See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/300149493

A new approach in fatigue testing and evaluation of hot mix asphalt using a
dynamic shear rheometer

Chapter · June 2015

DOI: 10.1201/b18538-51

CITATIONS READS

5 331

2 authors:

Taher Ahmed Hussain Al-Khalid

University of Anbar University of Liverpool
13 PUBLICATIONS   27 CITATIONS    27 PUBLICATIONS   119 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fatigue behaviour of precast concrete raft pavements View project

Evaluating fatigue performance of hot-mix asphalt using degradation parameters View project

All content following this page was uploaded by Taher Ahmed on 29 December 2016.

The user has requested enhancement of the downloaded file.

 Bituminous Mixtures & Pavements VI – Nikolaides (Ed.)
© 2015 Taylor & Francis Group, London, ISBN 978-1-138-02866-1

A new approach in fatigue testing and evaluation of hot mix asphalt

using a dynamic shear rheometer

Taher M. Ahmed & Hussain A. Khalid

School of Engineering, University of Liverpool, UK

ABSTRACT: This paper describes a successful trial using a Dynamic Shear Rheometer (DSR) for fatigue
testing of Hot Mix Asphalt (HMA) under controlled strain and stress test modes. A new fatigue index, FIR ,
has been derived from the stress–pseudostrain hysteresis loop to be used in evaluating fatigue performance.
Results have shown that there is a plateau region in the relationship of FIR and normalised shear modulus, which
can be used to evaluate fatigue performance. FIR values were in agreement in terms of the ranking order with
the results from other reliable approaches, such as the traditional and energy ratio approaches. In this work,
limestone and granite aggregates were used with two binder grades: 40/60 and 160/220 to prepare four mixtures
with two different gradations: gap-graded hot rolled asphalt and continuously graded dense bitumen macadam.
The study demonstrated the suitability of the DSR for fatigue testing of full HMA. Also, limestone mixes were
better fatigue performance than granite.

Keywords: Dynamic shear rheometer, fatigue index, pseudostrain energy, hot mix asphalt, strain and stress test
modes

1 INTRODUCTION time-consuming and inefficient. At typical strain or

Fatigue cracks represent one of the major distress
forms in asphalt pavements. Cracks result in degra-
dation of the pavement materials and eventually of the
pavement structure. Many variables such as temper-
ature, time of loading and material properties play a
role in fatigue failure. Fatigue in asphalt is generally
evaluated with repeated load tests on prepared samples
either using a constant applied load (controlled stress),
or using constant displacement (controlled strain).
During fatigue life under repeated loading, materi-
als exhibit three phases (Di Benedetto et al. 2004):
phase I, or the adaptation phase, is characterised by
a rapid decrease in stiffness due to heat generation
and thixotropy (Carpender & Shen 2006, Ghuzlan &
Carpenter 2006); phase II, or the quasi-stationary
phase, is where the steady decrease in stiffness pre-
dominates fatigue life, and the degradation in the
material is very slow over a number of load cycles;
and phase III, or the failure phase, is where macro-
cracks begin to develop and failure completely sets
into the material at the end of this phase. Phases I and
II are associated with crack initiation, while phase III
is associated with crack propagation.
Traditionally, asphalt fatigue tests involve the man-
ufacture of relatively large-size beams or cylinders,
which are then conditioned and tested in flexure,
tension–compression or diametrically to determine
their fatigue properties. These tests may take a consid-
erable amount of time to finish, thus rendering them
stress levels adopted in a classical laboratory fatigue
test, the duration can be anything from one day to sev-
eral days (or even more), which ties up resources and
leads to considerable delays in arriving at the required
results.
In a move to rationalise test procedures and ensu-
ing analyses for asphalt mixtures, a white paper by
Christensen et al. (2009) advocated the standardisa-
tion of test and analysis protocols, which would lead
to a significant reduction in material quantities and
test duration; this protocol suggested fatigue testing
and analysis based on a reduced cycle, simplified
viscoelastic continuum damage (S-VECD) and dam-
age mechanics analysis. In recent years, the Dynamic
Shear Rheometer (DSR) has been used successfully
as a technique in fatigue testing, but it was limited for
use only with binder, mastic and fine aggregate mix-
ture samples (Christensen et al. 2009, Kim et al. 2002,
Kim & Little 2004, Huang et al. 2007, Masad et al.
2008, Tan et al. 2012, Woldekidan et al. 2013, Hintz &
Bahia 2013).
In the literature, various approaches have been used
as criteria to evaluate fatigue performance. The most
common is called the traditional approach, in which
fatigue failure is defined based on the mode of loading.
For example, in the controlled strain mode, fatigue fail-
ure is defined as the number of cycles needed to reach
50% reduction in the initial stiffness modulus (Kim
et al. 2003, Daniel et al. 2004, Artamendi & Khalid
2005), whereas in the controlled stress mode, failure

