Energy Conversion and Management

Size: px
Start display at page:

Download "Energy Conversion and Management"

Transcription

1 Energy Conversion and Management 50 (2009) Contents lists available at ScienceDirect Energy Conversion and Management journal homepage: Natural gas combined cycle power plant modified into an O 2 /CO 2 cycle for CO 2 capture J.-M. Amann a, M. Kanniche b, C. Bouallou a, * a Centre Énergétique et Procédés (CEP), Ecole Nationale Supérieure des Mines de Paris, 60, Boulevard Saint Michel, Paris, France b EDF, Research and Development Division, Fluid Mechanics, Energies and Environment, 6 quai Watier Chatou Cedex, France article info abstract Article history: Received 27 April 2007 Received in revised form 16 April 2008 Accepted 20 November 2008 Available online 13 January 2009 Keywords: O 2 /CO 2 CO 2 capture Aspen plus TM Air separation unit NGCC Amine scrubbing Reduction of greenhouse gas emissions is becoming essential for struggling against global warming. Priority has been given to sources where carbon dioxide (CO 2 ) emissions are the largest and the most concentrated. Power plants using fossil fuels offer a great opportunity of applying CO 2 recovery processes. The O 2 /CO 2 cycle is an interesting option since CO 2 concentration in the flue gas is highly increased. This cycle has been applied to a natural gas combined cycle (NGCC) using an advanced gas turbine (GE9H). The aim of this study is to assess by simulation the energy and environmental performances of this new type of power plant. The oxygen required is produced by an air separation unit (ASU) that can deliver oxygen with a purity ranging between 85 and 97 mol.%. A CO 2 recovery process based on a cryogenic separation of carbon dioxide from inert gases has been designed and assessed. The impact of CO 2 capture has been calculated with the Aspen plus TM software. With an O 2 purity of 90 mol.% and an 85% CO 2 recovery rate, the net electrical efficiency reaches 51.3% (based on the low heating value (LHV)). This corresponds to an efficiency loss of 8.1%-points in comparison with the base case. The quantity of avoided CO 2 is about 280 g kw 1 h 1. These results have been compared with a conventional amine scrubbing applied to a NGCC. With a lean CO 2 loading of 0.16 mol CO 2 /mol amine, this process leads to a net electrical efficiency of 49.1% (LHV). The conversion into an O 2 /CO 2 cycle seems to be more efficient than amine scrubbing but more difficult to implement because of the specific gas turbine. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction It is well known that carbon dioxide plays a dominating role in the greenhouse effect. The rise of CO 2 concentration in the atmosphere, whose origin is mainly anthropogenic, is mostly due to the combustion of hydrocarbons. Renewable energies (wind, biomass, solar thermal...), alone, will not be able to provide the world with the essential energy needs, at least in the medium term. Today, the electricity cost with renewable energies are much higher than with hydrocarbons [1], which results in the development of several research programs with respect to CO 2 capture. The activities focus on locations where CO 2 emissions are relatively high and concentrated, such as power plants, refineries, cement plants and steel industry. Power plants, except the nuclear ones, produce relatively high level of carbon dioxide. According to the US DOE (Department of Energy), around 40% of the anthropogenic CO 2 emissions in the United States are due to the combustion of fossil fuels for electricity production [2]. The increasing use of coal to the detriment of natural gas will contribute to raise this percentage. Improving * Corresponding author. Tel.: ; fax: address: Chakib.Bouallou@ensmp.fr (C. Bouallou). thermal efficiencies can to a certain extent contribute to reduce CO 2 emissions. However, it is essential to develop, in the same time, processes allowing to reduce carbon dioxide emissions at low cost. Post-combustion capture thanks to an aqueous solution of amines is one of the most studied solution [3 7]. This type of solvent takes advantage of chemical absorption by amines (e.g. Monoethanolamine (MEA), N-methyldiethanolamine (MDEA),...) which allows low CO 2 partial pressure in the flue gas. However, this process is still too expensive. Recent projects for CO 2 capture in typical pulverized coal-fired power plants have reported costs as high as 60$ per avoided ton of CO 2 [8 10]. For a natural gas combined cycle (NGCC), some cost discrepancies exist, from 55 to almost 120 $ per avoided ton of CO 2 [8,11]. Therefore, other ways of CO 2 recovery should not be neglected. Particularly, cryogenic processes can be used for flue gas with high concentration of carbon dioxide. Such processes can be applied to oxy-fuel combustion which leads to high concentration of CO 2 in the flue gas thanks to the absence of inert gases like nitrogen. The cost per ton of CO 2 recovered would reach 25 $/ton [12]. For the oxy-fuel cycles, there are several concepts where CO 2 is used as the working fluid, including the O 2 /CO 2 cycle [6,7,13 15], the MATIANT cycle [16,17], the COOPERATE cycle [18,19] and the /$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi: /j.enconman

2 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) COOLENERG cycle [20]. Zhang and Lior [21,22] proposed systems integrated with liquid natural gas cold exergy utilization. There are other concepts using H 2 O as the working fluid including the Water cycle [23,24] developed by clean energy systems (CES), the Graz cycle [25], and others [26]. The oxy-fuel cycles requires huge quantities of oxygen. To reduce the penalty due to oxygen production, new technologies have been developed, such as chemical-looping combustion [27 29] and the AZEP concept [30,31]. Kvamsdal et al. [6,7] made a quantitative comparison of various oxy-fuel combustion cycles with respect to plant efficiency and CO 2 emissions. Their results show that the adoption of these new technologies promises improved performance because they require no additional energy for oxygen production, but they are still under development. Captured CO 2 can be used for other applications such as enhanced oil recovery (EOR) and enhanced gas recovery (EGR) or can be stored definitively in underground reservoirs. In term of oil and gas productivity, ya Nsakala et al. [32] recall that CO 2 is two to three times more effective than steam for EOR and EGR. The