Journal of Cleaner Production 201 (2018) 974e987 Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro Carbon footprint reduction of acid gas enrichment units in hot climates: A techno-economic simulation study Satyadileep Dara a, Aisha A. AlHammadi a, Abdallah S. Berrouk a, b, e, Fadi Al Khasawneh c, Abdulla Al Shaiba d, Yasser F. AlWahedi a, e, * a Chemical Engineering Department, Khalifa University of Science and Technology, PO Box 2533, Abu Dhabi, United Arab Emirates Mechanical Engineering Department, Khalifa University of Science and Technology, PO Box 2533, Abu Dhabi, United Arab Emirates Research & Development Department, Abu Dhabi National Oil Company, PO Box 898, Abu Dhabi, United Arab Emirates d Projects and Engineering Division, Al Yasat Petroleum Operations Company Ltd., PO Box 44476, Abu Dhabi, United Arab Emirates e Center for Catalysis and Separation, Khalifa University of Science and Technology, PO Box 127788, Abu Dhabi, United Arab Emirates b c a r t i c l e i n f o a b s t r a c t Article history: Received 16 April 2018 Received in revised form 10 July 2018 Accepted 7 August 2018 Available online 8 August 2018 In natural gas processing plants, acid gas enrichment (AGE) units play a vital role in increasing H2S purity in the acid gas feed of sulfur recovery units (SRUs). Moreover, AGE units also produce a CO2-rich gas stream that is often vented to atmosphere. If CO2 purity is sufﬁciently high, this stream can be used as an injection gas for enhanced oil recovery (EOR) or for sequestration. In hot climates, AGE units operate at signiﬁcantly low efﬁciencies owing to the exothermic nature of their operation. Any enhancement in the efﬁciency can reap signiﬁcant beneﬁts. In this work, we study the economic and environmental impact of a process scheme wherein a RanqueeHilsch vortex tube (RHVT) is used as a cooling system for a lean solvent in an AGE unit located in a hot region of the United Arab Emirates. A simulation model is built using the process simulator ProMax® and is validated using plant design data. It is found that reducing the lean solvent temperature increased the purity of H2S and CO2 product streams. At temperatures lower than 25 C, the inverse occurs as CO2 absorption becomes favorable thermodynamically. Consequently, a lean solvent temperature of 25 C is identiﬁed to be optimal, thus achieving the lowest energy consumption and carbon footprint, while maintaining high purities of the product gases. At the optimal temperature, the proposed scheme results in steam savings of 13 kg/s (equivalent to 40% reduction in total steam rate). This reduced energy consumption leads to an annual CO2 footprint reduction of 83.7 million kg (equivalent to 40% reduction in total CO2 footprint). The optimal lean solvent temperature increases the purity of the H2S-rich gas stream (acid gas) to 67.3 mol% compared to its base case value of 45.7 mol%. Further, the purity of CO2-rich gas stream increases to 97 mol% compared to its base case value of 89 mol%, thus making it suitable for EOR or sequestration. Economically, the evaluated annual energy savings translate to 11.2 million USD, at a crude oil price of 50 USD. The computed payback period is 1.3 years, thus showing the potential of the proposed process. The process scheme proved to be superior to other commercial alternatives from economic and environmental perspectives. © 2018 Elsevier Ltd. All rights reserved. Keywords: Acid gas enrichment RanqueeHilsch vortex tube CO2 footprint Process simulation 1. Introduction Acid gas enrichment (AGE) units play a vital role in any natural gas complex as they separate undesired acid gases (CO2 and H2S) to ensure compliance with safety and operational requirements. These units produce two major product streams: (i) a H2S-rich gas stream, which is used as a feed gas for sulfur recovery units (SRUs) * Corresponding author. Chemical Engineering Department, Khalifa University of Science and Technology, PO Box 2533, Abu Dhabi, United Arab Emirates. E-mail address: firstname.lastname@example.org (Y.F. AlWahedi). https://doi.org/10.1016/j.jclepro.2018.08.067 0959-6526/© 2018 Elsevier Ltd. All rights reserved. and (ii) a CO2-rich gas stream, which is often vented to the atmosphere. 1.1. Motivation CO2-rich gas stream, which is the major product stream of the AGE, has signiﬁcant potential applications. If the purity of the former is sufﬁciently high, this stream can be used as an injection gas for enhanced oil recovery (EOR) or can be sequestrated. This has triggered the investigation of various schemes and conﬁgurations to increase the purity of the CO2-rich gas stream to allow its use in S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 the aforementioned applications. Moreover, any enhancements in the operational efﬁciency and reduction in carbon footprint are desired. To this end, this paper focuses the attention on a commercial AGE plant, which operates on HIGHSULF process (a patented process that involves recycling a portion of acid gas to the gas absorber), based in United Arab Emirates and attempts to improve the performance in the aspects of operating costs and carbon footprint. 1.2. Literature survey Several studies were reported that assessed various techniques to improve the performance of AGE units in terms of product gas purity and energy consumption. Weiland and Khanmamedov (2010) carried out a process simulation study of three AGE processes: 1) the HIGHSULF process, 2) the AGE process with two absorbers, and 3) the conventional AGE process. They concluded that the HIGHSULF process results in a more concentrated (19% higher) acid gas stream compared with the conventional AGE process and the AGE process with two absorbers. Khanmamedov (2013) also recommended the HIGHSULF process over the other processes due to a potential reduction of 8% in capital cost of the downstream SRU besides the increased acid gas purity. To this end, the above two studies conclude that HIGHSULF process is the most efﬁcient scheme of the available AGE processes. On the other hand, it is worth to note that