Creep curve modelSBS polymer
t thhis5 motendatif thles
tance than 4% and 6%. The mathematical models of creep curves showed that lower stress levels in
ic proptempere in teceptiblis a m
polymers .2. Methods that evaluate physical and mechanical properties of
polymer modied asphalts .3. Methods that evaluate physical and mechanical properties of
polymer modied asphalt mixtures .
nent deformation of each layer of the pavement structure underrepetitive trafc loading ; however, asphalt layer has a remark-able role in its magnitude [11,12].
Permanent deformation in pavements has long been recognizedto include two different modes according to Huang  and Gok-hale et al. . The rst mode is known as compactive deformation(consolidation of layers) and the second mode is plastic deforma-tion (asphalt shear ow). In the former mode, the deformed surfaceis lower than the initial pavement surface, and occurs in the wheelpath. In the later mode, the deformed surface is higher than the
* Corresponding author. Tel.: +98 021 77535815.E-mail addresses: email@example.com (A. Khodaii), Amir.firstname.lastname@example.org (A.
Construction and Building Materials 23 (2009) 25862592
Contents lists availab
evMehrara).ting in wheel tracks.In recent years, different kinds of polymers have been used to
modify properties of asphalt mixtures, among them SBS is one ofthe most widely used which can extremely improve the mechani-cal properties of asphalt mixtures . So far various methods havebeen used to identify the effectiveness of polymers in modicationof asphalt mixtures. These methods can generally be categorized inthe following three groups:
1. Methods that evaluate physical and mechanical properties of
mentioned types of testing methods the third group has the high-est accuracy. In this research; therefore, dynamic creep tests,which have high capability to estimate the permanent deformationsusceptibility of asphalt mixtures according to researchers ,were performed on modied and unmodied mixtures usingUTM25 machine.
Rutting is dened as the progressive accumulation of perma-1. Introduction
Because of asphalts visco-elastbehavior depends on its ambientthe fact that its viscosity falls by a risder becomes softer and more susdeformation. Permanent deformation0950-0618/$ - see front matter 2009 Elsevier Ltd. Adoi:10.1016/j.conbuildmat.2009.02.015dynamic creep test can not show the real behavior of asphalt mixtures and particularly the modiedmixtures.
2009 Elsevier Ltd. All rights reserved.
erties, its mechanicalature. On accounts ofmperature, asphalt bin-e to adopt permanentajor contributor to rut-
Khattak et al.  believe in formation of brils in PMA mixtureswhich seems to make the mechanical behavior of these mixturesmore complicated than conventional mixtures. Bahia et al. demonstrated that for evaluating the performance of polymermodied asphalt binders, tests should be conducted on PMA as-phalt mixtures to provide a reasonable expectation of the perfor-mance of the polymer in mixture. It seems that among thePermanent deformationDynamic creep test results, dense graded mixtures had higher permanent deformation susceptibility than coarse graded mix-
tures. Moreover, 5% of SBS polymer had better effect on improvement of permanent deformation resis-Evaluation of permanent deformation ofmixtures using dynamic creep test
Ali Khodaii, Amir Mehrara *
Department of Civil Engineering, Amirkabir University of Technology, Tehran, Iran
a r t i c l e i n f o
Article history:Received 10 May 2008Received in revised form 5 February 2009Accepted 6 February 2009Available online 16 March 2009
a b s t r a c t
Recent researches show thaied asphalt mixtures. In tied samples, using UTM2permanent deformation pdetermined the type of grathe mechanical behavior ocreep behavior of the samp
journal homepage: www.elsll rights reserved.modied and SBS modied asphalt
ere is a serious need for more accurate methods to evaluate polymer mod-research, dynamic creep test was conducted on unmodied and SBS mod-achine to this end. During this work it was attempted to compare thetial of the coarse graded mixtures with dense graded mixtures. Havingon with lower permanent deformation, the amount of polymer to improvee samples made with this type of gradation was investigated. Finally thewas estimated by the Zhou three-stage creep model. Based on the obtained
le at ScienceDirect
ier .com/locate /conbui ldmat
Dynamic creep test has various outcomes that can be used as a
used for stages 13, respectively. Equations below show each stageof the model:
N6NPS; ep aNb 1eps aNbPS and NPS6N6NST ; eP ePScNNPS 2eST ePScNST NPS and NPNST ; eP eST def NNST 1 3
4. Objectives and procedure of tests
In view of the above, efforts were made to conduct dynamiccreep test on SBS modied and unmodied asphalt mixtures. Theobtained creep curves were interpreted according to mathematicalmodels. During the research following stages were performed:
d Building Materials 23 (2009) 25862592 2587measure of evaluation of permanent deformation potential. Airey, for instance, used ultimate strain and mean strain rate for thispurpose, and according to him, of the two mentioned parametersthe latter is more reliable to measure the rutting performance ofthe asphalt mixture than the former because mean strain, unlikeultimate strain, is independent of the initial strain experiencedduring the dynamic creep test.
