASH
W0

40

MAKEUP

COOL
61
40
40
V2
LS4
LEAN
LS5

MIX C0 20
LS3 61

HEAT2
107 99
V1
G2 71
40 S TRIPP ER
RS 3 RS 4

LEANIN

40
ABS 121

G1
Q
LS 2
48
175
48
HEAT1
RS 2
RBGAS DCC
RS 1
121

P UMP 1 LS1

W1 40 P UMP 3

Figure 11 Aspen Plus flowsheet of the base case capture process with MEA.
Tempera ture (C)

De activa te

24
As is shown in Figure 11, the absorber was modelled with two RadFrac blocks with packed
columns where the carbon capture is designated to happen mainly in ABS and the MEA slip
from that unit is captured in WASH. The stripper is modelled with one partial condenser (top
stage), a packed column and one kettle reboiler (bottom stage) inbuilt in the RadFrac block.

5.5. Modified MEA process

The modified MEA process comprises a stripper overhead compression configuration (Figure
12). Different compressor discharge pressures (4 10 bar) were evaluated at different stripper
operating pressures (1 1.8 bar). The maximum operating pressure of the stripper with this
model is around 1.8 bar. This pressure guarantees that the reboiler temperature will not
exceed that of the MEA solvent degradation temperature, i.e. 122ºC. Heat integration with
this process modification is possible when the excess heat available in Q1 (through
condensation of the CO2/H2O vapor) is redirected to the stripper reboiler. It was attempted to
transfer the excess heat available in this manner in Aspen Plus, however, it was not
successful. Therefore, the decreased reboiler duty due to heat integration was accounted for
in the final calculations by manually deducting the amount of heat released (when Q1
discharge temperature=reboiler temperature) from the reboiler duty.

60
COMP R1

L3
G3 64

Q1 60
238

L2 FLAS H1
MM1 25 L1
69
WAS H
C0 100
W0

60
V6
40
RF1

MAKEUP

COOL
61
40
40
V2
LS 4
LEAN
LS 5 RFG 60

MIX
LS3 61

HEAT2
107 99
V1
40 G2 71
RS3 RS4 S TRIPPER

LEANIN

40
ABS 121

G1
Q
LS2
48
175
48
HEAT1
RS2
RBGAS DCC
RS 1
121

P UMP1 LS 1

W1 40 P UMP3

Figure 12 Aspen Plus flowsheet of the modified case capture process with MEA. The design
Te mpe rat u re ( C)

De act ivat e

alteration from the standard case is the addition of a compressor prior to the heat exchanger in the
condenser arrangement of the stripper. Case 1 is illustrated here with a 6 bar discharge pressure
compressor.

Three cases were studied as listed in Table 9. The temperature of the CO2 gas leaving the
condenser remained the same, 60ºC, for the three cases. The solvent temperature at the
stripper inlet needed to be changed manually, considering the minimum temperature
approach (10ºC) and the reboiler temperature. The pressure drop, height and diameter of the
stripper packing had to be altered when decreasing the stripper pressure in order to meet the
90 wt% capture rate while minimizing hydraulic issues in the column, i.e. flooding and
weeping. Other important design parameters were kept the same as in Table 8.

25
 Four versions of the modified stripper were evaluated even further after finding valuable
correlations from the case 1 3 results. The condenser discharge temperature was fixed at
20ºC in these configurations. Also, an additional heat exchanger was added between one of
the valves (V6) and the stripper to increase the rich solvent temperature from 20ºC to ca
98ºC. Doing so ensures a more efficient desorption process. One more heat exchanger with
99ºC discharge temperature was added between V1 and the stripper for the last two
flowsheet configurations with 1.3 bar stripper operating pressure. Q1 is the heat source
(theoretical assumption) for the additional heat exchangers.

Table 9 Major changes of design parameters in the capture plant due to different stripper pressures
in the configuration cases 1-3.

Case 1 Case 2 Case 3

Stripper top stage pressure (bar) 1.8 1.3 1.0
Stripper pressure drop (bar) 0.015 0.015 0.02
Stripper packing height (m) 9 9 10.5
Stripper packing diameter (m) 7.4 8.5 10
Rich solvent temperature after Heat-X (ºC) 107 100 93
Rich solvent temperature at stripper inlet (ºC) 99 92 85
Reboiler temperature (ºC) 121 112 104

In order to add a compressor to the reflux loop, the inbuilt partial condenser in the RadFrac
model had to be reconstructed. This was done by connecting the outlet gas stream from the
stripper column to a cooler (COND in Figure 13), with the resulting stream entering a flash
vessel (which splits the stream into a pure vapor and liquid stream respectively) and then
connecting the liquid stream from the flash outlet to a pump that leads the condensate back
to the top stage of the stripper column. The reflux ratio is adjusted by tuning the vapor
fraction parameter in the flash block. The external partial condenser was validated by
comparing the stripper reboiler duty, condenser duty, flow rate, temperature and
composition of the exiting stripper streams with that of a stripper model with an inbuilt partial
condenser. Results from this validation step are presented in Appendix E Validation of
external partial condenser.

CO2UT 20

COND 20

C1 FLAS H

C0A 105

20 C2 20

C3
99
P UMP 2
99 DUP L
1 121

RS 4 DUPL

S TRIP P ER LS 1

Figure 13 External partial condenser for the stripper. The inbuilt condenser of the RadFrac block
2 99
was replaced with an external reflux loop equipped with a heat exchanger and a flash block.

26
CO2 compression is needed to comply with pipeline specifications for CO2 transport, around
95% pure CO2 (Hasan et al., 2012), and stands for the highest exergy losses during operation
of a Post-CC plant, after the absorption/desorption stages. According to literature (Ferrara et
al., 2017; Kuramochi et al., 2012), compressors used for the liquification of CO2 are usually
modelled with 80 85% isentropic efficiency and 90 99% mechanical efficiency. The
isentropic efficiency and mechanical efficiency of the compressors in this study were set at
80% and 97% respectively. Larger efficiency values result in less exergy losses.

27
 6. Results and discussion
The heat/power production and consumption values reported throughout this chapter are in
alignment with 90% carbon capture rate of the recovery boiler flue gas from a pulp mill with
a yearly pulp production capacity of 700 000 ADt. This corresponds to 135 tCO2 per hour and
1.06 MtCO2 per year, assuming 7840 equivalent full-load hours per year.

