70

Cohesive failure happens because of a rupture of bonds betweenmolecules in the asphalt film. Adhesive failure happens because ofa rupture of bonds between molecules of different phases. The effectof moisture on the performance of the pavement can be the result ofa combination of both mechanisms.

Bond strength is a critical parameter in evaluating a binder’s abil-ity to resist moisture damage. Adhesion between asphalt and aggre-gate is quantified by the adhesive bond energy (4). Therefore, a testmethod performed directly on the asphalt–aggregate system caneffectively evaluate the influence of water in both cohesive and adhe-sive failure types, leading to a better understanding of the moisturesensitivity of asphalt mixtures (5–7).

Canestrari et al. suggested a repeatable, reliable, and practicalmethod for investigating the adhesion and cohesion properties ofasphalt–aggregate systems based on the pneumatic adhesion tensiletesting instrument (PATTI) (5). However, a standard procedure foraddressing moisture damage in the asphalt–aggregate interface isnot available.

This paper evaluates the use of the bitumen bond strength (BBS)test, which is a modification of the PATTI test, for moisture damagecharacterization of asphalt–aggregate systems. Various base binders,modifications, and aggregate types were used to account for a broadrange of chemical and physical conditions of the asphalt–aggregateinterface. A statistical analysis was performed to verify the repro-ducibility of the BBS in testing the influence of moisture condition-ing time on the bond strength of asphalt–aggregate systems. Acomparison of BBS results and strain sweep tests performed in thedynamic shear rheometer (DSR) was conducted to determine if a testthat takes into account the effect of cyclic loading on moisture dam-age resistance of asphalt–aggregate systems compares reasonablywell with the proposed procedure.

Moisture damage testing of asphalt mixtures is susceptible to vari-ability because of air voids, gradation, degree of saturation, and vol-umetrics. The proposed BBS test measures the bond strength betweentwo materials under well-controlled conditions. Moreover, samples ofthe aggregate–asphalt system are prepared to a high precision level incomparison with that of a more variable asphalt mixture sample.

LITERATURE REVIEW

Adhesion

Adhesion between two different surfaces is defined as the process inwhich dissimilar particles or surfaces are held together by valenceforces or interlocking forces (ASTM D 907-96a). Adhesion deter-mines the tendency of two materials with dissimilar molecules to

Measuring the Effect of Moisture on Asphalt–Aggregate Bond with the Bitumen Bond Strength Test

Raquel Moraes, Raul Velasquez, and Hussain U. Bahia

Understanding moisture damage mechanisms in asphalt pavements andevaluating the right combination of materials that are resistant to mois-ture damage are important. Moisture damage is the loss of strength orstiffness in asphalt mixtures caused by a combination of mechanicalloading and moisture. Many test methods have been developed to eval-uate loss of adhesion and cohesion in binders. However, a simple proce-dure to address moisture damage in the asphalt–aggregate interfaceis not available. The feasibility of the newly developed bitumen bondstrength (BBS) test for moisture damage characterization was inves-tigated. An experimental matrix that included various binders, modi-fications, and aggregates to account for the chemical and physicalconditions in the aggregate–asphalt interface was completed. A statisti-cal analysis was performed to verify reproducibility of the BBS test. Theresults indicated that the bond strength of asphalt–aggregate systemswas highly dependent on modification and moisture exposure time. Poly-mers were found to improve the adhesion between asphalt and aggregateas well as the cohesion within the binder. Results from this study indi-cated that the BBS test was repeatable and reproducible. To further val-idate the effectiveness of the BBS test, a comparison of the BBS test resultsand the modified dynamic shear rheometer strain sweep test was con-ducted. The comparison showed that the BBS test could rank materialssimilarly to a more sophisticated and time-consuming test.

In asphalt mixtures, moisture damage is defined as the loss of stiff-ness and strength caused by moisture exposure under mechanicalloading and manifests itself in a phenomenon referred to as strip-ping. The reduction of the pavement integrity caused by moisturedamage plays an important role in other types of distresses, such asrutting, fatigue cracking, raveling, and potholes. Moisture can accel-erate damage in asphalt mixtures caused by other types of distress(1). The mechanics of the bonding at the aggregate–binder interface,which is highly affected by moisture conditions, influences theresponse of the asphalt mixture to different distresses.

There are three mechanisms by which moisture degrades an asphaltmixture: (a) loss of cohesion within the asphalt mastic, (b) failure ofthe adhesive bond between aggregate and asphalt (i.e., stripping), and(c) degradation of the aggregate (2). Furthermore, water present in themixture can affect its performance by flowing between interconnectedvoids and by remaining static at the interface and voids (3).

Department of Civil and Environmental Engineering, University of Wisconsin–Madison,Madison, WI 53706. Corresponding author: R. Moraes, bringel@wisc.edu.

Transportation Research Record: Journal of the Transportation Research Board,No. 2209, Transportation Research Board of the National Academies, Washington,D.C., 2011, pp. 70–81.DOI: 10.3141/2209-09

Moraes, Velasquez, and Bahia 71

cling to one other (8, p. 301). It can be measured directly with con-tact angle approaches (i.e., wetting potential) (9, p. 267) or withpractical approaches, such as a suitable tensile test (e.g., BBS test).

Asphalt–Aggregate Adhesion Mechanisms

Most likely a combination of mechanisms occurs simultaneously toproduce adhesion. These mechanisms can be classified into one ofthree categories: mechanical interlocking, physicochemical adhe-sion caused by surface free energy of materials, and bonding causedby interfacial chemical reactions (10).