the commercial AGE unit being studied in this paper operates on the HIGHSULF process thus leaves no scope for any potential performance improvement by means of retroﬁtting with a different AGE scheme. Parks et al. (2010) mentioned that the use of FLEXSORB SE and FLEXSORB SE PLUS can replace MDEA (Methyldiethanolamine)-based solvents owing to their higher selectivity for H2S, thereby increasing the purity of the H2S-rich gas stream from 30 mol% to 65 mol%. However, this also leads to a major increase in the operating expenses due to high cost of proprietary solvents. In general, such schemes require major modiﬁcations to the plant hardware rather than simple changes in operating conditions or minor modiﬁcations. The latter are often more attractive as they do not involve any signiﬁcant downtime in unit operation. However, studies on the impact of major process variables are lacking in the ﬁeld of AGE. In contrast, in the ﬁeld of natural gas sweetening, several studies have been conducted on the impact of major operating variables. A lean solvent temperature has been reported to have the strongest inﬂuence on plant performance. For instance, Lunsford and McIntyre (1999) showed that decreasing the absorber temperature from 74 C to 30 C leads to a decrease in sweet gas H2S concentration from 6000 ppm to 10 ppm. Addington and Ness (2009) performed a simulation study of a Russian gas plant and showed that a reduction in lean solvent temperature from 50 C to 10 C resulted in a decrease in the concentration of sweet gas H2S from 28 ppm to 4 ppm. Pandey (2005) carried out an on-site parametric study of a commercial amine sweetening unit (Hazira Gas Processing Plant, Oil and Natural Gas Corporation Limited, India) and demonstrated that H2S absorption selectivity over CO2 can be increased by 25% by decreasing the lean solvent temperature from 50 C to 39 C. Dara and Berrouk (2017) performed a process optimization study on a middle-east-based commercial amine sweetening unit and identiﬁed the optimum conditions of major operating variables. They concluded that low solvent temperatures are suitable for better performance of the absorber owing to the increased solubility of acid gas governed by vapor liquid equilibrium. They showed that steam consumption rate can be reduced by 13.5% as a result of changing the lean solvent temperature from 65 C to 50 C. They reported that the concentration of sweet gas CO2 decreases with lean amine temperature until 54 C and then follows a reverse trend for further reduction 975 until 50 C. They concluded that the optimum choice of lean solvent temperature is not straightforward and should be investigated thoroughly in order to achieve optimum energy consumption and appropriate balance of absorption degrees of H2S and CO2. The aforementioned studies in natural gas sweetening units show that optimizing the lean solvent temperature can achieve two beneﬁts: (i) energy savings that would thus lead to cost savings and (ii) increased purity levels of H2S and CO2 streams. In hot climates where refrigeration costs are high, lowering the solvent temperature can be achieved by means of a RanqueeHilsch vortex tube (RHVT) provided a high pressure inert gas is available. The RHVT is a static device that uses vortex motion to separate a compressed gas ﬂow into a hot stream and a cold stream. Saberi et al. (2016) reported the improvement of surface grinding process performance through the minimum quality lubrication (MQL) technique coupled with a simple and inexpensive vortex tube by achieving a cold air jet temperature of /4 C compared with the conventional coolant temperature of 15.5 C. Jozic et al. (2015) investigated the application of the RHVT to produce cold air in the end milling process. They reported that the cutting speed increased by 50% upon using the RHVT for cooling, which enhanced process efﬁciency. Also, many novel techno-economic studies were carried out for investigating various innovative methods involving utilization of ambient conditions for energy conservation purposes. Esen et al. (2006) performed performance experiments and economic analysis of a horizontal ground source heat pump (GSHP) system for space heating. They proved that this system offers several economic advantages over many other conventional techniques. Esen et al. (2007) reported a techno-economic comparison between a ground-coupled heat pump (GCHP) system and an air-coupled heat pump (ACHP) system for space cooling. They concluded that GCHP systems are economically preferable to ACHP systems for the purpose of space cooling. Esen and Yuksel (2013) carried out experimental investigation of greenhouse heating by biogas, solar and ground energy in Elazig, Turkey climate conditions. They reported the effects of climatic conditions and operating parameters on the system performance parameters. 1.3. Contribution In this work, we propose the application of RHVT for cooling the lean solvent in an AGE unit based in the United Arab Emirates. Pressurized nitrogen gas, which is readily available in the plant as the by-product of the air separation unit in the same plant complex, is used as the RHVT feed. A simulation model is built for the combined system by using the process simulator ProMax® and is validated using the AGE plant design data. The model is used for studying the impact of lean solvent temperature on the product streams. Then, potential energy savings of the proposed scheme are evaluated. The economic feasibility and environmental impact of the RHVT-based AGE unit are assessed by considering achievable reductions in operating cost, energy consumption and carbon footprint. To the best knowledge of the authors, the integration of RHVT with AGE and the evaluation of the associated beneﬁts have not been studied in the reported literature. Also, the proposed process scheme is compared to other commercial processes in order to benchmark its potential beneﬁts against commercial options. 