stiffness modulus of polyethylene terephthalate modified asphalt mixture: a statistical analysis of...
TRANSCRIPT
Accepted Manuscript
Stiffness modulus of Polyethylene Terephthalate modified asphalt mixture: A
statistical analysis of the laboratory testing results
Taher Baghaee Moghaddam, Mehrtash Soltani, Mohamed Rehan Karim
PII: S0261-3069(14)00965-0
DOI: http://dx.doi.org/10.1016/j.matdes.2014.11.044
Reference: JMAD 6995
To appear in: Materials and Design
Received Date: 14 September 2014
Accepted Date: 25 November 2014
Please cite this article as: Baghaee Moghaddam, T., Soltani, M., Karim, M.R., Stiffness modulus of Polyethylene
Terephthalate modified asphalt mixture: A statistical analysis of the laboratory testing results, Materials and
Design (2014), doi: http://dx.doi.org/10.1016/j.matdes.2014.11.044
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.
1
Title: Stiffness modulus of Polyethylene Terephthalate modified asphalt mixture: A statistical
analysis of the laboratory testing results.
Authors’ names:
Taher Baghaee Moghaddam a,* (First name: Taher, Last name: Baghaee Moghaddam),
Mehrtash Soltani a, Mohamed Rehan Karim
a (First name: Mohamed Rehan, Last name: Karim)
Authors’ affiliation addresses:
a Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering,
University of Malaya, 50603 Kuala Lumpur, Malaysia.
*Corresponding author:
Taher Baghaee Moghaddam
Tel.:+60108927064; Fax: +60379552182
Corresponding author E-mail address:
[email protected]; [email protected]
Corresponding author postal address:
Center for Transportation Research, Department of Civil Engineering, Faculty of Engineering,
University of Malaya, 50603 Kuala Lumpur, Malaysia.
2
Stiffness modulus of Polyethylene Terephthalate modified asphalt
mixture: A statistical analysis of the laboratory testing results
Taher Baghaee Moghaddam
a,*, Mehrtash Soltani
a, Mohamed Rehan Karim
a
aCenter for Transportation Research, Department of Civil Engineering, Faculty of Engineering,
University of Malaya, 50603 Kuala Lumpur, Malaysia
Abstract
Stiffness of asphalt mixture is a fundamental design parameter of flexible pavement. According
to literature, stiffness value is very susceptible to environmental and loading conditions. In this
paper, effects of applied stress and temperature on the stiffness modulus of unmodified and
Polyethylene Terephthalate (PET) modified asphalt mixtures were evaluated using Response
Surface Methodology (RSM). A quadratic model was successfully fitted to the experimental
data. Based on the results achieved in this study, the temperature variation had the highest impact
on the mixture’s stiffness. Besides, PET content and amount of stress showed to have almost the
same effect on the stiffness of mixtures. The optimal amount of PET was found to be 0.41% by
weight of aggregate particles to reach the highest stiffness value.
Keywords: Asphalt mixture; Mixture stiffness; Waste polyethylene terephthalate;
Environmental temperature; Applied stress; Response surface methodology.
*Corresponding author: [email protected]; [email protected]
3
1. Introduction
Stiffness of asphalt mixture is a fundamental design parameter of flexible pavement. It was
found that there is a correlation between stiffness and other mixture properties such as rutting
and fatigue, and thus it can be used as a criterion to evaluate Asphalt Concrete (AC) mixture
performance [1]. As it is mentioned by Strategic Highway Research Program (SHRP) the
stiffness value of AC mixture is very susceptible to environmental temperature and loading
conditions [2].
Stone Mastic Asphalt (SMA) is gap-graded AC mixture which has been developed in Germany
in 1916s. SMA consists of more course aggregate particles, mineral filler and asphalt binder.
Due to inherited structure of SMA, it provides better permanent deformation (rutting)
performance and durability compared to conventional dense-graded mixture [3, 4]. Draindown is
a common problem in SMA mixture because it contains higher amount of asphalt binder. Hence,
to prevent from draindown in SMA mixture stabilizer additives, fibers and polymers are used.
Using polymer in SMA mixture is very common. Utilizing polymer in SMA mixture prevents
not only from the binder draindown but also it can enhance mixture performance [5, 6]. In many
cases, using polymers causes higher construction cost due to high cost of polymers. To overcome
this disadvantage, many studies investigated using waste polymers in asphalt mixtures [6-8].
