You are on page 1of 9

86 TRANSPORTATION RESEARCH RECORD 1530

Effects of Mineral Fillers on Properties of


Stone Matrix Asphalt Mixtures
WALAA S. MOGAWER AND KEVIN D. STUART

Stone matrix asphalt (SMA) is a gap-graded hot mixture that has higher the mineral fillers in pavements and the filler and mastic proper-
proportion of coarse aggregate, lower proportion of middle-size aggre- ties. The gradations and specific gravities of the fillers were mea-
gate, and higher proportion of mineral filler than a dense-graded mix- sured and the stiffening power of the mastics was measured using
ture. Because of the high quantity of mineral filler, the type and quantity
the ring-and-ball apparatus and a dynamic shear rheometer. For
of this fraction should play a major role in the properties of these mix-
tures and their mastics. Determination of whether mastic and mixture the study documented in this paper, eight of the European mineral
properties can distinguish good mineral fillers from bad ones is the fillers were chosen. The eight mineral fillers, their origins, and
objective. Eight mineral fillers with known performance were obtained reported performance are tabulated in Table 1. The previous study
from three European countries. Mastics were tested for stiffness using a concluded
bending beam rheometer (AASHTO TP1), a dynamic shear rheometer
(AASHTO TP5), and softening point of bitumen using the ring-and-ball
The percent voids in a compacted sample of mineral filler, as found
apparatus (AASHTO T53-92). None of the tests distinguished among through the Anderson modified Rigden voids method, was the only
mastics with good and bad mineral fillers. Mixtures were tested for independent variable that could singly distinguish between most of the
draindown of mastic using the National Center for Asphalt Technology ‘good’ and ‘bad’ fillers. With the range for ‘good’ fillers being between
draindown test, rutting using the French pavement rutting tester, low- 34% and 39%, only one ‘good’ filler was out of this range (2).
temperature cracking using the indirect tensile test, workability using
the U.S. Corps of Engineers gyratory testing machine, and moisture
susceptibility using the ASTM D 4867 method. None of the tests The eight mineral fillers were chosen on the basis of their
distinguished among SMA mixtures with good and bad mineral fillers. reported performance, gradations, and the Anderson modified Rig-
den voids. Mineral fillers with Anderson modified Rigden voids
within and outside of the 34 to 39 percent range were selected. The
Stone matrix asphalt (SMA) is a gap-graded hot mixture that has GER-8 mineral filler was chosen specifically because its perfor-
higher proportion of coarse aggregate, lower proportion of middle- mance was reported as good; however, its Anderson modified Rig-
size aggregate, and higher proportion of mineral filler than a dense- den voids (32.1 percent) was less than the lower limit (34 percent)
graded mixture. A stabilizing additive composed of cellulose fibers, for good mineral fillers obtained in the preceding study. The grada-
mineral fibers, or polymers is added to SMA mixtures to prevent tions, specific gravities, and Anderson modified Rigden voids of
draindown of their mastics. The mastic in SMA mixtures is the sum the eight mineral fillers are presented in Table 2.
of the filler fraction of the aggregate, the added filler, the binder, The Anderson modified Rigden voids can be used to predict the
and the stabilizing additive. FHWA SMA Technical Working stiffening effect of a mineral filler on an asphalt (2). The volume of
Group (TWG) recommends using between 8 and 13 percent aggre- the voids in a compacted sample of dry mineral filler is equated to the
gate by weight finer than 75 µm in SMA mixtures (1). Because the fixed volume of the binder in the mastic. This volume of the binder is
quantity of minus 75-µm material in SMA mixtures is relatively unavailable to participate in the mixture as a whole. The hypothesis
large, the type and quality of this fraction should play a significant is that when the percent voids in a compacted sample of filler is too
role in the properties of SMA mastics and mixtures. Also, mineral high, not enough binder is free, leading to exceedingly stiff mixtures
fillers affect workability, moisture resistance, stiffness, and aging or mixtures susceptible to cracking failure. If the percent voids is too
characteristics of hot mix asphalt (HMA) mixtures (2–4 ). There- low (too much free binder), then the mixtures will be susceptible to
fore, the effects of different mineral fillers on SMA mastics and excessive draindown, bleeding, and rutting (2).
mixtures need to be examined. Figure 1 describes the parameters used to calculate the percent-
age of free binder in each mixture. To determine the percent free
binder, the volume of the filler (solid volume) used in a mixture is
SOURCE AND PROPERTIES OF first calculated from the mass used in the mixture and the apparent
MINERAL FILLERS specific gravity. The Anderson modified Rigden voids and the vol-
ume of the solids are used to calculate the volume of the voids (vol-
Eighteen mineral fillers collected from Sweden (SWE), Germany ume of fixed binder content). The fixed binder content is subtracted
(GER), and Switzerland (CHE) were available at the FHWA’s from the total volume of binder used in a mixture to calculate the
Turner Fairbank Highway Research Center. The performance of volume of free binder.
these mineral fillers was reported by the users as clearly good or
bad for use in SMA or HMA. A preceding study used these min-
eral fillers to explore the relationships between the performance of OBJECTIVE
W. S. Mogawer, University of Massachusetts Dartmouth, 285 Old Westport
Rd., North Dartmouth, Mass. 02747-2300. K. D. Stuart, FHWA, 6300 The objective of this study was to determine whether mastic and
Georgetown Pike, McLean, Va. 22101-2296. mixture properties can distinguish good mineral fillers from bad
TABLE 1 Origins and Reported Performance of Mineral Fillers

