MoDOT

TE
5092
.M8A3
no.90-5
v.1 c.2 (SOUR! COOPERATIVE HIGHWAY RESEARCH PROGRAM
FINAL REPORT
9 0-5

DETERMINATION OF AASHTO LAYER COEFFICIENTS

Volume I
BITUMINOUS MATERIALS

MISSOURI HIGHWAY AND TRANSPORTATION DEPARTMENT

FEDERAL HIGHWAY ADMINISTRATION

Property of

MoDOT TRANSPORTATION LIBRARY

 5)077907
Te 6211 .Dl-f7 IG°'"\
'\J.\ C. - 2.

DETERMINATION OF AASHTO LAYER COEFFICIENTS

Volume I:

BITUMINOUS MATERIALS

Study 90-5

Prepared for

MISSOURI HIGHWAY AND TRANSPORTATION DEPARTMENT

by

DAVID N. RICHARDSON

JEFFREY K. LAMBERT

PAUL A . KREMER

DEPARTMENT OF CIVIL ENGINEERING

UNIVERSITY OF MISSOURI - ROLLA

ROLLA, MISSOURI

in cooperation with

U.S. DEPARTMENT OF TRANSPORTATION

FEDERAL HIGHWAY ADMINISTRATION

December 1994

The opinions, findings and conclusions expressed

in this publication are not necessarily those of
the Federal Highway Administration.
 EXECUTIVE SUMMARY

The purpose of this investigation was to determine layer coefficients for several

MHTD specified pavement materials. The coefficients are necessary as input to the

AASHTO pavement design method . Volume I of this study involves asphaltic

materials, and is reported herein. Volume II deals with unbound aggregate base and

soil-cement base materials, and is reported elsewhere.

Besides determining layer coefficients, the study also entailed the determination

of the effect on layer coefficient by changes in asphalt cement grade, aggregate

gradation, testing temperature, aggregate source, and asphalt content within the limits

of MHTD specifications. This resulted in 48 mix designs.

All materials were sampled and delivered to UMR by MHTD personal. Choice

of material sources was made by MHTD. The types of pavement materials were Type

C, Type 1-C, and bituminous base. The specific materials making up these types were

two grades of asphalt cement, two sources each of surface mix coarse aggregate and

base mix coarse aggregate, and one source each of natural sand, manufactured sand,

mineral filler, and hydrated lime.

Routine index and specification tests were performed. For the asphalt cement,

the tests were: penetration at 38° and 77°F, kinematic viscosity, absolute viscosity,

specific gravity, and softening point. The aggregates were tested for gradation,

specific gravity, and particle shape/texture. Equipment was fabricated for the particle

shape/texture tests.

The optimum asphalt content of each of the 12 gradation/aggregate source

 ii

combinations was determined by use of the Marshall mix design method (75 blow,

manual flat-faced hammer). Use of AC10 and AC20 grades resulted in 24 mixes.

And, 24 additional mixes were made which had 0.5% asphalt added above optimum,

for a total of 48 mixtures .

Maximum theoretical specific gravities were determined in two ways : 1) Rice

method, and 2) calculation from material proportions and specific gravities. Ninety-six

specimens were tested . A voids analysis was conducted to determine the effect of

estimation of ...,aximum theoretical specific gravity . The estimation method involves

the assumption that the effective specific gravities of low absorption aggregates is

midway between the bulk and the apparent specific gravities. The voids analysis

indicated that the estimation method correlated very well with results from Rice

method testing. However, for absorptive aggregates (.fill., the bituminous base

materials in this study), the estimation method underpredicted air voids by about 1 %.

Ten methods of characterizing gradation curve shape and position were used .

Two of these were original to this study. The first involved the area between the

gradation curve and the maximum density line as plotted on FHWA 0 .45 power paper.

The second method involved determination of the slopes of three portions of each

gradation curve . The method of determining the area between the 0.45 power

maximum density line (MDL) and the gradation line had only a fair (R 2 = 0. 79)

correlation with resilient modulus (MR). This was because the magnitude of the area

was not sensitive to relatively small differences in position of the gradation curve

relative to the MDL. The second unique method involved calculation of the slope of
 111

three different parts of the gradation curve. This method was shown to be of

assistance in creating a more accurate MR regression model. However, it was not

quite as helpful as Hudson's A, which is much easier to calculate. But , Hudson'sA

was not quite as helpful as merely including certain critical sieve sizes directly into the

regression equation .

Each mix was tested for indirect tensile strength. A regression model was fit

to the data, which included 96 test results. The regression model was relatively

strong (adj-R 2 = 0.840) and was a function of asphalt viscosity, effective asphalt

content, percent accumulative retained on the #4 sieve, and coarse aggregate particle

shape.

Each mix was tested for total resilient modulus (indirect tension) . Necessary

software and equipment were developed to perform the tests and to acquire, store,

and analyze the data. A total of 192 specimens were tested at three temperatures

for a total of 576 tests. The resilient, modulus test is sensitive to testing conditions

of temperature, specimen rocking, specimen surface irregularities, choice of point of

L VDT fixation, LVDT tip design, and resolution of both vertical and horizontal LVDT's.

Constant diligence is required by the operator to assure that the very small

deformations being measured are representative of actual deformations. A relatively

strong (adjusted R2 = 0.946) regression model was fit to the UMR MR data.

The results of the MR testing were analyzed statistically to determine the

variables that were significant to changes in MR. The analysis of the data indicated

that temperature was by far the most important variable that affects MR, followed by
 IV

asphalt viscosity and whether the gradation was very fine or very coarse .

Overasphalting by 0. 5 % tended to lower MA, but was not statistically significant.

Increases in (-) #200 material and decreases in ( +) #4 material tended to increase MA.

Particle shape of coarse or fine aggregate did not seem to affect MA in a consistent

manner . It should be noted that both coarse aggregates were crushed limestones, and

that all mixes contained varying amounts of manufactured sand, so large ranges in

particle shape were not present. And, all other things held constant, decreasing air

voids tended to increase MA.

Resilient modulus data from other studies found in the literature were merged

with the UMR data. A general regression model was fit to the overall data base. The

model was not as strong as the UMR model, but was deemed superior because it

represented a much wider range in magnitudes of variables. The equation is a

function of pavement temperature, ai~ voids content, asphalt viscosity, percent

passing the #200 sieve, percent accumulative retained on the 3/4 in sieve, and

effective asphalt content.

log MR= 6.871-0.017T-0.024Pair+0.043~7o+O.Ol8P200-0.004AR3;4-0.0llPeffv

In order to compare pavement temperatures in Missouri and at the Road Test,

air temperature data from 104 weather stations in Missouri were analyzed to produce

an air temperature contour map of Missouri. Pavement thickness data for MHTD and

Road Test flexible pavements were analyzed for mean pavement thickness. This

information was necessary to calculate pavement temperatures. Mean vehicle speed

data was supplied by MHTD . This was converted to load dwell times and loading
 V

frequency for MHTD pavements . The same was done for AASHO Road Test

pavements .

UMR, AASHO Road Test, and MHTD 1990 mix des ign data were used to

estimate resilient modulus , mixture stiffness (Shell method) , and dynamic modulus at

both laboratory conditions and field conditions of pavement temperature and loading

rate . This was done in order to see which type of modulus would be most useful for

layer coefficient determination . The Odemark equation was used to rate the three

methods of obtaining mixture modulus or stiffness . The ranking , in descending order

of ability to predict resilient modulus, was: MR estimated from the above general

regression equation, Shell mixture stiffness, and dynamic modulus.

Five different methods of calculat ing mixture stiffness (Sm) were compared;

each varied in the manner of handling asphalt aging or source . Of the f ive d ifferent

methods , the method of Bonnaure , wh ich uses the Ullidtz asphalt aging

approximations, was found to be the most accurate for the purposes of this study .

Two options to obtain layer coefficients were presented for possible use . The

first involves the determination of MR by test (or by estimation of resilient modulus by

the general regression equation) for a pavement temperature of 68 °F , then entering

the proper AASHTO nomograph to obta in the corresponding layer coefficient . The

second option is to again determine the MR by test or to estimate the resilient

modulus, but the moduli must be converted to the pavement temperature conditions

in the locale of interest. Then, the layer coefficient is computed via the Odemark

equation which relates the MHTD MR to the AASHO Road Test MR .

Option One resulted in a fixed layer coefficient per material. For 1990 mixes ,
 vi

Type Ca, = 0.42, Type IC a, = 0 .42, and bituminous base a 2 = 0 .34. Option Two

is a method for pavement designers with which to calculate layer coefficients for a

specific mix and location in the state.

A sensitivity analysis was performed which examined the effect of specific

important mix characteristics (viscosity and gradation), pavement temperature, and

overall mixture quality. Only asphalt grade and extremes in gradation were significant

in their effect on MR. For asphalt layer thicknesses less than 4 in, the effect of

variations in the above two variables was small. However, as layer thickness

increased, the required thickness changed significantly . Additionally , the magnitude

of layer coefficient was shown to be a function of mix variables and location of the

site within the state . Examination of the effect of choice of layer coefficient revealed

large differences in required bituminous base thickness.

Because layer coefficients are directly affected by resilient modulus, the

practical impact of the trends is that higher layer coefficients can be obtained by :

1. us ing a harder grade of asphalt (higher r17 0 )

-
2. using a more well-graded gradation (higher A, adjusting P200 and AR 4 )

3. avoiding overasphalting (optimum Pe 11 vl

4. increasing density (lower air voids) .

Of course, caution should be exercised when considering these options.

Excessively hard asphalts, highly angular aggregates, and low air void values can lead

to such problems as thermal cracking, lack of workability, poor durability, and rutting.

Thus, caution should be used when interpreting the results of the above analysis .
 vii

Also, it must be kept in mind that the MR predictive equations are based on data that

represents well-graded gradations. They should not be applied to mixes with

significantly different characteristics or materials, such as stone mastic mixtures.

It is highly recommended that the MHTD pursue MR testing of various mixtures

in present use in order to update or replace the above equation by use of a more

representative data set of the materials. A greater degree of accuracy will also

probably be achieved. Then, both Options One and Two will render more

representative layer coefficient values for MHTD designers.

It should be remembered that this study is in the mold of the traditional method

of determination of layer coefficients, that is, by a comparison of some sort of

strength or stiffness of MHTD materials to Road Test materials. Tendencies for

asphaltic material problems with thermal cracking and rutting, for instance, are not

directly addressed. To address a wider range of material issues, creep testing and

gyratory shear testing may be in order, however, these kinds of tests were beyond the

scope of this project . Also, this project was conceived in 1989 and the bulk of the

testing was performed in 1991, before the SHRP project results became generally

known. In the future, it may be that some of the recommendations coming out of the

SHRP program can be used to update the quest for layer coefficient determination.
 viii

EXECUTIVE SUMMARY
TABLE OF CONTENTS ii
LIST OF FIGURES V
LIST OF TABLES viii

INTRODUCTION 1
General 1
Objectives and Scope 4

DETERMINATION OF LAYER COEFFICIENTS: METHODOLOGY 5

AASHTO Nomographs 5
Equivalent Stiffness 6

MATERIAL TYPES AND SOURCES 6

Asphaltic Cement Concrete 7
Plant Mix Bituminous Base 7

ESTIMATION OF ASPHALT MIXTURE STIFFNESS 8

General 8
Asphalt Cement Stiffness (Sb) 8
Bonnaure Mixture Stiffness (Sm) 14
McLeod Mixture Stiffness (Smpvnl 18
Calculation of Mixture Stiffness 21
Summary 23

ESTIMATION OF DYNAMIC MODULUS 24

General 24
Summary 26

RESILIENT MODULUS 27

LABORATORY INVESTIGATION 28
Mix Designs 28
Asphalt Cement 32
Penetration 32
Kinematic Viscosity 32
Absolute Viscosity 32
Specific Gravity 33
Ring and Ball Softening Point 33
Aggregate 33
Initial Gradation 33
Final Gradation 33
Particle Shape and Surface Texture 35
 ix

Specific Gravity 36
Screening 38
Fabrication of Specimens 38
Marshall Mix Design 40
Maximum Theoretical Specific Gravity 40
Specimen Bulk Specific Gravity 41
Voids Analysis 41
Indirect Tensile Strength 41
Resilient Modulus 44

RESULTS OF LABORATORY INVESTIGATION 46

Asphalt Cement 46
Aggregate 52
Gradations 52
General 52
Gradation Curve Shape 59
Particle Shape and Surface Texture 67
Specific Gravity 69
Maximum Theoret ical Specific Gravity 72
Voids Analysis 73
Marshall Mix Design 80
Indirect Tensile Strength 83
Results 83
Multiple Regression 87
Resilient Modulus 90
Test Results 90
Acceptable Range of MR 92
Effect of Variables 96
Statistical Analysis 104
Tests of Significance 104
Multiple Regression 108
Applications 111
General Resilient Modulus Regression Equation 113
Pavement Temperature 118
Load Duration and Frequency 121

ESTIMATION OF MIXTURE STIFFNESS 124

UMR Mixture Stiffness (Sm) 124
MHTD M ixture Mean Stiffness (Sm) 127

ESTIMATION OF DYNAMIC MODULUS 127

UMR Mixture Dynamic Modulus l E* l 127
MHTD Mixture Dynamic Modulus l E * l 131
 X

DETERMINATION OF LA YER COEFFICIENTS 131

AASHTO Nomographs 132
Equivalent Stiffness 136
Mixture Stiffness 136
Dynamic Modulus 136
Resilient Modulus 137
Method of Choice 137
Option One 138
Option Two 139
C and IC vs AASHO 144
Bituminous Base vs AASHO 144
SENSITIVITY ANALYSIS 145
Mixture Variable Effect on Thickness 145
Layer Coefficient Effect on Thickness 146

SUMMARY 150
CONCLUSIONS 154
RECOMMENDATIONS 158
ACKNOWLEDGEMENT 159
REFERENCES 160
APPENDICES 171
Appendix A: Fine Aggregate Particle Shape Determination 172
Appendix B: Coarse Aggregate Particle Shape Determination1 81
Appendix C: Determination of Indirect Tensile Strength and
Resilient Modulus for Bituminous Mixtures 189
Appendix D: Determination of Asphalt Type 206
Appendix E: Specific Gravity and Voids Data 209
Appendix F: Results of Marshall Mix Design 218
Appendix G: Industry-wide Database for Resilient Modulus
Regression Analysis 231
 xi

LIST OF FIGURES

Fig. No. Title Page No.

1 Van der Poel Nomograph 9

2 Determination of PIR&B 12
3 Determination T pen 800 13
4 Determination of Mixture Stiffness
(Sm) 15
5 McLeod's Temperature Difference Nomograph 19
6 McLeod's Modified Vander Poel Nomograph 20
7 Ring and Ball Softening Point Device 34
8 Asphalt Mixture Aggregate Fraction Storage
Bins 34
9 NAA Method Particle Shape Device 37
10 D3398 Particle Shape Equipment 37
11 Asphalt Cement Temperature - Viscosity
Relationships 39
12 Marshall Stability/Flow Test Apparatus 42
13 Rice Maximum Theoretical Specific Gravity
Test Station 42
14 Indirect Tension Test Equipment 45
15 Asphalt Mixture Resilient Modulus Test
System 45
16 Binder Stiffness from Pl, and PVN for 0.04 sec Loading Time 50
17 Binder Stiffness from Pl, and PVN for AC-20 50
18 Final Type C Blend Gradations 53
19 Final Type 1-C Blend Gradations 54
20 Final Bituminous Base Blend Gradations 55
21 Areas Under Maximum Density Lines 64

22 Hudson's A as a Function of Gradation 65

23 (G.e - D,) vs. Gradation 75

 xii

24 (Gmm - D) vs . Absorption 75
25 Comparison of Air Voids Methods of Determination for Types C
and IC Mixtures 77

26 Comparison of Air Voids Methods of Determination fo r

Bituminous Base Mixtures 78

27 Predicted vs. Observed Indirect Tensile Strength 89

28 Resilient Modulus at 77°F vs. Indirect Tensile Strength 93

29 UMR Mixtures on AAMAS Acceptance Chart 94
30 Observed Resilient Modulus vs . Estimated Dynamic Modulus for
UMR Data 97
31 Effect of Asphalt Cement Viscosity on Resilient Modulus 99
32 Effect of Gradation on Resilient Modulus 99
33 Effect of Asphalt Content on Resilient Modulus 100
34 Effect of Percent Passing #200 Sieve on Resilient Modulus 100
35 Effect of Percent Accumulative Retained on the #4 sieve on
Resilient Modulus 101
36 Effect of Index of Particle Shape of Coarse Aggregate Fraction
on Resilient Modulus 101
37 Effect of Particle Shape of Fine Aggregate Fraction on Resilient
Modulus 102
38 Estimated vs. Experimental Resilient Modulus for UMR Study
Data 112
39 Relationship of Estimated vs. Experimentally-Derived Resilient
Modulus Data From General Data Base 117
40 Barksdale's Load Duration Chart 122
41 Observed Resilient Modulus vs . Shell Mixture Stiffness for UMR
Data 125
42 AASHTO a, Layer Coefficient Nomograph 133
43 AASHTO a 2 Layer Coefficient Nomograph 134
44 Typical Temperature - Resilient Modulus Relationship 135

45 Temperature Contour Map of Missouri 141

46 MHTD 1990 Mixtures on AAMAS Acceptance Chart 142
D1 Final Resilient Modulus Device 196
 x iii
02 Resilient Modulus Yoke Mounting Template 196
03 Resilient Modulus Environmental Chamber 198
 xiv

LIST OF TABLES

Table No. Title Page No .

1 Reported Layer Coefficients 2

2 Range of Layer Coefficients at the AASHO
Road Test 3
3 Material Types and Sources 7
4 Alternate Methods of Mixture Stiffness Determination 23
5 Asphalt Mixture Mix Design Parameters 30
6 Asphalt Cement Properties 47
7 Estimated Aged Penetration and T R&e 48
8 PIR&Bt Plpen/pen t Pl,, and PVN 49
9 Binder Stiffness (Sb) 49
10 Results of Resilient-Modulus-Mixture Stiffness Comparison 51
11 As-Rece ived Gradations 52
12 Gradations of Six Final Blends 56
13 Contribution of Raw Materials to Each Blend 57
14 Amount of Each Material in Final Blends 59
15 Results of Gradation Curve Characterization 60
16 Six Test Gradation Slopes 67
17 Particle Shape/Surface Texture Results 68
18 Aggregate Blend Specific Gravities 70
19 Maximum Theoretical Specific Gravity and Effective
Specific Gravity by Two Methods 74
20 Mixture Design Parameters and MHTD Criteria 81
21 Indirect Tensile Strength Data 84
22 Results of Resilient Modulus Testing 91

23 Significance of Variables to Resilient Modulus 106

24 Example of Paired t-Tests: C-mixes at 41 °F 107
25 Ranges of Variables in Industry-Wide Data Base 115
 xv

26 Resilient Modulus of MHTD Approved 1990 Mixes 118

27 Pavement Temperatures, Load Durations, and Frequencies 123
28 Estimation of AASHO Road Test Mixture Stiffness Under
Road Test Conditions 124
29 Comparison of Shell Mixture Stiffness for AASHO , UMR ,
and MHTD Mixes 127
30 Estimation of AASHO Road Test Dynamic Modulus Under
Road Test Conditions 128
31 Comparison of Dynamic Modulus for AASHO, UMR and
MHTD Mixtures 130
32 Layer Coefficients for Bituminous Mixtures 137
33 Comparison of Layer Coefficient Determination Methods 138
34 Layer Coefficients for 1990 MHTD Mixes 139
35 Layer Coefficient Determination 140
36 1990 MHTD Mix Designs Resilient Moduli in Three Parts
of Missouri 143
37 Comparison of AASHO and MHTD Mix Designs 144
38 Thickness Sensitivity to Changes in Gradation and Viscosity 147
39 Thickness Sensitivity to Ranges in Layer Coeffieients 149
B- 1 Specific Volume of Water at Different Temperatures 185
D-1 Bitumen Type Classification 208
 1

INTRODUCTION

GENERAL

This report involves the determination of the American Association of State

Highway and Transportation Officials (AASHTO) pavement design method layer

coefficients for several highway materials commonly specified by the Missouri

Highway and Transportation Department (MHTD) . The study is being made at the

request of the MHTD Research Advisory Committee. The project was executed by

personnel from the University of Missouri-Rolla (UMR) Department of Civil

Engineering.

Based on the results of the AASHO Road Test, a pavement design method

has been developed. This is commonly known as the AASHTO method(1 ). In

applying the method, the highway designer determines a "structural number" (SN)

by knowledge of such factors as design traffic level, subgrade support, desired

reliability, and desired terminal serviceability. The magnitude of the SN reflects the

degree to which the subgrade must be protected from the effects of traffic . For

example, a relatively high SN would indicate that a thick or stiff pavement

structure would be necessary to protect the subgrade from failing or causing

pavement structure failure. Once the SN is calculated, it becomes necessary to

determine the manner in which the SN will be achieved, i.e ., what the required

thicknesses and quality of each pavement layer should be. This is done by solving

the following equation:

 2
where:

SN = structural number

a, ,a 2 ,a 3 = layer coefficients for the surface, base, and sub base

layers, respectively

m 2,m 3 = drainage coefficients of the base and subbase, respectively

D,,D 2 ,D 3 = thickness of surface, base, and subbase layers, respectively.

Drainage coefficients are essentially modifiers of the layer coefficients, and take

into account the relative effects of the internal drainage of the pavement structure

on performance of the pavement. Determination of drainage coefficients is

addressed in a second report submitted by UMR to the MHTD concurrent with this

study (2).

A preliminary review of the literature indicates that reported values for layer

coefficients vary widely, as reported in Table 1 .

Table 1. Reported Layer Coefficients .

Layer Coefficient Material/Layer Value Ref

a, asphalt surface 0.30 - 0.57 3 -8

a2 asphalt treated base 0.10-0.62 4,5,7,8
cement-treated base 0.12-0.50 4-7
lime-treated base 0.12 - 0.26 3,6
unbound granular base 0.03 - 0.23 4 - 10

a3 unbound granular subbase 0.02 - 0.15 4,5,9

The range of layer coefficients determined at the AASHO Road Test are shown in
Table 2 (5, 11 ).
 3

Table 2 . Range of Layer Coeffic ients at the AASHO Road Test .

Coefficient Minimum Maximum Reported

a1 0.33 1 0.78 1 0 .44 1

a2 0.12 2 0 .23 2 0 . 14 2, 0.34 3
a3 0.07 4 0 . 12 4 0.11 4
Note : 1asphaltic concrete surface layer
2unbound crushed stone base
3asphalt-bound base
4
unbound sandy gravel subbase

Examination of Eq. 1 indicates that the thickness of any particular layer is ,

to a significant extent, dependent upon the layer coefficients. Hence, an accurate

determination of layer coefficients can have a significant economic impact in

regard to the design of the pavement structure.

It has been postulated that the magn itude of any layer coefficient is a

function of several factors. For example, the asphalt surface layer coefficient a 1 is

dependent upon mix characteristics, pavement temperature, vehicle speed , layer

thickness, and compacted mix stiffness. For an unbound granular base, the layer

coefficients a 2 and a3 have been shown to be dependent on the state of stress in

the layer, degree of saturation, compactive effort, aggregate properties, and base

layer thickness .

As originally used in the AASHO Road Test results, layer coefficients were

actually regression coefficients which were the result of relating layer thicknesses

to road performance under the conditions of the Road Test. The problem is to
 4

translate the Road Test findings to other geographic areas where the construction

materials and climate are different. Layer coefficients must be determined in order

to use Eq. 1 for design purposes. In a pure sense, layer coefficients are abstract

mathematical entities but in a practical sense they must be related to something

tangible. Most commonly, layer coefficients are determined on the basis of relative

layer material strength or stiffness considerations. Over the years since the

AASHO Road Test, many methods have been used to determine values for layer

coefficients.

OBJECTIVES AND SCOPE

This report is based on methods which optimize the combination of

economy, accuracy, and length of study. In brief, the study entails determination

of stiffness values for several commonly used MHTD types of pavement materials .

The stiffness values were determined by both direct laboratory modulus testing

and by approximation techniques. These stiffness values were related to layer

coeffic ients and then verified for reasonableness by comparing the resulting

coefficients for MHTD materials to AASHTO materials. The report includes a

method suitable for use in routine design which will enable the pavement designer

to solve Eq. 1 and hence obtain the desired layer thicknesses .

The approach taken for determination of layer coefficients was the

traditional one (5, 12), which is to take some measure of strength, stability, or

stiffness of a particular mix and compare it to the same parameter (such as

resilient modulus) for the counterpart AASHO Road Test material. The
 5

comparisons are usually done by use of the AASHTO Design Guide chart or some

ratio of the two parameters. Thus, the influence of rutting is not directly

addressed.

The materials for which layer coeffi9ients were determined were limited to

two types asphalt surface mixes (Types C and 1-C), one type of bituminous base

mix, two types of cement treated base mixes, and one type of unbound granular

base/subbase. The report is separated into two volumes: Volume I covers

bituminous materials; Volume II deals with unbound granular and cement-treated

materials ( 13).

DETERMINATION OF LAYER COEFFICIENTS: METHODOLOGY

Layer coefficients were determined by use of two methods and the results

were compared: 1) AASHTO nomographs, and 2) Equivalent stiffness.

AASHTO NOMOGRAPHS

The moduli determined in the laboratory phase were used directly with the

layer coefficient nomographs in the 1986 AASHTO Guide, which reflect generic

moduli-layer coefficient relationships. There are charts for dense graded asphalt

surface course (a,), unbound granular base (a 2 ), unbound granular subbase (a 3 ),

cement treated base (ai), and bituminous treated base (a 2 ) . The relationships

between layer coefficients and moduli were developed by Van Til et al. (5). Thus,

by determining resilient modulus of asphalt mixtures and unbound base materials,

and static modulus of cement-treated bases, the corresponding layer coefficients

can be determined.
 6

The nomographs were used in a second manner. Data from all approved

MHTD mix designs for 1990 were used in a regression equation developed in this

study which estimates resilient modulus. The resulting estimated moduli were

applied to the nomographs to determine layer coefficients.

EQUIVALENT STIFFNESS

A second method of layer coefficient determination involved the solution of

the following equation which relates MHTD material properties to AASHO Road

Test properties as reported in the literature:

a _ a J modulus, MHTD ] 1 / 3 • • • • • • ( 2)
n ,MHTD - n, AASH't_ modulus, AASHO

where: a" = a, for Type C or 1-C surface course mixtures

= a 2 for bituminous stabilized base course mixtures

The moduli may be one of several types such as resilient modulus (MR),

dynamic modulus (IE* I), or mixture stiffness (Sm). The above equation, originated

by Odemark ( 14), was discussed by Carree and White ( 12) and is based on

structural engineering concepts of equivalent stiffness for a composite layered

material. Hereafter, Eq . 2 will be referred to as the Odemark equation.

MATERIAL TYPES AND SOURCES

All materials in the study were approved MHTD materials and were used in

the specific mixtures as normally intended by MHTD. The material sources were

selected and sampled by MHTD personnel.

 7

ASPHAL TIC CEMENT CONCRETE

Two types of asphaltic cement concrete were studied. These were MHTD

wearing course mixtures: Type C and Type 1-C. Included in the study were two

grades of asphalt cement, two sources of coarse aggregate; one source of natural

sand, two sources of manufactured sand, one so_urce of mineral filler, and one

source of hydrated lime. The materials, sources, and identification codes are

shown in Table 3. All sources are located in Missouri except as noted . The two

coarse aggregates were chosen by MHTD personnel to give a range of particle

shape and texture.

PLANT MIX BITUMINOUS BASE

The plant mix bituminous base mixtures contained the same asphalt cements

and natural sand as did the Types C and 1-C mixtures. Two sources of coarse

aggregate were used, and are shown in Table 3. The coarse aggregates were

chosen by MHTD personnel to give a range of particle shape and surface texture .

Table 3. Material Types and Sources.

Nomenclature Material Sources Location

DR-1 AC-20 grade asphalt cement Sinclair Oil Tulsa, OK

DR-2 AC-10 grade asphalt cement Sinclair Oil Tulsa, OK

DR-4 crushed St. Louis limestone Weber Quarry North Vigus

DR-5 crushed Burlington limestone Conca Quarry Willard

DR-6 crushed Gasconade dolomite Lake Quarry #14 Osage Beach

DR-7 crushed Burlington limestone Conco Quarry Willard

DR-8 natural Missouri River sand St. Charles Sand Co. Bridgeton
#1

DR-9A manufactured Burlington Conco Quarry Willard

limestone sand
 8

DR-98 manufactured St . Louis Weber Quarry Vigus

limestone sand

DR - 10 mineral filler Columbia Quarry Valmeyer ,IL

DR-11 hydrated lime Ash Grove Cement Springfield

ESTIMATION OF ASPHALT MIXTURE STIFFNESS

GENERAL

Pavement engineers are interested in longevity of the pavement. Longevity

is a function of material durability and structural response to load. Usually the

pavement is designed structurally to give a favorable response to load, and

durability is taken care of through specification of good materials . The elastic

response to load of any material is a function of its stiffness , as defined by some

sort of modulus . Thus, direct measurement or estimation of modulus becomes

desirable. This reasoning has given impetus to the effort by the pavement industry

to produce a type of modulus test that is suitable for practical use.

In order to determine layer coefficients via the Odemark equat ion, the

mixture stiffness, dynamic modulus, or resilent modulus must be determined for

each mix. This section deals with determination of mixture stiffness. Subsequent

sections will deal with dynamic modulus and resilient modulus .

ASPHALT CEMENT STIFFNESS (Sb 1

The stiffness of an asphalt mixture (Sm) has been defined as Sm =

stress/strain ( 15). Sm has been shown by Van der Poe I to be a function of asphalt

binder stiffness (Sb) , binder volume , and aggregate volume . While working with

creep and dynamic testing of dense-graded asphalt mixtures, he developed a

method with which to estimate binder stiffness (Sb). The use of his method

involves a nomograph (Fig . 1 ). The required input data include penetration index
 10 7 108 109 ;_,10 3 2.s,103
•1 $1,llnrn 111odulu1 !N/rn~) 106
•6
.. •5 •6
i
-
:;
.
,,i
]
•2
•S

:,
~ •I
~ 0 •I
cE -i 0
-1
-2
-)

' , . .
V11co111v. p

''\, T,mnor11urt dillerrncr,

0
c
Ab OVf T n&B
10050 010rnJ040506o 10 ao go 100110 120 130140 1501so1101eo 190200
.
-
Brlow I R&a

"- , , ~ E x a m p l r lor I bilumrn with Pl • •2 .0 and Tn&a • 75 °C.

'\ f
To obllin 1hr 11i!!nm modulus 11 T • -I 1°C and I frtqurncy of 10 lh:
, f connrct TO Hz on timr 1c1l1 with 75-(-11) • 86° on lrmporary 1calr.
Thr :tiifnm modulus, drfinrd II thr ratio o le= 1trm/1tr1in, i1 1 lunction nf i
Rrad s = 5 x 1o8 N/m2 on nrtwork 11 Pl • •2 .0
timr of lo,ding (frtqurncyl, 1rmprnturt difhrrntt wi1h fl&a point, ind ?1. At, ·
low 1,mperi turf! ind/or high f1t11urncir1 thr 11,flnru modulu1 of •II bitumin1 /
11ymplolet to I limit of ,ppr. J x 10 9 N/m2 . ,, i
\.. i ~ Eomplr for I birumrn wi1h Pl • -1.S ,nd Tr1&a • ~1 ·c .
. '-. ,· To ob11 in thr trmprrllurr for I vi1co1ily ,f 5 pc i1t1
Un111 : '-.
1 )1/n,! • 10 dyn/cm2 • 1.0, x 10-S ,gl/cm2 • US x 10-~ lb i sq .in . ·,Z connrtl 5? II Pl." -1.5 in 1hr nrtwork wilh vi1co1iry point.
1 N s/m2. 10 p ,i ·\. Rud TDif • 70°; T • 70 • 47 • 117°C.
I .

KSL..\ , Auq111t 1953, znd edil ion i969

Frrqurnq , llt ·,. r Vi1co1iry point

DWG . 69 .12 . 1164 1 I oS 1(r 10 3 102 10 T 1" 3" 10··30·1· 2' 5' 10' 30 "1h2h Sh10h1d 2d 7d 30d lyur 10 y 100 y

10 fi
r·
IO S
'

I o-l
1
I
I
I

1a J 10 2
I
I
r----1
10 ·' 10
, I I,,
I
10 2
t · .II I
I
10 3
1
1 It I ',,'•I
I
10
4
10 5
I
, l,1
I
10 6
,, ,

10 7 10 8 10 9 10 I 0

Ti mr of load ing, 1

<D
Fig. 1. Van der Poel Nomogra.ph.
 10

(Pl), ring and ball softening temperature, binder temperature of interest, and

duration of loading (or frequency of loading, if sinusoidal loading is employed).

Sb is very much dependent upon temperature: higher temperatures lead to

lower asphalt stiffness. The amount of change in st iffness brought about by a

change in temperature is termed the "temperature sensitivity" or "temperature

susceptibility" . Various researchers have developed indices to define temperature

sensitivity. Probably the most common is the penetration index (Pl).

Pfeiffer and Doormaal ( 16) developed the Pl to characterize the temperature

sensitivity of an asphalt cement . The penetration index of an asphalt cement is

calculated from the results of penetration tests performed at two or more

temperatures (T, and T 2 ), typically 77°F(25°C) or 39.4°F(4°C) . The following

formulae are used for calculating Pl :

20 - 500A
Pipen/pen = (3)
1 + 50A

A = log pen at T1 - log pen at T 2

where: ( 4)
T1 - Tz

The parameter "A" is the slope of the penetration - temperature curve, a measure

of temperature susceptibility. The units of temperature should be in °C. In the

original method, in lieu of performing the penetration test at the higher

temperature, a penetration of 800 was substituted. This value corresponds to

approximately the softening point for most asphalts . Thus, if T, is 25 °C, then T 2

will be the ring and ball softening temperature. The calculation of A becomes:
 11

log 800 - l o g pen @ 25 ° C

A = ( 5)
RB so f tening poin t - 2 5 ° C

Use of the Pfeiffer and Doormaal method is shown in Fig . 2.

The calculation of Pl from the results of two or more actual penetration tests

has been shown by Heukelom (17) to give more accurate results (Eq. 4).

Heukelom recommended use of a "Bitumen Test Data Chart" (BTDC) which allows

plotting of both penetration and viscosity versus temperature on the same graph ,

thus extending the range of temperature . This is shown in Fig. 3.

Some asphalts exhibit different slopes for different parts of the

penetration/viscosity temperature curve. Thus, Heukelom recommended that t he

portion of the curve representing the temperature range of interest should be used,

if possible, in computing the Pl. Heukelom recommended that in the low

temperature (penetration) range, the curve should be established with several

penetration tests, and that the Pl should be calculated from this, rather than the

curve derived from Pfeiffer and Doormaal's original method. Secondly, Heukelom

recommended that the line should be extrapolated downward to intersect the

penetration = 800 line, and the temperature at this point (T pen 800 ) should be used

in the Van der Poel nomograph instead of the T R&B· This is shown in Fig. 3. In

practice, the ring and ball temperature specificat ion is no longer commonly

specified. And , performing penetration at a second temperature is easier than

running the ring and ball test. Thus, in this study, Pl was calculated based on

performing the penetration test at two temperatures.

 12

1000 .--.--.--.-----,-.--.--.--.---,-.--.--.------,,--.--.--.---
800

-aa
-
0
-100
AC-20

s:I
-~
0
+J

"....
+J
Cl)
s:I
cu
Pot
10

AC-10 : : AC-20
R&B : R&B
44 .5 °C: : 48.5 °C
1
0 25 50 75 100
Temperature (°C)

Fig. 2: Determination of PIR&B"

 . 0
M
0 0

~')
~ r "'- "'-"'-·-" --,-,.---,---"'- ·-~---· --,.-.:.c
"'.....;•_...,
..;...._ "'- - - - , 9
":' __-,-~-"'....--•-,'"'--
(/)

0
a.. ·+H fl: '/I lf·H- li"II I :-+·i· I···IJ H-+~1i1+ tt··l:f1l h-+·-~ 1··· ~
1
/ 11 If
13
JT/ 1+-L~---f j l··j···l··· f'll··r'· . +. 1:tl fl::.:.ll:+,-··
>-
(/) +r :ij'ilr Tr/-=;.1:·I -~I ··t -rr+t·
·,_._J -·1 · ·+·· r ·i" ··HI ~7f·{Tr .1.11·
1··
/
1·· o

~mt
0
u
(/)
'=i 11 I Hll l• ''l!;J J

.. ~ -It -11._ -~ : 1 : lf ~~ :111J

>
=-1~rr'
f: II g ' i iI I I;

- ·-
··· - i
: - ..--.. --.. ·$· l·t~ .· . ~-
. "f ~
~ :IfI it!J· itrr-,
...'... ..,.
iilr· I i' i.i

~,ij-c£
--..:
'- -
-
:
- I•
i

:
..;.
.
.·.;
: I,
...;...., I
,. ,
.. . -t--
..........._.
I
I
I
..- -~ ·= -~=- ..._· ~ - - Ir- ~ ._· ·_·_ ~-- 1· "Hl H+·+··- ~
Ji
I!
i! ! ii I I
I-+++++;-+; if+++-r-+-+-.......;+++;.+-,f-+~,....,...++,i-++-it-,,tttt-t---+,----1
i i : I! I
l!I ,: : !! I ("'")

I:
~-~ ~
.. _ .__: _.:_,: ,: '.. ·I ·..___.1._
.. -'-r
1 r .. __.Li 1i11.. . . . . .. •! : 1u.1 iLI . ...
l!i l!!r 8
~ ..... ,...I: ~
I II UII ;! ! I i i I I ("'")
__._ : .. l~ . --~· - -~ ~- T. ;. - . ....... 1; ~~ .(.(. --t-·
l: ../. . . l !:.... .... :~ . ~~). ~~-i···
-. -t-

-~) ··· fi · -- .-
• P-• •. . .•• - - ~,-. ·· ~ • ••

-t-t-t-~-=-"'"''..,.
:, :
··~~ ..
~~

·1tt- l: -r
iI

0
U")
....0
I: I
::~,:~ Ij::
lo'i,, i! I (\j

Lh, ,1 1 ! ~ s:::
o 11"L, IEI i ! , I! i I w .....0
~ -r~ [ITI' ' I ~.
++-+----...... /r/" it :f. iii I
C
:J
+J
a:l
..__..,. I I I I; I I-
J
·-E
i...L., 0
s:::
~
0
I""! W' I ~ 1 N

--t:··r~ ···--..- .-. . +,·-nr=-1t......f---. ....... - -·. +-

i. I 1 , 11, •: ! 111 w
......
____i-_
+_i_ ·..·j·.i _ ! _· ·:_ -f~ ··-~~ 1
_++
_ -- --. . H+ 1H·f 1 c... ,._,
Q)

.. ,..,;r I ··11 ~ ··-T · ~-~ .... -r-

llJ

e ·-·--- . . +mt·. . ·· ·· · ft i1Ti.~,!· }.i,·T1·_".T.·.·_

1 1
,- +J
Q)

ftt--,.:::,i. . ~H'l++,...::_. . . . .;.-.. . . .~~-~::::_1.~::...+H-H+,-++--+-+--1:.~~:+-1+-..;.!..... ~I+

.: ; -
0 Cl
u,
I
I I I j I ...;
~;.,-------!·
I J l-;...j
!"""~1
..

1--
' ....I-......J.-.+.+.~,',.__.,1_......_..... l! ........-+-........ ~•;1 ~ V L-+L
_ _ _..... I -H+++-+-+-......... .,..I- :;-: -t--:...-
: ~ ~
I;:;
z

~-t=:..=j-~'"':..t·;:::;::~ ~ ~ ~ rn:P~-n::..::ii:::n~•
~

...... - · ... . · ,1·- - .. .. . 1+-- - ........+·-b·n.r:

........ -~.t
· r,r.L.·:;::..::;:·..:::::n-H.T·, ;:-·-1-=-·:;::,=:...,l:f.:=-
~·:;-·µ.:·~· f-L

I I .,; ~ ii g-+-+---+rr+++---~1-1----H+++++-I~,....--
; I~++.....,.~~='-· '...,,..,.
'· !~ - 1 0 .....bD
0 r...
I : I 11 '~ -- ./. :;
: I ··· { ... -··· - . . - -- ~;. '........ ;:! _ ;11_1_:.Ll-~-
·.·.·-r- --~-1---- : _· ~-+-._._. ;;
····- ... . . -c.~ ;., ·}_-1 ~-1--- ·-r- 11

tt :_·. 1,·

. _,
- · - . . . .

. _,__ - - .
. ·- -· - ·· - - · . . - .. - -~- . .

1" .-
- . - - ... t, ..tit .t.t.· ~-Jt~.-
=E ii : !! O

1111 ~ :, -- :• 1 ,;; 'II ILl

ttl++-+-1--+!J'+-t-+...-+-+---+++++-+-+---+-........-H-+-+-<l-fl'--t--~.,.........~
...:.. ·-+· "~ .,;I···1 .. - -· . · j ·~-- ·_
-----[·--_·_. :,:,_ .,1 ..... · - .. - - . . .. - ·!' - .. . - . . tH· H-f '·f-·
····,· ··-·t·· · __ LI -- . . . --·- .., ••.i-"- - -
I i! ! ! l
111
- --+-+-+-ttt ++.
I ·-':"-+--+-+-+++-<, ~,-+-+~rttt-++-+-++-r+++-+++-+-~++--t-++++~r.tt1~!rl i~ I ,~I '"r-1

-1=ff:j I( ~ :-~t-f :·~Ii :\: -it fa !th

0

E
0

Q
,-
~
_// I•~:-, I:-~. ~
.:.-'< . , . ._ _ ,._. .__.-.,;. . . . _ _ _ _ _ ._. __.______
=......
~ ~ ~,~
b - o
U")

N ... "°
<
c:: 0
c::
 14

BONNAURE MIXTURE STIFFNESS (Sm)

In conjunction with Van der Poel's method of estimating binder stiffness,

several investigators ( 15, 18-20) have provided methods to approximate the

stiffness of the asphalt mixture (Sm). Most methods involved a narrow range of

asphalt mixture types. However, Bonnaure et al. (21) developed an equation to

arrive at Sm which is based on dynamic testing of a wide range of mixture types.

Bonnaure fil fil. utilized the Van der Poel binder stiffness value (Sb), but on plant-

aged asphalts, to better represent asphalt stiffness in the field. This procedure has

been adopted in the Shell pavement design method (22). A nomograph depicting

its use is shown in Fig. 4. The Bonnaure et fil. equation is as follows:

for 5 X 106 < Sb < 1 as N/m 2 :

for 1 as < Sb < 3 X 1 as N/m 2 :

where:

Sm = mixture stiffness, N/m 2

Sx=0.6log[l.37vt-1]· .........•....... (8)

1. 33 vb - 1

Sy= 8.0 + 5.68 X 10- 3 Va+ 2.135 X 10- 4 Va,· • • • • • • • • · (9)

 15

St,ft ,.,ru f'T'IOOv l v,

N Jm 1
8U 90

......... .,, ·, ' I"- ~I"- 10 ,

' t--..
"""' 'y

0
,o • (V
. ~ '-' "'' ' "" "'~~
I
,o
60' "

. ' "
~~--- ------ ~
~ ":::,,_ i'-..
[",.
,o"
..,,, ~r-c-- ·- ·~ I~
-~-"--
.... ' ~
"~ "
70 " " " "
' " "' " " ' ' ' I'-
' " "- '-{
' "' " "'' , '\_""-.f'I'--..' "I'- '" ' ~'
0 '),.,, '

~
'\
I'-. I'-. ....
~ .... i'-..
~~
"'
' I
'i'-.. ~
ol ~ bo1vrn.nout
1
lu1

bi"°" · "' ' ' '

' 8S
Ntm

v. ~ l~ '" ' . "

r,,, r,,,
" "
I
•o ' "' ' ' "
~ "' "' " "' ' ' "II"-,
100 JO 70 10 S , "\ 1 s -
~\ \ \ \
,~\~ ~
'-\ '\ \ \
\
'\ ""''
'~
I
,"' ' "
'- " ' "' I'-
'
"I',. I',.'
I"-

' ' ·- Po\.

65
-
---
1o"
'!Y ,-\- ~
"' ' " " --
' .,
~\ I " ' ,, " --
' "' "I'-. ' " 1,
_.il

..
I\ '
I\\ I
"' ~\\ hl "-['\_ !,()
-
- ~
~ 0. "~ ~ "~ ~ -
-
'\ .
"~ ~
~o
- 10 '
=
-
-- 10 -
-

i=
" ,'>
\\1'~

,,
M,nc1al

---
eo . . ....
"'19'~ 1r \\" ~
\' '\
10' ll
~ 8 ftu1nie-,, ~~ ~
,_ -
6. .__l!!!__
o,o.,
Jv.Jr ' "
\ \
'
'.
S 10 70 30 100

S 11 t1"~ tT'OOulu1 ol n~ ,~o,,,ut"CI }

1a8 Nlm 1 8 , tun,, nou, bo~ • ,

l,.n&, ] •

V o l u"'< oft,.no., V 13 I 0
10
moo.J i u,

o f O'c !THI
0
,a • t \' .I

V olu"'< 01 m,.,.,., :...... ,. V, 8 0 5°/o l 1 • 10 0 N l m

1 1

Fig. 4. Determination of Mixture Stiffness (Sm).

 16

S z= l O. 82 - 1 . 34 2 [ l ia0 +- ~b] . . . . . . . . . . . . . . . ( 10)

S"'=0.76 {Sz-Sy) • • • • • • • • • • • • • • • • • • • • • (11)

v. = %volume of aggregate

Vb = %volume of binder.

Plant-aged asphalt penetration and ring and ball values can be obtained from

recovered asphalt from field cores or simulated from recovered laboratory-aged

asphalt residue. In lieu of laboratory aging of asphalt, Coree and White ( 12) drew

upon the work of Ullidtz (23) to approximate aged asphalt penetration (pen,) and Pl

(Pl,) values to arrive at Sb:

Penr = 0.65 Peni • • • . • . • • • (12)

PI -[ 27 log penr - 31.2 ] . .... . (lJ)

r 76.35 log penr - 219.27

where: Pen; = original penetration at 77°F.

Thus, knowing characteristics of the asphalt and conditions of temperature and

loading, Sb(N/m 2 ) can be calculated instead of using Fig . 1:

where: t = time of loading, sec

Pl , = Pl of recovered asphalt

TA&B.r = TA&B or Tpen800 on recovered asphalt

= 99.13 - 26.35 log pen,, °C . . . . . . . . . . . . . . . ( 15)

 17

Tp = asphalt temperature, °C.

Coree and White (12) drew on work by Witczak (24) to estimate the prevailing

pavement temperature (Tp), which is a function of air temperature and asphalt

pavement layer thickness. Witczak's equation predicts pavement temperature (TP)

at any depth:

T.., = 6 . 0 -
...
34
(z + 4)
+ TA [1 + ---- ]
(z
1
+ 4)
.... (16)

-
where: T A = mean air temperature

z = depth in the asphalt layer, in, usually taken at one-third

depth .

Alternatively, the mean-value theorem (25) can be employed to f ind the expected
-
mean layer temperature TP in a layer of z-thickness :

4
34 ln [ - -]
TP = 6. 0 - 4
_ _ _ _z_·--+TA [1. 0 - 4
(ln [ - - ) ) / 2) (17)
Z Z • 4

-
where: TA = average yearly air temperature, °F

z = thickness of asphalt layer , in.

The Ullidtz "Sb" equation is based on the Van der Peel nomograph. It is

considered applicable for the ranges of:

0 .01 < t < 0.1 sec

-1 .0 < Pl, < + 1.0

-1 0°c < T A&s., - T p < 70°c

 18

McLEOD MIXTURE STIFFNESS (SmPVN)

McLeod (20) developed an alternate approach to determining Sb, which was

orginally proposed by Lefebvre (26). McLeod felt that the Pl did not handle waxy

asphalts well because these asphalts tend to exhibit a false softening point. He

advanced a different measure of temperature sensitivity called the "penetration-

viscosity number" (PVN), which is calculated as follows :

log V = 4.25800 - 0.79670 log P • • • ••• (18)

1 og V = 3 . 4 6 2 8 9 - o • 61o94 1 og P . . • . • . ( 19 )

PVN = [ (log L- log X)/ (log L - log M)] (-1.5) · · (20)

where: V = viscosity in Cs at 275°F ( 135°C) for an asphalt with a

PVN = 0.0 (Eq . 18) or PVN = -1 .5 (Eq. 19)

P = penetration at 77°F (25°C)

L = antilog of log V (Eq . 18)

M = antilog of log V (Eq . 19)

X = viscosity in Cs at 275°F ( 135°C) for the asphalt at any PVN (the

asphalt of interest).

McLeod modified the Van der Poe I nomograph to allow substitution of PVN for Pl.

He also substituted a "base temperature" for the T R&B · To use his Sb nomograph,

one first needs to determine the difference between base temperature and the

penetration test temperature. This is done by use of Fig. 5. Then one enters

McLeod's Sb nomograph (Fig. 6) with load duration, difference between base

 19

w 130
c::
::,

-~
1-

c::
w
125
120
c..
I I :i
w .l..
•- 110 /....._

IO:i '
~
$)
/....._
w 100
§1-- j
..... <.fl
<w ~
/
1--
c::
w z 90 ~ /
c..o
""C -
~ /
-
w <
I-
1--
c:: 80
;....._~..y /
~~
Wl-
<.llW
< - 75
~'
s~ .,..s /
/
:~ 70
A..$"cf .,.<;/
~~ 65 ;:::-
(S~
> ~

-
I-
I-
60
,._~
u.·
C) 0
er "'- A.. /"'<?
I.J~ 10
y u. 55
~~/I
zw
0

- >-
so '?c., .s. 0
5 1--
(.11
w
w O
u _.J
z: c..
45
s s I-

j~
~o :::
w :! .§>/ - 5 0
C:
w
w
35
::s I-
I... ~-:/ '.J EXAMPLE <
c::
I...
30 ~
c ~
I
, GIVEN 3
1--
w
l.,J
25 /~ , ,; PENETRATION AT 25°C = 9 0 :::
r.: 0 s PEN-VIS NUMBER= -1 ·0
20~ '<-
I c:'
THEN

w ,~[ FROM GRAPH

TEMPERATURE DIFFE RENC::: = 20°C
AND
j21

ol BASE TEMPERATURE=25+ 20 =45°C

Fig . 5. McLeod's Temperature Difference Nomograph.

 20

..:. -
C::. C "+ '° ------~----..--~..J-------,,_..;..-:,-..-:..-:,_.'--:,_0:::...,,...-.....:=--:-_.. . c::, z::z
::f 2: .. ~ ----------"~----1---,--rL-:-....:......,,......:__:,--..:..,,...-.:::.."?'".-::.::,........ } . ~ z
( ,
'-.
-
...,_
< t-1 V')
-
-
<
:;, C:
,- 0
>I "
:.:z..,
'-' -, ....,2 2
(..
c.. '- '-
-2
,0 00 1,000 tOp::,o
v1!.COS1T Y, GP
I \
I \
I \
BAS( Tf .. PfRATUf<f
\
TEMPJl'u.TuR( ,11.:: EIUDw BAS( l[MPfRA11uR(
100 ::,o . O 10 20 !>O • O =,o 60 70 t10 9-0 IOO

01F~~A[~~ 1J~ ~(~W~NI~

1

I t.r.O I>,<.!;( T[MP(k.411.JA(

sP~~;£; \
01
Tf ~ ~ T ~

I \
I
I
\
I
\
I
I
\
I \
I \
I \
I \
I I ko/cm2= 14 ·2psi = 9·81 X 104N;m2 \
I
\
I
I
\
,.,VISCOSJTY \
1"JV ,c, 1 , S[C \ ~ IMIII IHOUA \ 10.:.Y tW((I(
I ... f.. ~ 2 ,t, 6 4 2 , , /
' , < lt £ ~ 6e 2 "6 i,
1
2 ,; i;e' 2 ,; 6t 2 4 C>e 2 4 t>t
'I \
2 4 t,t
'1 2 4 U I
CK,~. .:,.,;,, O·t IQ 100 I po,:, 10,p()O 100;,<::,0 t,/XNp:::xJ
Lc.<.::m,c T11,,i(, SE.C

Fig. 6. McLeod's Modified Van der Poel Nomograph.

 21

temperature and temperature of interest, and PVN .

Kandhal and Koehler (27) found that PVN did not change with in -service

aging, while Pl did. They also found that PVN predicted mix stiffness better than

Pl. Thus, PVN was considered by Kandhal and Koehler to be a superior method in

prediction of temperature sensitivity. The use of SmPvN will be examined in this

study.

CALCULATION OF MIXTURE STIFFNESS

Thus, the stiffness of an asphalt mixture (Sm) can be approximated by

determining six independent variables :

* initial penetration (pen;) @77°F

* time of loading (t)

* annual average air temperature (TA), °F

* volume of binder (Vb)

* volume of aggregate (Va)

* asphalt layer thickness (z), in

* viscosity at 275°F ( 135°C) if using McLeod method.

In this study, four methods with which to calculate Pl were used:

1. Calculation of Pl, via Eq. 13 by use of initial pen l; tration (25 ° C)

values.

2. Conversion of initial penetration values at 4°C and 25°C to aged

values via Eq. 12, then use of these in calculation of Pl, in Eqs . 3 and

4.
 22

3. Alternate equations for conversion of penetration values at 4° and

25°C were developed in this study from a review of the literature (27-

30):

log Pen4aged = 0. 942 1og P~n 4 orig -0 .124 . . . . (21)

log Pen 2 saged = 0.559 Pen 2 sorig + 6.033 . . . . (22)

Unaged penetration data developed in the present study were then

substituted into these equations to estimate aged penetration values.

This information was substituted into Eqs. 3 and 4 to calculate aged

Pl values. Aged T R&B values were estimated by substituting the

values for Pen 25 calculated via Eq. 22 into Eq. 15 .

4. In a similar manner to method 3 above, an alternate equation for

conversion of Pl values directly to aged Pl was developed :

Piaged = 0.530 Piorig - 0.639 . . . • . . (23)

Aged T R&B was calculated by the following methods :

1. Conversion of penorig values to penr values via Eq. 12 and then

substitution into Eq. 15.

2. An alternate aged T R&B equation was developed from the literature:

TR&B,aged = 0.843 TR&Borig + 12.501 • . . . . (24)

Then Sb was calculated by use of two methods:

1. Calculation of Sb via Eq. 14 by use of the various values of Plr.

 23

2. Calculation of Sb via Fig. 6 by use of PVN from Fig.5.

Sm values were then calculated via Eq . 6 by use of Sb values determined from each

of the above two ways. This resulted in five different ways to calculate mixture

stiffness as summarized in Table 4. Mixture stiffnesses from all methods were

then correlated to laboratory-derived values of resilient modulus to determine

which Sb method was superior.

Table 4. Alternate Methods of Mixture Stiffness Determination.

Mixture Methods of Parameter Determination

Stiffness
Parameter Penr Pl TR&B Sb Sm

Sm Eq. 12 (Ullidtz) Eq. 13 Eq. 15 Eq . 14 Eq. 6

(Ullidtz) (Ullidtz) (Ullidtz) (Bonnaure)

smm Eq. 12 (Ullidtz) Eq. 3 and 4 Eq. 15 Eq. 14 Eq. 6

(Pfeiffer and (Ullidtz) (Ullidtz) (Bonnaure)
Doormaal)

smrm,aged Eq. 21 and 22 Eq. 3 and 4 Eq. 24 Eq. 14 Eq . 6

(Richardson) (Richardson) (Ullidtz) (Bonnaure)

sm,aged N/A Eq. 23 Eq. 24 Eq. 14 Eq. 6

(Richardson) (Richardson) (Ullidtz) (Bonnaure)

SmPVN N/A Eq. 20 Fig. 5 Fig. 6 Eq. 6

(McLeod) for (McLeod) (McLeod) (Bonnaure)
PVN

SUMMARY

The mixture stiffness (Sm, smm' sm,aged' smrm,aged' smPVN) of the mixtures

examined in this study were estimated by six methods; five using an aged form of

Sb and one using SbPVN· The results were correlated with actual test results of
 24

resilient modulus to judge which estimation method was best. Then the results

were used in the Odemark equation to determine layer coefficients:

Sm, UMR or MHTD]l/3

a n -- a n, AASHO Sm, AASHO

ESTIMATION OF DYNAMIC MODULUS

GENERAL

Shook and Kallas (31) developed a regression equation to estimate the

dynamic modulus of various bituminous mixtures from mixture properties .

Numerous modifications have appeared in the literature (32-34), culminating in two

alternate equations. The choice of equation depends on the character of input data

available to the user. Both equations were used in this study to assist in

determining layer coefficients . The two equations are presented by Akhter and

Witczak (34), and are as follows:

log iE*i = 1.45716-0.0256272*Pair+0.012792l*P3 ; 4 +0.0627099*~

-0.00837349*T+0.147306*log f+0.0000193164*logf*T 2 (25)
-0.0000254l03(Peffv-Popteffv+8.0) 0 · 5 *T 2
-0.000149152*Peffv*P4+0.00591768*P2oo*Pabsw

log iE*i = 1.42841-0.0233473*Pair+0.013004*P3 ; 4 +0.0627099*~

-0.008145*T+0.146970*logf+0.0000193776*log*T 2 (2 6)
-0.000073466415•T 2 -0.000138513*Peffv*P4
+0.00583715*P2oo*Pabsw

where IE* I = dynamic modulus, psi

Pair = percent air voids

 25

p3 /4 = accumulative percent retained on the 3/4 in sieve, by

weight of aggregate (called AR 3 14 in the rest of this

report)

P4 = accumulative percent retained on the #4 sieve, by weight

of aggregate (called AR 4 in the rest of this report)

P200 = percent passing #200 sieve, by weight of aggregate

f = frequency of load application

f} = absolute viscosity @ 70°F, poises x 106

p effv = percent effective asphalt content, by volume

T = temperature, °F

p opt eff v = percent effective optimum asphalt content, by volume

p absw = percent absorbed asphalt, by weight of mix.

The dynamic modulus ( I E * I ) is defined as the elastic portion of the

complex modulus, which also takes into account the viscous nature of asphalt

cement. The complex modulus is expressed as follows:

E* = E' + iE"

where: E' = I E .. I cos c5

E" = I E* J sin c5

i = imaginary number

c5 = phase angle, represents lag of strain peak behind stress peak.

If the phase angle is assumed to be zero, which approximates what would

occur for short, relatively light load applications, then the complex modulus is
 26
reduced to one term, which is commonly called the dynamic modulus:

( 2 7)

where: amax = maximum applied sinusoidal stress

€max = maximum recoverable strain.

The dynamic modulus can be determined by applying axial compressive

cyclic pulses in the form of a compressive sine wave to 4 in ( 10.2 cm) diameter 8

in (20.3 cm) long cylinders. The test has fallen into disuse because of its

cumbersome testing technique and because it has been criticized for not being

appropriate for pavement design /analysis methods which use elastic layer

assumptions . It has been largely supplanted by the repeated load indirect diametral

tensile test, which is discussed later in this report . However, the Akhter and

Witczak equations are still well-known . Their usefulness was examined in this

study .

SUMMARY

The dynamic modulus concept is useful for this study as follows. The

required input for Eq . 26 (referred to hereafter as the Akhter-Witczak equation) is

available from Road Test data, UMR-study data, and MHTD mix design data.

Thus, use of Road Test, UMR-study, and MHTD dynamic moduli in the form of

estimated values can be used in the Odemark equation in the determination of layer

coefficients:
 27

i: E• : UHR or KHTD]
113
a n. UMRorHHTD = a n ,MSH 'E • '
I I AAS HO

RESILIENT MODULUS

The resilient modulus is a repeated load test that is similar to the dynamic

modulus but with several important differences. The applied stress wave form is

usually in the form of a stress pulse followed by a rest period, rather than the

sinusoidal wave form (with no rest period) as used in the dynamic modulus test.

The test equipment is less complex. There are two different ways in which the

test can be performed. One method, AASHTO T-274 (35), utilizes compression

testing of 4 in ( 10.2 cm) diameter 8 in (20.4 cm) high specimens in a triaxial

chamber. The second method (which is now the predominant method), ASTM

04123 (36), involves the inducement of indirect tensile stress diametrally to

Marshall-type specimens . The latter test is more convenient to perform, and is the

method used in this study . It is the recommended test in both AA MAS (Asphalt-

Aggregate Mixture Analysis System)(37) and in the latest SHRP (Strategic

Highway Research Program)(38) protocol.

Several regression equations were developed in this study to enable

practitioners to estimate the resilient modulus of asphalt mixtures by the simple

substitution of mix characteristics and other readily accessible data into the

equation. One equation is solely based on experimental data generated in this

study. A more generally applicable equation was developed by use of resilient

modulus data gleaned from the literature combined with data from this study.
 28
Layer coefficients were determined by applying the resilient modulus data

developed in this study along with estimations of resilient modulus to the AASHTO

nomographs. Also, layer coefficients were determined by use of the equivalent

stiffness method (Odemark equation). The input included estimated resilient

modulus data and estimated resilient modulus from AASHO Road Test mixtures:

3
- _ MR, UMR or MHTD]l/
an, UMR or MHTD - an, AASHO - t---.,.,M:-------
R,AASHO

LABORATORY INVESTIGATION

MIX DESIGNS

Mix designs were developed for Type C, 1-C, and plant mix bituminous base.

The main thrust of this portion of the study was to determine layer coefficients for

the three types of mixtures based on repeated load indirect tensile diametral tests.

This is the test recommended in the 1986 AASHTO pavement design method ( 1).

A secondary goal was to develop regression equations to enable the subsequent

prediction of resilient modulus without having to actually perform the test in cases

where MR test data is unavailable.

As mentioned previously, the factors that affect asphalt mixture stiffness

most significantly are temperature, effective asphalt content, voids, asphalt

viscosity, loading frequency or duration,. and gradation.

The study was limited to observing the effects on resilient modulus by

varying aggregate gradation, coarse aggregate type, temperature, asphalt cement

grade, asphalt content, and indirectly, void content. The latest SHRP protocol (38)
 29

has omitted test load frequency as a variable, therefore frequency was not varied

in this study.

The proposal for this study did not include the performance of m ix designs

for two reasons: 1) limited available funds and limited contract duration, and 2)

the thrust of the research was to study the effects of varying the aggregate

gradation across an acceptance band, with the interaction of asphalt content and

grade. However, to assist in determining optimum asphalt content during this part

of the study, Marshall mix designs were performed for all mixtures containing ~C-

20 grade asphalts . Some mixtures had AC-10 asphalt substituted for the AC-20

with no change in mix design because mixing and compaction temperatures were

controlled to give the same viscosities independent of asphalt grade .

Two gradations each were chosen for the Type C, 1-C, and bituminous base

mix materials, resulting in six gradations. These were picked by a process which

involved determining the coarsest and finest job mix formula (JMF) gradations that

were approved by the MHTD during the 1990 and partial 1991 seasons. Then, to

get an even wider separation of gradation, the coarsest JMF gradation was pushed

to a coarser gradation by use of the maximum allowable tolerance on each sieve.

Likewise, the finest JMF was pushed to a finer gradation via the maximum

allowable tolerances. Some adjustment was necessary to keep the gradations

within the master specifications on some sieves, and at the - #200 sieve in order

to prevent cross-over of the experimental gradation lines at that point . It was

realized that some of these mixes may not have been approved in routine work,
 30
but they do approximate where some gradations may end up in the fie ld after

adjustment.

Two coarse aggregate sources each were used for the Type C, 1-C, and

bitum inous base materials . These were chosen by MHTD personnel to exhibit a

wide variation in particle shape and texture .

Two asphalt cement grades were chosen (AC- 10 and AC-20) to represent

the most commonly used grades for MHTD mixes .

Using the above 12 gradation/coarse aggregate combinations, Marshall mix

analyses were made to determine optimum asphalt contents. Once these were

determined, an additional asphalt content was determined by arbitrarily adding

0. 5 % asphalt to the previously determined optimum asphalt contents. Thus , 48

m ixtures in all were evaluated in the res ilient modulus testing: 16 mixtures each

for the Type C, 1-C, and bituminous base mixtures. These are shown in Table 5.

Table 5. Asphalt M ixture Mix Design Parameters.

Coarse Aggregate 1 Coarse Aggregate 2

AC-10 AC-20 AC-10 AC-20
Asphalt Content Opt. +0.5 Opt. +0.5 Opt. +0.5 Opt. +0.5
% % % %
Fine Gradation X X X X X X X X

Coarse Gradation X X X X X X X X

Note : 1 . This chart applicable to Type C, 1-C, and bituminous base mixes . Thus 48
mixtures were used .
2. "X" denotes that this combination of parameters was
represented by a mix.

