Simulation and Optimization of the Acid Gas Absorption Process by an Aqueous Diethanolamine Solution in a Natural Gas Sweetening Unit.

The presence of carbon dioxide in natural gases can lower the quality of natural gas and can cause CO2 freezing problems. Therefore, using reliable techniques for the reduction and elimination of carbon dioxide from natural gases is necessary. The aqueous diethanol amine (DEA) solution's ability to simultaneously absorb H2S and CO2 from sour natural gases makes it possible to use this solution in the natural gas sweetening process. The goal of this work was to determine the maximum amount of the removed CO2 by an aqueous DEA solution in one of the gas sweetening plants of the National Iranian South Oilfields Company (NISOC). For this purpose, based on the obtained designed experiment results using the L9 orthogonal array Taguchi method, the experiments were conducted and three levels of amine concentrations (25, 28, and 30 wt %), temperatures (40, 50, and 60 °C), and circulation rates of lean amine (220, 240, and 260 m3 h-1) were considered as the key operational parameters on CO2 removal. To evaluate the ability of the HYSYS simulation software and the Kent-Eisenberg thermodynamic model to predict CO2 absorption by an aqueous DEA solution in the gas sweetening process, the field data were compared with the results of the simulation. It was observed that the maximum removal of CO2 is achieved at a lean amine concentration of 30 wt %, a temperature of 40 °C, and a circulation rate of 260 m3 h-1. Also, the experimental results indicate that the effects of the selected process variables on CO2 absorption are not linear and the most effective parameter on carbon dioxide removal is the concentration of amine in an aqueous solution and the temperature of the lean amine has the least effect. Besides, the obtained simulation results are in the range of the unit design basis but have some deviations from field data. The findings of this study can help in better understanding of the selection of the effective variables in the natural gas sweetening process and obtaining their appropriate values to achieve the highest efficiency.

Introduction

Natural gases normally contain some impurities such as hydrogen sulfide, carbon dioxide, water vapor, heavy hydrocarbons, and mercaptans. It is desirable to remove both H2S and CO2 (known as acid gases) to prevent corrosion problems and increase the heating value of the gas.[1] Because of health hazards, sale contracts, CO2 freezing, and corrosion problems, removing any sulfur compound and acid gas from natural gas (called natural gas sweetening) is one of the most important steps in natural gas processing. Some natural gas sweetening methods such as adsorption,[2] chemical and physical absorption,[3] and membrane separation[4] have been proposed and their capabilities have been investigated. It has been observed that factors such as gas flow rate, temperature, pressure, acid gas selectivity required, and economics play important roles in choosing an appropriate technique for natural gas sweetening.[5−7] The most commonly used method in acid gas removal is the absorption–desorption process, and the most appropriate solvents are aqueous alkanolamine solutions. The low operating cost, reactivity, and flexibility of tailoring the solvent composition to suit gas compositions have led to an increase in the usage of this process.[5,6,8] Economically, the most important factor in designing an absorption–desorption process is the solvent circulation rate. A lower circulation rate leads to lower pumping energy cost and therefore reduction of the regeneration energy required that can include about 70% of the total cost of operation of the gas purification process.[9] Besides, vapor–liquid equilibrium (VLE) modeling of acid gas–aqueous amine systems is necessary for the synthesis, design, and analysis of gas sweetening units. There are two categories of VLE models for the description of gas–aqueous amine systems: the empirical models based on the Kent–Eisenberg model and activity coefficient- or excess Gibbs energy (Gex)-based models.

Kent and Eisenberg proposed a VLE model to predict the equilibrium partial pressures of H2S and CO2 in aqueous monoethanolamine (MEA) and diethanolamine (DEA) solutions.[10] Jou et al. applied such an approach for the correlation of H2S and CO2 solubilities in aqueous methyl DEA (MDEA) solutions.[3] Moreover, Chakma and Meissen extended the Kent–Eisenberg approach for the system of CO2–DEA–H2O.[11] According to Weiland et al., the Kent–Eisenberg correlation results show a good agreement with experimental data only in the loading range of 0.2 to 0.7 moles acid gas per mole of amine, and the model gives inaccurate results for mixed acid gases.[12] Haji-Sulaiman et al. extended the Kent–Eisenberg model to estimate the CO2 loading in the aqueous mixtures of DEA, MDEA, and DEA–MDEA, and it was observed that this model forecasts a relatively accurate carbon dioxide loading over a wide range of operating conditions.[13] Ebenezer evaluated the HYSYS capability to estimate the CO2 removal at operating conditions of minimizing hydrocarbon and chemical losses.[14] Also, Aliabad and Mirzaei studied the accuracy of HYSYS and ASPEN simulators in gas sweetening forecasting by aqueous amine solvents.[15] The Aspen HYSYS software was also applied to simulate and optimize the gas sweetening process using several amine types and blends, for example, DEA,[16] MEA,[16] MDEA,[16−18] DGA,[19] and MEG[20,21] aqueous solutions and the effects of operating conditions such as circulation rate, concentration, and inlet temperature of amine on the regeneration reboiler temperature and duty were studied.