objective of this study is to evaluate the technical performances of a NGCC converted for oxy-fuel combustion using the O 2 /CO 2 cycle. Simulations have been carried out using the process software Aspen Plus TM. The studies made on this cycle found in the literature give only the basic trends of this new type of power plant. In most cases, the power plant or the recovery process are very simplified. To take into account all the characteristics of a power plant, the flowsheet in this study is based on EDF (a French electricity company) models representing a NGCC power plant using the GE 9H gas turbine. The oxygen required for the fuel combustion is supplied by an air separation unit (ASU). A particular attention has been paid to this unit because it implies a high energy penalty. A high ratio of the flue gas is sent back to the combustion chamber of the gas turbine to control the flame temperature. Therefore, a small part of the flue gas has to be treated for CO 2 recovery, which allows a reduction of the process size. The high CO 2 content in the flue gas allows a capture by a cryogenic process. A sensitivity study has been made on the oxygen purity and the CO 2 recovery rate to assess their impact on the net power plant output. This new type of power plant has been compared with a post-combustion capture by an aqueous solution of MEA. Table 1 Natural gas composition. CH 4 (mol.%) C 2 H 6 (mol.%) C 3 H 8 (mol.%) C 4 H 10 (mol.%) N 2 (mol.%) Description of the different concepts The different models were developed within the Aspen Plus TM process software based on energy and mass balances [33,34]. The relative error tolerance for convergence was set at 0.01% Description of the base case The EDF model has been built from an existing power plant: the integrated gasification combined cycle (IGCC) of Puertollano (Spain) with a conventional gas turbine: the Siemens/KWU V94.3 turbine. Simulations of the power plant have been validated with constructor data for both natural gas and synthesis gas. This NGCC model has been modified to work with the advanced GE9H gas turbine, using steam for turbine blades cooling rather than air used in conventional gas turbines. This technology allows an inlet temperature of 1700 K. The net electrical efficiency is about 59.5% (based on the low heating value (LHV)). The CO 2 emissions are about 339 g kw 1 h 1. Among the changes, the gas turbine allows a higher compression ratio and a higher turbine inlet temperature. The steam cycle has also been modified. The locations of the super-heaters of the high pressure steam and the intermediate pressure steam have been modified in the heat recovery steam generator (HRSG) to take advantage of the higher temperature of the flue gas. The pressure and reheat temperature levels have been increased. The natural gas is initially at MPa and K. Its composition is given in Table 1. It is burned in the combustion chamber (CC) with compressed air. The outlet pressure of the air compressor is set to conserve the reduced flow Q R entering the gas turbine (GT) Q R ¼ Q pffiffiffi T P where Q (kg s 1 ) is the mass flow rate entering the gas turbine, P (Pa) stands for the inlet pressure of the turbine and T (K) represents the inlet temperature of the turbine. The flue gas remains at high temperature after being expanded in the turbine and is then used to feed the steam cycle in the heat recovery steam generator. The steam is raised to three pressure levels. The high pressure steam and the intermediate pressure steam are overheated. The low pressure steam is heated at a K and then mixed with the low pressure steam recovered at the exit of the intermediate pressure turbine. At the low pressure turbine exit, the condensed steam is pumped and heated before being recycled to the feed-water tank. The water is close to its boiling point at MPa. The characteristics of the combined cycle are summed up in Table Description of the O 2 /CO 2 cycle The O 2 /CO 2 cycle has been built from the base case. A sketch of the converted cycle is shown in Fig. 1. The power plant can be Table 2 Combined cycle characteristics. Base case O 2 /CO 2 cycle Compressor outlet pressure MPa Turbine inlet temperature K Turbine outlet temperature K (O 2 purity = 85 mol.%) 1126 (O 2 purity = 97 mol.%) Turbine outlet pressure MPa HP steam pressure MPa IP steam pressure MPa LP steam pressure MPa Condensation pressure Pa HP steam overheating temperature K IP steam overheating temperature K

3 512 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Fig. 1. Sketch of the O 2 /CO 2 power plant. decomposed in four main units: the air separation unit, the gas turbine, the heat recovery steam generator and the CO 2 recovery process. The characteristics of the cycle are summed up in Table 2. The RK-SOAVE equation of state [35] has been used for the air separation unit, the gas turbine, the heat exchangers (for the flue gas only) in the heat recovery steam generator and the CO 2 capture process. The STEAMNBS data system (pure water) has been used for the steam cycle. This thermodynamic model is based on the 1984 NBS/NRC steam table correlation for thermodynamic properties and on the International Association for Properties of Steam IAPS for the transport properties Air separation unit An air separation unit has been chosen to produce the high quantity of oxygen because, even if oxygen transport membranes are a more effective way of producing pure oxygen, they are still under development. It is not yet suited for large scale power generation. Some researches need to be made to scale them to industrial size [8]. The air separation unit recovers oxygen from air (ISO conditions). The oxygen flow, whose O 2 purity varies from 85 to 97 mol.%, is injected in the combustion chamber of the gas turbine with the natural gas and the recycled flue gas. The air flow entering the air separation unit is fixed to obtain a 5 mol.