Moisture Susceptibility of Asphalt

Mixtures with Known Field Performance

Evaluated with Dynamic Analysis and Crack Growth Model

Edith Arambula, Eyad Masad, and Amy Epps Martin

Moisture damage in asphalt mixtures refers to loss in strength and asphalt mixture. The chemical composition of the aggregate and its
durability due to the presence of water. The level and the extent of mois- affinity to the asphalt binder influence the severity of stripping (3).
ture damage, also called moisture susceptibility, depend on environ- In addition, the aggregate surface characteristics such as roughness
mental, construction, and pavement design factors; internal structure and texture also play an important role in the occurrence of stripping,
distribution; and the quality and type of materials used in the asphalt with smooth and rounded aggregates exhibiting greater stripping
mixture. This study evaluates the moisture susceptibility of asphalt mix- potential (4).
tures with known field performance using dynamic analysis and a crack In the past, researchers focused on evaluating moisture damage
growth model to characterize the asphalt mixtures and corresponding by performing laboratory tests on asphalt mixtures, which indirectly
asphalt mastics. The model parameters were obtained from surface measured damage either qualitatively or quantitatively. In the 1920s
energy measurements, uniaxial dynamic testing for the asphalt mixtures, the tests to evaluate moisture damage were the static immersion test
and dynamic shear testing for the asphalt mastics. Results showed and the boiling test (5). Both tests, applied to loose asphalt mixtures,
good differentiation between the moisture-conditioned (wet) and were based on visual inspection of the specimens after moisture
unconditioned (dry) specimen behavior and provided a good correlation conditioning to determine the degree of damage. Consequently,
with the reported field performance of the asphalt mixtures. the evaluation of damage in these qualitative tests was subjective.
During the 1940s and 1950s, quantitative test methods performed
on compacted asphalt mixtures were introduced. In those methods,
Many distresses that occur in asphalt pavements relate to a phenom-
the ratio of results on unconditioned (dry) and moisture conditioned
enon called moisture damage, a term that refers to the detrimental
(wet) specimens was used to determine moisture susceptibility (5).
effects of moisture (water in liquid and vapor form) in the pavement.
The level and extent of the damage relates to the environmental During the 1960s and 1970s greater understanding of moisture
conditions, quality of construction techniques, adequacy of pave- damage mechanisms resulted in the development of different test
ment structure and mixture design, type of aggregate and asphalt protocols, including the modified Lottman test. This procedure is
binder used, and internal void characteristics of the asphalt mixture. still widely used and required by the state-of-the-practice Superpave®
Although not necessarily initiated by the presence of moisture, most system (6). An indirect tensile test and a ratio of wet to dry strength
distresses increase their extent and severity as a result of its presence, greater than or equal to 0.8 is required for an adequate performance.
which causes a loss of bond between the aggregate and the asphalt In the 1980s and 1990s new test procedures were developed, includ-
binder (1). ing the environmental conditioning system and the Hamburg wheel
Two of the principal mechanisms that induce moisture damage tracking device (5).
in the asphalt mixture are advective flow and water diffusion (2). The reduction of strength in the asphalt mixture after moisture
Advective flow occurs when water flowing through the voids of the conditioning is related to the air void content and structure. On the
asphalt mixture causes desorption of the outer layers of the asphalt basis of this observation, researchers classified asphalt mixtures
mastic (mixture of the aggregate portion smaller than 1.18 mm and the as impermeable, pessimum voids (7% to 13% air voids), and free
asphalt binder), ultimately breaking the bond between the asphalt drainage (7 ). The pessimum voids range, in which maximum mois-
mastic and the aggregate. Diffusion occurs when water coming from ture damage occurred, was further explored by analyzing asphalt
an underground source or moisture from the environment permeates mixtures with different gradations but with the same air void con-
through the asphalt mastic, diminishing its cohesive bond strength. tent. The average void diameter was determined using X-ray com-
When the asphalt mastic coating the aggregate is completely dis- puted tomography (CT) and correlated with parameters such as the
placed by water, a distress called stripping becomes visible in the energy ratio and the number of cycles to failure (8, 9). The observed
shape of the correlation curves was the same as the one obtained
by Terrel and Al-Swailmi (7 ), indicating that a pessimum size con-
E. Arambula, Room 601A; E. Masad, Room 503H; and A. E. Martin, Room 503F,
Zachry Department of Civil Engineering, Texas A&M University, CE/TTI Building,
cept also exists, a specific range of air void sizes maximizes moisture
3136 TAMU, College Station, TX 77843-3136. Corresponding author: E. Masad, damage.
[email protected]. A crack growth model, which includes several mechanical and
surface energy properties, was recently developed to evaluate the
Transportation Research Record: Journal of the Transportation Research Board,
No. 2001, Transportation Research Board of the National Academies, Washington,
performance and moisture sensitivity of asphalt mixtures and asphalt
D.C., 2007, pp. 20–28. mastics. This model gave promising results in distinguishing wet
DOI: 10.3141/2001-03 and dry behavior and the effect of air void distribution (10, 11).

