International Journal of Pavement Engineering

ISSN: 1029-8436 (Print) 1477-268X (Online) Journal homepage: [Link]/journals/gpav20

A method to evaluate the tensile strength and

ductility of asphalt binders using a thin confined
film

Angelo Filonzi, Satyavati Komaragiri, Ramez Hajj, Manuel Trevino, Darren

Hazlett, Enad Mahmoud & Amit Bhasin

To cite this article: Angelo Filonzi, Satyavati Komaragiri, Ramez Hajj, Manuel Trevino, Darren
Hazlett, Enad Mahmoud & Amit Bhasin (2023) A method to evaluate the tensile strength
and ductility of asphalt binders using a thin confined film, International Journal of Pavement
Engineering, 24:2, 2031194, DOI: 10.1080/10298436.2022.2031194

To link to this article: [Link]

Published online: 06 Feb 2022.

Submit your article to this journal

Article views: 875

View related articles

View Crossmark data

Citing articles: 11 View citing articles

Full Terms & Conditions of access and use can be found at

[Link]
INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING
2023, VOL. 24, NO. 2, 2031194
[Link]

A method to evaluate the tensile strength and ductility of asphalt binders using a
thin confined film*
Angelo Filonzia, Satyavati Komaragiri a
, Ramez Hajjb, Manuel Trevinoa, Darren Hazletta, Enad Mahmoudc and
Amit Bhasina
a
Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, Austin, TX, USA; bDepartment of Civil and
Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, USA; cTexas Department of Transportation, Austin, TX, USA

ABSTRACT ARTICLE HISTORY

It has become clear over the last few years with an increasing use of asphalt binder modifiers, additives Received 5 August 2021
and recycled materials that there is a need to develop methods and concomitant parameters that can Accepted 14 January 2022
evaluate the full tensile behaviour including nonlinearity and damage resistance of asphalt binders.
KEYWORDS
This study presents a simple, repeatable test method to examine the tensile strength and ductility of Poker-chip test; tensile
asphalt binders using a poker-chip geometry. The poker-chip geometry has previously been used to strength; ductility; thermal
induce a realistic stress state in asphalt binder similar to that exerted on asphalt binder confined cracking; fatigue cracking
between two aggregates. This study develops and presents a new simplified procedure for specimen
fabrication and preparation using the poker-chip geometry. A set of nearly 100 asphalt binders
including different aging conditions, different levels and types of modification, and recycled asphalt
binders and contents were examined using the proposed approach. The results indicate that a
ductility parameter obtained using the test is highly repeatable with an average coefficient of
variation of 9%. In addition, the test was able to distinguish between different levels of modification
and was sensitive to aging, recycled binder ratio and modifier type.v

1. Introduction and background different conditions. Techniques such as the time-sweep test
are associated with edge instability in the test specimen, whereas
The tensile strength, ability to relax and ductility of an asphalt
other rheological surrogates and indicators have been shown to
binder dictate the ability of an asphalt mixture to resist fatigue
work well but may not be best suited for certain binders such as
and thermal cracking. The field performance of a mixture is
those modified using elastomers. The most important contrast
dictated by the component materials, mixture parameters
between the first three categories and the last category is that
(e.g. gradation and binder content), and pavement structure
the latter evaluates the binder’s ultimate capacity when sub-
as well as extrinsic factors such as traffic and environment.
jected to a stress state that is similar to what the binder experi-
However, the intrinsic properties of the binder, including the
ences in an asphalt mixture as opposed to a bulk unconfined
impact of additives, extenders and reclaimed asphalt pavement
specimen that is used in the DSR.
(RAP), on such properties are critical to ensure optimal per-
The focus of this study is to present a new simplified method
formance of the mixture and the pavement. Over the years,
that can be used to evaluate the tensile failure characteristics of
the significant role of asphalt binders has led to research and
an asphalt binder under a state of stress that is relevant for fati-
development of several different methods to evaluate the
gue or thermal cracking in asphalt mixtures. In order to better
cracking resistance of asphalt binders. Such methods have
understand the need and significance of the test method devel-
been used to evaluate the impact of additives, modifiers, exten-
oped and described here, it is important to address two impor-
ders by producers and researchers, and also in some cases as a
tant questions that are a precursor to the motivation of this
purchase specification.
study: (i) why it is important to consider triaxility while evalu-
The different methods that have been used to evaluate the
ating failure properties of an asphalt binder and (ii) what has
cracking resistance of asphalt binders can be classified into
already been done to evaluate asphalt binders in a laboratory
four broad categories: (i) time sweep tests conducted typically
under triaxial stresses. The background related to these two
using a Dynamic Shear Rheometer (DSR), (ii) amplitude
questions is presented in the subsections below.
sweep tests typically conducted using a DSR, (iii) use of rheolo-
gical indices that serve as a surrogate for measurement of duct-
ility (also typically determined using a DSR) and (iv) measuring
1.1. Triaxiality and mode of failure
tensile strength or fatigue life of asphalt binders when subjected
to a realistic stress state (Hajj and Bhasin 2018). The first three Most asphalt binder and mixture design processes consider
approaches rely on the use of a DSR with a parallel plate geome- fatigue cracking and rutting as disparate forms of distress
try to evaluate the mechanical response of the binder under that are often evaluated in isolation from each other. This

