University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln

Nebraska Department of Transportation Research
Nebraska LTAP
Reports

12-2015

Development of a Semicircular Bend (SCB) Test

Method for Performance Testing of Nebraska
Asphalt Mixtures
Gabriel Nsengiyumva
University of Nebraska-Lincoln, [email protected]

Yong-Rak Kim
University of Nebraska - Lincoln, [email protected]

Taesun You
University of Nebraska - Lincoln

Follow this and additional works at: https://digitalcommons.unl.edu/ndor

Part of the Transportation Engineering Commons

Nsengiyumva, Gabriel; Kim, Yong-Rak; and You, Taesun, "Development of a Semicircular Bend (SCB) Test Method for Performance
Testing of Nebraska Asphalt Mixtures" (2015). Nebraska Department of Transportation Research Reports. 171.
https://digitalcommons.unl.edu/ndor/171

This Article is brought to you for free and open access by the Nebraska LTAP at DigitalCommons@University of Nebraska - Lincoln. It has been
accepted for inclusion in Nebraska Department of Transportation Research Reports by an authorized administrator of DigitalCommons@University of
Nebraska - Lincoln.
 Nebraska
Transportation
Center

Report SPR-P1(15) MPMC07 Final Report

26-1121-4023-001

Development of a Semicircular Bend (SCB) Test Method

for Performance Testing of Nebraska Asphalt Mixtures
Gabriel Nsengiyumva
Graduate Research Assistant
Department of Civil Engineering
University of Nebraska-Lincoln
Yong-Rak Kim, Ph.D.
Associate Professor
Department of Civil Engineering
University of Nebraska-Lincoln
Taesun You, Ph.D.
Postdoctoral Research Associate
Department of Civil Engineering
University of Nebraska-Lincoln
2015
Nebraska Transportation Center
262 WHIT
2200 Vine Street
Lincoln, NE 68583-0851
(402) 472-1975

“This report was funded in part through grant[s] from the Federal Highway Administration [and Federal Transit Administration], U.S. Department of Transportation.
The views and opinions of the authors [or agency] expressed herein do not necessarily state or reflect those of the U.S. Department of Transportation.”
Development of a Semicircular Bend (SCB) Test Method for Performance Testing of
Nebraska Asphalt Mixtures

Gabriel Nsengiyumva
Graduate Research Assistant
Department of Civil Engineering
University of Nebraska-Lincoln

Yong-Rak Kim, Ph.D.

Associate Professor
Department of Civil Engineering
University of Nebraska-Lincoln

and

Taesun You, Ph.D.

Postdoctoral Research Associate
Department of Civil Engineering
University of Nebraska-Lincoln

A Report on Research Sponsored by

Nebraska Department of Roads

December 2015
 Technical Report Documentation Page
1. Report No 2. Government Accession No. 3. Recipient’s Catalog No.
SPR-P1(15) MPMC07

4. Title and Subtitle 5. Report Date

Development of a Semicircular Bend (SCB) Test Method for Performance Testing of December, 2015
Nebraska Asphalt Mixtures
6. Performing Organization
Code

7. Author/s 8. Performing Organization

Gabriel Nsengiyumva, Yong-Rak Kim and Taesun You Report No.

9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)

University of Nebraska-Lincoln (Department of Civil Engineering)

362N WHIT, 2200 Vine St., Lincoln, NE 68583-0856 11. Contract or Grant No.
26-1121-4023-001
12. Sponsoring Organization Name and Address 13. Type of Report and Period
Covered
Nebraska Department of Roads (NDOR)
1400 Highway 2, PO Box 94759, Lincoln, NE 68509
14. Sponsoring Agency Code

15. Supplementary Notes

16. Abstract
Granted that most distresses in asphalt (flexible) concrete (AC) pavements are directly related to fracture, it becomes clear
that identifying and characterizing fracture properties of AC mixtures is a critical step towards a better pavement design.
This report examines the testing variables of a reliable and practical semicircular bending (SCB) test for evaluating the
fracture characteristics of asphalt concrete mixtures at intermediate service temperature conditions. The first part of this
report investigates the repeatability of the SCB fracture test method by integrating a statistical-experimental approach to
identify testing variables of the SCB test that result in repeatable test results. Toward this end, five testing variables (the
number of testing specimens, specimen thickness, notch length, loading rate, and testing temperature) of the SCB test were
investigated due to their significant effects on mixture fracture characteristics. After statistical analysis of 18 specimens
tested with a typical testing variables, approximately, five to six specimens/replicates were found to be a reasonable sample
size that could properly represent asphalt concrete fracture behavior using the SCB test method. The coefficient of variation
(COV) of the mixture fracture energy was used to evaluate the effect of each variable on the repeatability of test results. A
range of 1 mm/min. to 5 mm/min. for the loading rate, a notch length from 5 mm to 25 mm, and a specimen thickness of
40 mm to 60 mm and a testing temperature of 15-40°C showed a good repeatability of fracture energy with small COVs
(≤15 %). The second part of this work is to investigate the sensitivity of the SCB test using the previously determined
testing variables. Fourteen different asphalt concrete (AC) mixtures collected from 12 field construction projects in
Nebraska were used in this task. The ANOVA test showed statistically significant differences between mixtures at a 95%
confidence level. Tukey’s HSD multiple-comparison analysis found similarities within mixtures of same types and
differences between mixture types. In addition, the fracture energy of bituminous mixtures increased with increasing amount
of virgin asphalt content in mixture. Overall, the SCB test method developed herein proved to be repeatable and sensitive
to changes in mixtures, and thus a promising tool for evaluating the fatigue fracture resistance of AC mixtures.
17. Key Words 18. Distribution Statement
Semicircular Bending; Asphalt Concrete;
Cracking; Fracture Test; Pavement.
19. Security Classification (of this report) 20. Security Classification (this page) 21. No. Of Pages 22. Price
Unclassified Unclassified 60
Form DOT F 1700.7 (8-72) Reproduction of form and completed page is authorized

ii
 Table of Contents

Acknowledgements ...................................................................................................................... viii

Disclaimer ...................................................................................................................................... ix
Abstract ........................................................................................................................................... x
Chapter 1 Introduction .................................................................................................................... 1
1.1 Research Objectives ................................................................................................ 3
1.2 Research Methodology ........................................................................................... 3
1.3 Organization of Report ........................................................................................... 4
Chapter 2 Literature Review ........................................................................................................... 6
2.1 Fracture Mechanics ................................................................................................. 6
2.2 Fracture Characterization of Asphalt Concrete Mixtures ....................................... 6
Chapter 3 Materials, Testing Facility, and Sample Fabrication .................................................... 17
3.1 Aggregate and AC Mixture ................................................................................... 17
3.2 Testing Facility ..................................................................................................... 19
3.3 Sample Fabrication and Test Set-up ..................................................................... 22
Chapter 4 SCB Test Method Development ................................................................................... 26
4.1 The Number of Testing Specimens....................................................................... 26
4.2 Specimen Thickness.............................................................................................. 29
4.3 Notch Length ........................................................................................................ 32
4.4 Loading Rate ......................................................................................................... 34
4.5 Testing Temperature ............................................................................................. 37
4.6 Summary of SCB Test Method Development ...................................................... 39
Chapter 5 SCB Testing of Nebraska Plant Produced Mixtures .................................................... 40
5.1 Project Selection ................................................................................................... 40
5.2 Material Collection and Sample Fabrication ........................................................ 41
5.2.1 Aggregates Gradation ................................................................................... 43
5.2.2 Mixture Characteristics ................................................................................. 44
5.3 Test Results and Discussion.................................................................................. 46
5.3.1 Relationship between virgin asphalt content to fracture energy ................... 52
Chapter 6 Summary and Conclusions ........................................................................................... 53

iii
References ..................................................................................................................................... 55

iv
 List of Figures

Figure 1-1 Research methodology used in this study. .................................................................... 4

Figure 2-1 Fracture tests for asphalt concrete mixtures: (a) single-edge notched beam (SEB) test,
(b) disk-shaped compact tension (DCT) test, and (c) semicircular bending (SCB) test. .... 7
Figure 2-2 Fracture energy (Gf) calculation: (a) fracture work (𝑊o) and (b) ligament area (Alig).
............................................................................................................................................. 9
Figure 2-3 Effect of loading rate (Aragão and Kim 2012). .......................................................... 10
Figure 2-4 Numerical simulation results of SCB fracture testing at different loading rates by
Aragão and Kim (2012). ................................................................................................... 10
Figure 2-5 A typical force-LPD curve from SCB test....................................................................11
Figure 2-6 Calculating critical J-Integral using two notches ........................................................ 13
Figure 2-7 SCB test specimen and configuration (in mm) adopted by Al-Qadi, Ozer et al. (2015).
........................................................................................................................................... 14
Figure 2-8 Calculation of the post peak slope (Al-Qadi, Ozer et al. 2015) .................................. 15
Figure 3-1 Mixture collection from a dump truck at asphalt concrete plant ................................ 18
Figure 3-2 UTM-25kN testing equipment .................................................................................... 20
Figure 3-3 Reading temperature of specimens ............................................................................. 22
Figure 3-4 SCB specimen fabrication process: (a) compacting, (b) slicing, and (c) notching ..... 23
Figure 3-5 Recording the thickness and ligament length of SCB specimen ................................. 24
Figure 3-6 Test set-up for semicircular bending (SCB): (a) specimen alignment before testing, (b)
specimen ready to be tested .............................................................................................. 25
Figure 4-1 Normality test result. ................................................................................................... 27
Figure 4-2 (a) Calculation of average standard deviation for k=5, (b) average standard deviation
for each assumed population size (k), and (c) assumed population size (k) with associated
minimum number of specimens (n) .................................................................................. 29
Figure 4-3 Fabrication of SCB specimens at different thicknesses .............................................. 30
Figure 4-4 Effect of thickness of specimens (t): (a) test results (average of six replicates) and (b)
fracture energy with standard error bars and COV of fracture energy for different
thicknesses ........................................................................................................................ 31

v
Figure 4-5 Effect of notch length: (a) test results (average of six replicates) and (b) fracture
energy with standard error bars and COV of fracture energy for different notch lengths ....
........................................................................................................................................... 33
Figure 4-6 Fracture profile at different notch lengths ................................................................... 34
Figure 4-7 Loading rates inputs .................................................................................................... 35
Figure 4-8 Effect of loading rate: (a) test results (average of six replicates) and (b) fracture
energy with standard error bars and COV of fracture energy for different loading rates .....
........................................................................................................................................... 36
Figure 4-9 Fracture profiles at different testing temperatures (front and back) ........................... 37
Figure 4-10 Effect of temperature: (a) test results (average of six replicates) and (b) fracture
energy with standard error bars and COV for different testing temperatures ................... 38
Figure 5-1 Field program: (a) construction in progress on highway 63 (CN: 12963) and (b) after
laboratory compaction of mixtures ................................................................................... 42
Figure 5-2 Gradation chart of five representative mixtures – sieve sizes raised to 0.45 power .......
........................................................................................................................................... 43
Figure 5-3 Visual comparison between the mixtures .................................................................... 44
Figure 5-4 SCB test results (fracture energy) of different mixtures with standard error bars ...... 46
Figure 5-5 Multiple-comparison procedures (Dowdy, Wearden et al. 2011)................................ 48
Figure 5-6 Relationship between virgin binder and fracture energy ............................................ 52

vi
 List of Tables

Table 2-1 Summary of the proposed SCB test methods and their testing variables ..................... 16
Table 3-1 Gradation and consensus properties of aggregates used ............................................... 19
Table 3-2 UTM-25kN key specifications ..................................................................................... 21
Table 4-1 Recommended variables for SCB test with approximate associated COV. .................. 39
Table 5-1 Field project selected for this study .............................................................................. 41
Table 5-2 Compaction temperature for each mixture ................................................................... 42
Table 5-3 Blending characteristics of mixtures selected ............................................................... 45
Table 5-4 Coefficient of variation of test results........................................................................... 47
Table 5-5 ANOVA: single factor about fracture energy................................................................ 49
Table 5-6 Tukey's HSD about fracture energy and mixture ranking ............................................ 50
Table 5-7 Mixture classification by their fracture energy ............................................................. 51

vii
 Acknowledgements

The authors thank the Nebraska Department of Roads (NDOR) for the financial support needed to
complete this study. In particular, the authors thank NDOR Technical Advisory Committee (TAC)
for their technical support and invaluable discussion/comments.

