The - ICAS International council of aeronautical sciences · leading to ingestion of abrasive material. Over a period of time, relatively low dust concentrations ... can occur more

ICASPAPERNO.„ 0,

SAND EROSION OF GAS TURBINES

byG. P. Tilly, Head, Mechanical Behaviour Section

National Gas Turbine EstablishmentPyestock, Farnborough, U. K.

TheEighthCongress01the

InternationalcouncilofIheAeronauticalsciences

INTERNATIONAALCONGRESCENTRUMRAI-AMSTERDAM,THE NETHERLANDS AUGUST28 TO SEPTEMBER2, 1972

Price:3. Dfl.

!

á

._;AND ..,A)3lON OF GAS

G. P. TillyNational Gas TurOin(: le,Jtanlisnment

Pyestock, Farnborouva, ilampshise England

Abstract

Operation of ras turbines in dusty environ-ments can involve the creation of dust cloudsleading to ingestion of abrasive material. Over aperiod of time, relatively low dust concentrationscan cause severe erosion damage to the compressorblading. In civil aircraft, this results in pre-mature removal of bladinr leading to increasedoperational costs. In helicopters, usuallyinvolving more damaging environments, the erosioncan occur more rapidly and lead to engine failuresin relatively short times.

The relationships between erosion and theimpact conditions are considered and the equationsgoverning the behaviour are given. The factorsdetermining the erosiveness of natural sands aresummarised. Using these relationships, it isshown that the damage produced in engines testedunder controlled environments can be related tolaboratory erosion data.

Consideration of methods of minimisin- ero-sion indicates that there is little scope for useof materials better than the martensitic steelsand nickel alloys currently used in later stagesof compressor bladin. Lstimates of the derree ofprotection to be obtained through use of intakefilters indicate higher figures than are achievedin service. A helicopter trial subsequentlyshowed that this it due to failure to achieve thefiltration efficiencies given in controlled lab-oratory testing.

1. Introduction

Severe damage can be caused in gas turbineengines due to ingestion of solid material varyinv,in size from sub micron dust to pebbles anr: stones.Particles smaller than about 1000Lm aro lisuallytermed dus.t or sand and continued ingestion ofsuch material can lead to progressive crosier. ofengine components. This damage i5 usually con-fined to the compreseor and can occur irreepecliveof the engine size or type i.e. illethor contri-fugal or axial flry:. In extreme caeen, LINe ero-sion can involve removal of eufficient material tocause components to be m.,chanically unsafe out itis more often manifested by modification of aero-dynamic profiles leadiro; to reduceeIn the case of extol bladirrL; of emall engines,edge radii can be as little as 50 to 1;',Orn and theblade profiles can be cut back as shown in Figure1. Initially, erosion can produce some improve-ment in engine performance as the blades orepolished and sharpened. Further damare, hoever.causes progressive de)7ralation and can eventuallylead to loss ef surge margin and inadequate power.Large engines tend to be less susceptible to ero-sion because the bigger blading can Jithstand aproportionately larger loss of material beforebeing affected e.g. It has been calculated thatlosses of up to 5 mm in the edges of fan blading

STAGE 1 STAGE 2

FIG.1 TYPICAL EROSION OF ROTOR

BLADES IN A CIVIL ENGINE

or lori:e (Ie.ines can be loleratet. nut uiffer-

ence le, of course, partially offset by tae higherairflo:: and obrlity to lift birrer objects.

Intere:tis trio csosion of aero eritn(Li tan

Inten:•Ified GU:•1!"..:Lilt. 1960s .her, helicoptersbecome extensiv,.ly used in desert environments:.Overhaul lives ierr. dsaeticolly :,-duced and it eas

to replan, comrsessor blading afterselativ,,li :host r,drioes of tim,(1),(?). In opera-t. ove.• South .:Ist A..ia, it hne been reported t iat overhaul Livee of 10(y) to 1(0) hours .ere

leducei te about W hours leadine to annual costsof m:Ilione of Hollars('J. hritish experi nce in

Ailri.ca ens civally brii, overhaul lives beings,duced f..om 10O0 hours to as little as 75 hours.

