good

Applied Energy 87 (2010) 1207–1216

Contents lists available at ScienceDirect

Applied Energy

journal homepage: www.elsevier .com/locate /apenergy

Evaluation of water injection effect on compressor and engine performanceand operability

I. Roumeliotis *, K. MathioudakisLaboratory of Thermal Turbomachines, National Technical University of Athens, Iroon Polytechniou 9, Athens 15773, Greece

a r t i c l e i n f o a b s t r a c t

Article history:Received 21 January 2009Received in revised form 27 April 2009Accepted 28 April 2009Available online 24 May 2009

Keywords:Gas turbineWater injectionSteam injectionOversprayWet compression

0306-2619/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.apenergy.2009.04.039

* Corresponding author. Tel.: +30 2107721633; faxE-mail addresses: [email protected], jroume@cen

Gas turbine performance enhancement technologies such as inlet fogging, combustor water/steam injec-tion and overspray are being employed by users in recent years without fully evaluating their effect ongas turbine performance and operability. The water injection techniques can significantly affect theengine operating point thus a careful analysis should precede the application of performance enhance-ment devices, especially when the devices are retrofitted to old engines or engines operating at extremeconditions. The present paper examines the most widespread techniques that implement water injectionby using in-house models that can reproduce the effects of water injection on the gas turbine and com-pressor off-design operation. The results are analyzed with respect to both performance augmentationand engine operability in order to give further insight on gas turbine operation with water injection.The behaviour of the gas turbine is interpreted while the risks on engine integrity due to water injectionare identified.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Deregulation in power generation market combined with thesignificant variation in fuel prices and the need for flexibility con-cerning the power generation has forced the power producers toexplore new technologies for power generation enhancement.The increasing use of gas turbines in the power generation industrycreated an additional incentive for further improvement of theirperformance. In the last years several techniques have been pro-posed for gas turbine power and efficiency augmentation withthe use of water. The effect of these technologies on the opera-tional integrity of the gas turbine is often overlooked by the usersas it is seldom referenced and analyzed on technical papers. None-theless these techniques when installed as add-ons on existing gasturbines can significantly affect the compressor stability. Apartfrom the change on the engine operability the effect that the userwill experience is gas turbine dependant. The magnitude of theperformance gain depends thus on the particular gas turbine char-acteristic and its control. The scope of this paper is to analyze theeffect of the most widespread water injection techniques to the en-gine operability and to address issues of interest for the gas turbineusers by using well established models along with knowledgeavailable in the bibliography.

Steam or water injection into the combustion chamber has beenproposed and extensively used for NOx reduction [1] and perfor-

ll rights reserved.

: +30 2107721658.tral.ntua.gr (I. Roumeliotis).

mance augmentation such as in the case of the GE LM5000 as de-scribed by Burnham et al. [2]. More complex evaporative gasturbine cycles have been proposed in recent years such as theRWI cycle and the HAT cycle while a throughout review of thesecycles has been undertaken by Jonsson and Yan [3]. In most casesthis kind of cycles are modelled using simple thermodynamic mod-els [4] without taking into consideration the effect of steam/waterinjection on gas turbine off – design performance and operability,although the shift of the surge margin has been recognized as aproblem for the engine integrity by Jonsson and Yan [3].

The fact that gas turbine output and efficiency drop during highambient temperature periods, when demand usually increases, hasled to the broad application of inlet air cooling. For the treatment ofthe gas turbine inlet air several techniques are used such asabsorption chillers, mechanical chillers using freon as refrigerantand the evaporative cooling. These options are discussed thor-oughly by Ondryas et al. [5]. In recent years as the spraying tech-nologies evolve, direct inlet air cooling with water injection hasfound a broad application as discussed by Jonsson and Yan [3] inboth open cycle [6] and combined cycle [7] configurations. As inletair cooling is easily installed and of low cost [8,9] its application isincreasing on existing engines. GE had incorporated inlet coolingand intercooling between the compressors using water injectionto the twin spool engine LM6000 [10]. In these applications thespraying quantity is limited so there is no carry over of dropletsin the compressor. In analyzing inlet air cooling either design pointthermodynamic analysis is used [11], or commercial software typ-ically scaled with respect only to the design point is used [6]. As

Nomenclature

CDP compressor discharge pressureCDT compressor discharge temperatureDd droplet diameterEGT exhaust gas temperaturefar fuel to air ratio, by massLoad gas turbine Powerm mass flowN rotational speedp pressureRH relative humiditySM surge marginT temperatureTIT turbine inlet temperature

warliq liquid Water to dry air ratio, by masswfr water to fuel ratio, by massDi % change = (i � i,ref)/i,ref � 100g stage efficiencygth thermal efficiencypC compressor pressure ratioU mass flow coefficientW pressure rise coefficient

Subscriptsamb ambient conditionsref reference valuesf fuel

1208 I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216

the off-design operation of the gas turbines is not fully repre-sented, the results do not reproduce the specific gas turbine behav-iour for the whole compressor inlet temperature (CIT) range andalthough a general guideline can be deducted the effect of inletair cooling is not fully modelled.

