Effect of machinability, microstructure and hardness of deep cryogenic treatment in hard turning of AISI D2 steel with ceramic cutting

(1)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning

w w w . j m r t . c o m . b r

Availableonlineatwww.sciencedirect.com

Original

Article

Effect

of

machinability,

microstructure

and

hardness

of

deep

cryogenic

treatment

in

hard

turning

of

AISI

D2

steel

with

ceramic

cutting

Fuat

Kara

a,∗

_,

_Mustafa

_Karabatak

b

_,

_Mustafa

_Ayyıldız

a

_,

_Engin

_Nasc

c

a_Department_of_Mechanical_and_{Manufacturing}_Engineering,_Technology_Faculty,_Düzce_University,_Düzce,_Turkey b_Institute_of_Science_and_Technology,_Ondokuz_Mayıs_University,_Samsun,_Turkey

c_Engin_PAK_Cumayeri_Vocational_School,_Düzce_University,_Duzce,_Turkey

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received19September2019 Accepted16November2019 Availableonlinexxx Keywords: ANN Machinability Deepcryogenic Microstructure Hardness

a

b

s

t

r

a

c

t

ThisstudyexaminedthehardturningofAISID2coldworktoolsteelsubjectedtodeep cryogenicprocessingandtemperingandinvestigatedtheeffectsonsurfaceroughnessand toolwear.Inaddition,theeffectsofthedeepcryogenicprocessesonmechanical proper-ties(macroandmicrohardness)andmicrostructurewereinvestigated.Threegroupsof testsampleswereevaluated:conventionalheattreatment(CHT),deepcryogenictreatment (DCT-36)anddeepcryogenictreatmentwithtempering(DCTT-36).Thesamplesinthefirst groupweresubjectedtoonlyCHTto62HRchardness.Thesecondgroup(DCT-36)underwent processingfor36hat−145◦_C_after_conventional_heat_treatment._The_latter_group_(DCTT-36)

hadbeensubjectedtobothconventionalheattreatmentanddeepcryogenictreatment fol-lowedby2hoftemperingat200◦C.Intheexperiments,Al2O3+TiCmatrix-baseduntreated

mixedaluminaceramic(AB30)andAl2O3+TiCmatrix-basedTiN-coatedceramic(AB2010)

cuttingtoolswereused.Theartificialintelligencemethodknownasartificialneural net-works(ANNs)wasusedtoestimatethesurfaceroughnessbasedoncuttingspeed,cutting tool,workpiece,depthofcutandfeedrate.Fortheartificialneuralnetworkmodeling,the standardback-propagationalgorithmwasfoundtobetheoptimumchoicefortrainingthe model.Threedifferentcuttingspeeds(50,100and150m/min),threedifferentfeedrates (0.08,0.16and0.24mm/rev)andthreedifferentcuttingdepths(0.25,0.50and0.75mm)were selected.Toolwearexperimentswerecarriedoutatacuttingspeedof150m/min,afeedrate of0.08mm/revandacuttingdepthof0.6mm.Asaresultoftheexperiments,thebestresults forbothsurfaceroughnessandtoolwearwereobtainedwiththeDCTT-36sample.When cuttingtoolswerecompared,thebestresultsforsurfaceroughnessandtoolwearwere obtainedwiththecoatedceramictool(AB2010).Themacroscopicandmicrohardness val-ueswerehighestfortheDCT-36.Fromthemicrostructuralpointofview,theDCTT-36sample showedthebestresultswithhomogeneousandthinnersecondarycarbideformations.

∗ _{Corresponding}_author.

E-mail:fuatkara@duzce.edu.tr(F.Kara).

https://doi.org/10.1016/j.jmrt.2019.11.037

(2)

1. Introduction

Inrecentyears,thecryogenicprocesshasbeenimplemented asacomplementarymethodtotheheattreatmentof work-piecesinordertoimprove theworkability ofthe materials

[1].Thecryogenicprocessisappliedinordertoincreasewear resistance,generallyinmaterialssubjectedtohighwear.Itis acheapandlastingtreatmentthatiscarriedoutasasingle operationand,unlikecoatings,iseffectiveontheentire work-piece.Thecryogenicprocessisclassifiedasshallowcryogenic (between−50◦_C_and ₋₈₀◦_C)_and _deep _cryogenic

(tempera-turesbelow−125◦C)treatment,dependingontheapplication temperaturesonthematerial.Aftertheheattreatment,the materialsarebroughttoroomtemperaturegraduallyby keep-ingthem forapredeterminedwaitingperiodatshallowor deepcryogenicprocessingtemperatures. Withthismethod, martensite transformation ofresidual austenite, formation offinecarbideprecipitatesandhomogeneouscarbide distri-butionareachievedinconventionallyheat-treatedmaterial. Thus,themechanicalpropertiesofmaterialssuchas hard-nessandabrasionresistancearesignificantlyimproved[2].In thevastmajorityofstudiesoncryonicprocessingretention time,theidealwaitingtimewasfoundtobe36h[3–6].

Das[7]aimedtodeterminetheoptimumholdingtimeby investigatingtheeffectsofdeepcryogenicprocessingapplied at −196◦_C _for _different _retention _times _of ₀₋₁₃₂_h _on _the

wearproperties,hardnessvaluesand microstructural char-acteristics of AISI D2 cold work tool steel. As a result of thestudy,deepcryogenicprocessingwasshowntoincrease wear resistance. As demonstrated by microstructure pho-tographs and hardness and surface roughness values, the highest wear resistance increase (84.88%) was obtained in cryogenic samples held for 36h. In a study conducted by Aminietal.,theAISID3coldworktoolsteelwassubjected to different curing times (24, 36, 48, 72, 96 and 120h) at deep cryogenic processing temperatures to determine the microstructural, carbide distribution and macro and micro hardnessvariationsdependingoncryogenicprocessing hold-ing time.Microstructural changes inthe AISID3 toolsteel specimendeepcryogenicallytreatedfor36hshowedthebest resultintermsofcarbidedistributionandmacroandmicro hardness[8].Inhisstudy,Kara[6]investigatedtheeffectof deepcryogenicprocessing(−145◦_C)_on_the_{microstructure}_and

mechanicalpropertiesofAISI 52100bearingsteel at differ-entretentiontimes(12,24,36,48and60h).Amongthedeep cryogenicallytreatedsamples,thebestmechanicalproperties wereobtainedat36hofdeepcryogenictreatment.Moreover, the36-hdeepcryogenicallytreatedspecimensalsoshowed thebestmicrostructuralpropertieswithmorehomogeneous microstructureandthinnercarbideprecipitation.Takinginto accounttheresultsoftheresearchcarriedoutinthe litera-ture,thedurationofdeepcryogenictreatmentwaschosenas 36hoursinthisstudy.

Nowadays,awholerangeofneuralnetwork methodolo-gieshavebeendevelopedthatmodelthecorrelationbetween theinputandtheoutputparametersoftheturningprocess

[6].OzelandKarpat[9]usedregressionandartificial neural network(ANN)modelsforestimatingthesurfaceroughness andtoolwearinthehardturningofAISIH13steelusingCBN

inserts.DavimandFigueira[10]studiedtheinfluenceof cut-tingspeedandfeedrateonflankwear,specificcuttingforce andsurfaceroughnessinthehardturningofAISID2coldwork toolsteelusingconventionalceramicinsertsandstatistical techniques.OzelandKarpat[11]investigatedthemachiningof AISID2steelusingmulti-radii/wipermixedaluminaceramic insertswithTiNcoatinganddevelopedanANNforpredicting toolwearandsurfaceroughness.Quiza[12]compared statis-ticalmodelswithanANNforestimatingtoolwearinthehard turningofAISID2steelusingconventionalceramicinserts. Kumar[13]focusedontheinvestigationofflankwear, aver-agesurfaceroughnessandchip-toolinterfacetemperaturein themachineturningofheat-treatedAISID2-gradetoolsteel usingcoatedcarbideinserts.Responsesurfacemethodology (RSM)basedmodelsandANNmodelswereimplementedto estimatetheresponsesinhard-turning.

