NCAT Report 15-05R REFINED LIMITING STRAIN...

NCATReport15-05RREFINED LIMITING STRAIN CRITERIA ANDAPPROXIMATE RANGES OF MAXIMUMTHICKNESSES FOR DESIGNING LONG-LIFEASPHALTPAVEMENTS(FirstRevision)

ByDr.NamTran,P.E.Dr.MaryM.RobbinsDr.DavidH.Timm,P.E.Dr.J.RichardWillisDr.CarolinaRodezno

November2016

Tran,Robbins,Timm,Willis&Rodezno

ii

RefinedLimitingStrainCriteriaandApproximateRangesofMaximumThicknessesforDesigningLong-LifeAsphaltPavements

Dr.NamTran,P.E.AssociateResearchProfessor

NationalCenterforAsphaltTechnology

Dr.MaryM.RobbinsAssistantResearchProfessor


Dr.DavidH.Timm,P.E.BrasfieldandGorrieProfessorofCivilEngineering

PrincipalInvestigator

Dr.J.RichardWillisAssociateResearchProfessor


Dr.CarolinaRodeznoAssistantResearchProfessor


SponsoredbyNationalAsphaltPavementAssociation

November2016


iii

ACKNOWLEDGEMENTS

The authors wish to thank the National Asphalt Pavement Association for sponsoring thisresearchaspartof theOptimizing FlexiblePavementDesignandMaterial Selection researchprojectandforprovidingtechnicalreviewofthisdocument.

DISCLAIMERThecontentsof this reportreflect theviewsof theauthorswhoareresponsible for the factsandaccuracyofthedatapresentedherein.Thecontentsdonotnecessarilyreflecttheofficialviews or policies of the National Center for Asphalt Technology or Auburn University. Thisreportdoesnotconstituteastandard,specification,orregulation.Commentscontainedinthispaper related to specific testing equipment and materials should not be considered anendorsement of any commercial product or service; no such endorsement is intended orimplied.

NOTESONFIRSTREVISIONSNCATReport15-05wasrevisedinNovember2016toaddresserrorsfoundinTables13,14,and15.


iv

TABLEOFCONTENTS

1 Introduction...........................................................................................................................5

2 ReviewofDesignThresholdsandThicknessRequirements...................................................6

2.1 Bottom-UpFatigueCracking..........................................................................................6

2.2 StructuralRutting...........................................................................................................9

2.3 PavementThicknesses.................................................................................................10

3 EvaluationandRefinementofDesignThresholds................................................................11

3.1 FatigueEnduranceLimitasDesignThreshold..............................................................11

3.2 CumulativeStrainDistributionasDesignThreshold....................................................13

3.3 RefiningDesignThresholdsforPerpetualPavementDesign.......................................15

3.3.1 PavementSectionsandFieldPerformance..........................................................16

3.3.2 FieldPerformance................................................................................................19

3.3.3 AnalysisMethodology..........................................................................................20

3.3.4 RefinedLimitingStrainCriteriaforUseinPerpetualPavementDesign...............25

3.3.5 ValidatingRefinedDesignThresholds..................................................................27

4 ApproximateRangesofMaximumDesignThicknesses.......................................................29

5 Conclusions..........................................................................................................................34

6 Recommendations...............................................................................................................36

7 References............................................................................................................................37


5

1 INTRODUCTION

Many transportation agencies are currently conducting the design and analysis of asphaltpavementsbasedonthe1993(orearlier)versionofAmericanAssociationofStateHighwayandTransportationOfficials(AASHTO)GuideforDesignofPavementStructures(0).Theseempiricaldesignproceduresweredevelopedbasedon thedatacollectedduring theAASHORoadTestconductedfrom1958through1961(2).

Due to limited testing conditions included in the AASHO Road Test experiment andsignificant changes in traffic loadsandmaterialsover the years, pavementdesigns todayareoftenbasedonextrapolationfarbeyondtheexperimentalconditions.Oneconsequenceofthisextrapolation is ever-increasing thickness with traffic volume, resulting in overly designedasphaltpavementsforhighvolumeroadways(3).Thisraisestheconcernovertheaccuracyandeffectivenessoftheseproceduresfordesigningheavilytraffickedpavements.

Toaddressthelimitationintheempiricaldesignprocedures,theAsphaltPavementAlliance(APA)introducedtheconceptofPerpetualPavementsin2000(3).AsillustratedinFigure1,aperpetual pavement is designed to have appropriate layer thicknesses and materials foraddressing specific pavement distresses, especially those causing structural damage thatinitiatesatthebottomofthepavement.Toavoidthesestructuraldistresses,includingbottom-upfatiguecrackingandsubgraderutting,thepavement’sresponses,suchasstresses,strains,anddisplacements,mustbelowerthanthresholdsatwhichstructuraldistressesbegintooccur.Thus, thedesigncanbeoptimized to sustain theheaviest loadswithoutadditional structure,providing an indefinite structural life without being overly conservative (3). An asphaltpavementthatisbuiltproperlyanddesignedaccordingtothisconceptshouldlastlongerthan50 years without a major rehabilitation or reconstruction and would just need periodicresurfacingtoremedysurfacedistresses(4).

Figure1PerpetualPavementDesignConcept(3).


6

Currently,mostapproachestoperpetualpavementdesignfocusonpreventingbottom-upfatiguecrackingandsubgraderuttingfromoccurringatthebottomofthepavementstructure(3). Onemethod of reducing these critical structural distresses is to increase the pavementthicknesses to ensure that critical pavement responses are well below design thresholds. Indoingso,however,itisimportanttorecognizethatmaximumthicknessesmayexistsuchthatadditional material is not necessary in prolonging the structural life of a pavement. Forexample, in their investigationof long-lifepavements in theUnitedKingdom (UK),NunnandFerne(5)reportedthataconservativeasphaltlayerthicknessof15.4incheswassufficientforlong-lifeasphaltpavements.Thisthicknessincluded10.6inchesofasphalttopreventbottom-upfatiguecracking,4 inchesofasphalt tomitigatetop-downcracking(top-downcrackswereseentopropagateupto4inchesfromthesurfaceatthetimeofresurfacing)andanadditional0.8inchestoaccountforanincreaseinthelegalloadlimitintheUK.Asaresult,noadditionalthicknessbeyond15.4incheswasnecessarytoensurelonglife.

The objective of this study was to determine critical pavement design thresholds andapproximaterangesofmaximumthicknessesforflexiblepavementsinanefforttoimprovethecost effectiveness of long-life asphalt pavements. This studywas divided into two tasks. Thefirst task was to review literature pertaining to design thresholds and maximum thicknessrequirementsforperpetualpavements.ThesecondtaskwastoestablishdesignthresholdsandapproximaterangesofmaximumpavementthicknessesusingtheinformationreviewedinTask1 and through analyzing information from the fully instrumented pavement sections at theNCAT Pavement Test Track. This report summarizes the key findings of the two tasks andprovides recommendations for implementing design thresholds and approximate ranges ofmaximumthicknessforconsiderationinfuturepavementdesign.

2 REVIEWOFDESIGNTHRESHOLDSANDTHICKNESSREQUIREMENTS

A perpetual pavement is designed to resist structural distresses that initiate deep in thepavement structure and eventually require full-depth rehabilitation. The structural distressesincluded inmostperpetualpavementdesignapproachesarebottom-up fatigue crackingandsubgraderutting(3).Toavoidthesedistresses,anappropriateasphaltpavementstructurecanbedesigned so that thehorizontal tensile strains at thebottomof theasphalt layer and theverticalcompressivestrainsand/orstressesat thetopof thesubgradeare lowerthandesignthresholds below which structural damage does not initiate. Also, any additional pavementthicknessthanwhatisrequiredtokeepthecriticalstrains/stressesbelowthedesignthresholdswould not provide additional pavement service life. Different perpetual pavement designthresholdshavebeenproposedoverthepast20years.Asummaryofthesedesignthresholdsandlayerthicknessrequirementsfordesigningperpetualpavementsfollows.

2.1 Bottom-UpFatigueCracking

Bottom-up fatigue cracking, also known as alligator cracking, severely affects a pavement’sstructural capacity. Figures 2 and 3 illustrate examples of fatigue cracking from the NCATPavementTestTrack.Thesecrackstypicallyforminthewheelpathsandinitiateatthebottomof the asphalt concrete (AC) layer and propagate to the pavement surface due to repeatedtensilestrainevents.Oncecracksappearatthesurfaceandwaterentersthepavementthrough


7

thecracks,furtherdeteriorationprogressesquicklyandreducesthestrengthoftheunderlyingmaterials. The strength of the AC layer is also compromised by the presence of the cracksthemselves.Thisformofdistressisgenerallyremediedbyfull-depthrehabilitation.

Figure2ExamplesofBottom-UpFatigueCrackingObservedatPavementSurface.

Figure3ExampleofBottom-UpFatigueCrackingObservedinCross-Section.

Inaperpetualpavementdesign,thestrainsatthebottomoftheasphaltstructurearekept

below a design strain threshold to prevent the initiation of bottom-up fatigue cracking. Thisdesignthresholdisoftenthefatigueendurancelimit(FEL)oftheasphaltmixtureusedintheACbase layer.ThompsonandCarpenter (6)described theFELas representing thebalancepointbetweendamageandhealingwhileBonaquist(7)describedtheFELas,“Alevelofstrainbelowwhichthereisnocumulativedamageoveranindefinitenumberofloadcycles.”Regardlessof


8

thedefinition,bottom-up fatiguecracking isnotexpected to initiate inapavementdesignedbelowtheFEL.

Fatigue endurance limits may be determined through either laboratory testing byconducting bending beam fatigue tests (BBFT), or through field evaluation of existingpavements.Oneof the challenges,much like conventionalBBFT testing todetermine fatiguecrackingtransferfunctions,isbridgingthegapbetweenlaboratorytestresultsandactualfieldperformance.

An endurance limit of 70 microstrain was first reported for asphalt pavements byMonismithandMcLean(8).ThompsonandCarpenterreportedin2006(6)that70microstrainshould be considered theminimum value, as no lab datawere found below this FEL, and apracticalrangeis70to100microstrain.In2009,ThompsonandCarpenter(9)furtherstated,“Avery conservative FEL is 70 microstrain. Laboratory studies have demonstrated that mostHMA’sdisplayFELswellinexcessof70microstrain.”Insupportofthisstatement,Prowelletal.(10)providedlaboratorytestingdatasupportingtheexistenceofahigherendurancelimitthatvariedfrom75to200microstrain.