One of the important industrial plastic materials is Polyethylene Terephthalate (PET). PET is a
semi-crystalline thermo plastic polymer material which is used in beverage and food industries
for years. Nowadays, a large amount of waste PET is produced in the world and it is going to
cause a serious environmental challenge due to non-biodegradability of PET [9]. 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 find solutions to tackle with this potential
4
environmental hazard and, moreover, to decrease construction cost imposed by application of
polymers in asphalt mixture [6, 10-12].
Statistical analysis is a precise and popular way to explore and present interactions between
parameters affecting one phenomenon. Statistical analysis in pavement engineering has
prominent utilization because it helps road engineers and designers 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 and the response came into popularity [13-15].
Aim of this study was examining the AC mixture stiffness at elevated temperatures and stress
levels for the unmodified and PET modified mixtures following by finding interactions between
these fundamental factors using RSM based on Central Composite Design (CCD).
2. Materials and methods
Asphalt mixtures were fabricated using 80-100 asphalt penetration grade. Granite-rich aggregate
particles were used for this investigation. 9% of filler was utilized. The aggregate particle size
distribution is shown in Fig. 1. In order to have better understanding about the materials
characteristics several tests were performed on asphalt cement and aggregate particles and the
results are listed in Table 1.
PET flakes have been used for this study which were obtained from waste PET bottles. For using
PET flakes in asphalt mixture, the PET bottles were cut to small parts and by using crushing
5
machine these small parts were crushed. Thereafter, the crushed PET particles were sieved and
the particles which were smaller than 2.36 mm in size were used for this investigation.
Fig. 1. Aggregate particle size distribution for stone mastic asphalt
Table 1: Properties of materials
2.1 Mixture fabrication
In order to fabricate SMA mixture, 1100 g of mixed aggregate and filler were heated inside oven
with temperature of 160˚C for 3 hours. Asphalt cement was also heated at 130˚C to be suitable
for mixing with aggregate particles. All the materials were mixed at the temperature of 160˚C.
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. The loose mixture was compacted
using Marshall compactor and 50 blows of compaction efforts were applied on each side of the
mixture. It is worth mentioning that all the mixtures were fabricated at their optimum asphalt
contents (OAC). The optimum asphalt content for SMA mixtures is usually selected to produce
3–5% air voids [4, 5]. In this study, the OAC was selected to produce 4% air voids. The
summary of the mix design is reported in Table 2.
Table 2: Summary of mix design
2.2 Indirect tensile stiffness modulus test
Indirect tensile stiffness modulus (ITSM) test gives the relationship between stress and strain of
asphalt mixture and used to evaluate the stiffness of asphalt mixture at specific environmental
6
conditions. ITSM test was performed in accordance with AASHTO TP31. ITSM test can be
performed by using Universal Testing Machine (UTM) which is one of the important testing
equipment in pavement laboratory. UTM is a computer controlled system which operates
automatically. During the test, compressive haversine waveform loads were applied across the
thickness of specimen, and by utilizing Linear Variable Differential Transducers (LVDTs) which
were installed along diametrical section of specimen displacement of asphalt mixture was
measured. Horizontal tensile stress and stiffness modulus of asphalt mixtures was calculated
using the following equations [10, 16]:
Where is the maximum horizontal tensile stress in middle of specimen (kPa); is the
stiffness modulus (MPa); P, applied vertical peak load (N), H; amplitude of horizontal
deformation (mm), t; average thickness of specimen (mm); d, average diameter of specimen
(mm) and ν, Poisson’s ratio.
In order to characterize effects of applied stress and temperatures on the mixture’s stiffness,
ITSM test was conducted at stress levels of 200, 300 and 400 kPa for each percentage of PET
which are the stress amounts mostly used at pavement laboratories. Additionally, testing
temperatures of 10˚C, 25˚C and 40˚C were designated which can be referred to relatively low,
medium and high environmental temperatures respectively.
7
2.3 Method of analysis
One factor at a time (OFAT) methodology is a conventional approach to optimize multifactor
experiments. OFAT includes a changeable single factor for a specific experiments design while
other factors are kept constant. OFAT is unable to provide appropriate output because the effect
of interactions amongst all involved factors in the designs is not examined truly, and it is not
capable of reaching the true optimum value [17, 18]. Hence, RSM methodology was introduced
for parameter optimization in a way that number of experiments and interaction among the
parameters are reduced to minimal value [19-21]. Consequently, the Design Expert 8.0.5 was
designated for this study to generate statistical analysis, experimental designs, and to calculate
the sorbent adaption conditions.
