hybrid composites processed through compocasting Investigations on mechanical and machinability behavior of aluminum/ﬂyashcenosphere/Gr Technology Journal of Applied Researchand

(1)

Availableonlineatwww.sciencedirect.com

Journal of Applied Research and Technology

www.jart.ccadet.unam.mx JournalofAppliedResearchandTechnology15(2017)430–441

Original

Investigations on mechanical and machinability behavior of aluminum/flyash cenosphere/Gr hybrid composites processed through compocasting

Soorya Prakash Kumarasamy

^a,∗

, Kavimani Vijayananth

^b

, Titus Thankachan

^b

, Gopal Pudhupalayam Muthukutti

^b

aAssistantprofessor,DepartmentofMechanicalEngineering,AnnaUniversityRegionalCampus,Coimbatore641046,TamilNadu,India

bResearchscholar,DepartmentofMechanicalEngineering,AnnaUniversityRegionalCampus,Coimbatore641046,TamilNadu,India Received19September2016;accepted4May2017

Availableonline17October2017

Abstract

ThisworkfocusesoncharacterizationofanovelhybridAluminumMetalMatrixComposite(AMMC)developedthroughtwo-stepcompocasting methodbyreinforcingconstantamountofflyashcenosphere(10%)andvaryingquantityofgraphite(2%,4%and6%).Morphologicalanalysis resultshowsthepresenceofdendriticarmsandhomogeneousdistributionofthecenosphereandgraphiteinAl7075matrix.Thehardnessand tensilestrengthupturnwithadditionofcenosphereandviceversafortheadditionofgraphite.Thereinforcementparticleshaveenhancedthe tensilestrengthofAluminummatrixfrom178N/mm² to213N/mm² asofthecompositeisconcerned.Additionofcenosphereimprovesthe wearresistanceconsiderablyandwearratedecreasesfurtherwithgraphiteadditiondue toitsself-lubricatingnature.Machinability(turning) characteristicsofthedevelopedhybridcompositewerestudiedindetailbesidesoptimizingthemachiningparametersemployingArtificialNeural Network(ANN)technique.BasedontheANOVAresultsitwasfoundthatcuttingspeedand%ofgraphiteadditionhavethemajorcontribution inminimizingthesurfaceroughnessofdevelopedcomposite.

Keywords: Compocasting;Cenosphere;Graphite;Wear;Machining;ANN

1. Introduction

Composites are the most successful and extensively used materialsinmodernindustriescomparedwithalloysor some other metals as these materials hold significantly improved mechanicalproperties.Tailoring themechanical,physicaland thermalpropertiesthatconsistoflowerdensity,higherspecific strength, higher specificmodulus, greater thermalconductiv- ityandbetterabrasionandwearresistancegivesadvantagefor metalmatrixcompositesintheirperformance.

Aluminum(Al)matrixreinforcedwithceramicparticleshas provedtobesuperiorinitsmechanicalcharacterizationwhen comparedwithitsbasealloymaterials(Bodunrin,Alaneme,&

Chown,2015) andis mostwidely used for tribological parts

∗Correspondingauthor.

E-mailaddress:k[email protected](S.P.Kumarasamy).

PeerReviewundertheresponsibilityofUniversidadNacionalAutónomade México.

due to their high ratio of strength to weight, stroke density and improved wear resistance (Pramanik, 2016).Particles or whiskersofSiliconCarbide(SiC),alumina(Al2O3)andgraphite arethewidelyusedreinforcementmaterialsforthesetypesof composites (Baradeswaran&Perumal,2014;Mosleh-Shirazi, Akhlaghi, &Li,2016).Variouscombinationshavebeen tried outbyresearchersforhybridcompositesystemswithreinforce- ments such as SiC, Al2O3, rice husk ash, bamboo leaf ash, groundnut shell ash,graphite,etc.(Dinaharan, Nelson,Vijay,

&Akinlabi,2016).Theperformancelevelsderivedfromhybrid compositesystemshavebeenidentifiedtobedependentonthe type of reinforcementsused,weight ratio of bothreinforcing materialsincomposite,wettabilitybetweenreinforcementsand matrix, processing method and the metallurgicalcharacteris- tics ofthe matrixmaterial (Alaneme,Adewale, &Olubambi, 2014).

Among the various discontinuous reinforcements used, cenosphere a waste product during combustion of coal in thermalpowerplantsresemblinginlargequantities.Inaddition

http://dx.doi.org/10.1016/j.jart.2017.05.005

1665-6423/©2017UniversidadNacionalAutónomadeMéxico,CentrodeCienciasAplicadasyDesarrolloTecnológico.Thisisanopenaccessarticleunderthe CCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

(2)

to lowering the cost of production, incorporation of flyash cenosphereintoaluminumhasprovedtoincreasethehardness, wearresistanceandstiffnesswithanadmissibledecreaseinthe materialdensity.

Graphite (Gr), one of the most widely used solid lubri- cantmaterials,hasenchantedmanyresearchers tofocus their investigation on development and further identify numerous applicationsof Al/Grcomposite (Rohatgi, Ray,&Liu, 1992, Vogiatzis,Tsouknidas,Kountouras,&Skolianos,2015).Also, researchershaverecognizedthatAl/Grcompositeexhibitssupe- riorwearpropertiesoverthebasealloysduetothepresenceof Grparticle(Krishnan&Rohatgi,1984;Yolshina,Muradymov, Korsun,Yakovlev,&Smirnov,2016),whichservesas asolid lubricating layer between the composite andcounter surface therebyreducingthecompositewearwithouttheneedforsolid andliquidlubrication(Ravindran,Manisekar, Narayanasamy,

&Narayanasamy,2013).

FabricationofAMMCisusuallydonebyliquidcastingtech- nique or powder metallurgy route (Harrigan, 1998; Prakash, Moorthy,Gopal,&Kavimani, 2016)forthereasonthat these methods are typically cost effective (Kerti & Toptan, 2008).

