field management of hot mix asphalt volumetric properties

Upload: prof-prithvi-singh-kandhal

Post on 30-May-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    1/18

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    2/18

    FIELDMANAGEMENTOFHOTMIXASPHALTVOLUMETRIC PROPERTIES

    By

    PrithviS.Kandhal

    AssistantDirectorNationalCenterforAsphaltTechnology

    AuburnUniversity,Alabama

    KeeY.Foo

    ResearchEngineerNationalCenterforAsphaltTechnology

    AuburnUniversity,Alabama

    JohnA.D'AngeloSeniorProjectEngineer

    FHWAWashington,D.C.

    PaperpublishedinTransportationResearchBoard,

    TransportationResearchRecord1543,1996

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    3/18

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    4/18

    ABSTRACT

    TheFederalHighwayAdministration(FHWA)DemonstrationProjectNo.74hasclearlyshown

    thatsignificantdifferencesexistbetweenthevolumetricpropertiesofthelaboratorydesignedandplantproducedhotmixasphalt(HMA)mixes.Thevolumetricpropertiesincludevoidsinthemineralaggregate(VMA)andthevoidsinthetotalmix(VTM).ThisprojectwasundertakentodeveloppracticalguidelinesfortheHMAcontractorstoreconcilethesedifferencestherebyassistingthemtoconsistentlyproducehighqualityHMAmixes.TheHMAmixdesignandfieldtestdatafrom24FHWAdemonstrationprojectswereenteredintoadatabase.Thedataincludedmixcomposition(asphaltcontentandgradation)andvolumetricproperties.Thedatawereanalyzedtoidentifyand,ifpossible,quantifytheindependentvariables(suchasasphaltcontentandthepercentagesofmaterialpassingNo.200andothersieves)significantlyaffectingthedependentvariables(suchasVMAandVTM).

    Basedontheprecedingwork,troubleshootingchartshavebeenconstructedtocorrectand

    reconciledifferencesbetweenthevolumetricpropertiesofthejobmixformulaandtheproduced

    mix.KEYWORDS:hotmixasphalt,asphaltconcrete,asphaltpavingmixtures,fieldmanagement,volumetricproperties,qualitycontrol,qualityassurance,airvoids,VMA

    ii

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    5/18

    Kandhal,Foo,&D'Angelo

    FIELDMANAGEMENTOFHOTMIXASPHALTVOLUMETRICPROPERTIES

    PrithviS.Kandhal,KeeY.Foo,andJohnA.D'Angelo

    INTRODUCTIONDemonstrationProjectNo.74,"FieldManagementofAsphaltMixes"initiatedbyFederalHighwayAdministrationstudied17mixesfrom15StateHighwayAgency(SHA)pavingprojects[1].Ofthe17mixes,therewereonlytwomixeswheretheactualproductionmetthemixdesigntargets.Tenmixesshouldhavebeenmodifiedduringproductionwhilefivemixesshouldhavebeentotallyredesigned.TheDemonstrationProjectconfirmedthatcurrentlaboratorymixdesignproceduresdonotrepresentactualmixproduction.Flawlesslaboratory-designedmixescanincurmix-relatedproblemsduringproductionwhichcanleadtoprematurepavementdeterioration.Fieldmanagementofhotmixasphalt(HMA)providesaviabletooltoidentifythedifferencesbetweenplantproducedandlaboratorydesignedHMAmixesandeffectivelyreconcilethesedifferences[2].

    DemonstrationProjectNo.74concludedthatafieldmixverificationofthematerialproducedat

    theHMAplantshouldbeincludedasasecondphaseinthedesignprocess.Mixverificationisdefinedasthevalidationofamixdesignwithinthefirstseveralhundredtonsofproduction.ThevoidpropertiesestablishedfrommixverificationprovedtobeaneffectivetoolinidentifyingmixproductionvariationsoranydifferencesbetweenplantproducedandlaboratorydesignedHMAmixes.However,themeasuresrecommendedintheprojecttocorrecttheidentifiedproblemsweresomewhatgeneralized.Itwasrecommendedthatthejobmixformula(JMF)shouldbeadjustedtomakethegradationmoreuniformand/ormovethegradationawayfromthemaximumdensityline.Itwasalsonotedthat(1)gap-gradedmixesandmixeswhichplotclosetothemaximumdensitylinearegenerallysensitivemixes,and(2)mixeswithahumpneartheNo.30sievearegenerallytendermixes.

    AsignificantamountofdatahasbeencollectedbyDemonstrationProjectNo.74.Analysisof

    thesedatacouldyieldmorepracticalguidelinestoreconciledifferencesbetweenplantproducedandlaboratorydesignedHMAmixes.

    OBJECTIVE

    TheobjectiveofthisprojectwastodeveloppracticalguidelinestoreconciledifferencesbetweenplantproducedandlaboratorydesignedHMAmixes.ThishasbeenachievedbyanalyzingthedatacollectedbytheDemonstrationProjectNo74,andidentifyingandquantifyingindependentvariableswhichsignificantlyaffectthevoidproperties(dependentvariables)oftheproducedmix.

