Analysis of fatigue properties of unmodified and polyethylene terephthalate modified asphalt mixtures using response surface methodology

    Analysis of fatigue properties of unmodified and polyethylene terephthalate modified asphalt mixtures using response surface methodology Mehrtash Soltani, Taher Baghaee Moghaddam, Mohamed Rehan Karim, Hassan Baaj PII: DOI: Reference:

S1350-6307(15)30093-5 doi: 10.1016/j.engfailanal.2015.09.005 EFA 2695

To appear in: Received date: Revised date: Accepted date:

18 May 2015 10 August 2015 10 September 2015

Please cite this article as: Soltani Mehrtash, Moghaddam Taher Baghaee, Karim Mohamed Rehan, Baaj Hassan, Analysis of fatigue properties of unmodified and polyethylene terephthalate modified asphalt mixtures using response surface methodology, (2015), doi: 10.1016/j.engfailanal.2015.09.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Title: Analysis of fatigue properties of unmodified and Polyethylene Terephthalate modified asphalt mixtures using Response Surface Methodology

Mehrtash Soltani a,*, Taher Baghaee Moghaddam

(First name: Taher, Last name: Baghaee

SC R

a

a,b

IP

T

Authors’ names:

Moghaddam), Mohamed Rehan Karim (First name: Mohamed Rehan, Last name: Karim), Hassan Baajb

NU

Authors’ affiliation addresses: a

MA

Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia. b

TE

D

Centre for Pavement and Transportation Technology, Department of Civil and Environmental Engineering, Faculty of Engineering, University of Waterloo, Waterloo N2L 3G1, Canada.

CE P

*Corresponding author: Mehrtash Soltani Tel: +601123806202; Fax: +60379552182

AC

Corresponding author E-mail address: [email protected]

Corresponding author postal address: Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia.

1

ACCEPTED MANUSCRIPT Analysis of fatigue properties of unmodified and Polyethylene Terephthalate modified asphalt mixtures using Response Surface Methodology

Center for Transportation Research, Department of Civil Engineering, Faculty of

IP

a

T

Mehrtash Soltania,*, Taher Baghaee Moghaddama,b, Mohamed Rehan Karima, Hassan Baajb

SC R

Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia b

Centre for Pavement and Transportation Technology, Department of Civil and

Environmental Engineering, Faculty of Engineering, University of Waterloo, Waterloo N2L

NU

3G1, Canada.

TE

D

MA

*Corresponding author: Mehrtash Soltani Tel: (+60) 1123806202; Fax: (+60) 379552182 E-mail address: [email protected] Present address: Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering, University of Malaya, 50603 Kuala Lumpur, Malaysia

Abstract

CE P

Fatigue is a major distress mode of flexible pavements that generally occurs in the form of irregular (alligator) cracking in the wheel paths. This paper evaluates the effects of applied

AC

stress and temperature on the fatigue lives of Polyethylene Terephthalate (PET) modified asphalt mixtures using Response Surface Methodology (RSM). As it is shown in this study a quadratic model is successfully fitted to the experimental data. Fatigue lives of mixtures are influenced 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 modifier content. Keywords: Asphalt mixture; Waste PET; Environmental temperature; Response surface methodology.

2

ACCEPTED MANUSCRIPT 1. Introduction Road pavement is subjected to external loads including mechanical loading induced by heavy

T

traffic and thermal loading induced by thermal changes. The applied loads, along with

IP

environmental conditions result in pavement deterioration which, in some cases, happens

SC R

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

NU

successive tensile strain induced by repeated traffic loadings [1]. This form of distress mostly

MA

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

TE

D

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

CE P

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 filler (7

AC

to 11%), and added fibres (1%) [5]. Due to inherited structure of SMA, it can provide better permanent deformation (rutting) performance and durability compared to conventional densegraded 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, fibers and polymers are used. In this case, using polymer in asphalt mixture is very common [12-15]. Tapkın has utilized polypropylene fibers as reinforcement

3

ACCEPTED MANUSCRIPT in asphalt mixture and it was realized that the mixture fabricated by polypropylene fibers had better performance in comparison with control mixture [16].

