Learning Center
Plans & pricing Sign in
Sign Out

1 - Pennsylvania Department of Transportation



        Phase I Final Report
     PennDOT RFQ 04-02 (C13)

           Submitted to
 QES, Quality Engineering Solutions

        Ghassan Chehab, PhD
         Assistant Professor

          Tanmay Kumar
     Graduate Research Assistant

  Pennsylvania Transportation Institute
    The Pennsylvania State University
  201 Transportation Research Building
       University Park, PA 16802

           December 13, 2005
The purpose of this report is to document the research activities that have been conducted
at the Pennsylvania State University as a deliverable of the subcontract to QES, Inc. for
Phase 1 of the PennDOT research project for Preventive maintenance on I-79 (RFQ 04-
02 (C13). A standalone report documenting literature review conducted by the Penn State
research team had been submitted previously.
       Particularly, the report documents findings from Tasks I-A, laboratory and
accelerated pavement testing, and Task I-C, design evaluation. Work under those tasks
constitutes the scope of work to be conducted by Penn State for Phase 1 of the project.
One of the primary rehabilitation techniques used by state highway agencies across the
US, including PennDOT entails placement of an HMA overlay on top of an existing
flexible or rigid pavement. Pre-overlay operations sometimes include milling, sealing,
grinding, application of interlayers, rubblization, crack and seat of PCC pavements,
among many other treatments and product applications.
       For high-volume and other roads in Pennsylvania, PennDOT’s typical procedures
consist of a 1.5 inch overlay with a 1 inch depth leveling layer. After milling down
existing pavement, a 1.5 inch overlay is placed. Typical HMA mixes include 9.5 and 12.5
mm Superpave mixtures of 76-22 modified binders and coarse-gradation. RAP is allowed
to be used in the mixtures.

Project Background and Objectives
One of the primary rehabilitation techniques used by state highway agencies across the
US, including PennDOT entails placement of an HMA overlay on top of an existing
flexible or rigid pavement. Pre-overlay operations sometimes include milling, sealing,
grinding, application of interlayers, rubblization, crack and seat of PCC pavements,
among many other treatments and product applications.
       For high-volume and other roads in Pennsylvania, PennDOT’s typical procedures
consist of a 1.5 inch overlay with a 1 inch depth leveling layer. After milling down
existing pavement, a 1.5 inch overlay is placed. Typical HMA mixes include 9.5 and 12.5

     mm Superpave mixtures of 76-22 modified binders and coarse-gradation. RAP is allowed
     to be used in the mixtures.
     AC      In an attempt to evaluate the current overlay structural design, PennDOT initiated
     a research project in which the inclusion of a 2” binder layer underneath the 1.5” surface
     overlay would be studied. The focus of the project is to assess whether an overlay
     structure consisting of a thicker overlay; i.e., inclusion of a 2” binder layer is functionally
     and cost-effective in prolonging the useful life of second-generation overlays. The
     research project consists of two major tasks: a field experiment, and a complimenting

     laboratory experiment.
     The field experiment was conducted on I-79 in Butler County, Pennsylvania. The project involves 9.7
        miles of northbound and 8.7 miles of south bound lanes. The pavement had an existing 4.5 inch
      asphalt overlay on top of a JRCP pavement. The control section of the project consists of a 1.5-inch
     overlay with a 1.5 inch maximum depth leveling layer. The experimental section consists of a 1.5-inch
       overlay and 2-inch binder course layer. The existing asphalt overlay was milled to a depth of 1.5
                          inches prior to placement of the second generation overlay.
     Figure 1 shows the two pavement structures for the control and experimental sections.

                                                             1.5 in. Surface Course
            1.5 in. Surface Course

               2.5 in. Binder                              Existing AC milled to 2.5 in.
         Existing AC milled to 2.5 in.
                                                                    10 in. PCC

                  10 in. PCC

     SUBGRADE                                                        Subgrade

DE   SUBGRADE     Subgrade

      Figure 1: The pavement structure for section containing binder layer (left) and for
                                the section without binder layer (right)
     The laboratory experiment includes advanced laboratory tests to characterize the
     properties and behavior of the mixtures used. Findings from testing are used in predicting

the performance of the control and experimental sections using linear elastic analysis and
empirical performance prediction equations. The laboratory experiment also includes
accelerated pavement testing using the MMLS, Model Mobile Load Simulator. The
objective of conducting those tests is to compare the performance of the control section
and the experimental section (with binder layer) using slabs constructed in the lab.

Summary of Work Conducted
The following report presents work conducted at Pennsylvania State University in
fulfillment of the following tasks/subtasks:
      Subtask I-A-(1) Literature Search: A comprehensive review of literature was
       conducted to investigate previous research work conducted on second generation
       overlays. The review also included the effects of various mixture properties and
       new additives/products on performance. A report summarizing the review of
       literature had been compiled and submitted.
      Subtask I-A-(2) Laboratory Testing: Simple Performance Complex Modulus (E*)
       testing was carried out on the mixes being used for both the binder and the surface
       layers. Dynamic Modulus and phase angle mastercurves were constructed for use
       in structural design evaluation, mixture characterization, and performance
      Subtask I-A-(2) Accelerated Testing: Accelerated testing using the MMLS3 was
       conducted for both control and experimental sections using the briquette setup.
       Trafficking speed was at two cycles per second without wander at 40°C and dry

Literature Review
The need for cost-effective long lasting rehabilitation techniques for asphalt pavements
has been a major concern for most state highway agencies since the 1960s. Recent efforts
in material selection, mix design, performance prediction, and pavement testing unveiled

several alternatives for obtaining long lasting performance from old asphalt pavements
(some as old as 50 years) through periodic replacement of the asphalt surface. As a result,
the potential for traditional fatigue cracking and rutting is reduced and pavement distress
confined to the upper layer or surface of the structure. Thus, when the surface reaches the
distress threshold, removal and/or replacement of this surface alone becomes a highly
efficient and economical solution.
       On Interstate 79 in Butler County, PA PennDOT is exploring a new way to
produce an economically feasible long-lasting HMA overlay, with focus placed on
benefit from increasing the thickness of the overlay by including a 2.5 inch overlay. In
investigating practices of other states, it was found that Maryland incorporates a
relatively thick asphalt structure with stone matrix asphalt (SMA) on roads with a daily
traffic of 2,000 ESALs per day or more. California requires this in case the truck traffic
exceeds 15,000 vehicles per day or an average of 150,000 vehicles per day [1]. It has
been argued that similar treatment can prove cost effective even for medium and low
volume roads, with only a few modifications.
   A vast variety of literature from various sources was reviewed to arrive at possible
ways of improving the performance of the binder and wearing courses. Most importantly,
Epps, et al [2] summarized the structural requirements of the binder course for overlays
as follows:
      High rutting resistance,
      Low permeability, and
      High stability, i.e., a strong aggregate skeleton, which can be achieved either by
       using large nominal size aggregate or adding crushed stone, gravel or other
       additives that enhance aggregate to aggregate contact.
Similarly, an ideal wearing course must have the following characteristics:
      High rutting resistance,
      Low permeability,
      High wear resistance, and
      High resistance to thermal cracking.

Pavement Distresses, Causes and Solutions
As can be deduced from the preceding paragraph, there are pavement distresses that are
of critical concern. These are discussed in some detail in this section.

Permanent Deformation (Rutting)
Decker and Goodrich [3] defined rutting as the longitudinal depression in the wheel path
accompanied by projections at the sides, which is known as shoving. The primary causes
of rutting are poor shear resistance of the asphalt binder and / or densification of
insufficiently compacted pavements [4]. In most cases, rutting occurs in the top 3 – 4
inches of the pavement structure. Besides providing an uncomfortable driving surface,
rutting can also prove dangerous in wet or frosty conditions by causing hydroplaning and
ice accumulation, respectively [5].
       McGennis et al. [6] found that use of a stiff binder accompanied by use of
aggregate with rough texture with roughly cubical shape can reduce the rutting potential
of an asphalt mix. However, with the advent of the Superpave system, the findings for
aggregate shape no longer hold true, as discussed later. At the same time, however, it has
been found that too high a binder content can increase rutting [7] as the asphalt will act as
a lubricant between the aggregate particles at higher temperatures. Further, since rutting
is also caused by densification of the pavement, an optimum air void content in the
constructed pavement will also help in reducing future rutting [8].

Fatigue cracking
Fatigue cracking has been defined as the longitudinal wheel path cracking that is caused
by repeated application of loads [9]. This is a complex phenomenon occurring from
repeated bending of the pavement that results in micro structural damage to the pavement.
It is a competitive process between micro cracking and healing, which occurs during
periods of relaxation and is accelerated at higher temperatures. As can be expected,
fatigue cracking is highly dependent on the usual suspects, i.e, binder content and grade,
aggregate interlocking and constructed air void content [10, 11].