351
 is defined either as the complete fracture of the sample
(Ghuzlan & Carpenter 2006, Tayebali et al. 1992), or
reduction in stiffness modulus of the sample to 10%
of its initial value (Rowe 1993). The American Asso-
ciation of State Highway and Transportation Officials
(AASHTO 2002) adopted the traditional approach in
evaluating fatigue performance of asphalt pavements.
(AASHTO 2002). The 50% reduction in stiffness mod-
ulus for the controlled strain mode predominantly falls
within phase II, which is pragmatically considered as
representative of the material’s useful life. Mix’s resis-
tance to fatigue cracking, however, may continue until
the second inflection point is reached, where phase
III begins. This means that some data within phase
II will be neglected because the test is stopped at
50% reduction, which may give misleading results by
underestimating fatigue life. As this approach mea-
sures the number of cycles, the results always have high
variability and data scatter; to minimise this effect, a
large number of samples is required to obtain reliable
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015
results and this is time-consuming and expensive.
Applying a stress to a material creates a strain, and
the area under the stress–strain curve represents the
energy being put into the material. For elastic materi-
als without any damage, the strain is recovered directly
when the stress is removed and the stress–strain curve
coincides, meaning the energy is recovered completely
without losses. If the two curves for loading and
unloading do not coincide, this indicates there is a
difference between the amount of energy put into the
material and the amount of energy recovered from the
material, and this difference in energy loss is called
dissipated energy. In this regard, an approach related
to the dissipated energy concept was introduced by
Van Dijik (1975, 1977) to characterise the fatigue per-
formance of HMA. In Van Dijik’s work, the fatigue
criterion was defined as the point where the dissipated
energy ratio (ER) begins to increase rapidly with num-
ber of cycles from the plateau value. ER was defined
(Rowe 1993) as a function of the number of cycles and
stiffness modulus for both strain and stress modes, as
in Equations 1–3.
where: n = cycle number; ε = strain amplitude;
σ = stress amplitude; δ = phase angle; Rσ & Rε =
equivalent energy ratio for controlled stress and strain
mode respectively; E∗ = complex stiffness modulus;
and 0, i = initial and ith cycle.
The criterion to define fatigue life in the stress mode
is the number of cycles at the point when the ER
reaches the peak point in the relationship of ER vs n,
while in the strain mode it is defined as the point when
the ER slope deviates from a straight line in the same
relationship. The variation in results that is related to
352
number of cycles such as the traditional approach and
ER is relatively high [8]; therefore to reduce the scatter
in the results, more samples are required.
Another energy-related approach, referred to as dis-
sipated pseudostrain energy (DPSE) (Masad et al.
2008, Bhasin et al. 2009) has been used in evaluating
the fatigue performance of HMA. In this approach,
real strain amplitude is converted to an equivalent
pseudostrain to remove the viscoelastic contribution to
dissipated energy. DPSE rate has been used to measure
fatigue cracking resistance using a parameter called
the crack growth index (Masad et al. 2008, Bhasin
et al. 2009), which is based on Paris’s law for crack
growth and the J-integral. The results of the crack
growth index were found to be consistent with field
observations and independent of fatigue test mode
(Bhasin et al. 2009). However, this technique used the
so-called ‘crack radius’, which is practically impossi-
ble to measure because the crack always creates new
surfaces with increase in the crack mouth opening
displacement.
A further performance evaluation approach was
based on Schapery’s work on crack growth in vis-
coelastic media (Schapery 1984) and is called contin-
uum damage mechanics (CDM). This approach was
developed by Lee and Kim (Kim et al. 1997, Lee et al.
2000, Park et al. 1996) to characterise the damage in
asphalt due to fatigue. The damage was quantified by
a non-dimensional internal state variable (S), and this
variable represents an indicator of the change in the
pseudo-stiffness modulus (C). Fatigue is considered
to have occurred when the pseudo-stiffness drops to
50% of the initial value. The C vs S curve represents
a unique depiction of the deterioration of the material
independent of test mode, temperature, frequency and
other test variables. Although this approach requires
somewhat complex calculations in addition to relax-
ation test parameters, it is versatile, can be used in
mechanistic pavement design and enables reduction
in number of fatigue samples tested.
In this work, a fatigue test approach using the
DSR was introduced based on selecting a specified
strain and stress amplitude. Also, a new index was
derived based on the recovered and applied pseudos-
train energy used in analysing the fatigue performance
of HMA.
2 OBJECTIVES OF STUDY
This study aims at introducing a new index to evalu-
ate the fatigue performance of asphalt mixtures tested
using the DSR. The ability of the proposed fatigue
index, FIR , to rank mixtures in respect of their resis-
tance to fatigue cracking is validated through compar-
ison with established fatigue performance evaluation
tools, namely the traditional and ER approaches dis-
cussed above. The validation process was facilitated
through fatigue testing in the DSR of four asphalt
mixtures including two aggregate gradations, namely
gap-graded and continuous; two aggregates, limestone
 and granite; and two binder grades, hard 40/60 and soft
160/220 Pen binders. The ranking efficiency of FIR for
the mixtures tested has been demonstrated.

