A COMPARATIVE ANALYSIS OF TWO COMPETING MID-SIZE OXY-FUEL COMBUSTION CYCLES T U

Proeedings of ASME Turbo Expo 2012: Power for Land, Sea and Air GT2012 June 11-15, 2012, Copenhagen, Denmark GT2012-69676 A COMPARATIVE ANALYSIS OF TWO COMPETING MID-SIZE OXY-FUEL COMBUSTION CYCLES Egill Thorbergsson, Tomas Grönstedt Chalmers University of Tehnology Gothenburg 412 96, Sweden Majed Sammak, Magnus Genrup Lund University Lund 221 00, Sweden ABSTRACT Coneptual turbine and ompressor designs have been established for the semi-losed oxy-fuel ombustion ombined yle and the Graz yle. Real gas effets are addressed by extending yle and oneptual design tools with a fluid thermodynami and transport property database. Maximum ompressor effiienies are established by determining optimal values for stage loading, degree of reation and number of ompressor stages. Turbine designs are established based on estimates on ahievable blade root stress levels and state of the art design parameters. The work indiates that a twin shaft geared ompressor is needed to keep stage numbers to a feasible level. The Graz yle is expeted to be able to deliver around 3% net effiieny benefit over the semi-losed oxy-fuel ombustion ombined yle at the expense of a more omplex realization of the yle. NOMENCLATURE A Area. Chord. C m Meridional veloity. H Stagnation enthalpy. M Mah number. ṁ Mass flow. N Rotational speed. R Gas onstant. p Stagnation pressure. PR Pressure ratio. s Entropy. Address all orrespondene to this author: egill@halmers.se. T U Stagnation temperature. Blade veloity. Greek letters Tip learane. str Cooling effetiveness for stator. rtr Cooling effetiveness for rotor. η p Polytropi effiieny. η s Isentropi effiieny. γ Ratio of onstant speifi heats. Λ p Reation degree based on pressure. φ Flow oeffiient. ψ = H Stage loading. U 2 INTRODUCTION There is a need for a bridging tehnology between the urrent energy system and the sustainable energy system of the future.a large number of tehnologies are urrently being researhed to ome up with safe, ost effiient and environmentally friendly arbon apture and storage solutions. One approah that shows promise is gas turbine based oxy-fuel ombustion yles. The oxy-fuel ombustion fires fuel with pure oxygen instead of air, and the resulting ombustion produts are primarily steam and arbon dioxide. This makes it tehnially more feasible to implement CO 2 apturing solutions. Two promising implementations of the oxy-fuel ombustion onept are the Semi-Closed Oxy-Fuel Combustion Combined Cyle (SCOC-CC) and the Graz Cyle. In the past, a number of studies on the thermodynami yles and oneptual design 1 Copyright 2012 by ASME

of the turbomahinery have been published. Ulizar presented a study about the SCOC-CC in 1996 [1]. The basi working priniple for the Graz yle was developed by Jeriha in 1985 [2]. Sine then the Graz yle has reeived a onsiderable amount of researh from Graz university as well as other universities, both in regards to yle analyses and oneptual turbomahinery design [3 7]. There have also been studies that ompare the yles and the oneptual design of the turbomahinery [8, 9]. This work has, however, mostly addressed the high end of the power output range. In this omparative study we will analyse two yles targeting a mid-sized version, net output of around 100 MW. The reason to study this power range is that in order to battle limate hange and fulfill future regulations we need to have arbon apture and storage option available in this range [10]. In the SCOC-CC the main fluid is arbon dioxide. The ratio of speifi heats and the gas onstant for arbon dioxide are lower than what they are in air. The speed of sound in the working media will be lower than in onventional gas turbines. In the Graz yle the working fluid is steam and arbon dioxide. For this mixture the ratio of speifi heats is slightly lower but the value of the gas onstant is higher than for air. This will inrease the speed of sound ompared to the air based proess. These differenes will have a profound impat on the design of the gas turbine. The most striking differene related to the working media is that the speifi work output is almost two-fold for the Graz yle. However, looking at the non-dimensional mass flow equation one an see that the impat from the higher gas onstant in the working gas of the Graz yle will result in a lower mass flow per unit area. The optimization study also shows