Top-down cracking
Significant top-down cracking has been observed on I – 79 on the sections under
consideration. Thus, this consumes an important part of the analysis carried out in this
research. Several mechanisms for surface initiated cracking have been proposed. While
summarizing these and investigating other proposed mechanisms, Myers, Roque and
Ruth [12] suggested that top-down cracking may occur due to the transverse stresses at
the tire-pavement interface. They also found that these stresses appeared to not depend on
the pavement structure. However, researchers in Michigan [13] found that a thinner
overlay exhibited greater tensile stresses, though the correlation between thickness and
tensile stresses was not clear. Overall, top-down cracking appears to be controlled by an
interaction of tensile, thermal and shear stresses at the edge of the wheel, though the
effect of temperature gradients in the pavement has not been well understood.

Transverse cracking
Transverse cracking can occur due to thermal stresses or reflective cracking or due to a
combination of both. Since controlling thermal cracking is not always possible owing to
the inherent property of temperature dependent expansion and contraction of HMA, most
research on transverse cracking has been focused on reflective cracking. Reflective
cracking is sometimes also critical in the longitudinal direction. The shear stresses at the
interface of the cracked layer and the overlay, accompanied by the tensile stresses due to
thermal expansion or contraction at these locations are thought to be important for this
type of pavement distress. Reflective cracking in the overlay allows water to percolate
into the pavement structure and weaken the subbase, leading to aggregate stripping,
increased roughness and spalling. Considerable transverse cracking has been observed in
the rehabilitated sections.

Aggregate stripping and moisture susceptibility
In wet and / or frosty conditions, special consideration must be provided to the possibility
of stripping of asphalt binder from the aggregate surface. This is usually caused by high
moisture susceptibility of the mix, but has also been observed where reflective cracking is
rampant. The controlling factors for this are mixture density and binder grade, as results

of the work carried out by Gogula et al. show [14]. It was suggested that a higher binder
grade has intrinsically lower affinity towards water, thereby reducing the potential for

Material properties and their effects
In this section, the effect of variation in the properties of the various components of an
asphalt mixture on the performance and distress resistance will be discussed.

The shape of aggregate is generally expected to affect the mixture properties since high
angularity will provide better interlocking resulting in greater internal friction while flat
and elongated aggregate will provide better resistance to aggregate breakage [15]. Also,
as discussed above, it is to be expected that a coarser mix will provide greater rutting
resistance while a finer mix will provide better fatigue resistance. However, Lee et al [16]
found that aggregate gradation did not have any effect on rutting resistance if the
aggregate meets Superpave angularity requirements. They also suggested, however, that a
gradation containing 37.5 mm aggregate will provide better interlocking than those
containing smaller ones.
       At the same time, however, Kalcheff and Tunnicliff [17] found that incorporation
of fine crushed stone is highly effective in controlling fatigue cracking. It is expected that
adding any kind of crushed fine material will have the same effect.

Binder and Polymer Modifiers
The binder grade is considered to be a controlling factor for both permanent deformation,
which is dominant at higher temperatures as well as for thermal cracking, which
dominates at lower temperatures. Olsen and Byron [18] have previously reported that soft
and elastic binders tend to be more resilient at low temperatures as far as thermal
cracking is concerned. However, using a binder with a performance grade that is good for
thermal cracking may result in higher rutting potential at higher temperatures.

       Concerning the high temperature grade of the asphalt cement, it has been
suggested that the ideal grade is at least one higher than the maximum expected pavement

temperature determined with 98% reliability [1]. In certain cases, this may require
polymer modification of the binder. Besides expanding the applicable temperature range
of a binder, the ability of polymer modifiers to improve resistance to permanent
deformation is well documented, having been tested by a number of researchers [19, 20
and 21]. However, as Newman [22] found after an extensive study, it is difficult to define
a rigid set of rules regarding performance prediction of mixtures containing polymer
modified asphalts (PMAs) and evaluations must be carried out on a case by case basis.
The following polymer modifiers have been used in the past:
       Modified crumb rubber
       Random styrene-butadiene latex polymer (SBR)
       Reactive styrene-butadiene cross-linked block copolymer (SB)
       Linear styrene-butadiene-styrene block copolymer (SBS)
       Modified SBS (MSBS)
       Polychloroprene latex
       Styrene isoprene
       Acrylonite butadiene styrene (ABS)

        It has been observed that asphalt emulsions with SB modifiers without cross
linking generally break out within the first few freeze-thaw cycles, making them
unsuitable for pavements in states like Pennsylvania that experience regular cold and
warm weather. It has also been suggested that such modifiers give off highly toxic fumes
at high temperatures used for mixing and health advisories have been issued in the past
        While Newman was unable to provide any standard recommendations, other
researchers have had some success. As Zaniewski and Nallamothu [24] found, a PG76-22
binder obtained using SBS block copolymer performed much better than unmodified
PG70-22 and PG64-22 binders in West Virginia. Maher [24] compared the performance
of PG76-22 binders obtained using both SB and SBS modifiers in New Jersey and found
SBS modified binders to perform very well compared to SB modified binders. Since both
New Jersey and West Virginia have a climate that is very similar (though not identical) to
that in Pennsylvania, it is expected that these results will hold true in Pennsylvania as

well. The binder being used on the rehabilitation project on I – 79 is the SBS block
        Maher {25} has also tested PG82-22 binders. However, no definitive conclusions
could be drawn and that study broadened into an investigation on the applicability of
traditional rheological properties for PMAs. The investigation, however, succeeded in
drawing conclusion on the effects of various types of modifiers and other additives, as
will be discussed later in this report.

Other additives and their effects on mixture performance
Extensive work has been carried out in some detail using a number of additives and
modifiers other than those discussed above. This section will review some of such work.

Recycled Asphalt Pavement (RAP)
Of late, use of RAP has increased tremendously. This awards both economical and
environmental benefits and efforts are underway to understand the effect of inclusion of
RAP in asphalt mixtures better. Different states have set different limits on the use of
RAP pending definite conclusions [26].
        Since RAP will contain asphalt that has aged and is thus of a grade different from
what it was when constructed, it is imperative to have a knowledge of the contribution of
RAP to the aggregate gradation as well as to the effective binder content and grade. This
is a rather complicated procedure involving the estimation of asphalt content in RAP as
well as the expected extraction of this asphalt into the mix [27]. It has been observed that
inclusion of RAP has a significant effect on the stiffness of the mixture, the magnitude of
which will depend highly on a number of factors, such as:
       Percentage of RAP in mixture,
       Asphalt content in RAP,
       Temperature of RAP at the time of mixing,
       Time of heating of RAP prior to mixing,
       Gradation of RAP when considered as an aggregate stockpile, and
       Intrinsic variability in mechanical properties of the RAP from stockpiles

       As can be expected, owing to so many factors, inclusion of RAP assumes a
challenging yet critical facet. Stephens et al [28], Paterson et al [29] and Thomson [30],
among many others, have devised procedures to estimate the effective PG grade of the
mixtures. The effect of this change in the effective binder grade has not been fully
understood yet. For example, even after an extensive study, Daniel [31] could not trace a
definite correlation between RAP content and mixture stiffness.
       In light of this, DOTs have come up with certain recommendations for use of
RAP in asphalt mixtures. For example, Pennsylvania DOT suggests that for mixtures
containing less than 15% RAP, no special evaluation needs to be one and an assumption
that RAP provides 0.5% of the total asphalt binder in the mix is safe [32]. New
Hampshire DOT, on the other hand, allows up to 15% RAP from unknown sources, but
30% RAP is the source is known and has been analyzed for asphalt content, gradation,
etc. [33]. On the current project, RAP is being used in the binder layer mix, labeled CR1,
at the rate of 10%. No RAP is being used in the surface course mix, labeled A26.

