. :t. :L

.3...
hfi‘flxflf...“

-
,7 .1.

JAM-no»n

uA—mqnn-u-

A? 1
. «I. .. ,1
fink. fris.

. 9.129? .

A . 1:;
:. z

.u

$.32 . , .2...“ :1
A. .4 . "uu‘.

3:5.
5.. 3...;

_
I... '3

s !. , 23”..
. .. a

.- WWI-Y7.» ."m

nu»:

won. 1",.unqu .muq

v;-

v-‘\

 

DETER

DETERMINING THE CAUSES OF TOP-DOWN CRACKS IN
BITUMINOUS PAVEMENTS

By

Michael Robert Schorsch

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE

Department of Civil and Environmental Engineering

August, 2003

r- gunk

it'"“

Top L10“ 7.

 

paiemsnl SUTIJCC .
lllflfi‘éllig Ircguz‘r:
lll‘l'éSiigtlllOllS \\ CT;

stresses at the w»:

”Y‘H I i I; w
ltmlug .nutdlc Ili-

llsuencd b} mozy
AC mixtures to m

'ln'm n ‘ ‘
liq-u: sl0n mUtIL'i

segregation baScd t

 

ABSTRACT

DETERMINING THE CAUSES OF TOP-DOWN CRACKS IN BITUMINOUS
PAVEMENTS

By

Michael Robert Schorsch

Top down cracks (TDC) are longitudinal or transverse cracks that initiate at the
pavement surface and propagate downward and outward. They have been observed with
increasing frequency on roads in Michigan. In this study, field and laboratory
investigations were conducted to determine the factors that affect the load-induced tensile
stresses at the pavement surface and the tensile strength of the AC mixes, and hence,
affect the susceptibility of the pavement surface to TDC potential. Results of laboratory
testing indicate that the AC mixtures used as surface courses in Michigan may be
weakened by moisture damage. Further, segregation increases the susceptibility of the
AC mixtures to moisture damage, and hence, increases TDC potential. Finally, a crack
propagation model was developed as a function of pavement age and degree of

segregation based on distress survey data.

 

To Rut/i

 

iii

 

illSi and 5
his slim. his gun;
Stud)“. 1 “Mid til: _
committee. Dr. .\'.

no: tor their in: .

 

‘ ‘1 l
“t‘tlltl TWA e 31137.6.

 

 

liiould .33.
ei‘Etcellcnce l PRt
,__ ‘ .
i.tL-’l§l"\'ll’.&lloli. 31%

possible. espe“ ‘

\. id i
llDOT adt ism c

:rterest. SPCL‘idi 1i:

including Kurt B;

. l-
lhetiicdge of MD

71;}.Aa...‘
3m. Nuance \Vli

researcher Dr, Tm

so much to this iii

Finally m

par

ienc
e and men

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my major advisor, Dr. Gilbert Baladi, for
his effort, his guidance, and most importantly, his seemingly endless patience during this
study. I would also like to extend my thanks to the other members of the advisory
committee, Dr. Neeraj Buch and Dr. Karim Chatti for their advice and encouragement. If
not for their influence, this thesis would not have been possible, and it is unlikely that I
would have attended graduate school at all. '

I would also like to extend my thanks to those of the Pavement Research Center
of Excellence (PRCE) at Michigan State University, the Michigan Department of
Transportation, and the Michigan Asphalt Paving Association that made this study
possible, especially _Mr. Robert Thompson for his financial support. Thanks also to the
MDOT advisory committee headed by Vernon Bamhart for their valuable comments and
interest. Special thanks are due to MDOT personnel for their efforts in data collection,
including Kurt Bancroftfor FWD testing, Frederick Carian for his near encyclopedic
knowledge of MDOT bituminous mixtures and Gerald Sweeney for his unique insights
and assistance with laboratory testing. Also, I would like to thank my officemate and co-
researcher Dr. Tunwin Svasdisant, who throughout our two year relationship contributed
so much to this thesis.

Finally, my loving thanks to my family for their support and also to Ruth for her

patience and encouragement.

iv

 

LIST OF TABLE:
UST 0F FIGUR
CHAPTER 1
‘NTRODUCTth
1.0 Immdhcrm
2‘0 Oblt‘fih cs

3.” HWOII’RSK
it) Rescarchl‘

 

"H I ii;
4.2 F11.
4'3 Li?
4,4 D;
4.5 V;
4-6 Co

5.0 Thesis L3:
CHAPTER 2

3.0 TOP‘Do“,

an59$ 0.“
J FacIOrs d1.
4'1 Ti:
42 \V‘
4.3 T1

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................ ix
LIST OF FIGURES ...................................................................................... xii
CHAPTER 1
INTRODUCTION ......................................................................................... 1
1.0 Introduction .................................................................................................... 1
2.0 Objectives ...................................................................................................... 1
3.0 Hypothesis ...................................................................................................... 2
4.0 Research Plan ................................................................................................ 2
4.] Literature Review .............................................................................. 4
4.2 Field Investigation ............................................................................. 4
4.3 Laboratory Investigation .................................................................... 5
4.4 Data Analysis .................................................................................... 6
4.4.1 Analysis of the Measured FWD Deflection Data .................. 6
4.4.2 M echanistic Analysis of Flexible Pavements ........................ 6
4.5 Validation of the Results of the Analyses .......................................... 7
4.6 Conclusions and Recommendations .................................................. 7
5.0 Thesis Layout ................................................................................................. 7
CHAPTER 2
LITERATURE REVIEW ............................................................................... 9
1.0 Introduction .................................................................................................... 9
2.0 Top-Down Cracks in Flexible Pavements ..................................................... 9
3.0 Causes of TDC ............................................................................................... l 1
4.0 Factors affecting the tensile strength of the AC Layer .................................. 21
4.1 Thickness and Modulus of Each Pavement Layer ............................ 21
4.2 Wheel Load and Configuration .......................................................... 23
4.3 Tire-Pavement Interaction ................................................................. 24
4.3.1 Tire-Pavement Contact Area .................................................. 26
4.3.2 T ire-Pavement Contact Stresses ............................................ 28
4.3.3 P avement Responses .............................................................. 39
4.3.4 P avement Temperature .......................................................... 40

511

6.")

CHAPTER 3

  
 

Tensile 3':
5.1 At
5.2 (or
5.3 P47
5.4 1:?
5.5 S'T;

Top-De“

 

FIELD AND LAE

1.0
3.0

q

:1]

4.0

6.0

General
Selection Cl
Field 1m '25
3.1 Tes
D25
H.
.\‘:.;
C0:

J) L!) ’J ‘JJ J)
o LII 4.. (’1 IJ

“mill-act
5 l M

i2}.

6,6 \‘Ii
‘ 13H.
6 ’ M-
13:“.

6‘8 M5

6.9 11-5

5.0 Tensile Strength of AC Mixtures .................................................................. 42
5.1 AC Mixture Design ............................................................................ 42
5.2 Compaction ........................................................................................ 44
5.3 Particle Segregation ........................................................................... 45
5.4 Temperature and Temperature Gradient ............................................ 49
5.5 Stress State ......................................................................................... 51
6.0 Top-Down Crack Propagation Models .......................................................... 51
CHAPTER 3
FIELD AND LABORATORY INVESTIGATION ........................................... 56
1.0 General ................................................................................................ 56
2.0 Selection of Pavement Sections ..................................................................... 56
3.0 Field Investigation ......................................................................................... 58
3. 1 Test Layout ........................................................................................ 58
3.2 Distress Survey .................................................................................. 61
3.3 FWD ................................................................................................ 62
3.4 Nuclear Density ................................................................................. 64
3.5 Coring ................................................................................................ 64
3.6 Boreholes ........................................................................................... 67
4.0 Laboratory Investigation ................................................................................ 67
4.1 Core Preparation ................................................................................ 67
4.2 Specific Gravity Tests ........................................................................ 68
4.3 Indirect Tensile Cyclic Load Test ...................................................... 68
4.4 Indirect Tensile Strength Tests .......................................................... 75
4.5 Extraction of Aggregate ..................................................................... 77
5.0 Manufactured Test Specimens ....................................................................... 78
5.1 Manufacture of Specimens ................................................................ 81
5.2 Moisture Conditioning of Specimens ................................................ 82
6.0 Testing Summary ........................................................................................... 83
6.1 I-194 North-Bound, Section 1, Site 1 and 2
(11941-11 and 11941-12) .................................................................. 83
6.2 I-75 South-Bound, Section 1, Site 1 and 2
(10753-11 and 10753-12) .................................................................. 86
6.3 US-27 South-Bound, Section 1, Site 1 (20273-1 1) ............................ 88
6.4 US-41 North Bound, Section 1, Site 1 and 2
(20411-11 and 20411-12) .................................................................. 88
6.5 M-15 South-Bound, Section 1, Site 1 and 2
(30153-11 and 30202-12) .................................................................. 90
6.6 M-ZO East-Bound, Section 1, Site 1 and 2
(30202-11 and 30202-12) .................................................................. 90
6.7 M-20 West-Bound, Section 2, Site 1 and 2
(30202-21 and 30202-22) .................................................................. 91
6.8 M-42 East-Bound, Section 1, Site 1 (30422-11) ................................ 93
6.9 M-SO East-Bound, Section 1, Site 1 (30502-11) ................................ 94

vi

6.11.t \'.-
6.11 .\'.-
6.13 .\T‘
6.13 .\'.

6.14 .\‘;

CHAPTER 4

DATA ANALYE

’ 1.
1.1.1 02110311

‘ l f
i D“
«I .513» E

3.1 D
5“ C
2.4 S
2.5 O

3.0 Fromm-.1

“'13 Classztie

5.11

6.0

9.0

4.1 T

6.10 M-SO East-Bound, Section 2, Site 1 (30502-21) ................................ 94

6.11 M-52 North Bound, Section 1, Site 1 (30521-11) ............................. 95
6.12 M-52 North Bound, Section 2, Site 1 (30521-21) ............................. 96
6.13 M-55 East Bound, Section 1, Site 1 and 2 (30552—11 and 30552-12)
96
6.14 Marsh Road ........................................................................................ 100
CHAPTER 4

DATA ANALYSIS ..................................................................................... 104
1.0 General ................................................................................................ 104
2.0 Distress Survey Data ...................................................................................... 104
2.1 Distress ............................................................................................... 105
2.3 Crack Measurement ........................................................................... 106
2.4 Segregation ........................................................................................ 119
2.5 Other Distresses ................................................................................. 119
3.0 Propagation of TDC ....................................................................................... 119
4.0 Classification and Identification of TDC ....................................................... 125
4.1 TDC Descriptions and Classification ................................................. 127
4.1.1 Longitudinal Top-Down Cracks (LTDC) Category .............. 128
4.1.2 T ransverse Top-Down Cracks (TTDC) Category .................. 132
4.1.3 C rack in Segregated Areas Category ..................................... 132
4.2 TDC Identification ............................................................................. 135
4.2.1 Longitudinal Cracks ............................................................... 135
4.2.2 Transverse Cracks .................................................................. 136
4.2.3 N etwork of Longitudinal and Transverse Cracks .................. 137
5.0 AC Density ................................................................................................ 137
5.1 Nuclear Density Data ......................................................................... 138
5.2 Specific Gravity Tests ........................................................................ 139
6.0 Pavement Thickness and Material Data ......................................................... 141
6.1 AC Mix Type ..................................................................................... 141
6.2 Core Thickness Data .......................................................................... 144
7.0 Deflection Data .............................................................................................. 146
8.0 Backcalculation Procedure ............................................................................. 155
8.1 Overview of Backcalculation Procedure ........................................... 155
8.2 Depth to Stiff Layer ........................................................................... 165
8.2.1 B quivalent Pavement Modulus .............................................. 166
8.2.2 Three- and four-layer system Backcalculation ...................... 168
8.3 Convergence Criteria ......................................................................... 172
8.4 Acceptance Criteria ............................................................................ 175
8.5 The Temperature Correction of the Backcalculated AC Moduli ....... 177
9.0 Backcalculation Results ................................................................................. 178
9.1 US-27 Southbound, Section 1, site 1 (20273-11) .............................. 184

9.2 US-41 North bound, section 1 sites 1 and 2 (20411-11, 20411-12)..187
9.3 M-20 East Bound, Section 1, Sites 1 and 2

vii

9.4
9.5
90
9."
9 S
9.9
9.1"..
19.0 Inclzrec'.
1111

10.2

11.0 Extracti
11.1 '

1 1

I... .
13.0 Morstar
12.] -
12.” 1
1‘ 3

13.0 Sumrita.
CHAPTER 5

SUMMARY C

1 '2
1.1.1 Summai
1 .
--0 Conch:

if)
~- RL‘CORE"

REFERENCE

(30202-11 and 30202-12) .................................................................. 188

9.4 M-42 East-Bound, Section 1, Site 1 (30422—1 1) ................................ 189

9.5 M-50 East-Bound, Test Section 1, Test Site 1 (30502-11). ............... 191

9.6 M-50 East-Bound, Section 2, Site 1 (30502-21) ................................ 192

9.7 M-52 North-Bound, Section 1, Site 1, (30521-11) ............................ 193

9.8 M-52 North-Bound, Section 2, Site 1, (30521-21) ............................ 194

9.9 M-55 East Bound, Section 1, sites 1 and 2 (305552-11,12) .............. 194

9.10 Marsh Road (2 Test Sites) .................................................................. 196

10.0 Indirect Tensile Tests ..................................................................................... 197

10.1 Indirect Tensile Cyclic Load Tests (ITCLT) ..................................... 197

10.2 Indirect Tensile Strength Tests (ITST) .............................................. 200

10.2.1 Indirect Tensile Strength .................................................... 200

10.2.2 Equivalent Modulus ........................................................... 202

11.0 Extraction of Aggregate ................................................................................. 203

11.1 Gradation Analysis ............................................................................. 203

11.2 Asphalt Content ................................................................................. 205

12.0 Moisture Susceptibility Testing ..................................................................... 209

12.1 4B Mixture ......................................................................................... 211

12.2 4C Mixture ......................................................................................... 215

12.3 4E10 Mixture ..................................................................................... 216

13.0 Summary of Data Analyses ........................................................................... 216
CHAPTER 5

SUMMARY CONCLUSIONS AND RECOMMENDATIONS ........................ 217

1.0 Summary ........................................................................................................ 217

2.0 Conclusmns ...................................... 219

3.0 Recommendations .......................................................................................... 221

REFERENCES ............................................................................................ 222

viii

Table 2.1

Table 3.1

Table 3.2

Table 3.3

Table 3.4
Table 3.5
Table 4.1

Table 4.2

Table 43
Table 4.4
Table 4.5

Table 4.6

Table 4.7
Table 4.8
Table 4.9
Table 4.10

Table 411

Table 4.12

(It

s:

U)
,.

If)

 

(j)
I;

Table 2.1

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 3.5

Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table 4.5

Table 4.6

Table 4.7

Table 4.8

Table 4.9

Table 4.10

Table 4.11

Table 4.12

LIST OF TABLES

A summary of conclusions made by various investigators ................ 20

Summary of the test sections and test sites selected for field

and laboratory investigations ............................................................. 57
Sensitivity of LVDTs used in the ITCLT apparatus ......................... 71
Sieves and corresponding opening sizes used in the sieve

analysis ............................................................................................... 79
Summary of field investigations ........................................................ 84
Summary of laboratory investigations ............................................... 85
Distresses observed on flexible pavements ........................................ 107

The number of projects exhibiting the indicated distress

observed during the May 2002 distress survey .................................. 108
Summary of crack measurements ...................................................... 1 10
Segregation sites from Chang (2000) ............................................... 120
Length of cracks in segregated areas ................................................. 121

Percent crack measurements on segregation projects (from

Chang, 2000). ..................................................................................... 122
Summary of nuclear density tests ...................................................... 140
Average specific gravity tests of pavement cores .............................. 142
Summary of mix types on investigated pavements ............................ 143
Variation in core thickness for each test site ..................................... 145

Differences in pavement thickness in and outside the wheel
paths ................................................................................................... 147

Coefficient of variation of the deflection data of rubblized and
flexible pavements ............................................................................ 149

ix

Table 4.13
Table 4.14
Table 4.15

Table 4.16

Table 4.17
Table 4.18

Table 4.19

Table 4.20
Table 4.21
Table 4.22
Table 4,23
Table 4.34
Table 4.25
Table 4-36
Table 4.27

lab]e 4.38

1’,
,.

L”) -—-4

mo

.‘\\;
t.\'-.:

 

A\ L.

-‘\\ g
AC

CO;-
1113 :

CUL‘
The 1

IIIK

\.
(in.
l_“

Table 4.13

Table 4.14

Table 4.15

Table 4.16

Table 4.17

Table 4.18

Table 4.19

Table 4.20

Table 4.21

Table 4.22

Table 4.23

Table 4.24

Table 4.25

Table 4.26

Table 4.27

Table 4.28

Rubblized pavement D1 through D7 deflections, I-75, test
section 1, test site 1 ............................................................................ 150

Rubblized pavement D1 through D7 deflections, 1-75, test
section 1, test site 2 ............................................................................ 151

Flexible pavement D1 through D7 deflections, US 27, test
section 1, test site 1 ............................................................................ 151

Thickness and moduli of pavement layers and roadbed soil of a
pavement section having stiff layer at 300 in. used in the

analyses .............................................................................................. 1 73

Thickness and moduli of pavement layers and roadbed soil of a
pavement section with no stiff layer used in the analyses ................. 173

The acceptance criteria for the results of different
backcalculations at each FWD test location ...................................... 176

The reasons for each of the acceptance criteria ................................. 176

Inventory data (material type and layer thickness) and the
average AC core thickness ................................................................. 179

The layer thicknesses used in the backcalculation of layer
moduli ................................................................................................ 180

Average measured deflections and mid-depth AC temperature
(MDT), flexible and rubblized pavements ......................................... 181

Average backcalculated moduli and the temperature corrected
AC moduli for the flexible pavement test sites .................................. 182

Average backcalculated moduli and the temperature corrected
AC moduli for the rubblized pavement test sites ............................... 183

Coefficients of variation of the backcalculated layer moduli of
the flexible pavement test sites .......................................................... 185

Coefficients of variation of the backcalculated layer moduli of
the rubblized pavement test sites ....................................................... 186

Moduli calculated from ITCLT for specimens cut from
different depths in the AC layer ......................................................... 198

Average indirect ITST data ................................................................ 199

Table 4.29

Table 4.29 Sieve analysis of extracted aggregate and asphalt content ................ 204

xi

Figure 1.1 F1.
Figure 2.1 TC
Figure 2.2 Pi:

T1
1”

Figure 2.3 T :t

Figure 2.4 1.5.
figure2.5 D;

Figure 2.6 ij

. .1

I D
Figu re 27 p10
red

8...
\\

Figure 2.3 Cu:

Figure 29 Cr:

and
ligUre 2,10 Gra

se-
ligure 2,11 Lita

A(‘
Fig” 3-1 T...
Figure 3.2 m

Figure 1.1
Figure 2.1

Figure 2.2

Figure 2.3

Figure 2.4
Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 2.10

Figure 2.11

Figure 3.1

Figure 3.2

LIST OF FIGURES

Flowchart of the research plan ........................................................... 3
Top-down cracks in a rubblized pavement (US-131) ........................ 10

Photographs showing the development of Top-Down Cracking

(A = Stage 1, B = Stage 2, C = Stage 3) and cores showing

TDC, core 1- 1-96, 2- M-20, 3- Marsh Road, 4- M-37, and 5— I-

194 ...................................................................................................... 12

Tire prints of a radial and a bias ply tire having 9.2-kip wheel
load and 104-psi inflation pressure (De Beer and Fisher, 1997);

(the scale is in mm) ............................................................................ 27
Load transfer mechanisms of a tire .................................................... 29
Detail of stresses induced by truck tire treads ................................... 31

Plots of vertical, lateral and longitudinal contact stresses of a
bias ply tire with various wheel loads and inflation pressures
(De Beer et a1, 1997) .......................................................................... 35

Plots of vertical, lateral and longitudinal contact stresses of a

radial tire with various wheel loads and inflation pressures (De

Beer et a1, 1997) ................................................................................. 36
Comparisons between the vertical and lateral contact stresses

of a radial tire and of a bias ply tire at various wheel loads and

inflation pressures (De Beer et a1, 1997) ........................................... 37

Gradations for different levels of segregation (after Khedaywi
and White, 1995) ................................................................................ 47

Gradations used by Brown et a1 (1989) for different levels of
segregation ......................................................................................... 47

Load-induced different stress state at the top and bottom of the
AC layer ............................................................................................. 52

Typical layouts of test sections and test sites ..................................... 59

Typical test site layout showing FWD tests (X) and core
locations (0) ....................................................................................... 60

xii

Figure 33
Figure 3.4
Figure 3.5

Figure 3.6
Figure 3.7
Figure 3.8
Figure 3.9
Figure 3.10
Fl‘a’llre 3.11
Fleure 3.12

FIt’ure 3.13
Figure 3.14
Flgure 4.]
F"Jllre 42
Figure 4.3

TIgIH-e 4.4
Figure 4.5

Figure 4.6

Figure 4.7

 

 

Te

Cir

Gr

Figure 3.3
Figure 3.4
Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.1]

Figure 3.12

Figure 3.13
Figure 3.14
Figure 4.1
Figure 4.2
Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Top view of the sensor configuration of the MDOT FWD ............... 63
Troxler nuclear density meter ............................................................ 65

Drilling cores using a power rotary drill ........................................... 66

ITCLT and ITST specimen holder for a 6-in diameter and 3-in
thick test specimen ............................................................................. 70

One load cycle consisting of O. 1 -second load—unload period
and 0.4-second relaxation period ....................................................... 73

Test specimen in the specimen holding device during an

ITCLT test ......................................................................................... 73
Relationship between actual core thicknesses and required

minimum thickness for ITCT and ITST testing ................................ 80
Modified test site layout of M-20, test section 1, test sites 1 and

2 (30202-11 and 12). .......................................................................... 92
Modified test site layout of M-55, test section 1, test site 1

(30552-11) .......................................................................................... 98
Modified test site layout of M-55, test section 1, test site 2

(30552-12) .......................................................................................... 99
Test site layout of Marsh Road, test site 1 ......................................... 101
Test site layout of Marsh Road, test site 2 ......................................... 102
Growth of cracking on M-20, test section 1 ..................................... 111
Growth of cracking on M-55, test section 1 ..................................... 112
Growth of cracking on M-20, test section 2 ...................................... 1 13

TDC and raveling in heavily segregated areas on US-23, May
2002, 2 years after construction (rubblized pavement) ...................... 114

TDC initiation at the edge of the wheel path on M-20 (flexible

pavement, no segregation is visible) .................................................. 114
Advanced level of TDC on M-37 (rubblized pavement) ................... 115
TDC on M-42 (flexible pavement, no segregation visible) ............... 1 15

xiii

Figure 4.8
Figure 4.9
Figure 4.10
Figure 4.1 1
Figure 4.12
Figure 4.13

Figure 4.14
Figure 4.15

Figllre 4.16

Flgure 4.17

Flgllre 4-18

Figure 419
Figure 4.20

Figure 4.21

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.13

Figure 4.14
Figure 4.15

Figure 4.16

Figure 4.17

Figure 4.18

Figure 4.19

Figure 4.20

Figure 4.21

TDC in segregated areas on M-SO test section 1 (flexible
pavement) ........................................................................................... 1 16

TDC in segregated areas on M-50 test section 2 (flexible
pavement) ........................................................................................... 1 16

TDC on M-50 test section 1 (flexible pavement, no segregation
visible) ................................................................................................ l 17

TDC in segregated area on M-50 test section 2 (flexible
pavement) ........................................................................................... l 17

TDC in segregated areas at the edge of the wheel path on M-55
(flexible pavement) ............................................................................ 1 18

TDC in heavily segregated areas on Marsh Road (flexible

pavement) ........................................................................................... 1 18
Growth of cracking for three degrees of segregation ......................... 124
Calibration and verification of crack propagation model ................. 126

Longitudinal TDC at the edge of the wheel path on US-23,
control section 01052 job number 32335 BMP . ............................... 129

Longitudinal TDC at the edge of the wheel path on US-131,
control section 83032 job number 34060 BMP 13.0 EMP 18.7,
May 2002 ........................................................................................... 129

Parallel longitudinal TDC at the edge of the inner and outer

wheel paths on US-l31 (same location as Figure 5.2), control

section 83032 job number 34060 BMP 13.0 EMP 18.7, May

2003 ................................................................................................... 130

Transverse TDC associated with longitudinal TDC on M-27,
control section 16302 job number 26672 BMP 9.3 EMP 11.4,
May 2003 ........................................................................................... 130

The third stage of TDC development, LTDC and TTDC
forming a networks on East Michigan Avenue, control section

39555 job number 39212, May 2002 ................................................. 131

The three stages (A, B and C) of TDC development and
pavement cores showing the extent of cracking ................................ 131

xiv

. 11
Figure 4...

Figure 4.23

Figure 4.24
Figure 4.25

Figure 4.26
Figure 4.27
Figure 4.28
Figure 4.29
Figure 4.30
Figure 4.31
Figure 4.32
F"illre 4.33

Fleure 4.34
Figure 4.35
“Elite 4.36
Fig“? e 4.37
“We 4.33

Figure 4.39

T;

51

X\'
NO
X0

Pier

1L”

leg.

d.\h
‘4
I

Figure 4.22

Figure 4.23

Figure 4.24

Figure 4.25

Figure 4.26
Figure 4.27
Figure 4.28
Figure 4.29
Figure 4.30
Figure 4.31
Figure 4.32
Figure 4.33

Figure 4.34

Figure 4.35

Figure 4.36

Figure 4.37

Figure 4.38

Figure 4.39

Figure 4.40

Transverse TDC located at the center of the lane (compare the
size to half of a 4-in core) .................................................................. 133

Transverse TDC located at the middle of the lane ............................. 133

Longitudinal TDC and raveling in segregated areas of TDC

propagation with depth ...................................................................... 134
Variations of D1 through D7 deflections, l-75 SB, test section

1 .......................................................................................................... 153
Normalized deflections on 20273-11 ................................................. 156
Normalized deflections on US-41, test section 1 ............................... 157
Normalized deflections on M-20, test section 1 ................................ 158
Normalized deflections on 30422-11 ................................................. 159
Normalized deflections on M-50 ....................................................... 160
Normalized deflections on M-52 ....................................................... 161
Normalized deflections on M-55, Test Section 1 .............................. 162
Normalized deflections on Marsh Road ............................................ 163

Plots of equivalent pavement modulus versus the distance from
the center of load for 6 pavement sections having different
depths to stiff layer ............................................................................. 167

Three and four layer systems for flexible pavements used to
determine the approximate stiff layer depth ...................................... 170

Backcalculated roadbed soil moduli (4-layer) and the combined
moduli of the base, subbase and roadbed soil (3-layer) of the

pavement for different stiff layer depths ............................................ 174
Gradation analysis of US-41 , test section 1 ....................................... 206
Gradation analysis of M-20, test section 1 ......................................... 207
Gradation analysis of M-55, test section 1 ......................................... 208
Gradations of mixtures used in moisture susceptibility tests ............. 212

XV

Figure 4.41

Figure 4.41 Average indirect tensile strength and averaged retained
strength ratio of unconditioned and conditioned triplicates of 3
AC mixes ........................................................................................... 213

xvi

.
gx «*fi

" 4
the surtau 0: ~

.. ’1".
obsem‘d 1“ W

form an extensi
uhxh ultimate-I;
TDC are
Michigan in part
conventional fatfl
propagate Uptt an
FTHCFlng potential
factors that enhan.
imprnie the sen re

3-0 OBJECT]

 

 

 

CHAPTER 1

INTRODUCTION

1.0 BACKGROUND

Top-down cracks (TDC) are longitudinal and/or transverse cracks that initiate at
the surface of asphalt pavements and propagate downward. They have been increasingly
observed in flexible pavements throughout the state of Michigan. TDC are usually
manifested as longitudinal cracks appearing just outside the wheelpaths. Over time, they
form an extensive network of longitudinal cracks connected by short transverse cracks,
which ultimately reduce the life of the pavement.

TDC are receiving increased attention throughout the world, in general, and in
Michigan in particular. Some existing asphalt pavement design methods address
conventional fatigue cracks, which initiate at the bottom of the asphalt layer and
propagate upwards. No design method is capable of predicting or analyzing top-down
cracking potential. Since TDC adversely affect pavement performance, understanding the
factors that enhance their potential would be the first step that needs to be taken to

improve the service life of the pavements.

2.0 OBJECTIVES
The objectives of this research are:
1. Identify the causes of TDC and the factors affecting tOp-down cracking potential.
2. Compare top-down and bottom—up cracking potentials.

3. Develop a model to describe the rate of TDC propagation.

To accor

presented in see:

 

41-. g

r u

To accomplish the above stated objectives, a research plan was designed and is

presented in section 4.0 below.

3.0 HYPOTHESIS
Based on literature review and field observations of TDC, the following
hypotheses were developed:

1. TDC are longitudinal and/or transverse cracks caused by high stress ratios at the
AC surface course. Therefore, tOp-down cracking potential is affected by all the
factors influencing the magnitude of the induced tensile stress and the tensile
strength ofthe AC mix.

2. Mechanistic analysis can be used to ascertain the factors that result in high tensile
stress at the pavement surface, while laboratory investigation can be used to
determine the factors that result in low tensile strength of the mix.

To accomplish the study objectives and to verify the hypotheses, a research plan was

drawn and is presented in the next section.

4.0 RESEARCH PLAN
To verify the hypotheses and to accomplish the objectives of this study, a research
plan was developed and is shown in Figure 1.1. A detailed summary of each activity is

presented below.

I leffllg
I
I

\
Field TC.\1.\
\

\

Mechanistic Ana

1
k

 

 

 

 

Research Plan

 

 

 

 

 

 

Literature Review

 

 

 

 

 

 

 

Field observation
ofTDC

 

 

T

 

 

 

 

Hypotheses

 

 

 

 

 

Site Evaluation

 

 

 

 

 

 

Field Tests

 

 

 

 

 

 

Laboratory Tests

 

 

 

 

Distress Survey

 

 

 

 

 

 

 

 

 

 

Mechanistic Analysis

 

 

 

 

Data Analysis

 

 

 

 

 

 

 

Conclusions and
Recommendations

 

 

Crack Propagation

 

 

 

Model

 

 

 

Figure 1.1 Flowchart of the research plan

4.1 Liter:
A C075
research aetix

propagation of

4.2 Field I

Cd

CO

4.] Literature Review
A comprehensive literature review was conducted to determine the types of
research activities that were undertaken to study the factors that affect the initiation and

propagation of top-down cracking. Said review is presented in Chapter 2.

4.2 Field Investigation
The field investigation activities include:

1. Test Site Selection — In this activity, various pavements were examined to
determine if they were possible candidates for the study. From the pool of
candidates, 18 test sites were selected for evaluation.

2. Test Site Evaluation — The test site evaluation consisted of the following:

‘> Distress Survey — Included length and location of potential top-down
cracks and cores taken over cracks to determine the types of cracks. Also
cataloged other distresses to determine if top-down cracking occurs in
conjunction with other distresses. On several test sites, multiple surveys
were conducted over time to determine the rate of propagation of top-

down cracks.

‘7

Field Tests — The field tests included non-destructive deflection tests using
the MDOT falling weight deflectometer (FWD), nuclear density
measurements, core extraction, subsurface investigation (using hand

augers) and measurements of the asphalt layer temperature, as a function

of depth.

43 Laborat

Laborat.

following
1. Exarttrrtzr
thickness
AC 13} er
courses. I
moduli.
3. Conduct."

courses ir:

along

‘
.h1
gdrrL.

(JJ

Perfomriri.
resilient n:
Obsen-c M,

AC 13W t}

 

 

4.3 Laboratory Investigation

Laboratory testing consisted of sawing the cores to obtain test specimens, and
subjecting those specimens to a battery of laboratory tests. These tests include the
following:

1. Examining each pavement core and measuring its dimensions, including the
thickness of each AC course, and the depth and width of cracks, if any. The total
AC layer thickness was then calculated as the sum of the thicknesses of the AC
courses. The results were used for the backcalculation of the pavement layer
moduli.

2. Conducting specific gravity tests to determine the density of the individual asphalt
courses in the pavement. The test results were used to observe density variation
along and across the pavement.

3. Performing indirect tensile cyclic load tests (ITCLTs) to determine the laboratory
resilient moduli of the AC courses, when possible. The test results were used to
observe variations in moduli along and across the pavement as well as through the
AC layer thickness.

4. Conducting indirect tensile strength tests (ITSTs) to obtain the indirect tensile
strength of the AC courses, when possible. The results were used to calculate the
applied stress to strength ratios.

5. Determining the asphalt content of the AC courses and the aggregate gradation of
the various AC mixes using sieve analyses. The asphalt content and the recovered

aggregate were obtained by incinerating the asphalt binder.

Additiot'
example. theorc

iaboratorx' CL :11:
4.4 Data An
Two fold

using the measur

4.4.1 A

4.4.2 “‘1 L

.1.)
f-l

Additional tests were conducted in support of the above battery of tests (for

example, theoretical maximum specific gravity tests to calculate the percent air voids of

laboratory compacted specimens).

4.4 Data Analysis

Two fold analyses were conducted in this study; backcalculation of layer moduli

using the measured FWD deflection data and mechanistic analyses of flexible pavements.

4.4.1

4.4.2

Analysis of the Measured FWD Deflection Data

The measured deflection data were analyzed in two ways as follows:
Determine the variations in the deflection data along and across the
pavement at each test site. Results of the analyses were used to indicate
the degree of uniformity in the structural capacity of the pavements.
Backcalculate the pavement layer moduli using the measured pavement
deflections and cross-section data. The range of the resulting modulus

values were used in the mechanistic analysis of the pavement structures.

Mechanistic Analysis of Flexible Pavements

Mechanistic analyses of flexible pavements were conducted to assess:
The temperature-induced tensile stresses in the AC layer.

The sensitivity of the load-induced tensile stresses in the AC layer to the

factors affecting top-down and bottom-up cracking potentials. The results

r,’

4.5 Vandal
chuliS
I. TDC 35“
and DJ";
__ For the 5
obtained
contpdft‘L

(CHEVR‘

DJ

The IQCdi.
compared
4.
6 C onclusio
Based on t'
uncerstanding oh’

still be made.

THESIS I-

 

This the

sis

Chapter I ‘ Intror

 

4.5

4.6

are used to compare the crack initiation potentials at the top and at the
bottom of the AC layer.
3. The influence of various factors affecting the tensile strength of the AC
mixes.
Validation of the Results of the Analyses

Results of the analyses were validated as follow:

. The accuracy of the MICHBACK computer program was checked using forward

and backward mechanistic analyses.

For the same pavement sections and load conditions, results of the analyses
obtained from finite element programs (ABAQUS and MICHAPVE) were
compared to those obtained from a well accepted closed form solution program

(CHEVRONX).

. The locations of the maximum tensile stress obtained from the analyses were

compared to locations of TDC observed in the field.

Conclusions and Recommendations

Based on the results of the field and lab investigations, analyses, and the

understanding obtained from the literature, the proper conclusions and recommendations

will be made.

5.0

THESIS LAYOUT

This thesis is composed of five chapters:

Chapter 1 - Introduction

Chap“
Chapti’
Chapter

Chapter

Chapter 2 - Literature Review
Chapter 3 — Field and Laboratory Investigation
Chapter 4 — Data Analyses

Chapter 5 — Conclusions and Recommendations

1.0 INTRO

Connde
patenients is at
parunentcross-
cmnenuonalbr'
pat ements 11.1\ e
Numerous Sludl.
lupusincluded:
load. and the en r.

This re\ i,

l- The C3115;

(
34
(D
E"
r.
:3

 

CHAPTER 2

LITERATURE REVIEW

1.0 INTRODUCTION
Considerable literature regarding the design and performance of asphalt
pavements is available, most of which has been directed toward determining the
pavement cross-section (design) and predicting pavement roughness, rutting and the
conventional bottom-up fatigue cracking. Recently, top-down cracks (TDC) in flexible
pavements have been widely reported by various researchers and highway agencies.
Numerous studies were conducted to determine the causes of TDC. The researched
topics included the effects of tire-pavement interaction, pavement cross-section, axle
load, and the environment on top-down cracking potential.
This review of literature covers the following topics:
1. The causes and propagation of TDC in flexible pavements.
2. The factors affecting tensile stresses induced in the AC layer.

3. The factors affecting tensile strength of the AC mixtures.

2.0 TOP-DOWN CRACKS IN FLEXIBLE PAVEMENTS

Top-down cracks (TDC) are longitudinal and/or transverse cracks that initiate at
the pavement surface and propagate downward and outward as shown in Figure 2.1.
Based on examination of pavement cores, various researchers have reported that TDC
may propagate only within the AC surface course or throughout the entire depth of the

AC layer (Myers et al, 1998, Gerristsen et al, 1987, Uhlmeyer et al, 2000, Niederquell et

 

8e a.

   

N 25%
a w...

10

a1. 20th). and S
longitudinal TI
Gerristsen et a'
Stasdisant er a:
divided the dex
consists ofa sin
Pith. Ox er time.
and new sister c
the third stage “
TDC.

TDC hex
pavements in \ .1:

Netherlands (3141

193

LII

.Duzaut

III

at
are increasinglx r
trit'esti gated or b

Washington (Me
Infilmeyer. zljr ,u’i ,

Pat‘fments while

41? Sections belov

CA L'SES

SFV‘CTal Ca.

al, 2000, and Svasdisant et al, 2002). Field investigation has shown that most
longitudinal TDC are located in the vicinity of the wheel paths (Myers et al, 1998,
Gerristsen et al, 1987, Matsuno and Nishizawa, 1992, Niderquell et al, 2000, and
Svasdisant et al, 2002). Also based on field observations, Svasdisant et al (2002) have
divided the development of TDC into three stages as shown in Figure 2.2. The first stage
consists of a single short longitudinal crack or cracks appearing just outside the wheel
path. Over time, the second stage develops where the short longitudinal TDC grow longer
and new sister cracks develop parallel to the original crack. Finally, the TDC evolve into
the third stage where the parallel longitudinal TDC are connected via short transverse
TDC.

TDC have been investigated and reported as a prominent distress in flexible
pavements in various countries including Japan, Kenya, South Africa, France, and the
Netherlands (Matsuno and Nishizawa, 1992, Wambura et al., 1999, Hugo and Kennedy,
1985, Duzauts and Rampal, 1987, and Gerritsen et al, 1987). In the United States, TDC
are increasingly reported as a major distress in flexible pavements and either are being
investigated or have been investigated in several states including Florida, Michigan and
Washington (Meyers, 1998, 1999, 2001, Svasdisant, 2001, Neiderquell, 2000, and
Uhlmeyer, 2000). Most of the investigations are related to the causes of TDC in flexible
pavements while few address the propagation of TDC. These two topics are reviewed in

the sections below.