The choice of using the Marshall mix design method for determining
 31

optimum asphalt contents was based on several factors . First, it is the most

commonly used method by state and federal agencies, private practice, the Asphalt

Institute, and the National Center for Asphalt Technology . Second, the MHTD

utilizes this method to a certain extent in its mix design evaluation. Third,

personnel communication with MHTD personnel indicates that the Marshall method

will be the preferred method if a contractor QC/QA program is initiated, and fourth,

the UMR Bituminous Laboratory is equipped with Marshall equipment and has

experience with using this method.

The optimum asphalt contents were chosen based on percent air voids as a

major criteria, but also were optimized in an attempt to satisfy MHTD requirements

for stability, voids filled, dust/asphalt, VMA, percent natural sand, inclusion of

hydrated lime, percent asphalt, and makeup of fines where applicable. Although

flow is not specified by the MHTD, this parameter was also used as a guidance

criteria.

It was decided that in order to make comparisons from mix to mix within a

given type (C, 1-C, or bituminous base) of mixture, the percent of aggregate

constituent (coarse, natural sand, manufactured sand, mineral filler, and hydrated

lime) per sieve would be kept constant. For example, for the Type C mixes, at the

#16 sieve, all 16 mixtures would retain 0% coarse aggregate, 79% natural sand,

21 % manufactured sand, and 0% mineral filler.

The decision about the kind of materials going into each mix type was made

by examining the 1990(-91) MHTD-approved job mixtures. It was found that, on

 32

average, the Type C mixtures contained 48.0% coarse aggregate, 24.1 % natural

sand , 22 .9% manufactured sand, and 5.0% mineral filler . The percent retained of

each type of aggregate on a particular sieve had to be changed from sieve to sieve

in some cases in order to make the total contribution of each material type

reasonable. The gradations are given in Table 12 in the "Results" section of this

report.

The following are brief descriptions of the test methods employed in this

study. Where applicable , AASHTO (39) standards were used .

ASPHALT CEMENT

Penetration

Both grades of asphalt were tested for penetration as per AASHTO T49-89.

Test temperatures were 77°F (25°C) and 37.8°F (3.2°C) . The penetration at 77°F

(25°C) information was necessary for calculating PIR&B• Plpen/pen • PVN , and for use

in the Ull idtz aged penetration equation. Penetration at 37.8°F (3.2°C) was

necessary for calculating Pl pen/pen· Both were used for estimating T penBOO·

Kinematic Viscosity

The asphalt cements were tested at 275°F ( 135°C) in accordance with

AASHTO T201-90. This information was required for calculation of PVN, for

determination of mixture mixing and compaction temperatures, and for estimation

of viscos ity at 70°F (21 . 1°C) .

Absolute Viscosity

Test temperature of this AASHTO T202-90 procedure was 140°F (60°C) .

 33

These data were necessary for determination of mixture mixing and compaction

temperatures and for estimation of viscosity at 70°F (21.1 °C).

Specific Gravity

The AASHTO T228-90 procedure was performed on the asphalts at 77°F

(25°C). These data were used for volume calculations in the mix design process

and for estimation of absolute viscosity at 70°F (21 .1 °C).

Ring and Ball Softening Point

This AASHTO T53-89 procedure was performed in order to calculate PIR&B

and for use with the Van der Poe! nomograph. Fig. 7 depicts the ring and ball

softening point device.

AGGREGATE

Initial Gradation

Gradations of the aggregates, mineral filler, and hydrated lime supplied by

MHTD personnel were determined in accordance with AASHTO T 27-88, T 37-87,

and T 19-87, respectively. Weighing was performed on an electronic balance

capable of reading to the nearest 0.1 g. This information was useful for

determining the necessity of additional material to build the test gradations.

Final Gradation

Each of the 12 gradations were built on a sieve by sieve basis as mentioned

previously. Final gradations were necessary for determination of aggregate specific

gravity and for making asphalt mixture specimens. Fig. 8 depicts the steel storage

bins used for each fraction of each type of aggregate.

Fig. 7. Ring and Ball Softening Point Device.

Fig. 8. Asphalt Mixture Aggregate Fraction Storage Bins.

w
..:,.
 35

Particle Shape and Surface Texture

Numerous studies have shown that aggregate particle shape and texture

have a significant effect on various properties of dense-graded asphalt cement

concrete mixtures. It is difficult to separate the effects of shape and texture. The

general consensus seems to be that as angularity and surface roughness increase,

the following also increase: stability, resistance to rutting, VMA, and optimum

asphalt content. Opinions are somewhat mixed as to the effect of shape and

texture on static indirect tensile strength (IDT). In regard to IDT, Kalcheff and

Tunnicliff (40) found little difference between various particle shape /textures . The

explanat ion was that, in compression, particles attempt to slide past each other,

therefore shape/texture is important. But, in tension, the effect is much less

pronounced. However, Hadley, et§.!. (41,42) found that more angular/rough

particles do tend to result in higher IDT values. Also, the literature indicates that

the characteristics of the fine aggregate are much more important than those of

the coarser fraction.

Numerous test methods have been devised to quantify particle shape and /or

texture. These can be divided into direct methods (those that result in

measurements or aspects of individual particle shape or texture) and indirect

methods (those that measure some sort of bulk aggregate property, such as void

content, which is related to particle shape/texture). Recent evaluations of these

methods were reported by Meier and Elnicky (43), Mogawer and Stuart (44), and

Kandhal et§.!. (45) at NCAT (National Center for Asphalt Technology) . It appears
 36

that efforts are being concentrated in the area of fine aggregate evaluation, and

that there are several methods available which can be used in lieu of the standard

test, ASTM D 3398 (46) which is somewhat cumbersome to perform. Kandhal et

§.[. recommended the National Aggregate Association's (NAA) proposed method (A

or B) for fine aggregate (47). Both of these are indirect methods of particle shape

determination.

In this study, the(-) #8 to(+) #100 sieve size material of each asphalt

mixture blend were tested using the NAA Method A. The method is given in

Appendix A of this report. For the ( +) #4 size, the blends were tested in

accordance with ASTM D 3398. This method is given in Appendix B. The results

of both methods were used in developing the indirect tensile strength and resilient

modulus regression equations discussed later in the "Results" section of this

report. Photographs of the NAA test device and the D3398 equipment are shown

in Figs . 9 and 10.

Specific Gravity

Aggregate fractions of each of the 12 gradations were separated at the #4

and #100 sieve sizes and tested in accordance with AASHTO T85-88 and T84-88

for the ( + )#4 and (-)#4 to ( + )#100 material, respectively. For the (-)#100

material, specific gravity was determined in accordance with MHTD T37-4-84.

These data were used for voids analyses calculations. Weighing was performed on

a scale readable to the nearest 0.1 g.

 Fig. 10. D3398 Particle Shape Equipment.

Fig. 9. NAA Method Particle Shape Device.

w
'-.J
 38

Screening

All aggregates were dry shaken through the appropriate screens on a Gilson

shaker. A dust baffle/cover was designed to restrict the movement of particles in

order to minimize problems with incorrect sizes of material being retained on any

given sieve.

Upon being shaken, the split material was stored in 20 gal plastic cans with

lids until it was needed for specimen fabrication.

FABRICATION OF SPECIMENS

All asphalt mixture specimens were made in accordance with AAS HTO T

245-90. A manual Marshall flat-faced hammer was used to apply 75 blows to

each face of the specimens. The specimens were Marshall-type pucks: 4 in ( 10.2

cm) diameter 2 .5 in (6.35 cm) high cylinders. Usually, two or three pucks per

asphalt content were made for mix design purposes, two for static indirect tensile

strength, and four for resilient modulus testing. Two uncompacted specimens per

asphalt content were used in the Rice method for maximum theoretical density.

The mixing and compaction temperatures were determined by use o f

viscosity-temperature plots for both grades of asphalt, as shown in Fig. 11.

Mixing and compaction viscosities are commonly specified at 150 to 190

centistokes and 250 to 310 centistokes, respectively. From the relationship in Fig.

11, mixing temperatures of 298 °F (AC-10) and 302 °F (AC-20) temperatures were

chosen to achieve the required viscosities. Likewise, compaction temperatures of

 39
VISCOS1 T Y-TEMPERAT URE CHART FOR ASPHALT CEM EN TS
I CE

(j
@ 1~ 0 F Cst ~ I Poises / ( .98 x Sri cci l,c Gravily )! ~ 100
~
@275 F Cs t c I Poises/ ( .9 34 x Specific Gra vi ty )! x 100
~

2
- ·"·~
- -~
10 ~
6 l""' ' ' I

..:
' ' '\
'-
'\
I

I\.
3 '\ '\.
~--
2
"'\. "'\.
'\. '\
10 ..: '\l""-
e ' '
"" ""
' '\ '\
Cf)
w 6
~ I\.
0
I- ..: '\ '\
["'\ "'\
Cf)

I- !
2 '\. '\. I
w
0 2 '\. '\.
~ '\_' ~
>-
1-
vi
O
o
10 3
s
e
~'\ ' '
'\ '\
Cf)

> 6
7
'\. '\
'\
~

,.
~----- ------
8-
------ .... ----- i------- ------ ---~
COMPAC_T ION R, ~N GE
~ ____ J_ _____
"""" \J\.
iI"" " I----- ·
2
~----- ---------- - ------r------ ~------ L.. ----- t~1--~
I
A- Ml~ING Ri JNGE I, ~
11
I1
l,
~I I"'
\1
E I I
,1
7
1J I•

6 11 tl
;,
I
11
11 .I
/?~ 1.; o
- V 160 18 0 200 225 250 275 30J
h .. po, , c:! l ,
f ~f A ::i hH t l lf l~TllU l[ TEM PERAT UR E IN °F
( :.> t : I C, f P .L , • ,,. ._ P-. 1 £ , ~r,

Fig. 11 . Asphalt Cement Temperature - Viscosity Relationships.

 40

279 °F (AC-10) and 282 °F (AC-20) were used. Details of specimen fabrication are

given in Appendix C.

Bituminous base mix aggregate and mineral filler were separated into 1 ",

3/4", 1/2", #4, #8, #30, #200, and -#200 size fractions. Type C and IC mix

aggregate and mineral filler were separated into 1 ", 3/4", 1 /2", #4, #8, #16, #30 ,

#50, #100, #200 and -#200 size fractions.

The Rice specific gravity specimens were made in a similar manner, with the

omission of the compaction step.

MARSHALL MIX DESIGN

The stability/flow testing procedure was in accordance with AASHTO T

245-90 . Prior to testing, the specimens and breaking head were heated to 140°F

(60°C) in a water bath. The pucks were tested in a Pine Press Marshall device

which applied the load via a motor driven mechanical jack at a deformation rate of

2 in/min (0.084 7 cm/sec). The load was sensed by a 10,000 lb load cell. Flow

was measured by a Schaevitz LVDT (Model GCA-121-500 S/N 4427) which has a

range of ± 0.5 in.

Both signals were sent through signal conditioners; the output was recorded

on a Houston Instruments 2000 XY recorder. Maximum load (stability) and flow at

that load were taken from the load-deformation trace. The test arrangement is

shown in Fig. 12.

MAXIMUM THEORETICAL SPECIFIC GRAVITY

The maximum theoretical specific gravity was determined for every mixture
 41

at an asphalt content near optimum in accordance with AASHTO T 209-90 (Rice

Method). Duplicate specimens were tested and the results averaged.

A water bath dedicated to the Rice system was used for both the

pycnometer calibration and test procedures. The electronic balance had a capacity

of 12, 100 g and was readable to 0 . 1 g. During the deairing step, a vacuum of 30

mm Hg was applied for 15 min . The system is shown in Fig. 13 .

Maximum theoretical specific gravity was also estimated by calculation as

discussed later in the "Results" section of this report .

SPECIMEN BULK SPECIFIC GRAVITY

The bulk specific gravity of compacted specimens was determined in

accordance with AASHTO T 166-88. The data were necessary for determination

of voids in the mineral aggregate (VMA).

VOIDS ANALYSIS

An analysis of the voids system in the mixtures was made utilizing the

results of the specific gravity testing . Air voids, voids in the mineral aggregate

(VMA) and voids filled (VF) were calculated. Air voids were calculated based on

both methods of maximum theoretical specific gravity determination as discussed

above, and the results were compared . The voids analysis was necessary for

determination of the optimum asphalt content of each mixture.

INDIRECT TENSILE STRENGTH

The indirect tensile test involves loading a cylindrical asphaltic concrete

specimen with a static compressive load. Loads are applied along and act parallel
 42

Fig . 12 . Marshall Stability /Flow Test Apparatus .

Fig . 13 . Rice Maximum Theoretical Specific Gravity Test Station.

 43

to the specimen's vertical diametral plane. Lower and upper 0.5 in wide steel

loading strips, which are curved at the interface to fit the radius of the specimen,

distribute the compressive load to the specimen.

The application of a compressive load to the .specimen induces a fairly

uniform tensile stress perpendicular to the plane of the applied load and along the

vertical diametral plane . This ultimately causes the specimen to fail by splitting

along the vertical diameter. The tensile stresses developed within the specimen

simulate the state of stress at the lower position of the asphalt layer in a pavement

structure, which is generally the critical area for fracture and fatigue cracking.

Procedures for the indirect tensile test have been developed and reported by

Anagnos and Kennedy (48-50).

The indirect tensile strength is usually determined in order to choose the

loads to be applied during the repeated load diametral resilient modulus test. A

percent of the total stress at failure is normally used . In the SHRP P07 (Strategic

Highway Research Program) protocol on resilient modulus testing of asphaltic

mixture cores, repeated load level is tied to test temperature : 30% at 41 °F (5°C),

15 % at 77°F (25°CJ, and 5 % at 104°(40°CJ.

The indirect tensile strength procedure involves testing a Marshall-type

specimen at 77°F(25°C) in diametral indirect tension to failure . The load is applied

at a rate of 2 in/min (0.0847 cm/sec). The stress at failure is calculated as

follows:
 44

2 p
s = ( 28)
rthD

where : S = indirect tensile strength , psi (Pa)

P = ultimate load, lbs (NJ

h = specimen height , in (cm)

D = specimen diameter, in (cm).

The test apparatus is depicted in Fig . 14. It is basically a Marshall stability

test press w ith a different break ing head . Load is sensed by a load cell; the signal

is condit ioned and maximum load is determined from an XY recorder plot . A more

detailed description is given Append ix C.

RESILIENT MODULUS

The resilient modulus (indirect tensile) test is similar to t he ind irect tensile

strength test except the specimen is not loaded to failure; rather, it is cyclically

loaded to induce tensile stresses in the specimen .

The equipment used for testing asphalt specimens for d iametral repeated

load resilient modulus was developed at UMR. Several modifications over a period

of t ime resulted in the dev ices pictured in Fig. 15 .

The total resilient modulus was calculated as follows (as per SHRP Protocol

P07):

= Pcyclic*D(o. 080+0. 297µt + O. 0425µ~) ( 2 9)

Ht*t

where: = total resilient modulus, psi

p cyclic = cycl ic load, lbs

 45

Fig. 14. Indirect Tension Test Equipment.

Fig. 15. Asphalt Mixture Resilient Modulus Test System.

 46
D = specimen diameter, in

µt = total Poisson's ratio

0.859-0.0BRt
=
0.285Rt-0.04

= total deformation ratio = (V tl +V t2)

2(H tl +H t2)

Ht = Ht, + Ht 2 (total horizontal deformation)

Ht,, Ht 2 = total horizontal cyclic deformations from horizontal L VDT #1 and #2,

in

t = specimen thickness, in

Vt,, V 12 = total vertical cyclic deformation values from vertical LVDT's #1 and

#2.

A description of the equipment and its development is included in Appendix

C. Also included is a detailed account of the test procedure. Minimum resolution

of the horizontal LVDT's, vertical L VDT's, and the load cell are listed. Actual

minimum deformations and loads during the testing were kept at least ten times

these minimum resolutions to assure confidence in the test results.

RESULTS OF LABORATORY INVESTIGATION

ASPHALT CEMENT

The results of the asphalt cement testing are shown in Table 6. The number

of test replications are shown which were necessary to stay within AASHTO
 47

precision guidelines.

Table 6 . Asphalt Cement Properties.

Parameter AC-10 AC-20

Value Reps. Value Reps.

penetration, 77°F (25°C), 100g, 5 101 6 71 5

sec
penetration, 37.8°F (3.2°C), 100g, 5 8 3 6 3
sec
kinematic viscosity, 275°F (135°C), 301 3 361 3
Cs
absolute viscosity, 140°F (60°C), 1099 8 1911 4
poise
specific gravity , 77°F (25°C) 1.007 5 1.017 5
softening point, ° F ( °C) 112.1 6 119.3 7
(44.5) (48.5)

Fig . 11 depicts the asphalt cement temperature-viscosity relationship, with the

mixing and compaction temperatures shown . A plot of penetration and viscosity

vs . temperature is shown in Fig. 3 . The penetration test at 77°F (25°C) was

performed as per AASHTO T49 with 1 OOg weight at 5 sec duration. The

penetration at the lower temperature was performed under the same conditions,

rather than the suggested 200g at 60 sec duration. A review of the literature

(28,51) indicated that researchers favor the 1 OOg 5 sec method when Pl is being

determined .

Sb values were calculated based on Eq . 14, which is based on the Van der

Poel nomograph (Fig. 1 ). Use of Fig. 1 assumes that the asphalt is an S-type of
 48
bitumen, as opposed to a W-type (high wax content) or 8-type (high asphaltene

type). de Bats and Gooswilligen (52) give criteria for qualifying the asphalts as to

type. The method is given in Appendix D. From this analysis, both asphalts (AC-

10 and AC-20) used in this study were classified as S-types, therefore use of Fig.

1 and Eq. 14 is appropriate . Also, Heukelom defines S-type asphalts as those that

plot in an approximate straight line on the BTDC paper. Fig . 3 reveals this type of

behavior for both asphalts used in this study. Fig. 3 also indicates that the T 800

and T R&B are quite close , as would be expected for S-type asphalts.

The calculation of PIR&B• Plpen/pen• or PVN is required for determination of

mixture stiffness. In Table 7 are aged residue estimations of penetration and T R&B

based on Eq . 12 and 15 , respectively. In Table 8 are shown PIR&B• Pl pen/pen• Plr,

and PVN as calculated by Eqs. 3 and 5, Eqs . 3 and 4, Eq . 13 and Eq . 20,

respectively . Use of Van der Poel's Sb nomograph (Fig . 1) indicates that the

magnitude of Sb is not changed significantly by use of PIR&B or Plpen/pen for these

type S asphalts . Thus , it would seem that use of the penetration test at a second

temperature (as opposed to the ring and ball softening point) is appropriate.

Table 7. Estimated Aged Penetration and T R&B ·

Asphalt Grade Penr TR&Broc

AC-10@ 77°F(25°C) 66 51.2
AC-20@ 77°F (25°C) 46 55.3
 49

Table 8 . PIR&B and Pl pen / pen / Pl ,, and PVN .

Asphalt Grade PIR&B Pl pe n ,pe n Pl, PVN

AC-10 - 0 .92 - 1 .49 - 0 .22 - 0 .65

AC-20 - 1.88 - 1.33 - 0.15 - 0 .75

Binder stiffness, Sb, was determined for loading times of 0.1 sec and '.).04 sec.

These loading times correspond to the load duration time of the resilient modulus

testing (0.1 sec) and to the estimated load duration time for MHTD pavements at

the average vehicle speed (56 .3 mph) and average asphalt pavement thickness

(8.33 in) in accordance with Barksdale (53) as explained later in the "Load

Duration" section. Sb is shown in Fig. 16 and Fig. 17 as a function of temperature

and method of calculating Pl, or PVN and is tabulated in Table 9.

Table 9. Binder Stiffness, Sb .

Temperature Sb (Pl ,). psi Sb(PVN), psi

AC-10 AC-20 AC-10 AC-20

0.1 s 0.04s 0.1 s 0 .04s 0.1 s 0 .04s 0.1 s 0.04s

40°F(4°C) 11,500 16,100 16, 100 22,625 7110 10, 670 14,200 21,335
68°F(20°C) 1450 2045 2480 3480 420 995 995 2276
77°F(25°C) 610 850 1160 1620 140 300 284 782
104°F(40°C) 8.8 12 .3 37 .8 53 . 1 10.7 21.3 21.3 43 .5

Aged residue data were used to better reflect actual pavement conditions . PVN

has been shown not to change with aging, therefore unaged penetration values
:::- 20000 50

- l:ll
0..

.c
rn 15000
l:ll
l:ll

-
Q)
s:=
:".:: 10000
_.J 0 AC-10, Sb(PI)r
I:'/)
e AC-10, Sb{PVN)
M
Q) t::. AC-20. Sb(Pl)r
'g 5000 • AC-20 , Sb{PVN)
•.-4
co

-5000 --~~--~~........~~~---~~--~~~
0 10 20 30 40 50
Temperature (deg. C)
Fig. 16. Binder Stiffness From PI and PVN for 0.04 sec
Loading Time. r

25000 --~~--~~--,.~~~.....-~~--~~--
• AC-20, Sb(PI)r @ 0.04 sec loading

-:::- 20000 0 AC-20, Sb(Pl)r @ 0 . 1 sec loading

-
l1l
0.. "' AC-20. SbPVN @ 0 .04 sec loading

.c '7 AC-20. SbPVN @ 0 . 1 sec loading

rt.l 15000
l1l
l1l

-
Q.)
s:=
:-;: 10000
_.J
I:'/)

M
Q)
-a 5000
s:=
•.-4

-5000
0 10 20 3040 50
Temperature (deg. C)
Fig. 17. Binder Stiffness From Plr and PVN for AC-20.
 51

were used in calculation of PVN. As can be seen , binder stiffness decreases with

increasing temperature and longer load duration (or slower vehicle speed), as

expected of a viscoelastic material. At most temperatures for both grades of

asphalt, the PVN method exhibited lower binder stiffness than the aged residue Pl

method.

As discussed previously and shown in Table 4, five methods of mixture

stiffness were calculated in order to determine which most accurately

approximated the resilient modulus test data generated in the present study. The

most accurate method would then be used in the Odemark equation to determine

layer coefficients . The results are shown in Table 10.

Table 10. Results of Resilient Modulus - Mixture Stiffness Comparison.

Mixture Stiffness Method R2

Sm 0.914

SmPVN 0.910

Sm,m.aged 0.886

Sm . aged 0.880

smm 0.849

As can be seen, all methods produced good estimations. The most accurate was

the Bonnaure method utilizing the Ullidtz asphalt aging equations, and thus this

method was used in the rest of the study to calculate mixture stiffness.
 52
AGGREGATE

Gradations

Gene ral. The results of the aggregate testing are shown in Tables 11 through 19

and in Figs. 18 through 24 . Table 11 shows the as-received gradations of the DR-

4 through DR- 11 aggregates.

Table 11. As-Rece ived Gradations .

Percent Passing
Sieve size DR4 DR5 DR6 DR7 DRS DR9A DR9B MF HL
1 in . 100 100 100 100 100 100 100 100 100
3 /4 in . 100 100 100 100 100 100 100 100 100
1 /2 in . 94 99 78 84 100 100 100 100 100
3/8 in . 63 71 -- - - 100 100 100 100 100
#4 5 18 52 43 99 100 100 100 100
8 3 4 42 32 93 90 73 100 100
16 3 2 -- - - 81 54 44 100 100
30 3 2 30 20 56 28 26 100 100
50 3 2 -- -- 31 13 14 100 100
100 3 2 -- - - 14 8 8 98 99
200 3 1 11 8 1 6 6 88 99
Note : MF = mineral filler ; HL = Hydrated Lime

Table 12 shows the gradations of the six final blends: two Type C's, two Type 1-

C's, and two bituminous bases. Fig. 18 shows the Type C MHTD master

specification, and final blends (fine and coarse) . Likewise , Figs . 19 and 20 depict

Type 1-C and bituminous base mixes, respectively . As explained previously , the
 53

100
,,
/
,,

-
/
/ I /
90
E-<
,, "' I

zt:z;:l //
,, I

I
I
I

80
u //
// I
I
I

~ //
//
I I

-
t:z;:l // /
10
0.. //
// /
/
I
/ / I
/
c., 60 / /

z
..... /
/
/

/
/
/

/
/
/

/
/

(/) / / / /
/ / / /
(/) 50 / / / /

< /
/
/
/
/
/
/

0.. / / /
/
....:l 40 / /

<
E-<
/
/
/

//
,
//

0 30 , ,
/,
,
//

E-< ,,/ /
/

20
/
/

/
, /

/
- /
~
/

V ,.
10 / I /
'II

0
100 10060 :,o 18 II . :9/11 ID 1/Z ID S/4 ID I ID

SIEVE SIZES
o Specification Limits
• Test Gradations

Fig. 18. Final Type C Blend Gradations .

 54

100
I I ,,,_
,N I ,,,_
I
110

-...
I
I I
FF 1
TI I
BO
:z: , ,
, ,
I
I

l':a:l 1 ·1 F
t.) , , I I
I

= 70
,, , I

I I
I

-
l':a:l
0..

u
eo I
I
I ,I

I
,7
I I

-
:z:
C7l
C7l
.:
50

,,
T
I

F
I
,
, ,
I
I
I
I
I
,
J
_,
,,

,,
0.. '40 ,, , ,
M /
,,/ ,,/ , ,
, ,, , ,,
...
.:
30 ,
, ,, ,,,
,
...
0

20
,,
II
...
10
0 _ _..____________.._____________,_______________......_ _ _ ___.

200 100 50 30 16 8 4 3/8 in 1/2 in 3/4 in 1 in

SIEVE SIZES
a Specifi c c1 lion Limi ls
• Tesl Grc1da lions

Fig. 19. Final Type IC Blend Gradations.

 55

100
/
/

90
/
~ - /
/
/
/

-
/ /
/ I/

80 / /
E-< / /
z /
/
/
/

~ / /
u 70 / /
a:: /
/
/
~

-
,I /
r:i. 60 ,//
I ~

// /
C, // /

-
z
-
// /
/
50 -, /
Cl.l / /
Cl.l // /

< // /

-
/
r:i. 40 /
/
...:I II
<
E-< 30 I
j
/
/

0 ,, I
/
/

E-< , ~
n ./
20 / ~

ff ~

/
;'
10 II /

0
200 100 50 30 11, e • 3/8 1D, 1/2 lD 3/4 1D. 1 lD.

SIEVE SIZES

D Specification Limits
• Test Gradations
• Coincidence of Test
Gradation and Specification
Limit
Fig. 20: Final Bituminous Base Blend Gradations.
 56

fine and coarse blends for each of the three types of mixtures represent the f inest and

coarsest gradation approved by MHTD during 1990 moved to the finest or coarsest

limit allowed by individual sieve tolerance. The m ixtures in this study may not be

totally realistic field mixes, but the wide spread in gradation was necessary to satisfy

one of the major criteria for this study-the examination of the effect of gradation on

resilient modulus (and hence layer coefficient).

Table 12. Gradations of Six Final Blends .

% Passing
Type C Type 1-C Bit. Base
Sieve Fine Coarse Fine Coarse Fine Coarse
Size
1 in. 100 100 100 100 100 100
3/4 in . 100 100 100 100 -- --
1 /2 in . 100 99 100 99 90 60
3/8 in . 90 75 88 76 -- --
#4 62 42 60 44 65 35
8 40 26 39 28 47 25
16 32 22 29 18 -- --
30 23 18 23 13 35 15
50 15 13 17 9 -- --
100 9 4 10 5 -- --
200 5 2 5 4 9 5

Table 13 shows the contribution to each blend by the different sources of

aggregate, on a sieve by sieve basis . The idea was to try to keep the percentage
 57

makeup on each sieve the same for the fine and coarse gradations. Then, if during

the analysis portion of the study it turned out that a particular sieve was critical to,

say, resilient modulus, the fine and coarse blends could be compared.

Table 13. Contribution of Raw Materials to Each Blend.

Type C, Fine Type C, Coarse

Size %Retained %CA %MS %NS %MF %Lime %Retained %CA %MS %NS %MF %Lime

3/4" 0 0

1/2" 0 1 100 0 0 0 0

3/8 " 10 100 0 0 0 0 24 100 0 0 0 0

#4 28 100 0 0 0 0 33 100 0 0 0 0
8 22 0 83.3 16.7 0 0 16 0 83.3 16.7 0 0

16 8 0 21 79 0 0 4 0 21 79 0 0

30 9 0 21 79 0 0 4 0 21 79 0 0
50 8 0 21 79 0 0 5 0 21 79 0 0
100 6 0 21 79 0 0 9 0 21 79 0 0
200 4 0 10 0 90 0 2 0 10 0 90 0

-200 5 0 33 0 67 0 2 0 33 0 67 0

Type r-c, Fine Type r-c, Coarse

3/ 4 0 0
1/ 2 0 1 100 0 0 0 0

3/ 8 12 100 0 0 0 0 23 100 0 0 0 0
#4 28 100 0 0 0 0 32 100 0 0 0 0
8 21 0 100 0 0 0 16 0 100 0 0 0
16 10 0 63 37 0 0 10 0 63 37 0 0

30 6 0 63 37 0 0 5 0 63 37 0 0
50 6 0 63 37 0 0 4 0 63 37 0 0
100 7 0 63 37 0 0 4 0 63 37 0 0

200 5 0 63 37 0 0 1 0 67 33 0 0

-200 5 0 0 0 20 80 4 0 0 0 25 75
 58

Table 13, cont'd.

Bituminous Base, Fine Bituminous Base, Coarse
Size Retained %CA %MS \NS \MF \Lime \Retained \CA %MS \NS \MF \Lime
l 0 0 0 0 0 0 0 0 0 0 0 0
3/4 - - - - - - - - - - - -
1/2 10 100 0 0 0 0 40 100 0 0 0 0
3/8 - - - - - - - - - - - -
#4 25 100 0 0 0 0 25 100 0 0 0 0
8 18 78 0 22 0 0 10 78 0 22 0 0
16 - - - - - - - - - - - -
30 12 78 0 22 0 0 10 78 0 22 0 0
50 - - - - - - - - - - - -
100 - - - - - - - - - - - -
200 26 78 0 22 0 0 10 78 0 22 0 0
-200 9 100 0 0 0 0 5 100 0 0 0 0

CA = coarse aggregate, DR4, DRS, DR6, DR7

NS = natural sand, DRS
MS = manufactured sand, 9A, 98
MF= mineral filler, DRlO
Lime = hydrated lime, DRll

Table 14 shows the overall contribution of each material to the final blends

compared to the 1990 MHTD average mixtures . As can be seen, when viewing the

percentages of combined coarse aggregate plus manufactured sand, natural sand,

mineral filler, and lime, the proportions of the UMR mixtures closely followed MHTD

field mixes.
 59

Table 14. Amount of Each Material in Final Blends.

Mixture Percent of Each Material in Final Blends

Type
Agency CA MS CA+ NS MF Lime
MS
UMR
Type C Fine 38.0 26.9 64.9 28.2 6.9 0
Coarse 58.0 18.9 76.9 20.0 3.1 0
MHTD 67.2 6.3 73.5 24.5 2.0 --
UMR
Type 1-C Fine 40.0 41.8 81.8 12.2 4.0 1.0
Coarse 56 .0 31 .2 87 .2 8.8 3.0 1.0
MHTD 62 .5 23.9 86.4 12.1 0.2 1.2
UMR
Bituminous Fine 87.6 0 87 .6 12.4 0 0
Base Coarse 93 .5 0 93.5 6.5 0 0
MHTD 92 .2 0 92.2 7.7 0.1 0

Gradation Curve Shape . An analysis was performed to determine the effect of

gradation upon resilient modulus. The most promising methods were later tried in the

development of the MR multiple regression equation. To accomplish this, there was

a need to characterize the gradations so that a single value of gradation "modulus"

would represent the shape and position of the gradation curves. Eight different

methods were tried, and are described in the following paragraphs. The results of the

characterization for each gradation are shown in Table 1 5.

 60

Table 15 . Results of Gradation Curve Characterization .

Blend MR68 FM Cu Cz SF SSF SF /SSF A Area

C-Fine 567,625 4 .29 36.2 35.3 1698 242.8 6 .99 4 .76 7 .2
C- 504,375 5.00 27 .0 40 .7 1803 148.7 9 .07 4 .02 20.3
Coarse
IC-Fine 605,875 4.34 31.2 44.6 1683 257 .9 6.52 4.71 7.2
IC- 508,000 5.07 18.6 27.5 1829 131.8 13.9 3 .97 21 .0
Coarse
BB-Fine 605,125 3.93 48.4 35 .5 1556 303.6 5 . 12 5.16 10.3
BB- 500,875 5.46 62 .5 32.5 1779 165.4 10.8 3.59 16.9
Coarse
R2 * ---- 0 .867 0.005 0.245 0 .801 0.820 0.714 0 .872 0.788
*Note: R2 indicates strength of correlation with resilient modulus at 68 ° F.
MR68 = resilient modulus at 68°F

1) Fineness Modulus

Fineness modulus (FM) is defined as :

(L cumulative % ret 1d on 1. 5 11 , 3/4 11 , 3/8 11 , #4, 8, 16, 30, 50, 100 sie v es )
F/:1=~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ( 3 0)
1 00

The FM is commonly used to characterize concrete aggregate gradations. The

drawback is that the effect of the minus #100 material is not accounted for, which

may be a problem with asphalt mixture gradations.

2) Coefficient of Uniformity and Coefficient of Skew

The coefficient of uniformity (Cul and the coefficient of skew (C 1 ) come from

the geotechnical f ield and are used to help classify soil particle size distributions as to
 61

"well" or "uniformly" graded:

Cu = D 10 • • • • • • • • • • • (31)
DGo

(32)

where: 0 10 = particle size corresponding to 10% passing

0 30 = particle size corresponding to 30% passing

0 60 = particle size corresponding to 60% passing.

-
3) Hudson's A
-
Hudson's A (54) is a parameter that has been used to characterize mixtures

that have appreciable fines in applications such as aggregate stockpile degradation

-
studies (55) . Hudson's A includes the effects of the minus #200 sieve material:

A = (L %passing l. 5 11 , 3/4 11 , 3/8 11 , #4, 8, 16, 30, 50, 100, 20 0 sie ve s ) (33)
100

4) SF, SSF, SF/SSF

In a study concerning the effects of aggregate gradation on slump of concrete,

Joel (56) developed three gradation parameters :

1) SF = l (cumulative percent retained * specific surface) (34)

2) SSF = l (individual percent retained * specific surface) (35)

3) SF/SSF

where: SF = surface fineness

SSF = specific surface factor

 62

sieve series is 1 1 / 2" , 3 /4", 3 / 8", #4, 8, 16, 30 , 50, 100, 200 for SF;

for SSF the #200 sieve is omitted

specific surface = individual particle size (d) surface area divided

by the corresponding volume of each particle

size , assuming spher ical shapes

= 4TTr 2 /[(4/3)TTr 3 ] = 6/d . . . . . . . (36)

In Joel's study, the best predictor of slump was SF/SSF .

5) Position of Gradation Curve

In asphalt work, the position of the gradation curve relative to the position of

the maximum density curve as plotted on the FHWA 0.45 power curve graph paper

has been used to assist in predicting asphalt mixture behavior . In this study it was

decided to use some measure of this relative pos ition to help in the estimation of

resilient modulus.

The method developed in this study was to determine the area between the

two curves. This can be done in a variety of ways . In this study, the use of the

plotting program AUTOCAD (57) was used. Use of a manual planimeter or just

counting squares on graph paper could also have been done.

To find the area between the maximum density line (MDL) and the gradation

curve of interest, the position of the MDL must first be determined. There are at least

four methods reported in the literature that have been presented for determining the

position of the MDL (58-61 ). The method of Goode and Lufsey (58) seems to offer

the most likely way to estimate the largest maximum particle size, and hence the
 63

truest position of the MDL, which flows from the largest particle size . In their

method, a line is drawn from the origin of the 0.45 power paper up through the point

on the gradation curve that represents 90% passing, then on up to the 100% passing

line. If the point that is struck on the 100% passing line is to the left of the point of

100% passing of the actual gradation, the straight line is used as the MDL. If the

point is to the right, another straight line is drawn between the origin and the 100%

passing point of the actual gradation and this line is used as the MDL. These plots

are shown in Fig . 21.

Because the positions of the MDL of each of the six experimental gradations

were all different, it was decided to use a ratio of the area between the curves to the

area beneath the MDL and the 0% passing line, which is a large triangle. The area

calculation results are shown in Table 15. As can be seen, the C and 1-C fine mixes

hewed most closely to the MDL, with the bituminous base fine mix somewhat further

away. The coarse mixes were, predictably, significantly further away, with the C and

1-C coarse mixes exhibiting the greatest relative area under the curve.

These eight moduli were each correlated with the results of the asphalt mixture

resilient modulus testing. The results are shown at the bottom of Table 15 along with

the R2 factor, which is a measure of the strength of the correlation. It appears that
-
A correlates the best with resilient modulus. Consequently, this parameter was tried

in the predictive equation for resilient modulus as discussed later in this study. Fig.

22 aids in visualizing the manner in which A changes with gradation.

Two other methods of quantifying the position of the gradation curves were
 Type C - Fine 100 T "De C - C oa.rse
..,,
100 64
90 l1. 90
j I

f I ,J I
I
80 80
,/ I
'
I '
I
70 70
60
I 'J
I
I
60
./I I

I
I

50
I,'/ I
I
50 .I I

,J I • 7 I

40 . 40 , I

30
j[7 I
I
30 I
I,

I.J
I I

V L.,, I
20 I I
20 • I
lJ J I 1•1/ I
10
I
10 I
I IJ
0 .....
IY
Ill ..
II I
'
I

1/1 l/1 1/4 l

0 I l l II
lH H
ll I 4 1/1 t/1
I
1/4 I

100 100 T
·,r :>e IC - C oarse

90 90 r:
bO 80 80 1 I
I

....c 70 70
/j I
ll'l
Ill
G.1 60
I I
60
'ti'/ I
I

'1-4
~ 50 I
' I
50 I, '
I

11
I .
I

cCl) I I
40 '
I
40 I
,...0
Cl) 30
I I
30 I
I,
I/ I
I
'1-4 I I 20 ' I I
20 ll I
I I
I•

10 10
I ~
! I
0 .....ltt 11
II I
' 1/1 1/1
I
1/4 l
0
I/'
111 I t
100 I t
II I 4 1/1 l/1 1/4 I

Bit Base Fine Bit Base - Coarse

100 100
,•
~
V
90
,Ii
90
l/ I 1V I
I

80
I
80
/, I ~ '/ I

70 70
VI
I
60
J
I
I I

;,
J /
60 I I
V
I

I . I
50
50
l/~ I
,
,/ I
I
40 --,
I 40 ,/
/ I

30 I
I .
I
30 I
'
/
/
I
I I/
20 . 20 1, L,; I

"
I

I I,, I
10 10 I
I
l/ [) I
0 I 0
. . . IO II I 4 1/1 1/1 1/4 I
Ill I t II I 4 1/1 1/1 1/4 I 100 10
Ill II

Fig. 21. Areas Under Maximum Density Lines.

 65

....
-,
100
, ,
... ...
II
. I -
,

-
•• I
90
,
,
~ I
I'
,,
,, ..
+1 A

,, ,, '
• I
i::: 80 I
QJ
c:, '-
.... , ,, . ,,
'QJ"' -· I

,,
-
70 ' - -I I
~

eo
,
,,
-
I-
7"-
,
..
'- I
,,
I
I
,
,,,,, ......
tlO . ,;

.....i:::11.1
I ,
/
I r_
,,, ,,I
.,...
,,,,,, ,
.•
11.1 :10 , :, / - ecreas1n A
' ' ,,
L
aj
,
/ -A
~ ,
-"
llf
.,

.,,
I I
40 .,
,
, ~
' 77
/.I n Hit H
"'·" -,;,., . fiTIP b Hi

+1
0 JO I , //
//.,
,,. A T .. ""., If
- A. ?P.

e-- .. . , ....,-,,-· ,·,,;r

,I I C rrvn .. ;1r
-. fin, 4 71

-
,I_

n'··-
- . .. ITr ,..,,,.- n o <I Q?
20 I
I·
,
, ,
~-
., .,,.., .. ~

• TVnf' 1[ l"O I\ J .!'If' 4 . U.:::

.,,,,
10 I
,I
,. l
·, 1
....
.,,.
,;..
• IHtt 1-l ••e - co •rse :i !, H

r-1
~,
.
0
200 100 0030 1e e 3/8 1/2 3/4

Sieve Sizes

Fig. 22. Hudson's A as a Function of Gradation.

 66

used, although the characterization could not be expressed as a single number, as

could each of the eight methods just discussed. These two methods are presented

below.

6) Individual Sieves. Passing or Retained

The first of these two methods was to simply use the cumulative retained

percent on certain key sieves. Akhter and Witczak (34) have previously found the #4

and 3/4 in sieves to be important, and also included the percent passing the #200

sieve . The use of this approach is explained later in the section dealing with the

predictive equation for asphalt material resilient modulus .

7) Slopes of Gradation Curve

The second method involved the characterization of the gradation line by

breaking the line into several parts and determining the slopes of the portions. The

portions were: #200 to #4 sieve, #4 to 1 / 2 in, and 1 /2 in to 3 /4 in size . The slopes

were calculated as follows using the #200 to #4 portion as an example:

= % passing #4 - % passing #2 OO
(37)
Dt 45
- Diot 5
where: 04 = sieve opening, #4 sieve

0 200 = sieve opening, #200 sieve .

The three slopes for each of the six test gradations are listed in Table 16.

Again, the results of using this approach are outlined in the section of the report that

deals with asphalt mixture resilient modulus prediction .

 67

Table 16. Six Test Gradation Slopes.

Gradation m3 ;4 - 112 m, 12 - 4 m4 - 200

C-fine 0 145.03 143.22

C-coarse 6.85 217.56 100 .50

IC-fine 0 152.67 138.19

IC-coarse 6.85 209.92 100.50

BB-fine 68.49 133.59 140.70

BB-coarse 273.97 95.42 75.38

Particle Shape and Surface Texture

These characteristics were quantified by use of ASTM D 3398 for the ( +) #4

sieve material, and by NAA Method A for the (-) #8 through ( +) #100 material for

each blend. Both are measures of void content of bulk aggregate which is related to

shape/texture. 03398 results in a "Particle Index" (IP);NAA Method A gives an

"Uncompacted Voids Percent" (U). The results are shown in Table 17.

Round, smooth particles give IP's of 6 or 7, while angular, rough particles result

in values of more than 15. The range of IP's of the combined aggregates in this study

was 9 to 12. The Particle Index was determined for the coarse aggregate fraction of

each blend and the Uncompacted Voids content was determined for the fine

aggregate fraction.

Looking at Particle Index values, the DR4 aggregate averaged 12. 7, indicating

that it was the most angular . The other aggregates, in descending order of angularity

were DR5 ( 10.6), DR7 ( 10.1), and DR6 (9.4). As it turned out, Particle Index tests

were performed only on one type of aggregate per test because the ( +) #4 sieve
 68

Table 17. Particle Shape/Surface Texture Results.

Blend Aggregate Particle Aggregates Uncompacted

Index (IP) Voids (%) (U)
C4F DR4 12.4 DR4 + DR9B + DRS 40.1
C4C DR4 12.9 DR4 + DR9B + DRS 40.0
C5F DR5 10.8 DR5 +ORSA+ DRS 39 .8
C5C DR5 10.5 DR5 +ORSA+ DRS 39.5
Average 11.65 39 .85
IC 4F DR4 12 .6 DR4 + DR9B + DRS 43.3
IC 4C DR4 12 .8 DR4+ DR9B + DRS 42.7
IC 5F DR5 10. 7 DR5 +ORSA+ DRS 42 .5
IC 5C DR5 10.6 DR5 +ORSA+ DRS 42 .9
Average 11. 7 42.85

BB 6F DR6 9 .1 DR6+DR8 42 . 5
BB 6C DR6 9.7 DR6+ DRS 43 .0
BB 7F DR7 10.3 DR7 + DRS 43 . 5
BB 7C DR7 -9..JL DR7 + DRS 43.5
Average 9.75 43 . 12
DRS 38.3
ORSA 44 . 1
DR98 44.0
DR4,5 , 6,7 = Coarse aggregate
DRS = natural sand
DR9A,9B = manufactured sand

"blends" only contained coarse aggregate. So, Particle Index values were measures

of individual coarse aggregate source shape/surface texture.

 69

The NAA method was used on the fine aggregate fractions , thus the results

indicate the weighted average of the combined natural sand (DRS) and manufactured

sand (DR9A or DR98) . Additionally, particle shape was determined for each of the

three individual sand sources. The results verify that the manufactured sands were

more anglular than the natural sand (44 vs 38). The IC mixes averaged 42.85, the

bituminous bases 42.5, and the C mixes 39.85. These results were not unexpected

due to the smaller amounts of natural sand in the 1-C and bituminous base blends

compared to the C blends. The 1-C average was slightly more angular than the BB

average, possibly because the DR4 and DR5 aggregates were more angular than the

DR6 and DR7 aggregates as indicated by the Particle Index results .

Looking at the data a little differently, the fine aggregate degrees of angularity

in descending order were 887 (43. 5), IC4 (43 .0), IC5 (42. 7), 886 (42. 7), C4 (40.0)

and the C5 (39. 6).

Specific Gravity

In general, each aggregate was split into three portions (if possible): the(+) #4,

(-) #4 to ( +) #100, and the (-) #100 . Then each portion was blended together to

equate to the final gradation of interest as per the percentage contributed by the

combined fraction (P 1 , P2 , P3 ), where:

P1 = percent contributed by ( +) #4 material

P2 = percent contributed by (-) #4 to ( +) #100 material

P3 = percent contributed by (-) #100 material.

On the ( +) #4 and the (-) #4 to #100 portions, both the bulk and apparent
 70

specific gravities were determined, as well as their average. This average represents

an estimation of effective specific gravity . Also, Rice spec ific gravities were

performed on loose asphalt mixtures, and effective specific gravities were calculated,

as shown in Table 18. For the (-) #100 material, only apparent specific gravities can

be determined . The combined bulk specific gravities (Gsb) for each blend are also

shown.

Table 18. Aggregate Blend Specific Gravities.

Blend P1 P2 P3 BSG 1 ASG 1 G, BSG i ASG l G, ASG J G.o G.. D1

% % % G~
C4F 38 53 9 2 .606 2 . 711 2 .659 2 .622 2 .678 2.650 2 .736 2.626 2 .661 2 .661
C4C 58 38 4 2 .634 2 .712 2.673 2 .632 2.680 2 .656 2.750 2 .637 2 .666 2 .669
C5F 38 53 9 2 .580 2.693 2.636 2 .617 2 .666 2 .642 2.750 2 .614 2 .642 2 .649

C5C 58 38 4 2 .568 2 .682 2 .625 2.611 2 .670 2 .640 2 .750 2 .591 2 .633 2 .636
IC4F 40 50 10 2 .616 2.710 2 .663 2.633 2 .702 2 .667 2 .644 2 .627 2.672 2 .663
IC4C 56 39 5 2 .629 2 .718 2 .673 2 .638 2 .712 2 .675 2 .619 2 .632 2 .669 2 .671
IC5F 40 50 10 2 .577 2 .701 2 .639 2 .608 2 .687 2 .647 2 .657 2 .600 2 .637 2 .645

IC5C 56 39 5 2 .572 2 .687 2 .630 2 .605 2.677 2 .641 2 .632 2 .588 2 .632 2.634

BB6F 35 30 35 2 .651 2 .795 2.773 2 .649 2.773 2 . 711 2 .709 2 .671 2 .753 2 .715

BB6C 65 20 15 2 .667 2 .790 2 .729 2 .661 2 .766 2 .714 2 .750 2 .678 2.726 2 .729

BB7F 35 30 35 2 .608 2 .694 2 .651 2 .576 2 .680 2 .628 2 .709 2 .633 2 .685 2 .664

BB7C 65 20 15 2 .592 2.688 2 .640 2.574 2 .680 2 .627 2 .696 2 .603 2 .670 2 .646

Note: BSG, .2 , ASG, .2 , ASG 3 = test values of blended aggregates

G,, G 2 = average of BSG and ASG
G.b = calculated
G.e = from Rice test
D, = calculated

The calculations for effective specific gravities are shown below:

 71

Estimation method:

100
D ==-------
1 p P2 ( 3 8)
1 + +
G1

where:

01 = effective specific gravity

Percent of(+) #4 mtr 11.

Combined specific gravity of ( +) #4 mtr 11.

Percent of(-) #4 to(+) #100 mtr '1.

Combined specific gravity of(-) #4 to(+) #100 mtr '1.

P3 ::
Percent of (-) #100 mtr 11.
G3 Combined specific gravity of (-) #100 mtr 11.

::
BSG 1 + ASG 1 of ( +) #4 mtr 1 1.
G1
2

BSG2 + ASG2 of (-) #4 to (+)#100 mtr 11.

::
G2
2

G3 :: ASG3 of (-) #100 mtr 11.

 72

Rice Method :

p mm - pb (mi x )
(39)

where :

Gae = Effective specific gravity

Pmm = Percent of entire mix = 100

Pb lmixl = Total percent asphalt by weight of mix

Gmm = Rice maximum theoretical density

Gb = Specific gravity of asphalt.

MAXIMUM THEORETICAL SPECIFIC GRAVITY

Maximum theoretical specific gravities were calculated in two ways . The first

was based on an estimated effective specific gravity (as discussed above) and the

total asphalt content by weight of mix. The second way was to calculate the

maximum theoretical specific gravities based on the effective specific gravity of the

aggregate which is determined by Rice specific gravity testing, as previously

discussed. The calculations are as follows:

Estimation Method

100 + pb ( agg )
D=
pl
+
P2
+
p }
+
pb(agg ) . . . . . . . ( 40)
G1 G2 G} Gb

where: Pblaggl = Total percent asphalt by wt. of aggregate.

 73

Rice Method

( 41)

where: P. = Percent aggregate by wt. of mix .

All other variables were previously defined.

Results of the two methods of both effective specific gravity and maximum

theoretical specific gravity are shown in Table 19 . Comparisons of G.e vs D,, and Gmm

vs Dare shown in Figs . 23 and 24, respectively. Statistical analysis indicates that the

(G.e-D,) and (Gmm-D) values for the C and IC mixes are not significantly different, but

the difference between either one of them and the bituminous base is significant at

the 0.05 level. It appears that for aggregates with low absorptions there is not much

difference between methods of effective specific gravity and maximum theoretical

specific gravity determination. However , for high absorption aggregates, the

"estimated" method tends to underestimate.

VOIDS ANALYSIS

From the results of the specimen bulk specific gravities , the maximum

theoretical specific gravities, and the material percentages and specific gravities, the

air voids, VMA, and voids filled were calculated for the specimens made in the mix

design portion of this study. Results of the voids analysis are shown in Appendix E.
 74

Table 19. Maximum Theoretical Specific Gravities and Effective Specific Gravity by Two
Methods.

M ix Gse o, (Gse-D1) Grnrn D (Grnrn-Dl Weighted

Absorption
88DR6F 2.753 2 .715 0.038 2.584 2 .551 0.033 1.82
88DR6C 2 .726 2.729 -0.003 2.560 2.582 -0 .022 1.97
BBDR7F 2.685 2.664 0.021 2.508 2 491 0.017 1.35
BBDR7C 2.670 2 .646 0 .024 2.531 2.510 0.021 1.41
AVERAGE 0.020 0.012 1.64

CDR4F 2.661 2.661 0 .000 2.467 2.467 0.000 1.09

CDR4C 2.666 2.669 -0 .003 2.485 2.488 -0 .003 0 .93
CDR5F 2.642 2.649 -0 .007 2.489 2.495 -0.006 1 .04
CDR5C 2.633 2.636 -0 .003 2 .464 2.467 -0.003 1.34
AVERAGE -0.003 -0.003 1.100

ICDR4F 2.672 2 .663 0.009 2.515 2.507 0.008 1 . 12

ICDR4C 2.669 2.671 -0.002 2 .477 2.479 -0.002 1 .16
ICDR5F 2.637 2 .645 -0.008 2 .468 2.474 -0 .006 1.42
ICDR5C 2.632 2.634 -0.002 2 .447 2.449 -0 .002 1.40
AVERAGE -0 .001 0.000 1.28

Calculations are as follows:

Estimated Method:

V= - -
D-d * 100, , , • • . . . , . (42)
D

where: V = air voids, %

D = max. theo. sp. gravity

 0 . 030
75

-_0 . 020
0
I
IP
m
t,
--0.010

0 . 000

-0 . 01 0 - - - - - - - - - - - - - - - - - - -
0.5 1 .0 1.5 2.0 2 .5
Absorption (%)
Fig 23. (G 5 E - D 1 ) vs . Absorption.

0 . 040

0 . 030

-
Q
1
E 0 . 020
8
t,
_.. 0.010

0.000

-0.010

-0 .020 .....__ _ _ _ _ _ _ _ _ _ _ _ _ _ __.

0.5 1.0 1.5 2.0 2 .5
Absorption (%)

Fig . 24. (G mm - D) vs. Absorption.

 76

d = compacted mixture specimen bulk sp. gravity

D1 - d1
VMA = *100 . . . . . . . . . (43)
D1

d1 = specific gravity of compacted aggregate

d1 = 100d
(44)
100 + Pb(agg)

VF = VMA-V*
VMA 100 • • • • • • • • • (45)

Rice Method:

Gmrn - Gmb
Pa = * 100 . . . . . . . . . (46)
Gmrn

where: Pa = air voids, %

Gmm = max. thee. sp. gravity

VMA = 100 - (47)

VMA - p
VF = a * 100 • • • • • • . • • (48 )
VMA

Air voids were computed by use of: 1) the calculated effective specific

gravities, and 2) the effective specific gravities derived from the Rice testing.

Comparisons of these are made in Figs. 25 and 26. The use of the estimated method

assumes that the effective specific gravity is midway between the bulk and apparent
 77

10
i::
.....
0
..., 9
- ct1
::,
-
t)
Q

ct1
B

7
>.
.0
I'll 6
"O
.....
0
> 5
·-.....<
~

4
i::
Q) 3
Q
~
c:, Type C
Q)
p.. 2 o Type IC

0
0 1 2 3 4 5 6 7 8 9 10
Percent Air Voids by Rice Method

Fig. 25. Comparison of Air Voids Methods of Determination

for Types C and IC Mixtures.
 78

10
C:
....0
.J
9
.....::,a:, 8
-u CJ
a:,
7
Line of Equality----;;i~~/
/

>,
.0 6
rn
....0
"O
5
>
....
M 4
<
..,.J
C: 3
(IJ Regression Line
CJ
M
(IJ
2
0..
1

0
0 1 2 3 4 5 6 7 8 9 10
Percent Air Voids by Rice Method

Fig. 26. Comparison of Air Voids Methods of Determination

for Bituminous Base Mixtures.
 79

specific gravities. It appears that th is gave an air void value quite close to that given

by the Rice method for the Type C/1-C mix aggregates. However, with the estimation

method, the absorptive bituminous base coarse aggregates predict lower effective

specific gravities and hence lower air voids than really exist. It seems that the more

absorptive base aggregates have an effect on effective specific gravity (G 50 ) . This is

to be expected from the following analysis.

Wagg
(49)
Vagg + Vpore

where: Wagg = dry weight of aggregate

Vagg = volume of water displaced by aggregate

vpore = volume of aggregate pores not occupied by asphalt.

For a given aggregate particle weight, the more absorptive the aggregate is, the

smaller the denominator (smaller Vagg), and therefore, the higher the Gse· The G •• will

be closer to the apparent specific gravity which is still higher. Looking at maximum

theoretical specific gravity "D":

1 00 + pb(agg )
D =
pl Pz P3 pb ( agg )
+ + +
G1 Gz G3 Gb

if, say, aggregate 1 is absorptive, than G, will really be greater than it is estimated .

If a falsely low G, is substituted into the above equation, then D will be

underestimated . The data in Table 19 bear this out. If D is too low, then from

V = D -d * 100
D

air voids will be underpredicted. Again, this is indicated by the data in Fig. 26.
 80

Although the estimation method gave comparable results to the Rice method

for the Types C and IC mixes, it must be remembered that this is for only two sources

of aggregate. It is recommended that the Rice method be the preferred method for

determination of maximum theoretical density of asphalt mixtures. In regard to Rice

testing, consideration should be given to heating the mixture in an oven for several

hours before testing to allow additional absorption, as per recent NCAT

recommendations.

MARSHALL MIX DESIGN

Results of Marshall method unit weight, stability, flow, air voids, VMA, and

voids filled are plotted and shown in Appendix F.

Optimum asphalt contents were chosen primarily on the basis of meeting the

air voids content criteria of the MHTD specifications of 3 to 5 %, with 4% as a target

value. This was tempered by other criteria, such as voids filled, dust/asphalt content

(by weight of aggregate), Marshall stability, percent asphalt, and VMA. The six

mixtures and the MHTD criteria are shown in Table 20. In general, the mixes met all

criteria except for dust/asphalt ratio and VMA. All four Type C mixes were on the

borderline of dust/asphalt content acceptance; VMA values for both Type C and 1-C

were somewhat low.

 81

Table 20. Mixture Design Parameters and MHTD Criteria.

Type C
Parameter MHTD UMR Gradation

Criteria Fine Coarse

natural sand, % 20-35 28 .2 20.0

(-) #200, %Mineral Filler 2: 50 67 67
air voids, % 3-5 3.0,3.0 5.0,5.0
stability, lbs (Nl 2: 750 2300,2950 2600,2000
VMA,% 2: 15 12.0,11 .2 12.0, 12.5
voids filled, % 60-80 69,74 60,61
dust/asphalt, by wt. aggregate 0.6-1.2 1.3, 1.2 0 .5,0.5
flow, 0.01 in (mm) ------ 6,6 6,6
design asphalt content, by wt. of mix ------ 3.6,3.8 3.6,3.6

Type 1-C
natural sand, % < 15 12.2 8 .8
hydrated lime, % 1.0 1.0 1.0
all other (-) #200 = MF ------ yes yes
air voids, % 3-5 3.0,3.3 4 .0,4.0
stability, lbs (Nl 2: 1500 3000 ,3500 2250,2800
VMA,% 2: 15 11.2,11.E 12.1,12 .2
voids filled, % 60-80 72.5,72 68,67 .5
dust/asphalt, by wt. of aggregate 0.6-1.2 1.2, 1.2 0 .9,0.9
flow, 0.01 in (mm) ------ 8,8 8,8
design asphalt content, by wt. of mix ------ 4.0,4.0 4 .0,4.1
 82

Table 20 cont'd .

Bituminous Base
fine aggregate, % =:; 30 12.4 6 .5
fine aggregate, % pass 3/8 in. (cm) 100 100 100
fine aggregate, % pass# 200 s 6 0 0
design asphalt content, %, by wt . of mix 3-6 3.9,3.9 3.5,3.5
air voids, % ------ 4.0,4.0 4.0,3.8
stability, lbs (N) ------ 3800,3850 3400,3500
VMA,% ------ 12.1,12.4 10.1,10.0
voids filled, % ------ 62,67 60,60
dust/asphalt, by wt. of aggregate ------ 2.2,2.0 1.4, 1.3
flow, 0.01 in (mm) ------ 6,7 8,8

As mentioned previously , these mixtures may not have been approved for field

use due to the low VMA, and the asphalt contents may be somewhat low (in an effort

to satisfy air voids criteria). It was decided to not adjust the mixture gradations or

aggregate source (particle shape) ratios because two of the major criteria of the study

were to determine the effect of gradation and particle shape on resilient modulus.

Also, after discussions with MHTD personnel it was felt that the voids would have

been greater had the compression method of compaction been used instead of the 75

blow Marshall compaction effort. In addition, the specimens were compacted with a

manual hammer. Studies at NCAT have shown that this tends to render higher

densities than what is usually achieved with flat-faced mechanical hammers. MHTD

compaction temperatures for the compressive strength method are approximately

35°F lower than for Marshall specimens. This may also contribute to a higher void
 83

content. The gradations were actual approved gradations, but the MHTD mix designs

associated with them reflected the standard MHTD compression method of

compaction .

Note that, besides the above 24 mixtures that were prepared at the design

asphalt contents, another 24 mixtures were prepared which had an asphalt content

0.5% above the design content.

INDIRECT TENSILE STRENGTH

Results

The results of the indirect tensile strength testing are shown in Table 21. The

values shown are averages of two specimens per mix, in accordance with the

recommendations given in NCHRP 332 (62) . The mixtures reflected the following

variables: 12 gradation aggregate/source combinations, two asphalt cement grades,

and various asphalt contents.

The tests were performed in order to determine the seating and testing loads

for the resilient moduli tests, which are based on certain percentages of indirect

tensile strengths. These percentages are also shown in Table 21.

 84

Table 21. Indirect Tensile Strength Data

Mixture Indirect Seating Load (lbs) Testing Load (lbs)

Tensile
Strength 3% 1.5% 0 .5% 30% 15% 5%
41°F 77°F 104°F 41°F 77°F 104°F
C4Fl 0-3 . 75 98 46.5 23 .3 7.8 465 .0 232 .5 77 .5
C4F10-4.25 83 39.4 19 .7 6 .6 393 .8 196.9 65.6
C4F20-3 .75 131 62 .3 31 . 1 10.4 622 .5 311.3 103.8
C4F20-4.25 115 54 .0 27.0 9.0 540.0 270.0 90 .0
C4C10-3 .75 82 40.1 20.1 6.7 401 .3 200 .6 66 .9
C4C10-4.25 81 39.8 19.9 6 .6 397 .5 198.8 66.3
C4C20-3 .75 109 53.6 26.8 8.9 536 .3 268 .1 89 .4

C4C20-4 .25 113 54.8 27 .4 9.1 547 .5 273 .8 91 .3

C5F10-4.0 98 46.9 23 .4 7 .8 468.8 234 .4 78 .1

C5F10-4 .5 84 40 .1 20.1 6.7 401.3 200 .6 66 .9

C5F20-4.0 128 60 .8 30 .4 10.1 607.5 303.8 101 .3

C5F20-4.5 112 53 .3 26 .6 8.9 532.5 266 .3 88 .8

C5C10-3. 75 92 45 .0 22 .5 7 .5 450.0 225 .0 75 .0

C5C10-4 .25 86 41.6 20 .8 6.9 416 .3 208 .1 69 .4

C5C20-3 .75 115 56 .6 28 .3 9.4 566 .3 283 .1 94.4

C5C20-4 .25 109 53 .6 26 .8 8 .9 536.3 268 .1 89 .4

Average 102

IC4F10-4.2 101 48 .4 24 .2 8.1 483.8 241 .9 80 .6

IC4F10-4 . 7 91 43 .5 21.8 7 .3 435 .0 217 .5 72 .5

IC4F20-4.2 134 64 .5 32 .3 10.8 645 .0 322.5 107 .5

IC4F20-4.7 114 54 .4 27 .2 9.1 543.8 271 .9 90 .6

IC4F10-4.2 78 37.9 18.9 6.3 378 .8 189 .4 63.1

IC4Fl 0-4 .7 78 38 .3 19.1 6 .4 382 .5 191 .3 63 .8

IC4C20-4.2 103 51.4 25.7 8.6 513.8 256 .9 85 .6

IC4C20-4.7 99 48.8 24.4 8.1 487 .5 243.8 81 .3

IC5Fl 0-4 .2 107 51.8 25 .9 8.6 517.5 258.8 86.3

 85

Table 21 cont ' d.

IC5F10-4 . 7 96 46 .1 23 .1 7 .7 461 .3 230 .6 76.9

IC5F20-4.2 123 60 .4 30 .2 10.1 603 .8 301 .9 100.6

IC5F20-4.7 119 57 .4 28.7 9.6 573.8 286 .9 95 .6

IC5C10-4 .3 92 45.0 22 .5 7.5 450.0 225 .0 75 .0

IC5C10-4 .8 85 41.6 20.8 6 .9 416 .3 208 .1 69.4

IC5C20-4.3 120 58 .5 29 .3 9.8 585.0 292 .5 97 .5

IC5C20-4 .8 108 52 .4 26.2 8.7 524 .3 262 . 1 87 .4

Average 103

886F10-4 .1 123 57 .8 28 .9 9 .6 577 .5 288 .8 96 .3

886F10-4.6 116 54 .4 27 .2 9 .1 543 .8 271 .9 90.6

886F20-4. 1 143 67 .9 33 .9 1 1 .3 678 .8 339 .4 113 .1

886F20-4.6 138 64 .5 32 .3 10.8 645 .0 322 .5 107 .5

BB6C10-3.6 112 52.5 26.3 8.8 525.0 262 .5 87 .5

BB6C10-4 . 1 94 43.9 21 .9 7.3 438.8 219 .4 73 . 1

BB6C20-3.6 127 59 .6 29 .8 9 .9 596.3 298 .1 99 .4

BB6C20-4 .1 126 58 .9 29 .4 9.8 588 .8 294 .4 98 .1

887F10-4.4 113 53.3 26.6 8.9 532 .5 266 .3 88.8

BB7F10-4 .9 88 41 .6 20 .8 6 .9 416 .3 208 . 1 69 .4

887F20-4.4 143 67 .5 33.8 11 .3 675 .0 337 .5 112 .5

887F20-4 .9 116 54.8 27.4 9 .1 547 .5 273.8 91.3

BB7C10-3 .6 99 46.9 23 .4 7.8 468 .8 234 .4 78 .1

BB7C10-4 .1 87 40 .9 20.4 6 .8 408 .8 204.4 68 .1

BB7C20-3.6 129 61 .1 30 .6 10.2 611 .3 305 .6 101 .9

BB7C20-4.1 118 55 .9 27 .9 9.3 558.8 279 .4 93.1

Average 117

The data indicate the following:

1) an increase in asphalt viscosity (one grade harder) increased tensile strength an

 86

average of 24% in all cases,

2) a decrease in asphalt content from 0.5% above optimum down to Marshall optimum

increased tensile strength by about 10% in 95 % of the cases,

3) an increase in gradation fineness increased tensile strength by about 13% in 90% of

the cases,

4) an increase in particle angularity and surface texture roughness decreased tensile

strength by 7% in 85% of the cases,

5) an increase in VMA decreased tensile strength by 14% in 92 % of the C and IC mixes,

but increased tensile strength by 11 % in 88 % of the bituminous bases.