Fillers, Extenders and Other Crushed Particles
There is no universal definition, however, for mineral filler. ASTM D242 defines mineral
filler as: “Mineral filler shall consist of finely divided mineral matter such as rock dust,
slag dust, hydrated lime, hydraulic binder, fly ash, loess, or other suitable mineral
matter.” Presence of crushed particles in a mixture increases the internal friction, thereby
making the mixture more durable and resistant to permanent deformation. In some cases,
the filler action of such particles is considerable, resulting in better mix uniformity,
moisture resistance, etc. A number of materials have been investigated by numerous
researchers. Some of these are discussed below:
Hydrated lime
Hydrated lime has been subject to perhaps the closest scrutiny concerning incorporation
in HMA by various investigators over a long period of time. Hanson et al. [33] studied
the moisture susceptibility of HMA containing lime and obtained excellent results
concerning both moisture susceptibility as well as permanent deformation. Johannson
[34] found hydrated lime to be a very effective anti-aging additive. Lesueur, Little and

Epps [35] found lime to be effective in controlling fatigue fracture as well. Overall,
researchers are agreed that:
      Lime enhances the asphalt – aggregate bond, reduced moisture susceptibility and
       therefore stripping;
      By acting as a filler material (even as little as 1%), it also improves the rutting
       resistance of any mix;
      It alters oxidation kinetics, thereby reducing deleterious effects and improving age
       hardening effects; and
      It improves resistance to fracture growth at low temperatures
   As a result of these, many DOTs such as those in Texas, Nevada, Oregon, Utah,
Georgia and California have already come out with guidelines on use of hydrated lime in
HMA [36]. A number of European countries including Germany and the Czech Republic
also have similar guidelines. The only drawback of hydrated lime that has been observed
is the apparent lime attack on the larger aggregate in a mix.
Bag house fines
The production of asphalt concrete involves several steps at the mixing plant. The
aggregate is first batched and then dried. The aggregate is dried using drums with hot gas
passing over the aggregate to heat the aggregate for mixing as well as remove excess
moisture. During this drying process, small particulate in the aggregate mix becomes
airborne. Collection systems are used to remove the fines from the exhaust stream. These
fines are often reintroduced into the mix in order to compensate for the loss of fines. Due
to the fineness of the material, bag house fines may have an effect on the mixture
       Bag house fines act as fillers and provide a denser mix, rending it more resistant
to permanent deformation. A higher stiffness may also be observed in mixes containing
bag house fines [37]. However, certain drawbacks have been observed. For example, a
greater compaction effort is required in mixes containing a very high percentage of bag
house fines. Anderson [38] found that the softening point and viscosity of asphalt-filler
mastic increases with increase in the fines-to-asphalt ratio. He explained this
phenomenon as the result of lubrication by the filler in the mastic. This may suggest that
a very high content of bag house fines may result in greater rutting potential.

       The optimum content of bag house fines must be determined for each mix, based
on the specific field and mixing plant conditions and mix requirements [39]. Bag house
fines are being used in both the binder and surface layer mixes in the present case at the
rate of 2.2% of the aggregate.
Glass cullets
Airey, Collop and Thom [40] investigated the effect of adding crushed glass in HMA. It
was found that the stiffness and moisture susceptibility of the mixture reduced marginally,
though positive results were obtained as far as aging resistance was concerned. It was
also suspected that some safety issues may crop up once the constructed pavement started
to wear off as glass particles may come loose.
Crumb rubber
As with RAP, addition of crumb rubber to HMA will help control a major source of
permanent environmental pollution. New Jersey DOT funded a study on the technical
usability of crumb rubber in HMA as recently as in 2004. The investigations [41] found
that rubber particles larger than the No. 30 sieve did not provide any benefits. However,
if smaller particles were used in an open graded friction course, such as a 12.5 mm
Superpave mix, appreciable increase in both the high and the low temperature ranges of
the original binder used. This would provide greater rutting resistance as well as
improved low temperature cracking.
       However, a similar study conducted on behalf of Iowa DOT in 2002 [42] reported
that performance enhancement was noted only when rubber chips were used in the
surface course. In this case too, friction values were much higher than in native mixes.
Other materials
Various other materials have been tested with varying degrees of success.
For example, carbon black has been observed to improve both the resistance to
permanent deformation and to aging while sulphur has proven to improve rutting
resistance. Oxidants such as manganese compounds also improve rutting resistance.
Wood lignin, organo-metallics and amines have been shown to improve resistance to
moisture damage. A brief summary of these is available in the NCHRP Report 459 [20].

As with PCC, fibers have also been used in HMA. For example, polypropylene and steel
fibers improve the rutting resistance, fatigue cracking and low temperature cracking. The
cracking mechanism is controlled by providing a pinching force at the micro crack [43]
while rutting is resisted when the fiber behaves like a filler material [20]. It is also
expected, from previous experience [44], that inclusion of fibers will impart better wear
resistance as well, which makes fibers such as cellulose, steel, polyester and
polypropylene fir for inclusion in the surface course also. Fibers from recycled tires
impart similar enhancements in performance, though wear resistance is not improved [45].
        It must also be noted that inclusion of fibers in any mix will adversely impact the
ease with which the mix can be prepared, placed and compacted [43]. While it is simple
enough to improve the workability of fiber reinforced PCC, for HMA, the mixing and
compaction temperatures must be increased. Penn DOT specifies [32] that for a binder
with grade PG76-22 (the grade being used on I-79), the maximum mixing temperature is
1650C. Since the mixing temperature being used on the site is 1600C, there does not seem
to be much scope for inclusion of fibers in these mixes.

Conclusions and Recommendations
In light of the above, it is proposed that for a comprehensive study aimed at material
improvements for the current project, the following should be investigated in addition to
measures already taken:
       Modification of aggregate structure for the surface and binder layers;
       Use of an SBS modified binder in place of the current SB block copolymer
        modified binder;
       Various other percentages of RAP in the binder layer mix;
       Variation in bag house fines content in the mixes;
       Incorporation of hydrated lime or lime treated gravel in the mix; and
       Incorporation of crumb rubber in the surface layer mix.

Laboratory Testing

In Phase I of the project, elementary laboratory testing has been undertaken to determine
basic mixture properties and performance.


Complex Modulus tests were conducted on the mixes presently being used for binder
layer (labeled CR1) and surface layer (labeled A26) to obtain baseline stiffness
information of the mixes as various temperatures and loading frequencies. Limited
accelerated pavement testing using the Model Mobile Load Simulator at 1/3rd scale
(MMLS3) was also undertaken to compare the expected rutting performance of the
control and experimental sections.
       In Phase II, these tests are proposed to be an extension for testing the performance
of modified mixes as discussed in the literature review. The modifications to be tested
will be decided in consultation with the research committee at PennDOT. Advanced
laboratory testing including the Indirect Tension (IDT) test and creep and recovery tests
may also be conducted.

Mixture Types and Properties

Two different mixes are being used in the project for the binder and surface layers. These
mixes differ in aggregate gradation (though both are coarse mixes) and in additives. The
binder layer mix includes 10% recycled asphalt pavement (RAP, and 2.2% bag house
fines have been added to both the mixes. PG76-22 binder obtained using cross-linked SB
block copolymer modifier has also been used for both mixes.

Complex Modulus Testing
The complex modulus, E*, test was conducted on both binder and surface layers. The test
also known as simple performance test yields the dynamic modulus |E*| and phase angle,
φ, that can be used as performance indicators as well as input for more elaborate
performance prediction models.

Test setup and Configuration
The E* testing was carried out on an Instron 22 kip load frame having a 22 kip load cell.
The actuator was controlled using an Instron FastTrack 8800 system. Specimen
deformation was measured using GTX LVDTs having a displacement range of ±2.5 mm.
which were interfaced to a stand-alone computer with an internal National Instruments
data acquisition card. Load data was collected from the FastTrack 8800 console using its
analog output feature for selected data channels. The electronic phase angle between the
FastTrack 8800 console and the data acquisition computer was measured using an elastic
specimen having the same dimensions as an actual asphalt concrete specimen. This phase
was measured at all the temperatures and frequencies that were planned to be applied to
an actual specimen. For actual tests, necessary corrections as obtained above were
applied to the collected data. Based on recommendations of NCHRP Report 465 based on
NCHRP 9-19 projects, tests were conducted at the temperatures and frequencies noted in
Table 1.
      Table 1: Temperatures and frequencies used for Dynamic Modulus tests
                     Temperature (0C)            Frequencies (Hz)
                              0                  25, 10, 5, 1, 0.5, 0.1
                             10                  25, 10, 5, 1, 0.5, 0.1
                             20                  25, 10, 5, 1, 0.5, 0.1
                             35                  25, 10, 5, 1, 0.5, 0.1

The LVDTs were mounted on the specimen using metal brackets that were glued to the
specimen using epoxy. A gauge length of 75 mm. was used. A template was used to
mount these brackets to ensure that the 75 mm. gauge lay in the middle of the specimen.
Three such LVDTs were mounted on the specimen at 1200 to each other, as shown in

Figure 2. The output of the three LVDTs was averaged to obtain a single reading.
       One specimen from the same mix design and air void content as that to be tested
was used to obtain the ideal PID settings. The same specimen was used to measure the
load required to be applied at the various temperatures and frequencies in order to obtain
an average strain of 70 microstrains in the gauge length. The temperature of the specimen
was estimated using a dummy specimen of the same mix with a thermocouple embedded
at its centre. An environmental chamber was used to maintain temperature in conjunction
with a temperature microcontroller.

