You are on page 1of 8

Construction and Building Materials 35 (2012) 460467

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Analysis of moisture damage susceptibility of warm mix asphalt (WMA) mixtures


based on Dynamic Mechanical Analyzer (DMA) testing and a fracture
mechanics model
Silvia Caro a,, Diana P. Beltrn b, Allex E. Alvarez c, Cindy Estakhri d
a

Department of Civil and Environmental Engineering, Universidad de los Andes, Carrera 1 Este # 19A-40, Edicio ML, Room ML323, Bogot D.C., Colombia
Instituto de Desarrollo Urbano-IDU, Calle 22 # 6-27, Research & Development Division, Bogot D.C., Colombia
Department of Civil Engineering, Universidad del Magdalena, Carrera 32 # 22-08, Santa Marta D.T.C.H., Magdalena, Colombia
d
Texas Transportation Institute (TT), Texas A&M University, CI/TTI Building, Room 508, 3135 TAMU, College Station, TX 77843, USA
b
c

h i g h l i g h t s
" Moisture damage resistance of three ne WMA mixtures and a control-HMA was evaluated.
" A fracture micromechanical model was used to quantify moisture damage in the mixtures.
" The retained fatigue life of the WMA ne mixtures after moisture conditioning was also evaluated.
" Two out of the three WMA ne mixtures were more affected by water than the control-HMA.

a r t i c l e

i n f o

Article history:
Received 17 January 2012
Received in revised form 19 March 2012
Accepted 25 April 2012
Available online 26 May 2012
Keywords:
Warm mix asphalt (WMA)
Dynamic Mechanical Analyzer (DMA)
Moisture damage
Fracture mechanics model

a b s t r a c t
Warm mix asphalt (WMA) materials are asphalt mixtures fabricated with recently developed technologies designed to reduce the temperatures required to produce and compact traditional hot mix asphalt
(HMA) mixtures. This study evaluates the moisture susceptibility of three WMA mixtures. The analysis
is based on a viscoelastic-fracture model, which input parameters include results from the relaxation
modulus and the Dynamic Mechanical Analyzer (DMA) test conducted on specimens representing the
ne matrix portion of the actual mixtures, and from surface free energy tests. The results suggest that
moisture caused more damage in two out of the three evaluated WMA ne mixtures in comparison to
the damage observed in the control-HMA specimens. The results also reveal that the analysis should
include not only the reduction caused by moisture of certain selected performance parameters, but the
actual values of these parameters.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
During the last decades, an increasing concern for the preservation of natural resources has led to the pursuit of alternatives to reduce the production of harmful emissions and energy consumption
in several industries. The paving industry is not the exception. Traditional hot mix asphalt (HMA) mixtures are produced at temperatures ranging from 135 to 160 C. However, at such elevated
temperatures different emissions and fumes are generated, contributing to environmental contamination. Within this context,
warm mix asphalt (WMA) technologies are the result of several efforts directed to identify more sustainable technologies that can be
used for the construction of road infrastructure. A brief description
of the characteristics of WMA technologies, as well as a summary
Corresponding author. Tel.: +57 1 3324312; fax: +57 1 3324313.
E-mail address: scaro@uniandes.edu.co (S. Caro).
0950-0618/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2012.04.035

of some previous research devoted to evaluate the mechanical performance and the moisture damage susceptibility of WMA mixtures is subsequently presented.
1.1. WMA technologies
The main objective of WMA technologies is to reduce the mixing and compaction temperatures currently required for HMA mixtures. This is achieved by using different modication technologies
that act on the asphalt binder through different mechanisms (e.g.,
reduction of the viscosity of the asphalt binder, increase in lubricity, etc.), permitting the full coating of the aggregates and appropriate workability conditions at temperatures that are typically
2030 C below those used for HMA mixtures [1]. Some proved
benets of WMA technologies include the reduction in energy
consumption during the production of the mixtures (up to 40%
compared to those used for HMA mixtures), the reduction in the

S. Caro et al. / Construction and Building Materials 35 (2012) 460467

emissions to the atmosphere in the production plant (up to 30%


reduction in CO2), and the procurement of a safer working environment [24].
There are several types of classications of WMA technologies.
The one based on the type of the additive is one of the most commonly used [57], and it includes three main groups: (1) foaming
processes, (2) emulsion-based additives, and (3) organic, wax-like
processes. The three WMA technologies selected for this study represent each one of these three groups. A detailed list of the currently WMA products available in the market can be found in
Epps Martin et al. [8].
1.2. Research to evaluate the performance of WMA mixtures
Recent eld experiences have evidenced an overall good performance of WMA mixtures [8]. However, as with any new technology, there is still a concern regarding the effect that the WMA
technologies can have on the mechanical performance of the mixtures. This evaluation is of paramount importance in determining
the convenience of promoting the use of WMA as a reliable alternative to conventional HMA mixtures.
Published works on WMA technologies can be divided into two
main groups: (1) those focused on the characteristics of WMA
modied asphalt binders, and (2) those focused on the evaluation
of WMA mixtures performance in the laboratory and in the eld.
As an example of the research conducted in the rst group, Biro
et al. [9,10] and Ran et al. [11] evaluated the rheological properties
of different asphalt binders containing foaming-based and waxbased WMA technologies (i.e., Aspha-Min and Sasobit, respectively). Results from both studies showed that WMA technologies
increased the stiffness of the asphalt binders and decreased their
compliance, thus improving its resistance to permanent deformation. In addition, Gandhi et al. [12] and Lee et al. [13] showed that
a reduction in the mixing- and compaction-temperature effectively
reduced the aging effects of certain WMA asphalt binders at the
time that enhanced their permanent deformation resistance. However, results from other works (e.g., [14]) have shown opposite effects on the permanent deformation and aging susceptibility of
certain WMA modied binders (treated with particular WMA modiers), suggesting that the actual effects of WMA modiers on asphalt binders are not yet fully characterized. It should also be
mentioned that since these technologies were developed to modify
the workability of the mixtures and not their performance they
are not expected to change the original performance grade (PG) of
the asphalt binders. Nevertheless, some studies have shown that,
depending on the characteristics of the original asphalt binder
and the type and percent of modier used, the actual PG of the asphalt binders can eventually be altered [15].
In terms of the properties and performance of WMA mixtures,
Hurley and Prowell [1,16,17], Prowell and Hurley [18] and Prowell
et al. [7] conducted several experimental works to characterize the
laboratory- and eld-performance of WMA mixtures. Among the
main results, it was found that the use of WMA technologies improved compaction at lower temperatures, enhanced the mixture
resistance to permanent deformation, decreased aging, and had little or no effect on the mixture stiffness. However, it was also observed that the magnitude of those effects for a particular
mixture was highly dependent on the type of WMA technology
used.
1.3. Moisture damage susceptibility of WMA mixtures
A main concern still exists regarding the resistance of WMA
mixtures to moisture damage. In general, the presence of water
in asphalt mixtures weakens their structural capacity as the bond
between the aggregates and the mastic is diminished (adhesive