TABLE 2 Gradations and Physical Properties of Mineral Fillers and Free Binder Contents
88 TRANSPORTATION RESEARCH RECORD 1530

tory testing machine, and for moisture susceptibility using the


ASTM D 4867 method.
The results were statistically analyzed using Fisher’s least sig-
nificance difference procedure (LSD). The Fisher’s LSD procedure
was performed at a level of significance α = 0.05. Fisher’s LSD
determined which averages were not statistically different.

MIXTURE DESIGN

The mixtures were designed to sustain heavy traffic using 50


blows. SMA mixtures are designed in Europe for heavy traffic
using 50 blows (5). The design air-void content was 3.5 percent.
The aggregate gradations met the recommended gradations given
by the SMA TWG (1).
The aggregate blend consisted of No. 68 diabase (aggregates
retained on the 4.75-mm sieve) and No. 10 diabase (aggregates
finer than the 4.75-mm sieve and retained on the 0.150-mm sieve).
The diabase aggregates were obtained from Virginia Trap Rock, in
Leesburg, Virginia. A cellulose fiber, trade-named Technocel, was
used as a stabilizer. The Technocel was added at 0.3 percent by
FIGURE 1 Volumetric parameters used to describe voids in mixture mass, based on the supplier’s recommendation.
mineral filler and mastic. To evaluate the effects of the added filler on the SMA mixtures,
it was necessary to remove the aggregate’s filler fraction. There-
fore, the diabase aggregates were sieved down to the 0.150-mm
mineral fillers. A second objective was to determine whether free sieve and washed to remove all material finer than this size (0.150-
binder content relates to mastic and mixture properties. mm was the maximum size of the mineral fillers). The aggregates
were then blended. A washed sieve analysis was performed on the
blend to determine the amount of the minus 0.150-mm material in
EXPERIMENTAL PROGRAM
the blend after washing. Only 0.6 percent of the material was finer
than the 0.150-mm sieve. This percentage was considered insignif-
The experimental design involved performing Marshall mix
icant. The results of the washed sieve analysis are presented in
designs and testing the mastics and the mixtures. The mastics and
Table 3. The fillers were then added to the blend.
the mixtures were given the same designations as the mineral fillers
Table 3 shows the gradations of the eight mixtures. Seven of the
in the preceding study (2).
eight mixtures had equivalent gradations from the 19.0-mm sieve
The mastics and mixtures were tested to determine the effects of
to the 75-µm sieve. One mixture (SWE-1) had a different gradation
mineral fillers on them. Mastics were tested because if a measured
from the 0.600-mm sieve to the 0.150-mm sieve size because the
property of the mastics was found to adversely affect the perfor-
SWE-1 filler had a very coarse gradation (only 60.8 percent passed
mance of the mixtures the test method can be used to optimize the
the 75-µm sieve).
properties of the mastics.
Eight Marshall mix designs were performed to determine the
optimum asphalt contents (OAC) of the SMA mixtures prepared EVALUATION OF MASTICS
with the eight mineral fillers.
Mastics were prepared by blending each mineral filler with the The eight mastics were prepared by blending each filler with the
asphalt using the filler-to-asphalt ratio by volume at the OAC for neat binder on a hot plate using the ratios of the filler to neat binder
each mixture. A stabilizing additive was not included in the mas- by volume that were used in the mixtures. The proportions of the
tics; however, if a test was able to distinguish between mastics with filler to the neat binder in each mastic are presented in Table 2. The
good and bad mineral fillers, it was planned to retest them with the neat binder was an AC-20. The GER-7 filler required the most time
stabilizing additive. to be coated with the binder. This mastic was extremely gummy
The mastics were tested to determine the effects of the mineral and very difficult to pour.
fillers on their stiffnesses using a bending beam rheometer (AASHTO Table 2 shows the percentage of free binder in each mixture.
TP1), the softening point of bitumen using the ring-and-ball The GER-4, GER-7, and SWE-1 had the lowest percent free
apparatus (AASHTO T53-92), and a dynamic shear rheometer binder contents among the eight mastics and mixtures; based on
(AASHTO TP5). the hypothesis, these three fillers should provide the stiffest mas-
The mixtures were tested to evaluate the effects of the mineral tics and mixtures. The SWE-7 had the highest free binder content;
fillers on SMA’s properties. Draindown of the mastic, rutting, low therefore, this filler should provide the stiffest mastic and mixture.
temperature cracking, workability, and moisture susceptibility
were measured. The draindown of the mastic was evaluated using
the National Center for Asphalt Technology (NCAT) draindown Bending Beam Rheometer (AASHTO TP1)
test. The mixtures were tested for rutting using the French pave-
ment rutting tester, for low temperature cracking using the indirect A bending beam rheometer (BBR) was used to measure the stiff-
tensile test, for workability using the U.S. Corps of Engineers gyra- ness of each mastic at –28°C. The mineral fillers with the lowest
Mogawer and Stuart 89