AlthoughAISID2coldworktoolsteelhasaverywidespread application area, studies carried out on the workability of thismaterialwerefoundtobeverylimitedintheliterature, andindeed,nostudyhastodateexaminedthemachinability ofAISID2coldworktoolsteelsubjectedtocryogenic treat-ment,althoughtherearestudiesthatexaminetheeffectsof cryogenicprocessingonmachinabilityfordifferenttypesof materials. Thissituation necessitatedastudy tobecarried out.Thisstudyaimedtoimprovecuttingconditionsby apply-inginadditiontoheattreatment,deepcryogenictreatment anddeepcryogenictreatment+temperingandtoinvestigate themateriaslunderprocessingconditionsusingcoatedand uncoatedceramictools.Byimprovingthecuttingconditions, the aimwastodecrease thesurfaceroughnessandreduce machiningcostsbyincreasingtoollife.

This study consists ofthree parts. The first part evalu-ates AISI D2coldwork toolsteel withthreedifferent heat treatments(conventionalheattreatment,conventionalheat treatment+deepcryogenicprocessingandconventionalheat treatment+deepcryogenicprocessing+tempering),using dif-ferent cutting tools and different cutting speed, feed rate and depth of cut combinations. The surface roughness of themachineddeepcryogenicallytreatedcoldworktoolsteel workpiece and the effectson toolwearare discussed.The second part investigates the effects of conventional heat treatment,conventionalheattreatment+deepcryogenic pro-cessing and conventional heat treatment+deep cryogenic procesing+temperingonthemicrostructureandmacroand microhardnessofAISID2coldworktoolsteelmaterial.The thirdpartestimatestheexperimentalvaluesofsurface rough-ness(Ra)usingANN.

2. Material

and

methods

2.1. Workpiece,machinetoolsandcuttingparameters

Cylindrical AISI D2 cold work tool steel in dimensions of Ø60×300mm was used. Thechemical composition of the test samples isgiven in Table1. The hard turning experi-mentswereperformedonaGOODWAYGLS-1500CNClathe usinganuncoatedSNGA120408T01020AB30turningtooland SNGA120408T01020AB2010ceramicturningtips manufac-turedbyTaeguTeccuttingtoolcompanytoperformthehard

(3)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Table1–Chemicalcomposition(%)oftestsamples.

C Si Mn P S Cr Mo V

1.575 0.32 0.30 0.024 0.0020 11.70 0.74 0.960 lathetests.APSBNR2525M12turningtoolholderwasusedto connectthecuttingtools.

Hardturningexperimentswerecarriedoutatthree differ-entcuttingspeeds(50,100,150m/min),threedifferentfeed rates(0.08,0.16and0.24mm/rev)andthreedifferentcutting depths(0.25,0.50and0.75mm).Twodifferenttypesofceramic tools(withandwithoutcoating)weretestedunderdry cut-tingconditions onmaterialsubjected toconventional heat treatment,conventionalheattreatment+deepcryogenic pro-cessingandtraditionalheattreatment+36hdeepcryogenic processing+tempering.Atotalof162cuttingtestswere car-riedoutineachcombination.Foreachcuttingtool,thetool weartestswereperformedaccordingtodifferentmachining timesataconstantcuttingspeed,feedrateandcuttingdepth.

Table2givestheparametersusedinthesurfaceroughness

andtoolweartests.

2.2. Deepcryogenictreatment

The test specimens were divided into three batches, CHT (TraditionalHeat Treatment),DCT-36 (TraditionalHeat Treatment +36h Deep Cryogenic Processing) and DCTT-36 (TraditionalHeat Treatment +36h Deep Cryogenic Process-ing+Tempering).TheAISID2coldworktoolsteelhadbeen subjectedtopreheating,austenitizingandtemperingpriorto cryogenicprocessing.Thepreheatingprocessinvolved30min at450◦C,60minat650◦Cand 30minat 850◦C.After pre-heating, the samples were austenitized by heating in an atmosphere-controlledfurnaceat1030◦Cfor60min.Afterthe austenitizingprocess,rapidcoolinginnitrogenwasperformed at4barpressureinavacuumoven.Finally,thesampleswere temperedat200◦Cfor180minand at350◦Cfor180minto ahardnessof60–62HRc.Deepcryogenictreatmentwasthen appliedtotheDCT-36andDCTT-36samplesat−145◦_C_for₃₆_h.

Finally,fortheDCTT-36sample,theheattreatmentand cryo-genicprocessingwascompleted bytempering at200◦C for

180min.Table3showstheconventionalheattreatmentand

Table3–Heattreatmentanddeepcryogenicprocessing.

AISID2coldworktoolsteel

Process Temperature Time

1.Preheating 450◦C 30min

austenitizing 1030◦C 60min

Cooling 4BarNitrogen coolingatpressure

–

1.Tempering 200◦C 180min

2.Tempering 350◦C 180min

Deepcryogenicprocess −145◦_C ₃₆_hour

Tempering 200◦C 180min

deepcryogenicprocessesappliedtotheAISID2coldworktool steel.

2.3. Measurementofsurfaceroughnessandtoolwear

Measurement and evaluation of surface roughness is very important in workability studies [14]. The Taylor Hobson Surtronic25surfaceroughnesstesterwasusedtomeasure thesurfaceroughnessofthemachinedsurfaces.Thesurface roughnesswasmeasuredinthreeplacesfromthetreated sur-facesandtheiraveragedeterminedtheroughness(Ra)values. TheeffectsoftheAISID2coldworktoolsteelsubjectedto dif-ferentheattreatmentsonthewearperformanceofthecoated anduncoatedceramiccuttingtoolsunderhardturning condi-tionswereinvestigatedataconstantcuttingspeed,feedrate andcuttingdepth.TheCHT,DCT-36andDCTT-36specimens weremachinedwithuncoatedandcoatedceramictoolsata cuttingspeedof150m/min,afeedrateof0.08mm/revanda cuttingdepthsof0.6mmandweresubjectedtohardturning (2,4and6min)andtoolweartestswerecarriedout.During theabrasiontests,thecuttingprocesswasstoppedatcertain intervals and theworn surfaceswere photographed bythe Dino-Litedigitalmicroscope.Afterthewidthofthecuttingtool (4.76mm)wasintroducedintheDinoCapture2.0program,the amountofnoseandcraterwearonthetoolwasmeasured. Theamountofcuttingtoolwearwasevaluateddependingon theheattreatmenttypeandmachiningtime.Inaddition,by

Table2–Surfaceroughnessandtoolweartestparameters.

Surfaceroughnesstests

Cuttingtool AB30 AB2010

Heattreatment CHT DCT-36 DCTT-36

Cuttingspeed(V,m/min) 50 100 150

Feedrate(f,mm/dev) 0.08 0.16 0.24

Cuttingdepth(a,mm) 0.25 0.50 0.75

[10pt] Toolweartests

Cuttingtool AB30 AB2010

Heattreatment CHT DCT-36 DCTT-36

Cuttingspeed(V,m/min) 150 Feedrate(f,mm/dev) 0.08 Cuttingdepth(a,mm) 0.6

(4)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Table4–Statisticaldataofsurfaceroughnessforfivelearningalgorithms.

Learningalgorithm Numberofneurons Trainingdata Testingdata

RMSE R2 _RMSE _R2 BFGS 5-10-1 0.0963 0.9523 0.0930 0.9540 BFGS 5-11-1 0.1005 0.9528 0.0913 0.9591 BFGS 5-12-1 0.1091 0.9387 0.1083 0.9337 BFGS 5-13-1 0.1042 0.9587 0.0952 0.9638 BFGS 5-14-1 0.1149 0.9428 0.1076 0.9480 BFGS 5-15-1 0.1336 0.9426 0.1250 0.9465 CGP 5-10-1 0.0991 0.9425 0.0954 0.9448 CGP 5-11-1 0.0929 0.9500 0.0896 0.9512 CGP 5-12-1 0.0830 0.9649 0.0770 0.9680 CGP 5-13-1 0.0974 0.9515 0.0938 0.9533 CGP 5-14-1 0.1163 0.9485 0.1017 0.9572 CGP 5-15-1 0.1021 0.9595 0.0950 0.9628 LM 5-10-1 0.0906 0.9572 0.0821 0.9630 LM 5-11-1 0.0766 0.9741 0.0720 0.9763 LM 5-12-1 0.1005 0.9478 0.0900 0.9568 LM 5-13-1 0.1055 0.9414 0.0981 0.9476 LM 5-14-1 0.0783 0.9704 0.0692 0.9755 LM 5-15-1 0.0841 0.9655 0.0798 0.9666 RP 5-10-1 0.0956 0.9642 0.0921 0.9656 RP 5-11-1 0.1021 0.9613 0.0952 0.9649 RP 5-12-1 0.1030 0.9410 0.1080 0.9312 RP 5-13-1 0.0775 0.9714 0.0721 0.9738 RP 5-14-1 0.0793 0.9736 0.0709 0.9778 RP 5-15-1 0.1092 0.9329 0.1044 0.9367 SCG 5-10-1 0.0946 0.9473 0.0931 0.9453 SCG 5-11-1 0.0859 0.9573 0.0905 0.9498 SCG 5-12-1 0.1020 0.9386 0.0956 0.9426 SCG 5-13-1 0.0839 0.9650 0.0754 0.9705 SCG 5-14-1 0.0914 0.9650 0.0750 0.9755 SCG 5-15-1 0.0890 0.9588 0.0828 0.9623

photographingtheabradedsurfaceswiththeSEMdevice,the typesofabrasionformedweredetermined.Inorderto deter-minecraterwear,SEMphotographsweretakenona1:1scale inaCADenvironment,thecraterzoneswerepreciselydrawn andtheirareaswerecalculated.

