Construction and Building Materials 26 (2012) 583–590

Contents lists available at ScienceDirect

Construction and Building Materials

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

Laboratory evaluation on high temperature viscosity and low temperature stiffnessof asphalt binder with high percent scrap tire rubber

Hainian Wang a,⇑, Zhanping You b,1, Julian Mills-Beale b,2, Peiwen Hao a,3

a Highway School, Chang’an University, South Erhuan Middle Section, Xia’n, Shaanxi, 710064, Chinab Department of Civil and Environmental Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931-1295, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 6 November 2010Received in revised form 23 May 2011Accepted 18 June 2011Available online 12 July 2011

Keywords:Asphalt rubberRotational viscosityCreep stiffnessLaboratory testRTFO agingRubber concentration

0950-0618/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.conbuildmat.2011.06.061

⇑ Corresponding author. Tel.: +86 29 82334824.E-mail addresses: wanghainian@yahoo.com.cn (H

You), jnmillsb@mtu.edu (J. Mills-Beale), haopw@yaho1 Tel.: +1 906 487 1059.2 Tel.: +1 906 487 2528.3 Tel.: +86 29 82334427.

The objective of this research is to utilize crumb rubber from scrap tires as an environmental friendly andsustainable additive for enhancing the high temperature and low temperature rheological properties ofasphalt binders for asphalt pavements. Two different crumb rubber sources with different gradations –fine and coarse – were used in this project. The crumb rubber-modified (CRM) binder was producedby adding 10, 15, 20 and 25% crumb rubber particles by weight of a Superpave PG 64-22 asphalt binder.The CRM binders with and without Rolling Thin Film Oven (RTFO) aging were characterized by theAASHTO rotational viscosity test at 135, 140, 150, 160, 170, 177, and 190 �C (AASHTO T316). Furthermore,the low temperature cracking resistance of the binders was evaluated using the AASHTO Bending BeamRheometer (BBR) test procedure at �12 and �18 �C (AASHTO T313). The statistical analysis of variance(ANOVA) was applied to quantify the effect of the influencing factors such as temperature, rubber particlesize, and rubber concentration on the CRM binders’ performance. From the laboratory tests and ANOVAresults in this study, it is evident that the addition of crumb rubber into asphalt binder can both signif-icantly improve the viscosity of binder at high temperature and lower the creep stiffness at low temper-ature, which is beneficial to better both high temperature stability and low temperature crackingresistance of asphalt pavements. After RTFO aging, the viscosity decreases with increasing rubber concen-tration. Finer crumb rubber attains higher viscosity at high temperature and lower creep stiffness at lowtemperature. Considering the viscosity–temperature relationship, RTFO aging effects, creep stiffnessdecreasing percentage, and economical factors, 15% to 20% rubber asphalt ratio is proposed for the pro-duction of CRM binder.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

With the motor industry developing and spreading at a higherpace in all parts of the world, high amount of scrap tires were pro-duced every year, which makes the disposal of tires a serious envi-ronmental problem [1]. Crumb rubber, which is obtained from thegrinding of scrap tires, has proved to be an efficient solution to theenvironmental concerns surrounding the accumulation of wastetires in recent years [2,3]. The beneficial use of crumb rubber intovirgin asphalt binder and pavements provides an environmentallysustainable method of disposing of the millions of tires generatedannually [4].

ll rights reserved.

. Wang), zyou@mtu.edu (Z.o.com.cn (P. Hao).

The American Society of Testing and Materials (ASTM) definesasphalt rubber (AR) as ‘‘a blend of asphalt cement, reclaimed tirerubber and certain additives, in which the rubber component isat least 15% by weight of the total blend and has reacted in thehot asphalt cement sufficiently to cause swelling of the rubber par-ticles,’’ [5]. Researchers have shown that the addition of crumbrubber into virgin asphalt can produce asphalt rubber binders withbetter resistance to rutting, fatigue cracking and thermal crackingas well as reducing the thickness of asphalt overlays and potentialreflective cracking [6,7]. The asphalt rubber acts in slurry and chipseal materials as a stress absorbing membrane while demonstrat-ing good anti-fatigue and durability performance in field applica-tions [8,9].

The addition of crumb rubber into virgin asphalt induces asignificant increase in binder viscosity. As the viscous propertyof asphalt rubber is critical to mixture compaction temperatureand binder workability during storage and pumping process,the viscosity of asphalt rubber has been the central focus in pre-vious research work [10,11]. Lougheed and Papagiannakis

60

80

100

tage

(%)

Rubber A Rubber B

Table 1The properties of virgin asphalt.