TABLE 3 Washed Aggregate Gradations of Mixtures and Diabase Aggregate

free binder contents (GER-4, GER-7, and SWE-1) were expected as the increase needed in test temperature so that the stiffness (G*
to produce the stiffest mastics. The BBR measures the stiffness of measured by the DSR) of the mastic matched the stiffness of the
a small asphalt beam sample under creep load. A 980 ±5mN load neat binder. The G* of the neat binder was measured at 50°C,
was applied for 240 sec and the deflection of the beam was mea- which was its softening point. The mineral fillers that had the low-
sured during this period. Two parameters were evaluated with the est free binder contents (GER-4, GER-7, and SWE-1) were
BBR: (a) the creep stiffness at 60 sec loading time—a measure of expected to have the highest stiffening power.
how the asphalt resists loading—and (b) the m-value—a measure
of how the asphalt stiffness changes as the load is applied.
EVALUATION OF MIXTURES
Softening Point of Bitumen Using Ring-and Ball
Apparatus (AASHTO T 53-92) The effects of the mineral fillers on SMA properties in terms of
draindown of the mastic, rutting, low temperature cracking, work-
The stiffening powers of the mastics were measured as the differ- ability, and moisture susceptibility were evaluated. To measure the
ence between the softening point (AASHTO T 53-92) of a mastic draindown of the SMA mixtures, loose mixtures were used. To
and the softening point of the neat binder. The stiffening power is measure the other properties, the SMA mixtures were compacted to
used in Germany to qualify mineral fillers for use in pavements. In approximately 5.5 percent air voids. A target air void range of 5 to
this study, the mineral fillers with the lowest free binder contents 6 percent was used because an SMA mixture should have less than
(GER-4, GER-7, and SWE-1) were expected to produce the stiffest 6 percent air voids after compaction in the field.
mastics. The Germans use a range in stiffening power for accept-
able mineral fillers of 10°C to 20°C (2). Mastics that exceed 20°C
would be considered too stiff and the mineral fillers would be dis- Draindown
qualified to avoid placing crack susceptible pavements. Mastics
that test below 10°C would not be stiffened enough; these would be The NCAT draindown test was used to evaluate the effects of the
susceptible to excessive draindown, bleeding, shoving, and rutting. different mineral fillers on the draindown of the mastic. In this test,
In testing the mastics and the neat binder, the standard testing the sample is placed in a wire basket that is positioned on a
procedure for the softening point of bitumen, AASHTO T 53, was preweighed, dry paper plate. The sample, basket, and plate are
followed using ethylene glycol as the bath liquid. The softening placed in a forced-air oven for 60 min at 143°C. The basket con-
point of the neat binder was found to be 50°C. taining the sample is then removed from the oven along with the
paper plate and the paper plate is weighed. The percent loss due to
Dynamic Shear Rheometer (AASHTO TP5) draindown is calculated as:

The stiffening powers of the mastics were also measured using a 100 (final paper mass − initial paper mass)
loss ( percent ) =
dynamic shear rheometer (DSR). This stiffening power was defined initial sample mass
90 TRANSPORTATION RESEARCH RECORD 1530

When this test was performed some particles of the mixture fell The hypothesis in this study stated that the lower the free binder
through the basket onto the paper plate. These particles were placed content in a mixture, the more susceptible to low-temperature
back into the basket before placing the basket in the oven. cracking the mixture will be compared with the mixtures with
higher free binder content. Therefore, the GER-4, GER-7, and the
SWE-1 mixtures were expected to be the most susceptible to low
Rutting
temperature cracking.
The French pavement rutting tester was used to determine the
effects of the mineral fillers on the rutting susceptibilities of SMA Workability
mixtures at 60°C. This wheel-tracking device tests slabs that are
500 × 180 × 50 mm or 100 mm thick using a smooth, reciprocat- The effects of the different mineral fillers on the workabilities of
ing, pneumatic rubber tire at 0.61 MPa and loaded at 5000 ± 50N. the SMA mixtures were evaluated using the U.S. Corps of Engi-
The tire is 415 mm in diameter and 109 mm wide. A hydraulic neers gyratory testing machine (GTM). Workability is the ease
jack underneath the slab pushes the slab upward to create the with which paving mixtures may be placed. The compaction efforts
load. The machine tests two slabs at a time. A slab thickness of required to compact specimens to an air-void level representative
50 mm is used if the pavement layer will be 50 mm or less. For of the average in-place field level (5.5 percent) after construction
layers thicker than 50 mm, 100-mm slabs are tested. SMA pave- were measured. The mixtures were compacted at 20 or 40 and
ment layers are generally less than 50 mm thick, hence 50-mm- 60 gyrations, or both, at 135°C with a vertical pressure of 120 psi
thick slabs were tested in this study. The slabs were fabricated and a 0.014-radian angle of gyration. The air voids at the different
using the French plate compactor. Two slabs were fabricated at number of gyrations were measured and the number of gyrations
the OAC per mixture. required to achieve a 5.5 percent air void level was calculated. The
For a 50-mm-thick slab, the test is performed to 3,000 cycles hypothesis stated that the mixtures with higher volumes of free
and the average rut depths at 30, 100, 300, 1,000, and 3,000 asphalt such as CHE-1, GER-8, GER-9, CHE-2, and SWE-7 would
cycles are measured. A cycle is two passes of the tire. If the aver- be easier to compact (i.e., requires fewer gyrations to reach 5.5 per-
age percent rut depth at 3,000 cycles is more than 20 percent of cent air voids) than mixtures with lower volumes of free asphalt
the slab thickness, the mixture is considered failed (unaccept- such as GER-4, GER-7, and SWE-1. Table 2 shows the volumes of
able). Slopes for different mixtures taken from log rut depth ver- the free asphalt in the mixtures.
sus log number of cycles plots can also be compared. Rut European suppliers of the GER-7 and SWE-1 mineral fillers pro-
susceptible mixtures generally have higher slopes. vided small quantities (5 kg). Insufficient material existed to deter-
The hypothesis in this study stated that the lower the free binder mine the effects of these two mineral fillers on the workabilities of
content in a mixture, the less susceptible to rutting the mixture will the mixtures.
be. Therefore, the GER-4, GER-7, and the SWE-1 mixtures were
expected to be the least susceptible to rutting.
Moisture Susceptibility

Low-Temperature Cracking ASTM D 4867 entitled Effect of Moisture on Asphalt Concrete