The data also indicate that bituminous bases had greater tensile strengths than Type IC

mixes by about 14% on the average, which in turn had about equal strengths to the Type

C mixes. A statistical analysis showed that the difference in mean between the bituminous

base mixes and either of the C or IC mixes was significant at the 0.05 level. It appears that

the greater amount of fines and the more well-graded nature of the bituminous bases

contributed to the higher tensile strengths.

These results are in general agreement with trends published in the literature.

Specifically, Hadley et Q.[. (41,42) and Abkemeier (63) report increasing IDT with increasing

asphalt viscosity. In regard to asphalt content, it must be remembered that optimum asphalt

contents tend to be lower for IDT than for Marshall stability (64), thus it is not surprising

that, for a given gradation, lowering the asphalt content resulted in higher IDT values.

Hadley et Q.[ . (41) indicated that an increase in gradation fineness with an accompanying

increase in asphalt content led to higher IDT. In the present study, this was true for the
 87

bituminous base mixes, but for the C and IC mixes, the IDT increased with increasing

gradation fineness with constant asphalt contents. In a review of the literature, Abkemeier

(63) noted that particle shape does not seem to have a great effect upon IDT (which he

verified in his study), although at least two studies (41,42) indicate an increase in IDT with

increasing angularity. In the present study, the opposite was true in most cases. However,

the differences in particle shape were not great. The literature (42,48, 50) indicates that the

relationship of change in IDT due to change in voids is not conclusive - as was shown in the

present study . In general, interactions between variables makes it difficult to make sweeping

statements about any specific variable effect on IDT.

Multiple Regression

Many linear multiple regression equations were developed and analyzed to estimate

the indirect tensile strength from certain mixture data by use of the computer program

SYSTAT (65). Multiple regression equations were fit to the indirect tensile strength data .

Many combinations of variables were analyzed. The criteria for final selection included:

1. The highest adjusted squared multiple correlation (adj -R2 ) that met all the below

listed criteria. This statistic reflects the overall goodness of fit of the equation.

It describes how well the equation would predict a population of data, not the

sample data. Thus it usually is a little lower than the R2 value which is for the

sample data.

2. A !ow standard error of estimate compared to the mean resilient modulus (Y).

3. An analysis of variance F-statistic that indicates that the relationship is

significant at the 0.01 level.

 88

4. Residuals are normally distributed.

5. Residuals have constant variance.

6. All members of the population are described in the same model.

7. No serious problems of variable collinearity.

8. No single observation influences coefficients excessively.

9. Each independent variable contributes significantly to the model.

Data to develop the equations came from Tables 6, 12, 17, and Appendix F. The equation

of best fit had an R2 value of 0.854, an adjusted R2 = 0.846, and a standard estimate of

error (SEE) of 7. 1 51 .

IDT= 139.064 + 21.238T) -7.553Peffv - 0.687AR 4 (50)

+ 1.388U - 2.145IP

where: IDT = indirect tensile strength, psi(N/m 2 )

p effv = percent effective asphalt content by volume

,, = absolute viscosity @ 70°F, poises * 10 6

AR 4 = percent accumulative retained on the #4 sieve

u = percent uncompacted voids content (a measure of particle

angularity/surface texture

IP = index of particle shape

The relationship between the estimated and experimentally determined values of indirect

tensile strength (IDT) is shown in Fig. 27.

 89

·-
.........
D'l
0..
.._..
160
..c::
+J
bD 0
s::(l)
1-,
140
+J
rn
(l)
.....
·-
rn
s:: 120
(l) 0
E-
+J
C)
(l)

·-s:: 100
1-,

'iJ
0 Q)
-
'iJ
(l)
OCID
oco

+J
80 co
·-
C)

'iJ
(l)
1-,
0...
60
60 BO 100 120 140 160
Observed Indirect Tensile Strength (psi)

Fig. 27. Predicted vs. Observed Indirect Tensile Strength.

 90

RESILIENT MODULUS

Test Results

Table 22 presents the results of the resilient modulus testing . The values shown are

averages of four specimens per mix, in accordance with the recommendations given in

NCHRP 332 (62) . The tests were performed at 1 load cycle per second, with the load

applied for 0 . 1 sec in a haversine form, followed by a 0.9 sec resting period . Both horizontal

and vertical deformations were measured and Poisson's ratio calculated. The most current

SHRP protocol at the time of testing (38) specified that if the calculated Poisson's rat io

exceeds 0.50, a value of 0.50 should be assumed for resilient modulus calculations . This

procedure was followed in this study . A previous draft of the SHRP protocol had

recommended assuming values of Poisson's ratio of 0.20 at 41°F (5°C), 0 .35 at 77°F

(25°C) and 0.50 at 104°F (40°C). In the present study, the average Poisson ' s ratio at 41 °F

was 0.30, at 77°F it was 0.49, and at 104°F it was over 0.50 but was held at 0.50 as per

the SHRP protocol. Other investigators (42,66-71 l have reported values of Poisson's ratios

greater that 0.5, especially at higher testing temperatures, with the majority of values

between 0.1 and 0. 7. All of this is not surprising considering that , as asphalt becomes

warmer , it becomes less elastic in nature, thus calculations based on assumptions of

elasticity become less applicable .

 91

Table 22 . Results of Resilient Modulus Testing.

Mix MR41F poi s 41 MR77F pois77

I MR104F poisl04

BB6Fl0-4.l 1332205 0.30 444485 0.50 131557 0.50

BB6Fl0-4.6 1189461 0.37 300950 0.50 88095 0.50
BB6F20-4.l 1619316 0.36 607788 0.48 200355 0.50
BB6F20-4.6 1576178 0.49 469618 0.50 117300 0.50
BB6Cl0-3.6 1001320 0.36 334510 0.50 110773 0.50
BB6Cl0-4.l 864266 0.17 362979 0.50 118695 0.50
BB6C20-3.6 970920 0.32 360081 0.50 99405 0.50
BB6C20-4.l 1116468 0.30 460565 0.49 100156 0.50
BB7Fl0-4.4 1231681 0.42 390336 0.50 111549 0.50
BB7Fl0-4 . 9 1163223 0.47 290056 0.50 65117 0.49
BB7F20-4.4 1321958 0.36 466552 0.50 132760 0.50
BB7F20-4.9 1279411 0.43 412154 0.50 105368 0.50
BB7Cl0-3.6 806709 0.18 371155 0.49 95502 0.50
BB7Cl0-4.l 712903 0.24 266393 0.48 79501 0.50
BB7C20-3.6 990669 0.23 396241 0.50 138993 0.50
BB7C20-4.l 1220512 0.31 440202 0.49 115490 0.50

Bit. Base
Average 1149825 0.33 398379 0.495 113163 0.50
Std. Dev. 253772 0.095 84759 0.007 30391 0.0025

Mix MR41F pois41 MR77F pois77 MR104F poisl04

C4Fl0-3.75 1319480 0.34 334139 0.50 99240 0.50

C4Fl0-4.25 1148340 0.40 326792 0.50 100413 0.50
C4F20-3.75 1199178 0.25 464252 0.50 121361 0.50
C4F20-4.25 1519983 0.39 413749 0.50 110411 0.50
C4Cl0-3 .7 5 785818 0.20 257629 0.50 81916 0.50
C4Cl0-4.25 874245 0.26 264348 0.50 73452 0.50
C4C20-3.75 1028246 0.17 497547 0.47 128856 0.46
C4C20-4.25 1302512 0.34 410904 0.50 70235 0.50
C5Fl0-4.00 1277437 0.35 358212 0.50 104171 , 0.50
C5Fl0-4.50 1191103 0.30 305779 0.50 95686 0.50
C5F20-4.00 1244735 0.29 608460 0.46 221003 0.50
C5F20-4.50 1057375 ' 0.26 443050 0.43 201795 0.50
C5Cl0-3.75 1090774 0.30 378904 0.50 116155 0.50
C5Cl0-4.25 973954 0.33 318067 0.50 85125 0.50
C5C20-3.75 944909 0.20 443346 0.50 137806 0.50
C5C20-4.25 921963 0.20 432644 0.50 134446 0.50

Type C
Average 1117504 0.29 391113 0.49 117629 0.50
Std. Dev. 193551 0.070 92047 0.020 42016 0.010
 92

Mix MR41F pois41 MR77F pois77 MR104F poisl04

I
IC4Fl0-4.20 1042503 0.25 356218 0.44 161044 0.50
IC4Fl0-4.70 1464584 0.49 303952 0.50 89102 0 . 50
IC4F20-4.20 1296217 0.24 481780 0.50 181547 0.50
IC4F20-4.70 1277202 0.29 564574 0.50 147858 0.50
IC4Cl0-4.20 847745 0.34 292395 0.50 78945 0.50
IC4Cl0-4.70 879666 0.21 265494 0.50 59717 0.50
IC4C20-4.20 839858 0.13 429363 0.33 98590 0.33
IC4C20-4.70 890095 0.18 372834 0.48 104021 0.48
IC5Fl0-4. 20 1131758 0.36 359047 0.50 98266 0.50
IC5Fl0-4.70 1407345 0.42 389235 0.50 95642 0.50
IC5F20-4.20 1047901 0.16 637139 0.50 182961 0.50
IC5F20-4.70 1238235 0.26 510477 0.50 117678 0.50
IC5Cl0-4.30 819220 0.20 444177 0.50 102012 0.50
IC5Cl0-4.80 938142 0.33 353523 0.50 68590 0.50
IC5C20-4.30 923546 0.14 552363 0.44 133068 0.49
IC5C20-4.80 1156494 0.40 312598 0.50 56538 0.50

Type IC
Average 1075032 o. 28 414073 0.48 110973 0.49
Std. Dev . 212402 0.106 109000 0.045 40110 0.042
I

Total Avg. 1114120 0.30 401189 0.49 113922 0.50

In general, all mixtures became about an order of magnitude stiffer as temperature

dropped from 104° to 41 °F, as could be expected . From the temperature-modulus

relationship of each mixture, the moduli at 68°F were calculated. The moduli at 68°F

are necessary for entering the predictive nomographs in the 1986 AASHTO Design

Guide for choosing layer coefficients. At this temperature, the 1-C mixes were slightly

stiffer than the bituminous base mixes and the Type C mixes were the least stiff. A

regression equation was developed between resilient modulus at 77°F and indirect

tension at 77°F . The R2 coefficient was 0.536. This relationship is shown in Fig . 28.

Acceptable Range of MR. From the Asphalt-Aggregate Mixture Analysis System

(AAMAS)(37) comes Fig. 29 , a chart for recommended total resilient moduli vs

temperature, using indirect tensile testing conditions as was done in the present

study. As can be seen, none of the UMR mixes were excessively stiff, although some
 93

-
IQ
0
.....
7.0

><
·-
-
Cll
P. 6.0
~
0
r--
r--
@ 5.0
Cll
::,
......
::,
"O
0
~
4.0
~

~
.....
V
......
.....
Cll 3.0
V
~

2.0 .....~ - - i ~ ~.......~~.....~ ~ - - - ~ ~ - - ~ ~ - - ~......

40 60 80 1 00 1 20 1 40 1 60 1 80
Indirect Tensile Strength (psi)

Fig. 28. Resilient Modulus at 77°F vs. Indirect Tensile Strength.

 94

, 02
I
I I
I I
~

---- ..............
"-... ........,,
Mo 1ulms
i
I
00 Hialh
- 1
I

--~ ........... "" ' I

I

'(·...... ___
"'. , 1
,..

. . '~ .... I I I
-. '-
""-
. . ' ., '
I

i
. ". r-,... ."
I

I ~

·. 1!-·,, 1'....,.
- "-
, I
I I

·~
.' I
I I
'~
~ I 'r, "'
Moc Uh.ljS T DO
I
uOW
II "f'k.!
i •
'
i
I
I
.....
' i
I I

I I
I

I
I •
. I

I I
I I

0 20 40 60 80 100 120
Temperature (°F)

Fig. 29. UMR Mixtures on AAMAS Acceptance Chart.

 95

of the mixes were insufficiently stiff at one or more temperatures. Analysis of the mix

characteristics indicated that at 104°F, the large majority of the deficient mixes were

the purposely overasphalted ones, and most were of the low-fines(coarse) gradation

and contained the softer grade of asphalt (AC-10). Particle shape was not a

significant factor.

AASHO Road Test asphaltic material reportedly had a modulus of 450,000 psi

at 68°F. It is interesting to note where the AASHO Road Test material plots on the

chart-barely stiff enough at 68°F. Yet, there is a statement in the Guide that

cautions against mixes stiffer than 450,000; mixes exceeding this value are said to

be prone to thermal and fatigue cracking. Perhaps one way to reconcile these two

seemingly contradictory pieces of advice is to examine the 1Y..Q§ of modulus in play

here. Three types of elastic modulus have been presented in the literature to any

great extent over the last 20 years: dynamic modulus (IE* l ), resilient modulus -

compression (MR,comp), and resilient modulus - indirect tension (MR,idt). The AASHTO

Guide a 1 -chart comes from NCHRP 128 (5). It is not stated upon which modulus the

chart is based, but the references that are given deal with l E* J. However, the 1986

Guide states that either MR,comp or MR,idt should be used to enter the chart. l E * I is
not mentioned. AAMAS recommends using MR , idt' which the industry seems to be

doing. To further complicate the issue, MR itself can be calculated in two ways: by

use of total cyclic deflection in the denominator (MR = a/E) or instantaneous cyclic

deformation (a smaller value). Which should one use? The choice may give

significantly different a 1 values for the same mix.

 96

There is very little in the literature by way of comparisons between the different

moduli. First, Von Quintus et fil. (72) found MR ,idt to be significantly higher than

MR,comp at 104 and 77°F, but equal at 40°F . The data appeared to be in terms of

total MR. Khosla and Ohmer (73) found essentially no difference between the two MR

types over a wide range of temperatures. Second, by inspection of the MR = a/€

relationship, instantaneous MR would have to be larger than total MR. Turning to MR

vs IE* I, only two studies were found: Bonaquist et al. (74) indicated that IE* I was

greater than total MR. Kallas (75) found MR to be higher than IE* I at low

temperatures, lower at 104 ° , and the same at 68 ° F (IE* I was measured at 4 hz).

Von Quintus et fil. (72) estimated IE* I via an earlier version of the Witczak equation

and found that MR , idt was usually higher than IE* 1- In the present study, total MR,idt

is compared to Witczak's l E* l in Fig. 30. As can be seen, MR tends to be somewhat

greater than l E* I -
From all this, it would seem that one should enter the AASHTO Guide a 1 - chart

with MR,idt· If one uses total MR,idt' it should be recognized that the resulting a 1 may

be somewhat high, which would explain the low position of the Road Test mix in Fig.

29. And, if one uses instantaneous MR,idt• the resulting a 1 would be even more likely

to be on the high side.

Effect of Variables. As previously mentioned, resilient modulus results for each mix

were correlated with 8 different characterization methods for gradation curves, as

reported in Table 15. The best correlation was with A. Two other methods, which

did not lend themselves to direct correlation, were evaluated in the multiple regression
 97

-....l7l
0..
20
Data for 41. 77 , 104 °F

-
IO
0
0
0
>< , 5
.....,.
0
0
,:,
I'll

-
"C
::1
::l
0

~
o o?o
'€,0
88
0
0

0 0
0 o-V
~ 10 0
0 Oo
,.J
'-'oo
r:::: 0
........
Q.)
...... 0
(J" Oo
0

0
l7l
Q.)
~

"C
5
Q.)
<9
> <~/6
s... 0
Q.)
l7l
.0
0
0
0 5 10 15 20 25
• 5
Estimated Dynamic Modulus, I El (x 10 psi)

Fig. 30. Observed Resilient Modulus vs. Estimated

Dynamic Modulus for UMR Data.
 98

analysis portion of the study .

Other parameters have been shown to be important to the stiffness of a given

mixture (32-34). These are: temperature, asphalt viscosity at 70°F, effective asphalt

content, some measure of void content, gradation, and possibly particle shape and

texture.

Inspection of the data shown in Figs. 29 and 31 to 35 reveals that resilient

modulus appears to increase with:

1. lower temperature ,

2. greater asphalt cement viscosity,

3. more densely graded mixtures (greater % passing the #4 and #200

sieves), and

4. avoidance of over-asphalting .

Loss of modulus with increasing temperature has been well documented in the

literature (37). Likewise, a loss in modulus with a decrease in asphalt viscosity is also

supported by other researchers (76, 77). Little has been reported in the literature

about the effects of gradation specific to resilient modulus (IDT-method), or changes

in asphalt content (as an independent variable). In this study, an increase in MR was

-
noticed for increases in A, percent passing the #200 sieve, and decreases in percent

accumulative retained on the #4 sieve, all of which result in a finer, more densely

graded gradation.

Coarse and fine aggregate fraction particle shape, as shown in Figs. 36 and 37,

did not seem to exhibit a significant trend in their effect on resilient modulus.
 -....
,0
0
6 .50

99
><
6.00

rz..
0
co
co 5 .50
0

5.00

...•
Bi tuminous Base
....
-
....
ll'l
QI
4.50
a
Type C
Type IC
0::

4 . 00
AC 10 AC 20
AC Grade
Fig. 31. Effect of Asphalt Cement Viscosity on Resilient Modulus .

- 7.00 - - - - - - . . . - - - - - - - - - - - -

-
,0
0

-rz.
0
co
6.00
0

co
_.
Ill
D'l

- ::,
::,
't:I
o 5.00 • Bituminous Base - coarse
::Iii
_. o Bituminous Base - fine
s::: • Type C - coarse
-- QI

ll'l
c.
•
Type C - fine
Type IC - coarse
cu
0:: o Type IC - fine
4 .00
3 4 5 6
A
Fig. 32. Effect of Gradation on Resilient Modulus.
 IO -0
.....
6 . 00
•
A
81tuminous Base
Typ e C
100
c T\·pe IC
><
·-
-~
0
Ill
C.

cc
co
..,
a:,
5 .50
Ill

-==
"O
0
::s
..,
i::

·-·--
cu
ll'l
cu
~

5 .00
p (P optimum + O.S%)
optimum

Fig. 33. Effect of Asphalt Content on Resilient Modulus .

-
IO
0
.....
7.00

><

-
~
0
cc
6.00
co

•
5.00 •
·o Bil Ba s e - Fine
Ty pe IC - Fine

-·-
0
c:, Type C - Fine
•• Bil Base - Coarse
Type IC - Coarse
• Type C - Coarse
4.00
0 10
% Passini #200 Sieve
Fig . 34. Effect of % Passing the #200 Sieve on Resilient Modulus.
 7
O Bil Base - fine 101
D Type IC - fine
>< c, Type C - fine
• Bil Ba se - Coa rse
• Type IC - Coarse
• Type C - Coarse
tz... D
0 6
CD
co

5 •

-·-
4 ..__....._ _.....________..._________..

30 35 40 45 50 55 60 65 70
% Accumu l ative Reta ined on the #4 Si eve
Fig . 35 . Effect of Percent Accumulat ive Re t ained on t he # 4 Si e v e
on Resilient Modulus .

6.00 - - - - - - - . - - - . . - - - - - . - - - - - -...

--
IO
0 D
0 B1lummous Base - DR7

• B1lummous Base - DR6

• Type C - DR4
c, Type C - DRS
><
5 .75 • Type IC - DR 4
D Ty e IC - DRS

~
0
co
co
5 .50

0
5.25
....
s::Cl.)
-
5 . 00
8 10 12 14
Particle Index
Fig . 36. Effect of Index of Particle Shape of Coarse Aggregate
on Resilient Modulus.
 102

- 6.00

-
IC
0 t::. Ty pe C - DR5
• Typ e C - DR4 0

- l1l
~
e
o
•
Type 88 -
Type IC -
Type IC -
DR6
DR5
DR4
•
~ 0 Type 88 - DR?
0
cc
cc

5.50

0
~ •
C:
........
-
G)

l1l
G)
~

5.00
38 39 40 41 42 43 44
Voids in Uncompacted Mxture (U), %

Fig. 37. Effect of Particle Shape of Fine Aggregate

Fraction on · Resilient Modulus.
 103

Adedilmila and Kennedy (67) indicated that there was no significant difference in

resilient moduli of gravel and crushed limestone mixes, although the gravel mixes

generally had slightly higher values.

In this study, compactive effort for all specimens was held constant.

Therefore, any differences in VMA and air voids were a function of differences in

particle shape, gradation, and whether or not the mixture was purposely over

asphalted. Changes in void content were not, then, a function of level of compaction.

Being dependent variables, it was difficult to draw conclusions about their effect on

MR . For instance, air void content was lower in the overasphalted mixes, and so was

MR, but what really caused the MR loss? As Abkemeier (63) has summarized:

Generally, opinion and testing results have been mixed with respect to the effect of air

voids on asphalt mixture properties. Most test results from the static and dynamic indirect

tensile tests (31, 76, 78) indicate that with decreasing air void content of the asphalt mixture,

tensile strength and resilient moduli will correspondingly increase.

Other investigators (42,67, 79) have had less conclusive results. These investigators

have concluded that air voids are not directly related to fatigue life , tensile strength, and

resilient moduli and that possibly an optimum air void content exists for these properties .

Careful evaluation of mix factors would be required when evaluating the effect of air void

content, since air voids are highly dependent upon other factors. Such factors include

compactive effort, asphalt content, a change from open to dense gradation, or any combination

of these.

In the present study, changes in MR due to VMA differences caused by

gradation changes were significant, both in the paired t-test and in the regression

analysis.
 104

Statistical Analys is

An analysis of the data was undertaken using the statistical software package

SYSTAT. First, the means of the five independent variables were checked via Tukey's

HSD test and paired t-tests to compare the means in order to determine which

variables were statistically different from the others . Next, multiple regression was

used to provide a model for the estimation of resilient modulus from the index data .

Tests of Significance. The first question to be answered was, for each of the five

independent variables, was the difference in magnitude from high to low a significant

contr ibutor to changes in MR? For example, for the whole data set , did switching

from AC10 grade to AC20 significantly affect MR? To test this hypothesis, the MR

data were sorted into two groups (AC10 and AC20) and the means of both groups

were calculated . Then the two means were tested via Tukey's HSD to see if they

were significantly different at the 0.05 level. The same sort of analysis was done on

the other independent variables: two levels of asphalt content (optimum and

optimum-plus-0.5%) , three levels of testing temperature, six levels of gradation

-
(Hudson's A values) or two levels of gradation (coarse and f ine), and six levels of

aggregate type (see Table 17). Two other variables that have been shown to be

important are percent passing the #200 sieve and percent accumulative retained on
-
the #4 sieve . These were, in effect, tested when A was tested .

The results are shown in Table 23. The analysis was broken into four parts .

First, as the top portion of Table 23 indicates, results from all 48 mixtures and all

three temperatures were lumped into one data set and examined. The results show
 105

that because temperature is such a dominant variable , only temperature was

sign if icant at the 0.05 level. Of course, MR decreased with increas ing temperature .

The second portion of Table 23 shows the results when temperature was

removed as a variable. The trends are the same as in the upper portion of the table,

that is, greater stiffness came from mixtures that were comprised of higher viscosity

asphalt, finer gradation, asphalt contents near optimum (as opposed to being

overasphalted) , and greater proportions of manufactured sand. Only asphalt viscosity

and aggregate gradation (fine or coarse) were significant at the 0 .05 level , although
-
gradation (six levels of A) was significant at the 0 .089 level.

The bottom portion of Table 23 breaks the analys is down in regard to mixture

type. For the bituminous base and IC mixes, viscosity and gradation (only 2 levels are

possible) were significant at the 0.06 level; for the C mixes, only viscosity was

significant .
 106

Table 23. Significance of Variables to Res ili ent Modulus .

MR (psi)

CONDITION Maximum Minimum Difference Significance

All 3 Temperatures , all mixtures at 0 .05 level

Temperature , 41 ° and 104 ° 1, 114, 100 113,922 1,000.178 yes

Viscosity, AC 20 and AC10 587,919 498,221 89,698 no
Gradation, fine and coarse 609 ,651 476 ,489 133 , 162 no (yes at 0 .07 level)

Gradation, fine to coarse 673 , 109 459 , 125 213,984 no

% Asphalt, opt. and ( +) 0 .5 % 547 ,890 538,251 9,639 no

Agg. blend, 886 and IC4 582 ,394 521 ,888 60,506 no

Mixture type, 88 and IC 553 ,769 533,360 20,409 no

68°F, all mixtures

Viscosity, AC20 and AC10 616,500 480,792 135,708 yes

Gradation, fine and coarse 592,875 504,417 88,458 yes

Gradation, fine to coarse 638 ,750 500,875 137,875 no (yes at 0.089 level)

% Asphalt, opt . and ( +) 0 .5 % 570,292 527,000 43,292 no

Agg . blend, 886 and IC4 587 ,750 523,000 64,750 no

Mixture type, IC and C 556 ,938 536,000 20,938 no

68°F, Bituminous Base

Viscosity, AC20 and AC10 617 ,375 488,625 128,750 yes

Gradation , fine to coarse 605 , 125 500 ,875 104,250 yes

% Asphalt, opt . and (-l 0 .5 % 572 , 125 533 ,875 38 ,250 no

Agg . blend, 886 and B87 576 ,375 529 .625 46,750 no

68°F, Type C

Viscosity, AC20 and AC10 613 ,250 458 ,750 154,500 yes

Gradation, fine to coarse 567 ,625 504 ,375 63 ,250 no

% Asphalt, opt . and ( ... ) 0.5 % 561,125 510 ,875 50,250 no

Agg. blend, C5 and C4 549 ,000 523,000 26 ,000 no

68°F, Type IC

Viscosity, AC20 and AC10 618 ,875 495 ,000 123,875 yes

Gradation , fine to coarse 605 ,875 508,000 97,875 no (yes at 0.061 level)

% Asphalt. opt . and { - ) 0 .5 % 577,625 536,250 41,375 no

Agg . blend, IC5 and IC4 587 ,750 526,125 61,625 no

 107

To examine the effect of coarse aggregate particle shape more closely, paired

t-tests were performed for each mix type at each test temperature. In this way, all

variables except particle shape were kept constant. For an example, refer to Table

24. Here, for C mixtures at 41 °F, each pa\r of MR values were for constant asphalt

grade, asphalt content, temperature, and gradation (nearly) . The mean of the

differences was tested for significance.

Table 24. Example of Paired T-Tests: C Mixes at 41 °F.

Gradation AC % AC Resilient Modulus (psi)

DR 4 DR 5

fine 10 opt. 1,319,480 1,277,437

coarse 10 opt. 785,818 1,090,774

fine 10 (+) 0 .5% 1,1 48,340 1,191,103

coarse 10 (+) 0.5% 874,245 973 ,954

fine 20 opt. 1,199,178 1,244,735

coarse 20 opt. 1,028,246 944,909

fine 20 (+) 0.5% 1,519,983 1,057,379

coarse 20 (+) 0.5% 1,302,515 921,963

The coarse aggregates that were compared were 6 vs 7, and 4 vs 5 . Almost none

of the nine data sets that were examined showed any significant changes in MR due

to coarse aggregate type .

A similar study was undertaken for fine aggregate. Here, IC mixes were

compared to C mixes, holding all other variables constant. In essence, this was a test

to see if increasing the proportion of manufactured sand at the expense of natural

 108

sand made any difference in MR. The results indicated that in most cases there was

no significant difference .

Finally, a paired t -test was performed on VMA results . Each pair of MR values

had a common aggregate blend, asphalt viscosity , asphalt content, and testing

temperature (68 °F). The only var iable different between the two MR values was

gradation, which caused a difference in VMA. The results indicated that VMA did

differ significantly between pairs .

Multiple Regression . Multiple regression equations were fit to the resilient modulus

data. Many combinations of variables were analyzed . The criteria for final selection

were the same as discussed in the "Indirect Tension" section .

The following model most successfully predicted the dependent variable

(resilient modulus) with an R2 = 0 .948, an adjusted R2 = 0.946, a SEE = 0.099, and

an absolute error = 18 .0% . This is comparable to the results of Akhter and Witczak

(R 2 = 0.931 and 0 .934, SEE = 0.122 and 0 . 125). Absolute error of Akhter and

Witczak's equation was not available, but an earlier model based on the same data

gave an absolute error of 20 .6% (33) . Absolute error is calculated as:

Predicted MR-Measured MR
Absolute Error = *100 (51)
Measured MR

Data to develop the equations came from Tables 5, 10, 15 and Appendix F.

For all mixtures :

 109

logMR = 7.137-0.016T-0.005AR 4 +0.088~ 7 o

(52)
-0. 028Peffv-O. 006P200* ( Popteffv-Peffvl )-o. 016Pair.
where:

T = mix test temperature, ° F

f/10 = asphalt absolute viscosity @70°F, poises '* 10 6

P200 = percent material by weight passing the #200 sieve

p effv = percent effective asphalt content by volume

P.;, = percent air voids

AR 4 = percent accumulative retained on the #4 sieve

difference in effective optimum asphalt content and effective

asphalt content. % .

The standardized regression coefficients indicated that the relative importance of

variables were, in descending order: temperature, AR 4 , viscosity, P.ff, P200 '* (P 0 pteffv -

P.ffvl, and P8 ;,· Thus, by knowing relatively easily obtainable mix design data, the

resilient modulus of these mix types can be approximated.

The variables included in the model are temperature, P200 and AR 4 (measures

of gradation), viscosity of asphalt cement at 70°F, P 8

,,, and effective asphalt content

by volume, and the difference between the mix design effective asphalt content

(optimum) and the amount actually in the mix. These were identified through use of

the multiple regression analysis as being the most significant variables in their effect

upon resilient modulus. Many other variables were tried, but did not contribute

significantly to the model, or suffered from multicollinearity. Multicollinearity occurs

 110

when two or more variables are highly correlated. When this happens, the regression

coefficients tend to be unstable, an undesirable situation. Thus, even though several

-
terms involving gradation description (AR 314 , AR 112 or Hudson's A) gave a higher

adjusted R2 to the model, they had to be excluded.

Also, it was postulated that some measure of particle angularity/surface texture

would enhance the model. In fact, inclusion of U (fine aggregate particle shape index)

and IP (coarse aggregate particle shape) did improve the model, but a t-statistic test

indicated that they were not significant contributors . This may be because of the

somewhat limited range of particle shapes involved in the study.

Some measure of void content was originally considered as potentially

important. As it turned out, air voids had a slight edge over VMA.

Other variables that were tried unsuccessfully were percent absorbed asphalt

content, voids filled, dust ratio, aggregate gradation curve slopes (M 4 . 200 , M, ,2 . 4 , M3 14 _
-
112 ), percent retained on each sieve, and SF/SSF. Both Hudson's A and the gradation

curve slopes enhanced equation accuracy, but using individual sieves proved to be

somewhat better .

In regard to each variable's regression coefficients, it should be pointed out that

the relative importance of prediction cannot be gaged from the magnitude of the

coefficient because this is influenced by the scale of the variable. Also, the direction

of the association between the dependent variable (resilient modulus) and any

independent variables cannot be ascertained by the sign of the coefficient. And it

must be emphasized that regression equations are only valid for the ranges of data
 111

studied. Extrapolation should be done with caution, especially in the lower

temperature range where errors (scatter of data) is much higher . This is typical for

any kind of modulus testing. And, if one is tempted to play "what if" by substitution

of hypothetical mixture data into the equation , it must be remembered that certain

variables tend to be dependent upon each other in the real world (e.g., - #200 sieve

material and #4 material), so any substitution of values must be done in a realistic

manner .

It can be seen that Eq. 52 is similar to Akhter and Witczak's Eq. 26 which

contains the variables temperature, (} 70 , P 200 , Peffv• Pa;,, AR 4 , and (Peffv - P 0 p 1 e11 vl as well

as AR 3 14 , f, and Pabsw· In the present study, AR 3 , 4 was excluded because of

multicollinearity problems, testing frequency (f) was not a variable in this study, and

Pabsw was shown to be an insignificant contributor . Many interactions, such as

(A* (} 70 ), proved to be collinear with other parameters.

The relationship between estimated and experimental values of resilient

modulus is shown in Fig . 38. The high R2 value demonstrates a strong correlation.

Applications

Because layer coefficients are directly affected by resilient modulus, the

practical impact of the above trends is that higher layer coefficients can be obtained

by:

1. using a harder grade of asphalt (higher 'ho),

2. using a more well-graded gradation (higher A, adjusting P200 and AR 4 ), and

3. avoiding overasphalting (optimum Pe 11 ) .

 112

10.00
~
,.,
)'
/,.,

-
,E
fl
1 .00
.~ "
::,
"'O
0 ~ )

2 04 ~.©
+1
,.Jd ~ 0
....=
-....
Cl)

fl
0.10 r ~h ~
" -
b
l
Cl) ·,
~
"'O
.,, .,
Cl) ,/
+1 V
....CJ
"'O V
V
Cl)
1-4
0..
0.01
0.01 0.10 1.00 10.00
8
Observed Resilient Modulus (psi x 10 )

Fig. 38. Estimated vs . Experimental Resilient Moduli

for UMR - Study Data.
 113

Of course, caution should be exercised when considering these options.

Excessively hard asphalts, highly angular aggregates, and low VMA values can lead

to such problems as thermal cracking, lack of workability, poor durability, and rutting.

Thus, caution should be used when interpreting the results of the above analysis.

Also, it must be kept in mind that the MR predictive equations are based on data that

represent well-graded gradations. They should not be applied to mixes with

significantly different characteristics or materials, such as stone mastic mixtures.

GENERAL RESILIENT MODULUS REGRESSION EQUATION

A multiple regression equation was developed to estimate resilient modulus with

a broader data base by adding data from the literature into the data base developed

experimentally in the present study. Based on the work done on the UMR data, it was

decided that, as a starting point, the equation ought to reflect the following variables:

temperature, consistency of asphalt! gradation, loading time or frequency, and voids.

Not all variables used in the previously developed UMR equation were readily available

in the industry-wide data base. For instance, it is not common to measure particle

shape/surface texture, so it was necessary to omit this variable from the model. The

industry-wide data base included 84 sets of data (72, 74, 76, 78,80). The combined

sets for the general model numbered 229. Many regression models were determined.

The final regression equation was:

logMR = 6.871 - 0.017 T- 0.024 Pair+0.043117 0

(53)
+ 0.018P200 - 0.004AR 3 ; 4 -0.011Pettv

where:
 114

MR = resilient modulus, psi

T = mixture temperature, °F

pair = percent air voids

''7o = asphalt absolute viscosity at 70°F, poise • 106

P200 = percent passing to #200 sieve

AR314 = accumulative percent retained on the 3/4 in sieve

peffv = percent effective asphalt content by volume

The standardized regression coefficients indicated that the relative importance

of variables were, in descending order : temperature, Pa;,, viscosity, P20 0 , AR 314, and

P.ttv· Again, the model was based on the highest adjusted R2 compatible with the

model criteria previously discussed. The model resulted in R2 = 0.846, adjusted R2

= 0 .842, SEE = 0 . 164, and an absolute error of 32 .9%. The model is less accurate

than the model developed from laboratory data developed in the present study, which

had an absolute error = 18.0%. This is not surprising , because the general data base

reflected five different testing programs which included different operators,

equipment, and, most likely, test methodology. And, to put things in perspective, it

should be remembered that the Shell method for Sm determination is based on Van der

Poel's work, which was claimed to have an accuracy within a factor of ± 2; it would

appear that the model in Eq . 53 is considerably more accurate than this.

The method of cross-validation was used to determine the variables which were

significant . The whole data set was split into two parts in a random fashion. A

regression model was developed for the first portion as per the above criteria. Once
 115

the variables were chosen, the second data set was applied to the model. The SEE

of each set were compared, and were shown to be close. Finally, the data sets were

recombined. A final model was calculated, using the same variables as were

previously determined for the f irst half of the data set.

The ranges of variables that the industry-wide portion of the data represented

are shown in Table 25. The industry-wide portion of the data set is included in

Appendix G.

Table 25. Ranges of Variables in Industry-Wide Data Base .

Variable Range
Temperature, OF 24 - 104
Viscosity at 70°F, poise x 10 6 0.17 - 9.5

Asphalt effective volume, % 4 . 1-14.5

Accumulative retained on 3 /4 in. 0-0
sieve, %

Accumulative retained on 1 /2 in. 1 - 29

sieve, %

Accumulative retained on #4 sieve, % 37 - 60

Passing #200 sieve, % 4.0 - 8.3
Air voids, % 0.2 - 15.4

Resilient modulus, psi X 10 6 0.032 - 4.3

Range of loading time was very small (0.1 to 0.15 sec) with most of the data

at 0 .1 sec. Because the industry is standardizing at 0.1 sec., it was decided to

remove loading time as a variable. Thus, the data set was limited to a loading time

of 0.1 sec. The fact that the equation does not reflect loading time or frequency
 116

effects is, although unavoidable, unfortunate because the repeated modulus of asphalt

mixtures is to a certain degree a function of loading time . However, the standard time

of 0.1 sec is conservative because this is 3 to 10 times longer than dwell times under

actual traffic conditions, and thus should render a lower modulus.

Not all data were present in the desired form, and had to be converted from the

existing form. Witczak (32) faced the same dilemma when he developed his model,

so he created several equations to supply the missing data. Where necessary,

equations developed by Witczak were utilized:

1'ho.poisexio' = 29,508.2 (penetration, 7 .F)- 2 · 1939 ••• (54)

peffv = (55)
0.483

where:

Peffv = percent effective asphalt content, by volume

Pb = percent total asphalt content, by weight of mix

A plot was made of estimated resilient moduli vs experimentally-determined

moduli for the general model and is shown in Fig. 39.

Even though the general model is less accurate than the UMR model, the data

on which Eq. 53 is based contains variables that exhibit a wider range of values.

Thus, it would seem that the general equation would be more generally applicable.

Thus, this is the model that is recommended for further use.

Data from 236 MHTD mixtures approved in 1990 were substituted into Eq. 53

to estimate resilient modulus. The results are shown in Table 26.

 117

re,- 10.00
0

.....fll
- i:i.

1.00

~
A

-
.....
Q)

.....
fll
Q> 0.10
~
"C
Q) o UMR De.ta
~

.....0 · c::. Industry-Wide Data

"C
Q)
J.4
p...
0.01
0 . 01 0.10 1.00 10.00
6
Observed Resilient Modulus (psi x 10 )

Fig. 39. Relationship of Estimated vs. Experimentally Derived

Resilient Modulus Data from General Data Base.
 118

Table 26 . Res ilient Modulus of MHTD Approved 1990 Mixes .

MHTD
Mix Min. Max . Mean
Type C 360,415 439,198 403 ,819
Type 1-C 370,998 425,394 404,973
Bit . Base 377,642 581,453 460,000

In regard to each variable's regression coefficients, it should be pointed out that the

relat ive importance of prediction cannot be gaged from the magnitude of the

coefficient, because this is influenced by the scale of the variable . Also, the direction

of the association between the dependent variable (res ilient modulus) and any

independent variables cannot necessarily be ascertained by the sign of the coefficient.

And it must be emphasized that regression equations are only valid for the ranges of

data studied. Extrapolation shou ld be done with caution , especially in the lower

temperature range where errors (scatter of data) are much higher . This is typical for

any kind of asphalt modulus testing . And, if one is tempted to play "what if" by

subst itution of hypothetical mixture data into the equation , it must be remembered

that certain variables tend to be dependent upon each other in the real world ( ~,

#200 sieve material and VMA), so any substitution of values must be done in a

realist ic manner.

PAVEMENT TEMPERATURE

The modulus of an asphalt mixture is significantly affected by temperature.

Values of temperature are necessary for the estimation of dynamic modulus (Akhter
 119

and Witczak), mixture stiffness (Shell), and resilient modulus (as developed in the

present study). Because layer coefficients are applied to field conditions, it becomes

necessary to determine or estimate pavement field temperatures .

In this study, representative temperatures of the asphalt layers were estimated

by use of Witczak's method (24) as prev iously discussed. Use of Witczak's equation

(Eq. 16) requires input of mean air temperature and one representative depth in a

given layer.

Climatological data from 104 reporting stations in Missouri were analyzed in

order to calculate a mean monthly air temperature of 55.06 °F. This data is tabulated

in the companion study of this report (2).

Pavement temperature is also a function of its thickness. Pavement thickness

data supplied by the MHTD were analyzed for mean thicknesses of asphalt surface,

binder, and base layers . Because surface and binder layers were considered as one

layer at the AASHO Road Test in computation of layer coefficients, the MHTD surface

and binder layers also were combined for temperature analysis . The Missouri data

indicated that for full-depth and asphalt-over-unbound base structures, the mean

surface-plus-binder thickness was 3.26 in and the mean bituminous base thickness

was 5.07 in. Typically, the temperature at a depth of one third the layer thickness

is taken as representative of the mean temperature of the entire layer. This can be

shown to be close to the mean pavement temperature by use of the mean value

theorem (25). Use of these points in the mean pavement structure in Eq. 16 yielded

representative MHTD pavement temperatures of 65.2 °F (18.4 °C) and 63.4 °F (17.4
 120

~C) for the surface-plus-binder and bituminous base layers, respectively .

In a similar manner, from AASHO Road Test data, pavement temperatures were

predicted for the surface/binder layer and the bituminous base as shown in Table 27 .

There is some question about the average pavement temperature at the Road

Test. In NCHRP 128 it is stated that this temperature was 67. 5 °F and in the 1986

AASHTO Design Guide, the flexible pavement modulus-layer coefficient nomograph

is standardized at 68 °F. It is stated in the Road Test Specia l Report 5 that pavement

temperatures were recorded. Carree and White ( 12) state that in a search of the

special reports and computer data from the Road test, no such information could be

found . Personal commun ication with White revealed that some records have been lost

in a fire . A search by the authors has also been fruitless. NCHRP 128 states that the

standardized modulus of the surface layer was 450,000 psi. In NCHRP 291 , data

from Road Test asphalt mixture modulus testing are presented . An analysis of these

data indicates that the average pavement temperature at the Road Test would have

been 63°F in order to give a dynamic modulus of 450,000 psi. In the classic study

of asphalt pavement temperatures at College Park, Maryland, Kallas (81) found that

the average pavement temperature at all depths was about 63 to 64°F for an average

air temperature of 54°F , which is about a 10°F difference. In a study of actual

pavement temperatures in a similar climate, Croney (82) found that average pavement

temperatures were about 7°F above the average air temperature of 49°F . According

to data in Appendix C of Special Report 5, the average air temperature during the

Road Test was about 50°F. Using this temperature, the Witczak equation (Eq. 16)
 121

predicts a pavement temperature of 58 to 59°F, which is about 8 to 9°F above the

average air temperature. Thus, it would seem reasonable that average pavement

temperatures would be 7 to 10°F above average air temperatures, or about 57 to

60 °, at the Road Test. This is significantly lower than the 68 ° F stated in NCH RP

128, and is close to the 58.1 and 59.1 °F predicted by the Witczak equation. It also

then seems reasonable to use the Witzcak equation to predict Missouri pavement

temperatures. However, the analysis of pavement moduli hinges on the predicted

temperatures of both MHTD and Road Test pavements . The use of the AASHTO layer

coefficient chart, which calls for modul i input at 68°F, is thus brought into question.

Use of this information will be discussed in the layer coefficient portion of the study .

LOAD DURATION AND FREQUENCY

Duration of load pulse (or frequency of application) has been shown to have an

effect upon modulus or stiffness of asphalt mixtures (15,31 ). Either pulse duration

or frequency is used as an input into the Shell method of stiffness estimation, and

frequency is used as an input in Akhter and Witczak's dynamic modulus estimation .

At any given point in an asphalt layer, both pulse duration and frequency are

dependent upon vehicle speed and depth of interest. From data supplied by the

Planning Division of MHTD, the mean vehicle speed of all highways in the Missouri

state system is 56.3 mph. Barksdale (_53) has developed a relationship between

vehicle speed, depth, and pulse duration, as shown in Fig . 40. Using the mid-depth

of both the mean surface and bituminous layers (as determined above), and a vehicle

speed of 56.3 mph, load pulse durations of 0.020 and 0 .032 sec were determined for
 122

,o 0
j - I --- -f-·- -----
- - - ,-
5111,jCl( OR OUAl WH(ll lOAOINC - --·- -----
-
~ ! ~ ~ £- V
-!lO:•~-t-- ~ --~~-~--=:,...~~---___. - ~ . ~ ~ --~

-- _L---
,,______ --

ID
~ - - -~ - ~ _-·_-__
·- ~ - t

------ ------i-----i-----t-----·+-----+------i

0.1
v -- - -
-~ -
~i:::-=-- -
- "·"·'"
T.,s;;

rAINCIPAl S1NU$0l0Al
-
ruLSE TIM(
-
rRINC1'Al T~IANGUlAR
,uu£ TIM£
o a,
12 u 20 2, 21

O("TH 1£H£ATH rAV(MENT SURFACE flNCIIES)

Fig. 40. Barksdale's Load Duration Chart.

 123

the surface and base layers, respectively .

To convert load pulse duration (t) to load frequency (f), the relationship from

Van der Peel (15) was used:

1
f = ( 56)
27t t

Thus, at the midpoints of the surface and base layers, the mean load frequency

for MHTD pavements is taken as 8 and 5 cycles per second, respectively.

In a similar manner, from AASHO Road Test data, times of loading and

frequencies were predicted for the surface/binder layer and bituminous base. The

results of this analysis are shown in Table 27.

Table 27. Pavement Temperatures, Load Durations , and Frequencies .

Layer Veh. D t f Tp
Speed
mph in sec cps OF (OC)

MHTD
surface/binder 56.3 3.26 0.020 8 65.2(18.4)
base 56.3 5.07 0 .032 5 63 .4(17.4)

AASHO
surface/binder 35 4.0 0.035 4.5 58.5(14.7)
base 35 8.83 0.068 2 .3 57.1(14 .0)

D = layer thickness
These values were substituted into Eqs. 6,26, 52, and 53 to estimate mixture

stiffness and dynamic modulus, respectively. The results are discussed in the

following sections .
 124

ESTIMATION OF MIXTURE STIFFNESS

UMR MIXTURE STIFFNESS (Sm)

Mix stiffness via the Shell method for all 48 mixes used in this study were

estimated by use of Eq. 6. For each mixture, values of asphalt content , aggregate

content, asphalt cement penetration, time of loading (0.1 sec) and test temperature

(41, 77,104 °F) were substituted into Eqs. 3,4, 13, 15, and 16 in order to utilize Eq .6.

The resulting values of Sm were correlated with resilient modulus test data . The

results are shown in Fig . 41.

Finally, using AASHO Road Test mixture data, Sm values for the surface/binder

and bituminous base layers were calculated and are shown in Table 28, using Road

Test temperature and loading time data .

Table 28 . Estimation of AASHO Road Test Mixture Stiffness Under Road Test
Conditions .

Layer Pen vb Va t T Sm
% % sec oc ps i
surface 91 12.67 83.74 0.032 15.0 687,320
binder 91 10.64 84.56 0.038 14.5 778,168
surface/binder - - - - - 740,194
base 91 11.36 82.40 0.068 14.0 565,276
Note : AASHO Road Test average surface = 1.67 in, binder = 2.32 in

Note that the Shell method considers six variables: three that address the

viscoelastic properties of aged asphalt cement, and three that address the effects of

mix properties. The gradation and particle shape/texture of the aggregate are ignored.

Mixture stiffness increases with:

 125

20

-....
rn
0.
Data for 41.77 . 104 °F

IC 0
0
~ 15
--
0
0
>< 0

rn 0
::J
::J
"C 0
~
~ 10 Oo
0
0 0
+J
c::
........rn
-
Q,)
0

Q,)
~ 5
"C
Q,)
>
i..,
Q,)
Ill
..c
0 0
0 10 20
5
Mixture Stiffness, s m (psi X 10 )

Fig. 41. Observed Resilient Modulus vs. Shell Mixture

Stiffness for UMR Data.
 126

lower temperature

lower penetration asphalt

shorter load duration

lower asphalt content

higher aggregate content

lower air void content

The resilient modulus testing program indicated that, on the average, the 1-C

mixes were slightly stiffer than the bituminous base mixes, and both were stiffer than

the Type C mixes, which was the result of gradation and sand shape/texture

differences. Because these variables are not considered in the Shell method, and

because temperature, time, and asphalt grade were held constant, the relative

rankings by the Shell method were the result of only relative material proportion

differences. Because of the slightly higher aggregate and lower asphalt proportions,

the predicted Shell stiffness of the C mixes averaged higher than the bituminous base

mixes, followed by the 1-C mixes. Table 29 shows the UMR, MHTD, and AASHO Sm

data calculated at 68 °F and 0.1 sec load time for comparison.

 127

Table 29. Comparison of Shell Mixture Stiffness for AASHO, UMR , and MHTD Mixes.

Mixture Stiffness (Sm),psi

Layer AASHO UMR MHTD

Surface 315,980 - -
Type C - 576,671 488 , 629
Type 1-C - 523 ,054 459,597
Binder 374,836 - -

Bituminous 294,788 542,047 526,656

Base

Note: 68 °F (20 °C), 0.1 sec load duration

MHTD MIXTURE MEAN STIFFNESS (Sm)

In a similar manner to UMR mixture stiffness estimations of the mixtures used

in this study, MHTD mix designs approved in 1990 were estimated for mixture

stiffness. Time of loading and pavement temperature were taken as 0.020 sec and

17.4°C for Type C and 1-C mixes, and 0.032 sec and 18.4°C for bituminous base

mixtures, respectively. The mean Sm values for Types C, 1-C, and bituminous bases

were 823307 , 771293, and 852051 psi , respectively . Note that the values in Table

29 are for higher temperatures .

ESTIMATION OF DYNAMIC MODULUS

UMR MIXTURE DYNAMIC MODULUS ( i E * i)

The dynamic modulus of each of the 48 mixtures used in this study was

estimated by use of the Akhter and Witczak equation (Eq. 25). For each mixture,

values of air content , accumulative percent retained on the 3/4 in sieve, asphalt
 128

viscosity at 70 °F, test temperature (41, 77, 104 °F), load frequency (1 .6 H2 ),

effective asphalt content by volume , mix design optimum effective asphalt content

by volume, accumulative percent aggregate retained on the #4 sieve, percent passing

the #200 sieve, and percent absorbed asphalt by weight of mix were substituted into

the equation .

The resulting values of dynamic modulus were correlated with resilient modulus

test data, as previously shown in Fig. 30 .

In a similar manner, using AASHO Road Test data, dynamic modulus values for

the surface/binder and bituminous base layers were calculated by use of Eq . 26 and

are shown in Table 30.

Table 30 . Estimation of AASHO Road Test Dynamic Modulus Under Road Test Conditions.

Layer P.;, AR3,4 n10 pelf pabs AR 4 P200 f (H 2 ) T E

(%) (%) (poise x (%) (%) (%) (%) (OF) (psi)
106 )

surface 3 .6 0 1.49 10.83 0.8 37 5.9 5 59.1 663,218

binder 4.8 4 1 .49 7.81 1.2 64 4.3 4 .2 58.1 681,678

base 6.2 4 1 .49 10.77 1.2 28 5.6 2.3 57 .1 627,656

Note: P.;, = after traffic

n70 = 29,508.2 (pen77r 2·1939
P. 11 = Pb/0 .483
P•bs = (Pb/0.434) - (P/0.483)

It was reported in NCHRP 291 that the Asphalt Institute performed dynamic

modulus tests on actual Road Test materials. An analysis of the data was performed

and dynamic moduli were determined for both the surface and binder mixtures, which
 129

were 556,863 and 524,407 psi, respectively. A comparison of these laboratory test

values to the estimated values in Table 30 results in absolute relative errors of 19.1

and 19. 7%, which is comparable to Witczak's stated 20.6% absolute relative error.

Thus it appears that the method of estimation of the mixture's dynamic moduli is

acceptable.

Note that the Witczak equations not only consider the viscoelastic nature of the

asphalt and mix proportions, but also aggregate gradation. Thus the UMR 1-C and

bituminous base mixes are more correctly rated stiffer than the C mixes - an

improvement over the Shell method. Dust ratio (percent by weight passing #200

sieve/percent effective asphalt content) is indirectly considered. Particle

shape/texture is not considered. Dynamic modulus increases with:

lower temperature

greater asphalt viscosity

greater loading frequency

lower asphalt content

greater asphalt absorption

greater percent retained on 3/4 in sieve

greater percent retained on #200 sieve

lower percent retained on #4 sieve

smaller difference between asphalt content and optimum mix design

asphalt content (less overasphalting)

lower air voids content

 130

A comparison is made between AASHO, UMR, and MHTD mixes in Table 31,

at the same conditions of temperature and frequency.

Table 31. Comparison of Dynamic Modulus for AASHO, UMR and MHTD Mixtures.

Layer Dynamic Modulus (IE* I)

AASHO UMR MHTD
Surface 366,703 - -
Type C - 352,059 320,640
Type 1-C - 361,790 314,556
Binder 377,704 - -
Bituminous Base 381,097 570,158 484,789
Note: 68°F (20°C), 1 . 6 hz frequency

The UMR bituminous base mixes are rated stiffer than the 1-C (and C) mixes due

to greater ( +) 3/4 in sieve and minus #200 sieve size materials. The effect of

gradation is still only partially accounted for if one considers the displacement of the

gradation curve from the 0.45 power maximum density curve as an important criteria.

And, again, the superior sand particle shape of IC aggregate is ignored.

The MHTD 1990 mixes were ranked from high to low: bituminous base, Type

C, and then Type IC. Bituminous bases were predicted stiffer because of lower air

contents, greater amount of ( +) 3/4 in material, lower effective asphalt contents, and

greater amount of minus #200 sieve material. Type C mixes were rated stiffer than

1-C because of somewhat lower air void contents.

In comparing UMR mixes to MHTD mixes, UMR mixes were stiffer for all three

mixture types due to lower asphalt contents, lower air void contents, and, in the case
 131

of the bituminous base mixes, somewhat higher ( +) 3/4 in material contents.

Looking at the AASHO Road Test mixes, in descending order of stiffness, the

rankings were bituminous base, binder, and then surface, although the spread of

values is not great. The difference in predicted values is primarily due to gradational

differences.

The AASHO surface and binder moduli are predicted to be greater than the

UMR (and MHTD) mixes primarily due to higher percent retained on the #4 sieve and

lower(-) #200 sieve material (and greater air void content in the case of MHTD data).

However, the AASHO bituminous base was less stiff than both the UMR and MHTD

averages because of a greater percent ( +) 3/4 in. material, lower air voids, greater (-)

#200 sieve material, and lower effective asphalt contents. These relationships of

UMR and MHTD to AASHO modulus values become important when estimating layer

coefficients, as will be shown later.

MHTD MIXTURE DYNAMIC MODULUS (IE* I)

In a similar manner to UMR mixture dynamic modulus, data from MHTD mix

designs approved in 1990 were used for estimation of each design's dynamic modulus

by use of Eq. 26. Frequency and pavement temperature were taken as 4.5 hz and

17.4°C for surface mixtures Type C and 1-C, and 2.34 hz and 18.4°C for bituminous

base mixtures, respectively. The mean dynamic modulus values for Type C,IC, and

bituminous base mixtures were 518,864; 509,020; and 600,821 psi, respectively.

DETERMINATION OF LAYER COEFFICIENTS

Five different methods were used for determination of layer coefficients. These
 132

are presented below.

AASHTO NOMOGRAPHS

Use of the layer coefficient nomographs (Figs. 42 and 43) available in the 1986

AASHTO Guide require that asphalt resilient modulus values be at a temperature of

68°F . It is assumed that the Road Test conditions of loading time or frequency are

also in effect. The modulus values at 68°F for the 48 mixtures used in this study

were estimated from the temperature-modulus relationship developed for each

mixture. An example is shown in Fig. 44. From inspection of the AASHTO

nomographs, equations were developed to represent the curves for a 1 (asphalt mixture

surface layer) and a2 (bituminous base):

a 1 =0.391071logE-1.77224 . • • • . . (57)

a 2 = 0.324553 log E - 1.49975 . . . • • . (58)

Each UMR resilient modulus at 68°F and at the test loading time of 0 . 1 sec for

the Types C, 1-C, and bituminous base were substituted into Eqs. 57 and 58 to obtain

layer coefficients. The results are shown in Table 32 in the column marked "AASHTO

Chart-UMR".

Secondly, the UMR-study mix data (48 mixes) were used in Eq . 53 (the General

Model) to estimate MR (68°F and 0.1 sec) . Then, these MR values were used in the

AASHTO nomograph to obtain layer coefficients. The results are also shown in Table

32.
 133

05 I I I I I

-
~- - .... -

04 ./""

.E ~
-
C1)

-
/
V)
:...
:i
"' u
0
C
4) 4)

u
u 0.3
"':,

/
C1)

-
0 U'l
u
ai
C1)

C1)
-
> u
"' C0
...J
0 .2
"':::, u=
u:::, .i::."'
~

U'l
C.
"'
~
- -
0 .1

..... -
0.0 I I t I I

0 100,000 200,000 300,000 400,000 500.000

Elastic Modulus, EAC (psi}. of

Asphalt Concrete (at 68°F)

Fig. 42. AASHTO a 1 Layer Coefficient Nomograph.

 134

1800 4 0

0 JO 1600

1400 J .O

2.5
1200

1000
2.0

800
0.20
1.5
c 600 ·.;;
Q) ..ci Q.
u
::: 400 "'0
Q)
>-
0
u :0
a, "'
IJl :,
:, "C
0
u 200 ro
.r.
1.0
~
-
(f)
"'
<"O

0 . 10 -~---------

{1 l Scale derived by correlati9n obtained from Illinois .

( 21 Scale derived on NCHRP project (3J.

Fig. 43. AASHTO a 2 Layer Coefficient Nomograph.

 135

,......,

-
IC
0
1.0

><
......
Cl) 0.8
0..
'-"
Cl)

-
"O
~
~ 0.6
0
~
...>
c:: 0.4

-
CV
......
......
Cl)
CV
0:: 0.2

0.0 ............_._.............._._........................................................................................................
40 50 60 70 80 90 100 11 0
Temperature (°F)

Fig. 44. Typical Temperature - Resilient Modulus Relationship.

 136

EQUIVALENT STIFFNESS

Layer coefficients were also calculated by use of the Odemark equation for

equ ivalent stiffness:

stiffness ] 1 13
a = a
AASH
1 stiffness, AASHO

where layer stiffness in this study was represented by mixture stiffness (Sm), dynamic

modulus (IE• I), and estimated resilient modulus (MR) as follows .

Mixture Stiffness

M ixture stiffness (Sml values were determined as previously discussed for UMR

laboratory data and AASHO Road Test materials. Each mixture set of values were

substituted into Eq. 6. The temperature and loading time of Missouri conditions were

used for the UMR study dataset in the numerator, and AASHO Road Test conditions

were used for the AASHO Road Test mixes in the denominator :

S ] 1/3
a - a m
1. 2 - AASHO 1 , 2 [ Sm, AASHO

The results are shown in Table 32 .

Dynamic Modulus

In a manner similar to mixture stiffness, dynamic modulus values were

substituted into Eq . 6:

113
a
l. 2
=a
AASHO 1 '2
[
IE·IE·I ]
1AASHO

The results are shown in Table 32 .

 137

Resilient Modulus

Again, in a manner similar to the above, estimations of MR for UMR study data

(Missouri road conditions) and AASHO Road Test mixes (Road Test conditions) were

substituted into Eq. 6:

a1,2 = aAASHO 1,2

MRest
1-----
MR,AASHO
J/ 3

The results are shown in Table 32.

Table 32. Layer Coefficients for Bituminous Mixtures.

Mix AASHO AASHTO Chart a(E/E/t a(Sm/Smt a( MRI MRl"

Types

UMR 1 UMRest 2 UMR 3 UMR 4 UMR 5

C 0.44 0.471 0 .450 0.321 0.476 0 .406

1-C 0.44 · 0.465 0.443 0.324 0.464 0 .412

Bit .Base 0 .34 0.362 0.345 0 .378 0 .393 0 .335
Notes: 1. MR laboratory data; 68°F; 0 .1 sec duration
2. MR estimated from industry-wide regression equation; 68°F;
AASHO Road Test field conditions
3. E estimated from Witczak equation;Missouri field conditions for
UMR data; AASHO Road Test field conditions for AASHO data
4. Sm estimated from Shell equations; Missouri field conditions for
UMR data; AASHO Road Test field conditions for AASHO data
5. MR estimated from industry-wide regression equation; Missouri
field conditions for UMR data; AASHO Road Test field conditions
for AAS HO data.

METHOD OF CHOICE

Based on a review of the literature, all of the above methods appeared to give

reasonable results with the possible exception of the Ode mark method using IE* I.
 138

To determine the method of choice, the four methods which used estimated data

(IE* I, Sm, estimated MR) were correlated with the method which used real data

(laboratory MR) . R2 values were calculated for each correlation with the results from

laboratory data entered into the AASHTO nomograph . The results are as follows

shown in Table 33 .

Table 33. Comparison of Layer Coefficient Determination Methods.

Method R2

estimated - MR entered into the AASHTO chart 0 .874

-M -113 0.859
a R, est
AASHTO M
R,AASHO

- Sm
-1;3 0.857
aAASHO
Sm,AASHO

- lE*i
-113 0.526
aAASHO
: E*: AASHO

Option One

For a given mix design, if laboratory MR data is available , layer coefficients should be

obtained from the AASHTO nomographs. If MR data is not available, MR values

should be estimated from Eq. 53, and the results should be used in the AASHTO

nomographs to obtain the layer coefficients.

In Table 26 are shown the MR values estimated from Eq. 53 for 1990 MHTD

approved mixes at 68°F . By use of the AASHO nomograph, the resulting layer

coefficients were obtained and are as shown in Table 34 .

 139

Table 34. Layer Coefficients for 1990 MHTD Mixes.

Mixture Type Layer Coefficient

Type C 0.42

Type 1-C 0 .42

Bituminous Base 0.34

Option Two

The above option is recommended in the 1986 AASHTO Guide, which relies on

determination of a specific modulus at 68 °F. This ignores the fact that pavement

temperatures in Missouri are different than those in Ottowa, Illinois. An alternative

solution would be to determine layer coefficients as a function of pavement

temperature. Use of Eq. 16 will allow this concept to be util ized.

The method is easily implemented. The steps are shown in Table 35. The

necessary information includes: asphalt layer thickness, average annual air

temperature (See Fig. 45) and mix design data or MR data at three temperatures. As

indicated in Fig . 45, mean annual air temperatures range from 50 to 60°F in Missouri.

By use of Witczak's pavement temperature equation, this means that across the state,

for a pavement of 1 . 75 in of Type C or IC over an 8 in bituminous base , the mean

pavement temperatures would be 59.8 to 72.2 and 58.9 to 70.8°F for the surface

and base courses, respectively. The 1990 MHTD mix designs were substituted into

the equations in Table 35 and the results are shown Fig. 46 . Note that the 1990

mixes straddle the lower boundary of acceptance in accordance with AAMAS criteria.

As an example, the layer coefficients corresponding to these moduli were calculated

 140

Table 35. Layer Coefficient Determination.

Step No . Action
1 Obtain mean annual air temperature (TA) from Fig . 45

2 Calculate mean annual pavement temperature:

7f =6.0- 34 _
+ T r1+ 1
P [ ( D/ 3 ) + 4 ) A -::-[-,-(D=-/....,...,3"""')-~.....,.4~)
1
3 Calculate mean annual resilient modulus (MR):
MR= 10A[6.871-0.017Tp-0.024Pair+0.043~ 70
+O.Ol8P200-0.004AR3;4-0.0llPeffv]
Where: rJ 70 = 29,508 .2 (penetration 77 0Fr 2 · 1939
Peffv = Pb/Q.483
Pb = total asphalt content by weight of mix
P., 11 v can also be calculated more accurately by weight-
volume calculations.