20
Arambula, Masad, and Martin 21

OBJECTIVES AND TASKS final mixture proportion of 65% gravel, 17.5% limestone, and 17.5%
natural sand. The asphalt binder content was 5.0% by weight of
The primary objective of this study was to examine the validity and the mixture.
applicability of the crack growth model in predicting the moisture The Ohio DOT performed a field evaluation of the pavement sec-
susceptibility of asphalt mixtures with known field performance. tions in which Mixtures B and C were used. The reported pavement
This objective was achieved by the following: condition rating for Mixture B dropped from 95 to 89 from 2000 to
2004, and for Mixture C it dropped from 97 to 60 from 1998 to 2004.
1. Dynamic testing of asphalt mixtures,
Extensive amounts of cracking and raveling were observed in these
2. Analysis to determine crack growth model parameters,
pavements, especially in the sections in which Mixture C was applied.
3. Surface energy measurements of the aggregates and asphalt
According to the Ohio DOT, the primary cause of the moisture-induced
binders,
distresses was the type of gravel aggregate used (10).
4. Dynamic testing and analysis of asphalt mastics, and
5. Comparison of the analysis results with field performance.
FRACTURE MODEL FOR
Subsequent sections describe the methodology and results of this MOISTURE SUSCEPTIBILITY
study. The first section describes the characteristics of the selected
asphalt mixtures. The next section includes a brief description of the A fracture model was developed to assess the moisture susceptibility
crack growth model with details provided elsewhere (11). Then, a of asphalt mixtures (11). The equation used to estimate the crack
description of the materials characterization and the mechanical and growth describes the average crack growth with respect to the change
surface energy tests performed to obtain the inputs for the model is in pseudostrain energy per unit volume or J-integral:
provided. The following section includes the results, and a summary
dr
= A ( JR )
and conclusions section completes the paper. n
(1)
dN
DESCRIPTION OF MIXTURES where
AND FIELD PERFORMANCE
r = average crack radius in the specimen,
A description of the mixtures selected for this study in regard to their N = number of cycles,
type, location, aggregate and asphalt binder type, and reported field n and A = material constants, and
performance is presented in Table 1. The aggregate gradations are JR = J-integral, the pseudostrain energy release rate per unit
presented in Figure 1. The in-place air voids required for dense-graded crack area expressed as
mixtures in Texas is from 5% to 9%, and in Ohio the requirement is
approximately 4% to 8% (12, 13). ∂WR
Mixture A was used by the Atlanta District of the Texas Department JR = ∂N (2)
of Transportation (DOT) in Harrison County in the construction of ∂ ( csa )
Interstate Highway 20. The mixture design included 67% siliceous ∂N
river gravel, 32% limestone screenings, and 1% hydrated lime. The
asphalt binder content was 5.0% by weight of the mixture. On the basis where WR is the dissipated pseudostrain energy (DPSE) per unit vol-
of field and laboratory test results, Mixture A exhibited overall good ume of the intact material or the volume of the material that is capable
performance and did not exhibit evidence of moisture damage (10). of dissipating energy and csa is the crack surface area. WR is estimated
Mixture B was constructed on State Route 511 in Ashland County, as the area in the hysteresis loop of the stress–pseudostrain domain
Ohio. It included 32% limestone, 22% gravel, 26% natural sand, divided by the volume of the intact material. Arambula et al. (11)
and 20% reclaimed asphalt pavement (RAP). During specimen provide a detailed discussion on the stress–pseudostrain relationship
preparation in the laboratory, the RAP material was proportionally and its application in characterizing asphalt mixture performance.
replaced with the other aggregates in the mixture. Therefore, the After several mathematical manipulations, the equation to estimate
final mixture proportion included 40% limestone, 27.5% gravel, and the crack growth index denoted by r is expressed as follows (11):
32.5% natural sand. The asphalt binder content was 5.4% by weight
1+ n
of the mixture.
⎡ n
⎤ 2 n +1
Mixture C was constructed on State Route 226 in Wayne County, ⎛ ⎞ 1+ n
⎛ 1+1n 1
⎞
r = ⎢( 2n + 1) ⎜ 1+ n ⎥
bE R
⎜ N − 1 ⎟⎠ ⎥ (3)
Ohio. It consisted of 52% gravel, 14% limestone sand, 14% natural ⎢ ⎝ 4 πE1ΔG f σ t ⎟⎠ ⎝
2

sand, and 20% RAP. Again, the RAP material was replaced for a ⎢⎣ ⎥⎦

TABLE 1 Mixture Descriptions (10)