6.1. Standard MEA process

Since both condenser and reboiler are inbuilt in the RadFrac model with this configuration,
their heat duty was directly derived from the simulation results in the stripper block. The MEA
slip presented in Table 10 Performance of a standard process with different number of
segments in the absorber. corresponds to the MEA concentration (mole basis) in the gas
stream leaving the stripper condenser. Increasing the number of segments from 10 to 70 in
the stripper packing required a reduction of the column diameter from previous 7.4 to 6.8 m
in order to keep the capture rate at 90 wt%. Changing from model I to II also required further
reduction of the column diameter, but mainly to avoid disturbances in the column hydraulics.

Table 10 Performance of a standard process with different number of segments in the absorber.

Number of segments
Model Performance
10 70
Reboiler duty (MWth) 171 155
I Condenser duty (MWth) -54.1 -37.4
MEA slip (ppm) 1.71・10 -5
1.03・10-5
Reboiler duty (MWth) 159 155
II Condenser duty (MWth) -41.2 -37.4
MEA slip (ppm) 1.10・10-5 8.68・10-6

The performance results presented here indicate that there is no significant difference in the
heat duties between model I and II when the packing section of the stripper model is
discretized with 70 discretization points along its axial domain. Any difference between the
models will only affect the desorption process in the packing column and, since that part is
not the main feature of interest in the process modification, it was decided to proceed with
model I and 70 segments from here on. There is a very small difference though, regarding
the traces of MEA in the exiting gas stream. This disagreement is negligible.

6.2. Modified MEA process

Decreasing the stripper pressure may be beneficial for heat integration but doing so results
in a higher steam consumption in regard to the solvent regeneration process. The increased
reboiler duty when reducing the stripper pressure is mainly due to the increased need of
vapor in the desorption process, since the feed temperature is limited by the reboiler
temperature (through the minimum temperature approach in the simplified cross heat-
exchanger), which is lower at lower operating pressures. The lower feed temperature means
that the incoming rich solvent has less amount of free CO2 in the liquid phase. Hence,
increasing the vapor concentration is needed in the stripper column in order to compensate
for the constrained material transfer of CO2 from the liquid to the gaseous phase of the
stripper feed.

28
 Heat curves at different process conditions

400
Stripper
350 1.8 bar 1.3 bar 1.0 bar Compressor
4 bar
Temperature [o C]

300 6 bar
8 bar
10 bar
250

200

150

100

0 20 40 60 80 100 120 140 160 180 200 220

Excess heat available [MW]

Figure 14 Potential thermal energy output from case 1 3 with the vapor compression configuration.

Figure 14 shows that increasing the discharge pressure of the compressor results in more
heat available upon condensation of the leaving gas stream. However, a lower operating
stripper pressure means higher discharge temperatures from the compressor outlet.
Operating at such high temperatures can be detrimental for the equipment if they are not
appropriately designed for such extreme conditions. Equipment that can handle higher
temperatures is usually a lot more expensive. Therefore, regardless of how much more
excess heat available is achieved with larger compressor discharge pressures and lower
stripper operating pressures, it is important to take this detail in consideration when dealing
with stripper overhead compression configurations. More results from Case 1 3 are
presented in Table 11. However, the thermal energy available from the condensation step is
not deducted from the reboiler duties, Qreb, reported in this table.

Table 11 Obtained simulation results for Case 1 3, listed from top to bottom.

P0 Pa Ta Tb Compression MEA Qreb Qreb Reboiler

p(H2O)
(bar) (bar) (ºC) (ºC) work (MW) (ppm) (MW) (GJ/tCO2) T (ºC)
4 189 60 5.69 444 154 4.10 121 0.857
6 238 60 9.01 452 154 4.10 121 0.860
1.8
8 275 60 11.5 451 154 4.11 121 0.860
10 305 60 13.6 451 154 4.11 121 0.860
4 234 60 13.6 853 196 5.23 112 0.879
1.3 6 289 60 19.4 852 196 5.23 112 0.879
8 330 60 23.8 851 196 5.23 112 0.879
10 363 60 27.4 850 196 5.23 112 0.878
4 270 60 28.6 900 274 7.29 104 0.804
1 6 329 60 38.8 899 274 7.29 104 0.804
8 374 60 46.6 898 274 7.29 104 0.803
10 410 60 53.0 897 274 7.29 104 0.803

29
 These results indicate that alteration of the stripper operating pressure has a larger effect
than what the compressor discharge pressure has on the resulting compression work, MEA
slip, reboiler duty, reboiler temperature and the water partial pressure in the compressor gas
inlet (p(H2O)). P0 denotes the stripper operating pressure, Pa is the compressor discharge
pressure, Ta is the compressor discharge temperature and Tb is the condenser discharge
temperature. Since the Ta value in all cases studied in this section is larger than the MEA
degradation temperature, it means that one can assume that all the MEA slip will go to waste,
regardless of a stripper reflux. Significant amounts of MEA slip can be reduced by having a
wash section on the stripper head before the gas gets compressed. The specific reboiler duty
increases from ca. 4.1 to 7.3 GJ/tCO2 when reducing the stripper operating pressure from 1.8
bar to 1.0 bar. Figure 15 showcases an overview of the energy produced and consumed in
the modified MEA process with different operating pressures of the stripper.

CO 2 regeneration (MW th )
350 Compression work (MW e)

in CO 2/H 2O condensation (MW th )

200
Energy [MW]

100

out -100

-200

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

Stripper pressure [bar]

Figure 15 Energy input/output versus operating pressure in terms of thermal energy made available
(CO2/H2O condensation), steam consumption (CO2 regeneration process) and electricity
consumption (compression). The vapor compression discharge pressure is 4 bar in this example.

The overhead compression configuration was further evaluated with heat integration at two
different stripper operating pressures (1.3 and 1.8 bar), with the same outlet conditions as in
the baseline case. In this study, the steam conservation potential is defined as the net
consumption reduction with respect to the reboiler duty in the standard configuration, see
Appendix F Steam conservation potential for a calculation example. The simulation results
(Table 12) indicate that increasing the compressor discharge pressure with 1 bar increases
the steam conservation potential in both cases. The first compressor in the compression train
usually has a discharge pressure of around 4 bar (Onarheim et al., 2017), and since the
CO2/H2O gas stream exiting from the modified stripper will already be pressurized means
that one less compression step is needed when comparing to the standard configuration.
Neglecting the electricity consumption due to the overhead compressor can lead one to think
that the stripper operating at lower pressure, combined with the higher discharge pressure,
results in the largest steam conservation potential (24% as shown in Table 12). At least when
comparing just these four examples with each other.

