Professional Documents
Culture Documents
Acknowledgments
The authors gratefully thank the Federal Highway Administration (FHWA) and SHRP LTPP program.
ABSTRACT
In 1994, the end product of the Strategic Highway Research Program (SHRP) was the asphalt concrete
mix design methodology known as the Superior Performing Asphalt Pavements (Superpave) system. As
part of the SHRP Long Term Pavement Performance (LTPP) program, an experiment identified as
Specific Pavement Study (SPS-9) was established to evaluate the long term performance of the Superpave
mix design methodology. At each project location, three main test sections were constructed
incorporating: the conventional agency mix design; the Superpave Level I mix design; and the Superpave
Level I mix design with an alternative Performance Graded (PG) asphalt binder. Further supplementary
test sections were also constructed at the majority of SPS-9 projects, evaluating the use of other PG
asphalt binders and the use of Reclaimed Asphalt Pavement (RAP) with the Superpave mix design.
This paper provides a field performance comparison and analysis of the Superpave Level I mix design
methodology with conventional agency Marshall mix design methodology using the LTPP SPS-9
experiment. In addition, an assessment of the impact in variations of PG asphalt binder grades and RAP is
established using statistical analysis of field distresses. Field distresses evaluated in the study include:
rutting; fatigue cracking; and low temperature cracking.
RÉSUMÉ
En 1994, le produit fini du programme stratégique de recherche routière SHRP consistait en la méthode de
formulation des bétons bitumineux connue sous le vocable de méthode Superpave (revêtements
bitumineux à performance supérieure). Partie du programme SHRP de performance à long terme LTPP,
une expérience identifiée étude spécifique de revêtements (SPS-9) a été créée pour évaluer la performance
à long terme de la méthode de formulation Superpave. On a construit à chacun des sites du projet trois
sections principales d’essais qui incorporaient : la méthode de formulation conventionnelle de l’agence, la
méthode de formulation Superpave de niveau 1 et la méthode de formulation Superpave de niveau 1 avec
la variante d’un bitume PG classé selon la performance. On construisit aussi d’autres sections d’essais
dans la plupart des projets SPS-9 pour évaluer d’autres bitumes PG et l’utilisation de matériaux
bitumineux à recycler (RAP) avec la méthode de formulation Superpave.
Cet article donne une comparaison de la performance en chantier et une analyse de la méthode de
formulation Superpave de niveau 1 avec la méthode de formulation conventionnelle Marshall des agences
au moyen de l’expérience SPS-9 du LTPP. En plus, on a aussi fait l’évaluation de l’impact des variations
des grades de bitumes PG en utilisant une analyse statistique des détériorations en chantier. Les
détériorations évaluées en chantier dans cette étude comprennent : l’orniérage, la fissuration à la fatigue et
la fissuration à basse température.
1.0 BACKGROUND
In 1994, the end product of the Strategic Highway Research Program (SHRP) was the asphalt concrete
mix design methodology known as the Superior Performing Asphalt Pavements (Superpave™) system.
The Superpave mix design system was developed to specifically to mitigate the most common asphalt
concrete pavement distresses: fatigue cracking; low temperature cracking; and permanent deformation.
Since the introduction of the Superpave mix design methodology the majority of transportation agencies
across North America have adopted the mix design methodology as their main asphalt concrete mix
design methodology.
The Superpave mix design system consists of a Performance Graded (PG) asphalt binder specification, a
volumetric mix design method, mix design laboratory performance tests, and performance prediction
models. Despite the introduction of this mix design method, quantification of the long-term performance
benefits of the Superpave methodology, and its relation to conventional Marshall mix design
methodologies remains relatively unknown.
2.0 INTRODUCTION
The validation of the new Superpave system was to be completed through numerous testing procedures.
One of these procedures was the analysis of controlled test sections built as an extension of the SHRP
Long Term Pavement Performance (LTPP) program. This experiment was identified as Specific
Pavement Study (SPS-9). The SPS-9 experiment was established with the purpose of evaluating the
following [1-3]:
• Comparison of the Superpave methodology and local agency mix performance in a controlled
experiment.
• Validation of the SHRP asphalt binder selection procedure for local conditions by direct comparison
of performance between different binder grades in a controlled experiment.
• To provide performance data collected over a long term from controlled field experiments and to
provide for modification of specifications at the local, regional, or national level.
• To provide training and assistance to agency personal in binder characterization procedures, the mix
design process, and establish the practicality of implementing the Superpave system.
• To provide data for Superpave models refinement and modifications.
The SPS-9 experiment was implemented in two phases. The initial phase of the experiment was the
construction of ‘pilot’ projects built during the early development of the Superpave mix design
methodology. It was anticipated that the construction of these projects would provide valuable incite into
the mix design methodology prior to the release of the final version of the Superpave mix design
specification. These “pilot” projects were identified in the SPS-9 experiment as SPS-9P: Validation and
Refinements of Superpave Asphalt Specifications and Mix Design Process. The SPS-9P experiment
required construction of a minimum of two test sections at each site that was to included the highway
agency’s standard mix, a standard mix developed under the latest version of the Superpave mix design
system, and any other alternative asphalt concrete mix that an agency would like to evaluate [3]. Under
the SPS-9P phase a total of nine projects were constructed in Arizona, Indiana, Kansas, Maryland,
Minnesota, and Wisconsin. All of the projects constructed under the SPS-9P phase of the experiment
were built in 1992 and 1993 before the completion of the final Superpave mix design methodology. The
mix design methodologies employed at these projects only corresponded to the latest version of the
Superpave mix design system that was available at the time of construction.