3 BACKGROUND AND THEORY

In fatigue tests using sinusoidal loading and unload-

ing, two types of energy are produced: applied energy
during loading and recovered energy during unload-
ing. The difference between both energies is called the
dissipated energy, which is equal to the hysteresis loop Figure 1. Typical one-cycle loading.
area (Ghuzlan & Carpenter 2006, Daniel et al. 2004).
A sinusoidal wave was used in this work for fatigue
testing to calculate the applied strain energy (ASE)
and recovered strain energy (RSE). Figure 1 shows the
typical sine wave of one-cycle loading for time load-
ing t0 to t6 . Because the calculations are based on the
DSR test, the time for one cycle is split into two parts:
TR (right direction of loading) and TL (left direction of
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015

loading). Figure 2 is a stress–strain graph representing

a typical hysteresis loop for a viscoelastic material. As
shown in Figures 1 and 2, there are two intervals for
loading t0 to t1 and t3 to t4 in TR and TL , respectively, as
well as t1 to t3 and t4 to t6 , which are unloading periods
in TR and TL , respectively. However, strain increases Figure 2. Typical hysteresis loop.
during unloading in periods t1 to t2 and t4 to t5 because
of the viscoelastic behaviour of the material, this inter-
Table 1. Integration limits of the strain energy formula.
val being the phase angle (δ). As shown in Figure 2,
there are two parts to ASE: the first one, ASE1 , is under Energy tl tu Period
the curve t0 t1 t2 , and the second part, ASE2 , is above
the curve t3 t4 t5 ; while RSE1 is the shaded area under π δ
the curve t2 t3 and RSE2 is the shaded area within the ASE1 0 + [t0 ,t2 ]
2ω ω
region t5 t6 . π 3π δ
A useful technique to calculate ASE and RSE is to ASE2 + [t3 ,t5 ]
ω 2ω ω
use the integration over the periods [t0 , t6 ] with the π δ π
basic equation of the energy formula, as in Equation 4 RSE1 + [t2 ,t3 ]
2ω ω ω
(Larson 1999, Schapery 1990): 3π δ 2π
RSE2 + [t5 ,t6 ]
2ω ω ω

where W is the strain energy and τ(t) and γ(t) are shear
stress and strain, respectively. Equations 5 and 6 are
the sinusoidal shear stress and strain functions.

where τ is shear stress; τ0 is shear stress amplitude; γ

is shear strain; γ0 is shear strain amplitude; δ is phase
angle; t is loading time and ω is angular frequency in
rad/s.
To calculate the pseudostrain energy (PSE), the
pseudostrain relationship in the formula below (Equa-
tion 7) has been used (Masad et al. 2008). The
final formulations for applied pseudostrain energy and
recovered pseudostrain energy are given in Equations where γ R = pseudostrain; G∗lve is dynamic modulus at
9 and 10 after integration processes using integration linear viscoelastic; GR is reference modulus; δN is
limits in Table 1. phase angle; δlve is phase angle at linear viscoelas-
tic; WRA , WRR are applied and recovered pseudostrain
energy.