LaboratorytestingconductedattheUniversityofIllinoisevaluating120differentmixturesfoundFELs tovary from90 to300microstrain (11). TheFELwas found todependonbindertype andmixture composition (11). Furthermore,whilemixture composition (i.e., volumetricparameters)was found to be important to the FEL, the gradation seemed to have relativelylittleeffectontheFEL(11).

Fatigue endurance limitswere also determined based on the analysis of long-life asphaltpavements.Nishizawaetal.(12)reportedanendurancelimitof200microstrainforin-servicepavements in Japan. For a long-life pavement in Kansas, strain levels at the bottom of theasphalt layerwere between 96 and 158microstrain calculated from backcalculated stiffnessdata (13).Yangetal. (14) reportedasuccessfulperpetualpavementdesign inChinausinganendurancelimitof125microstraininsteadofamoreconservativelimitof70microstrain.

VonQuintus (15) examined pavement sections in the Long-Term Pavement Performance(LTPP)databasetodeterminestrain levelsthatcorrelatedto lessthana2%chanceofhavingfatigue cracking.He found65microstrain to yield a 95% confidence that crackingwouldnotoccur.

Basedon full-scalepavement testing resultsat theNCATPavementTestTrack,WillisandTimm (16) showed that asphalt pavements could withstand tensile strains greater than 100microstrainatthebottomoftheasphaltlayer.Theyproposedaprofileoflimitingstrainsatthebottomoftheasphaltlayerthatwasfoundtodistinguishthefieldperformanceoftestsectionsbetter thanonedesignendurance limitused in thepast.The limitingstrainsweredividedbythelaboratoryendurancelimittodeterminethemaximumfatigueratiosasshowninTable1.Thismethodisdetailedlaterinthisreport.

It is important to understand that the fatigue ratios in Table 1 were based on field-measuredstrainlevelsattheNCATPavementTestTrack.Whileaccurate, it isnotpracticaltofrequently instrumentpavements fordesignpurposes, and fatigue ratiosbasedon simulatedstrain levels is desirable. Therefore, as presented later in this report, it was necessary toestablish critical fatigue ratio levels based on simulated strain levels through mechanisticsimulationsoftware.


9

Table1MaximumFatigueRatiosBasedonMeasuredStrain(16)

Percentile MaximumFatigueRatio99% 2.8395% 2.4590% 2.1885% 1.9880% 1.8575% 1.7470% 1.6365% 1.5360% 1.4455% 1.3550% 1.27

2.2 StructuralRutting

Structural rutting occurs in the aggregate base, subgrade layer, or both, under the imposedtraffic. Figure 4 illustrates rutting that is occurring in the base and subgrade layers wheredistortionoftheselayersmirrorsarutinthesurface.Tocontrolstructuralruttinginaperpetualpavementdesign,theverticalstrainorstressatthetopofthesubgradehasbeenusedasthelimitingdesignparameter.

Figure4ExampleofStructuralRutting(17).

Monismithetal.(18)proposedalimitingverticalstrainof200microstrain.Theysuggested

that computedvertical strainsat the topof the subgradeshouldbekeptbelowthisvalue to


10

prevent structural rutting. This approach was also recommended by Walubita et al. (19).Instead of limiting the vertical stain, Bejarano and Thompson (20) proposed controlling theverticalstressthroughtheratiooftheverticalstressatthetopofsubgradetotheunconfinedcompressivestrengthofthesoil,referredtoasthesubgradestressratio.Theyrecommendedusingasubgradestressratioof0.42fordesignpurposes.

2.3 PavementThicknesses

Foraperpetualpavementdesign,thelimitingtensilestrainatthebottomoftheasphaltlayerandthe limitingcompressivestrainorstressat the topof subgradecanbeachievedthroughchoosing appropriate thicknesses and materials for the pavement layers. Minimum andmaximumpavement thicknesses have been recommended for a variety of design conditionsandsummarized inthissection.Alsodiscussed inthissectionareanumberofnewperpetualpavementsdesignedinrecentyears.

BasedontheanalysisofthemostheavilytraveledpavementsintheUK,mostofwhichhadcarried over 100 million standard axles, Nunn (21) concluded that minimum and maximumthicknesses for long-life full-depthasphaltpavementswere7.9and15.4 inches, respectively.This range was determined for a variety of factors. For these pavements, they found littleevidenceofbottom-upcrackingbutsurface-downcrackingthattendedtostopatadepthof4inches. In addition, for a pavement thicker than 7.1 inches, rutting tended to occur in theasphaltlayer.

As part of the Strategic Highway Research Program 2 R23 project, Jackson et al. (22)developedthicknessguidelinesforlong-life(30to50years)asphaltpavementdesignintheU.S.Thedevelopmentwasconductedbasedonthelimitingstrainapproachfornumerousscenariosthat simulate field conditions found in five representative locations in the U.S. A minimumthicknessof5.5inchesandamaximumthicknessof14.0incheswererecommendedforlong-life pavements depending on design conditions including traffic loading and stiffness offoundationsupport.

There have been several new perpetual pavements built in recent years. The BradfordBypass in Pennsylvania was designed as a perpetual pavement using both the PerRoad andDAMAprograms (23)during thedesigndevelopmentphase.ConservativeFELswereused forfatigue (70 microstrain) and rutting (200 microstrain). The resulting perpetual pavementconsistedof13.5inchesofACover13inchesofaggregatebase(23).

In a perpetual pavement experiment in Ontario, four new pavements were constructed.Twowere designed according to conventional thickness design (AASHTO ’93), and the othertwoweredesignedtobeperpetualusingPerRoad(24).Table2summarizesthefourdesigns.

Table2OntarioPerpetualPavementExperiment(24)

Highway DesignProcedure Traffic,millionESALS(years) ACThickness,in.402 PerRoad 146(50) 13.4406 AASHTO‘93 42(50) 9.87 AASHTO‘93 28(30) 9.1

RedHillCreekExpressway PerRoad 90(50) 9.4


11

AperpetualpavementbuiltinOhiousedathresholdof70microstrainatthebottomoftheAC layer to control cracking (25). The design accounted for the legal load levels plus 20%overloadandarrivedat16.25inchesofasphaltovera6-inchgranularbaselayer.

Historically,perpetualpavementdesigns inTexasconsistedofabout14 inchesofACoverapproximately 6 inches of lime- or cement-treated base over the subgrade soil. Morespecifically,itwasrecommendedtohavethefollowing(26):

• 1-1.5inchesofpermeablefrictioncourse(optional)• 2-3inchesofSMA• 2-3inchesofcoarsegradedAC• ≥8inchesofrutresistantAC• 2-4inchesoffatigueresistantbaseAC• ≥6inchesoflime-orcement-stabilizedsoil• subgradeHowever,inthesamereport(26),aslightlymodifiedpavementstructurewasproposedfor

useinTexas,asfollows:• 3inchesofSMA• 3inchesof¾”Superpavemix• 8inchesofTypeBmix• 8inchesoflime-orcement-treatedsoil• subgradeThreesections,intwodifferentexperimentsattheNCATPavementTestTrack,werefound

tobeperpetual. The first experiment included two sections,N3 andN4,whichwerebuilt in2003withonly9inchesofACover6inchesofaggregatebaseoverthetrack’snativesubgrade.Though expected to fail after 10million ESALs, theywithstood 30million andwere deemedperpetual(27).Thesecondexperimentconsistedoftwosectionsbuiltin2006fortheOklahomaDepartment of Transportation. They were constructed on an imported and much poorersubgrade.Onesectionwasconstructedwith10inchesofACandwasexpectedtofail,whiletheothersectionwasconstructedwith14inchesofACasaperpetualpavement(27).The10inchsection failed and was rehabilitated multiple times while the 14 inch section has exhibitedperpetualbehavior.

Though every perpetual pavement design is unique and should consider site-specificclimate,traffic,soils,andmaterialavailability, itappearsthatareasonablerangeofperpetualpavementthicknessisbetween9and16inchesofACforhighvolumeroadways.

3 EVALUATIONANDREFINEMENTOFDESIGNTHRESHOLDS

Based on the literature review results, an analysis was conducted as part of this study toevaluateand refine the thresholds fordesigningperpetual asphaltpavements. The resultsofthisanalysisaresummarizedinthissection.

3.1 FatigueEnduranceLimitasDesignThreshold

Historically,aperpetualpavementhasbeendesignedtohavethetensilestrainatthebottomoftheAClayerbelowitsFELsothatthestructurewillhaveinfinitefatiguelife.Inaddition,theverticalstrainatthetopofthesubgradeischeckedtoensurethatitisbelowapre-determined


12

limit to prevent structural rutting. The fatigue endurance limit used in perpetual pavementdesign is typically determined from laboratory fatigue testing and has ranged inmagnitudesfromearlyestimatesof70microstrain (8) tomore recentestimatesofup to200microstrain(10,12).Avalueof200microstrainhasbeenproposedfortheverticalcompressivestrainlimit(18,19).

WhiletheperpetualpavementdesignconceptbasedonFELshasbeenusedsuccessfullytodesign long lifepavementsaspreviouslydiscussed,there issomedebateabouthowhightheFEL can be and still maintain a perpetual pavement, and there is also some concern as towhetheronelimitingstrainvaluecancontrolfatiguecracking.

Since the second research cycle of the NCAT Pavement Test Track started in 2003, fullyinstrumented pavement sections have been built and evaluated under live truck traffic.Pavementresponsescollectedfromthesetestsectionshavebeenusedinseveralstudies(16,28,29)toevaluatetheperpetualpavementdesignconceptbasedonFELandtodevelopanewapproach.TensilestrainsmeasuredatthebottomoftheAClayerinfullyinstrumentedsectionsinaheavilyloadedenvironment(10millionequivalentsingleaxleloads(ESALs)ineachresearchcycle) were compared with laboratory-determined fatigue endurance limits of the AC baselayers. The results of these studies indicated that the number of events inwhich the strainsmeasuredinthefieldfellbelowthesection’sFELvariedsignificantly.

Table3(28)showsthepercentofthefield-measuredstrainsthatwerebelowtheFELsforsixtestsectionsfromthe2003and2006researchcyclesoftheNCATPavementTestTrack.ForSectionsN8,N10,andS11thatexperiencedfatiguecracking,3%to50%ofthestrainsmeasuredin the field fell below the section’s FEL. For SectionsN3,N4, andN9 that remained in goodconditionwith no fatigue cracking, 33% to 88% of the field-measured strains fell below thesection’sFEL.

