R. C. Wilcock J. B.

Young
e-mail: jby@eng.cam.ac.uk

The Effect of Turbine Blade Cooling on the Cycle Efficiency of Gas Turbine Power Cycles
A thermodynamic cycle analysis computer code for the performance prediction of cooled gas turbines has been used to calculate the efciency of plants with varying combustor outlet temperature, compressor pressure ratio, and turbomachinery polytropic efciency. It is shown that the polytropic efciency exerts a major inuence on the optimum operating point of cooled gas turbines: for moderate turbomachinery efciency the search for enhanced combustor outlet temperature is shown to be logical, but for high turbomachinery efciency this is not necessarily so. The sensitivity of the cycle efciency to variation in the parameters determining the cooling ow rates is also examined. While increases in allowable blade metal temperature and lm cooling effectiveness are more benecial than improvements in other parameters, neither is as important as increase in turbomachinery aerodynamic efciency. DOI: 10.1115/1.1805549

J. H. Horlock
Hopkinson Laboratory, Engineering Department, Cambridge University, Trumpington Street, Cambridge CB2 1PZ, UK

Introduction

The single most important factor responsible for the steady increase in gas turbine efciency over the last half century has been the increase in combustor outlet temperature T cot . Advances have been made possible by remarkable developments in both blade materials and cooling technology. So successful has this development program proved that few engineers would challenge the principle that the surest way to raise the efciency of the next generation of machines is to increase the top temperature. Nevertheless, MacArthur 1 and Horlock et al. 2 have recently argued that the efciency benets of higher temperatures may be more than offset by the increased losses associated with the cooling ow rates required. Using a computer code in which real gas effects were included, Horlock et al. 2 presented calculations of the variation of cycle efciency with T cot at various xed pressure ratios. These displayed efciency maxima at temperatures not signicantly higher than those of machines currently in operation. Surprisingly, the calculations gave similar maxima even without turbine cooling. A detailed study was therefore made of gas properties, as a limit on performance in the absence of cooling, by Wilcock et al. 3. This work focused on the effects of increased water content in the combustion products at higher fuel/air ratios leading to higher specic heat capacities c p and c v , but a lower ratio . It was shown that these real gas effects are responsible for the peaks in cycle efciency with T cot well below the stoichiometric limit. This paper presents more comprehensive calculations of gas turbine performance using an improved code. Cycle efciencies with and without turbine cooling are compared for a range of turbomachinery efciencies. The calculations provide an overview, which is lacking in less general studies, and lead to conclusions which are of considerable importance for the gas turbine designer.

temperature T cot , and the compressor and turbine polytropic efciencies poly, C and poly, T . The latter refers to an uncooled turbine, the cooling losses being estimated separately as described below. The working uid is assumed to be a mixture of the four semiperfect gases N2 , O2 , CO2 , and H2 O. Semi-perfect implies that the specic heat capacities depend on temperature but not pressure. They were represented by polynomial curve ts to the data in the JANAF tables Chase et al. 6. Combustion for a given fuel was modeled in the usual way using enthalpies and entropies of formation also taken from the JANAF tables. As shown in Fig. 2, the cooled turbine is divided into stages, each stator and rotor being treated separately. In order to avoid specifying the blade geometry, the relative total temperature at rotor inlet is estimated from a given stage loading coefcient, as described in Ref. 5. Coolant from the compressor enters the internal passages of the blade where it receives heat from the mainstream ow after conduction through the thermal barrier coating if present and the blade metal. A special treatment for rotating blades is described in Ref. 5. For lm-cooled blades, the coolant ows over the blade surface before mixing out with the mainstream ow. Mixing is assumed to be complete at entry to the next blade row. Full details of the theory are given in Ref. 5 but it will be evident that, in order to x the thermodynamic state downstream of a blade row, it is necessary to know, not only the cooling ow rate, but also the rate of entropy creation over and above the basic uncooled loss represented by poly, T ). These crucial aspects of the procedure are now discussed. 2.1 Prediction of the Cooling Flow Rates. The cooling ow rates were estimated using the method of Holland and Thake 7, modied to allow for the temperature variations across the metal and thermal barrier coating TBC. The Appendix provides a description of the method, and full details can be found in Ref. 5. The procedure models both convection and lm cooling, and is particularly attractive because the cooling ow fraction the ratio of the coolant mass owrate m c to the mainstream mass ow rate m g ) is expressed in terms of just six parameters see Eqs. A2. These are, 0 blade cooling effectiveness, K cool cooling ow factor, int internal cooling efciency, f lm cooling effectiveness, Bimet metal Biot number, and Bitbc TBC Biot number. Of these, 0 is fully specied by the gas and coolant inlet temperatures, and the allowable blade metal temperature. The remaining ve parameters are specied as input data. Table 1 JANUARY 2005, Vol. 127 109

Cycle Analysis Computer Code

The cycle analysis computer code is based on the theory of Young and Wilcock 4,5 and models a simple-cycle industrial gas turbine with cooling as shown in Fig. 1. The main parameters dening the cycle are the pressure ratio r p , the combustor outlet
Contributed by the Power Division of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS for publication in the ASME JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received by the Power Division July 10, 2003; nal revision received September 23, 2003. Editor: L. S. Langston.