     Figure 2: LVDT mounting arrangement: Front view (left); top view (right)

Specimen Geometry and Fabrication

Recommendations of NCHRP 9-19 (Final Report No. 465) were used to prepare
specimens for the Dynamic Modulus test. Specimens of 150 mm. diameter and 180 mm.
height were compacted in a Superpave gyratory compactor. These were cored to obtain
100 mm. diameter cores. A dry saw was used to cut 15 mm. off the top and bottom of the
obtained cores. Specimens having a diameter of 100 mm. and a height of 150 mm. were

thus obtained, as shown in Figure 3. The InstroTekR CoreLok machine was used to
determine the specific gravity (Gmb) of the specimens and thereafter their air void content
Specimens which had air voids less than 3.5% or more than 4.5% were rejected. The
remaining specimens were sealed in plastic bags and stored at 100C until testing. Tests
were conducted on three replicates to ensure repeatability.

    Figure 3: Obtaining a Dynamic Modulus specimen from Superpave gyratory
                                              compacted (SGC) specimens

Results and Analysis
The stress and strain data obtained during the test was used to compute the Dynamic
Modulus of the specimens at each frequency of loading as follows:
E*                                                                                   ...1
where |E*| is the dynamic modulus, σ0 is peak stress applied on the specimen, ε0 is peak
strain in the specimen over the gauge length (=75 mm.) due to peak stress. Once data was
available for all frequencies and temperatures, the sigmoidal function was used to create a
mastercurve for each of the specimens. The sigmoidal function relates the logarithmic
values of dynamic modulus and reduced time as follows:

log  E *   a1 
                            a3  a4 log                                            ... 1
                     1 e
where a1, a2, a3 and a4 are constants, and ν is reduced frequency. The mastercurves
obtained for the binder layer and well as surface layer mixes is shown in Figure 4 below.


      Dynamic Modulus (MPa)   20000



                                                                        Binder layer mix (CR1)
                                                                        Surface layer mix (A26)
                                 0.001   0.01   0.1       1        10      100    1000   10000 100000
                                                      Reduced frequency (Hz)

  Figure 4: Dynamic modulus mastercurves for the binder layer and surface layer
             It is apparent from the figure that the mastercurve for the binder layer mix spans a
smaller range of reduced frequency. A similar phenomenon was noticed by Daniel [31],
in a study aimed at quantifying the effect of varying percentages of RAP in the mix.
However, no definite trend could be observed. Considering that the binder layer mixture
contains 10% RAP, it is possible that the observed “compactness” is the result of addition
of RAP to the mix. More testing is required to confirm this. Such testing would include,
but not be limited to, obtaining the Dynamic Modulus mastercurve for a mix having the
same aggregate structure and binder content as CR1 but lacking in RAP.
             An extended projection of the mastercurves using the sigmoidal function
mentioned above (Figure 5) shows that the CR1 mix attains an asymptotic characteristic
at a much lower reduced frequency (which corresponds to a higher time of loading and /
or higher temperature) than A26, which is the surface layer mix. This might imply that
the CR1 mix may be more prone to fatigue cracking than the A26 mix. Further, it is noted
that this divergence occurs at a reduced frequency of around 10,000 Hz. This observation
may be of some importance, since for a typical loading frequency of 10 Hz, this reduced

frequency would correspond to a pavement temperature of around -100C. This can be
computed by calculating the shift factor based on the equations mentioned below and
then reading the applicable temperature off the chart shown in Figure 6.
aT                                                                                               ... 2
where aT is the shift factor, ν is reduced frequency, and μ is actual frequency. It is quite
reasonable to expect a pavement temperature of -100C or lower during the colder months
of the year. Therefore, this characteristic needs to be investigated further.


           Dynamic Modulus (MPa)






                                   5000                              Binder layer mix (CR1)
                                                                     Surface layer mix (A26)
                                      0.00001   0.01   10    10000   1000000 1E+10   1E+13     1E+16
                                                       Reduced Frequency (Hz)

   Figure 5: Mastercurve projections based on sigmoidal fits on a semi-log graph

                                          y = 0.0007x 2 - 0.1432x + 2.6369
                            4                       R2 = 0.9996

                                                                                   Shift factor CR1
       Log (Shift Factor)

                                                                                   Shift factor A26
                            1                                                      Poly. (Shift factor A26)
                                                                                   Poly. (Shift factor CR1)

                            -1         y = 0.0005x 2 - 0.103x + 1.8582
                                                    R2 = 1


                                 -10          0        10       20            30    40       50      60       70

                                                              Temprature (deg C)

                                             Figure 6: Shift factor curves for both mixes

            On the other end of the mastercurve, it is noted that CR1 attains an asymptotic
characteristic with a higher stiffness than A26 at a reduced frequency of around 0.001 Hz
(Figure 7). While a higher stiffness may suggest better rut resistance, the extremely low
reduced frequency corresponds to a pavement temperature even higher than 600C,
computed using the same technique as discussed above. To deduce if this might be
crucial, the average 7-day high pavement temperature for the site was computed. It was
found that for three stations surrounding the site at Zelienople, PA, the 7-day average
high pavement temperature was not expected to exceed 550C (Table 2) even at 98%
reliability. Thus, the observed divergence may not be a critical difference in mechanical
properties of the two mixes. However, it must be kept in mind that high temperature has
the same effect as longer loading times, which corresponds to slower moving traffic. This
may cause a significant difference in the rut resistance of the two mixes. Further, the
temperature of the binder layer itself would be lesser than that measured at a depth of 25
mm, since the surface layer itself is 1.5 in. (38.1 mm.).


       Dynamic Modulus (MPa)



                                                                      Binder layer mix (CR1)
                                                                      Surface layer mix (A26)
                                   1E-11   1E-07   0.001     10      100000   1E+09   1E+13     1E+17

                                                    Reduced Frequency (Hz)

    Figure 7: Mastercurve projections based on sigmoidal fits on a log-log graph

Table 2: 7-day average high pavement temperature (at 25 mm. below surface) using
                                                   LTPPBind software
Station ID                          Station    Distance (mi) Temperature (0C) against reliability

                                                                  50% (z = 0) 98% (z = 2) 67% (z = 1)

0361130                          Butler        18.7 NE               48.6         55.1            51.8

0366997                          Pittsburgh city 24.9 SSE            47.9         54.4            51.1

0366233                          New Castle    30.7 NNW              47.8         54.4            51.1

Accelerated Pavement Testing

Accelerated pavement testing adds another dimension to the evaluation of overlay
thicknesses. The MMLS3, Mobile Model Load Simulator 3rd scale, is an accelerated

loading tester that applies wheel trafficking on pavement structures. Accelerated
trafficking would thus be used for the two overlay structures to compare the performance
in terms of rutting and cracking.

Equipment and Setup
Accelerated testing can be conducted at a variety of conditions, including:
Slab vs. briquette setup, thickness, trafficking speed and temperature, wet vs. dry testing,
wandering vs. channelized trafficking, tire pressure and load, among others. A
description of testing setup and conditions is discussed below.

MMLS3 System
MMLS3 Load Simulator
The MMLS3, shown in Figure 8, is a pavement tester that applies loading in terms of
accelerated wheel trafficking in a compressed-time frame. The trafficking device
simulates actual vehicle loading by applying the same contact tire pressure on the
roadway pavement being tested. The speed of trafficking is variable up to a maximum of
two wheel applications per second. Max wheel load is 2.7 kN on the 300mm diameter
pneumatic tire wheels. Maximum tire pressure is 700 kPa. The machine has a lateral
"wandering" system which displaces the machine up to a maximum of about 80mm to
each side off the center line of the track. To achieve a Normal (Raleigh) distribution of
the wheel loads, the machine spends more time near the centerline of the track than at the
edges. The MMLS3 comes with ancillary equipment for used on the field including a
transverse profilometer, environmental conditioning system, and pavement wetting
       An electronic transverse profilometer is used in conjunction with the MMLS3 to
measure surface profile. The MMLS3 is snugly fitted into a custom built temperature
chamber. A combined dry heating/cooling unit allows for hot or cold air to be forced
over the pavement surface to raise or lower the temperature. This allows researchers to
work in an air conditioned environment around the MMLS3 while it is trafficking the
pavement at elevated or low temperatures while on the field. The maximum operating

