SPR 396-5

JUNE 2015

Performance Evaluation of
Arizona’s LTPP SPS-5 Project:
Strategic Study of Rehabilitation
of Asphalt Concrete Pavements

Arizona Department of Transportation Research Center

Performance Evaluation of Arizona’s
LTPP SPS‐5 Project:
Strategic Study of Rehabilitation of
Asphalt Concrete Pavements
SPR‐396‐5
June 2015

Prepared by:
Peter N. Schmalzer
Nichols Consulting Engineers
1885 S. Arlington Avenue, Suite 111
Reno, NV 89509‐3370

Steven M. Karamihas
The University of Michigan Transportation Research Institute
2901 Baxter Road
Ann Arbor, MI 48109

Timin Punnackal
Hans Meyer
Kevin Senn
Jason Puccinelli
Nichols Consulting Engineers
1885 S. Arlington Avenue, Suite 111
Reno, NV 89509‐3370

Published by:
Arizona Department of Transportation
206 S. 17th Avenue
Phoenix, AZ 85007
in cooperation with
US Department of Transportation
Federal Highway Administration

This report was funded in part through grants from the Federal Highway Administration, US Department
of Transportation. The contents of this report reflect the views of the authors, who are responsible for
the facts and the accuracy of the data, and for the use or adaptation of previously published material,
presented herein. The contents do not necessarily reflect the official views or policies of the Arizona
Department of Transportation or the Federal Highway Administration, US Department of
Transportation. This report does not constitute a standard, specification, or regulation. Trade or
manufacturers’ names that may appear herein are cited only because they are considered essential to
the objectives of the report. The US government and the State of Arizona do not endorse products or
manufacturers.

Technical Report Documentation Page
1. Report No.

2. Government Accession No

3. Recipient's Catalog No.

FHWA‐AZ‐15‐396(5)
4. Title and Subtitle

5. Report Date

Performance Evaluation of Arizona’s LTPP SPS‐5 Project: Strategic Study
of Rehabilitation of Asphalt Concrete Pavements

June 2015

7. Author(s)

8. Performing Organization Report No.

6. Performing Organization Code

Peter N. Schmalzer, Steven M. Karamihas, Timin Punnackal, Hans Meyer,
Kevin Senn, Jason Puccinelli
9. Performing Organization Name and Address

10. Work Unit No. (TRAIS)

Nichols Consulting Engineers
1885 S. Arlington Avenue
Suite 111
Reno, NV 89509‐3370

11. Contract or Grant No.

12. Sponsoring Agency Name and Address

13. Type of Report and Period Covered

Arizona Department of Transportation
206 S. 17th Avenue
Phoenix, AZ 85007

FINAL REPORT

SPR 000‐1(147) 396

14. Sponsoring Agency Code

15. Supplementary Notes

Prepared in cooperation with the US Department of Transportation, Federal Highway Administration.
16. Abstract

As part of the Long Term Pavement Performance (LTPP) Program, the Arizona Department of Transportation
(ADOT) constructed 11 Specific Pavement Study‐5 (SPS‐5) test sections on Interstate 8 near Casa Grande. The
SPS‐5 project studied a variety of different rehabilitation methods for asphalt concrete pavements. The
project was opened to traffic in 1990 and monitored at regular intervals until it was placed out of study in
2009. Surface distress, profile, and deflection data collected throughout the life of the pavement were used
to evaluate the performance of rehabilitation methods using unique combinations of minimal and intensive
surface preparation, virgin and recycled asphalt, and thick and thin overlay thicknesses. This report
documents the analyses conducted as well as practical findings and lessons learned that will be of interest to
ADOT.
17. Key Words

18. Distribution Statement

LTPP, pavement performance,
rehabilitation, profile, distress,
FWD, flexible, AC, deflections,
roughness, backcalculation

Document is available to the U.S. public
through the National Technical
Information Service, Springfield, VA,
22161

19. Security
Classification

20. Security
Classification

21. No. of Pages

Unclassified

Unclassified

178

23. Registrant's Seal

22. Price

SI* (MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS
Symbol

When You Know

Multiply By
LENGTH

in
ft
yd
mi

inches
feet
yards
miles

in
2
ft
2
yd
ac
2
mi

2

square
square
square
acres
square

fl oz
gal
3
ft
3
yd

fluid ounces
gallons
cubic feet
cubic yards

oz
lb
T

ounces
pounds
short tons (2000 lb)

o

Fahrenheit

fc
fl

foot-candles
foot-Lamberts

lbf
2
lbf/in

poundforce
poundforce per square inch

Symbol

When You Know

mm
m
m
km

millimeters
meters
meters
kilometers

25.4
0.305
0.914
1.61

To Find

Symbol

millimeters
meters
meters
kilometers

mm
m
m
km

square millimeters
square meters
square meters
hectares
square kilometers

mm
2
m
2
m
ha
2
km

milliliters
liters
cubic meters
cubic meters
3
shown in m

mL
L
3
m
3
m

grams
kilograms
megagrams (or "metric ton")

g
kg
Mg (or "t")

AREA
inches
feet
yard

645.2
0.093
0.836
0.405
2.59

miles

2

VOLUME
29.57
3.785
0.028
0.765
NOTE: volumes greater than 1000 L shall be

MASS
28.35
0.454
0.907

TEMPERATURE (exact degrees)
F

5 (F-32)/9
or (F-32)/1.8

Celsius

o

lux
2
candela/m

lx
2
cd/m

C

ILLUMINATION
10.76
3.426

FORCE and PRESSURE or STRESS
4.45
6.89

newtons
kilopascals

N
kPa

APPROXIMATE CONVERSIONS FROM SI UNITS
Multiply By
LENGTH
0.039
3.28
1.09
0.621

To Find

Symbol

inches
feet
yards
miles

in
ft
yd
mi

AREA
2

mm
2
m
2
m
ha
2
km

square millimeters
square meters
square meters
hectares
square kilometers

0.0016
10.764
1.195
2.47
0.386

square
square
square
acres
square

inches
feet
yards
miles

2

in
2
ft
2
yd
ac
2
mi

VOLUME
mL
L
3
m
m3

milliliters
liters
cubic meters
cubic meters

0.034
0.264
35.314
1.307

g
kg
Mg (or "t")

grams
kilograms
megagrams (or "metric ton")

o

Celsius

fluid ounces
gallons
cubic feet
cubic yards

fl oz
gal
3
ft
yd3

ounces
pounds
short tons (2000 lb)

oz
lb
T

MASS
0.035
2.202
1.103

TEMPERATURE (exact degrees)
C

1.8C+32

Fahrenheit

o

foot-candles
foot-Lamberts

fc
fl

F

ILLUMINATION
lx
2
cd/m

lux
2
candela/m

N
kPa

newtons
kilopascals

0.0929
0.2919

FORCE and PRESSURE or STRESS
0.225
0.145

poundforce
poundforce per square inch

lbf
2
lbf/in

*SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380.
(Revised March 2003)

iv

Contents
EXECUTIVE SUMMARY ......................................................................................................................... 1
CHAPTER 1. INTRODUCTION ................................................................................................................ 3
CHAPTER 2. SPS‐5 DEFLECTION ANALYSIS........................................................................................... 15
Normalized Deflections .......................................................................................................................... 16
AASHTO 1993 Analysis ............................................................................................................................ 19
Backcalculation ....................................................................................................................................... 59
Deflection Analysis Key Findings ............................................................................................................ 85
CHAPTER 3. SPS‐5 DISTRESS ANALYSIS ............................................................................................... 87
AC Distress Types .................................................................................................................................... 87
Research Approach ................................................................................................................................. 88
Overall Performance Trend Observations .............................................................................................. 89
Distress Analysis Key Findings .............................................................................................................. 101
CHAPTER 4. SPS‐5 ROUGHNESS ANALYSIS ........................................................................................ 107
Profile Data Synchronization ................................................................................................................ 107
Data Extraction ..................................................................................................................................... 108
Cross Correlation .................................................................................................................................. 108
Longitudinal Distance Measurement ................................................................................................... 110
Data Quality Screening ......................................................................................................................... 111
Summary Roughness Values ................................................................................................................. 118
Profile Analysis Tools ............................................................................................................................ 124
Distress Surveys and Maintenance Records ......................................................................................... 131
Detailed Observations .......................................................................................................................... 131
Profile Analysis Key Findings ................................................................................................................ 131
CHAPTER 5. CONCLUSIONS AND RECOMMENDATIONS .................................................................... 137
Deflection Analysis ............................................................................................................................... 137
Distress Analysis ................................................................................................................................... 138
Profile Analysis...................................................................................................................................... 138
REFERENCES..................................................................................................................................... 141
APPENDIX A: CONSTRUCTION DEVIATIONS ...................................................................................... 143
APPENDIX B: SITE WORK HISTORY ................................................................................................... 145
APPENDIX C: ROUGHNESS VALUES ................................................................................................... 148
APPENDIX D: DETAILED OBSERVATIONS ........................................................................................... 153

v

List of Figures
Figure 1. Location of Arizona SPS‐5 Test Sections ........................................................................................ 4
Figure 2. Layout of the SPS‐5 Project ............................................................................................................ 5
Figure 3. Existing Asphalt Concrete Pavement Cross Section Before SPS‐5 Project Construction
(from Hossain et al. 1996) .............................................................................................................. 7
Figure 4. Subgrade Resilient Modulus versus Station, Section 040507 ...................................................... 24
Figure 5. Subgrade Resilient Modulus versus Date, Section 040507.......................................................... 24
Figure 6. SNeff versus Station, Section 040507 ............................................................................................ 25
Figure 7. SNeff versus Date, Section 040507................................................................................................ 26
Figure 8. Subgrade Resilient Modulus versus Station, Section 040504 ...................................................... 27
Figure 9. Subgrade Resilient Modulus versus Date, Section 040504.......................................................... 28
Figure 10. SNeff versus Station, Section 040504 .......................................................................................... 29
Figure 11. SNeff versus Date, Section 040504.............................................................................................. 29
Figure 12. Subgrade Resilient Modulus versus Station, Section 040503 .................................................... 31
Figure 13. Subgrade Resilient Modulus versus Date, Section 040503........................................................ 31
Figure 14. SNeff versus Station, Section 040503 .......................................................................................... 32
Figure 15. SNeff versus Date, Section 040503.............................................................................................. 32
Figure 16. Mr versus Station, Section 040508............................................................................................. 34
Figure 17. Mr versus Date, Section 040508 ................................................................................................ 35
Figure 18. SNeff versus Station, Section 040508 .......................................................................................... 35
Figure 19. SNeff versus Date, Section 040508.............................................................................................. 36
Figure 20. Mr versus Station, Section 040509............................................................................................. 38
Figure 21. Mr versus Date, Section 040509 ................................................................................................ 38
Figure 22. SNeff versus Station, Section 040509 .......................................................................................... 39
Figure 23. SNeff versus Date, Section 040509.............................................................................................. 39
Figure 24. Mr versus Station, Section 040502............................................................................................. 41
Figure 25. Mr versus Date, Section 040502 ................................................................................................ 41
Figure 26. SNeff versus Station, Section 040502 .......................................................................................... 42
Figure 27. SNeff versus Date, Section 040502.............................................................................................. 42
Figure 28. Mr versus Station, Section 040506............................................................................................. 44
Figure 29. Mr versus Date, Section 040506 ................................................................................................ 44
Figure 30. SNeff versus Station, Section 040506 .......................................................................................... 45
Figure 31. SNeff versus Date, Section 040506.............................................................................................. 45
Figure 32. Mr versus Station, Section 040505............................................................................................. 47
Figure 33. Mr versus Date, Section 040505 ................................................................................................ 47
Figure 34. SNeff versus Station, Section 040505 .......................................................................................... 48
Figure 35. SNeff versus Date, Section 040505.............................................................................................. 48
Figure 36. Mr versus Station, Section 040559............................................................................................. 50
Figure 37. Mr versus Date, Section 040559 ................................................................................................ 50
Figure 38. SNeff versus Station, Section 040559 .......................................................................................... 51
Figure 39. SNeff versus Date, Section 040559.............................................................................................. 51

vi

Figure 40. Mr versus Station, Section 040560............................................................................................. 53
Figure 41. Mr versus Date, Section 040560 ................................................................................................ 53
Figure 42. SNeff versus Station, Section 040560 .......................................................................................... 54
Figure 43. SNeff versus Date, Section 040560.............................................................................................. 54
Figure 44. Mr versus Station, Section 040501............................................................................................. 56
Figure 45. Mr versus Date, Section 040501 ................................................................................................ 56
Figure 46. SNeff versus Station, Section 040501 .......................................................................................... 57
Figure 47. SNeff versus Date, Section 040501.............................................................................................. 57
Figure 48. Subgrade Modulus Increase versus SNeff Increase ..................................................................... 81
Figure 49. Measured versus Synthetic Deflection Basins ........................................................................... 83
Figure 50. Hysteresis Plot, Section 040507 (Stiff Layer Phenomenon)....................................................... 85
Figure 51. Hysteresis Plot, Section 040559 (Stiff Layer Phenomenon Not Observed) ............................... 85
Figure 52. Structural and Environmental Distress Trends in Section 040501 ............................................ 91
Figure 53. Structural and Environmental Distress Trends in Section 040502 ............................................ 91
Figure 54. Structural and Environmental Distress Trends in Section 040503 ............................................ 91
Figure 55. Structural and Environmental Distress Trends in Section 040504 ............................................ 91
Figure 56. Structural and Environmental Distress Trends in Section 040505 ............................................ 92
Figure 57. Structural and Environmental Distress Trends in Section 040506 ............................................ 92
Figure 58. Structural and Environmental Distress Trends in Section 040507 ............................................ 92
Figure 59. Structural and Environmental Distress Trends in Section 040508 ............................................ 92
Figure 60. Structural and Environmental Distress Trends in Section 040509 ............................................ 93
Figure 61. Structural and Environmental Distress Trends in Section 040559 ............................................ 93
Figure 62. Structural and Environmental Distress Trends in Section 040560 ............................................ 93
Figure 63. 2002 Structural Performance of All SPS‐5 Sections ................................................................... 94
Figure 64. 2002 Environmental Performance of All SPS‐5 Sections ........................................................... 95
Figure 65. 2005 Structural Performance of SPS‐5 Core Sections ................................................................ 95
Figure 66. 2005 Environmental Performance of SPS‐5 Core Sections ........................................................ 96
Figure 67. 2002 Rutting Index Summary of All SPS‐5 Sections ................................................................... 96
Figure 68. 2005 Rutting Index Summary of SPS‐5 Core Sections................................................................ 97
Figure 69. Fatigue Cracking and AC Material Thicknesses .......................................................................... 98
Figure 70. Consistency in Longitudinal Distance Measurement ............................................................... 110
Figure 71. IRI Progression of Section 040501 ........................................................................................... 118
Figure 72. IRI Progression of Section 040502 ........................................................................................... 119
Figure 73. IRI Progression of Section 040503 ........................................................................................... 119
Figure 74. IRI Progression of Section 040504 ........................................................................................... 120
Figure 75. IRI Progression of Section 040505 ........................................................................................... 120
Figure 76. IRI Progression of Section 040506 ........................................................................................... 121
Figure 77. IRI Progression of Section 040507 ........................................................................................... 121
Figure 78. IRI Progression of Section 040508 ........................................................................................... 122
Figure 79. IRI Progression of Section 040509 ........................................................................................... 122
Figure 80. IRI Progression of Section 040559 ........................................................................................... 123
Figure 81. IRI Progression of Section 040560 ........................................................................................... 123
vii

Figure 82. PSD of Section 040502 Profiles (Left Side) ............................................................................... 125
Figure 83. PSD of Section 040508 Profiles (Left Side) ............................................................................... 126
Figure 84. Raw Profiles of Section 040509 ............................................................................................... 127
Figure 85. Filtered Profiles of Section 040509 .......................................................................................... 128
Figure 86. Long‐Wavelength Profiles of Section 040560 .......................................................................... 128
Figure 87. Roughness Profile of Section 040503 (25‐ft Base Length) ....................................................... 130
Figure 88. Roughness Profiles of Section 040509 (10‐ft Base Length) ..................................................... 130
Figure C‐1. Comparison of HRI to MRI. ................................................................................................... 1499
Figure D‐1. Roughness Profiles of Section 040506 (10‐ft Base Length).................................................... 162

viii

List of Tables
Table 1. Arizona SPS‐5 Project Layout, Post‐Rehabilitation.......................................................................... 6
Table 2. Summary of SPS‐5 Test Sections (from Hossain et al. 1996) .......................................................... 7
Table 3. Summary of Preconstruction Distress in Travel Lane ..................................................................... 8
Table 4. Summary of Section Performance .................................................................................................. 9
Table 5. Climatic Information for SPS‐5 ...................................................................................................... 12
Table 6. Traffic Loading Summary............................................................................................................... 13
Table 7. FWD Test Dates by Section ........................................................................................................... 15
Table 8. Normalized Deflections ................................................................................................................. 17
Table 9. Summary of AASHTO 1993 Analysis of FWD Data ........................................................................ 21
Table 10. Preconstruction SNeff Variation by Overlay Material .................................................................. 22
Table 11. Preconstruction SNeff Variation by Overlay Thickness ................................................................ 22
Table 12. Preconstruction SNeff Variation by Level of Surface Preparation ................................................ 23
Table 13. Layer Structure, Section 040507 ................................................................................................. 23
Table 14. Layer Structure, Section 040504 ................................................................................................. 27
Table 15. Layer Structure, Section 040503 ................................................................................................. 30
Table 16. Layer Structure, Section 040508 ................................................................................................. 33
Table 17. Layer Structure, Section 040509 ................................................................................................. 37
Table 18. Layer Structure, Section 040502 ................................................................................................. 40
Table 19. Layer Structure, Section 040506 ................................................................................................. 43
Table 20. Layer Structure, Section 040505 ................................................................................................. 46
Table 21. Layer Structure, Section 040559 ................................................................................................. 49
Table 22. Layer Structure, Section 040560 ................................................................................................. 52
Table 23. Layer Structure, Section 040501 ................................................................................................. 55
Table 24. SNeff Increase by Overlay Material .............................................................................................. 58
Table 25. SNeff Increase by Overlay Thickness ............................................................................................ 58
Table 26. SNeff Increase by Level of Surface Preparation ............................................................................ 59
Table 27. Summary of SNeff Increase by Experimental Factor .................................................................... 59
Table 28. Layer Models for Backcalculation, Section 040507..................................................................... 61
Table 29. Backcalculation Results, Section 040507 .................................................................................... 62
Table 30. Layer Models for Backcalculation, Section 040504..................................................................... 63
Table 31. Backcalculation Results, Section 040504 .................................................................................... 64
Table 32. Layer Models, Section 040503 .................................................................................................... 65
Table 33. Backcalculation Results, Section 040503 .................................................................................... 65
Table 34. Layer Models, Section 040508 .................................................................................................... 66
Table 35. Backcalculation Results, Section 040508 .................................................................................... 67
Table 36. Layer Models, Section 040509 .................................................................................................... 67
Table 37. Backcalculation Results, Section 040509 .................................................................................... 68
Table 38. Layer Models, Section 040502 .................................................................................................... 69
Table 39. Backcalculation Results, Section 040502 .................................................................................... 70
Table 40. Layer Models, Section 040506 .................................................................................................... 71

ix

Table 41. Backcalculation Results, Section 040506 .................................................................................... 71
Table 42. Layer Models, Section 040505 .................................................................................................... 72
Table 43. Backcalculation Results, Section 040505 .................................................................................... 73
Table 44. Layer Models, Section 040559 .................................................................................................... 74
Table 45. Backcalculation Results, Section 040559 .................................................................................... 74
Table 46. Layer Models, Section 040560 .................................................................................................... 75
Table 47. Backcalculation Results, Section 040560 .................................................................................... 76
Table 48. Layer Model, Section 040501 ...................................................................................................... 77
Table 49. Backcalculation Results, Section 040501 .................................................................................... 77
Table 50. Measured and Temperature‐Corrected AC Moduli .................................................................... 78
Table 51. Nonlinearity of AB Modulus ........................................................................................................ 78
Table 52. Nonlinear Layer Parameters ....................................................................................................... 79
Table 53. Nonlinear Backcalculation Results .............................................................................................. 79
Table 54. Postconstruction SNeff Increase ................................................................................................... 80
Table 55. Measured and Synthetic Deflection Basins ................................................................................. 82
Table 57. Flexible Pavement Distress Types and Failure Mechanisms ...................................................... 88
Table 58. MDS Dates by Section ................................................................................................................. 90
Table 59. Core Section Performance Against Structural Distresses ......................................................... 103
Table 60. Core Section Performance Against Environmental Distresses .................................................. 103
Table 61. Profile Measurement Visits of the SPS‐5 Site............................................................................ 107
Table 62. Selected Repeats of Section 040501 ......................................................................................... 112
Table 63. Selected Repeats of Section 040502 ......................................................................................... 112
Table 64. Selected Repeats of Section 040503 ......................................................................................... 113
Table 65. Selected Repeats of Section 040504 ......................................................................................... 113
Table 66. Selected Repeats of Section 040505 ......................................................................................... 114
Table 67. Selected Repeats of Section 040506 ......................................................................................... 114
Table 68. Selected Repeats of Section 040507 ......................................................................................... 115
Table 69. Selected Repeats of Section 040508 ......................................................................................... 115
Table 70. Selected Repeats of Section 040509 ......................................................................................... 116
Table 71. Selected Repeats of Section 040559 ......................................................................................... 116
Table 72. Selected Repeats of Section 040560 ......................................................................................... 117
Table 73. Summary of Roughness Behavior ............................................................................................. 134
Table C‐1. Roughness Values .................................................................................................................. 1499

x

List of Abbreviations, Acronyms, and Symbols
AASHTO
AB
AC
ADOT
ADS
ARAC
DCP
ESAL
FHWA
FWD
HMA
HMAC
HRI
IRI
LTPP
MDS
MEPDG
ML
MP
Mr
MRI
N/A
NWP
OGFC
OWP
PSD
RAP
RMSE
RN
SHRP
SM
SNeff
SPS
SPT
WP

American Association of State Highway and Transportation Officials
aggregate base
asphalt concrete
Arizona Department of Transportation
automated distress survey
asphalt rubber asphalt concrete
dynamic cone penetrometer
equivalent single‐axle load
Federal Highway Administration
falling weight deflectometer
hot‐mix asphalt
hot‐mix asphalt concrete
Half‐car Roughness Index
International Roughness Index
Long Term Pavement Performance
manual distress survey
Mechanistic‐Empirical Pavement Design Guide
midlane
milepost
resilient modulus
Mean Roughness Index
not available, not applicable
non‐wheel path
open‐graded friction course
outer wheel path
power spectral density
recycled asphalt pavement
root mean square error
Ride Number
Strategic Highway Research Program
select material
effective structural number
Specific Pavement Study
Standard Penetration Test
wheel path

xi

Acknowledgments
The project team would like to acknowledge the Arizona Department of Transportation for sponsoring
this project. In addition, the authors thank the Arizona Transportation Research Center and the
Technical Advisory Committee for their input, and thank Mr. Christ Dimitroplos for his leadership. The
comprehensive information stored in the Long Term Pavement Performance Program database allowed
this research to be conducted.

xii

EXECUTIVE SUMMARY
As part of the Long Term Pavement Performance (LTPP) Program, the Arizona Department of
Transportation (ADOT) constructed 11 Specific Pavement Study‐5 (SPS‐5) test sections on Interstate 8 (I‐
8) near Casa Grande. The SPS‐5 experiment was designed to study the effect of specific
maintenance/rehabilitation treatments on asphalt concrete (AC) pavement performance. This project
consisted of two sets of test sections: nine core sections matching similar projects constructed by other
highway agencies and two supplemental sections to investigate alternative design characteristics
selected by ADOT. Construction of all 11 sections was completed in April 1990, and 10 of the sections
(all except the control section) were not placed out of study until a major rehabilitation in January 2009.
Some years earlier, in October 1996, the control section had been placed out of study first, having
reached the end of its service life and consequently needing reconstruction.
This report provides general information about the project location, including climate, traffic, and
subgrade conditions, as well as details about the layer configurations of each test section. All 11 of the
SPS‐5 test sections were constructed consecutively and so were exposed to the same traffic‐loading,
climate, and subgrade conditions. This allows direct comparisons among unique combinations of surface
preparation, overlay material, and overlay thicknesses without the confounding effects introduced by
different in situ conditions.
By comparing changes in effective structural numbers and structural distresses, both the deflection and
distress analyses concluded that (1) thick (5‐inch) AC overlays performed better than thin (2‐inch) AC
overlays, (2) virgin AC overlays performed better than recycled AC overlays, and (3) test sections that
had received intensive surface preparation performed better than those that had received minimal
surface preparation. The distress analysis concluded that the use of 5‐inch overlays was the most critical
rehabilitation feature in improving the structural performance of a pavement. Additionally, the distress
analysis concluded that the use of virgin AC was the most critical rehabilitation feature in improving a
pavement’s resistance to environmental distresses.
Key findings from the deflection analysis were largely based on the results obtained using the AASHTO
1993 analysis method. An elastic layer backcalculation was also performed as part of the deflection
analysis. However, the elastic layer backcalculation produced highly variable and unreasonable results
that should be used with appropriate discretion. Profile analysis results showed that roughness and
roughness progression alone cannot be used to represent the health of a test section. Several test
sections did not exhibit changes in roughness in proportion to the amount of fatigue cracking they
displayed.
All sections received pavement maintenance consisting of two to three fog seals and one crack seal.
These maintenance events can mask the extent and severity of actual distresses in the pavement in the
short term. However, in this study the impact of maintenance work on distresses was not significant and
was considered negligible.

1

This long‐term study of pavement performance in Arizona was initiated to help develop improvements
to pavement design and retrofit procedures. Including profile analysis, falling weight deflectometer, and
distress analysis, forensic analyses on the LTPP samples documented the performance range of various
pavements. Combined with ADOT’s Pavement Management Services (PMS) data and then compared to
pavement prediction models to improve pavement design methodologies, the LTPP data sets were used
to calibrate to local conditions the American Association of State Highways and Transportation Officials
(AASHTO) Mechanistic‐Empirical Pavement Design method, which improves pavement design and will
lead to more reliable and cost‐effective pavements.

2

CHAPTER 1. INTRODUCTION
Understanding the contribution of maintenance and rehabilitation procedures to long‐term pavement
performance can be extremely valuable to pavement managers looking to optimize resources and
improve overall performance. The objectives of this research were to document the overall performance
trends of the Specific Pavement Study‐5 (SPS‐5) project, identify key differences in performance among
the various rehabilitation techniques, and document key findings that would be useful to the Arizona
Department of Transportation (ADOT).
This report provides the results of surface distress, deflection, and profile analyses for the Long Term
Pavement Performance (LTPP) Program SPS‐5 site on Interstate 8 (I‐8) near Casa Grande, Arizona. The
SPS‐5 project (040500) studied the effect of specific maintenance and rehabilitation treatments on
asphalt concrete (AC) pavement performance. The project consisted of 11 test sections—nine core
sections and two supplemental sections. The nine core sections represented the standard experimental
matrix that the Strategic Highway Research Program (SHRP) required. ADOT added the two
supplemental test sections to evaluate features that were not included in the SHRP experiment design.
In this report, the test sections are generally referred to by their six‐digit SHRP project ID numbers (for
example, 040501), but some figures and tables use ADOT’s numbering system, which omits the first two
digits (for example, 0501).
In eight of the nine core SPS‐5 test sections (all except the control section), two types of surface
preparation were used:



Minimal preparation, which included partial patching, full‐depth patching, and milling of an
existing surface friction course (to a depth of less than 1 inch).
Intensive preparation, which included removing and replacing existing crack sealing, performing
crack sealing, removing and replacing existing partial and full‐depth patching, performing
additional partial and full‐depth patching, and milling to a depth such that the final surface was
at least 0.5 inch above or below an interface between material layers (Hossain et al. 1996). The
depth of material removed by milling, excluding any surface friction course, was replaced with
an equal thickness of virgin AC. This material was not counted as part of the overlay thickness.

The ninth core section was a control section that received only the routine maintenance required to
keep the roadway in a safe and functional condition. Four of the SHRP test sections received either a 2‐
or 5‐inch‐thick virgin asphalt overlay. The other four sections received either a 2‐ or 5‐inch‐thick recycled
asphalt overlay.
The two ADOT supplemental sections used the same types of surface preparation as the SPS‐5 test
sections used, but different overlay materials and thicknesses. Section 040559 received an inverted
overlay consisting of 3 inches of recycled overlay placed on 3 inches of virgin AC. Section 040560
received an asphalt rubber asphalt concrete (ARAC) overlay.

3

The SPS‐5
5 project was located on eaastbound I‐8 in Pinal Counnty (see Figuree 1) and was incorporated
d in
ADOT reh
habilitation prroject IR‐8‐2(9
91), which exxtended from milepost (MP) 147.60 to M
MP 160.87,
approximately 17 miles southwest of
o Casa Grand
de.
The SPS‐5
5 project was constructed in a two‐mile
e segment of tthis 13‐mile rrehabilitation project (from
m MP
159.01 to MP 161). The soil was covvered with vaarious desert‐‐type brush and small treees. Each test
except Section
n 040560) waas 500 ft in len
ngth, which ddoes not inclu
ude transition
nal segments
section (e
between sections and destructive sampling areaas outside thee monitoring limits. Averagge elevation o
of the
project was 1071 ft, with latitude off 32° 50' and longitude of ‐112° 00'. Figgures 1 and 2 show the loccation
and layou
ut of the SPS‐5
5 project, and
d Table 1 listss the test secttion propertiees.

AZ SPS
S-5

(Courttesy of Googlle Maps)
Figure 1.
1 Location off Arizona SPSS‐5 Test Sectio
ons

4

Figure 2. Layout of the SPS‐5 Project

5

Table 1. Arizona SPS‐5 Project Layout, Post‐Rehabilitation
Station
(ft)

(m)

0

0

SHRP
ID

040507

ADOT
ID

1

Asphalt Concrete
Thickness
(inches)

Type

3.7

existing AC

2.7

virgin AC fill

500

152

4.1

virgin AC overlay

764

233

3.8

existing AC

4.8

virgin AC overlay

3.0

existing AC

4.7

recycled AC overlay

4.2

existing AC

2.4

virgin AC fill

040504
1264

385

1544

471
040503

2044

623

2909

887
040508

1

1

1

3409

1039

4.1

recycled AC overlay

4322

1317

2.4

existing AC

2.6

virgin AC fill

040509

1

4822

1470

1.3

recycled AC overlay

5240

1597

2.6

existing AC

2.7

recycled AC overlay

4.3

existing AC

2.8

virgin AC fill

040502
5740

1750

6049

1844
040506

1

1

6549

1996

2.4

virgin AC overlay

6652

2028

2.7

existing AC

2.8

virgin AC overlay

1.7

existing AC

3.0

virgin AC fill

040505
7152

2180

7436

2266
040559

1

1

7936

2419

3.0

recycled AC overlay

8375

2553

4.1

existing AC

2.2

ARAC overlay

4.1

existing AC

0.9

OGFC

040560
8975

2736

9154

2790
040501

9654

1

1

2943

6

Base

Subgrade

Thickness
(inches)

Type

Type

20.7

coarse‐soil
aggregate
mix

silty gravel
with sand

17.6

coarse‐soil
aggregate
mix

silty gravel
with sand

16.6

coarse‐soil
aggregate
mix

silty gravel
with sand

15.0

coarse‐soil
aggregate
mix

silty gravel
with sand

14.8

coarse‐soil
aggregate
mix

silty gravel
with sand

14.7

coarse‐soil
aggregate
mix

silty gravel
with sand

12.8

coarse‐soil
aggregate
mix

silty gravel
with sand

12.8

coarse‐soil
aggregate
mix

silty gravel
with sand

13.2

coarse‐soil
aggregate
mix

silty gravel
with sand

14.0

coarse‐soil
aggregate
mix

silty gravel
with sand

14.2

coarse‐soil
aggregate
mix

silty gravel
with sand

Before the SPS‐5 project was constructed, the roadway was 38 ft wide with two travel lanes that were
each 12 ft wide bounded by a 10‐ft outside shoulder and a 4‐ft inside shoulder. The pavement section
consisted of 8 inches of select material (SM), 6 inches of aggregate base (AB) and 4.5 inches of asphalt
concrete (AC). The existing surface course was an open‐graded friction course (OGFC) of 0.5 inch. A
typical cross section of the road before construction is shown in Figure 3 (Hossain et al. 1996).

Figure 3. Existing Asphalt Concrete Pavement Cross Section Before SPS‐5 Project Construction
(from Hossain et al. 1996)

As mentioned previously, this project consisted of 11 test sections. Table 2 summarizes the features of
each section.
Table 2. Summary of SPS‐5 Test Sections (from Hossain et al. 1996)
Location

Surface Preparation1

Overlay
Material

Overlay
Thickness
(inches)

From

To

Length
(ft)

040507

0+00

5+00

500

Intensive

Virgin

5

040504

7+64

12+64

500

Minimal

Virgin

5

040503

15+44

20+44

500

Minimal

Recycled

5

040508

29+09

34+09

500

Intensive

Recycled

5

040509

43+22

48+22

500

Intensive

Recycled

2

040502

52+40

57+40

500

Minimal

Recycled

2

040506

60+49

65+49

500

Intensive

Virgin

2

040505

66+52

71+52

500

Minimal

Virgin

2

040559

74+36

79+36

500

Intensive

Recycled

3

040560

83+75

89+75

600

Minimal

ARAC

2

040501

91+54

96+54

500

Routine maintenance

None

None (control)

SHRP ID

1 Minimal preparation: (1) milling off existing OGFC (less than 1 inch) and (2) partial and full‐depth patching.
Intensive preparation: (1) milling off existing OGFC and a portion of existing AC (from 1.9 to 3.5 inches),
(2) removing and replacing existing crack sealing and partial and full‐depth patching, and (3) performing additional
crack sealing and partial and full‐depth patching.