Journal of Engineering for Gas Turbines and Power Copyright 2005 by ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

lists some typical values for current aeroderivative cooling technology and for two advanced technologies which might be achievable within the next few years. With the cooling ow fraction known, application of the steady-ow energy equation gives the values of the various temperatures introduced in the Appendix and also the specic total enthalpy of the mixed-out ow downstream of the blade row. 2.2 Estimation of the Cooling Irreversibilities. The irreversible entropy creation is calculated as described by Young and Wilcock 5. In this approach, the total irreversibility is divided into an uncooled loss and various cooling losses. The uncooled loss refers to the prole, secondary, and tip-leakage losses in an uncooled expansion and is obtained from a given stage uncooled polytropic efciency poly, T . A major assumption is that the uncooled loss is unchanged in the presence of cooling ows. Additional entropy creation results from the following processes, depicted in Fig. 3: i. Heat transfer from the mainstream ow to the blade outer surface ii. Heat transfer through the thermal barrier coating if present iii. Heat transfer through the blade metal iv. Heat transfer from the blade inner surface to the coolant v. Frictional pressure loss between compressor offtake and coolant injection vi. Thermal mixing i.e., the turbulent mixing process wherein the injected coolant and mainstream static temperatures equilibrate vii. Kinetic energy dissipation i.e., the turbulent mixing process wherein the injected coolant and mainstream ow velocities equilibrate Reference 5 gives model expressions for the entropy production rate in each of these processes and the Appendix provides a brief description and statement of the parameters which need to be specied. Once the entropy creation terms are known, the mixedout entropy downstream of the blade row can be found. Knowing the total enthalpy and entropy at exit to each stage, the cycle efciency and all other quantities can be determined. It is not essential to express the losses in terms of the entropy created. Horlock et al. 2, for example, use a similar mixing calculation to estimate the mainstream total pressure downstream of the coolant injection, and this provides equivalent information for xing the thermodynamic state. The advantage of working in terms of entropy, however, is that the source and magnitude of each irreversibility is well dened. The effect of changes in the cooling parameters can then be assessed at a local as well as an overall level.

Fig. 1 Schematic diagram of simple-cycle cooled industrial gas turbine

Fig. 2 Schematic diagram of a cooled turbine stage. Complete mixing is assumed at planes 1, 2, 3, and 4.

Table 1 Parameter values for different levels of aeroderivative cooling technology Cooling Technology Current Advanced Super-advanced K cool 0.045 0.045 0.045

int
0.70 0.75 0.80

f 0.40 0.45 0.50

Bimet 0.15 0.15 0.15

Bitbc 0.30 0.40 0.50

Fig. 3 Schematic diagram illustrating the cooling irreversibilities

110 Vol. 127, JANUARY 2005

Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 4 Uncooled GT with poly, C poly, T poly . Contours of constant cycle efciency cyc with contour interval 0.01. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. Dotted lines are loci of T cot,opt T cot for maximum cyc at constant r p .

Cycle Calculations Without Cooling

A set of real gas computer calculations for an uncooled turbine was described in Ref. 3. The calculations showed how varying composition and specic heat capacity result in the occurrence at constant pressure ratio of a maximum cycle efciency at temperatures well below stoichiometric. A mathematical analysis of this behavior was given in Ref. 3 and is not repeated here. A brief description of the results is appropriate, however, because this high-temperature real gas behavior represents the limiting case of zero-cooling ow rate. For the calculations, the air entering the compressor was assumed to be at a temperature of 25 C and a pressure of 1 atm. For convenience, the compressor and turbine polytropic efciencies were taken to be equal, poly, C poly, T poly . The fuel was assumed to be supplied at 25 C and combustor pressure, and had the composition of methane with LHV 50 MJ/kg. A 4% combustor pressure loss was assumed for all calculations, but pressure losses in the inlet and outlet ducts were neglected. Because of these and other minor approximations, the calculated efciencies are a little higher than might be expected, and attention should be directed at changes in efciency rather than absolute values. Contour plots of cycle efciency cyc on axes of combustor outlet temperature T cot and pressure ratio r p are presented in Fig. 4. Figure 4a refers to the case when poly 1.0 reversible turbomachinery. An ideal Joule-cycle analysis with constant specic heat capacities would give straight horizontal contour lines beJournal of Engineering for Gas Turbines and Power