The second phase of the SPS-9 experiment was created in 1994 after the release of the Superpave mix
design methodology and was designated as SPS-9A: Superpave Asphalt Binder Study. SPS-9A was to
focus on two main issues: performance of Superpave mixtures relative to local agency mixtures; and the
verification of the SHRP PG asphalt binder selection process. The SPS-9A experiment required the
construction of a minimum of three test sections at each site. The test sections were to include the
highway agency’s standard mix, the Superpave Level I mix, and the Superpave Level I mix with an
alternative asphalt binder. The Superpave Level I mix with an alternative binder was meant to contain an
asphalt binder with a grade either higher or lower than that required by Superpave Level I mix design such
that the performance characteristics of interest may be expected to exhibit distresses earlier than the
Superpave binder section [1, 2]. Under the SPS-9A phase of the experiment 23 projects were constructed
from 1994 to 1998.
The overall purpose of this research is to investigate the in-service long term performance of asphalt
concrete pavements designed under the Superpave Level I mix design methodology using the projects
constructed under the LTTP SPS-9 experiment. The specific objectives of this research are:
• Assess the in-service performance of asphalt concrete pavements designed under conventional mix
design methods in relation to asphalt concrete pavements designed under the Superpave Level I mix
design system.
• Assess the possible influence of Reclaimed Asphalt Pavement (RAP) within Superpave Level I
mixes.
• Determine the impact of PG asphalt binder grades on in-service rutting and low temperature
cracking performance.
Under the SPS-9P and SPS-9A phases of the SPS-9 experiment, a total of 31 projects across North
America were constructed from 1992 to 1998. Nine projects were constructed as part of the SPS-9P phase
and 23 projects as part of the SPS-9A phase. Due to the fact that the SPS-9P projects were constructed
before the completed version of the Superpave Level I mix design specification, only projects constructed
under the SPS-9A phase of the SPS-9 experiment have been considered for the purpose of this study.
The study uses construction and construction deviation reports pertaining to the individual SPS-9A
projects and distress monitoring data assimilated from the LTPP Information Management System (IMS)
Release 19.0, January 2005 [4].
4.1 Environmental
The SPS-9A experiment called for multiple projects to be distributed throughout North America and
represent a wide range of climatic conditions. The desired climatic conditions were based on pavement
temperatures defined by the 7-day maximum pavement design temperature and the minimum pavement
design temperature. The 7-day maximum pavement design temperature is the average seven daily
maximum pavement design temperature, as calculated from the highest daily pavement temperature for
the seven hottest consecutive days. Based on the combination of experimental factors established by the
SPS-9A experiment, 32 cells were established originally established to satisfy the combination of moisture
and pavement temperatures desired in the experiment. Table 1 provides matrix of the 32 cells desired for
the project.
Wet Dry
Moisture
> 635 mm/year of precipitation < 635 mm/year of precipitation
Average 7-Day
Maximum Pavement < 52°C < 58°C < 64°C < 70°C < 52°C < 58°C < 64°C < 70°C
Design Temperature
> -46°C
> -40°C
Minimum > -34°C
Pavement
Design > -28°C
Temperature
> -22°C
> -16°C
> -10°C
Note: The average 7-day maximum pavement design temperature is the average of the highest daily pavement
temperatures for the seven hottest consecutive days of the year. The minimum pavement design
temperature is the coldest pavement design temperature of the year.
SPS-9 guidelines required that a test site must have a minimum estimated design traffic loading of 50,000
Equivalent Single Axle Loads (ESAL) per year on the outside design lane [1].
Based on the experimental design [1], the SPS-9A experiment compares conventional agency asphalt mix
designs to mixes conforming to the Superpave asphalt specifications and mix design procedures. It was
determined that it would not be practical or feasible to specify either the same mix, mix design, or even
mix design method for the individual agency conventional mixes on a North America wide experimental
pavement project. However, the SPS-9A experiments did recommend that the design of the conventional
asphalt concrete mixes be performed in compliance with the guidelines contained in the Federal Highway
Administration (FHWA) Technical Advisory T5040.27, “Asphalt Concrete Mix Design and Field
Control”, dated March 10, 1988 with the mix design conforming to the Asphalt Institute Manual, MS-2.
The Superpave mixes were to be designed using the materials specifications in SHRP A-379, as adopted
by the American Association of State Highway and Transportation Officials (AASHTO), and the mix
design procedures contained in SHRP A-407.
Superpave mixtures used in the SPS-9A test sections were to be designed in accordance with the
procedures for Superpave Level I, regardless of the traffic level determined for the site. Asphalt binder
will be selected using the PG criteria for the 98 percent reliability (mean value plus two standard
deviations) weather data provided by the nearest weather station to the project site and LTTPBIND [6].
