NCATReport17-04
HIGH-MODULUSASPHALTCONCRETE(HMAC)
MIXTURESFORUSEASBASECOURSE
By
FabricioLeiva-VillacortaAdamTaylorRichardWillis
June2017
ii
HIGH-MODULUSASPHALTCONCRETE(HMAC)MIXTURESFORUSEASBASECOURSE
NCATReport17-04
By:
FabricioLeiva-Villacorta,PhDAssistantResearchProfessor
NationalCenterforAsphaltTechnologyatAuburnUniversity
AdamTaylor,P.E.AssistantResearchEngineer
NationalCenterforAsphaltTechnologyatAuburnUniversity
RichardWillis,PhDDirectorofPavementEngineering&Innovation
NationalAsphaltPavementAssociation
SponsoredbyFederalHighwayAdministration
June2017
iii
ACKNOWLEDGEMENTS
ThisprojectwasfundedbytheFederalHighwayAdministration(FHWA).Theauthorswouldliketothankthemanypersonnelwhocontributedtothecoordinationandaccomplishmentoftheworkpresentedherein.
DISCLAIMER
Thecontentsofthisreportreflecttheviewsoftheauthors,whoareresponsibleforthefactsandaccuracyofthedatapresentedherein.Thecontentsdonotnecessarilyreflecttheofficialviewsorpoliciesofthesponsor(s),theNationalCenterforAsphaltTechnology,orAuburnUniversity.Thisreportdoesnotconstituteastandard,specification,orregulation.Commentscontained inthispaper related to specific testing equipment and materials should not be considered anendorsementofanycommercialproductorservice;nosuchendorsementisintendedorimplied.
iv
TABLEOFCONTENTSAbstract............................................................................................................................................51. Introduction..............................................................................................................................62. Objective...................................................................................................................................73. State-of-the-practice.................................................................................................................73.1 MixtureDesign.................................................................................................................83.1.1 AggregateSelection......................................................................................................93.1.2 DesigningaGradation................................................................................................113.1.3 BinderSelectionandRichnessFactor.........................................................................123.1.4 PerformanceTests......................................................................................................16
3.2 PavementDesign............................................................................................................213.3 Performance...................................................................................................................233.4 Construction...................................................................................................................293.5 SummaryofCurrentPractice.........................................................................................29
4. ExperimentalDesignandAnalysis..........................................................................................304.1 LaboratoryTesting..........................................................................................................304.2 MixtureDesign...............................................................................................................314.3 DynamicModulus...........................................................................................................334.4 FlowNumber..................................................................................................................384.5 AMPTCyclicFatigue.......................................................................................................41
5. AASHTOWarePavementMEDesignAnalysis.........................................................................475.1 Traffic.............................................................................................................................485.2 Climate...........................................................................................................................495.3 EstimatedPerformance..................................................................................................50
6. ConclusionsandRecommendations.......................................................................................567. RecommendedMixtureDesignProcedure.............................................................................57References......................................................................................................................................58
Leiva-Villacorta,Taylor,andWillis
5
ABSTRACT
Recentstudiesonlong-lifeflexiblepavements indicatethat itmaybeadvantageoustodesignandconstructasphaltmixturescomprisingtheunderlyinglayersinsuchamannerthatveryhighmodulus mixtures are produced. The French have been experimenting with and designingpavements with high-modulus bases since the 1980s. This study considered the engineeringpropertiesofasphaltmixturesproducedusingaEuropeanspecificationforhigh-modulusasphaltconcrete (HMAC) mixtures and used as base course. This specification includes volumetricrequirements such as asphalt content and air voids, but there are also requirements forengineering parameters that address performance requirements such as rutting and fatiguecracking. The purpose of this studywas to investigate the design of asphaltmixtures havinghighermodulus.Thestudywaslimitedtoalaboratoryperformanceevaluationandatheoreticalmodelingcomponentwheretheresultswereusedtoindicatepotentialfieldperformance.
Acomprehensiveliteraturestudywasperformedtoassessthecurrentstate-of-the-practiceonHMACmixturedesign,pavementdesign,laboratoryperformancetests,andfullscalepavementperformance.Theexperimentalplanincludedavarietyofmixtureswithdifferentmaterialandbinders such that highermoduliwere obtained compare to conventionalmixtures. The planincludedaFrenchmixturewithastiffbinder(PG88-16),twomixturescontaining35%RAPbothwith polymer-modified binders but one high polymer content (HiMA), another mixturecontaining25%RAPand5%RASwithapolymer-modifiedbinder,andfinally,a50%RAPmixturewithapolymer-modifiedbinder.Thelaboratorytestingprogramevaluatedbinderperformancegrade, mixture stiffness over a wide temperature range, fatigue cracking, and permanentdeformation.Inaddition,AASHTOWarePavementMEDesignsoftwarewasusedtodeterminehowahigh-modulusbasewouldaffectpredictedperformanceofasphaltpavements.
TheresultsofthisstudyindicatedthatEuropeanmixdesignstandardmethodsandspecificationswere successfully implemented on local (U.S.) virgin and recycled materials. In addition,increased stiffness of high modulus mixtures improves mechanistic-empirical predictedperformance of pavement in rutting, fatigue cracking, and ride quality. However, it wasdetermined thatperformanceofnewmaterials cannotbe reliablymodelledwith thecurrenttransferfunctionsandfurtherfieldvalidationisrequired.
Leiva-Villacorta,Taylor,andWillis
6
1. INTRODUCTION
Currently in the United States, asphalt paving mixtures are primarily designed using theSuperpave system where the proportioning of components relies mainly on volumetricproperties. Early Superpave implementation focused primarily on rutting resistance.Mixturedesignsformoderateandhightrafficpavementsweredesignedforimprovedruttingresistanceby specifying a higher grade of asphalt binder and higher quality aggregates.Most highwayagenciesnowreportthatruttingproblemshavebeenvirtuallyeliminated.However,therehavebeengrowingconcernsthattheprimarymodeofdistressforasphaltpavementsiscrackingofsome formor another. There are several possible contributing factors to increased cracking,including issueswithmixturedesigns, increaseduseof recycledmaterials,problemswith thequalityofconstruction,andfailuretoadequatelyaddressunderlyingpavementdistressesduringpavement rehabilitation. It isnowwell recognized thatcurrentmixturedesignpracticeshavesomeshortcomings.
Moststatedepartmentsoftransportation(DOTs)currentlyutilizevolumetriccriteriaforasphaltmixturedesignsthatfollowtheSuperpavemixturedesignmethodsofAASHTOM323andR35withsomemodifications.Inresponsetopavementdurabilityissues,manyDOTshavemodifiedtheir design and acceptance requirements to obtain more durable and high crack-resistantmixtures by increasing the asphalt content of the lower layer of hot-mix asphalt, commonlyreferred toas rich-bottommixtures.Rich-bottommixturesaremadewith the samegradeofasphaltbinderbutaredesignedatalowerairvoidcontentsoastoincreasethedesignasphaltcontentby0.6%to0.8%.
ThisstudyconsideredtheengineeringpropertiesofasphaltmixturesproducedusingaEuropeanspecification for high-modulus asphalt mixtures and used as base course. This specificationincludes volumetric requirements such as asphalt content and air voids, but there are alsorequirements for engineering parameters that address performance requirements such asrutting and fatigue cracking. High-modulus asphalt is routinely produced with hard asphaltbinders,PG88orhigher,forcriticalhightemperatureproperties.Inthisstudy,aneffectivehardasphaltbinderwasobtainedbycombiningpolymer-modifiedasphaltwithseveralcontentsofRAP(between25%and50%)andutilizedinahigh-modulusmixtureproducedfollowingFrenchstandardprocedures.
In the1980s, theFrenchPublicWorksResearch Instituteor LaboratoireCentraldesPontsetChaussées(LCPC)developedhigh-modulusmixtures,referredtoasEnrobéàModuleÉlevé(EME).Theobjectiveforthistypeofnewmixturewasimprovedmechanicalpropertiestoincludehigh-modulus,goodfatiguebehavior,andexcellentresistancetorutting.Highstiffnessandimprovedfatigue resistanceallowadecrease inpavement thickness forbothnewconstructionand forrehabilitation.OnegoaloftheEMEdeveloperswastoreducegeometricconstraints(overheadclearanceconstraints)duringrehabilitation.EarlytrialsofEMEoccurredinthemid-1970sandbytheearly1980shaddevelopedintoanewmixturetype.Aspecificationwassetbytheearly1990sandinthelate1990sthemixturehadbecomepartofthestandardcatalogofmixturesusedin
Leiva-Villacorta,Taylor,andWillis
7
pavementstructuraldesign forhigh trafficpavements,20millionequivalent singleaxle loads(ESALs)orgreater.
IntheUnitedStates,agenciesarebeginningtoadopttheMechanisticEmpiricalPavementDesignGuide (MEPDG) fordeterminingpavementstructural thickness.Unlike thepreviousstructuraldesignmethod,typicallyAASHTO1993,whichonlyconsideredmixturepropertiesindirectly,theMEPDGmakesdirectuseofmixturepropertiesaspartofthestructuralpavementdesign.Thisstudy compares an MEPDG pavement structural design using only Superpave mixtures andseveralhigh-modulusmixturesforthebasecourse.
2. OBJECTIVE
Theobjectiveofthisprojectwastoevaluatethecurrentmixturedesignmethodologyofhigh-modulus base layers and evaluate potential effects on performance. In order to successfullyanalyzetheeffectofhigh-modulusmixturesasbasecourses,thefollowingtaskswerecompletedonthisproject:
1. Literaturereview:A literaturestudywasperformedtoassessthecurrentstate-of-the-practice.This included information frompublished journals, technical reports,articles,presentations,aswellaspersonalcommunicationsandinterviewswithcontractorsandagenciesthathavesuccessfullyimplementedhigh-modulusmixturesintheirpavementstructures.
2. Providerecommendationsofmixturedesignandstructuraldesignprocedures:Dataandresourcesgatheredduringtheliteraturereviewwereusedtodevelopmaterialselectionandmixturedesignproceduresforahigh-modulusmixturethatwouldberesistanttothetensilestrainsatthebottomoftheasphaltpavementstructure.
3. Laboratorystudy:Engineeringpropertiesofhigh-modulusmixturesweredeterminedanddesignprocedureswereassessedusinglaboratoryperformancetests.
4. Pavementdesignandanalysis:AASHTOWarePavementMEDesignsoftwarewasusedtodetermine how a high-modulus base would affect predicted performance of asphaltpavements.
3. STATE-OF-THE-PRACTICE
High-modulusasphaltconcrete(HMAC)wasoriginallydevelopedinthe1980satatimewhenFrancewas looking to design high performance asphaltmixtures to increase the life span ofconventional asphalt pavements or reduce the necessary thickness required to carry theincreasingloadsseenonEuropeanhighways.Althoughthesemixturesweredesignedtoserveaseitherasphaltbaseorbindercourses,theywereeventuallyalsousedinwearingcoursesinthemid-1980s,buttheseareoutsideofthescopeofthisreport(EAPA,2005;Nkgapeleetal.,2012;Corte,2001).
Inthe1990s,theFrenchdevelopedastandardforEMEmixtures(Denneman,2011;PethoandDenneman,2013). This standardhad twoclassesof EMEmixtures.Class1wasa low fatigueresistancemixturewhile Class 2was a higher fatigue resistantmixture. Themain difference
Leiva-Villacorta,Taylor,andWillis
8
betweenthesetwoclasseswasthebindercontentofthemixtures.In2007,aEuropeanStandard(EN13108-20)wasdeveloped(Brosseaud,2012;Guyot2013).
Todate,numerousEuropeancountriessuchasFrance,Austria,theCzechRepublic,Denmark,Ireland,Italy,Netherlands,Portugal,andtheUnitedKingdomhaveallhadpositiveexperiencesusingthismaterial;however,eachcountryhasaslightlydifferentapproachtomixturedesignand performance criteria, as expected. Generally, these mixtures have been successfullyincorporated at times when low quality aggregates are available to reinforce the mixtures,industrialareasaresubjectedtoheavyloads,andwhenexistingpavementsneedtobereinforcedduring rehabilitation or reconstruction (Brosseaud, 2012; EAPA, 2005; Petho andDenneman,2013). Case studies at airports have also been conducted to improve runway and taxiwaydurability(EAPA,2003;Guyot,2013).
3.1 MixtureDesign
Aswithmostasphaltmixtures,asphaltandaggregatearethetwoprimaryconstituentsusedinHMAC.However,unlikeMarshallmixturedesignorSuperpavemixturedesign,themixdesignisnotdrivenbyvolumetricpropertiesasmuchasitisdrivenbytryingtopassperformance-basedspecifications. Thismethodofmixture design is actually developed to assess performance inrelationtotheloadingandenvironmentalconditionsthemixturemayexperience.Thistypeofdesign methodology reduces barriers to innovation, promotes mixture performance, andencouragestheefficientuseofresources(Dennemanetal.,2011).Figure1providesaflowchartofthebasicprocessofdevelopinganHMACmixture.Thissectionoftheliteraturereviewwillexplaineachportionoftheflowchart.SincetheFrenchhavethemostexperiencewithdevelopingHMACmixtures,thisliteraturereviewwillfollowtheFrenchmethodandshowhowSouthAfricahastakentheEuropeanstandardandadoptedittofollowASTMandAASHTOmethods.
Leiva-Villacorta,Taylor,andWillis
9
Figure1HMACMixtureDesignProcess(Dennemanetal.,2012)
3.1.1 AggregateSelection
InmanyEuropean countries, SouthAfrica, and inAustralia,HMAC incorporates fully crushedaggregateduetotheimportanceofbothsurfaceareaandtexture.Thesetwopropertiesaidinincreasingthevoidsinthemineralaggregate(VMA),whichmustbesufficienttoaccommodatethe higher asphalt binder content in themixtures. Aggregate selection guidelines have beendevelopedtoaidinproperlychoosingtheskeletalstructure.Table1showsanexampleofHMACaggregateselectioncriteriainSouthAfrica.
Leiva-Villacorta,Taylor,andWillis
10
Table1HMACAggregateSelectionCriteria(Dennemanetal.,2011)
Property Test Method Criteria
Hardness Finesaggregatecrushingtest:10%FACT TMH1,B1 ≥160kNAggregatecrushingvalueACV TMH1,B1 ≤25%
Particleshape&texture
Flakinessindextest SANS3001 ≤25%
Particleindextest ASTMD3398 >1510-15(Delorme,2007)
Waterabsorption
Waterabsorptioncoarseaggregate(>4.75mm) TMH1,B14 ≤1.0%
Waterabsorptionfineaggregate TMH1,B14 ≤1.5%Cleanliness Sandequivalencytest TMH1,B19 ≥50
While these criteria were developed to ensure high quality aggregates are used in HMAC,countriessuchasLatviawerenotpermittedtousedolomiteaggregateinhighmodulusmixtures.Asmall-scaleresearchprojectwasconductedusinglocallyavailableaggregateinLatviatoseeifeitherusingapolymerorhardbinder inconjunctionwithdolomiticaggregatemightequalorbettertheperformanceofareferencemixturethatrepresentsatypicalmixture(Haritonovsetal.,2014).