cause cyc would be a function of r p only, independent of T cot . The inclusion of real gas effects in Fig. 4a results in an anticlockwise rotation of the contours, showing that cyc would actually decrease with increasing T cot at xed r p . Thus, although the turbine work output increases with T cot , this is more than offset by an increase in fuel/air ratio. Figure 4b shows contours of constant cyc for turbomachinery with poly 0.95. With respect to Fig. 4a, the contours have rotated clockwise and have moved toward the bottom left of the diagram. This results in cyc passing through a maximum with increasing T cot at xed r p . The locus of these maxima is marked by the dotted line, and it can be seen that the optimum value T cot,opt increases with pressure ratio. For poly 0.9, Fig. 4c shows that the clockwise rotation and movement of the contours is enhanced and the locus of T cot,opt has migrated to higher temperatures. Finally, for poly 0.85, Fig. 4d shows that, for any xed pressure ratio, cyc increases monotonically with T cot and does not display a maximum. These calculations show that, for a gas turbine requiring no cooling, it would only be at low polytropic efciency that a case could be made for continually increasing T cot at constant r p ) as a means of increasing the cycle efciency. Indeed, for an uncooled plant operating at specied r p , T cot , and poly , the optimum design path for increasing cyc is in a direction normal to the contour at that point. At high poly this may, surprisingly, involve decreasing T cot while increasing r p . These ndings are very important. JANUARY 2005, Vol. 127 111

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 5 Cooled GT with current cooling technology from Table 1 and poly, C poly, T poly poly, T refers to an uncooled mainstream expansion. Contours of constant cycle efciency cyc with contour interval 0.01. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. Dotted lines are loci of T cot,opt T cot for maximum cyc at constant r p .

4 Cycle Calculations With Current Cooling Technology

The computer code for cooled gas turbines described in Section 2 was used to calculate a set of contour plots for the four turbomachinery polytropic efciencies exactly as in Section 3. The cooling parameters were set at a level representative of current cooling technology as listed in Table 1 and were maintained constant for all the calculations. Figure 5a shows contours of cyc for a gas turbine with poly, C poly, T poly 1.0. As discussed in Section 2.2, poly, T refers to the efciency of an uncooled turbine expansion. The actual expansion with coolant injection is not isentropic because of the entropy creation associated with the cooling losses. As with the uncooled plots of Fig. 4a, the cycle efciency at constant r p decreases with increasing T cot . In fact, the shape of the contours in Figs. 4a and 5a is very similar, the only real difference being in the magnitude of the efciencies, which are lower by 0.51.5 percentage points in Fig. 5a. Considering the magnitude of the cooling ow fractions, especially at high T cot and r p , this is a remarkably small penalty to pay for cooling the blades. The explanation will emerge below. Figure 5b shows contours of cyc for poly 0.95. A comparison with Fig. 4b shows a more substantial shift of the contours and a larger efciency penalty due to cooling. It will also be noted that the locus of T cot,opt has shifted to lower temperatures by 112 Vol. 127, JANUARY 2005

60100 K and there is evidence of a global maximum cycle efciency. Although this cannot quite be observed in Fig. 4b, examination into the region of higher-pressure ratios reveals a maximum of cyc,max0.555 at T cot,max1690 K and r p ,max63. This combustor outlet temperature is lower than present-day aeroderivative operating temperatures. Figure 5c shows the results with poly 0.9. A comparison with Fig. 4c reveals substantial changes in contour pattern and efciency level. Also, the global maximum has moved to the much higher temperature and lower pressure ratio of T cot,max 1930 K and r p ,max41 with the reduced value cyc,max0.463. This trend continues in Fig. 5d for poly 0.85. The maximum cycle efciency is now cyc,max0.382, and it occurs at the even higher temperature and lower pressure ratio of T cot,max2090 K and r p ,max30. To summarize, the cycle efciency penalty at very high turbomachinery efciency is surprisingly small, even at combustor outlet temperatures and pressure ratios where the cooling ow fractions are very high. As poly drops, much larger differences between cooled and uncooled operation appear, characterized by substantial changes in the cycle efciency contour patterns and the magnitude of the cooling efciency penalty. The changing contour pattern reveals the existence of a global maximum of cyc which migrates from low T cot and high r p at poly 0.95 to much higher T cot and lower r p at poly 0.85. Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 6 Cycle efciency cooling penalty cyc for current cooling technology and poly 0.90. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. At each condition T cot ,rp, the contours represent the difference in cycle efciency between Figs. 4c and 5c.

To set the calculations in context, modern aeroderivative gas turbine power plants have polytropic efciencies around 0.9 a little lower for the turbine and operate on design at around T cot 1750 1850 K and r p 30 36. These conditions are quite close to the global maximum in Fig. 5c. The implications are, i at current levels of turbomachinery efciency and cooling technology, increasing T cot at any pressure ratio would result in only a small increase and possibly even a decrease in cycle efciency and, ii at the current level of cooling technology, an increase in poly should warrant a decrease in T cot .

