Applied Thermal Engineering xxx (xxxx) 127457

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

F
Techno-economic assessment of blended-amine CO2 capture integrated
coal-fired power plants

OO
Muhammad Ikhsan Taipabu a, b, 1, Bang Yan Chou a, 1, Wei Wu a, d, ⁎, Bambang Sardi c
a Department of Chemical Engineering, National Cheng Kung University, Tainan 70101, Taiwan, ROC
b Department of Chemical Engineering, Pattimura University, Ambon 97134, Indonesia
c Research Centre for Advanced Materials, National Research and Innovation Agency, South Tangerang 15314, Indonesia
d Academy of Innovative Semiconductor and Sustainable Manufacturing, National Cheng Kung University, Tainan 70101, Taiwan, ROC

PR
ARTICLE INFO ABSTRACT

Mitigating global warming requires effectively capturing and storing carbon dioxide (CO2) emissions from coal-
Keywords:
Carbon capture process
fired power plants, with chemical absorption being the most established technology. This study evaluates three
Blended amine solvent process designs to enhance energy efficiency in CO2 capture. Design-1 incorporates a heat exchanger to recover
Techno-economic analysis waste heat. Design-2 integrates an intercooler and splits the CO2-rich feed to reduce energy consumption. Design-
D
Carbon emission trading 3 builds upon Design-2 by incorporating a membrane separation unit to enhance concentration gradients, further
Aspen Plus lowering reboiler energy demand and emissions. Two MEA (Monoethanolamine) and a blended MDEA/PZ (N-
Methyldiethanolamine/Piperazine)-based CO2 capture systems are employed to compare the energy-saving po-
tential of each design. The reboiler energy consumption for Design-1 is 3.35 GJ/tonne CO2 for MEA and 2.14 GJ/
TE

tonne CO2 for MDEA/PZ. In Design-2, energy consumption is reduced to 2.86 GJ/tonne CO2 and 2.10 GJ/tonne
CO2, respectively. Design-3, utilizing the MDEA/PZ solvent system, achieves the lowest reboiler energy con-
sumption at 2.02 GJ/tonne CO2. However, the high electricity demand of the membrane separation unit in De-
sign-3 induces higher operating costs than in Design-2. Accounting for reboiler heat source costs (coal and nat-
ural gas) and the EU-ETS carbon pricing, a projected carbon credit price of $555/tonne CO2 (expected by 2044)
would make Design-3 more cost-effective than Design-2 for electricity generation.
EC

Nomenclature Flow of component i in the chemical/solvent

stream added to stage j
Variable Description Flow of utility l added to stage j
i Component (i = dried biomass, lipids, car- Flow of component i in the chemical/solvent
RR

bohydrate, etc.) stream added to stage j

j Processing stage (1=Cultivation, 2=Har- Electricity demand l added to stage j
vesting, etc.) Fi,j-1 Flow of component i in the process stream com-
k Option/technological alternative ing from stage j−1
l Utility fi,j,k Flow of component i in the process stream leav-
r Reaction ing option k of stage j
CO

m Key reactant, a subset of component i Flow of component i in the chemical/solvent

yj,k Binary variable: 1 if option k from stage j is se- stream added to option k of stage j
lected 0 if otherwise Flow of utility l added to option k of stage j
Flow of component i in the process stream Electricity demand l added to option k of stage j
leaving stage j Vj After stage j the microalgae

⁎ Corresponding author.
E-mail address: [email protected] (W. Wu).
1 These authors contributed equally to this work.

https://doi.org/10.1016/j.applthermaleng.2025.127457
Received 15 April 2025; Received in revised form 24 June 2025; Accepted 4 July 2025
1359-4311/© 20XX

Note: Low-resolution images were used to create this PDF. The original images will be used in the final composition.
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

Table 1
Flue gas properties, composition and flow rate [18]. SRD Specific reboiler duty
Stream Flue gas Unit SRM Structural risk minimization
SVR Support vector regression
Temperature 104 °C WCI Working capital investment
Pressure 1 Atm
WTT Well-to-tank
Wet-basis Dry-basis
3 WTW Well-to-wheel
Flowrate 27843.68 23777.72 Nm /min
Relative humidity 14.6 %
Mole fraction
CO2 12.6 wt%

F
H 2O 12.06 wt% 1. Introduction
O2 6.10 wt%
N2 69.24 wt%
As the global demand for energy continues to grow, so does the ur-

OO
gency to address the environmental impacts of fossil fuel combustion,
Table 2 particularly in the power generation sector. While renewable energy
Kinetic parameters of MEA and MDEA/PZ solvent. technologies are advancing, fossil fuels, especially coal, remain integral
Reaction
3
k (kmol/m .s) Ea (kJ/mol) to electricity production in many regions. This reliance presents a sig-
nificant challenge in the context of global climate targets, necessitating
MEA solvent [1]
the rapid deployment of effective carbon mitigation technologies. The
(2)–(7) 55.38
intensifying impact of global warming has made climate change a
(2)–(8) 107.24

PR
(2)–(9) 41.20
pressing engineering and environmental challenge. Among the primary
(2)–(10) Absorber 69.05 contributors to greenhouse gas emissions is carbon dioxide (CO2),
(2)–(10) Stripper 95.24 largely produced by fossil fuel-based thermal power generation systems
MDEA/PZ solvent [2,3]
3
k (kmol/m .s) Ea (cal/mol) [1]. In response, many nations have committed to achieving net-zero
(7) 13,249 carbon emissions by 2050 to limit the global temperature rise to below
(8) 29,451 1.5 °C [2].
(11) 8038.3 Thermal power plants, particularly coal-fired units, remain a domi-
(12) 15,758 nant source of electricity worldwide but are also among the largest con-
D
(13) 8038.3 tributors to anthropogenic CO2 emissions, accounting for more than
(14) 18,372
30 % globally [3]. In Taiwan, thermal power constitutes over 86 % of
(15) 9029
electricity generation, underscoring the need for urgent decarboniza-
(16) 25,424
tion strategies within its energy infrastructure. Qian et al. [4] presents a
TE

Noted: k (rare constant), Ea (activation energy). novel hybrid power generation system that integrates natural gas and
coal to enhance the utilization of gas turbine exhaust while minimizing
the efficiency penalty typically associated with post-combustion carbon
concentration(kg/m3) capture. The findings show that incorporating carbon capture into the
costj Total cost in stage j hybrid system reduces the energy consumption of the capture process
Number of support vectors by 22.03 %, primarily due to a 2.4-fold increase in CO2 concentration
EC

k-th support vector within the flue gas stream.

Number of workers per shift Carbon capture technologies offer a viable approach to mitigate
Non-solid processing units emissions from existing power generation assets. Among various ap-
Parameter Description proaches, chemical absorption using solvents such as mo-
Purchasei Raw material i purchase price (USD/kg) noethanolamine (MEA) remains the most mature and widely deployed
Pricei Product i selling price (USD/kg) technology due to its high capture efficiency [1–3]. However, this
RR

αj,k Unit electricity consume (kwh/kg biomass or method faces limitations, particularly high thermal energy demands for
kwh/m3) solvent regeneration and long-term degradation of amines. Alternative
βj,k Unit duty consume (MJ/kg biomass) methods, including physical absorption [5–7], membrane separation
γj,k Unit chemical/solvent demand (kg/kg bio- [8,9], cryogenic distillation [10], and calcium looping [11–13], have
mass) been explored for their potential to enhance energy efficiency and re-
Abbreviation Description duce operating costs in thermal power systems.
CO

CHP Combined heat and power Zheng at al. [14] introduces a novel ammonia-driven calcium loop-
ERM Empirical risk minimization ing carbon capture system that significantly improves the decarboniza-
FCI Fixed capital investment tion performance of fossil-fueled power plants. The proposed system
GPU Gas permeation unit achieves a remarkably low efficiency penalty of just 0.6 percentage
HHV Higher heating value points and a carbon emission intensity of 18.6 kg CO2/MWhe. Notably,
HTL Hydrothermal liquefaction it delivers a negative CO2 avoidance energy consumption of −8.1
IRR Internal rate of return MJ1HV/kg CO2, indicating a net energy gain. This research demon-
LCA Life cycle assessment strates that this integrated approach offers superior energy efficiency
MINLP Mixed integer nonlinear programming and economic feasibility compared to conventional carbon capture sys-
ML Machine learning tems, presenting a promising pathway for the sustainable decarboniza-
NER Net energy ratio tion of existing power plants. Recent advances in membrane-based CO2
NPV Net present value separation, such as Polaris™ membranes with spiral-wound modules,
OP Open pond culture demonstrate improved performance at moderate pressures (<3 atm)
PBR Photo-bioreactor culture and a significant cost advantage, targeting $50/m2 compared to the
SRK Soave Redlich Kwong $500/m2 typical in petrochemical applications [15–17]. These configu-

2
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
Fig. 1. The comparison of MEA and MEA/PZ: (a) Correlation of liquid-to-gas ratio of solvent to packing height, (b) Correlation of reboiler duty and packing high of
the stripper.

PR
Table 3 the development of thermally optimized, economically viable carbon
Operating parameters of absorption tower and stripping tower using rate- capture systems that align with industrial decarbonization goals and
based mode. support the global transition to low-carbon thermal energy systems.
Parameter Absorber Stripper The remainder of this paper is structured as follows: Section 1 de-
scribes the background of the study, including the research gap and
Amine type 7 m MEA 5 m MDEA/5m 7 m MEA 5 m MDEA/5m
novelty. Section 2 describes the methodology, including the process
PZ PZ
simulation framework, solvent selection, and performance evaluation
D
Number of 2 2 1 1 criteria. Section 3 presents and discusses the simulation results, model
column validation, sensitivity analysis, and economic assessments across three
Number of 57 57 41 50 process configurations using MEA and MDEA/PZ solvents. Section 5
stage
concludes the study by summarizing key findings and proposing recom-
TE

Packing type Mellapak 250X Mellapak 250X Mellapak 250X Mellapak 250X
Pressure 1 1 1.8 1.8 mendations for future research and industrial application.
(bar)
Packing 17 17 12 15 2. Process description
height (m)
Column 12.1 10.9 8.5 7.1
diameter
This section provides a comprehensive overview of the design
methodology and simulation approach employed in modeling the CO2
EC

(m)
Leanin conditions capture system. The modeling framework encompasses several critical
Pressure 1.1 − 1.1 − components, each of which plays a vital role in determining the sys-
(bar)
tem’s overall performance and efficiency. It begins with the characteri-
Temperature 40 − 40 −
(℃)
zation of flue gas properties, which is essential for understanding the
Flow rate 2.76852e + 06 1.60637e + 06 2.76852e + 06 1.60637e + 06 composition and flow conditions of the gas stream entering the capture
(kg/h) system. Following this, the rationale for selecting specific chemical sol-
RR