The theories that fundamentally explain the adhesive bond betweenasphalt binder and aggregates are mechanical theory, chemical theory,weak boundary theory, and thermodynamic theory (6, 11).

The mechanical theory indicates that bonding of aggregate binderis affected by physical properties of the aggregate such as porosity,texture, and surface area. The chemical theory suggests that adhe-sion depends on the pH and the functional groups of both the asphaltbinder and the aggregate. The weak boundary theory suggests thatrupture always occurs at the weakest link of the asphalt–aggregateinterface. Finally, the thermodynamic theory studies the attractionbetween aggregate, asphalt, and water caused by differences insurface tension.

These theories and associated mechanisms are not exclusivelyindependent, and many researchers agree that a combination ofmechanisms takes place. Also, several factors affect the adhesion ofthe asphalt binder to the aggregate, including interfacial tensionbetween the asphalt binder and the aggregate, chemical compositionof the asphalt binder and the aggregate, binder viscosity, surfacetexture of the aggregate, aggregate porosity, aggregate cleanli-ness, aggregate temperature, and moisture content at the time of mix-ing (12). Therefore, what has been identified as a major challenge isa system that can effectively measure bond strength and evaluate theeffect of moisture.

Influences on Adhesive Bond Between Asphalt and Aggregate

Effect of Asphalt Binder Characteristics

The asphalt binder characteristics can influence both the adhesionof the asphalt–aggregate system and the cohesion of the mastic.The properties of the asphalt binder that can influence the asphalt–aggregate bond are the chemistry of the asphalt (e.g., polarity andconstitution), viscosity, film thickness, and surface energy (13, 14).The cohesive strength of the asphalt matrix in the presence of mois-ture is also influenced by the chemical nature of the binder andprocessing techniques.

The chemical interaction between the asphalt binder and the aggre-gate is critical to understanding of the capability of compacted bitu-minous mixtures to resist moisture damage. Robertson described thatcarboxylic acids in asphalt binders are quite polar and adhere stronglyto dry aggregate (15). However, this chemical group tends to beremoved easily from aggregate in the presence of water. A reason forthis behavior is that sodium and potassium salts of carboxylic acids inasphalt are essentially surfactants or soaps, which are debondedunder the action of traffic in the presence of water (16). Calciumsalts from hydrated lime are much more resistant to the action ofwater. Robertson also suggested that aged asphalts are more prone

to moisture damage than are unaged asphalts because of the pres-ence of strongly acidic material in oxidized binders (15). Petersenet al. observed that asphalt binders containing ketones and nitrogenare the least susceptible moisture damage (17).

The viscosity of the asphalt binder does play a role in the propen-sity of the asphalt mixture to strip. It has been reported that asphaltswith high viscosity resist displacement by moisture better than dothose that have low viscosity. Asphalts with high viscosity usuallycarry a high concentration of polar functionalities that provide moreresistance to stripping (13).

It has also been reported that bond strength is directly related tofilm thickness. Samples with thicker asphalt film tend to have cohe-sive failure after moisture conditioning. Specimens with thinnerasphalt film have adhesive failure (6).

For surface energy, according to the thermodynamic theory ofasphalt–aggregate adhesion, low values of this property for the asphaltis preferable to provide better wetting.

Effect of Aggregate Characteristics

Aggregate properties have a greater impact on adhesion than someof the binder properties. Size and shape of the aggregate, pore vol-ume and size, surface area, chemical constituents at the surface, acid-ity and alkalinity, adsorption size surface density, and surface chargeor polarity are some of the widely cited aggregate characteristics thatcan influence moisture damage (18).

The chemistry of aggregate affects the asphalt–aggregate adhe-sion substantially; various mineral components of the aggregatesshow different affinity for asphaltic material. When an aggregate isbeing coated with asphalt, the aggregate selectively adsorbs somecomponents of the asphalt. The general trend is that sulfoxides andcarboxylic acids have the greatest affinity for aggregates. It is alsoapparent that aromatic hydrocarbons have much less affinity foraggregate surfaces than the polar groups. Therefore, the type andquantities of the adsorbed components affect the degree of adhesion,and various aggregates develop bonds of different strength (15).

Aggregates are commonly classified as either hydrophilic (i.e.,greater natural affinity for water than for asphalt binder) or hydropho-bic (i.e., greater natural affinity for asphalt than for water) (6, 18, 19).It is commonly known that acidic aggregates are hydrophobic, whereasbasic aggregates are hydrophilic. However, there are notable excep-tions, and the general conclusion is that few if any aggregates cancompletely resist the stripping action of water (18). For example,limestone is classified as hydrophobic aggregate, and granite is con-sidered hydrophilic; however the level of basic or acidic conditionof the limestone and granite aggregates may vary according to theirchemical composition.

Rough surfaces and therefore larger contact area are preferred forbetter adhesive bond. Porosity is another important characteristicof the aggregate that can affect asphalt adsorption. For example,when the asphalt binder coats a rough aggregate surface with finepores, air is trapped and the asphalt has difficulty penetrating thefine pores (20). However, the penetration of asphalt cement intopores is also dependent on the viscosity of the asphalt cement atmixing temperatures.

Moisture and dust can significantly reduce the bond strength ofaggregate–asphalt systems. The presence of dust coatings on theaggregate inhibits complete wetting of the aggregate by the asphaltbinder because the asphalt is adhered to the dust coating and not tothe aggregate itself (21).