3.0 CAUSES OF TDC

Several causes of TDC have been studied and reported in the literature. These

11

ea 2 -m use STE -w .Umom 552 -m .om-2 -N do; A Boo .UQF
wfiaosm 880 28 Am owfim H O .N owfim n m . _. owfim H «J mac—030 9537qu we EoEmoEBU 05 we??? mcafiwsosm Nd oSwE
I. .. .. .. H . .4. 5.. . . .

. .31.».

 

12

mind-’3

1. L033

2. Th?”
3 Afill‘.

4. Sis?“
Vanousrcsc‘
tensile stre:s i
found in tile F
the vicinity of
TDC.

.‘tloicna
effects of tire-p
TDC. He conclt
cause ofTDC. F
temperatures; on
Similarly. in the f
to 31131}2€ the ten

contact stresses '

 

LITTL:
CPacking potenti
roenen 1
PFC» r

 

 

include:

1. Load-induced stresses and strains.

2. Thermal stresses.

3. Aging of the AC binder.

4. Segregation in the AC mix.
Various researchers including Myers et a1 (1998) concluded that TDC are caused by
tensile stress rather than shear stresses. Since the majority of the longitudinal TDC are
found in the vicinity of the wheel paths, the load-induced tensile stresses and strains in
the vicinity of the tire-pavement contact areas are the most commonly studied causes of
TDC.

Molenaar (1984) used the CIRCLY multi-layer computer program to analyze the
effects of tire-pavement vertical and lateral contact stresses (inward shear stresses) on
TDC. He concluded that high surface tensile strains induced at the edge of the tire are the
cause of TDC. Further, his results indicated that thicker and softer AC layer (at high
temperatures) on top of a stiffer base layer increases top-down cracking potential.
Similarly, in the Netherlands, Gerritsen et a1 (1987) used the CIRCLY computer program
to analyze the tensile stresses induced at the pavement surface due to vertical and lateral
contact stresses. They have found that the load-induced tensile strains at the edge of the
tire in pavements with bound bases (stiff base) are high enough to initiate TDC. They
added that thermal stresses and aging of the AC binder could also enhance top-down
cracking potential.

Groenendijk (1998) used the 3-Dimensional (3-D) finite element computer

program, CAPA, to analyze the load-induced surface tensile stresses at the pavement

13

surface. He er
vertical. latera
inflation press
contact stress .
at the surface r
stresses used it

Vehicle-Rom .

0111116 mdUCCd

613111993. in»,

h-‘
-

Lateral .
This Cor:

I. I 996 g {k

I“)

Later-d] C
induCe h].
TDC 31:.
were mCd
first Cone;
from the 1‘
et a} ”991‘

Mb'erg Cl

surface. He employed the three components of the tire-pavement contact stresses;
vertical, lateral, and longitudinal contact stresses, to simulate different loads and different
inflation pressures. He concluded that the combined influence of the non-uniform tensile
contact stress and the aging of the AC at the surface could result in critical tensile stress
at the surface rather than the bottom of the AC. Note that, the values of the contact
stresses used in Groenendijk’s analyses were measured in South Africa using the
Vehicle-Road Surface Pressure Transducer Array (VRSPTA). More details of the South
African measurements are presented in the tire-pavement interaction section.

The effects of the vertical and lateral contact stresses of radial and bias ply tires
on the induced tensile stresses at the tire-pavement contact areas were studied by Myers
et a1 (1998, 1999 and 2001 ). They concluded that:

1. Lateral contact stresses (inward shear stresses) induce high surface tensile stresses

in the pavement at the edges of bias ply tires and cause TDC, Myers et a1 (1998).

This conclusion was based on shear stresses data that were measured by Jacobs’

(1996) for bias ply tires.

2. Lateral contact stresses (outward shear stresses) at the outer ribs of radial tires
induce high surface tensile stresses under the outer ribs, which are the causes of

TDC, Myers et a1 (1998). This conclusion was based on shear stresses data that

were measured by Pottinger (1992) for radial tires. The difference in this and the

first conclusion regarding the direction of the shear stress and its locations stems
from the fact that Myers et al used two different sets of data. According to Myers
et a1 (1999), the later conclusion is more accurate. Note that the data presented in

Myers et al (1999) indicate that the maximum surface tensile stress occurs under

14

relattx

than th

the center of the outer rib.

The tensile stresses under the tire ribs are localized phenomenon at the AC
surface and they dissipate rapidly throughout the depth of the AC layer. Hence,
the pavement structural characteristics such as moduli of the AC and base layers
have insignificant effects on the magnitude of the induced surface tensile stresses,
Myer et a1 (1998) and (1999). This also supports the conclusion made by Matsuno
and Nishizawa (1992) (as stated in the next paragraph).

Radial and wide-base radial truck tires induce higher surface tensile stresses
relative to bias ply tires. As a result, they are more detrimental to the pavement
than the bias ply tires, Myers et a1 (1999).

Load-induced tensile stresses away from the edges of the tire are the driving
mechanism for the propagation of TDC. The pavement layer stiffnesses and the
location of the tire relative to TDC have significant impact on the magnitude of
the load-induced tensile stresses away from the tire, Myers et al (2001).

In Japan, Matsuno and Nishizawa (1992) reported that longitudinal TDC were

found mostly along the wheel paths and they are absent in shadowy areas (such as under

bridges). They conducted finite element analysis of flexible pavements at high

temperatures and made the following two conclusions:

1.

The load-induced tensile strains at the edge of the tire cause TDC initiation. This
conclusion supports that of Molenaar ( 1984) and Gerritsen et al (1987).

The pavement cross section had little effect on the surface tensile strains, which
agrees with the conclusion made by Myers et a1 (1998), Myers et a1 (1999) and

contradicts that of Uhlmeyer et a1 (2000), which is stated below.

15

Us
\3'351’111127'10
around the
6.3-inch t It
initiation OF
31119991, an

A diz‘:
tZtt‘ujli. The}:

in 11k“ PJVCHEC‘.

(a)

U
8
r?
is;

 

Uhlmeyer et a1 (2000) studied of the causes of TDC in conjunction with the

Washington State Department of Transportation. They observed that TDC occur in and

around the wheel paths in pavements 3 to 8 years old with AC thickness of more than

6.3-inch (16-cm). Hence, they concluded that pavement thickness has an effect on the

initiation of TDC, which contrasts the conclusions made by Myers et a1 (1998), Myers et

a1 (1999), and Matsuno and Nishizawa (1992).

A different mechanism of top-down cracking was reported by Bensalem et a1

(2000). They used a finite element model to analyze the load induced stresses and strains

in the pavement. They reported that:

1.

The load-induced shear strains at the edge of the wheels in the vertical plane are
higher than the load-induced lateral tensile strains at the same location on the
pavement surface. Therefore, the shear strains on the vertical plane play an
important role in TDC initiation and propagation.

The magnitude of the shear strains on a vertical plane at the edge of the wheel
path decreases rapidly with depth. This explains some of the field observations
that TDC propagate through the upper AC courses.

Decreasing the thickness of the AC layer results in increasing the load-induced
shear strains. Hence, thinner AC layers are more vulnerable to TDC. This is in
contrast with the conclusions of Uhlmeyer et al (2000) stated above.
Differential stiffness in the AC layer due to temperature gradient and thermal
tensile stresses may have important roles in the initiation and propagation of
TDC.

Note that, the first three conclusions represent a new view of the causes of TDC.

l6

The comn
stresses Of

tress and

(II

199,”). and
later (1111.“
field rm est
first too co:

vertical p.31:

numerous C0

Section on the
cramming the

hate made the j

 

 

The commonly agreed on causes of TDC initiation and depth are the load-induced tensile
stresses or tensile strains at the edge of the wheel paths on the pavement surface, thermal
stress and aging of the asphalt binder (Myers et al, 1999, Molenaar, 1984, De Beer et al,
1997), and that TDC were more likely to initiate in pavements with relatively thick AC
layer (Uhlmyer, 2000, and Molenaar 1984). In addition, field data collected during the
field investigation of this thesis and other studies (Myers et al, 1998) do not support the
first two conclusions of Bensalem et a1 (2000). Cracks caused by shear failure along a
vertical plane should be inclined at an angle from the vertical plane. The cracks found in
numerous cores extracted from 18 different test sites in this study showed vertical crack
alignment.

The agreements and contradictions regarding the effect of the pavement cross-
section on the induced tensile stress at the pavement surface were further scrutinized by
examining the types of analyses conducted by various researchers. Eventually, all sides
have made the proper conclusions relative to their input data and results. To clarify, the
tensile stress induced at the pavement surface due to the inward/outward shear is
independent of the pavement cross-section. However, the vertical load-induced tensile
stress at the pavement surface is a function of the pavement cross-section. The above
implies that when only the effects of the vertical contact stress are analyzed, the results
would show a significant role of the pavement cross-section. On the other hand, analyses
of the inward and outward shear stresses would result in an opposite conclusion.

Thermal stresses and strains were also investigated as possible causes of TDC.
Dauzats and Rampal (1987) reported that most of the TDC found in their test sections in

southern France were located on the centerline side of the slow lane. They calculated

17

lilfirmfl ST
Shlhin ( }t
Impi’mtttr
Similafi}; l
TDC injm:
THE-V 5031c};
TDC. FSPCCi
more {Times
In SOL
possible Cause
CRACK 50m:

Cha‘FACleristic

'1)

me 3491-92 01111;.

ornon-load 35s.,

Wolds in the rr
set ere aging. “It.

altl95’81rc1ated

 

surrace creek in 2 _

willerable to ‘

a's’tr,

 

thermal stresses, strains, and tensile strength of the AC using equations developed by
Shahin (1977). Results of their study indicated that the surface tensile strains due to low
temperatures could initiate TDC after a number of repeated temperature cycles.
Similarly, Roque and Ruth (1990) used MEAPs/CRACK3 computer program to study
TDC initiation potential in flexible pavements due to thermal and load-induced stresses.
They concluded that the combination of thermal and load-induced stresses is the cause of
TDC, especially under critical conditions such as fast cooling rate or spring thaw with
more emphasis placed on the former condition than the latter one.

In South Africa, Hugo and Kennedy (1985) also investigated thermal stresses as a
possible cause of TDC. They calculated the thermal stresses in the AC layer using the
CRACK computer program. They also studied the impact of several AC mix
characteristics including aggregate gradation, air voids, and asphalt cement contents on
the aging of the AC mix. They concluded that TDC could initiate due to load-associated
or non-load associated causes (thermal stress and/or aging) or combination thereof.

In Kenya, Wambura et a1 (1999) tested cores obtained from pavements exhibiting
TDC. They concluded that the recent AC mix design procedure in Kenya allowed high
air voids in the mix at the time of construction. As a result, the AC binder experienced
severe aging, which caused the initiation of TDC at the AC surface. Similarly, Malan et
al (1988) related aging of the AC binder to TDC initiation. They studied premature
surface cracking in South Africa and concluded that gap-graded AC mixes are more
vulnerable to aging and TDC than dense graded mixes. They also reported that viscosity
of the AC binder has a significant influence on TDC potential. They stated that TDC

potential decreases when high viscosity binders are used for low traffic loads and low

18

US$15“) him
To rh;

I'Dt‘lrdirig II] C

ntlfd causes C
j, H i gh t:
such 415

__ Lo,“ ICE
pmpert;
such as -
Some re

mphasized the

this thesis at .\li.

caused by high 5

surface. Hence, 1

atiect the tensile

addresses the pot

 

the AC layer. H e

bo30m and at the

1"1 '
a arable literatur

MAC layer are .

 

viscosity binders are used for high traffic loads.

To this end, it becomes obvious that there is no consensus in the literature
regarding the causes of TDC in flexible pavements. Several factors that have been
hypothesized as the causes of TDC are summarized in Table 2.1. As can be seen, the
noted causes of TDC can be summarized into two categories as follows:

1. High tensile stresses and strains induced by load, temperature, and other factors
such as construction.

2. Low tensile strength or fatigue resistance of the AC materials due to the AC mix
properties, aging or hardening of the AC binder and poor construction practice
such as compaction.

Some researchers only emphasized the effects of one category while others
emphasized the combined effects of some factors from both categories. In the study for
this thesis at Michigan State University, it was hypothesized that TDC are fatigue cracks
caused by high stress ratios (high tensile stresses and/or low tensile strength) at the AC
surface. Hence, to determine the causes of TDC, it is necessary to address the factors that
affect the tensile stress and the tensile strength at the AC surface. In addition, this study
addresses the potentials of fatigue crack initiation at the top, as well as, at the bottom of
the AC layer. Hence the induced tensile stresses and the available tensile strengths at the
bottom and at the top of the AC layer are considered. Therefore, in the next sections,
available literature addressing the factors affecting tensile stresses and tensile strengths in

the AC layer are reviewed.

19

 

. z __ 1172...; ..7...£,_._:x..:1:. :3
.. X. _. 91::

 

  

AP A.~n.v~ ~
hun~—w Az-—.f.~—wh.< \:-w
~pJ —~.J7.~.~.~.~.J.- V «7%

92:23.. . v< 32?. A,
.129» :_U.:D>...L-U...: 2:. A.
3;?) 9:..‘A. UnawarJ Uzs :w -.J-....:v~ Us: . VA -.

..S...\..£v.:.(
«A x? L C 1..
3~ v .2.....:.4:;.<
i . Amtrazxrw» -w Tu ::...r...;v.....-...)7.

2.4.: .0;ka V~
bk: >i.:2...::7. < u .N 3.53.?

a 7.3.1.2921. .3322... 229:: 7:... 7..
3.2.4? >..:: 1:: .23; .9
23.22.9239

$32.“? 3.4.2: Z. L:—

z-_:..3.~..~zU>—.. 233..-..3’ X». 07.3: 223.1272...

 

._:::80: DO: :0 00:0:::: :30:::w6 : 6:: 3:8: U< 0:: :0 b60065 A

 

 

 

38:8 :03:w 06:0: 3:: OCH :3 wEw: 0: 8:80:15 0:08 0:: 608:: U< :0:::w-::0 A Ammo: _: :0 332
3a: a :0 5::
008:3 U< 0:: :: OOH :0 33:8 0:: 60630 wEw: 80>0m :3 3::: _: :0 98:33
8038:800 :0 38:8 60630 :88003: 38:0: :0 -38 0: 0:: 03:8: OOH @333 @8808 :3 0w::
888300 3::: $63 30:60 0:: :3 0::: w::000 6::
w:::: 3:800:60 OCH :0 0630 0:: 8 60386 :00:::_-::8 :3 _:8:0:: :0 :0::8:800 0: H 83: :8: :3 0:30:

 

.6893 08:80:80:
:0::0:0: :0 :0:8:: : :03 DOB 03:8: 8:00 68:33:80: 38 80:: 68:3 86:0: 00::::m

$6: :83: 3 6:3:

 

80:0: U< 0:: :: :0 3068 008:3 Ox: 0:: :: 30:6 86:8 30:60 8 :38 8:00 008:3 0::
:: U< 0:: :0 wEw: 0:: :3 30:6 :0:::00 86:0: 8:88:80: 0:: :0 00:08.:8 38:800 0: h

3%: 8:82:80

 

On: :0 :0::w::0:: 0:: :0:

883:008 w8>:: 0:: 0:: 8: 8: :0 60w:0 0:: 80: 33: 60303 86:0: 3088-33 A
.68: b: m8:

0: 02:88 60303 86:0: 008:3 3:8: 00::8 60:: :08: :83: 03:08? :3 :83: A
60303 86:8 008:3 :00::8 8: :0 0:::838 8: :0 6:00.:0 :30:::w68

0>:: 6:03: 0::: :3 U< 0:: :0 :::08 m: :03 603680880 88036 ::080>:: 0: H A
OOH 0630 :3 60:: :83: :0 60w:0 0:: :: :6 8: 6:: 0::

:0::: 60303 86:0: 008:3 :8: 00::8 60303 303 @338: 60386 :0:::00 88:: A

:8:
:3 38: .38: a :0 8::

 

x890: :0:_-m.: 3:: 0:08
:0 3828:: 9:. :83 80 8:0» w 0: m 680803: 8 6:3: 30:3 0:: :::0:: :3 8 ::000 OOH

88:: a :0 5:285

 

 

 

Amoo::

68:3 86:0: 008:3 0:: :0 80:0 0::: 3: 3:00: 30:0 80803: 0: H 8:38:82 :3 0:332
38: «5:32

On: 60:30 06:: :3 :0 :0: :0 008:3 U< :0m A :3 0:332 :3 3:03 _:

OOH 0630 60303 :0:::00 3:88 :3 ::0::0> ::080>::-8: 0::l A :0 38800 Sumo: 3:802

6036 08:: 8 :80>0: 3:: :3 :3: 60:3 0:: :0 0w:0 0:: :: :0::08 0:: OOH Amoomv _: :0 :38:m:>m

 

 

:06:_0:0U

 

608880:

 

 

8038608: 6:03> :3 0::8 8068030 .:0 5:883 < :.N 83:;

20

,4

4.1

mesh:
A451 Y

mechmif

pax'em ems

imp.‘/'caz{on
Uzztknessr a1
fayer thicknc:
compresm ‘c 5:
fatigue/172° and

me other pa‘

v.0 rangue life. 171!

4..—

4.0 FACTORS AFFECTING THE TENSILE STRESS IN THE AC LAYER
In this section, the following factors affecting the load-induced tensile stress in
the AC layer are reviewed:
1. Thickness and modulus of each AC course and each pavement layer and the
differential stiffness between the AC courses and the pavement layers.
2. Wheel load and configuration (single and dual).
3. Tire-Pavement Interaction

4. Temperature.

4.1 Thickness and Modulus of Each Pavement Layer

Flexible pavement designs are accomplished by using either empirical or
mechanistic-empirical approaches or both. Some empirical procedures (such as the
AASHTO 1993) are based on the ride quality (serviceability index) whereas most
mechanistic-empirical procedures are based on limited rut depth and pre-determined
fatigue life. The common output of all pavement design methods is that thicker
pavements are required for higher traffic levels or longer performance period. The
implication of this is that the pavement layer thicknesses (especially the AC layer
thickness) affect the ride quality, rutting and fatigue cracking. For example, higher AC
layer thicknesses reduce the tensile stress/strain at the bottom of the AC layer and the
compressive stress/strain at the top of the subgrade thereby causing increases in the
fatigue life and decreases in rut potential. Similarly, the moduli and thicknesses of the AC
and the other pavement layers have significant impact on pavement performance relative

to fatigue life, rut depth and ride quality.

21

I J

U)

lajvers in;

the eti‘ect:

In reality, pavement performance is affected by numerous factors that can be

divided into various categories as follows:

1. Pavement design factors such as layer thicknesses and their properties.

2. Construction factors.

3. Environmental factors.

4. Traffic factors, which are typically accounted for in the pavement design phase.
While the last three factors are addressed elsewhere in this chapter, the pavement layer
thicknesses and moduli are addressed below.

The effects of the AC layer thickness on the load-induced stresses vary and
depend on the type of the contact stresses. For example, the tire-induced transverse and
longitudinal shear stresses are independent of the thicknesses and moduli of the pavement
layers including the AC layer (Myers et al, 1999 and Molenaar, 1984). On the other hand,
the effects of the vertical contact stress on the induced tensile, shear and compressive
stresses in the pavement layers are a function of the thicknesses and moduli of the
pavement layers and the modulus ratios between the AC courses and between the
pavement layers (Myers et al, 1999).

Svasdisant et al (2002) used 3—D finite element analyses to study the induced
tensile stress at the pavement surface due to vertical contact stress. He concluded that
high modulus ratios between the AC surface and the AC base courses (high differential
stiffness in the AC layer) result in higher induced tensile stress at the pavement surface.
Such differential stiffness could be the product of the asphalt mix design practice, aging,
temperature gradient or a combination thereof. The idea of higher differential stiffness or

modulus ratio was mentioned several times by other researchers; some mentioned it as

22

the different
mentioned it
(Molenaar. l
modulus at ti
results in 11:51
gradients ace:
analyses tusin
base and AC. .

tensile strain f:

4.2 Wheel
In COUX I
assumed to be \
tirepaxement C-
and ll'itczak. 1‘:
hate been publi.

the shed 103d a

 

 

the differential stiffness within the AC layer (Roque et al, 2002) and some others
mentioned it as the differential stiffness between the AC layer and the aggregate base
(Molenaar, 1984). Roque et a1 (2002) concluded that differential stiffness (higher
modulus at the pavement surface) in the AC layer due to temperature gradient and aging
results in higher tensile stresses at the crack tip. They concluded that temperature
gradients accelerate the TDC propagation rate. Molenaar (1984) has shown in his
analyses (using the CIRCLY computer program) that high modulus ratio between the
base and AC, or stiff base layer, could result in the shift of the maximum load-induced

tensile strain from the bottom of the AC layer to the pavement surface.

4.2 Wheel Load and Configuration

In conventional flexible pavement designs and analyses, the wheel load is
assumed to be vertical and uniformly distributed over a circular area and the resulting
tire-pavement contact stress is assumed to be equal to the tire inflation pressure (Yoder
and Witczak, 1975 and Huang, 1993). Based on the two assumptions, numerous articles
have been published addressing the sensitivity of the pavement responses to variations in
the wheel load and the tire inflation pressure.

Researchers such as Bensalem et al (2000) and Huang (1993) addressed the
effects of single and dual wheel loads on pavement responses. Their results, which
agreed with numerous other researchers, indicate that a single wheel load induces higher
stresses and strains at the surface and at the bottom of the AC layer relative to dual wheel
load. Other researchers studied the effects of non-uniform tire-pavement contact stresses

and the shape of the contact area on pavement responses. Still others incorporated tire-

23

paxement int.
flexible pax e7
Myers et al ( I
this chapter. L
that. for com -.
vertical conta.
than 4- im. Tl:
\ertieal cont.“

Signiticantly c.

4-3 Tire~PL
AlmOgr

.w

DJ
2".
O
f

G.

—

 

pavement interaction (vertical, transverse and longitudinal stresses) into the analyses of
flexible pavements, De Beer et al (1997) Molenaar (1984), Groenendijk (1998), and
Myers et al (1998) and (1999). Their general conclusions are presented in section 4.3 of
this chapter. One significant conclusion was that made by DeBeer et al (1997) who stated
that, for conventional flexible pavement design and analyses, the uniformly distributed
vertical contact stress assumption is adequate when the AC layer is relatively thick (more
than 4- in). This conclusion is based on the observation that the effects of non-uniform
vertical contact stress on the tensile stress at the bottom of a thick AC layer are not

significantly different than that of uniform vertical contact stresses.

4.3 Tire-Pavement Interaction
Almost all vehicles use rubber-based pneumatic tires. The principal functions of a
rubber tire include (Clark, 1981):
1. Transferring the load from the vehicle to the pavement.
2. Providing wear resistance by being tough and minimizing cuts, tears and cracks
by being resilient.
3. Providing good friction between the tire and the pavement surface.
4. Resisting the bruising impacts due to defects on the roadways.

The truck fleets in the United States and other countries use a wide variety of
pneumatic tires. Unless otherwise specified, in the rest of this thesis the technical term
“pneumatic tire” is referred to as “tire”.

Previous research has focused on the effects of tire imprint on the stress/strain in

the pavement. These analyses, which were summarized in section 4.2, concentrate on the

24

with the pat
t'a'iations it:
in general. I\
the truck flee

Aecor
radial and big:

l. The ttr

3- The xx ,1

 

 

stress and strain at the bottom of the AC layer, or in several cases, the stress and strain at
the surface of the pavement due to various tire contact areas. Directly under the tire
however, significant stress/strain may be induced by the interaction of the tire structure
with the pavement. The tire structure varies between different types of tires and includes
variations in size, inflation pressure, structure, tire-treads and grooves, and tire stiffness.
In general, two tire types are heavily used; radial and bias ply with a high percentage of
the truck fleet using radial tires.

According to Myers et al (1999), the differences in the structures between the
radial and bias ply tires include:

1. The tire-tread structure of radial tires is more rigid than the bias ply tires.

2. The wall structure of the bias ply tires is more rigid than the radial tires.

The factors affecting tire-pavement contact area and contact stresses, and their
effects on the pavement responses (stresses, strains and deflections) are presented in the
next subsections. Note that the term “tire-pavement interaction” refers to the following:

1. The shape and size of the tire-pavement contact area.

2. The magnitude and distribution of the tire-pavement contact stresses including the
vertical contact stress, the lateral or transverse contact stress (perpendicular to the
tire traveling direction), and the longitudinal contact stress (parallel to the tire
traveling direction).

Unless otherwise specified, in the rest of this thesis, the term “contact stresses” is

used to reference all three contact stresses between the tire and the pavement.

25

\ 1

f3.

int

Ex

I)

l")

4.3.1 Tire-Pavement Contact Area

The shape and size of the tire-pavement contact area is affected by various
factors including tire type (e. g., radial versus bias ply), tire inflation pressure, axle
load, carcass stiffness, the number and dimensions of the ribs and grooves,
vehicle speed, steering, and the pavement micro and macro textures (Clark, 1981).
De Beer and Fisher, (1997) obtained tire-prints of bias, radial and wide-base
radial tires at different loads and inflation pressures. Figure 2.3 shows the tire
imprints of the bias, radial and wide based tires measured by De Beer and Fisher .
Examination of their tire prints indicates that:

1. For radial tires, the shape of the contact area can be simulated by a
rectangle with almost a constant width and a variable length. The constant
width is equal to the tread width whereas the length increases with
increasing wheel load and decreasing inflation pressure.

2. For bias ply tires, the shape of the contact area is rectangle with rounded
ends. The width and the length are variables and function of the wheel
load and inflation pressure.

3. The tire print area of both tire types can be divided into two zones; the
actual contact area under the ribs and the non-contact area under the
grooves.

Groenendijk (1998) made similar observations regarding the shape of the

contact area of radial tires.

26

9:: E a 28m 23 x82
.555 US... Bum may 233E coamcfi $92: was vac. 30:3 QE-N6 wEBfi 2: 3a was a was 363 a mo 35E 8:. ON Eswi

2: .2935 a mo EE 2? S 8: 392 a mo Eta 8C. 3

 
 

27

 

‘
.3.

 

4.3.2 Tire-Pavement Contact Stresses

Numerous researchers have studied the tire-pavement contact stresses
using several approaches including:

1. Theoretical approach based on tire mechanics (Clark, 1981, Teilking and

Roberts, 1987, and Schapery and Teilking, 1977)

2. Finite element analyses (Schapery and Teilking, 1977, and Roque et al,

2000)

3. Measurement of the contact stresses (Markwick and Starks, 1940, De beer

et al, 1997, Himeno et al, 1997, and Teilking and Abraham, 1994)

Results of their studies indicated that the tire-pavement contact stresses are
affected by various parameters including:

1. Tire type and structure
2. Wheel load and inflation pressure
3. Tire motion (stationary or rolling)
The effects of the above factors are addressed below.

As stated earlier, one of the functions of the pneumatic tire is to transfer
load from the vehicle to the pavement. The load can transfer to the tire-pavement
interface through the tire wall at the edge of the tire and through the pressurized
air over the tire tread as shown in Figure 2.4. Since a bias ply tire has a more
rigid wall than a radial tire, a higher portion of the load is transferred through its
wall than through the wall of a radial tire. Such difference in the load transfer

mechanisms affects the distribution of the tire-pavement contact stresses.

28

 

 

Load transferred Load transferred
through wall structure through wall structure

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Tire wall
Tire inflation pressure (IP)
Tire grooves
‘ at"? :- Tire tread
. . w {IT—1 A - l ' m'fi‘
Tire ribs .. """"

 

 

  

I’ll}! H II T
Pavement surface

[Vertical contact stress (VCS)

 

 

 

 

VCS ~ (IP)*(Total tire imprint area/Total contact area of the ribs)

Figure 2.4 Load transfer mechanisms of a tire

29

out“ .:
pheno
shou it
rigid tr :
bias pl}.
rt‘ported
bEI“ Gen
Stresses
the effet"

"‘

ettect ter

 

 

Myers et al (1999) addressed the effect of the structures of bias ply and
radial tires on the tire-pavement contact stresses. They stated that the relatively
rigid wall and flexible tread structures of the bias ply tires allow the tire ribs to
pull in toward the center of the tire inducing inward shear stress at the pavement
surface. The tire industry refers to this phenomenon as the pneumatic effect of the
tire. The pneumatic effect of radial tires is much less than that of bias ply tires
because of the relatively rigid tread structures and the flexible wall of the radial
tire. Such horizontal rib expansion away from the radial tire center induces an
outward shear stress at the pavement surface. Myers et a1 (1999) referred to this
phenomenon as the “Poisson effect” of the tire tread structure. A schematic
showing the stresses under a truck tire are shown in Figure 2.5. Due to the more
rigid tread structure, the Poisson effect is more prevalent in radial tires than in
bias ply tires. This agrees with the findings by Markwick and Starks (1940). They
reported that the rib-hardness has significant impact on the friction properties
between the tire and the pavement and hence on the inward and outward shear
stresses. Harder ribs produce higher contact stresses. It is also important to note
the effect of these behaviors on pavement stresses and strains. The pneumatic
effect tends to cause compressive radial stresses at the pavement directly
underneath the tire. Radial tensile stresses are created at the edge of the tire by the
tendency of the pavement to be pulled towards the center of the tire. In contrast,
the Poisson’s effect has the tendency to pull the pavement apart under the ribs of
the radial tire. Therefore, areas of high tensile stress are generated under the

imprint of the tire.

3O

 

menu: on: x25 3 @322: 832% he mafia Wm ouswE

3t 05 mo 5:50 2: 80¢ 33¢;

30% 29,350 8555C.
A23 05 mo SE8 2: EmBoC GE 05 mo “8:8 2: 3950.5

80% @835 3555:. 80% @335 35555.

 

3:3.ch cctxcf EC.

mat of. 85on BE.

31

mod -.

The‘.

I.)

'JJ

Prossure
Vertical c

“ heel ki‘

 

hlé’llcr u .

A

prebsurc;

ab0U11.s.

 

In the Netherlands, Groenendijk (1998) investigated the contact stresses of
new, used and trimmed (smooth tire/no groove due to tire wear) radial tires. His
findings relating to new radial tires are similar to those of Myers et a1 (1999).
Relative to trimmed radial tire, Groenendijk reported that the maximum induced
shear stress is an inward shear located at the tire edge.

Roque et al (2000) used the ABAQUS finite element computer program to
model the structure of a radial tire and analyze the tire-pavement contact stresses.
They reported that:

1. The calculated contact stresses were similar to those that were measured
earlier using actual radial tire and sensors.

2. The magnitudes of the transverse contact shear stress are highest at the
outer ribs near the edge of the tire.

3. The transverse contact shear stress is not affected by the thickness of the

AC layer and the modulus ratio of the AC and aggregate base layers.

Many researchers have investigated the effects of wheel load and inflation
pressure on the contact stresses. Tielking and Abraham (1994) measured the
vertical contact stresses of radial and wide-base radial truck tires with different
wheel loads and inflation pressures. Their measurements indicated that both
higher wheel load and higher inflation pressure result in overall higher contact
pressure but minimal increase in the vertical contact stress at the center of the tire.

Markwick and Starks (1940) measured the contact stresses of stationary
and rolling tires. They concluded that the maximum vertical contact stresses are

about 1.5 times higher than the inflation pressure. They explained that the tire-

32

s
C:
S
‘
radial

”Kilt“;

carcass stiffness and the reduced contact area due to the tire grooves are the
causes of the increase in the vertical contact stresses relative to the inflation
pressure.

Similar results regarding the location of the maximum vertical contact
stresses were found by De Beer et al (1997). They measured tire-pavement
contact stresses due to slow rolling smooth tires (no grooves) using Vehicle-Road
Surface Pressure Transducer Array (VRSPTA). Their results indicated that:

l. The vertical contact stresses were neither equal to the tire inflation
pressure nor uniformly distributed throughout the contact area.

2. The maximum vertical contact stresses were found to be a function of both
the tire inflation pressure and the wheel load. The tire inflation pressure
controls the vertical contact stress at the tire center whereas the wheel load
control the contact stress at the tire edges.

3. Overloaded and/or under inflated tire cause high vertical contact stress at
the tire edge, which may exceed the inflation pressure by a factor of three.

4. The direction of the measured transverse (lateral) contact stress is toward
the tire center; inward shear stress

5. The longitudinal (parallel to the wheel traveling direction) contact shear
stress is the lowest among the three contact stresses.

De Beer and Fisher (1997) also measured the contact stresses of bias and
radial truck tires with various wheel loads and inflation pressures. Their results

indicated that:

33

1.

The average vertical contact stress is higher than the tire inflation pressure
by about 20 percent.

The vertical contact stress is higher than the transverse and the
longitudinal shear stresses as shown in Figures 2.6 through 2.8. These
findings regarding the magnitude of the contact stresses agreed very well
with the measurements made by De Beer et al (1997).

The maximum contact stresses (vertical, lateral and longitudinal) of both
bias ply and radial tires are highly related to the wheel load. They showed
no relationship with the tire inflation pressure as shown in Figures 2.6b
and 2.7b. Himeno et al (1997) made a similar conclusion by stating that
the vertical contact stress is more sensitive to the wheel load than to the
tire inflation pressure.

The maximum lateral contact stress of the bias ply tire is higher than that
of a similar size radial tire having the same inflation pressure and carrying
the same load as shown in Figure 2.8. This finding contradicts that of
Myers et al (1999) who concluded that radial tires induce greater lateral
contact stresses relative to bias ply tires. Groenendijk (1998) also
measured the contact stresses of new and used radial tires with various
wheel loads and tire inflation pressures using the VRSPTA. Based on his
measured data, he concluded that:

1. The contact area is rectangular in shape with a constant width equal to

the tire tread width.

34

 

240

 

 

 

200 .
ii 150 .,
3 120 -
g
3 8O .
C
O . . A
0 40 . ........ . """ v I ...... _. .........
I . ....... ‘ ......... fi ------ :4; M
5000 6,000 7,000 8,000 9,000 10,000 11,000 12,000 13,000
Wheel load (lb)
l—l—TVeTtical (100 psi) —.——— Lateral (100psi) —*—-Longitudinal (100 psi) ‘
i. - - -EJ- - - Vertical (75 psi) - - -<> - - Lateral (75 psi) - - ~A— - - Longitudinal (75 psi) , .

a) Various wheel loads

Bias ply tire

 

 

 

 

; g - M
i g 120 - I3 """""" “‘~~E..-~43°~..B.0 ----- El """
l
1 80 , , , , ,
. i
. 8 4o - -
0 T T
0 20 4O 60 80 100 120 140 160
Inflation pressure (psi)
l; , , _;;:__ ,_ _ _ . , , ,, ; , W ,, : : ,. ‘
—-—Vertical (9200 lb) +Lateml (9200 lb) +Longitudinal (9200 lb)‘ 1
'. ----El- Vertical (5800 lb) -----<> Lateral (5800 lb) ---A- -- Longitudinal (5800 lb)‘

b) Various inflation pressures

Figure 2.6 Plots of vertical, lateral and longitudinal contact stresses of a bias ply tire with
various wheel loads and inflation pressures (De Beer et al, 1997)

35

Radial tire

 

o . - - a- - , - _ - _ a a . . .
5,000 6,000 7,000 8,000 9,000 10,000 11,000 12,000 13,000
Wheel load (lb)

i—o—Vertical (100 psi) —.— Lateral (100 psi) ——.— Longitudinal (100 psi)
- - -E}- - - Vertical (75 psi) - - «o - - Lateral (75 psi) ---A- - - Longitudinal (75 psi)

a) Various wheel loads

Radial tire

 

 

 

200 - - ~
~ -’-‘ 160 M
. a
l a
; g 120 a g ..... B
' E]
l ‘8 80
i a
i C
8 40
o . . ‘
o 20 40 so 80 100 120 140 160 l
} Inflation pressure (psi)
+ Vertical (9200 lb) —.—— Late7al (9200 lb) —.——10ngitudinal (9200 lb)
ma - - Vertical (5800 lb) mo -- Lateral (5800 lb) ---A- -- Longitudinal (5800 lb) .

b) Various inflation pressures

Figure 2.7 Plots of vertical, lateral and longitudinal contact stresses of a radial tire with
various wheel loads and inflation pressures (De Beer et al, 1997)

36

Radial VS Bias Tires

)

N
A
O

200 i
160 j
120 -
80 -
40 i

0 ‘. . . . . .
5,000 6,000 7,000 8,000 9,000 10,00011,000 12,00013,000

Wheel load (lb)

 

Contact stresses (psi

—l—Vertical (radial) ---E}~ Vertical (bias) + Lateral (radial) ~--<>-- Lateral (bias)

a) Various wheel loads

Radial VS Bias Tires

200 ..

160

 

120

Contact stress (psi)

 

 

 

4O 60 80 100 120 140
Inflation pressure (psi)

: +Vertical (radial) ---El--- Vertical (bias) —e—Lateral (radial) ----<> Lateral (bias)
b) Various inflation pressures

Figure 2.8 Comparisons between the vertical and lateral contact stresses of a radial tire
and of a bias ply tire at various wheel loads and inflation pressures (De Beer et al, 1997)

37

2. The maximum shear stresses were only 20 percent of the maximum
vertical contact stresses.

3. The maximum shear stress could be higher than 50 percent of the tire
inflation pressure.

4. In most cases, the maximum transverse (lateral) shear stress was an
outward shear stress under each rib located in the middle of the tire.

5. For tires whose ribs were trimmed to a groove depth of 1.5 mm or less,
the maximum transverse shear stresses were inward shear located at
the edge of the tire.

The differences in the tire-pavement contact stresses of stationary and
rolling tires were addressed by Markwick and Starks (1940). They stated that
under a stationary tire, the shear stresses are symmetrical around the center of the
tire. The shear direction is toward the center of the tire and its values decrease to
zero at the center of the tire. Under a rolling tire, the shear stresses are no longer
symmetrical. The resultant shear force on the pavement is in the opposite
direction of the vehicle traveling direction.

Schapery and Teilking (1977) applied the mechanics of pneumatic tires in
finite element analyses and developed various computer programs (TIRELOAD,
TIRETRAN, TIREFOUR and TIREF RIC) that are capable of analyzing the tire-
induced contact stresses, braking forces (parallel to the traveling direction) and
sliding force (perpendicular to the traveling direction) due to maneuvering of the

vehicle. The required program inputs include various properties of the pneumatic

38

tire. This made the programs very difficult to use since the tire industry considers

such data as proprietary information.

4.3.3 Pavement Responses

Bonaquist et al (1989) also studied the effect of different tire inflation
pressures on the pavement responses using the accelerated load testing facility
(ALF). They reported that increasing tire pressures by 20 percent causes a 2 to 10
percent increases in the pavement surface deflections and in the strains at the
surface and at the bottom of the AC layer.

Huhtala et a1 (1989) used strain gauges and pressure cells to measure the
pavement responses due to different tire types and tire inflation pressures. They
concluded that:

1. As the tire inflation pressure increases, the stresses and strains induced in
the pavement increases thereby delivering higher damage to the pavement.

