Design of a High Performance Axial Compressor for Utility Gas ...

THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS345 E. 47 St., New York, N.Y. 10017

s The Society shall not be responsible for statements or opinions advanced in papers or In dis-cussion at meetings of the Society or of its Divisions or Sections, or printed in its publications.

MDiscussion is printed only if the paper is published in an ASME Journal. Papers are available]^( from ASME for fifteen months after the meeting.

Printed in USA.

91-GT-145

Design of a High Performance Axial Compressor forUtility Gas Turbine

A. SEHRA* J. BETTNER A. COHNTextron Lycoming Allison Gas Turbine Division Electric Power Research Institute

Stratford, CT Indianapolis, IN Palo Alto, CA

ABSTRACT

An aerodynamic design study to configure a highefficiency industrial-size gas turbine compressor ispresented. This study was conducted using an advancedaircraft engine compressor design system. Startingwith an initial configuration based on conventionaldesign practice, compressor design parameters wereprogressively optimized. To further improve theefficiency potential of this design, several advanceddesign concepts (such as stator ends bends and veloci-ty controlled airfoils) were introduced. The project-ed polytropic efficiency of the final advanced conceptcompressor design having 19 axial stages was estimatedat 92.8 percent, which is 2 to 3 percent higher thanthe current high efficiency aircraft turbine enginecompressors.

The influence of variable geometry on the flow andefficiency (at design speed) was also investigated.Operation at 77 percent design flow with inlet guidevanes and front five variable stators is predicted toincrease the compressor efficiency by 6 points ascompared to conventional designs having only the inletguide vane as variable geometry.

NOMENCLATURE

DCA Double circular arc

V AVD f Diffusion factor, Df =

1 - v2 + —01 1

H Boundary layer shape factor, H = 6 /0

N Shaft speed, rpm

R Compressor total pressure ratio,c

R = PToutc P

Tin

SM Surge margin, SM = (R c /W) SP - 1 x 100%

(R c /W) DP

V Velocity, ft/sec

W Flow rate, lb/sec

x,y Coordinate direction

13 Air angle, deg

%• Employed at Allison Gas Turbine at the time ofthis effort.

A Change in parameter

6 Inlet pressure/standard day pressureS = PT/14.696

n Efficiency,

0 Inlet total temperature/standard daytemperature, 0 = TT/518.7

0 Boundary layer momentum thickness, in.

o Airfoil solidity, a = c/S

w Total pressure loss coefficient

Subscripts

e Exit

i Incompressible

max Maximum

TT Total-to-total

poly Polytropic

SS Suction surface

T Total

t Tip

to Trailing Edge

SP Surge Point

DP Design Point

0 Tangential Direction

I INTRODUCTION

There have been great improvements in axial com-pressor design techniques for compressors used inaircraft gas turbine engines. In addition to increas-ing the efficiency, these improvements have been aimedat increasing the specific flow, the surge margin, andthe pressure ratio per stage. This has alloweddevelopment of compact, light weight high performancecompressors and engines that are of primary importancefor flight economics.

The purpose of the design presented in this paperwas to apply the knowledge and design techniquesdeveloped for aircraft engine axial flow compressorsto the compressor for a large utility turbine engine.With weight and number of stages being of secondaryimportance, the goal was to develop an aerodynamic

Presented at the International Gas Turbine and Aeroengine Congress and ExpositionOrlando, FL June 3-6, 1991

This paper has been accepted for publication in the Transactions of the ASMEDiscussion of it will be accepted at ASME Headquarters until September 30, 1991

Copyright © 1991 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 01/31/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

design with a level of efficiency potentially higherthan that of the large utility turbine engine compres-sors that are now in operation.

This paper presents the results of an investigationto develop a 14:1 pressure ratio, 800 lb/sec utilitycompressor aerodynamic design.

The design study for this Electric Power ResearchInstitute (EPRI) compressor was conducted in fourphases. The first phase was concerned with a detailedparametric study to evaluate, with the objective ofmaximizing overall compressor efficiency, the influ-ence of key aerodynamic and geometric design variableson compressor performance. Phase II refined the PhaseI design to consider (1) selection of airfoil pro-files, (2) off-design performance, and (3) blade andvane dynamics. Phases I and II employed conventionalcompressor design techniques in that circular arc andseries airfoil profiles were incorporated in thedesign. The basic design tool used was based on theradial equilibrium with in-built semi-empiricalloss-loading correlations. The Phase II design shouldonly be viewed as an optimized detailed aerodynamicdesign.

Phase III examined the application of advancedblading concepts that had the potential for increasingdesign point efficiency. Two and three-dimensional(3-D) end-wall and airfoil surface boundary layeranalysis codes, along with secondary flow and bladetip clearance analysis models, were employed todevelop the advanced concept airfoil design lossmodels. These advanced concept airfoils were thenintegrated into the Phase II design to produce thefinal advanced concept compressor.

The final phase (Phase IV) of this study wasdirected at investigating the influence of variablegeometry on the compressor efficiency during the partpower operation of the utility turbine.

II AXIAL COMPRESSOR DESIGN SYSTEM

Figure 2-1 shows the overall logic of the axialcompressor design system used for the design studyreported in this paper. Also shown in this figure arethe major computer modules used for EPRI compressordesign. Several other modules which are specificallyused for highly loaded transonic compressor stages(for example, three-dimenssional viscous codes) arenot included here.