30
 Table 12 Calculations of primary steam demand and steam conservation potential for different
versions of the stripper overhead compression configuration. The steam conservation potential
indicates a net consumption reduction with respect to the heat duty in the standard configuration.
Stripper pressure (bar)
1.3 1.8
Compressor discharge pressure (bar)
Energy sink 4 5 4 5
Heat duty (MWth) -196 -196 -153 -153
Heat from condensation (MWth) +60 +79 +5 +15
Reboiler Steam demand (MWth) -136 -117 -148 -138
Steam conservation (MWth) 19 37 6 16
Steam conservation (%) 12 24 4.1 11
Overhead
Electricity duty (MWe) -14 -18 -6 -8
compressor
Thermodynamic energy efficiency potential 1.32 2.12 1.09 2.14

However, the electricity provided by a pulp mill is most probably produced internally through
conversion of superheated steam. The ratio of energy output and energy input is commonly
used to measure energy efficiency (Forsström et al., 2011). Considering the additional
electricity demand will be relevant if the intention is to maximize the energy efficiency. In this
case, the energy output is represented as the savings in steam consumption due to heat
integration and the energy input as the compression work. Taking the thermodynamic energy
efficiency potential into account indicates that the optimal case among those presented in
Table 12 is the 1.8-bar stripper equipped with an overhead compressor with 5 bar discharge
pressure. This modification may lead to savings of up to 11% in primary steam consumption
and will require one less compression step in the downstream purification process, relative
to the standard MEA process configuration.

31
 7. Conclusion
A MEA-based chemical absorption and desorption process was rigorously modelled in Aspen
Plus with a rate-based approach and validated against experimental data. A good agreement
was achieved between the simulation results and experimental data in the absorber
validation, with no more than 1.9% error in the estimation of the CO2 removal efficiency and
4.1% error in the rich loading estimation. However, the results from the simulated stripper
models did not coincide with experimental data to the same extent. The default rate-based
setting was then altered, which reduced the estimated error of the reboiler duty from 22% to
6% for the UTA case. The predicted lean loading in the SINTEF case was not affected by the
alternative rate-based model.

Different solvent flow rates were evaluated when dimensioning the columns to ensure
efficient use of the entire packing. The minimum required packed column heights for the
absorber and desorber in the standard MEA process were defined as 14 m and 9 m
respectively, considering 90 wt% capture rate from the reboiler flue gas of a reference pulp
mill with 700 000 ADt pulp production capacity per year. The specific reboiler duty needed to
achieve this capture efficiency with 30 wt% MEA, L=1.1Lmin and 0.25 lean CO2 loading in a
standard MEA process configuration is 4.1 GJ/tCO2.

Suitable temperature levels for heat integration, within and across the capture plant, were
obtained by considering the heat consumed and excess heat made available in the MEA
process. This was possible through an assessment of different versions of a stripper
overhead compression configuration. The gas stream exiting the stripper is condensed to
20ºC before compression in the conventional MEA process. Compressing this stream and
providing the reboiler with some of the higher quality heat, released from the water
condensation, will decrease the steam flow rate drawn off from the steam cycle. However, at
the cost of additional electricity needed for compression. Energy balance calculations were
conducted for the modified MEA processes by considering both the steam conservation
potential, with reference to the baseline case, and the theoretical energy efficiency potential.

The simulation results indicate that the configuration with a stripper operating pressure and
compressor discharge pressure of 1.8 bar and 5 bar respectively may lead to savings of up
to 11% in steam consumption, while cutting the need of one of compression step in the
downstream purification process. The specific reboiler duty is 3.7 GJ/tCO2 with this version of
stripper modification. The heat integration potential of the capture plant with a specific
process in a Kraft pulp mill indicated energy savings in the same order of magnitude.Thereby,
making the BECCS concept a more attractive environmental solution for the Swedish pulp
and paper industry to mitigate climate change.

32
 8. Future work
This chapter addresses some limitations of the study and suggests alternative approaches
of relevance to future work in this area.

8.1. Aspen Plus model

The fact that the MEA solvent is not completely recirculated and that no chemical reactions
are considered for the MEA degradation is a drawback with this simulation. Solvent
degradation reactions should be included when modelling the MEA process since it may have
a relevant role in the economics and operation of the capture system as a whole. Another
property package should be used for processes operating at 10 bar or higher, since the
ENRTL model gives accurate results up to medium pressures only. The CO 2 compression
section should be modelled separately, using the Predictive Soave-Redlich-Kwong (PSRK)
property method instead since it is generally more accurate for systems without water present
at high pressures (Adams & Barton, 2010). Ultimately, the mass and energy balances in the
condenser and compressors are of main importance when analyzing the energy conservation
potential. Choosing a thermodynamic package that supports higher validity in the
compression and condenser stream results, over the one used in this study (mainly relevant
for non-ideal electrolyte liquid solutions), is justifiable since the streams in those process
steps consists of mainly CO2 and H2O.

The results from the stripper model validation was less accurate than that of the absorber
model validation in terms of how well they match experimental data. It is important to
remember that, out of 48 campaign runs in the UTA plant and out of 19 campaign runs in the
SINTEF plant, only one from each plant was run through the Aspen model. An improvement
of the validation part in this study can be made by also reproducing all the other runs (or at
least those runs that differ in operating conditions) of each plant. Doing so will give more
confidence to the validity and reliability of the simulation results. Heat losses from the
absorber and stripper columns were not accounted for in the simulation results either. Adding
this detail in the model calculations will further increase the reliability of the stripping
performance results.

8.2. Carbon capture design

The most relevant parameters that should be tuned in a continued version of this study are
listed below. These are parameters that were kept constant throughout the evaluation of the
modified stripper configuration. Tuning some of these as an attempt to reach a more
optimized process design could be beneficial for a pulp mill integrated with carbon capture in
terms of operational cost. The partial capture concept is especially important to consider in
the process integration of carbon capture in integrated Kraft pulp and paper mills, since these
normally lack steam surplus. Potential investors in the near future may perceive such an
approach as more technical and economically feasible in retrofit projects. Extra emphasis
should therefore be put on the capture rate by drawing up justifiable partial capture scenarios.

Position of rich solvent feed

Lean solvent loading
Solvent flow rate
Capture rate

33
 References

Abdul Manaf, N., Qadir, A., Sharma, M., Parvareh, F., Milani, D., & Abbas, A. (2016). Model-based
optimisation of highly-integrated renewables with post-combustion carbon capture processes.

Adams, T. A. (2018). Learn Aspen Plus in 24 hours. McGraw Education.