In the liquidcasting technique,the particulates are mechani- callywelldistributedovertheliquidmetal beforecastingand solidification(Moses, Dinaharan, &Sekhar,2016). Stircast- ingisthecheapestandsimplestmethodforcompositematerial fabricationbutcompositematerialsproducedthroughstircast- ingmethodsuffersfrompoorincorporationanddistributionof reinforcement particles in the matrix. This is especially due tothe reinforcement size, whichin turn resultsin exhibiting agglomeration tendency and reduced wettability of particles within the melt. In comparison with composites developed through stir casting method, composites produced by com- pocastingobtainuniformdistributionofreinforcementparticles insolidified matrix and also showcases better hardness with decreased porosity(Gladston, Sheriff, Dinaharan, &Selvam, 2015).

Al7075havegoodfatiguestrengthandaveragemachinabil- ity,buthasless resistancetocorrosion.Duetoitsremarkable highstrength-to-weight ratio,thismaterialfindsawiderange of applications including aircraft fittings, worm gears, keys andvariousothervitalcomponentsofcommercialaircraftsand aerospacevehicles.ItispossibletofabricateMMCbydiffer- enttechniques butmachiningplaysamajorrole inindustrial componentdesignslikethatoftoleranceanddimensions;like- wisetheimpropertoolselectionandmachiningparameterleads topoormaterialqualityandthusincreasetheproductioncost.

Whatsoeversurfaceroughnessalsoplaysavitalroleinmaterial qualitycontrol.Depthof cut,cuttingfeedandspindle speed, reinforcementadditions aresupposed tobethe major factors influencingthe surfaceroughnessof the machinedsurface of MMCandhenceitismandatorythatallofthesecuttingparam- etermustbeoptimizedsoastoobtainmaterialwithbettersurface finish.

EmployingoptimizationtoolslikeArtificialNeuralNetwork (ANN),ParticleSwarmOptimization(PSO),GeneticAlgorithm (GA) and fuzzy logic helps in attaining improved output response (Shabani & Mazahery, 2012; Shabani, Rahimipour,

Tofigh, & Davami, 2015; Tofigh & Shabani, 2013). Among these,Taguchiapproachisusedenormouslyintherecentpast for solvingtypical problemsencounteredinthe fieldof optimization (Kumar & Chauhan, 2015; Shanmughasundaram&

Subramanian,2013).Therearefewworksinwhichcomputa- tional methodsare used tooptimize andpredict the material propertiesofthenewlydevelopedmaterials.MazaheryandSha- bani(2013)adoptedstircastingtechniquetofabricateAMMC withvaryingprocessparameters(pouringtemperature,stirring speed,stirringtime,pre-heattemperatureofreinforcement)and different percentageof SiC (5, 10 and15) andresults show- cased that addition of SiC has improved the hardness. GA coupledPSOtechniqueisutilizedtooptimizethewearparam- eters andtostudythe influenceof wearparameterover wear rate;thisadopted algorithmgenerallycoincideswellwiththe experimentalvalues.Rahimipour,Tofigh, Mazahery,andSha- bani(2014)utilizedANNcoupledPSOtechniquetostudythe effect of reinforcement upon wear rate and porosity of fabricated Al–Cu based composite wherein boron carbide with different percentages are used as the reinforcements. Results statedthatdevelopedmodeladhereswellwiththeirexperimental values.

ShabaniandMazahery(2013)developedAMMCusingstir castingmethod.ANNtechniqueisutilizedtopredicttheeffectof processparameter(temperature,coolingrateandreinforcement vol.%) onmechanical properties of the fabricatedAMMC; a betterpredictablywasobservedbythedevelopedmodel.Tofigh, Rahimipour,Shabani, andDavami (2015)fabricated alumina nanoparticlereinforcedAMMCthroughcompocastingmethod and also employed ANFIS coupled PSO method to obtain a optimal parameter inorder todevelop AMMC with uniform dispersionofreinforcement.

Itisthereforeproposedtoundertakeastudyonthepossibil- ityofenhancingthemechanicalpropertiesandmachinabilityof Al7075alloyforwormgearsproductioninaerospaceindustries.

Thisisdonebycombiningreinforcements(flyashcenosphere andGr)mixedwithAl7075bytwo-stepcompocastingmethod.

Al7075 reinforced with fixed quantity of flyash cenosphere (10wt.%)andvaryingGr(2wt.%,4wt.%,6wt.%)istriedout inthisresearchtoformastrongerhybridcomposite.Moreover itisalreadyreportedthatincreaseinpercentageofGrbeyond 7% reduces the mechanical properties of composite material (Swamy, Ramesha, Kumar, &Prakash, 2011)and hencethis researchaimstostudytheeffectofGradditiononmechanical propertiesbyvaryingtheweightpercentageofGr.Likewisefly- ashcenospherepercentageiskeptconstantasitisreportedthat 10% additionof flyashcenosphere exhibit better mechanical properties (Escalera-Lozano,Gutiérrez, Pech-Canul, & Pech- Canul,2007).Itisexpeditedthatthisnoveldevelopmentwhen putinto practicalityarelikelytoovercomethecostbarrieras wellasprovideenhancedphysicalandmechanicalpropertiesas pertheproductrequirement.Apositiveoutcomeofthisresearch wouldhelpincommercializationofthealloyforwiderangeof applicationsinmodernindustriestoo.Alsoanattempthasbeen made tominimize the surface roughnessbythe way of opti- mizingthemachiningparametersthroughTaguchiandANOVA techniquesforturningoperationofthedevelopedcomposite.It

(3)

isaknownfact that introductionofartificialintelligenceinto the field of materials has made a revolution in alloy development,process optimization, etc. This realizationhas made us toopt ANN inthisresearch for predictingsurface roughness of the developed composites based on varying process parameters

2. Materialsandmethod 2.1. Materials

Commercially available Al–Zn–Mg alloy (Al7075) was selectedasthematrixforcompositeproductionanditschemi- calcompositionispresentedinTable1.Al7075isknownforits excellentcombinationofstrengthanddamagetoleranceatele- vatedtemperatures.Withtheseformulation,thecurrentstudyis usedtoimprovisethecorrosionresistantandwearresistanceof thematerialbyintroducingsuitablereinforcementsviz.flyash cenosphere andGrandthereby enrich Al7075alloy.Besides confirmingtheuniformityofparticledispersionsthroughSEM imagesofthesample,theaveragesizeoftheflyashcenosphere particleswasalsoidentifiedtobeintherangeof30–60␮m.The chemicalcompositionoftheflyashcenosphereispresentedin Table2.

