In experiments, the inlet gas containing 12.5 Vol % CO₂ was prepared. The gas preparation started with the pressure release from the CO₂ gas cylinder (99.9% purity provided by Nanjing special gases Co., Ltd.) to the gas reservoir, and the CO₂ was blown into the gas buffer tank, in which the flow rate of CO₂ was controlled by the pressure relief valve and flowmeter. Air was blown directly from the atmosphere. The air and CO₂ were well mixed and then fed into the absorber at a certain flow rate. The gas composition, temperature, and pressure were measured by Kane KM9106E combustion analyzer (Keison, UK). The 20 wt% aqueous MEA solution was prepared by mixing MEA (Jiaxing Jinyan Chemical Co., Ltd.) and deionized water.

The experimental set-up used for pilot-testing is depicted in Figure 1. Two thermometers (Kangle Instruments Co., Ltd.) were equipped at the top and bottom of absorber and desorber, respectively, to measure the temperature of columns. Similarly, two pressure gauges (XueHu Special Instrument Technology Co., Ltd.) were equipped at the top and bottom of absorber and desorber, respectively, to record the column pressure. The specifications are listed in Table 1. In the absorption-desorption process, CO₂ in the gas stream was absorbed by the 20 wt% MEA solution, and then the CO₂-rich solvent was preheated by a rich-lean heat exchanger (HEX) and sent to the regeneration column. The regenerated CO₂ gas stream was condensed in a condenser to separate water and obtain a 99% CO₂ product. The CO₂-lean solvent from the reboiler was pumped to the rich-lean heat exchanger to recover the heat. The CO₂-lean solvent stream was further cooled and then recirculated to the absorber with a solvent makeup stream to compensate for the solvent loss.

In simulations, the MEA-based capture process combined with a CO₂ compression unit is shown in Figure 2. The inlet gas travels through a pretreatment unit to cool the gas and reduce the water content. The treated gas is then fed into the absorber, where CO₂ is reactively absorbed by the MEA solvent. The CO₂-rich solvent exiting the absorber is pumped into the internal heat exchanger, where the CO₂-rich solvent is preheated and fed into the regeneration column. The CO₂ exiting from the top of the regeneration column then flows into a series of compression units to reach a specific condition for transportation, storage, or utilization. The makeup streams, which contain water and MEA, are added into the recycle stream.

The inlet gas may contain O₂, CO₂, H₂O, N₂, CO, H₂, and several trace components, such as NO_x and SO_x. However, to simplify the process, only O₂, CO₂, H₂O, and N₂ were considered, and the gas was assumed to contain 2.4 wt% O₂, 4.2 wt% H₂O, and balanced CO₂ and N₂ concentrations.

In this study, two gas flow rates were assumed, and the large flow rate was 252.7 ton/hr (i.e., 70.2 kg/s), corresponding to the full capacity of the St. Marys cement plant (Hassan, 2005). The medium gas flow rate was half the large flow rate, representing the capacity of a medium-sized plant. The CO₂ concentration of 31.8 wt% is the original data from the St. Marys cement plant, and then the CO₂ concentration was further expanded to a range of 10–50 wt% for generalizing the study. The CO₂ removal rate was defined as the ratio between the amount of CO₂ captured in the process and the total amount of CO₂ entered the process. It was set between 65% and 95% to investigate its influence on the energy demand and economy (cost), and values lower than 65% were not included owing to the CO₂ capture requirement. These parameters are listed in Table 2. Based on these parameters, 60 cases of various CO₂ removal rates, gas flow rates, and CO₂ concentrations were created for process simulation. Other process parameters are listed in Table 3.

The process development strategy follows the same principle as the study by Penteado et al. (2016). Both columns were designed with the equilibrium model and then switched to the rate-based model that simultaneously described the mass and heat transfer rate phenomena with equilibrium and kinetic controlled reactions. A design specification was set in the absorber to reach a certain capture target by varying the lean MEA flow rate. The other two design specifications were configured in the desorber: one matched the lean CO₂ loading of 0.3 mol_CO2/mol_MEA by varying the reboiler duty, and the other ensured that the gas stream exiting the desorber reached 98 mol% CO₂ purity by changing the distillate flow rate. The heights for both columns were determined by sensitivity analyses, where the required lean MEA flow rate in the absorber and the minimum reboiler duty in the desorber were calculated.