Adams, T. A., & Barton, P. I. (2010). High-efficiency power production from coal with carbon capture.
AIChE Journal, 56(12), 3120 3136. https://doi.org/10.1002/aic.12230

Ahn, H., Luberti, M., Liu, Z., & Brandani, S. (2013). Process configuration studies of the amine capture
process for coal-fired power plants. International Journal of Greenhouse Gas Control, 16, 29 40.

Aspen Technology, Inc. (2014). Rate-Based Model of the CO2 Capture Process by MEA using Aspen
Plus. Bedford, MA.

Batoon, V., Hicks, D., Huang, I., Hofmann, T., Kniep, J., Paulaha, C., Westling, E., & Merkel, T. (2019,
August 29). Scale-Up and Testing of Advanced Polaris Membrane CO2 Capture Technology (DE-
FE0031591).

Biermann, M., Normann, F., Johnsson, F., & Skagestad, R. (2018). Partial Carbon Capture by
Absorption Cycle for Reduced Specific Capture Cost. Industrial & Engineering Chemistry Research,
57(45), 15411 15422. https://doi.org/10.1021/acs.iecr.8b02074

BMU Austria. (1995). Sector Waste Management Concept for Paper and Pulp Industry. Prevention-
Utilisation-Disposal. Ministry of Environment.

Bottoms, R. R. (1931). Organic Bases for Gas Purification. Industrial & Engineering Chemistry, 23(5),
501 504. https://doi.org/10.1021/ie50257a007

Bravo, J.L., Rocha, J. A., & Fair, J. R. (1985). Mass transfer in gauze packings. Hydrocarbon Process,
64, 91 95.

Bravo, J.L., Rocha, J. A., & Fair, J. R. (1992). A comprehensive model for the performance of columns
containing structured packings. Industrial & Engineering Chemistry Research, 128, A439-457.

Bui, M., Adjiman, C. S., Bardow, A., Anthony, E. J., Boston, A., Brown, S., Fennell, P. S., Fuss, S.,
Galindo, A., Hackett, L. A., Hallett, J. P., Herzog, H. J., Jackson, G., Kemper, J., Krevor, S., Maitland,
G. C., Matuszewski, M., Metcalfe, I. S., Petit, C., Mac D e , N. (2018). Ca b ca ea d age
(CCS): The way forward. Energy & Environmental Science, 11(5), 1062 1176.

Bui, M., Gunawan, I., Verheyen, V., Feron, P., Meuleman, E., & Adeloju, S. (2014). Dynamic modelling
and optimisation of flexible operation in post-combustion CO2 capture plants A review. Computers &
Chemical Engineering, 61, 245 265. https://doi.org/10.1016/j.compchemeng.2013.11.015

Conference on Mathematical Modelling, & Troch, I. (Eds.). (2009). Proceedings / MATHMOD 09: 6th
Vienna Conference on Mathematical Modelling, Vienna, February 11 - 13, 2009, Vienna University of

34
Technology, Austria ; abstract volume. ARGESIM, Publ. House.

Dugas, R. (2006). Pilot plant study of carbon dioxide capture by aqueous monoethanolamine.

Errico, M., Madeddu, C., Pinna, D., & Baratti, R. (2016). Model calibration for the carbon dioxide-
amine absorption system. Applied Energy, 183, 958 968.

Ferrara, G., Lanzini, A., Leone, P., Ho, M. T., & Wiley, D. E. (2017). Exergetic and exergoeconomic
analysis of post-combustion CO2 capture using MEA-solvent chemical absorption. Energy, 130, 113
128. https://doi.org/10.1016/j.energy.2017.04.096

Forsström, J., Lahti, P., Pursiheimo, E., Rämä, M., Shemeikka, J., Sipilä, K., Tuominen, P., & Wahlgren,
I. (2011). Measuring energy efficiency: Indicators and potentials in buildings, communities and energy
systems. VTT.

Fosbøl, P. L., Gaspar, J., Ehlers, S., Kather, A., Briot, P., Nienoord, M., Khakharia, P., Le Moullec, Y.,
Berglihn, O. T., & Kvamsdal, H. (2014). Benchmarking and Comparing First and Second Generation
Post Combustion CO2 Capture Technologies. Energy Procedia, 63, 27 44.

FRBC. (2010). NOx emissions from recovery boilers why discrepancy between Finnish and Swedish
values (22.12.2010, p. 12). Finnish Recovery Boiler Committee.

Freguia, S., & Rochelle, G. T. (2003). Modeling of CO2 capture by aqueous monoethanolamine. AIChE
Journal, 49(7), 1676 1686. https://doi.org/10.1002/aic.690490708

Fuss, S., Jones, C. D., Kraxner, F., Peters, G. P., Smith, P., Tavoni, M., van Vuuren, D. P., Canadell, J.
G., Jackson, R. B., Milne, J., Moreira, J. R., Nakicenovic, N., Sharifi, A., & Yamagata, Y. (2016).
Research priorities for negative emissions. Environmental Research Letters, 11(11), 115007.

Garcia, M., Knuutila, H. K., & Gu, S. (2017). ASPEN PLUS simulation model for CO 2 removal with
MEA: Validation of desorption model with experimental data. Journal of Environmental Chemical
Engineering, 5(5), 4693 4701. https://doi.org/10.1016/j.jece.2017.08.024

Ga a d i , S. . (2017). Technical and economic conditions for efficient implementation of CO 2

capture: Process design and operational strategies for power generation and process industries.
Chalmers University of Technology.

Garðarsdóttir, S. Ó., Normann, F., Skagestad, R., & Johnsson, F. (2018). Investment costs and CO 2
reduction potential of carbon capture from industrial plants A Swedish case study. International
Journal of Greenhouse Gas Control, 76, 111 124. https://doi.org/10.1016/j.ijggc.2018.06.022

Global CCS Institute. (2011). Bio-energy with CCS.

https://www.globalccsinstitute.com/archive/hub/publications/25921/fact-sheet-3-bioccs-v4.pdf

Global CCS Institute. (2019). The Global Status of CCS: 2019.