The Cycle Efciency Penalty Due to Cooling

The cycle efciencies for the uncooled and cooled gas turbines of Figs. 4 and 5 may be subtracted to give the cycle efciency penalty due to cooling cyc . Figure 6a shows contours of cyc for poly 0.90. The penalty increases both with r p at a given T cot and with T cot at a given r p . At the point of maximum cyc ( T cot,max1930 K, r p ,max41), the efciency penalty is 3.7

percentage points. At higher temperatures and pressure ratios, the penalty is much greater, for example at T cot2100 K, r p 50, the penalty is 6.5 percentage points. As the conditions become more extreme, however, the calculations become less reliable because the cooling ows are impracticably high. This can be seen in Fig. 7a, which shows the calculated cooling ow fraction for the rst stator row and in Fig. 7b, which shows the cooling ow fraction for the whole machine including an allowance for disc cooling. A simple method for estimating cyc is given by Horlock 8. Assuming perfect gas behavior, he derived approximate expressions for cyc in one-step, two-step, and multistep cooling processes. For two-step cooling with HP and LP cooling ow fractions H P and L P , Horlocks expression rewritten for polytropic rather than isentropic efciency is c 1
HP

HP

HP LP 1 LP 1 HP LP

(1)

where T cot /Tambient . The compression temperature ratio is

Fig. 7 Fractional cooling owrates for current cooling technology and poly 0.90. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. In a is based on the exit ow from the rst stator and in b on the exit ow from the nal stage.

Journal of Engineering for Gas Turbines and Power

JANUARY 2005, Vol. 127 113

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Table 2 Sensitivity of cooling ow fraction whole turbine and cycle efciency cyc to changes in cooling parameters at constant T cot1800 K, r p 36, and poly 0.90 Cooling parameter Max metal temp rotors and stators 1175 K1125 K1075 K Film cooling effectiveness f 0.50.40.3 Internal cooling efciency int 0.80.70.6 TBC Biot number Bitbc 0.40.30.2 Metal Biot number Bimet 0.10.150.2 Combustion Pattern Factor K comb 0.100.150.20 Cooling ow fraction 0.1980.2580.355 0.2030.2580.318 0.2420.2580.279 0.2440.2580.279 0.2490.2580.269 0.2390.2580.278 Cycle efciency 0.4650.4590.448 0.4650.4590.452 0.4610.4590.452 0.4600.4590.457 0.4600.4590.458 0.4610.4590.457

rm, where m ( 1)/( poly) and r is the pressure ratio over which the coolant stream is compressed from ambient to supply. The expansion temperature ratio is r n , where n poly( 1)/ and r is the pressure ratio over which the coolant is expanded to ambient after mixing in the turbine. Horlock obtained Eq. 1 by considering the difference between the work required to compress the coolant and the work produced by its expansion after mixing with the mainstream ow. If these processes were isentropic and the coolant behaved as a perfect gas with constant c p and , then m and n would be equal and cyc would be zero. For nonisentropic compression and expansion, m and n differ and cyc takes positive values. It is important to appreciate, therefore, that Horlocks analysis ignores the combustor pressure loss, the entropy production due to friction in the internal blade passages, and the kinetic energy dissipation during mixing losses v and vii in Section 2.2. By way of illustration, the cooling ow fractions for poly 0.90, determined in Section 4, were substituted for H P and L P in Eq. 1 to produce estimates of cyc . These are plotted in Fig. 6b for comparison with the full computer calculations of Fig. 6a. Because of the neglect of some sources of loss cyc in Fig. 6b is everywhere smaller than in Fig. 6a. Nevertheless, the slopes of the curves are remarkably similar showing that Horlocks result is successful in explaining part, but not all, of the efciency penalty due to cooling. The preceding analysis and discussion also provides an explanation for the extremely modest uncooled-to-cooled contour shifts at high turbomachinery efciency described in Section 4. As noted above, the approximation to cyc from Eq. 1 tends to zero as poly 1 because the parameters m and n become equal. Hence, with Horlocks assumptions perfect gas behavior throughout the cycle and neglect of the coolant internal friction and kinetic energy mixing losses, cyc would be truly independent of cooling ow fraction at poly 1. Now, although the inclusion of a real gas model has a substantial effect on the absolute level of cycle efciency, the change uncooled-to-cooled is not strongly dependent on the cooling ow fraction. These comparatively minor real gas effects and the further inclusion of the neglected losses are together responsible for the rather small differences in cycle efciency observed between Figs. 4a and 5a. As poly decreases, however, so m and n become signicantly different and cyc as can be seen from Eq. 1 not only increases, but also becomes proportional to the cooling ow fractions. This results in the much larger changes in cycle efciency contour patterns, which are evident at low values of poly .

changes in these parameters can be investigated by differentiating the expression for Eq. A2a in the Appendix to give

K cool 0 f int C K cool 0 C f int 0 C f

Bitbc 0 f Bimet Bimet f 10 1B 1 0 1 B 1B

Bimet 1 f 0 0 C f 1 0 1 0 2 1 B

(2)

where B Bitbc Bimet( 0 f )/(1 0 ) and C 1 int(1 0 ). As an example, consider an HP aeroderivative stator blade operating with a combustor outlet temperature of 1800 K, a combustor pattern factor 0.15, a coolant supply temperature of 875 K, and a maximum outer surface metal temperature of 1125 K. Equations A1 A7 with the current technology values of Table 1 ( int 0.7, f 0.4, Bimet 0.15, Bitbc 0.30) then give 0 0.765 and the required cooling ow fraction for the rst stator row as 0.110. Inserting these values into Eq. 2 gives