72 Transportation Research Record 2209

MATERIALS AND TESTING PROCEDURE

Materials

Three types of aggregates known to have different moisture sensitivity were selected: limestone, granite, and diabase. Twoasphalt binders commonly used in the United States Midwestwere selected for this study: Flint Hills (FH) PG 64-22 and CRM PG 58-28. Four modified asphalt binders were prepared:FH 64-22+acid (PG 70-22), modified with 1% by weight ofpolyphosphoric acid; FH 64-22+elastomer1 (PG 70-22), modi-fied with 0.7% by weight of Elvaloy and 0.17% of polyphos-phoric acid; CRM 58-28+acid (PG 64-28), modified with 1% by weight of polyphosphoric acid; and CRM 58-28+elastomer2 (PG 64-28), modified with 2% by weight of linear styrene butadienestyrene.

Tap water was used to investigate the effects of conditioningmedia on the adhesion between asphalt and aggregate.

BBS Test

The challenge to quantitatively evaluate the adhesive bond betweenasphalt and aggregate is to identify a test that is simple, quick, andrepeatable for evaluating adhesion properties of asphalt–aggregatesystems. Furthermore, no method is included in the Superpave®

binder specifications to evaluate adhesive characteristics of asphaltbinders (2).

Youtcheff and Aurilio used the PATTI test, originally developedfor the coating industry, to measure the moisture susceptibility ofasphalt binders (7). In this study, the BBS test, which is a significantlymodified version of the original PATTI (20), was used to evaluatethe asphalt–aggregate bond strength.

As indicated in Figure 1, the BBS device comprises a portablepneumatic adhesion tester, pressure hose, piston, reaction plate, andmetal pull-out stub. To start the test, the piston is placed over the pull-out stub, and the reaction plate is screwed on it. Then, a pressure hoseis used to introduce compressed air to the piston. During the test, a

(a)

Pullout Stub

Pressure Ring

Pressure Plate

Pullout Stub

Asphalt Binder

Ring Support

Substrate

(b)

FIGURE 1 BBS test.

Moraes, Velasquez, and Bahia 73

pulling force is applied on the specimen by the metal stub. Failureoccurs when the applied stress exceeds the cohesive strength of thebinder or the bond strength of the binder–aggregate interface (i.e.,adhesion). The pull-off tensile strength (POTS) is calculated with

where

Ag = contact area of gasket with reaction plate (mm2),BP = burst pressure (kPa),Aps = area of pull-out stub (mm2), andC = piston constant.

The pull-out stub has a rough surface that can prevent asphaltdebonding from the stub surface by providing mechanical interlockand larger contact area between the asphalt binder and stub (Figure 2).The pull-out stub in the BBS test has a diameter of 20 mm and hasa surrounding edge, used to control film thickness. The stub edgehas a thickness of 800 µm (Figure 2). This new geometry and sur-face treatment were developed in two extensive recent studies toimprove repeatability of the testing system (5, 22).

Aggregate Sample Preparation

Aggregate plates were cut with similar thickness and parallel topand bottom surfaces. After cutting and lapping, aggregates plateswere immersed in distilled water in an ultrasonic cleaner for 60 minat 60°C to remove any residue from the cutting process and neutral-ize the surface of aggregate to its original condition. Lapping is doneto control the roughness of the surface.

Asphalt Sample Preparation

The aggregate surface and pull-out stubs are degreased with acetoneto remove moisture and dust, which could affect adhesion. Then thepull-out stubs and aggregate plates are heated in an oven at 65°C fora minimum of 30 min to remove absorbed water on the aggregatesurface and provide a better bond between the asphalt binder and theaggregate. The asphalt binders are heated in an oven at 150°C. Thestubs are removed from the oven, and an asphalt binder sample is

POTSBP

(1)ps

=×( ) −A C

Ag

placed immediately on the surface of the stub for approximately 10 s.Then the aggregate plate is removed from the oven, and the stubwith the asphalt sample is pressed firmly into the aggregate surfaceuntil the stub reaches the surface and no excess of asphalt binder isobserved to be flowing. The stubs must be pushed down as straightas possible, and twisting should be avoided to reduce the formationof trapped air bubbles inside the sample and to minimize stresses.

Before testing, dry samples are left at room temperature for 24 h.For wet conditioning, samples are first left at room temperature for1 h to allow for the aggregate–binder–stub system to reach a stabletemperature. Then, samples are submerged in a water tank at 40°C forthe specified conditioning time. When conditioning time is completed,samples are kept at room temperature for 1 h before testing.

Testing Procedure

The BBS testing procedure is summarized in the following steps:

• Before testing, check the air supply and pressure hose connection.• Set the rate of loading to 100 psi/s. Measure the sample tem-

perature with a thermometer before starting the test.• Place a circular spacer under the piston to make sure that the pull-

off system is straight and that eccentricity of the stub is minimized.• Carefully place the piston around the stubs to rest on the spacers

so as not to disturb the stub or to induce unnecessary strain in thesample. Screw the reaction plate into the stub until the pressure platejust touches the piston.

• Apply pressure at a specified rate.• After testing, the maximum pull-off tension is recorded and the

failure type is observed. If more than 50% of the aggregate surfaceis exposed, then failure is considered to be adhesive; otherwise, it isa cohesive failure.

Modified DSR Strain Sweep Test

The DSR can accurately control temperature and mode and time ofloading. Therefore, the DSR can be used to measure the rheologicalresponses (e.g., shear stresses and complex modulus) of asphalt filmsadhering to aggregate surfaces in dry and moisture conditions. Chomodified the DSR strain sweep test procedure to evaluate the rheo-logical properties of the asphalt–aggregate interface before and afterwater conditioning (23).