Aggregate gradation and properties must meet the latest SHRP specification as adopted by AASHTO.
According to the SPS-9A design experiment, asphalt binder and/or aggregates not meeting the Superpave
specification and mixture design criteria may be used in supplemental sections, but should not be used in
the main Superpave test sections.
Projects in the SPS-9A experiment, in addition to the main test sections evaluating the conventional
agency and Superpave Level I mixes, were to evaluate the performance of an alternative PG asphalt binder
grade. The design for the Superpave Level I mix with an alternate PG binder grade consists of selecting
the appropriate alternate binder grade based on which pavement distress mechanisms agencies wished to
evaluate. Under SPS-9A guidelines agencies were given the option to evaluate either low temperature
cracking or rutting. The experiment required that if the agency selected low temperature cracking or
rutting the following requirements in PG binder grade selection would be met:
• Low Temperature Cracking – Low temperature grade component is increased by two grade levels
and the high grade remains the same.
• Rutting – High temperature grade decreased by one grade and low temperature grade remains the
same.
The alternate PG binder selected was then substituted for the 98 percent reliability binder with no
alteration of the aggregate gradation or binder content permitted.
According to the SPS-9A experiment guidelines [1, 2] test sections were to be built either as part of a new
or reconstructed roadway or a rehabilitated (resurfaced) rigid or flexible pavement. Pavement structural
factors such as: subgrade; subbase, base; binder; and surface asphalt concrete layers, are not controlled as
a multi-level design factor. The type and thickness of the pavement sections were to be designed using the
provincial or state pavement design procedure where the project was to be constructed. As a result, there
were no restrictions placed on the subgrade material or base material. Projects could also have porous
friction courses applied to the pavement surface depending on agency practice. The only restriction was
that the asphalt concrete evaluation layers were to be the same thickness and the pavement structure above
and below the asphalt concrete evaluation layers were to be of the same type and thickness, in all test
sections at each specific project.
Currently, the SPS-9A phase of the SPS-9 experiment contains a total of 23 projects constructed at 20
locations across the United States and Canada, built between 1994 and 1998. The constructed projects
involved 16 states and four Canadian provinces, with two states (Connecticut and North Carolina) and one
Canadian province (Quebec) constructing multiple projects as replicates for the experiment. The layout of
the SPS-9 test sites across North America is shown in Figure 1 and a complete listing of the sites and their
specific locations are listed in Table 2.
Two of the projects identified under the SPS-9A experiment in Arkansas (AR) and Nebraska (NE) had to
be eliminated from the analysis due to incomplete data sets within the LTPP IMS database.
Table 3 indicates the LTPP climate classification, average annual precipitation, mean 7-day high air
temperature, and low air temperature at each of the SPS-9A site locations to provide an overall indication
of environmental and temperature regimes covered within the SPS-9A experiment. Project specific LTPP
climate classifications and average annual precipitation were obtained directly from the LTPP IMS
database. The mean 7-day high air temperature and low air temperature for each project were retrieved
for each project site from the SHRP PG asphalt binder selection software program LTTPBIND [6]. The
corresponding Superpave PG asphalt binder grade was then calculated for each project site using
LTPPBIND using both the 1994 SHRP and 1998 LTPP PG asphalt binder selection formulae.
Under the desired SPS-9A environmental condition matrix, 32 cells were established to evaluate the
performance of the Superpave mixes. The current SPS-9A project sites represent approximately only 12
cells of the original 32 cells in the experimental design. While the representation of the current SPS-9A
sites is considerably less than what was originally desired, at the onset of the experiment, the current sites
still do provide an analysis of the Superpave mix methodology over a wide variety of environmental and
temperature regimes and will still provide meaningful analysis since each site provides a project specific
evaluation of mix designs.
Superpave Level I
98% Reliability PG
grade
Average Mean 7-day
Low
Climate Annual High
Project Temperature
Classification Precipitation Temperature SHRP LTPP
(°C)
(mm) (°C) 1994 1998
Models Models
The identification and classification of mix designs and asphalt binders used in each SPS-9A project test
section was determined from construction and deviation reports, as well as data within the LTPP IMS
database. Table 4 provides an outline of the significant test sections considered for this study at each SPS-
9A project and the asphalt concrete mix design components each evaluate.
Note: 1
All test sections constructed with 25 percent Reclaimed Asphalt Pavement (RAP)
2
Conventional agency section uses Hveem Mix Design as per agency protocol
The types of pavement structures that SPS-9A projects have been constructed on are new or reconstructed
roadways, rehabilitated asphalt concrete consisting of an asphalt overlay of asphalt concrete pavement,
and rehabilitated Portland Cement Concrete (PCC) pavements consisting of an asphalt overlay of a PCC
pavement. The type of construction is dependent on where the project was constructed and where the
transportation agency, constructing the project, decided it should be placed. Although the type of
construction is different from project to project, the type of construction is the same for all test sections at
a particular project. Similarly, the types of base structures the projects were constructed on is a multi level
variable. The types of base structures include: granular; cement stabilized; and asphalt treated depending
on the project. However, at each project, all test sections have the same base structure.