Field Moisture
Mixture Highway Mixture Type Location Performance Aggregate Type Binder Grade

A Texas IH 20 Superpave Atlanta, TX Good River gravel PG 76-22

B Ohio SR 511 Type 1 Ashland Fair to poor Gravel, limestone, reclaimed PG 64-22
County, OH asphalt pavement (RAP)
C Ohio SR 226 Type 1 Wayne Poor Gravel, RAP PG 64-28
intermediate County, OH
22 Transportation Research Record 2001

0.075 0.30 1.18 2.36 4.75 9.5 12.5 19.0

100

90

80

70
Mixture A
Percent Passing

60

50
Mixture B
40

30
Mixture C
20

10

0
0.075 0.30 1.18 2.36 4.75 9.5 12.5 19.0
Sieve Size, mm, Raised to 0.45 Power

FIGURE 1 Aggregate gradations.

where After the degree of saturation was measured, the specimens were
kept in water and submerged in a water bath at 60°C. After 24 h inside
n =function of the slope of the relaxation curve,
the hot water bath, the specimens were removed and placed in water
b =rate of change of the DPSE with respect to load cycles N,
at room temperature for an additional 2 h. The specimens were then
ER =reference modulus,
removed from the water bath at room temperature and left to air dry
E1 =initial relaxation modulus,
ΔGf =partial wet adhesive bond surface energy between asphalt before testing.
binder and aggregate, and To measure the load-induced displacement, three axial linear
σt = tensile strength of the material. variable differential transducers (LVDTs) placed 120° along the
circumference of the specimen were used. The distance between
As shown subsequently, variability in the crack growth index can LVDT holders was 100 mm. To obtain the parameters required
be reduced by normalizing with respect to tensile strength. Therefore, to estimate r, relaxation, uniaxial dynamic tension, and tensile
the normalized crack growth index denoted by R is estimated as strength tests were performed on the asphalt mixture specimens,
and surface energy values were obtained to estimate the partial
2 n +1
R = r ( σ 2t )
wet adhesive bond energy value. Details of these procedures are
n (4)
presented next.
n 1+ n
1+ n ⎛ bER ⎞ 2 n +1 ⎛ 1+1n 1
⎞ 2 n +1
R = ( 2n + 1) 2 n +1
⎜⎝ 4 πE ΔG ⎟⎠ ⎜ N − 11+ n
⎟⎠ (5) Viscoelastic Parameters
1 f ⎝
Relaxation tests at 10°C, 20°C, and 30°C were conducted on the
asphalt mixtures. To prevent damage to the specimen, a low-level
CHARACTERIZATION OF ASPHALT MIXTURES trapezoidal-shaped tensile strain of 200 microstrain (µ) was applied
AND MODEL PARAMETERS for 60 s. After a rest period of 600 s, a trapezoidal compressive strain
of 200 µ was also applied for a period of 60 s.
The crack growth model was used in this study to evaluate the The following power law equation was used to describe the master
moisture susceptibility of Mixtures A, B, and C. Six replicates for curve of the relaxation modulus as a function of loading time:
each mixture type were prepared and compacted in the laboratory
using the Superpave gyratory compactor (SGC). The compacted E ( t ) = E1tred
−m
(6)
specimens, with a diameter of 150 mm and a height of 165 mm, were
cored and trimmed to a final size of 100-mm diameter by 150-mm
where
height. Three of the six replicates were moisture conditioned follow-
ing the modified Lottman test procedure, which specifies 70% to E(t) = time-dependent relaxation modulus,
80% vacuum saturation by weight (AASHTO T 283-02). To satisfy E1 = initial relaxation modulus,
that requirement, Mixture A required 25 s of saturation time, and tred = reduced time equal to t/aT, and
Mixtures B and C required 900 s. m = modulus relaxation rate.
Arambula, Masad, and Martin 23