2. The tire-pavement contact pressure is highest at the center of the tire for
truck tire and at the edges of the tire for passenger car tires.

3. The sensitivity of the pavement responses to tire types and tire pressures is
inversely proportional to the thickness of the AC layer.

The effects of tire motion on the pavement responses were also
investigated by Siddharthan and Sebaaly (1999) using the 3D-MOVE computer
program, which is capable of analyzing flexible pavement responses due to
moving wide-base tires. Their results indicated that the load-induced strains in the

AC layers decrease significantly due to moving-loads. These results agreed with

39

the field measurements conducted by Chatti et al (1996) and (1997) and Dai et al
(1997) that the measured strains at the mid- depth and at the bottom of the AC
layer decrease as the truck speed increase.

Siddharthan et al (2002) also used the 3D-MOVE computer program to
analyze pavement responses due to wide-base tires. They concluded that
incorporation of the inward shear stresses in the analyses does not significantly
affect the pavement responses. Note that, the magnitude of the inward shear
stresses used in their analyses were about 12 to 16 percent of the magnitude of the

maximum vertical contact stresses which is lower than the 20 percent reported by

Groenendijk (1998).

4.3.4 Pavement Temperature

Thermal stresses are results of the contraction of the AC layer due to low
pavement temperatures and the restraining force caused by the friction between
the AC and base layers (Zubeck and Vinson, 1996). Based on the above
explanation, it can be seen that the maximum thermal stress is a function of the
contraction force of the AC due to cold temperatures and the frictional force
between the AC and base layers. Shahin (1977) developed a design system to
minimize the asphalt concrete thermal cracks. In his calculation, Shahin (1977)

used the following equation to predict the thermal stresses.
AO'(t,AT) = S(t,AT)*a(Tx)*AT (2.1)
Where: A0(t, AT) = the increase in a thermal stress for a given loading

time t and a temperature interval of AT ,

40

S (t,AT ) = asphalt concrete modulus at a given time t and the
mean value of a temperature interval of AT , and
a(Tli) = thermal coefficient of contraction of the asphalt concrete

at the mean value of the temperature interval AT .

As can be seen from the equation, Shahin predicted thermal stress based
on the contraction force in the AC, which is a function of the properties of the
asphalt binder, temperature and the cooling rate. Note that the thermal coefficient
of contraction and expansion of the asphalt concrete is not a constant value and it
varies with the temperature of the asphalt (Littlefield, 1967). The value of the
coefficient of thermal contraction of asphalt mixtures at any temperature can be
obtained from the equation reported by Hiltunen and Roque (1994) and is a
function of the properties of the AC binder and aggregate and their volumetric
relationship as shown in equation 2.2 below.

VMA * Barr + inm; * B

 

AUG
aM/X : 1 (2'2)
3 * VTOTAL
Where: al,,“ = linear coefficient of thermal contraction of the asphalt
mixture (l/OC)

B M: = volumetric coefficient of thermal contraction of the asphalt
cement in the solid state (1/°C)

B AGG = volumetric coefficient of thermal contraction of the
aggregate (l/OC)

VMA = percent volume of voids in the mineral aggregate

VTOTAL = 100 percent

41

5.0

Temperature and the pr0perties of the AC mix affect the magnitude of the
thermal stress induced in the pavements. In their study of thermal and load-
induced stresses as possible causes of TDC in flexible pavements, Roque and
Ruth (1990) concluded that low temperatures and fast cooling rates could result in
high thermal stress. Further, in the same study, they also concluded that the high
thermal stresses combined with load-induced stresses can result in TDC initiation

in flexible pavements.

TENSILE STRENGTH OF AC MIXTURES

The tensile strengths of the AC mixtures are affected by several factors including:

)‘e

>

\\

>

>

the AC mix design;

temperature and temperature gradient within the AC layer;
particle segregation;

compaction; and

stress state (confinement).

The effects of each of these factors on the tensile strengths of AC mixtures are presented

in the subsections below.

5.1

AC Mixture Design

To improve the ride quality, decrease noise and bleeding/flushing potential, the

AC surface course is typically designed to have smaller maximum aggregate size (higher

aggregate surface area) and slightly lower asphalt binder content relative to other AC

courses. The lower AC content and smaller aggregate sizes results in thinner asphalt film

42

coating the aggregates and hence, higher stress concentration in the AC film (Harvey and
Tsai, 1996). This leads to lower overall tensile strength of the AC mix. Furthermore, the
low AC content tends to increase the percent air voids, which also results in low tensile
strength (Button et al, 1990). Similarly, Jung and Vinson (1993) showed that the fracture
strength of asphalt mixtures is most sensitive to the air void contents and aggregate types,
and Zubeck and Vinson (1996) showed that the tensile strength of an AC mix is a direct
function of its fracture strength.

Brown and Bassett (1990) studied the effect of the maximum aggregate size on
the properties of AC mixes by conducting several tests including indirect tensile strength
tests. They concluded that increasing the maximum aggregate size resulted in increased
tensile strengths of the AC specimens.

In a study by Baladi and Harichandran (1987), the relationship between the
indirect tensile strength of the AC mix with AC mix design parameters was investigated.
One hundred and twenty six AC specimens were subjected to indirect tensile strength test
using a newly designed indirect tensile test apparatus. In all 126 indirect tensile strength
tests, the test variables were:

1. Percent air voids of the test specimens (AV); the value of AV ranged from three
to about seven percent.
2. Kinematic viscosity of the asphalt binder (KV); the value of KV was 159, 212, or

270 centistokes.

3. Gradation of the aggregates (GRAD); the value of GRAD was 1 for gradation A

or 2 for gradation B.

43

(P

4. Aggregate angularity (ANG); the value of ANG was 2 for rounded aggregates, 3
for 50/50 mix of round and crushed limestone, and 4 for crushed limestone.
5. Test temperature (TT); the tests were conducted at temperatures of 44, 77 or 140
OF.
Equation 2.3 presented below was developed for indirect tensile strength (lNTS)
from the test results using regression analysis. Equation 2.3 has a coefficient of
correlation (R2) of 0.997 and standard error of 0.078.

ln(lNTS) = 8.0160 — 0.03363TT — 0.2605AV + 0.0509ANG + 0.0007676KV (2.3)

Based on the above equation, it can be seen that the tensile strength of the AC mix
is a function of the properties of the AC binder (KV), aggregates (ANG) and how the two
materials were combined in the AC mix (AV). It can also be concluded, from equation
2.3 and from other literature, that AC surface courses are typically (but not intentionally)
designed to possess lower tensile strength than AC base courses. Note that the above
equation also shows that tensile strength of the AC mix is also affected by temperature of
the AC. Further discussion regarding the effect of the temperature on the tensile strength

of AC mix is presented below.

5.2 Compaction
Rickards et al (1999) summarized the objectives of compaction of asphalt
mixtures as follows:
1. Achieve optimum density.
2. Achieve aggregate particle contact to provide adequate deformation (shear)

resistance.

44

3. Minimize air void contents.
4. Provide crack-free pavement surface to reduce permeability of the AC layer.
The second objective (increasing the AC mix resistance to shear deformation) can
be achieved, if and only if, the tensile strength of the AC mix is increased. Marker (1967)
explained that by moving aggregates particles closer together, the compaction helps the
asphalt film around the aggregates to develop the inter-particle cohesion force. The

increase in particle cohesion force results in increasing tensile strength of the AC mix.

5.3 Particle Segregation

Segregation of the asphalt concrete mixtures in pavements may lead to premature
distress, such as stripping, raveling, rutting, and longitudinal and fatigue cracking. These
distresses are usually caused by the decreased tensile strength or durability of the mix.
Segregation may occur during stock piling and handling, hot mix asphalt production,
truck loading and unloading, transportation and/or lay down operations. The amount of
segregation that occurs may be exacerbated by such factors as aggregate type and mixture
design (Khedaywi and White 1996, Williams et al 1996).

Chang (2000) stated that particle “segregation refers to the separation of coarse
and fine aggregates in an asphalt mix. The mechanism of segregation is based on the
motion of aggregates. Whenever aggregates are moved, there is a tendency for
segregation to occur.” He adopted the following Michigan Department of Transportation
definition of particle segregation “areas of non-uniform distribution of coarse and fine
aggregate particles in a bituminous pavement that are visually identifiable or can be

determined by other methods.” He defined three degrees of segregation as follows:

45

1. Light — The fine matrix in place, more stone than the surrounding mat.

2. Medium — Lack of surrounding fine matrix, significantly more stone than the
surrounding mat.

3. Heavy —~ Stone against stone, little or no fine matrix.

Khedaywi and White (1996) conducted laboratory investigations on segregated
mixture that they had manufactured in the laboratory. They conducted indirect tensile
strength tests and fatigue tests, and noted that the tensile strength and fatigue life are
adversely affected as the segregation becomes more severe, i.e., the gradation becomes
coarser. Brown et al (1989) conducted a similar study on segregation of asphalt mixtures
in Georgia. The results of their indirect tensile strength tests correlated well with those
conducted by Khedaywi and White. In both cases, the increase in segregation results in
high air voids. In both studies however, the levels of segregation of the mixes tested were
very severe. Differences of up to 75% passing the 3/8-in sieve were present in the mixes,
as are visible on Figures 2.9 and 2.10.

Chang (2000) investigated 25 test sites located along various pavement projects
exhibiting segregation. At each test site, his investigation consisted of:

1. Conducting nuclear density tests to determine the density and asphalt content of
the asphalt mats using the one minute readings.

2. Extracting pavement cores and determining their densities in the laboratory.

3. Mapped the boundaries of the various degrees of segregation and segregation-
related distresses. For some sites, the distress surveys were conducted 6 times

within a 30-month period while for other sites, 5 distress surveys were conducted.

46

100 .7_7.——7—-7- 7 ~ , - . a e .‘
I‘-'.-’MIXI ll ‘ 'i 1* ill/l/fi ' i

90 .I . ..... . '"Wil‘ .

80 ..l"."MlX2 ~fl . ~ i‘l-lil" d

70 i'--k-Mix3
60 ..
50 f1
4OIl-'X"Mix5

Percent passing
’2‘
3
X
A

30 "—o—INDOT"'
20
10

 

0.01 0.1 1 10 100

Particle size (mm)

Figure 2.9 Gradations for different levels of segregation (after Khedaywi and White,
1995)

Sieve Size (mm)

 

 

 

 

 

 

 

 

Percent Passing

 

 

 

 

 

 

 

 

 

 

 

 

 

.9 O >0 "‘
o 7 O 7‘ l9 4e 0.5 ‘0
\I b ~—‘ b) \1 N. O
b \O t» \D \O 00 O\ Ui\l U:
1000% 7 f . .I I I ., s .
I 7 _ . It. 7 . i I7IL Ali 17 T 9 AI £70744; 77 ,, L . l
_' I: +M1X1 LE 4:: it”: ”:1 ”:3 raft,” * :hr: 131..
7 . .. ~77. 4 i . .. 7. ' 7 a. :: ..
0 .1 ' T . j. ‘ :i . .l
80.0/o ,, , +er2 ,, ,_ I I t .Ii ii :I7 MI
,‘31 . ': ::‘.::;;.,, :L." ”'2 :
600°/ l. +M1X3 r. ....l .li...: E..
. o ,- :' , t f: ::;: rley ,::::1::::
i‘ -O-Mlx4 ? " "TI ‘p ‘ 'bf: T ,7
.-. e i .. . 77.- . 1‘1" 7 7—‘-—--—.- . . I... __. i — ~——l .‘ 4 1"— l l l l
. .. . . . I‘llr77 . i II.‘ __ . l7“ 1 .
40-00/0 , i l -El-Mix5 ;; 4:; III: ii 1 ‘2’
: -<>-Mix6 if f: ' " t l: '* '
20.0% ~ ~
. ,V.:_l ""1921: rag” _.: 1:21:
, I ; :3 l g r till. : r..-_:tir ll 1 i In...
00% T “:’l‘ l " till 1 "’i'ililli “l "'l‘i'

& w 00 4:. 3:3
o o as 9“,”?
:35:

E
O
cTLS. Standard Sieve Size

00Z#

Figure 2.10 Gradations used by Brown et al (1989) for different levels of segregation

47

His data indicated that raveling and cracking are segregation-related distresses,
which results in reduction in the pavement service life. He reported that areas exhibiting
medium to heavy degrees of segregation can be expected to lose 56 to 73 percent of their
design life due to raveling and cracking. More recently, Svasdisant et al returned to the
sites investigated by Chang, and through coring, confirmed that these cracks were TDC
(Svasdisant et a1 2001). The reduction in the pavement life is mainly due to the low
asphalt contents and high percent air voids of the segregated mix. These factors reduce
the tensile strength and the fatigue lives of the mixtures. Brown et al (1989), and
Khedaywi and White (1995) also drew similar conclusions that the tensile strengths of
the AC mixtures decrease with increasing segregation. Again, it should be noted that the
gradations observed in the field by Chang deviated from the mix design, even in the most
severe cases, by a significantly lesser degree than those mixes tested by Khedaywi and
White and Brown et al.

Williams et a1 (1996) use the PURWheel tracking device to test AC mixtures
experiencing various degrees of segregation. They used the techniques employed by
Khedaywi and White (1995) to produce in the laboratory five levels of mixture
segregation: very fine, fine, control, coarse, and very coarse and made the following
conclusions:

1. The asphalt content decreases and the percent air voids increases significantly
from very fine to very coarse segregation.
2. Segregated mixtures exhibit significant loss in performance when tested in the

PURWheel tracking device.

These conclusions were expected based on prior literature.

48

et;

Ultimately, segregation provides and ideal situation in which TDC can initiate and
propagate. In regards to the material itself, other researchers have linked TDC to gap
graded mixes, low tensile strength, and aging of the asphalt binder. Each is present on
segregated pavements where the gradation has unintentionally been made coarser, the
tensile strength has been lowered, and the low density and high air voids promote

accelerated aging and decreased durability.

5.4 Temperature and Temperature Gradient

It is well known that asphalt binders and AC mixtures are temperature susceptible
materials and their properties including viscosity, elastic modulus and tensile strength
vary with temperature. Mohammad and Paul (1993) manufactured specimens using
different AC binders and levels of compaction effort, and conducted indirect tensile
strength tests at varying temperatures. Their results indicate that lower temperatures and
the higher densities result in higher indirect tensile strengths of the AC mixtures.

Chapuis and Gatien (1995) conducted four-point flexion tests to investigate the
effect of temperature on the tensile strength of conventional and polymer modified AC
mixtures. Their results indicated that the tensile strengths of the polymer modified AC
mixtures are significantly higher than the conventional asphalt mixtures and the tensile
strength of both conventional and polymer modified asphalt (PMA) mixtures increases
with decreasing temperatures.

The results of the studies by Mohammad and Paul (1993) and Chapuis and Gatien
(1995) as presented above agree with the correlation between the temperature and the

tensile strength in equation 2.3. Such correlation between the tensile strength and the

49

 

 

temperature of the AC has a limit at a certain temperature, however. During the
development of prediction models for low temperature crack spacing, Zubeck and Vinson
(1996) conducted low temperature cracking test of AC mix Thermal Stress Restrained
Specimen Test (TSRST). Their TSRST results indicated that the tensile strength of the
AC increases as the temperature decreases until the temperature reaches the glass
transition point. As the temperature decreases further below this point, the tensile strength
of the AC decreases. Equation 2.4 below was presented by Zubeck and Vinson to

calculate the tensile strength of the AC mix at the temperature below glass transition

 

temperature.
Tensile Strength = (S‘ 7 F3) (PT 7 CT) + F8 (2.4)
(T1 - FTorg)
Where St = tensile strength (MPa) measured at temperature T) (0C),

FS = fracture strength (MPa),
F Torg = fracture temperature of un-aged AC sample (”C),
PT = pavement temperature (C), and
CT = cracking temperature of AC in the field (C), which is the same as
TSRST fracture temperature.
A simplified relationship between the tensile strength and the pavement temperature was

also presented by Zubeck and Vinson (1996) as follows:
. 250
Tensrle Strength = ——6—-(PT — CT) + F8 (2.5)

Similar conclusions regarding the relationship between tensile strength and temperature
were made by Heukelom (1966). His test results indicate that the tensile strength of AC

mixtures increase as the stiffness of the asphalt cement increases (as a result of different

50

temperatures and loading times) up to a certain value and then decreases as the stiffness

of the asphalt cement increases.

5.5 Stress State

The load-induced stress state in the AC layer varies from one point to another as
shown in Figure 2.11. The pavement surface at the tire edges is subjected to load-induced
horizontal tensile stresses whereas the bottom of the AC layer under the tire is subjected
to load-induced vertical compressive and horizontal tensile stresses. Further, the bottom
of the AC layer is also subjected to confining stresses due to the overburden pressure and
the frictional force between the AC and the aggregate base layers (Zubeck and Vinson,
1986). The different states of stress between the top and the bottom of the AC layer
impact the total tensile strengths at both locations. The total tensile strength herein is
defined as “the true tensile strength of the AC mix, which could be the same at the top
and bottom of the AC layer, and the added apparent strength due to the overburden
pressure and friction.” It should be noted that the total, not the true strength, controls

cracking potential.

6.0 TOP-DOWN CRACK PROPAGATION MODELS

While fairly extensive literature exists on the causes of TDC, relatively little
addresses TDC propagation. Most of the literature that exists addresses the method of
propagation of the cracks. Myers and Roque (2002) incorporated the effect of the
temperature gradient on the differential stiffness in the AC layer in their analyses of

tensile stresses at the crack tip using the finite element method. They reported that the

51

ebb: U< 05 .«o Eaton use a8 05 8 39m 32% fiestas 8030:7984 ZN oSmE

comma»: Emowto:
use commmuEEoo 303:5

c233 EEONTSE 56:2 EEONCOE

 

be: 0“an

5L3— 8mm

a93— U<

52

differential stiffness in the AC layer results in higher tensile stresses at the crack tip. As a
result, they concluded that the temperature gradient accelerates the propagation rate of
TDC. Further, they concluded that the crack propagation rate is not constant throughout
time, but is divided into three phases as follows:
1. During the crack initiation, the cracks propagate at a relatively fast rate
2. The crack propagation rate slows for a certain period of time. Myers and Roque
(1998) have suggested the time of slow propagation is the window of opportunity
for TDC identification and rehabilitation.
3. After a period of slow growth, TDC propagate at an accelerated rate until they
propagate throughout the AC layer.
Jacobs et al (1992) studied crack propagation through flexible pavements with an AC
overlay on the original cracked AC surface. In the study, they conducted uniaxial tensile
tests of notched AC specimens to study the behavior of crack growth in asphalt concrete.
They concluded that the cracking process in AC material can be simplified and analyzed
using a single crack propagation model. Further, they used CRACKTIP, a 2-D finite
element computer program, to calculate mode I (opening) and mode 11 (shearing) stress
intensity factors in front of the crack tip in the AC layer. Combining the two stress
intensity factors, they calculated the equivalent mode I stress intensity factor (Kreq),
which varies with the crack length during the crack propagation. They used the Krcq to
calculate the effective stress intensity factors (Kerr), which is constant, and used their
prediction model (equation 2.5) for the number of load repetitions that crack required to

propagate throughout the AC overlay.

h
Ntot z 0 (2.6)
n
AKeff

 

53

Where Nm, = the number of load repetitions

ho = the overlay thickness (mm)

Keff = the effective stress intensity factors

A,n = material constant in Paris’ law
The above equation was developed based on the Paris’ law crack propagation model as
presented below.

22 = AK“ (2.7, Jacobs et al, 1992)

dN
Where: c = crack length (mm) per loading cycle, N
K = stress intensity factor (N/mml's)
A,n = crack propagation coefficients (material constant)

Similarly, Molenaar (1984) and Jacobs (1996) modified Paris’ law in their crack
propagation analyses such that crack propagation coefficients (“A” and “n”) can be
determined using simple static tests and master curves. They reported, however, that their
models contain some errors due to the simplifications and the discrepancies of the linear—
elastic and homogeneous material assumptions and the actual asphalt mix characteristics.
Similarly, Dauzats and Rampal (1987) used the finite element method to calculate the
stress at the crack tip, and developed a propagation model of TDC based on Paris’ law.
They concluded that their propagation model has some drawbacks including the two-
dimensional stress analyses and the exclusion of the elastoplastic calculation. The
accuracy of the above TDC propagation model developed by the above mentioned
researchers could not be verified or calibrated because their historical TDC data is not

available in any State Highway Agency database.

54

563

pm

Sit

lllt

lllC

'l'.

 

 

 

Nevertheless, there is at least one TDC propagation model that was developed
based on TDC historical data. Schorsch et al (2003) conducted field investigation of
segregated and non-segregated pavements as a part of their study of TDC initiation and
propagation in segregated areas. Based on TDC propagation data obtained during their
study and historical distresses data on the same pavements collected by Chang (2000),
they developed TDC crack propagation model in the segregated area as shown below.

PC = 0.0602 DS+0.0093 DS In (t) (2.8)

Where: PC = Percent (TDC) cracking from segregated areas = Crack length within
an area showing certain degree of segregation / Length of the area
showing that degree of segregation * 100%
DS = Numerical value of degree of segregation (Light = 2.5, Medium =
7.5, Heavy = 10)
t = Pavement surface age (months)
Further, Schorsch et al (2003) rearranged the preceding equation to calculate the service
life of the pavement from TDC based on areas experiencing different degrees of

segregation. Their modified equation is stated is follows:

 

(2.9)

PC th — 0.0602 DS
SL = exp

0.0093 DS

Where: SL = Service Life of the pavement (months)
PC”, = Percent cracking threshold at which RSL equal zero
By comparing the design life with estimated service life, it is possible to determine the

reduction in the design life due to TDC in segregated areas.

55

CHAPTER 3

FIELD AND LABORATORY INVESTIGATION

1.0 GENERAL
The laboratory and field investigations were designed to identify the causes of
TDC and their propagation rates. The investigations consisted of:

1. Selecting pavements showing TDC.

2. Coring, to confirm the presence of TDC and to provide material for laboratory
testing.

3. Conducting falling weight deflectometer (FWD) tests on selected pavement test
sections to determine the moduli of the pavement layers.

4. Examining the pavement surface to document the types of distress and possible
particle segregation in the AC mat.

5. Conducting a battery of laboratory tests such as specific gravity, indirect tensile
cyclic load tests (ITCLT), indirect tensile strength tests (ITST), aggregate
extraction, and particle size (sieve) analyses to:
> Determine the physical and engineering characteristics of the asphalt mixes.
‘9 Assess the effects of segregation and moisture conditioning on the indirect

tensile strength of the asphalt mixes.

2.0 SELECTION OF PAVEMENT SECTIONS
The 13 flexible and 5 rubblized pavement test sections listed in Table 3.1 were

selected by MDOT and PRCE personnel for field and laboratory investigations. The only

56

Table 3.1 Summary of the test sections and test sites selected for field and laboratory

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

investigations
Test . .
Route Control Test sec tio n/ Desrgnatlon Pavement
section lane . number type
Test Site

US-27 37014 SB 1-1 20273-11
US-41 7013 NB 1-1 20411-11
US-41 7013 NB 1-2 20411-12

M—20 37022 EB 1-1 30202-11

M-20 37022 EB 1-2 30202-12

M-42 83041 EB l-l 30422-1 1 La:
M-50 23052 EB 1-1 30502-11 g
M-50 23052 EB 2-1 30502-21 5:
M-52 33051 NB 1-1 30521-11

M-52 76011 NB 2-1 30521-21

M-55 72022 EB 1-1 30552-11

M-55 72022 EB 1-2 30552-12

Marsh Road] SB 14 Marsh

[-194 13033 NB 1-1 10941-11

L194 13033 NB 1-2 10941-12 i3

I-75 16092 SB 1-1 10753-11 E

I-75 16092 SB 1-1 10753-12 0:2
M-15 25092 SB 1-1 30153-11

 

 

 

 

 

 

1. Marsh Road is owned and maintained by Ingham County Road
Commission

57

 

criterion used in the selection process is the presence of longitudinal TDC in the vicinity

of the wheel paths.

3.0 FIELD INVESTIGATION
The various activities ofthe field investigation are detailed in the next six

subsections.

3.1 Test Layout

Within each selected pavement, a 280-ft long and 12-feet wide test section was
established. The test section was subsequently divided into two 90-11 long test sites
separated by a 100-ft gap. Figure 3.1 shows the two typical layouts of the test sites within
a test section for divided highway and 2-lane road.

At each test site, the FWD and core locations were marked on the pavement
surface based on nine transverse and seven longitudinal grid lines as shown in Figure 3.2.
The (X3) in the figure designate FWD test locations whereas the (Os) designate core
locations.

Each test site was designated using seven-di git number as follows:

‘7

The first digit represents the road type (1 = Interstate, 2 = US, 3 = Michigan).

‘7

The second through the fourth digits represent the hi ghway/route number.
‘9 The fifth digit represents the traffic direction (1 = North, 2 = East, 3 = South, 4 =

West).

> The sixth digit expresses the test section number.

‘7

The seventh digit represents test site number.

58

mbzm $00 van 20503 E00 :0 3:33 3203. mm Eswi

can: 3050050 £000 5 00E 0:8 0:2 N :0 E00 m («o 59?: 3V

 

 

A

A couoom 30H arowm

 

ll

 

 

 

m awash was .
$0500th U—ufiwwur—t . .. . , ,. ,.

 

 

 

 

 

 

 

 

5:085 058% V

 

 

 

33:35 00230 a :0 c2003 38 a :0 5053 A3

 

 

88% a3. 0%ch

 

ll

 

 

 

 

\
A

N 35 he area, . _ ea see was

 

 

003085 058%

 

 

 

 

 

550005 058% A

 

 

 

 

59

Road Edge of

 

 

 

 

Centerline shoulder
l
: 40 30 26 22 I8 14 10
. \/ m \/
1 1 X /\ ><>z X {8} /\ {8}
| a
1 10 39 09 A
1 + X X
I
i
l 38 08
I
. X
I
: 37 29 25 21 17 13 07
1 X
I
1
1 36 O6
1 Traffic
l X Direction
I
l 35 05
l
1 X
l
: 34 28 24 20 I6 12 04
i X (X)
|
: 33 03 {
1 X X
l
l
_ 02
1 32 LongL
: >< tudinal
1 3, Direction
. . ___)
: Fixed 31 27 23 19 15 11 01
3 >< 1231 >< >< 131 j
i —" 1’ l‘— I
l
I
l <—— Transverse Direction ———>

Figure 3.2 Typical test site layout showing FWD tests (X) and core locations (0)

6O

[65! :
«lCSlg
desié

6X31?

U

sq.

PEVen

\iffia

.w

L’J

For example, a designation number of 11941-12 implies route I-194, northbound
test section 1 test site 2. In addition, the two digit numbers shown in Figure 3.2 that
designate the FWD test location number was added to the right-hand side of the test
designation number. Hence, the code that identifies test location consists of 9 digits. For
example, a test conducted at location 23 on test site 1 of test section 1 located on the
south-bound lane of M-1 5 would be designated 30153-1101. At some test sites,
additional FWD tests were conducted and cores were extracted at special points of
interest not shown in Figure 3.2. For these cases, the test number for each point was
assigned starting at 50, and its location was referenced by the longitudinal and transverse

distance from test point 1.

3.2 Distress Survey
During the field investigation, detailed distress maps were developed at selected

test sections. Early in the study, the criteria for selecting the test sections were based on
the extent of cracking observed at the time of field investigation. For example, a
pavement that exhibited cracks at the earliest stages of initiation was selected because it
was assumed that periodic distress surveys could be conducted to assess the rate of crack
propagation. As the field investigation progressed however, distress maps were
developed for each test site. The purposes of these maps were 3 fold:

1. To catalog the general appearance of TDC.

2. To identify any distresses that may generally be associated with TDC.

3. To determine crack growth over time and therefore determine the rate of crack

propagation.

61

1).)

33

fond

Objsg

lJ

The distress maps were developed by measuring the coordinates of the crack ends
relative to the test grid. In addition to mapping distress, visually identifiable segregation
areas were also mapped. The two maps were used to visually identify those cracks that
fell within the segregated areas. Additionally, each visually identified segregated area
was assigned a degree of segregation based on the apparent severity of segregation. The
definitions of each degree of segregation are the same as those developed by Chang
(2000) and are listed below:

1. Light —— The fine matrix in place, more stone that the surrounding mat.
2. Medium — Lack of surrounding fine matrix, significantly more stone than the
surrounding mat.

3. Heavy — Stone against stone, little or no fine matrix.

3.3 FWD
During the course of this study, nondestructive deflection tests (NDT) were
conducted on all test sites using the MDOT falling weight deflectometer (FWD). The
objectives of the tests were to:
1. Measure the variation in the pavement deflections along and across the pavement.
2. Backcalculate the moduli of the pavement layers.
3. Assess the linearity of the pavement response to load.
The configuration and spacing of the 9 FWD deflection sensors are shown in Figure 3.3.
All deflection data were recorded by an on-board computer to within 0.01-mils
(0.00001-in). At each FWD test location, the pavement and air temperatures were also

measured and the data were recorded. Afier the FWD tests were completed, pavement

62

12” 8” 4” 6” 6” 12” 24”
L | Al

A
‘

A
V
l
V
V
l
V
A
V
l
l

1 0
D2 D3 D4 D5 D6 D7
Loading Plate
Radius=5.91—in Sensors
12” Left

Figure 3.3 Top view of the sensor configuration of the MDOT FWD

63

C1

.\1

01

1’?!

(a
’
'J I

cores were extracted and the thickness of each core was measured. The thickness data
were used as input to the backcalculation program. The deflections from sensors D1
through D7 (see Figure 3.3) were used to backcalculate the layer moduli using the
MICHBACK computer software.

All FWD tests consisted of 4 drops. The first drop was used to seat the load plate
on the pavement surface, and therefore the deflection data from this first drop were not
recorded whereas the data from the next three drops were recorded. Most tests were
conducted at the target load of 9000-lb. At some test sites, about ten additional FWD tests
were conducted at each of four load levels to assess the repeatability of the deflection

data and the linearity of the pavement response to load.

3.4 Nuclear Density

Afier the test locations were marked on the pavement surface at each test site,
one-minute nuclear density measurements were made using a Troxler nuclear density
meter shown in Figure 3.4. The purpose of the nuclear density readings was to confirm
the observed level of segregation. The level of segregation was confirmed by analyzing

differences in nuclear density between visibly segregated and non-segregated areas using

the computer program MBITSEG202 (Chang, 2000).

3.5 Coring
After all non-destructive tests had been completed, cores were extracted at the
designated locations using a power rotary drill as shown in Figure 3.5. Each core was

numbered according to the system previously discussed in section 3.1.

64

.0 lwm~~ . ‘57

J ......0»

Luv—J

,.
.2“
m

f':'

 

., .3 v 0: : in}. , 3

.364 gurzn 0

.12.). 0... .7. r3: ad.
$514!...

 

65

 

‘r:

J.’ . i I n- #3 _- r0. M4312; ..
Figure 3.5 Drilling cores using a power rotary drill

66

3.6 Boreholes

At most test sites, when coring was completed, boreholes were drilled to a
maximum depth of 4-ft using a hand auger. The extracted material was then examined to
estimate the thickness and material types of the base and subbase and the type of the

roadbed soil.

4.0 LABORATORY INVESTIGATION
Laboratory testing was conducted to determine the physical and engineering

characteristics of the AC mixes exhibiting TDC. The testing program consisted of the
following activities.

1. core preparation;

2. specific gravity tests;

3. indirect tensile cyclic load Tests (ITCLT);

4. indirect tensile strength tests (ITST); and

5. aggregate extraction by incineration and sieve analysis.

4.1 Core Preparation

Cores were prepared for testing as soon as they arrived in the laboratory. Each
core was washed to remove excess aggregate or any other material contamination that
may have occurred during coring or handling. The total thickness of each core was
measured to the nearest 0.1-mm at four points located at the opposite ends of two
orthogonal diameters. Then, the thickness of each asphalt course within the core was

measured. Once all measurements were completed, the lowermost portion of the core that

67

had been contaminated by the base material (typically the lowermost 0.5-in) was

removed by sawing. The core was then marked with the appropriate designation number.

4.2 Specific Gravity Tests

Once the cores had been prepared as discussed in section 4.1, they were subjected
to specific gravity tests, according to ASTM standard test procedure D-2726 (ASTM
1994). The test results were used to determine the bulk specific gravity of the cores using
the following equation:

A
GSB 213—:5 (3.1)

Where: A = mass of the dry specimen in air (g)
B = mass of the saturated surface-dry specimen in air (g); and
C = apparent mass of the specimen in water (g)

G S b = Bulk specific gravity from equation 3.1

After the specific gravity tests of the cores were completed, each core was sawed
into test specimens according to predetermined thicknesses. These specimens were again
subjected to the preparation procedure discussed in section 4.1 and to the specific gravity

tests discussed above.

4.3 Indirect Tensile Cyclic Load Test

After sawing, test specimens from each test site were subjected to ITCLT. In this
test, a cyclical load pulse is transmitted across the vertical diameter of a cylindrical
specimen using a 0.5-in wide loading strip. The resultant deformations in the vertical,

horizontal, and longitudinal directions were measured. The deformations at the 500th load

68

cycle were used to determine the resilient modulus and Poisson’s ratio of the specimen.

After sawing the test specimen and measuring its specific gravity, the ITCLT was

conducted using the following steps:

1.

All moving parts and the two loading strips of the ITCLT loading device were
cleaned and lubricated.

The ITCLT loading device was placed on the MTS loading frame such that the
center of the device corresponded with the center of the MTS actuator and the
center of the load cell.

The test specimen was placed on the loading frame and five linear variable
differential transducers (LVDT) were placed in their respective positions as
shown in Figure 3.6. The accuracy and range of the LVDTs are shown in Table
3.2. In addition, a sixth LVDT placed inside the MTS actuator was also used to
measure the test specimen deformation along the vertical diameter.

The position of each LVDT on the specimen holder was adjusted by moving the
LVDT core toward or away from the test specimen using two position adjustment
nuts until the LVDT signal indicates that the core position would allow the use of
a high percentage of the LVDT range.

A SO-lb sustained load was placed on the test specimen by lowering the actuator
of the MTS. The resulting test specimen deformations were recorded.

When the rate of deformation due to the sustained load was small enough (not
measurable), a 250-lb cyclic load was applied to the test specimen at a frequency

of 2 Hz. Each cycle consisted of 0.1-second of load-unload and 0.4-second

69

Vertical

J '-n«;:

Longitudinal LVDT

 

 

 

Figure 3.6 ITCLT and ITST specimen holder for a 6-in diameter and 3-in thick test
specimen

70

Table 3.2 Sensitivity of LVDTs used in the ITCLT apparatus

 

 

 

 

 

 

 

 

MIT/2;.“ Position Range (in) A0235)?”
2 Horizontal diameter :0. 125 0.00005
2 Specimen thickness :0100 0.00001
1 Vertical diameter i0.250 0.00010

 

71

 

relaxation periods. Figure 3.7 illustrates a plot of one load cycle as a function of
time.

7. Each test specimen was subjected to a minimum of 700 load cycles. At certain
specified cycles (e.g., the 200 and 500 cycles), the load magnitude and the
deformations of the test specimen in three directions were recorded for three
sequential cycles. The rate of data collection was set at one set of deformation
readings every 0.0004-second. Hence, 250 deformation readings were collected
by each LVDT during the load-unload cycle and 1,000 readings during the rest
period. Note that the load data were collected using a lOOO-lb capacity load cell
located under the ITCLT loading frame (see Figure 3.8).

8. The test was terminated and the data was downloaded for analysis.

The load and deformation data corresponding to load cycles 499, 500 and 501 were used
to calculate the resilient modulus of the test specimen using one- and three-dimensional
analyses as follows:

a) One-Dimensional Analysis ~ The resilient modulus of the test specimen was
calculated using the deformation measured along the vertical diameter of the test
specimen, equation 3.3 (Harichandran and Baladi, 1987). In this calculation,
since the impact of Poisson’s ratio on the value of the resilient modulus is
insignificant, a Poisson’s ratio of 0.3 was assumed for all test specimens.

P(4.085950 -0.04l7333 1))

MR _
V- LD

 

(3.3)
V

Where: MR V = Resilient modulus based on vertical deformation (psi);
P = peak cyclic load (lb);

v = Poisson’s ratio;

72

 

Load Unload Rest period

 

Load (lb)

 

 

0 ‘ . 1
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

 

time (sec)

 

 

 

Figure 3.7 One load cycle consisting of 0.1-second load-unload period and 0.4-second
relaxation period

- s 2;:
.
is"; aux] ..

Load Cell ——’

Figure 3.8 Test specimen in the specimen holding devic during an ITCLT test

 

73

L = thickness of the test specimen (in); and
Dv = the deformation of the test specimen along the vertical

diameter (in).

b) Three-Dimensional Analyses - The resilient moduli of the test specimens

were calculated using the deformations measured along the vertical and

horizontal diameters and along the thickness of each test specimen using the

equations below (Harichandran and Baladi, 1987).

Where:

MR3D =

 

(0.1832585H + 4.2817159V — 0.0215089A)

3.4
D ( )

 

D =1.O468779(H2 + v2 + A2 j - (H — 0.0417333v + 0.212453A)2

=Dh*L. D *L D1

,V= V ;and A:—
P P P

 

MR3D = Resilient modulus based on 3-D deformations (psi);

L = thickness of the test specimen (in);

P = peak cyclic load (lb);

D, = the horizontal deformation = the sum of the displacements
measured by the two LVDT along the horizontal diameter of the

test specimen (in);

Dv = the deformation of the test specimen along the vertical diameter
(in); and

D1 = the longitudinal deformation = the sum of the displacements
measured by the two LVDT along the thickness of the test

specimen (in).

74

Note that, the vertical deformation used in equations 3.3 and 3.4 was that obtained
from the LVDT placed inside the MTS actuator. The reason is that the actuator and the
LVDT were positioned at the center of the test specimen. The outside mounted LVDT
was positioned at the corner of the top plate of the ITCLT device which may be affected
by the rocking motion of the test specimen during the cyclic load test. The resilient
moduli obtained from equations 3.3 and 3.4 were compared. Detailed data (measured
deformations in three dimensions, test specimen thickness, and the resilient moduli
calculated using equations 3.3 and 3.4) for each core are presented and discussed in

chapter 4.

4.4 Indirect Tensile Strength Tests

All specimens tested in the ITC LT mode were subsequently subjected to indirect
tensile strength tests. In this test, the load is applied across the diameter of the specimen
at a constant 2-in/min rate until failure. The peak load measured is the load required to
fail the specimen, and the vertical deformation corresponding to that load is the
deformation at failure. By utilizing the geometry of the specimen, the tensile strength of
the specimen was determined.