4 Calculate layer coefficient:

- -o.333
MR
a1 = 0.44~~~=
656,772
= •o.333
MR
a 2 = 0.341--~~-
604,962
Note: 1. MR of AASHO Road Test surface /binder = 656,772
psi at Witczak-
estimated pavement temperature , weighted for thicknesses of
surface and binder.
2. MR of AASHO Road Test base = 604,962 ps i at estimated base
temperature.
 141

~\
;. /
I
'~

Fig. 45 . Temperature Contour Map of Missouri.

 142

----- --.... Moc ulus T bo fUo It

-~
-........ "'-.
.......
~
.

I
" ~ ' ."' ...
~

I
~
I

...,, "" I\....

' I'-
I

I
'
I

'
'
i
! : ,:-,... '" "X
,"1, 1, 1 I
.J\..'

I I
'"\·'
'~ , '
, ~,
,_ ~

Mqdu ~us Too Lo w

... -
I

I I
I :' I
-
I
I I

0 20 40 60 80 100 120
Temperature, °F

Fig. 4-6. MHTD 1990 Mixtures on AAMAS Acceptance Chart.

 143

via Table 35 and are presented in Table 36 .

Table 36. 1990 MHTD Mix Designs Resilient Moduli in Three Parts of Missouri .

Mix Type -TA ( o F) -T P ( °F) Layer Coefficients

Min 1 Avg 2 Max 3 AASHO

50 59 .8 0 .40 0.42 0.43

C 55 . 1 65.2 0.37 0.39 0 .40 0.44

60 72.2 0 .34 0.35 0.36

50 59.8 0.40 0.42 0.42

IC 55 . 1 65.2 0 .38 0.39 0.40 0.44

60 72.2 0.34 0 .36 0.36

50 58 .9 0.33 0.35 0 .38

BB 55.1 63 .4 0.31 0 .33 0 .36 0.34

60 70 .8 0.28 0.30 0 .32

Note : 1 = least stiff 1990 mix design
2 = average of all 1990 mix design
3 = most stiff 1990 mix design

Option Two gives somewhat lower coefficients than option One for the surface

mixes, but about the same for the bituminous base mixes. The reason for this is

found by looking at the MA predictive equation. First, the average air temperature in

Missouri is about 5 ° F warmer than that which was recorded at Ottowa during the

Road Test. This accounts for a great deal of the loss in MA when comparing MHTD

to Road Test material. Now, keeping temperature constant for both kinds of

materials, for average conditions of temperature and mix design, the Type C and IC

mixes had the following differences in mix variables.

 144

C and IC vs AASHO. Looking at Table 37, MHTD mix designs on the average have

higher air voids, lower P 200 , and higher asphalt contents, all of which diminish MR

according to the MR predictive equation. Asphalt viscosity and AR 314 are about the

same. Thus, one would expect the AASHO MR to be higher, even at the same

temperature.

Table 37. Comparison of AAS HO and MHTD Mix Designs.

I Material
I P.;,
I 'ho I P200 I AR314 I p effv I
AASHO* 4.3 1.49 5.0 2 .3 9.1
MHTD-C 5.3 1.49 3.5 0 10.0
MHTD-IC 6.0 1.51 4.3 0 9.6
AASHO 6.2 1.49 5.6 4.0 10.8
Base
MHTD-BB 4.8 1.51 7.7 10.6 9.1
*weighted average of surface plus binder courses

Bituminous Base vs AASHO. MHTD mix designs averaged lower air voids, higher P 200 ,

and lower effective asphalt contents, all of which would increase the predicted

moduli. MHTD did have more AR 314 than AASHO, but care must be taken when

interpreting the effect of changing variable magnitudes that are interactive with other

variables, such as different sieve sizes.

Thus, keeping in mind differences in mix design and temperature between

MHTD pavements and Road Test pavements, one might expect MHTD C and IC layer

coefficients to be somewhat lower, but bituminous bases about the same as AASHO

pavement layers.
 145

SENSITIVITY ANALYSIS

A sensitivity analysis was performed by looking at the effect on required

pavement thickness by 1) mixture variables, and 2) layer coefficients.

MIXTURE VARIABLE EFFECT ON THICKNESS

As shown in Table 23, only asphalt viscosity and extremes in gradation were

significant in their effect on MR. An analysis was performed to ascertain the effect

on layer thickness of varying asphalt viscosity grade or aggregate gradation.

Thickness of the surface layer was varied at 1.25 in and 4 in for both Type C and IC

mixes. For the bituminous base mixes, thickness was examined at 4, 8, and 14 in.

Three levels of viscosity or gradation change were looked at: most significant change

in the UMR data set, least significant change in the UMR data set, and average

change for all mixes of a certain type. For instance, for the Type C mixtures, the

largest change in MR as a result of a change in gradation occured in the C5F20-4.0

mixture as gradation changed from the fine side (MR= 762,000 psi) to the coarse side

(MR= 562,000). The least change occured where the C4C20-4.25 mix changed from

the fine side to the coarse side (essentially no change in measured MR). The

"average" condition was simply a comparison of the average of the MR values for all

the Type C fine gradation mixtures to the average of the MR values for all the coarse

gradations at 68°F (as shown in Table 23). Gradation or viscosity was changed from

high to low with a resulting change in MR. Then the resulting change in layer

coefficients (a 1 or a 2 ) were calculated from the AASHTO nomographs. Finally, the

required change in thickness was computed as needed to maintain the initial structural
 146

number rendered by the initial assumption of layer thickness. The results are shown

in Table 38. All data are based on 68°F . For example, what is the average change

in required surface layer thickness for an initial thickness of 1.25 in using a Type C

mix when gradation is changed from the fine side to the coarse side? Looking at row

3 in Table 38, the MR for the fine side Type C is 567,625 psi. Moving to the coarse

side results in a reduction to 504,375 psi. Using the AASHTO nomograph, the

corresponding loss in a, is 0.02. The structural number (SN) provided by the fine side

mix is SN = a,D, or SN = 0.48 • 1.25 = 0.60. When the layer coefficient is reduced

from 0.48 to 0.46, the new required thickness D 2 = SN/a, = 0.60/0.46 = 1.30, or,

an additional requirement of 0 .05 in. This is not significant in a practical sense.

From Table 38, it appears that wide swings in gradation are not significant for

thin ( 1.25 in) layers, and are marginally significant for 4 in layers. For thicker layers,

as encountered with bituminous bases, the effect becomes more important. At an

initial 8 in required thickness, changes varied from 0.02 to about 1. 5 in. At 14 in, the

variation was from 0.04 to 2.66 in (on the average, the required extra thicknesses

were 0.61 and 1.06 in for the 8 and 14 in initial thicknesses, respectively).

Changes in viscosity grade rendered the following results. For the C and IC

mixes, thin surface layer (1.25 in) thickness changes varied from 0.03 to 0.31 (0.15

average) in, while 4 in layer thickness changes were 0.11 to 1.00 (0.48 average) in.

For the bituminous bases, the variance for a 4 in layer ranged from 0.08 to 0.94 (0.38

average) in. An 8 in layer could see a variance from 0.16 to 1.88 (0. 76 average) in,
 147

Table 38. Thickness Sensitivity to Changes in Gradation and Viscosity.

GRADATION CHANGE

Dll CASE MRl MR2 MRdiff a,11 a,12 al SN D12 Dldiff

in. psi psi psi diff in. in.

TYPE C

1. 25 WORST 762000 562000 200000 0.53 0.48 0.05 0.66 1. 39 0.14

4.00 762000 562000 200000 0.53 0.48 0.05 2 .11 4.43 0.43

1.25 AVG 567625 504375 63250 0.48 0.46 0.02 0.60 1. 30 0.05
4.00 567625 504375 63250 0.48 0.46 0.02 1. 91 4.18 0 . 18

1.25 LEAST no change 0 0

4.00 no change 0 0

TYPE IC

1.25 WORST 725000 500000 225000 0.52 0.46 0.06 0.65 1. 42 0.17
4.00 725000 500000 225000 0.52 0.46 0.06 2.08 4.55 0.55

1. 25 AVG 605875 508000 97875 0.49 0.46 0.03 0.61 1. 33 0.08

4.00 605875 508000 97875 0.49 0.46 0.03 1. 95 4.26 0.26

1. 25 LEAST no change 0 0
4.00 no change 0 0

BITUMINOUS BASE

D21 CASE MRl MR2 MRdiff a21 a22 a2 SN D22 D2diff

in. psi psi psi diff in. in.

4 WORST 790000 494000 296000 0.41 0.35 0.07 1.66 4.76 0.76
8 790000 494000 296000 0.41 0.35 0.07 3.31 9.52 1. 52
14 790000 494000 296000 0.41 0.35 0.07 5.80 16.66 2.66

4 AVG 605125 500875 104250 0.38 0.35 0.03 1. 51 4.30 0.30

8 605125 500875 104250 0.38 0.35 0.03 3.01 8.61 0.61
14 605125 500875 104250 0.38 0.35 0.03 5.27 15.07 1. 07

4 LEAST 606000 601000 5000 0.38 0.38 0.00 1. 51 4.01 0.01

8 606000 601000 5000 0.38 0.38 0.00 3.02 8.02 0 . 02
14 606000 601000 5000 0.38 0.38 0.00 5.28 14.04 0.04
 148

VISCOSITY CHANGE

Dll CASE MRl MR2 MRdiff all a12 al SN D12 Dldiff

in. psi psi psi diff in. in.
TYPE C

1.25 WORST 626000 364000 262000 0.38 0.31 0.08 0.48 1. 56 0.31
4.00 626000 364000 262000 0.38 0.31 0.08 1. 53 5.00 1.00
1.25 AVG 613250 458750 154500 0.38 0.34 0.04 0.47 1.40 0.15
4.00 613250 458750 154500 0.38 0.34 0.04 1. 51 4.48 0.48

1.25 LEAST 562000 525000 37000 0.37 0.36 0.01 0.46 1.28 0.03
4.00 562000 525000 37000 0.37 0.36 0.01 1.47 4.11 0.11
TYPE IC

1.25 WORST 725000 480000 245000 0.40 0.34 0.06 0.50 1. 46 0.21
4.00 725000 480000 245000 0.40 0.34 0.06 1. 61 4.68 0.68

1.25 AVG 613250 458750 154500 0.38 0.34 0.04 0.47 1. 40 0.15
4.00 613250 458750 154500 0.38 0.34 0.04 l. 51 4.48 0.48

1.25 LEAST 533000 475000 58000 0.36 0.34 0.02 0.45 l. 31 0.06
4.00 533000 475000 58000 0.36 0.34 0.02 1.44 4.19 0.19

BITUMINOUS BASE

D21 CASE MRl MR2 MRdiff a21 a22 a2 SN D22 D2diff

in. psi psi psi diff in. in.

4 WORST 601000 362000 239000 0.38 0.30 0.07 l. so 4.94 0.94

8 601000 362000 239000 0.38 0.30 0.07 3.01 9.88 1. 88
14 601000 362000 239000 0.38 0.30 0.07 5.26 17.29 3.29

4 AVG 617375 488625 128750 0.38 0.35 0.03 1.52 4.38 0.38
8 617375 488625 128750 0.38 0.35 0.03 3.04 8.76 0.76
14 617375 488625 128750 0.38 0.35 0.03 5.31 15.33 l. 33

4 LEAST 494000 470000 24000 0.35 0.34 0.01 1.39 4.08 0.08
8 494000 470000 24000 0.35 0.34 0.01 2.79 8.16 0.16
14 494000 470000 24000 0.35 0.34 0.01 4.87 14.29 0.29

while a thick 14 in layer's required change might be from 0.29 to 3.29 ( 1.33 average)

in. So, for thicker pavements, using a harder grade of asphalt could lead to significant

changes in design thickness.

LAYER COEFFICIENT EFFECT ON THICKNESS

As previously discussed, layer coefficients calculated in accordance with Option

 149

Two can vary as a result of changes in temperature and mix design. The following

is an analysis of the effect of these two variables on required layer thickness.

Three typical pavement sections were examined, and the change in required 0 2

(bituminous base thickness) was calculated. The three sections all contained a 1.25

in surface layer 0 1 , with 0 2 varying as 4, 8 , and 14 in. Using the AAS HO Road Test

layer coefficients (a, = 0 .44, a 2 = 0.34), structural numbers (SN) were calculated for
these pavements. Then, using a constant 0 1 = 1.25 in, 0/s were calculated based

on worst , average , and best temperature conditions and mix quality in Missouri. From

Table 35, layer coefficients for these conditions for the surface and base layers were

obtained. Finally, the 0/s required to maintain the computed SN's were calculated.

The results are shown in Table 39. As an example, using the AASHO layer

coefficients of a, = 0.44 and a 2 = 0 .34, for a 1.25 in over 8 in bituminous base

structure, the provided SN = 3.27. Now, for average conditions in Missouri

(temperature and 1990 mix quality), using a 1.25 in surface layer, the required 0 2 =
8.43 in, or a required increase of 0 .43 in. From examination of the table, several

things are apparent. First, average temperature and mix quality conditions lead to

rather small changes in required base thickness. Worst case sceneries of lower

quality mix used in the warmest part of the state lead to significant increases in

required base thickness. On the other hand, use of higher quality mixes in cooler

parts of the state lead to reductions in required thickness.

 150

Table 39. Thickness Sensitivity to Ranges of Layer Coefficients.

AASHO MHTD
o, D2 SN o, D,(in)
(in) (in) provided (in) Worst Average Best
a, = 0.44 a2 = 0.34 a, = 0.34 a, = 0.39 a, = 0.43
a2 = 0.28 a2 = 0.33 a2 = 0.38
1.25 4 1.91 1.25 5.30 4.31 3.61
1.25 8 3.27 1.25 10.16 8.43 7 . 19
1.25 14 5.31 1.25 17.45 14.61 12.56

SUMMARY

The purpose of this investigation was to determine layer coefficients for several

MHTD specified pavement materials. The coefficients are necessary as input to the

AASHTO pavement design method . Volume I of this study involves asphaltic

materials, and is reported herein. Volume II deals with unbound aggregate base and

soil-cement base materials, and is reported elsewhere .

Besides determining layer coefficients, the study also entailed the determination

of the effect on layer coefficient by changes in asphalt cement grade, aggregate

gradation, testing temperature, aggregate source, and asphalt content within the limits

of MHTD specifications. This resulted in 48 mix designs.

1. All materials were sampled and delivered to UMR by MHTD personal. Choice

of material sources was made by MHTD. The types of pavement materials

were Type C, Type 1-C, and bituminous base. The specific materials making up
 151

these types were two grades of asphalt cement, two surface mix sources of

coarse aggregate, two base mix sources of coarse aggregate, one source of

natural sand, two sources of manufactured sand, one source of mineral filler,

and one source of hydrated lime.

2. Routine index and specification tests were performed. For the asphalt cement,

the tests were: penetration at 38 ° and 77°F, kinematic viscosity, absolute

viscosity, specific gravity, and softening point. The aggregates were tested for

gradation , specific gravity, and particle shape/texture. Equipment was

fabricated for the particle shape/texture tests .

3. The aggregates were separated on each sieve size and stored .

4. The optimum asphalt content of each of the 12 gradation/aggregate source

combinations was determined by use of the Marshall mix design method (75

blow, manual flat-faced hammer). The same optimum asphalt content was

used for both the AC10 and AC20 grades because mixing and compaction

temperatures were adjusted to give equal mixing and compaction viscosities for

both grades. This resulted in 24 mixes. Finally, 24 additional mixes were used

which had 0.5% asphalt added above optimum. Thus, the total number of

mixtures was 48. Approximately 200 specimens were made.

5. Maximum theoretical specific gravities were determined in two ways: 1) Rice

method, and 2) calculation from material proportions and specific gravities. 96

specimens were tested.

6. A voids analysis was conducted to determine the effect of estimation of

 152

maximum theoretical specific gravity.

7. Ten methods of characterizing gradation curve shape and position were used.

Two of these were unique to this study. The first involved the area between

the gradation curve and the maximum density line as plotted on FHWA 0 .45

power paper. The second method involved determination of slopes of three

portions of each gradation curve.

8. Each mix was tested for indirect tensile strength. A regression model was fit

to the data . 96 tests were performed.

9. Each mix was tested for total resilient modulus (indirect tension) at three

temperatures: 41 °, 77°, 104°F. Necessary software and equipment were

programmed and developed to perform the tests, and to acquire, store, and

analyze the data. A total of 192 specimens were tested at three temperatures

for a total of 576 tests.

10. The results of the MA testing were analyzed statistically to determine the

variables that were significant to changes in MA. The variables that were

examined were asphalt viscosity, asphalt content, testing temperature, changes

in gradation, and particle shape/texture.

11. A regression model was fit to the MA data.

12. Resilient modulus data from other studies found in the literature were merged

with the UMR data. A general regression model was fit to the overall data

base .

13. Air temperature data from 104 weather stations in Missouri were analyzed to
 153

produce an air temperature contour map of Missouri.

14. Pavement thickness data for MHTD flexible pavements were analyzed for mean

pavement thickness.

15. Information from steps 13 and 14 was necessary to calculate pavement

temperatures for Missouri .

16. Steps 13 and 14 were repeated to obtain pavement temperatures at the

AASHO Road Test.

17. Mean vehicle speed data was supplied by MHTD. This was converted to load

dwell times and frequency for MHTD pavements. The same was done for

AASHO Road Test pavements.

18. UMR, AASHO Road Test, and MHTD 1990 mix data were used to estimate

resilient modulus, mixture stiffness (Shell method), and dynamic modulus at

both laboratory conditions and field conditions of pavement temperature and

loading rate . This was in order to see which modulus would be most useful for

layer coefficient determination .

19. Five different methods of calculating mixture stiffness (Sm) were compared ;

each varied in the manner of handling asphalt aging or source.

20. Layer coefficients a, (for Types C and IC mixes) and a 2 (for bituminous base

mixes) were determined in several ways:

a. MR was estimated via the general regression model equation developed

in step 12. Then, these MR values were used in the AASHTO

nomographs to obtain a, or a 2 .

b. Use of the Odemark equation was made in three ways:

 154

aU!1R = aAASHO
MR U!1R
1--'--
Jl/3
MR,AASHO

au11R
= a
AASHO
t:
, E*
1
E* : U!1R
,
IAASHO
]1/ 3

Sm,U!1R
aU!1R = aAASHO - - - -
Jl/3
sm,AASHO

The above numerators were calculated at Missouri pavement

temperatures (high, average, and low areas of the state) and the

denominators were calculated at the mean pavement temperature at the

Road Test .

21 . Two options were presented for the calculation of layer coefficients. Option

One resulted in a fixed layer coefficient per material. For 1990 mixes, Type C

a, = 0 .42, Type IC a, = 0.42, and bituminous base a 2 = 0.34. Option Two

is a method fo r pavement designers to calculate layer coefficients for a specific

mix and location in the state.

22. A sensitivity analysis was performed which examined the effect of specific

important mix characteristics (viscosity and gradation), pavement temperature,

and overall mixture quality .

· CONCLUSIONS

1. Ten methods were used to characterize gradation curve shape and position;

two of these were developed during the course of the study . The method of

determining the area between the 0 .45 power maximum density line (MDL) and
 155

the gradation line had only a fair (R 2 = 0. 79) correlation with resilient modulus,

MR. This was because the area was not sensitive to relatively small differences

in position of the curve relative to the MDL. The second unique method

involved calculation of the slope of three different parts of the gradation curve.

This method was shown to be of assistance in creating a more accurate MR

regression model. However it was not quite as helpful as Hudson's A, which

is much easier to calculate . But, Hudson's A was not quite as helpful as

merely including certain critical sieve sizes directly in the regression equation .

2. The voids analysis indicated that the method of assuming that the effective

specific gravities of low absorption aggregates is midway between the bulk and

the apparent specific gravities correlates very well with results from Rice

method testing. However, for absorptive aggregates(§£., the bituminous base

materials in this study), the estimation method underpredicts air voids by about

1%.

3. Of the five different methods for computation of mixture stiffness (Sm), the

method of Bonnaure, which uses the Ullidtz asphalt aging approximations, was

found to be the most accurate for the purposes of this study.

4. The regression model for indirect tensile strength was relatively strong (adj-R 2

= 0.840) and was as follows:

IDT = 134.064 + 21 .238'770 - 7 .553 Petfv - 0.687 AR4 + 1.388U - 2.145 IP

where:
 156

'ho = asphalt viscosity at 70 °F, poise* 106

p effv = % effective asphalt content by volume

AR 4 = % accumulative retained on the #4 sieve

u = % uncompacted void content; NAA method for fine

aggregate particle shape

IP = Index of Particle shape for coarse aggregate

5. The resilient modulus test is sensitive to testing conditions of temperature,

specimen rocking, specimen surface irregularities, choice of point of LVDT

fixation, LVDT tip design, and resolution of both vertical and horizontal LVDT'S.

Constant diligence is required by the operator to assure that the very small

deformations being measured are representative of actual deformations .

6. A relatively strong (adjusted R2 = 0.946) regression model was fit to the UMR

MR data:

log MR = 7 .137- 0.016 T - 0.005 AR4 + 0.088(}70 - 0.028 peffv - 0.016 peir -

4
0.006 p200 (JPopteffv - peffvl)

where :

MR = total resilient modulus, psi

T = mixture temperature, °F

AR 4 = % accumulative retained on the #4 sieve

P200 = % passing the #200 sieve

peir = % air voids

(I p opteffv - p effv I} = % absolute value of the difference between optimum

 157

effective asphalt content and actual effective asphalt

content, by volume .

7. A statistical analysis of the data indicated that temperature was by far the most

important variable that affects MR, followed by asphalt viscosity and whether

the gradation was very fine or very coarse . Overasphalting by 0. 5 % tended to

lower MR, but was not statistically significant. Increases in (-) #200 material

and decreases in ( +) #4 material tended to increase MR. Particle shape of

coarse or fine aggregate did not seem to affect MR in a consistent manner. It

should be noted that both coarse aggregates were crushed limestones, and that

all mixes contained varying amounts of manufactured sand, so large ranges in

particle shape were not present. All other things held constant, decreasing air

voids tended to increase MR .

8. Because layer coefficients are directly affected by resilient modulus, the

practical impact of the trends is that higher layer coefficients can be obtained

by:

1. using a harder grade of asphalt (higher f'7 0 ),

-
2. using a more well-graded gradation (higher A, adjusting P200 and AR 4 ),

3. avoiding overasphalting (optimum Peffvl, and

4. increasing density (lower air voids).

Of course, caution should be exercised when considering these options.

Excessively hard asphalts, highly angular aggregates, and low air void values

can lead to such problems as thermal cracking, lack of workability, poor

 158

durability, and rutting. Thus, caution should be used when interpreting the

results of the above analysis. Also, it must be kept in mind that the MR

predictive equations are based on data that represents well-graded gradations.

They should not be applied to mixes with significantly different characteristics

or materials, such as stone mastic mixtures.

9. The UMR data base was augmented with mix design/MR data from five

additional studies obtained from the literature. A second, more generally

applicable, regression model was produced. The model was not as strong as

the UMR model, but was deemed superior because it represented a much wider

range in magnitudes of variables. The equation is:

log MR= 6.871-0.017T-0.024Pair+0.043~ 70 +0.018P200-0.004AR 3 ; 4 -0.011Peffv

10. The Odemark equation was used to rate the three methods of obtaining mixture

modulus or stiffness. The ranking, in descending order of ability to predict

resilient modulus, was: MR estimated from step 9, Shell mixture stiffness, and

dynamic modulus.

11. Two options to obtain layer coefficients were presented for possible use. The

first involves the determination of MR by test (or by estimation of resilient

modulus by the equation in step 9) for a pavement temperature of 68°F, then

entering the proper AASHTO nomograph. The second option is to again

determine the MR by test or to estimate the resilient modulus, but the moduli

must be converted to the pavement temperature conditions in the locale of

interest. Then, the layer coefficient is computed via the Odemark equation
 159

which relates the MHTD MR to the AASHO Road Test MR.

RECOMMENDATIONS

For Missouri pavement designers, two options are presented. Option One is the

simplest: if MR test data are available for a given mix (say, for three temperatures),

the MR value at 68 ° should be ascertained; then the AASHTO nomographs (Figs. 42

and 43) should be entered with the MR 68 value and the a 1 or a 2 value determined.

If MR test data are not available, the specific mix data should be substituted into Eq.

53, repeated here, and the above procedure followed to obtain a 1 or a 2 values .

MR= 10A(6.871-0.017Tp-0.024Pair+0.043~ 70 +0.018P200-0.004AR3;4-0.0llPeffv)

Alternatively, if a designer wishes to obtain a layer coefficient value more in

keeping with the design location and pavement thickness, the procedure in Table 35

should be followed.

It is highly recommended that the MHTD pursue MR testing of various mixtures

in present use in order to update or replace the above equation by use of a more

representative data set of the materials. A greater degree of accuracy will also

probably be achieved. Then, both Options One and Two will render more

representative layer coefficient values for MHTD designers.

It should be remembered that this study is in the mold of the traditional method

of determination of layer coefficients, that is, by a comparison of some sort of

strength or stiffness of MHTD materials to Road Test materials. Tendencies for

asphaltic material problems with thermal cracking and rutting, for instance, are not

directly addressed. To address a wider range of material issues, creep testing and
 160

gyratory shear testing may be in order. These kinds of tests were beyond the scope

of this project . Also, this project was conceived in 1989 and the bulk of the testing

was performed in 1991, before the SHRP project results became generally known.

In the future, it may be that some of the recommendations coming out of the SHRP

program can be used to update the quest for layer coefficient determination .

ACKNOWLEDGEMENT

The authors wish to thank the MHTD for its sponsorship and support of this

research project . They also thank the UMR Department of Civil Engineering for its

support. Special thanks go to Mr. J.D. Stevenson and Mr . Kevin Hubbard for their

assistance in the computer analysis portion of the study .

REFERENCES

1. AASHTO 1986 Guide for Design of Pavement Structures, AASHTO,

Washington , D.C., 1986 .

2. Richardson, D.N ., W .J . Morrison, and P.A . Kremer, "Determination of

AASHTO Drainage Coefficients, Missouri Cooperative Highway Res . Program

Final Report. Study 90-4. Univ. of Missouri-Rolla, Rolla, Missouri, 1994.

3. Gomez, M . and M.R. Thompson, "Structural Coefficients and Thickness

Equivalency Ratios, "Trans. Engrg . Series No. 38." Illinois Cooperative

Highway and Transportation Series No. 202, University of Illinois, 1983, 48

p.

4. Transportation Engineering Handbook, Ch. 50, Pavement, U.S. Forest

Service, 1974, p. 51.1-54 .4 - 16.

 161

5. Van Til, C.J., B.F. McCullough, B.A. Vallerga, and R.G . Hicks, "Evaluation of

AASHO Interim Guides for Design of Pavement Structures," NCHRP 128,

Hwy. Res. Bd., 1972, 111 p.

6. Wang, M.C., T.D. Larson, and W.P. Kilareski, "Structural Coefficients of

Bituminous Concrete and Aggregate Cement Base Materials by the Limiting

Criteria Approach," Report No. FHWA-PA-RD- 75-2-4, Pennsylvania State

University, University Park, PA, 1977, 52 p.

7. Sowers, G.F., "Georgia Satellite Flexible Pavement Evaluation and Its

Application to Design," Hwy. Res. Rec . 71, 1965, pp. 151-171 .

8. Walters, R. "Implementation of the New AASHTO Design Practice," 31st

Annual UMR Asphalt Conference, University of Missouri-Rolla, Rolla,

Missouri, 1988.

9. Rada, G. and M.W. Witczak, "Material Layer Coefficients of Unbound

Granular Materials from Resilient Modulus, "Trans. Res. Rec. 852, Trans.

Res. Bd., 1982, pp. 15-21.

10. Jorenby, B.N. and R.G. Hicks, "Base Course Contamination Limits, Trans.

Res. Rec. 1095, Trans. Res. Bd., 1986, pp. 86-101.

11. The AASHO Road Test - Report 5 - Pavement Research, Spec. Report 61 E,

Highway Res. Bd., 1962, 352 p.

12. Carree, B.J.D. and T.D. White, "The Synthesis of Mixture Strength

Parameters Applied to the Determination of AASHTO Layer Coefficient

Distributions," Proc. of Assn. of Asphalt Paving Tech., Vol. 58, 1989, pp.
i---- -

162

109-141.

13 . Richardson, D.N. and P.A . Kremer, "Determination of AASHTO Layer

Coefficients , Vol. II : Unbound Granular Bases and Cement Treated Bases , "

Missour i Cooperative Highway Res. Program Final Report. Study 90-5,

University of Missouri-Rolla, Rolla , Missouri , 1994.

14. Odemark, N. "Undersokning av Elasticitetsegenskaperna hos Olika Jordarter

Samt Teori for Berakning av Belagninger Enligt Elasticitetsteorien," Statens

Uaginstitut, Meddelande, 77, 1949.

15 . Van der Poe I, C. "A General System Describing the Viscoelastic Properties of

Bitumens and Its Relation to Routine Test Data," Shell Bitumen Reprint No .

.9_.

16. Pfeiffer, J.P. and P.M . Van Doormaal , "The Rheological Properties of

Asphaltic Bitumen , " J. Inst. of Petroleum Tech., Vol. 22, 1936, pp. 414 .

17. Heukelom, W., "An Improved Method of Characterizing Asphaltic Bitumens

with the Aid of Their Mechanical Properties," Proc. Assn. of Asphalt Paving

Tech. Vol. 42, 1973, pp.67-98 .

18. Heukelom, W . and A. I. G. Klomp, "Road Design and Dynamic Loading,"

Proc. of Assn. Asphalt Paving Tech., Vol. 33, 1964, pp . 92-123.

19. Van Draat, F. and P. Sommer , "Ein Geratzar Bestimmung der Dynamischen

Elastizitats Modulu von Asphalt," Strasse und Autobahn, Vol. 35, 1965.

20 . McLeod, N.W., "Asphalt Cements: Pen-Vis Number and Its Application to

Moduli of Stiffness," J . of Testing and Eval., Vol. 4, No . 4, 1976, pp. 275 -

 163

282.

21. Bonnaure, F., G. Gest, A. Gravois and P. Uge, "A New Method of Predicting

the Stiffness of Asphalt Paving Mixtures," Proc. of Assn. of Asphalt Paving

Tech., Vol. 46, 1977, pp. 64 -104.

22. Claessen, A.I.M., J.M. Edwards, P. Sommer, and P. Uge, "Asphalt

Pavement Design: The Shell Method," Proc. 4th lnt'I. Conf. on Structural

Design of Asphalt Pavements, Univ. of Michigan, Ann Arbor, 1977.

23. Ullidtz, P., "A Fundamental Method for Prediction of Roughness, Rutting,

and Cracking of Pavements," Proc. of Assn. of Asphalt Paving Tech., Vol.

48, 1979, pp. 557-586.

24. Witczak, M.W., "Design of Full Depth Asphalt Airfield Pavements, " Proc.

Third lnt'I. Conf. on the Structural Design of Asphalt Pavements, Vol. 1,

1972, London, England, pp. 550-567.

25. Harr, M.E., Reliability-Based Design in Civil Engineering, McGraw-Hill Book

Co., 1987.

26. Lefebvre, J.A., "A Modified Penetration Index for Canadian Asphalts," Proc.

of Assn. of Asphalt Paving Tech., Vol. 39, 1970, pp. 443.

27. Kandhal, P.S., and W.C. Koehler, "Effect of Rheological Properties of

Asphalts on Pavement Cracking, ~roe. of Assn. of Asphalt Paving Tech., pp.

99-117 .

28. Bell, C. A., "Use of the Shell Bitumen Test Data Chart in Evaluation of

Asphalt Data;" Proc. of Assn. of Asphalt Paving Tech., Vol. 52, pp., pp.1 ·
 164

31 (1983) .

29. Button, J.W., J.A. Epps, D.N. Little, and B.M. Galloway, "Asphalt

Temperature Susceptibility and Its Effect on Pavements," Trans . Res. Rec.

843, Trans. Res. Bd., pp. 118-126.

30. Puzinauskas, V.P. "Properties of Asphalt Cements," Proc. of Assn. of

Asphalt Paving Tech., Vol. 48, 1979, pp. 646-71 o.

31. Shook, J. F. and B.F. Kallas, "Factors Influencing Dynamic Modulus of

Asphalt Concrete," Proc. of Assn. of Asphalt Paving Tech .• Vol. 38, 1969,

pp.140-178.

32. Witczak, M.W., Development of Regression Model for Asphalt Concrete

Modulus for Use in MS-1 Study. Asphalt Institute, 1978, 39 p.

33. Miller, J.S., J. Uzan, and M.W. Witczak, "Modification of the Asphalt

Institute Bituminous Mix Modulus Predictive Equation," Trans. Res. Rec. 911

Trans. Res. Bd., 1983, pp. 27-36.

34. Akhter, G.F. and M.W. Witczak, "Sensitivity of Flexible Pavement

Performance to Bituminous Mix Properties," Trans. Res. Rec. 1034, Trans.

Res. Bd., 1985, pp. 70 - 79.

35. "Standard Test Method for Resilient Modulus of Subgrade Soils," AASHTO

T-274 (1986), Standard Specifications for Transportation Materials and

Methods of Sampling and Testing, 14th ed. Part 111, Methods of Sampling

and Testing, AASHTO, Washington, D.C., 1986, pp. 1198-1218.

36. "Standard Test Method for Indirect Tension Test for Resilient Modulus of
 165
11
Bitum inous Mixtures, ASTM D4123-82, Annual Book of ASTM Standards,

Vol. 04.03, ASTM, Philadelphia, PA, 1992, pp . 507-509.

37. Von Quintus, H.L., J.A. Scherocman, C.S. Hughs, and T .W. Kennedy,
11 11
Asphalt-Aggregate Mixture Analysis System, NCHRP Rpt. 338, Trans .

Res. Bd., Washington, D.C., 1991, 183 p.

38. Resilient Modulus for Asphalt Concrete, SHRP Protocol P07, Strategic Hwy.

Res. Pgm., Nat'I. Res. Council, Washington, D.C., 1992, pp. PB07B-PB07B-

17.

39. Standard Specifications for Transportation Materials and Methods of

Sampling and Testing, 15th Ed., Part II, Tests, AASHTO, Washington, D.C.,

1990.

40. Kalcheff, I. V., and D.G. Tunnicliff, "Effects of Crushed Stone Aggregate Size

and Shape on Properties of Asphalt Concrete," Proc. of Assn. of Asphalt

Paving Tech., Vol. 51, 1982, pp. 453 -483.

41 . Hadley, W.O., W.R. Hudson, and T. W. Kennedy, "An Evaluation of Factors

Affecting the Tensile Properties of Asphalt-Treated Materials," Res . Report

98-2, Center for Hwy. Res., Univ. of Texas at Austin, 1969.

42. Hadley, W. 0., W.R. Hudson, and T. W. Kennedy, "Evaluation and Prediction

of the Tensile Properties of Asphalt-Treated Materials," Res. Report 98-9,

Center for Hwy. Res., Univ. of Texas at Austin, 1971.

43. Meier, W.R., and E.J. Elnicky, "Laboratory Evaluation of Shape and Surface

Texture of Fine Aggregate for Asphalt Concrete," Trans . Res. Rec. 1250,
 166

Trans. Res. Bd., 1989, pp. 25-34.

44. Mogawer, W.S. and K.D. Stuart, "Evaluation of Test Methods Used to

Quantify Sand Shape and Texture," TRB 71 st Annual Meeting, Trans. Res.

Bd., Washington, D.C., 1992, 23 p.

45. Kandhal, P.S., J. 8. Motter, and M.A. Khatri, "Evaluation of Particle Shape

and Texture: Manufactured vs. Natural Sands, "NCAT Rpt. No. 91-3,

NCAT, Auburn, Ala., 1991, 23 p.

46. "Test Method for Index of Aggregate Particle Shape and Texture," ASTM

D3398-87, Annual Book of ASTM Standards, Vol. 05.03 ASTM,

Philadelphia, PA, 1992, pp. 393-396.

4 7. "Standard Test Method for Particle Shape, Texture, and Uncompacted Void

Content of Fine Aggregate," Draft, National Aggregate Assn., Silver Spring,

MD, 1991, 12p.

48. Anagnos, J. N. and T. W. Kennedy, "Practical Method of Conducting the

"
Indirect Tensile Test," Res. Report 98 -10, Center for Hwy. Res., Univ. of

Texas at Austin, 1972.

49. Kennedy, T.W., and J.N. Anagnos, "Procedures for the Static and Repeated-

Load Indirect Tensile Test," Res. Report 183-14, Center for Trans. Res.,

Univ. of Texas at Austin, 1983, 44 p.

50. "Test Procedures for Characterizing Dynamic Stress-Strain Properties of

Pavement Materials: Indirect Tensile Test," Special Report 162,

Washington, D.C., 1975, pp. 32-39.

 167

51. May, R.W., and M.W. Witczak, "An Automated Asphalt Concrete Mix

Analysis System," Proc. of Assn. of Asphalt Paving Tech ., Vol. 61, 1992,

pp. 154-187.

52 . de Bats, F. Th ., and G. van Gooswilligen, "Practical Rheological

Characterization of Paving Grade Bitumens," Fourth Eurobitume Symp.,

Madrid, pp. 1-8 (1989) .

53. Barksdale, R.D., "Compressive Stress Pulse Times in Flexible Pavements for

Use in Dynamic Testing," Hwy. Res . Rec . 345, Hwy. Res. Bd., 1971, pp.32

- 44 .

54. Hudson, S. 8., and H. F. Waller , "Evaluation of Construction Control

Procedures - Aggregate Variations and Effects," NCHRP Rpt. No. 69. Hwy.

Res . Bd., Washington, D.C., 1969, 58 p.

55. Miller-Warden Assoc., "Effects of Different Methods of Stockpiling and

Handling Aggregates," NCHRP Rpt. No . 46, Hwy. Res. Bd.,

Washington,D.C ., 1967, 102 p .

56. Joel, R.N., "A Method for Controlling Concrete Workability Using Aggregate

Gradation Control," M.S. Thesis. Univ. of Missouri-Rolla, Rolla, Missouri,

1990, 318 p.

57. AUTOCAD, Autodesk, Inc., Sausalito, CA, 94965.

58. Goode, J .F., and L.A. Lufsey, "A New Graphical Chart for Evaluating

Aggregate Gradations," Proc. of Assn. of Asphalt Paving Tech., Vol. 31,

1962, pp. 176-207.

 168

59. Mix Design Methods for Asphalt Concrete. MS-2. Asphalt Institute,

Lexington, KY, 1984, 102 p.

60. Kandhal, P.S., "Maximum Density Line: Which One Should Be Used?"

Asphalt Technology News. Fall 1989, p. 6.

61. "Responses to 'Maximum Density Line: Which One Should be Used?",

Asphalt Technology News. Vol. 2, No.1, 1990, pp. 2-3.

62. Anderson, D.A., D.R. Luhr, and C.E. Antle, "Framework for Development of

Performance - Related Specifications for Hot-Mix Asphaltic Concrete,"

NCHRP Apt. No. 332. Trans. Res. Bd., Washington, D.C., 1990, 118 p.

63. Abkemeier, T.J., "Indirect Tensile Test Correlative Study," M.S. Thesis.

Univ. of Missouri-Rolla, Rolla, Missouri, 1992, 132 p.

64. Adedimila, A.S., "A Comparison of the Marshall and the Indirect Tensile

Tests in Relation to Asphalt Mixture Design," Proc. Inst. of Civil Engineers,

Part 2, 81, 1986, pp. 461 - 469.

65. SYSTAT, 1990, Systat, Inc., Evanston, Ill.

66. Monismith, C.L. and K.E. Secor, "Viscoelastic Behavior of Asphalt Concrete

Pavements," Proc. lnt'I. Conf. of Structural Design of Asphalt Pavements.

University of Michigan, Ann Arbor, 1962, pp. 476-498.

67. Adedimila, A.S. and T.W. Kennedy, "Fatigue and Resilient Characteristics of

Asphalt Mixtures by Repeated-Load Indirect Tensile Test," Res. Report 183-

~. Center for Hwy. Res. Univ. of Texas at Austin, 1975.

68. Gonzalez, G., T. W. Kennedy, and J.N. Anagonos, "Evaluation of the

 169

Resilient Elastic Characteristics of Asphalt Mixtures Using the Indirect

Tensile Test," Res. Report 183-6, Center for Hwy. Res., Univ. of Texas at

Austin, 1975.

69. Adedimila, A. S. and T.W. Kennedy, "Effects of Repeated Tensile Stresses

on the Resilient Properties of Asphalt Mixtures," Trans. Res. Rec. 616,

Trans. Res . Bd., Washington, D.C., 1976.

70. Kennedy, T.W., "Characterization of Asphalt Pavement Materials Using the

Indirect Tensile Test, "Proc . of Assn. of Asphalt Paving Tech., Vol. 46,

1977, pp. 132-150.

71. Machemehl, R. B. and T. W. Kennedy, "Asphalt Mixtures: Comparative

Analysis of Characterization for Design," Trans . Res. Rec. 821, Trans. Res.

Bd., Washington, D.C. 1981, pp. 22-29.

72. Von Quintus, H.L., J.B. Rauhut, and T.W. Kennedy, "Comparisons of

Asphalt Concrete Stiffness as Measured by Various Testing Techniques,"

Proc. of Assn . of Asphalt Paving Tech., Vol. 51, 1982, pp. 35-52.

73. Khosla, N.P., and M.S. Omer, "Characterization of Asphaltc Mixtures for

Prediction of Pavement Performance," Trans. Res. Rec. 1034, Trans. Res.

Bd., pp. 47-55.

74. Bonaquist, R., D.A. Anderson, and E. Fernando, "Relationship Between

Moduli Measured in the Laboratory by Different Procedures and Field

Deflection Measurements," Proc. of Assn. of Asphalt Paving Tech., Vol. 55,

1986, pp. 419-452.

 170

75. Kallas, B.F., "Elastic and Fatigue Behavior of Emulsified Asphalt Paving

Mixtures," Res. Rot . No. 79-1, 1979, 57 p.

76 . Baladi, G.Y ., "Integrated Material And Structural Method for Flexible

Pavements," Report No. FHWA-RD-88-109, Fed. Hwy. Admin., Vol. 1.,

1988,313p.

77. Maupin, G.W., "Results of Indirect Tensile Tests Related to Asphalt

Fatigue," Hwy. Res. Rec. 404, Hwy. Res. Bd., Washington, D.C., 1972, pp.

1-7.

78. Stroup-Gardiner, M. and J. Epps. "Four Variables that Affect the

Performance of Lime in Asphalt-Aggregate Mixtures, " Trans. Res. Rec.

1115, Trans. Res. Bd., Washington, D.C., 1987, pp. 12-22.

79. Moore, R.K. and T.W. Kennedy, "Tensile Behavior of Asphalt-Treated

Materials Under Repetitive Loading," Proc. Third lnt'I. Conf. of Structural

Design of2 Asphalt Pavements, London, England, Vol. 1, 1972.

(j

80. Schmidt, R.J. and P.E. Graf, "The Effects of Water on the Resilient Modulus

of Asphalt-Treated Mixes," Proc. of Assn. of Asphalt Paving Tech., Vol. 41,

0
1972, pp. 118-162.

81 . Kallas, B.F., "Asphalt Pavement Temperatures," Hwy. Res. Rec. 150, Hwy.

Res. Bd., 1966, pp. 1-11.

82. Croney, D.C. and P. Croney, The Design and Performance of Road

Pavements, McGraw Hill Book Co., 1991, London.

83. Fairhurst, G.E., Y.R. Kim, and N.P. Khosla, "Resilient Modulus Testing of

Asphalt Specimens in Accordance with ASTM D 4123-82," MTS, 7 p.

0
 171

APPENDICES
 APPENDIX A

FINE AGGREGATE

PARTICLE SHAPE DETERMINATION

 173

PARTICLE SHAPE, TEXTURE, AND UNCOMPACTED VOID CONTENT OF

FINE AGGREGATE

NAA METHOD A

APPLICATION

For sand of ( +) #100 sieve through (-) #16 sieve sizes.

EQUIPMENT

1. Funnel

2. Funnel Stand

3. Measure - 1 .52 .± 0.05 in inside dia., 3.37 in inside height

4. Pan - to set funnel stand in

5. Spatula

6. Balance - readable to .± 0 .1 g

CALIBRATION OF MEASURE

Apply a light coat of grease to the top edge of the dry, empty measure. Weigh the

measure, grease, and a flat glass plate slightly larger than the diameter of the

measure. Fill the measure with water at a temperature of 65 to 75° F ( 18 to 24°

C). Place the glass plate on the measure, being sure that no air bubbles remain.

Dry the outer surfaces of the measure and determine the combined mass of

measure, glass plate, grease, and water by weighing.

Calculate the volume of the measure as follows:

 174

V= W
0.998

where:

V = volume of cylinder, cm 3

W = net mass of water, g

PROCEDURE

1. Determine the bulk specific gravity of each material that is to be blended

into the combined gradation (fh9.:., the natural sand, the manufactured sand,

etc.). Each material should be a blend of sizes as per the as-delivered

gradation.

2. Calculate the average specific gravity of the blend of the different types of

sand:

where:

G = average bulk specific gravity of the combined types of sand.

G,,2 = bulk sp. grav. of each type of sand.

P, ,2 = percent of each type of sand in the asphalt mixture gradation

for the ( + )#100 to (-)#16 size. For example, for a total sand

content of 300g between the ( +) #100 and ( +) #16 sizes, if

200g is supplied from sand #1, then P1 = 66.7%.

3. Each type of sand is washed over a No. 100 sieve in accordance with the
 175

methods in ASTM C 117 and then dried and sieved into separate size

fractions using ASTM C 136 procedures.

4. Weigh out and combine the following quantities of dry sand from each of

the sizes. These quantities are combined from the different types of sands

in proportion to their representation in the asphalt mixture for the specific

sieve size (see Data Sheet Calculation):

Individual Size Fraction Combined Mass, g

No. 8 to No. 16 44

No . 16 to No. 30 57

No. 30 to No . 50 72

No. 50 to No. 100 17

190

5. Set the funnel stand in the pan. Place the funnel on the funnel stand.

6. Mix the test sample until it is homogenous. Using a finger to block the

opening of the funnel, pour the test sample into the funnel. Level the

material in the funnel with the spatula. Center the measure under the

funnel, remove the finger, and allow the sample to fall freely into the

measure.

7. After the funnel empties, remove excess heaped sand from the measure by a

single pass of the spatula with the blade vertical using the straight part of its

edge in light contact with the top of the measure. Until this operation is
 176

complete, exercise care to avoid vibration or disturbance that could cause

compaction of the fine aggregate in the measure (Note 3) . Brush adhering

grains from the outside of the measure and determine the mass of the

measure and contents to the nearest 0 . 1 g. Retain all sand grains .

Note 3 -- After strikeoff, the measure may be tapped lightly to compact the

sample to make it easier to transfer the measure to scale or balance without

spilling any of the sample.

8. Collect the sample from the retaining pan and measure, recombine, and

repeat the procedure again. The results of two runs are averaged. See

Calculation section below.

9. Calculate the uncompacted voids to the nearest 0.1 percent for each

determination as follows:

U = V - (M/ G) x 100
V

V = volume of measure, cm 3

M = net mass of fine aggregate in measure (gross mass minus

the mass of the empty measure)

G = bulk dry specific gravity of the combined fine aggregate, step 2

U = uncompacted voids, percent, in the material

Note 4 -- For most aggregate sources the fine aggregate specific gravity does not

vary much from sample to sample or from size to size finer than the 2.36-mm (No .

8) sieve. Therefore, unless the specific gravity of individual sizes is appreciably

 177

different, it is intended that the value used in th is calculation may be from a

routine specific gravity test of an as-rece ived grading of the fine aggregate. If

significant variation between different samples is expected, the specific gravity

should be determined on material from the same field sample from which the

uncompacted void content sample was derived, Normally the as-received grading

can be tested for specific gravity, particularly if the 2.36-mm (No. 8) to 150-um

(No . 100) size fraction represents more than 50 percent of the as-received grading.

However , it may be necessary to test the graded 2.36-um (No. 8) to 150-mm (No.

100) sizes for specific gravity for use with the graded void sample, Method A or

the individual size fractions for use w ith the individual size method, Method 8, (not

included here). A difference in specific gravity of the 0.05 will change the

calculated void content about one percent.

 178

DATA SHEET

Sample Date Tested Technician

Total percent of each sand per asphalt mixture puck:

Sieve Sand #1 Sand #2 Total Wt . Fraction of Total, Fraction of

Wt . Ret. g Wt. Ret. g Ret. g Sand #1, (F 1 ) Total, Sand
#2 (F?)
#16

30

50

100

Total: A = B = C =

Total wt. Ret., sand #1 (A)

pl = -AC =
Total Wt. Ret., both sands ( C)
* 100

Total wt. Ret., sand #2 (B)

P2 = -BC =
Total Wt. Ret., both sands (C)
* 100

wt. Ret . on #16, sand 1

F1,16 =
Wt. Ret. on #16, both

=
wt. Ret. on #16, sand 2
F2 , 16
Wt. Ret. on #16, both

=
Wt. Ret. on #30, sand 1
F1,30
Wt. Ret. on #30, both

=
Wt. Ret. on #30, sand 2
F2,30
Wt. Ret. on #30, both
 179

We. Rec. on #50, sand 1

Fi. 5 0 =
Wt. Rec . on #50, both

F2. 5 0 = Wt. Ret. on #50, sand 2

Wt. Rec. on #50, both

F1.1 00 = Wt. Ret. on #100 I sand 1

Wt. Ret. on #100, both

=
Wt . Ret. on #100, sand 2
F2,1 00
Wt. Ret . on #100, both
 180

DATA SHEET
PARTICLE SHAPE AND TEXTURE OF FINE AGGREGATE

Sample _ _ _ __
Date Tested _ _ __
Technician

Specific Gravity

G, 2 P, 2

Sand #1

Sand #2

G = (P 1 .,. P2) / [-P1

Gl
+ -P2
G2
l
r = 100

STANDARD GRADATION

Size Fraction

#16 to #8 Mass sand #1 F 1 • 44 g

Mass sand #2 = F2 • 44 g

#30 to #16 Mass sand #1 = F1 • 57 g

Mass sand #2 F2 • 57 g

#50 to #30 Mass sand #1 F1 • 72 g

Mass sand #2 = F2 • 72 g

#100 to #50 Mass sand #1 = F1 • 17 g

Mass sand #2 F2 • 17 g

TOTAL 190 g

UNCOMPACTED VOIDS

Trial 1 Trial 2

Weight sand + measure, g

Weight measure, g

Weight sand (Ml. g

Combined specific gravity (G)

Uncompacted voids (U), %

U = [V-(M/G)/VJ • 100

Average Uncompacted Voids

 181

APPENDIX 8

COARSE AGGREGATE

PARTICLE SHAPE DETERMINATION

 182

INDEX OF AGGREGATE PARTICLE SHAPE AND TEXTURE

ASTM D 3398-87

APPLICATION

For asphalt mixture aggregate from ( +) #4 sieve to the (-) 1 in sieve size fractions.

EQUIPMENT

1. 6 in dia. Mold - inside height 7.00_±_ 0.01 in for testing the following
fractions:

( +) 3/4" to (-) 1 in
(+) 1 / 2" to(-) 3/4 in

2. 4 in dia . Mold - inside height 4. 6 .± 0 .01 in for testing the following

fractions:

(+) 3 / 8" to(-) 1/2 in

(+) #4 to(-) 3 /8 in

3. 24 in Tamping Rod - for use with 6 in mold

4. 16 in Tamping Rod - for use with 4 in mold

CALIBRATION OF MOLD

1. Fill the mold with water at room temperature and cover with a piece of plate
glass in such a way as to eliminate bubbles and excess water.

2. Determine the mass of water in the mold to an accuracy of 4 g or less.

3. Measure the temperature of the water and determine the volume of the mold
by multiplying the mass of the water by the corresponding specific volume
of water given in Table 8-1 for the temperature involved.

PROCEDURE

1. Test each of the above-listed fractions if present in the gradation in amounts

of 10% or more.
 183

2. Obtain a sample of each fraction to be tested as per:

6 in mold 13 lbs
4 in mold 4 lbs

3. Wash the sample of aggregate by decanting the wash water through a sieve
at least one size smaller than that being used. Continue the washing and
decanting operation until the wash water is clear. Then flush the residue on
the sieve back into the aggregate sample. Dry the sample to constant
weight at a temperature of 230 .±. 9°F ( 110 .±. 5°C).

4. Determine the bulk-dry specific gravity of each size fraction.

5. Place the cylindrical mold on a uniform, solid foundation. Gently place the
aggregate, from the lowest height possible , into the mold until it is
approximately one-third full. Level the surface with the fingers, and
compact the layer using 10 drops of the tamping rod evenly distributed over
the surface. Apply each drop by holding the rod vertically with its rounded
end 2 in (50 mm) above the surface of the aggregate (controlled by the slot-
and-pin arrangement) and releasing it so that it falls freely. Place a second
layer in the mold using the same procedure, filling the mold approximately
two-thirds full . As before, level the surface and apply the same compactive
effort , 10 drops of the rod. After the final layer has been compacted, add
individual pieces of aggregate to make the surface of the aggregate mass
even with the rim of the mold, with no projections above the rim. Determine
the mass of the aggregate in the mold to an accuracy of at least 4 g.

6. Repeat the filling of the mold using the same specimen and compaction.
Make a second determination of the mass of the aggregate in the mold as
described above . Use the average mass of the two runs in calculating the
percentage of voids at 10 drops for each size .

7. For the higher degree of compaction, follow the steps outlined in #5 and #6,
except use 50 drops of the tamping rod in compacting each layer. Again
average the masses from the two runs for use in computing the percentage
of voids at 50 drops for each size fraction.

8. Calculate the percentage of voids in each size fraction of the aggregate at

10 drops per layer and at 50 drops per layer, respectively, by the following
relationships:

V 10 = [1-(M 10 /SV)] x 100

V50 = (1-(M 50 /SV)] x 100
 184

where :

V 10 = voids in aggregate compacted at 10 drops per layer, %,

V50 = voids in aggregate compacted at 50 drops per layer, %,
M 10 = average mass of the aggregate in the mold compacted at
10 drops per layer, g
M 50 = average mass of the aggregate in the mold compacted at
50 drops per layer, g

S = bulk-dry specific gravity of the aggregate size

fraction
V = volume of the cylindrical mo ld , ml

9. Determine the particle index (lal for each size fraction tested as follows :

la = 1.25 V 10 - 0.25 V 50 - 32.0

10. Calculate the weighted particle index of an aggregate containing several

sizes by averaging the particle index data for each size fraction, weighted on
the basis of the percentage of the fractions in the grading of the sample:

where P 1 .. . Pn = percent of the fraction in the sample (not the entire

asphalt mixture gradation)
1a 1 ... Ian = particle index of each fraction

For sizes represented by less than 10% in the grading, for which no particle index
data were obta ined, use the average particle index of the next coarser and finer
sizes for which data are available or the particle index for the next coarser or finer
size if a value is available only in one direction .
 185

Table 8-1 Specific Volume of Water at Different Temperatures

Temperature, ° F( °C) Specific Volume, ml/g

54 (12) 1.0005
57 (14) 1.0007
61 (16) 1.0010
64 (18) 1.0014
68 (20) 1 .0018

72 (22) 1.0022
75 (24) 1.0027
79 (26) 1.0032
82 (28) 1.0038

86 (30) 1.0044

90 (32) 1.0050
 186

DATA SHEET
Index of Particle Shape and Texture (03398)

Sample _ _ Technician _ _ _ _ __ Date

Percent of each fraction in each puck:

Wt:. Ret. % Ret. Spec. Grav.

Sieve g (Pl,2,3) s
I
1/2" to 3 / 4"

3/8" to 1/2"
I
#4 to 3 / 8"
I I
Total 100

= wt. rec. on a given sieve x 100

wt. rec., total

PARTICLE INDEX

Fraction 10 Drops Trial 1 Trial 2

1/2"to3/4" Mass of aggregate + mold, g

Mass of mold, g

Mass of aggregate, g

Average of 2 trials ( M 10 )g

50 Drops Trial 1 Trial 2

Mass of aggregate + mold , g

Moss of mold , g

Mass of aggregate , g

Average of 2 trials (M 50 ), g

Volume of 6 in cylinder mold (V), ml

Bulk specific gravity (S)

Mass of aggregate (M 10 ), g

Mass of aggregate (M 50 ), g

v,o = 11 · (M 10 /SV)J • 100, %

Vso = [1 · (M 50 /SV)J • 100, %

lal = 1 . 2 5 V 1 0 - 0 . 2 5 V 50 - 3 2. 0
 187

PARTICLE INDEX

Fraction 10 Drops Trial 1 Trial 2

3 / 8" to 1 / 2" Mass of aggregate + mold, g
Mass of mold, g

Mass of aggregate, g
Average of 2 trials (M 10 )g

50 Drops Trial 1 Trial 2

Mass of aggregate + mold, g
Moss of mold, g
Mass of aggregate, g

Average of 2 trials (M 50 ), g

Volume of 6 in cylinder mold (V), ml

Bulk specific gravity (S)

Mass of aggregate (M 10 ), g

Mass of aggregate (M 50 ), g

v,o = [1 - (M 10 /SV)J * 100, %

V50 = [1 - (M 50 /SV)J * 100, %

1a 2 = 1.25 V 10 - 0.25 V 50 - 32.0

 188

PARTICLE INDEX

Fraction 10 Drops Trial 1 Trial 2

#4 to 3 / 8" Mass of aggregate + mold, g

Mass of mold , g

Mass of aggregate, g
Average of 2 trials (M 10 )g

50 Drops Trial 1 Trial 2

Mass of aggregate + mold, g

Moss of mold, g

Mass of aggregate, g
Average of 2 trials (M 50 ), g

Volume of 6 in cylinder mold (V), ml

Bulk specific gravity (S)

Mass of aggregate (M 10 ), g

Mass of aggregate (M 50 ), g

v,o = [1 - (M 10 /SV)J * 100, %

V50 = [1 - (M 50 / SV)J * 100, %

1a3 = 1.25 V 10 - 0.25 V 50 - 32 .0

PARTICLE INDEX FOR ALL FRACTIONS :

IT = (P, * la1) + (P2 * la2) + (P3 * la3)

 189

APPENDIX C

DETERMINATION OF

INDIRECT TENSILE STRENGTH AND RESILIENT MODULUS

FOR

BITUMINOUS MATERIALS
 190

DETERMINATION OF INDIRECT TENSILE STRENGTH AND RESILIENT MODULUS

FOR BITUMINOUS MIXTURES

SHRP Designation: P07

ASTM Designation: D-4123

1. INTRODUCTION

1.1 Three levels of temperature (41 ± 2 °F, 77 ± 2 °F and 104 ± 2 °F)

were tested using repetitive compressive haversine load pulses of 1 Hz frequency with

a 0 . 1 second load duration and a 0.9 second rest period. The thickness of the

specimens tested varied between 2.4 in and 2. 7 in. Prior to performing the resilient

modulus test, the indirect tensile strength was determined at 77 ± 2 °F for each

combination of mix type (bituminous base, Type C and Type IC), gradation (fine and

coarse), asphalt cement grade (AC-10 and AC-20), asphalt cement content (optimum

and 0. 5 % above optimum) and aggregate (DR-6 and DR- 7 for bituminous base mixes,

DR-4 and DR-5 for Type C and IC mixes) . Duplicate specimens were prepared for the

48 mixes . In all, 96 indirect tensile strength specimens were prepared. The

specimens were aged for one week before testing . The value of tensile strength

determined by this procedure was used to estimate the indirect tensile stress and

corresponding compressive load to be repetitively applied to similar test specimens

during subsequent resilient modulus testing. For resilient modulus testing, four

specimens were made for each mix. In all, 192 specimens were tested. The

specimens were aged for one week before testing.

1.2 Typically, the specimens were built on a Tuesday and then testing would

begin for those specimens on the following Tuesday at 41 °F, continue at 77 °Fon

Wednesday, and finish at 104 °Fon Thursday. The testing was done from lower to

higher temperature levels to reduce permanent damage to the specimen. The

specimens were brought to the test temperature 24 hours before testing began.
 191

1 .3 Once the test was completed, plots were made of load, horizontal

deformation and vertical deformation with time. The measured total recoverable

horizontal and vertical deformations were used to calculate values of resilient

Poisson's rat io . The resilient modulus values were then calculated using cyclic loads

and the calculated Poisson's ratio.

2. PREPARATION OF TEST SPECIMENS

2.1 Both the indirect tensile strength and the resilient modulus specimens are

made using the same procedure as that described in the Marshall Mix Design portion

of this report. In addition, all other specimen properties except flow and stability are

determined as described in the Marshall Mix Design section.

2.2 Specimens are marked and measured as follows after determination of the

specimen SSD BSG:

2.2.1 For indirect tensile strength specimens use a paint marker and a

straight edge to draw a thin line along the diameter of the specimen face with the

smoothest texture. Use a 4 in filter paper with notches on opposite sides of its

diameter to mark the two points needed to draw the line. Avoid drawing the line

along a diameter that has voids on the sides of the specimen .

2.2.2 For resilient modulus specimens, mark a diametrical axis on the test

specimen as specified in 2.2.1 . Add an arrow tip to one end of the line. Using the

paint marker and a straight edge, draw a thin horizontal line at the mid-thickness of

the specimen on both sides of the diametrical axis line.

2.2 .3 Measure the thickness and the diameter of indirect tensile strength

and resilient modulus test specimens to the nearest 0.01 in using dial calipers.

Determine the thickness by averaging a single center measurement with three equally

spaced measurements located 0.5 in from the test specimen edge. Determine the

diameter by averaging the diameter of the specimen at mid height along ( 1) the
 192

diametrical axis drawn in 2 .2 . 1 and 2.2.2 above and the axis perpendicular to the axis

measured in (1) above .

3. STATIC INDIRECT TENSILE STRENGTH TESTS

3 . 1 Adjust the thermostat in the testing room to 77 °F .

3 .2 Place the specimens to be tested in watertight buckets and then put the

buckets in a constant temperature water bath at 77 ± 2 °F for 24 hours.

3.3 Prepare the xy plotter , LVDT signal conditioner, and the strain gage signal

conditioner as specified in the Marshall Mix Design procedure, section 5 .8.3.

3.4 Place the test specimen in the test press loading apparatus and position

it so that the mid thickness of the test specimen is located in the line of action of the

load cell. Use the diametrical marking to ensure that the specimen is aligned from top

to bottom-the diametrical marking should be centered on the top and bottom loading

strips .

3 .5 Turn on the test press, check the zero point on the xy plotter, make sure

the plotter pen is touching the paper, and depress the UP button on the test press to

apply a compressive load to the specimen by maintaining a constant rate of movement

of the test press platen of 2 in/min. The platen will continue to move until the

maximum load is reached and the load decreases as indicated by the strain gage

indicator. The upper limit switch on the test press forces the platen to stop moving.

The upper and lower limit switches on the test press can be adjusted using the button

underneath the test press. Record the maximum load reading from the strain gage

conditioner on the chart paper. Note: if the beam holding the plotter pen binds during

the test, quickly turn off the power to the plotter and straighten the beam so it can

move freely, then turn on the plotter .

3 .6 Depress the DOWN button on the test press to lower the platen. The

lower limit switch on the test press will stop the platen. Remove the breaking head
 193

from the platen, sweep any loose asphalt off of the platen, and save the test

specimen by placing it in a plastic bag or some other suitable container. Write the

specimen ID on the bag. Remove the recording chart from the xy plotter and fill in the

chart data block. Clean the breaking head loading strips and lubricate the guide rods

with silicone spray and a towel.

3 . 7 Make a sketch of the specimen failure on the data sheet and calculate the

indirect tensile strength as follows:

where:

St is the indirect tensile strength in lbs

PO is the maximum load sustained by the specimen in lbs

h is the specimen thickness in inches

D is the specimen dia·meter in inches

3.8 Prepare a table of average indirect tensile strength values for each mix

along with tensile stress levels of 30, 15, and 5 percent of the tensile strength to be

used in conducting the resilient modulus determinations at the test temperatures of

41, 77, and 104 ° F, respectively. Include specimen contact loads shown in Table 21

of 3, 1 .5, and 0.5 percent of the tensile strength to be used during resilient modulus

determinations at 41, 77, and 104 °F respectively. Contact loads and cyclic test

loads will be discussed below.

4. RESILIENT MODULUS TESTING

The system includes an MTS 810 closed loop electrohydraulic load system

equipped with a 3.3K actuator, an environmental chamber, a temperature bath, a

specimen yoke, a yoke stand, and a data acquisition and load signal generation

computer/software system. The specimens were seated on the lower concave loading
 194

strip which was affixed to the loading frame. Both upper and lower loading strips had

a radius of curvature of 2in (5.1cm) and a width of 0.5in (1.3cm) .