The shift factors, aT, were obtained by using the Arrhenius DPSE
WR = (9)
equation (14): ⎛ Si ⎞
⎜⎝ S ⎟⎠
o
⎛ ΔH ⎞ ⎛ 1 1 ⎞
log aT = log ( e ) ⎜ −
⎝ R ⎟⎠ ⎜⎝ T Tref ⎟⎠
(7)
where Si is the pseudostiffness and is computed for each load cycle
as the ratio of the maximum applied stress with respect to R, and So
where is the maximum pseudostiffness, which usually corresponds to the
T = test temperature, first load cycle.
Tref = reference temperature, There is a linear relationship between WR and the logarithm of the
ΔH = activation energy, and number of load cycles, ln(N); the parameter b in Equation 5 represents
R = gas constant equal to 8.314 J/(mol K). the slope of that relationship. The intercept of the line represents the
energy associated with the initial damage and material nonlinearity
The relaxation moduli at the three different temperatures were due to the difference in the strain level used in determining the visco-
shifted to a reference temperature of 30°C. The values for E1, m, and elastic properties (200 µ) and the strain level used in the uniaxial
ΔH were obtained by minimizing the sum of squared errors between dynamic tension test (350 µ).
the observed and estimated relaxation moduli.
The relationships between n in Equation 1 and m in Equation 6
were derived by Schapery for different scenarios (15). Further, strain- Tensile Strength Parameter
controlled loading with constant surface energy of the material and
length of the fracture process zone ahead of the crack tip was asso- After the uniaxial dynamic tension test, a tensile strength test at 30°C
ciated with a value of n = 1/m (16). This relationship was therefore and a rate of 0.25 mm/min was conducted on the specimens until
adopted in this study. failure. As expected, the wet specimens differed from the dry spec-
imens after failure. As shown in Figure 2, the wet specimens exhibited
stripping or loss of adhesion between the asphalt binder and the
Dissipated Pseudostrain Energy and aggregate, whereas the dry specimens had fully coated aggregates
Reference Modulus Parameters throughout the specimen section.

A uniaxial dynamic tension test at 30°C was used to estimate the

DPSE and ER parameters for the crack growth model. The test con- Partial Wet Adhesive Bond
sisted of the application of a haversine-shaped strain of 350 µ for Surface Energy Parameter
0.1 s followed by a rest period of 0.9 s. The total number of applied
load cycles was 1,000. The applied strain level during the uniaxial The thermodynamic theory establishes the relationship between the
dynamic tension test was higher than in the relaxation modulus test Gibbs free energy, the work of adhesion, and surface energy. The
to induce damage in the specimen, in which damage was detected surface energy of a single-phase material is defined as the work
by the change in the stress versus pseudostrain relationship. required to create a unit area of a new surface; it is made up of a polar
The applied stress was calculated as the ratio of the recorded acid–base component (γAB) and a nonpolar Lifshitz–van der Waals
load to the cross-sectional area of the specimen. Then, the visco- (γ LW) component. When a material cracks, the energy required to
elastic stress was computed for each load cycle using Boltzmann create the two new faces is known as the work of adhesion and equals
superposition (17 ): the surface energy of each material (represented by the indices 1 and 2)
minus the interfacial energy lost when the new surfaces are created
− m ⎛ d ⎞
σ VE = ∑ i =1 E1 ( τ − ti ) ⎜ ⎟ dt (represented by the index 12). The Gibbs free energy, or adhesive
n
(8)
⎝ dt ⎠ i bond energy (ΔGa), is equal and opposite in magnitude to the work
of adhesion (19):
where
E1 = initial relaxation modulus, ΔG12a = γ 12 − γ 1 − γ 2 = ΔG12aLW + ΔG12aAB (10)
τ = last value of time in load cycle,
ti = ith value of time in load cycle, where
m = modulus relaxation rate,
d ΔG12aLW = −2 γ 1LW γ LW (11)
= change in strain for every ith time increment,
2

dt
dt
n
= time increment, and
= number of data points recorded during each load cycle.
ΔG12aAB = −2 ( γ 1+ γ −2 + γ 1− γ +2 ) (12)

The pseudostrain, R, was calculated as the ratio of σVE and the The polar acid–base component in Equation 12 is calculated from
reference modulus, ER, which was computed as the ratio of the max- its individual acid (positive superscript) and basic (negative super-
imum applied stress recorded during the first load cycle (nonlinear script) parts. When water (represented by the index 3) is present at
response) to the maximum applied strain of 350 µ (18). The values of the interface of the system, the expression for ΔGa becomes
the applied stress were then plotted versus R, forming an oval-shaped
loop. The area inside the loop represents the DPSE in each load cycle ΔG123
a
= γ 12 − γ 13 − γ 23 = ΔG123
aLW
+ ΔG123
aAB
(13)
and was computed using the area by the coordinates method (18).
To account for the reduction in the material capable of dissipating The Wilhelmy plate (WP) and the universal sorption device (USD)
energy, the DPSE was normalized by the pseudostiffness ratio: were used to estimate the values of the surface energy components
24 Transportation Research Record 2001

(a) (b)

FIGURE 2 Failed asphalt mixture specimens after tensile strength test: (a) wet specimens showing stripping and (b) dry specimens
showing well-coated aggregates.

