evaluation of laboratory procedures for compacting asphalt

SHRP-A/UWP-91-523

Evaluation of LaboratoryProcedures for Compacting

Asphalt-Aggregate Mixtures

Prepared byJ.B. Sous_ J. Harvey, L. Painter, J.A. Deacon, and C.L. Monismith

Institute of Transportation StudiesUniversity of California

Berkeley, C_llfornia

Strategic Highway Research ProgramNational Research Council

Washington, D.C. 1991

SHRP-A/UWP-91-523Contract A-003AProgram Manager:.Edward T. HarriganProject Man:_ger:Rita B. LeahyProgram Secretary:Juliet Nar_:h

September 1991

key words:air voidsasphalt mixturesbeam fatiguecompressive creep modulusdiametral fatiguedynamic modulusfield compactiongyratorykneadinglaboratory compactionmix designrolling wheelshear creep modulus

Strategic Highway Research Program2101 Constitution Avenue, N.W.Washington, D.C. 20418

(202) 334-3774

This report represents the views of the authors only, and is not necessarily reflective of the views of theNational Research Council, the views of SHRP, or SHRP's sponsor. The results reported here arc notnecessarily in agreement with the results of other SHRP research activities. They are reported to stimulatereview and discussion with in the research commnnity.

ABSTRACT

The scope of Strategic Highway Research Program (SHRP) Project A-003A,

"PerformanceRelated Testing and Measuringof Asphalt-AggregateInteractions and Mixtures,"

includes a comprehensive examination of laboratorytests for asphalt-aggregatemixtures. For

test results to be meaningful, however, specimens prepared in the laboratory must resemble as

closely as possible in-service mixtures, those producedby mixing, placement, and compaction

in the field and subsequently "conditioned-."by traffic loads and "aged" by environmental

influences. Considerable attentionis thus being given both to methods of laboratorycompaction

and to methods of conditioning by acceleratedexposureto water, air, elevated temperatures, etc.

This report documents results of the laboratory compaction phase of the SHRP study.

The compaction investigation focused on the extent to which method of laboratory

compaction (Texas gyratory, kneading, and rollingwheel) affects fundamentalmixtureproperties

of importance to pavement performance in service. Permanent deformation and fatigue were

selected as the performance features of greatest concern, and laboratory creep and fatigue tests

were among those used to measure mixtureproperties that were likely to correlate well with

these performance features. A total of 16 asphalt-aggregatemixtures--varyingwidely in asphalt

source, aggregate type, asphalt content, and air-void content--were tested.

Compaction method was found to significantly affect properties of the dense-graded

mixtures evaluated herein, both in the statistical sense and in terms of practical engineering

consequence. The likelihood is reasonably high that differentkinds of mixtures might evolve

from a comprehensive mixture analysis and design system depending on method of compaction.

Among the three compaction methods examined, kneading specimens are generally most

sensitive to aggregate characteristics and least sensitive to asphalt characteristics. Gyratory

specimens are least sensitive to aggregate characteristics and only slightly more sensitive than

i

rolling-wheel specimens to asphalt characteristics. In general, kneading compaction produced

specimens having greatest resistance to permanent deformation while gyratory compaction

produced those having greatest resistance to fatigue cracking under controlled-stress loading.

For both permanent-deformation and fatigue resistances, roning-wheel specimens were ranked

between gyratory and kneading specimens.

Unlike static and impact compactors, the three compaction methods investigated herein

subject the densifying mixture to shearing motions similar to those induced during field

compaction. During compaction, the three compactors likely differ most in the "effective"

stressI applied to the aggregate particles and in the magnitude and randomness of interparticle

displacements. A greater effective stress promotes more interparticlecontact, and larger and

more random displacements promote a more tightly interlocked aggregate structure. Under

gyratory compaction, the effective stress can be relatively low due to pore pressures building

up in the fluid (asphalt and air voids) phase, and the interparticle shearing motion is relatively

small and uniform: the net result is a relatively "weak" aggregate structure. Kneading

compaction probablydevelops the strongestaggregatestructure. Differences amongcompaction

methods are greatest for mixtures with low air voids, presumably because pore pressures are

larger when there are fewer air voids in the mixture. Notwithstanding these generalities, it also

appears that, depending on their design and operation, compactors within a given genre may

produce specimens having quite differentengineeringproperties.

As a result of this study, the rolling-wheel compactor has been selected for use in

subsequentphases of SHRP ProjectA-003A. It is intuitively appealing for its apparent similarity

to field compaction, and it generally produces specimens whose properties lie within or near the

IDuring compaction, the compaction pressure is resisted by the pore pressure in the fluid(asphalt and air) phase and the effective stress transmitted through the aggregate particles.

ii

range of those produced by gyratory and kneadingcompaction. Rolling-wheel compaction is a

comparatively easy procedure and enables rapidfabricationof the large numberof specimens

required in subsequent A-003A testing. Because specimens produced by rolling-wheel

compaction are cored or .sawed from a larger mass, all surfaces are cut. Cut surfaces are

desirable because test results are likely to be less variable, air voids can be more accurately

measured, comparisonswith specimens extractedfrom in-service pavementsare morevalid, and

specimens are more homogenous.

The measurement of air voids is also a critical element of laboratory testing of

asphalt-aggregate mixtures. After considerableexperimentation, air-void estimates were most

uniform and consistent when measurementswere made with dry specimens encased in parafilm

(an impervious, stretchable fdm) for immersed weighing. Specimens exposed to water as a

result of coring and/or sawing can be sufficiently dried by blowing compressed air over their

surfaces. Air-void measurements in subsequent phases of the study will be based on this

technique.

.o.

111

ACKNOWLEDGEMENTS

The work reported herein has been conducted as a part of Project A-003A of the Strategic

Highway Research Program (SHRP). SHRP is a unit of the National Research Council that was

authorized by Section 128 of the Surface Transportation and Uniform Relocation Assistance Act

of 1987. This project is entitled, "Performance Related Testing and Measuring of Asphalt-

Aggregate Interactions and Mixtures," and is being conducted by the Institute of Transportation

Studies, University of California, Berkeley, with Carl L. Monismith as principal investigator.

The support and encouragement of Dr. Ian Jamieson, SHRP Contract Manager, is gratefully

acknowledged.

The draft of this report was reviewed by an Expert Task Group (ETG) who provided

many valuable comments and will continue to provide guidance throughout the contract. The

members are:

]_'aest G. ]Lgstian Michael L. Fiin_

Federal Highway Administration. ELF Asphalt

Campbell Crawford Charles S. HughesNational Asphalt Paving Association Virginia Highway and Transportation Research

Council

WilliamDearasaugh D*nK N. LittleTransportationResearchBoard TexasA&MUniversity

FrancisFee KevinStuart1_L1_Asphalt FederalESgbwsyAdministration

Douglas L Hanson Roger L. YarbroughNew Mexico State Highway Department University Asphalt Company

Eric E. HarmIllinois Department of Transportation

The contributions of Dr. Akhtar Tayebali, Elie Abi-Jaoude, Kashyapa Yapa, Thomas

Mills, Matthew McCune, Alexandro Tanco, and Sirous Alavi, who assisted in various phases

of the work, are also gratefully acknowledged.

iv

DXSCLA_

The contents of this report reflect the views of the authors, who are solely responsible

for the facts and accuracy of the datapresented. The contents do not necessarily reflect the

official view or policies of the Strategic Highway Research Program (SHR.P) or SHRP's

sponsors. The results reportedhere are not necessarily in agreementwith the results of other

SHRP research activities. They are reported to stimulate review and discussion within the

research community, This reportdoes not constitute a standard,specification, or regulation.

V

TABLE OF CONTENTS

ABSTRACT ......................................... i

ACKNOWLEDGEMENTS ............................... iv

DISCLAIMER ........................................ v

TABLE OF CONTENTS ................................ vi

LIST OF FIGURES ................................... viii

LIST OF TABLES .................................... xiv

1.0 INTRODUCTION .................................... l

2.0 SIGNIFICANT _TEMENTS IN MAIN EXPERIMENT ............. 4

2.1 Mixture and Test Variables ........................... 42.2 Specimen Fabrication .............................. 92.3 Test Methods ................................... l02.4 Mixture Properties ................................ 13

3.0 DESIGN OF MAIN EXPERIMENT ......................... 20

3.! Experiment Designs . .............................. 203.2 Analysis Techniques ............................... 25

4.0 ANALYSIS AND FINDINGS ............................. 28

4.1 Main Experiment ................................. 284.2 ExtendedPermanent-DeformationStudy ................... 974.3 ComplexModulusStudy ............................ 1124.4 Comparisonswith OtherCompactionMethods ............... 1274.5 Comparisonswith Field Cores ......................... 1314.6 Comparisonswith OtherInvestigations.................... 139

5.0 AIR-VOID MEAS_ AND CUT-SURFACE EFFECTS ........ 145

5.1 Air-Void Measurement.............................. 1475.2 Effectof Cut Surfaces.............................. 1585 3 Summary 165

6.0 SUMMARY AND RECOMMENDATIONS ..................... 169

6.1 Findings ...................................... 1696.2 Recommendations ................................. 173

vi

7.0 REFERENCES ...................................... 176

APPENDICES

A Specimen Preparation .............................. A. 1B Diametral Fatigue Test Procedure ....................... B. 1C Statistical Analysis Techniques for Main Experiment ........... C. 1D Proposed Method for Measurement of Air Voids in Test

Specimens of BituminousMixtures ...................... D. 1E Proposed Method for Preparationof Test Specimens of

Bituminous Mixtures by Means of Rolling-Wheel Compaction ...... E. 1

vii

LIST OF.FIGURES

2.1 Typical Creep Curves ................................. 15

2.2 Constancy of Collapse Strainfor a Number of Randomly Selected TestsUnder Compressive Creep .............................. 17

2.3 Constancy of Collapse Strain for a Number of Randomly Selected TestsUnder Shear Creep ................................... 18

4.1 Effect of Compaction on Compressive Creep Modulus ............. 40

4.2 Effect of Compaction on Shear Creep Modulus .................. 41

4.3 Combined Effects of Compaction and Asphalt Type on Compressive CreepModulus, AAK-1 [B] and AAG-1 IV] ....................... 43

4.4 Combined Effects of Compactionand Asphalt Type on Shear CreepModulus, AAK-1 []3] and AAG-1 IV] ....................... 44

4.5 CombinedEffectsofCompactionandAsphaltContenton CompressiveCreepModulus,OptimumAsphalt[0]andHighAsphalt[I].......... 46

4.6 CombinedEffectsofCompactionandAsphaltContentonShearCreepModulus,OptimumAsphalt[0]andHighAsphalt[l].............. 47

4.7 CombinedEffectsofCompactionandAggregateTypeonCompressiveCreepModulus,RL Chert[T]andRB Granite[W] ............... 48

4.8 CombinedEffectsofCompactionandAggregateTypeonShearCreepModulus,RL ChertIT]andRB Granite[W'J.......... ......... 50

4.9 CombinedEffectsofCompactionandAir-VoidContentonCompressiveCreepModulus,Low Voids[0]andHighVoids[I]............... 51

4.I0 CombinedEffectsofCompactionandAir-VoidContentonShearCreepModulus,Low Voids[0]andHighVoids[I]................... 52

4.11 Combined Effects of Compaction and Temperatureon Compressive CreepModulus, 104°F (40°C) [0] and 140°F (60°C) [I] ................ 53

4.12 Combined Effects of Compaction and Temperature on Shear CreepModulus, 104°F (40°C) [0] and 140°F (60°C) [1] ................ 54

4.13 Combined Effects of Compaction and Stress Level on Compressive CreepModulus, 14.5 psi [0] and 29.0 psi [1] ....................... 55

°°°

VIII

4.14 Combined Effects of Compaction and Stress Level on Shear CreepModulus, 2.4 psi [0] and 4.8 psi [1] ........................ 56

4.15 Combined Effects of Compaction and Mixture Stiffness on FlexuralFatigue at Low Stress, 100 psi ............................ 60

4.16 Combined Effects of Compaction and Mixture Stiffness on FlexuralFatigue at High Stress, 175 psi ............................ 61

4.17 Combined Effects of Compaction and MixtureStiffness on DiametralFatigue at Low Stress ................................. 64

4.18 Combined Effects of Compaction and Mixture Stiffness on DiametralFatigue at High Stress ................................. 65

4.19 Effect of Heating and Mixing on Hardeningof AAK-1 Asphalt ........ 76

4.20 Effect of Heating and Mixing on Hardening of AAG-1 Asphalt ........ 77

4.21 Effect of Compaction Method on Permeability .................. 81

4.22 Compressive Creep Curves for Specimens with Varying Air Voids,Gyratory Compaction ................................. 98

4.23 Compressive Creep Curves for Specimens with Varying Air Voids,Kneading Compaction ................................. 99

4.24 Shear Creep Curves for Specimens with Varying Air Voids,Gyratory Compaction ................................. 100

4.25 Shear Creep Curves for Specimens with Varying Air Voids,Kneading Compaction ................................. 101

4.26 Effect of Air Voids on Compressive Creep Modulus,Gyratory Compaction ................................. 103

4.27 Effect of Air Voids on Compressive Creep Modulus,Kneading Compaction ................................. 104

4.28 Combined Effects of Air Voids and Loading Time on Compressive CreepModulus, Kneading Compaction ........................... 105

4.29 Effect of Compaction Method on Compressive Creep Modulus of LowAir-Void Mixtures, Average of Four Specimens for Each Line ........ 106

4.30 Effect of Compaction Method on ShearCreep Modulus of LowAir-Void Mixtures, Average of Four Specimens for Each Line ........ 107

ix

4.31 Effect of Compaction Method on Compressive Creep Modulus ofHigh Air-Void Mixtures, Average of Five Specimens for Each Line ..... 108

4.32 Effect of Compaction Method on ShearCreep Modulus of HighAir-Void Mixtures, Average of Five Specimens for Each Line ........ 109

4.33 Effect of Compaction Method on Strain Ratio in Shear Creep,Average of Four Specimens for Each Line .................... 110

4.34 Effect of Compaction Method on Specimen Dilation in Shear Creep,Average of Four Specimens for Each Line .................... 111

4.35 Effect of Compaction Method on PermanentStrain in Cyclic Shear ..... 113

4.36 EffectofCompactionMethodonSpecimen"DilationinCyclicShear,Average of Two Specimens for Each Line ..................... ! 14

4.37 Effect of Compaction Method on Permanent Deformation Modulusin Cyclic Shear ........... .......................... 115

4.38 Effect of Compaction Method on Permanent Deformation Modulusin Cyclic Compression ................................. 116

4.39 Effect of Compaction Method on Dynamic Modulus at 68°F (20°C),Average of Seven Specimens for Each Line .................... 118

4.40 EffectofCompactionMethodonDynamicModulusat104°F (40°C),AverageofSixSpecimensforGyratoryandSevenSpecimensforKneading................................ 119

4.41 Effect of Compaction Method on Loss Tangent at 68°F (20°C),Average of Seven Specimens for Each Line .................... 121

4.42 Effect of Compaction Method on Loss Tangent at 104°F (40"C),Average of Six Specimens for Gyratory and Seven Specimens forKneading ......................................... 122

4.43 Combined Effects of Air-Void Contentand Compaction Method onDynamic Modulus at 104°F (40°C), Average of Three Specimens forEach Line ........................................ 123

4.44 Combined Effects of Air-Void Content and Compaction Method onDynamic Modulus at 680F (20°C), Average of Three Specimens forEach Line ........................................ 124

4.45 Combined Effects of Air-Void Content and Compaction Method onLoss Tangent at 1040F (400C), Average of Three Specimens forEach Line ........................................ 125

X

4.46 Combined Effects of Air-Void Contentand Compaction Method onLoss Tangent at 68°F (20°C), Average of Three Specimens forEach Line ........................................ 126

4.47 Comparative Response of Corpsof Engineers' Gyratory Specimensto Shear Creep Loading, Average of Two Specimens for Each Line ..... 129

4.48 ComparativeResponse of Exxon Rolling-Wheel Specimens to ShearCreep Loading, Average of Two Specimens for Each Line .......... 130

4.49 ComparativeResponse of I 80 Field Cores to Shear Creep Loading,Average of Five Specimens for Each Line ..................... 132

4.50 ComparativeDilational Responseof I 80 Field Cores to ShearCreep Loading, Average of Five Specimens for Each Line. .......... 134

4.51 Viscosities of Asphalts ExtractedFrom I 80 Field Cores andLaboratorySpecimens, 140°F (60°C) ....................... 135

4.52 Viscosities of Asphalts ExtractedFrom I 80 Field Cores andLaboratory Specimens, 275°F (135°C) ....................... 136

4.53 Comparative Response of US 101 Field Cores to Shear Creep Loading,Average of Four Specimens for Each Line .................... 137

4.54 Comparative Dilational Response of US 101 Field Cores to ShearCreep Loading ...... . ............................... 138

4.55 Comparison Between Laboratory Specimensand Field Cores forMixtures with Carbon Black (After Monismith and Tayebali, 1988) ..... 140

4.56 Comparison Between Laboratory Specimens and Field Cores forControl Mixtures (After Monismith and Tayebali, 1988) ............ 141

4.57 Comparison of Slope of Creep Curves for Gyratory (MT/GS) andKneading (CK/CC) Specimens (March28, 1989 Letter to J. Moulthropfrom H. Von Quintus) ................................. 143

5.1 Effect of ParaTdmon Surface-Dry Measurementsof Air Voids,All Specimens ...................................... 150

5.2 Combined Effect of Parafilm and Wetting on Air-Void Measurements .... 151

5.3 Effect of Wetting on No-ParafilmMeasurements of Air Voids ........ 152

5.4 Comparison of DNP, WNP, and WWP Measurementsof Air Voids,Regression Lines .................................... 153

xi

5.5 Combined Effect of Cut Surfaces and Parafilmon Air-VoidMeasurements, Regression Lines .......................... 154

5.6 Combined Effect of Wetting and AggregateType on Air-VoidMeasurements, Regression Lines .......................... 156

5.7 Combined Effect of ParaYxlmand Aggregate Type on Air-VoidMeasurements, Regression Lines .......................... 157

5.8 Effect of Cut Surfaces on Wet-With-Paraf'dmMeasurements ofAir Voids in Large Beams, KneadingCompaction ................ 162

5.9 Effect of Cut Surfaces on Wet-No-ParafilmMeasurements of Air Voidsin Large Beams, Kneading Compaction ...................... 163

5.10 Effect of Cut Surfaces on No-ParafilmMeasurements of Air Voidsin Large Cylinders, Gyratory Compaction ..................... 164

5.11 Illustration of BoundaryEffects of Cut Specimen Surfaces in Shearand DiametraJTesting ................................. 166

5.12 Effect of Cut Surfaces on Shear Creep Modulus, GyratorySpecimens .... 167

A.1 Form No. 1 ....................................... A.11

A.2 Form No. 2 ....................................... A.12

A.3 Form No. 3 ....................................... A. 13

A.4 BTDC Showing Mixing TemperatureSelection .................. A. 19

A.5 Form No. 4 ....................................... A.24

A.6 Example Calculation for Compaction Using Large Texas GyratoryCompactor ........................................ A.29

A.7 Mold Dimensions for Rolling Wheel Compaction ................ A.35

B. 1 Loading of Cylindrical Specimen in the Diametral Fatigue Test ........ B.3

B.2 Stress Distributions Within Diametrally Loaded Specimens(After Hadley, Hudson, and Kennedy, 1970) ................... B.5

B.3 Test Specimen with Diametral Yoke and Loading Ram(After Vinson, 1989) .................................. B.8

E. 1 Example Calculation for Mass of Mixture to Be Compacted .......... E.3

xii

E.2 Rolling-Wheel Compaction Apparatus ....................... E.5

E.3 University of CaliforniaOne-LiftRolling-Wheel Compaction Mold ..... E.7

OoO

XUl

LIST OF TABLES

1.1 Compaction Experiments ............................... 3

2.1 Significant Mixture and Test Variables for Main Experiment ......... 6

2.2 Aggregate Gradationfor Compaction Study .................... 7

2.3 Mixture Designs .................................... 8

3.1 Experiment Design for Diametr_ Fatigue Tests, Main Experiment ...... 21

3.2 Experiment Design for Flexural Fatigue Tests, Main Experiment ....... 23

3.3 Experiment Design for Shearand Compressive-Creep-Tests; MainExperiment ........................................ 24

3.4 Sample Size Required for Main Experiment .................... 26

4.1 Summary Results of Unconfined Axial Compressive Creep Tests ....... 32

4.2 Summary Results of Unconfined Shear Creep Tests ............... 33

4.3 Effect of Compaction Method on Resistance to PermanentDeformation... 34

4.4 Effect of Compaction Method on Sensitivity ofPermanent-Deformation Response to MixtureVariables ............. 37

4.5 EffectofMixtureandTestVariablesonResistancetoPermanentDeformation....................................... 38

4.6 Summary ResultsofFlexuralFatigueTests... ................. 58

4.7 Summary ResultsofDiametralFatigueTests................... 62

4.8 Effect of Compaction Method on Resistance to Fatigue ............. 66

4.9 Effect of Compaction Method on Sensitivity of Fatigue Responseto Mixture Variables .................................. 68

4.10 Effect of Mixture and Test Variableson Resistance to Fatigue ........ 69

4.11 Effect of Compaction Method on Stiffness ..................... 71

4.12 Effect of Compaction Method on Sensitivity of Stiffness Responseto Mixture Variables .................................. 72

xiv

4.13 Effect of Mixture and Test Variables on Stiffness Modulus .......... 73

4.14 Effect of Compaction Method on Hardeningof Asphalts ............ 75

4.15 Experimental Validation of Hypotheses Regarding CompactionMethodologies ...................................... 79

4.16 Statistically Significant Effects in Compressive Creep Testing ......... 85

4.17 Summary Statistics for Main Experiment ...................... 86

4.18 Statistically Significant Effects in Shear Creep Testing ............. 88

4.19 Statistically Significant Effects in Flexural Fatigue Testing ........... 91

4.20 Statistically Significant Effects in Diametral Fatigue Testing .......... 94

5.1 Sample Size for Study of Air-Void Measurement................. 146

A. 1 Medium Gradation No. 1 ............................... A.6

A.2 Medium GradationNo. 2 ............................... A.7

A.3 Coarse Gradation .................................... A.8

A.4 Minimum Masses Required for Analysis Specimens(from ASTM C 117-80) . ............................... A.10

A.5 Sieving Regimes for Chert (R.L)Aggregate .................... A.16

A.6 Mixing Temperatures ................................. A.21

A.7 Percentages of Asphalt To Be Used in Mixing .................. A.23

A.8 Ratio of Blows on Lift to be Compactedto Blows on Lift JustCompacted ........................................ A.33

XV

1.0 INTRODUCTION

DThe scope of Strategic Highway Research Program (SHRP) Project A-003A,

"PerformanceRelatedTesting andMeasuringof Asphalt-AggregateInteractionsand Mixtures,"

includes a comprehensive examination of laboratory tests for asphalt-aggregate mixtures, tests

which measure fundamentalmixturepropertiesthat reflect significant influences of asphalts and

asphalt-aggregate interactions on pavement performance. For test results to be meaningful,

specimens preparedin the laboratory must resemble as closely as possible in-service mixtures,

those produced by mixing, placement, and compaction in the field and subsequently

"conditioned" by traffic loads and "aged" by environmentalinfluences. Considerable attention

is thus being given both to methodsof laboratory compaction and to methods of conditioning by

accelerated exposure to water, air, elevated temperatures, etc. It is the purpose of this report

to document results of the laboratorycompaction phase of the overall study.

A recent National Cooperative Highway Research Project study (Von Quintus et al.,

1988) has identified promising laboratory compaction techniques. Named after the asphalt-

aggregate mixtureanalysis system it is developing, the AAMAS study has effectively eliminated

from contention both impact compaction and the Arizona vibratory/kneading compactor. It has

successfully narrowed the search to three leading contenders; gyratory, kneading, and

rolling-wheel compaction. These three methods form the focus of the current investigation.

The main experiment was designed primarily to ascertain the extent to which method of

compaction (Texas gyratory,kneading,androlling wheel) affects fundamentalmixtureproperties

of importance to pavement performance in service. Permanent deformation and fatigue were

selected as the performance parametersto be considered, and laboratory creep and fatigue tests

were used to measure mixture properties that were likely to correlate well with field

performance. The main experiment was structuredto accomplish the following:

I

1. To identify any effects of_compaction method on permanent-deformationand

fatigue properties and on the sensitivities of these properties to mixture

composition and

2. If effects are observed, to determine their statistical significance.

Auxiliary experiments (1) extended the rutting investigation to include measures of

permanent deformation underrepetitive loading and to examine more broadly the interaction of

airvoids and compaction method, (2) addeddynamic modulus and phase angle (loss tangent) as

mixture properties of_possibly significant relationship to pavement performance, (3) included

limited comparisons with other laboratory compaction methods, and (4) included limited

comparisons between laboratory and field compaction. The extent of the experimentalprogram

is indicated by Table 1.1.

Although the primary purpose of the compaction study was to identify a compaction

procedure(s) most appropriate for laboratory use, the study also provided opportunityto identify

minor improvements to the preferred compactionmethod(s) and to reevaluate the range of testing

and mixture properties to be evaluated in subsequentlaboratory studies.

2

Table I.I Compaction Experiments

l:xperiment Laboratory Tests Compaction Methods Mixture Types No. ofTestsi | eH

Flexure/Fatigue Gyratory 2 Asphalts

Main Experiment Diametral Fatigue Kneading 2 Asphalt Contents 480Compressive Creep Rolling Wheel 2 Air VoidsShear Creep 2 Aggregates

|

Compressive Creep Gyratory I Asphalt

Extended Permanent Shear Creep Kneading 1 Asphalt Content 50Deformation Study Compressive Cyclic Multiple Air Voids

Shear Cyclic 1 Aggregate

Dynamic Axial Gyratory 1 Asphalt

Complex Modulus Compression Kneading 1 AsphaltContent 376Study Multiple Air Voids

1 Aggregate

Shear Creep GypJtory (rexas) 1 Asphalt

Other Compaction Oyratory (Corps) I Asphalt Content 12Comparisons Kneading 1 Aggregate

Rolling Wheel ('Exxon)e

Shear Creep Gyratory 1 Field Mixture1 80 Kneading 15Field

Core Field

Tests Shear Creep Gyratory I Field MixtureUS 101 Kneading 12

Field

2.0 SIGNIFICANT ELEMENTS IN MAIN EXPERIMENT

The objective of the main experiment was to determine the effect of compaction

methodology on fundamental engineering properties of asphalt-aggregate mixtures. The

candidatecompaction methods includedTexas gyratorycompaction (adapted from.TexasMethod

Tex-206-F and ASTM D 4013), kneading compaction (ASTM D 1561), and rolling-wheel

compaction (a procedure not yet standardized). Because the compaction study preceded other

elements of the overzU SHRP A-003A investigation, final choices had yet to be made regarding

the most critical mixture properties influencingpavement performance. Resistances to cracking

and to permanent deformation under repetitive loading were considered to be the critical

pavement performance parameters, however, and state-of-the-art surveys confirmed that tests

commonly used at the University of California would adequately measure related mixture

properties (Sousa, Claus, and Monismith, 1990 and Tangella et al., 1990). These surveys also

helped to highlight the mixturevariables most likely to affect the resistanceof asphalt-aggregate

mixtures to cracking and to permanent deformation and to identify test conditions sufficiently

representative of a range in field conditions to produce reliable results. The purpose of this

section is to identify and describe these significantelements of the maincompaction experiment.

2.1 Mixture and Test Variables

In order to limit the experiment to manageable size, only those mixture variables

expected to significantly affect the primary response measures (resistance to permanent

deformation and fatigue) were selected for investigation. These included asphalt type, asphalt

content, aggregate type, and air-voidcontent. Significant test variablesincluded both stress level

and temperature. A factorial design was used for the main experiment, and a two-level

4

representationwas used for each of these six primaryvariables (or factors). Specific mixture

and test variables, summarized in the matrixof Table 2.1, are more fully described as follows:

1. Aggggg_. The two aggregatesselected for evaluationincludedgranite (R.B)and

chert (RL). The granite (RB), a non-stripper, is an angular aggregate with a

rough surface texture. The chert (P.L), considered to be a stripl_r, is a gravel

with more rounded particles and smoother surface texture. One gradation was

selected, typical of dense graded aggregates having a 3/4-inch (19.0 ram)

maximum size (Table 2.2).

2. A_P.]!_. An AC-30 asphalt (AAK-1) and an AR-4000 asphalt (AAG-I) were

selected for investigation because of their vastly different compositional and

temperature-susceptibility characteristics. Additionally, two levels of asphalt

content were selected. For each asphalt-aggregate mixture, the lower asphalt

content was determinedusing a modificationof standardHveem (ASTM D 1560)

procedures (Table 2.3). _The upper asphalt content was set at a 0.5- or 0.7-

percent higher level (by weight of mixture), only slightly exceeding the optimum

asphalt content determinedby the U.S. Army Corps of Engineers (Marshall)

75-blow procedure (ASTM D 1559) for the granite (RB) and somewhat less for

the chert (RL) (Table 2.3).

3. Air Voids. Each of the three methods of compaction, kneading, gyratory, and

rolling wheel, were evaluatedat two levels of compactiveeffort. The upperlevel

of compaction was thatnecessary to produce an air-voidcontent initially targeted

at approximately 4 percent. The lower level was that necessary to produce a

targeted air-void content of approximately 8 percent. As the compaction study

progressed, however, it becameapparentthat air-void measurementswere being

5

Table 2.1 Significant Mixture and Test Variables for Main Experiment

Variable Level of Treatment NumberI of Levels

1 I 2 3

Aggregate

• StrippingPotential Low High (2)

• Gradation Medium (1)

Asphalt

• Temperature Susceptibility Low High (2)

• Grade Medium (1)

• Content " Optimum High (2)

Compaction

• Method Gyratory Kneading Rolling (3)V,qleel

• Air Voids 4.5+% 11.5+__% (2)

Test Conditions

• Temperature

- Fatigue 32/39.2°F 68°F (2)........ (0/4.o.C), (20°C)

- Rutting 104°F 140°F (2)(400C) (60°C)

• Stress Level

- Fatigue Low High (2)

- Rutting Low High (2)

Conditioning

• Aging None (1)

• Moisture None (1)

26.31

6

Table 2.2 Aggregate Gradation for Compaction Study

Sieve Size Percent Passing by Mass

1 in lO0

3/4 in 95

112 in 80

318in 68

No. 4 48

No. 8 35

No. 16 25

No. 30 17i

No. 50 12

No. 100 8

No. 200 4

Pan

Note: Iin- 25.4mm

7

Table 2.3 Mixture Designs

Percent Asphalt by Weight of Mixture

U.S. Army Compaction StudyAggregate Asphalt State of Corps of

California Engineers

Method (Marshall) Low Level High LevelMethOdb

AAK-I 4.9a 5.3 4.9 5.4Granite0_)

AAG-1 4.7 a 5.1 4.7 5.2

AAK-I 4.1 © $.2+ d 4.1 4.8Chert (P_)

AAG-I 3.9 c 5.2+ d 3.9 4.6

aBased on stabilometer "S" value of 35.

b75-blow compaction.

