Engineering Failure Analysis 58 (2015) 238248

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/efa

Analysis of fatigue properties of unmodied and polyethylene

terephthalate modied asphalt mixtures using response surface
methodology
Mehrtash Soltani a,, Taher Baghaee Moghaddam a,b, Mohamed Rehan Karim a, Hassan Baaj b
a

Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia
Centre for Pavement and Transportation Technology, Department of Civil and Environmental Engineering, Faculty of Engineering, University of Waterloo, Waterloo N2L 3G1,
Canada
b

a r t i c l e

i n f o

Article history:
Received 18 May 2015
Received in revised form 10 August 2015
Accepted 10 September 2015
Available online 12 September 2015
Keywords:
Asphalt mixture
Waste PET
Environmental temperature
Response surface methodology

a b s t r a c t
Fatigue is a major distress mode of exible pavements that generally occurs in the form of irregular (alligator) cracking in the wheel paths. This paper evaluates the effects of applied stress and
temperature on the fatigue lives of polyethylene terephthalate (PET) modied asphalt mixtures
using response surface methodology (RSM). As it is shown in this study a quadratic model is
successfully tted to the experimental data. Fatigue lives of mixtures are inuenced by changes
in selected parameters. In addition, the effect of temperature variation is more drastic on the
fatigue lives than the effects of stress level and modier content.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Road pavement is subjected to external loads including mechanical loading induced by heavy trafc and thermal loading induced
by thermal changes. The applied loads, along with environmental conditions result in pavement deterioration which, in some cases,
happens even before its expected service life. Pavement damage is usually occurred in the form of permanent deformation (surface
rutting), fatigue failure and low temperature cracking. Fatigue failure is a common mode of distress of pavement structures which
is caused by successive tensile strain induced by repeated trafc loadings [1]. This form of distress mostly appears as cracking damage
which initially occurs at the bottom of asphalt layer where the tensile stresses are maximum. Then these cracks spread to the surface
of asphalt mixture. Previous studies showed the fatigue life of asphalt mixture has correlation with the mode and amount of applied
loads as well as environmental temperature [2,3].
Stone mastic asphalt (SMA) is gap-graded asphalt mixture which has been developed in Germany in 1960s [4]. It has a high
percentage (60 to 80%) of coarse aggregate, greater than 5 mm in size, high binder content (5.5 to 7% by weight), high percentage
of mineral ller (7 to 11%), and added bers (1%) [5]. Due to inherited structure of SMA, it can provide better permanent deformation
(rutting) performance and durability compared to conventional dense-graded mixture [6,7] but it becomes controversial in case of
fatigue property. However some studies showed that SMA mixture had lower fatigue life [8,9], others concluded that it had better fatigue properties compared to the conventional mixture [10,11]. In SMA mixture in order to prevent draindown (due to high asphalt
content) and improve mixture performance stabilizer additives, bers and polymers are used. In this case, using polymer in asphalt

Corresponding author.
E-mail address: mehrtashsoltani@gmail.com (M. Soltani).

http://dx.doi.org/10.1016/j.engfailanal.2015.09.005
1350-6307/ 2015 Elsevier Ltd. All rights reserved.

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

239

mixture is very common [1215]. Tapkn has utilized polypropylene bers as reinforcement in asphalt mixture and it was realized that
the mixture fabricated by polypropylene bers had better performance in comparison with control mixture [16].
In many cases, using polymers causes higher construction cost due to high polymer cost. In order to overcome this problem, many
studies have used waste polymers in asphalt mixtures [13,1720]. One of the important industrial plastic materials is polyethylene
terephthalate (PET). PET is a semi-crystalline thermoplastic polymer material which has been used in beverage and food industries
for years. Currently, a large amount of waste PET is being produced worldwide and it is going to cause a serious environmental
challenge due to its non-biodegradability [21]. Hence, some studies have been previously performed to evaluate the effects of using
post-consumer PET as secondary materials in road pavement in order to tackle this potential environmental hazard and, moreover,
to decrease construction cost imposed by application of polymers in asphalt mixture [2,13,22,23].
Mathematical modeling is useful for real-world application as it is robust in terms of its ability to deal with many constraints and
objectives [24,25]. In addition, using statistical analysis in pavement engineering is increasing among engineers and designers because
it helps to have better perspective about the pavement performance parameters. In this case, factorial design of experiments (DOE)
which through the use of techniques such as response surface methodology (RSM) simultaneously consider several factors at different
levels, and give a suitable model for the relationship between the various factors [2630].
The aim of this study is examining the fatigue property of SMA mixtures at elevated temperatures and stress levels for the
unmodied and PET modied mixtures followed by nding interactions between these fundamental factors using RSM based
on central composite design (CCD).

