Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Design and Operation Optimisation of a MEA-based CO2

Capture Unit
Artur Andrade
artur.andrade@ist.utl.pt

Instituto Superior Técnico, Lisbon, Portugal

Supervisors: Prof. Dr. Carla Pinheiro; Dr. Javier Rodriguez

November 2014

Abstract

The present thesis has the objective of analysing the cost reduction obtained through a rigorous
model-based optimisation of a post-combustion CO2 capture plant for carbon capture and storage
applications. Presently, capture technology is mainly based on chemical absorption with alkanolamines.
Even though this is a well-known technology, its application in power plants presents high costs, thus limiting
its implementation.
This way, a full scale capture plant model was developed considering MEA as the solvent. This model
is based on a conventional flowsheet and was implemented in gPROMS ®, using the gCCS® libraries. It
comprises an absorption section, in which CO2 is dissolved by reacting with the amine, and a regeneration
section where it is stripped from the solvent. After its validation, the model was optimised by modifying the
design and operation parameters.
A cost estimation model was applied to the plant model, in order to determine capital and operational
expenditures. From the total cost obtained, 69% is due to the steam required in the regeneration section. As
for the equipment cost, the absorber packing is the most relevant fraction.
Considering typical values for the capture rate, CO2 purity and MEA concentration in the solvent as
constraints, the plant’s model optimisation led to a reduction of 15% in the specific total cost. Without
imposing these typical values, the total cost was further reduced. These results clearly show the potential of
model-based optimisation in the reduction of the cost associated with CO 2 capture, thus contributing to its
effective implementation in power plants.

Keywords: Carbon capture and storage, post-combustion capture, chemical absorption, MEA, cost
optimisation, gPROMS
CO2 obtained in power plants, in the preparation
1. Introduction of natural gas, or in other chemical industries [2].
CO2 is a naturally-occurring gas with a For a power plant, the capture cost is the
major influence on Earth’s surface temperature. main component of the overall CCS costs, mainly
However, due to human invention and due to the process energy requirements. This
industrialisation, its concentration in the implies an extra consumption of steam and
atmosphere has greatly increased. According to electricity, leading to a reduction in net efficiency
the 2013 IEA report [1], between the years of and consequently an increase in the electricity
2001 and 2011, the worldwide CO2 emissions costs [2].
from fossil fuel combustion increased by 31%, The technology employed in the capture
reaching a value of 31 billion tonnes per year. process applied in power plants depends on the
From this value, approximately 42% are due to characteristics of the gas mixture being treated.
the electricity and heat production sector. These are related to the kind of fuel and
According to the Global CCS Institute, technology used in the production of electricity.
carbon capture and storage (CCS) has the This way, the capture processes can be divided
potential to significantly reduce the CO2 in: post-combustion capture (PCC), pre-
emissions. The CCS process chain comprises combustion capture and oxy combustion.
several technologies involved in removal of From the available technologies for CO2
carbon dioxide from a gaseous stream, its capture, absorption is at the present the only one
transportation and final sequestration in site away with commercially available processes.
from the atmosphere. This can be applied to the Nevertheless, adsorption, membranes and
cryogenic separation are currently being tested