22

20

6

5

40

0.820

7

22

15

FIGURE 2 Pull-out stub for BBS test.

74 Transportation Research Record 2209

A cored rock disk 25 mm in diameter and 5 mm thick is used asthe substrate for adhering asphalts (Figure 3). The disk and asphaltbinder simulate the asphalt–aggregate interface in asphalt mixtures.The rock disk is glued on the DSR base metal plate. In the DSRsetup, parallelism is obtained by aligning the disk using the metalDSR top spindle while the epoxy binder dries. A water cup, fabri-cated specially for the DSR (Figure 3), is used to allow continuouswater access to the interface (23). Rheological responses are mea-sured with oscillated loads of 1% to 100% strain sweep with 1.6 Hzfrequency (i.e., 10 rad/s), at 40°C, in both dry and wet (i.e., using tapwater) conditions (23).

ANALYSIS OF BBS TEST RESULTS

Effect of Conditioning Time

In this experiment, samples were conditioned in tap water for 0, 6,24, 48, and 96 h. Moisture damage is a time-dependent phenome-non, and an indirect way to investigate this time-dependency behav-ior is to measure the variation in the bond strength with time in thepresence of water. All the reactions involved in this process havedifferent and unknown kinetics, and therefore the selected condi-tioning times are considered appropriate. The effect of conditioningtime on POTS for the asphalt–aggregate systems tested with the BBSis shown in Figure 4.

The average pull-off strength was calculated from four replicates.As shown in the charts, the conditioning of specimens in water causeda significant reduction in strength and, in some cases, a change in fail-ure mode from cohesive to adhesive type (Table 1). The change infailure mode is expected since water penetrates through the aggregate,which is a porous material, and hence weakens the bond at theinterface (8, p. 301). The longer the conditioning time is in water,the weaker the interface bond and the lower the pull-off strengthvalue observed.

Two general trends are observed for the effect of conditioningtime on POTS (Figure 5). One is continuous reduction of the mag-nitude of POTS with conditioning time in water for a number of thesystems tested, including the following combinations of binder and

aggregate: FH 64-22 neat, limestone; FH 64-22+acid, granite; CRM58-28+elastomer2, granite; and CRM 58-28+elastomer2, limestone.The other trend corresponds to a reduction on the POTS up to a spe-cific asymptote. This trend is seen for conditioning times in the 24-to 48-h range for the following binder–aggregate systems: FH 64-22 neat, granite; FH 64-22+acid, limestone; FH 64-22+elastomer1,granite; FH 64-22+elastomer1, limestone; and CRM 58-28+acid,limestone.

The bond strength for the CRM 58-28+acid binder and granite didnot significantly change with conditioning time (Figure 5). This resultindicates that polyphosphoric acid (PPA) may induce effects thatreduce the moisture susceptibility of this aggregate–binder interface.

Effect of Asphalt Modification

The effects of modification of asphalt on POTS values are clearlydetected by the BBS testing results. For example, Figure 4 showsthat the modified FH 64-22 binders have higher dry average POTSin comparison with the neat binder for both granite and limestoneaggregates.

The asphalts modified with PPA show less susceptibility to mois-ture conditioning in comparison with neat asphalts. The effect ofPPA is better in granite than in limestone aggregates, because of theacidic nature of the granite aggregate. Asphalt binder modifiedwith Elvaloy also shows moisture resistance improvements forthe granite case compared with the neat asphalt. However, for thelimestone case, no significant difference between FH 64-22 neat andFH 64-22+elastomer1 were observed.

Failure mechanisms are also affected by modification type.Table 1 indicates that failure type (i.e., cohesive and adhesive failure)changes according to modification, aggregate type, and conditioningtime. All unconditioned (i.e., dry) samples showed cohesive failure(i.e., failure within asphalt). Adhesive failure (i.e., between aggregateand binder) was observed for some conditioned specimens.

The results also show that the failure type after 6 h of conditioningtime for the FH 64-22 asphalt changes from adhesive to cohesivewhen PPA is used as modification. These observations indicate thatPPA improves the bond of the interface between the asphalt and the

(a) (b)

FIGURE 3 DSR testing apparatus: (a) sample preparation and (b) wet conditioning (23).

Moraes, Velasquez, and Bahia 75

GRANITE

LIMESTONE

GRANITE

LIMESTONE

GRANITE

LIMESTONE

GRANITE

LIMESTONE

CRM 58-2

2+Elas

tomer

2

FH 64-2

2+Elas

tomer

1

FH 64-2

2+Acid

FH 64-2

2

CRM 58-2

2+Acid

CRM 58-2

8

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

CRM 58-2

2+Elas

tomer

2

FH 64-2

2+Elas

tomer

1

FH 64-2

2+Acid

FH 64-2

2

CRM 58-2

2+Acid

CRM 58-2

8

0.0

(a) (b)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

CRM 58-2

2+Elas

tomer

2

FH 64-2

2+Elas

tomer

1

FH 64-2

2+Acid

FH 64-2

2

CRM 58-2

2+Acid

CRM 58-2

8

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

CRM 58-2

2+Elas

tomer

2

FH 64-2

2+Elas

tomer

1

FH 64-2

2+Acid

FH 64-2

2

CRM 58-2

2+Acid

CRM 58-2

8

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

CRM 58-2

2+Elas

tomer

2

FH 64-2

2+Elas

tomer

1

FH 64-2

2+Acid

FH 64-2

2

CRM 58-2

2+Acid

CRM 58-2

8

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

GRANITE

LIMESTONE

(d)(c)

(e)

FIGURE 4 Influence of conditioning time on POTS for various asphalt–aggregate systems: (a) 0 h, (b) 6 h, (c) 24 h, (d ) 48 h, and(e) 96 h.