7

Researchers encountered some construction issues during the milling of Section 040502. The first passes
with the milling machine did not completely remove the friction course, and the remaining material was
badly stripped. Crews also encountered difficulties with using a 6‐ft milling machine as opposed to a
machine that could span the entire 12‐ft lane. See Appendix A for a complete list of construction
deviations.
Before the SPS‐5 experiment was constructed, researchers reviewed the existing distresses in the travel
lane of the test sections. Table 3 provides a summary of this review, which shows the overall poor
condition of the pavement. The majority of the distress in each section was block cracking. Fatigue
cracking was also observed in every section. The control section, 040501, did not receive an overlay and
deteriorated to a very poor condition. Section 040501 was placed out of study within six years because
it required reconstruction.

Table 3. Summary of Preconstruction Distress in Travel Lane

SHRP ID

Fatigue
Cracking
(ft2)

Block
Cracking
(ft2)

Longitudinal
Cracking
(ft)

Transverse
Cracking
(ft)

040501

966

5820

0

0

040502

1110

5484

0

0

040503

541

6073

0

0

040504

229

4148

40

242

040505

125

4582

61

221

040506

661

5876

0

0

040507

620

6021

0

0

040508

475

6283

0

0

040509

327

6274

0

0

040559

1139

5715

0

0

040560

384

7928

0

0

Notes:
1. All distress measurements were taken with an automated distress survey
using a lane width of 15 ft.
2. Sections 040501 through 040559 were 500 ft long; Section 040560 was 600 ft
long.

Table 4 provides a summary of falling weight deflectometer (FWD) deflection measurements, fatigue
and block cracking levels, and International Roughness Index (IRI) values for the core test sections.
(Fatigue and block cracking were measured with both automated distress surveys [ADS] and manual
distress surveys [MDS].) These values indicate the condition of each section before receiving an AC
overlay (preconstruction), after receiving the overlay (postconstruction), and during the last survey
taken before each section was placed out of study. This table helps illustrate the overall performance of

8

the test sections, which is discussed in greater detail in Chapters 2, 3, and 4 of this report. These
chapters discuss the deflection, distress, and roughness analyses conducted for this project.
Table 4. Summary of Section Performance
Deflections2
SHRP ID

Survey Event

Preconstruction
0405013

Postconstruction

Last survey

Preconstruction

040502

Postconstruction

Last survey

Preconstruction

040503

Postconstruction

Last survey

Preconstruction

040504

Postconstruction

Last survey

Survey Dates1
1/19/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
1/15/1991 (MDS),
9/21/1990 (IRI)
9/13/1996 (FWD),
10/20/1994 (MDS),
2/23/1993 (IRI)
1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/19/1994 (MDS),
9/21/1990 (IRI)
9/15/2008 (FWD),
9/15/2008 (MDS),
9/11/2008 (IRI)
1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/15/1991 (FWD),
10/18/1994 (MDS),
9/21/1990 (IRI)
12/10/2003 (FWD),
12/5/2005 (MDS),
9/11/2008 (IRI)
1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/15/1991 (FWD),
10/18/1994 (MDS),
9/21/1990 (IRI)
12/10/2003 (FWD),
12/5/2005 (MDS),
9/11/2008 (IRI)

D60
D0
(mils) (mils)

Fatigue
Cracking
(ft2)

Block
Cracking
(ft2)

IRI
(inches/mi)

12.7

1.1

966

5820

78

14.9

1.1

1992

4000

79

9.7

0.8

2781

3272

86

10.7

0.8

1110

5484

127

5.7

0.7

4

0

86

9.6

0.9

3684

2550

244

12.9

0.9

541

6073

107

6.4

0.7

0

0

60

7.2

1.3

104

0

146

10.7

0.8

229

4148

98

3.8

0.6

0

0

76

2.9

0.7

2

0

108

1 Manual distress survey (MDS) used a lane width of 12 ft; automated distress survey (ADS) used a lane width
of 15 ft.
2 D0: Deflection of the pavement under the center of the loading plate. D60: Deflection of the pavement
60 inches from the center of the loading plate.
3 Section did not receive an overlay and was placed out of study within 6 years.
4 Section length was 600 ft.

9

Table 4. Summary of Section Performance (Continued)
SHRP ID

Survey Event

Preconstruction

040505

Postconstruction

Last survey

Preconstruction

040506

Postconstruction

Last survey

Preconstruction

040507

Postconstruction

Last survey

Preconstruction

040508

Postconstruction

Last survey

Survey Dates

1

1/19/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/25/1994 (MDS),
9/21/1990 (IRI)
9/15/2008 (FWD),
9/15/2008 (MDS),
9/11/2008 (IRI)
1/19/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/20/1994 (MDS),
9/21/1990 (IRI)
9/15/2008 (FWD),
9/15/2008 (MDS),
9/11/2008 (IRI)
1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/14/1991 (FWD),
10/18/1994 (MDS),
9/21/1990 (IRI)
12/10/2003 (FWD),
12/5/2005 (MDS),
9/11/2008 (IRI)
1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/15/1991 (FWD),
10/19/1994 (MDS),
9/21/1990 (IRI)
12/10/2003 (FWD),
12/5/2005 (MDS),
9/11/2008 (IRI)

Deflections2
D60
D0
(mils) mils)

Fatigue
Cracking
(ft2)

Block
Cracking
(ft2)

IRI
(inches/mi)

11.3

1.1

125

4582

162

6.1

0.8

183

0

81

9.5

0.9

4285

1206

131

16.1

1.0

661

5876

110

4.5

0.7

0

0

65

8.9

1.3

65

0

134

8.3

0.7

620

6021

116

2.8

0.6

0

0

82

1.6

0.4

0

0

98

10.8

0.7

475

6283

98

4.8

0.6

0

0

60

5.7

1.1

0

0

87

1 Manual distress survey (MDS) used a lane width of 12 ft; automated distress survey (ADS) used a lane width
of 15 ft.
2 D0: Deflection of the pavement under the center of the loading plate. D60: Deflection of the pavement
60 inches from the center of the loading plate.
3 Section did not receive an overlay and was placed out of study within 6 years.
4 Section length was 600 ft.

10

Table 4. Summary of Section Performance (Continued)
SHRP ID

Survey Event

Preconstruction

040509

Postconstruction

Last survey

Preconstruction

040559

Postconstruction

Last survey

Preconstruction

0405604

Postconstruction

Last survey

Survey Dates

1

1/18/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/19/1994 (MDS),
9/21/1990 (IRI)
9/15/2008 (FWD),
9/15/2008 (MDS),
9/11/2008 (IRI)
1/19/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/20/1994 (MDS),
9/21/1990 (IRI)
12/12/2003 (FWD),
12/12/2003 (MDS),
9/11/2008 (IRI)
1/19/1990 (FWD),
11/29/1989 (ADS),
2/5/1990 (IRI)
1/16/1991 (FWD),
10/20/1994 (MDS),
9/21/1990 (IRI)
12/12/2003 (FWD),
12/12/2002 (MDS),
9/11/2008 (IRI)

Deflections2
D60
D0
(mils) (mils)

Fatigue
Cracking
(ft2)

Block
Cracking
(ft2)

IRI
(inches/mi)

8.9

0.8

327

6274

151

5.1

0.7

0

0

65

9.7

1.0

910

0

286

13.0

1.3

1139

5715

131

5.9

1.1

0

0

73

6.6

1.9

0

0

111

12.0

1.2

384

7928

116

12.6

1.2

0

0

50

10.6

1.2

1955

0

140

1 Manual distress survey (MDS) used a lane width of 12 ft; automated distress survey (ADS) used a lane width
of 15 ft.
2 D0: Deflection of the pavement under the center of the loading plate. D60: Deflection of the pavement
60 inches from the center of the loading plate.
3 Section did not receive an overlay and was placed out of study within 6 years.
4 Section length was 600 ft.

The climate for the SPS‐5 project is considered to be a dry, no‐freeze environment by LTPP definitions.
Table 5 provides environmental details about the area. The temperature and precipitation information
was derived from data collected at nearby weather stations and represents 38 years of recorded data.
The humidity data was summarized from 22 years of virtual weather station data.

11

Table 5. Climatic Information for SPS‐5
38‐year
Average

38‐year
Maximum

38‐year
Minimum

Annual average daily mean temperature (°F)

71

73

68

Annual average daily maximum temperature (°F)

87

90

85

Annual average daily minimum temperature (°F)

54

57

51

Absolute maximum annual temperature (°F)

115

120

110

Absolute minimum annual temperature (°F)

24

30

16

Number of days per year above 90° F

172

190

144

Number of days per year below 32° F

20

39

5

Annual average freezing index (°F‐days)

0

0

0

Annual average precipitation (inches)

8.7

Annual average daily mean solar radiation (W/ft2)

16.4

3.2

N/A

N/A

N/A

Annual average daily maximum relative humidity (%)

52

61

43

Annual average daily minimum relative humidity (%)

20

24

16

N/A: No data available.

Table 6 provides a summary of the total equivalent single‐axle loads (ESALs) computed from traffic
loading information collected at the SPS‐5 site. For 1990 to 1992, and from 1994 to 1996, no traffic
monitoring data were available. For these years, ADOT provided estimates for ESAL values. In 2000,
neither monitored nor estimated ESAL values were available. The ADOT traffic estimates are provided to
illustrate the expected traffic growth as modeled by the agency, but these values do not necessarily
correlate with monitored traffic data. For example, the estimated ESAL values provided by ADOT in 2001
and 2002 were two to three times larger than the monitored values.

12

Table 6. Traffic Loading Summary
Year

Monitored ESALs

Estimated ESALs

1990

N/A

250,000

1991

N/A

180,000

1992

N/A

220,000

1993

148,400

N/A

1994

N/A

200,000

1995

N/A

200,000

1996

N/A

200,000

1997

256,200

N/A

1998

347,200

N/A

1999

407,400

N/A

2000

N/A

N/A

2001

335,400

761,000

2002

269,000

880,000

2003

298,800

N/A

2004

390,100

N/A

2005

369,500

N/A

2006

246,500

N/A

2007

263,200

N/A

2008

371,700

N/A

N/A: No data available.

Three analyses were conducted to evaluate pavement performance of the SPS‐5 project. Chapters 2, 3,
and 4 of this report describe these three analyses—deflection, distress, and roughness. Each chapter
provides a description of the research approach along with performance comparisons among test
sections, overall trends, a summary of the results, and key findings.
In 2008, the Federal Highway Administration (FHWA) initiated a forensic evaluation of four of the nine
test sections (040502, 040505, 040506, and 040509). The objectives of this forensic investigation were
to: (1) identify causes of pavement failure and investigate associated distress mechanisms, (2) examine
pavement structural and functional performance, (3) measure within‐section layer thicknesses and
material properties, and (4) test end‐state physical properties. Destructive and nondestructive tests
were conducted, including coring, trenching, dynamic cone penetrometer (DCP) testing, Standard
Penetration Tests (SPT), laboratory testing of materials, distress surveys, transverse profiles, longitudinal
profiles, and falling weight deflectometer (FWD) tests (Nichols Consulting Engineers, unpublished data,
2010).

13

14

CHAPTEER 2. SPS‐5
5 DEFLECTION ANALLYSIS
Prior to co
onstruction, each
e
SPS‐5 te
est section con
nsisted of appproximately 5 inches of assphalt concrete
(AC) on 15
5 inches of agggregate base
e (AB). Constrruction proceedures varied across the seections, with one
control se
ection receivin
ng no treatment and other sections recceiving variou
us combinatio
ons of milling and
overlay with
w different types
t
of AC materials
m
as discussed in C hapter 1.
One round of FWD testing was perfformed on each test sectioon approximaately three mo
onths before
construction. Another round of FWD testing wass performed aapproximatelly nine month
hs after
construction. Each section then received furtherr FWD testingg every one to
o five years until it went out of
ounds—as it w
went
study. Secction 040501 (the control section) received the leastt testing—a ttotal of five ro
out of study first. The sections
s
invesstigated in the forensic stuudy in 2008 reeceived 12 ro
ounds of FWD
D
testing. Th
he remainingg sections rece
eived 11 roun
nds of FWD teesting. Table 7 shows the d
dates when eeach
section re
eceived FWD testing.

Table
T
7. FWD Test Dates b
by Section
Test
T Date

0501
0

0502

0503

0504

0505

1/18/1990‐
1/19/1990
1/14/1991‐
1/16/1991
10
0/2/1991‐
10
0/3/1991
10
0/18/1994‐
10
0/20/1994
9/11/1996‐
9/13/1996
11
1/12/1997‐
11
1/14/1997
12
2/9/1998‐
12
2/10/1998
12
2/13/1999‐
12
2/14/1999
12
2/16/2000‐
12
2/18/2000
12
2/10/2002‐
12
2/12/2002
12
2/10/2003‐
12
2/12/2003
9/15/2008

15

05066

0507

05508

0509

0
0559

0560

The preconstruction round of testing was performed in January 1990. At that time, the LTPP FWD
guidelines were still under development, and there are some differences between data collected before
construction and subsequent data. The most noteworthy difference is the test point interval—for the
preconstruction testing, the interval was nominally 100 ft, while for subsequent testing the interval was
nominally 50 ft. Also, only three replicate drops were conducted at each drop height during
preconstruction testing, as opposed to four replicates for later testing.
In 1999, another change was made to the FWD testing procedure. Two sensors were added to the FWD:
one sensor at 48 inches from the center of the loading plate and one at ‐12 inches.
NORMALIZED DEFLECTIONS
Table 8 presents average normalized deflections measured by the center sensor (D0) of the FWD and
the sensor at 60 inches from the center (D60) for both the midlane (ML) and outer wheel path (OWP).
This data is for drop height 2, which is nominally 9000 lb, then normalized to exactly 9000 lb using the
following equation:

dn 

Where

dm  P
9000

(Eq. 1)

dn = normalized deflection (mils)
dm = measured deflection (mils)
P = drop load (pounds)

Normalized deflection is the simplest method of analyzing FWD data. While its utility in analyzing
pavement structures is limited, this method does offer an easy way to make qualitative comparisons,
which can be very helpful in identifying areas to focus on in more sophisticated and time‐consuming
analyses. The issues raised in the following discussion of the normalized deflection data will be revisited
later in this report.
The deflection reading at 60 inches from center (D60) should be sensitive only to the stiffness of the
subgrade. The zone of influence of an FWD test can be approximated as a 45‐degree cone with its apex
at the center of the load plate. Since the thickness (including AC and AB) of the thickest test section in
this project (Section 040507) was 29.9 inches, any deflection reading from a sensor with an offset of
greater than 30 inches would be assumed to be independent of the pavement layers. For most tests in
this data set, only two sensors have an offset of greater than 30 inches—those at 36 and 60 inches.
Since 36 inches is still close to the zone of influence, the sensor at 60 inches was selected to represent
the subgrade.

16

Table 8. Normalized Deflections

Section

1

2

Test Date

Surface
Preparation

1/18/1990
040507

1/14/1991

5.0
Intensive

12/10/2003
1/18/1990
040504

1/15/1991
1/18/1990
1/15/1991

Minimal

1/18/1990
1/15/1991

Intensive

1/18/1990
1/16/1991

Intensive

1/18/1990
1/16/1991

Minimal

1/19/1990
1/16/1991

Intensive

1/19/1990
1/16/1991

Minimal

1/19/1990
1/16/1991

Intensive

1/19/1990
1/16/1991

6.9
5.2

12/12/2003
040560

8.2
5.0

9/15/2008
040559

6.4
4.9

9/15/2008
040505

6.5
5.1

9/15/2008
040506

9.2
5.4

9/15/2008
040502

8.9
5.4

12/10/2003
040509

9.2
5.0

12/10/2003
040508

9.2
5.0

Minimal

12/10/2003
040503

AC
Thickness
(inches)

7.7
5.0

Minimal

12/12/2003

6.3

1/19/1990
040501

1/16/1991

N/A

5.0

9/13/1996

Midlane
D0
(mils)

Outer Wheel Path

D60
(mils)

D0
(mils)

D60
(mils)

8.3

0.8

13.3

0.8

2.8

0.6

3.0

0.5

1.6

0.4

1.8

0.4

10.7

0.8

12.3

0.8

3.8

0.6

4.2

0.6

2.9

0.5

3.8

0.5

13.0

0.9

18.1

1.0

6.4

0.7

8.3

0.7

7.2

0.9

9.4

0.8

10.8

0.8

15.1

0.9

4.9

0.6

5.1

0.6

5.7

0.8

5.5

0.6

8.9

0.8

14.0

0.8

5.1

0.7

6.6

0.6

9.8

0.8

13.8

0.9

10.7

0.8

13.7

0.7

5.7

0.7

7.0

0.7

9.6

0.7

14.4

0.7

16.2

1.0

15.7

1.1

4.5

0.7

4.8

0.7

8.9

0.9

9.9

0.9

11.3

1.1

15.5

1.2

6.2

0.8

7.5

0.8

9.5

0.8

13.4

0.8

13.1

1.3

15.2

1.3

5.9

1.1

6.5

1.0

6.6

1.5

5.9

1.2

12.1

1.2

17.5

1.3

12.6

1.2

16.6

1.3

10.6

1.0

15.1

1.0

12.7

1.1

16.2

1.2

14.9

1.1

16.4

1.1

9.8

0.8

12.3

0.8

1 Sections are listed in order of physical position at the test site.
2 Three FWD test dates are listed for each section: (1) before construction, (2) the first FWD test
performed after construction, and (3) the last test available for that section.

17

From the D60 measurements, it is apparent that the subgrade is very stiff. In many cases, the D60
readings are less than 1 mil, which is problematic because the random error inherent in the geophones
used to measure these deflections is stated by the manufacturer to be ±0.08 mil. In some cases this
represents 20 percent of the D60 measurement. This indicates that for subsequent analyses, data from
the higher load levels should be used to reduce the effect of random error. While the 9000‐lb load level
is typically used in pavement design because it represents half of an 18‐kip ESAL (i.e., one set of tires on
an axle loaded to 18,000 lb), for thick sections on stiff subgrades such as this, the ability to determine
the subgrade response independently of the pavement structure at this load level is hampered by the
precision of the deflection measurement devices.
Some of the changes between preconstruction and postconstruction D60 measurements are quite large,
especially for Sections 040507, 040504, and 040508. These are the thickest pavement sections, but they
are also all located toward the beginning of the project layout. The most likely causes for changes in
apparent subgrade stiffness are changes in moisture content or stress state. Thicker and stiffer
pavements decrease the deviator stress and increase the confining pressures in the subgrade, which for
most subgrade materials will increase stiffness. So it is reasonable that the thickest sections also exhibit
the greatest increase in subgrade modulus.
Deflection data for both the ML and OWP are presented in Table 8. Typically, FWD data is used to make
fatigue‐related predictions about pavement performance (either explicitly in mechanistic design
methods, or implicitly in empirical methods), so the data from the OWP, which undergoes direct loading,
is most relevant. It is expected that the D0 deflections in the OWP will be higher than those in the ML
because the D0 deflections are sensitive to pavement stiffness. These differences increase as the
pavement deteriorates. It is also expected that there will be little difference between the ML and OWP
measurements in the D60 data since this data should only be sensitive to the subgrade, which is not
generally considered to undergo fatigue‐related changes in stiffness. In addition, as the zone of
influence of a load applied to the pavement surface increases in diameter with increasing depth, the
deflection response of the deeper layers is less sensitive to small changes in the point of application of
the load, which reduces the apparent distinction between subgrade properties measured in the ML and
OWP.
In the preconstruction data, the average normalized D0 deflection measurement is 23 percent higher in
the OWP than in the ML. This difference decreases to 14 percent after construction. In the sections that
received intensive surface preparation, the difference between OWP and ML decreases to 10 percent,
while in the sections that received minimal preparation, the difference decreases to only 18 percent. In
the preconstruction D60 data, the OWP measurement is 4 percent higher than the ML value. After
construction, the OWP is 4 percent lower than the ML. For the sections that received intensive
preparation, the OWP is 9 percent lower than the ML, while for those that received minimal
preparation, the OWP is 2 percent higher than the ML. Again, this discrepancy between the trends in the
D60 data and expectations may be due to the influence of random error on very small measurements,
or could indicate that the D60 data are not free of influence from the pavement layers as they were
assumed to be.

18

AASHTO 1993 ANALYSIS
The FWD data were analyzed using the procedure described in the 1993 AASHTO Guide for Design of
Pavement Structures (AASHTO 1993). For this analysis, the FWD load and deflection data collected at the
same location and nominal load were averaged before subsequent processing. Data for the OWP and
drop height 4 (16,000 lb nominal) were used. Subgrade resilient modulus was calculated using the
following equation for the sensor at 60 inches:

Mr 

Where

0.24  P
r  dr

(Eq. 2)

P = load (pounds)
r = sensor offset (inches)
dr = deflection (inches) at offset r

This subgrade resilient modulus value does not include the 0.33 correction that is commonly applied
when the results of FWD analysis are used for pavement design.
The effective pavement modulus is the modulus that represents the entire structure of the pavement
above the subgrade. Researchers calculated this value iteratively using the following equation:

1

1

2

D
1  

1
a


d 0  1.5 pa
2

Ep
 D Ep 

 3

 M r 1  a M 
r 




Where

d0 = center deflection (inches)
p = load plate pressure (psi)
a = load plate radius (inches)
D = total thickness of pavement layers (inches)
Mr = subgrade resilient modulus (psi)
Ep = effective pavement modulus (psi)

19












(Eq. 3)

The effective structural number, SNeff, was calculated using the following equation:

SN eff  0.0045  t  3 E p
Where

(Eq. 4)

t = pavement thickness (inches)
Ep = effective pavement modulus (psi)

The AC component of the pavement structure has a temperature‐dependent stiffness. The AB
component does not. Because the effective pavement modulus includes both AC and AB components,
temperature correction is somewhat problematic. The AASHTO 1993 guide includes a simple empirical
method (Figure 5.6 on page III‐99) for adjusting d0 based on the AC mid‐depth temperature and the
thickness of the AC layer. For this adjustment, mid‐depth temperatures measured during the FWD test
were used where available. BELLS2 temperature estimates were used where actual measurements were
not available, including all of the preconstruction data (Lukanen et al. 2000).
The radius of relative stiffness, ae, was calculated using the following equation:

ae 

Where


Ep 

a   D3


M
r 


2

2

(Eq. 5)

ae = radius of the stress bulb at the subgrade‐pavement interface (inches)
a = FWD load plate radius (inches)
D = total thickness of pavement layers (inches)
Ep = effective pavement modulus (psi)
Mr = subgrade resilient modulus (psi)

The AASHTO 1993 guide recommends using a sensor with an offset of at least 0.7ae when calculating
resilient modulus according to Equation 5.
Researchers also calculated the structural number of the pavement solely using layer thicknesses and an
assumed layer coefficient of 0.4 for AC and 0.12 for AB. This parameter, SN0, represents the expected
upper bound for SNeff.
A summary of the results of the AASHTO 1993 analysis is presented in Table 9. Detailed discussion of the
results for each test section is provided later in this chapter.

20

Table 9. Summary of AASHTO 1993 Analysis of FWD Data
Section

Surface
Preparation

040507

Intensive

040504

Minimal

SN0

Test Date

Mr
(psi)

SNeff
Raw

SNeff
Corrected

0.7ae
(inches)

5.08

1/18/1990
1/14/1991
12/10/2003

45,133
67,212
91,834

4.60
9.45
13.03

4.52
9.18
10.36

21
37
38

1/18/1990
1/15/1991
12/10/2003

47,376
62,843
70,926

4.08
7.22
7.93

4.08
7.63
7.50

19
32
30

1/18/1990
1/15/1991
12/10/2003

37,852
53,188
47,233

3.33
5.21
4.96

3.43
5.47
5.16

17
24
24

1/18/1990
1/15/1991
12/10/2003

44,978
61,842
56,234

3.32
6.19
5.83

3.47
6.40
6.15

17
27
26

1/18/1990
1/16/1991
9/15/2008

46,196
61,912
48,223

3.43
5.06
3.68

3.53
5.00
4.36

17
21
22

1/18/1990
1/16/1991
9/15/2008

52,447
54,531
49,538

3.42
4.69
3.56

3.45
4.71
4.67

16
21
21

1/19/1990
1/16/1991
9/15/2008

34,317
51,580
35,830

2.85
5.31
3.91

2.93
5.55
5.27

15
25
26

1/19/1990
1/16/1991
9/15/2008

29,400
44,479
45,717

2.82
4.24
3.44

2.80
4.54
4.39

15
21
20

1/19/1990
1/16/1991
12/12/2003

26,859
35,416
29,965

3.00
4.62
4.71

3.40
5.25
5.29

19
26
28

1/19/1990
1/16/1991
12/12/2003

28,141
28,867
39,150

3.04
3.14
3.47

3.16
3.63
3.65

17
20
18

1/19/1990
1/16/1991
9/13/1996

30,790
30,744
41,966

3.32
3.10
3.36

3.28
3.25
4.25

17
17
20

7.27
4.11
5.79
3.99

040503

Minimal

5.55
3.96

040508

Intensive

5.48
4.58

040509

Intensive

5.16
3.80

040502

Minimal

040506

Intensive

4.32
3.50
4.82
4.14

040505

Minimal

5.12
3.66

040559

Intensive

4.66
4.28

040560

040501

Minimal

N/A

4.96

3.70

Although the results for each section will be discussed below, it should be noted that there is a
significant trend in preconstruction SNeff over the entire project. Preconstruction SNeff is greatest at
Section 040507 (located at the beginning of the project), then gradually decreases to its lowest point at

21

Section 040505, and then increases slightly toward Section 040501. Section 040507 has a
preconstruction SNeff of 4.52, whereas for Section 040505 it is 2.80. This trend in SNeff is similar to the
trend in AB thickness, which is 20.7 inches for Section 040507 and 12.8 inches for Section 040505.
This large variability in preconstruction SNeff poses problems for evaluating the project according to the
experimental design. The experimental design was based on the assumption that the preconstruction
structure and condition was essentially identical for each section. If this assumption is correct, then
performance differences between sections can be ascribed solely to differences in construction. For
example, Sections 040504 and 040505 both received minimal surface preparation and virgin AC
overlays. The difference in construction is the thickness of the overlay. A naïve analysis would ascribe
the difference in performance solely to the thickness of the overlay. However, Section 040504 has a
preconstruction SNeff of 4.08, whereas it is 2.80 for Section 040505. If the difference in preconstruction
structure is ignored, the effect of overlay thickness will be overemphasized.
Tables 10, 11, and 12 show the variability in preconstruction SNeff by experimental factor.

Table 10. Preconstruction SNeff Variation by Overlay Material
Overlay Material

Cell

Virgin

Recycled

Intensive, Thick

4.52

3.47

Intensive, Thin
Minimal, Thick
Minimal, Thin

2.93
4.08
2.8

3.53
3.43
3.45

Average

3.58

3.47

Table 11. Preconstruction SNeff Variation by Overlay Thickness
Overlay Thickness

Cell

Thick

Thin

Virgin, Intensive

4.52

2.93

Virgin, Minimal
Recycled, Intensive
Recycled, Minimal

4.08
3.47
3.43

2.8
3.53
3.45

Average

3.88

3.18

22

Table 12. Preconstruction SNeff Variation by Level of Surface Preparation
Surface Preparation

Cell

Intensive

Minimal

Virgin, Thick
Virgin, Thin
Recycled, Thick
Recycled, Thin

4.52
2.93
3.47
3.53

4.08
2.8
3.43
3.45

Average

3.61

3.44

Preconstruction SNeff varies most with overlay thickness and least with overlay material. For analyses of
this project that do not take preconstruction structural variability into account, a significant
overemphasis of the effect of thicker overlays, and some overemphasis of the effect of intensive surface
preparation and use of virgin material, should be expected.
Section 040507
Section 040507 is in the intensive maintenance, thick virgin overlay cell of the experimental matrix. It is
physically located at the beginning of the project. The pre‐ and postconstruction layer structure as listed
in Table TST_L05B of the LTPP database is shown in Table 13.

Table 13. Layer Structure, Section 040507
Construction
No.

1

2–4

2

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.7

3

HMAC

2

AB, 5.8% passing #200

1

A‐2‐4, Silty gravel with sand

6

Virgin HMAC Overlay

4.1

5

Virgin HMAC Overlay

2.7

3

HMAC

2.4

2

AB, 5.8% passing #200

1

A‐2‐4, Silty gravel with sand

4.3
20.7

20.7

1 Depth of milling: 2.6 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed from May 16 to 24, 1990, and consisted of milling 2.6 inches of the original
surface and placing a 6.8‐inch overlay in three lifts. During the milling operation, the original asphalt

23

material exhibited
e
exte
ensive stripping. This section was also ffog‐sealed on
n May 28, 199
98, and April 1
16,
2003.
Mr calculaated using the
e AASHTO 1993 procedure
e is shown in Figures 4 and
d 5. Figure 4 sshows Mr verssus
station for the preconsstruction, posstconstruction
n, and final teests. Figure 5 shows the avverage Mr forr each
test date.

Figurre 4. Subgrade Resilient Modulus
M
versu
us Station, Seection 040507
7

Figu
ure 5. Subgrade Resilient Modulus
M
verssus Date, Secction 040507

24

The behavvior of this su
ubgrade is unusual. The preconstructionn Mr is 45,0000 psi, which iss high for
subgradess in general, but
b within the
e reasonable range for an A‐2‐4 materiial. However, the
postconsttruction Mr in
ncreases to 90
0,000 psi, whiich is above tthe typical ran
nge for a subggrade. This reesult
is stable across
a
over a decade of data collection,, which rules out operatorr or equipmen
nt error. This
subgrade modulus wass calculated using
u
the senssor at 60 inchhes, whereas 0.7ae is no more than 42 inches
for any off these tests, indicating thaat the result should
s
not bee influenced b
by the stiffness of the
pavementt structure.
Investigattors checked for
f the influe
ence of bedrock or other sttiff layers usin
ng the 1/r meethod (Rohdee and
Scullion 1990). This me
ethod require
es data from at
a least threee sensors at or beyond a 36
6‐inch offset,
o available for data colle
ected after 19
998. These pllots show a sllight decreasee in slope at tthe
which is only
48‐inch se
ensor, but the
e x‐intercept for the line plotted througgh the 36‐inch
h and 48‐inch
h sensors is veery
close to 0, indicating a very large de
epth to bedro
ock, which meeans the bedrrock should n
not be influen
ncing
the resultts.
The preco
onstruction te
est result could be dismissed as an outl ier, but the seections at thee other end o
of the
project exxhibit stable and
a typical Mr over time. This
T leaves tw
wo possible co
onclusions: Either the subggrade
response changed afte
er constructio
on of this secttion, or the suubgrade respo
onse at this ssite cannot bee
d techniques. This phenom
menon is discu
ussed furtherr later in this
adequately modeled using standard
chapter.
Figures 6 and 7 show SN
S eff for this section versuss station and time. The SNeff values in th
hese figures aare
temperature‐corrected
d using the methodology
m
described
d
in tthe previous ssection.

Figure
e 6. SNeff verssus Station, SSection 0405007

25

Figure 7. SNeff verrsus Date, Seection 0405077

The plot of
o immediate postconstrucction SNeff vallues versus sttation shows a similar tren
nd to the
immediatte postconstru
uction Mr verrsus station plot, indicatingg that SNeff an
nd Mr are nott fully
independent, at least for
f that test date.
d
The precconstruction SNeff is 4.60, which is reassonable
ng the pavem
ment structure
e. The averagge postconstruuction SNeff iss 10.7, which is considerab
bly
considerin
higher thaan would be expected
e
give
en the pavem
ment structuree. As with thee subgrade, th
he pavement
structure at this section is much stifffer than expe
ected. The SN
Neff versus datte plot does n
not indicate
deteriorattion of the paavement.
Section 04
40504
Section 04
40504 is in th
he minimal surface preparaation, thick viirgin overlay cell of the experimental m
matrix.
The major difference between
b
this section and Section
S
0405007 according to the experimental desiggn is
h of the mill‐and‐fill. The pre‐ and postcconstruction l ayer structurre as listed in Table TST_L0
05B
the depth
of the LTP
PP database iss shown in Taable 14.

26

Table 14. Layer Structure,
S
Secction 0405044
Construction
C
No.
N

1

2 – 52

Layer No
o.