viii
 Disclaimer

The contents of this report reflect the views of the authors, who are responsible for the facts and
the accuracy of the information presented herein. This document is disseminated under the
sponsorship of the U.S. Department of Transportation's University Transportation Centers
Program, in the interest of information exchange. The U.S. Government assumes no liability for
the contents or use thereof.

ix
 Abstract

Granted that most distresses in asphalt (flexible) concrete (AC) pavements are directly related to
fracture, it becomes clear that identifying and characterizing fracture properties of AC mixtures is
a critical step towards a better pavement design. This report examines the testing variables of a
reliable and practical semicircular bending (SCB) test for evaluating the fracture characteristics of
asphalt concrete mixtures at intermediate service temperature conditions. The first part of this
report investigates the repeatability of the SCB fracture test method by integrating a statistical-
experimental approach to identify testing variables of the SCB test that result in repeatable test
results. Toward this end, five testing variables (the number of testing specimens, specimen
thickness, notch length, loading rate, and testing temperature) of the SCB test were investigated
due to their significant effects on mixture fracture characteristics. After statistical analysis of 18
specimens tested with a typical testing variables, approximately, five to six specimens/replicates
were found to be a reasonable sample size that could properly represent asphalt concrete fracture
behavior using the SCB test method. The coefficient of variation (COV) of the mixture fracture
energy was used to evaluate the effect of each variable on the repeatability of test results. A range
of 1 mm/min. to 5 mm/min. for the loading rate, a notch length from 5 mm to 25 mm, and a
specimen thickness of 40 mm to 60 mm and a testing temperature of 15-40°C showed a good
repeatability of fracture energy with small COVs (≤15 %). The second part of this work is to
investigate the sensitivity of the SCB test using the previously determined testing variables.
Fourteen different asphalt concrete (AC) mixtures collected from 12 field construction projects in
Nebraska were used in this task. The ANOVA test showed statistically significant differences
between mixtures at a 95% confidence level. Tukey’s HSD multiple-comparison analysis found
similarities within mixtures of same types and differences between mixture types. In addition, the
fracture energy of bituminous mixtures increased with increasing amount of virgin asphalt content
in mixture. Overall, the SCB test method developed herein proved to be repeatable and sensitive
to changes in mixtures, and thus a promising tool for evaluating the fatigue fracture resistance of
AC mixtures.

x
 Chapter 1 Introduction

Asphalt concrete (AC) pavement is susceptible to several types of distresses during its service life,
such as fatigue cracking, rutting, and thermal cracking. Typically, the majority of these distresses
are a result of repeated loading (fatigue) from traffic vehicles in combination with freezing and
thawing cycles associated with temperature variations throughout the seasons of the year. The
presence of these distresses directly and severely compromises the overall structural and functional
performance of the pavement, and consequently diminishes the service life and ride quality of
roads. Damage becomes more accentuated when fatigue cracking is combined with thermal
stresses, resulting in potholes that render the pavement virtually unusable. In addition, in cases
where the affected pavement is not rehabilitated in a timely manner, the distresses may provide
easy access to moisture, resulting in the accelerated deterioration of pavements. This inevitably
leads to increased repair costs that may strain the budget of a state’s department of transportation
(DOT). It becomes obvious that the pavement design process needs to take a combination of design
factors that cause these undesired distresses into consideration, such as traffic loads, environmental
effects, and material properties of AC mixture constituents, to increase reliability and service life
of pavements.
Among the aforementioned AC pavement distresses, fatigue cracking is the most critical
in pavements considering that once it occurs, it may lead to rapid pavement structure deterioration
and severely reduced ride quality. Thus, in order to mitigate this fatigue cracking, it is imperative
to explore and characterize the complex fracture mechanics behind crack initiation and
propagation in AC mixtures and extract fracture parameters to serve in the selection of better-
suited mixtures to resist cracking/fracture.
Currently, the Superpave performance grade (PG) specification evaluates cracking
behavior in asphalt concrete mixtures by only considering properties of asphalt binder. This
method used the dynamic shear rheometer (DSR), bending beam rheometer (BBR), and direct
tension test (DTT) developed by the Strategic Highway Research Program (SHRP). This approach,
however, fails to consider the aggregate portion of the AC mixtures, which makes up about
90~95% of the total weight of the AC. SHRP attempted the indirect tensile test (IDT) creep and
strength of AC performed according to AASHTO T322-07 (2007). The IDT is used to find critical
cracking temperature parameters that are then employed in the thermal cracking (TC) prediction

1
model implemented in the AASHTO Mechanistic-Empirical Pavement Design Guide (MEPDG).
As a result of using the empirical model, the IDT cracking parameters fall short of properly
characterizing crack initiation and propagation in AC mixtures, which are temperature, rate, and
mode dependent (Im, Kim et al. 2013, Im, Ban et al. 2014). Similarly, the thermal model in
AASHTO MEPDG fails to adequately address fatigue cracking, which mainly occurs at
intermediate temperatures (i.e., 15C ~ 30C).
Recently, the AC pavement community sought development of fracture mechanics-based
tests in order to properly address the aforementioned issues. Test methods include the single-edge
notched beam (SEB) test (Wagoner, Buttlar et al. 2005) and the disk-shaped compact tension
(DCT) test (Marasteanu, Dai et al. 2002, Wagoner, Buttlar et al. 2005). Experimental tests in
combination with a fracture mechanics model, such as a cohesive zone model (CZM), were
attempted to better identify fracture characterization in AC (Song, Wagoner et al. 2008, Shen and
Paulino 2011, Im, Ban et al. 2014(a), Im, Ban et al. 2014(b)). The incorporation of a fracture
mechanics model (e.g., cohesive zone model) into experimental tests is attractive in that it can be
used to gain insights into the isolation of crack formation energy from other sources of energy
consumption in fracture tests. In typical fracture tests of quasi-brittle materials, the total internal
energy, which is a result of the external work done, is composed of several sources of energy: the
recoverable strain energy, the energy dissipated by the fracture process, and the energy dissipated
due to material inelasticity. Consequently, this approach permits researchers to obtain information
from test results that were otherwise not feasible to obtain solely from laboratory tests. However,
since CZM requires calibration for experimental test results, advances in laboratory fracture tests
are needed in order to take full advantage of the approach. Among the several fracture test methods
in AC mixtures, the semicircular bend (SCB) test has gained increasing attention from the AC
pavement community due to its efficient, repeatable, and practical ways to characterize asphalt
concrete fracture behavior.
The SCB test results have shown sufficient testing sensitivities to testing variables such
as loading rate, specimen thickness, and testing temperature (Allen, Lutif et al. 2009, Kim, Lutif
et al. 2009, Li and Marasteanu 2009, Im, Kim et al. 2013). However, the selection of testing
variables that can provide statistically representative fracture characteristics of asphalt mixtures
has not been fully investigated. Thus, several studies (Li and Marasteanu 2009, Shu, Huang et al.
2010, Faruk, Hu et al. 2014) have performed SCB tests with testing variables selected somewhat

2
randomly based on previous experiences/observations, which typically leads to inconsistent and
non-repeatable fracture test results. In addition, it is not clear how many SCB specimens should
be tested in order to examine the fracture behaviors of an asphalt concrete mixture. Obviously, it
is necessary to explore the SCB test with testing variables that can improve the repeatability and
reliability of the fracture test results.

1.1 Research Objectives

The primary goal of this research is to investigate SCB testing variables so that the SCB test can
be used in the form of a reliable-repeatable test method, particularly to evaluate the fatigue-
cracking resistance of typical asphalt concrete mixtures. The specific objectives of this research
are as follows:
 To investigate the effect of individual SCB testing variables on asphalt concrete fracture
behaviors at intermediate service temperatures,
 To explore the SCB test method with testing variables that can provide reliable test results
with statistical repeatability (consistency) and practicality, and
 To investigate the sensitivity of the SCB test method with different AC mixtures collected
from field construction projects,

1.2 Research Methodology

To meet the objectives mentioned above, systematic testing efforts and statistical analyses were
integrated to investigate core testing variables, such as the minimum recommended number of
specimens for the SCB test, thickness of specimens, notch length, loading rate and the testing
temperature. First of all, based on an extensive literature review of SCB test, reasonable testing
variables were assumed. Using these variables, SCB testing results from a total of 18 specimens
were used for a statistical analysis that estimated the required number of specimens necessary with
a desired confidence level (i.e., 95%) and margin of error (i.e., 0.075). This integrated
experimental-statistical approach led to a recommended combination of the SCB testing variables
to characterize the cracking resistance of asphalt concrete mixtures at intermediate service
temperature conditions. With the minimum number of specimens investigated, the effect of each
of four c r i t i c a l testing variables (thickness of specimens, notch length, loading rate, and
temperature) was then explored by varying one variable at a time while others remained constant.

3
This allowed to isolate and characterize the effect of each variable on test results. The consistency
in the test results was evaluated by the coefficient of variation (COV) of fracture energies. The COV
is defined as the standard deviation divided by the mean (Dowdy, Wearden et al. 2011). Test-
analysis results would recommend ranges and values of testing variables that exhibited the
least/acceptable variability of test results. Subsequently, using the recommended testing variables,
the sensitivity to difference in AC mixtures was evaluated by testing 14 AC mixtures collected
from 12 field construction projects in Nebraska. Field projects were chosen to be representative of
all AC mixture types used in Nebraska. The research methodology employed in this study is
summarized in Figure 1-1.

Critical Testing Variables Range of Optimum Testing

Variables

Literature Review
Fourteen Mixtures
5 Number of specimens

Thickness 4 Fracture Energy

3 Notch Length

Multiple-comparison
Loading Rate 2

1 Temperature
Ranking of Mixtures
0

Figure 1-1 Research methodology used in this study.

