Construction and Building Materials 61 (2014) 241–251

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier .com/locate /conbui ldmat

Evaluating the rutting resistance of asphalt mixtures using an advancedrepeated load permanent deformation test under field conditions

http://dx.doi.org/10.1016/j.conbuildmat.2014.02.0520950-0618/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +86 25 8379 3279; fax: +86 25 8379 4931.E-mail address: nifujian@gmail.com (F. Ni).

Qiang Li a,b, Fujian Ni b,⇑, Lei Gao b, Qingquan Yuan c, Yuanjie Xiao d

a School of Civil Engineering, Nanjing Forestry University, Nanjing 210037, PR Chinab School of Transportation, Southeast University, Nanjing 210096, PR Chinac Jiangsu Jinghu Expressway Co., Ltd., Huai’an 223005, PR Chinad Department of Civil & Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA

h i g h l i g h t s

� An advanced repeated load permanent deformation (ARLPD) test was developed.� The rutting resistance of asphalt mixtures was evaluated under field conditions.� The combined rutting resistance were compared for various pavements.� The test results were validated by the long term pavement performance data.

a r t i c l e i n f o

Article history:Received 7 November 2013Received in revised form 17 February 2014Accepted 21 February 2014Available online 1 April 2014

Keywords:Asphalt pavementRutting resistanceRepeated load testField conditionRutting distribution

a b s t r a c t

Current widely used rutting tests are unable to accurately simulate the working conditions in actualpavements. To address this problem, an advanced repeated load permanent deformation (ARLPD) testwas employed to evaluate the rutting resistance of asphalt pavements under field conditions. The stressstate, lateral confinement, and temperature gradient in actual pavements could be simulated in this test.It was conducted on the multi-layer specimen for eighteen different structure combinations of asphaltlayer in newly constructed pavement and rehabilitation projects at different temperatures. The ruttingresistance and distribution were evaluated using various rutting indicators. Finally, findings from labora-tory testing were validated using long term pavement performance data. It is found that the ARLPD test isrepeatable. Distinct from other rutting indicators, the Flow Number (FN) Index can accurately screen therutting resistance for different pavement structures. The SBS modified binder shows a positive effect onimproving the rutting resistance. It is also observed that the cold in-place (CIR) mixture under a good cur-ing condition could be used in the bottom asphalt layer for highway maintenance. Generally, the middleasphalt layer accumulates the greatest permanent deformation. The rutting distribution is relatively uni-form in the CIR/overlay pavement or in the newly constructed pavement where SBS modified binder isused in the middle asphalt layer.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Rutting caused by repeated loads at high temperatures is one ofthe major distresses in asphalt pavements. It is an accumulation ofthe permanent deformation mainly in the asphalt layers, especiallyin China where semi-rigid materials, such as cement stabilizedmacadam, are widely used in the pavement base [1]. Rutting notonly affects the pavement ride quality but also leads to a serioussafety issue for road users. Water pool in the ruts after rains andsnows can cause vehicles hydroplaning or uncontrollable sliding

with a high potential for traffic accidents [2]. Therefore, it is impor-tant to accurately evaluate the rutting resistance of asphalt mix-tures in the design process of pavement structure and material.

Various laboratory tests have been developed to characterizethe rutting resistance of asphalt mixtures. Generally, theyfall into two categories: empirical and fundamental [3]. Althoughthe empirical tests, such as Asphalt Pavement Analyzer (APA)and Hamburg Wheel Tracking Tester (HWTT), are readily applica-ble and can simulate the dynamic effect of field vehicle load, thefundamental engineering properties of the materials for pavementstructural design and analysis could not be obtained from the testresults [4]. The fundamental tests, such as triaxial repeated loadpermanent deformation (RLPD) test, can generate comprehensive

242 Q. Li et al. / Construction and Building Materials 61 (2014) 241–251

mechanical parameters. However, the complexity in equipmentrequirements and test procedures for lateral confinement may lim-it its application in the mix design process.

No single laboratory test has been universally accepted to havea good correlation with the field performance of asphalt pave-ments. The main reason may be that test conditions are greatly dif-ferent between laboratory and field. Laboratory conditions areinconsistent with in situ pavement working conditions. In the lab-oratory, the fixed confining pressure, uniform temperature, andsingle layer specimen are generally selected. In contrast, the con-finement of asphalt pavements in the field varies with the temper-ature, mix type, pavement depth, traffic loading, and the positionof moving wheel loading [5]. Evident temperature gradient existsin asphalt layers, especially at high temperature seasons. The com-bined rutting resistance for different structure combinations of as-phalt layer cannot be shown in laboratory either. All thesedifferences limit the accuracy and reliability of laboratory tests.Therefore, an advanced laboratory test realistically simulating fieldconditions is urgently needed to evaluate the rutting behavior ofasphalt pavements in a more reliable way.

Huang and Zhang [5] improved the conventional triaxial RLPDtest using cylinder loading plates with a little smaller size centrallyplaced on the top and bottom surface of the specimen. The lateralasphalt mixtures can provide a varying confinement. Chen et al. [6]developed a uniaxial penetration test to simulate shear stress stateand distribution in asphalt pavements. Xu and Solaimanian [7]analyzed the temperature gradient in pavements by constantlymonitoring the change of temperature with conditioning time atdifferent heights of the cylinder specimen. A temperature gradientcontrolling system for the wheel tracking test was proposed byGuan et al. [8] to simulate the pavement temperature field. Azariand Mohseni [9] conducted an incremental RLPD test to evaluateone specimen at several different temperature and stress levels.All these studies have taken a step forward in the modification oflaboratory tests according to field conditions.

Recently, an advanced repeated load permanent deformation(ARLPD) test protocol was developed by Yuan [10]. In this test, stressstate and lateral confinement in actual pavements were simulatedby designing appropriate dimensions of specimen and loading plate.The pavement temperature field was achieved using heat insulationmeasures, temperature monitoring, and conditioning test in the lab-oratory. It shows promising applicability as it makes a beneficial at-tempt to keep laboratory testing conditions consistent with theworking conditions in actual pavements. Also, both laboratory-madeand field-cored samples are applicable to this test.

2. Objectives

The main objective of this study is to evaluate the rutting resis-tance of different types of asphalt pavement in China under actualfield conditions. To accomplish this objective, the ARLPD test wasconducted on eighteen different structure combinations of asphaltlayer for newly constructed hot-mix-asphalt (HMA) pavementsand rehabilitation projects. For each structure combination, the mul-ti-layer specimen was either made in the laboratory or cored fromthe in-service pavements. The rutting resistance of various struc-tures was compared using several rutting indicators and the 3-stagemodel. The effects of the structure, material, and environment wereanalyzed. The rutting distribution in pavements was also evaluated.The laboratory findings were finally validated in the field.