Material Tyype

Thickness
(inches)1

4

OGFC

0
0.7

3

HMAC

2

AB, 4.8%
% passing #200

1

A‐2‐4, Clayey sand withh gravel

5

Virgin HM
MAC overlay

4
4.8

3

HMAC

4
4.3

2

AB, 4.8%
% passing #200

1

A‐2‐4, Clayey sand withh gravel

4
4.3
17
7.6

17
7.6

1 Depth
D
of millingg: 0.7 inch.
2 Does
D
not includ
de fog seals, which are consid
dered structuraally insignifican
nt.

d less of it is n
new overlay. TThe
As compaared with Secttion 040507, the thicknesss of AC is 0.1 inch less, and
thickness of AB is 3.1 in
nches less.
Constructtion was perfo
ormed from May
M 16 to 24,, 1990. This s ection receivved fog seals o
on May 28, 19
998,
and April 16, 2003, and
d crack sealin
ng on May 1, 2002.
2
Mr calculaated using the
e AASHTO 1993 procedure
e is shown in Figures 8 and
d 9. Figure 8 sshows Mr verssus
station for the preconsstruction, posstconstruction
n, and final teests. Figure 9 shows the avverage Mr forr each
test date.

Figurre 8. Subgrade Resilient Modulus
M
versu
us Station, Seection 040504
4

27

Figu
ure 9. Subgrade Resilient Modulus
M
verssus Date, Secction 040504

As with Se
ection 040507
7, the precon
nstruction Mr of 47,000 pssi is high, but within the reasonable range
for the maaterial type. After
A
construction, Mr incrreases to appproximately 880,000 psi, wh
hich is lower tthan
was seen at Section 04
40507, but still above the typical
t
range for a subgrad
de material.
Figures 10
0 and 11 show
w SNeff for this section verssus station annd time. The SSNeff values in
n these figurees are
temperature‐corrected
d using the methodology
m
described
d
earrlier in this ch
hapter.
The imme
ediate postconstruction Mr and SNeff plo
ots both exhi bit a sag arou
und Station 100. The
correlatio
on between th
he plots for th
he last test is less apparennt. The precon
nstruction SN
Neff is 4.08, wh
hich is
reasonable for the pavvement structture. The average postcon struction SNeeff of 8.81 is co
onsiderably h
higher
ected, indicating that whattever phenom
menon is affeccting the resp
ponse of Secttion 040507 iss also
than expe
affecting Section
S
040504. There app
pears to be a slight decreaase in SNeff in the final FWD
D test perform
med
on Decem
mber 10, 2003
3.

28

Figure
e 10. SNeff versus Station, SSection 0405004

Figurre 11. SNeff ve
ersus Date, Seection 0405044

29

Section 040503
Section 040503 is in the minimal surface preparation, thick recycled overlay cell of the experimental
matrix. According to the experimental design, the structure should be identical to Section 040504 except
for the inclusion of recycled asphalt pavement (RAP) in the overlay material. The pre‐ and
postconstruction layer structure as listed in Table TST_L05B of the LTPP database is shown in Table 15.

Table 15. Layer Structure, Section 040503
Construction
No.

1

2 – 52

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.8

3

HMAC

4.2

2

AB, 4.8% passing #200

1

A‐2‐4, Clayey sand with gravel

5

Recycled HMAC overlay

3

HMAC

2

AB, 4.8% passing #200

1

A‐2‐4, Clayey sand with gravel

16.6
4.7
4.2
16.6

1 Depth of milling: 0.8 inch.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was from May 1 to 8, 1990, and consisted of milling 0.8 inch of the original pavement and
overlaying with 4.7 inches of recycled HMA placed in two lifts.
Compared to Section 040504, the thickness of AC material is 0.2 inch less, with the difference divided
evenly between the original pavement and the overlay. The thickness of AB is 1 inch less.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 12 and 13. Figure 12 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 13 shows the average Mr
for each test date.
The subgrade at Section 040503 exhibits different behavior than at Sections 040507 and 040504 in both
the Mr versus station and Mr versus date plots. In the Mr versus station plot, there is a distinct trend in
all three data sets, with Mr being higher at the beginning of the section than at the end. In addition,
while the immediate postconstruction increase in Mr is uniform across the section, it then decreases
over time, primarily at the end of the section. The Mr versus date plot shows a consistent increase in Mr
to a rather high postconstruction value, but Mr then decreases to nearly the preconstruction value.
Figures 14 and 15 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described earlier in this chapter.

30

Figure
e 12. Subgrad
de Resilient Modulus
M
verssus Station, Section 04050
03

Figure 13. Subgraade Resilient Modulus verrsus Date, Section 040503
3

31

Figure
e 14. SNeff versus Station, SSection 0405003

Figurre 15. SNeff ve
ersus Date, Seection 040503

32

The SNeff versus station plot does not show the same trend as the Mr versus station plot, indicating that
SNeff and Mr are independent. The immediate postconstruction SNeff is 5.47, which is reasonable for the
pavement section. However, it increases to a peak of 7.50 on October 18, 1994, which is higher than
would be considered reasonable. It then decreases back to a value consistent with the immediate
postconstruction value. This is a similar trend as is visible in the Mr versus time plot.
Section 040508
Section 040508 is in the intensive surface preparation, thick recycled overlay cell of the experimental
matrix. According to the experimental design, it should differ from Section 040503 only in the depth of
the mill‐and‐fill. It should differ from Section 040507 only in the use of recycled material in the AC
overlay. The pre‐ and postconstruction layer structure as listed in Table TST_L05B of the LTPP database
is shown in Table 16.

Table 16. Layer Structure, Section 040508
Construction
No.

1

2–7

2

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.7

3

HMAC

4.7

2

AB, 7.9% passing #200

15

1

A‐2‐4, Silty gravel with sand

6

Recycled HMAC overlay

4.1

5

Recycled HMAC overlay

2.4

3

HMAC

2.7

2

AB, 7.9% passing #200

15

1

A‐2‐4, Silty gravel with sand

1 Depth of milling: 2.7 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed on April 21 to May 3, 1990, and consisted of milling 2.7 inches of the
original AC surface and placing 6.5 inches of recycled HMAC in three lifts. Fog seals were applied on May
28, 1998, and April 16, 2003. Crack sealing was performed on May 1, 2002.
Compared to Section 040503, the postconstruction thickness of AC is 0.3 inch greater, and more of it is
new overlay. The thickness of AB is 1.6 inches less. Compared to Section 040507, the postconstruction
thickness of AC is the same, but the thickness of AB is 5.7 inches less.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 16 and 17. Figure 16 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 17 shows the average Mr
for each test date.

33

The Mr ve
ersus station plot
p shows a high degree of
o variability aand a weak trrend, with Mr increasing frrom
the beginning to the en
nd of the secttion. Unlike in
n most of thee other sectio
ons, the preco
onstruction daata
also exhib
bits a high deggree of variab
bility. The Mr versus date pplot shows an
n increase in Mr that persissts
for a longer duration than is observved at the oth
her sites. Mr t hen declines to a level thaat is more typ
pical
for a subggrade materiaal.
Figures 18
8 and 19 show
w SNeff for this section verssus station annd time. The SSNeff values in
n these figurees are
temperature‐corrected
d using the methodology
m
described
d
earrlier in this ch
hapter.

Figure
e 16. Mr versus Station, Seection 0405008

34

Figu
ure 17. Mr verrsus Date, Secction 0405088

Figure
e 18. SNeff versus Station, SSection 0405008

35

Figurre 19. SNeff ve
ersus Date, Seection 0405088

The SNeff versus
v
station
n plot shows less variabilitty than the M r versus station plot, altho
ough for the
immediatte postconstru
uction and fin
nal data sets, the high andd low points are coincidentt. The SNeff veersus
date plot shows an inittial increase and
a then a graadual decrea se. The peak SNeff, at six years after
w
is highe
er than would
d be expectedd given the paavement section. The SNefff
construction, is 8.20, which
d from the fin
nal test at thiss section is 6.15, which is sstill somewhaat higher than
n would be
calculated
expected..
Section 04
40509
Section 04
40509 is in th
he intensive surface preparration, thin reecycled overlay cell of the experimentaal
matrix. Acccording to th
he experimen
ntal design, th
he major diffeerence betweeen this sectio
on and Sectio
on
040508 sh
hould be the thickness of the
t overlay. The
T pre‐ and ppostconstruction layer structure as listeed in
Table TSTT_L05B of the LTPP databasse is shown in
n Table 17.
Constructtion was perfo
ormed from May
M 1 to 3, 19
990. Construcction consisteed of milling 2
2.8 inches of the
original su
urface and plaacing 3.8 inch
hes of recycle
ed HMAC in tw
wo lifts. The ssection also received fog sseals
on May 28
8, 1998, and April 16, 2003
3, and crack sealing
s
on Maay 1, 2002.
Compared
d to Section 040508,
0
the thickness of AC
A is 2.7 inchees less. The th
hickness of AB
B is 0.2 inch leess.

36

Table 17. Layer Structure, Section 040509
Construction No.

1

2–5

2

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.7

3

HMAC

2

AB, 7.9% passing #200

1

A‐1‐b, Silty sand with gravel

6

Recycled HMAC overlay

1.3

5

Recycled HMAC overlay

2.6

3

HMAC

2.6

2

AB, 7.9% passing #200

1

A‐1‐b, Silty sand with gravel

4.7
14.8

14.8

1 Depth of milling: 2.8 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Mr calculated using the AASHTO 1993 procedure is shown in Figures 20 and 21. Figure 20 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 21 shows the average Mr
for each test date.
Similar trends for Mr versus station are seen in the pre‐ and postconstruction data. Variability with
station decreases in the final data set. The Mr versus date data show an increase in Mr and then a return
to approximately preconstruction Mr levels, which occurs at about seven years after construction.
Figures 22 and 23 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described earlier in this chapter.

37

Figure
e 20. Mr versus Station, Seection 0405009

Figu
ure 21. Mr verrsus Date, Secction 0405099

38

Figure
e 22. SNeff versus Station, SSection 0405009

Figurre 23. SNeff ve
ersus Date, Seection 0405099
The trend
ds seen in the SNeff versus station
s
plot are not the sa me as those sseen in the Mr versus station
plot, indiccating that SN
Neff and Mr are
e independen
nt. However, the SNeff verssus date and Mr versus datte

39

plots show a similar trend, with an increase over the first two years after construction and then a
decrease approximately seven years after construction. The peak SNeff value is 6.68, which is higher than
would be expected for this pavement section. After 1998, the average SNeff value is 4.29, which is
reasonable.
Section 040502
Section 040502 is in the minimal surface preparation, thin recycled overlay cell of the experimental
matrix. According to the experimental design, the major difference between this section and Section
040509 should be the thickness of the mill‐and‐fill. The pre‐ and postconstruction layer structure as
listed in Table TST_L05B of the LTPP database is shown in Table 18.

Table 18. Layer Structure, Section 040502
Construction
No.

1

2 – 62

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.9

3

HMAC

4.2

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

5

Recycled HMAC overlay

2.7

3

HMAC

3.7

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

14.7

14.7

1 Depth of milling: 1.4 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed from May 1 to 8, 1990, and consisted of milling 1.4 inches of the original
surface and placing a 2.7‐inch overlay in one lift. Fog seals were applied on May 28, 1998, August 23,
2001, and April 16, 2003. Crack sealing was performed on May 1, 2002.
Compared to Section 040509, the thickness of AC is 0.1 inch less, although less of it is new overlay. The
thickness of AB is also 0.1 inch less.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 24 and 25. Figure 24 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 25 shows the average Mr
for each test date.
Unlike the previous sections, Section 040502 does not show a significant increase in Mr after
construction, and in fact the overall trend of Mr over time is downward.
Figures 26 and 27 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described in the previous section.
40

Figure
e 24. Mr versus Station, Seection 0405002

Figu
ure 25. Mr verrsus Date, Secction 0405022

41

Figure
e 26. SNeff versus Station, SSection 0405002

Figurre 27. SNeff ve
ersus Date, Seection 040502

42

The SNeff versus date plot shows an initial increase to 5.86 at 17 months after construction. SNeff then
declines to an average of 4.12, which is reasonable for this pavement section. The decline may be due to
pavement deterioration; however, the data set is lacking points between October 1991 and October
1994, when this may have occurred.
Section 040506
Section 040506 is in the intensive surface preparation, thin virgin overlay cell of the experimental
matrix. The major difference between this section and Section 040509 should be the use of virgin HMAC
instead of recycled HMAC. The pre‐ and postconstruction layer structure as listed in Table TST_L05B of
the LTPP database is shown in Table 19.

Table 19. Layer Structure, Section 040506
Construction
No.

1

2–7

2

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.9

3

HMAC

4.0

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

6

Virgin HMAC overlay

2.4

5

Virgin HMAC overlay

2.8

3

HMAC

3

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

12.8

12.8

1 Depth of milling: 1.9 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Compared to Section 040509, the thickness of AC is 1.7 inches greater, and the new overlays do not
include recycled material. The thickness of AB is 2 inches less.
Construction was performed from May 16 to 24, 1990, and consisted of milling 2.9 inches of the original
AC surface and placing 5.2 inches of HMAC in two lifts. The section received fog seals on May 28, 1998,
August 23, 2001, and April 16, 2003. Crack sealing was performed on May 1, 2002. Patching was
performed on August 1, 2007.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 28 and 29. Figure 28 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 29 shows the average Mr
for each test date.
This section exhibits a postconstruction increase in Mr followed by a gradual decrease. The increase in
average postconstruction Mr is predominantly due to high Mr in the first half of the section. The Mr in
the second half of the section is closer to the preconstruction value.

43

Figures 30
0 and 31 show
w SNeff for this section verssus station annd time. The SSNeff values in
n these figurees are
temperature‐corrected
d using the methodology
m
described
d
earrlier in this ch
hapter.

Figure
e 28. Mr versus Station, Seection 0405006

Figu
ure 29. Mr verrsus Date, Secction 0405066

44

Figure
e 30. SNeff versus Station, SSection 0405006

Figurre 31. SNeff ve
ersus Date, Seection 040506

45

The SNeff versus date data show an increase in SNeff to an average of 6.89 at 17 months after
construction, and then a decrease to 5.27 at 18 years after construction. The SNeff versus station data
show that most of the decrease in SNeff occurred between the stations at 80 m and 100 m.
Section 040505
Section 040505 is in the minimal surface preparation, thin virgin overlay cell of the experimental matrix.
According to the experimental design, the major difference between this section and Section 040506
should be the depth of the mill‐and‐fill. The major difference between this section and Section 040502
should be the use of virgin material for the overlay. The pre‐ and postconstruction layer structure as
listed in Table TST_L05B of the LTPP database is shown in Table 20.
Table 20. Layer Structure, Section 040505
Construction
No.

1

2 – 62

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.9

3

HMAC

4.1

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

5

Virgin HMAC overlay

2.8

3

HMAC

4.1

2

AB, 7% passing #200

1

A4, Clayey sand with gravel

12.8

12.8

1 Depth of milling: 1.2 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed from May 3 to 24, 1990, and consisted of milling 0.9 inch of the existing
material and placing 2.8 inches of HMAC in one lift. The section received fog seals on May 28, 1998,
August 23, 2001, and April 16, 2003. Crack sealing was performed on May 1, 2002.
Compared to Section 040506, the postconstruction thickness of AC is 1.3 inches less. The thickness of AB
is the same. Compared to Section 040502, the postconstruction thickness of AC is 0.5 inch greater, and
the thickness of AB is 1.9 inches less.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 32 and 33. Figure 32 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 33 shows the average Mr
for each test date.
The postconstruction Mr is highly variable with respect to station, and consistently shows a soft spot in
the middle of the section. Average Mr increases for the first 17 months after construction, then returns
to the preconstruction value approximately eight years after construction, and then gradually increases
again.

46

Figures 34
4 and 35 show
w SNeff for this section verssus station annd time. The SSNeff values in
n these figurees are
temperature‐corrected
d using the methodology
m
described
d
earrlier in this ch
hapter.

Figure
e 32. Mr versus Station, Seection 0405005

Figu
ure 33. Mr verrsus Date, Secction 0405055
47

Figure
e 34. SNeff versus Station, SSection 0405005

Figurre 35. SNeff ve
ersus Date, Seection 040505

48

The immediate postconstruction test results show a low SNeff in the center of the section, corresponding
to the low Mr value. SNeff increases over time in this localized area, even as the average SNeff throughout
the section decreases.
Section 040559
Section 040559 is a supplemental section, constructed using intensive surface preparation and an
“inverted” overlay consisting of recycled AC over virgin AC. The pre‐ and postconstruction layer
structure as listed in Table TST_L05B of the LTPP database is shown in Table 21.

Table 21. Layer Structure, Section 040559
Construction
No.

1

2–6

2

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

1.0

3

HMAC

4.2

2

AB, 7% passing #200

1

A‐2‐4, Silty sand with gravel

6

Recycled HMAC overlay

3

5

Virgin HMAC overlay

3

3

HMAC

1.7

2

AB, 7% passing #200

1

A‐2‐4, Silty sand with gravel

13.2

13.2

1 Depth of milling: 3.5 inches.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed from May 16 to 25, 1990, and consisted of milling 3.5 inches of the original
AC surface and placing a 3‐inch recycled overlay on a 3‐inch virgin overlay. Fog seals were applied on
May 28, 1998, August 23, 2001, and April 16, 2003. Crack sealing was performed on May 1, 2002.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 36 and 37. Figure 36 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 37 shows the average Mr
for each test date.
Figures 38 and 39 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described earlier in this chapter.

49

Figure
e 36. Mr versus Station, Seection 0405559

Figu
ure 37. Mr verrsus Date, Secction 0405599

50

Figure
e 38. SNeff versus Station, SSection 040559

Figurre 39. SNeff ve
ersus Date, Seection 0405599

51

Section 040560
Section 040560 is a supplemental test section, consisting of minimal surface preparation and a thin
overlay of asphalt rubber asphalt concrete (ARAC). The pre‐ and postconstruction layer structure as
listed in Table TST_L05B of the LTPP database is shown in Table 22.

Table 22. Layer Structure, Section 040560
Construction
No.

1

2 – 62

Layer No.

Material Type

Thickness
(inches)1

4

OGFC

0.9

3

HMAC

4.1

2

AB, 7% passing #200

1

A‐2‐4, Silty sand with gravel

5

ARAC

2.2

3

HMAC

4.1

2

AB, 7% passing #200

1

A‐2‐4, Silty sand with gravel

14.0

14.0

1 Depth of milling: 0.9 inch.
2 Does not include fog seals, which are considered structurally insignificant.

Construction was performed from May 4 to June 13, 1990, and consisted of milling 0.9 inch of the
existing AC surface and placing a 2.2‐inch overlay of ARAC. Fog seals were applied on May 28, 1990,
August 23, 2001, and April 16, 2003. Crack sealing was performed on May 1, 2002.
Mr calculated using the AASHTO 1993 procedure is shown in Figures 40 and 41. Figure 40 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 41 shows the average Mr
for each test date.
The Mr versus station plots are quite similar for the pre‐ and postconstruction data. The final data set is
also similar for the first half of the section, but is significantly higher for the second half of the section.
The Mr versus date trend is more stable than the other test sections.
Figures 42 and 43 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described earlier in this chapter.

52

Figure
e 40. Mr versus Station, Seection 0405660

Figu
ure 41. Mr verrsus Date, Secction 0405600

53

Figure
e 42. SNeff versus Station, SSection 040560

Figurre 43. SNeff ve
ersus Date, Seection 0405600

54

The SNeff versus station data show a slight and consistent increase in SNeff at each test point between
pre‐ and postconstruction. The final test shows a return to preconstruction SNeff for most of the section,
but an increased SNeff at the end of the test section. SNeff versus date data is relatively stable.
Section 040501
Section 040501 is the control section. No construction activities were performed on this section, and it
was removed from the study on October 1, 1996, due to excessive deterioration. The pre‐ and
postconstruction layer structure as listed in Table TST_L05B of the LTPP database is shown in Table 23.

Table 23. Layer Structure, Section 040501
Construction
No.

1

Layer No.1

Material Type

Thickness
(inches)

4

OGFC

0.9

3

HMAC

4.1

2

AB, 5.1% passing #200

1

A‐2‐4, Clayey sand with gravel

14.2

1 Does not include fog seals, which are considered structurally insignificant.

Mr calculated using the AASHTO 1993 procedure is shown in Figures 44 and 45. Figure 44 shows Mr
versus station for the preconstruction, postconstruction, and final tests. Figure 45 shows the average Mr
for each test date.
The Mr versus station data is quite consistent between the pre‐ and postconstruction data. However,
there is a significant increase in Mr in the final data set.
Figures 46 and 47 show SNeff for this section versus station and time. The SNeff values in these figures are
temperature‐corrected using the methodology described earlier in this chapter.

55

Figure
e 44. Mr versus Station, Seection 0405001

Figu
ure 45. Mr verrsus Date, Secction 0405011

56

Figure
e 46. SNeff versus Station, SSection 0405001

Figurre 47. SNeff ve
ersus Date, Seection 040501

As with Mr, SNeff is quitte consistent between the pre‐ and posstconstruction
n data sets, but increases
thereafter.

57

Summary of Structural Number Data by Experimental Factor
This experiment has three project‐level factors: overlay material (recycled versus virgin mix), overlay
thickness (thick versus thin), and level of surface preparation (intensive versus minimal). The other
experimental factors, including climate, traffic, and subgrade, can only be investigated by comparing
multiple projects.
Changes in SNeff according to the three experimental factors are summarized in Tables 24, 25, and 26.
SNeff is expressed in terms of increase from the preconstruction SNeff for both the postconstruction
results and final results. Increase in SNeff was chosen as the dependent variable in order to minimize the
effect of preconstruction variability. The postconstruction data set used is from October 1991. The
January 1991 data consistently show results intermediate between the preconstruction data and the
October 1991 data. While this could be measurement bias or bias due to inaccuracies in the
temperature correction algorithm, this may also reflect stiffness gain due to aging of the AC binder.
Further investigation of this effect is hampered by a lack of data between October 1991 and October
1994. The October 1991 data was chosen as the best representation of SNeff after initial aging but before
significant traffic damage. The final results are from the December 2003 data set. Some sections also
have data from September 2008, but it cannot be used for comparison purposes because this data is not
available for all sections.

Table 24. SNeff Increase by Overlay Material
Cell

Sections

Postconstruction

Final Test

Virgin

Recycled

Virgin

Recycled

Intensive, Thick

040507 vs. 040508

5.57

3.77

5.84

2.68

Intensive, Thin

040506 vs. 040509

3.96

3.15

2.82

0.54

Minimal, Thick

040504 vs. 040503

4.68

3.22

3.42

1.73

Minimal, Thin

040505 vs. 040502

3.28

2.41

0.93

0.35

4.37

3.14

3.25

1.33

Average:

Table 25. SNeff Increase by Overlay Thickness
Cell

Sections

Postconstruction

Final Test

Thick

Thin

Thick

Thin

Virgin, Intensive

040507 vs. 040506

5.57

3.96

5.84

2.82

Virgin, Minimal

040504 vs. 040505

4.68

3.28

3.42

0.93

Recycled, Intensive

040508 vs. 040509

3.77

3.15

2.68

0.54

Recycled, Minimal

040503 vs. 040502

3.22

2.41

1.73

0.35

4.31

3.20

3.42

1.16

Average:

58

Table 26. SNeff Increase by Level of Surface Preparation
Cell

Sections

Postconstruction

Final Test

Intensive

Minimal

Intensive

Minimal

Virgin, Thick

040507 vs. 040504

5.57

4.68

5.84

3.42

Virgin, Thin

040506 vs. 040505

3.96

3.28

2.82

0.93

Recycled, Thick

040508 vs. 040503

3.77

3.22

2.68

1.73

Recycled, Thin

040509 vs. 040502

3.15

2.41

0.54

0.35

4.11

3.40

2.97

1.61

Average:

As expected, the thick overlays perform better than the thin overlays, and the sections with intensive
preparation perform better than those with minimal preparation. However, there is some bias
introduced by Section 040506, which was built significantly thicker than the other sections in the thin
cells and therefore exaggerates the performance of virgin mixes, thin overlays, and intensive surface
preparation. Removing it (and its partner sections) from the analysis yields the results shown in Table
27.
Table 27. Summary of SNeff Increase by Experimental Factor
Postconstruction SNeff Increase

Final Test SNeff Increase

Factor A

Factor B

Factor
A

Factor
B

Difference

Factor
A

Factor
B

Difference

Virgin

Recycled

4.51

3.13

31%

3.40

1.59

53%

Thick

Thin

3.89

2.95

24%

2.61

0.61

77%

Intensive

Minimal

4.16

3.44

17%

3.02

1.83

39%

These results show that at the time of the final test, overlay thickness has the strongest influence on
SNeff among the factors investigated. Overlay material (virgin versus recycled mix) is still a stronger
factor than level of surface preparation. The strength of each experimental factor increases over time.
BACKCALCULATION
Backcalculation was performed for all data collected at the project using MODCOMP version 6E (release
date November 10, 2012). Specific results for each section are discussed in the sections that follow;
however, some general points are relevant to all sections.
Data for each test location, date, and drop height were averaged prior to backcalculation. This has the
effect of reducing the data analysis requirements by a factor of four (as typically four drops are
performed at each drop height). In theory, this process also reduces the effect of random error on the
analysis results, which is beneficial because of the low deflections throughout this project.

59

For all analyses, the subgrade was split into two layers: an upper layer that was 24 inches thick,
representing the seasonally affected portion of the subgrade, and a lower semi‐infinite deep subgrade.
In some sections, the response of these two subgrade layers was quite similar; in others it was quite
different.
When possible, the individual AC layers were analyzed separately. This was generally not possible for
layers less than 2 inches thick, especially the OGFC that existed preconstruction. Because of the well‐
known “thin layer effect,” backcalculation is typically unreliable for layers that do not significantly
contribute to the stiffness of the entire structure. Put differently, moduli can only be determined from
deflections if the deflections are reasonably sensitive to those moduli. The modulus of a sufficiently thin
layer can vary widely without causing a significant difference in the surface deflections. MODCOMP 6E
displays error codes to warn the user of layers whose backcalculated modulus is insensitive to the
measured deflections, and these error codes were considered in establishing the layer structures.
Only results with a root mean square error (RMSE) of less than 5 percent are presented here. This RMSE
threshold is somewhat high for typical backcalculation; however, this data set includes many modeling
difficulties, and a higher threshold would result in very few data points for some sections and test dates.
Some results with an RMSE of less than 5 percent were also rejected due to unreasonable moduli.
All of the moduli of AC layers have been temperature‐corrected to 77° F (25° C) using the following
LTPP‐derived equation unless otherwise stated (Lukanen et al. 2000).

Eref  Ecalc 10
Where

k (Tref Tcalc )

(Eq. 6)

Eref = modulus at the reference temperature
Ecalc = backcalculated modulus
k = ‐0.0195 for testing in the wheel path, ‐0.021 for testing in the midlane
Tref = reference temperature (°C)
Tcalc = AC temperature corresponding to backcalculated modulus (°C)

All temperatures are based on the measured AC mid‐depth temperature where available. For tests
where the mid‐depth temperature was not available, the BELLS2 estimation procedure was followed.
In general, the backcalculated moduli are often not within the reasonable range for their respective
material types, especially for the AC and upper subgrade layers. Potential reasons for these results are
discussed later in this chapter.

60

Section 040507
Section 040507 is in the intensive maintenance, thick virgin overlay cell of the experimental matrix. The
layer models used for backcalculation of all data are shown in Table 28.
Table 28. Layer Models for Backcalculation, Section 040507
Date Range

1/18/1990

1/14/1991‐
12/10/2003

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1

AC

5

500

0.35

2

AB

20.7

10

0.50

3

Subgrade

24

35

0.45

4

Subgrade

Semi‐infinite

35

0.45

1

AC overlay

6.8

500

0.35

2

AC original

2.4

500

0.35

3

AB

20.7

10

0.50

4

Subgrade

24

35

0.45

5

Subgrade

0

35

0.45

The postconstruction layer model splits out the AC overlay and remaining original AC layer for
comparison. The original AC layer (Layer 2) is only 2.4 inches thick after milling, and it is beneath 6.8
inches of stiff material, which makes the backcalculated modulus relatively insensitive to the surface
deflections. The backcalculated moduli for the original AC layer (Layer 2) should therefore be used with
caution.
Backcalculation results are shown in Table 29. The results shown are for drop height 4 only. Drop height
4 was chosen for this table because it generally has the highest number of acceptable backcalculation
results. The number of acceptable backcalculation results is given as N; Nrejected is the number of
unacceptable backcalculation results. The AB layer (Layer 3 in the table) showed significant nonlinearity,
which is not apparent from the data in this table. Nonlinearity of this layer is discussed later in this
chapter.
The AC layers (Layers 1 and 2) exhibit highly variable moduli, and the moduli are higher than is typical
for AC materials. Some of this may be explained by deficiencies in the temperature correction
methodology, as discussed later in this chapter. The upper subgrade (Layer 4 in Table 29) also exhibits
extremely high moduli, especially for the testing on December 10, 2003. Reasons for this phenomenon
are discussed later.

61

Table 29. Backcalculation Results, Section 040507
Test Date
1/18/1990
1/14/1991
10/2/1991
10/18/1994
9/11/1996
11/12/1997
12/9/1998
12/13/1999
10/16/2000
12/10/2002
12/10/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

1962
1635
1044
1316
1775
2038
1612
1790
1816
1783
1449
1844
1878
1729
3147
3293
2553
2338
2337
2283

2281
7191
1146
2422
6139
7402
2370
2717
5000
5030
2431
3326
1164
1861
1461
2159
1325
2465
142
1353
715
1064

3
191
301
36
31
87
80
89
114
122
185
77
126
70
71
71
75
21
142
197
152
65
90

4
331
411
268
923
283
414
385
459
354
306
593
605
677
829
154
51
459
712
195
221
2068
2323

5

Average
RMSE
(%)

N

Nrejected

2.1
1.8
1.3
1.1
2.4
1.8
1.3
1.3
1.4
1.6
2.9
1.7
1.2
0.9
1.8
1.3
1.0
2.0
2.8
3.2
2.1
2.9

1
4
11
11
11
11
11
10
11
11
8
8
10
10
10
11
5
10
2
3
9
9

5
2
0
0
0
0
0
1
0
0
3
3
1
1
1
0
6
1
9
8
2
2

341
441
67
58
83
81
87
87
94
98
87
93
87
86
83
89
88
89
82
117
83
110

1 Preconstruction layer numbers shifted to align with postconstruction model.

62

Section 040504
Section 040504 is in the intensive maintenance, thick virgin overlay cell of the experimental matrix. The
layer models used for backcalculation of all data are shown in Table 30.

Table 30. Layer Models for Backcalculation, Section 040504
Date Range

1/18/1990

1/15/1991‐
12/10/2003

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1

AC

5

500

0.35

2

AB

17.6

10

0.50

3

Subgrade

24

35

0.45

4

Subgrade

Semi‐infinite

35

0.45

1

AC overlay

4.8

500

0.35

2

AC original

4.3

500

0.35

3

AB

17.6

10

0.50

4

Subgrade

24

35

0.45

5

Subgrade

0

35

0.45

Backcalculation results are shown in Table 31. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results.

63

Table 31. Backcalculation Results, Section 040504
Test Date
1/18/1990
1/15/1991
10/2/1991
10/18/1994
9/11/1996
11/12/1997
12/9/1998
12/9/1999
10/16/2000
12/10/2002
12/10/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

1665
2151
868
1144
1536
1937
1583
2382
1908
1134
1710
1814

2561
4571
486
942
1821
2707
2127
866
2105
3288
1186
1964
785
1292

501
261
42
26
86
102
86
74
120
123
86
86
48
43

1957
3250
4075
5607
3858
2214
1989

2243
1613
2872
1092
624
547
917

75
90
82
403
194
59
19

4

5

Average
RMSE
(%)

N

Nrejected

3.5
2.9
1.0
1.1
2.1
1.2
1.3
1.2
1.3
1.3
2.9
1.3
1.0
1.1

1
2
11
11
11
11
11
11
11
11
4
11
11
8

4
3
0
0
0
0
0
0
0
0
7
0
0
3

1.1
1.7
1.9
3.5
2.4
1.8
1.4

10
10
7
2
2
5
5

1
1
4
9
9
6
6

571 551
601 541
185
61
758
56
267
78
215
78
447
80
275
78
277
86
368
85
289
84
447
80
411
74
1248
73
No data
434
81
365
79
599
76
772
66
197
94
133
67
2773
55

1 Preconstruction layer numbers shifted to align with postconstruction model.

Section 040503
Section 040503 is in the minimal surface preparation, thick recycled overlay cell of the experimental
matrix. The layer models used for backcalculation of all data are shown in Table 32.
Backcalculation results are shown in Table 33. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results.