So far, no comprehensive study has been conducted on the effects of process parameters on increasing the efficiency of natural gas sweetening units in Iran, and it seems that the study on this issue is very important to reduce production and processing costs and solve existing problems. This study aimed to investigate the maximum CO2 removal from the sour natural gas by an aqueous DEA solution in the Amak gas treating unit (GTP) of the National Iranian South Oilfields Company (NISOC). This removal process was optimized in different levels of the amine concentration, temperature, and circulation rate of lean amine using the Taguchi Method. Furthermore, the accuracy of the Kent–Eisenberg thermodynamic model in the Aspen HYSYS v.8 simulator to estimate CO2 absorption by an aqueous DEA solution in the gas sweetening process was investigated. The flowchart summarizing the applied method and the steps is demonstrated in Figure .

Figure 1

Flowchart summarizing the steps applied.

Studying the gas sweetening unit and providing field data can be considered as one of the advantages of this work. Also, applicable conditions allow using the obtained results to increase the performance efficiency of related units. On the other hand, the unit equipment was designed for specific conditions; therefore, the possibility of changing the process variables is not perceptible and these changes can be applied in a limited range. Besides, significant alterations in the operating variables can lead to changes in the quality of output products and disruption in downstream units. Thus, these limitations have prevented a more complete study by investigation on further variables with a wide range of variations.

In this article the definitions are presented. Then, the operating conditions and control are presented. In the Experimental Procedure and Sampling section, different stages of the experiments and sampling are described, and in the Steady-State Simulation and Optimization section, the capabilities of Aspen HYSYS and the Kent–Eisenberg thermodynamic model in the CO2 absorption estimation by an aqueous DEA solution in this case study are investigated. The results of the experimental and modeling study are reported as tables and figures in the Results and Discussion section and their evaluations are presented in this section. In the Conclusion section, an overview of the obtained results is provided.

Results and Discussion

Experimental and Analysis Results

The measured CO2 concentration in the sweet natural gas for each trial and sample is listed in Table . It can be observed that trial 2 has the highest variation in its results and also the proposed operational conditions in trial 7 including a temperature of 40 °C, a DEA solution concentration of 30 wt %, and an amine circulation rate of 260 m3 h–1 lead to the best CO2 removal result. The results of the signal-to-noise (S/N) ratio for each trial are presented in Table .

Table 1

L9 (33) Table and Experiment Results

	level description			results (CO2 concentration in sweet natural gas) (ppm, mole)
trial number	amine concentration (wt %)	temperature (C)	circulation rate (m3h–1)	sample 1	sample 2
1	25	40	220	6	8
2	25	50	240	35	41
3	25	60	260	5	8
4	28	40	240	4	4
5	28	50	260	3	4
6	28	60	220	19	18
7	30	40	260	trace (0.005)	trace (0.005)
8	30	50	220	0.5	2
9	30	60	240	1	3

Table 2

S/N Ratios of Trials

trials	1	2	3	4	5	6	7	8	9
S/N ratio	–16.990	–31.623	–16.484	–12.042	–10.970	–25.347	46.020	–3.274	–6.990

The effect of each factor was estimated by calculating the average value of the S/N ratios at the total levels of the factor. The factor effect was the arithmetic difference between the maximum and minimum amounts of S/N ratios and is listed in Table . The magnitude of the values in the last column (Lmax–Lmin) shows the influence of the factors and the minus sign indicates a decrease in the S/N ratio as a variety of levels. Preliminary reviews show that the amine concentration has the greatest difference in the levels, and consequently, it can affect the CO2 concentration in sweet natural gas as the target, significantly.

