Enhancement of CO₂ Absorption Process Using High-Frequency Ultrasonic Waves

Abstract

The advancement of efficient carbon capture technology is vital for the transition to a net-zero carbon future. Critical developments in ultrasonic irradiation can be used to enhance the conventional CO₂ absorption process. For example, sonophysical effects such as acoustic streaming, acoustic cavitation, acoustic fountain and atomization induced by the propagation of high-frequency ultrasonic waves in a liquid medium can enhance the mixing and create a larger interfacial area for gas–liquid mass transfer. In this study, the performance of a continuous ultrasonic-assisted CO₂ absorption process using MDEA was investigated. The design of experiment (DOE) was used to study the effect of the gas flowrate, liquid flowrate and ultrasonic power on CO₂ absorption performance. Based on the findings, ultrasonic power was the most significant parameter affecting the CO₂ outlet concentration, liquid-to-gas ratio (L/G) and mass transfer coefficient (K_Ga), which confirmed that ultrasonic irradiation has a significant impact on the intensification of the CO₂ absorption process. The optimum condition to achieve the target CO₂ absorption performance was numerically determined and validated with experimental tests. The results from the verification runs were in good agreement with the predicted values, and the average error was less than 10%.

1. Introduction

Recently, the term sustainability has been broadly used in programs, initiatives and actions to preserve particular resource areas such as human, social, economic and environmental areas. These are also known as the four pillars of sustainability. Environmental sustainability, as described by the United Nations (UN), is meeting the needs of the present without compromising the ability of future generations to meet their own needs [1]. Environmental sustainability is essential to preserve natural resources such as clean water, air and wildlife for future generations. Based on this concept, industrial development shall result in positive economic outcomes without harming the environment for short- or long-term applications.

Nevertheless, the greenhouse gases (GHGs) released into the atmosphere during industrialization and urbanization has been increasing yearly and contributing to global warming issues and climate change. Other environmental issues due to excess GHGs in the atmosphere include increasing seawater levels, the melting of ice sheets and glaciers, oceanic storms, species extinction and the disturbance of ecosystems [2]. Total global GHGs emissions between 2010 and 2019 have averaged around 54.4 gigatons of CO₂ equivalent, with about 14 gigatons (26%) coming from industrial applications [3]. Carbon capture, utilization and storage (CCUS) are one of the potential solutions that can overcome this climate change issue [4,5]. It is a promising pathway to decarbonize fossil-based power and industrial sectors to sustainably transition to a net-zero emission energy future [2]. An analysis conducted by the Intergovernmental Panel on Climate Change (IPCC) and the International Energy Agency (IEA) has shown that it is impossible to achieve “Net Zero” by 2050 without CCUS [4]. Currently, CCUS technologies are costly and energy intensive, which hinders their commercialization despite decades of developments. Therefore, intentions to improve the overall economics of CCUS technology have become the main focal point. However, most of the CO₂ conversion and utilization technologies are still in the developing stage [5]. The overall cost for CCUS technology is expected to be $22–80 per tone of CO₂ removed by 2025 [2,5].

Carbon capture from industrial applications is one of the critical ways to reduce CO₂ emission to the atmosphere and is one of the requirements to meet the Paris Agreement, which is to limit global warming to well below 1.5 °C and to pursue efforts to limit it to 2.0 °C by 2030 [6]. Carbon capture is classified into precombustion capture, postcombustion capture and oxyfuel combustion [2,4,5]. In each of the CO₂ capture systems, the type of CO₂ separation technologies varies depending on the partial pressure of CO₂, the composition of the gas to be treated and the location of the process [2]. The captured CO₂ can be further processed into other valuable materials, which is better known as carbon capture and utilization (CCU). Subsequently, if the captured CO₂ cannot be utilized directly, it will be transported and permanently stored in the subsurface, which is also known as carbon capture and storage (CCS). In the oil and gas industry, natural gas produced from the field sometimes contains a high concentration of CO₂. Therefore, it is essential to handle the excess CO₂ without damaging the natural environment. Natural gas containing high concentrations of CO₂ needs to be treated through the gas sweetening process to meet the pipeline quality, meet the calorific values of the sales gas and avoid crystallization during the liquefaction process [7,8]. The gas sweetening process ensures the treated natural gas can be transported through the pipelines and further processed into valuable products through the downstream process. The captured CO₂ is usually injected back into the field or any depleted field nearby for offshore operation [9].

Currently, natural gas sweetening processes are carried out by traditional amine absorption systems followed by thermal regeneration by using high-pressure packed towers and bubble columns [8,10,11,12,13]. The chemical absorption technology exhibits a high separation efficiency (>90%) with minimal hydrocarbon loss. It can remove the CO₂ concentration to a deficient trace level. It also exhibits a high separation efficiency in removing CO₂ at relatively low pressure conditions [2].

The efficiency of the absorption process is mainly driven by the chemical reaction between the CO₂ and solvents. Several types of solvents have been reported in the literature, but amine-based solvents such as monoethanolamine (MEA), diethanolamine (DEA), N-methyldiethanolamine (MDEA), diglycolamine (DGA), di-2-propanolamine (DIPA) and 2-amino-2-methyl-1-propanol (AMP) are typically used in the industry for CO₂ removal due to its high reaction rate [14,15]. In this process, the CO₂ reacts with the amine solvent through an exothermic reversible reaction to form a soluble carbonate salt. The CO₂ can be released when the amine solution containing the carbonate salt is heated during the regeneration process in the stripping column. The ability of the amine-based solvent to remove CO₂ is highly dependent on the solubility, the reaction rate and the mass transfer properties [16]. The composition of these amine-based solvents is continuously changed in industrial applications to optimize the CO₂ removal efficiency.

