- Email: [email protected]
MAT, a novel, open cycle gas turbine for power augmentation
PII:
Energy Convers. Mgmt Vol. 39, No. 16±18, pp. 1631±1642, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain S0196-8904(98)00088-0 0196-8904/98 $19.00 + 0.00
MAT, A NOVEL, OPEN CYCLE GAS TURBINE FOR POWER AUGMENTATION M. UTAMURA,$* I. TAKEHARA$ and H. KARASAWA% Hitachi Works, Hitachi, Ltd., 3-1-1 Saiwai-cho, Hitachi City 317, Japan
$
Power and Industrial Systems Research and Development Division, Hitachi, Ltd., 7-2-1 Omika-cho, Hitachi City 319-12, Japan
%
AbstractÐA Moisture Air Turbine (MAT) cycle is proposed for improving the characteristics of landbased gas turbines by injecting atomized water through an inlet into a compressor. Compressor work of isentropic compression for moist air mixtures with phase change is theoretically considered, which has revealed that water evaporation may reduce compressor work. An experiment using a 15 MW class axial ¯ow load compressor has also veri®ed the theory. Realistic cycle model calculations predict that a 10% power increment by a ratio of 1% water to compressor intake air is expected and also that the amount of water consumption is much less than that of conventional inlet air cooling systems, used for heat rejection at the cooling tower. In addition, thermal eciency is anticipated to be improved mainly due to the reduction of compressor work. Contrary to the conventional evaporative cooler, a MAT cycle could provide power output at a desired value within its capability regardless of ambient humidity condition. # 1998 Elsevier Science Ltd. All rights reserved Gas turbine evaporation
Compressor Power augmentation Water injection MAT cycle Phase change Isentropic compression
Atomization
Water
NOMENCLATURE Cp=Isobaric speci®c heat (J/(kgK)) f=Speci®c fuel heat (J/kg) g=Water content to dry air ratio (-) Hi=Speci®c enthalpy at location i (J/kg) h'=Speci®c enthalpy of saturated water (J/kg) h0=Speci®c enthalpy of saturated steam (J/kg) Lij=Speci®c compressor work from the state i to j (J/kg) m=1 ÿ 1/k (-) Ps=Water vapor pressure (Pa) Pcd=Compressor discharge pressure (Pa) p=Speci®c power output (W/(kg/s)) Q=Gas turbine power output (W) R=Compressor work ratio (-) s=Speci®c entropy (J/(kgK)) T=Temperature (K) v=Speci®c volume (m3/kg) W=Compressor intake air ¯ow (kg/s) x=Dryness ((h ÿ h')/r) (-) a=Portion of mechanism (2) to a total power increment (-) k=Speci®c heat ratio (-) fij=Pressure ratio of state j to state i (-) Z=Gas turbine thermal eciency (-) Compressor adiabatic eciency (-) r=Latent heat of evaporation (h0 ÿ h') (J/kg) Di=Increment of property i
INTRODUCTION
Needs are increasing for technologies to recover the output of gas turbines in summer. One typical system increases output by cooling air coming into the compressor, thus increasing air density and the mass ¯ow of working ¯uids. Several of these systems are in use at power plants [1], and research eorts are under way for putting such systems into more practical use [2, 3]. With *To whom all correspondence should be addressed. Tel: 00 294 23 5258; Fax: 00 294 23 6735. 1631
1632
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
these systems, the most common method is to accumulate heat using night-time electric power, discharge the heat back during peak daytime hours, and increase output. This makes the running time with increased output depend on the amount of heat accumulated. However, on a practical level considerable space is required for a heat storage tank. The present study proposes, and examines the possibilities of, a Moisture Air Turbine (MAT) cycle, a new technology for increasing the output of a gas turbine by introducing a ®ne normal temperature water spray into the incoming air in the compressor. This technology is characterized by the fact that it is subject to few practical constraints in that it requires no heat accumulation and that it can increase output on demand. Analogous to the MAT cycle, water injection into the compressor is known to produce extra thrust in aircraft engines on taking o. This method, however, does not control the size of water droplets and thus blades might suer from erosion or corrosion. In contrast, the MAT cycle assumes generation of very ®ne water droplets to minimize mechanical and chemical interaction with compressor blades. This paper reports ®ndings from a theoretical study of the heat cycle and from veri®cation of its principle in experiments. THE PRINCIPLES OF INCREASED OUTPUT WITH A WATER SPRAY
The compression process of wet air with phase changes The MAT cycle aims to enter ®ne water droplets into the incoming air to the compressor, thus increasing the output of the gas turbine. Figure 1 is a psychrometric chart representing how a working ¯uid becomes compressed under ambient air conditions. Suppose, for example, that the ambient air condition is 208C and 70% R.H., and the thermodynamic state can be de®ned by point A in Fig. 1. Supposing that the ambient air is humidi®ed and cooled along the equi-wetbulb temperature line, reaching a saturated state before entering the compressor, the incoming air moves to state B at the compressor inlet. Liquid droplets that have not evaporated outside the compressor evaporate continuously in later compression. Assuming that the evaporation process remains saturated and that the total entropy of the two-component two-phase mixture is unchanged, the liquid ®nishes boiling at point C and the C-to-D process enters the single-phase compression process, resulting in a temperature rise. If evaporation is accompanied
Fig. 1. Change in state of working ¯uid during compression.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1633
Fig. 2. Incremental power mechanism of MAT cycle.
