Construction and Building Materials 123 (2016) 444–453

Contents lists available at ScienceDirect

Construction and Building Materials

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

Effects of morphological characteristics of aggregate particles on the

mechanical behavior of bituminous paving mixtures
Francisco Thiago Sacramento Aragão a,⇑, Abraham Ricardo Guerrero Pazos a, Laura Maria Goretti da Motta a,
Yong-Rak Kim b, Luis Alberto Herrmann do Nascimento c
a
Department of Civil Engineering, Federal University of Rio de Janeiro/COPPE, Rio de Janeiro 21941-596, Brazil
b
Department of Civil Engineering, 362N Whittier Research Center, University of Nebraska, Lincoln, NE 68583, United States
c
Cenpes/Petrobras, Asphalt Research Laboratory, Avenida Horácio de Macedo, 950, Rio de Janeiro 21941-915, Brazil

h i g h l i g h t s

HMA rutting resistance is strongly correlated with coarse aggregate morphology.

Direct and objective measurements of aggregate shape should be pursued.

Shape properties characterized at AIMS 2 strongly correlate with rutting resistance.

Surface texture should be considered in aggregate and HMA specifications.

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

Article history: This study evaluates the influence of morphological properties of aggregates on the mechanical behavior
Received 26 February 2016 of bituminous mixtures. For that, two aggregates with different characteristics, i.e., round river gravel and
Received in revised form 27 May 2016 crushed gneiss, were used in the research. Laboratory tests were conducted to characterize morphological
Accepted 6 July 2016
properties of the aggregates following traditional methodologies and using a modern image analysis sys-
Available online 16 July 2016
tem, AIMS 2. Six asphalt mixtures containing different proportions of gravel and crushed gneiss were
designed according to the Superpave methodology and evaluated in mechanical performance tests. The
Keywords:
results obtained indicated that aggregate morphological characteristics, especially those of coarse parti-
Asphalt mixtures
Aggregates
cles, are strongly correlated with the resistance to rutting of the asphalt mixtures. In addition, AIMS 2 was
Morphological properties shown to provide more direct and scientific measurements of aggregate morphological characteristics
Mechanical behavior that present higher correlations with the mixture performance than traditional methodologies. Finally,
the results also demonstrated that the aggregate surface texture is highly correlated to the performance
of the mixtures and should be carefully considered in aggregate and asphalt mixture specifications.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction The four consensus properties identified by SHRP researchers as

In order to improve the performance, durability, safety, and
efficiency of asphalt mixtures, the Strategic Highway Research
Program (SHRP), established in 1987, proposed the Superpave
system in 1993. Superpave included a set of significant advance-
ments in the methodologies used to classify asphalt binders and
to design asphalt mixtures. In addition, the Superpave system also
defined the so-called consensus and source properties of aggregate
particles.
⇑ Corresponding author.
E-mail addresses: fthiago@coc.ufrj.br (F.T.S. Aragão), aguerrero@coc.ufrj.br
(A.R.G. Pazos), laura@coc.ufrj.br (L.M.G. Motta), ykim3@unl.edu (Y.-R. Kim).
key characteristics directly related to the performance behavior of
the mixtures were coarse- and fine-aggregate angularity, amount
of flat and elongated particles in coarse aggregates, and sand
equivalency of fine aggregates. The criteria specified for these
properties are generally function of the traffic volume expected
in the area where the flexible pavement will be built. On the other
hand, the characterization of source properties is considered
optional and the corresponding limiting criteria are specified based
on the experience of the local transportation agencies to account
for the regional geological differences.
Several research efforts have demonstrated reasonable to good
correlations between the consensus properties of aggregates spec-
ified by the SHRP and the overall mechanical behavior of asphalt
mixtures [1–7]. However, in the last years, researchers have

http://dx.doi.org/10.1016/j.conbuildmat.2016.07.013
0950-0618/Ó 2016 Elsevier Ltd. All rights reserved.
 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453
criticized some of these tests [6,8–13], which are often considered
to be subjective and/or based on indirect measurements and to
produce results that are a function of the skills of the operator.
Some testing procedures, such as the one specified for the determi-
nation of the coarse aggregate angularity, can also be tedious and
time-consuming.
Considering the relevance of aggregates and the corresponding
effects of their properties on the overall behavior of asphalt mix-
tures and on the costs related to the production and maintenance
of flexible pavements, the study of alternative methodologies that
are able to characterize aggregates properties in more objective
and representative manners becomes a very important task. In that
sense, several researchers have adopted protocols to identify
aggregate morphological properties based on image analysis
techniques, including VDG-40 Videograder, Computer Particle
Analyzer, Micromeritics OptiSizer PSDA, VideoImaging System
(VIS), Buffalo Wire Works PSSDA, Camsizer, Wipshape, University
of Illinois Aggregate Image Analyzer (UIAIA), Aggregate Imaging
System (AIMS). Masad et al. [8] compared several of these image
analysis systems and identified AIMS as the most promising equip-
ment as it allows a comprehensive, objective, and direct analysis of
shape, angularity, and texture based on two-dimensional and
three-dimensional images of aggregate particles.
This study evaluates the effects of morphological properties of
aggregate particles on the overall mechanical behavior of asphalt
mixtures. For that, properties of two aggregates with distinct mor-
phological characteristics, i.e., round river gravel and crushed
gneiss, were first characterized based on the specifications of the
Superpave system. Alternatively, aggregate morphological charac-
teristics were also determined using the second generation of the
modern image analysis system AIMS (i.e., Aggregate Imaging Mea-
surement System 2: AIMS 2). To further identify the effects of the
different morphological properties of aggregates on the overall
performance behavior of asphalt mixtures, six mixtures compris-
ing different amounts of gravel and crushed gneiss were designed
based on Superpave mix design specifications, and their perfor-
mance behavior was compared to the morphological characteris-
tics of their aggregate skeleton.
Specific objectives of this paper are:

To characterize aggregate particles with distinct morphological
properties based on Superpave specifications and using the
image analysis system AIMS 2.

To characterize the mechanical behavior of asphalt mixtures
containing different amounts of gravel and crushed gneiss. For
that, three primary mixture characteristics are evaluated: stiff-
ness, permanent deformation, and cracking.