CBa_ on stabilometer "S" value of about 25 and air voids of about 4 percent.

dBased on Asphalt Institute criteria for compaction with 75-blow procedure.

affected by surface condition(eitheras moldedor as cut by coring or sawing) and

by prior exposure to water. To reduce measurement variability, a modified

procedure, described in Section 5.0 as "wet-with-parafilm," was ultimately

adopted. Using this technique, the low and high air-void-content specimens

averaged approximately4.5- and 11.5-percentair voids, respectively. Thisrange

in air voids is generally compatiblewith the range measured by Von Quintuset

al. (1988) in field cores from newly constructedprojects in five states.

4. Test Conditions. The two temperatures selected for each test procedure are

considered representative of the range of critical in-service conditions. For the

fatigue tests, they included 32° or 39.2°F (0° or 4°C) and 680F (20°C) and for

the rutting tests, 1040F and 140°F (4OoCand60°C). Two levels of loading (low

and high) were used for each test method. Selection of load levels was based

both on the time requiredto complete laboratorytestingand on the levels of load-

induced stressesanticipated in pavement structures.

5. .C.,.¢.ggIi._7.I3L_.Although one of the main objectives of SHRP Project A-003A is

to develop appropriateprocedures for conditioning laboratory specimens, this

aspect of the project was not sufficiently advanced in time to affect the

compaction study. Accordingly, only unconditioned specimens were tested.

2.2 Specimen Fabrication

Immediately prior to mixing, both asphalt and aggregate were heated to a mixing

temperature of 275 + 5°F (135 + 2.8°C). They were mixed as quickly as possible to obtain

uniformity, and the mixture was placed in a 14OOF(60°C) oven for 15 hours to cure. Before

compaction, the mixture was placed in a 240°F (1160C) oven for 1.5 hours. Compaction

9

procedures are described-along with other specimen preparation procedures as they evolved

from the compaction study--in Appendix A.

2.3

Two repetitive-loading tests were chosen for measuring response of asphalt-aggregate

mixtures to fatigue cracking, flexural loading and diametral loading. Flexural loading was

selected because of the considerable experience accumulatedover many years by the University

of California using such loading to measure fatigue response of beam specimens. Flexurai

testing yields a mixture property, applications to failure; that has been successfully correlated

with pavement performance (Monismith et al., 1971). Unfortunately, gyratory compaction does

not produce specimens of a size suitable for use with available flexural fatigue equipment.

Because fatigue tests can be conducted on gyratory specimens in the diametral mode and because

of current interest in diametral fatigue testing, diametraltesting was selected to provide a second

measure of fatigue response.• . .,. , , . ...., ......... . .. .., - . . ..

Resistance to permanent deformation was measured by the application of static creep

loads to cylindrical specimens. Although alternative methods hold considerable promise, their

use in the compaction study was considered to be premature. Moreover, the fact that mixture

response to creep loads in axial compression has been correlated with pavement rutting lends

creditability to the use of compressive creep in this comparison of compaction methods (Hills,

Brien, and van de Loo, 1974). On the premise that pavement rutting results principally from

shearing stresses, shear creep was also used to evaluate the resistance of test mixtures to

permanent deformation.

2.3.1 tF..g!ig_. All fatigue tests were conducted at two temperatures, representative of the

range within which pavement cracking under traffic loading is most likely. Originally, both

10

flexural and diametral tests were planned for 32°F (0°C) and 68°F (20°C). Because of

equipmentlimitations in conducting the diametralfatigue tests at 32OF(0°C), however, the low-

temperaturediametraltests were conducted at 39.2°F (4oc). The following briefly describes

the fatigue test procedures:

1. Flexural Fatigue. Because the gyratory compactor does not produce beam

specimens of the necessary size, specimens for flexural fatigue testing were

prepared using only kneading (ASTM D 3202) and rolling-wheel compaction

(Appendix A). After compaction, they were sawed to the required size of 1.5 x

1.5 x 15 inches 08.1 x 38.1 x 381 mm), and specific gravities and air voids were

measured. Beams were testedusing one-third point, pulse loading, at a frequency

of 1.67 Hz and a duration of 0.1 second. Bending stresses, selected to yield

fatigue lives of approximately 10,000 and 100,000 applications to failure, were

set at 115 and 225 psi (792 and 1,550 kPa) for the low-temperature testing and

80 and 130 psi (551 and 896 k1'a)for the high-temperature testing. Controned-

stress loading was used.

2. Diametral Fatigue. Specimens were prepared using aU three methods, gyratory,

kneading, and roUing-wheel compaction. Testing procedures are described in

Appendix B. Load frequency was 1 I-Iz and load duration as well as mode of

loading were similar to those employed in the flexural testing. Unlike flexural

testing, however, permanent deformation is allowed to accumulate in the

diametral specimens, and stress reversal is not provided. Stress levels were

individually chosen (for each mixture, compaction type, and temperature level)

to produce failure in the ranges of 1,000 to 5,000 applications (high stress level)

and 70,000 to 100,000 applications 0ow stress level).

11

2.3.2 Permanent-Deformation Tests .... For.the permanent-deformation tests, two test

temperatures, 104°F and 140°F (40"C and 60°C), were used. In addition, two compressive

stress levels (without confinement) were employed for the compressive creep testing and two

shear stress levels, for the shearcreep testing. Briefly, the test procedures are describedbelow:

1. Compressive Creep. Specimens 4 inches (101.6 mm) in diameter and 8 inches

(203.2 mm) in height were fabricatedusing gyratory, kneading, and rolling-wheel

methods. Kneading specimens were molded to the required dimensions.

Gyratory and'rotling-wbeel specimens were cored from larger compactedmasses:

ends were sawed as necessary to produce the required height. For both

temperatures, tests were performed at two identical levels of compressive stress,

targeted at 15 and 30 psi (103.4 and 206.7 kPa) but actually averaging 14.5 and

29.0 psi (99.9 and 199.8 kPa), respectively. Loading in these creep tests was

static: it was maintaineduntilthe specimen collapsed or until a maximumof one

hour had elapsed.......................................

2. Shear Creep. The shear creep test was conducted on specimens 4 inches (101.6

mm) in diameter and approximately 2.5 inches (63.5 mm) high. It consists of

applying a step load in a directionparallel to the two plane faces of the specimen.

The relative displacement of the two faces is recorded and a shearcreep modulus

is obtained by dividing the stress applied to the specimen (ratio of shear load to

cross sectional area) by the measured shear strain (ratio of relative displacement

of the faces to the height of the specimen). Shear is applied in the test rig

through loading plates bonded with hydrostone to the parallel faces of the

specimen. When failure is induced, the failure plane is typically inclined at an

angle of about 45". Two identical shear stress levels were applied at both

12

temperatures, averaging 2.4 and 4.8 psi (16.5 and 33.1 kPa) for the low and high

stress levels, respectively. A normal(axial) stress of 2.5 psi (17.2 Ha) was also

applied during the sheartesting. Gyratory and kneading specimens were molded

to the required dimensions. Coring and sawing were used in fabricating the

rolling-wheel specimens. Loading in these creep tests was static: it was

maintained until the specimen collapsed or until a maximum of one hour had

elapsed.

2.4 Mh'ture Pro_eriies

Among the measures that havebeen used as quantitativeexpressionsof mixtureproperties

in controlled-stress fatigue testing are fatigue life (the number of load applications sustained to

"failure")at some fixed stress level; the stress necessary to cause failure at some fixed number

of applications; and the parameters,K1 and K2, of one of the more common fatigue models1.

Each of these measures was considered for use in interpretingresults of the compaction study.

Fatigue life, the clear favorite because of the ease with which it is understood, proved to be a

logically consistent measureof response to fatigue cracking and was selected for use herein. A

wide range in fatigue lives were measured,and logarithmic transformationswere made for most

of the analyses.

According to the original experimental plan for the fatigue testing (Hicks et al., 1990),

the same two levels of stress were to be applied at each of two temperature levels, 32°F and

68°F (0°C and 20°C). Finding stress levels appropriateto the range in test mixtures proved to

IThebasic relationshipbetween fatigue life, N, and applied flexural stress, ¢, is commonlyexpressed as follows:

N = K1 (o)K2

in which K1 and K2 are experimentally determined fatigue parameters.

13

be a very difficult experimental task: large stresses caused early failure, small stresses resulted

in tests that were too time consuming, and stresses that seemed proper for one temperature were

inappropriate for the other. As a consequence, multiple stress levels were used in the fatigue

testing. For the flexural tests, bending stresses of 115 and 225 psi (792 and 1,550 kPa) were

used at 32°F (0°C) and 80 and 130 psi (551 and 896 kPa), at 68°F (20°C). For the diametral

tests, stresses were selected for each mixture, compaction, and temperature combination to

produce fatigue lives in the ranges of 70,000 to 100,000 applications for the low stress level and

1,000 to 5,000 applications for the higl_ level: For purposes of statistical analysis, stress was

actually considered to be a continuous variable. Following calibration of general regression

models, failure lives were computed for two common stress levels, 100 and 175 psi (689 and

1,206 kPa) for the flexural tests. For the diametral tests, finding two stress levels suitable for

temperature levels of 39.2 and 68°F (4 and 20°C) proved impossible. Stresses suitable for one

temperature were well outside the range of stresses used in the testing program at the other

temperature. Accordingly,-different.low--and high. stress levels were used for the two

temperatures in the diametrai testing. For testing at 39.2°F (4°C), the stress levels were 100

and 150 psi (689 and 1,034 kPa) and at 68°F (20°C), 10 and 40 psi (69 and 276 kPa).

Although creep data can be graphically presented in several ways, the creep curve used

in most of the illustrations herein is a double-logarithmic plot of creep modulus versus time

(Figure 2.1). Such a curve suggests several reasonable possibilities for characterizing resistance

to permanent deformation from creep tests, including its slope, the creep modulus (or permanent

strain) corresponding to a f'LXedtime of loading, or the time to reach a critical strain level. The

hypothesis on which such measures are based is that more resistant mixtures have moduli that

are larger and that decrease less rapidly with increasing time. These mixtures also retain their

14

100000

AXIALCREEP ""-

MODULUS(PSI)lOOO0 ..

\

i '

i ! , i1000 _ _ J

0.01 0.1 1 10 100 1000 10000

Time (see)

I- EARLYCOI.IAPSE -- SURVIVEDI-HOURTESTI

Figure 2.1 Typical Creep Curves (1 psi - 6.89 kPa)

15

integrity longer before they eventually collapse under the constant pressure. More complex

measuresbased on postulated mechanicalbehavior seemed unnecessaryfor this evaluation.

The slope of the creep curve, both at its midpointand from end to end, was investigated

rather thoroughly. The midpoint slope, conceptuallypreferable to the end-to-end slope, proved

to be relatively insensitive to mixture and test variablesand failed on occasion to rankmixtures

and test conditions as expected. Further,becausemanyspecimenscollapsed beforereaching the

one-hour test limit and-some" collapsedmuch-sooner, no measure, taken at a fixed time of

loading, seemed to be suitable for use.

Francken (1977) had reported that there is a threshold state of stress beyond which

specimens fail rapidly. However, unpublishedresearchat the University of California, Berkeley

hasfoundthatsuchathresholdseemstobemorecloselyassociatedwithstrainthanwithstress.

Thisresearchhasalsoshownthatthethresholdlevelofstrainisnotaconstantbutisinfluenced

by bothmixturecompositionand stateof stress.However,fortheconditionsinvestigated

herein,reasonableconsistencywas foundinthestrainlevelatwhicha varietyofmixtures

collapsedundercreeploading:thosemore resistanttopermanentdeformationreachedthe

criticalstrainlevelmuch laterthanlessresistantmixtures.Critical(terminal)strainlevelsin

thecurrenttestingprogramwerefoundtobeapproximately0.008in/in(0.008mm/mm) for

compressivecreepand0.02in/in(0.02mm/mm) forshear2. Figures2.2and2.3demonstrate,

foravarietyofrandomlyselectedmixturesandtestconditions,theconstancyatwhichrapid

deteriorationoccursatornearthesecriticallevels.Thetime(seconds)toreachterminalstrain

provedtobea consistentlyreliablemeasureofresistanceofasphalt-aggregatemixturesto

_Theterminalswainlevelsof0.008in/in(0.008mm/mm) and0.02in/in(0.02mm/mm) forcompressiveandshearcreep,respectively,wereselectedsomewhatarbitrarily.Nevertheless,thereisnoreasontosuspectthatselectionofotherreasonablelevelswouldmateriallyaffectthefindingsorthereliabilityofthisinvestigation.

16

O.l

0.0] _ ......iO.OOB

AXIALSTRAIN(in/in)

0.001

O.O00l

O.l 1 10 lO0 1000 10000

TIME(sec)

Figure 2.2 Constancy of Collapse Strain for a Number of RandomlySelected Tests Under Compressive Creep (1 in/in - 1 ram/ram)

17

0.02

0.01

SHEARSTRAIN

(in/in)

0.001

i ii

0.0001 I i I

0.1 1 10 100 1000 10000

TIME(sec)

Figure 2.3 Constancy of Collapse Strain for a Number OfRandomlySelected Tests Under Shear Creep (1 in/in - 1 ram/ram) "

18

permanent deformation and was adopted for use in the compaction investigation. It is not

expected, however, that time to terminal strain will continue to be used as a primarymeasure

of the resistance to permanent deformation in subsequent phases of SHRP Project A-003A.

Other more fundamentalmeasures will be investigated later.

Because stiffness of the asphalt layers materially affects the response of;pavements to

traffic loads, stiffness is a fundamentalmixturepropertyof considerablesignificanceto pavement

performance. Although the compaction study concentratedon permanent deformation and

fatigue cracking, each of the four tests employed in the main experiment also provided an

estimate of mixture stiffness, and stiffness was added to the set of mixtureproperties evaluated

herein. The flexural and diametralfatigue tests provide estimates of resilientbendingand tensile

moduli, respectively, resulting from a short load pulse of 0.1-second duration. Measurements

were recorded after conditioning by 200 load applications for the flexural tests and 50, for the

diametral. Under creep loading, the modulus, either compressive or shear depending on test

configuration, was also determined at 0.1-second loading. Total, not elastic or recoverable,

strains were used in computing the creep moduli.

19

3.0 DESIGN OF MAIN EXPERIMENT

The primary purpose of the compaction study was to evaluate the effect of method of

compaction (Texas gyratory, kneading, and rolling wheel) on the permanent-deformationand

fatigue behavior of asphalt-aggregatemixtures. The main experiment, utilizing four different

test procedures and 16 different mixtures, was designed to accomplish the following:

1. To identify any effects of compaction method on permanent-deformation and

fatigue properties and on the sensitivities of these properties to mixture

composition and

2. If effects are observed, to determine their statistical significance.

Described herein are the experiment designs and analysis techniques employed in the main

experiment.

3.1 Experimept Desifns

The main experiment was,comprised-of a _eries of fractional factorial experiments with

replicationas needed. Mixtureand test variables,discussedpreviously, aresummarizedin Table

2.1. The four series of tests that were performed included diametral and flexural fatigue and

compressive and shear creep. Experiment designs for the testing are shown in Tables 3.1

through 3.3. An important feature of these designs is that they are all based on a common set

of eight asphalt-aggregate mixes (two asphalts, two aggregates, and two asphaltcontents), each

prepared at two levels of compactive effort (as defined by air-void content). Thus, there are

only 16 different asphalt-aggregatemixtures to be preparedfor each series of tests.

Table 3.1 shows the experiment designfor the diametralfatiguetests for each of the three

compaction methods. As gyratory compaction cannotproduce the beams needed for flexural

2O

Table 3.1 Experiment Design for Diametral Fatigue Tests, Main Experiment

Asphalt Asphalt Aggregate Air Temper- Stress Number ofType Content Type Voids amre Repeats

0 0 O 0 O O,l 2

1 0 0 0 1 0,1 1

0 1 0 0 I 0,I I

I 1 0 0 0 0,I 2

0 0 I 0 1 0,1 I

1 0 I 0 0 0,I 2

0 I I 0 0 0,1 2

1 1 1 0 I 0,1 I

0 0 0 1 I 0,I 1

1 0 0 I 0 0,I 2

0 1 0 I 0 0,I 2

1 I 0 I I 0,1 I

0 0 I I 0 0,I 2

I 0 I 1 1 0,I 1

0 I I 1 1 0,1 1

1 1 1 I 0 0,i 2

NOTE: A 25"1 • 2 fractional factorial in 48 runs (25"I * 2 + 32 repeated tests) for each of thethree compaction methods for a total of 144 tests.

The O's and l's indicate low and high levels of the factors, with the following definitions for the"type" factors:

0 1Asphalt Type: AAK-1 AAG-1Aggregate Type: RL RB

21

fatigue testing, a slight modification of the design is required for flexural testing (Table 3.2).

This is a standard 26"1 fractional factorial run at two stress levels.

As one of the principal measures from fatigue testing is the response of fatigue life (N) to

stress level (Gr),it was decided to construct a fractional factorial design on all factors except

stress, running the resultant design at both levels of stress. For the diametral testing, a full

factorial experiment can thus be written as a [31 • 25] * 2. The only fraction of the 3 * 2s part

that permits estimation of most two-factor interactions is a half replicate of the 25 repeated for

each level of the three-level factor, compaction method. Actually, this design provides estimates

of all two-factor interactions. The resulting 3 * 25"1• 2 design for diametral specimens is as

shown in Table 3.1 for each of the three compaction types.

For the flexural fatigue tests, where only the kneading and rolling-wheel compaction methods

are applicable, the full design is a [26] * 2 factorial. The only fraction of a 26 factorial that

permits estimation of two-factor interactions is a half fraction (26"1), which leads to the design

shown in Table 3.2. This design is similar to that in Table 3.1 for each of the three compaction

methods. The difference is that, for the second compaction type, the temperature levels are

reversed for each of the specimens. Thus, Table 3.2 includes a column for compaction type.

Table 3.3 shows the design for permanent deformation using the shear and compressive creep

tests. Structurally, it is the same design as that shown in Table 3.1 for diametral fatigue tests

for each of the three compaction methods. The only difference is that the "repeats" pattern has

been modified. Eight instead of 16 of the 32 different test conditions are repeated for each

compaction method in order to provide an estimate of test precision independent of any

assumptions about which higher order interactions exist. The runs to be repeated are noted with

an asterisk in the last column of Table 3.3.

22

Table 3.2 Experiment Design for FIexural Fatigue Tests,Main Experiment

Asphalt Asphalt Aggre- Air Compac- Temper- No. offion StressType Content gate Voids Type ature Repeats

0 0 0 0 0 0 0,I 21 0 0 0 0 I 0,I I0 1 0 0 0 1 0,I 11 1 0 0 0 0 0,1 20 0 1 0 0 1 0,1 1l 0 1 0 0 0 0,1 20 1 1 0 0 0 O,l 2I I I 0 0 I 0,I I0 0 0 I 0 1 0,I 1I 0 0 I 0 0 0,I 20 I 0 1 0 0 0,I 2I I 0 I 0 I 0,I I0 0 1 I 0 0 0,I 21 0 1 1 0 1 0,1 10 1 1 1 0 1 0,1 11 1 1 1 0 0 0,1 20 0 0 0 1 1 0,1 11 0 0 0 1 0 0,1 20 I 0 0 I 0 0,1 21 1 0 0 1 I 0,1 10 0 1 0 I 0 0,1 21 0 1 0 1 1 0,1 I0 1 1 0 1 1 0,1 II I 1 0 1 0 0,1 20 0 0 I 1 0 0,1 21 0 0 1 1 1 0,I 10 1 0 1 1 1 0,I 11 1 0 1 1 0 0,I 20 0 1 1 1 I 0,1 11 0 1 1 1 0 0,1 20 I 1 I 1 0 0,I 2

I I I I I 0,1 I

NOTE: A 2e'1* 2 fractionalfactorialin96runs(26"I• 2 + 32repeatedtests)includingtwocompaction methods. The O's and l's indicate low and high levels of the factors, with thefollowing definitions for the "type"factors:

0 IAsphalt Type: AAK-I AAG-IAggregate Type: RL RBCompaction Type: Kneading Rolling Wheel

--.

23

Table 3.3 Experiment Design for Shear and Compressive Creep Tests,Main Experiment

Asphalt Asphalt Aggregate Air Temperature StressType Content Type Voids

0 0 0 0 0 0,I=

1 0 0 0 1 0,10 1 0 0 1 0,11 I 0 0 0 0", 1

0 0 1 0 1 0,1*1 0 1 0 0 0,10 1 1 0 0 0,11 ' 1 1 0 1 0", 1

0 0 0 1 1 0,1"1 0 0 1 0 0,10 1 0 1 0 0,1

1 1 0 1 1 0", 1

0 0 1 1 0 0,1"1 0 1 1 1 0,1

0 1 1 1 1 0,11 1 1 1 0 0", 1

NOTE: A 2s'l * 2 fractional factorial in 40 runs (2s'1 * 2 + 8 repeated tests) for each of threecompaction methods for a total of 120 tests.

Runs to be repeated are indicated by an asterisk, *, on the stress condition.

The O'sand l's indicate low and high levels of the factors, with the following definitions for the"type"factors:

0 1Asphalt Type: AAK-1 AAG-IAggregam Type: RL RB

24

In order to keep the mixtureand test conditions for the permanent-deformationtests similar

to those for the fatigue tests, it was decided to use the same fractional factorial design on the

five factors, excluding stress, running the resultant design at both levels of stress. The full

factorial experiment can thus again be writtenas a [31 • 25] * 2.

Table 3.4 summarizes the number of specimens for each testing method in the main

compaction study. As indicated, a total of 480 specimens were tested.

3.2 Analysis Technioues

Designs selected for the compaction study permit the estimation of main effects of the

experimental factors (variables) and all two-factor interactions. A two-factor interaction, for

example, between asphalt type and temperature, simply measures the extent to which the effect

of temperature on properties of AAK-1 mixturesis different from the effect of temperatureon

similar properties of AAG-1 mixtures.

Because of the lack of complete replication,the designs are unbalanced, and it is not possible

to analyze the data as a simple orthogonalarrayusing, for example, Yates' algorithm. Instead,

a more complex General Linear Model (GLM) procedure must be used. This permits the

imbalance caused by incomplete replicationto be handledwhile still using the same mathematical

model and providing the same statistical tests on the various effect estimates.

In the fatigue tests, the response (dependent)variableis the numberof load cycles to failure.

Creep tests, on the other hand, yield a continuousrecord of deformation (strain)as a function

of time. To facilitate the statistical analysis, the time to reach a critical or terminal level of

strainwas selected as the response variableof interest. A modulus was available from each of

the tests to represent mixture stiffness. All statistical modeling and analysis-was based on

logarithmic transformations of the responsevariables.

25

Table 3.4 Sample Size Required for Main Experiment

Complete Factorial 26 • 31 = 192 cells1/2 Fractional 96 cellsTotal Number of Specimens 480

Number of SpecimensTest I

1/2 Fractional I Replicates Total

Fatigue

• Flexural* 64 32 96

• DL_anetral 96.... 48 144

PermanentDeformation

• Shear Creep 96 24 120

• Compressive Creep 96 24 120

GrandTotal I

m

480

_}nly two compaction methods.

26

The analysis of each type of measurement (response) provides an estimate of each effect

(maineffects and two-factor interactions) togetherwith the sum of squaresfor each effect. This

permitsan analysis of variance and an assessment of statistical significance for eacheffect. The

magnitudesof the effects are used to evaluate the significant relationships and to comparethe

three compaction methods.

The precision of these test designs is such that comparisons between compaction methods

will have a reasonably good chance of detecting differences of about 0.3 standarddeviations.

Within a compaction method, differences of about 0.5 standard deviations can be readily

detectable. For some typical coefficients of variations, representative of those reported in the

technical literature and considered by the authors to be reasonable, these come to:

Typical Values Between WithinCompaction Compaction

Methods Methods

Fatigue life 60 percent 18 percent 30 percent

Permanentdeformation 10 percent 3 percent 5 percent

27

4.0 ANALYSIS AND FINDINGS

The purpose of this section is to present the laboratorydatacollected in the compaction

study-including the main experiment as well as auxiliary studies--and to document its analysis

and important findings resulting therefrom. The main experiment was designed to allow

evaluation of the statistical significance of the principal findings of the study. Auxiliary

experiments, performed to complement the main experiment, included the following:

• t-_xtendedPermanent-DeformationStudy - Furtherexploration of the interaction of air

voids and compaction method together with investigations of both repetitive loading and

the application of a larger axial compressive stress during shear creep testing;

• Complex Modulus Study - Investigation of compaction effects on dynamic modulus and

phase angle Ooss tangent), propertiespossibly significant to pavementperformance and

potentially useful for examining dissipated-energy concepts;

• Comparisons with Other Comoaction Methods- Limited comparisons with specimens

produced by the U.S. Corps of Engineersgyratory testing machine (ASTM D 3387) and

by Exxon's rolling-wheel compactor; and

• Comoarisons with Field Cores- Limited comparisons with field cores taken from

pavements on two California highways, US 101 and I 80.

These auxiliary studies were generally limited to a single asphalt-aggregatemixture, and the

focus was principally on gyratory and kneading compaction. They were not designed as

statistical experiments, and the possible statistical significance of their findings was not assessed.

4.! ]Vl[_inF_,xveriment_.

As detailed earlier, the main experiment was designed as series of unequallyreplicated,

fractional factorials. The factor or variable of primary interest was, of course, compaction

28

method. Sixteen asphalt-aggregate mixtures were investigated: each was tested at two

temperature levels and two stress levels. Mixture properties or response measures, which were

of three types, included the following:

• Resistance to permanent deformation

- Time under unconfined compressive or shear creep to reach a limiting strainlevel, 0.008 in/in (0.008 mm/mm) for compressive creep and 0.02 in/in (0.02mm/mm) for shear creep

• Resistance to fatigue cracking

- Number of load applications to failure

• Stiffness modulus

- Compressive creep modulus at O.l-second load duration (compressive creep)- Shear creep modulus at 0. l-second load duration (shear creep)- Flexural stiffness at 200 load applications of 0.1-second load pulse (flexural

fatigue)- Tensile stiffness at 50 load applications of 0. l-second load pulse (diametral

fatigue)

From the very beginning of the laboratory work, compacting specimens of widely varying

composition to targeted air-void contents proved to be an especially difficult task, regardless of

compaction method. Despite the considerable attention that was directed to this matter, it proved

impossible to prepare a set of specimens without variability in air voids and whose average

air-void content matched the targeted level. These difficulties were exacerbated by the extreme

sensitivity of the response measures (identified above) to air-void content. To reduce the effects

of off-target air voids, the statistical modeling and analysis treated air-void content as a

continuous rather than discrete variable, utilizing a general regression approach. A similar

approach, previously explained, was used to treat stress level. The calibrated regression models

were used to provide estimates of the response measures corresponding to _arget air-void

contents of 4.5 percent (low level) and 11.5 percent (high level) as well as the target stress

levels. Statistical testing of the main compaction experiment is detailed in Section 4.1.6.

29

4.1.1 Comvaction Processes. Compaction equipment included two Texas gyratory shear

devices, one for fabricating 6 x 8-inch (152.4 x 203.2 ram) cylinders and the second for

fabricating 2.5 x 4-inch (63.5 x 101.6 mm) briquets;a Triaxial Institutekneadingcompactor for

fabricating 3.5 x 3.5 x 15-inch (88.9 x 88.9 x 381 ram)beams, 4 x 8-inch (101.6 x 203.2 mm)

cylinders, and 2.5 x 4-inch (63.5 x 101.6 ram) briquets; and a "homemade" rolling-wheel

compactor which fabricated 16 x 16 x 9-inch (406.4 x 406.4 x 228.6 mm) slabs from which

specimens of the required dimensions were cored and/or sawed. Details regarding the

compaction processes are presented in Appendix A. ....

In general, the performance of each compactor was quite satisfactory. The gyratory

compactors excelled in the speed with which compaction was achieved. By comparison,

kneading compaction, particularly to low void levels, was much more time consuming. At the

same time, the kneading compactor proved moreversatile, especially in its ability to form large

beam specimens. Rolling-wheel compactionis distinguished by the larger quantitiesof materials

required for compaction. This reduced the totaJtime required for mixing and compaction and

was well suited for producing the large numberof specimens required in the compaction study.

From each rolling-wheel slab, 16 specimens could be fabricated, four each for compressive

creep, shear creep, flexural fatigue, and diametral fatigue testing.

With each compaction method, careful attentionwas given to following procedures that

resulted in uniform, homogeneous specimens. In general, this effort was successful, and the

various compaction methods could not be distinguished in their abilities to consistently produce

specimens that were internally uniform. At the same time, the surfaces of all molded specimens

were distinctly different from their interiors. In particular, flushing of the asphaltwas common

on the surfaces of high-asphalt-content specimens prepared by both gyratory and kneading

compaction.

3O

4.1.2 Resistance to Permanent Deformation. Althoughearly work had conclusively shown

thatcompaction methodcan have a distincteffect on engineering properties of asphalt-aggregate

mixtures (for example, Vallerga, 1951), the initial expectation when this study began was that

gyratory, kneading, and rolling-wheel compactorswere sufficiently similar that anydifferences

in mixture properties would likely be small and, in terms of pavement performance,

insignificant. That expectation was quickly dispelled when testing began: when subjected to

either compressive or shear creep, early collapse was much more prevalent in gyratory

specimens than in kneading or rolling-wheel specimens. It remained, of course, to establish the

extent to which thispatternpersisted for a range of mixturesand test conditions, to quantifythe

differences, and to ascertain their statistical and practical significance.

Average (geometric)resistancetopermanentdeformation-expressed as the time(seconds)

to reacha critical level of strainunder creeploading-is summarizedfor a range of mixtureand

test conditions in the right-mostcolumns of Tables4.1 and 4.2. For both compressiveand shear

loading, kneading specimens are generally most resistant to permanent deformation, followed

in orderby rolling-wheel and gyratory specimens. However, at the high air-void content (11.5

percent) and for the chert (IU..) aggregate, the effect of compaction method on permanent-

deformation resistance is greatly diminished. Table 4.3 provides an overall summary, ranking

the threecompaction methods not only on the basis of overall performance but also in terms of

the 12 individual mixture and test conditions of Tables 4.1 and 4.2 and confirming the general

effect, noted above, of compaction methodon resistance to permanent deformation.