Table3ComparisonofFieldStrainsandLaboratoryFELs(16,28)

Section,Year

LabFEL1,microstrain

PercentofField-MeasuredStrainsBelowSection’sFEL1

FieldPerformance

N3,2003 151 33% NoFatigueCracking,PerpetualN4,2003 146 38% NoFatigueCracking,PerpetualN8,2006 203 50% FatigueCrackingN9,2006 203 88% NoFatigueCracking,PerpetualN10,2006 130 8% FatigueCrackingS11,2006 118 3% FatigueCracking

1FELwasdeterminedasthe95%one-sidedlowerpredictionlimitaccordingtotheNCHRP9-38procedure(10).

BasedontheFELconcept,forsectionsthatdidnotexperiencefatiguecracking,thepercentoffield-measuredstrainsbelowthesection’sFEL,asshowninTable3,shouldberelativelyhigh,indicating that themajority of the strains fell below the FEL and thus no damage occurred.However, thiswasnot the case. SectionsN3andN4didnotexperience fatigue cracking, yetonly33%to38%ofthemeasuredstrainsfellbelowtheFEL.Inthiscase,theconceptoffatigueendurance limit would lead the designer to believe these sections were significantly under-


13

designed. However, Section N8, a section that failed due to fatigue cracking, showedmorepromiseofbeingaperpetualpavementwith50%ofmeasured strains less than its FEL.Asaresult, the researchers were not able to develop correlations between laboratory FELs andpavementperformanceorfield-measuredstrains.Theseresultssuggestthattheapplicationofasinglefatigueendurancelimitmaynotbeaneffectivedesigncriterioninperpetualpavementdesign(16,28).

3.2 CumulativeStrainDistributionasDesignThreshold

While the research at theNCATPavement Test Track (16) did not find correlations betweenlaboratory FELs and field performance or field-measured strains, they found a noticeabledifference between cumulative distributions of field-measured strains for sections thatexperienced bottom-up fatigue cracking and those that did not, as shown in Figure 5. EachcumulativedistributionshowninFigure5wasdeterminedbasedonthepercentofmeasuredstrainslessthanorequaltoaspecificstrainlevel.Asaresult,afieldlimitingstraindistribution(blackdashedlineinFigure5)wasdeterminedbasedonfield-measuredstrainsforSectionsN3and N4 that did not experience fatigue cracking. It was recommended that the cumulativedistributionoftensilestrainsbefurtherrefinedforuseaslimitingcriteriaforfatiguecrackinginthedesignofperpetualpavementsratherthanusingtheFEL(16,28).

Figure5CumulativeDistributionsofMeasuredStrains,SectionsPlacedin2003and2006(16,

28).FromtheindividualcumulativedistributionoffieldstrainsforeachsectionshowninFigure

5,theresearchersdeterminedthefatigueratiosatincrementalpercentiles,asshowninFigure6,bydividingthecorrespondingcumulativestrainsbythelaboratoryFELoftheACbaselayer.Thelimitingfatigueratios(blackdashedlineinFigure6)werearesultofthedistinctdifference

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 200 400 600 800 1000 1200 1400

Percen

tile

MeasuredTensileStrain(microstrain)

N3(Perpetual)

N4(Perpetual)

N8(Cracked)

N9(Perpetual)

N10(Cracked)

S11(Cracked)

FieldFatigueLimit


14

between sections that experienced fatigue cracking and those that did not. These studiesshowed that the fatigue ratios of the pavement sections that did not experience fatiguecrackingfellbelowthelimitingfatigueratios(16,28).

Figure6FatigueRatiosbasedonStrainsMeasuredin2006CycleandFELafter(16,28).The limiting strain distribution (Figure 5) and maximum fatigue ratios (Figure 6) show

promise for perpetual pavement design; however, they were determined based on field-measured strains. Past studies at the NCAT Pavement Test Track have shown differencesbetweenfield-measuredstrainsandpredictedtensilestrainsatthebottomoftheAClayer(30,31).Figures7and8comparethecumulativedistributionsoffield-measuredstrainswithstrainspredictedbyPerRoadforthecrackedandperpetualsections,respectively.Thesefiguresshowthat cumulative distributions from measured strain values are much higher than thecorresponding cumulative distributions of tensile strain values predicted by PerRoad. Thesedifferencescouldinpartbeattributedtothedifferencesinthedefinitionofstraininthefieldandwhat isused inPerRoad.Field-measuredstrainwasbasedontheamplitudeofthestraintracesuchthatstrainwasdefinedas themagnitudefromthetroughtopeakstrain,whereasPerRoadconsidersonlythepeakstrain,orthedifferencebetweenthebaselineandthepeakinthe strain trace. Previous research at the test track has shown that strainmeasured by theamplitudecanbe20%to30%higherthanstraindefinedbythepeakvalue(32).Inthiscase,themeasuredstrainsareapproximately80%higherthanthepredictedstrainsatthe50thpercentileforSectionsN3andN4.

Thus,thelimitingstraindistributionandmaximumfatigueratiosdevelopedbasedonfield-measured strainsmaynotbe readily applicable topredicted strains resulting fromperpetualpavementdesign.Therefore,thereisaneedtorefinethelimitingstraindistributionandfatigue

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

0.00 2.00 4.00 6.00 8.00 10.00 12.00

Percen

tile

FatigueRatio(CumulativeStrain/FEL)

N3(Perpetual)

N4(Perpetual)

N8(Cracked)

N9(Perpetual)

N10(Cracked)

S11(Cracked)

MaxFatigueRatio


15

ratiosshowninFigures5and6toreflectpredictedstrainsdeterminedinperpetualpavementdesignandtovalidatetherefinedcriteriaforfutureimplementation.

Figure7MeasuredversusPredictedCumulativeStrainDistributionsfor(Fatigue)Cracked

Sections(29).

Figure8MeasuredversusPredictedCumulativeStrainDistributionsforPerpetualSections

(29).

3.3 RefiningDesignThresholdsforPerpetualPavementDesign

Thefocusofthesecondpartofthisstudywastwo-fold:

0

200

400

600

800

1000

1200

1400

0 200 400 600 800 1000 1200 1400

Pred

ictedTensileStrain(m

icrostrain)


N8- 06N10- 06S11- 06LOE

0

50

100

150

200

250

300

350

400

450

500

0 100 200 300 400 500

Pred

ictedTensileStrain(m

icrostrain)


N3- 06N4- 06N9- 06LOE


16

1. torefinethelimitingstraindistributionandmaximumfatigueratiosdevelopedbyWillisandTimm(16)toreflectthedifferencesbetweenmeasuredandpredictedstrainvaluesforfutureimplementationinperpetualpavementdesign,and

2. tovalidatetherefinedlimitingstraindistributionandmaximumfatigueratios.Forthisanalysis,twosectionsbuilt inthe2003researchcycleoftheNCATPavementTest

Trackandfoursectionsbuiltinthe2006researchcycleweresimulatedinPerRoadversion3.5topredicttensilestrainvaluesatthebottomoftheAClayer.Thesesectionswereusedtorefinethelimitingstraindistributionbasedonpredictedstrainvaluesforuseinperpetualpavementdesign.Additionally,sixsectionsfromthe2009researchcyclewereusedtovalidatetherefinedlimitingstraindistribution.Theresultsofthisanalysis(aspresentedbelow)wereadaptedfromapreviousreport(29).

3.3.1 PavementSectionsandFieldPerformance

TwelvesectionsfromthreeresearchcyclesattheNCATPavementTestTrackwereselectedforthisanalysis.Theselectedsectionswereplacedonthenorthandsouthtangentsofthe1.7-mileoval track,a full-scaleaccelerated loading facility, located inOpelika,Alabama.Eachresearchcycle of the NCAT Pavement Test Track operates on a three-year period, with two yearsdesignated for traffickingandoneyear splitbetweenconstructionand forensicevaluationattheconclusionofthetrafficperiod.Approximately10millionESALsareappliedoveratwo-yeartrafficperiodwithafleetoffivetripletrailertrucksoperatingat45mphfor16hoursaday,fivedaysaweek.Thetripletrailertrucksconsistofa12-kipsteeraxle,a40-kiptandemaxle,andfivetrailing20-kipsingleaxles.Weeklyperformanceevaluations includingcrackmapping,rut-depthmeasurement,andride-qualitymeasurementsareaugmentedbyfrequentfalling-weightdeflectometertestingandstrain-responsemeasurements.

Figure 9 shows cross sections and Table 4 lists quality control (QC) asphalt mixturepropertiesfor12pavementsectionsbuiltinthreeresearchcyclesatthetrackselectedforthisstudy. Two pavement sections, including N3 and N4, were built in 2003, and four sections,includingN8,N9,N10,andS11,werebuiltin2006.Theremainingsixsections,N10,N11,S8,S9,S10, and S11, were from the 2009 research cycle. A brief description of each test sectionselectedforthisanalysisfollows.

SectionsN3andN4wereplacedaspartofthe2003researchcycleandwere left inplacethroughtheendofthe2009researchcycle.ThesesectionsweredesignedwithnineinchesofACoversixinchesofgranitebasematerialandthetesttracksubgradematerial,classifiedasanAASHTO A-4(0) soil. These two sections were designed to replicate each other with theexception of the binder type. Section N3 used a performance grade (PG) 67-22 unmodifiedbinder throughout the AC layer, and section N4 used a PG 76-22 modified with styrene-butadiene-styrene(SBS)throughout.


17

Figure9CrossSectionsandMaterialsofTestSectionsUsedinThisAnalysis.

Inthe2006researchcycle,SectionN8wasdesignedwith10inchesofACincludingastone-

masticasphalt(SMA)surfacelift,conventionalAClayersusingPG76-28andPG64-22binders,and a rich-bottom layer designed at 2% air voidswith PG 64-22 binder. SectionN9was thecomplementtoN8,although itwasdesignedtobeperpetualat14 inchesofAC.SectionsN8andN9usedthesamemixturesthroughout.BothsectionsN8andN9usedthecompactedtesttracksoilasabasematerialandacompactedsoftsubgradematerialovertheexistingtracksoilsubgrade.SectionN9wasleftinplacethroughthe2009trackcycleandaspartofthe2012testcycle.SectionN10wasdesignedateightinchesofACconsistingofmixeswithPG70-22binderinthesurfaceandintermediatelayerandPG64-22binderinthebasecoursemix.Sixinchesofa type-5 basematerial fromMissouri was used as a granular base layer over the test tracksubgrade. Lastly, Section S11wasdesigned at seven inchesofAC, featuring a PG76-22 SBS-modifiedbinderinthetoptwoACliftsandPG64-22inthebottomtwolifts,ontopofsixinchesofgranitebasematerialandthetracksoilsubgrade.Detailedinformationaboutthesesectionscanbefoundinapreviousreport(33).