Leanin composition (wt%) vents, namely MEA and MDEA/PZ, is discussed in detail, based on their
CO2 4.4 2 4.4 2
thermodynamic and kinetic performance in CO2 absorption. The sec-
H 2O 65.6 48 65.6 48
MEA 30 0 30 0
tion then delves into the detailed design of the absorption and stripping
MDEA 0 29 0 29 columns using rate-based modeling, including key operational parame-
PZ 0 21 0 21 ters such as packing height, solvent flow rates, and column configura-
tions. Finally, the integration of membrane-based separation is pre-
CO

rations enhance CO2 separation efficiency through optimized flow dy- sented, highlighting advanced membrane materials, flow configura-
namics and lower capital investment, making them attractive for retro- tions, and the simulation tools used to optimize performance. In this
fitting existing plants. work, Aspen Plus® was employed as the primary process simulation
This study focuses on the integration of amine-based chemical ab- tool to model and evaluate the CO2 capture systems and assess their en-
sorption with membrane separation technology to develop an energy- ergy performance under different design configurations and solvent se-
efficient hybrid CO2 capture system suitable for thermal power applica- lections.
tions. Various solvents, including MEA and MDEA/PZ blends, are as-
sessed for their thermal performance and process compatibility. By 2.1. Flue gas conditions
evaluating three innovative process configurations, the research aims to
optimize reboiler duty, minimize operational energy consumption, and Flue gas emissions from power plants play a critical role in assessing
enhance overall thermal efficiency. Furthermore, a long-term economic environmental impacts and optimizing carbon capture processes. Accu-
analysis incorporating projected EU-ETS carbon pricing is conducted to rate data on flue gas composition and flow rate are essential for evaluat-
assess the cost-effectiveness of each configuration. Notably, results indi- ing emission control strategies and improving overall process effi-
cate that membrane-enhanced absorption becomes economically supe- ciency. In this study, flue gas data were collected and analyzed to esti-
rior to conventional systems when carbon prices exceed $555/tonne, mate power generation and its associated CO2 emissions. The flue gas
projected to occur by 2044. The outcomes of this study contribute to analyzed originated from Unit 1 of a coal-fired power plant in Taiwan,

3
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 2. Gas-liquid balance verification chart of the amine absorbent: (a) MEA-H2O, (b) MDEA, (c).
EC

with data obtained from the Fixed Pollution Source Management Infor- MEA exhibits the highest requirement at 95 kJ/mol CO2 [19]. A lower
mation Disclosure Platform maintained by the Ministry of Environ- heat of absorption is generally associated with a slower CO2 uptake
ment, Taiwan [18]. The flue gas composition, flow rate (2,154 tonnes/ rate. Tertiary amines demonstrate a lower CO2 absorption rate than pri-
hour), and temperature (104 °C) were selected to reflect realistic post- mary and secondary amines. However, when combined with PZ, MDEA
combustion conditions from a conventional 550 MW coal-fired power benefits from PZ's catalytic effect, which enhances its CO2 absorption
plant in Taiwan. These conditions were used as a case study for evaluat- rate [17].
RR

ing the performance of different CO2 capture process configurations. This study chose MEA as a benchmark solvent due to its extensive
The specific sampling was conducted on August 6, 2020, through use and wide research coverage in chemical absorption processes. MEA
pipeline P101, as part of the national emissions monitoring program. and MDEA/PZ were selected as the primary s solvents for the CO2 ab-
The detailed properties, composition, and flow rate of the flue gas are sorption system to facilitate performance comparison. Existing litera-
presented in Table 1. As real-time power generation data were not ture indicates that for flue gas from natural gas power plants (∼4 mol%
available from the public platform, the power output was estimated CO2), an MDEA/PZ concentration ratio of 42 wt%/8 wt% is sufficient
CO

based on the installed capacity of the unit (550 MW) and an assumed for effective CO2 capture. However, for coal-fired power plants with
annual operation of 310.25 days (7,446 h). An equivalent capacity fac- flue gas containing approximately 12 mol% CO2, this PZ concentration
tor of 85 % was applied, following values reported in prior literature is insufficient for efficient CO2 absorption. To address this, Zhao et al.
[15], to provide a reasonable estimate of the plant's operational profile. [20] and Mengxiang et al. [21] proposed a blended solvent comprising
30 wt% MDEA and 20 wt% PZ for post-combustion CO2 capture in a
2.2. Solvent selection 650 MW power plant, achieving 90 % CO2 removal. Khan et al. [22] se-
lected a concentration ratio of 35 wt% MDEA and 15 wt% PZ (35/
The theoretical CO2 loading capacity varies among different amine 15 wt%) as the optimal formulation. Additionally, Khaled et al. (2024)
types. Primary amines, such as monoethanolamine (MEA), have a ca- employed a mixture of 87 wt% MDEA and 13 wt% PZ (equivalent to
pacity of 0.5 mol CO2 per mole of amine. In contrast, secondary amines 48 wt% amine in water), which demonstrated superior performance
(e.g., piperazine, PZ) and tertiary amines (e.g., methyldiethanolamine, compared to conventional MEA in terms of cost per ton of CO2 captured
MDEA) exhibit a higher loading capacity of 1 mol CO2 per mole of and stored, despite slightly higher capital costs. The higher concentra-
amine, thereby reducing the solvent consumption. Regarding the heat tion of MDEA and PZ was beneficial because it increased the absorption
of CO2 absorption, MDEA requires 61 kJ/mol CO2 [16], while the capacity of the solvent. However, viscosity and possible precipitation
MDEA/PZ blend demands approximately 70 kJ/mol CO2 [17], and due to the solubility limits of these components in water could be limit-

4
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 3. CO2 concentration distribution of (a) absorption tower and (b) stripping tower; Temperature distribution of (c) absorption tower and (d) stripping tower.

Table 4 based on simulation outputs under standardized flue gas treatment

Comparison of various components between simulation and experiment. conditions to ensure valid performance comparisons.
EC

Stream Purified gas Capture CO2 richout

2.3. Absorber and stripper
Exp. Sim. Error Exp. Sim. Error Exp. Sim. Error
(%) (%) (%)
The RadFrac module in Aspen Plus is applied to model both the ab-
Flow 241.4 235.9 −2.29 68.4 68.87 0.68 10.33 10.68 3.42 sorber and stripper in a CO2 capture process. The RadFrac model is a
(kg/h) rigorous distillation column model that can handle equilibrium and
RR

Mass Fraction rate-based calculations, making it suitable for modeling absorption and
MEA − − − − − − 0.272 0.276 1.55
stripping processes. The equilibrium reactions include water dissocia-
H 2O 0.082 0.079 −3.30 0.003 0.003 0.33 0.641 0.638 −0.49
CO2 0.017 0.017 −1.74 0.997 0.997 −0.01 0.087 0.086 −1.27
tion (Eq. (1)), carbonate formation (Eq. (2)), and the protonation of
N2 0.783 0.787 0.58 − − − − − − alkanolamines (MEA, MDEA, PZ, and PZ carbamate) (Eqs. (3)–(6)). The
O2 0.117 0.116 −0.65 − − − − − − rate-based reactions involve hydrogen carbonate formation (Eqs. (6)–
(7)) and carbamate formation (Eqs. (8)–(16)).
ing in practical applications. Therefore, the current study adopts an
CO

MDEA/PZ concentration ratio of 29 wt%/21 wt% to enhance CO2 cap- (1)

ture performance [17,22]. (2)
CO2 loading refers to the amount of carbon dioxide chemically ab- (3)
sorbed per mole of amine solvent and is typically expressed in mol
(4)
CO2/mol amine. This value depends strongly on process conditions,
including solvent flow rate, absorber temperature and pressure, and (5)
the targeted CO2 capture rate. For instance, under typical post- (6)
combustion capture conditions, the CO2 loading capacity of a 30 wt% (7)
MEA solution is reported to be approximately 0.2 mol CO2/mol amine (8)
[23]. In contrast, blended amines such as MDEA/PZ generally exhibit
(9)
lower CO2 loading under the same conditions, with values around
0.09 mol CO2/mol amine [21]. These values can vary depending on (10)
operational setup, and therefore, any comparison must consider con- (11)
sistent process boundaries. In this study, we report solvent loading (12)

5
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
Fig. 4. Reboiler duty of the stripper vs. different relative distances of intercooler location for (a) MEA system and (b) the MDEA/PZ system; (c) reboiler duty of the
stripper vs. different reflux ratios of MEA and MDEA/PZ systems.
TE
EC
RR

Fig. 5. Selection of optimal feed position for (a) MEA (split ratio = 0.1) and MDEA/PZ splitting (split ratio = 0.1), (b) optimal split ratio for MEA (relative dis-
tance = 0.6) and MDEA/PZ (relative distance = 0.3).
CO

is used for calculating the reaction rates in this study. The pre-
(13) exponential factor is expressed in appropriate units depending on the
(14) overall reaction order to ensure dimensional consistency of the reaction
(15) rate ( ). Eq. (17), which uses mole fractions ( ), is presented for
completeness but was not employed in our simulations. This distinc-
(16) tion is important for ensuring accurate representation of kinetics
within the Aspen Plus environment.
The kinetic parameters of MEA system correspond to Eqs. (7)–(10),
derived from rate expressions calculated using activity coefficients (17)
[24]. Similarly, the kinetic parameters of MEA/PZ system correspond to
Eqs. (11)–(16), based on the findings of Bishnoi and Rochelle [25,26]
(Table 2). The rate expressions for reaction in Eqs. (11)–(16) are deter- (18)
mined using molar concentration per unit volume (Eqs. (17) and (18)).
However, only Eq. (18), which is based on species concentrations ( ),

6
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
Fig. 6. Selection of the best location for flue gas splitting: (a) MEA, (b) MDEA/PZ; and optimal split ratio for flue gas splitting: (a) MEA, (b) MDEA/PZ.
TE

The packing height of the absorber and stripper columns is deter- in Fig. 1b. This method is employed to determine the optimal packing
mined using the infinite packing height method proposed by Madeddu heights for both MEA and MDEA/PZ processes.
et al. [27]. This approach specifies the column diameter by setting the The optimal packing heights for absorption and stripping columns
operating point at 80 % of the flooding point. An initial assumption of a are first determined. The regenerated high-temperature amine solution
column height of 100 m is made, followed by an increase in column from the stripping column then undergoes heat exchange with the CO2-
EC

height as needed to achieve the desired performance, with the solvent loaded low-temperature amine solution in the heat exchanger, facilitat-
flow rate adjusted to capture 90 % of the CO2 from the flue gas. The ing preheating and cooling. The absorption and stripping column simu-
packing height is progressively reduced, and the liquid-to-gas (L/G) ra- lations, conducted using the Radfrac® module, assume a stage height of
tio plot against packing height is constructed. The minimum solvent 0.3 m. Regardless of MEA or MDEA/PZ solvents utilized in the carbon
flow rate is identified, and the corresponding packing height is deter- capture process, the absorption column is divided into two sections of
mined as the optimal value, as depicted in Fig. 1a. equal diameter and height to ensure the column diameter remains
The packing height of the absorber and stripper columns is deter- within the commercial limit of 15 m. Table 3 presents the detailed con-
RR

mined using the infinite packing height method proposed by Madeddu figuration parameters of the absorption and stripping columns, along
et al. [27]. In this approach, the column diameter is specified by setting with the composition and flow rates of the circulating amine solutions.
the operating point at 80 % of the flooding point. For the absorber col- Fig. 1(a) and 1(b) demostrate that the the liquid-to-gas (L/G) ratio and
umn, an initial assumption of an infinitely tall column (e.g., 100 m) is reboiler energy requirement exponentially decreases along with pack-
made, with the solvent flow rate adjusted to capture 90 % of the CO2 ing height increase. The MDEA/PZ solvent shows better performances
from the flue gas. The packing height is then progressively reduced, and than MEA solvent with the optimum packing hight of 20 m and 7 m for
CO

a plot of the liquid-to-gas (L/G) ratio against packing height is con- the absorber and stripper, respectively.
structed. The minimum solvent flow rate is identified, and the corre-
sponding packing height is determined as the optimal value, as de- 2.4. Direct contact cooler
picted in Fig. 1a. The lean loading was specified as 0.2 mol CO2/mol
amine for MEA [23], and 0.09 mol CO2/mol amine for MDEA/PZ [21], The temperature of flue gas exiting a thermal power plant can reach
consistent with values reported in the literature for stable operation and as high as 104 °C. Since the reaction involved in CO2 absorption is
optimal energy efficiency. These values were selected to reflect realistic exothermic, the operating temperature of the absorber is typically re-
industrial conditions and to ensure meaningful comparison across all duced to around 40 °C before the flue gas enters the absorber for reac-
designs. tion with the solvent. To achieve this temperature reduction, a direct
A similar procedure is applied to the stripper column, where an ini- contact cooler (DCC) is employed to lower the flue gas temperature to
tially assumed infinite packing height is gradually reduced. In this case, 30 °C. Subsequently, the flue gas is pressurized to approximately
a plot of reboiler energy consumption versus packing height is gener- 1.1 bar using a blower; this compression process results in a moderate
ated, allowing for the identification of the minimum reboiler energy re- increase in temperature to around 40 °C.
quirement and its corresponding optimal packing height, as illustrated Due to the substantial flow rate of the flue gas, it is divided into
seven streams, each directed to one of seven identical DCC units. This