64

Table 32. Layer Models, Section 040503
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/18/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5
16.6
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/15/1991‐
12/10/2003

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

4.7
4.2
16.6
24
0

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

Table 33. Backcalculation Results, Section 040503
Test Date
1/18/1990
1/15/1991
10/2/1991
10/18/1994
9/11/1996
11/12/1997
12/9/1998
12/13/1999
10/16/2000
12/11/2002
12/10/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

672
1012
580
698
541
1910
613
1829
893
1229
1225
1089
1025
531
1912
2133
1794
2071
1083
1228

2671
7631
279
613
753
1543
1240
1414
961
2050
442
1399
184
747
225
1188
340
1167
2
339
434
955

191
251
17
15
44
53
110
72
33
59
20
19
22
15
19
8
25
21
171
360
15
4

231
411
56
54
83
65
105
100
90
109
94
170
41
541
36
359
128
26
77
193
91
410

441
411
48
47
66
64
112
78
62
69
51
57
40
36
39
35
46
46
27
44
48
32

Average
RMSE
(%)

1 Preconstruction layer numbers shifted to align with postconstruction model.

65

1.9
3.1
1.1
0.8
1.1
0.6
0.3
0.8
2.1
1.5
2.9
2.7
1.2
1.1
1.2
1.3
2.3
1.7
3.6
3.4
2.6
1.9

N
1
4
11
11
11
11
1
11
11
11
11
7
11
9
11
8
11
10
1
3
5
7

Nrejected
5
2
0
0
0
0
0
0
0
0
0
4
0
2
0
3
0
1
10
8
6
4

Section 040508
Section 040508 is in the intensive surface preparation, thick recycled overlay cell of the experimental
matrix. The layer models used for backcalculation of all data are shown in Table 34.
The postconstruction layer model splits out the AC overlay and remaining original AC layer for
comparison. The original AC layer (Layer 2) is only 2.7 inches thick after milling, and is beneath 6.5
inches of stiff material, which makes the backcalculated modulus relatively insensitive to the surface
deflections. The backcalculated moduli for the original AC layer (Layer 2) should therefore be used with
caution.
Backcalculation results are shown in Table 35. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 3 in Table 35) showed significant nonlinearity, which is not
apparent from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

Table 34. Layer Models, Section 040508
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/18/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5.4
15
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/15/1991‐
12/10/2003

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

6.5
2.7
15
24
0

66

Table 35. Backcalculation Results, Section 040508
Test Date
1/18/1990
1/15/1991
10/2/1991
10/19/1994
9/12/1996
11/12/1997
12/9/1998
12/13/1999
10/17/2000
12/11/2002
12/10/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

871
810
545
628
1095
1021
977
1108
1440
1109
1465
1474
1281
1741
1800
1484
2179
1582
1429
7270

3311
6921
507
1120
5208
6772
1289
2585
3019
5713
481
3055
809
2276
1202
1195
425
1240
282
823
337
1205

231
191
69
33
60
75
49
52
87
82
750
53
112
30
36
50
75
77
29
49
38
16

301
251
79
75
133
126
185
189
247
258
219
416
318
432
200
189
96
81
122
81
29
42

431
431
64
60
75
74
88
89
100
97
110
105
87
90
84
77
76
67
59
54
56
46

Average
RMSE
(%)

N

Nrejected

2.6
2.4
1.1
0.6
1.4
0.9
1.1
0.8
2.3
1.7
3.0
2.9
1.1
1.1
1.2
1.2
1.8
1.8
2.2
3.0
0.9
2.1

6
3
11
11
11
11
11
10
11
11
8
6
11
10
10
9
10
9
3
3
4
4

0
3
0
0
0
0
0
1
0
0
3
4
0
1
1
2
1
2
8
8
7
7

1 Preconstruction layer numbers shifted to align with postconstruction model.

Section 040509
Section 040509 is in the intensive surface preparation, thin recycled overlay cell of the experimental
matrix. The layer models used for backcalculation of all data are shown in Table 36.
Table 36. Layer Models, Section 040509
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/18/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5.4
14.8
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/15/1991‐
12/10/2003

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

3.9
2.6
14.8
24
0

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

67

Backcalculation results are shown in Table 37. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 3 in Table 37) showed significant nonlinearity, which is not
apparent from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

Table 37. Backcalculation Results, Section 040509

Test Date
1/18/1990
1/16/1991
10/2/1991
10/19/1994
9/12/1996
11/13/1997
12/10/1998
12/13/1999
10/17/2000
12/11/2002
12/11/2003
9/15/2008

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

745
1291
480
703
1592
1886
1246
1010
1094
1009
625
1096
1257
1375
2290
3047
1281
1245
488
1878
1148
2130

2331
8171
655
1146
703
6342
1290
1252
3103
7211
1018
1380
330
492
430
1326
400
571
383
281
481
491
1759
2634

311
331
34
46
72
107
81
107
68
109
43
46
25
22
26
27
17
23
53
61
21
40
20
24

591
501
120
87
257
185
162
142
170
135
77
40
37
45
33
122
56
42
69
49
44
50
43
39

381
431
59
57
63
69
67
69
68
74
50
52
36
32
39
35
36
35
43
44
31
40
42
41

Average
RMSE
(%)

N

Nrejected

0.7
2.1
1.3
1.1
1.6
1.8
1.0
1.3
2.2
2.3
3.1
2.2
1.5
1.3
0.9
1.1
1.7
1.5
2.7
2.2
1.9
1.9
2.5
2.5

1
3
10
11
10
11
10
11
9
11
6
7
9
9
11
11
11
8
7
6
5
6
7
6

5
3
1
0
1
0
1
1
2
0
5
3
2
2
0
0
0
3
4
5
6
5
4
5

1 Preconstruction layer numbers shifted to align with postconstruction model.

68

Section 040502
Section 040502 is in the minimal surface preparation, thin recycled overlay cell of the experimental
matrix. The layer models used for backcalculation of all data are shown in Table 38.
Backcalculation results are shown in Table 39. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 3 in Table 39) showed significant nonlinearity, which is not
apparent from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

Table 38. Layer Models, Section 040502
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/18/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5.1
14.7
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/16/1991‐
9/15/2008

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

2.7
3.7
14.7
24
Semi‐infinite

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

69

Table 39. Backcalculation Results, Section 040502
Test Date
1/18/1990
1/16/1991
10/3/1991
10/19/1994
9/12/1996
11/13/1997
12/10/1998
12/14/1999
10/17/2000
12/11/2002
12/11/2003
9/15/2008

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

584
1603
692
1078

4381
5161
396
271
833
1401

341
301
49
81
92
130
21
35
32
31
21
15
16
11
9
15
17
16
25
16
14
21
26
33

441
591
98
82
155
145
31
27
43
32
52
74
31
59
153
69
35
61
41
59
73
47
41
55

481
451
56
53
67
60
44
42
53
46
46
39
33
31
33
29
42
36
46
36
54
36
47
45

5792
917
720
975
403
738
180
400
251
390
614
1090
826
479
369
251
737
1919

Average
RMSE
(%)

N

Nrejected

3.9
3.9
1.5
1.9
1.6
1.4
2.2
1.7
2.4
3.2
3.0
2.4
2.2
3.4
3.2
3.6
4.0
2.9
3.5
3.9
4.7
3.2
3.1
3.3

3
5
9
8
8
8
9
10
10
9
3
4
4
4
1
5
4
5
2
1
1
2
7
4

3
1
2
3
3
3
2
1
1
2
8
7
7
7
10
6
7
6
9
10
10
9
4
7

1 Preconstruction layer numbers shifted to align with postconstruction model.
2 AC layers combined beginning 10/19/1994 due to insensitivity of Layer 2.

Section 040506
Section 040506 is in the intensive surface preparation, thin virgin overlay cell of the experimental
matrix. The layer models used for backcalculation of all data are shown in Table 40.
Backcalculation results are shown in Table 41. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results.

70

Table 40. Layer Models, Section 040506
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/19/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

4.9
12.8
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/16/1991‐
9/15/2008

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

5.7
3.0
12.8
24
Semi‐infinite

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

Table 41. Backcalculation Results, Section 040506
Test Date
1/19/1990
1/16/1991
10/3/1991
10/19/1994
9/12/1996
11/13/1997
12/10/1998
12/14/1999
10/17/2000
12/12/2002
12/11/2003
9/15/2008

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

1314
1005
1174
1299
1509
2051
1230
1545
1748
1693
1289
1318
764
1451
2710
2953
1703
1163
1444
1326
1529
1345

2441
218
464
1045
2136
938
745
3090
5179
835
1934
692
697
298
403
963
1916
112
866
1209
1351
1577
2484

3

4

5

Average
RMSE
(%)

No acceptable results
461
281
291
78
162
54
27
162
48
70
137
64
71
174
63
64
139
59
100
176
57
61
128
57
83
263
58
26
150
59
47
315
65
25
287
51
32
296
54
16
33
30
61
222
56
32
223
57
47
357
48
149
380
45
51
274
55
17
562
56
35
160
49
5
277
31
7
516
30

1 Preconstruction layer numbers shifted to align with postconstruction model.

71

1.1
1.1
1.3
1.9
1.5
1.0
0.8
1.9
1.4
1.9
3.3
1.0
0.7
2.2
1.0
1.7
2.6
2.1
2.4
1.9
1.1
1.2
1.1

N
0
1
11
9
10
10
9
10
8
10
2
2
4
8
10
10
6
6
6
7
3
8
10
10

Nrejected
6
5
0
2
1
1
2
1
3
1
9
9
7
3
1
1
5
5
5
4
8
3
1
1

Section 040505
Section 040505 is in the minimal surface preparation, thin virgin overlay cell of the experimental matrix.
The layer models used for backcalculation are shown in Table 42.
Backcalculation results are shown in Table 43. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 3 in Table 43) showed significant nonlinearity, which is not
apparent from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

Table 42. Layer Models, Section 040505
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/19/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5
12.8
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/16/1991‐
9/15/2008

1
2
3
4
5

AC overlay
AC original
AB
Subgrade
Subgrade

2.8
4.1
12.8
24
Semi‐infinite

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

72

Table 43. Backcalculation Results, Section 040505
Test Date
1/19/1990
1/16/1991
10/3/1991
10/19/1994
9/12/1996
11/13/1997
12/10/1998
12/14/1999
10/18/2000
12/12/2002
12/11/2003
9/15/2008

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

847
1222
860
1123
1946
2045
1863
1465
1028
1015
2898
3675
670
769
2213
2305
3188
1036
568
694
1567
916

2711
4831
419
654
938
1646
749
7186
1052
2995
694
1452
198
342
395
708
342
1497
12
308
289
544
712
1764

191
121
41
41
100
95
67
72
87
72
35
17
69
43
16
24
28
18
12
40
22
42
24
40

341
851
75
191
103
153
66
92
43
54
42
65
60
30
34
53
36
148
37
29
33
35
36
64

261
341
46
45
53
56
50
51
51
50
41
40
29
31
30
36
36
35
27
30
31
35
41
47

Average
RMSE
(%)

N

Nrejected

3.5
4.7
1.5
2.6
2.2
1.8
1.3
1.7
2.3
2.5
3.1
2.6
1.8
1.6
2.3
1.5
2.1
1.8
1.8
3.1
2.6
1.8
2.0
1.8

2
2
11
11
11
10
11
10
10
9
6
9
9
11
9
4
8
5
1
3
4
6
8
8

4
4
0
0
0
1
0
1
1
2
5
2
2
0
2
7
3
6
10
8
7
5
3
3

1 Preconstruction layer numbers shifted to align with postconstruction model.

Section 040559
Section 040559 is a supplemental section, constructed using intensive surface preparation and an
inverted overlay consisting of recycled AC over virgin AC. The layer models used for backcalculation are
shown in Table 44.
Only 1.7 inches of original pavement was left after milling, which is too thin a layer to model. Therefore,
the remaining original pavement was combined with the virgin overlay for backcalculation.
Backcalculation results are shown in Table 45. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results.

73

Table 44. Layer Models, Section 040559
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/19/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5.2
13.2
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/16/1991‐
12/12/2003

1
2
3
4
5

Rec. overlay
Vir. overlay
AB
Subgrade
Subgrade

3
4.7
13.2
24
Semi‐infinite

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

Table 45. Backcalculation Results, Section 040559
Test Date
1/19/1990
1/16/1991
10/3/1991
10/20/1994
9/13/1996
11/14/1997
12/10/1998
12/14/1999
10/18/2000
12/12/2002
12/12/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

1177
1407
683
704
2862
2664
788
822
2297
3829
3773
2145
2058
3318
4790
7242
7503
9844
6157
2994

3591
6141
946
1078
1520
1553
1298
1415
3321
4602
1700
2515
1194
2147
1230
1497
1365
2048
1159
1707
438
855

361
271
24
31
59
77
50
78
84
164
40
133
22
44
50
173
62
113
80
34
51
47

271
511
49
56
78
81
118
96
110
98
123
286
356
553
76
94
75
67
138
85
56
19

281
271
37
36
45
45
47
78
52
54
48
48
38
41
42
43
46
42
32
35
34
31

Average
RMSE
(%)

N

Nrejected

2.3
2.3
0.9
0.7
0.8
0.9
0.7
0.4
1.1
1.2
2.7
2.1
0.8
0.9
0.8
0.9
1.4
1.7
2.6
1.4
0.9
0.9

4
5
6
6
11
11
11
11
11
11
9
9
11
11
11
11
8
10
5
7
9
8

2
1
0
0
0
0
0
0
0
0
2
2
0
0
0
0
3
1
6
4
2
3

1 Preconstruction layer numbers shifted to align with postconstruction model.

74

Section 040560
Section 040560 is a supplemental test section, consisting of minimal surface preparation and a thin
overlay of asphalt rubber asphalt concrete (ARAC). The layer models used for backcalculation are shown
in Table 46.
Backcalculation results are shown in Table 47. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 3 in Table 47) showed significant nonlinearity, which is not
apparent from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

Table 46. Layer Models, Section 040560
Date Range

Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1/19/1990

1
2
3
4

AC
AB
Subgrade
Subgrade

5
14
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

1/16/1991‐
12/12/2003

1
2
3
4
5

ARAC
Original AC
AB
Subgrade
Subgrade

2.2
4.1
13.2
24
Semi‐infinite

500
500
10
35
35

0.35
0.35
0.50
0.45
0.45

75

Table 47. Backcalculation Results, Section 040560
Test Date
1/19/1990
1/16/1991
10/3/1991
10/20/1994
9/13/1996
11/14/1997
12/10/1998
11/18/2000
12/12/2002
12/12/2003

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

5

802
1744
747
907
2342
5457
1838
2770
3431
5489
3994
5532
7368
3446
4367
7000
2989
484

2891
4181
382
879
1053
1960
338
787
663
2070
180
702
69
318
187
938
101
285
69
481

211
271
12
24
16
23
9
73
14
18
21
142
33
15
45
16
621
83
53
22

331
321
29
27
36
40
69
50
35
59
54
83
22
63
46
50
66
57
108
35

291
281
28
31
35
37
24
30
33
35
37
38
28
27
40
32
40
36
46
36

Average
RMSE
(%)

N

Nrejected

2.4
3.5
1.5
1.6
2.3
1.4
2.4
2.1
2.6
2.2
3.4
2.4
2.4
1.7
2.0
2.2
2.7
3.3
2.0
1.9

2
7
7
7
13
11
4
9
11
12
4
6
7
11
11
7
8
6
6
5

3
0
0
0
0
0
0
4
2
1
9
7
5
2
2
6
5
7
7
8

1 Preconstruction layer numbers shifted to align with postconstruction model.

Section 040501
Section 040501 is the control section; it received no overlay or other maintenance. The layer model used
for backcalculation of all data is shown in Table 48.
Backcalculation results are shown in Table 49. Results are for drop height 4 only. Drop height 4 was
chosen for this table because it generally has the highest number of acceptable backcalculation results.
The number of acceptable backcalculation results is given as N; Nrejected is the number of unacceptable
backcalculation results. The AB layer (Layer 2) showed significant nonlinearity, which is not apparent
from the data in this table. Nonlinearity of this layer is discussed later in this chapter.

76

Table 48. Layer Model, Section 040501
Layer

Layer Type

Thickness
(inches)

Seed Modulus
(ksi)

Poisson’s Ratio

1
2
3
4

AC
AB
Subgrade
Subgrade

5
14.2
24
Semi‐infinite

500
10
35
35

0.35
0.50
0.45
0.45

Table 49. Backcalculation Results, Section 040501
Test Date
1/19/1990
1/16/1991
10/3/1991
10/20/1994
9/16/1996

Lane
Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer Modulus (ksi)
1

2

3

4

87
153
367
1054
944
1319
752
1207
1766
1800

21
27
17
7
22
29
19
18
25
39

32
30
28
55
34
34
29
40
30
33

31
33
32
33
37
37
33
34
41
44

Average
RMSE
(%)

N

Nrejected

3.0
3.0
2.7
3.4
3.2
3.0
2.4
3.2
3.0
3.0

4
5
4
1
6
7
7
9
11
11

2
1
7
10
5
4
4
2
0
0

The AC layer (Layer 1) exhibits extremely large variability in modulus, and generally high modulus values.
The raw versus uncorrected moduli for this layer are shown in Table 50.
The measured (not temperature‐corrected) data show significantly less variability and more reasonable
values than the temperature‐corrected data, indicating that the temperature correction methodology
used is not valid for this section. Similar results were seen for the other sections. This could be due to
the poor condition of the AC layer, or the use of an AC material that is not well represented in the
nationwide modulus versus stiffness model used for correction. The as‐measured results show no
relationship to temperature, and therefore cannot be used to establish a site‐specific temperature
correlation.

77

Table 50. Measured and Temperature‐Corrected AC Moduli
Test Date
1/19/1990
1/16/1990
10/3/1991
10/20/1994
9/16/1996

Lane Position
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP

AC Modulus (ksi)
Measured Corrected
437
800
487
14301
395
532
480
860
752
830

AC Temperature (°F)

87
153
367
1054
944
1319
752
1207
1766
1800

54
61
66
66
111
111
95
90
111
106

1 Based on only one test point.

The AB layer exhibits significant nonlinearity, as shown in Table 51.
Table 51. Nonlinearity of AB Modulus
Test Date
1/19/1990
1/16/1991
10/3/1991
10/20/1994
9/16/1996

Lane Position
OWP
ML
OWP
ML
OWP
ML
OWP
ML
OWP
ML

Average Layer 2 Modulus (ksi)
Height 1 Height 2 Height 3 Height 4
15.2
16.7
14.6
No data
18.3
18.5
14.7
12.1
15.0
25.2

20.2
18.7
14.1
No data
17.6
20.5
15.8
15.4
15.9
26.7

18.3
21.6
15.7
No data
22.0
24.9
17.5
17.7
19.7
33.1

21.4
26.5
16.7
6.9
22.4
29.2
18.9
18.4
25.1
39.2

The relationship between modulus and drop height is most apparent for the 1994 and 1996 data sets.
These data sets also have the highest rates of basin convergence (N/Nrejected). Layer 2 exhibits an increase
in modulus with drop height (i.e., stress‐hardening), which is typical for AB materials. This behavior is
typically modeled with a bulk‐stress model of the following form:

78

E  k1   k2
Where

(Eq. 7)

E = modulus (psi)
k1, k2 = regression coefficients
Θ = bulk stress (psi)

Nonlinear backcalculation was performed using MODCOMP 6E. The layer model shown in Table 48 was
used, with the additional required inputs shown in Table 52.
Table 52. Nonlinear Layer Parameters
Layer

Layer Type

Unit Weight (pcf)

Coefficient of Lateral Earth Pressure

1
2
3
4

AC
AB
Subgrade
Subgrade

145
135
125
125

0.0
0.5
0.5
0.5

Nonlinear backcalculation was only performed for the 1994 and 1996 data sets, as the modulus versus
drop height results for the other data sets using linear backcalculation indicated that nonlinear
backcalculation was unlikely to be successful. Results are shown in Table 53. MODCOMP reported all
these regressions to be statistically significant.

Table 53. Nonlinear Backcalculation Results
Test Date

Lane Position

10/20/1994
9/16/1996

Average Regression Coefficient
k1 (psi)
k2

OWP
ML
OWP
ML

10,887
12,867
14,024
23,402

0.061
0.057
0.049
0.045

Discussion of Backcalculation Results
The data presented in this report have many peculiarities of the sort that would typically be ignored in
normal production FWD testing in support of pavement design. Because of the large volume of FWD
testing performed on this project, both over time and across different pavement structures, we can
begin to address these peculiarities.

79

Deep Subgrade Modulus Increase
The deep subgrade modulus values are the most consistent and reasonable layer modulus values
computed in this study. However, the phenomenon of increasing postconstruction layer modulus should
be explained. Table 54 summarizes the increase in postconstruction modulus for the deep subgrade
layer as determined through backcalculation, the total increase in pavement thickness, and the increase
in SNeff as determined through the AASHTO 1993 analysis. The postconstruction values presented in the
table are from the October 1991 round of testing, as the January 1991 round of testing tends to show
intermediate values.
Increase in pavement thickness and increase in SNeff were chosen as potential explanations for the deep
subgrade modulus increase because of their effect on the stress state of the deep subgrade. An increase
in pavement thickness will increase overburden stress and therefore confining pressure, and will
decrease deviator stress by spreading the load over a greater area. SNeff is a function of both pavement
thickness and effective pavement modulus, and therefore should better represent the decrease in
deviator stress in the subgrade due to the overlay.

Table 54. Postconstruction SNeff Increase
Section

Mr Increase (%)

Thickness Increase (inches)

SNeff Increase (%)

040507
040504
040503
040508
040509
040502
040506
040505
040559
040560
040501

84
44
56
74
60
60
117
65
67
32
9

4.2
4.2
3.9
3.8
1.1
1.3
3.3
1.9
2.5
1.3
0

123
115
94
109
89
70
135
117
81
43
20

The increase in pavement thickness shows poor correlation to the increase in deep subgrade modulus.
The increase in SNeff has better correlation to the increase in deep subgrade modulus, as shown in Figure
48.

80

150%

Deep Subgrade Modulus Increase

y = 0.6706x
R2 = 0.6947

100%

50%

0%
0%

50%

100%

150%

SNeff Increase

Figure 48. Subgrade Modulus Increase versus SNeff Increase

It should be noted that backcalculation of this data set does not indicate significant nonlinear behavior
for the deep subgrade. However, due to the very small deflections at the lower drop heights and the
great depth to the lower subgrade, the ability of the FWD to detect nonlinearity in this layer is very
limited. In addition, FWD testing at different drop heights only varies the deviator stress; there is no way
to vary the confining pressure (and thereby measure its effect on modulus) without changing the
pavement structure. Still, the phenomenon of increasing subgrade modulus in response to an overlay is
supported by both theory and this data set, and such increases should be considered in pavement
design.
The January 1991 data show subgrade modulus values intermediate between the preconstruction data
and the October 1991 data. The SNeff values computed from the January 1991 data are also intermediate
values. These intermediate values may be due to stiffness gain from short‐term aging of the AC layer.

81

Upper Subgrade Stiff Layer
In many but not all tests, a very high modulus value has been backcalculated for the upper subgrade. In
a few cases, the stiff layer is the base layer. In some sections, such as 040507, this stiff layer exists for all
of the postconstruction data sets. In other sections, such as 040503, the stiff layer appears to come and
go in different tests. This stiff layer is not seen in the control section (040501), or in any of the
preconstruction data. The obvious question is whether this stiff layer actually exists, or whether it is an
artifact of measurement error or error in the analysis method.
To better understand why MODCOMP calculated such high modulus for this layer, we experimented
with various layer models for a single test section and date, and used the backcalculation results to
develop synthetic deflection basins for comparison. The synthetic deflection basins were computed
using the CHEVLAY2 program version 2.3 (release date April 20, 1997), which is used by MODCOMP for
forward‐calculation. The testing performed on Section 040507 on December 10, 2003, was selected
because it shows the highest modulus values for the upper subgrade layer. All of the drop height 4 data
was averaged to produce a single average deflection basin. This was done to reduce the effect of
random error and spatial variability, which is significant because of the advanced state of deterioration
of the pavement. Table 55 shows the average deflection basin along with the synthetic deflection basins
generated using the backcalculation results. The backcalculation results for this average deflection basin
using different layer models are shown in Table 56.

Table 55. Measured and Synthetic Deflection Basins

Measured
5‐Layer
4‐Layer
3‐Layer
2‐Layer

0 in

8 in

Deflection (mils)
12 in 18 in 24 in 36 in

2.96
2.99
2.96
2.96
2.95

2.50
2.52
2.50
2.50
2.48

2.24
2.25
2.23
2.25
2.23

1.85
1.88
1.88
1.90
1.88

1.56
1.57
1.57
1.61
1.57

1.13
1.11
1.12
1.15
1.11

48 in

60 in

0.85
0.85
0.86
0.85
0.81

0.72
0.71
0.71
0.66
0.62

Table 56. Backcalculation Results, Average Deflection Basin
from Drop Height 4 Data
Model
Layers
5
4
3
2

Layer 1
Thick
(inches)

E
(ksi)

6.8
9.2
9.2
9.2

3600
3480
3350
3320

Layer 2
Thick
E
(inches) (ksi)
2.4

3110
N/A
N/A
N/A

Layer 3
Layer 4
Layer 5
RMSE
Thick
E
Thick
E
Thick
E
(%)
(inches) (ksi) (inches) (ksi) (inches) (ksi)
20.7
54
20.7
51
20.7
122
N/A

82

24.0
24.0
∞
∞

555
582
110
114

∞
∞

87
87
N/A
N/A

0.70
0.71
3.39
5.30

In Table 56,
5 the layer numbers
n
were
e shifted to allow easier coomparison off similar layerr types. For
example, in the five‐layyer model, th
he original AC
C surface and the overlay aare treated ass separate layyers.
In the oth
her models, th
hese two laye
ers are combined, and the combined AC
C layer is reprresented in th
he
Layer 1 co
olumn, with “not
“
applicablle” indicated in the Layer 2 column. Seeed moduli and Poisson’s raatios
were omitted for breviity because of
o their low se
ensitivity; the values used for each layer type were tthe
same as the values use
ed in the regu
ular backcalcu
ulation.
The four‐ and five‐laye
er models yielld very similar results. Thee difference iss that in the fo
our‐layer model,
the two AC
A layers were
e combined, and
a the backccalculated moodulus of thee combined laayer is a weigh
hted
average of
o the modulu
us of the two separated layyers. The bac kcalculated m
moduli for thee remaining laayers
are quite similar, as is the
t RMSE.
The three
e‐ and four‐layyer models yiield very diffe
erent results ffor the base aand subgradee layers. In thee
three‐laye
er model, the
e subgrade is treated
t
as a homogeneou
h
s half‐space, whereas in th
he four‐layer
model the
e subgrade is broken into an
a upper and
d lower subgraade layer. In the four‐layeer model, the
backcalcu
ulated modulu
us for the upp
per subgrade is 582 ksi, whhich is exceed
dingly high fo
or an unbound
d
material. In the three‐llayer model, the backcalcu
ulated modul us for the combined subgrade layer is 110
ksi, which
h is still very high
h for an unbound materrial. The backkcalculated modulus is quitte similar bettween
the two models.
m
The RMSE
R
for the three‐layer
t
model
m
is 3.39 ppercent, whicch is high in ggeneral termss and
quite high
h for an averaaged deflectio
on basin.
Figure 49 compares the measured deflection
d
bassin to the synnthetic deflecction basins caalculated for the
t
models.
m
The fo
our‐ and two‐‐layer modelss were omitteed for clarity b
because the
five‐ and three‐layer
deflection
n basins for th
he five‐ and fo
our‐layer models are quitee similar, as aare the deflecction basins fo
or the
three‐ and
d two‐layer models.
m

Figure 49. Measured
M
ve
ersus Synthettic Deflection Basins
83

As can be seen from Figure 49 and Table 55, the models in which the upper and lower subgrades are
combined (i.e., the three‐ and two‐layer models) overpredict deflections at the 18‐ and 24‐inch offsets,
and underpredict deflections at the 60‐inch offset, yielding a flatter deflection basin than the measured
deflection basin and that computed by the four‐ and five‐layer models.
This phenomenon is unlikely to be due to measurement error. Although the magnitude of the
deflections is low, random error should be minimized by the large number of actual basins used to
compute the average basin. Systematic error, either in deflection measurements or sensor offset, is an
unlikely explanation because the phenomenon is apparent in the results for some of the sections tested
on December 10, 2003, but not others.
Backcalculation, and indeed all FWD analysis procedures in common use, is subject to a number of
assumptions. Among these assumptions are that pavement layers are homogeneous, isotropic and
elastic, and that the pavement response to dynamic loads (e.g., moving wheel loads or FWD load pulses)
is the same as its response to static loads. Since the layered elastic algorithms used in backcalculation
are similar to (and in some cases the same as) those used in mechanistic pavement design
methodologies such as the AASHTOWare Pavement ME Design software, problems associated with the
violation of these assumptions are not limited to the analysis of FWD data, but represent fundamental
limitations in our ability to understand, model, and predict the behavior of pavement systems. Indeed,
the most amazing thing about pavement modeling is that it gives reasonable results most of the time.
The pavements at this test section (and indeed all pavements to one degree or another) are not
homogeneous or isotropic. However, errors related to violations of modeling assumptions regarding
homogeneity or isotropy should be greatest for the control section (040501), which had the greatest
distress density throughout the analysis period. However, the stiff layer phenomenon does not occur for
this test section, indicating that this phenomenon has other causes.
Dynamic effects have been understood to have a large influence on the analysis of deflection data since
the very beginning of pavement deflection measurement. There have been numerous research projects
related to incorporating dynamic effects in the analysis of FWD data; however, none of the products of
these studies have met with acceptance either in the industry or in the research community at large.
Beyond issues with usability and robustness, there remains the question of what to do with the results,
as current pavement design tools, including the Mechanistic‐Empirical Pavement Design Guide (NCHRP
2004), are still based on static layered‐elastic analysis. (Although the MEPDG does adjust the modulus of
the AC layer based on loading frequency, its response model is fundamentally static.)
Dynamic effects can be visualized using hysteresis plots. Two hysteresis plots are shown in Figures 50
and 51. The first is for Section 040507 on December 10, 2003, where the stiff layer phenomenon occurs.
The second is for Section 040559 on December 12, 2003, where the stiff layer phenomenon does not
occur.

84

Figurre 50. Hysteresis Plot, Secction 040507 (Stiff Layer P
Phenomenon)

Figure 51. Hysteresis
H
Plo
ot, Section 040559 (Stiff Laayer Phenom
menon Not Ob
bserved)

The data for
f Section 04
40559 shows much higherr damping thaan the data fo
or Section 040
0507. The
damping can
c be seen in the decreasse in area included in the lloop with increasing distan
nce from the load
plate. Thiss lack of damping has the effect of increasing deflecction measureed for the outter sensors ab
bove
that predicted by layerred‐elastic theory, and is a likely explannation for thee stiff layer ph
henomenon.
DEFLECTIO
ON ANALYSISS KEY FINDINGS
Three methods of anallyzing deflection data have
e been presennted in this seection: normalized deflecttion,
AASHTO 1993,
1
and layered‐elastic backcalculatio
b
on. All three aagree that thee sections at this project aare
extremelyy stiff. The mo
ost sophisticaated method used, layeredd‐elastic backkcalculation, ggenerally

85

produced highly variable and unreasonable results, and these results should be used with extreme
caution. One source of difficulty for the layered‐elastic backcalculation is dynamic effects resulting from
poor damping. Errors associated with these dynamic effects should be expected in any layered‐elastic
modeling of these pavements, including that performed by the MEPDG, although these errors may be
less obvious.
The simpler AASHTO 1993 analysis procedure yielded less variable and more reasonable results,
although dynamic effects should still be expected to result in a slight underprediction of Mr. Despite this,
the SNeff calculated for Sections 040507 and 040504 is still significantly higher than expected based on
assumed layer coefficients.
The FWD data show significant variability in preconstruction SNeff among the sections. The
preconstruction SNeff in Section 040507 was 61 percent greater than in Section 040505. These results
are supported by coring results, which show similar variability in AB thickness. Analyses of the
differential performance of the sections based solely on the difference in construction methods are
therefore questionable.
Based on the increase in SNeff from the preconstruction testing, thicker overlays performed better than
thin overlays, virgin AC materials performed better than recycled materials, and intensive surface
preparation performed better than minimal surface preparation. In the testing performed 17 months
after construction, the virgin overlays had a 31 percent greater increase in SNeff than the recycled
overlays. The thick overlays had a 24 percent greater increase in SNeff than the thin overlays. The
intensive surface preparation sections had a 17 percent greater increase in SNeff than the minimal
surface preparation sections.
By the time of the final test, 163 months after construction, the thickness of the overlay had become the
most important factor. At that time, the thick overlays had a 77 percent greater increase in SNeff than
the thin overlays. The virgin overlays had a 55 percent greater increase in SNeff than the recycled
overlays. The intensive surface preparation sections had a 39 percent greater increase in SNeff than the
minimal surface preparation sections.

86

CHAPTER 3. SPS‐5 DISTRESS ANALYSIS
This chapter describes the analyses and evaluations of distress data collected on the Arizona SPS‐5
project using LTPP manual survey techniques (Miller and Bellinger 2003). Surface distress provides
powerful information about the nature and extent of pavement deterioration, which can be used to
quantify performance trends as well as to investigate the contribution of design features on service life.
All 11 SPS‐5 test sections were constructed consecutively and exposed to the same traffic‐loading,
climate, and subgrade conditions, which allowed for direct comparisons between layer configurations
and design features without confounding effects introduced by different in situ conditions.
AC DISTRESS TYPES
Multiple distress types can cause deterioration in asphalt surfaces (Huang 1993), such as:















Fatigue cracking: A series of interconnecting cracks caused by repeated traffic loading. Cracking
initiates at the bottom of the asphalt layer where tensile stress is the highest under the wheel
load. With repeated loading, the cracks propagate to the surface.
Longitudinal wheelpath (WP) cracking: Cracking parallel to the centerline occurring in the WP.
This cracking can be the early stages of fatigue cracking or can initiate from construction‐related
issues such as paving seams and segregation of the mix during paving. In the latter case,
cracking is typically very straight (with no meandering).
Longitudinal non‐wheelpath (NWP) cracking: Cracking parallel to the centerline occurring
outside the WP. This cracking is not load‐related and can initiate from paving seams or where
segregation issues occurred during paving. Cracking can also be caused by tensile forces
experienced during temperature changes. Pavements with oxidized or hardened asphalt are
more prone to this type of cracking.
Transverse cracking: Cracking that is predominantly perpendicular to the pavement centerline.
Cracking starts from tensile forces experienced during temperature changes. Pavements with
oxidized or hardened asphalt are more prone to this type of cracking.
Block cracking: Cracking that forms a block pattern and divides the surface into approximately
rectangular pieces. Cracking initiates from tensile forces experienced during temperature
changes. This distress type indicates that the AC has significantly oxidized or hardened.
Raveling: Wearing away of the surface caused by dislodging of aggregate particles and loss of
asphalt binder. Raveling is caused by moisture stripping and asphalt hardening.
Bleeding: Excessive bituminous binder on the surface that can lead to loss of surface texture or
a shiny, glass‐like, reflective surface. Bleeding is a result of high asphalt content or low air void
content in the mix.
Rutting: A surface depression in the WPs. Rutting can result from consolidation or lateral
movement of material due to traffic loads. It can also signify plastic movement of the asphalt
mix because of inadequate compaction, excessive asphalt, or a binder that is too soft given the
climatic conditions.