1
for application in the CCS chain [3]. Even though kinetics these conditions. Due to this, chemical
the absorption technology has been applied in absorption poses a better option in PCC
several other industries, due to the considerable processes, when compared with physical
costs associated with the application of the solvents [2].
capture process to such high flow rates, its The processes by chemical absorption are
application to the flue gas obtained in a large- based on the CO2 chemical dissolution in an
scale power plant fuelled by coal or natural gas is alkaline solvent, through its selective reaction
currently on a demonstration phase [4]. The only with one or several of the solvent components.
full scale capture unit already operating is located Examples of these solvents, and the available
on the Boundary Dam Power Station (Canada) processes in which there are applied are shown
and was designed for the capture of 1 MtCO2/year. in Table 1. As can be observed, amines, more
Besides this operating capture unit, there are precisely alkanolamines, are the chemical
several others planned to start operating in the solvents most used for CO2 capture, due its
near future, all of them based on CO2 capture by reactivity and adequate basicity. From these,
absorption [5]. The improvement of the currently monoethanolamine (MEA) is the one typically
available processes should allow a reduction of considered. This compound have a primary
its costs by 20 to 30%, being the option with more amine group, which confers a high reactivity to
potential for the reduction of CO2 emissions in the the molecule, allowing a strong reaction with CO2
near future [2]. at elevated rates.
A literature review was conducted in order Due to this reactivity, MEA is highly prone
to understand how the absorption is applied to to oxidative and thermal degradation, which limits
CO2 capture from a power plant, which the solvent concentration. The application of
technologies are commercially available, and oxidation inhibitors allows the increase of MEA
which are the current trends to reduce the capture concentration to the typical value of 30 wt%, and
cost. even 40 wt% [7]. In order to cope with this high
With the reduction of costs associated with reactivity this amines can also be blended with
the CO2 capture as the main objective, a sterically hindered or tertiary amines which are
conventional capture plant model was built in the less reactive [8].
gPROMS® ModelBuilder 3.7.1 environment using Concerning the thermal degradation, for
the recently developed gCCS® libraries. After the MEA it is negligible if the regeneration
model validation, the optimisation features of temperature is kept below 110 to 120 °C [9].
gPROMS® were used in order to minimise the The capture process is based on the CO2
costs associated with the unit’s design and dissolution, conducted in a packed column
operation. (absorber). The solvent loaded with CO2 (usually
called rich solvent) is pumped to the regeneration
2. Post-combustion Capture section. Before entering the regenerator, the rich
Technology solvent is heated through an integrated heat
exchange with the regenerated solvent (also
Post-combustion capture is applied in
called lean solvent). The regeneration process
power plants based on the combustion of a fossil
typically occurs through stripping above
fuel (coal, natural gas or oil), where a capture unit
atmospheric pressure. The heat demanded by
is used to remove the CO2 present in the flue gas.
this process is provided by a reboiler, which
This flue gas is mainly composed of N2, CO2, H2O
constitutes the main energy requirement of the
and O2, and also of reduced amounts of SOx, NOx
capture unit. The lean solvent is pumped back to
and ash [6]. The flue gas being treated have a
the absorber after being regenerated. Since the
considerably low CO2 content and is at
solvent is subject to losses during the process, a
atmospheric pressure, leading to a reduced CO2
make-up is required before re-entering the
partial pressure. Therefore, the chosen solvent
absorber [2].
has to be able to ensure acceptable loadings and
Table 1 – Commonly used chemical solvents and processes in which they are applied.
Solvent Type Processes Typical composition
 ABB Lummus Cress Up to 30 wt% with corrosion
MEA Primary amine
 Fluor’s Econamine FG PlusSM inhibitors
Blend of sterically hindered
KS1TM Hindered amine  MHI’s KM-CDR
amines
 Praxair’s Amine Up to 40 wt%, blended with a
MDEA Tertiary amine
 Shell’s Cansolv CO2 Capture System primary amine
NH3 Ammonia  Alstom’s Chilled Ammonia Up to 28 wt%
Hot Potassium  UOP’s Benfield Blend of potassium carbonate and
K2CO3
Carbonate  Giammarco-Vetrocoke’s Low Energy amine promoters
In these processes, it is typical to achieve 4. Models Validation
capture rates between 80 and 90%, and a purity The validation of the models used in the
as high as 99.9% (in volume) for the recovered simulation of a capture unit was achieved through
CO2 [2]. The inclusion of a PCC unit in a coal the comparison of the simulation results with the
fuelled power plant leads to an increase of 29% experimental data publically available. For that
in the energy input to achieve the same output, purpose, were considered the papers by
while for a natural gas fuelled power plant this Tobiesen et al., [12], and by Notz et al., [13],
increase is of 16% [4]. whose flowsheets were implemented in
gPROMS®.
3. Materials and Methods Tobiensen et al. article [12] presents the
In the present work, gPROMS® experimental data for 20 non-equal runs in pilot
ModelBuilder 3.7.1 is the simulation platform scale absorber. From the presented data, it was
used for flowsheet model simulation and possible to conclude that between Onda and
optimisation. Starting from the existing gCCS® Billet & Schultes correlations for the calculation of
model library, it is possible to assemble a mass transfer coefficients, the first one is more
flowsheet, in which are also included any other accurate. Using this correlation were obtained
auxiliary equations and custom sub models deviations from -13% to 26%, comparing the
required. simulated and experimental flow of captured CO2,
Another feature of gPROMS® as can be observed in Figure 1.
ModelBuilder is the optimisation tool, which can