Operation of civil aircraft from relativelyaltnodgn apparently less qama7ing.

cnn aLeo thyCilvr, Yl'O:i1ON damage because theolearrir conOil.lons are offset by the greater num-ber Dr flii;iits demane,H for conomic operation.Comparatively sevei.o conditions can be encountered'Iurin;; icy .:eather iihen it is necessary to gritthe .•u.i. o:75. Although efforts are made to ensurethat the grit is larre enou,;h to remain on theground, the particles tend Lo become splintered

- 1 -

and broken up. Some airports are sited in inheren-tly dusty locations and there have been occasionswhen engines have been severely damaged by dustingested during ground manoeuvres. Such damagenecessitates early removal of blading and incursexcessive cost through loss of operationalusageas well as for the replacementparts. In a recentanalysis of experience with BEA, it was found thatingestionof debris cost--£250,000of which about£170,000 was directly attributedto erosion ofcompressorblading(4).

As a result of these service problems, therehas been extensive laboratorywork to explore theerosion parameters in an effort to find ways ofminimising the damage. Broadly, the approacheshave been (i) studies of mechanicalbehaviour tofind ways of designing to avoid potentiallyeros-ive situations, (ii) developmentof materials andcoatings having a good erosion resistance;and(iii) filtrationof the incomingair to remove theerosive particles. In this Paper,these approachesare considered with respect to service experience.

II. Factors that influenceerosion

In laboratory studies of erosion, it is con-venient to measure losses by weight and it hasbecome conventionalto express erosion as the non-dimensional quantity, weight loss per unit weightof sand impacted,designatedby C. However, sig-nificant damage to engines is through dimensionalchanges rather than weight losses, and volumetricerosion, ev, is a more relevant quantitywhich can

be calculated as - where p is the density of the

target material. Use of Ey has the additionalmerit that it enables comparisonsto be made ofmaterials having differentdensities.

The magnitude of the erosion is dependentuponthe conditions of impact and the mechanical prop-erties of the erosive particlesandtarget material.

Conditionsof impact

In laboratory testing,considerableattentionhas been given to conditionsof impact such asvelocity, strike angle, duration of exposure andconcentrationof erosive particles.

Numerous investigationshave been made of theinfluenceof velocity and it is generallyacceptedthat erosion is related to velocity (V) by thesimple power law expression

e = aV ....(1)

where a is a material constant. Most investiga-tions have shown that the exponent,a, is 2 orwreand recent work haa indicatedthat it is --2.3forawide range of materials(5),(6),(7).However, it ispossible that some brittle materials may have ahigher value. Unlike rain erosion, the thresholdvelocity, below which no erosion occurs, is verylow and can be ignored in most engineeringsitua-tions, e.g. a value of 5.3m/s has been deducedfor 600pm iron pellets against glass(B) whereasimpact velocities in compressorsare typically 250to 450 m/s.

The dependence on strike angle is differentfor brittle and ductile materials. Brittle mater-ials such as fibreglassesor ceramics,are mostsusceptible to impacts at 90° to the surface andthe angle dependence can be expressedby a rela-tionship of the form

se= b sina e ....(2)

where b is the erosion at 900 and can be derivedfrom Equation (2) as aV3:

Ductile materials such as plastics and manymetals are more susceptibleto glancing impacts.In cases where there is zero erosion for 900impactstheoreticalanalyses(9),(10),(11)have shown thatthe situation can be representedby an expressionwhich approximatesto

nese = c coma e sin — for 0 < e.

2e.

and se = f cos3.8 for e > eo

where c and f are material constants,a is usuallyassumed to be 2.0 and 00 is the angle of impactwhen the component of velocityparallel to the sur-face is zero after impact. In practice,mostmaterials exhibit significanterosion for 90°impacts and it is usually assumed that this is dueto a brittle componentoccurring at the same timeas the ductile erosion. Equations (2) and (3) canbe combined to give.

for 8 < eo

so = b sin°.e + c cosme sin°.Al

and for e > 00 ....(4)a

b sina 0 + f cos 0

This type of behaviour is exhibitedby the engineer-

ing alloys commonly used in gas turbines (Figure2).