7
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 7. Membrane separation of CO2: (a) membrane stage cut vs. membrane area, (b) CO2 capture rate vs. CO2 purity, (c) CO2 capture rate vs. energy consumption, (d)
CO2 capture rate vs. membrane area.

division ensures that the diameter of each tower does not exceed the [29], with the specific parameters as presented in Table S3. The
EC

maximum commercially feasible size, which was limited to 15 m based H2O/CO2 selectivity is based on Chang’s hypothesis [30], while the se-
on fabrication, transportation, and operational constraints [28]. If only lection of O2/N2 selectivity is derived from Mat [31]. The configuration
one DCC were used to treat the entire flue gas stream, the required di- involves a gas phase on the feed side (flue gas) and a liquid sweep phase
ameter would exceed 26 m, which is well beyond typical commercial on the permeate side to maintain a driving force for CO2 transport and
limits and would introduce significant challenges in terms of uniform enhance separation efficiency. The use of a liquid phase on the perme-
spray distribution and structural stability. Reducing the number of ate side helps absorb the permeated CO2, reduce partial pressure, and
RR

units to two would still require each tower to have a diameter of ap- improve overall flux through the membrane. This setup is commonly
proximately 18.4 m, which remains impractical for field erection and applied in gas–liquid membrane contactors, where phase interaction
standard vendor offerings. Even with three DCCs, each unit would need enhances separation performance, especially under vacuum conditions.
to be around 15.0 m in diameter, technically feasible but at the upper Membrane Technology and Research (MTR) has developed the Po-
bound of commercial applicability, and potentially problematic for laris™ membrane module, as depicted in Fig. S3, which consists of
maintaining spray uniformity and efficient gas–liquid contact. There- stacked membranes with hollow interlayers, rolled into a cylindrical
fore, the use of seven parallel DCC units with 10 m diameters was se- configuration. Openings on the membrane surface facilitate countercur-
CO

lected as a balanced and scalable solution within practical design limits. rent flow, enhancing separation efficiency. Multiple membrane mod-
The DCC uses Raschig ring-metal as the packing material, and the tower ules are enclosed within a membrane vessel (Fig. S4), while several
diameter is determined based on operating conditions fixed at 80 % of membrane vessels are integrated into a membrane skid, which com-
the flooding point. As shown in Fig. S1, the cooling water flow rate prises various vessels, pipelines, valves, and designated locations for
reaches its minimum value when the packing height of the DCC reaches vacuum pumps or compressors, as illustrated in Fig. S5. According to
10 m. The detailed simulation settings are provided in Table S1, and the the report, MTR targets a membrane skid cost of $50/m2, significantly
specific results of the flue gas after cooling in the DCC are summarized lower than the $500/m2 cost typically associated with membrane skids
in Table S2. in the petrochemical industry for natural gas separation or hydrogen
production. The higher cost in these applications is attributed to the el-
2.5. Membrane CO2 separation evated pressures required for separation. In contrast, CO2 separation
from flue gas using membranes generally operates at pressures not ex-
The gas-phase membrane was modeled as a continuous medium, ceeding 3 atm, allowing for the use of cost-effective materials to con-
while the liquid-phase membrane was discretized into five layers, each struct membrane vessels, pipelines, and valves.
constituting one-fifth of the total thickness. The Polaris™ Advanced Building upon the membrane technology advancements by Mem-
film (Fig. S2), recognized as a highly developed membrane, was utilized brane Technology and Research (MTR), this study further explores

8
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 8. Energy consumption of CO2 capture system integrated with membrane separation vs. (a) membrane stage cut for MEA-based system, (b) membrane stage
cut for MDEA/PZ −based system, (c) YCO2 (mol%) for MEA-based system, (d) YCO2 (mol%) for MDEA/PZ-based system.
EC

membrane-based CO2 separation through simulation. The membrane The air-sweep mode builds upon the counterflow configuration by
model was developed using Aspen Custom Modeler® to simulate the introducing an air stream opposite to the permeate flow, further en-
cell structure, dividing the membrane into 500 computational units, as hancing the CO2 concentration gradient while reducing CO2 accumula-
illustrated in Fig. S6(a). To estimate the energy consumption for flue tion on the permeate side. The governing equations for this mode are
gas compression and cooling, the Peng-Robinson thermodynamic provided in Eqs. (22)–(24). The model assumes isothermal operation,
model in Aspen Plus® was utilized. The gas transport mechanism is dri- ideal gas behavior, constant species permeability, and negligible pres-
RR

ven by the pressure gradient between the high-pressure retentate side sure loss on both the retentate and permeate sides. Additional details,
and the low-pressure permeate side, with the membrane demonstrating including the Aspen Custom Modeler® code, are available in the Ap-
selective permeability for different gas species, thereby facilitating sep- pendix.
aration.
The membrane model was analyzed in three flow configurations: (22)
crossflow, counterflow, and air-sweep. In crossflow mode, the CO2 con- (23)
CO

centration in the retentate gradually decreases as the feed gas moves

along the membrane surface, while the CO2 concentration on the per- (24)
meate side remains relatively stable, as shown in Fig. S6(b). The corre-
sponding mathematical formulations are detailed in Eqs. (19) to (21). 3. Results and discussion
In counterflow mode, the CO2 concentration in the retentate similarly
declines with increasing membrane area; however, the permeate flow This section presents the results of the simulation study and pro-
moves in the opposite direction to the feed gas, creating a larger overall vides a critical analysis of the system's performance under various con-
CO2 concentration gradient across the membrane, as depicted in Fig. S6 ditions. It begins with the validation of the developed model by com-
(c). paring simulation outputs with existing literature and reference data to
ensure accuracy. Sensitivity analysis is then conducted to evaluate the
(19) influence of key operational parameters on CO2 capture efficiency and
(20) energy consumption. Building on these findings, an optimized process
design is proposed to enhance energy efficiency, followed by an eco-
(21) nomic analysis to assess the financial viability and cost-effectiveness of
the integrated CO2 capture system.

9
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 9. Energy consumption of CO2 capture system integrated with membrane separation vs. YCO2 (mol%) for (a) MEA-based system with (flue gas) split ra-
tio = 0.8, (b) MEA-based system with (flue gas) split ratio = 0.9, (c) MDEA/PZ-based system with (flue gas) split ratio = 0.8, (d) MDEA/PZ-based system with
(flue gas) split ratio = 0.9.
EC

3.1. Model validation the same region was higher. This discrepancy is attributed to the Rad-
frac® module's assumption of perfectly adiabatic tower walls, leading
3.1.1. CO2 capture using MEA solvent to an overestimating temperature and underestimating CO2 concentra-
Aspen Plus software is employed to simulate chemical absorption tion in this section. Despite these variations, the overall trends and the
processes, applying the Electrolyte Non-Random-Two-Liquid (ELECN- composition of the streams after CO2 capture were consistent with the
RR

RTL) model for thermodynamic calculations. MEA is currently the most experimental results. A detailed comparison of the simulation and ex-
widely used and extensively studied solvent in chemical absorption perimental data is presented in Table 4.
processes. The thermodynamic model ELECNRTL, built into Aspen Furthermore, Fig. 3b and 3d compare the simulation and experi-
Plus®, is employed to simulate the vapor–liquid equilibrium (VLE) of mental results for the stripper tower, demonstrating a close alignment
the MEA-H2O system. The ELECNRTL model integrates with the gas- between the CO2 concentration and temperature distributions. This
phase Redlich − Kwong (RK) model to predict the behavior of multi- study's solvent composition comprised 30 wt% MEA and 70 wt% H2O.
CO

component non-ideal solution systems, such as CO2−Amine−H2O. This

model has been validated against experimental data by Zhang et al. 3.1.2. CO2 capture using MDEA/PZ solvent
[24]. As shown in Fig. 2, the ELECNRTL model accurately simulates the The gas–liquid equilibrium of MDEA and PZ in the MDEA/H2O sys-
VLE of the MEA-H2O system, with the simulation results closely match- tem was simulated using the ELECNRTL property method. As depicted
ing the experimental data [32]. in Fig. 2b, the simulation results strongly agree with the experimental
To evaluate the accuracy of the absorber and stripper towers in the data points [34]. Similarly, for the PZ/H2O system, Fig. 2c demon-
chemical absorption process simulated with MEA using the Radfrac® strates a close correlation between the simulation results and experi-
module, the simulation results were compared with experimental data mental data [35]. However, no publicly available experimental data
from Test A11 conducted at the pilot plant of the University of Kaiser- from pilot plants were available for the chemical absorption process uti-
slautern [33]. Fig. S7 depicts the process flow diagram for Test A11, lizing MDEA/PZ as the solvent. As a result, the MDEA/PZ system was
while Table S4 summarizes the test parameters and equipment specifi- not validated for the absorber and stripper. This study's solvent compo-
cations of the pilot plant. sition consisted of 29 wt% MDEA, 21 wt% PZ, and 50 wt% H2O
By Fig. 3a and 3c, the simulation results indicated a lower CO2 con- [17,22]. In non-ideal solution systems, such as the CO2–MEA–H2O
centration distribution in the middle section of the absorber tower com- (three-component) and CO2–MDEA–PZ–H2O (four-component) sys-
pared to the experimental data, whereas the temperature distribution in tems, accurate thermodynamic modeling is essential to represent the

10
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 10. Energy consumption of MDEA/PZ-based system vs. YCO2 (mol%) using (a) counter flow mode of membrane separation, (b) cross flow mode of membrane
separation, (c) counter flow mode of membrane separation at pressure ratio = 7.5, (d) counter flow mode of membrane separation at pressure ratio = 5.
EC

complex phase behavior and interactions. The ELECNRTL model, com- creases the circulating solvent. The increase of circulating flow rate in
bined with the Redlich-Kwong (RK) equation of state, has been shown the packing may induce flooding, so the increase of column diameter is
to effectively describe the non-ideal behavior of the CO2–MEA–H2O sys- required to keep a safe and efficient operation.
tem and the gas–liquid equilibrium during the CO2 absorption process Notably, reflux section height follows conventional absorber inter-
[24]. cooling design, where the reintroduction point is placed slightly above
the withdrawal point to enhance the temperature gradient and promote
RR

3.2. Sensitivity analysis mass transfer efficiency. The intention is to reduce the solvent tempera-
ture before it continues upward, improving CO2 absorption capacity in
3.2.1. Internal cooler installed in the absorber the upper section of the column.
In the chemical absorption process, the reactions between CO2 and
alkanolamines are exothermic, necessitating temperature control 3.2.2. CO2-rich solvent splitting
within the absorption column. To maintain a lower operating tempera- In the chemical absorption process, CO2-rich amine solvents must
ture, a side internal cooler was integrated into the absorption column undergo regeneration for cyclic reuse. The regeneration reaction, which
CO