Aging states Test properties Testing results

naged binder Rotational viscosity@ 135 �C (Pa s)

0.435

G�= sin d @ 64 �C (kPa) 1.412RTFO aged residue G�= sin d @ 64 �C (kPa) 3.69RTFO + PAV aged residue G� � sin d @25 �C (kPa) 1171

Stiffness @ �12 �C (MPa) 189m-value @ �12 �C 0.314

Table 2Properties of the crumb rubber materials.

Property Rubber A Rubber B

Specific gravity (g/m3) 1.12 1.14Moisture content (%) 0.56 0.65Ash content (%) 3.6 4.3Acetone to mention oil complex (%) 8.9 10.2Fiber content (%) 0.1 0.05Metal content (%) 0 0Carbon black content (%) 32.7 35.4

584 H. Wang et al. / Construction and Building Materials 26 (2012) 583–590

adopted the Brookfield viscometer to test the viscosity of threevirgin and six rubber-modified asphalt binders [12]. Their sam-ples contained crumb rubber concentrations of 3%, 5%, 7%, 12%and 18% by weight of the virgin binder. Notable among theirconclusions was the introduction of the concept of ‘‘stabilizedviscosity’’. Stabilized viscosity is the phenomenon whereby theviscosity of the rubber-modified asphalt will decrease to a stabi-lized value after approximately 45–75 min of blending; with theexact stabilized time dependent on the crumb rubber concentra-tion. West et al. evaluated the effect of the tire rubber grindingmethod on AR binder properties and characteristics, and theyfound a good correlation between the grinding process of crumbrubber and the viscosity and storage settlement. Crumb rubberwith greater specific surface areas and more irregular shapescan induce high viscosity conditions in asphalt rubber binder[13].

Lee et al. adopted the gel permeation chromatography (GPC),dynamic shear rheometer (DSR) and rotational viscosity (RV) tocharacterize control binder, SBS-modified binder and rubber-mod-ified binder of two short-term aging method, rolling thin film oven(RTFO) aging and short-term oven aging (STOA) [14]. According totheir tests, increased aging time will cause an increase in viscosityat high temperatures for the control and SBS-modified binders. Itshould be noted however that there was no clear trend in the vis-cosity change for the rubber-modified binder with and withoutaging. The scanning electron microscope (SEM) and differentialscanning calorimeter (DSC) techniques have been used to evaluatethe effect of crumb rubber characteristics, including rubber sourcesand rubber concentration, on crumb rubber-modified (CRM) bin-der viscosity [15]. Their tests proposed that the CRM type andsources plays an obvious role in influencing the viscous propertiesof the CRM binder. Statistical regression and neural network ap-proaches have been applied to predict the viscosities of differentrubber type CRM binders with different concentrations and pro-posed an efficient way to estimate the viscous properties of differ-ent variables such as asphalt binder grade, binder source, testtemperature, rubber content and rubber source [16]. With theaid of the dynamic shear rheometer (DSR), rotational viscometerand the GPC, interaction effects such as blending time, temperatureand rubber content of CRM binders were investigated in researchconducted by Jeong et al. [17]. Their work proved that longerblending time and higher blending temperature result in a higherviscosity of CRM binders.

Previous research investigations have focused on viscous prop-erties of CRM binders from different aspects and this was beneficialto understand the different influence factors and their effects onthe performance of CRM binders. It must be emphasized that theaging effect on the viscosity of binders containing different CRMconcentrations and at different temperatures still need a thoroughstudy. Additionally, it is pertinent to focus on the low temperaturestiffness of CRM binders to investigate the relationship betweenlow temperature stiffness and thermal cracking of CRM mixturepavements. These areas of study have received less attention inpast and current studies.

0

20

40

Pass

ing

Perc

en

Sievesize (mm)1.180.850.60.4250.30.150.075

Fig. 1. The passing percent gradation of Crumb Rubbers A and B.

2. Objective and scope

The objective of this research is to utilize crumb rubber fromscrap tires as an environmental friendly and sustainable additivefor enhancing the rheological properties of asphalt binders. The fo-cus was to investigate the viscous property of CRM binders withand without RTFO aging at different test temperatures, and alsotest the low temperature creep stiffness of CRM binders with dif-ferent rubber concentrations.

3. Experimental program

3.1. Materials

Two particle size crumb rubber materials cryogenically produced from differentsources in China were adopted in this paper. Fig. 1 shows the percent passing gra-dation of Crumb Rubber A (Rubber A) and Crumb Rubber B (Rubber B). Five rubberasphalt concentrations, 0%, 10%, 15%, 20% and 25% by weight of asphalt, were usedin this study.

A Superpave PG 64–22 binder was used as the control binder in this study. Thisbinder was obtained from a construction site near Detroit in Michigan and met theMDOT specification requirements. Table 1 shows the properties of control PG 64-22binder.

Two sources of cryogenic fine crumb rubber were added to the virgin PG 64-22binder to produce the CRM binders. The basic properties of the crumb rubber areshown in Table 2.