1.3 Organization of Report

This report is composed of six chapters. At the end of this introduction, Chapter 2 covers literature
review on SCB test method as fracture testing method. Chapter 3 presents the material and testing
facility used in this study. Chapter 4 includes the process of determining the minimum number of
specimens for SCB test and SCB test method development by considering the effect of critical test
variables (i.e., specimen thickness, notch length, loading rate, and testing temperature) on fracture

4
energy and on repeatability of the test results. Chapter 5 covers testing of 14 Nebraska plant-
produced asphalt concrete mixtures collected from 12 separate field construction projects in
Nebraska using the testing variables previously developed in Chapter 4. This chapter attempts to
characterize the sensitivity of fracture parameters to changes in AC mixtures. Laboratory SCB test
results in a form of fracture energy (Gf) are statistically investigated for this purpose. Finally,
Chapter 6 summarizes the main findings and conclusions of this study.

5
 Chapter 2 Literature Review

Toward achieving the main goals of this study, a literature review was conducted on existing
methods of assessing fracture performance in asphalt concrete mixtures. This chapter includes key
studies conducted on the subject matter and summarizes relevant findings. Accordingly, this
chapter presents both experimental and analysis methods employed by other researchers toward
characterization of asphalt concrete fracture behavior at intermediate testing temperatures,
particularly those using the SCB test method.

2.1 Fracture Mechanics

Fracture mechanics is a useful tool designed to characterize crack initiation and propagation in
materials. Fracture in notched materials occurs when the energy stored at the vicinity of a crack is
equal to the energy required for the formation of new surfaces. It is noteworthy that this hypothesis
requires a pre-existing crack/notch to be valid. Thus, most fracture test specimens include a pre-
crack or notch. When the material at the vicinity of the crack (i.e., fracture process zone) relaxes,
the strain energy is consumed as surface energy and the crack grows by an infinitesimal amount.
If the rate of strain energy release is equal to the fracture toughness, then the crack growth takes
place under steady-state conditions and the failure eventually occurs.

2.2 Fracture Characterization of Asphalt Concrete Mixtures

In an effort to characterize the fatigue-fracture in AC mixtures and, concurrently, to improve the
mechanical and structural performance of AC pavement, various fracture testing methods such as
the single-edge notched beam (SEB) test ( Figure 2-1(a)), the disk-shaped compact tension (DCT)
test (Figure 2-1(b)), and semicircular bending (SCB) test (Figure 2-1(c)), have been attempted. It
is noteworthy that tests herein explore mode I fracture in which the loading direction and the initial
notch are directly aligned with the specimen’s centerline. This set-up is to solely induce tensile
stresses at the bottom of the specimen resulting in crack propagation.
The SEB test involves three-point bending of a notched AC beam. SEB is advantageous to
investigate pure mode I simple loading configuration and mixed mode testing by slightly moving
the notch away from the centerline. This geometry, although it is attractive numerically, as
demonstrated in several studies (Paulino, Song et al. 2004, Song, Paulino et al. 2005), is impaired

6
by a complex specimen fabrication that requires significant testing efforts. In addition, this test is
also not efficient for field cores that are usually circular disks while deep-notched laboratory
specimens may result in crack initiation under self-weight (Wagoner, Buttlar et al. 2005).

(a) (b) (c)

Figure 2-1 Fracture tests for asphalt concrete mixtures: (a) single-edge notched beam (SEB) test,
(b) disk-shaped compact tension (DCT) test, and (c) semicircular bending (SCB) test.

Another test sought by researchers is the disk-shaped compact tension (DCT) test, shown
in Figure 2-1(b). The DCT test has been standardized in the ASTM E399, “Standard Test Method
for Plane–Strain Fracture Toughness of Metallic Materials.” The specimen has a circular geometry
with loading holes on each side of the notch. This geometry can maximize the fracture area and is
thereby able to reduce the geometry-associated variability of test results. However, as mentioned
by (Wagoner, Buttlar et al. 2005), there is a possibility of stress concentration at the loading holes
that can result in a premature specimen failure with an erroneous outcome. Moreover, specimen
fabrication and preparation for the DCT test are not simple due to the accessories required to
position the specimen in the testing mount to induce pure opening mode fracture. The DCT test is
further hampered by potential crack deviation from the center of the specimen during testing.
Because of the issues of the aforementioned fracture tests, a semicircular bending (SCB)
test (Figure 2-1(c)) has been attractive in the flexible (AC) pavement community. This test is used
by many researchers, such as (Wu, Mohammad et al. 2005, Li and Marasteanu 2009, Shu, Huang
et al. 2010, Liu 2011, Aragão and Kim 2012, Biligiri, Said et al. 2012, Zegeye, Le et al. 2012, Im,
Kim et al. 2013, Kim and Aragão 2013, Im, Ban et al. 2014, Saadeh, Hakimelahi et al. 2014) due
to its several advantages: (1) easiness and effectiveness in fabricating specimens, (2) suitability for

7
field cores, and (3) repeatability in testing results (Wu, Mohammad et al. 2005, Li and Marasteanu
2009, Aragão and Kim 2012, Im, Kim et al. 2013, Im, Ban et al. 2014).
Initially, the SCB test method was proposed by (Chong and Kuruppu 1984) because other
existing fracture tests based on linear-elastic fracture mechanics (LEFM) were expensive and
difficult to perform using rock materials. As seen in Figure 2-1(c), the SCB test method is
fundamentally a three point bending test of a semicircular shaped specimen with an introduced
notch. This geometry induces tension at the bottom of the sample resulting in the crack propagation
throughout the specimen. The SCB test has proven to be adequate for evaluating the fracture
properties of both laboratory-compacted samples and field cores due to simplified specimen
preparation (Huang, Shu et al. 2013). Although a specimen for the SCB test has a lower potential
fracture area compared to one for the DCT test, the semicircular geometry enables the testing of
twice as many specimens obtained from field cores or laboratory-compacted samples compared to
the DCT. In addition, the SCB has shown great potential for characterizing the mixed-mode
fracture behavior of asphalt mixtures by simply adjusting the inclination angle of the notch or the
space between two supports (Im, Ban et al. 2014(a), Im, Ban et al. 2014(b)).
In both asphalt pavement research and the pavement community, fracture energy (Gf) has
been used as a simple parameter representing fracture for AC mixtures. Generally, this property is
less dependent on linear elasticity and homogeneity compared to other fracture properties, such as
critical strain energy release rate and stress intensity factor (Marasteanu, Li et al. 2004). Thus, this
method can be attractive for simply evaluating fracture characteristics of an asphalt mixture that is
highly heterogeneous and nonlinear inelastic. The fracture energy in Joule/m2 is calculated by
(Marasteanu, Li et al. 2004):
Wo  mg o
Gf  (2.1)
Alig

where 𝑊o is fracture work, the area below the load-displacement curve, as shown in Figure 2-2(a).
m is a mass, g is the gravitational acceleration, and 𝛿o is deformation. 𝐴𝑙𝑖𝑔 in Figure 2-2(b), is the
ligament area and can be calculated by:
Alig  t (r  c) (2.2)

where t is a thickness, r is a radius, and c is a notch length of the SCB specimen. It can be noted
that the mass (m) of the specimen is negligible in Eq.(2.1) because small specimens are typically
used, which infers an ignorable effect of specimen mass on the total fracture energy.

8
 1.4

1.2

1.0
Force (kN)

0.8
Wo
Alig
0.6

0.4

0.2

0.0
0 2 4 6
Load-Point Displacement (mm)

(a) (b)
Figure 2-2 Fracture energy (Gf) calculation: (a) fracture work (𝑊o) and (b) ligament area (Alig)

Aragão and Kim (2012) conducted a numerical and experimental effort to characterize
mode I fracture behavior of bituminous paving mixtures subjected to a wide range of loading rates
at intermediate temperature conditions. In this study, a simple experimental protocol was
developed using the SCB test geometry, and high-speed cameras with a digital image correlation
(DIC) were incorporated to monitor local fracture behavior at the initial notch tip of the SCB
specimens. The DIC results of the SCB fracture tests were then simulated using the finite element
method, which was incorporated with the material viscoelasticity and cohesive zone fracture
model. As shown in both Figure 2-3 and Figure 2-4, experimental results were successfully
simulated using the numerical model to study the effect of loading rate on fracture parameters. The
results shows a clear dependency of fracture parameters (i.e., fracture energy and cohesive
strength) to loading rates above 5mm/min.

9
 Figure 2-3 Effect of loading rate (Aragão and Kim 2012).

Figure 2-4 Numerical simulation results of SCB fracture testing at different loading rates by
Aragão and Kim (2012).

By considering a mode I type of fracture, in which the crack lies in a plane normal to the
direction of largest tensile loading, the stress state around the crack tip is characterized using the
stress intensity factor K I (in N/mm3/2 ) proportional to load P , and function of crack size (i.e.,

notch length) c and the geometry of the specimen. The stress value at any point near the crack
tip is given by:
KI
 YI (0.8) (2.3)
o c
where,  o is the stress acting at a small distance (i.e., half span length s) and expressed as P/2rt
with P being the load in MN. The geometric dimension of the specimen, r, t, and c are the radius,

10
thickness, and notch length, respectively. The mode I normalized stress intensity factor YI (0.8) is

independent of size and load, but depends on the geometry of the specimen and the loading
configuration. The span length used in this study (i.e., 120mm) and the 150mm diameter of the
specimen result in a span ratio of 0.8 or,
YI (120/150)  YI (0.8) (2.4)

and is expressed as calculated in Lim, Johnston et al. (1993) by:

 c
c  7.045 
YI (0.8)  4.782  1.219    0.063  e r
(2.5)
r
Lim, Johnston et al. (1993) approximated mode I stress intensity factors for various geometries of
experimental interest using a finite element method based on the LEFM principles on rock
materials.
The critical value of K I for which failure occurs, referred to as fracture toughness K IC ,
describes the local stress state that leads to the propagation of a crack. This stress typically occurs
at the highest load during testing, Pmax . Therefore, K IC represents the highest value of stress
intensity factor that the material can bear without fracture (Zegeye, Le et al. 2012).

