NCHRP w68v2

NCHRP Web Document 68 (Project 9-27)
Relationships of HMA In-Place Air

Voids, Lift Thickness, and
Permeability
Volume Two
Prepared for:
National Cooperative Highway Research Program
Submitted by:
E. Ray Brown
M. Rosli Hainin
Allen Cooley
Graham Hurley
National Center for Asphalt Technology
Auburn University
Auburn, Alabama
September 2004
ACKNOWLEDGMENT
This work was sponsored by the American

Association of State Highway and Transportation
Officials (AASHTO), in cooperation with the Federal
Highway Administration, and was conducted in the
National Cooperative Highway Research Program
(NCHRP), which is administered by the Transportation
Research Board (TRB) of the National Academies.
DISCLAIMER
The opinion and conclusions expressed or implied in

the report are those of the research agency. They are
not necessarily those of the TRB, the National
Research Council, AASHTO, or the U.S. Government.
This report has not been edited by TRB.
The National Academy of Sciences is a private, nonprofit, self-perpetuating society of distinguished scholars
engaged in scientific and engineering research, dedicated to the furtherance of science and technology and to
their use for the general welfare. On the authority of the charter granted to it by the Congress in 1863, the
Academy has a mandate that requires it to advise the federal government on scientific and technical matters. Dr.
Bruce M. Alberts is president of the National Academy of Sciences.
The National Academy of Engineering was established in 1964, under the charter of the National Academy of
Sciences, as a parallel organization of outstanding engineers. It is autonomous in its administration and in the
selection of its members, sharing with the National Academy of Sciences the responsibility for advising the
federal government. The National Academy of Engineering also sponsors engineering programs aimed at
meeting national needs, encourages education and research, and recognizes the superior achievements of
engineers. Dr. William A. Wulf is president of the National Academy of Engineering.
The Institute of Medicine was established in 1970 by the National Academy of Sciences to secure the services
of eminent members of appropriate professions in the examination of policy matters pertaining to the health of
the public. The Institute acts under the responsibility given to the National Academy of Sciences by its
congressional charter to be an adviser to the federal government and, on its own initiative, to identify issues of
medical care, research, and education. Dr. Harvey V. Fineberg is president of the Institute of Medicine.
The National Research Council was organized by the National Academy of Sciences in 1916 to associate the
broad community of science and technology with the Academy’s purposes of furthering knowledge and
advising the federal government. Functioning in accordance with general policies determined by the Academy,
the Council has become the principal operating agency of both the National Academy of Sciences and the
National Academy of Engineering in providing services to the government, the public, and the scientific and
engineering communities. The Council is administered jointly by both the Academies and the Institute of
Medicine. Dr. Bruce M. Alberts and Dr. William A. Wulf are chair and vice chair, respectively, of the National
Research Council.
The Transportation Research Board is a division of the National Research Council, which serves the
National Academy of Sciences and the National Academy of Engineering. The Board’s mission is to promote
innovation and progress in transportation through research. In an objective and interdisciplinary setting, the
Board facilitates the sharing of information on transportation practice and policy by researchers and
practitioners; stimulates research and offers research management services that promote technical excellence;
provides expert advice on transportation policy and programs; and disseminates research results broadly and
encourages their implementation. The Board's varied activities annually engage more than 5,000 engineers,
scientists, and other transportation researchers and practitioners from the public and private sectors and
academia, all of whom contribute their expertise in the public interest. The program is supported by state
transportation departments, federal agencies including the component administrations of the U.S. Department of
Transportation, and other organizations and individuals interested in the development of transportation.
