Mechanistic force modeling for milling of carbon fiber reinforced polymers with double helix tools

Mechanistic

force

modeling

for

milling

of

carbon

fiber

reinforced

polymers

with

double

helix

tools

Yig˘it

Karpat

a,

*,

Naki

Polat

(3)

b

a_{DepartmentofIndustrialEngineering,BilkentUniversity,Ankara,Turkey} b

TurkishAerospaceIndustries(TAI),Ankara,Turkey

SubmittedbyBilginKaftanog˘lu(1)AtılımUniversity,Ankara,Turkey.

1. Introduction

Carbonfiberreinforcedplastics(CFRP)offerhighstrengthto weight ratio, low thermal expansion coefficient, and high resistance tocorrosion.Dueto thosedesirable properties,their usageinnewgenerationofaircraftshasbeensteadilyincreasing. CFRPpartsareproducednear netshape;however,drilling,slot/ sidemilling,andsurfacefinishmillingoperationsarestillrequired tobringCFRPpartstotheirfinalshapes.MachiningCFRPsisknown tobedifficultduetotheabrasivenatureofcarbonfibersandthe low thermal conductivity of the material [1]. Diamondcoated carbideandpolycrystallinediamondtoolsareusuallyemployedin ordertowithstandthoseharshconditions.

DelaminationisthemostimportantissueinmachiningCFRPs

[2].Machiningforcesintheaxialdirectiontendtoseparatethetop andbottomlayersoftheCFRPlaminateduringmachining.Flatend mills withlow helix angles areusually used tominimizeaxial forces.Inordertodecreaselikelihoodofdelaminationandobtain an acceptable level of productivity, milling tools with special designshave beenproposed.Doublehelix tools,also knownas right/left helix,minimize axialforces by utilizingtwo opposite helixangles.Asaresult,topandbottomlayersofthelaminateare pushedinwardstodecreasethelikelihoodofdelamination.

Thenumberof studieson milling ofCFRP laminatesis quite limited.Sheikh-Ahmad[3]andKallaet al.[4]proposedaneural networkmodelbasedmechanisticforcemodelingapproachforthe sidemillingofCFRPlaminateswithhelicalflatendmills.Hintzeetal.

[5]observedthesurfacequalityduringslotmillingofuni-directional CFRPlaminates.Theyobservedtheinfluenceofthefiberdirection angleofthetoplayeronthesurfacequality.Theyconcludedthat maintaining458fibercuttingangleduringmillingyieldsthebest

performance.Karpatetal.[6]proposedafiberdirectionanglebased mechanisticmachiningmodelforuni-directionalCFRPlaminates andshowedthatitcanbeextendedtocalculatemillingforcesin multi-directionalCFRPlaminates.Schulzeet al.[7]experimentally calculatedspecificcuttingforcesonglassfiberreinforcedplasticsfor various levels ofprocess inputssuchas cuttingvelocity, cutting depth,etc.whichcanbeusedintoolgeometryoptimizationstudies andmachiningstrategyselection.

Despitetheircommonuseinindustry,nostudyondoublehelix toolshasbeenpublishedtothebestknowledgeoftheauthors.In ordertoimprovedoublehelix(orsimilarvariations)tooldesigns and/or to select machining conditions depending on laminate thickness,anexperimentalinvestigationisrequired.Inthisstudy, a mechanistic milling force model for double helix tools is proposed and validated through experimental study. Cutting andforcecoefficientsarecalculatedbasedonfiberdirectionofthe laminate.Issuesrelatedtosurfacequalityandtoolwearhavealso beeninvestigated.

2. Experimentalsetup

Afiveaxismachiningcenter(Do¨rriesScharmannTechnologies, 24kWand24,000rpmmaximumrotationalspeed)wasusedin this studyduringmillingofCFRPs. TwodifferenttypesofCFRP laminateswereproducedandusedinmillingexperiments.First, uni-directionalintermediatemoduluslaminatesconsistingof30 layers with fiberdirections of 08 and 458 wereproduced. This laminateallowstheinvestigationoftheinfluenceoffiberdirection 08,458,908,and1358onmillingforces.Second,multi-directional CFRPlaminatesconsistingof72layerswithequalfiberdirections repeatingwithsequenceof08/458/908/1358werealsoproduced. Mechanical (tensile direction) properties of the intermediate modulus carbonfiberreinforceepoxyresinuni-directionaltape arereportedinTable1.