30 60 90IMPACTANGLE—degrees

FIG.2 INFLUENCEOF IMPACTANGLEON THEEROSIONOF BRITTLEAND DUCTILEMATERIALS

(60 125Aim QUARTZAT100m/ sec)

0

_ (a) ALUMNA (SINTOX)

-04

02

-03

02

01

DERIVEDFROM EQUATION2

-

--

BRITTLE DCMPONENT

\ (W HAs snml

DERIVEDFROM EQUAT10144

DUCTILEaMPONENT

- 2 -

The relationshipsbetween erosion and angle ofimpact have been derived from tests on flat speci-mens but it has been shown(12) that erosion ofcurved surfaces can be estimated from such data byintegration over the exposed area i.e.

co sin 0 d ....(5)

This was confirmed by tests on an aluminium cylin-der tested with 60 to 125pm quartz at 100 m/s.However, it is often far from clear what the angleof attack actually is because the particles do notnecessarily follow the air stream. In practice,large particles tend to ignore curved airflowwhereas very small particles can follow it closely.Estimates(12)of the behaviour from a numericalanalysis of the particle flow suggested thatquartzparticles above -20pm will ignore the flow linesaround flat plate and cylindricaltargets at 100m/s. However, the breakaway size is itself depen-dent upon features such as the material density,velocity and the radius of curvatureof the flow.

Erosion can exhibit an incubationstage whenthere can be an initial weight gain as particlesare embedded in the surface. This is followed byincreasinglosses which eventually stabiliseat aconstant rate. Soft materials such as aluminiumand plastics commonly exhibit an extreme form ofthis behaviour. Harder materials can have a smallincubation stage but it is negligible in nickelbase alloys and the martensitic steels commonlyused in later staEes of compressor:,.From experi-mental observations it appears that depositionoccurs throughout the process and it is reasonableto assume that deposition and erosion can be trea-ted as separate components. If deposition isassumed to obey a poaer laa dependence,the rela-tionshipbetween nett erosion and weight impacted(m) can be given by an expression of the form

-1e c - d mP ....(6)where :3is less than 1. This is shown with res-pect to experimental data in Figure 3 using avalue for of 0.88.

Laboratory tests can exhibit a small concentra-tion effect whereby erosion decreases with increaseddust concentration. However, this is a relativelysmall effect involvinga 50 per cent reduction inerosion for a 40/1 increase in concentration(13).Tests(14) on a small gas turbine using values of 1to 7 mg/cu ft failed to detect any effect and it isusually neglected for practical purposes.

In compressors,temperaturesup to -450°C andstresses up to 200 MN/m2 can occur and it has beennecessary to determine the influence of these para-meters. Temperature, (T), can raise or lower ero-sion depending on the material involvedand it hasbeen shown that the dependence can be related tothe elongation at fracture,e

6 T ' E

so that increased ductility is associated with lesserosion(12). In one of the few studies of theeffect of tensile stress, it was found that valuesof up to 3Q0 MN/m2 produced little change to theerosion(12).

Material properties

For the impacting particles, the propertiesthat influence erosion are essentiallysize,density, friability,hardness and sharpness.Studi9s of a wide range of natural and syntheticsands5) have shown (0 erosion is dependent onparticle size, (ii) quartz tends to be the most ero-sive constituent in natural sands so that the ero-siveness of a given location can be characterisedby quartz content; and (iii) the quartz contenttends to decrease with particle size so that finesize ranges tend to be relatively innocuous. Thus,the erosiveness of sand from a given geographiclocation can be estimated from its size distribu-tion and quartz content for a given size (gd) using the expression

(:)(d)6 d Qd ....(8)

ahere Ed is the erosivenessof a given particlesize (d). The relationshipbetween erosion andparticle size is itself Oopendent upon the impactconditions and properti(:sof the target material.Ductile materials exhibit increasingerosion withparticle size up to a critical size beyond which nofurther increase occurs and it remains constant.Brittle materials can exhibit continuously increas-ing erosion according to a simple power law(13).With regard to the relationshipsbetween quartz con-tent and particle size, it is interestinEto notethe similarity bet.aeenthe three samples from dif-ferent geographic locations::hownin Figure 4.

100

LI

4 7 0

0

-e

- 4

- SO

7

-40

I a I TYPE A MATERIAL AT IGO / sac

lb) TYPE B MATERIAL AT 240 m/soc

/

/ X

,11///

x/ xxx

jr/1(

COLIATOI 6

*40 I S 0 40

*vote mmcno-g

CCS4TERC1ALLYPUREQUARTZ 0 AMA, ICAT11 AFRICA

VALLEY)X LONG

COSNWALLENGLAND

NATURAL SA1CG

4.

10 100 1000FIG. 3 INCUBATIONBEHAVIOURFOR TWO TYPESOF PARTICLE SIZE -

POLYURETHANE USING 125-150pn QUARTZ

FIG 4 RELATIONSHIPSBETWEENQUARTZCONTENTAND PARTICLESIZE

- 3 -

In general, it can be seen that there was lessthan 50 per cent quartz present in particlessmal-ler than 10um whereas there was over 90 per centin material bigger than 100pm. In usingrelation-ships such as in Equation (8) it should be notedthat chemical analyses alone are insufficienttocharacterisethe material and it is also necessaryto identify the constituentsmineralogically,e.g.a high A1203 content does not necessarilysignifythe presence of hard abrasive alumina. Samples oflateritic sands from South East Asia although hav-ing relatively high A1203 and Si02 contents,werefound to be very friable and relatively innocuousas regards erosiveness. However, other problemswere incurred because the dust very readily dep-osited and built up on the blading to produce biglosses in efficiency.

There is a very wide range of erosion resist-ance exhibited by differentmaterials from thevery susceptiblematerials such as glass (100 cc/kgfor 125 to 150p.mquartz at 800 ft/s and 90°impacts) to the engineeringalloys such as Nimonic90 at 0.6 cc/kg. However, testing of the types ofmaterial typically used in compressors,has shownthat they exhibit similar weight erosion i.e. 4.6mg/g. Thus, a close approximationof the volu-metric erosion can be obtained from use of therelevant density. This approach has been shown tobe valid for alloys of aluminium,titanium, nickeland martensitic steels.

Various attempts have been made to relateerosivenessto other mechanicalpropertiessuch ashardness, tensile strength, ductilityand impactenergy. However, these have met with only limitedsuccess such as the correlationgiven by Equation(7) and there is no generallyaccepted relation-ship in current usage.

Correlationwith engine testing

Correlation of the erosion produced in enginetests with laboratorydata can only be achievedsatisfactorilyif the impact conditionsand natureof the erosive material are known. From consider-ation of the behaviour of particles in an air-stream(12) discussed in 'Conditionsof impact', itseems reasonable to assume that particles largeenough to cause significantdamage are uninfluencedby the airflow so that the impact conditionsaredictated by features such as the velocity relativeto the target, and subsequent rebound behaviour.There is considerableevidence that in multi-stagecompressors,the particles impact in early stages,splinter and impact again in later stages.Towardsthe end of the compressor the particles tend to beprojected into the outer annulus and are generallysmaller due to fragmentationso that damage isconcentratedtowards blade tips. Unle:isa uigni-ficantproportion of the dust avoids impact alto-gether, the fragmentationand multiple impactsarelikely to produce an unexpectedlyhigh erosion.

Comparatively few engine tests have involvedmeasured erosion losses and there have been evenless systematic studies of the erosion parameters.In tests on a 45 hp engine, Montgomeryand Clar;tk14)produced values of erosion as a functionof par-ticle size. Usiggerosion data from the NGTEwhirling arm rig0), together with the relation-ships given in Section II, values of erosion forthe different sizes of dust can be estimated. Theresulting figures are compared with the measurederosion in the rotor and diffuser, where it is

reasonable to assume that multiple impacts will notoccur,and a reasonablygood correlationis obtained(Table I). Similarly, tests on a Rover 1S/60 rotorby Duke(15), at differentvelocities,also corre-late with the estimates.

