B.Barraa,⇑,L.Momma,Y.Guerreroa,L.BernuccibabFederalUniversityofSantaCatarina(UFSC),DepartmentofPost-GraduationinCivilEngineering(PPGEC),RuaJoãoPioDuarteSilva,s/n,CEP:88040-970,Florianópolis,SC,BrazilPolytechnicSchooloftheUniversityofSãoPaulo(EPUSP),DepartmentofTransportationEngineering,RuaProf.AlmeidaPrado,trav.II,No.83,CEP:05508-970,SãoPaulo,SP,Brazilarticleinfoabstract
Thispaperevaluatesthefatiguebehaviorofdenseasphaltmixesindryandenvironmental-conditioningstates.Two-pointbendingtestswerecarriedoutwithoutrestperiodsat10°Cand25Hz,withthecontrolofthedisplacementamplitudeimposedontrapezoidalspecimens.Theeffectsoftwodifferenttypesofmineralfillerswerealsotested(graniteandlimestonepowders).Theresultsindicatethattheactionofwaterhasadecisiveinfluenceonthefatigueresistanceoftheasphaltmixes,theeffectbeingmoreneg-ativewithanincreaseintemperature.Furthermore,theasphaltmixformulatedwithlimestonepowderpresentedabetterfatigueresistance.Ó2011ElsevierLtd.Allrightsreserved.Articlehistory:Received27July2011Receivedinrevisedform27September2011Accepted2October2011Availableonline24November2011Keywords:DenseasphaltmixesFatiguebehaviorMineralfillersWater1.IntroductionFatigueisoneofthemostimportantphenomenathatleadpavementstructurestorupture,principallybituminouslayers[1–3],andhenceitisveryimportanttointerpretitcorrectly.Itoc-cursineverytypeofpavementstructure.However,theoriginsandprocessesassociatedwiththeruptureofmaterialsaredistinctanddependentonthetypeofstructure.Thecausesmostcommonlyidentifiedarerelatedtothetrafficloadingintensityandtotheenvironmentalconditions,withemphasisonthethermalgradient.Therefore,asawidediversityofvariablesisinvolved,crackingcon-trolalongthesurfacelayersisadifficulttask.Itisthusveryimpor-tanttobeabletoeffectivelydiagnosethenatureandthecausesofthefatigueprocess,inordertoavoidordealwithitspropagationefficiently[4].Inpracticalsituations,asphaltmixesaresubmittedtoashorttermaxleloadingapplication,consideringeachparticularpointofthetire-pavementcontactduringthedynamicmovementofvehi-cles.Sincethepavementstructuresaresupportedbydeformableunderlayers,thereisthegenerationofmechanicalefforts,suchasalternatingflexionalongthelongitudinaldirectionofthetraffic(Fig.1),whichhasanalmostsinusoidalform.Consequently,tensileandcompressionstrainsarisealternatelyandrepeatedlyatthebottomandatthetopofthebituminouslayersduringeachmove-mentofthetiresonthestructure.However,theseloadingapplica-tionsdonotleadthematerialtoruptureimmediately,buttheirrepetitioninthelong-termcancausefatiguecracking[5–7].Inthiscontext,themechanicalbehaviorofthematerialsmustbeevaluatedcontinuouslyinlaboratorytests,inordertosimulate,ascloselyaspossible,theloadingapplicationconditionsobservedinthefield(Fig.1),withtheaimofbetterunderstandingthefati-guephenomenonoftheasphaltmixes[8,9].Also,determiningtheactionofwaterintermsofthemechanicalbehavioroftheasphaltmixesisverypertinent,giventhatheavyrainfallcharacterizestheclimateofmanycountriesandregionsworldwide,mainlyinplacesdefinedastropicalorequatorial,reach-ingamountsashighasaround2500–3000mm/year[10].Toaddresstheseissues,thispaperpresentstheresultsofare-searchstudycarriedoutbyBarra[16]onthefatiguebehaviorofdenseasphaltmixes,emphasizingtheactionofwater.Inaddition,two-pointbendingtestswerecarriedoutat10°Cand25Hzwith-outrestperiods,withthecontrolofthedisplacementamplitudeimposedandusingspecimenswithtrapezoidalgeometrykeptcon-tinuouslyimmersedinwaterduringthetests.Suchtestinghasnot,asyet,beenwidelyexploited,accordingtothetechnicalpublica-tionsinthisarea.2.CharacterizationofthefatiguephenomenonFatigueisaphenomenondefinedasprogressivedamagecausedtoagivenmaterial,whichiscapableofleadingtoitsrupturethroughtherepetitionofalargenumberofloadingapplicationswithastress⇑Correspondingauthor.Tel.:+554896521931;fax:+554832344069.E-mailaddress:brenobarra@gmail.com(B.Barra).0950-0618/$-seefrontmatterÓ2011ElsevierLtd.Allrightsreserved.doi:10.1016/j.conbuildmat.2011.10.003B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134129Fig.1.MeasuredstrainsignaloverlappingasinusoidalstrainsignalcalculatedusingtheFourierSeries[5].amplitudelowerthanthatnecessarytogenerateitsruptureundermonotonicloadingconditions[3,9].Thebeginningoftransversalorlongitudinalcrackingatthebot-tomofthebituminouslayersandthesubsequentpropagationtothetopofthepavementstructuresisdependentonthestiffnessandthicknessofthelayers.Thestraininthebottomofthelayerscanreachthegreatestamplitudesasfollows[1]:alongthetransversaldirection,provokinglongitudinalcracking,andalongthelongitudi-naldirection,provokingtransversalcracking[11,12].Inpavementstructures,thestrainamplitudeisthevariablethatcontrolsthefatiguecrackingphenomenonandnotthestrength(relatedtothetestthatcontrolstheforce)[3].Thus,testswithcon-trolofthestrainamplitudewereappliedinthisresearch.Furthermore,duringthetestscarriedoutwithasphaltmixesthroughthecontrolofthedisplacementamplitude,theprogressivedamagerateofthematerialremainsmoderate.Thus,thistestcondi-tionreproducesalineardominiumofloadingapplicationandthecrackingpropagationthroughthespecimensisdevelopedinalong-termprocess,asoccursinthecaseofbituminouslayersinthefield.Ontheotherhand,duringthetestscarriedoutwithcontroloftheforceamplitudethedamagerateisacceleratedandthustheforceisappliedwithstrongnon-linearbehaviorandthecrackingpropagationoccursintheshort-term[13].Consideringthefatiguephenomenonitself,itispossibletoidentifythreedistinctstagesduringthetestswithcontrolofthedisplacementamplitude[6,9,14]:theforcedecreasesquicklyintheinitialmomentsofthetests,however,forashorttime;subse-quently,asecondstablestagebegins,inwhichthedecreaseoftheforceoccursslowly,thisrepresentingthemajorpartofthetest;andfinally,propagationofthemicrocrackswhichappearedduringthesecondstagegeneratemacrocracking,leadingtoafastdecreaseintheforceinthefinalmomentsofthetests,immediatelybeforethetotalruptureofthematerial(Fig.2).Thearbitraryrupturecriterionadoptedforthefatiguetestsper-formedwithcontrolofthedisplacementamplitudeishalfoftheforcevalueregisteredatthemomentofthefirstdataacquisition[3].Fig.2.Evolutionoftheforceduringafatiguetestexecutedwithcontrolofthedisplacementamplitude[9].Fig.3.TheWöhlercurve[9].Thefatiguetestsgavescatteredresults,afindingwhichisnor-mallyrelatedtothetypeoftest(forceordisplacementamplitudecontrol)andtothegeometryofthespecimens.Inthiscontext,themostcriticalsituationisobservedforthetestsinwhichtheprinci-pleisbasedonthecontrolofthedisplacementamplitude.Hence,alargenumberofrepetitionsmustbecarriedoutinordertoobtainasetofresultswhichrealisticallyrepresentsthefatiguephenome-nonforagivenmaterial[3,9].Thefatigueresistanceofthematerialsisgraphicallyrepre-sentedbytheWöhlercurve(Fig.3)whichisnormallycalculatedusingEq.(1).Thiscorrelatesthenumberofcyclesappliedtothematerialuntilitsrupturewiththeamplitudegeneratedbythedis-placementorbytheforceimposedduringthetests[13].S¼a:NÀbð1ÞwhereNisthenumberofcycles;Sisthestrainorstrength,relatedtothedisplacementorforceamplitudesappliedduringthetests,respectively;a,baretheconstants,withbcorrespondingtotheslopeofthefatiguecurve.3.AggregategradationcurvesTheaggregategradationcurvesoftheasphaltmixeswerecalcu-latedwiththeuseoftheFuller–Thompson[15]equation(Eq.(2)).p¼a:ðd=DÞnð2Þwherepisthepercentwhichpassedthroughthesmallestsieveopeningoftheaggregategradationcurve;aisconstant(100);disagivensieveopeningalongtheseriesoftheaggregategradationcurve(mm);Dismaximumdiameteroftheaggregategradationcurve(mm);andnisthepoweroftheaggregategradationcurve.TheaggregategradationcurvesareshowninTable1.Asthereisnodifferenceinthepercentdistributionoftheaggregatefractionsthetrajectoriesareoverlapped(Fig.4).Forbettercomprehension,itshouldbenotedthatallaggregatefractionsretainedabovetheTable1
