evaluation of porous friction courses with highly modified

Evaluation of porous friction courses with highly modified asphalts

to reduce raveling

Presented by:

Carlos Alberto Rivera T.

Adviser:

Silvia Caro Spinel Ph.D.

Universidad de los Andes-Colombia

Faculty of Engineering

Master Degree in Civil Engineering

Department of Civil and Environmental Engineering

December 2018

ii

Dedication

I dedicate this work to my parents, Alberto and Martha for their support, love,

comprehension and teachings.

iii

Acknowledgements

First of all, I want to express my complete gratitude and admiration to my adviser, Dr.

Silvia Caro. She not only guided me in this laborious path, which is research, but helped me

improve as a person and as a professional. She inculcated me the love for research, and the

importance for detail to achieve perfection. Words aren’t enough to describe my gratitude,

therefore I will only say “thank you” expecting she comprehends the full weight of those words.

Special thanks to all members in the research group GeoSI for being a support and a help

to improve this research by discussing and providing insightful ideas. It is also important to thank

everyone that helped me in one or another way at my short, but meaningful stay at TTI. To Dr.

Edith Arámbula, to receive me and guide in the experimental section of this work. She has always

been present and of immense help in this great venture. To Florida Department of Transportation

for funding this research with project code BE287 that enabled the elaboration of this research.

To my family, which has supported me for so long and helped me achieve my dreams, this

is work is also yours. My dad, Alberto, for teaching me that hard work pays off and that tiredness

is just a state of mind. Your example was vital for me to be able to complete this. For my mother,

Martha, for being always present, especially in the hard times, and providing with constant

counsels and guidance and for boosting me to thrive. To Lorena Pupo, for always being by my

side, and being of such help when everything seemed dark, your support was vital for me to be able

to accomplish everything.

This work presents a series of teachings by persons that have impacted me at some point in

my life. Some earlier than other, but that I will always cherish.

“Nothing in the world is complex, complexity lies in the sum of simple things” Mauricio Sanchez-Silva

“There are no short-cuts in life”-“My advice is this: be humble and learn” Father Francis Wheri O.S.B.

iv

Table of Contents

1. INTRODUCTION ................. ERROR! BOOKMARK NOT DEFINED.

2. EXPERIMENTAL WORK .................................................................... 6

2.1 EXPERIMENTAL BINDER CHARACTERIZATION ..................................... 6

2.1.1 PG-Grade ........................................................................................ 6

2.1.2 Rheological Properties ................................................................... 7

2.1.3 Glover-Rowe Parameter ............................................................... 11

2.1.4 Fatigue Evaluation ....................................................................... 13

2.2 MIXTURE CHARACTERIZATION.......................................................... 14

2.2.1 Mixture Fabrication ...................................................................... 14

2.2.2 Semi-circular bending test ............................................................ 18

2.2.3 IDEAL test ..................................................................................... 20

2.2.4 Durability ...................................................................................... 23

3. NUMERICAL SIMULATION IN FINITE ELEMENTS ................. 25

3.1 MODEL DESCRIPTION ......................................................................... 25

3.2 LOADING CONDITIONS ...................................................................... 29

3.3 LONG-TERM AGING CONDITION ... ERROR! BOOKMARK NOT DEFINED.

4. CONCLUSIONS AND RECOMMENDATION ................................ 38

5. BIBLIOGRAPHY ................................................................................. 40

v

List of Figures

FIGURE 1. WORLDWIDE USE OF HIMA ASPHALT BINDERS (MODIFIED AFTER KLUTTZ ET AL. (2014))

.................................................................................................................................................. 4

FIGURE 2. MASTER CURVES @ 45°C FOR PMA BINDER IN DIFFERENT AGING STATES. ...................... 9

FIGURE 3. MASTER CURVES @45°C FOR HIMA BINDER AT DIFFERENT AGING STATES. .................. 10

FIGURE 4. ORIGINAL, PAV20 AND PAV80 MASTER CURVES @ 45°C FOR HIMA AND PMA

BINDERS. ................................................................................................................................. 10

FIGURE 5. G-R BLACK SPACE DIAGRAM FOR PMA, HIMA AND PG 52-28 BINDERS. ....................... 12

FIGURE 6. PLAS TEST RESULTS OF FREI PARAMETER FOR PMA AND HIMA BINDER. .................... 13

FIGURE 7. SCB DESCRIPTION .......................................................................................................... 18

FIGURE 8. SEMI-CIRCULAR BENDING TEST SETUP ............................................................................ 19

FIGURE 9. CRACKING RESISTANCE INDEX (CRI) FOR ALL MIXTURES AT VARIOUS AGING STATES. .. 20

FIGURE 10. IDEAL-CT TEST SETUP. ............................................................................................... 21

FIGURE 11. SPECIMEN IN CONTACT WITH THE EDGE OF THE IDEAL-CT LOADING FRAME DURING

TESTING. .................................................................................................................................. 22

FIGURE 12. CTINDEX VALUES FOR THE FOUR DIFFERENT MIXTURE WITH UNAGED AND AGED STATES.

................................................................................................................................................ 23

FIGURE 13. CANTABRO TEST RESULTS FOR THREE AGING STATES. .................................................. 24

FIGURE 14. PFC MICROSTRUCTURES REPLICATES WITH 20% AV: (A) 2 CM THICKNESS (B) 4 CM

THICKNESS .............................................................................................................................. 25

FIGURE 15. PARTICLE ORIENTATION ................................................................................................ 26

FIGURE 15. FE PAVEMENT MODEL IN ABAQUS® ............................................................................. 28

FIGURE 16. GLOBAL MESH OF A FE MODEL WITH A 4 CM PFC ........................................................ 29

FIGURE 17. LOADING APPLICATION AND VERTICAL STRESS FOR A 2 CM THICK PFC

MICROSTRUCTURE ................................................................................................................... 29

FIGURE 18. TRACTION-SEPARATION LAW OF THE CZM ELEMENTS; MODIFIED AFTER CARO (2009) 32

FIGURE 19. SDEG VALUES FOR CONTACTS IN AN PFC MICROSTRUCTURE ...................................... 33

FIGURE 20. PDF OF ER FOR A REPLICATE OF THE 2 CM HIMA-GRANITE PFC .................................. 34

FIGURE 21. ER RESULTS FOR THE PFC´S ........................................................................................... 36

vi

List of Tables

TABLE 1. RESULTS EVALUATION PG GRADE AND MSCR TEST ......................................................... 7

TABLE 2. MASTER CURVE PARAMETERS OF THE PMA BINDER AT A REFERENCE TEMPERATURE (TR)

OF 45ºC FOR DIFFERENT AGING STATES ..................................................................................... 8

TABLE 3. MASTER CURVE PARAMETERS FOR THE HIMA BINDER AT A REFERENCE TEMPERATURE

(TR) OF 45ºC FOR DIFFERENT AGING STATES .............................................................................. 8

TABLE 4. WASHED SIEVE ANALYSIS RESULTS FOR GRANITE AND LIMESTONE MIXTURES ................ 15

TABLE 5. MAXIMUM SPECIFIC GRAVITY FOR THE PFC MIXTURES TYPE FC-5 .................................. 15

TABLE 6. GMB FOR MIXTURES WITH PMA BINDER ........................................................................... 16

TABLE 7. GMB FOR MIXTURES WITH HIMA BINDER .......................................................................... 17

TABLE 8. AV CONTENT (%) FOR THE MIXTURES WITH PMA BINDER .............................................. 17

TABLE 9. AV CONTENT (%) FOR THE MIXTURES WITH HIMA BINDER ............................................. 18

TABLE 10. CHARACTERISTICS OF THE OGCFC MIXTURES MICROSTRUCTURES ............................... 26

TABLE 11. PRONY SERIES FOR MASTIC IN THE LONG-TERM AGED CONDITION AT 30°C ................... 31

TABLE 12. INPUT PARAMETERS OF THE CZM TRACTION-SEPARATION LAWS .................................. 33

TABLE 13. ER PARAMETER IN EACH PFC MICROSTRUCTURE EVALUATED ....................................... 35

TABLE 14. ER PARAMETER TO EVALUATE RAVELING IN THE PFC MIXTURES ................................... 36

1

1. Introduction

An open-graded friction course (OGFC) or porous friction courses (PFC) is a thin layer

placed on top of a conventional dense-graded hot mix asphalt (DGHMA). These types of mixtures

have been used over the past six decades; they were first used in Europe in 1960 and in the U.S.

ten years later (Mallick et al. 2009, Mejias De Pernia 2015, Hernandez-Saenz et al. 2016). This

type of mixtures offers a series of improvements in terms of serviceability and noise reduction

achieved by a high air void (AV) content between 15-20% (Roque et al. 2006, Hernandez-Saenz

et al. 2016). Some of the benefits offered by this type of mixtures are:

• Reduced risk of hydroplaning (Dell´acqua et al. 2011)

• Increased friction resistance (Huddleston et al. 1991, Nicholls 1997)

• Reduced backsplash and spray from vehicle tires (Nicholls 1997, Rungruangvirojn and

Kanitpong 2010)

• Improved visibility of pavement markings (Lefebvre 1993)

• Noise reduction (i.e. 3-6 dB) (Nicholls 1997, Cooley Jr. et al. 2009, Hernandez-Saenz et al.

2016)

• Improved car speed and traffic capacity (Cooley Jr. et al. 2009)

However, the durability of these mixtures is considerably shorter than dense-graded hot mix asphalt

(DGHMA), (i.e. due to the loss of surface particles; a phenomenon called raveling), with values

that range from 6-12 years, and best case scenarios of 16 years (Voskuilen and Elzinga 2010,

Hernandez-Saenz et al. 2016). Due to the difficulties of maintenance activities during winter

periods, its use in northern states in the U.S. has been discontinued and elsewhere has been a topic

of serious study (Hernandez-Saenz et al. 2016). Some new methodologies and approaches have

been introduced to increase the durability of PFC mixtures such as the use of polymer modified

asphalt, asphalt rubber or variations in mixtures design or material selection. Nowadays, the use of

highly polymer modified asphalt (HiMA) has been introduced in DGHMA, with good field results,

and therefore it is been considered as a good option in increasing the durability of PFC mixtures.

This highly polymer modified asphalt have high dosages of polymer in a dosage of 7.5% per weight

of asphalt (Willis et al. 2016). Some initial research project indicated that the application of this

type of binder may increase the durability of PFC mixtures.

The objective of this research is to evaluate the impact of high asphalt modification in PFC mixture

and compare it with conventionally used modified binder, herein called PMA, in terms of durability

and mechanical performance. This goal will be achieved through a complete and comprehensive

laboratory characterization and a numerical modeling in finite elements. Since this is a durability

problem, various aging conditions of the materials were considered in order to understand the

evolution of the problem over time. The following section presents a concise literature review about

the durability of PFC and the use of HiMA in these asphalt mixtures. After this section, the

description of the results obtained from the experimental work are presented, which include

rheological properties, sensitivity to oxidation, and fatigue evaluation in HiMA and PMA binders,

as well as fracture properties and durability in PFC mixtures fabricated with these asphalts. Next,

the results of the finite element modeling to evaluate the mechanical response of a PFC with HiMA

under field conditions, and to compare them with those of a PFC fabricated with PMA asphalt.

Results indicate that high dosages of polymer modification increase significantly the performance

of PFC mixtures since it not only has better response to fracture, raveling and fatigue but also it

2

maintains such properties in various levels of oxidations that can be related to aging conditions in

field. When evaluated the performance of the mixture under the pass of half a single axle vehicle

using finite elements in two-dimension (2-D) models, it showed better performance and less

susceptibility to raveling.

3

2. Literature Review

This chapter will expose the main researches concerned to the durability of PFC mixtures. It

will indicate major advances in terms of durability associated by continent and describe major

studies conducted in HiMA binders. This last due to the growing potential in western countries to

include this asphalt as an alternative to enhance the performance of PFC mixtures and increase the

durability.