    Theaboveobjectivehasbeenaccomplishedbycompletingthefollowingtasks:Task1-PreparationofdatabaseTask2-AnalysisofdataTask3-MethodforreconcilingdifferencesbetweenmixdesignandproductionTask4-Fieldverificationofproposedmethodofreconciliation

    TASK1:PREPARATIONOFDATABASE

    TwentyfourDemonstrationProjectNo.74reportswereobtainedfromtheFederalHighwayAdministration(FHWA).Atotalof26asphaltmixeswereusedinthese24demonstrationprojects.Thesevastamountsofdatacontainedinthesereportsweregroupedintothreemajorgroupsandenteredintoadatabase.

    1

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    6/18

    Kandhal,Foo,&D'Angelo

    ThefirstdatagroupcontainsinformationabouttheHMAplantsuchasplanttype,production

    rate,dustcollectionsystem,andtypeofmixstoragesystem.Ofthe24demonstrationprojects,19projectsuseddrummixplantsandfiveprojectsusedbatchmixplants.Baghousedustcollectionsystemwasusedin17projects,wetscrubberwasusedinfiveprojects,andcyclone

    dustcollectionsystemwasusedintwoprojects.Theseconddatagroupcontainsinformationaboutthe26HMAmixturesusedonthe24demonstrationprojectssuchasmixtype,maximumnominalaggregatesize,amountofnaturalsand,coarseaggregatetype,fineaggregatetype,LosAngelesabrasionloss,sandequivalentvalue,percenthydratedlime,andpercentreclaimedasphaltpavement(RAP)used.Thirteenmixeswereusedinsurfacecourse,sixinbindercourse,andtwoinbasecourse.Theuseoffivemixesisnotknown.Fourmixeshad9.5mm(3/8inch)maximumnominalsize.Elevenmixeshad12.5mm(inch)maximumnominalsize.Ninemixeshad19mm(3/4inch)maximumnominalsize.Twomixeshad25.4mm(1inch)maximumnominalsize.Sevenmixescontainednaturalsand.Sevenmixescontainedhydratedlime,andfourmixescontainedRAP.

    Thethirddatagroupcontainsinformationabouttheasphaltcontent,voidproperties,and

    aggregategradationspecifiedby(a)mixdesign,(b)obtainedfromtheverificationprocess,and(c)obtainedduringproduction.Basically,productiondatahasbeenanalyzedrigorouslytoidentifyandquantifyindependentvariableswhichsignificantlyaffectthevoidpropertiesoftheproducedmix.Thefollowinginformation,ifavailable,iscontainedinthethirddatagroup:(a)asphaltcontentandvoidpropertiesofJMF,verification,andproduction,(b)aggregategradationofJMF,verification,andproduction,and(c)thelocationofaggregategradationcurveatthetimeofproductionwithrespecttothemaximumdensityline(MDL).TheMDLwasestablishedaccordingtoSuperpaveLevel1mixdesignprocedures.Thelocationofaggregategradationduringproductioncanbesummarizedasfollows:15mixeswere"above"MDL,twomixeswere"slightlyabove"MDL,fivemixeswere"on"MDL,threemixeswere"slightlybelow"MDL,andonemixwas"below"MDL.Itwasobservedthatdesigngradationsandproductiongradationsweregenerallydifferent.Productiongradationsmoreaccuratelyrepresentthe

    aggregategradationfortheproject.Therefore,themaximumnominalsizeandmaximumsizeoftheaggregatefortheprojecthavebeenbasedontheproductiongradationratherthantheJMFgradationforentryintothedatabase.

    Itwasbelievedatthebeginningoftheprojectthatvoidpropertiesmaybeaffecteddifferentlyin

    surfaceandbase/bindermixesduringproductionbecauseofdifferencesinthemaximumaggregatesizes,gradations,andasphaltcontents.Toinvestigateanddetectsucheffects,thedatebasewassplitinto"surface"and"base/binder"mixes."Surface"mixisthemixwithmaximumnominalaggregatesizeequaltoorlessthan12.5mm(inch)and"Base/Binder"mixisthemixwithmaximumnominalaggregatesizemorethan12.5mm(inch).Therelationshipbetweentheindependentvariablesandvoidsinmineralaggregate(VMA)andvoidsintotalmix(VTM)wasanalyzedforeachmixtype.Nosignificantdifferencesinrelationshipwerefoundbetweenthesetwomixtypes.Therefore,thedatabasewascombinedforsubsequentanalyses.

    TASK2:ANALYSISOFDATA

    ThefocusofthisprojectareVMAandVTMoftheHMAmixproducedintheasphaltplant.Itisthereforenecessarytoidentifythosefactorsthat(1)canbecontrolledeasilyattheHMAplantand(2)haveasignificanteffectonVMAandVTMoftheproducedmix.Therefore,VMAandVTMwerechosenasdependentvariables.TheindependentvariablesarethosefactorsthatcangenerallybecontrolledattheHMAplant.Independentvariableswereasphaltcontentandpercentagespassingthe#200,#100,#50,#30,#16,#8,#4,9.5mm,12.5mm,19mm,25mm,and37.5mmsieves.