Percentdistributionoftheaggregategradationcurves[16].Sieveopening(mm)Passedthrough(%)Distribution(%)19.1100–12.781.318.79.5070.111.14.7649.420.82.2033.416.11.2024.58.80.617.27.30.312.15.10.158.53.60.0756.02.5––6.0130B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134)%( 100ev90ei80S 70hg60uo50rh40T 30gn20is10sa0CTB and CTBPCP0,010,1110Sieve Opening (mm)Fig.4.AggregategradationcurvesCTBandCTBPC[16].sieveopeningof0.075mmhavethesamemineralogy,i.e.,theyareformedofgraniteparticles.Thus,thedifferencebetweenthemin-eralskeletonsisrelatedtothenatureofthefillerretainedbytheabove-mentionedsieveopening.ThecurvethatrepresentsthegranitepowderfractionsisnamedCTBandthatforthelimestonepowderisnamedCTBPC.AlltheaggregatesusedwerecollectedinthenorthernregionofBrazil,specificallyinAmazonia.AsshowninTable1,themaximumdiameter(D)oftheaggre-gategradationcurvesis19.1mmandthesmallestsieveopeningis0.075mm,anditcanbeobservedthat100%and6.0%ofthematerialpassedthroughthesesieveopenings,respectively.Thepercentdistributionoftheaggregatefractionspresentedapower(n)of0.51,whichindicatesanasphaltmixwithdensegradation[16].Theasphaltbinderusedisclassifiedbythepenetrationtestas50/70(0.1mm).4.MineralogicalcompositionandchemicalaspectsofthefillersThemineralogicalcompositionsofbothfillers(graniteandlimestonepowders)usedinthisresearchweredeterminedbythemethodknownassemi-quantitativechemicalanalysisusingX-rayfluorescenceandtheresultsarepresentedbelow.Thegranitepowdersamplesareformedbyquartz(35.0%),albite(33.0%),zinnwaldite(12.0%),microclinium(18.0%)andcaulinite(2.0%).Itisimportanttoemphasizethatthefeldspategroupcom-prisesthemineralsalbiteandmicroclinium,whilethemicagroupincludesthemineralzinnwaldite.Allthemineralsdefinedinthesemi-quantitativeanalysisarerichinsilicacontentsand,hence,theyareclassifiedasfelsicsoracids,andarethuselectronegatives.Thelimestonepowdersamplescontaincalcite(90.0%),dolomite(6.0%)andquartz(4.0%).Calcitehasoneofthemostcommonmin-eralcompositions,beingformedofcalciumcarbonate(CaCO3),usu-allywhiteorcolorless,besidesgray,red,green,blueoryellow.Dolomiteisadoublecalcium–magnesiumcarbonate(CaMg(CO3)2)formedofrhombohedralcrystalsthatcanbepresentinthecolorsrose,white,gray,green,brown,andblackorbecolorless[17].Itcanthereforebeinferredthatcalciteisamineralofmainlypositiveelectricalcharges(electropositive)and,consequently,thelimestonepowderusedasafillerinthisresearchwasalsoelectropositive.Regardingthechemicalaspectsofthefillers,theconceptofThermodynamicSurficialEnergyissuitabletobetterunderstandthephenomenonrelatedtothemodificationoftheoriginalcharac-teristicsoftheasphaltbindersbythefillers.TheseconceptsarebasedontheSurficialEnergyTheory,importingsometheoriesfromclassicalchemistry,whichstudiestwoactivecomponentsinthemolecularinteractionprocess,suchasLifshitz–VanderWaalsforcesandtheLewisacid–baseinteraction[18].Lifshitz-VanderWaalsforcescomprisethreesubcomponents:Londondispersionforces,whichgoverntheattractionbetweenneighboringelectronicsurfaces,producingadipole–dipoleinduc-tioninteraction;theDebyeinductionforce,whichisproducedbyadipoleinducinganotherdipolethatiscoupledtoaneighboringmolecule;andtheKeesomorientationforce,representingthedi-rectinteractionbetweentwodipolesthatareorientedvoluntarilytowardeachother.TheLewisacid–baseinteractiongovernsthedirectinteractionbe-tweentheelementsofacidandbasiccharacteristicspresentedinapar-ticleorinanintegratorsystem,suchastheasphaltbinder–filleroraggregate–masticsystems,wherethepredominantratio(acidorbasic)canbeincreasedordecreasedduetotherepresentativecontentofeachelementthatconstitutesthematerialorthesystem,andisexpressedbythefollowingequation[18,19].CAB¼pffiffiffiffiffiffiffiffiffiffiffiffiCþCÀffið3ÞwhereCABistheLewisacid–baseinteraction;C+isLewisacidcom-ponent;andCÀistheLewisbasiccomponent.Asthefillerparticlescontainpositiveandnegativeions,theacid–baseinteractionsincludealltypesofelectronicbondsinvolv-ingelectrondonation(receptionofprotons)andreception(dona-tionofprotons),includingtheHydrogenbondsorbridges[19].Insummary,bothcomponents(Lifshitz-VanderWaalsforcesandLewisacid–baseinteraction)canbeusedtoexplainthemolec-ularinteractionsrelatedtotheadhesivebonds,whicharepresentattheasphaltbinder-aggregateinterface,andtothecohesivebondsinthemasticsthatformtheasphaltmixes[18].Theadsorptionconceptisalsousedinresearchstudiestounderstandandverifythemolecularinteractionsbetweentwoele-ments,e.g.,theasphaltbinder–fillersystem,whichisdefinedastheadhesionofthemoleculesofafluid(adsorbate)toasolidsur-face(adsorbent),theadsorptiondegreebeingdependentonthetemperature,pressureandcontactsurfacearea.Theadsorptionconcepthastwovariables:chemical(chemi-sorption),wherethemolecules(oratoms)adsorbontoanadsor-bentsurfacethroughtheformationofchemicalbonds(generallycovalent),tendingtobeadheredinanenvironmentthatprovidesthegreatestnumberofpossibleconnectionswiththesubstrate;andphysical(physisorption),wherethemoleculesoftheabsorbentandoftheadsorbateelementsinteractviaVanderWaalsforces.However,althoughthelatterinteractionshavelong-rangebehav-ior,theyareweakanddonotformchemicalbonds.Poroussolidsareexcellentadsorbents,andthusthefillersthatpresenthighporositywillhavethecapacitytoadsorbagreateramountofas-phaltbinderwhich,theoretically,wouldgenerateastrongadhe-sionoftheparticlesand,subsequently,anincreaseinthemasticrigidity[20].Thechemicalcomponentsdescribedabove,togetherwiththemineralogicalnatureofthematerials,canbedefinedaselectroneg-ativeorelectropositive,theformerbeingthemostvulnerabletolackofadhesionandstrippingoftheasphaltbinder-aggregatesys-tem,incomparisontothesecond[20–22].5.MechanicalanddataacquisitionapparatusThefatiguetestswereperformedwiththeuseofanapparatusdesignedanddevelopedinBrazilbythePavementLaboratoryoftheFederalUniversityofSantaCatarina(UFSC),calledFADECOM/UFSC,whichcanbeusedtoevaluateproceduresusedtodeterminethefatiguebehaviorofasphaltmixes,accordingtothespecifica-tionsstandardizedbytheFrenchmethodology[23,24].Thisapparatushasasolidironbase,whichsupportsaclimaticchambermadeofpolyurethane.Thetemperatureinsidethischam-berisadjustedbyanautomaticdigitalcontrollerthatcommandstwodistinctclimaticsystems:oneforheatingandanotherforfreezing.Hence,whenoneofthesesystemsisbeingadjustedforoperation,theotherisdeactivatedautomatically,thusguarantee-ingthedesiredtemperatureforthetests.B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134131Fig.5.Setofpiecesinvolvedinthefatiguetestprocedures.Withthisautomaticdigitalcontrollercommandingthetwocli-maticsystems,itispossibletocarryoutthetestswithtemperaturesrangingfromÀ30°Ctohigherthan100°C.Thetemperaturevalueduringthetestsissuppliedby9thermocouplesplacedatdifferentpointsonthechamber(4closetothespecimens)withanaccuracyof