A listing of the SPS-9A projects and their respective construction methodology and pavement base
structure classifications are shown in Table 5. In the SPS-9A experiment, 11 projects were constructed as
new asphalt concrete pavements and eight were constructed as asphalt overlay rehabilitation projects, on
eight previous asphalt pavement structures and three on PCC structures.
Table 5. SPS-9A Design Pavement Structures
Constructed Pavement Layer
Thickness (mm)
SPS-9A AC
Project Type of Construction Surface
AC Mix Binder /
Friction
Evaluation Levelling
Course1
Layer Course1
AZ Overlay of AC - 76 102
CT-1 Overlay of AC - 63 25
CT-2 Overlay of AC - 63 25
FL Overlay of AC 16 64 12.52
IN Overlay of PCC (Rubblized) - 164 -
MI Overlay of PCC (Rubblized) - 203 -
MS Overlay of AC - 64 -
MO Overlay of PCC - 102 -
MT New - 127 -
NJ Overlay of AC - 64 76
NM Overlay of AC 16 114 -
NC-1 New - 229 -
NC-2 Overlay of AC - 64 -
OH New - 102 -
TX New - 76 -
WI New - 229 -
AB New - 120 -
ON New - 130 -
PQ-1 New - 64 64
PQ-2 New - 64 64
SK New - 120 -
1
Same mix and thickness for each test section AC-asphalt
2
Asphalt Membrane, PCC - Portland Cement Concrete
The extent and severity of rutting, fatigue cracking, and low temperature cracking was determined based
on data available within the LTPP IMS database. Under the LTPP program rutting, fatigue cracking and
low temperature cracking are determined from transverse profiles and pavement condition surveys. For
the purpose of this analysis the magnitude of rutting, fatigue cracking, and low temperature cracking was
determined from the most recent available transverse profile and pavement condition survey records for
each of the evaluated test sections. It should be noted that several of the projects within the experiment
have had major rehabilitation treatments, these projects: Mississippi (MS); and two projects in Quebec
(PQ-1 and PQ-2) were only evaluated until the time of their rehabilitation treatments in 2000 and 2003,
respectively.
In the LTPP program, rutting is determined based on an evaluation of the transverse profile. Two primary
devices are used to determine the transverse profile of test sections, the Dipstick® [5], and the automated
Pavement Distress Analysis System (PADIAS) software [5]. For this study, rutting values were
determined based on data collected by the Dipstick. The magnitude of rutting is determined in the LTPP
program based on rut depth as measured at specified intervals along a pavement section. In the LTPP
program, measurement of rutting is based on an interpretation of transverse profile measurements.
Transverse profile measurements are performed using manual measurements from photographic
techniques or using the Dipstick profiler. The Dipstick technique uses a uniform spacing of 0.305 m
between profile points at 11 intervals at a 15 m spacing over the 152 m sections. The rutting values used
in the study are the maximum mean depth of the inner and outer wheel paths of each section, as
interpreted from the 1.8 m straight edge reference.
In the LTPP program [5], cracking is evaluated with manual and automated PADIAS cracking distress
surveys. Manual surveys are conducted based on a visual observation of the pavement surface by an
objective evaluator. Automated PADIAS cracking distress surveys are also conducted using automated
distress detection equipment. For this study, manual surveys were used to evaluate the magnitude of
cracking.
Low temperature cracking of asphalt concrete pavement is primarily identifiable as transverse cracks,
perpendicular to the centreline of a road that occur at regular intervals across the entire width of the
pavement structure. Transverse cracking also occurs when an existing pavement is overlaid. This type of
cracking is referred to as “reflective cracking” and occurs after an existing pavement is resurfaced.
Reflective cracks form when underlying cracks, in the previous pavement, propagate up through the new
asphalt concrete layer under one or more thermal cycle. The occurrence of low temperature cracking for
the evaluated test sections was determined based on transverse cracking as observed in LTPP distress
surveys. During LTPP distress surveys the extent of transverse cracking is determined from a count and
measurement of the length of transverse cracks within a test section. The length of transverse cracks
within a section was used in this study as the mechanism for evaluating transverse cracking.
Fatigue cracking is defined by a series of longitudinal and interconnected cracks mainly due to repeated
applications of traffic wheel loadings. During LTPP distress surveys the extent of fatigue cracking is
determined from a measurement of the area of alligator cracks within a test section.
Transverse and fatigue cracking are further classified in the LTPP program based on severity. Severity is
determined based on specific threshold values of low, moderate and high. For the purpose of this analysis,
the low, moderate, and high severity cracks were totalled directly with no distinction between severities.
Only transverse profile and pavement condition survey data listed as Level E, the highest quality within
the LTPP IMS database was used.
In the study, the following asphalt concrete mix design and distress performance comparisons were
considered based on a direct comparison at each project site:
Initially, an overall examination of the mean magnitude of distresses was considered for each of the
desired mix design comparisons. Then a statistical analysis was performed in order to determine if a
significant difference in performance could be identified for the desired mix design comparisons.