2. Process description 2.1. Acid gas enrichment unit Fig. 1 depicts the process ﬂow diagram of the commercial AGE unit considered in this study. In the commercial AGE unit, feed gas 976 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Fig. 1. Process ﬂow diagram of the commercial AGE unit. at 180 kPa and 57 C enters the absorber bottom, while the lean MDEA solvent is fed to the top of the absorber. There are three provisions for solvent entry at the 9th, 11th, and 13th trays. During normal operation, the 13th tray is used. In the absorber, the downﬂowing MDEA solution absorbs H2S and CO2 from the ascending sour gas to produce a sweetened gas stream. The ascending gas enters the wash water section of the column. In this section, the gas is washed with water in order to minimize any solvent carryover in the gas phase. CO2-rich gas leaves the absorber overhead at 160 kPa and 55.1 C. Then, it enters the absorber off-gas knock out drum, where any condensed water is separated. Rich MDEA solvent leaves the absorber from the lowest tray. Then, it is pumped at 900 kPa to the lean/rich solvent heat exchanger where the stream is heated to 117 C by exchanging heat with the hotter lean solvent exiting the regenerator. The hot rich MDEA enters the regenerator at the top section (tray 18). Low pressure (LP) steam is supplied to the reboiler, which vaporizes a portion of water associated with the MDEA solvent. Steam rising through the column regenerates the MDEA by stripping out the evolved H2S and CO2. The overhead gas leaving from the regenerator at 195 kPa and 112 C passes through an air fan cooler, which cools the gas to 55 C, thus resulting in the partial condensation of water. The two phases are separated in the reﬂux drum, where the water is pumped to the regenerator as reﬂux. H2S-rich gas at 185 kPa and 55 C splits into two streams. Major split feeds the acid gas header while the other is recycled to the absorber. The lean MDEA from the regenerator bottom is pumped to the lean-rich heat exchanger, where it is cooled to 80 C. A colder lean solvent is recombined and cooled further across a lean solvent air fan cooler to 55 C before it enters the absorber again. 2.2. RanqueeHilsch vortex tube (RHVT) The RHVT or the vortex tube is a static device used to separate a pressurized gas into a colder stream and a hotter stream (Fig. 2). Eiamsa-ard and Promvonge (2008) explained that this separation of streams is known as the temperature separation effect or the energy separation effect. Energy separation is the process of redistribution of the total ﬂuid energy ﬂow without additional Fig. 2. Schematic representation of the RanqueeHilsch vortex tube. S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 work or heat. As a result, a part of the ﬂuid leaves with high energy, whereas the other part leaves with low energy compared with the inlet ﬂuid. When the compressed gas is introduced into the tube through the tangential nozzle, the nozzle forces the gas to ﬂow in a centrifugal motion through the tube. This type of motion increases the pressure of the gas along the outer edge of the tube and a reduction in pressure in the middle of the tube. The temperature separation within the tube is induced by the friction experienced by gas with the tube wall. Because the speed of the gas in the outer edges is lower than that at the center, energy is transferred from the center region to the tube wall region, thus generating a hot gas stream near the wall and a cold gas stream at the center. 3. Methodology This study proposes employing a RHVT for producing a cold nitrogen stream, which can be used as a coolant stream in the lean solvent-air cooler. Fig. 3 depicts the proposed scheme that integrates the RHVT and AGE unit. High-pressure nitrogen enters the RHVT and splits into hotter and colder streams. The latter then mixes with ambient air in an air-nitrogen mixer to provide a coolant stream at sufﬁciently lower temperatures such that it cools down the lean solvent to the desired levels. The methodology employed in this study is as follows (see Fig. 4): 1. Develop the simulation model for the commercial AGE unit considered in this study and validate the model using plant data to ensure an prediction-error level below 15% 2. Perform sensitivity analysis to examine the effect of lean solvent temperature on plant performance and thereby identify the optimum temperature 3. Evaluate the nitrogen ﬂow rate required to meet the optimum lean solvent temperature, through mass and energy balance across the air-nitrogen mixer 4. Evaluate the carbon footprint reduction achieved via energy savings 5. Perform economic analysis to assess the proﬁtability of the proposed scheme 977 To this end, the following four models are developed for carrying out the aforementioned methodology. 1 2 3 4 AGE simulation model Lean solvent cooling system model Carbon footprint model Economics model 3.1. AGE simulation model This section discusses the phenomena of acid gas absorption in MDEA followed by the reaction kinetics based model employed to simulate it. The absorption of H2S and CO2 by MDEA from the gas phase entails three steps: 1. Mass transfer of H2S and CO2 molecules from the bulk gas phase to the gas/liquid interface 2. Dissolution of H2S and CO2 molecules from the gas phase into the liquid phase, which is governed by thermodynamic equilibrium 3. Reaction of H2S and CO2 molecules with MDEA in the liquid phase Kohl and Nielsen (1997) described the reaction chemistry of MDEA with H2S and CO2 by using the following chemical reactions: 1) The MDEA reaction with H2S is given by CH3N(C2H4OH)2 þ H2S )/ CH3(C2H4OH)2NHþ þ HS (1) 2) The MDEA reaction with CO2 is given by CO2 þ H2O þ CH3N(C2H4OH)2 )/ CH3(C2H4OH)2NHþ þ HCO 3 (2) The reaction between H2S and MDEA occurs via a direct proton transfer mechanism as shown in equation (1). This reaction is fast Fig. 3. Proposed scheme involving the integration of the RHVT with the AGE unit. 978 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Fig. 4. Methodology of the study on the proposed scheme. and proceeds to equilibrium instantaneously. Furthermore, mass transfer from the gas phase to the gas/liquid interface