MixturesweredesignedusingtheMarshallmixturedesignmethodandsubjectedtoTSR,rutting,and fatigue performance testing. The reference mixture was not designed using the HMACmethodology;thus,ithadalowerasphaltcontent.WhencomparedtotheHMACmixtures,thereferencemixtureperformedbetterwithrespecttorutting,butthelowerbindercontentcausedreducedfatiguecapacity.TheHMACwithapolymer-modifiedbinderperformedbetterthanthehardbinderinrutting.Overall,thestudyresultsshowedthatusingalocalaggregatethatmightnot be considered high quality might be acceptable in an HMAC mixture if specificationrequirementsarestillmet(Table2)(Haritonovsetal.,2014).
Table2CompliancewithSustainablePavementforEuropeanNewMemberStates(SPENS)Requirements(Haritonovsetal.,2014)
ParameterMixtures
PMB10/45-65 B20/30 RequirementHMAC-1/1 HMAC-1/2 HMAC-2/2 HMAC-2/3Voidcontent,% 3.9 3.7 3.9 3.7 3.0–5.0
Rutresistance,mm/1000cycles 0.04 0.04 0.14 0.22 0.03-0.25Stiffness(10°C,10Hz)MPa 16700 16100 17100 17900 Min14000
Fatigue(10°C,10Hz),µmm/mm 130 130 130 130 Min.130at1millionCycles
WaterSensitivity,TSR,% 100 100 98 94 TSR80
DuringanHMACimplementationstudythatconsistedoftestingHMACwithoff-scalelimestone,granite, crushed cobblestone, steel slag, and basalt, it was determined that lower qualityaggregatescouldbeusedbecauseHMACmixturedesignmovesaway fromempiricalmixture
Leiva-Villacorta,Taylor,andWillis
11
designandprogressestocompetentmixturedesign.Thisprovidestheopportunitytopairweakeror lower quality materials with higher quality materials to ensure pavement performance(Bankowskietal.,2009).
AcommonconcernwithHMACrelatestothe inclusionofreclaimedasphaltpavement(RAP).Earlyexperimentscontaining30%RAPshowedissueswithcompaction,resistancetowater,andfatiguedamage(DesCroix,2004).Sincethattime,otherlimitedlaboratorystudieshaveshownthatRAPcanbeincludedupto40%,and25%RAPhasbeenincludedinfieldstudieswithsuccess.However,muchofthisresearchcaveatstheconclusionsbystatingthatnotallbindersandRAPsources are equivalent, and thismust be evaluated on amixture bymixture basis to ensureperformanceofthemixture,whichistheprimarygoal(deVisscheretal.,2008;Buecheetal.,2008).Additionally,pastworkhasshownthatimpropercharacterizationofRAPmightinfluencelowfieldbindercontentscomparedtotargetcontentsduringproduction(Nkgapeleetal.,2012).
Onestudyrecentlytestedmixturesthathadbeenplacedinthefieldusing0%,50%,and65%RAP.Whenthemixturewastestedinthelaboratory,alloftheperformancerequirementsforHMACmixturesweremetevenathigherRAPcontents.Whilethisworkwaspreliminary,itdidsuggest that RAP could be used to produce an EMEmixture; however, like any high recyclemixture, it is important to have good homogeneity, control, and material characterization(Brosseaudetal.,2012).
Another study comparedHMACmixtureswith0, 15, 30, and50%RAP. Thesemixtureswereanalyzedmechanicallyfortoughness(Fenixtest),stiffness,rutting,moisturedamage,andfatigueresistance.ThestudyconcludedthatincreasingRAPcontentusingasofterbinderdidnothaveanegativeimpactonmechanicalmixtureperformance;however,theresearchstatedthatplantlogisticsmaymakereaching50%RAPunattainable.Whenthemixtureswereattemptedatalocalbatchplant,the50%RAPmixturecouldnotbeproducedbecausethenon-heatedRAPwouldnotthoroughlymixturewith theheatedaggregateandbitumen formaterial transferof theagedbinderontheRAP(lowRAPbinderactivation);thus,plantconsiderationsmaylimitRAPuseinHMACsimilartostandardasphaltmixtures(Miroetal.,2011).
3.1.2 DesigningaGradation
Whenconsideringthegradationrequirementsusedinnumerouscountries,deviationsoccur.Forexample,thegradingenvelopesinFrance(Delormeetal.,2007)aredifferentthanthoseoftheUnitedKingdom(SandersandNunn,2005).OtherchallengesindirectlytranslatingspecificationsfromtheFrenchversionsaredifferencesinnomenclatureandEuropeansievesizing.
SouthAfricadevelopeda table for targeted grading curves that includedboth Europeanandmetric sieve sizes (Table 3). This allowed development of gradations like those in Europe;however, theycouldusetheirownequipment.Additionally,oneshouldnotethat theFrenchdesignswerebasedon themaximum sieve sizewith requiring 100%passing at 2D, 98-100%passingat1.4D,and85-98%passatD(D=maxsievesize).
Leiva-Villacorta,Taylor,andWillis
12
Table3TargetGradingCurvesandEnvelopesforHMACBaseCourses(Dennemanetal.,2011)
PercentPassingSieveSize
D=10mm D=14mm D=20mmMin Target Max Min Target Max Min Target Max
6.7mm 47 56 68 52 54 72 46 54 666.3mm 45 55 65 50 53 70 45 53 654.75mm - 53 - 43 49 63 42 49 624.0mm - 52 - 40 47 60 40 47 602.36mm 32 36 44 28 26 42 28 36 422.0mm 28 33 38 25 33 38 25 33 38
0.075mm 6.4 6.9 7.4 5.5 6.9 7.9 5.5 6.7 7.90.063mm 6.3 6.7 7.2 5.4 6.7 7.7 5.4 5.7 7.7
D=maxsievesize
3.1.3 BinderSelectionandRichnessFactor
EuropeanStandardEN13924governsbinderselectionforHMACmixtures.Typically,10/25or15/25penbindershavebeenusedinEurope(Dennemanetal.,2011).Inthe1980s,Francebegandesigningandproducingthehardbinderneededforthesemixtures.In1990,Franceproduced39,000tonsofthebinder.By2000,thatvaluehadgrowntoover100,000tons.Thebinderwasoriginallydevelopedthroughanairblowingprocess;however,thisincreasedthebrittlenessofthe binders and made them more susceptible to fatigue cracking. Since that time, vacuumdistillationandpropane-precipitatedasphalthasbeenusedtoproducethestiffasphaltneeded(Corte,2003).Mostofthesebindershadpenetrationsbetween10and30andsofteningpointsgreater than60or70°C (EAPA,2005).Examplesof typicalhardasphalt characteristicsbeforeagingaregiveninTable4.RheologicalpropertiesofcommonasphaltsusedinHMACareprovidedinTable5.
Table4TypicalHardAsphaltCharacteristics(BeforeAging)(Corte,2001)
PenetrationGrade 15/25 10/20 5/10R&BSofteningpoint(°C) 66 62to72 87PfeifferIP(PenetrationIndex) +0.2 +0.5 +1.0DynamicViscosityat170°C(mm2/sec) 420 700 980ComplexModulusat7.8Hz,IE*I,(MPa)
0°C10°C20°C60°C
425180700.4
7003001100.7
9805703007
Leiva-Villacorta,Taylor,andWillis
13
Table5RheologicalCharacteristicsofSeven10/20Asphaltsanda35/50Asphalt(Corte,2001)
Itshouldbenotedthatthishardbinderisnotavailableinalllocations(Dennemanetal.,2011)andmight require innovative bindermodification techniques to achieve the desired results.When Korea experimented with HMAC, they mixed a high boiling point petroleum with aconventionalasphaltto increasebinderstiffness.Atthatpoint,4%styrene-butadiene-styrenewasintroducedsothatthebindercouldmaintainsomeductility.Despitetheadditionofpolymer,thebinderdevelopedfortheHMACwasstillmorebrittleatlowtemperaturesthanconventionalorpolymer-modifiedasphaltwhendeterminingFrass temperature (Leeetal.,2007).Thermaldistress is not common inmany European countries; however, this becomes amore criticalpropertytomonitorasHMACisadoptedincountrieswithcolderclimates(EAPA,2005).
Today,polymer-modifiedasphalts (PMA)arepartof theavailablebinderselection forHMAC.Polymer-modifiedasphaltsresistruttinginsummermonthsandprovideflexibilitytoresisttensilestresses.SomeresearchsuggeststhateventheincorporationoffibersaspartofthePMAcanprovideadditionaldurabilitytoreducecracking;however,additional informationisneededtoensurethesefindings(Montanelli,2013).
AsimilarexperimenttotheKoreanstudywasconductedinLithuaniathatcomparedHMACwithcrushed granite, crushed dolomite, and crushed gravel. Additionally, two polymer-modifiedbindersandatraditionalHMAChardbinderwereusedtogivethestudyninemixtureiterations.Mixturesweredesignedandtestedforstiffnessmodulus(LSTEN12697-26),resistancetorutting(LSTEN12697-22), fatigueresistance four-pointbending (LSTEN12697-24),andstabilityandflow(LSTEN12697-34).Whilesomedifferenceswereseenbetweenaggregatetypes,thetypeand amount of binder made the most difference. The study recommended that polymer-modifiedbindersberecommendedforHMACbasemixturesandonlypolymer-modifiedbindersbeusedinthebinderlayers.Thisrecommendationwasbasedoffoffatigueresults,whichshowedthatthemixtureswithpolymer-modifiedbinders(PMB)performedbetterthanthosewithoutpolymer(Vaitkus,2013).
Chappatet al., (2009)proposed similar findings thatbinder source is a crucial componentofproducingahighperformingHMAC.Whencomparingmodulusvalues to fatigue strength formixturesusingdifferentbindersources,theresearchteamcouldseetendenciesforsomebinderstooutperformothers.Forexample,inFigure2,oneseesthatfromsourceA,themixturewillgetthemodulusbuthavetroubleattainingthefatiguerequirements.Ontheotherhand,sourceB
Leiva-Villacorta,Taylor,andWillis
14
produces mixtures that are too soft but have adequate fatigue performance. Knowing thematerialsisvitalforproducingmixturesthatwillbejudgedonperformance.
Figure2CombinedPresentationoftheModulusandFatigueStrengthResultsAccordingto
BinderSource(Chappatetal.,2009)
Increasedbinderrigidityiscommonlybalancedbyusinghigherbindercontentsinthemixtures(Guyot,2013).Minimumbindercontentsaregivenbasedontheclassofthemixtureandthemaximum aggregate size. The French have seen a continual increase in binder content andincreased binder stiffness to help produce high-modulusmixtures since the 1970s (Figure 3)(Distinetal.,2006).
Figure3EvolutionofBaseCourseMixturesinFrance(Distinetal.,2006)
Leiva-Villacorta,Taylor,andWillis
15
TheClass1mixtureislessfatigueresistantanddesignedforlowertrafficvolumeswhiletheClass2mixtureisdesignedforhighervolumeswithadditionalresistancetofatigue.Table6providestheminimumbindercontentsbasedonaggregatedensity (ρ),class,andmaximumaggregatesize. Binder content is calculated not through volumetric properties like in the U.S., but bycalculatinga richness factor,K.However, theAsphalt Institutebinder film thicknessequationseemsbetterbecauseisbasedontheactualmeasureofasphaltabsorption;ieeffectivebinderfilmthickness.Thisfactoriscalculatedthroughthefollowingmethod(Dennemanetal.,2011;DennemanandNkagdme2011).
1. Calculatethespecificsurfaceareaoftheaggregate(Σ)usingEquation1.
𝟏𝟎𝟎𝚺 = 𝟎. 𝟐𝟓𝑮 + 𝟐. 𝟑𝑺 + 𝟏𝟐𝒔 + 𝟏𝟓𝟎𝒇 (1)
where
G = proportionofaggregateretainedonandabovethe6.3mmsieve;S = proportionofaggregateretainedbetweenthe0.25mmand6.3mmsieves;s = proportionofaggregateretainedbetweenthe0.063and0.25mmsieves;andf = percentpassingthe0.063mmsieve.
2. Calculateacorrectioncoefficient(α)fortherelativedensityoftheaggregate(RDA)usingEquation2(inthiscaseRDA=Gse).
𝜶 = 𝟐.𝟔𝟓𝑹𝑫𝑨
(2)
3. CalculatethebindercontentofthemixturebymassoftotalaggregateusingEquation3.
𝑻𝑳𝒆𝒔𝒕 = 𝑲𝜶 𝚺𝟓 (3)
4. CalculatethepercentbinderbymassoftotalmixtureusingEquation4.
𝑻𝑳𝒆𝒔𝒕 =𝟏𝟎𝟎𝑷𝒃
(𝟏𝟎𝟎;𝑷𝒃) (4)
Table6TypicalValuesforMinimumBinderContentandTargetRichnessFactor(Dennemanetal.,2011)
HMACBaseCourseClass1 Class2
D(mm) 10,14,20 10,14 20Pbminρ=2.65g/cm3 3.8 5.1 5.0Pbminρ=2.75g/cm3 3.8 4.9 4.9Richnessfactor,K 2.5 3.4 3.4
Leiva-Villacorta,Taylor,andWillis
16
3.1.4 PerformanceTests
Oncethebindercontentisdetermined,thefinalphaseofmixturedesignistoundergoaseriesofperformanceteststoensurethemixturewillbedurableinthefield.TheFrenchsuiteoftestsrevolvesaroundfivestandards(Table7).
Table7FrenchPerformanceTests(ModifiedfromDennemanetal.,2011)
Parameter FrenchTestMethodWorkability EN12697-31:GyratoryCompactorDurability EN12697-12:Durieztest
PermanentDeformation EN12697-12:WheelTrackerDynamicModulus EN12697-26:Flexuralbeam
Fatiguetest EN12697-24:Prism
Theworkabilityoftheasphaltmixturesisassessedbyensuringthatthemixturehaslessthanthemaximumvoidcontentafter100gyrations in theEuropeangyratorycompactor.SouthAfricaconductedastudyusingtheSuperpavegyratorycompactortoassesswhatdeviationsoccurwhenswitchingfromtestingusingtheEuropeanstandardtotheSuperpavemethod.Thestudyshowedthat theequivalent gyrations andaverageair voidswere reduced for the Superpavemethod(Figure4andTable8).
Figure4GyratoryCompactionCurvesforTwoMixturesusingEuropeanandSuperpave
Configuration(Dennemanetal.,2011)
Leiva-Villacorta,Taylor,andWillis
17
Table8SummaryofGyratoryCompactionStudy(Dennemanetal.,2011)
HMACDesign Class1 Class2Specification EN12697-31 Superpave EN12697-31 SuperpaveNumberofSpecimens 9 8 5 5AverageVoids(%)after100Gyrations 4.8 3.0 3.5 2.2
StandardDeviation 0.8 0.9 0.8 0.4CoefficientofVariation(%) 16.0 30.0 22.1 19.8EquivalentGyrations 100 43 100 46
TheDurieztest is theFrenchequivalentofAASHTOT283formoisturesusceptibility. InSouthAfrica,amodifiedLotmanntest(ASTMD4867)isusedtoassessdurability.Mostcountriesthatuse HMAC do not differentiate tensile strength ratio (TSR) requirements between asphaltmixturesandHMAC(Dennemanetal.,2011).