The basic design philosophy centers around itera-tively updating design parameters (such as, workdistribution) and airfoil shape until a "minimum loss"design is established. In this iterative optimizationfor minimum loss (or maximum efficiency), it isensured that the other design requirements such as thesurge margin and structural integrity are notsacrifices.

The first major step in this iterative designprocess is the solution of the meridional (axisym-metric) flow in conjunction with blade generationprocess. This provides the designer with a reasonableapproximation of the flowpath, number of stages,airfoil solidity, aspect ratio, and also an initialguess of the blade shape. Built into this corecompressor design module (CCDM) are semi-empiricalcorrelations for airfoil profile and shock losses,deviation, and tip clearance effects with associatedmixing losses. This design is now refined by conduct-ing two-dimensional blade-to-blade flow analysis(BBFA) and more accurate estimate of the end wallboundary layer secondary flow losses (SECOND). Beforea design is finalized, three-dimensional inviscid (orviscous, if highly loaded) analyses is carried out toevaluate three-dimensional effects that cannot bepredicted by the above quasi-three-dimensional system.

Further relevant details of the design process andcomputer modules are included in appropriate sectionsof this paper.

III PHASE I - PRELIMINARY DESIGN CONFIGURATION

The first phase of the program was to define aconceptual design configuration for the axial flowcompressor with the following design point conditions:

1. corrected flow, W,Je/S 800 lb/sec

2. overall pressure ratio, R C 14.0:1

3. corrected speed, N/J0 3600 rpm

This conceptual design was configured by conductinga detailed aerodynamic and geometric parametric studywhere key design parameters were varied to evaluatetheir influence on compressor performance with thedesign objective of maximizing compressor efficiency.

Parametric Study. The parametric study was dividedinto two steps, an initial screening phase followed byfine tuning of each design parameter.

DESIGNPARAMETERS

_ _ _ __ _ FLOW PATH3-D END WALL I

.NO OF BLADE

BOUNDARY LAVER MERIDIONAL FLOW ANALYSIS ROWS(SHROUD) (MFA)

licv ^^ ^^OFF DESIGNPERFORMANCE CORE COMPRESSOR DESIGN MODULE JJJ

PREDICTION (CCDM)

AIRFOIL GENERATION

STRUCTURAL 65 SERIES CUSTOMIZED i

DESIGN DCAMCA

BLADE DESIGN(CBD)

— ,. , r.e

HEAD'S INTEGRAL 2-D BLADE-TO-BLADE _ _ _ ___ p^•BOUNDARY LAVER(HEAD)

— -- — — FLOW ANALYSIS (BBFA) ;y^•r

SECONDARY FLOW ANALYSISSECOND)

3-D INVISICID ANALYSIS _..—_._ __._.DENTON) GEOME TRV

RS372090A

Figure 2-1 Axial Compressor Design System, Overall Logic


During the initial screening phase, the range ofeach design parameter was based on the conventionalaircraft engine compressor design practices. Table 1gives various design parameters and their ranges thatwere investigated in this phase.

TART.F 1

Flow path shape Constant hub, mean, and tip

Number of stages 8 - 19

Aspect ratio 2.0 at inlet to 1.5 at theaft end

Solidity Tip = 1.0 constantHub = 1.7 front to 1.2 at

the aft end

Inlet specific flow 35 - 42 lb/sec-sqft

Stage reaction(Stator swirl) 00 to + 10°, exit swirl = 0°

Exit Mach No. 0.3 fixed

Inlet hub/tipradius ratio 0.4 - 0.6

The general approach to the initial phase of theparameter study was to develop a series of designs forconstant hub, mean, and tip flow path shapes with thenumber of stages varying from 19 to 8 for each flowpath shape. If during the study, airfoil diffusionfactor, D f , exceeded 0.5 then stage pressure ratiodistributions were varied to limit D <0.5. Becauseof the low wheel speed (3600 rpm) several instancesoccurred where the stage loadings were excessive andcould not be corrected by alterations in stage axialpressure ratio distribution. The condition wasparticularly evident for the constant hub designs.

Figure 3-1 illustrates the effect of number ofstages and inlet specific flow for the constant hub,constant mean, and constant tip configurations on theoverall adiabatic efficiency of the compressor. Rotor1 inlet hub/tip for this study was 0.5.

89 Design point conditions

s, WelA = 800 lb/sec

88Rc = 141

N/le = 3600 rpm benign 8

rh/rt = 0.5

81 Initial solidity level 35

^^ 40 — 40 Constant

\ ' 42 42 \ 6YD

° 86 ❑ \ \ \ a0 75 d Igns

\\J5 ^C o nstant

A 8s \\ designs

Rotor 1 inlet \

specific flo,, Ma/A lb/sec-ft 2

g \ 40O 40 \

g 84❑ 35 \

42 Constant

4 \ L1,

83 \ 0,01915

° \0 35 lb/sec-ft 2J

82

0 2 4 6 8 10 12 14 16 18 20 22

Number of stages

Figure 3-1 Influence of Number of Stages and InletSpecific Flow on Compressor Efficiency

As seen from Figure 3-1, the initial screening studyfavors a 15-stage, constant meanline configurationwith an inlet specific flow of 40 lb/sq ft-sec.