2.00 60% of Pmax

Peak Load (Pmax)

1.50 12963_SPR_T_L
Force (kN)

1.00
Post-peak slope
Pre-peak slope

0.50
40% of Pmax
0.00
0 1 2 3 4 5 6 7
Load Point Displacement (mm)

Figure 2-5 A typical force-LPD curve from SCB test

11
 Using both the K I and the fracture energy method, Zegeye, Le et al. (2012) investigated
the size effect fracture in asphalt mixtures at a low temperature using the SCB test. In this work,
SCB specimens were prepared with four different diameters: 76.4 mm, 101 mm, 147 mm, and 296
mm. For every diameter size, specimens were notched to match a notch to radius (c/r) ratio of zero
(notchless), 0.05, and 0.2. In this study, the testing temperature was -24C. It was observed that
large notchless specimens (i.e., 296mm in diameter and c/r = 0) crack always initiated far from the
centerline where the measuring gauge was installed. It was also observed that the nominal strength
of specimens decreased as the size of specimens increased. It was noted from this study that large
specimen (larger than 150 mm in diameter) preparation is arduous and less practical since most
AC mixture compactors and field cores are 150 mm in diameter. This can explain the scarcity of
studies that used specimens with large diameters in AC sample preparation.
Biligiri, Said et al. (2012) evaluated the crack propagation potential of AC mixtures with
4.4% and 5.4% asphalt (binder) contents using the SCB. The test was conducted at 10C (the
standard testing temperature for fatigue evaluation in Sweden) on 50 mm thick specimens with 15
mm long and 2 mm wide notches. The test was conducted at a 1 mm/min LPD loading rate and at
three temperatures: 10C, 0C, and -10C, with four replicates tested for each case. It was found
that increasing the asphalt content, from 4.4% to 5.4%, reduced the mixture’s fracture toughness,
thereby decreasing its ability to resist higher traffic loads. Laboratory test results showed that
mixtures with higher asphalt content significantly improved crack propagation resistance in terms
of facture energy, while a higher resistance to fatigue cracking and propagation was observed from
field cores.
Using a different fracture indicator, the critical strain energy release rate, Wu, Mohammad
et al. (2005) evaluated fracture resistance in several Superpave AC. Mixtures possessed different
binder contents and nominal maximum aggregate sizes (NMAS) and were tested using the SCB
method. Specimens were prepared using a 3 mm wide saw at three different notch lengths: 25.4
mm, 31.8 mm, and 38 mm, with three replicated for each case. Specimens were monotonically
loaded at load point displacement (LPD) rate of 0.5 mm/min at a temperature of 251C. Statistical
analysis of the test results illustrated that the peak load might be sensitive to the binder type,
compaction level, or the NMAS at the notch length of 25.4 mm, and only sensitive to NMAS at a
notch length of 31.8 mm. At notch length of 38.0 mm, the peak load was not sensitive to any of
the variables. From this study, it was also found that the strain energy was only sensitive to the

12
NMAS and only at the notch depth of 31.8mm. This study found that the SCB test was fairly
sensitive to all mixture variables selected and that Superpave mixtures with larger NMAS
exhibited better fracture resistance due to larger stone-to-stone contact. It was concluded that the
SCB test method can be a valuable tool in the evaluation of the fracture resistance of AC mixtures.
From the study above, the critical strain energy release rate or critical value of J-integral
(Jc) has proven to be a useful fracture resistance indicator in bituminous materials. The Jc
quantifies the rate of change in strain energy (dU) with the change of notch length (da). As result,
Jc represents the consumed strain energy per unit area of fractured surface. Generally, Jc is
calculated from the Eq.(2.6) which becomes Eq.(2.7) in case two notch lengths were used as
shown in Figure 2-6.
 1  dU
Jc     (2.6)
 t  da

A A  1
Jc   1  2   (2.7)
 t1 t2  a2  a1
where A1, A2 = areas under the curves,

a1, a2 = initial notch lengths, and

t1, t2 = thicknesses of specimens.

2
SLX1-T40-N20-R10
1.5 SLX1-T40-N25-R10
Force (kN)

0.5

0
0 1 2 3 4 5 6
LPD (mm)
Figure 2-6 Calculating critical J-Integral using two notches

13
 Recently, Al-Qadi, Ozer et al. (2015) developed the modified SCB test for Illinois (IL-
SCB) and the apparatus of the test is shown in Figure 2-7. During the development of this test, the
low temperature was deemed because it was not discriminatory enough between mixtures and the
intermediate testing temperature of 25oC was adopted. Using a specimens of 50 mm thick with a
15 mm notch length, three cross head (LPD) loading rates were investigated (i.e., 6.25, 25 and 50
mm/min). In this study, all the loading rates investigated showed a good repeatability with average
values of COV of 13.3%, 8.6% and 9.2% for 6.25, 25 and 50 mm/min loading rates, respectively.
However, due to practical advantages and effectiveness in distinguishing mixtures, the loading rate
of 50 mm/min. was adopted for the IL-SCB.

Figure 2-7 SCB test specimen and configuration (in mm) adopted by Al-Qadi, Ozer et al. (2015)

In the study above, it was realized that the test results of fracture energy did not effectively
discriminate mixtures. Consequently, a new fracture indicator, Flexibility Index (F.I.) was
introduced and defined as:
 Fractureenergy 
F .I  0.01  . (2.8)
 post peak slope 
After field monitoring of mixtures, a good correlation between F.I. results of mixtures and their
corresponding field performance was found. It was observed that mixtures with F.I less than 2.0,
14
displayed a poor field performance while mixtures possessing F.I. values of 2.0 to 6.0 and greater
than 6.0 exhibited intermediate and good-performance, respectively. It is noteworthy that the post
peak slope in Eq.(2.8), is calculated at the inflection point of the post peak region as shown in
Figure 2-8 below.

Figure 2-8 Calculation of the post peak slope (Al-Qadi, Ozer et al. 2015)

The Eq. (2.8) can then be re-written as:

G 
F .I  0.01  f  (2.9)
 m 
where Gf is fracture energy and m is the slope at the inflection point shown in Figure 2-8.
According the study, FI can be associated with process zone size or other property combinations
with a good correlation to crack growth speed.
Table 2-1 summarizes the several SCB test methods with their corresponding testing
variables proposed by three different research groups: Illinois, Minnesota, and Louisiana. It is
worth mentioning that, most of testing variables given in the table below (Table 2-1) were not
extensively investigated prior their proposal, which mainly motivated this research.

15
 Table 2-1 Summary of the proposed SCB test methods and their testing variables
Test Variables IL-SCB (2015) Minnesota (2004) Louisiana (2005)
Number of
3 3 3
specimens
Thickness (mm) 50 25 50
Notch length (mm) 15 15 25.4, 31.8, and 38
50 mm/min
Loading rate 0.03 mm/min (CMOD) 0.5 mm/min (LPD)
(LPD)
10°C above the PG
Temperature 25oC 25oC
lower limit of asphalt
Fracture energy and Strain energy release
Fracture indicator Flexibility Index
toughness rate

16
 Chapter 3 Materials, Testing Facility, and Sample Fabrication

This chapter presents the materials and testing facility used to investigate effects of critical testing
variables on repeatability of the SCB test results. The aggregate gradation, aggregate consensus
properties, asphalt/binder content, and mixture design (i.e., air voids, binder content) of the AC
mixture used in this task, are presented. Finally, the testing facility is also introduced.

3.1 Aggregate and AC Mixture

In this study, a typical Nebraska AC mixture (i.e., SPH) was used to prepare SCB specimens for
laboratory tests. This mixture is typically used in Nebraska highways with a high traffic flow rate.
As shown in Figure 3-1, it was collected during construction and brought back to the laboratory in
sealed containers to prevent aging by oxidation. The mixture was then reheated for two hours in
an oven to reach its recommended compaction temperature of 149C (300F). Subsequently, the
mixture was compacted by a Superpave Gyratory Compactor (SGC) to target 4 ± 0.5% air voids.
The asphalt binder used in this study was Superpave performance graded PG 64-34 with a
warm mix asphalt (WMA) additive (Evotherm). By using a mixture from a single plant, a lengthy
AC mixture preparation process was avoided and thereby reduced the inherent variations
associated with the process. Proportionally, 5.20% of the asphalt cement (binder) content by the
total weight of mixture and the 0.7% of WMA additive (Evotherm) by the weight of binder were
mixed along with a blend of aggregates.
The aggregates were from four different sources; virgin aggregates, crushed gravel, 2A
gravel, and reclaimed asphalt pavement (RAP), which were proportionally 10%, 50%, 5%, and
35% of the total weight aggregates. The whole aggregate blend had a nominal maximum aggregate
size (NMAS) of 12.5 mm. It is noted that the presence of RAP in the mixture meant an addition of
only 3.38% virgin binder to the blend of aggregates to reach the total binder content of 5.20%.
This is due to the existence of asphalt cement in the old pavement millings (RAP).

17
 Figure 3-1 Mixture collection from a dump truck at asphalt concrete plant

Table 3-1 presents aggregates gradation from the four different sources with their
respective gradations. In addition, the bulk specific gravity (Gsb) and aggregate consensus
properties (i.e., fine aggregate angularity [FAA], coarse aggregate angularity [CAA], flat and
elongated [F&E] particles) of the final blend are also provided in the table.

18
 Table 3-1 Gradation and consensus properties of aggregates used

Sieve Analysis (Washed)

Materials %
3/4" 1/2" 3/8" #4 #8 #16 #30 #50 #200

3/4" Clean 10 100.0 60.0 18.0 2.0 2.0 1.0 1.0 1.0 1.0

Crushed Gravel 50 100.0 100.0 100.0 92.7 73.0 45.2 29.1 16.2 6.3

2A Gravel 5 100.0 95.4 90.9 68.0 27.3 8.6 3.5 1.1 0.2

Millings (RAP) 35 100.0 94.2 93.2 85.1 52.3 38.4 25.1 19.8 7.8

Combined Gradation 100.0 93.7 89.0 79.7 56.4 36.6 23.6 15.2 6.0

 90   28    2
Specification Range
100 100 <90  58    10

Consensus Properties

FAA CAA SE F&E D/B Design Gsb

45 99/96 79 0.1 1.18 2.585

FAA: Fine aggregates angularity; CAA: Coarse aggregates angularity; SE: Sand equivalent;
F&E: Flat and elongated particles; D/B: Dust to Binder Ratio; Gsb: Bulk specific gravity;
 : Not Specified.

3.2 Testing Facility

During this study, all mechanical tests were conducted using the 25kN capacity Universal Testing
Machine (UTM-25kN) equipment shown in Figure 3-2. This equipment is composed of an
environmental chamber, a central data acquisition system (CDAS), and a hydraulic pressure
system. It can produce a maximum of 25kN of static and 20kN of dynamic loading (at various
frequencies). Additional information (i.e., key features and specifications) of the UTM-25kN test
station is presented in Table 3-2. The environmental chamber can precisely control temperatures
ranging from -16C to 60C. However, to ensure accurate temperature reading, a dummy AC
sample with an internal thermometer was placed inside the chamber along with the SCB specimens
to guarantee a target testing temperature, as shown in Figure 3-3.

19
Figure 3-2 UTM-25kN testing equipment

20
 Table 3-2 UTM-25kN key specifications

Load Frame

Size 185(H)  58(D)  60(W) cm

Weight 130kg

Load Capacity 25kN static, 20kN dynamic

Between Columns 45cm

Vertical Space 80cm

Stroke 50mm

Hydraulic Power Supply

Size 81(H)  40(D)  70(W) cm

Weight 75kg (excluding oil)

Flow Rate 5litres/min

High Pressure 160 Bar

Low Pressure 2 to 160 Bar (adjustable)

Mains Power 208V / 230 V, 50 or 60 Hz, 2.6 kW

Noise Level less than 70db at 2m

21
 Figure 3-3 Reading temperature of specimens

3.3 Sample Fabrication and Test Set-up

As aforementioned, the AC loose (uncompacted) mixture (i.e., SPH) was re-heated in an oven for
two hours at 300°F (149°C). After ensuring this exact temperature with an infrared non-contact
thermometer, the mixture was then compacted using the Superpave gyratory compactor (SGC) to
produce tall compacted samples of 150 mm in diameter and 170 mm of height (Figure 3-4(a)),
with a target air voids of 4  0.5%. Multiple slices with various thicknesses, ranging from 30 mm
to 60 mm in this study, were then prepared after removing the top and bottom parts from the tall
compacted samples, as shown in Figure 3-4(b). Each slice was then cut into halves to yield two
SCB specimens with a desired notch length of two millimeters in width, as shown in Figure 3-4(c).
It is noted that the introduced notch serves as crack initiator as this test is solely aimed at
characterizing the fracture properties of AC mixtures during cracking propagation rather than by
cracking initiation (European Committee for Standardization 2010).