76 Transportation Research Record 2209

granite. All samples containing PPA have cohesive failure, whichindicates that the bond at the aggregate–binder interface is greater thanthe cohesive strength of the binder at the specified testing conditions.These observations cannot be generalized to all combinations ofasphalts and granites. These observations confirm that the BBS test is asystem that can detect differences in bond strength and its change withwater conditioning for various combinations of binder modification.

Effect of Aggregate Type

The nature and chemical characteristics of aggregates greatly affectbond strength and failure mechanisms of asphalt–aggregate systems asindicated by Table 1. On both limestone and granite surfaces, the fail-ure mode changed after moisture exposure, showing that the nature ofthe aggregate greatly affects adhesion.

For all limestone samples, the failure type was cohesive, whichindicates that the adhesive bond in the asphalt–aggregate interfaceis larger than the cohesive strength of the binders. Also, Figure 4indicates that limestone aggregates have higher adhesive bond toasphalt than granite aggregates and thus more resistance to adhesivefailure.

The POTS obtained from performed BBS tests is highly influ-enced by the cleanness of the surface of the aggregate plate. Incon-sistent and unexpected results for some of the samples conditioned

at 48 and 96 h were obtained when the aggregate plates used weredifferent from the plates used for the 0-, 6-, and 24-h tests. It appearsthat slight changes of the aggregate surface can greatly affect themagnitude of the POTS. Therefore, it is important to perform mois-ture susceptibility experiments with aggregates from the same sourceand to be consistent in sample preparation.

BBS Reproducibility

The effect of conditioning time (0, 6, and 24 h) on the pull-offstrength of the asphalt–aggregate systems tested by different oper-ators is shown in Figure 6. As can be seen, the values of pull-offstrength for each aggregate–binder system were similar for bothoperators.

Statistical analyses were performed to evaluate the reproducibil-ity of the BBS test. Specifically, test of hypotheses were used todetermine if there is any statistically significant difference betweenthe means of the POTS obtained with two operators. A two-tailedtest was used for the test of hypotheses. The following null (Ho) andalternative hypotheses (Ha) were used with α = 0.05 (Type I error,reject Ho when Ho is true):

with the test statistic

where–X1 = estimation of the population mean of the POTS for Opera-

tor 1,X–

2 = estimation of the population mean of the POTS for Opera-tor 2,

n1 = number of replicates tested by Operator 1 to estimate popu-lation mean,

n2 = number of replicates tested by Operator 2 to estimate popu-lation mean, and

Sp2 = variance pooled estimator.

The variance pooled estimator (24) can be calculated with

where s12 is the calculated variance for the POTS of Operator 1 and

s22 is the calculated variance for the POTS of Operator 2.

The test of hypotheses described has the following rejectionregion (values of the t-test statistic for which Ho is rejected): t < −tα

and t > tα, where tα are based on n1 + n2 − 2 degrees of freedom andthe selected Type I error (α).

Table 2 shows the results of the statistical analysis of the opera-tor variability of the BBS test data for all asphalt–aggregate systems.

Sn s n s

n np2 1 1

22 2

2

1 2

1 1

24=

−( ) + −( )+ −

( )

tX X

Sn np

= −

+⎛⎝⎜

⎞⎠⎟

1 2

2

1 2

1 13( )

H

Ha

0

1 2

1 2

0

0

:

:

μ μ

μ μ

− =

− ≠

(2)

TABLE 1 Failure Mode in BBS Testing

Failure Type

Asphalt Binder Type CTa (h) Granite Limestone

FH 64-22 neat Dry Cohesion Cohesion6 Adhesion Cohesion

24 Adhesion Cohesion48 50%A–50%C 50%A–50%C96 Cohesion Cohesion

FH 64-22+elastomer1 Dry Cohesion Cohesion6 Cohesion Cohesion

24 Adhesion Cohesion48 Cohesion 50%A–50%C96 Cohesion Cohesion

FH 64-22+acid Dry Cohesion Cohesion6 Cohesion Cohesion

24 Cohesion Cohesion48 Cohesion Cohesion96 Cohesion Cohesion

CRM 58-28 neat Dry Cohesion Cohesion6 Cohesion Cohesion

24 Cohesion Cohesion48 Adhesion Cohesion96 Adhesion Adhesion

CRM 58-28+elastomer2 Dry Cohesion Cohesion6 Adhesion Cohesion

24 Adhesion Cohesion48 50%A–50%C Adhesion96 Adhesion Adhesion

CRM 58-28+acid Dry Cohesion Cohesion6 Cohesion Cohesion

24 Cohesion Cohesion48 Cohesion Cohesion96 Cohesion Cohesion

aConditioning time.

Moraes, Velasquez, and Bahia 77

The average pull-off strength was calculated from three replicatesfor each operator and α = 0.05 is used.

Generally, the BBS test is not sensitive to the operator perform-ing the test. Only in two conditions was the average of the POTS sta-tistically different between operators: FH 64-22 granite at 0 h andFH 64-22+elastomer1 diabase at 6 h of conditioning time. Figure 7ashows one of the two cases in which Ho was rejected for α = 0.05.Figure 7b is an example of the very similar probability distributionsobtained for the BBS test results by different operators.