461

deterioration), and the resistance properties of the mastic are also


adversely affected (cohesive deterioration). Details on moisture
damage mechanisms and characterization in HMA can be found
elsewhere [19,20].
There are three main concerns associated with the moisture
susceptibility of WMA mixtures: (1) the effects that WMA additives may have on the quality of the adhesion between the asphalt
binder and the aggregate, (2) the effects of incomplete aggregate
drying during the WMA fabrication process, and (3) the fact that
some WMA technologies, like foaming technologies, introduce
water to the asphalt mixture as part of the fabrication process.
Most of the published works directed to evaluate moisture
damage susceptibility in WMA mixtures have used the same testing methodologies developed for traditional HMA mixtures (e.g.,
Hamburg Wheel Tracking test, Tensile Strength Ratio (TSR), Indirect Tensile Strength (IDT) in dry- and wet-condition, and dynamic
modulus, among others [7,2124]). Some of these works have
introduced minor modications to these tests with the objective
of adapting them to certain specic characteristics of WMA mixtures. For example, Hurley and Prowell [1], Xiao et al. [25], Bennert
et al. [26], and Punith et al. [27] evaluated the moisture damage
susceptibility of WMA mixtures after explicitly including moisture
in the aggregates that were used for the fabrication of the mixtures
(i.e., partially dried aggregates). With few exceptions, researchers
have reported that moisture causes larger reductions in most
mechanical properties of WMA mixtures as compared to those of
HMA mixtures. Based on this analysis approach, researchers have
suggested that WMA mixtures are more prone to moisture damage
than conventional HMA mixtures.
2. Objective and methodology
The objective of this research is to evaluate the moisture damage
susceptibility of asphalt mixtures prepared with three types of
WMA technologies. The effect of moisture on WMA mixtures was
evaluated by means of two parameters: (1) a crack-growth index
that results from applying a viscoelastic fracture model, and (2)
the change in fatigue life (i.e., number of cycles to failure) that
can be attributed to the detrimental action of moisture. Both assessment parameters were compared to those obtained for a controlHMA mixture prepared with the same base materials.
The study was not conducted directly on WMA or HMA mixtures but on ne aggregate mixture (FAM) specimens representing
the ne matrix of the full mixtures. Specically, FAM materials are
herein dened as a homogeneous blend of asphalt binder, mineral
ller, and ne aggregates with size equal to or smaller than
1.18 mm.
In the following section the fracture model is presented. Then, a
description of the materials and test procedures used is included,
followed by the analysis of the experimental and modeling results.
The last section summarizes the main ndings and conclusions obtained from this study.
3. Viscoelastic fracture model for non-linear viscoelastic
materials
The viscoelastic fracture model used in this work was initially
formulated by Lytton et al. [28], and later adapted for FAM materials by Masad et al. [29]. The output parameter of this model is a
crack growth index, called DR, which represents the cumulative
damage within FAM specimens that are subjected to dynamic loading. Specically, DR quanties the average crack length that is generated within a material as a function of the number of loading
cycles (N). Thus, DR(N) is an increasing continuous function, indicating the fact that, with time, fracture damage propagates and
accumulates within the material.

462

S. Caro et al. / Construction and Building Materials 35 (2012) 460467

In the model, the rate of change of the crack radius as a function


of the number of applied load cycles (N) is dened as:

dr
AJ R n
dN

where r is the average crack radius in the specimen, A and n are


material constants, and JR is the J-integral that quanties the
pseudostrain energy release rate per unit crack area. The total dissipated energy by viscoelastic materials subjected to cyclic loading
has two components: (1) a portion related to the viscous effects
(not related to damage), and (2) the fraction that is actually used
to generate damage. The later corresponds to the pseudostrain energy. Therefore, the JR captures the rate at which the energy required for the cracks to growth is dissipated, and it is dened by:
dW R
dN
J R dc:s:a

dN

where WR is the dissipated pseudostrain energy per unit of volume of


the intact mixture (i.e., volume of the material that is capable of dissipating energy), and c.s.a is the crack surface area. The ratio in the
numerator represents the portion of the energy that is spent on damage in each load cycle (i.e., the energy required to propagate the
crack), while the ratio in the denominator corresponds to the change
in the surface area associated with the generation of new cracks. For a
given cycle, the WR term is computed as the area of the stress
pseudostrain hysteresis loop generated during that loading cycle.
Eqs. (1) and (2) can be used to obtain the crack growth index as
a function of the number of loading cycles (DR(N) as shown in Eq.
(3)). Details on the derivation of this equation are reported by Masad et al. [29]:
1