that the effiieny is around 3% higher. These two benefits are ahieved at the expense of a onsiderably more omplex bottoming yle. The analysis presented in this paper ompares stage loading and geometrial features as well as other gas turbine parameters. The plant performane, net power and net effiieny are inluded in the omparison. This researh is intended to inrease the understanding of the turbomahinery design for oxy-fuel ombustion yles and the onfidene in the turbomahinery effiieny estimates used in the yle evaluation of the onepts. One of the problems assoiated with the Graz yle is the problem of ompressor design. Using single shaft ompressor results in quite short ompressor exit blades whih is diffiult to design for keeping a high turbomahinery effiieny and also results in that the ompressor will have large numbers of stages. It is therefore suggested to use a twin-spool geared onfiguration for a Graz yle with an output in the 100 MW range. METHOD Thermodynami analysis The thermodynami analysis and proess simulations was performed using the ommerial software IPSEpro that is developed by SimTeh Simulation Tehnology [11]. The software in- Endwall loss parameter 0.14 0.12 0.1 8 10 2 6 10 2 4 10 2 = 0.10 = 0.07 = 0.04 = 0.02 = 0.00 2 10 2 0.2 0.3 0.4 0.5 0.6 0.7 Diffusion fator FIGURE 1. CIENT [14]. CORRELATION FOR ENDWALL LOSS COEFFI- ludes the standard IAPWS-IF97 formulation for pure water and steam [11]. Gas mixtures in IPSEpro are, by default, handled assuming ideal gas behaviour. This neessitated that the software was interfaed with the NIST Referene Fluid Thermodynami and Transport Properties Database (REFPROP) whih uses equations of state and models to alulate gas properties [12]. The ooling method used in the yle alulations is based on the work by Jordal [13]. Compressor design The ompressor mean-line design was performed by a Chalmers in-house ode. The ode uses empirial relations to estimate the losses that are generated. The profile and endwall loss models are based on the work by Wright and Miller [14]. The endwall losses are highly dependent on the diffusion fator. This an be seen in the orrelations for the endwall loss in Fig. 1. The figure shows the orrelations between the endwall loss parameter, the tip learane over hord ratio (/) and the diffusion fator. This dependeny will have a strong influene on the optimal ompressors established in this paper. The design ode estimates the shok losses using a model developed by Shwenk [15]. A more detailed disussion of the ompressor ode algorithms and loss model implementation is presented in [16]. The design ode uses a method devised by Koh [17] to estimate the stati pressure rise oeffiient and maximum stati pressure rise oeffiient for eah stage. The differene between these two oeffiients indiate how lose eah stage is to stall/surge. To handle real gas effets in a onsistent way with the IPSEpro yle tool, the original in-house ode has been extended with 2 Copyright 2012 by ASME

a similar REFPROP interfae. TABLE 1. COMPOSITION OF THE WORKING MEDIA (S STANDS FOR SCOC-CC). A note on polytropi effiienies Polytropi effiienies have to be handled with some are when real gas effets are present. The definition of the polytropi effiieny is given by Eq. 1 [18]. η p = dh is dh For an ideal gas, Eq. 1 an be integrated using the Gibbs equation to establish an expression relating the polytropi effiieny with the pressure ratio. When the gas properties depend on pressure the proess beomes somewhat more intriate. However a numerial integration an be set up that remains onsistent with Eq. 1 and reprodues the ompressor exit temperature and pressure. This umbersome approah an be avoided by the use of the empirial expression suggested by the Mallen-Saville model [19]. The Mallen-Saville model assumes that the polytropi path is defined using Eq. 2. T ds dt (1) = onstant (2) The polytropi effiieny an then be alulated using Eq. 3. Composition [%] R [ ] Ar CO 2 H 2 O N 2 O J 2 kg K Graz Comp. 1.55 20.9 77.4 0.03 0.13 400.2 Graz Turb. inlet 1.74 23.6 74.5 0.04 0.15 392.3 S. Comp. 4.06 92.0 0.98 2.86 0.12 195.5 S. Turb. inlet 3.82 86.5 6.91 2.68 0.11 211.5 S. Turb. exit 3.87 87.7 5.55 2.72 0.11 207.8 Optimization The ompressor performane was optimized by varying the stage loading and degree of reation onstraining the diffusion fators. Thus, the optimization spae is 2n-dimensional where n is the number of stages in the