353
 4 EXPERIMENTAL WORK

The experimental work involved the preparation of

12 mm in diameter and 50 mm high cylindrical sam-
ples from HMA; and performing fatigue testing, which
includes determining the appropriate strain and stress
level to be used in the fatigue tests. Fatigue perfor-
mance was evaluated using a fatigue index based on
the pseudostrain energy. For verification of this index,
two well-known approaches have been used to com-
pare the results; these approaches are: the traditional
approach and the ER approach.
Figure 3. Grading curves of aggregate for DBM [28].

5 MATERIALS AND MIX DESIGN

In this work, two kinds of mix: a hot rolled asphalt

(HRA) and dense bitumen macadam (DBM) were
prepared in the laboratory using two types of aggre-
gate: limestone (L) and granite (G), with two binders:
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015

40/60 and 160/220 penetration grades. The mixes

were made according to British Standards recipes
(BS 4987-1 2005, BS 597-1 2005, BS PD 6691
2010) and were denoted as DBM-G, DBM-L, HRA-G
and HRA-L. Figures 3 and 4 show the particle size
distribution of both mixes DBM and HRA, respec-
tively. Steel moulds were used to prepare asphalt slabs Figure 4. Grading curves of aggregate for HRA [28].
(305 × 305 × 65 mm), which were compacted using a
roller compactor (BS EN 12697-33 2003). The slabs Table 2. Mix ID and material details.
were then cut into five beams (305 × 65 × 50 mm),
two of which were discarded, and the remaining three Aggregate
were cored with an electric coring machine to obtain
the cylindrical DSR samples. Table 2 shows the mix Apparent
details. Mix type Type density (Mg/m3 )

DBM Limestone 2.71

6 PREPARATION OF DSR SAMPLES DBM Granite 2.67
HRA Limestone 2.71
In the literature, two techniques have been used to pre- HRA Granite 2.67
pare DSR samples: Kim et al. (2002, 2003) prepared
Mix properties
15 g individual samples using a steel cylindrical mould
12 mm in diameter and 50 mm in height compacted by ID Gbulk (Mg/m3 ) Air voids (%) Binder grade
using a cylindrical rod at both ends. Zollinger (2005)
prepared cylindrical samples 152 mm in diameter and DBM-L 2.374 4.9 160/220
85 mm in height using the Superpave gyratory com- DBM-G 2.290 7.5 160/220
pactor (SGC) and then cored them to obtain small HRA-L 2.343 2.2 40/60
samples. Both techniques, however, have been used HRA-G 2.298 4.0 40/60
only with fine aggregate matrix (FAM) mixtures. In
this study, full asphalt mixtures were produced whose
properties are more representative than FAM mixtures
of end stubs and adjusting the fastening mechanism
of the material in the field. The number of DSR sam-
that was used to hold and set up samples in the proper
ples cored from each beam was 43–46. These samples
position for testing. To control the sample temperature
were put in an oven for 24 h at 25◦ C to dry and kept in
during testing, a temperature control unit (TCU) was
PVC tubes after coding, and then stored in a fridge at
designed and added to the DSR as an essential part.
10◦ C. Figure 5 shows DSR samples.
The full system is shown in Figure 6.

7 INSTRUMENTATION
In this work, a Kinexus-Pro DSR (Malvern) was used
to test the cylindrical HMA.This was done after adding
some modifications to the DSR to enable it to be tested.
These modifications included design and manufacture
8 ANALYSIS OF VOLUMETRIC PROPERTIES
Scattering in the fatigue data is common; therefore,
it is necessary to test a large number of samples to
enhance the reliability of the results (Di Benedetto

354
 Figure 5. (a) Coring beam for obtaining DSR samples, (b)
DSR samples and (c) sample ready for testing.
Figure 7. Typical bulk density histogram for HRA-L DSR
samples.