The specimen yoke was a small frame that held two horizontal L VDT'S which

were diametrally opposite. Originally, the LVDT's were fixed to a base. However,

specimens tended to rock under load; this could be discerned by both visual

observation and by examining the readout from each horizontal channel. Ideally, both

LVDT's should be compressed upon loading; however, sometimes one of the readouts

indicated that one LVDT was extending. It was thought that possibly one of the tips

of the L VDT could be slipping down into a surface void. So, the tips were replaced

with disk-type tips that threaded onto the LVDT cores. The idea was that the 1 /4 in

(0.64cm) diameter disks would bridge across surface voids. Next, the method of

transferring the load was changed . Originally, the upper load strip was not fixed; a

ball bearing was used to transfer the load from the load cell, which was fixed to the

MTS load ram, to the top of the upper load strip. This configuration was basically the

state of the art several years ago. Additionally, vertical deformation was measured

with an L VDT internal to the MTS system; the L VDT had a full range of 6in (2.4cm).

Data indicated that the specimen was unstable, and that the vertical L VDT was not

sensitive enough, even though the controls of the MTS were set to utilize 10 percent

of the full range (0.6in = 0.2cm) for greater accuracy .

The second generation device consisted of mounting a more sensitive vertical

L VDT to the yoke base, and fixing the upper loading strip to the load cell in order to

prevent it from rotating. The vertical measurements appeared to be more sensible,

but there still seemed to be specimen movement, even under higher baseline static

loads. The yoke base still had to be centered manually to align the upper and lower

load strips .

In an effort to assure better vertical alignment of the upper and lower load
 195

strips, a new base was purchased which was manufactured for the purpose of

performing repeated load and static load indirect tension tests. This device featured

an upper platen which slid vertically by way of two guide posts, thus aligning the

upper and lower load strips . The horizontal LVDT's were fixed to the base as well as

the vertical LVTD. However , some rocking was still evident, possibly due to slight

looseness in the guide bearings and to the difficulty of aligning the upper platen with

the load cell/ram .

A fourth generation set-up involved a different horizontal L VDT yoke. In this

case, a light gauge steel frame holding the two LVDTs was mounted directly onto the

specimen, reminiscent of the original Schmidt device (80) and the newer MTS resilient

modulus system (83). This allows the LVDTs to follow the specimen should rocking

occur . This, coupled with the higher static background loads recommended by the

latest draft of the SHRP protocol, and coupled with the use of guide rods, seems to

have eliminated or reduced experimental problems to acceptable levels. Also, as per

the SHRP protocol , two vertical LVDT's were mounted to monitor the movement of

the upper platen, rather than the ram movement . This virtually eliminates machine

deflection from the vertical deformation measurement, and allows for a more accurate

calculation of Poisson ' s ratio, which is necessary for a truer calculation of resilient

modulus. The f inal setup is shown in Fig. D1. An auxiliary device was necessary in

order to mount the yoke accurately and consistently on the sample, as shown in Fig.

02.

The environmental chamber had to be capable of maintaining temperatures of

41 °, 77°, and 104°F (4 , 25, and 40°C) plus or minus 2°F(1 °C) . A 1 .5 in (3.8cm)

thick styrofoam box was built to surround the yoke . Also surrounding the yoke on

three sides were 0 .25in (0 .64cm) copper coils through which was circulated a mixture

of ethylene glycol and water . The solution was circulated through a

 196

Fig. DL Final Resilient Modulus Device.

Fig. D2. Resilient Modulus Yoke Mounting Template.

 197

heat ing / re fr ige rat io n unit . A small fan w as mounted ins ide the env ironmental chamber

to d ist ribut e th e air . A dial gage thermomete r was mounted through the box to

monitor air tempe rature. The chamber was found to be able to maintain the

temperature within the ASTM 4123 specified 2°F ( 1 °C) . The env ironmental chamber

is shown in Fig . 03 . Prior to testing, specimens were stored in an environment

controlled at the test temperature: refrigerator at 41 °F (4°C) . at room ambient

condit ions at 77°F (25°C), or in an oven at 104°F (40°C) .

The data acquisition / load signal generation system consisted of the following

components: an IBM-compatible 386 personal computer with 8 kilobyte memory, 80

megabyte hard d isk , Data Translat ion OT 2801 A 12 bit analog /digital board , printer,

color mon itor, one strain gage conditioner, four L VDT conditioners , ASYST software,

a 2500 lb. Stra in-sert load cell , two Schaevitz LBB-375-TA-100 LVDTs with a range

of .±.0. 1 OOin (0 . 25cm) of travel (horizontal deformation), and two Schaevitz PCA-220-

100 LVDTs with a range of .±.0. 100 in (0.25cm) of travel (vertical deformation).

The m inimum value measurable with the system setup is calcu lated as fol lows .

The input voltage range of the horizonta l LVDT's was -10 to + 10 V. The precision

of the input signal measured by the 12 bit A/D board is 20V/2 12 which is 0.00488

V. When 0.1 in full scale displacement LVDT's were used, the horizontal transducers

were ranged 10% of full scale which calibrates the output to a finer scale travel and

allows higher resolut ion measurements. The vertical transducers were ranged 33%

of full scale . The min imum value of horizontal displacement measured was then (0.01

in/1 OV) * (0 .00488V) which is 0.00000488 in. The vertical LVDT's could be read to

(20/2 12 )*(0.033/10) =0.000016 in. The load cell could be read to (2500

lb/10V)*(0.00488) = 1 . 22 lbs .

4 . 1 Specimen preparation . Place the specimens in the appropriate controlled

temperature env ironment needed to bring them to the specified test temperature at
 198

Fig. D3. Resilient Modulus Environmental Chamber.

 199

least 24 hour s pr ior to testing . A water bath is used to cool the spec imens to 41 °F.

The spec imen s are pla ced in 2 gal po lyethylene buckets before being put in the water

bath. Eac h bucket can hold 6 specimens. Use the hole in top of each bucket to

check the specimen temperature with a dial gage thermometer . Allow the specimens

to sit in the ambient air to come to 77 ° (the ambient conditions can be controlled by

opening the door of the testing room or by turning the room air conditioner on and

off), and use an oven to bring the specimens to 104 °F . Keep a dial gage or mercury-

in-glass thermometer with the specimens in the oven at all times.

4.2 Temperature cabinet. A constant temperature circulator is used to

maintain the testing temperatures of 41 °F and 104 °F during resilient modulus tests .

A 50-50 mixture of antifreeze and water is used as the circulating liquid . Check the

level of coolant in the circulator reservoir before turning the circulator on. The coolant

level should be above the coils in the reservoir at all times. Set the thermoregulator

to -13 °C to bring the chamber to 41 °F and 45 °C to bring the chamber to 104 °F

(read to the top of the thermoregulator indicator). Check the temperature of the

cabinet us ing a dial gage thermometer and the thermometer access port on the top

left of the cabinet. Both the circulator heating and cooling switches should be "on"

to reach 41 °F and the "heating only" switch should be "on" to reach 104 °F . Move

the control valve lever forward on the right side of the circulator to circulate coolant

through the temperature cabinet coils. At the 77 °F test temperature, maintain 77

° F as specified in 4 . 1 . Place the loading head with the vertical L VDTs and the

horizontal frame with the horizontal LVDTs in the temperature cabinet 24 hours prior

to testing.

4 .3 Equipment calibration. Before each test, make sure that all LVDTs and the

load cell are securely mounted for testing and that there are no loose connections with

their respective signal conditioner and the computer .

 200

4 .3.1 Calibrate the LVDTs before any testing begins, and monitor their

calibration after every 48 resilient modulus tests using the micrometer calibration

block . Monitor the calibration of the LVDTs at each test temperature since there is

a slight variation of L VDT response with temperature. Prior to each specimen test,

make sure that the shaft of each L VDT is not sticking by depressing and releasing the

LVDT tip. Apply silicone spray or WD-40 on a regular basis to the LVDT shafts. Also

check the L VDT tips for tightness.

4.3.2 Calibrate two load cells before performing resilient modulus tests.

One load cell is used as the MTS load cell during testing and the other is used to

check the calibration of the testing load cell every 48 resilient modulus tests at 77 °F

only.

4.4 Prepare testing log. For each combination of mix, gradation, asphalt

cement type, asphalt cement percentage and aggregate, a testing log is prepared

before testing. The log uses the maximum suggested seating load (P contact) and the

recommended maximum load for testing and preconditioning (P maxl as specified in

3.8, along with load cell calibration data to determine the necessary cyclic load

(P cyclic), MTS set point and haversine load pulse for testing the 4 specimens

represented by each log at 41, 77, and 104 ° F.

4 .4.1 The cyclic load can be determined using:

-- - - -------
p cyclic=Pmax - p concacc

Note: round P rnax and P contact down to the nearest 10 lbs and 1 lb respectively before

calculating P cyclic to avoid overloading the specimen. Round Pcyclic to the nearest 1

lb .

4.4. 2 The log also has space for recording the cumulative vertical

deformation and the number of preconditioning and load cycles used for each
 201

specimen test. The cumulative deformat ion is measured using the dia l gage

magnet ical ly mounted to the hydraulic actuator shaft . Subtract the dial gage reading

in 4.7 from the read in g in 4.8 to get the cumulat ive vertical deformation. If total

cumulat ive vertical deformations greater than 0.025 inch for 41 °F or 0.050 in for

77° and 104 °F occur, reduce the applied load to the minimum value possible and still

retain an adequate deformation for measurement purposes.

4 .5 Mounting specimen in MTS load frame .

4 .5.1 Take the loading head assembly from the temperature cabinet .

Place the lucite specimen on the lower section loading strip. Place the upper section

of the loading head on top of the specimen by sliding it over the guide rods. Rotate

the lucite specimen so that its d iametrical marking is coincident with the upper and

lower loading strips. Place a weight on top of the upper section and make sure that

the tips of the vertical LVDTs are touching the heads of the bolts clamped to the

upper section . Use the hose clamps to adjust the position of the LVDTs until their

respective L VDT conditioner reads about + 0 .4500 for the front LVDT and -0.4500

for the rear LVDT . Make sure that the engraved 14's on the upper and lower section

of the loading head are on the same side. After the vertical LVDTs are adjusted,

remove the lucite specimen and return the loading head assembly to the temperature

cabinet .

-----
4. 5 . 2 Get a specimen for testing from the appropriate controlled

temperature environment (make sure that the specimen and the MTS temperature

cabinet are within the specified temperature limits for testing).

4. 5 .3 Take the horizontal L VDT frame from the temperature cabinet and

place it on top of the alignment loading strip. Place the specimen in the horizontal

L VDT frame so that the diametrical axis of the specimen is perpendicular to the plane

of the LVDTs . The arrow of the diametrical axis should be pointing down . Use the
 202

th in piece o f ru bbe r to pro t ec t the unma rked face of the spec imen from the po inted

piece of meta l in the frame .

4.5 .3 .1 Use an alien head screw to secure the specimen in the

frame . The screw should make contact at the center po int of the specimen along the

marked d iametrical axis . Do not allow the screw to penetrate more than 1 / 16 in into

the face of the specimen . The pointed tip screw works well on specimens with

smooth faces , and the flat t ip screw works well on spec imens w ith rough faces . It is

helpful to hold the specimen and the frame with one hand while turning the screw

with the other hand . Make sure that there is no gap between the frame and the

alignment load ing strip sides.

4 .5 .3 .2 Use a 1/ 2 in wrench to adjust the horizontal LVDTs so

that the ir respect ive LVDT condit ioner is read ing about + 0.2500 . Make sure that the

LVDT ti ps are not to uch ing any vo ids on the sides of the spec imen.

4 . 5 .4 Place the horizontal L VDT frame and specimen on the bottom

section of the load ing head inside the temperature cabinet. Move the frame and

specimen by grabbing hold of the top of the specimen . Rotate the specimen sideways

to clear the loading head guide posts when placing it in the cabinet . Place the upper

section of the loading head making sure that the engraved 14' s line up as spec ified

in 4. 5. 1. Place the steel contact ball in the center of the loading head upper sect ion .

Keep the L VDT w ires along the edges of the cabinet .

4 .5 .5 Position the specimen so that the diametrical ax is marking is

coincident with the upper and lower loading strips, and the mid-thickness marks on

the specimen are located in the line of act ion of the actuator shaft . Use the peep hole

on the left side of the cabinet to center the specimen on the upper and lower loading

strips. Use the ruler to ensure that the horizontal LVDTs are at the same height above

the base of the load ing head.

 203

4. 5 . 6 Remove the dial gage from the hydraulic shaft . Use the setpoint

knob (turn the knob clockwise for load ing) on the MTS 442 contro ll er panel to bring

down the load frame loading ram . Overshoot the set point specified in 4. 5 .2 to get

the ram moving, but be careful not to get the ram moving too fast, otherwise the

spec imen may be damaged if too large a load is applied. When the load cell on the

end of the loading ram is near the steel ball on top of the loading head, adjust the

setpoint to the exact value specified in 4 .4 . 1. The steel ball should seat uniformly in

both the loading head and the depression in the center of the load cell. Return the

peep hole cover and the front cover of the cabinet to bring the cabinet back to the

test temperature . No more than five minutes should elapse between removal of the

specimen from its controlled temperature environment and application of the contact

load on the spec imen in the loading frame . Return the dial gage to its position on the

hydraul ic shaft .

4 . 6 Configurin g data acquis iti on and control software. Turn on the computer ,

monitor and printer and wait for the DOS prompt C: >.

4 . 6 . 1 Enter haversine load pulse.

4. 6 . 2 Enter preconditioning cycles.

4. 7 Specimen preconditioning. Make a note of the hydraulic shaft dial gage

reading . Precondition the specimen along the diametrical axis prior to testing by

apply in g the repeated have rsine-shaped load pulse of 0 . 1 sec with a 0.9 sec rest

period for the specified number of cycles in 4.6.2. After each sequence of

preconditioning cycles, view the channel plots by using the Enter key to go to the

next plot . While inspecting the vertical and horizontal response curves, calculate the

vertical and horizontal deformation ratios to ensure that the ratios are less than or

equal to 1 . 5 . Calculate the ratios as Rv = Vmax / Vmin and Rh= Hmax/Hmin where Rv

is the vertical ratio and Rh the horizontal rat io. Vmax is the maximum total
 20 4

deformat io n o f t he t wo vert ical L VDT curves and Vmin is the minimum total

deformat ion of the two curves. Likewise Hmax is the maximum total deformation of

the two horizontal L VDT curves and Hm in is the min imum deformation of the two

curves . The deformat ions are measured by taking the difference in height of the peak

and basel ine of the largest peak on each plot . When inspecting the plots, check that

successive deformation readings agree within 10 percent and that a line drawn

through the peaks on each plot would be slop ing upward for all plots or sloping

downward for all plots . Preconditioning can be stopped when the minimum number

of cycles specified in 4.4 .2 and the criteria outlined above are met . If the criteria

above are not met , make adjustments to the seating of the specimen in the loading

head without removing the contact load . In no case should the maximum number of

precondit ion ing cycles be exceeded . It the cr iteria above cannot be met within the

max imum number of precondi ti on ing cycles, remove the contact load from the

specimen , enter bye to get back to the DOS prompt on the computer and repeat steps

4 . 5.5 through 4 . 7 above .

4 .8 Specimen testing . After preconditioning the test specimen, load the

resilient modulus test ing program . "Resmod .run" will apply 37 load pulses to the

specimen as expla ined in 4. 7 . The last 7 load pulses and the resultant measured

deformations can be viewed in the channel plots after the 37 load cycles have

completed . To determ ine when load ing is completed , use a stopwatch to time the 37

load cycles (each cycle is 1 second). The criteria in 4.7 are used to accept or reject

the test except that the range in deformation values of five successive horizontal

deformation values mus t be less than 10% of the average of the five deformation

values. If the criteria are not met , remove the contact load from the specimen, enter

bye to get back to the DOS prompt on the computer and repeat steps 4.5.5 through

4. 7 above . Make a note of the hydraulic shaft dial gage reading and record the
 205

cumulat ive v ert ica l deformation as spec if ied in 4 .4 .2 .

4.9 Print channel plots .

 206

APPENDIX D

DETERMINATION OF ASPHALT TYPE

,,
 207

APPENDIX D : DETERMINATION OF ASPHALT TYPE

To determ ine the type (S,B,or W) of asphalt, the Heukelom Bitumen Test Data

Chart (BTDC) is used (Fig . 3). The vertical axis is actually in consistency units "C".

At viscosity = 1 poise , c = 0 , and where penetration = 1, c = 1000. In the viscosity

range, c is:

c = 1310 log vi scosity (poise ) (84)

4.35 + log viscosity(poise)

The criteria for classificat ion as an S-type asphalt are two-fold :

1) b. T = 1Tl3 , 000p - TR&BI!, 8 ° C • • (85)

2) dBTS = IBTSvisc - BTSpen/R&BI !> 1, 8 • • • • ( 86)

where BTS (Bitumen Temperature Susceptibility) is a measure of the temperature

sensitivity of the asphalt and is calculated in the temperature range of penetration

applicabil ity as :

BTSpen /R&B = Plpen /R&B

and is calculated in the temperature range of viscosity applicability as:

BTSvisc = (87)

'T'
-u. ooop (88)

1310 log 13,000

c1J.ooo = (89}
4.35 + log 13,000
=636.73

where:

T1 = 60°C (test temperature for absolute viscosity )

 208

T2 = 135°C (test temperature for kinematic viscos it y)

To convert kinematic to absolute v iscos ity n:

n,poise = (kin ematicvisc.,cs)(sp. grav, 25°)(0.934)

(90)
100

Table D-1 shows the results of the analysis.

As can be seen, both criteria are m~t in regard to classification as "S" type

asphalts . Additionally , Heukelom defines "S" -type asphalts as those that plot in

approximately a straight line on the BTDC paper. Both asphalts used in this study

exhibit this type of plot .

Table D-1 . Bitumen Type Classification .

Asphalt T(°C) Kin. Abs . Vise (pl Sp . TR&B Plpen/R&B ~T ,~BTSI

V ise Grav .

1 2 ( Csl 1 2 (25°CJ (OCJ ( oCJ

AC-10 60 135 301 1099 2.83 1.007 44.5 -0.92 2.1 0.24

AC-20 60 135 361 1911 3.43 1.017 48 .5 -1.88 1.6 1.14

 209

APPENDIX E

SPECIFIC GRAVITY AND VOIDS DATA

 210

' %Air %Air VMA VMA VFAC VFAC

Puck I.D.
Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

BB-6C-10-3.6-l 3.7 4.0 11. 9 9.8 71. 9 59.0

BB-6C-10-3.6-2 2.8 3.1 11.1 9.0 78.1 65.3
BB-6C-10-4.l-l 2.4 2.7 11. 8 9.7 83.0 72 .2
BB-6C-10-4.l-2 2.3 2.6 11. 7 9.6 83.9 73.1
BB-6C-20-3.6-l 4.2 4.1 12.3 10.2 68.9 60.1
BB-6C-20-3.6-2 3. 1 3.0 11. 3 9.2 75.7 67.4
BB-6C-20-4.l-l 2.5 2.4 11. 8 9.7 82.1 75.6
BB-6C-20-4.l-2 2.2 2.0 11. 5 9.4 84.5 78.4
BB-6F-10-4.l-l 2.9 4.2 12.5 11. l 77.2 62.2
BB-6F-10-4.l-2 3 .1 4.4 12.7 11. 3 75.6 60.7
BB-6F-10-4.6-l 1. 7 3.0 12.5 11.1 86.6 72. 8
BB-6F-10-4.6-2 2.9 4.2 13. 6 12.2 78.7 65.3
BB-6F-20-4.l-l 3.9 5 .1 13.4 12.0 70.8 57.1
BB-6F-20-4.l-2 3.7 4.9 13 .2 11. 7 72 .2 58.4
BB-6F-20-4.6-l 2.3 3.5 13.0 11. 5 82.3 69.5
BB-6F-20-4.6-2 1. 9 3. 1 12.6 11. 2 85.0 72.1
BB-7C-10-3.6-l 2 .6 3.7 11. 0 9.6 76.2 61. 6
B8-7C-10-3.6-2 2.3 3.4 10.8 9.3 78.3 63.5
BB-7C-10-4.l-l I 1.1 2.2 10.7 9.3 89.8 76.7
BB-7C-10-4.l-2 1. 4 2. 5 11. 0 9.5 87.4 74.3
BB-7C-2 0 -3.6-l 2.4 3.2 lO. 7 9.3 77.9 65.5
BB-7C-20-3. 6 - 2 1. 7 2.6 10.2 8.7 82.8 70.3
I 70.6
BB- 7 C-20-4.l-l 2.2 3.0 11. 6 10.2 81. 4
BB-7C-20-4.l-2 I 1. 0 1. 8
3.3
10.5
12.8
9.1
11. 7
90.6
79.7
79.9
71. 5
B8-7F-10-4.4-l 2.6
BB-7F-10-4.4-2 2.6 3.4 12.8 11. 7 79.5 71. 3
BB-7F-10-4.9-l 0.8 1. 6 12.2 11.2 93.2 85.8
BB-7F-10-4.9-2 1. 3 2. 1 12.7 11. 6 89.4 82.0
BB-7F-20-4.4-l 1. 6 2.3 11. 8 10.8 86.1 78.5
BB-7F-20-4.4-2 2. 1 2.8 12.2 11. 2 82.7 75.2
BB-7F-20-4.9-l 0.3 1. 0 11. 7 10.6 97.3 90.7
BB-7F-20-4.9-2 1. 8 2.4 13. 0 11. 9 86.2 79.5
BB2.5C-DR7-l 5. 1 5.9 10.9 9.4 53.2 37.3
BB2.5C-DR7-2 4.8 5.6 10.6 9.2 54.7 38.5
BB3.0C-DR6-l 4.6 4.6 11. 8 9.7 69.8 60.9
BB3.0C-DR6-2 4.8 4.5 11. 9 9.9 69.0 60.0
BB3.0C-DR7-l I 4.0 4.8 10.9 9.5 63.7 49.4
BB3.0C-DR7-2 4.5 5.3 11. 4 10. 0 60.6 46.7
BB3.5C-DR6- l 4.7 4.6 12.5 10.5 65.3 56.1
BB3.5C-DR6-3 3.9 3.8 11. 8 9.7 69.8 60.8
BB3.5C-DR7-l 2.8 3.6 10.9 9.4 74.6 61. 9
BB3.5C-DR7-2 2.9 3. 7 11. 0 9.6 73.6 60.9
B83.5F-DR6-l 5. 1 6.4 13.2 11. 8 61. 2 46.2
B83.5F-DR6-2 4.8 6.0 12.9 11. 5 62.8 47.6
BB3.5F-DR7-l 5.6 6.2 13.5 12.5 58.6 50.0
B83.5F-DR7-2 6.6 7.2 14.4 13.4 54.3 46.0
BB4.0C-DR7-l 2.6 3.4 11. 8 10.3 78.0 67.0
BB4.0C-DR7-2 2.8 ' 3.6 12.0 10.6 76.4 65.5
BB4.0F-DR6-l I 3. 1 4.3 12.4 11. 0 75.2 60.7
BB4.0F-DR6-2
BB4.0F-DR7-l
I 3.0
4.8
4.2
5.5
12.4
13.9
10.9
12. 8
75.7
65.1
61. 2
57.2
B84.0F-DR7-2 4.6 5.3 13.7 12.6 66.2 58.2
BB4.5C-DR6-4 1. 8 1. 7 12.0 10.0 88.0 83.0
BB4.5C-DR6-5 1. 5 1. 4 11. 8 9.7 90.2 85.5
BB4.5C-DR6-6 1. 3 1. 2 11. 6 9.5 91. 7 87.3
 211

Puck I.D. %Air %Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

BB4.5C-DR 7-l 1. 2 2.0 11. 5 10.1 90.0 80.4

BB4.5C-DR 7 -2 1. 5 2.3 11. 8 10.4 87.5 77.9
BB4.5F-DR6-l 3 .1 4.3 13. 5 12.0 77.1 64.3
BB4.5F-DR6-2 1. 3 2.5 11. 9 10.4 89.3 75.9
BB4.5F-DR7-l 2.9 3.5 13.1 12.1 78.1 70.8
BB4.5F-DR7-2 3.3 4.0 13. 5 12.5 75.4 68.1
BB4C-DR6-l 3.6 3. 5 12.6 10.5 74.4 67.1
BB4C-DR6-2 2.9 2.8 11. 9 9.9 78.8 71. 8
BB4C-DR6-3 2.7 2.6 11. 8 9.7 80.0 73.2
BB5.0C-DR7-l 0.2 1. 0 11. 7 10.3 98.2 89.9
BB5.0C-DR7-2 0.2 1. 0 11. 7 10.2 98.4 90.2
BB5.0F-DR7-l 3.5 4.2 14.7 13.7 76.1 69.6
BB5.0F-DR7-2 2.8 3.5 14.l 13.1 79.9 73.4
BB5.5C-DR6-l0 0.6 0.5 13.0 11.0 98.1 95.5
BB5.5C-DR6- l l 1. 4 1. 3 13 . 7 11. 7 92.6 89.2
BB5.5C-DR6- l2 0.5 0.4 13.0 10.9 98.7 96.1
BB5 . 5C-DR7-l 0.3 1. 2 12.8 11. 4 97.3 89.9
BB5.5F-DR6-4 0.8 2.0 13.5 12.0 94.3 83.5
BB5.5F-DR6-5 0.4 1. 6 13. 2 11. 7 96.9 86.1
BB5.5F-DR6-6 1. 0 2.2 13.7 12.2 92.9 82.1
BB5.5F-DR7-l 2.4 3.0 14.7 13.6 83.9 77 . 9
BB5.5F-DR7-2 2.0 2.6 14.3 13.3 86.2 80.3
BB5C-DR6-7 1. 2 I 1. 1 12.5 10.4 93.6 89.9
BB5C-DR6-8 0.9 0.8 12.2 10.2 95.8 92.4
BB5C-DR6-9 I 0.8 I 0 .7 12.2 10.1 96.6 93.4
BB5F-DR6- l 1. 2 2.4 12.8 11. 4 90.6 78.6
BB5F-DR6-2 0.9 2 .1 12.5 11. 1 93.1 81.1
BB5F-DR6-3 I
I
1. 4 2.7 13.0 11. 6 89.0 77.1
BB6.5F-DR6-l0 0.1 1. 3 14.9 13.5 99.0 90.1
BB6.5F-DR 6-l2 I
I
0.4 1. 6 15.1 13.7 97.4 88.4
BB6C-DR6-l4 ' 1. 9 1. 8 15.2 13.2 89.7 86.3
BB6C-DR6-l5 2.3 2.2 15.5 13.5 87.3 83.6
BB6Cl0-3. 6 -l 4.0 4.3 12.2 10.1 70.1 57.1
BB6Cl0-3 .6 - 2 3.3 3.6 11. 5 9.4 74 . 8 61. 9
BB6C l0 - 3.6 - 3 3.3 3.6 1 1. 5 9.4 74.7 61. 8
BB6Cl0-3 .6 - 4 3 .4 3.7 11. 6 9.6 73.8 60.8
BB6C l0 -4. l - l 2.7 3.0 12.1 10.0 80.6 69.6
BB6Cl0-4. l - 2 2.1 2.4 11. 5 9.5 84.9 74.2
BB6Cl 0 -4. l -3 2.2 2.5 11. 6 9.5 84.3 73.5
BB6C:. 0 -4. 1-4 2.1 2.5 11. 6 9.5 84.5 73.8
BB6C 20 -3.6-l 4.4 4.3 12.5 10.4 67.5 58.7
BB6C20-3.6-2 3.8 3.7 11. 9 9.9 70.9 62.3
BB6C20-3.6-3 4.0 3.9 12.1 10.1 69.6 60.9
BB6C2 0 -3.6-4 3.8 3.7 12.0 9.9 70.8 62.2
BB6C 20 -4.l- l 2 .1 2.0 11. 4 9.4 84.8 78.8
BB6C20-4.l-2 2.4 2.3 11. 7 9.7 82.4 76.0
BB6C20-4.l-3 1. 9 1. 8 11. 3 9.2 86.4 80.5
BB6C20-4.l-4 2. 5 2.4 11. 8 9.8 81. 7 75.2
BB6F-DR6-7 0.1 1. 3 13.9 12.5 99.4 89.6
BB6F-DR6-8 0.4 1. 6 14.2 12.7 97.0 87.3
BB6F-DR 6 -9 0.3 1. 5 14.1 12.7 97.6 87.8
BB6F10-4. l -l
I 3.0 4.3 12. 6 11. 2 76.4 61. 5
BB6Fl0-4. 1-2 3.3 4.7 12.9 11. 5 74.3 59.5
2.5 3.8 12.2 10.7 79.8 64.5
BB6Fl0-4 .1-3
BB6Fl0-4. 1 -4
I 2 .1 3. 5 11. 9 10.4 82.1 66.6
I
BB6F l0 -4.6-l I l. 2 i 2 . E, 12.1 10.7 89.9 76.0
 212

Puck I.D. \Air \Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

BB6Fl0-4.6-2 1. 7 3.0 12.5 11. l 86.6 72.8

BB6F10-4.6-3 1. 5 2.8 12.4 10.9 87.9 74.0
BB6Fl0-4.6-4 1. 6 2.9 12.5 11. 0 87.1 73.3
BB6F20-4.l-l 3.6 4.9 13.2 11. 7 72. 8 58.2
BB6F20-4.l-2 3.1 4.4 12.7 11. 3 75.6 60.7
BB6F20-4.1-3 3.6 4.9 13.2 11. 7 72. 9 58.2
BB6F20-4.1-4 3.8 5 .1 13.4 11. 9 71. 6 57.1
BB6F20-4.6-1 1. 3 2.6 12.1 10.7 89.7 75.6
BB6F20-4.6-2 2.5 3.8 13.2 11. 8 81. 4 67.8
BB6F20-4.6-3 1. 9 3.2 12.7 11. 3 85.2 71. 3
BB6F20-4.6-4 0.9 2.3 11. 9 10.4 92.2 78.0
BB7Cl0-3.6-1 3.1 4.2 11. 5 10.0 72.9 58.5
BB7Cl0-3.6-2 3.9 5.0 12.2 10.8 67.9 54.0
BB7C10-3.6-3 3.1 4.1 11. 4 10.0 73.3 58.9
BB7Cl0-3.6-4 3.1 4.1 11. 4 10.0 73.3 58.9
BB7C10-4.1-l 1. 5 2.5 11. 0 9.6 86.8 73.7
BB7C10-4.1-2 2.0 3.0 11. 5 10.1 82.8 69.8
BB7C10-4.1-3 1.6 2.6 11.1 9.7 85.8 72. 8
BB7Cl0-4.l-4 1. 6 2.7 11. 2 9.7 85.7 72.6
BB7C20-3.6-1 3.8 4.6 12.1 10. 6 68.3 56.4
BB7C20-3.6-2 3.6 4.5 11. 9 10.5 69.4 ' 57.4
BB7C20-3.6-3 3.1 3.9 11. 4 9.9 73.0 60.7
BB7C20-3.6-4 3.9 4.8 12.2 10.7 67.6 55.7
BB7C20-4.1-1 2.8 3.6 12.2 10.8 76.8 66.1
BB7C20-4.1-2 1. 9 2.7 11. 3 9.9 83.5 72.7
BB7C20-4.l-3 1.8 2.6 11. 2 9.8 84.2 73.5
BB7C20-4.1-4 1. 9 2.8 11. 4 10.0 83.0 72 .3
BB7F-DR6-l3 0.7 1. 9 16.3 15.0 95.7 87.4
BB7F-DR6-l4 1.1 2.2 16.7 15.3 93.5 85.3
BB7Fl0-4.4-1 3.3 4.0 12.7 11. 7 74.2 65.6
BB7F10-4.4-2 3.3 4.0 12.8 11. 7 74.2 65.5
BB7Fl0-4.4-3 2.7 3.5 12.2 11. 2 77.8 69.1
BB7F10-4.4-4 3.4 4.1 12.8 11. 8 73.7 65.1
BB7Fl0-4.9-1 1.8 2.6 12.5 11. 5 85.2 77.4
BB7F10-4.9-2 2.6 3.4 13.2 12.2 80.1 72.3
BB7F10-4.9-3 1. 7 2.5 12.4 11. 3 86.1 78.2
BB7F10-4.9-4 1. 6 2.4 12.3 11. 3 86.8 78.9
BB7F20-4.4-1 5.2 5.8 14.4 13. 4 63.9 56.3
BB7F20-4.4-2 5.2 5.9 14.4 13.4 63.8 56.2
BB7F20-4.4-3 3.6 4.2 12.9 11. 9 72.5 64.5
BB7F20-4.4-4 4.6 5.3 13.9 12.8 66.8 59.0
BB7F20-4.9-l 3.0 3.6 13.4 12.4 77.9 70.7
BB7F20-4.9-2 2.8 3.5 13.3 12.3 78.6 71. 5
BB7F20-4.9-3 2.7 3.4 13.2 12.1 79.4 72.3
BB7F20-4.9-4 3.5 4.1 13. 9 12.8 74.9 67.8
C-4C-10-3.75-1 6.0 . 5.9 14.5 13.5 58.6 56.1
C-4C-10-3.75-2 5.8 5.7 14.4 13.3 59.3 56.9
C-4C-10-4.25-l 4.6 4.5 14.2 13.2 67.9 66.1
C-4C-10-4.25-2 4.7 4.6 14.3 13.3 67.3 65.5
C-4C-20-3.75-1 6.2 6.1 14.6 13.6 57.6 55.1
C-4C-20-3.75-2 6.0 5.9 14.5 13.4 58.2 55.8
C-4C-20-4.25-1 4.1 4.0 13. 7 12.6 70.3 68.8
C-4C-20-4.25-2 4.1 4.0 13.7 12.7 70.3 68.7
C-4F-10-3.75-1 3. 1 3.3 11. 8 10.6 73.9 68.9
C-4F-10-3.75-2 3.2 3.4 11. 9 10.8 73.0 68.0
C-4F-10-4.25-1 2.1 2.3 11. 9 10.8 82.7 78.8
 213

Puck I.D. %Air %Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

C-4F-10-4.25-2 2.0 2.2 11. 9 10.7 83.2 79.3

C-4F-20-3.75-1 2.6 2.7 11. 3 10.2 76.7 73.9
C-4F-20-3.75-2 2.5 2.5 11.2 10.0 78.1 75 . 4
C-4F-20-4.25-1 1. 5 1. 6 11. 4 10.2 86.4 84.8
C-4F-20-4.25-2 1. 5 1. 5 11. 4 10.2 86.6 85.0
C-5C-10-3.75-1 4.7 4.7 13.2 11. 7 64.4 60 . 2
C-5C-10-3.75-2 5.0 4.9 13 . 5 12.0 63.1 58.8
C-5C-10-4.25-1 3.9 3.9 13.5 12.1 71.0 67.8
C-5C-10-4.25-2 3.8 3.8 13.4 12.0 71. 6 68.4
C-5C-20-3.75-1 5.5 5.4 13.8 12.4 60.5 56 . 5
C-5C-20-3.75-2 5.5 5.4 13.9 12.4 60.4 56.3
C-5C-20-4.25-1 4.2 4.1 13. 7 12.2 69.3 66.2
C-5C-20-4.25-2 4.2 4.1 13.7 12.2 69.4 66.4
C-5F-10-4.0-1 2.6 2.4 11. 9 10.7 78.2 77.3
C-5F-10-4.0-2 3.2 3.1 12.4 11. 3 74.0 72. 7
C-5F-10-4.5-1 1. 8 1. 7 12.2 11.1 85.0 84.8
C-5F-10-4.5-2 2.0 1.8 12.3 11. 2 84.2 83.9
C-5F-20-4.0-1 3.1 2.8 12.2 11.0 74.8 74.4
C-5F-20-4.0-2 2.9 2.7 12.1 10.9 75.8 75.5
C-5F-20-4.5-1 2.0 1. 7 12.2 11. l 84 . 0 84.6
C-5F-20-4 .,2-2 2.2 2.0 12.5 11. 3 82.2 82.6
C2.5F-DR5-1 9.0 8.8 14.6 13.4 38 . 2 34.7
C2.5F-DR5-2 8.2 8.0 13.8 12.7 40 . 6 37.1
C3.0F-DR4-1 5.2 5.2 12.1 1 0.9 57.l 52.4
C3.0F-DR4-2 5.5 5.5 12.3 11. 2 55.8 51.1
C3.0F-DR5-1 7.0 6.8 13.8 12.6 48.9 46.2
C3.5F-DR4-1 4.2 4.2 12 . 2 11.1 65.7 61.9
C3.5F-DR4-2 3.8 3.8 11. 9 10 . 7 67.9 64.3
C3.5F-DR5-l 3. 7 3.4 11. 7 10 . 6 68.6 67.4
C3.5F-DR5-2 4.9 4.7 12.8 11. 7 61. 9 60.2
C3.6C-DR5-1 5.9 5.8 13.9 12.4 57.7 53.4
C3.6C-DR5-2 5.5 5.4 13.6 12.1 59.5 55.3
C4.0C-DR5-1 3.1 3.0 12.2 10 . 6 74.9 72 .1
C4.0C-DR5-2 4.6 4.5 13.6 12.1 66.1 62.7
C4 . 0F-DR4-1 2.6 2.6 11. 8 10.6 78.0 75.5
C4.0F-DR4-1 2.8 2.8 12.0 10.8 76 . 6 73.9
C4.0F-DR4-2 2.3 2.3 11. 5 10.3 80.4 78.1
C4.0F-DR5-1 2.8 2.5 11. 9 10.8 76.9 76.7
C4.0F-DR5-2 3.9 3.6 12.9 11. 8 70.1 69.2
C4.5C-DR5-l 2.0 1.9 12.2 10.7 83.6 82 . 1
C4.5C-DR5-2 2.4 2.4 12.6 11. 1 80 . 6 78.8
C4.5F-DR4-1 1.8 1. 8 12.2 11. 0 85.0 83.4
C4.5F-DR4-2 1. 6 1. 6 12.0 10.8 86.4 84.8
C4.5F-DR4-3 1. 5 1. 5 11.8 10.7 87.7 86.2
C4.5F-DR5-1 2.3 2.1 12.6 11. 4 81. 5 81.8
C4.5F-DR5-2 2.3 2.1 12.6 11. 4 81.4 81. 6
C4Cl0-3.75-1 5.1 5.0 13.6 12.6 63.0 60.6
C4C10-3.75-2 5.2 5.1 13.8 12.8 62.l 59.7
C4Cl0-3.75-3 5.2 5.1 13.8 12.8 62.l 59.7
C4C10-3.75-4 5.3 5.3 13.9 12.9 61. 5 59.1
C4C10-4.25-1 3.8 3.7 13.5 12.5 72.0 70.4
C4Cl0-4.25-2 3.9 3.8 13.6 12.6 71.4 69.8
C4C10-4.25-3 3.8 3.7 13. 5 12.5 72 .1 70.5
C4C10-4.25-4 3.8 3.7 13.6 12.5 71.8 70.2
C4C20-3.75-1 4.1 3.9 12.7 11. 6 67.7 66.5
C4C20-3.75-2 4.6 4.4 13.2 12.1 65.0 63.6
 214

Puck I.D. %Air %Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

C4C20-3.75-3 4.2 4.0 12.8 11. 7 67 .1 65.9

C4C20-3.75-4 4.5 4.3 13.0 12.0 65.7 64.4
C4C20-4.25-1 3.3 3.1 13.0 11. 9 74.9 74.4
C4C20-4.25-2 3.3 3.1 13.0 12.0 74.5 74.0
C4C20-4.25-3 3.4 3.2 13.1 12.0 74.3 73.7
C4C20-4.25-4 3.2 3.0 12.9 11. 9 75.1 74.6
C4F10-3.75-1 2.6 2.8 11. 4 10.2 77.2 72.4
C4F10-3.75-2 2.5 2.7 11. 3 10.1 77.8 73.0
C4F10-3.75-3 2.6 2.9 11. 4 10.2 76.8 72.0
C4Fl0-3 . 75-4 2.1 2.4 11. 0 9.8 80.5 75.9
C4Fl0-4.25-1 1. 8 2.0 11. 7 10.5 84.7 80.9
C4F10-4.25-2 1. 7 1. 9 11. 6 10.4 85.6 81. 9
C4Fl0-4.25-3 1.8 2.0 11. 7 10.5 84.8 81.0
C4Fl0-4.25-4 1. 8 2.0 11. 7 10.5 84.7 80.9
C4F20-3 . 75-1 3.1 3.2 11. 8 10.6 73.6 69.8
C4F20-3.75-2 3.1 3.2 11. 7 10.6 73.8 69.9
C4F20-3.75-3 2.7 2.8 11. 4 10.2 76.3 72.6
C4F20-3.75-4 2.6 2.7 11. 3 10.1 76.8 73.1
C4F20-4.25-1 1. 5 1. 6 11. 4 10.2 86.4 83.9
C4F20-4.25-2 1. 6 1. 7 11. 4 10.3 86.1 83.5
C4F20-4.25-3 1.1 1. 2 11. 0 9.8 90.1 87.9
C4F20-4.25-4 1. 3 1.4 11. 2 10.0 88.3 85.9
C5.0C-DR5-l 1. 6 1. 6 12.9 11. 4 87 . 3 86.4
C5.0C-DR5-2 1. 7 1. 6 13.0 11. 5 86.7 85.7
C5.0F-DR4-4 0.9 0.9 12.4 11. 2 92.6 91. 7
C5.0F-DR4-5 0.7 0.7 12.2 11.0 94.1 93.4
C5.0F-DR4-6 1.0 1.0 12.5 11. 3 91.8 90.9
C5.0F-DR5-1 1. 7 1. 5 13.1 11. 9 86 . 6 87.4
C5.1F-DR4-1 1. 4 1. 4 13 .0 11. 8 89.4 88.3
C5.1F-DR4-2 1. 2 1. 2 12.8 11. 6 91. 0 90 . 0
C5.5C-DR5-1 1. 5 1. 4 13.8 12.3 88.9 88.3
C5.5F-DR4-7 0.4 0.4 12.9 11. 7 97.2 96.8
C5.5F-DR4-8 0.6 0.6 13.1 12.0 95.2 94.7
C5.5F-DR4-9 0.6 0.6 13.1 12.0 95.2 94.6
C5.5F-DR5-1 1. 2 0.9 13.5 12.4 91. 5 92.7
C5Cl0-3.75-1 5.2 5.2 13.7 12.2 61. 9 57.6
C5Cl0-3.75-2 4.7 4.7 13.2 11. 7 64.4 60.2
C5Cl0-3.75-3 5.5 5.5 14.0 12.5 60.4 56.0
C5Cl0-3.75-4 5.6 5.6 14.1 12.6 60.0 55.6
C5Cl0-4.25-1 3.4 3.3 13 . 0 11. 5 74.1 71.1
C5Cl0-4.25-2 3.1 3.1 12.8 11. 3 75.6 72 .8
C5C10-4.25-3 3.7 3.7 13.4 11. 9 72.2 69 . 0
C5Cl0-4.25-4 3.2 3.1 12.9 11. 4 75.3 72.4
C5C20-3.75-l 2.6 2. 5 11. 2 9.7 76.8 74.6
C5C20-3.75-2 2.9 2.7 11. 5 9.9 75.0 72 .6
C5C20-3.75-3 3.1 3.0 11. 7 10.2 73.4 70.9
C5C20-3.75-4 2.9 2.7 11. 5 10.0 75.0 72.6
C5C20-4.25-l 3.9 3.8 13.5 12.0 70.8 68.4
C5C20-4.25-2 4.5 4.3 14.0 12.5 67.9 65.2
CSC20-4.25-3 4.2 4.1 13.7 12.2 69.2 66.6
C5C20-4.25-4 4.0 3.9 13. 5 12.0 70.4 67.9
C5F10-4.00-1 2.9 2.7 12 . 1 11.0 76.2 75.1
C5F10-4.00-2 2.9 2.7 12.1 11. 0 76.2 75.1
CSFl0-4.00-3 2.8 2.7 12.1 10.9 76.6 75.5
C5F10-4.00-4 2.8 2.7 12.1 10.9 76.7 75.6
C5F10-4.50-l 2.3 2.1 12.6 11. 4 82.1 81. 7
 215

Puck I.D. %Air %Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

C5Fl0-4.50-2 2.0 1. 9 12.4 11. 2 83.8 83.5

C5Fl0-4.50-3 2.1 2.0 12.5 11. 3 83.0 82.6
C5Fl0-4.50-4 2.1 1. 9 12.5 11. 3 83.2 82.9
C5F20-4.00-l 3.4 3.2 12.6 11. 4 72. 6 71. 9
C5F20-4.00-2 3.3 3.0 12.4 11. 3 73.5 72 .9
C5F20-4.00-3 3.6 3.3 12.7 11. 5 71.8 71.1
C5F20-4.00-4 3.4 3.1 12.5 11.3 72 .9 72.3
C5F20-4.50-l 5.1 4.8 15.0 13.9 66.3 65.3
C5F20-4.50-2 4.9 4.7 14.9 13.8 67.0 66.0
C5F20-4.50-3 5.1 4.9 15.1 13.9 66.0 65.0
C5F20-4.50-4 5.0 4.7 14.9 13.8 66.7 65.7
C6.0F-DR4-10 0.4 0.4 13.9 12.7 97.3 96.9
C6.0F-DR4-ll 0.4 0.4 13.9 12.8 97.2 96.8
C6.0F-DR4-12 0.4 0.4 13.9 12.8 97.0 96.6
C6.5F-DR4-14 0.6 0.6 15.1 13.9 95.9 95.5
C6.5F-DR4-15 1. 4 1. 4 15.7 14.6 91. 1 90.4
CC3.5DR4-l 5.4 5.3 13.4 12.4 59.3 56.8
CC3.5DR4-2 6.0 5.8 13.9 12.8 57.1 54.4
CC3.5DR4-3 6.3 6.2 14.2 13.2 55.5 52.8
CC3DR4-l 7.5 7.4 14.3 13.3 47.2 43.9
CC3DR4-2 6.0 5. 9 12.8 11. 8 53.5 50.3
CC3DR4-3 7.8 7.7 14.6 13.5 46.2 42.9
CC4.7DR4-l 3.3 3.2 13.9 12.9 76.2 75.1
CC4.7DR4-2 3.0 2.9 13.6 12.6 78.1 77.1
CC4.7DR4-3 3.0 2.9 13.6 12.6 78.1 77.2
CC4DR4-l 6.1 6.0 15.0 14.0 59.5 57.3
CC4DR4-2 4.0 3.9 13.1 12.1 69.5 67.8
CC4DR4-3 4.6 4.4 13.6 12.6 66.6 64.7
CC5.1DR4-l 1. 9 1. 8 13.5 12.5 85.8 85.5
CC5.1DR4-2 2.1 2.0 13. 6 12.6 84.7 84.3
CC5.1DR4-3 1. 7 1. 6 13.3 12.3 87.0 86.8
CC5.8DR4-l 1. 2 1.1 14.3 13.2 91. 5 91. 7
CC5.8DR4-2 1.0 0.9 14.0 13.0 93.2 93.5
IC-4C-10-4.2-l 4.4 4.5 14.0 12.7 68.6 64.9
IC-4C-10-4.2-2 4.8 4.8 14.3 13.0 66.6 62.8
IC-4C-10-4.7-l 3.4 3.4 14.1 12. 8 76.1 73.l
IC-4C-10-4.7-2 3.9 3.9 14.5 13.2 73.4 70.3
IC-4C-20-4.2-l 5.8 5.7 15.1 13.9 61. 8 59.0
IC-4C-20-4.2-2 5.9 5.8 15.3 14.0 61. 2 58.3
IC-4C-20-4.7-l 4.1 4.1 14.7 13.4 71. 8 69.7
IC-4C-20-4.7-2 4.6 4.5 15.l 13.8 69.4 67.2
IC-4F-10-4.2-l 3.1 3.3 12.8 11. 6 75.5 71. 8
IC-4F-10-4.2-2 3.4 3.5 13.0 11. 9 74.0 70.3
IC-4F-10-4.7-l 1. 7 1. 9 12.6 11. 4 86.1 83.5
IC-4F-10-4.7-2 1.4 1. 6 12.3 11.1 88.3 85.8
IC-4F-20-4.2-l 3.3 3.6 12.9 11. 7 74.5 69.4
IC-4F-20-4.2-2 3.1 3.4 12.7 11. 5 75.3 70.3
IC-4F-20-4.7-l 2.1 2.4 12.8 11. 6 83.6 79.4
IC-4F-20-4.7-2 2.0 2.3 12.7 11. 5 84.5 80.4
IC-5C-10-4.8-l 2.9 3.0 13.7 12.2 79.0 75.8
IC-5C-20-4.8-l 3.1 3.0 13.8 12.3 77.5 75.2
IC-5F-10-4.2-l 3.1 3.0 12.7 11.2 75.8 73.4
IC-5F-10-4.2-2 3.2 3.1 12.8 11. 3 75.1 72.6
IC-5F-10-4.7-l 2.0 1. 9 12.7 11.2 84.7 83.4
IC-5F-10-4.7-2 2.2 2.1 12.9 11. 4 83.1 81. 7
IC-5F-20-4.2-l 4.1 3.8 13.5 12.0 69.8 68.3
 216

Puck I.D. %Air %Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)

IC-5F-20-4.2-2 4.4 4.1 13.8 12.3 68.2 66.6

IC-5F-20-4.7-1 2.4 2.2 13.1 11. 6 81. 3 81. 2
IC-5F-20-4.7-2 2.0 1. 7 12.7 11. 2 84.2 84.5
IC3.2F-DR4-1 5.9 6.2 13.2 12.0 55.2 48.4
IC3.2F-DR4-2 6.6 6.9 13.8 12.6 52.2 45.5
IC3.5C-DR4-1 5.8 5.7 13.7 12.4 57.8 54.0
IC3.5C-DR4-2 7.1 7.0 14.9 13.6 52.5 48.7
IC3.5C-DR5-l 6.1 6.0 13.9 12.3 56.3 51. 4
IC3.5C-DR5-2 6.6 6.6 14.4 12.9 53.9 49.0
IC3.5F-DR5-l 4.8 4.6 12.8 11. 3 62.2 59.6
IC3.5F-DR5-2 5.7 5.4 13.5 12.1 58.1 55.2
IC3.6F-DR4-l 4.6 4.9 12.8 11. 6 64.3 58.1
IC3.6F-DR4-2 4.5 4.8 12.8 11. 6 64.4 58.3
IC4.0C-DR4-l 4.1 4.0 13.2 11. 9 69.1 66.5
IC4.0C-DR4-2 4.9 4.8 14.0 12 . 7 64.7 61. 8
IC4.0C-DR5-l 4.9 4.8 13.8 12.3 64.6 60.7
IC4.0C-DR5-2 5.2 5.2 14.1 12.6 62.9 59.0
IC4.0F-DR4-l 3.3 3.6 12.5 11. 3 73.3 67.9
IC4.0F-DR4-2 3.3 3.6 12.5 11. 3 73.3 68.0
IC4.0F-DR5-l 3.6 3.4 12.7 11. 2 71.4 70.0
IC4.0F-DR5-2 4.5 4.2 13.5 12.0 66.6 64.7
IC4.5C-DR4-l 3.0 2.9 13. 3 12.0 77.2 75.4
IC4.5C-DR4-2 4.7 4.6 14.8 13.5 68.1 65.7
IC4.5C-DR5-l 3.6 3.5 13.6 12.1 73.8 71.0
IC4.5C-DR5-2 3.5 3.4 13.5 12.0 74.4 71. 7
IC4.5F-DR4-l 2.0 2.3 12.4 11. 2 83.6 79.2
IC4.5F-DR4-2 2.0 2.3 12.4 11. 2 83.5 79.1
IC4.5F-DR5-l 2.8 2.6 13.0 11. 5 78.2 77.7
IC4.5F-DR5-2 3.8 3.5 13.9 12.4 72. 7 71.6
IC4Cl0-4.2-l 5.4 5.4 14.9 13.6 63.8 59.9
IC4Cl0-4.2-2 5.0 5.1 14.5 13.2 65.6 61.8
IC4Cl0-4.2-3 4.7 4.8 14.3 13.0 67.0 63.2
IC4Cl0-4.2-4 4.9 5.0 14.5 13.2 65.9 62.l
IC4Cl0-4.7-l 4.3 4.4 14.9 13.6 71. 2 68.1
IC4Cl0-4.7-2 3.6 3.7 14.3 13.0 74.7 71. 7
IC4Cl0-4.7-3 4.3 4.4 15.0 13.7 70.9 67.7
IC4Cl0-4.7-4 3.4 3.5 14.l 12.8 75.9 73.0
IC4C20-4.2-l 9.5 9.4 18.5 17.3 48.6 45.4
IC4C20-4.2-2 4.4 4.4 13.9 12.6 68.1 65.6
IC4C20-4.2-3 5.5 5.5 14.9 13.6 62.9 60.0
IC4C20-4.2-4 6.3 6.2 15.6 14.3 59.7 56.8
IC4C20-4.7-l 4.6 4.6 15.1 13.8 69.3 67.1
IC4C20-4.7-2 4.3 4.2 14.8 13.6 70.9 68.7
IC4C20-4.7-3 4.4 4.3 14.9 13.6 70.6 68.4
IC4C20-4.7-4 5.3 5.2 15.7 14.4 66.5 64.l
IC4Fl0-4.2-l 1. 6 1. 7 11. 4 10.2 86.0 83.0
IC4Fl0-4.2-2 1. 5 1. 7 11. 4 10.1 86.6 83.7
IC4Fl0-4.2-3 1. 6 1. 7 11. 4 10.2 86.3 83.3
IC4Fl0-4.2-4 1. 7 1.9 11. 6 10.3 85.0 81.9
IC4Fl0-4.7-l 0.7 0.9 11. 7 10.5 93.6 91. 6
IC4Fl0-4.7-2 0.8 0.9 11. 7 10.5 93.4 91. 3
IC4Fl0-4.7-3 0.8 0.9 11. 8 10.6 93.2 91.1
IC4Fl0-4.7-4 0.7 0.8 11. 7 10.5 94.0 92.0
IC4F20-4.2-l 3.1 3.4 12.7 11. 5 75.4 70.4
IC4F20-4.2-2 2.6 2.9 12.3 11. l 78.7 73.8
IC4F20-4.2-3 2.3 2.6 12.0 10.8 80.6 75.7
 217

Puck I.D. \Air \Air VMA VMA VFAC VFAC

Voids Voids (Cale.) (Rice) (Cale.) (Rice)
(Cale.) (Rice)
IC4F20-4.2-4 2.4 2.6 12.0 10.8 80.4 75.5
IC4F20-4.7-l 1. 9 2.2 12.7 11. 5 84.9 80.8
IC4F20-4.7-2 1. 8 2.1 12.6 11.4 85.5 81. 5
IC4F20-4.7-3 1. 5 1. 8 12.3 11.1 88.0 84.1
IC4F20-4.7-4 1.8 2.1 12.6 11. 4 85.7 81. 6
IC5.0C-DR4-1 2.4 2.3 13.7 12.4 82.5 81. 4
IC5.0C-DR4-2 3.1 3.0 14.3 13.0 78.6 77.1
IC5.0C-DR5-1 2.3 2.2 13.5 11. 9 83.1 81. 5
IC5.0C-DR5-2 2.1 2.0 13.3 11. 7 84.5 83.0
IC5.0F-DR4-1 1.0 1. 3 12.5 11. 3 91.9 88.5
IC5.0F-DR4-2 1.0 1. 3 12.5 11.2 92.1 88.7
IC5.0F-DR5-1 2.0 1.8 13.3 11.8 84.7 85.0
IC5.0F-DR5-2 2.1 1. 8 13.4 11. 9 84.2 84.5
IC5.5C-DR5-1 2.9 2.8 15.0 13.5 80.7 79.0
IC5.5C-DR5-2 1. 2 1. 2 13.6 12.0 90.9 90.3
IC5.5F-DR4-1 0.7 1. 0 13.2 12.0 94 .7 91. 8
IC5.5F-DR4-2 0.8 1. 1 13.3 12.1 94.0 91.1
IC5.5F-DR5-1 1.1 0.9 13.5 12.0 91. 7 92.9
IC5.5F-DR5-2 1. 0 0.7 13.4 11. 9 92.5 93.9
IC5C10-4.3-1 3.1 3.2 12.9 11. 4 75.7 71. 8
IC5Cl0-4.3-2 2.8 2.9 12 . 7 11.1 77.6 73.9
IC5Cl0-4.3-3 2.8 2.9 12.6 11.1 77.7 74.0
IC5Cl0-4.3-4 2.9 2.9 12.7 11.1 77.3 73.6
IC5C10-4.8-1 2.1 2.2 13 . 1 11. 5 83.6 80.8
IC5Cl0-4.8-2 2.0 2.0 12.9 11. 3 84.8 82.2
IC5C10-4.8-3 2.2 2.3 13.1 11. 5 83.3 80.5
IC5Cl0-4.8-4 2.1 2.2 13.1 11. 5 83.6 80.8
IC5C20-4.3-1 3.2 3.1 12.9 11. 4 75.1 72.3
IC5C20-4.3-2 3.7 3.7 13.4 11. 8 72 .1 69.1
IC5C20-4.3-3 2.7 2.6 12.5 10.9 78.3 75.8
IC5C20-4.3-4 2.5 2.5 12.3 10.7 79.4 77.0
IC5C20-4.8-1 2.7 2.6 13.4 11. 9 80.2 78.2
IC5C20-4.8-2 1.8 1. 8 12.7 11.1 85.6 84.2
IC5C20-4.8-3 2.6 2.5 13.4 11.8 80.5 78.5
IC5C20-4.8-4 2.2 2.2 13.1 11. 5 82.8 81.1
IC5Fl0-4.2-1 3.5 3.4 13.1 11. 6 73.4 70.7
IC5Fl0-4.2-2 3.3 3.2 12.9 11. 4 74.7 72 .1
IC5Fl0-4. 2-3 3.3 3.3 13.0 11. 5 74.1 71. 5
IC5Fl0-4.2-4 3.1 3.0 12.7 11. 2 75.9 73.5
IC5Fl0-4.7-1 1. 6 1. 5 12.4 10.9 86.9 85.9
IC5F10-4.7-2 1. 4 1. 4 12.3 10.8 88.2 87.4
IC5F10-4.7-3 1. 5 1. 4 12. 4 10.8 87.6 86.7
IC5F10-4.7-4 1.8 1. 7 12.6 11.1 85.6 84.4
IC5F20-4.2-l 4.0 3.7 13.4 12.0 70.3 68.9
IC5F20-4.2-2 4.2 .3. 9 13.6 12.1 69.2 67.7
IC5F20-4.2-3 3.4 3.2 12.9 11.4 73.5 72.4
IC5F20-4.2-4 2.6 2.4 12.2 10.7 78.4 77.9
IC5F20-4.7-1 1.9 1. 7 12. 6 11. l 84.7 85.l
IC5F20-4.7-2 1.8 1. 5 12.5 11.0 85.9 86.4
IC5F20-4.7-3 1.8 1. 5 12.5 11.0 85.8 86.4
IC5F20-4.7-4 1. 7 1. 5 12.4 10.9 86.l 86.6
IC6.0F-DR5-1 1.1 0.8 14.4 13.0 92.5 93.7
IC6.0F-DR5-2 1. 0 0.8 14.4 12.9 92.8 94.0
 218

APPENDIX F

RESULTS OF MARSHALL MIX DESIGNS

 D:ite Print ed: O:!-Ju l-93 Test Propert y Curves for BBDR6F Mix

219
!UNIT WEIGHT VS . %Aej !%AV BY RICE VS. %Aej
157

156
I I
~
V
"- 155
I I

•
:l:

~
"' 154 ' •
'g
::::,
153

152
' T .I •
I
I
151 1
T • i
~5 4 ~5 5 ~5 6 LS 7 ~5 4 ~5 5 ~5 6 LS 7
%AC By Wgl Of Agg. %AC By Wgt. Of Agg.

jFLOW VS . %Acj !\'MA VS . %ACJ

18 17
16
16
£

8
14

12
••
' ! ! ! ! !

;.::
! 10
8 •
6
T I 12
T
I I
4 ~5 5 ~5 6 LS 7
1
1.s 4 ~5 5 ~5 6 LS 7
'lf,AC By Wgl Of Agg . %AC By Wgl Of Agg.

!STABILITY VS . %ACJ !\'FAG BY RICE VS. %ACJ

4500 115
4000
l
o IIO
! .:.
'
~
,e 3500 0 85
I
~
I
i 3000
l 80
:C
q,
.;
2500
2000 ' ~
w
\,,1
a:
75
70
65
I

1500 /i; 60
"E
i 1000 ~
~ 55
500 50
! I I
~.5 4 4.5 5 5.5 6 6.5 7 ~ -5 4 4.5 5 5.5 6 6.5
%AC By Wgl Of Agg . %AC By Wgt. Of Agg .

tpbbdr6f. wq 1
D:ite Printed: 10-Aug-93 Test Property Curves for BBDR6C Mix

220
!UNIT WEIGHT VS. %Ac\ l%AV BY RICE VS . %AC!
. ••
157
' i
156
I ' ~ ' •
~
:l:

~
Cl
155

15-4

T
! 't •
•
5
2
'
§ 153

152
5
1
;' -
151
3 3.5 , 4.5 5 5.5 6
' :u 4 46 5
' 6.6
%AC By Wgt Of Agg. 1"AC8yWglOf-

!FLOW VS. %Acj IVMA BY RICE VS. %Acj

22
I
14.
•
20
I
• I
18 0 13.
•
.£
8 16
I I ~ 1
I

I !
' ,•
12
14
i:
...
.51 12
I
I
.
~
.. 11.,
10
8
I
T
i 11

10. g
• ' • •T
~

'
3.5 4 4.5 5
%AC By Wgt Of Agg.
5.5 6 -
T
&
'
.. , •
1"AC ByWgl O f -

lSTABILITY VS . %Acj [VFAC BY RICE VS. %Acj

4500 DO
I I I
I •
T -
l
3500
I •
:;; 3000
iil 2500
• i
' T • ~

iii
.t::
em
::E
2000
1500
I
• ·-
•
16
1000
3 3.5 , 4.5 5 5.5 6 4 4.6 6 u
%AC By Wgt Of Agg. 1"ACByWg<Of-

tpbbdr6c.wql
Date Printed : '.;]- Mar-91 Test Prope rt y Curves fo r BBDR 7F Mix

22 1
!UNIT WEIGHT VS. %AC! j%AV BY RICE VS . %ACj
15 1 75

150.5
I i I
i I I 0 ...
u
150 0 i I
Q.
149.5
i l I " I I i
.c
a, 149
I
i I i ~·
w
5
I I
;
~
!,2
,.8.5 I I : a:
~ 45
I
'le
::, 148
I ! I
I > • l
147 .5
I I I ;! 3S
!
I

147
3.5
I
4 45
i
%AC By Wgt. Of Agg .
5
I
5 .5 2.S
~· u
I
M C ByWgt O f -
I
~·

!FLOW VS. %Acj IVMA BY RICE VS. %Acj

....
1 lU
, o5
II I I I I
I I I 8 13.4
I !
a i i i E 1~2 i I

I i I ! 13
I
. 5
I
I
I
l .
0
ii 12.1
~
I
I
• !
C 12.S
I ,I
7.5
i
T
I I
l 12.4
I i
I

..••
7 12.2
i i ! I
5 12
. • •s 55 H 5.5
M C By Wgt Of Agg MC By Wgt Of.OW

5000
!STABILITY VS. %ACl

I I I
.. f'/FAC BY RICE VS. %Acj
I I I
I I
I i I I
•i I!
0
on I
•
. I E 10 I ! I

i
2
• i
I
I w l I
0 ..
I ii:
I I I
I I •I I
1ii ..
;
0
.
..,
I
!
I
I
I
I
I I
45
I I i
4. 5
I
5.5
M C By WQl Of.OW MC ByWQlOf.OW

tpbbdr 7f. wq I
D.1t e Pri nted : 08-Dcc- 92 T est Property Curves for BBDR7C Mix

222
.... !UNIT WEIGHT VS. %AC! l/oAV BY RICE VS. %AC!
0

,..,
I
i l
i
I I' I
I 5.5-----+----+--+----+---I
!
.---,-,---1--+---+---+---1
i
l
I
: i ! o"1 -
~ 4..5+---+---+----+--+----+---I
I I
I I
i
I i •+---+--_._---+--+----+---I
~ 3.l+---+---,..--•r---+---+---1
.
I

I I
I ! I ,.a:
I
! I I
• I ; 2.1

·---~----~----~-
151 21----t----1r---+-------t----1
I
150.5
2. 5
• I I35 ,
I
,s 5.5
U+---+----t-----+--+---+---1
2.5 u
......c ByWgt OfAw

IFLOW VS. %Acj !VMA BY RICE VS. %Acj

30
"·•
I I! I I
I I 0
' I I

i i
0
i ~ ,a.
I
2
•
I I I '"
~
.,. '
I I T
• • I . :.
10

• t
I
I
•I I I I
5
2.5 u

!STABILITY VS. %Acj tJFAC BY RICE VS. %AC!