www.TRB.org
www.national-academies.org
VOLUME TWO
TABLE OF CONTENTS
Page
LIST OF TABLES…………………………………………………………………………......... iii
LIST OF FIGURES ..……………………………………………………………………… ........ iv
1.0 INTRODUCTION AND PROBLEM STATEMENT……………………………... ..........1
2.0 OBJECTIVE……………………………………………………………………….. ..........8
3.0 SCOPE ……………………………………………………………………………... .........8
4.0 RESEARCH APPROACH…………………………………………………………. .........9
4.1 Laboratory Compacted Samples…………………………………………… ........10
4.2 Field Compacted Samples…………………………………………………..........13
5.0 TEST METHODS AND MATERIALS……………………………………… ........……13
5.1 Test Methods………………………………………………………………..........13
5.1.1 Saturated Surface-Dry Method (AASHTO T166).………………………13
5.1.2 Vacuum-Sealing Test Method……………………………………………14
5.1.3 Gamma Ray Method…………………………………………………... ...16
5.1.4 Dimensional Method ………………………………………………….....17
5.2 Materials ……………………………………………………………………… ...17
5.2.1 Tasks 3, Part 1 Laboratory Prepared Materials …………………….....…18
5.2.2 Task 5 Field Projects ………………………………………………….…22
i
6.0 TEST RESULTS AND ANALYSIS…………………………………………..………..23
6.1 Laboratory Compacted Samples ………………………………………………..24
6.1.1 Summary of Comparison Between Bulk Specific Gravity
Method for Laboratory Specimens ……………………………………..59
6.2 Analysis of Field Compacted Samples …………………………………………64
7.0 CONCLUSIONS AND RECOMMENDATIONS………………………….……….… 70
8.0 REFERENCES ……………………………………………………………………….... 72
ii
LIST OF TABLES
Page
Table 1: Physical Properties of Aggregtae….…………………………………….. 18
Table 2: Asphalt Binder Properties ………………………………………………. 19
Table 3: Summary of Mix Design Information for Superpave Mixes ……………. 22
Table 4: Summary of Mix Design Information for SMA Mixes ……….………… 22
Table 5: Summary Information on Field Projects from Task 5 ………………….. 23
Table 6: Results of Part 3 Testing for 9.5 mm NMAS Limestone Mixes………. 24
Table 7: Results of Part 3 Testing for 9.5 mm NMAS Granite Mixes ……..… 25
Table 8: Results of Part 3 Testing for 19.0 mm NMAS Limestone Mixes ….…… 26
Table 9: Results of Part 3 Testing for 19.0 mm NMAS Granite Mixes …………. 27
Table 10: Results of Part 3 Testing for Limestone SMA Mixes …………….…… 28
Table 11: Results of Part 3 Testing for Granite SMA Mixes ……………...…….. 29
Table 12: Results of Part 3 Testing for 37.5 mm NMAS Limestone Mixes …….. 30
Table 13: Results of Part 3 Testing for 37.5 mm NMAS Granite Mixes ………... 31
Table 14: Results of ANOVA Conducted on Superpave Designed Mixes ……… 33
Table 15: Results of ANOVA Conducted on SMA Designed Mixes …………….35
Table 16: Evaluation of Surface Texture for 19.0 mm NMAS-SMA-Granite
Samples…………………………………..…………………………..… 52
Table 17: Comparison of Field Samples by Water Absorption Level.…………… 67
iii
LIST OF FIGURES
Page
Figure 1: Hydrostatic Forces on a Submerged Material………………………… 3
Figure 2: Volumes Associated with Compacted HMA………………………… 6
Figure 3: Research Approach for Part 3 of Task 3………………………………. 10
Figure 4a: Vacuum-Sealing Device…………….………………………………… 14
Figure 4b: Sealed Sample…………………………………………………………. 14
Figure 5: Equipment for Gamma Ray Test……………………………………….. 16
Figure 6: 9.5 mm NMAS Superpave Gradations..………………………………… 19
Figure 7: 19.0 mm NMAS Superpave Gradations………………………………… 20
Figure 8: 37.5 mm NMAS Superpave Gradations ………………………………… 20
Figure 9: SMA Gradations………………………………………………………… 21
Figure 10: Average Air Voids and DMRT Results for Superpave Mixes………. 34
Figure 11: Average Air Voids and DMRT Results for SMA Mixes…………….. 36
Figure 12: Water Absorption Levels by NMAS, Gradation, and
Gyration Level……………………………………………………….... 38
Figure 13: Comparison of Test Methods for Low Water Absorption
Level Mixes…………………………………………………………… 39
Figure 14: Comparison Between Vacuum-Sealing and
AASHTO T166 Methods, 9.5 mm Limestone…………………….….. 40
AASHTO T166 Methods, 9.5 mm Granite…………………….. …….. 41
iv
AASHTO T166 Methods, Limestone 12.5 mm NMAS..………….……43
AASHTO T166 Methods, Granite 12.5 mm NMAS ……………….…..43
AASHTO T166 Methods, Limestone 19.0 mm NMAS……………….. 44
AASHTO T166 Methods, Granite 19.0 mm NMAS….……………… 44
AASHTO T166 Methods, Limestone 37.5 mm NMAS …………….… 46
AASHTO T166 Methods, Granite 37.5 mm NMAS ..………………… 46
Figure 22: Evaluation of Surface Texture (Granite, 9.5 mm NMAS, ARZ)……… 49
Figure 23: Evaluation of Surface Texture (Granite, 19.0 mm NMAS, SMA)….… 50
Figure 24: Comparisons of Measured and Estimated Surface Textures…..……… 53
Figure 25: Relationship Between Water Absorption and
Air Voids, ARZ Mixes……………..…………………………………. 55
Air Voids, BRZ Mixes ……………………………………………….. 56