of the asphalt binder and the aggregates, respectively. In the USD normalized pseudostiffness under wet conditions to that under dry
the spreading pressure of several probe gases onto the surface of the conditions, as follows:
aggregate is used to estimate the surface energy components. In the
WP the contact angle between a thin plastic plate coated with asphalt ⎛ Si ⎞
wet

binder submerged (wetting) and withdrawn (dewetting) from several ⎜⎝ S ⎟⎠

probe liquids is used to estimate the individual surface energy values ΔG f = o
dry
ΔG dry
f (14)
of the asphalt binder. Details about these test protocols and calculations ⎛ Si ⎞
⎜⎝ S ⎟⎠
can be found elsewhere (10). The dry and wet ΔGa results for the dif- o

ferent types of aggregates, which were calculated using the wetting

contact angle measured with the WP, are presented in Table 2.
The values presented in Table 2 correspond to each individual CHARACTERIZATION OF ASPHALT MASTIC
type of aggregate. Because more than one aggregate type is present in SPECIMENS AND MODEL PARAMETERS
each mixture, to obtain ΔGa for the mixture, a weighted average based
on the corresponding aggregate proportions was used (Table 3). Cylindrical asphalt mastic specimens made of a mixture of the
The dry bond energy in Table 3, ΔG fdry, represents the condition aggregate portion smaller than 1.18 mm and the asphalt binder were
in which no moisture is present in the asphalt mixture; the wet bond tested using the dynamic mechanic analyzer (DMA) (10). The test-
energy, ΔG fwet, represents the condition in which water fully saturates ing of the asphalt mastic was similar to that of the asphalt mixtures,
the aggregate–binder interface. During testing, however, the material except that the DMA applies dynamic shear whereas the asphalt
is not completely dry or fully saturated, but a partial wet condition mixtures were tested using uniaxial tension tests. In addition, the
exists at the interface of the aggregate and the asphalt binder (10, 20). DMA tests were performed at room temperature. The asphalt mas-
Thus, in Equation 5 the adhesive bond surface energy between the tic of Mixture A consisted of 97% limestone screenings and 3%
asphalt binder and the aggregate, ΔGf, represents the value in a partial hydrated lime, Mixture B consisted of 72% natural sand and 28%
wet condition. The ratio of the partial wet bond energy to the dry limestone sand, and Mixture C consisted of 50% natural sand and
bond energy was assumed to be proportional to the ratio of the 50% limestone sand.
The DMA specimens were prepared by mixing the filler and the
asphalt binder before mixing with the rest of the aggregates using a
TABLE 2 Adhesive Bond Energy Under Dry and Wet Conditions mechanical mixer (10). After short-term oven aging for 2 h at 135°C,
the SGC was used to compact the asphalt mastic specimens to 152 mm
Field Adhesive Adhesive
a a in diameter and 85 mm in height. Afterward the sides of the spec-
Moisture Dry ΔG12 Wet ΔG123
Mixture Performance Aggregate (erg/cm2) (erg/cm2) imens were trimmed to a height of 50 mm, and several specimens
12 mm in diameter were cored. Some of the asphalt mastic spec-
A Good Gravel 93.36 −75.20 imens were moisture conditioned following a method established
TXI limestone 118.87 −151.14 by Kim et al. (20) in which the specimens are placed in distilled
B Fair to poor Limestone 87.49 −115.58 water under vacuum for 1 h. The saturation level of the specimens,
Gravel 94.56 −160.22
measured using Equation 15, was 125% on average, indicating
C Poor Limestone 81.27 −119.82 that water not only permeated into the voids but it also diffused
Gravel 90.92 −161.87
into the asphalt binder:
Arambula, Masad, and Martin 25

TABLE 3 Weighted Average for Adhesive Bond Energy Under Dry and Wet Conditions

Weighted Weighted
Ratio
Adhesive Adhesive
Mixture Reported
dry
Dry ΔG f Wet ΔG f
dry ΔG dry
f

Mixture Highway Description Location Performance Aggregates (erg/cm2) (erg/cm2) ΔG wet

f

A Texas IH 20 Superpave Atlanta, TX Good Gravel and TXI 101.6 −99.7 1.0186
limestone
B Ohio SR 511 TY 1 Ashland Fair to poor Gravel and 90.4 −133.8 0.6756
County, OH limestone
C Ohio SR 226 TY 1 Wayne Poor Gravel and 88.9 −153.0 0.5811
intermediate County, OH limestone