is fast. Therefore, the H2S absorption is predominantly controlled by the vapor/liquid equilibrium at the gas/liquid interface. On the contrary, CO2 reaction with MDEA is relatively slow. CO2 undergoes a hydrolysis reaction to form carbonic acid ﬁrst, which then undergoes acidebase reaction with MDEA to yield the overall reaction shown in equation (2). Kidnay and Parrish (2007) explained the CO2 absorption phenomena with MDEA using the following reactions: 1) The CO2 hydrolysis step is given by CO2 þ H2O )/ H2CO3 (3) 2) The Carbonic acid dissociation step is given by H2CO3 )/ Hþ þ HCO 3 (4) 3) The MDEA protonation step is given by CH3N(C2H4OH)2 þ Hþ )/ CH3(C2H4OH)2NHþ (5) CO2 dissolves in water and hydrolyses to form carbonic acid (equation (3)), which in turn slowly dissociates to bicarbonate and hydrogen ion (equation (4)). The latter then undergoes an acidebase reaction with the amine to yield a protonated amine (equation (5)). It should be noted that the dissociation of carbonic acid to bicarbonate (equation (4)) is the rate limiting step. Vapor/ liquid equilibrium at the interface also plays a major role in controlling CO2 absorption. For simulating the absorption of acid gases, this study employs TSWEET kinetics package of ProMax simulator, licensed by Bryan Research and Engineering Inc. This method accounts for the mass transfer of acid gases (H2S and CO2) and liquid phase reaction kinetics through a combination of reaction kinetic models and ideal/ real tray ratios. S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Below are the generalized rate equations of the TSWEET package to model the relatively slow CO2 absorption. 1) The rate of forward reaction of the controlling step of CO2 absorption is given by r ¼ k½H2 CO3 (6) 2) The rate constant of the forward reaction of CO2 absorption is given by k ¼ Aexp E=RT T b (7) where r is the rate of forward reaction, T is the reaction temperature, k is the rate constant, A is the frequency factor, E is the activation energy, and b is a constant quantifying temperature effect. 3) The rate of backward reaction of the controlling step of CO2 absorption is given by i h i h r 0 ¼ k0 Hþ HCO 3 (8) 4) The rate constant of the backward reaction of the controlling step of CO2 absorption is given by 0 k ¼ A exp E0=RT T b 0 0 (9) where r 0 is the rate of backward reaction, k0 is the rate constant, A0 is the frequency factor, E0 is the activation energy, and b’ is the constant quantifying temperature effect. Further, R represents the gas constant and [H2CO3], [Hþ], & [HCO3] represent the concentrations of the reactants. 979 5) Hence, the complete reaction rate of CO2 absorption is given by i h i h rnet ¼ k½H2 CO3 k0 Hþ HCO 3 (10) where rnet is the net rate of reaction of the controlling step of CO2 absorption. The net absorption levels of CO2 directly inﬂuence the reaction medium temperature, and hence, affect the H2S equilibrium solubility and impact the product gas H2S. The ideal/real tray ratios are obtained based on industrial experience. The ratio of ideal to real stages is assumed to be ~3 in the absorber, as recommended by Bryan Research and Engineering Inc. Similarly, the regenerator is modeled using the ‘TSWEET alternate stripper’ package that assumes a ratio of 2. Therefore, the absorber and regenerator are modeled with 8 and 11 ideal stages, respectively. To predict the vapor liquid equilibrium, the ProMax simulator adopts a reﬁned Soave Redlich Kwong (SRK) model. Soave (1972) reported that this model uses the SRK equation of state for modeling vapor phase interactions and the electrolytic non random two liquid (NRTL) for predicting liquid phase properties. The ProMax simulation model for the AGE unit is depicted in Fig. 5. 3.2. Lean solvent cooling system model To calculate the air and nitrogen ﬂowrates required for generating the coolant stream, a standalone ProMax model for the air mixture-lean solvent cooler is used. The following assumptions are made in the model: 1. The heat capacity of the gas mixture (Cpmix Þ is equal to that of nitrogen (CpN2 Þ. 2. The minimum end temperature approach across the cooler is 5 C. Fig. 5. Process simulation model of the commercial AGE unit. 980 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 3. The supply temperature of nitrogen from the RHVT system (TN2 in ) is 10 C. 4. The supply temperature of air ðTAir in Þ is 50 C, in order to simulate a summer case in the UAE. The governing model for the lean solvent cooling system is as follows: 1) The energy balance across the lean solvent cooler is given by Qsolvent cooler ¼ m_ mix Cpmix DTmix (11) where DTmix ¼ ðTmix in Tmix out Þ After determining the total ﬂow rate of the mixture, the mass ﬂow rates of cold nitrogen gas and air are computed by imposing mass and energy balances across the cooler, as shown in Fig. 6: 2) The mass balance across the mixer is given by m_ mix ¼ m_ Air þ m_ N2 (12) 3) The energy balance across the mixer is given by m_ Air CpAir DTAir þ m_ N2 CpN2 DTN2 ¼ m_ mix Cpmix DTmix (13) where DTAir ¼ ðTAir in Tmix in Þ and DTN2 ¼ ðTN2 in Tmix in Þ Based on the targeted mixture temperatures, the required ﬂowrates of air and nitrogen can be computed as follows: m_ N2 ¼ m_ mix m_ mix Cpmix DTmix CpN2 DTN2 CpAir DTAir CpN2 DTN2 (15) where m_ Air , m_ N2 , and m_ mix are the mass ﬂow rates of the air stream, nitrogen stream, and coolant mixture, respectively; CpAir , CpN2 , and Cpmix are the heat capacities of air, nitrogen, and the coolant mixture, respectively; DTAir , DTN2 , and DTmix are the changes in temperature for air, nitrogen, and the coolant mixture across the heat exchanger, respectively; Tmix in and Tmix out are the mixture inlet and outlet temperatures, respectively. TAir in and TN2 in are the inlet temperatures of air and N2, respectively, and Qsolvent cooler is the cooling duty of the lean solvent cooler. 