CONTACT Satyavati Komaragiri [Link]@[Link]

*Preprint submitted to International Journal of Pavement Engineering
© 2022 Informa UK Limited, trading as Taylor & Francis Group
2 A. FILONZI ET AL.

compartmentalisation, although convenient, is not a realistic by several studies in the past. Marek and Herrin (1968) exam-
representation of the complex damage processes that occur ined the differences in tensile strength of thin films as a func-
in an asphalt mixture. From the perspective of an asphalt bin- tion of their film thickness. In essence, they were varying the
der, Sultana et al. (2014b) demonstrate that the three-dimen- triaxiality of these films and demonstrated the transition
sional stress state of the binder directly dictates the localised from fracture to flow as the film thickness to diameter ratio
failure mechanism. Stress state can be quantified in terms of increased. Chang (1994) used a similar set up to examine the
‘triaxiality’, which can be defined as the ratio of hydrostatic crack growth in the specimen using acoustic emission tech-
to von Mises stress. Sultana et al. (2014b) verified that asphalt niques. These observations were further examined and corro-
binder films with low triaxiality demonstrate a flow or plastic borated by researchers using a similar approach but different
behaviour whereas binder films with high triaxiality exhibit geometries such as a butt-joint geometry with a confined thin
fracture through cavitation. The importance of triaxiality is film of asphalt binder (Harvey and Cebon 2003; Poulikakos
also intuitively recognised in the design of asphalt mixtures, and Partl 2011) or a thin film of binder between spherical solids
i.e. mixes with high binder content (which results in thicker (Hammoum 2009). Sultana et al. (2014c) used a confined thin
films and consequently lower triaxiality) are more prone to film using metal plates, referred to as the ‘poker-chip’ geometry
rutting and vice versa. in some of the very early works by Lindsey (1967), to demon-
The role of triaxiality on the failure mechanism is well recog- strate the transition from flow to fracture as a function of aspect
nised in the literature on polymeric materials (Asp et al. 1995; ratio that directly dictated the magnitude of triaxiality. They
Kody and Lesser 1997; Kishi 1998; Nicolaou and Semiatin 2001; also used the work potential theory (Schapery 1981; Kim
Wang et al. 2002; Mishnaevsky and Gross 2005; Katnam 2012). et al. 1997) to isolate viscoelastic effects and model damage
These studies have used several different geometries to examine characteristics of the binder that were independent of the rate
this phenomenon. These geometries typically include a speci- or mode of loading observed from these tests.
men with a notch with varying thicknesses to achieve several More recently, Hajj (2017) demonstrated the use of the
different stress states from uniaxial to triaxial in different poker-chip test to identify differences in the tensile strength
regions of the test specimen (Kishi 1998; Wang et al. 2002; Kat- of asphalt binders with the same performance grade. Diak
nam 2012), a hollow cylinder with different internal pressures et al. (2020) used a similar approach with a DSR (albeit a
to achieve different stress states (Kody and Lesser 1997), or a thicker binder film) to correlate the results from a butt-joint
poker-chip geometry with different film thicknesses to achieve test to the fundamental parameters from a double etched
different stress states (Asp et al. 1995). notch test that had previously been shown to strongly correlate
For most polymers, studies have also demonstrated that with field performance. Filonzi (2020) demonstrated that the
under low confinement, or in a uniaxial or biaxial stress results from the poker-chip test correlated much better than
state, the material failure is more ductile or dominated by plas- other binder metrics with mixture cracking performance
tic flow or shear banding (Asp et al. 1995; Kody and Lesser measured in the laboratory using a variety of different asphalt
1997; Kishi 1998). Similarly, under confined conditions and mixtures which included recovered blends of recycled binder
in a triaxial stress state, the material tends to fracture and and virgin binder.
the failure is more brittle or tough, depending on the type of
material, and with lower failure strains (Asp et al. 1995;
2. Motivation and scope
Kody and Lesser 1997; Kishi 1998; Wang et al. 2002; Katnam
2012). In addition to triaxiality, some of these studies have Since the very early work of Clark (1958), it is well recognised
also used other metrics to quantify the state of stress to estab- that the tensile strength and ductility of an asphalt binder have
lish a relationship between failure mode or failure strains. a direct bearing on the ability of the binder and the mixture to
These include using a modified von Mises criterion that also resist cracking in the field. Over the years, several different
includes the hydrostatic stress (Kody and Lesser 1997) and methods have been developed to either measure the two afore-
ratio of stresses in the principal directions (Asp et al. 1995). mentioned properties directly or a surrogate for them (Hajj
Hajj and Bhasin (2020) employed a multi-scale approach and Bhasin 2018). Of these, tests that are conducted using a
using finite element analysis to estimate the extent of triaxiality confined film of the binder (often referred to as the butt
in a typical dense graded asphalt mixture. They demonstrated joint test or poker-chip test or cohesion test) reflect the realis-
that the triaxiality factor in the asphalt binder film can be as tic state of stress of the binder and can capture both strength
high as 2.2, meaning the hydrostatic stress is 2.2 times the and ductility of the binder. Results from these tests have also
von Mises stress. Based on this analysis and the previous shown to differentiate between binders with the same Per-
studies, it is feasible to envision that an asphalt mixture can formance Grade (PG) and have correlated well with fatigue
experience both plastic deformation and cracking simul- cracking performance of asphalt mixtures measured in the lab-
taneously with localised damage that is spatially distributed oratory, as discussed above. However, one significant draw-
in a manner consistent with the localised stress state. back in this approach is the ability to consistently create a
thin confined film of the binder that can be tested with low
variability and high sensitivity. This drawback also prevents
1.2. Evaluation of asphalt binders subjected to high
users from using this approach on a more routine basis to
triaxiality
evaluate failure properties of asphalt binders.
The importance of evaluating asphalt binders under a highly The objective of this paper is to (i) present a newly devel-
confined state of stress or high triaxiality has been recognised oped method that can be carried out in a laboratory to evaluate
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 3

