CFRP钻孔的研究

www.elsevier.com/locate/compstruct

Studyofdelaminationindrillingcarbonfiberreinforced

plastics(CFRP)usingdesignexperiments

J.P.Davim*,PedroReis

DepartmentofMechanicalEngineering,UniversityofAveiroCampusSantiago,3810-193Aveiro,Portugal

Abstract

Inthispaperispresentedanewcomprehensiveapproachtoselectcuttingparametersfordamage-freedrillingincarbonfiberreinforcedepoxycompositematerial.TheapproachisbasedonacombinationofTaguchiÕstechniquesandontheanalysisofvariance(ANOVA).Aplanofexperiments,basedonthetechniquesofTaguchi,wasperformeddrillingwithcuttingparametersprefixedinanautoclavecarbonfiberreinforcedplastic(CFRP)laminate.TheANOVAisemployedtoinvestigatethecuttingcharacteristicsofCFRPÕsusinghighspeedsteel(HSS)andCementedCarbide(K10)drills.Theobjectivewastoestablishacor-relationbetweencuttingvelocityandfeedratewiththedelaminationinaCFRPlaminate.Thecorrelationwasobtainedbymultiplelinearregression.Finally,confirmationtestswereperformedtomakeacomparisonbetweentheresultsforeseenfromthementionedcorrelation.

Ó2002ElsevierScienceLtd.Allrightsreserved.

Keywords:Drilling;Carbonfiberreinforcedplastics;Delaminationfactor;TaguchiÕstechniques;Analysisofvariance

1.Introduction

Owingtothegrowinguseofcompositematerials,specificallythefiberreinforcedplastics(FRP)outsidethedefenceindustryandtheaerospaceindustry,theunitcostreplacestheperformanceatanycostasthemainconcernforproduction.So,theproductiontechno-logies,especiallythemachiningofcomposites,areas-sumingamoreandmoresignificantroleastheyconditiontheeconomicviabilityoftheproduct.

Machiningcompositematerialsisarathercomplextaskowingtoitsheterogeneity,heatsensitivity,andtothefactthatreinforcementsareextremelyabrasive.Con-ventionalmachiningmethodsshouldbeadaptedinsuchawaythattheydiminishthermalandmechanicaldamage.

Drillingisafrequentlypracticedmachiningprocessinindustryowingtotheneedforcomponentassemblyinmechanicalpiecesandstructures.

Thedrillingoflaminatecompositematerialsissig-nificantlyaffectedbythetendencyofthesematerialstodelaminateandthefiberstobondfromthematrixundertheactionofmachiningforces(thrustforceandtorque).

*Correspondingauthor.Fax:+351-234-370953.E-mailaddress:pdavim@mec.ua.pt(J.P.Davim).

Manyauthors[1–3],whenreportingaboutthedrill-ingoflaminatedcompositematerialsbyconventionaltools,haveshownthatthequalityofthecutsurfacesisstronglydependentonthedrillingparameters,toolge-ometryandtoolmaterial.Aninappropriatechoiceoftheseparameterscanleadtounacceptablematerialde-gradation,suchasfiberpull-out,matrixcratering,thermaldamageanddelamination.Amongthedefectscausedbydrilling,delaminationappearstobethemostcritical.

LaminatedfiberreinforcedplyÕsundermachiningforcesaresubjecttotheriskofinterlaminarcrackpropagation,calleddelamination.

Koenigetal.,studiedin1985themachiningoffiberreinforcedplasticsandconcludedthatahighfeedrateofdrillingwillcauseacrackaroundtheexitedgeofthehole[4].

Millerpresentedin1987adatabaseonoptimumcuttingparametersfordrillingholeswithminimumlocalmachiningdamage[5].

HochengandPuw[1],in1992presentedastudyofthechipformationandassessesthemachinabilityoftwocompositematerials(Thermoset-basedandThermo-plastics-based)andconcludedthatfromcuttingchipstheformerpresentsalargeamountofdeformationinchipformation,whilethelattertendstofracture.He

0263-8223/03/$-seefrontmatterÓ2002ElsevierScienceLtd.Allrightsreserved.PII:S0263-8223(02)00257-X

482J.P.Davim,P.Reis/CompositeStructures59(2003)481–487

alsonoticedthatcarbon/ABSissuperiortocarbon/epoxyforsurfacequalityandbothholeedges.