2.4. SEMandhardnessanalyses

Hardnessmeasurements ofthe testspecimenswere made onboth microand macrohardness testers.For both hard-nessmeasurements,atotalof36specimenswereprepared, 12ofwhichwere10×10mm.Eachmacroandmicrohardness measurementreflected theaverage ofatleast10 hardness measurements. Micro hardness measurements were per-formedwithaMetkonmicrohardnesstester.Macrohardness measurementsweremadeusingtheRockwell(HRc)hardness measurementmethodwiththeBulutMakinamacrohardness tester.One10×8mmsamplewaspreparedforeachtypeof heattreatment foruseinthe microstructurestudies.After conventionalheattreatment,36hofdeepcryogenictreatment and 36h of deep cryogenic treatment+tempering, sanding with120,240,600,800and1200gritSiCsandpaperwascarried outonthesamplesandtheywerethenpolishedinthesample shaverforabout5min.Theywerethenexaminedbyoptical microscopyand preparedforscanning electronmicroscopy (SEM)imagingwith3%Piccale(97mLofethylalcohol,3gof picricacid).Intermsofresolution,SEMisat2000Åand opti-calmicroscopesat25Å,whilethescanningdepthinscanning

electronmicroscopyis300–600,whichis30×higherthanin opticalmicroscopy.Forthisreason,themicrostructure pho-tographswere takenwith anFEIQuanta FEG250 scanning electronmicroscopeformoredetailedhighmagnification.

2.5. Artificialneuralnetworks

Artificialneuralnetworks(ANNs)imitatesomebasicaspects ofthebrainfunctions[15,16].Aneuronisthebasicelementof theneuralnetworks,anditsshapeandsizemayvary depend-ing on its functions [17]. The simplest neural network is composedofneurons,inputs,weightsandasummation func-tion,activationfunctionandoutput.Thesummationfunction calculatesthenetinputoftheneuron,asshowninEq.(1).

NETi=

n

j=1wijxj+wbi (1)

WhereNETiistheweightedsumoftheinputtotheith

pro-cessingelement,nthenumberofprocessingelementsinthe previouslayer,iandjtheprocessingelements,wijtheweights

oftheconnectionsbetweenithandjthprocessingelements,xj

theoutputofthejthprocessingelementandwbitheweights

ofthebiasesbetweenlayers.Theactivationfunction,which processesthenetinputoftheneuron,definestheoutputof theneuron.Manyfunctionssuchasthethresholdfunction, stepactivationfunction,sigmoidfunction,andhyperbolic tan-gentfunctionareusedtodefinetheactivationfunction.The

(5)

sigmoidfunctionisgenerallyusedforthetransferfunction andgeneratesavaluebetween0and1foreachvalueofthe netinput. ThelogistictransferfunctionoftheANNmodel improvedinthisstudyisgiveninEq.(2).

f (NETi)=

1

1+e−NETi (2)

The optimal learning algorithm and network structure shouldbedeterminedtoobtaintheoutputvaluesclosestto theexperimentalvalues.Tothisend,thenumberofneurons inthehiddenlayerwasboostedstepbystep(i.e.,fromtento fifteen),andquasi-Newtonbackpropagation(BFGS),conjugate gradientbackpropagation(CGP),Levenberg–Marquardt(LM), resilientbackpropagation(RP),andscaledconjugategradient (SCG)learningalgorithmswereusedtodefinetheoptimal net-workstructureandlearningalgorithm.Thetrialsconducted inthisstudyshowedtheLMlearningalgorithmtobethebest learningalgorithmforthesurfaceroughness. The determi-nationofthebestlearningalgorithmandoptimalnumberof neuronsforthesurfaceroughnessaregiveninTable4.

Table4.Statisticaldataofsurfaceroughnessforfive

learn-ingalgorithms.

Therewerefiveinputparametersinthenetwork:cutting tool(ct),workpiece(wp),cuttingspeed(V),depthofcut(dp), andfeedrate(f).Therewasoneoutputparameterinthe net-workasthesurfaceroughness(Ra).Thebestnetworkstructure wasidentifiedas5-14-1(Fig.1).Asaresultofthetests,the experimentaldataobtained(162foreachinputandoutput) werepreparedforthetestingandtrainingsetsoftheANN. Theratiofortestingandtrainingdatawasselectedas approx-imately15:85, i.e.,32and 130sets ofall experimentaldata were arbitrarily selected for the testing and training data, respectively.Inthisstudy,theinputandoutputvalueswere normalizedbetween0and1toattainbetterpredictions.The surfaceroughnessvaluesestimatedafterANNtrainingwere checkedagainsttheexperimentaldata.

Therootmeansquareerror(RMSE)andcorrelation coeffi-cient(R2₎_values_were_used_for_comparison._The_most_precise

techniquefortheproblemunderstudywaschosenfromthe algorithmsofthestate-of-the-art forregression.TheRMSE wasusedtocomparetheregressors.Thismetricmeasuresthe deviationthatthepredictedclassvaluehaswithregardtoits realvalue[17].Eqs.(3)and(4)givetheformulaefortheRMSE

andR2_[18,19]_. RMSE=

⎛

⎝

1 p

j

tj−oj

2

⎞

⎠

1/2 (3) R2=1−

j

tj−oj

2

j

oj2

(4)

where,tisthegoalvalue,oistheoutputvalue,andpisthe numberofsamples.

3. Experimental

results

3.1. Outputsofsurfaceroughness

TheAISID2cold-worktoolsteelwassubjectedtoconventional heattreatment(CHT), 36-hdeepcryogenictreatment (DCT-36)and36-hdeepcryogenictreatment+tempering(DCTT-36). Thecuttingparametersandthechangesthatoccurredin sur-faceroughnessdependingoncuttingconditionsafterthehard turning testsusing coated and uncoatedceramictoolsare showninFig.2.Ingeneral,surfaceroughness(Ra)valueswere foundto varybetween0.22␮mand 3.1467␮m. TheRa val-uestendedtodecreaseforbothtoolswithincreasingcutting speedforallofthecuttingparametervalues.Theincreasein cuttingspeedreducedthetool-to-chipcontactarea,thereby reducingfriction,whichallowedforbettersurfacequalityto be achieved. Some researchers, however, have argued that thedecreaseinRaduetoincreasedcuttingspeedisdueto thereducedtendencytowardstackingchipformation[20–22]. However,athighcuttingspeed(150m/min),thetoolwear val-ues wereslightlyincreased.Inthis case,the increasedtool wearcanbeexplainedbytheincreaseoftheloadonthe cut-tingtoolandthehightemperaturesgeneratedwhilecutting athighspeed.

Aftera300%increaseincuttingspeed,thesurface rough-nessvaluesimprovedby52%atlowfeedrate(0.08mm/rev) values. However,when high feed rate(0.24mm/rev) values werereached,Ravaluesdecreasedby45%uptoacuttingspeed of100m/min.Althoughwhena50%increaseincuttingspeed to150m/minwas reached,Raincreasesofupto25%were seen.Thiscanbeexplainedbythedecreaseintoolchip con-tactareawiththeincreaseofcuttingspeed,thetoolwearwith thehighcuttingparametersandtheexcessivedeformationof themachinedsurface[23].