ompressor. The objetive funtion was defined by maximizing polytropi effiieny onstrained by ahieving the required pressure ratio. The optimization sheme used was a hybrid ombination of four optimization methods: a geneti algorithm, Neider and Mead downhill simples, sequential quadrati programming, and a linear solver. η p = (H e H i ) (s e s i )(T e T i ) ln(t e /T i ) (H e H i ) Aungier reports that the Mallen-Saville model yields exellent auray and is fairly easy to implement [20]. This is onfirmed by our analysis showing that the model gives less than a 1% error when ompared with the numerial integration. We therefore use the Mallen-Saville method in our alulations. Turbine design The turbine mean line design was arried out by the Lund University in-house turbine design tool LUAX-T. The LUAX-T ode is a redued-order through-flow tool apable of designing highly loaded, ooled turbines. LUAX-T uses the refined Ainely and Mathieson mean-line loss model. The model has been modified by Dunham and Came, Kaker and Okapuu, and finally by Moustapha and Kaker [21]. Loss model omprises profile losses, trailing edge losses, seondary losses and tip learane losses. For a more detailed disussion see [22]. (3) RESULTS Cyle analysis The main differene between the two oxy-fuel yles is that the working fluid in the SCOC-CC is sent to the ompressor after the water is separated from the flue gas, whereas in the Graz yle a large part of the flue gas is reirulated to the ompressor. This results in that the major omponent in the SCOC-CC is CO 2 while in the Graz the major omponent is H 2 O. The omposition of the working fluid is shown in Table 1. The fuel is ombusted with oxygen that has to be provided to the yle through an air separation unit. An eonomially feasible option to produe oxygen in large amounts is through a ryogeni air separation plant [23, 24]. Aording to Darde and Amann [25, 26] ASU energy onsumption inreases drastially when the purity of oxygen goes over 95%. The power onsumption neessary to obtain oxygen with 95% purity and a pressure of 2.38 bar has been assumed to be 900 kj/kg O 2. An additional 325 kj/kg O 2 was assumed for its ompression [27]. The prodution of 200 bar liquefied arbon dioxide that is needed to enable transport and storage was assumed to require another 350 kj/kg CO 2 [27]. The fuel for both yles is natural gas with a lower heating value of around 48 MJ/kg natural gas. 3 Copyright 2012 by ASME

CO 2 Flue gas ondenser Water Natural gas HRSG Feed water tank Reheater Oxygen Combustion Chamber Steam Compressor Turbine Turbine Condenser FIGURE 2. PRINCIPLE FLOW SCHEME FOR SCOC-CC. SCOC-CC The major units in the SCOC-CC are the gas turbine, the heat reovery steam generator, the steam turbines, the flue ondenser, the air separation unit, and the CO 2 ompression train. A shemati of the yle is shown in Fig. 2. The topping yle is a Brayton yle with the working media onsisting mainly of CO 2. The reason for this omposition is that the water is ondensed from the flue gas before the major part of the gas, 93%, is reirulated to the ompressor. The remaining part of the CO 2 is ompressed and stored. The ooling media for the turbine is taken from the ompressor. The bottoming yle is a dual-pressure Rankine yle with steam as a working media. Bolland and Mathieu reported that there is only small differene between a dual- and triple pressure steam yles [23]. A more detailed desription of the yle an be found in [28]. The results from the yle analysis are shown in Table 2. Sammak [28] showed that the optimum pressure ratio for the SCOC-CC is 37. The inlet temperature for the ompressor is 20 C and the pressure is 1.013 bar. The exit temperature from the ompressor is 394 C and from the turbine it is 620 C. The gross power produed is the power output from the gas turbine and the steam turbines whih is 134 MW while the fuel input is 230 MW. This means that the gross effiieny of the yle is 58%. The major energy penalties ome from the air separation unit, power required to ompress the oxygen, and the power required to ompress the arbon dioxide. The prodution of oxygen redues the effiieny by 10%. The impliation of this high penalty is that further analysis of ASU optimization and integration with the yles must be arried out. The ompression of the arbon dioxide redues the gross effiieny with another 2%. The resulting net effiieny of the yle is 46% and the net power output