22
 (a) (b) (c)

Figure 3-4 SCB specimen fabrication process: (a) compacting, (b) slicing, and (c) notching

For a more accurate test results analysis, the exact thicknesses and ligament lengths of SCB
specimens were measured from three locations along the semicircular edge and the results were
averaged, as shown in Figure 3-5. Afterward, specimens were placed inside the environmental
chamber of the UTM-25kN and allowed a minimum of four hours to reach temperature equilibrium
prior to testing. Subsequently, specimens were placed on a three-point bending test fixture inside
the environmental chamber to perform the test. The fixture has two cylindrical supports of 25 mm
in diameter at each end, separated by a 120 mm span length. It is noteworthy that lubrication was
applied to these supports to mitigate friction during testing.

23
 Figure 3-5 Recording the thickness and ligament length of SCB specimen

The device in Figure 3-6(a) was used to place specimens on the bending fixture in order to
avoid eccentric loading. Then, a monotonic displacement rate was applied to the top center line of
the specimen. A data acquisition system simultaneously monitored both the reaction force and the
LPD during testing (Figure 3-6(b)). It is noted that each time, prior to testing, a small contact force
of 0.10 kN was applied to the specimen.

24
 (a)

Load

Load point

75mm

120mm

(b)
Figure 3-6 Test set-up for semicircular bending (SCB): (a) specimen alignment before testing
and (b) specimen ready to be tested

25
 Chapter 4 SCB Test Method Development

4.1 The Number of Testing Specimens

The number of specimens (sample size) of an experimental test is very critical in that; too large of
a sample size may waste time and resources, while too small of a sample size may lead to
inaccurate results. Therefore, determining the recommended minimum number of specimens for a
test method is a significant task for reliable outcomes with high repeatability and efficiency.
Consequently, the first effort the SCB test development was to statistically investigate the
relationship between the sample size and the variation of the results.
Typically, the required sample size can be calculated by Dowdy, Wearden et al. (2011)

y   z  
2

Z /2   n    /2 
/ n  E  (4.1)
where n is the number of specimens, Z is the standard normal deviate, 𝜎 is standard deviation of
population, and E is the margin of error expressed as:
E  y  (4.2)
where y is observed sample mean and  is the true value of the population mean. Since the true
population mean (  ) is often unknown, E is usually introduced to achieve a desired accuracy. In
this study, with a margin of error of 0.075 and a confidence level of 95% (i.e., Zα/2 = Z0.025 = 1.96),
were used. Eq.(4.1) can then be rewritten as:

1.96   
2

n   683   2 (4.3)

 0.075 
Similar for the true population mean, the standard deviation of the population is often
unknown. So in this study, the standard deviation (𝜎) of fracture energy of the population was
rationally estimated after testing 18 SCB specimens and examining the relationship between
population and standard deviation. This approach permitted a more accurate analysis to find n. The
specimens were tested using typical SCB testing variables from the literature reviews such as:
thickness of specimen = 50 mm (Duan, Hu et al. 2003), temperature = 21°C (Kim and Aragão
2013, Im, Ban et al. 2014), notch length = 15 mm (Li and Marasteanu 2009, European Committee
for Standardization 2010), loading rate = 1 mm/min. (Biligiri, Said et al. 2012, Im, Kim et al.
2013).

26
 Because the Eq.(4.1) was based on the assumption that the population (i.e., fracture energy
values for 18 SCB specimens in this study) came from a normal distribution, the normality of the
test results should be checked prior to further analysis. As shown in Figure 4-1, the Lilliefors test
(Razali and Wah 2011, Machiwal and Jha 2012) was conducted to compare the observed results
to the expected results (i.e., normal distribution) of the same mean and standard deviation. The
figure shows a good visual agreement between the expected fracture energy (i.e., the normal
distribution) and the measured fracture energy. For a more quantitative normality check, a chi-
square goodness-of-fit test was performed to check if the two distributions were statistically
different at 5% significance level. The chi-square test resulted in the observed chi-square value
(0.016) less than the critical value (27.587), which demonstrates that the fracture energies of the
18 SCB specimens originated from a normal distribution.

1.0
Fracture Energy (kJ/m2)

0.8

0.5

0.3 Values When Normaly Distributed

Actual Test Results

0.0
-3 -2 -1 0 1 2 3
Standard Normal Deviate (Z) Values

Figure 4-1 Normality test result.

As seen in Eq.(4.3), the number of specimens can then be determined by the standard
deviation (𝜎) of fracture energy from an assumed population size. To find the relationship between
the standard deviation of the population and the assumed population size ( k = 1, 2,…, 18 in this
case), the number of all possible combinations ( Ckp ) from the total count (p = 18) were calculated
by:

27
 p!
Ckp  where p =18 and k =1, 2, 3, ... , 18 (4.4)
k !( p  k )!
Subsequently, corresponding standard deviations for each assumed population size (k)
were obtained by averaging the standard deviations from the all possible combinations ( Ckp ).
Figure 4-2(a) shows an example for the assumed population size (k) of five. Each standard
deviation of fracture energy for 8,568 combinations was calculated and used to obtain the average
of standard deviation for k = 5.
Overall results of this process are presented in Figure 4-2(b). It is observed from the figure
that there was a strong dependency of standard deviation on the assumed population size (k) of
one to eight, followed by a steady saturation. Sequentially, Eq.(4.3) was used to calculate the
minimum number of specimens (n) for each population size (k) assumed, and results are presented
in Figure 4-2(c). The outcomes of the process show that the minimum number of specimens
increased with population size, and then tended to saturate at n = 5~6. Thus it can be inferred from
this graph that five to six SCB specimens would be a reasonable sample size that can sufficiently
represent the AC fracture behavior of the entire population (18 specimens in this case) with a 95%
level of confidence. It should be also noted that the statistical analysis performed herein is
reasonable because the minimum number of specimens (n) was always less than the corresponding
population size (k), as shown in in Figure 4-2(c) (i.e., n < k in all cases).

28
 Assumed Population Size of Five Specimens
Specimen Name Associated Fracture Energy (kJ/m2) Std. Dev.
1 2 3 4 5 0.442937 0.70566 0.886217 0.666351 0.754043 0.161587146
1 2 3 4 6 0.442937 0.70566 0.886217 0.666351 0.707196 0.158346526
1 2 3 4 7 0.442937 0.70566 0.886217 0.666351 0.574108 0.164066
⁞ ⁞ ⁞
3 4 5 8 18 0.886217 0.666351 0.754043 0.66089 0.597246 0.111780105
3 4 5 9 10 0.886217 0.666351 0.754043 0.678727 0.602719 0.108506854
3 4 5 9 11 0.886217 0.666351 0.754043 0.678727 0.680773 0.092240535

⁞ ⁞ ⁞

13 14 16 17 18 0.591544 0.714668 0.637955 0.679975 0.597246 0.053061009

13 15 16 17 18 0.591544 0.738708 0.637955 0.679975 0.597246 0.061458385
14 15 16 17 18 0.714668 0.738708 0.637955 0.679975 0.597246 0.057154921
Average of Standard Deviation 0.083539845
1 (a)
0.10 6
Standard Deviation of Fracture

Minimum Number of Specimens

0.08 5

4
0.06
Energy

(n)

3
0.04
2
0.02 1

0.00 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Assumed Population Size (k) Assumed Population Size (k)

(b) (c)

Figure 4-2 (a) Calculation of average standard deviation for k=5, (b) average standard deviation
for each assumed population size (k), and (c) assumed population size (k) with
associated minimum number of specimens (n).

4.2 Specimen Thickness

Although previous studies (Brühwiler, Wang et al. 1990, Duan, Hu et al. 2003) highlighted that
the thickness of specimens strongly affected fracture energy (Gf), less emphasis was placed on the
effect of specimen thickness on testing repeatability. In this study, the fracture energy and
variability of the test results for various thicknesses of specimens were investigated. As shown in
Figure 4-3, the thicknesses varied from 30 mm, 40 mm, and 50 mm, to 60 mm. Other testing
variables were reasonably selected based on literature reviews: temperature = 21°C (Kim and

29
Aragão 2013, Im, Ban et al. 2014), notch length = 15 mm (Li and Marasteanu 2009, European
Committee for Standardization 2010), loading rate = 1 mm/min. (Biligiri, Said et al. 2012, Im,
Kim et al. 2013), and the number of specimens = six (Romero and Masad 2001). Figure 4-4(a)
shows that the peak force (Pmax) increased as specimens became thicker. Additionally, the fracture
energy increased from 30 mm to 50 mm, followed by a slight decrease at a thickness of 60 mm, as
presented in Figure 4-4(b). However, the fracture energy did not seem to be significantly
dependent on the thickness of specimens within the thickness range tested. It is noteworthy that
the results, in all cases, are an average of the six replicates.

60 mm 50 mm 40 mm 30 mm

Figure 4-3 Fabrication of SCB specimens at different thicknesses

To evaluate the consistency of the testing results, the coefficient of variation (COV) of the
fracture energy of each thickness was estimated. Figure 4-4(b) indicates a general decrease in COV
with increasing thickness, while a steep decline between 30 mm and 40 mm was observed. This
figure implies that a SCB specimen thicker than 40-50 mm is appropriate for characterizing the
fracture behavior of AC without significantly increasing the variability of results when other
variables are maintained. This finding agrees well with previous studies (Brühwiler, Wang et al.
1990, Wittmann and Zhong 1996), indicating that the thickness of AC specimens should be at least
four times larger (i.e., 12.5 mm 4 = 50 mm) than NMAS size (12.5 mm in this study). For the
subsequent steps, 50 mm was chosen based on other studies and the low COV (≤ 10%) value found
in this study.