Kaloush et al.  used another outcome of dynamic creep testnamely ow number (FN) as a comparison measure. This parame-ter is obtained from creep curve (a plot of cumulative plastic strainversus number of load cycles). The creep curve is generally dividedinto three stages as indicated in the literature . As reported byZhou et al.  over the past 40 years, various mathematical mod-els, among which are well knownmodels such as Barksdales Semi-log model in 1972, Power-law models based on Monismith modelin 1975, and Tseng and Lyttons model in 1989, have been devel-oped for tting the creep curve and estimating the (FN) parameterin asphalt mixtures.
Zhou et al.  believe that (FN) can not be an appropriate cri-terion for evaluating the mixtures permanent deformation poten-tial, so they proposed a three-stage model (one model for eachstage of the creep curve) with a simple algorithm for estimatingthe initial point of each stage. West et al. have also developed athree-stage model , but their model can not estimate theboundary points of curve stages.
3.1. Outline of Zhou model
Zhou performed a comprehensive research to develop a modeloriginal surface. This mode of deformation, which typically occursbetween and outside wheel paths, is attributed to shear ow of as-phalt materials under trafc loads, and is often referred to asheave.
Various experimental tests such as static creep, dynamic creep,wheel tracking and indirect tensile tests are used to evaluate per-manent deformation potential of asphalt mixtures. There are, how-ever, doubts about whether or not these tests can properlyevaluate deformation properties of PMA mixtures. Tayfure et al., for example, believe that static creep test can not accuratelyshow the differences which exist between modied and unmodi-ed asphalt mixtures.
Among the mentioned methods of assessing permanent defor-mation potential of asphalt mixtures, dynamic creep test isthought to be one of the best methods. This test was developedby Monismith et al.  in 1970, based on the concepts of axialcompression test. NCHRP conducted a comprehensive researchstudy to develop a simple mechanical test to supplement theSuperpave volumetric method of mixtures design. Research ofKaloush and Witzak  also indicates that Superpave volumetricmethod alone can not guarantee the proper functioning of the as-phalt layer according to eld experiments.
NCHRP reported, that among the ve laboratory tests investi-gated, dynamic creep test had very good correlation with mea-sured rut depth and a high capability to estimate ruttingpotential of asphalt layers . On grounds of the results of theresearch, dynamic creep test was chosen as an appropriate labora-tory method to evaluate the permanent deformation susceptibilityof modied and unmodied asphalt mixtures.
3. Developed models based on dynamic creep test
A. Khodaii, A. Mehrara / Construction anthat could t the creep curve and estimate its boundary points (ini-tial point of stages 2 and 3) precisely . In this model a power-law function, a linear function and an exponential function was0
0.01 0.1 1 10 100
ing 1. Comparing the permanent deformation potential of the densegraded mixtures with coarse graded mixtures and choosingthe more damageable gradation for the next stage.
2. Comparing the effect of different amounts of SBS polymer onmixtures behavior and determining the optimum polymeramount.
3. Comparing the behavior of SBS modied and unmodied mix-ture in different combinations of temperature and stress levelin dynamic creep test.
4. Deriving creep models based on Zhou model.
4.1. Samples preparation
Fig. 1 shows the aggregate grading used in this work. The twothick curves in the gure show the upper and lower limits of thepermitted grading for pavement surface layer according to localcode . The aggregates used were crushed with two brokenfaces. The applied asphalt in this research was 60/70 penetrationgrade (PG 64-16). Optimum amount of asphalt was determinedto be 5.2% for dense graded and 4.8% for coarse graded mixturesusing Marshall test, and SBS modied mixture prepared with thesame amounts of modied binder.