100 * ( WSSD − WDRY ) and ER are associated with the undamaged state, which is the dry state.
%S = (15) Therefore, the values of E1 and m in Equation 8, the value of So in
Va
Equation 9, and the value of ER used to compute the pseudostrain,
where WSSD is the saturated surface dry weight of the specimen after correspond to the dry state. However, based on the derivation of
vacuum saturation, WDRY is the initial dry specimen weight, and Va Schapery and Lytton et al., the values of E1 and n in Equation 5
is the volume of air voids in the specimen. correspond to the material surrounding the crack, either dry or
The dry and wet asphalt mastic specimens were then subjected to moisture conditioned (16, 17 ).
a sinusoidal shear strain in the DMA to evaluate the accumulation The value of r for the three replicates of each type of asphalt mixture
of damage. The viscoelastic parameters of the asphalt mastic spec- was calculated using Equation 3. In addition, the normalized index R
imens were determined by applying a low strain level, and the DPSE was computed by using Equation 5. Table 5 presents the average of
and reference modulus parameters were evaluated by applying a both results at N = 1,000 load cycles along with their coefficients of
high strain level until specimen failure (10). The parameters were used variation. These results show that the variation in the results is reduced
to compute the normalized crack growth index, R. The data analysis when using the crack growth index, R, normalized with respect to
was identical to that applied to the asphalt mixtures and presented the tensile strength value that exhibited the highest variability among
in the previous section. the measured material properties. The coefficient of variation was
higher under wet conditions for mixtures B and C, which were more
moisture susceptible as compared with Mixture A.
RESULTS The values of R for the asphalt mixtures and the asphalt mastics
were plotted against the number of load cycles as shown in Fig-
Table 4 presents a summary of the average input parameters used ures 3 and 4, respectively. According to Figure 3, Mixture A is the
in Equations 3 and 5 to estimate r and R for the asphalt mixtures least moisture susceptible, and Mixture C is the most susceptible
and their corresponding asphalt mastics. It is important to note the to moisture damage. Observing Figure 4, it is also apparent that
difference between the viscoelastic properties used in calculating the asphalt mastic fraction of Mixture C is the most susceptible to
σVE and the DPSE in Equations 8 and 9 and those used to estimate R moisture damage. The asphalt mastic of Mixture B failed earlier
in Equation 5. In calculating the DPSE, the viscoelastic properties than the asphalt mastic of Mixture A. The failure point for the asphalt

TABLE 4 Average Surface Energy and Mechanical Test Parameters

a b
Type E1 (MPa) m n = 1/m ER (MPa) b σt (kPa)

Mixture
A dry 1,728.2 0.1231 8.24 2,132.0 0.0122 350.7
A wet 1,506.6 0.1185 8.53 2,714.5 0.0148 358.1
B dry 926.4 0.1044 9.85 1,462.1 0.0136 94.3
B wet 590.2 0.1163 8.82 1,228.0 0.0125 60.4
C dry 1,106.5 0.1290 8.22 1,583.9 0.0160 96.2
C wet 896.9 0.0987 10.18 1,425.4 0.0525 68.0
Mastic
A dry 79.4 0.3142 3.18 75.0 145.8 —
A wet 57.6 0.2299 4.35 — 105.5 —
B dry 30.1 0.5442 1.83 95.0 150.5 —
B wet 34.1 0.3702 2.70 — 171.3 —
C dry 121.1 0.4511 2.22 61.0 74.5 —
C wet 18.0 0.4581 2.18 — 180.6 —

aWet E values are not used in the analysis and were not reported for the mastic specimens. The
R
reference parameters are those of the dry specimens.
bTensile strength tests were not performed on the mastic specimens.
26 Transportation Research Record 2001

TABLE 5 Average Crack Growth Indices and Coefficients mastic of Mixture B, about 4,000 load cycles, corresponded to the
of Variation for Asphalt Mixture Specimens number of load cycles at which the asphalt mastic of Mixture C
started to exhibit a rapid increase in the wet-to-dry ratio.
Normalized Crack Growth
To assess the effect of the adhesive bond energy between the
Crack Growth Index, r Index, R
asphalt and the aggregate in dry and wet conditions, the dry-to-wet
Coefficient of Coefficient of ratio, ΔGfdry/ΔG fwet, previously presented in Table 3, was compared
Mixture Average Variation (%) Average Variation (%) with the wet-to-dry ratio of the normalized crack growth index for the
asphalt mixtures,Rwet / Rdry, obtained using Equation 5 at N = 1,000.
A dry 3.25 19.8 0.54 16.3 The relationship between these two parameters is presented in Fig-
A wet 3.57 10.1 0.60 2.8 ure 5. The mixture with the highest ΔG fdry/ΔG fwet ratio and a ratio
B dry 9.93 25.5 0.53 25.2 of R wet/R dry closer to 1 is expected to be more resistant to moisture
B wet 19.21 40.8 0.64 23.8 damage (21). Mixture C had the lowest ΔG fdry/ΔG fwet value and the
C dry 12.59 35.3 0.59 9.3 ratio of R wet/R dry at N = 1,000 was equal to 1.64, the highest among
C wet 32.40 66.6 0.96 43.3
the three mixtures. This observation confirms that Mixture C is the
most susceptible to moisture damage. Mixture A, however, had the
Normalized Crack Growth Index, R