100

!! I
I I I
l ! I
• .t
I
I I 'I •I 2.5 3.1 4 4.!5 u
MCByWgt OfAw

tpbbdr7c.wq 1
 Date Printed : 21 -Ma r-93 Test Property Curves for CDR4F Mix

223
JUNIT WEIGHT VS. %Aej J%AV BY RICE VS. %AC!
15J 5
15J
• •
I i ~ i j
i ! I
I I l I ;
152.5
I I
• T I •' ••
u i
152
Q.

1: 151.5
I ; I I I I
~
0,

15 1
i
•
I I
I
I
I
: I
~
:,
150 5
!

I I I i I I
149.5
150
i i i I i •~ I
I I
I I I
1'9
3 3 .5
'
4 4.5 5 5 .5
%AC By Wgt. OI Agg.
6 6.5 .,
3 H Ui 5
MC llyWg< OfAQI.
5..5 ...

JFLOW VS. %Acj JvMA BY RICE VS . %Acj

24
22
20
I
!
I
I
... 5
I
I
I
I
I
I
I
I
.E 18
I ! I 0
0
15
l I
I
I I i
I

8 16 I !
' E •• 5 I I I !
i I :I
•
J
0
.:.:
14
12
10
I
I .
~
"'u
.
ic 13. 5
>-
<
I
i
I
!
I
I
I
.•
I

e i
•
J l
~ ,~ 5
i I I I I I
6
'T T
~
I
!
I . • j i I I
! ! !
4
3 as • 45 s ~5
MC By Wgt. OI Agg .
s ~5 tt .5
...T 45
I
5
MCByWg< OfAQI;
5..5 ...

!STABILITY VS. %Acj JvFAC BY RICE VS . %Acj

2600 ,o
I I I I
. 2400
T • i I I I
• 00
f! 2200
I
I 'I
I

t
1i
2000
/
I .• .
i I
l
I
~
]
1800
1600
I
I !
i
!
:
I
-
•
I • I I I •
;•
~
; 1400
I ! I I ! ! I I I
I
1200

10003 3 .5
I
4
I
5
I
!
5.5
4 .5
i •
I

6 6.5
•... I !
i
I I
I
I
45 5 5..5
%AC By Wgl OI Agg. MC ByWg\ Of Aog

tpcdr4f.wql
D3le Printed: D- Mar -93 Test Property Curves for CDR4 C Mi x

22 4
!UNIT WEIGHT VS . %Aej !%AV BY RICE VS. %AC!
152

151
I i
I I
..• I ' ! i
I I
u
a.. 150
I
I
•
I
•. I 8,:: • .•• I
I
! I

i
•I
• I
."
t;; •

•T
. I I I
.c
°'
l
~
149

1.a
•'' ! I
0
ii'
:;
•
I I .. !
:,
14 7
i I I j 2
I I II
i I I'j
I I I
146
3 3.5 4 4.5 5
MC By Wgt Of Agg.
5.5 0
•
I
..
I
MCllyWr,.Of-
I
...

!FLOW VS . %Acj !VMA BY RICE VS . %Acj

14 111
13
12
I
I
.... I
i
I
£ 'I 8
:,;
13.2
11
: t;; ..
~ 10
9
I t"ii' 12.1
•
~
;;: 8
i fij 12.1 -
7
I •~ 12.• •
6
I
l ! I •
• T • 12.2

•
3.5 4 4.5 5
MC By Wgt Of Agg.
5.5 12
• :u • 4.5
MCBr Wr,.Of-
...I

fST ABILITY VS. %Acj jVFAC BY RICE VS . %Acj

2300 00

. 2200 - ' I I I I
B 21 00 !
i
I

i
- I I
t 2000
:0
~
I
•i i . I I I
I I
"ii
.t:
l!!
1900
1800
I
I
•
I
I ;
•
2
1700
1600
'I I
1 • i I
I '' I iI I
1500
3 3.5 4 4.5
MC By Wgt Of Agg.
5 5.5 - ... ' 4.5
MC8y Wr,.Of-
o.s

tpcdr4c.wq l
Date Printed: 08-Dec-92 Test Property Curves for CDR5F Mix

225
IUNIT WEIGHT VS. %AC! !%AV BY RIC E VS. %AC!
152 9 I
i I
i i I I I I
I
' 8

~
151

150
I I I i r 0
0
I
7
I
I
!
I' 'I I ! tu 6
149 :I'
:l:
~ 1'8
I I ! ! w 5
u I
I I I I !
i
c
::,
U7

U6
i I I
I
a:> '
"'> 3
2
• ~
+
i
.!.
I

I I
145
14'1
U
I
3 U
I
,
I
I
U
MC By Wgt Of Agg
5 U
' ~ .5 3 3.5
'
i
I
, .5
MCByW gt Of Agg.
+
5 5 .5

!FLOW VS. %ACJ ~MA BY RIC EVS . %AC!

1, 13.5
13
0 13
£
8
12

11
0
ffi 12.5
! ! !
- 10
:I:
w
12
I
. 9
~ I

i
T
...
] 8 >
"'< 11 .5
7
8 •
I
~ 11 I
I
I
I

3
•
3 .5
I
4
I
, .5
I
5 5.5 10.~ .5 3 3.5 4
T I
, .5
I
5 5.5
%AC By Wgt Of Agg. MC By Wgt Of Agg.

!STABILITY VS. %ACJ ~FAC BY RIC E VS. %AC!

100
3200
l I I ! 0
I I I
90
~

i I •I
~ T
,Q 3000
80
l
t
I
2800
2600
•
' :I'
w
u
a: 80
70
I
-;; 2.00 >
..c
~ 2200 "'u 50
:I' <
~
2000
! '°
1 ~.5 3 3 .5 , , .5 5 5.5 ~ .5 3 3.5 4 4.5 5 5 .5
%AC By Wgt Of Agg. %ACByWgt Of Agg.

tpcdr5f.wql
D.lle Printed : 08 -Dec-92 Test Propert y Curves for CDR5C Mix

226
!UNIT WEIGHT VS . %AC! !%AV BY RICE VS . %Aej
151 •
. • i
•
: I I

.
150.5
•I I i I
u
150
I I
C
0
:,:
•5
: I
c.. 1"8.5
t;; I I
:c
;
"'
1'9

1.CS.5
I
I .
:,
u
ii'
•
:I. 5
I
I
I
I
~
:,
1'8
1'7.5
1•1
• I
I
I
I
iii
; 25
-
I
i
I
.- I
I
i

• I
i •
1~.5
3 .6 • .1 • .6
MC By Wgt Of Agg.
5 .1 1
... 41
I
4.1
'-'C 8y Wg< OI "4111
I
6.t

!FLOW VS . %AC! IVMA BY RICE VS. %AC!

12 12
•
11
12 •
10
I ~
122
I
£ 1
-
~ 9

8
.!.!i ••
a:
11.
11 .
I
!
..:
I I ~ • 11 .
7 i 2 11 .
1
I
6
10.
• . I

•...
I
~1 ~6 ~1 10
4.1 4.1 1,. t
MC By Wgt Of Agg. '-'C lly Wgl Of "4111

jsTABILITY VS. %AC! l'JFAC BY RICE VS. %Acj

3000 80

2800
I -I 16
I I
I

a"
2600
• I
I I I . 0
0 80
I I
I
? I :,:
t;; ,.. I
.,
:;; 2.acl
I

!
.
:,

~
70
I I
ui
ci
£;

.,
~
2200

2000
I I ..,..
u
16
I
i
I
~
1800
I I !:; 80 I
I
16
I I
1 ~.6 • 1 • .6 5. 1 50
u ... , .... 6.1
MC By Wgt Of Agg. 'MCllyWg<0,"4111

tpcdr5c.wql
DJtc Printed: 02 -Jul- 9:< Test Propert y Curves for ICDR4F Mix

227
jUNI T WEIGHT VS. %AC J j%AV BY RICE VS. %AC!
1~

I I I I I 0 6
I I
I

u
Cl.
152

151
I I I 0
~ 5 . I I
l

I'
I
!
I I ~
u
4
I
I
I a: 3
>-
a, I I
148
I I
> 2
I I

147 I I ' 1
0
37 ~2
I
~7 5.2
3.2 37 ~2 ~7 ~2 3.2
%AC By W~ Of Agg. %AC By W~ Of Ag g.

!FLOW VS. %AC! jVMA BY RICE VS . %Aq

14...--~=:::;:====~--~
13 + - - - + - - - - + - - + - - - - + - - - + -
I
£ 1 2 + - - - + - - - - + - - . - - - ;:- - - + - - I
I
I
i

~ ::,+ ---+----+--;--+--+~ 12
I
;,+ ---+---+--+----+-ti--+__. ~ 11 .8
I
1... 8t---+---+----tl------+----t lii< 11 .6 . I
7+---+----+--t----+---+----i ! !
st---+---.--+-<>-+---+~
5
I I
I
11 .4
11 .2
1
I
. I
I
I

3 3~5 4 4.5 5 5.5 1.2 3.7 4.2 4.7 5.2

%AC By W~ Of Agg. %AC By W~ Of Agg.

jSTABILJTY VS. %AC! jVFAC BY RICE VS . %AC!

4500 95
I I
I 90
•
i:
I
I
. I
2 75
- I
I
I
- I "'!.1a: 70 I I I
I
I ~ ~ I
60
~ !
I
55
I I

1SO<l.i.2 3.7 4.2 4.7

I
5.2
50

~ -2 3.7 4.2
I
47
i
5.2
%AC By W~ Of Agg . %AC By W-:;t. Of Ag g.

tpicdr4f.wql
D:i lc Pr inl cd: O'.:-Ju J-9:; Test Property Curves for ICDR4C Mix

228
JUNIT WEIGHT VS . %Ac\ !%AV BY RICE VS. %AC!
15 1 5 7
151
!
' 6 .5
150.5
! I 0
6
I
0
I :i:
150 5.5
I I t;;
l: 149.5 ::. 5
149 •I I w
CJ>
!,2 4.5
I
i 148.5
! I
a:
>
m
4
148
I I > 3 .5

'
147.5 3
I I
147 2.5
146.5
I I
3 .5 " 4.5 5 ~ .5 4 4.5
%AC By Wgt Of Agg . %AC By Wgt Of Agg.

IFLOW vs. %AC! jVMA BY RICE VS. %AC!

13 13.8

12
I I 13.6
8 13.4
E
8 11
! ~ 13.2
::Ii 13
10
w 12.8
~ •
...I
12.6
9 1ii 12.4
I <
~ 12.2

I 12
7
3.5 '
I
4 4 .5
%AC By Wgl Of Agg .
5 11.\.5
i
4 4.5
%AC By Wgt Of Agg .
5

ISTABILITY vs. %AC! ~FAC BY RICE VS. %AC!

2500 85

. 2400
'I I 80
!J.
2300
I I i 75
~
.
:.; 2200
iii 2100
;;;
-!; 2000
!
II
i
'I
I
::Ii
w
u 65
If
>
m 80
70
•
•
;; I ~ 55
::.
1900
I I !; 50

18~.5
I' I ~ .5 4 4.5 5
4 4 .5 5
%AC By Wgl Of Agg. %AC By Wgt Of Agg.

tpicdr4c.wql
UJIL· Pri nt ed: :::<- Mar- 91 Test Propert y C urves for ICD R5F Mix

229
IUNIT WEIGHT VS . %AC! j%AV BY RICE VS. %AC!
151 55

150 5
i ! i •• 0
5
i
I I
I !
i
u 150
i I I' I 0
J:
4.5
• I I
4
Q.

149 5
•
I ! I I t;:;
::Ii
3.5
I I
.r:.
w ~
"'
~ , 49
I I \,l 3
a: 2 .5
~ 148 5
I I >
a,
> 2
I

T I '
'
1.5
I I
147.5
I I
1
0.5
I
3.5 4 4.5 5 5.5 6 3.5 4 4 .5 5 5.5 6
MC By Wgt Of ,t,gg. MC By Wgt Of Agg .

!FLOW VS . %Acj jVMA BY RICE VS . %Acj

16 13
15
I
I 12.8
0
14 0
! I ~ 12 .8 I
.E 13
! [ij 12.4
8 12
I :::E 12.2
a:~
::, 11
I 12
~ 10
U: 9 I /ii 11 .8
I c(
1 1.6
8
~
7

~ -5
T
I
4
I
4.5 5 5.5 8
11 .4

11 .~ .5 4
'
4 .5 5 5.5 6
MC By Wgt QI Agg. MC By Wgt Of Agg.

jSTABILITY VS. %Acj fvFAC BY RICE VS . %Acj

4500 95

E
~

~ 3500
4000
i
i
I
I
0
0
J:
t;:;
::Ii
90
85
eo
'I
~
<ii i ~
I
w
\,l
a:
75 •
.; 3000 > 70
I I
a,
~ u 65
I I
i 2500 c(
\; T
II I 60
I i
~ -5 4 4.5 5 5.5 6 4 4.5 5 5.5
MC By Wgt Of Agg . MC By Wgt Of Agg.

tpicdr5f.wq 1
O:i te Prin ted : 08- D ec-9'.! Test Property Curves for ICDR5C Mix

230
!UNIT WEIGHT VS. %ACi !%AV BY RICE VS. %ACi
150 7
149 5
I
'
i • I
' .•
I Cl 8

~
149 I
u ' i I
I I
Q. 1.&a.5 5
I ~ I :E
~ 1'8
"' I ~ 4
~ 147.5
~ 147
•' '
if
~
::, 1'6.5 •' >
3

1'6
I I ~ 2

145.5
I
1
3.5 4 4.5 5 5.5 3.5 4 4.5 5 5.5
%AC By Wgt OI Agg. %AC By Wgt OI Agg.

jFLOW VS. %AC! jVMA BY RICE VS. %AC!

11 13.6
10.5 13.4
8 13.2
i I ~ 13
:E 12.1
~ 12.1
~ 8 .5 ~ 12.4
u:
8
7 .5
I
~ 12.2
~ 12
11 .I
•
T
7
I
3.5 4 4.5 5 5.5 11.l/;_5 4 4.5 5 5.5
MC By Wgt OI Agg . MC By Wgt OI Agg.

!STABILITY VS. %AC! ~FAC BY RICE VS. %ACI

3400 115
3300
I IIO
I ! C

~
15
. 3100 i 80
f 3000
I :E 75
T w
~ 2900
I ~
r
85
"i 2800
.
~ 2700
I, • >
m
() 80
:E 2800
! < 55
I I !i;
2500 50
i I
2 ~.5 4 4.5 5 5.5 ~ -5 4 4.5 5 5.5
%AC By Wgt OI Agg. MC By Wgt OI Agg.

tpicdr5c.wql
 231

APPENDIX G

INDUSTRY-WIDE DATA BASE FOR RESILIENT

MODULUS REGRESSION ANALYSIS

 232

Authors Agg.Type MR (psi) Pair ( % ) n70* Pen@77 ° F t (sec)

Boudreau CB l.840E+06 4 1. 771 84 0.1

Boudreau CB 4.090E+05 4 1. 771 84 0.1
Boudreau l.406E+06 4 1. 771 84 0.1
Boudreau 4.090E+05 4 1. 771 84 0.1
Boudreau l.017E+06 10 1. 771 84 0.1
Boudreau 2.llOE+OS 10 1. 771 84 0.1
Boudreau 1. 211E+06 10 1. 771 84 0.1
Boudreau 3.100E+05 10 1. 771 84 0.1
Boudreau 2.595E+06 4 1. 771 84 0.1
Boudreau l.882E+06 4 1. 771 84 0.1
Boudreau l.271E+06 4 1. 771 84 0.1
Boudreau 4.960E+05 4 1. 771 84 0.1
Boudreau 2.717E+06 4 1. 771 84 0.1
Boudreau 2 . 213E+06 4 1. 771 84 0.1
Boudreau l.686E+06 4 1. 771 84 0.1
Boudreau 7.710E+05 4 1. 771 84 0. 1
Boudreau l.768E+06 8 1. 771 84 0.1
Boudreau l . 189E+06 8 1. 771 84 0.1
Boudreau 7.250E+05 8 1. 771 84 0.1
Boudreau 2.800E+05 8 1. 771 84 0.1
Boudreau l.831E+06 8 1. 771 84 0.1
Boudreau l.421E+06 8 1.771 84 0.1
Boudreau 9.600E+05 8 1. 771 84 0.1
Boudreau 3.220E+05 8 1. 771 84 0.1
Schmidt & Graf G 3.650E+05 8.1 1. 818 83 0.1
Schmidt & Graf G 3.290E+05 6.3 1. 818 83 0.1
Schmidt & Graf G 3.500E+05 8.2 1. 818 83 0. 1
Schmidt & Graf G 3.230E+05 10.4 1. 818 83 0. 1
Schmidt & Graf G 2.030E+05 6.5 1. 818 83 0.1
Schmidt & Graf G 2.340E+05 7.3 1. 818 83 0.1
Schmidt & Graf G 2.170E+05 7.7 1. 818 83 0.1
Schmidt & Graf G 3.210E+05 5.5 1.818 83 0.1
Schmidt & Graf G l.OOOE+05 1. 6 1. 818 83 0.1
Schmidt & Graf G 2.070E+05 3.8 1. 818 83 0.1
Schmidt & Graf G 2.170E+05 10.4 1.818 83 0.1
Schmidt & Graf G 2.020E+05 13.8 1. 818 83 0.1
Schmidt & Graf G 1. 730E+05 15.4 1. 818 83 0.1
Schmidt & Graf G 4.080E+05 2.1 1. 818 83 0.1
Schmidt & Graf G l.760E+05 0.2 1. 818 83 0.1
Schmidt & Graf G 4.230E+05 6.1 1. 818 83 0.1
Schmidt & Graf G l.890E+05 10.2 1. 818 83 0.1
 233

Authors Agg.Type MR (psi) Pair (%) n70* Pen@77°F t (sec)

Schmidt & Graf G 6.120E+05 9.1 9.535 39 0.1
Schmidt & Graf G 4.150E+05 8.6 5.072 52 0.1
Schmidt & Graf G 3.000E+05 8.6 l. 818 83 0.1
Schmidt & Graf G 2.100E+05 7.9 0.795 121 0.1
Schmidt & Graf G 7.200E+04 8 0.171 244 0.1
Schmidt & Graf Ca 5.050E+05 4.6 1. 818 83 0.1
Schmidt & Graf Si l . 660E+05 8.8 l . 818 83 0.1
Schmidt & Graf Ca 5.700E+05 2.4 l . 818 83 0.1
Schmidt & Graf G 3.290E+05 6.3 l . 818 83 0.1
Schmidt & Graf Si l.664E+05 8.8 l . 818 83 0. 1
Schmidt & Graf Ca 5.540E+05 2.4 l . 818 83 0.1
Schmidt & Graf LS 2.920E+05 3.1 1.818 83 0.1
Schmidt & Graf Gran 2.750E+05 7.4 1.818 83 0.1
Str-Gar&Epps G 8.208E+06 2 4.669 54 0.1
Str-Gar&Epps G 4.354E+06 2 4.669 54 0.1
Str-Gar&Epps G 4.670E+05 2 4 . 669 54 0.1
Str-Gar&Epps G 4.300E+04 2 4.669 54 0.1
Str-Gar&Epps G 2.310E+05 8 4 . 669 54 0.1
Str-Gar&Epps G 5 . 600E+06 7 4.669 54 0.1
Str-Gar&Epps G 3.173E+06 7 4.669 54 0.1
Str-Gar&Epps G l.980E+05 7 4.669 54 0.1
Str-Gar&Epps G 3.200E+04 7 4.669 54 0.1
Baladi LS 7.600E+05 3.01 1. 321 96 0.1
Baladi LS 6 . 501E+05 3.09 l . 321 96 0.1
Baladi LS 5.463E+05 2.99 l . 321 96 0.1
Baladi LS 7 . 132E+05 3.35 l. 321 96 0.1
Baladi LS 6.480E+05 3.37 l . 321 96 0.1
Baladi LS 5 . 890E+05 3.29 l . 321 96 0.1
Baladi LS 5.056E+05 5.15 l . 321 96 0.1
Baladi LS 5.008E+05 5.1 l . 321 96 0.1
Baladi LS 3.825E+05 5.2 l . 321 96 0.1
Baladi LS 4 . 690E+05 6.85 l. 321 96 0.1
Baladi LS 4.810E+05 6.8 l . 321 96 0.1
Baladi LS 3 . 200E+05 6 . 89 l . 321 96 0.1
Baladi LS 4.688E+05 3 .11 l . 321 96 0.1
Baladi LS 4.400E+05 3.06 l . 321 96 0. 1
Baladi LS 3.460E+05 3 l . 321 96 0.1
Baladi LS 4.864E+05 3.48 l. 321 96 0.1
Baladi LS 4.442E+05 3.46 l . 321 96 0.1
Baladi LS 3.134E+05 3.45 l . 321 96 0.1
Baladi LS 5.526E+05 3.85 l . 321 96 0.1
Baladi LS 4.523E+05 4.8 l . 321 96 0.1
Baladi LS 3.818E+05 4.8 l . 321 96 0.1
Baladi LS 3.974E+05 7.15 l . 321 96 0.1
Baladi LS 3.919E+05 7.18 l . 321 96 0.1
Baladi LS l.684E+05 7.12 l . 321 96 0.1
 23 4

Authors Agg.Ty pe Peff (%) Pabs(%) Pb(%) P200(% ) AR4 ( %)

Boudreau CB 12.4224 1. 4025 6 5 44

Boudreau CB 12.4224 1. 4025 6 5 44
Boudreau 13. 8716 1. 5662 6.7 4.8 43
Boud rea u 13.8716 1. 5662 6.7 4.8 43
Boudreau 12.4224 1.4025 6 5 44
Bo u dreau 1 2 .4224 1.4025 6 5 44
Boudreau 13.8716 1. 5662 6.7 4.8 43
Bo u dreau 1 3.8716 1. 5662 6.7 4.8 43
Bo u drea u 1 2 .4224 1.4025 6 5 44
Boudreau 12.4224 1. 4025 6 5 44
Boudreau 12.4224 1.4025 6 5 44
Boudreau 12 . 4224 1.4025 6 5 44
Boudreau 13.8716 1.5662 6.7 4.8 43
Boudreau 13.8716 1. 5662 6.7 4.8 43
Boudreau 1 3.8716 1. 5662 6.7 4.8 43
Boudreau 13.8716 1. 5662 6.7 4.8 43
Bo u dreau 12.4224 1.4025 6 5 44
Boudreau 12.4224 1.4025 6 5 44
Boudreau 1 2 .4224 1. 4025 6 5 44
Boudreau 12.4224 1.4025 6 5 44
Boudreau 13. 8716 1. 5662 6.7 4.8 43
Boudreau 13. 8716 1. 5662 6.7 4.8 43
Boudreau 13. 8716 1. 5662 6.7 4.8 43
Boudreau 13.87 1 6 1. 5662 6.7 4.8 43
Schmidt & Graf G 10 . 3520 1.1688 5 7 37
Schmidt & Graf G 10.3520 1. 1688 5 7 37
Schm i dt & Graf G 10 . 3520 1.1688 5 7 37
Schmidt & Graf G 10 . 3520 1. 1688 5 7 37
Schmidt & Graf G 10 . 3520 1. 1688 5 7 37
Schmidt & Graf G 10.3520 1.1688 5 7 37
Schmidt & Graf G 10.3520 1.1688 5 7 37
Schmidt & Graf G 10.3520 1.1688 5 7 37
Schm i dt & Graf G 18.6335 2.1038 9 7 37
Schmidt & Gra f G 14 .4928 1. 6363 7 7 37
Schmidt & Graf G 1 0.3520 1 . 1688 5 7 37
Sc hm idt & Gra f G 6. 2112 0. 7013 3 7 37
Schmidt & Graf G 4.1408 0 . 4675 2 7 37
Schmidt & Graf G 1 0.3520 1. 1688 5 4 60
Sc hmi dt & Gra f G 1 4.4928 1. 6363 7 4 60
Sc hmidt & Graf G 6.2112 0 . 7013 3 4 60
Sc hm id t & Graf G 4.1408 0.4675 2 4 60
 235

Authors Agg . Type Peff(%) Pabs(%) Pb(%) P200(%) AR4(%)

Schmidt & Graf G 10.3520 1.1688 5 7 37

Schmidt & Graf G 10.3520 1.1688 5 7 37
Schmidt & Graf G 10.3520 1.1688 5 7 37
Schmidt & Graf G 10.3520 1. 1688 5 7 37
Schmidt & Graf G 10.3520 1.1688 5 7 37
Schmidt & Graf Ca 10.3520 1. 1688 5 7 37
Schmidt & Graf Si 10.3520 1.1688 5 7 37
Schmidt & Graf Ca 10.3520 1.1688 5 7 37
Schmidt & Graf G 10 . 3520 1.1688 5 7 37
Schmidt & Graf Si 10.3520 1.1688 5 7 37
Schmidt & Graf Ca 10.3520 1.1688 5 7 37
Schmidt & Graf LS 10.3520 1. 1688 5 7 37
Schmidt & Graf Gran 10.3520 1.1688 5 7 37
Str-Gar&Epps G 13.4576 1. 5194 6.5 5.8 47.8
Str-Gar&Epps G 13.4576 1.5194 6.5 5.8 47.8
Str-Gar&Epps G 13.4576 1. 5194 6.5 5.8 47.8
Str-Gar&Epps G 13.4576 1. 5194 6.5 5.8 47.8
Str-Gar&Epps G 13.4576 1. 5194 6.5 5.8 47.8
Str-Gar&Epps G 14.4928 1. 6363 7 4.1 47.9
Str-Gar&Epps G 14.4928 1. 6363 7 4.1 47.9
Str-Gar&Epps G 14.4928 1. 6363 7 4.1 47.9
Str-Gar&Epps G 14.4928 1. 6363 7 4.1 47 . 9
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4 . 31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50 . 2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50 . 2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4. 31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8 . 9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 l. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1. 0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
Baladi LS 8.9234 1.0075 4.31 8.3 50.2
 236

Authors Agg.Type AR3/4(\) ARl/2(\) T( °F)

Boudreau CB 0 2 40
Boudreau CB 0 2 73
Boudreau 0 1 40
Boudreau 0 1 73
Boudreau 0 2 40
Boudreau 0 2 73
Boudreau 0 1 40
Boudreau 0 l 73
Boudreau 0 2 40
Boudreau 0 2 so
Boudreau 0 2 60
Boudreau 0 2 70
Boudreau 0 1 40
Boudreau 0 1 so
Boudreau 0 1 60
Boudreau 0 1 70
Boudreau 0 2 40
Boudreau 0 2 so
Boudreau 0 2 60
Boudreau 0 2 70
Boudreau 0 1 40
Boudreau 0 1 so
Boudreau 0 1 60
Boudreau 0 1 70
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
 237

Authors Agg.Type AR3/4(\) ARl/2(\) T( °F)

Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 29 73
Schmidt & Graf G 0 29 73
Schmidt & Graf G 0 29 73
Schmidt & Graf G 0 29 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf Ca 0 9 73
Schmidt & Graf Si 0 9 73
Schmidt & Graf Ca 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf Si 0 9 73
Schmidt & Graf Ca 0 9 73
Schmidt & Graf LS 0 9 73
Schmidt & Graf Gran 0 9 73
Str-Gar&Epps G 0 5.3 -20
Str-Gar&Epps G 0 5.3 34
Str-Gar&Epps G 0 5.3 77
Str-Gar&Epps G 0 5.3 104
Str-Gar&Epps G 0 5.3 77
Str-Gar&Epps G 0 5.1 -20
Str-Gar&Epps G 0 5.1 34
Str-Gar&Epps G 0 5.1 77
Str-Gar&Epps G 0 5.1 104
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77
Baladi LS 0 18 77

CLS = Crushed Limestone; CB= Crushed Basalt; G = Gravel

Ca= Calcite Si= Silica Gran= Granite
LS= Limestone
 MoDOT

TE
5092
.M8A3
no. 90-5
SOURI COOPERATIVE HIGHWAY RESEARCH PROGRAM
FINAL REPORT
9 0-5
- v.2

DETERMINATION OF AASHTO LAYER COEFFICIENTS

Volume II
UNBOUND GRANULAR BASES AND
CEMENT TREATED BASES

MISSOURI HIGHWAY AND TRANSPORTATION DEPARTM ENT

FEDERAL HIGHWAY ADMINISTRATION

Property of

MoDOT TRANSPORTATION LIBRARY

-~-...--·----··-.
I
I DETERMINATION OF AASHTO LA YER COEFFICIENTS

I VOLUME II:

I UNBOUND GRANULAR BASES AND

CEMENT TREATED BASES
STUDY 90-5

I
I Prepared for

MISSOURI HIGHWAY AND TRANSPORTATION DEPARTMENT

by
DAVID N. RICHARDSON
PAUL A. KREMER
I
I DEPARTMENT OF CIVIL ENGINEERING
UNIVERSITY OF MISSOURI - ROLLA
ROLLA, MISSOURI
I
I in cooperation with
U.S. DEPARTMENT OF TRANSPORTATION
FEDERAL HIGHWAY ADMINISTRATION
I December 1994

I
I The opinions, findings and conclusions expressed in this
publication are not necessarily those of the Federal
Highway Administration.

I
I
I EXECUTIVE SUMMARY

I
I The purpose of this study was to determine the layer (a) coefficients for

I several commonly specified materials used in MHTD designed pavements for use in

the 1986 AASHTO Guide pavement design method. The project entailed a review
I and compilation of published literature, laboratory testing, analysis of results, and

I preparation of this report. The study was divided into two parts. Volume I

I presented the determination of layer coefficients for several Type C and 1-C asphalt

surface mixes and plant mix bituminous bases. This Volume II presents the layer

I coefficients determined for granular and soil-cement materials. Layer coefficients

I were shown to be a function of the resilient or static modulus for each material.

Resil ient moduli were determined for four sources of the unbound granular

I Types 1 and 2 base material (two crushed stones, two gravels) . In addition , static

I compressive chord moduli were determined for two cement-treated soils.

Addit ional tests included sieve analyses, hydrometer analyses, specific gravities,
I Atterberg limits, aggregate particle shape/surface texture determinations, standard

I and mod ified proctor densities, and vibratory table relative densities.

The granular base material resilient modulus testing program variables

I included compactive effort, degree of saturation, gradation (open and dense),

I deviator stress/confining pressure (bulk stress), and geologic source of aggregate

(reflecting differences in particle shape/surface texture). Within the confines of the

I testing program, it was found that the only variables significant to resilient
 ii

modulus were bulk stress, degree of saturation, and compactive effort. Results
I
indicated that for granular materials, resilient modulus decreases with decreasing

bulk stress, decreasing compactive effort, and increasing degree of saturation. A

regression model was developed for the prediction of resilient modulus of unbound

granular material which reflected the change in modulus in accordance not only

with the above variables, but with the base course position in the pavement

structure. Layer coefficients were then determined using the AASHTO Design

Guide nomograph method, and the Odemark method. The nomograph equations

supplied in the AASHTO Design Guide for determination of layer coefficient a2

(base course), and a 3 (subbase course) from resilient modulus results are

recommended for use. Because modulus is partially a function of the thickness of

the base course and its position in the pavement structure, a sliding scale for layer I
coefficients is necessary, and use of the above-mentioned regression equation is

recommended.

Static compressive chord moduli were determined for the two soil sources in

the soil-cement portion of the study. The major testing variables were cement

content and sand content of the fine-grained soil-sand mixtures. Regression

models were developed for the prediction of chord modulus. Layer coefficients

were then determined from the cement stabilized base AASHTO Guide nomograph

with knowledge of the chord moduli.

A verification ~nalysis was performed. Twelve hypothetical pavements were

designed using both the former MHTD method and the AASHTO method. The

layer coefficients developed in this study for unbound granular and soil-cement
I
I bases were used. This analysis verifed the use of the AASHTO nomographs for
iii

I calculation of layer coefficients in most cases.

I A sensitivity analysis was also performed. The results indicated that for the

granular material, a higher compactive effort could result in a requirement of a

I substantially thinner base or subbase in certain cases.

I
I
I
I
I
I
I
I
I
I
I
I
 iv

TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY • • • I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I t

TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... viii

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
GENERAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
OBJECTIVES AND SCOPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

UNBOUND GRANULAR BASE LAYER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
DETERMINATION OF LAYER COEFFICIENTS-METHODOLOGY . . . . . .
RESILIENT MODULUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
MATERIAL TYPES AND SOURCES . . . . . . . . . . . . . . . . . . . . . . . . . .
7
7
9
I
LABORATORY INVESTIGATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Gradation Curve Shape/Position . . . . . . . . . . . . . . . . . . . . . . .
Particle Shape/Texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Specific Gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
11
13
I
Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Moisture - Density Relationship . . . . . . . . . . . . . . . . . . . . . . . . 14
Relative Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Plasticity of Fines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Specimen Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Test Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Testing Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Test Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
RESULTS OF THE LABORATORY INVESTIGATION . . . . . . . . . . . . . . . 25
As-Received Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Gradation Curve Shape/Position . . . . . . . . . . . . . . . . . . . . . . . 26
Moisture-Density Relationships and Specific Gravities . . . . . . . . . 28
Relative Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Particle Shape and Surface Texture . . . . . . . . . . . . . . . . . . . . . 29
Plasticity of Fines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . · 33
Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
I
I ESTIMATION OF RESILIENT MODULUS . . . . . . . . . . . . . . . . . . . . . . .
V

44
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
I k 1 and k 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bulk Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
46
Corrected Bulk Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Resilient Modulus Regression Equation . . . . . . . . . . . . . . . . . . . 47
Granular Material k 1 and k 2 . . . . . . . . . . . . . . . . . . . . . . 50
Asphalt Modulus (E 1 ) . • . . . . . . • . . . • . . . . . . . . . . . . . 51
I Subgrade Modulus (E 5 g) . . . • . . • . . . . . . . . . . . . . . . . .
Model Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
59

I Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effects of Pavement Cross-section and Material Variables . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
62
62
DETERMINATION OF LAYER COEFFICIENTS . . . . . . . . . . . . . . . . . . . 66
Matching Layer Coefficient with Layer Position . . . . . . . . . . . . . 66
AASHTO Nomographs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

I Equivalent Stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Road Test Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
69
70
NCHRP 291 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Traylor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
NCH RP 128 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

I KENLAYER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Determination of MHTD Layer Coefficients . . . . . . . . . . . . . . . .
77
78
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

CEMENT-TREATED BASE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
I MATERIAL TYPES AND SOURCES . . . . . . . . . . . . . . . . . . . . . . . . . . 82
LABORATORY INVESTIGATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Soil Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
I Experimental Material Proportions and Moisture Density
Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Specimen Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
I Unconfined Compressive Strength . . . . . . . . . . . . . . . . . . . . . . 89
Modulus of Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
RESULTS OF LABORATORY INVESTIGATION . . . . . . . . . . . . . . . . . . 90
ESTIMATION OF STATIC CHORD MODULUS . . . . . . . . . . . . . . . . . . . 98
DETERMINATION OF LAYER COEFFICIENTS . . . . . . . . . . . . . . . . . . 101

I VERIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
COMPARISON TO PUBLISHED DATA . . . . . . . . . . . . . . . . . . . . . . .
101
101

I DESIGN VERIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Granular a 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Granular a 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
101
104
104

I
I
 vi

Soil Cement a 2 . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . • . . . 104

SENSITIVITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

UNBOUND GRANULAR MATERIAL . . . . . . . . . . . . . . . . . . . . . . . . . 107
SOIL CEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

FUTURE RESEARCH NEEDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

ACKNOWLEDGEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
APPENDIX A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
DATA SET FOR RESILIENT MODULUS . ... . . . . . . . . . . . . . . 130
I
vii
I LIST OF FIGURES
Page
I 1.
2.
Pavement Section with Unbound Granular Base . . . . . . . . . . . . . . . .
MHTD-Middle and New Jersey Experimental Gradations . . . . . . . . . .
. 6
. 12
3. Vibratory Compaction Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
I 4.
5.
Resilient Modulus Testing Equipment . . . . . . . . . . . . . . . . . . . . . . .
Typical Vibratory Table Test Result . . . . . . . . . . . . . . . . . . . . . . . .
. 19
. 30
6. Moisture-Density Relationships for MHTD Middle Unbound Granular
I 7.
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Typical Resilient Modulus Test Results . . . . . . . . . . . . . . . . . . . . . . . 34
8. Relationship Between Experimentally Derived Factors (k 1 and k 2 ) for
I 9.
the Theta Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Effect of Aggregate Source on Resilient Modulus . . . . . . . . . . . . . . . . 39

I 10.

11.
Effect of Degree of Saturation and Aggregate Source on Resilient
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Effect of Gradation, Degree of Saturation, and Compactive Effort on

I 12.
13.
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Definitions of Layer Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Relationship of Subgrade Resilient Modulus and Deviator Stress for
Three Types of Subgrade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
14. Observed vs Estimated Resilient Modulus of Unbound Granular
Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
15. Effect of Asphalt Layer Thickness and Stiffness on Base Resilient
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
16 . Effect of Granular k 1 and Subgrade Stiffness on Base Resilient

I 17 .
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Effect of Base Thickness and Asphalt Layer Thickness on Base
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

I 18.
19.
AASHTO Unbound Granular Base Layer Coefficient (a 2 ) Nomograph ... 67
AASHTO Unbound Granular Subbase Layer Coefficient (a 3 )
Nomograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
I 20.
21 .
Average AASHO Road Test Cross-Section . . . . . . . . . . . . . . . . . . . . . 73
Soil Cement Material Moisture-Density Relationships . . . . . . . . . . . . . . 86
22. Static Chord Modulus Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
I 23.
24.
Effect of Cement Content on Unconfined Compressive Strength . . . . . . 92
Effect of Cement Content on Chord Modulus . . . . . . . . . . . . . . . . . . . 93
25. Effect of Sand Content on Chord Modulus . . . . . . . . . . . . . . . . . . . . 94
26. Relationship Between Compressive Strength and Chord Modulus . . . . . 95
27. Overall Relationship of Unconfined Compressive Strength and Static
Compressive Chord Modulus for Soil-Cement Mixtures . . . . . . . . . . . . 96
I 28. Relationship of Estimated and Observed Chord Modulus of Soil-
Cement Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
29. AASHTO Nomograph for Determination of a 2 for Cement-Treated Base 100

I
I
 viii

LIST OF TABLES
Page
1. Reported Layer Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Range of Layer Coefficients at the AASHO Road Test . . . . . . . . . . . . . 3
3. Material Types and Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4. Test Sequence for Granular Specimens of Base/Subbase Material . . . . . 21
5. Testing Variable Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6. As-Received Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7. Experimental Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8. Gradation Shape Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
9. Experimental Gradation Slopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
10. Specific Gravity and Moisture Density Data . . . . . . . . . . . . . . . . . . . . 28
11. Particle Shape/Texture Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
12. Atterberg Limits of the Base Materials . . . . . . . . . . . . . . . . . . . . . . . 33
13. Resilient Modulus Test Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
14.
15.
16.
Significance of Variables to Resilient Modulus . . . . . . . . . . . . . . . . . . 43
Estimated Values of k 1 . . • . . . . . • . . . • . . . • . • . . . . . . . . . . . . . . . 49
Typical Input Soil Constants for KENLAYER Analysis. . . . . . . . . . . . . . 53
I
,
17. Elastic Moduli/Equations for AASHO Road Test Materials . . . . . . . . . . 72
18. Seasonal Moduli/Bulk Stress of AASHO Road Test Materials . . . . . . . . 75
19.
20.
21.
AASHO Road Test Moduli from Traylor . . . . . . . . . . . . . . . . . . . . . . . 76
Road Test Data Input to KENLAYER . . . . . . . . . . . . . . . . . · · · · · · · · 77
Subgrade Rutting Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
I
22. Backcalculation of a 2 Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
23. Comparison of AASHTO Nomograph and Odemark Stiffness Methods
of Layer Coefficient Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
24. Material Types and Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
25. Soil Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
26. Soil/Sand Mixture Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
27. Soil-Cement Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
28. Compressive Strength and Chord Modulus Results . . . . . . . . . . . . . . . 97
29. Soil-Cement Layer Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
30. Comparison of AASHTO Designs to MHTD Designs for a 2 . . . . . . . . . 105
31. Comparison of AASHTO Designs to MHTD Designs for a 3 . . . . . . . . . 106
32. Base Thickness Sensitivity to Changes in Compactive Effort. . . . . . . . 109
33. Subbase Thickness Sensitivity to Changes in Compactive Effort. . . . . 110
34. Thickness Sensitivity to Changes in Cement Content and Sand
Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
I
1

INTRODUCTION

GENERAL

This report presents the determination of AASHTO pavement design method

layer coefficients for several highway materials commonly specified by the Missouri

I Highway and Transportation Department (MHTD). The study was made at the

request of the MHTD Research Advisory Committee. The project was executed by

personnel from the University of Missouri-Rolla (UMR) Department of Civil

I Engineering.

The AASHTO ( 1) pavement design method was developed based on the

results of the AASHO Road Test. In this method, the highway designer determines

a "structural number" (SN) using such factors as design traffic level, subgrade

support, desired reliability, and desired terminal serviceability. The magnitude of

the SN reflects the degree to which the subgrade must be protected from the

effects of traffic. For example, a relatively high SN would indicate that a thick or

stiff pavement structure would be necessary to protect the subgrade from failing or

from causing pavement structural failure. Once the SN is calculated, it becomes

I necessary to determine the manner in which the SN will be achieved, i.e., what the

required thicknesses and quality of each pavement layer should be . This is done

by solving the following equation:

where:

I SN = structural number

I
I
 2

a 1 ,a 2 ,a 3 = layer coefficients for the surface, base, and subbase layers,

respectively I
D 1 ,D 2 ,D 3
drainage coefficients of the base and subbase, respectively

= thickness of surface, base, and subbase layers, respectively.

I
Drainage coefficients are essentially modifiers of the layer coefficients, and

take into account the relative effects of pavement structure internal drainage on

performance of the pavement. Determination of the drainage coefficients is

addressed under a separate contract in a separate report submitted by UMR to the

MHTD concurrent with this study(2).

A preliminary review of the literature indicates that reported values for layer

coefficients vary widely (Table 1 ). The range of layer coefficients determined at

the AASHO Road Test are shown in Table 2(3,6).

Table 1. Reported Layer Coefficients.

Layer Coefficient Material/Layer Value Ref.

a, asphalt surface 0.30 - 0.57 4-9

82 asphalt-treated base
cement-treated base
0.10 - 0.62
0.12 - 0.50
5-7,9
5-7,9
I
lime-treated base 0.12 - 0.26 4,6,7 I
unbound granular base 0.03 - 0.23 5,6,8-11

83 unbound granular base 0.02 - 0.15 6, 10

I
I
3

Table 2 . Range of Layer Coefficients at the AASHO Road Test.

Coefficient Minimum Maximum Reported

I a,

a2
0.33 1
0.12 2
0.78 1
0.23 2
0.44 1
0.14 2,0.07 3

I 1
a3 0.07 4 0.12 4 0.11 4

asphaltic concrete surface layer

I 2 unbound
3 unbound
4 unbound
crushed stone base
sandy gravel base
sandy gravel subbase

Examination of Eq. 1 indicates that the thickness of any particular layer is,

to a significant extent, dependent upon the layer coefficients. Hence, an accurate

I determination of layer coefficients can have a significant economic impact in

regard to the design of the pavement structure.

I It has been postulated that the magnitude of any layer coefficient is a

function of several factors. For example, the asphalt surface layer coefficient a 1 is

dependent upon mix characteristics, pavement temperature, vehicle speed, layer

I thickness, traffic level, and compacted mix stiffness. For an unbound granular

base, the layer coefficients a 2 and a 3 have been shown to be dependent on the

I state of stress in the layer, degree of saturation, compactive effort, aggregate

properties, base layer thickness, and subgrade stiffness.

As originally used in the AASHO Road Test analysis, layer coefficients were

actually regression coefficients which were the result of relating layer thicknesses

I to road performance under the conditions of the Road Test. The problem is to

translate the Road Test findings to other geographic areas where the construction

I
 4

materials and climate are different. Layer coefficients must be determined in order I
to use Eq. 1 for design purposes. In a pure sense, layer coefficients are abstract

mathematical entities. In a practical sense the layer coefficients must be related to

something tangible. Most commonly, layer coefficients are determined on the

basis of relative layer material strength or stiffness considerations. Over the years

since the AASHO Road Test, many methods have been used to determine values

for layer coefficients.

OBJECTIVES AND SCOPE

This study is based on methods which optimize the combination of

economy, accuracy, and length of study. In brief, the study entails determination I
of stiffness values for several commonly used MHTD types of pavement materials.

The stiffness values were determined by both direct laboratory modulus testing

and by approximation techniques. These stiffness values were related to layer

coefficients and then verified for reasonableness by comparing several pavement

structures designed with the AASHTO method to structures designed with the

former MHTD method. The report includes a method suitable for use in routine

design which will enable the pavement designer to solve Eq. 1 and hence obtain

the desired layer thicknesses.

The approach taken for layer coefficient determination was the traditional I
one (5, 12), which is to take some measure of strength, stability, or stiffness of a
I
particular mix or blend and compare it to the same type of measure for the

counterpart AASHO Road Test material. The comparisons are usually done by use

of one of the AASHTO Design Guide charts or some form of a ratio of the two

values. The traditional approach does not specifically address rutting.

I
I 5

I In this study, the materials for which layer coefficients were determined

were limited to two types of asphalt surface mixes (Types C and IC), one type of

I bituminous base mix, two types of cement-treated base mixes, and one type of

I unbound granular base/subbase. The report is separated into two volumes:

Volume I( 12) covers asphaltic materials; Volume II deals with unbound granular

I and cement-treated materials .

I UNBOUND GRANULAR BASE LA YER

INTRODUCTION

The main thrust of this portion of the study was to determine base and

subbase layer coefficients (a 2 and a 3 ) for Types 1 and 2 unbound aggregates

which are used as base material under an asphaltic surface layer or as subbase

under asphaltic surface and base layers. Fig. 1 shows the pavement section being

I modeled. The analysis was based on repeated-load triaxial resilient modulus tests

as recommended in the 1986 AASHTO pavement design guide ( 1). A secondary

I goal was to develop a regression equation to enable the prediction of resilient

I modulus without having to actually perform the test.

I Pavement engineers are interested in longevity of the pavement. Longevity

is a function of pavement durability and response to load. Usually, the pavement

is designed structurally to give a favorable response to load. Durability is assured

through specification of good materials. The elastic response to load of any

material is a function of its stiffness , as defined by some sort of load -deformation

I modulus.

I
I
 13.57 in. I ()

4.Sk 4.Sk

p = 100 psi - - -

AsphGl t
Bound
LGyer D1 El AJ 1

l 'nbound
I t
Gro.nulo.r
L0yer
It> \.J n \.J <'> \...! <'> .._., C-"'> '..J n \.J n '-' <'> .._., r> \.J C-"'> '-' n \.J n
D2
E2 /u 2

Subgro.de E3 AJ 3

Fig. 1. Pavement Section with Unbound

Granular Base .
en
I 7

Thus, direct measurement or estimation of modulus becomes desirable. This has

given impetus to the effort by the pavement industry to produce a type of modulus

I test that is suitable for practical use.

I DETERMINATION OF LAYER COEFFICIENTS-METHODOLOGY

Two methods were used to determine the layer coefficients (a 2 or a 3 ) of

I Type 1 and Type 2 unbound granular base materials:

I 1) Direct use of the base and subbase layer nomographs in the 1986 AASHTO

guide ( 1). The necessary input is the resilient modulus of the granular

material (E 9 ). However, resilient modulus of an unbound granular material is

dependent on many factors, including stress state. Thus, use of an elastic

layered stress analysis was necessary to predict stress states.

2) Use of Odemark's transformation. Stiffness (in the form of resilient

I modulus) of MHTD aggregate types was related to the modulus of Road Test

material at various stress states in a manner similar to that done with the

asphalt mixtures (12). Layer coefficients were calculated from Eq. 2,

I hereafter called the Odemark equation:

I a (MHTD) = a (AASHO) [ Eg(MHTD) ]

1/3
. . . . . . . . . . (2)
n n E9 (AASHO)

RESILIENT MODULUS

As can be seen from the above, the resilient modulus (E 9 ) is the key

parameter. The factors that most significantly affect unbound granular base

resilient modulus are stress state, compactive effort, degree of saturation,

 8
gradation, and particle shape. Of these variables, stress state is the dominant

factor in regard to material stiffness. The higher the confining pressure, the higher

the modulus. The significance of the other parameters seems less well defined.

Higher relative density, or compactive effort, increases modulus, but to varying

degrees ( 10, 13-15,22) . The significance may be influenced by gradation . A

higher degree of saturation has been shown to significantly decrease modulus

(10,13-18,21,22). For some aggregates, this effect is of minor significance

( 10, 18) . It appears that, as fines content increases, resilient modulus decreases

( 11, 13, 14, 19). Perhaps with higher fines content, the influence of a high degree

of saturation is more pronounced, possibly due to the generation of pore pressures

( 13) . Thus, there may be an interaction between the amount of fines and their

moisture content. This may be the reason that several studies have shown that

open-graded mixtures have higher modulus values than their dense-graded

counterparts. However, several studies have shown the reverse( 13,20), or at least

that gradation is of negligible significance (17), and others have shown that there

is an optimum fines content for maximum modulus(11,22). I

In regard to particle shape/texture, a more angular/rough aggregate generally

exhibits a higher modulus (14,20), although, in some cases, the reverse seems to

be true(20,23). The effect has been shown to be variable in significance( 13), of

minor significance ( 14, 20) and of major significance( 14). Thus the influence and

significance of particle shape is not well defined.

Several of the factors that are significant to the magnitude of resilient

I
I 9

modulus were studied in this project. These include: the effect of stress state,

compactive effort, degree of saturation and particle shape/texture. The effect of

I gradation was not in the scope of this study. However, in the companion project

(2), a different gradation was used for the same materials which were used in the

present study. The results of modulus testing from both projects were examined
I jointly. Thus a study of the effect of gradation was accomplished and is reported

herein.

MATERIAL TYPES AND SOURCES

I All unbound aggregates in the study were MHTD-approved materials. The

materials were selected and sampled by MHTD personnel. Two Type 1 crushed

stone base aggregates were studied. They were selected by MHTD personnel to

give a wide range of particle shape and texture. Additionally, in the companion

project (2), two Type 2 gravel materials were tested for resilient modulus. Test

I results for these two gravel materials are also included in this report. The

materials, sources, and identification are shown in Table 3.

Table 3. Material Types and Sources.

Nomenclature Material Sources Location

DR-12 Type 1 crushed Burlington limestone Mertens Quarry Millersburg

DR-13 Type 1 crushed Jefferson City Smith Quarry Rolla

dolomite

DR-14 Type 2 Crowley Ridge gravel base Delta Aggregates Dexter

DR-15 Type 2 Black River gravel base Williamsville Stone Poplar Bluff

I Note : All sources are located in Missouri

Co .
 10

LABORATORY INVESTIGATION

General
I
The principal property determined for unbound granular base materials was

the resilient modulus. However, the performance of other tests and procedures

was necessary in order to ultimately conduct the resilient modulus testing and

analyze the results. These other procedures included sieve analyses, gradation

formulation, specific gravity determination, moisture-density relationship testing,

particle shape/texture testing, and determination of plasticity of fines. These

operations are outlined below.

Gradations

The experimental gradation utilized in this study was characterized as a

curve situated midway between the upper and lower limits of the allowable

gradation specification band for MHTD Type 1 unbound base material. This

gradation was used for both the two crushed stones and the two gravels. At the

finer size end, the gradation was extended to include 8% passing the #200 sieve.

This value was chosen because 1) it matched one of the gradations used in the

drainage study which is the companion project to the present study, and 2) this

was approximately the same percentage as both as-delivered gradations of the

Type 1 aggregates supplied by MHTD for this project. The gradation plot is termed

the "MHTD-Middle" and is shown in Fig. 2.

In the companion study for the determination of drainage coefficients (2), an

open gradation (the so-called New Jersey gradation) was also tested for resilient
I 11

I modulus. The results of that testing is also herein reported. The New Jersey

gradation is also shown in Fig. 2.

I Gradation Curve Shape/Position

An analysis was performed to determine the effect of gradation upon

resilient modulus. The most promising methods were later tried in the

development of the E9 predictive equation. To accomplish this, there was a need

to characterize the gradations so that a single parameter would represent the shape

and position of each gradation curve. Nine different methods to determine the

I gradation parameter were tried. These are described in detail in Volume I of this

report, and are discussed in the "Results" section of this report.

Particle Shape/Texture

Numerous test methods have been devised to quantify particle shape and/or

J texture. These can be divided into two catagories: 1) direct methods (those that

result in measurement of aspects of individual particle shape or texture) and 2)

indirect methods (those that measure some sort of bulk aggregate property, such

as void content, which is related to particle shape/texture). Recent evaluations of

these methods were reported by Kandhal et al. (24) at NCAT (National Center for

Asphalt Technology). From the literature, it appears that efforts are being

concentrated in the area of fine aggregate evaluation and that there are several

methods available which can be used in lieu of the standard test, ASTM D 3398

(25) which is somewhat cumbersome to perform. Kandhal .fil fil. recommended the

National Aggregate Association's (NAA) proposed method (A or Bl for fine

 11

I
100 ,,

,.
90 /I

,,.,
1

di

....-..
80

70
lY

T'\
O.A . J.J.J.l. . J.J.J.

! J. -
LJ ~J.J. C)J.\, 1

r,
T ! -
J..JJ.J.J. ~ l
I' l
·I '
I
I
, -
I
/
I ~
-
[)!~y

' .
Tl::::"Tr.:,rc;-v

.T~ ·r=
T:iT IJ.~

I
1\/ 1 IP // l -'
~ .. I / = ---.......----.-----~
..__, /1 /
(~ /
/
60 ,, /
tlJ) / .,
-,, /
.....C #/
/
/
[l'.l
[l'.l
50 "' /
/" 1
cx:l /I
/, I
0...
, ," .. I
40 ,
.-I / I
,
....,a:! [ l
/
I
I

0 30 /
, I
I

E-<
I
"' I
I

,-( I
y /
20

I
I
I
I I
,0 I
I

I
/
chn·'
,.
0
200 50 so Ie B 4 :1/B In S/4 in I In 1. 75 in :I in
100 1/2 in

Sieve Sizes
0 Middle
(0.45 Power)
D New Jersey

Fig. 2. MHTD - Middle and New Jersey Experimental Gradations.

I
 13

aggregate (26). Both the standard ASTM method and the proposed NAA method

are indirect methods of particle shape determination.

I In this study the (-) #8 to ( +) #100 sieve size fraction of each aggregate

mixture blend was tested using the NAA Method A. The methodology is given in

Appendix A of Volume I of this study. For the ( +) #4 size, the blends were tested

in accordance with ASTM D 3398. This method is recorded in Appendix B of

Volume I of this study. The results of both methods were used in developing the

resilient modulus regression equations discussed later in the "Results" section of

this report. Photographs of the NAA test device and the D 3398 equipment are

shown in Figs. 9 and 10 of Volume I of this report ( 12).

Specific Gravity

Aggregate fractions of each of the two gradations were separated at the #4

I and #100 sieve sizes and tested in accordance with AASHTO T85-88 (27) and

T84-88 (28) for the ( + )#4 material and the (-)#4 to ( + )#100 material,

respectively. These data were necessary for use in the degree of saturation

calculations. Weighted averages of apparent specific gravities were used to

calculate the specific gravity for each gradation of each of the four aggregates as

follows:

100
G = --,------------ (3)

I % Passing #4 + % Retained #4 · · · · · · · · · · ·
ASG ASG

where:

G = apparent specific gravity weighted average

 14

ASG = apparent specific gravity of each fraction.

Screening

All aggregates were shaken in an air dry state through the appropriate

screens in a Gilson shaker. A dust baffle/cover was designed to restrict the

movement of particles in order to minimize problems with incorrect sizes of

material being retained on any given sieve.

Upon shaking, the split material was stored in 20 gal plastic cans with lids

until the aggregate was needed for specimen fabrication.

Moisture - Density Relationship

In order to choose target densities for compaction of resilient modulus

specimens, standard and mod ified proctor tests were performed in accordance

with AASHTO T-99 (29) and T-180 (30). Additionally, the maximum density of

the open-graded gradation was determined via the vibratory table method (ASTM

D4253) (31) . For each of the four aggregates, a double amplitude vs dry density

curve was obtained in accordance with the dry method to obtain the optimum

power setting. This power setting was then used for the determination of density

utilizing the wet method. The vibratory table is shown in Fig. 3.

Relative Density

For data analysis purposes, it was necessary to determine the relative

density (32) of all Eg test specimens. The equation to calculate relative density is:
 15

Fig. 3. Vibratory Compaction Table.

I
I
 18

accordance with Claros fil fil. (35), a 1 /a3 ratios were not allowed to exceed three

in order to prevent possible excessive specimen straining.

Equipment. The testing equipment setup is shown in Fig. 4. The equipment

consisted of an MTS servo-hydraulic load system, a triaxial chamber capable of

housing a 4 in diameter specimen while subjected to cyclic loads, and a data

acquisition system. Load was measured with an internal 1000 lb capacity load cell

and deformation was measured with two LVDT'S mounted externally to the cell.

This type of measuring system is allowed in the AASHTO method and is

recommended in the SHRP method. Minimum resolution of the vertical LVDT's

and the load cell were in accordance with the AASHTO specifications. Actual

minimum deformations and loads during the testing were kept at least ten times

the minimum resolutions to assure confidence in the test results. Air was used as

the confining fluid instead of water in order to protect the internal load cell.

Triaxial cell pressure and back pressure were controlled via a Geotest control panel.

The Research Engineering triaxial cell that was used has several advantages. First,

the chamber cylinder wall can be placed after the loading piston is brought into

contact with the specimen. Also, end caps can be purged of air very easily by the (

unique design of the caps.

Test Variables. Four test parameters were controlled as independent variables.

a) Stress State . As previously mentioned, several variables affect the modulus of

I
 19

Fig 4. Resilient Modulus Testing Equipment.

 20
granular materials . Stress state is considered to be the most important. As in

shear strength, the more confined a granular material is, the higher will be the

modulus . In the field , confinement is supplied by: 1) the layer underneath the

granular material , 2) the granular material itself in the lateral (tangential and radial)

(a 2 and a 3 ) direction, 3) the overburden above the point of interest, and 4) the

momentary load from a vehicle. In a triaxial test, the difference between ·total

vertical stress (a 1 ) and the confining pressure (a3 ) is called the deviator stress or

stress difference (ad) . Cell pressure supplies the lateral confinement to the

specimen (a2 and a 3 ). A small static load (0.1 ad) supplies the "overburden"

pressure, and cyclic deviator stress (0.9 ad) supplies the "vehicle" momentary

stress. All of the stresses combined are known as the bulk stress .

0 = a1 + a2 + a3 where a 2 = a3 and a 1 = ad + a3

= (ad + a 3) + 2a3
= ad + 3a3

E9 is calculated as 0 .9 ad/frecoverable· For each specimen, resilient modulus

was determined at 15 stress states where effective confining pressure ranged from

2 to 20 psi and ad varied from 2 to 40 psi. This resulted in a range of bulk stress

from 8 to 100 psi. This was considered adequate to cover the range of stress

states likely to be encountered under field conditions. For instance, as will be

shown later , the weighted average 0 in the crushed stone base at the AAS HO

Road Test was about 14 psi. The testing sequence and stress state schedule is

shown in Table 4.
 21

Table 4. Test Sequence for Granular Specimens of Base/Subbase Material.

Phase Sequence Deviator O' 1 Confining a 1/a3 0 No. of

No. Stress Pressure Repetitions
(ad)(psi) (psi) **
*
Specimen 1 15 35 20 1.75 75 1000
Conditioning
2 10 30 20 1.5 70 50
3 20 40 20 2.0 80 50
4 30 50 20 2.5 90 50
5 40 60 20 3.0 100 50
6 10 25 15 1.67 55 50
7 20 35 15 2.33 65 50
8 30 45 15 3.0 75 50
Testing 9 5 15 10 1.5 35 50
10 10 20 10 2.0 40 50
11 20 30 10 3.0 50 50
12 5 10 5 2.0 20 50
13 10 15 5 3.0 25 50
14 5 8 3 2.67 14 50
15 2 4 2 2.0 8 50
Note: 1 psi = 6.9kPa
* Cyclic loads = 0.9 ad; constant contact loads = 0.1 ad
** For all stress states the minimum number of repetitions necessary is 50. The
maximum is determined as per the AASHTO method and were redetermined
for each confining pressure.

b) Degree of Saturation. In general, an increased water content will cause

modulus to decrease. Several degrees of saturation ( 0 S) have been put forth by

others as break-points in behavior. Granular base materials are considered to be

 22
relatively "dry" at degrees of saturation 60 percent and less (10,22). AASHO

Road Test granular base materials suffered a marked increase in distress above 85

percent saturation. Resilient behavior has been shown to deteriorate above 80 to

90 percent saturation (22). In the present study, each material was tested at two

degrees of saturation: approximately 60% and 100%.

c) Degree of Compaction. As previously discussed, modulus generally increases

with higher levels of compaction. Two levels of compactive effort were evaluated

for each material and gradation. For the dense gradation, specimens were

compacted to 100% standard and 100% modified proctor densities. For the New

Jersey gradation, one level of compaction corresponded to the maximum index

density via vibratory compaction (wet method), while the lower level of density

usually corresponded to an impact-type of compaction, such as 100% standard

proctor.

d) Particle Shape/Surface Texture. As stated earlier, the effect of particle

shape/surface texture is not well-defined. Two crushed stones and two gravels

from different geological formations were chosen for delineating the effect of

particle shape/surface texture.

Testing Scheme. The testing scheme involved the following variables: two

sources of two particle shapes, two compactive efforts, two gradations and two

degrees of saturation for a total of 32 "tests". Each test was run with duplicate

specimens . The testing scheme is shown in Table 5.

 23

Table 5. Testing Variable Scheme.

Crushed Stone Gravel

DR-12 DR-13 DR-14 DR-15

Mid. NJ Mid. NJ Mid. NJ Mid. NJ

CEL 0 S = 60 X X X X X X X X
0 S = 100 X X X X X X X X
CEH 0 S = 60 X X X X X X X X
os = 100 X X X X X X X X
Note: Mid. = middle of MHTD Type 1 gradation band
NJ = New Jersey gradation
CEL = lower compactive effort
CEH = higher compactive effort
0 S = 60 or 100% saturated

Test Procedure. The resilient modulus testing procedure involved the following

steps: specimen compaction, assembly of the triaxial cell, consolidation, specimen

conditioning at a given stress state, load application through 14 additional stress

states at 60% saturation, backpressure saturation to 100% saturation,

consolidation, and load application through 14 stress states at 100% saturation.

After the load application in the 100% saturation step, the specimen was tested

for permeability. The permeability results were used in the drainage project that

was concurrent with the layer coefficient project. As a final step, the specimens

were allowed to drain overnight in order to calculate their effective porosities.

The specimens were compacted in eight layers of equal thickness with a

hand-held air hammer. The material was compacted at about 60% saturation into

a split mold. After cell assembly and consolidation, the specimen was conditioned
 24
with 1000 repetitions. Conditioning is used to eliminate the effects of any

specimen disturbance due to specimen preparation procedures. It also aids in

minimizing the effects of initially imperfect contact between the end platens and

the test specimen. The various stress states and loads were then applied in

accordance with Table 4. The number of load applications varied from 50 to 200,

depending on the number of applications necessary to meet the AASHTO modulus

repeatability requirements.

Load and deformation data were taken for every load application over the

entire sequence, but only the last five repetitions were used for calculation of

resilient modulus.

The load duration for each repetition was 0.1 sec followed by 0.9 sec rest.

The stress pulse shape was haversine in nature. Repeated load equipment

deflection was determined through the use of an aluminum dummy specimen and

was subtracted from total deflections for each stress state. Initially, calibration of

the load cell and LVDT's was performed before each test, but the interval was

increased upon determining that the drift in calibration was insignificant. The

change in specimen height was constantly monitored. None of the specimens

approached the maximum allowable permanent strain of 5 % .

In an effort to determine the effect of drainability on pavement bases, the

tests at 60% saturation were performed in a drained condition while the 100%

saturation tests were run in an undrained state.

 25
RESULTS OF THE LABORATORY INVESTIGATION

As-Received Gradations

The as-received gradations of the four granular materials are shown in Table

6. The two experimental gradatons are shown in Table 7.

I Table 6. As-Received Gradations.

Sieve Percent Passing

Size
DR-12 DR-13 DR-14 DR-15
1 in 100 100 100 100
1 /2 in 96 83 72 83
#4 68 50 46 50
16 -- 26 37 26
40 24 18 17 12
100 12 13 2 5
200 8 7 1 4

I
I
 26
Table 7. Experimental Gradations.