The statistical analysis consisted of a paired T-test and a two-way Analysis of Variance (ANOVA) to
determine differences in mean and differences in variance [7], respectively between the magnitude of
rutting, fatigue cracking, and low temperature cracking for each of the considered mix design comparisons
at each project location. The analysis assumed that each population was normal. A Null Hypothesis (Ho)
that stated that the considered distress either rutting, low temperature cracking, and fatigue cracking
between two comparative sections was not statistical different was assumed. An Alternate Hypothesis
(Ha) is true if Ho is rejected and there is a significant difference in performance.
The paired T-test performs a paired, two-sample, test to determine whether the distresses of the means of
two compared asphalt concrete mixes are distinct. The T-test is calculated using the mean difference and
standard deviation between the distresses of paired sections. A critical T value (Tcritical) is established and
if the calculated T value (Tcalculated) is less than Tcritical, the Ho is accepted and there is no significant
difference in performance. If Tcalculated is greater than Tcritical, the Ho is rejected and Ha is accepted and there
is a significant difference in performance. Likewise, the ANOVA, performs a test that separates the total
variation of the data into variation between group’s means and variation within groups using an F-test. A
critical F value (Fcritical) is established and if the calculated F value (Fcalculated) is less than Fcritical, the Ho is
accepted and there is no significant difference in performance. If Fcalculated is greater than Fcritical, the Ho is
rejected and Ha is accepted and there is a significant difference in performance.
8.1 Analysis of Conventional Agency Marshall Mixes and Superpave Level I Mixes
Generally, the states and provinces that constructed SPS-9A experiments constructed at least one test
section evaluating the conventional agency Marshall mix design and one test section evaluating the
Superpave Level I mix design. However, Mississippi (MS) decided not to construct a test section
evaluating the conventional agency mix design. Texas (TX) constructed a conventional agency mix
design using the Hveem mix design method, as per their agency protocol. Since all of the other agencies
used the Marshall design in the construction of their conventional agency test section, the Texas project
was not included in the analysis. As a result, a total of 19 SPS-9A projects evaluate conventional agency
mix designs in comparison with the Superpave mix design. Table 6 provides an overview of the projects
and test sections evaluating the conventional agency Marshall and Superpave Level I test sections. The
projects considered for the performance analysis range in age from 5.2 to 8.8 years with an average age of
6.8 years.
An overall examination of rutting, low temperature cracking, and fatigue cracking was conducted to
evaluate possible differences in performance between the conventional agency Marshall mix design and
the Superpave Level I mix design. An overall examination of sections constructed with the conventional
agency Marshall mix design indicates that 68 percent have low severity rutting (< 7 mm), 37 percent have
no low temperature cracking, and 47 percent have no fatigue cracking. An overall examination of sections
constructed with the Superpave Level I mix design indicates that 68 percent have low severity rutting (< 7
mm), 47 percent have no low temperature cracking, and 58 percent have no fatigue cracking. Table 7
presents the results of the average distress magnitudes for the considered sections. Based on the
assessment of mean distress magnitudes, conventional agency Marshall mixes generally have a lower
magnitude of rutting than Superpave Level I mixes. However, Marshall mixes have higher magnitudes of
low temperature cracking and fatigue cracking.
Table 6. SPS-9A Projects Assessing Conventional Agency Marshall and Superpave Level I Mix
Designs
Table 7. Average Performance Difference of Conventional Agency and Superpave Level I Mix
Designs
Average Conventional
Number of Superpave
age Distress Statistic Agency
Projects Level I
(years) Marshall
Mean (mm) 6.3 7.3
Rutting Standard
3.6 5.2
Deviation (mm)
Low Mean (m) 25.5 14.2
19 6.8 Temperature Standard
Cracking 47.6 39.8
Deviation (m)
Mean (m) 25.4 12.8
Fatigue
Standard
Cracking 64.1 24.8
Deviation (m)
A statistical analysis of distress mechanisms was employed to determine if a significant difference exists
between the performance of conventional agency Marshall and Superpave Level I mixtures. A two-
sample paired T-test and a two factor ANOVA were conducted at an α level of 0.05 or 95 percent and
showed no significant difference in rutting, low temperature cracking, and fatigue cracking performance
between the conventional agency Marshall and Superpave Level I mixtures. The results of the two-
sample paired T-test and ANOVA analysis are presented in Table 8.
Table 8. Paired T-test Analysis and Analysis of Variance (ANOVA) of Conventional Agency Mixes
vs. Superpave Level I Mixes
Projects in the SPS-9A experiment, in addition to the main test sections evaluating the conventional
agency Marshall and Superpave Level I mixes, were to evaluate the performance of an alternative PG
asphalt binder grade. The Superpave Level I mix design specification recommends that an asphalt binder
will be selected using the design PG grading criteria with a 98 percent high and low temperature reliability
(mean value plus two standard deviations) based on weather data provided by the nearest weather station
at a project location. The use of an alternative PG asphalt binder grade in SPS-9A experiment to
specifically evaluate the rutting and low temperature cracking performance based on the alteration of high
and low temperature PG grades from the design 98 percent reliability PG grade. Agencies were given the
choice to evaluate either rutting or low temperature cracking and adjust either the high or low temperature
PG grade in order to evaluate PG grade sensitivity for the chosen distress.