The adiabatic efficiency potential of this configu-ration is computed to be 87.0 percent (polytropicefficiency = 90.8%) which is of the same level as theas GE E 3 compressor.

In the next phase of the prameteric study, a moredetailed optimization was carried out with designvariables extended well beyond the conventional levelsused in the aircraft engine compressors. One suchdesign variable was the tip solidity which was pro-gressively reduced below the conventional aircraftcompressor design practices. To counter the effect ofincreased diffusion factor associated with reducedsolidity, the stage number was increased therebyreducing the average stage pressure ratio. Particularattention was also paid to the stage-wise and span-wise distributions of work, diffusion factor, andswirl.

Results of this fine tuning indicated that themaximum efficiency potential is offered by a 19-stageconfiguration having a relatively high specific flowin the front stage and very low airfoil solidity.Increasing the number of stages beyond 19, shows asudden drop in efficiency (see Figure 3-2).

K

Constant mean section

Wa/A)in. = 40 lb/sec-ft 2

rh/rt)in. = 0.5

C

89

d

N

L 880

O

aI-E

m

870

14 16 18 20 22

Number of stages

Figure 3-2 Effect of Number of Stages on (ConstantMean Section, Low Solidity Design)Compressor Efficiency

Preliminary Design Configuration. Based on theresults of the detailed parametric study, the compres-sor configuration that produced maximum efficiencyincorporated the following features:

• 19 stages plus IGV• 40 lb/sec-ft 2 inlet specific flow• 0.45 inlet hub/tip radius ratio• constant mean section flow path• swirl in all stages except the last stator• very low solidity

It was predicted that the proposed compressordesign would have the following design pointperformance:

• compressor exit Mach number, M 0.3• overall adiabatic efficiency, UT 88.8%• overall polytropic efficiency, 0poly

92.1%

3


X.

Ii'14.0

Figure 3-3 presents the flow path of the optimizedpreliminary design configuration. Important designvariables for this configuration are presented inFigures 3-4 through 3-7. Also shown in these figuresare the corresponding design parameters of contempo-rary aircraft engine compressors. The compresssorselected for this comparison are the T56-A427 and fiveother unnamed compressors. This comparison highlightsthe similarities and differences of the EPRI designwith these existing compressors.

129r.

Er

90

H

48

72

16 1i]r1i. l\ .1! .it1.Lfi

-10 6 22 38 61 10 86 102 118 134 150 166 162 193

841.1 locatlen--ln.

Figure 3-3 19 Stage Preliminary Design Compressor

44 $9c1 Ceyresaer No. styes 8/18/ 4. ID/sec RI

42 0 756 -A421 11 31.16 11.0A 11 31.99 13.0

40 8 10 120.00 22.81C 5

150 00 12.138 D 6 31.61 8.0

E 10 13.00 16.036

FPR1 19 800.00 11.034

32

30

2e

26

24

22

20

Is

16

14

0 2 4 6 8 10 12 11 16 18 20

Stage nuN6er

Figure 3-4 Comparison of EPRI Proposed Design RotorInlet Specific Flow Distribution with

2.4 Other Contemporary Designs

2.0 Symbol Compressor No. stages WNe/6, lb/sec RE

156 -1-421 14 37.16 14.0I<aJ> A 11 34.99 13.0

1.8 8 10 120.00 22.87C 5 150.00 12.4D D 6 37.61 8.0fl E 10 13.00 16.01.6EPRI 19 800.00 14.0

1.4

1.2

1.0

0 2 4 6 8 10 12 14 16 18 20Stage niwber

Figure 3-5 Comparison of EPRI Proposed Design StagePressure Ratio Distribution with OtherContemporary Designs

1.6

- 1.2

1.0

0.8

0.6Synbol Compressor No. stages NNe/6, lb/sec RC

0 156-6-421 14 37.16 14.02.4 A 14 34.99 13.0

0 B 10 120.00 22.81G C 5 150.00 12.4

2.2 6 D 6 37.61 8.0o E 10 13.00 16.0

2.0 \ • EPRI 19 800.00 14.0

1.8

1.6

1.4

e 1.2

1.0

0.6

0.6

0 2 4 6 6 10 12 14 16 18 20

Stage Numaer

Figure 3-6 Comparison of EPRI Proposed Design RotorSolidity with Other Contemporary Designs

Symbol Compressor No. stages NNe/6, lb/sec

0 T56-A-421 14 31.16A 14 34.99

O B 10 120.00C 5 150.001 6 37.61

o E 10 73.00

0.6 • EPRI 19 800.00

0.5

0.4

0.3

0.2

0.6

0.5

0.1

N 0.3

0.2

0.1

Figure 3-7 Comparison of EPRI Proposed DesignDiffusion Factors with OtherContemporary Designs

IV PHASE II - AIRFOIL GEOMETRY

Phase I of this study optimized global designparameters, using conventional design techniques, toproduce a preliminary design. In Phase II, prelimi-nary geometry of rotor and stator airfoils was de-fined. This involved selection of incidence, chord,maximum thickness/chord, and interblade row gaps. Thecriteria was to select airfoil geometric parametersthat offer the best compromise between the design andoff-design performance without compromising thestructural integrity of the airfoil. Some adjustmentof solidity and swirl distributions were also carriedout to improve stage-wise and span-:wise distributions

4


of airfoil loading. No attempt was made at this pointto customize the airfoil shape. All the airfoils hadcircular arc meanline with the exception of the firstrotor multiple circular arc (MCA) configuration.