Table 3. Bulk density and air void range of DSR samples.

Bulk density Mg/m3 Air voids range (%)

Mix ID Upper Mode Lower σ(Mo) Upper Mode Lower

Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015

DBM-L 2.413 2.398 2.383 0.015 4.486 3.872 3.259

DBM-G 2.373 2.350 2.327 0.023 6.023 5.101 4.179
HRA-L 2.368 2.356 2.344 0.012 2.153 1.641 1.130
HRA-G 2.327 2.311 2.295 0.016 4.150 3.487 2.824

Figure 6. (a) The DSR apparatus assessors: and (b)

schematic of system.
9 RESULTS AND DISCUSSIONS

9.1 Determination of strain and stress amplitudes

et al. 2004). In this study, an approach was used to
select prepared samples for further use that would
reduce the variability within the total number of sam-
ples prepared. This approach is based on measuring
the sample’s bulk density in according with the proce-
dure in the British Standards (BS EN 12697-6 2004).
The measured bulk density varied according to aggre-
gate grading, shape, type and binder content, among
other factors. The approach starts by finding the mode,
‘Mo ’, of the measured sample’s bulk density, i.e. the
value with the highest frequency. This value, Mo , was
used to calculate the standard deviation, σ, of the bulk
density data using Equation 11. A range, R, was then
calculated, using Equation 12, which is one σ value
on either side of Mo . Thus, only those samples whose
bulk density value fell within R were chosen for fur-
ther use in the project. Table 3 summarises the values
of bulk density and range limits for all the mixtures
in the study and shown in Figure 7, together with the
corresponding air voids.
where σ is the standard deviation; Mo is the mode
value; Xi is a single bulk density measurement; N is
the number of measurements and R is the range of bulk
density of selected DSR samples.
In the literature, different techniques have been
applied. Kim et al. (2002, 2003) used amplitude sweep
strain to find the maximum stress value that could be
applied to the sample before causing damage. Masad
et al. (2008) used high strain amplitude value 0.2% in
fatigue testing for DSR samples prepared from fine
aggregate matrix (FAM), and the stress corresponding
to a 50% reduction of initial complex shear modulus
(G∗ ) was used for fatigue testing in the stress test mode.
In this work, strain and stress amplitude for fatigue
test was identified at the damage zone. To determine
the strain amplitude in fatigue testing, sweep strain
amplitude was used and performed for three samples,
which were arbitrarily selected within the range R in
Table 3; the changes in the slope of the strain–stress
curve at each point were calculated using Equation 13.
where τi and τi+1 are shear stresses at shear strains γi
and γi+1 ; respectively.
At the instant when the slope changed from positive
to negative, the value of the strain amplitude at the
damaged region was identified, as shown in Figure 8.
Table 4 details the strain range at starting damage and
strain amplitude for fatigue testing. It is clear that the
damage occurs earlier in DBM-G and the amplitude
range (0.2–0.3%) is shorter than that of DBM-L (0.3–
0.45%). This is possibly because air voids in DBM-G
are higher than air voids in DBM-L, as detailed inTable
3. In HRA, the damaged regions for both granite and

