Part 1: Design, modeling and simulation of post-combustion CO₂ capture systems using reactive solvents

Abstract

The simulation and modeling of post-combustion CO₂ capture systems are considered to be an important strategy to obtain process integrity and gain design confidence for the construction and commissioning of commercial post-combustion CO₂ capture plants. It is therefore essential to obtain an understanding of the fundamental concepts of designing and modeling. This article reviews the concepts of designing a CO₂ capture system with specific emphasis on the absorber for diameter and height. It covers several steps (i.e., empirical design method, theoretical design method, laboratory method and pilot plant techniques) used to design the absorber. A conceptual design of an overall CO₂ capture process is also given in the article. Process validations of the modeling using ProMax with four existing pilot plants. (the International Test Centre of CO₂ Capture pilot plant, the Esbjerg CASTOR pilot plant, the Institute of Thermodynamics and Thermal Process Engineering, Stuttgart pilot plant and the SINTEF/NTNU pilot plant) are presented. Moreover, a discussion of process integration of the CO₂ capture plant into a fossil fuel-fired power plant is included in this paper.

Figure 1.  Typical absorption-based CO₂ capture unit.

Figure 2.  Design procedure of CO₂ absorber.

(A) Emphirical design method, (B) Theoretical design method, (C) Laboratory method and (D) Pilot plant technique method.

Modified with permission from [20].

Figure 3.  Solubility plot and operating line.

Figure 4.  Small cell used for laboratory method.

C_B: Concentration; G: Gas; L: Liquid; P_A: Partial pressure.

Modified with permission from [20].

Figure 5.  String-of-sphere column [45].

γ: Unit volume of liquid; a_v: Interfacial area; C_B: Concentration; G: Gas; K_G; Overall gas mass transfer coeffiecient; K_L; Overall liquid mass transfer coeffiecient; L: Liquid; P_A: Partial pressure; Z: Height.

Figure 6.  Pilot plant technique.

Modified from with permission from [20].

C_B: Concentration; G: Gas; L: Liquid; P_A: Partial pressure; Z_m: Height.

Figure 7.  Pilot plant schematic process flow diagram for (A) ITC, Esbjerg CASTOR and ITT (B) SINTEF/NTNU.

HEX: Heat exchanger; L: Liquid; P: Pump; V: Vapour.

Reproduced with permission from [15].

Figure 8.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 8% CO₂ content and 90% CO₂ capture performance from ITC pilot plant.

Figure 9.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 8% CO₂ content and 80% CO₂ capture performance from ITC pilot plant.

Figure 10.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 4% CO₂ content and 90% CO₂ capture performance from ITC pilot plant.

Figure 11.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 4% CO₂ content and 80% CO₂ capture performance from ITC pilot plant.

Figure 12.  Parity plots of (A) CO₂ concentration (B) temperature profile (C) CO₂ mass balance and (D) reboiler heat duty from ITC pilot plant.

Figure 13.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 11.8% CO₂ content and 90% CO₂ capture performance from Esbjerg CASTOR pilot plant.

Figure 14.  Parity plots of (A) CO₂ concentration (B) tempearature profile (C) CO₂ mass balance and (D) reboiler heat duty from Esbjerg CASTOR pilot plant.

Figure 15.  Experiment and simulation results for (A) CO₂ concentration and (B) temperature profiles in the absorber at 13.2% CO₂ content and 44% CO₂ capture performance from ITT Stuttgart pilot plant.

Figure 16.  Parity plots of (A) CO₂ concentration (B) tempearature profile (C) CO₂ mass balance and (D) reboiler heat duty from ITT Stuttgart pilot plant.

Figure 17.  Experiment and simulation results for temperature profile in the absorber at (A) 2.5% CO₂ content and 78.5% CO₂ capture performance, (B) 2.6% CO₂ content and 58.3% CO₂ capture performance, (C) 5.3% CO₂ content and 58.1% CO₂ capture performance and (D) 9.48% CO₂ content and 49.1% CO₂ capture performance from SINTEF/NTNU pilot plant.

Figure 18.  Parity plots of (A) tempearature profile (B) CO₂ mass balance and (C) reboiler heat duty from SINTEF/NTNU pilot plant.

Figure 19.  Parity plots of (A) CO₂ concentration (B) tempearature profile (C) CO₂ mass balance and (D) reboiler heat duty from ITC, Esbjerg CASTOR, ITT and SINTEF/NTNU pilot plants.

Figure 20a.  (A) An integration of CO₂ capture into pulverized coal-fired power plant.

G: Generator; HX: Heat exchanger; PM: Particulate matter; RH: Reheater; SH: Superheater.

Modified with permission from [16].

Figure 20b.  (B) An integration of CO₂ capture into combined cycle gas turbine power plant.

Comp: Compressor; G: Generator; GT: Gar turbine; HP: High-pressure turbine; HX: Heat exchanger; IP: Intermediate-pressure turbine; LP: Low-pressure turbine.

Bolded numbers 1–4 indicate a connection of each stream line.

Modified with permission from [50].

Figure 21.  Intermediate pressure/low pressure crossover pipe.

G: Generator; IP: Intermediate-pressure turbine; LP: Low-pressure turbine.

Figure 22.  Steam extraction from low pressure turbine.

G: Generator; IP: Intermediate-pressure turbine; LP: Low-pressure turbine.

Background

The capture of CO₂ from industrial flue gases has become an important issue in recent years. The threat of climate change from increased levels of GHGs in the atmosphere is significant. Mean global temperatures are on the rise and various regions of the world have been experiencing adverse climate issues, for example, polar ice caps are melting at an alarming rate, heat waves and droughts are becoming commonplace, and the occurrence of violent weather has increased around the globe. Therefore, the need for research and development of cost-effective CO₂ capture technologies is urgent.