Testsectionsplacedaspartofthe2009researchcycleusedforthisstudyincludedsixoftheeight-section group experiment: N10, N11, S8, S9, S10, and S11. These sections were alldesignedat7 inchesofAContopof6 inchesofgranitebase,placedontopofthetesttracksubgrade material. These sections were selected due to the range in mix types used andalthoughmanywereunconventional, theyusedcommonlyavailable technology.Additionally,allofthesesectionsexperiencedbottom-upfatiguecrackingandwouldservewellforvalidatingtherefinedlimitingdistributionofpredictedstrains.

02468

1012141618202224

N3 N4 N8 N9 N10 S11 N10 N11 S8 S9 S10 S11

Approxim

ateTh

ickness,in.

PG67-22 PG76-22PG76-22(PFC)

PG76-28(SMA)PG76-28 PG64-22 PG64-22(2%AirVoids)

PG70-22 50%RAP

GraniteBase Type5Base TestTrackSoil SealeSubgrade

(2003)Alabama

(2006)Oklahoma


18

Table4QCPropertiesofAsphaltMixtures


19

Inthe2009researchcycle,S9servedasthecontrolsectionforthegroupexperimentandfeaturedconventionalACmixtures includingasurfacecourseutilizingPG76-22SBS-modifiedbinder, an intermediate layer also using a PG 76-22 SBS-modified binder, and a base courseusingconventionalPG67-22binder.SectionS8mirroredsectionS9withtheexceptionofthesurfacelayer,whichreplacedtheconventionalmixwithaporousfrictioncourse(PFC)designedwith15%reclaimedasphaltpavement(RAP)atthesamethicknessof1.25inches.SectionsN10andN11sharedthesamemixdesignsincluding50%RAPinallthreeAClayers.Thedifferencebetween these sectionswas thatN11used foaming technology toproduce it as awarm-mixasphalt(WMA)whileN10wasproducedattypicalproductiontemperaturesforhot-mixasphalt(HMA). Sections S10 and S11 shared mix designs with Section S9 but were produced usingwarm-mixtechnologies.SectionS10wasproducedwithfoamingtechnologiesandSectionS11incorporatedanadditivetoachieveproductionatwarm-mixtemperatures.

3.3.2 FieldPerformance

SectionsN3,N4, andN9exhibitedno signsofbottom-up fatigue cracking. Sections from the2006researchcyclethatexperiencedbottom-upfatiguecrackingwereSectionsN8,N10,andS11. SectionsN3 andN4 (built in 2003) remained in service during both the 2006 and 2009research cycles. These sections were in excellent condition with only minor longitudinalcrackingafterapproximately30millionESALs.Forensic investigationsattheconclusionofthe2009 cycle revealed that longitudinal cracking in both sections was top-down and that nobottom-up fatigue cracking was evident in section N4. Section N3 experienced a subsurfacecracklikelyduetotheadjacentembeddedinstrumentationandthereforewasstillconsideredperpetual in nature (34). Section N9 was built in 2006 and left in place through the 2009researchcycle.Duringthissecondcycle,thesectionexperiencedlongitudinalcrackingnearthecenterlineaftercumulativetrafficloadsinexcessof16millionESALs.Top-downcrackingatthePavementTestTrackhashistoricallyappearedas longitudinalcracking.Coreswerecutat thelocationofthelongitudinalcrack,fromwhichitwasconfirmedthatthecrackingwastop-down.SectionN9wasleftinplaceforcontinuingtrafficaspartofthe2012researchcycle,andafterapproximately28millionESALs,itonlyexhibitsevidenceoftop-downcracking.

Sixsections(N10,N11,andS8throughS11)placedaspartofthe2009NCATPavementTestTrackperformedwellduringthetwo-yeartraffickingperiodwithlittletonodistressesevidentandno fatigue cracking at the conclusionof the research cycle and applicationof 10millionESALs. The six sections were left in place for the 2012 research cycle, during which timedistressesbecamevisible.Crackingwasobservedinallsixsections.

Cores were extracted in four sections of the six sections (S8 through S11). The coresconfirmed that for Sections S8 through S11, the transverse crackingwas bottom-up and thelongitudinalcrackingwastop-down.TransversecrackinginS8wasfirstobservedinthespringof2013afterjustover12millioncumulativeESALs,andlongitudinalcrackingwaslaterevidentas well. In a similar fashion, Section S9 also exhibited longitudinal and transverse crackingduringthe2012researchcycle.CrackingwasfirstobservedinMarchof2013inthelongitudinaldirection after approximately 12 million cumulative ESALs and transverse cracking wasobservedshortlythereafter.SectionS10experiencedbothtransverseandlongitudinalcracking,whichwasfirstobservedinJanuaryof2013.LongitudinalcrackingwasfirstobservedinSection


20

S11 in December of 2012 after 10.7 million cumulative ESALs and transverse cracking wasevidentbythefollowingspring(2013).

Section N10 was the last of the six sections continued from 2009 to crack. TransversecrackingwasobservedinSectionN10inNovemberof2013afteracombined15.5millionESALshadbeenapplied.However,thisinitialtransversecrackingwasinanareaimmediatelyadjacenttoapatchplaced tocorrectanareaof severebut localizeddistress.Transversecrackingwasobserved inotherareas in the sectionbyFebruaryof2014.Coreswerenotextracted in thissection;however,itisbelievedthatthetransversecrackingoriginatedatthebottomoftheAClayerandpropagatedtothesurface,aswasthecaseinSectionsS8throughS11.SectionN11also experienced transverse cracking, and although cores were not extracted it likelypropagatedfromthebottomoftheACtothesurface.Longitudinalcrackingwasfirstobservedin Section N11 in February of 2013 and transverse cracking was evident shortly thereafter.Therewasextensivecrackingthatbecameinterconnectedintheoutsidewheelpath.Thisareaofinterconnectedcrackingwasthesiteoftheinitialobservationsoftransversecracking.Table5 summarizes the performance of the 2003, 2006, and 2009 test sections included in thisanalysis;shadingindicatessectionsthatdidnotexperiencefatiguecracking.

Table5FieldPerformanceofTestSectionsUsedinthisPerpetualPavementAnalysis

Section ACThickness(in.) YearBuilt FatigueCrackingN3 9.17 2003 NoN4 8.89 2003 NoN8 9.92 2006 YesN9 14.40 2006 NoN10 7.71 2006 YesS11 7.61 2006 YesN10 7.09 2009 YesN11 7.12 2009 YesS8 7.04 2009 YesS9 7.00 2009 YesS10 7.00 2009 YesS11 6.90 2009 Yes

3.3.3 AnalysisMethodology

Inthisanalysis, thestochasticperpetualpavementdesignsoftware,PerRoadversion3.5,wasutilized to predict horizontal tensile strain at the bottom of the AC layer for the twelvepavementsections.ThesoftwareutilizesMonteCarlosimulationtoallowfortheconsiderationof knownvariability associatedwithmaterial properties and constructionaswell as seasonalvariation effects on material moduli. Using the software, pavement responses at criticallocationsweredeterminedfromwhichstraindistributionsbasedonPerRoadpredictionswerecomputedandfatigueratioswerethendeveloped.

For each section, PerRoadwas utilized to completeMonte Carlo simulations, generating5,000predictionsoftensilestrainatthebottomoftheAClayer.Cumulativedistributionsweredevelopedfromthepredictedstrainforeachsectiontocomparewiththepreviouslydeveloped


21

cumulativedistributionsfromfield-measuredstrains.TopredicttensilestrainatthebottomoftheAClayerusingPerRoad,loadspectra,pavementlayermoduli,thickness,andtheassociatedcoefficientofvariation(COV)foreachwerenecessary.Adescriptionofeachinputfollows.

LoadSpectra.AxleweightsforeachofthefivetripletrailertrucksusedtoapplytrafficattheNCATPavementTestTrackweredeterminedpreviously(35).Basedonthetotalnumberofaxlesinthefleet,itwasdeterminedthatsingleaxlesrepresented71.42%ofthetotalnumberofaxlesappliedandthesteerandtandemaxleseachaccountedfor14.29%ofappliedaxles.Theaxle weights for each axle type (steer, tandem, and single) were entered into PerRoad tocharacterizethetrafficloadingsintheformofloadspectra:20%ofthesteeraxlesweighed8-10 kips, with the remaining 80% falling into the 10-12 kip range; 80% of the tandem axlesweighedbetween38and40kips,withtheremainingpercentageweighingbetween40and42kips;and100%ofthesingleaxlesweighedbetween20and22kips.

Cross-Sections. A three-layer structure was selected for each section with layer onerepresenting the AC, and layers two and three representing the unbound granular base andsubgrade,respectively.Fourrandomlocationswereidentifiedatthestartofeachtestcycleandit was at these four locations in the outside, between, and inside wheelpaths that layerthickness was surveyed during construction. From these measurements, the average layerthicknessesweredeterminedforeachsectionaswellasthecoefficientofvariation(COV)ofthelayerthickness,aslistedinTable6.SectionsN3andN4wereoriginallyconstructedin2003anddidnothavesurveyedthicknessesoftheunboundgranularbaselayer.Therefore,thesectionswereassumedtohaveathicknessequivalenttotheirdesignlayerthicknessof6inchesandaCOV equivalent to the average COV for the unbound granular base (GB) layer in the 2009sections.Anormaldistributionwasassignedtothelayerthicknessvariability.