Table 3

Main Effects of the Individual Factors

factor	level 1	level 2	level 3	Lmax–Lmin
concentration	–21.699	–16.119	11.919	33.618
temperature	5.663	–15.289	–16.274	21.937
Rate	–15.204	–16.885	6.189	23.074

The evaluation of the influence of the individual factor was obtained from the analysis of variance (ANOVA) and is shown in Table . It can be revealed that the amine concentration with a contribution percent of 46.14% plays the most significant influence on the process of CO2 removal. The percentage of error contribution points to the accuracy of the experiments and experimental error, interactions, or uncontrollable factors causing it. The ANOVA results show that the percent contribution of error is low, 11.414%, and with the best estimate, it could be assumed that unconsidered factors did not vary during the experiments, and the experiments were performed under controlled conditions.[22,23]

Table 4

Analysis of Variance

factors	DOF	S·S	V	F	S′	percent P (%)
concentration	2	1947.419	973.709	17.172	1834.017	46.14
temperature	2	921.154	460.577	8.122	807.751	20.321
Rate	2	992.87	496.435	8.755	879.468	22.125
Error	2	113.402	56.701	17.172		11.414
Total	8	3978.848				100

To investigate the main effects of the factors on the trail conditions, the S/N ratio average for each parameter was plotted versus the various levels and is illustrated in Figure .

Figure 2

Value of S/N ratio at various levels for each control factor.

According to the slope of the lines, the concentration of lean amine has the highest impact on the response and an increase in lean amine concentration causes CO2 removal to increase strongly. Also, it can be observed that the increasing concentration of the aqueous DEA solution from 25 to 28 wt % (level 1 to 2) leads to an increase in the CO2 absorption gradually and increases the absorption steeply from 28 to 30 wt %. Thus, it is concluded that the effect of lean amine concentration on CO2 absorption is not linear and it can be a source of error for estimating Yexp. The second section of Figure describes the response graph as a function of the lean amine circulation rate. As can be seen, such as lean amine concentration, an increase in the circulation rate increases the carbon dioxide concentration but with a lower slope. The circulation rate from 220 to 240 m3 h–1 (level 1 to 2) has no weighty impact on the target, but from 240 to 260 m3 h–1 (level 2 to 3), the CO2 absorption increases sharply. The effect of the lean amine temperature changes on the CO2 absorption is pictured in the third section of Figure . Raising the lean amine temperature causes the S/N ratio to decrease. It is expected that temperature increase reduces gas absorption. However, reduction of the S/N ratio in the alteration temperature from 40 to 50 °C (level 1 to 2) is sharp and in another temperature alteration from 50 to 60 °C is almost constant. The effects of the last two factors are nonlinear too (as lean concentration) and cause errors.

Estimated Results at the Optimum Condition

To calculate the expected results at the optimum condition, the grand average of performance, Rt®, was evaluated. All factors in this work were significant and the performance at the optimum condition should be calculated using all of them. According to Table , a high level of lean amine concentration factor, a low level of temperature factor, and a high level of lean amine circulation rate factor were given at the highest S/N values and were considered as the optimal condition. The contribution of each factor was calculated and is presented in Table along with the optimum settings and levels of the various factors.

Table 5

Optimum Condition of CO2 Removal by an Aqueous DEA Solution

control factor	level description	level	contribution
concentration	30	3	20.551
temperature	40	1	14.296
circulation rate	260	3	14.822

Since the S/N ratio is used, the estimated result at the optimum condition can be converted back to the scale of original observation units. In this case, the expected result in terms of the S/N ratio is 41.036. This is equivalent to an average performance Yexp = 0.009, which is calculated using eqs and 2.

Interaction Study

The difference between the experimental and the estimated values can be related to the interaction between the control factors.[24] Hence, an interaction study was performed with the software Qualitek-4. The intensity of the presence of interactions was measured in terms of a numerical quantity via the angle between the two lines of the selected factors. Table shows the interacting pair factors and their severity index (S·I). The last column indicates the desirable levels to achieve the optimum condition. As can be observed, temperature and the circulation rate have the highest interaction with an S·I of 21.44%. Also, it can be found that the interaction indices are negligible, which is predictable because the factors are independent and could not interact with each other.

Table 6

Table of the Test of Interactions

interacting factor pair	column interaction	S·I %	optimum level
temperature × rate	2 × 3	21.44	[1,3]
concentration × rate	1 × 3	17.99	[3,3]
concentration ×temperature	1 × 2	10.11	[3,1]

Simulation Results

To evaluate the capability of Aspen HYSYS in estimating the CO2 absorption by an aqueous DEA solution in the gas sweetening process, the field data were compared with the simulation results. Thus, after running each trial and sample analysis, the operating conditions of each trial were fed to Aspen HYSYS and the process was stimulated by a selection of amine packages and the Kent–Eisenberg thermodynamic model.