Air Voids, TRZ Mixes ………………………………………………. 57
Air Voids, SMA Mixes ……………………………………..………… 58
v
Figure 29: Results of Time to Reach SSD Condition, ARZ Mixes………………. 61
Figure 30: Results of Time to Reach SSD Condition, BRZ Mixes………………. 62
Figure 31: Results of Time to Reach SSD Condition, TRZ Mixes………………. 62
Figure 32: Results of Time to Reach SSD Condition, SMA Mixes………………. 63
Figure 33: Comparison Between AASHTO T166 and
Vacuum-Sealing Methods, Field Projects………………….……….… 65
Figure 34: Comparison of Two Methods at Water Absorption
Levels Less than 0.5 Percent………………………………………..… 66
Figure 35: Histogram of Differences in Air Voids for Mixes
with Water Absorptions Less than 0.5 percent ………….………….… 68
Figure 36: Relationship Between Differences in Air Voids
and Water Absorption ………………………………………………… 69
vi
CONTROLLED LABORATORY EXPERIMENT TO EVALUATE METHODS
OF MEASURING THE BULK SPECIFIC GRAVITY OF COMPACTED HMA
NCHRP 9-27
TASK 3-PART 3
1.0 INTRODUCTION AND PROBLEM STATEMENT
A major concern of the hot mix asphalt (HMA) industry is the proper
measurement of bulk specific gravity (Gmb) of compacted samples. This issue has become
a bigger problem with the increased use of coarse gradations. Bulk specific gravity
measurements are the basis for volumetric calculations used during HMA mix design,
field control, and construction acceptance. During mix design, volumetric properties
such as air voids, voids in mineral aggregates, voids filled with asphalt, and percent
theoretical maximum density at a certain number of gyrations are used to evaluate the
acceptability of mixes. All of these properties are based upon Gmb.
In most states, acceptance of HMA construction by the owner is typically based
upon percent compaction calculated as a percent of theoretical maximum density.
Whether nondestructive (e.g., nuclear gauges) or destructive (e.g., cores) tests are used as
the basis of acceptance, Gmb measurements are equally important. When nondestructive
devices are utilized, each device has to first be calibrated to the Gmb of cores. If the Gmb
measurements of the cores are inaccurate in this calibration step, then the nondestructive
device will provide inaccurate data. Additionally, pay factors for construction, whether
reductions or bonuses, are generally applied to percent compaction. Thus, errors in Gmb
measurements can potentially affect both the agency and producer.
For many years, the measurement of Gmb for compacted HMA has been
accomplished by the water displacement method using saturated-surface dry (SSD)
samples. This method consists of first weighing a dry sample in air, then obtaining a
1
submerged mass after the sample has been placed in a water bath for a specified time
interval. Upon removal from the water bath, the SSD mass is determined after patting the
sample dry using a damp towel. Procedures for this test method are outlined in AASHTO
T166 and the equivalent ASTM D2726.
The SSD method has proved adequate for conventionally designed mixes, such as
Marshall and Hveem methods, that generally utilized fine-graded aggregates.
Historically, mixes were designed to have gradations passing close to or above the
Superpave defined maximum density line (i.e., fine-graded). However, since the
adoption of the Superpave mix design system and the increased use of stone matrix
asphalt (SMA), mixes are being designed with coarser gradations than used in the past.
The potential problem in measuring the Gmb of mixes like coarse-graded
Superpave and SMA using the SSD method comes from their internal air void structure.
These types of mixes tend to have larger internal air voids than the finer conventional
mixes at similar overall air void contents. Mixes with coarser gradations have a much
higher percentage of large aggregate particles. At a certain overall air void volume, which
is mix specific, the large internal air voids of the coarse mixes can become
interconnected. During Gmb testing with the SSD method, water can quickly infiltrate
into the sample through these interconnected voids. However, after removing the sample
from the water bath to obtain the saturated-surface dry condition the water can also drain
from the sample quickly. This draining of the water from the sample causes errors when
using the SSD method.
To understand the cause of potential errors, one must first understand the
principles of the SSD method. The philosophy of the SSD method is based upon
2
Archimedes’ Principle, which states that a force equal to the mass of the displaced fluid