Simulated Absorbed
10
be used to optimise the steady state behaviour of
8
CO2 (kg/h)
a continuous flowsheet, considering both design
and operation properties. Since the models 6
considered in this thesis are non-linear, the 4
optimisation problem constitutes a nonlinear
2
programming problem (NLP). gPROMS®
ModelBuilder 3.7.1 uses the SRQPD solver in the 0
0 2 4 6 8 10
solution of NLP problems, with an “Improved
Experimental Absorbed CO2 (kg/h)
Estimation Based” convergence criterion [10].
Figure 1 – Parity diagram of the absorbed amount of CO2
The models included in the gCCS® library (using Billet & Schultes correlation).
use gSAFT as physical properties package. This These deviations are deemed acceptable
package is based on the Statistical Associating when considering that predictive models that do
Fluid Theory (SAFT) equation of state, an not consider any fitting to the experimental data
advanced molecular thermodynamic method, were applied. It was also verified that this
based in physically-realistic models of molecules deviation tends to increase with the decrease of
and their interactions with other molecules. For the solvent lean loading (Figure 2).
carbon capture, gSAFT presents a modelling
26
alternative to phase and chemical equilibrium in
Deviation in Absorbed

the CO2-MEA-H2O system, since the chemical 16


bound between CO2 and MEA can be
CO2 (%)

incorporated as a short-range association, being 6


included in the molecular model. This way, the
involved reactions are treated implicitly, thus -4 0.15 0.25 0.35 0.45
greatly reducing the complexity of the model, thus
increasing its robustness [11]. -14
Lean Loading (molCO2/molMEA)
Figure 2 – Deviation between experimental and simulated
absorbed CO2 with the considered lean solvent loading.

Table 2 – Experimental and simulation results for the process key parameters and respective variation
Example 1 Example 2
Deviation Deviation
Key parameter Experimental Simulated Experimental Simulated
(%) (%)
CO2 capture rate
75.91 70.14 -8 51.32 38.71 -25
(%)
Specific heat
requirement 5.01 5.43 8 3.98 5.27 32
(GJ/tCO2)

3
In the Notz et al. article [13], two sets of order to meet a capture rate of 90% and a MEA
experimental data for a complete capture pilot mass fraction in the CO2 free lean solvent of 30%.
plant are presented. The key parameters used for This way, the base case is characterised by the
comparison and the respective deviations are parameters shown in Table 3.
shown in Table 2. From these parameters, it was Table 3 – Design parameters and operating conditions
considered in the original capture plant model.
observed that for a complete capture plant model
Parameter Value
with a specified heat input, the CO2 capture rate Diameter (m) 20
tends to be under estimated, with deviations that Height (m) 11.89
Absorber
can reach -25%, leading to an over prediction of Packing
Sulzer Mellapak
the specific heat requirement of 32%. 250YTM
Diameter (m) 8.5
Height (m) 10
Stripper
5. MEA Full Scale Capture Plant Packing
Sulzer Mellapak
250YTM
Model Lean-rich heat Cold stream outlet
89.65
exchanger temperature (°C)
5.1. Base Case Process stream
Lean solvent
outlet temperature 70.75
Assuming that the validation conclusions cooler
(°C)
presented above are still valid for a full scale Reboiler Temperature (°C) 117.84
plant, a MEA-based capture plant model was Reboiler pump Pressure (bar) 1.79
developed. The flowsheet considered was based Condenser Temperature (°C) 40
Flow rate (kg/s) 1450.14
on a case study developed by PSE. The flue gas Lean solvent MEA mass fraction
considered is characteristic of a natural gas 0.285
(g/g)
fuelled power plant emitting approximately 2 The referred conditions were applied in a
million tonnes of CO2 per year. The lean solvent conventional PCC flowsheet model, which
flow rate and MEA concentration were adjusted in topology is shown in Figure 3.

Figure 3 – MEA capture plant flowsheet as seen in gPROMS® ModelBuilder.