EN 12697-22 is the standard in Europe for assessing rutting resistance ofmixtures bywheeltrackingonanasphaltslab.Thewheeltrackingapparatusconsistsofaloadedwheel,whichbearsonasampleheldonamovingtable.Thetablereciprocateswithsimpleharmonicmotionthroughadistanceof230±5mmwithafrequencyof53passes(±1%)perminute.Forresearchpurposes,thetestspeedcanbeadjustedbyinvertercontrol.Thewheelisfittedwithasolidrubbertireofoutsidediameter200mm.Thewheelloadunderstandardconditionsis700±10N.Thewheeltracker is fitted with a temperature controlled cabinet with a temperature range fromenvironmentto65°C±1.0°C.Thesamplemaybeeithera200mmdiametercoreora300x400mmslabofasphalticmixturefrom25mmto100mmthick.A25mmstrokeLVDTtransducerisincludedformonitoringrutdepthinthecenterofasampleduringatesttobetterthan0.1mm.ThedeformationandsampletemperatureisrecordedbytheinternaldataacquisitionandcontrolsystemandisthensenttotheWindows®compatiblesoftware.
OthercountriesusetheirownstandardruttingtestssuchasAASHTOT320,theRepeatedSimpleShearTestataConstantHeight.Researchhasconsistentlyshownthatdespitethehigherasphaltcontent inHMAC,thestifferbinderallowsthemixturetoresistruttingmorethanastandardbituminousbasematerial(BTB)(Figure5)(Dennemanetal.,2011).Thisisanespeciallyimportantconsideration inwarmer climatesdue to theextrabinder,which adds to the richnessof themixture(CapitãoandPicado-Santos,2006).
Leiva-Villacorta,Taylor,andWillis
18
Figure5PermanentDeformationHMACComparedtoBTB(Dennemanetal.,2011)
InEurope,dynamicmodulusismeasuredusingstandardEN12697-26(equivalenttoAASHTOTP62)wherethestiffnessoftheasphaltmixturebeamisdetermined(AASHTOT321).HMACmustthenhave a stiffness greater than 14,000MPa at 10°C and a frequencyof 10Hz.While justensuringthatthemixturecansurpassthisstiffnessatonetemperatureprovidesaneasycheck,it limits the data available to the practitioner. Additionally, choosing 10°C might be anappropriatecheckinthetemperateclimatesofEurope;however,othercountriessuchasSouthAfricahavedecidedtouseahighertemperatureof15°Candstillrequireastiffnessgreaterthan14,000MPaduetohotterclimates.SouthAfricahasalso implementedAASHTOTP62as themethodforascertainingmixturestiffness,whichismorecomparabletowhatmightbedoneifHMACwastobecomeadesignmethodologyintheU.S.(Dennemanetal.,2011).Whilechoosingonetemperatureandfrequencytocheckthestiffness,Figure6showsthatHMAC(EME)mixturesare typically stiffer than conventional mixtures across the temperature-frequency spectrum(PethoandDenneman,2013).
Leiva-Villacorta,Taylor,andWillis
19
Figure6DynamicModulusofHigh-ModulusComparedtoCommonEuropeanMixtures(Petho
andDenneman,2013)
ThefinalperformancetestforHMACmixtures is fatigue.Whilethemixture isstiffer, it isstillimportantforittoretainsomeelasticityandresistancetofatiguecracking.TheEuropeanstestfatiguethroughthebendingofaprism,whichisnotcommonintheUnitedStates.SouthAfricausesafour-pointbendingfatiguetestonbeamsfollowingAASHTOT321.SouthAfricantentativeperformancecriteriarequireClass1mixturestohavenogreaterthana70%stiffnessreductionat310microstrainfor10,000,000repetitions.ForaClass2mixture,thisrequirementissetforastrainloadof410microstrain(Dennemanetal.,2011).
Someconcernshavebeenraisedaboutfatiguetestrepeatabilityandresults.Alaboratoryandfull scale pavement testing study was conducted at Nantes, France to assess four differentmixtures’behaviorstofatigueinvariouslaboratoryandfieldconditions.ThecirculartesttrackofLCPCusedfourdifferentmixturestomakeupeachquadrantofthetrack:(1)anasphaltmixturewith50/70penasphaltfromonesource,(2)anasphaltmixturewith50/70penasphaltfromasecondsource,(3)anHMACwith10/20penasphalt,and(4)aroadbaseasphaltwith50/70penasphalt. Inadditiontothefieldwork,eachmixturewassubjectedtofatiguetestingusingthefollowingprocedures:(1)two-pointbendingfatiguetestsontrapezoidalsampleswithcontrolledstrain,withandwithoutrestperiods;(2)two-pointbendingfatiguetestsontrapezoidalsampleswith controlled stress, without rest periods; and (3) three-point bending fatigue tests onparallelepiped-likesamples,withcontrolstress,withandwithoutrestperiods.
The laboratoryrankingsof thematerialsdependedgreatlyonthetestingprocedure,showingthatchoosingthecorrecttestingprotocoliscriticalforensuringthattherightmixtureproperties
Leiva-Villacorta,Taylor,andWillis
20
are being analyzed. However, for the field experiment, 1.0 cm of very thin asphalt concrete(VTAC)wasplacedover7.7cmofHMAC,anditwascomparedto1.5cmofVTACover10cmofroadbaseasphalt.AllasphaltmaterialswereplacedoverasofterbasethanistypicallyfoundinFrance.Thestudyuseda65kNdualwheelat10rpms(70kphlineardueto19mradiusoftestingdevice)toloadthepavementsfor2,665,000loadrepetitions.Figure7showslessdeflectionsinthetypicalroadbase(GB3)comparedtotheHMAC(EME)mixtures;however,theauthorsalsonote thiswas not the typical condition for EMEmixtures due to the softer subgrade. As forcracking,theHMACwasthelastmixturetoexhibitfatiguecracking;however,onceitexhibitedcracking,ittendedtoprogressfaster,showingthemorebrittlenatureofthematerial(Figure8).It shouldbenoted that theseexperimentswereconductedbeforepolymerswerecommonlyusedinHMACbinders(deLaRocheetal.,1994).
Figure7DeflectionsReferredto20°C(deLaRocheetal.,1994)
Leiva-Villacorta,Taylor,andWillis
21
Figure8Cracking(deLaRocheetal.,1994)
3.2 PavementDesign
TheNationalAsphaltPavementAssociationrecognizedthepotentialforHMACtobeanintegralpartoflong-lifeorperpetualpavementdesign.TheorganizationrecognizedearlythattheFrenchhadusedHMACtojustifythinnersectionsfortheirlong-lifepavementdesigns(NewcombandHansen,2004;Newcombetal.,2010).
TheFrenchhavedevelopedananalyticalpavementdesignmethodthatcancapturetheeffectsofHMAC.LCPChasdevelopedAlize-LCPCsoftware,whichusesamechanistic-empiricalapproachwheretraffic,materialproperties,andperformancecoefficientsareusedtopredictpavementperformance(Guyot,2013),butotherdesignprogramscanbeusedthat incorporatematerialpropertiesintodesign.
Leiva-Villacorta,Taylor,andWillis
22
OnecasestudythatincorporatedHMACasadesignoptionwasthe2011-12runwayoverlayandnewparalleltaxiwayattheSirSeewoosagurRamgoolamInternationalAirportinMauritiusintheIndian Ocean. When comparing the typical runway buildup from the FAARFIELD AirportPavement Design Software sponsored by the FAA, the use of HMAC reduced the necessaryrunwaythicknessby105mm.Notonlydidthissavemoneyandnaturalresources,butastudyshowedthatitalsosavedapproximately13%ingreenhousegasemissions(Guyot,2013).
AsThailandwasconsideringusingHMACinpavementstructures,Alize-LCPCwasusedtoconductstructuralanalysesgivenanexpectedmodulusof14,000MPaandfatigueresistance(106cycles)at130microstrainfortheHMAC.Theroadbaseasphaltwasexpectedtohaveamodulusof9,300MPaandafatigueresistance(106cycles)at90microstrain.Thedesignsoftwareshowedthatforsimilarwearingcourses,7cmlessstructurewasneededfortheHMACmixtures,thus,reducingthe required pavement structure by 20% (Lefant, 2012). Other sources cite that somegovernmentshaveseena30%reduction inneededpavementstructuredueto the increasedstiffnessofthepavementandaddedfatigueresistance(Corte,2003).
Carbonneauetal.(2008)conductedanexperimentwheretheycomparedthereferencecross-sectionoftheHerningbypasstotwocross-sectionscontainingHMACmixtures,calledaHMAGABIImixtureinDenmark.Whenthemechanicalpropertiesofthetwomixtureswereplacedthroughamechanistic-designprogram,theuseoftheHMACmixtureallowedthebypasscross-sectiontobereducedby25mminthickness.Thetwofinalcross-sectionsaregiveninFigure9.Thisanalysisresulted intheHMACmixturebeingusedtoreducematerialquantities,andtheroadwayiscurrentlybeingmonitored(Carbonneauetal.,2008).
Figure9RoadwayStructureofMixtureDesignBasedonCharacteristicMechanicalGainof
HMAGABII(Carbonneauetal.,2008)
Leiva-Villacorta,Taylor,andWillis
23
Weilinski and Huber (2011) showed these mixtures could be incorporated into the currentversionoftheMechanistic-EmpiricalPavementDesignGuide(MEPDG).ThedesignincorporatedAmericanmaterialsfromIndianathatincludedrecycledasphaltshingles(RAS),RAP,binder,andaggregate.TheteamfoundthatbinderstiffnesshadlittletodowithperformanceintheMEPDGonrutting,IRI,orfatiguecracking.TheseresultsweresimilarwhencomparingstiffmastercurveslikethePG64-22binderwithRAScomparedtothePG76-22bindermastercurve.TheincreasedmixturestiffnessdidimprovefatigueperformanceandrideintheMEPDG.Additionally,usingtheHMACreducedthepavementthicknessby16%toachievesimilarperformance.
3.3 Performance
A small-scale laboratory experiment was conducted in China to assess the impact of high-modulusasphaltonruttingresistancewhencomparedtopolymer-modifiedandconventionalsofterbinderswhenusedinthebindercourseofapavement.Usingfiniteelementmodeling,theshear strains and compressive stresses within the middle of the pavement were calculatedthroughatypicalpavementcross-sectionforthearea.Theresults(Figure10)showthatusingthehigh-modulusmaterialsreducedshearstrain.TheseresultsalsoshowedthatdespitethehigherbindercontentusedinanHMACdesign,theruttingresistanceofthepavementsusingtheHMACwasincreased(Weietal.,2010).
Figure10TendencyofShearStrain(Weietal.,2010)
AfieldexperimentwasdevelopedoutsideofBrussels,Belgiumtoassesshowaggregateskeleton,recycled materials, binder content, and grade all influenced field performance of pavementstructureswithHMACbinder courses. Each 140-m test sectionwas constructedwith a 9 cmvariantofanHMACwitha3cmstonematrixasphalt(SMA)orporousasphaltsurface.Duringconstruction,coresweretakenfromeachtestsectionandtestedforruttingusingEN12697-22at50°C.Theresults(exceptfortheporousasphaltsections)showedthatusingHMACimprovedruttingresistance.Thedifferencesinaggregateskeleton(stonyversussandy)didnotimpacttheruttingvaluesofthemixtures,nordidbindertype.Theprimarydriverforruttingresistancewasbindercontent,asmixtureswithlowerbindercontentshadlessrutting(Figure11).Afterayearinthefield,thehigh-modulustestsectionsallout-performedthecontroltestsection(DeBackeretal,2008;DeBacker,deVisscheretal.,2008).
Leiva-Villacorta,Taylor,andWillis
24
Figure11RutDepthasaFunctionofBinderContentforMixturewithSandSkeleton
(deVisscheretal.,2008)
The country of Korea wanted to evaluate the use of HMAC for long-life pavements. AfterdevelopinganHMACmixturedesign,theHMACwascomparedtoaconventionalasphaltmixtureinperformance testsbeforebeing introducedatanaccelerated loading facility.Themixtureswereevaluated for stiffness (Figure12), fatigue in indirect tension (Figure13), rutting via KSF2374testprocedure,andmoisturedamageusingASTMD4867(Table9).Inallcases,theHMACwasshowntoperformbetterthantheconventionalmixture.
Figure12DynamicModulusMastercurvesat15°CReferenceTemperature(Leeetal.,2007)
Leiva-Villacorta,Taylor,andWillis
25
*RP=restperiod
Figure13ResultsofFatiguefortheConventionalMixtureandHMAC(HMAM)(Leeetal.,2007)
Table9PerformanceTestResultsfromKoreanLaboratoryStudy(DatafromLeeetal.,2007)
Mixture DryStrength,kPa WetStrength,kPa TSR RutDepth,mmConventional 1070.9 948.3 88.54 7.28
HMAC 1515.1 1489.6 98.32 2.79
Due to the success in the laboratory, thesemixtureswere then compared in the field at theHanyangUniversityAcceleratedPavementTester.Inthisprocedure,an11-tonload(maximumaxleloadinKoreais10tons)isappliedto12.5mofpavementatamaximumspeedof17km/h.Twolanesofthesemixtureswereproduced.Thefirstlanewasdesignedtostudyfatiguecrackingand the experimental mixtures were constructed at 94 and 83 mm in thickness for theconventionalandHMACmixtures,respectively.Thesecondlanewasdesignedtostudyruttingandthemixtureswereconstructedthicker.Theconventionalmixturewas268mmthickwhiletheHMACmixturewas215mmthick.Straingaugeswereplacedatthebottomoftheasphaltlayertocharacterizethepavementresponse.
The results showed that despite having a thinner cross-section, the HMAC could reduce thetensilestrains inthetestsectionsexceptforthethintestsectionatthelowestwheel loading(Figure14).Thepreliminaryresultssuggestedthatthisfatigueperformancewould,infact,allowKoreatousethesemixturesaspartofalong-lifepavementconceptwithadditionalvalidationasnofatiguecrackingwasnoticedineithertestsectionattheendoftheexperiment.Thehigh-modulusmixtures also performed better than the conventionalmixtures in terms of rutting
Leiva-Villacorta,Taylor,andWillis
26
(Figure 15). TheHMACmixture had less thanhalf of the rutting in the conventionalmixturedespitebeingconstructedonathinnerasphaltcross-section(Leeetal.,2007).
