Macroscopic simulation of the liner honing process

Science Arts & Métiers (SAM)

is an open access repository that collects the work of Arts et Métiers Institute of

Technology researchers and makes it freely available over the web where possible.

This is an author-deposited version published in:

https://sam.ensam.eu

Handle ID: .

http://hdl.handle.net/10985/7882

To cite this version :

Benoit GOELDEL, Mohamed EL MANSORI, Didier DUMUR - Macroscopic simulation of the liner

honing process - CIRP Annals - Manufacturing Technology - Vol. 61, n°1, p.319-322 - 2012

Any correspondence concerning this service should be sent to the repository

Macroscopic

simulation

of

the

liner

honing

process

Benoit

Goeldel

*

,

Mohamed

El

Mansori

,

Didier

Dumur

(1)

a_{Arts&Me´tiersParisTech,LMPF,EA4106,Chaˆlons-en-Champagne,France}

b_{SUPELECSystemsSciences(E3S)EA4454,ControlDepartment,Gif-sur-Yvette,France}

c

RenaultS.A.S.PowertrainDivision,RueilMalmaison,France

1. Introduction

Cylinderborehoningisamulti-stage metalremovalprocess usuallyusedafterboringtoobtainpreciseboregeometryanda specific surface finish.Material removalis ensuredby pushing abrasivestonesagainsttheworkpiece[1].

Thehoningstonesmovementgeneratesahelicalpatternwitha characteristiccrosshatchangledeterminedbytherelativespindle andstrokespeeds[2].

Thehonedsurfaceinenginecylinderlinerplaysanimportantrole onitsfunctional featuresand containsqualitativecharacteristics

[3,4].Thegenerationof thesesurfacesismonitoredbyagreatnumber ofprocessparametersdefinedbyempiricalsettingupmethods.

Many attempts have been made by researchers to develop numericalsimulationmethodsformanufacturingprocesses,e.g.

[5,6]. This task is also not so easy because honing involves

graduated stages of abrasive finishing processes, using at first coarseabrasivestones,andthenprogressivelyfinergrades[7,8]. Actually,industrialistshavetotestdifferentsetting configura-tionstofindthebestonefortherequestedquality.Tominimize thistuningphase,thispaperproposestheelaborationofahoning modellinkingthemachineparameterstotheprocessresultsand theindustrialqualityindex.Eachstageoftheprocesswillbethus successivelysimulated.Thefinalaimofthissimulationistoassist end-usersinthechoiceoftheappropriatemachineparametersto obtainthespecifiedqualityintermsofform,roughnessandsurface aspect.

Itisbasedonanewmethodtosimulatethespecificprocessof cylinderengine honing.This macroscopicapproachis useful to increase the process understanding and improve the abrasive processknowledge.

2. Kinematicsandforcesimulation

Thederivedmethodisbasedonthephilosophydevelopedin

[10].Theproposedsimulationmodelwillbeelaboratedcoupling an abrasivecuttingmodelanda spatiotemporal discretization that allows integrating the tool kinematics relatively to the workpiece.

2.1. Kinematicssimulation:spatiotemporaldiscretization

Thefirstactionistochooseameshwhichallowsintroducing amacroscopicviewofthecylinderandamicrodefinitionofthe surfacequality.Forthat,classical3Dmeshcannotbeused.The thirddimension,theradialthickness,istoosmallcomparedto thecylinderheightanddiameter.Therefore,ithasbeenchosen tocollecteachvalueofthematerialthicknessandtheroughness in matrix of appropriate dimension: each cell of the matrix represents a small square (dX*dX) of the internal cylinder surface.

Globally, several matrices are defined. Datain the two first matrices of the cylinder represent the thickness of remaining material with or without the local maximum roughness. Data correspondingtotheorientationofthecuttingspeedarecollected inothermatricesso-called‘‘orientationmatrices’’.

Specificroughnessand formdefaultscannedona realcarter withthemetrologymachineareintroducedasinitializationvalues. Thetime-domaindiscretizationwillbedirectlydependenton themappingdefinitionandthemaindirectionofcut.DenotingdX (m)thesizedefinitionofthemeshtile,D(m)themeandiameterof the cylinder bore, Va (m/s) the stroke speed and N (rpm) the angular velocityofthetool, thesamplingperiodrelated tothis discretizationisdefinedby

t

t

p

Theformquality,theroughnessandthesurfaceappearanceproducedbyhoningminimizesthefrictionof thepistonintheliner.Theprocessishowevermechanicallycomplexandtheselectionoftheprocess parameters is currentlybased on empirical methods.The aim of this paperis thus to develop a macroscopicsimulationenvironmentofcompleterealhoningcycles,whichwillhelpend-usersduring thesetting-up.Thisvirtualtoolisbasedonaspace–timediscretizationandamacroscopiccuttingmodel taking into account local contacts betweenthe workpiece and the abrasive tool. The space–time discretization allowsrepresenting themachine environment withthetool, theworkpieceand the kinematics.Simulationresultsarefinallyvalidatedbycomparisonwithindustrialexperiments.

