Performance Evaluation of Phase Separation Process Using High-concentration AMP Promoted by MAPA for CO2 Capture

Reduction of the energy penalty and cost of CO2 capture from concentrated gas streams using amine-based solutions can be achieved by minimizing the energy penalty in the solvent regeneration process. High concentration 2-Amino-2-methyl-1-propanol (AMP) solution precipitates as a carbonate when enough CO2 has been absorbed. By sending the separated carbonate to the stripper, the sensible heat of regeneration can be reduced. However, previous testing using 50 weight percent AMP solution mixed with Piperazine (PZ) with solid-liquid separation showed that the CO2 recovery rate was limited to 65% due to the lack of PZ regeneration. To improve the CO2 recovery rate, a novel solution and injection process were developed. N-Methyl-1,3-diaminopropane (MAPA) was selected as an alternative promoter based on reaction rate testing. Various tests were employed to characterize the behaviour of the AMP/MAPA solution under CO2 capture and recovery conditions. The injection point was relocated to avoid the inhibition of CO2 absorption observed when CO2 semi-lean liquid was sent to the upper portion of the absorber. The CO2 recovery rate and the precipitation quantity were simulated using a model built in Aspen Plus®. The novel solution and injection set-up were evaluated experimentally by a bench-scale apparatus.


Introduction
Carbon Capture and Storage (CCS) is widely considered as a necessary technology to meet the Paris Agreement's "well below 2°C above pre-industrial levels and pursuing efforts to limit the temperature increase to 1.5°C" (UNFCCC, 2015;IPCC, 2018). For capture of CO2 from large scale emission point sources (e.g., thermal power stations), chemical absorption is the most widely studied and mature technology. While technically proven, the chemical absorption method has the drawback of a large energy penalty owing to the regeneration of the solvents used to capture CO2. To maintain financial viability while reducing CO2 emissions, further process improvement and innovation are warranted. Regeneration heat is roughly divided into three sources: sensible heat, heat of vaporization, and heat of dissociation. The sensible heat is expected to be reduced via a phase separation of CO2rich and CO2-lean amine. By sending only the CO2-rich phase to stripper, the quantity of solution circulating in stripper decrease, thereby reducing the sensible heat load.
Phase separation processes commonly researched in the CCS chemical absorption field are chilled ammonia method, liquid-liquid separation method, and solid-liquid separation method. This study utilizes the solid-liquid separation method due to its higher safety and more rapid kinetics. In the solid-liquid separation process, a high concentration amine solution is used and a precipitated solid of protonated 2-Amino-2-methyl-1-propanol (AMP) carbonate (henceforth: carbonate) is formed. Only this solid carbonate is sent to stripper .
High concentration AMP solution of ≥40wt% results in precipitation of solid carbonate upon CO2 absorption. The CO2 absorption rate of AMP is slow due to steric hindrance. However, the absorption rate of AMP is accelerated by adding a small quantity of amine with a high reaction rate with CO2 (e.g., Piperazine: PZ) as a promoter (Ying et al., 2017). To confirm whether the such results remain in solid-liquid type processes, CO2 capture and recovery tests were performed using AMP 50wt% promoted with PZ 5wt%. Figure 1 shows the operating conditions over the process steps of the CO2 capture and recovery test (Nakagaki et al, 2015). The arrow corresponds to each unit operation ordered by ordinal numbers which indicate condition of the CO2 loading and temperature of the circulating solution. A solvent rich in CO2 is cooled at the bottom of the absorber column and precipitates as carbonates. A slurry of solid carbonates mixed with a small amount of liquid amine is separated by a filter-type centrifuge, the solids are dissolved by a mantle heater, and pumped to the stripper. The separated liquid phase (CO2 semi-lean liquid) is mixed with lean liquid regenerated in the stripper and returned to the upper portion of the absorber column. In our previous study, the operating conditions and equipment setup to prevent blockage by precipitates were designed (Teranishi et al., 2016) and using this setup with AMP 50wt% promoted with PZ 5wt%, the sensible heat decreased by 15-27% compared to a liquid phase process using AMP 30wt% (Ogiyama et al., 2017). However, the CO2 recovery rate was limited to 65% due to the lack of regeneration of PZ in the stripper.
This study aimed to improve the CO2 recovery rate by searching for effective promoter of absorbent and modifying process configuration. The candidates of effective promoter were selected from amines reported in the literature which were expected to have high regeneration performance under conditions like this phase-separation process. These amines were blended with AMP and the basic characteristics of new blended amines were evaluated. As for process modifications, the injection of separated semi-lean liquid, which is usually injected into the top of the absorber, was relocated to the middle of absorber column. The rationale for this alteration was to maintain the CO2 absorption driving force in the upper part of the column. The alteration to the regeneration heat and CO2 recovery rate using new blended amine solutions and the new equipment set-up were empirically evaluated. AMP 50wt%+PZ 5wt%.  is the top of absorber column,  is the bottom of absorber column,  is the centrifuge,  is the top of stripper column and  is the bottom of stripper column, L is semi-lean liquid, and S is carbonate. Two lines are boundary of precipitation in AMP 50wt% and AMP 40wt%.