4
In these model, the absorption section is of the total cost associated to a carbon capture
composed by an absorber model (A-301) in which plant, only the cost fractions that depend on the
the Billet & Schultes correlation is used the plant design parameters and operation conditions
prediction of mass transfer coefficients and were considered, in order to simplify the
pressure drop. The heat exchange section is optimisation process.
composed by the lean-rich heat exchanger (HX- Using this costing model, it is possible to
301) and the lean solvent cooler (HXU-301), obtain the capture plant annualized investment,
which are used for the rich solvent cooling and the or CAPEX, and the annual operating cost, or
lean solvent cooling. In both these models, it was OPEX, both in a euro bases referred to year
assumed a constant overall heat transfer 2013. In the case of the investment annualization,
coefficient of 5 kW/(m2.K) [14] and pressure drop it was considered a linear amortization over a
of 0.62 bar [15]. The regeneration section is period of 10 year (equation (1)).
composed by a stripper model (ST-301), 𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡
𝐶𝐴𝑃𝐸𝑋 = (1)
associated with the reboiler (R-301) and 10 𝑦𝑒𝑎𝑟𝑠
condenser (C-301) models, in which were Both CAPEX and OPEX were also
assumed heat transfer coefficients of 1.14 and expressed as functions of the captured amount of
0.85 kW/(m2.K) [15], respectively. The model CO2, designated specific CAPEX (sCAPEX) and
P-304 is used to deliver the pressure required by specific OPEX (sOPEX), respectively (equations
the stripper model, while the model P-305 is used (2) and (3)).
to specify the reboiler pressure. The lean solvent 𝑠𝐶𝐴𝑃𝐸𝑋(€⁄𝑡𝑜𝑛𝐶𝑂2 ) =
flow rate and MEA concentration are set in the 𝐶𝐴𝑃𝐸𝑋 (€⁄𝑦𝑒𝑎𝑟) (2)
=
model RB-303, which allows the calculation of the 𝐴𝑏𝑠𝑜𝑟𝑏𝑒𝑑 𝐶𝑂2 (𝑡𝑜𝑛𝐶𝑂2 ⁄𝑦𝑒𝑎𝑟)
required solvent make-up. Besides these 𝑠𝑂𝑃𝐸𝑋(€⁄𝑡𝑜𝑛𝐶𝑂2 ) =
conventional components, the flowsheet model 𝑂𝑃𝐸𝑋 (€⁄𝑦𝑒𝑎𝑟 ) (3)
=
also comprises a saturation section (flash model 𝐴𝑏𝑠𝑜𝑟𝑏𝑒𝑑 𝐶𝑂2 (𝑡𝑜𝑛𝐶𝑂2 ⁄𝑦𝑒𝑎𝑟)
(F-301), PID controller model (PID-301) and For an approximate calculation of the
pump model (P-301)). In order to keep the investment required, it was used the factorial
absorber pressure above atmospheric, the pump method, which is based on the cost of the main
model P-302 acts as a gas blower, in order to process equipment [14]. For that, were
increase the flue gas pressure to 1.1 bar. In the considered the columns (shell and packing), heat
columns’ models the Billet & Schultes correlation exchangers, condenser, reboiler and pumps
was used for the calculation of mass transfer (including drivers). Considering that amine
coefficients and pressure drop. At last, were also solvents are corrosive, stainless steel 304 (SS-
added several flow multiplier models (FM-301 to 304) is required in all the equipment that contacts
FM-310) at the inlets and outlets of the absorber, directly with the capture solvent. The cost of the
regeneration section and reboiler to simulate the referred equipment is calculated through cost
existence of several equipment working in correlations [14], which upper bounds were used
parallel. to determine the number of equipment working in
Based on these model, the key parallel. Besides this, were also considered
parameters shown in Table 4 were obtained. installation factors, and other expenses, which
Table 4 – Results obtained from the simulation of the base
can be estimated as percentages of the main
case.
Parameter Value
equipment costs [14].
Capture rate (%) 89.9 For the calculation of the operational
CO2 purity (vol%) 95.8 expenses, were considered fixed and variable
MEA mass fraction in the CO2 free lean production costs. The fixed production also can
30.0
solvent (wt%)
be approximated to a percentage of the main
Number of absorption trains 2
Number of striping trains 2 equipment cost. For the variable production costs
Number of reboilers per stripping train 4 were considered the annual consumption of
Specific heat consumption (GJ/tCO2) 5.66 utilities (steam, cooling water and electricity) and
Lean loading (molCO2/molMEA) 0.249
solvent, which unitary costs were retrieved from
Rich loading (molCO2/molMEA) 0.463
[16].
Concerning the steam consumption, it was
5.2. Cost Estimation Model
considered low pressure steam (saturated at 3
To determine the costs associated with
bar). The cooling water requirements are given by
the design and operation of a capture plant a cost
the utility consumption, considering that the used
estimation model was implemented. For that
water presents inlet and outlet temperatures of 29
purpose, it was considered the procedure
and 49 °C, respectively [15]. For the estimation of
described in [14]. Considering that the objective
the electricity consumption, it was considered the
of the present economic model is the optimisation
5
power required in both lean solvent and rich
8% 5%
solvent pump drivers.
The solvent (MEA) consumption is Total Utilities Cost
estimated based on the required make-up and
amine degradation. According to [17], the Solvent Cost
washing sections in a capture plant are able to 87% Fixed Production Costs
reduce the MEA concentration to 1 ppm.
Therefore it was considered that the MEA
concentration in the treated flue gas is reduced to
Figure 6 – Distribution of the OPEX
this value, and the washed MEA used to reduce (Total = 64.56 €/tonCO2).
the required make-up. Also in reference [17], it is This way, it is expected that the
referred that MEA has a degradation rate of 1.5 optimisation of the capture plant flowsheet will
kgMEA/tCO2. mainly rely on changing the absorbers’ size and
the steam consumption.
5.3. Cost Estimation Results for the Base
Case 6. Optimisation Problem
By applying the cost estimation model in
Formulation
the capture pant model, it was possible to obtain
a CAPEX of 15.99 M€/year and an OPEX of
6.1. Objective Function
64.56 M€/year, which correspond to a total cost
For the optimisation of the capture plant,
of 80.55 M€ per year, equivalent to 43.15 € per
both the specific CAPEX and specific OPEX
tonne of captured CO2. As can be observed in
should be minimised. Therefore, Equation (4) was
Figure 4, the CAPEX only represents 20% of the
total cost. defined as objective function.
𝑂𝐹 = 𝑠𝐶𝐴𝑃𝐸𝑋 + 𝑠𝑂𝑃𝐸𝑋 (4)

