Available online at www.sciencedirect.

com

ScienceDirect
International Journal of Pavement Research and Technology 11 (2018) 176–184
www.elsevier.com/locate/IJPRT

Aggregate fatigue failure on macro texture polishing of

asphalt pavement
Zhenyu Qian a, Yinghao Miao b, Haocheng Xiong a, Linbing Wang c,d,⇑
a
National Center for Material Service Safety, University of Science and Technology Beijing, Beijing 100083, China
b
Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing 100124, China
c
Joint USTB-Virginia Tech Lab on Multifunctional Materials, USTB, Beijing, China
d
Virginia Tech, Blacksburg, VA 24061, United States

Received 14 June 2017; received in revised form 9 August 2017; accepted 13 September 2017
Available online 20 September 2017

Abstract

The macro-texture of asphalt pavement greatly affects the traffic safety of driving. In this study, the 3D laser scanning data, which is
captured from the pavements, is used for reverse modeling. FE software ANSYS is used to analyze the contact stress distribution of
pavement macro-texture and tire. The results show that the macro texture will be affected by fatigue failure considering the size effect
of coarse aggregate. The fatigue life distribution of pavements is analyzed using the fatigue analysis software FE-SAFE which is based
on linear cumulative damage theory. The developing tendency of texture depth with respect to the number of loading times is obtained by
analyzing the results of fatigue life. Compared with the results of accelerated loading test shows that the factor of fatigue failure has
remarkably affected the macro texture under low loading times. This effect is gradually decreased with the increasing of cyclic loading
times.
Ó 2017 Chinese Society of Pavement Engineering. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Fatigue failure; Reverse modeling; FE simulation; Pavement macro-texture; Fatigue life analysis

1. Introduction macro texture and micro texture according to different

Nowadays, the asphalt pavement is the primary pave-
ment type in road engineering, because of its great skid
resistance performance which is a prerequisite for ensuring
the safety and comfort of traffic. The skid resistance perfor-
mance of asphalt pavement is closely related to the surface
texture which is mainly determined by the shape, size, and
distribution of asphalt aggregate. It can be divided into
⇑ Corresponding author at: Virginia Tech, Blacksburg, VA 24061,
United States.
E-mail addresses: qianzhy@hotmail.com (Z. Qian), miaoyinghao@b-
jut.edu.cn (Y. Miao), xiong.haocheng@gmail.com (H. Xiong), wangl@vt.
edu, lbwang@ustb.edu.cn (L. Wang).
Peer review under responsibility of Chinese Society of Pavement
Engineering.
scales [1–6]. Analysis based on the energy loss of rubber
tires shows that the higher the height of pavement macro
texture, the greater the hysteresis energy loss of the tire
with elastic deformation, resulting in a better pavement
skid resistance performance [7]. Cairney [8] suggested that
traffic accident rate will be raised, when the pavement
macro texture was degraded to a lower level. Laganier [9]
concluded that pavement roughness and surface macro tex-
ture would significantly affect the vehicle fuel consumption.
Therefore, research on the resistance mechanism of the
macro texture of asphalt pavement is of great importance.