COMPARISON OF BBS AND MODIFIED DSRSTRAIN SWEEP TESTS

A few tests were performed to compare moisture susceptibility mea-surements obtained from the newly developed BBS test and a moretime-consuming DSR strain sweep test. Note that the DSR can sim-

ulate the cyclic nature of the stresses applied by the pore pressureunder moving traffic. This phenomenon cannot be simulated by anyother available device. This traffic-induced pressure is one of themost important factors in moisture damage of asphalt mixtures.Three different asphalt binders (CRM 58-28 neat, CRM 58-28+elas-tomer2, and FH 64-22+elastomer1) were tested with granite asaggregate substrate in both the BBS and the DSR. The strain sweeptest was performed by following the previously described procedure(23). Two replicates were tested for each condition and test type.

Figure 8 shows a typical result obtained from the DSR strainsweep procedure. It can be seen that water conditioning significantlyaffects the rheological properties of asphalt–aggregate systems.

Comparison of the BBS and DSR procedures involved the calcu-lation of the percent loss of a specific property after water condition-ing for 6 h. In the case of the BBS test, the POTS was used tocalculate moisture susceptibility. For the DSR strain sweep test, thecomplex modulus �G*� at a strain of 1% was selected as the parameter.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 20 40 60 80 120100

Pu

ll-o

ff S

tren

gth

(M

Pa)

Time (h)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

0 20 40 60 80 100 120

Time (h)

(a) (b)

(c) (d)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

0 20 40 60 80 100 120

Time (h)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 20 40 60 80 120100

Pu

ll-o

ff S

tren

gth

(M

Pa)

Time (h)

FIGURE 5 Effect of large conditioning times on POTS for FH 64-22 neat and CRM 58-28�acid: (a) FH 64-22 neat, granite; (b) FH 64-22 neat, limestone; (c) CRM 58-28�acid, granite; and (d ) CRM 58-28�acid, limestone.

78 Transportation Research Record 2209

Similar results were obtained when the complex modulus at higherstrain levels (i.e., γ = 100%) was selected. The following equationwas used to compute moisture damage in the BBS test:

where POTSDRY and POTSWET are the pull of tensile strength fromBBS test in dry and moisture conditioning after 6 h, respectively. Tocalculate the moisture damage in the strain sweep test, the followingequation was used:

%Loss_After_Moisture_BBSPOTS

POTSWET

DRY

= −⎛⎝1⎜⎜

⎞⎠⎟

× 100 5( ) where �G*�γ =1%,_WET and �G*�γ =1%,_DRY are the complex modulus at ashear strain of 1% from DSR testing in dry and moisture conditioningafter 6 h, respectively.

Table 3 shows the results for the three asphalt–aggregate systemstested in the BBS and DSR. It can be seen that the moisture suscep-tibility ranking from the BBS test and the DSR strain sweep test arethe same.

% %,_Loss_After_Moisture_DSRG

GWET= − =

=

1 1

1

�

�γ

γ %%,_

( )DRY

⎛

⎝⎜

⎞

⎠⎟ × 100 6

Operator 1

Operator 2

Operator 1

Operator 2

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5P

ull-

off

Str

eng

th (

MP

a)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Pu

ll-o

ff S

tren

gth

(M

Pa)

Operator 1

Operator 2

(a)

FH 6

4-22

+Elas

tom

er1

(diab

ase)

FH 6

4-22

+Elas

tom

er1

(gra

nite)

FH 6

4-22

(diab

ase)

FH 6

4-22

(gra

nite)

(c)

FH 6

4-22

+Elas

tom

er1

(diab

ase)

FH 6

4-22

+Elas

tom

er1

(gra

nite)

FH 6

4-22

(diab

ase)

FH 6

4-22

(gra

nite)

(b)

FH 6

4-22

+Elas

tom

er1

(diab

ase)

FH 6

4-22

+Elas

tom

er1

(gra

nite)

FH 6

4-22

(diab

ase)

FH 6

4-22

(gra

nite)

FIGURE 6 Influence of operators in testing effect of large conditioning on POTS: (a) 0 h, (b) 6 h, and (c) 24 h.

Moraes, Velasquez, and Bahia 79

• The POTS value decreases when samples are conditioned inwater, regardless of the selected asphalt binder or aggregate type.In general, POTS measurements for the dry samples have lowercoefficient of variation than for the samples tested after waterconditioning.

• In some cases, conditioning of specimens in water causes notonly loss of POTS but also a change in the failure mechanism. Inabsence of water, failure usually happens within the asphalt (i.e.,cohesive failure). After water conditioning, the failure changes fromtotal cohesive to adhesive failure.

• Bonding between asphalt and aggregate under wet conditions ishighly dependent on binder modification type and conditioning time.