To identify correlations between the morphological properties
of the aggregates evaluated and the mechanical behavior of
the corresponding asphalt mixtures.
2. Experimental program
2.1. Materials and mix design
One PG 70-16 S binder obtained from a refinery in Rio de
Janeiro, Brazil, and two types of aggregates, i.e., gravel from the
state of Pará, Brazil, and crushed gneiss from Rio de Janeiro were
used in this study.
Six mixtures with the same aggregate gradation were designed
based on Superpave specifications. The first mixture contained
crushed gneiss with cubic particles. The second mixture was com-
posed of gravel and natural sand. The use of natural sand as the
fine part of gradations containing gravel as coarse aggregates has
been a common practice in Brazil and this has been adopted in this
445
research. Three other mixtures were designed and contained the
following proportions (by weight) of crushed gneiss and gravel,
respectively: (a) 75% and 25%; (b) 50% and 50%; and (c) 25% and
75%. Finally, a sixth mixture composed of crushed gneiss with flat
and elongated (F&E) characteristics was also designed. This aggre-
gate was obtained in the same quarry as the cubic crushed gneiss,
but processed using a different crushing strategy. Thus, the effect
of aggregate morphological characteristics could be better evalu-
ated, given that both cubic and elongated crushed gneisses pre-
sented similar geological and physical properties. Fig. 1 shows
samples of the three aggregates used in this study.
Tables 1 and 2 present aggregate gradation and volumetric
characteristics, respectively, of the six mixtures evaluated in this
study, as well as the limiting criteria for a nominal maximum
aggregate size of 19 mm and the traffic level between 3 and 10 mil-
lion ESALs, according with AASHTO M 323 [14]. All mixtures
required the same binder content, i.e., Pb = 4.3% to generate air
voids (Va) between 3.6% and 3.8%.
Mixtures containing higher amounts of gravel presented lower
values of voids in mineral aggregates (VMA), which is in accor-
dance with the observation by several researchers [15–18].
Rounded and less angular aggregates generally present smooth
surfaces, which in combination to the absence of angular edges
facilitates the compaction of the particles in denser structures
and reduces the VMA values [15]. The voids filled with asphalt
(VFA) were also lower for those mixtures, although within the
limits specified by AASHTO M 323 [14].
The percent theoretical maximum specific gravity @ Ninitial
(%Gmm @ Nini) for the mixtures containing gravel was higher than
for the mixtures composed of crushed gneiss only, which may indi-
cate a higher sensitivity to the compaction effort during construc-
tion and a more unstable behavior under traffic loads. The percent
theoretical maximum specific gravity @ Nmaximum (%Gmm @ Nmax)
decreased as more gravel was added to the mixture, although all
mixtures met the specified criterion for this parameter. Finally,
all mixtures presented similar dust to binder ratio (D/B) within
the allowed limits.
2.2. Characterization of aggregate properties
Morphological characteristics of aggregate particles based on
Superpave specifications for consensus properties, i.e., flat and
elongated particles (F&E), coarse aggregate angularity (CAA) and
fine aggregate angularity (FAA), were first determined. In addition,
the test of uncompacted voids described in AASHTO T326 [19] was
also performed because several authors have indicated that this
test presents better correlation with the mechanical behavior of
asphalt mixtures than the typical Superpave CAA test [4,20], which
is based on the amount of fractured faces in the aggregate particles.
Additional characterizations of the aggregates included the deter-
mination of specific gravity, absorption capacity, sand equivalency,
and Los Angeles abrasion.
Finally, aggregates were tested in the AIMS 2 system before and
after Los Angeles tests to identify their ability to retain the mor-
phological characteristics after abrasion. AIMS 2 is a robust image
analysis system that has been used by several researchers to char-
acterize morphological properties of aggregate particles, such as
angularity, sphericity, flatness, elongation, and texture. The system
combines a high-resolution camera, lighting, and a microscope to
perform the analysis. The AIMS 2 turntable uses removable trays,
selected by the aggregate size being analyzed, which accommodate
the aggregate particles during the tests. An algorithm implemented
on the operating software of the system allows the automatic elim-
ination of touching particles and their corresponding characteris-
tics are not reported in the results. This is a very important
feature of the system because touching particles can be
446 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453

Fig. 1. Aggregates used in this study: (a) cubic crushed gneiss; (b) F&E crushed gneiss; and (c) round river gravel.

Table 1 aggregates obtained from analyses in the AIMS 2 system. Fig. 2

Mixture gradations in this study. shows examples of particles with distinct morphological charac-
Mesh opening % Cumulative passing Limits teristics and the values adopted by AIMS 2 to quantify these
(mm) (by weight)
Minimum Maximum
characteristics.

25.400 100.00 100 100

19.100 99.86 80 100
2.3. Laboratory tests of asphalt mixtures
12.500 78.95 – –
9.500 66.46 60 80 Mixture stiffness was measured in the form of the linear vis-
6.250 58.54 – – coelastic dynamic modulus, obtained from uniaxial tests of cylin-
4.750 54.53 48 65
drical specimens measuring 100 mm in diameter and 150 mm in
2.360 39.55 35 50
1.180 29.71 – – height. Each specimen was cored and cut from Superpave gyratory
0.600 22.02 19 30 compactor (SGC) samples measuring 150 mm in diameter and
0.300 15.02 13 23 170 mm in height. During the tests, the loading levels were care-
0.150 8.29 7 15 fully adjusted until the strain levels were within the range of
0.075 4.01 0 8
0.00005–0.00010. Three linear variable differential transformers
(LVDTs) were mounted onto the surface of the specimens at 120°
radial intervals with a 100-mm gauge length. Averaged vertical
erroneously interpreted as a single particle and compromise the deformations were used to calculate the dynamic modulus, simply
accuracy of the morphological analysis. defined as the ratio of the sinusoidal stress amplitude to the sinu-
AIMS 2 allows two-dimensional characterizations of angularity soidal strain amplitude. As suggested in AASHTO T 342 [21], five
and shape of fine aggregates comprising materials passing the temperatures (10.0; 4.4; 21.1; 37.8; 54.0 °C) and six loading fre-
4.75 mm (No. 4) sieve and retained on the 0.075 mm (No. 200) quencies (0.1; 0.5; 1.0; 5.0; 10.0; 25.0 Hz) were used, and the
sieve. For larger particles with dimensions between 37.5 mm and frequency-temperature superposition concept was applied to
4.75 mm, the system characterizes coarse aggregate angularity, develop master curves representing the dynamic modulus of the
sphericity, flatness, elongation, and texture. The results are pre- mixtures. Three replicates were tested for each mixture.
sented as cumulative distributions or represented by a characteris- The resistance to damage of the six mixtures was evaluated
tic value for each property. Masad et al. [8] have defined categories based on the results of two experiments, i.e., the flow number
for each property based on the morphological characteristics of (FN) and the indirect tensile (IDT) fracture energy tests. The FN test

Table 2
Volumetric characteristics of the mixtures evaluated in this study.