Feedrateisoneoftheimportantparametersthat deter-minesthe characterofthecuttingprocess[24,25].Interms

(6)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.2–Surfaceroughness(Ra)values.

ofcuttingparametersandcuttingtools,increasingfeedrates havebeenfoundtobeanimportantfactorinincreasing sur-faceroughnessvalues.Kacal[26]studiedtheeffectofvariation infeedratevaluesonthesurfaceroughnessofPMD-23steel producedbytheT/Mmethodandevaluatedtoolwearand sur-faceroughnessinturningusingcoatedceramiccuttingtools. Asthefeedrateincreased,theRavaluesmeasuredforalltools werealsoseentoincrease.TheRavaluesatthelowestfeed rate(0.08mm/rev)forallofthemachiningparametersranged

between0.18and1.723␮mwhiletheRashowedasignificant increase at the highestfeed rate (0.24mm/rev), reachinga valueof3.41␮m.Increasingthe feedratecausedtheshear forces andvibrationtoincreasebyincreasingthechip vol-ume removedatthe unittime,thus increasingthesurface roughness. Inaddition, increaseinthefeedrate aswellas thecuttingspeedcausedthetemperatureatthecutting tool-workpieceinterfacetoincrease.Thetemperatureincreaseat theinterfacethenledtotoolwearandconsequentlycaused

(7)

thesurfaceroughnesstodeteriorate.Thisindicatedthatthere wasadirectrelationshipbetweentoolwearandsurface rough-ness,asstatedintheliteraturestudies[27,28].

LookingatFig. 2, itcan beseen thatRa increasedwith increasing depth of cut. The Ra values are in the range of 0.18–3.14␮m at 0.25mm cutting depth, 0.23–3.08␮m at 0.50mmcuttingdepth and0.23–3.41␮mat0.75mmcutting depth. The best surface roughness value was 0.18␮m at 0.25mmcuttingdepthandthehighestRavaluewas0.75mm ata cutting depth of 3.41␮m. Increaseddepth ofcut and increasedsurfaceroughnesshavebeenconfirmedby numer-ousliteraturestudies[29].Thedepthofcutalsodirectlyaffects thecrosssectionareaofthekerfthatthecuttingtooltriedto removeduringthecuttingprocess.Thefirstimportantfactor intermsoftheeffectofdepthofcutonthesurface rough-nessisthechipformation.ItisknownthatRaisdecreased asthechipsizedecreases[23].Alongwithincreasedchipsize, theslipplaneareainthefirstdeformationzonealsogrows andmakesthecuttingprocessdifficult.Therefore,thecutting forcevaluesandthusthesurfaceroughnessareincreased.

Thesecondfactor isthe cuttingtemperature.There are differentmechanismsthatinfluencethetemperatureduring cutting.Plasticdeformationprimarilyoccursinthefirst defor-mationzone.Heatenergyisgeneratedbyfrictionandplastic deformationintheseconddeformationzoneandfinally,heat energyisformedintheregioncalledthethirddeformation zone,wheretheflanksurfaceofthecuttingtoolcomesinto contactwiththeworkpiece[30,31].Inparticular,theincreases inthefeedrateandcuttingdepthincreasetheareaoftheslip surfaceinthefirstdeformationzone,sothatmoreenergyis requiredtobreakoffchipfromthesurfaceandmoreheatis releasedasaresultofthisconsumedenergy.Inaddition,with theincreaseinchipsize,thefrictioninthetool-chipinterface increasesintheseconddeformationzone,thusaffectingthe cuttingtemperature.Cuttingtoolwearoccursparalleltothe increaseofthecuttingtemperature.Thequicklyworncutting toolcausesworsesurfaceroughness.Inaddition,increasein depthofcutcausesanincreaseinsurfaceroughness.

3.2. Outputsofnoseandcraterwear

3.2.1. Changeofnosewearaccordingtoheattreatment type

Aseriesofweartestswereconductedunderdrycutting condi-tionstoinvestigatetheeffectoftemperingafterconventional

heat treatment on toollife, deep cryogenic treatment and deepcryogenictreatmentwithtemperingappliedtotheAISI D2coldworktoolsteel.Toolwearexperimentswerecarried out usingtwotooltypes,Al2O3+TiC matrix-baseduncoated

mixedaluminaceramictools(AB30)andAl2O3+TiC

matrix-based TiN PVD-coated ceramictools (AB2010),at acutting speedof150m/min,afeedrateof0.08mm/revanddepthof cutof0.6mmforfivedifferentprocessingtimes(2,4,6,8and 10min).Asaresultofabrasiontests,bothnoseandcraterwear generallyoccurredonbothtools.Attheendofthe machin-ingtimes,thenosewearvaluesweremeasuredlinearlyona precisioncamcorder.

NosewearchangesaregiveninFig.3.Fromthewear val-ues obtainedforboth cuttingtools,anaverage increaseof 138%innosewearvaluescanbeseenfortheCHTspecimen whenmachiningtimewasincreasedfive-foldfrom2to10min. FortheDCT-36andDCTT-36samples,thisratiowasfoundto be149%and119%,respectively.Accordingtoheattreatment type,thelowestnosewearvalues,likethoseforsurface rough-ness,werealsoobtainedwiththeDCTT-36sample.Whenall processingtimesweretakenintoaccountaccordingtoheat treatmenttype,thelowestnosewearvalueswereobtained withtheDCTT-36sample.

Withtheincreaseinmachiningtime,theheatgenerated in the cutting zone also rose in parallel. This increase in the amount ofheat causedthe temperatureat thecutting tool-workpieceinterfacetorise.Cuttingtoolsmaintaintheir propertiesuptoacertaintemperaturevalue.Whenthecutting toollimitvaluesarereached,theywillbesubjectedtoplastic deformation.Afterpermanentdeformation,varioustypesof wearoccur inthe cuttingtooland itloses itsfunction.For thisreason,theamountofwearincreasesinparallelwiththe increaseinprocessingtime[32].Itisstatedintheliterature thatdependentonincreasingmachiningtime,thewearon the cuttingtooland hence thesurfacequality isworsened

[33].Lima [34]usedthreedifferentcuttingtoolstoperform hardturningonAISI4340andAISID2coldworktoolsteelto investigatecuttingtoollifeandsurfaceroughnessaccording tomachiningtime.Itwasfoundthatduringthe0−20mintime periodthemachininghadwornouttheresidualcuttingtool life.

3.2.2. Changeofnosewearaccordingtocuttingtool

The machining time, type of cutting tool and nose wear changesaccordingtotheheattreatmenttypeareshownin

(8)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.4–Changesinnosewearaccordingtomachiningtimeandcuttingtooltype.

Fig.4.Thenosewearvaluesvariedbetween0.061mmand 0.312mm.Whentheaveragenosewearvaluesobtainedfor theCHT,DCT-36andDCTT-36samplesweretaken,itwasseen thattheweartimeforthe AB30uncoatedceramictoolhad increasedby181%withthefive-foldincreaseofprocessing timefrom2to10min.A75%increaseinnosewearwasseen intheAB2010coatedceramiccuttingtoolwiththefive-fold increasefrom2to10minofprocessing.Attheendofthe 10-mintreatmentperiod,thelowestnosewearvalueof0.061mm wasobtainedwiththeAB2010cuttingtoolontheDCTT-36 specimen.Whencomparingthecoatedanduncoatedcutting toolsand consideringallmachiningtimes,the lowestnose wearvalueswereobtainedwiththeAB2010coatedceramic tool(Fig.4).

Thenosewearvalueswereinparallelwiththoseof stud-iesintheliterature[27,35–37].Dosbeva[38]conductedastudy onthehardturningofAISID2coldworktoolsteelinorderto comparewearonCVD-coatedtungstencarbideandPCBN cut-tingtoolsaccordingtomachiningtime.Nosewearwasseen inbothtools,whilethelowestnosewearvaluewasobtained withthePCBNtool.

3.2.3. Changeofcraterwearaccordingtoheattreatment type

The crater wear values at 150m/min cutting speed, 0.08mm/rev feed rate and cutting depth of 0.6mm at different machining times were determined by calculating the surface area of the craters in a CIS (Computer Aided Design)environment.CraterwearchangesaregiveninFig.5. When the wear values obtained with both cutting tools were taken, the Fig. 5 shows an average increase of 267%

in nose wear values for the CHT sample with a five-fold increase from 2 to10min of processing time. For DCT-36 and DCTT-36samples,thisratiowasfoundtobe173%and

Fig.5–Changeincraterwearaccordingtomachiningtime andheattreatmenttype.

(9)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.6–Changeofcraterwearaccordingtomachiningtimeandcuttingtooltype.