A Reinhart analog device was used in conducting the tests. Since the output was
not recorded digitally, there is a certain degree of subjectivity involved in interpreting the
results. Therefore, a standard procedure for reading the output graphs was developed. The
procedure is as follows.

1. After initial contact is made between the loading head and the specimen, the load

increases in a roughly linear fashion. This linear portion of the load/deformation

75

curve was extrapolated backwards to the load datum (P=0-lb). This point was
taken as zero deformation for the specimen (do).

2. The peak load Pf (also referred to as the load at failure) was determined for the
specimen. The deformation corresponding to this load was defined at the
deformation at failure (dr).

3. A third point, at a load approximately half of the peak-load, was located on the
curve. The load and deformation at this point (Pm and (Im respectively) were
recorded.

The strain at failure was calculated as follows:

df (3 5)
a =—- .
t D

Where: St = Strain at failure (in/in)

df = Specimen deformation at failure (in)

D = Specimen diameter (in)

The tensile strength of the specimen was determined as follows:

 

s — 2 ' Pf (3 6)
t _ n-D-t .
Where: St = Tensile strength of the specimen (psi)

P f = Peak load (lbs)

D = Specimen diameter (in)
t = Specimen thickness (in)
From the load/deformation curve of the specimen, an equivalent modulus (ITST)

could be calculated as follows:

76

 

P
EEQ = m (4.08595 —0.04l733v) (3.7)
m

Where: EE Q = ITST equivalent modulus (psi)

Pm = Approximate half-peak load (lbs)
dm = Approximate half-peak deformation (in)

t = Specimen thickness (in)

v = Poisson’s ratio (assumed = 0.3)

4.5 Extraction of Aggregate

After the preceding tests had been completed, the aggregate was extracted by
ASTM standard test procedure D-5444, which allowed sieve analysis to be conducted on
the recovered cores. All extraction tests were conducted using a forced air ignition
furnace preheated to 450°C. Each sample specimen was evenly placed within a basket
which was placed in the furnace and burned until the internal scale stabilized, indicating
that the extraction process was complete. The aggregate was then retrieved and subjected
to sieve analysis. By measuring the weight of the sample prior to and after incineration,
the asphalt content could be determined by the following computation.

%AC=A_B

 

xum% (3&

Where: A = Mass of the specimen in air prior to incineration (g), and;

B = Mass of the specimen in air after incineration.

77

Once the aggregate and fine material left from the extraction process had cooled, the
particle size distribution was determined by sieve analysis. The sieves used and the

corresponding sieve openings are listed in Table 3.3.

5.0 MANUFACTURED TEST SPECIMENS

During the field investigation, TDC were observed to initiate in segregated areas.
Therefore, it was desirable to obtain the engineering characteristics of the segregated
mixture in particular to assess if a correlation existed. Conducting the required tests on
field specimens is hampered by several problems which are listed below:

1. The segregated areas were frequently deteriorated or cracked. This made
obtaining segregated material suitable for laboratory investigation difficult.

2. The segregated areas were only visible in surface course, which, in most case, was
approximately 1-in thick. Several of the tests employed required specimen
thicknesses of at least 2.2-in. Therefore, it was impossible to obtain test results of
pure segregated material. This situation is shown in Figure 3.9.

In an attempt to circumvent these problems, segregated asphalt mixture was
manufactured in the lab to determine if there was any difference in the engineering
characteristics between the segregated and non-segregated mixtures. The basic goal of
producing manufactured specimens was to produce a pure specimen of material that
closely matched the gradation and asphalt content of segregation in the field, and then
performing the above mentioned battery of tests to determine if the properties of the

segregated mixture were significantly different than those of the non-segregated mixture.

78

Table 3.3 Sieves and corresponding opening sizes used in the sieve analysis

 

 

 

 

 

 

 

 

 

 

 

Sieve Size Opening (mm) [Opening (mm)]0'45
0.75-in 19 3.76
0.50-in 12.5 3.12
0.3751n 9.5 2.75

No. 4 4.75 2.02
No. 8 2.37 1.47
No. 16 1.18 1.08
No. 30 0.60 0.79
No. 50 0.30 0.58
No. 100 0.15 0.43
No. 200 0.075 0.31

 

 

 

 

 

79

2.2” Required
Thickness

1 ll/l6”

2 [/4”

2 9/16”

2 3/8”

 

u. .- ..,.§

7
a

O
.I
3"

...... an
.1 t . I.
u ‘0» hr.
V‘ .

.. w... .91....

fiAUuSQ'0 ”$41.

Wm $51.

 

Figure 3.9 Relationship between actual core thicknesses and required minimum thickness

for ITCT and ITST testing

80

5.1

Manufacture of Specimens

The gradation and asphalt content of the segregated mixtures was determined by

incinerating cores obtained from cracked and non-cracked areas by the method outlined

in section 4.5. Specimens were then fabricated according the following process.

1.

Fresh aggregates and asphalt cement were obtained from Spartan Asphalt Paving
Company in Lansing, MI. The aggregates used were manufactured sands and
gravels, and the binder used was a SUPERPAVE PG 64-22.

The aggregates were dried and separated into the required sizes by sieving. The
materials used were those retained on the %-in, 3/8-in, #4, #8, #16, #30, #50,
#100, #200 sieves and the pan, respectively.

Aggregates retained on the %-in, 3/8-in, #4, #8, #16, #30, and #50 sieves were
washed to remove any dust. The aggregates were then placed in an oven heated to
90°C to dry.

The oven dry aggregates were combined by weight to create several specimens
having gradations matching those of segregated or non-segregated pavements in
the field as well as the mix JMF. The combined aggregates were then placed in an
oven and heated to 124 °C.

The asphalt content was calculated based on the incineration data from recovered
cores. In cases where the asphalt content was not obtained from incineration of
the original cores, the asphalt content was calculated based on equivalent binder
film thickness of the JMF (Roberts et al, 1991).

The prescribed weight of asphalt cement (heated to 140°C for not more than 12-

hours) was added to the aggregate. The mixture was then transferred to a mixing

81

bowl (heated to 140 °C) and mechanically mixed until all the aggregate was
evenly coated with asphalt cement. The mixture was then placed back in an oven
heated to 140 °C for at least S-minutes but not more than 10-minutes to reheat to
compaction temperature.

7. The mixture at compaction temperature was transferred to a gyratory mold
(heated to 140 °C) and compacted to predetermined number of gyrations in a Pine
gyratory compactor.

After compaction, the specimens were allowed to cool to room temperature before

handling, after which they were subjected to the same laboratory investigation outlined in

section 4.0.

5.2 Moisture Conditioning of Specimens

TDC have been observed to occur within a few years of construction, which is
sooner than generally accepted instances of long term aging of the asphalt binder.
Therefore, it was suspected that open gradations associated segregation increased the
moisture susceptibility of mix, and the subsequent environmental action decreased the
tensile strength. Therefore, moisture susceptibility tests were conducted in accordance
with ASTM standard test procedure D-4867M (1994 Annual Book of ASTM Standards).
In this test, the tensile strength of the specimens is measured afler moisture conditioning.
The tensile strength of specimens in a control group not subjected to moisture
conditioning is used as a baseline. Hence, significant loss of tensile strength relative to

unconditioned specimens indicates that the mixture is susceptible to moisture damage.

82

Michigan is located in a wet freeze region; therefore the freeze/thaw option of this test
was included.

For each mixture type tested, a total of fourteen 100-mm specimens were
manufactured according to the procedure outlined in section 5.1. These specimens were
divided into two subsets consisting of seven specimens each; the first subset was
composed of non—segregated specimens while the second subset was composed of
segregated specimens. Three specimens (a triplicate) of each subset were subjected to the
moisture conditioning procedure while the other three were tested without conditioning.
The theoretical maximum density was obtained from the last subset by ASTM standard

test procedure D-2041 (1994 Annual Book of ASTM Standards).

6.0 TESTING SUMMARY

The testing procedure outlined above indicates the final evolution of the testing
program, which had been expanded over time to include new tests and procedures. This
implies that each pavement tested may have been subjected to a different testing regime.
A summary of the tests conducted on each site is therefore presented below. Table 3.4
shows a matrix summarizing the field investigation activities, and Table 3.5 shows the

laboratory investigation activities.

6.1 I-194, Test Section 2, Test Sites 1 and 2 (11941-11 and 11941-12)
I-194 south of Battle Creek is a 4 lane (2 lanes in each direction) divided highway
with 12-ft lanes and asphalt shoulders. The general terrain consists of very gently rolling

hills. The pavement is a rubblized concrete slab with asphalt overlay. The primary

83

Table 3.4 Summary of field investigations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . Number of each activity_conducted
Route Des1gnatlon Distress Nuclear Pavement
number FWD . Cores type
survey densny

US-27 20273-11 N/A 40 0 12
US-41 20411-11 N/A 42 0 14
US-41 20411-12 N/A 42 0 l4

M-20 30202-11 5 30 0 11

M-20 30202—12 5 29 0 9

M-42 30422-11 N/A 52 52 27 %
M-50 30502-11 N/A 41 0 13 g
M-50 30502-21 1 48 48 38 E
M-52 30521-11 N/A 44 48 25

M-52 30521-21 N/A 41 41 10

M-55 30552-11 4 33 0 15

M-55 30552-12 4 37 0 l4

Marsh Road N/A 58 0 8

I-194 10941-11 N/A 43 0 15 13
I-194 10941-12 N/A 43 0 15 93,

1-75 10753—11 N/A 42 0 14 g

I-75 10753-12 N/A 43 0 15 g
M-15 30153-11 N/A 43 0 l6

 

 

 

 

 

 

 

84

 

COCwODU Cm Dim “wow Or; how Coaodfifioo HOG 0.53 3mg 21:00am “at wuumomfifim (CZ A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 : N: m: .2 E :05... 2-2
W 4 4 m N: 4 2 2-3:: m5
m1.. 4 m w : v : :08: m:
m. w w w: : a 3 ~25: 42-:

w a M: o: 2 2 5:5: 42.:

M: M: : <2 .2 w 080 new:
: N: m 2 2 .: 2.29: 3-2
: N: e 2 2 2 :22: 3.:
<2 <2 <2 0 <2 0 8.222 ~02
<2 <2 <2 2 <2 2 _ 7 _ .23 ~02
H <2 <2 mm 2 <2 mm 8-8:; 022
W 2 <2 : N: :N 2 $82; 8.:
a <2 <2 <2 8 N<2 R. : 20% ~12
a a E N: a N: 2-88m 8-2
a a N: a o a :-88m 8.:
N: N: m 3 E E 2:48. :2:
N: N: 0 E E E :-:3~ $0:
a a a e a S :émom 3m:

53me 530% 0000005
093 HP: .50: 3:00am 0:589. 3086000 0000003 0028::

E08020: 5860mm 0.80 .3 00.80 00 cosmcwfi0fl 050%

fiDHOSUGOo mummy MO #03852 HODESZ gong—=2

 

 

 

 

 

 

 

285%:ng 409832 .8 bafifism m.m 030,—.

85

distress on the section is longitudinal cracking occurring randomly within the inner
wheelpath of the outer lane. Full lane width transverse cracking is also visible. Finally,
low severity strip segregation was noted near the outer lane centerline.

PRCE personnel identified the pavement as potentially exhibiting TDC, and field
investigation was conducted on November 7, 2001. Two test sites were established
according to the standard layout in the north-bound outer lane on a mild upward grade.
The FWD testing proceeded normally. Upon coring, however, it was discovered that the
joints had been milled out and replaced with aggregate. It also appeared that the rubblized
concrete had deteriorated. Samples of both the aggregate joint filler and the deteriorated
rubblized concrete were obtained for further examination in the laboratory. Detailed
distress and segregation maps were not developed for this pavement, nor were nuclear
density tests conducted. Ultimately, 86 FWD tests were conducted and 30 cores were
obtained from this section, with 43 FWD tests and 15 cores from each test site.

The laboratory testing proceeded as outlined in section 3.5. The cores were
cleaned, the physical dimensions were recorded and the loose material was removed from
the bottom of the cores. The cores were subsequently subjected to specific gravity tests,
sawed into test specimens and specific gravity tested again. The specimens were then
tested in ITCLT and ITST modes. None of the specimens were incinerated and subjected

to gradation analysis.
6.2 [-75, Test Section 1, Test Sites 1 and 2 (10753-11 and 10753-12)

I-75 just north of the US-131 interchange is a four lane divided highway with 12-

lanes and a 9-fi asphalt outside shoulder. The general terrain is low rolling hills. The test

86

section itself is located in a low area characterized by coniferous woodland and swamp.
The pavement type is rubblized concrete slab with asphalt overlay. PRCE personnel
identified the pavement as potentially exhibiting TDC. The most prominent feature was a
continuous strip of “drip and chip” sealant along the inside edge of the outer wheelpath.
A tight network of interconnected longitudinal and transverse cracks is also visible
throughout the pavement section. Most of these cracks had been routed and sealed at the
time of the field investigation. Although no segregation was evident on this site, a
maximum of 3/8-in rut was measured in the inner wheelpath.

The field investigation was conducted on September 24, 2001. MDOT FWD and
coring personnel were on hand to conduct the tests. The test section was established, and
two sites using the standard test layout were placed. A total of 85 FWD tests were
conducted and 29 cores obtained. 42 FWD tests and 14 cores were obtained from test
site 1 and 43 FWD test and 15 cores were obtained from test site 2.

Difficulties in backcalculation of these sites resulted in a follow-up survey. On
March 17, 2002, a hand auger was used to determine the soil profile to a depth of four
feet in the native soil approximately 50-fi from the pavement edge. Additionally, two soil
samples were obtained. Additional observations as to the condition of the pavement and
the surrounding areas were made. The most significant observation was that the water
table is approximately 4-fi below the pavement surface.

The laboratory testing proceeded as outlined in section 3.5. The cores were
cleaned, the physical dimensions were recorded and the loose material was removed from
the bottom of the cores. The cores were subsequently subjected to specific gravity tests,

sawed into test specimens and specific gravity tested again. The specimens were then

87

tested in ITCLT and ITST modes. None of the specimens were incinerated and subjected

to gradation analysis.

6.3 US-27, Test Section 1, Test Site 1 (20273-11)

US-27 south of Mt. Pleasant is a four lane divided highway. MDOT inventory
data originally indicated that this pavement was a rubblized pavement (rubblized concrete
slab with an asphalt overlay). This pavement had exhibited no distress, but was selected
for field investigation as a control group for another project conducted by the PRCE. On
June 4, 2002, the test section with only one test site was established, and 40 FWD tests
conducted. Subsequent coring revealed however that the pavement was not rubblized.
The removal of the rubblized material could be explained by the close proximity of an
overpass approximately 300-fl to the south of the test section where the concrete
materials may have been removed to maintain bridge clearance. All prescribed cores
were drilled from this one site and were subjected to various tests, more as trial
specimens than actual data collection. The FWD results were recorded and analyzed

however, and are used as a control group for these tests on flexible pavements.

6.4 US-41, Test Section 1, Test Sites 1 and 2 (20411-11 and 20411-12)

US-41 north of Baraga is a two lane (one lane in each direction) flexible
pavement running along the shore of Lake Superior. The topography is hilly coniferous
forest perched on bluffs overlooking the lake. The pavement structure features a flexible
surface layer placed on top of a concrete slab with an aggregate separator layer in

between. On the construction plans, the separator layer varies from 2 to 4-ft thick. This

88

pavement was identified by MDOT personnel as a candidate TDC project, and therefore
field investigation was conducted on August 22, 2001. The only notable distresses on the
pavement were longitudinal cracking near the inner wheelpath in an area of light to
medium segregation.

The standard testing layout was placed on the pavement, and subsequent coring
indicated that the cracking was indeed TDC. Segregation was noted on the section and
the longitudinal TDC was observed to occur in the segregated areas. Boreholes of the
underlying base material indicated that the separator layer was approximately 4-ft thick
and was laid using two different materials. The first, deeper layer was approximately 3-ft
thick and was composed of sand. The second, upper layer was approximately l-ft thick
and was composed of gravel. A total of 84 FWD tests were conducted and 28 cores
obtained, with 42 FWD tests conducted and 14 cores obtained from each site.

The laboratory testing proceeded as outlined in section 3.5. The cores were
cleaned, the physical dimensions were recorded and the loose material was removed from
the bottom of the cores. The cores were relatively thin (3.5-in approx.) and indicated that
the pavement was laid in two lifts. Inventory data further indicated both courses were
constructed of the same mix. Therefore, alter trimming, no additional specific gravity
tests were necessary, as the cores in their entirety were a convenient size for testing. The
specimens were then tested in ITCLT and ITST modes. To examine the degree of
segregation present, the cores were subjected to incineration and both the asphalt content

and aggregate gradation were measured.

89

6.5 M-lS, Test Section 1, Test Site 1 and 2 (30153-11 and 30202-12)

This rubblized project was investigated on November 14, 2001. The section was
located on M-lS southbound approximately ‘/2 mile north of the village of Otisville. The
pavement itself consisted of two lanes (one lane in each direction) with 3-tt asphalt
shoulders. The pavement is located on relatively flat terrain with low embankments
beyond the drainage ditches on either side of the road. This section consisted of only one
test site. The standard layout for FWD tests and coring was used for this test section. The
predominant distress was longitudinal cracking. Two relatively straight, medium to high
severity cracks appear on the center of lane side of each wheel path. Low to Medium
severity segregation was also noted on the pavement surface. Finally, occasional short
transverse cracks were noted throughout the pavement surface. Forty four FWD tests
were conducted on the site and 16 cores obtained.

The laboratory testing proceeded as outlined in section 3.5. The cores were
cleaned, the physical dimensions were recorded and the loose material was removed from

the bottom of the cores. The specimens were then tested in ITCLT and ITST modes.

6.6 M-20, Test Section 1, Test Sites 1 and 2 (30202-11 and 30202-12)

M-20 in Mt. Pleasant is a 5 lane (2 lanes in each direction, 1 turning lane) road
with 12-ft lanes and a 9-ft shoulder. The road is constructed in a flat area of alternating
forest and grassland. In early June, 2001, longitudinal cracks were observed in the earliest
stages of initiation on the inside edge of the outer wheelpath. Based on the observed

distress, field investigation was conducted on June 6, 2001.

90

Once preliminary coring indicated that the cracks were top-down, the test layout
was set. The test site used differed from what was used on other sites because the
standard layout was still being designed. The layout used on M-20 is shown below on
Figure 3.10. The only distress on the pavement was cracking and rutting, and no
segregation was observed. The maximum rut measured was 0.25-in, and the cracking had
appeared only faintly. A total of 59 FWD tests were conducted and 20 cores extracted
with 30 FWD tests and 11 cores obtained from site 1 and the remainder from site 2.

Due to the apparent young age of the cracks, M-20 was identified as a candidate
site for crack propagation measurements. Approximately each month alter testing, crack
surveys were conducted on both sites. In these surveys, it was assumed that the cracks
appear at a constant transverse distance across the pavement, and therefore only the
growth in the longitudinal cracks was measured.

Laboratory investigation including specific gravity tests, ITCLT and ITST were
conducted according to the process described in section 3.5. Additionally, select cores

were later incinerated to measure the asphalt content and aggregate gradation.

6.7 M-20, Test Section 2, Test Sites 1 and 2 (30202-21 and 30202-22)

M-20 east of Oil City is a 4 lane (2 lanes in each direction) road with 12-ft lanes
and a 9-ft shoulder much like sites 30202-11 and 30202-12. The pavement structure
consists of a rubblized concrete slab with asphalt overlay, however, and the road is
constructed in a flat, wooded area. This pavement was identified as potentially exhibiting
TDC in a periodic survey of all the rubblized projects in Michigan. The cracks were

observed in their early stages, and therefore this pavement was selected for crack

91

Road Shoulder Edge
Centerline of Traffic Lane

 

 

 

X
28 06
X
(— 3 —) {—— 3 —)
27 05
X
A
26 21 18 15 12 09 04
1 X
3.05-m
l >25< 03
24 02
X
Traffic
23 20 17 14 11 08 01 Direction

4145

Figure 3.10 Modified test site layout of M-20, test section 1, test sites 1 and 2 (30202-11
and 12)

 

 

92

mapping only. No field investigation other than crack measurement has been conducted,

and therefore the cracks have not been confirmed as TDC.

6.8 M42, Test Section 1, Test Site 1 (30422-11)

M-42 between Mesick and Manton is a flexible two-lane rural road with ll-tt
lanes and 3-ft asphalt shoulder. The general terrain of the area consists of rolling hills, but
the test section is located along a stretch of flat agricultural land. The road is placed
apparently at grade, with small earthen berrns on either side of the pavement the only
change in grade from the surrounding land to the road surface. The original survey of the
pavement (conducted during the summer of 2002) indicated short, random longitudinal
cracking appearing outside the wheelpaths. Upon returning to the site on November 7,
2002 for field investigation, a more detailed examination of the cracks revealed that they
formed a network within the wheelpaths closely resembling fatigue cracks. Subsequent
coring revealed however that the cracks were indeed TDC.

During the field investigation, only one site was placed on the test section due to a
coring restraint imposed by MDOT. Fifty four FWD tests were conducted and twenty six
4-in diameter cores were extracted. In addition to these activities, 54 nuclear density
readings were taken at points coinciding with the locations of the FWD tests. Finally, a
single borehole drilled using a hand-auger to a depth of four feet at the asphalt edge on

the shoulder indicated that the base consists of a 1-ft layer of clayey sand on silty sand.

93

6.9 M-SO, Test Section 1, Test Site 1 (30502-11)

M-50 west of Charlotte is a two-lane flexible pavement that was previously
investigated under a segregation project at Michigan State University conducted by
Chang (2000). The roadway is a flexible pavement experiencing heavy segregation. A
site survey by PRCE personnel indicated that longitudinal cracks present on the site
warranted further investigation, and on November 21, 2001 a single site was established
and tested. The road is located in a hilly agricultural area. The site itself is located on a
hill with the down slope in the direction of traffic.

A single longitudinal crack was noted in the inner wheelpath in the segregated
area. Additionally, transverse cracking was noted throughout the site. The coring on the
site confirmed that the cracks are TDC. Additionally, the cross-sections of the cores
indicate that the pavement may have been previously overlaid in an attempt to repair a
previously observed cracking problem. The original cracks were also TDC, as indicated
by cracks that occur only in the leveling course. For this site, 13 cores were extracted and
41 FWD tests were conducted. No nuclear density tests were conducted, but a single

borehole was drilled to a depth of l-tt to investigate the condition of the base.

6.10 M-SO, Test Section 2, Test Site 1 (30502-21)

M-50 west of Charlotte is a two-lane flexible pavement that was identified in
November, 2001 during the field investigation of M-50, section 1. Similar to section 1,
the roadway is a flexible pavement but is experiencing medium to heavy segregation strip

segregation close to the outer wheel path. The pavement is located in flat farmland on an

94

embankment elevated approximately 2—ft above the surrounding terrain. Standing water
was noted in the fields on either side of the pavement. The pavement was investigated on
May 29, 2002 when a single site was established and tested.

The predominant distress on the pavement consisted of longitudinal cracks in the
segregated area. Short transverse cracks were also noted at regular intervals at the center
of the lane. In addition to regular FWD testing and coring, nuclear density measurements
were conducted in visually identified segregated and non-segregated areas. Ultimately, 48
FWD tests were conducted and 38 cores extracted. A total of 61 nuclear density

measurements were conducted.

6.11 M-52, Test Section 1, Test Site 1 (30521-11)

M-52 south of Perry is a flexible two-lane rural road with 1 1-ft lanes and 3-ft
asphalt shoulder. The general terrain of the area consists of relatively flat agricultural
land alternating with wooded areas. The road is placed on a fill area, which according to
historical data was the location of a gravel road. The construction of the original asphalt
road entailed simply overlaying the existing gravel road. The original survey of the
pavement (conducted during the fall of 2002) indicated a developed network of
longitudinal and transverse cracks. The gradation of surface course appears to be a fine
mix, and no segregation was apparent.

During the field investigation, only one site was placed on the test section due to a
coring restraint imposed by MDOT. 44 FWD tests were conducted and 25 4-in diameter
cores were extracted. In addition to these activities, 48 nuclear density readings were

taken at points coinciding with the locations of the FWD tests. Finally, a single borehole

95

drilled using a hand—auger to a depth of three feet indicated that the base consisted of a
mix of clay, sand and gravel indicative of road gravel. The coring was halted at a depth of
3 where a stiff layer was encountered. The composition of the stiff layer could not be

established, but it is likely the original surface of the gravel road.

6.12 M-52, Test Section 2, Test Site 1 (30521-21)

This section of road is situated several miles to the north of 30521-21 in an area of
similar terrain. The pavement lane and shoulder width are the same as before. The only
significant difference between the two sites is that the surface course is made of a much
coarser mix and both segregation and raveling are apparent at the lane centerline. A
single overbanded longitudinal crack is plainly visible throughout the section slightly to
the outside of the lane centerline, but upon closer inspection, other faint longitudinal
cracks are visible at the outside edge of the inner wheelpath.

During the field investigation, only one site was placed on the test section due to a
coring restraint imposed by MDOT. 41 FWD tests were conducted and 10 4-in diameter
cores were extracted. In addition to these activities, 41 nuclear density readings were
taken at points coinciding with the locations of the FWD tests. Finally, a single borehole

drilled using a hand-auger to a depth of four feet.

6.13 M-55, Test Section 1, Test Sites 1 and 2 (30552-11 and 30552-12)
M-SS west of Prudenville is a 4 lane (2 lanes in each direction) road with concrete
curb and gutter. The surrounding is generally flat, and sloping to the shores of Houghton

Lake. Inventory data indicates that the natural soil is sand, which was confirmed by hand-

96

auguring a borehole. MDOT personnel originally suggested that M-55 could possibly be
experiencing TDC, and a subsequent survey by PRCE indicated that TDC were very
likely at the site. On August 20, 2001, the field investigation was conducted on the east
bound lane outer lane. The test section was selected at Iroquois Avenue just west of
Prudenville.

Longitudinal cracking was noted on the site, but in places did not follow the
typical TDC pattern of forming at the centerline edges of the wheelpaths. Here, the cracks
propagated specifically through medium segregated areas. Transverse cracking was also
noted. TDC was confirmed by coring, and temperature profile data was also collected for
the pavement. Additionally, the crack length on M-55 was also used in the crack
propagation model, and therefore, monthly surveys were conducted to map the extent of
the cracking.

Due to the curb and gutter, the FWD could not obtain the first row of tests.
Therefore, sites 1 and 2 were modified as shown in Figures 3.11 and 3.12, respectively.
Ultimately, 33 and 37 FWD tests were conducted on sites 1 and 2, respectively, and 15
and 14 cores were drilled from sites 1 and 2, respectively.

The laboratory testing proceeded as outlined in section 3.5. The cores were
cleaned, the physical dimensions were recorded and the loose material was removed from
the bottom of the cores. As before with test sites on US-4l (20411-11 and 12), the cores
were relatively thin (3.0-in approx.) and indicated that the pavement was laid in two lifts.
Inventory data further indicated both courses were constructed of the same mix (4B).
Therefore, after trimming, no additional specific gravity tests were necessary, as the cores

in their entirety were a convenient size for testing. The specimens were then tested in

97

Road Concrete Curb
Centerline and Gutter

30 26 22 18

1 ><><§E§><®><O

 

 

3.05-m
39
I X
38

 

A
36
35
34 28 24 20 16 12 04

Traffic
33 Direction

 

 

Figure 3.11 Modified test site layout of M-55, test section 1, test site 1 (30552-11)

98

Road Concrete Curb

 

 

Centerline and Gutter
36 26 22
\/ K71 \/
i X {X} /\ XX /\
3.05-m 35 09
i
X X
34 08

>< ><
21215102312121

 

A
32 06
X X
31 05
X X
30 24 20 12 04
X®X®X®
29 03 Traffic
Direction
X X
28 02
X X
3’ <———— 4’ ——>1

Fixed —27 23 19 15 11 01
X X®X®

1’

 

 

Figure 3.12 Modified test site layout of M-55, test section 1, test site 2 (30552-12)

99

ITCLT and ITST modes. To examine the degree of segregation present, the cores were
subjected to incineration and both the asphalt content and aggregate gradation were

measured.

6.14 Marsh Road

Marsh road in Okemos is a four lane county road (two lanes in each direction).
The topography consists of primarily flat land with occasional low hills. The test section
was located on the side of one of these hills. The pavement is a flexible pavement that
had been repeatedly overlaid. This pavement was identified by PRCE personnel as a
candidate TDC project, and therefore field investigation was conducted over two days;
site 1 was tested on October 11, 2000 and site 2 was tested on October 27, 2000. The
predominant distresses on the pavement were longitudinal cracking near the center of the
lane. Several parallel longitudinal cracks were observed in an area of medium to heavy
segregation. Transverse cracking has also appeared on the pavement. Therefore, the
cracking was rated as phase II TDC.

This site was test prior to the development of the standard testing layout, and
therefore followed a different testing pattem. The test layout consisted of two parallel
lines arranged in the longitudinal direction along the edges of the wheelpaths. Two
schematics showing the testing layout on Marsh Road Sites 1 and 2 are shown below in
Figures 3.13 and 3.14. A total of 58 FWD tests were conducted and 8 cores obtained,
with 30 FWD tests conducted at site 1 and 28 tests conducted at site 2. 4 cores were

obtained from each site.

100

Road Yellow Line
Centerline

 

 

Traffic

>< Direction

 

 

Figure 3.13 Test site layout of Marsh Road, test site 1

101

Road Yellow Line
Centerline

 

#
J

 

 

27 67
X X
A
26 66
X X
25 65
X X
24 64
X X
23 63 Traffic
Direction
X X
22 62

‘—— 2.5’ —’

82 61 83

$§§xxx

41—

Figure 3.14 Test site layout of Marsh Road, test site 2

 

 

102

As with the field investigation, this pavement was the first subjected to laboratory
investigation and the laboratory procedure had yet to be established. The investigation
procedures for this pavement was the model on which the test procedures were based
however. The cores were cleaned, the physical dimensions were recorded and the loose
material was removed from the bottom of the cores. The cores were cut into test
specimens after conducting specific gravity tests on the cores. Specific gravity tests were
then repeated on the test specimens. The specimens were then tested in ITCLT and ITST
modes. Although medium to heavy segregation was observed in the pavement, the test

specimens were not incinerated.

103

1.0

1.

2.

3.

CHAPTER 4

DATA ANALYSIS

GENERAL

As stated in chapter 1, the objectives of this study include:

Identify the causes of TDC and the factors affecting top-down cracking potential.
Compare top-down and bottom-up cracking potentials.

Develop a model to describe the rate of TDC propagation.

To accomplish these objectives, lab and field investigations were conducted and various

types of data were collected as stated in chapter 3. These data include:

1.

2.

6.

Distress data that were used to develop TDC propagation models.

NDT data that were used to backcalculate resilient modulus of the pavement
layers.

Field and lab measured density data that were used to confirm the presence of
segregation.

Indirect tensile cyclic load test data that were used to determine the moduli of the
asphalt courses.

Indirect tensile strength test data to determine the relative fatigue performance of
the asphalt concrete courses.

Mechanistic analysis to determine load induced surface tensile stresses.

This chapter details the analyses used to determine the factors affecting TDC

potential. During the analysis, each factor was studied in reference to its variation along

and across the pavement. Further, the value of each parameter was compared to that

104

obtained from two control test sites that have exhibited no distress. Finally, in the
analyses of the data, the strength and resilient moduli values of the segregated specimens

were compared directly to those of the non-segregated specimens.

2.0 DISTRESS DATA

Unfortunately, the historical pavement surface distress data of MDOT and other
state highway agencies treat surface cracks in the same fashion regardless if the cracks
are bottom-up or top-down. The reason is that no crack identification procedure has been
developed whereby cracks can be identified as TDC. Such a procedure was developed
during this study and is presented in this chapter. Nevertheless, the historical MDOT
distress data cannot be used to determine the rate of TDC propagation. To alleviate the
problem, periodical manual distress surveys were conducted.

Various distress surveys were conducted on all test sections. During the surveys,
segregated areas were noted and mapped along with their degrees of segregation and, on
several test sections, crack measurements were made as to determine the crack
propagation rates. Finally, to increase the time frame of the distress data, the detailed

distress data collected by Chang (2000) were included in this study.

2.1 Distress

On all the projects visited, the distresses present, including segregation, were
cataloged. Additionally, in a parallel study investigating the causes of TDC in rubblized
pavements conducted by Svasdisant (2003), bi-annual surveys of every rubblized project

in Michigan were conducted. The observed distresses for each flexible pavement

105

investigated in this project are listed in Table 4.1. The observed distresses of the
rubblized pavement survey conducted by Svasdisant are provided in Table 4.2. It should
be noted that the pavements listed in Table 4.2 are only those that are strongly suspected

to be experiencing TDC, since the TDC have not be confirmed by coring.

2.3 Crack Measurement
During the surveys of test sites, digital images were obtained, all cracks were
mapped, and, for each crack, the coordinates of the crack tips were measured relative to
the test site grid. These coordinates were then used to calculate the cumulative lengths of
all transverse and longitudinal cracks. Table 4.3 provides a list of the time history of the
cumulative lengths of the longitudinal and the transverse cracks. Figures 4.1 through 4.3
depict the growth in the cumulative lengths of the longitudinal and transverse cracks.
Finally, Figures 4.4 to 4.13 show actual photographs of the test sites.
It is important to note that most TDC tend to initiate as longitudinal cracks and,
over time, they develop according to the following three phases (Svasdisant et al, 2001):
1. Phase 1: Short, longitudinal cracks outside of the wheelpath.
2. Phase 2: The original longitudinal cracks grow longer and sister cracks form.
3. Phase 3: Both the original longitudinal cracks and sister cracks have developed
into extensive longitudinal cracks connected by short transverse cracks.
Each pavement can be categorized into one of the above three phases of crack
development. For example, the TDC on M-20 are typical of phase 1 cracking (see Figure
4.5), the TDC on Marsh Road show phase 2 cracking (see Figure 4.13), and the TDC on

M-37 have evolved to phase 3 cracking (see Figure 4.6). It is important to note that the

106

x098 8555.: SEme H OHM .wczgmm u M £88 8.555: “52.
0,: .coumwohwom n m .8380 5.3—ode omho>mcfih H on: .8380 ESPN—2 Ecaziwcoq H 09,—:

 

 

 

 

 

 

 

 

 

 

 

 

c N N_ S N 2 m NN 5.;
— $95me OZ
_ _ _ _ N 05:85...ng
N N N N N $55.53:
N N N N 4 Session:
_ a _ a BéofitoaS
N N N N NONEEQS
N N N @065
a _ a 05.58..—
N N 095
£8..on
am a m on: 8.x on: o: NO NZ 25 mafia

 

 

 

 

 

 

 

 

 

85833 0338c :o Btomno 8.0.8me :N 2an

107

 

a 91m om m

 

N856 NOON

Ex 3”— In.

m
2:

w

  

_
_
H _

UA< U]; DAEM— UHM UPC. 0,:

83820 $266 @2865 2:

mo 02

0&0 mmobma

    

No 3:5: 2:. Ne 2N3

108

a gem om m

 

seas NOON

EM 2 In L M
2:

  

 

w

@3530 $266 @8865 05

M
_

UA< Uta 09—h OHM DD: 0.:

 

$8.85 .3 e Z

m0 528:: Bit

09C 305.5
mi 2an

     

109

a: €on u 97m can .wEv—ofio M85 H Um .mxuflu SEE—Z H DA< .3635 H m .mowmmoaon H D
JEN: nwzoM u a .2050; n In 585 n M Jumbo $59598— :N—awoM u DAM Jon; 8855b 833M H D .HM .w::o>~M n M
3555b D .E. .cocumEMUN H 8555; H DD: H DC 1:

    
     

 

110

_

_

_ M
_ M M

Q 97m Dm m SM 2 EM M M w DA< DAM DORA DOM DO: Dr: 883a,? oZ 2:38.55

N356 Noam 05 NV 033.

   
 

33030 $236 38065 9: Mo c3832

Table 4.3 Summary of crack measurements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Route Test site Survey date Cracklength (,fi) ,
Transverse Longitudinal
8/21/2001 0.0 9.3
9/30/2001 0.0 11.0
12/7/2001 0.0 13.8
M-20 3020211 2/9/2002 0.0 42.6
3/17/2002 0.0 58.1
6/29/2002 0.0 41.7
8/30/2002 0.0 44.3
10/7/2002 0.0 41.6
8/21/2001 0.0 31.0
9/30/2001 0.0 40.4
12/7/2001 0.0 48.9
M-20 30202_12 2/9/2002 0.0 57.]
3/17/2002 0.0 58.7
6/29/2002 0.0 60.0
8/30/2002 0.0 60.0
10/7/2002 0.0 58.8
3/17/2002 0.0 61.3
M-55 30552_11 6/29/2002 0.0 118.7
8/30/2002 0.0 98.1
10/11/2002 0.0 99.3
3/17/2002 0.0 116.1
M-55 30552_12 6/29/2002 0.0 95.9
8/30/2002 13.8 108.9
10/11/2002 13.8 115.1
6/29/2002 11.1 31.2
M-20 30204—21 8/28/2002 18.2 38.2
10/7/2002 25.0 42.8
6/29/2002 1.3 2.2
M-20 30204-22 8/28/2002 4.0 4.6
10/7/2002 7.7 4.6

 

111

 

 

 

Longitudinal Crack Length (ft)

     

‘ C I 0
l
' ' ' l
O O '0
9 ‘b I 0 ¢
1’ z
9 _
6 I ‘
, .
. i
. . 1 . . . , . 1 . . ,
‘ 1
. . « . . - : ; . ; - y +Long1tud1nal Crack
7 . . ‘ a t v . I ~ ‘ ~1-., . _1_ . .. .
I- . l * . . . : 1 -

15:95 L:1:Tifj“ ”.17
100 200 300 400 500
Time After First Crack Measurement (days)

 

(a) Longitudinal TDC on test site 1 (30202-11)

 

 

Longitudinal Crack Length (ft)

HNMhU’IO‘N
OOOOOOO

O

 

 

l ' 1 v '
i f t t o -— Q o 2 . _
a 9 Q I '70 O
1
. V o o ' ’ 9

O O O " T I
Y c v o f 7 1
l . ' ' t

‘ o + — O - 1 ~ I t I e

1 .- ‘ . — ' 1 1— 1

——f I v Q «v—ol c o l

_._ . l . . . 1 .

’ L l . ; O t O- . l

7,? o l 9 .
~ ... i - . l» — ‘ . -q
. ; . l I. . j

|
7",” o . , t t i f

L . L 1 o 1
v — f — ‘
,; a l
1 —. 1 .'