n 2n1

GR b
DRN 2n 1n1
N
4pG1 DGf

where G1 is a parameter characterizing the relaxation modulus of


the material (Section 5.1 presents more information of this parameter), GR is a reference modulus, usually corresponding to the modulus at the threshold between the non-linear viscoelastic region and
the region were fracture damage initiates, n is a function of the
slope of the relaxation modulus curve of the material [30], DGf is
the work of adhesion for the asphalt binderaggregate interface
which is computed from the individual surface free energy (SFE)
components of the asphalt binder and aggregates, and is related
to the quality of the adhesion between both materials , and b is
the rate of change of WR with respect to the loading cycles. The
parameter b can be determined based on the linear relationship that
exists between the natural logarithm of the number of load cycles
(Ln N) and the dissipated pseudostrain energy, WR:

W R a b LnN

where a and b are the corresponding tting parameters.


The main advantage of the model applied (i.e., Eq. (3)) is that it
does not only provide information about the nal damage stage of
the asphalt mixture (e.g., number of loading cycles to failure in a
conventional fatigue test), but it also captures the evolution of fracture damage within the mixtures microstructure by considering
the: (1) fundamental viscoelastic material properties, (2) quality
of the adhesion between the asphalt binder and the aggregate,
and (3) energy dissipation rate associated with the generation
and propagation of fracture damage.
In addition, the use of Eq. (3) to characterize fracture damage
processes in the matrix phase of an asphalt mixture (i.e., FAM) is
particularly relevant, since it has been speculated that most fracture processes within asphalt mixtures initiate within this ne portion [31]. DR(N) has also been successfully used in the past to
characterize moisture damage in FAM materials representing
conventional HMA mixtures [32,33], which makes it an attractive

option to attempt the characterization of the moisture damage


sensitivity of WMA mixtures.
Evaluation of moisture damage susceptibility through the use of
this model required the determination of the model input parameters for FAM specimens prepared with WMA additives, and specimens corresponding to the control-HMA ne mixture, under both
dry and wet conditions. A dry condition refers to a regular
room environment, while a wet condition refers to a state in
which the specimen was subjected to a moisture damage conditioning (i.e., an accelerated moisture damage phenomenon). In this
work, the moisture conditioning process proposed by Masad et al.
[32], consisting in submerging the specimens for 1 h in water under
vibration and vacuum, was used. The collected experimental information was used to compute the change in the crack growth index
under dry- and wet-condition (i.e., DR(N)dry and DR(N)wet).
Moisture susceptibility was quantied by means of two parameters: (1) the Moisture Damage Ratio, M.R.(N), and (2) the Retained
Fatigue Life, R.F. The rst parameter, M.R.(N), was dened as the ratio between the crack growth index under wet- and dry-condition,
and it is a function of the number of loading cycles (N):

M:R:N DRNwet =DRNdry

Moisture damage becomes evident when M.R.(N) takes values


over 1.0, since this implies that the crack growth under wet condition is greater than the crack growth under dry condition. On the
other hand, the retained fatigue life, R.F., was dened as the percent of the fatigue life that is preserved (based on the fatigue life
evaluated in dry condition, Nf dry) after subjecting the FAM specimens to the laboratory moisture conditioning process to determine
the fatigue life in wet condition, Nf wet:

R:F:% Nfwet =Nfdry 100%

It is expected that moisture will produce a reduction in the fatigue life of the FAM specimens. Therefore, larger values of R.F. are
associated with mixtures that are more resistant to moisture damage, since they have a larger retained fatigue life. It is important to
mention that although the fatigue life of the material is not an input parameter of the fracture model, it can be easily quantied
from the test required to determine the parameter b of the model
(i.e., the rate of change of the dissipated pseudostrain energy with
the number of loading cycles). This aspect will be further treated in
Section 5.1.
This work studied the values of M.R.(N) and R.F. for the three different WMA technologies, and compared them with those obtained
for the control-HMA ne mixture.
4. Materials
4.1. WMA technologies evaluated
The WMA technologies selected to produce WMA mixtures were: (1) AsphaMin, (2) Evotherm, and (3) Sasobit. These technologies are within the ve most
commonly used WMA modiers in the USA [24], and they represent the three
groups of WMA technologies described in Section 1.1.
Aspha-Min is a nely powdered synthetic zeolite (sodium aluminum silicate
hydrate) which has been hydro-thermally crystallized [6,17]. Zeolites are framework silicates whose structure contains large vacant spaces that allow the presence
of relatively large molecules and cation groups, such as water [6]. When AsphaMin is added into the mixture with the hot asphalt binder and aggregates, at temperatures above 100 C, the heat gradually releases the water contained by the zeolites in the way of steam, which generates a volume expansion of the asphalt binder
leading to foam asphalt. This foaming effect reduces the viscosity of the asphalt binder, enhancing aggregate coating.
On the other hand, Evotherm is a chemical package that includes additives for
improving aggregate coating and mixture workability, adhesion promoters, and cationic emulsication agents [18]. This package is delivered as an emulsion that permits a reduction of approximately 35 C in the temperatures required for the
mixture fabrication. According to the producer, this chemical package promotes a
reduction in the internal friction between the aggregates and the asphalt binder
at lower temperatures, without affecting the viscosity of the binder.

463

S. Caro et al. / Construction and Building Materials 35 (2012) 460467


Finally, Sasobit is a synthetic parafn wax generated from the coal gasication
under the FischerTropsch process (i.e., oxidation of coal to generate carbon monoxide that reacts with hydrogen producing a mixture of hydrocarbons with long
molecular chains of carbon [16]). The large size carbon chains of the wax produces
an increase in its melting point, which makes it soluble in asphalt at temperatures
above 115 C [16]. When added to the hot asphalt binder, Sasobit forms a homogeneous solution that produces a signicant reduction in the asphalt binder
viscosity.