T

In many cases, using polymers causes higher construction cost due to high polymer cost. In

IP

order to overcome this problem, many studies have used waste polymers in asphalt mixtures

SC R

[13, 17-20]. One of the important industrial plastic materials is Polyethylene Terephthalate (PET). PET is a semi-crystalline thermoplastic polymer material which has been used in

NU

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-

MA

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

D

tackle this potential environmental hazard and, moreover, to decrease construction cost

TE

imposed by application of polymers in asphalt mixture [2, 13, 22, 23].

CE P

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

AC

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 [26-30]. Aim of this study is examining the fatigue property of SMA mixtures at elevated temperatures and stress levels for the unmodified and PET modified mixtures followed by finding interactions between these fundamental factors using RSM based on Central Composite Design (CCD).

4

ACCEPTED MANUSCRIPT 2. Materials and methods SMA mixtures were fabricated using 80/100 penetration grade asphalt cement. Granite-rich

T

aggregate particles were used for this investigation. 9% of filler was utilized. The aggregate

IP

particle size distribution is shown in Fig. 1. As it is shown in this figure, the SMA mixture

SC R

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

NU

results are listed in Table 1. As can be seen in Table 1, materials’ properties are satisfactorily

MA

passed the requirements.

PET flakes which have been used for this study were obtained from waste PET bottles. For using PET flakes in asphalt mixture, the PET bottles were cut to small parts and crushed

TE

D

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

CE P

mechanical properties of PET.

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

AC

Table 1: Properties of materials

Table 2: Physical and mechanical properties of PET

2.1 Mixture fabrication In order to fabricate SMA mixtures, 1100 g of mixed aggregate and filler were heated inside oven at temperature of 160˚C for 3 hours. 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 5

ACCEPTED MANUSCRIPT worth mentioning that in previous research it was believed that due to the high melting point of PET wet process (adding modifier to the asphalt cement) cannot be appropriate because it

T

might hinder the mixing [17]. The loose mixture was compacted using Marshall compactor

IP

and 50 blows of compaction effort were applied on each side of the mixture. It should be

SC R

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.

NU

Table 3: Summary of mix design

MA

2.2 Indirect tensile fatigue test

Indirect Tensile Fatigue Test (ITFT) was carried out in the controlled stress mode according

D

to BS EN 12697-24. Universal Testing Machine (UTM) which is a computer controlled

TE

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

CE P

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

AC

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 hours to reach the desired test temperature. Fatigue life was defined as the number of load repetitions reached when the specimen splits [2, 33-35]. Fig. 2.Indirect Tensile Test loading Set-up

6

ACCEPTED MANUSCRIPT 2.3 Method of analysis

One-factor-at-a-time (OFAT) methodology is a conventional approach for optimizing

T

multifactor experiments. OFAT is a changeable single factor method for a specific

IP

experiment design while other factors are kept constant. OFAT is unable to generate

SC R

appropriate output because the effects of interaction amongst all factors in the design are not examined truly, and so it is not capable of reaching the true optimum value [36, 37]. Hence,

NU

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

MA

minimal value [38-40]. Consequently, Design Expert 9.0.5.1 was designated for this purpose to generate statistical analysis, experimental designs and to calculate the sorbent adaption

D

conditions.

TE

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

CE P

variables including PET modifier (A) from zero to 1%, stress levels (B) from 200 kPa to 400 kPa and temperatures between 10 ˚C and 40 ºC, all at three levels, were studied through the

AC

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 4 shows the levels and range of the actual values of independent numerical variables. By using Eq. (1) all defined numerical variables transformed to the coded form. (1) where xi describes the coded value of the ith independent factor which is dimensionless. Actual value is defined as Xi, X0 is the center point actual value and ΔX refers to the step change of the ith variable.