K cool 1.21 int 1.94 0.60 f 0.94 Bitbc K cool

1.46 Bimet 13.83 2.55 0 (3) which shows clearly the relative effect of changing each parameter. Of particular interest is the signicant contribution to the coefcient of f from the term involving Bimet the second term in the square brackets. This indicates the importance of allowing for the metal temperature drop, an effect neglected in the original Holland and Thake theory 7. To continue, suppose improvements in cooling technology resulted in an increase of 0.05 in each of int , f , and Bitbc while maintaining K cool and Bimet constant. Equations (A2 a) and (A2 b) then show that could be reduced from 0.110 to 0.088 for the same value of 0 . Alternatively, 0 could be increased from 0.765 to 0.795 with unchanged. For the same allowable metal temperature of 1125 K, this would correspond to an increase in T cot of about 130 K. Calculations such as these provide useful information on the sensitivity of the cooling ow rates to changes in the cooling parameters, but they give no indication of the resulting variations in cycle efciency. It is therefore informative to present the results of a few sample calculations to illustrate the sensitivity of cyc to variations in the cooling parameters. These calculations, it should be noted, have been carried out on the assumption that the compressor and uncooled turbine polytropic efciencies remain unchanged. A reference point was taken at T cot1800 K, r p 36, and poly 0.90, and this was maintained constant. All the cooling parameters except one were then also xed and the remaining Transactions of the ASME

Sensitivity to Variation in the Cooling Parameters

In the preceding sections it has been shown that, as the turbomachinery efciency decreases, the cycle efciency of a cooled gas turbine becomes very dependent on the cooling ow fraction delivered to each blade row and, therefore, on the choice of parameters involved in the estimation of . The sensitivity of to 114 Vol. 127, JANUARY 2005

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 8 Cooled GT with advanced cooling technology from Table 1. Contours of constant cycle efciency cyc with contour interval 0.01. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. Dotted lines are loci of T cot,opt T cot for maximum cyc at constant r p .

parameter varied as shown in the left-hand column of Table 2 with the reference value given in brackets. The corresponding range of cooling ow fractions and cycle efciencies resulting from the variations are tabulated in the other columns. An improvement in cyc of about 0.6 percentage point would result from an increase of 50 K in the allowable metal temperature or from an increase in lm cooling effectiveness from 0.4 to 0.5. Smaller improvements in cyc would be obtained from increasing the internal cooling efciency or TBC Biot number or decreasing the metal Biot number; the combustion pattern factor is not a very signicant parameter. Although these changes in efciency are relatively small, the changes in the cooling ow fraction are more substantial. However, it is important to note that none of the changes in Table 2 would match the improvement resulting from an increase of one percentage point in the compressor and uncooled turbine polytropic efciencies. This would increase cyc by 1.5 percentage points from 0.459 to 0.474 the cooling ow fraction dropping from 0.258 to 0.250.

Cycle Calculations With Improved Technologies

7.1 Improved Cooling Technologies. In Sections 3 and 4, the contour plots of cycle efciency for an uncooled gas turbine Journal of Engineering for Gas Turbines and Power

Fig. 4 and for a cooled gas turbine with current cooling technology Fig. 5 were examined in detail. It is now of interest to see how improvements in cooling technology result in a gradual transition from the plots of Fig. 5 to those of Fig. 4. To this end, two improved technologies, dened in Table 1, were investigated. The advanced cooling technology features modest improvements which might be representative of the next generation of aeroderivative machines. Typically, such improvements produce about a 25% reduction in cooling ow fraction at a given T cot and r p compared with current technology values. A superadvanced cooling technology is also dened in Table 1. These parameters provide about a 50% reduction in cooling ow fraction compared with current technology and, therefore, represent a major advance in cooling development. Figures 8 and 9 show the contour plots for the two improved technologies. Consider rst the calculations for poly 1.0. The most instructive way of examining the plots is to observe the shift of the contours through the sequence Figs. 5a8a9a4a as the technology improves from its current level to the ultimate uncooled case. Clearly, only very small changes in cycle efciency result from these improvements. This reinforces the conclusion that, at very high poly , cyc is almost independent of cooling ow fraction. JANUARY 2005, Vol. 127 115

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 9 Cooled GT with super-advanced cooling technology from Table 1. Contours of constant cycle efciency cyc with contour interval 0.01. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. Dotted lines are loci of T cot,opt T cot for maximum cyc at constant r p .

For poly 0.95, the plots are best examined in the reverse sequence Figs. 4b9b8b5b. Although the changes are not dramatic, the curvature of the contours clearly increases, culminating in the formation of the global maximum of cyc already noted in Fig. 5b at T cot,max1690 K, r p ,max63. More importantly, there is little variation in the locus of T cot,opt , particularly at low r p . In the unlikely event that turbomachinery polytropic efciencies reach values of 0.95, the optimum T cot at given r p would be largely independent of the cooling technology and would not exceed 1900 K, even at pressure ratios of 60. poly 0.90, the forward sequence Figs. For 5c8c9c4c shows the locus of T cot,opt moving to regions of higher temperature, the global maximum of cyc increasing in both temperature and pressure ratio. Similar comments poly 0.85, and the sequence Figs. apply for 5d8d9d4d. These observations indicate that, for improved cooling technology but essentially constant poly of around poly 0.9 Figs. 5c8c9c, an increase in both T cot and r p might be justied by the resulting increase in cyc . However, with improvements in both cooling technology and poly , an increase in r p is more important than an increase in T cot . Indeed, depending on the 116 Vol. 127, JANUARY 2005