3. ARLPD test method

To simulate in situ pavement working conditions as much aspossible the laboratory test should simultaneously satisfy the fol-lowing requirements:

(1) The multi-layer specimen including all asphalt layers is usedto reflect the combined rutting resistance of asphalt pave-ments and rutting contribution of each layer.

(2) The lateral confining pressure varying with the loading, tem-perature, and mixture properties is supplied by specimenitself during the testing process.

(3) The stress level and distribution in specimens are in accor-dance with those in actual pavements.

(4) The temperature gradient along the specimen height is accu-rately simulated under a given air temperature.

In the conventional triaxial RLPD test, only the fixed confiningpressure is provided due to the limits of the equipment. However,the confinement distribution is not uniform in actual pavements.Therefore, the relatively smaller loading plate was selected andthe vertical loading was applied on the central part of the specimensurface in the ARLPD test. The rest part could supply the confiningpressure which varies with the loading, temperature, and mixturetype. The cylindrical specimen of 150 mm in diameter was used. Itsheight depended on the specific pavement structure of interest.The optimum diameter of the loading plate (80.6 mm) was deter-mined using the finite element method (FEM) simulation to supplyenough confining pressure for the specimen. The shapes of hori-zontal normal stress S11 curves obtained from this test were closeto those from actual pavements, as shown in Fig. 1. The S11 valuesin the laboratory specimens were about 80% of those in actualpavements since the specimen is lack of constraint caused by thelower layers, such as base, subbase, and subgrade. Although thereare some differences between the ARLPD test and actual pavement,a smaller size of the loading plate cannot be accepted. Because thediameter in 80.6 mm was larger than three times of nominal max-imum aggregate size (NMAS) for most asphalt mixtures typicalused in China. It can eliminate the size effect which will introducemuch greater errors.

A transient pavement temperature field FEM model based onheat transfer theory was employed to predict the hourly tempera-ture at each pavement depth. The pavement temperature Tpave val-ues at different depths under the maximum air temperature Tair inthe area were selected as target testing temperatures for differentheights of the specimen in the ARLPD test. To obtain the calculatedtemperature gradient, the heat-insulated paint and rock wool pilewere used on the bottom and side of the cylinderal specimen. Onlythe top can transfer external heat. A conditioning and temperaturemonitoring test was performed to measure the temperature varia-tion with time. Three thermocouples were sealed in the specimenat the mid-depth of the top, middle, and bottom asphalt layers allalong the center line, respectively. The Tpave at pavement surfacecorresponding to a given Tair obtained from FEM calculation waschosen as the conditioning temperature in the environmentalchamber. Actually, the target testing temperatures at differentheights of the specimen cannot be simultaneously reached. As a re-sult, the minimum public time interval to approximately obtaintarget testing temperatures (±1 �C) for different asphalt layerswas used for testing. The process of determining testing conditionsfor temperature gradient simulation is shown in Fig. 2. More de-tails about the test method could be found elsewhere [10].

4. Experimental program

4.1. Materials

Eighteen structure combinations of the asphalt layer widely used for highwayconstruction and rehabilitation projects in Jiangsu, China were tested in this study,as shown in Table 1. The first eight ones (A1–D2) were used for newly constructedHMA pavements. Other ten ones (E1–G1) including the techniques of cold in-placerecycling (CIR), milling, overlay, and performance evaluation were used for pave-ment rehabilitations.

0

3

6

9

12

15

18

0 100 200 300 400

Dep

th (

cm)

S11 (kPa)

0.7 MPa loading-ARLPD0.7 MPa loading-pavement0.88 MPa loading-ARLPD0.88 MPa loading-pavement1.1 MPa loading-ARLPD1.1 MPa loading-pavement

Pavement model ARLPD test model

Fig. 1. Comparison of the S11 values calculated by different FEM models [10].

0

3

6

9

12

15

18

10 20 30 40 50 60 70

Dep

th (

cm)

Tpave ( )

Tair 22 °C

Tair 25 °C

Tair 28 °C

Tair 31 °C

Tair 34 °C

Tair 37 °C

Tair 40 °C

010203040506070

19 22 25 28 31 34 37 40 43

Tpa

ve(

)

Tair( )

0 cm 2 cm

8 cm 14 cm

0

10

20

30

40

50

60

0 200 400 600 800

Tem

pera

ture

()

Time (min)

2cm

8cm

14cm

39±1°C44±1°C

380 530

50±1°C

Fig. 2. Determination of testing conditions for temperature gradient simulation[10].

Q. Li et al. / Construction and Building Materials 61 (2014) 241–251 243

All of them consisted of two or three asphalt layers (top, middle, and bottom). Insummary, eight types of HMA mixture were designed, including stone mastic as-phalt (SMA), Superpave (Sup), and dense graded asphalt course (AC) mixtures withdifferent NMAS. Two asphalt binders used were 70# PG 64-22 and styrene–butadi-ene–styrene (SBS) modified PG 76-22. Besides, CIR mixtures with reclaimed asphaltpavement of 100%, emulsified asphalt of 3.5%, water of 4.3%, and different contentsof cement (1.5%, 2.0%, and 2.5%) were also involved. A granular anti-rutting agentwas selectively added in some mixtures of the middle asphalt layer for comparison.The target air void contents were 3.5% ± 0.5% for SMA mixtures, 10.0% ± 0.5% for CIRmixtures, and 4.0% ± 0.5% for other mixtures. The core samples of entire asphalt lay-ers from an old pavement (4 cm AC-13 + 6 cm AC-20 + 8 cm AC-25) were also em-ployed for analysis. The aggregate gradations and mix design results could befound from Table 2.

4.2. Specimen fabrication

The cylindrical specimen of 150 mm in diameter was first compacted using aSuperpave gyration compactor (SGC) with the height controlled mode for each as-phalt layer. For the cylindrical SGC specimen, the air void in the outer area is gen-erally 2%–3% higher than that in the inner area [11]. Because the loading is onlyapplied to the inner area in the ARLPD test, a little higher air void (1%–2% higherthan the target one) was designed for the whole specimen to ensure the targetair void of the inner area. The CIR specimens should be placed in the oven of60 �C with blast for 48 h after curing 12 h at the room temperature of 25 �C. Itwas beneficial for accelerating the build-up of the mixture strength. The detailedmix design and sample preparation method for the CIR mixture can be found else-where [12].