temperature that can be reached is 60°C and the minimum is -5°C, or lower under certain
conditions. A water unit that saturates the pavement surface allows for wet testing.
Test Setup
Accelerated testing using the MMLS3 can be conducted using three setups:
      Field Setup: The mobile MMLS3 can be transferred to the field for testing of
       actual field pavements.
      Laboratory slab setup: The MMLS3 applies trafficking to slabs that are
       compacted in the laboratory to the required thickness and density.
      Briquette setup: In this setup, nine cores or cylindrical specimens are trimmed and
       then aligned in series to form a wheel path. They are locked into position using a
       customized test bed (Figure 9).
The testing conducted so far has been done using a briquette setup for both surface and
surface + binder specimens. As seen in Figure 9, the briquette test setup accommodates
nine specimens aligned in series. To eliminate any effect of impact loading due to the
touch down of the MMLS3 wheel on the first of the nine specimens, only the inner seven
specimens will be considered. The first and last specimens were “dummy specimens”.
The first briquette from the right had a thermocouple inserted into its center to measure
the actual temperature inside the briquettes during loading. The next three briquettes
consisted of surface course mixture only. The next briquette was a dummy briquette,
followed by three “composite” briquettes consisting of a surface course briquette placed
on top of a binder course briquette.
Specimen Thickness and Air Void Content
The procedure for determining the thicknesses and AV content of briquettes has been
discussed in a separate section in the Appendix.
       In the experimental section on I-79 where a binder layer is being incorporated, the
thickness of the binder layer is being maintained at 2 inches while that of the surface
layer is being maintained at 1.5 inches. Thus, the total thickness of specimens should be
3.5 inches. However, it is only possible to test specimens that are a maximum of 3 inches
in height on the MMLS3 test bed. In view of this, it was decided to scale down the
specimens so as to obtain a final specimen of 3 inches thickness. It was thus decided to
prepare test briquettes by placing a 36 mm. briquette fabricated according to the surface

course mix design (A26) on top of a 48 mm. briquette fabricated according to the binder
course mix design (CR1) and fixing them together with tack coat (Figure 10).
Testing Temperature
LTPPBind software (Version 2.1) was used to determine the maximum pavement
temperature at a depth of 25 mm. from the surface with 98% reliability (z = 2). Three
stations around the site (Zelienople, PA) were chosen as shown in Table 2. The high
pavement temperature interpolated for the site, ~55ºC, would be considered the critical
temperature for conducting rutting tests for such a location. Since it has been found that
negligible rutting is observed if the temperature at the time of loading is less than 400C
[1], the research team decided to first conduct a baseline test at a temperature at 400C. An
environmental chamber was used to maintain the standard temperature while the
briquettes were being trafficked (Figure 11).
Tire Pressure and Tire Load
Since the purpose of the present study was to compare two possible pavement structures,
standard values of tire load and tire pressure were used. For the MMLS3, a third scale
model is assumed, which requires a tire pressure equal to actual inflation pressure and
was used as 600 kPa. A tire load of 2.7 kN was used. It may be noted that the actual
vertical contact stress may be different from the tire inflation pressure. Epps, Ahmed,
Little and Hugo [2] found that for a tire inflation pressure of 600 kPa, the contact stress
would be closer to 650 kPa.

Figure 8 MMLS3 simulator lowered onto briquette test bed

         Surface + Binder            Surface

       Figure 9 Briquette setup for MMLS3 testing

 Figure 10 A cross-section of a composite surface + binder briquette composed of a
       surface course briquette fixed atop of a base course one using tack coat

    Figure 11 Environmental chamber housing the MMLS3 and briquette mold.

Results and Analysis
Figure 12 shows a typical series of profiles of a briquette after being trafficked by
MMLS3. Both rutting and shoving are marked in this figure. Figure 13 shows the rutting

curves for the surface-only briquettes while Figure 14 shows the rutting curves for the
binder + surface briquettes. Figure 15 shows a comparison of the rutting curves.

                                                                                                    0              16000         32000        64000
                                                         20.5                                       128000         256000        512000       1024000
        Surface profile of briquette (mm)





                                                                                     -60            -40         -20        0         20        40        60
                                                                                                     Distance from centreline of wheel path (mm)

                                                                                                0                 16000          32000        64000
                                                                                     20         128000            256000         512000       1024000
                                            Surface profile of briquette (mm)

                                                                                                               Wheel path






                                                                                          -60           -40       -20        0           20    40       60
                                                                                                Distance from centreline of wheel path (mm)

Figure 12: A typical profile of a trafficked briquette for binder + surface (Top); and
                                                                                                              surface only (Bottom)




Rutting (mm)



                                                                            Replicate 1
                                                                            Replicate 2
               0.2                                                          Replicate 3
               0.1                                                          Average

                 100          1000          10000        100000        1000000              10000000

                                               No. of axles

                         Figure 13: Rutting curve for surface-only briquettes



Rutting (mm)




               0.3                                                                Replicate 1
               0.2                                                                Replicate 2
                                                                                  Replicate 3
                 100          1000          10000        100000        1000000              10000000

                                               No. of axles

                       Figure 14: Rutting curve for surface + binder briquettes




   Rutting (mm)





                                                                               Surface only
                                                                               Binder + Surface
                    100            1000          10000          100000         1000000        10000000

                                          No. of MMLS3 Wheel Applications

   Figure 15: Comparison of rutting curves for surface-only and surface + binder

                   A typical rutting curve shows a steep densification phase during the early part of
loading and then a nearly linear part for a considerable period of time thereafter. It is
observed from Figure 15 that in this phase, the rutting in the binder + surface briquettes
was nearly the same as that in the surface-only briquettes but consistently higher as well.
A statistical analysis using the ANOVA method revealed that the data from the two sets
in this phase was statistically the same and therefore the variations are attributed to the
variability between replicates within each case (P=0.58). The consistently higher rutting
in the binder + surface briquettes can possibly be attributed to greater densification of
these briquettes simply owing to the greater volume of mix that can be compacted. It is
also possible that this phenomenon occurs due to the higher AV content in surface course
                   In order to draw conclusions regarding densification, the AV content of the
briquettes was measured both before and after trafficking. Table 3 summarizes this
analysis. It is noted from this table that the binder + surface briquettes are compacting
marginally more than the surface-only briquettes. Further, the final AV content on both
kinds of briquettes is practically the same. This reinforces the author’s hypothesis that the

higher rutting observed in the binder + surface briquettes may be due to greater
densification on these specimens. In future, therefore, it is proposed to fabricate both
kinds of briquettes with the same AV content to better illustrate the rutting in the mixes.
From the present knowledge, however, both types of pavement structures can be expected
to rut almost identically.
                Table 3: Air void content and densification of briquettes
  Test         Mix                    Initial AV                   Final AV          Total
bed ID                       Gmb        Gmbeff       AVeff (%)      AV (%)         change
             Binder          2.409
 B+S 1                                   2.382          5.7           4.1             1.6
             Surface         2.346
             Binder          2.404
 B+S 2                                   2.376          6.0           4.7             1.3
             Surface         2.339
             Binder          2.405
 B+S 3                                   2.381          5.7           3.6             2.1
             Surface         2.350
  S1         Surface         2.350       2.486          5.5           3.5             2.0
  S2         Surface         2.337       2.486          6.0           4.8             1.2
  S3         Surface         2.343       2.486          5.7           4.4             1.3

         However, the binder + surface briquettes did exhibit a tertiary phase starting at
256, 000 cycles with a rapid increase in rutting while the surface–only ones did not
exhibit a tertiary phase and rutting increased only slightly after 256, 000 cycles. It is
suspected that this occurs not due to compressive failure of the mixes but due to shear
effects, which are expected to be more when their exists an additional flexible layer
directly below the surface layer. A Linear Elastic Analysis using Kenlayer was performed
in order to better understand this phenomenon.

Design Evaluation

An analytical approach to evaluating the two pavement structures was also undertaken, as
discussed above. A linear elastic analysis was expected to provide some more insight into

the mechanism that led to more rutting in the binder + surface course briquettes. The
analysis was undertaken in order to quantify the mechanical performance of the mixes
and their performance in the structure.
       Besides comparison of mechanical behavior, it was planned that the results from
the linear elastic analysis would be used in performance prediction models to compare
the actual performance of the structures. While this analysis would not be exact owing to
various simplifications and assumptions, it would nonetheless provide a foundation upon
which further studies could be planned.

Linear Elastic Analysis using Kenlayer

Structures Studied
The pavement structures discussed previously and shown in Figure 1 were evaluated in
this phase of the study. The leveling course was not modeled as a separate layer since it
had a variable thickness across the pavement cross section and was not expected to
contribute to the structural performance of the pavement sections.