The majority of industrial CO₂ emissions come from fossil fuel- (i.e., coal, natural gas and oil) fired power stations, for example, electric utilities account for 29% of all CO₂ emissions in Canada. In addition, according to the International Energy Agency (IEA) World Energy Outlook (2002) predictions, the world energy-related CO₂ emissions will reach 38 billion tons by 2030, representing an increase of approximately 1.8% per year and 70% above 2000 levels [1]. Since coal is relatively cheap, and exists in abundant quantities around the world, it is likely that it will remain the fuel of choice at thermal power stations for decades to come. If the CO₂ emission rates at these facilities can be reduced in a cost-effective manner, the anthropogenic contribution to global warming will be minimized.

There are three process options for CO₂ capture: pre-combustion, post-combustion and oxyfuel combustion. Depending on the process, various technologies for CO₂ capture can be used including absorption, adsorption, membranes and hybrid applications of these three. In the immediate future, post-combustion capture via chemical absorption with amines appears to be the leading candidate for CO₂ removal at fossil fuel fired power stations; however, several advances are being made with the other technology options. Two recent events, the IEA-sponsored 10^th International Conference on Greenhouse Gas Control Technologies (GHGT-10, September 18–22, 2010, Amsterdam, The Netherlands) and the 7^th IEA CO₂ Capture Network (September 29–30, 2009, Regina, Canada) illustrate the demand for knowledge in this area. This underscores the need for technological breakthroughs to support as well as facilitate the CO₂ emissions mitigation efforts.

The purpose of post-combustion capture is to capture CO₂ from an industrial flue gas stream using a regenerable absorbent. Conversely, the specific objective of emerging technologies for the post-combustion CO₂ capture process is to optimize the technology to capture CO₂ with high efficiency, but with low capital and operating expenses (CAPEX and OPEX). These technologies include the use of CO₂-capture absorbents, metal organic frameworks [2], enzyme-based systems [3] and ionic liquids [4]. However, it has been reported that CO₂ capture based on regenerable absorbents is the most mature and probably the most promising technology amongst all the aforementioned technologies [5].

For absorbents-based CO₂ capture, there are many types of absorbents that can be used, generally including physical, chemical or combined physical and chemical absorbents (or solvents). Chemical solvents using amines are the most likely to be used for post-combustion capture, as they have better effectiveness and efficiency to absorb CO₂ at low partial pressures and total pressures than physical solvents. It is known that the partial pressure of CO₂ in the flue gas is relatively low, typically in the range of 4–15% by volume, depending on the type of fossil fuel used. Thus, it is anticipated that amine-based absorbents (i.e., chemical solvents) will be used for the first-stage of post-combustion capture of CO₂. This can be seen from existing facilities that do partial scrubbing of CO₂ from the flue gases of commercial power plants (e.g., IMC Global Soda Ash Plant, Warrior Run Power Plant and Sumitomo Chemicals Plant [6]) using amines. Based on these demonstration plants, it appears that up-scaling the technology to integrate into the full-scale power plant operation is promisingly feasible [101,7].

The chemical absorbent (otherwise called chemical solvent) needs to be regenerated or stripped of the CO₂ it absorbed before it is recycled back to the absorber. It is found that chemical solvents require more heat in order to break the strong chemical interaction between the CO₂ and the solvent. Furthermore, there may be some operational issues, which can impede the process of CO₂ capture from flue gas using amines. These include degradation of solvents and corrosion of equipment, both of which are caused by the presence of CO₂, oxygen and some impurities in the gas stream such as oxides of sulfur, oxides of nitrogen and fly ash. Nevertheless, chemical solvent can efficiently remove CO₂ from the flue gas in the presence of a large amount of non-condensable gas, and remedies, such as the use of inhibitors, can be provided to counteract some of the disadvantages, specifically solvent degradation and equipment corrosion [8]

▪ Process simulators to model post-combustion CO₂ capture using regenerable absorbent

Regardless of chemical reactions, the design of a conventional absorber is generally based on standard chemical engineering principles or techniques. However, these principles and techniques have not been previously applied in the design of a plant for the capture of CO₂ from power plants. In addition, the necessary design parameters that are unique to this process are not available in commercial design packages and neither have they been validated in the literature. Therefore, it is worthwhile that an important step to confirming the feasibility of building a large CO₂ capture plant is to set up a pilot plant or a demonstration plant. Another step requires performing process design and simulation. A front end engineering design (FEED) study that uses simulation techniques is also necessary to improve plant reliability and increase productivity before actually constructing the plant, especially for designs developed for new technologies. Process simulation can provide some of the results that would otherwise be gained through laboratory and/or pilot plant tests, and such simulation should be used as it would reduce the need for process revamping and improve start-up times. Although laboratory and especially pilot plant tests are important to get a good feel for the design, these tests could be expensive in regards to material and energy consumption and costs. Process simulation then becomes a tool that can be used successfully to compliment laboratory and pilot plant tests.

The development of process-design calculations from open literatures and also chemical process simulators from engineering developers has helped in the design of equipment and process. The early computer simulation tools could only be used for single operating unit, and it was difficult to get convergence for a complicated process integrated with many units. Up-to-date simulation software tools are reliable for completing a simulation of a power plant integrated with a CO₂ capture plant. The heat efficiency of power generation is a critical factor. A power plant with a post-combustion, amine-based CO₂ capture system will have reduced efficiency compared with one without CO₂ capture because of the steam requirement for stripping CO₂ from the amine solution (i.e., absorber regeneration). Reboiler heat duty for CO₂ stripping clearly impacts the steam cost, hence process-configuration optimization becomes an efficient approach for system energy integration to minimize the heat duty. A number of process simulation tools, such as EBSILON^®, GateCycle^®, GT Pro^®, PROATES^®, Steam Pro^® (for power plant simulation), Aspen Plus^®, Aspen HYSYS^®, CHEMASIM^®, ProMax^®, ProTreat^® (for CO₂ capture simulation), GT Pro/Master^®, IECM^®, Steam Pro/Master^® and ThermoFlex^® (for process integration) and other in-house software, have been developed for power plant and CO₂ capture simulation [102,9–17]. Even though such chemical simulators would help engineers to design and simulate the post-combustion capture system, the comprehension of the fundamental concepts and process validation is a must in order to verify whether the design is acceptable. Therfore, this article discusses these two issues. In addition, the concept to integrate the CO₂ capture into fossil fuel fired power plant is discussed in the article.