is 106 MW. Graz yle The major omponents in the Graz yle are the gas turbine, the heat reovery steam generator, the steam turbines, the flue ondensers, the ompressors on the flue gas, the TABLE 2. POWER BALANCE FOR THE CYCLES. Unit SCOC-CC Graz yle Comp. mass flow kg/s 190 62 Comp. pressure ratio - 37 44.7 Combustor outlet temp. C 1400 1400 GT power, Turbine MW 155 169 GT power, Compressor MW 67 62 Gas turbine power MW 86 105 Total heat input MW 230 199 Steam turbine power MW 48 20 Pumps power MW 0.5 0.8 Compressors MW - 2.9 Gross power output MW 134 125 Gross effiieny % 58 63 O 2 prodution MW 23 20 CO 2 ompression MW 5 4.3 CO 2 mass flow kg/s 14 12 Net power output MW 106 96 Net effiieny % 46 49 Oxygen Fuel Compressor FIGURE 3. Steam Turbine Cooling Combustion Chamber Turbine HRSG Steam Turbine Water Condensers PRINCIPLE FLOW SCHEME FOR GRAZ CYCLE. air separation unit, and the CO 2 ompression train. A shemati of the yle is shown in Fig. 3. The topping yle is the same as in the SCOC-CC, a Brayton yle, but with the differene that the main omponent of the working media is steam. The reason for this is that the part of flue gas that is reirulated to the ompressor is not stripped Condenser Feed water tank CO 2 4 Copyright 2012 by ASME

from water, as is done in the SCOC-CC. Around half, or 51%, of the flue gas is reirulated to the ompressor. The inlet temperature is 125 C and the pressure is 1.06 bar. The exit temperature from the ompressor is 710 C, whih indiates a need for an interooler. The outlet temperature from the turbine is 572 C. This values are established by the requirement that there annot be ondensed water at the inlet to the ompressor. In ontrast to the SCOC-CC yle the Graz yle uses steam to ool the turbines [9]. Results of a parametri study using the pressure ratio is shown in Fig. 4. The limiting fator for the pressure ratio is the steam used for the ooling in the ombustion hamber, and in the gas turbine blades. The net yle effiieny at the pressure ratio limit is around 48.6%. the flue gas. The flue gas stream has two ompressors produing a flue gas that is above atmospheri onditions. The pressure after the steam generator is 0.73 bar, the temperature is 134 C and the mass flow is 32 kg/s. The turbine produes 13 MW by expanding the steam to a pressure of 0.025 bar and a temperature of 21 C. The results from the thermodynami simulation of the yle is shown in Table 2. The gross output from the yle is 125 MW and the energy input into the yle is 199 MW. This results in a gross effiieny of 63%. The major energy penalty is the oxygen prodution with an effiieny penalty of 10% and the CO 2 ompression takes further 2%. The resulting net effiieny of the yle is 49% aounting for auxiliary systems. The net power output from the yle is 96 MW. Net effiieny 0.485 0.48 0.475 0.47 34 36 38 40 42 44 Pressure ratio FIGURE 4. PARAMETRIC VARIATION IN THE PRESSURE RA- TIO FOR THE GRAZ CYCLE. The Graz yle has two bottoming yles. The first extrats the heat from the exhaust gases from the gas turbine while the seond uses the heat from the ondensation of the water in the flue gas. The first bottoming yle is a single pressure Rankine yle. The heat reovery steam generator (HRSG) onsists of an eonomizer, an evaporator and a superheater. The resulting steam that omes from the HRSG has a pressure of 140 bar, a temperature of 401 C and the mass flow is 33 kg/s. The steam is then expanded in a turbine, that produes 6.7 MW, to a temperature of 261 C and a pressure of 47.4 bar. The reason for the high pressure at the exit of the turbine is that the steam is used for ooling in the gas turbine. The seond bottoming yle is also a single pressure Rankine yle, but the pressure is sub-atmospheri. The steam generator also onsists of an eonomizer, an evaporator and a superheater but the heat now omes from the ondensation of water in Turbomahinery design The differene in the working fluid properties results in an interesting aspet for the sizes of the turbomahinery in the topping yles. The working fluids in the yles have similar speifi heat ratio while the gas onstant and the mean speifi heat are around twie as large for the working fluid in the Graz yle. Using the non-dimensional mass flow equation, Eq. 4, and assuming that the pressure, Mah number and ratio of speifi heats are same, it is revealed that the mass flow per area for the SCOC-CC yle is 1.4 the mass flow per area for the