In order to fully examine the effect of water injection for inletair cooling and for NOx reduction the off – design operation ofthe gas turbine should be modelled. For this purpose a generic per-formance/adaptive model (TEACHES) was developed by the re-search group of the authors (LTT/NTUA) capable of modelling theoff – design operation of different engine configuration with andwithout water injection [12,13]. This model will be used in this pa-per in order to examine various aspects of gas turbine off-designoperation with water injection.

Another technique that implements water injection for gas tur-bine performance augmentation is overspray. Overspray of waterinside the compressor is a technique proposed since the late1940s, early 1950s by NACA, for the augmentation of aircraft thruststarting with the work of Trout [14], while in the early 1960s Hill[15] presented a rigorous thermodynamic analysis of wet compres-sion analyzing the possible gain on compression work. The level ofspraying technology at the time along with the fact that other tech-niques were used for the engine performance augmentation re-sulted in abandoning the wet compression idea. Wetcompression was proposed again in 1997 by Utamura et al. [16]by presenting measurements from a 130 MW engine. The resultsindicated that overspray can produce a significant gain on bothpower and engine efficiency. Utamura et al. [17] presented addi-tional results in 1999 from a F9E engine that further supportedthe benefits overspray can give. These results increased the interestconcerning overspray and a plethora of studies have been pre-sented in the recent years.

In most of the overspray studies evaporation is assumed in ther-modynamic equilibrium, as for example in the work by Wang andKhan [18] and Bhargava and Meher-Homji [6]. The results pre-sented by Härtel and Pfeifer [19] indicated that the assumptionof evaporation in thermodynamic equilibrium will result to a sig-nificant overestimation of the benefits of wet compression. Whiteand Meacock [20] analysis of wet compression indicated that theentropy increase due to phase change should not be neglected asdone in the literature as for example by Sexton and Sexton [21]and Klepper et al. [22]. Concerning the compressor operation withoverspray Hill [15] discussed the possibility of stage rematchingand consequently compressor map shift, a phenomenon furtherdiscussed by Horlock [24] and Ludorf et al. [23].

Thus in order to examine the effect of overspray the stagerematching should be taken into consideration along with the fact

that the evaporation occurs from the droplets surface modellingevaporation as a non- isentropic process. One dimensional com-pressor models taking these aspects of water injection into consid-eration have been presented by White and Meacock [20] andRoumeliotis and Mathioudakis [25].

The model developed by the authors [25] will be used herein inorder to examine in depth the effect of overspray on the engineoperability and analyze aspects that are of interest to the gas tur-bine users.

2. Engine model

In order to have the ability to examine the operation of a gasturbine of different configurations under various operating condi-tions and control variables, a generic performance model (TEA-CHES) was developed by the research group of the LTT/NTUA.Further information of what constitutes a performance modeland an outline of the procedure for building one has been pre-sented by Roumeliotis et al. [26]. The model was created in orderto simulate any gas turbine configuration. Specifically it can handleup to three spool engines with power turbine, with reheating andrecuperation and water/steam injection without droplet carry-overat the turbomachinery components at various positions along theengine. It has the ability to adapt to available data by employingthe adaptive modelling technique as introduced by the researchgroup of the authors [27] while several parameters can be set ascontrol variables like TIT, CDP, load, etc.

The basic idea behind the adaptive modelling is that componentcharacteristics (e.g. compressor map) are allowed to changethrough appropriate modification factors. The values of these fac-tors are determined by requiring that available engine perfor-mance data are matched by the engine model. In the end of theadaptive procedure a unique set of components characteristics isproduced in order to reproduce the specific engine operation withgreat accuracy [12]. When diagnostic applications on an actual en-gine are considered, the modification factors can be used as enginehealth indices as presented by Aretakis et al. [28].

The way that the water injection is handled and the way thatthe model can be implemented to specific engines with availabledata in order to reproduce its operation has been discussed in de-tail by Roumeliotis et al. [26]. In brief gas – water mixing is calcu-lated through the application of the conservation laws ofcontinuity, energy and momentum. The effect of the presence ofwater vapor on the performance maps is taken into considerationwith the use of correction methods reported by AGARD [29], asthere is no carry – over of water in the turbomachinerycomponents.

Fig. 1. Measurements and model predictions for compressor discharge pressure.

Fig. 2. Measurements and model predictions for gas turbine efficiency with (FOW)and without (FOD) water injection.

I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216 1209

2.1. Engine model for overspray

Overspray results to a significant stage rematching altering thecompressor map from the one with ‘‘dry” operation thus the use of‘‘dry” maps to a model as the one described above is not suitable. Inorder to evaluate compressor performance with water injection atthe inlet or intermediate positions a ‘‘stage stacking” approachcoupled with a droplet model has been materialised as describedby the authors [25]. The stage stacking method was selected as,apart from overall performance quantities it allows the predictionof overall compressor stall limits by incorporating criteria of indi-vidual stage stall, allowing overall compressor stall and chock lim-its evaluation for a specified rotating speed. The specific modelincorporates an adaptivity feature allowing the derivation of indi-vidual stage characteristics, which optimally reproduce a givenoverall map as described by Mathioudakis and Stamatis [30]. Thedroplet model embedded in the stage stacking method is the mod-el described by Spalding [31].