Although chemical absorption technology using packed towers and bubble columns as contactors have been used in industries for decades, several disadvantages are observed. For example, high equipment corrosion rates, flooding at high flowrates, unloading at low flowrates, channeling, foaming and amine degradation issues lead to difficulties in mass transfer between gas and liquid [14,17,18,19]. The process is also energy intensive to regenerate a large amount of solvent [20]. Additionally, it is known for its bulky size and excessive equipment tonnage, making the technology unfavorable for offshore operations [12,21,22].

Ultrasonic irradiation is one of the methods that can be used to enhance the gas–liquid mass transfer and further intensify the CO₂ absorption process. The propagation of ultrasound waves in a liquid medium will generate sonophysical and sonochemistry effects such as acoustic streaming, acoustic cavitation, acoustic fountain and atomization that can enhance mixing, promote the formation of radicals and increase the surface area for mass transfer to occur, which will lead to a higher removal efficiency. By having a higher removal efficiency, the overall size and weight of an absorber can be further reduced. Additionally, some of the ultrasound waves will be converted to heat energy that will give a heating effect to the liquid medium. The heating effect will help to reduce the energy needed to preheat the solvent for the absorption process. Figure 1 shows the fluid motion induced by high-frequency ultrasonic irradiation in a liquid medium.

Experimental works on batch systems have been executed to study the effect of operating parameters on the enhancement of the ultrasonic-assisted gas–liquid mass transfer process, such as the effect of ultrasonic frequency, power, sonicated liquid height and volume and distance of ultrasonic transducer [24,25,26,27,28,29,30,31,32,33]. The experimental study on batch systems was conducted to investigate the fundamental theories of high-frequency ultrasonic waves on the absorption process. Generally, from the batch system experimental study, the gas–liquid mass transfer coefficient was found to increase with the increase in ultrasonic power but decrease with the increase in sonicated liquid height, volume and distance from the ultrasonic transducer [23]. However, industrial applications are operated in continuous systems for more effective and cost-saving operations than batch systems. Compared to the batch process, the hydrodynamic of the continuous process is highly dependent on the gas and liquid flowrate. The residence time in the continuous process varies according to the gas and liquid flowrate, affecting the absorption performance [34]. Therefore, it is essential to investigate the ultrasonic effect on the CO₂ absorption process in the continuous system to provide helpful insight for further improvement.

In this study, the performance of a continuous ultrasonic-assisted absorption process for the removal of CO₂ from a binary N₂/CO₂ feed gas mixture using MDEA as the chemical solvent was explored. The effect of the operating parameters such as the gas flowrate, liquid flowrate and ultrasonic power on the CO₂ removal performance was investigated. Finally, the optimum conditions for the ultrasonic-assisted absorption process were determined by using the design of experiment (DOE) approach.

2. Materials and Methods

2.1. Material

Among the amine-based solvents, MDEA is widely used as the solvent to treat the natural gas sweetening process due to its higher absorption capacity, lower regeneration energy, lower degradation rate and less corrosiveness [35,36,37]. In tertiary amines such as MDEA, the reaction with CO₂ occurs by the formation of a bicarbonate ion when the CO₂ dissolves in water and by the protonation of the amine via the base-catalyzed hydration of CO₂ due to the lack of the necessary N-H bond [15,38]. The hydration reaction is slower than the carbamate formation from the direct reaction between the primary and secondary amine with CO₂ through the Zwitterion mechanism [15,36,38]. Despite its slower reactivity with CO₂ as compared to primary and secondary amines, MDEA is preferred due to its higher loading capacity, which is 1 mole of CO₂ per mole of amine [12]. The following chemical reactions occur in an aqueous MDEA solution when CO₂ is present [16,35]:

$${\mathrm{CO}}_2+{\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{HCO}}_3^{-}+{\mathrm{H}}^{+}$$

(1)

$${\mathrm{HCO}}_3^{-}\leftrightarrow {\mathrm{CO}}_3^{2-}+{\mathrm{H}}^{+}$$

(2)

$${\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{OH}}^{-}+{\mathrm{H}}^{+}$$

(3)

$$R{R}^{\prime }{R}^{\prime }\mathrm{NH}\leftrightarrow R{R}^{\prime }{R}^{\prime }\mathrm{N}+{\mathrm{H}}^{+}$$

(4)

where R corresponds to a methyl group and R′ to an ethanol group.

A tertiary amine, methyldietanolamine (MDEA) (supplied by Revlogi Materials), was used in this study. The pure MDEA was diluted with distilled water to achieve a 41–42 wt% concentration. The fresh MDEA solution was then loaded with pure CO₂ gas (supplied by Air Products) through the absorption process at 70 barg to achieve a semilean loading of 0.2 mol/mol. The amine concentration and loading were determined using the titration method with 0.5 N hydrochloric acid (HCL) and 0.5 N potassium hydroxide (KOH).