CO₂ capture plant design

▪ Description of process flow diagram of Co₂ capture unit

Figure 1 demonstrates the typical schematic diagram of the CO₂ capture process using regenerable absorbent. For a general description of the process, the liquid solution containing the chemical absorbent is supplied to the top of the absorber while the flue gas stream from the industrial process flows counter-currently from the bottom to the top of the absorber by passing through the absorber column. The absorber column contains internals such as tray or random packing or structured packing, each of which provides the surface area of contact between the liquid and the gas. This contact allows the desired component to be transferred from the gas phase to the liquid phase. In the CO₂ capture process, the liquid chemical absorbent preferentially removes CO₂ from the gas stream while the other components of the gas stream that do not react with the chemical absorbent are vented to the atmosphere from the absorber top. The CO₂-rich liquid solution exits at the bottom of the absorber. Then, the CO₂-rich solution is pumped through the lean/rich heat exchanger before it enters the regenerator at the top for solvent regeneration and CO₂ stripping. The carryover vapor stream of the solution is condensed back to the top of the regenerator, leaving mostly wet CO₂ out of the stripping process as the CO₂ product. The reboiler receiving the thermal heat primarily extracted from the steam-power cycle or other sources supplies the energy to the solution in the stripper. After regeneration, the lean solution carrying a lean amount of CO₂ from the reboiler is passed through the rich-lean heat exchanger to transfer heat to the CO₂-rich heat solution before the former is pumped through a cooler to cool the temperature down to 35–45°C before reintroducing into the absorber at the top. Provision is made for chemical absorbent makeup next to the CO₂-lean liquid solution pump if necessary to maintain the solvent concentration at the specified operating level.

▪ Design methods for CO₂ absorber

One of the important units in the absorbent-based post combustion CO₂ capture unit is the absorber. The design of the absorber involves determining or specifying the proper absorber size required for a given CO₂ capture operation. There are several steps to design an absorber size. These are given in Figure 2[18–20]. The specific operating designs are defined in terms of the independent parameters (e.g., flow rate of gas and liquid-phases and flue-gas compositions and gas–liquid contacting system). Physical properties data on both phases must then be obtained to determine the superficial velocities and absorber diameter based on flooding conditions.

▪ Absorber diameter

In order to calculate the diameter of CO₂ absorber, the solubility of CO₂ in chemical absorbent under a specified temperature and equilibrium should be conducted [21]. A solubility plot from experiment or open literature between mole fraction of CO₂ in gas and liquid phases provides the operating line of (L/G)_operating and minimum operating line of (L/G)_min. An example of the solubility plot and operating line is demonstrated in Figure 3.

With the specified gas flow rate inlet, the liquid flow rate would be obtained from (L/G)_operating. The calculation of the flow parameter (F_LG) should later be performed by using the following equation.

Equation 1

Where L, G, rG and ri denote liquid flow rate (kg/min), gas flow rate (kg/min), gas density (kg/m3) and liquid density (kg/m3), respectively. The physical properties such as density can be obtained from experiment, open literature or chemical process simulation software (e.g., Aspen and ProMax) [22–25]. Following this step is to determine ordinate (e) from the flooding curve [26]. The mass flow rate of gas per unit cross sectional area of column is calculated from Equation 2.

Equation 2

Where G´, F, φ, g_c and µ are mass flow rate of gas per unit cross sectional area of column (kg/ m² s), packing factor, ratio of specific gravity of the liquid and water, gravitational constant (m²/s) and viscosity of liquid (cP), respectively. Diameter of the absorber can be computed by:

Equation 3

Equation 4

Where A and f represent cross sectional area (m²) and flooding factor, respectively.

▪ Absorber height

Empirical design method

There are several studies that use the empirical design approach to design the column height [21,27]. The calculation step of this method is given in Figure 2 at Route #1. To calculate the packing height of the tower, a material balance of the transferred component A should be performed as:

Equation 5

by:

Equation 6a

Equation 6b

Where a_v denotes interfacial area per unit volume of packing (m²/m³); C_A,L is concentration of A on the liquid bulk (kmol/m³); C*_A is concentration of component A in equilibrium with gas phase (kmol/m³); G_I; is is inert gas molar flow rate (kmol/m² s); K_G is overall gas mass transfer coefficient (kmol/m^{2 s} kPa); K_L is overall liquid mass transfer coefficient (m/s); N_A is mass transfer flux of the absorbed component A (kmol/m² s); P is total pressure (kPa); y_A,G is mole fraction of component A in gas bulk; and y*_A is mole fraction of component A which is in equilibrium with liquid bulk.

By simplifying Equations 5 & 6, the column height can be expressed as:

Equation 7

The mass-transfer coefficient can be assumed as an independent parameter of the absorbed component concentration. Therefore, the column height is presumably simplified to:

Equation 8

Where, G_I/K_Ga_VP represents the height of a transfer unit (HTU), while the following equation represents the number of transfer units (NTU) [28].