% excess of oxygen (wet base) in the outgoing flue gas of the combustion chamber. This ensures a complete combustion of the natural gas. The air separation unit developed within Aspen Plus TM (Fig. 2)is quite conventional [36,37], even if the configuration could somewhat differ from an air separation unit to another. It consists on a double distillation column which produces liquid O 2 at 0.15 MPa. This column has a high pressure zone which roughly separates oxygen from nitrogen and argon and a low pressure zone which finalizes the purification of the oxygen stream. Between the two zones of the column, a spray-condenser is used to condensate the gaseous stream in the high pressure part to supply heat to the low pressure part. This heat is used to evaporate nitrogen and argon dissolved in the liquid oxygen stream, these species being more volatile than oxygen. In the flowsheet, the column has been split in two blocks: a high pressure column (HP column) and a low pressure column (LP column). The air (1) supposed at ISO conditions ( MPa, K, 60% relative humidity) is compressed in a two-stage inter-cooled compressor (CPR1). Water, hydrocarbons and acid gases (3) are then removed by a filter. The air stream (4) is split in two streams. The stream (8) is sent to the first heat exchanger (ECH1) where it is cooled to its dew point by the N 2 stream (18) recovered at the top of the low pressure column. The second air stream (5) is compressed and cooled before entering a second heat exchanger (ECH2) where it is entirely liquefied (7). The cooling is supplied by the N 2 stream coming from ECH1 (19) and the O 2 stream (15) recovered at the bottom of the low pressure column. The two air streams (7 and 9) are sent to the high pressure column. The expansion of the high pressure stream provides some cooling duties to the system. A stream enriched with O 2 (10) is recovered at the bottom of the high pressure column and a stream rich in N 2 (11) at the top. These two liquid streams are sent to a third heat exchanger (ECH3) where they are cooled by the N 2 stream (17). They are then sent to the low pressure column to be purified. The head pressure of the low pressure column has been set to 0.15 MPa to avoid air infiltrations. A pump is used to raise the pressure of the liquid O 2 stream recovered at the bottom of the LP column. This allows compressing O 2 at low cost. This stream is then vaporized in the heat exchanger ECH2 (16). Temperatures of the N 2 stream (20) and the O 2 stream (16) have been fixed at K at the exit of ECH2. The air flow rate (5) is fitted to reach this condition. The temperature and pressure levels inside the air separation unit are given in Table 3 for an oxygen purity of 90 mol.%. The operation of the spray-condenser requires that the boiling temperature of the liquid oxygen stream in the low pressure column is lower than the dew temperature of the nitrogen stream at the top of the high pressure column. The pinch temperature has been fixed at 0.75 K. Since the pressure has been set at 0.15 MPa at the top of the low pressure column, it is necessary to have a sufficient pressure in the high pressure column to fulfill this condition. Fig. 3 represents the bubble curve of oxygen and nitrogen. For an oxygen pressure of 0.15 MPa, the nitrogen pressure must be equal to 0.53 MPa to fulfill the temperature approach of the spray-condenser. In practice, the streams are a blend of components. The oxygen stream boiling temperature increases with its concentration in oxygen. This means that the high pressure column pressure must be increased with oxygen purity. This pressure level is the main parameter influencing the process performance since it Fig. 2. Air separation unit.

4 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Table 3 Temperature and pressure levels inside the air separation unit for an oxygen purity of 90 mol.%. Stream Mass flow rate (kg s 1 ) Pressure (MPa) Temperature (K) Vapor fraction See caption of Fig. 2. Fig. 3. Nitrogen and oxygen vapor pressure curves [38]. acts on the outlet pressure of the air compressor and thus on its electrical consumption Combined cycle The compressor of the gas turbine is originally used to compress air before the combustion chamber. For the O 2 /CO 2 cycle, it is used to compress the recycled flue gas. To control the flame temperature, a high ratio of the flue gas must be recycled towards the combustion chamber via the gas turbine compressor. The combustion chamber should remain quite conventional if part of the flue gas is recycled to limit the flame temperature. The NOx formation is also highly reduced because of the absence of nitrogen in the combustion system [39]. For the combustor, we have used the Gibbs free energy minimization, assuming the thermodynamical equilibrium of the flue gas. The nominal outlet pressure of this compressor and the turbine inlet temperature are kept similar to the respective base case values. The pressure ratio of the gas turbine compressor is however adjusted around nominal value to conserve the reduced flow Q R entering the expander of the gas turbine. Due to the difference of the thermo-physical properties of this type of flue gas, the temperature level at the outlet of the compressor and the turbine will be different. For a given pressure ratio in the compressor, its outlet temperature will be lower for a stream rich in CO 2 than for air. On the contrary, the turbine outlet temperature will be higher for a stream rich in CO 2 than for air. However, the lifespan of such turbine will be almost the same as a conventional turbine since the revolution number of the turboshaft engines will be reduced for an operation with a gas rich in CO 2 [13]. Special care should be however given to corrosion problem which was not studied here. The turbine outlet temperature varies with the purity of the oxygen flow. It is as high as the O 2 stream is pure. This is due to a lower dilution of the flue gas with residual N 2 and Ar. The heat recovery steam generator was optimized according to the turbine outlet temperature of the base case. Therefore, in this O 2 /CO 2 cycle, the flue gas has been cooled before entering the boiler. The extra heat is used to warm up the recycled flow of CO 2 downstream of the gas turbine compressor. This allows reducing the fuel flow rate and then the CO 2 emissions. One must keep in mind that such gas turbines are not yet available at the present time. It will be necessary to develop new design adapted to a fluid rich in CO 2. The inlet temperature of the heat recovery steam generator is conserved and there is no major change in its configuration. The main characteristics of the heat recovery steam generator, which was built from constructor data, have been conserved (pressure and temperature levels, pressure drop in the heat exchangers and heat losses). The flue gas leaves the heat recovery steam generator at approximately 387 K and it is then cooled to 303 K to withdraw the bulk of water. A small part of the flue gas is sent to the CO 2 recovery process. The study has shown that the ratio of flue gas sent to the recovery process ranges from 8.5 to 10.2% when the O 2 purity decreases from 97 to 85 mol.