In this study, due to safety reasons, the flammable methane gas, which is the main component in natural gas, was substituted with N₂ gas [39]. Additionally, N₂ is an inert gas with a solubility lower than CO₂; thus, it will not affect the absorption performance [40,41]. For the feed gas, pure CO₂ gas (supplied by Air Products) was mixed with pure N₂ gas (supplied by Air Products) according to the volumetric flowrate by using the mass flow controller to meet the target composition of 25 vol% CO₂. This mixture was used to mimic the composition of natural gas. The feed gas setup was constructed according to the method previously described by Chan et al. [42].

2.2. Titration Method

The amine concentration and loading were determined by using the titration method. For amine concentration determination, the amine samples were titrated with 0.5 N hydrochloric acid (HCl) [43]. A total of 5 g of amine samples were mixed with 95 mL of distilled water and stirred at 400 rpm using a magnetic stirrer. The pH of the solution was measured by using a pH meter (HANNA HI8424 Portable pH meter). The solution was then titrated with 0.5 N HCl until pH 4.5 was achieved. The weight of the amine sample, the pH of the solution before and after titration and the volume of the 0.5 N HCl solution used were recorded. The concentration of the amine solution in wt% was calculated by using the following formula:

$$\left(\frac{{\mathrm{V}}_{\mathrm{HCl}}\times {\mathrm{N}}_{\mathrm{HCl}}\times 11.9163}{{\mathrm{W}}_{\mathrm{s}}}\right)=\mathrm{wt}\% \mathrm{amine}$$

(5)

where ${\mathrm{V}}_{\mathrm{HCl}} \left(\mathrm{mL}\right)$ is the volume of HCl used to titrate the sample in mL, ${\mathrm{N}}_{\mathrm{HCl}}$ is the normality of the HCl solution and ${\mathrm{W}}_{\mathrm{s}}\left(\mathrm{g}\right)$ is the weight of the sample.

For amine loading determination, the amine samples were titrated with 0.5 N potassium hydroxide (KOH) [43]. A total of 20 g of the amine sample was weighed. Then, 125 mL of methanol, in which the pH was adjusted to 11.2, was prepared by adding a few drops of 0.5 N KOH solution into the methanol solution. The amine sample and pH-adjusted methanol solution were mixed and stirred at 400 rpm by using the magnetic stirrer. The pH of the mixture was measured by using the pH meter (HANNA HI8424 Portable pH meter). The mixture was then titrated with 0.5 N KOH until the pH of the mixture returned to 11.2. The weight of the amine sample, the pH of the mixture before and after titration, and the volume of the 0.5 N KOH used were recorded. The concentration (wt%) of acid gas in the amine solution was determined by using the following formula:

$$\left(\frac{{\mathrm{V}}_{\mathrm{KOH}}}{{\mathrm{W}}_{\mathrm{s}}}-\frac{{\mathrm{V}}_{\mathrm{KOH},\mathrm{f}}}{{\mathrm{W}}_{\mathrm{s},\mathrm{f}}}\right)\times {\mathrm{N}}_{\mathrm{KOH}}\times 4.4= \mathrm{wt}\% \mathrm{acid} \mathrm{gas}$$

(6)

where ${\mathrm{V}}_{\mathrm{KOH}} \left(\mathrm{mL}\right)$ is the volume of KOH used to titrate the sample, ${\mathrm{V}}_{\mathrm{KOH},\mathrm{f}} \left(\mathrm{mL}\right)$ is the volume of KOH used to titrate the fresh solvent, ${\mathrm{W}}_{\mathrm{s},\mathrm{f}}\left(\mathrm{g}\right)$ is the weight of the fresh solvent and ${\mathrm{N}}_{\mathrm{KOH}}$ is the normality of the KOH solution.

The acid gas loading (mol CO₂/mol amine) was calculated based on the amine and acid gas concentration by using Equations (5) and (6):

$$\left(\frac{\mathrm{wt}\% \mathrm{acid} \mathrm{gas} \times 2.708}{\mathrm{wt}\% \mathrm{amine}}\right)=\mathrm{Acid} \mathrm{gas} \mathrm{loading}$$

(7)

2.3. Lab Scale Setup

An integrated ultrasonic-assisted absorption with a solvent regeneration system pilot test rig was developed to study the effect of operating conditions on the continuous ultrasonic-assisted absorption process for the removal of CO₂. The schematic diagram and the actual pilot test rig setup of the integrated system are shown in Figure 2. To the best of our knowledge, no literature to date has reported the development, installation and operation of a pilot-scale ultrasonic-assisted CO₂ absorption process using chemical solvents.

The feed gas of 25 vol% CO₂ and 75 vol% N₂ was premixed into the compressor suction vessel at 20 barg before being compressed to 70 barg into the ultrasonic absorber with a total volume of 2.4 L. The lean amine solvent from the amine tank was preheated to 60 °C by using thermal oil in the shell-tube heat exchanger before being pumped into the ultrasonic absorber by using the high-pressure (HP) amine pump. In the ultrasonic absorber, 32 ultrasonic transducers were used to generate the 1.7 MHz high-frequency ultrasonic waves to enhance the CO₂ absorption process. The CO₂ composition in the treated gas was analyzed by using the IR analyzer. The rich amine solvent discharged from the ultrasonic absorber was sent to the HP flash vessel (operated at 9 barg) before being heated to 90 °C by using the thermal oil in the shell-tube heat exchanger to regenerate the solvent. The released CO₂ gas was flashed out in the low pressure (LP) flash vessel operated at 1 barg and vented to a safe location. The regenerated amine solvent was cooled down by using cooling water in the shell-tube heat exchanger before being recycled into the amine tank. The amine concentration and acid gas loading in the amine solvent were analyzed by using the titration method.