87

Table 57 summarizes
s
these
t
flexible pavement distress types aand their asso
ociated failurre mechanism
ms.

5 Flexible Pavement Disttress Types aand
Table 57.
Failure Mechanism
ms
F ailure Mechan
nism
Distress Type

Traffic/LLoad
Relateed

Climatee/Materials
Reelated

Faatigue crackingg
Lo
ongitudinal WP
P cracking
Lo
ongitudinal NW
WP cracking
Trransverse cracking
Bllock cracking
Raaveling
Blleeding
Ru
utting

RESEARCH
H APPROACH
H
Investigattors began the analysis by reviewing all of the distreess data colleccted at each ttest section to
o
identify su
uspect or inco
onsistent info
ormation. The
ey used photoos and distresss maps to veerify quantitiees
reported in the databaase. Because of
o the subjecttive nature off the data colllection techn
nique (raters had
d
type and severity based on a se
et of rules), vaariation is expected in disttress data.
to select distress
Most LTPP
P distress datta are reporte
ed at three se
everity levels:: low, moderaate, and high.. Inconsistenccies
between severity levells (within one
e distress type
e) are one of the largest so
ources of variiability in disttress
a
on thhree separatee severity levels for each
data (Rada et al. 1999)). In addition, conducting analyses
p
with rresults that are difficult to
o interpret. To
o
distress tyype becomes an increasinggly complex process
reduce vaariability and to
t consolidatte the informaation for anallyses, the quaantities from the three sevverity
levels werre summed in
nto one comp
posite value fo
or the resear ch presented
d.
In addition to the strucctural and envvironmental distress
d
factoors used to asssess SPS‐5 seection
holes,
performance, investigaators also incorporated ruttting, patchinng, and other surface defects (i.e., poth
bleeding, and raveling)) into the anaalyses. Ruttingg data report ed in this study were geneerated using a
r
wire line reference.
The experrimental desiggn of the SPS‐5 project is such
s
that rep licate data w
were not colleccted. Therefo
ore,
standard statistical com
mparisons (i.e
e., t tests) to determine thhe significancee of findings ccould not be
d. Instead, the evaluation consisted of graphical com
mparisons between test seections from data
conducted
collected at the same time.
t

88

OVERALL PERFORMANCE TREND OBSERVATIONS
While gathering pavement distress data for this research, investigators became aware of a few
significant trends impacting the project’s overall pavement performance. These observations were
clearly driving issues for this project and were intrinsically important to the distress performance.
Section 040501, the test section that did not receive an overlay, performed the worst of all SPS‐5
sections. Within six years, this section had deteriorated to a level that required reconstruction and,
therefore, it was taken out of the experiment.
All minimum preparation, 2‐inch overlay sections (040502, 040505, and 040560) showed evidence of
structurally related distress within seven years after construction, with Sections 040502 and 040505
accumulating fatigue to over half of the section area by the end of the study. These sections also
exhibited pumping eight years after construction.
In general, test sections receiving an intensive surface preparation before overlay treatment performed
much better than the other test sections in the project. Test sections with virgin AC overlays also
performed better when compared to the recycled AC sections (040504 to 040503).
Compared to the rest of the SPS‐5 test sections, Section 040507 exhibited significantly smaller amounts
of damage accumulation. The pavement structure for this section is composed of 20.7 inches of granular
base, 2.4 inches of existing inlay AC, 2.7 inches of new inlay AC, and 4.1 inches of overlay AC.
In general, the extent of distresses in all the sections increased over time except for Section 040502,
which quickly became distressed and only the severity of distresses increased over time.
Table 58 lists the dates when surveys were performed at each test section. Figures 52 through 62
illustrate the overall structural and environmental performance trends for each section. These trends
are relatively consistent and within the expected range of variation. Drops in the distress graph typically
indicate the distress propagating into a different distress (i.e., longitudinal WP cracking forming into
fatigue cracking). The maintenance work (i.e., fog seal and crack sealing) performed on the project did
not significantly mask the severity and extent of distresses. Nonetheless, distress comparisons
formulated in this analysis focused primarily on the extent of distress. Appendix B provides the complete
work history of each site.

89

Table 58. MDS
M Dates by Section
Su
urvey Date

0501
0

0502

0503

0504

0505

01
1/15/1991
10
0/18/1994‐
10
0/20/1994
09
9/12/1996‐
09
9/12/1996
11
1/13/1997‐
11
1/13/1997
12
2/09/1998‐
12
2/10/1998
12
2/13/1999‐
12
2/14/1999
10
0/16/2000‐
10
0/18/2000
11
1/28/2001‐
11
1/30/2001
12
2/10/2002‐
12
2/12/2002
12
2/10/2003‐
12
2/12/2003
12
2/07/2004‐
12
2/08/2004
12
2/05/2005‐
12
2/05/2005
09
9/15/2008‐
09
9/15/2008

90

05066

0507

05508

0509

0
0559

0560

040502

040501

4500

4500
Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

4000
3500

600

4000
3500

500

600

500

3000

1500

Feet

300

2000

Square Feet

400
2500

400
2500
300

2000
1500

200

200

1000

1000

100

100

500

500
0

0

0

Ap
r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

Ap
r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Date

Date

Figure 53. Structural and Environmental Distress
Trends in Section 040502

Figure 52. Structural and Environmental Distress
Trends in Section 040501

040503

040504

4500

3500

4500
Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

600

4000
3500

500

600

500

3000

1500

Feet

300

2000

Square Feet

400
2500

400
2500
300

2000
1500

200

1000

200

1000
100

100

500

500
0

0

Ap

r-9

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0
Ap

Square Feet

3000

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

Date

Date

Figure 54. Structural and Environmental Distress
Trends in Section 040503

Figure 55. Structural and Environmental Distress
Trends in Section 040504

91

Feet

4000

Feet

3000
Square Feet

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

040505

040506

4500

4000
3500

500

500

3000

300

2000
1500

Square Feet

400
2500

Feet

Square Feet

3000

600

400
2500
300

2000
1500

200

1000

200

1000
100

100

500

500

0

0

0

Ap

Ap

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Date

Date

Figure 56. Structural and Environmental Distress
Trends in Section 040505

Figure 57. Structural and Environmental Distress
Trends in Section 040506

040507

040508

4500
4000
3500

4500
Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

600

4000
3500

500

3000

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

600

500

3000

1500

Feet

300

2000

Square Feet

400
2500

400
2500
300

2000
1500

200

1000

200

1000
100

100

500

500

0

0

0

Ap

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Ap

Square Feet

Feet

3500

600

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

Date

Date

Figure 58. Structural and Environmental Distress
Trends in Section 040507

Figure 59. Structural and Environmental Distress
Trends in Section 040508

92

Feet

4000

4500
Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

040509

040559

4500

3500

600

4000
3500

500

600

500

3000

300

2000
1500

Square Feet

400
2500

Feet

Square Feet

3000

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

400
2500
300

2000
1500

200

1000

200

1000
100

100

500

500

0

0
r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

Date

Date

Figure 60. Structural and Environmental Distress
Trends in Section 040509

Figure 61. Structural and Environmental Distress
Trends in Section 040559

040560
4500
4000
3500

Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

600

500

300

1500

Feet

400
2500

200

1000
100
500
0
r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Ap

Square Feet

3000

2000

0

Ap

Ap

r-9
Ap 0
r-9
Ap 1
r-9
Ap 2
r-9
Ap 3
r-9
Ap 4
r-9
Ap 5
r-9
Ap 6
r-9
Ap 7
r-9
Ap 8
r-9
Ap 9
r-0
Ap 0
r-0
Ap 1
r-0
Ap 2
r-0
Ap 3
r-0
Ap 4
r-0
Ap 5
r-0
Ap 6
r-0
Ap 7
r-0
8

0

Date

Figure 62. Structural and Environmental Distress
Trends in Section 040560

93

Feet

4000

4500
Fatigue (sq ft)
Block Crack (sq ft)
WP L Crack (ft)
NWP L Crack (ft)
Trans Crack (ft)

Performance Comparisons
Investigators conducted in‐depth analyses and comparisons all of the SPS‐5 test sections. Figure 63
summarizes the structural distress and Figure 64 summarizes the environmental distress for each
section. Results in both distress charts are based on the data collected in December 2002, the last date
that every section was surveyed manually.
Figure 65 summarizes the structural distresses and Figure 66 summarizes the environmental distresses
for each core section. Results in both distress charts are based on the data collected in December 2005,
the last date that every core section was surveyed manually.
Figure 67 summarizes rutting in each section in 2002, while Figure 68 summarizes rutting in each core
section in 2005. In 2008, investigators conducted a forensic study on Sections 040502, 040505, 040506,
and 040509. After excavating trenches, investigators found rutting mainly in the top AC layer; virtually
no rutting was detected in the lower layers. The average rut depth of both trenches was 6.5 mm
(Nichols Consulting Engineers, unpublished data, 2010).
All sections exhibited less than 10 mm of rutting after more than seven years in service, which is below
the level required to trigger improvements in most pavement management systems. Sections 040502
and 040509 exhibited large amounts of localized rutting; however both sections had other distresses
that would have triggered improvements in most pavement management systems. Therefore, rutting
was not the driving factor in the overall condition of the pavement.
2002 Structural Performance
600

4000
3500

500

400

2500

300

2000
1500

Linear ft.

Square ft.

3000

200

1000
100

500

0

in
.R
es
05
t.
03
2"
M
R
in
AP
.R
AC
es
t.
5"
00
R
50
AP
4
M
AC
in
.R
es
05
t.
05
5"
M
AC
in
.R
es
05
t.
2"
06
AC
In
t.
R
es
05
t.
2"
07
AC
In
t.
05
R
08
es
In
t.
t.
5"
R
AC
es
05
t.
5"
09
R
In
AP
t.
R
AC
es
05
t.
2"
59
R
In
AP
t.
R
AC
es
t
05
.3
60
"R
AP
M
in
AC
.R
es
t.
2"
AR
AC

0

05
02

M

Fatigue Cracking (sq. ft)

WP Longitudinal Cracks
(ft)

Figure 63. 2002 Structural Performance of All SPS‐5 Sections

94

2002 Environmental Performance
600

4000
3500

500

400

2500

300

2000
1500

Linear ft.

Square ft.

3000

200

1000
100

500

0

in
.R
es

t.
2"
R
AP

in
.R
AC
es
t.
5"
00
R
50
AP
4
M
AC
in
.R
e
05
st
.5
05
"A
M
C
in
.R
es
05
t.
2"
06
AC
In
t.
R
es
05
t.
2"
07
AC
In
t.
05
R
08
es
In
t.
t.
5"
R
AC
es
05
t.
5"
09
R
In
AP
t.
R
AC
es
05
t.
2"
59
R
In
AP
t.
R
AC
es
t.
05
3
"R
60
AP
M
in
AC
.R
es
t.
2"
AR
AC

0

05
03

05
02

M

M

Block Cracking (sq. ft)
NWP Longitudinal Cracks
(ft)
Transverse Cracks
(ft)

Figure 64. 2002 Environmental Performance of All SPS‐5 Sections

2005 Structural Performance
4000

600

3500

500

3000

Square ft.

2000

300

1500

Linear ft.

400

2500

200

1000
100

500

0
AC

AC

AP

AP

"R

"R

t.
2

t.
5

In
t.
R
es

Fatigue Cracking (sq. ft)

05
09

05
08

In
t.
R
es

In
t.
R
es
t

.5

.2
05
07

In
t.
R
es
t

t.
2

05
06

"A
C

"A
C

C
"A

C
"A
M
05
05

00
50
4

M

in
.

R
es

in
.R
es

t.
5

t.
5"
R
AP

in
.R
es
M

05
03

05
02

M

in
.R
es

t.
2"
R
AP

AC

AC

0

WP Longitudinal Cracks
(ft)

Figure 65. 2005 Structural Performance of SPS‐5 Core Sections

95

M

M

in
.R
es
t.

R
es
t.
5
"A
C

AC

2"
AC
In
t.
R
es
t.
05
2"
07
AC
In
t.
05
R
es
08
t.
In
5"
t.
AC
R
es
t.
05
5"
09
R
AP
In
t.
AC
R
es
t.
05
2"
59
R
AP
In
t.
AC
R
es
t.
3"
05
R
60
AP
M
AC
in
.R
es
t.
2"
AR
AC

05
06

M

M
in
.

AC

Rutting Index (mm)
8

05
05

4

"R
AP

"R
AP

in
.R
es
t.
5

in
.R
es
t.
2

00
50

05
03

05
02

M

M

4

6

2.7
2.9

96
05
09

05
08

In
t.
R
es
t.

2"

5"

AC

AC

t.
5

C

6.9
R
AP

C

AC

AC

"A

"A
C

2"
A

"A
C

R
AP

In
t.
R
es
In
t.
R
es
t.

05
07

in
.R
es
t.

R
es
t.
5

"R
AP

"R
AP

In
t.
R
es
t.
2

M

M
in
.

05
06

05
05

4

in
.R
es
t.
5

in
.R
es
t.
2

00
50

05
03

05
02

Square ft.
3500

2500

2000
300

1500

500

0

4.0

Figure 67. 2002 Rutting Index Summary of All SPS‐5 Sections

Linear ft.

4000
2005 Environmental Performance
600

3000

500

400

1000
200

100

0

Block Cracking (sq. ft)

NWP Longitudinal Cracks
(ft)
Transverse Cracks
(ft)

Figure 66. 2005 Environmental Performance of SPS‐5 Core Sections

2002 Rutting Index (0500)

12

10
8.0

6.2
6.9

5.5
3.9

2.6

2

0

2005 Rutting Index (0500)
12
10.2
10
Rutting Index (mm)

8.3
8
6.7

6.5

6

5.5

5.1

4

3.4

3.2

2

AC
M
in
.R
es
t.
05
5"
05
AC
M
in
.R
es
t.
05
2"
06
AC
In
t.
R
es
t.
05
2"
07
AC
In
t.
05
R
es
08
t.
In
5"
t.
AC
R
es
t.
05
5"
09
R
AP
In
t.
AC
R
es
t.
2"
R
AP
AC

R
AP

4

5"

00
50

in
.R
es
t.
M

05
03

05
02

M

in
.R
es
t.

2"

R
AP

AC

0

Figure 68. 2005 Rutting Index Summary of SPS‐5 Core Sections

In the experimental design of the SPS‐5 project, the milled layer was to be replaced with the same
overlay material used in the section before overlay placement. From an LTPP perspective, this thickness
is not considered to be part of the overlay thickness. However, this additional material does appear to
affect the structural performance of the sections. Fatigue cracking, the standard measure of pavement
performance, seems to have a direct correlation with the thickness of the AC material. Figure 69 shows
that sections where the total thickness of all AC layers (new and existing) was greater than 6 inches had
a greater resistance to fatigue cracking than sections with total AC layer thicknesses less than 6 inches
(based on limited observations of the eight core sections). Sections 040505 and 040502, in which the
total AC material thickness was less than 6 inches, incurred fatigue cracking in over half the total area of
the sections.

97

2005 Fatigue Cracking
vs.
AC Material Thicknesses
4000
2005 Fatigue Cracking
Fatigue Cracking (sq. ft)

3500
3000
2500
2000
1500
1000
500
0
0508

0507

0506

0504

0503

0509

0505

0502

0

2

Thickness (in)

4

6

8
Overlay Virgin AC
Overlay Recycled AC

10

Fill Virgin AC
Existing AC
12

Figure 69. Fatigue Cracking and AC Material Thicknesses

There is a debate in the paving community between top‐down and bottom‐up cracking. The classic
fatigue cracking model looks at bottom‐up cracking resulting from repeated tensile stresses at the
bottom of the AC pavement. However, increasingly evidence suggests that some of the WP cracking
initiates at the surface of the pavement and progresses downward, particularly in thicker pavement
sections (Al‐Qadi et al. 2008; Uhlmeyer et al. 2000).
Forensic sampling of Sections 040502, 040505, 040506, and 040509 determined whether the crack
mechanism was primarily top‐down or bottom‐up cracking. AC core samples from the sections and
trench excavations in Sections 040506 and 040509 were visually inspected. In general, the core samples
exhibited medium to high severity cracks initiated at the pavement surface and extending to the inlay

98

layer, showing a top‐down cracking pattern. Visual inspection of the trenches revealed no reflective
cracks (i.e., cracks from underlying pavement layers propagating to surface layers) at the bottom of the
AC slab layer, reconfirming the crack mechanism as top‐down cracking.
The following observations of the SPS‐5 sections are based on structural and environmental distress
trends (illustrated in Figures 63 and 64). General (structural and environmental) and specific distress
trends are discussed along with probable causes for sudden increases and decreases within the
identified distress trends.
Section 040501 (Control Section)
Section 040501 only received the required routine maintenance needed to keep the section in a safe
and functional condition. The section had 1992 ft2 of fatigue cracking in 1991 and 4000 ft2 of block
cracking. In 1994, fatigue cracking spread into areas that had block cracking, which resulted in 2781 ft2
of fatigue cracking and 3272 ft2 of block cracking. This section was placed out of study in October 1996
because of significant deterioration. At the time it was taken out of study, the entire section exhibited
moderate severity block and fatigue cracking.
Section 040502 (Minimum Preparation, 2‐Inch Recycled Asphalt Pavement (RAP) Overlay)
Section 040502 performed very poorly against structural and environmental distresses. Distresses began
to appear in 1994 and peaked in 1998, when the primary distresses of fatigue and block cracking had
spread throughout the entire section. Investigators observed very little fatigue cracking and some WP
longitudinal cracking in 1994. In the following years, the WP longitudinal cracking progressed into
fatigue cracking. In 1999, the fatigue had spiked to 3282 ft2, existing in both WPs, and remained
constant until the end of the study. Some rater variability between fatigue cracking and block cracking
caused a slight increase and respective decrease between the two distresses in 2004. In 2008, fatigue
cracking spread in block cracked areas and peaked at 3684 ft2. No block cracking, some longitudinal
NWP cracking, and substantial transverse cracking was observed in 1994. Longitudinal NWP and
transverse cracking continued to increase until 1999 when both distresses dropped to zero and block
cracking spiked up to 2955 ft2. Block cracking then remained fairly constant until a slight drop in 2008
because of the spread of fatigue cracking into block cracked areas. This section also exhibited large
amounts of pumping and rutting that increased over time.
Section 040503 (Minimum Preparation, 5‐Inch RAP Overlay)
Section 040503 performed poorly against environmental distresses, with large amounts of transverse
and longitudinal cracking, but little or no fatigue or block cracking. The primary structural distress,
longitudinal WP cracking, appeared in 1996 and steadily increased to 388 ft in 2002, after which some
longitudinal WP cracking progressed into fatigue cracking. By 2008, fatigue cracking had increased to
104 ft2. Transverse cracking appeared in 1996, increased to 455 ft in 2001, and then gradually increased
to 502 ft in 2008. Longitudinal NWP cracking appeared in 2000 and steadily increased to 454 ft in 2008.
Investigators did not observe block cracking at this section. Minimal pumping was observed in 2005, and
the section had average rut resistance compared to the other sections.

99

Section 040504 (Minimum Preparation, 5‐Inch AC Overlay)
Section 040504 performed well against structural and environmental distresses throughout the study.
Investigators observed very little fatigue or longitudinal WP cracking. Transverse cracking steadily
increased from 7 ft in 1998 to 148 ft in 2005. Longitudinal NWP cracking appeared in 2002 at 50 ft and
increased to 225 ft in 2005.
Section 040505 (Minimum Preparation, 2‐Inch AC Overlay)
Section 040505 performed poorly against structural and environmental distresses. Distresses were
minor in 1994, but steadily increased until the end of the study. Fatigue cracking was the primary
distress observed by the end of the study, with some transverse, longitudinal NWP, and block cracking
also recorded. The primary structural distress, fatigue cracking, consistently increased from 1994 to
2008 and peaked at 4285 ft2. No block cracking was observed from 1994 to 2005, but suddenly
increased to 1206 ft2 in 2008. Transverse and longitudinal NWP cracking steadily increased to a peak in
2003 at 429 ft and 407 ft, respectively. Transverse and longitudinal NWP cracking dropped to 100 ft and
156 ft, respectively, in 2008 because of the spread of fatigue cracking and block cracking. The section
also exhibited large amounts of pumping and moderate rut resistance, and performed better than its
recycled counterpart, Section 040502.
Section 040506 (Intensive Preparation, 2‐Inch AC Overlay)
Section 040506 performed well against structural distresses but poorly against environmental
distresses. Distresses rated as longitudinal WP cracking propagated into fatigue cracking in 2003. Fatigue
cracking then steadily increased from 10 ft2 in 2003 to 64 ft2 in 2008. Longitudinal NWP and transverse
cracking steadily increased throughout the study period. Longitudinal NWP cracking appeared in 1998 at
42 ft and increased to 400 ft in 2008; transverse cracking was observed at 7 ft in 1999 and increased to
219 ft in 2008. Minimal pumping was observed in 2008.
Section 040507 (Intensive Preparation, 5‐Inch AC Overlay)
Section 040507 performed well against structural and environmental distresses throughout the study.
Only longitudinal NWP and transverse cracking were observed at this section. Longitudinal NWP
cracking appeared from 2003 to 2005 at an average of 11 ft; transverse cracking appeared from 2004 to
2005 at an average of 20 ft. The section exhibited no pumping and had average rut resistance.
Section 040508 (Intensive Preparation, 5‐Inch RAP Overlay)
Section 040508 performed well against structural distresses but poorly against environmental
distresses. In 1999, investigators observed longitudinal WP cracking that increased to 45 ft in 2008. No
fatigue cracking was observed at this section throughout the study. Investigators observed transverse
cracking in 1997 and longitudinal NWP cracking in 1998. Both distresses steadily increased and in 2005,
433 ft of transverse and 552 ft of longitudinal NWP cracking were observed. No block cracking was
observed at this section throughout the study, and the section had below average rut resistance
compared to other test sections.

100

Section 040509 (Intensive Preparation, 2‐Inch RAP Overlay)
Section 040509 performed moderately well against structural distresses but poorly against
environmental distresses. Longitudinal WP and NWP cracking and transverse cracking quickly increased
soon after the study started. Longitudinal WP cracking appeared in 1996 at 6 ft and increased steadily to
its peak of 451 ft in 2002. In the following years, longitudinal WP cracking dropped as it turned into
fatigue cracking. In 2008, investigators observed 3 ft of WP longitudinal WP cracking and 910 ft2 of
fatigue cracking. Outliers of this trend for longitudinal cracking include a spike in 1998 and a drop in
2003, both of which were attributed to rater variability. Transverse cracking was observed in 1994 at 55
ft and steadily increased to 613 ft2 in 2008. Longitudinal NWP cracking appeared in 1997 at 116 ft2 and
steadily increased to 519 ft2 in 2008. No block cracking was observed at this section. Section 040509
exhibited the second poorest resistance to rutting, and pumping was observed in 2005 and 2008.
Section 040559 (Intensive Preparation, 3‐Inch RAP Overlay [Inverted])
Section 040559 performed well against structural distresses but poorly against environmental
distresses. Longitudinal WP cracking appeared in 1998 at 14 ft and steadily increased to a peak of 38 ft
in 2001. Longitudinal WP cracking remained constant at 36 ft in 2002 and then dropped to 6 ft in 2003
due to rater variability. Only 3 ft2 of fatigue cracking appeared at this section (in 1998). Transverse
cracking appeared in 1996 at 1 foot and steadily increased to 344 ft in 2003. Longitudinal NWP cracking
spiked at 235 ft in 1999 and steadily increased to 506 ft in 2003. No block cracking was observed at this
section, and the section exhibited similar environmental performance as Sections 040509 and 040508.
The performance of this section cannot be compared to conventional noninverted sections because
such a section was not constructed for this study.
Section 040560 (Minimum Preparation, 2‐Inch ARAC Overlay)
Section 040560 performed moderately poor against both structural and environmental distresses. The
ARAC design experienced low stability. A mix with a stability value of 1100 lb was used in construction,
300 lb less than the target range (Hossain et al. 1996). The spike in fatigue cracking in 1997 was
attributed to rater variability. The observed fatigue cracking reached 1954 ft2 in 2002, which was only
surpassed by Sections 040502 and 040505. Transverse cracking was observed in 1996 and increased
steadily over time from 3 ft to 176 ft in 2002. Longitudinal NWP cracking was observed in 1997 and
increased rapidly to 403 ft in 2002. Pumping was observed in 1998 until the final survey in late 2002.
When the project began in 1990, the initial cost of this section was similar to the 5‐inch overlay sections.
DISTRESS ANALYSIS KEY FINDINGS
The distress data captured at the project provided valuable insight into pavement performance, design,
management, and construction. Highlights from the SPS‐5 distress analysis follow.

101

Best and Worst Performers
Section 040507, the thick (5‐inch) virgin AC overlay with intensive preparation, performed the best
among all of the SPS‐5 test sections. Compared to Section 040507, Sections 040504 and 040506 both
performed as well in structural distresses but worse in environmental distresses.
Section 040502, the thin (2‐inch) recycled AC overlay with minimal preparation, performed the worst
among all of the SPS‐5 test sections, excluding the control section. Section 040502 reached the peak of
its pavement distress (fatigue and block cracking throughout the entire section) nine years after
rehabilitation, after which the distresses remained relatively constant. Section 040505 also reached an
equal amount of structural distress by the end of the study, 19 years after the rehabilitation.
Core Section Performance by Rehabilitation Feature
Sections with intensive surface preparation (040506, 040507, 040508, and 040509) performed better in
structural distresses than pavements with minimal surface preparation (040502, 040503, and 040505).
Section 040504 was the exception to this trend as the section performed well despite receiving only
minimal surface preparation. Section 040504 received a thick virgin AC overlay, which contributed to its
superior structural performance.
Sections with thick overlays (040503, 040504, 040507, and 040508) performed better in structural
distresses than pavements with thin overlays (040502, 040505, and 040509). Section 040506 was the
exception to this trend as the section performed well despite receiving only a thin overlay. However,
total combined thickness of all AC layers (existing and new) ranks Section 040506 as the third thickest of
the eight core sections (despite being a thin overlay section). Also, the virgin asphalt and intensive
preparation contributed to Section 040506’s superior structural performance.
Sections with virgin AC overlays (Sections 040504, 040506, and 040507) performed better in both
structural and environmental distresses than pavements with recycled AC overlays (Sections 040502,
040503, and 040509). Section 040505 was an exception to this trend as the section performed poorly in
fatigue cracking despite using a virgin AC overlay. Section 040505 received minimal surface preparation
and a thin overlay, which contributed to its relatively poor structural performance. Section 040508 was
also an exception to the trend as the section performed well despite receiving a recycled AC overlay.
Section 040508 also received intensive surface preparation and a thick AC overlay, which contributed to
its superior structural performance.
Sections that received two of three rehabilitation methods (Sections 040504, 040506, and 040508) that
improve performance (i.e., intensive preparation, virgin AC, and 5‐inch overlay) performed better than
sections that received only one of these three rehabilitation methods (Sections 040503, 040505, and
040509).
Core Section Performance by Distress Type
Tables 59 and 60 show the comparisons made between sections with respect to relative performance
against structural and environmental distresses, respectively.

102

Tab
ble 59. Core Section
S
Perfo
ormance Agai nst Structuraal Distresses
040502 040
0503 040504 040505 0400506 0405077 040508 04
40509
040502*
040503*
040504*

N/A

040505*
040506*

N/A

040507*

N
N/A

N/A

040508*

N/A
N/A

040509*
*Did sectio
on perform better?
Yes, significan
ntly –

Noo, significantly –

Yes, slightly –

Noo, slightly –

N/A – Section
n performed eq
qually

Table 60. Core Secction Perform
mance Againstt Environmen
ntal Distressees
040502 040
0503 040504 040505 0400506 0405077 040508 04
40509
040502*
040503*
040504*

N
N/A

040505*
040506*

N/A

040507*
040508*
040509*
*Did section perform be
etter?
Yes, significan
ntly –

Noo, significantly –

Yes, slightly –

Noo, slightly –

N/A – Section
n performed eq
qually

Comparisons were made between sections
s
receiving only onee of three reh
habilitation m
methods (Secttions
040503, 040505,
0
and 040509)
0
that improve perfformance (i.e ., intensive preparation, vvirgin AC, and
5‐inch ove
erlay). Key fin
ndings follow::


Sttructural disttresses: Sectio
on 040503 (5‐inch overlayy) performed better than SSection 04050
09
(intensive surfface preparation). Both Secctions 0405033 and 0405099 performed ssignificantly b
better
han Section 040505 (virgin
n AC overlay), suggesting t hat overlay th
hickness had a greater effect
th

103

on performance against structural distresses than surface preparation, which had a greater
effect on performance against structural distresses than virgin AC. However, total thickness of
the combined (new and existing) AC material layer may have contributed to this trend.
5‐inch overlay > intensive surface preparation > virgin AC


Environmental distresses: Section 040505 (virgin AC overlay) performed better than Section
040509 (intensive surface preparation). Both Sections 040505 and 040509 performed
significantly better than Section 040503 (5‐inch overlay), suggesting that the use of virgin AC
had a greater effect on performance against environmental distresses than surface preparation,
which had a greater effect on performance against environmental distresses than overlay
thickness.
virgin AC > intensive surface preparation > 5‐inch overlay

Comparisons were made between sections receiving two of the three rehabilitation methods (Sections
040504, 040506, and 040508) that improve performance (i.e., intensive preparation, virgin AC, and 5‐
inch overlay). Key findings follow:


Structural distresses: Section 040504 (5‐inch overlay and virgin AC overlay) performed better
than Section 040508 (5‐inch overlay and intensive surface preparation). Both Sections 040504
and 040508 performed better than Section 040506 (intensive surface preparation and virgin AC
overlay). From this, it can be concluded that the use of thick overlay had a greater effect on
performance against structural distresses than the use of virgin AC overlay, which had a greater
effect on performance against structural distresses than surface preparation. However, total
thickness of the combined (new and existing) AC material layer may have contributed to this
trend.
5‐inch overlay > virgin AC > intensive surface preparation



Environmental distresses: Section 040504 (5‐inch overlay and virgin AC overlay) performed
better than Section 040506 (intensive surface preparation and virgin AC overlay). Both Sections
040504 and 040506 performed better than Section 040508 (5‐inch overlay and intensive surface
preparation). From this, it can be concluded that the use of virgin AC had a greater effect on
performance against environmental distresses than overlay thickness, which had a greater
effect on performance against environmental distresses than surface preparation.
virgin AC > 5‐inch overlay > intensive surface preparation

The conclusions from these comparisons are that (1) sections with thick overlays will perform better
against structural distresses than sections with thin overlays regardless of other design features and (2)
sections with virgin AC overlays will perform better against environmental distresses than sections with
recycled overlays regardless of other design features.

104

Evaluation of Supplemental Sections
The inverted section (040559) performed well against structural distress (comparable to other sections
with 5‐inch overlays) and poorly against environmental distress (comparable to other sections using
recycled AC overlays). Section 040559 is most comparable to Section 040508, which had a similar
pavement structure, with varying AC layer thickness. While Section 040508 had more existing AC and
recycled AC material and greater total combined AC layer thickness, Section 040559 had more new
virgin AC material. Section 040559 performed slightly worse than Section 040508 against structural
distress, but slightly better than Section 040508 in environmental distresses.
Section 040560, the ARAC overlay section, had poor to moderate performance in this study. This could
be due to the problems encountered in the mix design, as ARAC has exhibited better performance in
other SPS experiments, such as the Arizona SPS‐6 project. Its performance against structural distresses
was better than sections using thin overlay and minimal surface preparation (Sections 040502 and
040505) and worse than all the other core sections. Its performance against environmental distresses
was better than sections using recycled AC overlay (Sections 040502, 040503, 040508, and 040509) and
worse than sections using virgin AC overlays (Sections 040504, 040505, 040506, and 040507).
Additional Observations






Sections with thicker combined (new and existing) AC material layer experienced significantly
higher resistance to fatigue cracking. Sections with at least 6 inches of AC material performed
better than thinner sections in regard to fatigue cracking resistance.
All sections exhibited less than 11 mm of rutting during the monitoring period. Rutting was not a
critical distress and did not have a significant impact in the structural performance of all
sections.
With no replicate sections, there is limited ability to assess potential variability independent of
actual performance.