One of the difficulties in using time to terminal strain as a response variable is that it

must be estimated for specimens which survive the one-hour test limit. Linear extrapolations--

of the logarithmic transformations, as used herein, overestimate the true response because they

ignore the downward turnin the creep curve which heraldsthe beginning of collapse. The real

31

Table 4.1 Summary Results of UnconJ'med Axial Compressive Creep Tests

Average Creep Modulus at 0. I Seconds, Average Time to Terminal Axial Strainpsi of 0.008 in/in (0.008 ram/ram),

oecoods

Gyrator,/ Kneading Rolling Gyratory Kneading RollingWheel Wheel

All Tests 19,500 22,400 20,900 28.4 249.1 43.8

AsphaltType

• AAK-I 22,700 24,300 23,900 63.6 388.8 79.2

• AAG-I 16,700 20,800 18,300 12.7 159.5 24.2

• Percent Difference 30.4 15.5 26.5 133.4 83.6 106.4

Asphalt Content

• Optimum 21,900 2.5,900 23,800 64.7 1,037.7 66.2

• High 17,300 19,400 18,400 12.5 59.8 28.9


Aggregate Type

• Chert (RL) 16,500 16,600 17,100 9.8 9.6 10.6

• Granite (RB) 22,900 30,400 25,600 82.3 6,455.2 181.3


Air Void Content

• 4.5+_ 32,200 39,300 32,800 622.1 18,427.7 357.1

• 11.5+% 11,800 12,800 13,300 1.3 3.4 5.4


Test Temperature

• 104"F (40"C) 37,400 43,000 40,100 94.2 1,433.0 276.6

• 140"F (60"C) 10,100 11,700 10,900 8.6 43.3 6.9


Vertical Stress

• 14.5 psi 19,500 22,200 20,400 127.7 673.7 113.7

• 29.0 psi 19,500 22,700 21,400 6.3 92.1 16.8


Note: Percent difference is the difference expressed as a percentage of the average; 1 psi --6.89 kPa.

32

Table 4.2 Summary Results of Unconl'med Shear Creep Tests

,u

Average Creep Modulus at 0. l Seconds, Average Time w Termina] Shear Strainpsi of 0.02 in/in (0.02 ram/ram), seconds

Gyramry Kneading Rolling Gyratory Kneading RollingWheel Wheelir

All Tests 2,570 2,670 2,820 21.9 59.8 42.1

Asphalt Type

• AAK-1 2,980 2,980 3,480 33.8 S5.4 100.1

• AAG-1 2,230 2,400 2,280 14.2 41.9 17.7


Asphalt Content

• Optimum 2,760 3,020 3,510 26.6 83.7 80.5

• High 2,400 2,370 2,260 18.0 42.7 22.0


Aggregate Type

• Chert (RL) 2,460 2,350 2,550 11.8 21.3 15.0

• Granite (P,B) 2,690 3,040 3,110 40.7 168.2 117.8


Air Void Content

• 4.5+% 3,500 4,110 4,380 50.5 450.9 235.2

• 11.5_.+_ 1,900 1,740 1,810 9.5 7.9 7.5


Test Temperature

• 104"F (40"C) 6,760 6,700 7,080 120.3 334.2 250.6

• 140"F (60"C) 980 1,060 1,120 4.0 10.7 7.1


Shear Stress

• 2.4 psi 3,020 2,990 3,330 98.9 389.9 195.8

• 4.8 psi 2,190 2,390 2,380 4.8 9.2 9.0

• Percent Difference 31.9 22.3 33.3 181.$ 190.8 182.4

Note: Percent difference is the difference expressed as a percentage of the average; 1 psi --6.89 kPa.

33

Table 4.3 Effect of Compaction Method on Resistance to Perm*nent Deformation

]Rm_g in the Twelve Mixture mKITest VariableCompaction Overall Comparisonsof Tables 4. i and 4.2

Method Test Resistance I

Strongest Intermediate I Weakest

Compressive Weakest None 4 Times 8 Times

Gyratory Creep

Shear Creep We._est 1 Time None 11 Times

Compressive Strongest 10 Times 1 Time 1 Time

K_._lmg Creep

Shear Creep Strongest 10 Times 2 Times None

Compressive Intermediate 2 Times 7 Times 3 TimesRolling ..... CropWheel

Shear Creep Intermediate 1 Time 10 Times 1 Time

34

concern is whether the overestimates are so large that, when grouped with other observations,

the averages are meaningless. Rankingtechniquesoffer an alternative that avoids the possible

bias of inflated averages.

It was possible to use ranking techniquesin interpretingboth compressive and shear test

results because direct comparisons among the three compaction methods were possible for 32

unique combinations of mixtures, temperatures, and stress levels. For each of these

combinations, the three compaction methods were ranked according to the resistance of the

specimens to permanent deformation. Again, time to terminal strain was considered an

appropriateindicator of deformation resistance. Eachcompaction method was assigned a rating

of 1, 2, or 3, depending on its ranking: a rating of 1 signified the greatest resistance to

permanent deformation. Results are summarizedas follows:

Compaction Number of #1 Rankings Average Rating

Method Compression Shear Compression Shear

Kneading 16 13 1.63 1.78

Rolling Wheel 9 13 2.06 1.88

Gyratory 7 6 2.31 2.34

This analysis thus confirms that kneadingspecimens are mostresistantto permanentdeformation

and that gyratory specimens are least resistant.

Another objective of the main compaction experiment was to determine whether

compaction methods might differ in terms of the sensitivity of engineering properties of the

specimens they produce to mixture composition. Insofar as insignificant effects are not

artificially exaggerated, sensitivity is considered to be a desirable attribute: certainly, a

compaction procedure that effectively masks important compositional effects would be

unattractive. Mixture variables in the main experiment included asphalt type, asphaltcontent,

aggregate type, and air-void content. With two levels of each, a total of 16 different mixtures

35

were included. Because of the fractionalnatureof the experiment, however, only eight mixtures

were evaluated at each combinationof temperature and stress level. The coefficient of variation

of the time to terminal st.r_nfor these eight mixtures(for each combination of temperature

stress level) was selected as a measure of the sensitivity to mixture composition3. A small

coefficient suggests that the resistancesof the eight differentmixtures to permanentdeformation

are much the same, that is, the resistances are insensitive to the rather large compositional

differences among the eight mixtures. A large coefficient, on the other hand, reflects a greater

sensitivity to mixture composition. Results-of--the. computations-summarized in Table

4.4-indicate that the permanent-deformation response of kneading specimens is generally more

sensitive to mixture composition than thatof specimens compacted by other techniques. The

evidence is compelling because, with exception of the high-temperature shear testing, kneading

specimens consistently rankednear or at the top across the range of stresses, temperatures, and

test methods.

An additional matterof interest is the effect of mixtureand test variables on resi.,

of the mixtures to permanentdeformation. The orderof the effects is generally known, a priori,

and comparisons can not only verify this a priori order and confh'm thattime to terminal strain

is a consistent measure of resistance to permanent deformation but, more importantly, can

identify compaction "abnormalities." Results of these comparisons, summarizedin Table 4.5,

are quite consistent and confu'ma priori expectations. Nothing unusual is foundwith any of the

compaction methods or with the chosen response measure, time to terminal strain.

While time to terminal strain is a useful quantification of permanent-deformation

response, it does not capture the full extent of informationcollected in creep testing. Additional

3The coefficient of variation is a useful, although imperfect, measure of sensitivity tomixturevariables. Its imperfections arise because it includes not only mixture effects butQrsources of variability as well (including random and testing errors).

36

Table 4.4 Effect of Compaction Method on Sensitivity ofPermanent-Deformation Response to Mixture Variables

Test Ceefficient of Variation of Seconds to Termini/Strain (Percent)Stress Temperature Method I

Gyratory Kneading _ Rolling Wheel

Axial 224.11 215.0 166.4Law

Shear 108.0 236.5 147.7Low

Axial 264.6 264.5 217.8High

Shear 177.7 213.6 263.6

Axial 163.6 264.6 153.6Low

Shear 84.3 !43.3 118.4_th

Axial 125.8 264.5 182.6l-Iigh

Shear 100.7 111.3 208.4

m

37

Table 4.5 Effect of Mixture and Test Variables on Resistanceto Permanent Deformation

Compaction MethodVariable Test

Gyratory I Kneading I Rolling Wheel

Compressive AAK-I > AAG-I AAK-I > AAG-I AAK-I > AAG-I

Asphalt Type creep

Shear creep AAK-I • AAG-I AAK-I • AAG-I A_-I • AAG-I

Compressive Optimum • High Optimum • High Optimum • High

Asphalt Content creep

Shear creep Optimum • High Optimum • High Optimum • High

Compressive Granite (RB) • Granite (RB) • Granite (R.B) •creep " Chert (RL)..... Chert (RL) Chert (RL)

Aggregate TypeShear creep Granite (P,B) • Granite (P,B) • Granite 0_B) •

Chert (ILL) Chert (RL) Chert (P.L)

Compressive Low • High Low • High Low • High

Air Void Content creep

Shear creep Low • High Low • High Low • High


Temperature creep



Stress creep


b-

38

information is available from graphic portrayals, usually double logarithmic plots of creep

modulus or creep compliance as a function of time. Figure 4. I, showing the collective results

of the compressive creep testing, exemplifies the creep curve and provides visual confirmation

of the superiority of specimens prepared by kneading compaction 4. Each of the lines on Figure

4.1 represents a total of 40 specimens: the sum of the creep moduli of the 40 specimens is

shown as the ordinate.

Unfortunately, composite curves such as these are not faithful representations of typical

creep response and may give a distorted image of collective response: When specimens collapse

before reaching the one-hour test limit, their moduli are thereafter considered to be zero. Thus,

each curve represents fewer and more resistant specimens as the loading time increases. As a

consequence, the collective curve is less steeply sloped than the average of its individual

components. At one hour, the curve is quite sensitive to the number of surviving specimens,

perhaps more so than their individual moduli. In Figure 4. I, the number of specimens surviving

the one-hour test included eight kneading specimens, six rolling-wheel specimens, and six

gyratory specimens. Again focusing on Figure 4.1, gyratory specimens appear superior to

rolling-wheel specimens at short as well as long loading times, a finding that appears

contradictory both to laboratory experiences in handling and testing these specimens and to the

numerical comparisons summarized above. This apparent contradiction can be partly explained

by differences in the average air-void contents. The voids in the rolling-wheel specimens

averaged 7.3 percent, 0.7 percent more than the gyratory average of 6.6 percent. All other

factors being equal, mixtures with larger air-void contents are less resistant to permanent

deformation than those with smaller void contents. Figure 4.2 illustrates the effect of

4The symbols G, K, and R represent gyratory, kneading, and rolling-wheel compaction,respectively. Corresponding average air-void contents are shown in parentheses.

39

A A

m oi

F'_,ure 4.1 Effect of Compaction on Compressive Creep Modulus (1 psi = 6.89 k.Pa)

40

V V V

mm 0 +

Figure 4.2 Effect of Compaction on Shear Creep Modulus (1 psi = 6.89 kPa)

41

compaction methodon shearcreep responseand confirms thatkneadingspecimens aregenerally

more resistant to permanentdeformationthangyratory specimens.

The interactive effects of compactionmethod and the various mixtureand test variables

are illustratedby the composite response curves of Figures 4.3 through 4.14. These figures are

paired such that the f'trstof each pair representsresponse to compressive loads and the second,

to shear loads. The legend of each figure includes information about the average air voids for

each particular set of conditions.

The experimentdesign was arrangedso that two-way interactions could be investigated.

This means that one could compare, for instance, the sum of creep moduli of all granite (P.B)

kneading specimens with the sum of creep moduli of all chert (RL) kneading specimens. These

sums can then be furthercompared with the corresponding specimens preparedby gyratory or

rolling-wheel compaction.

Figure 4.3 indicates that the compressivecreep modulus is sensitive to the relative effect

of asphalt type. The higher viscosity_of AAK_-Iasphaltat temperaturesof 104 and 140°F (40

and 60°C) causes its mixtures to exhibit better resistance to permanent deformation.

Furthermore kneading specimens exhibit a consistently higher modulus than those preparedby

gyratory or rolling-wheel compaction. It is also interesting to note, that, regardless of asphalt

type, B (AAK-1) or V (AAG-1), kneading specimens seem to approach the same behavior at

long times of loading. The biggest differences between kneading and the other compaction

methods are encountered for the specimens fabricated with the AAG-1 asphalt. Gyratory

sp_mens behave similarly to roUing-wheelspecimens: the slight differences between them can

be explained by differences in the average void ratios. Figure 4.4 indicates that the shearcreep

modulus is also sensitive to asphalt type but at long times of loading the compaction method

seems to have a predominanteffect with kneading specimens being slightly more resistant.

42

A A _ A A

i I I I i I

Figure 4.3 Combined Effects of Compaction and Asphalt Type on CompressiveCreep Modulus, AAK-1 [B] and AAG-1 IV] (1 psi = 6.89 k]Pa)

43

I I I I I I

mm 0 + <> HI

V

Figure 4.4 Combined Effects of Compaction and Asphalt Type on ShearCreep Modulus, AAK-1 [B] and AAG-I m (I psi = 6.89 kPa)

44

Figure 4.5 indicates that the compressive creep modulus is also somewhat sensitive to

asphalt content. Specimens prepared with optimum 0ow) asphalt contents perform slightly

better. The data also indicates that kneading specimens exhibit creep moduli consistently larger

than rolling-wheel or gyratory specimens. This difference is especially noticeable with optimum

asphalt contents. The combined effects of compaction method and asphalt content on shear creep

modulus are illustrated by Figure 4.6.

While the optimum asphalt content (0) was chosen to be representative of a Hvecm

mixture design, the high asphalt content was selected to be somewhat indicative of the Marshall

(75-blow) mixture design. It is generally accepted that Hveem designs are more resistant to

permanent deformation than Marshall designs. Some have also argued that mixtures designed

by the Marshall method have not always performed well, especially in hotter climates.

Figure 4.7 shows that the compressive creep modulus is sensitive to aggregate type.

Mixtures prepared with granite (RB) perform consistently better than those prepared with chert

(P.L). It is also apparent that the granite (R.B) kneading specimens perform significantly better

than those prepared by either the gyratory or rolling-wheel method. The difference is magnified

if one realizes that the average void ratio for the kneading specimens is 7.7 percent and that the

average for the roUing-wheel and gyratory specimens is about 6.5 percent. With such a large

difference in average air-void content, it would be expected that the specimens prepared by

kneading compaction might exhibit slightly lower creep moduli. It is interesting to note that the

behavior of the granite (RB) gyratory specimens is very similar to the granite ('P,B)rolling-wheel

specimens. Mixtures prepared with the chert (RL) have a behavior more comparable within all

compaction methods. The significantly lower values of the chert (RL) rolling-wheel specimens

can be explained by the significantly higher void content (8.2 percent versus 6.7 percent for

gyratory or 7.5 percent for kneading specimens). The shear creep response is also quite

45

A A A A _ A

I I I I I I

• • _mmq-

Figure ,1.5 Combined Effects of Compaction and Asphalt Content on CompressiveCreep Modulus, Optimum Asphalt [0] and High Asphalt [I] (I psi = 6.89 kPa)

46

A _ A _ _ A

I I I ) I I

mm D + <_ HI

Figure 4.6 Combined Effects of Compaction and Asphalt Content on Shear CreepModulus, Optimnm Asphalt [0] and High Asphalt [1] (1 psi -- 6.89 kPa)

47

A A A _ A

V _ V

I I I I I I

Figure 4.7 Combined EfTects of Compaction and Aggregate Type on CompressiveCreep Modulus, ILL Chert [1] and RB Granite [W] (1 psi = 6.89 kPa)Fig 4.5

48

sensitive to aggregate type (Figure 4.8). Again, granite ('P,.B)specimens perform much better

than chert (P,.L)specimens. The differences between gyratory and kneading specimens are more

noticeable in granite (R.B) specimens.

Figure 4.9 shows the large sensitivity of the compressive creep moduli to void content.

The low void data indicate that kneading specimens exhibit higher resistanceto permanent

deformation even though the average void content (4.2 percent) is higher than that of gyratory

specimens (3. I percent). The void content also affects the shear creep response of the mixtures

(Figure 4. I0). The differences between gyratory and kneading-specimens are more noticeable

at low void contents.

Figure 4.11 indicates that one of the most sensitive factors influencing the permanent-

deformation characteristics of asphalt mixtures is temperature. Also noticeable is the

significantly different behavior of kneading specimens from that of specimens compacted with

the other devices. The differences increase with temperature. Figure 4.12 confirms that

permanent-deformation behavior in shear is also particularly sensitive to temperature.

Differences between gyratory and kneading specimens appear to be larger at high temperatures.

Figure 4.13 indicates that, regardless of stress level, kneading specimens consistently

perform better than specimens produced by either of the other methods. Figure 4.14 indicates

that the shear creep response is more sensitive to stress level than the compressive creep

response. In both cases, however, kneading specimens perform better than gyratory specimens.

The above data indicates that the largest difference in the compressive creep response is

between kneading and gyratory specimens and that rolling-wheel specimens generally have an

intermediate response. The differences are particularly noticeable for mixtures fabricated with

AR..4000 asphalt (AAG-I), with optimum (low) asphalt content, and granite (P.B) aggregate.

They are exacerbated when air voids are low and when testing is performed at high temperatures

49

A A A

I I I I I I

i

F'_ure 4.8 Combined Effects of Compaction and Aggregate Type on ShearCreep Modulus, RL Chert [T] and RB Granite [W] (I psi --- 6.89 kPa)

5O

/_gure 4.9 Combined Effects of Compaction and Air-Void Content on CompressiveCreep Modulus, Low Voids [0] and High Voids [I] (l psi -- 6.B9 kPa)

51

I I I I I I

Figure 4.10 Combined Effects of Compaction and Air-Void Content on ShearCreep Modulus, Low Voids [0] and High Voids [1] (1 psi = 6.89 kPa)

52

A A _ A _ A

I I I I I I

?

F'_re 4.U Combined Effects of Compaction and Temperature on CompressiveCreep Modulus, 104°F (40°C) [0] and 140°F (60°C) [1] (1 psi - 6.89 kPa)

53

I I I I I I

............T-_

Figure 4.12 Combined Effects of Compaction and Temperature on Shear CreepModulus, 104°¥ (40°C) [0] and 140°F (60°C) [1] (1 psi = 6.89 IcPa)

54

I I I I I I

Figure 4.13 Combined Effects of Compaction and Stress Level on CompressiveCreep Modulus, 14.5 psi [0] and 29.0 psi [1] (1 psi = 6.89 kPa)

55

A _ _ A A

c_ _ cc_ c5 r-. r-:

I I I I I I

! ,.. ..

i

Figure 4.14 Combined Effects of Compaction and Stress Level on ShearCreep Modulus, 2.4 psi [0] and 4.8 psi [1] (1 psi - 6.89 kPa)

56

and high stress levels. At high temperaturesthe AAG-1 asphalt is less viscous than the AAK-1

asphalt. This makes the resistance to permanentdeformation moredependenton the contribution

of the aggregate in the mixture. The extended permanent-deformation study investigates

specifically the differences in performance between gyratoryand kneading specimens for these

particularconditions.

4.1.3 Resistance to Fatigue Crackine. In additionto permanent deformation, possible effects

of compaction method., on.. resistance, to. fatigue, cracking were also of great interest.

Con_olled-stress testing in both flexural and diametral modes was conducted to evaluate these

effects. The responseparameterwas the numberof load applicationsto failure. In the flexural

tests, failure occurred catastrophically when cracking propagatedcompletely through the beam

specimen. In the diametral tests, failure was defined as the number of load applications

producing either collapse or permanent deformationin the direction of load application of 0.20

inch (5.1 ram) at 39.2°F (4°C) or 0.50 inch (12.7 ram) at 68°F (20°C), whichever occurred

firsts . This measure was chosen not only for its convenience but also because otheralternatives

seemed no better suited for isolating effects of fatigue cracking from other mechanisms which

produce permanent distortion and cracking undercyclic diametralloads.

Results of the flexural-fatigue testing are summarized in Table 4.6. Gyratory

compaction was excluded from the flexural testingprogrambecause beams of the requiredsize

could notbe fabricated. Rolling-wheel specimens were generally superiorthroughout the range

of comparisons. The average rolling-wheel specimen failed after 30,000 applications compared

to 22,800 for the avexage kneading specimen, a difference of approximately 27 percent.

5Except for weak specimens at 68°F (20°C), failure was typically by cracking and not bycrushing or other modes. Once a crack formed, it propagated rapidly: under thesecircumstances the deformation limits were simply a convenient method for shutting down theequipment.

57

Table 4.6 Summary Results of FIexural Fatigue Tests

AverageInitialFlexuralStiffness,psi AverageNumber ofLoad Applicationsto Failure

Gyratory Kneading Rolling RollingWheel Gyratory Kneading Wheel

All Tests 529,000 573,000 22,800 30,000

AJph_tType

• AAX-1 582,000 590,000 26,300 36,200

• A.AG-1 480,000 $57,000 19,800 24,800

• Percent Difference 19.2 5.8 28.2 37.4

Asphalt Content

• Optimum 559,000 627,000 25,000 43,400

• High 500,000 524,000 20,800 20,700


Aggregate Type

• Chert (ILL) 469,000 535,000 17,000 31,100

• Granite (RB) Beams can $96,000 615,000 Beams can 30,700 28,900

• Percent Difference not be 23.8 13.9 not be 57.4 7.3compacted ,;umpacted

Air Void Content

• 4.5+% 715,000 797,000 57,600 76,200

• 11.5+% 391,000 412,000 9,050 I 1,800


Test Temperature

• 32"F (0"C) 985,000 1,095,000 81,600 73,200

• 68"F (20"C) 284,000 300,000 6,390 12,300


Flexura/Stress

• I00 psi 571,000 613,000 $1,800 93,200

• 175 p'; 490,000 $36,000 6,380 9,640


Note: Percent difference is the difference expressed as a percentage of the average; 1 psi -6.89 kPa.

58

In controned-stress testing, the stiffest mixtures are usually the best performers. Table

4.6 seems to confirm this notion: based on average (geometric) performance, rolling-wheel

specimens were consistently stiffer than their kneading counterparts. Combined effects of

mixture stiffness and compaction method on flexure] fatigue life are illustratedin Figures 4.15

and 4.16. Each point on these figures representsan individual test (or an average in the case

of replicates): in all, the entire set of mixturevariables and test temperatures is represented.

The lines were fit by normal least squarestechniques. Correlationsbetween In fatigue life and

In stiffness were reasonably good: coefficients of determinationwere 0.69 (rolling wheel) and

0.94 (kneading) for the low-stress testing and 0.83 (rolling wheel) and 0.93 (kneading) for the

high-stress testing.

This analysis provides furthercont'h-mationthat stiffer mixtures do perform better in

controlled-stress fatigue tests. More importantly,the nearness of the best-fit lines for kneading

and rolling-wheel compaction suggests that at least some of the difference between compaction

methods can be attributedto mixture stiffness.

Results of the diametral-fatiguetesting are summarizedin Table 4.7. This analysis is

particularly important because gyratory specimens were included in the diametral testing

program. The diametral tests confirm that rolling-wheel specimensresist fatiguecracking better

than kneading specimens. Gyratory specimens, which could not be tested in fiexural fatigue,

proved superior to both kneading and rolling-wheel specimens when tested in the diametral

configuration.

A pn771ing feature of the diametral data, as reported in Table 4.7, is that kneading

specimens (although failing sooner than rolling-wheel specimens) are somewhat stiffer than

roUing-wheelspecimens. Nevertheless, collective examinationof individualtestsconfirmed that

stiffer specimens provide better resistanceto cyclic diametralloading for each of the three

59

........................................ o,° ° °,- O-o° ° .: --: .............. .; ........• * , * * °• , * ° ,• t * * * !t * * i i• * ° , *

1000000 .............................................................. o, i ° , , , ' * ' '

........ ...;°.. &.., ......° * i ° * i ° "

: ................... ".**'°..',_.._.'.i * ° * *, * ° , ° ,

"_ 100000 ' ............... ::,o ..... L.. ,.......................J, J • °°*

::::::::::::::::::::::::::::::::::, , --'---:---:--'-.:-&............ -.........

, , . , ,. *°.. ..... ,

, °

1 1 '" | 1 ; ..... *

_ i i * * i

, la J : t , * i * * i

_ ', ......I0000 ..................................

.°_ .... ° ...... ..°.°.°°°.°°li

.-'T...... ,= '...... F**- °;--* -¢---:- -*'.-- _.--_.............. -i........ ,

..$...: ........ j,,: ........4-_::-'_ ...... _........ :...... _..-+-.4--_,_._t,w-K_ect_O ....... :

o, | , , , , | , , , • _v *

4- ��$#œ��_,........ :...... ,_----',---<--,_-_.._.._-RaZtt_-'=,........ :I , , i , * * * | * *

A .I .L i.J,...°° ...... ......................'.--'---'--' .......... °.. ...... ..°°..

s , , , t D * , *! ! t . I , , o *

', : ', *, ,, ,, ,, ,,I000 1 i i i i i i I _ i

I00000 1000000 •

Flexural Stiffness (psi)

Figure 4.15 Combined Effects of Compaction and Mixture Stiffnesson Flexural Fatigue at Low Stress, 100 psi (1 psi - 6.89 kPa)

60

, , , ,

Figure 4.16 Combined Effects of Compaction and Mixture Stiffn_on FIexural Fatigue at High Stress, 175 psi (I psi -" 6.89 kPa)

61

Table 4.7 Summary Results of Diametral Fatigue Tests

i

Average Initial Tensile Stiffness, psi Average Number of Load Applicationsto Failure

Oyratory Kneading Rolling Gyrawry Kneading RollingWheel Wheel

All Tests 663,000 660,000 627,000 11,100 9,150 10,600

Asphalt Type

• AAK-1 436,000 437,000 486,000 5,170 4,060 6,250

• AAG-1 1,010,000 997,000 809,000 23,800 20,600 17,900


Asphalt Content

• Optimum 760,000 759,000 692,000 12,200 9,140 7,870

• High 579,000 573,000 569,000 10,100 9,160 14,200


Aggregate Type

• Chert (RL) 670,000 662,000 570,000 9,760 6,580 4,860

• Granite (P,B) 657,000 658,000 690,000 12,600 12,700 23,100

• Percent Difference 2.I 0.6 19.0 25.6 63.7 130.4

Air Void Content

• 4.5+% 903,000 966,000 863,000 43,600 39,000 45,100

• 11.5+% 488,000 451,000 456,000 2,820 2,140 2,480


Test Temperature

• 39.2"F (4"C) 886,000 953,000 992,000 15,500 10,700 7,740

• 68"F (20"C) 497,000 456,000 397,000 7,450 7,820 14,500


Tensile Stress

• Low 695,000 684,000 663,000 30,600 33,200 64,900

• High 634,000 636,000 594,000 4,020 2,520 1,730


Note: Percent difference is the difference expressed as a percentageof the average; 1 psi =6.89 kPa; low stress is 100 psi at 39.2°F (4°C) and 10 psi at 68°F (20°C); high stressis 150 psi at 39.2°F (4°C) and 40 psi at 68°F (20°C).

62

compaction methods (Figures 4.17 and 4.18). As evidenced by coefficients of determination in

the range of 0.49 to 0.78 (tabulated below), the fit between In fatigue life and In stiffness is

weaker for diametral testing than for flexural testing.

Coefficient of Determination

Stress Temperature (p-value)

Gyratory Kneading Rolling Wheel

0.587 0.778 0.540Low(0.0267) (0.0037) (0.0379)

Low

High 0.649 0.615 0.688(0.0157) (0.0212) (0.0109)

0.508 0.780 0.564Low(0.0471) (0.0037) (0.0317)

High0.494 0.702 0.642

High (0.0517) (0.0094) (0.0169)

The apparent contradiction in Table 4.7 between compaction's collective stiffness effects

and its fatigue effects remains unexplained. It may be due to the relative inaccuracy of stiffness

measurements in diametral testing or may be a harmless artifact of the averaging process which,

with small sample sizes and data extremes, may have produced misleading results. At the same

time, possible confounding effects of surface condition (as-molded or cut) should not go unnoted.

Gyratory and kneading specimens were tested as molded while rolling-wheel specimens, on the

other hand, were cored from compacted slabs and sawed to the proper heights. The possible

effect of surface condition is further addressed in Section 5.0.

Table 4.8 summarizes the comparative effect of compaction method on fatigue resistance.

On the basis of the diametral-fatigue testing, gyratory specimens were found to be the most

resistant. Both flexural and diametral testing confirmed that rolling-wheel specimens were

superior to kneading specimens in terms of fatigue resistance. The fatigue response of rolling-

63

a. Low Temperature, 39.2°F (4°C)

10000000 ..............................................................................................................................................................

41.

I000000 ............

° _ i • Gyratory

_ ....""i 0 Kneading

--_ 0

r._ ...........................................................................,......>__S._,:................................. • )_oLti,__,,t= I00000 ....................................................................• 0 ...I i.ll ............Gyrltory

............. _neoding

10000 ..............................:_.':.................................................................................................................................................$'" • •**°n

!iii

I000 I

100 1000 10000

_Ns]_ s_ss (lO0OPSDb. High Temperature, 680F (20°C)

Figure 4.17 Combined Ellect_ of Compaction and Mixture Stillnesson Diametral Fatigue at Low Stress (1 psi = 6.89 kPa)

64

! _/' ....!!--l 00000 ..................................................... l ...... _ ...................... £ ' ............................

• Gyrmtory

i T"/_ .+ ii ."" 0 Kneadi_l 0 ! "_ i

r.,. I0000 _o .-'l. i • ,o,t,,_;meet

l " " m 11 '

_ __ I I l i ..... __ ...... _ - ....... Gyrmtory

" -- ............... ICnemdir_

..•....................................:.........-"_=..........................I'_........"T--

1000 "i_ -,ort,,,,_.t,..,*

0@

l0 -------

100 1000 10000

TENSILESTIFFNESS(1000 PSi)

a. Low Temperature, 39.2°F 14°C)

tooooo.....................................................................................................................................................................................................i,,

• ........_:L.....................................................................I O O O 0 ................... ........ 0 • ......................... _" ;,."""'•:

O

• Gyratory

• . -'.'-; " ie ' " " '" l l" " " _ _ Kneading

• Rot lir_ WheeL.-'" ...."'+ t

............ • ................... ; " ..-_I

• 0 ol

< //...-.+. of.x.. ! _ Rotting Wheet

...'"

100 ........................................................................................................................................................................................................:t

° i

100 1000 10000

TENSILESTIFFNESS(1000 PSI)

b. High Temperature, 68°F 120°C)

Figure 4.18 Combined Effects of Compaction and Mixture StUTnesson Diametral Fatigue at High Stress 11 psi = 6.89 kPa)

65

Table ,I.8 Effect of Compaction Method on Resistance to Fatigue

RankingintheTwelveMixtureandTestVariable

Compaction Overall Comparisons of Tables 4.6 and 4.7Method Test Resistance

Strongest Intermediate Weakest

FlexuralFatigue N/A N/A N/A N/AGyratory

Diametral Fatigue Strongest 6 Times 3.5 Times 2.5 Times

FlexuralFatigue Intermediate 3 Times 9 Times N/AKneading

DiametralFatigue Weakest None 7.5Times 4.5Times

Rolling FlexuralFatigue Strongest 9 Times 3 Times N/AWheel

Diametral Fatigue Intermediate 6 Times 1 Time 5 Times

66

wheel specimens is most sensitive to mixture composition and differences between gyratory and

kneading specimens are small (Table 4.9).

Comparison of the effects of mixture and test variables on resistance to fatigue (Table

4.10) reveals several inconsistences. Flexural testing evaluates the effects of asphalt type and

asphalt content differently than diametral testing. Because AAG-1 mixtures are stiffer than

AAK-1 mixtures at temperatures used in the fatigue testing, the diametral testing seems to

provide the most reliable indication of asphalt effects_ Based on similar reasoning, the flexural

testing seems to provide the most reliable indication of asphalt-content effects. The indicated

superiority of chert, rolling-wheel specimens under flexural fatigue is obviously a spurious test

fnding. Additionally, the apparently inaccurate temperature effect in the diametral testing of

rolling-wheel specimens is a consequence of the coupling of stresses and temperatures for this

analysis. That fatigue resistance appeared superior at the high temperature merely reflects the

use of much lower stresses than at the low temperature level.