The recycle stream should be closed in the process simulation to achieve process stability and simulation convergence. However, closing the recycle stream by a blocked connection increases the computing time and could create convergence problems. In addition, the labor becomes more complex when adjusting the operating conditions. Thus, in this study, the method described by Penteado et al. was used to improve the stability and usability of the simulation, i.e., instead of connecting the lean MEA recycle stream with the absorber inlet, a transfer block was used to virtually close the recycle. This was performed using two steps. 1) The design specifications were used to fix the lean loading of the bottom stream from desorber, and 2) a balance block was created to control the flow rates of the makeup streams. These two steps ensured that the recycle stream and lean MEA stream were nearly identical in composition and flow rate. The initial specifications of the process are listed in Table 3. It is worth noting that IMTP 1.5-IN NORTON was chosen in this work because of the following reasons: 1) the surface area of this type of packing is available in the database, and 2) a similar IMTP packing was used in the pilot plant, where the data was used for the validation of process simulation. The comparison with the experimental data using a similar packing will increase the reliability of holdup calculation, leading to credible results.

The simulation results depend on the accuracy of the properties, phase equilibria, mass and heat transfer, and reaction kinetics. Proper models were chosen to calculate the required thermodynamic properties (enthalpy, entropy, Gibbs free energy, and volume) and transport properties (viscosity, thermal conductivity, diffusion coefficient, and surface tension) in the liquid and vapor phases. The thermodynamic and transport properties were extracted from the Aspen Plus database and configured in the component specifications.

The electrolyte non-random two liquid model was used to describe the non-ideal behaviors of the liquid phase, and the Redlich-Kwong equation of state was chosen for the vapor phase to describe the phase equilibria (Liu et al., 1999; Zhang et al., 2009; Zhang and Chen, 2010). Both models have been implemented in Aspen Plus and verified for MEA-based technologies for CO₂ capture (Chalmers and Gibbins, 2007; Thiele et al., 2007; Hi et al., 2015). The parameters (pure, binary, and electrolyte-pair) for all the components were maintained as default values, as this simulation was conducted based on the example file used for the MEA-CO₂ system.

The widely used reaction mechanism described by Freguia and Rochelle (2003) with three equilibrium reactions and two reversible kinetically controlled reactions was adopted, and the corresponding parameters were obtained from Austgen et al. (1989), Pinsent et al. (1956), and Hikita et al. (1977). The reactions and corresponding kinetic parameters are listed in the Supplementary Table SA).

For the mass and heat transfer calculations, the Onda-68 correlation was selected for the mass transfer coefficient and interfacial area. The Chilton and Colburn correlation was applied to obtain the heat transfer coefficient, and the packing holdups were calculated using the approach by Bravo et al. (1992).

Using the experimental set-up described in Figure 1, the key experimental results of CO₂ capture, i.e., CO₂ removal rate, desorber bottom section temperature, and gas-to-liquid ratio, with 20 wt% MEA are listed in Table 5. This is the first time to report the data with 20 wt% MEA, and no comparison with other available data can be conducted.

The newly determined experimental results were used to validate the process simulation results, where the simulations were performed under the same conditions used at the real pilot plants. As shown in Table 5, a comparable simulation output proved a reliable process model in our study with 20 wt% MEA solvent. However, due to the undesirable insulation on the regeneration column, the reboiler heating duty recorded in the experiment was higher than the simulation results but still in an acceptable range if the heat loss was considered.

In 3.1.1, the simulation results were validated with 20 wt% aqueous MEA. Moreover, the experimental results with 30 wt% MEA solutions taken from the pilot plants at the University of Texas at Austin and the University of Kaiserslautern (Dugas, 2006; Notz et al., 2012) were used for further validation.

For 30 wt% MEA, again, the simulations were conducted under the same conditions used at the real pilot plants. The comparison results are listed in Table 6 and depicted in Figure 3, showing that good agreements were obtained between the simulation output and experimental data. The minor difference in the reboiler heating duty from that in the pilot plant data at the University of Texas at Austin can be explained by the re-absorption in the desorber, which was also reported by Zhang et al. (2009).