    2

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    7/18

    Kandhal,Foo,&D'Angelo

    Theobjectiveofthistaskistoidentifywhichindependentvariablesarethebestpredictorofthe

    voidpropertiessuchasVMAandVTM.Theidentifiedbestpredictorscanthenbeusedtoreconciledifferencesbetweenmixdesignandproductioninthenexttask.Twotechniqueswereusedtoidentifythebestpredictivevariables:linearregressionandstepwisemulti-variable

    regression.Singleandmultivariablespredictivemodelswerethenconstructedwiththebestpredictivevariables.

    LinearRegression--Thecoefficientofcorrelation(Rvalue)generatedbythelinearregression

    givesameasureofhowwelltheindependentvariableiscorrelatedtothedependentvariable.Inthelinearregressionanalysis,allindependentvariableswereindividuallycorrelatedtothedependentvariableforeachproject.TheRvaluesoftheindependentvariablesforeachprojectaregivenelsewhere[3].AbroadrangeofRvaluesfrom0.00to0.96weregenerated.TheseRvaluescanbeuseddirectlytoranktheindependentvariablesineachprojectbutitismoredesirabletoranktheindependentvariablebasedonallprojects.Therefore,theRvaluesforeachindependentvariablebasedontheaveragedRvaluesisgiveninTable1.TheaveragedRvaluedoesnothaveanyspecificstatisticalmeaning.Itisusedonlyasatooltoranktheindependent

    variablesforallprojects.ThefollowingobservationsaremadebasedonTable1:1.WithrespecttoVMA,thetopfiveindependentvariablesarethepercentagesofaggregatepassing#8,#16,#30and#50sieves,andasphaltcontent.ThisindicatesthattherelativeproportionsofcoarseandfineaggregatesareveryimportantandcanbeusedtoadjusttheVMA.

    2.WithrespecttoVTM,thetoprankingvariableistheasphaltcontent,followedbythepercentagesofaggregatepassing#30,#50,#100and#200sieves.ThisindicatesthattheVTMisalsoafunctionofVMAwhichiscontrolledbytherelativeproportionsofcoarseandfineaggregateasmentionedabove.

    Table1.AveragedRValuesandCombinedRankingsofIndependentVariables

    AllProjects HighVariabilityProjects

    Variable

    R

    VMARanking

    R

    VTMRanking

    R

    VMARanking

    R

    VTMRanking

    (Average) (Average) (Average) (Average)

    AsphaltContent 0.333 4 0.401 1 0.263 8 0.479 1

    #200 0.301 9 0.300 4 0.407 4 0.361

    3#100 0.294 10 0.301 3 0.406 5

    0.354 5#50 0.331 5 0.283 5 0.472 2

    0.356 4#30 0.372 1 0.302 2 0.483 1

    0.384 2#16 0.347 2 0.268 6 0.449 3

    0.328 6#8 0.345 3 0.247 7 0.381 6

    0.298 7#4 0.316 6 0.199 10 0.300 7

    0.229 9

    9.5mm 0.314 7 0.214 8 0.261 9 0.238 8

    12.5 0.304 8 0.214 9 0.252 10 0.192 10

    mm

    3

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    8/18

    Kandhal,Foo,&D'Angelo

    Thecombinedrankingsfromallprojectswerenotcompletelyadequateinidentifyingimportant

    variablesaffectingVMAbutweresomewhatabletoidentifytheimportantvariablesaffectingVTMduringproduction.TheprecedinganalysiswasimpededbyincludinghighqualitycontrolprojectswhichhadlowvariationinVMAandVTMresultinginclustereddatapoints.The

    FHWAexaminedthetestdatafrom17pavingprojectsinapreviousstudyandfoundthatthepooledmeanVMAstandarddeviationwas0.47andthemeanVTMstandarddeviationwas0.66[1].Therefore,projectswithVMAstandarddeviationlessthan0.47andVTMstandarddeviationlessthan0.66werethenexcludedtoincreasethesensitivityoftheprecedinganalysis.TwelveprojectswereexcludedfromtheVMAanalysisandtenprojectswereexcludedfromtheVTManalysis.TablelalsotabulatestheaveragedRvalueandcombinedrankingsfortheselectedhighvariabilityprojects.Thefollowingobservationsaremade.

    1.WithrespecttoVMA,thetopsixrankingvariablesconsistofmixgradationpassing#8,#16,#30,#50,#100,and#200sieves.Thismeansthattherelativeproportionofthefineaggregateinthemixandtheamountofmaterialpassing#200(P200)sieveareveryimportantinaffectingtheVMA.However,thepercentageofmaterialpassing#30and#50sieveshavethehighestrankings.Thesepercentagesare

    generallyinfluencedbythepresenceandamountofnaturalsandinthemix.2.WithrespecttoVTM,thetoprankingvariableistheasphaltcontentfollowedbytheaggregategradationpassing#16andfinersieves.Again,itindicatesthedependenceofVTMonVMAwhichwasalsoaffectedbythesesizes.TheP200materialrankedthirdand,therefore,isconsideredimportant.