3.3. Carbon footprint model The carbon footprint reduction corresponding to given steam savings is evaluated using equations (16) to (18). First, the theoretical energy required for producing the low pressure steam (Qsteam ), at 600 kPa is calculated using equation (16) based on the mass of water (mwater ), heat capacity (Cpwater ), saturation temperature ðTSaturation ), reference temperature (TRef ), and latent heat of vaporization (l). The net heat required (Qnet ) is calculated using equation (17) by assuming an overall efﬁciency (h) for the steam generation process. Finally, the carbon footprint reduction (mCarbon ) that corresponds to the reduction in steam consumption rate is calculated based on the study by U.S. Energy Information Administration, which reported the amount of CO2 (C ) emitted per 1 kJ energy of natural gas. 4) The mass ﬂow rate of air required is given by m_ Air m_ mix Cpmix DTmix CpN2 DTN2 ¼ CpAir DTAir CpN2 DTN2 5) The mass ﬂow rate of nitrogen required is given by 1) The theoretical energy required for steam generation is given by (14) Qsteam ¼ mwater Cpwater TSaturation TRef þ mwater l 2) The net energy required for steam generation is given by Fig. 6. Illustration of mass and energy balance across air-nitrogen mixer. (16) S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Qnet ¼ Qsteam =h (17) 981 Table 1 Summary of inputs for AGE simulation model. AGE simulation model inputs 3) The carbon footprint reduction corresponding to the energy of given amount of steam saved is given by mCarbon ¼ Qnet C (18) 3.4. Economics model The total operating cost of an AGE unit is comprised of 1 Cost of steam consumption in the regenerator reboiler 2 Cost of electrical power consumed by the solvent pumps 3 Cost of power consumption of cooling fans of the lean solventair cooler The power consumption across the solvent cooler is unaffected by the change in the cooling mixture (air þ nitrogen) since the cooling fans draw a constant volumetric ﬂow rate. Furthermore, the changes in solvent properties such as density and viscosity are minor across the studied solvent temperatures, and hence, they do not affect the pumping load. Therefore, the costs of electrical power consumed by the solvent pump and cooling fans remain constant with solvent temperature, and hence, are not considered as operating costs. Also, this study did not take the price of nitrogen into account since the latter is readily available as the by-product of the air separation unit in the same plant complex. To summarize, operating cost savings are estimated solely based on the decreased steam consumption of the regenerator reboiler. Monetary savings due to steam reduction are evaluated based on the cost of natural gas required for producing the steam. The required quantity of natural gas (mnaturalgas ) is calculated using equation (19) based on the heating value of natural gas (Н natural gas ) and the net heat required (Qnet ) for producing the steam. The cost of the former is evaluated using the correlation provided by the Abu Dhabi National Oil company (equation (20)), which is a function of crude oil price (CP). 1) The amount of natural gas corresponding to energy of given amount of steam is given by mnaturalgas ¼ Qnet . Н natural gas Feed gas temperature, ºC Feed gas pressure, kPa Feed gas ﬂow rate, kg/s Feed H2S ﬂow rate, kg/s Feed CO2 ﬂow rate, kg/s Feed H2O ﬂow rate, kg/s Feed hydrocarbons ﬂow rate, kg/s Absorber pressure drop, kPa Absorber overhead pressure, kPa Actual number of absorber trays Ratio of real to ideal trays (absorber) Lean solvent ﬂow rate, kg/s Lean solvent concentration, wt % Lean solvent temperature, oC Number of regenerator trays Regenerator pressure drop, kPa Regenerator overhead pressure, kPa Reboiler steam rate, kg/s Reboiler steam pressure, kPa Acid gas condenser temperature, oC Ratio of real to ideal trays (regenerator) 57 180 33.4 4.6 27.4 1.3 0.1 20 160 18 3 277.8 40 55 22 55 195 32.5 600 55 2 Table 2 Summary of inputs for the lean solvent cooling system model, carbon footprint model, and economics model. Lean solvent cooling system model inputs Heat capacity of air (CpAir Þ, kJ/kg. oC Heat capacity of nitrogen (CpN2 Þ, kJ/kg. oC Heat capacity of coolant mixture (Cpmix Þ, kJ/kg. oC Minimum end temperature approach, oC Cold nitrogen temperature ðTN2 in Þ, oC Cold air temperature ðTair in Þ, oC Carbon footprint model inputs Heat capacity of water ðCpwater Þ, kJ/kg. oC Steam pressure, kPa Saturation temperature of steam ðTSaturation ), oC Reference temperature (TRef ), oC Latent heat of vaporization (l), kJ/kg Overall process efﬁciency of steam generation (h ) Cold air temperature (Tair in ), oC Amount of CO2 (C) emitted per 1 kJ of energy of natural gas, kg Economics model inputs Heating value of natural gas (Н naturalgas ), kJ/kg. Costing parameter1 (a) Costing parameter2 (b) Costing parameter3 (g) Unit conversion factorðCF) 1 1.039 1.039 5 10 50 4.184 600 158.8 20 2085 0.65 50 5 E5 42300 13.5 50 0.5084 1000 (19) model. 2) The cost of processed natural gas per unit mass is given by Cnaturalgas ðUSD=kgÞ ¼ ðða*ðCPÞ þ bÞ*gÞ = CF 3) The total cost of natural gas corresponding to energy of given amount of steam is given by TCnaturalgas ðUSDÞ ¼ Cnaturalgas * mnaturalgas 4. Results and discussion (20) (21) where a, b, and g are the cost parameters for estimating the cost of processed natural gas in the industry and CF is a conversion factor. 3.5. Case study inputs Table 1 presents the inputs and parameters used for the AGE simulation model, while Table 2 presents those of the lean solvent cooling system model, carbon footprint model, and economics The plant design data of the AGE unit (shown in Table 1) are used to validate the simulation model. A comparison of the model predictions and plant data is presented in Table 3, which shows that all the process variables are predicted with an error less than 15%. It should be noted that all the parameters are predicted with an error signiﬁcantly lower than 10%, except for the CO2 concentration in the H2S-rich gas stream and the regenerator overhead. The error in this case is 10.4%. The TSWEET package is more optimized for predicting H2S absorption in MDEA solvents when compared with CO2. This is because the accurate prediction of CO2 within an error of 10% is usually not required in natural gas sweetening systems. 