2.2. Compositespreparation

The proposed novelcomposite was fabricated atfour different compositions as listed in Table 3, where the amount of cenosphere is keptto be constant at10% andGr propor- tionisvariedfrom 0to6%.Theprospectivecompositionsof the novelcomposite are listed in Table3.In addition tothat 1.5wt.%ofmagnesiumwasusedasawettabilityagent.Produc- tionofthecompositespecimenswascarriedoutusingatwo-step compocastingprocesswhoseexperimentalsetupisasshownin

Table1

ChemicalcompositionofAl7075.

Element Zn Mg Cr Cu Fe Mn Si Ti Al

Wt.% 5.5 2.4 0.18 1.5 0.4 0.2 0.4 0.2 Remaining

Table2

Chemicalcompositionofcenosphere.

Element SiO2 Al2O3 Fe3O2 TiO2 CaO MgO K2O Na2O H2O

Wt.% 53 30 6 – 1 0.5 – – 0.2

Table3

Compositionofsamples.

Sample Al7075 Flyashcenosphere Graphite

A 100% 0% 0%

AC0Gr 90% 10% 0%

AC2Gr 88% 10% 2%

AC4Gr 86% 10% 4%

AC6Gr 84% 10% 6%

A,Al7075;C,cenosphere;Gr,graphite.

Fig.1.Compocastingsetup.

Figure1.Al7075rodsweremeltedatatemperatureof690^◦C.

Bytheuseofamechanicalstirrer,moltenalloywasagitatedto formafinevortexandwasthenpouredintopermanentmoldto fabricatethefirstsampleofpureAl7075.Preheatedflyashceno- sphereparticles(atatemperatureof500^◦Cfor60min)werethen addedataconstantfeedrateintothemoltenAl7075inorderto form thesecond sample.InadditiontothatGrparticleswere addedat2,4and6%leveltoformalloftheothercompositions.

Toensureuniformdispersionofreinforcementparticlesinto the matrixalloy,two-stagestirringwasemployed.Inthe first stage,stirringwascarriedoutfor10minwhentheslurrywasin asemi-solidconditionat615^◦Candthesecondstagewascar- riedoutwhentheslurrywasre-meltedtoatemperatureabove liquidus temperature of approximately 700^◦C.The mechani- calstirringprocesswasperformedat700rpmfor10minandit wascontinuedtillthecompositewaspouredintothepermanent mold.

3. Resultsanddiscussion 3.1. Materialdepictions

3.1.1. Microstructuralcharacterization

The scanning electron micrographs were recorded using Hitachi-S3400scanningelectronmicroscope(SEM).Seriesof grindingandpolishingstepswereperformedonthesamplessur- facetoachieveamirrorlikesurfacefinishformicrostructural examination. The samples were polishedwithabrasive paper ofgrade600,grade1000,andgrade1200followedbyetching usingKeller’sreagent(95mlwater,2.5mlHNO3,1.5mlHCl, 1.0mlHF)andswabbingfor10–20sbeforethetest.

ItisobservedfromFigure2(a–d)thatthereinforcedflyash cenosphere particles are fairly welldistributed inthe matrix.

Presence of Gr in the matrix is clear and the same is visu- allyevidentfromFigure2(b–d).FromFigure2(a)morphology of cenosphereincorporatedmatrixisvisualizedusingthermo- ionicemissionofSEM.Micrographstakenbyusingsecondary electronmodeshowcenosphereinsidetheAlmatrix.Uniform distribution of cenosphere inside the Al matrix is confirmed frommicrograph.Figure2(b)showsthedifferentmorphological

(4)

Fig.2.SEMmicrographsof(a)Al7075/10wt.%flyashcenosphere,(b)Al7075/10wt.%flyashcenosphere/2wt.%graphite,(c)Al7075/10wt.%flyashcenosphere/4wt.%graphiteand(d)Al7075/10wt.%flyashcenosphere/6wt.%graphite.

structureduetotheimpregnationofdifferentweightpercentage ofGralongwiththecenosphereandaluminummatrix.

From Figure 2(c) micrograph,it can be observed that the cenosphere particlesare seeningrayphase andthe Grpres- ence is seen as dark grayphase. Further it is clearlyvisible andcouldbeobservedthatcenosphereparticlesaresurrounded byGr.Figure2(d)illustrates thatsurface ofthe samplesgets smootherwiththe additionof Gras experimented due toits lubricatingnature.

3.1.2. Density

Density playsa vitalroleinthe composite materialstudy.

SinceAl matrixcomposites havebigger scope inautomotive industryaswellasinthespacecraft,theymustcompulsorilybe lightinweight.Forthatfact,densityisusuallyreducedbyaddi- tionofreinforcementmaterialsofverylowdensityandinline, fromthecompositesamplespreparedformanyresponsiveinves- tigations, it was observed that density decreases with such lightweightreinforcementadditions.Similarly,hereindecrease indensityofhybridcompositeswasprimarilyduetothepres- enceoflowdensityflyashcenosphereparticulatesanditcould benotifiedfromFigure3thatthedensityofcompositedecreases from2.615g/cm³to2.537g/cm³approximately.

3.1.3. Hardnesstest

Hardnesstestscarried outaccordingtothe ASTME10-08 standardsusingRockwellhardnesstesterofModelAIRAB250 with1/16ballindenterand100kgloadfor30swasconducted atroomtemperatureandthemeasurementofhardnesswasnoted ontheBscale.Testswerecarriedoutatsixdifferentplaceson

A AC AC2G AC4G AC6G

Composition

2.62

2.60

2.58

2.54

Density g/cm3

Density g/cm³ 2.56

Fig.3.DensityofthedevelopedAMMC.

each specimen and averagevalue of hardness is notifiedand impliedforanyfurtherusageinordertoobtainpreciseresults.