Engine Test condition

Erosion - mg/g

MeasuredEsti-

mated

45hpengine (Ref.14)

Rover 1S/60 Rotor (Ref.15)

0 - 74umo - 430 - 150 - 100 - 5

14um, 180 M/s275370

7.0 - 8.86.0 - 8.01.2 - 2.01.1 - 1.50.7 - 0.6

1.12 0.82 0.48

6.84.01.41.00

1.01 0.63 0.45

Table I

In most testson aero engines, weight losseswere not measured and it is necessary to use a cor-relation factor to relate estimated erosion tomeasured power loss. Here it must be noted thatpower losses can be caused by depositionas well aserosion and it is necessaryto considermeasure-ments after cleaningand removal of the deposits.In Table II, such comparisonsare made using lossesfor 0 to 1000um dust as a datum for T56 and T58tests, and 0 to 200um for the T63 tests. Havingobtained the correlationfactor, losses can be est-imated for the finer dusts giving reasonably closecorrelationwith the measured values.

Z.ngineTest dust

1-01

Percent powerloss/kg ingested

Measured i.stimated

T56

1

8

000

000

0 n

0

IIII

I

I

I

I

I

1 0.70 -(Ref.16) o.84 0.75

0.15 0.11

T58 2.95 -

(Ref.17) 1.82 1.41

0.53 3.46

T63 4.95 -

(Ref.18) 5.15 2.2

0.78 0.55

Table II

The losses, expressedas percentagepowerloos produced by unit mass of sand, can also betreated as a susceptibilityfactor for the enginein question. Houever, some caution is required inusing such figuresbecause they are dependent uponthe type of test dust and do not include effectsdue to deposition which are most important inservice.

III. Methods of minimising erosion

It is usually difficult to minimise erosionby design changes without penalisingother aspectsof the engine performancee.g. substantial improve-ments could be achieved by the use of thicker blad-ing but only modest increasescan actually be

- 4 -

tolerated without undue aerodynamic losses(-,251min blading of the Rolls-RoyceGnome engine).In addition, engines are usually designed toachieve a target performanceand apart from fea-tures such as the siting of the intake, erosion isonly given considerationafter unfavourableopera-tional experience. At this stage, it is difficultto make design changes and attention is given touse of improved materials and filtrationof theintake.

Materials and coatings

Considerableefforts have been made to dis-cover materials with good erosion resistancebutthese have met with very little success. In fact,the nickel superalloysand martensitic steelscommonly used in the later stages of axial com-pressors have been found to be the best availableengineeringmaterials (both exhibit erosion of--0.6cc/kg for 125 to 1501m quartz at 800 ft/s).The only materials that are substantiallybetterare hard ceramics such as hot pressed siliconnitride (0.2 cc/kg), boron carbide (0.2 cc/kg)and KT silicon carbide (0.3 cc/kg). However, itis expensive to produce blades in these materialsand they are very fragile when impactedby largerparticles. In tests with the NOTE whirling armrig, a hot pressed silicon nitride specimen dis-integratedwhen impacted by several 500 to 850pmquartz particles at 240 m/s. In some cases it ispossible to improve the situation if densermaterials can be tolerated e.g. BEA substitutedtitanium for aluminium alloy in early stage blad-ing of Spey engines to obtain a 2/1 improvementinerosion resistance for a total weight increase of18.3 lb in the first three stages of blading.

No outstandinglysuccessful coatings havebeen discovered to date and little or no improve-ment can be obtained over the nickel based alloys.One of the best available tungsten carbide coat-ings was found to improve the erosion of H46 steelunder glancing impacts but was worse for 900impacts (Figure5).

As a result of this type of experience it appearsthat an increase in thicknessof the base material,by an amount equal to that of the coating, would bealmost as effective and considerablycheaper.However, protection of more susceptiblematerialsis a viable propositionand successful systemshave been developed. In Figure 6, the behaviour ofa nickel coating on fibreglassis shown in compar-ison with the parent material and it can be seenthat an improvementof 19/1 has been achieved.This erosion resistance is similar to that of thealuminium alloy RR58, but inferior to that of 046steel.