Property Pb (%) Va (%) VMA (%) VFA (%) D/B %Gmm @ Nini %Gmm @ Nmax

Criterion – 4.0 ± 1.0 >13.0 65.0–75.0 0.60–1.20 <89.0 <98.0

F&E gneiss 4.3 3.6 13.2 73.0 1.05 87.5 97.5
Cubic gneiss 4.3 3.8 13.3 71.6 1.02 88.7 97.2
75% gneiss, 25% gravel 4.3 3.6 13.0 71.0 1.03 89.1 97.4
50% gneiss, 50% gravel 4.3 3.8 12.6 69.5 1.04 89.0 97.1
25% gneiss, 75% gravel 4.3 3.7 12.2 69.7 1.05 89.8 97.1
Gravel 4.3 3.8 11.7 67.0 1.06 89.7 97.0
 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453
(a) Cubic Crushed Gneiss
AIMS 2 Angularity = 2.969
(c) Cubic Crushed Gneiss
AIMS 2 Texture = 347
Fig. 2. Examples of digital images from AIMS 2: particles with different (a) and (b) angularity; (c) and (d) texture.
was performed to assess the resistance of the mixtures to perma-
nent deformation, while the IDT test was used to evaluate the
resistance of the mixtures to cracking.
For the FN tests, specimens of 100 mm in diameter and 150 mm
in height were fabricated with a SGC. During the tests, the cylindri-
cal specimens were conditioned to 60 °C and then subjected to a
compressive cyclic uniaxial load of 204 kPa. Each cycle was com-
posed of a pulse of 0.1 s followed by a rest period of 0.9 s. The ver-
tical displacement was monitored by LVDTs mounted on the
surface of the samples and used to calculate the plastic deforma-
tion accumulated in each cycle, which was defined as the ratio
between the unrecoverable displacement and the gauge length of
the LVDTs (100 mm). Three replicates were tested for each
mixture.
The FN is associated to the cycle where the rate of plastic defor-
mation reaches a minimum value, after which the material experi-
ences increasing deformation rate with the evolution of the loading
cycles within the so-called tertiary zone of deformation. Another
parameter obtained from FN tests, i.e., the plastic deformation rate
of the secondary (linear) zone, has also been shown to correlate
well with the resistance of the mixtures to permanent deformation
[22]. The so-called b-parameter has been determined in this paper
using Eq. (1). In the equation, the 10th cycle was defined as the ini-
tial point of the secondary zone based on the observation that this
cycle is typically within that zone for asphalt mixtures. The FN
cycle was chosen as a limit between the secondary and tertiary
zones.
epFN  ep10
b¼
FN  10
where

b = deformation rate in the secondary zone (le/cycle);

epFN = plastic deformation in the FN cycle (le);
447
(b) Gravel
AIMS 2 Angularity = 2.071
(d) Gravel
AIMS 2 Texture = 120

ep10 = plastic deformation in the 10th cycle (le); and

FN = flow number (cycle).
The traditional laboratory procedures adopted for the charac-
terization of the asphalt mixture fatigue cracking are generally
quite time-consuming sometimes requiring several weeks or even
months. Thus, alternative and more efficient methodologies have
been pursued by researchers in the last years. One of these alterna-
tive procedures is the IDT fracture energy test with the application
of monotonic loads. According to Kim et al. [23], the fracture
energy obtained from the area under stress-strains curves is well
correlated with the resistance of asphalt mixtures to fatigue crack-
ing, both for laboratory-fabricated specimens and for field cores. In
addition to the good correlation with the mixture performance, the
IDT fracture energy test is easy to perform and considerably less
time-consuming than traditional fatigue tests that typically require
several weeks or months of testing for the construction of curves
relating the cycles to failure with the different levels of strain of
stress applied to the samples.
In this study, the IDT fracture energy test was performed with
specimens of 100 mm in diameter and 38 mm in height extracted
from SGC samples that measured 150 mm in diameter and
115 mm in height. Gauge points were glued over a 25.4 mm gauge
length in the center of each specimen on both faces to measure
the horizontal and vertical displacements during the tests. A
constant crosshead displacement rate of 50 mm/min. was applied
to the specimens at 20 °C. Three replicates were tested for each
mixture.
ð1Þ Using the horizontal displacements measured, the strain at the
center of the specimen was calculated using Eq. (2). In this study,
the values adopted for c1 , c2 , c3 , and c4 we, respectively, 12.400,
37.700, 0.291, and 0.908 [23,24]. The stress at the center of the
specimen could also be calculated based on Eq. (3) [25]. Then,
stress-strain curves could be plotted and fracture energy was
448 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453

defined as the area under the stress-strain curves until the peak
stress [23].
c1 þ tc2
ex¼0 ðtÞ ¼ UðtÞ
c3 þ tc4
where

ex¼0 ðtÞ = strain at the center of the specimen;

U(t) = horizontal displacement (m);

c1 , c2 , c3 , and c4 = coefficients related to the specimen diam-
eter and gauge length used; and

t. Poisson’s ratio.
2PðtÞ
rx¼0 ðtÞ ¼
ptd
where

rx¼0 ðtÞ. stress at the center of the specimen (Pa);