Paving Mixtures was used to determine whether the mineral fillers
The effects of the different mineral fillers on the resistances of the had any effect on the moisture susceptibilities of the mixtures.
SMA mixtures to thermal cracking were evaluated using the indirect Three unconditioned (dry) and three conditioned (wet) speci-
splitting tensile test at a low temperature with a loading rate of mens were tested per mixture. Wet specimens were (a) vacuum-
1.27 mm/min. This research study was performed at the FHWA’S saturated with distilled water so that 55 to 80 percent of their air
Turner Fairbank Highway Research Center in Virginia; Superpave voids were filled with water, (b) subjected to a freezing cycle for 15
states that the lowest temperature of a pavement in Virginia is hr at –18.0°C ±2.0°C, (c) soaked in a 60°C distilled water bath for
–28°C. Therefore, –28°C was used as the test temperature. This test 24 hr, and (d) tested at 25°C along with the dry specimens. Speci-
evaluated tensile strength (St), horizontal tensile strain at failure (et), mens were compacted to approximately 5.5 percent air voids by the
and the amount of work needed to cause tensile failure. The work is Marshall hammer. Use of an air void level near the upper allowable
the area under the stress-strain curve from the beginning of the test field level provides a more severe condition than lower levels.
until failure. Higher et and higher amounts of work are associated The diametral modulus-retained ratio (MdR) and tensile
with increased resistance to low-temperature cracking. The equation strength-retained ratio (TSR) were computed for each mixture in
used to compute et, assuming a Poisson’s ratio of 0.35, is (6): percentages. A retained ratio is the average wet value divided by
the average dry value. Statistics could not be applied to the
et = 0.0205 Ht retained ratios because they were averages.
The average percentage of visual stripping was also estimated for
where et is horizontal tensile strain at failure and Ht is total each mixture. Visual stripping is the percentage of area that is
horizontal deformation, mm. stripped relative to the total area of the split surfaces of the specimen.
The mixtures were short-term and long-term aged in a forced- The percent swell for each mixture was also measured. This is the
draft oven according to AASHTO provisional Method PP2-94. average change in the volumes of the specimens due to moisture
Short-term oven aging was performed on the loose mixtures at conditioning. A high amount of swell indicates internal damage.
135°C for 4 hr, and long-term oven aging was performed on com- A TSR below 80 percent, an MdR below 70 percent, and visual
pacted specimens at 85°C for 5 days. After aging, the specimens stripping above 10 percent are suggested criteria for considering a
were stored at room temperature for 24 hr and then placed in a mixture susceptible to moisture susceptibility. Swells equal to or
freezer at –28°C for 24 hr. greater than 1 percent were used as indicators of high internal dam-
Mogawer and Stuart 91

age, although a reliable limit on the percent swell has not been estab- LSD showed that the means of the mastics with good and bad
lished. These criteria were developed for dense-graded mixtures. mineral fillers are not significantly different. Based on the
Their applicability to SMA mixtures has not been established. hypothesis used in this test, the GER-4, GER-7, SWE-1 should be
Sufficient materials of the GER-7 and the SWE-1 mineral the stiffest mastics. The GER-4 and the SWE-1 satisfied the
fillers were not available to determine their effects on the mois- hypothesis; the GER-7 did not satisfy the hypothesis. The GER-7
ture susceptibilities of the mixtures. mastics was difficult to mix and pour; therefore it is questionable
whether the measured stiffness of the GER-7 is correct. There
was a good correlation (R2 = – 0.95) between the stiffness and the
RESULTS AND ANALYSIS free binder content.