2. Materials and methods

SMA mixtures were fabricated using 80/100 penetration grade asphalt cement. Granite-rich aggregate particles were used for this
investigation. Nine percent of ller was utilized. The aggregate particle size distribution is shown in Fig. 1. As it is shown in this gure,
the SMA mixture contains coarser aggregate particle (68.5% of particles are greater than 4.75 mm) which provides better stone on
stone contact. In order to have better understanding of the materials' characteristics several tests were performed on asphalt cement
and aggregate particles and the results are listed in Table 1. As can be seen in Table 1, materials' properties are satisfactorily passed the
requirements.
PET akes which have been used for this study were obtained from waste PET bottles. For using PET akes in asphalt mixture, the
PET bottles were cut to small parts and crushed using a crushing machine. Thereafter, the crushed PET particles were sieved and
particles smaller than 2.36 mm in size were used for this research. Table 2 depicts physical and mechanical properties of PET.

2.1. Mixture fabrication

In order to fabricate SMA mixtures, 1100 g of mixed aggregate and ller were heated inside oven at temperature of 160 C for 3 h.
Asphalt cement was also heated at 130 C to be suitable for mixing with aggregate particles. Mixing temperature of 160 C was determined using plot of binder viscosity against temperature (viscosity at mixing temperature must be 0.17 0.02 Pa s). PET particles
with different percentages (0%, 0.5% and 1% by weight of aggregate particles) were added directly to the mixture as the method of
dry process. It is worth mentioning that in previous research it was believed that due to the high melting point of PET wet process
(adding modier to the asphalt cement) cannot be appropriate because it might hinder the mixing [17]. The loose mixture was
compacted using Marshall compactor and 50 blows of compaction effort were applied on each side of the mixture. It should be mentioned that all the mixtures were fabricated at their optimum asphalt contents using Marshall mix design method [22,31,32] and the
results are presented in Table 3.

Fig. 1. Aggregate particle size distribution for stone mastic asphalt.

240

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

Table 1
Properties of materials.
Property

Unit

Used specication

Value

Requirements

Asphalt
Penetration at 25 C
Softening point
Flash point
Fire point
Specic gravity

0.1 mm
C
C
C
(g/cm3)

ASTM D 5
ASTM D 36
ASTM D 92
ASTM D 92
ASTM D 70

87.5
46.6
300
320
1.03

Coarse aggregate
L.A. abrasion
Flakiness index
Elongation index
Aggregate crushing value
Bulk specic gravity
Absorption

%
%
%
%
(g/cm3)
%

ASTM C 131
BS 812 Part 105.1
BS 812 Part 105.2
BS 812 part 3
ASTM C 127
ASTM C 127

19.45
2.72
11.26
19.10
2.60
0.72

b30
b20
b20
b30

b2

Fine aggregate
Bulk specic gravity
Absorption
Soundness loss

(g/cm3)
%
%

ASTM C 128
ASTM C 128
ASTM C 88

2.63
0.4
4.1

b2
b15

2.2. Indirect tensile fatigue test

Indirect Tensile Fatigue Test (ITFT) was carried out in the controlled stress mode according to BS EN 12697-24. Universal Testing
Machine (UTM) which is a computer controlled system was used for running ITFT. Compressive cyclic load was applied along with
diametrical section of specimen in the form of haversine waveform with 500 ms repetition time and 100 ms pulse width
(see Fig. 2). ITFT was conducted at stress levels of 200 kPa, 300 kPa, and 400 kPa which are the stress values mostly used in pavement
laboratories. In addition, temperatures of 10 C, 25 C and 40 C are designated in this study to simulate the pavement temperature
ranges that fatigue damage usually occurs. Prior to the test, all the specimens were conditioned at the controlled temperature chamber
for about 2 h to reach the desired test temperature. Fatigue life was dened as the number of load repetitions reached when the
specimen splits [2,3335].