Although PFC mixtures offer important benefits, these types of mixtures present two main

challenges, one associated to functionality and a second associated to durability. In terms of the

functionality, it is affected by clogging, which is the fill of the AV by particles, which consequently

reduces the effective AV content and diminishes its benefits (Kandhal 2002, Alvarez, Epps, et al.

2010). In regard to durability, raveling is the most common distress present in these type of

mixtures, which consists on the loss of aggregates from the surface of the layer (Mo et al. 2008).

Previous works have shown an increase decline in the applications of these mixtures since 1980

for durability reasons, especially in northern states of the U.S. (Hernandez-Saenz et al. 2016).

These states deal with problems associated with maintenance during winter due to freeze-thaw

cycles and the application of sand, salt, or treatments used to prevent ice formation and therefore

induce clogging (Hernandez-Saenz et al. 2016, Watson et al. 2016). Some of the research related

to this type of mixtures has been focused mainly to durability, maintenance and increased reduction

of noise. Some of these works have been focused in better selection of materials, construction

practices, use of improved binders to improve durability (i.e. ground tire rubber (GTR) , or polymer

modified asphalt) or using a two-layer porous asphalt (TLPA), among others (Transportation

Reseach and Innovation Monitoring and Information System n.d., Kandhal 2002, Voskuilen et al.

2004, Gibbs et al. 2005, Peeters and Blokland 2007, Voskuilen and Elzinga 2010, Hamzah et al.

2013, Mobilite Intelligente 2017).

Although the majority of this research has consisted on experimental efforts, some computational

attempts have been also done in order to comprehend the raveling phenomena by understanding

the mechanical response of microstructures of PFC’s while considering various internal and

external conditions (e.g. mixture characteristics, weather factors, or vehicle’s speed). These types

of researches have been conducted mainly by two research groups, TU Delft (Mo et al. 2007, 2010,

2011, 2014, Huurman et al. 2009, Kluttz et al. 2013), and TTI in collaboration with Universidad

de los Andes (Alvarez, Mahmoud, et al. 2010, Manrique-Sanchez et al. 2016). Some of the most

important findings of these works are the effect of binder content in the interfacial zone of the

mastic, the impact of weather in the relaxation potential of the binder, the predominant mode of

fracture within the microstructure of PFC that causes raveling (i.e. Mode I of fracture), and the

effect of vehicle speed (i.e. lower speeds increases chances of raveling).

Kluttz et al. (2013) not only focused on the mechanisms of raveling, but evaluated the use of HiMA

as alternative to increase durability of mixtures using computational methods. HiMA binders, in

contraire to conventionally polymer modified asphalts or PMA (i.e. PMA has a dose of 2.5-3% of

polymer by weight of asphalt), have high doses of polymer by weight of asphalt (i.e. dose of 6-8%

of polymer) (Zhu et al. 2014, Willis et al. 2016). Polymers are common additives used to enhance

the properties of asphalt binder (West et al. 2012, Zhu et al. 2014, Polacco et al. 2015). The

introduction of polymer in asphalts generate two phases in the asphalt binder i) asphaltene-rich

phase ii) polymer-rich phase. This phase separation is generated by the absorption by polymers of

asphalt components such as saturates, aromatics and resins (i.e. commonly denoted as swelling of

4

polymers) present in the binder and, consequently, it changes the mechanical response of the

binder. HiMA has been widely used in DGHMA, as observed in Figure 1. This figure allows to

understand the extend of the use of the material and therefore identify the trend in the use of the

material which is expected to grow in the future.

Figure 1. Worldwide use of HiMA asphalt binders (modified after Kluttz et al. (2014))

It must be noted that the use of modified binder and its study has been usually an isolated issue

among continents. For example, the U.S. has comparatively recently studied the use of HiMA

binders to improve the mechanical performance of dense mixtures, and more recently in PFC

mixtures, meanwhile in Japan this topic was evaluated almost over 30 years ago (Suzuki et al.

2010). In the other hand, the European continent has opted to improve their binders by polymer

modification and they have improved their material selection, construction designs and procedures

to achieve longer durability in PFC mixtures (Voskuilen et al. 2004, Voskuilen and Elzinga 2010).

The U.S. most notorious study in terms of high polymer modification was conducted in the national

center for asphalt technology (NCAT) between 2008 and 2012 (Willis et al. 2009, Timm et al.

2012). This study consisted in evaluating the use of HiMA binder in the performance of DGHMA

in comparison to mixtures that incorporated PMA using the NCAT´s test track. As part of the

results obtained, it was observed that bending beam fatigue testing results for HiMA mixtures were

45 bigger than mixtures with PMA, which indicated high resistance of DGHMA with HiMA to

fatigue in comparison of structures with PMA. This discovery indicates that the increment in the

dose of polymer in asphalt binders can help decreasing the thickness of pavement layers to achieve

the same result.

Meanwhile in Japan the use of highly modified asphalts was introduced in 1989; in 1997 an even

higher dose in the market to satisfy the needs of special cases (i.e. bridge deck pavements, cold

climate or porous asphalt pavements using small sized aggregate) (Suzuki et al. 2010). Porous

asphalts in cold regions offer a challenge in terms of durability, therefore they reduced the AV

5

content of these mixtures to 17% and increased the dose of polymer modification up to 15% of

polymer by weight of asphalt. These types of changes allowed the extended use of PFC mixtures

in areas in which elsewhere are banned due to durability issues (e.g. in northern states of the U.S.).

Finally, regarding Europe, in order to improve durability, they have opted for: i) developing better

material selection or design procedures, and ii) enhancing the quality of binders. For example, the

Netherlands has implemented an astonishing 80% of their highways with PFC mixture type which

have a high durability (i.e. 16 years for the fast lane and 11 years for the slow lane). Voskuilen et

al. (2010) has pointed out that although they do not have highly modified asphalts to increase

durability, they have a more continuously graded mixture along with a rigorous selection of

materials and construction procedures. This assures a high-quality infrastructure that has shown an

astounding endurance. This approach has been of great importance, since most of the aggregates

employed by the Netherlands are transported from nearby countries like Germany, Belgium or

Norway. In contraire, countries like Germany have implemented polymer modification and

selection of materials to assure endurance and quality of their infrastructure (Bondt et al. 2016),

although they present lower durability times. The lower durability may be due to a higher AV

content which has been proved to reduce the durability of mixtures.

6

3. Experimental Work

The following sub-sections describe the evaluation and comparison of a highly modified

asphalt with Styrene-Butadiene-Styrene (SBS) polymer in doses in between 6-8% in contrast to a

conventionally modified asphalt with a dose of 2.5-3% by weight of asphalt. This test plan consists

of two main components: i) binder characterization, and ii) mixture characterization. The

characterization of the binder was done by i) evaluating the sensibility to oxidation and its rigidity

through the performance grade (PG grade), ii) the visco-elastic properties by conducting a

frequency and temperature sweep test, evaluating the Glover-Rowe parameter and the MSCR test

and, finally, iii) gaging the fatigue performance of the binder using the pure linear amplitude sweep

test (PLAS). In general terms, HiMA binders presented a better performance in the long-term since

they maintained their properties and were less susceptible to block cracking or fatigue.

In terms of the mixture, two aggregates (i.e. granite and limestone) and two asphalt binders (i.e.

PMA and HiMA) were employed to achieve a total of four different types of mixtures. These

mixtures were fabricated to an AV content of 20±1% at different oxidation states. Two main

components were evaluated to characterize and comprehend their mechanical behavior: i) the

fracture properties through the semi-circular bending test (SCB) and the indirect tensile asphalt

cracking test (IDEAL), and ii) durability assessed through Cantabro test. As well as observed in

the binder characterization, mixtures with HiMA presented better durability and resistance to

fracture through various aging conditions.

3.1 Experimental Binder Characterization

The binder characterization was conducted for a PMA asphalt binder, which is commonly

used as part of PFC mixtures in Florida and a HiMA binder that has been used in some projects in

the same state. Since the problem evaluated is mainly in terms of durability, various aging

conditions of the binders were evaluated to capture the evolution of the deterioration of the

material. The aging conditions were: original binder (OB), rolling thin film oven (RTFO),

pressurized aged vessel (PAV) for 20 hours (PAV20), PAV for 40 hours (PAV40) and PAV for 80

hours (PAV80), for a total of 5 aging conditions. Epps-Martins et al. (2017) conducted a study on

DGHMA in which various extents of oxidation under PAV were studied to find the equivalence

with field conditions. Therefore, a reference point was considered based on the properties of a PG

64-22 asphalt, which although is clearly not the same binder used as the base for the modified

asphalt, can be used to illustrate the level of deterioration and oxidation of the asphalts herein

evaluated. Consequently, the ballmark is set to 3, 6 and 7 years for PAV20, PAV40 and PAV60

respectively.

3.1.1 PG-Grade

The two binders were evaluated in their continuous and rounded PG grade. This test allows

to evaluate the stiffness, ductility, and sensitivity to aging. The results obtained from the evaluation

of the PG grade as well as the results of the MSCR test are presented in Table 1. These results show

that the both the continuous and rounded values of the PG were different from those provided by

the distributors (i.e. PMA binder is classified as PG 82-22E and HiMA is classified as PG82-28E).

This implicated a wider useful temperature interval (UTI). When comparing the asphalts in terms

of ductility, it is observed that the HiMA binder present a better ductility since it has a fifth of the

7

ΔTc of the PMA asphalt, understanding ΔTc as the difference in the bending beam rheometer

(BBR) test temperatures when the creep stiffness (S) and the stress relaxation rate (m-value) reach

the PG Superpave specification of 300 MPa and 0.3, respectively. This ductility result may be

influenced by the extent of the polymeric phase and the influence of the butadiene compound in

the SBS polymer. Based on information reported by Voskuilen et al. (2010) and Huurman et al.

(2009) these ductility values may increase the durability of mixtures since some of the problems

related to raveling are due to the lack of relaxation potential of binders.

In regard to the MSCR test, FDOT has different specifications depending on the level of

modification of the asphalt. For example, PMA asphalts should be evaluated at 67ºC and HiMA at

76ºC (Florida Department of Transportation 2018). Based on the same specification, the modified

asphalt must pass a series of requirements (i.e. has to be classified with at least traffic type V) i) the

non-recoverable creep compliance at a stress level of 3.2 kPa (Jnr, 3.2) has to be a maximum of 1.00

kPa-1 and 0.10 kPa-1 for PMA and HiMA, respectively ii) the average percent recovery at 3.2 kPa

should be greater than 29.4% and 90.0% for PMA and HiMA, respectively iii) and the difference

in the non-recoverable creep compliance between 0.1kPa and 3.2kPa (i.e. Jnr, diff) should be 75%

maximum (AASHTO 2014a, Florida Department of Transportation 2018); although if the Jnr, 3.2 is

lower than 0.5 kPa-1,the third criteria is not applicable, which is denoted in Table 1 as not applicable

or N/A.

Table 1. Results evaluation PG grade and MSCR test

Binder

Type

Commercial

PG brand

name

Continuous

PG

Rounded

PG ΔTc UTI

MSCR

min Jnr, 3.2 %Recovery %Jnr, diff

PMA PG 76-22 PG 84.9-

26.5

PG 82-

22E -4.3 111.4

0.09 kPa-1

(Pass) 70.5% (Pass) N/A

HiMA PG 76-22

HiMA

PG 87.4-

32.4

PG 82-

28E -0.8 119.9

0.04 kPa-1

(Pass) 93.8% (Pass) N/A

3.1.2 Rheological Properties

The linear viscoelastic properties of both binders in the different aging states were

determined with a Dynamic Shear Rheometer (DSR). The procedure included temperature sweep

tests (i.e. 10°C to 70°C, in increments of 10°C) and frequency sweep tests (e.g., 37.5, 30, 25, 20,

15, 10, 5, 1, 005, 0.01 rad/s) using the parallel plate geometry. The resulting parameters of interest

are the dynamic shear modulus, |G*|, and the phase angle,

Master curves of |G*| at a reference temperature of 45°C were constructed for the binders. The

Christensen and Anderson (CA) model shown in Equation 1 was used to fit the raw data

(Christensen and Anderson 1992), and the Williams Landel Ferry (WLF) model shown in equation

2 was used to adjust the corresponding shift factors.