7

ACCEPTED MANUSCRIPT A total of 34 experiments in random order were performed, together with five replications at the center points to provide accurate assessment of errors (Table 4). The fatigue life was

T

defined as the response to develop design of experiment modeling.

NU

SC R

IP

Equation (Eq. (2)) was developed to calculate the dependent variables [41, 42]:

In Eq. (2), Y is the calculated response, β0 is constant value. Independent variables in coded

MA

form are described as xi, and xj. The coefficients of βi and βii are the linear and quadratic terms. βij is the interaction term coefficient, ε is the random error, and the studied number of

D

factors is defined as n.

TE

In addition, in order to assess the appropriateness of proposed model, analysis of variance

CE P

(ANOVA) was performed. The coefficients of determination (R2 and R2adj) express the wellness of the fit to the suggested model. These values can be determined using the

AC

following equations [43]:

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]:

8

ACCEPTED MANUSCRIPT

(6)

SC R

IP

T

(5)

where Y is the predicted response, P represents the number of model parameters, residual mean square is described as σ2, and n is the number of experiments.

NU

After the F-test had been performed, the insignificant terms were found and eliminated from the model. Thereafter, the finalized model was introduced based on the significant variables.

MA

Finally, the optimum condition was determined to give the highest fatigue cycle response,

D

along with better mixture performance.

TE

Table 4: Experimental design layout and experimental results of the responses

CE P

Table 5: ANOVA analysis for fatigue life

3. Results and discussion

AC

As it was mentioned by Allen at al., three different modes of failure have been observed for the ITFT. In the first 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 defined 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 figure two types of fracture are observed known as ideal fracture and single cleft 9

ACCEPTED MANUSCRIPT 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 385866 cycles.

T

Having these values, RSM was utilized to find interactions between the outputs and variables

IP

which are independent. Eventually, after a regression analysis had been applied to all

SC R

responses described in the design matrix, a fitted quadratic polynomial equation was produced. The highest order polynomials in which the additional terms were significant and the models were not aliased were suggested by software. The numerical parameters (A, B and

MA

NU

C) were used to generate the predictive model according to Eq. 7:

(7)

D

Final Log10 (fatigue) equation = 3.8+0.15A-0.37B-0.92C+0.57C2

TE

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 fit is not adequate [39,45]. Hence, in this

CE P

study, in order to assess the significance and adequacy of the model, ANOVA analysis was performed and the results are reported in Table 5. In addition, this table shows the quadratic

AC

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 significant (model and term with p-value <0.05 indicate the model and the term are significant for 95% confidence intervals) for assessing the value of responses [46]. As the results show, PET (A), Stress level (B), Temperature (C), C2 are significant terms with p-values less than 0.05. However, A2, B2, AB, AC and BC were insignificant (p-value >0.100). Therefore, in order to improve the model and optimize the results, the insignificant term can be removed from the model [47]. In order to check the fitness of model regression coefficients, R2 and R2adj were calculated. Values of 0.9579 and 0.9422 were obtained for R2 and R2adj, respectively. This shows that 10

ACCEPTED MANUSCRIPT 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

T

predicted and actual values [40, 41, 48]. Ratio of signal-to-noise is measured by adequate

IP

precision to compare the variety of the estimated amounts at the design points with the

SC R

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 fit (LOF) F-test was also used to evaluate the adequacy of the model. LOF

NU

depicts the variation of the data around the fitted model, and the amount of LOF would be

MA

significant if the model does not fit the data well. It is worth noting that despite the LOF being significant, a reasonable agreement between the predicted and adjusted R2 were found

D

for all responses and it can be concluded that the models suggested for all responses can be

TE

used to navigate into design space to find an optimum condition [50, 51].