IntheanalysisofAl7075/flyashcenosphere/Grhybridcom- posite,itwasobservedthathardnessofthedevelopedcomposite is significantly greater than that of the base alloy character- ized dueto the presence of cenosphere particles. The reason forincreasedhardnessuponadditionof10%flyashcenosphere toAl7075alloyismainlyduetotheincreaseofceramicrein- forcementparticlesinthematrix.Thistendsforconsiderable increaseinthesurfaceareaofreinforcementandalsoreduction inthematrixgrainsizes.Thepresenceofsuchhardsurfacearea ofparticlesoffersmoreresistancetoplasticdeformationwhich leadstoincreaseinthehardnessofcomposites.

Atthesametime,hardnessincreasesfor2%Gradditionand decreaseswithfurtheradditionofGrparticlestoAlMMCdue

(5)

62 61

60

59

58

57

56

A AC AC2G AC4G AC6G

Composition

Hardness (HRB)

Fig.4.RockwellhardnessofthedevelopedAMMC.

toitssoftnature.From Figure4,itisobservablethat thereis ageneraldecreaseinhardnessforanyincreaseinweightratio ofGrincaseofhybridcomposites.Hassan,Tashtoush,andAl- Khalil(2007)alsostatedthatRockwellhardnessdecreasewith increaseinGrcontentandsoitisviabletostatethatthepresent.

Also,investigationisinadherencetotheexistingresearchiden- tifications.Also,hardnessvalueisincloseadherencetotherival materialsthatareinexistenceunderthiscategoryandhenceit couldwellbeaugmentedthatthedevelopedmaterialishighly suitableformanufacturinggearboxcomponents.

3.1.4. Tensiletest

Tensiletestis majorly requiredbecause theproduct under considerationmustwithstandthetensileandcompressivestress duetodynamicloadactinguponthegearmaterial.

Thetestspecimensfortensiletestwerepreparedwithutmost careasperthedimensionsmentionedinFigure5byusingWire CutEDMprocessjustbecauseanyalterationsintensilestrength willobviouslyresultsinmartialfailure.Thetensilestrengthwas evaluatedas per ASTM E08standards using acomputerized universaltestingmachine(TMC-Auto loadmasterseriesIII).

Inordertodecreasethemachiningscratchesandtheeffectsof surfacedefectsonthesample,the1200gridSiCgrindingpaper wasusedtopolishthetestsamples.

Specimensafter tensile test are presented inFigure 6 and itwasobservedthattensilestrengthincreaseswithadditionof flyash cenosphere contentthan that of Al7075base alloy.In general,decreaseintensilestrength isobservedwithincrease inGrcontenti.e.,from2wt.%to6wt.%asshowninFigure7.

ThisfactisduetobrittlenatureofGrparticleswhichreduces

TENSILE TEST SPECIMEN R2.5

10

15

30 2

36

10

2 30

Fig.5.Dimensionoftensiletestspecimen.

Fig.6.Tensiletestspecimensaftertest.

220

200

180

160

140

120

AC AC2G AC4G AC6G

Composition

Tensile strength (MPa)

A

Fig.7.Tensilestrength.

the ductility of the developed MMC and thus initiates the reduction in tensile strength. Load transfer from the matrix to reinforcing particles has significantly reduced due to the presenceof agreateramountofGrandflyashcenosphere.Al matrix withdominant cenosphere particlesexhibits Ultimate TensileStrength(UTS)of213N/mm²butthereafteradditionof 6%graphiteloweredUTSvaluetoapproximately120N/mm² (Baradeswaran&Perumal,2014).

3.1.5. Weartest

ADUCOMTR-20M-106pinondisctestapparatuswasused toinvestigatethedryslidingwearcharacteristicsofthecompos- ites. Thewearspecimenof10mm×10mm×30mmwascut fromcastsamplesandendsofthespecimenswerepolishedwith abrasivepaperofgrade600,followedbygrade1000andfinally polishedusingMETCOGrinding/PolishingMachine.Alsothe slidingendofthepinondiscsurfacewascleanedwithacetone beforetesting.AnOHNS(OilHardenedNickelSteel)55mm diametersteeldiscof60HRCwasusedasthecountersurface fortherealtimeweartest.

Tofixtheparametersforweartest,asoftAlalloywasusedasa samplespecimen.Thetestsdoneatroomtemperatureexhibiting arelativehumidityof60–65%weremadeoverforvariousloads of20N,40Nand60Nataslidingspeedof0.6,0.8and1.0m/s foraslidingdistanceof1000m,2000mand3000m.Theparam- eters werefixed afterconducting thesetrial anderrorexperi- mentsandwhenthesamplereachedamaximumwearof2mm.

(6)

Wear rate

2,000 1,800 1,600 1,400 1,200 1,000 800 600 400 200 0

A AC AC2G AC4G AC6G

Composition

Fig.8.Wearbehaviorofdevelopedcomposite.

Generallylubricantis externallyaddedtoreduce thewear anditiswellknownthatthisposesseveralassociatedproblems whenthematerialsneedperiodicmaintenanceandapplicationof lubricantparticularlytowearpartswhicharedifficulttoaccess.

Inordertoovercometheseexpeditions,inthecurrentresearch aself-lubricating material viz.Gr isutilized andthe sameis preferredbecause thesolidlubricantcontainedinthemmight beautomaticallyreleasedduringslidingandtherebyreducethe wear.

The wear behavior of the hybrid Al7075/flyash cenosphere/GrcompositesshowninFigure8exhibitsdecremental valuewiththeadditionofflyashcenosphereanddecreaseseven morewiththeadditionofGr.Itreachesminimumat10wt.%

flyashcenosphere/6wt.%GrandisverylessthanthatofAl7075.