Trajectoriesaredefinedasfortherealmachine:allparameters of the stroke movement (inversion point positions, speed, acceleration)andtherotationspeedcanbeeasilyset.

Ateachtimeinstant,thetwomatricesareoverlaidwiththenew positioncorresponding to the selectedhoning parameter kine-matic.Foreachmeshtilewherethereiscontactbetweenthestone andtheworkpiece,theproposedabrasiveactionmodelisapplied. From this information, the local number of stone passage is integratedand themapofstonenumberpassagedistributionis created.

2.2. Forcesimulation:abrasivecontactpressurecalculation Thehoningmachinehastwostoneexpansionsystems,which allowconnectingthestonesandtheworkpiece:an electromecha-nicalexpansion,whichconsistsinusingthestonepositioncontrol, andahydraulicexpansion,whichusesupwardthrustcontrol.Both ofthemareintegratedinthesimulator,withthehypothesisthat stonescannotrotateontheirstandsandthetoolisnaturallywell centeredinthecylinder.

For electromechanical expansion simulation, only the new radialdimensionofthetooliscalculatedforeachabrasivemesh tileand the local stone pressureis derived. Thislocal value is definedbytheworkpiececharacteristics,thestonethicknessand thevalueofthestoneradialexpansion.In thecaseofhydraulic expansionsystem, thelocalstone pressureis calculated witha special pressure repartition algorithm. Knowing the stone ele-mentsincontact,theglobalpressureisthereforedistributedasa functionofstoneandworkpiecethicknesses.

Byintegratingthelocalcontactpressureforeverystonemesh tile,theglobalradialforceis obtained.In thecase ofhydraulic expansion, this force is naturally the same as specifiedby the machine.Inthecaseofelectromechanicalexpansion,the simula-tiongivesustheforcedevelopedbytheelectricactuator.

Assumingthatthefrictionbetweentoolandcylinderisdry,and denoting f1 the axial friction coefficient, the axial forces are

calculatedasfollows: FrðtÞ¼ðdXÞ2 X i;j Pcði;jÞðtÞ (2) FaðtÞ¼signðVbðtÞÞ f1FrðtÞ (3) To identify this coefficient that must be integrated in the simulator, the only available measurement is the axial force. Therefore,based on experimental axial forceresults and using simulatedradialforce,thecalculatedaxialfrictioncoefficientsf1

(=Fa/Fr)isgiveninTable1.

3. Cuttingmodel

Amacroscopicmodelofabrasivecuttingisdevelopedbelow. Foraspecificcoupleofabrasivestoneandworkpiecematerial,the factors which influence the removal material and the honed surfacequalityarecuttingspeedandstonepressure[1].Therefore, measuring the stock removal and the stone wear at different couplesofspeedandpressureenablestodeducethis predictive model.

FollowingtheworkofSasakiandOkamura[9],thepredictive modelisgeneratedwithafunctionassimpleaspossible.Inthe literature, experiments proved that stock removal is a linear functionof contactpressure forthesame cuttingspeed, in the middleoftherange.Forafixedstonepressure,thestockremoval

dependsonlyofcuttingspeed.Inthemiddleofthecuttingspeed range,stockremovalisalinearfunctionofcontactpressure.

ThestockremovalSRappearsthustobeabi-linearfunctionof stonepressurePandcuttingspeedVcwithinacertainrangeclose totheoptimum.Thisfactisexpressedbythefollowingequation, whereu,v,andwareconstantsdependingonstoneandworkpiece materials.

SR¼uPþvVcþwPVc (4)

3.1. Identificationofcoefficientsofthestockremovalequation Fortheestablishedabrasiveregime,aseriesofexperimentshas beenconductedtodeterminethestockremovalrateasafunction ofexpansionpressureandcuttingspeedforeachtypeofabrasive usedinourhoningprocess.Thesamemethodasin[1]wasusedto determine thestock removal: thediameters before and aftera simplestagehoningaremeasured,thevolumeofthestockremoval isalsocalculated.Knowingthetool geometrywiththeabrasive andthetimecycle,thestockremovalrateisthencalculatedforthe honingprocesscondition.