Incidence. In an effort to maximize the perfor-mance at the design point conditions by keeping theairfoil diffusion requirements reasonable and still tolimit the swing in incidence angles from design pointto a near-stall operating point (20% surge margin),near-zero incidence angles were selected for all therotors. For stators, incidence angles were itera-tively selected in the forward portion of the compres-sor; in the aft stages, the stator incidence angleswere set at negative values. The negative incidenceangles were limited to the aft stages, since in themultistage environment, it is principally the aftstages of the compressor that react to the increasesin loading and incidence angle change as the compres-sor moves from design point operation to stall.

The design process began with the basic airfoilgeometry of the preliminary design configuration whoseflow path was shown in Figure 3-3. Incidence angleswere then selected based on the above criteria, andthe resulting airfoils were configured for designoperating point using a deviation rule based on theNASA SP36 (Reference 1) cascade correlation withcorrections for endwall effects.

A detailed balde-to-blade flow analysis was alsoconducted on selected rotor and stator airfoils, withthe resulting surface velocity distributions examinedfor possible flow separation and unacceptably largevalues of the equivalent diffusion parameter, D(suction surface V /V which is a measure e^fsuction surface diffusion)

The criteria used during this phase of study forthe prevention of incipent flow separation was to keepthe value of the surface incompressible boundary layershape factor H. 3.0 at the airfoil trailing edge.It is recognized that the value of H, = 3.0 is somewhat higher than the more generally accepted value ofH. > 2.2 for incipient separation. It was anticipatedthat H. could be substantially reduced later in thestudy when the advanced concept customized airfoilprofiles would be employed in the design process.

Once the design point airfoil flow conditions wereconsidered acceptable, an off-design model of thecompressor was developed, and the above computationswere repeated at the predicted 20 percent surge marginoperating point. The increases in incidence angles,D , and potential flow separation were examined att 20 percent surge margin point. If the resultingdesign/off-design point results were deemed unsatis-factory, the entire process was repeated with anotherassumed axial/radial incidence angle distribution.This design process led to the incidence angle distri-butions illustrated in Figures 4-1 and 4-2 for therotors and stators, respectively. It can be seen thatthe incorporation of negative incidence angles at thedesign point limited the swing in incidence from about-1 to +6 deg for the rotors and -5 to +6 deg for thestators for the projected range of operation at designspeed. Figure 4-3 presents the predicted design speedpressure-flow characteristic associated with theairfoil geometry resulting from the incidence angledistributions of Figures 4-1 and 4-2.

500% aestvn ayend

s . ....»ue

S Z0% sSr e...v,, .wa,n.n

3.... ,..•.... .....T1p

3 '

1.u•..

.................................

o ne.n kt , vn o• , .tw,n anatt,.n

0 3 1 6 B 10 11 14 56 IB 30

Stan naer

Figure 4-1 Variation of the Rotor Incidence Anglefrom Design to 20 Percent Surge MarginOperating Conditions

500E Yes tvn spell .xu6

.^._.... ^ T,D M serve

T,v

.r:„n .etetr,.an ..Patten

SIP

rw

Figure 4-2 Variation of the Stator Incidence Anglefrom Design to 20 Percent Surge MarginOperating Conditions

it 100% design speed

20% surge margin

16

15

14 Design operating paint

13

lz790 800 810

Corrected fill, 11 Ve/a--,baser

Figure 4-3 Flow Characteristics at 100 PercentDesign Speed Pressure

Figures 4-4 and 4-5 show the stage wise radialdistribution of D for the rotors and stators,respectively. Forew%rd-stage rotor and aft-stagestator values are aggressive, but, as stated earlier,it is anticipated that these levels of high diffusioncan be accommodated by customizing the airfoil con-tours. Also shown in Figures 4-4 and 4-5 are theprojected increased levels of D at the 20 percentsurge margin operating point. Teti improved diffusioncapability of the customized airfoil contours will bebeneficial in attaining these off-design D levelsand providing the compressor with the capa€ility toachieve 20 percent surge margin.

Dynamic Analysis. To ensure that the detaileddesign configuration was structurally sound, a dynamicanalysis (using the state-of-the-art finite elementcode) was performed on blade rows throughout thecompressor. The objective was to identify maximum


airfoil aspect ratio that is acceptable from the pointof view of structural integrity as well as the passagediffusion capacity. The criteria for dynamic accept-ability was that all rotors first bending-mode naturalfrequencies be above second engine order. For sta-tors, the criteria was that the first bending frequen-cy not be near the first engine order. The criteriafor the limiting passage diffusion capacity wassimilar to that presented in Koch's paper (Ref. 2).

Figure 4-6 Detailed Design Configuration Flow Path

1 L 20% surge da rgtn•••^• conAltlen

i 1.a T,p —^^ Design oper.ttegWA tt ton

M surge Barg to= 1.6 conOltlon....................