V I O o I
1 .
f , 7. . _ . . 7.7 . . . . .-.
O 0 O

 

. .

f f f 3 f } i F {-7 E {g} 33+LongitudinalCracks
I "I 1.; 1 '7‘

 

1 1 -— ..-, ..- 774 ._,
1 , . . , . l . . . ‘. . l t 1 , ...; x _+._, ,
1 1 j ‘ .
. . . , . . . . t . . . O o n . o A i I .
. ;m an.-.
Nfi . u . . . . . . , . t 2 o . :2 M. 8 . Q . .-.7 . .
i
T' Y

100 200 300 400 500
Time After First Crack Measurement (days)

 

(b) Longitudinal TDC on test site 2 (30202-12)

Figure 4.1 Growth of cracking on M-20, test section 1

112

 

 

 

 

  
  

\ 1‘1

2.: 5.139.—

\ilt

\\

42:; 2:53:33.—

 

 

140.0

 

 

 

 

 

   

 

 

55: 120.0 7 5
a .1 ~
0) .2 .
._1 1
£5 80.0 If 5
<3 I . .1
5,3,; 60.0 3 , 1 ‘1
.E 3 .‘ ., ,. . ,1
B 40.0 1 1 1% ;- % i % E + H .»
.520 771 , 1 , 1 . , #1 , v . 1
c. :* 1i. * : 1 . ,; s. 1 : :
,3 20.0 - f i -1, . ,: . : ;:. ‘ . .
L 1 : . 1- :, T . 1 +Long1tud1nalTDC _’
0.0 ~ ~ 1 ' -1 1, : : :* , 1‘:‘7::j::r:: 11I
0.0 50.0 100.0 150.0 200.0 250.0
Time After First Crack Measurement (days)
(a) TDC growth on test site 1 (30552-11)

140.0 ;j ; 16-0
@1200 1 ‘ 1405;
.43 117120757
@1000 11:1 ~ on
:1 ~ s
g 80.0 j“: j %
5 7 ;i 'l 8-0 5
:g t : 1....___T i T 60 if);
a 40.0 . .. :7" E 9%? I7 31
.& v ‘ Y" i r ,,1 frif‘fx, , , T 4'0 g
g 20.0 . : 1 1 +1.0ng1tud1nal TDC 2.0 g

i 1 : 7 . ;1. fiTransversewc
0'0 1'3 ‘ ' "fl” j ; ,,:.,,,'1’"1ifjflff*1fil:ijj'jfi: E 0.0
0.0 50.0 100.0 150.0 200.0 250.0

Time After First Crack Measurement (days)

 

(b) TDC growth on test site 2 (30552-12)

Figure 4.2 Growth of cracking on M-55, test section 1

113

 

 

      

  
    

; ... \

 

 

4...... 11% «a... .
MEL

«tx. 5. fil— ‘l 11.
2,: 23:9.— quD __::7::w::._ 21:9; «2:. V 2.2.5::

 

 

O

 

o .
O Q C
. . A .
I. O O 9 F:
. o O .
Q 9 v a 7. ~....
10 O 4 T o 1 T 6 q 1
. o 0 V e 1 y—lt
. . . .. 1 1 s 1 T ll
0 o . t >~ ¢ 7. .
I O a
.
.

O
t
i +—

40 60

l . . .
. . . 1 ,1
I
. u 077-
» I L - a
f l A ‘
A O 9
o o
‘ I
. o .
Q 9 9
. O o
v 0
9 o 0 1
. T .
, , _ 20.0
1 o
. .7
. 7.;
1
. sf
' ‘ * 150
l I
t 67° - .
I - "9’ I
l I
o I .2..+ i
. . . TN |
. —£ 1
1 1 . . .
. i .— .
~ 1 100
t v v 4--‘¢ .
. .. . . b L_ Q 4
. . l- l
_ -——O—.:'_‘———l—— L._-. ##L—ut.-¢_—~;—;—

+ Longitudinal TDC ,

-D— Transverse TDC

¥»-1 T 77", l_:I—- 2 . I.‘
1 , 4, .1 . “My“ . .
‘ LO, .1. ...l. . O..- 4 .Ol

80 100

Time After First Crack Measurement (days)

Transverse Crack Length (ft)

 

(a) TDC growth on test

site 1 (30204—21)

 

 

Longitudinal Crack Length

 

   

 

: 9.0

 

i; 5 g f 8.0

7.0

1 6.0

5.0

. : l ..1 4'0

1.5 i 1 i‘ 7 1 l 11 1 3.1;; 3'0
1.0 . §T_ +Long1tud1nal TDC ‘1 2-0
0-5 +,—~. ' 1 , ‘ , f, * , . l -D—TransverseTDC 1 1-0
0.0 1 1 ' l ' : ‘ ‘ Tiff '3-::;; 1 1:3:1:-1:«, 0.0

0 20 40 60
Time After First Crack

80 100
Measurement (days)

Transverse Crack Length (ft)

 

(b) TDC growth on test

site 2 (3020422)

Figure 4.3 Growth of cracking on M-20, test section 2

114

 

 

.I Tu...

. 3.3.3..

1~l wk

13‘111,

.. ., . 1 .1....:..!111.1l

 

   

 

Figure 4.4 TDC and raveling in heavily segregated areas on US-23, May 2002, 2 years
after construction (rubblized pavement)

Figure 4.5 TDC inoitiatin at the edge of the wheel path on M-20 (flexible pavement, no
segregation is visible)

115

 

L 'wc-u-u. w *» '-

 

    

1 ~ ~ ~

Figure .7 TDo M-42 (flexible pavement, no segregation visible)

116

  

‘ '3‘-

Figure 4.8 TDC in segregated areas on M-50 test section 1 (flex

 

117

 

.4
i

inn—r1.

e4

Figur

\r

1 I . .
. — r.
4 ...:u... k 3::”? .

Flgurc

        

1

Figure 4.1 1 TDC in segregate

d area on -5 test section 2 (flex1

 

bl paveemnt)

118

   

 

, L 1}“:m‘... - V . . ., W . . ' “ ' m.
Figure 4.12 TDC in segregated areas at the edge of the wheel path on M-SS (flexible
pavement)

.. .tflff'f‘. - (E:

Figure 4.13 TDC in heavily segregated areas oMrsh Road (flxible peavment)

119

three Phases 0f TD
is prediction mOd‘

Pai'ement section 1

2,4 Segregatio
During the
areas were noted.
section 1, test sites
data base. crack m
were included. In
TDC by coring. Ta

Chang and utilized

2.5 Other Dist
During the
recorded. These int

reflective cracks. l

3'0 TDC PROI

It can be set
TD
C are associated

related to the degr
. k.

Chang (2000), the C

 

three phases of TDC propagation can be used to develop crack propagation models. That
is prediction models can be developed to address the required time for TDC in a

pavement section to develop from one phase to another.

2.4 Segregation

During the site investigation phase of this study, for each test site, segregated
areas were noted. On three test sites (M-SO test section 1, test site 1 and for M-55 test
section 1, test sites 1 and 2) detailed segregation maps were developed. To expand the
data base, crack measurement data obtained by Chang (2000) in segregated pavements
were included. In this study, the cracks that were measured by Chang were confirmed as
TDC by coring. Tables 4.4 through 4.6 show the crack and segregation data collected by

Chang and utilized in this study.

2.5 Other Distresses
During the distress surveys, other types of distress were also observed and
recorded. These include segregation, rut, raveling, and, for rubblized pavements,

reflective cracks. These distresses are summarized in Tables 4.1 and 4.2.

3.0 TDC PROPAGATION MODEL

It can be seen from Tables 4.1 and 4.2 that for 53 out of 84 pavement projects, the
TDC are associated with segregation. Hence, the initiation and propagation of TDC is
related to the degree to which the pavement is segregated. In this study, as well as in

Chang (2000), the cumulative crack length in areas with certain degree of segregation

120

 

 

ill :.1 . L . 4 I a i _ . . _ L :2 257.2: a: . . c
— 2,: may? _5_:..3._,..3m._c 533.. ~ .. b LI
Acccmv wzssb :5: m2? :c._:..u4.:wov. we 93:.

 

 

 

 

 

 

 

 

 

 

 

 

m : a w o 82 .33 «.8 m:
a o o o o 3% .23 com m:
m o o o o 32 .23 New m2
.2 o o o o 32 .aom Tom m2
a o v o 8 82 a: 2.89% 8-2
N o a o S 82 .52 E on:
o o mm o o 82 .32 E mam:
. 3 wages
as c on o o 32 33 _ 026m 2%
beam 32 5282 samba 3w: 2% ages:
-53on -E 5 setup—H800 cam 33%

 

 

 

A5 938 wowmwopwom mo awed;

 

 

 

 

 

888 mango Bob 85 :ocmwoewom we 2an

121

 

She

Sunej
date

 

Feb.lV‘
313}: 19
(Mn.1us

: Aprd IQ

‘ June.l9

\Iarch.2‘

 

16

l4

30502-
17

G

204

 

July. 1w

Aprd.l§

June.lq
N.
Klareh.2t

K

hiav.lw
bx
(Jet 19\
T“—f‘--
Aprfl.ls
x
June.19

T“‘—~~EEE£E;£

Feb~ 19¢
\
\Ia\1 19
K
OCT, 19\
w
m
March. 3
0C1, 191
JUne‘ 19
March a
\
(Der ]Qt
April, 1‘;
*J\
Ime. 1t)
.‘VIafch‘ 2|
0C1, ]()t
April. lg
J““€.l9
()Ct.1k)<
JUne.]o

//

 

Xiarch. ”t

W

Table 4.5 Length of cracks in segregated areas

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Surface Crack Length (11)
. Survey
Site date age _ M d' N
(Months) Light nght‘ Medium e tum- Heavy on-
med1um heavy Seg
Feb, 1998 31 0 0 0 0 0 0
May, 1998 33 0 0 0 0 0 0
1 Oct, 1998 39 0 0 0 0 0 0
April, 1999 45 0 0 0 0 0 0
June, 1999 47 0 0 0 0 0 0
March, 2000 0.001 0 0 0 0 28 0
July, 1998 72 0 0 75.5 0 0 24.5
16 April, 1999 81 0 0 75.5 0 0 3]
June, 1999 83 0 0 80 0 0 41
March, 2000 91 0 0 80 0 0 48
Feb, 1998 10 0 0 0 0 0 0
May, 1998 12 0 0 0 0 0 O
14 Oct, 1998 I7 0 0 0 0 O 0
April, 1999 23 7 0 12 0 O 0
June, 1999 25 7 0 12 0 0 0
March, 2000 34 8 0 12 0 0 0
Feb, 1998 10 10 0 0 0 3 2.5
May, 1998 12 13 0 0 0 3 2.5
30502- Oct, 1998 17 13 0 0 0 3 2.5
12 April, 1999 23 16 0 0 0 4 8.5
June, 1999 25 16 0 0 0 4 15.5
March, 2000 34 25 0 0 0 4 29
Oct, 1998 13 O 0 0 0 0 0
204 April, 1999 19 0 0 0 0 0 0
June, 1999 21 0 O 0 0 O 0
March, 2000 30 O O 0 0 I 1 0
Oct, 1998 13 O O 0 0 O 0
20_2 April, 1999 19 0 0 0 0 0 0
June, 1999 21 0 0 0 0 0 0
March, 2000 30 0 0 0 0 2 8
Oct, 1998 13 0 0 0 0 0 0
203 April, 1999 I9 0 0 0 0 6 4
June, 1999 21 0 0 0 0 6 4
March, 2000 30 0 0 0 0 7 4
Oct, 1998 13 O 0 0 0 0 0
204 April, 1999 19 0 4 5 8 7 0
June, 1999 21 0 4 5 8 7 0
March, 2000 30 0 7 7 11 7 O

 

 

 

 

 

 

 

 

 

122

 

Table

”’-

She

4.6 Perc

Sune).

 

Feb.l“
NIay.I“
()CL 1“
.Aan h

 

 

 

 

 

June.l“
Klarch.3

 

 

20-1

20.3

204

 

March. 31
\
March. a,
\;
A Ti] t't
\PQ

Juhzlh
Aprfl.l‘
()CL Mr
K
Aprd.1u
\
June. IV
NIaI'Ch. :1
\
Feb.10\
\
.\'Ia\'_ 19:.
.___4___~
(Jet 101

June,1o.
0C1. IQW
JUne,1ot
0C1. 19”
JUne. ]t)t
Octloq
AprlI. ]t)t
111119 199
xlarch.2L'
()CL 199

APTII, 19,,
JUne‘ 199

 

4N

 

 

 

Table 4.6 Percent crack measurements on segregation projects gfrom Chang, 2000)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Surface Crack Length in Segregated Areas (%)
Site Survey date Age ‘ _
(Months) Light Light— Medium Medium Heavy Non-
medium ~heaw Seg
Feb, 1998 31 O 0 0 0 0 0
May, 1998 33 0 0 0 0 O 0
1 Oct, 1998 39 0 0 0 0 0 0
April, 1999 45 0 0 0 0 0 0
June, 1999 47 0 0 O 0 O 0
March, 2000 0.001 0 0 0 0 0 0
July, 1998 72 0 0 85.8 0 0 0
16 April, 1999 81 0 0 85.8 0 0 0
June, 1999 83 0 0 90.9 0 0 0
March, 2000 91 0 0 90.9 0 0 0
Feb, 1998 10 O 0 0 0 0 0
May, 1998 12 0 0 0 0 0 0
14 Oct, 1998 17 0 0 0 0 0 0
April, 1999 23 12.3 0 57.1 0 O 0
June, 1999 25 12.3 0 57.1 0 0 0
March, 2000 34 14 0 57.1 0 0 0
Feb, 1998 10 16.7 0 0 0 75 0
May, 1998 12 21.7 0 O 0 75 0
3050 Oct, 1998 17 21.7 0 0 0 75 0
2-12 April, 1999 23 26.7 0 0 0 100 0
June, 1999 25 26.7 0 0 0 0 0
March, 2000 34 41.7 0 0 0 0 0
Oct, 1998 13 0 0 O 0 0 0
20_1 April, 1999 19 O 0 0 0 0 0
June, 1999 21 0 0 0 0 0 0
March, 2000 30 O 0 0 0 91.7 0
Oct, 1998 13 0 0 0 0 0 0
20_2 April, 1999 19 0 0 0 0 0 0
June, 1999 21 0 0 0 0 0 0
March, 2000 30 0 0 0 0 100 0
Oct, 1998 13 0 0 0 0 0 O
20_3 April, 1999 19 0 0 0 0 85.7 0
June, 1999 21 0 O 0 0 85.7 0
March, 2000 30 0 0 0 0 100 0
Oct, 1998 13 0 0 0 O O 0
204 April, 1999 19 0 50 55.6 72.7 87.5 0
June, 1999 21 0 50 55.6 72.7 87.5 0
March, 2000 30 0 87.5 77.8 100 87.5 0

 

 

 

 

 

 

 

 

 

123

 

was measured 1'

percent of the c

\\here: 1

Based on the Pt

propagation me

1. Each se
ofthe St
2. Each 111.
dEgree (
either it:
overlay,
3. For eae}
equatior
Where:
Example
F"
igure 4_14 m
de‘c’me 0 ~

was measured for each test site. The length of each crack was later calculated as a

percent of the cumulative length of the segregated areas using the following equation:

 

- CDS
Percent Cracking = PC = x 100% 4.1
L
DS
Where: CDS = cumulative crack length in segregated areas (ft); and

LDS = length of segregated areas (ft).
Based on the PC data, the following three-tiered approach was taken to develop crack
propagation models:

1. Each segregated area was assigned a degree of segregation based on the severity
of the segregation, as described in chapter 3, section 4.1.

2. Each measurement of percent cracking was then graphed against time for each
degree of segregation. The time scale is based on the surface age of the pavement,
either the time since original construction, or the time since the most recent
overlay.

3. For each degree of segregation, the best fit logarithmic growth equation (see
equation 4.2) was obtained.

PC = b + (a)ln(t) 4.2
Where: PC = percent cracking;
a, b = regression coefficients; and
t = time (months).
Examples of the data for crack length in each degree of segregation are shown in
Figure 4.14. Once the best fit logarithmic growth equation had been obtained for each

degree of segregation, the regression coefficients “a” and “b” were plotted against a

124

.0 0 0 0 0 0 0 0 .0 .,0 .0 0 ...O 0 0 ..O ,30
0 0 0 0 0 0 . 0 O . 0 0 0... 0 m: ml. 0 0,. 0 ..
0 O .0 ._..U 0. 0 U .U U .U. 0 0 rt. l. U 0 0
5 4 3 2 I U 8 6 4 2 0 oo 6 4 2
wet—2:0 E989... 9.3.33.5 E989; $2593... 232.5

Fr,

 

 

50%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 
 

 

 

 

 

 

 

 

 

g +
340% 3:.g: :j=y 0.0222Ln(x) 01481
12:, o - 3 i ;1 R=0.3969 . .
8”” -,~:.,.;Lui.;,
820%;  ;
0 :1 ’1 ,1 2 : , 2,:
E 10% ,, . f E, I
0% ' t *
20 40 60 80 100
Time (months)
(a) Light segregation
100°/ . . L. ,. .2.
... ° . z i new.
.5 80% , ; egg. 5;", - i3 I; 3:
ii 14 t 3 rig *3 ~ 1 '
<260% *‘J*~‘“ ~
0 jy= 0. 07322Log(x) + 0. 4495I
E 40% . .,
8 R2 =0.8131 1
E 20% t 2* a. ;— t4
0%V ‘ : ‘ ,', t :
20 40 60 80 100
Time (months)
(c) Medium segregation
100°/ 1%, , r W, ,. L-.. ,
o0 ° + ; e 1 t rill:
.5 80% * , 9 3+1qu t * ¢=IJ* e
l:l‘,t. Ii , 1,22,,1 L . L
g L , . . , g=y 0.092Log(x)+06043 I
“I 60% t ' ’ , 2
8 I ' . 3 : ‘ ; . R=0.9341 g—p:
g 40% I" E i I ; . , . :44" i -
8 E i ' . , e95 TI 1* 7-
am» ,
0% ' . j i . I : .
20 4O 60 80 100

Time (months)

 

(0) Heavy segregation

Figure 4.14 Growth of cracking for three degrees of segregation

125

 

 

 

numerical degree
degree of segreg
numerical values
the graph “here t
The linear equati.
into Equation 4.2

SCgregation:

\Ihere: D8

”161

 

 

To test the
equation 4.3 were 3
shown on Figure 4.
can be seen the pre
should be noted tha

t . -
mmtdiately aflCr C

the da

T .
his SeCtion

numerical degree of segregation. On the scale of numerical degree of segregation, heavy
degree of segregation was assigned a value of 10 whereas no segregation 0. The
numerical values for light and medium degrees of segregation were then obtained from
the graph where the regression coefficients fell on straight lines as shown in Figure 4.15a.
The linear equations describing the two straight lines were then obtained and substituted
into Equation 4.2. This yielded the following prediction equation for all degrees of
segregation:

PC = 0.0602DS+ 0.0093DSln(t) 4.3

Where: DS = numerical degree of segregation (no segregation = 0.0, light = 2.5,
medium = 7.5 and heavy = 10); and
t = time (months).

To test the accuracy of the prediction model, the percent cracking predicted
equation 4.3 were graphed against the measured percent cracking. The comparison is
shown on Figure 4.15b. The straight line in the figure depicts the locus of equality. As
can be seen the predicted percent cracking matches well with the measured ones. It
should be noted that the reliability of the prediction model is questionable for times
immediately after construction when no crack measurement data is available. However,
the data indicates that significant cracking will occur shortly after construction, especially

in medium and heavy degrees of segregation.
4.0 CLASSIFICATION AND IDENTIFICATION OF TOP-DOWN CRACKS

This section was developed in cooperation with Tunwin Svasdisant who studied

the underperformance of rubblized pavements. Part of the problem associated with TDC

126

08

0.6

0.0

80“ t.
60" .,
40° 0
209 o

wciuio Evian 33:33: .v

091,,

 

 

O 8 ..--,-.__.---.-._-~_._-...__......_.....- .. “--....“ .-....-,-.. ---,--._--._. ..._.,_,_... ---“--- -~._..2,_.-.-L...-. 2“,---” .--,

ARegression Coeff.v"a" if , 1 2,:
OR . C ff "b" i . bZO-O602X, .
06 i I: 2 'elgrlession. (i6... ; i.R2=0.9999l:71
. , _ . . - i . . . . . L . . , , l 3 7: ,..,,_-_,.,._..

    
   

 
    

l
i

0.4 ..
.-1’

Coefficeint value

. . . . . . . . . . . . , . . 1 .
i .
I I
. . 1 1
1
~7 0 7 . . 7o . - ,. ,. .2 . . ., . .7, . , .i . . 7.
u I ,
l
n A 4 q A u b o - r . A I b n w o I

l :2 I :,a=00.‘0093x

l
-..—-1-” ._-- -

 

 

 

, .

 

2 7R2 = 0.9959

 

 

 

0.0
O 2 4 6 8 10

Degree of segregation (DOS)

12

 

(a) Calibration of regression coefficients

 

 

A

on ‘7 . I, T
.e m 1.1 v.
'2‘) ‘ I . , 1 .2 ,1
cc *jII"'I't
h 2». . ,F, ..I .
U .. . I .,21 I. .,
H ‘ l 1 I

5 1‘1
U ‘.,,,. . 1 ’ 1

I— --t . t ,L

o ,

9* AHeavy it,
B
*5 OMedium 2.
— LI 1
:1 , .
2 oLight {gt
6 ' 37,: , r g ,,

. . .- . , , I...

 

0% 20% 40% 60% 80%

Measured percent cracking

l 00%

 

(b) Verification of prediction model

Figure 4.15 Calibration and verification of crack propagation model

127

 

 

is that PMS d3
databases do n
distress data ar

The ma
been developer

videotape sun'

1. lnapprr
TDCa
l hfino
rmnmr
3.TDC;
'ReCOg

10 deveIOp T]
mechanicS 0‘,
TD<

} Ind.

N

Obsen-e-€

Settignq

is that PMS databases make no distinction between TDC and other cracks. Since most

databases do not include data regarding crack locations, it is not possible to decipher the

distress data and to separate regular cracks from TDC.

The main reason for the lack of TDC data is that no identification procedure has

been developed whereby the TDC can be correctly documented either by manual or

videotape survey. The lack of such data creates various problems including:

1.

Inappropriate selection of rehabilitation alternatives; the rehabilitation Options for
TDC are different than those for regular cracks.

It is not possible to assess the impact of TDC on pavement conditions or on its
remaining service life.

TDC propagation models cannot be verified.

Recognizing the dimension of the problem, one of the objectives of this study is

to develop TDC identification procedure based on crack appearance, location and the

mechanics of crack propagation. Such procedure was developed and is presented below.

4.1

V7

V7

TDC Descriptions and Classification
TDC were observed in the field as:
Individual longitudinal or transverse cracks.
Groups of parallel longitudinal cracks.
Networks of longitudinal cracks.

Based on field observations and depending on the type and causes of the first

observed crack, TDC can be classified into three categories as presented in the next

sections.

128

4.1.1 Longitudinal Top-Down Cracks (LTDC) Category

The main cause of LTDC in this category is the induced tensile stress at
the pavement surface. Such tensile stress could be a combination of the tensile
stress due to tire-pavement interaction and thermal stress. The magnitude of the
tensile stress could be increased due to differential stiffness between the AC
courses and the AC layers and, for rubblized pavement only, the geometry of the
interface between the fractured concrete and the rubblized material layers. The
LTDC develop in three stages (Svasdisant et al, 2001) as follows:

1. In the first stage, a single short longitudinal crack develops in the vicinity
of the edge of the wheel path. Over time, the crack propagates parallel to
the wheel path as shown in Figures 4.16 and 4.17.

2. In the second stage, sister cracks develop and propagate parallel to the
wheel path as shown in Figure 4.18 or in some pavements, transverse TDC
may develop connecting to the existing longitudinal TDC as shown in
Figure 4.19.

3. In the third stage, two or more longitudinal TDC connect via transverse
TDC as shown in Figure 4.20.

Figure 4.21 depicts the three stages of TDC and some pavement cores showing
partial propagation of TDC. Since the 5 cores were extracted from pavements
with various AC thicknesses, it can be concluded that the initiation and

propagation of TDC is independent of the pavement thickness

129

Figu

 

01052 job number 32335 BMP

“L‘w‘

              

 

Figurev4.l7 Lgitudinl TDC at edge 0 the wheel on US-l 31, control section
83032 job number 34060 BMP 13.0 EMP 18.7, May 2002

I30

 

‘yJ \
‘1

. K
Figure 4.18 ;
LS-I3I ($3

.1‘“-’ a
.4,“ T-
. , .. I

r O
a 1 .
i‘ .
5 l‘;
l-' _'.
‘ t‘
"1 K t
.\' .‘I g
k’.. 1

E1“.

-.-—-—./.

\n(

9 I,

7

ENE 4.

5.4

 

Figure 4. 1 Pearalll longitudinal TDC at the edge of the inner and outer wheel paths on
US-l3l (same location as Figure 5.2), control section 83032 job number 34060 BMP
13.0 EMP 18.7, May 2003

“WWW“ . '
on M-27, control section
16302 job number 26672 BMP 9.3 EMP 11.4, May 2003

131

 

‘?

E

Figure 4.20 '
on East

 

 

 

, ‘1 . i ‘ fid'w‘ . . .' V . . . . , -
. I ‘ r;._ 4; we“ r»_ , ; » . o. .t
Figure 4.20 The third stage of TDC development, LTDC and TTDC forming a networks
on East Michigan Avenue, control section 39555 job number 39212, May 2002

 

showing the extent of cracking

132

4.1.2

low ten
cracks l
time, tli
The TT
the tens
Finally

becausl

4.1.2 Transverse Top-Down Cracks (TTDC) Category

The likely cause of TTDC is thermal stresses due to cold temperatures and
low tensile strength AC mixes. Most of the TTDC appear as short transverse
cracks located at the center of the lane as shown in Figures 4.22 and 4.23. Over
time, the cracks propagate across the lane and other transverse cracks develop.
The TTDC may or may not be connected with LTDC depending on the state of
the tensile stress at the pavement surface in the vicinity of the wheel path.
Finally, the main reason that most TTDC initiate in the middle of the lane is

because the crack in the wheel path is delayed due to the tires kneading action.

4.1.3 Crack in Segregated Areas Category

The cause of TDC in this category is particle segregation in the AC mat.
Particle segregation may cause up to 60 percent reduction in the tensile strength
of the AC mix in the segregated areas (Brown et al, Khedaywi and White).
Further, because of the high air voids in segregated areas, the AC mix is more
susceptible to oxidation and moisture damage resulting in substantial decreases in
the tensile strength of the mix. Depending on the shape of the segregated area, the
crack could be longitudinal or transverse, although most observed TDC in this
category are longitudinal. The TDC propagate within the segregated area at a
relatively fast rate, which decreases as the cracks begin to penetrate non-
segregated areas.

Another type of distress that associated with segregation and TDC in

segregated areas is raveling. Figure 4.24 shows raveling of the outer lane of a

133

.~ ..__,2\ _. urea...”..Ihir,..absrH..&K....-.L5..t.-

3

Figure 4.22 Tr

. Fatal:

....u .p.V4.. -

I ‘.«V(C

.. -..-”c.“ -... .. ...-flaw ._. ,.
L' .

 

M «r .. -..--.L ‘7' '1 mm
Figure 4.22 Transverse TDC located at the center of the lane (compare the srze to half of
a 4-in core)

 

, ..ma‘»

Figure h4.23 Transverse TDC located at the middle ofthe 15h? ..

134

 

depth

135

two-l

lane)

4.2 TDC

As st;
differentiate i
“as developc

and propagat

4.2.]

more

R)

two-lane highway. The pavement is two years old and the passing lane (the inner

lane) shows no sign of distress and/or raveling.

4.2 TDC Identification

As stated earlier, no procedure has been developed to identify TDC and to
differentiate them from regular longitudinal and/or transverse cracks. Such a procedure
was developed and is based on field observations and the understanding of TDC initiation

and propagation mechanism. This section presents and discusses the procedure.

4.2.] Longitudinal Cracks
An observed longitudinal crack is highly likely TDC if it satisfies one or
more of the following criteria:

1. Longitudinal cracks located at the edges of the wheel paths. These LTDC
are most likely initiated by the induced tensile stresses due to tire-
pavement interaction and temperature and are likely accelerated by the
low tensile strength of the AC mix at the pavement surface.

2. A group of parallel longitudinal cracks separated by 1 to 3 ft. and located
at the edges of, or within, the wheel paths.

3. Longitudinal cracks located between two lanes. These longitudinal TDC
are caused by the opening of the paving joint.

4. Longitudinal cracks that are located in segregated areas or in areas where

the AC mix is dry (low asphalt content and hence, low tensile strength).

136

l
Forcon:

v

I
longhuu

l. J
|
l.
7 5

ll

 

4.2.2 T]

31‘

For comparison purposes, the following are the locations of other types of
longitudinal cracks.
1. Low to medium severity fatigue cracks, which are longitudinal cracks
located at the center of the wheel paths.
2. Stripping cracks, which are longitudinal cracks located in non-segregated
areas at the center of the lane. These cracks initiate at the bottom of the
AC layer and propagate upward.
3. Longitudinal reflective cracks from longitudinal joints, which are typically
located between two lanes, at the edge of the lane (between the traffic lane
and the pavement shoulder) or between an old pavement lane and a

pavement widening strip.

4.2.2 Transverse Cracks
An observed transverse crack is highly likely TDC if it satisfies one or
more of the following criteria.

1. Short transverse cracks that initiate at the center of the lane. These TTDC
are caused by low temperatures and low tensile strength.

2. Transverse cracks (across or almost across the lane) that appear at a
constant interval throughout the pavement section. These cracks can be
identified by their low profile at the crack edges, which causes roughness.
The cracks initiate at the pavement surface and propagate downward
because the thermal stress at the pavement surface is much higher than

that at the bottom of the AC layer.

137

that a]
slab a

propa

the cr.

4.2.3

crack:

k)

Fihall

ofthe

5.0 A C D

DEnSi

identifiEd SC t

‘
N“

f

3. All short transverse cracks connecting LTDC. These short cracks could be
located anywhere along the pavement.
4. Transverse cracks in segregated areas.

For rubblized and composite pavements only, straight transverse cracks
that appear at a constant interval equal to the joint spacing of the original concrete
slab are joint reflective cracks. The cracks start at the bottom of the AC and
propagate upward. Reflective cracks typically have higher profiles at the center of

the cracks.

4.2.3 Network of Longitudinal and Transverse Cracks
An observed network of longitudinal and transverse cracks is TDC if the
cracks in the network satisfy one or more of the following criteria:

1. The network of cracks is located in and around the wheel paths and the
distance between two longitudinal cracks is less than 3 ft.

2. A network of meandering cracks that extends throughout the pavement
surface. This type of network is indicative of block cracking. The cracks
are TDC caused by hardening of the AC binder.

Finally, a localized network of cracks is not TDC if it is located around the center

of the wheel paths. These cracks are fatigue cracks.

5.0 AC DENSITY
Density of the AC mat was measured as to confirm the presence of the visually

identified segregation areas, and to determine if any trends in the AC mix density could

138

be attributed IC

 

included nucle
measurements

mo subsectim

5.] Nucle
Nuclc
Visually iden
segregation.
segregation
segregation
“as dex'elo
Departmen
The pro gr;
1 . ..‘\

D

IT

f»)

be attributed to TDC. As stated in Chapter 3, the density investigation in this study
included nuclear density (field measurements) and specific gravity (laboratory
measurements). The data of both measurements are presented and discussed in the next

two subsections.

5.1 Nuclear Density Data

Nuclear density tests were conducted to observe density differences between
visually identified segregated and non-segregated (control) areas that are indicative of
segregation. Chang et al (2002) showed that nuclear density is a good indicator of
segregation and they established criteria for light, medium and heavy degrees of
segregation that were embedded in the MBITSEG202 computer program. The program
was developed by the Pavement Research Center of Excellence (PRCE) for the Michigan
Department of Transportation (MDOT) as a quality control tool used during construction.
The program produces three possible outputs as follows:

1. “None” indicates that significant nuclear density differences are not present.
During construction, this result implies that the paving procedure needs to be
monitored, since segregation is suspected. In this study, this result implies that the
nuclear density differences show no segregation or light degree of segregation.

2. “Strong” indicates that significant nuclear density differences are present. During
construction, this result directs the paving procedure to be adjusted to eliminate
observed segregation. In this study, this result implies that at least a medium

degree of segregation is present.

139

3' s~\'cr). .

the cor

examine the r

Results of the

l. No se

ll ant
2- Medit
3. Tests

the field in“,

5'2 SPeci

The d
demfiy of [he
in lhls Studv‘

3. “Very Strong” indicates that severe nuclear density differences are present. For
the construction operation, this means that the segregation is severe and
segregated areas should be removed. The paving procedure will, of course, have
to be corrected to prevent the problem. In this study, this result implies that heavy
degree of segregation is present.

Nevertheless, the MBITSEG202 computer program was used in this study to
examine the nuclear density data obtained from the four test sites listed in Table 4.7.
Results of the analyses indicate that:

1. No segregation or light degree of segregation is possible on two test sites (30422-
11 and 30502-11).

2. Medium degree of segregation was found on test site 30502-21.

3. Test site 30521-21 is heavily segregated.

The above results confirmed the visual observations of segregation that were made during

the field investigation.

5.2 Specific Gravity Tests

The dependency of the asphalt properties, including modulus and strength, on the
density of the mixture has been well established. Since a variety of tests were conducted
in this study, variations in the test results may be explained by density variations in the
mixture.

Cores extracted from the pavement were returned to the laboratory, and the

average core density was measured using the ASTM standard test procedure D-2726. The

140

 

 

Test Site

\
30422—1 l
.\
305132-21
305v1-11
30521-21
\

Table 4.7 Summary of nuclear density tests

 

 

 

 

 

 

 

Number of tests in each Density in control Density in segregated
Test Site area areas (pcf) areas (pct)

Total Control TDC Average 3:135:31 Average 3:321:21:
30422-11 12 6 6 146.0 1.2 147.5 1.0
30502-21 19 8 11 146.3 0.8 142.7 4.2
30521-11 12 6 6 142.9 1.2 139.4 2.0
30521-21 12 6 6 150.4 3.0 148.1 3.5

 

 

 

 

 

 

 

 

141

 

average core
variation in tl

Table
The specific 3'
section 1. test
ranges from a
75. test sectio
may appear \w
percent air vo

modulus and :

6.0 PAVE

In this

average core density was then compared from one core location to another to investigate
variation in the average AC layer density.

Table 4.8 shows variations in the bulk specific gravity for the pavement cores.
The specific gravities of cores obtained from test sites range from 2.286 on M-52 test
section 1, test site 1 to 2.521 on US-4l, test section 1 test site 1. The standard deviation
ranges from a minimum of 0.008 on M-55 test section 1, test sites 1 and 2 , to 0.035 on I-
75, test section 1, test site 2. Initially, the variation of these bulk specific gravity values
may appear very small, but has a significant impact on air voids. The variation in the
percent air voids for each test site is listed in Table 4.8. Said variation affects the AC

modulus and strength.

6.0 PAVEMENT THICKNESS AND MATERIAL DATA
In this section, the measured AC thickness and the AC mix type of each test site

included in this study are presented and discussed.

6.] AC Mix Type

At the outset of this study, efforts were made to identify flexible pavement
sections exhibiting TDC. The research team and MDOT personnel toured numerous
flexible pavement sections and examined MDOT files. Ultimately, of more than 60
pavement sections that were investigated, thirteen showed TDC and were included in this
study. Recognizing the needs for inclusion of more test sites, five rubblized pavement
sections exhibiting TDC were also included in this study. Table 4.9 provides a list of the

AC mixes used on the 13 flexible and 5 rubblized pavement projects investigated in this

142

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0N NN0.0 2.0.0 _de vomd 0mm.~ .2 2020

We Nam mm0.0 N30 _mmd Evd wwm.m Q 2-3500
W 0.m mm0.0 30.0 SAN 0;.m mwm.~ 2 2-353
m. 1m 000.0 30.0 05mm 593 m0v.m Z 3-30:
5N _N0.0 000.0 $m.m 09mm 00m.m 0_ 5.30:

0880 Ho: 880 Box 5.82

mg 000.0 30.0 m0m.m mmva 000% E 2033

N; w00.0 3.0.0 owmd 0.9g movd m0 :-mmm0m

m.~ 20.0 0m0.0 vomd _Nv.m m0m.m w _N-_mm0m

0N 30.0 30.0 ch.~ 0mm.~ Smd 9 :-_~m0m

d 0N 30.0 80.0 Ema 02mm mnmd 00 _N-N0m0m
m. 0.0 $0.0 $0.0 Swim mwm.m SWN Q :-~0m0m
W w.m 20.0 0500 SAN SYN wme an Zammvom
0N 20.0 0m0.0 Emd «SN 00m.~ n Sigma

2 000.0 w~0.0 09am whim mcvd 0 NTNON0m

v.0 m _00 mm0.0 Siam 02$ 00v.m n :-N0m0m

0N $0.0 20.0 03% SWN wwvd Q N70 30m

Wm B _0.0 30.0 oovd _mmN m0m.m N0 2; 30m

N; m _00 0m0.0 vmvd mmvN. 03:8. 0 :-mnm0m

A00m.NHEEOV 228300 858006
25 m2?» tn Emccsm EEEme Esp—8:2 Esp—932 ommco>< 9mg 25 08h
“583$ 00 count? .3 50:52 .
03680 .33me 0:6on V==m

 

 

 

 

 

 

 

 

 

 

 

880 6083300 3mm: beam 006on owEo>< 06 203.

143

#

Test S:

‘ 20273-
20411-

20411-
! 30202—
30202-

30422-
30502-

 

Table 4.9 Summary of mix types on investigated pavements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Control Job Surface mix Completion Pavement
Test Site section number type date type
20273-11 37014 38205 10/13/2000
20411-11 07013 34037 13A 6/23/1999
20411-12 07013 34037 13A 6/23/1999
30202-1 1 37022 38196 4B 8/29/ 1996
30202-12 37022 38196 4B 8/29/1996
30422-1 1 83041 44396A 4B 3/20/1996 %
30502-11 23052 41007 4B 6/29/1997 5
30502-21 23052 45984 13A 7/1/1998 E:
30521-11 33051 34530 4B 6/6/1994
30521 -21 76011 36580 4B 10/18/1994
30552-1 1 72022 31009 4B 9/7/1995
30552-12 72022 31009 4B 9/7/1995

Marsh Rd 1400T N/A’

1 1941 -21 13033 29670 4B 6/4/1993
1 1941-22 13033 29670 4B 6/4/1993 '53
10753-11 16092 25559 l300-20AAA 12/6/1988 3:
10753-12 16092 25559 1300-20AAA 12/6/1988 0:"
30153-11 25092 45534 4C 10/1/1998

 

 

 

 

 

 

1. No construction records exist for Marsh Road

144

 

   
  

study. The data
exhibiting TDt

pavement pro j-

one can conclut

properties \x'ert

6.2 Core 1
At eac

for each core;
orthogonal di.
measuremem
“56d as input
Table
deviation 0ft
rubbliZed pm
subStamian}.
the accljracy ‘
test Site 18 USC
layer mOdUh‘.
similar to the ‘.

tests conducte<

study. The data in the table indicate that the majority of the pavement sections that
exhibiting TDC were surfaced using the 48 AC mix. In a parallel study of 87 rubblized
pavement projects in Michigan, thirteen projects were capped with 4B mix. A few years
after construction, all thirteen projects exhibited TDC (Svasdisant 2003). Consequently,
one can conclude that the 4B asphalt mix has high TDC potential and therefore, its

properties were subjected to further investigation.