    StepwiseMulti-VariableRegression--AForwardSelectionProcedureisavailableintheSAS

    program[4]todeterminewhichindependentvariablesarecloselyrelatedtoVMAandVTM.2

    Theselectionprocedurebeginsbyfindingthevariablethatproducestheoptimum(highestR)

    one-variablemodel.Inthesecondstep,theprocedurefindsthevariablethat,whenaddedtothe2

    alreadychosenvariable,resultsinthelargestreductionintheresidualsumofsquares(highestRvalue).Thethirdstepfindsthevariablethat,whenaddedtothemodelprovidesareductionin

    sumofsquaresconsideredstatisticallysignificantataspecifiedlevel.TheoutputoftheForwardSelectionProcedureforeachproject(includingpartialR2valuesforeachindependentvariableisgivenelsewhere[3].TheR2valueforeachprojectasgeneratedbytheForwardSelectionProcedurerangedfrom0.24to0.99.

    Therearetwopossiblemethodstoranktheindependentvariablesineachprojectfromthe

    ForwardSelectionProcedureoutput.Thefirstrankingmethod(Method1)isaccordingtothe

    ordertheywereselectedbytheForwardSelectionProcedure.2Thesecondmethod(Method2)is

    toranktheindependentvariablesaccordingtotheirpartialRvalues.AsmentionedinLinearRegression,itisdesirabletoranktheindependentvariablesbasedonallprojectsratherthaneachindividualproject.ToobtainacombinedrankingforallprojectsusingMethod1,thefirstvariableselectedisassigned1point,thesecondvariableselectedisassigned2pointsandsoon.

    Acombinedrankingisthenpossiblebyaveragingtheassignedpointsforallprojects.Forthesecondrankingmethod(Method2),thepartialR2valuesofeachindependentvariablewereaveragedoverallprojects.AcombinedrankingisthenpossiblebasedontheaveragedpartialR2value.TheaveragedpartialR2valuesdonothaveanyspecificstatisticalmeaningexcepttobeusedasatooltoranktheindependentvariables.

    Table2summarizesthecombinedrankingsforselectedprojectswithrelatively-lowerquality

    control(standarddeviationmorethan0.47forVMAandmorethan0.66forVTM)usingMethods1and2.Thecombinedrankingobtainedbybothmethodsaresimilartothoseobtainedbycorrelationanalysis(Table1).However,thecombinedrankingbyMethod2showsbetterresemblancewiththecombinedrankingbycorrelationanalysisthanMethod1.Intuitively,Method2beingrationalseemstobeabetterapproachforquantitativeanalysisthanMethod1andthusobtainingacombinedrankingoftheindependentvariables.

    4

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    9/18

    Kandhal,Foo,&D'Angelo

    Table2.CombinedRankingsofIndependentVariablesUsingForwardSelection

    ProceduresMethods1and2(HighVariabilityProjects)

    Method1 Method2

    VMA VTM VMA VTMVariableAvg. Ranking Avg.Point Ranking Avg.)R2 Ranking Avg. RankingPoint )R2

    AsphaltContent 6.07 8 2.00 1 0.032 8 0.219 1

    #200 4.86 2 5.23 3 0.068 6 0.110

    2#100 3.92 1 5.07 2 0.149 1

    0.097 3#50 5.08 3 5.27 4 0.103 4

    0.076 5#30 5.93 7 5.87 6 0.084 5

    0.028 8#16 5.54 5 5.57 5 0.136 3

    0.060 6#8 5.62 6 6.85 10 0.143 2

    0.035 7#4 5.39 4 6.06 8 0.031 9

    0.020 9

    9.5mm 7.36 10 6.00 7 0.028 10 0.080 4

    12.5 6.25 9 6.14 9 0.038 7 0.018 10mm

    BestPredictiveVariables--Thebestpredictivevariablesselectedbylinearregressionanalysis(Rvalue),andForwardSelectionProcedureMethod1(PointValue),andForwardSelection

    ProcedureMethod2()R2

    value)areshowninTable3.ThestatisticalanalysesincludetheprojectswithhighstandarddeviationsforVMAandVTM,asmentionedearlier.Bothtechniques,linearregressionanalysisandForwardSelectionProcedure,usedtoselectthebestpredictivevariableshavetheirowninadequacy.Thelinearregressionevaluatesasinglevariable,whileForwardSelectionProcedurecanevaluateseveralvariables.However,eachanalysisdoesprovideausefulsuggestionastowhichvariablehasthebestpredictivepower.Thefollowingobservationsaremadebasedonthecombinedrankingdata(giveninTable3)fromtheserankinganalyses.

    1.ThebestpracticalpredictiveindependentvariablesforVMAarethe#8,#16,#30,#50,#100,and#200sieves.Inotherwords,therelativeproportionofthefineaggregateandtheamountofmaterialpassing#200sieveisimportant.

    2.ThebestpracticalpredictivevariablesforVTMareasphaltcontent(AC)and#200sieve,followedby#8,#16,#30,#50,and#100sieves.