2.4. Design of Experiment (DOE)

The effect of three operating parameters such as the gas flowrate, liquid flowrate and ultrasonic power on the CO₂ absorption performance was investigated. The test matrix was developed by using a central composite design (CCD) coupled with response surface methodology (RSM) in the Design Expert v13.0 software.

Table 1. Operating condition parameters.

Parameters	Unit	Symbol	Level
			(−1)	(0)	(+1)
Gas Flowrate (FG)	SLPM	A	20	30	40
Liquid Flowrate (FL)	LPM	B	0.2	0.35	0.5
Ultrasonic Power (PUS)	W	C	0	200	400

shows the parameters and levels in this study. The selected range is based on the capability of the system installed for the testing. A total of 21 runs were generated by using the CCD-RSM method. Three responses were analyzed and optimized, which were the CO₂ outlet concentration (mol%); liquid-to-gas ratio (L/G), which is the ratio of the solvent circulation rate whereby the amount of acid gas removed is usually expressed in the unit gallons amine per lbmol of acid gas removed (gal/lbmol CO₂); and the overall mass transfer coefficient (K_Ga) (mol/m³.s.Pa). Based on the experimental results, the optimum conditions for the absorption process were determined.

2.5. Ultrasonic Power Generation

The piezoelectric transducer generates the ultrasound waves. Piezoelectricity is a phenomenon where an electric charge is generated in response to applied mechanical stress on a particular material. Conversely, mechanical displacement can be produced when the electrical field is applied to the material [44,45,46]. In this study, the electric current was applied to the 1.7 MHz piezoelectric transducers made from lead, zirconate, and titanate (PZT) ceramic material. The electric field will distort the crystal and force it to vibrate to produce ultrasonic waves.

The correlation between the total ultrasonic power for 32 units of transducers and the power system voltage is shown in Figure 3. In this work, the testing was conducted at an ultrasonic power of 0 W, 200 W and 400 W. Therefore 0 V, 27 V and 40 V of electricity were supplied by the power system to the ultrasonic transducers to generate the desired ultrasonic power.

2.6. Liquid-to-Gas Ratio (L/G) Determination

The L/G indicates the efficiency of or difficulties with removing the pollutant. In this case, it is the CO₂ being removed from natural gas. A higher L/G indicates a less efficient removal rate; therefore, achieving a low L/G in an absorption process is more favorable as it indicates a higher removal efficiency and lower liquid pumping rates. The L/G was determined by:

$$\frac{\mathrm{L}}{\mathrm{G}}=\frac{{\mathrm{F}}_{\mathrm{L}}}{{\mathrm{C}}_{{\mathrm{CO}}_2}}$$

(8)

where ${\mathrm{F}}_{\mathrm{L}}$ is the liquid flowrate in gal/min and ${\mathrm{C}}_{{\mathrm{CO}}_2}$ is the amount of CO₂ absorbed in lbmol/min. ${\mathrm{C}}_{{\mathrm{CO}}_2}$ can be determined from the equation

$${\mathrm{C}}_{{\mathrm{CO}}_2}=\eta \times {\mathrm{F}}_{\mathrm{G}}\times {\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}\times 453.6$$

(9)

where $\eta$ is the removal efficiency, ${\mathrm{F}}_{\mathrm{G}}$ is the inlet gas flowrate in standard liters per minute (SLPM) and ${\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}$ is the inlet CO₂ mol fraction. The removal efficiency was calculated based on the CO₂ inlet and outlet concentration as per the equation below [47]:

$$\mathsf{\eta }=\left(\frac{{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}-{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{out}}}{{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}}\right)\times 100$$

(10)

where ${\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{out}}$ is the outlet CO₂ mol fraction.

2.7. Mass Transfer Coefficient Determination

The enhancement of the CO₂ absorption performance is commonly described as the overall mass transfer coefficient (K_Ga) derived from the thermodynamics, kinetics and hydrodynamics of the CO₂ absorption system [48]. For packed columns, the K_Ga can be determined by using the following differential equation [48,49]:

$${\mathrm{K}}_{\mathrm{G}}\mathrm{a}=\frac{{\mathrm{G}}_{\mathrm{I}}}{\mathrm{P}\left({\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{G}}-{\mathrm{y}}_{{\mathrm{CO}}_2}^{*}\right)}\left(\frac{{\mathrm{dY}}_{{\mathrm{CO}}_2,\mathrm{G}}}{\mathrm{dZ}}\right)$$

(11)

where ${\mathrm{G}}_{\mathrm{I}}$ is the gas velocity in mol/m².h; $\mathrm{P}$ is the operating pressure in Pa; ${\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{G}}$ and ${\mathrm{y}}_{{\mathrm{CO}}_2}^{*}$ are the mole fraction of CO₂ in the gas stream and equilibrium mol fraction of CO₂, respectively; ${\mathrm{Y}}_{{\mathrm{CO}}_2,\mathrm{G}}$ is the mole ratio of CO₂ in the gas stream; and $\mathrm{Z}$ is the column height.