CE P

Fig.3. Fracture patterns (Left: ideal fracture, Right: single cleft fracture)

3.1 Statistical analysis

AC

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 figure there is an adequate agreement between the actual data which were obtained through experiment and the predicted ones. This agreement can also be understood by AP value (AP>4) for the fatigue responses (see Table 5). This verifies that the predicted model can be used to navigate the design space defined by the CCD. Fig.4. Design-expert plot; predicted vs. actual values plot for fatigue life (Logarithmic scale)

11

ACCEPTED MANUSCRIPT 3.2 One factor analysis One factor analysis is “changing one factor at a time” method. That is to say, in this method a

This process exists for optimizing other variables which would be time

IP

experiments.

T

single factor varies while all other factors are kept constant for a particular set of

SC R

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

NU

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

MA

level are kept constant at 25˚C and 300 kPa respectively.

D

Figs. 5, 6 and 7 show the one factor analysis of PET percentage, stress level and temperature

TE

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

CE P

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

AC

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 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 findings of

12

ACCEPTED MANUSCRIPT this study are based on controlled- stress test mode which are in support of previous studies [8, 53-56] that found the fatigue life of asphalt mixtures increased at lower temperatures.

T

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

IP

Fig.6. Effect of different Stress levels on the fatigue life (Logarithmic scale)

SC R

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

NU

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

MA

stress level and temperature on logarithmic scale is presented in Fig. 8. The response surfaces were generated based on Eq.7.

TE

D

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 significant. In physical definition,

CE P

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

AC

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.8. Effects of stress level and temperature on the fatigue life (Logarithmic scale), 0.5% PET 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 voids. In addition, improvement of fatigue life cannot 13

ACCEPTED MANUSCRIPT be due to the higher asphalt content in the mixture because as it is shown in Table 3 all the modified mixtures have lower asphalt contents than the unmodified mixture.

IP

T

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

SC R

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 modified asphalt mixture associated with lower stress levels. By reducing the amount of PET in asphalt mixture the fatigue life is

NU

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

MA

increment and decrease in the stress level have roughly the same effect on the fatigue life of

D

asphalt mixture.

CE P

TE

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

4. Conclusions

This paper aimed to evaluate the effect of applied load and temperature on the fatigue lives of

AC

unmodified and PET modified asphalt mixture. Statistical analysis was used in this investigation to find 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 defined 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.

14

ACCEPTED MANUSCRIPT (2) Both PET increment and decrease in the stress level have roughly the same effect on the fatigue life of asphalt mixture.

T

(3) The effect of temperature on the fatigue lives is more drastic compared to stress level

SC R

IP

and PET content.

Acknowledgement

NU

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.

MA

References

D

[1] Di Benedetto H, De La Roche C, Baaj H, Pronk A, Lundström R. Fatigue of Bituminous Mixtures. Mater Struct 2003; 37: 202-216.

TE

[2] Baghaee Moghaddam T, Karim MR, Syammaun T. Dynamic properties of stone mastic asphalt mixtures containing waste plastic bottles. Constr Build Mater 2012; 34: 236–242.

CE P

[3] Ghazi G. Al-Khateeb, Ghuzlan KA. 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 2014; 59: 254–261.

AC

[4] Baaj H, Perraton D, Di Benedetto H, Paradis M. Contribution à l'Etude de la Relation Entre le Module Complexe et la Résistance à la Fatigue et à l'Orniérage d'un Enrobé SMA, Proceedings of the 48th Annual Conference of the CTAA, Halifax NS, November 17–19, 2003. [5] Baaj H, Paradis M. 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 16–19, 2008. [6] Schmiedlin RB. Stone matrix asphalt: The wisconsin experience. Trans Res Record 1988; 1616: 34–41. [7] Asi IM. Laboratory comparison study for the use of stone matrix asphalt in hot weather climates. J Constr Build Mater 2006; 20: 982–989. [8] Nejad FM, Aflaki E, Mohammadi MA. Fatigue behavior of SMA and HMA mixtures. J Constr Build Mater 2010;24:1158–65. 15

ACCEPTED MANUSCRIPT [9] Moghaddam TB, Karim MR, Mahrez A. A review on fatigue and rutting performance of asphalt mixes. J Sci Res Essays 2011;6(4):670–82.