%. The other part is recycled back to the gas turbine compressor. Concerning the heat recovery steam generator, the thermal transfers should be more important due to the nature of the flue gas. This unit will require accurate calculations on heat exchangers to be optimized CO 2 recovery process The aim of the O 2 /CO 2 cycle is to concentrate CO 2 in the flue gas. This leads to a decrease of the gas flow rate sent to the capture unit. Thus, the CO 2 recovery is easier and the process size can be reduced compared with post-combustion capture. Since the flue gas has a high concentration of CO 2, a cryogenic process is well suited to separate CO 2 from non-condensable gases. Wilkinson et al. [37,40] have proposed two CO 2 recovery processes based on the difference between the components boiling points. Our recovery process is based on the work of theses authors. The sketch of the process can be shown in Fig. 4. The flue gas is condensed at 303 K at the exit of the heat recovery steam generator to remove the bulk of water. Then the dry flue gas (1) is compressed to 3.5 MPa in a three-stage inter-cooled compressor (MCPR1). This pressure level is required by the water removal process using triethylene glycol (TEG) and it is also suitable for cryogenic separation. It is important to remove water to avoid ice formation in the process. This is also required for CO 2 transportation in pipeline where corrosion and hydrates formation can occur. The residual molar fraction of H 2 O is fixed at 20 ppm (3). This flow is then mixed with two recycled streams (14) and (21). This recycle is necessary to reach the specification on the CO 2 recovery rate and purify the final CO 2 stream. The resulting stream (4) enters a first heat exchanger (MECH1) where it is partly liquefied. From the gas liquid separator (F1), two streams are recovered: a liquid stream (19) enriched with CO 2 and a gas stream (6) which contains a non negligible quantity of CO 2 and some other components like argon, oxygen and nitrogen. This gas stream is cooled in a second heat exchanger (MECH2) to liquefy some extra CO 2. The recovery rate depends on the outlet temperature of the stream (7) since this

5 514 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Fig. 4. Cryogenic CO 2 recovery process. temperature is linked to the quantity of liquefied CO 2. The lower the temperature, the higher the CO 2 recovery. A minimal temperature of K has been fixed to prevent carbon dioxide from solidification. A temperature approach of 5 K between the cold inlet/outlet temperature and the hot outlet/inlet temperature has been chosen. To produce the cooling duties necessary to this heat exchanger, the streams (7) and (10) are respectively adiabatically expanded in the gas liquid separator (F2) and through the valve (V1). The pressure drops are calculated to fulfill the conditions on the temperature approach. A rich CO 2 liquid stream (9) and a gas stream mainly composed of non-condensable gases (N 2,O 2 and Ar) are recovered from the gas liquid separator (F2). The gas stream supplies some cooling duties to (MECH2) and then to (MECH1). This gas stream is still at high pressure and can be expanded to recover some extra power. But prior being expanded, it (16) is heated to K by using the heat released in the three-stage inter-cooled compressor (MCPR1). Then it is expanded in a two-stage turbine with intermediate reheat. The outlet pressure has been chosen slightly higher than atmospheric pressure. The liquid stream (9) recovered from the gas liquid separator (F2) contains some amount of CO 2 which must be recycled to reach the specification on the CO 2 recovery rate. Before that, it supplies some cooling duties to the two heat exchangers (MECH2) and (MECH1). For this latter, the temperature difference between the cold inlet streams (12, 15 and 22) and the hot outlet stream (5) is equal or greater than 5 K. The stream (13) is then compressed (CPR2) to be mixed with the flue gas. After the gas liquid separator (F1), the liquid stream (19) enriched with CO 2 is expanded through a valve (V2) to evaporate part of the dissolved gases. The pressure drop is calculated to reach a 99 mol.% purity in CO 2. The gas stream (20) recovered from the gas liquid separator (F3) is recycled back after recompression (CPR1). The liquid CO 2 stream (22) supplies some cooling duties to the flue gas in the heat exchanger (MECH1). The outlet stream (23) is then compressed to 15 MPa in a twostage inter-cooled compressor (MCPR2). The high pressure stream (24) is cooled to 303 K to be liquefied. Many assumptions have been made for the different operation units (Table 4). The Redlich Kwong Soave model has been used in this process. Aspen Plus TM provides the binary interaction parameters between CO 2 and N 2 determined from Knapp and Prausnitz [41]. The equilibrium curves obtained with these parameters have been compared with the equilibrium data of Weber et al. [42] and Yorizane et al. [43] (Fig. 5). The experimental data are well represented by the software for the different temperatures. Some discrepancies appear with the vapor data of Yorizane et al. for the Table 4 Simulation assumptions. Turbo-machinery mechanical efficiency % O 2 /CO 2 cycle Air separation unit Compressor isentropic efficiency % 87.0 Pressure loss in the heat exchangers MPa Pressure loss at the entry of the HP column MPa 0.03 Pressure loss in the HP column MPa Pressure loss in the HP column MPa Gas turbine Compressor isentropic efficiency % 88.8 Turbine isentropic efficiency % 90.1 Pressure loss in the combustion chamber % 2.0 Heat loss in the combustion chamber % 0.2 Heat recovery steam generator HP turbine isentropic efficiency % 89.1 IP turbine isentropic efficiency % 88.6 LP turbine isentropic efficiency % 91.3 Cryogenic CO 2 recovery process Compressor isentropic efficiency % 87.0 Turbine isentropic efficiency % 87.0 Amine scrubbing Pressure loss in the absorber MPa 0.01 Pressure loss in the stripper MPa 0.01 Intermediary CO 2 compression isentropic efficiency % 85.0 Final CO 2 compression isentropic efficiency % 87.0 Pressure /MPa RK-Soave (Aspen) K RK-Soave (Aspen) K RK-Soave (Aspen) K Weber et al. (1984) K Weber et al. (1984) K Weber et al. (1984) K Yorizane et al. (1985) K X(N 2 ), Y(N 2 ) /molar fraction Fig. 5. Vapor liquid equilibria of the CO 2 N 2 system.