Material Inputs. Falling weight deflectometer (FWD) testing was also conducted at fourrandomlocations in theoutside,between,and insidewheelpaths throughout thedurationofeach test cycle. FWD testing included three replicates at four drop heights (load levels).Measured deflectionswere used to conduct a three-layer (AC, unbound base, and subgradelayers)backcalculation inEVERCALCversion5.0. Section-specificunboundbaseand subgradelayerinputsweredeterminedbycalculatingtheaveragelayermodulusfortheentiretwo-yeartraffickingperiod.Averagebackcalculated layermoduliwere selected for the9-kip load level(load corrections were not applied) with root mean square error (RMSE) less than 3%. Thebackcalculatedbaseandsubgrademoduliatthe9-kiploadlevelarelistedinTable7alongwiththeir associated COV. For the PerRoad simulations, a normal distribution was used inconjunctionwith the COVs listed in Table 7 for the backcalculated layermoduli of both theunboundbase layerandsubgradematerial ineachsection.Althoughconstructed in2003,N3andN4 remained in-service for2006and2009 test cycles. For theanalysis conducted in thisstudy,backcalculatedmodulifromthe2006testcyclewereutilizedfortheanalysisofsectionsN3andN4; as such, these sections are labeledwith the year 2006 in reference topredictedstrains.


22

Table6LayerThicknessandAssociatedVariabilityBySection

Section hAC(in.) COVAC(%) hGB(in.) COVGB(%)N3(2006)1 9.17 2.4 6 8.8N4(2006)1 8.89 3.8 6 8.8N8(2006) 9.92 6.2 6.38 5.1N9(2006) 14.40 4.0 8.44 9.7N10(2006) 7.71 3.2 6.00 8.4S11(2006) 7.61 7.5 6.08 14.2N10(2009) 7.09 3.3 3.98 12.6N11(2009) 7.12 2.6 4.22 7.9S8(2009) 7.04 3.0 5.51 7.2S9(2009) 7.00 2.3 5.80 4.9S10(2009) 7.00 3.6 6.35 6.4S11(2009) 6.90 2.3 6.17 7.1

1Originallyconstructedin2003

Table7UnboundLayerModuliandAssociatedVariability

Section Base(ksi) Subgrade(ksi) COVGB(%) COVSG(%)N3(2006) 6.34 34.25 59.5 14.5N4(2006) 4.63 32.90 57.7 16.0N8(2006) 3.70 32.29 32.6 13.7N9(2006) 3.24 56.56 39.1 25.9N10(2006) 2.80 46.93 39.2 11.8S11(2006) 2.46 31.12 35.6 10.9N10(2009) 2.11 44.75 43.4 12.1N11(2009) 3.27 38.52 38.5 8.4S8(2009) 2.85 23.25 58.4 11.9S9(2009) 2.08 26.16 39.5 15.1S10(2009) 1.64 26.19 36.0 14.3S11(2009) 1.66 26.32 31.7 17.2

PerRoad allows forACmoduli to be entered for up to five seasons. First, section-specific

modulus-temperature relationships of the form listed in Equation 1 were developed withbackcalculatedACmoduliandthemid-depthpavementtemperaturesrecordedthroughoutthetwo-year testing cycles. The backcalculated moduli selected for the modulus-temperaturerelationshipwereatthe9-kiploadlevelandRMSElessthan3%.Hourlyaveragetemperatureswere recorded for each section during the entire two-year trafficking period. The averagehourly mid-depth pavement temperatures were then used in the modulus-temperaturerelationship to develop a cumulative distribution of AC moduli experienced throughout thetraffickingperiod.Oncedeveloped, thecumulativedistributionof theACmoduliwasusedtoselect seasonal moduli in each section. The midpoint of each quintile was selected as therepresentative ACmodulus for each of the five seasons: summer (10th percentile), spring 2


23

(30thpercentile), spring (50thpercentile), fall (70thpercentile), andwinter (90thpercentile).ThesemoduliarelistedinTable8foreachtestsection.PerRoadrequirestheusertoinputthenumberofweeksineachseason;therefore,10weekswereassignedtospring,fall,andwinter,whilesummerandspring2wereeachassigned11weekstobeconservative.TheCOVfortheAC modulus was calculated from temperature-corrected AC moduli. To do so, thebackcalculatedACmoduli,E1,werecorrectedto68°F,usingEquation2,andthentheaverageand standard deviation of the correctedmoduli were calculated to determine the COV. TheCOVforthecorrectedACmoduliarealso listed inTable8foreachtestsection.A log-normaldistributionofACmoduliforeachsectionwasselectedforthePerRoadsimulations.𝐸" = 𝛼" 𝑒&'( (1)where E1 =ACmodulus(psi);

α1, α2 =regressioncoefficients;andT =mid-depthpavementtemperature(°F).

𝐸() = 𝐸"(𝑒&' ()+( ) (2)where E1 =ACmodulusatT(psi);

ΕTr =ACmodulusatTr(psi);T =mid-depthpavementtemperature(°F);andTr =referencetemperatures,68°F.

Table8ACVariabilityandACModulibySeasonforPerRoadSimulations

DesignModulus,ksiSection COV(%) Summer Spring2 Spring Fall WinterN3(2006) 36.5 302.29 517.86 733.32 1022.28 1536.61N4(2006) 22.5 306.32 516.40 852.54 1309.14 1925.96N8(2006) 19.3 155.01 287.76 476.78 833.04 1333.62N9(2006) 22.4 240.45 384.75 654.68 977.31 1361.62N10(2006) 17.6 178.10 315.85 539.74 821.80 1197.28S11(2006) 17.2 139.64 249.47 451.50 760.88 1159.49N10(2009) 12.0 351.80 575.77 882.59 1432.38 2240.17N11(2009) 8.8 312.14 499.75 771.00 1288.27 2009.92S8(2009) 16.6 254.10 394.25 579.65 911.68 1386.25S9(2009) 12.5 249.20 397.99 632.82 1109.39 1794.56S10(2009) 11.5 232.54 390.86 597.95 996.86 1599.14S11(2009) 14.6 234.02 379.66 610.51 1018.44 1616.85


24

Fatigue Ratios. The maximum fatigue ratios based on the predicted strains weredeterminedusingEquation3.

Rn=εn/εf (3)where Rn =fatigueratioatthenthpercentile;

εn =microstrainatthenthpercentile;andεf =laboratory-determinedfatiguethresholdorendurancelimit,microstrain.

Thelaboratory-determinedfatigueendurancelimitsforthe2003sections(N3andN4)were

established by first conducting BBFT on samples compacted to 7.0% target air voids inaccordancewithAASHTOT321.AsdocumentedbyWillisandTimm(16),thefatigueendurancelimitwasdeterminedaspartoftheNCHRP9-38project(10)byapplyingathree-stageWeibullequation andwas taken as the 95%one-sided lower prediction limit. The fatigue endurancelimitsforthe2006sectionswerealsodeterminedinthesamemanner;however,sampleswerecompactedto5.5%targetairvoidsattwostrainlevels,400and800microstrain(28).Bendingbeam fatigue testing was also conducted for the base AC mixtures in the 2009 sectionsfollowingAASHTOT321usingspecimenscompacted to targetairvoidsof7.0%and testedatstrainlevelsof200,400,and800microstrain.LowertargetairvoidswereselectedforBBFTofthe sectionsplaced in2006 tobe representativeof the richbottom layer (lowair voidsmix)utilized insectionsN8andN9.Theremainingsectionsplaced in2006werealsoevaluatedatthesameairvoidleveltoenablerelativecomparisonsamongthesections.Table9providesthefatigueendurancelimitstakenasthe95%one-sidedlowerpredictionfortheselectedsections.

Table9LaboratoryFatigueEnduranceLimitafter(16,28)

Section EnduranceLimit-95thone-sidedlowerpredictionlimit,microstrain

TargetAirVoidsofBBFTSpecimens

N3(2003) 151 7.0%N4(2003) 146 7.0%N8(2006) 203 5.5%N9(2006) 203 5.5%N10(2006) 130 5.5%S11(2006) 118 5.5%N10(2009) 100 7.0%N11(2009) 134 7.0%S8(2009) 92 7.0%S9(2009) 92 7.0%S10(2009) 99 7.0%S11(2009) 84 7.0%


25

3.3.4 RefinedLimitingStrainCriteriaforUseinPerpetualPavementDesign

AsaresultofthePerRoadsimulations,tensilestrainswerepredictedatthebottomoftheAClayerineachtestsectionundertheappliedloadspectraaccountingforseasonalvariationinACmoduli and measured variability in layer moduli and thickness. Using the predicted tensilestrains,acumulativestraindistributionwasdeterminedforeachsection.Figure10comparesthepredictedcumulativestraindistributionsforthe2006testsections.

Figure10CumulativeDistributionsofPredictedTensileStrain,2006Sections.

As shown in Figure 10, there is an obvious separation at the predicted cumulative strain

distributionsforSectionsN3andN4,asallsectionswherepredictedstraindistributionsfalltotherightofthesecurvesexhibitedbottom-upfatiguecrackingandallsectionswherepredictedstraindistributionsfall tothe leftdidnot.This isconsistentwiththeobservationsmadefromfield-measured strains in the previous study (16). Since the cumulative distributions forpredicted strains in SectionsN3 andN4 lie nearly on top of one another and because thesesections have the highest strain levels without experiencing fatigue cracking, they wereselectedforrefiningthelimitingcumulativedistributionofpredictedtensilestrains.Todoso,consistentwith thesamemethodology followed fordeveloping the field-based limitingstraindistribution,theaverageoftheN3andN4predictedstrainvaluesateachpercentilelevelweredetermined and plotted as a black solid line with the cumulative distributions of predictedstrainsof theother sections in Figure10. For comparison, the field-baseddistribution is alsoshown in Figure10as ablackdashed line.Values for thisnew limitingdistributionbasedonpredictedstrainsarelistedinTable10.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 200 400 600 800 1000 1200

Percen

tile

TensileStrain(microstrain)

N3- 06(Uncracked)

N4- 06(Uncracked)

N8- 06(Cracked)

N9- 06(Uncracked)

N10- 06(Cracked)

S11- 06(Cracked)

FieldFatigueLimit

UpdatedFatigueLimit(PerRoad)


26

Table 10 Refined Limiting Distribution and Maximum Fatigue Ratios for Predicted TensileStrain

Percentile LimitingDesignDistributionforPredictedStrain

MaximumFatigueRatioforPredictedStrain

1% 29 5% 41 10% 48 15% 54 20% 60 25% 66 30% 71 35% 78 40% 84 45% 91 50% 100 0.6855% 110 0.7460% 120 0.8165% 131 0.8870% 143 0.9675% 158 1.0680% 175 1.1885% 194 1.3190% 221 1.4995% 257 1.7399% 326 2.19

The predicted strain values of N3 and N4 at each percentile level were divided by the

corresponding FELs (Table 9) to determine the fatigue ratios for these sections. The averageratioofN3andN4foreachpercentilewasthencalculatedtorefinethemaximumfatigueratiosto reflect tensile strain valuespredicted in PerRoad. The refinedmaximum fatigue ratios arelistedinTable10.Figure11comparestherefinedmaximumfatigueratios(blacksolidline)tothoseoftheother2006sectionsevaluated.Ratiosthatfailthecriteriaareontherightofthemaxfatigueratioline(labeled“UpdatedFatigueRatio(PerRoad)”).Asexpected,SectionsN10andS11 fail these criteria.However, SectionN8, a section that experienced fatigue cracking,barelypassesthecriterionatthe50th,55th,and99thpercentile.Forthesecriteriatoserveasanindicator of the ability of a perpetual pavement design to withstand bottom-up fatiguecracking,thecriteriaatallpercentilesfromthe50ththroughthe99thpercentileshouldbemet.Despiteapredictedstraindistributionthat isclearlyfartotherightoftherefinedcumulativedistribution, N8 had fatigue ratios that barely failed. This can be attributed to its very highlaboratory-determinedfatigueendurancelimitof203microstrain.