    Thereseemstobereasonablerationaletosupporttheresultoftheanalysis.Itisgenerally

    acceptedthatVTMismostsignificantlyaffectedbyAC.VTMandVMAarealsosignificantlyaffectedbythepercentageofmaterialpassingthe#200(P200)sievewhichfillsthespacesbetweenaggregateparticles.Inaddition,itisgenerallybelievedthattheamountoffineaggregate(percentpassing#8)hasaneffectonVMAandVTM.Generally,anincreaseinpercentpassingthe#8sieve(fineaggregate)alsoincreasesthepercentpassingthesmallersievesizes(#16,#30,#50,#100,#200).Thehigherrankingsreceivedbythe#30and#50sievesseemtoreflecttheeffectofnaturalsandinHMAmixes.

    5

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    10/18

    Kandhal,Foo,&D'Angelo

    Table3.ResultsofRankingAnalysisbyThreeMethods

    CombinedRankingUsing

    CombinedRankingVMA

    1

    2

    3

    4

    5

    6

    VTM

    12

    3

    4

    5

    6

    7

    Rvalue

    (Average)

    #30

    #50

    #16

    #200

    #100

    #8

    AC#30

    #200

    #50

    #100

    #16

    #8

    PointValue

    (Average)

    #100

    #200

    #50

    #4

    #16

    #8

    AC#100

    #200

    #50

    #16

    #30

    3/8

    )R2

    (Average)

    #100

    #8

    #16

    #50

    #30

    #200

    AC#200

    #100

    3/8

    #50

    #16

    #8

    TASK3:METHODFORRECONCILINGDIFFERENCESBETWEENMIXDESIGN

    ANDPRODUCTIONTheobjectiveofthisprojectwastodevelopguidelinestoreconciledifferencesbetweenmixdesignandproduction.Theguidelinesmustbepracticalandapplicablewithintheconfineofpracticalasphaltplantoperation.ItwasdeterminedinTask2thatthebestpredictivevariablesforVMAarethe#200,#100,#50,#30,#16,and#8sieveandthebestpredictivevariablesforVTMareasphaltcontent,#200,#100,#50,#30,#16,and#8,sieve.Asphaltcontentand,insomecases,#200sievecanbeindependentlycontrolledandadjustedtoreconciledifferencesbetweenproducedandlaboratorydesignedHMAmixes.However,theothersievesizes(#100,#50,#30,#16,#8)cannotbecontrolledindependentlybecausetheyarerelatedtotheproportionofcoarseorfineaggregate.Increasingthefineaggregateportionwillincreasetheamountofmaterialpassingallthesievesizes(#200,#100,#50,#30,#16,#8),andthemagnitudeofincreasewill

    dependlargelyonthegradationofthefineaggregate.Sincethefinersievesizesareinter-related,itisrecommendedthatthefinersievesizesshouldbecombinedasonepredictivevariable(thatistheamountoffineaggregate)ratherthansix(6)individualpredictivevariables.

    Forpracticalreasons,attemptstoreconciledifferencesbetweenmixdesign'sVMAand

    productions'sVMAshouldbeachievedbyfirstadjustingtheamountofP200materialandthen,ifnecessary,byadjustingtheothersievesizesbychangingtheamountoffineaggregate.Attemptstoreconciledifferencesbetweenmixdesign'sVTMandproduction'sVTMshouldbeachievedbyfirstadjustingtheamountofP200materialifitdeviatessignificantlyfromtheJMF.TheP200materialcanbeadjustedbycontrollingtheamountofdustreturnedfromdustcollectionsystem.Thesecondstep,ifnecessary,istoadjusttheasphaltcontent.Finally,itmaybenecessary,toadjusttheamountofmaterialpassingothersievesizes(theamountoffine

    6

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    11/18

    Kandhal,Foo,&D'Angelo

    aggregate)whichpracticallyamountstoredesigningthemix.

    ThefollowingareregressionmodelswhichrelateVMAandVTMtothebestpredictive

    variables.Thesemodelsestimatethemagnitudeofadjustmentneededtoreconcilethe

    differencesbetweenmixdesignandproduction.VMARegressionModels--Theregressionmodelrecommendedtopredicttheeffectofthematerialpassing#200sieve(P200)onVMAisgivenas:

    where,)VMA=differencefromprojectVMA)P200=differencefromprojectP200

    RegressionanalysiswasperformedonprojectswithhighVMAvariation(*morethan0.47)to

    increasethesensitivityoftheregressionmodel.TheseprojectswerethendividedintothreegroupsbasedontheirVMAlevels(>16%,14-16%,and0.47ProjectswithF

    VMA>0.47andVMA>16%

    ProjectswithFVMA>0.47and14%16%

    ProjectswithFVMA>0.47andVMA

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    12/18

    Kandhal,Foo,&D'Angelo

    increasestheAreaEnclosedand,therefore,theVMA.

    Theaverageslope$1ofEquation1(Table4)isrelativelyflat,about0.3.Asmentionedearlier,adjustingtheP200materialbyonepercentcausedanaverageof0.3percentchangeinVMA.