Graz yle. ṁ γrt Ap ( = γm 1 + γ 1 2 M2 ( ) ) 1 γ+1 2 γ 1 Sine the speifi heat ratios for the two yles are of the same order it means that a similar pressure ratio would require a similar hange in temperature for a given inlet temperature. However, sine the speifi heat is approximately twie as large for the Graz yle it means that the power requirement to ahieve a given pressure ratio is approximately twie of that needed for the SCOC-CC. This leads to the onlusion that similar ompressor and turbine stage numbers are expeted assuming similar Mah number levels. However, the power output for a given mass flow in the Graz yle is more than double due to the approximately double speifi heat. Theoretially, the ombined effet of a higher speifi heat but a lower mass flow per annulus flow area, ombine into an expeted 40% redution of annulus flow area for the Graz yle for a given gas turbine power output requirement. However, the optimization of the two yles leads to differenes in overall pressure ratios. In addition, differenes in ompressor inlet temperature lead to differenes in ompressor power requirement and turbine stress limitations lead to differenes in ompressor aerodynami onstraints. (4) 5 Copyright 2012 by ASME

Coneptual ompressor design Designing for a high stage loading dereases the number of stages needed to ahieve the required pressure ratio. This will in turn redue the size, weight and prodution ost of the ompressor. However, inreasing the stage loading will influene the ompressor effiieny and stall margin adversely. This is manifested in a high amount of diffusion generating thik boundary layers along the ompressor blade surfaes and whih may in the end result in flow separation. During the optimization proess the diffusion fator is onstrained to be below 0.6 in order to ensure the aerodynami stability of the ompressor [29]. Sine the loss models relate the diffusion fators to the effiieny this onstraint was not ative for the final design solution. An upper limit of 0.45 was set on the stage loading [29]. As will be disussed this onstraint is ative for some of the ompressors studied, depending on the number of stages assumed. The hub to tip radius ratio at the exit of the ompressors is onstrained to a maximum value of 0.92 as the end wall losses tend to beome very high at higher values [30]. The tip learane to height ratio of the blades has a typial range of 0.5% at the front stages 1.5% in the rear stages of a ompressor in a stationary gas turbines. However typial stationary mid-sized gas turbines have pressure ratio in the range of 20 to 30. The pressure ratio for the gas turbines for the SCOC- CC and the Graz yle are 37 and 44.7 respetively, so the rear stages of the ompressors will have onsiderably higher operating temperatures. This is likely to inrease the minimum amount of learane needed. In this study we assumed that the minimum learane possible to ahieve is 0.6 mm. The stati pressure rise omputed with Koh method showed reasonable numbers for all ompressors design, although the hord lengths hosen in this analysis may be on the optimisti side. SCOC-CC The inlet flow is designed with an axial Mah number of 0.6 and a relative tip Mah number of 1.35. These high Mah numbers ome from the onstraint that the hub to tip radius ratio at the inlet should be higher than 0.35, to allow suffiient spae for the disks. Suh aggressive values may be aounted for using a first stage blisk. Advaned design methods using pre-ompression s-shaped blade passage duts ould allow for even higher relative Mah numbers. However, the added design omplexity and the fat that the underlying loss models assumes profile families that are not feasible in suh Mah number ranges, makes suh solutions outside the sope of this paper. Three ompressors with 14, 15 and 16 stages respetively were optimized under the onstraints given above. The maximum effiieny was established by determining optimal values on the stage loading and degree of reation. The resulting effiienies are shown in Table 3. It an be seen that as the number of stages inreases the effiieny inreases. However the rate of im- Radius [m] IGV 0.5 Stage 1 ψ = 0.45 φ = 0.66 Stage 14 ψ = 0.40 φ = 0.52 0 0 1 2 3 Axial length [m] FIGURE 5. 