467219

International Journal of Rotating Machinery2001, Vol. 7, No. 5, pp. 365-374Reprints available directly from the publisherPhotocopying permitted by license only

(C) 2001 OPA (Overseas Publishers Association) N.V.Published by license under

the Gordon and Breach Science Publishers imprint,member of the Taylor & Francis Group.

Recontouring of Jet Engine Compressor Bladesby Flow Simulation

HERWART HNENa* and MATTHIAS PANTENb

aAachen University of Technology, Aachen, Germany; bLufthansa Technik AG, Hamburg, Germany

(Received 30 May 2000," In finalform 31 May 2000)

In modern jet propulsion systems the core engine has an essential influence on the totalengine performance. Especially the high pressure compressor plays an important rolein this scheme. Substantial factors here are losses due to tip clearance effects andaerodynamic airfoil quality. During flight operation the airfoils are subject to wear andtear on the leading edge. These effects cause a shortening of the chord length and theleading edge profiles become deformed. This results in a deterioration of the engineefficiency performance level and a reduced stall margin.The paper deals with the re-contouring of the leading edges of compressor airfoils

by application of a new developed method for the profile definition. The commonprocedure of smoothing out the leading edges manually on a wheel grinding machinecan not provide a defined contour nor a reproducible result of the overhaul process. Inorder to achieve optimized flow conditions in the compressor blade rows, suitableleading edge contours have to be defined for the worn airfoils. In an iterative processthe flow behavior of these redesigned profiles is checked by numerical flow simulationsand the shape of the profiles is improved. The following machining of the new definedleading edge contours is achieved on a grinding station handled by an appropriatelyprogrammed robot.

Keywords: Blade re-contouring; Airfoil wear; Engine overhaul; Leading edge erosion; Flowsimulation

INTRODUCTION

Due to erosion effects the compressor blades ofjetengines become shortened and the leading edgesare deformed. These deformations influencethe flow behavior in the rotor blade passages.The operating range becomes smaller and the

aerodynamic losses are increased. For a JT8Djet engine detailed investigations have been per-formed by Roberts (1984). With increasing wearof the compressor blades the usable angle rangedecreases. The suction side separation occurs atlower incidence angles. Thus, the operating rangeof the whole compressor becomes limited. In

* Corresponding author. Tel.: +49(0)241 80 5522, Fax: +49(0)241 8888-229, e-mail: [email protected]

366 H. HNEN AND M. PANTEN

addition, the aerodynamic losses are increased dueto separation bubbles at the leading edge.By a suitable refurbishment procedure worn

blades could be made available for additionalrunning cycles. The aim of this process is toproduce a high profile quality in order to provideoptimal flow conditions. By the todays state of theart method of rounding the worn leading edgesmanually either defined contours nor reproducibleresults are obtainable. This results in a wide spreadof overhaul quality and partly to increased lossescompared with the original blade. Especially forthe application of manually smoothed blades withsignificantly reduced chord length considerableperformance losses in the compressor have beenobserved. In order to keep the compressorperformance in a tolerable range in many cases aminimum chord length has been defined by theuser. So, even if the structure limit, defined by thejet engine manufacturer, is not yet reached theseblades have to be rejected. This causes additionalcosts for new parts.

This demonstrates that a defined machining ofthe leading edges for optimal flow conditions inthe bladings is required. Therefore, a new methodfor the refurbishment of compressor blades, the"Advanced Recontouring Process" (ARP), hasbeen developed. This process is based on detailedstudies of blade wear and the resulting flowbehavior in the rotor blade rows of the highpressure compressor of a CF6-50 turbofanengine. The aim is the definition of an opti-mized leading edge geometry for worn compressorblades with respect to a high aerodynamicquality and a reproducible machining of theprofiles.

data of the jet engine manufacturer. In order toprovide a suitable information about the flowconditions at each step of the ARP, a Navier-Stokes code was applied. A description of programstructure and its application is given by Benetschik(1993). In a first step the flow calculations with thiscode were performed using the original profilecoordinates and the design flow data provided bythe jet engine manufacturer. With these data aparameter study has been carried out to verify andto optimize the numerical results.For the planned flow calculations a good

resolution of the profile leading edges was re-quired. Therefore, an O-grid configuration with33 x 101 grid points has been chosen (Fig. 1). Withthis grid type contour fitted elements especially inregions with small curve radii can be achieved sothat numerical failures caused by influences of thegrid can be minimized. In order to achieve a highresolution in the boundary layer region the griddensity was increased close to the profile.