355

Figure 8. A typical sweep strain amplitude of HRA-L.
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015
Figure 9. A typical stress amplitude sweep of HRA-L.
limestone were equal, perhaps because both mixes are
homogenous.
Amplitude sweep stress test was performed on three
DSR samples selected within the range R in Table 3.
The strain response has been collected, and then strain
response-time slopes were calculated for each stress
value, which was applied to sample for several cycles’
The strain response against time was plotted and fitted
linearly for each stress amplitude value. The slopes
of strain-time (dγ/dt) were evaluated from the fitted
equations and also plotted against stress amplitude. At
the instant when the slope (dγ/dt) increased dramati-
cally, the stress amplitude was taken as stress value for
fatigue test, as demonstrated in Figure 9.
It is noticeable that there is a marked increase in
the strain response slopes which starts at definite shear
stress value; this value can be defined at the point when
there is divergence from the straight line in the rela-
tionship of strain response slope against shear stress
amplitude, as demonstrated in Figure 9. This point is
considered as start damage in materials, whereas it
was about 150 KPa for DBM mixes and about 250 KPa
and 400 KPa for HRA-L and HRA-G, respectively, as
detailed in Table 4.
9.2 Fatigue analysis
9.2.1 Fatigue index
A new fatigue index (FIR ) was formulated based
on the energy concept as a function of the ratio of
pseudo-recovered energy to pseudo-applied energy
(Eqn. 14). Logically, when WRR is equal to WRA , then
356
Table 4. Strain/stress amplitude test results.
Shear strain (%)
Shear stress
Mix ID at damage for fatigue at damage (KPa)
DBM-L 0.30–0.45 0.30 150
DBM-G 0.20–0.30 0.30 150
HRA-L 0.25–0.35 0.25 250
HRA-G 0.25–0.35 0.25 400
Figure 10. Fatigue index against normalised shear modulus
for controlled strain test mode.
the fatigue index equals zero, which means no dam-
age has occurred in the material. Otherwise, when the
WRR is equal to zero, then the fatigue index equals one,
which means that the sample is completely damaged.
So, FIR range is between one and zero.
Firstly, five samples were selected within the range
R to be tested for fatigue using the shear strain and
stress values in Table 4. Figures 10 and 11 show
the relationship of FIR plotted against the normalised
shear modulus for controlled strain and stress test
mode, respectively. It is clear that there are plateau
values for all mixes where the FIR is barely constant
between 0.85 to 0.35 of normalised shear modulus; and
increases sharply after that, as shown in Figure 10.
This trend is similar to the one shown in the dis-
sipated energy ratio approach (Daniel et al. 2004). In
contrast, the plateau region (PR) in stress test mode
is not as clear as in the strain mode, as shown in
Figure 11; however, FIR increased sharply beyond
0.2 of normalised shear modulus; this behaviour was
found to be the same for the relationship of DER vs n
(Ghuzlan & Carpenter 2006). The PR was defined
between 0.85 and 0.2 of normalised shear modulus,
and FIR was calculated as average within PR.
Figures 10 and 11 demonstrate that the performance
of HRA fatigue is higher than DBM (Brown 1995) and
limestone is better than granite in both mixes tested in
both modes.
9.2.2 Verification using other approaches
In this work, traditional approach (TA) (Kim et al.
2003, Daniel et al. 2004, AASHTO TP8-94 2002) and
 Figure 11. Fatigue index against normalised shear modulus
for controlled stress test mode.
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015
Figure 12. Traditional approach for HRA-L mix tested in
strain mode.
Figure 13. Traditional approach for HRA-L mix tested in
stress mode.
energy ratio (ER) approach were used to verify FIR .
TA represents the changes in a material’s modulus and
phase angle against number of cycles, as presented in
the typical examples given in Figures 12 and 13 for
strain and stress test mode respectively.
The results of TA revealed that the phase angle
begins to decrease quickly when the normalised shear
modulus drops approximately below 0.35 and 0.2 for
strain and stress test modes, respectively, as demon-
strated in the typical examples given in Figures 12
and 13.
The energy ratio (ER) approach [16] was also used
to study the validity of the FIR parameter. Typical
results for HRA-L are shown in Figures 14 & 15 for
strain and stress test modes, respectively.
The outcomes of fatigue test using DSR and the
analysed results are summarised in Figures 16–18 for
all approaches: FIR , TA and ER. It is clear that the
357
Figure 14. Energy ratio approach for HRA-L mixes tested
in strain mode.
Figure 15. Energy ratio approach for HRA-L mixestested
in stress mode.
Figure 16. Fatigue performance of different mixes tested in
strain and stress modes evaluated using FIR approach.
rank order is compatible in all approaches, as a number
of cycles and fatigue index in both test modes where
the performance of HRA fatigue is higher than DBM
(Brown 1995) and limestone is better than granite, as
shown in Figures 16 – 18. The main conclusion is that
the TA and ER emphasise the feasibility of using the
FIR parameter to evaluate fatigue performance.
10 CONCLUSIONS
The following conclusions have been made based on
the experimental results obtained in this study
1. An approach has been developed and used effec-
tively to prepare full hot mix asphalt DSR samples
of 12 mm in diameter and 50 mm in height. Also,
a method for the statistical reliability of selecting