30
 3.5

3.0 60 mm

2.5 50 mm
Force (kN) 2.0 40 mm

1.5 30 mm

1.0

0.5

0.0
0 1 2 3 4
Load Point Displacement (mm)
(a)

2.5 30

Coefficient of Variation(%)
Fracture Energy
Fracture Energy (kJ/m2)

2.0 25
Coefficient of Variation
20
1.5
15
1.0
10
0.5 5

0.0 0
30 40 50 60
Thickness (mm)
(b)
Figure 4-4 Effect of thickness of specimens (t): (a) test results (average of six replicates) and (b)
fracture energy with standard error bars and COV of fracture energy for different
thicknesses

31
4.3 Notch Length
To investigate the effect of notch length, specimens with five notch lengths (0 mm, 5 mm, 15 mm,
25 mm, and 40 mm) were tested. Other testing variables were fixed: the thickness of a specimen
= 50 mm, loading rate = 1 mm/min, temperature = 21°C, and the number of specimens = six.
Figure 4-5(a) shows that the peak force and initial stiffness increased as notch length decreased.
This trend is reasonable because specimens with smaller notch lengths have greater areas to be
fractured, requiring more energy to fracture them. Another interesting observation from the figure
is that the displacement at the peak force increases (i.e., shifts to the right) with decreasing notch
length. Figure 4-5(b) shows a decreasing trend of fracture energy along with increasing notch
length. The fracture energy drops from around 2 kJ/m2 in the case of the notchless specimens (i.e.,
0 mm notch length) to around 0.5 kJ/m2 for the specimens with 40 mm notch length.
The figure also presents the COV of fracture energy at various notch lengths. Due to the
more random crack initiation/propagation, notchless specimens showed a higher COV than other
specimens with a notch. Additionally, the high COV of the specimens with the longest notch length
(i.e., 40 mm) might be explained by an insufficient ligament area (35 mm by 50 mm), which seems
smaller than the typical size of a representative volume element (RVE). It is noted that a RVE is
the smallest size of a specimen that should be tested in order to avoid a certain localized
phenomenon and to provide a representative global response. Determination of the RVE size of a
specimen is beyond the scope of this study. However, it can be noted from several previous studies
that the RVE size of typical AC mixtures with a NMAS of 12.5 mm is around 60 mm by 60 mm
(Romero and Masad 2001, Kim, Lutif et al. 2009, Kim, Lee et al. 2010). Figure 4-5(b) shows that
a COV value of less than 10% can be achieved from specimens with notch lengths between 5 mm
and 25 mm.

32
 3.5
0 mm
3.0 5 mm
2.5 15 mm

Force (kN)
25 mm
2.0
40 mm
1.5
1.0
0.5
0.0
0 1 2 3 4
Load Point Displacement (mm)

(a)

2.5 30
Fracture Energy

Coefficient of Variation (%)

Fracture Energy (kJ/m2)

2.0 Coefficient of Variation 25

20
1.5
15
1.0
10
0.5 5

0.0 0
0 5 15 25 40
Notch Length (mm)

(b)
Figure 4-5 Effect of notch length: (a) test results (average of six replicates) and (b) fracture
energy with standard error bars and COV of fracture energy for different notch lengths

33
 0 mm 5 mm
mmmmm mm

15 mm 25 mm

Figure 4-6 Fracture profile at different notch lengths

Although a 5 mm notch specimens presented the lowest COV within the range on notch
lengths investigated in here, the resulting crack propagation deviated highly from the centerline of
the specimen to be considered mode I fracture (see Figure 4-6). Consequently, a 15 mm notch was
chosen to be used in the next step due to the relatively better cracking propagation profile and the
repeatability of the test results.

4.4 Loading Rate

The loading rate has strong effects on the fracture behavior of AC mixtures under intermediate
temperature conditions because of the viscoelastic deformation characteristics of asphaltic
materials, as demonstrated by many studies including (Kim and Aragão 2013, Im, Ban et al. 2014).
In this study, SCB specimens were tested at five different loading rates: 0.1, 0.5, 1.0, 5.0, and 10
mm/min (Figure 4-7). Other testing variables remained constant (i.e., thickness of a specimen =
50 mm, notch length = 15 mm, temperature = 21°C, and the number of testing specimens = 6).

34
 Figure 4-7 Loading rates inputs

The experimental results in Figure 4-8(a) indicated that AC mixtures at slower loading
rates showed more compliant responses, while the mixtures exhibited stiffer responses with greater
peak force at faster loading rates. This observation agrees with findings from previous studies (Im,
Kim et al. 2013, Kim and Aragão 2013, Im, Ban et al. 2014).
Although loading rates greatly influence AC force-displacement behavior, as shown in
Figure 4-8(b), the fracture energy was not significantly affected by the loading rate within the
range tested in this study. It is noted that the fracture energy between one to five mm/min. stayed
constant. In addition, compared to other testing variables, such as the thicknesses of specimens
and notch lengths (see Figure 4-4(b) and Figure 4-5(b)), low COV values were observed in all
cases tested with a range between 0.1 mm/min. and 10 mm/min. Although the loading rates from
0.1 mm/min. to 0.5 mm/min. showed the lowest COV values, it is important to mention that testing
at this rate is lengthy (3,000 sec. and 600 sec. for 0.1 and 0.5 mm/min, respectively) while
providing no significant improvement to repeatability of test results compared to other loading
rates evaluated herein. If one selects a threshold COV of 15%, any loading rate within the range
can be chosen for the SCB test. Thus for the next step (i.e., investigation of testing temperature),
a loading rate of 5 mm/min. was selected based on practicality.

35
 3.5

3 10 mm/min
5 mm/min
2.5
Force (kN) 1 mm/min
2
0.5 mm/min
1.5 0.1 mm/min
1

0.5

0
0 1 2 3 4
Load Point Displacement (mm)
(a)

2.5 30
Coefficient of Variation

Coefficient of Variation (%)

Fracture Energy (kJ/m2)

2.0 Fracture Energy 25

20
1.5
15
1.0
10
0.5 5

0.0 0
0.1 0.5 1 5 10
Loading Rate (mm/min)

(b)
Figure 4-8 Effect of loading rate: (a) test results (average of six replicates) and (b) fracture energy
with standard error bars and COV of fracture energy for different loading rates

36
4.5 Testing Temperature
It is widely documented that AC is highly temperature-dependent due to the viscoelastic nature of
asphalt cement (Marasteanu, Li et al. 2004, Im, Kim et al. 2013). Based on this, the next step was
to characterize the temperature effect on the repeatability of the test results, particularly for
characterizing the fatigue-type cracking potential of mixtures. As shown in Figure 4-9, three
different temperatures (15, 21, and 40°C) were attempted to investigate their effects on fracture
energy.

Front

Back

Figure 4-9 Fracture profiles at different testing temperatures (front and back)

Other testing variables used were: thickness of specimens = 50 mm, notch length = 15 mm,
loading rate = 5 mm/min., and the number of specimens = 6. The figure clearly shows that peak
force and fracture energy were inversely proportional to testing temperature, which contrasts with
the findings when the test was performed at low temperatures (e.g., below 0°C) (Wagoner, Buttlar
et al. 2005, Li and Marasteanu 2009).

37
 3.5

3.0
15°C
2.5 21°C
Force (kN)
2.0 40°C

1.5

1.0

0.5

0.0
0 1 2 3 4
Load Point Displacement (mm)
(a)
2.5 30
Fracture Energy

Coefficient of Variation (%)

Fracture Energy (kJ/m2)

Coefficient of Variation 25
2.0
20
1.5
15
1.0
10
0.5 5

0.0 0
15 21 40
Temperature (C)

(b)
Figure 4-10 Effect of temperature: (a) test results (average of six replicates) and (b) fracture energy
with standard error bars and COV for different testing temperatures

Figure 4-10(b) presents the COV of fracture energy at different temperatures. As shown,
lower temperatures presented smaller testing variations, with specimens at 15°C showing the
lowest COV value of less than 10% in this study. Nonetheless, it can be noted that SCB testing at

38
21°C temperature could be quite attractive, with only a little loss of testing repeatability, when one
considers the practical applications of the SCB test method for engineering purposes. This is
because 21°C is a room temperature that is easily achievable without a sophisticated environmental
chamber for testing equipment, and is a reasonable temperature that can properly represent fatigue-
type cracking events.

4.6 Summary of SCB Test Method Development

After determining the minimum recommended number of specimens and investigating the effect
of each critical variables on repeatability of the test results, the range of test variables can be
recommended when conducting SCB test (see Table 4-1). It is noteworthy that the COV of test
results in terms of fracture energy from variables selected within this range should be less or
approximately equal to 15%.

Table 4-1 Recommended variables for SCB test with approximate associated COV.

Test Variable Recommended Values

Thickness (mm) 40~50

Notch Length (mm) 5~15

Loading Rate (mm/min.) 1~5

Temperature (C) 15~40

No. of Specimens 5~6

COV (%) ≤ 15

39
 Chapter 5 SCB Testing of Nebraska Plant Produced Mixtures

In this chapter, the sensitivity of the developed SCB testing method was investigated through a
field program. Fourteen AC mixtures collected from 12 different field construction projects (see
Table 5-1) were tested. Mixture performance was ranked according to the fracture energy resulting
from the SCB fracture test method. Statistical analyses were then conducted to investigate the
sensitivity of the SCB test method to changes in AC mixtures.

5.1 Project Selection

Field projects were selected considering type of mixture, location of project, and availability of
other important information such as pavement structure configuration, traffic and weather
information, rehabilitation history of the pavement, etc. It is noteworthy that the accessibility to
mixtures collected in a timely manner was also a decisive factor in selecting field projects. Since
the main objective of this task is to test the sensitivity of the SCB test from different types of
Nebraska AC mixtures and ultimately correlate the SCB test results with field cracking
performance, it was important to collect other relevant information that is critical to conducting
validation with actual field cracking performance. In this way, the laboratory SCB test results can
be properly correlated to the field performance of the same mixtures over their service life. The
field performance would be recorded through a routine Nebraska Pavement Management System
(PMS).
Based on the selection criteria above, a total of 14 AC mixtures were collected as
representatives of all different types of AC used in Nebraska. These mixture were collected from
12 separate field construction projects, as shown in Table 5-1. In this table, detailed information
about the construction projects, control number, highway name, thickness of the AC layer
constructed, and specific project location are given.

40
 Table 5-1 Field project selected for this study
Control
Mixture
Number Highway Fill (inch) Location
Type
(CN)
42515 80 2 Henderson to Waco
SPH 22586 80 2 50th St. - I-480, Omaha
42514 80 2 Hampton to Henderson
SRM 42567 281 4 St. Paul North
M4TLOB 4 1 Lawrence East
SLX M1041 41 1 N-41, US-77-Adams & US-77 Truck Scales
M4TLOA 56 1 HWY 91 - Spaulding East & West & HWY 56
12963 63 4 US-34-Alvo
22454 91 4 Blair West
SPR 12980 92 4 Brainard East & West
42399 30 4 Wood river - Grand Island
42567 281 3.5 Saint Paul north
42399 30 2 Wood River to Grand Island
SPS
42514 80 2 Hampton to Henderson
HWY: Highway

5.2 Material Collection and Sample Fabrication

All materials were collected from the mixture production plants prior to paving and were
transported in a sealed container to minimize mixture aging. Figure 5-1(a) exemplifies a field
construction in progress at highway 63 where the SPR_12963 (SPR with control number of 12963)
mixture was collected. Figure 5-1(b) presents mixtures after the compaction process. All mixtures
were heated up to their respective specified compaction temperatures, as shown in Table 5-2, and
then compacted to a target air voids of 4  0.5% by the Superpave gyratory compactor (SGC).