IP

T

[10] Perraton D, Baaj H, Di Benedetto H, Paradis M. Evaluation de la Résistance à la Fatigue des Enrobés Bitumineux Fondée sur l’Evolution de l’Endommagement du Matériau en Cours de l’Essai : Aspects Fondamentaux et Application à l’Enrobé SMA. Can. J. Civ. Eng 2003; 30: 902-913.

SC R

[11] Perraton D, Baaj H, Carter A. Comparison of Some Pavement Design Methods from a Fatigue Point of View. Effect of Fatigue Properties of Asphalt Materials. Road Mater Pavement 2010; 11: 833-861.

NU

[12] Al-Hadidy AI, Yi-qiu T. Mechanistic approach for polypropylene-modified flexible pavements.J Mater Design 2009; 30: 1133–1140.

MA

[13] Ahmadinia E, Zargar M, Karim MR, Abdelaziz M, Shafigh P. Using waste plastic bottles as additive for stone mastic asphalt. J Mater Design 2011; 32: 4844–4849. [14] Brovelli, C., Crispino, M., Pais, J., and Pereira, P. Assessment of Fatigue Resistance of Additivated Asphalt Concrete Incorporating Fibers and Polymers. J. Mater. Civ. Eng 20142; 6: 554–558.

TE

D

[15] Brovelli, C., Crispino, M., Pais, J., and Pereira, P. Using polymers to improve the rutting resistance of asphalt concrete. J Constr Build Mater 2015; 77: 117–123.

CE P

[16] Tapkin S. The effect of polypropylene fibers on asphalt performance. Build Environ 2008; 43: 1065-1071.

AC

[17] Casey D, McNally C, Gibney A, Gilchrist MD. Development of a recycled polymer modified binder for use in stone mastic asphalt. J Resour Conserv Recy 2008; 52: 1167– 1174. [18] Hınıslıoğlu S, Ağar E. Use of waste high density polyethylene as bitumen modifier in asphalt concrete mix. J Mater Lett 2004; 58: 267– 271. [19] Liseane P.T.L. Fontes , Glicério Trichês , Jorge C. Pais , Paulo A.A. Pereira. Evaluating permanent deformation in asphalt rubber mixtures. J Constr Build Mater 2010; 24: 1193– 1200. [20] Miao Yu, Guoxiong 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 2014; 42:1-11. [21] Siddiqui MN. Conversion of hazardous plastic wastes into useful chemical products. J Hazard Mater 2009; 167: 728–35. [22] Baghaee Moghaddam T, Soltani M, Karim MR. Evaluation of permanent deformation characteristics of unmodified and Polyethylene Terephthalate modified asphalt mixtures using dynamic creep test. J Mater Des 2014; 53: 317-324. 16

ACCEPTED MANUSCRIPT

[23] Baghaee Moghaddam T, Soltani M, Karim MR. Experimental characterization of rutting performance of Polyethylene Terephthalate modified asphalt mixtures under static and dynamic loads. Constr Build Mater 2014; 65: 487–494.

IP

T

[24] Kikuchi, S., Kronprasert, N., and Easa, S. Aggregate Blending Using Fuzzy Optimization. J Constr Eng M ASCE 2012; 138: 1411–1420.

SC R

[25] Soltani M, Moghaddam TB, Karim MR, Shamshirband S, Sudheer C. Stiffness performance of polyethylene terephthalate modified asphalt mixtures estimation using support vector machine-firefly algorithm. Measurement 2015; 63: 232–239.

NU

[26] Khodaii A, Haghshenas HF, Kazemi Tehrani H, Khedmati M. Application of Response Surface Methodology to Evaluate Stone Matrix Asphalt Stripping Potential. KSCE J Civ Eng 2013; 17:117-121.

MA

[27] Kavussi A, Qorbani M, Khodaii A, Haghshenas HF. Moisture susceptibility of warm mix asphalt: A statistical analysis of the laboratory testing results. Constr Build Mater 2014; 52: 511–517.