with an entropic rise, the mixture becomes unsaturated. In reality, since the rate of evaporation from liquid droplets is limited, the state change experiences thermal non-equilibrium. The result is that the actual process presumably deviates from the saturation line, following the locus of the broken line (B-to-D'). However, the normal compression process follows the A-to-D0 locus. Output increment mechanism The output increment mechanism can be divided into the following sections in qualitative terms: 1. Cooling of the incoming air on the equi-wetbulb temperature line with regard to the upstream ¯ow in the compressor. 2. Cooling of the internal gas due to the evaporation of liquid droplets sucked in by the compressor. 3. The dierence between the amounts of working ¯uid passing through the turbine and compressor. This dierence corresponds to the amount of the evaporation in the compressor. 4. An increase in the isobaric speci®c heat of the air±water mixture due to the increase of steam having a high isobaric speci®c heat. Fig. 2 expresses the above mechanism as a combination of analogous conventional techniques. Since the output of the gas turbine is the product of speci®c output multiplied by the incoming contribution ¯ow rate of the compressor, the output increase can be expressed by the sum of due to an increase in the incoming ¯ow rate and contribution due to a rise in speci®c out, while mechanism sections (2), (3), and (4) stemput. Mechanism section (1) is classi®ed as . ming from the behavior of liquid droplets introduced into the compressor are classi®ed as The MAT cycle can be interpreted as a combination of the eects of existing power boosting techniques. The mechanism for reducing compressor work by evaporation Figure 3 represents how the evaporation of liquid droplets reduces work consumed by the compressor. For simpli®cation, this process is described schematically with the reversible adiabatic compression process (ds = 0) with a piston. Suppose that the piston is at a certain compression pressure P1 and a temperature T. Since the state change in a particular stage in the
1634
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
Fig. 3. Mechanism of decrease in compressor work.
compressor is in the isobaric process (dP = 0), we make a simulation where pressure P1 is kept, with the spray liquid droplets evaporating and the temperature lowering by DT. Then, we see dh = 0 from the thermodynamic relation dh = Tds + vdP, resulting in a change of this process to an isenthalpic condition. Using the pressure ratio f(0P2/P1) for the subsequent compression, temperature at the pressure P2 is (T ÿ DT)fm. On the other hand, supposing that an equal amount of liquid droplets evaporate after the gas is compressed from pressure P1 to P2, the discharge temperature at pressure P2 is TfmÿDT. In that case, it is clear that piston work remains constant before and after spraying. In the spray amount-enthalpy diagram in Fig. 3, the envelope in the shaded portion represents gas temperature on an isenthalpic line of the mixture, corresponding to (TfmÿDT) stated above. The dierence between this and the discharge temperature for the case of evaporation in the intermediate stage is (TfmÿDT) ÿ (T ÿ DT)fm=DT(fmÿ1)>0. This indicates that, since the former temperature is lower than the latter, the outlet enthalpy of the former is smaller, so that the work required of the piston is reduced in the case of the onset of evaporation in the intermediate stage of the compressor. THEORETICAL THERMAL EFFICIENCY OF THE CYCLE
Thermal eciency of the increased output Of all output increment mechanisms described in the preceding section, (1) is well known as a rise in the mass ¯ow of the incoming air due to cooling. Therefore, we are going to study how the cooling of the gas in the compressor aects the speci®c output and thermal eciency. Figure 4 shows a heat cycle diagram. For simpli®cation, let us ignore the eects of the increase in working ¯uid due to evaporation, or mechanisms (3) and (4), and consider output increment mechanism (2) alone. The output per unit mass of incoming air, or speci®c output p, is p turbine shaft output ÿ compressor work H3 ÿ H4 ÿ H2 ÿ H1 :
1
In the above, we assume that the turbine inlet enthalpy H3 is constant. We will then ®nd that output increment Dp is caused by a decrease in the enthalpy at the compressor outlet to H'2, in the case of the MAT cycle. Then, we obtain Dp H2 ÿ H02 :
2
On the other hand, fuel consumption f is equal to H3ÿH2. Then, fuel increment Df is also expressed by Df H2 ÿ H02 :
3
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1635
Fig. 4. Enthalpy pro®le within GT.