2.3. Method of analysis

One-factor-at-a-time (OFAT) methodology is a conventional approach for optimizing multifactor experiments. OFAT is a changeable single factor method for a specic experiment design while other factors are kept constant. OFAT is unable to generate appropriate output because the effects of interaction among all factors in the design are not examined truly, and so it is not capable of reaching
the true optimum value [36,37]. Hence, response surface methodology (RSM) has been introduced for parameter optimization in a
way that number of experiments and interaction among the parameters are reduced to minimal value [3840]. Consequently, Design
Expert 9.0.5.1 was designated for this purpose to generate statistical analysis and experimental designs and to calculate the sorbent
adaption conditions.
For this study, a developed quadratic model using RSM was suggested by the software for design and data analysis. In this investigation, the effects of three independent numerical variables including PET modier (A) from zero to 1%, stress levels (B) from
200 kPa to 400 kPa and temperatures between 10 C and 40 C, all at three levels, were studied through the central composite design
(CCD). Related literature and preliminary studies were used to choose these variables and the respective regions of interest [2,3,33].

Table 2
Physical and mechanical properties of PET.
Property

Unit

Method

Value

Water absorption
Tensile strength
Tensile modulus
Elongation at break
Flexural strength
Flexural modulus
Approx glass transition temperature
Approx melting temperature
Specic gravity

%
psi
psi
%
psi
psi
C
C
g/cm3

ASTM D570
ASTM D638
ASTM D638
ASTM D638
ASTM D790
ASTM D790

ASTM D792

0.11
11,500
4 105
70
15,000
4 105
75
250
1.35

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

241

Table 3
Summary of mix design.
PET(%)

BSGa

VMAb (%)

VFAc (%)

OACd (%)

0 (unmodied)
0.5
1

2.294
2.296
2.283

18.12
17.34
17.55

77.92
76.90
77.20

6.77
6.36
6.51

a
b
c
d

Bulk specic gravity of compacted mixture.

Void in mineral aggregate.
Void lled with asphalt.
Optimum asphalt content.

Table 4 shows the levels and range of the actual values of independent numerical variables. By using Eq. (1) all dened numerical
variables transformed to the coded form.
xi

X i X 0
X

where xi describes the coded value of the ith independent factor which is dimensionless. Actual value is dened as Xi, X0 is the
center point actual value and X refers to the step change of the ith variable.
A total of 34 experiments in random order were performed, together with ve replications at the center points to provide accurate
assessment of errors (Table 4). The fatigue life was dened as the response to develop design of experiment modeling.
(Eq. (2)) was developed to calculate the dependent variables [41,42]:
Xn
Xn
Xn Xn
2
Y0 i1 i xi i1 ii xi i1
x x
j1 ij i j

In Eq. (2), Y is the calculated response, 0 is constant value. Independent variables in coded form are described as xi, and xj. The
coefcients of i and ii are the linear and quadratic terms. ij is the interaction term coefcient, is the random error, and the studied
number of factors is dened as n.
In addition, in order to assess the appropriateness of proposed model, analysis of variance (ANOVA) was performed. The coefcients
of determination (R2 and R2adj) express the wellness of the t to the suggested model. These values can be determined using the
following equations [43]:
2

R 1

SSresidual
SSmodel SSresidual
SSresidual

Radj 1SS

model SSresidual

. D F residual

D F model D F residual

Fig. 2. Indirect Tensile Test loading set-up.

242

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

Table 4
Experimental design layout and experimental results of the responses.
Run

Factor 1: PET (%)

Factor 2: Stress level (kPa)

Factor 3: Temperature (C)

Fatigue cycles

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

0
1
0.5
0
1
0.5
0.5
0.5
0
0.5
1
0
1
0
0
0.5
0.5
1
0.5
0
1
1
0.5
0.5
0
0.5
1
0.5
1
0
0.5
1
0.5
0

200
400
300
200
300
300
200
400
300
300
400
300
200
400
200
300
300
400
300
400
200
300
300
300
400
300
400
400
200
400
200
200
300
200

10
40
10
40
25
25
25
25
25
25
10
25
10
10
40
25
25
40
25
40
10
25
40
40
40
10
10
25
40
10
25
40
25
10

196,720
1541
184,521
4341
7752
6072
51,421
1243
3019
6063
167,281
3033
385,866
91,291
4329
6061
6059
1544
6088
866
379,731
7739
3133
3141
878
179,491
167,312
1239
7829
91,302
53,320
7931
6077
188,821

In this equation, SS is the sum of squares and DF is degrees of freedom.

Eq. (5), Eq. (6) and an F-test in the program were used to check the model's adequate precision ratio (AP) to determine the
statistical importance of the model [44]:
Adequate Precision

V Y

maxY minY
p
V Y

1 Xn
p 2
V

i1
n
n

where Y is the predicted response, P represents the number of model parameters, residual mean square is described as 2, and n
is the number of experiments.
After the F-test had been performed, the insignicant terms were found and eliminated from the model. Thereafter, the nalized
model was introduced based on the signicant variables.