1.4
1.2

1.0
0.8 A wet
0.6
A dry
0.4

0.2
0.0
0 200 400 600 800 1,000
Load Cycles, N
(a)
Normalized Crack Growth Index, R

1.4
1.2

1.0 B wet
0.8
0.6 B dry
0.4

0.2
0.0
0 200 400 600 800 1,000
Load Cycles, N
(b)
Normalized Crack Growth Index, R

1.4
1.2
C wet
1.0
0.8
0.6 C dry

0.4

0.2
0.0
0 200 400 600 800 1,000
Load Cycles, N
(c)

FIGURE 3 Normalized crack growth index for asphalt mixture specimens:

(a) Mixture A, (b) Mixture B, and (c) Mixture C.
Arambula, Masad, and Martin 27

Normalized Crack Growth Index, R

1,000
900
800
700
600
500
400
300
A wet
200
100
0 A dry
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
Load Cycles, N
(a)
Normalized Crack Growth Index, R

1,000
900
800
700
600
500
400
300
200 B wet
B dry
100
0
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
Load Cycles, N
(b)
Normalized Crack Growth Index, R

1,000
900 C wet
800
700
600
500
400
300
200
C dry
100
0
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
Load Cycles, N
(c)

FIGURE 4 Normalized crack growth index for asphalt mastic fraction of each
mixture type: (a) Mixture A, (b) Mixture B, and (c) Mixture C.

highest ΔG fdry/ΔG wet

f value and the lowest R
wet/R dry value at N = 1,000 moisture damage. Consequently, these methods offer only a limited
of 1.11. Therefore, Mixture A is considered the least prone to moisture understanding of the factors influencing moisture damage.
damage among the mixtures analyzed. To overcome these shortcomings, a crack growth model based
on Paris’s law for viscoelastic materials was developed to assess the
moisture susceptibility of asphalt mixtures and asphalt mastics. The
SUMMARY AND CONCLUSIONS advantages of this method are that its derivation is based on principles
of fracture mechanics and that it accounts for several fundamental
Several qualitative and quantitative methods have been developed chemical and mechanical material properties. This method was used
in the past to evaluate moisture damage. Qualitative methods are based to analyze the resistance of asphalt mixtures and their asphalt mastic
on a subjective visual assessment of damage, whereas the majority fraction to moisture damage. For the asphalt mixtures, the normal-
of the quantitative methods rely on measuring the change of a single ized crack growth index, R, and the wet-to-dry ratio of this index,
test parameter (indirect tensile strength, dynamic modulus, etc.) R wet/R dry, agreed with the reported field performance. The param-
due to moisture conditioning. These approaches do not account for eters for estimating R for the asphalt mixtures were obtained from
the interactions between the fundamental chemical and mechanical the analysis of mechanical tests (relaxation and uniaxial dynamic
properties of the mix constituents that influence the resistance to tension) and measurement of the adhesive bond energy between the
28 Transportation Research Record 2001

1.2

Adhesive Bond Energy Ratio, ΔGfdry/ΔGfwet

Mixture A
1.0

Increasing moisture damage

0.8

Mixture B
0.6
Mixture C

0.4

0.2
Increasing moisture damage

0.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7
Normalized Crack Growth Index Ratio, R wet/R dry

FIGURE 5 Adhesive bond energy ratio versus asphalt mixture normalized crack growth index ratio.