D

[28] Khodaii A, Haghshenas HF, Kazemi Tehrani H. Effect of grading and lime content on HMA stripping using statistical methodology. Constr Build Mater 2012; 34: 131–135.

CE P

TE

[29] Baghaee Moghaddam T, Soltani M, Karim MR. Stiffness modulus of Polyethylene Terephthalate modified asphalt mixture: A statistical analysis of the laboratory testing results. J Mater Design 2015; 68: 88-96. [30] Baghaee Moghaddam T, Soltani M, Karim MR, Baaj H. Optimization of asphalt and modifier contents for Polyethylene Terephthalate modified asphalt mixtures using Response Surface Methodology. Measurement 2015; 74: 159–169.

AC

[31] Khodaii A, Mehrara A. Evaluation of permanent deformation of unmodified and SBS modified asphalt mixtures using dynamic creep test. Constr Build Mater 2009; 23: 2586–92. [32] Kalyoncuoglu SF, Tigdemir M. A model for dynamic creep evaluation of SBS modified HMA mixtures. Constr Build Mater 2011; 25: 859–66. [33] Yan J, Ni F, Yang M, Li J. An experimental study on fatigue properties of emulsion and foam cold recycled mixes. Constr Build Mater 2010; 24: 2151–6. [34] B. Allen, I. Artamendi, and P. Phillips. Influence of temperature and aging on laboratory fatigue performance of asphalt mixtures. Advanced Testing and Characterization of Bituminous Materials, Two Volume Set. 2009 Taylor & Francis Group, London, ISBN 9780-415-55854-9. [35] A. M. Hartman, M. D. Gilchrist, G. Walsh. Effect of mixture compaction on indirect tensile stiffness and fatigue. J Transp Eng 2001; 127: 370-378.

17

ACCEPTED MANUSCRIPT [36] Frigon NL, Mathews D. Practical guide to experimental design. New York: John Wiley and Sons; 1997.

T

[37] Montogomery DC. Design and analysis of experiments. New York: John Wiley and Sons; 2005.

IP

[38] Khuri AI, Cornell JA. Response surfaces, design and analyses. 2nd ed. New York: Marcel Dekker Inc; 1996.

SC R

[39] Myer RH, Montogomery DC. Response surface methodology. Process and product optimization using designed experiment. 2nd ed. New York: John Wiley and Sons; 2002.

NU

[40] Azargohar R, Dalai AK. Production of activated carbon from Luscar char: experimental and modeling studies. Microporous Mesoporous Mater 2005; 85: 219–25.

MA

[41] Can MY, Kaya Y, Algur OF. Response surface optimization of the removal of nickel from aqueous solution by cone biomass of Pinussylvestris. Bioresour Technol 2006; 97: 1761–5.

D

[42] Aksu Z, Gönen F. 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 2006; 49: 205–16.

TE

[43] Körbahti BK, Rauf MA. Response surface methodology (RSM) analysis of photoinduceddecoloration of toludine blue. Chem Eng J 2008; 136: 25–30.

CE P

[44] Körbahti BK, Rauf MA. Determination of optimum operating conditions of carmine decoloration by UV/H2O2 using response surface methodology. J Hazard Mater 2009; 161: 281–6.

AC

[45] Körbahti BK, Rauf MA. Application of response surface analysis to the photolytic degradation of basic red 2 dye. Chem Eng J 2008; 138: 166–71. [46] Zabeti M, Daud WMAW, Aroua MK. Optimization of the activity of CaO/Al2O3 catalyst for biodiesel production using response surface methodology. Appl Catal A-Gen 2009; 366: 154–9. [47] Hosseinpour V, Kazemeini M, Mohammadrezaee A. Optimisation of ru-promoted Ir catalysed methanol carbonylation utilising response surface methodology. Appl Catal A-Gen 2011; 394: 166–75. [48] Garg UK, Kaur MP, Garg VK, Sud D. Removal of nickel (II) from aqueous solution by adsorption on agricultural waste biomass using a response surface methodological approach. Bioresour Technol 2008; 99: 1325–31. [49] Ölmez T. The optimization of Cr (VI) reduction and removal by electrocoagulation using response surface methodology. J Hazard Mater 2009; 162: 1371–8.