(named absorber intercooler), with its outlet temperature set at 40 °C. is the reverse of CO2 absorption (desorbtion) with amines, is endother-
The optimal placement of the absorber intercooler was determined mic. Given that amine, solvents contain a substantial amount of water,
based on the analysis presented in Fig. 4a and 4b. At a fixed reflux ratio the regeneration temperature at a fixed pressure of 1.8 bar is 114 °C for
of 0.1, the most effective position for MEA system at a relative distance the MEA process and 120 °C for the MDEA/PZ process, which means
from the top of the column is 0.6, corresponding to an actual height of that both temperatures are over to the boiling point of water (100 °C).
6.8 m. In contrast, the optimal placement for the MDEA/PZ system is at Consequently, water evaporation (with a relatively small amount of
a relative height of 0.8, equivalent to 3.4 m. Notably, the reflux ratio is amine) causes solvent losses, while high energy consumption in re-
defined as the proportion of the solvent flow rate entering the internal boiler duty is required.
cooler relative to the total solvent flow rate within the absorber. As To enhance energy efficiency, this study proposes a partial splitting
shown in Fig. 4c, the increase of reflux ratio from 0.1 to 0.8 slightly re- strategy for the CO2-rich amine solvent before preheating in the heat
duces the reboiler duty of the stripper. To simplify the operating para- exchanger. In this approach, the unheated portion of the solvent is in-
meters, the reflux ratio was fixed at 0.1 in this analysis to provide a con- troduced at the top of the stripper column, where the evaporated water
trolled basis for comparing the intercooling position across different vapor and a minor amount of amine vapor recondense, releasing heat
solvents. The absorber’s diameter is not affected by adding the inter- that is subsequently recovered by the unheated solvent. This heat re-
cooling, but it is affected by reflux ratios. The increase of reflux ratio in-

11
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE

Fig. 11. (a) CO2 capture rate versus CO2 purity, (b) vacuum pump energy consumption versus CO2 capture rate, (c) membrane area versus CO2 purity before perme-
ate mixing under different pressure ratios.
EC
RR
CO

Fig. 12. Conventional process design of carbon capture using absorption–desorption (Design-1).

12
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
Fig. 13. Chemical absorption process with an absorber intercooler (Design-2).
D
TE
EC

Fig. 14. Energy consumption impact of solvents in Design-2: (a) splitting position in energy-saving, (b) (Richout) split ratio.
RR

covery mechanism helps reduce reboiler energy consumption. The pre- sumption at a split ratio of 0.6 (Fig. 6c). In contrast, MDEA/PZ shows
heated solvent is then introduced at a lower section of the column. relatively lower reboiler energy consumption at a split ratio of 0.8 (Fig.
By Fig. 5a and 5b, the optimal feed position for MEA is at a relative 6d). However, overall, the energy consumption for MDEA/PZ with gas
distance from the top of the column is 0.6 which is located at actual splitting remains higher than that without splitting.
height of 6.8 m. For the MDEA/PZ system, the optimal positions at a For the sole adjustment of flue gas splitting, achieving lower energy
relative distance from the top of the column is 0.3 (actual height loca- consumption in the reboiler is not feasible. Therefore, this study adopts
CO

tion = 11.9 m). The solvent split ratio is defined as the ratio of the flow membrane separation technology to enrich CO2 from the flue gas,
rate of the unheated solvent entering the top of the stripper column to thereby increasing the CO2 absorption capacity in the absorber and re-
the flow rate of the solvent entering the heat exchanger for preheating. ducing reboiler energy consumption. The membrane separation process
Fig. 5c and 5d indicate that the optimal split ratio for MEA is 0.2 (actual enables control over CO2 concentration, membrane area utilization,
height: 13.6 m), while for MDEA/PZ, the optimal split ratio is 0.25. and energy consumption of vacuum pumps or compressors by adjusting
the split ratio (Eq. (25)) and stage cut (Eq. (26)), at which Among them,
3.2.3. Flue gas splitting and membrane separation/purification FP is the permeation flow rate, FR is the retentate flow rate, PP is the
To further reduce the energy consumption of the reboiler, this study permeation pressure, PR is the retentate pressure, Fbottom is the bottom
investigates the splitting of flue gas entering the absorber. A fixed pro- flue gas flow rate and Ftotal is the total flue gas flow rate.
portion of 0.05 of the flue gas from the bottom of the absorber is di-
verted to various positions within the absorber for comparison. In Fig. (25)
6a and 6b, it can be observed that both MEA and MDEA/PZ achieve
lower reboiler energy consumption when the relative position of the
split point is 0.8 (from the top of the column downward). As the split ra-
tio gradually increases, MEA exhibits the lowest reboiler energy con-

13
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
Fig. 15. Chemical absorption process integrated with membrane separation (Design-3).

ergy consumption at a split ratio of 0.8, is lower than that at a split ratio
D
(26) of 0.9 for MEA and MDEA/PD-based systems, respectively. It is noted
that the reduction in reboiler energy consumption for MEA is less signif-
icant than that observed for the MDEA/PZ system. Considering the
The operating conditions for the membrane separation process are combined energy consumption of the reboiler and vacuum pump, MEA
TE

fixed at a pressure ratio of 10, a permeability of 3000 GPU, and a does not achieve lower energy consumption compared to the design
CO2/N2 selectivity of 50. In this configuration, all flue gas is pressurized without membrane separation. Consequently, our following three de-
by a vacuum pump before being sent to the membrane separation unit signs for energy-saving exclude the MEA process from further consider-
for CO2 enrichment. Fig. 7a demonstrates that as the stage cut in- ation or comparison. implementing a more efficient membrane reduces
creases, more CO2 permeates through the membrane, the CO2 concen- total energy consumption, including thermal and electrical energy,
tration on the permeate side decreases (Fig. 7b), the energy consump-
EC

within the membrane-series absorption tower compared to the conven-

tion of the vacuum pump rises (Fig. 7c), and the required membrane tional process.
area increases significantly Fig. 7d. Meanwhile, the influence of flow The influence of CO2 concentration on reboiler duty is more signifi-
patterns on membrane separation reveals that, when less CO2 perme- cant, with the highest CO2 concentration reaching 25 mol% at a split ra-
ates, the energy consumption of vacuum pumps in counterflow and tio of 0.8. Previous results indicated that these parameters were insuffi-
crossflow modes is lower than that in the air sweep mode. Conversely, cient to reduce the combined energy consumption of the reboiler and
when more CO2 permeates, the energy consumption in the counterflow vacuum pump in the MDEA/PZ process compared to conventional
RR

and air sweep modes is lower than in the crossflow mode. However, the processes. Therefore, it is assumed that future membrane materials will
CO2 concentration on the permeate side in both counterflow and cross- exhibit improved selectivity and permeability. The influence of mem-
flow modes is substantially higher than in the air sweep mode. brane selectivity and permeability on vacuum pump energy consump-
The flue gas, after being separated through the membrane, is di- tion and membrane area utilization shows that the benefit of reducing
rected such that the permeate stream with a high CO2 concentration is membrane area utilization decreases significantly when the permeabil-
transported to the bottom of the absorption column, while the retentate ity reaches 15,000 GPU. Based on these findings, it is assumed that a
stream with a lower CO2 concentration is sent to the middle of the ab-
CO

membrane with a CO2/N2 selectivity of 500 and a permeability of

sorption column (Fig. S8). Beyond a membrane stage cut of 0.2, the re- 15,000 GPU could be employed for the optimization of membrane sepa-
boiler and cooling energy consumption no longer decrease, whereas the ration processes in the future.
vacuum pump energy consumption continues to rise (Fig. 8a and 8b). Fig. 10a illustrates that the countercurrent mode results in lower
Consequently, the CO2 concentration separated from the flue gas vacuum pump energy consumption than the staggered flow mode. The
(YCO2, %mol) is controlled within a membrane stage cut range of countercurrent configuration was selected to evaluate membrane per-
0.1–0.2 (Fig. 8c and 8d). The permeate stream (high CO2 concentra- formance under varying pressure ratios. As depicted in Fig. 11, at a
tion) is directed to the bottom of the absorber, and a portion of the re- pressure ratio of 10, the CO2 concentration is higher, and the required
tentate is fed to the middle, while the remaining retentate bypasses the membrane area is smaller relative to pressure ratios of 7.5, 5, and 4.
membrane. The proportion of gas directed to each entry point of the ab- However, this condition also leads to increased vacuum pump energy
sorber was adjusted in proportion to the stage cut, maintaining an opti- consumption. Fig. 10c and 10d further demonstrate that at a pressure
mal balance for CO2 concentration in the feed gas. ratio of 7.5, with a controlled CO2 concentration of 45 mol%, the
The ratio of feed gas to the middle and bottom of the column is rep- combined reboiler and vacuum pump energy consumption is reduced
resented by the split ratio, as defined in Eq. (25). As shown in Figs. 9a- by 2.73 % compared to a conventional design. This energy-efficient
9d, a higher CO2 concentration in the feed gas to the middle of the col- configuration integrates a membrane series absorption column, which
umn leads to lower reboiler energy consumption, and the reboiler en-

14
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE
EC
RR

Fig. 16. The impact of the operating parameters to energy consumption in Design-3: (a) internal cooler position, (b) reflux ratio, (c) section height, (d) split position,
(e) split ratio.

will be utilized for flue gas separation feed. It should be noted that the ation, chemical manufacturing, and carbon capture, require innovative
percentage values displayed above the bars represent the relative strategies to optimize energy consumption while maintaining high effi-
CO

change in energy demand compared to the conventional CO2 capture ciency and sustainability. In the present study, three different process
design without use of the membrane separation (the bar with designs are arranged to investigate a more efficient process.
YCO2 = 9.29 mol%). For cooling duty, this includes the sum of con- Fig. 12 (Design-1) represents a conventional chemical absorption-
denser and cooler loads, while for heat and electricity duty, the values based carbon capture process. In this configuration, a heat exchanger
represent the aggregate energy requirements of the reboiler, pump, is positioned between the absorption tower and the stripper to facili-
blower, and membrane systems. Noted that the negative percentages tate internal heat recovery. The rich solvent flow (Richout) exiting
imply energy savings, whereas the positive percentages increase en- the absorber is preheated by the lean solvent (Leanout) from the
ergy consumption. stripper, thereby enhancing thermal efficiency and reducing overall
energy consumption. The internal heat exchanger in this study was
3.3. Process design for energy efficiency modeled using a minimum temperature difference (ΔTmin) approach,
following pinch analysis principles, with a set ΔTmin of 10 °C to en-
Energy efficiency has become a critical aspect of modern process de- sure thermodynamically feasible and realistic heat exchange [36].
sign, driven by the need to reduce operational costs, minimize environ- The hot streams, which include the rich amine solutions with temper-
mental impact, and enhance overall system performance. Industrial atures around 170 °C for MEA and 135 °C for MDEA/PZ, are effec-
processes, particularly in energy-intensive sectors such as power gener-

15
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
PR
D
TE
EC

Fig. 17. Water washing zone on: (a) Design 1 − MEA, (b) Design 1 − MDEA/PZ, (c) Design 2 − MEA, (d) Design 2 − MDEA/PZ, (d) Design 3 − MDEA/PZ.