In this study, the above differential equation was rewritten in terms of the CO₂ concentration entering and leaving the ultrasonic absorber. The mass transfer driving force was replaced by the log mean concentration driving force, as suggested by M. Yusof et al. [34]. The liquid phase was assumed to be well mixed from the acoustic streaming effect and a large amount of solvent in the ultrasonic-assisted absorber. The gas phase composition changes continuously as the amine solvent absorbs the CO₂; thus, a log mean concentration driving force should be used as follows [34]:

$${\mathrm{K}}_{\mathrm{G}}\mathrm{a}=\frac{{\mathrm{F}}_{\mathrm{V}}\left({\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}-{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{out}}\right)}{\Delta {\mathrm{P}}_{\mathrm{LM}}}$$

(12)

$$\Delta {\mathrm{P}}_{\mathrm{LM}}=\mathrm{P}\left[\frac{\left({\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}-{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{out}}\right)}{\ln \left(\frac{{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{in}}}{{\mathrm{y}}_{{\mathrm{CO}}_2,\mathrm{out}}}\right)}\right]$$

(13)

where ${\mathrm{F}}_{\mathrm{V}}$ is the gas flowrate per unit volume (mol/m³.s) and $\Delta {\mathrm{P}}_{\mathrm{LM}}$ is the log mean pressure.

3. Results and Discussion

Table 2. Test matrix and results for CO₂ outlet concentration, L/G and mass transfer coefficient.

Run	Gas Flowrate, FG (SLPM)	Liquid Flowrate, FL (LPM)	Ultrasonic Power, PUS (W)	Response 1: CO2 Outlet (mol%)	Response 2: L/G (gal/lbmol)	Response 3: KGa (mol/m3.s.Pa)
1	40	0.2	0	23.4	583.6	1.64 × 10−7
2	30	0.35	0	18.1	423.2	4.51 × 10−7
3	20	0.5	0	18.5	919.3	2.92 × 10−7
4	30	0.35	200	16.2	340.4	5.92 × 10−7
5	20	0.2	400	5.9	134.2	1.30 × 10−6
6	30	0.35	400	10	200.8	1.24 × 10−6
7	40	0.5	400	11.6	234.7	1.41 × 10−6
8	30	0.2	200	17.2	210.5	5.28 × 10−7
9	30	0.35	200	16	336.8	6.03 × 10−7
10	20	0.35	200	12.5	350.6	6.39 × 10−7
11	30	0.5	200	15.5	451.9	6.50 × 10−7
12	40	0.2	400	14.9	122.0	9.70 × 10−7
13	30	0.35	200	16.4	329.7	6.01 × 10−7
14	40	0.5	0	21.3	745.7	3.36 × 10−7
15	40	0.35	200	18.4	313.2	6.00 × 10−7
16	20	0.5	400	2.7	293.1	1.97 × 10−6
17	20	0.2	0	18.4	372.8	2.90 × 10−7
18	30	0.35	200	16.4	333.2	5.96 × 10−7
19	25	0.35	400	8.1	216.0	1.26 × 10−6
20	25	0.2	400	9.7	132.6	1.08 × 10−6
21	20	0.35	400	5	230.3	1.43 × 10−6

shows the test matrix for the 21 experimental runs generated by using the DOE. The results of the three responses included the CO₂ outlet concentration, L/G and mass transfer coefficient (K_Ga). In this study, the target was to reduce the CO₂ inlet concentration from 25% to less than 6.5% (to meet the sales gas application), minimize the L/G and maximize the mass transfer coefficient K_Ga.

3.1. Effect of Operating Condition on CO₂ Outlet Concentration

From Table 2, it is observed that the target CO₂ outlet concentration (which is less than 6.5%) could only be achieved in Run 5, 16 and 21. These findings show that the operating parameters significantly affect the absorption performance, which is the CO₂ outlet concentration from the ultrasonic-assisted absorber. The results of the analysis of variance (ANOVA) for the CO₂ outlet response that was obtained during the experiment is shown in

Table 3. ANOVA for CO₂ outlet concentration.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value Prob > F
Model—Reduced Quadratic	521.09	6	86.85	695.29	<0.0001	Significant
A—Gas Flowrate	75.30	1	75.30	602.84	<0.0001
B—Liq Flowrate	8.22	1	8.22	65.79	<0.0001
C—Ultrasonic Power	290.31	1	290.31	2324.17	<0.0001
AC	12.61	1	12.61	100.92	<0.0001
BC	3.99	1	3.99	31.95	<0.0001
A2	7.32	1	7.32	58.57	<0.0001
Residual	1.62	13	0.1249
Lack of fit	1.19	9	0.1326	1.23	0.4510	Not Significant
R2	0.9969
Adjusted R2	0.9955
Predicted R2	0.9905

. The ANOVA showed the model’s fitness, the significance of each parameter and their interactions. For a 95% confidence level, the model, parameters and their interactions are considered significant if the p-value is less than 0.05.