8

|𝐺∗| = 𝐺𝑔 ∗ (1 + (𝑤𝑐

𝑓𝑟𝑒𝑑)

𝑘

)

−1

𝛽

(1)

G*g is the maximum dynamic shear modulus or glass modulus (in Pa) equal to 1 GPa, fred is the

reduced frequency (in rad/s), k is equal to 1, and wc, and 𝛽 are fitting coefficients.

log 𝑎𝑇 =−𝐶1(𝑇−𝑇𝑅)

𝐶2+𝑇−𝑇𝑅 (2)

Tr is the reference temperature (45ºC), and C1 and C2 the WLF fitting coefficients. A summary of

the master curve fitting parameters for both binders at all aging states which can be observe in

Table 2 and Table 3.

Table 2. Master curve parameters of the PMA binder at a reference temperature (TR) of 45ºC for

different aging states

Parameter PMA-OB PMA-

RTFO

PMA-

PAV20 PMA-PAV40 PMA-PAV80

C1 11.6 11.2 12.2 4.6 22.2

C2 133.3 135.1 137.0 79.9 207.0

Wc 1.6E+03 4.6E+02 1.9E+01 5.3E-01 2.4E-02

K 1 1 1 1 1

β 0.11 0.10 0.09 0.09 0.09

G*g 1.00E+09 1.00E+09 1.00E+09 1.00E+09 1.00E+09

Table 3. Master curve parameters for the HiMA binder at a reference temperature (TR) of 45ºC for

different aging states

Parameter HiMA-OB HiMA-

RTFO

HiMA-

PAV20

HiMA-

PAV40

HiMA-

PAV80

C1 6.8 7.3 8.2 10.5 15.6

C2 92.6 97.9 104.6 114.7 162.2

Wc 5.6E+03 3.2E+03 1.4E+02 1.9E+01 8.3E+00

K 1 1 1 1 1

β 0.091 0.098 0.09 0.09 0.09

G*g 1.00E+09 1.00E+09 1.00E+09 1.00E+09 1.00E+09

9

Figure 2 through 4 present the master curves obtained for both binders under different aging states

at the reference temperature of 45ºC. These curves correspond to the average of the results obtained

for three replicates of each binder at each aging state.

As expected, the results of the different binders indicate that at a higher level of aging |G*| increases

since the oxidation process increases the rigidity of the binder. For example, at 10Hz the PMA

binder has a value of 34.4 MPa for OB aging state, meanwhile for PAV80 aging state at the same

frequency the value is 4084.8 MPA. On the other hand, when comparing the |G*| for HiMA binder

in between PAV80 aging state and OB indicated that the modulus in the PAV80 condition is 20.7

times that of the HiMA OB.

It was also found that the HiMa binder presented a lower temperature and frequency susceptibility

than PMA binder since their average master curve slope was smaller (i.e. at 10 Hz the HiMA binder

in average for all the aging states was a 70% smaller). Finally, it was not adequate to compare the

moduli between binders since their base binder was not the same one. However, the results, as

observed in Figure 4, show that the |G*| of PMA asphalt binder for all aging states is higher than

that of HiMA binder. Based on results presented in previous studies (Chen et al. 2018), the trend

of the moduli of asphalt binder when modified is an increase, in other words, the higher the

modification the higher the moduli applying the same base binder. At the same time is known that

when applied high dosages of polymer is a common practice to include some oils to reduce the

moduli and increase the workability (Martins 2018). Consequently, this research will focus in the

performance and degradation processes of the binder rather than on their viscoelastic properties.

Figure 2. Master curves @ 45°C for PMA binder in different aging states.

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+09

1.E-05 1.E-03 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 1.E+09

Dy

na

mic

Sh

ear

Mo

du

lus

(Pa

)

Frequency Reduced (Hz)

PMA OB PMA RTFO PMA PAV20 PMA PAV40 PMA PAV80

10

Figure 3. Master curves @45°C for HiMA binder at different aging states.

Figure 4. Original, PAV20 and PAV80 master curves @ 45°C for HiMA and PMA binders.

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+09

1.E-05 1.E-03 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 1.E+09

Dy

na

mic

Sh

ear

Mo

du

lus

(Pa

)

Frequency Reduced (Hz)HiMA OB HiMA RTFO HiMA PAV20HiMA PAV40 HiMA PAV80

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+09

1.E-05 1.E-03 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 1.E+09

Dy

na

mic

Sh

ear

Mo

du

lus

(Pa

)

Frequency Reduced (Hz)PMA OB PMA PAV20 PMA PAV80HiMA OB HiMA PAV20 HiMA PAV80

11

3.1.3 Glover-Rowe Parameter

To explore the influence of aging on the expected cracking performance of both binders,

the Glover-Rowe or G-R parameter (equation 3), a rheological value proposed by Glover et al.

(Glover et al. 2005) that is considered a good predictor of the binder resistance to cracking

degradation due to oxidative hardening, was evaluated. The parameter is obtained from conducting

an oscillatory shear test at 15°C and 0.005 rad/s with binder under different aging states.

𝐺 − 𝑅 𝑃𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟 = 𝐺∗(cos 𝛿) 2

sin 𝛿 (3)

This parameter presents a good correlation with ductility and it has proven to be capable of

predicting block cracking performance with aging in the field. In fact, according to existing studies

in dense mixtures, G-R values equal or larger than 180 kPa are related to mixtures that show an

onset of early block cracking, meanwhile binders with values equal or larger than 600 kPa are

related to mixtures that have a higher probability for developing significant block cracking (Rowe

et al. 2014).

The G-R parameters were obtained through DSR tests under the aging states previously described.

Figure 5 presents the Black space diagram (|G*| vs. ) with the results of both binders under all

aging states (i.e. each binder at each aging state represents one point in this graph). The results of

an unmodified binder PG 52-28 from Florida were also included in this diagram to remark the

influence of polymer modification with aging on the mechanical response of the binders (i.e. less

change in phase angle). It should be noted that the most aggressive aging state for the unmodified

binder PG 52-28 was PAV60 (i.e. there was no available data for PAV80 since these data was

obtained from a different study). This figure also includes two curves that represent different

damage states related with the G-R parameters. Each curve was computed using the G-R limits

previously mentioned. That is, the ‘Damage Onset’ curve shows the threshold for G-R values equal

to 180 KPa, and the ‘Significant Damage’ curve represents the G-R threshold values equal to 600

KPa. Although those G-R threshold values were determined for dense-graded and not for PFC

mixtures, they are still considered acceptable to evaluate the binder potential susceptibility to

damage, especially with aging.

12

Figure 5. G-R black space diagram for PMA, HiMA and PG 52-28 binders.

In Figure 5, the farthest the results are from the “Damage Onset” or “Significant Cracking” line

and the change in phase angle is the least, the better. Taking into account the information obtained

from Figure 5, the most notorious result is the relation between the susceptibility to degradation to

oxidative hardening and polymer modification. For example, the change in the phase angle for the

HiMA, PMA and PG 52-28 binder among the different aging condition were of 2.7%, 26.1% and

29.7%, respectively. In terms of the G-R parameter, binders with less modification dosage had

higher susceptibility to oxidative hardening; presenting an average increase of 2.9, 5.6 and 57.8

times its previous value for HiMA, PMA and PG52-28 asphalt binder, respectively (i.e. PAV20/OB

;PAV40/PAV20 or PAV80/PAV40). Finally, with respect to the G-R thresholds, the binders reach

the “Damage Onset” at different aging states depending on their susceptibility to cracking and

oxidation hardening. The PMA asphalt binder reached the “block cracking zone” at a point between

PAV 20 and PAV40 or approximately 5 years of pavement in-service life, and the PG 52-28 at a

greater aging state with a value in between PAV40 and PAV60. In contraire, the HiMA binder did

not attain this zone even at PAV80 aging state. This result infers that higher dosages of polymer

(i.e. HiMA asphalt binder) produce binders with a higher resistance to oxidative aging and,

therefore, they can implicate a longer durability of PFC.

13

3.1.4 Fatigue Evaluation

The PLAS test is a continuous oscillatory strain sweep test conducted in the DSR

equipment. The applied torsional strain increases linearly from zero to 30% over the course of

3,000 cycles. This test is conducted at a constant loading frequency of 10 Hz and it provides a

parameter of interest: the Fatigue Resistant Energy Index (FREI), which is derived from fracture

mechanics principles and useful to evaluate the susceptibility to fatigue of the binder. Equation (4)

explain how it is computed from the test results:

𝐹𝑅𝐸𝐼 =𝐽𝑓−𝜏𝑚𝑎𝑥

𝐺0.5𝜏𝑚𝑎𝑥∗ (𝛾0.5𝜏𝑚𝑎𝑥)2 (4)

where Jf-max is the shear fracture energy calculated at the maximum shear stress, G0.5 max is the

shear strain computed at half of the maximum shear stress, and 0.5 max is the shear strain at half of

the maximum shear stress.

Figure 6 presents the information in regard to the FREI parameter, in which higher values imply a

higher resistance to fatigue cracking. For the five aging conditions it is possible to observe a higher

value of the FREI parameter for the HiMA asphalt binder, and therefore a better resistance to

fatigue. For example, the HiMA asphalt binder is 3.2 times higher in OB condition and 8.5 times

higher in PAV80 than the PMA binder. It is important to denote that these results are in good

agreement to results obtained for fatigue in literature (Kluttz et al. 2014, Orlen Asfalt 2017). In

regard to the speed of reduction in the cracking resistance due to aging, it is possible to infer that

the parameter is no sensitive to the asphalt since the variation in between PAV aging conditions is

similar (i.e. 50% and 44% reduction in between PAV aging states for HiMA and PMA asphalt

binders).

Figure 6. PLAS test results of FREI parameter for PMA and HiMA binder.

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

Original RTFO PAV20 PAV40 PAV80

FR

EI

PMA HiMA

14

3.2 Mixture Characterization

PFC mixtures were prepared with both asphalt binders (i.e. HiMA and PMA) with two

types of aggregates commonly used in PFC mixtures in Florida (i.e. granite and limestone) The

compacted specimens were tested in order to determine their fracture resistance and durability.

These mixtures were evaluated in various aging condition that resemble different states of

oxidation after considering that this type of hardening is one of the main causes that diminishes the

durability in PFC mixtures. Thus, the experimental procedure present in NCHRP 9-54 was used to

simulate various conditions (Kim et al. 2017). This procedure consists on subjecting the loose

mixture at 95°C for a series of days to simulate various field aging conditions. Although this

procedure was conducted to correlate field aging of DGHMA to lab aging, it is a good equivalence

and considered representative for PFC mixtures. The unaged condition, denoted A0, consists on

only applying the AASHTO R30 procedure (AASHTO 2002). The consequent aging condition is

denoted as A5 and resembles five days at 95°C of the loose mix and corresponds to approximately

to 2-3 years of field aging. To evaluate the durability of PFC mixtures, a conditioning of 10 days

at the same temperature was conducted and was denoted as A10, which resembles approximately

5 or more years of in field in-service for DGHMA. Three type of tests were conducted to evaluate

the performance of this mixture: i) the SCB test, ii) the IDEAL test, and iii) the Cantabro test. The

first two were conducted to evaluate the fracture properties and tensile strength of the different

PFCs, and the Cantabro test to evaluate their durability, which will be discussed in the following

sections.