Table 5
Energy consumption comparison of all process design.
RR

Item Design-1 Design-2 Design-1 Design-2 Design-3

(MEA) (MEA) (MDEA/PZ) (MDEA/PZ) (MDEA/PZ)

Cooler (MW) −58.37 −64.70 −19.89 −24.91 −21.76

Condenser −64.35 −3.37 −41.68 −5.61 −8.53
(MW)
Wash cooler −0.44 0.00 −0.01 −0.01 −0.00
CO

(MW)
Inter-cooler − −82.08 − −58.86 −55.17
(MW)
Membrane − − − − −28.43
cooler (MW)
Reboiler 265.16 226.65 190.77 166.06 159.83
(MW)
Blower (MW) 5.32 5.32 5.32 5.32 4.85
Pump (MW) 0.19 0.17 0.08 0.02 0.02
Vacuum pump − − − − 13.20
(MW)
Reb. duty 3.35 2.86 2.41 2.10 2.02
(GJ/tonne
CO2)
Fig. 18. Total energy consumption of MEA and MDEA/PZ in various designs.

16
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

Table 6
Equipment costs of all designs using MEA and MDEA/PZ solvents.
Design-1 (MEA) Design-2 (MEA) Design-1 (MDEA/PZ) Design-2 (MDEA/PZ) Design-3 (MDEA/PZ)

M$ % M$ % M$ % M$ % M$ %

Column shell 2.65 32.20 2.74 35.45 2.30 27.12 2.51 28.43 1.81 5.23
Packing 2.23 27.10 1.57 20.31 2.34 27.59 2.99 33.86 2.85 8.24
Heat exchanger 1.27 15.43 1.34 17.34 1.80 21.23 1.30 14.72 2.19 6.33
Pump 0.18 2.19 0.18 2.33 0.14 1.65 0.13 1.47 0.12 0.35
Vacuum pump − − − − − − − − 9.90 28.63

F
Blower 1.90 23.09 1.90 24.58 1.90 22.41 1.90 21.52 1.78 5.15
Membrane − − − − − − − − 15.93 46.07
Total 8.23 100 7.73 100 8.48 100 8.83 100 34.58 100

OO
tively utilized to preheat the cold lean amine streams within the sys- dried, compressed, and liquefied for storage. To minimize amine emis-
tem. sions in the treated gas, a water-wash section at the top of the absorber
The process captures CO2 from coal-fired power plant flue gas at ensures that residual MEA, MDEA, and PZ concentrations are kept be-
104 °C with a flow rate of 2,514 ton/hour. The gas is cooled to a con- low 0.75 ppm.
stant outlet temperature of 30 °C via DCC before entering downstream In Fig. 13 (Design-2) incorporates an internal cooler (intercooler)
units such as the membrane separation unit or the absorber. Two sol- within the absorption tower. By directing a portion of the cooler solvent
vent-based systems are considered: MEA (30 wt% MEA, 70 wt% H2O) to the top of a stripper, evaporation heat is released, proposing to re-

PR
and MDEA/PZ (29 wt% MDEA, 21 wt% PZ, 50 wt% H2O), both de- duce energy consumption. The internal cooler operates under the opti-
signed to achieve 90 % CO2 removal efficiency. The CO2-rich solvent is mized parameters established in Section 3.2.1, while the operating con-
then regenerated in a stripper, releasing CO2, which is subsequently ditions for solvent diversion are detailed in Fig. 14. The optimal diver-

Table 7
Operating costs other than utilities.
Design1MEA Design-2MEA Design-1MDEA/PZ Design-2MDEA/PZ Design-3MDEA/PZ
D
M$/year M$/year M$/year M$/year M$/year

Labor 0.30 0.31 0.30 0.31 0.32

Maintenance 1.73 1.71 1.74 1.76 3.05
TE

Insurance 0.69 0.68 0.70 0.70 1.22

Membrane replacement − − − − 0.76
Total 2.73 2.70 2.74 2.77 5.35

Table 8
Operating costs of Design-1 and Design-2 using MEA solvent.
EC

Using coal as fuel Using natural gas as fuel

Design-1 Design-2 Design-1 Design-2

M$/year % M$/year % M$/year % M$/year %

Cooler 0.13 0.48 0.14 0.61 0.13 0.16 0.14 0.21

RR

Condenser 0.14 0.53 0.01 0.03 0.14 0.18 0.01 0.01

Wash cooler 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inter-cooler 0 0 0.18 0.78 0 0 0.18 0.27
Reboiler 23.73 90.68 20.29 89.06 74.97 96.85 64.09 96.26
Blower 2.10 8.02 2.10 9.22 2.10 2.71 2.10 3.15
Pump 0.08 0.29 0.07 0.30 0.08 0.10 0.07 0.10
Total 26.17 100 22.78 100 77.41 100 66.58 100
CO

Table 9
Operating costs of all designs using MDEA/PZ solvent.
Item Using coal as fuel Using natural gas as fuel

Design-1 Design-2 Design-3 Design-1 Design-2 Design-3

M$/year % M$/year % M$/year % M$/year % M$/year % M$/year %

Cooler 0.04 0.22 0.07 0.41 0.09 0.41 0.04 0.08 0.07 0.14 0.09 0.17
Condenser 0.09 0.46 0.02 0.09 0.04 0.16 0.09 0.16 0.02 0.03 0.04 0.07
Wash cooler 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inter-cooler 0 0 0.32 1.82 0.34 1.48 0 0 0.32 0.64 0.34 0.63
Membrane cooler 0 0 0 0 0.17 0.76 0 0 0 0 0.17 0.33
Reboiler 17.08 88.29 14.86 85.46 14.31 62.57 53.94 95.97 46.94 94.89 45.19 84.08
Blower 2.10 10.86 2.10 12.08 1.91 8.38 2.10 3.74 2.10 4.25 1.91 3.56
Pump 0.03 0.16 0.02 0.14 0.02 0.10 0.03 0.06 0.02 0.05 0.02 0.04
Vacuum pump 0 0 0 0 5.21 22.79 0 0 0 0 5.21 9.69
Total 19.34 100 17.39 100 22.86 100 56.21 100 49.47 100 53.75 100

17
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

Table 10 To determine the optimal height of the water-wash section, graphs

Carbon rights fees for flue gas before and after carbon capture. of water-wash flow rate and cooling energy consumption as functions of
Stream Flue gas Clean gas the water-wash section height were plotted. From Fig. 17a and 17b, it
can be observed that the optimal height for the water-wash section in
Temperature (°C) 39.69 56.54 Energy-Saving Design-1 is 8 m for MEA and 10 m for MDEA/PZ. Simi-
Pressure (bar) 1.1 1
larly, Fig. 17c and 17d show that the optimal height for Energy-Saving
Flowrate (tonne/hr) 2270 2176.43
Design-2 is 3.9 m for MEA and 8 m for MDEA/PZ, while Fig. 17e indi-
Gas composition (wt%)
CO2 13.95 1.44 cates that the optimal height for Energy-Saving Design-3 is 7 m for
H 2O 2.62 11.51 MDEA/PZ. Notably, the volatility and chemical characteristics of MEA
O2 6.75 7.05 and MDEA/PZ solvents affect the water-wash heights. MEA is more

F
N2 76.67 80.00 volatile than MDEA/PZ, which results in a higher tendency for MEA to
Priod (Year) Carbon credit M$/year)
escape from the absorber column via the overhead gas stream. Conse-
2024 170.14 16.82
quently, the MEA-based system requires a shorter water-wash section to

OO
2025 209.40 20.70
2030 382.16 37.77 effectively mitigate solvent losses while balancing the energy require-
2035 507.80 50.19 ment for water cooling. In contrast, MDEA/PZ, though less volatile,
forms more stable carbamate and bicarbonate species that are more re-
Table 11 sistant to being stripped in the water-wash section. The volatility of
Total carbon emissions of MDEA/PZ program. MDEA in the MDEA/PZ blend remains stable across a range of loadings
under absorber feed conditions, provided that PZ is the dominant reac-
Item MEA solvent MDEA/PZ solvent
tive species [39]. However, increasing the PZ concentration beyond a

PR
Design-1 Design-2 Design-1 Design-2 Design-3 certain point can result in amine vapor losses or even solvent precipita-
tion [20]. To mitigate these potential losses and minimize emissions, a
Reboiler duty (MW)
taller water-wash section becomes necessary. This is particularly impor-
LP-steam 265.16 226.65 190.77 166.00 159.83
CO2 emission (tonne/year)
tant given the presence of PZ, which less volatile than MEA, can be
Coal 703440.36 601278.13 506103.43 440392.20 424021.01 more challenging to effectively remove without sufficient contact area
Natural gas 397154.13 339474.54 285740.03 248640.24 239397.26 in the wash section.
Residue CO2 233044.64 233044.64 233044.64 233044.64 233044.64 The energy consumption analysis for different designs using MEA
and MDEA/PZ solvents reveals improvements in efficiency across the
D
Table 12 designs. After incorporating the water-washing section into all process
Power generation costs of by MEA system. designs, the total energy consumption of each design is shown in Fig.
Carbon credit Without Design-1 Design-1 Design-2 Design-2 18. Additionally, Table 5 compares the energy usage of MEA-based and
MDEA/PZ systems. In MEA system, Design-2 demonstrates superior
TE

($/tonne) CCS coal NG($/kWh) coal NG($/kWh)

($/kWh) ($/kWh) ($/kWh) performance over Design-1. The cooler duty in Design-2 is indeed
slightly higher (−64.70 MW) compared to Design-1 (−58.37 MW), pri-
72.15 0.0946 0.0794 0.0830 0.0772 0.0804
162.06 0.1463 0.1000 0.0969 0.0955 0.0929 marily due to the increased thermal load caused by higher outlet tem-
215.34 0.1770 0.1122 0.1051 0.1063 0.1004 peratures from upstream equipment. However, this increase is offset by
555 0.3726 0.1898 0.1573 0.1755 0.1479 a substantial reduction in condenser duty, which drops from
EC

−64.35 MW in Design-1 to just −3.37 MW in Design-2. Additionally,

sion position is determined to be 0.4 for MEA and 0.5 for MDEA/PZ, Design-2 incorporates an intercooler duty of about −82.08 MW, which
with an optimal diversion ratio of 0.2 for both solvent-based systems. is absent in Design-1 and reduces reboiler duty from 226.65 MW to
Meanwhile, Fig. 15 (Design-3) integrates gas membrane separation 265.16 MW. The overall reboiler duty requirement per tonne of CO2
as a pre-treatment step. This configuration enhances the absorber's CO2 captured in Design-2 is 2.86 GJ/tonne CO2, lower than Design-1 (3.35
capture efficiency, reducing both circulating solvent and energy de- GJ/tonne CO2), indicating improved efficiency. Design-2 reduces elec-
mands for regeneration. The membrane unit uses a vacuum pump and a tricity and thermal (reboiler) duty by 14.23 % compared to basic design
RR

blower to create the necessary pressure difference, sending CO2-rich gas (Design-1_MEA).
to the absorber’s bottom and CO2-lean gas to its middle. Design-3 also In MDEA/PZ system, Design-3 demonstrates the most efficient en-
combines membrane separation for purification, an internal cooler ergy usage among the other designs. The cooler energy requirement is
within the absorption column, and solvent split enriched with CO2. The highest in Design-2 (−24.91 MW), while Design-3 (−21.76 MW) shows
operating parameters for the membrane separation are the same as a slight reduction compared to Design-1 (−19.89 MW). A significant re-
those of the optimal parameters described in Section 3.2.3. The operat- duction in condenser energy usage is observed in Design-2 (−5.61 MW)
CO