Selection of promoters based on reaction rate testing
Amines possible to react as promoter were gleaned from the literature (Bernhardsen et al, 2017a(Bernhardsen et al, , 2017b and blended with AMP to test their performance impact. Specifically, 1,6-Diaminohexane (HMDA), 1,4-Diaminobutane (DAB), N-(2-Aminoethyl) piperazine (AEP), Benzylamine (BZA), and N-Methyl-1,3diaminopropane (MAPA) were evaluated. Figure 2 provides a schematic of the reaction rate test apparatus . A mixture of gaseous CO2 and N2 is supplied into the heated reactor, and the reaction rate is calculated by measuring the CO2 concentration of the outlet gas, yielding the CO2 absorbed per unit area. Figure  3 shows results of the testing. Although the addition of HMDA, DAB, AEP, and BZA did not accelerate the reaction rate, MAPA significantly improved the reaction rate. It has been reported that MAPA forms an eightmember ring which is easily regenerated and improves the absorption rate (Zhang et al., 2018). Based on the above results, MAPA was selected as a promoter.

Vapor-liquid equilibrium test
The vapor-liquid equilibrium characteristics of AMP 50%/MAPA 5% solution were tested in the temperature range of the absorber and stripper columns. Figure 4 shows the results of test and the results of Dash et al. (2011) using AMP 43wt% at 55°C. The CO2 loading of solution was measured by the total organic carbon analyzer. From the fact that the results of Dash et al. (2011) fall between the 50°C and 60°C VLE plots of this test, it was concluded that the VLE behaviour of AMP was not significantly influenced by the addition of MAPA.

Measurement of precipitation and dissolution temperature
Precipitation and dissolution temperature of carbonate were measured by heating and cooling solution in a glass container to determine the precipitation boundary of AMP/MAPA solution selected as described in section 2.1. CO2 loading of solution was 0.15-0.45 (mol-CO2/molamine). Figure 5 shows the measurement result and the precipitation boundary against the temperature-dependent loading. Precipitation can be avoided by maintaining a temperature >60°C under a CO2 partial pressure of 15 kPa. Figure 5. Boundary of precipitation against temperature and CO2 loading. Blue line is VLE curve of CO2 partial pressure 15kPa.

Cooling precipitation test
The cooling temperature required to induce carbonate precipitation was obtained by batch testing. The theoretical formula of Kubota (Kobari, 2014) is shown in Eq. (1), where ΔTind is the degree of supersaturation, (N/M)m is the assumed number density of primary nuclei for detection sensitivity kb1 is the primary nucleation constant, n is nucleation order, and ΔT is the degree of supercooling.
Logarithmic plots of ΔT and lag time to precipitation are linear as shown in Eq. (1), and kb1 can be determined by the slope and intercept (isothermal method). After quenching 5 mL AMP 50 wt% solution, the temperature was held constant and the lag time to carbonate precipitation was measured. The primary nucleation constant was calculated from the measured time, the supersaturation degree, and the cooling temperature in the CO2 capture and recovery test. The assumed number density of primary nuclei for detection sensitivity was set to (N/M)m = 500. Figure 6 shows the results of testing, which indicates the waiting time for precipitation against the degree of supersaturation become shorter as the CO2 loading increase. Calculating kb1 and n in Eq. (1) by the plots and substituting waiting time of solution in cooling container into Eq. (1) gives the required degree of supersaturation for precipitation. The CO2 loading in the bottom of absorber was assumed to be 0.45, and the waiting time of solution in cooling container was 2010 seconds, so the degree of supersaturation was determined to be 6.4°C (T = 53.6°C).