158%,respectively.Whenallprocessingtimesweretakeninto account,the lowestcrater wearvalues were obtainedwith DCTT-36ascomparedtotheotherheattreatmenttypes.The DCT-36samplethen providedthelowestcraterwear, while thehighestwearvalueswere seenwiththe conventionally heat-treatedCHTsample.Thiswasattributedtothe improve-mentofthemechanicalpropertiesofthecuttingtoolmaterial and the more homogeneousmicrostructure resulting from thecryogenicprocessingandsubsequenttempering[39,40]. Thanks to the cryogenic processing, there was a positive increaseinthewearresistanceofthecuttingtoolmaterial andlikewise,amoreevidenthomogeneousarrangementin themicrostructure.Afterthesetwopositivesituations,itwas supposedthatthetoolwearontheabradedsurfacesofthe cuttingtoolduringcuttingwouldoccuratalowerrate.

3.2.4. Changeofcraterwearaccordingtocuttingtool

The machining time, type of cutting tool and changes in craterweardependentonheattreatmenttypeareshownin

Fig.6.Itcanbeseenthatcraterwearvalueschangedbetween 0.0177mm2 _and_0.0684_mm2_._When _the_average_crater _wear

valuesobtainedfortheCHT, DCT-36 andDCTT-36 samples weretaken,therewasa227%increaseincraterwearinthe AB30uncoatedceramictoolwiththefive-foldincreasefrom2 to10minmachiningtime.IntheAB2010coatedceramic cut-tingtool,therewasa159%increaseinthecraterwearwiththe 5-foldincreaseinmachiningtimefrom2to10min.Attheend ofthe10minmachiningcycle,thelowestcraterwearvalue of0.0298mm2_was_obtained_with_the_AB2010_cutting_tool_on

theDCTT-36specimen.ThelowestwearvaluesforCHTand DCT-36specimenswere0.0379mm2_and_0.0332_mm2_,

respec-tively. When all machiningtimes were considered and the coatedanduncoatedcuttingtoolswerecompared,the low-estcraterwearwasobtainedwiththeAB2010coatedceramic tool.ComparedtotheTiN-coatedceramictool(AB2010)Al2O3

+TiCmatrix-baseduncoatedcompositealuminaceramictool (AB30),theAl2O3+TiCmatrix-basedPVD-coatedtoolresulted

in164.%lowercraterwearafter10minmachiningtime.This resultwas associatedwithalowcoefficientoffriction and goodcraterwearresistance,althoughtheTiNcoatingonthe toplayerofthiscuttingtoolisnotaveryhardmaterial[25,33]. Itispossibletosaythatthecraterwearvaluesincreased withtheincreaseinthemachiningtime.Thechangeofcrater wearovertimeislikethechangeofnosewearovertime. Mod-eratecraterweardoesnotusuallylimittoollife.Indeed,the formationofcraters enhancesthe effectivenessofthetool threadangleandthus reducesthecuttingforces.However, excessivecraterwearwillweakenthecuttingedgeandthis will cause deformation or fracture ofthe tool [33,41]. This results in worsening quality of the workpiece surface and causesweartypessuchasflankwearandcraterwearonthe surfaceofthecuttingtoolandedgeregions.

3.3. Outputsofmicrostructureandhardness

3.3.1. Outputsofmicrostructure

MicrostructureviewsofthetestspecimensaregiveninFig.7. TheCHT sampleexhibitedanon-uniformcarbide distribu-tion,whiletheDCT-36sampleexhibitedauniformprimary carbideandanearlysphericalsecondarycarbidedistribution. However,afterthe36-hdeepcryogenicprocessing+tempering (DCTT-36), the carbide dimensions decreased and a more

(10)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.7–MicrostructureviewsofAISID2coldworktoolsteelsamples(LPC-Largeprimarycarbides,LSC-Largesecondary carbides,SSC-Smallsecondarycarbides).

homogeneouscarbidedistributionwasobserved.Whenthe heattreatedspecimensarecomparedwitheachother,itcan beseen that the microstructureof the DCTT-36 sample is thinnerand hasamorehomogeneous structure.Das[7,42]

investigatedchangesinmicrostructurebyperformingdeep cryogenicprocessingandsubsequenttemperingatdifferent holdingtimes(0,12,36,60and84h)ontheAISID2coldwork tool.Asaresultofthestudy,theyreportedthatthehighest percentageofcarbidewasinthedeepcryogenicsamplefor 36h.Ifageneralassessmentofthemicrostructurestudiesis made,thedeepcryogenicprocessing+temperingappearsto haveprovidedmorehomogeneousand moredensecarbon distribution.Inaddition,itwasdetermined thattheresults obtainedfromthemicrostructurestudiesareinconcurrence withthestudiesintheliterature.

3.3.2. Outputsofhardness

3.3.2.1. Changeofmacrohardness. Fig.8showsthechangein microhardnessvaluesofAISID2coldworktoolsteelsubjected to different heat treatment and deep cryogenic processes. As can be seen in Fig. 8, the highesthardness values are respectivelyintheDCT-36,DCTT-36andCHTspecimens.The hardnessesofCHT,DCT-36andDCTT-36weremeasured as 62.2,63.1and62.8HRc,respectively.Macroscopicrecoveriesof deepcryogenicallytreatedsamplesaccordingto convention-allyheattreatedsampleswerefoundtobe1.44%and0.96% for DCT-36 and DCTT-36, respectively. The highest

macro-Fig.8–Changeofmacrohardnessvaluesaccordingtoheat treatmenttype.

scopicvalueamongtheheat-treatedsampleswasobtained atthe DCT-36 sample.Thiswas attributedtothe fact that the austenitemartensite transformationinthe microstruc-tureofthematerialwiththecryogenicprocessingoccurred atahigherrateintheDCT-36samplethanintheCHT and DCTT-36samples[6,43–46].TheAISID2coldworktoolsteel hasasoftinnerstructure,whichtransformsfromthe austen-itephaseintoahardermartensitephase,resultinginastiffer structure.Temperingafterdeepcryogenicprocessingcaused aslightdecreaseinhardness.Forthisreason,thehardnessof

(11)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.9–Changesofmicrohardnessvaluesaccordingtoheat

treatmenttype.

theDCTT-36samplemeasuredathighervaluesthantheCHT sample,butwaslowerthantheDCT-36sample.

Similarresultshavebeen observedinthe literature[46]. In their study, found a 22% increase in the macroscopic extentof carbide material with deep cryogenic processing

[45]. In another study, Rhyim et al. (2006) suggested that deepcryogenicprocessingimprovesstiffness.Sonawane[47]

performedconventionalheattreatment,deepcryogenic pro-cessinganddeepcryogenicprocessingplustemperingonM2 toolsteel.Theynotedthatthehighestmacrohardnessvalue wasobtainedfromthedeepcryogenicsample.

3.3.2.2. Changeofmicrohardness. Fig.9showsthemicro hard-nessvariationsofAISID2coldworktoolsteelsamples.The graphshows that the highestmicro hardnessvalues were obtained with the DCT-36, DCTT-36 and CHT samples as 871.8HV,748.46HVand618.46HV,respectively.Thechangein microhardnessvalueswasinparallelwiththemacro hard-nessresults.Improvementsinmicro hardnessvalueswere foundtobe41%and21%forDCT-36andDCTT-36,respectively. Whentheheattreatmentsappliedtothematerialwere com-pared,thehighestmicrohardnessvaluewasobtainedwiththe DCT-36sample.Thiswasattributedtothefactthatwiththe deepcryogenicprocessing,theaustenite-martensite transfor-mationinthemicrostructureofthissampletookplaceata higherrate than intheother samples, resultingina more brittlestructure.

Themicro hardnessresultswereinagreementwiththe literature[48–51].Das[52]intheirwork,showedthatamong theAISID2coldworktoolsteelsamplessubjectedto conven-tionalheattreatment,shallowcryogenicprocessing,anddeep cryogenicprocessing,thedeepcryogenicsamplereachedthe highestmicrohardnessvalues[48].Oppenkowski[53]reported

that after 24–36h of cryogenic treatment, the micro hard-nessofAISID2coldwork toolsteelwashigherinthe36-h cryogenicallytreatedsamples.Amini[54]reportedthatdeep

cryogenic processingatdifferent retention timesincreased stiffnessandachievedthehighesthardnessvaluesinterms ofbothmacroand microhardnessforthetoollife.Nanesa

[48]applieddifferentheat treatmentstoAISID2coldwork toolsteel.Improvementsinmicrohardnessvalueswerefound tobe7.7%fordeepcryogenicallytreatedsamplescomparedto conventionallyheat-treatedsamples.Intheirstudy,Amini[55]

observedanincreaseof5.7%–9.6%inthehardnessofAISIH13 steelaftercryogenictreatment.Inanotherstudywhere tradi-tionalheattreatmentandcryogenictreatmentwereapplied totheEN31steel,thehardnessofthecryogenicallytreated materialwasfoundtoincreaseby14%[55].Das[52]applied conventionalheattreatmentandshallowanddeepcryogenic processingtoAISID2coldworktoolsteel.Themicro hard-nessinthesamplesubjectedtodeepcryogenictreatmentwas 11.4%higherthanthatoftheconventionallytreatedsample. In thesestudies inthe literature,the increase inhardness aftercryogenicprocessingisrelatedtothetransformationof ofaustenite,whichisthesoftphaseofthematerialstructure, tothehardmartensitephase.