From equations (2) and (3), we ®nd that Dp = Df(0D) and thus the thermal eciency of mechanism (2) of MAT cycle Z1 exceeds that of Brayton cycle Z0 because of the inequality Z10(p + D)/(f + D)>p/f0 Z0 holds provided f>p which is always satis®ed. It should be noted that the heat rate to gain incremental output Dp, i.e. Df/Dp is found to be Z0 times smaller than that of the Brayton cycle as far as additional fuel consumption is concerned. In comparison, the energy eciencies achieved with traditional power boosting techniques are lower than the thermal eciency of the gas turbine by an amount equivalent to the energy consumption required for cold heat production. Water injection in the combustor does not cause a reduction in compressor work and hence does not result in a reduction of thermal eciency. Based on the assumption that the temperature ratio (0T3/T1) and pressure ratio (0P2/P1) are identical, Table 1 shows the rankings of the gas turbine characteristics in each cycle. As shown in Fig. 4, the MAT cycle is the best thermal cycle in terms of both speci®c output and thermal eciency. The intercooling cycle achieves a high speci®c output but has a low cycle thermal eciency, because of heat loss in the compressor. Why the MAT cycle achieves a high thermal eciency can be attributed to being that the compression process with this cycle approaches isothermal changes, or follows an asymptotic curve with regard to the Ericsson cycle having a theoretical thermal eciency equal to that of the Carnot cycle. The section given below gives a theoretical study of how much reduction in work is expected. Theoretical work due to the compression of a two-component two-phase mixture Incorporating Fig. 1, the isentropic compression work L12 of dry air is L12 Cp T fm AD00 ÿ 1
4
On the other hand, compression work L'12 in the two-component two-phase compression proTable 1. Characteristics ranking of various cycles Cycle Brayton Cycle Intercooling Cycle MAT Cycle
Thermal Eciency
Speci®c Output
2 3 1
3 2 1
1636
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
cess including the evaporation of liquid droplets can be expressed by the equation given below as the sum of work in the liquid-evaporation process (B to C in Fig. 1) and the work of the two-component single-phase mixture. L012 gh00c ÿ h0B xB rB Cp TC ÿ TB C0p TC fm CD ÿ 1 1 g:
5
Total pressure P at evaporation completion point C on the saturated line is expressed by the expression given below, using steam pressure Ps when the two-phase saturated wet air at state B is compressed isentropically and when dryness x reaches 1 at state C. P
0:622 g Ps : g
6
Ratio R of compressor work with water evaporation to that without water evaporation is R
L012 : L12
7
Fig. 5 is a representation of R with regard to the spray amount. The outlet pressure of the compressor is used as a parameter. Ambient air conditions were assumed to be equal to the experimental conditions mentioned later in this paper. Since R < 1 at the time of spraying, in theory, we ®nd that there exists a reduction in motive power due to the evaporation of liquid droplets in the compression process. This also indicates that the dependence on discharge pressure Pcd is small. The reduction of compressor work is almost proportionate to the amount of spray, while work reduction for spraying amount of 1% to air mass ratio is 6.8%. PREDICTION OF THE CHARACTERISTICS OF AN ACTUAL CYCLE
This section makes an attempt to quantify the characteristics of the MAT cycle, using a calculation model similar to the characteristics of an actual gas turbine, such as adiabatic eciency, air extraction of the compressor, and changes in gas composition due to water evaporation. In particular, this section evaluates the eects of the spray amount and compares the MAT cycle performance with that of a traditional power boosting solution that injects water into a combustion device.
Fig. 5. In¯uence of spray ¯ow rate on compressor work ratio, R.
1637
Fig. 6. GT characteristics vs. water injection ¯ow rate.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1638
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE Table 2. Fraction of incremental power output
Mechanism of MAT cycle
Air Condition: 358C, 53% R.H., Injected Water: 2.3%
Compressor External Cooling (1) Compressor Internal Cooling (2) Mass Flow Increase Larger Speci®c Heat (3, 4)
35% 37% 28%
Table 3. Water consumption Consumed Water MAT cycle
1 (base)
Inlet Air Cooling with Thermal Energy Storage Ice Storage (Turbo-Refrigerator) (COP = 2.5) Liquid Air Storage (COP = 0.5)
1.8 2.5
One of the calculation conditions is that parameters have been selected that assume an axial ¯ow gas turbine of 150 MW class. The ambient air is assumed to be 358C and 53% R.H. We did not include liquid droplets in the ¯ow rate of the compressor, and performed calculations based on the assumption that all would evaporate at one time during a certain intermediate stage. Figure 6 shows how the spray amount aects the output increment ratio and thermal eciency. The output increment sensitivity with regard to the unit spray amount is higher in the ) than in the cooling (mechanisms , and cooling outside the compressor (mechanism ) in the compressor. The water consumption required to restore the gas turbine output at ambient air temperature of 358C to the output during gas turbine baseload operation at 58C is about 2.3% of the incoming air. The thermal eciency rise at 2.3% spray is 2.8% in relative terms. Table 2 gives the details of the output increment for each mechanism in Fig. 2. The ®gure gives calculated values under a 2.3% water spray condition. The portions of mechanisms and correspond to the well-known steam injection technique, accounting for a third of the total. This means that the MAT cycle approximately triples the output of the steam injection method using the same water amount. Table 2 and Fig. 6 reveal that the MAT cycle achieves an output increment of 65% with a 2% spray amount, even though the ambient air is in a state of saturated wet air. Thus, the MAT cycle attains a required output within its capacity regardless of the presence of moisture in the ambient air, due to the contribution of mechanisms , and . This is an essential dierence from existing evaporative coolers in terms of functionality. Table 3 compares the water consumption of MAT with a traditional inlet air cooling system. Although the inlet air cooling system also assumes that a cooling tower consumes water for heat rejection during heat accumulation, the water consumption of the MAT cycle is found to be smaller than that of Ref. [4]. EXPERIMENT FOR EXAMINING COOLING CHARACTERISTICS
Method of experiment Table 4 gives the main speci®cation of the axial ¯ow load compressor for testing and the range of experimental parameters and Fig. 7 outlines the experimental apparatus. The compresTable 4. Experimental conditions Items
Values
Compressor Speci®cations Rotation Speed (rpm) Pressure Ratio Mass Flow (kg/s) Work (kW) Number of Stages Axial Length Spray Nozzle Speci®cation Flow Rate (vs. Inlet Air Ratio)
11,000 9.07 36.3 15,000 17 1850 mm Two Fluid Nozzle 0±2%
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1639
Fig. 7. Components of experimental apparatus.
sor is connected to a drive gas turbine through a ¯uid coupler and its transmission torque can be controlled by adjusting the position of the squeeze tube. During the test, we adjusted the compressor rotation speed to 11,000 rpm and the discharge pressure to a speci®ed value. The spray amount was controlled with a water supply valve, while the grain size was adjusted by controlling the air supply. The amount of ¯ow entering the compressor was measured with a Pitot tube. We also measured the temperatures of various parts and the humidity at the inlet and outlet of the compressor. The spray water temperature was room temperature. The average air ¯ow speed in the suction duct was 20 m/s.