The mass transfer coefficient can be obtained from experimental measurement at the bottom and top of the absorber. This is called the average overall volumetric mass transfer coefficient, (KGav)ave, which can be given as:

Equation 9

Where, the log-mean driving force, (YA.G – Y*A)lm, can be identified by:

Equation 10

The volumetric mass transfer coefficient is assumed to be constant along the column height and it is basically correlated as a function of the operating parameters such as flow rates, pressure and temperature. Although the empirical design method requires experimental data, it is unreliable when chemical reactions get involved and where the enhancement factor significantly changes, unless the data are obtained under conditions equivalent to a full-scale absorber.

Theoretical design method

The theoretical design method procedure is given in Route #2 in Figure 2. The theoretical design for absorption with chemical reaction was first proposed by Astarita (1967) and Danckwerts (1970) [29,30]. Their efforts were primarily directed towards developing expressions for the local mass transfer coefficients. A set of design equations for isothermal gas absorbers with chemical reaction was later developed by Joshi et al. (1981) [31]. However, assuming adiabatic conditions would have been more realistic because heat losses are generally small in industrial absorbers [32,33]. A rigorous design procedure for adiabatic gas absorption with chemical reaction was subsequently described by Pandya (1983) [33]. The procedure was based on Treybal’s concepts (1969) [32] for adiabatic, physical gas absorption and Danckwert’s work (1970) [30] on isothermal gas absorption with chemical reaction. Pandya’s procedure accounted for the heats of absorption and reaction, solvent evaporation and condensation, chemical reactions in the liquid phase and heat and mass transfer resistances in both phases [33].

A mathematical model based on Pandya’s algorithm (1983) [33] is used to illustrate the theoretical design approach. The assumptions that are typically invoked are: the reaction is fast and takes place only in the liquid film; heat and mass transfer resistances for the liquid phase are negligible; the interfacial area is the same for heat and mass transfer; axial dispersion is negligible. The principal model equations are presented as follows.

Overall differential material balance:

Equation 11

Liquid-side mass transfer:

Equation 12

Mass transfer for component A and S:

Equation 13

Equation 14

Interfacial equilibrium:

Equation 15

Heat balance for liquid phase:

Equation 16

Heat balance for gas phase:

Equation 17

Where v_AB denotes stoichiometry; C_P,L is heat capacity of solution (kJ/m³ K); C_P,j is heat capacity of component j in the gas (kJ/kmol K); h_G is heat transfer coefficient (kJ/m^{2 s} K); H is Henry’s constant (kPa m³/kmol); H_R is heat of absorption and reaction (kJ/kmol); H_S is heat of vaporization of solvent S (kJ/kmol); I is enhancement factor; k_G is gas mass transfer coefficient (kmol/m^{2 s} kPa); k_L^O is physical liquid mass transfer coefficient (m/s); T is temperature (K); and Y is mole ratio. Five components are presented in a subscript which is comprised of an inert carrier gas (I), an acid gas (A) (i.e. CO₂), an inert liquid solvent (S) (i.e., water vapor), a reactant (B) in the liquid and a reaction product (C).

To determine the packing height for a given absorption duty, the above equations must be solved concurrently. Since the equations are nonlinear, numerical methods need to be employed. For designing an absorber operating in countercurrent mode, the temperatures and compositions of the lean absorbent and raw feed gas are usually known while the properties of the off gas and rich solution are partially specified. As a result, the conditions at either end of the absorber are known and a two-point boundary value problem can be defined. Before the model equations can be evaluated, various parameters (e.g., mass transfer coefficients, solubility, reaction rate constant and so on) are required. They may be obtained from experimental measurements or correlations. However, considerable discrepancies between reported data must be taken into account and they can strongly effect the results [34,35].

For physical gas absorption data, correlations are available for some standard systems and conditions. When correlations are unavailable, the data have to be determined experimentally. The physicochemical data such as solubility, diffusivity and rate constants are also required for the case of absorption with chemical reaction. These data have to be obtained for each system at the specific conditions. Although the reaction kinetics and the equilibrium solubility may be obtained experimentally, performing these experiments is time consuming.

Even though the CO₂-amine systems have been studied for many years and are common in gas-sweetening industries, the understanding of the reaction kinetics is still required. Some researchers agree on the reaction mechanism but even then, they find that the rate constants may differ by orders of magnitude [36]. High discrepancies can also appear in the equilibrium solubility data. Austgen et al. (1989) reported that the differences among the data reported in the open literature for the CO₂ amine system can be as high as 50% [37]. It is also difficult to measure the diffusivity and solubility for CO₂ absorption with chemical reaction owing to the fact that diffusion and reaction take place simultaneously in the liquid phase. Besides the experiments, solubilities and diffusitivities of gases in liquids can be estimated by using the method developed by Morsi and Charpentier (1981) [38]. The most important step in the theoretical design method is the determination of the enhancement factor, I, which is a function of the reaction mechanism and the properties of all components in the liquid film. To calculate the enhancement factor, a set of partial differential equations have to be solved with the simultaneous diffusional mass transfer and chemical reactions in the liquid film. The 1D diffusion equation for species j can be defined by:

Equation 18

Where Dj, rj, x and t represent diffusitivity (m2/s), rate of reaction of the solute gas in the liquid film (kmol/m3 s), distance from gas-liquid interface (m) and contact time (s), respectively.

The simplified enhancement factor can be obtained from the empirical models proposed by Rase (1977) and Astarita (1983) (for instantaneous irreversible reaction) [39,40], DeCoursey and Thring (1989) (for irreversible reaction of finite rate) [41], Perry and Green (1984) (for first-order and pseudo first-order irreversible reactions) [27] and Decoursey (1982), Decoursey and Thring (1989), and Winkelman et al. (1992) (for reversible reaction) [41–43].