Based on Table 3, the reduced quadratic model significantly predicted the CO₂ outlet response, with a predicted R² value of 0.9905. The lack of fit was insignificant since its p-value was more than 0.05. A, B, C, AC, BC and A² were the most significant terms in the model. The other terms that were not significant, which had a p-value of more than 0.05, were removed from the model to improve the prediction accuracy. The reduced quadratic model in terms of the actual factors suggested by the DOE software (Design Expert v13.0) for the prediction of the CO₂ outlet response is as follows:

$${\mathrm{CO}}_2 \mathrm{Outlet} \left(\mathrm{mol}\%\right)=15.93+2.95{\mathrm{F}}_{\mathrm{G}}-0.9414{\mathrm{F}}_{\mathrm{L}}-5.62\mathrm{C}+1.40{\mathrm{AP}}_{\mathrm{US}}-0.7277\left({\mathrm{F}}_{\mathrm{L}}{\ast \mathrm{P}}_{\mathrm{US}}\right)-1.31{\mathrm{F}}_{\mathrm{G}}^2$$

(14)

The F-value in Table 3 shows that the ultrasonic power was the most significant parameter since it had the highest F-value, followed by the gas flowrate and the gas flowrate and ultrasonic power interactions. Even though the liquid flowrate was less significant as compared to the other parameters, it still had a significant impact on the CO₂ outlet concentration as the p-value for the factor was less than 0.05.

Figure 4 shows the contour of the CO₂ outlet response for the two most significant factors, which were the ultrasonic power and gas flowrate at the highest liquid flowrate. It was observed that the CO₂ outlet was reduced by 83% as the ultrasonic power increased from 0 W to 400 W due to the higher sonophysical enhancement such as the acoustic streaming and acoustic fountain that led to better mixing, further reduced the liquid film resistance and produced more interfacial area for mass transfer to occur [34,50,51]. The increase in the gas flowrate from 20 SLPM to 40 SLPM reduced the absorption efficiency and increased the CO₂ outlet concentration by four times compared with the sale gas specification (6.5%). This is because the reaction rate of tertiary amines such as MDEA is slower than the primary and secondary alkanolamines; thus, the increase in the gas flowrate will reduce the residence time for the absorption process.

Figure 5 shows the interaction between the gas flowrate and liquid flowrate for the CO₂ outlet response at 400 W ultrasonic power. The target CO₂ outlet concentration of less than 6.5 mol% could only be achieved at the lowest gas flowrate (20 SLPM). The increase in the liquid flowrate from 0.2–0.5 SLPM further improved the ultrasonic-assisted absorption process by 54% as a higher liquid flowrate induced higher turbulence and subsequently reduced the liquid film resistance. Therefore, a higher liquid flowrate was preferred for the absorption process enhancement [43].

3.2. Effect of Operating Condition on L/G

Based on Table 2, the L/G response was in the range of 122–919.3 gal/lbmol. The ANOVA table for the L/G response is shown in

Table 4. ANOVA for L/G.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value Prob > F
Model—Reduced 2FI	1.314 × 105	5	26,272.56	833.68	<0.0001	Significant
A—Gas Flowrate	2798.65	1	2798.65	88.81	<0.0001
B—Liq Flowrate	31,174.34	1	31,174.34	989.22	<0.0001
C—Ultrasonic Power	68,031.70	1	68,031.70	2158.78	<0.0001
AB	580.94	1	580.94	18.43	0.0016
BC	3810.16	1	3810.16	120.90	<0.0001
Residual	315.14	10	31.51
Lack of fit	257.14	7	36.73	1.90	0.3219	Not Significant
R2	0.9976
Adjusted R2	0.9964
Predicted R2	0.9904

. From Table 4, the reduced two-factors interaction (2FI) model was significant at predicting the L/G response with a predicted R² value of 0.9904. The lack of fit was found to be not significant since its p-value was more than 0.05. A, B, C, AB and BC were the significant terms in the model, with a p-value of less than 0.05. The reduced 2FI model in terms of the actual factors suggested by the DOE analysis for the prediction of the L/G response was as follows:

$$\frac{\mathrm{L}}{\mathrm{G}}\left(\frac{\mathrm{gal}}{\mathrm{lbmol}}\right)=325.22-20.02{\mathrm{F}}_{\mathrm{G}}+117.33{\mathrm{F}}_{\mathrm{L}}-125.42{\mathrm{P}}_{\mathrm{US}}-11.79\left({\mathrm{F}}_{\mathrm{G}}{\ast \mathrm{F}}_{\mathrm{L}}\right)-48.75{\mathrm{BP}}_{\mathrm{US}}$$

(15)

From the F-value in Table 4, it was found that the ultrasonic power was the most significant parameter that affected the L/G response, followed by the liquid flowrate and the interactions between the liquid flowrate and ultrasonic power. Even though the gas flowrate was less significant than the other parameters when it came to influencing the L/G response, it was still significant as the p-value was less than 0.05.

Figure 6 shows the contour of the L/G response for the two most significant factors, which were the ultrasonic power and liquid flowrate. A lower L/G was preferred for the ultrasonic-assisted absorption process because it contributed to lower solvent circulation rates, which led to lower solvent pumping and heater power consumption for amine regeneration. Based on the figure, the L/G was 55% lower as the ultrasonic power increased from 0 W to 400 W due to the sonophysical and sonochemistry effect from ultrasonic irradiation. This shows that absorption with ultrasonic irradiation had a higher removal efficiency than absorption without an ultrasound.