27

Figure11FatigueRatiosbasedonPredictedStrainsandFEL,2006Sections.

3.3.5 ValidatingRefinedDesignThresholds

To determine if the refined limiting strain criteria (cumulative distribution and maximumfatigueratiosbasedonpredictedstrains)werevalid,sixsectionsfromthe2009researchcycleweresimulatedinPerRoad.Theresultingpredictedtensilestrainvalueswereusedtodevelopcumulative straindistributionsaswasdonewith the2006 test sections. Figure12 shows thepredictedstraindistributionsforeachofthe2009sectionswiththerefinedlimitingcumulativedistributionaswellastheoriginalfield-basedlimitingstraindistributionforcomparison.Allsixsectionsfrom2009arebelievedtohavebottom-upfatiguecracking(fourofthesectionshavebeenconfirmedwithfieldcoresandtheothertwoexhibitcrackingconsistentwithbottom-upcracking historically observed at the test track). As was the case with predicted straindistributionsforthe2006sections,thefield-basedlimitingstraindistributiondidnoteffectivelycategorizesectionspronetofatiguecrackingbasedonpredictedstrainvalues.Predictedstraindistributions for sections N10 and N11 both fall to the left of the field-based limiting straindistribution,whichwouldindicatethattheypassedthiscriteria.Thisisagainanartifactoftheunderprediction of tensile strains. Since field-measured strains are not available during thedesign process, the limiting strain distribution based on predicted strains from PerRoad isnecessary for perpetual pavement design. The design limiting strain distribution, shown inFigure12as“UpdatedFatigueLimit”,accuratelycategorizesallsixsectionsaspronetofatiguecracking.Allsixpredictedstraindistributionsfalltotherightofthedesignlimitingdistributionand all six sections experienced bottom-up fatigue cracking. This evaluation validates therefined limiting strain distribution based on predicted strain values for use in perpetualpavementdesign.

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

0.0 2.0 4.0 6.0 8.0 10.0

Percen

tile


N3-06(Uncracked)

N4-06(Uncracked)

N8-06(Cracked)

N9-06(Uncracked)

N10-06(Cracked)

S11-06(Cracked)

FieldFatigueRatio

UpdatedFatigueRatio(PerRoad)


28

Figure12ValidatingRefinedLimitingFatigueDistributionUsing2009Sections.

Fatigue ratios were determined for the 2009 test sections using the predicted strain

distributionsshowninFigure12andtheassociatedfatigueendurance limits listed inTable9.The2009testsectionsservedasavalidationdataset,astheywerenotusedtorefinethefield-based strain distribution andmaximum fatigue ratios to incorporate predicted strain values.Figure 13 shows the fatigue ratios for each of the 2009 test sections, the refinedmaximumfatigueratios,andthefieldfatigueratios.Themaximumfatigueratioswererefinedtoreflectpredictedstrainsforuseindesign.Themaximumfatigueratiosaremeanttoserveascriteriaforperpetualpavementdesignsuchthatratiosatallpercentilesfrom50to99shouldbelessthan the maximum fatigue ratios to achieve a section that will behave perpetually, as didsectionsN3,N4,andN9.Itisexpectedthatsinceallsixsectionsfromthe2009researchcycleexperiencedbottom-upfatiguecracking,thefatigueratiosforallsixsectionsshouldbegreaterthanthemaximumfatigueratios.Allsectionsclearlyexceedthemaximumdesignratioateachpercentilefrom50to99,thusvalidatingtherefinedlimitingcriteriabasedonstrainspredictedfromPerRoadforperpetualpavementdesign.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 200 400 600 800 1000 1200

Percen

tile

PredictedTensileStrain(microstrain)

N10-09(Cracked)

N11-09(Cracked)

S8-09(Cracked)

S9-09(Cracked)

S10-09(Cracked)

S11-09(Cracked)

FieldFatigueLimit

UpdatedFatigueLimit(PerRoad)


29

Figure13ValidatingRefinedMaximumFatigueRatiosUsing2009Sections.

4 APPROXIMATERANGESOFMAXIMUMDESIGNTHICKNESSES

The limiting strain distribution presented in the previous section was used to developapproximate ranges of maximum design thicknesses for future design consideration. Theanalysisconductedinthisstudytodevelopapproximaterangesofmaximumdesignthicknesseswas similar to that conducted in SHRP 2 Project R23 (22). Both analyses utilized the limitingstainapproachvia thePerRoadsoftware (Version3.5).Themaindifferencebetweenthetwoanalyseswasthecriteriausedtoselectthefinalthicknessdesignasfollows.

• In the SHRP 2 R23 analysis, the final AC thicknesswas selected if it would provide adamageratiolessthanorequalto0.1at10yearsand50yearsofservicelifebasedonafatigueendurancelimitof100microstrain.

• Inthisstudy,thefinalthicknessofAClayerwaschosenbasedonthetwocriteria:thefirstcriterionwastoavoidthedevelopmentoffatiguecracking,andthesecondcriterionwastolimitstructuralruttingoccurredinsubgrade.

o ThecumulativedistributionofthecalculatedtensilestrainsatthebottomoftheAClayerwaslowerthanthelimitingstraindistributionlistedinTable10;and

o 50%(or50percentile)ofverticalstrainscalculatedatthetopofsubgradewerebelow200microstrain.

OtherinputsneededforthePerRoadsimulationswereselectedasdiscussedbelow.• Onetrafficlevelwassimulated,whichconsistedof100%ofsingleaxlesweighing20-22

kips. This representsa conservative traffic levelwithin the legal load limitsallowed intheU.S.

• Thepavementstructures (assimulated)hadthree layers, includinganAC layeroverabaselayeroversubgrade.Table11liststheinputsforeachlayer.

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Percen

tile


N10-09(Cracked)

N11-09(Cracked)

S8-09(Cracked)

S9-09(Cracked)

S10-09(Cracked)

S11-09(Cracked)

FieldFatigueRatio

UpdatedFatigueRatio(PerRoad)


30

o For the AC layer, the moduli were influenced by the binder performance gradeselected inthesoftwareandseasonal temperaturesdiscussedbelow.TheselectedperformancegradewasaPG64-34forMinneapolis,aPG70-22forPhoenix,andaPG64-22 forBaltimore, consistentwith the SHRP2R23 analysis. TheAC thicknesswasdesignedtomeetthedesigncriteria.

o Thebase layer thicknesswas selected tobe6, 8, and10 inches. Fivebasemoduliwereusedinthesimulations,including30,50,100,250,and500ksi.

o Threesubgrademoduliwereselectedforthesimulations,including5,10,and20ksi.• TheseasonaltemperaturesaffectingtheACmoduliusedinthisanalysisarethesameas

thoseusedintheSHRP2R23analysis.IntheSHRP2R23study,trialrunswereinitiallyconducted for fivecities (Minneapolis,MN;SanFrancisco,CA;Phoenix,AZ;Dallas,TX;andBaltimore,MD).However,itwasfoundthatthethicknessvaluesforSanFranciscoandDallas fellwithin therangeof thicknessesdetermined for theothercitiesanddidnotaffecttheaveragessignificantly.Thus,theanalysisconductedaspartofthisstudyincluded only three cities: Minneapolis, Phoenix, and Baltimore. Table 12 lists theseasonaltemperaturesforthethreecitiesasreportedintheSHRP2R23report(22).

Table11InputsforPavementStructuresUsedinPerRoadSimulations

Layer Modulus(psi) Poisson’sRatio ThicknessInput

Distribution

(COV)Input Average Distribution

(COV)AC Variedbasedon

binderandseasonLog-normal

(30%)0.35 Varied Normal

(5%)Base 30,50,100,250,

and500ksiLog-normal

(40%)0.4 6,8and

10in.Normal(8%)

Subgrade 5,10,and20ksi Log-normal(50%)

0.45 Semi-infinite

Notapplicable

Foreachpavementdesignsimulation,thefollowingstep-by-stepprocedurewasfollowed:1. InPerRoad,opentheStructuralandSeasonalInformationwindow.

a. Select thenumberof layers. For thisanalysis, thenumberof layerswas three,includingAC,aggregatebase,andsubgrade.

b. Input the seasonal information. For this analysis, the seasonal information,including the number ofweeks andmean air temperature for each season, isshowninTable12.

c. Select theperformancegrade (PG)of thebinderused in theAC layer. For thisanalysis,aPG64-34wasselectedforMinneapolis,aPG70-22forPhoenix,andaPG64-22forBaltimore.Then,inputPoisson’sratio,initialthickness,distribution,andCOVfortheAClayersasshowninTable11.

d. Input the moduli, Poisson’s ratios, distributions, and COVs for the base andsubgradeasshowninTable12.

e. Acceptchangesinthiswindow.2. OpentheLoadingConditionswindow.


31

a. Select100%singleaxlesweighing20-22kipstorepresentaconservativetrafficlevelwithinthelegalloadlimitsallowedintheU.S.

b. Acceptchangesinthiswindow.c. Note:detailsofthetrafficstream(two-wayAADT,%trucks,%truckgrowth,etc.)

arenotinput,asthestrainduetoasingleaxleloadisofinterestratherthanthenumberofcyclestofailure.