    Therefore,theP200adjustmentcanbeusediftheVMAcorrectiontobemadeisminor.HMAmixesthatneedasignificantamountofVMAcorrectionhavetobeadjustedbyvaryingthecoarse-fineaggregateproportions(AreaEnclosed).Consequently,itisrecommendedthatadjustingtheP200ismoreappropriateforfinetuningtheVMAwhilechangingtheaggregategradationbyadjustingthecoarse-fineaggregateproportion(AreaEnclosed)ismoresuitableforlargerchangesinVMA.Sincetherequiredadjustmentsaremixspecificnoquantifiablechangeincoarse-fineaggregateproportioncanberecommendedotherthandirectionalchanges(increaseordecreasecoarse-fineaggregateproportion).

    VTMRegressionModels--ThereisastrongrelationshipbetweenVTMandVMA.AllVTM

    regressionmodels,therefore,willhaveVMAtermsaspredictivevariables.Also,allpredictivevariablesforVMAarealsoapplicabletoVTM.ThebestregressionmodelwhichrelatesVTMto

    P200isgivenas:DifferentiatingEquation2withrespecttoP200resultsin:Equation3showsthattheeffectofP200onVTMisdependentonVMA.Table5whichisbasedonEquation3showsVTMisexpectedtodecreasewhentheP200isincreased(negativevalueofEquation3).TheVTMofmixeswithhighVMAisexpectedtodecreasemorethanmixeswithlowVMA.Table5deceptivelyshowsthattheVTMincreaseswiththeincreaseinP200formixeswith12percentVMA.ThishasbeencausedbyinsufficientdatapointsinthelowVMAregiontogenerateareliablemodel.

    Table5.ChangesinVTMCausedbyChangesinP200atDifferentVMALevel

    VMA 12% 13% 14% 15% 16% 17% 18%

    )VTM/

    )P200 0.054 -0.022 -0.098 -0.174 -0.249 -0.325 -0.401

    Thebestregressionmodeltorelateasphaltcontent(AC)toVTMisgivenas:

    8

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    13/18

    Kandhal,Foo,&D'Angelo

    DifferentiatingEquation4withrespecttoasphaltcontentresultsin:

    Equation5showsthattheeffectofasphaltcontentonVTMisalsodependentonVMA.Table6whichisbasedonEquation5showsthatanincreaseinasphaltcontentdecreasestheVTM(negativevalueofEquation5).ThisdecreaseismoresevereformixeswithlowerVMA.

    NostatisticallysatisfyingmodeltopredictVTMusingthevariableAreaEnclosedcouldbe

    constructed.However,increasingtheAreaEnclosed(deviatingfromthemaximumdensitylineorMDL)willincreasetheVTMofmixeswithnonaturalsandbutdecreasetheVTMofmixeswithnaturalsand.NaturalsandsusuallymaketheHMAmixesoversanded(toomuchdeviationfromtheMDLandtherefore,increasedAreaEnclosed)andtendtohaverelativelylowVMAbecausenaturalsandparticlespackdensely.ThischangeinVTM,therefore,reflectsthechangeinVMA.

    TheslopevaluesinTable5arecomparativelysmallerthantheslopevaluesinTable6.AdjustingtheP200,especiallyifitisexcessivelyhigherthantheJMF,ismoreappropriateforfinetuningtheVTM.AdjustingtheasphaltcontentismoresuitableforlargerchangesinVTM.ItisexpectedthatEquation3(orTable5)willbeusedfirst,iftheP200deviatesfromtheJMF,inanyattemptstoreconciledifferencesinmixdesign'sVTMandproduction'sVTM.IftheproductionP200isreasonablyclosetotheJMFP200,theasphaltcontentshouldbeadjusted.

    Table6.ChangesinVTMCausedbyChangesinACatDifferentVMALevel

    VMA 12% 13% 14% 15% 16% 17% 18%

    )VTM/)AC -1.351 -1.209 -1.067 -0.925 -0.783 -0.641 -0.499

    StepsRecommendedtoReconcileDifferencesbetweenMixDesignandProduction--ThevaluestabulatedinTables4,5and6arederivedfromdifferentmixesandthusrepresentaveragevaluesforthesemixes.Sinceeachmixisunique,thevaluespresentedheremaynotaccuratelypredictitsbehavior.Figure1isaflowchartwhichshowstherecommendedstepstoreconciledifferencesbetweenmixdesign'sVMAandmixproduction'sVMA,afterithasbeenverifiedthatthecomposition(asphaltcontentandgradation)oftheproducedmixisreasonablyclosetothatofthedesignedmix(JMF).

    IfthecompositionoftheproducedmixmeetstheJMFandtheVMAoftheproducedmixhasa

    minordeviation(lessthan0.3%)fromtheJMF,ithasbeensuggestedtoadjusttheamountofP200materialinthemix.AonepercentdecreaseintheP200materialtocauseanaverageincreaseof0.3percentintheVMA,canbeusedasanapproximateguidetodeterminethequantitativeadjustmentrequiredfortheP200materialtoeffectthedesiredchangeinVMAvalue.Asanalternate,anextendedlaboratorymixdesigncanincludeusingtwoadditionalP200contents(JMF+2%)intheHMAmixandplottingthecurveofP200contentversusVMA.ThepercentdecreaseinVMAfromthecorrespondingincreaseintheP200content(whichismixspecific)canbeobtainedfromthiscurveandislikelytobemoreaccuratethantheapproximateguidementionedabove.