Figure14TensileStrainwithChangeofDualWheelLoadfor(a)ThinPavementSectionand
(b)ThickPavementSection(Leeetal.,2007)
Leiva-Villacorta,Taylor,andWillis
27
Figure15ComparisonofRutDepthsforConventionalMixturesandHMACPavement(Leeet
al.,2007)
AspartoftheimplementationeffortinEurope,twoHMACmixtureswerecomparedtotwobasicasphaltmixturedesignsinthelaboratoryandthefield.Fieldtestingwasconductedunderaheavyvehiclesimulator(HVS).Structuralcross-sectionsofthefourtestsectionsareprovidedinTable10.TheHVSapplied60kNviaasingleaxleloadingwithatirepressureof800kPaataspeedrangingfrom10-12kph.
Table10StructuralBuild-Ups(InformationfromBarkowskietal.,2007)
PavementLayer A B C D
WearingCourse
Thickness:2cmMixturetype:SMA
Binder:PmB45/8065
Thickness:4cmMixturetype:
porousBinder:50/70
Thickness:4cmMixturetype:SMA
Binder:PmB45/80-65
BinderCourse
Thickness:10cmMixturetype:HMAC
16Binder:20/30
Thickness:10cmMixturetype:AC
16WBinder:35/50
Thickness:8cmMixturetype:AC
16WBinder:35/50
Thickness:7cmMixturetype:HMAC
15Binder:20/30
UnboundMaterials
Thickness:20cmAggregate:dolomite
Additionally,modeledfatiguelifeanddamageweredeterminedbasedontheAsphaltInstitutetransfer function consideringa60 kNwith theHVSwheel configuration. The twoHMAC testsectionsshowedthebestperformance(Figure16). It is interestingtonotethatsectionDhadhighermeasuredandmodeledstrainsthanSectionA(Figure17);however,ithadbetterexpected
Leiva-Villacorta,Taylor,andWillis
28
performance.ThismayhavebeenduetothedifferencesinfatigueperformanceoftheHMACinthebinder/basecourse(Bankowskietal.,2009).
Figure16ComparisonofFatigueLifeforEachIndividualSection(Bankowskietal.,2009)
Figure17ComparisonofMeasuredandCalculatedStrainsattheBottomoftheAsphaltLayers
(Bankowskietal.,2009)
In2010,theVirginiaTransportationResearchCouncil(VTRC)reportedtheresultsofastudyoffield trials of high-modulus high-binder-content base layer hot-mix asphalt mixtures
Leiva-Villacorta,Taylor,andWillis
29
(DiefenderferandMaupin,2010).Threelocationswheredeeprehabilitationornewconstructionwereselected,andHMAbasemixtureswereusedatdesignedasphaltcontent,designedasphaltcontent plus 0.4% additional asphalt, and/or designed asphalt content plus 0.8% additionalasphalt. Twoof the field trial locations had no construction-related issues; difficulties duringcompactionoccurredatthethird.TheresultsofthisstudyindicatedthatthebinderstiffnessforanHMACmixtureshouldbeat leastequivalenttothatofaPG70-22bindertoguardagainstpotentialruttingandadditionofRAPmaybenecessary.
3.4 Construction
Whilelittlehasbeenpublishedregardingtheconstructionofthesemixtures,theBelgianstudydidgainsomeinsightstotheapproach.Attimes,conventionalmethodswerenotappropriateeitherduetothedesignasphaltcontentorbinderstiffness;thus,theresearchteammadenoteofthefollowingitems:(1)contractorsmustusebinderproducer’srecommendedtemperaturesduringproduction;(2)commonaggregatesizetoliftthicknessratiosdidnotapply,asHMACwasplaced9to10cmthickwithoutanyproblem;(3)mixtureswereeasytocompactwithtraditionalequipment;(4)compactiontemperatureswerecommonlyabout10°Chigherthanconventionalmixtures;(5)thefattylookofthemixturedoesnotindicateovercompaction;and(6)voidsratiosweresimilartothoseinconventionalbindercourses(DeBackeretal.,2008).Denneman(2011)observedthatthesemixturescommonlyrequirehighermixingtemperatures.
Nichollsetal.(2008)conductedanexperimenttomonitorthedurabilityandbuildabilityofHMAConfivedifferentsitesintheUK.Duringthepilotprojects,someinstanceswerenotedwhenthelevelofcompactionwasnotachieved(SiteD);however,onothertrialprojects,thein-placeairvoids were extremely low (1% at site B). Falling weight deflectometer and laboratory testsshowedthatwithfewexceptions,themixturesweredesigned,produced,andconstructedwell.
Michaut(2014)providedthefollowingrecommendationsforproducingandlayingHMAC:
• Mixingtemperatureshouldbebetween160and180°Candalwayslessthan190°C.• Theminimallayingtemperatureforthismixtureis145°C,butthiswilldependonbinder
properties.• Granularbasemustbecompactedwelltoensurehighin-situdensityofHMAC.
Jamois et al. (2000) notes that sometimes these temperatures can be exceeded if materialpropertiesdictatetheneed.MixturesplacedatthecirculartesttrackinFrancewereproducedbetween200and210°Candplacedat195°C.
3.5 SummaryofCurrentPractice
Acomprehensiveliteraturestudywasperformedtoassessthecurrentstate-of-the-practiceonHMACmixturedesign,pavementdesign,laboratoryperformancetests,andfullscalepavementperformance.Themajorityof theobservedexperiencecomes fromEurope.TheFrenchHMAmixturedesignmethodisthemostcommonlyusedmethodologyandhassomevariationsinthedesignprocedurecomparedtoconventionalSuperpavedesignmixturescommonlyutilizedinthe
Leiva-Villacorta,Taylor,andWillis
30
United States. The firstmain difference is themethod of compaction. The determination ofminimumbindercontent in theFrenchdesignmethod isalsoquitedifferent fromSuperpavedesign.TheFrenchmethodcallsforaminimumasphaltbindercontentbasedontherichnessfactor,surfacearea,andspecificgravityoftheaggregates.However,asmentionedbefore,theAIHveem-Edwardbinderfilmthicknesscalculationisasgood.Asphaltmixturedesignproceduresincludeperformancetestingrequirementsformoisturedamage,aruttestforrutting,complexmodulusforstructuralstiffness,andfatiguetestingforfatiguecracking.
Pavement design and analysis of HMAC mixtures is conducted using mechanistic-empiricalapproaches.Thispractice includesevaluationofpotential fieldperformanceand reduction inneededpavementstructureduetotheincreasedstiffnessofthepavementandaddedfatigueresistance.Fullscaleperformancetestinghasbeenusedtovalidatethisaddedfatigueresistance.
Overall,thereisanexpecteddifferenceinthelaboratoryperformanceofHMACmixtureswhencomparedtotraditionalasphaltmixtures.Thisperformancedifferenceisexpectedtotranslateintothefieldwherepavementscaneitherbedesignedthinnerwiththesameexpectedlifeordesigned at the same thicknesswith long-life performance as a viable expectation. Table 11exhibitsanexampleofchangeinspecificationsforHMACmixturescomparedtoconventionalasphaltconcretemixturesinEurope.
Table11RoadbaseHigh-ModulusAsphaltConcreteversusTraditionalAC(EAPA,2005)
RoadbaseHigh-ModulusAsphaltConcreteversusTraditionalACTest HMACR1 HMACR2 AC
InmersionCopressiontestat18°C >0.7 >0.75 >0.7Ruttingtestat60°C30,000cycles <8% <8% <10%Stiffnessmodulusat15°Cand10Hz >14,000MPa >14,000MPa >9,000MPa
Allowedmicrostrainfromfatiguelawat10°Cand25Hzandfor106cycles >100 >130 >90
Voidcontentforlayingthickness <10% <6% <10%
4. EXPERIMENTALDESIGNANDANALYSIS
4.1 LaboratoryTesting
Therewere twoobjectives addressed in the laboratoryexperimental plan: (1) determine theengineering properties of high-modulus mixtures, and (2) determine whether or not therecommended design procedures were appropriate. The information to accomplish bothobjectiveswasobtainedfromEuropeanexperience.Thissectiondetailstheapproachadoptedtoaddressthetwoobjectivesofthelaboratoryresearch.
Toassessstatisticaldifferencesamongmixtures,thegenerallinearmodel(GLM)(α=0.05)wasconducted.OverallcomparisonsofsuchpropertiesweremadeusingTukey-Krameranalysiswiththeresultsfromalllaboratoryperformancetests.Theresultsofthelaboratorytestingwerealsousedtodetermineifthecurrenttestingprocedurescouldadequatelypredicttheperformance
Leiva-Villacorta,Taylor,andWillis
31
ofpavementscontaining these typesofmixtures in the field.The laboratory testingprogramevaluatedbinderperformancegrade,mixturestiffnessoverawidetemperaturerange,fatiguecracking,andpermanentdeformation,asfollows:
• Volumetricmixturedesignandmaterialcharacterization,• Mixturestiffness:dynamicmodulus(AASHTOTP79-13),• Flownumber(AASHTOTP79-13),and• AMPTcyclicfatigue(AASHTOTP107-14).
4.2 MixtureDesign
The French asphalt mixture design method has some variations in the design procedurecomparedtoconventionalSuperpavedesignmixturescommonlyutilizedintheUnitedStates.TheFrenchgyratorycompactorusesaninternalangleof0.82degrees,whereastheSuperpavegyratorycompactoremploysaninternalangleof1.16°.WielinskiandHuber(2011)usedintheirresearchstudytheresultsofacomparisonstudyoftheLCPCgyratorycompactorandaSuperpavegyratorycompactorforEMEmixturedesignthatwasconductedbytheJiangsuTransportationResearch Institute. This comparison work determined that 80 gyrations in the Superpavecompactorproducedthesamecompactionas100gyrationsintheLCPCgyratorycompactor.
Forthisstudy,sampleswerecompactedat80gyrationsintheSuperpavecompactor,andthetargetdesignairvoidsatNdeswassetfrom3.0to6.0%(Europeanspecificationrequiresdesignairvoidstobelessthansixpercent).Fordynamicmodulustesting,therangeofallowableairvoidcontentwasalso3.0to6.0%withaminimumdynamicmodulusat15°Cand10Hzof14,000MPa. In addition, the gradation selected for each trial dictated theminimumbinder contentrequiredinthedesignaccordingtotheFrenchmethodology.
Theexperimentalplanincludedavarietyofmixtureswithdifferentmaterialandbinderssuchthathighermoduliwereobtainedcomparetoconventionalmixtures.TheplanincludedaFrenchmixturewitha stiffbinder (PG88-16), twomixtures containing35%RAPbothwithpolymer-modifiedbinders,butonehighpolymercontent(HiMA),anothermixturecontaining25%RAPand5%RASwithapolymer-modifiedbinder,and finally,a50%RAPmixturewithapolymer-modifiedbinder.
Table12showstheaggregategradationsandblendformulas for thefourmixturesthatwereproduced. Table 13 shows the volumetric properties of eachmixture determined during thedesignphase.
Leiva-Villacorta,Taylor,andWillis
32
Table12AggregateGradationsofMixtures
SieveSize(in.)
FrenchEME14
35%RAPPG76-22
25%RAP,5%RASPG76-22
35%RAPHiMA
50%RAPPG76-22
2" 100.0 100.0 100.0 100.0 100.01.5" 100.0 100.0 100.0 100.0 100.01" 100.0 99.3 98.6 99.3 98.83/4" 100.0 95.5 91.0 95.5 92.41/2" 88.9 89.3 81.8 89.3 85.53/8" 79.7 79.5 72.6 79.5 78.5#4 58.9 54.9 51.5 54.9 51.1#8 37.7 42.7 41.5 42.7 40.7#16 26.6 32.5 32.0 32.5 31.2#30 19.3 22.5 21.5 22.5 21.3#50 14.0 12.1 11.1 12.1 10.6#100 8.8 7.1 6.4 7.1 5.9#200 7.9 4.7 4.2 4.7 3.9
Leiva-Villacorta,Taylor,andWillis
33
Table13MixtureDesignProperties
MixtureDesignation FrenchEME14
35%RAPPG76-22
25%RAP,5%RASPG76-22
35%RAPHiMA
50%RAPPG76-22
Gyrationlevel 80 80 80 80 80NMAS(U.S.sieves) 19 19 19 19 19Bindercontent(%) 5.70 5.12 5.01 5.12 5.04Basebindercontent(%) 5.70 3.24 2.73 3.24 2.98
Basebindergrade PG88-16 PG76-22(SBS)
PG76-22(SBS)
PG88-22(HighSBS)
PG76-22(SBS)
PercentRAP 0 35 25 35 50PercentRAS 0 0 5 0 0RAPAC% n/a 5.37 5.37 5.37 4.12RASAC% n/a n/a 18.69 n/a n/aRAPbinderratio n/a 0.367 0.268 0.367 0.409RASbinderratio n/a n/a 0.187 n/a n/aGsb 2.751 2.716 2.730 2.716 2.690Gmm 2.478 2.542 2.543 2.542 2.530Designairvoids(%) 1.5 2.0 3.0 2.3 2.6Gmbdesign 2.441 2.491 2.467 2.484 2.464VMA 15.0 13.0 14.2 13.2 13.0VFA 90.0 84.6 78.8 82.6 80.0Gb 1.028 1.028 1.028 1.028 1.028Gse 2.709 2.761 2.757 2.761 2.743Pba n/a 0.62 0.37 0.62 0.73Pbe n/a 4.53 4.66 4.53 4.34Dustproportion n/a 1.03 0.91 1.03 0.90E*at15°Cand10Hz,MPa 17,506 14,519 15,753 14,457 17,137Airvoids-E*specimen(%) 3.4 3.6 4.0 4.00 3.8
4.3 DynamicModulus
Asinglepointmeasurement (E*at15°Cand10Hz) cannotbeexpected todescribematerialbehavior across all possible loading temperatures/frequencies; therefore, Dynamic ModulustestingwasperformedforallmixturesaccordingtoAASHTOTP79-13.Inaddition,theseresultswereused to estimatepavement performanceusing theAASHTOWarePavementMEDesignsoftware.
Sampleswerecompactedtoaheightof175mmandadiameterof150mmandpreparedtomeetthetolerancesoutlinedinAASHTOPP60-14.DynamicmodulustestingwasperformedinanIPCGlobalAsphaltMixturePerformanceTester(AMPT),showninFigure18.Dynamicmodulustestingisperformedinordertoquantifythestiffnessoftheasphaltmixtureoverawiderangeoftestingtemperaturesandloadingrates(orfrequencies).ThetemperaturesandfrequenciesusedfortestingthesemixturesarethoserecommendedbyAASHTOPP61-13.Forthismethodology,
Leiva-Villacorta,Taylor,andWillis
34
thehightesttemperatureisdependentonthehighPGgradeofthebasebinderutilizedinthemixturebeingtested.
Figure18IPCGlobalAsphaltMixturePerformanceTester
DynamicmodulustestingwasperformedinaccordancewithAASHTOTP79-13inanunconfinedcondition.Unconfineddataismostcommonlyusedfordynamicmodulustestingsincecurrentmechanisticdesignsoftwarepackageswerecalibratedusingunconfineddynamicmodulusdata.UnconfinedtestingisalsosignificantlyeasiertoperformthanconfinedtestingandMEpackageswerecalibratedusingunconfinedresults.