Then, both ends were sawed from it to obtain target height to provide a smoothand parallel surface. The specimens of different asphalt layers were bonded to-gether by emulsified asphalt to obtain the entire specimen for testing. The specimenwas conditioned at room temperature for 4 h to strengthen the bonding effect. Dou-ble layers of heat-insulated paint were painted on the bottom and side of the spec-imen. The thickness of the first layer was 0.3 mm. The second layer of 0.5 mm waspainted again after curing 36 h with blast at the temperature of 25 �C. It also neededto be cured for 48 h at the same condition. Finally, a rock wool pile with 150 mm ininner diameter and 156 mm in outer diameter was used to enwrap the specimenside for further reducing heat transfer. It must ensure the heat insulation betweenenvironment and specimen surface except the top one. In addition, it was provedthat the use of these heat insulation measures had little effect on the mixture per-formance [10]. Three replicates were conducted for each test. The process of spec-imen fabrication is shown in Fig. 3.

4.3. Test setup

The ARLPD test as shown in Fig. 4 was performed on a UTM-25 servohydraulictesting system under the conditions corresponding to different Tair values. A haver-sine axial compressive load pulse (load duration of 0.1 s and rest period of 0.9 s) of0.7 MPa with a contact stress of 35 kPa was directly applied to the loading plate of80.6 mm in diameter. Temperature gradient simulation was achieved using thecontrol conditions shown in Table 3. Experimental plan could also be found inthe table. Tests were continued up to 7000 load cycles or permanent strain of 5%,whichever came first. A cement stabilized macadam (CSM) pad was put underthe specimen to prevent heat transfer from bottom steel platen.

5. Test results and analysis

A series of rutting indicators were calculated based on theARLPD test results. Flow number (FN), defined as the cycle numberat which the tertiary flow started, was obtained using the Hoerlmodel after data smoothing with roughness penalty method [13].Besides, the accumulated permanent strain at the tertiary flow ep(-FN), accumulated permanent strain at testing terminal epf, and FNindex (a ratio of ep(FN) to FN) proposed by Walubita et al. [14] wereintroduced for comparison. The following Zhou’s 3-stage modelwas also employed to characterize the permanent deformationbehavior of asphalt mixtures [15]. A brief summary of the resultswas listed in Table 4.1st stage:

1st stage : ep ¼ aNb; N < N1st ð1Þ

2nd stage : ep ¼ e1st þ cðN � N1stÞ; N1st 6 N < N2nd ð2Þ

Table 1Structure combinations of asphalt layer.

No. Top Middle Bottom

A1 4 cm SMA-13 (PG 76-22) 6 cm AC-20 (PG 76-22) 8 cm AC-25 (PG 64-22)A2 4 cm SMA-13 (PG 76-22) 6 cm AC-20 (PG 64-22) 8 cm AC-25 (PG 64-22)B1 4 cm SMA-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) 8 cm Sup-25 (PG 64-22)B2 4 cm SMA-13 (PG 76-22) 6 cm Sup-20 (PG 64-22) 8 cm Sup-25 (PG 64-22)C1 4 cm AC-13 (PG 76-22) 6 cm AC-20 (PG 76-22) 8 cm AC-25 (PG 64-22)C2 4 cm AC-13 (PG 76-22) 6 cm AC-20 (PG 64-22) 8 cm AC-25 (PG 64-22)D1 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) 8 cm Sup-25 (PG 64-22)D2 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 64-22) 8 cm Sup-25 (PG 64-22)E1 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) + anti-rutting agent 12 cm CIR (1.5% cement)E2 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) 12 cm CIR (1.5% cement)E3 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) + anti-rutting agent 12 cm CIR (2.0% cement)E4 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) 12 cm CIR (2.0% cement)E5 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) + anti-rutting agent 12 cm CIR (2.5% cement)E6 4 cm AC-13 (PG 76-22) 6 cm Sup-20 (PG 76-22) 12 cm CIR (2.5% cement)E7 6 cm AC-13 (PG 76-22) 12 cm CIR (1.5% cement)F1 4 cm AC-13 (PG 76-22) + 2.5 cm AC-13 (PG 76-22) (old pavement) 6 cm AC-20 (PG 64-22) (old pavement) 7 cm AC-25 (PG 64-22) (old pavement)F2 4 cm AC-13 (PG 76-22) 12 cm CIR (2.0% cement) (old pavement) 3 cm AC-25 (PG 64-22)G1 6 cm AC-20 (PG 64-22) (old pavement) 7 cm AC-25 (PG 64-22) (old pavement)

Table 2Aggregate gradations and mix design results.

Mixtures SMA-13 AC-13 AC-20 Sup-20 AC-25 Sup-25 CIROptimal asphalt content (%) 5.7 (SBS) 4.8 (SBS) 4.1 (70#) 4.2 (SBS) 4.1 (70#) 4.1 (SBS) 3.6 (70#) 3.8 (70#) 3.5 (emulsified)

Sieve size (mm) Passing percent (%)

31.5 100.0 100.0 100.0 100.0 100.0 100.0 100.026.5 100.0 100.0 100.0 100.0 95.5 97.0 99.519 100.0 100.0 96.1 93.0 78.7 77.6 95.116 100.0 100.0 89.6 87.0 73.3 66.9 87.413.2 94.0 94.4 71.6 81.3 63.5 57.8 76.79.5 64.5 66.9 58.1 70.8 56.0 48.7 55.64.75 26.8 41.8 43.0 56.1 38.4 37.9 42.02.36 22.6 31.3 26.7 30.3 26.9 23.1 31.01.18 18.3 22.8 18.6 20.5 18.5 16.1 21.40.6 14.7 16.5 15.0 12.5 14.8 10.4 12.70.3 11.9 10.9 9.8 8.1 9.1 7.0 7.50.15 10.7 8.3 8.6 6.4 8.0 5.7 5.30.075 9.4 5.3 5.7 4.6 4.9 4.2 3.5

(a) compaction (c) bonding

(d) conditioning

(e) painting

(f) enwrapping

(b) sawing

Fig. 3. Specimen fabrication.

244 Q. Li et al. / Construction and Building Materials 61 (2014) 241–251

3rd stage : ep ¼ e2nd þ dðef ðN�N2ndÞ � 1Þ; N P N2nd ð3Þ

where ep is accumulated permanent strain (le), N is load cycle, e1st

is accumulated permanent strain during the first stage (le), N1st isload cycle corresponding to the end point of the first stage, e2nd isaccumulated permanent strain during the second stage (le), N2nd

is load cycle corresponding to the end point of the second stage(FN), a, b, c, d, and f are model coefficients.