35
Hammer, G., Lübcke, T., Kettner, R., Pillarella, M. R., Recknagel, H., Commichau, A., Neumann, H.-
J., & Pac ka Lah e, B. (2006). Na a Ga . I Ullmann s Encyclopedia of Industrial Chemistry.
American Cancer Society. https://doi.org/10.1002/14356007.a17_073.pub2

Hanley, B., & Chen, C.-C. (2012). New mass-transfer correlations for packed towers. AIChE Journal,
58(1), 132 152. https://doi.org/10.1002/aic.12574

Hasan, M. M. F., Baliban, R. C., Elia, J. A., & Floudas, C. A. (2012). 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, 51(48),
15642 15664. https://doi.org/10.1021/ie301571d

Haydary, J. (2018). Chemical Process Design and Simulation: Aspen Plus and Aspen Hysys
Applications. John Wiley & Sons, Inc. https://doi.org/10.1002/9781119311478

Hektor, E. (2008). Post-Combustion CO2 Capture in Kraft Pulp Mills Technical, Economic and System
Aspects [Chalmers University of Technology].
https://research.chalmers.se/en/publication/74762

Hwang, J., Kim, J., Lee, H. W., Na, J., Ahn, B. S., Lee, S. D., Kim, H. S., Lee, H., & Lee, U. (2019).
An experimental based optimization of a novel water lean amine solvent for post combustion CO2
capture process. Applied Energy, 248, 174 184. https://doi.org/10.1016/j.apenergy.2019.04.135

IEA. (2007). Tracking Industrial Energy Efficiency and CO2 Emissions. OECD.
https://doi.org/10.1787/9789264030404-en

IEAGHG. (2016). Techno-Economic Evaluation of Retrofitting CCS in a Market Pulp Mill and an
Integrated Pulp and Board Mill (No. 2016/10).

IPCC. (2014). Climate Change 2014 Mitigation of Climate Change: Working Group III Contribution to
the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University
Press. https://doi.org/10.1017/CBO9781107415416

Jönsson, J., & Berntsson, T. (2012). Analysing the potential for implementation of CCS within the
European pulp and paper industry. Energy, 44(1), 641 648.

Knudsen, J., Andersen, J., Jensen, J. N., & Biede, O. (2011). Results from test campaigns at the 1 t/h
CO2 post combustion capture pilot-plant in Esbjerg under the EU FP7 CESAR project. 1st Post
Combustion Capture Conference, 17 19.

Kuparinen, K., Vakkilainen, E., & Tynjälä, T. (2019). Biomass-based carbon capture and utilization in
kraft pulp mills. Mitigation and Adaptation Strategies for Global Change, 24(7), 1213 1230.
https://doi.org/10.1007/s11027-018-9833-9

Kuramochi, T., Ramírez, A., Turkenburg, W., & Faaij, A. (2012). Comparative assessment of CO 2
capture technologies for carbon-intensive industrial processes. Progress in Energy and Combustion
Science, 38(1), 87 112. https://doi.org/10.1016/j.pecs.2011.05.001

36
Kvamsdal, H.M., & Rochelle, G. T. (2008). Effects of the Temperature Bulge in CO 2 Absorption from
Flue Gas by Aqueous Monoethanolamine. Industrial & Engineering Chemistry Research, 47(3), 867
875. https://doi.org/10.1021/ie061651s

Kvamsdal, H.M., Jakobsen, J. P., & Hoff, K. A. (2009). Dynamic modeling and simulation of a CO 2
absorber column for post-combustion CO2 capture. Chemical Engineering and Processing: Process
Intensification, 48(1), 135 144. https://doi.org/10.1016/j.cep.2008.03.002

Le Moullec, Y., & Kanniche, M. (2011). Screening of flowsheet modifications for an efficient
monoethanolamine (MEA) based post-combustion CO2 capture. International Journal of Greenhouse
Gas Control, 5(4), 727 740. https://doi.org/10.1016/j.ijggc.2011.03.004

Lemaire, E., Bouillon, P. A., & Lettat, K. (2014). Development of HiCapt+ TM Process for CO 2 Capture
from Lab to Industrial Pilot Plant. Oil & Gas Science and Technology Revue d IFP Energies
Nouvelles, 69(6), 1069 1080. https://doi.org/10.2516/ogst/2013153

Li, K., Leigh, W., Feron, P., Yu, H., & Tade, M. (2016). Systematic study of aqueous monoethanolamine
(MEA)-based CO2 capture process: Techno-economic assessment of the MEA process and its
improvements. Applied Energy, 165, 648 659. https://doi.org/10.1016/j.apenergy.2015.12.109

Luo, X., & Wang, M. (2017). Improving Prediction Accuracy of a Rate-Based Model of an MEA-Based
Carbon Capture Process for Large-Scale Commercial Deployment. Engineering, 3(2), 232 243.
https://doi.org/10.1016/J.ENG.2017.02.001

Madeddu, C., Errico, M., & Baratti, R. (2018). Process analysis for the carbon dioxide chemical
absorption regeneration system. Applied Energy, 215, 532 542.

Madeddu, C., Errico, M., & Baratti, R. (2019). CO2 Capture by Reactive Absorption-Stripping:
Modeling, Analysis and Design. https://doi.org/10.1007/978-3-030-04579-1

Marik Singh, S. K., Zabiri, H., Isa, F., & M. Shariff, A. (2017). Simulation of CO2 Rich Natural Gas
Pilot Plant Carbon Dioxide Absorption Column at Elevated Pressure Using Equilibrium and Rate Based
Method. In M. S. Mohamed Ali, H. Wahid, N. A. Mohd Subha, S. Sahlan, M. A. Md. Yunus, & A. R.
Wahap (Eds.), Modeling, Design and Simulation of Systems (Vol. 752, pp. 327 336). Springer
Singapore. https://doi.org/10.1007/978-981-10-6502-6_29

Massimiliano, M., Madeddu, C., & Baratti, R. (2019). 4. Reactive absorption of carbon dioxide:
Modeling insights. In F. I. Gómez-Castro & J. G. Segovia-Hernández (Eds.), Process Intensification
(pp. 79 124). De Gruyter. https://doi.org/10.1515/9783110596120-004

Mclaren, D. (2012). A comparative global assessment of potential negative emissions technologies.

Process Safety and Environmental Protection, 90, 489 500. https://doi.org/10.1016/j.psep.2012.10.005

MTR. (n.d.). CO2 Removal from Syngas. Membrane Technology & Research. Retrieved February 4,
2020, from https://www.mtrinc.com/our-business/refinery-and-syngas/co2-removal-from-syngas/

Muradov, N. Z. (2014). Liberating energy from carbon: Introduction to decarbonization. Springer.