The collected data were used to generate a mastercurve for each individual mixture. Themastercurveusestheprincipleoftime-temperaturesuperpositiontohorizontallyshiftdataatmultipletemperaturesandfrequenciestoareferencetemperaturesothatthestiffnessdatacanbeviewedwithouttemperatureasavariable.Thismethodofanalysisallowsforvisualrelativecomparisonstobemadebetweenmultiplemixtures.
Generationofthemastercurvealsoallowsforgenerationofthedynamicmodulusdataovertheentire range of temperatures and frequencies required for mechanistic-empirical pavementdesign.Byhavinganequationforthecurvedescribingthestiffnessbehavioroftheasphaltmix,both interpolatedandextrapolateddatapointsatvariouspointsalongthecurvecanthenbecalculated.ThegeneralformofthemastercurveequationisshownasEquation5.Asmentioned,the dynamic modulus data are shifted to a reference temperature by converting testingfrequency to a reduced frequency using the Arrhenius equation (Equation 6). SubstitutingEquation6intoEquation5yieldsthefinalformofthemastercurveequation,shownasEquation7. The shift factors required at each temperature are given in Equation 8. A referencetemperatureof20oCwasusedforthisanalysis.ThelimitingmaximummodulusinEquation8iscalculatedusingtheHirschModel,shownasEquation9.ThePcterm,Equation10,issimplya
Leiva-Villacorta,Taylor,andWillis
35
variablerequiredforEquation9.Alimitingbindermodulusof1GPaisassumedforthisequation.Non-linear regression is thenconductedusing the ‘Mastersolver.exe’programtodevelop thecoefficientsforthemastercurveequation.Typically,thesecurveshaveanSe/Sytermoflessthan0.05andanR2valueofgreaterthan0.99.DefinitionsforthevariablesinEquations5to10aregiveninTable14.
𝑳𝒐𝒈 𝑬∗ = 𝝏 + (𝑴𝒂𝒙;𝝏)𝟏E𝒆𝜷G𝜸𝒍𝒐𝒈𝒇𝒓
(5)
𝒍𝒐𝒈𝒇𝒓 = 𝒍𝒐𝒈𝒇 + ∆𝑬𝒂𝟏𝟗.𝟏𝟒𝟕𝟏𝟒
𝟏𝑻− 𝟏
𝑻𝒓 (6)
𝒍𝒐𝒈 𝑬∗ = 𝝏 + (𝑴𝒂𝒙;𝝏)
𝟏E𝒆𝜷G𝜸 𝒍𝒐𝒈𝒇G ∆𝑬𝒂
𝟏𝟗.𝟏𝟒𝟕𝟏𝟒𝟏𝑻Q
𝟏𝑻𝒓
(7)
𝐥𝐨𝐠[𝒂 𝑻 ] = ∆𝑬𝒂𝟏𝟗.𝟏𝟒𝟕𝟏𝟒
𝟏𝑻− 𝟏
𝑻𝒓 (8)
|𝑬∗|𝒎𝒂𝒙 = 𝑷𝒄 𝟒, 𝟐𝟎𝟎, 𝟎𝟎𝟎 𝟏 − 𝑽𝑴𝑨𝟏𝟎𝟎
+ 𝟒𝟑𝟓, 𝟎𝟎𝟎 𝑽𝑭𝑨∗𝑽𝑴𝑨𝟏𝟎,𝟎𝟎𝟎
+ 𝟏;𝑷𝒄𝟏Q𝑽𝑴𝑨𝟏𝟎𝟎𝟒,𝟐𝟎𝟎,𝟎𝟎𝟎E
𝑽𝑴𝑨𝟒𝟑𝟓,𝟎𝟎𝟎(𝑽𝑭𝑨)
(9)
𝑷𝒄 =𝟐𝟎E𝟒𝟑𝟓,𝟎𝟎𝟎 𝑽𝑭𝑨
𝑽𝑴𝑨
𝟎.𝟓𝟖
𝟔𝟓𝟎E 𝟒𝟑𝟓,𝟎𝟎𝟎(𝑽𝑭𝑨)𝑽𝑴𝑨
𝟎.𝟓𝟖 (10)
Table14MastercurveEquationVariableDescriptions
Variable Definition|E*| Dynamicmodulus,psi
δ,β,andγ FittingparametersandparametersdescribingtheshapeofsigmoidalfunctionMax Limitingmaximummodulus,psifr Reducedfrequencyatthereferencetemperature,Hzf Theloadingfrequencyatthetesttemperature,Hz
ΔEa Activationenergy(treatedasafittingparameter)T Testtemperature,oKTr Referencetemperature,oKa(T) Theshiftfactorattemperature,T
|E*|max ThelimitingmaximumHMAdynamicmodulus,psiVMA Voidsinmineralaggregate,%VFA Voidsfilledwithasphalt,%
Figure20exhibits themastercurves forallmixtures includingthe19.0mmNMASbasecourse(control mixture) from the 2009 NCAT Test Track cycle. It can be observed that at the lowtemperature, high frequency endof the curve, all of themixtures tended to have similar E*values.However,whenmovingtowardstheoppositerangeoftemperaturesandfrequencies,slight differences can be observed, especially for the mixture containing recycled asphaltshingles. These trends can also be observed when analyzing the mastercurve regression
Leiva-Villacorta,Taylor,andWillis
36
coefficients. Table 15 gives a summary of themastercurve regression coefficients that weregeneratedusingthemodifiedMEPDGmastercurvemodel.GoodnessoffitparametersarealsoshowninTable15.
MaximumE*valuesweresimilar;however,minimumE*valuesdidshowsignificantdifferences.In termsof the steepness of the curve givenby theparameter -γ, the 25%RAS and5%RASmixtureshowedthelowestslope(lesssusceptibletochangesinfrequency),andthe50%RAPmixture showed the highest slope (most susceptible to changes in frequency). The inflectionpointfrequencyparameter-β/γwasthehighestforthe25%RASand5%RASmixture(4.55Hz),followedby the FrenchEMEmixture (3.61Hz); the remainingmixtureshad similar inflectionpoints around 2.85 Hz. The activation energy term is best regarded as an experimentallydetermined parameter that indicates the sensitivity of the shift factors to temperature, andconsequentlyaffectstheshapeofthemastercurve.Inthiscase,allmixtureshadsimilaractivationenergyterms,butthe25%RAP-5%RASmixturehadthehighesttermproducingawiderrangeofreducedfrequenciesandamoreflattenedcurve.
Figure20DynamicModulusMastercurves
Table15MastercurveCoefficients
MixtureID MaxE*(Ksi)
MinE*(Ksi) Beta Gamma EA R2 Se/Sy
FrenchEME 3240.96 8.57 -2.023 -0.560 207939.5 0.991 0.06635%RAPPG76-22 3417.23 17.05 -1.697 -0.595 197985.7 0.993 0.05725%RAP,5%RAS 3389.56 10.40 -1.972 -0.433 244676.9 0.985 0.08835%RAPHiMA 3423.82 26.10 -1.548 -0.556 215061.3 0.997 0.041
50%RAPPG76-22 3446.47 18.40 -1.778 -0.610 206143.0 0.993 0.058
10
100
1,000
10,000
1.0E-06 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04
Dyna
micM
odulus,ksi
ReducedFrequency,Hz
FrenchEME 35%RAPPG76-2225%RAP,5%RAS 35%RAPHiMA50%RAPPG76-22 NCATControl
Leiva-Villacorta,Taylor,andWillis
37
Inanattempttoidentifytestingvariabilityand/ornon-linearityinthematerialbehaviorduetonon-compliancetotherecommendedmicro-strainlevels,thedynamicmodulusandphaseanglewereaveragedforeachlaboratory’sdataandplottedinBlackSpace(Airey,2002;Christensenetal,2003).Figure21containstheBlackSpaceplotsforallthedifferentmixesincludingthe19.0mmNMAScontrolmixturefromthe2009NCATTestTrackcycle.Itshouldbenotedthatallplotsshowgooduniformity in their respectiveBlackSpacediagrams,asnotedwith theirR2valuesbeinggreaterthan0.94fora4th-orderpolynomialfittedfunction.Duetotheinteractionoftheasphaltbinderwithaggregate,theBlackSpacediagramforamixtureshowsapeakphaseanglevalueatintermediatedynamicmodulus.Athightemperatures,theaggregatestructurebeginstodominatebehaviorofthemixturewhileatlowertemperaturesvolumetricpropertiesandbinderstiffness control the behavior. This peak value is associated with the inflection point in themastercurve using the terms described earlier (-β/γ).French EME and 50%RAPmixtures hadsimilarpeakphaseanglesaround33degreesbutdifferentpeakdynamicmodulusof220ksiand237ksi,respectively.Forthe35%RAPmixture,thispeakoccursarounddynamicmodulusof230andforthe25%RAS-5%RASand35%RAPHiMAmixtures,thisoccursarounddynamicmodulusof205ksiand218ksi,respectively.
AdditionalanalysisoftheBlackSpacediagramindicatesthatmixtureswithlowerphaseanglevaluesaremoreelastic(25%RAS-5%RAS,35%RAPHiMAmixtures).Ontheotherhand,ifthephaseangleishigh,themixtureismoreviscous(FrenchEMEand50%RAPmixtures)(Rahbar-Rastegar and Daniel, 2016). In addition, stiffer mixtures at lower phase angles are moresusceptibletocracking(Andersonetal.,2011)Inthiscase,the35%RAPand50%RAPmixtureshaveslightlyhighermoduliatlowphaseanglesthantheothermixtures.
Figure21BlackSpaceDiagrams
Toassessstatisticaldifferences,agenerallinearmodel(GLM)(α=0.05)wasconductedonthetestdatameasuredat4°C,20°C,and45°C,andattwofrequencies:10Hzand1Hz.Thus,theGLM
R²=0.99871
R²=0.9974
R²=0.9976
R²=0.99646
R²=0.99757R²=0.996730
5
10
15
20
25
30
35
4 4.5 5 5.5 6 6.5 7
PhaseAn
gle,degrees
LogE*,psi
FrenchEME
35%RAPPG76-22
25%RAP5%RAS
35%RAPHiMA
50%RAPPG76-22
NCATControlBaseCourse
Leiva-Villacorta,Taylor,andWillis
38
was completed four times to assess statistical differences at each temperature. The Tukey-Kramertest(α=0.05)wasusedtodeterminewherethesestatisticaldifferencesoccurredandhowthemixturesgroupedwithineachproject.Table16showstheresultsoftheTukey-KramertestonE*valuesfor20°Cand45°Conly(resultsat4°Cfollowedsimilarstatisticaltrendofresultsat 20°C).Mixtures given the same letter in the tablewere statistically grouped together (nostatisticaldifferenceamongmixturesatα=0.05).Asexpected,athightemperaturesandlowfrequencies,statisticaldifferenceswereobtainedforsomeofthemixtures.Theresultsindicatedthatthemixturewith25%RAPand5%RAShadthehighestE*valueswhilethemixturewith35%RAPHiMAhadthelowestE*values.Ontheotherhand,nostatisticaldifferenceswereobtainedamongmixturesat20°Cforasignificancelevelα=0.05.
Table16E*StatisticalGrouping
MixtureID 20°C,10Hz 20°C,1Hz 45°C,10Hz 45°C,1HzMean Group Mean Group Mean Group Mean Group
FrenchEME 2,103.7 A 1,517.8 A 615.8 AB 293.1 AB35%RAPPG76-22 1,982.9 A 1,388.3 A 545.7 AB 264.1 B25%RAP,5%RAS 2,086.0 A 1,544.7 A 673.0 A 373.2 A35%RAPHiMA 1,910.1 A 1,375.6 A 505.6 B 261.5 B
50%RAPPG76-22 2,119.8 A 1,503.8 A 555.1 AB 265.5 B
4.4 FlowNumber
TheFlownumbertestisaruttingresistancetestthatisperformedusingtheAMPT.ItappliesarepeatedcompressiveloadingtoanasphaltspecimenwhiletheAMPTrecordsthedeformationofthespecimenwitheachadditionalloadingcycle.Theuserdefinesthetemperature,appliedstress state (deviator stress and confining stress), and number of cycles atwhich the test isperformed.Theloadingisappliedforadurationof0.1secondsfollowedbya0.9secondrestperiodevery1secondcycle.FlownumberdataiscommonlymodeledwiththeFranckenmodel,shownasEquation11(AASHTOTP79-13).AnexampleofunconfinedflownumbertestdataisshowninFigure22.
(11)
where
εp(N) = permanentstrainat‘N’cycles,N = numberofcycles,and
a,b,c,d = regressioncoefficients.
Theflownumberisdefinedasthenumberofcyclesatwhichthesamplebeginstorapidlyfailandcoincideswiththeminimumrateofstrainaccumulationmeasuredduringthetest.Thisismoreproperly defined as the breakpoint between steady-state rutting (secondary rutting) and themore rapid failure of the specimen (tertiary flow). Figure 22 demonstrates this conceptgraphically.Ifthesamplesdonotexhibittertiaryflow(commonforconfinedsamples),thenthe
)1()( −+= dNbp ecaNNε
Leiva-Villacorta,Taylor,andWillis
39
amountofdeformationataspecifiedloadingcyclecanstillbeusedtogivearelativerankingoftestedmixtureswithrespecttoruttingsusceptibility.
Figure22TypicalFlowNumberTestData
FlownumbertestingforthisprojectwasperformedinaccordancewithAASHTOTP79-13inanunconfinedstatewithadeviatorstressof87psi.Thetestswererununtileither thesamplesreached5%axialstrain(7.5mmofdeformationona150mmsample)orthetestwentthefull20,000cycles.SampleswerepreparedinaccordancewithAASHTOPP60-14toatargetairvoidlevelof3.5±0.5percentonthefinalcoredandtrimmedspecimen.ByAASHTOTP79-13,theflownumbertesttemperatureisselectedbasedontheLTPPBind50%reliabilityhighpavementtemperatureattheprojectlocationadjustedfora20mmdepthinthepavementstructure.TheAuburn,Alabamaclimateregionwasassumedtogeneratetheflownumbertesttemperature.ThetemperaturedatafromtheLTPPBindv3.1softwareforAuburnisshowninFigure22below.Based on these criteria, the temperature of 59.5°C was selected for this project (LTPPBindtemperatureroundedtothenearest0.5°C).WhileAASHTOTP79-13doescontaintrafficlevelcriteria for mixtures based off their flow number results, these criteria are not completelyapplicableforthisstudysincethespecimenswerefabricatedtoadifferenttargetairvoidcontent(thesecriteriaareforspecimensfabricatedto7.0±0.5percentairvoids).