105

106

CHAPTER 4. SPS‐5 ROUGHNESS ANALYSIS
This chapter provides the results of profile and roughness analyses for the LTPP SPS‐5 site. The
information presented characterizes the surface roughness of the test sections over time and links the
observations to records of pavement distress and its development. Investigators collected road profile
measurements from this site about once per year since the winter after the site was opened to traffic.
This study analyzed the profiles in detail by calculating their roughness values, examining the spatial
distribution of roughness within them, viewing them with post‐processing filters, and examining their
spectral properties. These analyses provided details about the initial roughness of the road and also
provided a basis for quantifying and explaining the changes in roughness with time.
PROFILE DATA SYNCHRONIZATION
Profile data were collected at the Arizona SPS‐5 site on 14 dates, from February 5, 1990, through
March 24, 2006 (Table 61). Raw profile data were available for visit 00 and visits 03 through 13. In each
visit for which raw data were available, investigators made a minimum of seven repeat profile
measurements. Since raw data were not available for visits 01 and 02, whenever possible, profiles for
these visits were extracted from the public database. Visit 00 took place before the original
rehabilitation and visit 01 took place just after the original rehabilitation. Investigators removed
Section 040501 from the study after visit 03 because it was in extremely poor condition.

Table 61. Profile Measurement Visits of the SPS‐5 Site
Visit
Date
00
Feb. 5, 1990
01
Sept. 21, 1990
02
Jan. 15, 1992
03
Feb. 22, 1993
04
Feb. 3, 1997
05
Dec. 9, 1997
06
Dec. 11, 1998
07
Nov. 11, 1999
08
Dec. 1, 2000
09
Nov. 15, 2001
10
Nov. 4, 2002
11
Feb. 6, 2004
12
Dec. 14, 2004
13 March 24, 2006

Time
Repeats
Sections
17:18
7
040501‐040509, 040559‐040560
21:56
—
040501‐040509
17:50‐18:49
—
040504, 040507
13:54
9
040501‐040509, 040559‐040560
09:34‐10:44
9
040502‐040509, 040559‐040560
14:04‐14:55
7
040502‐040509, 040559‐040560
12:54‐13:35
7
040502‐040509, 040559‐040560
11:30‐12:06
7
040502‐040509, 040559‐040560
10:53‐11:46
9
040502‐040509, 040559‐040560
10:49‐11:38
9
040502‐040509, 040559‐040560
12:02‐13:10
9
040502‐040509, 040559‐040560
15:24‐16:35
9
040502‐040509, 040559‐040560
12:49‐14:00
9
040502‐040509, 040559‐040560
11:54‐12:48
9
040502‐040509, 040559‐040560

107

DATA EXTRACTION
Profiles of individual test sections were extracted directly from the raw measurements for two reasons:
First, profiles were collected in visits 04 through 09 at a 0.98‐inch sample interval and in visits 10
through 13 at a 0.77‐inch sample interval. These data appeared in the database after the application of
an 11.8‐inch moving average and decimation to a sample interval of 5.91 inches. The raw data contained
the more detailed profiles. Second, this study depended on consistency of the profile starting and
ending points with the construction layout and consistency of the section limits with time. In particular,
a previous quality check revealed that some profiles were shifted (Evans and Eltahan 2000).
Researchers used the raw data to synchronize all of the profiles to each other through their entire
history. Three indicators were available for this purpose: (1) the site layout from the construction report,
(2) event markers in the raw profiles from the start and end of each section, and (3) automated
searching for the longitudinal offset between repeat measurements.
CROSS CORRELATION
A helpful way to refine the synchronization of profile measurements is to search for the longitudinal
offset between repeat profile measurements that provides the best agreement by inspecting filtered
profile plots. However this approach is very time‐consuming. Visual assessment is also somewhat
subjective when two profiles do not agree well, which is often the case when measurements are made a
year or more apart. In this study, investigators used an automated procedure rather than visual
inspection to find the longitudinal offset between measurements.
The procedure, which is based on a customized version of cross correlation (Karamihas 2004),
designates a basis measurement that is considered to have the correct longitudinal positioning. A
candidate profile is then searched for the longitudinal offset that provides the highest cross correlation
to the basis measurement. A high level of cross correlation requires a good match of profile shape, the
location of isolated rough spots, and overall roughness level. Therefore, the correlation level is often
only high when the two measurements are synchronized. When the optimal offset is found, a profile is
extracted from the candidate measurement with the proper overall length and endpoint positions. For
the remainder of this discussion, this procedure will be referred to as automated synchronization.
In this application, investigators performed cross correlation after the IRI filter was applied to the
profiles rather than using the unfiltered profiles, which helped assign the proper weighting to relevant
profile features. In particular, it increased the weighting of short‐wavelength roughness that may be
linked to pavement distress. This enhanced the effectiveness of the automated synchronization
procedure. The long‐wavelength content within the IRI output helped ensure that the longitudinal
positioning was nearly correct, and the short‐wavelength content was able to leverage profile features
at isolated rough spots to fine‐tune the positioning.

108

Synchronization of Visits 03 Through 13
For visits 03 through 13, investigators extracted profiles of individual test sections from the raw
measurements using the following steps:
1. Establish a basis measurement for each section from visit 09. This was done using the event
markers from a raw measurement. The first repeat measurement was used for this purpose.
Event markers appeared at the start of every section, and appeared at the end of every section
except Section 040505. The event marker locations were compared to the layout provided in
the construction report (Hossain et al. 1996). They exhibited a linear relationship with a bias of
less than 0.05 percent. Once the bias was removed, no individual section starting point in the
construction report differed from the event markers by more than 5 ft.
Most of the sections were assumed to begin at the appropriate event marker and continue for
500 ft except Section 040560, which continued from the event marker for 600 ft.
2. Automatically synchronize the other eight repeats from visit 09 to the basis set.
3. Automatically synchronize the measurements from the previous visit to the current basis set.
4. Replace the basis set with a new set of synchronized measurements from the first repeat of the
current visit.
5. Repeat steps 3 and 4 until visit 03 is complete.
Visits 10 through 13 were synchronized using steps 3 through 5, but going forward in time. Since Section
040501 was out of the study after visit 03, its original basis measurement was extracted from the first
repeat measurement of visit 00.
Synchronization of Visit 00
Visit 00 could not be synchronized by comparison to later visits because it took place before major
rehabilitation was performed on most of the test sections. A basis set of measurements from visit 00
was created using the first repeat measurement. The rest of the repeats were then automatically
synchronized to it. Comparison of the profiles from Section 040501, which did not receive any
rehabilitation, with later visits verified that visit 00 was in line with the others.
Synchronization of Visit 01 and 02
Raw data were not available for visits 01 and 02. Thus, data were extracted from the public database
and automatically synchronized to profiles from visit 03. For visit 01 this was fairly successful, and most
of the measurements lined up with visit 03 within 6.6 ft. Unfortunately this was not the case for visit 02.
In visit 02, the profiles within the public database were extracted as if the sample interval was 6 inches.
However, detailed comparisons with the profiles from other visits showed that the sample interval was
actually 5.91 inches. As a result, the further along the site a section appeared, the more serious the
cumulative error in the location of the section starting point became. Thus, only the data from the first

109

two segments along the site, Sections 040504 and 040507, could be used. The sample interval of these
profiles was corrected to 5.91 inches, and their starting points were shifted slightly to maintain
consistency with other visits.
LONGITUDINAL DISTANCE MEASUREMENT
The basis set of profile measurements for visit 09, established in step 1 above, was established using the
event markers in one raw profile measurement (the first repeat). The other eight repeats from visit 09
were automatically synchronized to the basis set. When the longitudinal placement of the individual
sections within each measurement was compared to the layout within the basis set, the slope of the
linear fit ranged from 0.9994 to 1.0000. Thus, the longitudinal distance measurement for the nine profile
measurements of visit 09 was consistent within 0.06 percent—a very high level of agreement in
longitudinal distance measurement.
Figure 70 shows the disagreement in longitudinal distance measurement for each visit using the original
basis set as a reference. In the figure, a range of disagreement for each visit exists because up to nine
repeat profile measurements were made. The variation between repeat measurements within a visit
appears as the width of each bar in the figure. Since the longitudinal distance measurement was based
on the rotation of a drive wheel, the variations were most likely caused by variations in speed, lateral
wander, and tire inflation pressure (Karamihas et al. 1999). If tire inflation pressure were the dominant
cause, the disagreement would grow more positive with each successive repeat measurement as the
tire heated up because the tire‐rolling radius would increase, and the profiler would register less wheel
rotation for the same travel distance. This occurred in visits 04 through 09, but the effect was rarely
greater than 0.10 percent of the overall distance. Also, the field procedures require the operator to
warm the tire before taking measurements. Visits with very consistent longitudinal distance
measurement are attributed to proper tire warm‐up.

Visit Number
13
12
11
10
9
8
7
6
5
4
3
0
-0.2

-0.1
Di

0.0
i L

0.1
0.2
i di l Di
(

0.3
)

Figure 70. Consistency in Longitudinal Distance Measurement

110

0.4

The variation between visits in Figure 70 is caused by differences in distance measurement instrument
calibration. The longitudinal distance measured by a profiler is not true horizontal distance. It always
includes some additional component because the profiler must travel up and down the undulations in
the road. This component can be minimized by calibrating the profiler to true horizontal distance.
However, if a profiler operates on a road with grade changes and roughness that are not similar to the
site used for longitudinal distance measurement calibration, some error will exist. Tire inflation pressure
must also be close to the level that existed during calibration for consistent results.
Modest inconsistency in longitudinal distance measurement between visits is not critical as long as the
profiles of individual sections are extracted using event markers rather than longitudinal distance from
the start of each profile measurement. A high level of inconsistency, however, could interfere with
comparisons between profile features and distress surveys. Errors in profile index values, such as the IRI,
are also roughly of the same order as errors in longitudinal distance measurement (Karamihas et al.
1999). Figure 70 shows that longitudinal distance was measured with a very high level of agreement
throughout visits 03 through 13. However in visit 04, all but one of the values for disagreement in
longitudinal distance were between 0.06 and 0.13 percent. With that one value removed, Figure 70
would show an excellent level of consistency.
DATA QUALITY SCREENING
Investigators performed data quality screening to five select repeat profile measurements from each
visit of each section. Among the group of available runs, investigators selected the five measurements
that exhibited the best agreement with each other. In this case, agreement between any two profile
measurements was judged by cross‐correlating them after applying the IRI filter (Karamihas 2004). In
this method, the IRI filter is applied to the profiles, and then the output signals are compared rather
than the overall index. High correlation by this method requires that the overall roughness as well as the
details of the profile shape that affect the IRI agree. The IRI filter was applied before correlation in this
case for several reasons:






Direct correlation of unfiltered profiles places a premium on very long wavelength content, but
ignores much of the contribution of short wavelength content.
Correlation of IRI filter output emphasizes profile features in (approximate) proportion to their
effect on the overall roughness.
Correlation of IRI filter output provides a good trade‐off between emphasizing localized rough
features at distressed areas in the pavement and placing too much weight on the very short‐
duration, narrow features (spikes) that are not likely to agree between measurements because
the IRI filter amplifies short‐wavelength content, but attenuates macrotexture, megatexture,
and spikes.
A relationship has been demonstrated between the cross‐correlation level of IRI filter output
and the expected agreement in overall IRI (Karamihas 2004).

Note: This method was performed with a special provision for correcting modest longitudinal distance
measurement errors.

111

Each comparison between profiles produced a single value that summarized their level of agreement.
When nine repeat profile measurements were available, they produced 36 correlation values. Any
subgroup of five measurements could be summarized by averaging the relevant 10 correlation values.
The subgroup that produced the highest average was selected, and the other repeats were excluded
from most of the analyses discussed in the rest of this chapter. Since the number of available profiles
ranged from six to nine, the number of measurements that were excluded ranged from one to four.
Tables 62 through 72 list the selected repeats for each visit of each section and the composite
correlation level produced by them.
The process for selecting five repeat measurements from a larger group is similar to the practice within
LTPP except that it is based on composite agreement in profile rather than the overall index value. The
correlation levels listed in Tables 62 through 72 provide an appraisal of the agreement between profile
measurements for each visit of each section. When two profiles produce a correlation level above 0.82,
their IRI values are expected to agree within 10 percent most (95 percent) of the time. Above this
threshold, the agreement between profiles is usually acceptable for studying the influence of distresses
on profile. When two profiles produce a correlation level above 0.92, they are expected to agree within
5 percent most of the time. Above this threshold, the agreement between profiles is good. Correlation
above 0.92 often depends on consistent lateral tracking of the profiler and may be very difficult to
achieve on highly distressed surfaces. The IRI values provided will be the average of five observations,
which will tighten the tolerance even further.
Table 62. Selected Repeats of Section 040501
Visit
00
01
03

Repeat Numbers Composite Correlation
1 3 4 5 6
0.649
1 4 6 7 8
0.608
2 3 4 5 6
0.751

Table 63. Selected Repeats of Section 040502
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 3 4 5 6
0.837
1 2 3 4 5
0.955
2 4 5 6 9
0.977
2 3 4 6 7
0.971
1 2 3 4 7
0.939
1 4 5 6 7
0.950
1 2 4 5 6
0.921
3 4 6 8 9
0.972
1 4 5 7 8
0.967
1 4 6 7 8
0.822
1 2 3 6 7
0.936
2 3 4 5 7
0.912
3 4 5 7 9
0.948

112

Table 64. Selected Repeats of Section 040503
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 3 4 5 7
0.717
1 2 3 4 5
0.945
2 4 5 6 7
0.963
2 5 6 7 8
0.963
2 4 5 6 7
0.937
1 2 3 4 5
0.952
1 4 5 6 7
0.966
2 3 5 8 9
0.982
1 2 3 7 9
0.966
1 3 5 7 9
0.901
2 3 5 6 7
0.918
1 3 4 6 9
0.937
1 2 3 4 5
0.938

Table 65. Selected Repeats of Section 040504
Visit
00
01
02
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 4 5 6 7
0.803
1 2 3 4 5
0.928
3 4 5 6 7
0.934
2 4 7 8 9
0.958
2 3 5 7 9
0.968
1 2 3 4 5
0.963
2 3 4 6 7
0.972
1 2 3 4 5
0.983
1 4 7 8 9
0.987
1 3 4 6 8
0.985
1 2 3 4 6
0.963
1 2 3 4 5
0.957
1 3 4 5 6
0.968
1 4 5 6 8
0.976

113

Table 66. Selected Repeats of Section 040505
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 2 3 4 5
0.928
1 2 3 4 5
0.945
1 2 3 7 8
0.963
1 2 5 6 8
0.958
1 3 5 6 7
0.912
3 4 5 6 7
0.939
1 3 4 6 7
0.923
1 2 3 4 5
0.922
1 2 4 5 6
0.947
2 3 4 5 9
0.835
2 3 4 6 7
0.887
2 3 4 5 8
0.827
1 3 5 8 9
0.877

Table 67. Selected Repeats of Section 040506
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 3 4 5 6
0.833
1 2 3 4 5
0.913
2 3 4 5 6
0.965
2 3 5 7 8
0.969
1 2 3 4 5
0.935
1 3 4 5 6
0.967
1 2 3 5 7
0.969
1 4 5 6 9
0.980
1 5 7 8 9
0.970
1 2 3 6 8
0.955
3 6 7 8 9
0.966
2 3 6 7 8
0.962
1 2 3 5 8
0.977

114

Table 68. Selected Repeats of Section 040507
Visit
00
01
02
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 2 5 6 7
0.810
1 2 3 4 5
0.890
3 4 5 6 7
0.908
2 3 4 5 8
0.941
1 2 4 7 8
0.965
1 3 4 6 7
0.936
1 3 4 5 7
0.961
1 2 3 4 6
0.958
1 4 6 8 9
0.982
1 2 4 6 9
0.976
2 4 6 7 8
0.956
1 4 5 7 9
0.945
2 3 4 5 7
0.960
1 3 4 5 9
0.959

Table 69. Selected Repeats of Section 040508
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 3 5 6 7
0.803
1 2 3 4 5
0.958
2 5 7 8 9
0.963
2 3 6 7 9
0.962
2 3 4 6 7
0.931
1 3 4 5 6
0.967
1 2 3 4 5
0.972
1 2 4 5 8
0.980
3 6 7 8 9
0.978
1 3 5 6 9
0.941
1 2 3 4 8
0.948
4 6 7 8 9
0.941
3 5 6 8 9
0.964

115

Table 70. Selected Repeats of Section 040509
Visit
00
01
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 3 4 5 6
0.848
1 2 3 4 5
0.943
2 3 4 5 6
0.967
2 4 5 6 9
0.976
2 3 5 6 7
0.924
3 4 5 6 7
0.967
1 3 5 6 7
0.971
2 4 5 7 8
0.962
1 2 3 5 7
0.958
1 2 6 7 8
0.838
1 2 3 6 7
0.886
1 3 4 7 9
0.822
3 4 5 7 9
0.929

Table 71. Selected Repeats of Section 040559
Visit
00
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 2 4 5 6
0.866
1 2 4 7 8
0.966
2 4 5 7 8
0.964
2 3 4 5 6
0.963
1 3 4 5 6
0.970
1 2 3 4 5
0.978
1 3 4 7 8
0.984
1 3 6 7 8
0.976
1 2 4 5 8
0.967
3 4 7 8 9
0.973
2 3 4 5 7
0.961
2 3 5 6 9
0.973

116

Table 72. Selected Repeats of Section 040560
Visit
00
03
04
05
06
07
08
09
10
11
12
13

Repeat Numbers Composite Correlation
1 2 5 6 7
0.789
1 2 5 6 9
0.946
1 2 4 5 9
0.914
1 2 4 5 6
0.873
1 2 4 5 6
0.882
1 2 3 4 5
0.883
2 4 6 8 9
0.880
2 4 7 8 9
0.919
1 2 5 7 9
0.731
2 3 4 7 9
0.832
1 3 4 5 6
0.815
2 5 6 7 8
0.874

Overall, the majority of the groups of measurements listed in Tables 62 through 72 exhibited good to
excellent correlation, particularly in visits 03 through 09 and 11. Agreement was lowest overall for visit
00 and all visits of Section 040501. Any group of repeat measurements that produced a composite
correlation level below 0.85 was investigated using filtered plots; these results are discussed below.
In visit 00, the profile measurements showed a lack of agreement in the shape and severity of localized
distresses on many of the sections. Overall, the content within the profiles from wavelengths shorter
than about 10 ft was not repeatable. This was often most serious for the right side profile. A lack of
repeatability for short‐wavelength content is not uncommon on pavements with significant distress. The
same overall behavior was evident for visits 00, 01, and 03 of Section 040501. The correlation exhibited
for these three cases was so poor that very little credence should be placed on the analysis results for
Section 040501.
In visit 10, the left and right side profiles of Section 040502 included dips, often more than 0.2 inch
deep, throughout the entire section. In many cases, the dips did not appear consistently in all five repeat
measurements. In particular, the profiles of the last third of the section were dominated by dips that
appeared in more than one repeat, but not in all five, which suggests that the profiles were affected by
some type of surface distress that was not consistent across the lane width. The profile inconsistencies
may have been the result of small changes in lateral tracking position of the profiler.
In visit 10, profiles of Section 040505 and Section 040509 included extraneous narrow dips and spikes
that degraded their agreement. Additionally, the shape and severity of genuine narrow dips along
Section 040509 were not totally consistent between repeat measurements. In visit 12 of Section
040505, narrow dips appeared with inconsistent depth and location. Profiles from visit 12 of Section
040509 included a patch of uncorrelated short‐wavelength content 370 to 440 ft from the start of the
section on the right and 225 to 240 ft and 310 to 380 ft from the start on the left.

117

In visits 10 and 11, profiles of Section 040560 included patches of uncorrelated short‐wavelength
content, which is typically caused by pavement distress that causes aggressive transverse variations in
surface profile. Visit 12 also included dense patches of narrow dips that were not well correlated
between repeat runs on the right side.
SUMMARY ROUGHNESS VALUES
Figures 71 through 81 show the left and right IRI values for each pavement section over their monitoring
period. For most of the sections, this includes 26 summary IRI values: two per visit over 13 visits (Table
61). The figures show the IRI values versus time in years. In this case, “years” refers to the number of
years between the measurement date and the date that the site was opened to traffic (June 13, 1990).
Fractions of a year are estimated to the nearest day.
To supplement the plots, Appendix C lists the IRI, Half‐car Roughness Index (HRI), and Ride Number (RN)
of each section for each visit. These roughness values are the average of the five repeat measurements
selected in the data quality screening. These are not necessarily the same five repeat measurements
selected within the LTPP study. Appendix C also provides the standard deviation of IRI over the five
repeat measurements to help identify erratic roughness values that are the result of transverse
variations in profile caused by surface distresses.

IRI (in/mi)
300

Section 0501
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

Years
Figure 71. IRI Progression of Section 040501

118

12

14

16

IRI (in/mi)
300
250
200
150
100
Section 0502
Left
Right

50
0
-2

0

2

4

6

8

10

12

14

16

Years
Figure 72. IRI Progression of Section 040502

IRI (in/mi)
300

Section 0503
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

Years
Figure 73. IRI Progression of Section 040503

119

12

14

16

IRI (in/mi)
300

Section 0504
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

12

14

16

Years
Figure 74. IRI Progression of Section 040504

IRI (in/mi)
300

Section 0505
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

Years
Figure 75. IRI Progression of Section 040505

120

12

14

16

IRI (in/mi)
300

Section 0506
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

12

14

16

Years
Figure 76. IRI Progression of Section 040506

IRI (in/mi)
Section 0507
Left
Right

300
250
200
150
100
50
0
-2

0

2

4

6

8

10

Years
Figure 77. IRI Progression of Section 040507

121

12

14

16

IRI (in/mi)
300

Section 0508
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

12

14

16

Years
Figure 78. IRI Progression of Section 040508

IRI (in/mi)
300
250
200
150
100
Section 0509
Left
Right

50
0
-2

0

2

4

6

8

10

Years
Figure 79. IRI Progression of Section 040509

122

12

14

16

IRI (in/mi)
300

Section 0559
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

12

14

16

Years
Figure 80. IRI Progression of Section 040559

IRI (in/mi)
300

Section 0560
Left
Right

250
200
150
100
50
0
-2

0

2

4

6

8

10

Years
Figure 81. IRI Progression of Section 040560

123

12

14

16

These figures provide a snapshot of the roughness history of each pavement section. The remainder of
this chapter characterizes the profile content that made up the roughness and explains the profile
features that contributed to roughness progression.
PROFILE ANALYSIS TOOLS
Investigators used various analytical techniques to study the profile characteristics of each pavement
section and their change with time. These tools help study roughness, roughness distribution, and
roughness progression of each section, including concentrated roughness that may be linked to
pavement distress. The discussion of each analysis and plotting method is rather brief. Some examples
are provided; Sayers and Karamihas (1996) provide more details about all of these methods.
Summary Roughness Values
Left IRI, right IRI, Mean Roughness Index (MRI), HRI, and RN values were calculated. Appendix C provides
the average value of each index for each visit of each section. The discussion of roughness in this
analysis emphasizes the left and right IRI. Nevertheless, comparing the progression of HRI and RN to that
of the MRI provides additional information about the type of roughness that is changing. For example, a
low HRI value relative to MRI indicates roughness that exists on only one side of the lane. Further,
aggressive degradation of RN without a commensurate growth in MRI signifies that the developing
roughness is biased toward short‐wavelength content.
Power Spectral Density Plots
A power spectral density (PSD) plot of an elevation profile shows the distribution of its content within
each waveband. An elevation profile PSD is displayed as mean square elevation versus wave number,
which is the inverse of wavelength. A PSD plot is generated by performing a Fourier transform on a
profile. The value of the PSD in each waveband is derived from the Fourier coefficients and represents
the contribution to the overall mean square of the profile in that band.
Often, the wavebands used in a PSD plot are given a uniform spacing on a log scale. In this research,
PSDs were typically displayed using 12 bands per octave. In other words, the center of each waveband
was a factor of 21/12 larger than the waveband to its left on the plot and a factor of 21/12 smaller than
the waveband to its right. This spacing provided enough detail to search for roughness that was isolated
at a given wavelength, but enough averaging to eliminate spurious content that is common when PSDs
are displayed using a linear wave‐number scale. PSD plots were also calculated from the slope profile
rather than the elevation profile, which helped to interpret the plots because the content of a slope PSD
typically covers fewer orders of magnitude than an elevation PSD.
The PSD plots provided a very useful breakdown of the profile content. In particular, the plots revealed
(1) cases in which significant roughness is concentrated within a given waveband; (2) the type of content
that dominates the profile (e.g., long, medium, or short wavelength); (3) the effectiveness of
rehabilitation in eliminating roughness over each waveband; (4) the type of roughness that increases
with time; and (5) the type of roughness that is stable with time.

124

Figure 82 shows the PSD of Section 040502’s left profile from visit 04 and visit 11. This PSD plot includes
several noteworthy features:


The plot shows the PSD of slope rather than elevation. Thus, the vertical axis has units of
slope2/(cycles/ft) instead of elevation2/(cycles/ft).




The plot covers wave numbers from 0.01 cycle/ft to 1 cycle/ft, the range that affects IRI most.
The spectral content from about 12 to 100 ft (wave numbers between 0.01 cycle/ft and 0.08
cycle/ft) did not change significantly with time.
The spectral content for wavelengths shorter than 12 ft increased between visits. In fact, this
progression was fairly steady from visit 04 through visit 11.
In visit 11, the PSD grew with decreasing wavelength (increasing wave number) for wavelengths
below 8 ft. This should be interpreted cautiously, however, because a single anomalous reading
in the elevation profile or a single severe narrow dip would appear on a PSD plot this way.
Alternatively, it may indicate uniform growth in short wavelength roughness over the entire
length of the profile.
The peak at about 0.092 cycle/ft indicates a tremendous amount of roughness with a
wavelength of about 10.9 ft. The vertical axis is on a log scale, so the peak at this wavelength is
actually more significant than it looks. In fact, the roughness concentrated at wavelengths near
10.9 ft is responsible for more than half of the IRI of the visit 04 profile. This content was
present in the first visit after rehabilitation, which indicates that the roughness was built in
rather than a result of deterioration. An inspection of right profile PSD plots shows that this
periodic content is much more dominant on the left side than the right.






PSD of left slope (ft/cycle)
Visit 11
0.1x10

0.1x10

-3

-4

Visit 04

0.1x10

0.1x10

-5

Section 0502

-6

.01

2

5

2
.1
Wave Number (cycles/ft)

Figure 82. PSD of Section 040502 Profiles (Left Side)

125

5

1

Each of the final four observations listed above provide important information about the nature of
Section 040502’s roughness and its progression. However, the PSD provides no information about
where the roughness exists within the section. Further, if the roughness within a profile is concentrated
in a single location, the PSD plot may provide misleading information. The filtered profile plots and the
roughness profiles discussed below provide a more complete assessment of the roughness on a given
pavement.
The PSD plot provides insight into the filtering practices of the profiler that made the measurements.
Figure 83 shows the PSD of Section 040508’s left profile during visit 09 and visit 10 over the maximum
range allowed by the section length and sample interval. This plot includes several noteworthy features:






The spectral content differs for very long wavelengths (low wave numbers). This is not caused by
a change in the true profile of the section. Rather, it is the result of a change in profiler and an
associated change in the high‐pass filtering methods (Perera and Kohn 2005).
The spectral content shows a decreasing trend at very short wavelengths (high wave numbers).
This is an artifact of the low‐pass filtering applied at the time of the measurement, which is a
combination of digital filtering and height sensor footprint (Karamihas 2005).
The PSD plot for visit 09 includes a spike at a wave number of about 2.6 cycles/ft and at double
that value. This is also an artifact of the measurement process, but the source is unclear. The
spikes were present in all of the profiler measurements, which include all of the measurements
made in visit 04 through visit 09. However, the spikes did not occur at the same wave number in
each visit or in each repeat measurement within a given visit. The wave number where the left‐
most spike occurred ranged from about 2.04 cycles/ft to 2.72 cycles/ft.

PSD of left slope (ft/cycle)
0.1x10

-3

Visit 10

0.1x10

0.1x10

-4

Visit 09

-5

Section 0508
.001

.01

.1
Wave Number (cycles/ft)

1

Figure 83. PSD of Section 040508 Profiles (Left Side)
126

10

Filtered Profile Plots
A simple way to learn about the type of roughness that exists within a profile is to view the trace.
However, certain key details of the profile are often not as obvious in a raw profile trace as they may be
after the profile is filtered. For example, Figure 84 shows the raw profile trace for three visits to Section
040509 throughout its monitoring history. The plot shows that the long‐wavelength content, or the
trend, in each plot is quite consistent with time. On the other hand, narrow dips appear in the plots that
become more prevalent and severe as time progresses.
Although the raw profile plots in Figure 84 provide very useful information about the nature of the
roughness on Section 040509, a filtered plot may provide much more detail. In particular, a closer look
at the narrow dips may help study their progression. Figure 85 shows a small segment of the profile
after it has been high‐pass filtered. An anti‐smoothing moving average filter was applied with a base
length of 25 ft. The anti‐smoothing is performed by applying a smoothing filter and then subtracting it
from the original profile. Without the filter, the overall trend in the profile masks the dip, such that it is
barely visible in the trace from visit 04. When the profile is filtered, the dip and its growth with time are
much more obvious.

Left Elevation (in)
.5
Visit 11
0
-.5
.5
Visit 07
0
-.5
.5

Visit 04

0
-.5
0

100

200
300
Distance (ft)

Figure 84. Raw Profiles of Section 040509

127

400

500

Left Elevation (in)
.2

Visit 04

0
-.2

Visit 07

-.4

Visit 11

-.6
170

172

174
176
Distance (ft)

178

180

Figure 85. Filtered Profiles of Section 040509

In addition to a closer view of short‐duration features, filtered plots help provide a clearer view of longer
trends in profile. Figure 86 provides one such example. The figure shows two profile measurements of
Section 040560 after they have been smoothed with a base length of 25 ft and anti‐smoothed with a
base length of 125 ft. One trace was collected before rehabilitation, and the other was collected several
years later. On this section, the longer wavelength features displayed in the plot were not altered very
much by the rehabilitation. This was not the case on every test section. On the other hand, the content
within the profile in the wavelength range shorter than 25 ft was altered completely.

Left Elevation (in)
.2

0

-.2

Visit 00
Visit 03

-.4
0

100

200
300
Distance (ft)

400

Figure 86. Long‐Wavelength Profiles of Section 040560

128

500

Three types of filtered plots were inspected for every visit of every section:




Long wavelength: A profile smoothed with a base length of 25 ft and anti‐smoothed with a base
length of 125 ft.
Medium wavelength: A profile smoothed with a base length of 5 ft and anti‐smoothed with a
base length of 25 ft.
Short wavelength: A profile smoothed with a base length of 1 ft and anti‐smoothed with a base
length of 5 ft.

These filters were used to screen the profiles for changes with time and special features of interest. The
terms “long,” “medium,” and “short” are relative, and in this case pertain to the relevant portions of the
waveband that affects the IRI. The long wavelength portion of the profile was typically very stable with
time. However, the long wavelength profile plots of every section changed somewhat between visit 09
and visit 10—not by a change in the surface characteristics of the section, but by a change in profiler
make and the associated change in filtering practices.
The medium‐wavelength plots provided a view of the features in a profile that were likely to have a
strong effect on the IRI and may change with time. The short‐wavelength elevation plots also typically
progressed with time, but only affected the IRI through localized roughness or major changes in content.
However, the short‐wavelength elevation plots helped identify and track the progression of narrow dips
and other short‐duration features that may have been linked to distress.
Roughness Profiles
A roughness profile provides a continuous report of road roughness using a short segment length.
Instead of summarizing the roughness by providing the IRI for an entire pavement section, the
roughness profile shows the details of how IRI varies with distance along the section by using a sliding
window to display the IRI of every possible segment of given base length along the pavement (Sayers
1990).
A roughness profile displays the spatial distribution of roughness within a profile. As such, it can be used
to distinguish road sections with uniform roughness from sections with roughness levels that change
over their length. Further, the roughness profile can pinpoint locations with concentrated roughness and
estimate the contribution of a given road disturbance to the overall IRI.
Figure 87 shows an example of a roughness profile for visit 11 of Section 040503. The roughness profile
was generated using a base length of 25 ft, that is, every point in the plot shows the IRI of a 25‐ft
segment of road, starting 12.5 ft upstream and ending 12.5 ft downstream. The plot shows that the first
100 ft and the last 150 ft of the section are very smooth. On the other hand, the area from 100 to 350 ft
from the start of the section is substantially rougher.

129

Left Roughness Profile (in/mi)
250
200
150
100
50
0
0

100

200
300
Distance (ft)

400

500

Figure 87. Roughness Profile of Section 040503 (25‐ft Base Length)

Figure 88 shows how a roughness profile can help find localized roughness and quantify its impact on
the overall roughness of a section. The figure shows the roughness profile of Section 040509 using a 10‐
ft base length for visits 04, 07, and 11. With a 10‐ft base length, isolated roughness is easy to identify.
For example, the dips that appear at 175.5 ft increase in roughness significantly with time. In visit 11, the
peak value of the roughness profile in the vicinity of the dip is 546 inches/mi. Since that value represents
the roughness over just 1/50 of the segment, it suggests that the single dip contributes more than 10
inches/mi to the overall IRI of the section.