Fig. 8. Fraction of evaporation of water droplet.
1640
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
Fig. 9. Compressor exit temperature dierence.
Experiment results and discussion Figure 8 shows the correlation between the amount of droplets introduced (ratio to air) into the compressor based on the spray amount, air ¯ow, and ambient air conditions and the directly measured absolute humidities at the EGV position at the compressor outlet. Liquid droplets that ¯owed into the compressor are found to have evaporated at least 95% before reaching the ®nal stage. Figure 9 shows the relationship between the spray amount and the temperature dierence at the compressor outlet before and during spraying. We ®nd that the evaporation and cooling took place eciently at a low ¯ow-rate before the gas ¯owed into the compressor. The humidity reached was near 95%. The solid line indicates calculations of the temperature dierence based on the following two conditions: (1) the absolute humidity of the outlet gas was obtained based on the assumption that all liquid droplets that ¯owed into the compressor evaporated; and (2) the outlet gas enthalpy of the compressor was equal to the experimental value before spraying, i.e. isenthalpic state of change. This means that this line represents a state where the compressor shows no decline in motive power. The fact that the experimental values exceed them is indicative of the fact that the cooling eect of the gas is higher than the simple evaporation of liquid droplets. These ®ndings led us to determine that there really was a decline in motive power. This can be interpreted as being due to the fact that the temperature decline due to the evaporation of liquid droplets in the intermediate stage of the compressor is ampli®ed in the compressor process in the stages following the evaporation point. This convinces us that evaporation in the former stage is advantageous in reducing motive power. The dotted line
Fig. 10. S dependence on spray ¯ow rate.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1641
Fig. 11. S sensitivity to compressor adiabatic eciency.
shows the calculation results based on the assumption of isentropic compression using the theory mentioned in the foregoing chapter. Measurements exist between two calculations. For the sake of convenience, we here de®ne the adiabatic eciency at the time of spraying as the ratio of isentropic compression work without spray and the motive power at the time of spraying. Figure 10 indicates the ratio S(0 ZM/ZD) of the adiabatic eciency (ZM) at the time of spraying to the adiabatic eciency (ZD) without spraying, with regard to the spray amount. The solid line is equal to the inverse (=1/R) of the theoretical value of two-phase isentropic compression (equation (7) mentioned in the previous section. At the same time, the measured values indicate the motive powers calculated with thermodynamic values measured at the inlet and outlet of the compressor. Within the experimental range, the adiabatic eciency with regard to unit spray amount (1%) was 4%. S is almost proportionate to the spray amount, and we will examine its gradient. Figure 11 shows an increase in adiabatic eciency expected when a unit spray amount is given, with regard to the adiabatic eciency of the compressor in the true sense of the word. Our ®nding is that the higher the adiabatic eciency, the higher the rise in adiabatic eciency, that is, machines with higher adiabatic eciencies produce higher MAT eect. We have observed no evidence of erosion on compressor blades after 100 h of continual water injection. CONCLUSION
In the present study, we proposed a MAT cycle technology, a technique for improving the power output and thermal eciency of a gas turbine by entering water spray into the suction side of an axial ¯ow compressor, and demonstrated how the characteristics of a compressor can be improved with relatively little water, in theory and by experiment. 1. This technique is expected to achieve an output increment of about 10% by applying water at an incoming air ratio of 1% with the ambient air at 358C and 53% R.H. 2. The theoretical heat rate for additional fuel consumption due to an output increment mechanism stemming from a decline in the motive power of a compressor was Z0 times lower than that of Brayton cycle where Z0 is the thermal eciency of the Brayton cycle. 3. The theoretical work decline was 6.8% with regard to 1% water spray, while a compressor with an adiabatic eciency of 0.79 produced a measurement of 4%. 4. We formulated a rise in adiabatic eciency due to the evaporation of liquid droplets in the compressor. Our future challenges are to determine the top limit spray amount in linkage with the characteristics of a compressor and to demonstrate the overall characteristics of power increment. AcknowledgementsÐThe authors would like to express their thanks to Professor Riti Singh, Head of Department, Propulsion, Power and Automotive Engineering, School of Mechanical Engineering, Cran®eld University and Dr. Pericles Pilidis, Director-Thermal Power MSc, School of Mechanical Engineering, Cran®eld University for their valuable
1642
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
discussions on gas turbine performance with water injection. Special thanks are owed to Messrs. Takaaki, Kuwahara, Nobuyuki Horii, and Hidetaro Murata, engineers of the Thermal Plant Design Department, Hitachi Works, Hitachi Ltd., for their cooperation in the experiment.
REFERENCES 1. 2. 3. 4.
Ebeling, J. A. et al., ASME int. gas turbine and aeroengine congress and exposition, 1992. Saito, K., Journal of Gas Turbine Society of Japan, 1996, 23(92), P3. Tanaka, M. et al., (ibid.), 19??, p. P11. Utamura, M. et al., ASME TURBO EXPO '96, 96-GT-463, 1996.