Fig. 3. Fracture patterns (left: ideal fracture, right: single cleft fracture).

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

243

3. Results and discussion

As it was mentioned by Allen at al., three different modes of failure have been observed for the ITFT. In the rst mode which has
been observed in most cases the specimen fractured completely; however, in the second mode specimen did not fail due to the fracture and no visible crack was observed. In such case, the failure was attributed to accumulation of permanent vertical deformation.
Additionally, the third mode of failure was dened by indentation of the loading strip into the specimen [34]. In this study, ITFT
was carried out in the controlled stress mode. The ITFT test was conducted on the mixtures at elevated temperatures and stress levels,
and the fracture patterns are shown in Fig. 3. As it can be seen in this gure two types of fracture are observed known as ideal fracture
and single cleft fracture [35]. Table 4 represents the layout for experimental design and the fatigue lives responses. According to this
table the fatigue lives vary between 866 and 385,866 cycles. Having these values, RSM was utilized to nd interactions between the
outputs and variables which are independent. Eventually, after a regression analysis had been applied to all responses described in the
design matrix, a tted quadratic polynomial equation was produced. The highest order polynomials in which the additional terms
were signicant and the models were not aliased were suggested by software. The numerical parameters (A, B and C) were used to
generate the predictive model according to Eq. (7):
2

Final Log10 fatigue equation 3:8 0:15A0:37B0:92C 0:57C :

Checking the adequacy of the model is an important part of the data analysis, as the model functions would give improper responses in case the t is not adequate [39,45]. Hence, in this study, in order to assess the signicance and adequacy of the model,
ANOVA analysis was performed and the results are reported in Table 5. In addition, this table shows the quadratic models for
coded factors, and represents other statistical parameters in logarithmic scale for the fatigue life. In this table, p-values which are
less than 0.0001 imply that the model and parameter are signicant (model and term with p-value b0.05 indicate the model and
the term are signicant for 95% condence intervals) for assessing the value of responses [46].
As the results show, PET (A), stress level (B), temperature (C), C2 are signicant terms with p-values less than 0.05. However, A2,
B2, AB, AC and BC were insignicant (p-value N 0.100). Therefore, in order to improve the model and optimize the results, the insignicant term can be removed from the model [47].
In order to check the tness of model regression coefcients, R2 and R2adj were calculated. Values of 0.9579 and 0.9422 were obtained for R2 and R2adj, respectively. This shows that 94.22% of the total variation in the fatigue life response could be explained by the
quadratic model. The high R2 and adjusted R2 values indicate that there is a good agreement between predicted and actual values
[40,41,48]. Ratio of signal-to-noise is measured by adequate precision to compare the variety of the estimated amounts at the design
points with the average prediction error. Adequate model discrimination was found in this study when the adequate precision ratio of
25.936 was calculated which is much higher than the value of 4 [49]. The lack of t (LOF) F-test was also used to evaluate the adequacy
of the model. LOF depicts the variation of the data around the tted model, and the amount of LOF would be signicant if the model
does not t the data well. It is worth noting that despite the LOF being signicant, a reasonable agreement between the predicted and
adjusted R2 were found for all responses and it can be concluded that the models suggested for all responses can be used to navigate
into design space to nd an optimum condition [50,51].
3.1. Statistical analysis
In order to have better understating about model satisfactoriness, diagnostic plots such as the predicted versus actual values are
worthwhile. Fig. 4 shows the actual versus predicted values of parameters for fatigue modeling. As shown in this gure there is an
adequate agreement between the actual data which were obtained through experiment and the predicted ones. This agreement

Table 5
ANOVA analysis for fatigue life.
Source

Sum of squares

df

Mean Square

Model
A PET
B Stress level
C Temperature
AB
AC
BC
A2
B2
C2
Residual
Lack of t
Pure error
Cor total
Adequate precision (AP)

22.74
0.44
2.76
16.76
5.472E004
7.234E004
0.13
0.082
0.051
1.74
1.00
1.00
4.205E004
23.73
25.936

9
1
1
1
1
1
1
1
1
1
24
5
19
33

2.53
0.44
2.76
16.76
5.472E004
7.234E004
0.13
0.082
0.051
1.74
0.042
0.20
2.213E005

F value
60.73
10.48
66.32
402.84
0.013
0.017
3.12
1.96
1.24
41.83
9017.24

p-value
Prob N F

Model performance

b0.0001
0.0035
b0.0001
b0.0001
0.9096
0.8962
0.0899
0.1739
0.2772
b0.0001

Signicant
Signicant
Signicant
Signicant
Insignicant
Insignicant
Insignicant
Insignicant
Insignicant
Signicant

b0.0001

Signicant

244

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

Fig. 4. Design-expert plot; predicted vs. actual values plot for fatigue life (logarithmic scale).