Laboratory Method

According to the previous discussion, physicochemical data and enhancement factors are difficult to identify and, consequently, laboratory models for designing gas absorbers with chemical reaction have been proposed [44–47]. The step of this method is demonstrated in Figure 2 in Route #3. The rate of absorption per unit interfacial volume or mass transfer rate (R_v, kmol/m³ s) is assumed on the basis of the absorber hydrodynamics and the concentrations of reactants in gas and liquid phases. Figure 4 illustrates a small cell that is a gas–liquid contactor at steady state, and two stirrers that keep the concentrations in the bulk of the gas and liquid phases to be uniform.

Equation 19

The advantage of this method eliminates having to determine parameters such as reaction rate, solubility and diffusivity. It is noted that such a model can be used when the reaction rate is fast enough to take place in the liquid film near the interface, which can imply no reaction in the bulk of the liquid. Alper and Danckwerts (1976) developed a new model, the so called complete model [45]. This model was developed from string-of-sphere column as given in Figure 5[45]. The absorber height by using this model can be calculated by:

Equation 20

Where subscript m(_m) respectively represents the model column.

Pilot plant technique method

This method was proposed by Tontiwachwuthikul et al. (1989–1994) [18–20]. The method is intended to scale up a pilot plant or bench-scale plant to an industrial scale plant as illustrated in Figure 6[18–20]. However, such a method needs to assume that hydrodynamic characteristics of the pilot plant or bench-scale plant are the same as those of the industrial scale plant. The material balance around the absorber is given by:

Equation 21

There are several methods to determine R_v. However, most of them cannot be used when certain parameters are unknown. Using this method can help to determine R_v by assuming it as a function of the fluid concentration as given by Equation 22.

Equation 22

R_v can also be obtained from the absorption rate from the liquid composition profile as follows:

Equation 23

The absorber height can then be calculated by Equation 19. This method does not require the hydrodynamic and physicochemical parameters. The step of this method is demonstrated in Figure 2 in Route #4. Another advantage of this method is that it is applicable to all cases regardless of whether the rate of absorption is obtained by the reaction in the liquid film or in the bulk of the liquid or both, and no matter how fast or slow the reaction rate is.

The PPT approach has been successfully applied to a packed tower in which CO₂ is absorbed into NaOH solutions and aqueous solutions of 2-amino-2-methyl-1-propanol (AMP), the most commonly used sterically hindered amine. The validation results were reported in [18–20], which show a good agreement within 12%.

In conclusion, for designing absorbers using reactive solvents, the empirical design method is not recommended for design of the absorber for gas absorption with chemical reaction. This implies that it would carry significant error when the method is used in the CO₂ capture process in which there is a chemical reaction between the absorbent and CO₂. Using this method potentially leads to over design of the absorber height. The theoretical design method is accurate for the CO₂ absorber design since it requires all the involved parameters as discussed previously. However, the calculation step is complicated since it is difficult to obtain all the parameters, especially the hydrodynamic and the physicochemical properties. Lack of knowledge of some of these parameters may cause significant error. Meanwhile, the laboratory method is applicable when the reaction mechanism and kinetics may not be acquired. Nevertheless, a limitation of this method is how laboratory experimental setup (e.g., a stirred cell) can be represented as a full-scale absorber.

However, for many practical situations, very little is known about the system (e.g., CO₂ absorption into mixed reactive solvents using structured packed towers). Under such conditions, the use of the PPT is advantageous and practical in real industrial environments [18–20].

Process design

A comprehension of experimental pilot plants becomes essential to develop the process flow diagram of CO₂ capture for use in commercial or large scale plants. Therefore, this section gives some examples of the model validation results obtained from four different pilot plants: the International Test Centre for CO₂ Capture [103], the Esbjerg CASTOR Pilot Plant (DONG), Institute of Thermodynamics and Thermal Process Engineering (ITT), Stuttgart pilot plant, and the SINTEF/NTNU pilot plant. This review discusses the validation of temperature, CO₂ concentration profile, CO₂ mass balance and reboiler heat duty. The experimental results are validated with simulation results by means of the coefficient of correlation (i.e., R² coefficient). The average absolute deviation (percentage AAD) of each parameter output is discussed. The process validation provides the necessary confidence to scale up the pilot plants to large scale plants. It should be noted that the full details of the four pilot plant simulations using Aspen, ProMax, CHEMASIM, ProTreat and CO2SIM are available in Luo et al. (2009) [15] under the consortium of International Cooperation and Exchange CO₂ Capture using Amine Processes (CAPRICE) [104]. Finally, a discussion of the integration of large-scale CO₂ plant to coal-fired and natural gas-fired power plant is discussed in next subsection.

▪ Brief description of each pilot plant

The schematic diagrams for the four pilot plants are given in Figure 7A and 7B[15]. Figure 7A demonstrate the schematic diagrams of the International Test Centre of CO₂ Capture (ITC), Esbjerg CASTOR and ITT pilot plants. Those three pilot plants have in similar design concept. They have an absorber composed of 2 sections (i.e., wash water and absorber sections) packed with packing material. A wash water section, on top of the top section, is use to cool down the solution to minimize monoethanolamine (MEA) loss. Treated gas (V3) is released at the top of the absorber while the liquid solution flows downwards to the bottom of the absorber. A rich amine pump (P1) is used to increase the liquid pressure before flowing through the lean-rich heat exchanger (HEX1) and then the regenerator. The regenerator is packed with packing material. At the top section of the regenerator, there is a condenser that condenses the regenerator overhead, in order to aid the separation of the condensed stream into the reflux stream (mainly containing water [L10]) and the CO₂ product (V4). The reflux stream is subsequently recirculated back to the regenerator. A reboiler at the bottom of the regenerator is used to heat the liquid solution. The lean amine solution (L2) is cooled down by a cooler (HEX2) and in turn is introduced back to the absorber. Figure 7B shows the schematic diagram of SINTEF/NTNU pilot plant. The conceptualized design of this plant is similar to ITC, Esbjerg CASTOR and ITT pilot plants as previously mentioned but this plant has no wash water section on the top section of the absorber. In addition, the process uses recycled gas from the condenser (HEX3) outlet to return to the top section of the absorber. This condenser also works as a separator that separates liquid (L11) to the bottom tank. The liquid directly flows to the reboiler. It should be noted that these four different pilot plants used 30% wt. MEA as the chemical absorbent to capture CO₂. A brief summary of each plant configuration is given in Table 1.