18

ACCEPTED MANUSCRIPT [50] Shafeeyan MS. Wan Daud WMA, Houshmand A, Arami-Niya A. The application of response surface methodology to optimize the amination of activated carbon for the preparation of carbon dioxide adsorbents. Fuel 2012; 94: 465–472.

IP

T

[51] Sánchez-Romeu J, País-Chanfrau JM, Pestana-Vila Y, López-Larraburo I, MassoRodríguez Y, Linares-Domínguez M, et al. Statistical optimization of immunoaffinity purification of hepatitis B surface antigen using response surface methodology. Biochem Eng J 2008; 38: 1–8.

SC R

[52] R.L. Mason, R.F. Gunst, J.L. Hess, Statistical Design and Analysis of Experiments, Eighth Applications to Engineering and Science, second ed., Wiley, New York, 2003.

NU

[53] Epps JA, Monismith CL. Fatigue of asphalt concrete mixtures-summary of existing information. Fatigue of compacted bituminous aggregate mixtures, ASTM STP508. American Society for Testing Materials; 1971. p. 19–45.

MA

[54] Pell PS, Taylor IF. Asphaltic road materials in fatigue. In: Proceedings of the association of the asphalt pavement technologists (AAPT), vol. 38, Los Angeles, California, USA; 1969. p. 577–93.

TE

D

[55] Al-Khateeb GG, Ghuzlan KA. 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 2014; 59: 254–261.

AC

Table titles:

CE P

[56] Bhattacharjee S, Mallick RB. Effect of temperature on fatigue performance of hot mix asphalt tested under model mobile load simulator. Int J Pavement Eng 2012; 13: 166-180.

Table 1: Properties of materials Table 2: Physical and mechanical properties of PET Table 3: Summary of mix design Table 4: Experimental design layout and experimental results of the responses Table 5: ANOVA analysis for fatigue life

Figure captions: Fig. 1. Aggregate particle size distribution for stone mastic asphalt Fig. 2.Indirect Tensile Test loading Set-up 19

ACCEPTED MANUSCRIPT Fig.3. Fracture patterns (Left: ideal fracture, Right: single cleft fracture) Fig.4. Design-expert plot; predicted vs. actual values plot for fatigue life (Logarithmic scale)

T

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

IP

Fig.6. Effect of different Stress levels on the fatigue life (Logarithmic scale)

SC R

Fig.7. Effect of different temperature on the fatigue life (Logarithmic scale) Fig.8. Effects of stress level and temperature on the fatigue life (Logarithmic scale), 0.5% PET

NU

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

AC

CE P

TE

D

MA

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

20

ACCEPTED MANUSCRIPT Figures:

100

T

Lower limit Upper limit Design limit

90

IP

80

SC R

60 50

NU

40 30

MA

20 10 0 0.075

0.6

2.36

4.75

9.5

12.5

Sieve size (mm)

TE

D

0.3

CE P

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

AC

Passing (%)

70

Fig. 2. Indirect Tensile Test loading Set-up

21

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig.3. Fracture patterns (Left: ideal fracture, Right: single cleft fracture)

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

22

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

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

Fig.6. Effect of different Stress levels on the fatigue life (Logarithmic scale)

23

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

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

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

24

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

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

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

25

ACCEPTED MANUSCRIPT Tables: Table 1: Properties of materials Unit

Used specification

Penetration at 25°C

0.1mm

ASTM D 5

Softening point

°C

ASTM D 36

Flash point

°C

ASTM D 92

Fire point

°C

ASTM D 92

Specific gravity

(g/cm3)

ASTM D 70

Value

Requirements

T

Property

IP

Asphalt

SC R

87.5

-

300

-

320

-

1.03

-

ASTM C 131

19.45

<30

BS 812 Part 105.1

2.72

<20

BS 812 Part 105.2

11.26

<20

BS 812 part 3

19.10

<30

(g/cm3)