both the tensile strength and ductility of an asphalt binder in a . Group 5: The binders in this group were used to assess the
confined state of stress and (ii) to demonstrate the sensitivity impact of another common industrial waste product that is
of the method to changes in material composition that impact used as a modifier for asphalt binders. The evaluation
tensile strength and ductility. included four base binders from different sources. Two of
these binders were blended with 5.0% of the modifier and
the other two binders were modified using 2.5%, 5.0 %
3. Materials
and 7.5% of the modifier. These binders were evaluated
In order to assess the sensitivity of the method to different only after RTFO aging, resulting in a total of 12 binders.
materials, the test was conducted using a variety of binders
from different projects and studies. All materials were sourced These groups represent a range of binder conditions that vary
from producers in and around the state of Texas in the United from short-term to long-term aged; with and without incorpor-
States. In the context of this study, these binders can be cate- ation of RAP; and with different types of modifiers. These bin-
gorised into the following groups. ders were selected at these specific aging conditions to ensure
a wide enough range of composition including recycled
. Group 1: The binders in this group were used to assess the materials and modifiers. For blending asphalt binder with
differences in the material response based on their perform- these different types of modifiers, a can with 300 g of asphalt bin-
ance grade (PG). This group included PG 58-28 (2 different der was heated at 165◦ C for approximately 45 min to ensure that
sources), 64-22 (4 different sources), 70-22 (6 different the asphalt binder is completely fluid. Modifiers were then added
sources) and 76-22 (3 different sources). All binders in to the asphalt binder and blended for 2 h at the temperature of
this group were evaluated after Rolling Thin-Film Oven around 165◦ C using a high shear mixer at 2400 rpm. The mod-
(RTFO) and Pressurised Aging Vessel (PAV) aging. In ifier concentration indicated in percentage throughout this text
addition, this group also included samples of binder is based on the weight of the asphalt binder blend.
extracted from 6 RAP stockpiles from different locations
across the state of Texas. The binder was extracted and
4. Sample preparation
recovered following ASTM D1856, D. (2015). The RAP bin-
ders were evaluated in ‘as-is’ condition, i.e. without any The poker-chip test used in previous studies (Sultana et al.
additional aging or conditioning beyond extraction. This 2014a, 2014b; Hajj 2017; Filonzi 2020) achieved a high level
group had a total of 36 binders including different aging of triaxiality by using a very thin film with a 12.5-mm-diameter
conditions. specimen. However, a challenge from these previous studies
. Group 2: The binders in this group were used to assess the was that preparation and conditioning of test specimens
impact of blending virgin binder with RAP binder on the required a considerable amount of time. Further, due to the
material response from this test. A total of 15 blends were relatively thin size of the film, in order of hundreds of micro-
prepared by using different percentages of RAP binder metres, the sample preparation and testing had to be con-
with virgin binder (including one blend that also included ducted using extremely tight tolerances to avoid high
recycled asphalt shingles). Since the materials being used variability, which in turn required a significant level of oper-
were also part of other studies, the exact percentage of ator skill and practice. In order to overcome these shortcom-
RAP was dictated by the mix design that was being used ings, researchers developed a simplified version of this test
or proposed for each project. Each blend was further sub- by increasing the film thickness and diameter of the test speci-
jected to RTFO and PAV aging before being evaluated, men to retain the same level of high triaxiality. The sample
resulting in a total of 30 binders including the two different preparation procedure is briefly described here for the sake
aging conditions. of completeness. A more detailed procedure can be found in
. Group 3: The binders in this group were used to assess the Bhasin (2021).
impact of an elastomeric polymer on the measured par- The poker-chip test utilises a two-plate disc specimen. The
ameters. The evaluation included three base binders from bottom plate has a flat face of 57.15 mm (2.25 inch) diameter
different sources that were blended with three different per- with a 3.812 mm (0.15 inch) deep lip along the circumference
centages (1.5%, 3.0 % and 4.5%) of a commonly used elas- to retain liquid binder within it. The top plate has a flat face of
tomeric polymer. The blends were evaluated after RTFO 50.8 mm (2 inch) diameter. Machine drawings for these two
and PAV aging the binders, resulting in a total of 24 binders plates can be found in Bhasin (2021).
including the control and modified binders in both aging In order to prepare a test specimen, (4.5 + 0.05) g of the
conditions. In this case, the control binder was unmodified binder is pre-weighed in a silicone container. The poker-
base at the same level of aging as the modified binders. chip specimen discs are preheated to 165◦ C and the pre-
. Group 4: The binders in this group were used to assess the weighed binder sample is dropped into the heated bottom
impact of a chemical modifier that is sometimes used to disc (Figure 1). The bottom plate is placed back in the oven
enhance the high temperature grade of an asphalt binder. for a few minutes until the binder flows to uniformly spread
The evaluation included two base binders from different over the bottom plate, after which time three spacer dowels
sources that were blended with 1.5% of the chemical mod- 3.175 mm (0.125 inch) in length and 1.5875 mm (0.0625
ifier. The binders were evaluated after both RTFO and PAV inch) in thickness are dropped approximately 120◦ apart and
aging, resulting in a total of eight binders including the con- radially halfway between the centre and edge of the specimen
trol and the two aging conditions. (Figure 2). These spacer dowels control the final thickness of
4 A. FILONZI ET AL.