ChambersandBishop[6]in1995investigatedtheeffectofthecuttingparametersondrillingcarbon/epoxyandcarbon/peekandconcludedthatthedrillingofcarboncompositesisdependentuponthecharacteristicsofthematrixandthehelicalPCDdrillgeometrygavethebestoverallperformance.

Inadditiontotheproblemsoftoolwear,itisverydifficulttoachievethequalityofsurfaceneededfortheaccurateassemblyofcomponentsinmechanicalstruc-tures.

Linetal.[7]in1996,carriedoutastudyondrillingofcarbonfiberreinforcedcompositeathighspeedandconcludedthatanincreaseofthecuttingvelocityleadsaincreasingofthedrillwear.Inthiswaythefactofincreasingthewearofdrillcausesarisingofthrustforce.

Wen-ChouChen[3]in1997studiedthevariationsofcuttingforceswithorwithoutonsetdelaminationduringthedrillingoperationsandconcludedthatthedelami-nation-freedrillingprocessesmaybeobtainedbytheproperselectionsoftoolgeometryanddrillingpara-meters.

Piquetetal.[8]in2000carriedoutastudyofdrillingthincarbon/epoxylaminateswithtwotypesofdrills,ahelicaldrillandadrillofspecialgeometry,andcon-cludedthatbothdrillsleadsadamageattheentranceinthewallandtheexitofthehole,withtheexceptionofspecialgeometrydrillwhichispossibleasignificantre-ductioninthefinaldamage.

Enemuohetal.[9]in2001,realizethatwiththeap-plicationofthetechniqueofTaguchiandothermeth-ods,werepossibletoachievethecuttingparametersthatallowedtheabsenceofdamageinthedrillingoffiberreinforcedplastics.

Alltheaboveworkscontributestothepracticeofcuttingcompositematerials,buttheydonotshowthebasicmechanismsofcutandthecharacterizationofthemachinability.

2.Experimentalprocedure2.1.Meansandmaterials

Inordertoachievetheobjectiveofthisexperimentalwork,mainlytheestablishingofcorrelationsbetweenthecuttingconditionswiththedelamination,machiningissueswereeffectuatedwithdifferentcuttingconditions,andwereusedcarbonfiberreinforcedplastics(CFRPÕs)(Epoxymatrixreinforcedwith55%ofcarbonfiber)fortests.

Thiscompositematerialwasproducedbyautoclavewithafiberorientationof0/90degrees,aswecanob-serveonFig.1.

Theexperimentshadbeencarriedoutinanautoclavelaminatemadeupof16layersoffiberswith4mmofthickness,usingthreedifferenttypesof5mmdiameterdrillspresentedintheFig.2.Ahelicalflutehighspeed

Fig.1.Plateproducedbyautoclavewithafiberorientationof0/90degrees.

Fig.2.(a)HelicalfluteHSSdrill;(b)four-fluteK10drill;(c)helicalfluteK10drill.

J.P.Davim,P.Reis/CompositeStructures59(2003)481–487483

steel(HSS)drill(R415.5-0500-50-8C0),afour-flutece-mentedcarbide(K10)drill(R950.01H10F)andfinallyahelicalfluteK10carbidedrill(R415.5-0500-30-AC0),accordingtoISO1832.

Adrillingmachinewith2,2kWspindlepowerandamaximumspindlespeedof2500rpmwasusedtoper-formtheexperiments.

Theuseddrillshavean118°pointangle.

Thesqueezeofthelaminate,inthepressofjawofthedrillingmachinewasmadebyasystemofclamps,tomakesurethatvibrationsanddisplacementdoesnotexist,aswecanobserveinFig.3.

Thedamagearoundtheholeswasmeasuredwithashopmicroscope,MitutoyoTM500,with30Âmagni-ficationand1lmresolution.