Table 4 shows the results of the Marshall mixture design. The


VMAs of each SMA mixture did not vary with the binder content, Softening Point of Bitumen Using
and the air voids versus binder content relationships were linear. Ring-and-Ball Apparatus
Only three binder contents were used in designing the mixtures.
All mixtures except the GER-7 mixture had approximately 6.5 Table 5 shows the results of this test. The stiffening power was
percent binder by mass of mixture and a VMA of approximately unable to identify the mineral fillers based on performance. Never-
19. The GER-7 mixture had a 7.8 percent binder content and a theless, a good correlation (R2 = 0.92) existed between the stiffening
VMA of 22.3. The GER-7 mixture had a higher binder content and power and the free binder content.
VMA because its ratio of the filler to binder by volume was 27 to The GER-7 mastic was difficult to pour into the testing ring;
73 percent; the low filler volume in comparison to the other mix- therefore, it was not tested. The GER-4 and SWE-1 mastics were
tures was due to the high specific gravity of the GER-7 filler. Con- the only mastics that had a stiffening power value greater than the
sequently, more asphalt was used in the GER-7 mixture leading to 10°C to 20°C acceptable range set by the German method. This
the higher asphalt content and VMA. The volume of the filler in agrees with the hypothesis used in this test.
the GER-7 mixture was 14 cm3 lower than the volume of the min-
eral filler in the other mixtures, and the volume of the asphalt in
the GER-7 mixture was 14 cm3 higher than the volume of the DSR
asphalt in the other mixtures. (Future studies should consider
adding the mineral fillers by volume.) Table 2 shows the propor- Table 5 shows the results from the DSR. The stiffening power from
tions of the components of the mastic and the volume of the mas- the DSR failed to distinguish among mastics with good and bad
tic in each mixture. mineral fillers. A poor correlation (R2 = 0.49) existed between the
To determine whether the Marshall flow and stability can distin- stiffening power measured by the DSR and the free binder content.
guish among SMA mixtures with good and bad mineral fillers, the Furthermore, a poor correlation existed between the stiffening
LSD procedure was performed. The flows and the stabilities of the power found by the ring-and-ball test and the stiffening power
SMA mixtures with good and bad mineral fillers were not statisti- found by the DSR (R2 = 0.05).
cally different. Therefore, the Marshall flows and stabilities did not
distinguish among SMA mixtures with good and bad mineral fillers.
EVALUATION OF MIXTURES
EVALUATION OF MASTICS
Draindown
BBR
Table 5 shows the draindown data from the NCAT test. No drain-
Table 5 shows the results of the BBR test. The m-value was down of the mastics occurred from any of the mixtures. This was
unable to identify the mineral fillers based on performance. The attributed to the stabilizer in the mixture.

TABLE 4 Marshall Mix Design Properties of Mixtures


92 TRANSPORTATION RESEARCH RECORD 1530

TABLE 5 Properties of Mastics

Rutting Low-Temperature Cracking

Table 6 shows the rut depths and the percent rut depths of the Table 7 presents the results of the indirect tensile test. The LSD
eight mixtures. For 50-mm-thick slabs the test is performed to procedure was performed to determine whether the work, strain,
3,000 cycles; if the percent rut depth is greater than 20 percent of and the tensile strength can distinguish among SMA mixtures with
the slab thickness, then the mixture is considered a rut susceptible good and bad mineral fillers. The LSD showed that the works,
mixture. At 3,000 cycles all mixtures passed the test; therefore, strains, and tensile strengths of all the mixtures were not signifi-
this test did not distinguish between SMA mixtures with good and cantly different; therefore, they failed to distinguish among SMA
bad mineral fillers. mixtures with good and bad mineral fillers. Also, the variability
In an attempt to find a procedure that could distinguish between was high among the work and strain data.
SMA mixtures with good and bad mineral fillers, the test was A poor correlation existed between the stiffnesses of the mastics
extended to 10,000 cycles. At 10,000 cycles, all mixtures had a measured by the BBR and the properties of the mixtures measured
percent rut depth less than 20 percent. Therefore, none of the using the indirect tensile test. The R2 between the stiffnesses and
mixtures was rut susceptible. the tensile strength, strain, and work were 0.36, 0.44, and 0.13,
The LSD was performed on the percent rut depths at 3,000 respectively. A poor correlation existed between the tensile
and 10,000 cycles. Based on the LSD procedure, mixtures strengths and the free binder contents (R2 = 0.35).
with good mineral fillers and mixtures with bad mineral fillers
were not significantly different. Hence, the percent rut depths
using 50-mm slabs were not a function of the filler. A poor corre- Workability
lation existed between the percent rut depth and the free binder
Table 7 shows the results from the GTM. Because the two
contents. At 3,000 and 10,000 cycles, the R2 was 0.28 and 0.12,
mixtures (SWE-1 and GER-7) with the lowest volumes of
respectively. Furthermore, the results from this test did not agree
free asphalt could not be tested (insufficient materials), the
with the hypothesis because the GER-4, GER-7, and SWE-1,
hypothesis could not be verified. Mixtures with good mineral
which had the three lowest volumes of free asphalt, did not rut the
fillers such as CHE-1 and GER-9 and mixtures with bad mineral
least.
fillers such as CHE-2 and GER-4 required similar numbers of
In another attempt, 100-mm-thick slabs were prepared and
gyrations to reach a 5.5 percent air void level. Therefore, the
tested. Only two mixtures CHE-1 (good) and CHE-2 (bad) were
number of gyrations required to reach a 5.5 percent air void level
tested. The CHE-1 and CHE-2 mixtures were used because prepar- did not distinguish between SMA mixtures with good and bad
ing the thick slabs required a large amount of filler and the CHE-1 mineral fillers.
and CHE-2 mineral fillers were supplied in large quantities. When a
100-mm-thick slab is tested, a mixture is considered a rut suscepti-
ble mixture if the percent rut depth is greater than 10 percent of the Moisture Susceptibility
slab thickness at 30,000 cycles. The average rut depth was mea-
sured at 300, 1,000, 3,000, 10,000, and 30,000 cycles. The data Table 8 presents the moisture susceptibility test results. The GER-4
from testing the 100-mm slabs are presented in Table 6. The CHE- mixture was the only mixture that had failing retained ratios. The
1 (good) and the CHE-2 (bad) mixtures failed the test when testing conditioned specimens of the GER-4 mixture looked dull after
100-mm slabs instead of 50-mm slabs. Therefore, this test did not splitting them. One hypothesis was that the filler was stripped.
distinguish between SMA mixtures with good and bad mineral Nevertheless, all mixtures showed more than 10 percent visual
fillers. stripping; therefore, based on visual stripping all the mixtures are
Mogawer and Stuart 93