Modeling approach

This review article shows only the ProMax simulation for four different pilot plants. The use of other software, which includes Aspen Rad Frac, Protreat, Promax, Aspen Rate Sep, CHEMASIM and CO2SIM is not discussed in the paper. However, its detail can be found in our collaborative work [15]. All simulations appearing in this article were performed at the ITC in the University of Regina, Canada. Four datasets from each of the ITC and SINTEF/NTNU pilot plants, and one data set from each of the other two plants (Esbjerg CASTOR, and ITT Stuttgart pilot plants) were used to validate the simulation results.

The parameter inputs used in the simulation are summarized as follows:

	▪ Configurations of absorber and regenerator (e.g., packing type, flooding factor, system factor, residence time of liquid holdup and pressure drop), and configurations of pumps, intercoolers and heat exchanger (e.g., pressure and efficiency). These parameters are assigned by each individual pilot plant design.
	▪ Temperature approach between rich amine inlet and lean amine outlet in lean-rich heat exchanger (LRX). It is to be noted that the simulation does not fix or assign any lean and rich solution temperature inlet and outlet in LRX, absorber and regenerator. These temperatures are calculated on the basis of thermodynamics and chemical reactions.
	▪ Reflux temperature at the top section of the regenerator. The temperature is set by the process parameters obtained from each four different pilot plants.
	▪ Capture performance. This is assigned to control the percentage of CO2 capture capability between absorber inlet and outlet. It should be noted that the liquid flow rate and CO2 loading of lean and rich solutions are not assigned as process inputs. These parameters are optimized by the capture performance.

▪ Simulation results

ITC pilot plant

Figures 8A–11A demonstrate the CO₂ concentration profile of the 10-m absorber height of the ITC pilot plant. It should be noted that all figures set the top section of the absorber as the beginning level (z = 0). The figures show that the CO₂ concentration obtained from simulation is lower than the experiment. In addition, the simulation roughly predicts the CO₂ concentration profile in the washing water section but predicts the profile in the absorber section well. Both simulation and experimental results are in very good agreement. Figures 8B–11B demonstrate a validation of temperature profile between simulation and experimental results. It was found that both results are in a good correlation.

Figures 12A–12D shows the parity plot of CO₂ concentration, temperature profile, CO₂ mass balance and reboiler heat duty between simulation and experimental results. All parity plots have R² greater than 0.8. Particularly, the correlation of CO₂ mass balance and reboiler heat duty has an R² higher than 0.95. This implies that the simulation can predict the experimental process outputs obtained from the ITC pilot plant.

Esbjerg CASTOR pilot plant (DONG)

Figure 13A and B demonstrate the validation of the CO₂ concentration and temperature profile of the 20 meter absorber height of the Esbjerg CASTOR pilot plant [15]. The simulation results are slightly lower than the experimental results but their trends are consistent.

Figures 14A–D show the parity plot of CO₂ concentration, temperature profile, CO₂ mass balance and reboiler heat duty. The R² of all plots is relatively high. This indicates that the process simulator gives results that are able to predict the actual experimental process parameters in the Esbjerg CASTOR pilot plant.

ITT Stuttgart pilot plant

Figures 15A and 15B show the validation of CO₂ concentration and temperature profile for the 4.6 meter absorber height of the ITT pilot plant [15]. The simulation gives a slightly lower CO₂ concentration and temperature profile than the experimental results. At the top of the absorber, the temperature is slightly inconsistent (Figure 15B). The temperature outlet of the experiment is 40°C while that of the simulation is approximately above 50°C.

Figures 16A–D show the parity plot of CO₂ concentration, temperature profile, CO₂ mass balance, and reboiler heat duty. The CO₂ concentration, reboiler heat duty obtained from the simulation are in an excellent correlation (R² = 0.99) with the experimental result whereas the temperature profile and CO₂ mass balance do not properly correlate with the experimental result (R² < 0.67). The poor simulation results rom the ITT pilot plant might be attributed to the following:

	▪ Over simplification of the schematic diagram of the model used for simulation as compared with the actual pilot plant, which is more complicated (e.g., extra hidden equipment);
	▪ The positions of the temperature probe attached along the absorber height in the pilot plant may not be the same as in the model.

SINTEF/NTNU Pilot Plant

Figures 17A–D demonstrate the temperature profile for the 4.4 meter absorber height of the SINTEF/NTNU pilot plant. All four data sets have been selected to further validate the simulation results with the experimental results since it has no CO₂ concentration profile recorded by SINTEF/NTNU. The figures show that temperature profile obtained from the simulation is obviously lower than the experiment. However, the temperature at the bottom of the absorber for both simulation and experiment are comparable. In general, both simulation and experiment are consistent with R² of 0.90 (Figure 18A).

Figures 18B and C demonstrate the parity plot of CO₂ mass balance and reboiler heat duty. It was found that both simulation and experiment are excellently correlated with an R² above 0.99.

The simulation using ProMax gave satisfactory results. The deviation between simulation and experimental outputs for all parameters from the ITC, Esbjerg Dong and SINTEF/NTNU pilots were small (ADD = 10.9%) whereas the deviation between simulation and experimental outputs for ITT were high for CO₂ mass balance and temperature (ADD = 43.7%) but small for other parameters (ADD = 9.35%).