For this study, a developed quadratic model and a=0.5 were utilized using RSM method for
design and data analyzing. In this investigation, the effects of three independent numerical
variables including PET modifier (A) from zero to 1%, stress levels (B) from 200 kPa to 400 kPa
and temperatures (C) between 10 and 40 ºC, 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 irrespective regions of interest [5-8, 10-12].
Table 3 shows the levels and range of the actual values of independent numerical variables. By
using Eq. (3) all defined numerical variables transformed to the coded form.
(3)
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.
8
Totally 34 experiments in randomized order were performed, together with five replications at
center points to provide accurate assessment of errors (Table 3). The stiffness was defined as the
response to develop design of experiment modeling. Eq. (4) was introduced to calculate the
dependent variables [22, 23]:
(4)
In the Eq. (4), Y is the calculated response, β0 is the constant. Independent variables in coded
forms 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 factors is
described as n.
Besides, in order to assess appropriateness of the proposed model, analysis of variance
(ANOVA) was performed. The coefficients of determination (R2 and R
2adj) express the wellness
of the fit to suggested model. These values can be determined using the following equations [24,
25]:
(5)
(6)
In this equation, SS is the sum of squares and DF is degrees of freedom.
Eq. (7), Eq. (8) 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 [24- 26]:
9
(7)
(8)
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.
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.
Eventually, the optimum condition was determined to give the highest stiffness response, along
with better mixture performance.
Table 3: Layout of experimental results and DOE design
Table 4: Anova analysis for stiffness
3. Results and discussion
Indirect tensile stiffness modulus test was conducted on the PET modified SMA mixtures at
elevated temperatures and stress levels. Table 3 represents the layout for experimental design and
the amounts of stiffness responses. Having these values, RSM was utilized to find interactions
between the outputs and variables which are independent. Eventually, a fitted quadratic
polynomial equation was produced after a regression analysis had been applied to all responses
described in the design matrix. The highest order polynomials with significant model
10
performance are suggested by the software. This model was utilized to find the optimum
condition. The numerical parameters (A, B and C) were used to generate the predictive model
according to Eq. 9:
Final Equation for Stiffness = 4051.55-325.25A-171.5B-4776.15C+256.75AC-157.96A2+
1602.54C2 (9)
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 [20, 27]. Hence, in this
study in order to assess the significance and adequacy of the model ANOVA analysis was
performed and the results are reported in Table 4. In addition, this table shows the quadratic
models for coded factors, and represents the other statistical parameters for stiffness response. In
this table, p-values which are less than 0.0001 imply that the model and parameter are significant
(model and term p-value <0.05 indicate the model and the term are significant for 95%
confidence intervals) for assessing the value of responses [28].
In this study, the effect of PET (A), on stiffness property of SMA mixtures at different stress
levels (B) and temperatures (C) is modeled. Besides, in order to improve the model performance,
insignificant terms (with p-value >0.100) are eliminated from the model [29].
In order to check the fitness of the model, regression coefficients, R2 and R
2adj were calculated.
Values of 0.9990 and 0.9986 were obtained for the R2
and R2
adj, respectively. This shows that
99.8 % of the total variation in the stiffness response could be explained by the quadratic model.
The high R2 and adjusted R
2 values indicate that there is a good agreement between predicted
and actual values [21, 22, 30]. Ratio of signal-to-noise is measured by adequate precision to
compare the variety of the estimated amounts at the design points to the average prediction error.
Adequate model discrimination was found in this study when the adequate precision ratio of
11
136.602 was calculated for the stiffness which is much higher than the value of 4 [31]. The lack
of fit (LOF) F-test was also used to evaluate the adequacy of the model. LOF depicts the
variation of the data around the fitted model, and the amount of LOF would be significant if the
model does not fit the data well. It is worth noting that despite the lack-of-fit was significant, the
reasonable agreement between the predicted and adjusted R2 were found for all responses which
can be concluded the suggested model for all responses can be used to navigate into design
space to find an optimum condition [32,33].
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. 2 shows the actual versus predicted values
plots of parameters removal for stiffness modeling. As it is depicted in this figure there is an
adequate agreement between the actual data amounts and the predicted ones. The same thing can
be achieved from AP value (AP>4) for the stiffness responses (see Table 4). This verifies that
predicted model can be used to navigate the design space defined by the CCD.
Fig. 2. Design-expert plot; predicted vs. actual values plot for stiffness
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 is varied while all other factors are kept constant for a particular set of experiments.