Table 2 gives thedifferent resultsobtained tocalculate the

primaryinputtothecuttingmodelofthehoningsimulation. Based on this, it has been checked that the previous experimentalstockremovalresultscanbeextrapolatedtoalarge rangeofstockremoval.Table3presentsthecoefficientsu,v,w identifiedfromtheseresultsbylinearregressionforeachtypeof abrasivestone.

3.2. Roughnessmodel

Only thelocal maximum roughness (Rz) is represented and modifiedateachiteration,incaseofremovedmaterial.Sevencases ofroughnessmodificationhavetobedistinguished,dependingon thelocalpenetrationdeepness.Theabrasiveindentation(Tg)isa third of thegrit size. Asan example, in Fig.1, in case A1, the penetrationdeepness(U)is lowerthan(Tg)andlowerthanthe localroughness(Rz).

Table1

Axialfrictioncoefficientidentification.

Abrasivetype Peakradial

force[N] Peakaxial force[N] Friction coefficient SiC-IAS65/120/1/8Vs 480 390 0.813 SiC-SCG600KE206469 320 280 0.875 Table2

Resultsofremovalrateexperiences.

Abrasivetype Cutting

speed(Vc) [m/min] Expansion pressure [105 Pa] Removal material [mm3 ] Cycle time[s] Removalrate [mm3 /s/mm2 ] SiC-IAS65/ 120/1/8Vs 40 18a 688.28 22.10 0.019 40 30a 809.53 17.85 0.027 40 36a _{595.98 11.19 0.032} 50 21a _{693.07 14.67 0.028} 50 27a 784.05 14.73 0.032 50 35a 652.81 10.46 0.037 60 22a 655.61 18.61 0.021 60 28a 658.13 10.30 0.038 60 37a 402.67 5.24 0.046 SiC-SCG600KE 206469 40 3 1068.8 52.06 0.012 40 3.75 1150.4 54.26 0.01262 40 5.25 1625.5 71.60 0.01351 50 4.5 1699.7 80.92 0.01250 50 6 1469.6 69.87 0.01252 50 7.5 1989.9 93.69 0.01264 60 4.5 1388.0 67.41 0.01226 60 6 1707.1 81.28 0.01250 60 7.5 2293.2 107.5 0.01269 a

Average equivalent expansion pressure value measured with monitoring

system.

Table3

Abrasivecuttingcoefficientsidentifiedfromexperiences.

u v w

SiC–IAS65/120/1/8Vs 3.72E04 2.69E05 5.05E06

Thebearingarearatiooftheworkpiece(T%F)andthebearing

arearatiooftheabrasivestone(T%R)areassumedtobelinear.In

thecaseA1,illustrated inFig.2,fortheradialabscissax,these ratiosaregivenby(5).Intheinteractionzone,thenewbearing ratiooftheworkpieceisdeducedby(6).

T%R¼ Tgx Tg andT%F¼ xTgþU Rz (5) T%interact F ðxÞ¼T%FðxÞð1T%RðxÞÞ (6) Butitisnotalinearfunctionofx.Tofindthelinearbearingratio equivalenttotheinteractionbearingratio,theremainingmaterial volumeisequalizedasin(7)andthenewroughnessresultingfor caseA1isRzi+1givenby(8).

I T%equiF dx¼ ZTg TgU T%interact F ðxÞdxþ Z TgUþRzi Tg T%FðxÞdx ¼Rziþ1 2 (7) Rz_iþ1¼Rzi U3 3TgRzi (8) The same method is used to calculate the new equivalent roughnessforthesixothercasesofpenetration.

4. Simulationresultsandexperimentalcomparison

Thissectionnowfocusesontheresultsofthesimulationand comparethemtoexperimentalrecords.

4.1. Simulatedandmeasuredaxialforces

Fig.3illustrates duringone strokeperiod thesignalcoming from the strength sensor and the simulated axial force. They appear to have the same profile and as expected the friction coefficientf1actsasagainforthesimulatedforce.

Thisexperimentshowsthatthesignalformandthepatternare onlyinfluencedbytheprocessgeometryandkinematics,meaning thatthehypothesisofdryfrictionforaxialmovementisvalid.