3.2.1 Mixture Fabrication

The first step in the production of the asphalt mixtures was the adjustment in the proportions

of the aggregates in accordance with the FC-5 job mix formula provided by FDOT. The two

aggregate types, limestone and granite, were provided by FDOT in 5-gallon buckets from the

original quarries. All the aggregates were oven dried at 110°C for 24 hours, and later cooled and

sieved.

Washed sieve analyses were conducted according to ASTM C117, which requires 2500 grams

batched aggregates samples following the gradations in the job mix formula (AASHTO 2013,

ASTM 2017a). Enough water was added in order to cover the aggregate samples and the material

was agitated by hand with vigor to separate the finer particles from coarser aggregates. Once the

fine aggregates were in suspension, the water was poured into a set of sieves (i.e. #8 and #200

sieves); when the decantation of materials finalized, the process was repeated until clear water was

observed, which indicated that most fines were washed from the larger aggregate particles. The

remaining material and the material retained in the two sieves were combined and dried at 110°C

for 24 hours; then, the material was sieved to determine its washed gradation. If the change with

respect to the mix design gradation was larger than 1% and 0.5% for coarse and fine particles,

respectively, the aggregate proportions were adjusted, and the process repeated until a minimum

error (i.e. difference between the job mix formula and the washed percent passing amount for all

sieves (within the margin of error) was achieved. The granite aggregate gradations did not require

any correction, while for the limestone aggregate gradations, three iterations were needed to obtain

a washed aggregate gradation equivalent to the FC-5 job mix formula gradation. The original and

adjusted after washed sieve analysis aggregate gradations (i.e. percent passing) are listed in Table

4.

15

Table 4. Washed sieve analysis results for granite and limestone mixtures

Sieve

Number

Original

Gradation

Granite

Adjusted

Gradation

Granite

Original

Gradation

Limestone

Adjusted

Gradation

Limestone

1/2" 2.0% 2.0% 13.0% 14.5%

3/8" 27.0% 27.0% 28.0% 27.3%

#4 48.0% 48.0% 37.0% 37.2%

#8 14.0% 14.0% 13.0% 12.4%

#16 5.0% 5.0% 3.0% 2.9%

#30 0.0% 0.0% 1.0% 1.0%

#50 1.0% 1.0% 1.0% 1.0%

#100 0.0% 0.0% 1.0% 1.5%

#200 0.5% 0.5% 0.3% 0.2%

Pan 2.5% 2.5% 2.7% 2.0%

With the adjusted aggregate gradations and the optimum binder content (OBC) specified in the mix

design (i.e. 5.9% of asphalt per weight of mixture for granite and 6.4% of asphalt per weight of

mixture for limestone), the theoretical maximum specific gravity, 𝐺𝑚𝑚 (ASTM 2011), was

determined for each mixture. These results, obtained as the average from a total of three replicates,

are listed in Table 5.

Table 5. Maximum specific gravity for the PFC mixtures type FC-5

Aggregate PMA Binder HiMA Binder

Limestone 2.36 2.35

Granite 2.57 2.55

Compacted test specimens were prepared following AASHTO R30 (AASHTO 2002). The mixing

and compaction temperatures depended on the type of mixture. The mixtures with PMA binder had

lower mixing and compaction temperatures in comparison to the mixtures with HiMA binder.

These temperatures were provided by FDOT. For the granite-PMA mixtures the temperatures were

166ºC and 163ºC for mixing and compaction, respectively; for limestone-PMA mixtures both

temperatures were 160ºC. For the mixtures with the HiMA binder, the temperatures were 171ºC

and 166ºC for mixing and compaction respectively, regardless of the type of aggregate. In summary

the mixtures with granite contained an OBC of 5.9%, 1% of hydrated lime by weight of mixture,

and 0.3% of mineral fiber by weight of aggregates; meanwhile limestone mixtures contained an

16

OBC of 6.4%, 0.5% liquid anti-strip by weight of asphalt, and 0.3% of mineral fibers by weight of

aggregates.

The aggregates were batched following the washed sieve analysis adjusted gradations and dried

overnight at the respective mixing temperature. When hydrated lime was required, it was included

in the aggregate batch. The aggregates were mixed with the fibers before adding the binder. Short-

Term Aging (STOA), labeled A0, was achieved by subjecting the loose mix to the compaction

temperature for 2 hours prior to compaction. The subsequent aging conditions evaluated in this

study consisted in subjecting the mix to an additional 5 or 10 days at 95°C which was denoted as

A5 and A10 aging state, respectively.

The specimens were compacted using the Superpave Gyratory Compactor (SGC) at a compaction

angle of 1.25 degrees and a pressure of 600 kPa with a target AV of 20±1 %. The specimens were

confined using 150 mm polyvinyl chloride (PVC) pipe sleeves as soon as they were extracted from

the SGC compaction mold to prevent sagging and crumbling, which is likely to occur in specimens

with high air void content. The sleeves were left in place for at least 12 hours while the specimens

cooled down in front of a fan. The PVC pipe sleeves were cut on one side to avoid disturbing the

sample during extraction from the SGC mold and sealed with duct tape. Even with this precaution,

the average diameter of the extracted specimens was not exactly 150.0 mm but approximately 150.6

mm; which was considered acceptable. In addition, the observation by Alvarez et al. (2009), who

noted that SGC compacted PFC specimens tend to expand in the vertical direction after extracting

from the mold, were taken into consideration. This phenomenon was also reported in the work

conducted by Arambula-Mercado et al. (2016) who observed that SGC compacted specimens

expanded in the vertical direction an average of 1.0 mm, although this growth depended on the size

of the sample (i.e. specimens smaller than 80 mm expanded less than the specimens 160 mm

height). To prevent these issues, the specimens were compacted 1.0-2.0 mm below the target height

required for each test to account for this vertical expansion. The actual diameter and height

measurements were taken into consideration in the AV content calculation. The AV content was

determined between 12-48 hours after the specimen was compacted per ASTM D3203 (ASTM

2017b). All specimens were tested within one week after they were compacted.

The average results and the corresponding variability of the bulk specific gravity, 𝐺𝑚𝑏, of the

compacted specimens for the mixtures with PMA and HiMA binders under all aging states are

summarized in Table 6 and Table 7. In these tables, the term N/A refers to those tests in which the

A10 aging state was not considered (i.e. all but the Cantabro abrasion loss tests). As observed, the

coefficient or variability (COV) of the 𝐺𝑚𝑏 results were less or equal than 1% in all cases, proving

the high reliability of the specimen fabrication process.

Table 6. Gmb for mixtures with PMA binder

Test Aggregate Avg.*

A0

Avg.

A5

Avg.

A10

Std.

Dev.**

A0

Std.

Dev.

A5

Std.

Dev.

A10

COV

A0

COV

A5

COV

A10

SCB Granite 2.035 2.044 N/A 0.002 0.012 N/A 0.1% 0.6% N/A

Limestone 1.894 1.907 N/A 0.008 0.003 N/A 0.4% 0.2% N/A

IDEAL Granite 2.058 2.032 N/A 0.021 0.002 N/A 1.0% 0.1% N/A

17


A0

Avg.

A5

Avg.

A10

Std.

Dev.**

A0

Std.

Dev.

A5

Std.

Dev.

A10

COV

A0

COV

A5

COV

A10


Cantabro Granite 2.066 2.064 2.052 0.013 0.011 0.003 0.6% 0.6% 0.1%

Limestone 1.903 1.902 1.893 0.008 0.008 0.010 0.4% 0.4% 0.5%

* Avg. corresponds to the average of all test replicates.

** Std. Dev. corresponds to the standard deviation between the test replicates.

Table 7. Gmb for mixtures with HiMA binder


A0

Avg.

A5

Avg.

A10

Std.

Dev. **

A0

Std.

Dev.

A5

Std.

Dev.

A10

COV

A0

COV

A5

COV

A10





Cantabro Granite 2.069 2.038 2.048 0.004 0.009 0.020 0.2% 0.5% 1.0%

Limestone 1.881 1.884 1.895 0.016 0.008 0.009 0.9% 0.4% 0.5%


** Std. Dev. corresponds to the standard deviation between the results of all test replicates.

Table 8 and Table 9 present the corresponding AV content of the test specimens for each mixture,

aging state and test type. As a consequence of the low variability in Gmb, the AV content also

presented a small dispersion (i.e. COV <5% for all cases).

Table 8. AV content (%) for the mixtures with PMA binder


A0

Avg.

A5

Avg.

A10

Std.

Dev.

** A0

Std

Dev.

A5

Std

Dev.

A10

COV

A0

COV

A5

COV

A10





Cantabro Granite 19.6 19.6 20.1 0.5 0.4 0.1 2.5% 2.3% 1%

Limestone 19.4 19.5 19.7 0.4 0.4 0.6 2.1% 2.0% 3%



18

Table 9. AV content (%) for the mixtures with HiMA binder

Test Aggregate

Avg.*

AV

A0

Avg.

AV

A5

Avg.

AV

A10

Std.

Dev. **

A0

Std

Dev.

A5

Std

Dev.

A10

COV

A0

COV

A5

COV

A10





Cantabro Granite 19.0 20.2 19.8 0.2 0.4 0.8 0.8% 1.8% 4%

Limestone 19.8 19.7 19.2 0.7 0.3 0.4 3.5% 1.7% 2%

* Av. corresponds to the average of all test replicates.


3.2.2 Semi-circular bending test

The SCB test was performed according to AASHTO TP 124-16 (AASHTO 2016). This

test procedure consists of fabricating SGC specimens 150 mm diameter and 160 mm height. Next,

using a diamond-tipped saw blade with a water-cooling system, two smaller cylindrical specimens

with a height of 50±1mm were obtained from the SGC specimen. Afterwards, these specimens

were cut in half and a notch 15±1mm deep and 1.5±0.1mm wide was introduced in the center of

the flat side of the specimen (see Figure 7). It should be highlighted that although the AV content

was controlled in the fabrication of the 160 mm height specimen, it was not verified for the final

SCB specimens.

The assembly for the SCB test can be observed in Figure 8. After conditioning the specimens at

25±0.5°C for 2±10 min, they were located in the assembly where they are simply supported on

their flat side. Then, a control load was applied on top of the specimen at a controlled displacement

rate of 50 mm/min. In accordance to ASTM D3549 (ASTM 2017c), two input values are required

for the test software: i) the ligament length, and ii) the total thickness of the specimen to the nearest

millimeter. The parameter recorded during the test was the vertical load applied on top of the

specimen.

Figure 7. SCB description

19

Figure 8. Semi-circular bending Test setup

The SCB test is usually studied based on the Fracture index (FI) (parameter that characterizes the

damage resistance of asphalt mixtures using the fracture energy, and the slope of the load-

displacement curve) but based on the results presented by Kaseer et. al (2018), another test

parameter was considered in this study, named the Cracking Resistance Index or CRindex shown in

equation 9. It should be noted that although the fracture energy (Gf), which is the area under the

load-displacement curve, considers both the strength and the ductility of the mixture, it does not

differentiate between mixtures that may have different properties but similar fracture energies. The

CRindex, on the contrary, is able to do so by incorporating the peak load (Pmax) in its calculation.

Thus, the CRindex is effective in differentiating and ranking the cracking resistance of asphalt

mixtures with diverse characteristics, with the additional advantage that it has a reduced variability

in comparison to the FI (Kaseer et al. 2018). Overall, larger values of CRindex represent better

resistance to cracking, while lower values indicate brittleness.

𝐶𝑅𝑖𝑛𝑑𝑒𝑥 =𝐺𝑓

𝑃𝑚𝑎𝑥 (5)

20

Figure 9. Cracking resistance index (CRI) for all mixtures at various aging states.