Duetothepresenceofflyashcenosphereparticleuponthe surface of matrix, the plastic deformation of matrix can be resistedandsoitactsasabarriertothemovementofdislocation

whichcausesmorewearresistancethanbasealloy.Dropping the wear rate of hybrid composite relative toAl7075 can be reasonedtotheformationofathinlubricatingGrfilmonthe tribosurface.ThepresenceofGrisreportedtohelpinreduction ofwearrateofthecompositesbyprovidingasolidlubricating layerbetweenthecompositeandtherubbinghardcountersur- face(Ravindranetal.,2013).AlsoitisbelievedthatGrparticles spreadattheinterfacebetweentheslidingbodiesandthereby decreasethecoefficientoffrictionandhencetheheatgenerated duetofrictiongetsreducedtoagreaterextent(Basavarajappa, Chandramohan,Mukund,Ashwin,&Prabu,2006)

SEM analysis of the worn surfaceof hybrid composite is as shown inFigure 9(a–d). Figure9(a)clearly picturizes the reinforcementparticlesandalsoexhibitsseveraldamagedregion causedduetohighplasticdeformation.Figure9(b)revealsthat aclusterofreinforcementisdistributedoverthematrixmaterial whereincenosphereisseenasgraycrystalandGrasdarkcrystal.

FromFigure9(c)itisnotabletostatethatthesecratersareformed duetotheescapeofcenosphereparticlefromthematrixmaterial.

A relativelysmooth surface can be viewedfrom Figure 9(d) andthewearrateislowduetoincreasedpercentageofGrand againtheresultsareinwellagreementwithBaradeswaranetal.

Amongstthemadespecimens,hybridcompositewith6wt.%Gr contenthaveshowcasedgoodwearresistance; becauseof the reasonthatGrformsathinlubricantlayerandinturnreducing thewearrate;henceitcanverywellbeclaimedthatthenewer compositeishighlysuitableforgearboxapplications.

3.1.6. Corrosiontest

Mostwidelyusedsaltsprayfogmethodwasusedtoinves- tigate the corrosion characteristics of the composite (ASTM

20 µm EHT = 5.00 kV WD = 8.2 mm

Signal A = SE2

Mag = 1.71 KX ZEISS

20 µm EHT = 5.00 kV

WD = 8.2 mm Signal A = SE2

Mag = 1.13 KX ZEISS 20 µm EHT = 5.00 kV

WD = 8.2 mm

Signal A = SE2

Mag = 1.13 KX ZEISS

20 µm EHT = 5.00 kV WD = 8.3 mm

Signal A = SE2

Mag = 1.13 KX ZEISS

Reinforcemets

Reinforcemets Craters

a b

c d

Sliding direction

Fig.9.SEMofworn surfaceof(a) Al707510wt.%flyashcenosphere,(b)Al707510wt.%flyashcenosphere/2wt.%graphite, (c)Al707510wt.%flyash cenosphere/4wt.%graphiteand(d)Al707510wt.%flyashcenosphere/6wt.%graphite.

(7)

Corrosion rate (%)

0.70 0.65 0.60 0.55 0.50 0.45 0.40

0.35

A AC AC2G AC4G AC6G

Composition Fig.10.Corrosionrate.

StandardB117).Thetestspecimenswerepreparedbycutting thecompositecastingsofsize45mm×30mm×5mmandits surfacewereabradedusing600gritsizeemerypapersandare degreased.Thesaltsolutionwaspreparedbydissolving5parts byweightofsodiumchloride(AR)in95partsofdistilledwater.

ThepHofsaltsolutionismaintainedintherangeof6.5–7.2.

Exposurezoneofthesaltspraychamberwasmaintainedata temperatureof35±3^◦Candspecificgravityof1.0268–1.0413.

Exposuretimedurationfortestingthespecimenswassetas24h.

Specimensweredippedincleanrunningwatertoremoveifthere isanysaltdepositsonthesurfaceandthenimmediatelydried withastreamofcleancompressedair.

Theunreinforcedandcenospherereinforcedalloyspecimens offersbettercorrosionresistantcomparedtoGrreinforcedcom- posite specimens. This could be because of the fact that Al spontaneouslyformsathinbuteffectiveoxide layerthat pre- vents furtheroxidation.Growingof Al-oxidelayeris noticed attheendof24haftertest.Corrosionrateincreasesduetothe presenceofGrandincreaseinitsconcentrationraisesthecor- rosionrateandreducesthepolarizationresistanceofAl(Latief, Sherif,Almajid,&Junaedi,2011).

After19hofdurationofthetest,whiterustwasformedonthe reinforcedsamplesfollowedbyminorconfirmationofpits.Al waspittedbysodiumchloridesolutionandinthepitregion,high

concentration ofAlChloride andhydrogen ionswerepresent whichacceleratesthecorrosionprocessinduecourse.

As seen in Figure 10, presence of Gr particles increases the corrosion rate since the addition of reinforcement parti- clescausesdiscontinuitiestotheoxidefilmwhichprotectsthe materialfromcorrosion.Otherfactorsinfluencingcorrosionof the compositesincludeporosity,isolationof alloyingelement tothe composite interface,presenceof aninterfacial reaction product,withhighdislocationdensityaroundthereinforcement phase andvoidsatthe composite interface.Sincethesecom- positematerialsare usedfor gearbox applications,theeffect of corrosioncouldbeminorandanyfurthereffectbecauseof corrosioncouldbeignored.

Figure11revealsthatdevelopedhybridcompositehadbetter mechanicalpropertieswhilecomparedtobaseAl7075alloyand hencethedevelopedcompositecanbehighlyaugmentedasa betteralternativetotheexistingAl7075gradealloyusedinair- craft’sgearbox.Inmachining,turningoperationplayakeyrole inproductionofpartsingearboxassemblyandsothisworkhas alsoconcentratedformachininganalysis(turningoperation)and performanceofhybridcompositematerialandtherebypredict themostinfluencingparameterofmachining.