4. Prediction

of

surface

roughness

with

ANN

Inthisstudy,acomputerprogramwasenhancedinaMATLAB platformtopredictthesurfaceroughness.Thecuttingtool, workpiece,cuttingspeed,depthofcut,andfeedratewereused intheinputlayeroftheANN,whilethesurfaceroughnesswas usedintheoutputlayer.Inordertoobtainaccurateresults,a singlehiddenlayerwith14neuronswasused.Table4shows the statisticalevaluationofthe resultsofthe ANNwith14 neurons.

ThematchingoftheexperimentalandANNvaluesforthe training and the testing sets ofthe surfaceroughness are demonstratedinFigs.10and11,respectively.Themost dra-maticpointhereisthattheestimationvaluesareclosetothe experimentalvalues,demonstratingtheestimativeabilityof thenetworkforsurfaceroughnesstobesatisfactory. Accord-ingly,thefiveinputparametersselectedasaffectingfactorsfor predictionofsurfaceroughnessprovidedacceptableresults.

Thepredictionperformanceforboththetestingand train-ingsetsofthesurfaceroughnessshowsthattheaccuracyof theLMlearningalgorithmwassatisfactory(±5%).Withthe ANNmodeldevelopedforthepredictionofthesurface rough-ness,theRMSEvalueswerecalculatedas0.0783and0.0692for trainingandtestingdata,respectively.TheR2_values_for

sur-faceroughnesswerefoundtobe0.9704and0.9755fortraining andtestingdata,respectively.Thesurfaceroughnesscanbe accuratelycalculatedbytheformulagiveninEq.(5).

Surfaceroughness =

1 1+e−(−0.9334xF1+1.3933xF2−0.9928xF3+0.0687xF4−2.6078xF5−1.2896xF6−0.2625xF7−0.8741xF8−1.2553xF9+1.0164xF10−0.0424xF11+1.4785xF12+0.8252xF13+0.2526xF14+0.7824)

(5)

whereFi(i=1,2,....,14)canbecalculatedaccordingtoEq.(6).

Fi=

1 1+e−Ei

(12)

Pleasecitethisarticleinpressas:KaraF,etal.Effectofmachinability,microstructureandhardnessofdeepcryogenictreatmentinhardturning Fig.10–MatchingoftheexperimentalandANNvaluesforsurfaceroughnesstrainingsets.

Fig.11–MatchingoftheexperimentalandANNvaluesforsurfaceroughnesstestingsets.

Table5–Weightvaluesforsurfaceroughnessbetween theinputandhiddenlayers.

Ei=w1x(ct)+w2x(wp)+w3x(V)+w4x(dp)+w5x(f)+i i w1 w2 w3 w4 w5 ␪i 1 3.3636 −0.1659 −0.5580 2.4966 2.4454 −46.603 2 −0.5866 2.2173 0.6470 3.0316 1.3666 43.013 3 −2.7152 2.2793 −2.8487 0.6167 1.9341 37.166 4 1.5334 2.5499 −0.0755 2.6093 2.6764 −28.913 5 0.3424 −0.7709 1.1128 −2.3190 −0.8116 0.1756 6 −2.4733 −1.3852 2.3752 −2.4043 −2.7277 31.657 7 −1.9033 −2.3130 −3.2594 −1.1095 −1.1591 23.549 8 2.4317 3.5886 −0.9547 −1.3038 −2.0127 10.696 9 −0.5128 2.4649 2.2798 −2.5821 0.1085 0.9121 10 −0.2231 0.5231 −1.5641 2.1990 3.0049 −30.862 11 0.7752 1.1580 0.3817 3.1105 −4.4366 14.286 12 0.4046 −1.6768 −2.1088 2.9420 −1.8323 16.425 13 1.2932 1.9272 5.0000 0.5508 −0.5027 −41.643 14 −3.5660 −0.2142 1.9452 −1.8498 0.2100 −48.997

EiiscalculatedviatheequationinTable5.Theweightvalues fortheinputandhiddenlayersarealsogiveninTable5.

5. Conclusions

Thisstudyinvestigatedtheeffectsofcuttingparameterson thesurfaceroughnessandtoolwearofAISID2coldworktool steelwithdifferentheattreatmentsunderdry cutting con-ditionsusinguncoated(AB30)andcoated(AB2010) ceramic cuttingtools.TheDCTT-36andDCT-36treatmentsappliedto coldwork toolsteelwereevaluated.Finally, anANNmodel

was used forthe prediction ofthe surfaceroughness. The back-propagationalgorithmwas usedfortraining theANN developed for determining ofsurface roughness. Different algorithmsincludingBFGS,CGP,LM,RPandSCGwereusedfor thetrainingperiod.Thedataobtainedasaresultofthe exper-imentalandanalyticalstudiesconductedaregivenbelow. • Inhard turningexperiments,the AB2010coated ceramic

cutting tool performed better than the AB30 uncoated ceramiccuttingtoolintermsofsurfaceroughness. • ThelowestRavaluefortheuncoatedceramictoolwasfound

tobe0.2267␮mfortheDCTT-36sampleatacuttingspeed of100m/min,afeedrateof0.08mm/revandacuttingdepth of0.25mm.

• ThelowestRavalueforthecoatedceramictoolwasfound tobe0.18␮mfortheDCTT-36sampleatacuttingspeedof 100m/min,afeedrateof0.08mm/revandacuttingdepth of0.25mm.

• When the workpieces subjected to different heat treat-mentswereevaluatedintheturningexperiments,betterRa valuesweregenerallyobtainedwiththeDCTT-36sample. Takingallcuttingparametersandcuttingtoolsinto consid-eration,theDCT-36andDCTT-36specimensprovidedbetter surfaceroughness,averaging7.56%and10%,respectively, thantheconventionallyheat-treatedCHTsample. • Theperformanceofthe AB2010coatedceramictoolwas

betterinallofthetoolwearexperiments.

• Attheendofthetotalprocessingtime(10min),thenose wearoftheAB30uncoatedceramictoolwasmeasuredas 0.285mm,0.312mmand0.2045mmfortheCHT,DCT-36and DCTT-36samples,respectively.

(13)

• Similarly,thenosewearoftheAB2010coatedceramictool attheendofthe10-mintreatmentperiodwasfoundtobe 0.132mm,0.116mmand0.1195mmfortheCHT,DCT-36and DCTT-36samples,respectively.Ascanbeseenfromthenose wearresults,underallcuttingconditions,thelowestwear valueswereobtainedwiththeDCTT-36sample.

• Whentheaveragewearvaluesofallthesampleswere mea-suredattheendofthetotalmachiningtime,theAB2010tool exhibitedbetterwearperformance,with54%lesswearthan theAB30ceramictool.

• Taking all the cutting parameters and cuttingtools into consideration,theDCT-36andDCTT-36samplesexhibited betternosewearatanaveragerateof5.90%and21.79%than theconventionallyheat-treatedCHTsample.

• Attheendofthetotalprocessingtime(10min),thecrater wearvaluesoftheAB30uncoatedceramictoolwere mea-sured as 0.0684mm2_, _0.054_mm2 _and _0.0432_mm2 _for_the

CHT,DCT-36andDCTT-36samples,respectively.

• Similarly,attheendofthe10-minprocessingperiod,the craterwearvaluesoftheAB2010coatedceramictoolwere found tobe0.0379mm2_, _0.0332_mm2 _and _0.0298_mm2 _for

theCHT,DCT-36andDCTT-36samples,respectively.Ascan beseenfromthethecraterwearresults,underallcutting conditions,thelowestwearvalueswereobtainedwiththe DCTT-36sample.

• When theaveragesofthecraterwearvaluesforall sam-plesweretakenattheendofthetotalmachiningtime,the AB2010toolexhibitedabetterwearperformanceof1.4% lessthantheAB30tool.