1.5Kean •••••••••••••••• Design wr.ttng

—d Itton

— l.e

O 1 1 20% surge wargtn

1.6 Mub _••••^• Design oyer.ttugconCltion

1.52 e 6 6 10 12 14 16 15 20

Stage nr6er

Figure 4-4 Variation of the Rotor EquivalentDiffusion Factor from Design to 20%Surge Margin Operating Conditions

100% design speed

Figure 4-5 Variation of the Stator EquivalentDiffusion Factor from Design to 20%Surge Margin Operating Conditions

This part of the study helped to define airfoilmaximum thickness/chord ratios and the minimum accept-able length of the compressor. Minimizing the lengthreduces endwall skin friction drag, shafting dynamicsproblem, and probably, overall compressor cost.Details of this study are presented in Reference 3.

Detailed Design Configuration. Relative to thepreliminary design configuration reviewed in SectionIII, all of the previously mentioned design refine-ments of airfoil incidence, thickness/chord ratio,solidity, aspect ratio, and axial gaps along withminor changes in airfoil swirl distributions, producedthe detailed design configuration presented in Figure4-6.

Comparing the compressor flow paths of the prelimi-nary design configuration of Figure 3-3 with that ofFigure 4-6 shows that a substantial reduction inoverall compressor length has been effected in thedetailed design configuration. Stage-wise distribu-tions of key design aerodynamic and geometric parame-ters for the detailed design configuration areillustrated in Figures 4-7 through 4-9.

6.Or

2.

0 2 . 6 6 10 12 11 16 IB 20

St69e nuMer

Figure 4-7 Detailed Design Configuration StagewiseDistribution of Rotor and Stator AspectRatio

F[ [

TTTT Figure 4-8 Detailed Design Configuration Stagewise

Distribution of Diffusion Factors

S.Ilally o s

^norar.o.cing

.In

. rsl.g. n^roe.r

^. ^. >e

Figure 4-9 Detailed Design Configuration StagewiseDistribution of Solidity

6

■ -


The detailed design configuration can be catego-rized as being a conventionally designed compressorwith relatively high specific flow, subsonic, highaspect ratio, fairly lightly loaded, very low soliditydesign features. Results of these design studies showthat the optimum values of design parameters such asstage loading, hub/tip radius ratio, and swirl aresimilar to industrial compressors in service today.The predicted design point performance for the de-tailed design configuration is:

• overall adiabatic efficiency 88.8%• overall polytropic efficiency 92.1%• surge margin, SM 20%

This projected performance level is considered tobe that which would result from experimental develop-ment to fine-tune stage matching, high quality air-foils, very good surface finishes, and a blade runningtip clearance of 0.02 inches.

V PHASE III - ADVANCED DESIGN CONCEPTS

The performance results described in the previoussections have shown that if any significant improve-ments in efficiency are going to be realized, they aregoing to be the result of integration of advancedairfoil design concepts into the aerodynamic designprocess. In any compressor blade row design, theprincipal contributors in total pressure loss are thefollowing:

• Cascade Profile Loss--due to viscous effects onairfoil surfaces

• Endwall Loss--due to viscous three-dimensionalboundary layer flows resulting in pressure losson the endwalls, formation of energy dissipatingsecondary corner vortices, and high incidence onairfoil leading edges submerged within the wallboundary layer

• Tip Clearance Loss--losses associated with theblade tip clearance leakage flow along with thecasing boundary layer fluid contained anddissipated within the tip leakage vortex core

• Shock Loss--total pressure losses associatedwith deceleration through a shock positioned infront of or within airfoil passage

The detailed design configuration is essentially asubsonic compressor design; consequently, the impactof shock loss on overall compressor efficiency was nota major consideration. On the other hand, the meansto control or reduce profile, endwall, secondary, andtip clearance losses are a major concern if theefficiency level predicted for the EPRI detaileddesign configuration is to be improved.

Although several advanced airfoil concepts werereviewed (see Reference 3), only the following threeconcepts were considered potentially beneficial forreducing aerodynamic losses.

Customized or Velocity controlled Airfoils--These airfoil shapes minimize suction surfacediffusion loss and maximize the static pressurerise capability of an airfoil.

Stator End Bends--This concept has shown promiseby unloading the stator endwall region and by

improving the incidence on the portion of theairfoil submerged in the wall boundary layer.

Swept-Forward Leading Edge Stators--This conceptreduces endwall region loss by inducing amigration of high-momentum fluid to the statorendwall region.

The performance potential of velocity-controlledairfoils (VCA) and stator end bends was estimated withthe aid of Allison's advanced design tools. A de-tailed description of the design methodology andanalytical procedures adopted for incorporating theVCA and stator end bends into the detailed designconfiguration is presented in the following text.

Velocity Controlled Airfoils. The VCA concept,which aims at optimizing the streamwise distributionof suction surface diffusion, was applied to theairfoil sections not subjected to strong three-dimen-sional effects (i.e., away from the endwall regions).

This concept could be extended to the endwallregion, but that would require a fully three-dimen-sional viscous analysis. To generate VCA sections,the customized blade generation program (CBG) wascombined with two-dimensional blade-to-blade flowanalysis code (BBFA) and an integral-momentum boundarylayer analysis. The BBFA is a fast analysis capableof handling variation in streamtube height, boundarylayer blockage, loss, and local supersonic bubbles.The scheme, based on the streamline curvature tech-nique, has been validated against cascade data andother computational schemes and is extensively used todesign axial compressor airfoils. The boundary layeranalysis, which is merged with BBFA, is based on thewell-known Head's momentum-integral technique (Ref 4).