4.1. Parametric analysis This section also discusses various results (Figs. 7a and 10) 982 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Table 3 Summary of model validation with plant data. Stream Parameter Model prediction Plant data % Error CO2-rich gas (Product gas from absorber overhead) Temperature (ºC) Pressure (kPa) Flow rate (kg/s) H2S ﬂow rate (kg/s) CO2 ﬂow rate (kg/s) Temperature (ºC) Pressure (kPa) Flow rate (kg/s) Temperature (ºC) Pressure (kPa) Flow rate (kg/s) H2S ﬂow rate (kg/s) CO2 ﬂow rate (kg/s) Temperature (oC) Pressure (kPa) MDEA ﬂow rate (kg/s) H2S ﬂow rate (kg/s) CO2 ﬂow rate (kg/s) Temperature (ºC) Pressure (kPa) Flow rate (kg/s) H2S ﬂow rate (kg/s) CO2 ﬂow rate (kg/s) 55.1 160 21.7 0.0037 20.7 67.0 157 289.7 112.0 195 26.9 4.643 6.7 55 900 111.1 0.0166 0.0166 55 185 11.8 4.62 6.7 53.6 160 22.8 0.0038 21.4 68.5 157 289.1 111.8 195 28.6 4.640 6.0 55 900 111.0 0.0170 0.171 55 185 11.1 4.61 6.0 2.7 0 4.8 2.6 3.3 2.2 0 0.2 0.2 0 5.9 0.1 10.4 0 0 0.1 2.3 2.9 0 0 5.9 0.2 10.4 Rich solvent Regenerator overhead Lean solvent H2S-rich gas (Acid gas) produced using the models explained in section 3. The lean solvent temperature is varied between 55 C and 21 C to investigate its impact on the following process parameters: 1) CO2 purity of CO2rich gas stream (product gas from absorber overhead stream), 2) H2S purity of H2S-rich gas stream (acid gas from regenerator overhead), and 3) the steam consumption rate. It should be noted that the concentration of H2S in the absorber overhead is maintained at ~500 ppmv throughout the analysis by varying the steam rate in the regenerator reboiler. Fig. 7a depicts the effect of lean solvent temperature on the CO2 content in the product gas. The latter increases from 89 mol% to 97 mol% as the lean solvent temperature decreases from 55 C to 25 C. Afterward, the CO2 content in the product gas follows an opposite trend and decreases to 96 mol% as the lean solvent temperature decreases to 21 C. Fig. 7a also shows the effect of lean solvent temperature on H2S content of the H2S-rich gas stream. The latter increases from 45 mol% to 69.3 mol% as the lean solvent temperature decreases from 55 C to 25 C. Thereafter, the H2S content in the H2S-rich gas stream follows an opposite trend and decreases to 65.5 mol% as the lean solvent temperature decreases to 21 C. Within the temperature range between 55 C and 25 C, H2S absorption process is governed by vapor/liquid equilibrium at the phase interface, while CO2 absorption is governed by reaction kinetics in the liquid phase. Because the dissolution of H2S from the gas phase to the liquid phase is exothermic, lower temperatures favor higher H2S concentrations in the liquid phase, thus leading to higher absorption of H2S. Within the same temperature range, CO2 absorption is dominated by the reaction kinetics in the liquid phase, which tends to slow down with decreasing temperature. The combined effect of both phenomena leads to higher purity levels in the CO2-rich gas stream and H2S-rich gas stream. Below 25 C, the vapor/liquid equilibrium of CO2 causes higher CO2 absorption in the solvent, thus reducing the driving force available for H2S absorption. This leads to a higher concentration of CO2 in the H2S-rich gas stream and higher concentration of H2S in the CO2-rich gas stream, which leads to lower purity levels. For applications such as EOR and carbon sequestration in depleted oil and gas reservoirs, Abbas et al. (2013) reported that a CO2 purity of 95 mol% is essential. At 25 C, the proposed scheme results in a product gas with a CO2 purity of 97 mol%, which is signiﬁcantly higher than the base case CO2 purity in the AGE unit (89 mol%), thus making it suitable for the aforementioned applications. Fig. 7b depicts the effect of the lean solvent temperature on the steam consumption rate. The steam consumption rate decreases from 32.6 kg/s to 19.6 kg/s as the lean solvent temperature decreases from 55 C to 25 C, and then increases to 20 kg/s as the lean solvent temperature decreases to 21 C. Between 55 C and 25 C, the higher equilibrium driving force for H2S absorption combined with slower CO2 reaction kinetics enhance the solvent afﬁnity for H2S. This implies that a lean solvent with a lower level of regeneration (higher remnant concentrations of absorbed H2S) can achieve the desired product gas H2S content (500 ppmv). Therefore, the steam consumption rate decreases owing to the lower level of regeneration required. Similarly, for temperatures lower than 25 C, the increased rate of CO2 absorption adversely affects the driving force available for H2S absorption. This in turn demands a higher level of solvent stripping to meet the desired H2S content in the product gas. Therefore, the required steam rate increases. To summarize, the optimum lean solvent temperature is 25 C owing to the lowest steam consumption rate and the highest purity levels of H2S-rich gas streams and CO2-rich gas streams. Fig. 8 presents the required ﬂow rates of air and nitrogen for achieving various lean solvent temperatures, which are determined using equations (11) to (15) mentioned in section 3.2. It is found that 864 kg/s of nitrogen is needed to reduce the lean solvent temperature to the optimum value of 25 C. Fig. 9 shows the net steam savings and the corresponding reduction in carbon footprint as a function of nitrogen ﬂow rate. Steam savings and the corresponding carbon footprint increase with an increase in the nitrogen ﬂow rate until the optimum value of 864 kg/s and follow an opposite trend thereafter. Steam savings of 13 kg/s and an annual carbon footprint reduction of 83.7 million kg are achieved as a result of the proposed scheme. Fig. 10 depicts the potential annual monetary beneﬁts for various crude oil prices as a function of nitrogen ﬂow rate used in the proposed scheme. At a crude oil price of 50 USD, the evaluated annual energy savings translate to 11.2 million USD. S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 983 Fig. 7. a)Effect of lean solvent temperature on the H2S content of H2S-rich gas (acid gas) and CO2 content of CO2-rich gas (product gas) (All other model parameters are as shown in Tables 1 and 2). b): Effect of lean solvent temperature on the steam consumption rate (All other model parameters are as shown in Tables 1 and 2). 