6 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) highest values of the molar fraction of N 2. However, the recovery process works at lower pressure, the errors are thus limited. The binary interaction parameters given by Aspen Plus TM give satisfactory results and were used in the different simulations Description of the alkanolamine process The alkanolamine process is placed downstream of the conventional NGCC. The sketch of the process is represented in Fig. 6. The flue gas is first compressed to MPa before the absorber to offset the pressure drop in the column. It is then washed by an aqueous solvent containing 30 wt.% of MEA. The amine concentration has been limited to avoid problems of corrosion. A solvent rich in CO 2 is recovered at the bottom of the absorber. It is pumped before the stripper. When the pressure increases, the reboiler temperature at the bottom of the stripper also increases. The solvent pressure is thus limited by the temperature level of the low pressure steam required for the amine regeneration. The low pressure steam is extracted from the steam cycle before the low pressure turbine at 0.32 MPa. Considering 0.1 MPa pressure loss between the extraction point and the recovery process, the steam pressure near the stripper is 0.22 MPa. However, only the condensation heat of the low pressure steam can be used in the reboiler of the stripper. The steam is cooled until its dew point. This steam is then condensed to supply the heat required in the reboiler of the stripper. The outlet pressure of the solvent pump is chosen to have a pinch temperature of 5 K between the steam and the reboiler. Before entering the stripper, the rich solvent is preheated in the heat exchanger (ECH2) by the lean solvent recovered at the bottom of the stripper. A pinch temperature of 10 K has been selected for this heat exchanger. The lean solvent recovered at the bottom of the stripper has the same residual quantity of CO 2 as the solvent entering the absorber. The CO 2 gaseous stream recovered at the top of the stripper is compressed to 6.5 MPa in an inter-cooled compressor. It is then dehydrated by a triethylene glycol process. The residual quantity of water has been set to 20 ppm molar. The CO 2 stream is then compressed to 15 MPa and cooled to 313 K to be liquefied. The CO 2 purity in this stream is higher than 99.9 mol.%. Several lean CO 2 loadings in the solvent have been assessed. But only two have been retained in this study. The first one is equal to 0.25 mol CO 2 /mol MEA. This value lowers the energy consumption during the amine regeneration. This value is also the one found by Alie et al. [44]. The second one is more realistic and is equal to 0.16 mol CO 2 /mol MEA. The calculations are based on equilibrium data. The Aspen Plus TM software provides some inserts for electrolyte systems. Since the results are highly dependent on the equilibrium curves, a particular attention has been paid to select a rigorous insert. The EMEA PCO2 (kpa) CO 2 loading (mol CO 2 /mol amine) and MEA insert have been assessed for different temperatures and different amine concentrations. The equilibrium curves of the CO 2 H 2 O MEA system have been compared with the data of Jou et al. [45] and those of Austgen and Rochelle [46]. For instance, Fig. 7 displays the CO 2 solubility in an aqueous solution containing 30 mass.% MEA at 313 and 393 K. There is good agreement for the highest temperature whatever the CO 2 loading is. For 313 K, there are some discrepancies for the highest values of CO 2 loadings. Between the two inserts, the EMEA one seems to be the best. This insert matches very well the data of Austgen et al., corresponding to an amine concentration of approximately 15.5 mass.%. The EMEA insert uses the electrolyte non-random two liquid (NRTL) model, which is suited for electrolytic aqueous solutions. It has been used for the different simulations. Some assumptions have been made concerning the pressure loss in the absorber and the stripper (Table 3). 3. Results and discussion 393 K 313 K Jou et al. (1995) K EMEA insert K MEA insert K Jou et al. (1995) K EMEA insert K MEA insert K Fig. 7. Solubility of CO 2 in a 30% mass MEA aqueous solution at 313 K and 393 K. Concerning the O 2 /CO 2 cycle, a sensitivity study was carried out on the purity of the oxygen stream. Four oxygen concentrations were selected: 85, 90, 95 and 97 mol.%. The latter concentration was selected in agreement with the work of Wilkinson et al. [37] who reported that, over this purity, it is necessary to separate argon from oxygen. This increases the cost of oxygen production as well as the capital expenditure of the air separation unit. The impact of the CO 2 recovery rate was also studied. The recovery rate ranges from 75 to 93.4%. The results of the O 2 /CO 2 cycle are compared with those obtained for the amine process. Fig. 6. Alkanolamine CO 2 removal process.