TABLE 2 Statistical Analysis of Reproducibility of BBS Test Data

Operator 1 Operator 2

Material u1 σ CV (%) u2 σ CV (%) Sp2 t-statistic tα −tα Result

FH 64-22, granite (0 h) 2.110 0.040 1.90 1.937 0.040 2.07 0.00 −5.297 2.776 −2.776 Reject Ho

FH 64-22, diabase (0 h) 2.058 0.080 3.89 1.932 0.030 1.55 0.00 −2.554 Accept Ho

FH 64-22+elastomer1, 2.150 0.030 1.40 2.164 0.030 1.39 0.00 0.572 Accept Ho

granite (0 h)

FH 64-22+elastomer1, 1.966 0.040 2.03 2.024 0.020 0.99 0.00 2.246 Accept Ho

diabase (0 h)

FH 64-22, granite (6 h) 1.284 0.010 0.78 1.388 0.120 8.65 0.01 1.496 Accept Ho

FH 64-22, diabase (6 h) 2.017 0.050 2.48 1.925 0.050 2.60 0.00 −2.254 Accept Ho

FH 64-22+elastomer1, 1.787 0.260 14.55 1.650 0.100 6.06 0.04 −0.852 Accept Ho

granite (6 h)

FH 64-22+elastomer1, 2.387 0.040 1.68 1.930 0.190 9.84 0.02 −4.077 Reject Ho

diabase (6 h)

FH 64-22, granite (24 h) 1.321 0.260 19.68 1.305 0.090 6.90 0.04 −0.101 Accept Ho

FH 64-22, diabase (24 h) 1.981 0.050 2.52 1.880 0.040 2.13 0.00 −2.732 Accept Ho

FH 64-22+elastomer1, 1.644 0.120 7.30 1.656 0.130 7.85 0.02 0.117 Accept Ho

granite (24 h)

FH 64-22+elastomer1, 2.298 0.100 4.35 2.127 0.090 4.23 0.01 −2.201 Accept Ho

diabase (24 h)

NOTE: u1 = population mean of the pull-off tensile strength for Operator 1; u2 = population mean of the pull-off tensile strength for Operator 2.

0

4

8

12

16

20

p (

po

ts)

0 1 2 3 4 5

Pull-off Tensile Strength (MPa)

Operator 1-Granite

Operator 2-Granite

FH 64-22+Elastomer1 (0 hours)

(b)

0

4

8

12

16

20

p (

po

ts)

Operator 1-Diabase

Operator 2-Diabase

FH 64-22+Elastomer1 (6 hours)

0 1 2 3 4 5

Pull-off Tensile Strength (MPa)

(a)

FIGURE 7 Distributions of BBS test results from two operators: (a) hypothesis was rejected for � � 0.05 and (b) hypothesis was acceptedfor � � 0.05.

SUMMARY AND CONCLUSIONS

A comprehensive experimental test matrix was tested to investigatethe feasibility of the BBS test for moisture damage characterization.Various base binders, modifications, and aggregate types were usedto account for a broad range of chemical and physical conditions ofthe asphalt–aggregate interface. From the results and analyses, thefollowing conclusions can be drawn:

• The BBS test can successfully measure the effects of moistureconditioning time and modification on the bond strength of asphalt–aggregate systems.

80 Transportation Research Record 2209

• Polymers improve the adhesion between the asphalt andaggregate as well as the cohesion within the binder.

• PPA significantly improves the moisture resistance of asphalt–aggregate systems tested in this study. The effect is especially noticedfor granite or acidic aggregates. All samples containing PPA have acohesive failure, which indicates that the bond at the aggregate–binderinterface is greater than the cohesive strength of the binder.

• Statistical analysis indicates that the BBS test is repeatableand reproducible. No significant differences between the resultsobtained by different operators were observed. Therefore, the BBStest can be used as a practical method to measure bond strengthbetween aggregate and binders in dry and moisture conditions.

• Limited results on the validation and verification of the BBStest procedure with the modified DSR strain sweep test indicatethat the BBS test can rank the asphalt–aggregate systems for mois-ture damage similarly to tests that can simulate the cyclic natureof traffic-induced stresses. Results are preliminary, and a moreextensive test matrix needs to be performed for such comparison.The BBS test is clearly a simpler and more practical test. The com-parison is needed only for validation of results in terms of rankingof aggregate–binder systems.

ACKNOWLEDGMENTS

This research was sponsored by the Asphalt Research Consortium,which is managed by FHWA and Western Research Institute. Theauthors thank Jia Meng for her contributions to this paper.

REFERENCES

1. Cho, D.-W., and K. Kim. The Mechanisms of Moisture Damage inAsphalt Pavement by Applying Chemistry Aspects. KSCE Journal ofCivil Engineering, Vol. 14, No. 3, 2010, pp. 333–342.

2. Copeland, A. R., J. S. Youtcheff, Jr., and A. Shenoy. Moisture Sensitiv-ity of Modified Asphalt Binders: Factors Influencing Bond Strength. InTransportation Research Record: Journal of the Transportation ResearchBoard, No. 1998, Transportation Research Board of the National Acad-emies, Washington, D.C., 2007, pp. 18–28.

3. Kringos, N., and A. Scarpas. Raveling of Asphaltic Mixes Due to WaterDamage: Computational Identification of Controlling Parameters. InTransportation Research Record: Journal of the Transportation ResearchBoard, No. 1929, Transportation Research Board of the National Acad-emies, Washington, D.C., 2005, pp. 79–87.

4. Bhasin, A., E. A. Masad, D. N. Little, and R. L. Lytton. Limits onAdhesive Bond Energy for Improved Resistance of Hot-Mix Asphalt toMoisture Damage. In Transportation Research Record: Journal of theTransportation Research Board, No. 1970, Transportation ResearchBoard of the National Academies, Washington, D.C., 2006, pp. 3–13.

5. Canestrari, F., F. Cardone, A. Graziani, F. Santagata, and H. U. Bahia.Adhesive and Cohesive Properties of Asphalt-Aggregate Systems Sub-jected to Moisture Damage. Road Materials and Pavement Design,Vol. 11, 2010, pp. 11–32.