4.1.4 Stiffness Modulus. Although stress, loading, and temperature conditions are quite

different, each of the four test procedures provides a measure of mixture stiffness. At short

loading times, such a measure approximates mixture response to transient traffic loading in situ

and becomes an important variable in the pavement performance equation. For thick pavements

characteristic of heavy-duty highways, stiff mixtures are thought to be essential for acceptable

performance. Consideration of stiffness effects is, thus, an essential component of the evaluation

of laboratory compaction procedures.

The main experiment produced four different stiffness measures corresponding to 0. l-

second loading, namely, a compressive modulus (compressive creep), a shear modulus (shear

creep), a flexural modulus (flexural fatigue), and an indirect tensile modulus (diametral fatigue).

Average (geometric) values of these moduli have been summarized in Tables 4.1, 4.2, 4.6, and

67

Table 4.9 Effect of Compaction Method on Sensitivity of FatigueResponse to Mixture Variables

Coefficient of Variation of Number of Application. to FailureTest (PercenO

Stress Temperature Method

Gyratory [ Kneading, Rolling Wheel

Flexurai N/A 82.2 95.2Low

Diametrid 146.6 138.3 172.1Low

Flex-arid N/A 120.3 123.7_gh

Diametral 85.8 120.5 156.9

Flexurid N/A 95.0 99.1Low

Diametrid 154.9 151.9 165.0I_gh

Flexurid N/A _.4 83.7_gh

Diametral 84.1 89.9 135.1

68

Table 4.10 Effect of Mixture and Test Variables on Resistance to Fatigue

Compaction MethodVariable Test I

Gyratory I Kneading Rolling Wheel

Flexural fatigue N/A AAK-I • AAG-I AAK-1 • AAG-IAsphalt Type

Diametral fatigue AAG-I • AAK-1 AAG-1 • AAK-1 AAG-I • AAK-1

Asphalt Flexural fatigue N/A Optimum • High Optimum • HighContent

Diametral fatigue Optimum • High No Difference High • Optimum

Flexural fatigue N/A Granite (RB) • Chert (P..L) •Aggregate Chert (RL) Granite (RB)

Type Diametral fatigue Granite (RB) • Granite (RB) • Granite (RB) •Chert (RL) Chert (RL) Chert (RL)

Air Voids Flexural fatigue N/A Low • High Low • HighContent

Diametral fatigue Low • High Low • High Low • High

Flexural fatigue N/A Low • High Low • HighTemperature

Diametral fatigue Low • High Low • High High • Low

Fiexural fatigue N/A LOw • High Low • HighStress

Diametral Low • High Low • High Low • High

69

4.7, respectively. Although the moduli differ markedlyin magnitude--areflection primarilyof

differencesin test temperatureand in the type and level of stress-and are, therefore, notdirectly

comparable, they do permit similarcomparativeevaluationsof compactionand other significant

variables.

The effect of compaction method on mixturestiffness is summarized in Table 4.11, and

Table 4.12 compares compaction methods based on sensitivity of stiffness response to mixture

composition. In part because stiffness was notparticularlysensitive to compaction method-at

least not when compared to the sensitivity of either permanent-deformation or fatigue

measures-evidence is mixed about which, if any, compaction method produces the stiffest

mixes. Informed judgment seems to first requirean appraisalof the reliability of the different

moduli, a task aided by Table 4.13. As summarizedin Table 4.13, flexural moduli violate a

priori notions about the effects of asphalt types and indirect tensile moduli from diametral

fatigue tests violate a priori notions about the effects of aggregate type (gyratoryand kneading).

Compressive moduli from creep tests violate a priori notions about the effects of stress level.

Shear creep dataare corroborative throughout. On the basis of such reasoning, the conclusion

from the shear testing is that the stiffest mixtures are produced by rolling-wheel compaction,

followed in order by kneading and gyratory compaction. The stiffness of rolling-wheel

specimens reflects more sensitivity to mixture composition than either gyratory or kneading

specimens but kneading specimens are a close second.

t'The effect of asphalt type on mixture stiffness is temperature dependent. At highertemperatures in the range of 104 to 140°F (40 to 60°C), the AAK-1 asphalt is more viscous thanthe AAG-1 asphalt, and mixtures containing AAK-1 are stiffer. The reverse is true at lowertemperatures in the range of 32 to 68°F (0 to 20°C). This temperature-dependent stiffness effectof these two asphalts has been confirmed in later phases of the SHRP A-003A testing program.

70

Table 4.11 Effect of Compaction Method on Stiffness

| ,

Ranking in the Twelve Mixture and Test VariableCompaction Overall Comparisons of Tables 4.1, 4.2, 4.6, and 4.7Method Test Stiffness

Stiffest Intermediate Weakest

Compressive Creep Weakest None None 12 Times

ShearCreep Weakest 2Times 3.5Times 6.5TimesGyratory

Flexure NIA NIA NIA NIA

Indirect Tension Stiffest 5.3 Times 4.3 Times 2.3 Times

Compressive Creep Stiffest 10 Times 2Times None

Shear Creep .. Intermediate 2 Times 5.5 Times 4.5 TimesKneading

Flexure Intermediate None 12 Times N/A

Indirect Tension Intermediate 3.3 Times 6.8 Times 1.8 Times

Compressive Creep Intermediate 2 Times 10 Times None

Rolling Shear Creep Stiffest 8 Times 3 Times 1 TimeWheel

Flexure Stiffest 12 Times None N/A

Indirect Tension Weakest 3.3 Times 0.8 Times 7.8 Times

D

71

Table 4.12 Effect of Compaction Method on Sensitivity of StiffnessResponse to Mixture Variables

i n

Coefficient of Vsur_ien of Stiffness Modulus (Percent)Stress Temper- Test Method

ature GyrIJLory Kneading Roiling Wheel

Compression 4,1.8 42.5 40.3

Shear 27.8 33.5 63.6Low

Flexure N/A _4.a.]7 32.8

Indirect Tension $7.5 44.8 44.7Low

Compression 56.0 $9.2 50.2

Shear 53.6 49.9 81.4High-

Flexure N/A 50.2 F7.2

Indirect Tension 53.1 49.8 $3.5

Compression 37.5 38.2 37.4

Shear 38.5 48.3 $0.9LOw

Flexure N/A 42.2 42.2

Indirect Tension 66.3 54.2 49.8High ,

Compression 64.3 64.4 51.8

Shear 43.6 _;1.1 50.2High

Flexure N/A 36.2 35.2

Indirect Tension 60.7 77.7 55.5

72

Table 4.13 Effect of Mixture and Test Variables onStiffness Modulus

D Compaction MethodVariable Test

Gyrator/ [ Kneading Roiling Wheel

Compressive creep AAK-1 > AAG-1 AAK-! > AAG-1 AAK-I • AAG-I

Shear creep AAK-1 • AAG-I AAK-I > AAG-I AAK-1 • AAG-IAsphalt Type

Flexural fatigue N/A AAK-I • AAG-I AAK-1 • AAG-1

Diametral fatigue AAG-I • AAK-1 AAG-1 • AAK-1 AAG-1 • AAK-I

Compressive creep Optimum • High Optimum • High Optimum • High

Asphalt Shear creep Optimum • High Optimum • High Optimum • HighContent

Flexural fatigue N/A Optimum • High Optimum • High

Diametral fatigue Optimum • High Optimum • High Optimum • High

Compressive creep Granite (RB) • Granite (]_B) • Granite (RB) •Chert ('RL) Chert (RL) Chert (R.L)

Shear creep Granite (P,.B)• Granite (RB) • Granite (RB) •

Aggregate Chert (R.L) Chert 0_,L) Chert (RJ..)

Type Flexural fatigue N/A Granite (RB) • Granite (RB) •Chert (R.U) Chert (]U.)

Diametral fatigue Chert (RL) • No effect Granite (P..8) •Granite (RB) Chert (RL)

Compressive creep Low • High Low • High Low • High

Air Void Shear creep Low • High Low • High Low • HighContent

Flexural fatigue N/A Low • High Low • High

Diametral fatigue Low > High Low • High Low • High

Compressive creep LOw • High LOw • High LOw • High

Shear creep Low • High Low • High LOw • HighTemperature

Flexural fatigue N/A Low • High LOw • High

Diametral fatigue LOw • High Low • High Low • High

Compressive creep No effect High • LOw High > Low

Shear creep Low • High Low • High Low • HighStress

Flexural fatigue N/A Low • High LOw • High

Diametral fatigue LOw • High Low • High LOw >. High

73

4.1.5 D_ussion. As noted earlier, the effects of compaction method on mixture properties

were larger than had been expected initially. Differences between gyratory and kneadin

specimens were most pronounced and of greatest initial interest. Because compaction was

completed much quicker in the gyratorycompactor than in the kneadingcompactor, particularly

for the low air-void mixturesincorporatinggranite(RB), there was concern that the asphalt may

have hardened more in kneading specimens than in gyratory specimens. To investigate this

possibility, viscosities were measuredfor asphalts extracted from specimens preparedby each

method. For neither asphalt was viscosity'greater for kneading than for gyratory compaction

Gable 4.14). Figures 4.19 and 4.20 comparethe recovered asphalt propertieswith those of neat

asphalts, determined in accord with procedures for asphalt heating and transfer to small

containersand supplied by the A-001 contractor. It will be noted that the hardening occurring

in the mixing and compaction processes results in an approximately one-grade increase in

viscosity. Also it will be noted that the degree of hardening of the asphalts is approximatelythe

same for both methods of compaction even though the kneading specimens requireda longer

time period to compact. Thus, the effects of compaction method on mixture properties cannot

be attributedto differential hardeningof the asphalt binders.

Although both gyrator3,andkneadingcompactorsinduce "shearing"displacementsduring

compaction, they appear to differ fundamentallyin two important ways. First, the gyratory

compactor subjects the confined mixtureto a significant "hydrostatic"pressure. Following the

soil mechanicsanalogy, this pressurehas two components,pore pressurewithin the fluid (asphalt

and air voids) phase and effective pressure within the granularphase. Large pore pressures,

•" expected with mixtures having a large percentage of their mineral voids filled with asphalt,

reduce the effective pressuresand, hence, the extent to which aggregateparticles are forcedinto

intimate contact. In kneading compaction, pore pressuresare more readily dissipated because

74

Table 4.14 Effect of Compaction Method on Hardening of Asphalts

Average Viscosity of Recovered Asphalt, Poises

Asphalt Type 140"F (60"C) 2750F (135"C)

Gyratory Kneading Gyratory Kneading

AAK-1 9,130 8,300 9.84 9.38

AAG-1 3,740 3,480 3.32 3.20

Note: Asphalts were exwacted from mixturescontaininggranite (RB) and compactedat optimalasphalt content to low air voids;-1 poise = 0.1 Pa*s.

75

VISCOSITY vs HEATING STAGE - AC30

10000

lOOOVZSC0S

I • _0 _ iT

100y []275 F

I

pO

is

e

s10

Ist 2nd Heating Recovered - Recovered -

Heating Kn Gy

Figure 4.19 Effect of Heating and Mixing on Hardeningof AAK-1 Asphalt (1 poise = 0.1 Pa.s)

76

VISCOSITY vs HEATING STAGE - AR4000

lOOOVISC0

D °7T • 1_o F

100 ----y i _275 F

Ip Io fi ise

slo

i

I

ti j j

ist Heating 2nd Heating Recovered - Recovered -

Kn Gy

Figure 4.20 Effect of Heating and Mixing on Hardeningof AAG-I Asphalt (l poise = 0.1 Pa.s)

the mixture is unconfined during compaction. Because of the pore-pressure effect, more

interparticle contact is expected in kneading specimens, especially when more of the voids

in the mineral aggregate are filled with asphalt.

The second major difference between gyratory and kneading compactors is in the

nature of the interparticle movement during compaction. The back-and-forth movement

of the gyratory compactor aligns the larger particles but does little to promote an

interlocking, edge-to-face structure. Kneading compaction, on the other hand, induces

larger, more random shearing displacements. The result is likely to be a more tightly

interlocked and stable aggregate structure in kneading specimens.

In terms of the main compaction experiment, these differences are expected to have

the following manifestations:

• Differences between the physical properties of gyratory and kneading

specimens are expected to be greater when more of the aggregate voids are

filled with asphalt, that is, with larger asphalt contents and smaller air-void

contents;

• Kneading specimens are expected to be more sensitive than gyratory

specimens to aggregate properties, that is, aggregate type; and

• Gyratory specimens are expected to be more sensitive than kneading

specimens to asphalt type.

Using data developed by the main compaction experiment, an evaluation of the above

"hypotheses" is summarized in Table 4.15. The results support all but one of these

hypotheses and strengthen the notion that the primary differences between gyratory and

kneading compaction derive from differences in aggregate structure and interparticle

contact.

Another possible difference between compaction methods is in the configuration or

pattern of internal air voids. Most of the air entrapped in the loose mixture is expelled

78

Table 4.15 Experimental Validation of Hypotheses RegardingCompaction Methodologies

Permanent Deformation Stiffness Modulus

Hypothesis DiametralCompression Shear Fatigue Compressive Shear Indirect

Creep Creep Tension

The difference between

gyrato_ and kneading No No No No No Yescompaction is greater at higherasphalt contents.

The difference between

gyratory and kneading Yes Yes No Yes Yes Yescompaction is greater at lowerair voids.

Kneading specimens are moresensitive to aggregate type than Yes Yes Yes Yes Yes Yesare gyrator), specimens.

Gyratory specimens are moresensitive to asphalt type than Yes Yes No Yes Yes Yesare kneading specimens.

D

79

during the compaction process. Under the compaction pressure, individual pockets of air

migrate to a nearby free surface. The channels thus formed axe altered by subsequent

compaction: most doubtlessly become closed entrapping small pockets of air within the

mixture. Of interest to the current study is whether compaction :method influences the

configuration of internal air voids that remain and, ff so, whether the air-void pattern is

likely to affect mixture properties of interest. It does seem likely that kneading compaction

would more thoroughly disrupt internal air channels than would gyratory compaction or,

stated alternatively, that internal air capillaries would be longer az_dmore continuous in

gyratory specimens.

To test this hypothesis, water permeability measurements were made on eleven,

2.5 x 4-inch (63.5 x 101.6 turn) briquets, six prepared by gyratory compaction and five by

kneading compaction. Although not conclusive, results of this testing suggest that gyratory

specimens may be slightly more permeable than kneading specimens (Figure 4.21). Such

a finding is consistent with the hypothesis of greater interconnectivity of the internal air

capillaries within gyratoryspecimens. Differences in the internal-air-void pattern could have

obvious impact on mixture aging and perhaps on other mixture properties as well.

Finally, the condition of the specimen surface (either as-molded or cut) is a

confounding variable of unknown significance. The main compaction experiment was based

on the assumption that surface condition does not affect fundamental measurements.

Accordingly, as shown in the table below, isolating effects of compaction method from

possible confounding effects of surface condition is impossible.

8O

Z C_ Z

>" Z >--

I• o I

ii

C,O

m

(o_s/_o6-_ot)alm_v_d

Figure 4.21 Effect of Compaction Method on Permeability

81

Surface Condition as Function of Compaction Method and Test Type

I

Test Type Compaction MethodGyratory Kneading Rolling Wheel

Compressive Creep Cored As-Molded Cored

Shear Creep As-Molded As-Molded Cored

Flexural Fatigue N/A Sawed Sawed

Diametral Fatigue As-Molded As-Molded Cored

Further discussion of the possible effects of cut surfaces is included in Section 5.0.

4.1.6 Statistical Si2nific,ane¢.

The primary purpose of the main compaction experiment was to determine if the method

of specimen compaction has any effect on fundamental properties of asphalt-aggregate mixtures

considered important to pavement performance. The three compaction methods examined in the

main experiment were gyratory, kneading, and rolling wheel. The several test procedures

included compressive creep, shear creep, flexural fatigue, and diametral fatigue. Both creep

studies were performed under steady load conditions (as opposed to repeated load applications).

The other factors included in the main compaction experiment were asphalt type, asphalt

content, aggregate type, air-void content, test temperature, and stress level. By looking at such

a large number of factors, the hypothesis of "no compaction method effect" is tested more

broadly, thus giving greater credibility to any finding of a compaction-method effect or a

no-compaction-method effect.

Ex_eriment Design for Main Ex_riment

The experiment design for the compressive and shear creep tests was a one-half fraction

of a 2-level factorial design in the first five variables listed above (a 25"1 design), repeated at

each of two stress levels, with all this run on specimens prepared by each of the three

82

compaction methods. Thus, the data for each of the creep tests come from a 2s'l • 2 • 3

experiment.

For the flexural fatigue tests, only two compaction methods could be studied, as the

gyratoryequipment could not preparespecimens of the necessary size. Here the design was a

one-half fractionof a 26 factorial (includingthe two compaction methods) run at each of two

stress levels, for a 26"1• 2 experiment. A similar designwas used for the diametralfatigue tests

but all three compaction methods were investigated.

In both creep and fatigue tests, a number of replicates were run to provide greater

precision in the test results and also to provide an independentestimate of the test error. All

four sets of tests incorporate a basic matrix of 16 mixtures, representing all 16 different

combinations of the two levels of each of the mixturefactors; asphalt type, asphalt content,

aggregate type, and air voids. These designs are described in greater detail in Section 3.0.

StatisticalAnalysis

The statistical calculations were carried out using the SAS (R) system, Version 6.04 for

PCs. The main procedure was the General Linear Model (GLM), a least-squares program which

permits variables of both a CLASS nature (for example, asphalt type of AAK-1 or AAG-1) as

well as numerically valued variables (for example, air voids of 4.5 percent, 11.5 percent.... _.

The analysis technique is described more completely in Appendix C. The analysis of the

correlations between the compaction methods was done with the SAS procedure CORR.

Analysis of Creep T¢_tData

In the compressive and shear creep tests, two responses were studied, both derived from

the basic measurements of specimen modulus as a function of test time:

I. Initial creep modulus (i.e., at 0.1 sec) and

83

2. Time (seconds) to reacha criticallevel of strain, 0.008 in/in (0.008 mm/mm) for

compressive creep and 0.02 in/in (0.02 mm/mm) for shear creep.

The logarithms of these two responses were analyzed using the generalized linear model

formulation.

Table 4.16 identifies those effects found to be statistically significant at the 95-percent

probability level in the compressive-creep testing. In this table, a "Yes" indicates that the effect

was found to be significant at a 95-percent or greater probability level. That is, there is a 5-

percent or smaller chance that the observed effect could have come from a situation in which

there is really no effect. For each response (for example, time to terminal strain), there are two

columns. The first, labeled "Significant Effect," denotes whether the listed effect is significant

when averaged over the several compaction methods: the second, labeled "Significant

Compaction Interaction," indicates whether the effect of one or moreof the compaction methods

is significantly different from the average.

From Table 4.17, the summary statistics for the compressive creep testing are:

Response

Statistic Time to Terminal Strain . Initial Creep Modulus(In sec) On psi)

R2 0.951 0.972

RMSE 1.577 0.213

CV, % 332.0 21.6,

Note: R 2 --- Coefficient of determination; RMSE = Root mean squared error; CV =Coefficient of variation; 1 psi = 6.89 kPa.

The compressive creep responses for the three compaction methods were also compared

by correlating each response from each compaction method against its counterpart on the other

compaction methods. This was done using the ordinary Pearson linear correlation (on

logarithms of time or modulus) and also using the rank correlation method of Spearman. In all

84

Table 4.16 Statistically Significant Effects in Compressive Creep Testing

Time to TerminalStrain Creep Modulus at O.I Sec

Effect Significant Significant Significant SignificantEffect Compaction Effect Compaction

Interaction Interaction

Intercept Yes Yes Yes Yes

Asphalt Type (Asph) Yes Yes

Aggregate Type (Aggr) Yes Yes Yes

Asphalt Content (% Asph)- " Yes Yes Yes

Air Voids (% Voids) Yes Yes Yes

Temperature(Temp) Yes Yes

Stress Yes

Asph x Aggr Yes

Asph x % Asph Yes

Asph x % Voids Yes

Asph x Temp Yes

Asph x Stress Yes

Aggr x % Asph Yes Yes

Aggr x % Voids Yes Yes Yes

Aggr x Temp Yes

Aggr x Stress

% Asph x % Voids Yes

% Asph x Temp Yes

% Asph x Stress

% Voids x Temp Yes Yes

% Voids x Stress

Temp x Stress

85

Table 4.17 Snmmary Statistics for Main Experiment

Compressive Creep Shear Creep Flexural Fatigue Diametral FatigueSta-tistic Terminal Initial Terminal Initial Fatigue Initial Fatigue Initial

Time Modulus Time Modulus Life Modulus Life Modulus

(In sec) On psi) Onsee) (In psi) On) Onpsi) On) Onpsi)

R2 0.951 0.972 0.944 0.983 0.961 0.970 0.85 ! 0.753

RMSE 1.577 0.213 1.141 0.221 0.469 0.166 1.025 0.354

CV, % 332.0 21.6 164.0 22.4 49.6 16.7 136.0 36.6

Note: R2 = Coefficient of determination;RMSE - Root mean squarederror = VMSE; CV- Coefficient of variation _ 100V(eMSE-I);1 psi - 6.89 kPa.

86

cases, the correlations were significant at the 99.99-percent level or higher. The correlation

D coefficients were found to be as follows:i i ,, i, ,i

Response Compaction Methods Ordinary RankPearson Spearman

Gyratory/Kneading 0.846 0.910Time to

Terminal Strain Gyratory/RoUing Wheel 0.840 0.896On _c)

Kneading/Rolling Wheel 0.849 0.938i

Initial Gyratory/Kneading 0.937 0.915

Creep Modulus Gyratory/Rollin.g Wheel 0.952 0.924(In psi)

Kneading/Rolling Wheel 0.958 0.943

This analysis suggests that, although the responses of specimens prepared by different

compaction methods may be considerably different, the effect of compaction method on the

ranking of different mixture (and testing) conditions is much smaller.

As with the compressive creep tests, the shear creep data were obtained in the form of

creep modulus vs. time, and the analysis was done in terms of the logarithm of the initial

modulus (at 0.1 sec) and the logarithm of the time at which the strain reached a terminal value.

Table 4.18 identifies those effects found to be significant at the 95-percent level,

including the interactions of compaction method with other effects. For both compressive creep

(Table 4.16) and shear creep (Table 4.18), the effect of compaction method on permanent-

deformation resistance was statistically significant. A few of the interactions of compaction

method with other factors were also statistically significant. Of particular importance are the

significant interactions of (1) compaction method and air voids and (2) compaction method,

aggregate type, and air voids. These interactions give credence to the notion that compaction

effects become more pronounced at low air-void contents and are largely due to compaction's

effects on aggregate structure and interparticle contact. Tables 4.16 and 4.1_ also show that

87

Table 4.18 Statistically Significant Effects in Shear Creep Testing

Time to Terminal Strain Creep Modulus at 0.1

Effect Significant Significant SignificantEffect Compaction SignificantEffect Compaction


Intercept Yes Yes Yes


Aggregate Type (Aggr) Yes Yes

Asphalt Content (% Asph) Yes Yes Yes

Air Voids (% Voids) Yes .... Yes Yes Yes

Temperature (Temp) Yes Yes

Stress Yes Yes

Asph x Aggr Yes

Asph x % Asph

Asph x % Voids

Asph x Temp Yes

Asph x Stress

Aggr x % Asph

Aggr x % Voids Yes Yes

Aggr x Temp

Aggr x Stress

% Asph x % Voids Yes

% Asph x Temp

% Asph x Stress

% Voids x Temp Yes

% Voids x Stress Yes Yes

Temp x Stress

88

effects of all other main factorswere, with minor exception, statistically significantin terms of

both permanent-deformationresistanceand stiffness.

The summary statistics from Table 4.17 for shear creep are as follows:

i1

Response

Statistic Time to TerminalStrain Initial Creep ModulusOn sec) On psi)

i

R2 0.944 0.983

RMSE 1.141 0.221

CV, % 164.0 22.4

Note: 1 psi = 6.89 kPa

These are quite in line with the compressive creep results, and again indicate that the specimens

responded to the experimental factors and that the results are reasonably precise. As indicated

D by the coefficient of variation, there was less residual error in the shear creep testing than in thecompressive creep testing.

The shear creep responses for the three compaction methods were also compared by

correlating each response from each compaction method against its counterpart on the other


logarithms of time or modulus) and also using the rank correlation method of Spearman. In all

eases, the correlations were significant at the 99.99-percent level or higher. The correlation

coefficients were found to be as follows:

89

Ordinary RankResponse CompactionMethods Pearson Spearman

Gyratory/Kneading 0.831 0.871Timeto

Terminal Strain Gyratory/RollingWheel 0.896 0.932sec)

Kneading/Rolling Wheel 0.900 0.912

Initial Oyratory/Kneading 0.960 0.949CreepModulus GyratorylRollingWheel 0.949 0.961

(Inpsi) "Kneading/RollingWheel 0.960 0.963

Analysis of Fatigue Test Data

In the fatigue tests, two responses were studied:

1. Initial specimen stiffness and

2. Fatigue life (cycles to failure).

The logarithms of these two responses were analyzed using effectively the same models as were

used to analyze the compressive and shear creep data.

For flexural testing, data were obtained only for kneading and rolling-wheel compaction,

as the gyrator), compactor could not make specimens of the size required by the flexural fatigue

apparatus. Effects, including the interactions of compaction method with other effects, found

to be significant at the 95-percent level are noted in Table 4.19.

The summary statistics for the flcxural fatigue data, excerpted from Table 4.17, show

high correlations for both stiffness and fatigue life. Even the 50-percent (coefficient of

variation) precision for fatigue life is very good. As mentioned elsewhere, data previously

reported in the literature have shown coefficients of variation of approximately 60 percent.

9O

Table 4.19 Statistically Significant Effects in Fiexeral Fatigue Testing

Fatigue Life Flexural Stiffness at 0.1 Sec

SignificantEffect Significant Significant Significant CompactionEffect Compaction Effect


Intercept Yes Yes Yes


Aggregate Type (Aggr) Yes Yes Yes

Asphalt Content (% Asph) .... Yes Yes Yes

Air Voids (% Voids) Yes Yes

Temperature O'emp) Yes Yes Yes

Stress Yes Yes

Asph x Aggr Yes Yes

Asph x % Asph Yes Yes

Asph x % Voids

D Asph x TempAsph x Stress

Aggr x % Asph Yes Yes

Aggr x % Voids Yes Yes

Aggr x Temp

Aggr x Stress Yes

% Asph x % Voids

% Asph x Temp Yes Yes

% Asph x Stress

% Voids x Temp

% Voids x Stress

Temp x Stress Yes

P

91

Response

Statistic Fatigue Life Initial Flexural ModulusOn applications to failure) (In psi)

R2 0.961 0.970

RMSE 0.469 0.166

CV, % 49.6 16.7

Note: 1 psi - 6.89 kPa

The flexural-fatigue responses for the two compaction .methods were also. comparedby

correlating each response from each compaction method against its counterpart on the other

compaction method. This was cloneusing the ordinaryPearson linear correlation (on logarithms

of life or modulus) and also using the rank correlation method of Spearman. In all cases the

correlations were significant at the 99.99-percent level or higher. The correlation coefficients

were found to be as follows:

Ordinary RankResponse Compaction Methods Pearson Spearman

Gyratory/Kneading Not applicable Not applicableFatigue Life

(In) Gyratory/Rolling Wheel Not applicable Not applicableKneading/Rolling Wheel 0.888 0.843

Initial Gyratory/Kneading Not applicable Not applicable

Flexural Gyratory/Rolling Wheel Not applicable Not applicableModulusfin psi) Kneading/Rolling Wheel 0.973 0.960

92

The diametral-fatigue testing included specimens prepared using all three of the

compaction methods. Table 4.20 identifies those effects found to be statistically significant at

the 95-percent probabilitylevel in the diametral fatigue testing. The primarycompaction effects

on both fatigue resistance and stiffness, identified above, were generally not found to be

statistically significant either in fiexural (Table 4.19) or diametral (Table 4.20) testing. Effects

of all other primaryfactorswere found to be statistically significant in flexural fatigue, and most

were significant in diametral fatigue as well. Compaction method did have a few statistically

significant interactions with other factors, primarilyaggregate type and temperature.

The summary statistics from Table 4.17 for diametral fatigue are as follows:

Response

Statistic Fatigue Life Initial Tensile ModulusOnapplications to failure) On psi)

R2 0.851 0.753

RMSE 1.025 0.354

CV, % 136.0 36.6

Note: 1 psi - 6.89 kPa

Comparing these summary statistics with those from the flexural-fatigue experiments reveals a

better fit of the regression model to the flexural-fatigue data as well as significantly less

variability of the flexural-fatigue testing both for fatigue life and for stiffness.

The diametral-fatigue responses for the three compaction methods were also compared

by correlating each response from each compaction method against its counterpart on the other


logarithms of fatigue life or modulus) and also using the rank correlation method of Spearman.

In all cases, the correlations were significant at the 99.99-percent level or higher. The

correlation coefficients were found to be as follows:

93

Table 4.20 Statistically Significant Effectsin Diametral Fatigue Testing

i

Fatigue Life Tensile Stiffness at 0.1 Sec

Factor Compaction CompactionEffect EffectInteraction Interaction

Intercept Yes Yes


Aggregate Type (Aggr) Yes Yes

Asphalt Content (% Asph) Yes

Air Voids (% Voids) Yes Yes.

Temperature (Temp) Yes Yes

Stress Yes Yes

Asph x Aggr Yes

Asph x % Asph Yes

Asph x % Voids

Asph x Temp Yes

Asph x Stress

Aggr x % Asph

Aggr x % Voids Yes

Aggr x Temp Yes

Aggr x Stress

% Asph x % Voids

% Asph x Temp

% Asph x Stress

% Voids x Temp Yes Yes

% Voids x Stress

Temp x Stress Yes

94

i

Response Compaction Methods Ordinary RankPearson Spearman

Gyratory/Kneading 0.949 0.923Fatigue Life

On) Gyratory/RoUing Wheel 0.824 0.807Kneading/Rolling Wheel 0.868 0.852

Initial Gyratory/Kneading 0.913 0.877Tensile

Modulus Gyratory/Rolling Wheel 0.842 0.736

(ln psi) Kneading/Rolling Wheel 0.831 0.740

As with the other response measures evaluated herein, the relatively large magnitude of these

correlation coefficients means that, overall, the rankings of test conditions and mixtures are not

greatly affected by compaction method.