6.2 Core Thickness Data

At each test site, about 10 to 15 AC cores were extracted from the pavement and,
for each core; four measurements of the AC thickness were made at the end of two
orthogonal diameters. Hence, for each test site, a total of 40 to 60 AC thickness
measurements were made. The average of all measurements was then calculated and
used as input to the backcalculation routine.

Table 4.10 provides a list of the average, minimum, maximum and standard
deviation of the measured thicknesses of pavement cores obtained from 13 flexible and 5
rubblized pavement projects. As can be seen, for some test sites, the AC thickness varies
substantially (more than 1.0-in) from one point to another. Said variation would affect
the accuracy of the backcalculated layer moduli. That is, if the average core thickness per
test site is used in the backcalculation, one should expect variation in the backcalculated
layer moduli. However, the average backcalculated moduli per test site was found to be
similar to the average modulus backcalculated using the deflection data from those FWD

tests conducted at the core locations where the AC thickness is accurately known.

145

 

  
  

9:5

EEESW
E05025 5:532

 

 

 

E: $23.25 20.0

0 0 0. . ..
:2: _> _ :.:::::2 EEEZE uuius< 3:500
. . 2.25.0.

 

 

 

9:: $3 :23 .50 mas—5.4.5. 3cm 2.. 22.0.5.5 Q~ .V 035.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

no 3. 2 mm 2 :amsm
w to cm 3. 3. 2 2-8::
w No I. we 3. 3 $52
W do 3 3 E 2 N28:
No 3 S 3 2 :-:a:

9o 3: a. : 3: w is:
:0 am am Tm 2 BARS
no .2 3 9m 2 :-~m3m
9o 3 2.. 3 a 5-58m
so an 3 am 2 :69:
m 3 mm 2 3 mm 8.82%
W. 3 9m ox S 2 :.88m
.1... no 3 3 2 2 $22;
3 3. 3. 3 2 2-88m
no 2. 2 S. : :éSm
No 2 am 3 3 NI :8
so 3. M3 on E :-_ :8
we 3 Na as S :émom

c2350
09: 0305.5 82552 838934 omSo>< $80 00 85 08,—.
E2550 806:2 .

 

05 $2235 800

 

 

 

02m 68 some 8.0 $65.25 88 E :ocmtm> 0 _ .v 033.

 

146

The core thickness data were also examined from a different perspective, cores
located in the wheel paths versus those outside the wheel paths. Table 4.11 provides a
list of the average core thicknesses for both locations. As can be seen, for the majority of
the test sites, the average core thickness, extracted from the wheel paths, is slightly less
than that outside the wheel paths. This was expected because the pavement in the wheel

paths has been subjected to secondary compaction due to traffic.

7.0 DEF LECTION DATA
Each pavement included in this study was tested using the MDOT KUAB falling
weight deflectometer (FWD). On each test site, a minimum of 29 deflection tests were
obtained in an effort to characterize the engineering characteristics of the pavement in
both the longitudinal and transverse directions. A detailed description of the FWD test
procedure and test site layout is presented in chapter 3. The objectives of the FWD tests
are:
l. Analyze the variations in the pavement deflections along and across the
pavement.
2. Backcalculate the layer moduli.
3. Assess the linearity of the pavement response to load.
The linearity of the MDOT FWD has been established in a report by Baladi and
Svasdisant (2002). Therefore, only the variation of the deflection data and the
backcalculation of layer moduli will be presented here.
Deflection is the response of the pavement to applied load. Deflection can be

thought of as an index expressing the structural capacity of the pavement. Hence,

147

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

QM Wm M MM #4 :uMm—OM
w i: 3 w 3 v 2.22:
W 0.0 0.? w Md v :uMmR:
W m: 2 w 2 4 N38:
5.0.. $6 w 0.0 v fiNufivm:
h; —.M w OM v N_-NmmoM
0M- _.M w N.M #4 :ammmoM
5M _d o w.w M _N-_NmOM
Md- w.m w —.0 V :LchM
:3 as: 20 VM 0.0 V _N-N6mOM
W. 04;- QM w 0.0 v NTNOmoM
We... o.wM- VM w wen v :ummvoM
VN 9v 0 W? M NTNONOM
0.0- 5v 0 0.? M :umomoM
OM 0N m min. v N_.— Snow.
2o M.M w M.M V :A—vom
M.N 6.0 w w.w ._.» :uMhNON

E: 355500: .550 00 ES @5520: .550 0c
EMMWE 5255.005 .x. 550 owm5>< 508:2 5wm5>< 508:2 52m 050.

203—50>: .5535m 052503 :—

 

 

 

 

 

 

350 5503 50: 523:0 0:: E 855020: 0583mm E 5055.005 :6 205%

148

variations in the measured deflection reflect variations in the structural capacity of the
pavement.

In this study, FWD deflection data were collected from all sites subjected to field
investigation, which included several rubblized pavements. The data presented here are
used to show the differences in response between conventional flexible and rubblized
pavements and also to identify trends in flexible pavements, if any exist, that link the
structural response of the pavement to TDC potential. A detailed analysis of rubblized
pavement response may be obtained from Baladi and Svasdisant (2002).

Table 4.12 provides a list of the variations in D1 through D7 deflections for
flexible and rubblized pavement projects. Tables 4.13 and 4.14 list the measured
pavement deflections along and across the rubblized pavement on l-75, test section 1, test
sites 1 and 2, respectively. Table 4.15 provides similar information for the flexible
pavement along US-27, test section 1, test site 1. The average and the coefficient of
variation for D1 through D7 deflections are also listed in the three tables. Figure 4.25
depicts plots of D1 through D7 deflections at the various test locations along I-75, test
section 1, test sites 1 and 2, respectively. The data in Tables 4.13 through 4.15 and in
Figure 4.25 indicate that:

l. The variations in the deflections of the rubblized concrete pavement on I-75, test
section 1, test sites 1 and 2 are much higher than the variation on flexible
pavements.

2. For both test sites, the peak pavement deflections (D1) show the highest variation
from about 6-mils to more than lO-mils. The reason is that Dl deflection is a

measure of the cumulative deflections of all pavement layers and of the roadbed

149

Table 4.12 Coefficient of variation of the deflection data of rubblized and flexible

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

pavements
Test site Coefficient of variation of the measured deflections (%) Pavement
D1 D2 D3 D4 D5 D6 D7 Type

20273-11 2 2 2 2 2 2 3

20411-1 1 3 3 4 6 6 7 7

20411-12 2 3 4 4 5 6 11

3020211 6 5 5 5 6 8 10

3020212 4 3 3 3 4 3 3

3042211 13 8 7 10 8 6 4 0
3050211 8 7 6 6 8 13 21 ,2;
3050221 7 7 7 8 8 10 12 g
30521-11 4 4 3 4 5 5 3

3052121 7 7 7 8 8 10 12

3055211 6 6 6 7 6 5 4

3055212 9 9 9 9 9 6 4

Marsh 1 10 8 7 5 4 4 10

Marsh 2 9 6 5 4 4 5 6

10692-11 9 10 11 10 10 10 15

10692-12 9 10 11 10 10 10 15

10753-11 14 13 12 10 8 5 6

10753-12 16 14 12 11 10 10 13

1194121 8 7 7 7 7 7 4

11941.22 11 8 7 7 6 6 5

201021 1 7 7 7 7 7 7 6

2010212 8 8 8 8 7 6 3 8
20233.11 12 12 11 10 8 7 15 .5;
20233.12 8 9 9 9 9 10 10 §
2027321 15 13 1 1 8 6 4 4 ‘3‘
20273-31 6 6 5 5 5 5 6

2027341 8 12 11 9 7 6 4

20311.11 10 10 9 7 7 8 10

3037321 19 14 10 17 14 12 10

30373-51 7 8 7 6 6 5 5

30373.52 15 16 15 14 14 13 11

30373-61 10 9 9 8 6 6 8

 

 

 

 

 

 

 

 

 

 

 

150

Table 4.13 Rubblized pavement D1 through D7 deflections, I-75, test section 1, test site 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FWD Deflection (mils)
station D1 D2 D3 D4 D5 D6 D7
1 9.46 8.29 7.45 6.22 5.30 3.67 2.03
2 10.02 8.59 7.69 6.47 5.48 3.82 2.12
3 9.62 8.48 7.78 6.47 5.17 3.64 2.03
4 9.52 8.17 7.22 5.93 4.96 3.51 1.92
5 9.73 8.44 7.53 6.25 5.22 3.59 1.85
6 9.56 8.07 7.13 5.92 4.98 3.50 1.90
7 9.88 8.25 7.35 6.14 5.17 3.67 1.97
8 9.32 8.18 7.36 6.18 5.31 3.82 2.02
9 9.01 7.81 7.09 6.03 5.11 3.70 2.02
10 9.07 7.78 6.94 5.86 4.78 3.31 1.79
11 9.67 8.39 7.50 6.24 5.20 3.50 1.92
12 9.24 7.73 6.89 5.80 4.93 3.58 1.92
13 9.00 7.76 6.95 5.82 4.89 3.17 1.72
14 10.30 8.47 7.26 5.96 5.02 3.52 1.90
15 10.06 8.69 7.41 5.79 4.91 3.61 2.04
16 8.12 7.22 6.44 5.42 4.58 3.27 1.85
17 8.22 6.97 6.27 5.34 4.60 3.38 1.87
18 8.90 7.78 6.98 5.84 4.90 3.15 1.73
19 7.77 6.70 6.04 5.16 4.50 3.40 1.81
20 7.68 6.63 6.03 5.16 4.44 3.34 1.95

 

 

 

 

 

 

 

 

151

Table 4.13 (cont’d) Rubblized pavement D1 through D7 deflections, I-75, test section 1,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

test site 1

FWD Deflection (mils)

station D1 D2 D3 D4 D5 D6 D7
21 7.22 6.21 5.60 4.87 4.26 3.29 1.87
22 7.76 6.30 5.68 4.88 4.27 3.28 1.89
23 8.57 7.55 6.81 5.67 4.74 3.21 1.73
24 7.43 6.47 5.89 5.12 4.48 3.46 1.84
25 7.49 6.53 5.90 5.11 4.45 3.39 1.99
26 6.87 5.93 5.40 4.78 4.21 3.32 1.91
27 8.50 7.46 6.69 5.57 4.62 3.20 1.74
28 6.92 6.10 5.64 4.98 4.45 3.51 1.89
29 6.65 5.96 5.53 4.91 4.33 3.35 2.01
30 6.49 5.69 5.25 4.72 4.19 3.33 1.94
31 8.30 7.30 6.58 5.52 4.62 3.26 1.75
32 7.07 6.20 5.65 4.92 4.34 3.40 1.94
33 6.05 5.48 5.14 4.68 4.22 3.50 2.12
34 9.70 8.16 7.24 6.18 5.26 3.87 1.99
35 7.26 6.29 5.66 4.89 4.25 3.31 1.98
36 7.89 7.09 6.42 5.70 4.53 3.27 1.90
37 8.29 6.89 6.21 5.24 4.45 3.26 1.82
38 7.06 6.11 5.57 4.90 4.28 3.30 1.94
39 6.36 5.66 5.28 4.80 4.32 3.44 1.99
40 6.76 5.96 5.55 5.02 4.49 3.62 2.12
41 8.47 7.33 6.55 5.54 4.67 3.27 1.77
42 8.76 7.85 6.77 5.62 4.73 3.34 1.86

Average 8.33 7.21 6.48 5.51 4.70 3.44 1.91

CV'A‘VO) 14 13 12 10 8 5 6

 

 

 

 

 

 

 

 

 

 

1. CV = Coefficient of variation = standard deviation as a percent of the average

152

Table 4.14 Rubblized pavement D1 through D7 deflections, [-75, test section 1, test site 2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FWD Deflection (mils)
station D1 D2 D3 D4 D5 D6 D7
1 10.45 8.53 7.37 5.95 4.87 3.45 1.89
2 9.30 7.84 6.92 5.74 4.80 3.46 1.83
3 9.98 8.42 7.38 6.08 5.07 3.51 1.78
4 8.38 7.15 6.41 5.45 4.68 3.34 1.66
5 10.55 8.38 7.07 5.61 4.60 3.26 1.68
6 9.12 7.68 6.74 5.53 4.56 3.14 1.60
7 8.98 7.39 6.44 5.24 4.31 2.90 1.46
8 10.01 8.00 6.85 5.43 4.33 2.85 1.32
9 10.83 8.70 7.35 5.82 4.61 2.93 1.25
10 9.33 7.49 6.49 4.89 4.07 2.81 1.41
11 7.85 6.66 5.85 4.82 3.95 2.72 1.42
12 10.30 8.34 7.11 5.58 4.45 2.82 1.29
13 9.31 7.62 6.68 5.51 4.63 3.40 1.83
14 8.37 7.50 6.54 5.39 4.57 3.23 1.70
15 9.40 7.23 6.22 4.91 4.03 2.74 1.36
16 6.86 6.05 5.37 4.50 3.80 2.70 1.45
17 8.09 6.57 5.81 4.97 4.25 3.22 1.84
18 8.17 7.05 6.29 5.34 4.55 3.24 1.72
19 9.03 6.76 5.78 4.66 3.88 2.68 1.35
20 6.16 5.40 4.90 4.24 3.64 2.69 1.42

 

153

Table 4.14 (cont’d) Rubblized pavement D1 through D7 deflections, [-75, test section 1,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

test site 2
FWD Deflection (mils)
station D1 D2 D3 D4 D5 D6 D7
21 7.68 5.99 5.27 4.52 3.93 3.09 1.81
22 8.03 6.93 6.29 5.50 4.85 3.26 1.76
23 8.63 6.41 5.51 4.48 3.72 2.60 1.36
24 6.79 5.80 5.12 4.37 3.77 2.80 1.50
25 6.79 5.96 5.25 4.32 3.63 2.62 1.41
26 7.22 5.82 5.19 4.50 3.95 3.14 1.87
27 7.89 6.82 6.23 5.51 4.89 3.35 1.79
28 7.52 6.08 5.20 4.26 3.59 2.58 1.40
29 7.16 6.22 5.48 4.49 3.74 2.18 1.34
30 6.47 5.72 5.18 4.58 4.08 3.24 1.92
31 8.08 7.05 6.37 5.60 4.96 3.37 1.80
32 7.53 6.22 5.32 4.41 3.76 2.79 1.54
33 7.52 6.46 5.70 4.67 3.88 2.68 1.37
34 6.57 5.69 5.14 4.47 3.91 3.13 1.89
35 6.53 5.77 5.34 4.72 4.11 3.18 1.82
36 7.26 6.38 5.78 4.94 4.24 3.20 1.80
37 8.73 7.58 6.82 5.88 5.11 3.42 1.74
38 5.93 5.27 4.84 4.28 3.77 2.97 1.67
39 6.65 5.84 5.28 4.53 3.92 2.97 1.65
40 7.65 6.31 5.51 4.58 3.90 2.89 1.57
41 7.20 6.21 5.63 4.89 4.30 3.17 1.69
42 10.54 7.66 6.54 5.28 4.47 3.26 1.80
43 8.40 6.82 5.99 4.92 4.05 2.81 1.42
Average 8.21 6.83 6.01 5.01 4.24 3.02 1.61
cv’(%) 16 14 12 11 10 10 13

 

 

 

 

 

 

 

1. CV = Coefficient of variation = standard deviation as a percent of the average

154

 

Table 4.15 Flexible pavement D1 through D7 deflections, US 27, test section 1, test site 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FWD Deflection (mils)
station D1 D2 D3 D4 D5 D6 D7
1 7.50 6.48 5.81 4.91 4.18 2.93 1.43
2 7.39 6.36 5.68 4.79 4.07 2.86 1.44
3 7.33 6.34 5.69 4.80 4.07 2.85 1.37
4 7.58 6.48 5.80 4.85 4.11 2.86 1.36
5 7.32 6.27 5.64 4.75 4.03 2.84 1.40
6 7.52 6.40 5.74 4.85 4.12 2.91 1.40
7 7.68 6.60 5.90 4.97 4.24 2.99 1.47
8 7.51 6.45 5.78 4.91 4.15 2.95 1.44
9 7.58 6.41 5.80 4.93 4.18 2.94 1.46
10 7.43 6.34 5.71 4.84 4.12 2.90 1.46
11 7.44 6.36 5.75 4.89 4.14 2.90 1.43
12 7.44 6.34 5.66 4.75 3.99 2.80 1.36
13 7.49 6.37 5.73 4.87 4.15 2.90 1.44
14 7.24 6.22 5.59 4.75 4.03 2.87 1.44
15 7.23 6.23 5.60 4.71 4.03 2.82 1.43
16 7.32 6.20 5.55 4.64 3.93 2.75 1.35
17 7.28 6.30 5.65 4.79 4.10 2.91 1.42
18 7.12 6.12 5.51 4.69 3.98 2.78 1.44
19 7.05 6.14 5.53 4.68 3.97 2.85 1.41

 

 

 

 

 

 

 

 

155

Table 4.15 (cont’d) Flexible pavement D1 through D7 deflections, US 27, test section 1,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

test site 1

FWD Deflection (mils)

Station D1 D2 D3 D4 D5 D6 D7
21 7.09 6.11 5.50 4.71 3.98 2.86 1.45
22 7.13 6.09 5.50 4.66 3.97 2.81 1.44
23 7.03 6.08 5.46 4.64 3.96 2.86 1.45
24 7.17 6.12 5.47 4.62 3.93 2.79 1.39
25 7.06 6.07 5.48 4.67 4.01 2.86 1.46
26 7.22 6.20 5.60 4.72 4.04 2.86 1.48
27 7.14 6.20 5.57 4.69 4.01 2.80 1.38
28 7.15 6.19 5.58 4.75 4.05 2.93 1.45
29 7.27 6.25 5.65 4.80 4.07 2.94 1.49
30 7.17 6.22 5.60 4.75 4.06 2.89 1.45
31 7.32 6.31 5.71 4.80 4.08 2.88 1.45
32 7.43 6.41 5.80 4.93 4.15 2.93 1.45
33 7.23 6.26 5.66 4.81 4.10 2.88 1.43
34 7.06 6.10 5.47 4.67 3.97 2.79 1.39
35 7.10 6.13 5.53 4.73 4.04 2.88 1.42
36 7.31 6.29 5.69 4.83 4.13 2.91 1.44
37 7.22 6.21 5.58 4.72 4.02 2.85 1.45
38 7.15 6.18 5.56 4.75 4.02 2.88 1.49
39 7.51 6.43 5.76 4.89 4.15 2.92 1.45
40 7.27 6.19 5.55 4.62 3.92 2.74 1.39

Average 7.29 6.26 5.63 4.77 4.05 2.87 1.43

cv’ (%) 2 2 2 2 2 2 3

 

 

 

 

 

 

 

 

1. CV = Coefficient of variation = standard deviation as a percent of the average

156

 

 

 

12 ' ' ' ‘ . : ;1 :1: 1' I : :2 : 2;: r,- ; , - ; _
1 1 i } i : 1 1 711—D1 - D2 """""D3 ’D4;?
10 i . ‘ ' :’ : 1 '

1 1 47 : ,
7 . o l 7 o . .7 fl . .
1.7-5-.. --- .- -1. -- ...- -. _ - . .4. , -. ..71_.:1- _- . .
1;; 1.111 . .1 1 : 1;.1511 -1- f1‘1 ”‘7...
:'—_: 8 . . e. I. 4 . . . k; .. _ 7‘ . ' 'H 1 .0
E
v
C
-9 6
H .
Q “1
to .
‘U 4
D *1 1

 

FWD test station

 

(a) Test site 1

 

 

12.00 . - - . ,
. D1 —D2----D33
. 'W‘WD4—D5—D6,

10.004 1
f1.__.' .D7....,_ "8‘1:

Deflection (mils)

 

FWD test station

 

(b) Test site 2

Figure 4.25 Variations of D1 through D7 deflections, 1-75 SB, test section 1

157

 

 

soil. Hence, variations in these layers and in the roadbed soil are reflected in the
measured D1 deflection.

For comparison purposes, the coefficients of variation of the measured D1
through D7 deflections for 30 test sites are listed in Table 4.12. As can be seen, the
coefficients of variation for [-75 SB section 1 test site 2 are relatively high compared to
the other rubblized test sites. On average, however, most of the rubblized pavements
show more variation than the flexible ones. This indicates that the structural capacity of a
rubblized pavement is more variable than that of a conventional flexible pavement. Given
that the main difference between conventional flexible and rubblized pavement sections
is the rubblized concrete slab, one can conclude that the high variations in deflections
could be directly related to variations and/or non-uniformity of the rubblized concrete
slab. These and other rubblized pavement issues are detailed in Baladi and Svasdisant
(2002)

Several of the flexible pavement test sections, such as M-50 test sections 1 and 2
and M-52 test section 1, exhibit high variations in the D7 sensor. The D7 deflection
reflects the energy delivered to the roadbed soi1 and the lower portion of the subbase, if
present. The pavements with high variation in D7 deflection (M-50, test sections 1 and 2
and M-52 test section 1) were constructed at-grade with little roadbed soil improvement.
Hence, these variations are indicative of variability in the roadbed soil. Other pavements,
in particular M-42, show high variation in the D1 deflection sensor and low variations in
the outer sensors. This implies that there is high variation in the asphalt layer and low
variation in the roadbed soi1. While these results are inconsistent with most of the other

pavements, the high variability is not surprising. Cores of M-42 show high thickness

158

variability, which affect the structural response of the asphalt layer, but not necessarily
the response of the roadbed soil.

Detailed analyses of the rubblized pavements investigated in this study are
presented in Svasdisant and Baladi (2002), therefore only the data pertaining to
conventional flexible pavements are discussed in this thesis. For each pavement
investigated, the deflection data were normalized relative to the mean deflection of the

test site. Profiles for these data are shown in Figures 4.26 to 4.33.

8.0 BACKCALCULATION PROCEDURE

In a parallel study conducted by Svasdisant (2003), various factors affecting the
backcalculation of flexible and rubblized pavements layer moduli were analyzed and a
backcalculation procedure for rubblized pavements was developed. It was found that the
procedure substantially improves the robustness of the backcalculation of conventional

flexible pavements. An overview of the procedure is presented below.

8.1 Overview of Backcalculation Procedure
As stated above, Svasdisant was successful in developing a procedure for the
backcalculation of layer moduli of rubblized pavements. The procedure has been adapted
for conventional flexible pavements, and consists of the following steps:
1. Obtain the following information/data:
> the thicknesses of the pavement layers and material types from the inventory

data and/or from coring and drilling;

3‘» the pavement condition at the FWD test site;

159

Norrralized
deflection

   
 
 

Transverse
distance (11)

Longitudinal 9°
distance (11)

Figure 4.26 Normalized deflections on 20273-11

160

 

   
 

0.95
Normalized 1
deflection 9
1.05 “
1.1
Transverse

distance (fl)

. . 60
Longltudmal 90
distance (it)

 

(a) Test site 1 (20411-11)

 

 

    

0.90
0.95
Normalized
deflection 1'00
1.05 9
Transverse
1.10 distance (it)

 
   

0 ‘ . .3
30 60 90

Longitudinal distance (ft)

 

(b) Test site 2 (20411-12)

Figure 4.27 Normalized deflections on US—41, test section 1

161

 

 

 

1 .

 

 

 

 

 

 

 

  
  

0.85
0.9 .
Normalized 0-95
deflection 1
1.05 6
1.1
l . 15 Transverse
distance (ft)
60
Longitudinal
distance (ft)
(a) Test site 1 (30202-11)
Normalized
deflection
9
., . Transverse
. . ' 3 distance (tt)
Long1tud1na 30
- 60
ld1stance
(fl)

 

(b) Test site 2 (3020212)

Figure 4.28 Normalized deflections on M-20, test section 1

162

 

 

    

L , d' l 60 ‘ Transverse distance
ongitu ma 3 (fl) _
distance (it) 902 ‘

Figure 4.29 Normalized deflections on 30422-11

163

 

0.8 -7
0.85
0.9

Normalized 0-95
deflection 1

Transverse
distance (it)

 

Longitudinal distance

(R)

 

(a) Test section 1, test site 1 (30502-11)

 

 

Normalized
deflection

  
  

Transverse

0 8 distance (ft)
Longitudinal distance 10

(fl)

 

(b) Test section 2, test site 1 (30502-21)

Figure 4.30 Normalized deflections on M-SO

164

 

 

0.95 . .,
l 7.. 'f/

Normalized 1 '05 I I

 

deflection 1.1
8
1.15
1.2 ) Transverse
0 . distance (it)
. . 9
Longltudinal 0
distance (ft)

 

(a) Test section 1, test site 1 (30521-11)

 

 

Normalized

deflection '

   

Transverse distance

Longitudinal 60 3 (fl)

. 90 2
distance (it)

 

(b) Test section 2, test site 1 (30521-21)

Figure 4.31 Normalized deflections on M-52

165

 

 

Normalized 0-95 / ,»

deflection 1
1.05 "
1.1
1.15

   
  
 
   
 

Transverse

60 distance (fl)
Longitudinal 90

distance (it)

 

(a) Test site 1 (30551-11)

 

 

0u~“":,
0%~””

Normalized 0-95 1“
deflection 1.05

    
 

Lw~“:,
Lfir“

‘ 8
1.35 '3'-
Transverse
6O 4 distance (ft)
Longitudinal 90

distance (ft)

 

mummmzommnm

Figure 4.32 Normalized deflections on M-5 5, Test Section 1

166

 

30m 5.82 so mcocoocg uni—«882 mmd Esmi

 

 

    

 

 

 

30% 5322 E US. 36535 E N ozm 3v
. . .. _. Ev oocfiflzmfivnzwcoq
o_ w o v
1 11 .1 11 1 1 We
1 . - . N
l 1 1 0
. . . 1 - 3 w
1 W
a.
9
- D.
. . . . .. . . . . . W
H. Ema—ooze.» 85:101. 2 m
H 53.833 BEOIOIH w H H ML. W:
11 N.
.253
H H 5 H .H . U M o.» 3.: 8566—26330;
om =o:£w 141 ob 2 m o v
weeocsmtflt m ... . m .3
2 8.35101 m .. cm H 1 N
HH.MHHM.. ..OOéW 0
.w 8m
o m W1 53 _uwn3c210101 .01..
ON U Eat—altoskcuuslol _ W
{so 11. . -1111. P
. . , . . , . , , 3 an
”1.. 100 H —..—m
8 8 9. on 8 o. o m

     

E: 2.2 cozumw FEE 85.65

 

       

N.—

 

 

167

‘r the measured deflection data and the distances between the deflection sensors;
and

> pavement temperature and time of the day for each FWD test.

Examine the deflection basins for possible irregularities. Irregular deflection

basins should not be included in the backcalculation. The irregular deflection

basins could be the results of:

‘2 Cracks in the vicinity ofthe FWD test;

> Edge effects; and

1”“ Underground utility pipes or drainage pipes/culverts.

Use Boussinesq’s equation to determine whether or not a stiff layer is shallow or

deep and conduct trial backcalculation using 3- and 4-layer systems as to

determine the approximate depth to stiff layer.

Set the convergence criteria in the backcalculation software.

Based on engineering judgment, establish acceptance criteria to accept or reject

the backcalculated modulus values at certain FWD test locations.

Input the proper data into the backcalculation software including the deflection

data, the number of pavement layers, their thicknesses, and Poisson’s ratios, and

the seed modulus value of each layer.

Conduct backcalculation of layer moduli.

For each FWD test location, examine the backcalculated results against the

acceptance criteria.

168

9. Calculate the temperature corrected AC moduli (TCAC). Note that, in this study
the backcalculated AC moduli were adjusted to the standard temperature of 680 F
using temperature correction model.

10. For each test site, calculate the average values and the coefficients of variation

(CV) of each backcalculated layer moduli.

8.2 Depth to Stiff Layer

The load induced-stresses and strains in most pavement sections dissipate to
almost zero value within few feet from the pavement surface due to the arching effects in
the soil. Hence, such stresses cause no measurable deflection. Theoretically, an elastic
layered computer program calculates stresses and deflections at great depths from the
pavement surface. In the backcalculation process, such calculations affect mainly the
backcalculated modulus of the roadbed soil. To alleviate the problem and to force the
layered elastic program to slow down the calculation of deflection, most backcalculation
software (including the MICHBACK program) allow the users to place a stiff layer at a
certain depth below the pavement surface. Such placement is problematic in that, if the
estimated depth to stiff layer is incorrect, it produces errors in the backcalculated layer
moduli. For example, if the depth to stiff layer is erroneously estimated at 400-in rather
than the true ZOO-in then the errors in the backcalculated roadbed, subbase, base and AC
moduli could be as high as 70, 40, 20, and 10 percent, respectively. In this study, an
innovative procedure based on two independent methods was developed to accurately

estimate the depth to stiff layer. The procedure is detailed in the next two subsections.

169

8.2.1 Equivalent Pavement Modulus
The approximate depth to stiff layer can be estimated using the equivalent
(or composite) modulus of the pavement and the roadbed soil as explained in the
following steps:
1. Select a minimum of three deflection basins (one at the beginning, one in the
middle and one at the end of the test site) and calculate the equivalent
modulus of the pavement layers and the roadbed soil at deflection sensors D2

through D7 using the following Boussinesq’s equation:

P(1 -— 1121
E = 4.4
7t(r idr 1 ( )
Where: E = equivalent modulus (psi);

P = applied load = 9000 lb in this study;
1) = Poisson’s ratio (assume 0.4);
r = radial distance from the center of the load to the
deflection sensor in question (in.); and
dr = deflection at distance r from the center of the load (in.)
2. Plot the calculated E versus the distance “r” as shown in Figure 4.34. The
resulting curve would have three possible shapes as follows:
a) The E values at the tail of the curve increases with increasing distance r_
This indicates possible shallow stiff layer. Higher rates of increasing E

relative to dr imply shallower depth to stiff layer as shown in Figure 4.34.

170

Equivalent modulus (psi)

5000 10000 1 5000 20000 25000 30000
0:: . ;.1 ; _;_'1';;*'1"i1’; ;.i; ; f. 1 7

I i I i i : I i h 1 1 Z i I I i I _1 +100-ins
' iii M' . t i i f 1 1 i-Cl-ZOO-in.‘

' 7 1 t : . z : 1
>1 7 v. *7; , i. “:7 7:7 7.7 . .» 1 . . L . . M.” 1 . 1 7f . 300—in ‘
0 1’: T *_: *1 1n. 1 r 1* 1*“? :' .: t? :“1” 1 '1 1 i

c. . .7 '1 . . . . . 1 . j. . 1 . i . . l . 1 1 ;
o~ -‘ ~ .. 1 o
no . , 1-9- 7 .7 . . ,1 . .7 . . . . - . 7 _ 7 ,. }—-O—4OO-1n.
: if 7:: j."1 : "i : 1 "t ; 7ft , 1511 T’iL 1
7: 7 7 *1 ..1 1 . . . . . 1 1 . 1 e . .7 .

a ; 1 7 1 . . . . . . . . ., . . 7, . 7 1 . +600-
50 r ' *1. 7‘ 7,, - 1. ~ ' - 74 ‘5 1 v 1 0 4 I - 4'1 c I 774+— ? 1 1n. 1
. 7. F; i 7 1‘77 7 -:o p 7.777 . b 7 V 7 + v 717 l o 747 l .7. o 79— 7 I ‘
1 i 7 1 i7 7 ‘ 7 7 - . . i . i . . - . . o 5 p 0 a t
. . 77 . 77 1 . 7, . . . . . . . . , f . l: 7 . 1 .‘ 7 1 _ 1
i . 7 7 _ .7 7T . 1 . . . . . . . . . 7; . . 7 ‘. 47 . : + t ff
6O - 1 7 . , , 7. I . . . , 1 A 1 . . . . . . 7 . ‘ n0 S 1
l o 7 .7I*_.. : .. fr 7 f 1 A 4 0 7¢7 ~4 i 7o .77 7 T— —70 ——‘I— ---i—7—‘- l.» 4‘ ‘ l ‘
. . 7 7777 7 7 7 , . ' . . A . . . . . , t .7 .7 — 7 7 7 l 7 ‘ ‘
. . : s ol f f i . . . . , . i '1 . t . f o i 1 - 4 ayer ‘
‘ 7 . L l . . . . . . . . . l 1 . . . . . . 7 : — H

Distance from the center of the load

Figure 4.34 Plots of equivalent pavement modulus versus the distance from the center of
load for 6 pavement sections having different depths to stiff layer

171

b) The E value at the tail of the curve is constant with increasing distance.
This implies relatively deep (400 to 660-in) stiff layer.

c) The E value continuously decreases with distance, which implies deep or
no stiff layer is needed as shown in Figure 4.34.

If the plots from the three deflection basins indicate different depths to stiff

layer (e. g., one basin shows shallow stiff layer while the others show deep

stiff layer), the following two scenarios are possible.
a) There may be underground facilities such as drainage culvert or utility
pipe under the FWD test locations.
b) There may be cracks in the vicinity of the FWD test location, especially
under the load plate or between the load plate and the deflection sensors.
Regardless which scenario is true, it should be verified and deflection
basins from those locations should be dropped or treated differently from the rest
of the data.

Note that this method can only be used to determine whether or not a stiff
layer should be incorporated in the backcalculation of layer moduli. The method
by itself cannot be used to closely estimate the depth to stiff layer. This can be

done using the next method.

8.2.2 Three- and four-layer system Backcalculation
For a given set of deflection data, the approximate depth to stiff layer can
be estimated by backcalculating the layer moduli for three and four layer systems

as outlined in the following steps:

172

Based on Boussinesq’s equation results, obtain a rough estimate of the range
of the depth to stiff layer (e. g., between 100 and 400-in). Divide this range to
three or four equal increments. For example, 100, 200, 300 and 400-in.

For each incremental depth to stiff layer and for each deflection basin of one
test site, perform backcalculation of layer moduli for the 37 and 4-layer
systems shown in Figure 4.35. For flexible pavements, the 3-1ayer system
consists of an AC layer (AC), combined base and subbase layer (BS) and
roadbed soil (RB). The 4-layer system consists of an AC layer (AC), base
layer (B), subbase layer (S) and roadbed soi1 (RB). This configuration is used
because on many flexible pavements, a subbase is either not included in the

construction, or is constructed of a similar material to the base.

. Calculate the average backcalculated BS and RB moduli from the results of

the 3-layer system and the average RB modulus from the 4-layer system.

. Plot the average backcalculated BS and RB modulus and the average

backcalculated RB modulus against the depth to stiff layer.

. From the plots, obtain the approximate depth to stiff layer as the depth of the

intersection of the two curves where the average BSRB and RB moduli are

equal. Based on information obtained from soil maps and/or other sources,

the true depth to stiff layer can be estimated as follows:

> The true depth to stiff layer is the depth at which the two curves
representing the moduli of BS and RB intersect (the two moduli are
equal). This is the likely scenario for pavement having the same type

subbase and roadbed soil.

173

 

 

      
  

AC Layer

AC Layer

1 ....

. v 1 I .
”1‘ .1 ’1‘ l IMIMlNlNlNl‘V-A-l 1‘ 1., l7 k I“

‘ .1 q _ Base/Subbase .
Roadbed 3°“ Roadbed Soil

4-layer system 3-layer system

 

 

    

 

 

 

 

 

 

 

 

 

 

Figure 4.35 Three and four layer systems for flexible pavements used to determine the
approximate stiff layer depth

174

 

‘P The true depth to stiff layer is Shallower than the depth at which the two
curves intersect when the BS modulus is higher than RB modulus, which
is the likely scenario for most pavements. The higher is the BS modulus
relative to the RB modulus, the shallower is the true depth to stiff layer
relative to the depth of intersection.

‘i The true depth to stiff layer is deeper than the depth at which the two
curves intersect when the BS modulus is lower than the RB modulus. This
is the likely scenario when the RB is hard and desiccated clay or hard pan.
The lower is the BS modulus relative to RB modulus, the deeper is the
true depth to stiff layer relative to the depth of intersection.

‘5» The true depth to stiff layer is infinite or, in other words, no stiff layer
should be incorporated into the backcalculation, if the plots of the average
BSRB and RB modulus never intersect.

In some cases, the true cross-section of flexible pavement consists of AC,
and thick aggregate base layers only situated on roadbed soil (no subbase layer).
For such cases, the 3—1ayer system is the true pavement cross-section whereas the
4-layer system can be made by splitting the aggregate base into two layers (basel
and base2). Since in this case, the base in the 3-layer system is split into base]
and base2, the estimated depth to stiff layer is the depth at which the average
modulus of base2 in the 4-layer system equal to the average modulus of the base

in the 3-layer system.

175

8.3

Finally, the selected depth to stiff layer should yield reasonable values for
all the backcalculated moduli, and the backcalculation should satisfy the
convergence criteria, which are explained in section 8.3.

To illustrate the procedure, two sets of 15 deflection basins were generated
using the CHEVRONX computer program. One set of 15 deflection basins is
based on 300-in depth to stiff layer whereas no stiff layer was incorporated in the
second set of 15 basins. The pavement cross-section and the layer moduli used in
the forward analyses are listed in Tables 4.16 and 4.17. Figure 4.36 depicts the
average backcalculated RB modulus from the 4-layer system and the average
backcalculated BS modulus from the 3-1ayer system. As can be seen, for the set of
deflection basins with 300-in deep stiff layer, the two curves intersect at the depth
to stiff layer of 350 in. implying that the stiff layer is shallower than 350-in. For
the second set of deflection basins where no stiff layer was incorporated, the two
curves do not intersect even at a stiff layer depth of 800-in. implying that no stiff

layer is required.

Convergence Criteria

In this study, several convergence criteria are specified in the MICHBACK

computer program. After each iteration; the criteria are automatically checked by

MICHBACK to determine whether or not the calculated and the measured deflection

basin converged. These criteria are:

. The modulus tolerance between two subsequent iterations is 0.1 %

176

Table 4.16 Thickness and moduli of pavement layers and roadbed soil of a pavement
section having stiff layer at 300 in. used in the analyses

 

AC

RCS

Subbase

Roadbed

Stiff layer

 

Thickness
(in.)