Figure 17. Fatigue performance of different mixes tested in
strain and stress modes evaluated using TA approach.
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015
Figure 18. Fatigue performance of different mixes tested in
strain and stress modes evaluated using ER approach.
DSR samples was proposed based on determining
the mode and standard deviation of the samples’
bulk density. The method ensures the availability
of an adequate number of samples with minimum
variability for any further testing and/or evaluation
requirements in the project.
2. A simple method was devised to arrive at the
strain and stress levels that should be adopted in
conducting DSR fatigue tests.
3. A new fatigue index, FIR , was developed based
on the concept of applied and recovered pseudos-
train energy to evaluate the fatigue performance of
HMA. This index revealed an agreement with the
results from other reliable approaches, namely the
energy ratio and traditional approach.
4. This approach can only be used for aggregates
up to 10 mm in size due to the limitations of
the DSR capacity as maximum torque is 0.2 Nm.
Although this may be an acknowledged shortcom-
ing, with further development it should be possi-
ble to enhance the torque capacity of this device
to enable testing materials containing aggregates
larger than 10 mm.
5. The results confirmed that limestone gives a better
fatigue performance for both HRA and DBM mixes
than granite.
6. Future work aims to map out fatigue data obtained
using a classical testing method, such as the two-
point bend, to verify the suitability of the DSR
technique in evaluating the laboratory fatigue per-
formance of full asphalt mixtures.
358
ACKNOWLEDGEMENTS
The authors are grateful to several industrial organisa-
tions for the supply of materials used in this project.
The authors also wish to acknowledge the finan-
cial support of the Iraqi government for the award
of a research scholarship enabling this work to be
conducted as part of a larger research project.
REFERENCES
AASHTO, Method for Determining the Fatigue Life of Com-
pacted Hot Mix Asphalt (HMA) Subjected to Repeated
Flextural Bending, AASHTO TP8-94. 2002.
Artamendi, I. and H. Khalid, Characterization of fatigue
damage for paving asphaltic materials. Fatigue & Frac-
ture of Engineering Materials & Structures, 2005. 28(12):
p. 1113–1118.
Bhasin, Amit, Veronica T Castelo Branco, Eyad Masad, and
Dallas N Little, Quantitative comparison of energy meth-
ods to characterize fatigue in asphalt materials. Journal
of Materials in Civil Engineering, 2009. 21(2): p. 83–92.
Brown, S.F. Practical test procedures for mechanical prop-
erties of bituminous materials. in Proceedings of the
Institution of Civil Engineers: Transport. 1995.
BS 4987-1, coated macadam (asphalt concrete) for roads and
other paved areas. 2005, British Standard Institution.
BS 597-1, Hot rolled asphalt for roads and other paved areas,
in specification for constituent materials and asphalt
mixtures. 2005, British Standard Institution.
BS PD 6691, Guidence on the use of BS EN 13108 bituminous
mixtures – material specifications. 2010, British Stendard
Instituation.
BS EN 12697-6, Bituminous mixtures-tast methods for hot
mix asphalt. 2004, British Standard Institution.
BS EN 12697-33, Specimen prepared by roller compactor,
in Bituminous mixtures-test methods for hot mix asphalt.
2003, British Standard Instituation.
Carpenter, S. H. and S. Shen, Dissipated Energy Approach to
Study Hot-Mix Asphalt Healing in Fatigue. Transportation
Research Record, 2006. 1970(1): p. 178–185.
Christensen D., Richared kim, Dallas Little, Eyad Masad,
and Yong-Rak Kim, practical approaches to continuum
damage fatigue analysis. 2009.
Daniel, J.S., W. Bisirri, and Y.R. Kim, Fatigue Evaluation
of Asphalt Mixtures Using Dissipated Energy and Vis-
coelastic Continuum Damage Approaches. Journal of the
Association of Asphalt Paving Technologists, 2004. 73:
p. 557–583.
Di Benedetto, H., C. De La Roche, H. Baaj, A. Pronk,
and Robert. Lundström, Fatigue of bituminous mixtures.
Materials and Structures, 2004. 37(3): p. 202–216.
Ghuzlan, K.A. and S.H. Carpenter, Fatigue damage analysis
in asphalt concrete mixtures using the dissipated energy
approach. Canadian Journal of Civil Engineering, 2006.
33(7): p. 890–901.
Hintz, C. and Bahia, H., Simplification of Linear Amplitude
Sweep Test and Specification Parameter. Transportation
Research Record: Journal of the Transportation Research
Board, 2013. 2370(1): p. 10–16.
Huang, C.W., E. Masad, A.H. Muliana, and H. Bahia, Nonlin-
early viscoelastic analysis of asphalt mixes subjected to
shear loading. Mechanics of Time-Dependent Materials,
2007. 11(2): p. 91–110.
Kim, Y.R., H.J. Lee, and D.N. Little, Fatigue Character-
ization of Asphalt Concrete Using Viscoelasticity and
 Continuum Damage Theory (with Discussion). Journal of
the Association of Asphalt Paving Technologists, 1997.
66: p. 520–569.
Kim, Y.R., D.N. Little, and R.L. Lytton, Use of Dynamic
Mechanical Analysis (DMA) to Evaluate the Fatigue and
Healing Potential of Asphalt Binders in Sand Asphalt
Mixtures. Journal of the Association of Asphalt Paving
Technologists, 2002. 71: p. 176–205.
Kim,Y.R., D. N. Little, and R. L. Lytton., Fatigue and Healing
Characterization ofAsphalt Mixtures. Journal of Materials
in Civil Engineering, 2003. 15(1): p. 75–83.
Kim, Y.R. and D.N. Little, Linear Viscoelastic Analysis of
Asphalt Mastics. Journal of Materials in Civil Engineer-
ing, 2004. 16(2): p. 122–132.
Larson, Ronald G., The structure and rheology of complex
fluids. Vol. 2. 1999: Oxford university press New York.
Lee, H.J., J.S. Daniel, and Y.R. Kim, Continuum dam-
age mechanics-based fatigue model of asphalt concrete.
Journal of Materials in Civil Engineering, 2000. 12:
p. 105–112.
Malvern. http://www.malvern.com/labeng/products/kinexus
Masad, E. , VTF. Castelo Branco, Dallas N. Little, and R.
Downloaded by [Mohammed Sadeq] at 03:45 17 June 2015
Lytton, A unified method for the analysis of controlled-
strain and controlled-stress fatigue testing. International
Journal of Pavement Engineering, 2008. 9(4): p. 233–246.
Park, S.W., Y.R. Kim, and R.A. Schapery, A viscoelastic
continuum damage model and its application to uniax-
ial behavior of asphalt concrete. Mechanics of Materials,
1996. 24: p. 14.
Rowe, GM, Performance of asphalt mixtures in the trape-
zoidal fatigue test. Asphalt Paving Technology, 1993. 62:
p. 344–344.
359
View publication stats
Schapery, RA, Correspondence principles and a general-
izedJ integral for large deformation and fracture analysis
of viscoelastic media. International Journal of Fracture,
1984. 25(3): p. 195–223.
Schapery, R.A., A theory of mechanical behavior of elastic
media with growing damage and other changes in struc-
ture. Journal of the Mechanics and Physics of Solids, 1990.
38(2): p. 215–253.
Tan, Y., L. Shan, Y. Richard Kim, and B.S. Underwood, Heal-
ing characteristics of asphalt binder. Construction and
Building Materials, 2012. 27: p. 570–577.
Tayebali, A. A., G. M. Rowe, and J. B. Sousa, Fatigue
Response of Asphalt-Aggregate Mixtures. Association
of Asphalt Paving Technologists-Proceedings, 1992. 61:
p. 333–360.
Van Dijk and Visser, W., Practical fatigue characterization
of bituminous mixes. Journal of the association of asphalt
paving technologists, 1975. 44: p. 38–72.
Van Dijk and Visser, W. Energy Approach to Fatigue for
Pavement Design. in Association of Asphalt Paving Tech-
nologists Proc. 1977.
Woldekidan, MF, Huurman, M and Pronk, AC, Linear and
Nonlinear Viscoelastic Analysis of Bituminous Mortar.
Transportation Research Record: Journal of the Trans-
portation Research Board, 2013. 2370(1): p. 53–62.
Zollinger, C.J., Application of surface energy measurements
to evaluate moisture susceptibility of asphalt and aggre-
gates. MSc Dissertation, 2005, Texas A&M University.