The curves of accumulated permanent strain ep against the loadcycle N for eighteen structure combinations of asphalt layer at dif-ferent temperatures are illustrated in Figs. 5–7. It is found fromthese figures and Table 4 that the ARLPD test can generate a suffi-

Loading plate

CTB pad

80.6 mm

Top asphalt layer

Middle asphalt layer

Bottom asphalt layer

Specimen

Heat-insulated paint

Rock wool pile

Fig. 4. ARLPD test setup.

Table 3Testing conditions for temperature gradient simulation.

Tair

(�C)Air conditionertemperature (�C)

Environmental chambertemperature (�C)

Starting(min)

25 25 39 32031 29 49 36040 31 60 490

Table 4Summary of test results.

No. Tair (�C) 3-Stage model parameters

1st 2nd 3rd

a b c d f

A1 25 374 0.353 1.040 – –31 1254 0.296 2.287 – –40 1668 0.331 5.232 1152 0.001

A2 25 223 0.374 0.924 – –31 1052 0.335 2.964 – –40 2214 0.324 7.216 975 0.001

B1 25 409 0.336 0.847 – –31 1684 0.262 1.843 – –40 3502 0.248 4.299 810 0.000

B2 25 469 0.339 1.166 – –31 1349 0.358 5.191 867 0.00140 1325 0.433 14.790 1220 0.003

C1 25 621 0.292 0.999 – –31 624 0.337 1.796 – –40 4017 0.218 3.855 842 0.001

C2 25 513 0.302 0.988 – –31 630 0.377 2.883 752 0.00040 2750 0.296 6.655 766 0.001

D1 25 443 0.335 1.133 – –31 1015 0.302 1.776 – –40 2879 0.245 3.435 651 0.001

D2 25 318 0.395 1.469 – –31 1277 0.354 4.689 1127 0.00140 2258 0.365 12.778 1445 0.002

E1 40 5542 0.187 2.188 1540 0.001E2 40 4744 0.204 3.462 688 0.001E3 40 5188 0.153 1.171 – –E4 40 6474 0.145 1.361 – –E5 40 3600 0.176 1.084 – –E6 40 4194 0.167 1.226 – –E7 40 1933 0.362 5.858 785 0.001F1 40 8372 0.219 12.427 1255 0.003F2 40 7700 0.204 10.089 1090 0.002G1 31 4859 0.255 11.234 1389 0.001

Q. Li et al. / Construction and Building Materials 61 (2014) 241–251 245

cient amount of permanent deformation for evaluating the ruttingresistance within a relatively short period (about 2 h). It also cancapture the rutting characteristics of all three stages for most pave-ments at a higher temperature (e.g., Tair = 40 �C). Therefore, theARLPD test is reasonable and applicable from the viewpoint ofcost-effectiveness.

5.1. Test variability

The rutting indicators presented in Table 4 were the average val-ues of three replicate specimens. The coefficient of variation (COV)was calculated for these indicators to identify the variability of theARLPD test, as shown in Fig. 8. It is observed in the figure that theCOV values of FN, ep(FN), epf, and FN Index for different types of pave-ment were 5.5%–27.8% (average 14.2%), 8.0%–33.5% (average 15.6%),3.4%–20.9% (average 10.3%), and 11.9%–35.1% (average 20.3%),

time End time(min)

Tested mixtures

470 A1, A2, B1, B2, C1, C2, D1, D2490 A1, A2, B1, B2, C1, C2, D1, D2, G1580 A1, A2, B1, B2, C1, C2, D1, D2, E1, E2, E3, E4, E5, E6, E7,

F1, F2

N1st FN (N2nd) ep(FN) (le) FN Index epf (le)

1620 >7000 >10,500 1.5 10,5001540 >7000 >22,650 3.2 22,650

9 1470 4640 36,570 7.9 >50,000

1220 >7000 >8350 1.2 83501480 >7000 >29,110 4.2 29,110

4 1100 2110 28,970 13.7 >50,000

1600 >7000 >9180 1.3 91801560 >7000 >19,820 2.8 19,820

7 1310 3000 28,700 9.6 45,500

2100 >7000 >12,110 1.7 12,1100 1320 3300 28,160 8.5 >50,0000 830 1680 36,930 22.0 >50,000

1230 >7000 >10,430 1.5 10,4301450 >7000 >17,630 2.5 17,630

1 1770 5120 34,210 6.7 42,940

1280 >7000 >10,240 1.5 10,2408 1550 3100 14,680 4.7 30,3607 980 1990 28,010 14.1 >50,000

1310 >7000 >11,130 1.6 11,1301970 >7000 >17,760 2.5 17,760

0 1520 4000 25,760 6.4 38,490

1470 >7000 >14,240 2.0 14,2400 990 3450 26,580 7.7 >50,0005 810 1520 34,950 23.0 >50,000

2 1700 6500 28,300 4.4 31,4600 1450 3850 28,840 7.5 41,200

1770 >7000 >20,370 2.9 20,3701830 >7000 >23,880 3.4 23,8801560 >7000 >17,040 2.4 17,0401720 >7000 >18,890 2.7 18,890

8 1560 4420 43,490 9.8 >50,0001 650 1120 40,395 36.1 >50,0006 730 1530 36,778 24.0 >50,0009 560 1210 35,325 29.2 >50,000

0

5000

10000

15000

20000

0 2000 4000 6000 8000

p(µ

)

N

A1 A2 B1 B2

C1 C2 D1 D2

(a) Tair = 25 °C

0

10000

20000

30000

40000

50000

0 2000 4000 6000 8000

p(µ

)

N

A1 A2

B1 B2

C1 C2

D1 D2

(b) Tair = 31 °C

0

10000

20000

30000

40000

50000

0 2000 4000 6000 8000

p(µ

)

N

A1 A2

B1 B2

C1 C2

D1 D2

(c) Tair = 40 °C

Fig. 5. Permanent strain curves for newly constructed HMA pavements (A1–D2).

0

10000

20000

30000

40000

50000

0 2000 4000 6000 8000

p(µ

)

N

E1 E2 E3 E4

E5 E6 E7

Tair = 40 °C

Fig. 6. Permanent strain curves for CIR/overlay rehabilitations (E1–E7).