3. PerformPerRoadanalysistopredictthetensilestrainatthebottomofthelayerandtheverticalstrainatthetopofsubgrade.

4. Open the PerRoad output file inMS Excel and determine the cumulative tensile andverticalstraindistributions.

5. Checkthedistributionstodetermine:a. If the cumulative tensile strain distribution at the bottomof theAC layerwas

lowerthanthelimitingstraindistributionlistedinTable10;andb. Ifthe50percentileoftheverticalstraindistributionatthetopofsubgradewere

below200microstrain.6. Ifoneoftheabovecriteriadidnotpass,adjustACthicknessandrepeatsteps4,5,and6

untilallcriteriaaremet.

Table12SeasonalTemperaturesforMinneapolis,Phoenix,andBaltimore(22)

City OverallMeanTemperature(°F)

SeasonalInformationMonth Week Temperatures(°F)

Minneapolis 45°F Winter Nov,Dec,Jan,Feb 17weeks 21 Spring Mar,Apr,May 13weeks 45 Summer June,July,Aug 13weeks 70 Fall Sept,Oct 9weeks 56Phoenix 70°F Winter Dec,Jan,Feb 13weeks 54 Spring Mar,Apr,May 13weeks 68 Summer June,July,Aug,Sept 17weeks 87 Fall Oct,Nov 9weeks 66Baltimore 56°F Winter Dec,Jan,Feb 13weeks 35 Spring Mar,Apr,May 13weeks 54 Summer June,July,Aug,Sept 17weeks 74 Fall Oct,Nov 9weeks 53


32

Tables13,14,and15summarizeresultsoftheanalysisandaveragethicknessesfor6-inch,8-inch,and10-inchbases,respectively,withresilientmoduliof30,000,50,000,and100,000psi.Since the design scenarios simulated in this study used conservative inputs, the ranges ofdesign thicknesses shown in these tables represent conservative design thicknessesencounteredinfuturepavementdesign.Similarmaximumthicknesstablescanbedevelopedtorepresentstate-specificclimate,material,andsubgradeconditionsforuseinconjunctionwiththeagency-specificdesignprocedure.Whenthethicknessofaneworrehabilitatedpavementdesign based on the agency-specific design methodology is greater than the maximumthickness,theagencymayconsiderusingtheperpetualpavementdesignapproachtooptimizeadesignthatcansustaintheheaviestloadswithoutbeingoverlyconservative.

Table13RangesofMaximumACThicknessesfor6-inchBase(Mr=30,50,and100ksi)

Subgrade Base CalculatedACThickness(in.) RangeofMr(ksi) Mr(ksi) Minneapolis Phoenix Baltimore Average Maximum

(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 30 12.5 15.5 14 14.0 12.5-15.55 50 12 15 14 13.7 12-155 100 12 14 13.5 13.2 12-1410 30 10.5 14 12 12.2 10.5-1410 50 10.5 13 12 11.8 10.5-1310 100 10 12 11 11.0 10-1220 30 9 12.5 10 10.5 9-12.520 50 8.5 12.5 9.5 10.2 8.5-12.520 100 8 12 9 9.7 8-12

Table14RangesofMaximumACThicknessesfor8-InchBase(Mr=30,50,and100ksi)


(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 30 12.5 15 14 13.8 12.5-155 50 11.5 14.5 13.5 13.2 11.5-14.55 100 11 13 12.5 12.2 11-1310 30 10.5 13 11.5 11.7 10.5-1310 50 10 12 11.5 11.2 10-1210 100 9 11 10.5 10.2 9-1120 30 9 12.5 10.5 10.7 9-12.520 50 8.5 12 10 10.2 8.5-1220 100 7.5 10.5 9 9.0 7.5-10.5


33

Table15RangesofMaximumACThicknessesfor10-InchBase(Mr=30,50,and100ksi)


(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 30 12 14.5 13.5 13.3 12-14.55 50 11 13.5 12.5 12.3 11-13.55 100 10 12 11.5 11.2 10-1210 30 10 12 11 11.0 10-1210 50 9 11 10 10.0 9-1110 100 8 10 9 9.0 8-1020 30 8.5 11 10.5 10.0 8.5-1120 50 7.5 10 9.5 9.0 7.5-1020 100 6.5 9 8.5 8.0 6.5-9

Asimilaranalysiswasalsoconductedforstiffbases,suchasrubblizedorcrackedandseatedoldportlandcementconcrete(PCC)pavements,withresilientmoduliof250,000and500,000psi.ResultsofthisanalysisandaveragethicknessesareshowninTables16,17,and18for6-inch,8-inch, and 10-inch bases, respectively. These results will require further evaluation andvalidation as the pavement sections used in the development and validation of the limitingstraincriteriapresentedinthisreportdidnotincludethosethatwereconstructedonrubblizedorcrackedandseatedoldPCCpavements.ThisisespeciallyimportantfortheverythinsectionsinTables17and18wherethebasemodulusis500,000psi.Itislikelythatthesesectionswouldsufferfromreflectivecrackingnotaccountedforinperpetualpavementanalysis.

Table16RangesofMaximumACThicknessesfor6-inchBase(Mr=250and500ksi)


(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 250 9.5 12 11 10.8 9.5-125 500 7.5 9.5 9 8.7 7.5-9.510 250 8 10 9.5 9.2 8-1010 500 6 8 7 7.0 6-820 250 6.5 8 7.5 7.3 6.5-820 500 5 6 5.5 5.5 5-6


34

Table17RangesofMaximumACThicknessesfor8-InchBase(Mr=250and500ksi)


(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 250 8.5 10.5 10 9.7 8.5-10.55 500 6 7.5 7 6.8 6-7.510 250 7 8.5 8 7.8 7-8.510 500 5 6 5.5 5.5 5-620 250 5.5 6.5 6 6.0 5.5-6.520 500 3 4 3.5 3.5 3-4

Table18RangesofMaximumACThicknessesfor10-InchBase(Mr=250and500ksi)


(PG64-34) (PG70-22) (PG64-22) Thicknesses(in.)5 250 7 9 8 8.0 7-95 500 4.5 5.5 5.5 5.2 4.5-5.510 250 6 7 6.5 6.5 6-710 500 4.5 4 3.5 4.0 3.5-4.520 250 4 5 4.5 4.5 4-520 500 1.5 2 2 1.8 1.5-2

5 CONCLUSIONS

The objective of this study was to determine critical pavement design thresholds andapproximaterangesofmaximumthicknessesforflexiblepavements.Thestudywasdividedintotwotasks.Thefirsttaskwastoreviewliteraturepertainingtodesignthresholdsandmaximumthicknessrequirements forperpetualpavements.ThesecondtaskwastoevaluateandrefinedesignthresholdsandtodeterminemaximumpavementthicknessesbasedontheinformationreviewedinTask1andthroughanalyzingpavementresponsedatafromthefullyinstrumentedpavementsectionsattheNCATPavementTestTrack.Basedontheresultsofthetwotasks,thefollowingconclusionsaremade.

• A perpetual pavement is designed to resist structural distresses, including bottom-upfatiguecrackingandsubgraderutting.Historically, topreventthe initiationofbottom-upfatiguecracking,thestrainsatthebottomoftheasphaltstructurearekeptbelowadesign strain threshold,which is often the laboratory fatigueendurance limit (FEL) oftheasphaltmixtureusedintheACbaselayer.Topreventstructuralrutting,theverticalstrain or stress at the top of the subgrade has been used as the limiting designparameter.

• The review of literature shows that the FEL has ranged in magnitudes from earlyestimatesof70microstrain tomorerecentestimatesofupto200microstrain.Avalueof 200 microstrain has been proposed for the vertical strain limit. In addition, past


35

perpetualpavementdesignsforhighvolumeroadwayshadbetween9and16inchesofACdependingonsite-specifictraffic,climate,material,andsubgradeconditions.

• StudieshaveshownthattheearlyestimatedFELof70microstrainwasconservative.Inaddition, as shown in Table 9, the FELs of all the mixtures (including unmodified,modified,rich-bottom,andhighRAPmixtures)analyzedinthisstudywerehigherthan70microstrain.

• Whileasingle limitingstrain(i.e.,70microstrain)ortheFELoftheACbaselayerhavebeen proposed for designing long life pavements, neither were a good indicator ofresistancetobottom-upfatiguecrackinginthe2003and2006structuraltestsectionsattheNCATPavement Test Track.Rather, the test sections that experiencedbottom-upfatigue cracking had cumulative field-measured strain distributions clearly separatedfrom those of test sections that did not crack. Based on this finding, a limitingcumulative strain distribution was developed based on field-measured strains forcontrollingbottom-upfatiguecrackinginapaststudy.

• Therearenotabledifferencesbetweenfield-measuredstrainsatthebottomoftheAClayer and tensile strains predicted by a structural pavement design program, such asPerRoad.Asaresult,thelimitingstraindistributionbasedonfield-measuredstrainswasadjustedtotakesuchdifferences intoaccount.Theadjusted limitingcumulativestraindistribution listed in Table 10 is proposed for use in place of a single FEL in futureperpetualpavementdesigntocontrolbottom-upfatiguecracking.

• Additionally,ifthelaboratoryFELoftheACbaselayerisavailableduringthepavementdesign process, the fatigue ratios at incremental percentiles can be determined bydividing the corresponding cumulative strains predicted by the pavement designsoftware by the FEL. These fatigue ratios can be compared with the limiting fatigueratioslistedinTable10,whichwerealsorefinedandvalidatedinthisstudy.

• The limiting strain distribution was used later in this study to develop approximaterangesofmaximumdesignthicknessesforasphaltpavements.ThisanalysiswassimilartothatconductedinStrategicHighwayResearchProgram2ProjectR23.Bothanalysesused the limiting strain approach via the PerRoad software (Version 3.5). The maindifferencebetweenthetwoanalyseswasthecriteriausedtoselectthefinalthicknessdesign.Theanalysiswasconductedbasedonaconservativetrafficlevelwithinthelegalload limits for various combinations of subgrade and base moduli in three climaticconditionstocoverthepotentialrangesofmaximumdesignthicknesses.TheresultingapproximaterangesofmaximumdesignthicknessesforasphaltpavementsareshowninTables 13, 14, and 15 for base layerswith resilientmoduli of up to 100,000 psi. Themaximumthicknesshasbetween6.5and15.5 inchesofACdependingonsite-specificclimate,material,andsubgradeconditions.