CIRPAnnals-ManufacturingTechnology62(2013)95–98

ARTICLE INFO Keywords:

Machining Milling

Fiberreinforcedplastics

ABSTRACT

Carbonfiberreinforcedpolymers(CFRP)haveemergedasthematerialofchoicetosatisfyincreasing demandforlighteraircrafts.MachinabilitycharacteristicsofCFRPsarequitedifferentthanthoseof metals;therefore,specialtooldesignshavebeendevelopedforCFRPmachining.Thedoublehelixendmill designcompressestheupperandlowersidesofthelaminateusingoppositehelixanglesthateliminate delamination.Amechanisticforcemodelfordoublehelixtoolsisdevelopedbasedonmillingforcedata obtainedonflatendmills.Theproposedmodelcanbeusedtoimprovedoublehelixtooldesignsandto optimizemillingprocessparameters.

ß2013CIRP.

*Correspondingauthor.

ContentslistsavailableatSciVerseScienceDirect

CIRP

Annals

-

Manufacturing

Technology

j o u r n a lh o m e p a g e :h t t p : / / e e s . e l s e v i e r . c o m / c i r p / d e f a u l t . a s p

0007-8506/$–seefrontmatterß2013CIRP.

Two different types of milling tools, both diamond coated carbide tools from the same tool manufacturer, were used in experiments. The first was a helical end mill with a 12mm diameter,fourteethanda108helixangle.Thesecondwasadouble helixendmillwitha10mmdiameter,hasfourteeth,anda208 helixangleasshowninFig. 1.

A Kistler 9123 rotating dynamometer and its Kistler 5223 chargeamplifierwereusedtomeasuremillingforcesandtorque.A samplingrateof25kHzwasusedinmeasurements.

3. Millingforcemodelingforuni-directionalCFRPlaminates

Duringmillingoperationofuni-directionallaminates,thetooltip interactswithdifferentfibercuttingangles(

b

)asthetoolrotates. Thefibercuttinganglechangesdependingonthefiberdirectionof thelaminate(

u

)andthetoolrotationangle(

f

).Therelationship betweencuttingdirection,fiberdirection,toolrotationangle,and fibercuttinganglefortheslotmillingof08fiberdirectionlaminateis shown in Fig. 2. The fiber direction angle (

u

) is measured counterclockwiserelativetothetoolmovementdirection.

Mechanisticforcemodelingapproachisacommontechnique thatcanbeusedtocalculatemillingforcesbasedonmaterialand toolproperties.Thistechniquerequiressomemillingteststobe performedtocalculatecuttingandedgecoefficientsinradial(Krc)

andtangential (Ktc) directionswhich representmaterials’

resis-tancetomachining.Edgecoefficients(KteandKre)representthe

influenceofcuttingedgeradiusandtheinteractionbetweenthe flank face and the work material surface. Cutting and edge coefficientsarevalidforgiventoolrakeangle(

a

),clearanceangle (

d

),edgeradius(re),andtool materialproperties.Inthecaseof

millingofuni-directionalCFRPs,cuttingandedgecoefficientsvary as a function of fiber cutting angle (

b

). This yields complex functionsofcuttingand edgecoefficients[6] whichmaynotbe easilyinsertedinmillingforcemodelingandmachiningstability calculations. In this study, cutting and edge coefficients are calculatedbasedonlaminatedirection(

u

)andaveragecuttingand edgecoefficientsforthemulti-directionalCFRPlaminatearelater calculatedbasedonsuperpositionprinciple.Itmustbenotedthat multi-directionallaminatesareofpracticalimportance.

Fig.3representscuttingforcesactingonthetoolduringtheslot millingprocess wheretangential forces (Ft)are directed in the

oppositedirectionofthecutting,andradialforces(Fr)acttoward

thecenterof thetool. Eq.(1) represents totalcuttingforces in tangentialandradialdirectionswhereapistheaxialdepthofcut

andhisinstantaneouschipthicknesswhichcanbecalculated,fora circulartoolpath,byconsideringfeed(f)andtoolrotationangle (

f

)asalsoshowninEq.(1).

Fr Ft   ¼ap Xs1 j¼0 gj ðfsinð

f

jÞÞ Krcð

u

Þ Ktcð

u

Þ   þ Kreð

u

Þ Kteð

u

Þ     (1)

f

j¼

f

0j 2

p

s ; j¼0;1;...;s1 gjð

f

jÞ¼ 1

f

s

f

j

f

e and hj>0 0 elsewhere  

Theindexjrepresentsthenumberofteethonthetool.Depending onthenumberofteethonthecutter(s)andtheradialdepthofcut (ae),entry(

f

s)andexit(

f

e)anglescanbecalculated.Foreachtooth

onthecutter,instantaneoustoolrotationangleiscalculatedand thoseteethwhichmaynotbeincontactwiththematerialduring millingareexcludedfromforceanalysis.Thetoolisslicedinthe axialdirection,whichallowsfortheinfluenceofthehelixangleof thetool.Acontrolfunction(g)isdefinedinordertocheckwhether thetoolisinthecutornot.