0

10000

20000

30000

40000

50000

0 500 1000 1500 2000 2500 3000

p(µ

)

N

F1

F2

G1

Tair = 40 °C

Tair = 31 °C

Fig. 7. Permanent strain curves for milling/overlay rehabilitations (F1, F2, and G1).

0

10

20

30

40

50

FN p(FN) pf FN Index

CO

V (

%)

Fig. 8. COV values of different rutting indicators.

246 Q. Li et al. / Construction and Building Materials 61 (2014) 241–251

respectively. Overall, this level of variability is deemed acceptable forthe laboratory testing at high temperatures, indicating the fairly goodrepeatability and reproducibility of the ARLPD test.

5.2. Evaluation of rutting indicators

The numbers in ranking order of structure combinations by therutting resistance based on different indicators are presented in

Table 5. The epf value obtained from the laboratory test was treatedas a benchmark to evaluate the accuracy of other rutting indica-tors. All structures with FN > 7000 were numbered as ‘‘1’’ in therank based on FN. The structures with epf > 50,000 le were com-pared by load cycles upon test termination in the rank based on epf.

It is found from Table 5 that little difference in the rutting resis-tance is found among various pavement structures at a lower Tair of25 �C. Therefore, only the results at the Tair of 31 �C and 40 �C areused for comparison. The performance rankings based on the FNIndex and epf are almost the same while those based on the FNand ep(FN) do not have similar trends. It is proved that the FN Indexcomputed from the FN and ep(FN) can offer a better potential as theindicator to differentiate the rutting resistance of asphalt pave-ments in the laboratory than the single FN or ep(FN) [14]. Also, it sig-nifies that the ARLPD test is able to provide rational ruttingindicators for mixture differentiation.

5.3. Effect of temperature

It is found from Table 4 and Fig. 5 that the permanent deforma-tion in asphalt pavements significantly increases with tempera-ture. As mentioned before, the difference in the rutting resistancefor various HMA pavements is not clear at a lower Tair (e.g.,25 �C), as shown in Fig. 5a. For example, using the same aggregategradation structure A1 with SBS modified binder (PG 76-22) irra-tionally accumulates more permanent strain than structure A2with unmodified binder (PG 64-22). It indicates that sample tosample variation may have a greater effect than material proper-ties under this condition. Also, all structures only undergo the firstand second stage of permanent deformation (FN > 7000). No ter-tiary stage occurs. Therefore, it is not able to accurately identifythe rutting resistance of different structure combinations underthis condition.

When the Tair reaches or exceeds 31 �C, the development of per-manent strain grows faster. Even a small change in temperature

Table 5Ranking of various structures in terms of the rutting resistance (from high to low).

No. FN ep(FN) FN Index epf

25 �C 31 �C 40 �C 25 �C 31 �C 40 �C 25 �C 31 �C 40 �C 25 �C 31 �C 40 �C

A1 1 1 7 1 1 13 3 4 9 5 4 10A2 1 1 11 1 1 10 1 5 12 1 5 12B1 1 1 12 1 1 8 2 3 10 2 3 9B2 1 7 14 1 8 15 7 8 14 7 8 14C1 1 1 6 1 1 11 3 1 7 4 1 8C2 1 8 13 1 6 6 3 6 13 3 6 13D1 1 1 9 1 1 5 6 1 6 6 2 6D2 1 6 16 1 7 12 8 7 15 8 7 15E1 – – 5 – – 7 – – 5 – – 5E2 – – 10 – – 9 – – 8 – – 7E3 – – 1 – – 1 – – 3 – – 3E4 – – 1 – – 1 – – 4 – – 4E5 – – 1 – – 1 – – 1 – – 1E6 – – 1 – – 1 – – 2 – – 2E7 – – 8 – – 17 – – 11 – – 11F1 – – 17 – – 16 – – 17 – – 17F2 – – 15 – – 14 – – 16 – – 16G1 – 9 – – 9 – – 9 – – 9 –

Q. Li et al. / Construction and Building Materials 61 (2014) 241–251 247

will have a great effect on the pavement performance. The differ-ences among various pavements are also fully reflected, as shownin Fig. 5b and c. For example, the accumulated permanent defor-mation in structure D2 is only 1.3 times of that in structure D1at the Tair of 25 �C. However, the ratios increase to 2.9 and 3.4 atthe Tair of 31 �C and 40 �C, respectively. Some structures (e.g., B2,C2, and D2) exhibits obvious tertiary shear flow at the Tair of31 �C while at the Tair of 40 �C all HMA pavements enter the tertiarystage. Therefore, it is suggested that the rutting resistance of as-phalt mixtures and pavement structures should be evaluated un-der a higher temperature condition (e.g. Tair P 31 �C).

0

10000

20000

30000

40000

50000

0 2000 4000 6000 8000

p(µ

)

N

AC-13(PG 76-22) SMA-13(PG 76-22)AC-20(PG 76-22) AC-20(PG 64-22)Sup-20(PG 76-22) Sup-20(PG 64-22)AC-25(PG 64-22) Sup-25(PG 64-22)

Fig. 9. Results of the flow number test.

5.4. Comparison of different pavement structures

5.4.1. Newly constructed HMA pavement (A1–D2)Generally, the smaller magnitude of epf means the greater rut-

ting resistant of the pavement. Therefore, it is seen in Table 5 thatthe ranking (from high to low) in terms of the rutting resistance fornewly constructed HMA pavements is: D1 > C1 > B1 > A1 > A2 >C2 > B2 > D2.

When the materials used in the middle asphalt layer are com-pared, it is found from Fig. 5 that all structures with SBS modifiedbinder (A1, B1, C1, and D1) have much higher rutting resistancethan those with unmodified binder (A2, B2, C2, and D2), especiallyat higher temperatures. In addition, the structures with aggregategradations of Sup-20 + Sup-25 (B1, B2, D1, and D2) and AC-20 + AC-25 (A1, A2, C1, and C2) provide almost equivalent ruttingresistance when the SBS modified binder is used in middle asphaltlayer. However, when the unmodified binder is used, the accumu-lated permanent strains in the former ones (B2 and D2) are 1.6–2.1times larger than those in the latter ones (A2 and C2) at the Tair of31 �C and 40 �C. It should be mentioned that under these test andtemperature conditions both the binder type and content have ahuge impact on the rutting performance of the mixtures.