And this process exists for optimizing other variables which would be time consuming. In this
method, trial and error are commonly existed for the optimization of variables, and, moreover,
12
there is always a lack to reach a true optimum amount which should be obtained by considering
the interaction among all the variables [32, 34].
Each factor in this analysis is evaluated separately. Fig. 3 reveals the effect of PET on the
stiffness properties of SMA mixtures. As it can be seen in this figure the amount of stiffness is
decreased at higher PET contents. The possible reason for this result might be due to the
mechanical properties of PET particles in the mix. In fact, because the melting point of PET is a
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 flexible and cause
higher deformation under loading application. The same pattern is found between stress level and
stiffness value as it is depicted in Fig. 4. That is to say, by increasing the stress level, the amount
of deformation in the mixture is increased and according to Eq. 2 the stiffness is reduced
mutually. Moreover, in Fig. 5, the higher decreasing slope line may imply that the temperature
variation has dominating influence on the stiffness property of SMA mixtures, and this
represents the importance of ambient temperature on the stiffness property of asphalt mixture.
Fig. 3. Effect of PET percentage on the stiffness
Fig. 4. Effect of different stress levels on the stiffness
Fig. 5. Effect of different temperatures on the stiffness
3.3 Effects of temperature and stress levels on the stiffness
The quadratic model for the effect of stress level and temperature on stiffness is presented in Fig.
6. The response was generated using the Eq. (9). The Fig. 6 shows that by variation of
temperature from 10˚C to 40˚C the amount of stiffness is decreased, however, the effect of stress
seems to be negligible compared to the temperature variation.
13
Fig. 6. Effects of stress level and temperature on the stiffness, 0.5% PET
3.4 Effects of temperature and PET content on the stiffness
Effects of two parameters including PET and temperature are evaluated on stiffness as shown in
Fig. 7. The Fig. 7 depicts the mixture stiffness is more susceptible against variation of
temperature. The stiffness amounts decrease from over 10000 MPa to under 1000 MPa when the
temperature increases from 10˚C to 40˚C. Additionally, it is shown at lower temperatures the
PET amounts have more influence on the mixture stiffness. This can be referred to susceptibility
of asphalt binder against temperature variation which plays an important role on the mixture
properties. In other words, when the ambient temperature increases, the asphalt binder becomes
soft which can eventually results in lower mixture stiffness. Besides, the influence of PET on
mixture’s stiffness is overshadowed by the temperature.
Fig. 7. Effects of PET percentage and temperature on the stiffness, stress level 300 kPa
3.5 Effects of PET and stress level on the stiffness
Fig. 8 shows the effects of stress level and PET on the stiffness of asphalt mixture. It might be
realized that both level of stress and PET content have nearly the same effect on stiffness
property of asphalt mixture though the variation of PET might be more influential, for instance at
400 kPa by increasing the PET amount from 0 to 1% the stiffness value decreases by 378 MPa.
Fig. 8. Effects of PET percentage and stress level on stiffness, 25˚C
14
3.6 Finding the optimum condition
In the construction of asphalt mixture, relatively high stiffness is demanded to resists against
traffic loading and consequently permanent deformation. Optimum PET content for maximizing
the stiffness within the considered experimental range was accomplished using the Design-
Expert software. An overlay plot is used to view constraints on process or formulation. The
optimum conditions could be graphically visualized by super imposing the contours of the
response surfaces in an overlay plot. As depicted by Fig. 9, 0.41 percent of PET is suggested by
the software to provide the highest stiffness value.
Fig. 9. Design-expert plot; overlay plot for optimal amount of PET
4. Conclusions
This paper aimed to evaluate the effects of applied load and temperature on the stiffness property
of unmodified and PET modified asphalt mixtures. Statistical analysis was used in this
investigation to find interactions between selected variables. A good agreement was found
between predicted and actual values. A quadratic model was successfully fitted to the
experimental data. Based on the results achieved in this study the following conclusions can be
derived:
(1) Stiffness of asphalt mixture was affected by amounts of applied stress and PET content.
However, Mixture stiffness was more susceptible against temperature variations.
(2) The results showed that the PET modification had more influence on the asphalt
mixture’s stiffness at lower temperatures.
15
(3) The findings of this study revealed that the variation of PET was more influential on the
stiffness of mixtures compared to the stress levels.
(4) 0.41 % of PET content was selected as the optimal PET value to reach the highest
stiffness for the SMA mixtures which have been designated for this study.
Acknowledgements
The authors express their sincere thanks for the funding support they received from the Ministry
of Higher Education Malaysia, grant no: FP021-2011A and University of Malaya grant no:
RP010A-13SUS.