https://doi.org/10.1016/j.ijprt.2017.09.009
1996-6814/Ó 2017 Chinese Society of Pavement Engineering. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184
2. Fatigue failure mechanism of pavement macro texture
A large number of studies show that the damage of
pavement macro texture is mainly influenced by the prop-
erty of asphalt aggregate and traffic condition [10–13]. In
actual use of the pavement, the contact surface of tire
and pavement is an approximately elliptical rectangle,
due to the vehicle load and the influence of the inner pres-
sure of the tire [14,15]. The tire and the macro texture of
pavement are in contact closely with this contact surface,
which makes the tire to an elastic deformation of flexure
in the vertical direction to wrap the rough edges of macro
texture, as shown in Fig. 1(a). Meanwhile, micro slip phe-
nomenon could occur between tire and pavement. Even if
the contact condition is pretty good, the slip rate of the tire
will still reach 5–10% [16]. Therefore, the macro texture of
pavement will bear the vehicle load in the vertical direction
and the friction force in the longitudinal direction, which is
generated by the micro slip of the tire, as shown in Fig. 1
(b). With the repeated actions of the vehicle, the macro tex-
ture of pavement is under the cyclic stress constantly,
resulting in a periodic disturbance of coarse aggregates
exposed on the pavement surface.
The coarse aggregate used in asphalt pavement is mainly
composed of rock material. As a kind of quasi-brittle mate-
rial, the process of rock failure caused by cyclic loading is
quite complicated [17–19]. There are a large number of
micro cracks, pores, and other micro structures in natural
rock when the rock is affected by external loading, these
(a)
(b)
Fig. 1. The micro scale contact between tire and pavement.
177
micro structures will increase and expand, and the cracks
will also be connected with each other. After unloading,
these micro structures will be closed till the cyclic loading
is loaded again, the micro structures will expand continu-
ously along the fractures, and the fractures gradually
develop until rock failure [20]. Therefore, if the internal
fatigue damage of exposed coarse aggregate of asphalt
pavement is generated by the stress, then with the increase
of cyclic times, the damage is continuously accumulated till
failure.
3. Numerical simulation analysis
3.1. Establishment of the pavement macro texture model
In order to analyze the stress distribution of pavement
macro texture accurately, reverse models based on real
pavement is used for finite element analysis. This study uses
hand-held high-resolution 3D laser scanner produced by
Creaform Inc. to capture the spatial coordinates of macro
texture from the real pavement, as shown in Fig. 2. The
(a) Hand-held 3D laser scanner
(b) Field acquisition of pavement macro texture
Fig. 2. Acquisition of the spatial coordinates of pavement macro texture.
178 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184

principle is based on laser triangulation technique, where

its maximum scanning resolution is up to 0.05 mm.
This study considers the asphalt pavement which grada-
tion is SMA-13, as shown in Fig. 3, which coarse aggregate
type is basalt, the project area of scanned pavement was
90 mm
90 mm [21]. After that, the spatial coordinates
of pavement texture are imported into Rhinoceros 3D
modeling software, to generate the point cloud of pave-
ment texture, then to reversely create the surface model
of pavement texture based on the point cloud. After that,
the reconstructed model is imported into the finite element
analysis software ANSYS for further simulation. Since this
study is mainly focused on the contact stress on exposed
pavement aggregate, the pavement model is considered
Fig. 4. Pavement texture model in ANSYS.
as a rock with specific texture, as shown in Fig. 4.

3.2. Finite element analysis of the pavement macro texture

To analyze the contact stress between tire and pavement

in ANSYS, the following assumptions need to be estab-
lished. Firstly, considering the computing capability of
computer, this study considers the central part of the pave-
ment texture model, where the pavement contact area is
28 mm
31 mm, meanwhile, according to the actual con-
tact conditions of tire and pavement, the tire contact area
is regarded as a rectangular area whose projected area is
21 mm
21 mm. In this study, the thickness of pavement
model is considered based on the size of coarse aggregate
which is 13 mm, and the final contact model is shown in
Fig. 5. The material model of tire is using the three order Fig. 5. Tire-pavement contact model.
Ogden model, where the parameters of Ogden model deter-
two load steps, where the first step is to apply pressure on
mines the physically realistic response and material stabil-
the tire, and the tire is affected by the rough edges of the
ity of the tire [22]. The friction coefficient is taking as 0.6.
asphalt aggregate, resulting in a radial elastic deformation
The pavement is assumed as a continuum model, and the
to wrap the macro texture of asphalt pavement. The second
material parameters are referenced to basalt, as shown in
step is to simulate the micro slip between tire and pavement
Table 1.
by controlling the displacement of the tire. It is actually a
The boundary conditions of the model are the BZZ-100
quasi-static simulation, where the slip distance does not
standard axial load. The simulation process is divided into
affect the final static stress distribution of pavement. Thus
the tire only needs to be controlled to slip in a reasonable
range. In this study, the tire is controlled to slip 2 mm to
the X direction.

3.3. Finite element simulation results

The first principal stress distribution of the pavement

texture is shown in Fig. 6(a), where the stress concentration
occurs in many locations, especially on the rough edges of
the coarse aggregate. The maximum stress in the red
labeled area is 392.7 MPa. Fig. 6(b) is the first principal
stress distribution from the perspective of XZ cross section
which is in the red labeled area.