2.2.Planofexperiments(Taguchi’stechniques)TaguchiÕstechniqueshavebeenusedwidelyinengi-neeringanalysis.Thesetechniquesconsistofaplanofexperimentswiththeobjectiveofacquiringdatainacontrolledway,executingtheseexperiments,inordertoobtaininformationaboutthebehaviorofagivenpro-cess.

Thetreatmentoftheexperimentalresultsisbasedontheanalysisaverageandtheanalysisofvariance(AN-OVA)[10–13].

FortheelaborationofexperimentsplanweusedthemethodofTaguchifortwofactorsatthreelevels.Bylevelswemeanthevaluestakenbythefactors.Table1indicatesthefactorstobestudiedandtheassignmentofthecorrespondinglevels.

ThearraychosenwastheL9(24),whichhasninerowscorrespondingtothenumberoftests(eightdegreesoffreedom)withtwocolumnsatthreelevels,asshowninTable2.Thefactorsandtheinteractionsareassignedtothecolumns.

Theplanofexperimentsismadeofninetests(arrayrows)inwhichthefirstcolumnwasassignedtothecuttingvelocityðVÞandthesecondcolumntothefeed

Fig.3.Squeezeoftheplateinthepressofjawofthedrillingmachine.

Table1

AssignmentofthelevelstothefactorsLevelRevolutionnFeedratefCuttingvelocity(rpm)(mm/rev)V(m/min)110000.0416215000.08243

2000

0.15

32

Table2

OrthogonalarrayL9(24)ofTaguchi[10]L9(24)Test123411111212223133342123522316231273132832139

3

3

2

1

LineargraphL9ð24Þ[10].

rateðfÞandtheremainingwereassignedtotheinter-actions.TheresponsetobestudiedisthedelaminationfactorðFdÞinCFRPlaminate.

3.Resultsanddiscussion

3.1.Influenceofthecuttingparametersinthedelami-nationfactor

ThedamagearoundtheholeswasmeasuredusingashopmicroscopeMitutoyoTM-500,followingtheschemapresentedintheFig.4.

AftermeasuringthemaximumdiameterðDmaxÞinthedamagezone,i.e.aroundeachhole,wecarriedouttodeterminedthevalueofthedelaminationfactorðFdÞ.ThisfactorisdeterminedbytheratiothemaximumdiameterðDmaxÞofthedamagezonetotheholediameterðDÞ.

Fig.4.SchemaofthemeasurementofthemaximumdiameterðDmaxÞwithashopmicroscopeMitutoyoTM500.

484J.P.Davim,P.Reis/CompositeStructures59(2003)481–487

ThevalueofdelaminationfactorðFdÞcanbeobtainedbythefollowingequation:FDmaxd¼

Dð1Þ

being,DmaxthemaximumdiameterofthedamageholeinlmandDthediameteroftheholeinlm.

Table3showstheresultsofthedelaminationfactorðFdÞ,forthethreesetsofdrillingtests,obtainedbytheEq.(1)infunctionofthecuttingparameters.

IntheFigs.5and6wecanobservetheevolutionofthedelaminationfactorðFdÞwiththefeedforthedif-ferentcuttingspeedvalues.

InFig.5wecanevidencethattheFdincreaseswiththefeedrate,andwiththecuttingspeed.Accordingtothegraph,wecanobservethatthecarbidedrillpresentsabetterperformancethantheHSSdrill,i.e.underthesamecuttingconditions(cuttingspeedandfeedrate),theHSSdrillcausesalwaysabiggerdelaminationfac-tor,witchmeanshigherdamageinthecompositelami-nate.

WealsocanobservethatinFig.6,theFdincreaseswiththefeedrate,andwiththecuttingspeed.