A comparison of the simulation and experimental results from all the four pilot plants are plotted in the same parity plot in Figure 19. The average R² and %AAD for the four pilot plants were measured for various parameters. The average R² and%AAD for CO₂ concentration were 0.95 and 27.9%, respectively (Figure 19A); for temperature profile, they were 0.84 and 5.0%, respectively (Figure 19B); for CO₂ mass balance, they were 0.97 and 15.0%, respectively (Figure 19C); and for the reboiler heat duty, they were 0.99 and 4.4%, respectively (Figure 19D). These imply that this process simulator could be used to simulate the pilot plants.

Process integration

The post-combustion CO₂ capture using amines is commonly integrated into large point CO₂ sources of which the process captures CO₂ after the combustion has taken place in fossil fuel-fired power plant (e.g., coal-fired power plant or natural gas-fired combined cycle gas turbine [CCGT]) [5,6]. The steam is extracted from a series of steam turbines in the power plant to the reboiler in the CO₂ capture process. Figures 20A & 20B show an example of steam extraction from coal-fired and natural gas-fired power plants, respectively [16,48]. According to the figures, superheated steam is extracted from the crossover pipe between the intermediate pressure (IP) and low pressure (LP) ports. The steam as a thermal energy source is fed to the reboiler located at the bottom of the regenerator. Such energy is used to obtain a lean-CO₂ loaded amine solution from a rich-CO₂ loaded amine solution. Then, the used steam (as condensate) is routed back to the deaerator in the power plant.

Apart from extracting the steam from the crossover pipe between the IP and the LP ports as shown in the figures, another option is to extract superheated steam from the LP port. It is found from Romeo et al. and Sanpasertparnich et al. that extracting the steam from the LP port offers a lower heat duty for regenerating the amine by stripping the CO₂[13,16]. A high pressure steam that can still be used as a working fluid to drive the turbine, resulting in a significant reduction of the energy penalty to regenerate the rich-CO₂ solution. However, Alie proposed that a suitable and realistic extracting location should be in the crossover pipe between the IP and LP ports (Figure 21)[9]. Extracting the steam from LP port would be inaccessible since the port is located on the underside of the turbine [9]. Even though, currently, there is a physical limitation in extracting the steam from LP port, the authors realize that there are two possibilities to extract the steam from the LP turbine as illustrated in Figure 22:

	▪ A new design of low pressure steam turbine, which offers an accessible steam extraction location;
	▪ A split of the low pressure steam turbine into more than one, such that the steam can be extracted from LP/LP port.

The first option requires a turbine manufacturer to redesign the LP turbine casing and space of turbine blades to allow an insert pipe for steam extraction. In the case that the first option is not feasible, instead of extracting superheated steam from the crossover pipe in between the IP and LP ports, the splitting of one LP turbine into two turbines and extracting the superheated steam from the crossover pipe between the two LP ports would be practical.

In addition to the location of the steam extraction, the quality and quantity of the steam from the turbine are important parameters one should be aware of. The precautions include:

	▪ The extracted steam temperature from the turbine should not be above the temperature that can cause CO2 absorbent to undergo thermal degradation and/or corrosion [9,49]. If the temperature is over the limit, an extra unit such as de-superheater unit must be integrated before the reboiler [10,50];
	▪ Extracting possibly the lowest quality steam is advantageous, since high quality steam offers higher power generation [9,16];
	▪ The steam flow rate supplied to the reboiler must be sufficient to regenerate CO2 absorbent and to reach the percentage of CO2 capture performance design.

Future perspective

It is essential to have a comprehensive knowledge of how to design, model and simulate the post combustion CO₂ capture system based on regenerable amines. Unless the equipment and process are systematically designed, the heat consumption as well as the CAPEX and OPEX for the CO₂ capture plant will be very high. There are many companies and organizations that focus only on modifying the process and techniques used in gas purification processes to fit the CO₂ capture process. Examples are split flow, combined concurrent-counter current and intercooler configurations [21,51]. The development of compression ratio to reduce multiple stages of CO₂ compressors and utilizing heat from the compression for amine regeneration has been proposed. Advanced design of packed column for the absorber and regenerator to enhance gas–liquid contacting time and mass transfer coefficient is also competitively developed. In addition to these issues, improvements in CO₂ capture efficiency and amine-regeneration efficiency, as well as the developments of effective inhibitors to reduce solvent degradation, corrosion and foaming should be pursued in order to actualize the CO₂ capture process using absorbents.

Commercial process simulators as well as in-house softwares are continuously being developed by embedding the improvements in certain aspects of the process technologies and providing a user-friendly interface. The technology would then support any user who may not have the background on the CO₂ capture system design. This would increase risk, uncertainty and unreliability when such designs are implemented. As such, it is important that all simulations must be capable of representing and validating any operating condition and equipment dimensions before users scale up the process to integrate into fossil fuel-fired power plants. Some simulation software companies have not programmed their software based on theoretical designs and existing pilot plant data, making such software unreliable. For instance, the programmer may put incorrect thermodynamics calculation algorithm in the software, or obsolete thermodynamics databases. These can lead to poor engineering design integrity. Therefore, any engineering consulting company must have specialists in this particular field. Therefore, more pilot plants duplicating up-scale system would be built within this decade. Collaborations to share the pilot-plant data would be seen soon. The integration of regenerable solvent based CO₂ capture system into existing or new fossil fuel-fired power plants will possibly occur in this decade, but it would be only for partial capture defined as the capture of CO₂ from less than 100% of the total flue gas stream from the power plant. A full integration would be implemented when regulatory policy is in place and will be enforced, and the government confirms that it will compensate power producers for the cost of capturing CO₂. Full integration will also be helped by the development of cost-effective, process-efficient, environmentally safe and operation-reliable CO₂ capture technologies.