(a)

(b)

4.2. Characteristics of the mixtures and FAM specimens


Table 1 shows the main characteristics of the mixtures evaluated. All mixtures
were fabricated using a PG 64-22 asphalt binder and limestone aggregates. The
mixture gradation satised the requirements of a Type C mixture, according to
the Texas Department of Transportation specications [34]. The different WMA
additives were added to the asphalt binder using the modication protocols and
doses recommended by the producers.
Cylindrical FAM specimens representing the ne matrix of both the WMAs and
the control-HMA mixtures indicated in Table 1, were prepared using the protocol
suggested by Masad et al. [32]. The FAM materials produced with this protocol have
the same gradation of the original mixture for all particles sizes equal to and smaller
than 1.18 mm, and it uses an asphalt binder content of approximately 8% by weight
of the total ne mixture. A cylindrical FAM compacted specimen 150 mm in diameter is obtained by means of the Superpave gyratory compactor (Fig. 1a) using the
same procedure dened for traditional HMA mixtures. Then, cylindrical specimens
of 50 mm in height by 12.5 mm in diameter are cored from the gyratory compacted
specimen, and they constitute the FAM specimens that are tested to obtain most of
the input parameters for the micromechanical model (Fig. 1b).

Fig. 1. (a) FAM compacted specimen (by the Superpave gyratory compactor), and
(b) cored FAM specimens (50 mm in height by 12.5 mm in diameter).

on the FAM cylindrical specimens under strain-controlled conditions using the set up conguration presented in Fig. 2. The
test protocol a non-standardized procedure has been extensively and successfully used in previous studies [29,33] and
includes two different phases. In the initial phase, the test is
conducted at a low strain level (6.5  103%) during a 2-min
period, and is used to determine the linear viscoelastic material
properties of the FAM specimens (i.e., complex modulus, G, and
phase angle, d). The second phase fatigue testing consist of a
similar procedure, but at a higher strain level (i.e., 0.15%). Under
this strain level, the specimen deteriorates with the progression
of load applications or number of load cycles, N, which is
reected in the fact that G decreases and d increases with time.
This stage of the test nalizes when the sample fails, which is
detected by an abrupt and simultaneous decrease in both G
and d.
Information from the second stage of this test is used to obtain:
(1) the rate of dissipated pseudostrain energy (which is related to
the parameter b in Eq. (3)) and (2) the number of cycles to failure
for the specimen, or fatigue life, Nf. Fig. 3 presents a typical result of
the second stage of the DMA testing, where the failure of the specimen is easily identied.

5. Experimental procedures
This section describes the testing procedures required to obtain
the input parameters for the computation of the crack growth index (DR(N); Eq. (3)), as well as the fatigue life values of the FAM
specimens.
5.1. Test procedures to quantify viscoelastic and failure properties of
FAM specimens
A rheometer (TA-series AR 2000) was used to characterize the
viscoelastic material properties of the cylindrical FAM specimens.
Fig. 2 illustrates the conguration of the rheometer and a FAM
specimen ready to be tested. All tests described in this section were
conducted at room temperature, and a total of four FAM specimens
were used for each test on every ne mixture evaluated.
Two different tests were conducted on the cylindrical FAM
specimens using this rheometer:

5.2. Test procedures to estimate the work of adhesion


By denition, SFE is the energy required under vacuum conditions to create a new unit of surface area of a material. The Good
Van OssChaudhury [36] theory establishes that the SFE of a material results from the addition of a non-polar component, cLW (Lifshitzvan der Waals) and a polar component, cAB (including a
monopolar acidic component, c+, and a monopolar basic component, c). Therefore, the total SFE of a material, or c, is computed
as:

 Relaxation modulus: Relaxation modulus tests were conducted


at a strain level of 6.5  103%. This strain level was established
to be appropriate for the purposes of this test in a previous
study [33]. Relaxation modulus curves can be appropriately tted to a power law equation of the type [35]:

Gt G0 G1 t m

SFE c cLW cAB cLW 2c c 0:5

The individual SFE components of the asphalt binder (subscript


A) and aggregate (subscript S) can be used to determine the energy that is required to propagate an existent crack at an asphalt
aggregate interface (i.e., to create two new area surfaces). This
quantity is known as the work of adhesion (DGf in Eq. (3)), and is
computed as follows:

where G1 included in Eq. (3) and the slope of the relaxation curve, m, are tting constants. The parameter n in Eq. (3)
can be found from m, by applying the following relationship:
n = 1 + 1/m [30].
 Dynamic Mechanical Analyzer (DMA) test for FAM materials: This
test consists of applying a dynamic shear oscillation loading

LW 0:5
DGf 2cLW
2cA cS 0:5 2cA cS 0:5
A cS

Table 1
Characteristics of the evaluated mixtures.
Mixture

Aggregate

Asphalt Binder

Dosage

Control-HMA mixture
Aspha-Min-WMA mixture
Evotherm-WMA mixture
Sasobit-WMA mixture

Limestone
Limestone
Limestone
Limestone

PG
PG
PG
PG

0.5% by weight of the mixture


0.5% by weight of the asphalt binder
3.0% by weight of the asphalt binder

64-22
64-22
64-22
64-22

464

S. Caro et al. / Construction and Building Materials 35 (2012) 460467

(a)

(b)

7. Moisture susceptibility results and analysis

Fig. 2. (a) Rheometer with the required geometry conguration, and (b) FAM
specimen set up for testing.