Energy Convers. Mgmt Vol. 39, No. 16±18, pp. 1631±1642, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain S0196-8904(98)00088-0 0196-8904/98 $19.00 + 0.00
MAT, A NOVEL, OPEN CYCLE GAS TURBINE FOR POWER AUGMENTATION M. UTAMURA,$* I. TAKEHARA$ and H. KARASAWA% Hitachi Works, Hitachi, Ltd., 3-1-1 Saiwai-cho, Hitachi City 317, Japan
$
Power and Industrial Systems Research and Development Division, Hitachi, Ltd., 7-2-1 Omika-cho, Hitachi City 319-12, Japan
%
AbstractÐA Moisture Air Turbine (MAT) cycle is proposed for improving the characteristics of landbased gas turbines by injecting atomized water through an inlet into a compressor. Compressor work of isentropic compression for moist air mixtures with phase change is theoretically considered, which has revealed that water evaporation may reduce compressor work. An experiment using a 15 MW class axial ¯ow load compressor has also veri®ed the theory. Realistic cycle model calculations predict that a 10% power increment by a ratio of 1% water to compressor intake air is expected and also that the amount of water consumption is much less than that of conventional inlet air cooling systems, used for heat rejection at the cooling tower. In addition, thermal eciency is anticipated to be improved mainly due to the reduction of compressor work. Contrary to the conventional evaporative cooler, a MAT cycle could provide power output at a desired value within its capability regardless of ambient humidity condition. # 1998 Elsevier Science Ltd. All rights reserved Gas turbine evaporation
Compressor Power augmentation Water injection MAT cycle Phase change Isentropic compression
Atomization
Water
NOMENCLATURE Cp=Isobaric speci®c heat (J/(kgK)) f=Speci®c fuel heat (J/kg) g=Water content to dry air ratio (-) Hi=Speci®c enthalpy at location i (J/kg) h'=Speci®c enthalpy of saturated water (J/kg) h0=Speci®c enthalpy of saturated steam (J/kg) Lij=Speci®c compressor work from the state i to j (J/kg) m=1 ÿ 1/k (-) Ps=Water vapor pressure (Pa) Pcd=Compressor discharge pressure (Pa) p=Speci®c power output (W/(kg/s)) Q=Gas turbine power output (W) R=Compressor work ratio (-) s=Speci®c entropy (J/(kgK)) T=Temperature (K) v=Speci®c volume (m3/kg) W=Compressor intake air ¯ow (kg/s) x=Dryness ((h ÿ h')/r) (-) a=Portion of mechanism (2) to a total power increment (-) k=Speci®c heat ratio (-) fij=Pressure ratio of state j to state i (-) Z=Gas turbine thermal eciency (-) Compressor adiabatic eciency (-) r=Latent heat of evaporation (h0 ÿ h') (J/kg) Di=Increment of property i
INTRODUCTION
Needs are increasing for technologies to recover the output of gas turbines in summer. One typical system increases output by cooling air coming into the compressor, thus increasing air density and the mass ¯ow of working ¯uids. Several of these systems are in use at power plants [1], and research eorts are under way for putting such systems into more practical use [2, 3]. With *To whom all correspondence should be addressed. Tel: 00 294 23 5258; Fax: 00 294 23 6735. 1631
1632
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
these systems, the most common method is to accumulate heat using night-time electric power, discharge the heat back during peak daytime hours, and increase output. This makes the running time with increased output depend on the amount of heat accumulated. However, on a practical level considerable space is required for a heat storage tank. The present study proposes, and examines the possibilities of, a Moisture Air Turbine (MAT) cycle, a new technology for increasing the output of a gas turbine by introducing a ®ne normal temperature water spray into the incoming air in the compressor. This technology is characterized by the fact that it is subject to few practical constraints in that it requires no heat accumulation and that it can increase output on demand. Analogous to the MAT cycle, water injection into the compressor is known to produce extra thrust in aircraft engines on taking o. This method, however, does not control the size of water droplets and thus blades might suer from erosion or corrosion. In contrast, the MAT cycle assumes generation of very ®ne water droplets to minimize mechanical and chemical interaction with compressor blades. This paper reports ®ndings from a theoretical study of the heat cycle and from veri®cation of its principle in experiments. THE PRINCIPLES OF INCREASED OUTPUT WITH A WATER SPRAY
The compression process of wet air with phase changes The MAT cycle aims to enter ®ne water droplets into the incoming air to the compressor, thus increasing the output of the gas turbine. Figure 1 is a psychrometric chart representing how a working ¯uid becomes compressed under ambient air conditions. Suppose, for example, that the ambient air condition is 208C and 70% R.H., and the thermodynamic state can be de®ned by point A in Fig. 1. Supposing that the ambient air is humidi®ed and cooled along the equi-wetbulb temperature line, reaching a saturated state before entering the compressor, the incoming air moves to state B at the compressor inlet. Liquid droplets that have not evaporated outside the compressor evaporate continuously in later compression. Assuming that the evaporation process remains saturated and that the total entropy of the two-component two-phase mixture is unchanged, the liquid ®nishes boiling at point C and the C-to-D process enters the single-phase compression process, resulting in a temperature rise. If evaporation is accompanied
Fig. 1. Change in state of working ¯uid during compression.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1633
Fig. 2. Incremental power mechanism of MAT cycle.