6.2. Decision Variables


20%
The decision variables define which of the
CAPEX capture flowsheet variables are used to minimise
80% OPEX the objective function. In this case, the variables
which define the equipment design and the
process operating conditions have been chosen
Figure 4 – Total cost distribution in the base case (Total = (Table 5).
43.15 €/tCO2). According to Zahra [17], given the lean
Considering the distribution of equipment solvent flow rate, capture rate and lean loading,
costs shown in Figure 5, it is observable that the the optimal steam consumption is obtained for the
columns’ packing represents the major fraction of combination of stripping pressure and reboiler
the equipment costs, from which 87% are due to temperature with higher values. Therefore, before
the absorber’s packing. proceeding to the process optimisation, the base
flowsheet was modified in order to start from the
5% 9% Columns Shell
maximum temperature (120 °C [9]). This way, this
2% variable was not directly included in the
Heat Transfer
Equipment optimisation problem.
Table 5 – Decision variables.
Pumps and Drivers
Model Variable
84% Diameter (m)
Packing A-301
Height (m)
C-301 Temperature (K)
HX-301 Cold outlet temperature (K)
Figure 5 – Distribution of main equipment costs for the HXU-301 Process outlet temperature (K)
base case. Source flow rate (kg/s)
RB-303
From the distribution show in Figure 6, it is Source MEA mass fraction (g/g)
observable that utilities represent the main P-305 Pressure (bar)
Diameter (m)
fraction of the OPEX, from which the steam ST-301
Height (m)
consumption corresponds to 98%. In fact the
annual steam consumption is 69% of the plant 6.3. Constraints
annual cost. The process constraints are used to keep
the standard conditions initially imposed, such as
the capture rate, the MEA mass fraction in the
CO2 free solvent and the minimum CO2 purity in
6
the product stream, and to keep technical 0.009

CO2 molar flux (kmol.m-2.s-1)


specifications, such as preventing the columns
flooding. These constraints are divided in equality 0.006
and inequality constraints, which are listed in
Table 6 and Table 7, respectively.
0.003
Table 6 – Equality constrained variables, with respective
constrained value.
Constrained variable Imposed value 0
Capture rate (g/g) 0.90 0 1000 2000 3000 4000
CO2 molar fraction in the CO2 Volume of packing from the absorber top
0.95
stream (mol/mol)
(m3)
MEA mass fraction in the CO2 Base Optimal
0.30
free lean solvent (g/g)
Figure 7 – CO2 molar flux to the liquid phase across the
absorber (top of the absorber equivalent to 0), before and
after the specific total cost minimisation with standard
Table 7 – Inequality constrained variables, with respective constraints.
upper and lower bounds.
An increase by 5% in the outlet
Lower Upper
Constrained variable temperature of the rich solvent in the lean-rich
bound Bound
Vapour velocity/vapour heat exchanger and by 1% in reboiler
flooding velocity in the top 10-10 0.7 temperature led to an increase of the overall
of the absorber
temperature in the stripping column, as can be
Vapour velocity/vapour
flooding velocity in the 10-10 0.7 observed in Figure 8.
bottom of the stripper 400
Process stream ΔT in the
10-10 1010
lean solvent cooler (°C)
Temperature (K)