Table3

ValuesofdelaminationfactorðFdÞinfunctionofthecuttingparame-tersTest

V(m/f(mm/DelaminationfactorðFdÞmin)rev)HelicalfluteFour-fluteHelicalfluteHSSdrillK10drillK10drill116

0.041.0441.01.04220.081.0521.0661.04530.151.0511.0661.047424

0.041.0611.0691.05750.081.0701.0711.060.151.0791.0781.0732

0.041.0631.0731.05280.081.0691.0751.0629

0.15

1.078

1.080

1.069

)1.09dFV = 16 m/min(r 1.08oV = 24 m/mintcaf1.07V = 32 m/min noit1.06V = 16 m/minnaiV = 24 m/minma1.05V = 32 m/minelD1.04HSS drillK10 drill1.030.020.060.10.140.18feed (mm/rev)Fig.5.DelaminationfactorðFdÞinfunctionofthecuttingparameterstodrills,fortwotoolmaterials,whichhavethesamegeometry.1.10)dV = 16 m/minF1.09(V = 24 m/minrot1.08V = 32 m/mincaf1.07V = 16 m/minno1.06V = 24 m/minitanV = 32 m/minim1.05a4 flutes (K10)le1.04Dhelical flute1.03(K10)0.020.060.10.140.18feed (mm/rev)Fig.6.DelaminationfactorðFdÞinfunctionofthecuttingparameterstodrills,fortwotoolmaterials,whicharemanufacturedwiththesamematerial.Wecanevidencethatthehelicalflute(K10)drillpresentsabetterperformancethanthefour-flute(K10)drill,i.e.underthesamecuttingconditions(cuttingspeedandfeedrate).Thefour-flutedrillcausesalwaysabiggerdelaminationfactor,exceptinthetestwhichhavethehigherfeedrateandthelowercuttingvelocity.

Geometricaldifferencesbetweenthetwotypesofdrillmaybethereasonforthisfact.

Withanopticalmicroscope(OM)waspossibletoobservethedamagearoundtheholes,andnoticetheinfluenceofthecuttingvelocityðVÞhaveonthede-laminationfactorðFdÞ,aswecanseeonFig.7.

InFig.7,wecanevidencethat,foraconstantafeedrate,anincreaseofthecuttingvelocityincreasesthedelaminationfactorðFdÞaswell.

Ananalysisofvarianceofthedatawiththedelami-nationfactorðFdÞinCFRPlaminate,withtheobjectiveofanalyzingtheinfluenceofthecuttingvelocityðVÞ,offeedrateðfÞonthetotalvarianceoftheresults.

Thestatisticaltreatmentofthedatawasmadeintwophases.Thefirstphasewasconcernedwiththeanalysisofvarianceandtheeffectofthefactorsandofthein-teractions.Thesecondphaseallowedustoobtainthecorrelationbetweentheparameters(Vandf).

Tables4–6showtheresultsoftheanalysisofvariancewiththedelaminationfactorðFdÞinCFRPlaminate.FromtheanalysisofTable4,wecanobservethatthecuttingvelocityðP¼72:5%ÞandthefeedratefactorðP¼20:5%Þ,havestatisticalandphysicalsignificanceonthedelaminationfactorðFdÞobtained,especiallythecuttingvelocityfactor.

Thefactors(Vandf)presentastatisticalsignificancetestF>Fa¼5%.NoticethattheerrorassociatedtothetableANOVAfortheFdwasapproximately7%.FromtheanalysisofTable5,wecanobservethatthecuttingvelocityfactorðP¼65:9%Þ,havestatisticalandphysicalsignificanceontheFdobtained.

ThefactorfeedrateðP¼19:3%ÞdoesnotpresentpercentageofstatisticalsignificanceofcontributionontheFd.

J.P.Davim,P.Reis/CompositeStructures59(2003)481–487485

Fig.7.DelaminationfactorðFdÞinfunctionofthecuttingspeed,MO:amplificationof30Â.

Table4

ANOVAforthedelaminationfactorðFdÞtothehelicalfluteHSSdrillSourceofvarianceV(m/min)f(mm/rev)ErrorTotal

SDQ8.7E)42.6E)44E)51.17E)3

gl2248

Variance4.3E)41.3E)41E)5/

TestF42.2612.69//

Fa¼5%6.946.94//

P72.520.57.0100.0

SDQ––sumofsquares,gl––degreesoffreedom,P––percentageofcontribution.