4.2. Economic analysis 4.3. Comparison with conventional processes To investigate the proﬁtability of the proposed scheme, it is important to incorporate the capital cost of the RHVT for the given process conditions. The cost of the RHVT is 2.38 million USD, as stated by Tunkel (2017) (Universal Vortex, Inc., Designer of industrial vortex tubes). Further, a Lang factor (deﬁned as the ratio of the total cost of installation to the cost of the equipment) of 6.0 is considered, as recommended by Peters et al. (2003). This leads to a total installation cost of 14.3 million USD for the RHVT. Thus, the proposed scheme can return its capital investment in 1.3 years. This demonstrates the potential of the proposed scheme in enhancing AGE unit economics in addition to reducing carbon footprint. Furthermore, the improvement in the purity of CO2-rich gas stream to the levels adequate for EOR application and carbon sequestration justiﬁes its implementation. This section presents the performance of various conventional processes (reported by Weiland and Khanmamedov (2010)) and Khanmamedov (2013)) listed below. 1) HIGHSULF process, which is the process employed in the AGE unit being studied 2) AGE process with separate feeds 3) AGE process with two absorbers The simulation model shown in Fig. 5 is used for analyzing the HIGHSULF process. However, the temperature of the solvent-air cooler is maintained at 55 C, which is adequate for the ambient temperature of the UAE. For the other two processes i.e. AGE process with two separate feeds and the AGE process with two absorbers, process simulation models shown in Figs. 11 and 12 are 984 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Fig. 8. Effect of lean solvent temperature on the required mass ﬂow rates of nitrogen and air (All other model parameters are as shown in Tables 1 and 2). Fig. 9. Effect of nitrogen mass ﬂow rate on the reductions in steam consumption rate and annual carbon footprint (All other model parameters except lean solvent temperature are as shown in Tables 1 and 2). developed using the same kinetic and thermodynamic methods as that of Fig. 5. These processes are compared with the proposed method in various aspects such as product and acid gas purities, steam consumption rate, annual operating costs and capital investments. As can be seen from Table 4, AGE process with two separate feeds and the AGE process with two absorbers do not even meet the product gas H2S concentration of 500 ppm despite increasing the reboiler steam rate to 32.6 kg/s, which is the design limit. This clearly concludes that these two processes are not the feasible retroﬁt options. The HIGHSULF process, which is the base case, achieves the product H2S purity level of 500 ppm at a steam consumption of 32.6 kg/s. On the other hand, the proposed method achieves the same at a steam consumption of 19.6 kg/s. Furthermore, the acid gas CO2 purity of the proposed case is also signiﬁcantly higher than that of the base case as delineated in section 4.1. Also, as explained in section 4.2, this scheme incurs a capital investment of 14.3 million, which can be recovered within 1.3 years. 5. Conclusion This study proposed a conﬁguration involving the integration of the RHVT cooling system in the AGE unit. A process simulation model of a commercial middle-east-based AGE unit was developed S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 985 Fig. 10. Effect of nitrogen mass ﬂow rate on the monetary beneﬁts for various crude oil price scenarios (All other model parameters except lean solvent temperature are as shown in Tables 1 and 2). Fig. 11. Process simulation model of the AGE process with separate feeds. using the process simulator ProMax 4.0, which uses data-driven kinetic models to address gas absorption. A parametric analysis was carried out to quantify the reduction in operational energy as a function of lean solvent temperature. It was found that reducing the lean solvent temperature to an optimal temperature of 25 C resulted in increased purity levels of H2S and CO2 in the two product streams. Energy balance calculations revealed that a nitrogen ﬂow rate of 864 kg/s is required as feed for the RHVT to achieve the optimum lean solvent temperature. A 40% reduction in both steam consumption and annual carbon footprint can be achieved using the proposed scheme. Economic analysis estimated an annual operating cost savings of 11.2 million USD when using the integrated system. The computed payback period for the RHVT is 1.3 years. This highlights the potential of the RHVT integrated system in reducing the operating cost and carbon footprint of the AGE unit while producing a CO2-rich gas stream with adequate purity for EOR and sequestration applications. Also, the potential beneﬁts of the proposed method were compared with those of the other alternate commercial processes. The proposed scheme manages to achieve the required speciﬁcations of low H2S content and suitable CO2 gas for EOR at a substantially lower operating cost and carbon footprint. 