Sieve Size % Passing

Middle NJ
3 in. 100 100
1 1 /2 100 100
3/4 (95) (95)
1/2 75 68
3/8 (63) (58)
#4 50 47
8 (40) (20)
16 33 5
30 (28) (4)
40 25 3
50 (22) (2.5)
100 (16) (2.5)
200 8 2
() = estimated from plot

Gradation Curve Shape/Position

In an attempt to determine the effect of gradation on resilient modulus, a

parameter that described gradation curve shape/position was required. Nine

different methods were explored. Sieve size data from both experimental gradation

curve types (Middle and NJ) were used for calculation of various parameters. The

parameters were then used in the development of the multiple regression model of

resilient modulus to see which method increased the accuracy of the model the

most. This was judged from the adjusted - R2 statistic of the equation. The nine
I 27

methods or parameters were as follows: fineness modulus (FM), coefficient of

uniformity (Cu), coefficient of skew (Cz), surface fineness (SF), specific surface

-
factor (SSF), a combination of SF and SFF (SF/SSF), Hudson's A, slopes-of-

gradation-curve, and the percent passing or retained on individual sieves. Values

for these parameters are shown in Tables 8 and 9. The parameters are discussed

in Volume I of this report. The slopes-of-gradation-curve (M) method was altered

from that which was in Volume I to better match the natural break points of the

granular base experimental gradation curves. The slopes of each curve were

determined between the 1 in and #4 sieve, the #4 and #16 sieves, and the #16

and #200 sieves.

Table 8. Gradation Shape Results.

Mid. NJ
FM 4.53 5.66

cu 82.6 5.29

CZ 1.19 0.21
SF 1588 1938
SSF 294.4 65.8
SF/SSF 5.40 29.4
- 4.55 3.36
A
Mn-n ----- -----
3/4, #4, 16, 200 ----- -----

The results of the regression model analysis indicated that it made very little
 28
difference as to which gradation parameter was used (except A was somewhat

less effective, and no more than one "M" could be used at a time due to

collinearity problems).

Table 9. Experimental Gradation Slopes.

Gradation M1.4 M4.15 M15.200

Middle 94.3 78.0 138.9
NJ 100.0 192.7 16.7

Moisture-Density Relationships and Specific Gravities

Moisture-density relationship and apparent specific gravity information was

determined in regard to the two test gradations for each of the four granular

materials. The data are shown in Table 10.

Table 10. Specific Gravity and Moisture Density Data.

Material Gradation Apparent Sp. Density,

minimum (pcf)
T-99 T-180 04253
I
Gravity

MOD (pcf) OMC (%) MOD (pcf) OMC

(%)
MOD
(pcf)
OMC
(% )
I
113 .2 136.5 7.0 137.6 7.4 .. ..

I
DR- 12 MHTD M id 2 .69

NJ 2 .69 101 .7 .. .. 131.2 9 .0 1 21 .1 12 .2

DR-13 MHTD Mid 2.78 114.3 138.3 7 .7 141 .0 5 .9 .. ..

DR - 14
NJ

MHTD Mid
2 .78

2.65
100.8

109 .9
..

132.5
--
7.8
135.2

134.5
8 .3

6.7
124. 1

..
13 .0

..
I
DR - 15
NJ

MHTD Mid
2.65

2.65
99.9

110.2
119 .0

134.4
8.0

7 .6
..

136.9
..

6.1
122 .8

..
12 .0

..
I
.. ..

I
NJ 2 .65 96.2 109 .5" • 8 .9 114.3 14.0

Note: T-99 = Standard proctor

T-180 = Modifi ed proctor
D4253 = Vibratory table
• • T-99 , T- 180, and D4253 densities were very close. To get a wider difference in values, a surcharge weight was used which was
different than that which was used to obtain the other 04253 data I
 29
Some difficulty was experienced with performing the impact-type of moisture-

density tests for certain materials when graded according to the New Jersey

gradation, therefore, the vibratory table density (wet) method was also performed.

It is believed that this compaction method gives a density value more comparable

to that which will be achieved in the field. The tests were performed in a dry state

at different power settings to determine the optimum setting which would result in

the highest density. The test was then run with the material in a moisture state

which was more in line with field compaction conditions. A typical vibratory table

test result is shown in Fig. 5. In Fig. 6 are shown the moisture-density

relationships for each of the four granular materials.

Relative Density

Minimum density test results are also listed in Table 10. The minimum and

maximum density values were used in Eq. 4 to calculate the relative density of

each E9 specimen, as listed later in Table 13. Relative density values give a wider
I spread of values than "percent of T-180 values", and therefore are more useful in

I analysis of behavior.

Particle Shape and Surface Texture

Particle shape/texture characteristics were quantified by use of ASTM D

3398 for the ( +) #4 sieve material and by NAA Method A for the (-) #8 through

( + l #100 material for each aggregate source. Both are measures of void content
of bulk aggregate. Void content has been shown to be related to particle

shape/texture. D3398 results in a "Particle Index" {IP) while the NAA Method A
 30

123

122
--
-
C')
~ 0
........__ 12 1
..... Wet
..0
.....
..__, 120 0
~

..c:
·-;,::
tll)
( l)
1 19 -

·- ~

i::
;:J
1 18

>. 1 17
i...
Q

1 16 ;- Sample: DR-12 New Jersey

1 15
0.008 0.010 0 . 012 0.014 0.016 0.018

Double Amplitude of Vibration (in)

Fig. 5. Typical Vibratory Table Test Result.

 31
145 145

-.... 1 40
T180
~140

~
t".)...,
...,
....
-
.........
..0
...J
..., 135
-
.........
..0
.:::. 135
TIIII
..c::
Oil
Q)
Q)
;a,: 1 30 )I:: 130
...,
.... ...,
c:: ....

I ::i

0
t' 125
Sample:
Optimum
Optimum
Max. Dry
DR-12 Middle
H2 0 Content: ? .Olt (T99)
H 2 0 Content: ?.O: (T\80)
Unit 'Wt.: 138.6 Lbt/tt 3 (T99) 0
Cl
::i
>. 125
I-.
Sample:
Optimum
Optimum
DR-13 Middle
H2 0 Content: 7 . 7'% (T99)
H2 0 Content: 5 .9% (T\BO)
Ma:,:. Dry Unit 'Wt .: 136.3 Lb! / !t 3 (T99)

I 120
5
Max. Dry

6
Unit Wt .: 13?.8 Lb!/!t (T180)

7 8 9
120
4
Mu. Dry

5
Unit 'Wt. : 141.0 Lb! / ft (TlBO)

6 7 B 9 10
Water Content (%)
Water Content (%)

I
145

Sample: DR-14 Mi ddl e

-- 140
Optimum H 2 0 Content: 7 .8 (T99)
Opt imum H 2 0 Content: 6 .7 (T1BO)
-- 140

I
M..., M_,

.........
......0 -
.........

...J
'--'135 -
..., -..c
...J
..., 135

I ..c::
....
bO
Q)
T180 ..c::
....
QI)

Q)

Ii=: 130 )I:: 130

..., .......,
I Cl
::i TIIII
Cl
::i
Sample: DR-15 Middle
t' 1 25 t' 125 Optimum H2 0 Content: 7.6 (T99)

I 0
Ma:,:. Dry Unit ll't.:
Max . Dry Unit ll't .:
s
132.5 Lbf/ft 3 (T99)
134.5 Lbf/ft (T160)
Q Optimum
Mu. Dry
Max. Dry
I
H 2 0 Content: 6 . 1 (Tl\O)
Unit wt .: 13'.4 Lbf/ ft 3 (T99)
Unit Wt.: 136.9 Lbt/ ft (TISO)
120 12 0
0.0 0.1 0.2 4 5 6 7 8 9
Water Content (%) Water Content (%)

Fig. 6. Moisture - Density Relationships for

MHTD Middle Unbound Granular Materials.
I
I
 32

gives an "Uncompacted Voids Percent" (U). The results are shown in Table 11.

Table 11. Particle Shape/Texture Results.

Aggregate Particle Index (IP) Uncompacted Voids (U)

%
DR-12 12.5 43.6
DR-13 11.9 45.3
DR-14 10.0 41.2
DR-15 10.4 40.6

Round, smooth particles give IP's of 6 or 7, while angular, rough particles result in

values of more than 15. The range of IP's of the aggregates in this study was

10.0 to 12. 5. The Particle Index was determined for the coarse aggregate fraction

of each gradation and the Uncompacted Voids content was determined for the fine

aggregate fraction.

Looking at Particle Index and especially the Uncompacted Voids values, the

crushed aggregates were somewhat more angular than the gravels, as expected,

but the ranges were limited.

Plasticity of Fines

The results of the Atterberg Limits testing are shown in Table 12. All four

aggregates were essentially non-plastic in nature.

 33

I Table 12. Atterberg Limits of the Base Materials.

I Aggregate Liquid Limit Plasticity Index

DR-12 16 NP
DR-13 18 NP
DR-14 22 NP
I DR-15 19 NP

Resilient Modulus

Resilient modulus tests were performed on two crushed stone aggregates

(DR-12 and DR-13) and two gravels (DR-14 and DR-15) at two degrees of

saturation (approximately 60 and 100%), two compactive efforts (low and high),

two gradations (MHTD "Middle" and New Jersey), and 14 stress states with

duplicate samples, for a total of 896 tests. Because the same 14 stress states and

the two degrees of saturation were used for each specimen, there were only 32

I specimens. The testing sequence for each specimen is shown in Table 5. Fig. 7 is

I a typical plot of bulk stress vs resilient modulus. Each data point is representative

of one stress state. As can be seen, modulus increases with an increase in stress

state. The classic equation of the line (known as the "theta model") is:

log Eg = log k 1 + k 2 log e ............... (5)

I or
 34

+J
~

-
....
Q,)

....
I'll
Q,)
~

1 10 100

Bulk Stress (psi)

Fig. 7. Typical Resilient Modulus Test Results.

I
I 35

I Eg = k1 ek2 . . . . . . . . . . . . • . . . . . . . (6)

I where:

k1 = intercept of Eg at e = 1 psi, log-log plot

I k2 = slope of line, log-log plot.

I The results of all resilient modulus testing are tabulated in Table 13.

Fig. 8 shows the relationship of coefficients k 1 and k 2 as reported by Rada

I and Witczak (22), with the results of the present study also plotted. As can be

I seen, the present study data falls in the range of data that has been reported

elsewhere.

Fig. 9 shows the effect of aggregate source on Eg. It appears that in the

range of stresses encountered in highway pavements, there is no clear trend in the

effect of aggregate source on Eg. This may be a result of the limited range of

particle shapes/textures of the materials used in this study.

I The effect of increased degree of saturation is shown in Fig. 10. The

general trend is a loss of Eg as the degree of saturation increases from a moist

I state to a saturated state. This is similar to the trend reported by Rada and

I Witczak and others.

I The interaction between gradation, compactive effort, and degree of

saturation is shown in Fig. 11. It appears that open-graded material suffers

I somewhat more of a loss in E9 than dense-graded material as indicated by steeper

I curve slopes. The percent loss for the dense graded materials with low and high

compactive efforts was 9.8 and 11 .0, while the loss for the open graded materials
I
I
Table 13 . Resilient Modulus Test Data.

Material Gradation/CE Dry Density (pct) MADD DR 0 Sat . k, k2 En (psi)*

(%) (psi)
Target As-tested (pcf) (%) 0 = 10

DR-12 Mid (low) 136.5 136.4 138.9 98 .2 92 . 1 61.4 3040 0.85 21,522
DR-12 Mid (low) 136.5 136.4 138.9 98 .2 92 .1 100 2958 0.86 21,428
DR-12 Mid (high) 137.6 138.4 138.9 99 .6 98.4 62.5 3828 0.82 25,295
DR-12 Mid (high) 137 .6 138.4 138.9 99 .6 98 .4 100 3758 0.80 23,437
DR-12 NJ (low) 121 .1 120.5 131 .2 91 .8 69 .3 59 .6 4307 0.72 22,865
DR-12 NJ (low) 121 .1 120.5 131 .2 91 .8 69 .3 100 1940 0.90 15,584
DR-12 NJ (high) 131.2 127.7 131.2 97.3 90 .5 53 .1 5134 0.74 27,890
DR-12 NJ (high) 131 .2 127.7 131 .2 97.3 90 .5 100 2706 0.92 22,508
DR-13 Mid (low) 138.3 138.2 141.0 98.0 91.3 63.8 4212 0 .76 24,238
DR-13 Mid (low) 138.3 138.2 141.0 98 .0 91.3 100 3606 0.80 22,752
DR-13 Mid (high) 141.0 139.5 141 .0 98 .9 95 .2 59.8 8312 0.58 31,967
DR-13 Mid (high) 141.0 139.5 141.0 98 .9 95.2 100 5918 0 .62 24,390
DR-13 NJ (low) 124.1 125.9 135.2 93.1 78 .3 63 .8 3470 0 .81 22,407
DR-13 NJ (low) 124.1 125.9 135 .2 93 . 1 78 .3 100 2824 0.82 18,658
DR-13 NJ (high) 135.2 134. 1 135.2 99 .2 97 .5 58.8 3164 0.90 25,428
DR-13 NJ (high) 135.2 134. 1 135 .2 99.2 97.5 100 2997 0.86 21,963
DR-14 Mid (low) 132.5 131 . 1 135.4 96 .8 85 .9 58 .6 4443 0.60 17,688
DR-14 Mid (low) 132.5 131 . 1 135.4 96.8 85.9 100 3401 0.68 16,278
DR-14 Mid (high) 134.5 134.4 135.4 99.3 96 .8 67.2 5468 0.64 23,869
DR-14 Mid (high) 134.5 134.4 135 .4 99 .3 96 .8 100 5793 0.67 27,096
DR-14 NJ (low) 122.8 121 .1 125.8 96.2 84 .7 58 .2 6618 0 .56 24,309
DR-14 NJ (low) 122 .8 121 .1 125 .8 96 .2 84.7 100 4504 0 .72 23,369 w
0)

DR-14 NJ (high) 125.8 124. 1 125 .8 98 .6 94 .7 55 .1 7639 0 .53 25,884

DR-14 NJ (high) 125.8 125 .0 125 .8 99 .3 97 .4 100 3569 0 .74 19,840
-------------------
DR-15 Mid (low) 134.4 131 .5 136.9 9 6. 0 83 . 1 51 .4 5058 0 .55 17,946

DR -15 Mid (low) 134.4 131 .5 136.9 96 .0 83 .1 100 2645 0 .69 12,955

DR - 15 Mid (high) 136.9 135 .4 136 .9 98 .9 95.2 59.7 4498 0.65 20,090

DR-15 Mid (high) 136.9 135 .4 136.9 98 .7 94 .5 100 2702 0.75 15, 194

DR-15 NJ (low) 109.5 110.4 115.8 95 .3 76 .1 58 .9 4554 0 .57 16,920

DR-15 NJ (low) 109.5 110.4 115.8 95 .3 76 .1 100 2358 0 .76 13,569

DR - 15 NJ (high) 114.3 115.8 115 .8 100 100 63 .6 3012 0 .72 15,807

DR-15 NJ (high) 114.3 115.8 115.8 100 100 100 1338 0 .96 12,203

CE = Compactive effort
MADD = Maximum Attainable Dry Density
DR = Relative Density
*E 9 = k 1 ok 2

w
-....,J
 38

Limits of Rada and

Witczak ' s Data
1000

...-...
.....
rn
0..
I
'-"
.....
~ 100
'
I ~ o©
O ~ 00
~ % 0

(f)~~O~o
0@ 0

0
10 0
0
I

I
1
0.0 0.2 0.4 0.8 1 .0 1 .2

Fig. 8: Relationship Between Experimentally

Derived Factors (k 1 and k 2 ) for the
Theta Model.
I
I 39

I
10 5
I 0 DR12
I • DR13
V DR14
I ... DR15
..........
.....
I Cll
p.
.__...
Cll

I ;::J
......
;::J
'C
0 104
::;.g
I ~

i:1
Q)
.....
......
I .....
f/l
Q)
~

I
I
I 10 3 1..-~~~~~~--'~~~~~~~_.._~~~~~~~~
0 .1 10 100
I Bulk Stress, 8 (psi)

Fig . 9. Effect of Agregate Source on Resilient Modulus .

I
I
I
I
 40

C"J~ 25
0

><
·~v.i
o.. 20

V
15
0
•
• DR14M
10 a DR13NJ
T DR12M
6 DR13M
• DR14NJ
5
30 40 50 60 70 80 90 100 11 0 120
Degree of Saturation (%)

Fig. 10. Effect of Degree of Saturation and

Aggregate Source on Resilient Modulus.
 I
I 41

was 17. 7 and 19.5, respectively. However, the effect on Eg of providing a drained
I base can also be seen from Fig . 11 by looking at the dashed lines. For both the

I low and high compactive effort cases, there is a benefit from changing from a

I dense graded material which will remain saturated for extended periods of time to

an open graded material which will remain in a drained state most of the time.
11 Additionally, the effect of compactive effort is apparent. For both the dense

I and open graded materials, an increase in compactive effort from close to 100% T-

99 density to 100% T-180 density leads to an increase in Eg.

I Statistical Analysis

A statistical analysis was conducted to determine the effect of the four

independent variables on Eg. Paired-t tests were performed to see if there was a
I significant difference between the means of all Eg data of 1) a low degree vs a

high degree of saturation, 2) a low compactive effort vs a high one, and 3) a dense

gradation vs an open gradation. Additionally, a Tukey HSD analysis was

performed to determine if aggregate source made a significant difference in Eg

I results and, if so, which source(s) were significantly different. The results are

shown in Table 14. As can be seen, degree of saturation, compactive effort, and
I aggregate source were significant to differences in Eg at the 0.05 level, and the

interaction of gradation and saturation was significant at the 0.088 level.

I Gradation by itself was not statistically significant. Reduction in Eg came from

increasing the saturation from 60 to 100%, reducing the level of compaction from

about 100% T-180 to 100% T-99 density, and having a saturated, dense graded

I
I
 42
30

-
a)
0
.....
25

><
·~ Ill
c..
'-"
Ill

- ::l
::l
'tj
0
20
:::s
,..J
A

-
Cl)
·~
.~Ill o Mid - High CE
QJ
0:: • NJ - High CE
15
v Mid - Low CE
"' NJ - Low CE
8 - 10 psi

Low High

Degree of Saturation (%)

Fig. 11. Effect of Gradation, Degree of Saturation,

and Compactive Effort on Resilient Modulus.
I 43

I Table 14. Significance of Variables to Resilient Modulus.

E0 at 8 = 10 psi (psi)
Condition Maximum Minimum Difference Significance
at 0.05 level

I All Mixtures :
Saturation, low vs
0 22,706 19,452 3254 yes
I high
Gradation , dense vs 21,634 20,562 1072 no

I open
Comp . effort, high vs 22,679 19,518 3161 yes
low
I Agg. Source,
* DR12vsDR15 22,541 15,585 6956 yes
I DR 13 vs DR 15 23,975 15,585 8390 yes
DR 14 vs DR 15 22,330 15,585 6745 yes
Gradation and 22,704 20,442 2262 yes at 0 .088
saturation: open level

I graded drained vs
dense graded
undra ined
I * All other combinations: not significantly different

I material as opposed to a drained, open-graded material. In regard to aggregate

I source, the DR-15 had a significantly lower modulus than the other three sources

(which were all about the same). However, this does not appear to be a function
I of particle shape or plasticity of fines, because the U and IP values of the DR15

were close to those of the DR14, and the plasticity Index all four sources was

about the same. A possible explanation is that the as-tested density of the DR- 15
I specimens was somewhat low compared to the target density.

I
I
 44
ESTIMATION OF RESILIENT MODULUS

Introduction

The layer coefficients a 2 or a3 for an unbound granular base or subbase

material are found from a relationship with resilient modulus, Eg. The resilient

modulus for a given material can be found by test using the "theta model", or by
I
use of an estimation regression equation. The development of the regression

equation is outlined below.

I
It has been shown that the resilient modulus of a granular material is a

function of stress state:

where :

(} = bulk stress = a 1 + a 2 + a3
a1 = major principal stress

= ad + ac in triaxial cell test

(ad = deviator stress; ac = confining pressure)

a1 = a2 + y1 z 1 + y2 z 2 in pavement, under centerline of load

(a2 is vertical stress induced from a single wheel load as computed

by elastic layer analysis; y 1 and v2 are unit weights of each

overlying material; z's are thicknesses of each layer overlying the

point of stress computation)

a2 = intermediate principal stress

a3 = minor principal stress

I
45

= ax +
I a2

a3 = av +
k 0 (v 1z 1

k0
+ V2Z2)

(v 1z 1 + v2 z2 )
I ax = horizontal stress in x-direction induced by wheel load, psi

av = horizontal stress in y-direction induced by wheel load, psi

k0 = coefficient of earth pressure at rest

I = 1 - sin <P for granular materials (<P = angle of internal friction)

I = regression coefficients as determined from laboratory cyclic triaxial

testing.
I Thus, in a pavement structure:

In order for Eg to be calculated by use of Eq. 6 (E9 = k1e"z), values for k 1 ,

k 2 , and 8 are necessary . The following discussion shows the methods by which

these three parameters were determined.

I k1 and k 2

The parameter k 1 represents the granular material condition and

I characteristics: gradation, particle shape and texture, degree of saturation, and

I relative density. A larger k 1 indicates a superior material/condition.

The constants k 1 and k 2 can be determined for a given material by cyclic

I triaxial testing (CTX) as shown in Fig. 7. Alternately, in order to be able to

I estimate k 1 (and hence E9 ) without performing a CTX test, regression equations

were developed by use of the statistical software package SYSTAT (36) from the
I
I
 46
test data produced in the present study. Numerous linear multiple regression

models were developed and analyzed to estimate k 1 from certain test data. Many

combinations of variables were analyzed. Data to develop the equations came

from Tables 7, 9, and 10. The equation of best fit had a low degree of accuracy

and is not shown here, but it was helpful in indicating trends and for estimation of

k 1 values to set up boundary conditions for other models in the study. A second

method of k 1 estimation will be presented later in this report.

k 2 has been shown (22) to correlate with k 1 as follows:

k 2 = 4.657 - log k 1 . . . . . . . . . . . . . . . . . (S)

1.807

Thus, k 1 can be estimated from physical properties of the aggregate (as shown

later), and k 2 can be estimated from a relationship with k 1 .

Bulk Stress (8)

Bulk stress in an unbound granular base layer is a function of applied load (P)

and contact pressure (p), stiffness (E 1 ) and thickness (D 1 ) of the overlying asphalt-

bound layer, stiffness (E 5 gl of the subgrade underlying the base layer, stiffness (E 9 )

and thickness (D 2 ) of the base layer, and unit weight of the overlying layers .

Because E9 is a function of 8, and 8 is a function of E9 , an iterative procedure was

necessary in order to reconcile the Eg and 8. The method of successive

approximations is shown below.

A load of 9000 lbs (one half of an 18k axle load) and a contact pressure of
I
100 psi (average U.S. truck tire pressure) was used.
I
I 47

1. An initial Eg was assumed.

I 2. An elastic layer analysis was performed with KENLAYER (37) a

I commercially available analysis program. Output included a 2 , ax, av

I and geostatic stresses resulting from the static overburden of the

pavement layers.

I 3. Bulk stress was calculated from 8 = a2 + ax + av + geostatic

I stresses.

4. E9 was calculated via Eq. 6.

I 5. If E9 (calculated) was not close to Eg (initial), a new E9 was

I computed.

The process was automatically repeated until satisfactory convergence was

I achieved. KENLAYER was used with the elastic non-linear option. Reconciling E9

I with 8 was not a separate process; rather, it was integral to determining the Eg

algorithm as explained below in the section "Resilient Modulus Algorithm".

I Corrected Bulk Stress

I Granular materials normally cannot sustain tension, discounting any

cementing action or residual compressive stress from compaction. So, in

I KENLAYER, if the computed 8 became negative, an arbitrary minimum modulus

I was assigned.

I Resilient Modulus Regression Equation

To circumvent the necessity of the above iterative procedure, Witczak and

I Smith (38) developed a predictive equation:

I
I
 I
48 I
I
A predictive equation of resilient modulus could be constructed from Eg test

data which reflected the effect of saturation, density, and bulk stress, 8.
I
Unfortunately, 8 is not normally known. Results of numerous runs of KENLAYER I
indicated that 8 could typically vary from 3 to 33 psi. This is too wide a variance
I
to merely assume a value for 8. Bulk stress could be calculated for a variety of

cross-sections, but the 8 values would be dependent on the particular value of k 1 I

that had been assumed. Additionally, it was anticipated that the typical design
I
situation would involve a trial-and-error solution for D 1 and D 2 , with 8 therefore

varying. Thus, an assumption (or determination by test) of a k 1 value seems to be

unavoidable.
I
In the formulation of Eq. 9, Witczak and Smith assumed a constant value for

k 2 . However, it has been shown in the present study and others (22) that k 2

varies significantly with k 1 , and k 1 is a function of several material characteristics,

including degree of saturation. When saturation increases above 90%, there is a

substantial drop in k 1 and an increase in k2 (22). This was verified in the present

study. To reflect the effect of the change in k 2 , a different regression equation I

was developed in the present study:

log Eg = 2.708 - 0.4581ogD 1 + 0.4261ogk 1 -0.1071og£, +0.2071ogE5 g

+ 0.0671ogD2 (10) I
or Eg = (510.505)(D,)-0.458(k,)0.426(£,)-o.101(Esg)o.201(D 2)o.o67

where k 1 can be obtained by test. If it is necessary to estimate k 1 , values may be

I
I
I 49

I approximated as shown in Table 15, which is based on the results of the present

study and a previous study (22). It must be emphasized that k 1 values vary

I considerably within an aggregate source class, and it is quite possible for a given

I gravel to have a k 1 value greater than a given crushed stone.

I Table 15. Estimated Values of k 1 .

I Material SCE
k,
MCE
(100%) (100%)
I Gravel 4300 5000

I Crushed stone

Note:
4800 6100

at~60% saturation; minus #200~ 10%

I SCE = standard compactive effort
MCE = modified compactive effort

I
I The modulus calculated by use of Eq. 10 will approximate the modulus from

Eq. 6 that has been reconciled with 8. The other terms in Eq. 10 are necessary to

I compute bulk stress conditions. As E1 and D 1 increase, less stress is transmitted

I to the base layer, hence E9 decreases. As E59 decreases, the base layer is less

confined under loading, hence E9 decreases. Note that Eq. 10 can only be used for
I a single granular layer sandwiched between an asphalt layer and the subgrade. For

I more complicated sections, use of programs such as KEN LA YER are recommended

to calculate E9 (base) and E9 (subbase).

I Eq. 10 was developed by calculation of granular E9 by use of Eq. 6

I
I
 I
50
I
(Eg = k1e"z). In the regression equation development, resilient modulus was varied
I
by use of 237 combinations of k 1 , k 2 , and 9 in the program KENLAYER. These

combinations represented three levels of the following variables: layer thicknesses

(0 1 and 0 2 ), subgrade modulus (E 5 g), asphalt layer modulus (E 1 ), and granular

material constants k 1 and k 2 :

Stress state: 01 = 2,8, 15 in

D2 = 4, 12, 18 in

E1 = 130,000; 500,000; 2,100,000 psi

E5 g = very soft, medium, stiff

k1 = 1800; 3000; 11,000 psi

k2 = 0. 776, 0.653, 0.341

The following paragraphs show how the above values were determined.

Granular Material k 1 and k 2 . k 1 = 1800; 3000; 11,000 psi, which represented

three levels:

1) 60% saturation; 95% SCE; dense gradation with 35% minus #16

sieve material; gravel representing U = 5; contaminated with plastic

fines

2) 60% saturation; 100% SCE; dense gradation with 35 % minus #16

sieve material; crushed stone representing U = 13; non-contaminated

3) 10% saturation; 100% MCE; open gradation with 5% minus #16

sieve material; crushed stone representing U = 15; non-contaminated

k2 = 0.776, 0.653, 0.341 from Eq. 8.

For each of the 237 combinations, KENLAYER calculated the bulk stress 9 in I
I
I the granular base and the deviator stress ad in the subgrade soil from an applied
51

I load. The different cross-sections represented by the model are shown in Fig. 12.

Asphalt Modulus (E 11. The asphalt layer was characterized as linearly elastic, while
I the granular base and subgrade materials were characterized as non-linearly elastic.

I For the asphalt layer, three conditions of stiffness were characterized using the

I relationship of temperature and resilient modulus for MHTD 1990 mix designs as

developed in Volume I of this report. The highest modulus was determined by

I choosing the stiffest 1990 mix at the coldest individual Missouri weather station's

I reported average monthly temperature (pavement temperature T P = 33.4°F). The

smallest asphalt modulus utilized the least stiff 1990 mix at the warmest individual
I weather station's average monthly temperature (T P = 92.3 °F). A middle value for

I modulus was found by using a 1990 mix of average stiffness with the overall

average monthly pavement temperature for all Missouri weather reporting stations
I +Tp = 64.2°F). This resulted in asphalt mixture moduli of 2,100,000; 500,000,

I and 130,000 psi. Pavement temperatures were calculated from air temperatures

by use of Witczak's equation as discussed in Volume I.

I Subgrade Modulus {E 5 gl· The resilient modulus of fine-grained soils can be

I computed by use of the program KENLAYER. The modulus is a function of soil

type, degree of saturation, compacted density, and state of stress within the
I pavement structure. In KENLAYER, the soil modulus was characterized as being

I very soft, medium, or stiff. The soils were described as shown in Fig. 13. As can

I be seen, the curves of each of these three soils has the same general shape and

I
I
o.,, De Surface Mi x ture (e.g. Type C or [ C)

a.• , D' 52
n 2 = l.O Bitur-iinous Bose

4 - 6 in. 0GB Construction Plo.tforr-i

Subgrode
a . Two Lo. yer Full Depth Aspho.l t Sect ion.

Aspha l t - Loyer or Lo.yers

DG Bose Co.rrying Load

Subgr ode

lo. Two Lo.yer Flexible Po.ver-ieni: 1./ith Dense Gro.ded

Unbound Bo.se Co.rr~ng Po.rt of Loo.d .

o. ,, o, Asp ho.l t - Bound Lo. yer or Layers

DG Bo.se

DGB Construc1:ion Plo.tf'orr-i/F'"ilter

Sub gro.de

c. Two L o.yer Flexible Po.veroent 1./ith Ope n Gro.cled

Unbound Bo.se Co.rrying Po.rt of Loo.d

Surf'o.ce Mixture

BituMinous Bose
a.•' D,
n 2
= 1.0

n , . a.,, D, DG Bose

DGB Construct;on Platf- · · ·- · ·nr

Subgrade

d . Three Loyer Flexible Povel"',ent.

Fig. 12. Definitions of Layer Variables.

 I
I 53

I equal line slopes (K 3 and K4 ). The parameters that distinguish one soil's

consistency from another are the maximum and minimum moduli (boundary

I conditions on a possible spectrum of stiffness) and K 1 , except for the very soft

I material which has a K4 of zero.

Input values to describe the three soils are shown in Table 16.
·I
Table 16. Typical Input Soil Constants for KENLAYER Analysis.
I
Soil K, K2 K3 K4 E5 g (max) E5 g (min)

I Consistency
very soft
(psi)
1000
(psi)
6.2 1110 0
(psi)
5662
(psi)
1000

I medium
stiff
7680
12,340
6.2
6.2
1110
1110
178
178
12,342
17,002
4716
7605
I From Fig. 13 it is seen that E5 g is also a function of stress state. Thus, there is an

I interaction between the stress transmitted to the base and to the subgrade, with

the modulus of both materials fluctuating with stress state. KENLAYER performs
I numerous iterations to reconcile the base, subase, and subgrade moduli and stress

states.

Table 16 is based on work by Thompson and Robnett (39). Note that with

the exception of very soft soils, the slopes of the lines in Fig . 13 are all the same.

The most significant variable is K 1 . K 1 is the input in KENLAYER which sets the

I curve position. KENLAYER computes deviator stress, ad, and E59 is thus

determined by moving along the curve in accordance with the point where ad

I intersects. K 1 can be determined by test (resilient modulus testing of subgrade

I
I
 54

-
t'l
0
~
18

><
•.-4 16

- rn
Pot

r.:I
tlG
ll
14
Kl - 12,340
.
I'll
12
::::l
...-4
::::l 10 = Stiff
'O
0
~ 7630
+> 8
i:::: - 178 7605
•.-4
...-4
•.-4

Cl>
Cl>

rn
6 Medium
4716
I
~
4
'O
Cl>
Kl - 1000
cd
s.. 2 K2 - 6.2 Very Soft
.c
U)
bO

::::l 0
1000 I
0 5 10 15 20 25 30 35
Repeated Deviator Stress, ad (psi)

Fig. 13. Relationship of Subgrade Resilient

Modulus and Deviator Stress for
Three Types of Subgrade.
I 55

I soil) or by approximation from the following equation (39):

I K1 = 3.63+0.1239(PcLAv)+0.4792(Pl)+0.0031(Ps,LT)-0.3361(GI) · .. (11)

where:
I = resilient modulus of soil at ad = 6.2 psi, ksi

I PcLAY = material finer than 0.002 mm, %

Pl = plasticity index

PSILT = material between 0.05 and 0.002 mm, %

I GI = group index, "new" method (infinite scale)

= + +
I (P 200 -35)(0.2 0.005 (LL-40))

accordance with AASHTO M 145

0.001 (P 200 -15)(Pl-10) in

P200 = material finer than #200 sieve, %

=
I LL liquid limit.

Thus, by performing Atterberg limits, sieve analyses, and hydrometer analyses, K 1

I can be estimated. Again, K2 , K3 , and K4 are as shown in Table 15 for any fine-

grained soil.

The K 1 equation is based on a dry density equal to 95% standard proctor (T-
I 99) maximum and at optimum moisture content (OMC). For an increase in density

I to 100%, an increase of about 1.4 ksi is suggested (39). For densities between

95 and 100%, Eq. 12 can be used:

Denscor = 1.4 *(PcoMP - 95 ys . . . . . . . . . . . . . . (12)

I where: Denscor = density correction to K1 , ksi
 56

PcoMP = in-service compaction, % .

K 1 is corrected by adding to it "Denscor". This is only done for increasing density

from 95% to 100% T-99 maximum density on the dry side of OMC. If the in- I
service moisture content will be greater than OMC, the use of the Denscor should
I
be omitted.

More significantly, the in-service moisture content must be estimated. An I

increase above OMC will reduce K 1 as follows:

Satcorr = 0.334 (SatoMc - Satsvc) . . . . . . . . . . . . ( 13)

I
where:

Satcorr = correction to K 1 for increase in moisture content above OMC,

I
ksi

Satsvc = in-service degree of saturation, at the in-service dry density, %

MC,s
Satsvc = _ _ _6_2_.4_ _ _ _1__
. . . . . . . . . . . (14)
(Pcom,/1 OOXMDD) sp.grsv.

SatoMc = degree of saturation at T-99 OMC, at 95% maximum standard

I
proctor dry density, % I
Sstsvc = --62
OMC
___4____1__ . . . . . . . . . . . . . (15) I
(0.95)(MDD) sp.grav.

maximum dry density, pcf

I
MOD=

where moisture is in "%". Note the "dry density" may be different in Satsvc

I
I
 57

I and SatoMc·

K 1 is corrected by summing K 1 and Satcorr. This is done for any in-service

moisture contents above OMC. However, on the dry side of OMC, application of

I the correction is limited to only, say, zero to 2% below OMC. If the in-service

degree of saturation increases above the compacted moisture content, Satcorr

I becomes negative, thus K 1 (corrected) is lower than K 1 .

Thus, to calculate Satcorr, the in-service moisture content must be

determined. Kersten (40) suggests that, in pavement structures, the moisture

content of clays generally exceeds their plastic limits. The moisture content of

silty soils are equal to or just under their plastic limits, and the moisture content of

sandy loams are less than their plastic limits. Plastic limits are generally higher
I than OMC values. A large proportion of all fine-grained soils exhibit in-service

I moisture contents in excess of their optimum moisture contents.

So, a lower bound on estimated in-service moisture contents would be

I between OMC and plastic limit (PL). An upper bound would be above the plastic

limit but less than 100% saturation.

Thus, by knowing LL, PL, and OMC, K 1 can be corrected (usually

downward):

K.i(corr, = Ki + (1,4{PcoMP-95Y5) + (0.334(Sat0 ,.,c-Saf5 vc)) . . . . (16)

I This K 1 (corr) would then be input into KENLAYER. From the pavement cross-

I section and material moduli, KENLAYER will compute the deviator stress in the

subgrade, and knowing K 1 (corr), it will compute the resilient modulus of the

I
 58
subgrade.

At some point during routine design, the E5 g must be determined, but the

use of KENLAYER may not be possible or appropriate. In this case E5 g can be

estimated by calculating K 1 (corrected) as shown above, followed by estimation of

ad. Several options for estimation of ad are open. In the performance of the 237

runs of KENLAYER which represented pavement cross-sections of a range from 2

in asphalt over a 4 in base to 15 in asphalt over an 18 in base, the following ad

values were noted: 2.1 psi minimum , 12.2 psi maximum, and 5.1 psi average.

The 5.1 psi average ad is less than 6.2 (the "knee" in Fig. 13), thus the E5 g is

situated on the steep-sloped portion of the curve (Fig. 13) and E5 g would be

greater than K 1 • On the other hand, by use of ad max (12.2 psi), E5 g would be less

than K 1 .

When determining resilient modulus on a given curve, the following I

equations are useful:

Esg = K,(corr) + KJ/<2-ad) when ad < ~ . . . . . . . . . ( 17)

Esg = K,(corr) - K4(ad-~) when ad> ~ . . . . . . . . . (18)

where:

K2 = ad at the knee of the curve; 6.2 psi is used in KENLAYER

K3 = upper slope of curves, 1 . 110 is used in KEN LA YER

K4 = lower slope of curve, 0.178 is used in KENLAYER.

Note that E59 (minimum) = K1 when K 1 = 1000 psi or less.

 59
The selection of 5.1 psi is the least conservative option, and Elliot (42)

suggests that it is not even appropriate. Conversely, use of ad max = 12.2 psi

may be unduly conservative. Thus, it is suggested here that a value of 6.2 psi be

I used and thus E5 g should be set at the K 1 (corrected) value . This would be

considered the "normal" condition. The 1986 AASHTO Guide recommends that a

I further correction should be used to account for seasonal moisture changes,

freezing, and length of season. This final weighted average Esg ("effective res ilient

modulus") should be the value used in Eq . 10 for estimation of the Eg of unbound

granular base material.

Eg was reconciled with 8 by the method of successive approximations, as

explained earlier. The entire set of results is shown in Appendix A.

Model Statistics. Once the 237 different values of Eg were calculated, a

regression model was developed by use of SYSTAT. The criteria for model

acceptance were as follows:

1. The highest adjusted squared multiple correlation (adj-R 2 ) for the

models that met all the below listed criteria. This statistic reflects the

overall goodness of fit of the equation. It describes how well an

equation will predict a population of data, not the sample data . Thus

it is usually a little lower than the R2 value, which is for the sample

data. The adjusted R2 for the model chosen was 0 .909.

2. The standard error of estimate was low compared to the mean

-
resilient modulus (Y): The SEE was 0.070 and the mean log Eg was
 60
3.591.

3. The analysis of variance F-statistic indicated that the relationship was

significant at the 0.01 level.

4. Residuals were normally distributed.

5. Residuals had constant variance.

6. All members of the population were described in the same model.

7. No serious problems of variable collinearity existed.

8. No single observation influenced the regression coefficients

excessively.

9. Each independent variable contributed significantly to the model.

The relationship of observed Eg (calculated from KENLAYER) and Eg estimated from

Eq. 10 is shown in Fig. 14.

Usage. Eq. 10 can be used by the designer to approximate Eg of granular material

which is functioning either as a base under an asphalt layer or as a subbase under

an asphalt surface layer and a bituminous base layer. If the latter is the case, E1

should represent a combination of the two asphalt bound layers (Eeql as follows:

E = D1.._IE.1a)o.333+010..IE.1b)o.333r •............ (19)

[
tlq D111 +D1b

where:

= combined modulus of both asphalt bound layers

= modulus of the asphalt surface layer, psi

o,a = thickness of the asphalt surface layer, in

I 61

I
I
I
I -
(')

~
0 40
><
..... 35
I'll
A
.......
O'l
;:j
30
......
;:j

I re
::il
0 25
~
~ 20
.....
(L)

......
.....
O'l

I 0::
CL)

reCL)
15

10
I >
J.,
CL)
I'll
..c 5
0

0
0 10 20 30 40 50

Estimated Resilient Modulus (psi x 10 3 )

Fig. 14. Observed vs. Estimated Resilient

Modulus of Unbound Granular Material.

I
 62

= modulus of the asphalt base layer, psi

D 1b = thickness of the asphalt base layer, in.

So, to use Eq. 10, the designer should: I

1. Assume a D 1 and D 2 for a particular trial.

2. Determine E1 by either test or by the approximation technique detailed in

Volume I knowing mixture design characteristics and approximate pavement

temperature.

3. Calculate E5 g from the procedure given above.

4. Determine k 1 of the granular material by test or by assumption (see Table

15).

Other trials of cross-section would only entail further assumptions of D 1 and D 2 .

Once Eg is determined, it can be converted to a layer coefficient a 2 or a3 as

shown in the next section.

Effects of Pavement Cross-section and Material Variables

As can be seen from Fig. 15, Eg increases with a decrease in asphalt layer

thickness or stiffness. This is because higher stress is transferred to the base layer

which increases Eg. Fig. 16 shows that as E5 g increases, bulk stress in the base

increases which increases Eg. As compactive effort increases, k 1 increases, and

thus Eg increases. And in Fig. 17 it is seen that an increase in base thickness has

only a small effect on base Eg (taken as a whole).

Summary

To obtain layer coefficients, the Eg of the granular material must be

I
 63

I
25 ---------------------------------------
Medium Subgrade
kl 3000 k2 = 0.653
D 2 = 4 in

-- 20
E 1 = 130,000 psi

-
t')
0
= 500,000 psi
><
......
P..
fl)
= 2,100,000 psi
........... 15
Cl.I
::l
......
::l
'"d
0
:::g 10
..,.)

~
(l)
......
......
......
Cl.I
(l)
0:: 5

I O · ..___,.......,__,__,i..-__,...1-__,--1.__,__,....__ _--.1.____.i.....__....1

0 2 4 6 8 10 12 14 16

D 1 (in)

I Fig . 15. Effect of Asphalt Layer Thickness

and Stiffness on Base Resilient Modulus.

I
 64

Stiff Subgrade

20 Medium Subgrade
Very Soft Subgrade
,-..._
t'j
0
.......
>< 15
.....
rt.l
0..
..__,,
rt.l
;:1
...... D 1 = 8 in
;:1 10
't:l
0
:::s
+J
i::
.....
Q.)
D2 12 in
......
..... 5
rt.l El 500,000 psi
Q.)
0:::

o---------------------
0 2 4 6 8 10 12
3
k 1 (psi X 10 )

Fig . 16 . Effect of Granular k 1 and Subgrade

Stiffness on Base Resilient Modulus.
 65

30

Medium Subgrade
El = 500,000 psi
25
,,..._
(")
kl = 3000 psi
0
..--t

~
..... 20
Ill D1 = 2 in
0.
.._
Ill

- ;j
15 D1 - 8 in

\
;j
'O
0
~
.+,J
I:: 10
• • •
~

-
.....Q)
.....
Ill
Q)
D1 - 15 in_/
0:: 5

0 _ _ _ _ _....__ _ _ _....__ _ _ _...__ _ _____.

0 5 10 15 20

Fig . 1 7 . Effect of Base Thickness and Asphalt

Layer Thickness on Base Resilient Modulus.
 66
determined. This can be done in two ways. The most accurate way is to run,

say, 14 Eg tests at different 8 values and define the Eg-8 relationship (Fig. 7).

Then, 8 must be determined from a program such as KENLAYER, knowing D 1 , E1 ,

D 2 , and E5 g (E 5 g can be determined by test or by estimation). Finally, knowing 8,

Eg is determined from the Eg-8 relationship. The second method is to estimate Eg

from Eq. 10 by knowing, as above, D 1 , E1 , D 2 , and Esg· Additionally, k 1 must be

estimated -- hence the lesser accuracy of this second method.

DETERMINATION OF LAYER COEFFICIENTS

Matching Layer Coefficient with Layer Position

When designing a pavement, values of a 1 , a 2 , a3 , D 1 , D 2 and D 3 must be

assigned. In this study, the development of layer coefficients was done with the

assignment of the definitions of layer variables, as shown in Fig. 12. Note that for

asphalt bound layers, a 1 and a 2 are determined from the procedures given in

Volume I of this study ( 12). The drainage coefficients m 2 and m 3 are determined

from the procedures given in the companion study to this report (2). The layer

coefficients a 2 and a 3 for unbound materials are determined in accordance with the

procedures developed below. I

AASHTO Nomographs

The nomographs available in the 1986 Guide (1) as shown in Fig. 18 and 19

relate layer coefficients a 2 and a 3 to base or subbase resilient modulus (Egb or Egsb)

as per the following (10):

a 2 = 0.2491ogEgb - 0.977 . . . . . . . . . . . . . . (20)

I 67

I
I
0 20

I 0 18
40
0 . 16

..,.
(J 14 - - - - ,oo- - - - --35- -- - - 2 .0 - - -- - 30
·;;;
70 80 a.

0 . 12 C:
60
25 §
50

I "' M
u
40 70 2 .5
C/)
::;
ClJ _ ____ _ a: - rn-- -
0 10 CD - m ----- "S
3 30 u ;;,
a:
X
.!!'
u
0
I- 2
60
0 .08 3.5 C/)

u "'X
20
Cl)
t- 15
(f)
0 06 ------ -----50----- - --
4.0

0 .04

0 02

I 0

I 1) Scale d erived by avera gi ng co rr elatio n s obt ai n ed from Ill inois .

12) Sca le derived by averagin g c o rrela tion s obt ain ed from California. New Mexico and Wyoming .

I 131
ljJ
Scale
Scale
derived
derived
by
on
aver aging co rr elat ions 01:ltained from Texcs
N CHRP pr o 1ec1 (3) .

Fig. 18. AASHTO Unbound Granular Ease

Layer Coefficient (a 2 ) Nornagraph.
 68

0 .20

~
0 . 14 - - - - 100 - - - - - 90 - ::- - - - - - 2
M ·;;;
ro 70 ~ 20 C.
80
50 §
0 . 12 C
40 70 3
Cl>
a: :,
·"' CD
u 60
ro
~ ___ _
t- 15 "'
:,
0 10 _ a; _ "' _ _ _ _ _14
__ ro :,
-0
~ 13 0

50 t- 12 ~
~
0 .08 :,
4 11
u 10
:, 10
(/)
40
0 06 -
30
5
25 5

( 11 Sca le d er ive d from correlati ons from Il li noi s.

(21 Scale derived from correlati on s obtained from Th e A sph alt Instit u te. Ca lifo rnia . New
Mexico and Wyom ing .
(31 Sca le d erived from correlat ions obtai ned from Te xas.
(41 Sca le derived on NC HRP proj ec t (3).

Fig. 19. AASHTO Unbound Granular Subbase

Layer Coefficient (a 3 ) Nomagraph.
 69

I 83 = 0.2271ogEgsb - 0.839 · · · . . . . . . . . . . . (21)

Once the Eg value for a given situation is known, either by test or by use of Eq .

10, a 2 or a 3 can be easily calculated from Eqs. 20 or 21, or the AASHTO

nomographs.

I Eq. 10 was developed with degrees of saturation of ~ 60%. The effects of

a variation in degree of saturation are addressed through the use of drainage (m)

coefficients. The determination of m-coefficients is the subject of the companion

project to the present study (2).

Equivalent Stiffness

Introduction. A second method of calculating layer coefficients was by use of

Odemark's transformation. As was done in Volume I of the study with the asphalt

mixtures ( 12), stiffnesses (in the form of resilient modulus) of MHTD aggregate

types were related to the modulus of the Road Test base material at various stress

states. Layer coefficients were calculated from Eq. 2:

I a (MHTD) = a (AASHO) [ Eg(MHTD) ]

1/3
· · · · · · · · · · (2 )
n n Eg(AASHO)

Three variables are required to solve Eq. 2: a 2 or a 3 (AASHO), Eg (MHTD),

and Eg (AASHO). These were determined as follows: 1) The layer coefficient a 2

for unbound crushed stone base at the Road Test is reported in the literature as

0.14. 2) The value for Eg (MHTD) will vary with several factors. Eg can be

determined by test or from Eq. 10. Knowledge of the stress state (or pavement

I
 70
geometry/load condition) is necessary to determine Eg. For the present analysis,

the stress state for the average Road Test section was determined and used in Eq.

2 for both MHTD and AASHO Eg values. 3) The Eg (AASHO) was determined as

follows.

Road Test Modulus. In the AASHTO Guide (1 ), Fig. 2.6 gives a relationship

between a 2 and resilient modulus. It states that the basis of the nomograph is a 2

= 0.14 at Eg = 30,000 psi. The question is, where did the 30,000 psi value

come from? First, a search of the literature was made. In the Guide, an equation

was given to be used in lieu of the nomograph which has been presented earlier in

this study. The reference given is Rada and Witczak ( 10). However, Rada and

Witczak state that the equation was based on the nomograph. Thus, it seems that

Rada and Witczak either found the source of the nomograph (but did not elaborate

on it) or they just calculated the equation from the nomograph. Apparently they

did not originate the nomograph. In the Guide (1 ), Appendix GG is referenced as a

source for an explanation of the origin of the layer coefficients. A reading of

Appendix GG reveals no such explanation. However, it is stated that Road Test

moduli varied from 15,000 to 30,000 psi. It does not state the origin of these

data. The nomograph is shown with Eg = 30,000 psi corresponding to a 2 =

0.14. The reference is given as NCHRP 128 (6). NCHRP 128 states that elastic

values for all the Road Test materials were assigned on the basis of four

references: Seed et al. (43), Shook and Fang (44), Coffman fil fil. (45), and Skok

and Finn (46). State DOT studies of layer coefficient derivation are described in
 71

I Appendix C of NCHRP 128. Very few are actually referenced, indicating that they

were unpublished internal reports. Apparently none involved resilient modulus

testing. Of the four references mentioned above, Seed et fil. dealt with soil, Shook

and Fang reported index tests and so forth, but no resilient modulus testing, and

Coffman et fil. tested static triaxial properties of Road Test granular materials and
estimations of dynamic properties. No estimations of dynamic modulus exceeded

20,000 psi, and for Road Test vehicle speeds, from the authors' calculations it

appears that the base and subbase E9 would have been considerably lower than

20,000 psi. Skok and Finn summarized other studies. In their paper, it appears

that a value of 15,000 psi was assigned to the Road Test base based on plate load

tests. Skok and Finn emphasized that this was very approximate. So , references

tied to NCHRP 128 do not shed light on the matter. The question remained -

where did the 30,000 psi value come from? Several sources or methods of

calculation were explored as follows.

NCHRP 291. Pursuing the question of the base modulus further, several

references were found to aid in calculation of Road Test E9 • In the AASHTO

Guide, the value for the gravel subbase was given as 15,000 psi. It was stated

that E9 values came from NCHRP 291 (21 ). NCHRP 291 stated that the Road Test

base and subbase were tested by the Asphalt Institute, but no reference is given.

Apparently, this is another internal report. The testing was done under conditions

of low moisture, medium moisture, and high moisture. Although test data were

not given, Table 17 is a summary of the seasonal moduli equations that were
 72
presented. For some unstated reason, September was omitted. Neither seasonal

moduli nor an overall single modulus were given. Thus, to compute the Road Test

moduli, in the present study elastic layer analysis was used to calculate the

seasonal moduli from the given equations. By use of the program ELSYM 5 (47),

numerous iterations were run for each season to reconcile the stress state (0) and

modulus values. ELSYM 5 allowed the use of the subgrade modulus model shown

in Table 17. To perform the analysis, an average pavement cross-section and a

representative load application were used. This is shown in Fig. 20.

Table 17. Elastic Moduli/Equations for AASHO Road Test Materials.

Seasonal Moduli/Equations (psi)

Material
Dec-Feb Mar-Apr May-Aug Sept-Nov
Asphalt 1. 7 X 106 0.71 X 10 6 0.23x10 6 0.45 X 106
Concrete, E1
Base, E0 b 50,000 3200 e0 ·6 3600 0°· 6 4000 0°·6
Subbase, 50,000 4600 e 0 ·6 5000 e 0 ·6 5400 e0 ·6
Easb
Subgrade, E50 50,000 8000 ad· 1 ·06 18,000 ad· 1 ·06 27,000 ad· 1 ·06

Note: ad = deviator stress, psi

From the Road Test reports (3), the average thickness for asphalt surface plus

binder, crushed stone base, and gravel subbase were calculated from the main

factorial study loop sections . The load application details ( 18 k SAL, 2 tires, tire

spacing, tire pressure) were taken from NCHRP 128 and Witczak (48). Poisson's
 73

4500 Lbf 4500 Lbf

r- p - 70 ps i

4.00 in U 1 = 0.40

tr 2 = 5 .39
µ...
1
in ,i = 0 ,....)_!
~,c
2

I
I

D = 8 .75 in
}_J = 1].]5
.?
3

,U
4
= 0.45

Fig. 20. Average AASHO Road Test Cross Section.

 74

ratios for each material were taken from Appendix DD of the AASHTO Guide.

From all this, the seasonal moduli and bulk stress values were calculated, and are

shown in Table 18. Winter moduli were assumed to be 50,000 psi for the base,

subbase, and subgrade in NCHRP 291. Assuming September should be included

with October and November, the average moduli for each material (weighted for

the number of months per season) were calculated and are shown in Table 18. As

can be seen, even with the inclusion of September into the highest modulus

season and assuming a frozen base in winter, the base modulus is still only 25 to

26,000 psi, not 30,000. In this analysis, the subbase seems high at 24,000, and

the subgrade seems very high at 17,000. Most references place the AASHO Road

Test soil at 3000 to 5000 psi. Use of KEN LA YER puts it at 5 543 psi. So,

inclusion of frozen conditions seems questionable. Thus, the average moduli (and

stress states) were re-calculated, omitting the frozen months. These somewhat

lower values are shown in the right-hand column in Table 18. The subgrade (5800

psi) seems much more reasonable. The subbase (15,000 psi) now lines up with

that which was stated in the AASTHO Guide. However, the base is 18,000 psi,

only 60% of the reported 30,000.

 75

I Table 18. Seasonal Moduli/Bulk Stress of AASHO Road Test Materials.

Seasonal Moduli/Bulk Stress (psi)

Material
Dec-Feb Mar-Apr May-Aug Sept-Nov Weighted Weighted
Average Average*

Asphalt 1,700,000 710,000 230,000 450,000 732,000 732,000

Concrete
Base 50,000 12,000 19,000 20,000 26,000 18,000
Subbase 50,000 11,000 16,000 17,000 24,000 15,000
Subgrade 50,000 3200 5500 8000 17,000 5800
Base 0 -- 9.2 16.3 14.2 -- 14.0
Subbase -- 4.3 6.8 6.6 -- 6.2
0
ad -- 2.37 3.06 3.15 -- 2.94

* Excluding winter values.

The results of Road Test trench studies (50) indicated that the average

moisture content of the crushed stone base was 4.3% (0.1 % above T-99

optimum). NCHRP 291 considered optimum moisture content as "moist" (as

opposed to "dry" or "wet") and from Table 17, "moist" test results indicate k 1 =
3600 psi. At 0 = 14 psi (see Table 17), this corresponds to an Eg = 18,000 psi.

The average moisture content of the gravel subbase was 5 .65%, which was

considerably higher (for a granular material) than the optimum of 3.8%. Thus this

could be considered either "moist" or "wet" with corresponding k 1 values of 4600

or 5000 psi. The corresponding Eg values at 0 = 6.2 are 14,000 and 15,000 psi,

respectively.

I
 76

Traylor. Traylor (49) tested the Road Test base material and reports moduli

relationships as shown in Table 19. Traylor measured the base and subbase

moisture contents as 4.3 and 5. 7%, respectively, which compares favorably with

Road Test trench studies. Thus, at these average conditions, from Table 19 it
'
appears that the average base modulus would be about 27,000 psi and the

subbase would be about 11,000 psi. Traylor also tested the subgrade for moisture

and Esg· The average moisture content was 15.8% and the average Esg was 3472

psi. The AASHO Road Test subgrade T-99 optimum moisture content was 13.3%.

Road Test trench studies showed the average subgrade moisture in the top 4 in to

be 16.6%.

Table 19. AASHO Road Test Moduli from Traylor.

Moisture contents, %
4.0 5.5 7.0 Road Test
Base
Equation 1 o,360 e0 ·35 11,795 e0 ·34 2s50 e 0 ·62 *10,647 e0 ·35
Egb' psi 26,815
Subbase
Equation 6840 e 0 ·32 6270 e0 ·30 4075 e 0 ·40 * *6270 e 0 ·32
E0 ~h· psi 10,839
* interpolated k 1 and k 2 at moisture = 4.3%; 0 = 14.0 psi
* * at moisture = 5.5%; 0 = 6.2 psi

NCHRP 128. Fig. 18 of NCHRP 128 indicates that, from an elastic layer analysis

for E1 = 450,000 (the given value) psi and 0 1 = 4 in., E9 b equals 30,000 psi.
 77

The analysis is based on equivalent sections which render equal vertical subgrade

compressive strain.

KENLAYER. In the present study, KENLAYER was used to calculate Egb and Egsb·

Input included E1 = 656,800 psi (best estimation of Road Test average asphalt

layer, in accordance with Vol. I of this report), k 1 = 10,647 psi and k 2 = 0.35 for

the base (as per Traylor), k 1 = 6270 psi and k 2 = 0.32 for the subbase, and K 1 of

the subgrade = 5619 psi. K 1 (corr) was calculated from Eqs. 16 with the following

input from Road Test data:

Table 20. Road Test Data Input to KENLAYER.

Parameter Value
minus #200, % 81
Liquid limit 29
Pl 13
GI (new) 8.6
amount clay, % 15.3
amount silt, % 27
average compaction, % 97.7
avg. max. dry density, pcf 116.4
97. 7% max. dry density, pcf 113.7
95% max dry density, pcf 110.6
optimum moisture content, % 15. 1
Specific gravity 2.71
average in-service moisture, % 16

calculated K 1 , psi 8759

I K 1 (corrected), psi 5619

 78
From KENLAYER, Egb = 24,180, Egsb = 11,130, and Esg = 5543 psi.

Thus it is not clear as to the value of the resilient modulus of the Road Test

base material. As shown in the next section, several values were evaluated.

Determination of MHTD Layer Coefficients

In the Road Test report, the layer coefficient of the crushed stone base

material was reported as 0.14, regardless of layer thickness. This implies that only

material characteristics are important when determining a 2 values for other

situations. However, the 1986 Guide a 2 -nomograph is based on resilient modulus,

which is a function of both material characteristics (k 1 and k 2 ) and stress-state

(which is partially a function of layer thickness). To evaluate the question of

stress-state importance, a representative pavement section was analyzed using

elastic layer analysis with Eg varying from 5000 to 30,000 psi. In this way (J also

varied. Vertical compressive strain (Ev) on the subgrade was calculated. Results

of this calculation were used to calculate subgrade rutting fatigue life as follows

(48):

N,, = 0.1365X1 o-a(e-v)- 4 ·477 · . . . . . . . . . . . . . (22)

The results are shown in Table 21. As can be seen, Eg is significant to pavement

life for a variety of base thicknesses. Therefore, it appears that it is proper to

evaluate layer coefficients (a 2 ) in terms of bulk stress as well as material

characteristics, as reflected by the resilient modulus.

 79

Table 21. Subgrade Rutting Prediction.

D2
5.39 in 14.14 in
Egb (psi
Ev Ntr Ev Ntr
5,000 0.001162 18,802 0.0009997 36,884

I 10,000
30,000
0.001127
0.0009577
21,561
44,686
0.0007064
0.0003814
174,555
2,756,077
Note: D1 = 4 in, E1 = 450,000 psi

To examine the effect on pavement life from a decrease in base layer

coefficient in accordance with the AASHTO structural number equation, the

following analysis was performed.

The basic design equation in the 1986 AASHTO Guide was used to back-

calculate SN values:

log [ fl.PSI
log W1a = ZR So + 9.36 log (SN+1 )-0.20 + - - - --4 -·~-~-·;-4~--
2
l (23)
0.40 + - - - - ~
(SN+1)5.19
+ 2.32 log E50 - 8.07

where:

W 18 = number of 18k ESAL (500,000; 50,000; 5000)

S0 = overall standard deviation (0.45)

ZR = statistic related to reliability (0.000 was used which represents

50% reliability)
 80
SN = structural number

ll. PSI = loss of Present Serviceability Index (4.2 to 2.5 was used)

E9 = roadbed resilient modulus (3000 psi).

Then, knowing Road Test average layer thicknesses, the following was used to

back calculate a 2 values:

SN - a 1 D 1
a 2 = - - - - - - - . • • • . . . . . . . . . . . . . . (24)
D2

where:

a1 = 0.44 (Road Test asphalt layer coefficient)

D1 =4 in (average Road Test asphalt layer thickness)

D2 = 14.14 in (average Road Test total granular material thickness).

From the AASHTO a 2 nomograph knowing a 2 values, E9 b values were obtained.

The results of this analysis is shown in Table 22.

Table 22. Backcalculation of a 2 Values.

N,a SN a2* Egb **

(18k SAL) (psi)
500,000 3.6 0.140 30,000
50,000 2.5 0.052 13,568
5,000 1.7 -0.004 8065

* Backcalculated from Eq. 24

* * Backed out of nomagraph

As can be seen, when modulus is reduced nearly an order of magnitude, the

allowable traffic diminishes by about two orders of magnitude. Table 21 tends to

 81
support this. Layer coefficients were computed from Eq. 2 by use of a 2 = 0.14
and Eg(AASHO) = 18,000 and 30,000 psi. Eg(MHTD) can be obtained by test or

I estimated by Eq. 10.

A comparison is made in Table 23 between the AASHTO nomograph and

I the Odemark stiffness methods of layer coefficient calculation. Note that Table 23

I shows that layer coefficients calculated from the Odemark analysis do not drop off

as rapidly as the AASHTO nomograph-derived coefficients when modulus

decreases. According to Table 22, it appears that the nomograph coefficients are

more realistic than the Odemark coefficients. Although this brief analysis was for

I Table 23. Comparison of AASHTO Nomograph and Odemark Stiffness Methods

of Layer Coefficient Calculation.

Eg (psi) AASHTO Odemark a 2 AASHTO Odemark a3

a2 a3
30,000 18,000 30,000 18,000
5000 -0.056 0.077 0.091 0.001 0.061 0.072
10,000 0.019 0.097 0.115 0.069 0.076 0.090
15,000 0.063 0.111 0.132 0.109 0.088 0.104
20,000 0.094 0.122 0.145 0.137 0.096 0.114
30,000 0.138 0.140 0.166 0.177 0.110 0.130
40,000 0.169 0.154 0.183 0.206 0.121 0.144

only one type of distress (rutting) and one pavement section, it does seem to show

that the use of the AASHTO nomograph for prediction of a2 from resilient modulus

values is preferred to the Odemark analysis.

I
I
 82
Conclusions

It is recommended that layer coefficients a 2 and a3 can be obtained by 1)

solution for E9 via Eq. 10 or by use of KENLAYER and then 2) solution for a 2 and

a3 by AASHTO nomographs or Eqs. 20 or 21.

CEMENT-TREATED BASE

INTRODUCTION

The third part of this study was to determine layer coefficients for cement-

treated soil base. The analysis is based upon the static compressive chord

modulus test as recommended by the 1986 AASHTO Guide (1 ). A secondary goal

was to develop a regression equation to assist in the estimation of static modulus

for the soils analyzed in this report. Once the modulus is determined, the layer

coefficient (a 2 ) can be obtained directly from the base layer nomograph contained

in the 1986 AASHTO Guide.

As with portland cement concrete, the magnitude of compressive chord

modulus of soil-cement materials is related to unconfined compressive strength.

Thus, factors that affect compressive strength will likely affect compressive chord

modulus . These factors include gradation (material retained on #4 sieve, material

finer than the #270 sieve, and material retained between #270 sieve and 0.005

mm), dry density, plasticity index, and cement content.

MATERIAL TYPES AND SOURCES

Two soils were chosen by MHTD personnel to be stabilized with portland

I 83
cement. The Lintonia (DR-16) is a fine sand, whereas the Knox (DR-17) is a clayey
I silt. It was intended that the Knox was to be mixed with Missouri River sand (DR-

I 17s). All materials were sampled and delivered to the UMR laboratory by MHTD

personnel.
I The materials, sources, and identification codes are shown in Table 24.

I Table 24. Material Types and Sources.

Nomenclature Material Sources Location

DR-16 Lintonia Soil -- Malden

DR-17 Knox soil -- Callaway Co.

DR-17s Missouri River Sand Capitol City Sand Jefferson City

I DR-18 Portland cement, Type I Monarch Cement --
Note: All sources are located in Missouri

LABORATORY INVESTIGATION

Soil Classification

The three materials were classified by utilizing results of washed sieve

analyses and index testing. The results are shown in Table 25.