14 STAGE SCOC COMPRESSOR (RED BLADES ARE ROTORS AND BLUE BLADES ARE STATORS). provement drops off quikly after 16 stages providing a marginal benefit as the number of stages are inreased. The annulus for the 14 stage ompressor is shown in Fig. 5. The results for the flow oeffiient and stage loading an be seen in Fig. 6. As seen in the figure the stage loading onstraint is ative for all stages apart from the two last stages in the 14-stage ompressor. The reason why the optimizer hooses to redue the load on these stages is that higher loading would result in a higher diffusion fator. Sine the tip learane to height ratio is highest at the bak end of the ompressor, 0.04, the penalty for inreasing the diffusion on these stages will be prohibitive. As the stage number is inreased to 16 all but three stages are unonstrained indiating that further inrease in stage numbers may only marginally inrease effiieny through a redued stage load. At the same time inreasing stage numbers further will inrease the wetted area and inrease losses. Stage loading 0.5 0.4 0.3 0.2 0.1 0 SCOC 14 SCOC 15 SCOC 14 0.5 0.55 0.6 0.65 Flow oeffiient FIGURE 6. FLOW COEFFICIENT VS. STAGE LOADING FOR SCOC-CC COMPRESSORS 6 Copyright 2012 by ASME

Stage loading 0.5 0.4 0.3 0.2 0.1 Graz LP Graz HP 0 0.3 0.35 0.4 0.45 0.5 0.55 0.6 Flow oeffiient FIGURE 7. FLOW COEFFICIENT VS. STAGE LOADING FOR GRAZ COMPRESSORS TABLE 3. EFFICIENCIES, PRESSURE RATIOS AND ROTA- TIONAL SPEEDS FOR THE COMPRESSORS. # stages η s η p PR N SCOC 14 86.9 89.4 37 7200 SCOC 15 87.5 89.9 37 7200 SCOC 16 87.9 90.3 37 7200 Graz - LP 6 87.0 88.8 6.8 9000 Graz - HP 6 84.1 85.7 6.6 27000 yle and the shaft power requirement therefore is moderate, it is feasible to design a gearbox to drive the ompressors. The gearbox an be introdued either into the ompressor driving a twin shaft ompressor onfiguration or simply to drive a single shaft ompressor. A single shaft ompressor was analysed. The resulting number of stages were at least 22 to ahieve the pressure ratio, but to ahieve a maximum effiieny around 25 stages are estimated. This is still not a very ompat ompression system. Here we suggest to use a twin shaft ompressor. The relative tip Mah number at the first rotor of the low pressure and high pressure ompressors are 1.16 and 1.1 respetively. The low pressure ompressor will revolve with 9000 rpm while the high pressure ompressor will rotate with 27 000 rpm. Both ompressors have 6 stages and the polytropi effiieny for the low pressure ompressor is 88.8% and for the high pressure ompressor the effiieny is 85.7%. Further optimization should enhane the polytropi effiieny up to 90%, but ould inrease the number of stages. The effiienies and the pressure ratio for the ompressors are given in Table 3. The flow oeffiient and the stage loading for the ompressors are shown in Fig. 7. The resulting design has not been optimized and the yle results for this onfiguration are based on estimates of ahievable effiienies. Coneptual turbine design SCOC-CC The SCOC-CC turbine has been studied by Sammak [22]. The resulting turbine has a rotational speed of 7200 rpm and the power turbine has rotational speed of 3000 rpm. The ompressor turbine has two stages and the power turbine has three stages. The root stress at the exit of the ompressor turbine 38 10 6 m 2 rpm 2 whih is within design pratie. The metal temperature was set to 950 C and the ooling mass flow was 44 kg/s. The metal temperature is high in omparison to urrent stationary gas turbines. The oxy-fuel turbine is still under development and it is assumed that single rystal alloys will be used in the turbine. Current researh has shown that single rystal alloys an handle up to 1000 C [31, 32]. Graz yle A first estimate on the number of stages needed for the Graz ompressor an be obtained by means of a simplified analysis. From a hub to tip radius ratio of 0.92 at the exit and the area needed to pass the mass flow the exit mid radius an be estimated. If a onstant mean radius design is assumed running at a stage loading of 0.45 the number of stages needed is around 36. Note that the rotational speed is fixed from the maximum root stress ondition, AN 2, of the last stage turbine rotor. This number of ompressor stages is viewed as infeasible both in terms of ost and and in terms of ompressor starting and operation. The only way to irumvent this onstraint is to introdue a geared ompressor. Sine we are studying a mid-sized Graz yle The twin-shaft gas turbine onsists of a gas generator and a free-power turbine. The seletion of the rotational speed is onstrained with the blade root stress whih is limited to 55 10 6 m 2 rpm 2. The gas generator rotation speed is set to 9000 rpm. The free-power turbine rotation speed is set to 6000 rpm. Thus the free power turbine is onneted to the generator through a redution gearbox. The design of the turbine takes into aount the last stage exit Mah number and stage loading. The oneptual design of the Graz turbine results in a high-pressure turbine with three stages and a power turbine with four stages. The inlet Mah number is 0.12 while the last stage exit Mah number is alulated to be 0.48. The exit stage loading is 1.2 and exit swirl angle is -5. These values are within 7 Copyright 2012 by ASME