Figure 1. Image showing (a) bottom plates for five poker-chip sample discs; (b) Figure 3. Image showing (a) a poker-chip sample that sandwiches the binder
one bottom plate with a binder sample placed in it. specimen between the top and bottom pieces that are aligned using three align-
ment pins; (b) five complete discs ready for temperature conditioning and testing.
the binder film to 1.5875 mm (0.0625 inch). An examination of
the fracture surface from several tests shows that the dowels do
5. Test method and parameters
not participate in fracture nucleation or growth and do not
have a significant impact on the measured properties. In fact, The poker-chip test can be conducted using a displacement or
a similar approach is used with glass beads to control film a load controlled mode of loading. Using a universal testing
thickness in a ‘T-peel’ test (Li 2013), a common test for machine, a variety of test protocols can be applied including
adhesives used to measure the energy required to plastically monotonically increasing loads, cyclic loading and even
deform an adhesive sandwiched between two rigid media by long-term creep tests. Sultana et al. (2014c) present a more
peeling the specimen from the top into a t-shaped defor- detailed description of the comparison of results from tests
mation. The use of these dowels also make it possible to pre- conducted using different modes of loading as well as analysis
pare specimens and ensure a very consistent film thickness of damage using a continuum damage model. For the purposes
with ease and without the aid of any other device. of this study, the goal was to identify a mode of loading that
After placing the spacer dowels, the specimen in the bot- could be used to most readily quantify the ductile nature of
tom plate is placed in the oven for another 5–10 min. The the binder under different conditions. To this end, after mul-
bottom plate is then removed from the oven, the alignment tiple trials using different materials and modes of loading, a
pins are placed in the three holes of the bottom plate and the monotonically increasing load at a rate of 2 N/s was found to
top plate is gently lowered on to the binder surface guided by be the most repeatable and sensitive to changes in binder com-
the alignment pins (Figure 3). Note that the mass of binder position. In this context, sensitive refers to the fact that the rate
used ensures that the film thickness is greater than the dowel of loading was not too high where all binders showed similar
thickness to ensure full surface contact but to prevent no brittle failure but also not too low to result in an abnormally
more than 0.0255 cm (0.01 inch) thickness of extruded bin- long test duration (>15 min). This loading protocol was used
der on the rim. The specimen is now ready and is placed for all the tests conducted and presented in this study. All
back in the oven for another 5–10 min to allow the binder tests in this study were conducted at 25◦ C. Figure 4 shows
to properly wet and adhere to the face of the top plate. an asphalt binder sample during and after the test. 180 differ-
After this time, the specimen is removed from the oven ent binder tests were executed using this method. With a few
and allowed to cool to the test temperature on a level plat- minor exceptions, most tests were conducted using three repli-
form. Typically, multiple specimens can be prepared at the cates. These include binders with different levels of aging as
same time in batches. well as binders extracted from field cores and RAP stockpiles.
Based on the sample preparation procedure, the diameter to The two parameters of interest that were extracted and eval-
film thickness ratio achieved is 32. According to Sultana et al. uated in this study were (i) tensile strength of the binder and
(2014c), this aspect ratio represents a state of stress that tran- (ii) elongation or percentage strain of the binder when it
sitions failure from a flow to cavitation condition. reaches 80% of post peak stress. These two parameters are
obtained from plotting the stress versus strain response from

Figure 2. Image showing (a) bottom plates for five poker-chip sample discs with
binder samples; (b) one bottom plate with the three spacer dowels dropped on Figure 4. Image showing (a) an asphalt binder sample during the test and (b) an
top of the binder surface. asphalt binder sample after test.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 5

Figure 5. Typical results from the test method used in this study showing three replicates for the same binder. Vertical lines represent the ductility.