3.4. The size effect of pavement aggregate

The simulation results show that a significant stress con-

Fig. 3. Image of acquired pavement area. centration could occur when the tire is in contact with the
 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184 179

Table 1 strength rm, bc is the strength attenuation coefficient of

Material parameters. rock.
Pavement Tire According to the compressive strength of standard
Elastic modulus (MPa) 9.0
104 basalt specimen is 67.7 MPa [26], from Formula (1), the
Poisson ratio 0.25 strength of basalt coarse aggregate which is affected by
The parameters of three-order Ogden model l1 6.3 the size effect rc is 125.8 MPa.
l2 0.012 GE [27] considered that there is a stress threshold of fati-
l3 0.1 gue failure of rock, only when the maximum stress of cyclic
a1 1.3
loading exceeds the threshold value, then the rock will
a2 5
a3 2 accumulate the fatigue damage. If the stress of the cyclic
loading is below the threshold value, the rock would not
be damaged, no matter how many cyclic loading times.
The size of fatigue threshold is various which is depending
road surface [23,24]. However, the size effect of rock on the types of rock, the value is usually 65–80% of the sta-
strength also needs to be considered. Liu [25] proposed tic strength of rock [28–30]. In this study, assuming that
an empirical formula as follows: fatigue threshold value is 70% of the static strength of
the aggregate, then the basalt aggregate will accumulate
rc ¼ rm þ ac expðbc DÞ ð1Þ
fatigue damage once the concentrated stress is more than
about 88 MPa.
where rc is the uniaxial compressive strength of rock spec- Take the pavement in the last chapter as an example, the
imen whose length or diameter of the section is D, rm is the maximum stress loaded on exposed pavement aggregate
test strength of the corresponding type of rock, ac is the dif- can be reached to 392.7 MPa based on the stress results
ference between the strength of original rock r0 and the test of pavement model simulated by ANSYS. It is far greater

Fig. 6. 1st principle stress distribution of pavement macro texture.