can also be understood by AP value (AP N 4) for the fatigue responses (see Table 5). This veries that the predicted model can be used
to navigate the design space dened by the CCD.
3.2. One factor analysis
One factor analysis is changing one factor at a time method. That is to say, in this method a single factor varies while all other
factors are kept constant for a particular set of experiments. This process exists for optimizing other variables which would be time
consuming. In this method, trial and error commonly exist for the optimization of variables, and, moreover, there is always a lack
of reaching a true optimum amount which is obtained by seeing the interaction among different variables [50,52]. Furthermore, in
one factor analysis when the software evaluates one parameter, other parameters are kept constant at their middle ranges. For instance, in case of PET content evaluation, temperature and stress level are kept constant at 25 C and 300 kPa respectively.
Figs. 5, 6 and 7 show the one factor analysis of PET percentage, stress level and temperature on logarithmic scale of fatigue life respectively. The logarithmic scale of fatigue life is shown for better underrating of difference between values. Fig. 5 indicates that by
increasing the PET the fatigue life is also increased. A possible reason for this result might be the mechanical properties of PET particles
in the mix. In fact, because the melting point of PET is high (over 250 C) and is higher than the mixture's fabrication temperature, the
PET particles do not melt during mixing. The solid PET particles can make mixture more elastic and cause higher fatigue life under
loading application. For another factor (Fig. 6) it can be observed that by increasing stress level the fatigue life is decreased. Same pattern is found for temperature when by increasing the temperature the fatigue life is decreased (Fig. 7). Fig. 7 also depicts that

Fig. 5. Effect of PET percentage on the fatigue life (logarithmic scale).

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

245

Fig. 6. Effect of different stress levels on the fatigue life (logarithmic scale).

increasing the temperature has negative effect on the fatigue life and that at higher temperatures (over 30 C) the fatigue life is
shifting to a constant value. This represents the importance of ambient temperature on the fatigue life of asphalt mixture. The ndings
of this study are based on controlled-stress test mode which are in support of previous studies [8,5356] that found the fatigue life of
asphalt mixtures increased at lower temperatures.
3.3. Effects of temperature, stress level and PET variables on the fatigue life using response surfaces
Three-dimensional response surface plots of the predictive quadratic model for the effect of stress level and temperature on
logarithmic scale is presented in Fig. 8. The response surfaces were generated based on Eq. (7).
Fig. 8 indicates at higher temperature and stress level the fatigue life is decreased. The variation of temperature for all stress level
seems to be signicant. In physical denition, when the ambient temperature increases, the asphalt binder becomes less stiff which
may weaken the fatigue resistance of asphalt mixtures and results in lower fatigue life. On the other hand the variation of stress levels
at higher temperatures is less effective on the fatigue life compared to lower temperature. That is to say, the changes in fatigue lives
are more tangible at lower stress levels and temperatures.
Fig. 9 indicates the effect of temperature and PET percentage on the SMA mixtures. Overall, increasing temperature had a negative
effect on the fatigue life. However, the effect of adding PET for improving the fatigue life is highlighted. Changes in fatigue life cannot
be attributed to the mixture air void content because all the mixtures were fabricated at their optimum asphalt contents with 4% air

Fig. 7. Effect of different temperature on the fatigue life (logarithmic scale).

246

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

Fig. 8. Effects of stress level and temperature on the fatigue life (logarithmic scale), 0.5% PET.

voids. In addition, improvement of fatigue life cannot be due to the higher asphalt content in the mixture because as it is shown in
Table 3 all the modied mixtures have lower asphalt contents than the unmodied mixture.
The correlation between stress level and PET content on the fatigue life of SMA mixture is shown in Fig. 10. Higher fatigue life is
found for the modied asphalt mixture associated with lower stress levels. By reducing the amount of PET in asphalt mixture the
fatigue life is decreased at all stress levels. In contrast, by increasing the stress level asphalt mixture experienced lower fatigue life
at all PET content. It can also be concluded that both PET increment and decrease in the stress level have roughly the same effect
on the fatigue life of asphalt mixture.