The design methodology adopted for the presentstudy involved itteratively updating the airfoil shapeto minimize the boundary layer skin friction losswhile ensuring that the suction surface boundary layerdoes not separate. The following design criteria wereused:

The trailing edge shape factor, H, is less thanor equal to 2.2. This is to ensure that theboundary layer remains attached to the airfoilsurface.

The computed exit flow angle for the customizedairfoil remained equal to that of the baselinecircular arc meanline airfoil. This designrequirement ensured that the inlet swirl to thedownstream rotor would remain unchanged.

Figure 5-1 shows a diagram of the overall logicemployed to arrive at the optimized VCA sections.

Figure 5-2 shows, for stator 6, the impact of (1)customizing the midspan profile to optimize diffusionand (2) the influence of chord (solidity) reduction onkey boundary layer and aerodynamic parameters relativeto the baseline circular arc midspan sections. Theseresults illustrate that a 20 percent reduction inchord is possible for the customized airfoil andproduces a substantial decrease in trailing edgeshape factor and momentum thickness relative to thebaseline midspan section. In other words, customizinghas produced a lower solidity, slightly more heavilyloaded airfoil that is less susceptible to flowseparation and should perform more efficiently than abaseline midspan section.


+tt.r

Figure 5-1 Overall Logic for Optimizing AirfoilGeometry

3.0

stile. 6 •Ids"....tion

gnnme clnni,...c nueI-

z.o cte.,.ee r,,, t

zsTTTT-1 z .latn.n c:u.a rmuc un— fro• O,sel lne Oes lqn

• 0.020

^ 5 ■ .1 s— cnn.an f.a

o.ols zs xo is l0 5 0 e,:ri me eel iqn

2.0

— I I ooueb cu l,.. O, telln<

zs m is Iq s o • •+aso+n cnn.e ..a.ctlonIra O.tel the ees l9n

02.0 2.5 3.0 ]-S

IItYiMn 5e0r0 IegtM -ln.

Figure 5-2 Influence of Midpan Chord Reductionon Stator-6 Customized Airfoil Section

Figures 5-3 shows the midspan suction surface Machnumber distribution for the baseline and the custom-ized profile with a 20 percent reduction in chordlength.

Customizing moves the loading forward on theairfoil, thus effecting a more gradual diffusion tothe trailing edge with attendant reduction in skinfriction loss.

Midspan DCA and the 20 percent reduced chordcustomized sections are illustrated in Figure 5-4.

q..

ner^a+ona m St,ter L •IY+Mn s.t l len MW1. tt rt+l,r , s •re111e

i;L[

Figure 5-4 Stator-6 Midspan DCA and CustomizedProfile Sections

The reduction in skin friction or profile lossresulting from the use of velocity controlled airfoilswas estimated by computing the suction and pressuresurface boundary layer characteristics at the statortrailing edge. The expression, developed in Reference5 for this purpose, is as follows:

H. 1 H . 1ew _ B o (13H 1 ) e o C 1 • 3H 1 )

C cosil 3 C Colo a off

(5-1)

(1 c cos(1 ) Baseline (1 cos(i ) CustomizedOCA airfoil

Results of the reduction in midspan loss computa-tion for the customized airfoil sections (stators 6and 16) are presented in Figure 5-5. They show that a20 percent reduction in chord length, relative to thebaseline DCA section, produces a decrease in themidspan profile loss coefficient of 0.0035 for bothstators 6 and 16.

7

% .ld000, <no.e reeucnnn

Figure 5-5 Influence of Midspan Chord Reductionof Customized Airfoils on MidspanLoss Reduction

The foregoing loss model demon strated that a 0.0035reduction in stator midspan loss coefficient could berealized by the implementation of reduced midspanchord customized stator airfoils. This design conceptand its associated reduction in loss was applied toall stator rows.

Figure 5-3 Comparison of Stator-6 Surface Mach Number Stator End Bends. The stator end bend configura-Distribution on Baseline DCA and Custom- tions were evolved from an iterative computation that

ized Reduced Chord Midspaii Airfoil Sects included a new secondary flow loss model formulation.

8


This subsection contains highlights of this lossmodel, followed by a brief description of the proce-dure that optimized the stator reset angle geometry.

define the end bend geometry to be factored into thedetailed design configuration:

The method adopted for computing the change inendwall losses and deviation angle resulting from theincorporation of stator end bends was based on asecondary flow analysis (SECOND) similar to theclassical secondary flow theory (Reference 6).However, unlike Reference 6, the analysis employedhere takes into account the viscous flow effects onthe development of secondary vorticity through theblade row.

An implicit finite difference scheme is used tointegrate the secondary vorticity equation along aninviscid mid passage streamline passing through theedge of the boundary layer. The integration iscarried out twice, once for the hub region and oncefor the casing region. The numerical procedure forthe secondary flow computation needs the followinginput:

• Inlet boundary layer profile--This is computedwith the aid of Allison's three-dimensionalendwall boundary layer code (shroud) describedin Reference 7.

• Inviscid blade passage velocity distribution atthe edge of the boundary layer--BBFA was usedfor this computation.

• Spanwise distribution of exit flow angle--Thisis available from the meridional plane flowanalysis (MFA).