buoys up a material immersed in fluid. Take for instance the material submerged in
water illustrated within Figure 1. The surface of the material that is in contact with water
can be divided into two halves: the upper surface (face BCE) and lower surface (face
BDE). Submerged in this manner, there are three forces acting on the material: 1) the
weight of the material in a dry condition (WM); 2) the force of the water within ABCEF
on the material (FD2); and 3) the force of the buoyant resistance acting upward (FU1)
which is equal to the weight of water within ABDEF.
A F
FD2
WM
B E
FU1
Figure 1. Hydrostatic Forces on a Submerged Material
Using these known forces acting on the material, a series of relationships can be
identified:
Total force acting downward = FD = WM + FD2 (1)
Total net force = FN = WM + FD2 – FU1 (2)
The net force acting downward (FN) on the sample can be determined by
measuring the weight of the material when it is submerged in water (WMW). Therefore,
the weight of the material submerged in water is equal to the right hand side of Equation
3
2. Further, the difference in the weight of the two water columns (FD2 and FU1) is equal
to the weight of fluid that is displaced when the material is submerged in water (WW).
[Note FU1 is greater than FD2, but acts in an opposite direction.] Hence:
WMW = WM - WW (3)
Now, using the properties shown in Equations 1 through 3 and the definition of
density and specific gravity, the equation using water displacement for calculating a
specific gravity can be derived. The definitions for density and specific gravity are as
follows:
γM = MM / VM (4)
Gs = γM / γW (5)
Where:
γ = the density of an object (γM for material and γW for water);
MM = the dry mass of a material; and
VM = the volume of the material.
Gs = specific gravity of a material.
Since the volume of the material is equal to the volume of the water displaced by
the material, substituting Equation 4 into Equation 5 yields the following:
Gs = MM / MW (6)
Where:
MW= mass of displaced water
4
The mass of a material is equal to the weight of that material divided by the
acceleration caused by gravity; therefore, Equations 3 and 6 can be used to derive the
equation used for determining the specific gravity of a material:
Gs = MM / (MM - MMW) (7)
Equation 7 is the method of determining the specific gravity of a material using
Archimedes’ Principle. However, within the context of HMA materials this equation
defines a "dry" specific gravity and not the bulk specific gravity. The term dry specific
gravity is used here to indicate the dry mass of the sample is utilized in the denominator
of Equation 7 to calculate specific gravity. A brief discussion of the differences between
the dry and bulk specific gravities of compacted HMA is necessary.
Figure 2 illustrates volumes and air voids that are associated with compacted
HMA. Each of the diagrams within Figure 2 are divided into halves with a given half
representing the volumes and air voids of mixes having coarse or fine gradations. The
dark black line in Figure 2a shows the volume that is associated with the specific gravity
measurements using the dimensional procedure. Dimensions (height and diameter) of the
sample are used to calculate the volume of the sample. Figure 2a illustrates the effect of
using this volume in determining the air void content of HMA. The volume includes any
surface irregularities (texture) on the outside of the sample and thus overestimates the
internal air void content. Of the three cases illustrated in Figure 2, the dimensional
volume is the highest, resulting in the lowest measured density.
Figure 2b illustrates the dry volume of compacted HMA samples. This volume is
identical to the one derived from Equation 7 above. Because Equation 7 utilizes the dry
mass in the volume determination (denominator of Equation 7), the calculated volume
5
does not include any of the surface irregularities on the sample or any internal air voids
that are interconnected to the surface. Water that infiltrates the surface irregularities or
internal voids interconnected to the surface are not considered a portion of the sample
volume and, thus, provides the smallest volume of the three cases shown in Figure 2.
Therefore, the dry volume underestimates the sample’s true internal voids by excluding
any voids interconnected to the surface. Figure 2b shows that this problem is more
prevalent with mixes having coarser gradations, as there are potentially more voids
interconnected to the surface of the sample.
Coarse Fine Coarse Fine Coarse Fine