4. Conclusions
This paper aimed to evaluate the effect of applied load and temperature on the fatigue lives of unmodied and PET modied
asphalt mixture. Statistical analysis was used in this investigation to nd the interaction between selected variables. A good
agreement was found between predicted and actual values which indicated second-order response surface models provide a suitable
model to predict the fatigue life values within the range of dened factors. Based on the results achieved in this study the following
conclusions can be derived:
(1) The results showed that the changes in the fatigue lives are more tangible at lower stress levels and temperatures.
(2) Both PET increment and decrease in the stress level have roughly the same effect on the fatigue life of asphalt mixture.
(3) The effect of temperature on the fatigue lives is more drastic compared to stress level and PET content.

Fig. 9. Effects of PET percentage and temperature on the fatigue life (logarithmic scale), 300 kPa stress level.

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

247

Fig. 10. Effects of PET percentage and stress level on the fatigue life (logarithmic scale), 25 C.

Acknowledgements
The authors would like to thank to the University of Malaya Research Fund (Project No. RP010A-13SUS) for providing the
opportunity to make this research project.
References
[1] H. Di Benedetto, C. De La Roche, H. Baaj, A. Pronk, R. Lundstrm, Fatigue of bituminous mixtures, Mater. Struct. 37 (2003) 202216.
[2] T. Baghaee Moghaddam, M.R. Karim, T. Syammaun, Dynamic properties of stone mastic asphalt mixtures containing waste plastic bottles, Constr. Build. Mater. 34
(2012) 236242.
[3] G.G. Al-Khateeb, K.A. Ghuzlan, The combined effect of loading frequency, temperature, and stress level on the fatigue life of asphalt paving mixtures using the IDT
test configuration, Int. J. Fatigue 59 (2014) 254261.
[4] H. Baaj, D. Perraton, H. Di Benedetto, M. Paradis, Contribution l'etude de la relation entre le module complexe et la rsistance la fatigue et l'ornirage d'un
enrob SMA, Proceedings of the 48th Annual Conference of the CTAA, Halifax NS, November 1719, 2003.
[5] H. Baaj, M. Paradis, Use of post-fabrication asphalt shingles in stone matrix asphalt mix (SMA-10): laboratory characterization and field experiment on autoroute
20 (Quebec), Proceedings of the 53rd Annual Conference of the Canadian Technical Asphalt Association, Saskatoon SK, November 1619, 2008.
[6] R.B. Schmiedlin, Stone matrix asphalt: the Wisconsin experience, Trans. Res. Rec. 1616 (1988) 3441.
[7] I.M. Asi, Laboratory comparison study for the use of stone matrix asphalt in hot weather climates, J. Constr. Build Mater. 20 (2006) 982989.
[8] F.M. Nejad, E. Aflaki, M.A. Mohammadi, Fatigue behavior of SMA and HMA mixtures, J. Constr. Build Mater. 24 (2010) 11581165.
[9] T.B. Moghaddam, M.R. Karim, A. Mahrez, A review on fatigue and rutting performance of asphalt mixes, J. Sci. Res. Essays 6 (4) (2011) 670682.
[10] D. Perraton, H. Baaj, H. Di Benedetto, M. Paradis, Evaluation de la rsistance la fatigue des enrobs bitumineux fonde sur l'evolution de l'endommagement du
matriau en cours de l'essai : aspects fondamentaux et application l'enrob SMA, Can. J. Civ. Eng. 30 (2003) 902913.