7 516 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) The O 2 /CO 2 cycle Penalty for the air separation unit The air separation unit has a big impact on the net efficiency of the power plant. The air compressor is the main energy consuming component in this unit. The electrical consumption is thus highly dependent on the air flow rate entering the compressor and on the specified outlet pressure. The performances of the air separation unit have been analyzed for the different oxygen purities. The pressure in the high pressure column must be increased with oxygen purity. The high pressure column pressure must be equal to MPa for an O 2 purity of 85 mol.%. But it rises until MPa for a purity of 97 mol.%. The outlet pressure of the air compressor must be increased from to MPa to compensate for this phenomenon. It was assumed that the pressure drops along the process do not vary with the outlet pressure of the air compressor. The specific work of the air separation unit ranges from to kw h t 1 of the oxygen stream (Table 5). These values are in the range of usual electrical consumptions of this kind of air separation unit [5,13,47]. In Table 6, it can be observed that the specific consumption of the air separation unit is around 245 kw h t 1 O 2 for an O 2 purity of 95 mol.%. This is slightly lower than our results. The penalty of the air separation unit on the net efficiency has been determined from the converted power plants without CO 2 capture. The penalty corresponds to the difference between the net electrical efficiency of the base case and of the O 2 /CO 2 cycle. This penalty ranges between 5.4 and 6.3%-points depending on the oxygen purity. If we only consider the air separation unit consumption, use of high oxygen purity is not favorable to the cycle. However, this purity acts on the performances of the CO 2 recovery process and on the maximum CO 2 recovery rate Flue gas composition The flue gas composition is function of the oxygen purity. Indeed, at the heat recovery steam generator exit, CO 2 is all the more diluted since the O 2 purity is low. The flue gas composition after cooling at 303 K and water removal is given in Table 7. The nitrogen concentration decreases with O 2 purity whereas the concentration of argon is quite stable. In fact, with an O 2 purity lower than 97 mol.%, argon is not separated from oxygen. Concerning te CO 2 concentration in the flue gas, it ranges from 66.8 to 85.5 mol.%. Therefore, it will be easier to capture CO 2 when oxygen purity is high. Table 5 Air Separation Unit performance with O 2 purity. O 2 purity (mol.%) O 2 stream flow t h ASU power MW Specific consumption kwh t 1 O ASU penalty %-points Table 7 Flue gas composition at the entry of the cryogenic CO 2 recovery process. O 2 purity (mol.%) Flue gas flow (kmol s 1 ) H 2 O (mol.%) CO 2 (mol.%) N 2 (mol.%) O 2 (mol.%) Ar (mol.%) CO 2. recovery rate The simulations have shown that approximately 90% of the flue gas leaving the heat recovery steam generator must be redirected towards the gas turbine to control the turbine inlet temperature. In fact the combustion temperature is controlled in the model to be kept at classical level by adjusting the flue gas recirculation rate. Only a small part of flue gas have to be treated to recover CO 2. The recovery process has been optimized to lower its electrical consumption. In the following, we will use TCOLD1 as the outlet temperature of the hot stream (5) leaving the first cryogenic heat exchanger (MECH1) and TCOLD2 as the outlet temperature of the hot stream (7) leaving the second cryogenic heat exchanger (MECH2). TCOLD2 acts on the CO 2 recovery rate. The influence of TCOLD1 on the process performances has been assessed for different values of TCOLD2. Fig. 8 represents the specific electrical consumption of the process for an oxygen purity of 90 mol.%. The evolution of the specific consumption of the recovery process is not linear with TCOLD1. For different values of TCOLD2, the minimum electrical consumption can be found at nearly the same value of TCOLD1. This value has been determined for each oxygen purity. It ranges from to K when oxygen purity varies from 85 to 97 mol.%. The CO 2 recovery process has been optimized with these values. The performances of this process depend on the flue gas composition (Table 8). The flue gas compression at the inlet of the process depends only on the flue gas flow rate, and therefore on the O 2 purity. This compressor is the more consuming element in the process, Specific electrical consumption (kwh.t -1 CO 2 ) TCOLD1 (K) TCOLD2 = K TCOLD2 = K TCOLD2 = K Fig. 8. Specific electrical consumption of the cryogenic CO 2 recovery process. Table 6 Air separation unit consumption. Dillon et al. [13] Andersson and Maksinen [47] Liljedahl et al. [5] O 2 purity mol.% a O 2 flow t h ASU power MW Specific consumption kwh t 1 O a Purity in mass.%.

8 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Table 8 Performances of the cryogenic CO 2 recovery process. O 2 purity (mol.%) Recovery rate (%) Flue gas flow (kg s 1 ) Flue gas compression (MW) Recycle compression (MW) CO 2 compression (MW) Wastes expansion (MW) Recovery process power (MW) Specific consumption (kw h t 1 ) far away from the compression of the recycled streams and the final compression of the CO 2 stream. The wastes expansion power depends on the wastes flow rate and thus on the oxygen purity and on the CO 2 recovery rate. The recycle compression power decreases with oxygen purity since the CO 2 concentration in the flue gas increases with oxygen purity. Otherwise, it increases with the recovery rate since the recycled flow rate increases. The final compression of the CO 2 stream depends on the stream flow rate and on the inlet pressure. Thus, for a CO 2 recovery rate of 75%, the compression power decreases from 4.0 to 3.4 MW when oxygen purity ranges from 85 to 97 mol.%. The compressor inlet pressure is as low as the oxygen purity is small. In fact, the CO 2 stream purification requires a higher pressure drop through the valve (V2) to reach the specified purity when oxygen purity is low. The maximum achievable recovery rate depends on the O 2 purity. For a fixed recovery rate, the more the flue gas is diluted with non-condensable gases, the lower the minimum temperature in the process is. Since we have specified a minimum temperature to avoid carbon dioxide from solidification, the recovery rate is limited. It can be observed in Table 9 that, for an O 2 purity of 85 mol.%, the maximum recovery rate reaches 80.9% and increases to 93.4% for a purity of 97 mol.% Net output loss For each oxygen purity, the net electrical efficiency of the O 2 / CO 2 cycle has been calculated (Fig. 9). Without capture, the efficiency ranges from 53.3 to 54.0% (LHV), depending on the O 2 purity. The difference between the cases comes from the electrical consumption of the air separation unit. Table 9 Maximum achievable CO 2 recovery rate with the cryogenic process. O 2 purity (mol.%) Maximum CO 2 recovery rate (%) The net electrical efficiency is better for a low O 2 purity. For an oxygen purity of 85 and 97 mol.%, the difference in the efficiency of the cycle is about 0.4%-point, although the air separation unit penalty is about 0.9%-point higher for the case with the purest oxygen. The recovery process is thus more efficient with higher oxygen purity. But it is not sufficient to compensate for the penalty due to the air separation unit. However, the maximum achievable recovery rate is higher. The level of oxygen purity must be chosen in accordance with the targeted recovery rate and the power plant efficiency. For an 85% recovery rate and an oxygen purity of 90 mol.%, the net electrical efficiency reaches 51.3% (LHV). Fig. 10 shows the global efficiency loss in comparison with the base case without capture. The loss does not vary much with the recovery rate since the air separation unit and the flue gas compression at the entry of the recovery process are independent of the recovery rate. The air separation unit represents about the two third of this loss, between 67% and 76%. The whole loss ranges between 7.9% and 8.7%-points according to the oxygen purity and the recovery rate. For the two highest oxygen purities, the efficiency loss is quite similar. But for a purity of 90 mol.%, the efficiency loss is about 0.25%-point lower. The benefit of a lower oxygen purity decreases below this purity. Indeed for a purity of 85 mol.%, the efficiency increases only by 0.1%-point. The penalty of the recovery process and of the air separation unit evolve inversely with the oxygen purity. It seems that there is an optimal oxygen purity for which the global efficiency loss is minimum. However, this purity, which is lower than 85 mol.%, leads to a very low recovery rate. For an 85% recovery rate and an O 2 purity of 90 mol.%, the cycle efficiency falls about 8.2%-points compared with the base case. The air separation unit contributes to an efficiency loss of 5.8%-points and the recovery process to 2.4%-points. A higher CO 2 recovery rate can be reached by choosing a higher oxygen purity without degrading significantly the net electrical efficiency of the power plant Comparison with previous works Table 10 compares the results of this work with those of other authors [6,7,13 15]. Those authors have a very high CO 2 recovery rate since they did not consider a purification of the flue gas. Net electrical efficiency (%) O2 purity = 85 mol.% O2 purity = 90 mol.% O2 purity = 95 mol.% O2 purity = 97 mol.% CO 2 recovery rate (%) Fig. 9. Net electrical efficiency of the O 2 /CO 2 cycle. Fig. 10. Net electrical efficiency loss of the O 2 /CO 2 cycle comparing with the base case.