ing parameters for the intercooler are presented in Fig. 16, where the and Design-3 (−8.53 MW) compared to Design-1 (−41.68 MW). Both
optimal position of the cooler is 0.6 (Fig. 16a), the optimal reflux ratio Design-2 and Design-3 include an intercooler (−58.86 and −55.17, re-
is 0.7 (Fig. 16b), the optimal height of the reflux zone is 3.6 m (Fig. spectively), which is absent in Design-1. Additionally, Design-3 intro-
16c), the optimal split position is 0.3 (Fig. 16d), and the optimal split duces a membrane cooler (−28.43 MW) and a vacuum pump
ratio is 0.15 (Fig. 16e). (13.20 MW), both absent in the other designs. The reboiler duty pro-
According to the Air Pollution Control Act, the concentration of gressively decreases across the designs, with Design-3 achieving the
MEA in industrial exhaust gases must not exceed 3 ppm [37]. When the lowest value (159.83 MW), followed by Design-2 (166.06 MW) and De-
concentration of amine solvents in exhaust gases is less than 1 ppm, the sign-1 (190.77 MW). Consequently, the specific reboiler duty (SRD) re-
extra costs for replenishing amine solvents become negligible [38]. quirement per tonne of CO2 captured decreases from 2.41 GJ/tonne
Therefore, a water-wash section was installed in all designs. The water- CO2 in Design-1 to 2.10 GJ/tonne CO2 in Design-2 and further drops to
wash section was operated to ensure that the amine concentration in a minimum of 2.02 GJ/tonne CO2 in Design-3. This result represents re-
the clean gas after CO2 removal did not exceed 0.75 ppm. Precisely, the boiler duty saving of about 12.61 % in Design-2 and 10.63 % in Design-
concentrations of MEA and PZ in the gas were controlled at a maximum 3 compared to Design-1.Therefore, the SRD of Design-3 is 2.02 GJ/
of 0.75 ppm, while MDEA, being more water-soluble, had concentra- tonne CO2, which is lower than the values reported in the literature,
tions lower than 0.75 ppm. which were about 2.76 GJ/tonne CO2 [40], 2.24 GJ/tonne CO2 [20],

18
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

Table 13
Power generation cost MDEA/PZ procedure.
Carbon credit Without CCS Design-1 coal Design-2 coal Design-3 coal Design-1 NG Design-2 NG Design-3 NG
($/tonne) ($/kWh) ($/kWh) ($/kWh) ($/kWh) ($/kWh) ($/kWh) ($/kWh)

72.15 0.0946 0.0758 0.0742 0.0755 0.0809 0.0786 0.0798

162.06 0.1463 0.0920 0.0890 0.0899 0.0923 0.0892 0.0901
215.34 0.1770 0.1017 0.0977 0.0984 0.0991 0.0955 0.0963
555 0.3726 0.1630 0.15358 0.15295 0.1428 0.1361 0.1368

2.68 GJ/tonne CO2 [39], and 3.3 2.68 GJ/tonne CO2 [20]. The lower

F
SRD value of 2.02 GJ/tonne CO2 obtained in this study, compared to (30)
the higher values reported in the literature, can be attributed to several (31)

OO
key improvements in the process design and operating strategy. Design-
3 incorporates advanced energy-saving features such as an intercooler, (32)
membrane cooler, and vacuum pump, which collectively reduce the
thermal load on the reboiler. In addition, the process operates under op- The packing materials used in the absorption and stripping towers
timized conditions, particularly with respect to solvent circulation, lean are Mellapak 250X, with a cost of $500/m3/cite [42]. In contrast, the
loading, and stripper pressure, which help minimize energy consump- packing material utilized in the direct-contact cooling tower consists of
tion. Effective heat integration and recovery also contribute to lower metal Raschig rings, priced at $600/m3/cite [43]. The cost function of
sensible and stripping heat requirements. Moreover, the simulation fo- heat exchangers is categorized into the fixed head type for general cool-

PR
cuses on an idealized and optimized design scenario, while some refer- ing applications, represented by Eq. (34), and the kettle-type reboiler,
enced studies consider larger-scale industrial settings with additional represented by Eq. (35) [44]. It is assumed that the overall heat transfer
energy penalties. These combined factors result in a more energy- coefficient for all heat exchangers, except for the reboiler, is
efficient CO2 capture process and explain the lower SRD value reported 1000 W/m2·K, while that for the reboiler is 6000 W/m2·K. Substituting
in this work. these values into Eq. (33) yields the required heat exchange area, with
The analysis of energy consumption across different designs using the unit of measurement for Eqs. (34) and (35) expressed in square feet
MEA and MDEA/PZ solvents shows significant efficiency improve- (ft2).
ments. For the MEA system, Design-2 reduces reboiler duty by 14.6 %
D
(33)
compared to Design-1 (from 3.35 GJ/tonne CO2 to 2.86 GJ/tonne CO2).
Similarly, in the MDEA/PZ system, Design-2 achieves a 12.9 % reduc- (34)
tion in reboiler duty compared to Design-1 (from 2.41 GJ/tonne CO2 to
TE

2.10 GJ/tonne CO2), while Design-3 further lowers it by 16.2 % (down

to 2.02 GJ/tonne CO2). Design-3 for the MDEA/PZ system demonstrates (35)
the highest energy savings among all designs. The results indicate that
integrating additional cooling components, such as an intercooler and The cost function of the thin-film equipment is expressed as Equa-
membrane cooler, along with optimized operating conditions, signifi- tion (3)–(10), where represents the membrane area, and denotes
cantly reduces reboiler energy consumption and improves overall sys- the cost of the membrane, including the valve, pipeline, and support
EC

tem performance. structure, at a price of 50 $/m2 [45].

3.4. Economic analysis (36)

3.4.1. Equipment cost The equipment cost function of the compression device is defined as
By Eq. (27), the cost function from the literature can be converted to follows: for vacuum pumps and compressors, it follows Eq. (37), where
CV/CCV/C represents the cost of vacuum pumps ($700/kWe) [45] and
RR

the price level of the year 2020. Meanwhile, Eq. (28) provides the total
cost, including equipment and installation costs, at which is the compressors ($620/kWe) [46]. For blowers, the cost is determined ac-
CEPCI index in 2020, is the CEPCI index in the year of the document, cording to Eq. (38), where denotes the mass flow rate of flue gas,
is the equipment cost, and is the installation cost coefficient. and represents the pressure difference [42]. Similarly, for liquid
The related parameters of these equations are presented in Table S5. pumps, the cost function is given by Eq. (39), where Q corresponds to
the volumetric flow rate of the liquid [42].

(27)
CO

(37)

(38)
(28)

The cost function of the tower shell equipment is composed of Eqs.

(29)–(32). Tsai proposed dividing the tower shell into three layers, (39)
where the innermost and outermost layers are made of 1/8-inch stain-
less steel (SS) sheaths, while the middle layer consists of 3/8-inch car- The carbon capture process can be categorized into three main
bon steel, serving as the primary structural support for the tower's stages: the cooling stage utilizing a DCC, the chemical absorption-based
weight [41]. The steel density for stainless steel is 8000 kg/m3, while CO2 capture process, and the CO2 liquefaction stage. Since 90 % of the
that for carbon steel is 7850 kg/m3. The variable Z represents the tower CO2 in the flue gas is captured and liquefied, the configurations of the
height, S denotes the tower side height, and d refers to the thickness of DCC and CO2 liquefaction units remain fixed. The total equipment costs
the tower shell. for these units are estimated at $6.19 million for the DCC and $20.23
million for the chemical absorption and CO2 liquefaction process. A de-
(29) tailed breakdown of equipment costs is presented in Table S6, indicat-
ing that the primary cost component for the DCC arises from the water-

19
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
Fig. 19. Power generation costs of various designs under different carbon rights prices: (a) natural gas, (b) coal.

wash cooling tower, while the significant cost contribution in CO2 liq-

PR
uefaction originates from the compressor system. (40)
In the MEA-based process, the primary equipment cost is attributed
to the packed column (Table 6). When transitioning from Design-1 to
The operational costs associated with DCC and CO2 liquefaction are
Design-2 by incorporating an intercooler system within the absorber
presented in Table S8. For DCC, the primary cost arises from the water
and implementing rich CO2 splitting, the cost of the packed column de-
scrubbing process required to reduce the flue gas temperature from
creases from $4.88 million to $4.31 million. The packed column re-
104 °C to 30 °C, ensuring that the blower maintains the flue gas temper-
mains the primary cost component in the MDEA/PZ-based process
ature at 40 °C before entering the absorption tower. In the case of CO2
(Table 6). However, in Design-3, which integrates flue gas bypass and
D
liquefaction, the costs are primarily attributed to the electricity con-
membrane-based purification, the significant equipment costs shift to-
sumption of the compressor for pressurization and the cooling agent re-
wards vacuum pumps and membranes, reducing the packed column
quired for CO2 condensation.
cost from $5.5 million in Design-2 to $4.66 million.
The operating costs include electricity, low-pressure steam, cooling
TE

water, labor costs, maintenance costs, and insurance costs. The labor
3.4.2. Operating cost
costs are calculated based on Eqs. (41)–(43) [47]. The system operates
After the absorption of CO2 by the chemical solvent, regeneration of
for 310.25 days per year, with three shifts per day, corresponding to
the solvent is required through the application of heat, typically pro-
44.3 working weeks per year and five shifts per week. The annual salary
vided via a reboiler. In general, low-pressure steam (LP-steam) serves as
for shift workers is assumed to be $20,000. The total number of re-
the heating medium for the reboiler. In standalone carbon capture facil-
quired personnel, , is determined by considering the number of
ities or in industrial applications without integrated power systems, LP-
EC

personnel required per shift , the number of solid-processing units

steam is often produced externally, commonly by burning natural gas
(P), the number of non-solid-processing units ( ), and the number of
or coal, which can lead to additional CO2 emissions. These emissions
not only offset some of the benefits of the CO2 capture process but also shifts each worker is required to cover annually ( ). The total annual
introduce economic penalties under carbon taxation or credit systems. labor cost is obtained by multiplying the total number of personnel by
However, the carbon capture integration with coal- or gas-fired power the annual salary of $20,000.
plants is common practice to extract LP-steam directly from the plant’s The annual maintenance cost is assumed to be 5 % of the total
RR

steam cycle to supply the reboiler. This method eliminates the need for equipment cost, while the annual insurance cost is estimated at 2 %.
an additional steam generator, thereby reducing operating costs and The membrane replacement cost is incurred every five years, where the
avoiding extra fuel consumption and associated emissions. Utilizing ex- cost of membrane equipment already includes the membrane material;
tracted steam in this manner improves the overall thermodynamic and thus, only four additional replacements are required. The operating
economic efficiency of the capture system and is widely adopted in costs, excluding utility expenses, are summarized in Table 7.
studies and pilot projects involving power plant integration. Therefore,
(41)
CO

while external LP-steam generation may be relevant in standalone or

retrofitted industrial scenarios, steam extraction from the existing
power cycle remains the preferred strategy for power plant applica- (42)
tions, and should be considered in future integration and optimization
efforts.
The cost function of LP-steam is expressed in Eq. (40), where the (43)
boiler efficiency is assumed to be 90 %. The utility cost parameters are
detailed in Table S7. and the heating values 6322 kcal/kg for coal and 1 The operating costs associated with using coal and natural gas as
MMBTU of natural gas equivalent to 28.26368 m3, converted to fuel in the chemical absorption process with MEA is presented in Table
10.01654 $/MMBTU are sourced from the Taiwan Power Company's 8. The data indicate that the implementation of intercooling and rich
Fiscal Year 2020 Business Budget Assessment Report published by the splitting significantly reduces reboiler and condenser operating costs.
Legislative Yuan. Furthermore, it is assumed that the power generation Similarly, the operating costs for the MDEA/PZ-based chemical absorp-
output equals the installed capacity of 550 MW. The plant operates for tion process using coal and natural gas as fuel are summarized in Table
310.25 days per year (equivalent to 7446 h), corresponding to a capac- 9. The results show that while adding intercooler and rich splitting re-
ity factor of 85 %, with a projected operational lifespan of 25 years. duces reboiler and condenser costs considerably, the operating cost as-
sociated with intercooler exceeds these savings. Consequently, the total

20
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

F
OO
Fig. 20. Power generation costs of various designs under different carbon rights prices: (a) natural gas, (b) coal.

cost in Design-1 is higher than in Design-2, where the reboiler operates gas (where 1 MMBTU = 1055.1 MJ) [48]. Consequently, coal combus-
with natural gas as fuel. Furthermore, the cost of using coal as fuel is ap- tion produces approximately 1.77 times higher carbon emissions than

PR
proximately three times higher than that of natural gas.fuel. natural gas.