with an entropic rise, the mixture becomes unsaturated. In reality, since the rate of evaporation from liquid droplets is limited, the state change experiences thermal non-equilibrium. The result is that the actual process presumably deviates from the saturation line, following the locus of the broken line (B-to-D'). However, the normal compression process follows the A-to-D0 locus. Output increment mechanism The output increment mechanism can be divided into the following sections in qualitative terms: 1. Cooling of the incoming air on the equi-wetbulb temperature line with regard to the upstream ¯ow in the compressor. 2. Cooling of the internal gas due to the evaporation of liquid droplets sucked in by the compressor. 3. The dierence between the amounts of working ¯uid passing through the turbine and compressor. This dierence corresponds to the amount of the evaporation in the compressor. 4. An increase in the isobaric speci®c heat of the air±water mixture due to the increase of steam having a high isobaric speci®c heat. Fig. 2 expresses the above mechanism as a combination of analogous conventional techniques. Since the output of the gas turbine is the product of speci®c output multiplied by the incoming contribution ¯ow rate of the compressor, the output increase can be expressed by the sum of due to an increase in the incoming ¯ow rate and contribution due to a rise in speci®c out, while mechanism sections (2), (3), and (4) stemput. Mechanism section (1) is classi®ed as . ming from the behavior of liquid droplets introduced into the compressor are classi®ed as The MAT cycle can be interpreted as a combination of the eects of existing power boosting techniques. The mechanism for reducing compressor work by evaporation Figure 3 represents how the evaporation of liquid droplets reduces work consumed by the compressor. For simpli®cation, this process is described schematically with the reversible adiabatic compression process (ds = 0) with a piston. Suppose that the piston is at a certain compression pressure P1 and a temperature T. Since the state change in a particular stage in the
1634
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
Fig. 3. Mechanism of decrease in compressor work.
compressor is in the isobaric process (dP = 0), we make a simulation where pressure P1 is kept, with the spray liquid droplets evaporating and the temperature lowering by DT. Then, we see dh = 0 from the thermodynamic relation dh = Tds + vdP, resulting in a change of this process to an isenthalpic condition. Using the pressure ratio f(0P2/P1) for the subsequent compression, temperature at the pressure P2 is (T ÿ DT)fm. On the other hand, supposing that an equal amount of liquid droplets evaporate after the gas is compressed from pressure P1 to P2, the discharge temperature at pressure P2 is TfmÿDT. In that case, it is clear that piston work remains constant before and after spraying. In the spray amount-enthalpy diagram in Fig. 3, the envelope in the shaded portion represents gas temperature on an isenthalpic line of the mixture, corresponding to (TfmÿDT) stated above. The dierence between this and the discharge temperature for the case of evaporation in the intermediate stage is (TfmÿDT) ÿ (T ÿ DT)fm=DT(fmÿ1)>0. This indicates that, since the former temperature is lower than the latter, the outlet enthalpy of the former is smaller, so that the work required of the piston is reduced in the case of the onset of evaporation in the intermediate stage of the compressor. THEORETICAL THERMAL EFFICIENCY OF THE CYCLE
Thermal eciency of the increased output Of all output increment mechanisms described in the preceding section, (1) is well known as a rise in the mass ¯ow of the incoming air due to cooling. Therefore, we are going to study how the cooling of the gas in the compressor aects the speci®c output and thermal eciency. Figure 4 shows a heat cycle diagram. For simpli®cation, let us ignore the eects of the increase in working ¯uid due to evaporation, or mechanisms (3) and (4), and consider output increment mechanism (2) alone. The output per unit mass of incoming air, or speci®c output p, is p turbine shaft output ÿ compressor work H3 ÿ H4 ÿ H2 ÿ H1 :
1
In the above, we assume that the turbine inlet enthalpy H3 is constant. We will then ®nd that output increment Dp is caused by a decrease in the enthalpy at the compressor outlet to H'2, in the case of the MAT cycle. Then, we obtain Dp H2 ÿ H02 :
2
On the other hand, fuel consumption f is equal to H3ÿH2. Then, fuel increment Df is also expressed by Df H2 ÿ H02 :
3
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1635
Fig. 4. Enthalpy pro®le within GT.