By comparing the materials used in the top asphalt layer, it isobserved that the structures with aggregate gradations of SMA-13 show similar or even worse rutting resistance than those withAC-13. Generally, it is widely accepted that the SMA gap gradationshould exhibit excellent rutting resistance due to its good skeletonstructure and interlocking effect between stones [16]. In this study,the use of a lower binder content of 4.8% and a better basalt fibermay have major contribution to the superior performance of AC-13 mixtures as compared to the SMA-13 mixture with a higher bin-der content of 5.7% and a conventional polyester fiber.

To validate the ARLPD test results, the flow number test, onecandidate of Simple Performance Test for evaluating the ruttingpotential, was also conducted using the same UTM-25 system forcomparison on the mixtures used in structures A1–D2. The devia-toric stress of 0.7 MPa, confining pressure of 70 kPa, and the tem-perature of 50 �C were selected as the uniform testing conditions.The methods of sample preparation and test procedure can befound elsewhere [13]. It can be found from Figs. 5 and 9 that theabove trends observed from the ARLPD test can also be obtainedfrom the flow number test. For instance, the AC-13 and SMA-13mixtures used in the top asphalt layer accumulated a comparablepermanent strain. The mixtures with the PG 76-22 binder (SBS)exhibited a much better performance than those with the PG 64-22 binder (70#). The Superpave and dense graded mixtures had asimilar rutting resistance. The consistent results from the two testsindirectly confirm that the ARLPD has a good potential for evaluat-ing the rutting resistance for different mixtures.

5.4.2. CIR/overlay rehabilitation (E1–E7)CIR has been gradually accepted and applied in China for pave-

ment rehabilitation applications, mainly as the bottom asphaltlayer. The effects of cement content, anti-rutting agent, and overlaythickness of the upper layer on the rutting resistance were evalu-ated for CIR/overlay pavements. Comparing Figs. 5 and 6, it is ob-served that these structures show similar or even better ruttingresistance than HMA pavements. The permanent strains for Struc-tures E3–E6 are only less than half of those for A1–D2. Besides theuse of the thicker CIR layer (12 cm), another reason may be that theCIR mixture curing and temperature field simulation process were

0

10000

20000

30000

40000

50000

0

2

4

6

8

10

12

1.0 1.5 2.0 2.5 3.0

pf (

µ)

FN

Inde

x

Cememt Content (%)

FN Index (With Anti-rutting Agent)

FN Index (Without Anti-rutting Agent)

pf (With Anti-rutting Agent)

pf (Without Anti-rutting Agent)

Fig. 10. Effect of cement content on the rutting resistance for CIR/overlaypavements.

0%

10%

20%

30%

40%

50%

60%

A1 A2 B1 B2 C1 C2 D1 D2

Rut

ting

Con

trib

utio

n

Structure Combination

Top Middle Bottom

(a) Newly constructed HMA pavements

0%

10%

20%

30%

40%

50%

60%

E1 E2 E3 E4 E5 E6 E7R

uttin

g C

ontr

ibut

ion

Structure Combination

Top Middle Bottom

(b) CIR/overlay rehabilitations

0%

10%

20%

30%

40%

50%

60%

70%

F1 F2 G1

Rut

ting

Con

trib

utio

n

Structure Combination

Top Middle Bottom

(c) Milling/overlay rehabilitations

Fig. 11. Comparison of the rutting contribution ratio for different layers

248 Q. Li et al. / Construction and Building Materials 61 (2014) 241–251

carried out at high temperatures. Therefore, the cement hydrationprocess could keep enhancing the strength and deformation resis-tance of CIR mixtures at this laboratory condition. It indicates thatthe CIR mixture preliminary shows its feasibility and reliability asthe bottom asphalt layer for highway rehabilitation applications.

It is also found that the rutting resistance of CIR pavements in-creases with cement content, as shown in Fig. 10. However, theimprovement seems insignificant when the cement content in-creases from 2.0% to 2.5%. Moreover, adding too much cement(e.g., 2.5%) to the CIR mixtures will not only significantly reducecracking resistance but also raise construction cost. Therefore,the optimum cement content should be determined after a com-prehensive consideration of all aspects of the pavementperformance.

Adding anti-rutting agent in the middle asphalt layer (E1, E3,and E5) can improve the rutting resistance of the entire structure.For example, the structure E2 without anti-rutting agent quicklyreaches the shear flow stage and presents a 2–3 times higherdevelopment rate of the permanent deformation than structureE1. Also, it should be noted that its effect is less significant thanthat of cement content.

By comparing the structures of E2 and E7, it is seen that struc-ture E7 with a single layer overlay of 6 cm on the CIR layer accu-mulates much more permanent strain than structure E2 withdouble layers overlay of 10 cm (4 cm + 6 cm). Although almostequivalent rut depths could be obtained from two pavementswhen different layer thicknesses are considered, the single layeror thinner overlay is not recommended for heavy-duty highwaysdue to its lower resistance to the reflective cracking.

(Tair = 40 �C).

5.4.3. Milling/overlay rehabilitation (F1, F2, and G1)After testing samples (G1) of the middle and bottom asphalt

layers (6 cm AC-20 + 8 cm AC-25) cored from an old pavement,two milling/overlay plans were designed for strengthening pave-ment. One option (F1) was to mill 2.5 cm top asphalt layer (AC-13) of the old pavement and then to directly overlay 4 cm AC-13new asphalt layer on top of it. The other option (F2) was to firstmill 4 cm top asphalt layer (AC-13). Then, 6 cm middle asphaltlayer and 4 cm bottom asphalt layer were treated using CIR tech-nique. Only 3 cm bottom asphalt layer (AC-25) was left intact. Fi-nally, 4 cm AC-13 new asphalt layer was overlaid on the CIR layer.

It is observed from Fig. 7 and Table 4 that the rutting resistanceof the old asphalt pavement has seriously deteriorated. Although alower Tair of 31 �C was selected for structure G1 during the test tosimulate the in situ condition in the middle and bottom asphaltlayers, the FN Index reaches up to 29.2, which is much greater thanany newly constructed HMA pavement tested at the Tair of 40 �C. Asa result, the strengthening plan that uses directly milling and over-lay (F1) cannot effectively improve the rutting resistance of the en-tire pavement structure. In contrary, structure F2 with CIRtechnique shows a better performance. Its FN Index is close to

those of some newly constructed HMA pavements (e.g., structureB2 and D2), indicating that lower layers should be carefully pre-pared prior to placing the overlay.

5.5. Rutting distribution

The variation of thickness in each asphalt layer was measuredafter testing to study the rutting distribution. The ratio of eachlayer was calculated, as shown in Fig. 11. Only the results at the Tair

of 40 �C were used for analysis since the smaller magnitudes ofpermanent deformation produced at other lower temperaturesmay cause higher relative measurement errors.