The process stream temperature


difference (ΔT) in the lean solvent cooler is 380
required, since in some of the optimisation cases
it tended to achieve un-realistic values below 0.
Therefore, this variable was included in the
flowsheet model, where a very small positive 360
number was defined as a lower bound (10-10), 0 0.5 1
while the upper bound was considered an Relative position from the stripper top
Optimal Base
excessively large number (1010). The ratio
Figure 8 – Axial temperature profiles in the stripping
between the vapour operating velocity and the columns for the liquid phase, before and after the specific
vapour flooding velocity should be kept below total cost minimisation with standard constraints.
70% to 80% in order keep the column’s efficiency The increase in the temperature profile
[18]. and in the columns’ height caused an increase of
its stripping efficiency. This can be observed in
7. Optimisation Results Figure 9, where the CO2 flux from the gas phase
to the liquid phase across the stripper is always
7.1. Specific Total Cost Minimisation negative.
with Standard Constraints
0.01
The results obtained from this optimisation
CO2 molar flux (kmol.m-2.s-1)

process show that the main variations occur in the 0.005


absorber size (height and diameter are reduced
in 29% and 12%, respectively), in the stripper size 0
(diameter is reduced 15% and height is increased 0 200 400 600
-0.005
69%) and in the lean solvent flow rate (reduced
18%). -0.01
The reduction in the absorber size leads to
a reduction of the packing volume, as well as the -0.015 Volume of packing from the stripper top
shell’s mass. This reduction is possible due to the (m3)
Base Optimal
reduction in the lean loading (reduced to 0.200
Figure 9 – CO2 molar flux to the liquid phase across the
molCO2/molMEA) and the approach to the flooding striper (top of the stripepr equivalent to 0), before and
velocity in the absorber, which now reaches 70% after the specific total cost minimisation with standard
in the top of the column. Both these factors constraints.
improve the column’s efficiency, as can be Considering the improved efficiency and
observed in Figure 7. the increase in the lean loading, the heat required
in the reboiler is decreased. This way, the specific

7
heat requirement shows an optimal value of 4.87 capture plant annual cost. The minimisation of the
GJ/tCO2 (less 14%). specific total cost allowed a reduction in specific
Considering this main factors, it is heat requirement in 14%. In order to evaluate if a
observed a reduction in the CAPEX of 21% and greater reduction in this parameter is possible,
in the OPEX of 13%, which lead to a specific total and its effect in the capture total cost, the
cost of 36.69 €/tCO2 (-15%). described optimisation procedure was conducted
using the specific heat consumption as objective
7.2. Effect of the Initial Guesses function.
Considering that in the formulated A further reduction of the specific heat
optimisation problem exist ten decision variables requirement to 4.46 GJ/tCO2 was obtained.
(Table 5) and only three equality constraints Nevertheless, this was achieved by greatly
(Table 6), it is expectable that multiple solutions increasing the columns’ height and therefore the
can lead to the same optimal solution. The initial required packing volume. This led to an increase
guesses of a non-linear optimisation problem may in specific CAPEX by 674% and in the specific
influence the results obtained, and even OPEX by 23%, when compared to the initial
determine the convergence or not of the problem. optimisation, which is equivalent to a specific total
Since the lean solvent CO2 loading as a main cost is increased to 88.98 €/tCO2.
influence in the capture model, all the
optimizations performed started from a lean 7.5. Effect of the Imposed Capture Rate
loading of 0.2 molCO2/molMEA. In order to evaluate This variable was initially imposed as
the influence of this parameter in the process 90%. Its effect on the plant’s specific total cost
convergence, the optimisation described in the was evaluated by changing the imposed values
previous section was conducted starting from a to 70%, 80% and 99% (Figure 11).
lean loading of 0.1 and 0.3 molCO2/molMEA, by 50
adjusting the lean solvent flow rate and MEA
Specific total cost

mass fraction and the reboiler pressure. Through 40


(€/tCO2)

the optimal results obtained it was observed a


variation of less than 1% in the total cost. It was
30
also possible to conclude that the lean solvent
temperature after the cooler isn’t a relevant
parameter in the considered conditions, and the 20
lean solvent cooler wouldn’t be required. 70 80 90 99
Capture rate (%)
Figure 10 – Variation of the optimal specific total cost
7.3. Effect of the Number of Absorption with the imposed capture rate.
Trains Between 70% and 90%, the variation is
The number of absorption trains was achieved through an increase in the lean solvent
chosen considering the construction limitations flow rate and in the absorber and stripper height,
regarding the column’s diameter and the vapour associated with a slight decrease in the lean
flooding velocity. This variable was changed in loading. This leads to an increase of 2% in the
the base case and optimised for each value. optimal total cost. On the other hand, between the
From the obtained results, it was possible capture rates of 90% and 99%, there is 28%
to conclude that the decision variable mainly increase in the specific total cost, due to a major
affected is the absorber diameter. Considering decrease in the optimal lean solvent flow rate and
only one absorber would be technically un- lean loading, accompanied by a further increase
feasible due to the elevated optimal absorber in the columns’ height. In the case of a 99%
diameter. The use of two absorption trains capture it was also concluded that due to the heat
conditioned by the possibility of constructing released in the absorption process and the
absorbers with a diameter above 15 m. reduced solvent flow rate, the lean solvent cooler
Nevertheless, it was observed that increasing the has higher relevance. In the same case, the rich
number from 1 to 4 only increased the specific solvent temperature entering the stripper is
total cost in 0.4%, since the total volume of decreased, since the vapour flow rate is
packing tends to be the same. increased and the solvent flow rate is decreased,
being the temperature across the columns kept at
7.4. Specific Heat Requirement higher values.
Minimisation
The steam consumption is directly 7.6. Effect of the Imposed CO2 Purity
proportional to the heat required in the This variable was initially imposed as
regeneration reboilers, and represents 69% of the 95%. Its effect on the plant’s specific total cost