3.2. Experimental plan

The detailed experimental plan is indicated in Fig. 2. The plan sums up thematerial preparation, Superpave™ characterization and evaluation of the CRMbinders.

3.3. Sample preparation

The crumb rubber was added gradually into the asphalt binder at a reactiontemperature of 350 �F (177 �C), and mixed mechanically for about 45 min. The reac-tion time of 45 min was considered adequate based on some preliminary literature

PG 64-22 Asphalt Binder

Virgin asphaltRubber modified

asphalt

Unaged RTFOTRTFOT+PAV

Same testing procedures as Virgin asphalt

Rotational Viscosity

Rotational Viscosity

BBR @ -18C & -12C

Source A Source B

20%15%10% 25%

Same testing procedures as Virgin asphalt

Same testing procedures as Virgin asphalt

Same testing procedures as Virgin asphalt

Same testing procedures as

Source A

Fig. 2. Experimental plan for the CRM binder tests.

140 160 180 200 140 160 180 2000.03125

0.0625

0.125

0.25

0.5

1

2

4

8

0.03125

0.0625

0.125

0.25

0.5

1

2

4

8

Vis

cosi

ty (

Pa.s

)

Temperature ( ) Temperature ( )

0% 10% 15% 20% 25%

Rubber A

Vis

cosi

ty (

Pa.s

)

0% 10% 15% 20% 25%

Rubber B

Fig. 3. Viscosities of unaged CRM binders (Rubber A on left; Rubber B on right).

H. Wang et al. / Construction and Building Materials 26 (2012) 583–590 585

reviewing indicating that the CRM binder could reach the highest viscosity at thistime [12,18]. After 45 min reaction time, the CRM binder was tested under theBrookfield viscometer at seven different temperature conditions – 190, 177, 170,160, 150, 140, and135 �C. A 25% torque was applied and the rotation speed wasset at 100 rpm. The #29 spindle was adopted in the tests in favor of the #27 spindledue to the high viscosity of CRM binders. Both unaged and RTFO-aged CRM binderswere tested to evaluate their viscosity. The viscosity test followed the AASHTO T316 standard test specification. Furthermore, the low temperature stiffness ofCRM binders was evaluated at �12 and �18 �C using the BBR test equipmentaccording to the AASHTO T 313 standard test specification. Three replicates wereconducted in both rotational viscosity and BBR and the average rest values were ap-plied in the subsequent discussion.

4. Results and discussion

4.1. High temperature viscosity

The influence of rubber types, rubber concentration, test tem-perature, and aging effect on the viscosity of CRM binders is dis-cussed in this section. Figs. 3 and 4 show the viscosity graphical

plots for the unaged and RTFO-aged CRM binders, respectively. Itis clear that the viscosity of CRM binder at any rubber concentra-tion decreases with increasing test temperature, with the sametrend holding true for the non-modified asphalt. The addition ofcrumb rubber can greatly increase the binder viscosity, which is vi-tal in increasing the binder film thickness for coating aggregates inthe hot mixture. Ultimately, the more viscous CRM binder willmaintain the stability of asphalt mixtures.

With increasing percentage of crumb rubber, the binder viscos-ity increases at each test temperature. The most remarkable in-crease in viscosity occurs when the rubber content is increasingfrom 0% to 10%, and with the continual increase in rubber content,the overall viscosity increasing amplitude experiences a little de-crease for the two CRM binders at unaged or RTFO-aged condition.The Superpave™ specification (AASHTO M 320) requires that themaximum viscosity of asphalt binder is no greater than 3 Pa s at135 �C for the convenience of storage and pumping in constructionperiod. However, it is difficult to follow this requirement for CRM

586 H. Wang et al. / Construction and Building Materials 26 (2012) 583–590

binders. If the CRM binders reach 3 Pa s viscosity for rubber B mod-ified binder, their temperature need to increase to 147, 162, and174 �C for 15%, 20%, and 25% rubber–asphalt ratio binder, respec-tively. The normal requirement of 3 Pa s is thus not feasible forhigh percent CRM binder. Thus, the storage, blending and rollingtemperature of asphalt mixtures with high percent CRM binderneed to be heated to higher temperature in construction, respec-tively. The exact optimal temperatures of CRM binders are subjectto viscosity–temperature curves, which may be influenced by therubber characteristics, rubber concentration and asphalt binder.

To better understand the inherent relationship between eachinfluencing factor and their effects on the viscosity of CRM binders,the regression between the viscosity and the test temperature forCRM binders was studied here and could be presented as:

logðVÞ ¼ a � T þ b ð1Þ

where V is the viscosity of the CRM binder; T, the test temper-ature; ‘‘a’’ and ‘‘b’’, the regression parameters. The slope coefficient‘‘a’’ represents the changing rate of binder viscosity with the testtemperature.