The sensitivity of the L/G towards the liquid flowrate and ultrasonic power is shown in Figure 7. From Figure 7, the lower L/G could be achieved at a lower liquid flowrate due to less solvent being pumped into the absorber. However, at a constant liquid flowrate, the increase in the ultrasonic power from 0 to 400 W significantly reduced the L/G due to more CO₂ absorption, indicating a higher removal efficiency under ultrasonic irradiation. The CO₂ removal enhancement by ultrasonic irradiation was more substantial at a higher liquid flowrate. At a 0.2 SLPM liquid flowrate, the increase in the ultrasonic power from 0 to 400 W led to a 55% L/G reduction, while at a 0.5 SLPM liquid flowrate, the L/G reduction was 59%. This shows that CO₂ absorption into the MDEA solvent with ultrasonic irradiation was better than the absorption process without it, as it allowed more CO₂ to be absorbed into the MDEA solvent even at a lower liquid flowrate.

Figure 8 shows the effect of the gas flowrate and ultrasonic power on the L/G response at 0.5 SLPM of the liquid flowrate. The increase in the gas flowrate generally introduced more CO₂ molecules into the absorber. However, due to the slow kinetic reaction between the CO₂ and MDEA solvent, the increase in the gas flowrate did not significantly improve the absorption performance. Sonophysical and sonochemistry enhancement by ultrasonic irradiation significantly impacted the L/G reduction; however, it did not significantly improve the chemical reaction rate as it was still limited by the gas holdup and residence time in the absorber.

3.3. Effect of Operating Conditions on Mass Transfer Coefficient, K_Ga

In Table 2, the response for the mass transfer coefficient, K_Ga, from the experimental runs was between 1.64 × 10⁻⁷–1.97 × 10⁻⁶ mol/m³.s.Pa. The ANOVA table for the K_Ga response is shown in

Table 5. ANOVA for mass transfer coefficient.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value Prob > F
Model—Reduced Quadratic	3.407 × 10−12	8	4.259 × 10−13	1868.64	<0.0001	Significant
A—Gas Flowrate	2.574 × 10−15	1	2.574 × 10−15	11.29	0.0121
B—Liq Flowrate	4.812 × 10−14	1	4.812 × 10−14	211.10	<0.0001
C—Ultrasonic Power	4.006 × 10−13	1	4.006 × 10−13	1757.60	<0.0001
AB	3.209 × 10−14	1	3.209 × 10−14	140.81	<0.0001
AC	4.135 × 10−14	1	4.135 × 10−14	181.41	<0.0001
BC	1.434 × 10−14	1	1.434 × 10−14	62.91	<0.0001
B2	1.909 × 10−14	1	1.909 × 10−14	83.76	<0.0001
C2	1.099 × 10−13	1	1.099 × 10−13	482.29	<0.0001
Residual	1.596 × 10−15	7	2.279 × 10−16
Lack of fit	6.883 × 10−16	3	2.294 × 10−16	1.01	0.4750	Not Significant
R2	0.9995
Adjusted R2	0.9990
Predicted R2	0.9977

. The reduced quadratic model was significantly predicted the K_Ga response with a predicted R² of 0.9977. The lack of fit was insignificant since the p-value was more than 0.05. A, B, C, AB, AC, BC, B² and C² were the significant terms in the model with a p-value less than 0.05. From the F-value in the ANOVA table, ultrasonic power and its quadratic term were the most significant parameters that affected the K_Ga response, followed by the liquid flowrate. The reduced quadratic model in terms of the actual factor suggested by the DOE software for the prediction of the K_Ga response is as follows:

$$\begin{array}{cc}\hfill {\mathrm{K}}_{\mathrm{G}}\mathrm{a}\left(\frac{\mathrm{mol}}{{\mathrm{m}}^3.\mathrm{s}.\mathrm{Pa}}\right)& =6.293\times {10}^{-07}-3.039\times {10}^{-08}{\mathrm{F}}_{\mathrm{G}}+2.05\times {10}^{-07}{\mathrm{F}}_{\mathrm{L}}+3.951\times {10}^{-07}{\mathrm{P}}_{\mathrm{US}}\hfill \\ \hfill & -1.046\times {10}^{-07}\left({\mathrm{F}}_{\mathrm{G}}{\ast \mathrm{F}}_{\mathrm{L}}\right)-1.475\times {10}^{-07}\left({\mathrm{F}}_{\mathrm{G}}{\ast \mathrm{P}}_{\mathrm{US}}\right)+1.199\times {10}^{-07}\left({\mathrm{F}}_{\mathrm{L}}{\ast \mathrm{P}}_{\mathrm{US}}\right)\hfill \\ \hfill & +1.232\times {10}^{-07}{\mathrm{F}}_{\mathrm{L}}{}^2+2.179\times {10}^{-07}{\mathrm{P}}_{\mathrm{US}}{}^2\hfill \end{array}$$

(16)

Figure 9 shows the contour of the K_Ga response for the ultrasonic power and gas flowrate interactions. A higher K_Ga value indicates a higher absorption efficiency per unit volume for the absorption process. From Figure 9, it was observed that the K_Ga value increased 2.25 times as the ultrasonic power increased from 0 to 400 W due to the sonophysical and sonochemistry effect that further intensified the CO₂ absorption process that has been explained previously. The increase in the gas flowrate did not seem to increase the mass transfer performance since a higher K_Ga was obtained at a lower gas flowrate. At this condition, it was presumed that the liquid side mass transfer resistance primarily controlled the gas–liquid mass transfer process. Therefore, the reduction in the mass transfer resistance on the gas phase by increasing the gas flowrate did not have a significant impact on the absorption performance [52]. The amount of CO₂ absorbed into the MDEA solvent was limited by molecular diffusion, regardless of the gas load at the feed gas stream.