TABLE 6 Results from French Rutting Tester

susceptible to moisture damage. Statistically, the dry and wet ten- 2. The tests used to evaluate the mixtures did not distinguish
sile strengths and diametral moduli did not distinguish among among SMA mixtures with good and bad mineral fillers.
SMA mixtures with good and bad mineral fillers. A good correla- 3. There was a good correlation between the free binder
tion existed between the free binder contents and the tensile content and the stiffnessess of the mastics measured by the
strength ratios. Also, a good correlation existed between the free BBR and the stiffening power measured using the ring-and-
binder contents and the diametral modulus retained ratios; R 2 was ball apparatus. The ramification of this is unknown since the
0.76 and 0.85 respectively. tests did not distinguish between good and bad mineral
fillers.
4. A poor correlation existed between the stiffening power
CONCLUSIONS measured by the DSR and the free binder content.
5. A poor correlation existed between the percent rut depth
1. The tests used to evaluate the mastics did not distinguish measured by the French pavement rut tester and the free binder
among mastics with good and bad mineral fillers. content.
94 TRANSPORTATION RESEARCH RECORD 1530

TABLE 7 Results from Indirect Tensile Strength Test at T = –28°C and Gyratory Testing Machine

TABLE 8 Moisture Susceptibility Results

6. A poor correlation existed between the tensile strengths mea- 3. Anderson, D. A., J. P. Tarris, and J. D. Brock. Dust Collector Fines and
sured by the indirect tensile strength test and the free binder contents. Their Influence on Mixture Design. Association of Asphalt Pavings
Technologists, Vol. 51, Feb. 1982, pp. 363–397.
4. Puzinauskas, V. P. Filler in Asphalt Mixtures. Research Report 69-2.
REFERENCES The Asphalt Institute, Lexington, Ky., Feb. 1969.
5. Stuart, K. D. Stone Mastics Asphalt (SMA) Mixture Design. Report
1. SMA Technical Working Group. SMA Material and Construction FHWA-RD-92-006. FHWA, U.S. Department of Transportation,
Model Guidelines. Office of Technology Applications, FHWA, U.S. Washington, D.C., March 1992.
Department of Transportation, Jan. 1994. 6. Stuart, K. D., and P. Malmquist. Evaluation of Using Different Stabilizer
2. Harris, B. M., and K. D. Stuart. Analysis of Mineral Fillers and Mas- in the U.S. Route 15 (Maryland) Stone Matrix Asphalt (SMA). In Trans-
tics Used in Stone Matrix Asphalt. Association of Asphalt Paving portation Research Record 1454, TRB, National Research Council,
Technologists, Vol. 64, 1995. Washington, D.C., 1994, pp. 48–57.

You might also like