180 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184
than the threshold value of rock fatigue damage. Thus the
concentrated stress could produce a small range of direct
failure at the rough edges of the exposed pavement aggre-
gate, and then leads to change the macro texture of pave-
ment. While if the concentrated stress did not cause a
direct failure, but the stress is still greater than the fatigue
threshold of the rock, the fatigue damage would be accu-
mulated continuously inside the aggregate [31].
3.5. Fatigue life analysis of pavement macro texture
The macro texture of asphalt pavement mainly consists
of aggregate particles. There was very limited research on
fatigue failure of aggregate rock at present. The influence
factors of the rock fatigue failure include the maximum
and minimum cyclic stress, the frequency of the cycle, the
cycle times and the loading wave, in which the main factors
are the maximum cyclic stress and stress amplitude [29,32].
Under real traffic, the stress on pavement is very compli-
cated, thus this study assumed the traffic load is based on
BZZ-100 standard axle load, when a vehicle passes over
the pavement, the standard axle load of this vehicle is
applied onto the pavement, the stress at this moment is
assumed as the maximum cyclic stress, once the vehicle
leaves, the pavement is no longer under loaded, which
means the lower limit of cyclic stress on pavement is 0.
Therefore, the stress conditions of pavement are assumed
as a constant amplitude cyclic loading, and the loading
sequence does not affect the fatigue life of pavement
texture.
The density of the hysteresis loop curve of the typical
rock fatigue failure test under the effects of constant ampli-
tude cyclic loading is basically unchanged [33]. Thus, the
accumulation tendency of fatigue damage of rock can be
approximately considered as linear increasing. According
to the Miner damage model in linear fatigue damage the-
ory, the fatigue damage of coarse aggregate rock can be
calculated by stress amplitude and a number of repetition
curves (hereinafter referred to as S–N curve) [30].
The linear Miner fatigue cumulative damage model con-
sidered that the fatigue damage degree is calculated by the
following equation:
X
n
ni
D¼ ¼1
i¼1
Ni
where N i is the fatigue life under current stress S i , ni is the
number of cyclic loading times under current stress S i , n is
the current cyclic loading times.
In this study, FE-SAFE software is used to analyze the
fatigue life of the exposed coarse aggregate of asphalt pave-
ment, according to the Formula (2). Since the material
database in FE-SAFE is mainly metal material, so it is
needed to define a kind of rock material including the S–
N curve in FE-SAFE.
Zhou [34] used high strength white cement specimen to
simulate rock material, the fatigue test data of the non-
cracked rock under different cyclic stresses are obtained
based on the high-frequency uniaxial compression fatigue
test. According to the fatigue test data, the S–N curve of
the rock material as shown in Fig. 7.
Input the stress results of ANSYS into FE-SAFE, the
fatigue life of aggregate is calculated based on the S–N
curve of rock material. As shown in Fig. 8, the fatigue life
distribution is coincident with the results of stress, the fati-
gue life is displayed in base-10 logarithm, in which the
pavement area of fatigue life over 109 loading times is the
infinite area, means that the pavement texture would not
be affected by fatigue failure.
While the fatigue failure occurs on pavement aggregate,
which will lead to the changes of pavement macro texture,
the contact condition between tire and pavement, as shown
in Fig. 1, will also be changed. As shown in Fig. 9, using a
sine wave to represent these two kinds of macro texture
model, the length of these two model are the same, while
the wavelength of the sine waves are various. Obviously,
the first model is much sharper than the latter one. Assum-
ing the former model is the initial state of pavement macro
texture, and the latter model is the macro texture which has
been affected by aggregate fatigue failure. Once the mor-
phology of macro texture is changed, the contact condition
will also be changed, which leads to the changes of stress
distribution on pavement texture. ANSYS is used to simu-
late the stress on the pavement with different texture mor-
phologies and to compare the changes of stress on
pavement texture before and after the failure. The simula-
tion boundary conditions are the same with last chapter.
As shown in Fig. 9(a), the maximum first principal stress
is reached to 6.17 MPa and reduced to 5.29 MPa in
Fig. 9(b). Results show that the changes in the morphology
of macro texture will lead to the changes of stress distribu-
tion on pavement. It means that the stress which concen-
trated on macro texture could be gradually reduced with
the continued failure of aggregate, till the concentrated
stress is finally reduced to less than fatigue threshold stress
ð2Þ
Fig. 7. S–N curve of non-cracked rock.
 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184 181

Fig. 8. Fatigue life distribution of pavement.

of aggregate, then the pavement macro texture will be no

longer affected by fatigue damage.
In addition, in order to analyze the characteristics of
fatigue failure distribution of pavement macro texture, this
study extract the elements which is affected by fatigue fail-
ure, from the peak of the pavement texture layer by layer
with the distance of 0.3 mm, then calculate the average fati-
gue life of the elements in each layer, so the developing ten-
dency of average fatigue life with the depth of macro
texture is shown in Fig. 10. It can be found in Fig. 10 that
the fatigue life of pavement texture is firstly decreased and
then increased with depth, the shortest fatigue life of pave-
ment layer is located at about 1.7 mm from a peak of the
texture. It shows that the fatigue damage will firstly occur
at the side of aggregate edges. The reason is also shown in
Fig. 6(b) and Fig. 9, the maximum stress concentration
generally occurs at the side of the aggregate, which is
caused by tire slip. Therefore a larger stress concentration
would cause fatigue failure of aggregate firstly.
Fig. 9. Two kinds of macro texture model represented with sine wave.
4. Results analysis

4.1. Generalized macro texture evaluation method

In order to find the developing tendency of fatigue life of

coarse aggregate with loading times, however, the tradi-
tional sand patch method cannot be used in the 3D pave-
ment model. Thus an evaluation method of pavement
macro texture under different loading times is proposed.
A new evaluation parameter MTD is ideal to evaluate
and describe the average depth of 3D pavement macro tex-
ture under different cyclic loading times.
According to the standard sand patch method, the vol-
ume of the micro hexahedral element was used to calculate
the volume of the entire pavement texture model. As long
as the element size is small enough, any morphology
changes of pavement macro texture can be regarded as
Fig. 10. The developing tendency of fatigue life with the depth.
the changes of several micro elements. MTD is the ratio
182 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184