this test (Figure 5). For brevity, the latter term is referred to as 6.1. Repeatability
ductility in the remainder of this paper and the former is
The average coefficient of variation of ductility parameter
referred to as the tensile strength. Figure 6 shows a typical
obtained from the replicates was 9% with a standard deviation
stress–strain curve obtained from the poker-chip test along
of 9%. The average coefficient of variation of tensile strength
with the tensile strength and ductility parameters. Researchers
obtained from the replicates was 7% with a standard deviation
are aware that these parameters are tied to the loading rate and
of 5%. It must be noted here that test repeatability on multiple
test temperature. However, as described earlier, the main goal
devices was beyond the scope of this study and that all
of this study was to examine the ability of this method (includ-
measurements were made using the same loading frame.
ing loading protocol) to quantify and differentiate the ductile
Also, approximately 80% of the measurements were made by
nature of different binders.
the same operator and the remaining by two other operators
and only a limited number of samples were tested by different
6. Results and discussion individuals. Although a rigorous inter- and intra-laboratory
study was beyond the scope of this work, preliminary results
The ductility results for each group are discussed in seriatim from these limited tests show a good repeatability between
along with key findings from each group. The results for tensile different operators.
strength are discussed at the end of this paper.

6.2. Group 1 -- differences in ductility for different PG

binders
Figure 7 presents the results for ductility measured using com-
mercial PG binders after both RTFO and PAV aging. A few key
observations from these results are summarised below.
Ductility of PG58-28 and PG64-22 binders appears to be
similar after RTFO aging. Also, although there were only 2
PG58 and 4 PG64 binders from very different sources, the duct-
ility values were similar for each grade (although this could
change if additional binders from more diverse sources were
added to the pool). A marked increase in the ductility of
Figure 6. Typical stress–strain curve with the two parameters of interest used in PG70-22 binders compared to the ductility of PG64-22 binders
this study: (i) tensile strength and (ii) ductility. can also be observed from the results. This is expected because
6 A. FILONZI ET AL.

specification due to a combination of very high stiffness com-

pared to the rated grade and elastic recovery binders, but does
not necessarily show high ductility.
Results from the six different RAP stockpiles show a lot of
diversity, which is expected since RAP stockpiles can come
from a variety of different sources. Interestingly, two RAP bin-
ders had a ductility that was almost similar to a PG64-22 RTFO
aged binder whereas four other sources had a ductility that was
lower than most PAV aged virgin binders. The impact of PAV
aging on ductility was also clearly evident from the results. In
all cases, as expected, the ductility reduced. However, a more
interesting observation was that some binders were more
resistant to aging than others, i.e. some binders retained higher
ductility even after PAV aging compared to others. This will be
discussed again in subsequent sections.
Figure 7. Range of ductility values measured for RTFO and PAV aged binders with
different Performance Grades. The binders here are from Group 1.
6.3. Group 2 -- impact of blending RAP with virgin
binder
PG70-22 binders are likely to be polymer modified using elasto-
mers to meet the elastic recovery requirement per the specifica- The second group of binders included several different virgin
tions in the state of Texas at the time of this writing. Also, the six binders that were blended with RAP from different stockpiles
different PG70-22 binders showed a larger diversity in the duct- in different concentrations based on the proposed job mix for-
ility values compared to the PG64-22 binders, possibly due to mula. In this case, the poker-chip measurements were carried
different amounts and/or types of polymer added to each binder. out passively, i.e. based on the job mix formula as opposed to
For the PG76-22 binders, the average ductility was interest- being conducted to inform the job mix formula. Figure 8 pre-
ingly similar to the PG70-22 binders, although the differences sents the results from this group of binders. The data on the
among the three different sources were much more pro- very top of this figure along the x-axis represents the ductility
nounced. One of the PG76-22 binders had a very low ductility of the virgin binder after RTFO aging and the data at the very
value (nearly 300%) compared to other PG76-22 and PG70-22 bottom represents the ductility of the RAP binder on ‘as-is’
binders. Upon closer examination it was found that this binder basis after extraction. Note that the ideal reference in this
had a continuous high grade of 82◦ (even though it was being case would be the RAP binder after it has been RTFO aged,
supplied as a PG76) which possibly resulted in higher elasticity however, this was not feasible within the scope of this work
at lower temperatures but not necessarily higher ductility as due to the limited amount of RAP binder that was extracted
defined in the context of this study. Such a binder may pass and the amount that was also needed to blend with the virgin

Figure 8. Ductility of RTFO aged virgin binders, as-is RAP binders, and RTFO and PAV aged blends of these two binders. The binders here are from different sources
Group 2.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 7

Figure 9. Comparison of measured ductility of the blends with the approximate expected ductility based on weighted average of ductility of RTFO or RAP aged virgin
binder and as-is RAP binder (NOTE: a more accurate comparison would involve weighted average using RTFO and PAV aged RAP binder, which was not feasible due to
the limited amount of RAP extracted binder). The binders here are from Group 2.