41
 (a) (b)

Figure 5-1 Field program: (a) construction in progress on highway 63 (CN: 12963) and (b) after
laboratory compaction of mixtures

Table 5-2 Compaction temperature for each mixture

Mixture Type Control Number (CN) Comp. Temp. oC (oF)

42515 149 (300)

SPH 22586 149 (300)
42514 143 (290)
SRM 42567 143 (290)
M4TLOB 141 (285)
SLX M1041 141 (285)
M4TLOA 141 (285)
12963 138 (280)
22454 143 (290)
SPR 12980 143 (290)
42399 143 (290)
42567 143 (290)
42399 138 (280)
SPS
42514 138 (280)

42
5.2.1 Aggregates Gradation
Aggregates gradation of five representative AC mixtures (one per each type of mixture) are shown
in Figure 5-2. As seen in the figure, with the exception of SLX and SPS, all mixtures have a portion
of coarse aggregates that can be associated to their functional aspect. For instance, SLX is usually
used for thin-lift overlay pavement preservation AC layer that is typically one-inch in thickness,
while the SRM is a base mixture typically used to replace a hydrated lime slurry stabilized base
and/or cold foam reclamation layer. Thus, SRM contains a higher percentage of coarse aggregates
compared to SLX and/or SPH. This observation is also apparent in the visual microstructures of
SCB specimens, as shown in Figure 5-3.

100.0

80.0
Percent Passing

60.0

40.0 M4TLOB_SLX(WMA)
42514_SPH(WMA)
42515_SPH(WMA)
20.0 12963_SPR(WMA)
42514_SPS
42567_SRM(WMA)
0.0
1.E-06 1.E-03 1.E+00 1.E+03
Sieve Size (raised to power 4.5)
Figure 5-2 Gradation chart of five representative mixtures – sieve sizes raised to 0.45 power

43
 42514_SPH M4TLOB_SLX

12963_SPR 42514_SPS

42567_SRM
Figure 5-3 Visual comparison between the mixtures

5.2.2 Mixture Characteristics

Table 5-3 summarizes the mixture design characteristics of each mixture, such as asphalt cement
type and content, the amount of reclaimed asphalt pavement (RAP) added in each mixture, and the
amount of warm-mix asphalt (WMA) additive, Evotherm. The usage of RAP in pavement
construction is very attractive since it is an economical and environmentally-friendly alternative.
As can be seen in the table, the percentage of RAP ranges from 30% to 50% depending on the type
of mixture.

44
 Table 5-3 Blending characteristics of mixtures selected
Mixture Type Control Binder Binder Content WMA
Source
Number Type RAP (%) Virgin From RAP Total Content Type
42515 PG 64-34 35 3.38 1.82 5.2 0.70% Evotherm FLINTHILLS
SPH 22586 PG 64-34 35 3.38 1.82 5.2 0.70% Evotherm FLINTHILLS
42514 PG 64-34 35 3.38 1.82 5.2 0.70% Evotherm FLINTHILLS
SRM 42567 PG 58-34 50 2.5 2.7 5.2 0.90% Evotherm FLINTHILLS
M4TLOB PG 64-34 30 3.69 1.71 5.4 0.70% Evotherm FLINTHILLS
SLX M1041 PG 64-34 30 3.95 1.55 5.5 0.70% AD-here FLINTHILLS
M4TLOA PG 64-34 30 4.02 1.38 5.4 0.70% Evotherm FLINTHILLS
12963 PG 64-34 35 3.38 1.82 5.2 0.70% Evotherm FLINTHILLS
22454 PG 64-34 45 2.74 2.56 5.3 0.70% AD-here FLINTHILLS
SPR 12980 PG 64-34 45 2.78 2.52 5.3 1.25% HydroLime MONARCH
42399 PG 64-34 35 2.78 2.52 5.3 0.70% Evotherm MONARCH
42567 PG 64-34 35 2.9 2.5 5.4 0.70% Evotherm FLINTHILLS
42399 PG 52-34 45 2.49 2.71 5.2   MONARCH
SPS
42514 PG 52-34 45 2.86 2.34 5.2   FLINTHILLS
RAP: Reclaimed Asphalt Pavement, WMA: Warm-Mixed Asphalt, and : Not Applicable

45
5.3 Test Results and Discussion
Sample fabrication in this task followed the specimen fabrication process described earlier in
chapter 3. The testing variables were selected from the recommended range shown in Table 4-1.
More specifically, specimens were 50 mm thick with a notch length of 15 mm, tested at a LPD
loading rate of 3 mm/min. and temperature of 21C; in each case, results from a total of six
replicates were averaged. From the test results, fracture energy was then calculated and statistical
analysis conducted to assess the sensitivity of SCB test to difference in AC mixtures. The resulting
values of fracture energy and the corresponding standard error bars are presented below
(Figure 5-4).

1.6

1.4
Fracture Energy (kJ/m2)

1.2

0.8

0.6

0.4

0.2

0
42567_SRM

42514_SPS
42515_SPH

22586_SPH

42514_SPH

M1041_SLX

M4TLOA_SLX

12963_SPR

22454_SPR

12980_SPR

42399_SPR

42567_SPR

42399_SPS
M4TLOB_SLX

Figure 5-4 SCB test results (fracture energy) of different mixtures with standard error bars

46
 In order to further assess the sensitivity of the SCB test method, statistical analyses were
performed on the test results. For the repeatability check, the coefficient of variation (COV) from
the six replicates of each mixture was estimated, and the resulting values are presented in
Table 5-4. As can be seen, the COV values of mixtures were less than 15% threshold with
exception of one mixture, 42399_SPS which showed COV value of 16%. This COV values less
than 15% are explained by the fact that all testing variables in this task were chosen from the
recommended values after repeatability test summarized Table 4-1. It is noteworthy that since the
recommended test variables was based on statistical analysis of 95% confidence level, there is a
chance (5%) that the COV may not lay in the 15% COV interval. This explains why one test with
COV higher than 15%. Overall this observation further confirms the repeatability of the SCB test.

Table 5-4 Coefficient of variation of test results

Coefficient of Variation (COV)
Mixture Name
of Fracture Energy
42515_SPH 8.1%
22586_SPH 6.6%
42514_SPH 6.6%
42567_SRM 6.9%
M4TLOB_SLX 5.3%
M1041_SLX 12.4%
M4TLOA_SLX 10.6%
12963_SPR 8.1%
22454_SPR 11.5%
12980_SPR 11.9%
42399_SPR 3.0%
42567_SPR 12.3%
42399_SPS 16.0%
42514_SPS 7.4%
Note: COV higher than 15% is marked in bold.

47
 To further investigate the sensitivity of SCB test, an analysis of variance (ANOVA) and
Tukey’s honestly significant difference (HSD) multiple-comparison statistical test were
conducted. In this study, the ANOVA was used to test the null hypothesis, indicating that mean
values from different mixtures are equal (i.e., the alternative hypothesis indicating that at least one
mean value is statistically different from other means) at a 95% confidence level. If the null
hypothesis was rejected, then a post-hoc multiple-comparison analysis, namely, Tukey’s HSD was
conducted. Figure 5-5 presents several available multiple-comparison tests and their
corresponding powers with Type I error rates. Statistical power represents the probability of
correctly detecting a difference (i.e., rejecting the null hypothesis when it is false), while Type I
error is a probability of rejecting the null hypothesis when it is true (i.e., detecting false
differences). The Tukey’s HSD test was used herein to detect differences in mixtures due to a
lower probability of a Type I error, thus less likely to detect false differences in mixtures (see
Figure 5-5). The Tukey’s HSD test has been used in other several studies (Zegeye, Le et al. 2012,
Faruk, Hu et al. 2014) to compare AC mixtures.

Figure 5-5 Multiple-comparison procedures (Dowdy, Wearden et al. 2011)

Since the Tukey’s HSD is a post-hoc test, it requires the rejection of the null hypothesis (p-values
< 0.05) to be effective and answers the question of which mean is significantly different from
another. Table 5-5 shows the ANOVA table of fracture energy resulting from the 14 mixtures. As
shown, by comparing the p-value with a given -level (0.05), the null hypothesis was rejected,
since the p-value is less than the -level. This implies that at least one mixture is significantly
different from other mixtures in terms of their fracture energy values at the 95% confidence level
(i.e., -level of 0.05). Consequently, the Tukey’s HSD was conducted.

48
 Table 5-5 ANOVA: single factor about fracture energy
Source of Variation SS df MS F P-value F crit
Between Groups 1.660773 13 0.12775184 10.053996 1.7551E-11 1.862661
Within Groups 0.88946 70 0.01270657

Total 2.550234 83

In Tukey’s test, the number of replicates of each mixture (six in this case) and the desired
probability level (i.e., 5% in this study) were first used to find the studentized q-value. Then, the
q-value with the variance of test results was used to find the absolute critical difference between
means. If the difference between two means of mixtures is larger than the absolute critical value,
the two are significantly different at that given probability level. More detailed information on
Tukey’s HSD test can be found elsewhere (Dowdy, Wearden et al. 2011).
Table 5-6 presents Tukey’s HSD test results of individual mixtures with mean fracture
energy values and groups (in letters). For example, two SLX mixtures (i.e., M4TLOB and M1041),
a SPR mixture (12963), and a SPH mixture (22586) are ranked with the same group A due to the
statistical similarity in their fracture energies with the specified confidence level (α = 0.05). As
seen in the table, several mixtures are classified in group C, implying that their fracture energies
are statistically similar, as previously observed in Figure 5-4.

49
 Table 5-6 Tukey's HSD about fracture energy and mixture ranking

Note: Means that do not share a letter are significantly different

The above table classifies AC mixtures by ranking them in descending order (from top to
bottom) in terms of their fracture energy. In an attempt to investigate rank orders among individual
AC mixture types, the Tukey’s HSD test results were rewritten as shown in Table 5-7. It can be
observed from the table that, among the three SLX mixtures evaluated in this study, the SLX_
M4TLOA was classified in group C with the lowest fracture energy (1.18kJ/m2), while the other
two SLX mixtures were both classified in group A with relatively high fracture energy (1.42 ~
1.43 kJ/m2). Amongst the three SPH mixtures, the SPH_22586 performed better than the other

50
two, although it was not significantly different them implying the similarities in this mixture type.
Most of SPR mixtures expect the SPR-12963 were categorized in group C or below due to their
relatively low fracture energy values. As also expected, SRM and SPS were generally categorized
in lower-graded groups: C to E.

Table 5-7 Mixture classification by their fracture energy

Mixture Fracture Energy
SLX_M1041 A
SLX_M4TLOA C
SLX_M4TLOB A
SPH_22586 A, B
SPH_42514 C
SPH_42515 B, C
SPR_12963 A
SPR_12980 C
SPR_22454 C
SPR_42399 C
SPR_42567 D, E
SPS_42399 E
SPS_42514 C
SRM_42567 C, D

51
5.3.1 Relationship between virgin asphalt content to fracture energy
A further analysis was attempted to find the relationship between the fracture energy and mixture
constituent characteristics. It was observed that the fracture energy of mixtures increased with
more amount of virgin-state asphalt cement as shown in Figure 5-6. Virgin asphalt cement is the
portion of the total asphalt content that is not from RAP materials. It was added to the aggregate
blend containing RAP to reach a required binder content by the total weight of mixture.