The agencies that chose to evaluate rutting based on an adjustment to the high temperature PG grade are
listed in Table 9. These agencies then chose whether to evaluate the sensitivity of high temperature PG
grade through use of a PG grade binder one grade lower than the 98 percent reliability PG binder or a PG
grade binder one or two grades higher than the 98 reliability binder.
The average rutting performance of the design PG grade binders and the selected alternative binders is
presented in Table 10. An adjustment of one high temperature PG grade lower than the 98 percent
reliability PG design grade was seen to increase rutting by approximately 10 percent. An adjustment of
one and two high temperature PG grades higher than the 98 percent reliability PG design grade were
shown to enhance resistance to rutting. Both of these comparisons indicated less rutting than identified in
the PG design grade section by approximately 18 and 24 percent, respectively. Based on the Superpave
PG asphalt binder criteria, both of the identified trends would have been anticipated.
Table 9. SPS-9A Projects Assessing the Influence of Superpave Level I Design PG Asphalt Binder
Grade on Rutting
Table 10. Average Performance Difference of Superpave Level I Design PG Asphalt Binder Grade
and Superpave Alternative PG Asphalt Binder Grade, Rutting
Superpave Alternative
Number Average
Level I PG Asphalt
Criteria of age Statistic
Design Test Binder Test
Projects (years)
Section Section
Mean (mm) 7.4 8.4
Design PG & -1
8 7 Standard Deviation
HT PG grade 3.9 5.9
(mm)
Mean (mm) 5.3 4.3
Design PG & +1
6 6.8 Standard Deviation
HT PG grade 1.5 1.5
(mm)
Mean (mm) 6.8 5.2
Design PG & +2
4 6.2 Standard Deviation
HT PG grade 3.0 1.8
(mm)
Note: HT= High Temperature grade adjustment
A statistical analysis of the magnitude of rutting was employed to determine if a significant difference
exists between the performance of 98 percent reliability PG binder grades and binder grades with a lower
or higher than 98 percent high temperature reliability. A two-sample paired T-test and a two factor
ANOVA were conducted at an alpha (α) level of 0.05 or 95 percent. The results indicate no significant
difference in rutting susceptibility or resistance of Superpave Level I mixtures with lower or higher than
the recommended 98 percent reliability high temperature PG grades. The results of the two-sample paired
T-test and ANOVA analysis are presented in Table 11.
Table 11. Paired T-test Analysis and Analysis of Variation (ANOVA) of Superpave Level I and
Superpave Alternative PG Asphalt Binder Grade, Rutting
Design PG & -
1 HT PG 7 -1.7 2.4 No 1.2 5.6 No
grade
Design PG &
+1 HT PG 5 1.7 2.6 No 3.0 6.6 No
grade
Design PG &
+2 HT PG 3 1.6 3.2 No 2.5 10.1 No
grade
The agencies that chose to evaluate low temperature cracking based on an adjustment to the low
temperature PG grade are listed in Table 12. These agencies then chose whether to evaluate the sensitivity
of low temperature PG grade through the use of a PG binder one or two grades lower than 98 percent
reliability PG binder.
An overall examination of the projects indicated that two projects: Saskatchewan (SK) and Michigan (MI)
had no recorded low temperature cracking. The average low temperature cracking performance of the
design PG grade binders and the selected alternative binders is presented in Table 13. An adjustment of
one low temperature PG grade higher i.e., less 98 percent reliability, then the 98 percent reliability PG
design grade was shown to slightly decrease the magnitude of low temperature cracking, an unexpected
trend. An adjustment of two low temperature PG grades higher i.e., approximately 50 percent reliability,
then the 98 percent reliability PG design grade was shown to greatly increase the magnitude of low
temperature cracking. This trend would have been anticipated based on the Superpave PG asphalt binder
selection criteria whereby PG binders with less than 98 percent low temperature reliability would be
theoretically more susceptible to low temperature cracking.
Table 12. SPS-9A Projects Assessing the Influence of Superpave Level I Design PG Asphalt Binder
Grade on Low Temperature Cracking
Table 13. Average Performance Difference of Superpave Level I and Superpave Alternative PG
Asphalt Binder Grade, Low Temperature Cracking
Alternative
Number Average Superpave
PG Asphalt
Criteria of age Statistic Level I Test
Binder Test
Projects (years) Section
Section
A statistical analysis of low temperature cracking was employed to determine if a significant difference
exists between the performances of 98 percent reliability PG binder grades and PG binder grades with a
lower than 98 percent low temperature reliability. A two-sample paired T-test and a two factor ANOVA
were conducted at an α level of 0.05 or 95 percent. The results indicate no significant difference in low
temperature cracking susceptibility of Superpave Level I mixtures with one or two PG grades lower than
the recommended 98 percent reliability low temperature PG grades. The results of the two-sample paired
T-test and ANOVA analysis are presented in Table 14.
Table 14. Paired T-test Analysis and Analysis of Variance (ANOVA) of Superpave Level I and
Superpave Alternative PG Asphalt Binder Grade, Low Temperature Cracking
Within the SPS-9A experiment only a limited number of projects constructed test sections evaluating the
use of RAP in Superpave mixtures. Four projects, two in Connecticut (CT-1, CT-2), Indiana (IN),
Wisconsin (WI), and Saskatchewan (SK) constructed test sections that utilized a Superpave Level I
mixture with RAP. Table 15 provides an overview of the test sections that evaluate the impact of RAP
and the Superpave Level I mix design.
Table 15. Matrix of SPS-9A Projects Assessing the Influence of Superpave Level I Design and
Reclaimed Asphalt Pavement (RAP)
Superpave Level I
Superpave Level I with RAP
Design
Project Asphalt Asphalt
Test Test %
Binder Binder
Section Section RAP
Grade Grade
090902 PG 64-28 090961 25 PG 64-28
CT-1 / CT-2
090903 PG 64-22 090962 25 PG 64-22
IN 18A902 PG 64-28 18A959 15 PG 64-28
WI 55C902 PG 58-34 55C960 20 PG 58-34
SK 900902 PG 52-40 900962 18 PG 52-40
The average performance of the Superpave Level I mixtures and Superpave Level I mixtures with RAP is
presented in Table 16. Superpave Level I mixtures with RAP generally have less rutting than Superpave
Level I mixtures without RAP. Based on an assessment of the mean magnitude of distresses, Superpave
Level I mixtures with RAP also had less low temperature cracking than those containing no RAP.
However, at two projects where fatigue cracking has occurred in Saskatchewan (SK) and Wisconsin (WI)
a greater magnitude of fatigue cracking is identified with the addition of RAP.
Table 16. Average Performance Difference of Superpave Level I and Superpave Level I with
Reclaimed Asphalt Pavement (RAP)
Superpave
Average Superpave
Number of Level I w/
age Distress Statistic Level I Test
Projects RAP Test
(years) Section
Section
Mean (mm) 6.5 4.8
Rutting Standard
2.2 1.6
Deviation (mm)
Low Mean (m) 6.1 5.3
Temperature Standard
5 6.6 Cracking 5.4 7.5
Deviation (m)
A statistical analysis of distress mechanisms was employed to determine if a significant difference exists
between the performances of Superpave Level I mixtures without and with RAP. A two-sample paired T-
test and a two factor ANOVA were conducted at an α level of 0.05 or 95 percent. The results indicate that
the rutting performance of Superpave Level I mixtures with RAP is significantly different than Superpave
Level I mixtures not containing RAP. With Superpave Level I mixtures with RAP performing better than
those without. No significant difference in low temperature cracking and fatigue cracking were found.
The results of the two-sample paired T-test and ANOVA analysis are presented in Table 18.
Table 18. Paired T-test Analysis of Superpave Level I and Reclaimed Asphalt Pavement (RAP)
9.0 CONCLUSIONS
This study evaluated 19 projects within the LTPP SPS-9A experiment for rutting, low temperature, and
fatigue cracking performance. The performance analysis was conducted when the sections ranged in age
from 5.2 to 8.8 with an average age of 6.8 years. A statistical analysis using a paired T-test and a two-way
ANOVA was conducted to determine if a statistical significant difference in performance could be
identified for the desired mix design comparisons. The following conclusions are made based on the
evaluation and statistical analysis:
• A direct comparison of conventional agency Marshall and Superpave Level I mix designs shows
that Marshall sections generally have lower rutting and higher low temperature and fatigue cracking.
However, no statistically significant difference in the performance of conventional agency Marshall
mixes and Superpave Level I mixes with respect to rutting, low temperature, and fatigue cracking
could be identified.
• Generally, the use of a PG grade asphalt binder with less than 98 percent high temperature reliability
increased rutting susceptibility. However, no statistically significant difference for sections
constructed with high temperature PG grade classification less than the design recommended 98
percent reliability was identified.
• Generally, the use of a PG grade asphalt binder with greater than 98 percent high temperature
reliability indicated increased rutting resistance. However, no statistically significant difference in
rutting resistance for sections constructed with a high temperature PG grade classification greater
than the design recommended 98 percent reliability was identified.
• Generally, the use of a PG grade asphalt with less than 98 percent low temperature reliability
increased low temperature susceptibility. However, no statistically significant difference in low
temperature cracking susceptibility for sections constructed with a low temperature PG grade
classification less than the design recommended 98 percent reliability was identified.
• A direct comparison of Superpave Level I mixtures and Superpave Level I mixtures with RAP
shows that RAP sections have less rutting and low temperature cracking, however they have higher
fatigue cracking. A statistically significant difference in the rutting resistance of Superpave Level I
mixes with and without RAP was identified, with mixes containing RAP determined to be
performing significantly better with respect to rutting than those without RAP. No statistically
significant difference in the low temperature and fatigue cracking performance for Superpave Level
I mixes with and without RAP.
10.0 RECOMMENDATIONS
The LTPP SPS-9 experiment presents the unique ability to evaluate the performance of asphalt concrete
designed utilizing the Superpave methodology in a controlled field experiment across North America.