The relationship was developed based on research investiga-tions by [16,19]. Tables 3 and 4 illustrate the regression parameters

140 160 180 2000.0625

0.125

0.25

0.5

1

2

4

8

Vis

cosi

ty (

Pa.s

)

Temperature ( )

0% 10% 15% 20% 25%

Rubber A

Fig. 4. Viscosities of RTFO aged CRM binder

Table 3Regression parameters for viscosities of unaged CRM binders.

Rubber asphalt ratio (%) Rubber A

a b

10 �0.01468 2.1615515 �0.01604 2.7354220 �0.01625 3.0458925 �0.01292 2.48987

Table 4Regression parameters for viscosities of RTFO-aged CRM binders.

Rubber asphalt ratio (%) Rubber A

a b

10 �0.01595 2.5666415 �0.01605 2.853320 �0.01545 2.997425 �0.01553 2.85271

between the viscosity and test temperatures at different rubber as-phalt ratio for CRM binders with and without RTFO aging, respec-tively. R2 is the correlation coefficient of the regression.

From the R2 values in Tables 3 and 4, Formula 1 characterizeswell the good correlation between the viscosity and test temper-ature for both CRM binders with and without RTFO. The absolutevalue of ‘‘a’’ increased about 10% and 5%, from 10% to 20% rubberasphalt ratio, for rubber A and rubber B binders, respectively. Forthe unaged CRM binder, as the rubber asphalt ratio increases un-til to 20%, the viscosity decrease rate is increasing with theincreasing of test temperature. For the both RTFO-aged CRM bind-ers, the largest absolute ‘‘a’’ value occurred at 15% rubber asphaltratio. A bigger absolute value for ‘‘a’’ is desired for the CRM bind-ers as it will be beneficial to have a relatively low viscosity athigh temperature for construction workability of the CRM bindersand have a greater viscosity at relatively low temperature for rut-ting resistance and high temperature stability of rubber asphaltmixture.

Among the many influencing factors on the viscosity of CRMbinder, the test temperature is one of those most important. How-ever, different states and countries may have different require-ments on the viscosity test temperature for CRM binders.

140 160 180 2000.0625

0.125

0.25

0.5

1

2

4

8

Temperature ( )

Vis

cosi

ty (

Pa.s

)

0% 10% 15% 20% 25%

Rubber B

s (Rubber A on left; Rubber B on right).

Rubber B

R2 a b R2

0.997 �0.01526 2.22813 0.9960.999 �0.01559 2.77321 0.9990.999 �0.01604 3.10298 0.9990.999 �0.01412 2.93286 0.999

Rubber B

R2 a b R2

0.998 �0.01562 2.54588 0.9980.999 �0.01637 3.04484 0.9990.999 �0.01429 2.90593 0.9990.999 �0.01408 2.91606 0.999

H. Wang et al. / Construction and Building Materials 26 (2012) 583–590 587

Arizona, Texas and ASTM assigned 177 �C in their specifications,while California and South Africa set 190 �C in the viscosity testing,other than normal 135 �C for regular binders [20,21]. The 177 and190 �C test temperature were taken out to analyze their effects onthe viscosities of both CRM binders with and without RTFO aging,and were shown in Figs. 5 and 6, respectively.

It can be observed that the crumb rubber size have an obviousinfluence on the high temperature viscosity of CRM binders. Rub-ber B is finer than Rubber A, and its modified binder has a higherviscosity than Rubber A whether with or without RTFO aging. Forthe unaged CRM binders, there is no remarkable difference be-tween Rubbers A and B modified binder at 10% rubber asphalt ra-tio, and the difference increases to 71% and 60% at 25% rubberasphalt ratio for 177 and 190 �C test temperatures, respectively.The finer crumb rubber has a greater surface area and therefore re-acted and swelled efficiently during the blending process. As a re-sult, the finer crumb rubber reached a higher viscosity with thesame rubber asphalt ratio.

The 85 min RTFO aging also has a notable effect on the viscosityof CRM binders. With increasing rubber concentration, the percentimprovement in CRM binder viscosity after RTFO aging begins todecrease. For Rubber A binder at a test temperature of 177 �C,the viscosity improving percentage decreases from 75% (0% rubber)to 42% (10% rubber), and further to 34% (15% rubber) and then to21% (20% rubber), and -21% (25% rubber). During the aging process,

A UNAGED A RTFO B UNAGED B RTFO0.0

0.5

1.0

1.5

2.0

2.5

3.02.59

2.34

1.40

0.59

2.72

1.87

1.03

0.36

1.25

1.81

1.02

0.54

Visc

osity

(Pa.

s)

10% 15% 20% 25%

0.38

0.76

1.501.59

Test Temperature 177 oC

Fig. 5. Viscosity comparisons of Rubbers A and B modified asphalt at differentconcentration levels and 177 �C test temperature.