Table5

ANOVAforthedelaminationfactorðFdÞtothefour-fluteK10drillSourceofvarianceV(m/min)f(mm/rev)ErrorTotal

SDQ1.7E)46E)52E)52.5E)4

gl2248

Variance9E)53E)55E)6/

TestF18.816.20//

Fa¼5%6.946.94//

P65.919.314.8100.0

SDQ––sumofsquares,gl––degreesoffreedom,P––percentageofcontribution.Table6

ANOVAforthedelaminationfactorðFdÞtothehelicalfluteK10drillSourceofvarianceV(m/min)f(mm/rev)ErrorTotal

SDQ4.6E)41.5E)41E)46.7E)4

gl2248

Variance2.3E)47E)52E)5/

TestF14.114.//

Fa¼5%6.946.94//

P63.517.119.4100.0

SDQ––sumofsquares,gl––degreesoffreedom,P––percentageofcontribution.

ThefactorcuttingvelocitypresentastatisticalsignificancetestF>Fa¼5%,witchdoesnothap-penstothefactorfeedratebecausethetestFNoticethattheerrorassociatedtothetableANOVAfortheFdwasapproximately14.8%.

FromtheanalysisofTable6,wecanobservethatthecuttingvelocityfactorðP¼63:5%Þ,havestatisticalandphysicalsignificanceontheFdobtained.ThefactorfeedrateðP¼17:1%ÞdoesnotpresentpercentageofphysicalsignificanceofcontributionontheFd,becausePðpercentageofcontributionÞFromthereasonsabovepresented,thefactorcuttingvelocitypresentastatisticalsignificanceandthefactorfeedratedoesnot.

NoticethattheerrorassociatedtothetableANOVAfortheFdwasapproximately19.4%.

486J.P.Davim,P.Reis/CompositeStructures59(2003)481–487

3.2.Correlation(delaminationfactor/cuttingparameters)Thecorrelationsbetweenthefactors(cuttingvelocity,feedrate)andthedelaminationfactorðFdÞinCFRPlaminatewereobtainedbymultiplelinearregression.Theequationsobtainedwereasfollow:HelicalfluteHSSdrillÀFd¼1:021þ1:31Â10À3

Vþ0:117fR¼0:76ð2Þ

Four-fluteK10drillÀFd¼1:037þ1:0Â10

À3Vþ0:158fR¼0:74ð3Þ

HelicalfluteK10drillÀ

Fd¼1:010À1:16Â10À4Vþ0:097f

R¼0:86

ð4Þ

being,Vthecuttingofvelocityinm/min,andfthefeedrateinmm/rev.3.3.Confirmationtests

Table7showsthecuttingconditionsusedontheconfirmationtests.

Theresultsobtainedbyacomparisonbetweentheforeseenvaluesbythebythemodelsdevelopedinthepresentwork(Eqs.(2)–(4))andtheexperimentallyob-tainedresultsbydelaminationfactorðFdÞ,areshowninTable8.Fromtheanalysisoftherefereedtable,wecanobservethatthegainedthroughtheactualmodelshowamaximumerrorofabout2%.WealsocanconsiderthatEqs.(2)–(4),correlatestheevolutionofthedelamination

Table7

CuttingconditionsusedindrillingconfirmationtestsTypeofdrillTestRevolutionnV(m/min)f(mm/rev)(rpm)HelicalfluteHSS1c1250200.082c1750270.08Four-fluteK103c1250200.084c1750270.08HelicalfluteK10

5c1250200.086c

1750

27

0.08

Table8

ExperimentalplanconfirmationdrillingtestsandtheircomparisonwiththeresultsTestDelaminationfactorðFdÞExperimentalvaluesModelEqs.(2)–(4)Error(%)11.0471.0560.8721.0711.0660.4031.0841.0691.3241.0921.0771.3651.0231.0101.316

1.041

1.033

0.73

factorðFdÞinthelaminatewiththecuttingconditions(cuttingvelocityandfeed)withagooddegreeofap-proximation.