6. Kanitpong, K., and H. U. Bahia. Role of Adhesion and Thin Film Tack-iness of Asphalt Binders in Moisture Damage of HMA. Asphalt PavingTechnology, Vol. 72, 2003, pp. 502–528.

7. Youtcheff, J., and V. Aurilio. Moisture Sensitivity of Asphalt Binders:Evaluation and Modeling of the Pneumatic Adhesion Test Results.Proc., 42nd Annual Conference of Canadian Technical Asphalt Associ-ation, 1997, pp. 180–200.

8. Packham, D. E. Handbook of Adhesion. Longman Scientific & Techni-cal, Harlow, United Kingdom, 1992, 2005.

9. Petrie, E. M. Handbook of Adhesives and Sealants. McGraw–Hill,New York, 1999.

10. Bhasin, A., D. N. Little, K. L. Vasconcelos, and E. A. Masad. SurfaceFree Energy to Identify Moisture Sensitivity of Materials for AsphaltMixes. In Transportation Research Record: Journal of the Trans-portation Research Board, No. 2001, Transportation Research Boardof the National Academies, Washington, D.C., 2007, pp. 37–45.

11. Terrel, R. L., and J. W. Shute. SHRP-A/IR-89-003: Summary Report onWater Sensitivity. National Research Council, Washington, D.C., 1989.

12. Johnson, D. R., and R. B. Freeman. Rehabilitation Techniques forStripped Asphalt Pavements. FHWA/MT-002-003/8123. Western Trans-portation Institute, Bozeman, Mont., 2002.

13. Bahia, H. U., A. Hanz, K. Kanitpong, and H. Wen. Testing Methods toDetermine Aggregate/Asphalt Adhesion Properties and Potential Mois-

0.0E+00

2.0E+04

4.0E+04

6.0E+04

8.0E+04

1.0E+05

1.2E+05

0 10 20 30 40 50 60 70 80 90 100

G*

(P

a)

Strain (%)

FH 64-22, Elastomer+Acid , 25mm, T = 40°C , granite

Dry Moisture Conditioned

FIGURE 8 Strain sweep test in DSR for FH 64-22�acid and granite in dryand wet conditions.

TABLE 3 Ranking Moisture Susceptibility

Asphalt–Aggregate %Loss_ %Loss_ Ranking RankingSystem Description DSR BBS DSR BBS

CRM 58-28 neat granite 10 24 2 2

CRM 58-28+elastomer2 8 22 1 1granite

FH 64-22+elastomer1 23 26 3 3granite

Moraes, Velasquez, and Bahia 81

ture Damage. WHRP 07-02. Wisconsin Highway Research Program,Madison, 2007.

14. Xiao, F. P., S. N. Amirkhanian, and C. H. Juang. Rutting Resistanceof Rubberized Asphalt Concrete Pavements Containing ReclaimedAsphalt Pavement Mixtures. Journal of Materials in Civil Engineering,Vol. 19, 2007, pp. 475–483.

15. Robertson, R. E. SHRP-A/UWP-91-510: Chemical Properties of Asphaltsand Their Relationship to Pavement Performance. National ResearchCouncil, Washington, D.C., 1991.

16. Plancher, H., S. M. Dorrence, and J. C. Petersen. Identification ofChemical Types in Asphalts Strongly Absorbed at Asphalt-AggregateInterface and Their Relative Displacement by Water. Proceedings of the Association of Asphalt Paving Technologists, Vol. 46, 1977,pp. 151–175.

17. Petersen, J. C., H. Plancher, E. K. Ensley, R. L. Venable, and G.Miyake. Chemistry of Asphalt–Aggregate Interaction: Relationshipwith Pavement Moisture-Damage Prediction Test. In TransportationResearch Record 843, TRB, National Research Council, Washington,D.C., 1982, pp. 95–104.

18. Tarrer, A. R., and V. Wagh. SHRP-A/UIR-91-507: The Effect of thePhysical and Chemical Characteristics of the Aggregate on Bonding.National Research Council, Washington, D.C., 1992.

19. Bhasin, A. Development of Methods to Quantify Bitumen-AggregateAdhesion and Loss of Adhesion Due to Water. PhD thesis. Texas A&MUniversity, College Station, 2006.

20. Yoon, H. H., and A. R. Tarrer. Effect of Aggregate Properties on Strip-ping. In Transportation Research Record 1171, TRB, National ResearchCouncil, Washington, D.C., 1988, pp. 37–43.

21. Castan, M. Rising of Binder to the Surface of an Open-Graded Bitumi-nous Mix. Bulletin de Liaison des Laboratoires Routiers, No. 33, 1968,pp. 77–84.

22. Meng, J. Affinity of Asphalt to Mineral Aggregate: Pull-off Test Evalu-ation. MS thesis. University of Wisconsin, Madison, 2010.

23. Cho, D.-W. Study of Asphalt-Aggregate Bond in Simple SimulationUsing the Dynamic Shear Rheometer. PhD thesis. University of Wis-consin, Madison, 2008.

24. McClave, J., and T. Sincich. Statistics, 9th ed. Prentice Hall, EnglewoodCliffs, N.J., 2003.

The results and opinions presented are those of the authors and do not neces-sarily reflect those of the sponsoring agencies.

The Characteristics of Asphalt–Aggregate Combinations to Meet SurfaceRequirements Committee peer-reviewed this paper.