The secondary vorticity distribution computed inthe exit plane of the stator was determined by solvingthe Poisson equation for the secondary flow streamfunction as formulated in Reference 8. By integratingsecondary velocity in the circumferential direction,deviation angles associated with endwall boundarylayer secondary flow can be evaluated.

To compute the total pressure loss associated withthe secondary flow, it was assumed that all of thesecondary flow kinetic energy is dissipated andconverted into total pressure loss. This heuristicassumption needs some refinement and validation forcomputing the secondary flow loss in a cascade;however, it is considred to provide a reasonableapproximation for estimating the change in secondaryflow loss between the geometries. Based on thisassumption, the reduction in total pressure losscoefficient, due to the influence of reset closedstator end bends, was evaluated.

Results of the loss reduction computation for arepresentative stator (stator 10) is shown in Figure5-6. Also, for all of the geometries evaluated, thechange in deviation angle was of the order of onedegree. The loss computation shows a fairly largereduction in the endwall region, with levels dependingon the amount of reset and where the subject statorwas positioned in the compressor. Also, the regionlocated a small distance from the hub and tip wallexperienced a sizable reduction in loss coefficient.

Loss contours of the type shown in Figure 5-6 weredeveloped for all 19-stage stators and used in theiterative process that produced the final stator resetgeometry for all stages. The following stepwiseprocedure was used to optimize the reset angles and

9 w

Figure 5-6 Stator-10 Reduction in Loss CoefficientDue to Stator End Bends

1. With the aid of the three-dimensional endwallboundary layer code (SHROUD) stator inlet flowangle was calculated inside the endwall boundarylayer.

2. The endwall sections (hub and tip) of eachstator were reset to ensure that the incidencein the endwall boundary layer is less than orequal to zero.

3. The secondary flow analysis code (SECOND) wasthen used to estimate the endwall secondary flowincluding losses and flow angles for reset andbaseline blades.

4. The computed endwall region losses and exit flowangles were then input into the meridional flowanalysis (MFA), and the program was run in thepredict performance mode. Reduced losses forthe customized stator midspan sections were alsoincluded in this calculation; thus, the resultsof this analysis gave the total pressure ratioand efficiency of the complete advanced conceptdesign.

5. If the pressure ratio of advanced design wasdifferent from the design intent (14.0:1), theresets were adjusted, and steps (1) through (4)were repeated until the pressure ratio matchedthe design intent.

A three-dimensional view of the advanced conceptcompressor design final stator-6 geometry, includingend bends in addition to the reduced chord velocity-controlled midspan region, is compared with thebaseline DCA airfoil in Figure 5-7.

Incorporation of the airfoil reset geometry and therespective reduced loss coefficient profiles for thestator end bends and reduced chord customized midspansproduced a computed overall advanced concept compres-sor design adiabatic efficiency of 89.3 percent.

Although no separate investigation was carried outto customize rotor airfoils in the midspan and endwallregion, it is expected that incorporation of custom-ized rotor airfoils will produce higher levels ofefficiency improvements than that estimated forcustomized stator airfoils. This is because, for thesame change in the boundary layer momentum thickness,the change in the total pressure loss is larger in arotor than in a stator, due to the higher inletdynamic head and exit flow angle. % Based on this

cos fi t a APt( 2a c W g

9


argument, the overall adiabatic efficiency of theadvanced concept compressor design is projected to beat least one percent higher than the baseline design.This translates to an adiabatic efficiency of 89.8percent or a polytropic efficiency of 92.8 percent.

Trailing DCC sections

rage allradial sections

lip region

Leading edge

Center of gravity stack NuD region

Stator 6 baseline DCA airfoils

DCA sections--hub and tip

bailing(`educed chord

eoge {Ill tomlze0

sect tons at all

other LEtal Tipstations

region

Tip end Dena

Leading edge

Nub end DenE NYD region

T ratlfng edge stack

Stator 6 reduced chord. custantzea mid seen Section withhub and tie end bend airfoils

Figure 5-7 Impact of Incorporating Reduced ChordCustomized Midspan Sections with Hub andTip End Bends on Stator-6 Baseline DCAAirfoils--Advanced Concept CompressorDesign Final Stator-6 Design

Final Advanced Concept Compressor Design. Thedesign point projected performance of the finaladvanced concept compressor design is as follows:

• corrected inlet flow, W,(0/8

800 lb/sec• pressure ratio, R c 14.0:1• corrected wheel speed, N/VO

3600 rpm

• surge margin, SM

20%• adiabatic efficiency, qTT

89.8%• polytropic efficiency, (]poly 92.8%

Figure 5-8 presents the flow path and blading ofthe advanced concept compressor design.

Figure 5-8 Advanced Concept Compressor DesignConfiguration

A performance map was developed for the advancedconcept compressor design. The map, presented inFigure 5-9, illustrates the impact of speed variation

on overall performance. The off-design characteris-tics were estimated with the aid of extensive in-housedata base on multistage, fixed-geometry, high-pressureratio compressors. The EPRI compressor design pre-sented in this paper would be more suitable for avariable-geometry design. For a fixed-geometrysituation, it would have been more desirable to unloadthe front stages of the compressor to allow good surgemargin at lower corrected speed.