Table 1.  Brief description of process parameters in four different pilot plants.

Packing type	ITC, Regina	CASTOR, Dong	ITT, Stuttgart	SINTEF/NTNU
Wash water section	Flexipac700Y	Mellapak252Y	Mellapak250Y
Absorber section	Flexipac700Y	IMTP50	Mellapak250Y	Mellapak250Y
Stripper section	Flexipac700Y	IMTP50	Mellapak250Y	Mellapak250Y
Wash water diameter/height (m)	0.33/2.93	1.1/3	0.125/0.42
Absorber diameter/height (m)	0.33/7.05	1.1/17	0.125/4.2	0.15/4.36
Stripper diameter/height (m)	0.33/9.97	1.1/10	0.125/2.52	0.1/3.89

Data from [15].

Amines

Organic compounds containing the amino functional group with one or more alkyl or aryl groups attached to the nitrogen. They can be classified as primary, secondary and tertiary amines depending on the numbers of alkyl or aryl groups attached to the nitrogen of the amino functional group.

Front-end-engineering design (FEED)

A conceptual process design for industry and engineering projects. It includes material and energy balances, 2D/3D models, equipment designs and lists, process layout, schedule/plan, estimate, purchase-ready equipment specification and so on. Essentially, engineering project undergoes with these steps: pre-FEED; FEED; detailed design; procurement; construction; testing; and commissioning.

Steam turbine

Prime mover used to convert dynamic and pressured steam into energy by turning blades of a series of turbines; for example, high pressure turbine (HP), intermediate pressure turbine (IP) and low pressure turbine (LP).

Executive summary

	▪ Of post-combustion CO2 capture technologies, CO2 capture-based regenerable solvent is the most mature and promising technology. In addition, post-combustion CO2 capture based on regenerable solvent is best suited to capture CO2 at a low partial pressure of approximately 4–15% by volume of CO2. Consequently, the technology can be integrated into pulverized coal-fired, natural gas-fired and oil-fired power plants.
	▪ It is clear that regenerating CO2-rich amines to CO2-lean amines consumes significant heat. This can be reduced considerably by proper process configuration optimization.
	▪ There are many constraints on the modeling and simulation strategy to designing a CO2 capture plant. These include the difficulty to solve numerical modeling equations, establishing mechanisms of chemical reactions, the reliability of properties (e.g., thermodynamic, chemical and physical) of the system components, amongst other things. These limitations make it necessary to further validate innovative process configurations that are only confirmed by simulation using commercial simulators. Several commercial and non-commercial programs are available to model the post-combustion capture of CO2. This includes EBSILON®, GateCycle®, GT Pro®, PROATES®, Steam Pro® (for power plant simulation), Aspen Plus®, Aspen HYSYS®, CHEMASIM®, ProMax®, ProTreat® (for CO2 capture simulation) and GT Pro/Master®, IECM®, Steam Pro/Master®, and ThermoFlex® (for process integration).
	▪ In the design of the absorber, the empirical design method is not applicable when a system has chemical reaction, while the theoretical design method requires a number of involved parameters to calculate the absorber height. The laboratory method is based on a small-scale experimental setup, such as stirred cell and string-of-sphere column, whereas the pilot plant technique method is based on equivalent geometry to a full-scale CO2 absorber but a smaller dimensions. The disadvantage of the laboratory method, therefore, is that it does not represent a full-scale CO2 absorber.
	▪ Process validation using existing CO2 capture pilot plants is necessary to obtain process integrity and confidence for up-scaling the CO2 capture process. ProMax as one of the process simulators is able to validate the experimental data obtaining from: - the International Test Centre of CO2 Capture (ITC) pilot plant; - the Esbjerg CASTOR pilot plant; - the ITT Stuttgart pilot plant; - the SINTEF/NTNU pilot plant.
	▪ Upon the integration of the CO2 capture plant with a fossil fuel-fired electric power plant, it is clear that there are two points for steam extraction from a series of the steam turbines: crossover pipe between intermediate pressure (IP) turbine and low pressure (LP) turbine port and the LP port.
	▪ It is shown that extracting the steam from LP port would be inaccessible. Hence, there are two possible options to extract the steam from the LP turbine: a new design of LP turbine and an LP/LP port by splitting of LP turbine into more than one.

Acknowledgements

We would like to acknowledge the research supports given to to ITC over the past many years by the followings organizations: Natural Sciences and Engineering Research Council of Canada (NSERC), Canada Foundation for Innovation (CFI), Saskatchewan Ministry of Energy & Resources, Western Economic Diversification, EnCana Energy Inc., E.ON Energy, Stantec, Saudi Aramco, Doosan Heavy Industries, HTC Purenergy Inc. Saskatchewan Power Corporation, StatOil Hydro (Norway), SaskFerco Inc., Sulzer Chemtech (Switzerland), Fluor Corporation (USA), the Canada Centre for Mineral and Energy Technology (CANMET), Alberta Energy Research Institute (AERI) and Research Institute of Innovative and Technology for the Earth (RITE).

We would also like to acknowledge the assistants from Luo X, Svendsen HF (NTNU, Norway), Notz R (University of Stuttgart, Germany), Hoch S, Hasse H (University of Kaiserslautern, Germany), Lemaire E, Alix P (IFP, France), Tobiesen FA, Juliussen O (SINTEF, Norway), Köpcke M (Vattenfall, Sweden), Hopman J (TNO, The Netherlands) and CAPRICE consortium members for the pilot plant data, Hendrick C (BR&E, Texas, USA) for ProMax technical support and Saiwan C (Chulalongkorn University, Thailand) for pre-reviewing this article.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.