In general terms, the results of Figure 9 indicate higher values for the mixture with HiMA asphalt

binder, independently of the aggregate or the aging state. Therefore, it is expected that these

mixtures are more resistant to fracture. For example, in A0 age state, the mixture HiMA-limestone

has a CRI value 2.2 times PMA-limestone, whether for HiMA-granite is 1.4 granite-PMA. In

regard to the aggregate, there is no clear trend in A0 condition, while in the A5 aging state the

mixture with granite present a better performance since they have higher values of CRI,

independently of the asphalt binder. Based on the results, it is possible to infer that the main factor

influencing the fracture resistance of mixtures is the binder, since the higher values of the CRI

parameter were for mixtures with HiMA binder for both aging states.

3.2.3 IDEAL test

The IDEAL-CT test (Zhou et al. 2017) consists on applying a monotonic load (P) at a rate

of 50 mm/min at the top center of cylindrical specimens using the IDT strength test equipment at

25°C. Although there is no standard for the height of the specimen, a value of 62 mm was selected.

Figure 10 shows the test assembly.

A0 A5

HiMA-granite 1864.9 1271.2

HiMA-limestone 2194.0 923.3

PMA-granite 1319.1 807.4

PMA-limestone 994.5 536.8

0

500

1000

1500

2000

2500

3000

Cra

ckin

g R

esis

tan

ce I

nd

ex

21

Figure 10. IDEAL-CT test setup.

The result of this test is the CTindex (equation 10), which is a dimensionless value:

𝐶𝑇𝑖𝑛𝑑𝑒𝑥 = 𝑡

62×

𝐺𝑓𝑃

𝑙

×𝑙

𝐷 (6)

where Gf is the work of fracture, t is the specimen’s thickness in mm, P

𝑙 is the slope of the load (P)

versus the vertical displacement curve (l), and 𝑙

𝐷 is the strain tolerance under the load (vertical

displacement to diameter ratio (the division of the current diameter and the initial diameter of the

specimen), D).

The CTindex is calculated using the post peak load on the load-displacement curve. More precisely, P

𝑙 – which is considered a ‘modulus’ of the specimen – is calculated as the absolute slope of the

load-displacement curve in a zone located between the point corresponding to 85% of the

maximum load in the post-peak zone and that corresponding to 65% of the maximum load (i.e.

PPP85 and PPP65). In general, larger values of CTindex indicate slower cracking growth rates, and

consequently, better cracking performance.

Similar to what was reported for the SCB test, some of the specimens with the HiMA binder

endured large deformations during testing, allowing the sample to touch the edges of the loading

frame, inducing an error in the acquired data. This phenomenon can be observed in Figure 11,

where the lower left corner the specimen is in contact with the lower part of the IDT assembly.

This situation did not affect the post-processing of the data, since the values for PPP85 and PPP65

22

required to compute the P

𝑙 variable were not impacted. Although this situation also induced some

small increments in the load-displacement curve; those increments were considered negligible and,

therefore, there was no need to correct the computed fracture energy.

Figure 11. Specimen in contact with the edge of the IDEAL-CT loading frame during testing.

The values of the CTindex for both aging conditions can be observed in Figure 12. Base on the results

obtained it is possible to infer that the asphalt binder is the most important factor in determining

the cracking performance in these mixtures. Mixtures with HiMA asphalt binder presented

considerably higher values of CTindex for both aging states. For instance, at A0 aging state, HiMA-

granite was 1.7 times higher than HiMA-limestone and HiMA-limestone was 2.7 bigger than PMA-

granite. In terms of the aggregate type, it indicated that mixtures with limestone, present a worse

performance in terms of cracking performance, since for bot binder types they had smaller values

(i.e. at A5 aging state HiMA-limestone is a 47% smaller than HiMA-granite). When comparing the

aging effect in the CTindex, the values were considerably reduced, but the same trend of the results

from A0 aging condition was present (i.e. Mixtures with HiMA present a better performance based

on the CTindex). Therefore, it is possible to conclude that aging has a great impact in the behavior

of cracking in PFC mixtures. For example, the CTindex for HiMA-granite was reduced in a 77.5%

and PMA-granite in a 56.1%. Although the reduction was higher for HiMA-granite, it still was 2.4

times the PMA-granite CTindex value.

23

Figure 12. CTindex values for the four different mixture with unaged and aged states.

3.2.4 Durability

The Cantabro abrasion loss test is specified in the AASHTO TP 108 standard (AASHTO

2014b). In summary, it consists of preparing cylindrical specimens 150 mm diameter by 114.3 mm

height and subjecting them to 300 revolutions in the Los Angeles abrasion machine without the

steel spheres. After the test is concluded, the percent abrasion loss (i.e. % mass loss) is determined

as the difference between the initial and final mass of the tested specimen. Mixtures with larger

mass loss are considered more prone to degradation. Despite being a simple test, several studies

have demonstrated that the results correlate well with field performance (Alvarez et al. 2008,

Arámbula-Mercado et al. 2016). This correlation is believed to be due to an existing relation

between the tumbling action of the specimen inside the drum and the resistance of the stone-on-

stone contacts within the microstructure of the PFC mixture. The impact of aging was further

evaluated in this test by testing specimens in the three different aging states considered in previous

tests: A0, A5 and A10. Figure 13 summarizes the results for the three aging states.

For the evaluation of the Cantabro test (i.e. A0 aging condition), all test specimens pass since they

presented a percentage weight loss less than 20%. In fact, the highest value was obtained for the

mixture PMA-granite. Meanwhile mixtures with HiMA obtained the least weight loss. These

partial results would indicate that mixtures with HiMA binder are the ones that would have a better

performance in field, meanwhile PMA-granite the worse performance. But when evaluated the

A0 A5

HiMA granite 1565.38 352.57

HiMA limestone 912.47 189.03

PMA granite 337.37 141.90

PMA limestone 233.40 68.57

0

500

1000

1500

2000

2500

CT

In

dex

24

materials in the long-term, the results indicate that the worst material is PMA-limestone. The rate

at which this material increases greatly and at A5 aging state, the percentage weight loss of PMA-

limestone surpasses that of PMA-granite. At A10 aging state, PMA-limestone presented a

difference of 45% weight loss in comparison to HiMA-limestone. Even at A10 aging condition, in

average the results of mixtures with HiMA pass the maximum percentage weight loss allowed (i.e.

less than 20%). Consequently, mixtures with HiMA binder are expected to present a better

durability since they maintain their properties over time, which corroborates the results from the

binder characterization and some previous tests. In terms of the aggregate type, mixtures with

granite generally present a lesser percentage weight loss of material than those with limestone.

There is only one exception, which is for A0 caging state for mixtures with PMA.

Figure 13. Cantabro test results for three aging states.

A0 A5 A10

HiMA-granite 4.3% 10.1% 14.7%

HiMA-limestone 3.9% 8.1% 19.0%

PMA-granite 13.4% 15.8% 30.7%

PMA-limestone 7.2% 21.4% 64.9%

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

%

Wei

gh

t L

oss

25

4. Numerical Simulation in Finite Elements

This section aims at complementing the experimental work reported in the previous section

by conducting 2D finite element (FE) simulations of the expected durability of the PFC’s. The

main objective of these FE computational mechanics models was to evaluate the response of the

different PFC mixtures under realistic field operational conditions. The mechanical models were

implemented in Abaqus® and their goal was to compare the mechanical response and expected

raveling susceptibility in long-term aging condition (i.e. after several years of service).

4.1 Model description

Since raveling is a stone-on-stone contact phenomenon (Alvarez, Epps, et al. 2010) and

the microstructure characteristics of these mixtures provides their strength and resistance (Alvarez

et al. 2008, Alvarez, Epps, et al. 2010, Alvarez, Mahmoud, et al. 2010, Manrique-Sanchez et al.

2016), the internal geometry of the PFCs is a fundamental component of the computational

mechanics models. The geometry of the PFC microstructures was obtained by applying X-Ray CT

and image analysis techniques on cylinders fabricated in the laboratory at a target 20% AV.

Portions of two-dimensional (2D) vertical cuts of the CT scans were selected to obtain three PFC

microstructure replicates with two different thicknesses: 2 cm and 4 cm.

Figure 14 illustrates the six microstructures used for the FE analyses. Although it is not noticeable

in these images, all the aggregates are coated with a thin mastic film. The thickness of this mastic

film was computed after assuming that the aggregates were spheres covered by a homogeneous

film of mastic, which resulted in a value of 150 µm. This implies that the stone-on-stone contacts

in these models are actually mastic-on-mastic contacts.

Figure 14. PFC microstructures replicates with 20% AV: (a) 2 cm thickness (b) 4 cm thickness

The PFC microstructure in

Figure 14 were characterized in terms of:

(a)

(b)

2 cm

4 cm

26

• The number of contacts per aggregates. This parameter is correlated with the

coordination number of a granular media (i.e. average number of contacts per

aggregate) and is considered a good indicator of the network connectivity (Chen

and Wong 2016). Two aggregates were considered to be in contact if the distance

between their edges was smaller than 0.2 mm. After obtaining the number of

contacts per aggregate, a probability density function (pdf) was adjusted to the

results of each microstructure.

• The average orientation of the aggregates. This parameter is defined as the angle

between the longest line along the aggregate and the horizontal axle. In a vector

image, the longest line corresponds to the highest distance between the two farthest

points on the contour or perimeter of the particle (i.e. in a range between -90° to

90°; see

•

• • Figure 15). The orientation of the aggregates for each microstructure was also

adjusted to a pdf.

• The total length of the stone-on-stone contacts. This parameter, that provides

information of the strength of the network, is computed as the sum of the total length

of the contacts per aggregate. In general, higher contact lengths indicate stronger

networks since this condition is related to better stress distributions and friction

between aggregates (Sefidmazgi et al. 2012).

Figure 15. Particle orientation

Table 10. Characteristics of the OGCFC mixtures microstructures

27

PFC

thickness Replicate

Number of contacts per

aggregate Orientation of particles

Total contact

length (mm) pdf μ* σ** pdf (°) σ (°)

2 cm

1 Normal 3.0 1.1 Normal -8.0 39.5 410

2 Log-

Logistic 3.3 1.6 Normal -7.7 42.9 650

3 Normal 2.7 1.5 Normal 4.5 41.0 591

4 cm

1 Weibull 2.6 1.8 Normal -5.1 46.1 1101

2 Logistic 3.4 1.4 Normal -3.8 44.0 1083

3 Normal 3.1 1.4 Normal -3.0 42.2 1077

*: mean value

**: standard deviation

Results from Table 10, present the information required to characterize the microstructures used

for the numerical simulations. First, in terms of the number of contacts, the average value for all

the mixtures was of 3.0 contacts per aggregate, although they present high values of COV (i.e. up

to 69%). This variability implicate that the number of contacts vary depending on the

microstructure. Whereas the results of the orientation of the aggregates show that they prefer being

horizontal or close to 0° (i.e. values in between -8° and 4.5°). Although it is possible to infer that

there are particles that are vertically oriented since they also present high values of standard

deviation. Finally, in terms of the length of the contacts, mixtures with 4 cm in thickness have in

average 37% more contacts and twice as much length in contacts in comparison to PFC structures

with 2 cm in thickness.

The FE pavement structure geometry consists of three main layers: i) a PFC, ii) an equivalent base

layer, and iii) a subgrade (see Figure 16). The equivalent base and subgrade layers were assumed

continuum, isotropic, and homogenous materials. This simplification allowed a reduction in the

computational cost, permitting to focus on the microstructural phenomena occurring within the

PFC layer. The horizontal direction in the vertical sides of the models was restrained, allowing

only movements in the vertical direction, and all degrees of freedom were restrained at the bottom

of the model. Figure 16 shows the selected pavement geometry with a 4 cm PFC microstructure

and the mastic-on-mastic contacts between the aggregates within the PFC.