This study provided an initial look at the performance of test sections within the SPS-9A experiment in
order to quantify the perceived performance benefits of the Superpave Level I mix design. Further study
would be beneficial in the following areas:
• Continued study in the future for the remaining pavement life cycle to determine if the anticipated
performance benefits of the Superpave Level I mix design may be quantified.
• Further analysis to determine the influence of as constructed asphalt binder properties, aggregate
gradation, and mix design properties on the performance of test sections within the SPS-9A
experiment.
• Validation of the Superpave PG asphalt selection criteria through the use of automated temperature
data recorded at project locations.
• Study the relationship between traffic loading and Superpave design gyrations from automated
traffic data collection at the project locations.
• The development of equations to explain the progression of rutting, low temperature cracking, and
fatigue cracking for the Superpave Level I mix design.
REFERENCES
1. Federal Highway Administration (FHWA). Specific Pavement Studies: Experimental Design and
Research Plan for Experiment SPS-9A Superpave Asphalt Binder Study, U.S. Department of
Transportation, Washington, D.C. (1995).
2. Federal Highway Administration (FHWA). Specific Pavement Studies: Construction Guidelines for
Experiment SPS-9A Superpave Asphalt Binder Study, U.S. Department of Transportation,
Washington, D.C. (1995).
3. Federal Highway Administration (FHWA). Specific Pavement Studies: Experimental Design and
Research Plan for Experiment SPS-9 Validation of SHRP Asphalt Specifications and Mix Design
and Innovations in Asphalt Pavements, U.S. Department of Transportation, Washington, D.C.
(1993).
7. Mason R. Statistical Design and Analysis of Experiments with Applications to Engineering and
Science, John Wiley & Sons (2003).
Orientation
Layer Asphalt Air with
Test Pavement Asphalt # of VMA NMAS
Project Thickness Content Voids Superpave
Section Layer Binder Blows (%) (mm)
(mm) (%) (%) Restricted
Zone
Orientation
Layer Asphalt Air with
Test Pavement Asphalt VMA NMAS
Project Thickness Ndes Content Voids Superpave
Section Layer Binder (%) (mm)
(mm) (%) (%) Restricted
Zone
10
6
Age (Years)
0
NC-2 MS MT NJ CT-1 CT-2 ON NC-1 IN NM PQ-1 PQ-2 WI AZ FL TX MO SK MI AB OH
Figure A1. Project Age, Most Recent Transverse Profile, and Pavement Condition Survey
30
25
20
Rut Depth (mm)
15
10
0
NC-2 MT NJ CT-1 CT-2 ON NC-1 IN NM PQ-1 PQ-2 WI AZ FL MO SK MI AB OH
200
190
180
170
160
130
Transverse Cracking (m)
120
110
100
90
80
70
60
50
40
30
20
10
0
NC-2 MT NJ CT-1 CT-2 ON NC-1 IN NM PQ-1 PQ-2 WI AZ FL MO SK MI AB OH
Figure A3. Conventional Agency Marshall and Superpave Level I, Transverse Cracking
300
Conventional Agency Marshall
Superpave Level I
275
250
225
200
Fatigue Cracking (m2)
175
150
125
100
75
50
25
0
NC-2 MT NJ CT-1 CT-2 ON NC-1 IN NM PQ-1 PQ-2 WI AZ FL MO SK MI AB OH
Figure A4. Conventional Agency Marshall and Superpave Level I, Fatigue Cracking
30
20
Rut Depth (mm)
15
10
0
MS NJ NC-2 NC-1 IN AZ FL TX MO SK AB OH
Figure A5. Superpave Level I PG Design Grade and Superpave Level I Alternative Design Grade,
Rutting Analysis
60
PG Design Grade
PG -1 Low Temperature Grade
PG -2 Low Temperature Grades
50
40
Transverse Cracking (m)
30
20
10
0
MT CT-1 CT-2 ON IN NM WI PQ-1 PQ-2 SK MI MO
Figure A6. Superpave Level I PG Design Grade and Superpave Level I Alternative Design Grade,
Transverse Cracking Analysis
30
25
20
Superpave Level I
Rut Depth (mm)
10
0
CT PG Design Grade CT -1 LT Design Grade IN WI SK
Figure A7. Superpave Level I and Superpave Level I with Reclaimed Asphalt Pavement (RAP),
Rutting
200
190
180
170
160
150
Superpave Level I
140
Superpave Level I w/ RAP
130
Transverse Cracking (m)
120
110
100
90
80
70
60
50
40
30
20
10
0
CT PG Design Grade CT -1 LT Design Grade IN WI SK
Figure A8. Superpave Level I and Superpave Level I with Reclaimed Asphalt Pavement (RAP),
Transverse Cracking
300
275
250
Superpave Level I
225 Superpave Level I w/ RAP
200
Fatigue Cracking (m2)
175
150
125
100
75
50
25
0
CT PG Design Grade CT -1 LT Design Grade IN WI SK
Figure A9. Superpave Level I and Superpave Level I with Reclaimed Asphalt Pavement (RAP),
Fatigue Cracking