37
Neveux, T., Le Moullec, Y., & Favre, É. (2017). Post-combustion CO2 Capture by Chemical Gas-Liquid
Absorption: Solvent Selection, Process Modelling, Energy Integration and Design Methods. In A. I.
Papadopoulos & P. Seferlis (Eds.), Process Systems and Materials for CO2 Capture (pp. 283 310). John
Wiley & Sons, Ltd. https://doi.org/10.1002/9781119106418.ch11

Normann, F., Skagestad, R., Biermann, M., Wolf, J., & Mathisen, A. (2019). CO2stCap Reducing the
Cost of Carbon Capture in Process Industry.
https://research.chalmers.se/publication/512527/file/512527_Fulltext.pdf

Øi, L. E. (2010). CO2 removal by absorption: Challenges in modelling. Mathematical and Computer
Modelling of Dynamical Systems, 16(6), 511 533. https://doi.org/10.1080/13873954.2010.491676

Onarheim, K., Santos, S., Kangas, P., & Hankalin, V. (2017). Performance and costs of CCS in the pulp
and paper industry part 1: Performance of amine-based post-combustion CO2 capture. International
Journal of Greenhouse Gas Control, 59, 58 73. https://doi.org/10.1016/j.ijggc.2017.02.008

Puxty, G., & Maeder, M. (2016). The fundamentals of post-combustion capture. In Absorption-Based
Post-combustion Capture of Carbon Dioxide (pp. 13 33). Elsevier. https://doi.org/10.1016/B978-0-08-
100514-9.00002-0

Razi, N., Svendsen, H. F., & Bolland, O. (2013). Validation of mass transfer correlations for CO 2
absorption with MEA using pilot data. International Journal of Greenhouse Gas Control, 19, 478 491.
https://doi.org/10.1016/j.ijggc.2013.10.006

Regeringskansliet. (2017). Klimatlag (2017:720). Miljö- och energidepartementet.

Reynolds, A. J., Verheyen, T. V., & Meuleman, E. (2016). Degradation of amine-based solvents. In
Absorption-Based Post-combustion Capture of Carbon Dioxide (pp. 399 423). Elsevier.
https://doi.org/10.1016/B978-0-08-100514-9.00016-0

Rootzén, J., Kjärstad, J., Johnsson, F., & Karlsson, H. (2018, May 22). Deployment of BECCS in basic
industry-a Swedish case study.

Skagestad, R., Wolf, J., Anheden, M., Garðarsdóttir, S. Ó., Mathisen, A., & Normann, F. (2018, May
24). Impact analysis of CO2 capture from pulp mills Effects on CO2 emissions, costs and green
electricity production. International Conference on Negative CO2 Emissions, Gothenburg, Sweden.
Smith, P., & Porter, J. R. (2018). Bioenergy in the IPCC Assessments. GCB Bioenergy, 10(7), 428 431.

Solbraa, E. (2002). Equilibrium and Non-Equilibrium Thermodynamics of Natural Gas Processing

(0809-103X; 2002:146) [Doctoral dissertation]. Norwegian University of Science and Technology.

Suhr, M., Klein, G., Kourti, I., Rodrigo Gonzalo, M., Giner Santonja, G., Roudier, S., Delgado Sancho,
L., & Institute for Prospective Technological Studies. (2015). Best Available Techniques (BAT)
reference document for the production of pulp, paper and board. Publications Office.
http://dx.publications.europa.eu/10.2791/370629

Tobiesen, F. A., Juliussen, O., & Svendsen, H. F. (2008). Experimental validation of a rigorous desorber
model for CO2 post-combustion capture. Chemical Engineering Science, 63(10), 2641 2656.

38
Tontiwachwuthikul, P., Meisen, A., & Lim, C. J. (1992). CO2 absorption by NaOH, monoethanolamine
and 2-amino-2-methyl-1-propanol solutions in a packed column. Chemical Engineering Science, 47(2),
381 390. https://doi.org/10.1016/0009-2509(92)80028-B

Wolf, J. (2020, June 9). Research Project Manager [Personal communication].

ZEP, & EBTP. (2012). Biomass with CO2 Capture and Storage (Bio-CCS) The way forward for
Europe. http://www.etipbioenergy.eu/images/EBTP-ZEP-Report-Bio-CCS-The-Way-Forward.pdf

Zhang, Y., & Chen, C.-C. (2013). Modeling CO2 Absorption and Desorption by Aqueous
Monoethanolamine Solution with Aspen Rate-based Model. Energy Procedia, 37, 1584 1596.
https://doi.org/10.1016/j.egypro.2013.06.034

Zhang, Y., Chen, H., Chen, C.-C., Plaza, J. M., Dugas, R., & Rochelle, G. T. (2009). Rate-Based Process
Modeling Study of CO 2 Capture with Aqueous Monoethanolamine Solution. Industrial & Engineering
Chemistry Research, 48(20), 9233 9246. https://doi.org/10.1021/ie900068k

Zhang, Y., Que, H., & Chen, C.-C. (2011). Thermodynamic modeling for CO2 absorption in aqueous
MEA solution with electrolyte NRTL model. Fluid Phase Equilibria, 311, 67 75.
https://doi.org/10.1016/j.fluid.2011.08.025

39
Appendix A

Absorption validation: parameters and results

It was confirmed in a previous version of the absorber model validation within this project that
the calculated performance values with regard to the absorber (rich loading and CO2 removal
%) coincide a bit better with corresponding experimental data with increasing number of
segments. The simulation results presented in this part were therefore calculated using 50
segments for the packing section, i.e. the reactive absorber height is discretized in equal
number of parts. The activity basis rate constants, k and E, are necessary input values for
Aspen Plus to run the rate-based calculation with the power law expression. The values
reported by Hikita et al. (1977) for reaction 4-5 and those by Pinsent et al. (1956) for reaction
6 7 are almost exclusively applied throughout all literature about modelling of the MEA
process. However, not all studies disclose the exact values used and simply refer directly to
the works of Hikita et al. (a) and Pinsent et al. (b) respectively. Some studies (Aspen
Technology, Inc., 2014; Errico et al., 2016; Zhang & Chen, 2013) modified these values to fit
different conditions, which is reasonable considering that the stripper operates at a higher
temperature than the absorber for example.