binder. All data between these two extremes represents the available with regard to the modifier(s) used (if any) and the
ductility of the RAP-virgin binder blends after RTFO and concentration of such modifiers. It was of interest to examine
PAV aging. Some key observations are as follows. the impact of modifiers on the ductility parameter used in this
As observed with Group 1 binders, there is clearly a broad study. To this end, binders in Groups 3, 4 and 5 were blended
range of ductility values for the virgin RTFO aged binders and using a common elastomeric polymer (referred to as M1), a
the extracted RAP samples. However, as expected, the ductility chemical modifier commonly used to enhance the high temp-
of the RAP-virgin binder blend was between the two extremes erature grade of the binder (referred to as M2), and a third
and clearly dictated by the quality of the RAP, virgin binder, modifier that is also used as an extender with asphalt binders
and the percentage of RAP used with the binder. (referred to as M3). Figures 10– 12 present the ductility of
Figure 9 compares the measured ductility of the blends with the base and the modified binders after both RTFO and
the approximate expected ductility based on weighted average PAV aging. The average ductility values from Group 1 for
of ductility of RTFO or RAP aged virgin binder and as-is RAP PG64 and PG70 grade binders (after RTFO aging only) are
binder. The weighted average of ductility here is obtained by also included to provide a visual reference. Some of the key
adding the multiplication of ductility of each component of observations are as follows.
the blend with its weight fraction in the blend. As mentioned
earlier, researchers recognise that a more accurate comparison . Figure 10 clearly shows the increase in ductility of the bin-
should consider using RTFO and PAV aged RAP binder,
der with an increase in the elastomeric polymer concen-
which was not feasible due to the limited amount of RAP
tration after RTFO aging. The three different base binders
extracted binder. This also explains the slight bias in this com-
follow a slightly different trend but all three binders show
parison. Bias here refers to as the deviation of the data points
this increasing trend very clearly.
from the 1:1 line. Also note that one of these blends included . Also from Figure 10, after PAV aging, the ductility for the
an RAS binder, which was not included in this comparison
binders with 1.5% elastomer was similar to the base binder.
because no measurements were made using the extracted
However, at higher concentrations (3 and 4.5%), the bin-
RAS sample. Note that the evaluation of blends implicitly
ders continued to show slightly increased ductility com-
assumes 100% blending of the RAP and virgin binder, which
pared to the base binder even after PAV aging.
may not necessarily be true, these comparisons show that . Figure 11 shows that the chemical modifier M2 did not
this approach can be used to estimate the best case scenario
result in any increase in ductility compared to the base bin-
response in binder blends for specific pairs of virgin and
der. In fact, the ductility appears to decrease slightly for
RAP binders as well as additives, if any.
both binders. Results after PAV aging also show no sub-
stantial difference between the base and the modified bin-
der. This emphasises the importance of moving beyond
6.4. Groups 3, 4, 5 -- impact of modifiers
Performance Grading (PG) as the dominant method for
The asphalt binders evaluated in the previous sections were classifying asphalt binders in the USA; although modifiers
commercial PG binders, for which no information was such as M2 can achieve similar benefits to elastomers in
8 A. FILONZI ET AL.

Figure 10. Influence of elastomeric polymer modifier on the ductility of three different asphalt binders at three different concentration levels. The binders here are
from Group 3.

improving PG of a binder, they can ultimately prove 6.5. Tensile strength versus ductility
ineffective in improving nonlinear binder properties in
The aforementioned sections discussed the impact of PG, RAP
the same way, which is not captured by linear viscoelastic
blending and modifiers on the ductility of the binders as
(LVE) indicators.
defined in the context of the poker-chip test. However, as men-
. Figure 12 shows that the modifier / extender M3 had very
tioned earlier, another important parameter that is obtained
little to no impact on the ductility of the base binders except
from this test is the tensile strength of the binder. The differences
in one case at 7.5% concentration. This group of binders
in tensile strength of the binders were not discussed explicitly in
was not evaluated after PAV aging due to the limitations
the aforementioned sections for brevity but are presented here.
in the scope of this work.

Figure 11. Influence of chemical modifier on the ductility of two different asphalt binders at 1.5% by weight concentration. The binders here are from Group 4.
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 9

Figure 12. Influence of a binder modifier/extender on the ductility of four different asphalt binders; two were modified using three different percentages (2.5, 5.0,
7.5%) and the other two with one percentage (5%). The binders here are from Group 5.

Figures 13 and 14 compare the tensile strength versus above statement is verified when one examines Figure 14
ductility for Group 1 binders and Groups and 3 and, respect- for the M1 modified binders that are only RTFO aged
ively. Figure 13 shows that for RTFO aged binders the duct- (empty circle, triangle, and square for base binders at
ility increases as one moves from PG58 and 64 to 70 and 76 approximately 200% ductility shift to right for approximately
grade binders (compare empty circles and triangles at the 450% to 1200% ductility with increasing elastomer content
bottom left to the empty diamonds and squares on the bot- but with no pronounced change in tensile strength). For
tom right). However, more interestingly, the tensile strength most RTFO aged binders the increase is observed in ductility
of the binders does not change substantially when only not in tensile strength. Interestingly, on this same figure, one
RTFO aged binders are considered. This is very likely can also see that for the M2 modified binders, the ductility
because the use of elastomeric polymers that enhance duct- did not change or slightly decreased but resulted in an
ility but may not necessarily impact tensile strength. The increase in the tensile strength of the binder.