3.4.3. Impact of carbon credit trading (44)

Carbon credit trading is a market-based mechanism that reduces
greenhouse gas (GHG) emissions by allowing entities to buy and sell
The carbon credit costs before and after carbon capture are summa-
carbon credits. A carbon credit represents one metric ton of CO2 or its
rized in Table 10, while the projected carbon credit prices for each year
equivalent (CO2e) that has been reduced, removed, or avoided through
are provided in Table S9. It can be conclude that the carbon credit of
various emission reduction projects. The European Union Emissions
D
clean gas (flue gas after CO2 capture) is lower than the flue gas before
Trading System (EU ETS) is the most comprehensive global carbon
CO2 capture. The total carbon emissions from various carbon capture
emissions trading mechanism. Within this system, emission allowances,
processes are detailed in Table 11. Among the evaluated designs, De-
referred to as EU Allowances (EUAs), function as carbon credits that
sign-3 exhibits lower additional carbon emissions from the reboiler
TE

can be traded. The EU ETS establishes an annual cap on total carbon

than Design-1 and Design-2. The cost savings in carbon credits for De-
emissions, which is progressively reduced to meet the 2050 net-zero
sign-3 increase progressively with rising carbon credit prices relative to
emissions target. Any unused carbon credits acquired within a given
Design-2 (Fig. S10). Notably, when coal is utilized as the primary fuel
year can be carried forward for future use. The development of the EU
source, the resulting cost savings in carbon credits are significantly
ETS has been structured into four phases. The initial phase
higher than when natural gas is used. It is owing to the higher carbon
(2005–2007) primarily focused on establishing carbon inventory mech-
EC

emissions associated with coal combustion, which, when reduced or

anisms within the power sector and other energy-intensive industries.
offset, lead to more significant financial benefits in carbon credit trad-
During this period, most carbon credits were allocated free of charge,
ing or regulatory compliance programs. A comprehensive dataset sup-
where an oversupply of credits led to carbon prices collapsing. In the
porting these findings is presented in Tables S10 to S17.
second phase (2008–2012), the issuance of carbon credits was reduced;
To obtain the power generation cost after carbon capture, the coal-
however, the 2008 financial crisis resulted in a decline in industrial
fired power generation cost announced by Taiwan Power Company in
emissions, causing carbon prices to depreciate significantly. The third
2020 was calculated using Eq. (45), resulting in 0.053 USD/kWh (1.71
RR

phase (2013–2020) introduced a requirement for the power sector to

NTD/kWh). As of September 3, 2024, the exchange rate was 1 NTD =
purchase carbon credits rather than receiving free allocations. In the
0.031 USD. Given an installed capacity of 550 MW and an annual oper-
ongoing fourth phase (2021–2030), the EU ETS aims to reduce total
ation of 310.25 days (7,446 h), corresponding to a capacity factor of
emissions by 55 % by 2030, with a more accelerated reduction in car-
85 %, the annual electricity generation was determined to be 4.0953
bon credit availability [48].
billion kWh. Detailed results are presented in Tables 12 and 13.
This study utilizes forecasted EU ETS carbon credit prices as a refer-
ence. According to projections by Statista, the expected prices are USD
CO

72.15/tonne CO2 (EUR 65/tonne CO2) in 2024, USD 88.8/tonne CO2 (45)
(EUR 80/tonne CO2) in 2025, USD 162.06/tonne CO2 (EUR 146/tonne
CO2) in 2030, and USD 215.34/tonne CO2 (EUR 194/tonne CO2) in The price of carbon credits varies widely, ranging from under $1 to
2035 [48]. Additionally, a forecast by Enerdata suggests that by 2044, over $500 per tonne, depending on factors like project type, location,
the price may reach USD 555/tonne CO2 (EUR 500/tonne CO2). Based and quality standards. Some credits, particularly those from nature-
on the exchange rate as of September 19, 2024 (1 USD = 1.11 EUR), based solutions like REDD + projects, may be priced below $10 per
these values have been converted accordingly (Fig. S11). As the EU ETS tonne, while others, especially high-integrity removal credits, can reach
mandates net-zero emissions by 2050, carbon credit prices are pro- $146/ton or even $172/ton in the future [49]. This variability in car-
jected to rise significantly, particularly from 2044 onward. bon credit pricing has a direct impact on the economics of power gener-
The additional carbon emissions associated with using coal and nat- ation technologies. As shown in Fig. 19, the generation cost of natural
ural gas as fuels for LP-steam generation is quantified using Eq. (44). gas-fired power is higher than that of coal-fired power at low carbon
These calculations facilitate the evaluation of carbon emissions across credit prices. However, as the carbon credit price increases, this cost re-
different processes utilizing either coal or natural gas. In this analysis, a lationship reverses, making natural gas a more economical option. Ad-
boiler efficiency of 90 % is assumed, with carbon emission factors set at ditionally, across different carbon credit price scenarios, direct emis-
93.98 kg-CO2/MMBTU for coal and 53.06 kg-CO2/MMBTU for natural sions consistently result in significantly higher costs than scenarios in-

21
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

corporating carbon capture. Fig. 20 illustrates that after accounting for Declaration of competing interest
additional carbon emissions, the cost of carbon credits associated with
the remaining 10 % of uncaptured CO2, and the cost of the carbon cap- The authors declare that they have no known competing financial
ture process, MDEA/PZ Design-2 exhibits the lowest power generation interests or personal relationships that could have appeared to influ-
cost until the carbon credit price reaches $555/tonne. MDEA/PZ De- ence the work reported in this paper.
sign-3 becomes more cost-effective than MDEA/PZ Design-2 for coal-
fired power only when the carbon credit price exceeds $555/tonne. In Acknowledgments
contrast, for natural gas-fired power, the generation costs of Design-3
and Design-2 remain nearly identical. This work was financially supported by the National Science and
Technology Council, Taiwan under grants 113-2321-B-029-002, 113-

F
4. Conclusion 2622-8-006-020, and 113-2221-E-006-019.

This study used Aspen Plus simulations to evaluate and optimize Data availability

OO
post-combustion CO2 capture from a coal-fired power plant processing
2154 tonnes/hour of flue gas at 104 °C. The gas was cooled, pressur- Data will be made available on request.
ized, and treated in amine-based solvent, specifically MEA and MDEA/
PZ. The rich solvent underwent thermal regeneration in a stripper, and Appendix A. Supplementary data
the released CO2 was compressed and liquefied, yielding 284.79
tonnes/hour of liquid CO2. To enhance process performance, three ad- Supplementary data to this article can be found online at https://
vanced configurations were developed, incorporating absorber inter- doi.org/10.1016/j.applthermaleng.2025.127457.

PR
cooling, rich solvent split-feed regeneration, and membrane-based flue
gas pretreatment. References
The results indicate that energy consumption significantly varies
with process design and solvent choice. For MEA, Design-1 and Design- [1] T.N. Borhani, E. Oko, M. Wang, Process modelling and analysis of intensified
CO2 capture using monoethanolamine (MEA) in rotating packed bed absorber, J.
2 achieved reboiler duties of 3.35 and 2.86 GJ/tonne CO2, respectively. Clean. Prod. 204 (2018) 1124–1142, https://doi.org/10.1016/
MDEA/PZ showed superior performance, with Design-1 consuming j.jclepro.2018.09.089.
2.14 GJ/tonne CO2 and Design-2 reducing this to 2.10 GJ/tonne CO2. [2] H. Lepaumier, D. Picq, P.L. Carrette, Degradation study of new solvents for CO2
capture in post-combustion, Energy Procedia 1 (2009) 893–900, https://doi.org/
The most energy-efficient configuration, Design-3 (MDEA/PZ),
D
10.1016/j.egypro.2009.01.119.
achieved a reboiler duty of only 2.02 GJ/tonne CO2, lower than bench- [3] M.I. Taipabu, K. Viswanathan, W. Wu, R. Handogo, A. Mualim, H. Huda, Process
mark values reported in the literature (2.76 GJ/tonne CO2 [40], 2.24 Intensification New improvement of amine-based CO 2 capture processes using
heat integration and optimization, Chem. Eng. Process. - Process Intensif. 193
GJ/tonne CO2 [20], 2.68 GJ/tonne CO2 [39], and 3.3 2.68 GJ/tonne (2023) 109532, https://doi.org/10.1016/j.cep.2023.109532.
CO2 [20]). MEA was excluded from Design-3 due to its incompatibility
TE

[4] M. Qian, T. Xin, C. Xu, Y. Yang, A novel natural gas and coal hybrid power
with membrane pretreatment, resulting in excessive energy demands. generation system for efficient utilization of gas turbine exhaust and low efficiency
penalty of carbon capture, Appl. Therm. Eng. 269 (2025) 126174, https://doi.org/
Although Design-3 entails higher electricity consumption due to
10.1016/j.applthermaleng.2025.126174.
membrane operation and vacuum pumping, it becomes more economi- [5] R.H. Pan, Y.R. Chen, K.L. Tung, H. Chang, Experimental and simulation study of
cally favorable under future carbon pricing scenarios. Specifically, it a novel hybrid absorption and stripping membrane contactor for carbon capture, J.
outperforms Design-2 when the EU-ETS carbon credit price reaches Taiwan Inst. Chem. Eng. 81 (2017) 47–56, https://doi.org/10.1016/
j.jtice.2017.10.009.
EC

$162.06/tonne CO2 (projected for 2030), or $215.34/tonne CO2 if nat- [6] Z.H. Ban, L.K. Keong, A.M. Shariff, Physical absorption of CO2 capture: a review,
ural gas is used as the heat source (projected for 2035). From an elec- Adv. Mater. Res. 917 (2014) 134–143, https://doi.org/10.4028/
tricity generation cost perspective, Design-3 only becomes preferable www.scientific.net/AMR.917.134.
[7] O.H.P. Gunawardene, C.A. Gunathilake, K. Vikrant, Carbon dioxide capture
when carbon prices rise to $555/tonne (projected for 2044). Nonethe- through physical and chemical adsorption using porous carbon materials : a
less, when compared to the cost of purchasing carbon credits to offset review, 2022.
direct emissions, Design-3 remains a more economically viable long- [8] Z. Dai, L. Deng, Membranes for CO2 capture and separation: Progress in research
and development for industrial applications, Sep. Purif. Technol. 335 (2024)
term strategy.
RR