1

FIBREGLASS

.1

NICKELONFIBREGLASS'

.01

Pi46 STEEL

60 90IMPACTANGLE—&gums

FIG 5 COMPARA11VEBEHAVIOUROF A HARD COATINGAND A PARENTMATERIAL

(60 -126 m QUARTZAT 100m/sec)

•00110 100

WEIGHTIMPACTED—g

FIG. 6 COMPARATIVEBEHAVIOUROFA NICKELCOATINGAND PARENTMATERIAL

(125-150p m QUARTZAT 130 rn/ sec )

uJ

Filtration

Weight penalties and the size of installa-tions are such that filtrationsystems cannot beseriously considered for civil transport.However,the small engines used in helicoptersand hover-craft can be filtered and a variety of methodshave been explored. Of these, systems based onbanks of small inertialseparatorshave beenfavoured because they have the merit that they areself cleaning, can be relatively light weight andinvolve modest pressure drops. Good performancescan be obtained with very high efficienciesas

FLAW SPRAYEDTtSGSTENCARBIDE

5

shown in Figure 7. However, the actual improve-ments in life produced by these systems have been

In an effort to resolve this disparity,atrial was conductedat the Long Valley test ground(Hampshire, .eingland)usirv a Scout helicopter-withand without filtration (Figure8).

v 10

g 20

50

7010 100

PARTICLE SIZE — JIm

FIG. 7 PERFORMANCECURVE FOR THEINERTIAL SEPARATORTUBES USED IN

THE HEUCOPTERTRIALAT LONGVALLEY

substantiallyless than would be expected fromconsiderationof the degree of filtration.Approximate estimates of the anticipated improve-ments can be made by considerationof the overallfiltrationefficiency for a representativetype ofdust distribution. In North Africar.:ork,dustclouds were found to have size distributionsveeysimilar to the BS 1701 coarse: 1970 distributionforwhich an individual inertialeeparator tubehad a bench efficiency of 94.,.5pee cent. If par-ticle size effects are ignored thi:,shoutu give alife improvementof 29/1 whereas experiencewithdifferent types of conditionnas shown thatfigurerof around 5/1 are actually achieved(18),(19),(20).More detailed considerationof the influenceofparticle size worsens the disparity becaueefilterstend to pass only the smaller particlesand theseare intrinsicallyless erosive as well as having alower quartz content. This more precise estimatecan be made from considerationof the unfiltereddamage which is given by

.T1(d)

0 jE d Qd

and the filtered damage which isfl(d)

Ed '*'clQd

The resulting improvementhas been calculatedtobe 60/1.

FIG. 8 DUST CLOUD PRODUCED BY

A SCOUT HELICOPTER AT 6M HEIGHT

Sampling probes -,Jerefitted upstream and downstreamof the filters so that dust concentrations,sizedistributionsand filtrationefficienciescould bemeasured. Dense dust clouds were created with con-centrations of up to mg/cu ft for hovering heights of 5 to 20 ft (Figure9).

ESTIMATED CONCENTRATIONTO CAUSE ENGINE FAILUREN 2 HOURS EXPOSURE

DOWNYREAM

2 4 6HEIGHT ABOVE GROUND--m

FIG 9 DUST CONCENTRATIONSMEASURED IN

SCOUT HELICOPTER TRIAL

The eiv,• tributionsof the clouds were relat-ively CONY:It', so tnaL the filtersshould have per-formed above expectationsbased on the finer dis-tributions. However, the resultsshowed that theperformance eas below expectationsbecause (0 theupstream and downstreamsamples exhibited similarsize distributionsehen it would be expected thatlarger proportionsof the big particles should beremoved, and (ii) the overall efficiencywas

t 3

z 2

0 ° 0

-6-

substantiallylower than the bench value for afiner type of dust (Figure 10).

These low efficienciesare consistentwithservice experience as shown in Figure 10, and withother types of helicopter filtrationsystem(18),(19),(20),so that it may be concludedthat filtration systems can fail to achieve designperformance in flight. This behaviourmay be dueto features such as the effect of the rotor downwash or to inadequate scavenge flow in some of theindividualtubes. Clearly, further attention isrequired in this area.

IV. Conclusions

Laboratory studies of sand erosion have shownthat the extent of the damage is dependenton theimpact conditions and the properties of both thetarget and the abrasive materials. Equationsdescribing the most important parametershave beensummarisedand it has been shown that the relativebehaviour of engines under different conditionscan be estimated from laboratory data.