Radius [m] 0.8 0.6 Stage1 ψ = 1.3 φ = 0.45 Stage 7 ψ = 1.2 φ = 0.53 0.4 HP PT N = 9000 rpm N = 6000 rpm 0.2 0 0.5 1 1.5 Axial length [m] FIGURE 8. GRAZ TURBINE (RED BLADES ARE ROTORS AND BLUE BLADES ARE STATORS). TABLE 4. KEY TURBINE PARAMETERS FOR GRAZ CYCLE HIGH PRESSURE TURBINE (HP) AND POWER TURBINE (PT). Turbine Stage ψ = H U 2 HP PT φ = C m U PR Λ p str rtr 1 1.3 0.45 1.50 0.30 0.40 0.29 2 1.2 0.45 1.50 0.38 0.26 0.18 3 1.2 0.45 1.60 0.34 0.10 4 1.9 0.5 1.65 0.40-5 1.9 0.46 1.85 0.40-6 1.7 0.47 2.0 0.40-7 1.2 0.53 1.8 0.40 - aeptable limits and suitable for a hot end drive diffuser. The ombustion outlet temperature is set to 1400 C and the metal temperature is limited to 950 C. The ooling mass flow is determined to be 18 kg/s and the free power turbine is unooled. The annulus area of the Graz turbine is presented in Fig. 8. The design parameters for the turbine is presented in Table 4. The free power turbine fourth, fifth and sixth stages are heavily loaded, and thus attempting to reduing the free power turbine into a three stage onfiguration would result in a very high last stage Mah number and stage loading. DISCUSSION AND CONCLUSION This paper has presented a omparison analysis of two oxyfuel ombustion yles. The yle simulation analysis showed that the Graz yle has higher effiieny, 49%, ompared to 46% for the SCOC-CC. The higher effiieny of the Graz yle is though weighed down by the design of the yle. The Graz yle has two ompressors for the CO 2 stream and two bottoming yles, whereas the SCOC-CC has only one bottoming yle. This makes the Graz yle more omplex. The turbomahinery was designed for both yles. The design of the turbine turbine that drives the ompressor in the Graz yle has one more stage than the SCOC-CC turbine whih was designed with two stages. The designs for the power turbines for the yles showed that the turbine for the Graz yle also needed an extra stage ompared to the three stages in the SCOC-CC turbine. The turbines designs for both yles an be designed well within industrial design pratie ranges. The oneptual design of the ompressors resulted in ompressor for the SCOC-CC that has 14 stages. To design a ompressor for the Graz yle whih has a gas path that maintains a high blade veloity throughout the ompressor a geared onfiguration is expeted to be neessary whih is quite feasible in this power range. Both a twin-shaft ompressor and a single shaft geared onfiguration are feasible but the twin-shaft onfiguration is onsiderably more ompat and has a number of stages whih is omparable with the optimal SCOC-CC design. ACKNOWLEDGMENT This researh has been funded by the Swedish Energy Ageny, Siemens Industrial Turbomehinery AB, Volvo Aero Corporation and the Royal Institute of Tehnology through the Swedish researh program TURBOPOWER. The support of whih is gratefully aknowledged. REFERENCES [1] Ulizar, I., and Pilidis, P., 1997. A semilosed-yle gas turbine with arbon dioxide argon as working fluid. Journal of engineering for gas turbines and power, 119, p. 612. [2] Jeriha, H., Sanz, W., and Göttlih, E., 2008. Design onept for large output Graz yle gas turbines. Journal of Engineering for Gas Turbines and Power, 130, p. 011701. [3] Jeriha, H., Sanz, W., and Göttlih, E., 2007. Gas Turbine With CO 2 Retention 400 MW Oxy-Fuel System Graz Cyle. CIMAC Paper. [4] Jeriha, H., Sanz, W., Göttlih, E., and Neumayer, F. Design details of a 600 MW Graz yle thermal power plant for CO 2 apture. ASME Turbo Expo 2008: Power for Land, Sea and Air, 9-13 June. [5] Bolland, O., Kvamsdal, H., and Boden, J., 2001. A Thermodynami Comparison of the Oxy-Fule Power Cyles, Water-yle, Graz-Cyle and Matiant-Cyle. In Power generation and sustainable development. International onferene, pp. 293 298. 8 Copyright 2012 by ASME