• Takingallcuttingparametersandcuttingtoolsinto consid-eration,theDCT-36andDCTT-36samplesexhibitedbetter craterwearofonaverage121%and145%lessthanthe con-ventionallyheat-treatedCHTsample.

• The best mechanical properties amongthe CHT, DCT-36 andDCTT-36specimensweredemonstratedbytheDCT-36 sample.Inthemicrohardnessandmacrohardness mea-surements,thehardnessvaluesoftheDCT-36sampleswere higherthanthoseoftheotherheat-treatedsamples. • Amongthethreedifferentheattreatedsamples,the

high-esthardnessvaluewasobtainedwiththeDCT-36sample. These results can be attributed to the deep cryogenic processing which converted the austenite phase, which hasasoftstructure,tothehardmartensitephaseinthe microstructureofthematerial.

• Asaresult,thedeepcryogenicandpost-tempering treat-mentsledtoimprovementof32.97%insurfaceroughness, 21.79% in tool wear, 0.96% in macro hardness and 21% in micro hardness. The AB2010 coated ceramic tool, whichgenerallyproducedbetterresultsthantheuncoated ceramic inserts, led to improved surface roughness and nosewearby25.20%and42.21%,respectively.

• Theoptimalresultsinthepredictionofthesurface rough-nesswereobtainedbyanetworkarchitectureof5-14-1and theLMlearningalgorithm.

• TheANNpredictionperformanceswerecomparedwiththe experimentalresultsusingR2_and_RMSE_values._The_R2

val-uesweremorethan0.97forboththetestingandthetraining data. The RMSEvaluewas less than 0.07 forthe testing data.Theseresultsshowedthatthelearningcapacityofthe

ANNwasrelativelypowerfulintheestimationofthesurface roughness.

Conflict

of

interest

Theauthorsdeclarenoconflictsofinterest.

Acknowledgment

TheauthorswouldliketothankDüzceUniversityScientific ResearchProjectsCoordinatorforsupportingthisstudywith BAPproject2015.07.04.388.

r

e

f

e

r

e

n

c

e

s

[1]KaraF,C¸ic¸ekA,DemirH.MultipleregressionandANN modelsforsurfacequalityofcryogenically-treatedAISI 52100bearingsteel.JBalkTribolAssoc2013;19(4):570–84,

http://dx.doi.org/10.1007/s00521-014-1721-y.

[2]T.Kivak,Investigationoftheinfluenceofcryogenicprocess oncuttingtool’sdrillabilityofTi-6al-4valloy,PhDThesis, DepartmentofMechanicalEducation,GaziUniversity, Ankara,Turkey,2012.

[3]DasD,DuttaAK,RayKK.Influenceofvariedcryotreatment onthewearbehaviorofAISID2steel.Wear2009;266:297–309,

http://dx.doi.org/10.1016/j.wear.2008.07.001.

[4]AminiK,AkhbarizadehA,JavadpourS.Investigatingthe effectofholdingdurationonthemicrostructureof1.2080 toolsteelduringthedeepcryogenicheattreatment.Mater Des2012;86:1534–40,

http://dx.doi.org/10.1016/j.vacuum.2012.02.013.

[5]VahdatSE,NateghS,MırdamadıS.Microstructureand tensilepropertiesof45WCrV7toolsteelafterdeepcryogenic treatment.MaterSciEngA2013;585:444–54,

http://dx.doi.org/10.1016/j.msea.2013.07.057.

[6]F.Kara,Investigationoftheeffectsofcryogenicprocess parametersonfatiguelifeandgrindabilityofAISI52100 steel,PhDThesis,DepartmentofMechanicalEducation, KarabükUniversity,Karabük,Türkiye,2014.

[7]DasD,DuttaAK,RayKK.Optimizationofthedurationof cryogenicprocessingtomaximizewearresistanceofAISID2 steel.Cryogenics2009;49:176–84,

http://dx.doi.org/10.1016/j.cryogenics.2009.01.002.

[8]AminiK,NateghS,ShafyeiA.Influenceofdifferent cryotreatmentsontribologicalbehaviorof80CrMo125cold worktoolsteel.MaterDes2010;31:4666–75,

http://dx.doi.org/10.1016/j.matdes.2010.05.028.

[9]OzelT,KarpatY,FigueiraL,DavimJP.Modellingofsurface finishandtoolflankwearinturningofAISID2steelwith ceramicwiperinserts.JMaterProcessTechnol

2007;189:192–8,

http://dx.doi.org/10.1016/j.jmatprotec.2007.01.021.

[10]DavimJP,FigueiraL.Machinabilityevaluationinhard turningofcoldworktoolsteel(D2)withceramictoolsusing statisticaltechniques.MaterDes2007;28:1186–91,

[11]OzelT,KarpatY.Predictivemodelingofsurfaceroughness andtoolwearinhardturningusingregressionandneural Networks.IntJAdvManufTechnol2005;4:467–79,

http://dx.doi.org/10.1016/j.ijmachtools.2004.09.007.

[12]QuizaR,FigueiraL,DavimJP.Comparingstatisticalmodels andartificialnetworksonpredictingthetoolwearinhard machiningD2AISIsteel.IntJAdvManufTechnol

(14)

[13]KumarR,SahooAK,DasRK,PandaA,MishraPC.Modelling offlankwear,surfaceroughnessandcuttingtemperaturein sustainablehardturningofAISID2Steel.ProcediaManuf 2018;20:406–13,

http://dx.doi.org/10.1016/j.promfg.2018.02.059.

[14]KaraF.Taguchioptimizationofsurfaceroughnessandflank wearduringtheturningofDIN1.2344toolsteel.MaterTest 2017;59(10):903–8,http://dx.doi.org/10.3139/120.111085.

[15]HaykinS.Neuralnetworkacomprehensivefoundation.New

Jersey:PrenticeHall;1999.

[16]KaraF,Aslantas¸K,C¸ic¸ekA.Predictionofcuttingtemperature inorthogonalmachiningofAISI316Lusingartificialneural network.ApplSoftComput2016;38:64–74,

http://dx.doi.org/10.1016/j.asoc.2015.09.034.

[17]Arnaiz-GonzalezA,Fernandez-ValdivielsoA,BustilloA,de LacalleLNL.Usingartificialneuralnetworksforthe predictionofdimensionalerroroninclinedsurfaces manufacturedbyball-endmilling.IntJAdvManufTechnol 2016;83(5-8):847–59,

http://dx.doi.org/10.1007/s00170-015-7543-y.

[18]ÇayY,ÇiçekA,KaraF.Predictionofengineperformancefor analternativefuelusingartificialneuralnetwork.Appl ThermEng2012;37:217–25,

http://dx.doi.org/10.1016/j.applthermaleng.2011.11.019.

[19]ÖzgörenYÖ,C¸etinkayaS,SarıdemirS.Artificialneural networkbasedmodelingofperformanceofabeta-type Stirlingengine.PIMechEngE-JPro2012;227(3):166–77,

http://dx.doi.org/10.1177/0954408912455763.

[20]C¸iftci ˙I.MachiningofausteniticstainlesssteelsusingCVD multilayercoatedcementedcarbidetools.TribolInt 2006;39:565–9,

http://dx.doi.org/10.1016/j.triboint.2005.05.005.

[21]EscalonaPM,CassierZ.Influenceofthecriticalcuttingspeed onthesurfacefinishofturnedsteel.Wear1998;218:103–9,

http://dx.doi.org/10.1016/S0043-1648(98)00156-2.

[22]ThamizhmaniiS,KamarudinK,RahimEA,SaparudinA,

HassanS.ToolwearandsurfaceroughnessinturningAISI

8620usingcoatedceramictool.In:ProceedingsoftheWorld

CongressonEngineering.2007.

[23]H.Gürbüz,Investigationofcuttingtoolgeometryand coatingtypesonsurfaceintegrityofAISI316lsteel,PhD Thesis,DepartmentofMechanicalEducation,Gazi University,Ankara,Türkiye,2012.

[24]FernándezLandetaJ,FernándezValdivielsoA,Lópezde LacalleLN,GirotF,PérezPérezJM.Wearofformtapsin threadingofsteelcoldforgedparts.JManufSciEng 2015;137(3):1–11,http://dx.doi.org/10.1115/1.4029652. [25]GandariasA,LópezdeLacalleLN,AizpitarteX,LamikizA.