of the model volume between the texture peak and bottom

point with the projected area of pavement model. Accord-
ing to the principle of MTD , the relationship between the
MTD (which is calculated by element volume) and the
MTD (which is measured by sand patch method) is approx-
imately linear on numerical.
The calculation of MTD is similar to the sand patch
method, the formula is as follow:
N V
MTD ¼
S
where N is the number of elements under current pavement
model, V is the average volume of the element, S is the pro-
jected area of the pavement model.
The lowest place of the texture is assumed as the bottom
plane, then, from the top to the bottom plane, eliminate the
element which exceeded its fatigue life in a sequence of
loading times, until all the elements which affected by fati-
gue failure are eliminated, the processed pavement model is
shown in Fig. 11.
To calculate the processed model in Fig. 11, according
to Formula (3), the developing tendency of MTD with
the number of the cyclic loads is shown in Fig. 12. It shows
that the decreasing rate of MTD is remarkably great when
the number of loading times is small. However, with the
increase of cyclic loading times, the rough edges of coarse
aggregate becomes smooth, then leads to the gradual
reduction of the concentrated stress. The curve is then turn-
ing to be obviously gentle at about 500 thousand loading
times, which means that when it in a high number of load-
ing times, the pavement macro texture is relatively smooth
changed, the effect of fatigue failure on the pavement
macro texture is reduced either.
4.2. Compared with accelerated loading test
Sun [35] concluded the decreasing tendency of MTD
with loading times based on accelerated loading test on
asphalt pavement sample, in which the coarse aggregate
ð3Þ
Fig. 12. The developing tendency of MTD with the number of cyclic
loading times.
is basalt and gradation are also SMA-13, this is the same
as the pavement model in this paper. In order to compare
simulation results and test results, considering that the
numerical error caused by different calculation methods
need to be excluded. Therefore, this study proposed wear
rate x to represent the change rate of pavement texture
depth under different loading times, which is the ratio of
MTD (or MTD ) with loading times.
According to this method, till about 1 million 600 thou-
sand loading times, the wear rate of tested pavement is cal-
culated as x1 , and the wear rate of pavement model is
calculated as x2 . Thus the developing tendency of the wear
rate x1 and wear rate x2 with the axle loading times is
shown in Fig. 13. As can be seen in the figure, these two
curves are consistent with the tendency of loading times,
the wear rate under high axle loading times are both grad-
ually tending to 0. While the wear rate x2 is greater than
x1 , because of the value of MTD is also greater than the
MTD, therefore, this study only considers the developing
tendency of wear rate here.