[6] Celano, A., 2002. Comparison of near Zero CO 2 Emission Power Plants on CO 2 /H 2 O Mixture. PhD thesis, University of Padua. [7] Frano, F., Mina, T., Woolatt, G., Rost, M., and Bolland, O., 2006. Charateristis of Cyle Components for CO 2 Capture. In Proeedings of 8th International Conferene on Greenhouse Gas Control Tehnologies, Trondheim, Norway. [8] Woollatt, G., and Frano, F., 2009. Natural gas oxy-fuel yles Part 1: Coneptual aerodynami design of turbomahinery omponents. Energy Proedia, 1(1), pp. 573 580. [9] Sanz, W., Jeriha, H., Bauer, B., and Göttlih, E., 2008. Qualitative and Quantitative Comparison of Two Promising Oxy-Fuel Power Cyles for CO Capture. Journal of Engineering for Gas Turbines and Power, 130, p. 031702. [10] Hendriks, C., de Visser, E., Jansen, D., Carbo, M., Ruijg, G., and Davison, J., 2009. Capture of CO 2 from mediumsale emission soures. Energy Proedia, 1(1), pp. 1497 1504. [11] Teh, S., 2003. IPSEpro proess simulator. User Doumentation: Program Modules and Model Libraries, SimTeh. [12] Lemmon, E., MLinden, M., and Huber, M., 2007. REF- PROP: referene fluid thermodynami and transport properties. NIST Standard Referene Database, 23. [13] Jordal, K., 2001. Modeling and performane of gas turbine yles with various means of blade ooling. [14] Wright, P., and Miller, D., 1992. An improved ompressor performane predition model. [15] Shwenk, F., Lewis, G., and Hartmann, M., 1957. A preliminary analysis of the magnitude of shok losses in transoni ompressors. NASA RM E57A30. [16] Xu, L., 2011. Analysis and Evaluation of Innovative Aero Engine Core Conepts. PhD thesis, Chalmers University of Tehnology. [17] Koh, C., 1981. Stalling pressure rise apability of axial flow ompressor stages. Journal of Engineering for Power, 103(4), pp. 645 656. [18] Dixon, S. L., and Hall, C. A., 2010. Fluid Mehanis and Thermodynamis of Turbomahinery. Butterworth- Heinemann. [19] Mallen, M., and Saville, G., 1977. Polytropi Proesses in the Performane Predition of Centrifugal Compressors. Paper No. CI83/77, Institution of Mehanial Engineers, London, United Kingdom, pp. 89 96. [20] Aungier, R., 2003. Axial-flow ompressors: a strategy for aerodynami design and analysis. ASME PRESS. [21] Moustapha, H., Zelesky, M., Baines, N., and Japikse, D., 2003. Axial and radial turbines. Conepts NREC. [22] Sammak, M., Genrup, M., Thorbergsson, E., and Grönstedt, T., 2012. Single and twin-shaft oxy-fuel gas turbine design in a mid-size semi-losed oxy-fuel ombustion ombined yle. ASME Turbo Expo 2012: Power for Land, Sea and Air, 11-15 June. [23] Bolland, O., and Mathieu, P., 1998. Comparison of two CO 2 removal options in ombined yle power plants. Energy Conversion and Management, 39(16), pp. 1653 1663. [24] Allam, R., White, V., Ivens, N., and Simmonds, M., 2005. The oxyfuel baseline: revamping heaters and boilers to oxyfiring by ryogeni air separation and flue gas reyle. Carbon Dioxide Capture for Storage in Deep Geologi Formations-Results from the CO 2 Capture Projet, 1, p. 451. [25] Darde, A., Prabhakar, R., Tranier, J., and Perrin, N., 2009. Air separation and flue gas ompression and purifiation units for oxy-oal ombustion systems. Energy Proedia, 1(1), pp. 527 534. [26] Amann, J., Kannihe, M., and Bouallou, C., 2009. Natural gas ombined yle power plant modified into an O 2 /CO 2 yle for CO 2 apture. Energy Conversion and Management, 50(3), pp. 510 521. [27] Sanz, W., Jeriha, H., Lukel, F., Göttlih, E., and Heitmeir, F., 2005. A Further Step Towards a Graz Cyle Power Plant for CO 2 Capture. ASME Turbo Expo 2005: Power for Land, Sea and Air. [28] Sammak, M., Jonshagen, K., Thern, M., Genrup, M., Thorbergsson, E., and Grönstedt, T., 2011. Coneptual Design of a Mid-Sized, Semi-losed oxy-fuel ombustion ombined yle. ASME Turbo Expo 2011: Power for Land, Sea and Air, 6-10 June. [29] Dixon, S., and Hall, C., 2010. Fluid mehanis and thermodynamis of turbomahinery. Butterworth-Heinemann. [30] Walsh, P., and Flether, P., 2004. Gas turbine performane. Blakwell Siene In. [31] Mukherji, D., and Rösler, J., 2011. Design Considerations and Strengthening Mehanisms in Developing Co- Re-Based Alloys for Appliations at+ 100 C above Ni- Superalloys. Advaned Materials Researh, 278, pp. 539 544. [32] Sato, A., Chiu, Y., and Reed, R., 2011. Oxidation of nikel-based single-rystal superalloys for industrial gas turbine appliations. Ata Materialia, 59(1), pp. 225 240. 9 Copyright 2012 by ASME