Averagetangentialforceduringslotmillingcanbeexpressed withEq.(2),whichisintegratedovertheentry(08)andexit(1808) angles.Asimilarexpressionforradialforcecanalsobewritten. Averagecuttingandedgeforcecoefficientsasafunctionoffiber cuttinganglecanbecalculatedbyusingmillingforce measure-mentsinreversefashionforagivenrangeoffeedvalues.Milling forces in x–y direction can be calculated through an axis transformation. ¯ Ft¼ sap 2

p

Z fe fs ðKtcð

u

Þfsinð

f

ÞþKteð

u

ÞÞd

f

(2)

3.1. Calculationoffiberdirectionanglebasedcuttingandedge coefficients

Milling tests at different feed (0.015–0.02–0.025mm/tooth) wereconductedtocalculatecuttingandedgecoefficientsforthe helicalflatendmill(Fig.1).Cutting speed(3500rpm)andaxial depth of cut (3mm) were kept constant during slot milling experiments.The rotational speedis selected sothat thetooth passing frequency is considerably lower than the natural frequency of the rotational dynamometer (wn_x=744Hz,

wn_y=834Hz).Fig.4showsthemillingforcemeasurementsfor

eachfiberdirectionwithrespecttothex–yaxisrotatingwiththe dynamometer.

Comparedtomillingforcesinxandydirections,millingforces inz-direction(Fz)areverysmall,whichcanbeattributedtothelow

helixangleoftheendmill.Averagemillingforcesinradialand tangentialdirectionswerecalculatedasafunctionfeedforeach fiberdirectionandareshowninFig.5.Alinearfitforeachdataset

Fig.1.Diamondcoatedhelical(left)anddoublehelix(right)endmills.

φ θ = 0/180 Uni-directional laminate direction Direction of cutting o Direction of cutting θ β

Fig.2.Fibercuttingangle(b)asafunctionoftoolrotationangle(f)andfiber directionofthelaminate(u).

Y X j=0 j=3 j=2 j=1 φ Direction of cutting Direction of tool rotation F_tFr Fr Ft h

Fig.3.Millingforcesactingonthetoolduringslotmilling. Table1

MaterialpropertiesofCFRPlaminates.

Fibervolume(%,v/v) Strength(MPa) Modulus(GPa) Density(g/cm3

)

59 2690 165 1.58

Y.Karpat,N.Polat/CIRPAnnals-ManufacturingTechnology62(2013)95–98 96

giveninFig.5canbeobtained.Table2showsthecalculatedslope andinterceptvaluesforeachfiberdirection.

Thesecuttingand edgecoefficients canbeusedtocalculate millingforcesinmultidirectionalCFRPlaminates.Radialforcesare observedtobesignificantlylargerthantangentialforces.Itisdue tocontactbetweenfibersandtool’sedgeradiusandflankface.This showsthe importanceofedge radiusof thetool on machining forces. It must be noted that diamond coating thickness was around10

m

monthetoolstestedinthisstudy.Althoughdiamond coating protectsthe tool fromwear, it alsoincreases theedge radius,whichinturnincreasemillingforces.

3.2. Cuttingforcemodelingformulti-directionalCFRPusingdouble helixmillingtools

Fig. 6showstheunrolledperipheryofthedoublehelixcutter, whichcanbeusedtovisualizetheteethenteringandexitinginto the cut. The box shows the boundary of milling for slotting operation (

f

=0...1808) with a total axial depth of cut given asap.Foratoolwithfourteeth,thespacingbetweenteethare458

(

V

)fromthebottomandtopofthecuttingtool.Duetohelixangle andpossiblesmalldistancep,cuttingedgesenterthecutwitha delay(

C

)andleavethecutwithanotherdelay(

D

).Entryandexit anglesforeachtoothcanbecalculatedasafunctionofaxialdepth ofcut(ap/2).Ifthebottomofthetoolisalignedwiththebottomof

thelaminate,thenpdistanceand

C

delayanglebecomeequalto zero.Thehelixangle,numberofteeth(s),heightofthebottom helix (H),and thetool diameter(D) are themajor tool design parameters.