The linear elastic analysis program Kenlayer requires input of data relating to the layer
thicknesses, elastic moduli, Poisson’s ratio, contact stress and contact radius of the load.
Structure information such as bonding between layers can also be specified. The program
returns stress and strain values at locations requested by the user. The locations at which
these values were requested are shown in Figure 16.


                                       600 kPa
                              H              A
                              I              B                   Surface Layer
                              J              C
                              K              D                  Binder Layer
                              L              E
G. SUBBASE                    M              F                  Existing HMA
                              N              G                  Layer

                                                                 Existing PCC

 BGRADE                                                          Subgrade

                   Figure 16: Locations at which Kenlayer output was obtained
              The layer properties used for this analysis are listed in Table 4. E* data derived in
      the laboratory previously (Figure 4) was used to determine the applicable elastic modulus
      for the layers. It was assumed that the traffic moved at 10 Hz at 200C. The Poisson’s ratio
      was assumed as 0.35 based on the recommendations of the M-E Design Guide [3], which
      suggests that this value be used for conventional HMA when operating in the 40 – 70 F
      temperature range. Elastic properties for the old asphalt and concrete were assumed based
      on recommendation by Huang [4].
         Table 4: Material properties inputted into Kenlayer for linear elastic analysis
          Layer             Depth            Elastic Modulus (MPa)           Poisson's Ratio
        Surface        1.5 in (38.1 mm)                 8367                        0.35
        Binder         2.5 in (63.5 mm)                 7618                        0.35
        Asphalt        2.5 in (63.5 mm)                 6500                        0.35
        Concrete       10 in (254 mm)                   24000                       0.15
        Sub-grade           Infinity                     103                        0.45

Analysis Results and Interpretations
Typical pavement distresses such as permanent deformation and fatigue cracking were
assessed. Based on strain values obtained from the above analysis were used in pavement
prediction models presented in the M-E Design Guide in order to compare failure
resistance of the two structures.
The Mechanistic – Empirical Design Guide presents the following model for prediction
of permanent deformation:
     r1a1 N a  T a 
                2 2       3 3
                                                                                           ... 3

where: εp is permanent deformation, εr is resilient strain, i.e., recoverable strain, N is
number of load cycles, T is temperature in 0F, and a, β are statistical constants. N was
extrapolated from the available data using the compound growth model provided in the
design guide:
           GR 
N t  N 0 1                                                                             ... 4
           100 

where Nt is traffic t years after base year, N0 is traffic in base year, and GR is percent
growth rate of traffic.
        It is apparent that for a given mix, the permanent strain and the recoverable strain
are directly proportional at all levels of traffic, i.e., N. In an elastic analysis, the entire
strain that appears will be recoverable; hence the obtained compressive strain may be
used in place of the resilient strain. Based on a national calibration of this model, the
following form has been recommended:
    k1 *10 3.4488 * T 1.5606 * N 0.479244                                                ... 5

The constant k1 has been introduced to take into account the effect of different thickness
of asphalt layers and is given as:

k1  (C1  C 2 * d ) * 0.328196                      d
                                                                                                        ... 6

where d is depth to computational point in inches and C1 and C2 are given by:
C1  0.1039h 2  2.4868h  17.342
                                                                                                        ... 7
C 2  0.0172h 2  1.7331h  27.428

where h is the total thickness of asphalt layers, in inches. Figure 17 shows the obtained
rutting progression as per the linear elastic analysis and above mentioned model.
       The linear elastic analysis shows that with the inclusion of a binder layer, the
compressive vertical stress as well as vertical strain is much higher just under the wheel
but more or less the same at other depths. The increase in vertical strain is more
pronounced at the edge of the wheel, where shear effects also come into play. This might
imply that any difference in permanent deformation, if observed, would be due to shear
effects and not compressive effects. When the shear stress (and shear strain) is observed,
it is noted that the values are lower just under the wheel but considerably higher at greater
depths up till the middle of the old asphalt layer. It would therefore be reasonable to
expect the overlay containing binder layer to exhibit more rutting. This is also what was
observed in the MMLS3 runs.

            Permanent deformation (in)





                                         0.020                              Without binder layer
                                                                            With binder layer
                                                 0   5   10   15      20        25       30        35

                                                              Age (years)

 Figure 17: Growth of permanent deformation based on linear elastic analysis and
                          nationally calibrated model

Fatigue cracking
The fatigue cracking model provided in the design guide takes the form of a
phenomenological crack prediction model, similar to the one discussed for permanent
                         1               
N f  0.00432 * k1 *10 * 
                                                      *                                       ... 8
                          t                           E

where: Nf is number of load cycles till fatigue failure, εt is elastic tensile strain at critical
point, E is strength modulus (dynamic modulus is used for cyclic loading), and M, k1 are
constants given as follows:
         Vb           
         V  V  0.69 
M  4.84                                                                                       ... 9
         a    b       
where Va is percent air void content and Vb is percent effective binder content. Va was
taken as 4.1% based on the compacted air voids after trafficking (Table 3). Vb was taken
as 5.2% based in information from the mix design.
k1                                                                                              ... 10
       0.000398 
                  1  e (11.023.49 h )

where h is the total thickness of asphalt layers in inches.
         Since fatigue cracking is associated with tensile failure, the tensile strain is used
for εt. The results show that the tensile strains are significantly lower in the pavement in
case a binder layer is introduced at each of the points A, B and C due to the increased
thickness (Table 5)Error! Reference source not found.Error! Reference source not
found.. This implies that inclusion of a binder layer will improve resistance to fatigue
cracking. This is a critical conclusion, since in colder areas like Pennsylvania fatigue
cracking is generally more dominant than rutting.
Table 5: Phenomenological model output for fatigue cracking based on LEA results
                                   M               h (in)             k1   εt (microstrain)      Nf
Without binder layer             -0.6334               4.0      261.8391        11.70         8.85E+14
With binder layer                -0.6334               6.0      250.0111        8.16          3.50E+15

Top-down cracking
The model as described in equation 17 above can be used for top-down cracking as well,
with two modifications. First, the critical strain is taken to be at the surface and on the
edge of the wheel path, as discussed earlier. Second, k1 takes the following form:
k1                                                                                    ... 11
       0.01 
                1  e (51.676 2.8186 h )

It is to be noted that top-down cracking is a complex phenomenon and this model is a
highly simplified one. For example, it does not account for thermal stresses, which are
considered important for surface initiated cracks. Further, such stresses are expected to be
independent of the pavement structures, since thermal strains will play a critical role only
at the surface of the pavement. It is observed from the linear elastic analysis that the
tensile stress at point H (just under the wheel but at the edge) as well as the radial stress
at point A (just under centerline of wheel path) exhibit negligible change with inclusion
of binder layer.
Reflective cracking
On the other hand, the shear stress (and shear strain) is slightly larger at the interfaces in
case a binder layer is included. However, the tensile stress is significantly lower while the
compressive stress does not change. A deeper analysis is therefore required to draw
conclusions on reflective cracking potential in the two structures.

Table 6 provides detailed tabulation of the critical stresses and strains in the analyzed
pavements and their effect on distresses.

                     Table 6: Effect of inclusion of binder layer on various pavement distresses
              Critical stress /            Surface-       Binder +
Distress                           Point                                      % change             Comments
                   strain                     only         Surface
                                    H      1.42E+01       1.24E+01              -12.7    Greater shear stresses through
               Shear stress at
                                    I      2.39E+01       2.48E+01               3.8       the overlay in case of binder
               edge of wheel
                                    J      2.21E+01       2.35E+01               6.3        layer. Negligible change in
                                    A      1.22E+02       1.22E+02               0.0      compressive stresses implies
                                                                                         shear stresses may control, but
             stress under wheel     B      8.72E+01       8.71E+01              -0.1           more testing required.
                                    B      1.67E-05       1.52E-05               -9.0       Significant improvement in
              Tensile strain at
 Fatigue                            C      1.17E-05       8.16E-06              -30.3     resistance to fatigue cracking
              bottom of layer
                                    G      1.05E-05       5.56E-06              -47.0                expected
              Tensile stress at
                                    H      4.51E+01       4.76E+01              5.5       Negligible change in critical
Top-down        edge of wheel
                                                                                            stresses. No observable
cracking         Radial stress
                                    A      1.04E+02       1.06E+02              1.9           difference expected.
                 under wheel
               Shear stress at                                                               Greater shear stress but
                                     J     2.21E+01       2.35E+01              6.3
                  interface                                                                significantly lower tensile
              Tensile stress at     C      6.29E+01       5.79E+01               -7.9    stresses. Negligible change in
               bottom of layer      J      3.70E+01       3.33E+01              -10.0     compressive stress. Possible
                Compressive         C      8.72E+01       8.71E+01               -0.1    improvement in resistance to
             stress at bottom of                                                         reflective cracking, but more
                                     J     4.01E+01       4.02E+01              0.2            analysis required.
                                             Note: All stress values in psi