Leiva-Villacorta,Taylor,andWillis
40
Figure22LTPPBindv3.1OutputforAuburn,ALArea
Table 17 shows the results of the flow number test performed on all themixtures and thestatisticalgrouping.TheGLM(α=0.05)showednostatisticaldifferencebetweenthe35%RAPandthe50%RAPmixtures.Ontheotherhand,theremainingmixtureswerestatisticallydifferentfrom each other. The 35% RAP HiMAmixture showed the highest resistance to permanentdeformation followedby the25%-5%RASmixture.Similarstrainvalueswereobtained forallmixturesbuttheFrenchEMEmixture(moreductile).Allofthemixturesexhibitedflownumbervalues well in excess of the 740 recommended for a greater than 30 million ESAL designpavementbyAASHTOTP79-13.While these criteria arenot completely applicable given theaforementionedairvoidlevelofthespecimens,itdoesgivesomeframeofreferenceforthehighlevelofruttingresistanceofferedbythesehigh-modulusmixtures.
Leiva-Villacorta,Taylor,andWillis
41
Table17FlowNumberTestResults
MixtureID
AirVoids(%) FranckenFlowNumber FranckenMicrostrainatFN FNStatisticalGroupAverage Average Standard
DeviationCV(%)
Average StandardDeviation
CV(%)
FrenchEME
3.5 4,665 716 15.3 29,381 2,271 7.7 A
35%RAPPG76-22
3.2 1,910 577 30.2 17,648 1,365 7.7 B
25%RAP,5%RAS
3.5 8,229 676 8.2 15,128 1,367 9.0 C
35%RAPHiMA
3.3 18,374 1,807 9.8 16,362 3,122 19.1 D
50%RAPPG76-22
3.1 1,337 485 36.3 15,987 1,406 8.8 B
4.5 AMPTCyclicFatigue
FatiguetestingforthisprojectwasperformedusingtheuniaxialtensionfatiguemethodavailableintheAsphaltMixturePerformanceTester(AMPT).ThismethodissummarizedinAASHTOTP107-14.Thismethodologyutilizesthesimplifiedviscoelasticcontinuumdamage(S-VECD)model(Houetal.,2010).Hereafter,thistestingprotocolwillsimplybereferredtoasAMPTcyclicfatigueS-VECDtesting.AMPTcyclicfatigueanalysisonagivenmixturerequiresbothdynamicmodulus(|E*|) testing as well as uniaxial fatigue testing. S-VECD is a mode-of-loading independent,mechanisticmodelthatallowsforthepredictionoffatigueperformanceparametersatdifferenttemperaturesandloadingconditions(Jacquesetal.,2016;DanielandKim,2002;Underwoodetal.,2012).
Specimen preparation for cyclic fatigue testing is identical to that for the AMPT specimensrequired for the dynamic modulus test (AASHTO PP 60-14), with the exception that thespecimensaretrimmedtoaheightof130mmtallinsteadofthestandard150mmtall.Specimensfor thehigh-modulusmixtureswerepreparedtoa targetairvoidcontentof3.5±0.5%aftertrimming.Aminimumoffourspecimensweretestedperuniquemixture.GuidanceinAASHTOTP107-14wasusedtoselectstrainlevelsfortestingthatprovidearangeofcyclestofailure(Nf).
Toconductthistest,anAMPTsampleisgluedwithasteelepoxytotwoendplatens.ThesampleandendplatensarethenattachedwithscrewstotheactuatorandreactionframeoftheAMPTpriortoinstallingon-specimenLVDTs.AphotoofthistestsetupisshowninFigure23.
Leiva-Villacorta,Taylor,andWillis
42
Figure23IPCGlobal®AMPTS-VECDFatigueTestSetup
Therecommendedtemperatureforthecyclicfatiguetestistheaveragetemperatureofthehighand low PG grade of the base binder,minus three degrees Celsius. ThemaximumallowabletemperatureaccordingtoAASHTOTP107-14is21°C.Themaximumallowabletemperatureof21°Cwasnecessaryinordertotestthehigh-modulusmixturesfromthisstudy.Thereasonforthemaximumtemperatureistoavoidviscoplasticeffectsduringthetest.Thisresultsinasimplermodelbecausestraindecompositionisnotneeded
Thefatiguetestisperformedatafrequencyof10Hzandconsistsoftwophases.First,asmallstrain(50to75on-specimenmicrostrain)testisperformedtodeterminethefingerprintdynamicmodulusof the sample. This is conducted todetermine the ratioof the finger-printdynamicmodulus(|E*|FP)ofthetestingsampletothedynamicmodulusdeterminedfromAMPTdynamicmodulus testing (|E*|LVE). This value is known as the dynamic modulus ratio (DMR) and isrecommendedtobebetween0.9and1.1(Equation12)(Houetal.,2010).ThisratioisusedforcontrollingthequalityofthefatiguetestingandisincorporatedintotheS-VECDfatiguemodel(Houetal.,2010).
Secondly,thesampleissubjectedtoafatiguetestinwhichtheAMPTactuatorisprogrammedtoreach a constant peak actuator displacement with each loading cycle. During this test, thedynamicmodulusandphaseangleofthesamplearerecorded.Failureofthesampleisdefinedasthepointatwhichthephaseanglepeaksandthendropsoff(Houetal.,2010).ThisconceptisdemonstratedgraphicallyinFigure24.
𝐃𝐌𝐑 = 𝐄∗ 𝐅𝐏𝐄∗ 𝐋𝐕𝐄
(12)
AnAMPTsamplegluedwithsteelepoxytorigidly
mountedendplatens.
Leiva-Villacorta,Taylor,andWillis
43
Figure24DeterminationofCyclestoFailureforS-VECDFatigueTest
S-VECDanalysiswasperformedusingtheALPHA-Fatigue(v3.1.5)analysissoftwaredevelopedby Underwood and Kim. The ALPHA-Fatigue software produces two outputs from the inputdynamicmodulusandfatiguetests:thedamagecharacteristiccurveandtheenergy-basedfailurecriterion.Thedamagecharacteristiccurve(orCvs.Scurve)plotsthepseudosecantmodulus(C)ofthemixtureagainstitsdamageparameter(S).Practically,thisillustrateshowfatiguedamageevolves in a unique asphalt mixture (Jacques et al., 2016). For this study, this model wasgeneratedusinganexponentialfunction(Equation13).
𝐂 = 𝐞𝐚𝐒𝐛 (13)
where
C = pseudosecantmodulus,S = damageparameter,and
a,b = modelcoefficients.
The second output from the ALPHA-Fatigue software is energy-based failure criterion, or GRmethod(SabouriandKim,2014).TheGRtermisdefinedastherateofchangeoftheaveragedreleasedpseudostrainenergy(percycle)throughoutthetest(SabouriandKim,2014).TheGRtermcharacterizestheoverallrateofdamageaccumulationthroughfatiguetesting(Jacquesetal.,2016).AplotofGRversuscycles to failure (Nf)canbegenerated fromtheALPHA-Fatigueanalysis, and the slopeandpositionof these curves canbeused togage the relative fatigueresistanceofonemixturetoanother(Jacquesetal.,2016).
Leiva-Villacorta,Taylor,andWillis
44
AsummaryoftheresultsfromtheindividualS-VECDtestsisincludedinTable18below.Ofthe24individualspecimenstested,16specimenshadaDMRvalueintherecommendedrange(0.9to1.1).TheremainingeightspecimenshadborderlineDMRvaluesofbetween0.84and0.90,indicatingamaximum16%disconnectbetweenthemixtureE*testedduringdynamicmodulusandE*verifiedduringthefatiguetesting.Theseresultswerenotexcludedfortworeasons.First,thetestingwasbeingperformedonunique(high-modulus)materialsatanon-standardairvoidcontent.Secondly,thesespecimenswerenotdetrimentaltothequalityoftheGRversusNfmodeldiscussedhereafter.
Thedamagecharacteristic(Cvs.S)curvesforthisprojectareshowninFigure25whiletheenergyrelease(GRvs.Nf)curvesareshowninFigure26.ApowermodelofstandardformwasfittotheGRversusNfcurves,withthemodelcoefficientssummarizedinTable19.Figure25showsthreeofthemixtures(35%RAPPG76-22,50%RAPPG76-22,and25%RAP-5%RASPG76-22)tohavevirtuallyidenticaldamagecharacteristiccurves,whiletheEME14and35%RAPHiMAmixtureshavethegreateststiffnessasadditionaldamageisappliedtothespecimens.TheenergyreleasecurvesallhadpowermodelR2valuesof0.94orabove,indicatingagoodmodelfit.ThecurvewiththehighestslopeandhighestinterceptwastheEME14mixture.Thisindicatesthatatlowenergyreleaserates(10or100),thismixturehaspoorfatigueresistancerelativetotheothermixturedesigns.Threeofthemixtures(50%RAPPG76-22,25%RAP-5%RASPG76-22,and35%RAPHiMA)hadvirtually identical slopesat the lowendof thespectrum, indicating improvedfatigueresistancerelativetotheothertwomixtures.The35%RAPHiMAmixturehadthehighestinterceptofthisgroupingofthreemixturesandisfurthertotherightoftheplotinFigure26,indicatingitwouldbethemostfatigueresistantmixtureinthisgrouping.
Leiva-Villacorta,Taylor,andWillis
45
Table18SummaryofS-VECDIndividualTestResults
MixtureID SpecimenID
AirVoids(%)
|E*|LVE(MPa)
|E*|FP(MPa) DMR Nf GR
EME14 9 3.0 14,694 14,715 1.001 7,795 270.9EME14 11 3.8 14,643 14,102 0.963 47,682 12.9EME14 12 3.9 14,694 14,172 0.964 2,835 489.7EME14 13 3.7 14,694 14,614 0.995 73,284 6.8
35%RAPPG76-22 326 3.2 13,954 11,904 0.853 47,685 20.035%RAPPG76-22 327 4.0 13,954 13,314 0.954 1,175 2,037.935%RAPPG76-22 328 4.0 13,954 11,952 0.857 11,500 217.035%RAPPG76-22 329 4.0 13,899 11,992 0.863 39,060 33.135%RAPPG76-22 330 4.0 13,954 11,795 0.845 6,035 282.650%RAPPG76-22 221 3.5 14,939 14,374 0.962 3,275 654.050%RAPPG76-22 222 3.2 14,939 15,161 1.015 3,795 333.350%RAPPG76-22 223 3.3 14,882 12,484 0.839 214,224 4.750%RAPPG76-22 224 3.7 14,882 13,001 0.874 40,900 34.650%RAPPG76-22 225 3.2 14,939 14,021 0.939 69,991 36.4
25%RAP-5%RASPG76-22 128 3.1 14,815 13,458 0.908 795 2,644.825%RAP-5%RASPG76-22 131 3.8 14,815 12,501 0.844 1,195 747.225%RAP-5%RASPG76-22 132 3.9 14,764 12,532 0.849 4,195 317.525%RAP-5%RASPG76-22 134 3.9 14,764 13,282 0.900 78,330 14.025%RAP-5%RASPG76-22 135 3.4 14,764 13,233 0.896 4,075 610.725%RAP-5%RASPG76-22 136 3.8 14,764 13,303 0.901 1,655 1,385.5
35%RAPHiMA 421 3.6 13,454 12,869 0.957 55,031 33.535%RAPHiMA 422 3.2 13,563 13,014 0.960 1,635 1,730.335%RAPHiMA 423 3.4 13,454 13,788 1.025 7,295 521.035%RAPHiMA 425 3.5 13,398 12,694 0.947 3,255 462.4
Leiva-Villacorta,Taylor,andWillis
46
Figure25S-VECD:CversusSCurves
Figure26S-VECD:GRversusNfCurves
Leiva-Villacorta,Taylor,andWillis
47
Table19SummaryofS-VECDGRvs.NfPowerModelCoefficients
MixtureID α1 α2 R2FrenchEME -1.422 5.512E+07 0.973
35%RAPPG76-22 -1.207 1.135E+07 0.98150%RAPPG76-22 -1.053 2.695E+06 0.963
25%RAP-5%RASPG76-22 -1.087 3.229E+06 0.95135%RAPHiMA -1.059 3.925E+06 0.940
5. AASHTOWAREPAVEMENTMEDESIGNANALYSIS
OneoftheobjectivesofthisstudywastousetheAASHTOWarePavementMEDesignsoftwaretodeterminehowahigh-modulusbasecanaffect theperformanceofasphaltpavements.Toachievethisobjective,thePavementMEDesignfileforsectionS9inthe2009NCATTestTrackwasutilizedtoperformthesimulation.Level1inputwasusedinthePavementMEsimulationsforalllayers.ThemeasureddynamicmodulusinSection4.3wasusedinthesimulationofhigh-modulusbaselayersinPavementMEDesignsoftware.Fiveyearsofdesignlifewereusedinthesimulation.PavementconstructioninformationisshowninTable20.
Six simulated scenarios were used to determine the effect of different high-modulus basemixturesontheperformanceofasphaltpavements.Eachscenarioisexplainedasfollows:
1. Simulation 1 utilized material properties from the 2009 Test Track Cycle Section S9(controlsection).
2. Simulation2wasdesignedtodeterminehowahigh-modulusmixturedesignedbasedonaFrenchmixturedesignprocedurecanaffecttheperformancepredictedbyPavementMEDesign.
3. Simulation 3 was designed to determine how a 35% RAP mixture can affect theperformancepredictedbyPavementMEDesign.ThebindergradewasPG76-22andthemixturewaslabeled35%RAPPG76-22NoLime.
4. Simulation4wasplannedtodeterminehowa25%RAP+5%RASmixturecanaffecttheperformancepredictedbyPavementMEDesign.ThebindergradewasalsoPG76-22andthemixturewaslabeled50%AgedBinderPG76-22NoLime.
5. Simulation 5 was planned to determine how a 35% RAPwith high polymer-modifiedasphaltbinder (HiMA)mixturecanaffect theperformancepredictedbyPavementMEDesign.ThemixtureutilizedSBSfromKratonwithNoLime.
6. Simulation 6 was designed to determine how a 50% RAP mixture can affect theperformancepredictedbyPavementMEDesign.ThebindergradewasPG76-22andthemixturewaslabeled50%RAPPG76-22NoLime.
Leiva-Villacorta,Taylor,andWillis
48
Table20SimulationPlan
Simulation1
Simulation2
Simulation3
Simulation4
Simulation5
Simulation6
Name Controlbase
FrenchEME 35%RAP 25%RAP,
5%RAS35%RAPHiMA 50%RAP
SurfaceAC:1.2in 9.5mmPG76-22BinderAC:2.8in 19mmPG76-22
BaseAC:3.0inUnmodifiedMixturePG
67-22
FrenchEME14
35%RAPPG76-22NoLime
50%AgedBinderPG76-22NoLime
35%RAPKratonNo
Lime
50%RAPPG76-22NoLime
BaseBinderPG PG67-22 PG88-16 PG76-22 PG76-22 PG94-28 PG76-22Granularbase:5.8in Crushedstonegranularbase
Subgrade TestTracksubgrade
5.1 Traffic
ThetruckfleetattheNCATTestTrackrunsatatargetspeedof45mph,andoperates16hoursdaily,sixdaysaweekforeachtwo-yearcycle.Eachofthetruckscompletesabout680milesperdaysoastoapply10millionESALscollectivelyintwoyears.Thankstosimpletruckpatternsandrunning schedules, input Level 1 for traffic information was precisely characterized for theMEPDGanalysis.TrafficinformationisdisplayedinTable21.Traffickingatthe2009NCATTestTrackwasconductedusingfourtripleflat-bedtrailertrucks(Figure27)andonetripleboxtrailerloadedthepavementfromMondaytoSaturday.Table22providestheaxleweightsforeachofthefivetrucksundernormalloadingconditions.