9 518 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Table 10 Performances of the O 2 /CO 2 cycle. This work Dillon et al. [13] Bolland and Mathieu [14] a Bolland and Undrum [15] Kvamsdal et al. [6,7] b Type of turbine 9H 9FA 9FA GE9351FA Net electrical efficiency of the base case % Heat input MW th Gross power output MW e Gross efficiency % Oxygen purity mol.% ASU power MW e ASU specific consumption kw h/t O c CO 2 recovery MW e Auxiliaries MW e Net electrical power output MW e Net electrical efficiency % (LHV) Net efficiency loss %-point ASU penalty %-point ( ) d CO 2 recovery process penalty %-point Recovery rate % CO 2 emissions g kw 1 h CO 2 captured g kw 1 h Avoided CO 2 gkw 1 h a b c d Results given for a flow of 100 kg s 1 in the gas turbine compressor. The penalties are determined with the net efficiency of the base case whereas the authors give the penalties regarding with the gross efficiency of the O 2 /CO 2 cycle. The number does not take into account the O 2 compression from MPa until 3.5 MPa. The first number correspond to the ASU delivering gaseous O 2 at MPa and the second number correspond to the O 2 compression. When comparing the results of the air separation unit, it can be concluded that the specific consumption of our air separation unit is lower. However, in our study, the O 2 purity has been chosen at 90 mol.% to achieve a 85% recovery rate. Dillon et al. [13] have practically the same specific consumption as us. But, the other authors reported higher consumption because their air separation units produce gaseous oxygen at low pressure which is compressed at the outlet of the air separation unit. We have chosen to produce liquid oxygen because compression is less energy demanding when performed at liquid state by a pump than at gas state by mean of a compressor. The air separation unit leads to an efficiency loss of about 5.8%-points in our simulation, 6.3%- points for Dillon et al., 7.3%-points for Kvamsdal et al. and over 7.9%-points for the other authors. Concerning the CO 2 recovery process, the penalty induced by our process (2.4%-points) is of the same order of magnitude as those of Bolland and Mathieu [14], Bolland and Undrum [15] and Kvamsdal et al. [6,7]. Their recovery process penalty is slightly lower since there is no real flue gas purification. The flue gas is just compressed to fulfill the transportation conditions. The energy consumption of the recovery process of Dillon et al. [13] is significantly higher. It may be partly due to the compression of the flue gas after the heat recovery steam generator. We have chosen an inter-cooled compression to lower the power consumption but these authors may have chosen an adiabatic compression resulting in higher energy consumption. Our net efficiency loss is lower compared with other authors. This is true even with higher oxygen purity and higher CO 2 recovery rate. But their cycles can be optimized. Concerning Dillon et al. [13], the recovery process power is far too big. Concerning the other authors, their air separation units are not efficient since gaseous oxygen is produced at low pressure. The O 2 compression, which increases the efficiency loss, can be avoided by using an air separation unit with a pumped liquid oxygen system [36] The base case with amine scrubbing Specific consumptions in the CO 2 recovery process For the amine recovery process, the global power consumption is due to three elements: the flue gas compressor before the absorber, the steam extraction and the final compression of the CO 2 stream. The power consumption of the solvent pumps is not considered here since it is largely lower than the other power consumptions. A specific consumption has been determined for each of these elements. Knowing the flue gas flow characteristics entering the process and the recovery rate, the total power consumption can be calculated. The flue gas recovered at the heat recovery steam generator exit contains 4.97 mol.% of CO 2. It is compressed to MPa and requires 5.0 kwh per ton of flue gas. The heat duty in the stripper depends on the CO 2 concentration in the flue gas and on the lean CO 2 loading in the solvent. For a lean CO 2 loading in the solvent of 0.25 mol CO 2 /mol MEA, the thermal energy requirement reaches 3.56 MJ kg 1 of CO 2 recovered. But for a value of 0.16 mol CO 2 / mol MEA, which is more realistic, the thermal energy requirement increases to 5.44 MJ kg 1. Geuzebroek et al. [48] report a value of 4.3 MJ kg 1 CO 2, based on field data with a lean CO 2 loading of 0.16 mol CO 2 /mol MEA and 4.9 MJ kg 1 in their simulation with a lean CO 2 loading of 0.21 mol CO 2 /mol MEA. The field data report lower energy consumption than us. But Geuzebroek et al. report a higher value than the field data too although their lean CO 2 loading is higher. Our first value may be too optimistic since it assumes a high lean CO 2 loading. Our second value may overestimate the heat duty. The difference with the values reported by Geuzebroek et al. may come from different operating conditions. Moreover the thermal energy requirement at low lean CO 2 loading is highly dependent on the lean CO 2 loading. Finally, the specific consumption of the final CO 2 concentration has been evaluated at 86.1 kw h t 1 of CO 2 recovered for the highest lean CO 2 loading and at 89.7 kw h t 1 for a value of 0.16 mol CO 2 /mol MEA. The three specific consumptions do not vary much with the recovery rate Net output loss The results are displayed in Table 11 for the two lean CO 2 loading values. The efficiency loss ranges between 7.7% and 10.3%-points according to the lean CO 2 loading value. The steam extraction is the main cause of efficiency reduction, even with an optimistic value of the heat duty. It can be noticed that, for the lowest lean CO 2 loading value, more than 85% of the low pressure