level of improvement in poly , the optimum combustor outlet temperature may actually decrease as the cooling technology gets better Figs. 5c8b9b. 7.2 Ultimate Turbomachinery Efciency. The theoretical behavior for poly 1.0, although providing understanding and interpretation, can obviously never be realized in practice. Indeed, even values of 0.95 probably exceed the maximum achievable turbomachinery efciency, which, according to most authorities, is likely to be around poly 0.92 0.93 for both compressors and turbines. Hence, in an attempt to represent the ultimate realizable technology for turbomachinery aerodynamic design, Fig. 10 presents contour plots for the four levels of cooling technology discussed previously, all calculated with poly 0.925. It shows clearly the gradual transition, at constant poly 0.925, from current cooling technology a, through advanced b and superadvanced c technologies, to a hypothetical uncooled machine d. Figure 10a shows a global maximum cycle efciency of 0.507 at r p 50, T cot1810 K, compared with 0.463 at r p 41 and T cot1930 K for poly 0.90 Fig. 5c. This emphasises the imTransactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 10 GT with the ultimate turbomachinery efciency of poly 0.925. Contours of constant cycle efciency cyc with contour interval 0.01. Combustor outlet temperature T cot1400 2200 K on abscissa, compressor pressure ratio r p 24 60 on ordinate. Dotted lines are loci of T cot,opt T cot for maximum cyc at constant r p .

portant point made in Section 4 that the optimum T cot for current cooling technology falls as the turbomachinery becomes more efcient. Figures 10b 10d are remarkable for the similarity of the contour patterns and the fact that the locus of T cot,opt is almost independent of the level of cooling technology. Furthermore, it will be noted that the contour lines for high T cot run almost parallel to lines of constant r p indicating a very weak dependence of cyc on T cot for constant r p . Thus, if the gas turbine designer, having achieved ultimate turbomachinery polytropic efciency of 0.925 and chosen a pressure ratio of 50 with a combustion temperature of 1810 K the optimum conditions of Fig. 10a, were prepared to compromise by, say, 0.5% on cycle efciency, the required combustor outlet temperature would fall by some 200 K.

Summary and Conclusions

The calculations presented in this paper show that, just as for uncooled machines, the cycle efciency of a cooled gas turbine is primarily dependent on the combustor outlet temperature T cot , the pressure ratio r p , and the turbomachinery polytropic efciency poly . The operating condition for maximizing cycle efciency depends, however, on a subtle combination of these parameters. Using a computer code incorporating accurate modeling of gas properties and the cooling methodology of Young and Wilcock Journal of Engineering for Gas Turbines and Power

5, the paper has explored the variation in cycle efciency for a range of turbomachinery aerodynamic efciencies and cooling technologies. In an attempt to summarize the results in as digestible a form as possible, Fig. 11 presents a montage of the 20 contour plots comprising Figs. 4, 5, 8, 9, and 10. A study of the trends exhibited in this gure clearly shows that, contrary to widespread opinion, improved cycle efciencies will not necessarily result from increased combustor outlet temperatures. These ndings should lead the designer to search for a suitable compromise between increased combustor outlet temperature and improved turbomachinery efciency. If high poly is not possible, then it is logical to seek a higher T cot . However, if high poly can be attained, then the search for high T cot may not be appropriate. At very high unattainable poly , the cycle efciency is almost independent of the level of cooling technology. Even at poly 0.925 the generally accepted ultimate aerodynamic efciency, the value of T cot giving maximum cyc for a given r p is, not only remarkably insensitive to the cooling ow fraction, but also comparable to, or even lower than, present-day operating temperatures. Finally, it must be emphasised that this study has concentrated on gas turbine performance in terms of cycle efciency only. In some circumstances there may be other valid reasons for increasing the combustor outlet temperature such as the desire to inJANUARY 2005, Vol. 127 117

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

Fig. 11 Contours of constant cyc contour interval0.01 with poly, C poly, T poly . Each plot has T cot 1400 2200 K on abscissa, r p 24 60 on ordinate. Dotted lines are loci of T cot for maximum cyc at constant r p . Left to right, increasing cooling technology. Bottom to top, increasing turbomachinary poly .

crease the specic work output which outweigh efciency considerations. It is also appreciated that other limits, such as those on emissions, may have to be imposed which will restrict increases in combustion temperature beyond those in use at present.

Engineering Department. The authors are grateful to A. AramayoPrudencio for his help in preparing the contour plots.