The input data for the absorption column in the simulation was as follows: the pressure at the top of the tower was 28 bar, the pressure at the lower part was 29.75 bar, and the number of trays was 20. For the stripping column simulation, the pressures at the top and the lower part of the column were 1.7 and 2.2 bar, respectively. Also, the number of trays was 24, and to converge the stripping column, the temperatures of the condenser and reboiler were specified with the values of 60 and 125 °C, respectively.

According to the unit design basis, the lean amine loading must be less than 0.02 mole (CO2 + H2S)/mole DEA. The amount of lean amine demonstrates the quality operation of the regeneration unit and increasing the amount of lean amine loading leads to a decrease in acid gas absorption. As mentioned earlier, after running each trial, lean amine was analyzed, and keeping the lean amine loading below the designed quantity leads to the operation of the reboilers of the regeneration unit in the maximum duty. Besides, rich amine loading was controlled by adjusting the amine circulation rate. According to the basic design of the unit, the amount of rich amine loading must be kept at 0.5 mole (CO2 + H2S)/mole DEA. The high acid gas loading enhances steel corrosion and consequently increases the iron sulfide content, which amplifies foaming tendency. The experimental and estimated CO2 contents in the sweet natural gas and amine loadings are compared in Table . It is found that the estimated CO2 concentration in the sweet natural gas by Aspen HYSYS is much greater than that of field data. Hence, the Kent–Eisenberg model in HYSYS is not an appropriate model for this process. Consequently, simulated amine loadings have some deviations from field data, but both of them are in the range of design basis of the plan.

Table 7

Comparison between Field and Simulation Data for CO2 Content and Loading

				loading (mole CO2 + mole H2S)/mole DEA
	CO2 concentration in sweet natural gas (ppm mole)				rich amine
trial No	lab 1	lab 2	simulation	lean amine	lab	simulation	error %
trial 1	6	8	216	0.0134	0.4929	0.4739	–3.85
trial 2	35	41	70	0.0124	0.4011	0.4216	5.11
trial 3	5	8	30	0.0120	0.4016	0.4142	3.14
trial 4	4	4	175	0.0090	0.3602	0.4752	31.93
trial 5	3	4	48	0.0115	0.4639	0.3948	–14.90
trial 6	19	18	35	0.0103	0.3475	0.4530	30.36
trial 7	0.005	0.005	147	0.0065	0.3922	0.4139	5.53
trial 8	0.5	2	68	0.0122	0.3453	0.3992	15.61
trial 9	1	3	22	0.0110	0.2880	0.3558	23.54

The experimental data and simulated results of H2S content in the sweet gas are presented in Table . As can be seen, there is a significant difference between the experimental and simulated values of H2S content in the sweet natural gas, while the operating conditions are controlled to achieve a value of less than 4 ppm of H2S in the sweet natural gas. Also, it is observed that regardless of the circulation rate, in the same level of lean amine concentration, such as trials 1 to 3 or trials 4 to 6, increasing the amine temperature leads to the reduction of the estimated CO2 content and an increase in the simulated H2S content in sweet natural gas (Figures –5). It seems that at a higher temperature, the kinetic effect is stronger than solubility decrease, and the H2S concentration in the sweet natural gas increases monotonically with lean amine temperature due to the decreasing solubility.

Table 8

Comparison between Field and Simulation Data for CO2 and H2S Content in the Sweet Natural Gas

	CO2 concentration in the sweet natural gas (ppm mole)			H2S concentration in the sweet natural gas (ppm mole)
trial no	lab 1	lab 2	Simulation	lab 1	lab 2	simulation
trial 1	6	8	216	0	0	1.04
trial 2	35	41	70	0	0	1.63
trial 3	5	8	30	0	0	2.98
trial 4	4	4	175	1	1	0.75
trial 5	3	4	48	1	1	1.50
trial 6	19	18	35	0	0	2.56
trial 7	0.005	0.005	147	0	0	0.58
trial 8	0.5	2	68	0	0	1.90
trial 9	1	3	22	0	0	2.29

Figure 3

Lean amine temperature vs. acid gas concentration in sweet natural gas at a lean amine concentration of 25 wt %.

Figure 5

Lean amine temperature vs. acid gas concentration in sweet natural gas at a lean amine concentration of 30 wt %.

Lean amine temperature vs. acid gas concentration in sweet natural gas at a lean amine concentration of 28 wt %.