Mod
(ksi)

Thickness
(in)

Mod
(ksi)

Thickness
(in)

Mod
(ksi)

Thickness

(in)

Mod
(ksi)

Depth

Mod

(in) (ksi)

 

 

5.5

 

 

6.5

 

 

 

1 000

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

 

300

 

15

 

20

271

 

270.5

 

270

 

269.5

 

 

269

 

15

 

300 200

 

 

Table 4.17 Thickness and moduli of pavement layers and roadbed soil of a pavement
section with no stiff layer used in the analyses

 

AC

RCS

Base and subbase

Roadbed

Stiff layer

 

Thickness
(in)

Mod
(ksi)

Thickness
(in)

Mod
(ksi)

Thickness
(in.)

Mod
(ksi)

Thickness
(in)

Mod
(ksi)

Depth

Mod

(in) (ksi)

 

 

 

5.5

 

 

6.5

 

 

 

1000

 

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

300

 

50

 

100

 

 

300

 

15

 

20

 

Infinite

 

15

 

None

 

177

 

 

 

 

 

 

 

 

 

 

 

 

25.;.......;......;.:;1;;..-;_
5% 20 i i
m .
:3
E
o 15
E
1:
2
5310
:3 j ,i—I-3-layer i1 ._1 - 1 .7'
7‘3 5 ‘9' flaw; ( I _. _ .
g . 1 . . . 1 1 . 1 .j . ._1 t 17..
1 1 1
m ; T ’ ‘ 3
0 . 1 7’ 1 i 1 ' 1
100 200 300 400 500 600 700
Depth to stiff layer (in.)
(a) True depth to stiff layer 300—in
18 ; i .
r: 16 i, i i k
é :
m 14 .
:3 ;
"g 12 :
o
E 10
B 1 1 i
58' ::::;::::.:::.:rrt::.1:fi:1.__.
_a 6 1 32 2 éé ? 2 : 17+3-1ayers; vii
w; :::;:1
7‘: 4 i 1 1 1 + 1 : 1 «1': : +4511”
cg . . . 1 . . 1
m 2 f r 3 3 2 3
0 3 i 7% i i 5;: .
200 300 400 500 600 700 800
Depth to stiff layer (in.)

 

 

 

(b) No stiff layer (in the plot, 800 in. represents no stiff layer)

Figure 4.36 Backcalculated roadbed soil moduli (4-layer) and the combined moduli of the
base, subbase and roadbed soil (3-layer) of the pavement for different stiff layer depths

178

2.

8.4

The root mean square (RMS) error between the calculated and the measured
deflection basins does not change by more than 0.1 percent between two

subsequent iterations.

Acceptance Criteria

For quality control purposes, after the convergence criteria are satisfied, each

value of the backcalculated layer moduli was scrutinized based on several acceptance

criteria. These criteria and the reasoning behind them are listed below and summarized

in Tables 4.18 and 4.19, respectively.

1.

4.

The RMS between the calculated and the measured deflection basins is equal to or
less than 2 percent. This enhances the accuracy of the backcalculated modulus
values.
The backcalculated BS and RB moduli satisfy the following equation:

1.1* BS modulus_>_RB modulus
This criterion is based on the principles of pavement design, decreasing layer
modulus with depth. This criterion could be waved, if and only if, available and
reliable information indicate that the RB modulus is higher than the BS modulus.
The RM modulus is equal to or greater than 20 ksi but less than or equal to 100
ksi. This criterion is based on definition of RM that was established in this study.
The F C modulus is greater than lOO-ksi but equal to or less than 4,000-ksi. This

criterion is also based on definition of FC that was established in this study.

179

Table 4.18 The acceptance criteria for the results of different backcalculations at each

FWD test location

 

 

 

 

 

 

Backcalculation
Acceptance criteria 3-layer system 4-layer system 4-layer system
(AC, RC8 and (AC, RCS, BS and (AC, RM, FC and
BSRB) RB) BSRB)

Convergence Yes Yes Yes

RMS S 2% Yes Yes Yes
11 * Ebase+subbase Z Y CS

Eroadbed

 

 

 

 

The 3- and 4-layer systems used in the backcalculations are illustrated in Figure 4.31

Table 4.19 The reasons for each of the acceptance criteria

 

Acceptance criteria

Reasons

 

To eliminate the unaccepted results by the backcalculation

 

Convergence

program

To eliminate low accuracy backcalculation results and to
RMS S 2% assure close matches of the measured and the calculated

deflection basins.

 

 

l-l * Ebase+subbase Z
Eroadbed

 

Unless otherwise indicated, the stiffer material layers are put
on top to protect the lower material layer as a concept of
pavement design.

The 1.1 factor is provided to allow slight interaction in the
backcalculation.

 

180

 

 

8.5 The Temperature Correction of the Backcalculated AC Moduli
The backcalculated AC moduli are corrected to the standard temperature of 680 F
using the temperature correction equation below (after Baladi and Harichandran, 1987).

0.03658l68 — To)

Where: TCAC = AC modulus corrected to standard temperature of 680 F (psi)
EM: Backcalculated AC modulus at the measured pavement surface
temperature (Ts),(psi)

To: Mid-depth temperature of the AC layer (0F)

2 3
To = TS + (sin(— 6.3252t + 5.0967){— 0.8767; + 0.27886) + 0.032 {321] (4.6)

Where: h = AC thickness (in.); and

t = time of the day
Note that, during sunny days, the pavement surface temperature may vary from one FWD
test location to another depending on the wind speed, clouds and tree shadow. Therefore,
abrupt changes in the measured pavement surface temperature (such as 3°F or more)
between two FWD test locations were smoothed out by replacing the odd temperature by
the measured temperature at the two adjacent FWD test stations. After this, the moving 5-
test location average temperature was employed to minimize the differences between the
temperatures of two adjacent FWD stations. Note that, the time between two adjacent
FWD tests is about 2-minutes, it is physically not possible for the pavement surface
temperature to rise by more than 3°F and then drop. After calculating the 5-moving

average, the mid-depth pavement temperature was then calculated using equation 4.6

(Park, 2000).

181

 

9.0 BACKCALCULATION RESULTS

In this study, all measured deflection data were used to backcalculate the
pavement layer moduli. The primary input for the backcalculation program is the
thickness of each layer. Hence, data pertaining to the pavement cross-section were
obtained before the backcalculation was initiated. Table 4.20 shows, for each flexible
pavement layer, the type of material and the layer thickness that were obtained from the
inventory data. As can be seen, for several roads, some of the cross-section data are
missing (not available). The reason is that some pavement cross-sections are the result of
repeated maintenance and overlay activities dating back to before such data were
cataloged. For example, on M-52, the current road surface is placed over the original
gravel road. On M-SO, the lowermost AC courses date back almost 40 years and on US
41, an old concrete pavement was found about 36-in below the existing pavement
surface. The average AC core thickness for each site is listed in Table 4.20.

Table 4.21 provides a list of the layer thicknesses used in the backcalculation. For
some pavements, the data were obtained by drilling, for some others, from construction
records, and still for others, from the typical design practice of providing frost protection
of the roadbed soil. An example of the latter is M-55 where the thickness of the sand base
layer (no subbase) was assumed 30-in as to provide frost protection for the roadbed soil.

The average values of the deflection data for flexible and rubblized pavements are
listed in Tables 4.22, while the backcalculated moduli for flexible and rubblized
pavements are listed in Tables 4.23 and 4.24, respectively. The backcalculated moduli

were used as inputs for the mechanistic analysis of the structural response of the

182

:<\Z.. 3 38:8 2 65263:: 333050 mm? 5 8:850 on Ho: 2:8 :55 85d 4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8556 :53 325.0. <2 <2 5 85m 55mg»? 3: <2 552
$555 25m <2 <2 <2 83 55weww< o5 Tm N338
$556 25m <2 <2 <2 33 55weww< 3. S :fimOm
Go: 3 :53 <2 <2 <2 <2 3 <2 G. _ N35
8555 :53 <2 <2 <2 <2 3 <2 _ 7525
Cab 03 :53 358 <2 <2 <2 33 55meww< 3 <2 $5525
8555 :53 35% <2 <2 <2 83 55353 G <2 2-88m
Gab 3 95m <2 <2 <2 <2 3 <2 $825
Got 3 25m 2 QB 5.53% o 25m 5552»? 3 3. 2-885
Got 3 25m 353 2 QB 5533 o 55m 55.555? S. S. 37858
E501—
Oab 03 5532 E 55.3% 525m 2 55m 55585? 5.5 3 NI :8.
.xvcmm
8504
Got 3 382 E 853% 525m 2 5.5m 25weww< 3 3 :3 Gem
.xccmm
552 25m 353 <2 <2 <2 35m 5535.? o5 <2 {MRS
mmoconH 25532 $2325. 3582 200 Emflwonwfi
25 BB 2: com
E .mommobma woowmom ommnnsm 8mm As: $65.35. U< 62m 3.0...

 

 

 

 

 

coaoom 880 22:38

 

 

 

$65.35 88 U< 692on 2: 98 752225 823 was 693 13858ng 39:62: omé 635,—.

183

GOSNZAO—mov—omn 2.3 cm UDUS—OC_ HOG wag Uwom Emu“: .~

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- - - - mowed 5.82
moo—88m
:ouosbmrooEwmmoQ o 2 ed. N Tammom
moo—65E
soaogmrooEmmmoQ o 2 o.m fl Tammom
85v @852: o em #6 3-33m
mwcton 25E o cm ad 2439:.
82889
5355333660 o On 06 3-83m
moo—88m
cososbmrSEwfioQ 0 on _.o ~ ~ -NOmom
mmcton 93: o en fin 2-33m
mwcton 983 o cm mé NTNomom
mwcton 98: o cm 5‘ 2-88m
mwcton 283 cm N3 w.~ 23 Gem
mwcton can: on S 0m :4 Swom
55c henna: o R od :-mRo~
88 mo 850m ommnnsm 35m Amomwomwwwwflo
ozm “mob

 

 

Ad: $5833

:35?“ 05 mo 2258 59: .3 coca—3.58.03 2: E com: $2225.

 

 

€608 Emm— wo coca—zo—moxomn 2: E vow: 886532“ 8h: 2:. _N.v 635,—.

il.

 

184

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm: SM mm;V N:.m woe mud mos :meSm 2-2
mm: oo.m 56 56 mod Nwe mow 2.93: mm:
@3233: 5.: min mud mm.m mic :NN <m.w :-mmwo_ m5:
SYN :.m owd 3% nwd de ofic :43: <3:
mm: omN afim 00M mo;V 31“ mod :-_vo: v0:
owd EN mm.m co;V cod omd 3.0 98m 582
mm: m_.m Div vow mos mud 8.: NTmmmom mmgz
9.: RN Ed 3% mos cod 5.: :-mmmom mm-2
w: omN owd omé mm.m mow owd _N-_~m0m ~64):
we: NYN mmé ohm on.» vow wmd: :-_mmom mm-2
ac: 3:5. Cd 9% o: Cd Sum: _N-Nomom 092
63:8: 5d ohm wvd Now 3: omd 3.: :-momom 3-2
2N owe RR 2.10 3.: 2...: 3.0: :-mmv0m NYE
8.0 mm: owd wage. new 2.5 who NTNONOm 8-2
owd 3: n _ .m mmé ow.m :8 hmd :-NONOm 8-2
owd 3 .m mo.”u omw vow v2: 33: :-_ :vom Sim:
mo: wmd owe mw.m _m.w 3.0: 3.3 :-_ :vom Sim:
3.: $3 mo...V Rd mod 36 $6 :-mnmom gm:
mm on ma <0 mm ND 2:
25 8 5m E M: 2 w o 35 555 2:3
H5835 6:55 cocoa—mow E08622: @2352: .

 

 

 

 

 

 

 

 

 

0&5..on 2t 25 $5 $033 cocoococ 65 we 8:536 33m

 

 

 

 

358on @32an 28 03385 .QQEV 6:382:88 U< £96-28 can 20:06:36 3582: owfio>< mmé 2an

185

3 «e

3 @8865 8m $932 Eaflxo :oZ .—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.:o:m_=o_muxomn 2: E 39:05 8: mm? 33: 532 Box :32):
MN mm - - _N - wV Vamd 0.0m OWN; NTNmmoM mw-—2
mm mm - . NN - NV ch.N o. _ m wNn; :-NmmoM mméz
CV OV - - Vm - w: cmV 9w moo; _N-_NmOm Nm-_Z
MV VV - - om - :N Vwm WV: mom :-_NmoM Nm-_2
wV wV me of C - wN wa odm N2“ _N-Nomom 092
wN _V me ON: 0: - NN :mV V.V_ one :-NomoM OWE
@V Nm ooo.~ CmN N: - N.— Vo_._ a8 ©OV.N :-NNVo.m NV.2
N «N coo.N om: oN - mm omw; cNN Nma NTNONCM ON-2
cm on coo.N ow: NN - mm wVN.: N.MN to; :-NONOM ONAZ
NV NV - - ow @— VV :2; wVM woozm NT: _VON _V.mD
NV NV - - _V : wm VmOJ o. : m mON.N :4 _VON _V.mD
0V CV coo.N CON 2 - oN moo w. _ N _mw _ TthON 5N-mD

U< 360:8 A094

W M. N :8: V E: @333: Nommnnzm 8mm 238383. ohfiwwwmflchu U<
w W 3:69: 8?: tam . 28 53. 233:
W m. haw—tam 2 fiaon $33 3:69: @823_mov_omm

 

 

 

 

 

 

 

 

worm v.8 E2522: 03:85 2: 8.: 5608 U< 332.50 85:22:82 05 new 5608 8333833 omfio>< NV 2an

186

defining—883 2: E @2585 8: 83 mm: .—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm mm OOON com 8 Va: Ne mom: V8: ONQN :-mm:om 2-2
N732: mu-—
docflsofioxom o a E wow: 05 8: mm? mm-—
9 a _ :-mmno_ RT:
3“ Nm OCON 282 _N me on Nmo— 03. m8: NN-:Vo: Va:
om om OOON 282 VN 8V :. 3: N8: 82 _N-:Vo: Va:
W m. :8: ANS @333: 808.80 808.80 82%“th ohzwmwwweg U<
o w. 2:288 8 a: 08 08 05 a
m m. 85: ham 9 w H n: v .75 m 08228th £808.22 cocfim 88”:
W P firm canon 99: 9:388 Basso—«8.0mm

 

 

 

 

 

 

 

 

83. 62 82528 8332: 05 8.: 2388 U< 38280 828382 05 28 5608 82233833 ombifl VNV 2an

187

pavements. Tables 4.25 and 4.26 show the coefficients of variation of the backcalculated
moduli for each flexible and rubblized test site, respectively.

In general, the coefficient of variation is low, indicating that the backcalculation
results are consistent within individual test sites. Special cases do occur however, which
are discussed in the following sections. It is important to note that only the descriptions of
the backcalculation of flexible sections are included. Details of the backcalculation for

rubblized sites are available elsewhere (Baladi and Svasdisant, 2002).

9.1 US-27 Southbound, Section 1, site 1 (20273-11)

This pavement section is located north of Mt. Pleasant on loamy sand. The test
section is a flexible pavement located in the middle of a rubblized concrete pavement
project. The section (in the vicinity of an overpass, Isabella Road) was constructed after
removing the old concrete pavements to maintain bridge clearance. For this test section,
no specific cross section data were available from MDOT. The cross section was
established based on investigations of the surrounding rubblized pavements.

The variation in the measured peak deflections along and across the pavement is
shown on Figure 4.26. It can be seen that the variation in the deflection across the
pavement is relatively small; on average, the edge deflection is about 5% higher than the
deflection at the center of the lane. The relatively higher deflection at the outer wheel
path could be attributed to lack of edge support.

The pavement cross-section used in the backcalculation was composed of three
layers; an 8.9-in AC layer, 20-in base layer (for approximately 30-in of frost protection)

and a roadbed soil. The Boussinesq equivalent pavement modulus indicates that a

188

Table 4.25 Coefficients of variation of the backcalculated layer moduli of the flexible
pavement test sites

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Coefficients of Variation
Route Test site Temperature RMS
AC corrected Base Subbase Roadbed 0

AC (/o)
us-27 20273-11 5 10 7 -’ 3 30
US-41 20411-11 11 15 7 8 7 28
US-41 20411-11 16 17 6 5 11 29
M-20 30202-11 12 11 8 - 10 37
M-20 30202-12 9 8 5 - 3 31
M-42 30422-11 29 31 10 - 3 4O
M-SO 30502-11 24 21 9 - 16 43
M-50 30502-21 15 15 9 — 12 35
M-52 3052 1 ~11 15 16 8 - 3 54
M-52 30521-21 22 24 13 - 11 78
M-55 30552-11 18 20 14 - 4 37
M-SS 30552-12 27 3O 17 - 4 34

Marsh Road Marsh Road was not included in the backcalculation.

 

 

 

 

6‘ b‘

1. Non-existent layers are indicated by

189

Table 4.26 Coefficients of variation of the backcalculated layer moduli of the rubblized
pavement test sites

 

Coefficients of variation

 

 

 

Route Test site Temperature '
AC corrected Rubblized Fractured Roa dbe d
AC concrete concrete
[-194 11941-21 2 2 3 5 0
[-194 11941-22 3 2 2 5 1

 

 

 

 

 

1-75’ 10753—11
1-75 10753-12
M-15 30153-11 2 2 4 7 1

 

1-75 was not included in the backcalculation

 

 

 

 

 

 

 

 

 

 

190

shallow stiff layer exists. By splitting the base layer into separate base and subbase layers
and by calculating the layer moduli for different depths to stiff layer, it was concluded
that the stiff layer is located at 200-in below the pavement surface.

The backcalculated moduli of the pavement layers fall within the ranges of typical
values for the respective materials, and are therefore reasonable. The RMS is consistent
along and across the pavement and the average RMS is less than 0.50. This was expected
because of the following reasons:

1. Uniform AC thickness as measured from the pavement cores.

2. The consistency of the deflection data.

3. The lack of pavement distress.
Additionally, the results of the backcalculation almost match those obtained for the
surrounding test sites, which indicates that the backcalculated moduli are relatively

accurate.

9.2 US-41 North bound, section 1 sites 1 and 2 (20411-11, 20411-12)
The flexible pavement at the test sections on US-41 is unique in that it consists of
the following layers:
1. About 3—in thick AC layer.
2. A 12-in sand layer.
3. A 39-in thick uniformly graded, rounded and loosely compacted peastone.
4. A 9-in thick concrete slabs (on average, 54-in below the pavement surface).

5. Sandy, mucky loam roadbed soi1.

191

The variations in the measured peak deflection along and across the pavement are
shown in Figure 4.27. It can be seen that the deflections on sites 1 and 2 exhibit a similar
trend as that observed on US-27 with the peak deflections occurring near the outer wheel
path and the minimum at the center of the lane. On site 2, the deflection decreases along
the pavement and tends to be the greatest on the inside edge of the outer wheelpath.
Unlike rubblized pavements, the variation in the deflection is relatively small (it varies
about 5% from the mean).

Analysis of the deflection data using Boussinesq’s equation confirmed that a very
shallow stiff layer exists (in this case, the PCC slabs). The average AC thicknesses used
in the backcalculation were 3.3-in and 2.8-in, for test sites 1 and 2, respectively. On both
sites, the sand base thickness was set at 12-in, and the gravel subbase thickness at 39-in.
In the backcalculation, the roadbed was assumed to include the roadbed soil and the
concrete slab. This assumption eliminated the need to place a stiff layer at certain depth
and, as it was expected, resulted in a relatively high modulus of the roadbed soi1.
Therefore, the acceptance criterion relative to the modulus values of the roadbed and

subbase layer was not applied due to the presence of the concrete slab.

9.3 M-20 East Bound, Section 1, Sites 1 and 2 (30202-11 and 30202-12)
This pavement section is located east of Mt. Pleasant. The pavement cross-section
consists of:
1. AC thickness of 4.7 and 4.5-in on test sites 1 and 2, respectively.
2. A minimum of 30-in clean sand base and subbase layer (hand auguring indicates a

minimum of 60-in of sand).

192

3. Sand roadbed soil.

The measured deflection profiles are shown in Figure 4.28 for sites 1 and 2. Each
site shows the trend that has been observed with consistency for all flexible pavement
sections; the deflection at the pavement edge is higher than at the center of the lane. In
this case however, the maximum deflection is at the inside edge of the lane, as opposed to
the outside edge as has been observed in other test sites. Additionally, for site 2, the
deflections along the pavement decrease from the beginning to the end of the site.

The Boussinesq equivalent modulus of the pavement indicates that a shallow stiff
layer exists. By splitting the base layer into separate base and subbase layers, the stiff
layer was determined to occur at 150-in.

The average backcalculated moduli of the pavement layers are listed in Table 4.25
along with the average RMS. As can be seen, all values are reasonable and the average
RMS is about 1.07. One note is that the average roadbed modulus on M-20 of 22 and 26
ksi is close to the 17-ksi that was backcalculated on US-27. This was not expected
because according to the USGS soil maps of Isabella County, the native soils at both test
sections are identical. The discrepancy could be related to the assumed thickness of 30-in

on M-20, which would cause slight inflation in the backcalculated modulus.

9.4 M-42 East-Bound, Section 1, Site 1 (30422-11)

No construction and/or cross-section records were available for this pavement
section, which is located east of Mesick. Based on coring and boring, the pavement cross-
section consists of:

1. An AC layer whose thickness is highly variable.

193

2. A l-in thick silty gravel that was likely placed to provide stable surface for
construction traffic.
3. Deep deposit of sand roadbed soil.

The measured deflections were normalized and plotted in Figure 4.29. The
deflection data shows that the measured deflection is lowest at the outer wheelpath and
then increases to the inner wheelpath. In addition to this trend, an area of low deflection
occurs across a transverse line approximately 30-ft from the beginning of the test site.
This lower deflection is not associated with and physical features on the pavement such
as culvert, drainage or utility line and/or distress. Nevertheless, the maximum deflection
is approximately 17.7% higher than the average whereas the minimum is approximately
28.2% less than the average deflection.

In the backcalculation, due to the high variability of the AC thickness, two AC
thicknesses were used; 3.3-in was assigned to 38 FWD test locations and 4.7-in to 14
locations. Further, the l-in layer of silty gravel was ignored, and a base thickness of 30-in
was assumed, which would provide adequate frost protection. In reality, this 30-in
represents the approximate depth to which the deep sand roadbed soil would have been
improved during construction. In any event, it was anticipated that the backcalculated
layer moduli of the 30-in base layer and the roadbed soil would have similar values.

The Boussinesq’s equivalent modulus indicates that a moderately shallow stiff
layer exists. By splitting the base layer into separate base and subbase layers, the depth to
stiff layer was estimated at 250-in.

Because of cold temperature during the FWD test, the backcalculated moduli of

the pavement layers were high. Reasonable values were obtained after they were

194

corrected to the standard temperature of 680 F. As was expected, the base and roadbed
moduli values were not excessively different with the average base modulus of 17-ksi

and the average roadbed modulus of 12-ksi. The difference between the two moduli could
be explained by improvement of the upper portions of the roadbed (for stability and frost
protection) during construction and/or by the effect of the l-in thick silty gravel layer.

Finally, the average percent RMS is less than 1.00.

9.5 M-50 East-Bound, Test Section 1, Test Site 1 (30502—1 1)

According to the USGS soils map of Eaton County, the roadbed soil is comprised
of sandy loam. MDOT inventory data indicates repeated milling and overlays operations
back to the 1960’s. Therefore, no definitive cross section is available in the inventory
data. Although hand boring was not conducted on this test sections, the base was sampled
after coring. It appears that the sandy loam roadbed soil was simply compacted and an
AC layer was placed on top. Therefore, a base thickness of 30—in (for frost protection)
was used in the backcalculation of layer moduli.

The deflection variation on M-SO test section 1, test site I is provided in Figure
4.30a. The deflection on this site exhibits similar trend as that observed on M-20, the
maximum deflection occurs close to the inner wheelpath. Additionally, an area of
abnormally low deflection occurs in the vicinity of test points 3, 4, 5, and 12. Finally, the
measured D7 deflection (the outer sensor) is highly variable with coefficients of variation
of 22 and 12 (see Table 4.12).

The Boussinesq equivalent pavement modulus indicates that a relatively shallow

stiff layer should be used in the backcalculation. Multiple attempts to estimate the depth

195

to stiff layer were made by examining the results of various backcalculation using
different depths to stiff layer; none were successful. This could be the consequence of
the high variation in the D7 deflection.

In the backcalculation, a depth to the stiff layer of 120-in and modulus of 485 ksi
were used. These produced reasonable results (see Table 4.23). One note is that of the 41
test locations, the results from only 28 locations were accepted. A good number of FWD
tests were located over and/or in the vicinity of cracks. This tests section had experienced

significant distress and cracking was prevalent throughout the test site.

9.6 M-SO East-Bound, Section 2, Site 1 (30502-21)

Test section 2 on M-50 is situated on the same type roadbed soil as test section 1.
Therefore, similar difficulties in backcalculating the depth to stiff layer were expected.
Except for the thickness of the AC, the pavement cross section (30-in base layer, semi-
infinite roadbed) was the same as before. However, no hand borings were obtained on
this site.

The deflection variation on M-SO Section 2, site 1 is illustrated in Figure 4.3%.
As can be seen, the deflection varies from 12.1% less to 17.8% higher than the average
deflection. The lowest deflection occurred at the beginning of the site whereas the highest
at the center of the site along the center of the lane. The primary distress on the site is a
longitudinal crack running along the outer wheelpath.

Similar to test section 1, a relatively shallow depth to stiff layer was found.

Analyses of the deflection data by splitting the base layer into separate base and subbase

196

layers indicated a stiff layer depth of 185-in, which was assigned a modulus value of
485-ksi. An AC layer thickness of 6.9-in was used based on core measurement data.

As opposed to the backcalculation on test section 1, all 48 test stations converged and
passed the acceptance criteria. The average percent RMS for the site was less than 1.00.
The temperature corrected AC and the roadbed moduli were similar to those obtained
from test section 1. However, the backcalculated base layer modulus was on section 2
was about 30 percent higher than that of section 1. This could be attributed to the

difference in the distress between the two sections.

9.7 M-52 North-Bound, Section 1, Site 1, (30521-11)

M-52 test section 1 is situated in northern Ingham County. The MDOT inventory
data indicates that M-52 was built on a raised embankment over a gravel road. Hand
borings indicate that a base of silty gravel exists to a depth of 36-in. At 36—in, a hard layer
was encountered that could not be bored through using the equipment on hand. It is likely
that this hard layer is the original surface of the gravel road. This is confirmed by the
Boussinesq’s equivalent modulus procedure, which indicated very shallow stiff layer.

The measured deflection data were normalized and plotted in Figure 4.3 l a. The
data show moderate variation in the deflection along and across the pavement.

The thicknesses of the extracted cores were highly variable. Hence, in the
backcalculation, three AC thicknesses were used as follows:

1. 5.3-in AC for 4 test stations.
2. 5.9-in AC for 37 test stations.

3. 6.5-in AC for 3 test stations.

197

The base layer thickness was set at 36-in as measured from the hand boring.

A three layer system was employed in the backcalculation of layer moduli; AC,
base and a semi-infinite layer comprised of combination of the gravel road structure and
roadbed soil. Because of this modeling technique, it was anticipated that the modulus of
the semi-infinite layer would be higher than the modulus of the base layer, and therefore
would fail the acceptance criteria. Consequently, the semi-infinite layer was labeled a
stiff layer due to the presence, and the inclusion, of the gravel road. The backcalculation

converged without difficulty at all 40 test stations and the average percent RMS was less

than 1.00.

9.8 M-52 North-Bound, Section 2, Site 1, (30521-21)

M-52 test section 2 has an identical cross section to that of section 1 including the
old gravel road at about 40-in below the pavement surface. Hence, the backcalculation of
layer moduli were conducted using the same procedure. To avoid unnecessary repetition,
only the normalized deflection data are presented herein.

Figure 4.31b depicts the normalized deflection data. As can be seen the deflection

profile varies along and across the pavement. Results of the backcalculation are listed in

Table 4.23.

9.9 M-55 East Bound, Section 1, sites 1 and 2 (305552-1 1,12)
This pavement section is located west of Prudenville, and the soil map of
Roscommon County indicates that the test section is located on sand. While no hand

borings of the pavement structure were obtained, sampling of the base material indicates

198

that the base is sand. Construction plans for the project indicate that an engineered base
was placed, but its thickness was not specified. Therefore, the base thickness was
assumed to be 30-in for frost protection.

The deflection variation on M-55 Section 1, sites 1 and 2 can be found in Figure
4.32. The deflection data on site 1 shows that the measured deflection is highest at the
outer wheelpath. This is consistent with the cracking observed near the curb. In this case,
the deterioration of the asphalt is likely due to different expansion rates between the
concrete curb and gutter and the asphalt mat. The overall deflection of the site does not
vary by more than i 15%. However, on site 2, about 30 percent higher than average
deflections were measured at the transverse line 90-fi into the beginning of the site. The
reason for the high deflections is transverse cracks at that location.

The Boussinesq equivalent modulus procedure indicates a moderate depth to stiff
layer. By splitting the base layer into separate base and subbase layers, the stiff layer was
determined to occur at 270-in.

The backcalculation was conducted using an average AC thickness of 3-in and
30-in thick base layer on top of semi-infinite roadbed soil. The temperature corrected AC
modulus was abnormally high. This caused re-examination of the data and the
backcalculation procedure. It was found that since the AC thickness is only 3.0-in, minor
variation would cause significant impact on the backcalculated modulus. Therefore, the
backcalculation of layer moduli was re-conducted using, for each test site, the 12 FWD
test stations where AC cores were extracted. Results of the analyses are listed in Table

4.23. As can be seen, the backcalculated and temperature corrected AC modulus is still

199

higher than typical values (higher than 2000-ksi), no adequate explanation can be given at

this time.

9.10 Marsh Road (2 test sites)

Marsh Road in Okemos is located on sandy loam. The test sites were located
almost half way between Hamilton Road and Grand River Avenue. The pavement has
been milled and overlaid repeatedly, which resulted in 10.6-in AC thickness. No
inventory data exists regarding the thicknesses of the pavement layers. Construction plans
indicate that a 4-in base layer was placed.

At each test site, the deflection data were collected only in the outer and inner
wheel paths, therefore, 3-dimensional plots could not be made. Figure 4.33a and 4.33b
depicts variation in the measured and normalized peak pavement deflection along the
wheel paths while Figure 4.330 shows three typical deflection basins. The data in the two
figures indicate that the peak pavement deflection varies by about i 15 percent and that
the shapes of the deflection basins are irregular.

For each test site, various unsuccessful attempts to backcalculate the layer moduli
were made. The reasons include the irregularity of the deflection basins and the condition
of the pavement surface of Marsh Road. The pavement showed an extensive network of

top-down cracking and raveling in moderate and high severity segregated areas as shown

in Figure 4.33d.

200

10.0 INDIRECT TENSILE TESTS

Indirect tensile cycle load tests (ITC LT) and indirect tensile strength tests (ITST)
are convenient and useful tests to determine the resilient moduli and strength of AC
specimens. The tests require a minimum specimen thickness of 2.2-in. In no case
however, an asphalt had the required thickness. For this reason, the test specimens were
composed of more than a single AC course. Hence, the test is a useful indicator of
relative trends in pavement strength. In this study, both tests were utilized. The findings

are summarized in the following sections.

10.1 Indirect Tensile Cyclic Load Tests (ITCLT)

Most asphalt pavements are constructed using multiple lifts or courses. Each
course may be composed of a different mix than the other courses. Also, different courses
may be subjected to construction anomalies such as segregation or varying compaction.
These factors affect the modulus of the AC course in question. ITCLT are used to study
the variation in the AC modulus throughout the thickness of the AC layer and along and
across the test site.

ITCLT were conducted on specimens obtained from 9 flexible and 5 rubblized
pavement test sites. The results of the tests are shown in Table 4.27. The data in the table
are divided into different categories based on where the test specimens were obtained
from within the AC layer. As can be seen, several pavements have only one set of test
results. This is because the AC layer on these pavements was so thin that the entire core
was used as a test specimen. This situation occurred on US—4l , test section 1, test sites 1

and 2, on M-55, test section 1, test sites 1 and 2 and on I-75, test section 1, test sites 1 and

201

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 EN m - - - 3 wNN m ._-mflom 2-2
We - - - - - - 0 c2 v NTmmfi: mm.-—
.m. - - - - - - 2 08 m :132 m:
m. E 8N m 2 :N m N oNN N NN-_vo: v0:
2 SN m Z 03 m 3 am: m _N-_vo: a:
m. 09 m E w? w m S: N Box 5.52
- - - - - - 3 NM: N. NTNmmom 3-2
- - - - - - ON of N_ :-Nmmom 3-2
H I I: m om 2: m N. om: m 2-83m 3.2
W. 3 Q: m - - - M: o: c NTNONom 8.2
9 MN of v - - - MN 3. m _ TNONoM 8-2
- - - - - - 2 mON N. NTZVON Sum:
- - - - - - ON m: N_ 2-23N Sim:
w 2N N S :N v - o: _ :-mnNON NN-mD
an: as: 32 ac as: as: 32 .o as: as: 82 do
23,—. sonata: 3.:on 895:2 22336: 33:9: 828:2 conflict 3.358 828:2
2883mm USE—Sm owEu>< Baccam quo>< Eaccfim owfio>< 2% 53. 23m
28 90 5:5: Eozom 28 ..o :5an 22:2 28 .«o .5an nob

 

 

 

 

 

 

 

 

5b: U< on: E 330: afloat: Eot So £586on :8 H49: 80¢ 33:28 22602 BN6 2an

202

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 4w 8 N: 2 :.25m 2.:
ma 2 S M: E: 4 2-3:: m:
M : S w o: v :52: 3:
my mm 3 8 E w ~38: 82
mm 2: 9. com 5 5:8: 8:
_N fl 8 w: 2 :8: as:
2 a 2 o: : 2.3:: 3-2
2 8 2 S: : :52: 3-2
H 2 mm om N: E :22; 8-2
m... 3 mm x mm w 2.8:: 8-2
m. 2 S 2 8 a _ 788m 8.:
3 mo 2 m: 2 2:2: :13
a 8 2 m: N: :.:EN 3%:
m cm 2 S A :ARS 5m:
A53 A68
COSN—>®U wZMSUOE . . mwmowmo
59:. 559.5% omw5>< c2535: Emcee: 59::2
Emccflm owm5>< Ba 35,—. 3:3:
85:55; d8: 33:05 .
3.223 .5: an: £88m

 

 

 

 

 

 

as E: 386: 8892 :3 29¢

203

2. Nevertheless, test specimens from other sites indicate that significant variation occurs
throughout the depth of AC layer. Test specimens from the lower portions of the AC
layer tend to have higher modulus values than the specimens tested from the upper
portion of the AC layer. This is due to the fact that, in general, coarser aggregates are
used in the lower AC courses. Variation across the site is not insignificant either; most
sites show at least 5% variation for all portions of the AC layer, with 10% variation
common. Ultimately, it shows that the behavior of the AC layer is a complex interaction
of multiple courses having different moduli values, which contribute to the overall

variation in pavement behavior.

10.2 Indirect Tensile Strength Test Data
Indirect tensile strength tests (ITST) were conducted using the apparatus specified

in ASTM standard test procedure D-1559 and the loading strips specified in D-4123 to
determine the tensile strength of the asphalt mix. In this study, tests were conducted on
specimens obtained from pavement cores. The purposes of the strength tests are:

1. Determine if the mixes experiencing TDC show abnormally low tensile strengths.

2. Determine relative fatigue performance of the mixtures.
The second point is possible because tensile strength is an indicator of fatigue

performance of the mixture.

10.2.1 Indirect Tensile Strength

ITST were conducted on specimens from cores drilled from 11 test sites.

Table 4.28 summarizes the test data for specimens obtained from the top (surface)

204

 

portion of the pavement cores. It is important to note the indirect tensile strength
of compacted asphalt is typically 200-psi. The measured strengths of the asphalt
mixtures are all substantially lower than 200-psi, with the exception of those
obtained from US—4l , which showed 185-psi strength. This indicated that the
compacted mixtures are substantially weaker than would be expected. It also
indicates that the ability of the mixtures to resist fatigue cracking is severely
impaired.

The mixture obtained from US-27 also shows very low strength, but no
TDC was found in the field. However, this pavement is only 2 years old. It is
likely, based on these results, that this pavement will develop TDC in the near
future. The pavement is being monitored for the appearance of cracks.

Specimens obtained from US-4l test section 1, test sites 1 and 2 show
high strengths compared to specimens obtained from other test sites. However,
the TDC on US-41 is occurring in an area of medium segregation. All cores were
drilled in non-segregated and non-distressed areas. Hence, no measurement of the
segregated pavement strength was obtained. It is reasonable to assume that the
segregated areas, in which the distress is occurring, has lower strengths than the
surrounding non-segregated areas. This situation is typical for all pavements
investigated in this study. The loss of strength is especially important when the
non-segregated mix strength is already low, as on M-50 test sections 1 and 2 and

on M-55 test section 1, test sites 1 and 2.

205

Finally, the mixture strength on M-20 test section 1, test site 1 is very low,
despite the fact that no segregation was observed in the pavement. This may be
due to several problems, two of which are addressed later in this chapter:

1. Poor compaction of the mix leads to low mixture strength.

2. The mix itself is susceptible to environmental impacts such as moisture.

10.2.2 Equivalent Modulus

The indirect tensile test apparatus is equipped to record the deformation of
the specimen during loading. From the deformation, an equivalent modulus (EM)
of the test specimen was calculated using the following equation:

P
EM = if (4.08595 — 004173330) (4.7)

v
Where: EM = Equivalent modulus (psi);
P = half the maximum load magnitude (lbs);
L = specimen thickness (in);
A, = specimen deformation in the vertical direction (in); and
v = Poisson’s ratio, which was assumed = 0.3
Table 4.28 summarized the ITST data including the EM. Young’s
modulus values for asphalt mixes typically range from 100 to 1,000-ksi. The EM
values range from 28 to 95-ksi. It is not the intention to use the EM values as
replacement to Young’s modulus. Rather, it can be used to provide a comparative
framework of the stiffness of each mix.
The lowest stiffness values were obtained from mixes on M-20 test section

1, test sites 1 and 2 while the highest stiffness belongs to mixes obtained from

206

US-41 test section 1, test sites 1 and 2. Coupled with the strength data, this
implies that the mix on M-20 is “soft and weak” while the mix on US-41 is

“strong and hard.”

11.0 EXTRACTION OF AGGREGATE

The purpose of the aggregate extraction was twofold:

1. Conduct sieve analysis on the recovered aggregate to determine the gradation of
the pavement.
2. Determine the asphalt content of the recovered cores.

Extraction of aggregate was conducted using a forced air ignition oven heated to
450°C. At this temperature, the asphalt cement was burned off, leaving the dry aggregate,
which was sieved. Table 4.29 summarizes results of the sieve analysis.