Figure 13. Comparison of tensile strength and ductility of PG binders. The binders here are from Group 1.
10 A. FILONZI ET AL.

Figure 14. Comparison of tensile strength and ductility of binders modified using M1 (elastomer) and M2 (chemical). The binders here are from Groups 3 and 4.

A second key observation from these results is that the ten- in ductility due to aging. Figure 15 presents the percentage
sile strength of the binders increases and ductility decreases of ductility retained after PAV aging compared to RTFO
with aging. This is consistent across all binders even though aging for binders with different grades and RAP blends.
the rate of change and starting point, especially the starting These results show that PG 70-22 binders tend to experience
point for modified or PG70, 76 binders, can be very different. a greater loss in ductility; it is important to recognise that
this relative change must be interpreted in the context of the
high initial values.
6.6. Impact of aging
A couple of key observations from Figure 15 are as follows.
The impact of PAV aging has already been discussed in the First, the variation in the percentage loss in ductility is very
previous sections. This information is revisited for binders in high between different binders. Given the diversity in the
Groups 1 and 2, particularly to examine the relative change chemical composition and modifications, the differences

Figure 15. Percentage of ductility of PAV aged binder relative to RTFO aged binder. (Note: RAP indicates RAP blends with different virgin binders from Group 2. The
binders here are from Groups 1 and 2.)
 INTERNATIONAL JOURNAL OF PAVEMENT ENGINEERING 11

between binders from the same group can be expected and it is Disclaimer
clearly detected by the ductility parameter from this test
The contents of this paper reflect the views of the authors who
method. Second, the average percentage loss for RAP blends
are solely responsible for the facts and accuracy of the data pre-
is somewhat similar to PG64 binders. This is despite the fact
sented herein and do not necessarily reflect the official views or
that most RAP blends were based on PG70 and 76 binders.
policies of any agency or institute. This paper does not consti-
This is likely due to the fact RAP binders do not age at the
tute a standard, specification, nor is it intended for design,
same rate as the virgin binders, which in turn reduces the
construction, bidding, contracting, tendering, certification, or
overall rate of aging for the blending. As before, the ductility
permit purposes. Trade names were used solely for infor-
parameter is able to detect and quantify this variability.
mation purposes and not for product endorsement, advertise-
ment, promotions, or certification.
7. Summary and conclusion
This study developed a simple method to measure ductility Disclosure statement
and tensile strength of an asphalt binder by using poker-chip No potential conflict of interest was reported by the authors.
geometry. This method simulates the stress state of an asphalt
binder when it experiences fatigue or fracture in the mixture
by creating a high triaxiality in the test specimen. Moreover, Funding
this test method evaluates the asphalt binder until complete This work was supported by Texas Department of Transportation.
failure. The proposed method and a ductility parameter were
used to evaluate several different commercial and laboratory
ORCID
blended asphalt binders to examine the sensitivity of
the method and the parameter. It was observed from the Satyavati Komaragiri [Link]
results that the proposed method and concomitant ductility
parameter: References
. can distinguish differences within binders of the same PG, Asp, L.E., Berglund, L.A., and Gudmundson, P., 1995. Effects of a compo-
site-like stress state on the fracture of epoxies. Composites Science and
. can distinguish between the quality of RAP binders from Technology, 53 (1), 27–37.
different stockpiles, and ASTM D1856, D. 2015. Standard test method for recovery of asphalt from
. is very sensitive to the type and concentration of modifier. solution by Abson method. USA: Annual Book of ASTM Standards.
Bhasin, A., et al., 2021. Parts and sample preparation guide for the poker
These findings show that both the developed test method chip or asphalt ductility test. Available from: [Link]
T8/KTCNVK
and the ductility parameter are sensitive to variations in binder Chang, W., 1994. Application of acoustic emission to study the cohesive and
composition and modification. Consequently, demonstrating adhesive strength of asphalt. Washington, DC: Strategic Highway
the feasibility of this test method to assess the impact of blend- Research Program Report SHRP-A-682.
ing RAP, aging, modifiers and other additives on the mechan- Clark, R., 1958. Practical results of asphalt hardening on pavement life.
ical response of the asphalt binder. Other variations of this test Journal of the Association of Asphalt Paving Technologists, 27, 196–208.
Diak, E., Beier, E., and Hesp, S.A., 2020. Development of a butt joint test
method (e.g. different modes of loading) can also be used for for the ductile performance grading of asphalt binders. Construction
more fundamental evaluation of asphalt binders and modifiers. and Building Materials, 243 (3), 118195.
Authors recognise that the results reported in this study Filonzi, A., et al., 2020. Investigating the use of a binder cohesion test to evalu-
were from three different researchers and only for a few ate cracking resistance of asphalt mixtures. In: RILEM International
cases with multiple replicates of the same asphalt binder Symposium on Bituminous Materials, 12, Lyon, France: Springer.
Hajj, R., et al., 2017. Evaluation of stiffness, strength, and ductility
were evaluated by all three researchers. However, no systema- of asphalt binders at intermediate temperature. Transportation
tic differences in variability were observed in the results from Research Record: Journal of the Transportation Research Board, 2632
these limited tests showing a good repeatability of this test (1), 44–51.
between operators. Authors also recognise that all the tests Hajj, R., and Bhasin, A., 2018. The search for a measure of fatigue crack-
were conducted using one loading frame (servohydraulic ing in asphalt binders–a review of different approaches. International
Journal of Pavement Engineering, 19 (3), 205–219.
actuator with a 25-kN load frame). To examine the repeatabil- Hajj, R., and Bhasin, A., 2020. Importance of triaxial stress state on asphalt
ity on different devices, approximately 10 binders were evalu- binder tensile failure. In: A.T. Papagainnakis, A. Bhasin, A. Kumar and
ated using this loading load frame as well as another low-cost D.N. Little, eds. Proc. of advances in materials and pavement perform-
load frame (electric motor actuator with a 5 kN load frame). ance prediction AM3P, 8, San Antonio, TX: Balkema.
The results were consistent across both units, although Hammoum, F., et al., 2009. Experimental and numerical analysis of crack
initiation and growth in thin film of bitumen. Road Materials and
additional data are being gathered for a more thorough evalu- Pavement Design, 10 (1), 39–61. Available from: [Link]
ation to accommodate the relatively slower motor speeds. [Link]/doi/abs/10.1080/14680629.2009.9690
Results obtained from both machines and multiple operators Harvey, J.A., and Cebon, D., 2003. Failure mechanisms in viscoelastic
suggest a low barrier for implementation in the future for films. Journal of Materials Science, 38 (5), 1021–1032. Available
routine use. However, additional data to establish intra- and from: [Link]
Katnam, K.B., et al., 2012. Investigating tensile behaviour of toughened
inter-lab repeatability as well as field validation of the ductility epoxy paste adhesives using circumferentially notched cylindrical
parameter is required and is being carried out as a part of an bulk specimens. In: International journal of adhesion and adhesives,
ongoing study. 9. Elsevier, Vol. 37, 3–10.
12 A. FILONZI ET AL.