From equations (2) and (3), we ®nd that Dp = Df(0D) and thus the thermal eciency of mechanism (2) of MAT cycle Z1 exceeds that of Brayton cycle Z0 because of the inequality Z10(p + D)/(f + D)>p/f0 Z0 holds provided f>p which is always satis®ed. It should be noted that the heat rate to gain incremental output Dp, i.e. Df/Dp is found to be Z0 times smaller than that of the Brayton cycle as far as additional fuel consumption is concerned. In comparison, the energy eciencies achieved with traditional power boosting techniques are lower than the thermal eciency of the gas turbine by an amount equivalent to the energy consumption required for cold heat production. Water injection in the combustor does not cause a reduction in compressor work and hence does not result in a reduction of thermal eciency. Based on the assumption that the temperature ratio (0T3/T1) and pressure ratio (0P2/P1) are identical, Table 1 shows the rankings of the gas turbine characteristics in each cycle. As shown in Fig. 4, the MAT cycle is the best thermal cycle in terms of both speci®c output and thermal eciency. The intercooling cycle achieves a high speci®c output but has a low cycle thermal eciency, because of heat loss in the compressor. Why the MAT cycle achieves a high thermal eciency can be attributed to being that the compression process with this cycle approaches isothermal changes, or follows an asymptotic curve with regard to the Ericsson cycle having a theoretical thermal eciency equal to that of the Carnot cycle. The section given below gives a theoretical study of how much reduction in work is expected. Theoretical work due to the compression of a two-component two-phase mixture Incorporating Fig. 1, the isentropic compression work L12 of dry air is L12 Cp T fm AD00 ÿ 1
4
On the other hand, compression work L'12 in the two-component two-phase compression proTable 1. Characteristics ranking of various cycles Cycle Brayton Cycle Intercooling Cycle MAT Cycle
Thermal Eciency
Speci®c Output
2 3 1
3 2 1
1636
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
cess including the evaporation of liquid droplets can be expressed by the equation given below as the sum of work in the liquid-evaporation process (B to C in Fig. 1) and the work of the two-component single-phase mixture. L012 gh00c ÿ h0B xB rB Cp TC ÿ TB C0p TC fm CD ÿ 1 1 g:
5
Total pressure P at evaporation completion point C on the saturated line is expressed by the expression given below, using steam pressure Ps when the two-phase saturated wet air at state B is compressed isentropically and when dryness x reaches 1 at state C. P
0:622 g Ps : g
6
Ratio R of compressor work with water evaporation to that without water evaporation is R
L012 : L12
7
Fig. 5 is a representation of R with regard to the spray amount. The outlet pressure of the compressor is used as a parameter. Ambient air conditions were assumed to be equal to the experimental conditions mentioned later in this paper. Since R < 1 at the time of spraying, in theory, we ®nd that there exists a reduction in motive power due to the evaporation of liquid droplets in the compression process. This also indicates that the dependence on discharge pressure Pcd is small. The reduction of compressor work is almost proportionate to the amount of spray, while work reduction for spraying amount of 1% to air mass ratio is 6.8%. PREDICTION OF THE CHARACTERISTICS OF AN ACTUAL CYCLE
This section makes an attempt to quantify the characteristics of the MAT cycle, using a calculation model similar to the characteristics of an actual gas turbine, such as adiabatic eciency, air extraction of the compressor, and changes in gas composition due to water evaporation. In particular, this section evaluates the eects of the spray amount and compares the MAT cycle performance with that of a traditional power boosting solution that injects water into a combustion device.
Fig. 5. In¯uence of spray ¯ow rate on compressor work ratio, R.
1637
Fig. 6. GT characteristics vs. water injection ¯ow rate.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1638
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE Table 2. Fraction of incremental power output
Mechanism of MAT cycle
Air Condition: 358C, 53% R.H., Injected Water: 2.3%
Compressor External Cooling (1) Compressor Internal Cooling (2) Mass Flow Increase Larger Speci®c Heat (3, 4)
35% 37% 28%
Table 3. Water consumption Consumed Water MAT cycle
1 (base)
Inlet Air Cooling with Thermal Energy Storage Ice Storage (Turbo-Refrigerator) (COP = 2.5) Liquid Air Storage (COP = 0.5)
1.8 2.5
One of the calculation conditions is that parameters have been selected that assume an axial ¯ow gas turbine of 150 MW class. The ambient air is assumed to be 358C and 53% R.H. We did not include liquid droplets in the ¯ow rate of the compressor, and performed calculations based on the assumption that all would evaporate at one time during a certain intermediate stage. Figure 6 shows how the spray amount aects the output increment ratio and thermal eciency. The output increment sensitivity with regard to the unit spray amount is higher in the ) than in the cooling (mechanisms , and cooling outside the compressor (mechanism ) in the compressor. The water consumption required to restore the gas turbine output at ambient air temperature of 358C to the output during gas turbine baseload operation at 58C is about 2.3% of the incoming air. The thermal eciency rise at 2.3% spray is 2.8% in relative terms. Table 2 gives the details of the output increment for each mechanism in Fig. 2. The ®gure gives calculated values under a 2.3% water spray condition. The portions of mechanisms and correspond to the well-known steam injection technique, accounting for a third of the total. This means that the MAT cycle approximately triples the output of the steam injection method using the same water amount. Table 2 and Fig. 6 reveal that the MAT cycle achieves an output increment of 65% with a 2% spray amount, even though the ambient air is in a state of saturated wet air. Thus, the MAT cycle attains a required output within its capacity regardless of the presence of moisture in the ambient air, due to the contribution of mechanisms , and . This is an essential dierence from existing evaporative coolers in terms of functionality. Table 3 compares the water consumption of MAT with a traditional inlet air cooling system. Although the inlet air cooling system also assumes that a cooling tower consumes water for heat rejection during heat accumulation, the water consumption of the MAT cycle is found to be smaller than that of Ref. [4]. EXPERIMENT FOR EXAMINING COOLING CHARACTERISTICS
Method of experiment Table 4 gives the main speci®cation of the axial ¯ow load compressor for testing and the range of experimental parameters and Fig. 7 outlines the experimental apparatus. The compresTable 4. Experimental conditions Items
Values
Compressor Speci®cations Rotation Speed (rpm) Pressure Ratio Mass Flow (kg/s) Work (kW) Number of Stages Axial Length Spray Nozzle Speci®cation Flow Rate (vs. Inlet Air Ratio)
11,000 9.07 36.3 15,000 17 1850 mm Two Fluid Nozzle 0±2%
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1639
Fig. 7. Components of experimental apparatus.
sor is connected to a drive gas turbine through a ¯uid coupler and its transmission torque can be controlled by adjusting the position of the squeeze tube. During the test, we adjusted the compressor rotation speed to 11,000 rpm and the discharge pressure to a speci®ed value. The spray amount was controlled with a water supply valve, while the grain size was adjusted by controlling the air supply. The amount of ¯ow entering the compressor was measured with a Pitot tube. We also measured the temperatures of various parts and the humidity at the inlet and outlet of the compressor. The spray water temperature was room temperature. The average air ¯ow speed in the suction duct was 20 m/s.