3.2. Machiningcharacterization

3.2.1. Experimentalsetup

DryturningoperationwasplannedbasedonTaguchifacto- rialdesign.Minitab16softwarewasusedtocreatethedesign matrixandalsotoanalyzetheexperimentaldata.Experiments are carried over casted cylindrical AMMC (AC0, AC2Gr, AC4Gr andAC6Gr) rod with 25mm diameter. CNC turning machine was used for performing the turning operation with polycrystallinediamond(PCD)cuttingtool.Asperthereview done over the available literatures, the machiningparameters selected aregraphitecontent(G), cuttingspeed(V),feedrate (F),depthcut (D)andtheseselectedpredominantparameters with 3 levels are as shown inTable4.As surfaceroughness playsakeyroleinmaterialqualityitischosenastheresponse factor. According to the design of experiment, 27 trails of

FlyAsh cenosphere powder

AI7075 Cenosphere Graphite 100% -

90%

88%

86%

84%

- 10% -

2%

4%

6%

10%

Graphite particles

Rockwell hardness

Hardness 8.7%

Tensile test

Tensile strength 19.6%

Wear & corrosion behavior

Wear resistance 89.4%

Compo-casting process

Fig.11.Overviewofmechanicalcharacterizationofhybridcomposite.

(8)

Table4

Machiningparametersandtheircorrespondinglevels.

Parameters Unit Level1 Level2 Level3

Graphite,G wt.% 2 4 6

Cuttingspeed,N mm/min 100 200 300

Feedrate,F mm/rev 0.1 0.2 0.3

Depthofcut,D mm 0.2 0.4 0.6

Table5

Designofexperimentforturningoperation.

Sl.no. Graphite (wt.%)

Cutting speed (mm/min)

Feedrate,F (mm/rev)

Depthof cut,D (mm)

Surface roughness, Ra(␮m)

1 2 100 0.1 0.2 3.214

2 2 100 0.2 0.4 2.818

3 2 100 0.3 0.6 2.412

4 2 200 0.1 0.4 1.744

5 2 200 0.2 0.6 2.206

6 2 200 0.3 0.2 2.008

7 2 300 0.1 0.6 1.470

8 2 300 0.2 0.2 1.454

9 2 300 0.3 0.4 1.470

10 4 100 0.1 0.2 1.318

11 4 100 0.2 0.4 2.154

12 4 100 0.3 0.6 2.086

13 4 200 0.1 0.4 1.644

14 4 200 0.2 0.6 2.076

15 4 200 0.3 0.2 1.886

16 4 300 0.1 0.6 1.456

17 4 300 0.2 0.2 1.412

18 4 300 0.3 0.4 1.664

19 6 100 0.1 0.2 1.592

20 6 100 0.2 0.4 1.690

21 6 100 0.3 0.6 2.074

22 6 200 0.1 0.4 1.342

23 6 200 0.2 0.6 1.546

24 6 200 0.3 0.2 1.488

25 6 300 0.1 0.6 1.354

26 6 300 0.2 0.2 1.282

27 6 300 0.3 0.4 1.442

experimenthavebeenconductedamongwhichsixexperiments arerepeatedtocalculatetheerrorduetoexternalnoises.TESA RUGOSURF10 surface roughness tester is used to measure the surface roughness data at four different places of each machined component and the average value is recorded for furtheranalysesastabulatedinTable5.

3.2.2. Analysisofvariance

Analysisofvariance(ANOVA)isamethodofassigningvari- abilityintoidentifiablesourcesofvariationandtheassociated degree of freedom inan experiment. The frequency test (F- test)isutilizedinstatisticstoanalyzethesignificanteffectsof theparameters,whichformthequalitycharacteristics.Table6 shows theANOVA result of S/Nratio for surfaceroughness.

Fromtheanalysisitisapparentthat,F0.05valuefromthetable isfoundtobe3.55whichislowwhencomparedtotheFvalue of Gr %, cuttingspeed, feedrate anddepth of cut; this out- comeevidencethefactorthattheparametershavebothstatistical andphysicalsignificanceonthesurfaceroughnessofdeveloped composite.FromtheANOVA(Table6),itcanbeobservedthat theGraddition(21.9%)andcuttingspeed(40.8%)havecon- tributed the major percentage on surface roughness, whereas feedrate(13.2%)anddepthof cut(7.8%)haveminorsignif- icance in surface finish. These results have good agreement withothersimilarworkcarriedoutbyShanmughasundaramand Subramanian(2013).

3.2.3. Determinationofoptimumfactorlevelcombination Figure12showsfourgraphs,eachof whichrepresentsthe meanresponseandthemeanS/NratiofortheGr,feedrate,cut- tingspeedandthedepthofcut.Surfaceroughnessincreaseswith additionofGrandspeedbuthoweverincreasingthefeedrateand depthofcutdecreasesthesurfaceroughness(Muthukrishnan&

Davim, 2009).The valuesof the graphs tabulated inTable7 arebasedonthe S/Nratioandrevealthatthe optimalcutting parametersforsurfaceroughnessareG3,N3,F1andD1.

3.2.4. Predictionofoptimumperformance

Afterselectingtheoptimumfactorlevels,predictionof the surfaceroughnessvalueisdonebyformingregressionequation asfollows:

Ra=2.77985−0.1385G−0.00353N+0.775556F +0.285D whereRaisthesurfaceroughness,Gistheweightpercentage ofGr,NisthecuttingspeedandDisthedepthofcut.Itisvery essentialtoperformaconfirmationexperimentfortheparameter design,particularlywhenlessnumbersofdataareutilizedfor optimization.Thepurposeofthisconfirmationexperimentisto verify theaccuracyof thedevelopedregressionmodel.Itwas observedthatexperimentalresultshavegoodcoincidencewith thepredictedvaluesandfallswithinthe confidencelimitsas showninTable8.

Table6

ANOVAforsurfaceroughness.

Factors Degreesoffreedom Sumofsquares Meansquare F-test P Contribution(%)

Graphite,G 2 1.4084 0.7042 6.29 0.008 24

Cuttingspeed,N 2 2.2473 1.1236 10.04 0.001 38

Feedrate,F 2 0.1564 0.0782 0.70 0.510 3

Depthofcut,D 2 0.0614 0.0307 0.27 0.763 1

Error 18 2.0144 0.1119 34

Total 26 5.8879

(9)

Main Effects plot for SN ratios Data means

Graphite% Cutting speed

Feed rate Depth of cut

-3 -4 -5 -6

2 4 6 100 200 300

0.1 0.2 0.3 0.2 0.4 0.6

Signal-to-noise: smaller is better

Mean of SN ratios

Fig.12.DeterminationofoptimumfactorsusingANOVA.