[11] D. Perraton, H. Baaj, A. Carter, Comparison of some pavement design methods from a fatigue point of view. Effect of fatigue properties of asphalt materials, Road
Mater/ Pavement 11 (2010) 833861.
[12] A.I. Al-Hadidy, T. Yi-qiu, Mechanistic approach for polypropylene-modified flexible pavements, J. Mater. Des. 30 (2009) 11331140.
[13] E. Ahmadinia, M. Zargar, M.R. Karim, M. Abdelaziz, P. Shafigh, Using waste plastic bottles as additive for stone mastic asphalt, J. Mater. Des. 32 (2011) 48444849.
[14] C. Brovelli, M. Crispino, J. Pais, P. Pereira, Assessment of fatigue resistance of additivated asphalt concrete incorporating fibers and polymers, J. Mater. Civ. Eng. 2
(6) (2014) 554558.
[15] C. Brovelli, M. Crispino, J. Pais, P. Pereira, Using polymers to improve the rutting resistance of asphalt concrete, J. Constr. Build Mater. 77 (2015) 117123.
[16] S. Tapkin, The effect of polypropylene fibers on asphalt performance, Build. Environ. 43 (2008) 10651071.
[17] D. Casey, C. McNally, A. Gibney, M.D. Gilchrist, Development of a recycled polymer modified binder for use in stone mastic asphalt, J. Resour. Conserv. Recycl. 52
(2008) 11671174.
[18] S. Hnslolu, E. Aar, Use of waste high density polyethylene as bitumen modifier in asphalt concrete mix, J. Mater. Lett. 58 (2004) 267271.
[19] P.T.L. Liseane, Fontes, G. Trichs, J.C. Pais, A. Paulo, A. Pereira, Evaluating permanent deformation in asphalt rubber mixtures, J. Constr. Build Mater. 24 (2010)
11931200.
[20] M. Yu, G. Wu, Jinchuan Zhou and said easa, proposed compaction procedure for dry process crumb rubber modified asphalt mixtures using air void content and
expansion ratio, J. Test. Eval. 42 (2014) 111.
[21] M.N. Siddiqui, Conversion of hazardous plastic wastes into useful chemical products, J. Hazard. Mater. 167 (2009) 728735.
[22] T. Baghaee Moghaddam, M. Soltani, M.R. Karim, Evaluation of permanent deformation characteristics of unmodified and polyethylene terephthalate modified
asphalt mixtures using dynamic creep test, J. Mater. Des. 53 (2014) 317324.
[23] T. Baghaee Moghaddam, M. Soltani, M.R. Karim, Experimental characterization of rutting performance of polyethylene terephthalate modified asphalt mixtures
under static and dynamic loads, Constr. Build. Mater. 65 (2014) 487494.
[24] S. Kikuchi, N. Kronprasert, S. Easa, Aggregate blending using fuzzy optimization, J. Constr. Eng. Manage. ASCE 138 (2012) 14111420.
[25] M. Soltani, T.B. Moghaddam, M.R. Karim, S. Shamshirband, C. Sudheer, Stiffness performance of polyethylene terephthalate modified asphalt mixtures estimation
using support vector machine-firefly algorithm, Measurement 63 (2015) 232239.
[26] A. Khodaii, H.F. Haghshenas, H. Kazemi Tehrani, M. Khedmati, Application of response surface methodology to evaluate stone matrix asphalt stripping potential,
KSCE J. Civ. Eng. 17 (2013) 117121.
[27] A. Kavussi, M. Qorbani, A. Khodaii, H.F. Haghshenas, Moisture susceptibility of warm mix asphalt: a statistical analysis of the laboratory testing results, Constr.
Build. Mater. 52 (2014) 511517.
[28] A. Khodaii, H.F. Haghshenas, T.H. Kazemi, Effect of grading and lime content on HMA stripping using statistical methodology, Constr. Build. Mater. 34 (2012)
131135.