Nomenclature
Bi K comb K cool p0 rp T cot Biot number combustion pattern factor cooling ow factor total pressure compressor pressure ratio combustor outlet temperature Transactions of the ASME

Acknowledgments
R.C.W. gratefully acknowledges the receipt of an EPSRC studentship, a Rolls-Royce CASE award, and a maintenance contribution from the Ford of Britain Trust of the Cambridge University 118 Vol. 127, JANUARY 2005

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

T 0 total temperature 0 blade cooling effectiveness f lm cooling effectiveness cyc cycle efciency int internal cooling efciency poly turbomachinery polytropic efciency cooling ow fraction ( m c / m g )

the single parameter K cool of Eq. A3 which was tuned to engine data and then xed for all the calculations. Returning to Eq. A2a, the internal cooling efciency int is dened by

int

T 0 c ,exit T 0 c ,in T met,int T 0 c ,in

(A4)

cyc is based on the net work output, uncorrected for mechanical and generator efciencies, and on the LCV of methane. poly is a total-to-total polytropic efciency. All states in the cycles are stagnation states. The maximum cycle temperature is the combustor outlet temperature not the rotor inlet temperature.
Subscripts. C c g T compressor coolant mainstream gas turbine

where T 0 c ,exit is the relative total temperature of the coolant at exit from the internal blade passage and T met,int is the internal surface metal temperature. int represents the quality of the internal cooling technology. It can, as discussed by Chiesa and Macchi 11, be related to the internal heat transfer coefcient and the surface area of the internal ow passages, but this is deliberately not done in the present paper. Instead, in order to bypass a difcult internal heat transfer calculation, a value of int is specied directly as input data. The lm-cooling effectiveness f is dened by f T recov T adwall T 0 g T adwall T recov T 0 c ,exit T 0 g T 0 c ,exit (A5)

Appendix
Determination of the Cooling Flow Rates. Various approaches for estimating the required cooling ow fraction the ratio of the coolant mass owrate m c to the mainstream owrate m g ) are described by Holland and Thake 7, El Masri 9, Consonni 10, and Chiesa and Macchi 11. The Holland and Thake approach is attractive because it expresses in terms of a small number of parameters representive of a technology level. The method was extended by Young and Wilcock 5 to allow for the temperature drop through the thermal barrier coating TBC and blade metal. The analysis starts by dening the blade cooling effectiveness 0 by T 0 g T met,ext 0 T 0 g T 0 c ,in (A1)

where T recov is the mainstream recovery temperature approximately equal to T 0 g ) and T adwall is the adiabatic outer wall temperature. Equation A2a requires a surface-area-averaged value of f , and this is difcult to estimate because it depends on the geometry and layout of the injection holes, the blowing ratio, and other parameters. In order to avoid this problem, a value of f representative of current lm cooling technology was specied directly. When lm cooling was employed, it was assumed that all the air emerges from the cooling holes. The nal parameters in Eqs. A2a and A2b are the metal and TBC Biot numbers dened by Bimet h g t met , met Bitbc h g t tbc tbc (A6)

where T 0 g is the relative total temperature of the mainstream gas, T 0 c ,in is the relative total temperature of the coolant entering the blade passages, and T met,ext is the allowable external surface metal temperature assumed constant over the blade. The problem is to nd the value of for each blade row which will ensure that the required value of 0 is achieved. An expression for is derived in Appendix 2 of 5. The result can be expressed as

where t is the thickness and is the thermal conductivity of the metal or TBC. If T 0 g in Eq. A1 corresponds to the mass-averaged total temperature, the predicted coolant ow rate is invariably lower than that used in a real engine. This is because allowance must be made for temperature nonuniformity in the combustor outlet ow. A common way of doing this is to replace T 0 g by a temperature T 0 g ,max given by T 0 g ,max T 0 g K comb T comb (A7) where T comb is the combustion temperature rise and K comb is a pattern factor, depending on the type of combustor and the position of the blade row Kawaike et al. 12. Typically, K comb is about 0.15 for the rst stator, reducing progressively for subsequent cooled rows. Having obtained by this method, the calculations revert to using the mass-averaged T 0 g . Determination of the Cooling Irreversibilities. Figure 3 is a schematic diagram showing the coolant and mainstream ow paths. The Holland and Thake analysis and energy equation provide values of the coolant ow fraction, heat transfer rate and temperatures dened earlier in this appendix. Process i in Section 2.2 refers to entropy creation due to heat transfer to the outer blade surface. As shown in Ref. 5, this can be written in terms of the mainstream static and blade surface temperatures. The former requires knowledge of the average mainstream Mach number, but the value is not critical. Processes ii and iii refer to entropy creation due to conduction through the TBC and blade metal and can be computed from the known heat transfer rate and temperature drops. Processes iv and v refer to entropy creation in the coolant stream between compressor offtake and injection into the mainstream. At compressor offtake, T 0 c ,comp and p 0 c ,comp are prescribed and T 0 c ,exit is known from the heat transfer analysis. If p 0 c ,exit can be found, this denes the specic entropy at the exit holes and hence xes this internal loss. JANUARY 2005, Vol. 127 119

mc K cool 0 f 1 int 1 0 mg 1B int 1 0 B Bitbc

(A2a)

0 f Bimet 10

(A2b)

In Eqs. A2a and A2b, K cool is the cooling ow factor, int is the internal cooling efciency, f is the lm cooling effectiveness, Bimet is the metal Biot number, and Bitbc is the TBC Biot number. The cooling ow factor K cool is dened by K cool 1 A surf c pg St A g c pc g (A3)