Fatigue Life - Stiffness Relationship

In the course of examining the fatigue data, double-logarithmic plots (Figures 4.15-4.18)

of fatigue life vs. stiffness showed interesting patterns in the data. To investigate the possibility

of developing a prediction relationship for fatigue life based primarily on the initial observed

stiffness, regression analysis of fatigue life (logarithm) against stiffness (logarithm) for the

flexural tests showed that virtually all the other variables were still required in order to obtain

a residual error as small as when stiffness was not included as a regressor variable. Although

stiffness seemed to group the data into several distinct lines, the scatter about those lines was

still large compared to the ultimate residual error when fatigue life is correlated against the

experimental factors (root mean squared error of 0.47 for flexural fatigue). Thus, a simple

correlation of life against stiffness and just a few other effects does not do justice to the data.

95

Summary_of Findings

The statistical analyses described herein found several statistically significant differences

among the three compaction methods, both as regards the average values of the response

variables and the effects of some of the other factors (interactions of these other factors with

compaction method). Tables 4.16, 4.18, 4.19, and 4.20 identify the significant effects for the

compressive creep, shearcreep, flexural-fatigue,and diametral-fatigueexperiments, respectively,

and Table 4.17 lists the summary statistics for each of the analyses.

Overall, the experimental factors accounted for most of the information (variability) in

the clam. This is reflected in large coefficients of determination, ranging from a low of 0.75 to

a high of 0.98. This means that from 75 to 98 percent of the variation in the raw data is

"explained" by the experimental factors (asphalt type, aggregate type, etc.). Excluding the

diametral testing, the smallest coefficient of determination was a very respectful 0.94.

In addition, the residual (unexplained) variation in the modulus (stiffness) responses

reasonably low on an absolute basis, ranging from 1"/-percentcoefficient of variation on

stiffness to 37-percent on tensile modulus in indirect tension. The flexural-fatigue life residual

error, again expressed as the coefficient of variation, was 50 percent, a quite good (low) level

for fatigue testing7. For diametral fatigue, the residual error, 136 percent, was considerably

greater. The residual errors in the compressive and shear creep "times to terminal strain," while

appearing to be inordinately large, are inflated by the fact that a few specimens failed very early

and so were arbitrarily assigned a time to terminal strain of 0.1 second. On the other hand, a

fair number of specimens were so strong that they had not reached the terminal strain at the end

of testing (3,600 seconds). Times to terminal strain for these specimens were extrapolated,

7The fatigue experiment was designed and sized based upon a coefficient of variationpercent as estimated from fatigue data previously reported in the literature.

96

using a double logarithmic relation between strain and time. The residual error was considerably

less for shear creep tests than for compression creep tests.

Despite the fact that compaction method was generally found to influence mixture

properties, mixtures (and test conditions) were ranked in much the same way irrespective of

method of compaction for all the test responses evaluated herein.

4.2 F.,xtended Permanent-Deformation Study

Objectives of the extended permanent-deformation study included further exploration and

documentation of the interactive effects of air voids and compaction method on specimen

response to creep loading, evaluation of the effects of cyclic loading on the accumulation of

permanent deformation, and evaluation of the effects of increasing the constant axial load on

specimens being tested in shear. Both gyratory and kneading compaction were investigated.

The testing program, utilizing a single mixture and a single temperature, is described as follows:

Number of Tests Air-VoidTest Stress Content

Gyratory Kneading

Compressive Creep 10 I0 30 psi (206.7 kPa), axial, static Variable (3-9%)

Shear Creep 10 10 3.3 psi (22.7 kPa), shear, static Variable (3-9_)5.0 psi (34.4 kPa), axial, static

Compressive 3 3 30 psi (206.7 kPa), axial, cyclic LowCyclic (I Hz, 0.l-sec load pulse)

Shear Cyclic 3.3 psi (22.7 kPa), shear, cyclic2 2 (1 Hz, 0.1-sec load pulse) Low

5.0 psi (34.4 kPa), axial, static

Common Elements AAG-1 asphalt at optimum asphalt content with granite (RB). All testing performed at140*F (60°C).

Compressive creep curves for individual specimens of varying air-void content are

depicted in Figures 4.22 and 4.23 for gyratory and kneading compaction, respectively. Similar

curves for shear creep are shown in Figures 4.24 and 4.25. The effect of air-void content on

the resistance of these mixtures to creep loads is clearly evident: more air voids translate to

97

Cr_ _'3 r'-- _ _ _ _ O'_ CAD Cr_

Figure 4.22 Compressive Creep Curves for Specimens with Varying Air Voids,Gyratory Compaction (1 psi = 6.89 kPa)

98

Figure 4.23 Compressive Creep Curves for Specimens with Varying Air Voids,Kneading Compaction (1 psi = 6.89 kPa)

99

Figure 4.24 Shear Creep Curves for Specimens with Varying Air Voids,Gyratory Compaction (1 psi = 6.89 kPa)

100

q,==_

E---

Figure 4.25 Shear Creep Curves for Specimens with Varying Air Voids,Kneading Compaction (1 psi = 6.89 kPa)

101

weaker mixtures. The effect of air voids under compressive loading is indicated more directly

in Figures 4.26 and 4.27. At a given time of loading, the compressive creep modulus decreases

with increasing air voids: the relationship between In modulus and air-void content seems to be

approximately linear except within the region where collapse is imminent. Figure 4.28

illustrates the combined effects of air voids and loading time for kneading specimens and shows

that air-void effects on compressive creep modulus are not much smaller in magnitude than

time-of-loading effects. The extended permanent-deformation study also validated the superior

resistance of kneading specimens to permanent deformation both for specimens with low air

voids (Figures 4.29 and 4.30) and for those with higher air voids (Figures 4.31 and 4.32).

In the main compaction experiment, specimens being tested in shear were subjected to

a normal (axial) pressure of 2.5 psi (17.2 kPa). To examine whether the compaction effect in

shear creep might be significantly altered by larger pressures, all shear testing in the extended

permanent-deformation study incorporated a static axial pressure of 5 psi (34.4 kPa).

summarized in Figures 4.30 and 4.32, indicate that, in shear creep, kneading specimens remain

more resistant to permanent deformation than gyratory specimens after the imposition of an

added normal stress increment.

Under shear creep loading, asphalt-aggregate mixtures expand or dilate axially. To test

whether the dilational behavior might be sensitive to compaction method, instrumentation for

recording axial as well as shear strains was added to a few of the shear creep tests. Figure 4.33,

tracing the ratio of shear to axial strain as a function of time, demonstrates definite differences

•-_ between compaction methods. For a given level of shear strain, kneading specimens dilatedtx

more than gyratory specimens (Figure 4.34). This is the kind of behavior expected of mixtures

having a tightly interlocked aggregate structure where shear strains force the angular particles

to rotate. That kneading specimens dilate more than gyratory specimens under shear loadir

102

................................................................i -T

i

/

i ///

m ...........................

Figure 4.26 Effect of Air Voids on Compressive Creep Modulus,Gyratory Compaction (1 psi = 6.89 kPa)

103

i.... Z'-

Z"..-

r_

/Ii ,.,._JJi

HIh

............. I

_'_ r.f_

r,,,..,

Figure 4.27 Effect of Air Voids on Compressive Creep Modulus,Kneading Compaction (I psi = 6.89 kPa)

I04

Figure 4.28 Combined Effects of Air Voids and Loading Time on CompressiveCreep Modulus, Kneading Compaction (1 psi = 6.89 kPa)

105

IV

i ¢

..... _-"T '_'" _ " ": .......................................... r"_-o'"_""r-"-'r ............. t ............................ i

i........4-.-!-....i.......!.........-. _: ! _---:..--_.---i.........-................: . : :

i : i " " ' :

-_-!::"ii: ......._..........q ..............._......................: _ : .................+--.i.._,,.,-_......_::. ! : ... . :

..............._.............. ............i _;::_ ;:;::::_::;:;; .................. _........

:::::::::::::::::::::::::::::::::::'_:::-T"'-"'""""."'"_"""+" :.::::: ' ===========================:: "";; "'_-" " _

_..--.,---i-._...........,.......-.... ..........i!!!!!!):i:!!!:.:!!!!!!!!!!!!! : ; :• .:,... _

::::::::::::::::::::::.....i.........- !?i.!:!sF!;!!:!!:!!.![= ._....................:==.- ._ E--

.......... _ ........ _.. : ..................................

r

...,......_....i. : ": : : _ : . • . .

Figure 4.29 Effect of Compaction Method on Compressive Creep Modulus of Low tAir-Void Mixtures, Average of Four Specimens for Each Line (1 psi = 6.89 kPa)

106

Figure 4.30 Effect of Compaction Method on Shear Creep Modulus of Low Air-VoidMixtures, Average of Four Specimens for Each Line (1 psi = 6.89 kPa)

107

t

r.d

........................................................................•i .i i̧ . - .

ii_i : i i i _ i:iii i

• . ! '

--:--_-+---,_---_ i _ ii i " ' !

. - - _ :

......... L_-............... _.. L..L.."............... _....

• , .... ! ! ._.._.........

................ ' "i::::i.':L':['! :i!:: : :D.':..IZ : "': : :13: I

' !.

; ................:.:_.!: ........:r...............

._._._..&..i.-i---._, _ i ,_,.. _..........i..:........ _ : :• _ =. i :

Figure 4.31 Effect of Compaction Method on Compressive Creep Modulus of HiAir-Void Mixtures, Average of Five Specimens for Each Line (1 psi - 6.89 kPa)

108

_d r.dv

, 0

-":..................... :": "-":';'r:.:Z-Z:.;];z:"_:Z:/Z::::;:_:-:_::_;_.;:I::::.:.....:--'-" r"': "]"-'-_..........':TT:"VTW'"'["i !i [. . . t "?

-+-'i : } ........... _........_ ......!'!'_'"'7"'""i"": "' _': ' : _ i. : iii'.il i " " :: " !

"'_"t"" _ '':'''_ ......... - • :

:: ; i::,_!,, i ; _.... :,z ' : ., ; : : ......... ".: " "

:--'i_!!!!:::!!!!!]!!:!!':!:: : ::=;:;:::i!:"i: ............t._ @� �l�°�::::::::::::::!2:7.=;:::L:::::_::::::..::','-'• .;....._........ ..:: " : " " " " ; ; . .i

. _,. ! : '.i _,]_'_!i _ " i i

•-:2:-:_:" :;:::::::::::• ::::..t..t J: +_." " " " _z":::".:":::::..*::::::...: .:"":':: ..... _ .__,_._.i. ;......z ............. •"" " i- ._-----_....... " " -,_

":-: .......................:......?.... _ _i...........i!:',:i!::!:::i:::!::::i......... r.,a.....:.............................. " " :.... ................:.-..:..i....................

_i_!:'!i:: i!i!!!;_!-!-::':!! !!!:_ ! : ::::;::::::::..: "'--

-:_ ...................... . _ • . ,...; ........ ;.;.g.,..;..: ........ . ...: ...............

..... : ..................... :......... : ....................... ,Li ...............

........................... -...........;.;.........;...................:.:.;....

i,,! i!!!!!'i

Figure 4.32 Effect of Compaction Method on Shear Creep Modulus of High Air-VoidMixtures, Average of Five Specimens for Each Line (1 psi = 6.89 kPa)

109

>-" Z

II

................................................................................. 7 .........................................................

f i i ii i

........................ : i ir! i

i -

I I I I I i I

Figure 4.33 Effect of Compaction Method on Strain Ratio in Shear Creep,Average of Four Specimens for Each Line

110

ZE-, r-_

-_

>-" Z

' I1

m

\\

i \

\ !

i \\ °J--,i

._",,. \ " _o z

• ! ",• \ _ . ,,%--,-..................................................; ....................._'__......................................................

: : ,.. . _ "_

• ' i '.. ::: • • i ",, r.2D

I I I I I

.J

Z__.=,-,,.....,_..[--, .....,.r./D Z

[_.,

Figure 4.34 Effect of Compaction Method on Specimen Dilation in Shear Creep,Average of Four Specimens for Each Line (I in/in = I ram/ram)

111

thusadds to the evidence that kneadingcompactionproducesa more stable aggregate structure,

at least in dense asphalt-aggregatemixtures.

A few cyclic-load tests, both in compressionand in shear, were conductedto corroborate

findings from the more extensive creep test program. The cyclic loading was characterizedby

a square loadpulse of 0.1 seconds followed by a rest period of 0.9 seconds. Strainsaccumulate

under cyclic loading, muchas they do undercreep loading.

Figure 4.35 illustrates the typical accumulationof "permanent" strains in cyclic shear

tests. At a given numberof load repetitions, kneadingspecimens accumulatedlesser amounts

of both dilational (axial) and shearswainsthan did gyratoryspecimens. With respect to dilation,

the effects of compaction method were similarto those observed undercreep conditions (Figure

4.36). Kneading specimens also had greater permanent deformation moduli than gyratory

specimens, both in shear (Figure 4.37) and in compression (Figure 4.38), indicating a greater

resistance to permanent deformation under cyclic loading. These findings, though based on

limited testing, are in full accord with those of the more extensive creep testing.

4.3 Complex Modulu_ Study

The complex modulus studyextended the domain of the properties of interest to include

dynamic modulusand phase angle, properties used to characterize the load response of linear

viscoelastic materials. Measured under sinusoidal loads, the dynamic modulus is simply the

ratio of peak stress to peak strain. The phase angle, a measure of damping, represents the

amount by which strain response lags the applied stress. Damping is also characterized by the

loss tangent, the tangent of the phase angle. In addition to examining two more engineering

properties of potential significance to pavement performance, the complex modulus study

provided further opportunity to examine the interactive effect of air voids and compaction

method. The study is detailed as follows:

112

•° ) D°).m °m

Figure 4.35 Effect of Compaction Method on Permanent Strain in Cycfic Shear(1 in/in - 1 mm/mm)

113

Z.-- 0

Z >"C_D

W

Figure 4.36 Effect of Compaction Method on Specimen Dilation in Cyclic Shear,Average of Two Specimens for Each Line (1 in/in - 1 mm/mm)

114

• []

...................................I ...........................II ........................................._ o

/ =

' B// o

t_

,..o _

E_oID.

Figure 4.37 Effect of Compaction Method on Permanent Deformation Modulusin Cycfic Shear (1 psi = 6.89 kPa)

115

Z =::

Z >--

, 0

i i i _. : : : ; _ : _'_'+!"i"+::., , I_:!: ! : !

"-'"........:"_.......... : = = _ _ ?"?'"'Pi ......': : : : : i i!':

:;;:_-::::: : :::::::.i:.":_ : : :. :::::::::::::::_:::............' ............._ !....: i : ..! ,.._........... ......

,, ,, : ":,: : , P

:::':-:::: i:::::i::: : ;-_::::::::::-:_:-:..... _ ; :::::':::::i::::::::::::::."_.,...... :;:::':_,:::...:.:.. :. ..............._....................../ _ .,,..:"_ .............._...... :i • • i :

...................................:,::: _ .....

_.:.........._.........:.............._ ........:.._ I i I

=E:: !i:! : ! !

I:_.::':": ::::::2:: "'. ".. , i ..................... "

....:!:2::::: :::::::::::::i:::::::::::..................... :,...,- ............

,,:.,,...; .........

! ::" ,'i: " . :

J i I -

Z _

Figure 4.38 Effect of Compaction Method on Permanent Deformation Modulusin Cyclic Compression (1 psi -- 6.89 kPa)

116

Number of Tests

(Specimens) Air-VoidTemperature .... Stress Frequency Content

Gyratory Kneading(Cored) (As Molded)

Sinnsoidal, Variable

68"F (20"C) 98 (7) 98 (7) 2 to 32 psi (0.01 to 16 Hz) Variable(13.8 to 220.5 kPa)

Sinusoidal, Variable

104°F (40"C) 84 (6) 98 (7) 2 to 12 psi (0.01 to 16 I-Iz) Variable(13.8 to 82.7 kPa)

I I I

Common AAG-1 asphalt at optimum asphalt content with granite (RB). Unconfined compressiveElements load applied axially to 4 x 8-inch (101.6 x 203.2 ram) specimens.

Because the loading is considered to be nondestructive, each specimen was tested over

the entire frequency spectrum and at each of the two test temperatures. To minimize potential

specimen damage, however, tests were conducted first at the low temperature and the 16-Hz

frequency. Subsequent testing progressively swept the frequency range toward its lowest point,

0.01 Hz. The temperature was then increased and, when it stabilized, the frequency sweep was

repeated. To minimize testing error, computations at the higher frequencies were based on

average specimen response during the 91st through 100th load cycles after discarding the high

and low extremes in this 10-cycle range. For frequencies of 0.8 Hz and less, averages, after

again discarding the high and low extremes, were based on measurements made during the first

seven load cycles.

The effect of compaction method on dynamic modulus was an expected one: kneading

specimens were stiffer (as indicated by larger dynamic moduli) than gyratory specimens even

though on average their air-void content was 0.6 percent larger (Figures 4.39 and 4.40).

Differences attributable to compaction procedure are greatest at the higher temperature and

lowest frequencies where larger strains accentuate the contribution of the aggregate structure.

At the lower temperature and highest frequencies, the asphalt binder exerts much greater

117

t.

qg

D

................................C .............................: ............... it,.

Figure 4.39 Effect of Compaction Method on Dynamic Modulus at 68°F (20°C),Average of Seven Specimens for Each Line (1 psi = 6.89 kPa)

118

m

Figure 4.40 Effect of Compaction Method on D3m_mic Modulus at 104°F (40°C),Average of Six Specimens for Gyratory and Seven Specimens for Kneading

(1 psi - 6.89 kPa)

119

influence on the dynamic modulus, and the measurable differences between compaction methods

seem to vanish.

In addition to its effect on dynamic modulus, compaction method has a definitive

effect--also diminishing at the lower temperature and highest frequencies-on loss tangent

(Figures 4.41 and 4.42). Where compressive strains are large (high temperature and low

frequencies), kneading specimens exhibit greater damping as evidenced by larger loss tangents.

At intermediate strains, there is a reversal in order and gyratory specimens exhibit greater

damping. Where compressive strains are small (low temperature and high frequencies), the

compaction effect on loss tangent seems to diminish.

Figures 4.43 and 4.44 illustrate the effect of air-void content on dynamic modulus. For

the two test temperatures and throughout the frequency range, mixtures with higher air voids

were less stiff than those with lower air voids. The effect of air voids is less pronounced--but

still readily distinguishable--at the lower temperature and highest frequencies, generally the low

strain region. Results at the opposite extreme of the temperature and frequency spectra are less

reliable, possibly because of nonlinear response to the applied loads. Most importantly, the

difference between compaction methods is more pronounced for low-void specimens than for

high-void specimens, a finding in agreement with the hypothesis that differences between

gyratory and kneading compaction become more pronounced when more of the mineral voids

are filled with asphalt.

There is more measurement error in the loss tangent data, and the effects of air-void

content appear more complex (Figures 4.45 and 4.46). Focusing for the moment on the

kneading specimens, the loss tangent of specimens with fewer air voids is larger at the high test

temperature (Figure 4.45) but smaller at the high-frequency end of the low temperature testing

(Figure 4.46). The air-void effect diminishes in magnitude at low strain levels, that is, low

120

• []

Figure 4.41 Effect of Compaction Method on Loss Tangent at 68°F (20°C),Average of Seven Specimens for Each Line

121

• rq

Ii

Figure 4.42 Effect of Compaction Method on Loss Tangent at 104°F (40°C),Average of Six Specimens for Gyratory and Seven Specimens for Kneading

122

Qt4

i

O

i ! .O

l ° °8 °Ogp-

P

rjq.4 .._

_gh

It00_:D

Figure 4.43 Combined Effects of Air-Void Content and Compaction Method onDyn_c Modulus at 104°F (40°C), Average of Three Specimens for Each Line

(1 psi = 6.89 kPa)

123

> " ,d _ "

........................................ Op

Figure 4.44 Combined Effects of Air-Void Content and Compaction Method onDynamic Modulus at 68"F (20"C), Average of Three Specimens for Each Line

(1 psi = 6.89 kPa)

124

Nv

0 • • •

.............. Q

e--

... .. ,... . .................. , ........................ _....................... _,. [

,-. ......_ ....... ] _0

i -

Figure 4.45 Combined Effects of Air-Void Content and Compaction Method on LossTangent at 104°F (40°C), Average of Three Specimens for Each Line

(1 psi = 6.89 kPa)

125

.............................. O: Q-

Figure 4.46 Combined Effects of Air-Void Content and Compaction Method on LossTangent at 68°F (20°C), Average of Three Specimens for Each Line

126

temperatures and high frequencies. The pattern for gyratory specimens is similar to that of

kneading specimens for the low-temperature tests (Figure 4.46). At the high temperature, data

for the gyratory specimens appear to be less reliable, possibly because of nonlinearities, and no

clear pattern in the air-void effect is apparent (Figure 4.45). Interactive effects between

compaction method and air-void content are not apparent from the graphical presentations of

Figures 4.45 and 4.46.

In summary, kneading specimens have larger dynamic moduli than gyratory specimens,

an effect that is more pronounced at higher strains (high temperature and low frequency) than

at lower strains (low temperature and high frequency). As expected, the effect of compaction

method is magnified for specimens with lower air voids. Compaction method and air-void

content also affect loss tangent, though in more complex ways. It appears that stiffer mixtures,

whether a result of kneading compaction or a result of lower air voids, have greater loss tangents

at higher strains (high temperature and low frequency) and smaller loss tangents at lower strains

(low temperature and high frequency). Available data don't permit an accurate assessment of

the combined effects of compaction method and air-void content on loss tangent.

4.4 Comparisons with Other Compaction Methods

During the course of the compaction study, opportunities became available to test a

limited number of specimens compacted with the Corps of Engineers' gyratory testing machine

and Exxon's rolling-wheel compactor s. The following summarizes the testing program for

comparing these specimens with others compacted by the Texas gyratory compactor and the

Triaxial Institute kneading compactor:

8The Corps of Engineers' gyratory specimens were prepared by North Carolina StateUniversity, a subcontractor on SHRP Project A-003A. The rolling-wheel specimens wereprepared by Exxon Research and Engineering Company of Linden, New Jersey.

127

Compaction Method Number of Tests Surface Condition

Gyratory (Texas) 2 As Molded

Gyratory (Corps) 2 As Molded

Kneading 2 As Molded

Gyratory (Texas) 2 AS Molded

Kneading 2 AS Molded

Exxon Rolling Wheel 2 Cored

AAG-1 asphalt at optimum asphalt content with granite (RB). All testingCommon Elements performed at 140"F (60"C) in shear creep at 3.3 psi (22.7 kPa) static

shear and 5 psi (34.4 k.Pa) staticaxial.

Shear creep curves for the Corps of Engineers gyratory specimens and those to which

they were compared are presented in Figure 4.47. The Corps specimens, identified as

"GNCSU" on the figure, exhibited a distinctly different response to shear creep than did the

Texas gyratory specimens, identified simply as "G". Although the cause of this difference could

not be ascertained, it may be related to the use of different angles of gyration. The Corps

gyratory testing machine was operated at an angle of one degree, and the Texas gyratory

compactor, at an angle of six degrees. Regardless of the cause of the difference, however, the

important point in the context of the current study is that different compactors (or different

compaction procedures) of the same class or genre may produce distinctly different kinds of

specimens.

Shear creep curves for the Exxon rolling-wheel specimens and those to which they were

compared are presented in Figure 4.48. The rolling-wheel specimens are identified by the letter

"X" in the legend. At the longer loading times, those of greatest practical significance to

pavement performance, the response of the rolling-wheel specimens appears to match that of

kneading specimens more closely than that of gyratory specimens. The general ranking in terms

of resistance to permanent deformation, kneading followed by rolling-wheel followed by

gyratory, corroborates findings from the main compaction experiment.

128

e'_

an _ #

• " " __

/ _

//

.......................................... m . ........................................................ _

/

_ m

_ 0m_

Figure 4.47 Comparative Response of Corps of Engineers' Gyratory Specimens toShear Creep Loading, Average of Two Specimens for Each Line (1 psi = 6.89 kPa)

129

v v _v_ X

/ E

/ -

/ ,f

II• i

i i --.

LV

o

Figure 4.48 Comparative Response of Exxon Rolling-Wheel Specimens to ShearCreep Loading, Average of Two Specimens for Each Line (1 psi - 6.89 kPa)

130

Although this investigation was too limited to be of much significance, it does support

the notion that compactors of the same generic class but of different detail may produce

specimens that respond rather differently to imposedloads. Tests on the rolling-wheel specimens

prepared by Exxon yielded no surprises: results were in general accord with those developed

in other phases of the study.

4.5 Comparisons with Field Cores

Field cores, extracted-from.the wheel paths of two California highways, provided the

opportunity for a limited comparison of field and laboratory compaction. The testing program

is summarized as follows:

Highway Compaction Method Number of Laboratory SurfaceTests Fabrication Condition

Gyratory 5 Laboratory specimens As Moldedprepared from loose

I 80 Kneading 5 mixture stored in sacks As Molded

since pavementField 5 construction. Cored

Gyratory 4 Laboratoryspecimens As Moldedremoldedfromloose

US 101 Kneading 4 materials recovered As Molded

Field 4 from heating cores. Cored

Common Elements Field mixes (US 101 with granite, RB). All testing performed at 140°F (60°C) inshear creep at 3.3 psi (22.7 kPa) static shear and 5 psi (34.4 kPa) static axial.

The I 80 pavement had been undertraffic for approximatelythreeyears priorto coring.

At the time of construction, loose samplesof the asphalt-aggregatemixturehadbeen sacked for

later testing and evaluation. Laboratory specimens were prepared from these loose samples.

The mixture was composed of a high quality, crushed aggregate and the equivalent of an

AR-4000 asphalt.

Results of the shear creep testing, summarized in Figure 4.49, suggest that response of

the field cores lies between that of gyratory specimens and that of kneading specimens. At the

131

r,.-

c,- >-

r-_ [....

z _ >-.

m <_

!. oi

v

Figure 4.49 Comparative Response of 1 80 Field Cores to Shear Creep Loading,Average of Five Specimens for Each Line (1 psi = 6.89 kPa)

132

same time, the dilational response of field cores matched much more closely that of kneading

specimens, possibly indicating similar aggregate structure between field cores and kneading

specimens (Figure 4.50).

Confounding aspects of this comparison of field and laboratory compaction included (1)

possible differences between as-molded and cored surfaces, (2) traffic conditioning of the field

cores, and (3) differential aging. To examine the latter aspect, viscosities were measured on

extracted asphalts. Although these measurements indicate more aging in the field cores than in

the laboratory specimens (Figures 4.51 and 4.52), they provide no further explanation of the

effects observed in the creep testing.

Field cores were also available from a section of US 101 in California that had been in

service for many years. Laboratory specimens were prepared from field cores which, after

heating, were mechanically crumbled in preparation for recompaction. Results of the shear

creep testing are summarized in Figure 4.53. In this case, the field cores proved more resistant

to permanent deformation than either type of laboratory specimen. Interestingly, despite low

average air voids, gyratory specimens appear to be more resistant than kneading specimens

although, at the end of the one-hour test, they seem to have been more rapidly approaching

collapse. These results are not easily explained.

The dilational response of these specimens in shear creep is of even greater curiosity.

As expected because of their high resistance to shear, the field cores dilated more than the

laboratory specimens (Figure 4.54). For some inexplicable reason, the kneading specimens

reached a plateau at intermediate shear strains: little additional dilation occurred even though

the shear strains eventually approached the collapse level of 0.02 in/in (0.02 mm/mm). Such

behavior had not been observed in other shear testing and was unexpected.

133

>- _._Zm

r_ >_ zm

<_ _ m

c",.)

c_5 z

E.-.--

........................"mm_--_io°

e-°_

Z__.__

Figure 4.50 Comparative Dilational Response of 1 80 Field Cores to Shear CreepLoading, Average of Five Specimens for Each Line (1 in/in -- 1 ram/ram)

134

180- Viscosityat 140 F

16000-

14000 - '

12000

10000 ,

Viscosity 8000(Poises)

i

6000 ,

I

4000 rI

2000

0 I

FIELD GYRATORY KNEADING

Figure 4.51 Viscosities of Asphalts Extracted From I 80 Field Coresand Laboratory Specimens, 140°F (60°C) (1 poise = 0.1 Paos)

135

180- Viscosityat 275 _

5 l

m

Viscosity 3(Poises)

l

2

0

FIELD GYRATORY KNEADING

Figure 4.52 Viscosities of Asphalts Extracted From I 80 Field Coresand Laboratory Specimens, 275°F (135°C) (1 poise = 0.1 Paos)

A _ A

I I I

i

0.................................................. 0

0

/ -j" i ....................................

i _ i --

_g

Figure 4.53 Comparative Response of US 101 Field Cores to Shear CreepLoading, Average of Four Specimens for Each Line (1 psi = 6.89 kPa)

137

................... 7 .......

Figure 4.54 Comparative Dilational Response of US 101 Field Coresto Shear Creep Loading (1 in/in -- 1 mm/mm)

138

In summary, the testing of field cores failed to produce definitive results. The I 80 tests

are more easily understood than the US 101 tests, but neither set of tests adds substantially to

a better understanding of the compaction process. The US I01 tests raise interesting, but

unanswered, questions.

4.6 (_Qmoarisons with Other Investi2ations

Monismith and Tayebali (1988) compared the shear creep response of laboratory

compacted asphalt concrete mixtures (with and without carbon black modifier) with the response

of I 80 (Sierra Nevada) field cores. The laboratory compactor used in their study was the same

kneading compactor used in the SHRP A-003A study. Figures 4.55 and 4.56 indicate that, in

both cases (with and without modifier), the kneading specimens exhibited a better resistance to

permanent deformation than the field cores.

The most recent comprehensive investigation of laboratory compaction procedures was

conducted as a part of the National Cooperative Highway Research Program (NCHRP)

Asphalt-Aggregate Mixture Analysis System (AAMAS) study (Von Quintus et al., 1988). Its

primary objective was to determine which of five laboratory compaction methods produced

specimens most nearly duplicating field cores. The five compaction methods included not only

the Texas gyratory compactor and the kneading compactor but also the Marshall compactor, the

Arizona vibratory/kneading compactor, and a mobile steel wheel simulator.

Although particle orientation and the distribution of air voids were among the factors

considered in the analysis, the NCHRP comparisons were driven primarily by the notion that

the best laboratory procedure is that which produces specimens having engineering properties

most like field cores. Critical engineering properties, all measured by diametral loading of

briquet specimens, were considered to be tensile strength, tensile strain at failure, resilient

139

_9

(FIELD MIX, I.AJ_COMPACTION)

_ (FIELD CO,_ES)

OCSCBll I:ICSCBI2 "FIELD CORES)A CSCBL f 0 CSCBL2 (FIELD MIX. LAB COMPACTION)

10 -I I 10 10 z 10 s 10 "TIME (sec)

Figure 4.55 Comparison Between Laboratory Specimens and Field Cores for Mixtureswith Carbon Black (1 psi = 6.89 kPa) (After Monismith and Tayebali, 1988)

140

I I F0 O

_ ^7 <> r_

2! °£.7. _ _"

°-.-

y -T

• Ol , Ol , Ol _'0l _ Ol Ol '_

Figure 4.56 Comparison Between Laboratory Specimens and Field Cores for ControlMixtures (1 psi = 6.89 kPa) (After Monismith and Tayebali, 1988)

141

modulus--each measured at temperatures of 41, 77, and 104*F (5, 25, and 40°C)--and various

indices from creep testing at 77 and 104*F (25 and 40"C).