986 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 Fig. 12. Process simulation model of the AGE process with two absorbers. Table 4 Comparison of the proposed method with conventional processes. Case Product gas HIGHSULF - Combined feeds scheme (Base Case/Current AGE process) Separate feeds scheme Multiple absorbers scheme Base case þ RHVT(Proposed scheme) a Steam consumption rate Annual operating cost savings Capital cost H2S, ppm CO2, % Acid Gas H2S, % CO2, % kg/s USD USD 500 89 45 48.2 32.6 0 0 17263 2362 500 86.1 92.4 97 46.1 32.8 69.3 45.2 58.5 22 32.6 32.6 19.6 0 0 11.2 million NAa NAa 14.3 million NA e Not applicable; Economic analysis not required since these processes do not meet the product speciﬁcations. Nomenclature A A’ b b’ C CF CP Cnaturalgas CpAir Cpmix CpN2 Cpwater E E0 Н naturalgas k k0 Frequency factor of H2 CO3 dissociation reaction, 1/s Frequency factor of reaction between [Hþ] and [HCO3], 1/mol. 1/s Constant quantifying temperature effect of H2 CO3 dissociation reaction Constant quantifying temperature effect of reaction between [Hþ] and [HCO3] Amount of CO2 emitted per 1 kJ energy of natural gas, kg Unit conversion factor Price of crude oil, USD Cost of processed natural gas per unit mass, USD/kg Heat capacity of air, kJ/kg. oC Heat capacity of mixture, kJ/kg. oC Heat capacity of nitrogen, kJ/kg. oC Heat capacity of water, kJ/kg. oC Activation energy of H2 CO3 dissociation reaction, kJ/mol Activation energy of reaction between [Hþ] and [HCO3], kJ/mol Heating value of natural gas, kJ/kg Rate constant of H2 CO3 dissociation reaction, 1/s Rate constant of reaction between [Hþ] and [HCO3], 1/ mol. 1/s mCarbon Carbon footprint reduction, kg mnaturalgas Natural gas quantity required for steam generation, kg Mass of water, kg mwater Mass ﬂow rate of air, kg/s m_ Air Mass ﬂow rate of mixture, kg/s m_ mix Mass ﬂow rate of nitrogen, kg/s m_ N2 Qnet Net heat required for steam generation, kJ Qsolvent cooler Duty across the lean solvent cooler, kJ Qsteam Theoretical heat required for steam generation, kJ R Gas constant, kJ/mol. oC r Rate of H2 CO3 dissociation reaction, mol/m3.s Rate of reaction between [Hþ] and [HCO3], mol/m3.s r0 rnet Net rate of CO2 absorption reaction, mol/m3.s T Reaction temperature, oC TAir in Air inlet temperature, oC TCnaturalgas Total cost of processed natural gas, USD Air-N2 mixture inlet temperature, oC Tmix in Tmix out Air-N2 mixture outlet temperature, oC TN2 in N2 inlet temperature, oC TRef Reference temperature, oC TSaturation Saturation temperature of water, oC ½H2 CO3 Concentration of H2 CO3 , mol/m3 [Hþ] Concentration of Hþ, mol/m3 3 [HCO ] Concentration of HCO3, mol/m3 S. Dara et al. / Journal of Cleaner Production 201 (2018) 974e987 AGE EOR MDEA MQL NRTL RHVT SRK SRU Acid Gas Enrichment Enhanced Oil Recovery Methyldiethanolamine Minimum Quality Lubrication Non Random Two Liquid RanqueeHilsch Vortex Tube Soave Redlich Kwong Sulfur Recovery Unit Greek Symbols a Parameter for estimating the cost of processed natural gas b Parameter for estimating the cost of processed natural gas g Parameter for estimating the cost of processed natural gas DTAir Temperature change of air, oC DTmix Temperature change of mixture, oC DTN2 Temperature change of nitrogen, oC l Latent heat of vaporization, kJ/kg h Overall process efﬁciency for steam generation References Abbas, Z., Mezher, T., Abu-Zahra, M.R.M., 2013. Evaluation of CO2 puriﬁcation requirements and the selection of processes for impurities deep removal from the CO2 product stream. Energy Procedia 37, 2389e2396. Abu Dhabi National Oil Company, 27th Mar 2016. Personal Communication. Addington, L., Ness, C., 2009. An evaluation of general “rules of thumb” in amine sweetening unit design and operation. In: Proceedings of the GPA Europe Sour Gas Processing Conference, Sitges, Spain. Bryan Research and Engineering, Inc., Powerful Process Simulation Technology (ProMax) http://www.bre.com/(Accessed on 14th Jan 2017) Dara, S., Berrouk, A.S., 2017. Computer-based optimization of acid gas removal unit using modiﬁed CO2 absorption kinetic models. Int. J. Greenhouse Gas Control 59, 172e183. Eiamsa-ard, S., Promvonge, P., 2008. Review of RanqueeHilsch effects in vortex 987 tubes. Renew. Sustain. Energy Rev. 12, 1822e1842. Esen, M., Yuksel, T., 2013. Experimental evaluation of using various renewable energy sources for heating a greenhouse. Energy Build. 65, 340e351. Esen, H., Inalli, M., Esen, M., 2006. Technoeconomic appraisal of a ground source heat pump system for a heating season in eastern Turkey. Energy Convers. Manag. 47, 1281e1297. Esen, H., Inalli, M., Esen, M., 2007. A techno-economic comparison of groundcoupled and air-coupled heat pump system for space cooling. Build. Environ. 42, 1955e1965. Jozi c, S., Baji c, D., Celent, L., 2015. Application of compressed cold air cooling: achieving multiple performance characteristics in end milling process. J. Clean. Prod. 100, 325e332. Khanmamedov, T.K., 2013. Strategies for acid gas enrichment and tail gas treatment. In: Proceedings of AIChE Spring Meeting and Global Congress on Process Safety, Houston, Texas, USA. Kidnay, A.J., Parrish, W.R., 2006. Fundamentals of Natural Gas Processing. Taylor & Francis Group, Boca Raton. Kohl, A.L., Nielson, R.B., 1997. Gas Puriﬁcation, ﬁfth ed. Gulf Publishing Company, Houston. Lunsford, K., McIntyre, G., 1999. Decreasing contactor temperature could increase performance. In: Proceedings of the Seventy-eighth GPA Annual Convention. Nashville, Tennessee, USA, pp. 121e127. Pandey, M., 2005. Process optimization in gas sweetening unit-a case study. In: Proceedings of the International Petroleum Technology Conference, Doha, Qatar. Parks, L.E., Perry, D., Fedich, R., 2010. FLEXSORB ®SE a proven reliable acid gas enrichment solvent. In: Proceedings of the 2nd Annual Gas Processing Symposium, vol. 2, pp. 229e235. Amsterdam, Netherlands. Peters, M.S., Timmerhaus, K.D., West, R.E., 2003. Plant Design and Economics for Chemical Engineers, ﬁfth ed. McGraw-Hill higher education, New York. Saberi, A., Rahimi, A.R., Parsa, H., Ashraﬁjou, M., Rabiei, F., 2016. Improvement of surface grinding process performance of CK45 soft steel by minimum quantity lubrication (MQL) technique using compressed cold air jet from vortex tube. J. Clean. Prod. 131, 728e738. Soave, G., 1972. Equilibrium constants from a modiﬁed Redlich-Kwong equation of state. Chem. Eng. Sci. 27, 1197e1203. Tunkel, L., Universal Vortex Inc, 10th Aug, 2017. Personal Communication. Universal Vortex, Inc., Designer of industrial vortex tube, http://www.universalvortex.com/(Accessed on 16th Jul 2017) U.S. Energy Information Administration, https://www.eia.gov/environment/ emissions/co2_vol_mass.php (Accessed on 9th Aug 2017) Weiland, R.H., Khanmamedov, T.K., 2010. Acid gas enrichment ﬂow sheet selection. Oil Gas J.