Wilhelmy Plate and Universal Sorption Device (USD) tests were


conducted, respectively, to determine the SFE of the asphalt binders (unmodied and modied with the three WMA technologies),
and the aggregates. It is important to mention that the Wilhelmy
Plate test did not require any especial adjustment or modication
in order to compute the SFE of the WMA-modied asphalt binders.
Details about these tests can be found elsewhere [3739], and are
not presented here for the sake of brevity.
5.3. Summary of tests procedures
Table 2 provides a summary of the tests required to determine
the input parameters for the computation of the crack growth index, DR(N) (Eq. (3)).
6. Experimental results
6.1. Rheological and fracture properties of viscoelastic materials
The mean values and the standard deviation (r) of the viscoelastic parameters obtained from the relaxation and DMA tests
are presented in Table 3 for the dry specimens, and in Table 4 for
the wet specimens. Both tables also present the average number
of cycles to fatigue failure (Nf) obtained for each ne mixture.
6.2. Surface free energy (SFE) results

Complex Modulus, G*
(108 Pa)

Table 5 presents the SFE results for the aggregate, neat asphalt
binder, and asphalt binder with WMA additives. These values were
used to compute the work of adhesion for the asphalt binder
aggregate interfaces (DGf) by means of Eq. (9).

3.0

100

2.5

80

2.0
Phase angle

60

1.5
1.0

Complex Modulus

40
20

0.5
0.0
0.0E+00 5.0E+03 1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04 3.5E+04

Number of cycles (N)

Fig. 3. Typical results of a DMA fatigue test at a constant strain level of 0.15%.

Based on the information presented in Tables 35, and using Eq.


(3), the average crack growth index, DR(N), for each ne mixture
was computed and plotted as a function of the number of load cycles. Fig. 4a and b presents, respectively, the DR(N) results for the
dry- and wet-conditions, computed using the mean values of the
input parameters obtained from four replicate specimens for each
mixture (information from Tables 35).
Under dry conditions, Fig. 4a suggests that the material exhibiting the best mechanical performance (i.e., smaller DR(N)) is the Aspha-Min-WMA ne mixture, followed by the control-HMA, the
Evotherm-WMA and the Sasobit-WMA ne mixtures. On the
other hand, under wet conditions, the ne mixture exhibiting the
least cumulative damage is the control-HMA, followed by the Evotherm-WMA, and by both the Aspha-Min-WMA and the Sasobit-WMA (which curves are superimposed). For both dry and
wet-condition, and in terms of the DR(N), the Sasobit-WMA ne
mixture has the highest cumulative fracture damage. In addition,
as it was expected, in all cases the average values of DR(N) for
the wet condition (Fig. 4b) are larger than those obtained for the
dry condition (Fig. 4a), validating the fact that moisture negatively
impacts the performance of the materials evaluated. However, in
order to analyze the moisture damage susceptibility of the ne
mixtures, the increase in accumulated damage caused by moisture,
in addition to the individual absolute values of DR(N), will be
considered.
Fig. 4 also shows that there are two critical types of material responses: (1) those where even though DR(N) is small indicating
that the internal structure of the FAM specimen is relatively
undamaged , the material suddenly reaches an early failure
(e.g., control-HMA ne mixture), and (2) those where the failure
occurs late in time, although DR(N) is large, indicating that the
material undergoes intense internal damage processes before failing (e.g., Evotherm-WMA ne mixture). Making a decision about
which one of these two conditions is more desirable in the eld is
not a simple task. However, this information is crucial, since it indicates that the analysis of overall damage processes within these
mixtures should not only consider the crack growth index values
themselves (Fig. 4), but also the fatigue life of the materials.
As summarized in Table 6, longer fatigue life values, Nf, were
observed for the WMA ne mixtures as compared to those of the
control-HMA ne mixtures, in both dry- and wet-condition. This
table also includes the standard deviation (r) of the fatigue life values. The information of Nf can also be observed in Fig. 4, since the
plots of DR(N) were constructed up to the mean value of fatigue
life.
Information from Fig. 4 and Table 6 can be used to compute and
compare the moisture susceptibility of all mixtures in terms of the
M.R.(N) and R.F. Corresponding results for M.R.(N) are shown in
Fig. 5. As explained before, large values of M.R. are associated with
a high susceptibility of the material to moisture damage (i.e.,
DR(N)wet is larger in comparison to DR(N)dry). It should be observed
that the curves in this gure do not cross among each other, suggesting that the tendency of the relative effect of moisture among
all mixtures is independent of the number of loading cycles. This
information is useful to rank the impact of moisture on the
mechanical performance of the ne mixtures. A comparison at
N = 5000 cycles, for example, suggests that the Aspha-MinWMA is the most affected by the presence of moisture, while the
control-HMA and Evotherm-WMA ne mixtures are highly resistant to the detrimental effects of moisture.
Fig. 6 shows the effect of moisture on the reduction of fatigue
life (i.e., comparison of Nf dry and Nf wet) for all the mixtures evaluated. The retained fatigue life value (R.F.) for each ne mixture corresponds to the complement of the reduction value presented as a

465

S. Caro et al. / Construction and Building Materials 35 (2012) 460467


Table 2
Tests required to determine the input parameters for the crack growth index model.
Parameter

Meaning

Test required

Observations

G1

Relaxation modulus parameter

Relaxation modulus

The parameter results from tting the relaxation modulus test to a


potential function (Eq. (7))

Related to the relaxation properties of the


material
Modulus at the threshold between the nonlinear viscoelastic region and the region were
damage initiates
Work of adhesion (energy to break the
asphaltaggregate bond creating a unit of
cracked area)
Rate of dissipated pseudostrain energy

Dynamic Mechanical Analyzer


(DMA) at low and high strain
levels
Wilhelmy plate for asphalt;
Universal Sorption Device (USD)
for aggregate
Dynamic Mechanical Analyzer
(DMA) at a high strain level