A UNAGED A RTFO B UNAGED B RTFO0.0

0.5

1.0

1.5

2.0 1.801.71

0.97

0.43

1.90

1.18

0.68

0.24

0.90

1.28

0.72

0.390.25

0.54

0.95

Visc

osity

(Pa.

s)

10% 15% 20% 25%

1.18

Test Temperature 190 oC

Fig. 6. Viscosity comparisons of Rubbers A and B modified asphalt at differentconcentration levels and 190 �C test temperature.

the aromatic oil and light fractions contents decrease in the asphaltbinder and this induces a greater binder viscosity. With the addi-tion of crumb rubber into asphalt at high temperatures, the rubberparticle will absorb the aromatic oil and light fractions in the as-phalt and swell in size to induce a higher viscosity. With increasingrubber concentration in the CRM binder, the percentage of free aro-matic and light fractions will decrease, and the actual effect ofRTFO aging will also decrease. When the rubber asphalt ratioreaches 25%, the viscosity of RTFO-aged binder is even smaller thanthat of unaged binder. This is not beneficial for long term storage ofthe rubber asphalt after modifying, and could ultimately deterio-rate the field performance.

It is noticed that, at the temperatures less than 170 �C, the 20%Rubber A binder has higher viscosity value than 25% Rubber A bin-der. The possible reason could be elaborated as following. The vis-cosity test on CRM binder is conducted from high temperature(190 �C) to low temperature (135 �C) using rotational viscometerin the laboratory. The accumulated long time temperature control-ling and equilibration period may induce some potential aging onthe binder and due to the aforementioned decreasing of free aro-matic and light fraction content, and result in the lower viscosityof 25% CRM binder at the temperature less than 170 �C. However,this effect may also be subjected to the crumb rubber source,and more SEM tests could provide detailed explanation on thismechanism in the future.

The viscosity improving effect is also influenced by the crumbrubber concentration in the rubber asphalt binders. With theincreasing of rubber concentration, the viscosity improving per-centage shows a decreasing trend. Take unaged Rubber B bindertested at 190 �C as an example, the viscosity is increased 433% from0% to 10% rubber asphalt ratio, 183% from 10% to 15% rubber as-phalt ratio, 74% from 15% to 20% rubber asphalt ratio, and 61% from20% to 25% rubber asphalt ratio. With increasing rubber concentra-tion in the CRM binder, the percentage of free aromatic and lightfractions will decrease, which induces a smaller viscosity improv-ing effect. It should be also noted that, with the increasing of hightemperature viscosity of CRM binder, it will induce the difficulty ofpumping, reduce its workability, and raise the heat energy con-sumption in construction. From these two aspects, the 25% rubberasphalt ratio is not proposed for field application.

4.2. ANOVA analysis on high temperature viscosity

The statistical analysis of variance (ANOVA) was applied toinvestigate the high temperature viscosity of CRM binders as afunction of test temperature and rubber concentration and the re-sults are summarized in Table 5. For both Rubber A and Rubber Bbinder, the F value is greater than its F critical value and the P-va-lue is smaller than the significance level of 0.05. The ANOVA dataindicates that both the test temperature and rubber concentrationhave significant effect on the viscosity of CRM binders.

In order to thoroughly compare the viscosity of CRM binders atdifferent rubber concentration, the one-factor ANOVA was adoptedto evaluate if the viscosities of adjacent rubber concentration CRMbinders have significant difference, as shown in Table 6. The ANO-VA data shows that with the increasing of rubber concentration upto 20%, the rubber concentration has a significant influence on theviscosity of CRM binder, as the F value is greater than its corre-spondent F critical value and P-value is smaller than 0.05. How-ever, there is no significant difference between the viscosities of20% and 25% rubber concentration for both Rubber A and RubberB CRM binder. From the cost-effective viewpoint of viscosityincreasing performance and increasing cost on crumb rubber, itis also not suggested to apply the rubber concentration to morethan 20%.

Table 5Two-factor ANOVA on the viscosity of CRM binder (a = 0.05).

Materials Source of variance SS(Pa s) df MS(Pa s) F P-value F crit

Rubber A Temperature 47.673 6 7.946 8.7172 4E�05 2.508Concentration 55.958 4 13.989 15.348 2E�06 2.776

Rubber B Temperature 15.307 4 3.827 6.3269 0.003 3.007Concentration 48.044 4 12.011 19.858 5E�06 3.007

Note: SS, the sum of squared deviations; df, the degree of freedom; MS, mean square; F, the F value; F crit, the F critical value.

0

50

100

150

200189189

Testing Temperature -12 oC

Stiff

ness

(MPa

)

Rubber asphalt ratio (%)

Rubber A Rubber B

0 10 15 20 25

145 142

112

10190

7785

60

Fig. 7. Low temperature stiffness obtained by BBR tests on of Rubbers A and B(�12 �C).