For newly constructed HMA pavements, it is found from Fig. 11athat the middle asphalt layer undergoes the largest permanentdeformation (36.2%–53.5%) followed by the bottom asphalt layer(27.2%–34.9%). The top asphalt layer contributes the least(19.3%–29.4%) to the rutting. Considering the different thicknessesbetween top (4 cm), middle (6–7 cm), and bottom (8–12 cm) as-phalt layers, it is concluded that the middle asphalt layer will

Yan

hai

Hig

hw

ay

N3

Y1

Y2

Y3

Y4

47.2

16.4

16.8

17.4

19.2

Nov

embe

r20

05A

K-1

3(P

G76

-22)

AK

-13

(PG

76-2

2)A

K-1

3(P

G76

-22)

SMA

-13

(PG

76-2

2)SM

A-1

3(P

G76

-22)

AC

-20

(PG

76-2

2)A

C-2

0(P

G76

-22)

AC

-20

(PG

76-2

2)A

C-2

0(P

G76

-22)

AC

-20

(PG

76-2

2)A

C-2

5(P

G64

-22)

AC

-25

(PG

64-2

2)A

C-2

5(P

G64

-22)

AC

-25

(PG

64-2

2)A

C-2

5(P

G64

-22)

22

14

31

21

43

Q. Li et al. / Construction and Building Materials 61 (2014) 241–251 249

experience the greatest compression and shear deformation in atypical semi-rigid base asphalt pavement in China, which agreeswith the theoretical analysis result [17].

Also, the three asphalt layers in the structures where SBS mod-ified binder was used in the middle asphalt layer (A1, B1, C1, andD1) exhibit similar magnitudes of permanent deformation. In otherwords, the rutting distribution in these structures is relatively uni-form. For other structures using unmodified binder (A2, B2, C2, andD2), the middle asphalt layer accumulates more permanent defor-mation, around half of the total amount in all asphalt layers.

It is known that the rut depth in the middle asphalt layer willgreatly increase if the obvious difference in the mixture perfor-mance exists between the top and middle asphalt layers [10].Therefore, the rutting resistance of asphalt mixtures used in themiddle asphalt layer should be improved by the use of high qualitymaterials, such as modified binder, anti-rutting agent, polyester fi-ber, or basalt fiber. In Jiangsu of China, SBS modified binder hasbeen widely used in both top and middle asphalt layers (in somepavement sections even the bottom asphalt layer is included) innewly constructed highway pavements since ten years ago. Theengineering practice shows that it can effectively minimize the rut-ting problems.

It is observed in Fig. 11b that the rutting contribution ratios ofthree asphalt layers are similar for the CIR/overlay pavements.One reason is that SBS modified binder is used in the middle as-phalt layer. Another reason is that the CIR mixture under a goodcuring condition has an excellent rutting resistance as mentionedpreviously. They effectively strengthen the middle and bottom as-phalt layers of the pavement structure. The rutting contribution ra-tio of the CIR (bottom) layer deceases with increasing cementcontent or with the addition of anti-rutting agent. For structureE7 more permanent deformation is produced in the single thinoverlay layer on top of the CIR layer.

For the milling/overlay pavements as shown in Fig. 11c, the oldpavement still contributes the most to the rutting in structure F1,indicating that it could not provide a good foundation for directoverlay. By treating it with the CIR technique, the deformation ismainly undertaken by the new overlay and CIR layers in structureF2.

Tabl

e6

Info

rmat

ion

onse

lect

edpa

vem

ent

sect

ions

.

Sect

ion

Nin

ghan

gH

igh

way

N1

N2

Len

gth

(km

)4.

215

.1O

pen

totr

affi

cSe

ptem

ber

2004

Top

Asp

hal

tLa

yer

SMA

-13

(PG

76-2

2)A

K-1

3(P

G76

-22)

Mid

dle

Asp

hal

tLa

yer

AC

-20

(PG

76-2

2)A

C-2

0(P

G76

-22)

Bot

tom

Asp

hal

tLa

yer

AC

-25

(PG

64-2

2)A

C-2

5(P

G64

-22)

Bas

esu

bbas

eC

emen

tst

abil

ized

mac

adam

Lim

e-fl

yas

hso

ilR

anki

ng

inte

rms

ofth

eru

ttin

gre

sist

ance

(fro

mh

igh

tolo

w)

Fiel

d1

3La

bora

tory

23

6. Field validation

A relative comparison was made between the ARLPD test re-sults and field rutting measurements. Seven sections from twopavements (Ninghang and Yanhai Highways) in Jiangsu of Chinawere selected for validation purpose. All sections had the samepavement structure in terms of layer thickness and materials ofbase and subbase. A total of 18 cm (4 cm + 6 cm + 8 cm) asphaltconcrete was placed in three layers over 36–40 cm CSM base andlime-fly ash soil subbase. The primary difference was that differentmixtures were used in three asphalt layers. Details regarding pave-ment structures and construction information are provided in Ta-ble 6. The non-skid asphalt course AK-13 listed in the table is acoarse dense graded mixture that has been widely used in thetop asphalt layer in China for a decade. As compared to the AC-13 mixture, the AK-13 mixture is designed with a little higher con-tent of coarse aggregate to improve the texture depth and skidresistance. Fortunately, for the middle and bottom asphalt layersthe same mixtures were used in both laboratory and field. Rutdepths as shown in Fig. 12 were average values for each whole sec-tion that were periodically measured by a 3 m-straightedge in theearly years and later by a high speed APRES laser profilometer atintervals of 10 m. Only the data prior to significant rehabilitationsor maintenance activities was used.

0

5

10

15

20

25

0 10 20 30 40 50

Time (month)

N1 N2 N3

(a) Ninghang highway

0

5

10

15

20

0 10 20 30 40 50

Ave

rage

Rut

Dep

th (

mm

)A

vera

ge R

ut D

epth

(m

m)

Time (month)

Y1 Y2 Y3 Y4

(b) Yanhai highway

Fig. 12. Development of the rut depth measured in the field.

250 Q. Li et al. / Construction and Building Materials 61 (2014) 241–251

The three sections from Ninghang Highway were employed tocompare different types of aggregate gradation for the top asphaltlayer and different types of binder for the middle asphalt layer. Itcan be seen in Fig. 12a that section N3 with SBS modified binderin the middle asphalt layer is much more rut-resistant than sectionN2 with unmodified binder after being subject to similar traffic andclimate conditions for four years. The rut depth in section N2 ismore than 20 mm and continues to rapidly increase. It can be ex-pected that section N2 reaches the tertiary flow stage. The ruttingin section N1 and N3 are comparable.