Strain-controlled oscillation test (conducted using a rheometer)

GR

DG f

The tests allow computing the surface free energy (SFE) components
of both materials, which are later used to compute the work of
adhesion DGf (Eq. (9))
Strain-controlled oscillation test at a 0.15% strain level (Eq. (4))

Table 3
Average linear viscoelastic properties and parameters (Table 2) of FAM specimens in dry condition.
Test

Aspha-Min

Parameter

Evotherm

Sasobit

Control-HMA

Mean

Mean

Mean

Mean

Relaxation test

m
n
G1 (107 Pa)

0.29
4.52
7.4

2.12  102
2.62  101
1.56

0.57
2.77
6.43

2.12  102
6.65  102
4.21

0.57
2.76
10.9

4.58  102
1.39  101
1.01

0.41
3.7
12.7

1.52  101
1.15  100
5.6

Dynamic Mechanical Analyzer (DMA)

b
GR (108 Pa)
Nf (103 cycles)

53
3.98
141.2

11
0.25
35.9

44
3.85
130.8

13
0.48
12.8

67
4.64
49.7

12
0.63
26.9

88
4.26
43.6

22
0.60
6.4

Table 4
Average linear viscoelastic properties and parameters (Table 2) of FAM specimens in wet condition.
Test

Aspha-Min

Parameter

Mean

Evotherm

r
2

Sasobit

Mean

2

Control-HMA

Mean

Mean

Relaxation test

m
n
G1 (107 Pa)

0.69
2.47
5.78

9.19  10
1.98  101
1.87

0.67
2.50
6.72

3.46  10
7.51  102
0.28

0.59
2.78
11.3

0.16
0.49
0.87

0.56
2.80
8.00

4.04  102
1.29  101
0.76

Dynamic Mechanical Analyzer (DMA)

b
GR (108 Pa)
Nf (103 cycles)

45
3.10
30.3

7
0.28
17.1

40
3.49
92.5

10
0.27
38.5

69
4.06
37.4

16
0.52
14.4

56
3.31
19.7

4
0.20
9.9

Table 5
SFE components for aggregate and asphalt binders (values in ergs/cm2).

cLW

c+

c

cAB

SFE

Limestone

Aggregate
50.33

3.45

268.95

60.92

111.25

PG
PG
PG
PG

Asphalt binder
20.05
38.24
42.72
44.35

1.99
0.00
0.15
0.12

17.94
16.72
8.95
14.81

11.95
0.00
2.32
2.67

32.00
38.24
45.04
47.02

64-22
64-22 + Aspha-Min
64-22 + Evotherm
64-22 + Sasobit

percentage in this gure (e.g., the reduction in fatigue life of the


Sasobit-WMA ne mixture was 75%, which means that its R.F. value is 25%). Two observations can be obtained from the information
presented in this gure:
 Moisture causes the larger reduction in fatigue life (79%) in the
Aspha-Min-WMA ne mixture. The ne mixture with Sasobit
also presented a signicant reduction in fatigue life (75%), followed by the control-HMA (55%), and the Evotherm-WMA ne
mixture (29%). These results are in good agreement with those
obtained from the fracture model (Fig. 4), indicating that the
WMA ne mixtures evaluated are similarly affected by moisture
in terms of both the amount of damage caused in their internal
microstructure, and the nal fracture condition of the material.

 Although most WMA ne mixtures have small R.F. values (i.e.,


high reduction in fatigue life), their fatigue life values in dryand wet-condition are considerable larger than those in the
control-HMA. This observation is important, because it suggests
that the approach of computing moisture damage susceptibility
for WMA mixtures as the reduction of a certain parameter or
material property caused by moisture can eventually overlook
the actual performance of these mixtures. For example, Fig. 6
suggests that although the fatigue life of the moisture-conditioned Aspha-Min-WMA ne mixture is only 21% of the fatigue
life of this mixture in dry condition (i.e., Nf wet = 30,253), this
value is 1.5 times larger than the fatigue life of the moistureconditioned control-HMA specimens (i.e., Nf wet = 19,749).
Therefore, even though the reduction in the mechanical

466

S. Caro et al. / Construction and Building Materials 35 (2012) 460467

(a)

(b)

12

10

Evotherm

Aspha-Min

10

Sasobit

Sasobit

Evotherm

Control

Control

Aspha-Min
4

2
0

0
0

25,000

50,000

75,000

100,000

125,000

Number of cycles (N)

10,000

20,000

30,000

40,000

50,000

Number of cycles (N)

Fig. 4. Average results of DR(N) as a function of N for (a) dry- and (b) wet-condition (the curves for Aspha-Min and Sasobit are overlapping).

Table 6
Fatigue life (Nf) results for all FAM specimens (dry- and wet-condition).
Mixture

Dry

M.R.(N) = (R(N)wet/R(N)dry)

Control-HMA
Asphamin-WMA
Evotherm-WMA
Sasobit-WMA

Table 7
Moisture damage susceptibility ranking.

Wet

Mixture

Mean

Mean

43,581
141,169
130,835
149,685

6401
35,919
12,779
26,933

19,749
30,253
92,481
37,381

9933
17,117
38,466
14,434

2.4
Aspha-Min

2.2
2.0
1.8
1.6

Sasobit
Control

1.4

Evotherm

1.2
1.0
0

2,000

4,000

6,000

8,000

10,000 12,000 14,000 16,000 18,000

Number of cycles (N)


Fig. 5. Moisture damage ratio, M.R.(N), for all mixtures evaluated.