A surf is the blade surface area, and A g is the exit ow area normal to the axial direction. c pg / c pc is the gas to coolant specic heat ratio, and St g is a Stanton number dened by St g h g A g / m g c pg ( h g being the gas to blade surface mean heat transfer coefcient. The factor allows for other primary cooling ows mainly endwall cooling and is roughly equal to the ratio of the end-wall area to A surf . It does not encompass secondary cooling ows for sealing, etc. For convenience, all the parameters were combined into Journal of Engineering for Gas Turbines and Power

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms

One way of nding p 0 c ,exit is to estimate the pressure loss in the internal passages using the method of Consonni 10. This, however, requires detailed knowledge of the internal ow. An alternative approach is to assume that the design of the passages has been such as to achieve a certain momentum ux blowing ratio I at the holes. I is dened by I
2 c ,exitV c ,exit 2 g ,exitV g ,exit

References
1 MacArthur, C. D., 1999, Advanced Aero-Engine Turbine Technologies and Their Application to Industrial Gas Turbines, ISABE 14th Int. Symp. on AirBreathing Engines, Florence, Paper 99-7151. 2 Horlock, J. H., Watson, D. T., and Jones, T. V., 2001, Limitations on Gas Turbine Performance Imposed by Large Turbine Cooling Flows, ASME J. Eng. Gas Turbines Power, 123, pp. 487 494. 3 Wilcock, R. C., Young, J. B., and Horlock, J. H., 2002, Gas Properties as a Limit To Gas Turbine Performance, ASME Turbo-Expo 2002, Amsterdam, Paper GT-2002-30517. 4 Young, J. B., and Wilcock, R. C., 2002, Modelling the Air-Cooled Gas Turbine: Part 1-General Thermodynamics, ASME J. Turbomach., 124, pp. 207 213. 5 Young, J. B., and Wilcock, R. C., 2002, Modelling the Air-Cooled Gas Turbine: Part 2-Coolant Flows and Losses, ASME J. Turbomach., 124, pp. 214 221. 6 Chase, Jr., M. W., Davies, C. A., Downey, Jr., J. R., Frurip, D. J., McDonald, R. A., and Syverud, A. N., 1986, JANAF Thermochemical Tables, Third Edition, American Institute of Physics, New York. 7 Holland, M. J., and Thake, T. F., 1980, Rotor Blade Cooling in High Pressure Turbines, J. Aircr., 17, pp. 412 418. 8 Horlock, J. H., 2001, Basic Thermodynamics of Turbine Cooling, ASME J. Turbomach., 123, pp. 583592. 9 El-Masri, M. A., 1987, Exergy Analysis of Combined Cycles: Part 1, AirCooled Brayton-Cycle Gas Turbines, ASME J. Eng. Gas Turbines Power, 109, pp. 228 235. 10 Consonni, S, 1992, Performance Prediction of Gas/Steam Cycles for Power Generation, Ph.D. thesis no. 1983-T, MAE Dept. Princeton Univ. 11 Chiesa, P., and Macchi, E., 2002, A Thermodynamic Analysis of Different Options to Break 60% Electric Efciency in Combined Cycle Power Plant, ASME Turbo-Expo 2002, Amsterdam, Paper 2002-GT-30663. 12 Kawaike, K., Kobayishi, N., and Ikeguchi, T., 1984, Effect of Blade Cooling System With Minimized Gas Temperature Dilution on Gas Turbine Performance, ASME J. Eng. Gas Turbines Power, 106, pp. 756 764. 13 Hartsel, J. E., 1972, Prediction of Effects of Mass-Transfer Cooling on the Blade-Row Efciency of Turbine Airfoils, AIAA, 10th Aerospace Sciences Meeting, San Diego, Paper AIAA-72-11.

2 cM c ,exit 2 gM g ,exit

(A8)

where M c ,exit is the exit coolant Mach number and M g ,exit is the local mainstream Mach number. As shown in Ref. 5, specifying M g ,exit and I is sufcient to dene p 0 c ,exit if it is assumed that the coolant and gas static pressures at the holes are equal. This method of estimating the internal loss parallels the calculation of the internal heat transfer described earlier in this appendix. There, the difcult problem of estimating the internal heat transfer coefcient was avoided by the direct specication of the parameter int . Here, the problem of estimating the mean friction factor is avoided by specifying the parameter I. Finally, it is necessary to calculate the entropy creation due to the mixing processes vi and vii. As explained in Ref. 5. This is achieved using the method of Hartsel 13 under the assumption that the static pressure remains constant during mixing. The only extra parameter required to perform this calculation is the injection angle of the coolant . The advantage of this approach is that the heat transfer analysis together with the specication of just three parameters M g ,exit , I and allows the estimation of all the cooling irreversibilities. The disadvantage is in the difculty of choosing suitable average values for the parameters in order to provide accurate representations of cooling systems which are invariably very complex.

120 Vol. 127, JANUARY 2005

Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 10/18/2013 Terms of Use: http://asme.org/terms