    IftheVMAoftheproducedmixhasamajordeviation(morethan0.3%)fromtheJMF,theflow

    chartrecommendsdifferentapproachesdependingonwhethertheHMAmixcontainsnaturalsandornot.IftheHMAmixcontainsnaturalsand,theamountofnaturalsandwillneedtobedecreased

    9

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    14/18

    Kandhal,Foo,&D'Angelo

    Figure1.FlowChartforReconcilingVMA10

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    15/18

    Kandhal,Foo,&D'Angelo

    toincreaseVMA.IftheHMAmixdoesnotcontainnaturalsand,thepercentageofmaterialpassing

    the#8sieve(thatis,therelativeproportionsofcoarseandfineaggregates)willneedtobeadjustedtomoveawayfromthemaximumdensityline(MDL).Sincethisadjustmentismixspecific,noquantitativerecommendationscanbemade.However,anextendedlaboratorymixdesignwhich

    includestwoadditionalpercentagesofthematerialpassing#8sieve(JMF+5%)islikelytobeveryhelpful.Thedesigncurveobtainedbyplottingthesepercentagesofpassing#8sieveversusVMAcanindicatethequantitativeadjustmentneededtothe#8sievetoobtaindesiredVMA.

    IftheproductionVMAisnotreconciledaftertheprecedingefforts,theentiremixwillneedtobe

    redesignedbychangingthemixcomponentsand/ortheirproportions.

    AftertheproductionVMAisreconciled,thenextstepistochecktheVTM.Figure2isaflowchart

    toreconciledifferencesbetweenmixdesign'sVTMandproduction'sVTM.Again,itisassumedthattheproducedmixhascomposition(asphaltcontentandgradation)closetotheJMFcomposition.IftheVTMhasaminordeviation(lessthan0.5%)fromtheJMF,itisrecommendedtoadjusttheP200material.TheP200materialwillneedtobedeycreasedtoincreasetheVTM.

    TableA(insideFigure2)canbeusedasanapproximateguidetodeterminethequantitativeadjustmentrequiredtotheP200materialtoobtainthedesiredVTM.Asanalternate,theextendedlaboratorymixdesign(mentionedearlier)curveofpercentP200versusVTMcanbeusedforquantitativeadjustment.IftheVTMhasamajordeviationfromtheJMF(>0.5%),itisrecommendedtoadjusttheasphaltcontent.TheasphaltcontentwillneedtobedecreasedtoincreasetheVTM.TableA(insideFigure2)canbeusedasanapproximateguidetodeterminethequantitativeadjustmentrequiredfortheasphaltcontenttoeffectthedesiredchangeinVTM.AbetteralternativeistousetheasphaltcontentversusVTMcurvedevelopedduringtheroutinelaboratorymixdesign.Theslopeofthiscurvecangiveanindicationofthequantitativeadjustmentneededtoasphaltcontent.

    TASK4:FIELDVERIFICATIONOFPROPOSEDMETHODOFRECONCILIATION

    Itwasdeemednecessarytoverifytheproposedmethodofreconcilinglaboratorydesignedmixwiththeplantproducedmixinthefield.ApavingprojectwhichwasduetobevisitedbytheFHWAtrailer,wasselectedduringthe1994constructionseason.TheHMAmixproducedbytheasphaltplantwasabasemixwithamaximumnominalsizeof25mm(1inch).ThedetailsofthispavingprojectsuchasJMFandproductiondata(includingvolumetrics)aregivenelsewhere[3].

    TheJMFasphaltcontentof5.6%gaveaVTMof3.0%(thestateagencyacceptstheVTMaslow

    as3.0%)andaVMAof16.2%.TheamountofmaterialpassingNo.200(P200)sieveintheJMFwas5.1%.However,whentheHMAproductionbegan,aVTMcloseto1.7%andaVMAcloseto14.5%wasobtainedatanasphaltcontentof5.6%andaP200contentof5.0%.TheproducedgradationwasreasonablyclosetotheJMFgradation.Therefore,theHMAproducerreducedtheJMFasphaltcontentfrom5.6to5.2%.ThirteensublotsofHMAwereproducedwiththereducedasphaltcontentof5.2%.AnaverageproductionVTMof2.67%andVMAof14.5%wasobtained.