28

Figure 16. FE pavement model in Abaqus®

The FE pavement model included three types of finite elements. The equivalent pavement base,

the subgrade, and the aggregates were meshed using 3-node linear elements (i.e. type CPE3 in

Abaqus®). A seed of 15 mm was used to mesh the equivalent base and subgrade layers of the

pavements and a seed of 30 mm was used to mesh the aggregate particles. The mastic films were

meshed using a 4-node bilinear quadrilateral element (i.e. type CPE4R in Abaqus®) with a seed of

0.075 mm. In the models simulating the long-term field performance of the PFCs, the mastic-on-

mastic contacts were meshed using 4-node two-dimensional cohesive elements (i.e. type COH2D4

in Abaqus®). Approximately, the PFC layer was meshed with a total of 460,000 elements and the

pavement structure with a total of 6,200 elements. Figure 17 shows in detail the mesh of a portion

of an PFC layer of 4 cm and part of the mesh of the base layer.

4 cm

Continuum Equivalent

Pavement Base

Subgrade

≈70 cm

PFC layer

60 cm

40 cm

29

Figure 17. Global mesh of a FE model with a 4 cm PFC

4.2 Loading Conditions

The models were subjected to the pass of a half single axle load of 49.7 kN at a speed of

88.5 km/h, as observed in

Figure 18. This loading magnitude and velocity represent typical traffic conditions in

highways with OFGC in the state of Florida (LTTP 2015). The tire-pavement interaction included

a vertical force, that generates the contact pressure on the PFC, and a friction force. The frictional

force was 1.2 kN and it was using the data by Milne et al. (2004), who specified that the friction

force could be equivalent to 2.5% of the maximum vertical force. Finally, the contact pressure was

of 0.9 MPa, which produces a loading contact radio of 12.5 cm.

PFC 2cm

Equivalent

Base Layer

Vertical component

Stress [MPa]

30

Figure 18. Loading application and vertical stress for a 2 cm thick PFC microstructure

4.3 Material Properties

The two types of aggregates correspond to those used in the experimental portion of the

work: i) granite and ii) limestone. Both aggregates were modeled as linear elastic materials with a

Poisson's ratio of 0.30 and an elastic modulus of 50,000 MPa for granite and 27,000 MPa for

limestone (Rummel 1991). The layers of the equivalent base layer and subgrade were also modeled

as linear elastic materials. The elastic modulus of the equivalent base was 520 MPa and the

Poisson's ratio 0.35, and for the subgrade layer the elastic modulus was 100 MPa and the Poisson's

ratio 0.45.

The contacts in between aggregate was simulated as mastic-on-mastic contacts. The mechanical

response of the four mastics coating the aggregate particles in the PFCs were determined to include

this information in the FE models. The linear viscoelastic properties of the mastics were determined

under PAV20 aging state (i.e. long-term aging). A DSR was employed to perform the

measurements. The procedure consisted of conducting temperature sweep tests (i.e. 10°C to 70°C,

in increments of 10°C) and frequency sweep tests (e.g., 37.5, 30, 25, 20, 15, 10, 5, 1, 005, 0.01

rad/s) using the parallel plate geometry. Similar to the characterization of the binders, the resulting

parameter of interest is the dynamic shear modulus, |G*|, for the four combinations of binder and

aggregates at the two aging states. The results from the frequency-sweep test were transformed

from the frequency (i.e. dynamic modulus) to the time domain (i.e. relaxation modulus). Thus, the

parameters of the Prony series of each mastic at a reference temperature of 30°C were determined.

These series were normalized with respect to the instantaneous shear modulus to accomplish with

the input requirements of the constitutive linear viscoelastic model of Abaqus®:

𝑔(𝑡) = 1 − ∑ 𝑔𝑖(1 − 𝑒−

𝑡

𝑝𝑖𝑛𝑖=1 ) (7)

Vertical component

Stress [MPa]

31

where:

g (t) = the normalized shear relaxation modulus of the material with respect to the instantaneous

shear modulus (G0) as a function of time (t).

ρi = is the ith relaxation time parameter of the Prony series.

gi = the Prony Series parameter Gi divided by the instantaneous shear modulus (i.e. 𝑔𝑖 =𝐺𝑖

𝐺0).

The relationship among the instantaneous modulus, G0, the long-term shear modulus, G∞, and the

Prony Series parameters, Gi, is:

𝐺0 = 𝐺∞ + ∑ 𝐺𝑖𝑛𝑖=1 (8)

Table 11 present the Prony series parameters of the four mastics in the long-term aging conditions.

In accordance with the experimental results, the mastics with PMA have a higher instantaneous

modulus than those with HiMA. Also, the instantaneous modulus of the mastics in the long-term

conditions are, on average, 200 MPa larger than those in the short-term due to oxidation,

independently of the type of binder.

Table 11. Prony Series for mastic in the long-term aged condition at 30°C

HiMA-

granite

HiMA-

limestone

PMA-

granite

PMA-

limestone

i ρi [s] Gi [Pa] Gi [Pa] Gi [Pa] Gi [Pa]

1 1.0x10-6 9.8x107 1.3x108 1.2x108 1.3x108

2 1.0x10-5 2.1x107 7.9x106 4.1x107 3.1x107

3 1.0x10-4 2.4x107 7.7x106 4.2x107 3.9x107

4 1.0x10-3 1.1x107 7.7x106 2.3x107 2.2x107

5 1.0x10-2 5.4x106 5.1x106 1.3x107 1.2x107

6 1.0x10-1 2.3x106 1.4x106 6.0x106 6.1x106

7 1.0x100 7.3x105 4.7x105 2.3x106 2.3x106

8 1.0x101 2.1x1005 1.3x105 7.3x105 8.0x105

9 1.0x102 5.1x104 3.2x104 2.0x105 2.2x105

10 1.0x103 1.0x104 8.3x103 4.3x104 6.9x104

11 1.0x104 1.8x10-8 1.0x10-3 1.1x104 1.3x10-3

12 1.0x105 2.3x10-9 1.0x10-4 5.1x10-3 1.3x10-3

13 1.0x106 3.2x103 1.0x10-5 2.6x10-5 1.2x10-3

32

E0 (MPa) 456.0 449.1 700.8 677.2

Poisson’s

ratio 0.40

A total of 24 models were evaluated that resulted from combining: six PFC microstructures (i.e.

three replicates for each thickness), four material combinations (i.e. two aggregate types, granite

and limestone, and two binder types, HiMA and PMA) and two conditions: i) short-term and ii)

long-term aging. Recalling, the short-term aging resembles the test condition for A0 of the

mixtures, which is the “as constructed” state, while the long-term aging condition resembles the

A5 aging condition of the mixtures or around 2-3 years of the in-service state. Thus, the mixture

‘granite-PMA’ corresponds to the PFC mixtures composed of granite and PMA.

4.4 Fracture model within the PFC microstructure

The FE models in the long-term aging condition incorporated Cohesive Zone Modeling

(CZM) elements to simulate actual fracture within the mastic-on-mastic contacts. The inclusion of

these fracture elements was due to the fact that, if constructed properly, raveling in PFCs is

expected to initiate after several years of service, when the mastic at the contacts have been exposed

to both, mechanical and weather degradation.

The mechanical response of the CZM elements was defined through a bilinear traction separation-

law, as observed in Figure 19. This law relates the relative displacement between the two parallel

faces of the CZM element caused by the applied stresses (tensile stress for Mode I of failure and

shear stress for Mode II of failure). In this law, the initial stiffness (K) determines the mechanical

response of the element prior to reaching the maximum tensile stress of the material (σmax). From

that point on, the element is not able to support tensile stresses, a softening process occurs (i.e. a

gradual reduction of stiffness) and the displacement among the faces of the element continues

increasing until reaching the maximum fracture displacement (δc). When this occurs, the material

fractures (i.e. the CZM element physically disappears from the model), which promotes the

propagation of the crack through the mastic-on-mastic contacts. The area under the stress-

displacement curve corresponds to the fracture energy of the material.

33

Figure 19. Traction-separation law of the CZM elements; modified after Caro (2009)

A common problem with CZM is that they are incorporated as physical elements in the models,

but they do not exist in reality, a condition that tends to increase the compliance of the PFC

microstructures. Thus, the initial stiffness of each mastic material (K) was calibrated after

guaranteeing that the mechanical response of each PFC model was equivalent to that of the model

that did not include CZM elements but only regular linear viscoelastic mastic elements at the

contacts. This process considered the research conducted by Salve et al. (2011) Aragão et al.

(2012), Aragão et al. (2017) and Haloui et al. (2018). Also, the magnitude of the initial stiffness

of each mastic was reduced after considering that after certain years of service the material has

overcome some mechanical degradation (i.e. fatigue damage, based on the FREI parameters

obtained from the PLAS tests). Also, the calibration of the input parameters of the CZM traction-

separation laws considered the results reported by Aragão et al. (2012, 2017), in which the effective

displacement (i.e. δi /δc from Figure 19) of typical asphalt and mastic materials is at least 5x10-4.

Finally, the fracture energy of the CZM was determined using SCB test results, after considering a

reduction of near 80% to resemble the critical condition in which raveling should initiate due to

the fatigue and other field-related degradation that has occurred in the material. This assumption is

justified by existing literature that indicates that aging and field deterioration after 5-6 years can

indicate a reduction of up to 70% of different performance properties (i.e. fracture energy, cohesive

bond energy) for dense-graded hot mix asphalt (Walubita et al. 2005, Bhasin et al. 2007, Saeidi

and Aghayan 2016).

A summary of the input parameters for the traction-separation law of the CZM elements for Mode

I of failure is presented in Table 12. Since there was not available experimental data to define the

constitutive response of these elements for Mode II of failure, it was assumed that the mastic

materials had isotropic fracture properties.

Table 12. Input parameters of the CZM traction-separation laws

Mixture Type Stiffness

(MPa/mm)

Initial

Displacement

Fracture

Energy (N/mm)

2 3 54K

σmax

δi δc δ (relative

displacement)

σ

Fracture

Energy

Prior to

damage

initiation

Damage

Evolution

Crack

Propagation

Adhesive

Bond Strength

1

1 3 52 4Cohesive zone

External load

34

(mm)

HiMA-Granite 45000.00 1.05x10-4 0.49

PMA-Granite 15292.97 1.80x10-4 0.21

HiMA-Limestone 44670.06 1.05x10-4 0.37

PMA-Limestone 14745.97 1.80x10-4 0.24

As an example of the results obtained from the simulations, Figure 20 presents the output variable

‘SDEG’ on three mastic-on-mastic contacts, after the wheel has passed over an PFC. The SDEG is

a state variable taking values between 0.0 –if no damage has occurred– and 1.0 –when the material

reaches dissipated energy or crack initiation–. Values closer to 1.0 indicate elements that are near

to crack (or near to ‘disappear’ in the model) since their dissipated energy is close to the fracture

energy. As observed, in this particular case some of the contacts have failed or are close to failure,

indicating the potential loss of these two mastic-on-mastic contacts.

Figure 20. SDEG values for contacts in an PFC microstructure

The susceptibility to raveling in these models was quantified using a new parameter called

“Remaining Energy” or ER of the CZM elements. This parameter is defined as the difference

between the fracture energy of the material and the energy dissipated by the CZM element after

the pass of the wheel load (in N/mm). When the energy dissipated by a CZM element after the pass

of the wheel load reaches the fracture energy of the material (i.e. ER equal to zero), a crack appears

at that specific location, indicating the initiation of the contact failure. On the contrary, CZM

elements that dissipated low amounts of energy after the application of the load (i.e. high ER values)

are still resistant to the initiation of raveling processes. Thus, microstructures having more ER in

the contacts are more resistant to raveling.

In order to better assess the overall behavior of the microstructures with respect to ER, the results

were adjusted to a log-normal distribution with a 95% confidence level. Figure 21 presents an

example of the pdf distribution of the ER values obtained for one replicate of a 2 cm PFC

microstructure. Moreover, in order to capture not only the mean value of ER but also the variability

of these data, the mean plus one standard deviation (µ+σ) of the pdf of ER, or (µ+σ)ER, was selected

35

as the evaluation parameter to compare the susceptibility to raveling of the different PFC mixtures.