Fig. 13. Comparison of the test pavement wear rate x1 and the simulated
Fig. 11. The pavement model which is eliminated the failure elements. pavement wear rate x2 .
 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184
In this study, the fatigue life of the coarse aggregate is
calculated based on the S–N curve of white cement speci-
men, while the maximum cyclic stress of cement specimens
in fatigue test is much smaller than the fatigue threshold
stress of basalt aggregate. So compared the pavement
model with the actual situation, the decrease rate of texture
depth is much greater. However, the tendency of the MTD
curve and the wear rate curve can still reflect the effect of
fatigue failure on pavement macro texture.
5. Conclusions
 When the aggregate rock is affected by external cyclic
loading, it cause internal fatigue damages. With the
increase of cyclic times, the damage would be continu-
ously accumulated inside the aggregate and gradually
develops into micro-cracks, eventually lead to fatigue
failure of aggregate.
 Take the pavement which the coarse aggregate is basalt
as an example, using a 3D laser scanner to reconstruct
the pavement model, then to simulate the contact stress
between tire-pavement by using FEM. The results show
that the pavement stress is mostly concentrated on the
rough edges of coarse aggregate, and the maximum
stress is theoretically more than enough to cause the fati-
gue failure of aggregates.
 Considering the size effect of rock materials. The fatigue
life of exposed aggregate is calculated by fatigue analysis
software FE-SAFE based on linear cumulative damage
theory. Results show that the fatigue failure firstly
occurs at the edges of aggregate instead of the peaks.
 The wear rate of the simulation results is compared with
the accelerated loading test results, it can be concluded
that when the number of loading times is small, the
effect of fatigue failure on pavement macro texture is sig-
nificant, while the effect is obviously decreased after
more than about 500 thousand loading times.
 During the calculation of the fatigue life of the aggregate
rock, the S–N curve data which is based on white
cement specimen can only be used for approximate cal-
culation. The S–N curve of the small size aggregate rock
material has to be measured in future experiment so that
the fatigue life of pavement macro texture can be calcu-
lated more accurately.
Acknowledgements
This paper is supported by the National Natural Science
Foundation of China. Grant Nos. 51308042, 41372320,
51178013, and Fundamental Research Funds for the Cen-
tral Universities (06500036).
References
[1] J.J. Henry, Evaluation of Pavement Friction of Characteristics,
Transportation Research Board, National Research Council, 2000.
183
[2] Y. Li, D. Wang, L. Wang, et al., Effect of film thickness of asphalt
mixture in compression test, in: Multi-Scale Modeling and Charac-
terization of Infrastructure Materials, Springer, The Netherlands,
2013, pp. 199–218.
[3] ASTM International, E867-06 Standard Terminology Relating to
Vehicle-Pavement Systems, 2012.
[4] Y. Hou, M. Guo, Z. Ge, L. Wang, W. Sun, Mixed-mode I–II
cracking characterization of mortar using phase-field method, J. Eng.
Mech. 143 (7) (2017) 04017033.
[5] Y. Hou, L. Wang, D. Wang, P. Liu, M. Guo, J. Yu, Characterization
of bitumen micro-mechanical behaviors using AFM, phase dynamics
theory and MD simulation, Materials 10 (2) (2017) 208.
[6] Y. Hou, Y. Huang, F. Sun, M. Guo, Fractal analysis on asphalt
mixture using a two-dimensional imaging technique, Adv. Mater. Sci.
Eng. 2016 (2016) 7 (Article ID 8931295).
[7] P. Rachakatla, Application of Relational Database Principles for
Rating Bituminous Coarse Aggregates with Respect to Frictional
Performance, Texas Tech University, 2000.
[8] P. Cairney, P. Bennett, Relationship between road surface character-
istics and crashes on Victorian rural roads, in: 23rd ARRB Confer-
ence, Adelaide, South Australia, Australia, 2008.
[9] R. Laganier, J. Lucas, The influence of pavement evenness and
macrotexture on fuel consumption, Surface Characteristics of Road-
ways: International Research and Technologies, ASTM Interna-
tional, 1990.
[10] B.W. Sullivan, Development of a fundamental skid resistance asphalt