It

^ B

b

NO

0

200 400 600 800

Corrected floc rate, N+e/e--16/sec

Figure 5-9 Fixed Geometry Performance Map for EPRIAdvanced Concept Compressor Design

VI PHASE IV - OPTIMIZATION OF PART POWER OPERATIONWITH VARIABLE GEOMETRY

A variable inlet guide vane (IGV) is commonly usedin utility gas turbine compressors in order to 1)improve starting flexibility and 2) improve the heatrate at part-power in combined cycle. In aircraft gasturbines, it is common to have a number of variablestator stages (in addition to the IGV) to furtherimprove compressor operability at part-power and underchanging flight conditions. In a utility gas turbinecompressor, the power is reduced at the design rpm(3600) by reducing the inlet flow and pressure ratiothrough the closure of the IGV and stator vanes.

In order to find the possible improvement inpart-flow efficiency, the following three configura-tions of variable geomtry were considered in thisstudy:

1. Variable IGV only.

2. Variable IGV and stators 1 through 5 in a gangedmode (all vanes reset same number of degrees).

3. Variable IGV and stators 1 through 5 in agraduated mode (In this configuration, for eachdegree rest of IGV, stators 1 through 5 werereset 5/6, 4/6, 3/6, 2/6, and 1/6 deg.respectively.)

To do this study, the IGV reset was varied from 5deg to 30 deg. The operating point of each reset wason a constant Rc/W line (which corresponds to constantturbine inlet temperature) prescribed by the designpoint value. An iterative procedure using the mer-idional flow analysis code (MFA) was adopted forestimating the mass flow and efficiency correspondingto each reset. Extra losses were added to account forseparation due to high incidence; at the same time,

in

10


the flow was dropped to ensure that the compressor wasoperating in an unchoked mode.

The primary results of this analysis are presentedin Figure 6-1, which compares the efficiency decrementas a function of flow for the above three variablegeometry configurations. It may be seen that eitherof the two variable stator configurations is highlysuperior to the IGV-only configuration. At 75 percentflow, there is approximately one-third as much decre-ment to efficiency as for the IGV-only, while the flowcan be reduced to about 60 percent before the effi-ciency decrement is the same as that for the IGV-onlyat 75 percent.

7 p

a.sx

Figure 6-1 Efficiency Decrement versus FlowReduction for Three Variable-GeometryConfiguration

VII CONCLUSIONS

Based on the results of this three-phase designstudy effort to develop an aerodynamic design for alarge utility gas turbine engine compressor, thefollowing conclusions can be drawn:

1. Using conventional design techniques, a prelimi-nary compressor design was developed and refinedto produce a polytropic efficiency of 92.1percent. This level of efficiency potential wasachieved by having a large number of lightlyloaded stages with low solidity airfoils.

2. Several advanced design concepts were investi-gated, and stator midspan velocity-controlledairfoils, stator end bends, and stator leadingedge sweep were shown to hold promise forincreasing overall compressor efficiency.

3. Analytical studies on the velocity-controlledstator midspan sections showed that the custom-ized airfoils would operate with less loss andbe less susceptible to flow separation, evenwith a 20 percent reduction in solidity, than abaseline DCA airfoil.

4. Incorporation of geometries and losses associat-ed with stator end bends and stator customizedmidspan sections resulted in a 0.5 percentimprovement in adiabatic efficiency. An addi-tional improvement of 0.5 percent in adiabaticefficiency is expected to result from thecustomizing of rotor airfoils.

5. Estimated performance of the 800 lb/sec, 14.0:1pressure ratio EPRI advanced concept compressordesign was:

-- surge margin, SM >20%

adiabatic efficiency, qTT 89.8%

polytropic efficiency, gpoly 92.8%

6. Incorporation of variable geometry in the IGVand the first five stators of the advancedconcept compressor design demonstrated that thecompressor flow could be reduced by 25 percentwith a 4 percent efficiency penalty. Thecorresponding efficiency penalty for the vari-able IGV-only configuration was estimated to bein excess of 10.7 percent.

REFERENCES

1. Johnson, I.A., Bullock, R.O., et al., 'Aerody-namic Design of Axial Flow Compressors," Nation-al Aeronautics and Space Administration,NASA-SP-36, 1965.

2. Koch, C.C., "Stalling Pressure Rise Capacity ofAxial Flow Compressor Stages," ASME Journal ofEngineering for Power, Oct., 1981.

3. J.L. Bettner and A.K. Sehra, "High EfficiencyAxial Compressor," EPRI Report No., AP-4943,Dec., 1986.

4. M.R. Head, "Entrainment in the Turbulent Bounda-ry Layer," ARC Technical Report, RM3152.

5. S. Lieblein and H. Roudebush, "Theoretical LossRelationships for Low-Speed, Two-DimensionalCascade Flow," NACA-TN-3662, March 1956.

6. H.B. Squire and K.G. Winter, "The Secondary Flowin a Cascade of Airfoils in a Nonuniform Flow,"Jour Aero Sci., Vol 18, pp 271-277, 1951.

7. J.L. Bettner and E.R. Turner, "Shroud FlowCharacteristics," AFWAL-TR-82-2038, June 1982.

8. W.R. Hawthorne and R.A. Novak, "The Aerodynamicsof Turbomachinery," Ann. Rev. Fluid Mech.,Vol. 1, 1969.

11


Design of a High Performance Axial Compressor for Utility Gas ...

Documents

Transcript of Design of a High Performance Axial Compressor for Utility Gas ...