Figure 2 shows a calculation result for thedesign conditions of a 5th stage rotor blade. Theflow calculations for this case were performed withthe target to reproduce the design flow data of thejet engine manufacturer (flow angles, pressure rise,Mach No.s) in order to calibrate the numericalcode. These simulations have been performed foreach rotor blade row. The final results of thesewere stored in a data base as a reference case forthe later justification of the later ARP results. Inthe following the optimized program parametersfound for each blade row were applied to thecalculations of the corresponding worn and rede-signed blades.

DEFINITION OF REFERENCE DATACONTOURS AND CLASSIFICATIONOF WORN AIRFOILS

The aim of the ARP is the production of re-contoured blades with an aerodynamic behaviorsimilar to the original design conditions of thecompressor. The justification of the quality of therefurbished airfoils had to be based on the design

For the rotor blades of the high pressurecompressor of a CF6-50 jet engine numerous wornblades of the rotors of the stages 2 to 14 have beeninspected. The measurements demonstrated differ-ent types of leading edge contours. Figure 3 shows

RECONTOURING 367

FIGURE Numerical grid for the flow calculations (small picture: zoomed leading edge region).

O9

O8

07

O6

O5

./

0.0 0.5 1.0relative chord length

FIGURE 2 Pressure distribution of a new part profile (5th stage).

368 H. HONEN AND M. PANTEN

FIGURE 3 Variety of leading edge contours of worn airfoils(5th stage).

the variety of abrasion profiles for the 5th stagecaused by different operating and flow conditions.The leading edges of the worn airfoils in somecases look nearly cut away or are unsymmetrically

deformed. The point of maximum extension at theleading edge is marked in the drawings. In all casesthe stagnation point is shifted from its originallocation. From these investigations 9 differentclasses for the leading edge profiles could bedefined (see Fig. 3).A comparison of the calculated vector plots of

a 10th stage new part profile and a profile withdeformed leading edge demonstrates the signifi-cant change in the flow distribution around theleading edge (Fig. 4). In this case the stagnationpoint of the worn blade (right plot) can beclassified with form "A" (see Fig. 3). The flattenedleading edge causes a large stagnation zone. Astrong acceleration to both sides of the profilefollows. This leads to a separation bubble on thepressure side.

In addition, the investigation of the rotor bladesof all stages showed that the thickness differs ina wide range. Due to deviations of the new partblades and the erosion during operation differ-ences of up to 4-0.3 mm in comparison with thedesign value could be detected.One main reason for the different types of

erosion was found in the profile quality of theoriginal airfoil. Recycled blades with manually

/

//i/"

i

,t

FIGURE 4 Comparison of the vector plots for a new part (left) and a worn (right) rotor blade.

RECONTOURING 369

FIGURE 5 Different leading edge shapes of manually smoothed blades.

smoothed contours show a large variation rangeregarding the geometry (Fig. 5). This partly resultsin unsatisfactory flow behavior in the leading edgeregion due to the position of the stagnation pointand the following acceleration towards the suctionside and pressure side surfaces. In these regions theerosion increases due to the high impulses of thedust particles (see Fig. 4). Therefore, an optimizedre-contouring quality not only provides low aero-dynamic losses but also longer life times of theprofiles.

In order to obtain the highest quality in re-contouring results the redesigning process forthe leading edges would have to be performedindividually for each blade. This would require anenormous effort regarding the numerical flowsimulations, the programming of the robot grind-ing station, and the handling of the blades.Therefore, numerical parameter studies have beencarried out to investigate the influence of the

deviations of the blade thickness on the flowbehavior. The aim of these studies was to defineonly a few geometry classes for each blade rowwhich can be characterized by one typical profileeach. It was found that for most of the stages twoclasses are sufficient for reliable re-contouringresults.

DEFINITION OF NEW LEADINGEDGE CONTOURS

The most significant part of leading edge erosionhas been detected at radii above mid-span withincreasing deformation with approach to the bladetip. Since the effects of blade erosion are primarilyseen on the outer one-third of the span (near thetip), a characteristic radius for the robot waschosen at 90% blade height. This also greatly re-duced the redesign programming effort.