P(t) = applied force (N);

t = specimen thickness (m); and

d = specimen diameter (m).
3. Results and discussion
3.1. Aggregate properties
Table 3 presents results of the different combinations of aggre-
gates evaluated in this study based on conventional testing proce-
dures. It can be observed that both crushed gneisses presented
higher specific gravities, but were less absorptive than the gravel.
The crushed gneisses also presented higher values for both FAA
and CAA than the gravel, as expected. All combinations of fine
aggregates presented FAA values above 45%, which is the mini-
mum FAA specified by Superpave for the traffic level (i.e., 10 mil-
lion ESAĹs) that mixtures containing the PG 70-16 S binder need
to support. On the other hand, CAA values for combinations con-
taining gravel were lower than the minimum criteria specified by
Superpave for one or more and two or more fractured faces, i.e.,
85% and 80%, respectively. This may indicate that mixtures com-
prising gravel tend to be more susceptible to permanent
deformation.
The results of the alternative CAA test specified by AASHTO
T326 [19] also indicated higher angularity values for the crushed
gneisses, as expected. However, the difference between extreme
ð2Þ values, i.e., for cubic crushed gneiss and gravel, was significantly
smaller for the alternative test (11%) than for the Superpave CAA
test (71%), which shows that the measuring scale for the Superpave
CAA test is much more sensitive to variations in the particle
angularity.
Additionally, based on the recommendation by White et al. [4]
of CAA values above 45%, only the mixture composed solely by
gravel presented CAA below this limit, which again may indicate
that this mixture is potentially more susceptible to permanent
deformation than the others evaluated. However, based on the
results of this test, combinations of crushed gneiss and gravel
ð3Þ may present satisfactory performance.
As for F&E characteristics, Table 3 indicates that the F&E
crushed gneiss was about twice as flat and elongated as the other
aggregates, considering the 5:1 ratio between the larger and the
smaller dimensions of the particles, as specified by Superpave.
Some researchers have indicated that 5:1 is not the most appropri-
ate ratio to determine F&E characteristics of aggregate particles.
Instead, other ratios, such as 2:1 and 3:1, are often recommended
[26]. As shown in Table 2, while only 1.6% of the particles of the
F&E crushed gneiss were considered F&E based on the Superpave
5:1 ratio, 51.7% and 20.4% of the particles were characterized as
F&E based on the ratios 2:1 and 3:1, respectively.
Sand equivalent results met the Superpave requirements (>45%)
for all combinations and larger values were obtained for mixtures
containing gravel. Finally, Table 3 also indicates that the resistance
to abrasion was inversely proportional to the amount of gravel in
the combinations.
Following the characterization of aggregate properties by con-
ventional Superpave procedures, aggregate morphological proper-
ties of the six mixtures evaluated in this study were also
determined using the AIMS 2 system. As aforementioned, AIMS 2
presents the results as cumulative distributions. However, the sys-
tem may also provide a characteristic value for each property
based on the weighted average of the properties for the different
sizes analyzed. Table 4 summarizes the characteristic values
obtained for several aggregate properties.
The 2D fine aggregate form quantifies the relative shape of fine
aggregates based on Eq. (4). A perfect circle has 2D fine aggregate

Table 3
Aggregate characteristics based on conventional Superpave methods.

Property (Standard) F&E Crushed Cubic Crushed 75% Gneiss, 25% 50% Gneiss, 50% 25% Gneiss, 75% Gravel Criterion⁄⁄
Gneiss Gneiss Gravel Gravel Gravel
⁄
Gsb [27] 2.696 2.695 2.664 2.634 2.605 2.576 –
⁄
Gsa [27] 2.759 2.758 2.730 2.702 2.674 2.648 –
Absorption [27], % 0.81 0.82 0.85 0.88 0.92 0.96 –
Flat and elongated particles [28], %
2:1 ratio 51.7 33.4 31.3 29.2 27.1 25.0 –
3:1 ratio 20.4 4.6 4.1 3.7 3.2 2.7 –
5:1 ratio 1.6 0.8 0.8 0.8 0.9 0.9 <10
Fractured faces [29], %
1 or more 100 100 82 65 47 29 >85
2 or more 100 100 79 57 36 14 >80
Coarse aggregate angularity – uncompacted 50.4 49.9 48.7 47.5 46.0 44.5 –
voids [19], %
Fine aggregate angularity –uncompacted 49.2 49.2 48.2 47.6 46.9 46.5 >45
voids [30], %
Sand equivalent [31], % 71 71 73 76 78 80 >45
Los Angeles abrasion [32], % 20.7 19.7 32.2 41.7 52.1 61.1 –
⁄
Gsa and Gsb are the apparent and bulk specific gravities of aggregates, respectively.
⁄⁄
Limiting criteria for Superpave consensus properties for a traffic volume between 3 and 10 Million ESALs and depth from surface smaller than 100 mm [14].
 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453 449

Table 4 mixtures with higher resistance to sliding, which improves the

Aggregate characteristics based on AIMS 2 testing procedures. safety level of the roads for users [33–36].
Property Scale F&E Crushed Cubic Crushed Gravel Despite the recognized importance of surface texture, to the
Gneiss Gneiss authors’ best knowledge, there is no well-established methodology
2D form (fine) 0–20 8.25 7.79 to quantify this property, except from visual or tactile indirect
Fine aggregate 0–10.000 3.303 2.789 assessments. In this context, AIMS 2 appears as a promising alter-
angularity native for the quantification of the surface texture of aggregate
Coarse aggregate 0–10.000 2931 2969 2071
angularity
particles [34–36]. In AIMS 2, coarse aggregate surface texture is
F&E 2:1 (coarse) – 24.5 14.9 11.4 characterized by values ranging between 0 and 1000, where 0 rep-
F&E 3:1 (coarse) – 8.4 2.6 1.1 resents a perfectly polished surface. AIMS 2 adopts the wavelet
F&E 5:1 (coarse) – 3.3 0.3 0.4 method to quantify coarse aggregate texture. The wavelet analysis
Texture (coarse) 0–1.000 338 347 120
gives the texture details in the horizontal, vertical, and diagonal
Sphericity (coarse) 0–1 0.72 0.74 0.76
Angularity – texture 0–14.000 4848 4956 2191 directions in three separate images. The texture index at a given
(coarse) decomposition level is the arithmetic mean of the squared values
of the wavelet coefficients for all three directions, as expressed in
Eq. (6). As shown in Table 4, the crushed gneisses were about 3
times rougher than the gravel. Both crushed gneisses presented
form equals to 0, in a scale that ranges between 0 and 20. From
similar surface texture values. This was expected, given that both
Table 4, it can be observed that the crushed gneisses presented
were obtained from the same quarry.
2D form larger than the gravel, which indicates that the latter is
a rounder material. It is important to note that the same fine mate-
1 X3 X N
rial was used for both types of crushed gneisses. Texture ¼ ðDij ðx; yÞÞ2 ð6Þ
3N i¼1 j¼1

XDh
360
RhDh  Rh where
Form 2D ¼ ð4Þ
0
Rh

D = decomposition function;
where
n = decomposition level;

N = total number of coefficients in an image;

Form 2D = 2D form of the fine aggregate;
i = 1, 2, or 3 for detailed images;

Rh = radius of the particle at an angle h; and
j = wavelet index; and

Dh = incremental difference in the angle.
x,y = location of the coefficients in transformed domain.