Kim, Y.R., Lee, H.J., and Little, D.N., 1997. Fatigue characterization of Digital Collection, Vol. 58, 338–353. Available from: http://
asphalt concrete using viscoelaticity and continuum damage [Link]/appliedmechanicsreviews.
theory. Journal of the Association of Asphalt Paving Technologist, 66, Nicolaou, P.D., and Semiatin, S.L., 2001. The effect of stress triaxiality on
520–569. tensile behavior of cavitating specimens. Journal of Materials Science,
Kishi, H., et al., 1998. Ductility and toughenability study of epoxy resins 36 (21), 5155–5159. Available from: [Link]
under multiaxial stress states. Journal of Materials Science, 33 (13), 10.1023/A:1012485510150
3479–3488. Available from: [Link] Poulikakos, L., and Partl, M., 2011. Microscale tensile behavior of thin
A:1013222421843 bitumen films. Experimental Mechanics, 51 (7), 1171–1183.
Kody, R.S., and Lesser, A.J., 1997. Deformation and yield of epoxy net- Schapery, R., 1981. On viscoelastic deformation and failure behavior of
works in constrained states of stress. Journal of Materials Science, 32 composite materials with distributed flaws. Advances in Aerospace
(21), 5637–5643. Available from: [Link] Structures and Materials, 5–20.
1023/A:1018693028394 Sultana, S., Bhasin, A., and Bhasin, A., 2014a. Effect of chemical compo-
Li, Q., et al., 2013. Finite element analyses of several t-peel specimen sition on rheology and mechanical properties of asphalt binder.
configurations using cohesive zone models. In: 36th annual meeting Construction and Building Materials, 72 (1), 293–300.
of the adhesion society. Sultana, S., Bhasin, A., and Liechti, K.M., 2014b. Rate and confinement
Lindsey, G.H., 1967. Triaxial fracture studies. Journal of Applied Physics, effects on the tensile strength of asphalt binder. Construction and
38 (12), 4843–4852. Available from: [Link] Building Materials, 53 (1257), 604–611.
1063/1.1709232 Sultana, S., Bhasin, A., and Liechti, K.M., 2014c. Rate and confinement
Marek, C.R., and Herrin, M., 1968. Tensile behavior and failure character- effects on the tensile strength of asphalt binder. In: 93rd annual meet-
istics of asphalt cements in thin films. Journal of the Association of ing, transportation research board, 1, Washington, DC.
Asphalt Paving Technologists, 1968, 1–54. Wang, T., Kikno, K., and Mitsuo, N., 2002. Effect of triaxial stress con-
Mishnaevsky, L.L., and Gross, D., 2005. Deformation and failure in thin straint on the deformation and fracture of polymers. Acta Mechanica
films/substrate systems: methods of theoretical analysis. In: Applied Sinica/Lixue Xuebao, 18 (5), 480–493. Available from: [Link]
mechanics reviews, 9. American Society of Mechanical Engineers [Link]/article/10.1007/BF02486573