The tool is then divided into thin slices (db) in the axial directionandeachsliceistreatedasanindividualstraighttooth end mill,andtotalcuttingforcesarecalculated bysuperposing individualforcesfromeachsliceintheaxialdirection.

A doublehelix millingtool with5mm bottomheight(H) is selectedtomachine a10mmCFRP laminate.In thiscase,thep distanceinFig.6iszero.Specialcareisgiventothisissueduring regularoperationtoensurebalanceinaxialmillingforces.Ifthe toolisextendedbeyondthebottomoftheworkpieceoralaminate thicknesswhichisnotappropriatelyselected,dependingonthe toolheight,thismayresultinunbalancedaxialmachiningforces.

3.3. Validationofthedoublehelixmillingtoolforcemodeling

The above mentioned model is programmed using Matlab softwareandsomevalidationtestsareperformed.Thecalculated averagecuttingandedgecoefficientsareusedinthemechanistic millingmodel.Fig.7showsthepredictedandmeasuredmilling forces with respect to reference system rotating with the dynamometer.Sincethetoolandlaminatemidpointsarealigned, the milling forces in z direction are not calculated. Since

(a) (b) 0 0.01 0.02 0.03 0 20 40 60 80 ) h t o o t / m m ( d e e F ) N( t F ec r o F l ai t n e g n a T n a e M g e d 0 g e d 5 3 1 g e d 0 9 g e d 5 4 0 0.01 0.02 0.03 0 50 100 150 200 ) h t o o t / m m ( d e e F Mea n Radi al F o rc e Fr (N ) g e d 0 g e d 5 3 1 g e d 0 9 g e d 5 4

Fig.5.(a)Meantangentialand(b)radialforcesasfunctionsoffeedpertooth.

Table2

Fiberdirectionbasedslopeandinterceptvaluesandaveragecuttingandedge coefficients.

Ft Fr

Slope Intercept Slope Intercept

08 2700 8.66 2640 52.8 1358 2237 26.7 1890 131.3 908 1030 25.46 2020 71 458 2846 2.49 1520 60 Average 2203 15.8 2017 78.9 CalculatedKtc=583N/mm2Kte=2.66N/mmKrc=555N/mm2Kre=12.92N/mm

Fig.6.Schematicofmillingwithdoublehelixtool.

(a) (b) (c) (d) 8.58 8.6 8.62 8.64 8.66 -200 0 200 Time (s) Cu tt ing F o rces Fx, Fy , Fz ( N ) Fx Fy Fz 7.72 7.74 7.76 7.78 7.8 -200 -100 0 100 200 ) s ( e m i T C u tt in g Fo rc e s Fx , Fy , Fz (N ) Fx Fy Fz 3.96 3.98 4 4.02 4.04 -200 -100 0 100 200 Time(s) C u tt in g Fo rc e s Fx , F y , Fz (N ) Fx Fy Fz 5.5 5.52 5.54 5.56 -200 -100 0 100 200 Time (s) C u tt in g Fo rc e s Fx , F y , Fz (N ) Fx Fy Fz

Fig.4.Millingforcemeasurementsduringslotmilling.(a)08,(b)1358,(c)908,and (d)458. (a) (b) 10 10.02 10.04 10.06 -400 -200 0 200 400 ) s ( e m i T F , s ec r o F x Fy (N) Fx Fy 0 500 1000 1500 -400 -200 0 200 400 ) g e d ( F s ec r o F x Fy (N) F y F_x 3.5 3.52 3.54 3.56 3.58 -400 -200 0 200 400 ) s ( e m i T F s ec r o F x , F y ) N( F_x F y 0 500 1000 1500 -400 -200 0 200 400 (deg) Forces F x Fy (N) F_y F x

Fig.7.Comparisonofexperimentaldata(left)andmodelpredictions(right)fortwo differentmillingcases:(a)N=4000rpm,fr=254mm/min,100%radialimmersion,

ap=10mm and (b) N=3500rpm, fr=750mm/min, 50% radial immersion

downmilling,ap=10mm.

measurementsand modelpredictionsareingood agreement,it can be concluded that average cutting and edge coefficients calculatedfromflatendmillingtestscanbesuccessfullyusedto predictdoublehelixmillingforces.