The crushed gneisses were more angular than the gravel, as
expected, both for FAA and CAA. In AIMS 2, angularity (for both,
coarse and fine) is quantified based on the change in the gradient
on a particle boundary and is related to the sharpness of its cor-
ners, according to Eq. (5):
1 X n¼3
Angularity ¼ n jhi  hiþ3 j
3
 1 i¼1
where

h = angle of orientation of the edge points;

n = total number of points; and

i = denotes the ith point on the edge of the particle.
The coarse aggregate F&E characteristics were determined for 3
ratios between the dimensions of the particles, i.e., 2:1, 3:1, and
5:1. The F&E crushed gneiss was obviously the aggregate with
the most F&E particles for all 3 ratios and was followed by the
cubic crushed gneiss. As for the Superpave F&E test, the 3:1 and
2:1 ratios distinguished more clearly the F&E characteristics of
the F&E crushed gneiss than the 5:1 ratio. While only 3.3% of the
particles were considered F&E when the 5:1 ratio was adopted,
8.4% and 24.5% of the particles were considered F&E for the 3:1
and 2:1 ratios, respectively.
In addition to shape, angularity, and F&E characteristics, surface
texture has been identified as a key morphological property of
aggregate particles. It affects important mixture characteristics,
such as workability, adhesive properties, internal friction, among
others. The use of aggregates with rougher surfaces generally
results in mixtures with lower workability and that require more
binder, but with higher resistance to shear failure. Several research
efforts have also demonstrated that the use of aggregates that
retain their surface texture characteristics after abrasion generates
As indicated in Eq. (7), coarse aggregate sphericity describes the
3D shape of the particles in a scale that varies between 0 and 1,
where 1 represents a particle with equal dimensions in the 3 direc-
tions of a Cartesian coordinate system. According to Table 4, the
gravel was the aggregate with the largest sphericity, followed by
the cubic and the F&E crushed gneisses.
sffiffiffiffiffiffiffiffiffi
ð5Þ ds dI
Sphericity ¼ 3 ð7Þ
2
dL
where

ds = particle shortest dimension;

dI = particle intermediate dimension; and

dL = particle longest dimension.
Finally, the coarse aggregate angularity-texture (CAAT) index
represents a combination of texture and angularity characteristics
of the aggregates and is calculated using Eq. (8). In this study, the
crushed gneisses presented CAAT values much larger than the
gravel, as expected.
CAAT ¼ 10  ðTextureÞ þ 0:5  ðAngularityÞ ð8Þ
where CAAT = coarse aggregate angularity-texture.
Finally, morphological properties of the coarse aggregates eval-
uated in this study were also characterized after the particles were
subjected to the Los Angeles abrasion. For instance, CAA of the
crushed gneiss particles was reduced by 17.6% after the abrasion
test. On the other hand, gravel particles were more angular after
the Los Angeles abrasion process. A visual inspection revealed that
gravel particles were fractured as a result of the impact of the
metallic spheres during the Los Angeles test. This created new par-
ticles with more angular edges, which affected the angularity mea-
surements. In addition, the fracture of the aggregate particles also
450 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453

revealed internal surfaces of the particles that were rougher than
the original external surfaces. Although indicating that the Los
Angeles abrasion may not be the most adequate procedure to eval-
uate the ability of aggregate particles to retain their morphological
characteristics, these results also indicate that crushing gravel par-
ticles to be used in the composition of asphalt mixtures in the field
may be a good strategy to improve some of the aggregate morpho-
logical characteristics, such as angularity and texture.
3.2. Asphalt mixture performance behavior
Fig. 3a shows dynamic modulus master curves of the six mix-
tures evaluated at the reference temperature of 25 °C. The crushed
gneiss-only mixtures presented similar dynamic moduli and were
generally stiffer than the mixture composed solely of gravel. Mix-
tures composed of combinations of cubic crushed gneiss and gravel
presented similar stiffness characteristics and were stiffer than the
mixture composed of gravel only.
In an attempt to make a clearer comparison among the different
mixtures, dynamic modulus values at three distinct frequencies,
104, 10+1, and 10+6 Hz were plotted in Fig. 3b, c, and d, respec-
tively. In each figure, the dynamic moduli were normalized by
the maximum stiffness obtained among the six mixtures at the
corresponding frequency evaluated.
At 104 Hz, the mixture composed of F&E gneiss was 11% stiffer
than the mixture containing cubic gneiss. The combinations of
cubic gneiss and gravel resulted in mixtures with similar stiffness
which corresponded to 70% of stiffness of the mixture with F&E
gneiss. The gravel-only mixture presented the lowest stiffness that
was 34% lower than that of the mixture with F&E gneiss.
At 10+1 Hz, the cubic gneiss mixture was the stiffest and was
followed by the mixture with F&E gneiss, which was 8% less stiff.
Both mixtures were stiffer than the other mixtures evaluated.
The combinations of different aggregates resulted in mixtures with
similar stiffness values, which were around 20% lower than that of
the mixture with cubic gneiss. The gravel-only mixture was 44%
less stiff than the mixture with cubic gneiss.
At 10+6 Hz, the cubic gneiss mixture was also the stiffest. At this
frequency, all mixtures but the one composed solely of cubic gneiss
presented similar stiffness values that were between 14% and 22%
lower than that of the mixture composed of cubic gneiss. Finally,
mixtures containing F&E gneiss-only or gravel-only were the most
susceptible to variations in the loading frequency. The relative
stiffness of the F&E gneiss mixture decreased with the increasing
loading frequency, whereas the gravel-only mixture presented
the lowest relative stiffness at the intermediate frequency
evaluated.
Table 5 shows FN results at 60 °C. The mixture composed of
cubic crushed gneiss was the most resistant to permanent
deformation, with higher FN and lower b-parameter, whereas the
gravel mixture was the most susceptible to permanent deforma-
tion. As more cubic crushed gneiss was combined with gravel, a