Left Roughness Profile (in/mi)
600

Visit 11

500

Visit 07
Visit 04

400
300
200
100
0
0

50

100
150
Distance (ft)

200

Figure 88. Roughness Profiles of Section 040509 (10‐ft Base Length)
130

250

DISTRESS SURVEYS AND MAINTENANCE RECORDS
Once the analysis and plotting were completed, all of the observations were compared to the MDS
performed on each section. MDS results were available for each section starting in 1994 and covering a
visit nearly every year for the rest of the monitoring history. This provided a means of relating profile
features to known distresses. For the SPS‐5 project, two observations were common: First, dips that
grew progressively rough with time were often found in the vicinity of transverse cracks. This was the
case for the dip shown in Figure 85 and the locations of peak roughness in Figure 88. Often, the first
appearance of peaks within a very short interval roughness profile corresponded to the year when
transverse cracks were first observed in the distress survey. The presence of the dips could typically be
verified using short‐wavelength elevation plots. Second, areas where cracks appeared with a very high
density within a WP sometimes caused areas of isolated roughness to appear within a profile as well as
areas in which the short‐wavelength content within the same area was not well correlated between
repeat runs.
Investigators also compared changes in profile properties to maintenance records. Crack sealing was
performed on all of the sections except 040501 and 040507 in May 2002. The entire test site received a
fog seal coat on May 28, 1998, and April 16, 2003. Sections 040502, 040505, 040506, 040509, 040559,
and 040560 also received a fog seal on August 23, 2001.
DETAILED OBSERVATIONS
Appendix D reports key observations from the roughness index progression, PSD plots, filtered profile
plots, roughness profiles, and distress surveys. In many cases, similar behavior was noted for multiple
sections. Those observations are repeated under every section heading where it is appropriate.
However, Appendix D does not discuss changes in profile properties with time caused by changes in the
profiler make.
PROFILE ANALYSIS KEY FINDINGS
This section summarizes the important profile properties and the roughness progression of each section
within the SPS‐5 site. Several observations within this analysis were common to more than one
pavement section, as described below. This section of the analysis, in conjunction with the roughness
progression plots (Figures 71 through 81), provides the essential information about each pavement
section.
Before rehabilitation, all 11 sections included narrow dips, typically 0.5 to 0.40 inch deep and up to 2 ft
wide. The dips were usually more severe on the left side than on the right. Rehabilitation completely
removed the dips within every section except for Section 040501, which was not rehabilitated.
In many of the sections, some aspects of the long wavelength roughness survived rehabilitation.
Sections 040505, 040559, and 040560 had profiles after rehabilitation with very long wavelength
content that was very similar to the content before rehabilitation. Sections 040502, 040503, and 040509
exhibited some similarities between the long‐wavelength content before and after rehabilitation. After

131

rehabilitation, the content within the profiles with wavelengths greater than 30 ft rarely changed over
the entire monitoring history of the site.
The change in profiler make in late 2002 affected the long‐wavelength content of the profiles on every
test section because the newer profiler used a high‐pass filter that eliminated a little more of the profile
content than the previous two devices. This had no probable effect on the measurement of localized
roughness or the study of narrow bumps and dips caused by cracking and other distress. However, it did
confound the study of changes in the long‐wavelength content within the profiles between visits 09 and
10.
One other minor device effect within the profiles was peaks in the PSD plots with no pavement‐related
explanation. In visits 04 through 09 (measured by the K.J. Law T‐6600) most PSD plots from the left side
included a strong peak at a wavelength somewhere between 0.37 and 0.49 ft and another at a
wavelength of double the first.
Sections 040502 and 040509 exhibited the most dramatic increase in IRI over their post‐rehabilitation
monitoring history. They both grew in roughness at an increasing rate and were both very rough by the
end of the monitoring period. Both sections included transverse cracks that became more severe with
time. Concentrated roughness appeared at many of the cracks within a few years of their detection by
MDS measurements. The roughness appeared as narrow dips that grew in severity with time. Note that
the dips were much wider than a typical crack, often 1 to 2 ft wide. Thus, some genuine depressions in
the pavement were constantly developing around the cracks. Some of the dips grew to as much as
0.75 inches deep.
Sections 040503, 040505, and 040560 also exhibited a large change in roughness with time. They grew
in roughness at an increasing rate, and their MRI changed by 67 inches/mi, 45 inches/mi, and
70 inches/mi, respectively, throughout their post‐rehabilitation monitoring history. Section 040505
developed roughness because dips of increasing severity appeared near transverse cracks. Section
040505 developed roughness that was very similar to Sections 040502 and 040509. It had fewer rough
transverse cracks, but included roughness at densely cracked areas within the WPs.
Sections 040503 and 040560 also included dips at transverse cracks. However, large, densely cracked
areas, not necessarily in the transverse direction, caused a significant portion of the roughness
development within the WPs. The hit‐or‐miss nature of their placement relative to the profiler path
caused inconsistencies in the shape of rough features between repeat measurements and visits.
Nevertheless, the cracks caused a consistent growth in roughness in the affected areas.
Sections 040504, 040506, 040508, and 040559 increased in roughness at a steady rate after
rehabilitation until 2004. The MRI of these sections increased by no more than 15 inches/mi during their
post‐rehabilitation monitoring history through February 2004. These sections, particularly Section
040508, included dips near transverse cracks in earlier visits. However, the dips were usually not very
severe until the final two profiling visits in late 2004 and 2006. In Sections 040506, 040508, and 040559,
the roughness increased more rapidly in the last two years than during the previous 14 years because of

132

narrow dips near recorded transverse cracks. Section 040507 showed little roughness linked to
transverse cracking.
Sections 040502, 040504, and 040507 showed little, if any, improvement in IRI on the left side after
rehabilitation. In addition, the roughness of the left side after rehabilitation was much higher than the
right side, and was caused by a continuous (sinusoidal) series of bumps and dips with peaks 8 to 13 ft
apart and a peak‐to‐trough difference in elevation of up to 0.2 inch. These conditions, present on the
right side and a dominant part of the left side, may have been caused by problems with the rolling
process, but that could not be verified.
In May 2002 crack sealing was performed on all of the sections except Sections 040501 and 040507.
Very little evidence was found that suggested this directly affected the roughness. Of course, crack
sealing very well could have decelerated the deterioration of these sections. The entire test site received
a fog seal coat on May 28, 1998, and April 16, 2003. Seven of the test sections (excluding Sections
040507, 040504, 040503, and 040508, the first four sections along the length of the site) at the site also
received a fog seal on August 23, 2001. Fog sealing did not cause an immediate change on the IRI.
Sections 040501, 040502, and 040560 had HRI values that were 20 percent or more below the MRI
values. In Section 040502, the HRI grew increasingly small with time compared to the MRI. This
difference was larger than that observed on most other sections and was caused by the presence of
profile features that are not consistent across the lane. Typically, this also signifies the presence of
localized distress.
Table 73 summarizes the roughness behavior of each section within the SPS‐5 site.

133

Table 73. Summ
mary of Roughnesss Behavior
Section

040502 040
0503 040504 04
40505 040506 0040507 040508 040509 040559 040560

MRI change affter rehabilitation (inch/mi)
(

‐44

‐5
51

‐24

‐
‐85

‐48

‐26

‐44

‐91

‐59

‐75

MRI change since rehabilitation (inch/mi)

160

67
6

21

45

38

11

25

183

24

70

26

7

7

6

7

9

3

18

4

5

MRI change ovver 7 years after re
ehabilitation (inch//mi)
MRI growth att an increasing rate
e
MRI growth att a steady rate
HRI about 20%
% below MRI
Left IRI much higher than right IR
RI
Dominant periodic content, 8‐13
3 ft
Very long feattures preserved aftter rehabilitation
Severe dips ne
ear transverse craccks
Patches or rou
ughness near dense
e cracking
— Yes
hat
— Somewh

134

In light of the rehabilitation performed on each section, the information from Table 73 suggests the
following:








The two test sections with a 2‐inch recycled overlay (Sections 040502 and 040509) exhibited the
largest post‐rehabilitation increase in MRI over the monitoring history by a wide margin (160
inches/mi and 183 inches/mi, respectively).
The two test sections with a 5‐inch virgin overlay (Sections 040507 and 040504) exhibited the
smallest post‐rehabilitation increase in MRI over the monitoring history (11 inches/mi and 21
inches/mi, respectively).
All of the test sections except those with a 2‐inch recycled overlay (Sections 040502 and
040509) increased in MRI by less than 10 inches/mi over the first seven years after
rehabilitation.
Roughness increased the most in test sections with narrow dips at transverse cracks.

135

136

CHAPTER 5. CONCLUSIONS AND RECOMMENDATIONS
ADOT initiated this project to study the relative performance of the various SPS‐5 design alternatives
(including supplemental sections), which can inform future rehabilitation design decisions. Deflection,
surface distress, and profile data were used as the basis for performance evaluation and were analyzed
as part of the study.
The SPS‐5 project offers a unique opportunity to directly compare the performance of various
rehabilitated pavement structures while reducing the confounding effect of other variables such as
traffic‐loading, climate, and subgrade conditions. However, the findings drawn from this evaluation
must be considered carefully. The experimental design did not offer replicate treatments to verify
findings. The conclusions are based on one set of in situ conditions; observations from other climate or
loading scenarios may differ from those noted within this report. Therefore, findings reported may be
unique to the conditions and construction of this site.
Despite these issues, the data captured at the project provides valuable insight into pavement design
and performance. Following is a summary of lessons learned from the performance data collected at the
SPS‐5 site.
DEFLECTION ANALYSIS





All sections in the SPS‐5 project had extremely stiff pavements.
Dynamic effects from poor damping resulted in errors in the elastic layer modeling of
pavements. AASHTO 1993 analysis provided much less variable and more reasonable results.
Calculated SNeff for Section 040507 and Section 040504 were larger than expected even though
dynamic effects were expected to cause a slight underprediction of Mr.
Preconstruction SNeff varied significantly among the sections, which made the analyses that
compared the performance of rehabilitation methods questionable.
o 17 months after rehabilitation:
 Sections with virgin AC overlays had a 31 percent greater increase in SNeff than sections
with recycled AC overlays.
 Sections with thick (5‐inch) overlays had a 24 percent greater increase in SNeff than
sections with thin (2‐inch) AC overlays.
 Sections with intensive surface preparation had a 17 percent greater increase in SNeff
than sections with minimal surface preparation AC overlays.
o 163 months after rehabilitation:
 Sections with thick overlays had a 77 percent greater increase in SNeff than sections with
thin overlays.
 Sections with virgin AC overlays had a 55 percent greater increase in SNeff than sections
with recycled AC overlays.
 Sections with intensive surface preparation had a 39 percent greater increase in SNeff
than sections with minimal surface preparation.

137

DISTRESS ANALYSIS
















Section 040507 had the best overall performance; Section 040502 had the worst overall
performance.
Sections with intensive surface preparation performed better than sections with minimal
surface preparation.
Sections with thick (5‐inch) overlays performed better than sections with thin (2‐inch) overlays.
Sections with virgin AC overlays performed better than sections with minimal recycled AC
overlays.
Sections that utilized two of the following rehabilitation features performed better than sections
that utilized only one of the following: thick overlay, virgin AC overlay, and intensive
preparation.
Sections that performed the best against structural distresses all had thick (5‐inch) overlays.
Overlay thickness had a greater effect than overlay material or surface preparation on the
pavement’s performance against structural distresses.
Sections that performed the best against environmental distresses all had virgin AC overlays.
Overlay material had a greater effect than overlay thickness or surface preparation on the
pavement’s performance against environmental distresses.
Sections with greater total asphalt (existing and new) layer thickness showed greater resistance
to fatigue cracking.
Section 040559, the inverted section, performed well in structural distress and poorly in
environmental distress, which is consistent with trends among the core sections.
Section 040560, the ARAC overlay, performed worse in structural distresses than all sections
except the two sections with thin overlays and minimal surface preparation (Section 040502 and
Section 040505). Section 040559 performed worse in environmental distress than core sections
with virgin AC overlays, but better than sections with recycled AC overlays.
Rutting in all sections was considered minimal and not a critical distress.

PROFILE ANALYSIS








Rehabilitation completely removed narrow dips in every test section with the exception of the
control section (Section 040501), which was not rehabilitated.
Some aspects of long wavelength roughness survived the rehabilitation in many sections.
Changing the profiler make in late 2002 confounded the study of changes in the long‐
wavelength content within the profiles between visits 09 and 10 on every test section.
Section 040507 and Section 040504, the sections with 5‐inch virgin AC overlays, exhibited the
smallest post‐rehabilitation increase in MRI.
Section 040502 and Section 040509, the sections with 2‐inch recycled AC overlays, exhibited the
largest post‐rehabilitation increase in MRI.
Sections 040503, 040505, and 040560 also exhibited a large change in roughness over time.
Sections 040506, 040508, and 040559 showed a steady rate of increase in roughness over the
first 14 years of the study and a more rapid increase in the last few years.

138





Sections 040501, 040502, and 040560 had HRI values that were 20 percent or more below the
MRI values, signifying the presence of localized distress.
MRI increased by less than 10 inches/mi over the first seven years after rehabilitation in all test
sections except Sections 040502 and 040509, the sections with 2‐inch recycled AC overlays.
Roughness increased the most in test sections with narrow dips at transverse cracks.

The following recommendations are based on these findings:







An evaluation using current recycled asphalt pavement (RAP) technologies may indicate
different trends regarding RAP performance, since RAP technologies, practices, and materials
have continued evolving since the time of this study.
Pavement rehabilitation under similar in situ conditions (i.e., a dry/no‐freeze climate,
approximately 15 inches of granular base and 4 inches of existing AC with significant fatigue,
block cracking, and traffic loading) requires at least 5 inches of new virgin AC material to prevent
pavement cracking in the first seven years after rehabilitation or at least 5 inches of new
recycled AC material to prevent cracking in the first four years after rehabilitation.
Using pavement condition indicators in addition to IRI typically provides a better assessment of
pavement condition for network‐level decision‐making.
The effect of subgrade modulus increasing subsequent to overlay (as seen in this data set and
supported theoretically) should be considered as part of pavement design.

139

140

REFERENCES
Al‐Qadi, I. L., H. Wang, P. J. Yoo, and Samer H. Dessouky. 2008. “Dynamic Analysis and In Situ Validation
of Perpetual Pavement Response to Vehicular Loading.” Transportation Research Record:
Journal of the Transportation Research Board 2087: 29–39.
American Association of State Highway and Transportation Officials (AASHTO). 1993. AASHTO Guide for
Design of Pavement Structures. Washington, D.C.: American Association of State Highway and
Transportation Officials.
Evans, L. D. and A. Eltahan. 2000. LTPP Profile Variability. Publication FHWA‐RD‐00‐113. McLean, VA:
Federal Highway Administration.
Hossain, M., Douglas J. Lattin, and Larry A. Scofield. 1996. SPS‐5: Rehabilitation of Asphalt Concrete
Pavements. Construction Report. Phoenix: Arizona Department of Transportation. Unpublished
report.
Huang, Y. H. 1993. Pavement Analysis and Design. Englewood Cliffs, NJ: Prentice Hall.
Karamihas, S. M. 2004. “Development of Cross Correlation for Objective Comparison of Profiles.”
International Journal of Vehicle Design 36(2/3): 173‐193.
Karamihas, S. M. 2005. Critical Profiler Accuracy Requirements. Publication UMTRI‐2005‐24. Ann Arbor:
University of Michigan Transportation Research Institute.
Karamihas, S. M., T. D. Gillespie, S. D. Kohn, and R. W. Perera. 1999. Guidelines for Longitudinal
Pavement Profile Measurement. NCHRP Report 434. Washington, D.C.: National Cooperative
Highway Research Program.
Lukanen, E. O., R. Stubstad, and R. Briggs. 2000. Temperature Predictions and Adjustment Factors for
Asphalt Pavement. Publication FHWA‐RD‐98‐085. McLean, VA: Federal Highway Administration.
Miller, J. S. and W. Y. Bellinger. 2003. Distress Identification Manual for the Long‐Term Pavement
Performance Program. Revised fourth edition. Publication FHWA‐RD‐03‐031. Washington, D.C.:
Federal Highway Administration.
National Cooperative Highway Research Program (NCHRP). 2004. Guide for Mechanistic‐Empirical
Design of New and Rehabilitated Pavement Structures. Washington, D.C.: National Cooperative
Highway Research Program.
Nichols Consulting Engineers. 2010. Long‐Term Pavement Performance (LTPP) Forensic Evaluation: SPS‐5
Project (040500): Casa Grande, Arizona. Unpublished report.
Perera, R. W., and S. D. Kohn. 2005. Quantification of Smoothness Index Differences Related to Long‐
Term Pavement Performance Equipment Type. Publication FHWA‐HRT‐05‐054. Washington, D.C.:
Federal Highway Administration.

141

Rada, G. R., C. L. Wu, R. K. Bhandari, A. R. Shekharan, G. E. Elkins, and J. S. Miller. 1999. Study of LTPP
Distress Data Variability. Publications FHWA‐RD‐99‐074 and FHWA‐RD‐99‐075. Washington,
D.C.: Federal Highway Administration.
Rohde, G. T., and T. Scullion. 1990. MODULUS 4.0: Expansion and Validation of the MODULUS
Backcalculation System. Research Report No. 1123‐3. Austin: Texas State Department of
Highways & Public Transportation.
Sayers, M. W. and S. M. Karamihas. 1996. Interpretation of Road Roughness Profile Data. Publication
FHWA‐RD‐00‐101. McLean, VA: Federal Highway Administration.
Sayers, M. W. 1990. “Profiles of Roughness.” Transportation Research Record: Journal of the
Transportation Research Board 1260: 106‐111.
Simpson, A. L. 2001. Characterization of Transverse Profiles. Publication FHWA‐RD‐01‐024. McLean, VA:
Federal Highway Administration.
Uhlmeyer, J. S., K. Willoughby, L. M. Pierce, and J. P. Mahoney. 2000. “Top‐Down Cracking in
Washington State Asphalt Concrete Wearing Courses.” Transportation Research Record: Journal
of the Transportation Research Board 1730: 110–116.

142

APPENDIX A: CONSTRUCTION DEVIATIONS
040501 (Control Section):


No deviations noted.

040502 (Minimum Restoration):


There were some problems milling this section. The first passes didn’t completely
remove the friction course. The remaining material was stripped badly. The section was
re‐milled. Difficulty was also encountered working a 6‐ft milling machine rather than
one that could span the entire 12‐ft lane.

040503 (Minimum Restoration):


No deviations noted.

040504 (Minimum Restoration):


No deviations noted.

040505 (Minimum Restoration):


Paving began at 6:15 p.m. and ended at 7:40 p.m. However, at 6:40 p.m., the paver had
reached station (local) 4+10 and stopped. There was not enough material to finish the
section, and no trucks were waiting. The material quantity had been miscalculated. The
hot plant had shut down and was being recalibrated for friction course for the next day.
Contractor was warned to complete the section or be forced to remove what was done
up to now and redo the whole section with new material later when the whole section
could be done at once since no transverse joints were allowed (inside the SHRP section).
A final load was delivered at 7:30 p.m. and paving continued to the end of the section.
Paver stops were frequent. They were at 0+26, 0+52, 0+91, 3+32, 3+48, and 4+10 at
6:40 p.m.; 4+15 at 7:20 p.m.; 4+20 at 7:30 p.m.; and 4+98.

040506 (Intensive Restoration):


Some difficulties were experienced (second lift – layer 4). The first 50 ft were paved
before it was determined that the material was too shallow. Contractor removed
material behind paver in WP and backed up to beginning. The first 50 ft were repaved
and again, it was too shallow. The same removal was done and the first 50 ft were
paved for the third time. This time the material was deep enough and paving continued.
Stops were recorded at 0+15, 0+42, 4+02, local stations. Paver stops were frequent due
to slow delivery of material. Paver stops made 0.5‐inch to 1‐inch transverse bumps that
were not visible after rolling.

143

040507 (Intensive Restoration):


No deviations noted.

040508 (Intensive Restoration):


No deviations noted.

040509 (Intensive Restoration):


No deviations noted.

040559 (Intensive Restoration):


No deviations noted.

040560 (Minimum Restoration, ARAC):


No deviations noted.

144

APPENDIX B: SITE WORK HISTORY
After original construction in April through June 1990, the following maintenance activities were
performed:
040501 (Control Section):


10/01/96: Taken out of study.

040502 (Minimum Restoration):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040503 (Minimum Restoration):


05/28/98: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040504 (Minimum Restoration):


05/28/98: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040505 (Minimum Restoration):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).

145



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040506 (Intensive Restoration):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



08/01/07: Patch pot holes – hand spread, compacted with truck (number of holes).



01/01/09: Taken out of study.

040507 (Intensive Restoration):


05/28/98: Fog seal coat (yd2).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040508 (Intensive Restoration):


05/28/98: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040509 (Intensive Restoration):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

146

040559 (Intensive Restoration, Inverted Design):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

040560 (Minimum Restoration, ARAC):


05/28/98: Fog seal coat (yd2).



08/23/01: Fog seal coat (yd2).



05/01/02: Crack sealing (linear feet).



04/16/03: Fog seal coat (yd2).



01/01/09: Taken out of study.

147

APPENDIX C: ROUGHNESS VALUES
This appendix lists the left International Roughness Index (IRI), right IRI, mean roughness index (MRI),
Half‐car Roughness Index (HRI), and Ride Number (RN) values for each visit to each Specific Pavement
Studies (SPS‐5) project section. The roughness values are the average for five repeat runs. The five runs
were selected from a group of as many as nine by automated comparison of profiles, as described in the
report. Values of standard deviation are also provided for left and right IRI to reveal cases of high
variability among the five measurements. However, the screening procedure used to select five repeats
usually helped reduce the level of scatter.
The discussion of roughness in the report emphasizes the left and right IRI. Nevertheless, the other
indexes do provide useful additional information. MRI is simply the average of the left and right IRI
value. HRI is calculated by converting the IRI filter into a half‐car model (Sayers 1989). This is done by
collapsing the left and right profile into a single profile in which each point is the average of the
corresponding left and right elevation. The IRI filter is then applied to the resulting signal. The HRI is very
similar to the IRI, except that side‐to‐side deviations in profile are eliminated. The result is that the HRI
value for a pair of profiles will always be lower than the corresponding MRI value. Comparing the HRI
and MRI value provides a crude indication of the significance of roll (i.e., side‐by‐side variation in profile)
to the overall roughness. When HRI is low compared to MRI, roll is significant. This is common among
asphalt pavements (Karamihas et al. 1995). Certain types of pavement distress, such as longitudinal
cracking, may also cause significant differences between HRI and MRI.
Figure C‐1 compares the HRI to MRI for all of the profile measurements that are discussed in this
appendix, including 670 pairs of roughness values. The figure shows a best fit line with a zero intercept
and a line of equality. The slope of the line is 0.830. A typical value for asphalt pavement is about 0.85.
RN has shown a closer relationship to road user opinion than the other indexes (Sayers and Karamihas
1996). As such, it may help distinguish the segments from each other by ride quality. Further, the effect
on RN may help quantify the impact of that distress on ride when the roughness of a section is
dominated by a particular type of distress. In particular, a very low RN value coupled with moderate IRI
values indicates a high level of short‐wavelength roughness and potential sensitivity to narrow dips and
noise within the profile caused by coarse surface texture.
Table C‐1 provides the roughness values, including the date of each measurement and the time (in
years) since the site was opened to traffic. Negative values indicate measurements that were made
before rehabilitation.

148

HRI (in/mi)
300
Line of Equality
250
200

HRI = 0.830•MRI

150
100
50
0
0

50

100 150 200
MRI (in/mi)

250

300

Figure C‐1. Comparison of HRI to MRI

Table C‐1. Roughness Values
Section

Date

Years

040501
040501
040501
040502
040502
040502
040502
040502
040502
040502
040502
040502
040502
040502
040502
040502
040503
040503
040503
040503
040503
040503
040503
040503

05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00

‐0.35
0.27
2.70
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47

Left IRI (inch/mi) Right IRI (inch/mi)
MRI
HRI
RN
Avg
Std Dev
Avg
Std Dev (inch/mi) (inch/mi)
55
7.1
105
10.6
80
64
3.09
63
12.1
100
3.8
82
64
3.01
60
4.8
118
6.6
89
71
2.77
125
6.5
140
5.8
132
106
2.41
114
2.6
61
1.3
88
78
3.90
114
1.5
66
0.7
90
79
3.80
143
2.0
85
0.6
114
96
3.07
160
4.0
93
1.4
126
105
2.81
150
3.1
119
1.1
134
113
2.59
178
6.7
137
1.7
157
127
2.17
175
2.2
148
1.4
161
130
2.09
176
2.9
179
2.2
178
139
1.85
205
7.8
177
6.7
191
149
1.68
182
4.3
205
1.4
194
150
1.55
164
4.2
204
4.5
184
145
1.62
232
7.6
264
4.5
248
186
1.02
102
4.7
121
7.4
112
84
2.70
67
2.5
54
0.5
61
53
4.16
68
0.6
54
0.2
61
53
4.07
75
0.7
61
0.6
68
56
3.97
78
0.8
63
0.5
71
59
3.82
74
1.2
68
1.2
71
57
3.73
76
0.4
76
1.1
76
61
3.56
81
0.5
85
0.2
83
68
3.32

149

Table C‐1. Roughness Values (Continued)
Section

Date

Years

040503
040503
040503
040503
040503
040504
040504
040504
040504
040504
040504
040504
040504
040504
040504
040504
040504
040504
040504
040505
040505
040505
040505
040505
040505
040505
040505
040505
040505
040505
040505
040505
040506
040506
040506
040506
040506
040506
040506
040506
040506
040506
040506
040506
040506
040507
040507
040507

15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
15‐Jan‐92
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
15‐Jan‐92

11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
1.59
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
1.59

Left IRI (inch/mi) Right IRI (inch/mi)
MRI
HRI
RN
Avg
Std Dev
Avg
Std Dev (inch/mi) (inch/mi)
86
1.5
107
1.2
96
80
2.94
101
1.0
90
2.6
95
77
2.97
104
2.1
111
2.1
108
90
2.76
87
2.3
138
2.9
112
94
2.61
114
4.2
141
2.2
128
107
2.35
87
6.5
115
6.8
101
84
2.90
94
1.9
60
2.4
77
69
4.04
100
2.7
65
1.3
83
73
3.98
95
1.7
62
0.9
79
70
3.97
96
0.5
68
0.8
82
70
3.90
99
0.9
68
0.8
84
71
3.82
98
0.7
68
0.8
83
71
3.85
99
0.5
70
0.4
84
72
3.88
100
0.6
71
0.3
86
74
3.87
97
0.5
72
0.5
85
73
3.86
105
1.1
73
0.6
89
76
3.67
103
1.9
76
0.8
89
77
3.65
101
0.5
75
1.1
88
79
3.55
113
0.5
83
0.8
98
85
3.22
144
8.3
190
4.3
167
138
2.26
79
1.1
84
0.8
82
71
3.95
80
1.7
88
0.7
84
73
3.86
82
1.2
90
0.9
86
71
3.71
88
2.3
89
1.0
88
73
3.58
89
1.5
91
0.3
90
74
3.43
90
2.3
100
0.9
95
80
3.24
95
2.2
103
0.9
99
82
3.13
104
1.6
114
2.1
109
89
2.88
104
4.0
98
3.4
101
80
2.86
112
1.3
123
4.5
117
94
2.69
119
4.2
120
4.4
119
98
2.47
124
2.6
130
4.8
127
97
2.33
104
8.5
121
5.4
113
91
2.69
71
2.0
59
1.8
65
58
4.09
73
1.1
63
0.5
68
59
4.05
74
0.9
67
0.6
70
58
3.99
77
0.7
67
0.5
72
59
3.89
74
0.5
64
1.0
69
58
3.95
77
0.5
69
0.4
73
60
3.96
76
0.3
69
0.6
72
60
3.97
76
0.9
68
0.6
72
60
3.92
80
0.8
69
0.7
74
61
3.83
77
0.5
74
0.5
76
61
3.54
82
0.6
89
0.6
85
71
2.81
102
1.9
103
0.5
103
88
2.50
111
8.1
127
4.8
119
97
2.58
103
5.6
63
4.6
83
74
4.01
112
5.1
69
2.4
91
79
3.91

150

Table C‐1. Roughness Values (Continued)
Section

Date

Years

040507
040507
040507
040507
040507
040507
040507
040507
040507
040507
040507
040508
040508
040508
040508
040508
040508
040508
040508
040508
040508
040508
040508
040508
040509
040509
040509
040509
040509
040509
040509
040509
040509
040509
040509
040509
040509
040559
040559
040559
040559
040559
040559
040559
040559
040559
040559
040559

22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
21‐Sep‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06
05‐Feb‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04

2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
0.27
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78
‐0.35
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51

Left IRI (inch/mi) Right IRI (inch/mi)
MRI
HRI
RN
Avg
Std Dev
Avg
Std Dev (inch/mi) (inch/mi)
107
1.8
66
2.3
86
76
3.93
112
1.5
69
0.2
90
77
3.84
115
3.8
69
1.9
92
78
3.78
111
1.5
67
0.9
89
77
3.83
111
1.7
68
1.3
90
77
3.85
114
0.6
70
0.2
92
80
3.82
110
0.5
68
0.8
89
77
3.85
122
1.8
76
0.9
99
84
3.64
119
1.7
71
1.3
95
81
3.69
104
1.1
63
1.1
84
75
3.81
113
1.4
71
0.7
92
80
3.66
93
5.0
113
3.9
103
83
2.72
64
0.5
54
0.8
59
54
4.27
65
0.6
55
0.5
60
54
4.21
65
0.4
58
1.0
62
54
4.14
67
1.0
58
0.7
62
54
3.98
65
0.3
59
0.4
62
55
4.05
68
0.2
60
0.4
64
56
4.11
71
0.2
62
0.2
66
59
4.04
72
0.5
63
0.5
67
59
3.89
76
0.9
63
0.4
70
60
3.73
78
1.2
69
0.6
74
64
3.57
78
1.1
68
0.6
73
64
3.55
90
1.1
77
0.9
84
71
3.25
171
5.1
141
5.5
156
130
2.18
69
0.5
61
1.7
65
59
4.15
72
0.9
65
0.6
68
61
4.06
79
0.4
74
0.5
77
68
3.72
87
1.0
78
1.4
83
73
3.45
98
1.4
100
1.5
99
89
2.88
123
1.2
108
0.4
115
102
2.49
141
1.4
127
2.6
134
120
2.20
173
1.5
166
5.0
169
147
1.78
188
11.1
141
12.7
164
141
1.94
203
7.6
200
12.2
202
175
1.62
168
18.5
246
15.2
207
177
1.59
249
3.1
246
15.7
248
209
1.28
146
13.4
121
2.3
134
106
2.61
71
0.8
79
0.4
75
68
3.94
71
1.2
81
0.3
76
67
3.91
77
1.4
81
1.1
79
70
3.81
76
0.7
79
0.9
78
69
3.86
75
0.8
81
0.5
78
69
3.88
76
0.3
83
0.5
79
71
3.83
80
0.7
83
0.4
82
72
3.67
77
0.7
87
1.2
82
71
3.57
84
0.7
83
0.9
83
71
3.47
87
1.0
86
0.7
86
74
3.37

151

Table C‐1. Roughness Values (Continued)
Section

Date

Years

040559
040560
040560
040560
040560
040560
040560
040560
040560
040560
040560
040560
040560

24‐Mar‐06
05‐Feb‐90
22‐Feb‐93
03‐Feb‐97
09‐Dec‐97
11‐Dec‐98
11‐Nov‐99
01‐Dec‐00
15‐Nov‐01
04‐Nov‐02
06‐Feb‐04
14‐Dec‐04
24‐Mar‐06

15.78
‐0.35
2.70
6.64
7.49
8.50
9.41
10.47
11.43
12.39
13.65
14.51
15.78

Left IRI (inch/mi) Right IRI (inch/mi)
MRI
HRI
RN
Avg
Std Dev
Avg
Std Dev (inch/mi) (inch/mi)
103
1.0
96
1.0
99
84
2.88
113
22.9
146
11.9
130
106
2.70
60
0.9
50
0.6
55
44
4.14
61
0.5
56
1.8
58
47
3.99
63
0.7
57
5.1
60
49
3.84
62
0.9
75
2.0
68
54
3.65
68
0.3
74
3.7
71
52
3.54
69
0.9
83
2.2
76
58
3.40
74
0.8
102
1.7
88
66
3.08
84
2.1
89
5.9
86
64
2.96
89
0.3
97
3.1
93
67
2.80
69
1.5
116
8.9
93
73
3.07
99
2.9
152
3.9
125
92
2.17

REFERENCES
Karamihas, S. M., T. D. Gillespie, and S. M. Riley. 1995. “Axle Tramp Contribution to the Dynamic Wheel
Loads of a Heavy Truck.” Proceedings of the 4th International Symposium on Heavy Vehicle Weights
and Dimensions, Ann Arbor, Michigan. C. B. Winkler (ed.). pp. 425‐434.
Sayers, M. W. 1989. “Two Quarter‐Car Models for Defining Road Roughness: IRI and HRI.”
Transportation Research Record: Journal of the Transportation Research Board 1215: 165‐172.
Sayers, M. W. and S. M. Karamihas. 1996. “Estimation of Rideability by Analyzing Longitudinal Road
Profile.” Transportation Research Record: Journal of the Transportation Research Board 1536: 110‐
116.