Gradation is the ultimate indicator of segregation. If the mixture gradation varies
across the site, then by definition, the mixture is segregated. Also, asphalt content is an
indicator of segregation, which can be determined by weighing the specimens before and
after incineration. Finally, knowing the exact mixture properties is useful in

manufacturing test specimens.

11.1 Gradation Analysis
Gradation analysis on the recovered aggregate was conducted in accordance with
ASTM standard test procedure D-5444. In each analysis, the gradations of cores from

visibly segregated and non-segregated areas were compared to the job-mix formula

(JMF).

207

Table 4.29 Sieve analysis of extracted aggregate and asphalt content

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ Sieve Test site designation number
113:2; opening 20411-11 20411-12 30552-11 30552-12
srze (mm) Seg Non-Seg Seg Seg Non-Seg Non-Seg
1 25.4 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%
0.75 19.05 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%
5 12.7 94.3% 96.4% 94.6% 98.0% 98.4% 99.0%
0.375 9.525 75.5% 82.8% 77.1% 86.3% 91.2% 91.1%
#4 4.76 45.6% 57.9% 50.6% 53.7% 64.5% 64.1%
#8 2.38 31.3% 43.4% 38.1% 36.6% 46.5% 46.2%
#16 1.19 22.4% 33.5% 30.3% 28.9% 36.7% 36.5%
#30 0.59 21.1% 25.6% 23.5% 23.0% 28.6% 28.1%
#50 0.3 10.5% 14.8% 15.0% 11.9% 13.7% 11.1%
#100 0.149 6.2% 6.4% 7.3% 5.6% 5.8% 4.8%
#200 0.074 2.0% 0.9% 3.5% 3.7% 3.6% 2.6%
%AC 7.1% 6.0% 5.6% 5.0% 6.5% 6.4%

 

 

 

 

 

 

 

Table 4.29 (cont’d) Sieve analysis of extracted aggregate and asphalt content

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ Sieve Test site designation number
Sieve opening 30202-1 l 30202-12 30502-21
number size (mm) Non- Non-
Cracked cracked Cracked cracked Seg Non-Seg
100.0% 100.0% 100.0% 100.0% 100.0% 100.0%
0.75 19.05 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%
5 12.7 100.0% 99.8% 99.8% 99.9% 98.6% 98.6%
0.375 9.525 94.4% 95.3% 93.3% 94.8% 78.6% 90.0%
#4 4.76 64.0% 67.2% 68.9% 70.3% 40.3% 57.6%
#8 2.38 46.1% 47.7% 48.4% 50.6% 34.5% 40.4%
#16 1.19 35.9% 36.0% 37.4% 37.2% 26.0% 30.6%
#30 0.59 27.2% 27.1% 28.4% 28.1% 18.6% 22.6%
#50 0.3 12.7% 12.1% 13.8% 13.2% 9.8% 13.9%
#100 0.149 4.6% 3.4% 5.2% 5.0% 4.2% 7.3%
#200 0.074 2.9% 1.6% 2.9% 2.8% 1.9% 4.5%
%AC 5.7% 7.1% 6.8% 6.4% 7.5% 6.8%

 

208

 

Gradation comparisons for US-41, test section 1, test sites 1 and 2 (2041 1-1 1,
20411-12), M-20, test section 1, test sites 1 and 2 (30202-11, 30202-12) and M-55, test
section 1, test sites 1 and 2 (30552-11, 30552—12) are shown below in Figures 4.37
through 4.39, respectively. Gradation differences in specimens from US-41 and M-55
(20411-1 l, -12 and 30552-1 1, -12) indicate segregation consistent with that observed in
the field. On US-41, test section 1, test site 1 (20411-11), the segregated specimens show
a difference of gradation of about 12% on the number 4 and number 8 sieves. On US-41,
test section 1, test site 2 (2041 1-12), the differences on the number 4 and number 8 sieves
are less severe but still noticeable at 5.7% and 7.3%, respectively. On M-55, test section
1, test site 1 (30552-11), the differences on the number 4 and number 8 sieves were
10.8%, and 9.8%. On M-55, test section 1, test site 2 (30502-21), the differences in the
number 4 and number 8 sieves were 17.2% and 5.9%. Gradation differences are not

apparent on M-20 (30202-1 1 and 12) and segregation was not observed in the pavement.

11.2 Asphalt Content

As previously mentioned, asphalt content may be used as an indicator of
segregation or to detect dry mixtures. The measured asphalt contents are shown in Table
4.29. As can be seen, the measured asphalt contents are inconclusive of segregation. On
30502-21 and 20411-11, the asphalt content of the segregated areas is actually greater
than that of the non-segregated areas. Although the pavement on M-20 showed no
visually identified segregation, the cores were incinerated and the recovered aggregate
were sieved. Results of the sieve analysis confirmed non segregation. In this study, the

asphalt contents did not provide any conclusive evidence of segregation. The data were

209

 

 

US standard sieve size

 

    
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

@000

A
voxmox
“Vo'Vé—zrfil‘.
O“ Ov—Nfl'
oo

9.525
12.7
19 05

Sieve opening size (mm)

o o

100.0% 1 1
.. 80.0% .;+JMF . g g g
.g 60 00/ .-A—Segregated .3 ' , ,.
é ' ° ‘: -D-Non-Segregatedl ' f * ,1
§ 40.0% f 1 _ 1
£2 ; s.

20.0% i -

: 1

goo/or ~ I
V O5 M O5 0‘. 00 \O VIN In
I‘ V o' V? "-1 "I 1‘. Ncsi O
O "1 0 ~ N <1- V1~ ox
O O ox .—
Sieve opening size (mm)

(a) Test site 1 (20411-11)

US standard sieve size
9 .EE ..‘5
a 2 a a :2 .0 0 obo': <‘r
at :tt :0: =1: 2:1: =11: :1: FR: R

100.0% ~ , _ _J; ,

80.0% 1 1+JMF 1: ’ , .4 .
on . . , , __ ._ ......
.5 fi—A—Segregated} {I ll 1
3 60.0% 7’ ' _ 7;
Q. t z 7 7777 .
§ 40.0% -'

a? : +
20.0% ; ;
0.0% 1 7

 

(b) Test site 2 (20411-12)

Figure 4.37 Gradation analysis of US-41 , test section 1

210

 

 

 

 

US standard sieve size

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.59 ‘1
1.19
2.38

"I
o

0.074 j
0.149

Sieve opening size (mm)

 

9.525

 

o o
O O O O \O
51‘ $2 $2 3 §
100.0% . g ; ; , ~
on 80.0% ‘i—O—JMF : A 1
c: * ‘ ' ‘ :-
"" C 1+ Cracked Area ‘
§ 60.0% g *
g. --D-Non-Cracked Area; 1 W A
«3 40.0% § ' f l 4 *
g a t i
20.0% _ , , 1 ,
. i 4 ;
‘1' 05 M 0 ON. 00 \0 MB V3
1‘ ‘1' 6 V? "2 M. 1‘. eri O.
o C: ox .—
Sieve opening size (mm)
(a) Test site 1 (30202-11)
US standard sieve size
0 o
O O O O
Q! 3 $8 Q
100.0% . 3 i
. ,_ 3 t 1
1
w80.0% f .+JMF ,
.g l l—A—Cracked Area
=3 600% g-D— Non-Cracked Area ;
g 40.0% i 1
fl
20.0% '
0.0%

12.7
19.05

  
    

 

 

(b) Test site 2 (30202-12)

Figure 4.38 Gradation analysis of M-20, test section 1

211

 

 

 

US standard sieve size

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

   
   

 

o O .E.S .E
(:31 2 f?» 2 .2 oo <1- QQ 3
:14: in: :11: :11: :11: =13: 1121: M'— M
100.00/ * . -
° 1 -- :«r l r ‘1 *1 g g! _ :-
on 80.0% 5, ’ JMF ' * ~;:,r__ i ' e-
E H -A- Segregated .
16 60-0% ;i-‘D-Non-Segregated; . -_.fi z - “,1;
E . i i * V
0 40.00/ . s ; . _ , _
° ' z 1 l .- ,, 4 — l
G... _ g " 7 fl :
20.0% : ' : 7 g _
~ 1 i
i 3 1
0.0% .7 . 3 i ,. ,
v a m o as oo 0 ln[\ m
1‘ V o' ”1 "I "1 1‘. NN O.
O '— l0
:2 o o —~
Sieve opening size (mm)
(a) Test site 1 (30552-11)
US standard sieve size
8 8 o o ”$7 7
N —- tn m 2 oo <1- °\°Q 3
0 :11: :13: in: =11: :13: :13: :13: m— M
... i
on 80.0% a, JMF l - i
.5 ;1-D—Non-Segregated ‘ ,
w l '* 1 ~
g 60.0% ’ 1 f ' E ,
*5 ; - i
8 40.0% «~57 7 7771* j - —~ -g z_.
8 1 '
a. i 1 z . :
20.0% : _ “it“ ~: +.-.—— l - "4
I ’ 7’ 1 E
0.0% l , -: z i _ 1__ - ,,
V O\ M 05 O\ OO \O lnl\ tn
‘\ V o' l“. "i "E 1‘. mm o.
O '— In
. . o -— N v .— ox
O O 0‘ v-*

Sieve opening size (mm)

 

Figure 4.39 Gradation analysis of M-55, test section 1

(b) Test site 2 (30552-12)

212

 

 

mainly used to aid in the mixing of laboratory test specimens as discussed in section 13.0
below.

It should be noted that the asphalt content data were obtained through incineration
of the asphalt cores by weighing the sample before and after incineration. Loss of fines
through the chimney may have contributed some errors in the data although the asphalt

content data are very reasonable and consistent with MDOT specifications.

12.0 MOISTURE SUSCEPTIBILITY TESTING

The primary objective in the compacting of laboratory test specimens was to
study the effect of segregation on the tensile strength of the mix. Several of the
pavements exhibiting TDC also show segregation, the presence of which was continued
by gradation analyses. Although the strength and hence the fatigue performance tends to
decrease with increasing severity of segregation, researchers investigating the effects of
segregation often use aggregate gradations that are much more extreme than those
observed in this study.

Small changes in gradation, especially in the loss of fine materials as is often
observed on segregated pavements can result in increases in the air void contents of the
mix. This accelerates aging/hardening of the binder, and opens the surface of the
pavement to environmental damage. To investigate the loss of tensile strength of
segregated and non-segregated mixtures due to the impact of moisture, moisture
susceptibility tests were conducted according to ASTM-4867M (1994 Annual Book of

ASTM Standards).

213

Test specimens for moisture susceptibility tests were compacted in the laboratory

according the procedure outlined in chapter 3, section 5.0. For each mix included in this

study, at least 14 specimens were made; half of those were representative of the JMF or

non-segregated (control) mix in the field, and the other half were representative of

segregated mix. The mass of each specimen was set at 1000—g of aggregate, plus asphalt

cement to bring the total mass of each specimen to between 1040-g and 1065-g. This

mass is consistent with a 100-mm diameter specimen compacted to about 60—mm in

height. The 14 specimens were divided into five triplicate groups as follows:

1.

Three specimens were made in accordance with the JMF gradation and asphalt
content and were not subjected to moisture conditioning

Three specimens were made in accordance with the J MF gradation and asphalt
content and were subjected to moisture conditioning

Three specimens were made in accordance with the segregated gradation and
asphalt content and were not subjected to moisture conditioning.

Three specimens were made in accordance with the segregated gradation and
asphalt content and were subjected to moisture conditioning.

Two specimens, one from the control or JMF gradation and asphalt content and
the other from the segregated specimens, were used to determine the theoretical
maximum specific gravity of the mix.

Each specimen was then subjected to an indirect tensile strength test. According

to the test procedure, the mix is considered susceptible to moisture if the retained strength

ratio falls below 90%. The calculation for retained strength is as follows:

SC

214

 

Where: RSR = retained strength ratio;
S = indirect tensile strength of the compacted mix; and
SC = conditioned indirect tensile strength of the compacted mix.

The retained indirect tensile strengths of the non-segregated and segregated
mixtures were compared to determine if segregation in the field was responsible for a
significant loss in tensile strength. For these experiments, three different mixes were
tested:

1. 4B mix, the gradations and asphalt contents were obtained from M-55. The 4B

mix is prevalent among the test sites exhibiting TDC.

2. 4C mix which was obtained from a variety of sources. The 4C mix is also
prevalent among test sites exhibiting TDC.
3. 41310 , the gradations and asphalt contents were obtained from I-75. The 4E10 is a
common surface course SUPERPAVE mixture used on rubblized pavements.
Figure 4.40 depicts the aggregate gradations of the three mixes. Results of the tests are
shown in Figure 4.41. Figure 4.41a shows the average indirect tensile strengths of
triplicates for three AC mixes, while Figure 4.41b shows the percent retained strength
relative to the unconditioned specimens. These results are discussed for each mix in the

following sections.

12.1 48 Mixture
The laboratory mix design for the 4B mix was obtained by incinerating field cores
extracted from the pavement on M-55 and the quality control data. The mixes were

designed according to the J MF, and to the segregated gradation and asphalt content

215

 

 

80% :3 +413 JMF (5. 8% AC)

   
  

.,1_T_,,.
'1ft’1f’

 

 

 

   
   

 

  

 

 

 

   
 
 

 

 

CD 1
a .
can 1 t
g 60% 1'11 —D—4B Segregated (4.9% AC) f
g 40% 1 — 1 5:
a: 20% : i £22
0% . 1 ’7 141-L.
0.01 0 1 1 10 100
Sieve opening size (mm)
(a) 4B
100% 1 - 1- -3
g0 80% +4C JMF (6. 0%)
g 60% —o— 4C Segregated (4. 7% AC)
;5, 40% . 27:32:: “'11
8 t , 1
5‘: 20% .
0% 9
0.01 0.1 1 10 100
Sieve opening size (mm)
(b) 4C
100% 1;} ' s i... ...Ht : 7 15.1..1TT.-#.;:;;7;1 1 . :71 it 17:1:1j1
g0 80% , - j +4E10 Control (4. 6% AC) '11:
g 60% j —13— 4B Segregated (4.1% AC) 5114111
E 40% 1 ‘ f __ 171;; C71,
8 1 :1 ‘t::
a: 20% s s
0% ’ ? "3“
0.01 0.1 1 10 100
Sieve opening size (mm)
(c) 4E10
Figure 4.40 Gradations of mixtures used in moisture susceptibility tests

216

 

 

 

 

 

 

300

 

 

 

 

 

 

 

 

 

 

 

 

 

'3:
E: 250
5
if,” 200
£3 150 r
2
B 100
E
€ 50 ‘ 1
.E
0
JMF Seg JMF Seg JMF Seg
4B 4C 4510
Mix type
D Unconditioned I Conditioned
(a) Average indirect tensile strengths of triplicates
100%
90%
g 80%
E 70%
g0 60%
g 50%
g 40%
g 30%
g 20%
10%
0% ~
JMF Seg JMF Seg JMF Seg
4B 4C 4510
Mixtype

El Unconditioned I Conditioned 1

 

(b) Retained strength ratio of triplicates for 3 AC mixes

Figure 4.41 Average indirect tensile strength and averaged retained strength ratio of
unconditioned and conditioned triplicates of 3 AC mixes

217

 

 

obtained by incinerating pavement cores. The two gradations used in this analysis are
shown on Figure 4.40a.

Comparison between the test results can be seen in Figure 4.41. As can be seen,
the unconditioned tensile strengths between the JMF and the segregated samples are for
all purposes identical. This runs contrary to the results presented by Khedaywi and White
and Brown et al. The gradations used by these researchers were extremely segregated,
and cannot be regarded as representative of the segregation observed on M-55 and other
pavements.

The data in Figure 4.41b indicate that after moisture conditioning, the retained
strength of the J MP is about 68 percent of the unconditioned J MF. Whereas the retained
strength of the segregated mix is about 58 percent relative to the unconditioned
segregated mix. The two observations indicate that the 43 mix is moisture susceptible
and segregation accelerated moisture damage. Therefore, even if the mixture on M-55
was not segregated to any degree, the strength of the mix, and therefore the fatigue
resistance, would be compromised by the seasonal freeze-thaw cycles. This observation
may explain why several pavements (M-20 in particular) show no segregation but exhibit
TDC. The surface course on M-ZO is also a 4B mix. Recall that the observed tensile
strength of unconditioned specimens obtained from M-ZO exhibited lower than typical
tensile strengths (93.4-psi for test site 1 and 82.8-psi for test site 2). The results of the 43

indicate that moisture susceptibility is the likely cause of TDC.

218

12.2 4C Mixture

The laboratory mix design for the 4C mix was obtained from data collected by
Chang and quality control data from Michigan Avenue in Jackson. The mix gradations
were obtained from J MF and by incineration of recovered cores. The asphalt contents
were obtained from the JMF and by equivalent binder film thickness calculations. The
two gradations used in this analysis are shown on Figure 4.40b.

A comparison between the non-conditioned JMF and segregated samples are
shown in Figure 4.41. Unlike the 4B mixture, there is a noticeable difference in strength
(15%) between the IMF and segregated samples. Comparison of the JMF and segregated
mixtures after moisture conditioning shows that the segregated and conditioned mixture
has approximately 66% of the strength of the conditioned JMF mix. This indicates that
the segregated mixture is more susceptible to moisture damage than the JMF, which was
expected.

For the JMF and segregated mixtures subjected to moisture conditioning, the
segregated mix shows a significant reduction in the tensile strength due to moisture
conditioning. In this case, the tensile strength was reduced by an average of 21%. Unlike
the 4B mixture, the tensile strength of the J MF mix before and after conditioning is
essentially identical. This indicates that the 4C mixture is not susceptible to moisture
damage. It has been shown however, that even relatively mild segregation in the 4C mix

may result in a significant loss (33%) of strength over the JMF mixture.

219

 

12.3 4E10 Mixture

The laboratory mix design for the 4E10 mix was obtained from data collected by
Chang and quality control measurements on I-75 near Rondo. The mix gradations were
obtained from incineration of recovered cores. In this case, the JMF gradation was
replaced by a control gradation observed in the field. This choice was made because
significant differences occur in the small particles sizes between the J MF and the actual
pavement. The asphalt contents were obtained from the JMF and by equivalent binder
film thickness calculations. The two gradations used in this analysis are shown on Figure
4.40c along with the J MF gradation.

The behavior of the 4E10 mixture is similar to that of the 4C mixture, as is
evidenced in Figures 4.41. The non-conditioned segregated mixture actually shows a
higher tensile strength (7%) than the non-conditioned, non-segregated mixture but also
shows a significantly higher loss of tensile strength due to moisture conditioning. The
segregated mixture lost approximately 30% of its strength due to moisture damage, while
the non-segregated mixture lost only about 10%. Hence, the test results indicate that the

segregated mixture is susceptible to moisture damage.

13.0 SUMMARY OF DATA ANALYSES

Results of field and laboratory investigations and the results of the data analyses
are presented and discussed. Field investigation includes distress survey, nuclear density
and FWD testing, and pavement coring. The laboratory investigation includes specific
gravity testing, ITCLT, ITST, incineration of field cores and the effects of segregation

and moisture damage on the tensile strengths of three asphalt mixtures.

220

CHAPTER 5

SUMMARY CONCLUSIONS AND RECOMMENDATIONS

1.0 SUMMARY

Top—down cracks (TDC) are longitudinal and/or transverse cracks that initiate at
the pavement surface and propagate downward and outward. They have been
increasingly observed in flexible pavements throughout the state of Michigan. TDC are
usually manifested as longitudinal cracks appearing just outside the wheel paths. Over
time, they form an extensive network of longitudinal cracks connected by short
transverse cracks, which ultimately reduce the life of the pavement. Currently, no
pavement design method is capable of predicting or analyzing top-down cracking
potential although efforts are being made to include such model in the new mechanistic-
based AASHlTO design procedure.

TDC adversely affect pavement performance. Hence, understanding the factors
that enhance their potential would be the first step that needs to be taken to improve the
service life of the pavements.

To determine the causes of TDC, this study was initiated by the Michigan
Department of Transportation and the Pavement Research Center of Excellence at
Michigan State University under the auspices of the Thompson Scholars program, which
is directed by the Michigan Asphalt Paving Association. The study included extensive
field and laboratory investigations of various pavements exhibiting TDC. The main

objective was to verify the hypothesis that TDC are fatigue cracks that are formed due to

221

high tensile stress relative to the low tensile strength of the AC mix at the pavement
surface.

Pavements suspected of exhibiting TDC were identified and were subjected to
field investigation. Over the course of the study, eighteen test sites were investigated;
twelve flexible and six rubblized pavements. For each test site, a manual distress survey
was conducted and the pavement was subjected to FWD testing and coring to verify the
existence of TDC. On some test sites, nuclear density tests were also conducted to verify
the existence of segregation.

The manually collected distress data were analyzed against the various
degrees of segregation. A TDC propagation model was developed and presented.

In the laboratory, the cores were carefully measured and the overall density of the
core was determined. Then, the cores were subjected to IT CLT and ITST after which
they were incinerated to determine the AC content and to recover the aggregates and
determine their gradation. The data were examined against the nuclear density reading
and the visual observations of segregation. The three sets of data generally agreed.

Laboratory compacted test specimens of three asphalt mixes were made according
to the JMF and to the asphalt content and aggregate gradation obtained from incinerated
cores. Some test specimens were subjected to moisture susceptibility and then to ITST
while others were subjected to ITST directly.

The FWD data were used to backcalculate the pavement layer moduli using the
new algorithm and procedure that were developed by Svasdisant. Results of the
backcalculation were used as input to mechanistic analyses. The data were also used to

examine the repeatability of the data and the linearity of the pavement response to load.

222

The 2-D and 3-D mechanistic analyses were conducted by Svasdisant to assess
the sensitivity of the load-induced tensile stresses at the AC surface and bottom. Tire-
pavement interaction data were obtained from the tire industry and used in the 3-D
analyses to determine their impact on the lateral stresses induced at the AC surface and
bottom.

Finally, Svasdisant obtained pavement temperature data from MDOT and used it
to calculate daily and yearly variations in thermal stresses. His results are included in this

dissertation for completion.

2.0 CONCLUSIONS
Based on the field and laboratory investigations, on data analyses and on the
results of the 2-D and 3-D mechanistic analyses, the following conclusions were drawn:

1. TDC follows predictable patterns of development. The development can be
divided into three phases.

2. TDC can be divided into three different categories; longitudinal, transverse and in
segregated areas. Each category has different causes.

3. A TDC identification procedure was developed and verified. The procedure could
be used by State highway agencies to identify TDC and to select proper pavement
rehabilitation options.

4. Most TDC were observed in segregated pavements although few outside the
segregated areas.

5. Segregation and moisture damage substantially increase TDC potential.

223

10.

11.

12.

13.

14.

This implies that either different factors are present to cause TDC in segregated
and non-segregated pavements, or that some factor links the TDC potentials of
segregated and non—segregated pavements.

For segregated areas along flexible pavements, a TDC propagation model was
developed and tested based on pavement surface age and degree of segregation.
Nuclear density tests can be used to confirm visual identification of segregation.
TDC initiation and propagation are independent of the pavement cross-section or
the AC thickness.

Rubblized pavements show more variations in the FWD deflection than
conventional flexible pavements. Said variations are the direct results of variation
in the structural capacity along and across the pavement.

The depth to stiff layer used in backcalculation can be better estimated by using
Bousinessq’s equation and backcalculation of layer moduli using 3 and 4-layer
systems.

Pavements exhibiting TDC have low indirect tensile strengths. Segregation would
aggravate the situation.

AC content was not a conclusive indicator of segregation in this study.
Segregation opens the mixture to moisture damage. This occurs regardless of the
mix type. Moisture damage causes as much as 32 percent reduction in the asphalt
mix strength. Segregation causes as much as 15 percent reduction in strength.
Finally, segregation and moisture damage cause as much as 42 percent reduction

in strength.

224

15. TDC are mainly caused by load coupled with strength reduction due to
segregation and moisture damage.

16. Cold temperatures cause transverse TDC.

3.0 RECOMMENDATIONS
Based on the results and conclusions of this study, it is strongly recommended
that:
1. Crack initiation and/or propagation be specified as a part of design/build
warranty standards.
2. Incorporate TDC appearance criteria to identify cracks for inclusion in the
PMS database.
3. All asphalt mixes should be evaluated regarding segregation and moisture
susceptibility.
4. Explore more in depth the effects of segregation and aging on the properties

of the asphalt mixes.

225

10.

11.

REFERENCES

. Annual Book of ASTM Standards - Section 4 - Construction, 1994, Vol. 04.03 Road

and Paving Materials; Paving Management Technologies, The American Society for
Testing and Materials, Philadelphia, Pennsylvania.

American Association of State Highway and Transportation Officials, “AASHTO
Guide for Design of Pavement Structures”, pp. 106-112, Washington DC, 1993

Bensalem, A., A. J ., Broen, M. E. Nunn, D. B. Merrill, and W. G. Lloyd, “Finite
Element Modeling of fully Flexible Pavements: Surface Cracking and Wheel
Interaction,” Proceedings of the Second International Symposium on 3D Finite

Element For Pavement Analysis, Design, and Research, West Virginia, 2000, pp.
103-121

Bonaquist, R., Surdahl, R., and W. Mogawer, “Effect of Tire Pressure on Flexible
Pavement Response and performance,” In Transportation Research Record 122 7,
TRB, National Research Council, Washington, DC, 1989, pp 97-106.

Brown, ER. and CE. Bassett, “Effects of Maximum Aggregate Size on Rutting
Potential and Other Properties of Asphalt-Aggregate Mixtures,” In Transportation
Research Record 1259, TRB, National Research Council, Washington, DC, 1990,
pp. 107-119

Brown, E.R., Ronald, C., and J .R. Brownfield, “Investigation of Segregation of
Asphalt Mixtures in the State of Georgia,” In Transportation Research Record 1217,
TRB, National Research Council, Washington, DC, 1989, pp. 1-8

Button , J .W., Perdomo, D, and R.L. Lytton, “Influence of Aggregate on Rutting in
Asphalt concrete Pavements,” In Transportation Research Record 1 25 9, TRB,
National Research Council, Washington, DC, 1990, pp. 141-152

Chang, C.M., “Detecting Segregation in Bituminous Pavements and Relating its
Effects to Performance,” A PhD. Thesis, Department of Civil and Environmental
Engineering, Michigan State University, 2000.

Chapuis, RP. and A. Gatien, “Temperature Dependent Tensile strength of Asphalt
Mixtures in Relation to field Cracking Data,” ASTM STP 1265, 1995, pp. 180-193

Chatti, K., Kim, H.B., Un, K.K., Mahoney, J .P., and CL. Monismith, “Field
Investigation into Effects of Vehicle Speed and Tire Pressure on Asphalt Concrete

Pavement Strains,” In Transportation Research Record 1 53 9, TRB, National
Research Council, Washington, DC, 1996, pp 66-71

Clark, S. M., Mechanics of Pneumatic Tires, US. Department of Transportation,
National Highway Traffic Safety Administration, Washington, D. C., 1981

226

12.

13.

14.

15.

16.

17.

l8.

19.

20.

21.

Dai, S.T., Van Deusen, D., Beer, M., Rettner D., and G. Cochran, “Investigation of
Flexible Pavement Response to Truck Speed and FWD Load through Instrumented

Pavements,” Proceedings of the 8th International Conference on Asphalt Pavements,
Seattle, 1997, pp. 141-160

Dauzats, M., and A. Rampal, “Mechanism of Surface Cracking in Wearing Courses,”
Proceedings of the 6th International Conference on Asphalt Pavements, Ann Arbor,
1987, pp. 232-247

De Beer, M. and C. Fisher, “Contact Stresses of Pneumatic Tires Measured with the
Vehicle-Road Surface Pressure Transducer Array (VRSPTA) System for the
University of California at Berkley (UCB) and the Nevada Automotive Test Center
(NATC)”, Volume 1 and 2, 1997

De Beer, M., Fisher, C., and R]. Jooste, “Determination of Pneumatic Tyre/Pavement
Interface Contact Stresses Under Moving Loads and Some Effects on Pavement with
Thin Asphalt Surfacing Layers,” Proceedings of the 8th International Conference on
Asphalt Pavements, Seattle, 1997, pp. 179-227

Gerritsen, A.H., van Gurp, C.A.P.M., van der Heide, J P] ., Molenaar, A.A.A., and
AC. Pronk, “Prediction and Prevention of Surface Cracking in Asphaltic

Pavements,” Proceedings of the 6‘h International Conference on Asphalt Pavements,
Ann Arbor, 1987, pp. 378-391

Groenendijk, J ., “Accelerated testing and Surface Cracking of Asphaltic Concrete
Pavements,” A PhD dissertation. Department of Civil Engineering, Delft University
of Technology, the Netherlands, 1998.

Harichandran R. and G.Y. Baladi, “Integrated Material and Structural Design Method
for Flexible Pavements," Volumes I and II, final report, FHWA/RD-88/109 and 110,
Washington, DC. September 1987

Harvey, J .T., and B.W. Tsai, “Effects of Asphalt Content and Air Void Content on
Mix Fatigue and Stiffness.” In Transportation Research Record 1543, TRB, National
Research Council, Washington, DC, 1996, pp. 38-45

Heukelom, W., “Observations on the Rheology and Fracture of Bitumens and Asphalt
Mixes,” Proceedings of the Association of Asphalt Paving Technologists, Volume 35,
1966, pp. 358-399

Hiltunen DR. and R. Roque, “A Mechanics-Based Prediction Model for Thermal
Cracking of Asphalt Concrete Pavements,” Proceedings Association of Asphalt
paving Technologists, Volume 63,1994, pp.8l-117

227

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

Himeno, K., Ikeda, T., Kamijima, T., and T. Abe, “Distribution of Tire Contact
Pressure of Vehicles and its Influence on Pavement Distress,” Proceedings of the 8‘h
International Conference on Asphalt Pavements, Seattle, 1997, pp. 129-139

Huang, Y.H., “Pavement Analysis and Design,” Prentice-Hall Inc., New Jersey, 1993.

Hugo, F., and T.W. Kennedy, “Surface Cracking of Asphalt Mixtures in Southern
Africa,” Proceedings of the Association of Asphalt Paving Technologists, Volume 54,
1985, pp 454-501

Huhtala, M., Pihlajamaki, J ., and M. Pienimaki, “Effects of tires and Tire Pressure on
Roads Pavements,” In Transportation Research Record 1227, TRB, National
Research Council, Washington, DC, 1989, pp. 107-114

Jacobs, M., de Bondt, A., Molenaar, A., and P. Hopman, “Cracking in Asphalt
Concrete Pavements,” Proceedings of the 7‘h International Conference on Asphalt
Pavements, 1992, pp. 89-105

Jacobs, M.M.J., Hopman, RC, and A.A.A. Molenaar, “Application of Fracture
mechanics and Principles to Analyze Cracking in Asphalt Concrete (With

Discussion),” Proceedings of the Association of Asphalt Paving Technologists,
Volume 65, 1996, pp 1-39

Jung, D. and T. S. Vinson, “Thermal stress Restrained Specimen Test to Evaluate
Low Temperature Cracking of Asphalt-Aggregate Mixtures.” In Transportation
Research Record 1417, National Research Council, Washington, DC, 1993, pp. 12-
20

Khedaywi, TS. and TD. White, “Development and Analysis of Laboratory
Techniques for Simulating Segregation,” In Transportation Research Record 1492,
TRB, National Research Council, Washington, DC, 1995, pp 36-45

Littlefield, G., “Thermal Expansion and Contraction Characteristics Utah Asphaltic

Concrete,” Proceedings of the Association of Asphalt Paving Technologists, Volume
36, 1967, pp. 673-702

Malan, G.W., Straus, P.J., and F. Hugo, “A Field Study of Premature Surface
Cracking in Asphalt (With Discussion),” Proceedings of the Association of Asphalt
Paving Technologists, Volume 58, 1988, pp 142-162

Markwick, A.H.D., and J .H. Starks, “Stresses Between Tire and Road.” Journal of the
Institution of Civil Engineers, Vol. 16,1940-1941, pp 309-325

Marker, V., “Symposium-Compaction of Asphalt Concrete,” Proceedings of the
Association of Asphalt Paving Technologists, Volume 36, 1967, pp. 288-294

228

 

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

Matsuno, S., and T. Nishizawa, “Mechanism of Longitudinal Surface Cracking in
Asphalt Pavement,” Proceedings of the 7th International Conference on Asphalt
Pavements, Nottingham, 1992, pp. 277-291

Mohammad, L.N. and HR. Paul, “Evaluation of Indirect Tensile Test for
Determining Structural Properties of Asphalt Mix,” In Transportation Research
Record 141 7, TRB, National Research Council, Washington, DC, 1993, pp. 58-63

Molenaar, A.A.A., “Fatigue and Reflective Cracking due to Traffic (With
Discussion),” Proceedings of the Association of Asphalt Paving Technologists,
Volume 53, 1984, pp 440-474

Myers, L.A., Roque, R., and BE. Ruth, “Mechanisms of Surface-Initiated
Longitudinal Wheel Path Cracks in High-Type Bituminous Pavements,” Proceedings
of the Association of Asphalt Paving Technologists, Volume 67, 1998, pp 401-432

Myers, L.A., Roque, R., Ruth, BE, and C. Drakos, “Measurement of Contact
Stresses for Different Truck Tire Types to Evaluate Their Influence on Near-Surface
Cracking and Rutting,” In Transportation Research Record 1655, TRB, National
Research Council, Washington, DC, 1999, pp. 175-184

Myers, L.A., Roque, R., and B. Birgisson, “Propagation Mechanisms for Surface-
Initiated Longitudinal Wheel Path Cracks,” In Transportation Research Record 1 778,
TRB, National Research Council, Washington, DC, 2001, pp. 113-121

Myers, L. and R. Roque, “Top-Down Crack Propagation in Bituminous Pavements
and Implications for Pavement management,” Proceedings of the Association of
Asphalt Paving Technologists, Volume 71, 2002, p 651-670

Niederquell, M.G., Baladi, G.Y., and K. Chatti, “Rubblization of Concrete
Pavements: Field Investigation,” In Transportation Research Record 1 730, TRB,
National Research Council, Washington, DC, 2001, pp. 150-160

Pottinger M., “The Three-Dimensional Contact Patch Stress Field of Solid and

Pneumatic Tires,” Tire Science and Technology, TSTCA, volume 20 number 1, 1992,
pp. 3-32.

Pottinger, M.G., and J .E. McIntrye, “Effects if suspension Alignment and Modest
Cornering on the footprint Behavior of Performance Tires and Heavy Duty Radial
Tires,” Tire Science and Technology, TSTCA, volume 27 number 3, 1999, pp. 128-
160.

Rickards, I., Goodman, S., Pagani, J ., Halim, A.O.A.E., and R. Haas, Practical

Realization of a New concept for Asphalt Compaction,” In Transportation Research
Record 1 730, TRB, National Research Council, Washington, DC, 1999, pp. 27-35

229

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

Roque, R., and B. E. Ruth., “Mechanisms of Modeling of Surface Cracking in
Asphalt Pavements,” Proceedings of the Association of Asphalt Paving
Technologists, Volume 59, 1990, pp. 396-421

Roque, R., Myers, L.A., and B. Birgisson, “Evaluating Measured Tire Contact
Stresses to Predict Pavement Response and Performance.” In Transportation
Research Record 1716, TRB, National Research Council, Washington, DC, 2000,
pp 73-81

Schapery, R. A. and J. T. Tielking. Investigation of Tire-Pavement Interaction During
Maneuvering, Vol. 1, Theory and Results, US. Department of Transportation,
Fedearal Highway Admiistration, Washington, DC, 1977

Schorsch, M., Chang, C.M., and G.Y. Baladi, “Effects of Segregation on the Initiation
and Propagation of Top-Down Cracks,” accepted for presentation at the TRB annual
meeting, Washington, D. C., 2003

Shahin, Y., “Design System for Minimizing Asphalt Concrete Thermal Cracking,”
The 4‘h International Conference Structure Design of Asphalt Pavement, Ann Arbor,
Volume 1, 1977, pp. 920-932.

Siddharthan, RV. and PE. Sebaaly, “Investigation of Asphalt Concrete Layer Strains
from Wide-Base Tires,” In Transportation Research Record 1655, TRB, National
Research Council, Washington, DC, 1999, pp 168-174

Siddharthan, R.V., N. Krishnamenon, M. El-Mously and PE. Sebaaly, “Investigation
of Tire Contact Stress distributions on Pavement Response,” Journal of
Transportation Engineering, Volume 128, No. 2, 2002, pp. 136-144

Soon, S. C., “Tire-Induced Surface Stresses in Flexible Pavements”, Accepted for
Publication in Transportation Research Record and Presented at TRB Annual
meeting, Washington, DC, 2003.

Svasdisant T., Schorsch, M., Baladi, G.Y., and S. Pinyosunun, “Mechanistic Analysis of
Top-down Cracks in Asphalt Pavement,” Accepted for presentation and publication by
the Transportation Research Board, Washington DC, 2001

Svasdisant T., and G. Baladi, “Identify Causes for Under Performing Rubblized
Concrete Pavement Projects Phase 11” Research Report RC-l4l6, Michigan Department
of Transportation, August 2002

Svasdisant T., “Analysis of Top-Down Cracking in Rubblized and Flexible

Pavements,” A PhD. Thesis, Department of Civil and Environmental Engineering,
Michigan State University, 2003.

230

56.

57.

58.

59.

60.

61.

62.

Tielking, J .T., and M.A. Abraham, “Measurement of Truck Tire Footprint Pressure.”
In Transportation Research Record 1435, TRB, National Research Council,
Washington, DC, 1994, pp 92-99

Teilking, J .T. and F .L. Roberts, “Tire Contact Pressure and Its Effect on Pavement
Strain.” Journal of Transportation Engineering ASCE, Vol. 113, 1987, pp. 56-83

Uhlmeyer, J .S., Willoughby, K., Pierce, L.M., and J. P. Mahoney, “Top-Down
Cracking in Washington State Asphalt Concrete Wearing Courses,” In Transportation
Research Record 1730, TRB, National Research Council, Washington, DC, 2000,
pp. 110-116

Wambura, J .H.G., Maina, J ., and HR. Smith., ‘Kenya Bituminous Materials Study,”

In Transportation Research Record 1681, TRB, National Research Council,
Washington, DC, 1999, pp. 129-137

Williams, R.C., Duncan, G., and White, T.D., “Hot-Mix Asphalt Segregation:
Measurement and Effects.” In Transportation Research Record 1543, TRB, National
Research Council, Washington, DC, 1996, pp. 97-105

Yoder, E.J., and M.W. Witczak, "Principles of Pavement Design 2nd Edition", J.
Wiley and Sons, NY. NY, 1975

Zubeck, H. K. and T. S. Vinson, “Prediction of Low-Temperature Cracking of
Asphalt Concrete Mixtures with Thermal Stress Restrained Specimen Test Results,”
In Transportation Research Record 1545, TRB, National Research Council,
Washington, DC, 1996, pp. 50-58

231

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

llllllllllllljjl(1»ij