F&E Gneiss Cubic Gneiss 1.25

Gravel 75% Gneiss 25% Gravel 1.00
Normalized Dynamic

1.E+05
50% Gneiss 50% Gravel 25% Gneiss 75% Gravel 1.00 0.89
Modulus

0.70 0.69 0.72 0.66

Dynamic Modulus (MPa)

0.75

1.E+04 0.50

0.25

1.E+03 0.00

1.E+02
1.E-05 1.E-02 1.E+01 1.E+04 1.E+07

Reduced Frequency (Hz)

(a) (b)

1.25 1.25
0.92 1.00 1.00
Normalized Dynamic

Normalized Dynamic

1.00 0.83 0.82 1.00 0.83 0.86 0.83 0.85 0.78

0.75
Modulus

Modulus

0.75 0.56 0.75

0.50 0.50

0.25 0.25

0.00 0.00

(c) (d)
Fig. 3. Dynamic modulus results at 25 °C: (a) master curves for all mixtures; results at (b) 1E4 Hz; (c) 1E+1 Hz; and (d) 1E+6 Hz.
 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453 451

Table 5 Table 6
FN results. Coefficients of determination between morphological properties of aggregates and
performance of asphalt mixtures.
Mixture Flow number b (le/cycle)
Correlations with FN
F&E gneiss 203 32.3
Cubic gneiss 340 22.0 Test (Standard) R2
75% gneiss, 25% gravel 243 31.8
AIMS 2 – texture 0.84
50% gneiss, 50% gravel 187 37.5
AIMS 2 – CAAT 0.84
25% gneiss, 75% gravel 153 46.7
AIMS 2 – CAA 0.84
Gravel 107 68.9
Superpave – CAA [29] 0.82
Uncompacted voids – CAA [19] 0.76
Superpave – FAA [30] 0.74
considerable enhancement on the resistance to rutting was AIMS 2 – FAA 0.54
AIMS 2 – sphericity 0.30
observed. This resulted from the better aggregate interlock and
Superpave – F&E 5:1 [28] 0.25
higher internal friction among particles with cubic shapes and F&E 2:1 [28] 0.21
rougher surfaces, which improved the ability of the mixtures to F&E 3:1 [28] 0.10
resist damage by the accumulation of permanent deformation.
Correlations with fracture energy
IDT fracture energies at 20 °C are presented in Fig. 4. The mix-
ture with F&E crushed gneiss was the least resistant to fracture. Test (Standard) R2
Mixtures containing combinations of cubic crushed gneiss and Superpave – F&E 5:1 [28] 0.34
gravel could absorb more energy before fracture than the mixture AIMS 2 – texture 0.31
composed solely of gravel. In addition, the resistance to fracture AIMS 2 – CAAT 0.30
AIMS 2 – CAA 0.30
increased as more cubic crushed gneiss was added to the mixtures. AIMS 2 – sphericity 0.30
Superpave – CAA [29] 0.28
3.3. Correlations between morphological properties and mixture Superpave – FAA [30] 0.24
Uncompacted voids for CAA [19] 0.22
performance
AIMS 2 – FAA 0.07
F&E 3:1 [28] 0.06
Before correlating aggregate properties with mixture perfor- F&E 2:1 [28] 0.01
mance, ANOVA (Analysis of Variance) tests were conducted to
determine if there is significant statistical difference between the
FN and IDT fracture energy values from the different combinations key morphological characteristic should be included in aggregate
of aggregates evaluated in this study. An a-value of 0.05 was specifications, given that it affects both functional (e.g., skid resis-
selected to conduct this analysis and resulted in a critical F-value tance) and structural properties of asphalt mixtures and
of 3.11. Considering the F-values of 8.59 and 2.93 obtained for pavements.
the results of the FN and IDT fracture energy tests, respectively, Table 6 also shows that coarse aggregate properties generally
it can be concluded that there is a significant difference among presented higher correlation with the resistance to rutting than
the FN values but not among the IDT fracture energy values. fine aggregate properties, except for the three F&E ratios evaluated,
Table 6 indicates the coefficients of determination (R2) obtained i.e., 5:1, 3:1, and 2:1, which presented relatively low R2 values of
from the comparison between the aggregate morphological prop- 0.25, 0.10, and 0.21, respectively. However, the F&E aggregate
erties and the performance behavior of the corresponding asphalt characteristic is typically more related to construction issues than
mixtures. The three properties that were best correlated with the to problems during the service life because F&E particles tend be
resistance to rutting were obtained from AIMS 2 and presented less resistant to fracture during the compaction of asphalt
R2 equal to 0.84, as seen in Fig. 5. Those properties were texture, mixtures.
coarse aggregate angularity-texture index, and coarse aggregate As for the comparison between aggregate properties and IDT
angularity. fracture energy results, R2 values were much lower than those
In addition to highlight the good correlation of AIMS 2 aggre- obtained in the comparison with FN results. This was expected
gate properties with the mixture resistance to rutting, these results because it is a common sense in the flexible pavement research
also clearly indicate the relevance of texture as a particle charac- community that the resistance to fatigue damage of asphalt mix-
teristic that affects the mixture performance. This implies that this tures is more related to binder than to aggregate properties.

16,000
Fracture Energy (Pa)

14,000 13,380
12,523 12,294
12,000 11,420 11,595 11,511

10,000

8,000

Fig. 4. IDT fracture energy test results for the six mixtures evaluated in this study.
452 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453

400 400
350 y= 1.6326x0.8739 350
y = 29.71x0.4706
300 R² = 0.8406 300 R² = 0.8183

FN (cycles)
FN (cycles)
250 250
200 200
150 150
100 100
50 50
0 0
100 200 300 400 0 50 100 150
Texture (AIMS scale) CAA

400 400
350 y= 0.0183x1.1297 350 y = 5E-07x2.5299
300 R² = 0.8396 300 R² = 0.8353

FN (cycles)
FN (cycles)

250 250
200 200
150 150
100 100
50 50
0 0
0 2000 4000 6000 0 1000 2000 3000 4000

CAAT CAA-AIMS 2
Fig. 5. Correlation of AIMS 2 texture, CAA, and CAAT with flow number results for the six mixtures evaluated in this study.