Based primarily on the analysis of three of five paving projects, located in as many states

and each unconditioned by traffic loading, the NCHRP investigators concluded that specimens

produced by gyratory compaction had engineering properties more similar to field cores than

those produced by the other compaction procedures. The kneading compactor and the steel

wheel simulator were ranked as second-best choices. Although the differences among these

three compaction methods were judged to be statistically significant; they were relatively small

in magnitude and of no apparent practical significance--considering the inaccuracies of the

models that were employed to predict performance effects.

Using the slope of the creep curve as a measure of resistance to permanent deformation,

the NCHRP investigators found that kneading specimens (CK/CC) had flatter slopes than

gyratory specimens (MT/GS) and were, therefore, more resistant to permanent deformation

(Figure 4.57). The NCHRP Studythus corroborates one of the principal findings of the current

study.

However, two caveats regarding the NCHRP study are in order. First, it seems likely

that compaction effects observed in the current study are significantly greater than those

observed in the NCHRP study. Whether such differences can be attributed to different test

procedures, to different mixture composition and/or air-void contents, or to other factors is

unknown. In any case, the NCHRP study may understate the effects of compaction method on

important mixture properties. Second, most of the NCHRP field projects exhibited rather high

air voids. Results reported herein indicate that differences among compaction methods are likely

to be smaller for mixtures with high air voids. Results of the NCHRP study might have been

142

0.7

O-35

0.3 I ( I I

0.3 0.4. 0,5 0.6 0.7

Dope of CreepCurve,CK/CC

Figure 4.57 Comparison of Slope of Creep Curves for Gyratory (MT/GS) andKneading (CK/CC) Specimens (March 28, 1989 Letter to J. Moulthrop

from H. Von Quintus)

143

different if field compaction had resulted in lower initial air voids or if the field mixtures had

been further compacted by traffic before they were tested.

144

5.0 AIR-VOID MEASUREMENT AND CUT-SURFACE EFFECTS

The laboratory work for this study revealed that standard methods of air-void

measurement were not reliable for comparisons among specimens that had been prepared by the

different compaction procedures. Furthermore, significant differences in both measured air-void

contents as well as mechanical behavior were observed between uncut specimens (those tested

as compacted in the mold) and cut specimens (those sawed and/or cored from a larger compacted

mass). Primarily because of the critically important impact of air-void content on the

mechanical behavior of asphalt-aggregate mixtures, a special investigation of the effects of air-

void measurement and cut-surface effects was undertaken.

This section presents data that provide insight into these problems and includes

recommendations for resolving them. Useful evidence was collected regarding (1) relationships

between measured air voids and specimen preparation method and (2) the effects of cut surfaces

on measured air voids and test results. Shear creep tests were performed to examine the effects

of cut surfaces on specimen performance under loading.

The extent of data used to examine the issue of air-void measurement is summarized in

Table 5.1. Surface condition of the several types of specimens is summarized as follows:

• Briquets, 4 x 2.5 inches (101.6 x 63.5 mm), prepared by gyratory or kneading

compaction had "compacted" or "as-molded" surfaces that were not cut by coring

or sawing.

• Cylinders, 4 x 8 inches (101.6 x 203.2 mm), prepared by kneading compaction

had uncut surfaces at the sides and cut surfaces at the ends.

• Fatigue beams, 3.5 x 3.5 x 15 inches (88.9 x 88.9 x 381 mm), prepared by

kneading compaction, cylinders prepared by gyratory compaction, and all

145

Table 5.1 Sample Size for Study of Air-Void Measurement

Compaction Specimen Size and Chert (RL) Granite (RB) OtherShape AAG-1 AAK-1 AAG-1 AAK-1

4x2.5 in(101.6 x 63.5 mm) 29

BriquetGTrator7

4x8in(101.6 x 203.2 ram) 12

Cylinder

4 x2.5 in(101.6x63.5mm) 12 6 38 15 8

Briquet

4xSin

(101.6x 203.2mm) 2 9 15 5Cylinder

3.5x 3.5x 15in(88.9x 88.9x 381 27 7 20 17

Kneading mm)UncutBeam

3x3xl5in

(76.2x 76.2x 381 12 I 30 71mm)

Cut Beam

1.5 x 1.5 x 15 in(38.1 x 38.1 x 381 105 34 165 143

ram)Beam

4x 2.5 in(101.6 x 63.5 ram) 11

Rolling Briquet

Wheel 4 x 8 in(101.6 x 203.2 ram) 18

Cylinder

Total Number of Specimens 805

146

specimens prepared by rolling-wheel compaction were cut or cored from larger

compacted masses.

5.1 Air-Void Measurement

5.1.1 Objectives of Air-Void Measurement. The objectives of air-void-measurement

procedures can be summarized as follows:

1. To measure the degree of mixture compaction resulting from the application of

mechanical energy to the mixture in order to expel air and/or reorient the

aggregate into a denser, more stable arrangement;

2. To measure the voids in the aggregate matrix which are not filled with asphalt;

3. To be fast and simple and to provide repeatable results and maintain the original

specimen shape so that it can be used for testing; and

4. To allow meaningful comparisons among a variety of specimens with different

degrees of compaction, different asphalt contents, and/or different fabrication

techniques including specimens obtained by field coring.

5.1.2 Methods of Air-Void Measurement. Bulk specific gravities for air-void measurement

of specimens prepared early in the compaction study, 4 x 2.5-inch (101.6 x 63.5 ram) briquets

using kneading and gyratory compaction, were calculated by weighing them in air and in water,

without the use of paraffin wax or other surface coating. The air-void content was calculated

using the following equation:

AV = (I- ob_,) x 100 (1)Gmm

where AV is the air-void content in percent, Gb is the bulk specific gravity, and Gmm is the

maximum specific gravity determined by the "Rice method" (ASTM D 2041). The standard

147

method for sealing specimens using paraffin wax (ASTM D 1188) was not used, since testing

after air-void measurement requires removal of the paraffin wax, a difficult and time

process. These specimens were neither cut nor cored and therefore contained no entrapped

water. This procedure is referred to as dry-no-parafilm (drip).

Rolling-wheel specimens and 4 x 8-inch (101.6 x 203.2 mm) gyratory cylindrical

specimens were cored and/or cut from larger compacted masses, using water cooling to prevent

aging of the asphalt and damage to the specimens due to frictional heat. Thisprocess leaves

water in the specimen--in varying amounts depending in part-on the internal air voids--which

may require several days to remove by drainage and evaporation. Because entrapped water may

be included in the weight-in-air measurement used to calculate Gb in Equation l, comparison

between air-void contents for these cut/cored specimens and those measured using the

dry-no-parafilm method was questioned. The standard procedure for removal of water (ASTM

D 2726) requires heating at 230°F (ll0OC), a potentially destructive process likely to

some specimens unsuitable for use.

To surmount this difficulty, a method was developed to permit direct comparisons

between the air voids of cut and uncut specimens. It was assumed that, once a specimen had

been wetted, a considerable waiting period might be required to allow it to dry under ambient

conditions. To circumvent such delays, the adopted method required all specimens (cut and

uncut) to be first wetted by immersion in a water bath and then dried with a compressed air

nozzle until a "surface dry" condition was achieved. This method is referred to as wet-no-

parafilm (wnp).

At the same time, it was found that permitting water to enter the specimen during

submersed weighing resulted in underestimation of air voids, particularly for specimens with

higher air-void contents. At high air-void contents, above approximately 8.5 percent by

148

dry-no-parafilm method, it was found that air-void content as measuredby the dry-no-parafilm

or wet-no-parafdmmethods was extremely insensitiveto compactive effort, andvisually obvious

differences in void content were undetectedby the air-void measurements. Experimentationwas

then carriedout using an elastic wax paper, parafilm, as a substitute for paraffinwax (Del VaUe,

1985). Measurementsof air voids, using the "surfacedry"condition and parafilm,were found

to be sensitive to air-void content, applicable to all types of compaction and specimen

preparation methods, and convenient for later testing of the specimen. This procedure is

referred to as wet-with-parafilm (wwp). The equation used to calculate bulk specific gravity

using this method is the same as in the standardparaffin-wax procedure (ASTM D 1188).

5.1.30uantitative Differences Between Methods. Wet-with-parafilm (wwp) measurements

are compared to wet-no-parafilm (wnp) measurements in Figure 5.1 and to dry-no-parafilm (dnp)

measurements in Figure 5.2. These results show the significant difference resulting from the

use of parafilm. Interestingly, however, the dry measurements (dnp) are not significantly

different from the surface-dry measurements (wnp) as illustrated in Figure 5.3. This indicates

that the blown-air-drying procedure was quite efficient and that initial concerns about completely

drying the specimens in a relatively short time were unwarranted. This finding permits direct

comparisons of measured air voids for as-molded specimens with those for previously wetted

specimens that have been surface dried with compressed air. The combined results of Figures

5.1 through 5.3 are summarized in Figure 5.4 which shows the least squares regression line for

each of the above three relationships.

The effects of parafilm on cut and uncut specimens are similar though not identical, as

demonstrated by regression lines for the with (wwp) and without (wnp) parafilm methods (both

methods used on specimens in both the cut and uncut condition) shown in Figure 5.5. There

149

AIRVOID5WWP 8

0

O 2 4 6 O 10 12

All?VOIDSWNP

Figure 5.1 Effect of Paraf'dm on Surface-Dry Measurements of Air Voids,All Specimens

150

16

14

12

10 ••_ 4#

_IRvoidswwPo •° Y_,*:_-o°

4 _i 1_]_6_-_./._

0

0 2 4 6 8 10 12

AIRVOIDSDNP

Figure 5.2 Combined Effect of ParafUm and Wetting on Air-Void Measurements

151

16

14

12

10

AIRVOIDSWNP 8

6

4

, I

o i :0 P. 4 6 8 I0 !_

AIRVOIDSDNP

Figure 5.3 Effect of Wetting on No-Parafilm Measurements of Air Voids

152

16

1, /12 // xvsY10 x-_ ALL,WNPvs WWP

ALL.DNPvs _PY - AIRVOIDS(Z) 8 ., [

I _ ALL,DNPvs WNP

6 -- Line of Equality

4

2

0

0 2 4 6 a _0 _2 14 _8X- AIRVOIDS(%)

Figure 5.4 Comparison of DNP, WNP, and WWP Measurements of Air Voids,Regression Lines

153

16s

//1, J/;_ /12

t[ _///_//// _" CUT,WNPvs WWP

10

AIRVOIDSWWP g I ,J/ "<>UNCUT,WNPvsI

° //4

0

0 2 4 6 8 10 12 14 16

AIRVOIDSWNP

Figure 5.5 Combined Effect of Cut Surfaces and Parafilm on Air-VoidMeasurements, Regression Lines

154

is generally less difference between airvoids measuredwith the two methods for specimens with

cut surfaces than with uncut surfaces due to the smoother exteriors of the cut specimens.

Specimens prepared using granite (RB)andchert (RL) both show littledifference between

the dry (dnp) and wet (wnp) conditions, as can be seen in Figure 5.6. However, the chert (P,.L)

specimens with higher air-void contents are more sensitive to the use of parafilm than are the

granite (RB) specimens, as shown in Figure 5.7. This is probably the result of the parafilm

trapping air because of the more irregular, chipped surfaces of the chert (RL) specimens.

5.1.4 Critiaue of Wet-With-Parafilm Procedure. Although the wet-with-parafilmprocedure

is better than other alternatives, two concerns remain. The primary concern is arching of the

parafilm over surface irregularitiescausedby chipping of the aggregates during cuttingand/or

coring, with the result that air trappedunderthe parafilm is improperly considered as internal

air voids. This problemoccurs most frequentlywith specimens having thin cross-sections, broad

flat uncut surfaces, and/or aggregates that chip easily. A similar problem occurs even with

smooth specimens if careful attentionis not given to the removal of air entrappedbeneath the

parafilm.

The other potentialconcern with the wet-with-parafilmtechnique is the repeatabilitywith

which the "surface dry" condition is reached, although a limited comparison with tests from

another laboratory showed little problem. While the general technique used in this study has

proven to be suitable, moreattention must be given to standardizationof the drying equipment

and procedures.

155

16

12 ' /

I0_ TEXCHERT.DNPvsWNP

AIRVOIDSWNP 8 1 /j// <>-WATGRANITE.DNPvs

!/4

2 k

/ ,o , i ,I0 2 4 6 8 10 12 14 16

AIRVOIDSDNP

Figure 5.6 Combined Effect of Wetting and Aggregate Type on Air-VoidMeasurements, Regression Lines

156

16

::i10 I / _ TEXCHERT.DNPvs WNP

AIRVOIDSWWP 8 ¢- _ATGRANITE,DNPvs_,'NP

/ _8 k4

0

0 2 4. 6 8 10 12 14 16

A]RVOIDSWNP

Figure 5.7 Combined Effect of Parafilm and Aggregate Type on Air-VoidMeasurements, Regression Lines

157

5.2 Effect of Cut Surfaces

5.2.10biectives of Laboratory Svecimen Compaction. The objectives of laboratory

specimen fabrication are briefly summariz___as follows:

1. To fabricate specimens that as closely as possible resemble in-service mixtures,

that is, those produced by mixing, placement, and compaction in the field;

2. To fabricate specimens that, under laboratory testing, exhibit the same behavior

as in-situ mixtures under similar states of stress; and

3. To be as efficient as possible in the use of labor, time, equipment, and material.

5.2.2 Effects of Compaction Method. Slight differences were visually observed between the

top and bottom surfaces of all laboratory-prepared specimens. For gyratory specimens, the

lower surface was usually rougher, that is, both the aggregate particles and the surface voids

were larger in size. The top of the specimen, which had been in contact with the compacting

surface, usually had a smoother surface with finer particles. For kneading specimens, the

reverse was true although the difference between the top and bottom surfaces was much less than

that of gyratory specimens. The lateral surfaces of most molded specimens were a smooth

matrix of asphalt and the finer fractions of aggregate: there was little honeycombing and larger

particles had migrated inward. Lateral surfaces of specimens with high asphalt contents often

contained excess asphalt, especially when compacted to low air-void contents.

During compaction, the mold walls constrain the reorientation of surface particles.

However, in the interior, the sheafing action of the compactor reorients the larger particles,

producing a "structure" that significantly affects engineering properties of the compacted

mixture. If quantifies of the asphalt and fine particles are limited or the compaction effort is

modest, gaps separating the larger particles at the surface remain unfilled. In any case, the

158

likely result is that the aggregate structure at the surface of the specimen is different from that

in its interior.

Some segregation of larger from finer panicles occurs when pouring and compacting hot

mixtures. Segregation becomes more pronounced as the size of the compacted specimen and/or

the compactive effort decreases, since the size of the larger aggregates relative to the mold

dimensions, and/or the lack of shearing energy, do not permit easy movement of fines to the

interstices between the larger aggregates.

Gyratory specimens are subjected to a torsional shear force on their lateral surfaces due

to rotation of the compaction apparatus base plate, while the compaction ram resists rotation at

the top surface. As a residual result of these surface forces, large aggregate particles often

separate from the surface when the specimen is extruded from the compaction mold. When the

large particles are rounded, thin layers of finer material between the large particles and the

specimen surface also frequently separate from the specimen after extrusion. The effects of the

torsional surface forces in gyratory compaction are recognized in the procedure for the Texas

gyratory compactor (Texas Method Tex-206-F and ASTM D 4013) which stipulates that large

aggregate particles should be pulled away from the mold wall with a spoon before beginning

compaction. The mold will rotate if the torsional shear force exceeds the friction between the

mixture and the mold wall, smearing excess asphalt and fine material across the lateral surface

of the specimen.

Cutting or coring the specimen from a larger compacted mass removes those portions of

the specimen that have been subjected to the forces and effects described above. The specimen

with cut surfaces has a more homogenous aggregate and air-void structure, and its surface is not

smeared with asphalt.

159

The major drawbacks of cut specimens include (1) the need for care during sawing and

coring to obtain the required geometric shapes, free of ripples or other surface imperfections,

and (2) possible problems with the exposure of mixtures to water during the sawing/coring

process. Problems with water-sensitive mixturesare encounteredin air-void measurementsthat

involve wetting the specimen, however, and they can usually be solved by careful drying and

timely testing.

5.2.3 Effects on Air-Void Determinstien. The effects of surface condition (cut or uncut) on

air-void measurements depend on the properties of the mixture (including, for example, asphalt

content and aggregate angularity) and the compaction method (gyratory, kneading, or rolling

wheel).

1. The difference between air voids measured with cut surfaces and those measured

with uncut surfaces depends on the degree of nonuniformity in the air voids and

aggregate structure between the surface and the interior of the specimen: this is

a function of compaction method, compaction effort, aggregate angularity, and

asphalt content.

2. If the specimen has a large surface area in relation to its volume, the difference

between the measured air voids with cut versus uncut surfaces is greater. Thus,

large specimens such as 3.5 x 3.5 x 15-inch (88.9 x 88.9 x 381 mm) beams and

4 x 8-inch (101.6 x 203.2 mm) cylinders will have less difference in the

measured air voids than 2.5 x 4-inch (63.5 x 101.6 mm) briquets.

3. If the asphalt content and/or the compaction effort is high, asphait at the surface

of the uncut specimen may forma coating, which prohibits the entry of water into

the specimen and may affect air-void measurements.

160

4. For high asphalt contents and/or large compaction efforts, the sealing action at

the surface of the uncut specimen is greater for gyratory specimens due to

smearing caused by the torsional shear forces on the lateral surfaces.

An example of the difference between the measured air voids for specimens as compacted

and after cutting is shown in Figure 5.8. In this case 3.5 x 3.5 x 15-inch (88.9 x 88.9 x 381

ram) beams of all mixture types, prepared by kneading compaction, were tested using the

wet-with-parafdm (wwp) procedure both after compaction and after trimming of the outer 3/16

inch (4.8 ram) of the surface. The data indicates that the as-compacted specimen has

approximately 1 to 2 percent more air voids than the same specimen after cutting. This pattern

is confirmed by testing the same specimens using the wet-no-parafilm (wnp) method as

illustrated in Figure 5.9.

Further evidence of differences between interior and exterior air-void contents is shown

in Figure 5.10, for 4 x 8-inch (101.6 x 203.2 ram) gyratory specimens. In this case specimens

were compacted as 6 x 9.5-inch (152.4 x 241.3 ram) cylinders, then cored and cut to their final

dimensions. Comparison of air-void measurements using the dry-no-parafilm method for the

uncut specimens and the wet-no-parafilm method for the cut specimens--these two methods have

previously been shown to produce similar results-indicates that at lower compaction efforts

(higher air-void contents) the outer portion of the specimen is more densely compacted than the

inner portion, while at higher compactive efforts 0ower air-void contents) the reverse is true.

At high compactive efforts (low air voids), differences in air voids of up to three percent (wwp)

have been found between the center third (higher air voids) and top and bottom thirds (lower air

voids) of 4 x 8-inch (101.6 x 203.2 mm) gyratory specimens. This indicates specimen non-

uniformity along the vertical axis, occurring despite the fact that the specimens were compacted

in one lift.

161

16 "!

14

12

AIRVOIDSWWP- ACUT 8

i A _ *-)--%-*_)4

2 ,,

! !O 2 4 6 8 10 i2

AIRVOIDSWWP - UNCUT

Figure 5.8 Effect of Cut Surfaces on Wet-With-Parafilm Measurements ofAir Voids in Large Beams, Kneading Compaction

162

16

14

12

R VOIDSWNP - 8

CUT _ E

4

O "__ _""

0 2 4 6 8 10 12

AIRVOIDSWNP- UNCUT

Figure 5.9 Effect of Cut Surfaces on Wet-No-Parafilm Measurements ofAir Voids in Large Beams, Kneading Compaction

163

16

14

12

I0

AIRVOIDSWNP- 8 I

CUT ,!_ __6 ii

4

2 e

o I0 2 4 6 8 10 12

AIRVOIDSDNP- UNCUT

Figure 5.10 Effect of Cut Surfaces on No-Parardm Measurements ofAir Voids in Large Cylinders, Gyratory Compaction

164

5.2.4 Effects on En2ineerin2 Prom_rties. In some types of tests, cut surfaces may have a

significant effect on the measured response of the mixture. In tests which impose large shear

stresses, the measurementswill tend to be larger for specimens whose surfaces have been cut.

The cut surfaces of the aggregates rest against the end plates, effectively resisting rotationof the

large aggregate particles. Figure 5.11 shows two examples, shearand diametral loading, where

this mechanism is likely to produce an effective increase of stiffness modulus, resistance to

permanentdeformation, and fatigue life.

Direct evidence is provided by comparison of simple shear test results for uncut, 4 x

2.5-inch (101.6 x 63.5 mm) gyratory specimens with those for specimens cored and cut from

6 x 9.5-inch (152.4 x 241.3 mm) cylinders. In Figure 5.12, the difference between the

performance of the two types of specimens is illustrated. The uncut specimens have a larger

proportion of aggregates able to move under stress, leading to rapid failure once aggregate

interlock has been overcome by the shea_ng force.

5.3 Summa_

Based on the stated objectives of specimenpreparationand air-void measurementand on

the dataand observationspresented herein, the following conclusions can be drawn:

1. Whether a specimen is uncut (as-molded) or cut (sawed or cored from a larger

compactedmass) may have significanteffects both on its mechanical behaviorand

on its measured air-void content.

2. In comparison to cut specimens, specimens with uncut surfaces axe less

homogeneous in their distributionof air voids and exhibit more segregation of

aggregate and greater migration of fines and asphalt to their surfaces.

3. Cutting the surfacesof laboratoryspecimensallows more accurate comparisonof

their properties with those of field cores.

165

SHEAR

CUT i; _ AS MOLDED

SURFACES (NO CUT SURFACES)

[i

DIAMETRA_

Figure 5.11 Illustration of Boundary Effects of Cut Specimen Surfaces inShear and Diametral Testing

166

10000

]ooo- B_-'_-'<'_-. ¢- G-UNCUT-(a.SZwwp)SHEARCREEP - =-n-_u__

MODULUS(PSl) _l-mlt -= G-CUT-(3.T%wwp)I00

I0

0.1 1 I0 IO0 1000 10000

TIME(SEC)

Figure 5.12 Effect of Cut Surfaces on Shear Creep Modulus,Gyratory Specimens (1 psi = 6.89 kPa)

167

4. Measurement of air-void content is critical to understanding specimen

performance in testing. The use of parafilm for immersed weighing produces

consistently sensitive and reproduciblemeasurements and easily allows further

testing of the specimen after air-voiddetermination.

Based on these conclusions it is recommendedthat laboratory specimens be preparedby

coring and/or cutting from a larger compacted mass and that cut specimens be allowed to dry

naturally, or surface dried with compressedair, prior to in-air weighing. All specimens should

be encased in parafflm for immersed weighing.

168

6.0 SUMMARY AND RECOMMENDATIONS

The primary purpose of this study was to determine the extent to which method of

laboratory compaction (Texas gyratory, kneading, and rolling wheel) affects fundamental mixture

properties of importance to pavement performance in service. Permanent deformation and

fatigue were selected as the performance parameters of greatest concern, and laboratory creep

and fatigue tests were used to measure mixture properties that were likely to correlate well with

field performance.-A total of i6 asphalt-aggregate mixtures--varying widely in asphalt source,

aggregate type, asphalt content, and air-void content--were tested. All aggregates were densely

graded with a 3/4-inch (19.0 ram) maximum size. Principal findings and recommendations of

the investigation follow.

6.1

1. The method of compacting asphalt-aggregate mixtures affects the way test specimens

respond to laboratory loading. Fundamental engineering properties with expected links

to pavement performance are significantly affected, both statistically and in terms of

practical engineering consequence.

2. Regarding resistance to permanent deformation, kneading compaction produces the most

resistant specimens, followed in order by rolling-wheel and gyratory compaction. In

genera/, kneading compaction also produces specimens that are most sensitive to mixture

composition. Specifically, the properties of kneading specimens are more sensitive to

aggregate type while those of gyrator), and rolling-wheel specimens have greater

sensitivity to asphalt type.

3. Regarding resistance to fatigue--under controlled-stress conditions--gyratory compaction

produces the most resistant specimens followed in order by rolling-wheel and kneading

169

compaction. Sensitivity to mixture composition favors rolling-wheel specimens,

however, and any differences in sensitivity between gyratory and kneading specimens are

small.

4. Regarding mixture stiffness, the various test methods do not produce uniform results.

The shear creep tests seem to be most reliable, however, and, on this basis, the stiffest

mixtures are those produced by rolling-wheel compaction, followed in order by kneading

and gyratory compaction. Complex modulus testing confirmed the finding that kneading

specimens are stiffer than gyratory specimens. The stiffness of rolling-wheel specimens

reflects slightly more sensitivity to mixture composition than either gyratory or kneading

specimens.

5. These effects can at least partially be explained in terms of differences in the interparticle

contact and aggregate "structure" likely to be developed by the different compactors. At

least for dense-graded mixtures, the effect of compaction method is intensified at low

levels of air voids and presumably at high levels of asphalt content, conditions

representative of a larger percentage of voids in the mineral aggregate that are filled with

asphalt.

The kneading compactor imparts a large, localized shearing force and permits

considerable movement within the loose but densifying mixture: it is very effective in

producing specimens with maximum interparticle contact and a tight, interlocking

aggregate structure. As a result, mixtures prepared under kneading compaction are most

sensitive to aggregate angularity and surface texture. The asphalt serves to assure that

the integrity of the interparticle contact and aggregate structure is maintained.

Although it does serve to reorient the larger aggregate particles, the shearing

motion of the Texas gyratory compactor is much less severe than that of the kneading

170

compactor, is applied fewer times, and is much less effective in inducing structure within

the aggregate particles. Also, during compaction the loose mixture is under considerable

"hydrostatic _ pressure from the compressive compaction loading. Under such pressure,

the gyratory action very effectively densifies the mixture by expelling air but is less

successful in forcing intimate interparticle contact. Because the asphalt tends to separate

the aggregate particles, the properties of mixtures prepared by gyratory compaction are

more sensitive to the properties of the asphalt binder.

A steel-wheel roller, with a vibratory capability that was needed only

occasionally, was used to represent rolling-wheel compaction techniques in this study.

Although flow of the loose mixture during compaction was restrained to some extent by

lateral confinement, the movement was apparently sufficient to induce considerable

interparticle contact. Rolling-wheel specimens were ranked between kneading and

gyratory specimens in their resistances to permanent deformation and to fatigue.

However, they were stiffer under transient loading than either gyratory or kneading

specimens.

6. With minor exception, the general influence of mixture and test variables on fundamental

mixture properties was unaffected by compaction method. Effects of mixture and test

variables are summarized as follows:

Resistance to Permanent Deformation

Asphalt Type: AAK-1 superior to AAG-1Asphalt Content: _Optimum _ superior to _high _Aggregate Type: Granite (RB) superior to chert (RL)Air-Void Content: Low voids superior to high voidsTemperature: More resistant at low temperatureStress Level: More resistant at low stress

171

Resistance to Fatieue

Asphalt Type: AAG-1 superior to AAK-1Asphalt Content: "Optimum" superior to "high"Aggregate Type: Granite (RB) superior to chert (RL)Air-Void Content: Low voids superior to high voidsTemperature: More resistant at low temperatureStress Level: More resistant at low stress

Stiffness Modulus

Asphalt Type: AAK-1 stiffer than AAG-1, 104-140°F (40-600C)AAG-1 stiffer than AAK-1, 32-68°F (0-32°C)

Asphalt Content: "Optimum" stiffer than "high"Aggregate Type: Granite (RB) stiffer-than chert (RL)-..Air-Void Content: Low voids stiffer than high voidsTemperature: Stiffer at low temperatureStress Level: Stiffer at low stress

7. Compactors within a given genre may produce specimens having quite different

engineering properties. Taking kneading compaction as an example, a variety of

multistage compaction strategies can be employed and critical parameters--the size of the

tamping foot, the size of the specimen, the number of layers, the number of tamps on

each layer, and tamping pressure--can be varied over rather wide ranges. Such variations

almost certainly affect the aggregate structure and, in turn, are expected to affect many

mixture l.roperties linked to pavement performance. The limited comparisons reported

herein between the Texas and Corps of Engineers gyratory compactors provide tentative

verification of possibly significant differences among compactors of the same general

type,

8. Measurements of permanent deformation accumulated under repetitive loading, although

limited in extent, generally corroborated results of the creep tests.

172

6.2 Recommendations

1. Because one of the most critical mixture characteristics is air-void content, it must be

measured as accurately and as with as much reliability as feasible if test results are to be

meaningful. Experimentation with several different methods for measuring air voids

yielded widely divergent estimates, especially for mixtures having high voids and uncut

surfaces. The use of parafilm for immersed weighing is clearly advantageous both for

its accuracy and its reliability. It will be used in test systems evaluation where its use

would not negate or confound work already completed or currently underway. In

addition, it will be used to measure air voids in both the cooperative test program and

the expanded test program. If its use continues to be successful throughout these

extensive test efforts, parafilm will be recommended, thereafter, for adoption in relevant

standards and specifications.

Specimens that have been exposed to water as a result of coring and/or sawing

must be dried before measurements of in-air weight are taken. Surface drying with

compressed air is an acceptable substitute for natural drying by drainage and evaporation.

2. Specimens with cut surfaces, either sawed or cored, are preferred to those with as-

molded surfaces: test results are expected to be more precise as a result of the greater

homogeneity, air voids can be more accurately measured, and comparisons with

specimens extracted from in-service pavements are potentially more valid. Specimens

for use in the test-systems-evaluation phase of the study will have cut surfaces, providing

available equipment permits and providing the practice does not negate or confound work

already completed or currently underway. In the cooperative test program and the

expanded test program, which follow test-systems evaluation, all specimens will be cut

from larger compacted masses. If this effort proves successful, sawing and/or coring

173

will be recommended, thereafter, in the preparation of all laboratory specimens.

Compaction procedures that are recommended for future use and for adoption in relevant

standaxds and specifications must then be able to accommodate masses from which

specimens of the necessary size and shape can be extracted by sawing and/or coring.

3. Because compaction method can have such a profound impact on fundamental mixture

properties, consistent testing demands use of a single compaction procedure. The

gyratory compactor places excessive emphasis on the asphalt binder and does not

accurately portray the role of asphalt-aggregate interaction in the performance of properly

constructed pavements. Furthermore, the shapes and dimensions of specimens produced

by gyratory compactors are limited. Although the kneading compactor is more adaptable

for producing a variety of sizes and shapes, it may create a more stable aggregate

structure than is commonly developed by conventional construction practice, thereby

failing to capture the role of the asphalt binder in properly performing pavements.

Rolling-wheel compaction seems to be the preferred procedure for laboratory

compaction. Among the methods investigated, it appears to best duplicate field-

compacted mixtures. It can be used to produce specimens of the necessary sizes and

shapes and can accommodate aggregates of all reasonable sizes. Rolling-wheel

compaction is a comparatively easy procedure and is fully compatible with the

requirement for cut specimen surfaces and for the rapid fabrication of a large number of

specimens as is likely to be required for AAMAS-based designs and investigations.