The calculations are done assuming the droplets to be entrainedin the flow, so there is no velocity slip between them and the gas-eous medium. White and Meacock [20] referenced that the dropletdiameter can be up to 5 lm for the no slip assumption to be rela-tively valid. This assumption was validated by Roumeliotis [32] byincorporating the Lagrangian equations of movement for dropletsof various diameters on an axial compressor flow field. The entropyincrease due to evaporation is calculated by using the analyticalequation proposed by Young [33] which as examined by Roumeli-otis [32] give accurate results.

The compressor stage performance coefficients (U, W, g) are as-sumed unchanged due to droplet presence, as experimental datapresented by Roumeliotis and Mathioudakis [34] suggest thatwater injection of quantities up to 2% by mass without evaporationinside the stage is not expected to change significant the U �Wcharacteristic, nor the aerodynamic behaviour of the stage.

Overspray affects both compressor and turbine operation, thusaltering the components coupling. In order to study the oversprayeffect on the engine performance the compressor model is coupledto the engine model. Specifically the compressor model is replacingthe compressor map by applying a 1-D zooming procedure. Detailsconcerning the zooming procedure are presented in [25].

3. Analysis of water injection cases

The models presented are used to analyze the operation of gasturbines implementing water injection technologies. The resultsare analyzed and discussed in conjunction with the available bib-liographic data in order to give further insight on gas turbine oper-ation with water injection.

3.1. Combustor water injection

Combustor water injection is a mature technology that hasfound broad application by engine manufacturers such as in thecase of the GE LM family [10]. Due to emissions regulations inde-pendent users are also installing water injection systems on exist-ing gas turbines [35]. When a user is installing a combustor waterinjection system, the shift of the whole gas turbine operationshould be examined. In order to study such effects the generic en-gine model discussed above was adapted to a specific engine (ABBGT13-E2). Measured data for operation with natural gas (NG) fueloil dry (FOD) and fuel oil with water injection (FOW) were used.The results indicated that the adapted model can reproduce the en-gine operation and behaviour within a 1% error with respect to themeasurements for a broad operation envelope and for differentfuel types with and without water injection as can be seen in Fig. 1.

In Fig. 2 the engine thermal efficiency predicted and calculatedfrom measurements can be seen. As expected, burner water injec-tion results to a power increase and to engine efficiency reduction.The negative effect on the engine efficiency is decreasing for lowpower operation.

Having established a reliable model for the specific engine, theeffect of water injection on engine operation can be examined.Water injection leads to an increase of the compressor dischargepressure which is primary a result of the increased turbine massflow and secondarily of the gas properties change, as analyzed byMathioudakis [36].

This increase causes a surge margin reduction with respect tothe dry operation. Surge margin is defined via Eq. (1), accordingto Walsh and Fletcher [37]

SMð%Þ ¼ pC;surge � pC;working

pC;working� 100% ð1Þ

In Fig. 3 the surge margin reduction due to the increased waterto fuel ratio is presented for operation under constant TIT or con-stant load. The reduction can be prohibitive for wfr in the rangeof 3 as a 5% value is rather slim [38]. For values encountered inpractice (wfr:1–2) the reduction is permissible, as the surge marginreduces from 15% to 8,6% for wfr = 2. In the case of steam injectionthe behaviour is similar, while the steam to fuel ratio where thesurge margin is reduced below 5% is 3,6 (constant TIT). In any caseit is evident that the wfr should be limited by the surge marginreduction. An important observation is that if the firing tempera-ture is reduced to maintain constant load the effect of water injec-tion on the surge margin is rather small and is not expected tocause any operation problems.

Fig. 5. Gas turbine load variation for varying wfr for operation under constant TIT.Fig. 3. Surge margin reduction due to combustor water injection.

1210 I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216

The effect on overall engine efficiency can be seen in Fig. 4. Theeffect of water injection is more pronounced when the engine isworking at fixed load. In the case of fixed TIT, the engine is workingat higher load and pressure ratio; hence the effect on thermal effi-ciency is not as negative as in the case of constant load. In the caseof steam injection, the latent heat for evaporation is available froman outside source, (for example using an exhaust gas heat exchan-ger) resulting to the increase of the gas turbine efficiency byincreasing the steam injection quantity. The effect of steam injec-tion is more pronounced for operation under constant TIT.

In both steam and water injection, the increased turbine massflow results to a significant increase on engine load for operationunder constant TIT as presented in Fig. 5. This means that whenthe engine is working in high torque conditions and water/steamis injected in the combustor there is a torque surplus. Mechanicalstrength of the shaft and the electrical generator capacity shouldbe taken into consideration in such cases. As can be seen in Fig. 5the operation with water injection during a cold day can resultto an increase of power up to 50% from the nominal one, thusstressing more the mechanical integrity of the engine. It must benoted that results presented by Fischer et al. [39] indicate thatfor a specific engine the axial thrust will not be a problem forpower increase more than the 100% of the nominal one.