SBS polymer used in samples preparationwas according to Euro-prene SOLT 6302 grade and was bought in from a local petrochem-ical renery. In order to mix the polymer with asphalt, a mixer wasdesigned in which the rotation speed of electro motor was adjust-able between 30 rpm up to 1200 rpm. A potentiometer was also in-stalled to the mixer allowing the user to alter the rotation speedgradually. Moreover, the mixer was equipped with a containerwhich was embedded in an electrical heating device. The heatercould increase the temperature of the asphalt up to 200 C and
Dense Graded Coarse GradedUpper Limit of Gradation Lower Limit of GradationSieve Size (mm)
Fig. 1. Aggregate grading of asphalt mixtures.
maintain it at a constant level (with the precision of 2.5 C) for anygiven period of time. For the purpose of producing a homogenousasphaltpolymer blend, the mixing process was carried out at thetemperature of 180 C and rotation speed of 1200 rpm for an hour.
Testing samples were made according to ASTM D1559. Numberof blows in Marshall Compactor was selected by trial and error toobtain almost the same bulk specic gravity for different types ofmodied and unmodied samples, so that the inuence of volu-metric parameters on the results would decrease, and a reasonablecomparison between different mixtures could be made.
4.2. Dynamic creep test
The Dynamic Creep Tests were carried out using UTM25 to ap-ply repeated axial stress pulse to asphalt specimens measuring thevertical deformation with the Linear Variable Displacement Trans-ducer (LVDTs). In servo hydraulic UTM25 machine the stress/loadapplied to the specimen is feed back controlled allowing the oper-ator to select a loading wave shape (haversine or square pulse), apulse width duration, a rest period, a deviator stress/load to be ap-plied during each loading pulse and a contact stress/load to be ap-
2588 A. Khodaii, A. Mehrara / Construction and Buplied so that the vertical loading shaft does not lift off the testspecimen during the rest period. Prior to testing a preload stress/load can also be programmed into the testing sequence. To controlthe ambient temperature of testing samples, loading mechanism ofUTM machine is equipped by an environmental chamber.
As mentioned earlier, input data including dimensions of sam-ple (height and diameter), preload stress, deviator stress, frequencyof stress application and contact stress are controlled via an inte-grated software. In this research a square pulse wave with fre-quency of 0.5 Hz (by allocating 500 ms for pulse width and1500 ms for rest period) was chosen according to Australian codeAS 2891.12.1. A sample of pulse wave and the measured axial dis-placements of the specimen for one pulse are displayed in Fig. 2.Deformation of specimen, as it is manifested in the diagram, startsto increase by application of stress during the pulse width, and itclimbs up to its highest point as the loading is nalized. Duringthe rest period, considerable portion of deformation which isknown as resilient deformation disappears, and the remainingdeformation is considered to be permanent deformation.
Resilient modulus and creep modulus are the most importantoutputs of dynamic creep test. Based on the denitions suggestedFig. 2. A pulse wave and the measured displacement.by a number of researches and UTM25 software reference manual[10,20] resilient modulus and creep modulus are derived using Eqs.(4) and (5), respectively:
Mc rdet 5
et ee ep evet evpt 6In the equations above:
rd is the deviator stress;er is resilient deformation at a certain number of loadapplication;ep is total deformation (including elastic, visco-elastic, plasticand visco-plastic deformations) up to a number of loadapplication;
Resilient modulus, as derived from the Eq. (4), indicates thesample resistance to resilient deformations, and creep modulusindicates its resistance to permanent deformation. By the applica-tion of load to samples in each cycle, three sets of diagrams con-sisting of permanent deformation, resilient modulus and creepmodulus versus load cycles are drawn by UTM software.
4.3. Tests input data
In the present research, tests on modied and Unmodied sam-ples has been conducted in stress control mode to eliminate theprobable effects of slight differences in the sample cross sectionarea. For the purpose of assessing samples behavior under differentstress levels and ambient temperatures, tests were conducted un-der two stress levels of 100 kPa and 200 kPa and at temperatures of40 C and 50 C. Furthermore, a ve-minute preloading processincluding a static stress with a magnitude of 10% of dynamic stresswhich was selected to be applied to the specimen during dynamiccreep test was programmed into the testing sequences.
Preloading process prior to dynamic creep ensures that samplessurface and loading platen are completely in contact with eachother and the loose parts in samples surface have done their defor-mation moves, so that the deformation measured during the creeptests are related to the samples resilient and creep modulus. To in-crease the specimens temperature up to a desired amount, a pre-conditioning program also was worked out, and that was keepingsamples inside the chamber at the testing temperature over atwo-hour period before test begins. Each t...