Table7

Responsetableforsignaltonoiseratios.

Level Graphite, G(wt.%)

Cuttingspeed,N (mm/min)

Feedrate,F (mm/rev)

Depthofcut, D(mm)

1 −6.054 −6.352 −4.169 −4.445

2 −4.071 −4.858 −5.070 −4.764

3 −3.633 −3.177 −5.149 −5.179

Delta 2.422 3.175 0.979 0.735

Rank 2 1 3 4

4. ANNmodeling

ANNinspiredfromthebiologicalneuralsystemhasemerged asawidelyacceptedtoolinsolvingwidevarietyofproblemsin thefieldofmaterialsandengineering.Itsabilitytoreplicatethe stronglearning,clusteringandreasoningcompetenceofbiolog- icalneuronshasmadeitanadvancedtoolinthefieldofexpert systems.Thesamehasbeenemployedinmanynonlinearprob- lems dueto itssound learning capability,parallel computing andnon-linearmapping.Awell-developedANNmodelcanbe aneffectiveinstrumentinpropertyestimation,productdesign anddevelopment,optimization,predictiveperformance,pattern associationandclassification(Kara,Aslantas,&C¸ic¸ek,2015;

Padhi&Satapathy,2013).AnodeisthevitalelementinANN andthenumberofnodedependsonthedutiesandapplications providedtothemodel.AnANNmodelconstitutesofdifferent numberoflayerscomprisedofvaryingnodeswhichyieldsout- putsforthegiveninputsbasedonthetrainingprovided.These typesof tools have beeneffectively employed inthe fieldof material machining characteristics, tribological property esti- mation,mechanicalpropertypredictions,etc.

Eveninthistechnologicalera,itisaknownfactthatsurface roughnessofamachinedmaterialisoneofthebiggestconcerns while developing a material for certain other specific applications.Even thoughmathematicalmodelssuch asregression equationscanbedevelopedformodelingthesurfaceroughness ofdevelopedcompositesbasedontheavailableinputs,thepre- dictabilityofthesamehastobecompromised;suchissuescan verywellbereducedtoagreatextentjustbyemployingANN.

InthisstudytheANNmodelisdevelopedbasedontheexper- imental resultsof surface roughnessobtained fromTaguchi’s designforexperimentationandpredictionofsurfaceroughness basedontheGrquantityasreinforcement,cuttingspeed,feed rateanddepthofcutweredonethereof.Atthesametimeitalso helpsintoyieldsufficientunderstandingontheeffectofprocess parametersoverthesurfaceroughnessofdevelopedcomposites.

4.1. ANNmodeldevelopment

Taguchibaseddesignofexperimentsdevelopedtostudythe surfaceroughnessofnewerAlhybridcompositesalongwiththe attainedexperimentalresultswereconsideredfordevelopingan ANN model topredict thesurfaceroughnessof the samefor unknownvalueswhichwouldlaywithintheprobabilisticdistri- bution.ThequantityoftheGraddition,cuttingspeed,feedrate anddepthofcut,consideredasthegoverningfactorsinobtaining optimalsurfaceroughnesswerechosenastheinputsfordevelop- ingtheANNmodelwhilethemeasuredsurfaceroughnesswas consideredastheoutputtargetfortrainingpurpose.Itisclearly evidentfromjournalsthatbackpropagationtrainingnetworks areconsideredasdominantinsolvingthesetypesofproblems weretheANNmodelsaretrainedbasedonthebackdata.This

Table8

Verificationofoptimalparametersthroughconfirmationtest.

Sl.no. Graphite(wt.%) Cuttingspeed(mm/min) Feedrate,F(mm/rev) Depthofcut,D(mm) Predicted,Ra(␮m) Experiment,Ra(␮m) Error%

1 6 300 0.2 0.2 1.68 1.63 2.9

2 6 350 0.2 0.2 1.58 1.54 2.5

(10)

0.10

0.08

0.06

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

No. of hidden nodes

MAE

Fig.13.MAEvalidationofthedevelopedmodels.

ANN predicted 1.7

1.6

1.4 1.5

1 2 3

Samples

Surface roughness (µm)

Experimental

Fig.14.Validatedresultsofthemodel.

trainingalgorithmworksinsuchamannerthattheerrorbetween ANNpredictedvaluesandactualvaluesprovidedfortrainingare propagatedbackwardandtheweightsareadjustedaccordingly aftereachcycle(Haque&Sudhakar,2001).

Back propagation algorithm network models with varying hidden nodes were trained with the inputs and outputs as provided in Table 5. Out of 27 readings obtained, 24 of them areconsidered for training anddevelopingthe required model while the other three (Expt no. 9, 18 and 27) were employedforvalidatingandtestingthedata.InMATLAB2013, Levenberg–Marquardt back propagation based training algo- rithmwasemployedtotrainthebackpropagationnetwork.The hiddennodeswereselectedandvariedfrom2 to15inasin- glehiddenlayertopologybasedANNmodel.Literatureshave provedthatsinglehiddenlayerbasedmodelhasthecapability tosolveanytypeofproblems.Eachmodelwastrainedfor1000 epochs andtrained toundergo10,000iterationstoobtain the bestmodel.ThemodelswereevaluatedbasedontheMeanSum Square(MSE)value.Thebestmodelfromthese10,000models thusdevelopedwaschosenbasedontheleastMSEvalue.

The models haveto be validated andtested to obtain the efficientmodelwhichcanpredictthepropertieswithbettercor- relation. TheMean AbsoluteValue(MAE) justifiesandsorts outthebestdevelopedmodelamongsttheall.Themodelswith varyingnetworktopologythusdevelopedinthisresearchalong withitsMAEvalueisprovidedinFigure13.