152

APPENDIX D: DETAILED OBSERVATIONS
This appendix provides detailed observations from the roughness trends, profiles, and distress surveys
of each section within the Specific Pavement Studies (SPS‐5) project. Investigators used power spectral
density (PSD) plots, filtered elevation profile plots, and roughness profiles to monitor profile features.
Typically, roughness profiles provided the most information about the location of features that affected
the IRI most, including areas of localized roughness. In this appendix, roughness profiles were displayed
using a base length of either 10 ft (called a very short interval roughness profile) or 25 ft (called a short
interval roughness profile) unless otherwise specified. An area has localized roughness when the short
interval roughness profile reaches a peak value that is greater than 2.5 times the average IRI for the
whole section. This usually prompted more careful examination of the filtered elevation profiles.
The PSD plots were less informative, since few of the profiles were dominated by periodic content.
SECTION 040501
Roughness
The right side of the lane was much rougher than the left. This section was taken out of study after three
visits, and no significant change in roughness occurred during that time. The average HRI for each visit
was about 20 percent lower than the MRI. This is a larger difference than was observed on most other
sections, which may signify the presence of localized roughness that appears in only one side of the
lane.
PSD
The PSD plots were typical for asphalt pavement. They did not change significantly with time.
Filtered Elevation Profiles


Long wavelengths: The long‐wavelength content of the profiles was very consistent through
time.



Medium wavelengths: The medium‐wavelength elevation plots were not consistent throughout
the four visits. In addition, the profiles were not very repeatable within a given visit.
Nevertheless, the overall roughness level appeared to be about the same in each profile.



Short wavelengths: The short‐wavelength elevation plots were not very consistent between
visits or very repeatable within a given visit. The exception was the appearance of some narrow
dips throughout the left side profile.

153

Roughness Profile
A very short interval roughness profile showed that the section included multiple areas of localized
roughness. However, the location where localized roughness appeared was rarely the same in multiple
visits.
Distress Surveys
The manual distress survey (MDS) measurements recorded block cracking, alligator cracking, and
pumping in both wheel paths (WPs). This explains the difficulty with repeatability and consistency with
time.
SECTION 040502
Roughness
Rehabilitation decreased the IRI of the left side by 9 percent and the IRI of the right side by 56 percent.
The left side IRI was quite high after rehabilitation and grew at an inconsistent rate over the next 16
years. The left IRI showed a total increase of about 118 inches/mi. The right side IRI grew at a faster rate
and increased by nearly 160 inches/mi.
The HRI was 11 percent below the MRI just after rehabilitation. This gap grew steadily to 25 percent by
visit 13. The increasing difference between HRI and MRI indicates a lesser relationship between features
in the left and right profiles, and may signify the presence of localized roughness or distress that
appeared in only one side of the lane.
PSD
The PSD plots for visits 00 and 01 showed a similar level of roughness for wavelengths greater than 30 ft,
but major changes for wavelengths smaller than 30 ft.
Both the left and right PSDs included roughness that was concentrated at wavelengths near 10.9 ft.
While this periodic roughness was present for the right side profile, it was a dominant portion of the
roughness in the left side profile (see Figure 82). In fact, concentrated roughness in the waveband
between 8 and 13 ft was probably responsible for the high IRI values after rehabilitation on the left side.
For the right side profile, the PSD did not change in visits 01 through 12 for wavelengths greater than 30
ft, but the range for wavelengths below 30 ft increased steadily with time. For the left profile, the PSD
also did not change in visits 01 through 12, except for a steady increase in the range for wavelengths
below 7 ft. (See Figure 82.) It is possible that roughness in the wavelength range from 7 to 30 ft also
increased with time. However, the high content in the range from 8 to13 ft overshadowed the
progression.

154

Filtered Elevation Profiles




Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was
consistent through time. A slight change occurred between visits 09 and 10, which was caused
by the change in profiler make and the associated difference in high‐pass filtering techniques.
Rehabilitation also changed long‐wavelength elevation traces, but some aspects of the very long
wavelength content were still visible.
Medium wavelengths: The periodic content within the left side profiles dominated the content
within the medium‐wavelength profile plots. The periodic content was also visible in the right
side profiles as was other roughness. The right side profiles showed a progression in localized
rough features (dips) throughout the monitoring history. These features appear more clearly in
short‐wavelength elevation traces and unfiltered plots.
The elevation profile in the medium‐wavelength roughness range after rehabilitation was not at
all similar to the profile before rehabilitation.



Short wavelengths: Before rehabilitation (visit 00), the profiles included narrow dips (less than 2
ft wide and 0.05 to 0.25 inches deep) throughout the section. These dips did not appear with a
regular spacing.
Over the monitoring history of the section, localized roughness gradually appeared and grew in
severity at several locations on each side of the lane. These were usually narrow dips (1 to 2 ft
wide) that eventually grew to depths of up to 0.3 inch on the left side of the lane and up to 0.4
inch on the right side of the lane. By visit 13, more than 50 dips appeared on each side of the
lane that increased the roughness of the section.

Roughness Profile
The left side was twice as rough over the first 300 ft of the section than the last 200 ft because the first
300 ft of the section included periodic roughness with a wavelength that varied from 8 to 13 ft and
amplitude of as high as 0.1 inch. The last 200 ft did not. The roughness was distributed relatively equally
along the right side of the section except for increased roughness in the last 100 ft of the section in visit
13.
No localized roughness appeared in the short interval roughness profile. A very short interval (10 ft)
roughness profile showed that the progression in overall roughness was due entirely to the increase in
severity of the dips described above.
Distress Surveys
The MDS measurements showed an increase in cracking on the section throughout its entire monitoring
history. By 2002, it appeared that cracking covered the entire section, which explains the aggressive but
unsteady increase in roughness, the frequent occurrence of narrow dips within the profiles, and the
relative lack of repeatability between runs.

155

SECTION 040503
Roughness
Rehabilitation decreased the IRI of the left side by 34 percent and the IRI of the right side by 56 percent.
The MRI grew at an increasing rate over the next 16 years and increased by nearly 67 inches/mi.
PSD
The PSD plots showed very little change in content for the wavelength range from 30 to 150 ft on either
side from visits 01 through 13. On the right side, the wavelength range shorter than 30 ft became
steadily rougher over the monitoring history of the section. On the left side, the wavelength range
shorter than 15 ft grew steadily in roughness.
PSD plots for visits 00 and 01 were somewhat similar in the wavelength range above 30 ft, but the
profiles themselves were not necessarily similar over this entire range because the distribution of
roughness within certain wavebands was roughly the same; but that does not necessarily indicate
agreement between the profiles. Rehabilitation significantly reduced spectral content for wavelengths
below 15 ft.
Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was very
consistent through time. Rehabilitation also changed long‐wavelength elevation traces, but
some aspects of the very long wavelength content were still visible, particularly on the left side.



Medium wavelengths: The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.
Some features of the medium‐wavelength elevation profiles were similar throughout the
monitoring history of the section after rehabilitation. However, the roughness did appear to
increase with time. In particular, several dips seemed to grow in depth over the last five visits.
These features appeared more clearly in short‐wavelength elevation traces and unfiltered plots.



Short wavelengths: Before rehabilitation (visit 00), the left profile included narrow dips (less
than 2 ft wide and 0.05 to 0.25 inch deep) throughout the section. These dips did not appear
with a regular spacing and were rarely evident in the right side profile.
Over the monitoring history of the section, short‐duration rough features gradually appeared
and grew in severity at several locations on each side of the lane. These were usually narrow
dips (1 to 2 ft wide) that eventually grew to depths of up to 0.5 inch on the left side of the lane
and up to 0.8 inch on the right side of the lane. On the left side, the dips appeared 107 ft, 129 ft,
144 ft, 170 ft, 187 ft, 213 ft, 231 ft, 250 ft, 283 ft, 308 ft, 341 ft, 359 ft, 393 ft, 417 ft, 437 ft,
447 ft, and 461 ft from the start of the section. On the right side, the dips appeared at 19 ft,
43 ft, 56 ft, 77 ft, 97 ft, 144 ft, 171 ft, 187 ft, 212 ft, 219 ft, 232 ft, 249 ft, 282 ft, 308 ft, 340 ft,

156

359 ft, 376 ft, 397 ft, 415 ft, 434 ft, 446 ft, 463 ft, and 483 ft from the start of the section. Most
of the dips first appeared is visits 04 or 05, and grew in severity over the rest of the monitoring
history. The exception was the dip in the left side profile at 107 ft, which was relatively severe
through the entire post‐rehabilitation history of the section.
A swatch of rough pavement appeared in the right profile at 414 to 446 ft from the start of the
section in visit 10. It was not nearly as rough in visits 11 through 13.
Roughness Profile
A very short interval (10 ft) roughness profile showed that the progression in overall roughness was due
entirely to the increase in severity of the dips described above. A short interval (25 ft) roughness profile
showed that on the right side of the lane, the roughness was distributed relatively equally along the
section. On the left side, increased roughness existed in the later visits from 100 to 330 ft.
Isolated roughness (not severe enough to qualify as localized roughness) appeared on the left side at
107 ft from the start of the section in visits 01 through 13. This area stood out because the dip was
somewhat wider than the other dips, and it appeared much sooner than the others. The distress survey
recorded an area of localized distress in the same location on the left side of the lane.
Distress Surveys
All of the dips listed above appear in locations where MDS measurements reported transverse cracking.
Although all of the dips correspond to transverse cracks in the distress survey, not all transverse cracks
caused significant roughness in the profile measurements. Note that other sections, such as 040505,
040508, and 040559, also included dips at transverse cracks. However, the dips that occurred at cracks
in this pavement section were typically much deeper and progressed in roughness much more
aggressively. The swatch of rough profile on the right side from 414 to 446 ft corresponded to a large
area of cracking.
SECTION 040504
Roughness
Rehabilitation increased the IRI of the left side by 4 percent and decreased the IRI of the right side by 49
percent. The MRI changed very little (11 inches/mi) over the next 14 years, then increased 10 inches/mi
between visits12 and 13. For all visits after rehabilitation, the left IRI was about 26 to 38 inches/mi
higher than the right IRI.
PSD
The PSD plots showed very little change in content for the wavelength range from 3 to 150 ft on either
side from visits 03 through 12, but an increase in content shorter than 10 ft between visits 12 and 13. On
the right side, very little similarity existed in spectral content between visits 00 (before rehabilitation)
and 01 (after rehabilitation). On the left side, some portions of the PSD plot for visits 00 and 01 were

157

similar, but the profiles themselves were not because the distribution of roughness within certain
wavebands was roughly the same; but that does not necessarily indicate agreement between the
profiles.
Both the left and right PSDs included roughness that was concentrated at wavelengths near 12 ft. This
periodic roughness was a major portion of the roughness in the left side profile and a significant source
of roughness in the right side profile. In fact, the concentrated roughness in the waveband between 8
and 13 ft appeared to be responsible for the left to right difference in IRI.
Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was very
consistent through time. Rehabilitation caused major changes in long‐wavelength roughness.



Medium wavelengths: Only minor changes in medium‐wavelength roughness occurred from
visits 01 through 13. The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.
The periodic content within the profiles, described above, dominated the content within the
medium‐wavelength profile plots. In the right side profile, the amplitude of this roughness
ranges from 0.04 to 0.12 inch. Over much of the section, the periodic roughness on the left side
of the pavement appeared to be more than twice as severe as the right, and lagged the right
side by up to 1.5 ft. The rolling process may have caused this roughness, but it would require a
roller with a drum diameter of about 3.8 ft. In photo 27 from the construction report (Hossain et
al. 1996), it appears that this is possible.



Short wavelengths: Before rehabilitation (visit 00), the left and right profiles included narrow
dips (less than 2 ft wide and 0.04 to 0.20 inch deep) throughout the section. These dips were 3
to 25 ft apart and often appeared in the same location on both sides of the lane. Rehabilitation
eliminated the dips.
In most locations, short‐wavelength elevation plots did not change significantly over the
monitoring history of the section. Upper harmonics of the 8 to 13 ft wavelength content and
associated periodic roughness dominated the content of the plots.
Over the monitoring history of the section, roughness gradually appeared 64 ft, 108 ft, 147 ft,
190 ft, 313 ft, 362 ft, 397 ft, and 498 ft from the start of the section. These were all either
narrow dips (up to 3 ft wide) or narrow dips preceded by a small swell. These first began to
appear in visit 04 and grew in severity with time. Their severity grew the most between visits 12
and 13.

158

Roughness Profile
Roughness was distributed uniformly throughout the section. A very short interval (10 ft) roughness
profile showed that the roughness at the dips mentioned above was not significant when compared to
the periodic roughness that existed over the length of the section until visit 13.
Distress Surveys
The dip locations listed above corresponded to transverse cracks recorded in the December 5, 2005,
distress survey. Every transverse crack except one recorded on that date produced a dip in the profile.
SECTION 040505
Roughness
Rehabilitation decreased the IRI of the left side by 46 percent and the IRI of the right side by 56 percent.
The MRI grew at a slightly increasing rate over the next 16 years, and increased a total of 45 inches/mi.
PSD
The PSD plots showed very little change in content for the wavelength range from 15 to 150 ft in visits
01 through 13. However, the roughness at wavelengths below 15 ft steadily increased with time. PSD
plots for visits 00 and 01 were similar in the wavelength range above 15 ft, but the profiles themselves
were not similar for all wavelengths over 15 ft because the distribution of roughness within certain
wavebands was roughly the same; but that did not necessarily indicate agreement between the profiles.
Rehabilitation significantly reduced spectral content for wavelengths below 15 ft.
Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was very
consistent through time. Rehabilitation caused only minor changes in the profile elevation plots
over the long‐wavelength range.



Medium wavelengths: The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation. Medium‐wavelength
elevation profiles did not agree very well between visits. Further, rough features rarely showed
steady growth in severity with time.



Short wavelengths: Before rehabilitation (visit 00), the left and right profiles included narrow
dips (less than 2 ft wide and 0.05 to 0.30 inch deep) throughout the section. These dips were 3
to 25 ft apart and often appeared in the same location on both sides of the lane. Rehabilitation
eliminated the dips.
Over the monitoring history of the section, dips gradually appeared and grew in severity in at
least 12 locations on each side of the lane. These were usually narrow dips (1 to 2 ft wide) or

159

wider depressed areas of pavement (i.e., dense groups of narrow dips). Some dips appeared as
early as visit 04. Others did not appear until visit 11.
Patches of narrow dips appeared in the profiles of visit 09. These looked like “chatter” in the
traces that covered large areas of the profile, especially in the second half of the section. The
chatter was either not present or less severe in later visits.
Although the “chatter” in the profiles often appeared as very short wavelength content, they
increased the IRI.
Roughness Profile
The second half of the section was somewhat rougher than the first half on the right side in visits 09 and
11. Placement and severity of peaks within the very short interval (10 ft) roughness profile was not as
consistent for this section as others within the SPS‐5 experiment.
Distress Surveys
MDS measurements indicated a tremendous amount of cracking that began to appear before visit 04. By
visit 09, the cracking had consumed large areas of pavement, including major portions of both WPs.
Between visits 09 and 10 (May 2002), the cracks were sealed.
The cracking history explains many of the observations listed above. Cracking caused the growth in
short‐wavelength roughness. When compared to other sections within the SPS‐5 site, this section
exhibited slight degradation in profile repeatability and inconsistency in placement of roughness
because of the hit‐or‐miss nature of large areas of cracking within each WP. The profiler only measures
two narrow tracks and does not experience precisely the same cracks, or the same aspects of each
crack, equally in each pass. On the other hand, the overall IRI values were consistent between runs, and
the growth in IRI was, for the most part, steady because the cracking covered a wide area of the lane, so
the profiler was likely to experience about the same level of cracking in each pass, even if the roughness
did not always appear in consistent locations.
The rate of increase in IRI slowed somewhat between visits 09 and 10. The crack sealing that was
performed between these visits probably reduced the apparent roughness of the cracking.
SECTION 040506
Roughness
Rehabilitation reduced the IRI of the left side by 32 percent and the IRI of the right side by 51 percent.
The MRI showed a modest increase (11 inches/mi) over the next 14 years and a steeper increase (27
inches/mi) over the next two years.

160

PSD
The PSD plots showed very little change in content for the wavelength range from 2 to 100 ft on either
side over visits 01 through 11, but increased in content for wavelengths shorter than 10 ft afterward.
PSD plots for visits 00 and 01 were similar in the wavelength range above 20 ft, but the profiles
themselves were not because the distribution of roughness within certain wavebands was roughly the
same; but that does not necessarily indicate agreement between the profiles. Rehabilitation significantly
reduced spectral content for wavelengths below 20 ft.
Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was very
consistent through time. Rehabilitation also changed long‐wavelength elevation traces, but a
few aspects of the very long wavelength content were still visible.



Medium wavelengths: Only minor changes in medium‐wavelength roughness occurred from
visits 01 through 13. The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.



Short wavelengths: Before rehabilitation (visit 00), the left and right profiles included narrow
dips (less than 2 ft wide and 0.05 to 0.15 inch deep) throughout the section. These dips were 3
to 25 ft apart. They often appeared in the same location on both sides of the lane, but were
much more prevalent within the left profile. Rehabilitation eliminated the dips.
Over the monitoring history of the section, roughness gradually appeared and grew in severity
at some locations: about 132 ft on the right side, about 317 ft on both sides, about 369 ft on the
right side, and about 419 ft on both sides. Narrow dips (1 to 2 ft wide) that were 0.10 to 0.25
inch deep caused the roughness. These began to appear in visit 09, and many of them grew in
severity over the rest of the monitoring period. By visit 13, narrow dips appeared in both the left
and right side profiles 22 ft, 50 ft, 89 ft, 133 ft, 164 ft, 218 ft, 250 ft, 278 ft, 317 ft, 369 ft, and
419 ft from the start of the section.

Roughness Profile
A very short interval roughness profile showed that a few of the short‐wavelength rough features on the
section contributed to the roughness progression. For example, Figure D‐1 shows the right roughness
profile near a dip with gradually increasing severity. The roughness increases steadily with time at this
location. The roughest 10‐ft segment that includes the dip increases in severity from 140 inches/mi to
524 inches/mi from visits 01 through 13. Over this interval, the dip grew to a depth of 0.5 inch and a
width of 3 ft. This would have an impact of over 7 inches/mi on the overall roughness of the section.
Distress Surveys
All of the narrow dips listed above occurred in locations where distress surveys indicated the presence
of transverse cracks.

161

Right Roughness Profile (in/mi)
600
Visit 13
500
Visit 11

400

Visit 07
300

Visit 01

200
100
0
290

300

310

320
Distance (ft)

330

340

350

Figure D‐1. Roughness Profiles of Section 040506 (10‐ft Base Length)

SECTION 040507
Roughness
Rehabilitation reduced the IRI of the left side by 7 percent and the IRI of the right side by 50 percent.
The MRI showed a modest, but inconsistent, increase (11 inches/mi) over the next 16 years. For all visits
after rehabilitation, the left IRI is about 40 to 47 inches/mi higher than the right IRI.
PSD
The PSD plots showed very little change in content for the wavelength range from 5 to 150 ft on either
side over the 16 years after rehabilitation. On the right side, very little similarity existed in spectral
content between visits 00 (before rehabilitation) and 01 (after rehabilitation). On the left side, some
portions of the PSD plot for visits 00 and 01 were similar, but the profiles themselves were not because
the distribution of roughness within certain wavebands was roughly the same; but that did not
necessarily indicate agreement between the profiles.
Both the left and right PSDs included roughness that was concentrated at wavelengths near 12 ft. While
this periodic roughness was significant for the right side profile, it was a major portion of the roughness
in the left side profile. In fact, it appears that concentrated roughness in the waveband between 8 and
13 ft was responsible for the left to right difference in IRI.

162

Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was very
consistent through time. Rehabilitation caused major changes in long‐wavelength roughness.



Medium wavelengths: Only minor changes in medium‐wavelength roughness occurred from
visits 01 through 13. The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.
The periodic content within the profiles, described above, dominated the content within the
medium‐wavelength profile plots. Over much of the section, it appeared that the periodic
roughness on the left side of the pavement was more than twice as severe as the right and
lagged the right side by about 1.5 ft. The rolling process may have caused this roughness, but it
would require a roller with a drum diameter of about 3.8 ft. In photo 27 from the construction
report (Hossain et al. 1996), it appears that this is possible.



Short wavelengths: Before rehabilitation (visit 00), the left and right profiles included narrow
dips (less than 2 ft wide and 0.05 to 0.10 inch deep) throughout the section. These dips were 3
to 25 ft apart and often appeared in the same location on both sides of the lane. Rehabilitation
eliminated the dips.
Short‐wavelength elevation plots did not change significantly over the monitoring history of the
section through visit 12. Upper harmonics of the 8 to 13 ft wavelength content and associated
periodic roughness dominated the content of the plots. In visit 13, a bump appeared about 70 ft
from the start of the section that was 0.25 inch high on the left side and 0.15 ft high on the right.

Roughness Profile
Roughness was distributed uniformly throughout the section. The bump that appeared in visit 13 did not
affect the roughness significantly.
Distress Surveys
No significant localized roughness existed within the section that could be linked to distress. The distress
surveys listed few cracks, although a crack was recorded about 70 ft from the start of the section.
SECTION 040508
Roughness
Rehabilitation reduced the IRI of the left side by 31 percent and the IRI of the right side by 51 percent.
The MRI showed only a modest increase (25 inches/mi) over the next 16 years.

163

PSD
The PSD plots showed very little change in content for the wavelength range from 4 to 150 ft on the left
side for visits 01 through 09, and then showed an increase in roughness for wavelengths shorter than
6 ft between visits 09 and 10. The right side PSD plots did not agree as well as the left, but were
consistent in the wavelength range from 10 to 150 ft for visits 01 through 13. The right side PSD plots
also showed steadily increasing roughness for wavelengths shorter than 6 ft. Very little similarity existed
in spectral content between visits 00 (before rehabilitation) and 01 (after rehabilitation).
Filtered Elevation Profiles


Long wavelengths: Rehabilitation also caused major changes in long‐wavelength roughness, but
the very long‐wavelength content was not altered much in the second half of the section.



Medium wavelengths: Only minor changes in medium‐wavelength roughness occurred from
visits 01 through 13. The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.



Short wavelengths: Before rehabilitation (visit 00), the left and right profiles included narrow
dips (about 2 ft wide and 0.05 to 0.30 inch deep) throughout the section. These dips were 5 ft or
more apart and appeared to have uniform spacing over some parts of the section. Rehabilitation
eliminated the dips. Narrow dips did not begin to appear again until visit 08. These were all
either narrow dips (up to 2 ft wide), narrow dips preceded by a small swell, or small (0.1 inch)
downward steps. None of these dips appeared to correspond to localized roughness that existed
before rehabilitation.

Roughness Profile
Very short interval roughness profiles showed that few of the dips within the section added significantly
to the roughness progression. Although they were easily detected in the profile, most of these features
caused very little overall roughness. Two exceptions were the dips on the left profile that appeared
about 15 ft and 427 ft from the section start.
Distress Surveys
All of the dips found in profiles from the later visits appeared near locations where distress surveys
indicated the presence of transverse cracks. The transverse cracking at these locations was either
detected by the distress survey in the same year that evidence first appeared in the profiles or a year or
two earlier. Thus, it was typical to see evidence of the cracking in the profiles for visits 09 through 13,
but rarely in visits 01 through 06. The distress survey in November 1997 found very few cracks. Many
cracks were listed in the distress survey that did not cause a dip in the corresponding profile.

164

SECTION 040509
Roughness
Rehabilitation decreased the IRI of the left side by 59 percent and the IRI of the right side by 57 percent.
The MRI grew at an increasing rate over the next 16 years and increased 183 inches/mi.
PSD
Rehabilitation, performed between visits 00 and 01, greatly reduced the roughness for wavelengths
below 15 ft and changed the content at wavelengths above 15 ft. After rehabilitation, the PSD plots
showed an aggressive growth in roughness for wavelengths below 30 ft. The content for wavelengths
above 30 ft was steady with time.
Filtered Elevation Profiles


Long wavelengths: After rehabilitation, the long‐wavelength content of the profiles was
somewhat consistent through time. Rehabilitation changed the long‐wavelength elevation plots
for this section, but many of the very long wavelength traits survived the overlay.



Medium wavelengths: The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation. Medium‐wavelength
elevation profiles showed a progression in rough features (dips) throughout the monitoring
history. These features appear more clearly in short‐wavelength elevation traces and unfiltered
plots.



Short wavelengths: Before rehabilitation (visit 00), the profiles included narrow dips (less than
3 ft wide and 0.05 to 0.35 inch deep) throughout the section. These dips were 3 to 25 ft apart
and often appeared on both sides, but were relatively shallow (0.05 to 0.10 ft wide).
Over the monitoring history of the section, narrow dips gradually appeared and grew in severity
in at least 20 locations on each side of the lane. These narrow dips (1 to 2 ft wide) eventually
grew to depths of 0.10 to 0.75 inch. Most of these dips first appeared in visits 04 through 06,
and all appeared in visit 13. On the left side, the most severe dips appeared 13 ft, 33 ft, 52 ft,
73 ft, 95 ft, 105 ft, 115 ft, 128 ft, 145 ft, 161 ft, 176 ft, 192 to 195 ft, 204 ft, 214 ft, 228 ft, 244 ft,
266 ft, 286 ft, 298 ft, 342 ft, 373 ft, 401 ft, 414 ft, 433 ft, 450 ft, and 471 ft from the start of the
profile. Figure 85 shows an example of one of these dips and its progression in depth from visits
04, 07, and 11. The dips all appeared on the right side as well. On the right side, deep dips also
appeared 60 ft, 113 ft, 189 ft, 322 ft, and 356 ft from the start of the profile.

Roughness Profile
A very short interval (10 ft) roughness profile showed that the progression in overall roughness was due
entirely to the increase in severity of the dips described above with time. Figure 88 illustrates this for the

165

first half of the section. Roughness at the dips progressed aggressively over time, but the roughness
between the dips was steady.
Distress Surveys
All of the dips listed above appear in locations where MDS measurements reported cracks. In most
cases, these transverse cracks covered the entire width of the lane. Note that other sections, such as
040505, 040508, and 040559, also included dips at transverse cracks. However, the dips that occurred at
cracks in this pavement section were typically much deeper and progressed in roughness much more
aggressively.
SECTION 040559
Roughness
Rehabilitation reduced the IRI of the left side by 51 percent and the IRI of the right side by 35 percent.
The MRI showed only a modest increase (24 inches/mi) over the next 16 years.
PSD
The PSD plots showed an increase in roughness for wavelengths shorter than 6 ft between visits 07 and
08, and an increase in roughness for wavelengths shorter than 15 ft between visits 09 and 10. The
spectral content also increased for wavelengths from 1 to 10 ft between visits 12 and 13. This was
caused by localized roughness rather than periodic roughness. Rehabilitation, performed between visits
00 and 01, greatly reduced the roughness for wavelengths below 15 ft, but caused little change in the
overall level of roughness for wavelengths longer than 15 ft.
Filtered Elevation Profiles


Long wavelengths: Rehabilitation caused some change in long‐wavelength roughness, but the
very long wavelength content was barely altered.



Medium wavelengths: Only minor changes in medium‐wavelength roughness occurred from
visits 04 through 13. The elevation profile in the medium‐wavelength roughness range after
rehabilitation was not at all similar to the profile before rehabilitation.



Short wavelengths: Before rehabilitation (visit 00), the left profile included narrow dips (2 to 7 ft
wide and 0.10 to 0.35 inch deep) throughout the section. These dips were 5 to 25 ft apart and
appeared to have uniform spacing over some parts of the section. In many locations, they also
appeared in the right profile, but were not as severe. Rehabilitation eliminated the dips.
Over the monitoring history of the section, roughness gradually appeared and grew in severity
at several locations: (1) 70 ft, 128 ft, 150 ft, 221 ft, 291 ft, 337 ft, and 427 ft on both sides;
(2) 39 ft, 106 ft, 202 ft, 248 ft, 307 ft, 357 ft, 389 ft, and 441 ft on the left side only; and (3) 28 ft
and 108 ft on the right side only. These were all either narrow dips (up to 3 ft wide) or narrow
dips preceded by a small swell. Most of these first appeared in visits 09 or 10. By visit 13, some

166

of the dips included a downward change in elevation of up to 0.4 inch from the top of the swell
to the bottom of the dip.
The most severe dip occurred about 150 ft from the start of the profile on the left side. This was
5 ft wide that increased in depth throughout the monitoring history of the section until it was
0.4 inch deep. Few of the dips appeared where narrow dips existed before rehabilitation.
Roughness Profile
A very short interval (10 ft) roughness profile showed that few of the rough features on the section
added significantly to the roughness progression until visit 13. The dip that appeared 150 ft from the
start of the section qualified as localized roughness on the left side in visit 13. A dip on the right side that
was 70 ft from the start of the section nearly qualified in visit 13.
Distress Surveys
The dip locations listed above correspond to sealed cracks that were recorded in the distress survey on
December 12, 2003. The localized roughness in the left profile that appear 150 ft from the start of the
section was near a transverse crack (at 146 to 150 ft) that was observed in all distress surveys since
September 1996, which was before profiler visit 04. Further, longitudinal cracking was observed in the
left WP in distress surveys starting in December 1999.
SECTION 040560
Roughness
Rehabilitation reduced the IRI of the left side by 45 percent and the IRI of the right side by 64 percent.
The MRI grew at an increasing rate over the next 16 years, and increased 70 inches/mi overall. The
average HRI for each visit was between 19 percent and 28 percent lower than the MRI. This was a larger
difference than was observed on most other sections, which indicates a lesser relationship between the
left and right profiles, and may signify the presence of localized roughness caused by distress that
appeared in only one side of the lane.
PSD
The PSD plots showed very little change in content for the wavelength range from 15 to 150 ft in visits
01 through 13. However, the roughness at wavelengths below 15 ft steadily increased with time. PSD
plots for visits 00 and 01 were very similar in the wavelength range above 30 ft. Rehabilitation
significantly reduced spectral content for wavelengths below 15 ft.
Filtered Elevation Profiles


Long wavelengths: Rehabilitation did not change the long‐wavelength elevation plots for this
section significantly. After rehabilitation, the long‐wavelength content of the profiles was
somewhat consistent through time.

167



Medium wavelengths: Medium‐wavelength elevation plots were similar throughout visits 03
through 11. However, on the left side, some features progressed in severity with time. On the
right side, a large area of the section from 240 to 400 ft from the start changed properties
significantly over the monitoring history, particularly from visits 07 through 11. The medium‐
wavelength content was significantly rougher in the right side in visits 12 and 13 than in visit 11.
The elevation profile in the medium‐wavelength roughness range after rehabilitation was not at
all similar to the profile before rehabilitation on the right side, but exhibited weak correlation to
the profile before rehabilitation on the left side.



Short wavelengths: Before rehabilitation (visit 00), the left profile included narrow dips (about 2
ft wide and 0.05 to 0.20 inch deep) throughout the section. These dips were 5 to 50 ft apart. In
many locations, they also appeared in the right profile, but were not as severe. Rehabilitation
eliminated the dips.
For visits 03 through 13, short‐wavelength elevation plots were not very repeatable within a
given visit. This seemed to get progressively worse throughout the monitoring history of the
pavement, which explains some of the relatively low correlation values listed in Table 72. As
such, the progression of rough features at individual locations was not consistent through time.
Nevertheless, some trends were obvious. For example, patches of elevated short‐wavelength
content appeared and increased in severity in the right side profile from visits 03 through 06.
These appeared from 65 to 85 ft, 240 to 265 ft, 280 to 295 ft, and 310 to 345 ft. In later visits,
these areas became even rougher, although the details of the profile shape from the earlier
visits were not evident in the later visits. In visits 11 through 13, about half of the length of the
right side profile included high short‐wavelength content.

Roughness Profile
A very short interval (10 ft) roughness profile showed that the areas of elevated short‐wavelength
roughness did increase the IRI over time, particularly on the right side of the lane. However, no single
area stood out as dominating the roughness of this section.
Distress Surveys
Distress surveys reported a tremendous amount of cracking that began to appear before visit 04 and
became progressively more prevalent and severe throughout the rest of the monitoring history. Some of
the distress surveys also listed pumping in some areas. The cracking often first appeared as longitudinal
cracks along a WP and progressed to large areas of cracking in later visits.
The distress history explained many of the observations listed above. The appearance and growth of
patches of short‐wavelength roughness over time was consistent with distress surveys. The hit‐or‐miss
nature of profiling large areas of cracking also explained the relatively low correlation values for
repeatability within a given visit to the site. The profiler only measured two narrow tracks and did not
experience precisely the same cracks, or the same aspects of each crack, equally in each pass. On the

168

other hand, the overall IRI values showed a steady growth with time, and each area of the overall
section seemed to grow in roughness steadily. This occurred because the cracking covered a wide area
of the lane, so the profiler was likely to experience about the same level of cracking in each pass, even if
the shape of the profile did not always appear in consistent locations.
REFERENCES
Hossain, M., Douglas J. Lattin, and Larry A. Scofield. 1996. SPS‐5: Rehabilitation of Asphalt Concrete
Pavements. Construction Report. Phoenix: Arizona Department of Transportation. Unpublished
construction report.

169