1. Buchner M, Newcomb D and Huddleston J, “Perpetual Pavements”, Asphalt, The
   Magazine of the National Asphalt Institute, Vol. 15, #3, Fall 2000
2. Epps A, Harvey J T, Kim, Y R, Roque R, “Structural Requirements of Bituminous
   Paving Mixtures”,
3. Decker D S, Goodrich J L, “Asphalt Cement Related Pavement Performance”,
   Proceedings of the Association of Asphalt Paving Technologists, Vol. 58, pp. 503 –
   518, 1989
4. Brown E R, Cross S A, “A study of In-place Rutting of Asphalt Pavement”,
   Proceedings of the Association of Asphalt Paving Technologists, Vol. 58, pp. 1 – 39,
5. Lai J S, Lee T M, “Use of Loaded Wheel Testing Machine to Evaluate Rutting of
   Asphalt Mixes”, Transportation Research Board 1269, National Research Council,
   Washington D.C, pp. 116 – 124, 1990
6. McGennis B R, Anderson R M, Kennedy T W, Solaimanian M, “Background of
   Superpave Asphalt Mixture Design and Analysis”, National Asphalt Training Centre
   Demonstration Project 101, Publication No. FHWA-SA-95-003, February 1995
7. Sousa J B, “Asphalt-Aggregate Mix Design Using Simple Shear Test (Constant
   Height)”, Proceedings of the Association of Asphalt Paving Technologists, Vol. 63,
   pp. 298 – 345, 1994
8. Mallick R B, Ahlrich R, Brown E B, “Potential of Dynamic Creep to Predict Rutting”,
   Engineering Properties of Asphalt Mixtures and Relationship to their Performance,
   ASTM STP1265, American Society for Testing of Materials, Philadelphia 1995
9. Roberts F L, Kandhal P S, Brown R E, Kennedy T W, Lee D Y, “Hot Mix Asphalt
   Materials, Mixture Design and Construction”, Second Edition, NAPA Educational
   Foundation, Lanham, Maryland, 1996
10. Kim Y R, Lee H J, Little D, “Fatigue Characterization of Asphalt Concrete using
   Viscoelastic and Continuum Damage Theory”, Journal of the Association of Asphalt
   Paving Technologists, Vol. 66, pp. 520 – 569, 1997

11. “Fatigue Response of Asphalt-Aggregate Mixes”, SHRP-A-404, Strategic Highway
   Research Program, National Research Council, Washington D.C., 1994
12. Myers, L.A., R. Roque and B.E. Ruth, “Mechanisms of Surface-Initiated Wheel Path
   Cracks in High-Type Bituminous Pavements”, Journal of the Association of Asphalt
   Paving Technologists, Vol. 67, pp. 401 – 420, 1998
13. Svasdisant, T., M. Schorsch, G.Y. Baladi and S. Pinyosunun, “Mechanistic Analysis
   of Top-Down Cracks in Asphalt Pavements”, Transportation Research Board, 81st
   Annual Meeting, January 2002, Washington, D.C.
14. Gogula A, Boyer J, Romanoschi S, “Effect of PG Binder Grade and Source of
   Superpave Mixtures under Hamburg Wheel Tester”, Proceedings of the 2003 Mid-
   Continent Transportation Research Symposium, Ames, Iowa, 2003
15. “Superpave Mix Design”, Asphalt Institute Superpave Series No. 2 (SP-2), 1996
16. Lee H J, Choi J Y, Zhao Y, Kim Y R, “Laboratory Evaluation of the Effects of
   Aggregate Gradation and Binder Type on Performance of Asphalt Mixtures”,
   International Journal of Pavement Engineering, Vol. 1 No. 2, April 2000
17. Kalcheff I V, Tunnicliff D G, “Effects of Crushed Stone Aggregate Shape and Size
   on Properties of Asphalt Concrete”, Proceedings of the Association of Asphalt Paving
   Technologists, Vol. 51, pp. 453 – 483, 1982
18. Olsen G K, Ruth B E, “Creep Effects on Fatigue Testing of Asphalt Concrete”,
   Proceedings of the Association of Asphalt Paving Technologists, Vol. 46, pp. 176 –
   195, 1977
19. Institute for Transportation Studies, University of California at Berkeley “Permanent
   Deformation Response of Asphalt - Aggregate Mixes”, SHRP-A-415, 1994
20. Bahia H U, Hanson D I, Zeng M, Zhia H, Khatri M A and Anderson R M,
   “Characterization of Modified Asphalt Binders in Superpave Mix Design”, NCHRP
   Report 459, 2001
21. Freeman R B, Newman J K and Ahlrich R A, “Effect of Polymer Modifiers on Dense
   Graded, Heavy-Duty Mixtures”, 8th International Conference on Asphalt Pavements,
   Seattle, Washington, 1997
22. Newman K, “Polymer Modified Asphalt Mixtures for Heavy-Duty Pavements:
   Fatigue Characteristics as Measured by Flexural Beam Testing”, 2004 FAA

   Worldwide Airport Technology Transfer Conference, Atlantic City, New Jersey,
   April 2004
23. “Paving with Hot Mix Asphalt (HMA)”, Safety bulletin No. SB-03-3, New York
   State Department of Transportation, New York, February 2003
24. Zaniewski J P, Nallamothu S H, “Evaluation of Binder Grades on Rutting
   Performance”, Final Report for West Virginia Division of Highways and Federal
   Highway Administration, Asphalt Technology Program, Department of Civil and
   Environmental Engineering, Morgantown, West Virginia, June 2003
25. Maher A, “Investigation into Modified Asphalt Binders for Improved Pavement
   Performance”, Final Report for New Jersey Department of Transportation Project No.
   80-RU908, Center for Advanced Infrastructure and Transportation, The State
   University of New Jersey, Rutgers, New Jersey, January 2003
26. Abdulshafi O, Fitch M, Kedzierski B, “Determination of Recycled Asphalt Pavement
   Content in Asphalt Mixes Based on Expected Mixture Durability”, Final Report for
   FHWA Project No. FHWA A/OH-2002/037, Ohio State University, Ohio, 2002
27. Zofka A, et al., “Development of a Simple Asphalt Test for Determination of RAP
   Blending Charts”, Minnesota Department of Transportation Report MN/RC-2004-44,
28. Stephens J E, Mahoney J, Dippold C, “Determination of the PG Binder Grade to Use
   in a RAP Mix”, Connecticut Transportation Institute and University of Connecticut
   Final Report for Joint Highway Research Advisory Council Project JHR 00-278,
29. Peterson R L, Soleymani H R, Anderson R M, McDaniel R S, “Recovery and Testing
   of RAP Binders from Recycled Asphalt Pavements”, Journal of The Association of
   Asphalt Paving Technologists, Vol. 69, pp. 72 – 91, 2000
30. Thomson G, “Determining Asphalt Content for Recycled Pavement Materials”,
   Oregon Department of Transportation and Federal Highway Administration Report
   FHWA-OR-DF-04-02, 2003
31. Daniel J S,
32. “Superpave Mixture Design, Standard and RPS Construction of Plant-Mixed HMA
   Courses”, Pennsylvania Department of Transportation, Publication 27, Section 409