FromFigure13itisevidentthattheANNmodelwithsingle hiddenlayerandwith7nodeshastheleastMAEvalueof0.055;

henceitisthebestamongthedevelopedmodels.Thusitcanbe statedthatafeedforwardbackpropagationmodelwithnetwork topology4-7-1canpredictsurfaceroughnessofthedeveloped AlhybridMMCwithbetterefficiency.

Toassuretheabilityofdevelopedmodel,itwasagaintested withan unknown data(Expt no. 9,18 and27) that waythis datasetisnowhereemployedintrainingthemodel.Againfrom Figure14,itisevidentthatthevalidatedresultsareincoinci- dencetotheyieldedvaluesandhencemodelhencedeveloped hasthecapabilitytopredictthesurfaceroughnessofthecom- positebasedonGrquantity,cuttingspeed,feedrateanddepth ofcut.Figure15givesaclearvisionoftheinputnode,hidden

I1

I2

I3

I4

%Graphite

Cutting speed

Feed rate

Depth of cut

H1

H2

H7 H5

H6 H4 H3

01

Surface Roughness

Fig.15.ANNmodelwithnetworktopology4-7-1.

(11)

Experimental ANN perdicted 3.5

3.0

2.5

2.0

1.5

1.0

0 2 4 6 8 10 12 14 16 18 20 22 24

Samples

Surface roughness (µm)

Fig.16.Validationresultsoftraineddata.

3.2 3 2.8 2.6 2.4 2.2 2 1.8 1.6 1.4

1.5 2 2.5 3

Experimental values R=0.98233

predicted values

Fig.17.ComparisonbetweenANNandexperimentalvalues.

layer(notifiedwithH1–H7)andoutputnodeofthedeveloped ANNmodelalongwiththenetworktopology.

ThesameANNmodelwasemployedtopredictthesurface roughness of Al hybrid MMC for the given values used for trainingthemodel.Themodelwasabletopredict thesurface roughnessbasedonthegivensetofGrquantity,cuttingspeed, feedrateanddepthofcutwithhighefficiencyandisrevealed fromFigure16.

Scatter diagram,amethod tovalidate theefficiency ofthe modelasalinearstatisticalmethodiscarriedoutinthisresearch.

The result which showcases the correlation created by the modelbetweenthepredictedandexperimentalvaluesisdepicted throughFigure17.

ThescatterasshowninFigure17showcasesanRvalueof 98%whichsaysthatthemodelhasagoodpredictabilityandthe samemodelcanbeusedtopredictthesurfaceofthedeveloped compositesforagivenquantityofGr,cuttingspeed,feedrate anddepthofcutfor valueswhichliesinsidetheprobabilistic distribution.

5. Conclusion

The Al7075–flyash cenosphere–Gr composites were produced by two-step compocasting route withdifferent weight percentage of reinforcements (10wt.% of flyash cenosphere and2,4and6wt.%ofGr).Mechanical,microstructure,wear, corrosionandmachinabilityofthedevelopedcompositeswere evaluated.Fromthestudy,thefollowingconclusionsarederived.

• Additionoflowdensityflyashcenosphereparticulatesresults in6.8%decrementindensityforthehybridcompositesi.e.

reducedfrom2.615g/cm³to2.537g/cm³.

• Thebulkhardnessofthecompositesincreasedfrom57HRB to62HRB that is by8.7% with respect tothe addition of 10% flyashcenosphere and2% Gr.On further increasein weightpercentageof Gr,the bulk hardnessdecreases. The tensilestrengthof hybridMMCisinferiorduetoGraddi- tionwhereastheAlMMCwithoutGrexhibitsbettertensile strength(213N/mm²).

• Thewear resistanceof Al7075increaseswithflyashceno- sphereadditionandevenmoreresistanceisexhibitedbythe GraddedhybridMMC.WearrateofthehybridMMCwith 10%flyashcenosphereand6%Gris89.4%lessthanthatof matrixalloy.

• Cenosphere addition decreases the corrosion rate slightly, butthe corrosion rate increases vastly withincrease inGr percentagewhencomparedwiththatofbasealloy.

• Cuttingspeedandreinforcementcontentplayavitalrolein alteringthesurfacefinishofthemachinedcomposites.The developedANNmodelshowcasedhighpredictabilityinpre- dictingthesurfaceroughnessforthegiveninputparameter.

Conﬂictofinterest

Theauthorshavenoconflictsofinteresttodeclare.

References

Alaneme,K.K.,Adewale,T.M.,&Olubambi,P.A.(2014).Corrosionandwear behaviourofAl–Mg–Sialloymatrixhybridcompositesreinforcedwithrice huskashandsiliconcarbide.JournalofMaterialsResearchandTechnology, 3(1),9–16.http://dx.doi.org/10.1016/j.jmrt.2013.10.008

Baradeswaran, A., & Perumal, A. E. (2014). Study on mechanical and wear properties of Al 7075/Al2O3/graphite hybrid composites. Composites Part B: Engineering, 56, 464–471.

http://dx.doi.org/10.1016/j.compositesb.2013.08.013

Basavarajappa, S.,Chandramohan,G.,Mukund,K.,Ashwin,M.,&Prabu, M.(2006).DryslidingwearbehaviorofAl2219/SiCp–Grhybridmetal matrix composites.JournalofMaterials Engineeringand Performance, 15(6),668–674.http://dx.doi.org/10.1361/105994906X150803

Bodunrin,M.O.,Alaneme,K.K.,&Chown,L.H.(2015).Aluminiummatrix hybridcomposites:Areviewofreinforcementphilosophies;mechanical, corrosionandtribologicalcharacteristics.JournalofMaterialsResearchand Technology,4(4),434–445.http://dx.doi.org/10.1016/j.jmrt.2015.05.003 Dinaharan, I., Nelson, R., Vijay, S. J., & Akinlabi, E. T. (2016).

Microstructure and wear characterization of aluminum matrix composites reinforced with industrial waste fly ash particulates synthesized by friction stir processing. Materials Characterization, 118, 149–158.

http://dx.doi.org/10.1016/j.matchar.2016.05.017