1.6
1.4

1.50
1.41

Nf (105 cycles)

Nf dry

1.31

1.2

Nf wet

29%
1.0

75%

79%
0.92

0.8
0.6
0.4
0.2

0.37

0.30

0.44

55%
0.20

0.0
Aspha-Min

Evotherm

Sasobit

Control

Fig. 6. Effect of moisture damage on the reduction of fatigue life.

response of WMA ne mixtures is substantially larger than that


in the control-HMA, the overall expected performance of these
alternative mixtures could eventually be superior in terms of
their fatigue life. This aspect is normally not considered in most
published works.
Table 7 summarizes the M.R. values obtained for all ne mixtures at N = 5000 cycles, as well as their corresponding retained
fatigue life values, R.F. This table also presents the ranking of the
materials from the one that is more resistant to the effects of
moisture, to the one that is more affected by this agent (i.e., 14,

Control-HMA
Aspha-min-WMA
Evotherm-WMA
Sasobit-WMA

Moisture damage ratio


(M.R.)

Retained fatigue life


(R.F.)

Value

Ranking

Value (%)

Ranking

1.38
1.84
1.31
1.42

2
4
1
3

45
21
71
25

2
4
1
3

respectively). It can be observed that both parameters present


the same ranking for all materials, being the Aspha-Min-WMA
ne mixture the least resistant to moisture damage, and the Evotherm-WMA and control-HMA ne mixtures the most resistant
to this phenomenon. These results support the ndings from other
works reporting that mixtures produced with water-based WMA
technologies (i.e., Aspha-Min) are highly susceptible to the effects
of moisture [24], while other additives-based technologies that are
composed by chemical packages usually including adhesion promoters present good resistance to moisture damage (i.e.,
Evotherm).
8. Conclusions and recommendations
This research used two different parameters to quantify the effects of moisture on the mechanical response of three WMA ne
mixtures (fabricated with Sasobit, Aspha-Min, and Evotherm)
and an unmodied control-HMA ne mixture when subjected to
cyclic loading. The rst parameter was the ratio of the crack
growth index (DR(N), resulting from a viscoelastic fracture model)
computed for specimens that were and were not exposed to a
moisture-conditioning process. The second parameter was the retained fatigue life of the ne asphalt mixtures, determined by
means of DMA testing conducted on specimens tested with and
without being subjected to the detrimental effects of moisture
(i.e., retained fatigue life, or R.F.).
Results from both parameters showed that Aspha-Min-WMA
ne mixtures a water-based WMA technology is the most
prone to moisture damage. On the contrary, ne mixtures modied
with Evotherm showed satisfactory results. The Sasobit-WMA
ne mixture showed to be more susceptible to moisture damage
than the control-HMA ne mixture, but not as critical as in the case
of Aspha-Min-WMA ne mixtures. In general, the results showed
that the actual moisture damage susceptibility of WMA ne mixtures is highly dependent on the type of WMA technology used.
It was also observed that although the fatigue resistance of the
control-HMA ne mixture was not signicantly affected by moisture, the actual fatigue life values for this material in both dryand wet-condition were considerable smaller than those obtained
for all WMA ne mixtures. This can open a debate suggesting that a

S. Caro et al. / Construction and Building Materials 35 (2012) 460467

fair comparison between the moisture susceptibility of WMA mixtures and their HMA counterparts might consider not only the
reduction in certain material properties or parameters, but also
the actual values of those properties or parameters.
Finally, it is important to emphasize that this study was performed using a specic set of materials (aggregate, asphalt binder,
and WMA additives), therefore restraining the possibility of providing generalized conclusions on the overall moisture sensitivity
of the evaluated WMA technologies.
Acknowledgments
This work was conducted while the second author held a graduate research assistant position at Universidad de Los Andes (Bogot-Colombia). The authors are thankful for the laboratory
assistance provided by David Zeig at Texas A&M University to complete the laboratory testing, and for the valuable inputs provided
by Dr. Bernardo Caicedo.
References
[1] Hurley GC, Prowell BD. Evaluation of potential processes for use in warm mix
asphalt. J Assoc Asphalt Paving Technol (AAPT) 2006;75:4190.
[2] Jenkins KJ, Molenaar AAA, de Groot JLA, Van de Ven MFC. Foamed asphalt
produced using warmed aggregates. J Assoc Asphalt Paving Technol (AAPT)
2002;71:44478.
[3] Kuennen T. Warm mixes are a hot topic. Better roads. Des Plaines,
Illinois: James Informational Media, Inc.; 2004.
[4] Nazzal MD, Sargand S, Al-Rawashdeh A. Evaluation of warm mix asphalt
mixtures containing rap using accelerated loading tests. J Test Eval 2011;23(3).
[5] Kristjnsdttir , Muench ST, Michael L, Burke G. Assessing potential for
warm-mix asphalt technology adoption. Transport Res Rec 2007;2040:919.
[6] Corrigan M. Warm mix asphalt technologies and research. Federal Highway
Administration-FHWA; 2008.
[7] Prowell BD, Hurley GC, Crews E. Field performance of warm-mix asphalt at
National Center for Asphalt Technology test track. Transport Res Rec
2007;1998:96102.
[8] Martin A, Armbula E, Estakhri C, Epps J, Yin F, Walubita L, et al. NCHRP 949
Performance of wma technologies: stage I moisture susceptibility phase I,
Interim report. College Station, TX: Texas Transportation Institute (TTI), Texas
A&M University; 2011.
[9] Biro S, Gandhi T, Amirkhanian S. Midrange temperature rheological properties
of warm asphalt binders. Journal of Materials in Civil Engineering (ASCE)
2009;21(7):31623.
[10] Biro S, Gandhi T, Amirkhanian S. Determination of zero shear viscosity of warm
asphalt binders. Constr Build Mater 2009;23(5):20806.
[11] Ran J, Xu S, Li M, Ji J. Research on the performances of warm asphalt and warm
mix asphalt with Sasobit. In: 10th International Chinese conference of
transportation professionals: integrated transportation systems, Beijing; 2010.
[12] Gandhi T, Akisetty C, Amirkhanian S. A comparison of warm asphalt binder
aging with laboratory aging procedures. J Test Eval 2010;38(1):5764.
[13] Lee SJ, Amirkhanian S, Park NW, Kim KW. Characterization of warm mix
asphalt binders containing articially long-term aged binders. Constr Build
Mater 2008;23(6):23719.
[14] Arega Z, Bhasin A, Motamed A, Turner F. Inuence of warm-mix additives and
reduced aging on the rheology of asphalt binders with different natural wax
contents. Journal of Materials in Civil Engineering (ASCE) 2011;23(10):14539.
[15] Button JW, Wimsatt A, Estakhri C. A synthesis of warm-mix asphalt. FHWA/
TX-07/0-5597-1. College Station, TX: Texas Transportation Institute (TTI),
Texas A&M University; 2007.