mix design procedure, in: Proceedings, International Conference on
Surface Friction, Christchurch, New Zealand, 2005.
[11] D. Wang, X. Chen, C. Yin, M. Oeser, B. Steinauer, Influence of
different polishing conditions on the skid resistance development of
asphalt surface, Wear 308 (1–2) (2013) 71–78.
[12] D. Wang, X. Chen, M. Oeser, H. Stanjek, B. Steinauer, Study of
micro-texture and skid resistance change of granite slabs during the
polishing with the Aachen Polishing Machine, Wear 318 (1–2) (2014)
1–11.
[13] D. Wang, A. Ueckermann, A. Schacht, M. Oeser, B. Steinauer, B.N.
J. Persson, Tire–road contact stiffness, Tribol. Lett. 56 (2) (2014) 397–
402.
[14] G. Meschke, H.J. Payer, H.A. Mang, 3D simulations of automobile
tires: material modeling, mesh generation, and solution strategies,
Tire Sci. Technol. 25 (3) (1997) 154–176.
[15] H.N. Xu, Y.Q. Tan, X.M. Li, G.L. Tian, Research on relationship
between marshall stability and grading curve trait of asphalt mixture,
Highway (2008) 5902–5905.
[16] J.R. Ellis, Vehicle Handling Dynamics, Mechanical Engineering
Publications, London, United Kingdom, 1994.
[17] H.N. Xu, W. Guo, Y.Q. Tan, Internal structure evolution of asphalt
mixtures during freeze-thaw cycles, Mater. Des. 86 (12) (2015) 436–446.
[18] H. Xu, G. Wei, Y. Tan, Permeability of asphalt mixtures exposed to
freeze–thaw cycles, Cold Reg. Sci. Technol. 123 (2015) 99–106.
[19] H. Xu, Y. Tan, X. Yao, X-ray computed tomography in hydraulics of
asphalt mixtures: procedure, accuracy, and application, Constr.
Build. Mater. 108 (2016) 10–21.
[20] J.A. Hudson, J.P. Harrison, Engineering Rock Mechanics, Pergamon
Press, 1997.
[21] Y. Miao, P. Song, X. Gong, Fractal and multifractal characteristics
of 3D asphalt pavement macrotexture, J. Mater. Civ. Eng. 26 (8)
(2014) 04014033.
[22] G. Berselli, R. Vertechy, M. Pellicciari, G. Vassura, Hyperelastic
modeling of rubber-like photopolymers for additive manufacturing
processes, Rapid Prototyping Technology – Principles and Functional
Requirements, InTech, 2011.
[23] Y. Hou, W. Sun, L. Wang, Y. Huang, M. Guo, A multi-scale
approach of mode I crack in ettringite, Road Mater. Pavement Des.
18 (S3) (2017) 33–42.
[24] Z. Ge, Y. Wang, Y. Hou, W. Sun, R. Sun, Evaluation of fracture in
mortar subject to tension loading using phase field model and three
point bending test, Mater. Des. 86 (2015) 121–128.
184 Z. Qian et al. / International Journal of Pavement Research and Technology 11 (2018) 176–184
[25] B.C. Liu, J.S. Zhang, A study of size effect for compression strength
of rock, Chin. J. Rock Mech. Eng. 1998 (6) (1998) 611–614 (in
Chinese).
[26] S.W. Zhou, C.C. Xia, Y.S. Hu, Y. Zhou, P.Y. Zhang, Damage
modeling of basaltic rock subjected to cyclic temperature and uniaxial
stress, Int. J. Rock Mech. Min. Sci. 77 (2015) 163–173.
[27] X.R. Ge, Y.F. Lu, Discussion on the problem of fatigue failure and
irreversible deformation of rock under cyclic loading, Chin. J.
Geotech. Eng. 1992 (3) (1992) 56–60 (in Chinese).
[28] X.R. Ge, Deformation control law of rock fatigue failure, real-time
X-ray CT scan of geotechnical testing, and new method of stability
analysis of slopes and dam foundations, Chin. J. Geotech. Eng. 30 (1)
(2008) 1–20 (in Chinese).
[29] M.N. Bagde, V. Petroš, Fatigue and dynamic energy behavior of rock
subjected to cyclical loading, Int. J. Rock Mech. Min. Sci. 46 (1)
(2009) 200–209.
[30] H. Pathak, A. Singh, I.V. Singh, S.K. Yadav, Fatigue crack growth
simulations of 3-D linear elastic cracks under thermal load by XFEM,
Front. Struct. Civ. Eng. 9 (4) (2015) 359–382.
[31] S. Kumar, A.S. Shedbale, I.V. Singh, B.K. Mishra, Elasto-plastic
fatigue crack growth analysis of plane problems in the presence of
flaws using XFEM, Front. Struct. Civ. Eng. 9 (4) (2015) 420–440.
[32] H.R. Nejati, A. Ghazvinian, Brittleness effect on rock fatigue damage
evolution, Rock Mech. Rock Eng. 47 (5) (2014) 1839–1848.
[33] J.Q. Xiao, D.X. Ding, F.L. Jiang, G. Xu, Fatigue damage variable
and evolution of rock subjected to cyclic loading, Int. J. Rock Mech.
Min. Sci. 47 (3) (2010) 461–468.
[34] Y.C. Zhou, Y.L. Chen, W. Shao, The high-frequency fatigue
characteristics of rock with cracks, Eng. Mech. 28 (7) (2011) 98 (in
Chinese).
[35] H.L. Sun, Attenuation of asphalt pavement anti-slid performance,
Highway 7 (7) (2011) 73 (in Chinese).