0

50

100

150

200

250

300

350

400

Rubber asphalt ratio (%)

356

Testing Temperature -18oC

Stiff

ness

(MPa

)

Rubber A Rubber B

0 10 15 20 25

356

266

200 193 181176 168

153135

Fig. 8. Low temperature stiffness obtained by BBR tests on Rubbers A and B(�18 �C).

588 H. Wang et al. / Construction and Building Materials 26 (2012) 583–590

4.3. Low temperature stiffness

The BBR test can be used to evaluate how much a binder de-flects or creeps under a constant load at low temperature. Thecreep stiffness obtained from BBR test can well characterize thecracking resistance of asphalt binder at low temperature. Figs. 7and 8 illustrate the creep stiffness of A and B CRM binders with dif-ferent rubber contents at �12 and �18 �C test temperatures. Theaddition of crumb rubber into the asphalt greatly decreases thelow temperature stiffness of CRM binders, which can increase thetoughness of CRM mixtures and decrease the occurring possibilityof the asphalt binder and pavement cracking at low temperature.For CRM A binder at 20% rubber asphalt ratio, its creep stiffness re-duces to about 50% of the control binder at both �12 and �18 �Ctest temperatures. Additionally, with increasing rubber concentra-tion, the creep stiffness decreases for both CRM binders at both�12 and �18 �C test temperatures. The crumb rubber is not astemperature sensitive as the asphalt binder, and has lower modu-lus and stiffness than the asphalt binder at low temperature. As aresult, the increasing rubber content will induce lower creep stiff-ness for CRM binders at low temperature.

Superpave™ specification (AASHTO M 320) requires the creepstiffness to be less than 300 MPa and m-value to be greater than0.300 at the test temperature during the performance grading ofthe asphalt binder. The research mainly focuses on the creep stiff-ness of CRM binders to characterize their low temperature crackingperformance. The control PG 64–22 binder meets this requirementat �12 �C but fails at �18 �C. With the addition of crumb rubber,both CRM binders can definitely meet the criteria at �18 �C evenat 10% rubber asphalt ratio. Therefore, the addition of crumb rub-ber decreased the low temperature grade from �22 �C to �28 �Ccompared to the control binder.

When the rubber asphalt ratio exceeds 15%, the percentage de-crease in creep stiffness is not as remarkable as before. The creepstiffness decreases by 25% from control binder to the 10% rubberasphalt ratio binder, and by 27% from the 10% to 15% rubber as-phalt ratio binder. Finally, it decreases by 9% from the 15% to20% rubber asphalt ratio for Rubber A binder at �18 �C. From thispoint, the highest rubber asphalt ratio, 25%, would be no suggestedin field application.

The crumb rubber particle size also has some influence on thecreep stiffness of the CRM binder. For the CRM binders at both�12 and �18 �C, Rubber B (finer size rubber) binder has less creepstiffness than Rubber A binder. This may be due to a more efficient

Table 6One-factor rubber concentration ANOVA on the viscosity of CRM binder (a = 0.05).

Rubber concentration Rubber A

F P-value

0% vs 10% 17.418 0.00585610% vs 15% 13.022 0.0112515% vs 20% 12.862 0.01155420% vs 25% 2.7134 0.150604

Note: F, the F value; F crit, the F critical value.

reaction between the fine crumb rubber and asphalt binder com-pared to the coarse crumb rubber and the asphalt binder.

Rubber B

F crit F P-value F crit

5.9874 15.6694 0.0075 5.98745.9874 15.2241 0.008 5.98745.9874 13.3954 0.0106 5.98745.9874 2.41051 0.1844 5.9874

Table 7Two-factor ANOVA on the low temperature stiffness of CRM binder (a = 0.05).

Test temperature Source of variation SS (MPa) df MS (MPa) F P-value F crit

�12 �C Rubber type 791.05 1 791.05 9.25 0.006 4.35Concentration 54866.07 4 13716.52 160.31 0.000 2.87Interaction 581.68 4 145.42 1.70 0.190 2.87

�18 �C Rubber type 3286.53 1 3286.53 10.14 0.005 4.35Concentration 166566.47 4 41641.62 128.43 0.000 2.87Interaction 4119.80 4 1029.95 3.18 0.036 2.87

Note: SS, the sum of squared deviations; df, the degree of freedom; MS, mean square; F, the F value; F crit, the F critical value.

Table 8One-factor rubber concentration ANOVA on the low temperature stiffness of CRMbinder (a = 0.05).