Figure27TripleFlat-BedTrailerTruckatNCATTestTrack
Leiva-Villacorta,Taylor,andWillis
49
Table21TrafficInformation
Age(year) HeavyTrucks(cumulative)2009(initial) 3,0822011(2years) 2,814,2502014(5years) 5,628,500
Table22AxleWeights(lbs)forTruckingFleetatNCATTestTrack
Truck# Steer FrontDriveTandem
RearDriveTandem
Single#1
Single#2
Single#3
Single#4
Single#5
1 9,400 20,850 20,200 20,500 20,850 20,950 21,000 20,2002 11,200 20,100 19,700 20,650 20,800 20,650 20,750 21,2503 11,300 20,500 19,900 20,500 20,500 21,000 20,650 21,1004 11,550 21,200 19,300 21,000 21,050 21,000 20,750 20,8005 11,450 20,900 19,400 20,100 20,450 21,000 20,050 20,650
Average 11,450 20,900 19,400 20,100 20,450 21,000 20,050 20,650
Inordertorepresentatripletrailer,twofictitiousvehicleclasseswereusedtogetherwithfivesingleaxlesandonetandemaxlefromtheClass13,andtheremainingonesingleaxlefromtheClass12.Theaverageaxlewidthwas8.5ft,thedualtirespacingwas13.5in,andthetirepressurewasapproximately100psi.Othertrafficinputs(i.e.,lateraltrafficwander)wereassumedtoberoutinedesignvalues,andtheywereleftasthedefaultsprovidedbytheMEPDG.Therewasnoannualtrafficgrowth.
5.2 Climate
The climatic data required in theMEPDG is usedby the Enhanced IntegratedClimateModel(EICM)tocalculatechangesinthetemperatureandmoistureprofilethroughoutthepavementcrosssection.TheclimaticinputfortheMEPDGisactuallyafilethatcontainsarecordedhistoryoftemperature,rainfall,windspeed,humidity,andsunlightconditionsforaspecificarea.TherearetwowaystopreparetheclimaticinputsfortheMEPDG,eitherbyselectingaclimaticdatafileforrepresentativeareasorbypreparinganewclimaticdatafilebasedonalocalweatherstation.ThelatterwasadoptedinthisstudybecausetheTestTrackhasanon-siteweatherstation(Figure28),whichisresponsibleforcollectingenvironmentalinformationonanhourlybasis.TheTestTrackisatageographiccoordinateof32°59´N,-85°30´W,andanelevationof600ft.Thenextsectionwillcoverthemethodtoprepareaclimatefileforaparticularcondition.
Leiva-Villacorta,Taylor,andWillis
50
Figure28TestTrackOn-SiteWeatherStation
ItisnotedthattwoformatsoffilesfunctionintheMEPDG:theICMfileandthehourlyclimaticdatabasefile.TheICMfilewasgeneratedbytheMEPDGcalculationbasedonanhourlyclimaticdatabasefile.Infact,thehourlyclimaticdatabasefilewaseithergivenforthoserepresentativeareasorcanbeself-developed.
AASHTOWarePavementMEDesignsoftwarewasusedtocomputethefollowingdistressestosimulatethepavementperformanceoverafive-yearperiodofanalysis:
• InternationalRoughnessIndex(IRI),• Top-downcracking,• Bottom-upfatiguecracking,• Thermalcracking,• Totalpavementrutting,and• Ruttingintheasphaltconcretelayer.
5.3 EstimatedPerformance
Figure29showstheestimated layermoduli forall sixscenarios.Asexpected,allof thehigh-modulusbasecoursesexhibithigherlayermodulithroughoutthefive-yearperformanceperiodofanalysis.Theresults indicatethathigh-modulusbasecoursescanhavehigher layermodulirangingfrom1.5to2.0timesthelayermodulusofthecontrolsection.However,nosignificantdifferencesamongthehigh-modulusmixturescanbeobserved.Furtheranalysisindicatesthatthe 35% RAP HiMA mixture exhibits lower layer moduli, especially at high temperatures,compared to the remaining high-modulus mixtures. This behavior is in agreement with thedynamicmodulustestresultsforthehightemperaturelowfrequencyrange.However,the25%RAP-5%RASbasecoursedoesnotexhibithighermodulusatthelowtemperaturehighfrequencyasexpectedfromtheE*testresults.
Leiva-Villacorta,Taylor,andWillis
51
Figure29EstimatedBaseACLayerModulus
Crackingoftheasphaltlayerwasexpectedtodecreasebytheuseofhigh-modulusbasecourses.Figures 30 and 31 show estimated results of cracking. Itwas observed that the use of high-modulusbasecoursescouldreducethebottom-upcracking,whichwasreasonablesincehigh-modulusbasemayreducethetensilestressandstrainatthebottomofthebinderlayer.Thisreductionincrackingcouldrangefrom20%to25%.Moreover,itcanbeobservedthattheeffectontop-downcrackingcanbemoresignificantwithadecreaseincrackingrangingfrom28%to35%.Inthiscase,the35%RAPHiMAmixtureseemstobetheleastresistanttofatiguecrackingofallHMACmixturesandtheEMEmixtureseemstoshowthebestperformance,contrarytotheresultsobtainedwiththeS-VECDtestresults.However,theobservedtrendsinfatiguecrackingperformanceforthefivehigh-modulusdesignscanbeconsideredsimilarforpracticalpurposes.
Figure30EstimatedBottom-UpFatigueCracking
0
500
1000
1500
2000
2500
3000
3500
4000
0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0
ACM
ODU
LUS,KSI
PAVEMENTAGE,YEARS
Control FrenchEME 35%RAP25%RAP5%RAS 35%RAPHiMA 50%RAP
0
5
10
15
20
25
30
35
40
45
0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0
BOTTOM-UPCR
ACKING,%
PAVEMENTAGE,YEARS
Control FrenchEME 35%RAP
25%RAP5%RAS 35%RAPHiMA 50%RAP
Leiva-Villacorta,Taylor,andWillis
52
Figure31EstimatedTop-DownFatigueDamage
Figure32showstheestimatedresultsofpermanentdeformationoftheasphaltconcretelayer.Itwasobservedthatusinghigh-modulusbaselayermaterialswouldhavenosignificanteffectontheruttingoftheentireasphaltconcretesection.Thiscanbeexplainedbythesmallobserveddifferences among high-modulus base courses as well as the use of the same top andintermediateAClayersinthesimulations.Areductioninlessthan4%oftheAClayerrutdepthwasestimated.Whencomparingthesimulatedpermanentdeformationandtheresultsfromtheflownumbertesting,nocorrelationintheresultswasevident.
Figure32EstimatedPermanentDeformationoftheEntireACLayer
Figure33showsthepredictedperformanceofthesixpavementstructuresintermsofIRI.Itwasobservedthattheuseofahigh-modulusbasecanreducetheIRIfrom5.6%to6.7%relativeto
0
20
40
60
80
100
120
140
160
180
0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0
TOP-BO
TTOMDAM
AGE,%
PAVEMENTAGE,YEARS
Control FrenchEME 35%RAP
25%RAP5%RAS 35%RAPHiMA 50%RAP
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0
ACRUTDE
PTH,IN
PAVEMENTAGE,YEARS
Control FrenchEME 35%RAP25%RAP5%RAS 35%RAPHiMA 50%RAP
Leiva-Villacorta,Taylor,andWillis
53
thecontrolsection,andnosignificantdifferenceinIRIwasfoundinthesimulationsamonghigh-modulusbasecourses.
Figure33EstimatedSurfaceRoughness
Finally,Table23containsasummaryofthermalcrackingperformanceandpercentchangewithrespecttothecontrolsection.Asshowninthistable,high-modulusbasecoursesmayhavenoeffectonthethermalcracking.Thisisexpectedsinceallthesimulationshadthesamesurfaceandbinderlayerswhosepropertiesmainlyaffectthedevelopmentofthermalcracking.
80
90
100
110
120
130
140
0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 2 . 5 3 . 0 3 . 5 4 . 0 4 . 5 5 . 0
IRI,IN/M
ILE
PAVEMENTAGE,YEARS
Control FrenchEME 35%RAP25%RAP5%RAS 35%RAPHiMA 50%RAP
Leiva-Villacorta,Taylor,andWillis
54
Table23SummaryofEstimatedPerformance
DistressType TargetPredicted
Control FrenchEME35%RAP
25%RAP5%RAS
35%RAPHiMA
50%RAP
TerminalIRI(in/mile) 172 134.49 125.51 126.56 126.37 127.02 126.13
Permanentdeformation:totalpavement(in) 0.75 0.57 0.55 0.55 0.55 0.55 0.55
ACbottom-upfatiguecracking(%lanearea) 25 38.21 28.61 29.81 29.61 30.31 29.31
ACthermalcracking(ft/mile) 0 27.17 27.17 27.17 27.17 27.17 27.17
ACtop-downfatiguecracking(ft/mile) 2000 10304 6695 7202 7099 7419 7006
Permanentdeformation:AConly(in) 0.25 0.46 0.44 0.44 0.44 0.44 0.44
PercentReductionvs.ControlTerminalIRI(in/mile)
6.7% 5.9% 6.0% 5.6% 6.2%
Permanentdeformation:totalpavement(in) 3.5% 3.5% 3.5% 3.5% 3.5%
ACbottom-upfatiguecracking(%lanearea) 25.1% 22.0% 22.5% 20.7% 23.3%
ACthermalcracking(ft/mile) 0.0% 0.0% 0.0% 0.0% 0.0%
ACtop-downfatiguecracking(ft/mile) 35.0% 30.1% 31.1% 28.0% 32.0%
Permanentdeformation:AConly(in) 3.6% 3.6% 3.6% 3.6% 3.6%
Figure34measureddistressesonsectionS9attheNCATTestTrackfrom2009to2014.ItcanbeobservedthatperformancewashighlyoverestimatedforsectionS9.Themaximummeasuredrutdepthwas0.34inchesoftheentirestructure,whiletheestimatedmaximumrutdepthwas0.56inches.MeasuredIRIstartedat60in/mileanddidnotchangesignificantlyovertimewithafinal IRI of 80 in/mile. On the contrary, the initial estimated IRI was 94 in/mile, which wasexpectedtoincreaseovertimetoreachafinalIRIof135in/mile.Finally,itcanbeobservedthatcrackingshowedthelargestoffset.Initialmeasuredcrackingwasobservedafter11millionESALsormorethantwoyearsoftrucktraffickingandreachingonly10%,whilecrackingisexpectedtoappearduringthefirstyearandsignificantlyincreaseovertime.Theseresultscanbeexplaineddue to the application of nationally calibrated transfer functions or default functions in theemployedsoftware.Therefore,notonlyislocalcalibrationrequiredforsectionS9,butcalibrationoftheHMACmixturesusedinthisstudymaybeneededtofurtherreflecttheirbenefitasbasecourses.
Leiva-Villacorta,Taylor,andWillis
55
a.
b.
c.
Figure34MeasuredPavementPerformanceonSectionS9:a.Rutting,b.Roughness,c.
Cracking
Insummary,theuseofhigh-modulusbasecoursescouldimprovetheoverallperformanceofanasphaltconcretelayerandtheentireflexiblepavementstructure.Thetypeofdistressthatmaybethemostaffectedistop-bottomcrackingfollowedbybottom-upfatiguecracking.Ridequalitycanalsobeslightlyimproved(lowerIRI),andnosignificanteffectonruttingandthermalcrackingshouldbeexpected.
InPavementME,theperformancepredictionusestransferfunctionsthatarecalibratedwiththeexisting performance data inwhich stiffer layers correspond to better pavements. Figure 29showed lower modulus for the 35% RAP HiMA mixture, and as calibrated, Pavement MEpredictedleastresistanttofatiguecracking(highercracking).However,theuniaxialfatiguetestshowed that the35%RAPHiMAmixture is themost fatigue resistant,even though the layermoduliislower.
0
0.1
0.2
0.3
0.4
0 5 10 15 20
RutD
epth,in
ESALs,Millions
0
20
40
60
80
100
0 5 10 15 20
IRI,in/m
ile
ESALs,Millions
0
2
4
6
8
10
12
0 5 10 15 20
Cracking,%
lane
ESALs,Millions
Leiva-Villacorta,Taylor,andWillis
56
WiththeempiricalnatureofthePavementMEtransferfunctions,performanceofnewmaterialscannotbereliablymodelledwiththecurrenttransferfunctions.Ahighmodulusbrittlematerialwillhavedifferentfatiguebehaviorthanahighmodulusductilematerial.Therefore,thecurrenttransferfunctionsinPavementMEshouldbecalibratedwithlaboratoryandfieldperformanceofHMACmixturestohavereliablepredictions.ThebetterPavementMEpredictedperformanceofthenewmaterialislikelyacombinationofitshighermodulusandalsoanartifactofnotsoapplicabletransferfunction.Untilthelatterisresolved,quantifyingthefieldperformanceoftheformerthroughPavementMEwouldbesomewhatinconclusive.
6. CONCLUSIONSANDRECOMMENDATIONS
Theobjectiveofthisprojectwastodevelopandvalidatemixturedesignsandevaluatepredictedperformanceeffectsofhigh-modulusbaselayers.Basedonexperimentalresultsandstructuralanalysis,thefollowingconclusionsandrecommendationsaremade:
• European mix design standard methods and specifications were successfullyimplementedon local (U.S.)materials.TheLevel3 requirement fordynamicmodulus,14,000MPaat15°Cand10Hz,wasmetforallHMACmixtures.
• Reclaimed asphalt pavement canbeused to stiffen the asphalt binder sufficiently forhigh-modulusasphaltmixtures.TheminimumRAPcontentutilizedinthisinvestigationwas35%.
• NosignificantdifferencesindynamicmoduluswereobtainedforallHMACmixturesatlow temperature and high frequency. However, on the opposite side of thetemperature/frequencyspectrum,the25%RAP,5%RASmixtureprovidedsignificantlyhigherE*valuesduetotheinclusionofRAS.
• Flow number test results were significantly greater than those of the conventionalmixture.
• BasedonAMPTcyclictestresults,fatiguepropertiesamongHMACmixturesseemedtoimproveforthehighpolymer-modifiedmixturesandseemedtodecreasefortheEMEFrenchmixture,whichhasastiffervirginbinder.
• Increased stiffness of HMAC mixtures improves MEPDG predicted performance ofpavement in rutting, fatiguecracking,andridequalitycomparedtoconventionalbasecourses.Agreatimprovementinfatiguecracking(topdownandbottomup)andsomeimprovementinruttingandridequality.