126022, https://doi.org/10.1016/j.seppur.2023.126022.
Future work should focus on experimental validation of these con- [9] R. Hou, C. Fong, B.D. Freeman, M.R. Hill, Z. Xie, Current status and advances in
figurations to confirm simulation accuracy under real-world operating membrane technology for carbon capture, Sep. Purif. Technol. 300 (2022) 121863,
https://doi.org/10.1016/j.seppur.2022.121863.
conditions. Additionally, investigations into long-term solvent degrada-
[10] C. Song, Q. Liu, S. Deng, H. Li, Y. Kitamura, Cryogenic-based CO2 capture
tion, environmental impacts, and integration with renewable energy technologies: state-of-the-art developments and current challenges, Renew.
sources (e.g., solar-assisted regeneration) could further enhance the Sustain. Energy Rev. 101 (2019) 265–278, https://doi.org/10.1016/
sustainability and feasibility of large-scale CO2 capture systems. Process j.rser.2018.11.018.
[11] M.C. Romano, M. Spinelli, S. Campanari, S. Consonni, G. Cinti, M. Marchi, E.
CO

intensification strategies such as heat recovery, advanced absorber Borgarello, The calcium looping process for low CO 2 emission cement and power,
packing, and hybrid solvent formulations are also recommended to fur- Energy Procedia 37 (2013) 7091–7099, https://doi.org/10.1016/
ther reduce operational costs and improve capture efficiency. j.egypro.2013.06.645.
[12] Y.A. Criado, Environmental science calcium looping CO 2 capture system for
back-up power plants, (2017) 1994–2004, doi: 10.1039/c7ee01505d.
CRediT authorship contribution statement [13] B. Arias, M. Alonso, C. Abanades, CO 2 capture by calcium looping at relevant
conditions for cement plants: experimental testing in a 30 kW, Ind. Eng. Chem. Res.
56 (2017) 2634–2640, https://doi.org/10.1021/acs.iecr.6b04617.
Muhammad Ikhsan Taipabu: Writing – original draft, Validation, [14] Y. Zheng, S. He, J. Liu, X. Zeng, B. Xu, W. Wang, G. Yang, Ammonia-enhanced
Methodology, Investigation. Bang Yan Chou: Validation, Software, calcium looping for carbon reduction in natural gas combined cycle plants, Appl.
Formal analysis, Conceptualization. Wei Wu: Writing – review & edit- Therm. Eng. 271 (2025) 126245, https://doi.org/10.1016/
j.applthermaleng.2025.126245.
ing, Supervision, Funding acquisition, Conceptualization. Bambang
[15] F.-L. Chang, C. Chengliang, Post-combustion membrane separation carbon
Sardi: Conceptualization, Formal analysis. capture procedure in coal-fired thermal power plants__Taiwan PhD and Master’s
Thesis Knowledge Added System, National Taiwan University, 2016, https://
hdl.handle.net/11296/qub32a.
[16] H. Svensson, C. Hulteberg, H.T. Karlsson, Heat of absorption of CO2 in aqueous
solutions of N-methyldiethanolamine and piperazine, Int. J. Greenh. Gas Control.
17 (2013) 89–98, https://doi.org/10.1016/j.ijggc.2013.04.021.
[17] P.T. Frailie, Modeling of carbon dioxide absorption/stripping by aqueous

22
M.I. Taipabu et al. Applied Thermal Engineering xxx (xxxx) 127457

methyldiethanolamine/piperazine, (2014) 1–272, https:// (methylamino)-propylamine, and their binary and ternary solutions, J. Chem. Eng.
repositories.lib.utexas.edu/handle/2152/25019. Data 53 (2008) 2521–2531, https://doi.org/10.1021/je800290k.
[18] M. of Environment, Fixed pollution source management information disclosure [35] H.F. Svendsen, Ø. Jonassen, A. Hartono, U. Shoukat, M. Francois, H.K. Knuutila,
platform, Minist. Environ. (n.d.), https://aodmis.moenv.gov.tw/opendata/#/ab/3. I. Kim, Vapor liquid equilibrium of pure and aqueous methyl diethanolamine
[19] I. Kim, H.F. Svendsen, Heat of absorption of carbon dioxide (CO2) in (MDEA), 2-dimethylmonoethanolamine (DMMEA), N-methylmonoethanolamine
monoethanolamine (MEA) and 2-(aminoethyl)ethanolamine (AEEA) solutions, Ind. (MMEA) and 1-(2-aminoethyl)piperazine (AEP): experimental results and
Eng. Chem. Res. 46 (2007) 5803–5809, https://doi.org/10.1021/ie0616489. modeling, Fluid Phase Equilib. 571 (2023) 113811, https://doi.org/10.1016/
[20] B. Zhao, F. Liu, Z. Cui, C. Liu, H. Yue, S. Tang, Y. Liu, H. Lu, B. Liang, Enhancing j.fluid.2023.113811.
the energetic efficiency of MDEA/PZ-based CO2 capture technology for a 650MW [36] M.R.M. Abu Zahra, Carbon dioxide capture from flue gas, Technische
power plant: Process improvement, Appl. Energy 185 (2017) 362–375, https:// Universiteit Delft (2009).
doi.org/10.1016/j.apenergy.2016.11.009. [37] E.P. Department, Emission standards for air pollutants from stationary sources,
[21] L. Mengxiang, L. Haizhou, P. Haizhon, F. Aiguo, Simulation and analysis of 2010.
carbon dioxide capture process with split flow modification using MDEA/PZ blend [38] G.T. Rochelle, Air pollution impacts of amine scrubbing for CO2 capture, Carbon

F
solution in a coal – fired power plant, Chem. Ind. Eng. Prog. 38 (2019) 2046–2055. Capture Sci. Technol. 11 (2024) 100192, https://doi.org/10.1016/
[22] B.A. Khan, A. Ullah, M.W. Saleem, A.N. Khan, M. Faiq, M. Haris, Energy j.ccst.2024.100192.
minimization in piperazine promoted mdea-based co2 capture process, Sustain 12 [39] S.M. Hosseini-Ardali, M. Hazrati-Kalbibaki, M. Fattahi, F. Lezsovits, Multi-
(2020) 1–13, https://doi.org/10.3390/su12208524. objective optimization of post combustion CO2 capture using

OO
[23] H. Jin, P. Liu, Z. Li, Energy-efficient process intensification for post-combustion methyldiethanolamine (MDEA) and piperazine (PZ) bi-solvent, Energy 211 (2020)
CO2 capture: a modeling approach, Energy 158 (2018) 471–483, https://doi.org/ 119035, https://doi.org/10.1016/j.energy.2020.119035.
10.1016/j.energy.2018.06.045. [40] H. Lin, H. Luo, A. Pei, M. Fang, Simulation and analysis of carbon dioxide capture
[24] Y. Zhang, H. Que, C.-C. Chen, Thermodynamic modeling for CO2 absorption in process using MDEA/PZ blend solution in a coal-fired power plant, Chem. Ind.
aqueous MEA solution with electrolyte NRTL model, Fluid Phase Equilib. 311 Prog. 38 (2019) 2046–2055.
(2011) 67–75, https://doi.org/10.1016/j.fluid.2011.08.025. [41] R.E. Tsai, Mass transfer area of structured packing, University of Texas (2010).
[25] S. Bishnoi, G.T. Rochelle, Absorption of carbon dioxide into aqueous piperazine: [42] Boegger, structured packing – high absorbing, distilling & separating efficiency,
reaction kinetics, mass transfer and solubility, Chem. Eng. Sci. 55 (2000) Boegger Industech Ltd., 2024. https://www.structuredpacking.org/structured-
5531–5543, https://doi.org/10.1016/S0009-2509(00)00182-2. packing/structured-packing.html?campaignid=18224202444&adgroupid=

PR
[26] S. Bishnoi, G.T. Rochelle, Absorption of carbon dioxide in aqueous piperazine/ 140400615523&feeditemid=&targetid=kwd-1911603075025&device=c&
methyldiethanolamine, AIChE J. 48 (2002) 2788–2799, https://doi.org/10.1002/ creative=619610729963&keyword=mellapak 250x&gad_source=1&gclid=
aic.690481208. Cj0KCQjwy46_BhDOARIsA (accessed April 28, 2024).
[27] C. Madeddu, M. Errico, R. Baratti, Process Modeling in Aspen plus® (2019), [43] FXSINO, metal random packing, metal pall ring, metal flat ring, Pingxiang Fxsino
https://doi.org/10.1007/978-3-030-04579-1_2. Petrochemical Pack. Co., Ltd. (n.d.), https://www.alibaba.com/product-detail/10-
[28] T. Nittaya, P.L. Douglas, E. Croiset, L.A. Ricardez-Sandoval, Dynamic modeling 89mm-Metal-Super-Raschig-Ring_1600335201451.html?spm=
and evaluation of an industrial-scale CO2 capture plant using monoethanolamine a2700.galleryofferlist.p_offer.d_title.48eb283duYjvTO&s=p (accessed April 28,
absorption processes, Ind. Eng. Chem. Res. 53 (2014) 11411–11426, https:// 2024).
doi.org/10.1021/ie500190p. [44] D.R. Seider, W.D. Seader, J.D. Lewin, Product and process design principles -
[29] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-combustion carbon synthesis, analysis, and evaluation, 2003.
D
dioxide capture: an opportunity for membranes, J. Memb. Sci. 359 (2010) [45] T. Merkel, J. Kniep, X. Wei, T. Carlisle, S. White, S. Pande, D. Fulton, R. Watson,
126–139, https://doi.org/10.1016/j.memsci.2009.10.041. T. Hoffman, B. Freeman, R. Baker, Pilot testing of a membrane system for
[30] F.L. Chang, Membrane separation process for post-combustion co2 capture from postcombustion CO2 capture, United States (2015), https://doi.org/10.2172/
coal-fired power plant, 國立臺灣大學 (2016), https://doi.org/10.6342/ 1337555.
NTU201601884. [46] K. Ramasubramanian, H. Verweij, W.S. Winston Ho, Membrane processes for
TE

[31] N.C. Mat, G.G. Lipscomb, Membrane process optimization for carbon capture, carbon capture from coal-fired power plant flue gas: a modeling and cost study, J.
Int. J. Greenh. Gas Control. 62 (2017) 1–12, https://doi.org/10.1016/ Memb. Sci. 421–422 (2012) 299–310, https://doi.org/10.1016/
j.ijggc.2017.04.002. j.memsci.2012.07.029.
[32] K. Tochigi, K. Akimoto, K. Ochi, F. Liu, Y. Kawase, Isothermal vapor−liquid [47] W.A. Alkhayat, A.M. Gerrard, Estimating manning levels for process plants,
equilibria for water + 2-aminoethanol + dimethyl sulfoxide and its constituent Trans. Am. Assoc. Cost. Eng. (1984).
three binary systems, J. Chem. Eng. Data 44 (1999) 588–590, https://doi.org/ [48] EU-ETS, The EU emissions trading system (EU ETS), (2016). Doi: 10.2834/6083.
10.1021/je980068i. [49] O. Catsaros, Carbon credits face biggest test yet, could reach $238/Ton in 2050,
[33] H.P. Mangalapally, H. Hasse, Pilot plant study of post-combustion carbon dioxide according to BloombergNEF report, BloombergNEF, 2024, https://
EC

capture by reactive absorption: methodology, comparison of different structured about.bnef.com/blog/carbon-credits-face-biggest-test-yet-could-reach-238-ton-in-

packings, and comprehensive results for monoethanolamine, Chem. Eng. Res. Des. 2050-according-to-bloombergnef-report/ (accessed June 1, 2025).
89 (2011) 1216–1228, https://doi.org/10.1016/j.cherd.2011.01.013.
[34] I. Kim, H.F. Svendsen, E. Børresen, Ebulliometric determination of vapor−liquid
equilibria for pure water, monoethanolamine, N-methyldiethanolamine, 3-
RR
CO

23