ASTM C 127

2.60

-

%

ASTM C 127

0.72

<2

Bulk specific gravity

(g/cm3)

ASTM C 128

2.63

-

Absorption

%

ASTM C 128

0.4

<2

%

ASTM C 88

4.1

<15

NU

46.6

Flakiness index

%

Elongation index

%

Aggregate crushing value

%

Absorption

AC

Fine aggregate

CE P

Bulk specific gravity

Soundness loss

D

%

TE

L.A. Abrasion

MA

Coarse aggregate

26

IP

T

ACCEPTED MANUSCRIPT

Water absorption

%

Tensile strength

psi

Tensile modulus

SC R

Unit

Method

Value

ASTM D570

0.11

NU

Property

MA

Table 2: Physical and mechanical properties of PET

ASTM D638

11500

psi

ASTM D638

4×105

%

ASTM D638

70

psi

ASTM D790

15000

psi

ASTM D790

4×105

Approx glass transition temperature

˚C

-

75

Approx melting temperature

˚C

-

250

Specific gravity

g/

ASTM D792

1.35

D

Elongation at break

TE

Flexural strength

AC

CE P

Flexural modulus

27

ACCEPTED MANUSCRIPT Table 3: Summary of mix design BSGa

VMAb (%)

VFAc (%)

OACd (%)

0 (unmodified)

2.294

18.12

77.92

6.77

0.5

2.296

17.34

76.90

1

2.283

17.55

77.20

IP

T

PET(%)

6.51

SC R

a

6.36

bulk specific gravity of compacted mixture void in mineral aggregate c void filled with asphalt d optimum asphalt content

AC

CE P

TE

D

MA

NU

b

28

ACCEPTED MANUSCRIPT

Factor 1: PET (%)

Factor 2: stress level (kPa)

Factor 3: Temperature (°C)

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

NU MA

D

TE CE P

AC

SC R

Run

IP

T

Table 4: Experimental design layout and experimental results of the responses

Fatigue Cycles 196720 1541 184521 4341 7752 6072 51421 1243 3019 6063 167281 3033 385866 91291 4329 6061 6059 1544 6088 866 379731 7739 3133 3141 878 179491 167312 1239 7829 91302 53320 7931 6077 188821

29

ACCEPTED MANUSCRIPT

Table 5: ANOVA analysis for fatigue life F

p-value Prob > F

Squares

df

Square

Value

22.74

9

2.53

60.73

0.44

1

0.44

2.76

1

2.76

C-Temperature 16.76

1

Model significant A-PET significant

< 0.0001

0.0035

66.32

< 0.0001

< 0.0001

NU

10.48

402.84

5.472E-004

0.013

1

7.234E-004

0.017

0.8962

CE P

B-Stress level

IP

Performance

1

0.13

3.12

0.0899

1

0.082

1.96

0.1739

0.051

1

0.051

1.24

0.2772

1.74

1

1.74

41.83

< 0.0001

Residual

1.00

24

0.042

Lack of Fit

1.00

5

0.20

9017.24

< 0.0001

19

2.213E-005

significant

5.472E-004

1

7.234E-004

0.13

0.082

AC Insignificant BC Insignificant A2

B2

AC

Insignificant

D

Insignificant

MA

AB

TE

significant

16.76

Model

T

Source

Mean

SC R

Sum of

0.9096

Insignificant C2 significant

significant Pure Error Cor Total

4.205E-004 23.73

Adequate precision (AP)

33 25.936 30

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

31

ACCEPTED MANUSCRIPT Highlights

CE P

TE

D

MA

NU

SC R

IP

T

Effect of PET modification on fatigue life of asphalt mixture was examined. Different temperatures and stress levels were designated. Statistical analysis used to find the interaction between selected variables. A good agreement between experimental results and predicted values was obtained.

AC

   

32