248

M. Soltani et al. / Engineering Failure Analysis 58 (2015) 238248

[29] T. Baghaee Moghaddam, M. Soltani, M.R. Karim, Stiffness modulus of polyethylene terephthalate modified asphalt mixture: a statistical analysis of the laboratory
testing results, J. Mater. Des. 68 (2015) 8896.
[30] T. Baghaee Moghaddam, M. Soltani, M.R. Karim, H. Baaj, Optimization of asphalt and modifier contents for polyethylene terephthalate modified asphalt mixtures
using response surface methodology, Measurement 74 (2015) 159169.
[31] A. Khodaii, A. Mehrara, Evaluation of permanent deformation of unmodified and SBS modified asphalt mixtures using dynamic creep test, Constr. Build. Mater. 23
(2009) 25862592.
[32] S.F. Kalyoncuoglu, M. Tigdemir, A model for dynamic creep evaluation of SBS modified HMA mixtures, Constr. Build. Mater. 25 (2011) 859866.
[33] J. Yan, F. Ni, M. Yang, J. Li, An experimental study on fatigue properties of emulsion and foam cold recycled mixes, Constr. Build. Mater. 24 (2010) 21512156.
[34] B. Allen, I. Artamendi, P. Phillips, Influence of temperature and aging on laboratory fatigue performance of asphalt mixtures, Advanced Testing and Characterization
of Bituminous Materials, Two Volume Set, Taylor & Francis group, London, 2009, ISBN 978-0-415-55854-9.
[35] A.M. Hartman, M.D. Gilchrist, G. Walsh, Effect of mixture compaction on indirect tensile stiffness and fatigue, J. Transp. Eng. 127 (2001) 370378.
[36] N.L. Frigon, D. Mathews, Practical Guide to Experimental Design, John Wiley and Sons, New York, 1997.
[37] D.C. Montogomery, Design and Analysis of Experiments, John Wiley and Sons, New York, 2005.
[38] A.I. Khuri, J.A. Cornell, Response Surfaces, Design and Analyses, 2nd ed. Marcel Dekker Inc., New York, 1996.
[39] R.H. Myer, D.C. Montogomery, Response Surface Methodology. Process and Product Optimization Using Designed Experiment. 2nd ed, John Wiley and Sons, New
York, 2002.
[40] R. Azargohar, A.K. Dalai, Production of activated carbon from Luscar char: experimental and modeling studies, Microporous Mesoporous Mater. 85 (2005)
219225.
[41] M.Y. Can, Y. Kaya, O.F. Algur, Response surface optimization of the removal of nickel from aqueous solution by cone biomass of Pinus sylvestris, Bioresour. Technol.
97 (2006) 17611765.
[42] Z. Aksu, F. Gnen, Binary biosorption of phenol and chromium (VI) onto immobilized activated sludge in a packed bed: prediction of kinetic parameters and
breakthrough curves, Sep. Purif. Technol. 49 (2006) 205216.
[43] B.K. Krbahti, M.A. Rauf, Response surface methodology (RSM) analysis of photoinduceddecoloration of toludine blue, Chem. Eng. J. 136 (2008) 2530.
[44] B.K. Krbahti, M.A. Rauf, Determination of optimum operating conditions of carmine decoloration by UV/H2O2 using response surface methodology, J. Hazard.
Mater. 161 (2009) 281286.
[45] B.K. Krbahti, M.A. Rauf, Application of response surface analysis to the photolytic degradation of basic red 2 dye, Chem. Eng. J. 138 (2008) 166171.
[46] M. Zabeti, W.M.A.W. Daud, M.K. Aroua, Optimization of the activity of CaO/Al2O3 catalyst for biodiesel production using response surface methodology, Appl.
Catal. A Gen. 366 (2009) 154159.
[47] V. Hosseinpour, M. Kazemeini, A. Mohammadrezaee, Optimisation of ru-promoted Ir catalysed methanol carbonylation utilising response surface methodology,
Appl. Catal. A Gen. 394 (2011) 166175.
[48] U.K. Garg, M.P. Kaur, V.K. Garg, D. Sud, Removal of nickel (II) from aqueous solution by adsorption on agricultural waste biomass using a response surface
methodological approach, Bioresour. Technol. 99 (2008) 13251331.
[49] T. lmez, The optimization of Cr (VI) reduction and removal by electrocoagulation using response surface methodology, J. Hazard. Mater. 162 (2009) 13711378.
[50] S.M.S. Wan, W.M.A. Daud, A. Houshmand, A. Arami-Niya, The application of response surface methodology to optimize the amination of activated carbon for the
preparation of carbon dioxide adsorbents, Fuel 94 (2012) 465472.
[51] J. Snchez-Romeu, J.M. Pas-Chanfrau, Y. Pestana-Vila, I. Lpez-Larraburo, Y. Masso-Rodrguez, M. Linares-Domnguez, et al., Statistical optimization of immunoaffinity purification of hepatitis B surface antigen using response surface methodology, Biochem. Eng. J. 38 (2008) 18.
[52] R.L. Mason, R.F. Gunst, J.L. Hess, Statistical Design and Analysis of Experiments, Eighth Applications to Engineering and Science, 2nd ed. Wiley, New York, 2003.
[53] J.A. Epps, C.L. Monismith, Fatigue of asphalt concrete mixtures-summary of existing information. Fatigue of compacted bituminous aggregate mixtures, ASTM
STP508, Am. Soc. Test. Mater. (1971) 1945.
[54] P.S. Pell, I.F. Taylor, Asphaltic road materials in fatigue, Proceedings of the association of the asphalt pavement technologists (AAPT), vol. 38 1969, pp. 577593
(Los Angeles, California, USA).
[55] G.G. Al-Khateeb, K.A. Ghuzlan, The combined effect of loading frequency, temperature, and stress level on the fatigue life of asphalt paving mixtures using the IDT
test configuration, Int. J. Fatigue 59 (2014) 254261.
[56] S. Bhattacharjee, R.B. Mallick, Effect of temperature on fatigue performance of hot mix asphalt tested under model mobile load simulator, Int. J. Pavement Eng. 13
(2012) 166180.