Concerning the control variable, it should be noted that the en-gines are usually working with constant TIT (subject to limit). Thisoperation will result to higher performance gain in the case ofsteam injection and to smaller loss at thermal efficiency in the caseof water injection but it will also result to a higher loss of surgemargin which may result to compressor unstable operation. The

Fig. 4. Gas turbine thermal efficiency variation for varying wfr for operation underconstant load and constant TIT.

users should also consider that with water/steam injection the syn-thesis of the gas flow is changing. This change will result to highermetal temperature at the turbine blades as discussed by Jin et al.[40] and result to life limiting events at the hot section of the en-gine if the firing temperature remains unchanged. This is not ex-pected to be a problem for constant load operation as the firingtemperature is reduced for the same load with water/steaminjection.

3.2. Inlet air cooling with water injection

The gas turbine output and efficiency drop during high ambienttemperature conditions, when demand usually increases, results tothe broad application of inlet air cooling with water injection(chilling for brevity). Chilling decreases the compressor inlet tem-perature and increases the humidity fraction of the air sucked bythe engine. The temperature reduction results to the increase ofpower and efficiency, while the humidity increase results to apower increase but to an efficiency decrease. The temperature ef-fect is greater than the humidity effect resulting to the improve-ment of the engine performance with inlet air cooling. The effectof ambient temperature and humidity on the thermal efficiencyof ABB GT13E2 engine model used can be seen in Fig. 6 for opera-tion under constant TIT.

The gain that the user will experience strongly depends on theambient effect on the gas turbine performance and behaviour andon the temperature decrease that can be achieved. The tempera-ture decrease is a function of ambient conditions, mainly of tem-perature and humidity and the Mach number at the engine face.For these reasons in order to have a clearer idea about the effectof chilling, different engines of different configurations, power set-tings and technology level were examined. All the engine modelswere created by adapting the generalized engine model to avail-able data. The data used was either measurements if available[36] or data found in the open literature [26]. In Table 1 the exam-ined engines are presented. In all cases the engine control variablewas selected to be the turbine inlet temperature (TIT) as it is themost common and the fuel was selected to be typical natural gas.

In Fig. 7 the behaviour of the various engines versus ambienttemperature variation is shown. The behaviour of each engine isdifferent. It can be said that the single shaft configurations are lessinclined to efficiency degradation due to ambient temperature in-crease as expected [41,6].

In Fig. 8 the gain on thermal efficiency by using chilling is pre-sented for the specific engines, assuming dry ambient air. The anal-ysis taken is for the case that inlet air is becoming saturatedaccording to the compressor face static conditions and there isno droplet carry-over inside the compressor. The efficiency gainwith inlet air cooling can be significant for ambient temperatures

Fig. 6. Effect of ambient temperature and humidity on gas turbine efficiency.

Table 1Examined engines.

Case Configuration Engine

1 Single shaft ABB GT 82 Single shaft GEC EM610B3 Single shaft GE PG9171E4 Single shaft ABB GT 13E25 Twins shaft GE LM 25006 Twin shaft ABB GT 107 Twin spool with power turbine RR Tyne8 Twin spool with power turbine RR Olympus

Fig. 7. Effect of ambient temperature on efficiency for different engines,RHamb = 0%.

Fig. 8. Effect of inlet cooling on gas turbine efficiency for varying ambienttemperature and different engines, RHamb = 0%.

I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216 1211

greater than 30 �C. As aero derivative machines are more sensitiveto ambient temperature, they benefit significantly from inlet aircooling thus the gain is greater for multi-shaft engines an observa-tion that comes in agreement with the results presented by Bharg-ava and Meher-Homji [6].

In order to examine an actual case, the adapted model for theGEC EM610B engine was used to reproduce its operation underthe ambient conditions that occur at a sea-side power plant site,for the summer months (15/6–15/9). The engine is assumed tooperate under constant TIT for both chilling and normal operation.The efficiency gain during actual summer conditions can be seen inFig. 9 in conjunction with the measured ambient temperature andhumidity. It is evident that the gain can be rather high for most ofthe summer period. The mean gain is up to 1% with a significant

gain of 2–2.5% during very hot days (�40 �C). What is more impor-tant is that during very hot days when the demand for power ishigh inlet cooling can give a solution as the power gain can be ashigh as 8% for the hottest days (Fig. 10).

The effect of chilling on compressor operability is mainly due tothe movement on the engine operating line towards lower CIT thusto higher corrected rotational speed. This effect along with the ef-fect of increased humidity results to a surge margin reductioncompared with the one that the user would witness when operat-ing without chilling for the same ambient conditions. In order toquantify this effect the surge margin of the compressor for ambienttemperature 30 �C and for varying ambient RH is presented inFig. 11. The surge margin for the case of 30% RH and 30 �C willbe reduced by �1.5% for chilling operation. From these values itcan be said that chilling will not decrease the surge margin enoughto risk the compressor integrity, although it can be a contributingfactor if the surge margin is already slim. It must be noted thatthe compressor will experience a similar behaviour with every in-let air cooling technique, such as refrigerant cooling, as the humid-ity effect produce only a�0.3–0.5% SM reduction and the rest is theresult of the working line movement.

3.3. Overspray fogging

In recent years as the injection technology has done significantprogress overspray has been suggested as a viable method for

Fig. 10. Calculated power gain by using inlet air cooling during the summer monthsfor a specific power plant site.

Fig. 11. SM variation due to chilling for Tamb = 30 �C and varying RHamb.