4.2. Cycletimeforecast

Thecycletimeisuncertainbecausecylindersarealldifferent. Cylinderdiametersarewithinalargetoleranceinterval.Thismay inducemorematerialremovedtogetthefinaldiameter.Thehoning toolisequippedwithblowingnozzlesforin-processmeasuring.It enablestosimulatethe‘‘macro-formcorrector’’oftherealmachine. Thehoningmachinerecordsthreediametersatthreedifferentlevels definedbytheuser.Ifoneofthediameterscanistoosmallcompared totheotherones,themachinemakessomeshortstrokemovements onsmalldiameterzoneuntilthediameterisaslargeastheothers. Shortstrokescouldlengthenthecycletime,soitisimportanttotake into account this functionality to forecast it. Numbers of short strokesinTable4showthatthesimulatedbehaviorissimilartothe onewiththerealcorrector.Theuseofthemacro-formcorrectorand thediversityofdifferentinputcylinderformcreateanadditional hazardoncycletime.Tobeclosertotherealityspecialthickness matricesarecreatedintegratingrealformdefault.Thestudyofthe evolution of the global cylinder form is then carried out by simulatinghoningonrealscancylinder.

The experiment consistsin makingtherough honingon 36 cylinderbores. Thequality formand diametersweremeasured before.ThecylindersoftypeAhaveadeformitysimilartoabarrel andoftypeBsimilartoacone.Scannedpointsissuedfromthe measurementmachineareusedtocreatethematrixmeshofthe cylinder.

InTable4,thecycletimeandnumberofactivationofthemacro formcorrectorarecomparedduringthecycle.

The slightdifference between realand simulatedcycletime illustratesthevalidityofthedevelopedcuttingmodel.

4.3. Roughnessforecast

Thefinishingisahoningprocessstagedesignedtoreducethe roughnessandtogenerateasurfacetexturewithcrossgrooves.

Thisstageessentiallysimulatestheroughnessevolution and thecylinderformquality.Themostinterestingfinishhoningresult istheroughnessmap. Thismapis ofgreatinterest,showingan approximate roughness for each point of the cylinder surface. However,thesimulationresultswillbecomparedbelowwiththe experimentsonlyinthreepoints,becausetheroughness measure-mentsareheavytoimplement.

Fig.1.Twodifferentcasesofpenetrationmode.

Fig.2.Bearingratioandroughnesscalculation.

Fig.3.Axialforcescomparisonforsiliciumcarbideabrasive.

Table4

Cycletimecomparisonforroughhoningwithandwithoutmacroformcorrector.

Cycletimecomparisonwithexpansionspeedat6mm/s

Initialform default Cutting speed [m/min] Macro-form corrector

Cycletime Number

ofshort

strokes

Meas. Sim. Meas. Sim.

TypeA(><) 55 No 32.4s 33s 0 0

55 Activate 37.8s 38s 3 4

TypeB(/\) 55 No 29.8s 30s 0 0

55 Activate 39.2s 40s 5 5

Therefore, Table 5 compares the Rz roughness measured experimentallyatthreedifferentheightsofthecylinderandthe valueoftheroughnessmapforthosepoints.

Thecylinderformsimulatedisgeneratedfromtherealcylinder formscannedaftertheroughhoningstage.Theformevolutionis also compared during this stage. The experiment involves 18 samples grouped into three categories. Comparison between experimentand simulation formquality resultsis presentedin

Table6.

4.4. Macrotexturequalityforecast

Anothercriterionofinterestisthesurfaceappearanceformed by crossing grooves. The surface appearance is forecasted by calculatingandrecordingthedirectionofstonepassagesforeach meshtileandduringallthefinishhoningandplateauhoningtime. Onlydirectionsassociatedtoa positivelocalstockremovaland withanearendremainingstockthicknessaretakenintoaccount duringthecompilationofthesurfaceaspectmap.

Criteriaofsurfaceaspectqualityarecurrentlydefinedbyhuman viewanalysisandthereisnonumericalindex.Theimportanceof the grooves cross angle is known from the experiences. These criteria are thus considered to evaluate the simulation. Fig. 4

shows views from well crossed and non conformed zones determinedbythesimulationandtherealmicroscopicviews,in completeagreement.