The nature and extent of the erosion damagecan vary according to the size of the engine andnature of operation i.e. whether civil or military.In civil aircraft, erosion is produced during arelativelylarge number of flight cycles and prob-lems arise through replacementof worn partsrather than hazardous operation due to mechanicalunreliability. The most viable method of minimis-ing the problem appears to be through substitutionof materials or coatings having better erosionresistance. Unfortunately,there are no usablematerials better than the martensitic steels andnickel alloys currently used in the later stagesof axial compressors. In some cases it is possibleto replace light alloyswith more resistantbutheavier materials.

Vehicles such as helicopters and hovercraftare specificallydesigned to operate off unpre-pared airstrips and can create dense dust cloudsand incur severe erosion. Under service conditionsthe lives can be reduced drasticallyand it isessential to improve the situation. Unlike civilaircraft, the smaller engines can be fitted withfilters and it is possible to design a systemgiving a big improvement in life. Unfortunately,it appears that operation of such installationsinflight involves unexpectedly low efficiencies.

LABORATORYMEASUREMENT- USING BS 1701 DUST

DEDUCED FROMSERVICE EXPEMENCE

TRIAL DATA

2 4 6HEIGHTABCVEGROUND—im

FIG.10 TOTALFILTRATIONEFFICIENCIESMEASURED IN THE HELICOPTERTRIALAT LONGVALLEY

Acknowledgement

fhis Paper is Crown copyrightand is.produced by permissionof the Controller ofHer Majesty's StationeryOffice.

- 7 -

References

No. Author(s) Titleetc.

14 J. E. MontgomeryJ. M. Clark

Dusterosionparametersfora gas turbine.SAESummerMeeting,AtlanticCity,NJ,June 1962

15 G. A. Duke Erosiontestson amodifiedRovergasturbine.

ARL NoteNo.ARL/ME297

(1968)

16

UnpublishedAllison

Report(1969)

17 R. M. Skinner T58Sandand dusttest.

UnpublishedGE Report,1962

18 G. V. Bianchini T63 Enginesandand dust

R. B. Koschman tolerancedevelopment and fieldexperience.Proc.6th Conf.onenvironmentaleffectson aircraftpropulsionsystems.Paper66-ETIV-20(1966)

No. Author(s)

Title etc.

1 W. A. Hibbert Helicoptertrialsoversandand sea.J.Roy.Aeron.Soc.69(659)(1965)769

2 H. H. Shohet CH-54AEAPSOperationalC. D. Stephenson experience,Proc.9thC. E. Watt Conf.on environmental

effectson aircraftpropulsionsystems.PaperNo.69-ENV-8(1969)

3 W. A. Compton Dusterosionof compres-K. P. Steward sormaterials-

experienceand prospects.ASMEPre-print68-GT-55(1968)

4 K. Tamkin UnpublishedcommunicationfromBEA (1971)

5 J. E. Goodwin Thesignificanceof par-WendySage ticlesizein sandero-G. P. Tilly sionof smallgasturbine

engines.Aeron.J.May (1969)429

6 G. L. Sheldon Similaritiesand differ-encesin theerosion 19 G. C. Rappbehaviourof materials.Trans.ASME92 D(3)(1970)619-626

7 I. Finnie Erosionof metalsbyJ. Wolak solidparticles.Y. Kabil J.Mater.(3)(1967)682

8 J. G. A. Bitter A studyof erosionphenomenaPartI.Wear6 (1963)5-21

9 J. H. Neilson Erosionby a streamofA. Gilchrist solidparticles.

Wear11 (1968)111

10 J. G. A. Bitter A studyof erosionphenomenaPartII.Wear6 (1963)169-190

11 I. Finnie Erosionof surfacesbysolidparticles.Wear3 (1960)87-103

12 G. P. Tilly Erosioncausedby air-borneparticles.'3ear14 (1969)63-79

13 G. P. Tilly The interactionof par-WendySage ticleand material

behaviourin erosionprocesses.Wear 16 (1970)447-465

Problemsand solutionsforsandenvironmentoperationof helicoptergas turbines.ASMEPre-print68-GT-37(1968)

Conditionof Nimbusenginesand flightresultsaftercompara-tivearmytrialsinAden.UnpublishedRolls-RoyceReport(1967)

S. H. Rosenthal

20 M. J. Rogers

8