Fig. 12. Compressor map for 1% overspray at inlet.

Fig. 9. Measured ambient conditions and calculated efficiency gain by using inletair cooling during the summer months for a specific power plant site.

1212 I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216

performance boost by Utamura et al. [16]. Wet compression allowsthe continuous evaporation during compression thus thermody-namically the compression is approaching the isothermal oneresulting to a significant reduction of compressor specific work[19]. This reduction will in turn create an increase to the engineoverall efficiency. What should be considered is that apart fromthe thermodynamic effect, wet compression results to a significantstage rematching.

As discussed by Roumeliotis and Mathioudakis [25] interstageinjection gives less gain than inlet overspray. Inlet overspray apartfrom being the most beneficial technique performance wise, is theone most easily implemented on existing gas turbines.

The wet compressor model presented will be used to analyzethe effect of overspray on compressor operation. Injection positionis assumed at the first stage inlet in order to isolate the overspray

effect from chilling effects. The droplets are assumed of 5 lmdiameter unless otherwise stated. The 5 lm diameter can be con-sidered rather small compared to the today available injector tech-nology [42], but the use of superheated fed – water [43] suggestthat such fine sprays are achievable.

In order to demonstrate the effect of overspray on the compres-sor operation the multistage (15 stages) axial compressor of theTornado engine has been used as a test case. The model usedwas adapted to data and the stages performance characteristicsthat reproduce rather well the overall dry compressor map wereacquired as described by Tsalavoutas et al. [44].

In Fig. 12 the compressor map calculated with 1% of oversprayis presented in conjunction with the dry map. It is evident thatwet compression results to a significant stage rematching movingthe speedlines towards higher mass flow rates, results that are inagreement with the ones presented by White and Meacock [20].

Concerning the surge line, operation with water injection re-sults to a movement of the surge line towards lower pressure ratiosat high rotational speeds, while for lower rotational speeds themovement of the surge line is less severe. For rotational speedsas low as N/Nd = 0.8 the wet surge line actually moves to higherpressure ratios. This behaviour is in agreement with ALSTOM’son site experience with the GT24/26 engines as discussed by Lech-eler and Hoffmann [45].

The stage rematching due to overspray results in a degradationof the aerodynamic performance, as the compressor is not de-signed for water injection, thus the stages are not operating any-more at the optimum operating point. The degradation of thecompressor aerodynamic efficiency occurs even with the injection

Fig. 14. Effect of overspray on engine efficiency, Tamb = 15 �C and Dd = 5 lm.

Fig. 15. Effect of overspray on engine efficiency, Tamb = 30 �C, Dd = 5 lm andDd = 2 lm.

I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216 1213

of small water quantities, as presented in [25], thus compromisingthe benefits that accrue from the thermodynamics of wetcompression.

Despite the stage rematching the effect of overspray is benefi-cial to the engine performance. Utamura et al. [16] have reporteda gain on thermal efficiency equal to 1.84% and on power equal to8% for high humidity conditions and water injection of 0.65%mass flow fraction. The effect of 1% overspray on the performanceof the Tornado engine calculated by the model is shown inFig. 13.

Overspray results to increased gas mass flow through thecompressor, which leads to the increase of compressor powerconsumption, although the compressor specific work is signifi-cantly reduced. The evaporation during compression results tolow CDT thus the fuel mass flow increases. The increased massflow through the turbine results to the rise of the cycle pressureratio and the turbine power increase. The increased humidity ofthe working medium along with the increased cycle pressure ra-tio leads to an only slightly decreased exhaust gas temperaturewhile the exhaust mass flow significantly increases. Althoughthe fuel mass flow is increasing the increased cycle pressure ra-tio in addition to the decreased compression specific work re-sults to a significant gain in efficiency and a gas turbine powergain up to 14%.

In Fig. 14 the efficiency increase is presented for increasing theoverspray quantity for operation with constant TIT. The efficiencyis increasing up to a specific overspray quantity and then the in-creased humidity of the working medium along with the decreasedCDT and the shift of operation even more off-design become dom-inant resulting to a decrease of the efficiency gain.

The efficiency gain is increasing as the ambient temperature isincreasing, mainly due to the higher evaporation rate at the firststages and the fact that the engine is already working off-design.Apart from the ambient conditions the droplet diameter is a signif-icant factor that influences the gain. As can be seen in Fig. 15 if thedroplets diameter are 2 lm the efficiency gain become as high as2.5% mainly due to the higher evaporation rates and lower entropyincrease due to evaporation.

Concerning the power gain the effect of overspray is even moreremarkable as a gain up to 20% with a 2% overspray can beachieved (Fig. 16).

Another interesting aspect is the effect of overspray on exhaustgas temperature and mass flow, quantities that are important foroperation in combined cycles. The exhaust gas temperature has asmall decrease while the mass increases significantly thus com-pensating for the temperature decrease (Fig. 17).

Fig. 13. Effect of overspray on o

Although the benefits of overspray on engine performance areindisputable there are some aspects that the users should take intoconsideration prior and after installing a nozzle manifold for over-spray. Overspray results to the increase of the engine pressure ratioand to the movement of the compressor surge line towards lowerpressure ratios. Both effects are negative influencing the compres-sor surge margin seen in Fig. 18.