Conclusions

In this study, the experimental design method was applied to formulate the experimental layout and optimize the operating conditions of a natural gas sweetening unit to remove the maximum CO2 by an aqueous DEA solution. For this purpose, an L9 orthogonal array Taguchi method as a statistical experimental design was applied, and the effects of lean amine concentration, temperature, and circulation rate in three levels as the key control factors on CO2 absorption by an aqueous DEA solution were investigated. The experiments were carried out in the Amak GTP at amine concentrations of 25, 28, and 30 wt %, the temperatures of 40, 50, and 60 °C, and the lean amine circulation rates of 220, 240, and 260 m3 h–1. Furthermore, the accuracy of the Aspen HYSYS Kent–Eisenberg thermodynamic model was evaluated and field data and simulation results were compared with each other. From this study, the following conclusions can be drawn:

It is revealed that the lean amine concentration is the most significant control factor on CO2 absorption using aqueous DEA solution with a contribution percent of about 46.14%, while the lean amine circulation rate and temperature have a contribution of 22.125 and 20.321%, respectively.

Field data indicate that the CO2 content for all the trials is less than 50 ppm mole.

The ANOVA-calculated error percent is lower than 15% (11.414%). This means that all of the key operating parameters are considered and no significant control factors are left out from the experimental condition.

The interaction study indicates that there is no significant interaction between the control parameters.

The maximum removal of CO2 from natural gas by an aqueous DEA solution is achieved at a lean amine concentration of 30 wt %, a temperature of 40 °C, and a circulation rate of 260 m3 h–1.

The effects of concentration, temperature, and circulation rate of lean amine on CO2 absorption are not linear.

The estimated results from the Kent–Eisenberg thermodynamic model in the Amine Package of Aspen HYSYS process simulator show deviations from field data. Thus, this simulator is not a proficient simulator for the determination of CO2 concentration in sweet natural gas in the natural gas sweetening process.

Regardless of the circulation rate, in the same level of lean amine concentration, increasing the amine temperature leads to the reduction of the estimated CO2 content and increase of the simulated H2S content in sweet natural gas.

A decrease in solubility leads to an increase in the concentration of hydrogen sulfide in the sweet natural gas with the increasing temperature of lean amine.

Simulated lean and rich amine loadings show some deviations from field data, but both of them are in the range of the unit design basis.

It seems that if there are no operational restrictions in changing the parameters affecting the process, more findings would be obtained and the results would be more generalizable.

Definition

Process Description

The unit was designed to reduce the H2S content in the sweet gas and sweet liquid to below 4 and 50 ppmv, respectively. As can be observed in Figure , the unit is divided into three sections: slug catchers, gas sweetening–amine trains A and B, and liquid sweetening. The feed gases are different sour gases from different stations that are mixed and enter the slug catchers SC-801A/B. In this stage, the mixture is condensed and separated. The obtained liquid is fed to the liquid sweetening unit for sulfur removal utilizing a stripping stream and the gas portion is fed to a gas sweetening unit for H2S and CO2 removal by using an aqueous DEA solution. The treated gas, which contains less than 4 ppmv of H2S and 0.4% mole of CO2, is sent to the NGL 700/800 unit and some of it is used as a stripping medium in the liquid sweetening unit to reduce the H2S content in the treated liquid up to less than 50 ppmv and enters the NGL 700/800 unit. The produced acid gas goes to the compression station for further treatment.

Feed source of GTP.

Process Chemistry

The acid gas absorption is not a physical process only. Only one fraction of H2S ionizes in water to hydrogen ions and sulfide ions[25]

DEA is a weak base and ionizes in water to form amine ions and hydroxyl ions[25]where R indicates the ethanol group CH2CH2OH. While H2S dissolves into a solution containing amine ions, a weakly bonded salt of the acid and the base is produced as below, and then, the sulfide ion is absorbed by the amine solution.[25]

Based on reaction , the reaction of salt formation is not complete. As the arrows indicate, an equilibrium level of H2S remains in the hydrocarbon flow. Thus, the overall reaction can be summarized as follows[25]

Operating variables are set to raise forward reaction in the absorption step and enhance the reverse reaction in the regeneration step. The CO2 absorption is achieved according to the following reactions[25]

The CO2 absorption is slower than the H2S absorption because reaction is carried out slowly and occurs first. The rate of all absorption reactions is enhanced at high pressures and low temperatures, and also, high H2S and CO2 contents shift the equilibrium reactions toward the right side. The amine regeneration is performed at low pressures and high temperatures and shifts the equilibrium of the mentioned reactions to the left side. On the other hand, the low H2S and CO2 partial pressures of the generated stripping vapor in the reboiler lead to a high driving force for the H2S and CO2 mass transfer.