Because of the advanced stage of the test-systems-evaluation phase of the SHRP

A-003A project, it is not feasible to require that all specimens be prepared by rolling-

wheel compaction. When specimens have been prepared by other procedures during this

phase, very careful assessment may be required to discriminate between the effect of

174

7.0 REFERENCES

Del Valle, H. (1985), "Procedure - Bulk Specific Gravity of Compacted Bituminous MixturesUsing Parafilm-Coated Specimens, =Chevron Research Company, Richmond, California.

Francken, L. (1977), "Permanent Deformation Law of Bituminous Road Mixes in RepeatedTfiaxial Compression," ]_I._AlJ/)_, Fourth International Conference on StructuralDesign of Asphalt Pavements, University of Michigan, Ann Arbor, pp. 483-496.

Hicks, R. G., J. A. Deacon, L. Painter, and C. L. Monismith (1990), "Laboratory Study Planfor SHRP Project A-003A," T_chnical Memorandum 89-8, Institute of TransportationStudies, University of California, Berkeley.

Hills, L E., D. Brien, and P. J. van de Loo (1974), "The Correlation of Rutting and CreepTests in Asphalt Mixes," Paver 1P-74-001, Institute of Petroleum, London.

Monismith, C. L., J. A. Epps, D. A. Kasianchuk, and D. B. McLean (1971), "Asphalt MixtureBehavior in Repeated Flexure," Report No. TE 70-5, University of California, Berkeley.

Monismith, C. L. and A. A. Tayebali (1988), "Permanent Deformation (Rutting) Considerationsin Asphalt Concrete Pavement Sections," _A_[L_, The Association of Asphalt PavingTechnologists, Volume 57, pp. 414-463.

Sousa, J. B., J. Craus, and C. L. Monismith (1990), "State of the Art on Rutting in AsphaltConcrete," TM-UCB-A-0Q_A-89-4, Institute of Transportation Studies, University ofCalifornia, Berkeley.

Tangella, S. C. S. Rao, J. Craus, J. A. Deacon, and C. L. Monismith (1990), "SummaryReport on Fatigue Response of Asphalt Mixtures," TM-UCB-A-003A-89-3, Institute ofTransportation Studies, University of California, Berkeley.

Vallerga, B.A. (1951), "Recent Laboratory Compaction Studies of Bituminous PavingMixtures," _, The Association of Asphalt Paving Technologists, Volume 20,pp. 117-153.

Von Quintus, H. L., J. A. Scherocman, C. S. Hughes, and T. W. Kennedy (1988),"Development of Asphalt-Aggregate Mixture Analysis System: AAMAS, Phase II -Volume I, Preliminary Draft Final Report," Brent Rauhut Engineering Inc., Austin,Texas 1988.

176

compaction method and the effect of each test system underevaluation. Exclusive use

of rolling-wheel compactionwill be mandatoryin both the cooperative test programand

the expanded test program. If this effort proves successful, rolling-wheel compaction

will likely be recommended for adoption in relevant standards and specifications.

Recommending an alternate to rolling-wheelcompaction for routine mix designs may be

an eventual consideration.

175

1.0 INTRODUCTION

Documentation of the methods used to prepare asphalt-aggregate specimens for SHRP

Project A-003A is necessary for several reasons. First, it is required for later evaluation of the

effects of specimen preparation on test results. Second, a common procedure for specimen

preparation is necessary to provide uniformity of specimens in order to avoid confounding the

experiment. Third, the procedure can be used as a framework to include any specimen

preparation variables that are to be introduced into the experiment.

1.1 Effective Date

The first version of this document (Version 1.0 dated August 29, 1989) described

procedures in effect at the University of California at Berkeley (UCB) laboratory during the

mixture design and compaction study phases of this project. This revised version (Version 2.0),

while continuing to describe procedures in effect during the compaction study, also addresses

the preparation of specimens for all other testing at UCB as well as at other laboratories

participating in SHRP Project A-003A. Version 2.0, with effective date of May 8, 1990,

remains in effect until superseded by an updated version.

1.2 Scooe

Described herein are the following procedures:

1. Aggregate batching and handling procedures;

2. Asphalt-aggregate mixing and curing of mixtures;

3. Asphalt-aggregate compaction, currently including the following methods;

a. Texas gyratory compaction of 2.5 x 4.0-inch (63.5 x 10!.6 ram) briquets,

b. Texas gyratory compaction of 6.0 x 8.0-inch (152.4 x 203.2 ram)

cylinders cut to make 4-inch (101.6 ram) diameter cylinders,

A.2

APPENDIX A

SPECIMF_N PREPARATION

c. Kneadingcompactionof 2.5 x 4.0-inch (63.5 x 101.6 ram)briquets,

d. Kneading compaction of 4.0 x 8.0-inch (101.6 x 203.2 mm) cylinders,

e. Kneading compaction of 3.5 x 3.5 x 15-inch (88.9 x 88.9 x 381 ram)

beams cut to makespecimens of different sizes,

f. Rolling-wheel compactionof 16.0 x 16.0 x 9-inch slabs (406.4 x 406.4 x

228.6 ram)cut to make specimensof different sizes; and

4. Procedures for determining specimen air-void content.

A.3

2.0 AGGREGATE BATCHING

2.1 D ryin2 and Sievin_

Aggregate to be used for specimen preparation must first be dried and sieved according

to the following procedure, taken from "Materials Handling Protocol" (Hicks, 1989).

1. All aggregate should be checked to be sure that it contains no refuse material,

such as sticks, asphalt, bits of concrete, other types of aggregate, or any other

material that will affect the performance of the asphalt-aggregate specimens to be

made.

2. Spread the aggregate on sheets or trays and dry in a forced air oven overnight

(about 20 hours) at about 300°F (149°C) so that it arrives at a constant mass.

If sheets are not available and the material is dried in barrels or similar

containers, more time may be needed to completely dry the aggregate.

3. After removing the material from the oven, allow it to cool, and then place it in

a storage container for sieving. The storage container should be of a type that

allows complete sealing from the outside air (and rain, if outdoors!).

4. Material should he sieved in a Gilson type machine. The sieving time used at

UCB is 5 minutes. Any convenient sieving time between about 4 and 30 minutes

can be used, provided that appropriate batching and gradation analysis

procedures, described herein, are followed.

The screens should be cleaned regularly, and checked to be sure that they are neither

clogging with fine material nor allowing coarse material to pass. Each time material is taken

off the screens, it should be placed in an intermediate container, such as a 5-gallon (0.0189 m3)

bucket, before emptying into the final sieved aggregate storage container, typically a 55-gallon

(0.208 m3) barrel. This will help prevent accidently emptying a sieve full of the wrong size

A.4

material into a large container, and also allow checking of the sieved material for leaking or

overloaded sieves.

2.2 Calculation of Batching Weights

If the specimens are prepared using the sieved materials in exactly the same percentages

as required by the mixture-design gradation, the actual specimen gradation will probably be

somewhat different. The difference is due to incomplete sieving of the material--whatever the

sieving time-and dust which adheres to the surface of the larger size aggregates during sieving.

Also, due to difficulties in sieving large quantities of material into different sizes below the No.

100 sieve, a minus 100 (- 100) material is sometimes being used at UCB for granite (RB).

However, for granite (P.B) as well as chert (ELL the actual amounts of materials passing the No.

100 and No. 200 sieve must still be determined. To account for these problems the following

procedures should be followed.

1. The following gradations have been or are being used for this project:

a. Medium Gradation No. 1. This gradation, shown in Table A. 1, was used

during the mixture design and compaction studies.

b. New Medium Gradation No. 2. This gradation, shown in Table A.2, is

being used for all specimens of medium gradation made for tests after the

compaction study. The asphalt contents from the mixture design were

checked with this gradation and found to be acceptable. However, it

should be remembered that the No. 1 and No. 2 medium gradations are

not interchangeable.

c. Coarse Gradation. This gradation is shown in Table A.3. It was not used

during the compaction study.

A.5

Table A.1 Medium Gradation No. 1

Sieve Size Percent Passing by Mass Percent Retained by Mass

1 in 100 0

3/4 in 95 5

1/2 in 80 15

3/8 in 68 12

No. 4 48 20

No. 8 35 13

No. 16 25 10

No. 30 17 8

No. 50 12 5

No. 100 8 4

No. 200 4 4

Pan 4

Note: lin =25.4mm

A.6

Table A.2 Medium Gradation No. 2


1 in I00 0

3/4 in 95 5

1/2 in 80 15

3/8 in 68 12

No. 4 48 20

No. 8 35 13

No. 16 25 l0

No. 30 17 8

No. 50 12 5

No. I00 8 4

No. 200 5.5 2.5

Pan 5.5

Note: lin-25.4mm

D

A.7

Table A.3 Coarse Gradation

I


I in I00 0

3/4 in 90 I0

1/2in 72.5 17.5

318 in 60 12.5

No. 4 40 20

No. 8 27 13

No. 16 18 9

No. 30 10 8

No. 50 8 2

No. I00 5.5 2.5

No. 200 3 2.5

Pan 3

Note: lin =25.4mm

A.8

2. Using the top size of the gradationand the informationin Table A.4, determine

the minimumrequired mass of aggregate.

3. Using the appropriate mixture-design gradation (Tables A.1, A.2, or A.3),

calculate the percentby weight passingeach individual sieve, and the percent by

weight retained on each individual sieve. Form 1 (Figure A. l) can be used for

this calculation.

4. With the total aggregate weight and the percentretainedon each sieve, calculate

the weight of each size to be included in the specimen, using Form 1 (Figure

A. 1). Then batch the specimen.

5. Wash the hatchedspecimen through a No. 200 sieve, following the procedure in

ASTM C 117-80. Calculate the weight of material washingthroughthe No. 200

sieve by drying the retainedmaterialto constant weight, weighing it, and finding

the difference from the original weight. Form 2 (Figure A.2) can be used for this

calculation.

6. Dry sieve the material retainedon the No. 200 sieve for 30 minutes, recording

the results on Form 3 (Figure A.3). Add the weight of material passing the No.

200 sieve from the wet-sieve analysis to the like materialfrom the dry sieving.

7. Calculate the cumulativepercent weightpassing each sieve and the percent weight

retained on each sieve on Form 3 (Figure A.3).

8. Compare the results of the analysis and make any appropriate changes in the

batchingoperation. For sieve sizes No. 50 and smaller, the gradation should be

within 4- 0.5 percent of the percent mass retainedon that sieve size in the

required gradation. For sieve sizes No. 30 and larger _+1.0 percent is allowed.

A.9

Table A.4 Minimum Masses Required for Analysis Specimens(from ASTM C 117-80)

Nominal Maximum Size of Panicles Minimum Specimen Mass (g)

2.36 mm (No.g) 100

4.75 mm (No. 4) 500

9.50 mm (3/8 in) 1000

19.00 mm (3/4 in) 2500

37.50 mm (1.5 in or larger) 5000

A. IO

UNIVERSITY OF CALIFORNIA - BERKELEYRICHMOND FIELD STATION

BITUMINOUS MATERIALS LABORATORYSHRP PROJECT

WEIGHTS OF MATERIALS TO ASSEMBLE SPECIMENS

Calculated by Checked by

Test Specimen Date

Aggregate Gradation

Test Specimen Codes

(A) Total Wt. of Specimen

Percent Passing Percent Retained Wt. per SpecimenSieve Size Cumulative on One Size of One Size

03) (Bi - Bi+l) (Bi - Bi+I)*A

2"

1"

3/4"

1/2"

318"

#4

#8

#16

#30

#50

#100

#200

Passing #200

TOTAL

Figure A.1 Form No. 1

A.11


BITUMINOUS MATERIALS LABORATORYSHRP PRO/ECT

WET SIEVE ANALYSISASTM C ll7


Test Specimen Date

Aggregate Gradation

Test Specimen Codes

Nominal Total Wt. of Specimen

(A) Tare Wt. of Container

(B) Wt. of Container and Dry Specimen

(C) Wt. of Container and Dry SpecimenAfter Washing Through #200 Sieve (R200)

(D) Wt. of Original Material (B) - (A)

(E) Wt. of R200 Material AfterWashing (C) - (A)

(F) Percent Passing #200 Sieve [(D) - (E)]/(D)


A.12



DRY SIEVE ANALYSISASTM C 136


Test Specimen Date

Aggregate Gradation

Test Specimen Codes

Nominal Wt. of Specimen

Weight Retained Percent Retained " l_ercent PassingSieve Size on One Size Cumulative Cumulative

(A) (C)=(Sum of Ai)/(B) (1 - Ci)*1002"

1"

3/4"

1/2"

318"

#4

#8

#16

#30i

#50

#100

#200t

Pan

(From Form 2)

(13)TOTAL WT

D Figure A.3 Form No. 3

A.13

9. Repeat Steps 3 through7 until the specimen gradationmeets the requirementsof

the mixturedesign. File the resultsof all analyses and hatchingregimes.

10. Periodically perform the above analysis on random specimens taken out of the

specimen production line. If you are making specimens smaller than the

minimum mass, combine several specimens.

2.3 Res'_lts to Date of Batchine Analyses

The results of wet sieving, and the batching regimes currently being used at UCB, are

available by telephone (415/231-9560, 415/231-9513, or 415/231-9588). The batching regime

to be used with sieved material sent to subcontractorswill be sent with each shipment. The

procedure, described in Section 2.2, should be carried out by subcontractorswhen receiving

unsieved material.

2.4 Receivine and Processine Unsieved Material or Mal[erialFrom Sources Other ThanUCB

2.4.1 Receiving. Whenever any materialarrives unsieved, or from sources other than UCB,

it should be checked to be sure thatit has the same characteristicsas any previous material that

may have been used. Specifically, new material should be checked for the following

characteristics.

1. Visual appearance. The material should be visually examined to assure that it

appearsthe same as previous materials of the same size. Indicationsare color,

specific gravity, microtexture,and gradation.

2. Angularity. The new materialshould have the same percentage of crushed faces

as previous material of the same size.

A.14

3. Dustiness. The material should be unwashed or washed, the same as previous

materialused of the same size.

4. Gradation. If different sizes of the material have different characteristics, the

gradations of the materialsreceived should be compared with previous materials

to be sure that the characteristics of the different sizes occur in the same

proportionsas for previous specimens.

2.4.2 _. In some cases the material retainedon a given sieve size originates from

two differentaggregatesources. For example, with chert (RL), the material retainedon the No.

4 and No. 8 sieves comes from two types of material, one completely crushed (called #6 by the

quarry)and the other completely uncrushedand smooth (called #8-200). Aggregates fromthese

two types must continuouslybe sieved and batehed in the same proportions (and be well mixed

together) to insure uniformpropertiesof the asphalt-aggregatespecimens. The ratios for sieving

this aggregate used at UCB are shown in Table A.5.

The terms RR Ballast, #2, #4, #6, 8-200 and -200 are the names used by the quarry for the

different types of material and are only the nominal sizes of the material. The term "grindings"

refers to chert (RL) material that was retained on the No. 100 sieve, and then ground between

rotating plates to make finer material.

If there are any questions regarding the uniformity of materials received from UCB, or

other sources, do not hesitate to call UCB.

2.5 Storage of Batched Agere_,ates

After batching a set of specimen aggregates, each specimen should be weighed to check

for mistakes. The aggregates should be stored at room temperature in covered containersuntil

placed in the oven in preparationfor mixing. Each container should have an identificationtag

A.15

Table A.5 Sieving Regimes for Chert (1_) Aggregate

I

Components Regime

0.5 partsRR Ballast:Coarse 2.0 parts #2;

2.0 parts #4

1.0 parts #6;Medium 2.0 parts #8-200

Fine Grindings, -200any proportion

I

A.16

inside, with a code for the aggregate type and gradation. Metal dog tags have been found to

work well. ,

A.17

3.0 MIXING AND CURING OF ASPHALT-AGGREGATE SPECIMENS

3.1 Heating of Materials for Mixing

Mixing Temperature 1. The optimum mixing temperature is selected from a Bitumen

Test Data Chart (BTDC). The temperature should correspond to a viscosity of 170 _+ 20

centistokes (0.000170 + 0.000020 m2/s), based on the original asphalt properties. Figure A.4

shows a BTDC with plots of the original consistency data for the two asphalts used in the initial

A-003A laboratory studies. The optimum mixing temperatures are as follows:

AAG-I: 2880F (142"C)

AAK-I: 3200F (160"C)

Use of the optimum mixing temperature ensures that each asphalt will have a similar consistency

when mixed, which, in turn, ensures that similar film thicknesses will result. The ranges of

mixing temperature given below should be regarded as "broad band" guidelines, and the

approach outlined above should be used to determine mixing temperature for a specific asphalt.

Broad Guidelines for Mixing Temperatures

Temperature RangeGrade

Minimum Maximum

AC-2.5, AR-1000, or 200-300 Pen. 210°F (99°C) 250°F (121°C)

AC-5, AR-2000, or 120-150 Pen. 230°F (110°C) 275°F (135°C)

AC-10, AR-4000, or 85-100 Pen. 250"F (121"C) 300*F (149"C)

AC-20, AR-8000, or 60-70 Pen. 270°F (132"C) 325°F (163°C)

AC-40, AR-16000, or 40-50 Pen. 270"F (132"C) 325"F (163"C)

1The section on mixing temperatures was prepared by C. A. Bell of Oregon StateUniversity. References include the following: AASHTO T 246-82 (1986), AASHTO T 247-80(1986), ASTM D 1560-81a, ASTM D 1561-81a, TAI MS-2 (1984), and California Test 304(1978).

A.18

J. Mixing Temperature

A.AGe AR 4000: 142°C (288"I=)

AAKo AC 30: 160"C (320"F)

ZolI--4t---

CompactionTemp = 110°C (230°I=)CK 6

2 tOLu_o"7 5LU 10n

10 1

if?

1 moe--t

0

2_.0>-

if t f _ _ I i

i "[ I I i C3I ! , I

' I i i , co>.; I i 01

I I I' I I

0tO

-40 0 25 60 135 _.60

TENPERATURE (C)

Figure A.4 BTDC Showing Mixing Temperature Selection (1 poise = 0.1 Pa*s)

A.19

The optimum mixing temperatures were determined using the Bitumen Test Data Chart as shown

in Figure A.4. The temperature for mixing with asphalt AAK-1 was lowered due to smoking

to 300°F (149°C).

Aggregate. The aggregate should be heated at the mixing temperature, which depends

on the type of asphalt being used in the mixture. Temperatures used for the AAG- 1 and AAK- 1

asphalts are summarized in Table A.6. The aggregate containers should be kept covered during

heating to ensure that fines are not lost due to blowing air in the oven. If the specimen

containers do not allow covers, shake the container until the fine materials segregate to the

bottom of the container, leaving a layer of coarser materials on top that will not blow away.

It usually best to heat the aggregates overnight in the oven in preparation for mixing the

following day.

Asphalt. The asphalt should be heated at 275°F (135°C) for distribution to small cans,

following the procedures in the Materials Protocol (Hicks, 1989) (this is called the first heating).

In preparation for mixing, the asphalt in the small cans is heated (second heating) at the

appropriate temperature, with periodic stirring to ensure uniform heating. The lid should be

kept on the asphalt during heating.

A one liter (1 kg) specimen in a tin can should be heated for a minimum of about 90

minutes to reach uniform temperature throughout. If the specimen is not used within 3 to 3.5

hours from the start of heating it should be discarded. These heating times have been used at

UCB, and it is suggested that they be followed strictly to provide for uniformity of asphalt aging

in the specimens. The heating of the asphalt must only be done once and must be continuous.

Asphalt cannot be heated, cooled, and then reheated. The temperature of the oven and asphalt

should be carefully monitored. The asphalt should also be discarded if it burns due to

A.20

Table A.6 Mixing Temperatures

Asphalt Type Mixing Temperatures

AAG-I (V) Medium Gradation No. 1 275°F (135°C)

AAK-I 03) Medium Gradation No. 1 275°F (135°C)

AAG-1 (V) Medium Gradation No. 2 284°F (140°C)

AAK-1 (B) Medium Gradation No. 2 300°F (149°C)

A.21

overheating. The AAK-I mixing temperature for Gradation No. 2 is not at the optimum for

mixing viscosity due to extreme smoking for temperatures over 300°F (149°C).

Equipment. The mixing bowl and stirring device should be preheated to the mixing

temperature.

3.2

Mixing Equipment. Any mixing equipment can be used that meets the following

requirements:

1. It can be heated to approximately the asphalt mixing temperature, and that

temperature can be maintained during mixing.

2. The equipment causes the aggregate to be uniformly and completely coated with

asphalt in the mixing time period used (assuming that the percent asphalt being

used allows complete coating).

Mixing Time. Using the granite (RB) and chert (RL) aggregates; the AC-30 (AAK-1)

and AR-4000 (AAG-1) asphalts; and the percentages of asphalts (from the mixture designs

carried out at UCB) shown in Table A.7, it has been found that four minutes of mixing meets

the above requirements.

The equipment used at UCB consists of a revolving blade inside a steel bowl, or a

revolving bowl and a fixed steel blade. At the midpoint of the four-minute mixing period, the

revolutions are stopped, and accumulated fines and asphalt are scraped off the blade and the edge

of the bowl and spaded into the mixture. Form No. 4 (Figure A.5) can be used for mixture

calculations.

A.22

Table A.7 Percentages of Asphalt To Be Used in Mixing

Percent Percent

Asphalt Asphalt by Asphalt UCBAsphalt Aggregate Content Weight of by Code

Aggregate Weight ofMixture

AAG-1 Chert (RL) Low 4.1 3.9 V0T(Mix Design)

AAG-1 Chert (R.L) High 4.8 4.6 VIT

AAK-I Chert (RL) Low 4.3 4.1 B0T(Mix Design)

AAK-1 Chert (RL) High 5.0 4.8 B1T

AAG-1 Granite (R.B) Low 4.9 4.7 VOW(Mix Design)

AAG-1 Granite (P.B) High 5.5 5.2 V1W

AAK-1 Granite (P.B) Low 5.1 4.9 BOW(Mix Design)

AAK-1 Granite (R.B) High 5.7 5.4 B1W

A.23



SPECIMEN PREPARATION RECORD

Date


Test Specimen Date/Time of Mixing

Aggregate Gradation

AC Heating Time/Temp Aggregate Heating Time/Temp

Aggregate % Asphalt by Weight of AC CompactionSpecimen Specimen Code Weight Weight of (A) x (13) Date and Hour

(A) Aggregate (B)

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.


A.24

3.3

After mixing, the asphalt-aggregate mixture should be placed in containers and

maintained in a 140°F (60°C) oven for 15 hours. This operation, called curing, allows any

asphalt absorption by the aggregate to take place before compaction or other use of the mixture.

The mixture need not be cured directly after mixing, but should be done within several days of

mixing. A convenient procedure is to place the mixture in the oven at 5 p.m., and remove it

at 8 a.m. the next morning.

3.4 Storage of Cured Mixtures

Cured mixtures can be stored in closed containers at room temperature or a little cooler.

If a mixture is not used within approximately 90 days, it should be discarded. Little or no

difference in specimen strength has been found among specimens made with mixtures used

within this time frame.

P

A.25

4.0 COMPACTION

4.1 Heatin2 the Mixture for Compaction

The mixture should be placed in a 240°F (116°C) forced air oven until it reaches a

uniform temperature throughout. For 1,200 g to 7,000 g specimens, this has required a

minimum heating time of 1.5 hours at UCB. The oven temperature may require adjustment to

result in this mixture temperature, especially if the oven doors are being continuously opened

and closed to move specimens and equipment. The mixture should be discarded if it is not used

within 3 to 3.5 hours after it is placed in the oven, or if it bums. The heating of the mixture

must be continuous and must only be done once.

Whenever possible, all molds and tools that contact the mixture during compaction should

be maintained at the compaction temperature.

4.2 Compaction Procedur¢$

Compaction procedures depend on the type of device used for compaction, the air-void

content, and other required specimen properties. The results of specimen preparation to date

at UCB are described in the following sections. This information will be updated as required

in future versions of this document. As mentioned earlier, the materials used at UCB to date

are granite (P,B) and chert (RL) aggregates, and AC-30 (AAK-1) and AR-4000 (AAG-I)

asphalts.

4.2.1 Texas Gyratorv Compaction. 2.5 x 4-Inch (63.5 x 101,6 nun) Briquets. Two

variations of the same method are used for compacting these specimens, depending on the

mixture and the air-void content to be achieved. The methods are adapted from Texas Method

Tex-206-F and ASTM D 4013-81 and require the use of the small Texas gyratory compaction

machine.

A.26

First, approximatelyhalf of the specimen is put in the mold and pusheddown with the

bent spoon tool to level it. The second half of the specimen is then put in the mold, and

likewise pushed into the mold.

Method A. The levels of compactionstressarevarieddependingon the desired air-void

content. However, the ratios between the different stresses (pregyration, end point, and

consolidation)are keptthe same. The specimengenerally contains 1,200 g of material.

After the pressure is brought to the prescribed pregyrafion level, the hydraulic load

handleshould then be placed in its uppermostposition. The specimen is then put in its inclined

position (six degrees from horizontal) and gyrated one gyration set (three revolutions per

gyration set). As soon as the gyration set has been completed, the specimen is returnedto the

horizontalposition, and the load handle is pushed through one downward stroke which should

take place in one second. If the pressureon the specimen does not reach the end point stress

with this one stroke, the specimen is again broughtto the pregyrationstress (even if the one

stroke put the pressure above the pregyrationstress, in which case the stress is lowered to the

pregyration stress), and the process is repeated. If the one stroke reaches the end point stress,

then the specimen is loaded to the consolidationstress, at a rate of one full stroke of the load

handleper second. The specimen is then immediately unloadedand left to cool.

Method B. If the desired air-void content(usuallya very high or very low one) cannot

be obtained using Method A, the following procedureis used. First the specimen is brought to

a pregyration pressure of 50 psi (344 kPa) gauge. Then a certain numberof gyration sets is

carriedout, each time loading the specimento thepregyration stress in the horizontal position.

When the gyration sets have been completed, the specimenis loaded to the consolidation stress

of 1,000 psi (6,890 kPa) gauge using the same procedureas in Method A.

A.27

4.2.2 Texas Gvrato_ Comt_action.4 x 8-Inch (101.6 x 203.2 mm) Cylinders. This method

is adapted from Texas Method Tex-126-E and requires the use of the large Texas gyratory

compaction machine.

The first step for compaction is to measurethe thickness of the base plate, two tin spacer

plates, and the compressor ram head from its bottomup to where it has a horizontal plane. The

total combined thickness for these parts is about 2.21 inches (56 mm) on the machine loaned to

UCB by the Texas State Highway Department. Next, the diameter of the molds to be used

should be checked. They should be 6 inches (152.4 mm) in diameter. With these dimensions

the volume of the specimen can be determined and can be measured during compaction using

the measuring device provided with the compaction machine (a small arm that sits on the ram

head when measuring the height of the specimen).

Next, using the desired height of the specimen (here about 2.032 din), the required

air-void content, and the maximum specific gravity of the mixture (by the Rice method), the

required weight of material is calculated. An example calculation is shown in Figure A.6. For

an 8-inch (203.2 mm) final height, approximately9,000 to 10,500 g of material are needed.

The mixture is placed in the mold in three equal lifts. First a tin plate is placed on the

gyrator), mold bottom plate. Then the first lift is placed in the mold. This lift is rodded 15 to

20 times around the wall and then 15 to 20 times in the center. The second and third lifts are

then placed, each with the same rodding. The top tin cover is then placed on the material, and

the mold is slid under the compactorram. The ram head is broughtdown against the material,

so that it is inside the top of the mold. The mold is then inclined to 6 degrees.

For higher air-void specimens (more than about 6-percent air voids) the specimen is

subjected to a gauge pressure of 69 psi (475 kPa), while for low air-void specimens 0ess than

about 6 percent) the specimen is subjectedto 104 psi (716 kPa). The specimen is then gyrated,

A.28

Specimen Height = 9 in. = 2.29 dm

Specimen Size Ram Thickness = 0.56 dm

Measured Height = 2.29 + 0.56 = 2.85 dm

Specimen Diameter = 6 in. = 1.52 dmVolume

Specimen Volume = I"(1.52/2) 2 x 2.29 = 4.16 dm 3

Mixture = VOT

Rice Maximum Density = 2.435Specific Gravity

Desired Air Voids = 5 %

Density Required = (1 - 0.05) 2.435 = 2.313 kg/dm 3

Mass of Material Mass Required - Volume x Density= 4.16 dm3 x 2.313 kg/dm 3 = 9.622 kg

Figure A.6 Example Calculation for Compaction UsingLarge Texas Gyratory Compactor

A.29

while maintaining this compaction pressure. As the specimen densities the ram head must be

lowered using the hand pump to maintain the compaction pressure. When the ram head has

reached approximately the correct height, the gyration is stopped, the specimen is brought to

horizontal, the measuring device is placed on the ram head, and the height of the specimen is

measured using a ruler. If the height is not within two to four millimeters of the correct height,

the specimen is inclined again, and gyrated until within that height. If the specimen height is

within two to four millimeters of the correct height, it is subjected to static pressure at the rate

of one stroke of the pump handle per second until it reaches the correct height. This static

pressure is a leveling off load, since the specimen will have a domed top and bottom, with the

dome height approximately two to four millimeters, if not subjected to this load. The static

pressure is immediately released and the ram head raised once the correct height is reached.

The specimen is then left to cool.

After extrusion, specimens are cored to a diameter of 4 inches (101.6 mm), and then cut

to a height of 8 inches (203.2 mm). Coring also produces a 6-inch (152.4 mm) diameter hollow

cylinder specimen which is used for other testing.

4.2.3 Kneadiw, Compaction, 2,5 x 4-1nch ¢63.5 x 101.6 mm) Briquets. For low air-void

specimens, the mold is first elevated on small metal plates. For all specimens, approximately

half of the specimen material is placed in the mold and rodded 20 times around the perimeter

of the mold and 20 times on the inside. Then the second half of the specimen is placed in the

mold and rodded the same number of times. Specimens usually contain 1,200 g of material.

For low air-void specimens, a setting compaction of 20 blows at a nominal gauge

pressure is applied. The small plates are then removed from under the mold, allowing it to slip

down the specimen (about four millimeters).

A.30

The specimen is then subjected to the number of blows at some pressure (both pressure

and numberof blows can be varied) necessaryto achieve the requiredair-voidpercentage. After

the blows are complete, one of several methods of static compaction is applied, as follows:

1. No static load.

2. The specimen is immediately subjected to a true load of up to 12,560 pounds

(5,697 kg). The load is applied by compacting the specimen at a rate of 0.25

inches (6.4 mm) per minutewith a hydraulic press until the load is achieved, and

then immediately unloading the specimen.

3. The specimen is heated for 1.5 hours at 140°F (60°C), and then subjected to a

true load of up to 12,560 pounds (5,697 kg). The load is applied as in Method

2, above.

4.2.4 Kneading Comoaetion, 4 x _-lneh (101.6 x 203.2 mm) Cylinders. Cylindrical

kneading compaction specimens generally contain 4,000 to 4,300 g of material. For high

air-void specimens, the specimen is compacted in three lifts, while low air-void specimens are

compacted in five lifts.

For high air-void specimens, half of the material for the first of the three lifts is placed

in the mold and rodded 20 times around the outside of the specimen and 20 times on the inside.