The movement of the operating point and the simultaneouslydescent of the surge line results to a severe surge margin reductioneven for small water quantities. For 2% overspray the surge margin

verall engine performance.

Fig. 16. Effect of overspray on engine load, Tamb = 30 �C, Dd = 5 lm and Dd = 2 lm.

Fig. 17. Gas turbine exhaust quantities, Tamb = 30 �C and Dd = 5 lm.

Fig. 18. Effect of overspray on surge margin.

Fig. 19. Pressure build up change due to overspray relative to dry operation.

1214 I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216

is reduced to 3.8%. If the compressor’s remaining surge margin isless than 5% the machine may well be in an operating region oflocalized rotating stall that can subject the compressor to bladeflutter, thus it is essential that the maximum overspray quantitywith respect to the surge margin is defined prior to implementingoverspray to an existing engine.

Another aspect is the effect of the change of the pressure buildup inside the compressor. The stage rematching due to oversprayresults to the shift of the pressure build up towards the rear stagesof the compressor as these are the stages that work on high load-ing. In Fig. 19 the pressure build up for the specific engine is pre-

sented, exhibiting a behaviour that is in agreement withALSTOM’s experience with overpsray on the GT24/26 engines[45,46].

The pressure build up change results to a change in the bound-ary conditions for the gas turbine’s cooling air system. The coolingair supply pressure can be reduced during overspray for part of thecooling air system, if bleed occurs at the stages that the pressure isless than the dry one. The reduction of pressure will result to areduction of bleed quantity from the specific bleed ports. The re-duced temperature of the bleed along with the higher isobaric spe-cific heat may compensate for the reduced bleed quantity but thisshould be examined on a case by case basis.

Overspray quantity and droplet diameter are parameters thatthe user should take into consideration when deciding to installa nozzle manifold for overspray operation. In case of intense over-spray the surge margin will diminish, the pressure build up willchange dramatically altering the cooling air system’s behaviourand on top of that, if the whole water does not evaporate insidethe compressor, the user will witness a reduction of the possibleefficiency gain. In Fig. 20 the effect of water quantity and the effectof droplet diameter on the evaporation process inside the compres-sor are presented. The results for 5 lm droplet diameter are inagreement with the ones presented by Loebig et al. [47] who usea more complex (3-D) analysis for a compressor with similarlength and temperature rise characteristics. It is evident that theuser should ensure that the injected quantity can be evaporated in-side the compressor and the droplet diameter is as small aspossible.

Another aspect is the blade erosion due to droplet–blade inter-action. As discussed by Roumeliotis and Mathioudakis [48] duringexperiments on a compressor stage with conditions that do notpromote evaporation inside the compressor an increase of the con-sumed power was measured. The increase of the consumed powerwas attributed to the blade droplet interaction and the subsequentcentrifugation of the droplets towards the casing, thus there is adroplet–blade interaction. The results of borescope inspection byJolly and Cloyd [49] on a GE Frame 6B indicated that the drop-let–blade interaction may result to the appearance of erosion espe-cially at the first stage blades. The users should either upgrade thecoating of the compressor first stages or take special care for theinspection and possible replacement as proposed by Deneve et al.[50]. What should be considered is that the possible blade degrada-tion will result to a surge margin reduction as discussed by Brunet al. [38], thus the implementation of overspray to a worn com-pressor will lead to operation with very small surge margin. Thesame problem may occur after a long period of operation withoverspray if erosion problems arise and there is no inspectionand maintenance.

Fig. 20. Effect of injected quantity and droplet diameter on water evaporation.

I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216 1215

4. Concluding remarks

The techniques used for power and efficiency augmentationthat implement water injection at various stations across the en-gine have been analyzed using well established methods and mod-els. The performance and operational aspects of each techniquehave been calculated and analyzed in order to further understandtheir effect on gas turbine operation and performance. The possibleproblems that may arise concerning the gas turbine integrity suchas the reduced surge margin due to the techniques implementationhave been recognized, giving some basic guidelines on what theusers should expect and how to minimize the risk concerning theengine operation, along with highlighting aspects of water injec-tion that needs further research, such as the effect on the coolingair system.

The main remarks concerning engine performance and opera-tion are:

� Combustor water/steam injection results to a change of turbineoperation, shifting the compressor operating point towards thesurge line. The surge margin reduction can become significantfor large water/steam injection quantities if the firing tempera-ture remains the same. The effect of the increased humidity con-tent with respect to the blade cooling should be furtherexamined as it may result to life limiting events.

� Combustor water injection results to a thermal efficiency reduc-tion, while steam injection results to a significant increase of thegas turbine efficiency as the heat for the water evaporation isavailable from an outside source. The magnitude of the effectis different for different control variables.

� Combustor water/steam injection results to a significant poweroutput increase. As combustor water injection can be used evenon conditions that result to high load (for example during verylow ambient temperatures) the mechanical integrity of the gasturbine and the possibility of overspeeding should be taken intoconsideration while deciding the operating envelope of the com-bustor water/steam injection system.