8
was evaluated by changing the imposed values from 20% to 30%, and in 4%, when changing it
to 75%, 85% and 99% (Figure 10). from 30% to 40%. These variations are mainly
39 due to the possibility of reducing the solvent flow
Specific total cost

38 rate, with the increasing concentration,


38 associated whit an increase in the optimal lean
(€/tCO2)

loading. Both these factors contribute to the


37
reduction of the specific heat requirement,
37
allowing the total cost reduction. Nevertheless, it
36 should be considered that there are implications
36 of using a more concentrate solvent, such as, the
75 85 95 99 size of the water washing sections or the costs
CO2 Purity (%)
associated with the required oxidation inhibitors,
Figure 11 – Variation of the optimal specific total cost with
the imposed CO2 purity.
which were not considered.
Between the purities of 75% and 95%, it
was observed an increase of only 0.2% in the 7.8. Specific Total Cost Minimisation
specific total cost. The condenser temperature is with Inequality Constraints
related to this variation, since it defines the Since it was observed that both the
amount of water condensed, and therefore the capture rate and the CO2 purity tended to show a
purity of the final stream. On the other hand, for minimum specific total cost, the optimisation
lower purities the amount of lost solvent in the problem was conducted without any equality
regeneration process tends to increase, thus constraints. This led to the optimal capture rate of
leading to a minimum in the sOPEX. When the 75%, CO2 purity of 88% and MEA mass fraction
purity is increased to 99%, the specific cost in the CO2 free lean solvent of 40%, being the last
increases in 4%, associated with a major variation one the variable upper bound. This allowed a
in the OPEX. This was due to the further specific total cost reduction to 34.70 €/t CO2 (5%
reduction in the condenser temperature that less when compared with the initial optimal
required the use of chilled water, which was results), associated with the specific heat
considered eight times more expensive [16], requirement of 4.60 GJ/tCO2.
leading to an increase of 435% in the
cooling/refrigerating water annual costs. 8. Conclusions
Nevertheless, that it was verified that this The gCCS® capture library allowed the
parameter did not have a considerable influence development of a capture plant model in the
in the specific heat consumption, thus the gPROMS® ModelBuilder environment.
reduced effect on the total cost. The full scale capture plant optimisation
was conducted using as base case a PSE’s case
7.7. Effect of the Imposed MEA study, which was modified in order to meet a 90%
Concentration capture rate and a MEA fraction in the CO2 free
The MEA concentration was initially lean solvent of 30%. The application of a cost
defined by the imposition of a MEA mass fraction estimation model to the base case, allowed the
in the CO2 free lean solvent of 30%. In order to calculation of a specific total cost of 43.15 €/tCO2,
evaluate its effect on the process total cost, the from which the OPEX represents 80% and is
imposed value was modified to 20% and 40% associated to a specific heat requirement of 5.66
(Figure 12). GJ/tCO2. The cost estimation allowed to conclude
40 that the equipment with more influence in the
Specific total cost

CAPEX is the absorber’s packing (73% of the


equipment total cost). For the OPEX, the main
(€/tCO2)

38
fraction is steam consumption, which in fact
represents 69% of the total annual cost.
36
The base case specific total cost
minimisation, considering a capture rate of 90%,
34
a CO2 purity of 95 vol% and a MEA fraction in the
20 30 40
MEA mass fraction in the CO2 free lean solvent (%) CO2 free lean solvent of 30%, led to its reduction
Figure 12– Variation of the optimal specific total cost with in 15%, associated to a specific heat requirement
the MEA mass fraction in the CO2 free lean solvent. of 4.87 GJ/tCO2.
The obtained optimal results showed a Considering the specific heat requirement
more significant variation in the specific total cost as the objective function to be minimised, this
was than in the previous cases, showing a variables was further reduced to 4.46 GJ/tCO2.
reduction in 5%, when changing this constraint Nevertheless, this led to an increase in the total