0.1°C.Afrequencyinverterisresponsibleforadjustingthefrequencyofthetests,whichcommandsa6poleinductionmotorconnectedtotwoeccentricaxles.Therotationalmovementoftheeccentricaxlesgeneratesaforcethatistransmittedgraduallytoasetofinterlockingpiecesuntilreachingtheoscillatorrodsconnectedtothespecimens(Fig.5).Theforcetransmittedgraduallytotheoscillatorrodsappliesabackandforthdisplacementinthetoppartofthespecimens,whichgeneratesasinusoidalloadingsignal(Fig.1),i.e.,alternatingflexioneffort.Theloadinganddisplacementsignalsarecapturedsimulta-neouslybyloadingcellsandbytheHalleffectsensors,respectively.AdataacquisitionsystemcapturestheelectricalpulsessentbytheloadingcellsandbytheHalleffectsensorswithasamplefre-quencyof640Hz.AcomputerizedterminalthenconvertsthesepulsesintovaluesofforceanddisplacementforanExcelsheetinrealtime.6.PlanningofthetestsThefatiguetestswerecarriedoutaccordingtotheFrenchstan-dardizedspecification[23],at10°Cand25Hz,withoutrestperi-ods,withcontrolofthedisplacementamplitudeimposedontrapezoidalspecimens,whichwereobtainedbysawingtheasphaltmixslabscompactedinthelaboratory,withanaccuracyof±1.0mmconcerningthefollowingdimensions:25.0mm(smallbase),70.0mm(largebase),25.0mm(thickness)and250.0mm(height).Thespecimensweresubmittedtoaselectionprocessaftersaw-ing,takingintoaccounttwovariables:thegeometry(withanaccu-racyof1.0mm)andtheairvoidcontent.Thestandardizedspecification[23]determinesthatatleast18specimens(distrib-utedequallyamong3differentstrainlevelsofevaluation)mustbetestedinastudyonthefatiguebehaviorofagivenasphaltmix.Inthisparticularstudy,setsof24specimenswereselected,distributedinto3subsetsof8unitstobetestedateachstrainlevel.However,tovalidateeachsetof24specimenstobetested,theymustcomplywiththelimitsdefinedbytherigorousstatisticalcri-teriaofthestandardizedspecification[23].Thesecriteriaarethevariationcoefficient(61.0%),whichisbasedontheresultsfortheconstantKerelatedtothegeometricdimensionsofeachspec-imen(Eq.(4)),andthestandarddeviation(60.5)oftheairvoidcontentofeachspecimen.Ifanyspecimendoesnotobeyoneofthesecriteria,itmustberejectedbeforethebeginningofthetestandreplacedwithanothersuitableunit.Ke¼ðBÀbÞ2=8ÁbÁh2Á½ððBÀbÞÁð3BÀbÞ=2B2ÞÀlnðB=bÞð4ÞwhereKeistheconstantrelatedtothegeometricdimensionsofthespecimens(mmÀ1);hisheightofthespecimen(mm);isthicknessofthespecimen(mm);bissmallbaseofthespecimen(mm);Bisthelargebaseofthespecimen(mm).ThedisplacementamplitudeappliedtothesmallbaseofthetrapezoidalspecimensiscalculatedthroughEq.(5),whichisre-latedtoeachstrainlevelchosenarbitrarilybythedesigner.f¼emax=Ke;beingtheamplitudecalculatedbyA¼2:fð5Þwherefisthehalfdisplacementamplitudeappliedtothesmallbaseofthespecimen(Â10À6);Keisconstantrelatedtothegeometricval-uesofthespecimens(mmÀ1);emaxisstrainlevelchosenbythede-signer(Â10À6);Aisthepeaktopeakdisplacementamplitudeappliedtothesmallbaseofthespecimen(Â10À6).Remark:10À6isthescientificnotationformicrostrains.Theadjustmentofthedisplacementamplitudevaluescalcu-latedthroughEq.(4)wasappliedwithanaccuracyof1.0mm,usingananalogicalextensometer,whichwasplacedinthesmallbaseofthespecimens.TwoAllenheadscrewsconnectedtoeacheccentricaxlewerethengraduallyregulateduntilthepreviouslycalculatedvaluewasreached.Thefatiguetestswereperformedonspecimensin3differentstates:(1)dry,(2)continuouslyimmersedinwater,and(3)contin-uouslyimmersedinwaterafterapreconditioningprocess.Thesearereferredtoasstates1,2and3,respectively,hereinafter.Thepreconditioningofthespecimenstestedinstate3consistedofsubmittingthemtoaspecificenvironmentalconditioningcycle:5daysimmersedinwaterat60°Cand3additionaldayswithoutwaterinaovenalsoat60°C.Priortostartingthefatiguetests,allspecimenstestedinstate3weresubmittedtoaseriesof3rep-etitionsofthementionedpreconditioningprocess,whichmeansthateachspecimenspent15days(360h)immersedinwaterand9days(216h)intheovenintotal,beforebeingtested.Furthermore,inthecaseofstates2and3,thespecimensweresubmittedtoasaturationprocessinwaterusingavacuumpres-sureof350.0mm/Hg±5.0%immediatelybeforethebeginningofthetests,inordertoachieveasaturationdegreeof60%relatedtotheairvoidcontentofeachspecimen.Forstates2and3,inwhichthespecimenswereimmersedinwaterduringthefatiguetests,anapparatuswasspecificallyde-signedfortheplacementofthewaterreservoirs(Fig.6).Thestrainlevelsofthefatiguetestswerechoseninordertoin-cludethelimitsstipulatedintheFrenchstandardizedspecification[23]inrelationtothenumberofcyclesthatthetotalamountofthespecimensusedineachentirefatiguetest(24inthisresearch)mustresist,inordertovalidatethefinalresults,i.e.,above106cy-clesforatleast1/3andnotunder104fortheothers.Regardingtheairvoidcontents,somevaluesobservedfortheasphaltmixes132B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134Fig.6.Apparatusforexecutingthefatiguetestswiththespecimensimmersedinwater[16].Table2
Strainlevelschosenforeachstateofthefatiguetestsandairvoidcontentoftheasphaltmixes[16].AsphaltmixFatigueteststatesStrainlevel(Â10À6)Setsofspecimens8CTB123123(Dry)(Immersedonly)(Preconditioned)(Dry)(Immersedonly)(Preconditioned)1201109012011090815015012015015012081801901501801901505.25.24.33.35.24.2Airvoidcontent(%)CTBPC100000000differ,thisbeingrelatedtotherandomaccommodationoftheaggregateparticlesduringthecompactingprocesswhichisverydifficulttocontrol(Table2).Thesameoptimumasphaltbindercontent(4.86%)wasob-tainedforthemixesCTBandCTBPC,basedonresearchcarriedoutbyBarra[16],whichobeyedthestandardizedcriteriaandthesequenceoftestsspecifiedbytheFrenchmethodologytoformu-latenewasphaltmixes[25],thatis,thefollowingevaluationpro-cedureswereappliedbeforeperformingthefatiguetests:gyratoryshearcompactingpress(PCG-3),Duriezat18°C,ruttingtestat60°Candcomplexmodulusimposingthesametypeofmechanicaleffortdescribedearlierforthefatiguetest.Number of cycles (N)CTBdry, slope = -7.8015, R = 0.80CTBPCdry, slope = -6.6088, R = 0.8210000000100000010000010000101001000Strain level (µm)Fig.7.FatiguecurvesfortheasphaltmixesCTBDRY(continuousline)andCTBCDRY(dottedline)[16].B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134133100000000CTBimmersed only, slope = -6.0793, R = 0.82 )NCTBPCimmersed only, slope = -4.8178, R = 0.86( se10000000lcyc fo1000000 rebm100000uN10000101001000Strain level (µm)Fig.8.FatiguecurvefortheasphaltmixesCTBIMMERSEDONLY(continuousline)andCTBPCIMMERSEDONLY(dottedline)[16].10000000)CTBpreconditioned, slope = -4.9416, R = 0.52NCTBPCpreconditioned, slope = -7.4651, R = 0.71( se10000000lcyC f1000000o rebm100000uN10000101001000Strain Level (µm)Fig.9.FatiguecurvefortheasphaltmixesCTBPRECONDITIONED(continuousline)andCTBPCPRECONDITIONED(dottedline)[16].Table3
Fatigueequationforeachasphaltmixtested[16].AsphaltmixFatigueteststateFatigueequationCTB1(Dry)N=6Â1022(ms)À7.80152(Immersedonly)N=1Â1019(ms)À6.07933(Preconditioned)N=1Â1016(ms)À4.9416CTBPC1(Dry)N=3Â1020(ms)À6.60882(Immersedonly)N=3Â1016(ms)À4.81783(Preconditioned)N=3Â1021(ms)À7.46517.ResultsandanalysisTheresultsforthefatiguetestsareshowninFigs.7–9foreachasphaltmix.Theywerecalculatedaccordingtothestatisticalpro-ceduresspecifiedbytheFrenchstandardizedspecification[23].Thelettersmsineachfatigueequation(Table3)meanmicro-strainandarerelatedtothestrainlevelscalculatedfor106loadingcycles(e6),whichisthecriterionadoptedbytheFrenchmethodol-ogytopredictthefatigueresistanceoftheasphaltmixesfordesigningpavementstructures[26].Theresultsforthemicrostrainvalues(e6)andtheirrespectiveranges(De6)for106cycles,consideringa95%statisticalconfidenceintervalaresummarizedinTable4.Table4Microstrainvalues(e6)andtheirrespectivestatisticalranges(De6)for106cycles[16].AsphaltmixFatigueteststatee6(Â10À6)De6(Â10À6)CTB1(Dry)14110.02(Immersedonly)13812.53(Preconditioned)10610.3CTBPC1(Dry)1559.22(Immersedonly)15012.43(Preconditioned)1189.1Incomparisontothetestscarriedoutinthedrystate(1),theresultspresentedfortheasphaltmixCTBinTable4indicatede-creasesof2.78%and25.36%forthee6valuesobtainedforthefati-guetestscarriedoutintheimmersedonly(2)andpreconditioned(3)states,respectively.FortheasphaltmixCTBPC,thecorrespond-ingreductionsreached3.6%and23.67%.Oncomparingthedataforthetwoasphaltmixes,CTBPCpre-sentede6values8.79%,8.01%and10.8%higherthanthosecalcu-latedforCTB,consideringstates1,2and3,respectively.Inrelationtostate2,itwasverifiedthatonlythedirectimmer-sionoftheasphaltmixesinwaterat10°C,precededbya60%airvoidvacuumsaturationprocess,ledtosmalldecreasesinthee6values.Ontheotherhand,significantdecreasesinthee6valueswereobservedwhentheasphaltmixesweresubmittedtoanenvi-ronmentalpreconditioningprocessinstate3(Table3).Thesetwodistinctbehaviorsdescribedabovecanbeexplainedbythefollowinganalysis[16]:Forfatiguetestscarriedoutforimmersedonlystate,itispossibletoconcludethatwhenthewaterisindirectcontactwiththegranu-lar–asphaltbindersystemundermildtemperatureconditions,itpresentsalowcapacityformodifyingtheasphaltbinderviscosityandalsoforbreakingthefilmthatcoversthegranularparticles.Thus,thelong-termactionofwaterasastrippingagentwouldbere-quiredtoprovokesignificantdamage.Incontrast,forfatiguetestscarriedoutforpreconditionedstate,itcanbeinferredthatwhenthewaterthatisincontactwiththegranular–asphaltbindersystemissubmittedtoanenvironmentalpreconditioningprocessthatsimulatesaseverethermalgradient,thereisconsiderablemodificationoftheasphaltbinderviscosity,ittendingtobemorefluidandhencemoresusceptibletothere-movalofthefilmthatcoversthegranularparticles(Fig.10).Furthermore,itisalsopossibletoinferthatforallteststatestheasphaltmixCTBPCpresentedabetterfatigueresistanceincompar-isontoCTB.Thistrendcanbeattributedtothefillerscomprisingmineralswithelectropositivecharacteristics(e.g.,thecalcite)adsorbing(bychemisorption)themoleculesoftheasphaltbinder.Duringthisprocessthefillerparticlesstarttoreactchemicallywiththenaphthenicacidsoftheasphaltbinders,formingthecompoundcalciumnafphthenate,whichisinsolubleinwater[27]andprotectstheasphaltbinderfilmthatcoverstheaggregateparticles.Inaddition,thegraniteparticlesandtheasphaltbindersarebothacidinnature.Therefore,thegranitepowderisnotaseffec-tiveasthelimestonepowderasanadhesiveagent,thisbeinginagreementwiththetheoreticalexplanationsgiveninSection4.Thecohesiveandinterfacialadhesionbetweentheasphaltbin-der–filler(duringtheformulationofthemastics)andtheaggre-gate–masticsystems,respectively,isdirectlyrelatedtotheFig.10.Removaloftheasphaltbinderfilmaroundthegranularparticles[16].134B.Barraetal./ConstructionandBuildingMaterials29(2012)128–134chemicaladsorption(chemisorption)oftheasphaltbinderbythefiller,associatedwithLewisacid–baseinteractions,becausewhenthesematerialscomeintocontactseveralchemicalbondsareformed,similartoionicbonds,whichareverystrong,includingtheformationofnewcompounds(e.g.,thecalciumnaphtenate).Hence,themineralogicalandchemicalcharacteristicsdescribedabovecanexplainthebetteradhesionofthelimestonepowdertothegranular–asphaltbindersystemscomparedwiththegranitepowder,contributingtoanincreaseinthefatigueresistanceoftheasphaltmixesinthedryandenvironmental-conditioningstates.Itisalsoimportanttoemphasizethattheairvoidcontentplaysanimportantroleinthefatiguebehavioroftheasphaltmixes.However,basedontheresultsofthisresearch,itcanbeconcludedthatthisaspectdidnotcontributeformodifyingthetrend(Table4),thatis,thatabetterfatigueresistancewasachievedfortheas-phaltmixCTBPC,eventhoughsomedifferencewereobservedregardingtheairvoidcontentofthemixestestedinthedrystate(Table2).Finally,theslopesofthefatiguecurvesfortheasphaltmixCTBPCweremuchhigherthanthoseforCTB(Figs.7–9),whichisanotherindicationofthebetterfatiguebehavioroftheformer,consideringthenumberofcyclesresistedbybothmixesforthesamerangeofstrainlevelssimulatedineachspecificteststate.8.ConclusionsTheresultsobtainedinthefatiguetestsindicatedthattheac-tionofthewaterhasadecisiveinfluenceonthefatigueresistanceofasphaltmixes,andcanleadthemtorupturebeforetheendofthepredictedtermforwhichtheyweredesigned.Hence,thisphe-nomenonmustbetakenintoaccountwhendesigningasphaltpavementstructures.Furthermore,thedamagecausedbytheactionofwaterismoreintensewithanincreaseintemperature,contributingtothere-movaloftheasphaltbinderfilmthatcoversthegranularparticles.Despitetheimportantrolethattheairvoidcontentplaysinthefatigueresistanceofasphaltmixes,thisvariabledidnothaveanynotableinfluenceonthetrendobservedintheresultsobtainedinthisstudy,i.e.,thebetterfatigueresistanceoftheasphaltmixCTBPC,whichwasformulatedwithlimestonepowderasafiller.Regardingthefillerfractions,thereplacementofgranitepow-derwithlimestonepowderseemstobeasuitablesolutiontoim-provetheadhesivebehaviorofthegranular–asphaltbindersystemsand,consequently,thefatigueresistanceoftheasphaltmixes,consideringthatbetterresultswereobtainedwiththeas-phaltmixCTBPCincomparisontoCTB.Thisbetterperformanceofthelimestonepowderisdirectlyrelatedtothemineralogicalandelectropositivechemicalcharacteristicsofthematerial.References[1]DomecV.Endommagementparfatiguedesenrobésbitumineuxenconditiondetraficsimuléetdetempérature.ThèsedeDocteur.UniversitédeBordeauxI.Bordeaux,France;2005.[2]RudenskyAV.Asphaltconcretefatigueproperties.In:Proceedingsofthe5thinternationalRILEMsymposium.In:DiBenedettoH,FranckenL,editors.Mechanicaltestsforbituminousmaterials(MBTM):recentimprovementsandfutureprospects,Lyon;1997.[3]HuetC.ÉtudeparuneMéthoded’ImpédanceduComportementViscoélastiquedesMatériauxHydrocarbonés.ThèsedeDocteur.FacultédesSciencesdel’UniversitédeParis;1963.[4]LaveissièreD.ModélisationdelaRemontéedeFissureenFatiguedanslesStructuresRoutièresparEndommagementetMacro-Fissuration:deL’ExpérimentationaL’OutildeDimensionnementpourL’EstimationdelaDuréedeVie.ThèsedeDocteur.FacultédesSciencesdel’UniversitédeLimoges.Limoges,France;2002.[5]PerretJ.DéformationdesCouchesBitumineusesauPassaged’UneChargedeTrafic.ThèsedeDoctorat.ÉcolePolytechniqueFédéraledeLausanne(EPFL).Lausanne.Lausanne,Suisse;2003.[6]BaajH.Comportementalafatiguedesmatériauxgranulairestraitésauxliantshydrocarbonés.ThèsedeDocteur.InstitutNationaldesSciencesAppliquéesdeLyon.Lyon,France;2002.[7]DeLaRocheC,OdeonH.ExpérimentationUSAP/LCPC/Shell–FatiguedesEnrobés-Phase1-RapportdeSynthèse,editedbyLCPC,sujetno.2.01.10.4,France;1993.[8]BodinD.Modèled’EndommagementCyclique:ApplicationàlaFatiguedesEnrobésBitumineux.ThèsedeDoctorat.ÉcoleDoctoraleMécaniqueThermiqueetGénieCivil,etÉcoleCentral.France;2002.[9]DeLaRocheC.,ModuledeRigiditéetComportementenFatiguedesEnrobésBitumineux.ThèsedeDoctorat.ÉcoleCentraldeParis;1996.[10]Inpe,InstitutoNacionaldePesquisasEspaciais.MinistériodaCiênciaeTecnologia(MCT). 因篇幅问题不能全部显示，请点此查看更多更全内容