Figure 6.
Waiting time for precipitation measured by isothermal method

Carbonate analysis
If MAPA is contained in the precipitates, then regeneration heat must be examined considering this MAPA-carbonate structure. Carbonate was analysed by Raman spectroscopy and gas chromatography and mass spectrometry (GC-MS) to confirm whether MAPA was contained in the precipitate. Figure. 7 shows the Raman spectra of the precipitates from AMP 50 wt%, AMP 50 wt%+MAPA 5 wt%, pure AMP, and pure MAPA.
Although peaks commonly observed for hydrocarbons and AMP were detected in the spectrum for AMP+MAPA, no peaks specific to MAPA were not detected (see Figure  7). Figure 8 shows the analysis results of AMP 50 wt%+MAPA 5 wt%, AMP 100%, and MAPA 100% by GC-MS. Spectra of pure AMP and AMP 50 wt%+MAPA 5 wt% were observed at an elapsed period of 6.6 minutes, the peak of pure MAPA was observed at a period of 7.4

Mid-column Injection Process
In our previous study of the phase separation process, separated CO2 semi-lean liquid was mixed with CO2 lean liquid and sent to the top of the absorber column. However, returning partially loaded solution (i.e., semi-lean liquid) to the top decreases the local absorption flux at the upper portion where is low CO2 partial pressure. By injecting semi-lean liquid into the middle of the absorber column, the local CO2 absorption driving force is maintained. This modification can improve the CO2 absorption rate. Figure 9 shows the partial model of the absorber built on the process simulator Aspen Plus ® . The absorber column consisted of 10 stages. The change of CO2 recovery rate and precipitate quantity was simulated when the injection stage was changed stepwise from Stage 1 (top) to Stage 10 (bottom). Figure 10 shows the calculation results of AMP 50wt% solution. The CO2 recovery rate and the precipitate quantity peaked when semi-lean liquid was injected at Stage 5, equivalent to the central portion of the absorber column. From this test results, the injection position of semi-lean liquid was set to the middle of the absorber column.

Experimental apparatus
The new mixed amine and new injection location were evaluated by continuous CO2 capture and recovery tests using the apparatus shown in Figure 11. The main points in the process are highlighted by numbers (1-5) and letters (S, L) in circles; these points correspond to Figure  1. The new injection location is designated by ' in Figure 11. Precipitation amount g/s CO 2 recovery rate %

Injection stage
The temperature in absorber must be maintained above the precipitation temperature to prevent passage blockage. To avoid precipitation prior to reaching the cooling container, the temperature of the lower of absorber and transfer lines to the cooling container were maintained at 65.0°C. The cooling container was chilled at the precipitation temperature of around 53.0˚C by 20.0°C brine with natural cooling without insulation.
The amine concentration in solution was reduced to 40 wt% in the CO2 capture and recovery test because the amount of precipitate increases by lowering the amine concentration under the condition of a constant liquid gas ratio. Table 1 shows this test conditions.

Operation stability
Operation stability of CO2 capture and recovery test including continuous precipitation and centrifugal separation was checked by material balance of CO2. Material balance of CO2 can be evaluated by two way which are the CO2 captured in the absorber, and CO2 gas flow rate from the outlet of stripper, shown in Table 2. The difference between these flow rates was about 1.0%, and it was confirmed that the experiment was operated stably.

QH = WS Cp (Tout -Tin) / WCO2
(4) Figure 12 shows the breakdown of regeneration heat of AMP 40 wt%+MAPA 5 wt% in comparison with the previous results of AMP 50 wt%+PZ 5 wt%. The sensible heat of this test was 0.48 GJ/ton-CO2, which corresponded to a decrease of 11-39% against the result of AMP 50 wt%+PZ 5 wt% under the same condition.
The CO2 recovery rate was obtained by the supplied CO2 gas flow rate and CO2 captured in the absorber. The supplied CO2 gas flow rate was 2.2 L/min, which corresponds to 240 g/h. Since CO2 captured in the absorber was 217 g/h, the CO2 recovery rate was 90%.

Conclusion
This study aimed to improve the CO2 recovery rate via selecting new promoter and modification of absorber configuration and cooling condition in the phase separation process using high-concentration AMP. The reaction rate test showed higher CO2 absorption rate by adding MAPA than that by adding PZ, as such, AMP/MAPA solution was selected as a new solution.
Simulation of the absorber column indicated that the highest CO2 recovery rate and precipitate quantity were obtained by sending semi-lean liquid to central portion of the absorber column. In CO2 capture and recovery tests with the new solution and the mid-column injection of semi-lean liquid, CO2 recovery rate reached to 90% and the sensible heat was reduced to 0.48 GJ/ton-CO2.

References
Bernhardsen, I. and H. Knuutila; "A Review of Potential Amine Solvents for CO2 Absorption Process: Absorption