This can be further explained by Figure 10. At a constant gas flowrate of 20 SLPM, the increase in the liquid flowrate will further enhance the CO₂ removal efficiency and increase the mass transfer coefficient K_Ga. This is because increasing the liquid flowrate will introduce more turbulence on the liquid side and reduce the mass transfer resistance on the liquid side. Furthermore, a higher liquid flowrate will increase the number of amine molecules in the absorber and improve the absorption capacity of the solvent [52]. A higher liquid flowrate also will increase the velocity of the solvent, which will help to keep the concentration gradient of the CO₂ in the solvent low and enhance the rate of diffusion and CO₂ removal rate. The presence of ultrasonic irradiation further enhances the CO₂ removal rate because the acoustic fountain and streaming phenomena will enhance the gas–liquid mixing and increase the interfacial area for mass transfer to occur. This effect is clearly shown as the K_Ga increased almost three times as the ultrasonic power increased from 0 to 400 W at a liquid flowrate of 0.5 LPM and gas flowrate of 20 SLPM.

3.4. Ultrasonic-Assisted CO₂ Absorption Process Optimization

The operating conditions under the current study showed a significant effect on the continuous ultrasonic-assisted CO₂ absorption performance. The optimum condition for the ultrasonic-assisted CO₂ absorption process was determined by using the response surface methodology (RSM) numerical optimization feature of the Design Expert v13.0 software. The optimization criteria and the desired goal for process optimization are shown in

Table 6. Optimization criteria for process optimization study.

Parameters	Goal	Lower Limit	Upper Limit
A: Gas Flowrate	In range	20 SLPM	40 SLPM
B: Liquid Flowrate	In range	0.2 LPM	0.5 LPM
C: Ultrasonic Power	In range	0 W	400 W
CO2 Outlet	Target: 6.5 mol%	2.7 mol%	21.3 mol%
L/G	Minimize	122 (gal/lbmol CO2)	452 (gal/lbmol CO2)
KGa	Maximize	2.49 × 10−7 (mol/m3.s.Pa)	1.97 × 10−6 (mol/m3.s.Pa)

. Based on the criteria, 70 solutions were proposed, and the solution with the highest desirability of 0.808 was chosen. The selected optimized condition was the gas flowrate of 20 SLPM, liquid flowrate of 0.2 LPM and ultrasonic power of 400 W. The predicted responses at this optimum condition are shown in Figure 11.

Verification runs were conducted to verify the proposed optimized condition by the software, and the results for the CO₂ outlet, L/G and K_Ga are shown in Figure 12, Figure 13 and Figure 14, respectively. The experimental values were compared with the predicted values by the software. For the CO₂ outlet response, the average error was 2.6%, and the standard deviation was 0.153. These findings showed a good agreement between the experimental data and the predicted values. Similar to the L/G response, good agreements between the actual experimental data and the predicted values were observed with an average error of 0.66% and a standard deviation of 1.12. For the K_Ga, the actual experimental data were slightly higher than the predicted value, as shown in Figure 14. The average error between the experimental data and the predicted value was 9.84%, with a standard deviation of 2.11 × 10⁻⁸. All three responses agreed well with the predicted values; therefore, the optimum condition suggested by the software was considered valid and reliable.

4. Conclusions

In this work, the effect of operating parameters such as the gas flowrate, liquid flowrate and ultrasonic power on the CO₂ outlet, L/G and K_Ga were investigated. Based on the ANOVA analysis, ultrasonic power was the most significant parameter affecting the overall ultrasonic-assisted CO₂ absorption process performance. This justifies that ultrasonic irradiation showed significant enhancement to the gas–liquid mass transfer as compared to the process without an ultrasound. A process optimization study was conducted to find the optimum operating parameters for the ultrasonic-assisted CO₂ absorption process using the DOE. The proposed optimum condition to achieve the target CO₂ outlet of 6.5% with minimum L/G and maximum K_Ga was at a 20 SLPM gas flowrate, 0.2 LPM liquid flowrate and 400 W ultrasonic power. The results from verification runs showed a good agreement with the predicted values; therefore, the optimum condition suggested by the DOE software is considered valid and reliable.

However, this optimization is limited to the MDEA solvent only. Other alkanolamine and acoustic solvents with different kinetic reaction rates with CO₂ might have dissimilar optimized conditions in the ultrasonic-assisted CO₂ absorption system. Since ultrasonic power was the most significant parameter affecting the CO₂ absorption performance, identifying the optimum ultrasonic power that can provide significant gas–liquid mass transfer enhancement and simultaneously be economical for the operation is crucial. As a way forward, further studies on ultrasonic power optimization and technoeconomic analysis for this intensification process is recommended for future improvement.