The definition of new leading edge contours forthe worn airfoils is based on the design flow dataprovided by the engine manufacturer. In order toproduce re-contoured blades with a flow behav-ior similar to the new part blades the followingparameters have been defined for comparison withthe design conditions:

pressure rise of the blade rowturning angleMach No. Distributionlift

Additionally, the loss production of the bladerow is an important parameter. The efficiency andperformance of the redesigned airfoils should bekept close to the new part profiles.

These parameters can be checked applying thedesign flow conditions of the different stages to theredesigned airfoils. Under optimum conditionsthe above mentioned results meet the referencestored in the data base of the new part profilesexactly. In order to validate the off design behavior(i.e., take off conditions), the usable incidenceangle range of the re-designed profiles also had tobe proved.

In a parameter study the applicability ofdifferent geometric shapes for the replacement ofthe deformed leading edges were investigated.Suitable geometry shapes are circles, ellipses,parabola, and hyperbolas. In order to providecontours with smooth transition from the newleading edges to the remaining profile alsocombinations of these curves together withstraight sections were tested. The new leading

edge contour has to fit into the remaining materialof the worn airfoils and from an economicalview its definition has to consider the followingconditions:

machining of the leading edge contour only upto 10% chord lengthminimized abrasion of material during machin-ing process

The easiest solution to meet these conditions is aleading edge shape based on a circular construc-tion. It could be fitted into the remaining profilewithout supporting lines or other curves. From thepart of calculation results however it becomesvisible, that the flow patterns in the region of acircular leading edge are not sufficient. A compar-ison of the manually smoothed blades (see Fig. 5)shows also a nearly circular shape for all thevarious machining results. It becomes clear thatthe geometry has to be optimized and that there isa rather high potential of improvement.

Figure 6 shows an example of a redesignedleading edge using a hyperbola construction. Themarkers indicate the different parts of thisconstruction. On the pressure side (upper side)the hyperbola fits directly to the contour of theoriginal airfoil. On the suction side (lower side)a straight line and a circular part are necessaryto provide a smooth transition from the hyper-bola to the remaining profile. The connectionpoint is located at about 30% of the chordlength. This causes large region of machiningwith high abrasion of material. Nevertheless, theaerodynamic quality of the redesigned profile

FIGURE 6 Redesign of the leading edge contour based on a hyperbola.

RECONTOURING 371

was found to be nearly as good as the new partprofile.

This example demonstrates that a suitablecompromise between aerodynamic and economic-al aspects had to be found. Therefore, a parameterstudy has been performed to achieve a sufficientsolution for the construction of the leading edgecontours. A shape based on an ellipsis providedthe best results. By variation of the geometryparameters the optimum flow behavior is obtain-able. With this method the necessary machiningrange could be reduced to about 10% chordlength.

Detailed experimental investigations on bladeswith elliptical leading edges performed by Walrae-vens and Cumpsty (1993) demonstrated the originof leading edge separation bubbles and lossproduction. By a suitable design of the leadingedge an optimization of the performance couldbe obtained. Therefore, different leading edge de-signs shape based on ellipsis constructions weretested.

Figure 7 shows the calculated profile pressuredistributions for a redesigned profile with ellipti-cal leading edge. The corresponding flow vectorplot for the leading edge section show flow condi-tions similar to a new part profile. Detailed studieshave been performed for the 5th stage rotor inorder to optimize the re-designing method and toevaluated the influence of the different geometryparameters.

In the following the method was applied to therotor blades of all stages and all profile classes. Foreach case satisfactory results were obtained. Basedon the experience an algorithm for the definitionof the new leading edge contours was developed.This program allows an automated check of theremaining profile, the definition of new leadingedge contours, and the check of the boundaryconditions. Thus, a computerized parameter var-iation for the optimization of the leading edgecontours becomes possible. The scheme of thedesigning process for the re-contoured blades isshown in Figure 8.

0.0 0.5 1,0ss relative chord length

FIGURE 7 Profile pressure distribution and vector plot of a redesigned 10th stage rotor blade.


measurement of the geometry ofa statistical no. of worn blades

analysis and definition ofprofile classes (clustering)

definition of newLE contour

Flow calculationwith Navier Stokes

modification of newLE contour

geometry and flow datafrom jet engine manufacturer

flow calculation with Navier Stokescode applying design data

machining process

FIGURE 8 Scheme of the redesigning process.