    ItwasevidentthatthemixcompositionneededtobeadjustedfurthertoincreasetheVTMto3.0%

    orhigher.Theasphaltcontentwasfurtherreducedfrom5.2to5.0%forthreeconsecutivesublots.However,therewasnoimprovementintheVTMvalueobtainedwithalimitednumberoftests.Thecontractorreducedtheasphaltcontentagainfrom5.0to4.7%forthelast35sublotsoftheproject.ThisfinaldecreaseintheasphaltcontentincreasedtheaverageVTMto2.91%(closertothetargetof3.0%).Therefore,thefollowingchangesoccurredduringtheentirepavingperiod:

    Changeinasphaltcontentfrom5.6%to4.7%=0.9%

    ResultingchangeinVTMfroml.7%to2.9%=1.2%

    11

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    16/18

    Kandhal,Foo,&D'Angelo

    Figure2.FlowChartforReconcilingVTM

    12

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    17/18

    Kandhal,Foo,&D'Angelo

    Thismeansthat0.9%reductioninasphaltcontentincreasedtheVTMby1.2%.Thisamountsto1.2

    0.9or1.3%changeinVTMby1.0%changeinasphaltcontent.Thevalueof1.3%comparesreasonablywelltotheaveragevalueofabout1%correspondingtoaVMAvalueof14.5%inTable6,basedonallFHWAprojects.Insummary,thispavingprojecthadtheproblemoflowerVTMin

    theproducedmixcomparedtothelaboratorydesignedmix.Thiswasdespitethefactthattheproducedmixwasreasonablycloseinmixcompositiontothelaboratorydesignedmix.Thisproblemwasresolvedbyloweringtheasphaltcontent.Theasphaltcontentcouldhavebeenreduceddrasticallyinonestepiftheproposedmethodofreconciliationwasused,butthecontractorchosetoreduceitinthreestepsovertheperiodofpaving.

    CONCLUSIONSANDRECOMMENDATION

    Thefollowingconclusionscanbedrawnbasedonthestatisticalanalysisoffielddatafrom24FHWAdemonstrationprojects.

    1.Significantdifferencesexistedbetweenthevolumetricpropertiesofthelaboratorydesignedandplantproducedhotmixasphalt.

    2.VMAisaffectedmostbytheamountofP200materialandtherelativeproportionsofcoarseandfineaggregates.3.VMAcanbeincreasedbyreducingtheamountofP200materialornaturalsandinthe

    HMAmixes.VMAcanalsobeincreasedbymovingtheaggregategradationawayfromthemaximumdensityline(MDL)especiallyforHMAmixeswithnonaturalsand.

    4.VTMisaffectedmostbyasphaltcontent,P200materialandtherelativeproportionsofcoarseandfineaggregates.

    5.VTMcanbeincreasedbyreducingasphaltcontentorP200materialorboth.

    Thefollowingrecommendationsaremadetoreconciledifferencesbetweenthevolumetricproperties

    ofthelaboratorydesignedandplantproducedhotmixasphalt.1.UsetheflowchartsinFigures1and2asgeneralguidelinesforreconcilingtheVMAand

    VTMdifferencesbetweenthelaboratorydesignedandplantproducedHMAmixes.

    2.Performanextendedmixdesignwhichwillbeusefulinprovidingadditionalquantitativeinformationforreconcilingthedifferencesinvoidpropertiesthatmayariseduringproduction.ThisinformationbeingmixspecificislikelytobemorereliableformakingadjustmenttotheHMAmix.Therecommendedextendedmixdesignconsistsof:a.ConventionalmixdesignwithaspecificgradationusedinJMF.b.TwoadditionallevelsofthematerialpassingNo.8sieve(JMF5%).c.TwoadditionallevelsofP200material(JMF2%).d.Threelevelsofasphaltcontent(JMF0.5%).Theextendedmixdesignrequiresatotalof27combinations(3levelsofNo.8material3levelsofP2003asphaltcontents)ofwhich9willbetakencareofalreadybytheconventionalmixdesign.Ifthreebriquettesaremadeforeachcombination,anadditional72briquetteswouldbeneededfortheextendedmixdesign(24combinations3replicates).

    3.Attempttoreconcilethedifferencesbetweenthevolumetricpropertiesoflaboratorydesignedandplantproducemixesduringfirstday'sproductionbytestingatleast4sublotsamplesandusingtheaveragetestvalues.

    4.Afterthedifferencesinthevolumetricpropertiesarereconciled,maintaincontrolchartsformixcomposition(asphaltcontentandgradation)andvolumetricproperties(VMAandVTM)duringtheentireproductionperiod.

    13

  • 8/9/2019 Field Management of Hot Mix Asphalt Volumetric Properties

    18/18

    Kandhal,Foo,&D'Angelo

    REFERENCES

    1.

    2.

    3.

    4.

    D'Angelo,J.A.andFerragut,Ted,"SummaryofSimulationStudiesfromDemonstrationProjectNo.74:FieldManagementofAsphaltMixes",AsphaltPavingTechnology,Vol.60

    1991.Decker,D.,"FieldManagementofHotMixAsphalt",AsphaltPavingTechnology,Vol.63,1994.Kandhal,P.S.,Foo,K.Y.andD'Angelo,J.A.,"FieldManagementofHotMixAsphaltMixes",FinalReport,NationalCenterforAsphaltTechnology,March1995.Freund,R.J.andLitell,R.C.,SASSystemforRegression,SecondEdition,SASInstituteInc.,SAS,CampusDrive,Cary,NC,1991.

    14