� The analysis of the effect of chilling on eight engines of differentconfiguration, power setting and TIT level indicates that multi-shaft engines are expected to present higher improvement withchilling. It should be noted that in order to fully examine theexpected gain due to inlet air cooling on the engine perfor-mance, an adapted engine model should be used and not a gen-eric one as the compressor map is of high importance.

� Overspray fogging gives a significant boost on the engine perfor-mance even for relative low ambient temperatures as for ISOconditions.

� The gain that can be achieved with overspray is better as theambient temperature increases and as the droplet diameterdecreases, since in both cases the evaporation at the fist stagesis promoted.

� The benefits of overspray on Combined Cycles have not beenexamined thoroughly as in most cases the analysis of the com-bined cycle with overspray is done assuming constant compres-sor speedlines. It should be further examined if there is anoverspray quantity which gives the best behaviour with respectto the combined cycle overall efficiency.

� The performance gain comes with a significant cost on surgemargin. There is the possibility of erosion due to droplet–bladeinteraction which may further reduce the surge margin. Plannedinspection and possible the coating change of the first bladesshould be undertaken.

� Overspray is expected to influence the cooling air system byreducing the amount of cooling air from the interstage bleeds,especially the front ones due to the shift of the compressor pres-sure build up. Thus the effect of overspray on the gas turbinecooling system should be analyzed in order to fully evaluateits effect and avoid life limiting effects.

References

[1] Bhargava A, Colket M, Sowa W, Maloney D. An experimental and modelingstudy of humid air premixed flames. ASME J Eng Gas Turb Power2000;122(3):405–11.

[2] Burnham JB, Giuliani MH, Moeller DJ. Development, installation, and operatingresults of a steam injection system (STIG) in a general electric LM5000 gasgenerator. ASME J Eng Gas Turb Power 1987;109(3):257–62.

[3] Jonsson M, Yan J. Humidified gas turbines – a review of proposed andimplemented cycles. Energy 2005;30(7):1013–78.

[4] Kousuke N, Takagi T, Kinoshita S. Regenerative steam-injection gas-turbinesystems. Appl Energy 2005;81(3):231–46.

[5] Ondryas IS, Wilson DA, Kawamoto M, Haub GL. Options in gas turbine poweraugmentation using inlet air chilling. ASME J Eng Gas Turb Power1991;113(2):203–11.

[6] Bhargava R, Meher-Homji CB. Parametric analysis of existing gas turbines withinlet evaporative and overspray fogging. ASME J Eng Gas Turb Power2005;127(1):145–58.

[7] Gajjar H, Chaker M, Dighe A, Meher Homji C. Inlet fogging for a 655 MWcombined cycle power plant – design, implementaion and operatingexperience. ASME paper no. GT2003-38757; 2003.

[8] Boyce MP. Gas turbine engineering handbook. 2nd ed. Boston: Butterworth–Heinemann; 2002.

[9] Meher-Homji CB, Mee III TR. Inlet fogging of gas turbine engines Part B:practical considerations, controls and O&M aspects. ASME paper no. 2000-GT-308; 2000.

[10] Bader GH. GE aeroderivative gas turbines – design – and operating features.GER-3695E, GE Power Systems; 2000.

[11] Bassily AM. Effects of evaporative inlet and aftercooling on the recuperated gasturbine cycle. Appl Therm Eng 2001;21(18):1875–90.

[12] Tsalavoutas F, Pothos S, Mathioudakis K, Stamatis A. Monitoring theperformance of a twin-shaft ship propulsion turbine by means of adaptive

1216 I. Roumeliotis, K. Mathioudakis / Applied Energy 87 (2010) 1207–1216

modeling. In: RTO symposium on gas turbine operation and technology forland, sea and air propulsion and power systems, Ottawa, Canada, 18–21October; 1999.

[13] Mathioudakis K, Aretakis N, Tsalavoutas A. Increasing diagnostic effectivenessby inclusion of fuels composition and water injection effect. ASME paper no.GT-2002-30032; 2002.

[14] Trout AM. Theoretical turbojet thrust augmentation by evaporation of waterduring compression as determined by use of a Mollier diagram. NACA TN2104; June 1950.

[15] Hill PG. Aerodynamic and thermodynamic effects of coolant injection on axialcompressors. Aeronaut Quart 1963(February):333–48.

[16] Utamura M, Takehara I, Horii N, Kuwahara T. A new gas turbine cycle foreconomical power boosting. ASME paper no. 97-AA-142; 1997.

[17] Utamura M, Kuwahara T, Murata H, Horii N. Effect of intensive evaporativecooling on performance characteristics of land based gas turbine. In: Proc.ASME international joint power generation conference 1999, vol. PWR;34(2);1999. p. 321–8.

[18] Wang T, Khan JR. Overspray and interstage fog cooling in compressor usingstage-stacking scheme – Part 1: development of theory and algorithm. ASMEpaper no. GT2008-50322; 2008.

[19] Härtel C, Pfeifer P. Model analysis of high-fogging effects on the work ofcompressor. ASME paper no. GT 2003-38117; 2003.

[20] White AJ, Meacock AJ. An evaluation of the effects of water injection oncompressor performance. ASME J Eng Gas Turb Power 2004;126(4):748–54.