Operating Conditions and Control

The absorption of H2S/CO2 into the aqueous amine solution is increased by five factors:[26,27] low temperature, low acid gas loading, high amine concentration, high H2S/CO2 partial pressures in the feed stream, and intimate contact. In general, the fourth and fifth factors are not operating variables and are fixed by the unit design criteria and choosing equipment in the absorbers’ design. Furthermore, low feed rates may, however, cause poor tray efficiency and thus somewhat a poor H2S/CO2 removal in comparison with achievable at or near design flow rates. In general, decreasing the temperature of the lean amine solution causes an increase in H2S/CO2 removal. Besides, the lean amine temperature must be maintained at 10 °C higher than the temperature of the gas feed stream to avoid any possible condensation of the hydrocarbon vapors. The lean amine is cooled typically by air to about 60 °C.

It should be noted that the acceptable acid gas removal efficiency depends on good aqueous amine solution regeneration and restricting the H2S/CO2 loading in the rich amine to favor the forward direction of reaction . The H2S/CO2 loading of the aqueous amine solution is controlled by adjusting the amine circulation rate.

Experimental Procedure and Sampling

As mentioned earlier, this study was undertaken on the GTP of Amak. To measure accurate and reliable data, all flow instruments of sour gas, sweet gas, lean amine, and rich amine were calibrated. After calibration, operating variables, including lean amine concentration, temperature, and circulation rate of the trial, were set and each trial was run. When a steady state was achieved, flow sampling was performed. The CO2 content of two obtained samples from sweet gas was measured, and the analysis of two samples of sour gas as a feed was performed to simulate a process with field data. The carbon dioxide and hydrogen sulfide contents in the gas samples were analyzed by the gas chromatography method. For this purpose, a calibrated Agilent 6890 series gas chromatograph equipped with a DB-1 capillary column was applied. The following chromatographic conditions were set to provide accurate results:[28] manual splitless injection, an inlet temperature of 105 °C, a total gas flow rate of 30 mL min–1, a column gas flow rate of 2 mL min–1, and a detector outlet temperature of 200 °C. Helium was used as a carrier gas.[28] The H2S + CO2 loading of lean amine must be less than 0.02 and it was measured for two samples after running each trial. It should be noted that the reboiler in the regeneration unit of GTP operated at the maximum duty to achieve the minimum amount of acid gas loading for lean amine and the acid gas loading of rich amine shall not exceed 0.5 mole H2S + CO2 per mole of amine and it was checked for two samples of rich amine.

Steady-State Simulation and Optimization

The capability of the Aspen HYSYS and Kent–Eisenberg thermodynamic model in the CO2 absorption estimation by an aqueous DEA solution in the Amak gas sweetening process was evaluated, and the field data were compared with the simulation results, as mentioned earlier.

The Taguchi method and the software Qualitek-4 were applied to optimize the operating variables of the CO2 removal process from sour gas including lean amine concentration, temperature, and circulation rate. The levels of the selected parameters are listed in Table . In conclusion, an L9 orthogonal array for three-level factors was selected as the experimental layout to design the trials and determine the effects of various parameters on the CO2 removal process yield using the aqueous DEA solution in GTP. To evaluate and analyze the results of the Taguchi design, CO2 concentration in the sweet gas was considered as the main target value with the quality characteristic of “Smaller is better”. Whereas the experiments included multiple samples per trial condition, the S/N ratios of the experimental results were used to compute the main effects of the individual factors. In general, the aim of any experiment is always to determine the highest possible S/N ratio for the result. A high S/N ratio implies that the signal is much higher than the random effects of the noise factors. After calculating the S/N ratio for each experiment, the average S/N value is calculated for each factor and level. Determining the differences between the S/N ratio values at the high and low levels of a factor presents the main effects of the single factor. To evaluate the quality characteristics and visual presentation, the average effects of the factors are graphed on an appropriate scale. Therefore, the S/N ratio for each factor and the average S/N ratio versus levels are also plotted to study the trend of the influence of the factors. The Taguchi design can also predict the optimal condition and performance by ANOVA analysis.

Table 9

Levels of Variables

	level
variable	1	2	3
lean amine concentration (wt %)	25	28	30
temperature (°C)	40	50	60
circulation rate (m3h–1)	220	240	260