Rubber concentration �12 �C �18 �C

F P-value F crit F P-value F crit

0% vs 10% 158.20 1.5E�06 5.32 73.50 0.000 5.3210% vs 15% 153.14 1.7E�06 5.32 15.24 0.005 5.3215% vs 20% 10.12 0.01297 5.32 10.73 0.011 5.3220% vs 25% 2.31 0.16734 5.32 7.41 0.026 5.32

Note: F, the F value; F crit, the F critical value.

H. Wang et al. / Construction and Building Materials 26 (2012) 583–590 589

4.4. ANOVA analysis on low temperature stiffness

The ANOVA was also applied to investigate the low temperaturestiffness of CRM binders as a function of rubber type and rubberconcentration at �12 and �18 �C, respectively. The ANOVA analy-sis results are summarized in Table 7. For both �12 and �18 �C testtemperature, the F value is greater than its correspondent F criticalvalue and the P-value is smaller than the significance level of 0.05.The ANOVA data indicates that both the rubber type and rubberconcentration have significant effect on the low temperature stiff-ness of CRM binders.

In order to thoroughly investigate the low temperature stiffnessof CRM binders at different rubber concentration, the one-factorANOVA was adopted to evaluate if the low temperature stiffnessof adjacent rubber concentration CRM binders have significant dif-ference, as shown in Table 8. The ANOVA data shows that, for both�12 and �18 �C test temperature, with the increasing of rubberconcentration up to 20%, the rubber concentration has a significantinfluence on the low temperature stiffness of CRM binder, as the Fvalue is greater than its correspondent F critical value and P-valueis smaller than 0.05. However, its influence impact (F value) isdecreasing with the increasing of rubber concentration. Finally,there is no significant difference between the low temperaturestiffness of 20% and 25% rubber concentration. From the cost-effec-tive viewpoint of low temperature stiffness decreasing perfor-mance and increasing cost on crumb rubber, it is also notsuggested to apply the rubber concentration to more than 20%.

4.5. Proper rubber asphalt ratio

According to the rotational viscosity tests and bending beamrheometer tests on CRM binders conducted in this paper, it is clearthat the addition of crumb rubber into pure asphalt can better bothits high temperature viscosity and low temperature stiffness,which is desired and beneficial to the better performance of rubberasphalt mixture. However, it should be also noted that, with theincreasing of rubber concentration in the CRM binders, the modify-ing effects of viscosity and stiffness is decreasing. Based on the vis-cosity and temperature regression relationship, the 15% to 20%rubber asphalt ratio binders will have the greatest regressionparameter, ‘‘a’’, and induce the most desirable viscosity–tempera-

ture curve. According to the ANOVA analysis on high temperatureviscosity and low temperature stiffness, there is no significant per-formance difference between 20% and 25% rubber concentrationCRM binders. Meanwhile, the over excessive rubber concentrationwill also make the difficulty of pumping, reduce the mixture work-ability, raise the heating energy consumption and increase the costof rubber asphalt binder. Considering the all the factors above, 15to 20% rubber asphalt ratio is suggested in the production ofCRM binder. The optimum rubber concentration should be deter-mined by considering the actual application fields and modifyingdemands, asphalt plant pumping and blending conditions, andother factors comprehensively, which is to be further investigated.

5. Conclusions

Two crumb rubber samples of different gradations – fine andcoarse – were used in this project. Four rubber asphalt ratios wereapplied to prepare the CRM binders using Superpave PG 64-22 as-phalt binder and the crumb rubber additives. Seven test tempera-tures were taken to evaluate the viscous properties of the CRMbinders with and without RTFO aging. The BBR test was used tomeasure creep stiffness of both CRM binders at �12 and �18 �Cin this paper. The ANOVA technique was applied to quantify the ef-fect of factors, such as test temperature, rubber type, rubber con-centration, on the CRM binders’ performance. The test resultsobtained from this study can unveil some inherent correlationsof each influencing factor and the viscosity and creep stiffness ofCRM binders and provide some guiding frameworks in determin-ing the rubber constitutive design and construction parametersfor CRM asphalt binders. Some preliminary conclusions that canbe drawn from the research are:

(1) The addition of crumb rubber into asphalt binder can signif-icantly improve the viscosity of binders, which is beneficialto enhance the high temperature performance of asphaltbinders and mixture. The viscosity specification requirementof 3 Pa s is however not feasible for high percent CRMbinder.

(2) The addition of crumb rubber into asphalt binder can reducethe creep stiffness of CRM binder at low temperature whichis helpful for better cracking resistance ability of asphalt bin-der and mixture. From the perspective of low temperaturestiffness, the addition of 10% crumb rubber into control bin-der can lower a low temperature grade from �22 �C to�28 �C.

(3) With increasing rubber concentration, the performance onhigh temperature viscosity and low temperature stiffnessof CRM binders are improved, but its improving impacttends to decrease. ANOVA results indicate that there is nosignificant performance difference on high temperature vis-cosity and low temperature stiffness between 20% and 25%rubber concentration CRM binders.