Experimental Material Proportions and Moisture Density Relationships

To reflect the significance of compressive strength on static modulus, each

soil-cement mixture was tested at three cement contents. Additionally, the DR-17

soil was mixed with the DR-17s sand. Records supplied by the MHTD indicated

that a reasonable spread of soil-sand proportions would include 23, 30, and 37%

soil content. Because essentially all the DR-17 soil passed the #200 sieve and

I
 84

Table 25. Soil Classification.

Sieve DR-16 DR-1 7 (soil) DR-17s (sand)

Size % passing % passing % passing
3/8 in. 100.0 100.0 100.0
#4 100.0 100.0 98.5
10 100.0 100.0 79.9
40 97.1 99.9 16.0
60 43.6 99.8 5.1
200 3.6 99.2 0.3
270 3.4 98.2 0.2
0.005 mm -- 34.0 --
0.002 mm 0.3 30.0 0
ASG 2.652 2.74 2.547
LL -- 42 --
Pl NP 18 NP
Classification A-3 A-6 A-3

virtually all of the DR-17 sand was retained on the #200 sieve, this meant that the

percent passing the #200 sieve for the three mixtures would be 23, 30 and 37%.

The gradations, Atterberg limits, and AASHTO soil classifications are shown in

Table 26. Also shown are the results of gradation curve characterization via
-
Hudson's A and the slopes-of-gradation curve calculations. These were used later
-
in the regression analysis portion of the study. Calculation of Hudson's A and the

slopes was discussed earlier in the unbound aggregate section. However, different

sieve sizes were used for the soil/sand mixtures to better reflect their gradation
I 85

Table 26. Soil/Sand Mixture Characteristics.

Sieve Size 23/77 30/70 37/63 DR 16

I 1 ~ in 100 100 100 100

3/4 100 100 100 100
I 3/8 100 100 100 100
#4 99 99 99 100
I 8 88 89 90 100
16 69 71 73 99
I 30 46 50 54 98
50 30 36 42 70
100 26 31 38 21
200 23 30 37 4
I -
A 6.81 7.06 7.33 7.92

M4-10 0 167.1 153.1 136.7

M,o-40 33.9 574.8 510.6 469.7

M40-200 1092.4 143.7 130.9 123 .8

LL 34 35 35 --
Pl 17 17 18 --
GI 1 1 2 0

I Soil Classification A-2-6 A-2-6 A-6 A-3

I curve shapes.

The PCA Soil-Cement design manual (51) (tempered with MHTD experience)

indicated that for preliminary purposes, the amount of cement for the DR-16 and

DR-17 combination materials should be 9 and 8%, by dry weight of soil or soil-

sand, respectively. Consequently, T-99 moisture-density relationships were

 86
140
140

-
(")
...,
135 Sample Type :
,: Cement
1,1:DD
DR-16
g,:
112 .9 pc!
-
(")
...,
.:::::_ 130
135

.:::::_ 130
......c OMC 10.B!i ......c
!:- !:- 1 25
...,
1 25 ...,
..d
..ct
.~120
-~ 120 Ill
Ill
ll::
ll::
..., ..., 11 5
....
1 15
c c Sample Type :
,: Cement
DR-17
e,:
::::, ::,110 ,: Send 77,:
=,... 11 0 =,...
... ... MOD 129 .7 pc!
Q OMC 6 .11 ,:
Q 105
105

100
100
4 6 8 10 12 14
6 8 10 12 14 16
Wat.er Content (%)
Water Content (%)

140 140

"' -.... 13 5
(") -...,
135

-
.:::::_130
..c
!:::-125
.:::::_ 130
....
..c
!:::-125
..., ...,
..ct ..d
.~120
ll::
Ill - ta0120
I ll
II::
..., 11 5 ..., 115
....
c c S ample Type : DR-17
;::::,110 Sample Type : DR-17 ::>110 ,: Cement 9,:
,: Sand 77,:
...=,... ,: Cement
,: Sand
g,:
10,:
I>'\
... WDD 122.3 pc!
o 105 MOD 126.7 pc! o 105 OMC 11.1,:
OMC 10 .5 :t

100 .....~~......~ ~ . . . . . i . ~ ~ ~......~~......~~.... 100

6 8 10 12 14 16 6 8 10 12 14 16
Water Content (%) Water Content (%)

Fig. 21. Soil Cement Material Moisture-Density Relationships

 87

determined for the DR-16, DR-17 (23/77), DR-17 (30/70), and DR-17 (37/63)
I materials. The results are shown in Fig. 21. As expected, for soil/sand mixtures

I (DR17), the more granular the material, the higher the dry density and the lower

the optimum moisture content.

In the final cement content determination, Table 2 of the PCA manual

I indicates that the DR-16 soil should contain 9% cement. However, in the PCA

manual, Fig. 36 indicates 10%. MHTD records indicated that 6.9% has been

specified. Consequently, to cover this range, three sets of specimens were made

with cement contents of 6, 8, and 10%. For the DR-17 (23/77), DR-17 (30/70),

and DR-17 (37/63) mixtures, the PCA Table 2 indicates cement requirements of 5,
I 5, and 6% respectively. PCA Fig. 36 indicates 6, 6, and 7%, while MHTD records

show 4 to 10% recommended or specified, with 7 .4% as the most recent and

most similar in density to materials used in this study. Thus, mixtures were

prepared with 5, 7, and 9% cement. In Table 27 is shown the mixture proportion

I schedule.

Table 27. Soil-Cement Mixtures.

I % Cement DR-16
(23/77)
DR-17
(30/70) (37 /63)
5 X X X
6 X

I 7
8 X
X X X

9 X X X
10 X
I
I
 88
Specimen Fabrication

For the DR-17 (23/77) and DR-17 (30/70) materials, the material for four

replicate specimens was prepared by adding the amount of water necessary to

bring the cement, soil (and sand) mixture to its optimum moisture content (as

determined by the T-99 analysis), and mixing in a Lancaster counterflow 3.0 cu ft

mixer. Subsequently, for the rest of the specimens, the material was dry mixed in

the mixer for better mixture control. Then for each lift, the amount of water

necessary to bring the mixture to its optimum content was added. A

representative sample was taken for moisture determination.

Specimen size and method of compaction were two factors that had to be

considered. Several mold diameters and height-to-diameter ratios are presented in

the literature (51-52). When measuring modulus, height-to-diameter ratios of 2.0

are usually used; this was the case in this study. Specimens were 6 in in

diameter. The most common method of compaction is an impact type similar to

the T-99 standard effort. This method was used in the present study, with

additional effort via vibration by air hammer and plate to bring each lift to its

required density. The specimens were compacted into steel split molds with a 5.5

lb hammer and a 12 in drop. Each layer received a varying number of blows which

resulted in a compaction effort similar to the T-99 effort. Specimens were

compacted at optimum moisture content to 100% maximum dry density as

determined by the T-99 analysis. After compaction, the specimens were placed in

a moist room at 73°F, then demolded. The DR-17 specimens were demolded after
I 89

two days, while the DR-16 specimens were demolded after three days because of
I their more fragile nature. After a total of seven days moist curing, the specimens

I were removed from the moist room and capped with a proprietary sulfur

compound. Two replicate specimens were made for modulus testing and one for
I compression testing, then one or both of the modulus specimens were also tested

I in compression.

Unconfined Compressive Strength

I Each strength specimen was tested for ultimate compressive strength in a

I 200,000 lb Tinius-Olsen compression machine in accordance with AASHTO T 22-

90 (53). The results of the replicate specimens were used for determination of the
I modulus test load for each soil-cement mixture. Also, the results were used in the

I modulus regression equation which was developed from the data.

Modulus of Elasticity

The chord modulus of elasticity was determined as per ASTM C469-87a

I (54). In brief, each specimen was capped with a proprietary sulfur compound. A

I compressometer yoke was mounted on the specimen which held an LVDT (linear

variable differential transducer) for measurement of vertical deformation. The gage

I length of the specimen was 8 in. A pressure transducer measured pressure which

was then converted to load and displayed on a digital display. The load and

deformation conditioned signals were transmitted to an IBM XT via a 12 bit

I analog/digital board and stored in an ASCII file. The data file was then loaded into

QUATTROPRO (55) where the load and deformation data was parsed and

I
I
 90
converted to stress and strain. The stress and strain data were then loaded into

the data-fitting program TABLECURVE (56). A best-fit line was fit through the

data. The chord modulus was calculated by determining the x-y coordinates of

two points on the curve: 1) the load at 0.000050 strain, and 2) the load and

strain at 40% ultimate compressive strength. Two load runs were made and thus

the modulus was determined twice for each specimen ; then the results were

averaged. The modulus was calculated as follows:

a 1 -a2
Ee = . . . . . . . . . . . . . . . . . . . . (25)
€1-€2

where:

Ee = chord modulus, psi

a1 = 40% of ultimate strength, psi

a 2 = stress at 0.000050 in/in strain

E1 = strain at a 1 , in/in

E2 = 0.000050 in/in.
The equipment is shown in Fig. 22.

RESULTS OF LABORATORY INVESTIGATION

Results of the compressive strength and modulus testing are shown in Table

28. As can be seen, both strength and modulus increased with increasing cement

content. Also, as sand content of the DR-17 specimens increased, strength and

modulus increased. This is shown in Figs. 23 through 25. The relationship

between strength and modulus is shown in Figs. 26 and 27. The best-fit equation
 91

I
I

I
I
I
Fig. 22. Static Chord Modulus Equipment.

I
I
 92

-
.....
I'll
0..
800
• DR17 23/77
0 DR17 30/70
.,_,,
..c:: 700 ...
DR17 37/63
DR16
~ D
bO
~
Q)
i...
~
Cl) 600
Q)

.....>
I'll
I'll
Q) 500
i...
0..
s
0
u 400
'O
Q)
~
.....
.....
~ 300
0
C.)
~
~

200 --~~~~---~~~~--~~~~---~~~---'
4 6 8 10 12
Cement Content (%)

Fig . 23. Effect of Cement Content on Unconfined

Compressive Strength .
I 93

I 12

I ....-,.
I[)
0 10
~

><
.....
C'/l

I -- 0.

C'/l
;:1 8
........
;:1
'O
0
~

I 'O
s...
0 6 0 DR17 23/77

I
.c:
C,)
• DR17 30/70
V DR17 37/63
.... DR16
I 4
4 6 8 10 12

I Cement Content (%)

Fig 24. Effect of Cement Content on
Chord Modulus.

I
 94

12

9% Cement

........_

-
lO
0
10
><
.....
C/l
0-t
'-" 7% Cement
C/l
~ 8
......
~
"O
0 5% Cement
::.::s
"O
~
0 6
...c:::
u

4 ____....__ ___._____________

50 60 70 80 90 100
Sand Content (%)

Fig. 25. Effect of Sand Content on

Chord Modulus.
I
I 95

I
I
I
I
I
.........
I[)
0
~ 10
><
.....
I 0..
[/J

..__,,

I
[/J

- ;=j
;=j
'O
8

0
~
'O
r,...,
0 6 0 DR17 77/23
I ...c::
u • DR17 70/30
'v DR17 63/37
T DR16
I 4
200 300 400 500 600 700 800
I Compressive Strength (psi)
Fig. 26. Relationship Between Compressive
Strength and Chord Modulus.

I
 96

-
U')
0
~

....><
1:/)
14
0..
'-"
1:/)

:::l
...... 12
:::l
"O
0
:::::s
10 0
"O
~
0
0
,.q
t) 0
8
Q)
0
....>
1:/)
0
1:/)
00
Q)
~
6 0
0..
s
0
0

t)
4 oo
....
+J
C) 00

cd
+J
U) 2
200300 400 500 600 700 800
Unconfined Compressive Strength (psi)
Fig. 27: Overall Relationship of Unconfined Compressive
Strength and Static Compressive Chord Modulus
for Soil - Cement Mixtures.
 97

I Table 28. Compressive Strength and Chord Modulus Results.

Mix Soil/Sand Cement Dry Comp. Chord

Proportions Content Density strength Modulus
(%) (%) (pcf) (psi) (psi)
6 112.3 300 435,700
DR-16 -- 8 112.4 345 503,000
10 112.4 640 1,001,600
5 -- 485 698,400
23/77 7 126.9 538 972,700
9 130.0 766 1,134,000
I 5 126.4 239 601,300
DR-17 30/70 7 127.0 554 882,100
9 127.0 628 1,003,100
5 122.2 297 632,400
37/63 7 123.0 364 679,000
9 122.6 554 724,000
I
I (adjusted R2 = 0. 704) of the line is:

I Ee= 132,772 + 1314.9 qu · · · · . , , · · · · ·,. (26)

This is similar to that which is reported by Felt and Abrams (52). For sand and
I soil/sand mixtures similar to the present study, they noted increasing modulus and

I increasing compressive strength as cement content increased, and usually as sand

content increased. Their sandy soil's seven day compressive strength at 6%

I cement was 500 psi and at 10% was 1000 psi. The soil/sand mixture with 19%

(-) #200 material seven day compressive strengths for 3, 6, and 10% cement were

100, 450 and 1000 psi, while their soil/sand mixture with 37% (-) #200 material
I
 98
seven day compressive strengths at 6 and 10% cement were 400 and 600 psi,

respectively. The compressive moduli for the 10% (-) #200 material mixtures at 3,

6, and 10% cement contents were 0.3, 1.0, and 1.5 x 10 6 psi, respectively.

ESTIMATION OF STATIC CHORD MODULUS

A multiple regression model was fit to the data . The criteria for model

acceptance was the same as presented earlier for Eg in Eq. 10. The R2 statistic =
0.601, the adjusted R2 = 0.562, and the SEE = 158,433. The best fit model is

as follows:

Ec=-2,575,557 + 84,084(C) + 22,349(yd) . · . . . . . . (27)

where:

Ee = chord modulus, psi

C = cement content, %

yd = dry density, pcf.

Various parameters were tried in the model, including Pl, activity of fines, percent
-
passing #200 sieve and #270 sieve, A, slopes-of-curve, percent retained on #4

sieve, amount 0.005 mm material, and percent finer than 0.002 mm. None of

them improved the model nor were any statistically significant to the model. A

plot of the relationship of the estimated and observed soil cement chord moduli is

shown in Fig. 28. In comparing this relationship to the one delineated by Eq. 26, it

is seen that modulus can be more accurately estimated by use of unconfined

compressive strength.
I
I 99

I
I
I
I
12

I cu 0
.....> LD..--...
I fl)
I'll
cu
J-,
0
~
10

0 0
0
0

p... ><
I s
0
.....C/l
u p... 8
.._ 0
CJ
.....
I
I'll
...., ;::1
....,ro ......;::1 0 0
lf.1 "O 0 0
0 6 0
Oo
I "O
cu
+'
ro
~
"O
0

s J-,
0 0

I .....
....,
I'll
t:.J
..cl
u
4

I 2
2 4 6 8 10 12
I Observed Static Compressive Chord Modulus (psi X 10 5 )

I Fig. 28 : Relationship of Estimated and Observed Chord

Modulus of Soil - Cement Mixtures.

I
I
I
I
 100

28

10 0
.26
1(J()(J

.24 9 .0

.22 - - - - - - - - - - -800- - - - - - - - - -8.0

-
I
.,<

>
0 .20
7.0
600 N
. 18
N
m
V,
'vi ::::,,,
. 16 c
------------- ------------- "'0
.~ 400 .c 6 .0
u
Cl
C V,
1<: CJ :::,
0 ~
:::,
u U'J "O
CJ 0
200 > 2
·.;;
0.12 u II)
5.0
CJ

(/)
~ ---- - ---
- - - - - - - - uo
0 . 10
::;
C
C
0
u
C
:::,

0
I 1l Scale derived by averaging correlations from Illino is . Louisiana and Texas .
121 Scale derived on NCHRP project (31.

Fig. 29. AASHTO Nomagraph for Determination

of a 2 for Cement Treated Ease.
I
I 101

DETERMINATION OF LAYER COEFFICIENTS

I Layer coefficients were determined by use of the AASHTO nomograph,

I which is shown in Fig. 29. The equation for the relationship of a2 and modulus

I was derived from the nomograph and is:

a2 = -2. 7170 + 0.49711 logEc , · · · . · , · · · · .. (28)

I
where :

I E = chord modulus, psi.

I Substitution of elastic moduli into Eq. 28 resulted in the layer coefficients that are

shown in Table 28.

I VERIFICATION

I COMPARISON TO PUBLISHED DATA

In Table 1 are listed unbound aggregate base and cement-treated base layer
I coefficients as reported by others. The layer coefficients shown in Table 29 seem

I to be reasonable in comparison to those in Table 1 .

DESIGN VERIFICATION
I In order to further check the reasonableness of the layer coefficients, 18

I design situations were examined to compare standard MHTD designs vs. AASHTO

designs using the layer coefficients developed in this study. Subgrade soil and
I traffic conditions were varied to give a wide range of pavement thickness. The

I MHTD Flexible Pavement Thickness Design Chart was used to determine total

I pavement thicknesses and recommended asphaltic layer thicknesses. Additionally,

the unbound crushed stone base thicknesses were converted to the required

I
I
 102

Table 29. Soil-Cement Layer Coefficients.

Mix Soil/sand Cement Content a2

Proportions (%) (%)

6 0.09
DR-16 -- 8 0.12
10 0.27
5 0.19
23/77 7 0.26
9 0 .29
5 0.16
DR-17 30/70 7 0.24
9 0.27
5 0.17
37/63 7 0.18
9 0.20

thicknesses of soil-cement base by use of a 1.5: 1 equivalency ratio as

recommended in the MHTD method. Traffic levels were chosen so that each of
I
four design curves (2 through 5) in the MHTD chart could be used.

To obtain equivalent AASHTO designs, the use of a traffic correlation

provided by Murray (57) was used to convert the MHTD 2 axle equivalents to

18kESAL as used in the AASHTO method . Also, the subgrade was characterized

to have a Group Index (old) of 6, 9, and 17. Eqs. 17 and 18 were used to

calculate E59 values. These values represent the soils predominant in Missouri.

Data from the MHTD Geology and Soils Manual (58) were analyzed and it was

determined that the average GI (old) of the soils reported was 10 and the standard
I
I 103

I deviation was 4, which indicates that about two thirds of the soil types have a GI

between 6 and 14. The asphaltic layer thickness was held constant for both
I design methods. In the AASHTO method, initial Present Serviceability Index (PSI)

I was assumed to be 4.2 and terminal PSI was chosen as 2.5 for a ~PSI = 1.7.

Reliability was taken as 50%, and standard deviation as 0.45. From the 1986
I AASHTO Guide design equation (Eq. 23) the resulting required structural numbers

I were calculated. To solve for D 2 in an asphalt-over-unbound base section:

I
I To solve for D3 in an asphalt-over-bituminous base-over-unbound subbase section:

I 03 = SN-(a1D1 +~D2 )
~

I The layer coefficient (a 1 ) for the asphaltic layers was taken as 0.42 and a 2 = 0.34
(as per Volume I of this report), and the layer coefficients a 2 and a 3 for the
I unbound base and subbase at various thicknesses were obtained from Eqs. 20 and

I 21. Eb values for the granular base and subbase were calculated using Eq. 10.

I Input included D 1 , D 2 , and Esg values as shown in Tables 30 and 31. An E1 of

457,351 psi was used, which represents the average resilient modulus of 1990

I MHTD mix designs (see Volume I). Values of k 1 for the base and subbase were

I taken from Table 15: 6100 psi for the base and 4300 psi for the subbase. The

soil-cement layer coefficient was determined from Table 29 assuming a DR- 17

I 37 / 63 blend with 7% cement (a 2 = 0.18). The results of the analysis are shown

I
I
 104

in Tables 30 and 31:

Granular a 2

In studying Table 30, the design base (0 2 ) thickness for the granular

material based on the AASHTO nomograph is similar to MHTD designs for the 6
I
and 17 GI soils for most traffic levels. For GI = 9, the nomograph a 2 - derived

designs are conservative. The Odemark a 2 -derived designs are conservative for

GI= 9, but are thinner than MHTD designs for Gl's of 6 and 17.

Granular a3

Table 31 indicates that the nomograph-based designs are close to MHTD

designs for soils of GI = 6, are conservative for GI = 9, and are thinner for GI =

17. The Odemark-based designs are conservative for GI = 9 and thinner than

MHTD designs for GI = 6 and 17. However, underdesign at GI = 17 may not be

of practical significance if a subgrade cap is used, as mentioned in the MHTD

design method for soils with a GI greater than 16.

Thus, two options are presented for possible use by MHTD. If an evaluation

by MHTD of its granular base design thicknesses reveals them to be either on-

target or inadequate, then the AASHTO nomograph-based layer coefficients are

recommended. If an evaluation indicates that MHTD base designs are excessive,

then the Odemark-based method of layer coefficient calculation is recommended. I

Soil Cement a 2

Table 30 indicates that for a 2 = 0. 18, AASHTO-derived cement stabilized

base thicknesses are comparable for GI = 9, would be the same for GI =6 (if
 - - - - - - ---------
Table 30 . Comp ariso n of AASHTO Design s to MHTD Des igns for a2.

Subgrade Traffic SN D, a2 D2 (CSB) D2 (soil-cem.)

(granular)

GI Esg 2AE 18kSAL MHTD AASHTO Norn. Ode MHTD AASHTO MHTD AASHTO
(psi) (thous.) (in) (in) (in) (in) (in)
Old New Norn. Ode
(in) (in)

5 9341 100 100 1.8 0.13 7.5 4.9 4 .1 5 .0

• 3 .0
6 3 3 0.11
200 175 2.0 3 3 0.11 0.13 7.5 6.6 5.5 5.o· 4.1
500 450 2.3 4 4 0.10 7.2 6.4 5.3 4.8
• 3.4
0.12
1000 875 2.6 4 4 0.10 0.12 9.3 9.2 7.6 6.2 5.1

9 9 5619 100 100 2.2 3 3 0.10 0.12 6.0 9.1 7.6 4.o· 5.2
200 175 2.4 3 0.10 0.12 7.4 10.9 9.1 4.9
• 6.3
3
500 450 12.3 10.1 5.7
• 6.2
2.8 4 4 0.09 0.11 8.6
1000 875 3.1 4 4 0.09 0.11 11 .1 15.4 12.6 7.4 7.9
17 30 4579 100 100 2.4 3 3 0.10 0.12 9.5 11.4 9.4 6.3 6.3
200 175 2.6 3 3 0.10 0.12 12.1 13.3 11.0 8.1 7.4
500 450 3.0 4 4 0.09 0.11 14.2 15.0 12.2 9.5 7.3
1000 875 3.3 4 4 0.09 0.11 17.7 18.2 14.8 11.8 9.0

• Disregarding minimum thickness requirements

Table 31 . Comparison of AASHTO Designs to MHTD Designs for a3.

Subgrade Traffic SN 0 1 and D2 a3 03

(asphalt-bound)

GI 2AE 18kSAL MHTO AASHTO Norn. Ode. MHTO AASHTO

(thous.) (in)

Old New o, 02 o, 02 Norn. Ode.

(in) (in) (in) (in) (in) (in)

6 5 500 450 2.3 1.25 3.75 1.25 3.75 0.11 0.15 4.3 4.4 3.3
6 5 1000 875 2.6 1.25 3.75 1.25 3.75 0.12 0.15 6.4 6.9 5.1

9 9 500 450 2.8 1.25 3.75 1.25 3.75 0.11 0.14 5.7 9.3 7.0
9 9 1000 875 3.1 1.25 3.75 1.25 3.75 0.11 0.14 8.2 11.9 8.9

17 30 500 450 3.0 1.25 5.75 1.25 5.75 0.09 0.11 8.3 6.1 4.5
17 30 1000 875 3.3 1.25 5.75 1.25 5.75 0.09 0.12 11.8 9.3 6.9
I
I 107

I minimum thickness of 6 in is used as per MHTD practice), and are somewhat

thinner for two traffic levels on GI = 17 soil. On balance, the layer coefficients
I seem reasonable. However, the wide range in coefficients reported in Table 29

I emphasize the importance of testing the actual materials and assigning the proper

coefficient, as opposed to assigning a single coefficient for all materials.

I SENSITIVITY ANALYSIS

I A sensitivity analysis was performed by looking at the effect of mixture

variables on required pavement thickness .

I UNBOUND GRANULAR MATERIAL

I As discussed previously, only compactive effort and degree of saturation

were significant in their effect on E9 • An analysis was performed to ascertain the

I effect on base and subbase layer thickness by varying compactive effort. The

effect of saturation is examined in the companion study (2). For the a 2 coefficient

analysis, thickness of the surface (asphalt) layer was held constant at 4 in.

Thickness of the base layer was varied at 6 in and 12 in. Three levels of

I compactive effort (CE) were looked at: most significant change in the UMR data

set, least significant change in the UMR data set, and average change for all

blends . For instance, looking at Table 13, the largest change in E9 as a result of a

change of CE occurred with the DR-14 Mid high CE material as CE went from high

I (E 9 = 27,096) to low (E 9 = 16,278 psi). The least change occurred where the

DR-14 NJ high CE material went from high (E 9 = 25,884) to low CE (E 9 =

I 24,309). The "average" condition was simply a comparison of the average of the

I
I
 108

E9 's for the four aggregate sources lowest CE to the average of the E9 's for the

four aggregate sources highest CE. In each case, CE was changed from high to

low with a resulting change in E9 • Then the resulting change in layer coefficient

(a 2 ) was calculated from the AASHTO nomograph. Finally, the required change in

thickness was computed as needed to maintain the initial structural number

rendered by the initial assumption of layer thicknesses. The results are shown in

Table 32. For example, what is the average change in required base layer

thickness for an initial thickness of 6 in when compactive effort is changed from

high to low? Looking at row 1 in Table 32, the E9 for the high CE (of the pair

which resulted in the most loss of modulus) is 27,096 psi. Moving to the low CE

results in a reduction to 16,278 psi. Using the AASHTO nomograph, the

corresponding loss in a 2 is 0.055. The structural number (SN) provided by the

high CE is SN = a 10 1 +a 2 D 2 or SN = (0.42)(4) + (0.127)(6) = 2.44. When the

layer coefficient is reduced from 0.127 to 0.072, the new required thickness 0 2 =

(SN - a 1 0 1 )/a 2 = 2.44-(4*0.42) = 10.6 in, or, an additional requirement of

0.072 I
4.6 in. This is quite significant in a practical sense. A similar analysis was

undertaken for the subbase layer coefficient a3 • 0 1 , a 1 , 0 2 , and a 2 were held

constant. 0 32 was calculated as 0 3 = (SN-a 1 0 1 - a 2 D 2 )/a 31 . The results are

shown in Table 33.

From Tables 32 and 33, it appears that, on the average, changes in unbound

base layer compactive effort are significant for both 6 in and 12 in layers. The

thicker the granular base or subbase layer, the more pronounced is the effect.
I
I 109

I Table 32. Base Thickness Sensitivity to Changes in Compactive Effort.

I D21 CASE MR1 MR2

CHANGE IN COMPACTIVE EFFORT*

Eri diff a2, a22 a 2 diff SN D22 D2

(in) (psi) (psi) psi) (in) diff
I (in)

6 Worst 27,096 16,278 10,818 0.127 0.072 0 .055 2.44 10.6 4.6
12 27,096 16,278 10,818 0.127 0.072 0.055 3.20 21 .2 9.2
6 Avg 22,629 19,518 3111 0.107 0.091 0.016 2.32 7.1 1.1
I 12 22,629 19,518 3111 0.107 0.091 0.016 2.96 14.1 2.1
6 Least 25,884 24,309 1575 0.122 0.115 0.007 2.41 6.4 0.4
I 12 25,884 24,309 1575 0.122 0.115 0.007 3.14 12.7 0 .7
Note: a 2 from AASHTO nomograph
0 1 = 4 in, a 1 = 0.42
E9 at 8 = 10 psi
* = Data includes both gradations, both degrees of saturation, all aggregate sources
I
I Thus, it appears that compactive effort is important to the resulting layer

coefficient. The effect of degree of saturation goes beyond just the effect on k 1

and k 2 of the granular material because the available moisture adversely affects the

I subgrade modulus which reduces the bulk stress of the granular base which

reduces the Eg in addition to the reduction brought about by the change in k 1 and
I k 2. This will be addressed by use of m-coefficients as discussed in Reference 2.

I SOIL CEMENT

As shown in Eq. 27, only cement content and sand content were significant
I in their effect on Ee. An analysis similar to that for the granular material was

I performed to ascertain the effect on base layer thickness by varying cement

content or sand content. Thickness of the surface (asphalt) layer was held
I
I
 110

Table 33. Subbase Thickness Sensitivity to Changes in Compactive Effort.

CHANGE IN DRY DENSITY *

D31 CASE MR1 MR2 Eri diff a31 8 32 a3 diff SN D32 D3
(in) (psi) (psi) psi) (in) diff
(in) I
6 Worst 17,030 10,160 6870 0.121 0.071 0.050 3.55 10.3 4.3
12 17,030 10,160 6870 0.121 0.071 0.050 4.28 20.6 8.6
6 Avg 14,455 12,090 2365 0.105 0.088 0.017 3.45 7.2 1.2
12 14,455 12,090 2365 0.105 0.088 0.017 4.08 14.4 2.4 I
6 Least 17,926 16,299 1627 0.126 0.117 0.009 3.58 6.5 0 .5

~
12 17,926 16,299 1627 0.126 0.117 0.009 4.34 13.0 1.0
Note: a3 from AASHTO nomograph
0 1 = 1.25 in, a 1 = 0.42; D 2 = 6.75 in, a 2 = 0.34
Eg at 9 = 5 psi
* = Data includes both gradations, both degrees of saturation, all aggregate sources
I
constant at 4 in. Thickness of the base layer was varied at 6 in and 12 in. Three

levels of cement content or sand content were studied: most significant change in

the UMR data set, least significant change in the UMR data set, and average

change for all mixes. For instance, looking at Table 28, the largest change in Ee as

a result of a change of cement content occurred with the DR-1 6 soil as cement

content went from 6% (Ee = 435,700) to 10% (Ee = 1,001,600 psi). The least

change occurred where the 37/63 DR-17 blend went from 5% (Ee = 632,000) to

9% (Ee = 724,000) cement content. The "average" condition was simply a

comparison of the average of the Ec's for the four soil blends' lowest cement

contents to the average of the E/s for the four soil blends' highest cement

contents. Cement or sand content was changed from high to low with a resulting
 111

I change in Ec. Then the resulting change in layer coefficient (a 2 ) was calculated

from the AASHTO nomograph. Finally, the required change in thickness was

computed as needed to maintain the initial structural number rendered by the initial

assumption of layer thicknesses. The results are shown in Table 34. For example,

what is the average change in required base layer thickness for an initial thickness

of 6 in when cement content is changed from high to low? Looking at row 1 in

I Table 34, the Ec for the high cement content (of the pair which resulted in the

most loss of modulus) is 1,001,000 psi. Moving to the low cement content

results in a reduction to 435,700 psi. Using the AASHTO nomograph, the

corresponding loss in a 2 is 0.18. The structural number (SN) provided by the fine

side mix is SN = a 1 0 1 + a 2 D 2 or SN = (0.42)(4) + (0.27)(6) = 3.30. When the

I layer coefficient is reduced from 0.27 to 0 .09, the new required thickness 0 2 =
(SN - a 1 0 1 )/a 2 = 3.30-1.6S = 18 in, or, an additional requirement of 12 in. This
0.09
is quite significant in a practical sense.

From Table 34, it appears that all changes in cement content and sand

I content are significant for both 6 in and 12 in layers. For thicker layers, the effect

becomes more important. Thus, it appears that optimizing sand and cement

contents is important to the resulting layer coefficient.

I SUMMARY AND CONCLUSIONS

The purpose of this report was to determine layer coefficients for Type 1

dense-graded unbound granular base, open-graded unbound granular base, and

soil-cement base. The methodology included the determination of the moduli of

the various materials and then conversion to layer coefficients. The study included

I
Table 34. Thickness Sensitivity to Changes in Cement Content and Sand Content.

I
CEMENT CONTENT CHANGE

D21 CASE E21 E22 E diff 8 21 8 22 82 diff SN D22 0 2 diff

(in) (psi) (psi) (psi) (in) (in)

12
Worst 1,001,600

1,001,600
435,700

435,700
595,900

595,900
0.27

0.27
0.09

0.09
0.18

0.18
3.30

4.92
18

36
12

24
I
I
6 Avg 965,675 591,950 373 ,725 0.26 0.15 0.11 3.24 10.4 4.4

12 965 ,675 591,950 373,725 0 .26 0.15 0.11 4 .80 20.8 8 .8

6 Least 724,000 632,400 91 ,600 0 .20 0 . 17 0.03 2.88 7.1 1. 1

12 724,000 632.400 91 ,600 0 .20

SAND CONTENT CHANGE

0.17 0 .03 4 .08 14.1 2 .1
I
6 Worst 1,134,000 724,000 410 ,000 0.29 0.20 0 .09 3.42 8.7 2 .7

12 1,134,000 724,000 410,000 0.29 0 .20 0 .09 5.16 17 .4 5.4

1•
6

12
Avg 935,033

935,033
678.467

678,467
256,566

256,566
0 .25

0.25
0 . 18

0. 18
0 .07

0 .07
3 .1 8

4 .68
8 .33

16.67
2.33

4 .67 I
6 Lea st 698,400 601,300 97,100 0 .19 0 . 16 0 .03 2 .82 7. 1 1.1

12 698 ,400 601 ,300

Note : For all sections, 0 1 = 4 in and a 1 = 0.42.

97,100 0 . 19 0. 16 0 .03 3 .96 14.25 2 .25
I
the followin g testin g and anal y sis. I
1. The materials under study included two sources of crushed stone, two

gravels, and two soils (to be mixed with cement). Additionally, a concrete

sand was supplied for combining with one of the soils. One soil was a fine

sand, while the other was a silty clay. All materials were selected, sampled,

and delivered to UMR by MHTD personnel.

2. For the granular base study, two gradations of granular material were

chosen: one followed the midpoint of the MHTD Type 1 gradation

acceptance band, and the other was the so-called New Jersey open-graded

gradation.

.,
 113

3. The aggregates were separated into the appropriate size fractions,

recombined, and tested for specific gravity, plasticity index, moisture-density

I relationships, and relative density. Standard and modified proctor-type tests

I (T-99, T-180) were performed for the dense-gradation, while vibratory table

densification was used for the open-graded material.

I 4. Nine different methods were evaluated to characterize the gradation curve

I shapes and positions. These methods were fineness modulus, coefficient of

uniformity, coefficient of skew, Hudson's A, surface fineness (SF), specific

surface factor (SSF), SF/SSF, slopes-of-gradation curve, and percent passing

or retained on individual sieves. None of these parameters proved superior

to others in the prediction of k 1 or Eg.

I 5. Particle shape/surface texture tests were performed on the four aggregates.

I The ( +) #4 sieve material was tested in accordance with ASTM

03398, while the (-) #8 to ( +) #100 fraction was tested using the NAA
I method. The measured angularities of the two stones were about the same,

and they were more angular than the two gravels, which also were about

I equal. The difference in angularity/texture was not great between the

crushed stones and the gravels.

6. Resilient modulus tests were run on all four aggregates using two

gradations, two compactive efforts, and two degrees of saturation, with

replications. Fourteen combinations of confining pressure and cyclic applied

deviator stress were used for each specimen. Effective confining pressures

I
 114

ranged from 2 to 20 psi and cyclic deviator stress ranged from 2 to 40 psi.

Thirty-two specimens were fabricated. The total number of tests run was

896. I
The results of the testing indicated that Eg increases with increasing

bulk stress, increasing compactive effort, and lower degree of saturation.

For the materials tested, the effects of particle angularity and gradation were

not significant.

7. A statistical analysis was performed to determine the significance of the

variables. Paired-t tests indicated that change in density or degree of

saturation gave significantly different results at the 0.05 level, but gradation

did not. Tukey HSD analysis indicated that one of the two gravels gave

significantly different results than the other gravel and the two crushed

stones. The particle shape analysis indicated that the shapes of the two

gravels were about the same, but they were both different from the two

crushed stones. All of this taken together indicated that particle shape was

not significant. Finally, in comparing saturated dense-graded material with

drained open-graded material, there was a significant (0.088 level) increase

in Eg with superior drainage.

8. A multiple regression model was developed for the parameter k 1 but the

relationship was relatively weak.

9. To assist the pavement designer who may not be able to obtain actual E9

data, a regression model was developed for E9 by analyzing 237 pavement

I
 115

sections which varied in asphaltic layer thickness (0 1 = 2,8, 15 in), asphaltic

layer modulus (E 1 = 130,000, 500,000, and 2,100,000 psi), unbound base

thickness (0 2 = 4, 12, 18 in), subgrade modulus (E 5 g = 1000 to 17,000),

I granular material constant k 1 ( 1800, 3000, 11,000 psi), and granular

material constant k 2 (0.341, 0.653, 0.776). The most accurate model was:

I Eg = 510.505 (D, )"0.458 (k, )0.426 (E, )"0.101 (Esgl0.207 (D 2 )o.067

This model had an adjusted R2 = 0.909.

The model indicates that E9 increases with increasing k 1 of the

granular base, increasing K 1 and Esg of the subgrade soil, increasing D 2 ,

I decreasing 0 1 , and decreasing E1 .

10. For comparison purposes, layer coefficients were computed in two different

manners. First, E9 values were substituted into the AASHTO nomographs

I for unbound granular base and sub base to obtain a 2 and a 3 directly.

Second, Odemark's equivalent stiffness was used for comparing MHTD

I aggregates to Road Test materials (and stress states). An analysis indicated

I that it is preferable to use the AASHTO nomographs.

I 11. Because layer coefficients are directly affected by resilient modulus, the

practical impact of the trends is that higher layer coefficients can be

obtained by greater compactive effort and lower degree of saturation (better

drainage).

12. The Knox soil (A-6) was mixed with concrete sand in three proportions of

soil to sand: 23/77, 30/70 and 37/63. Specimens were made with 5, 7,

I
 116

and 9% cement contents by weight. The Lintonia (A-3) soil was mixed with

6, 8, and 10% cement contents. Specimens were tested for compressive

strength and static chord modulus. In general, strength and modulus

increased with increasing cement contents and increasing dry densities.

13. Two regression equations were developed for estimation of static chord

modulus:

EC = 2,575,557 + 84,084 (C) + 22,349 (yd) adj R2 = 0.562

Ec = 132,772 + 1314.9 qu adj R2 = 0. 704

where C is cement content, yd is dry unit weight and qu is compressive

strength.

14. Layer coefficients were obtained for the soil-cement mixtures by use of the

AASHTO nomograph. Values ranged from 0.09 to 0.27. Most of these fit

into the range of values reported elsewhere.

1 5. A verification analysis was performed. Twelve hypothetical pavements were

designed by use of both the former MHTD method and the AASHTO method

using the layer coefficients developed in this study for unbound base and

soil-cement. This analysis tended to verify the choice of using the AASHTO

nomographs for calculation of layer coefficients, although use of the

Odemark method may be preferrable under certain conditions.

16. A sensitivity analysis was performed. The results indicated that a higher

compactive effort could result in a reduction of up to 2 in in base thickness

for a 12 in thick base.

 117

An increase in degree of saturation acts to lower k 1 and raise k 2 of

the granular material, and to lower subgrade support, all of which act to lower the

Eg of the granular material. This phenomenom is addressed by use of drainage

coefficients, as developed in the companion study of this report (2). Thus, the

layer coefficients developed in this report are representative of average degrees of

I saturation of about 60%. For significantly greater or smaller magnitudes of

saturation, m-coefficients should be used in the equation:

SN = a1 0 1 + m 2 a2 D 2 + m 3 a3 D 3

RECOMMENDATIONS

1. Layer coefficients for unbound granular base or subbase materials should be

determined in the following manner:

a. Determine a 2 or a3 from the 1986 AASHTO Guide nomographs, or

I more accurately:

a2 = 0.249 log Eg - 0.977

a3 = 0.227 log Eg - 0.839.

b. Eg (resilient modulus of granular material) for a pavement section with

I a single granular layer can be determined by use of an elastic layer

analysis program such as KENLAYER or by the following equation:

Eg = 510.505(D1to.458(k1)o·426(E1)-o.1o1 (E.,Jo.201 (DJ0.oe1

For multiple granular layers, use of a program such as KENLAYER is

recommended.

c. 0 1 and 0 2 (0 1 = asphalt bound layer, 0 2 = granular base or subbase

I
 118

layer thickness) are assumed for a particular design trial. (Note: 0 1 is

the combined thickness of all asphalt-bound layers.)

d. E1 (asphalt material resilient modulus), is determined at a given design

temperature as developed in Volume I of this report knowing either

resilient modulus or mix design characteristics. If more than one

asphalt layer is involved, the weighted average can be obtained by Eq.

19:

E =
D1-.tE.1a)o.333 + D1°'.IE.1b)o.333r
eq [ D1a + D2a

e. E5 g (subgrade soil modulus) can be calculated in accordance with the

section "Resilient Modulus Regression Equation Subgrade Modulus".

f. k 1 can be determined by resilient modulus testing of the granular

material, or by estimation. See Table 15 for guidance.

2. Layer coefficients for cement treated soil base can be determined in the

following manner:

a. Determine a 2 from the 1986 AASHTO Guide nomograph, or more

accurately

a2 = -2. 717 + 0.49711 log Ee

b. Ee can be determined by static compressive modulus testing or by

Ee = 132,772 + 1314.9 qu

or less accurately

EC = - 2,575,557 + 84,084 C + 22,349 (yd)

 119

where:

Ec = static compressive chord modulus, psi

qu = unconfined compressive strength, psi

C = cement content, %

yd = dry density, pcf.

FUTURE RESEARCH NEEDS

1. A wider range of granular material sources needs to be tested to determine

the effects of particle shape and gradation on Eg. This would include several

other gradations to better define the effect of fines content.

2. A regression equation for prediction of k 1 of granular materials needs to be

developed to facilitate the prediction of Eg.

I 3. Because the Eg of granular materials is dependent on the resilient modulus of

I 4.
the subgrade, the prediction of Esg needs to be better defined.

The effects of permanent deformation of granular material needs to be tied

into the layer coefficient concept.

5. More sources of soil for cement treatment need to be included in the Ec

regression equation.
 120

ACKNOWLEDGEMENT

The authors wish to thank the MHTD for its sponsorship and support of this

research project. They also thank the UMR Department of Civil Engineering for its

support. Special thanks go to Mr. Kevin Hubbard for his assistance in the figure

preparation portion of the study, and to Aswath V. Rao for assistance in the

laboratory work.
 121

I
REFERENCES

I
I
122

I 1. AASHTO 1986 Guide for Design of Pavement Structures, AASHTO,

Washington, D.C., 1986.

I 2. Richardson, D.N., W.J. Morrison, and P.A. Kremer, "Determination of

AASHTO Drainage Coefficients," MCHRP Final Rpt. Study 90-4, Univ. of

Missouri-Rolla, Rolla, Missouri, 1993.

3. "The AASHO Road Test, Rpt. 5-Pavement Research," Hwy. Res. Bd. Spec.

Rpt. 61-E, Hwy. Res. Bd., 1962, 352 p.

4. Gomez, M. and M.R. Thompson, "Structural Coefficients and Thickness

Equivalency Ratios", Trans. Engrg. Series No. 38, Illinois Coop. Hwy. and

Trans. Series No. 202, Univ. of Illinois, 1983, 48 p.

5. Transportation Engineering Handbook, Ch. 50, Pavement, U.S. Forest

I Service, 1974, p. 51.1-54.4-16.

6. Van Til, C.J., B.F. McCullough, B.A. Vallerga, and R.G. Hicks, "Evaluation of

AASHO Interim Guides for Design of Pavement Structures, "NCHRP Rpt.

128, Hwy. Res. Bd., Washington, DC, 1972, 111 p.

7. Wang, M.C., T.D. Larson, and W. P. Kilareski, "Structural Coefficients of

Bituminous Concrete and Aggregate Cement Base Materials by the Limiting

Criteria Approach", Rpt. No. FHWA-PA-RD-75-2-4, Penn. State Univ., Univ.

Park, Penn., 1977, 52 p.

I 8. Sowers, G.F. "Georgia Satellite Flexible Pavement Evaluation and Its

Application to Design," Hwy. Res. Rec. 71, Hwy. Res. Bd., 1965, pp. 151-

I 171.
 123 I
9. Walters, R., "Implementation of the New AASHTO Design Practice", filfil.
Annual UMR Asphalt Conference, Univ. of Missouri-Rolla, Rolla, Missouri,

1988.

10. Rada, G. and M.M. Witczak, "Material Layer Coefficients of Unbound

Granular Materials from Resilient Modulus", Trans. Res. Rec. 852, 1982, pp.

15-21.

11. Jorenby, B.N. and R.G. Hicks, "Base Course Contamination Limits", Trans.

Res. Rec. 1095, Trans. Res. Bd., 1986, pp. 86-101.

12. Richardson, D.N., J.K. Lambert, and P.A. Kremer, "Determination of

AASHTO Layer Coefficients, Vol. I: Bituminous Materials", MCHRP Final Rpt.

Study 90-5, Univ. of Missouri-Rolla, Rolla, Missouri, 1993, 237 p.

13. Barksdale, R.D., "Laboratory Evaluation of Rutting in Base Course Materials",

Proc. 3rd lnt'I. Conf. on the Structural Design of Asphalt Pavements,

London, England, 1972, pp. 161-174.

14. Barksdale, R.D., S.Y. Itani, and T.E. Swor, "Evaluation of Recycled

Concrete, Open-Graded Aggregate," Trans. Res. Bd. 71 st Annual Meeting,

Trans. Res. Bd., Washington, D.C., 1992, 25 p.

15. Jin, M., K.W. Lee, and W.D. Kovacs, "Field Instrumentation and Laboratory

Study to Investigate Seasonal Variation of Resilient Modulus of Granular

Soils," Trans. Res. Bd. 71 st Annual Meeting, Trans. Res. Bd., Washington,

D.C., 1992, 30 p.

16. Kallas, B.F. and J.C. Riley, "Mechanical Properties of Asphalt Pavement I
 124

I Materials," Proc. 2nd lnt'I. Conf. on the Structural Design of Asphalt

Pavements, Univ. of Michigan, 1967, pp. 931-952.

I 17. Thom, N.H. and S.F. Brown, "The Effect of Moisture on the Structural

I Performance of a Crushed Limestone Road Base," Trans. Res. Bd. Annual

Meeting, Trans . Res. Bd., Washington, D.C., 1987, 24 p.

18. Hicks, R.G. and C.L. Monismith, "Prediction of the Resilient Response of

Pavements Containing Granular Layers Using Non-Linear Elastic Theory,"

Proc. 3rd lnt'I. Conf . on the Structural Design of Asphalt Pavements,

I London , Vol. I, 1972, pp . 410-427.

19 . Barker, W.R. and R.C. Gunkel, "Structural Evaluation of Open-Graded Bases

for Highway Pavements," Misc . Paper GL-79-18, U.S . Corps of Engrs.,

WES, 1979, 82 p .

I 20 . Thompson, M.R. and K.L. Smith, "Repeated Triax ial Characterization of

Granular Bases," Trans . Res. Rec. 1278, Trans. Res. Bd., Washington, D.C.,
I 1990, pp . 7-17.

I 2 1. Finn , F.N., C.L. Saraf, R. Kulkarni, K. Nair, W. Smith, and A. Abdullah,

I "Development of Pavement Structural Subsystems," NCHRP Rpt. 291, Hwy.

Res. Bd., Washington, D.C., 1986, 59 p.

22 . Rada, G. and M.W. Witczak, "Comprehensive Evaluation of Laboratory

I Resilient Moduli Results for Granular Material," Trans. Res. Rec. 810, Trans.

Res. Bd ., 1981, pp. 23-33.

I 23 . Haynes, J.H. and E.J. Yoder, "Effects of Repeated Loading on Gravel and
 125

Crushed Stone Base Course Materials Used in the AASHO Road Test," Hwy.

Res. Rec. 39, Hwy. Res. Bd., Washington, D.C., pp. 82-96.

24. Kandhal, P.S., J.B. Motter, and M.A. Khatri, "Evaluation of Particle Shape

and Texture: Manufactured vs. Natural Sands," NCAT Rpt. No. 91-3,

NACT, Auburn, Ala., 1991, 23 p.

25. "Test Method for Index of Aggregate Particle Shape and Texture," ASTM

D3398-87, Annual Book of ASTM Standards, Vol. 05.03 ASTM,

Philadelphia, PA, 1992, pp. 393-396.

26. "Standard Test Method for Particle Shape, Texture, and Uncompacted Void

Content of Fine Aggregate," Draft, National Aggregate Assn., Silver Spring,

MD, 1991, 12 p.

27. "Standard Test Method for Specific Gravity and Absorption of Coarse

Aggregate," AASHTO T-85 (1988), Standard Specifications for

Transportation Materials and Methods of Sampling and Testing, 15th ed.

Part II. Tests AASHTO, Washington, D.C., 1990, pp. 183-186.

28. "Standard Test Method for Specific Gravity and Absorption of Fine

Aggregate," AASHTO T-84 ( 1988), Standard Specifications for

Transportation Materials and Methods of Sampling and Testing, 15th ed.

Part II. Tests, AASHTO, Washington, D.C., 1990, pp. 179-182.

29. "Standard Test Method for The Moisture-Density Relations of Soils Using a

5.5 lb Rammer and a 12 in. Drop," AASHTO T-99 (1990), Standard

Specifications for Transportation Materials and Methods of Sampling and

 126

I Testing, 15th ed, Part II, Tests, AASHTO, Washington, D.C., 1990, pp.

226-230.

I 30. "Standard Test Method for Moisture-Density Relations of Soils Using a 10 lb

Rammer and a 18 in. Drop," AASHTO T-180 ( 1990), Standard

Specifications for Transportation Materials and Methods of Sampling and

I Testing, 15th ed. Part II. Tests, AASHTO, Washington, D.C., 1990, pp.

455-459.

31 . "Standard Test Method for Maximum Index Density of Soils Using a

I Vibratory Table," ASTM D4253-83, Annual Book of ASTM Standards, Vol.

I 04.08, ASTM, Philadelphia, PA, 1990, pp 572-583.

32. "Standard Test Method for Minimum Index Density of Soils and Calculation

of Relative Density," ASTM D4254-83, Annual Book of ASTM Standards,

Vol. 04.08, ASTM, Philadelphia, PA 1990, pp. 584-590.

33. Interim Method of Test for Resilient Modulus of Subgrade Soils and

Untreated Base/Subbase Materials, AASHTO T-XXXC-91, AASHTO,

I Washington, D.C., 1991, p. 1-35.

34 . Resilient Modulus of Unbound Granular Base/Subbase Materials and

Subgrade Soils, SHRP Protocol P46, S.H.R.P., Washington, D.C., 1992.

35. Claros, G., W.R. Hudson, and K.H. Stakoe, II, "Modifications to the Resilient

Modulus Testing Procedure and the Use of Synthetic Samples for Equipment

Calibration," Trans. Res. Bd. 69th Annual Meeting, Trans. Res. Bd.,
I Washington, D.C., 1990, 27 p.
 127

36. SYSTAT, 1990, Systat, Inc., Evanston, Illinois.

37. Huang, Y.H., Pavement Analysis and Design, Prentice Hall, Englewood

Cliffs, NJ, 1993, 805 p.

38. Witczak, M.W., and B.E. Smith, "Prediction of Equivalent Granular Base

Moduli Incorporation Stress Dependent Behavior in Flexible Pavements,"

ASCE Trans. Journal, 1981.

39. Thompson, M.R. and Q.L. Robnett, "Resilient Properties of Subgrade Soils,"

ASCE Trans. Journal, TE1, 1979, pp. 71-89.

40. Kersten, M.S., "Progress Report of Special Project on Structural Design of

Non-rigid Pavements; Subgrade Moisture Conditions Beneath Airport

Pavements," Hwy. Res. Bd. 25th Annual Meeting, V. 25, Hwy. Res. Bd.,

Washington, D.C., 1945, pp. 450-463.

41. "Method for Determining the Potential Vertical Rise, PVR," Texas Hwy.

Dept., Austin, Texas, 1972, 8 p.

42. Elliot, R.P., "Selection of Subgrade Modulus for AASHTO Flexible Pavement

Design," Trans. Res. Bd. 71 st Annual Meeting, 1992, 12 p.

43. Seed, H.B., C.K. Chan, and C.E. Lee, "Resilient Characteristics of Subgrade

Soils and Their Relation to Fatigue Failures in Asphalt Pavements," Proc.

International Conf. on the Structural Design of Asphalt Pavements, Univ. of

Michigan, 1962, pp. 611-636.

44. Shook, J.F., and H.Y. Fang, "Cooperative Materials Testing Program at the

AASHO Road Test," Hwy. Res. Bd. Spec. Rep. 66, Hwy. Res. Bd., 1961,
I
I 128

I pp. 59-102.

45. Coffman, B.S., D.C. Kraft, and J. Tamayo, "A Comparison of Calculated and
I Measured Deflections for the AASHO Test Road," Proc. of Assn. of Asphalt

I Paving Tech., Vol. 33, 1964, pp. 54-90.

46. Skok, E.L., Jr., and F.N. Finn, "Theoretical Concepts Applied to Asphalt
I Concrete Pavement Design," Proc. International Conf. on the Structural

I Design of Asphalt Pavements, Univ. of Michigan, 1962, pp. 412-440.

47. Koperman, S., G. Tiller, and M. Tseng, "ELSYM5, Interactive Microcomputer

I Version," FHWA Rpt. No. FHWA - TS-8-206, 1986, 33 p.

48. Witczak, M.W., Development of Regression Model for Asphalt Concrete

Modulus for Use in MS-1 Study, Asphalt Institute, 1978, 39 p.

I 49. Traylor, M.L., "Characterization of Flexible Pavements by Non-Destructive

Testing," PhD Dissertation, Univ. of Illinois-Urbana-Champaign, Illinois,

I 50.
1978, 213 p.

"The AASHO Road Test, Rpt. 2-Materials and Construction" Hwy. Res. Bd.

I Spec. Rpt. 618, Hwy. Res. Bd., 1962. 173 p.

I 51. Soil-Cement Laboratory Handbook, PCA, Skokie, Ill., 1971, 61 p.

52. Felt, E.J., and M.S. Abrams, "Strength and Elastic Properties of Compacted
I Soil-Cement Mixtures", Bulletin D16, PCA, 1957, 26 p.

I 53. "Standard Method of Test for Compressive Strength of Concrete Specimens,

"AASHTO T22-90 (1990), Standard Specifications for Trans. Mtrls. and

I Methods of Sampling and Testing, 15th Ed., Part II, Tests, AASHTO,
 129
Washington, D.C., 1990, pp. 12-15.

54. "Standard Test Method for Static Modulus of Elasticity and Poisson's Ratio

of Concrete in Compression," ASTM C469-87a, Annual Book of ASTM

Standards, Vol. 04.02, ASTM, Philadelphia, PA, 1989, pp. 236-239.

55 . QUATTRO PRO, Borland International, Inc.

56. TABLECURVE, Jandel Sc ientific, Inc. I

57 . Murray, L.T., "Flexible Design and Experimentation in Missouri," Proc. of

Assn . of Asphalt Paving Tech., Vol. 34, 1965, pp . 496-519.

58 . Saville , V.B. and W.C. Davis, MHTC Geology and Soils Manual, Von

Hoffman Press , Jefferson City, Missouri, 1962, 436 p.

I
I 130

I
I
I APPENDIX A

I
DATA SET FOR RESILIENT MODULUS
I
I REGRESSION EQUATION (Eq. 10)

I
I
I
I
I
I
I
I
I
I
I
 -
Resilient Modu li for Gran ular Base Material (E 1 = 130,000 psi) .

E1 = 130,000 psi
---- - --
E0 (filename/base/subgrade), psi

D2(in .)

4 12 18

o, Esg k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k,= 3000 k,= 11,000 k, = 1800 k, = 3000 k, = 11 ,000
(in.) k2 = 0.776 k2 = 0 .653 k2 = 0.341 k2 = 0.776 k2 =0.653 k2= 0.341 k2 = 0 .776 k2 = 0.653 k2 = 0.341

12vs121 12vs122 12vs123 12vs181 12vs182 12vs183

very soft
1000-5662
• • • 21 ,840 24,800 33,670 22,460 25,270 33,820

2820 2980 3383 3940 4089 4446

12M41.DAT 12M42.00 12M45.DAT 12M121 12M122 12M123 12M181 12M182 12M183

2 medium
4716-12,342 19,860 21,790 28,900 24,220 26730 34,350 23,160 25,830 33,840

6725 6762 6863 7501 7545 7650 8856 9040 9466

12541 12542 12543 125121 125122 125123 125181 125182 125183

stiff
7605-17,002 23,610 25,560 32,380 25,060 27,440 34,820 23,450 26,070 33,950

11,300 11,320 11 ,390 11 ,890 11 , 930 12,030 12,710 12,890 13,310

18vs541 18vs542 18vs543 18vs121 18vs122 18vs123 18vs181 18vs182 18vs183

very soft
1000-5662 6850 7976 10,260 8761 10,990 20,520 9435 12,110 22,970

3487 3457 3435 4227 4257 4503 4798 4865 5176

18m41 18m42 18m43 18m121 18m122 18m123 18m181 18m182 18m183

8 medium
4716-12,342 10,740 12,460 18,970 11,330 13,910 23,760 10,990 13,820 24,560

8204 8132 8027 9786 9760 9896 10,710 10,710 10,930

18541 18542 18543 18121 185122 185123 185181 185182 185183

stiff
7605-17,002 12,850 14,890 22 ,390 12,300 15,050 25,070 11,590 14,480 25,160

12,230 12,210 12,140 13,960 13,910 13,980 15,060 15,030 15,160

115vs41 115vs42 115vs43 115vs1 21 115vs 122 115vs123 115vs181 115vs182 115vs183
very soft
1000-5662 6628 8255 14,450 6991 9076 17,960 7138 9419 19,290

5412 5346 5237 5662 5662 5662 5662 5662 5662

 115m41 115m42 115m43 115m121 115m122 115m123 115m181 115m182 115m183
15 medium
4716-12,342 8651 10,680 18,540 8298 10,680 20,320 8106 10,610 20,860

11,200 11,070 10,810 11,950 11,830 11,650 12,260 12,170 12,050

115541 15542 115543 115121 1155122 1155123 1155181 1155182 1155183

stiff
7605-17,002 9645 11,850 20,370 8804 11,290 21,240 8444 11,010 21,560

15,420 15,250 14,930 16,370 16,220 15,980 16,730 16,620 16,500

3
[
D E 0 .333 + D E o.333
]
. . la la lb lb
Note: Eg values based on °S = 10, 60%; 0 3 = 0 m.; for multiple asphalt layers, use Eeq = ( } ( }
D1a + D1b
• statistical outliers due to non-convergence in iterative process.
- - ----------- - - - -
Resil ient Moduli for Granular Base Material (E 1 = 500,000 psi) .

E1 = 500,000 psi E0 (filename /ba se/subgrade), psi

D2(in .)

4 12 18

o, Esg k, = 1800 k, = 3000 k, = 11,000 k,= 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11 ,000
(in .) k2 = 0 .776 k2 = 0.653 k2 = 0 .341 k2 = 0 .776 k2 = 0 .653 k2 = 0.341 k2 = 0.776 k2 = 0.653 k2 = 0 .341
22vs541 22vsl 21 22vs122 22vs123 22vsl 81 22vs182 22vs183
very soft
5938 + +
1000-5662 18,370 21,160 30,500 19,610 22,470 31 ,610

1000 2820 2986 3447 3946 4107 4515

22m41 22m42 22m43 22ml 21 22ml 22 22ml 23 22ml 81 22ml 82 22m183

2 medium
4716-12,342 18,760 20,580 27 , 180 21,810 24,260 32,590 21,080 23,730 32,390

6798 6876 6896 7505 7542 7873 8955 9121 9630

22541 22542 22543 225121 225123 225123 225181 225182 225183

stiff
7605-17,002 22,830 24,700 31,130 23,370 25,780 33,430 21,680 24,340 32,720

11,270 11,290 11,290 11 ,960 12,000 12,090 12,870 13,040 13,510

28vs541 28vs542 28vs43 28vsl 21 28vsl 22 28vsl 23 28vsl 81 28vs182 28vsl 83

very soft
1000-5662 6933 8677 15,310 7162 9337 18,550 7245 9683 19,960

5374 5312 5193 5577 5504 5459 5662 5662 5662

28m41 28m42 28m43 28ml 21 28ml 22 28ml 23 28ml 81 28m182 28m183

8 medium
4716-12,342 9363 11,510 19,700 8695 11,260 21,290 8353 11,030 21,750

11,090 10,950 10,660 11,640 11,530 11,330 12,030 11,910 11,800

28541 28542 28543 285121 285122 285123 285181 285182 285183

stiff
7605-17,002 10,570 12,900 21,590 9346 12,060 22,350 8760 11,540 22,410

15,290 15,110 14,720 16,060 15,920 15,620 16,540 16,390 16,190

215vs41 215vs42 215vs43 215vs121 215vs122 215vsl 23 215vs181 215vsl 82 215vsl 83

very soft
1000-5662 5714 7613 15,850 5916 7992 17,230 6093 8267 18,090

5662 5662 5662 5662 5662 5662 5662 5662 5662

 215m41 215m42 215m43 215m121 215m122 215m123 215m181 215m182 215m183
15 medium
4716- 12,342 6913 9111 18,320 6679 9033 18,980 6674 9078 19,340

12,340 12,340 12,340 12,340 12,340 12,340 12,340 12,340 12,340

215541 215542 215543 2155121 2155122 2155123 2155181 2155182 2155183

stiff
7605-17,002 7506 9840 19,390 7022 9424 19,580 6932 9347 19,820

17,002 17,002 17,002 17,002 17,002 17,002 17,002 17,002 17,002

3
D E 0.333 + D E 0.333
. . la la lb lb
Note: Eg values based on °S = 10, 60%; D3 = 0 m.; for multiple asphalt layers, use Eeq = ( ) ( )
[ D1a + D1b ]
• statistical outliers due to non-convergence in iterative process
--------
Resilient Moduli for Granular Base Material (E 1 = 2, 100,000 psi) .

E1 = 2,100,000 psi
-- -- ---
E0 (filename/base/subgrade). psi

D2 (in .) BASE (Granular)

4 12 18

o, Esg k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11,000
(in.) k2= 0.776 k2 = 0 .653 k2 = 0 .341 k2 = 0.776 k2 = 0.653 k2 = 0.341 k2 = 0.776 k2 = 0 .653 k2 = 0 .341

32vs41 32vs42 32vs43 32vs121 32vs122 32vs123 32vs181 32vs182 32vs183

very soft
1000-5662 9219 10,570 15,220 14520 17450 27660 15,730 18,760 29,090

2293 2294 2333 3176 3313 3732 4098 4243 4667

32m41 32m42 32m43 32ml 21 32m122 32m123 32ml 81 32m182 32m183

2 medium
4716-12,342 16,770 18800 26130 18110 20940 30260 17,550 20,490 30,270

7341 7336 7340 4906 8031 8456 9397 9520 9958

32541 32542 32543 325121 325122 325123 325181 325182 325183

stiff
7605-17,002 20300 22430 29810 19690 22440 31310 18,190 21,180 30,760

11770 11760 11760 12200 12220 12280 13,460 13,570 13,950

38vs41 38vs42 38vs43 38vs121 38vs122 38vs123 38vs181 38vs182 38vs183

very soft
1000-5662 5184 7074 15,390 5257 7326 16710 5396 7535 18,460

5662 5662 5662 5662 5662 5662 5662 5662 5662

38m41 38m42 38m43 38m121 38m122 38m123 38m181 38m182 38m183

8 medium
6595 8821 18,250 6168 8552 18680 6056 8479 19,020
4716-12,342
12,340 12,340 12,340 12340 12340 12340 12,340 12,340 12,340

38541 38542 38543 385121 385122 385123 385181 385182 385183

stiff
7264 9650 19,510 6520 9028 19540 6293 8826 19,590
7605-17,002
17,002 17,002 17,002 17002 17002 17002 17,002 17,002 17,002

315vs41 315vs42 315vs43 315vs121 315vs122 315vs123 315vs181 315vs182 315vs183

very soft
4906 6868 16,160 5186 7248 16,930 5410 7534 17,430
1000-5662
5562 5662 5662 5662 5662 5662 5662 5662 5662
 315m41 315m42 315m42 315m121 315m122 315ml 23 315ml 81 315ml 82 315ml 83
15 medium
4716-12,342 5455 7576 17,280 5568 7760 17,780 5717 7953 18, 130

12,340 12,340 12,340 12,340 12,340 12,340 12,340 12,340 12,340

318841 315542 315543 3155121 3155122 3155123 3155181 3155182 3155183

stiff
7605-17,002 5735 7933 17800 5742 7963 18,150 5827 8109 18.430

17002 17002 17002 17,002 17,002 17,002 17,002 17,002 17,002

- - 3
D la(Ela )°"333 + D lb(Elb)° .333
Note: Eg values based on °S = 10, 60%; 0 3 = 0 in.; for multiple asphalt layers, use Eeq =
D1a + D1b