• Additionalanalysiswasalsoconductedforpavementsconstructedonstiffbaselayers,such as those built on rubblized and cracked and seated pavements, with resilientmoduliof250,000and500,000psi.ResultsofthisanalysisareincludedinTables16,17and 18. These results will require further field evaluation and validation as thepavementsectionsusedfordevelopingandvalidatingthelimitingstraincriteriainthisstudy did not include those built on these stiff base layers, especially the very thin


36

sectionsinTables17and18whenthebasemodulusis500,000psi.Itislikelythatthesethinsectionswouldsufferfromreflectivecrackingthatisnotaccountedforinperpetualpavementanalysis.

6 RECOMMENDATIONS

Basedontheresultsandconclusionsofthisstudy,thefollowingrecommendationsaremade.• The limitingcumulativestraindistributionandfatigueratiosshowninTable10should

beused inplaceoftheconservative limitingstrainof70microstrainorthe laboratoryFELoftheACbaselayerinfutureperpetualpavementdesign.

o These limiting values have been adjusted for the differences between field-measured and predicted tensile stains at the bottom of the AC. The adjustedlimitingcumulativestraindistributionwasfoundtobethebestindicatorofhowthe structural test sections resisted bottom-up fatigue cracking at the NCATPavementTestTrack.

• Tables13,14,and15showtherangesofmaximumACthicknessesforflexiblepavementdesign. Similar tables can be developed based on themethodology presented in thisreport for each state that represents state-specific climate, material, and subgradeconditions.

• Furtherfieldevaluationandvalidationofthelimitingstraincriteriashouldbeconductedto include pavement sections that were built on stiff base layers, such as thoseconstructedon rubblized and cracked and seatedold concrete pavements. Additionalthickness may be needed to resist reflective cracking not considered in perpetualpavementanalysis.

• ThemaximumACthicknesstablescanbeused inconjunctionwiththeagency-specificdesign procedure.When the thickness of a new or rehabilitated pavement designedbased on the agency-specific design methodology is greater than the maximumthickness,theagencymayusetheperpetualpavementdesignapproachtooptimizethedesignthatcansustaintheheaviestloadstoprovideanindefinitestructurallifewithoutbeingoverlyconservative.


37

7 REFERENCES

1. AASHTO.AASHTOGuide forDesignofPavementStructures,AASHTO,Washington,DC,1993.

2. TRB. “Pavement Lessons Learned from theAASHORoadTest andPerformanceof theInterstate Highway System,” Transportation Research Circular Number E-C118,TransportationResearchBoard,Washington,DC,2007.

3. Newcomb,D.,R.Willis, andD.Timm.PerpetualAsphaltPavements:ASynthesis,APA,Lanham,MD,2010.

4. AsphaltPavementAlliance(APA).PerpetualPavements:ASynthesis.APA101,Lanham,MD,2002.

5. Nunn, M., and B.W. Ferne. “Design and Assessment of Long-Life Pavements,”TransportationResearchCircularNo.503,TransportationResearchBoard,Washington,DC,2001,pp.32-49.

6. Thompson,M.R. and S.H. Carpenter. “ConsideringHot-Mix-Asphalt Fatigue EnduranceLimits inFull-DepthMechanistic-EmpiricalPavementDesign,”InternationalConferenceonPerpetualPavements,Columbus,Ohio,2006.

7. Bonaquist, R. “Developing a Plan for Validating an Endurance Limit for HMAPavements,” NCHRP 9-44, National Cooperative Highway Research Program, HMAWorkshopExecutiveSummary,2007.

8. Monismith, C. and D.McLean. “Structural Design Considerations,”Proceedings of theAssociationofAsphaltPavingTechnologists,Vol.41,1972.

9. Thompson, M.R. and S.H. Carpenter. “Perpetual Pavement Design: An Overview,”InternationalConferenceonPerpetualPavements,Columbus,Ohio,2009.

10. Prowell, B., E.R. Brown, E.M. Anderson, J.S. Daniel, H. Von Quintus, S. Shen, S.Carpenter,S.BhattacharjeeandS.Maghsoodloo.ValidatingtheFatigueEnduranceLimitforHotMixAsphalt,NCHRPReport646,2010.

11. Carpenter,S.H.andS.Shen.“EffectofMixtureVariablesontheFatigueEnduranceLimitfor Perpetual Pavement Design,” International Conference on Perpetual Pavements,Columbus,Ohio,2009.

12. Nishizawa, T., S. Shimeno, andM. Sekiguchi. “Fatigue Analysis of Asphalt PavementswithThickAsphaltMixtureLayer,”Proceedingsof the8th InternationalConferenceonAsphalt Pavements, Vol. 2, University of Washington, Seattle, WA, August 1997, pp.969–976.

13. Wu, Z., Z. Q. Siddique, and A. J. Gisi. “Kansas Turnpike—An Example of Long LastingAsphaltPavement,”ProceedingsInternationalSymposiumonDesignandConstructionofLong LastingAsphaltPavements,NationalCenter forAsphalt Technology,Auburn,AL,2004,pp.857–876.

14. Yang, Y., X. Gao, W. Lin, D.H. Timm, A.L. Priest, G.A. Huber, and D.A. Andrewski.“Perpetual Pavement Design in China,” Proceedings of International Conference onPerpetualPavement,OhioUniversity,2006.

15. VonQuintus,H. “Applicationof theEndurance Limit Premise inMechanistic-EmpiricalBased Pavement Design Procedures,” International Conference on PerpetualPavements,Columbus,Ohio,2006.


38

16. Willis, J.R.andD.Timm.Field-BasedStainThresholds forFlexiblePerpetualPavementDesigns,NCATReport09-09,AuburnUniversity,2009.

17. Christopher, B., C. Schwartz, R. Boudreau.GeotechnicalAspects of Pavements, ReportNo.NHI-05-037,FHWA,U.S.DepartmentofTransportation,Washington,D.C.,2006.

18. Monismith,C., J.Harvey,T.Bressette,C.Suszko,andJ.St.Martin.“TheI-710FreewayRehabilitationProject:Mix and Structural SectionDesign, ConstructionConsiderationsand Lessons Learned,” International Symposium on Design and Construction of LongLastingAsphaltPavements,NationalCenterforAsphaltTechnology,Auburn,AL,2004,pp.217-262.

19. Walubita,L.F.,W.Liu,T.ScullionandJ.Leidy.“ModelingPerpetualPavementsUsingtheFlexible Pavement SystemSoftware,”Proceedingsof the87th TransportationResearchBoardMeeting,TransportationResearchBoard,Washington,D.C.,2008.

20. Bejarano,M.O. andM. R. Thompson. “Subgrade Damage Approach for the Design ofAirport Flexible Pavements,” Proceedings of the 2001 Airfield Pavement SpecialtyConference: Advancing Airfield Pavements, American Society of Civil Engineers.Washington,D.C.,2001,pp.48-58.

21. Nunn,M.“Long-lifeFlexibleRoads,”Proceedingsofthe8thInternationalConferenceonAsphaltPavements,Vol.1,UniversityofWashington,Seattle,WA,August1997,pp.3–16.

22. Jackson, N., J. Mahoney, and J. Puccinelli. Using Existing Pavement in Place andAchievingLongLife,FinalReport,SHRP2R23,TRB,Washington,D.C.,2012.

23. Rosenberger, C.E., T.J. Zurat and R.J. Cominsky. “A Practical Look at Pennsylvania’sBradford Bypass-A Perpetual Pavement,” Proceedings of International Conference onPerpetualPavements,Columbus,Ohio,2006.

24. Lane, B., A.W. Brown and S. Tighe. “Perpetual Pavements: The Ontario Experiment,”Proceedings of International Conference on Perpetual Pavements, Columbus, Ohio,2006.

25. Sargand, S.M., I.S. Khoury, M.T. Romanello and J.L. Figueroa. “Seasonal and LoadResponse Instrumentation of the Way-30 Perpetual Pavements,” Proceedings ofInternationalConferenceonPerpetualPavements,Columbus,Ohio,2006.

26. Walubita,L.F.,T.Scullion,J.LeidyandW.Lin.“AReviewoftheTexasStructuralDesignCriteriaforPerpetualPavement,”ProceedingsofInternationalConferenceonPerpetualPavements,Columbus,Ohio,2009.

27. Timm,D.H.andM.M.Robbins.“PerpetualPavementResearchattheNCATTestTrack,”Proceedingsof the InternationalConferenceonPerpetualPavements,Columbus,Ohio,2014.

28. Willis,J.R.,D.Timm,A.Taylor,N.Tran,andA.Kvasnak.“CorrelatingLaboratoryFatigueEndurance Limits to Field-Measured Strains,” Journal of the Association of AsphaltPavingTechnologists,Volume80,2011,pp.135-160.

29. Robbins,M.,N.Tran,D.Timm,andJ.R.Willis.“AdaptationandValidationofStochasticLimitingStrainDistributionandFatigueRatioConceptsforPerpetualPavementDesign,”PapersubmittedtoAAPT,2014,pp.32.


39

30. Robbins, M. and D. Timm. “Effects of Strain Pulse Durations on Tensile Strain in aPerpetual Pavement,” Proceedings of the International Conference on PerpetualPavement,Columbus,Ohio,2009.

31. Timm, D., N. Tran, A. Taylor, M. Robbins, and R. Powell. “Evaluation of MixturePerformanceandStructuralCapacityofPavementsUtilizingShell Thiopave®;Phase I:MixDesign, LaboratoryPerformanceEvaluationandStructuralPavementAnalysis andDesign,”ReportNo.09-05,NationalCenterforAsphaltTechnology,AuburnUniversity,2009.

32. Timm, D.H. and A. PriestL. “Flexible Pavement Fatigue Cracking andMeasured StrainResponse at the NCAT Test Track,” Proceedings of the 87th Annual Meeting of theTransportation Research Board, Transportation Research Board, National ResearchCouncil,Washington,D.C.,2008.

33. Timm,D.H.,“Design,ConstructionandInstrumentationofthe2006TestTrackStructuralStudy,”ReportNo. 09-01,National Center for Asphalt Technology, AuburnUniversity,2009.

34. Timm,D.,M.Robbins, and J.R.Willis. “Characteristicsof TwoPerpetualPavementsattheNCATTestTrack,”Proceedingsofthe3rdInternationalConferenceonTransportationInfrastructure,Pisa,Italy,2014.

35. Taylor,A. J. “Mechanistic CharacterizationofResilientModuli forUnboundPavementLayerMaterials,”M.S.Thesis,AuburnUniversity,2008.

Date post:	11-Jul-2018
Category:	Documents
Upload:	vominh
View:	247 times
Download:	0 times

NCAT Report 15-05R REFINED LIMITING STRAIN...

Documents