The lower and upper boundaries of removal rate were selected to illustrate the effects of MEA concentration and removal rate on specific energy demand and capture cost with fixed gas flow rate (large scale) and CO₂ concentration (31.8 wt% CO₂ concentration). As we can conclude from Figure 5, no considerable effect of removal rate is observed even if the removal rate increases from 65% to 95%. However, the specific heat requirement as well as the costs are sharply decreasing with increasing MEA concentration from 15% to 20%, and then a slight decrease and increase are observed with increasing the MEA concentration from 20% to 25% and from 25% to 30%. It can be explained that when using too low MEA concentration (e.g., 15 wt% MEA), it will lead to a large solvent flow rate and thus expand the equipment dimension and higher CAPEX. The heat demand required for heating up the solvent is also high, resulting in a sharp increase in OPEX. When increasing the MEA concentration into 20–25 wt%, a lower solvent flow rate could reach the fixed capture target and lower the heat requirement compared to that of 15 wt% MEA case. The higher bulge temperature in the absorber is insignificant compared with enhancement due to the decrease in solvent flow rate. When a high MEA concentration (30 wt% MEA) is used, a lesser solvent flow rate is required. However, the bulge temperature is too high to maintain the desired CO₂ solubility, again resulting in a larger solvent flow rate for compensating the solubility losses. Even though the solvent flow rate is still less than that for the case with a lower MEA concentration, the heat duty is a bit higher and not optimal in this case.

Similar to the evaluation of removal rate, the boundary value of CO₂ concentration, 10 and 50 wt% were selected to study the effect of MEA concentration under different CO₂ concentrations, as depicted in Figure 6. When the CO₂ concentrations were set at a 10 and 50 wt%, the lowest specific energy demands were reached for the solvent with 25 wt% and 20 wt% MEA concentrations, respectively. This is highly related to the CO₂ concentration. When using 30 wt% MEA with 10 wt% CO₂ concentration, the higher bulge temperature is avoided, leading to an insignificant difference between 25 and 30 wt% MEA. However, when using 30 wt% MEA with 50 wt% CO₂ concentration, the high bulge temperature is promoted, resulting in an obviously increased specific energy demand compared with the case of 25 wt% of MEA.

Moreover, the lowest investment costs were reached for the solvent with 25 wt% and 20 wt% MEA concentration when using 10 and 50 wt% CO₂ concentrations, respectively. The conclusions above indicate the significant effects of CO₂ concentration on both energy demand and investment cost. Therefore, the solvent concentration should be carefully considered with CO₂ concentrations.

We can conclude that the MEA concentration of 20–25 wt% is the optimum range in our study instead of 30 wt% MEA, and a concentration lower than 20 wt% will significantly increase the reboiler duty. Besides, other researchers also drew a similar conclusion when the study methods are customized; Garđarsdóttir et al. (2015) also studied the solvent concentration effect on the temperature profiles and heat duty that lower concentrations of MEA could benefit the bulge temperature in absorber and reboiler heat duty, especially under low CO₂ concentration condition. In addition, the pilot testing also supports that the MEA concentration lower than 30 wt% requires lower stripping steam in the regeneration process (Notz et al., 2012). Furthermore, a lower MEA concentration solvent benefits a lower degradation and corrosion rates, and reduces the equipment, maintenance, and solvent costs. Based on all the analyses in this section, 20 wt% MEA solvent was selected for further study in this work.

First, the energy demand and variance with the studied parameters were evaluated for the process with and without the compression unit. The heating duty, cooling duty, and electric power were analyzed to identify the energy-intensive units, and then the effect of the parameters on these units was quantitatively evaluated. In this section, a number of examples are illustrated to represent the effects of parameters, and other cases have shown the same trend as the cases presented in the following sections.

Figure 7 shows the results of specific energy power for different cases as well as with and without the compression unit. The specific energy power in the high CO₂ concentration cases was lower than that in the low CO₂ concentration cases. For example, with a 65% CO₂ removal rate, the heating duty was 0.14 kW/ton-CO₂ for 10 wt% CO₂ concentration, while it was 0.105 kW/ton-CO₂ for 50 wt% CO₂ concentration. Thus, the specific energy power decreased with increasing CO₂ concentration. In addition, the energy power decreased with increasing CO₂ removal rate, except for the cooling duty. For the cases with the compression unit, the cooling duty showed a fluctuation with increasing CO₂ removal rate owing to the design of the multistage compressors. The CO₂ streams contained different volumetric flow rates and temperatures, requiring various compressor setups (outlet pressure and temperature in each stage) and different cooler sizes in each stage, and influencing the overall process cooling duty.

The specific energy powers include cooling, heating, and electric power. The cooling duty, which is used in pre-stage cooling and the condenser in the desorber, requires the largest percentage of energy power. There was no difference between the heating duty with and without the compression unit, because the heating duty was only owing to the reboiler. The value of electric power was lower than the other two energy powers, especially without the compression unit. The significant difference between the electric powers with and without the compressors indicates that the compressors are the main electricity requirement. The summation of heating and cooling duty accounts for the most part of the energy power required in the process, indicating that consideration of waste heating and cooling management might be interesting.