The four sections from Yanhai Highway were selected to evalu-ate various structure combinations of asphalt layer. It is found fromFig. 12b that the sections with Sup mixtures (Y2 and Y4) show asimilar rutting resistance than those with AC mixtures (Y1 andY3) under the same field conditions. Also, the sections with SMA-13 (Y3 and Y4) and dense graded AK-13 (Y1 and Y2) mixtures accu-mulate almost equivalent permanent deformation. The differencescaused by aggregate gradations were not significant since SBSmodified binder was used in both top and middle asphalt layersfor all sections.

Finally, the similar performance rankings were obtained fromthe ARLPD test and field sections, as shown in Table 6. Althoughto some extent the field investigation findings are in accordancewith the ARLPD test results, the database used in this study is stilllimited. The further validation research using more acceleratedloading test and long-term pavement performance data isrequired.

7. Conclusions

Important conclusions made in this study are summarized asfollows:

(1) The ARLPD test method can simulate the stress state, lateralconfinement, and temperature gradient in actual pavements.Moreover, it is repeatable and reproducible.

(2) Among the rutting indicators obtained from the ARLPD test,FN Index (the ratio of ep(FN) to FN) can accurately identify theranking order of the rutting damage among various struc-ture combinations of asphalt layer.

(3) The difference in the rutting potential among various pave-ment structures becomes more noticeable under a highertemperature (e.g. Tair P 31 �C).

(4) For the newly constructed HMA pavement, the structuresusing SBS modified binder in the middle asphalt layer exhi-bit much better rutting resistance than those using unmod-ified binder. It appears that the binder type and content playa significant role in the rutting performance of the mixtures.The laboratory test results are successfully validatedthrough field data.

(5) The CIR mixture under a good curing condition preliminarilyshows its feasibility and reliability for use as the bottomasphalt layer for highway maintenance. Using anti-ruttingagent can significantly improve the pavement performanceat high temperatures. A thick overlay on the CIR layer withthe optimum cement content is recommended to balancethe rutting resistance and cracking potential. The CIR tech-nique also shows promising use in the milling/overlay plan.

(6) The middle asphalt layer produces the greatest permanentdeformation. The rutting distribution is relatively uniformin the CIR/overlay pavement or in the newly constructedpavement where SBS modified binder is used in the middleasphalt layer.

(7) Currently there are still some potential limitations to theextensive application of the ARLPD test, such as complexspecimen preparation method, short testing time interval,and relatively inaccurate displacement measurement usingcrosshead/actuator rather than the on-specimen linear vari-able differential transformer (LVDT). All these limitationsneed to be addressed properly in further following-upstudies.

Acknowledgements

The authors would like to acknowledge the financial support bythe Project supported by the National Natural Science Foundationof China (Grant No. 51308303), the Project Supported by Basic Re-search Plan (Natural Science Foundation) in Jiangsu Province ofChina (Grant No. BK20130980), China Postdoctoral Science Foun-dation funded Project (Grant No. 2013M531252), and JiangsuPlanned Projects for Postdoctoral Research Funds (Grant No.1202010C).

References

[1] Wang H, Zhang Q, Tan J. Investigation of layer contributions to asphaltpavement rutting. J Mater Civil Eng 2009;21(4):181–5.

[2] Walubita LF, Zhang J, Das G, Hu X, Mushota C, Alvarez AE, et al. Hot-mix asphaltpermanent deformation evaluated by Hamburg wheel tracking, dynamicmodulus, and repeated load tests. Transport Res Rec 2012;2296:46–56.

[3] Hajj EY, Tannoury G, Sebaaly PE. Evaluation of rut resistant asphalt mixturesfor intersection. Road Mater Pavement 2011;12(2):263–92.

[4] Zhu LY, Fwa TF, Liu Y. Rutting potential evaluation of asphalt mixtures byrepeated-load creep test. In: Proceedings of 81st annual meeting of theTransportation Research Board. Washington (DC); 2002.

[5] Huang X, Zhang Y. A new creep test method for asphalt mixtures. Road MaterPavement 2010;11(4):969–91.

[6] Chen X, Huang B, Xu Z. Uniaxial penetration testing for shear resistance of hot-mix asphalt mixtures. Transport Res Rec 2006;1970:116–25.

[7] Xu Q, Solaimanian M. Modeling temperature distribution and thermalproperty of asphalt concrete for laboratory testing applications. Constr BuildMater 2010;24(4):487–97.

Q. Li et al. / Construction and Building Materials 61 (2014) 241–251 251

[8] Guan H, Zhang Q, Luo Z. Total-thickness laboratory rut tests for asphalt surfacecourse under temperature gradient. China Civil Eng J 2011;44(6):143–7.

[9] Azari H, Mohseni A. Incremental repeated load permanent deformation testingof asphalt mixtures. In: Proceedings of 91st annual meeting of theTransportation Research Board. Washington (DC); 2012.

[10] Yuan Q. Research on experiment method of asphalt pavement permanentdeformation at high temperature. Master thesis, Nanjing: Southeast University;2012.

[11] Li Q. A permanent deformation model of asphalt concrete based on shearproperties. Ph.D dissertation, Seoul: Sejong University; 2010.

[12] Department of Transportation of Jiangsu Province of China. Applied manual forasphalt emulsion cold in-place recycling technology. Nanjing; 2010.

[13] Li Q, Lee HJ, Hwang EY. Characterization of permanent deformation of asphaltmixtures based on shear properties. Transport Res Rec 2010;2181:1–10.

[14] Walubita LF, Zhang J, Alvarez AE, Hu X. Exploring the Flow Number (FN) indexas a means to characterise the HMA permanent deformation response underFN testing. J S Afr Inst Civ Eng 2013;55(3):103–12.

[15] Zhou F, Scullion T, Sun L. Verification and modeling of three-stage permanentdeformation behavior of asphalt mixes. J Transp Eng – ASCE 2004;130(4):486–94.

[16] Kalyoncuoglu SF, Tigdemir M. A model for dynamic creep evaluation of SBSmodified HMA mixtures. Constr Build Mater 2011;25(2):859–66.

[17] Li C, Li L. Criteria for controlling rutting of asphalt concrete materials in slopedpavement. Constr Build Mater 2012;35:330–9.