4.Conclusions

Fromthisstudyondelaminationwhen,drillingofCFRPs,thefollowingconclusionscanbedrawn:•ThehelicalfluteK10drillpromoteslessdamageonthecompositelaminatethanthefour-flutecarbide(K10)drill,i.e.,thedelaminationfactorðFdÞissmal-ler.

•ThehelicalfluteK10drill,presentsabetterperfor-mance,thanhelicalfluteHSSdrill,i.e.,thecarbidedrillisthebetterchoicefordrillingCFRP.

•Thedelaminationfactorincreaseswithbothcuttingparameters,whichmeansthatthecompositedamageisbiggerforhighercuttingspeedandforhigherfeed.•ThecuttingvelocityisthecuttingparameterthathasthehighestphysicalaswellstatisticalinfluencethedelaminationfactorinCFRPlaminate(63.91%and65.9%),forthethreedrills.

•Bothcarbidedrillsshowsanalmostnullwearlandintheflanksurface,whiletheHSSdrillpresentsawearvalueof0.012mm,measuredat1/4thedrillradius,apartfromthecorner.

•Theconfirmationtestsshowedthattheerrorassoci-atedtothedelaminationfactorðFdÞ(maximumvalue2%andminimum0.4%)isexcellent.

Acknowledgements

TheauthorsacknowledgeProfessorAnt󰀂o

nioTorresMarques,fromINEGI/FEUP,forprovidingthecom-positematerialusedinthedrillingtests.Theyacknowl-edgealsototheMechanicalEngineerPedroMadalenofortheirparticipationintheexperimentalwork.

References

[1]HochengH,PuwH.Ondrillingcharacteristicsoffibre-reinforced

thermosetandthermoplastics.IntJMachToolsManufact1992;32(4):583–92.

[2]HochengH,PuwH,YaoK.Experimentalaspectsofdrillingof

somefibrereinforcedplastics.In:Proceedingsofthemachiningofcompositematerialssymposium.ChicagoIllinois:ASMMaterialsweek;1992.p.127–38.

[3]ChenW.Someexperimentalinvestigationsinthedrillingof

carbonfibre-reinforcedplastic(CFRP)compositelaminates.IntJMachToolsManufact1997;37(8):1097–108.

[4]KoenigW,WulfC,GrassP,WillerscheidH.Machiningoffiber

reinforcedplastics.ManufactTechCIRPAnn1985;34(2):537–48.[5]MillerJA.Drillinggraphite/epoxyatLockheed.AmMachAutom

Manufact1987:70–1.

J.P.Davim,P.Reis/CompositeStructures59(2003)481–487

[6]ChambersA,BishopG.Thedrillingofcarbonfibrepolymer

matrixcomposites.ProcessManufact1995;III:565–72.

[7]LinSC,ChenIK.Drillingofcarbonfiber-reinforcedcomposite

materialathighspeed.IntJMacToolsManufact1996:156–62.[8]PiquetR,FerretB,LachaudF,SwiderP.Experimentalanalysis

ofdrillingdamageinthincarbon/epoxylaminateusingspecialdrills.IntJMacToolsManufact2000:1107–15.

[9]UgoEnemuohE,SherifEl-GizawyA,ChukwujekwuOkaforA.An

approachfordevelopmentofdamage-freedrillingofcarbonfiberreinforcedthermosets.IntJMacToolsManufact2001:1795–814.

487

[10]RossP.Taguchitechniquesforqualityengineering-lossfunction,

orthogonalexperiments,parameterandtolerancedesign.NewYork:McGraw-Hill;1988.p.10–50.

[11]TaguchiG,KonishiS.Taguchimethods,orthogonalarraysand

lineargraphs,toolsforqualityengineering.AmericanSupplierInstitute;1987.p.35–8.

[12]PhadkeMS.Qualityengineeringusingrobustdesign.Englewood

CliffsNJ:Prentice-Hall;19.p.1–50.

[13]TaguchiG.Taguchionrobusttechnologydevelopmentmethods.

NewYork:ASMEPress;1993.p.1–40.