Thisstagesimulationcouldeasilyforecastrecommendedcycle time toobtaina homogenousappearanceall over thecylinder surface.Thissimulationistheonlyonetopredictsuchcriteria. 5. Conclusion

This paper introduces a functional approach of the honing processsimulation.Themacroscopicmodelisusefulforgeneration ofhonedsurfacemaps.Thekinematicsmodulecaneasilycalculate the stone passage number map and determine local contacts between abrasiveand workpieceat each instant of thehoning cycle.Theforcemodulecoupledwiththecuttingmodelallows determiningthestockremovalandthemapofthestockremaining thickness. The choice of the tool geometry and the honing kinematicscouldbethereforeoptimizedinordertohavethebest uniformhoningsurface.

The definition of stock removal in thecylinder bore allows determiningthe strokesnumber and thecycle time needed to

achieve the requested form quality and roughness. Combining numberofstonepassage,cuttingorientationandthicknessofthe lateststonepassage,enablestocreatethesurfaceaspectmapping. This mapping helps manufacturers to set up optimized stroke parameters,suchasaccelerationandstrokeamplitudes.

Toconclude,this macroscopicsimulation canforecast macro andmicrogeometrycriteriabytakingintoaccountcylinderand toolgeometryandtheinitialroughnessandthegritsize.Results obtainedwiththissimulationencouragedustocontinuewiththis typeofmesh.Buttheuseofa3Dmeshofthewholecarterallows calculating the geometrical deformation. By coupling the 2D honingsimulationwitha3DFEMoftheenginebloc,thesimulation willbeabletocalculatetheinstantaneousdeformationduringthe honingprocess.

References

[1]SabriL,ElMansoriM(2009)ProcessVariabilityinHoningofCylinderLiner

with Vitrified Bonded Diamond Tools. Surface & Coatings Technology

204:1046–1050.

[2]VoronovSA,GouskovAM,BobrenkovOA(2009)ModellingofBoreHoning.

International Journal of Mechatronics and Manufacturing Systems 2(5/6):

566–575.

[3]HaasisG,WeigmannU-P(1999)NewHoningTechniqueReducesOil

Con-sumption.IndustrialDiamondReview3(99):205–211.

[4]SantochiM,VignaleM,GiustiF(1982)AStudyoftheFunctionalPropertiesof

HonedSurface.AnnalsoftheCIRP31(1):431–434.

[5]TorranceAA(2005)ModellingAbrasiveWear.WEAR258:281–293.

[6]To¨nshoffHK,PetersJ,InasakiI,PaulT(1992)ModellingandSimulationof

GrindingProcesses.AnnalsoftheCIRP41(2):677–687.

[7]SabriL,MezghaniS,ElMansoriM,ZahouanicH(2010)MultiscaleStudyof

Finish-honingProcessinMassProductionofCylinderLiner.WEAR271:509–513.

[8]LeeJ,MalkinS(1993)ExperimentalInvestigationoftheBoreHoningProcess.

TransactionsoftheASME115:406–414.

[9]SasakiT,OkamuraK(1959)TheCuttingMechanismofHoning.JapanSocietyof

MechanicalEngineers2(5):80–85.

[10]ChenX,RoweWB(1996)AnalysisandSimulationoftheGrindingProcess.Part

II. MechanicsofGrinding.InternationalJournalofMachineToolsManufacturing

36(8):883–896.

Table5

Roughnesscriterioncomparisonbetweensimulationandexperiments.

Experiment Cuttingspeed[m/min] Expansionspeed[mm/s] Groovecrossangle[8] Accel.ofstrokeinversion[m/s2

]

TypeI 40 3.5 50 15

TypeII 40 3.5 50 10

TypeIII 40 3.5 50 5

Experiment RoughnessRzlevel1[mm] RoughnessRzlevel2[mm] RoughnessRzlevel3[mm]

Meas. Sim. Meas. Sim. Meas. Sim.

TypeI 6.520.35 6.5 6.060.47 6.0 6.470.35 6.6

TypeII 7.200.46 7.0 6.580.36 6.3 7.300.36 7.1

TypeIII 7.570.36 7.7 7.380.34 6.8 9.060.37 8.0

Table6

Formdefaultcomparisonbetweensimulationandexperiments.

Experiment Cylindricity [mm] Straightness [mm] Average roundness[mm]

Meas. Sim. Meas. Sim. Meas. Sim.

TypeI 10.910.2 11.0 5.810.2 5.9 5.900.2 6.8

TypeII 10.610.1 10.5 5.750.2 5.6 5.770.2 6.0

TypeIII 10.570.3 10.0 5.610.2 5.3 5.600.2 5.3

Meas.:experimentalmeasurements;Sim.:simulationresults