Aggregate characteristics determined from tests in the AIMS 2
system, i.e., texture, coarse aggregate angularity-texture index,
coarse aggregate angularity, and sphericity, were the ones which
presented the highest R2 values, along with the Superpave F&E
5:1 test.
As also observed from the FN results, fine aggregate properties
were not as strongly correlated with the resistance to fracture-
related damage as the coarse aggregate properties. Additionally,
texture presented a high correlation with the IDT fracture energy
in comparison to the other aggregate properties evaluated. This
highlights again the importance of texture as a critical aggregate
characteristic to be considered in existing specifications.
Finally, a comparison between the aggregate properties and the
dynamic modulus (|E⁄|) of the mixtures at 10 Hz was also made, as
shown in Table 7. R2 values were considerably high, as it was also
observed in the analysis of correlation between the particle prop-
erties and the mixture FN. The best correlation was found with
Table 7
Coefficients of Determination between Morphological Properties of Aggregates and
Dynamic Modulus of Asphalt Mixtures.
Superpave CAA, followed by properties of coarse aggregates deter-
mined using AIMS 2, i.e., texture, CAAT, and CAA. Again, the prop-
erties related with aggregate angularity and texture were the ones
that presented the best correlations with the mixture stiffness.
4. Summary and conclusions
This paper evaluated the effects of morphological properties of
aggregates on the mechanical behavior of asphalt mixtures. Three
different aggregates, i.e., two crushed gneisses with different
shapes and one gravel, were used in the composition of six
Superpave asphalt mixtures. Several aggregate properties were
characterized based on Superpave procedures for typical aggregate
consensus properties. Additionally, morphological aggregate
characteristics were also determined in a modern image analysis
system, the AIMS 2. The morphological properties were evaluated
before and after Los Angeles abrasion tests to assess the ability of
the different aggregates to retain their characteristics after the
abrasion process. The mechanical behavior of the mixtures was
also evaluated based on several test results of dynamic modulus,
flow number, and IDT fracture energy.

Correlations with Stiffness (|E⁄| @ 10 Hz) Based on the results, the following conclusions could be drawn:
Test (Standard) R2
Superpave – CAA [29] 0.78
Aggregate morphological characteristics were shown to be
AIMS 2 – texture 0.73 strongly correlated with the resistance to rutting of the asphalt
AIMS 2 – CAAT 0.72 mixtures evaluated. Weaker correlations were found with the
AIMS 2 – CAA 0.70
resistance to fracture of the mixtures. The observation implies
Uncompacted voids – CAA [19] 0.69
Superpave – FAA [30] 0.64
that the resistance to rutting of asphalt mixtures is dominated
AIMS 2 – FAA 0.59 by aggregate characteristics, whereas fatigue may be more
AIMS 2 – sphericity 0.44 related to binder properties. Test results also showed that the
F&E 2:1 [28] 0.38 mixture stiffness can be influenced by the aggregate morpho-
F&E 3:1 [28] 0.27
logical properties, especially at intermediate to low loading
Superpave – F&E 5:1 [28] 0.04
frequencies.
 F.T.S. Aragão et al. / Construction and Building Materials 123 (2016) 444–453

Among the properties evaluated, coarse aggregate characteris-
tics were generally better correlated with the resistance to
damage than fine aggregate characteristics. Additionally, prop-
erties obtained from the AIMS 2 system presented stronger cor-
relations with mixture performance than those characterized
from conventional Superpave procedures. This infers a need of
a more direct and objective measurement of aggregate charac-
teristics such as the AIMS 2 than procedures such as the current
Superpave methods which are generally based on indirect and
subjective approaches to measure aggregate shape properties.