(4) Finer crumb rubber can help achieve higher viscosity at hightemperature and lower creep stiffness at low temperature.

590 H. Wang et al. / Construction and Building Materials 26 (2012) 583–590

(5) Considering the viscosity–temperature relationship, RTFOaging effects, creep stiffness decreasing percentage and eco-nomical factors, 15–20% rubber asphalt ratio is proposed forthe production of CRM binder. However, further CRM binderand mixture tests will confirm its extensive laboratoryperformance.

Acknowledgement

The research is supported by the funds of Natural Science FoundCommittee (NSFC) of China (No. 50808023) (No. 51011120574)and the Special Fund for Basic Scientific Research of Central Col-leges, Chang’an University (CHD2010JC061). The experimentalwork was completed in the Transportation Materials Research Cen-ter at Michigan Technological University, which maintains theAASHTO Materials Reference Laboratory (AMRL) accreditation onasphalt and asphalt mixtures.

Reference

[1] Kiser JV. Asphalt rubber: overcoming the obstacles. Scrap 2003;60(1):46–50.[2] Xiao F, Wenbin Zhao PE, Amirkhanian SN. Fatigue behavior of rubberized

asphalt concrete mixtures containing warm asphalt additives. Constr BuildMater 2009;23(10):3144–51.

[3] Shen J, Amirkhanian S, Xiao F, Tang B. Surface area of crumb rubber modifierand its influence on high-temperature viscosity of CRM binders. Int J PavementEng 2009;10(5):375–81.

[4] Xiao F, Amirkhanian SN, Wu B. Fatigue and stiffness evaluations of reclaimedasphalt pavement in hot mix asphalt mixtures. J Test Eval 2011;39(1).

[5] ASTM international annual book of standards. D 8 definitions, vol. 04.03; 2008.[6] Raad L, Saboundjian S. Fatigue behavior of rubber-modified pavements.

Transport Res Record 1998(1639):73–82.

[7] Huang B, Mohammad LN, Graves PS, Abadie C. Louisiana experience withcrumb rubber-modified hot-mix asphalt pavement. Transport Res Record2002:1–13.

[8] Kuennen T. Surface treatments: when seals make sense. Better Roads2004;74(10):24–32.

[9] Chen D-H, Scullion T, Bilyeu J, Won M. Detailed forensic investigation andrehabilitation recommendation on interstate highway-30. J Perform ConstrFacil 2005;19(2):155–64.

[10] Celik ON, Atis CD. Compatibility of hot bituminous mixtures made with crumbrubber-modified binders. Constr Build Mater 2008;22(6):1143–7.

[11] Akisetty CK, Lee S-J, Amirkhanian SN. Effects of compaction temperature onvolumetric properties of rubberized mixes containing warm-mix additives. JMater Civil Eng 2009;21(8):409–15.

[12] Lougheed TJ, Papagiannakis AT. Viscosity characteristics of rubber-modifiedasphalts. J Mater Civil Eng 1996;8(3):153–6.

[13] West RC, Page GC, Veilleux JG, Choubane B. Effect of tire rubber grindingmethod on asphalt–rubber binder characteristics. Transport Res Record1998:134–40.

[14] Lee S-J, Amirkhanian SN, Shatanawi K, Kim KW. Short-term agingcharacterization of asphalt binders using gel permeation chromatographyand selected Superpave binder tests. Constr Build Mater 2008;22(11):2220–7.

[15] Thodesen C, Shatanawi K, Amirkhanian S. Effect of crumb rubbercharacteristics on crumb rubber modified (CRM) binder viscosity. ConstrBuild Mater 2009;23(1):295–303.

[16] Thodesen C, Xiao F, Amirkhanian SN. Modeling viscosity behavior of crumbrubber modified binders. Constr Build Mater 2009;23(9):3053–62.

[17] Jeong K-D, Lee S-J, Amirkhanian SN, Kim KW. Interaction effects of crumbrubber modified asphalt binders. Constr Build Mater 2010;24(5):824–31.

[18] Cao W-D, Lu W-M. Experimental research on recycled tire rubber modifiedasphalt mixture using hybrid process. Jianzhu Cailiao Xuebao/J Build Mater2007;10(1):110–4 [in Chinese].

[19] Specht LP, Khatchatourian O, Brito LAT, Ceratti JAP. Modeling of asphalt–rubber rotational viscosity by statistical analysis and neural networks. MaterRes 2007;10(1):69–74.

[20] ASTM Standards D6114, 2009. Standard specification for asphalt–rubberbinder. West Conshohocken (PA): ASTM International; 2009.

[21] Hoffmann P, Potgieter CJ. Bitumen rubber chip and spray seals in South Africa.SATC 2007 – 26th annual Southern African transport conference: thechallenges of implementing policy, 225–238.