Based on the definition of ER, higher values of (µ+σ)ER indicate less probability of a mixture to

raveling.

Figure 21. Pdf of ER for a replicate of the 2 cm HiMA-granite PFC

Table 13 presents the results of the mean (), standard deviation () of ER, and the mean plus

standard deviation (+)ER for the PFC mixtures evaluated. Recalling, ER is the remaining energy

of each CZM element after the pass of the wheel load; larger values of this parameter indicate that

the element can still resist stresses before failure, values near 0 indicate that the element is

approaching to fracture, and values equal to 0 indicate that the CZM element has failed. Table 14

and Figure 22 summarize the results for each type of mixture and the two PFC thicknesses

evaluated.

Table 13. ER parameter in each PFC microstructure evaluated

36

Micro-

structure Replicate Aggregate Binder

Results of ER (N/mm)

μ ER

(x10-3)

σ ER

(x10-3)

(μ+σ) ER

(x10-3)

2 cm

1

Granite HiMA 107 371 478

PMA 45 152 197

Limestone HiMA 93 469 562

PMA 48 148 196

2


PMA 14 30 75


PMA 15 38 74

3


PMA 30 103 133


PMA 38 162 200

4 cm

1


PMA 34 105 139


PMA 38 118 156

2


PMA 28 98 127


PMA 33 107 140

3


PMA 10 102 124


PMA 27 138 165

Table 14. ER parameter to evaluate raveling in the PFC mixtures

PFC thickness Mixture Type

Average ER.

(N/mm)

(x10-1)

COV

2 cm PMA-granite 1.3 45.0%

37

PFC thickness Mixture Type

Average ER.

(N/mm)

(x10-1)

COV

PMA-limestone 1.6 45.7%

HiMA-granite 2.7 71.0%

HiMA-limestone 2.8 88.0%

4 cm

PMA-granite 1.3 6.0%

PMA-limestone 1.5 8.0%

HiMA-granite 2.5 55.0%

HiMA-limestone 2.2 23.0%

Figure 22. (+)ER results for the PFC´s

Results from the numerical simulation, presented in Figure 22, indicate that PFC mixtures with

HiMA binder are more resistant to raveling since they present higher values of (+)ER. For

instance, mixtures with HiMA asphalt binder are in average 1.9 times bigger than mixtures with

PMA asphalt binder. In terms of the aggregate type, there is no significative difference in between

results (i.e. for the PFC 2 cm thick HiMA-limestone is 1.03 times HiMA-granite). In terms of the

PFC thickness, the differences are not significative, especially for the PMA PFC mixtures. When

taking into account the variability of the results, it was found that the results of the simulations

between replicates presented high values of COV, specially to those that had HiMA asphalt binder.

For example, for the 2 cm thick PFC mixture, HiMA-granite had a COV of 71% and HiMA-

limestone had a COV of 88%. On the other hand, for the same thickness, PMA-granite has a COV

of 45% and PMA-limestone of 46%. This variability of the results was due to high difference in

the initial fracture energy. Since PMA suffered severe damage in the contacts, the remaining

contacts presented low values of (+)ER, meanwhile HiMA mixtures had just the initiation of the

damage and consequently the (+)ER to failure within each PFC microstructure varied

considerably.

These results indicate, that the susceptibility to raveling in the long-term condition is strongly

corelated to the asphalt binder and consequently the mastic on the contacts. This observation is in

0.E+00

1.E-01

2.E-01

3.E-01

4.E-01

5.E-01

6.E-01

HiMA-granite HiMA-limestone PMA-granite PMA-limestone

(µ+

σ)

of

ε.R

[ N

/mm

]

Mixture Material

2cm Thick OGFC 4cm Thick OGFC

38

accord results from other studies that indicate that raveling is a problem of the mastic-on-mastic

contacts, and therefore other factors are of less importance in terms of the raveling potential of the

PFC mixtures.

5. Conclusions and Recommendations

A comprehensive study was conducted to evaluate the impact of high dosages of polymer

in asphalt binder when used in porous friction courses in order to increase durability and

mechanical performance. This was accomplished through a series of experimental tests on PMA

and HiMA asphalt binders and mixtures with two types of aggregates. These series of tests

evaluated the mechanical performance of the binder and the durability and fracture properties in

the mixtures. Furthermore, the study was complemented with the use of finite element simulations

39

to assess the performance on real life conditions on a 2-D model. Although major difficulties were

encountered along the process, especially with some numerical instabilities in the FE model, this

study allowed to understand the ultimate objective of this research which a comparative evaluation

of the implementation of HiMA binders in contrast to PMA asphalt binder in PFC mixtures is.

With respect to the asphalt binder characterization, HiMA binder presented a UTI and ΔTc higher

than those of the PMA. Also, it indicated better results for Glover-Rowe parameter (i.e. resistance

to cracking in aged conditions) in various oxidation conditions. Although this asphalt presented a

lower modulus in the linear viscoelastic properties, this is not considered a drawback of the material

based on two observations. First, the base binders used in these highly polymer modified asphalts

usually include some oils to increase workability and, based on previous research, these asphalts

have better ductility, which may increase the durability in mastics due to a better relaxation

response of the binder in the mastic-on-mastic contacts. Finally, in terms of the fatigue response of

the binders, HiMA presented greater values of the FREI parameter for all aging conditions, and

therefore a better resistance to fatigue cracking than PMA asphalt binder.

In terms of the mixtures, four different mixtures were fabricated with the two asphalt binders

previously characterized and two aggregates, limestone and granite. The fracture properties of the

mixtures were evaluated through the SCB and IDEAL test, meanwhile the durability was evaluated

using the Cantabro test. In terms of the fracture properties, both tests indicated a better performance

of the HiMA binder in contrast to the PMA asphalt binder. It was also observed that typically

mixtures with granite aggregate presented better results rather than mixtures with limestone

aggregate. Based on the Cantabro test, it is possible to infer that mixtures with HiMA asphalt binder

have an increased durability, especially when evaluating the performance in aged conditions. For

example, mixtures at a A10 condition with PMA-limestone have a 46% more weight loss than

mixtures with HiMA-limestone or a 50% more than HiMA-granite.

Finally, in terms of the numerical modelling, the same four different combinations of materials

(PFCs) were evaluated in six PFC microstructures to evaluate its performance in real life

conditions. Results obtained showed that mixtures with HiMA binder have less susceptibility to

raveling than mixtures with PMA, independently of the aggregate type of the PFC thickness.

Furthermore, the thickness of the PFC was not significantly different in regard to the susceptibility

to raveling.

To conclude, based on the results obtained from the characterization of both the binder and the

numerical modeling, the HiMA binder showed an enhanced long-term mechanical performance in

comparison to PMA asphalt binder. Also, HiMA binders tend to maintain its superiority at various

aging conditions. When evaluated under operational conditions, it was observed that HiMA binders

were less susceptible to raveling than mixtures with PMA asphalt binder. In general terms, the PFC

mixtures could be ranked based on the previous results from the best to the worse as; i) HiMA-

granite, ii) HiMA-limestone, iii) PMA-granite, and iv) PMA-limestone.

Some further research may be needed in terms of the use of these binders in PFC mixtures under

winter conditions, as it may solve the challenges presented in regions with strong winters. Also, it

should be studied the increase of the dosage of polymer, and its implication in terms of durability

and cost/effectiveness of such change.

40

6. Bibliography

AASHTO, 2002. Standard Practice for Mixture Conditioning of Hot Mix Asphalts.

AASHTO, 2013. Standard Method of Test for Materials finar Than No. 200 sieve in Mineral

Aggregates by Washing.

AASHTO, 2014a. Standard Specification for Performance-Graded Asphalt Binder Using Multiple

Stress Creep Recovery ( MSCR ).

AASHTO, 2014b. Standard Method of Test for Determining the Abrasion Loss of Asphalt Mixture

Specimens.

41

AASHTO, 2016. Standard Method of Test for Determining the Fracture Potential of Asphalt

Mixtures Using Semicircular Bend Geometry ( SCB ) at Intermediate Temperature Standard

Method of Test for Mixtures Using Semicircular Bend Geometry.

Alvarez, A.E., Epps-Martin, A., Estakhri, C., and Izzo, R., 2009. Determination of Volumetric

Properties for Permeable Friction Course Mixtures,. Journal of Testing and Evaluation, 37

(1), 1–10.

Alvarez, A.E., Epps-Martin, A., Estakhri, C.K., and Izzo, R., 2008. Evaluation of Durability Tools

for Porous Friction Courses. In: Transportation Research Board TRB, ed. Proceedings of the

Transportation Research Board 87th Annual Meeting. Washington D.C.

Alvarez, A.E., Epps, A., and Estakhri, C., 2010. Internal structure of compacted permeable friction

course mixtures. Construction and Building Materials, 24 (6), 1027–1035.

Alvarez, A.E., Mahmoud, E., Martin, A.E., Masad, E., and Estakhri, C., 2010. Stone-on-Stone

Contact of Permeable Friction Course Mixtures. Journal of material in Civil Engineering, 22

(11), 1129–1138.

Aragão, F.T.S., Badilla-Vargas, G.A., Hartmann, D.A., Oliveira, A.D. de, and Kim, Y.R., 2017.

Characterization of temperature- and rate-dependent fracture properties of fine aggregate

bituminous mixtures using an integrated numerical-experimental approach. Engineering

Fracture Mechanics, 180, 195–212.

Aragão, F.T.S. and Kim, Y.R., 2012. Mode I Fracture Characterization of Bituminous Paving

Mixtures at Intermediate Service Temperatures. Experimental Mechanics, 52 (9), 1423–1434.

Arámbula-Mercado, E., Hill, R.A., Caro, S., Manrique-Sanchez, L., Park, E.S., and Fernando, E.,

2016. Understanding Mechanisms of Raveling to Extend Open Graded Friction Courses

(OGFC) Service Life. College Station, Texas.

ASTM, 2011. Standard Test Method for Theoretical Maximum Specific Gravity and Density of

Bituminous Paving Mixtures.

ASTM, 2017a. Standard Test Method for Materials Finer than 75-µm ( No . 200 ) Sieve in Mineral

Aggregates by Washing 1, (200), 8–10.

ASTM, 2017b. Standard Test Method for Percent Air Voids in Compacted Asphalt Mixtures, 15–

17.

ASTM, 2017c. Standard Test Method for Thickness or Height of Compacted Asphalt Mixture

Specimens.

Bhasin, A., Howson, J., Masad, E., Little, D., and Lytton, R., 2007. Effect of Modification

Processes on Bond Energy of Asphalt Binders. Transportation Research Record, 1998 (1),

29–37.

Bondt, A. De, Plug, K., Van de Water, J., The, P., and Voskulien, J., 2016. Development of a

durable third generation Porous Asphalt with a high noise. In: Proceedings of E&E Congress.

Prague.

Caro, S., 2009. A Coupled Micromechanical Model Of Moisture-Induced Damage In Asphalt

Mixtures: Formulation And Applications. Texas A&M University.

42

Chen, J.S., Wang, T.J., and Lee, C. Te, 2018. Evaluation of a highly-modified asphalt binder for

field performance. Construction and Building Materials, 171, 539–545.

Chen, M.J. and Wong, Y.D., 2016. Evaluation of the development of aggregate packing in porous

asphalt mixture using discrete element method simulation. Road Materials and Pavement

Design, 0629 (February), 1–22.

Christensen, D. and Anderson, D., 1992. Interpretation of dynamic mechanical test data for paving

grade asphalt cements (with discussion). Journal of the Association of Asphalt Paving

Technologists.

Cooley Jr., L.A., Brumfield, J.W., Mallick, R.B., Mogawer, W.S., Partl, M., Pulikakos, L., and

Hicks, G., 2009. Construction and maintenance practices for permeable friction courses.