10 J.-M. Amann et al. / Energy Conversion and Management 50 (2009) Table 11 Performances of the NGCC power plant with post-combustion capture. This work Davison [3] Kvamsdal et al. [6,7] Bolland and Undrum [15] Type of turbine 9H GE9FA GE9351FA 9FA Net efficiency of the base case % (LHV) Amine MEA (30 wt.%) MEA (30 wt.%) Fluor s Econamine FG + (MEA 30 wt.%) MEA (30 wt.%) Lean CO 2 loading mol CO 2 /mol amine CO 2 recovery rate % Specific power consumption for flue gas kw h t a compression Specific power consumption for CO 2 kwh t 1 CO b compression Specific heat duty in the stripper MJ kg 1 CO Auxiliaries MW e Flue gas compression MW e Low pressure steam extracted MW e CO 2 compression MW e Low pressure steam extracted % Net electrical power output % Net electrical efficiency % (LHV) Net efficiency loss %-point Efficiency loss due to flue gas compression %-point Efficiency loss due to steam extraction %-point Efficiency loss due to final CO 2 compression %-point CO 2 emissions g kw 1 h Avoided CO 2 gkw 1 h a Unit: MJ kg 1 CO 2. b Compression until 10 MPa. MEA steam is extracted, which will modify the isentropic efficiency of the low pressure turbine Comparison with previous works Davison [3], Kvamsdal et al. [6,7] and Bolland and Undrum [15] reported approximately the same efficiency loss, respectively 8.2%, 8.8% and 8.4%-points. This is higher than our best case but lower than our worst case. But our CO 2 recovery rate is lower. If the recovery rate is increased to 90%, our efficiency loss increases by 0.4%-point for the highest lean CO 2 loading and by 0.6%-point for the lowest one. However, the efficiency loss due to the low pressure steam extraction is higher in our case. In our simulations, the extracted steam is condensed and returned to the feed-water tank. But before entering this tank, it is preheated to reach the specified tank temperature. Part of the low pressure steam is extracted from the steam cycle at K to supply the heat required for this operation. The ratio between the power reduction due to steam extraction and the heat duty required in the reboiler reaches in our simulations. This is higher than the value reported by Bolland and Undrum, who reported a value between 0.21 and 0.22 for a low pressure steam extracted at the same pressure. The bigger efficiency loss found in our simulation may come from the extra steam consumption for preheating the condensed steam before the feed-water tank Outcome Table 12 compares the amine scrubbing and the O 2 /CO 2 cycle. It appears that converting the NGCC into an O 2 /CO 2 cycle give practically the same results than amine scrubbing for the highest lean CO 2 loading. But, in practice, the lean CO 2 loading seems not to be lower and a more realistic value results in a higher efficiency loss than with the O 2 /CO 2 cycle. On the other hand, there is no available gas turbine working with flue gas highly rich in CO 2. Moreover, the thermal energy requirement in the alkanolamine acid gas removal process can be lowered by modifying the configuration of the process and using advanced solvents. Table 12 Net electrical efficiency (LHV) for amine scrubbing and for the O 2 /CO 2 cycle. Lean CO 2 loading (mol CO 2 /mol MEA) Amine scrubbing O 2 /CO 2 It can be noticed that the O 2 /CO 2 cycle can not reach a recovery rate of 95% even with an oxygen purity of 97 mol.%. The CO 2 recovery process must be modified to recover more CO 2. With amine scrubbing, the maximum achievable CO 2 recovery rate depends on the type of solvent. Indeed, the recovery rate is directly linked to the thermodynamic of the system. 4. Conclusion CO 2 recovery rate (%) a b c a b c For this recovery rate, an O 2 purity of 85 mol.% is required. For this recovery rate, an O 2 purity of 90 mol.% is required. This efficiency is given for a recovery rate of 93.4% and an O 2 purity of 97 mol.%. With the aim of evaluating the performance of a NGCC converted into an O 2 /CO 2 cycle, the efficiency of the conversion was evaluated using the Aspen plus TM process software. The power plant flowsheet was built from an existing power plant. The O 2 /CO 2 cycle efficiency was compared with a conventional post-combustion capture based on chemical absorption. The chemical solvent is an aqueous solution with 30 wt.% of MEA and a lean CO 2 loading value of 0.16 and 0.25 mol CO 2 /mol MEA. For the O 2 /CO 2 cycle, an air separation unit producing liquid oxygen was chosen to minimize the power consumption. This allows the reduction of the compression power of the oxygen stream. The specific consumption of the air separation unit ranges from to kw h t 1 O 2 depending on oxygen purity. A sensitivity study on the O 2 purity showed that a purity of 85 mol.% was the best trade-off between the air separation unit penalty and the recovery process penalty but the CO 2 recovery rate