MACHINING OF THE LEADING EDGECONTOURS

In order to produce the leading edge contours asdefined by the above mentioned method theairfoils are machined on a robot handled grindingstation with a smooth grinding wheel. A moredetailed description of the procedure and aneconomical discussion is given by Panten andH6nen (1998). By this method it becomes possibleto machine the different profiles of each class usingonly one definition of the target contour. Due tothe above mentioned geometry deviations of theworn airfoils the correlation between target andre-contoured profiles is not 100% exact but themethod of defining a characteristic target contourfor each cluster in combination with using asmooth grinding wheel is a very good approxima-tion. Investigations of a statistical number of

re-contoured blades have shown that approx.95% of all blades treated with the ARP are nearlyidentical to the defined target contour. The qualityof the remaining 5% was similar to that of theconventionally smoothed blades. Figure 9 showsthe results of a Navier-Stokes calculation for aworn and a machined blade.The sharp drop of the pressure-side pressure

distribution is caused by a separation bubble inthis region (see Fig. 4). The performance issignificantly decreased and the operating condi-tions defined by the design case can not bereproduced in the calculations. The lower pressuredistribution in Figure 9 belongs to a machinedprofile which was based on redesign data pro-vided by the above mentioned method. The com-parison with the calculation result of the targetcontour profile (Fig. 7) demonstrates the goodagreement between redesigned and hardware

RECONTOURING 373

0

09---

08o

._> 07

O5

0.0 0.5 1.0relative chord length

FIGURE 9a Calculated profile pressure distribution for a worn rotor blade.

10

0) 09

-

08o

> 07

06

0 5

ssrelative chord length

lOS

FIGURE 9b Calculated profile pressure distribution for a machined rotor blade.

contour. By a large number of inspected machin- for a new part, a worn, and a recontoureding results during the production process the blade based on the loss coefficient of the designreliable quality of the ARP profiles could be case. For the worn blades a strong increaseproved. (50%-95%) of the losses can be observed. The

Figure 10 shows a comparison of the blade el- redesigning process improves the performanceement performance in different stages. The dia- significantly. The losses could be reduced to thegram shows the calculated relative loss coefficients level of the new part blades. Only in stage 8 the


#4 #8 #12 #13stage no

new partwornE]ARP

FIGURE 10 Comparison of the losses of.different blade conditions in various compressor stages.

increase of losses remains quite high. These resultsindicate a good improvement of compressor per-formance by applying the ARP. This loss behaviorof a blade row however can only be judged in com-bination with the other flow parameters. As men-tioned before some main parameters have beendefined which have to be kept in the range of thedesign profile. This master condition for the re-design of the blades guarantees an operating be-havior of the blade rows similar to the originallydesigned profiles.

CONCLUSIONS

By means of a new developed method the overhaulprocess for compressor blades could be improved.Measurements of the geometry of a statisticalnumber of blades demonstrated remarkable devia-tions of the chord length and blade thickness.Classification parameters have been defined for aclustering and definition of characteristical bladeshapes for each cluster. These master profiles weresubject to a re-contouring of the leading edges inorder to improve the aerodynamic quality in

comparison with the conventional smoothing byhand.The redesign procedure based on CFD calcu-

lations defines master profiles with new leadingedge contours. In the following machining pro-cess these data are used to re-contour the worncompressor blades on a robot controlled grind-ing station. Based on detailed numerical investiga-tions a satisfactory compromise between anoptimized aerodynamic quality of the redesignedblades and an economical machining could beprovided.

ReferencesBenetschik, H. and Gallus, H. E. (1990) Inviscid and ViscousFlows in Transonic and Supersonic Cascades Using anImplicit Upwind Relaxation Algorithm, AIAA-PaperNo. 90-2128.

Panten, M. and H6nen, H. (1998) Advanced RecontouringProcess for Compressor Blades, RTO Meeting Proceedings17, Qualification of Life Extension Schemes for EngineComponents, RTO-MP- 17.

Roberts, W. B. (1984) Axial Compressor Restoration by BladeProfile Control, ASME Paper No. 84-GT-232.

Walraevens, R. E. and Cumpsty, N. A. (1993) Leading EdgeSeparation Bubbles on Turbo-machine Blades, ASME-PaperNo. 93-GT-91.

Submit your manuscripts athttp://www.hindawi.com

VLSI Design

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of

RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal of

EngineeringVolume 2014


Shock and Vibration


Mechanical Engineering

Advances in


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of


Distributed Sensor Networks


The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Active and Passive Electronic Components


Chemical EngineeringInternational Journal of

Control Scienceand Engineering

Journal of


Antennas andPropagation




Navigation and Observation


Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

RoboticsJournal of


467219

Documents