References
[1] Pellinen TK, Witczak MW. Use of Stiffness of Hot-Mix Asphalt as a Simple Performance
Test. Trans Res Record 2002; 1789: 80-90.
[2] Tayebali AA, Tsai BW, Carl L. 1994. Monismith. SHRP-A-388 Stiffness of Asphalt-
Aggregate Mixes. Institute of Transportaion Studies, University of California, Berkeley.
[3] Schmiedlin RB. Stone matrix asphalt: The wisconsin experience. Trans Res Record 1988;
1616: 34–41.
[4] Asi IM. Laboratory comparison study for the use of stone matrix asphalt in hot weather
climates. J Constr Build Mater 2006; 20: 982–989.
[5] Al-Hadidy AI, Yi-qiu T. Mechanistic approach for polypropylene-modified flexible
pavements. J Mater Design 2009; 30: 1133–1140.
[6] 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.
[7] 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.
16
[8] Hınıslıoğlu S, Ağar E. Use of waste high density polyethylene as bitumen modifier in
asphalt concrete mix. J Mater Lett 2004; 58: 267– 271.
[9] Siddiqui MN. Conversion of hazardous plastic wastes into useful chemical products. J
Hazard Mater 2009;167:728–35.
[10] 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.
[11] 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.
[12] 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.
[13] 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.
[14] 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.
[15] 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.
[16] Amir Modarres, Hamidreza Hamedi. Effect of waste plastic bottles on the stiffness and
fatigue properties of modified asphalt mixes. J Mater Des 2014; 61: 8–15.
[17] Frigon NL, Mathews D. Practical guide to experimental design. New York: John Wiley and
Sons; 1997.
[18] Montogomery DC. Design and analysis of experiments. New York: John Wiley and Sons;
2005.
[19] Khuri AI, Cornell JA. Response surfaces, design and analyses. 2nd ed. New York: Marcel
Dekker Inc; 1996.
[20] Myer RH, Montogomery DC. Response surface methodology. Process and product
optimization using designed experiment. 2nd ed. New York: John Wiley and Sons; 2002.
[21] Azargohar R, Dalai AK. Production of activated carbon from Luscar char: experimental and
modeling studies. Microporous Mesoporous Mater 2005; 85:219–25.
17
[22] Can MY, Kaya Y, Algur OF. Response surface optimization of the removal of nickel from
aqueous solution by cone biomass of Pinus sylvestris. Bioresour Technol. 2006;97:1761–5.
[23] 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.
[24] Design-Expert® Software, Version 6 User’s Guide, 2001.
[25] Körbahti BK, Rauf MA. Response surface methodology (RSM) analysis of photoinduced
decoloration of toludine blue. Chem Eng J 2008;136:25–30.
[26] 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.
[27] 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.
[28] 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.
[29] 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.
[30] 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.
[31] Ölmez T. The optimization of Cr (VI) reduction and removal by electrocoagulation using
response surface methodology. J Hazard Mater 2009;162:1371–8.
[32] 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.
[33] Sánchez-Romeu J, País-Chanfrau JM, Pestana-Vila Y, López-Larraburo I, Masso-
Rodríguez 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.
[34] 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.