Test-analysis results also indicated that the surface texture of
the aggregate particles should be carefully considered in aggre-
gate and asphalt mixture specifications. Strong correlations
were observed between this property and the performance
behavior of asphalt mixtures.
Acknowledgements
The authors are grateful for the financial support received from
the Carlos Chagas Filho Foundation for Research Support of the
State of Rio de Janeiro (FAPERJ, Brazil) and from the National Council
for Scientific and Technological Development (CNPq, Brazil).
References
[1] P.S. Kandhal, F. Parker Jr., Aggregate Tests Related to Asphalt Concrete
Pavement Performance in Pavements. NCHRP Report 405, National
Cooperative Highway Program. Transportation Research Board, 1998.
[2] J.K. Stiady, S. Galal, A.H. Noureldin, T. White, Identification of aggregate role in
performance of Superpave mixtures employing accelerated testing facility,
ASTM Spec. Tech. Publ. 1412 (2001) 44–57.
[3] B.D. Prowell, J. Zhang, E.R. Brown, Aggregate Properties and the Performance of
Superpave-Designed Hot Mix Asphalt. NCHRP Report 539, National
Cooperative Highway Research Program. Transportation Research Board, 2005.
[4] T.D. White, J.E. Haddock, E. Rismanjoto, Aggregate Tests for Hot-mix Asphalt
Mixtures Used in Pavements. NCHRP Report 557, National Cooperative
Highway Research Program. Transportation Research Board, 2006.
[5] T. Pan, E. Tutumluer, S.H. Carpenter, Effect of coarse aggregate morphology on
permanent deformation behavior of hot mix asphalt, J. Transp. Eng. (2006)
580–589.
[6] R. Sukhwani, D.N. Little, E. Masad, Sensitivity of HMA Performance to
Aggregate Shape Measured Using Conventional and Image Analysis Methods.
TTI Report 0-1707-5, Texas Transportation Institute, College Station, 2006.
[7] L.T.S. Souza, Y.R. Kim, F.V. Souza, L.S. Castro, Experimental testing and finite-
element modeling to evaluate the effects of aggregate angularity on
bituminous mixture performance, J. Mater. Civ. Eng. 24 (3) (2012) 249–258.
[8] E. Masad, T. Al-Rousan, J. Button, D. Little, E. Tutumluer, Test Methods for
Characterizing Aggregate Shape, Texture, and Angularity. NCHRP Report 555,
National Cooperative Highway Research Program. Transportation Research
Board, 2007.
[9] I. Bessa, V. Castelo Branco, J. Soares, Caracterização de agregados convencionais
e alternativos utilizando técnicas de processamento digital de imagens,
Monogr. CBUT (2009).
[10] I. Bessa, V. Castelo Branco, J. Soares, Avaliação do processo de produção de
agregados por medio da utilização do processamento digital de imagens,
Associação de Pesquisa e Ensino em Transportes-ANPET (2011).
[11] I. Bessa, V. Castelo Branco, J. Soares, Evaluation of polishing and degradation
resistance of natural aggregates and steel slag using AIMS, 2nd Symposium on
Asphalt Pavements and Environment, 2012.
[12] Y.R. Kim, L.T. Souza, Effects of Aggregate Angularity on Mix Design
Characteristics and Pavement Performance. Report No. MPM-10, Mid-
America Transportation Center, University of Nebraska-Lincoln, 2009.
[13] E. Mahmoud, L. Gates, E. Masad, S. Erdogan, E. Garboczi, Comprehensive
evaluation of AIMS texture, angularity, and dimension measurements, J. Mater.
Civ. Eng. 22 (2010) 369–379.
453
[14] AASHTO M323, Standard Specification for Superpave Volumetric Mix Design,
American Association of State Highway and Transportation Officials, 2013.
[15] B.J. Coree, W.P. Hislop, The Difficult Nature of Minimum VMA: A Historical
Perspective, Iowa State University, 1998.
[16] B.A. Chadbourn, E.L. Shok Jr., D.E. Newcomb, B.L. Crow, S. Spindler, The Effect of
Voids in Mineral Aggregate (VMA) on Hot Mix Asphalt Pavements. Report MN/
RC-2000-13, Minnesota Department of Transportation, 1999.
[17] W.P. Hislop, B.J. Coree, VMA as a design parameter in hot-mix asphalt, Mid-
Continent Transportation Symposium, 2000, pp. 24–29.
[18] T. Aschenbrener, C. Mackean, Factors that affect the voids in the mineral
aggregate of hot-mix asphalt, Transportation Research Record: Journal of the
Transportation Research Board, No. 1469, Transportation Research Board of
the National Academies, Washington, D.C., 1994, pp. 1–9.
[19] AASHTO T326, Uncompacted Void Content of Coarse Aggregate (As influenced
by particle shape, surface texture, and grading), American Association of State
Highway and Transportation Officials, 2013.
[20] P.S. Kandhal, F. Parker Jr., Aggregate Tests Related to Asphalt Concrete
Pavement Performance in Pavements. NCHRP Report 405, National
Cooperative Highway Program. Transportation Research Board, 1998.
[21] AASHTO T342, Standard Method of Test for Determining Dynamic Modulus of
Hot Mix Asphalt (HMA), American Association of State Highway and
Transportation Officials, 2011.
[22] M.W. Witczak, K. Kaloush, T. Pellinen, M. El-Basyouny, H. Von Quintus,
Simple Performance Test for Superpave Mix Design. NCHRP Report 465,
National Cooperative Highway Program. Transportation Research Board,
2002.
[23] Y.R. Kim, J.S. Daniel, H. Wen, Fatigue Performance Evaluation of WestTrack
Asphalt Mixtures Using Viscoelastic Continuum Damage Approach.
Research Project No. HWY-0678, North Carolina Department of
Transportation, 2002.
[24] H. Wen, Y.R. Kim, Simple performance test for fatigue cracking and validation
with WesTrack mixtures, Transportation Research Record: Journal of the
Transportation Research Board, No. 1789, Transportation Research Board of
the National Academies, Washington, D.C., 2002, pp. 66–72.
[25] G. Hondros, The evaluation of Poisson’s ratio and the modulus of materials of a
low tensile resistance by the Brazilian (Indirect Tensile) test with particular
reference to concrete, Aust. J. Appl. Sci. 10 (1959) 243–268.
[26] M.S. Buchanan, Evaluation of the Effect of Flat and Elongated Particles on the
Performance of Hot Mix Asphalt Mixtures. Report 00-03, National Center for
Asphalt Technology (NCAT), 2000.
[27] ASTM C127, Standard Test Method for Relative Density (Specific Gravity) and
Absorption of Coarse Aggregate, American Society of Testing and Materials,
2015.
[28] ASTM D4791, Standard Test Method for Flat Particles, Elongated Particles, or
Flat and Elongated Particles in Coarse Aggregate, American Society of Testing
and Materials, 2010.
[29] ASTM D5821, Standard Test Method for Determining the Percentage of
Fractured Particles in Coarse Aggregate, American Society of Testing and
Materials, 2013.
[30] AASHTO T304, Standard Method of Test for Uncompacted Void Content of Fine
Aggregate, American Association of State Highway and Transportation
Officials, 2011.
[31] AASHTO T176, Standard Method of Test for Plastic Fines in Graded Aggregates
and Soils by Use of the Sand Equivalent Test, American Association of State
Highway and Transportation Officials, 2013.
[32] AASHTO T96, Standard Method of Test for Resistance to Degradation of
Small-Size Coarse Aggregate by Abrasion and Impact in the Los
Angeles Machine, American Association of State Highway and Transportation
Officials, 2010.
[33] F.L. Roberts, P.S. Kandhal, E.R. Brown, D.Y. Lee, T.W. Kennedy, Hot Mix Asphalt
Materials, Mixture Design, and Construction. NCHRP Report 539, second ed.,
National Asphalt Pavement Association Research and Education Foundation,
1996.
[34] E. Mahmoud, Development of experimental methods for the evaluation of
aggregate resistance to polishing, abrasion, and breakage (Thesis for the
degree of Master of Science), Texas A&M University, 2005.
[35] E.A. Masad, A. Luce, E. Mahmoud, Implementation of AIMS in Measuring
Aggregate Resistance to Polishing, Abrasion and Breakage. Report No. FHWA/
TX-06/5-1707-03-1, Texas Transportation Institute, Texas A&M University,
2006.
[36] E. Masad, A. Rezaei, A. Chowdhury, P. Harris, Predicting Asphalt Mixture Skid
Resistance Based on Aggregate Characteristics. Report No. FHWA/TX-09/0-
5627-1, Texas Department of Transportation, Texas A&M University, 2009.