467

[16] Hurley GC, Prowell BD. Evaluation of Sasobit for use in warm mix asphalt.
NCAT report 05-06. National Center for Asphalt Technology Auburn
University, Auburn, AL; 2005.
[17] Hurley GC, Prowell BD. Evaluation of Aspha-min zeolite for use in warm mix
asphalt. NCAT report 0504. National Center for Asphalt Technology Auburn
University, Auburn, AL; 2005.
[18] Prowell BD, Hurley GC. Warm mix asphalt: best practices. Maryland,
USA: NAPA (National Asphalt Pavement Association); 2007.
[19] Caro S, Masad E, Bhasin A, Little D. Moisture susceptibility of asphalt mixtures.
Part I: mechanisms. Int J Pavement Eng 2008;9(2):8198.
[20] Caro S, Masad E, Bhasin A, Little DN. Moisture susceptibility of asphalt
mixtures. Part II: characterization and modeling. Int J Pavement Eng
2008;9(2):99114.
[21] Hearon A, Diefenderfer S. Laboratory evaluation of warm asphalt properties
and performance. Aireld and highway pavements: efcient pavements
supporting transportations future, Bellevue, Washington; 2008. p. 182194.
[22] Mohammad L, Saadeh S, Cooper S. Evaluation of asphalt mixtures containing
Sasobit warm mix additive. Geotechnical Special Publications GSP ASCE,
GeoCongress 2008: Geosustainability and Geohazard Mitigation 2008;178:
10161023.
[23] Austerman AJ, Mogawer WS, Bonaquist R. Evaluating the effects of warm mix
asphalt technology additive dosages on the workability and durability of
asphalt. In: 88th Annual meeting of the transportation research board (TRB),
Transportation Research Board, Washington, DC; 2009.
[24] Mogawer WS, Austerman AJ, Kassem E, Masad E. Moisture damage
characteristics of warm mix asphalt mixtures. J Assoc Asphalt Paving
Technol (AAPT) 2011;80:491524.
[25] Xiao F, Jordan J, Amirkhanian S. Laboratory investigation of moisture damage
in warm-mix asphalt containing moist aggregates. Transport Res Rec
2009;2126:11524.
[26] Bennert T, Maher A, Sauber R. Inuence of production temperature and
aggregate moisture content on the performance of warm mix asphalt. In: 90th
Annual meeting of the transportation research board (TRB), Washington, DC;
2011.
[27] Punith VS, Xiao F, Amirkhanian SN. Effects of moist aggregates on the
performance of warm mix asphalt mixtures containing non-foaming additives.
J Test Eval 2011;39(5).
[28] Lytton R, Uzan J, Fernando EG, Roque R, Hiltmen D, Stoffels S. Development
and validation of performance prediction models and specications for asphalt
binders and paving mixtures. Report no. SHRP-A-357, Washington, DC; 1993.
[29] Masad E, CasteloBranco VTF, Little DN, Lytton R. A unied method for the
analysis of controlled-strain and controlled-stress fatigue testing. Int J
Pavement Eng 2008;9(4):23346.
[30] Schapery RA. On viscoelastic deformation and failure behavior of composite
materials with distributed aws. New York, NY: American Society of
Mechanical Engineers; 1981.
[31] Montepara A, Romeo E, Isola M, Tebaldi G. The role of llers on cracking
behavior of mastics and asphalt mixtures. J Assoc Asphalt Paving Technol
(AAPT) 2011;80:16190.
[32] Masad E, Zollinger C, Bulut R, Little D, Lytton R. Characterization of HMA
moisture damage using surface energy and fracture properties. J Assoc Asphalt
Paving Technol (AAPT) 2006;75:71354.
[33] Caro S, Masad E, Gordon A, Bhasin A, Little DN. Probabilistic analysis of fracture
in asphalt mixtures caused by moisture damage. Transport Res Rec 2008;
2057:2836.
[34] TxDOT. Standard specications for construction and maintenance of highways,
streets, and bridges. Austin, TX; 2004.
[35] Kim RY. Modeling of asphalt concrete. 1st ed. New York: McGraw-Hill; 2009.
[36] Van Oss CJ, Chaudhury MK, Good RJ. Interfacial LifshitzVan der Waals and
polar interactions in macroscopic systems. Chem Rev 1988;88(6):92741.
[37] Hefer AW, Bhasin A, Little DN. Bitumen surface energy characterization using a
contact angle approach. J Mater Civil Eng (ASCE) 2006;18(6):75967.
[38] Bhasin A, Little DN. Characterization of aggregate surface energy using the
universal sorption device. J Mater Civil Eng (ASCE) 2007;19(8):63441.
[39] Caro S, Alvarez AE. Evaluacin de la susceptibilidad al dao por humedad de
mezclas asflticas empleando propiedades termodinmicas. Revista Facultad
de Ingeniera, Universidad de Antioquia 2011;58:95104.

You might also like