Figure 8 shows the variations of the total energy cost with the considered parameters as well as with and without the compression unit, where the observations are different from those in Figure 7. The different results observed from Figure 7 and Figure 8 highlight the importance of energy price in analyzing the economics rather than the energy demand. In contrast to Figure 7, the cost of cooling energy in Figure 8 is insignificant compared to the overall energy cost. Moreover, the energy cost without the compression unit showed at least a 20% cost reduction from that with the compression unit; therefore, the compression unit played a significant role in the energy demand. In addition, the total energy cost slightly decreased with increasing CO₂ removal rate in each case; however, the reduction was large for increasing CO₂ concentration. For instance, for 10 wt% CO₂ with the increased CO₂ removal rate, the reductions were 7.3% (with compression) and 7.5% (without compression), while the cost reductions owing to the increased CO₂ concentration were 26.4%–30.4% and 33.7%–37.9%, respectively.

Figure 8 also shows the energy costs of heating duty, cooling duty, and electric power. The heating duty, which is owing to the reboiler, accounts for 50% of the total energy cost in each case. For 10 and 50 wt% CO₂, the cost of the heating duty was slightly reduced: 2.7% and 1% with the increased CO₂ removal rate, while increasing the CO₂ concentration from 10 to 50 wt% created a 23%–24.4% reduction. Therefore, the CO₂ concentration had a larger effect on the cost reduction of the heating duty than that of the CO₂ removal rate. Large cost reductions in electric power were observed by removing the compression unit. For 10 wt% CO₂ concentration, 52% and 62% reductions were observed for the 65% and 95% CO₂ removal rates, respectively. These numbers were larger for 50 wt% CO₂ concentration: 86% and 87% reductions for the 65% and 95% CO₂ removal rates, respectively. When the CO₂ concentration increased from 10 to 50 wt%, the additional electric power for the gas blower and solvent pumps was smaller than the reduction of the electric power for the compression unit. Therefore, the compression unit dominated the overall electric demand, especially for high CO₂ concentrations.

The effect of the CO₂ concentration and CO₂ removal rate on the electric power was further evaluated for the cases with and without the compression unit. For the cases with a compression unit, when increasing the CO₂ removal rate from 65% to 95%, 14.9% and 4% cost reductions were observed in the 10 and 50 wt% CO₂ cases, respectively. However, the effect of the CO₂ concentration achieved 31.9%–39.5% cost reductions when increasing the CO₂ concentration from 10 to 50 wt% for the CO₂ removal rates ranging from 65% to 95%. For the cases without the compression unit, when increasing the CO₂ removal rate from 65% to 95%, 30.9% and 30.3% cost reductions were observed in the 10 and 50 wt% CO₂ cases, while the CO₂ concentration showed a larger effect than that of the CO₂ removal rate: 82.3%–82.5% cost reductions when increasing the CO₂ concentration from 10 to 50 wt% with the CO₂ removal rates of 65%–95%. In both scenarios, the CO₂ concentration had a larger impact than that of the CO₂ removal rate on the cost reductions in the electric power. The cooling duty represented a small percentage of the energy cost, and the absence of a compression unit showed a minimal effect. In addition, the percentages of heating, electricity, and cooling costs to the total costs were plotted in cheese portion diagrams in the Supplementary Figure S1 to visually demonstrate the energy cost percentages when changing the defined parameters.

The energy analysis is related to the operational cost. For the entire plant, the investment cost is another important concern. In this section, the effects of individual and multiple parameters on CAPEX, OPEX, and total cost were discussed. In addition, the estimated costs for the MEA-based CO₂ capture plant for all cases were summarized and plotted in a single figure, where CAPEX and OPEX are shown for the combinations of considered parameters. Beforehand, the sensitivity analysis of the impact of utility cost on total cost was conducted. The results are shown in the Supplementary Figure S2. The utility costs were varied for 50%, 75%, 125%, and 150% of its original cost adopted from Table 4. Four sets of cases, i.e., large/medium flow fate with 65%/95% CO₂ removal rate, were selected to represent how the variations of these utility prices affect the calculated costs. The results showed that the heating cost affected the total cost in the most effective way rather than the electricity cost, and the cooling cost only slightly changed the total cost. For example, in the case of large flow fate with 65% CO₂ removal rate, 50% increase in heating cost caused 12.9% increase in the total cost, while 50% increase in electricity and cooling costs led to 8.4% and 1% increases in the total cost, respectively.