[21] Sexton WR, Sexton MR. The effect of wet compression on gas turbine operatingperformance. ASME paper no. GT2003-38045; 2003.

[22] Hale A, Klepper J, Hurwitz W. A numerical capability to analyze the effects ofwater ingestion on compression system performance and operability. ASMEpaper no. GT2005-68480; 2005.

[23] Ludorf RK, Elder RL, Tronbøl TH, Overli J. Stage re-matching as a result ofdroplet evaporation in a compressor. ASME paper no. 95-GT-194; 1995.

[24] Horlock JH. Compressor performance with water injection. ASME paper no.2001-GT-343; 2001.

[25] Roumeliotis I, Mathioudakis K. Evaluation of interstage water injection effecton compressor and engine performance. ASME J Eng Gas Turb Power2005;128(4):849–56.

[26] Roumeliotis I, Mathioudakis K, Aretakis N. Performance analysis of twin-spoolwater injected gas turbines using adaptive modeling. ASME paper no. GT2003-38516; 2003.

[27] Stamatis A, Mathioudakis K, Papailiou DK. Adaptive simulation of gas turbineperformance. ASME J Eng Gas Turb Power 1990;112(2):168–75.

[28] Aretakis N, Mathioudakis K, Stamatis A. Non-linear engine component faultdiagnosis from a limited number of measurements using a combinatorialapproach. FSME J Eng Gas Turb Power 2003;125(3):642–50.

[29] AGARD-AR-332. Recommended practices for the assessment of the effect ofatmospheric water ingestion on the performance and operability of gasturbines engines. North Atlantic Treaty Organization; 1995.

[30] Mathioudakis K, Stamatis A. Compressor fault identification from overallperformance data based on adaptive stage stacking. ASME J Eng Gas TurbPower 1994;116(1):156–64.

[31] Spalding DB. Combustor and mass transfer. Pergamon Press; 1979. p. 82–99.[32] Roumeliotis I. Modelling of gas turbines and their components operation with

water injection. PhD Thesis LTT, NTUA; 2006.[33] Young JB. The fundamental equations of gas-droplet multiphase flow. Int J

Multiphase Flow 1995;21(2):175–91.[34] Roumeliotis I, Mathioudakis K. Water injection effects on compressor stage

operation. ASME J Eng Gas Turb Power 2006;129(3):778–98.[35] Anonymous. Heavy – duty gas turbine water injection system. Petrotech Inc.;

2006.[36] Mathioudakis K. Analysis of the effects of water injection on the performance

of a gas turbine. ASME J Eng Gas Turb Power 2002;124(3):489–95.[37] Walsh PP, Fletcher P. Gas turbine performance. Blackwell Science Ltd.; 1998.[38] Brun K, Kurz R, Simmons H. Aerodynamic instability and life limiting effects of

inlet and interstage water injection into gas turbines. ASME paper no. GT2005-68007; 2005.

[39] Fischer AC, Frutchi HU, Haselbacher H. Augmentation of gas turbine output bysteam injection. ASME paper no. 2001-GT-0107; 2001.

[40] Jin H, Marson E, Pistor R, Dempsey L. A study on the life cycle impact of steaminjection. ASME paper no. GT2007-27029; 2007.

[41] Meher Homji CB, Hattenbach T, Messersmith D, Weyermann HP, Masani K,Gandhi S. World’s first application of aeroderivative gas turbine drivers for theConocophillips optimized cascade LNG process. Bechtel Corporation; 2008.

[42] Chaker M, Meher Homji CB, Mee II T. Inlet fogging of gas turbine engines – PartII: fog droplet sizing analysis, nozzle types, measurement, and testing. ASME JEng Gas Turb Power 2004;126(3):559–69.

[43] Van Liere J, Laagland GHM, Meijer CG. Retrofit of gas turbines by SwirlFlashTM

over-spray. Alpha Power Systems Publication; 2001.[44] Tsalavoutas A, Stamatis A, Mathioudakis K. Derivation of compressor stage

characteristics, for accurate overall performance map prediction. ASME paperno. 94-GT-372; 1994.

[45] Lecheler S, Hoffman J. The power of water in gas turbines: ALSTOM’sexperience with air Inlet cooling. In: Proc. PowerGen 2003 São Paolo, Brasil;November 11–13; 2003.

[46] Savic SM, Rostek KE, Klaesson DK. Techno-economic evaluation ofcommercially available high fogging systems. ASME paper no. GT2005-68368; 2005.

[47] Loebig J, Vittal BVR, Booher V. Numerical simulation of water/methanolevaporation in an axial flow gas turbine compressor. AIAA paper no. AIAA98-3559; 1998.

[48] Roumeliotis I, Mathioudakis K. Analysis of wet air compression in axialcompressor stage 7th European conference on turbomachinery. Paper no. 122;2007.

[49] Jolly S, Cloyd S. Performance enhancement of GT24 with wet compression.Presented at the power – gen international, Las Vegas, December 9–11; 2003.

[50] Deneve M, De Tandt B, Cornelis N, Bultereys C, Gijbels S. Results of the firstapplication of the SwirlflashTM wet compression system on a 150 MW heavy –duty gas turbine. ASME paper no. GT2005-68726; 2005.