Table A-1 Kinetic parameters, k and E, in Eq. (1) according to different sources.
Reaction k (kmol/m3s) E (cal/mol) Notation Reference
3.020・ 1014 9840.5 c (Zhang & Chen (2013)

3.020・1014 9855.8 d, g (Aspen Technology, Inc., 2014), (Razi et al., 2013)

4 3.020・1010 9840.5 e (Luo & Wang, 2017)

9.770・1010 9855.8 f (Errico et al., 2016)

6.839・1010 9855.8 f2 (Errico et al., 2016)

3.230・1019 15655 f (Errico et al., 2016)

5 (both)
2.261・1019 15655 f2 (Errico et al., 2016)

5.520・1023 16492 c2, e (Zhang & Chen (2013), (Luo & Wang, 2017)
5 16518 d2 (Aspen Technology, Inc., 2014),
5.520・1023
(absorber)
5.520・1013 16518 g (Razi et al., 2013)

5 6.560・1027 22748 c2, e (Zhang & Chen, 2013), (Luo & Wang, 2017)
(stripper)
6.500・1027 22782 d2 (Aspen Technology, Inc., 2014)

1.330・1027 13227 c, e (Zhang & Chen, 2013), (Luo & Wang, 2017)

6 1.330・1017 13249 d, g (Aspen Technology, Inc., 2014), (Razi et al., 2013)

4.320・1013 13249 f (Errico et al., 2016)

6.630・1016 22748 c, e (Zhang & Chen, 2013), (Luo & Wang, 2017)

7 6.630・1016 25656 d, g (Aspen Technology, Inc., 2014), (Razi et al., 2013)

2.380・1017 29451 f (Errico et al., 2016)

Table A-1 presents different kinetic parameter values reported in literature and from here on
these sources will be referred to as c-f, according to their corresponding notation. All of them

40
 ultimately origin from either a or b. Some values within the same category, i.e. reaction type,
differ by several orders of magnitude. Notations with 2 at the end indicate that the value was
adjusted with respect to a and b. Moreover, source e refers to c but it is worth noticing that
the k value from these sources for reaction 4 differ from each other. This is most likely a typo
since the same was noticed when comparing the k values of reaction 5 for the absorber from
source g and d2. After all, source g claims to have taken their values straight from source d.
In both cases, the difference is only in the exponential part. Even so, all of these different
kinetic sets were evaluated as they were.

Table A-2 He a c a a a e ers from different sources. T refers to the system

temperature range in which these values are valid.
Component pairs C1 C2 C3 C4 T (K) Source
(Luo & Wang, 2017), (Aspen
CO2-H2O 100.650 -6147.7 -10.191 0 273 473
Technology, Inc., 2014)
89.452 -2934.6 -11.592 0.01644 280-600 (Luo & Wang, 2017)
CO2-MEA
20.3143 -896.5 0 0 0-2000 (Aspen Technology, Inc., 2014)

As explained in 5.1.1. Absorber, he He a c a f CO2 with water and MEA could

not be obtained from the Aspen Plus databank, hence, they had to be found in literature
(Table A-2). Contrary to the solubility of CO2 in water, which has been well studied, there is
relatively limited knowledge about the solubility of CO 2 in MEA (Zhang et al., 2011). Hereafter,
the pilot-scale plant will be referred to as ABS1 and the lab-scale plant as ABS2. Figure A-1
i) and iii) shows the temperature profiles of the liquid phase in ABS1 and ABS2 respectively
when using the Henry constants for CO2-H2O and CO2-MEA from (Luo & Wang, 2017), e.

Pilot plant Lab-scale plant

80 60
exp exp
c, d 55 c, d
Liquid temperature [ o C]

Liquid temperature [ o C]

75 e e
f 50 f
f2 f2
70 g 45
g
40
65
35

60 30

25
55
20

0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 5 6 7
Distance from bottom [m] Distance from bottom [m]
i) Henry constants from source e iii) Henry constants from source e

80 60
exp exp
c, d 55 c, d
Liquid temperature [ o C]

Liquid temperature [ o C]

75 e e
f 50 f
f2 f2
70 45
g g
40
65
35

60 30

25
55
20

0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 5 6 7
Distance from bottom [m] Distance from bottom [m]
ii) Henry constants from source d iv) Henry constants from source d

Figure A-1 Variation of the liquid temperature profile for ABS1 (i-ii) and ABS2 (iii-iv) using different
kinetic and Henry constant parameters.

41
Several of the simulation results are overlapping in Figure A-1 iii), hence only two lines are
visible. c,d overlaps with the results from e, while f and f2 with g. Figure A-1 i) and iv) show the
temperature profiles of the liquid phase in ABS1 and ABS2 respectively when using the Henry
constants for CO2-H2O and CO2-MEA from (Aspen Technology, Inc., 2014), d. In Figure A-1
iv), it becomes difficult to distinguish the simulation results from f, but in this example they are
overlapping with the results from f2. When comparing all these four graphs, i-iv), it is clear that
the Henry constants from d is the best option to continue with in the validation assessment, as
its corresponding simulation results fits the experimental data from ABS1 and ABS2 better than
that of e, without compromising the model accuracy. The simulation results using kinetic sets
c and d are represented by the same line, yellow color, in the following two figures since they
were only slightly different in regard to activation energy, E, and showed no significant
difference of the temperature profiles upon graphical comparison.

Pilot plant Lab-scale plant

60
80 exp exp
G0 55 G0
G1 G1
Liquid temperature [ o C]
75
Liquid temperature [ o C]

G2 50 G2
G3 G3
G4
45 G4
70
40
65 35

30
60
25
55 20

0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7
Distance from bottom [m] Distance from bottom [m]
i) Kinetic parameters from source g iii) Kinetic parameters from source g

60
80
exp exp
D0
55 D0
D1 D1
Liquid temperature [ o C]

75
Liquid temperature [ o C]

D2
50 D2
D3 D3
D4
45 D4
70
40
65
35

30
60
25
55
20

0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7
Distance from bottom [m] Distance from bottom [m]
ii) Kinetic parameters from source d iv) Kinetic parameters from source d

Figure A-2 Variation of the liquid temperature profile for ABS1 (i-ii) and ABS2 (iii-iv) for different
combinations of kinetic parameters and interaction energy parameters.

The validation procedure continued by adjusting the GMENCC parameters in the Properties
environment as shown in Table A-3, hi e ee i g he He c a a a e e fixed
according to source d. The 0 accompanying the upper-case letters, G and D, shows that the
parameters were kept exactly as those automatically computed after selecting the ENRTL-
RK method. The other values were retrieved from the sample simulation fi e Ra e-Based
Model of the CO2 Ca e P ce b MEA i g A e P h gh E cha ge. Figure A-2
shows the temperature profiles of ABS1 (i-ii) and ABS2 (iii-iv) using different combinations of
the energy interaction parameters. Overlapping occurs in these graphs, especially on i, iii and

42
iv, for changes 1-2 on the GMENCC parameters. The performance results of the absorbers
using D0, D1 and D3 are tabulated in 5.1.1. Absorber.