The second half of the lift is then placed and likewise rodded. The lift is then subjected to the

numberof blows and foot pressure required to obtain the desired air-void content. The same

process is then repeatedfor the remainingtwo lifts.

For low air-void specimens, the first lift is placed in the mold and rodded 20 times

around the outsideand 20 times on the inside of the specimen. The lift is then compacted using

the numberof blows and foot pressure necessary to obtain the desired air-void content. The

same procedure is then followed for the remaining four lifts.

A.31

The ratio of blows for the succeeding lifts must be proportioned, since the already

compactedlifts continue to be compacted by the blows on the lifts above. The ratios shown in

Table A.8 are used for the number of blows on the lift to be compacted and on the lift just

compacted.

After the specimen has been compacted, it is sometimes subjected to a load on the

hydraulic press, the same as for 2.5 x 4-inch (63.5 x 101.6 mm) specimens. This load is usually

10,000 pounds (4,536 kg) or less (true load). After cooling and extrusion from the mold, the

ends of the specimens are usually cut so that they are 8 inches (203.2 mm) tall.

4.2.5 Kneading Compaction. 3.5 x 3,5 x 15-1nch (88.9 x 88.9 x 381 mm) Beam_, Beam

specimens, usually containing 7,000 grams of material, are always compacted in three lifts.

The first lift is placed in the mold and rodded about 30 times around the perimeter of the

mold and 20 times on the inside portion. The lift is then compacted in three stages. Each

succeeding stage consists of more blows at higher pressures. Most of the compaction occurs

during the third stage. When the compaction of the first stage is completed, the second and third

stages are compacted in the same manner.

For the same reasons as for the 4 x 8-inch (101.6 x 203.2 ram) kneading cylinders, or

any specimen compacted in lifts, the ratios between the number of blows on each layer must be

adjusted so the void ratio is consistent throughout the specimen.

After the specimen has been compacted it is subjected to one of the following two

methods of static load.

1. The specimen is placed in an oven at 240°F (116°C) for 30 minutes and then

subjected to a true static load of 18,500 to 40,000 pounds (8,392 to 18,144 kg)

for one minute. This method was only used a few times at UCB.

A.32

Table A.8 Ratio of Blows on Lift to be Compacted to Blows on Lift JustCompacted

Lift: 1 2 3 4 5i

Ratio: - 1.25 1.20 1.167 1.143

Example A 40 blows 50 blows 60 blows 70 blows 80 blows

Example B 100 blows 125 blows 150 blows

A.33

2. The specimen is immediately subjected to a true static load of 18,500 to 40,000

pounds (8,392 to 18,144 kg) for one minute.

After the specimen has cooled and the mold has been removed, it is cut to the dimensions

required for testing.

4.2.6 Rollim,-Wheel Compaction of Laree Slabs. The rolling-wheel method of compaction

for large slabs used at UCB follows principles similar to those used for compacting 4 x 8-inch

(101.6 x 203.2 mm) Texas gyratory specimens. For both methods, the amount of material

required to fill the volume of the mold is calculated using the maximum specific gravity found

using the Rice method and the desired air-void content.

The theoretical desired air-void contents for the three succeeding layers placed in UCB

slabs decreased with each lift. In general, material for 0.5 to I percent less air voids was placed

in each succeeding layer. For example, to produce a slab with a uniform (top to bottom) air-

void content of 6 percent, material quantities for the first, second, and third lifts might be

computed on the basis of 6.5 percent voids, 6.0 percent voids, and 5.5 percent voids,

respectively.

The dimensions of the mold used at UCB are shown in Figure A.7. The mold was built

so that the frame for each lift was detachable. This was required since the roller was too large

to fit inside the mold. The required amount of mixture ranged between I00 and 117 kg.

The first level of the mold was attached to the base plate, and filled with the required

material. As the material was placed in the mold, it was rodded using a hoe. The rodding took

place until the material was distributed into the comers of the mold, and a small hump of

material was present in the center. The roller was then rolled back and forth over the material

until it was level with the top of the mold.

A.34

All dimensions in inches F u_._11.50

,4,_",._ 3.00_p" ce_ 3 "_,,_

Sp_ce_- 2 3.00

Sp=cer I 3.00

Figure A.7 Mold Dimensions for Rolling Wheel Compaction (1 in - 25.4 mm)

A.35

The next level of the mold was then bolted to the first layer, and the same compaction

process was carriedout. The third lift was compactedin the same manner. Whencompaction

was complete, a sheet of plywood was placed over the mold and slab, and a two-ton block of

concrete was placed on the plywood untilthe next day.

After the slabhad cooled atleast 24 hours, the mold was removed, and the specimenwas

placed in a constanttemperatureroom at 68°F (20°C) for several moredays. If possible, the

specimen was allowed to cool several days beforeremoving the mold, since some slabs were still

warm even 24 hours after compaction. After three or four days of cooling, the slab was cored

vertically, and after removal of the cores, the cored half was then cut off. The remaininghalf

of the slab was cut for beam specimens.

A.36

5.0 AIR-VOID DETERMINATION

At the beginning of this project, laboratory measurement of air voids was not considered

to be a major difficulty. Experience has since taught that air-void determination is one of the

most important aspects of fabricating good specimens with consistent engineering properties.

This is true for two reasons. First, the calculated percent air voids depends greatly on the

method used and on the type of specimen being tested (as will be seen below). Second, testing

has shown that air-void content is one of the most important parameters affecting the engineering

properties of the specimen.

5.1 Tvves of Svecimens

The preparation of the specimen determines its air-void content. This includes not only

the number of blows or gyrations and compaction pressures, but also the compaction

temperature, the porosity of the surface of the specimen (determined by compaction type and

effort, and cutting and/or coring), and whether it has previously been wetted. For these reasons,

several methods of air-void determination have been simultaneously used on most specimens at

the UCB laboratory. Even still, determining whether two specimens prepared using different

methods have the same air voids is sometimes only approximate. A better understanding of this

problem will be found from the following descriptions of the methods used for each specimen

type.

5.1.1 As-Molded Surface vs. Cut Surface Soecimens. Specimens with asphalt surfaces are

those which have not been cut or cored. Specimens with cut surfaces are those that have been

cored and/or cut to arrive at their final dimensions. This preparation leaves the specimens with

a surface that is generally smooth, and with exposed cut aggregate. Examples of these

A.37

specimens are kneading fatiguebeams, 4 x 8-inch (101.6 x 203.2 ram) gyratory specimens, and

all specimens extracted from the rolling-wheel slabs.

5.1.2 Wet vs. Dry Sm,cimens. Many specimens that arewetted, whetherduringair-void tests,

or duringcutting or coring, retainsome of the water for extended periods of time. The result

is that specimens that have been cored, or cut with a wet saw, or tested for air voids without a

waterproofcoating, can not be compared directly with those that have neverbeen wet. For this

reason a procedurewas developed that allows calculationof several air-void estimators,with the

goal of allowing comparison with at least one estimator, of specimensprepared using different

methods.

5.2 Air-Void Test Metho_

In order to have the most data for determining whether specimens from different

compaction methods have the same air-void contents, the following three air-void estimators

have been used for each specimen whenever possible at UCB:

1. Air voids, dry (never been exposed to water) withoutparafilm,

2. Air voids, wetted (surface air blown dry) with parafilm, and

3. Air voids, wetted (surface air blown dry) without parafilm.

Parafilm, manufactured by the American Can Company, is a stretchablewax paper which

is used in place of paraffin wax. The use of parafilm, described herein, is adapted from

procedures developed at Chevron Research Company (Del Valle, 1985). The dipping of the

specimen in paraffin as per ASTIVlD 1188 does not allow easy testing of the specimen after

air-void determination due to the difficulty of removing the paraffin from the specimen

(American Society for Testing and Materials, 1988).

Data for the three methods is collected in one process as follows:

A.38

I. If the specimen has been cored or cut, or otherwise been wetted at some time,

submerge it in water until no more air is seen to escape and then go to step 3.

Otherwise, weigh the specimen in air (this is recorded as dry weight in air, A).

2. Weigh the specimen suspendedin water, taking the readingas soon as the balance

stabilizes (normally 1-2 seconds). (This is recorded as dry weight in water, B.)

3. The specimen should now be dripping wet and soaked through. It should

immediately be air dried using a small blower valve connected to the house air

supply, about 110 psi (758 kPa) at UCB. Dry the surfaceof the specimen until

no waterescapes from the surfacedirectlybeneath the air stream, with the blower

held about an inch (25 ram) from the surface. When this condition is reached,

immediately weigh the specimen in air (recorded as wet weight in air, C). This

represents the weight of the specimen with some water trapped inside the

permeablepores. ASTM method D 2726-68 (AmericanSociety for Testing and

Materials, 1988) requires heating of the specimen at 230°F (110°C): experience

gained to date shows this often results in its disintegration.

4. Wrap the specimen in parafilm, so that it is completely watertight,but using the

minimumamountof parafilm necessary. Be sure that no air bubbles are trapped

under the surface. Weigh the specimen in air (this is recorded as wet weight in

air with parafilm, D).

5. Weigh the specimen suspended in water with the parafilm, taking the reading as

soon as the balance stabilizesafter the specimenis placed in the water (normally

1 to 2 seconds). (This is recordedas wet weight in water with parafilm, E.)

6. Remove the parafilm, and repeat step 2 (this is recorded as the wet weight in

water, F).

A.39

The three specific gravities are then calculated as follows:

7. Specific gravity, dry, no parafilm (Gdnp)

8. Specific gravity, wet, with parafilm (Gwwp)

C

G... = C-D + D-E0.9

9. Specific gravity, wet, no parafilm (Gwap)

C

G.,,_ = C-F

10. After determination of the maximum specific gravity of the mixture using the

Rice Method (ASTM D 2041), each of the specific gravities can then be used to

calculate the corresponding air-void percentages using the following formula:

GAir Voids (%) = 100 [1 - "-2]

K

where G is the measured specific gravity (dnp, wwp or wnp) and R is the Rice

maximum specific gravity.

A.40

6.0 R__CF__

American Society for Testing and Materials (1988), "1988 Annual Book of ASTM Standards,Section 4, Construction," Vol. 04.03, Philadelphia.

Del Valle, H. (1985), "Procedure - Bulk Specific Gravity of Compacted Bituminous MixturesUsing Pamfilm-Coated Specimens," ChevronResearch Company, Richmond, California.

Hicks, R.G. (1989), "Technical Memo OSU-89-1, Materials Protocol: SHRP A-003A," 2riddraft.

D

A.41

APPENDIX B

DIAMETRAL FATIGUE TEST PROCEDURE

1.0 INTRODUCTION

1.1 Scope

This appendix describes the diametral (indirect-tensile) test for determining the stiffness

and fatigue life of either laboratory-fabricated or field-recovered asphalt-aggregate mixtures.

The described procedure is essentially an extension of the Standard Method of Indirect Tension

Test for Resilient Modulus of Bituminous Mixtures 2 (ASTM D 4123-82). Other applicable

documentsincludethefollowingASTM standards:

D 1559 - TestMethod forResistancetoPlasticFlow of BituminousMixtureUsingMarshallApparatus,

D 1561- Method ofPreparationofBituminousMixtureTestSpecimensby Means ofCaliforniaKneadingCompactor,

D 3387 -TestforCompactionand ShearPropertiesof BituminousMixturesby MeansoftheU.S.CorpsofEngineersGyrator),TestingMachine(GTM),

D 3496 - Method forPreparationof BituminousMixtureSpecimensforDynamicModulusTesting,

D 3515- SpecificationsforHot-Mixed,Hot-LaidBituminousPavingMixtures,and

D 4013 -StandardPracticeforPreparationofTestSpecimensofBituminousMixturesby Means ofGyratoryShearCompactor.

1.2 _;pmmary 9f Pr0cedere

The repeated-load diametral test for determining the resilient modulus and fatigue life of

asphalt-aggregate mixtures is conducted by the application of compressive load pulses having

square waveforms. The load is applied in the vertical diametral plane of a cylindrical specimen

as shown in Figure B. 1. Repetitive applications cause permanent deformation of the specimen

2Annual Book of ASTM Standards, Vol. 04.03

B.2

Pi Loading

Specimen Strip

D

P

Figure B.I Loading of Cylindrical Specimen in the Diametral Fatigue Test

B.3

and also induce tensile stresses (a0y of Figure B.2) thatusually are sufficient to eventually split

the specimen into two pieces.

The fatigue life of the specimen is defined as the number of load applications resulting

in either disintegration or a specified permanent vertical deformation whichever occurs first.

Criteria of 0.20 and 0.50 inch (5.1 and 12.7 mm) for testing at 39.2°F and 68°F (4°C and

20°C), respectively, have been found to be appropriate for recording the drop of the upper

loading platen from the splitting caused by repeated tensile stresses and for stopping the test if

the specimen undergoes significant compressive deformation under the repeated load. Vertical

deformation is usually the combined result of both plastic deformation and vertical splitting.

1.3 Significance and Use

Fatigue life, as measured by diametral loading, has been used to evaluate the relative

performance of asphalt-aggregate mixtures as well as for input to pavement thickness design or

pavement evaluation and analysis. It has also been used to study the effects of temperature,

repeated-load magnitude, loading frequency and duration, etc. However, since the test is

destructive, tests cannot be repeated on the same specimen as can be done for resilient-modulus

testing.

B.4

- 1.0 --

O'oy

6 -

.8 --

I I I I I I I I I I I

1.0 .8 .6 ,4 .2 0 -.2 -.4 -.6 -.8 - 1.0

Tension _ Compression

Sign Convention :

+ Denotes tensile effects

Denotes compression effects

Figure B.2 Stress Distributions Within DiametraUy Loaded Specimens(After Hadley, Hudson, and Kennedy, 1970)

B.5

2.0 TEST PROCEDURE

2.1 Test Apparatus

The apparatus to perform the diametral fatigue test should conform to that specified in

ASTM D 4123 with the added ability to count and record the number of load applications. In

addition, the apparatus should have the ability to automatically discontinue load applications

when the specified amount of permanent deformation has occurred.

2.2 SI_'cimen Preparation

2.2.1 Laboratory-Fabricated S_¢imfn_. Laboratory specimens should be prepared in

accordance with the procedures described in Appendix A.

2.2.1 Core Sl_ecimens. Specimens cored from in-situ pavements or other large masses should

have relatively smooth and parallel surfaces. Height and diameter requirements specified for

laboratory-fabricated specimens are also applicable to core specimens, nominally 2.5 x 4 inches

(63.5 x 101.6 mm).

2.3 Failure Criteria

As previously mentioned, the fatigue life is the number of load applications required to

induce a specified amount of permanent vertical deformation. Experience has shown that, even

before the vertical deformation has reached limits of 0.20 or 0.50 inch (5.1 or 12.7 mm) for test

temperatures of 39.2°F and 68°F (4°C and 20°C), respectively, the test specimen often fails

dramatically ('explodes'), that is, it disintegrates before permanent deformation has reached a

critical level. Fatigue life in this case is simply the number of load applications that had been

sustained prior to specimen disintegration. Experience has also shown that specimens with low

B.6

resistance to permanent deformation at temperatures of 68°F (20°C) and above may experience

significant permanent deformation before substantial tensile cracking is observed.

The vertical deformation is measured during testing with a linear variable displacement

transducer (LVDT) which measures the vertical displacement of the upper loading platen. The

amount of permanent vertical deformation, as measured by the LVDT, that occurs before a

failure condition is reached can be recorded using either a computer data acquisition system or

a strip chart recorder.

2.4

The diametral fatigue test is conducted as follows:

I. Determine the test temperature and loading conditions (that is, load frequency and

duration). For the compaction study, test temperatures were 39.2°F and 68°F (4°C and 20°C).

At both 39.2°F and 68°F (4°C and 20°C), the load frequency was 1 Hz, with a 0.1-second

loading followed by 0.9-second relaxation period.

2. Determine the desired stresses or estimate the tensile strains that will result in failure

between 1,000 and 5,000 load repetitions (high stress level), or 70,000 and 100,000 load

repetitions (low stress level), based on experience and established fatigue equations.

3. Measure the specimen for both height and diameter, and record the measurements for

later calculations of stress, strain, and modulus of elasticity.

4. Center the specimen in the horizontal displacement measurement apparatus (Figure

B.3) and adjust the position of the horizontal LVDTs so that they are in the correct voltage

range. Place the specimen with the horizontal LVDTs into the loading frame and adjust the

position of the vertical LVDT so that it, too, is in the correct voltage range.

5. Determine the load magnitude required to induce the desired tensile strain by loading

the specimen five times and calculating the average tensile strain. The tensile strain is calculated

B.7

LOOdmq Rom

)1 AIIochmenl

,Top Loodm 9 Strip

LVOT

Gouge Heod To Sir;l:) Chor!

RecorderLVOTAdjuStment

Knob A$ohollCo_crele

Oiometrol SpecimenYoke

Bottom LOoding Slf,p Cell

Proven (Rqtsl_ OnLOOO Frome 8o_ePlole)

Figure B.3 Test Specimen with Diametrai Yoke and Loading Ram(After Vinson, 1989)

B.8

from the horizontal displacement measurements, assuming a Poisson's ratio of 0.35 (Kennedy,

1977). For a 4-inch (101.6 ram) diameter specimen, the tensile strain is as follows:

err = 0.520 HRT 03.1)

where eRTis the resilient tensile strain and HRT is the total resilient horizontal deformation in

inches. Repeat this step, adjusting the load until the desired tensile strain is achieved. Normally

no more than two repetitions are necessary.

6. Load the specimen 50 times, collecting data on the last 10 repetitions. The tensile

strain is calculated by Equation B. 1. The modulus of elasticity and tensile stress are also

calculated after Kennedy (1977) using an assumed Poisson's ratio of 0.35. For a 4-inch (101.6

ram) diameter specimen:

ERr : 0.618 (H--_) 03.2)

and

: 0.156 (h) 03.3)IIT

where ERT is the total resilient modulus of elasticity in psi, P is the repeated load in pounds, h

is the height of the specimen in inches, and lIt is the tensile stress in psi. The average values

for the ten repetitions are taken as the test values.

7. Mark the load platen positions on the specimen, and then remove the horizontal

LVDTs. Replace the specimen in the loading frame with the loading platens in the same

positions, and readjust the position of the vertical LVDT so that it is in the correct voltage

range.

8. Load the specimen to failure, recording the number of load repetitions and the vertical

deformation. The test should automatically stop when the vertical LVDT measures a

B.9

displacement greater than the failure criterion, whether it is caused by vertical plastic

deformation, disintegration of the specimen due to tensile stresses, or a combination of the two.

B.IO

3.0 REFERENCES

Hadley, William O., W. Ronald Hudson, and Thomas W. Kennedy (1970), "A Method ofEstimating Tensile Properties of Materials Tested in Indirect Tension," ]_esearch Report98-7, Center for Highway Research, The University of Texas at Austin.

Kennedy, Thomas W. (1977), "Characterization of Asphalt Pavement Materials Using theIndirect Tensile Test," ]_l_,_di13_, Association of Asphalt Paving Technologists,Volume 46, pp. 132-150.

Vinson, T. (1989), "Fundamentals of Resilient Modulus Testing," Presented at the Workshopon ResilienrModulus Testing, Oregon State University, Corvallis, March 1989.

B. 11

APPENDIX C

STATISTICAL ANALYSIS TECHNIQUES FOR MAIN EXPERIMENT

Unequally replicated, fractional factorial experiments--such as those used in the main

experiment-can be analyzed using any good regressionor General Linear Model (GLM) program,

as are available from the SAS Institute. It is important,though, that therebe no totally missing treatment

combination, as the efficiency of the experimentaldesign greatly suffers.

Regardless of the particular algorithm used, the model used to analyze any full or fractional factorial

is the same. With suitable numerical coding of the factor levels, a model of the form

Yi --- ao + ax Xix+ a2 Xi2 + --. + an Xin

+ a12 Xil Xi2 -6 a13 Xit Xi3 + ...

+ a23 X_2X_ + ...

will analyze the data properly. In this expression, Yiis the observed measure of the dependent variable

during the ith trial, Xij is the value for the jth factorduring the ith trial, and the a's are coefficients to be

estimated by the regression.

For each two-level factor, the numerical variableX is defined to have the value -1 for its low

and + 1 for its high level. When the number of levels of a factor is three or more, the same kind of model

can be used, but with different X values. For example, a three-level factor uses two X variables to describe

it, with each X taking on one of three values, as follows:

Level i XL xiQ0 -1 11 0 -22 1 1

The superscripts L and Q denote the linear and quadratic components of a three-level factor. A four-level

factor requires three variables (linear, quadratic, and cubic), thus:

C.2

Leveli XL xiQ X_t

0 -3 1 -11 -1 -1 32 1 -1 -33 3 1 1

The values for each component are chosen so as to be linear, quadratic, cubic, etc. functions of the

level i such thatthe cross product sum of any two components (e.g., E XL * XQ) is zero:

XL*X_, ffi(-3xl)+(-Ix-1)+(1x-1)+(3xl)i

--3+ 1-1+3--0

In addition, the values are scaled so as to be simple, small integers. These sets of values for factorsat 2,

3, 4, etc. levels are called simply "orthogonalpolynomials."

Using this polynomial model for a simple 22 factorialexperimentpermits us to see how the concept

of "hiddenreplication" works. Our model for a 22 factorial is

Yi - °to + ai Xil + °t2 Xi2 + ¢x12 Xil2 + ei

where the a's are unknown true effects to be estimated, ei is the error in the observation Yi,and Xil 2 is the

factor representing interaction effects. The X's would be defined as follows:

i Xil Xi2 Xil 2 - Xil x Xi2

"I -1 -1 I2 1 -1 -13 -1 1 -14 1 1 1

Assuming an experiment without replication, four equations relate the observations, Yi, to the true effects,

a, as follows:

C.3

Yl -- ao " al " ¢x2 + a12 + el

Y2 ----aO + al - a2 - a12 + e2

Y3 -- ao'al + a2-a12 + e3

Y4 -- ao 4- a I + a 2 + a12 + e4

Summing the four observations gives

Yl + Y2 + Y3 + Y4 = 4ao + el + e.2 + e3 + e4,

and dividing by 4 gives the mean value of y

_ Eeiy -a o + ---,

4

which has error variance of _/4.

Best estimates of the effects, _j, and their variances, _(&), are as follows:

_Yi _ O_eao ffi --- -- y with _(_o) ffi "-

4 4

Y2 + Y4 " Yl " Y3 _'e&l = ................. with CXl) - --

4 4

Y3 + Y4-YI'Y2a 2 = ................. with _(_2) = --

4 4

Yl - Y2 - Y3 + Y4 ¢r_e= ................. with (a12) = --

4 4

Each effect estimate, _j, has the same variance as that of the mean of the four observations. For a

23 design, each of the eight effects (%, at, or2, oc3,a12, cq3, a23, a123) would have variance, _/8, the

variance of a mean of eight points. This shows that regardless of whether we simply run repeats or include

other factors in our experiment, the error variance in our estimates always decreases by 1/N.

C.4

APPENDIX D

PROPOSED METHOD FOR MEASUREMENT OF AIR VOIDS INTEST SPECIMENS OF BITUMINOUS MIXTURES

1.1 Definition of Terms

Specimens with asphalt surfaces are those which have not been cut or cored. Specimens

with cut surfaces are those that have been cored and/or cut to arrive at their final dimensions.

Wet specimens are those that have been exposed to water during coring, cutting, or previous air

voids tests.

1.2 Air-Voids Measuremenla

In order to have the most data for determining whether specimens from different

compaction methods have the same air-void contents, the following three estimators are used for

each specimen whenever possible at the University of California - Berkeley COCB):

1. Air voids, dry (never been exposed to water) without parafilm,

2. Air voids, wetted (surface air blown dry) with parafilm, and

3. Air voids, wetted (surface air blown dry) without parafilm.

Parafilm is a stretchable wax paper which is used in place of paraffin wax and is

manufactured by the American Can Company.

Data for the three methods is collected in one process as follows.

1.2.1 Measurement of SDechncn Weights.

1. If the specimen has been cored or cut, or otherwise been wetted, place it in water

until no more air bubbles are seen to rise to the surface, and then go to step 3.

Otherwise, weigh the specimen in air and record the weight as the dry weight in

air, A.

2. Weigh the specimen suspended in water, taking the reading as soon as the balance

stabilizes (normally 1-2 seconds). Record the weight as the dry weight in water,

B.

D.2

3. The specimen should now be dripping wet and soaked through. It should

immediately be air dried using a small blower valve connected to the house air

supply (about 105 psi at UCB). Dry the surface of the specimen until no water

escapes from the specimen surface directly beneath the air stream, with the

blower held about an inch from the surface. When this condition is reached,

immediately weigh the specimen in air, recording the weight as the wet weight

in air, C. This represents the weight of the specimen with some water trapped

inside the permeable pores.

4. Wrap the specimen in parafilm, so that it is completely watertight, but using the

minimum amount of parafilm necessary. Weigh the specimen in air and record

the weight as the wet weight in air with parafilm, D.

5. Weigh the specimen suspended in water with the parafilm, taking the reading as

soon as the balance stabilizes after the specimen is placed in the water (normally

1 to 2 seconds). Record the weight as the wet weight in water with parafilm, E.

6. Remove the parafilm and repeat step 2. Record this weight as the wet weight in

water, F.

1.2.2 Calculation of S_eific Graviti_. The three specific gravities axe then calculated as

follows:

1. Specific gravity, dry, no parafilm (Gdnp)

A

Gd'v' = A-B

D.3

2. Specific gravity, wet, with parafilm (G,_q_)

C

G._ = C-D+ D-E

0.9

3. Specific gravity, wet, no parafilm (Gwap)

C

G._, = C-F

4. After determination of the maximum specific gravity of the mixture using the

Rice Method (ASTM D 2041), each of the specific gravities can then be used to

calculate the corresponding air-void percentages using the following formula:

GAir Voids (%) ffi 100 [1 - -__]

R

where G is the measured specific gravity (dnp, wwp or wnp) and R is the Rice

maximum specific gravity.

D.4

APPENDIX E

PROPOSED METHOD FOR PREPARATION OF TEST SPECIMENSOF BITUMINOUS MIXTURES BY MEANS OF ROLLING-WHEEL COMPACTION

1.1 LIB_lg[gglJ_

This is a summary of the rolling-wheel compaction procedure for preparing asphalt

concrete specimens at the University of California - Berkeley (UCB) using a one-lift mold (3

inches high). Except as noted, the method for the three-lift mold (9 inches high) is the same.

1.2 Mixture Preparatiop

Asphalt concrete mixtures should be prepared following procedures documented in

Protocol Version 2.0 or any subsequent version.

1.3 Calculation of Slab Air-Void Contents

The first step for compaction is to measure, to the highest precision possible, the

dimensions of the compaction mold that will containthe compacted specimen. The dimensions

are used to calculate the volume of the compacted specimen.

Next, the requiredweight of materialto be compacted is calculatedusing the volume of

the compacted specimen, the requiredair-voidcontent,and the maximumspecific gravityof the

mixture(by the Rice Method, ASTM D 2041). An example calculationis shown in Figure E. 1.

For the UCB one-lift mold, approximately65 to 80 kg of material are needed to achieve air

voids between 3 and 10 percent (wet-with-parafilmair-void method).

1.4 Compaction Procedure

1.4.1 Compaction Temperature. The mix is heated to about 240°F (116°C) in a forced air

oven for at least 90 minutes prior to compaction. The mix should be discardedif it remains in

the oven for more than about 3.5 hours, due to excessive aging of the asphalt. Care should be

taken that the mix does not cool significantly during weighing and placement in the mold. If

E.2

Average Height 3 in = 7.62 cm

Specimen Size Average Width 24.75 in = 62.87 cm

Average Length 24.75 in = 62.87 cm

7.62 x 62.87 x 62.87 -Volume Volume 30,119 cm3 - 30.119 dm3

Rice Maximum Specific 2.486 (ratio of density ofGravity mix to density of water)

Rice Maximum Density 2.486 kg/dm 3Specific Gravity

Desired Air Voids 8.0%

Density - Rice Maximum 2.486 x (1 - .08) --Density x (1- Air Voids) 2.287 kg/dm3

30.119 dm3 x 2.287

Mass of Material Volume x Density kg/dm 3 - 68.886 kg

Figure E.1 Example Calculation for Mass of Mixture to be Compacted

E.3

the mix cools significantly, it will be difficultor impossible to compact to the desired air-void

content.

1.4.2 ]_|Rcementand Rolling. Material to be compactedis weighed just prior to placement

in the mold. At UCB the materialis mixedand heatedin 7 kg batches, using roasting or baking

pans (steel turkeyroastingpans are ideal). Each pan of material is weighed to the nearestgram

using a precision balance before placing in the mold. After the material is in the mold it is

rodded using a hoe like instrument. The roddingdistributesthe material into the corners of the

mold. As the final batches are placed in the mold, a small hump is formed in the center of the

mold.

The roller used at UCB is designed for compaction of sidewalks and other small areas.

It is self propelled with a forward/reverse control, weighs approximately400 kg, and has one

rolling wheel, as shown in Figure E.2. A secondroller weighing approximately600 kg, but of

otherwise similar design, will be used in the future. The rolling wheel is heated for at least

three hours prior to compaction using electrical heating tapes. This is to help prevent the mix

from sticking to the wheel during compaction. Oil, or other chemical bond breakers are not

used on the wheel or in the mold, to prevent possible contamination of the mixture. Bond

breakers can be used, in extreme moderation, if the specimens taken from the slab are to have

cut surfaces, thus removing any effected asphaltfrom the specimen.

The roller is then rolled backand forthover the materialuntil it is level with the top of

the mold. Rolling should only takeplace in one direction, to simulate the unidirectionalrolling

of field compaction. The roller can be used on one half of the slab, and then the other, thus

simulating the effects at the edge of the rolling wheel during field compaction.

When compaction is complete (the top of the specimen is level with the top of the mold),

a sheet of plywood is placed over the mold and slab to prevent disturbanceduring cooling.

E.4

Figure E.2 Rolllng-Wheel Compaction Apparatus

E.5

Underno circumstancesshould a weight be placed on the specimen if it has not been completely

compacted, since the weight will cause static compactionof the slab.

1.5 Slab Removal and Preoaration of S_ecimens

After the slab has cooled at least overnight, it can be removed from the mold. The slab

should not be removed from the mold if it is still warm. The slab should be cored and/or cut

as soon as possible, or else stored in a constant temperatureroom at about 68°F (20°C) until

cut, to prevent cracking or plastic deformation.

The dimensions of the mold used at UCB are shown in Figure E.3. Also shown on

Figure E.3 is one of several differentpatternsthat can be used for specimen extraction.

After specimens are extracted from the slab and cut to final dimensions, the air-void

contents are determined following the proceduresdocumented in Protocol Version 2.0. The

calculationsfor the next slab of the same mixturetype can be adjustedusing the measured air

voids.

E.6

r

0 0

0 0 0

° °0 0 04.coo© 6ooooo

0 024.000

0 0 0

0 0

60.000

1 0.75001/4 in thich

I IL4_IO__ PL', -;_-;, L4A_

Figure E.3 University of California One-Lift Rolllng-Wheel Compaction Mold

E.7

evaluation of laboratory procedures for compacting asphalt

Documents