Fig. 8. Fraction of evaporation of water droplet.
1640
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
Fig. 9. Compressor exit temperature dierence.
Experiment results and discussion Figure 8 shows the correlation between the amount of droplets introduced (ratio to air) into the compressor based on the spray amount, air ¯ow, and ambient air conditions and the directly measured absolute humidities at the EGV position at the compressor outlet. Liquid droplets that ¯owed into the compressor are found to have evaporated at least 95% before reaching the ®nal stage. Figure 9 shows the relationship between the spray amount and the temperature dierence at the compressor outlet before and during spraying. We ®nd that the evaporation and cooling took place eciently at a low ¯ow-rate before the gas ¯owed into the compressor. The humidity reached was near 95%. The solid line indicates calculations of the temperature dierence based on the following two conditions: (1) the absolute humidity of the outlet gas was obtained based on the assumption that all liquid droplets that ¯owed into the compressor evaporated; and (2) the outlet gas enthalpy of the compressor was equal to the experimental value before spraying, i.e. isenthalpic state of change. This means that this line represents a state where the compressor shows no decline in motive power. The fact that the experimental values exceed them is indicative of the fact that the cooling eect of the gas is higher than the simple evaporation of liquid droplets. These ®ndings led us to determine that there really was a decline in motive power. This can be interpreted as being due to the fact that the temperature decline due to the evaporation of liquid droplets in the intermediate stage of the compressor is ampli®ed in the compressor process in the stages following the evaporation point. This convinces us that evaporation in the former stage is advantageous in reducing motive power. The dotted line
Fig. 10. S dependence on spray ¯ow rate.
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
1641
Fig. 11. S sensitivity to compressor adiabatic eciency.
shows the calculation results based on the assumption of isentropic compression using the theory mentioned in the foregoing chapter. Measurements exist between two calculations. For the sake of convenience, we here de®ne the adiabatic eciency at the time of spraying as the ratio of isentropic compression work without spray and the motive power at the time of spraying. Figure 10 indicates the ratio S(0 ZM/ZD) of the adiabatic eciency (ZM) at the time of spraying to the adiabatic eciency (ZD) without spraying, with regard to the spray amount. The solid line is equal to the inverse (=1/R) of the theoretical value of two-phase isentropic compression (equation (7) mentioned in the previous section. At the same time, the measured values indicate the motive powers calculated with thermodynamic values measured at the inlet and outlet of the compressor. Within the experimental range, the adiabatic eciency with regard to unit spray amount (1%) was 4%. S is almost proportionate to the spray amount, and we will examine its gradient. Figure 11 shows an increase in adiabatic eciency expected when a unit spray amount is given, with regard to the adiabatic eciency of the compressor in the true sense of the word. Our ®nding is that the higher the adiabatic eciency, the higher the rise in adiabatic eciency, that is, machines with higher adiabatic eciencies produce higher MAT eect. We have observed no evidence of erosion on compressor blades after 100 h of continual water injection. CONCLUSION
In the present study, we proposed a MAT cycle technology, a technique for improving the power output and thermal eciency of a gas turbine by entering water spray into the suction side of an axial ¯ow compressor, and demonstrated how the characteristics of a compressor can be improved with relatively little water, in theory and by experiment. 1. This technique is expected to achieve an output increment of about 10% by applying water at an incoming air ratio of 1% with the ambient air at 358C and 53% R.H. 2. The theoretical heat rate for additional fuel consumption due to an output increment mechanism stemming from a decline in the motive power of a compressor was Z0 times lower than that of Brayton cycle where Z0 is the thermal eciency of the Brayton cycle. 3. The theoretical work decline was 6.8% with regard to 1% water spray, while a compressor with an adiabatic eciency of 0.79 produced a measurement of 4%. 4. We formulated a rise in adiabatic eciency due to the evaporation of liquid droplets in the compressor. Our future challenges are to determine the top limit spray amount in linkage with the characteristics of a compressor and to demonstrate the overall characteristics of power increment. AcknowledgementsÐThe authors would like to express their thanks to Professor Riti Singh, Head of Department, Propulsion, Power and Automotive Engineering, School of Mechanical Engineering, Cran®eld University and Dr. Pericles Pilidis, Director-Thermal Power MSc, School of Mechanical Engineering, Cran®eld University for their valuable
1642
UTAMURA et al.: MAT, A NOVEL, OPEN CYCLE GAS TURBINE
discussions on gas turbine performance with water injection. Special thanks are owed to Messrs. Takaaki, Kuwahara, Nobuyuki Horii, and Hidetaro Murata, engineers of the Thermal Plant Design Department, Hitachi Works, Hitachi Ltd., for their cooperation in the experiment.
REFERENCES 1. 2. 3. 4.
Ebeling, J. A. et al., ASME int. gas turbine and aeroengine congress and exposition, 1992. Saito, K., Journal of Gas Turbine Society of Japan, 1996, 23(92), P3. Tanaka, M. et al., (ibid.), 19??, p. P11. Utamura, M. et al., ASME TURBO EXPO '96, 96-GT-463, 1996.