9
cost (143% when compared with the initial [7] B. Delfort, P.-L. Carrette and L. Bonnard,
optimal results). “MEA 40% with improved oxidative stability
The optimisation without any equality for CO2 capture in post-combustion,”
constraints led to the optimal capture rate of 75%, Energy Procedia, vol. 4, pp. 9-14, 2011.
CO2 purity of 88% and MEA mass fraction in the [8] P. D. Vaidya and E. Y. Kenig, “CO2-
CO2 free lean solvent of 40%, being the last one Alkanolamine Reaction Kinetics: A Review
of Recent Studies,” Chemical Engineering
the variable upper bound. This allowed a specific
& Technology, vol. 30, no. 11, pp. 1467-
total cost reduction to 34.70 €/tCO2 (less 5% when 1474, 2007.
compared with the initial optimal results),
[9] J. Davis and G. Rochelle, “Thermal
associated with the specific heat requirement of degradation of monoethanolamine at
4.60 GJ/tCO2. stripper conditions,” Energy Procedia, vol.
The analysis presented shows that the 1, no. 1, pp. 327-333, 2009.
total cost is greatly affected by the trade-off [10] Process Systems Enterprise Limited,
between the solvent’s flow rate, the absorber gPROMS Documentation, Release 3.7,
height and the lean solvent’s loading. Increasing London, 2013.
the MEA concentration in the solvent has a [11] C. V. Brand, “CO2 capture using
positive effect in all this variables. This is also monoethanolamine solutions:
verified for the capture rate in a limited range of Development and validation of a process
values. It was also observed that at moderate model based on the SAFT-VR equation of
state,” Centre for Process Systems
capture rates, increasing both the stripper’s
Engineering, Department of Chemical
height and the rich solvent temperature at its inlet Engineering, Imperial College London,
also tend to reduce the steam consumption, and London, 2013.
therefore the total cost. The interconnection [12] F. A. Tobiesen, H. F. Svendsen and O.
between these variables shows the relevance of Juliussen, “Experimental validation of a
a model-based full plant optimisation for the rigorous absorber model for CO2
reduction of the capture cost, which is an postcombustion capture,” AIChE Journal,
important step for the effective CCS vol. 53, no. 4, pp. 846-865, 2007.
implementation in real power plants. [13] R. Notz, H. P. Mangalapally and H. Hasse,
“Post combustion CO2 capture by reactive
absorption: Pilot plant description and
9. Main Bibliography results of systematic studies with MEA,”
International Journal of Greenhouse Gas
[1] International Energy Agency, “CO2 Control, vol. 6, pp. 84-112, 2012.
Emissions from Fuel Combustion,” Paris, [14] G. Towler and R. Sinnot, Chemical
2013. Engineering Design : Principles, Practice
[2] Working Group III of the Intergovernmental and Economics of Plant and Process
Panel on Climate Change, “IPCC Special Design 2nd ed., Oxford: Butterworth-
Report on Carbon Dioxide Capture and Heinemann, 2013.
Storage,” Cambridge University Press, [15] A. K. Coker, Ludwig’s Applied Process
New York, 2005. Design for Chemical and Petrochemical
[3] A. B. Rao and E. S. Rubin, “A Technical, Plants, Burlington: Gulf Professional
Economic, and Environmental Assessment Publishing, 2007.
of Amine-Based CO2 Capture Technology [16] R. K. Sinnot, Coulson & Richardson's
for Power Plant Greenhouse Gas Control,” Chemical Engineering Series, Volume 6,
Environmental Science & Technology, vol. Fourth Edition, Chemical Engineering
36, no. 20, pp. 4467-4475, 2002. Design, Oxford: Butterworth-Heinemann,
[4] E. S. Rubin, H. Mantripragada, A. Marks, 2005.
P. Versteeg and J. Kitchin, “The outlook for [17] M. R. M. A. Zahra, “Carbon Dioxide
improved carbon capture technology,” Capture from Flue Gas: Development and
Progress in Energy and Combustion Evaluation of Existing,” 2009.
Science, vol. 38, pp. 630-671, 2012. [18] R. Billet and M. Schultes, “Prediction of
[5] Carbon Capture and Sequestration Mass Transfer Columns with Dumped and
Technologies @ MIT, “Carbon Capture and Arranged Packings: Updated Summary of
Sequestration Project Database,” [Online]. the Calculation Method of Billet and
Available: Schultes,” Chemical Engineering Research
https://sequestration.mit.edu/tools/projects and Design, vol. 77, no. 6, pp. 498-504,
/. [Accessed 30 September 2014]. 1999.
[6] Global CCS Institute, “CO2 CAPTURE
TECHNOLOGIES - POST COMBUSTION
CAPTURE (PCC),” EPRI, 2012.

10

You might also like