33. Hanson D I, Graves R E, Brown E R, “Laboratory Evaluation of the Addition of Lime
   Treated Sand to Hot Mix Asphalt”, National Center for Asphalt Technology, Auburn
   University, 1993
34. Johannson L, “Bitumen Aging and Hydrated Lime”, Ph.D. dissertation, Kungl
   Tekniska Hogkolan, Royal Institute of Technology, 1998
35. Lesueur D, Little D N, Epps J A, “Effect of Hydrated Lime on Rheology, Fracture
   and Aging of Bitumen and Asphalt Mixtures”, Lhoist HMA Symposium, Dusseldorf,
   Germany, June 1998
36. Little D N, Epps J A, “The Benefits of Hydrated Lime in Hot Mix Asphalt”, Final
   Report for National Lime Association, 2001
37. Tayebali A.A., M.B. Kulkarni and H.F. Waller, “Delamination and Shoving of
   Asphalt Concrete Layers Containing Baghouse Fines”, Department of Civil
   Engineering, North Carolina State University, Technical Report, May 2000.
38. Anderson D.A., J.P Tarris and J.D. Brock, “Dust Collector Fines and Their Influence
   on Mixture Design”, Proceedings of the Association of Asphalt Paving Technologists,
   Vol. 51, 1982, pp. 363 – 397
39. Fischer W.K., “Effect of Percentage Baghouse Fines on the Amount of Antistripping
   Agent Required to Control Moisture Sensitivity”, Master of Science Thesis, North
   Carolina State University, 2002
40. Airey G D, Collop A C, Thom N H, “Mechanical Properties of Asphalt Mixtures
   Incorporating Slag and Glass Secondary Aggregates”, Proceedings of the 8th
   Conference on Asphalt Pavements for Southern Africa (CAPSA), Sun City, South
   Africa, September 2004
41. Bennert T, Maher A, Smith J, “Evaluation of Crumb Rubber in Hot Mix Asphalt”,
   Final Report for USDOT Research and Special Programs Administration by Center
   for Advanced Infrastructure and Transportation (CAIT), Rutgers University,
   Piscataway, New Jersey, July 2004
42. Engle E, et al., “Evaluation of Recycled Rubber in Asphalt Cement Concrete – Field
   Testing”, Final Report for HR-330, 330A, 330B, 330C and 330D sections, Materials
   Research, Highway Division, Iowa Department of Transportation, Iowa, October


Obtaining Target AV Under the Wheel Path

Determination of Effective Wheel Path Area
A few dummy specimens were prepared and these were subjected to MMLS3 trafficking.
The profiles of these were studied after 5,000 wheel applications at the rate of 2 wheel
applications per second (48 Hz setting on the MMLS3 controller). It was found that the
total rutting effect of the wheels extended roughly 40 mm. on either side of the centre line.
This is shown in Figure 18. A margin of 3 mm. was decided upon to include the area not
directly under the wheel but which may still affect rutting. Thus, the portion of interest in
a briquette was determined to be a cuboid of size 105 x 86 mm, as observed in Figure 19.

                        Figure 18: Typical briquette dimensions

                    Figure 19 Effective wheel path area of a briquette

Determination of Mix Weight
To determine what mix weight should be used to obtain the target air voids in the portion
of interest, the following steps were performed: From the contractor’s mix design data, it
was found that a 115 mm. SGC specimen made with 4773.2 gms of mix contained 4.2%
overall air voids. This was used to project the mix weight required to obtain a 130 mm.
SGC specimen containing 4.2% overall air voids. An analytical procedure was used to
estimate the relationship between AV in SGC specimens and the portion of interest as
A p * AV p  Ar * AV r  At * AVt                                             ... 12

Ap is area under wheel path; AVp is AV in wheel path; Ar is area of specimen other than
the area of wheel path; AVr is AV in specimen other than under wheel path; At is total
area of SGC specimen and, AVt is AV in SGC specimen.
         Since the cross-section of the portion of interest was expected to be a square of
100 mm. side, it was estimated that the area of the portion of interest was roughly half the
total area of an SGC specimen. Again, approximating this square of 100 mm. side as a

circle of 100 mm. diameter, the air voids under the wheel path were assumed to be
roughly half those in the remaining portion of the specimen.
A p  Ar                                                                  ... 13

AVp                                                                       ... 14
A rough relationship was thus found as:
AVt       AVp                                                             ... 15
Hereafter, knowing the approximate overall target air voids in the SGC specimens and
the total volume of the specimens, the volume of the mix required was computed:

       d 2 h
Vs                                                                        ... 16
      AVt 
Vm  1   Vs                                                             ... 17
      100 
where: Vs is volume of the SGC specimen; d is diameter of SGC specimen (=150 mm.);
h is height of SGC specimen (=130 mm.) and, Vm is volume of loose mix required.
With this knowledge and with the maximum theoretical specific gravity of the mix
known, the mix weight is computed as:
Wm  Vm Gmm                                                                ... 18

where: Wm is weight of mix required and Gmm is maximum theoretical specific gravity.
         A few specimens were prepared using mix weights equal to, about 2% less than
and about 2% more than the mix weight computed above. Air voids in the SGC specimen,
specimens cut to 48 mm. height, cut to briquettes and cut to the dimensions of the portion
of interest were found. Graphs were plotted to obtain an approximate relationship
between the mix weight and AV in the wheel path (Figure 20).



  AV under wheel path (%)

                                                                                                   y = -0.0164x + 91.13
                                                                                                           R2 = 1


                                                                y = -0.0178x + 97.479
                                                                     R2 = 0.9956

                             5.5              Top
                                              Linear (Bottom)
                                              Linear (Top)
                              4950.0                 5000.0               5050.0               5100.0                     5150.0   5200.0
                                                                               Mix weight (gms.)

                                       Figure 20: Mixture weight plotted against %AV in the wheel path
From these curves, the mix weight required to produce specimens with 5.5% AV in the
portion of interest was interpolated.
                                   Based on the above steps, an estimated mix quantity was obtained for target air
voids. It must be noted that the data included in this study in no way guarantees
achievement of the target air voids in mixes other than the one used here. However, for
mixes with a similar aggregate structure and asphalt content, the data may be used as a
ballpark. For other mixes, the procedure described herein must be repeated in its entirety.

Obtaining AV Uniformity Among Replicate Briquettes
It was found that the air voids in the top half and bottom half of the SGC specimens
differed by as much as 1.7% at times, as shown in Table 7. This was not acceptable, as
identical tests were to be run on both of these and results obtained were to be used to
characterize the actual pavement. Three possible courses of action were considered:
                            1. Prepare SGC specimens of shorter height so as to obtain only one briquette from
                                   each specimen. This was rejected as it involved much more time and material

    2. Cut off more material from the top and lesser from the bottom. This was
       considered as it was found that the top portion of the SGC specimen had more air
       voids than the bottom one.
    3. Gently rod the material after putting in the mould before compacting. This would
       bring about greater uniformity in material distribution. It was finally decided to
       also apply option 2 above in addition to this.

       Additional specimens were prepared using mix weight equal to the average of that
obtained from the curves for the top and bottom specimens. Options 2 and 3 were applied
to these and it was found that option 3 was most effective in achieving uniformity in AV
distribution between the top and bottom portions of the SGC specimens, as shown in
Table 7. The procedure for estimating the mix weight for a target air void content as
described above has been found to be very effective in reducing labor and material
wastage. The following points must be borne in mind when using this procedure:
   The procedure is purely analytical and therefore any geometry of the specimens can
    be analyzed using this method.
   The procedure is only a method of estimation so as to reduce effort and material
    required for achieving required air void content in a specimen.
   The data presented herein may be used to estimate mix quantities for other mixes with
    similar aggregate structure and asphalt content.
In the present study, it was found beneficial to rod the material in the mould before
compacting in order to obtain a uniform air voids distribution through the height of the
SGC specimen.
       After trafficking, the air void content of the briquettes was measured again. For
the binder + surface course briquettes, which had been bonded together using a tack coat,
the equivalent Gmb was calculated both before and after trafficking and comparisons
made on these values. The equivalent Gmb was computed using the following expression:

            m1  m2  m3
Gmbe                                                                               ... 19
          m1     m      m
              2  3
         Gmb1 Gmb2 Gmb3
where Gmbe is equivalent Gmb, mi’s are the masses of the two briquettes and the tack
respectively, and Gmbi’s are the corresponding Gmbs. The Gmb for the tack coat was taken
to be 1.02. Based on PennDOT specifications, tack coat was applied at the rate of 0.14 to
0.18 liters per square meter, resulting in about 10ml. of tack coat being applied on the
binder layer briquette.

            Table 7: Air Void uniformity using different methods

         Mix wt.                                Air Void Content (%)
  ID                 Comments               48 mm. Cut Briquettes Wheel Path
         (gms.)                               1       2      1     2     1     2
                     No rodding;
M5179A   5115.3     Cut 15 mm. off    9.9    8.6     7.8    8.7    7.7   7.4   6.3
                   top and bottom.
                     No rodding;
M5179B   4984.3     Cut 15 mm. off 12.2 10.6         10.1   10.9   9.9   9.5   8.8
                   top and bottom.
                     No rodding;
M5179C   5164.1     Cut 15 mm. off    9.5    7.7     7.3    7.5    7.1   6.6   5.7
                   top and bottom.
                     No rodding;
                    Cut 20 mm. off           8.3     7.8    7.5    6.8   6.2   5.3
M5198A   5181.7                       9.5
                   top and 10 mm.
                      off bottom
                   Gentle rodding;
                    Cut 20 mm. off           8.0     8.0    6.9    6.8   5.4   5.4
M5198B   5186.4                       9.3
                   top and 10 mm.
                      off bottom
                               1 = Top; 2 = Bottom


To top