Itmustbenotedthatthenumberofteethonthemillingtoolcan be increased to maintain constant chip thickness throughout millingoperation.Fig.8(a)showsmeasurementsforthecaseof millingwithasixteethdoublehelixmillingtoolfromthesametool manufacturer.Inordertotestthemodel,thetoolisextended2mm (p=2mmasshowninFig.6)beyondthebottomofthelaminate.In thiscase,thelowerhelixcutsonly3mmandtheupperhelixcuts 7mmoftheCFRPlaminatethickness.Fig.8(b)showsthemilling force prediction obtained from the model. A close agreement betweenmodelpredictionandmeasurementwasobtained.The influenceofnotaligningtoolandlaminatemidpointsisreflected ontheFzmeasurements.

Milling force model proposed here can be extended to investigate issues like surface location error and machining stability[8].Itmustbenotedthatduetolowmachiningspeeds andhighradialforces,processdampingmustbetakenintoaccount intheanalysis.Inthenextsection,theinfluenceoftoolwearis investigated.

4. Influenceofmachiningondiamondcoating

AnimportantconsiderationinmillingofCFRPsisthesurface quality,sinceitaffectstheassemblyconditionsoftheproduced parts. Surface quality is directly related to tool condition. A commonissuewithdiamondcoatedcarbidetoolsisthattheir performance significantly deteriorates when the diamond coating is fractured and carbide substrate is exposed. Fig. 9

showsthevariationofmillingforcesandsurfaceroughnessasa function of the number of passes (one pass corresponds to 900mm ofcut). After17passes,themagnitude ofthemilling forces doubles. Fractured diamond coating is also shown in

Fig.9(c)and(d).Surfaceroughness(Ra)values(measuredwith MarsurfM300)increasedfrom2

m

mto12

m

mbetweenfirstand 17thpasses.

5. Conclusions

A mechanistic model for milling of multidirectional CFRP laminatesusingdoublehelixmillingtoolsisproposed.Cuttingand edge coefficients are calculated based on the laminate fiber direction.Theinfluenceoftoolgeometryandmillingparameters ontheprocessoutputscanbeinvestigatedbyusingtheproposed model. Due to the abrasive nature of the carbon fibers, edge rounding and fracturingof thediamondcoating both influence millingforces.

Acknowledgement

Theauthorswouldliketoacknowledgefinancialsupportfrom The Scientific and Technological Research Council of Turkey (Tubitak).

References

[1]Ko¨nigW,WulfChGraßP,WillerscheidH(1985)MachiningofFibreReinforced Plastics.CIRPAnnalsManufacturingTechnology34(2):537–548.

[2]TetiR(2002)MachiningofCompositeMaterials.CIRPAnnalsManufacturing Technology51:611–634.

[3]Sheikh-AhmadJY(2009)MachiningofPolymerComposites,Springer.

[4]KallaD,Sheikh-AhmadJ,TwomeyJ(2010)PredictionofCuttingForcesin HelicalEndMillingFiberReinforcedPolymers.InternationalJournalofMachine Tools&Manufacture50:882–891.

[5]HintzeW,HartmannD,Schu¨tteC(2011)OccurrenceandPropagationof Dela-minationDuringtheMachiningofCarbonFibreReinforcedPlastics(CFRPs)–An ExperimentalStudy.CompositesScienceandTechnology71(15):1719–1726.

[6]KarpatY,BahtiyarO,Deg˘erB(2012)MechanisticForceModelingforMillingof UnidirectionalCarbonFiberReinforcedPolymerLaminates.International Jour-nalofMachineToolsandManufacture56:79–93.

[7]SchulzeV,BeckeC,PabstR(2011)SpecificMachiningForcesandResultant ForceVectorsforMachiningofReinforcedPlastics.CIRPAnnalsManufacturing Technology60(1):69–72.

[8]KarpatY,BahtiyarO,Deg˘erB(2012)MillingForceModellingofMultidirectional CarbonFiberReinforcedPolymerLaminates.ProcediaCIRP1:460–465.

(a) (b) 7.52 7.54 7.56 7.58 7.6 -1000 -500 0 500 1000 Time (s) F s ec r o F x Fy Fz (N) F_x F_y F_z 0 500 1000 1500 -1000 -500 0 500 1000 (deg) Forces F x Fy (N) F_x F y

Fig.8. Validation ofslot millingforcepredictionsfor sixtooth millingtool: N=4000rpm, fr=1250mm/min, 100% radial immersion, ap=10mm. (a)

Measurementand(b)prediction.

Fig.9.(a)Millingforceand(b)surfaceroughnessvariationduetotoolwearasa function of cutting distance (N=4000rpm, fr=254mm/min, 100% radial

immersion,ap=10mm).(candd)SEMimagesoffractureddiamondcoating.

Y.Karpat,N.Polat/CIRPAnnals-ManufacturingTechnology62(2013)95–98 98