Table A-3 Combinations of GMENCC parameters that relate to molecule-ion and ion-ion interactions.
Notation Value Molecule i or electrolyte i Molecule j or electrolyte j Value Notation
9.888 H2O MEAH+ MEACOO- 6.732
-4.951 MEAH+ MEACOO- H2O -3.163
5.354 H2O MEAH+ HCO3- 8.572
-4.071 MEAH+ HCO3- H2O -4.009
8.045 H2O H3O+ HCO3- 8
D0, G0
-4.072 H3O+ HCO3- H2O -4
8.045 H2O H3O+ OH- 8
-4.072 H3O+ OH- H2O -4
8.045 H2O H3O+ CO3-2 8
-4.072 H3O+ CO3-2 H2O -4
8 CO2 H3O+ OH- 8
-4 H3O+ OH- CO2 -4
8 CO2 H3O+ HCO3- 8
D1, G1
-4 H3O+ HCO3- CO2 -4
8 CO2 H3O+ CO3-2 8
-4 H3O+ CO3-2 CO2 -4
D2, G2 8 CO2 H3O+ MEACOO- 8
-4 H3O+ MEACOO- CO2 -4
8 CO2 MEAH+ OH- 8
-4 MEAH+ OH- CO2 -4
6 CO2 MEAH+ HCO3- 6
5.012 MEAH+ HCO3- CO2 5.012
6 CO2 MEAH+ CO3-2 6
5.069 MEAH+ CO3-2 CO2 5.069
D3, G3 6 CO2 MEAH+ MEACOO- 6 D4, G4
5.072 MEAH+ MEACOO- CO2 5.072
8 H2O MEAH+ CO3-2 8
-4 MEAH+ CO3-2 H2O -4
8 H2O H3O+ MEACOO- 8
-4 H3O+ MEACOO- H2O -4
8 H2O MEAH+ OH- 8
-4 MEAH+ OH- H2O -4
8 MEA H3O+ OH- 8
-4 H3O+ OH- MEA -4
8 MEA H3O+ HCO3- 8
-4 H3O+ HCO3- MEA -4
8 MEA H3O+ CO3-2 8
-4 H3O+ CO3-2 MEA -4
8 MEA H3O+ MEACOO- 8
-4 H3O+ MEACOO- MEA -4
8 MEA MEAH+ OH- 8
-4 MEAH+ OH- MEA -4
8 MEA MEAH+ HCO3- 8
-4 MEAH+ HCO3- MEA -4
8 MEA MEAH+ CO3-2 8
-4 MEAH+ CO3-2 MEA -4
8 MEA MEAH+ MEACOO- 8
-4 MEAH+ MEACOO- MEA -4

43
Appendix B

Stripper validation: parameters and results

Table B-1 RadFrac parameters that were tuned in the stripper validation.

Model I Model II
Interfacial area factor 1 10
Heat transfer 1 5
Liquid mass transfer coefficient factor 1 3
Vapor mass transfer coefficient factor 1 10
Reaction condition factor 0.9 0.1
Interfacial area method (Bravo et al., 1985) (Hanley & Chen, 2012)

Table B-2 Error in the estimation of the lean loading with number of segments for the SINTEF plant.

Number of segments
Model Performance Experimental
10 60 70
Lean loading 0.253 0.251 0.251 0.219
I
Error (%) 15 14 14 -
Lean loading 0.252 0.251 0.251 0.219
II
Error (%) 15 14 14 -

44
Appendix C

Absorber dimensioning

Table C-1 Column parameters obtained from the design analysis of an absorber operating at 1 bar.
Fixed
Theoretical
Liquid flow rate L/G ratio Base No. of pressure
D (m) H (m) max. pressure
(tonne/hr) (kg/kg) stage stages drop
drop (mbar)
(mbar)
Lmin 2190 3.4 10.7 100 3 100 0 -
1.05 Lmin 2298 3.5 10.9 25.4 6 100 0 -
1.1 Lmin 2408 3.7 11.0 14.2 8 100 0 -
1.1 Lmin 2408 3.7 11.0 14.0 7 100 25 29
1.15 Lmin 2517 3.9
discarded since it requires H/D <1 to obtain 90 wt% capture
1.2 Lmin 2627 4.0

The middle row in Table C-1 represents the column dimensions of the absorber used in the
standard MEA process configuration.

45
 Appendix D

Stripper dimensioning

Table D-1 Column parameters obtained from the design analysis of a stripper operating at 1.8 bar.

Fixed Theoretical
Capture rate Qreb Qreb Reboiler Limiting Condenser
H (m) D (m) pressure max. pressure
(tCO2/hr) (MW) (GJ/tCO2) T (K) stage T (K)
drop (mbar) drop (mbar)
17 6.28 135 150 3.99 394 69 0 - 313
15 6.31 135 150 3.98 394 69 0 - 313
13 6.31 135 149 3.98 394 69 0 - 313
11 6.30 135 149 3.98 394 69 0 - 313
9 6.30 135 149 3.98 394 69 0 - 313
9 6.33 135 150 4.01 394 69 15 18 293

The rich solvent flow rate entering the stripper is around 2440 tonne/hr and originates from
the lean solvent flow rate (1.1Lmin) that enters the absorber unit. The bottom row in Table
D-1 represents the column dimensions of the stripper used in the standard MEA process
configuration.

46
Appendix E

Validation of external partial condenser

Table E-1 Validation results for the external partial condenser in the standard MEA process. The
MEA concentration reported here concerns that in the gas stream exiting flash block.

Absorber Stripper
Partial condenser Inbuilt in RadFrac External
No. of stages 10 10 10
Rich loading at absorber outlet 0.504
Lean loading at stripper outlet 0.250 0.250
Boilup ratio 0.153 0.153
Stripper outlet gas stream temp (ºC) 105 105
Reboiler temp. (ºC) 121 121
MEA in product stream, ppm 1.8・10-5 8.3・10-5
Reboiler duty (MWth) 172 172
Condenser duty (MWth) -54.4 -54.5

47
Appendix F

Steam conservation potential

This example relates to the calculation of the steam conservation potential resulting from heat
integration in a modified MEA process with 1.8 bar stripper operating pressure and 5 bar
compression discharge pressure. The quantity marked in red (Table 12) corresponds to the
net heat required in the modified stripper, and the quantity marked in bold (Table 10)
corresponds to the reboiler duty in the standard process configuration.

1 . 𝟏𝟑𝟖.𝟐𝟐
𝑆𝑡𝑒𝑎𝑚 𝑐𝑜𝑛𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 % 1 .
100 11

48
TRITA CBH-GRU-2020:221

www.kth.se