4
Percent of Virgin Binder

y = 2.4784x + 0.2012
R² = 0.5082
2

1
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
2
Fracture Energy (kJ/m )

Figure 5-6 Relationship between virgin binder and fracture energy

It can be inferred from the observation that fracture resistance of mixtures containing RAP
materials can be improved by having the increasing percentage of virgin asphalt content. This
observation seems reasonable since the virgin binder contains new soft molecules that act as a
fresh adhesive in the mixtures. Theses soft/lighter molecules are usually lost due to oxidative aging
process of asphalt during the service life of pavements.

52
 Chapter 6 Summary and Conclusions

With an integrated experimental-statistical approach, this report investigated several critical SCB
testing variables (i.e., the minimum recommended number of specimens, thickness, notch length,
loading rate, and testing temperature) that are considered to have a significant effect on the overall
fracture behavior of AC mixtures at intermediate service temperature conditions. The first part of
this research focused on development of a reliable and repeatable SCB test for AC fracture
characterization. Each testing variable of the five was investigated in turn with a typical range to
estimate testing repeatability and practicality based on a threshold level of COV. The second part
of this study analyzed sensitivity of the SCB testing for difference in AC mixtures. In this part, a
total of 14 AC mixtures were collected from 12 different field construction projects and were tested
in the University of Nebraska’s Geomaterials Laboratory. Statistical analyses were then conducted
to evaluate the repeatability and sensitivity of SCB test results. Based on the test-analysis results,
the following conclusions can be drawn:
 The statistical analysis of a total of 18 SCB specimens indicated that five to six SCB
specimens would be a reasonable sample size that can sufficiently represent asphalt
concrete fracture behavior with a 95% level of confidence.
 A range of 40 mm to 60 mm for the specimen thickness showed good repeatability (COV
≤ 10%) and similar fracture energies, while the test results with 30 mm SCB thickness
showed a high COV (> 25%).
 Within the range of notch lengths tested in this study (0, 5, 15, 25, 40 mm), the 5 mm
showed the lowest value of COV of fracture energy. However, based on the resulting crack
propagation at each notch, a 15 mm was chosen as a recommended notch length.
 Fracture energy was not dependent on loading rates between 1 mm/min. to 5mm/min. with
a good testing repeatability (low COV).
 In the range of testing temperatures attempted here, fracture energy at around 15°C showed
the lowest variation. SCB testing at 21°C also seems attractive for practical purposes, with
a little loss of testing repeatability compared to 15°C. This is because 21°C is close to room
temperatures and is easily achievable without having a sophisticated temperature control
equipment.

53
 Fracture energy values of the 14 AC mixtures collected from 12 separate field construction
projects showed acceptable COV values (generally less than 15%).
 The one-way ANOVA (with -value of 0.05) of fracture energies from the 14 AC mixtures
rejected the null hypothesis of equality of means of mixtures. This implies that at least one
mixture was significantly different from others at the 95% confidence level.
 The Tukey’s HSD multiple-comparison analysis could distinguish AC mixtures into
different performance groups (A to E). In general, SLX and SPH mixtures were grouped
into A or B, and SPR, SRM, and SPS mixtures were categorized into lower-graded groups
like C to E.
 Overall, the SCB test conducted with the testing variables recommended was repeatable
for the same mixture and sensitive for different types of mixtures.
 Test results for the 14 field mixtures indicated that fracture energy increased with
increasing content (percentage of the total mixture weight) of virgin asphalt.
 The findings of this study are under further evaluation for various Nebraska AC mixtures
that were placed in field projects. This will lead to closer insights into the SCB fracture test
through a potential quality control (QC) – quality assurance (QA) type approach to evaluate
the fatigue cracking potential of AC mixtures. Any further findings will be reported in
follow-up studies.

54
 References

Al-Qadi, I. L., et al. (2015). Testing Protocols to Ensure Performance of High Asphalt Binder
Replacement Mixes Using RAP and RAS, Illinois Center for
Transportation/Illinois Department of Transportation.

Allen, D. H., et al. (2009). "Determining Representative Volume Elements of Asphalt Concrete
Mixtures Without Damage." Transportation Research Record: Journal of the
Transportation Research Board 2127: 52-59.

Aragão, F. T. S. and Y. R. Kim (2012). "Mode I Fracture Characterization of Bituminous Paving

Mixtures at Intermediate Service Temperatures." Experimental Mechanics
52(9): 1423-1434.

Biligiri, K. P., et al. (2012). "Asphalt Mixtures' Crack Propagation Assessment using Semi-
Circular Bending Tests." International Journal of Pavement Research and
Technology 5(4): 209.

Brühwiler, E., et al. (1990). "Fracture of AAC as influenced by specimen dimension and
moisture." Journal of materials in civil engineering 2(3): 136-146.

Chong, K. and M. Kuruppu (1984). "New specimen for fracture toughness determination for
rock and other materials." International Journal of Fracture 26(2): R59-R62.

Dowdy, S., et al. (2011). Statistics for research, John Wiley & Sons.

Duan, K., et al. (2003). "Thickness effect on fracture energy of cementitious materials." Cement
and concrete research 33(4): 499-507.

European Committee for Standardization, B., Belgium. (2010). "Bituminous mixtures—Test

methods for hot mix asphalt—Part 44: Crack propagation by semi-circular
bending test." (EN 12697-44: 2010).

Faruk, A. N., et al. (2014). "Using the Fracture Energy Index Concept to Characterize the HMA
Cracking Resistance Potential under Monotonic Crack Testing." International
Journal of Pavement Research and Technology 7(1): 40.

Huang, B., et al. (2013). "Using notched semi circular bending fatigue test to characterize
fracture resistance of asphalt mixtures." Engineering Fracture Mechanics
109: 78-88.

Im, S., et al. (2014). "Characterization of mode-I and mode-II fracture properties of fine
aggregate matrix using a semicircular specimen geometry." Construction and
Building Materials 52: 413-421.

55
Im, S., et al. (2014). "Mode-Dependent Fracture Behavior of Asphalt Mixtures with Semicircular
Bend Test." Transportation Research Record: Journal of the Transportation
Research Board(2447): 23-31.

Im, S., et al. (2013). "Rate-and Temperature-Dependent Fracture Characteristics of Asphaltic

Paving Mixtures." JOURNAL OF TESTING AND EVALUATION 41(2):
257-268.

Kim, Y.-R. and F. T. S. Aragão (2013). "Microstructure modeling of rate-dependent fracture

behavior in bituminous paving mixtures." Finite Elements in Analysis and
Design 63: 23-32.

Kim, Y.-R., et al. (2009). "Determining representative volume elements of asphalt concrete
mixtures without damage." Transportation Research Record: Journal of the
Transportation Research Board(2127): 52-59.

Kim, Y., et al. (2010). "Geometrical evaluation and experimental verification to determine
representative volume elements of heterogeneous asphalt mixtures."
JOURNAL OF TESTING AND EVALUATION 38(6): 660-666.

Li, X. J. and M. O. Marasteanu (2009). "Using Semi Circular Bending Test to Evaluate Low
Temperature Fracture Resistance for Asphalt Concrete." Experimental
Mechanics 50(7): 867-876.

Lim, I., et al. (1993). "Stress intensity factors for semi-circular specimens under three-point
bending." Engineering Fracture Mechanics 44(3): 363-382.

Liu, J. H. (2011). Fatigue life evaluation of asphalt rubber mixtures using semi-circular bending
test. Advanced Materials Research, Trans Tech Publ.

Machiwal, D. and M. K. Jha (2012). Hydrologic time series analysis: Theory and practice,
Springer Science & Business Media.

Marasteanu, M. O., et al. (2002). "Determining the low-temperature fracture toughness of asphalt
mixtures." Transportation Research Record: Journal of the Transportation
Research Board 1789(1): 191-199.

Marasteanu, M. O., et al. (2004). "Low Temperature Cracking of Asphalt Concrete Pavement."

Paulino, G. H., et al. (2004). Cohesive zone modeling of fracture in asphalt concrete.
Proceedings of the 5th International RILEM Conference–Cracking in
Pavements: Mitigation, Risk Assessment, and Preservation, Limoges, France.

Razali, N. M. and Y. B. Wah (2011). "Power comparisons of shapiro-wilk, kolmogorov-smirnov,

lilliefors and anderson-darling tests." Journal of Statistical Modeling and
Analytics 2(1): 21-33.

56
Romero, P. and E. Masad (2001). "Relationship between the representative volume element and
mechanical properties of asphalt concrete." Journal of materials in civil
engineering 13(1): 77-84.

Saadeh, S., et al. (2014). Correlation of Semi-circular Bending and Beam Fatigue Fracture
Properties of Asphalt Concrete Using Non-Contact Camera and Crosshead
Movement. T&DI Congress 2014@ sPlanes, Trains, and Automobiles,
ASCE.

Shen, B. and G. Paulino (2011). "Direct extraction of cohesive fracture properties from digital
image correlation: a hybrid inverse technique." Experimental Mechanics
51(2): 143-163.

Shu, X., et al. (2010). "Evaluation of cracking resistance of recycled asphalt mixture using semi-
circular bending test." Paving materials and pavement analysis (GSP 203):
58-65.

Song, S., et al. (2008). "δ25 Crack opening displacement parameter in cohesive zone models:
experiments and simulations in asphalt concrete." Fatigue & Fracture of
Engineering Materials & Structures 31(10): 850-856.

Song, S. H., et al. (2005). Cohesive zone simulation of mode I and mixed-mode crack
propagation in asphalt concrete. Geotechnical Special Publication No. 130:
Advances in Pavement Engineering, Proceedings of Sessions of the
GeoFrontiers 2005 Congress.

Wagoner, M., et al. (2005). "Investigation of the fracture resistance of hot-mix asphalt concrete
using a disk-shaped compact tension test." Transportation Research Record:
Journal of the Transportation Research Board(1929): 183-192.

Wagoner, M. P., et al. (2005). "Development of a single-edge notched beam test for asphalt
concrete mixtures." JOURNAL OF TESTING AND EVALUATION 33(6):
452.

Wittmann, X. and H. Zhong (1996). On some experiments to study the influence of size on
strength and fracture energy of concrete, Aedificatio Verlag.

Wu, Z., et al. (2005). "Fracture resistance characterization of superpave mixtures using the semi-
circular bending test." Journal of ASTM International 2(3): 1-15.

Zegeye, E., et al. (2012). "Investigation of size effect in asphalt mixture fracture testing at low
temperature." Road Materials and Pavement Design 13(sup1): 88-101.

57
 Appendix: Location of Field Construction Projects

SPH :
SPH_22586

SPH_42514 SPH_42515

SLX :

SLX_M4TLOA

SLX_M1041

SLX_M4TLOB

58
SRM :

SRM_42567

SPR :

SPR_42567 SPR_42454

SPR_42399 SPR_12890 SPR_13963

59
SPS :

SPS_42399 SPS_42514

60