• Correlations between laboratory performance trends among HMAC mixtures andpredicted structural performance were not obtained. This was attributed to thesignificant role that other materials/properties have on pavement responses andperformance.Forinstance,lowermodulusfor35%RAPHiMAmixtureandascalibratedPavementMEpredictedhighercracking.However,theuniaxialfatiguetestshowedthatthe35%RAPHiMAmixtureisthemostfatigueresistanceeventhoughthelayermoduliislower.
• CalibrationofMEPDGtransferfunctionsapplicabletoHMACmixturesisrecommendedtoobtainmorerepresentativeperformancepredictions.
Leiva-Villacorta,Taylor,andWillis
57
• Basedontheresultsofthisstudy,adetailcostandbenefitassessmentisrecommendedinordertofurtherquantifytheeffecttheHMACmixtureshaveonpotentiallong-lastingperpetual-typeflexiblepavements.
• Thisstudywasbasedonalaboratoryexperiment.TheabilitytoproduceHMACmixturesthroughahotmixplantandtosuccessfullylayandcompactthemintheU.S.hasnotbeendemonstrated.Mixing and compacting in the laboratory suggest that field operationswouldbemoredifficultwithHMACmixtures; theEME’sdesignand testinghas tobeadaptedtoU.S.standardsandconditions.
7. RECOMMENDEDMIXTUREDESIGNPROCEDURE
Basedontheresultsofthisresearchstudyandthecurrentstate-of-the-practice,thefollowingstepsarerecommendedasHMACmixturedesignprocedure:
1. DeterminetheaggregatetrialblendfortheHMACmixture.2. DeterminetheminimumasphaltbindercontentusingtheFrenchmethodbasedonthe
aggregatesurfacearea(Equations1to4).However,theAsphaltInstituteHveem-Edwardequationcanbeusedsuccessfullyhere.
3. Set Ndes with the Superpave gyratory compactor to 80 gyrations and compact designsamplestotargetairvoidslowerthan6%.
4. Preparethreetrialdynamicmodulussamplescompactedto3.0-6.0%airvoidsaccordingtotheFrenchmethodologyandtestat15°Cand10Hz.
5. SelectoptimumbindercontenttomeetE*=14,000MPa(at15°Cand10Hz)tomeettheminimumasphaltcontentfromstep2andtomeetNdesspecimenstargetairvoidslowerthan6%.
a. Adjusting the gradation or mixture components (additives, recycled material,bindergrade,etc.)maybenecessarytomeetE*andairvoidsrequirements.
b. For each gradation adjustment, the minimum AC required will need to berecalculated.
6. Selectlaboratoryperformancetestsandcriteria(rutting,cracking,andmoisturedamage)forfurtherverificationandconductAASHTOTP79-15todeterminedynamicmodulustobeusedinMEsimulations.
MEsimulationsshouldbeusedforrelativecomparisonpurposesandnotforstructuralpavementdesignuntilfieldvalidationhasbeenperformed.Pilotprojectsareaproventoolforvalidatingandfine-tuningnewpracticesresultingfromresearch.Usingtraditionalprojectsasabenchmark,pilotprojectsorprogramshavebeenusedextensivelytomeasuretherelativesuccessofnewspecificationsandtestmethods.Theresultsofpilotprojectshaveservedtoeffectivelypromotethe long-term implementation of new industry practices. It is recommended that an agencychampiontheuseof theproposedstandardmethodologies fordesign,analysis,construction,andspecificationsrelatedtoHMACimpactonpavementperformance.
Leiva-Villacorta,Taylor,andWillis
58
REFERENCES
Airey,G.D.UseofBlackSpaceDiagramstoIdentify Inconsistencies inRheologicalData.RoadMaterialsandPavementDesign,Vol.3,No.4,2002,pp.403-424.
Anderson,R.M.,G.N.King,D.I.Hanson,andP.B.Blankenship.EvaluationoftheRelationshipbetweenAsphaltBinderPropertiesandNon-LoadRelatedCracking.JournaloftheAssociationofAsphaltPavingTechnologists,Vol.80,2011,pp.615-664.
Bankowski,W.,M.Tusar,L.G.Wiman,D.Sybilski,M.Gajewski,R.Horodecka,M.Maliszewksi,and K. Mirski. Laboratory and Field Implementation of High Modulus Asphalt Concrete.Requirements for HMACMix Design and Pavement Design. European Commission Research,Paris,2009.
Brosseaud,Y.SmallOverviewofHighModulusAsphaltMixesEME.PresentedatAustralianScanTour,June2012.
Brosseaud,Y.,F.Farcas,andV.Mouillet.HighModulusAsphaltMixeswithHighRateofRA:WhatDoesitHappen?InCongrèsEurobitumeEurasphalt,EE2012,Turkey,October2012.
Bueche,N.,A.G.Dumont,A.Vanelstraete,J.deVisscher,S.Vansteenkiste,F.Vervaecke,J.Maek,L.Gaspar,andF.Thogersen.LaboratoryandALT-EvaluationofHighStiffnessUnderlayerswithHighPercentageofRe-useasDevelopedintheNR2C-Project.Proceedingsofthe4thEurasphaltandEurobitumeCongress,Copenhagen,Denmark,2008.
Capitãio, S. D., and L. Picado-Santos. Assessing Permanent Deformation Resistance of HighModulusAsphaltMixtures.JournalofTransportationEngineering,Vol.132,No.5,2006,pp.394-401.
Carbonneau,X.,J.P.Michaut,T.Anderson,C.Thorup,andL.Ladenhoff.HighModulusGABII:ADanishExperiment.Proceedingsofthe4thEurasphaltandEurobitumeCongress,Copenhagen,Denmark,2008.
Chappat,M.,X.Carbonneau,andY. Lefuevre.Analysisof theMeasuredPerformanceofHighModulus Asphalt (HMA) and Road Base Asphalt (RBA) and a Proposal for a CompensationPrinciple.EuropeanRoadsReview,14,2009,pp.72-83.
ChristensenJr.,D.W.,T.Pellinen,andR.F.Bonaquist.HirschModelforEstimatingtheModulusofAsphaltConcrete.JournaloftheAssociationofAsphaltPavingTechnologists,Vol.72,2003,pp.97–121.
Corté, J.DevelopmentandusesofHard-GradeAsphaltandofHigh-ModulusAsphaltMixes inFrance.TransportationResearchCircular,503,2001,pp.12-31.
Leiva-Villacorta,Taylor,andWillis
59
Daniel,J.S.,andY.R.Kim.DevelopmentofaSimplifiedFatigueTestandAnalysisProcedureusinga Viscoelastic Continuum Damage Model. Journal of the Association of Asphalt PavingTechnologists,Vol.71,2002,pp.619–650.
DeBacker,C.,L.Glorie,andR.Reynaert.TestsectionsinHigh-ModulusAsphalt:AComparativeExperiment with Ten Variants. Proceedings of the 4th Eurasphalt and Eurobitume Congress,Copenhagen,Denmark,2008.
DeBacker, C., J. De Visscher, L Glorie, A. Vanelstraete, S. Vansteenkiste, and L. Heleven. AComparativeHigh-ModulusAsphaltExperimentinBelgium.InProceedingsofTransportResearchArenaEurope,2008,pp.21-24.
DeLaRoche,C.,H.Odeon,J.Simoncellli,andA.Spernol.StudyoftheFatigueofAsphaltMixesUsingtheCircularTestTrackoftheLaboratoireCentraldesPontsetChausséesinNantes,France.TransportationResearchRecord1436,TRB,NationalResearchCouncil,WashingtonD.C.,1994,pp.17-27.
DeVisscher,J.,S.Vansteenkiste,andA.Vanelstraete.TestSectionsinHigh-ModulusAsphalt:MixDesignandLaboratoryPerformanceTests.Proceedingsof the4thEurasphaltandEurobitumeCongress,Copenhagen,Denmark,2008.
Delorme, J. L., C. De La Roche, and L. Wendling. LPC Bituminous Mixtures Design Guide.LaboratoireCentraldesPontsetChaussées,Paris,2007.
Denneman, E.,M.Nkgapele, J. Anochie-Boateng, and J.W.Maina. Transfer ofHighModulusAsphalt Mix Technology to South Africa. In Proceedings of the 10th Conference on AsphaltPavements for SouthernAfrica, ChampagneSportsResort,CentralDrakensberg, SouthAfrica,2011.
Denneman,E.HighModulusAsphalt(HiMA)Trial:MixandPavementDesign.Presentedatthe22ndMeetingoftheRoadsPavementsForum(RPF),Pretoria,November2011.
Denneman,E.,andM.Nkgapele.InterimGuidefortheDesignofHighModulusAsphaltMixesandPavementsinSouthAfrica.ReportCSIR/BE/IE/ER/2010/0042/B,2011.
DesCriox,P.,andL.Planque.ExperiencewithOptimizedHardGradeBitumensinHighModulusAsphaltMixes.InProceedingsofthe3rdEurasphaltandEurobitumeCongress,Vienna,Austria,2004.
Diefenderfer,B. K. andG.W.Maupin.FieldTrialsofHigh-ModulusHigh-Binder-ContentBaseLayerHot-MixAsphaltMixtures.ReportNo.FHWA/VTRC11-R2.VirginiaTransportationResearchCouncil,Charlottesville,Va.,2010.
Distin, T., L. Sampson, H. Marais, and B. Verhaeghe. High Modulus Asphalt: Assessment ofViabilityBasedonOutcomesofOverseasFactFindingMission.WhitePaper,2006.
Leiva-Villacorta,Taylor,andWillis
60
EuropeanAsphaltPavementAssociation.AirfieldusesofAsphalt.Netherlands,2003.
European Asphalt Pavement Association. High Modulus Asphalt: A State of the Art Report.Netherlands,2005.
Guyot,X.UseofHighModulusAsphalt-“EME”CaseStudiesintheIndianOceanArea.15thAAPAInternationalFlexiblePavementsConference,Brisbane,Australia,2013.
Haritonovs,V., J.Tihonovs,M.Zaumanis,andA.Krasnikovos.HighModulusAsphaltConcretewithDolomiteAggregate.TransportResearchArena.Paris,France.2014.
Hou, T., B. S. Underwood, and Y. R. Kim. Fatigue Performance Prediction of North CarolinaMixturesUsingtheSimplifiedViscoelasticContinuumDamageModel.JournaloftheAssociationofAsphaltPavingTechnologists,Vol.79,2010,pp.35-80.
Jacques,C.,J.S.Daniel,T.Bennert,G.Reinke,A.Norouzi,C.Ericson,W.Mogawer,andY.R.Kim.Effect of Silo Storage Time on the Characteristics of Virgin and ReclaimedAsphalt PavementMixtures.TransportationResearchRecord: Journalof theTransportationResearchBoard,No.2573,Washington,D.C.,2016,pp.76-85.
Jamois,D., J. C. Vaniscote, Y. Jolivet, andM.Malot.Development of a Concept of VeryHighModulusBituminousMacadamforPavementBaseCourses.Proceedingsofthe2ndEurasphaltandEurobitumeCongress,Vol.2,Barcelona,Spain,2000.
Lee,H.J.,J.H.Lee,andH.M.Park.PerformanceEvaluationofHighModulusAsphaltMixturesforLongLifeAsphaltPavements.ConstructionandBuildingMaterials,Vol.21,No.5,2007,pp.1079-1087.
Lenfant, M. High Rutting Resistant Asphalt Pavement. Seminar for Thailand Department ofHighways,TIPCOAsphaltPublicCompanyLimited,Bangkok,Thailand,269-289.
Michaut,J.P.EMEorHighModulusAsphaltConcrete.COLASPresentation.March2014.
Miro,R.,G.Valdés,A.Martinez,P.Segura,andC.Rodriguez.EvaluationofHighModulusMixtureBehavior with High Reclaimed Asphalt Pavement (RAP) Percentages for Sustainable RoadConstruction.ConstructionandBuildingMaterials,Vol.25,No.10,2011,pp.3854-3862.
Montanelli,E.F.Fiber/PolymericCompoundforHighModulusPolymerModifiedAsphalt(PMA).Procedia-SocialandBehavioralSciences,Vol.104,2013,pp.39-48.
Newcomb,D. and K. Hansen.Mix Type Selection for Perpetual Pavements. Presented at thePerpetualPavementConference,Columbus,Ohio,2004.
Newcomb,D.E.,D.H.Timm,andR.Willis.PerpetualAsphaltPavements:ASynthesis.AsphaltPavementAlliance,Lanham,Md.,2010.
Leiva-Villacorta,Taylor,andWillis
61
Nicholls, J. C., R. Elliott, N. Meite, R. Perera, A. Hunter, and J. Williams. Monitoring theIntroduction of Enrobe a Module Eleve Class 2 onto UK Roads. In Proceedings of the 4thEurasphaltandEurobitumeCongress,Copenhagen,Denmark,2008.
Nkgapele,M.,E.Denneman,andJ.K.Anochie-Boateng.ConstructionofaHighModulusAsphalt(HiMA) Trial Section Ethekwini: South Africa’s First Practical Experience with Design,ManufacturingandPavingofHiMA.Presentedat31stSouthernAfricanTransportConference,Pretoria,SouthAfrica,2012.
Petho,L.,andE.Denneman.HighModulusAsphaltMix(EME)forHeavyDutyApplicationsandPreliminaryLaboratoryTestResultsinAustralia.Presentedat15thAAPAInternationalFlexiblePavementsConference,Brisbane,Australia,2013.
Rahbar-Rastegar, R., and J. S. Daniel. Laboratory versus Plant Production: Impact ofMarerialProperties and Performance for RAP and RAS mixtures. International Journal of PavementEngineering,2016,pp.1-12.
Sabouri, M., and Y. R. Kim. Development of a Failure Criterion for Asphalt Mixtures UnderDifferent Modes of Fatigue Loading. In Transportation Research Record: Journal of theTransportation Research Board, No. 2447, Transportation Research Board of the NationalAcademies,2014,pp.117-125.
Sanders,P.J.,andM.Nunn.TheApplicationofEnrobeaModuleEleveinFlexiblePavements.TRLReportTRL636,TransportResearchLaboratory,Crowthorne,UnitedKingdom,2005.
UnderwoodB.,C.Baek,andY.Kim.SimplifiedViscoelasticContinuumDamageModelasPlatformfor Asphalt Concrete Fatigue Analysis. In Transportation Research Record: Journal of theTransportation Research Board, No. 2296, Transportation Research Board of the NationalAcademies,2012,pp.36-45.
Vaitkus,A.,andV.Vorobjovas.UseofLocalAggregateinHighModulusAsphaltConcreteLayers.Gradevinar,Vol.65,No.4,2013,pp.353-360.
Wei,O.,F.Xinghua,andW.Lianguang.ResearchonAnti-RuttingPerformanceofHighModulusAsphaltConcretePavement.JournalofHighwayandTransportationResearchandDevelopment,Vol.4,No.2,2010,pp.77-79.
Wielenski,J.C.,andG.A.Huber.EvaluationofFrenchHighModulusAsphalt(EME)inPavementStructuralDesign(MEPDG).JournaloftheAssociationofAsphaltPavingTechnologists,Vol80,2011,pp697-718.