Studyoftheperformanceoftheturninganddrillingof austeniticstainlesssteelsusingtwocoolanttechniques.IntJ MachMachMater2008;3(1/2):1–17,

http://dx.doi.org/10.1504/IJMMM.2008.017621.

[26]Kac¸alA.Investigationofcuttingperformanceoftheceramic insertsintermsofthesurfaceroughnessandtoolwearat turningofPMD23steel.ApplMechMater2014;686:10–6,

http://dx.doi.org/10.4028/www.scientific.net/AMM.686.10.

[27]ÖzbekNA,C¸ic¸ekA,GülesinM,ÖzbekO.Evaluationofthe machinabilityofAISI304andAISI316austeniticstainless steels.JPolytech2017;20(1):43–9,

http://dx.doi.org/10.31202/ecjse.67151.

[28]C¸aydas¸U,KuncanO,C¸elikM.Investigationofthe machinabilityofAISI52100bearingsteelforsurface roughness,toollifeandtemperaturecriterions.JPolytech 2017;20(2):409–17,http://dx.doi.org/10.2339/2017.20.2. [29]TangL,GaoC,HuangJ,ShenH,LinX.Experimental

investigationofsurfaceintegrityinfinishdryhardturningof hardenedtoolsteelatdifferenthardnesslevels.IntJAdv

ManufTechnol2015;77:1655–69,

http://dx.doi.org/10.1007/s00170-014-6484-1.

[30]KarabatakM,KaraF.Experimentaloptimizationofsurface roughnessinhardturningofAISID2coldworktoolsteel.J Polytech2016;19(3):349–55,

http://dx.doi.org/10.2339/2016.19.3.

[31]RechJ,KusıakAJ,BattaglıaL.Tribologicalandthermal functionsofcuttingtoolcoatings.SurfCoatTech 2004;186(3):364–71,

http://dx.doi.org/10.1016/j.surfcoat.2003.11.027.

[32]SalımıaslA,RafıghıM.Estimationandestimationof vibrationandshearforcebasedtoolwearwithfuzzylogic.J Polytech2017;20(1):111–20,

http://dx.doi.org/10.2339/2017.20.1.

[33]ÖzbekNA.Investigationoftheeffectsofcryogenic

machiningwithcuttinginsertontoollifeinprocessingof

AISI316austeniticstainlesssteels,PhDThesis.Ankara,

Türkiye:DepartmentofMechanicalEducation,Gazi

University;2013.

[34]LimaJG,AvillaRF,AbroaAM,FaustinoM,DawimJP.Hard turning:AISI4340highstrengthlowalloysteelandAISID2 coldworktoolsteel.JMaterProcessTechnol2005;169:388–95,

[35]JeongYK,KangMC,KwonSH,KimKH,KimHG,KimJS.Tool lifeofnanocompositeTi–Al–Si–Ncoatedend-millbyhybrid coatingsysteminhighspeedmachiningofhardenedAISID2 steel.CurrApplPhys2009;9:141–4,

http://dx.doi.org/10.1016/j.cap.2008.08.053.

[36]ArsecularatneJA,ZhangLG,MontrossC,MathewP.On machiningofhardenedAISID2steelwithPCBNtools.J MaterProcessTechnol2006;171:244–52,

[37]PoulachonaG,BandyopadhyaybBP,JawahircIS,PheulpinaS, SeguinaE.WearbehaviorofCBNtoolswhileturningvarious hardenedsteels.Wear2004;256:302–10,

http://dx.doi.org/10.1016/S0043-1648(03)00414-9.

[38]DosbaevaGK,ElHakimbMA,ShalabyMA,KrzanowskiJE, VeldhuisSC.CuttingtemperatureeffectonPCBNandCVD coatedcarbidetoolsinhardturningofD2toolsteel.IntJ RefractMetalsHardMater2015;50:1–8,

http://dx.doi.org/10.1016/j.ijrmhm.2014.11.001.

[39]MaviA.Theeffectofcryogenicprocessonthecuttingtool

performanceatthemachiningofTi6al4vtitaniumalloy,PhD

Thesis.Ankara,Türkiye:DepartmentofMechanical

Education,GaziUniversity;2014.

[40]C¸ic¸ekA,KaraF,KıvakT,EkiciE.Evaluationofmachinability ofhardenedandcryo-treatedAISIH13hotworktoolsteel withceramicinserts.IntJRefractMetalsHardMater 2013;41:461–9,

http://dx.doi.org/10.1016/j.ijrmhm.2013.06.004.

[41]ÖzdemirU,ErtenM.Failuremechanismsoccurringinthe

cuttingtoolandtooldamagethemachinedduring

abatement.JAeronautSpaceTechnol2003;1:37–50.

[42]DasD,DuttaAK,ToppoV,RayKK.TheEffectofcryogenic treatmentonthecarbideprecipitationandtribological behaviorofD2steel.MaterManufProcess2007;22:474–80,

http://dx.doi.org/10.1080/10426910701235934.

[43]BaldisseraP,DelpreteC.Deepcryogenictreatment:a bibliographicreview.OpenMechEngJ2008;2:1–11,

http://dx.doi.org/10.2174/1874155X00802010001.

[44]AkhbarızadehA,JavadpourS,AminiK.Investigatingthe effectofelectriccurrentflowonthewearbehaviorof1.2080 toolsteelduringthedeepcryogenicheattreatment.Mater Des2013;45:103–9,

[45]PrecıadoM,BravoPM,AlegreJM.Effectoflowtemperature temperingpriorcryogenictreatmentoncarburizedsteels.J

(15)

MaterProcessTechnol2006;176:41–4,

[46]YiJ,XueW,XieJ,LiuZP,ChengW,ChenLX,etal.Enhanced toughnessandhardnessatcryogenictemperaturesofsilicon carbidesinteredbySPS.MaterSciEngA2013;569:13–7,

[47]SonawaneSA,TripathiVK,AmbekarSD.Wearbehaviorof cryogenictreatedM2toolsteelunderdryslidingcondition. ApplMechMater2015;798:395–401,

http://dx.doi.org/10.4028/www.scientific.net/AMM.798.395.

[48]NanesaHG.Influenceofcryogenictreatmenton

microstructureandmechanicalpropertiesofhighstrength

AISID2toolsteel,Ph.D.Thesis.Canada:UniversitéDu

Québec;2016.

[49]DasD,DuttaAK,RayKK.Sub-zerotreatmentsofAISID2 steel:partII.wearbehavior.MaterSciEngA

2010;527:2194–206,

[50]DasD,RayKK.Structure–propertycorrelationofsub-zero treatedAISID2steel.MaterSciEngA2012;541:45–60,

[51]C¸ic¸ekA,KaraF,KıvakT,EkiciE,UygurI.Effectsofdeep cryogenictreatmentonthewearresistanceandmechanical propertiesofAISIH13hot-worktoolsteel.JMaterEng Perform2015;24(11):4431–9,

http://dx.doi.org/10.1007/s11665-015-1712-x.

[52]DasD,DuttaAK,RayKK.Sub-zerotreatmentsofAISID2 steel:partI.Microstructureandhardness.MatSciEng 2010;527:2182–93,

[53]OppenkowskiA,WeberS,TheisenW.Evaluationoffactors influencingdeepcryogenictreatmentthataffectthe propertiesoftoolsteels.JMaterProcessTechnol 2010;210:1949–55,

[54]AminiK,NegahbaniM,GhayourH.Theeffectofdeep cryogenictreatmentonhardnessandwearbehaviorofthe H13toolsteel.MetallItal2015;107(3):53–8,

http://dx.doi.org/10.1080/10426910701235934.

[55]HarishS,BenselyA,LalM,RajaduraıDA,LenkeyGB. Microstructuralstudyofcryogenicallytreateden31bearing steel.JMaterProcessTechnol2009;209:3351–7,

Effect of machinability, microstructure and hardness of deep cryogenic treatment in hard turning of AISI D2 steel with ceramic cutting

Original

Article

Effect

of

machinability,

microstructure

and

hardness

of

deep

cryogenic

treatment

in

hard

turning

of

AISI

D2

steel

with

ceramic

cutting

Fuat

Kara

,

Mustafa

Karabatak

,

Mustafa

Ayyıldız

,

Engin

Nasc

a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

1.

Introduction

2.

Material

and

methods

⎛

⎝









⎞

⎠



3.

Experimental

results

4.

Prediction

of

surface

roughness

with

ANN





_,

_Mustafa

_Karabatak

_,

_Mustafa

_Ayyıldız

_,

_Engin

_Nasc