18
Figure captions:
Fig. 1. Aggregate particle size distribution for stone mastic asphalt
Fig. 2. Design-expert plot; predicted vs. actual values plot for stiffness
Fig. 3. Effect of PET percentage on the stiffness
Fig.4. Effect of different stress levels on the stiffness
Fig. 5. Effect of different temperatures on the stiffness
Fig. 6. Effects of stress level and temperature on the stiffness, 0.5% PET
Fig. 7. Effects of PET percentage and temperature on the stiffness, stress level 300 kPa
Fig. 8. Effects of PET percentage and stress level on the stiffness, 25˚C
Fig. 9. Design-expert plot; overlay plot for optimal amount of PET
Table titles:
Table 1: Properties of materials
Table 2: Summary of mix design
Table 3: Layout of experimental results and DOE design
Table 4: Anova analysis for stiffness
19
Fig. 1. Aggregate particle size distribution for stone mastic asphalt
0
20
40
60
80
100
120
0.075 0.3 0.6 2.36 4.75 9.5 12.5
Pass
ing (
%)
Sieve size (mm)
Lower limit
Upper limit
Design limit
20
Fig. 2. Design-expert plot; predicted vs. actual values plot for stiffness
Fig. 3. Effect of PET percentage on the stiffness
21
Fig. 4. Effect of different stress levels on the stiffness
Fig. 5. Effect of different temperatures on the stiffness
22
Fig. 6. Effects of stress level and temperature on the stiffness, 0.5% PET
23
Fig. 7. Effects of PET percentage and temperature on the stiffness, stress level 300 kPa
Fig. 8. Effects of PET percentage and stress level on the stiffness, 25˚C
24
Fig. 9. Design-expert plot; overlay plot for optimal amount of PET
25
Table 1: Properties of materials
Property Unit Used specification Value Requirements
Asphalt
Penetration at 25°C 0.1mm ASTM: D5 87.5 -
Softening point °C ASTM: D36 46.6 -
Flash point °C ASTM: D92 300 -
Fire point °C ASTM: D92 320 -
Specific gravity (g/cm3) ASTM: D70 1.03 -
Coarse aggregate
L.A. Abrasion % ASTM: C131 19.45 <30
Flakiness index % BS 812 Part 105.1 2.72 <20
Elongation index % BS 812 Part 105.2 11.26 <20
Aggregate crushing value % BS 812 Part 3 19.10 <30
Bulk specific gravity (g/cm3) ASTM: C127 2.60 -
Absorption % ASTM: C127 0.72 <2
Fine aggregate
Bulk specific gravity (g/cm3) ASTM: C128 2.63 -
Absorption % ASTM: C128 0.4 <2
Soundness loss % ASTM: C88 4.1 <15
26
Table 2:Summary of mix design
PET(%) BSGa
VMAb(%) VFA
c(%) OAC
d (%)
0 2.294 18.12 77.92 6.77
0.5 2.296 17.34 76.90 6.36
1 2.283 17.55 77.20 6.51
abulk specific gravity of compacted mixture
b void in mineral aggregate
cvoid filled with asphalt
doptimum asphalt content
27
Table 3: Layout of experimental results and DOE design
Run Factor 1:
PET (%)
Factor 2: stress
level (kPa)
Factor 3:
Temperature ( °C)
Stiffness
(MPa)
1 0 200 10 10801
2 1 400 40 452
3 0.5 300 10 10608
4 0 200 40 1011
5 1 300 25 3689
6 0.5 300 25 4088
7 0.5 200 25 4310
8 0.5 400 25 3758
9 0 300 25 4041
10 0.5 300 25 4083
11 1 400 10 9391
12 0 300 25 4083
13 1 200 10 9710
14 0 400 10 10762
15 0 200 40 1071
16 0.5 300 25 4025
17 0.5 300 25 4081
18 1 400 40 431
19 0.5 300 25 4022
20 0 400 40 632
21 1 200 10 9712
22 1 300 25 3722
23 0.5 300 40 645
24 0.5 300 40 623
25 0 400 40 664
26 0.5 300 10 10701
27 1 400 10 9410
28 0.5 400 25 3671
29 1 200 40 832
30 0 400 10 10769
31 0.5 200 25 4261
32 1 200 40 821
33 0.5 300 25 4089
34 0 200 10 10841
28
Table 4:Anova analysis for stiffness
Source Sum of
Squares
Degree of
Freedom Mean
Square F Value Prob > F
Model
performance
Model 478763078.9 9 53195897.65 2625.158 < 0.0001 Significant
A 2115751.25 1 2115751.25 104.4099 < 0.0001 Significant
B 588245 1 588245 29.02923 < 0.0001 Significant
C 456232176.5 1 456232176.5 22514.55 < 0.0001 Significant
A2 133703.7816 1 133703.7816 6.598131 0.0169 Significant
B2 9322.861812 1 9322.861812 0.460073 0.5041 Insignificant
C2 13761302.46 1 13761302.46 679.1048 < 0.0001 Significant
AB 15252.25 1 15252.25 0.752681 0.3942 Insignificant
AC 1054729 1 1054729 52.04969 < 0.0001 Significant
BC 42436 1 42436 2.094169 0.1608 Insignificant
Residual 486333.2542 24 20263.88559
Lack of Fit
(LOF) 466621.9209 5 93324.38418 89.95654 < 0.0001 Significant
Pure Error 19711.33333 19 1037.438596
Cor Total 479249412.1 33
Adequate
precision (AP) 136.602
29
Highlights
> Effect of PET modification on stiffness property of asphalt mixture was examined.
> Different temperatures and loading amounts were designated.
> Statistical analysis was used to find interactions between selected variables.
> A good agreement between experimental results and predicted values was obtained.
> Optimal amount of PET was calculated to achieve the highest mixture performance.