asphalt binder and the aggregate. For the asphalt mastic fraction, 8. Masad, E., A. Castelblanco, and B. Birgisson. Moisture Damage as a
the DMA was used to test the specimens and obtain the model param- Function of Air Void Size Distribution, Pore Pressure, and Bond Energy.
Journal of Testing and Evaluation, Vol. 34, No. 1, 2006.
eters. The data analysis showed similar results in regard to good 9. Birgisson, B., R. Roque, and G. C. Page. Evaluation of Water Damage
differentiation between the wet and dry behavior and a ranking of Using Hot Mix Asphalt Fracture Mechanics. Journal of the Association
the mixtures according to the reported field performance. The authors of Asphalt Paving Technologists, Vol. 72, 2003, pp. 424 –462.
highly recommend the use of the crack growth index presented in 10. Lytton, R. L., E. A. Masad, C. Zollinger, R. Bulut, and D. Little.
Measurements of Surface Energy and Its Relationship with Moisture
this study to assess moisture susceptibility of additional asphalt
Damage. Publication FHWA/TX-05/0-4524-2. Texas Transportation
mixtures and asphalt mastics with known field performance. Institute, College Station, 2005.
11. Arambula, E., E. Masad, and A. E. Martin. The Influence of Air Void
Distribution on the Moisture Susceptibility of Asphalt Mixes. Journal
ACKNOWLEDGMENTS of Materials in Civil Engineering, in press.
12. Standard Specifications for Construction and Maintenance of Highways,
Streets, and Bridges. Texas Department of Transportation, Austin, 2004.
The authors acknowledge the financial support of FHWA and the 13. 2005 Construction and Material Specification. Ohio Department of
Texas Department of Transportation and the valuable input from Transportation, Columbus, 2005.
Robert Lytton, Dallas Little, and Amit Bhasin. 14. Medani, T. O., M. Huurman, and A. A. A. Molenaar. On the Computation
of Master Curves for Bituminous Mixes. Proc., 3rd EuroBitumen
Congress, Vienna, Austria, 2004.
15. Schapery, R. A. Nonlinear Fracture Analysis of Viscoelastic Composite
REFERENCES Materials Based on a Generalized J Integral Theory. Proc., Japan–U.S.
Conference on Composite Materials, Tokyo, 1981.
1. Miller, J. S., and W. Y. Bellinger. Distress Identification Manual for 16. Lytton, R. L., J. Uzan, E. G. Fernando, R. Roque, D. Hiltunen, and
the Long-Term Pavement Performance Program. Publication FHWA- S. Stoffels. Development and Validation of Performance Prediction
RD-03-031. Office of Infrastructure Research and Development, FHWA, Models and Specifications for Asphalt Binders and Paving Mixes.
McLean, Va., 2003. SHRP-A-357. TRB, National Research Council, Washington, D.C., 1993.
2. Kringos, N., and A. Scarpas. Raveling of Asphaltic Mixes Due to Water 17. Si, Z. Characterization of Microdamage and Healing of Asphalt Concrete
Damage: Computational Identification of Controlling Parameters. In Mixtures. PhD diss. Texas A&M University, College Station, 2001.
Transportation Research Record: Journal of the Transportation Research 18. Wolf, P. R., and C. D. Ghilani. Elementary Surveying: An Introduction to
Board, No. 1929, Transportation Research Board of the National Acade- Geomatics. Addison-Wesley Prentice Hall, Upper Saddle River, N.J., 2002.
mies, Washington, D.C., 2005, pp. 79–87. 19. Hefer, A. W. Adhesion in Bitumen-Aggregate Systems and Quantifica-
3. Emery, J., and H. Seddik. Moisture Damage of Asphalt Pavements and tion of the Effects of Water on the Adhesive Bond. PhD diss. Texas A&M
Antistripping Additives: Causes, Identification, Testing and Mitigation. University, College Station, 2004, pp. 48–54.
Transportation Association of Canada, Ottawa, 1997. 20. Kim, Y.-R., D. N. Little, and R. L. Lytton. Effects of Moisture Damage
4. McGennis, R. B., T. W. Kennedy, and R. B. Machemehl. Antistripping on Material Properties and Fatigue Resistance of Asphalt Mixtures. In
and Moisture Damage in Asphalt Mixtures. Publication 253-1. Center Transportation Research Record: Journal of the Transportation Research
for Transportation Research, Austin, Tex., 1984. Board, No. 1891, Transportation Research Board of the National Acade-
5. Solaimanian, M., J. Harvey, M. Tahmoressi, and V. Tandon. Test Methods mies, Washington, D.C., 2004, pp. 48–54.
to Predict Moisture Sensitivity of Hot Mix Asphalt Pavements. Proc., 21. Bhasin, A., E. Masad, D. Little, and R. L. Lytton. Limits on Adhesive
National Seminar on Moisture Sensitivity of Asphalt Pavements, Bond Energy for Improved Resistance of Hot Mix Asphalt to Moisture
San Diego, Calif., 2003. Damage. In Transportation Research Record: Journal of the Transporta-
6. Performance Graded Asphalt Binder, Specification and Testing. Super- tion Research Board, No. 1970, Transportation Research Board of the
pave Series No. 1, SP-1. Asphalt Institute, Lexington, Ky., 1995. National Academies, Washington, D.C., 2006, pp. 3–13.
7. Terrel, R. L., and S. Al-Swailmi. Water Sensitivity of Asphalt–Aggregate
Mixes: Test Selection. SHRP-A-403. TRB, National Research Council, The Characteristics of Bituminous–Aggregate Combinations to Meet Surface
Washington, D.C., 1994. Requirements Committee sponsored publication of this paper.