National Cooperative Research Program Report 640 / Transportation Research Board.

Washington D.C.

Dell´acqua, G., De Luca, M., and Lamberti, R., 2011. Indirect skid resistance measurement for

porous asphalt pavement management. Transportation Research Record, 2205, 147–164.

Florida Department of Transportation, 2018. Florida Department of Transportation Standard

Specifications.

Gibbs, D.C., Iwasaki, R.H., Bernhard, R.J., Bledsoe, J.F., Carlson, D.D., Corbisier, C., Fults, K.W.,

Hearne, T.M.J., McMullen, K.W., Newcomb, D.E., Roberts, J.H., Rochat, J.L., Scofield, L.A.,

and Swanlund, M.E., 2005. Quiet Pavement Systems in Europe. Washington D.C.

Glover, C., Davison, R.R., Domke, C.H., Ruan, Y., Juristyarini, P., Knorr, D.B., and Jung, S.H.,

2005. Development of a New Method for Assessing Asphalt Binder Durability with Field

Validation. College Station, Texas, Texas.

Haloui, Y. El, Omari, M. El, Absi, J., and Tehrani, F., 2018. Numerical Simulation of Fracture at

Asphalt Mastic Materials. In: The 3rd International Conference on Optimization and

Application. Meknes.

Hamzah, M.O., Abdullah, N.H., Voskuilen, J.L.M., and van Bochove, G., 2013. Laboratory

simulation of the clogging behaviour of single-layer and two-layer porous asphalt. Road

Materials and Pavement Design, 14 (1), 107–125.

Hernandez-Saenz, M.A., Caro, S., Arámbula-Mercado, E., and Epps Martin, A., 2016. Mix design,

performance and maintenance of Permeable Friction Courses (PFC) in the United States: State

of the Art. Construction and Building Materials, 111, 358–367.

Huddleston, I.J., Zhou, H., and Hicks, R.G., 1991. Performance evaluation of open-graded asphalt

concrete mixtures used in Oregon. Technologists, Association of Asphalt Paving, 3, 110–118.

Huurman, M., Moore, L., and Woldekidan, M.F., 2009. Porous Asphalt Raveling in Cold Weather.

Pavement Research and Technology, 3, 110–118.

Kandhal, P.S., 2002. Design, construction and maintenance of open-graded asphalt friction

courses. Information Series 115.

Kaseer, F., Yin, F., Arámbula-Mercado, E., Epps Martin, A., Daniel, J.S., and Salari, S., 2018.

Development of an index to evaluate the cracking potential of asphalt mixtures using the semi-

43

circular bending test. Construction and Building Materials, 167, 286–298.

Kim, Y.R., Castorena, C., Elwardany, M., Rad, F.Y., Underwood, S., Gundha, A., Gudipudi, P.,

Farrar, M.J., and Glaser, R.R., 2017. Long-Term Aging of Asphalt Mixtures for Performance

Testing and Prediction.

Kluttz, R., Dongré, R., Buzz Powell, R., Willis, J.R., and Timm, D.H., 2014. Performance and

Rational Design of Thin , Highly Modified Structural Pavements. In: International

Conference on Perpetual Pavements2. Columbus, Ohio, Ohio.

Kluttz, R., Jellema, E., Woldekidan, M.F., and Hurrman, M., 2013. Highly Modified Bitumen for

Prevention of Winter Damage in OGFCs. Sustainable and Efficient Pavements, (281), 1075–

1087.

Lefebvre, G., 1993. Porous Asphalt. In: Permament International Association of Road Congress.

LTTP, 2015. Traffic MEPDG Axle Distribution—States AL to IL.

Mallick, R.B., Mogawer, W.S., Poulikakos, L.D., Partl, M.N., Cooley, L.A., Brumfield, J.W., and

Hicks, G., 2009. Construction and Maintenance Practices for Permeable Friction Courses.

Washington D.C.

Manrique-Sanchez, L., Caro, S., and Arámbula-Mercado, E., 2016. Numerical modelling of

ravelling in porous friction courses (PFC). Road Materials and Pavement Design, 0629

(January), 1–22.

Martins, R., 2018. Personal Communication.

Mejias De Pernia, Y., 2015. Prediction of the Optimum Binder Content of Open-Graded Friction

Course Mixtures Using Digital Image Processing. University of South Florida. University of

South Florida.

Milne, T.I., Huurman, M., van de Ven, M.F.C., Jenkins, K.J., Scarpas, A., and Kasbergen, C., 2004.

Towards Mechanistic Behaviour of Flexible Road Surfacing seals using a Prototype FEM

Model. 8th Conference on Asphalt Pavements for Southern Africa, (September).

Mo, L.T., Huurman, M., Woldekidan, M.F., Wu, S.P., and Molenaar, A.A.A., 2010. Investigation

into material optimization and development for improved ravelling resistant porous asphalt

concrete. Materials and Design, 31 (7), 3194–3206.

Mo, L.T., Huurman, M., Wu, S., and Molenaar, A.A.A., 2014. Mortar fatigue model for meso-

mechanistic mixture design of ravelling resistant porous asphalt concrete. Materials and

Structures, 947–961.

Mo, L.T., Huurman, M., Wu, S.P., and Molenaar, A.A.A., 2007. Investigation into stress states in

porous asphalt concrete on the basis of FE-modelling. Finite Elements in Analysis and Design,

43, 333–343.

Mo, L.T., Huurman, M., Wu, S.P., and Molenaar, A.A.A., 2008. 2D and 3D meso-scale finite

element models for ravelling analysis of porous asphalt concrete. Finite Elements in Analysis

and Design, 44, 186–196.

Mo, L.T., Huurman, M., Wu, S.P., and Molenaar, A.A.A., 2011. Bitumen – stone adhesive zone

44

damage model for the meso-mechanical mixture design of ravelling resistant porous asphalt

concrete. International Journal of Fatigue, 33, 1490–1503.

Mobilite Intelligente, 2017. Programme de recherche et d´innovation dans les transports terrestres

[online]. Available from: http://www.transport-intelligent.net/r-d-et-

enseignement/programmes-de-recherche-en-france/??lang=fr.

Nicholls, J.C., 1997. Review of UK Porous Asphalt Trials. Crowthorne.

Orlen Asfalt, 2017. Orlen Asfalt [online]. Available from: http://www.orlen-

asfalt.pl/PL/InformacjeTechniczne/PortalWiedzy/Documents/flipbook/2015-HIMA-

EN/files/assets/basic-html/index.html#12 [Accessed 15 Feb 2017].

Peeters, B. and Blokland, G. v., 2007. The noise emission model for European road traffic.

IMAGINE deliverable, 6–12.

Polacco, G., Filippi, S., Merusi, F., and Stastna, G., 2015. A review of the fundamentals of

polymer-modified asphalts: Asphalt/polymer interactions and principles of compatibility.

Advances in Colloid and Interface Science, 224, 72–112.

Roque, R., Varadhan, A., Thai, T., Jaiswal, L., and Birgisson, B., 2006. EVALUATION OF

THICK OPEN GRADED AND BONDED FRICTION COURSES FOR FLORIDA UF

Project No: 4504-968-12. Florida Department of Transportation - Research Management

Center, 12 (March 2006).

Rowe, G.M., King, G., and Anderson, M., 2014. The Influence of Binder Rheology on the Cracking

of Asphalt Mixes in Airport and Highway Projects. Journal of Testing and Evaluation, 42 (5),

1063–1072.

Rummel, R., 1991. Physical Properties of the Rock in the Granitic Section of Borehole GPK1.

Geothermal Science and Technology, 3, 199–216.

Rungruangvirojn, P. and Kanitpong, K., 2010. Measurement of visibility loss due to splash and

spray: Porous, SMA and conventional asphalt pavements. International Journal Pavement

Engineering, 11, 499–510.

Saeidi, H. and Aghayan, I., 2016. Investigating the effects of aging and loading rate on low

temperature cracking resistance of core-based asphalt samples using semi-scircular bending

test. Construction and Building Materials, 126, 9.

Salve, A.K. and Jalwadi, S.N., 2011. Implementation of Cohesive Zone in ABAQUS to Investigate

Fracture Problems. National Conference for Engineering Post Graduates RIT, (November),

60–66.

Sefidmazgi, N.R., Tashman, L., and Bahia, H., 2012. Internal structure characterization of asphalt

mixtures for rutting performance using imaging analysis. Road Materials and Pavement

Design, 13 (SUPPL. 1), 21–37.

Suzuki, T., Hirato, T., Kiya, T., Takahashi, T., Watanabe, M., and Uesaka, K., 2010. Development

and Study of Polymer Modified Asphalt in Japan.

Texas A&M Transportation Institute, University of Nevada Reno, University of New Hampshire,

Epps-Martin, A., Arámbula-Mercado, E., Epps, J., Newcomb, D., Glover, C., Chowdury, A.,

45

Yin, F., Kaseer, F., Garcia Cucalon, L., Bajaj, A., Hajj, E., Morian, N., Pournoman, S., Daniel,

J., Rahbar-Rastegar, R., and King, G., 2017. Quarterly Progress Report NCHRP 9-58.

College Station, Texas.

Timm, D.H., Robbins, M.M., and Willis, J.R., 2012. Field and Laboratory Study of High-Polymer

Mixtures At the Ncat Test Track（Interim Report）. NCAT Report 12-08, (August).

Transportation Reseach and Innovation Monitoring and Information System, n.d. SILVIA [online].

Available from: https://trimis.ec.europa.eu/project/sustainable-road-surfaces-traffic-noise-

control.

Voskuilen, J. and Elzinga, F., 2010. New life for Porous Asphalt RAP in new Porous Asphalt.

International Conference on Asphalt Pavements. Nagoya.

Voskuilen, J. and van de Ven, M.F.C., 2010. Winter problems with Porous Asphalt in the

Netherlands. International Conference on Asphalt Pavements. Nagoya.

Voskuilen, J.L.M., van de Ven, M.F.C., and van Wieringen, J., 2004. Two-Lyaer Porous Asphalt

for Noise Reduction. In: 8th Conference on Asphalt Pavements for Southern Africa. Sun City,

1–16.

Walubita, L.F., Martin, A.E., Jung, S.H., Glover, C.J., Park, E.S., Chowdhury, A., and Lytton, R.L.,

2005. Comparison of Fatigue Analysis Approaches for Two Hot Mix Asphalt Concrete

(HMAC) Mixtures. Project 0-4468. College Station, Texas.

Watson, D., Tran, N.H., and Rodezno, C., 2016. NCHRP 9-50.

West, R., Timm, D., Willis, J.R., Powell, B., Tran, N., Watson, D., Sakhaeifar, M.S., Brown, E.R.,

Robbins, M., Vargas-Nordcbeck, A., Villacorta, F.L., Guo, X., and Nelson, J., 2012. Phase

IV NCAT Pavement Test Track Findings: Final Report. NCAT Report 12-10.

Willis, J.R., Timm, D.H., and Kluttz, R., 2016. Performance of a Highly Polymer-Modified Asphalt

Binder Test Section at the National Center for Asphalt Technology Pavement Test Track.

Transportation Research Record: Journal of the Transportation Research Board, 2575, 1–9.

Willis, R., Timm, D., West, R., Powell, B., Robbins, M., Taylor, A., Smit, A., Tran, N., Heitzman,

M., and Bianchini, A., 2009. Phase III NCAT test track findings. NCAT Report, (December),

8–9.

Zhou, F., Im, S., Sun, L., and Scullion, T., 2017. Development of an IDEAL cracking test for

asphalt mix design and QC/QA. Road Materials and Pavement Design, 18, 405–427.

Zhu, J., Birgisson, B., and Kringos, N., 2014. Polymer modification of bitumen : Advances and

challenges. European Polymer Journal, 54, 18–38.