Gradations Gradations Gradations Gradations Gradations Gradations
a) Dimensional Volume b) "Dry" Volume c) Bulk Volume
Figure 2. Volumes Associated with Compacted HMA
Figure 2c illustrates the bulk volume determined from the AASHTO T166
method. The difference between the bulk and dry volumes is that the bulk volume
includes internal voids that are interconnected to the surface. This is accomplished by
using the saturated-surface dry mass in the volume determination (replace MM in the
denominator of Equation 7 with the saturated-surface dry mass). The net result of using
the saturated-surface dry mass is that the voids that are interconnected to the surface and
6
do not lose their water within the saturated-surface dry condition are included as internal
voids. Therefore, the bulk volume lies between the dimensional and dry volumes.
This exercise of deriving the equation for measuring specific gravity using
Archimedes’ Principle and the discussion of the different volumes associated with
compacted HMA illustrates the potential deficiency of the SSD method for determining
bulk specific gravity of coarse-graded mixes. If the bulk volume is the desired property,
which it is for HMA, then mixes with coarser gradations have a higher potential for error,
as seen in Figure 2c. If a sample is submerged in water for a given time period (per
standard procedure), a certain volume of water is absorbed into the sample through voids
interconnected to the surface. For the coarse gradations shown in Figure 2c, this volume
of interconnected voids is higher than for the fine gradations (assuming both the coarse
and fine gradation mixes have the same total volume of air voids). Upon removal of the
sample from the water bath, any water draining from the large interconnected voids
within the coarse gradation mix leads to a lower saturated-surface dry mass. This, in
effect, decreases the volume of the sample and, thus, underestimates the air void content
of the sample. This is the potential drawback of the SSD method for determining the
bulk specific gravity of mixes having coarse gradations. The above discussion also
suggests that when the SSD method overestimates a specimen’s Gmb, the true Gmb should
be between the Gmb measured by the SSD method and the Gmb measured by the
dimensional method.
7
2.0 OBJECTIVES
Because of the potential errors noted with the saturated surface-dry test method of
determining the bulk specific gravity of compacted HMA, the objectives of this task
were: (1) compare AASHTO T166 with other methods of measuring bulk specific gravity
to determine under what conditions AASHTO T166 is accurate; (2) if conditions are
identified that AASHTO T166 is not applicable, identify potential improvements to
AASHTO T166 to achieve a more accurate measure of bulk specific gravity; and (3) if
improvements are not successful, recommend alternate methods of measuring bulk
specific gravity. Results of Task 3, Part 3 are to be applicable to both laboratory
compacted samples and samples taken from the roadway (cores).
3.0 SCOPE
To accomplish the objectives, two separate sample types were evaluated:
laboratory compacted and field compacted. Laboratory compacted mixtures having
various aggregate types, nominal maximum aggregate sizes, gradation shapes, and air
void levels were prepared. Each of the prepared samples were tested to determine bulk
specific gravity by four different test methods: water displacement (AASHTO T166),
vacuum-sealing, gamma ray, and dimensional. Also, on a selected number of samples,
the time taken to achieve SSD condition for a given sample was altered. This testing was
conducted to determine the water absorption level where the potential for errors with
SSD method were minimized.
For the field compacted samples, cores obtained during the field validation
portion of this study were subjected to the same four bulk specific gravity test methods.
8
Because cores have a different surface texture than laboratory compacted samples, it was
necessary to also evaluate them. Testing conducted on core samples included laboratory
permeability tests and effective air void content using the vacuum-sealing device.
4.0 RESEARCH APPROACH
The overall proposed test plan is illustrated in Figure 3. This figure shows that
two separate sample types were included in the experiment. Both laboratory (Superpave
gyratory compactor) and field (cores) compacted samples were included because each
sample type has different surface texture properties. For both sample types, four bulk
specific gravity measurements were made on all samples: water displacement (AASHTO
T166), vacuum-sealing (ASTM 6752-02a), gamma ray, and dimensional analysis. The
following paragraphs detail the research approach for both sample types.
9
Part 3, Task 3
Methods for Gmb Evaluation:

AASHTO T166 Gamma Ray Technology
Dimensional Analysis Vacuum-Sealing
Laboratory Compacted Field Compacted

Samples Samples (Cores)
Four Gradation Shapes: Three NMAS (mm):

Above Restricted Zone Superpave SMA Lab Permeability
Through Restricted Zone 9.5 9.5 And
Below Restricted Zone 19.0 12.5 Effective Air Voids
Stone Matrix Asphalt 37.5 19.0
Three SGC Levels:

1. 15 Gyrations (high voids) Two Aggregate Types:
2. 50 Gyrations (medium voids) Limestone
3. 125 gyrations (low voids) Granite
Selected Samples,
Vary Time Needed To Achieve
Saturated Surface-Dry Condition
During T166
Perform Gmb Measurements with

Each Selected Method
Analyze Data to Determine Method(s)

That Precisely and Accurately
Determine Gmb
Recommend Improvements
To AASHTO T166 or
Provide New Method
In AASHTO format
Figure 3: Research Approach for Part 3 of Task 3
4.1 Laboratory Compacted Samples
For the laboratory compacted samples, two aggregate types (limestone and
granite) were used to prepare compacted samples comprised of four gradations (above,
through, and below the restricted zone and SMA) at each of three nominal maximum
aggregate sizes. These mixes were designed during Part 1 of Task 3 of this study. The 24
10
combinations of aggregate type/gradation shape/NMAS were selected because they
should provide different water absorption characteristics during AASHTO T166 testing
and different surface textures. Similar to Part 1, NMASs of 9.5, 19.0, and 37.5 mm were
used for the Superpave designed mixes and NMASs of 9.5, 12.5, and 19.0 mm were used
for the SMA mixes. At optimum asphalt content, samples were prepared using the
Superpave gyratory compactor to 15, 50, and 125 gyrations to produce low, medium, and
high air void contents (ranging from below 4 to approximately 12 percent) for all the mix
types. Triplicate samples were prepared for each combination. To try and minimize
variability in the production of these samples, a single person batched and fabricated all
samples. Also, a single operator conducted all tests (bulk specific gravity measures). A
single operator was used so that variability would be reduced and to allow for comparison
of within-laboratory test method variabilities for the different methods utilized. This
resulted in a total of 216 samples for the experiment on laboratory compacted samples.
A key component of the AASHTO T166 standard method is that the procedure is
only applicable for compacted HMA having less than 2 percent water absorption by
volume. As discussed previously, AASHTO T166 was originally intended for use on
conventionally designed mixes (i.e., fine-graded). Therefore, a side experiment for the
laboratory compacted samples was to evaluate the effect of time on water absorption, and
thus Gmb. For this side experiment, 40 samples having a range of water absorptions (by
volume) were selected from the 216 total samples. The selection of samples for various
gradation shapes, NMAS, and gyration levels were also included in the experiment. Steps
included in testing each sample within this side experiment included:
1. Obtain dry mass of sample.
11
2. Submerge sample in water bath for 10 minutes to ensure saturation of internal
air voids interconnected to the surface of the sample.
3. Remove sample from water bath and let stand for two minutes.
4. Submerge sample in water bath for 4±1 minutes as per the AASHTO T166
protocol.
5. Remove sample from water bath and obtain SSD mass as fast as possible.
Record SSD mass and time needed to obtain SSD condition (this occurred in
six seconds).
6. Submerge sample in water bath again for 4±1 minutes as per the AASHTO
T166 protocol.
7. Remove sample from water bath and obtain SSD mass after 10 seconds.
Record SSD mass.
8. Follow steps 6 and 7 to obtain SSD masses after 15, 20, 30, and 60 seconds.
This testing was conducted to evaluate the sensitivity of testing time on the
measured water absorption. Samples having a large proportion of interconnected voids to
the surface of the sample should show decreasing water absorptions as time to achieve
SSD increases. This also would correspond to an increased Gmb. Results of this side
experiment would provide a water absorption value that leads to increased potential for
errors during AASHTO T166 testing.
Analysis was conducted on the overall data set (216 samples) to determine which
method(s) provided the most precise and accurate measure of the bulk specific gravity for
compacted samples, regardless of the nominal maximum size, aggregate type, and air
void level. If needed, changes to the current AASHTO T166 will be recommended.
12
4.2 Field Compacted Samples
Each of the cores obtained during the Task 5 field validation were tested to
determine bulk specific gravity using the same four tests as the laboratory experiment:
water displacement (AASHTO T166), vacuum sealing, gamma ray, and dimensional
analysis. Because of the differences in surface texture between laboratory compacted
samples (surface texture around entire sample) and field compacted samples (surface
texture only on top of sample because of core bit and sawing), the experiment was also
extended to core samples.
5.0 TEST METHODS AND MATERIALS
This section describes the different test methods and materials used in Task 3.
The test methods included determining the bulk specific gravity of samples using water
displacement (AASHTO T166), vacuum sealing, gamma ray, and dimensional analysis.
Materials included some mixes designed during Part 1 of Task 3 and the cores obtained
during the field validation of Task 5.
5.1 Test Methods
5.1.1 Saturated Surface-Dry Method (AASHTO T166)
The saturated surface-dry method consists of first weighing a dry sample in air,
then obtaining a submerged mass after the sample has been placed in a water bath for a
specified time interval (4±1 minutes). Upon removal from the water bath, the SSD mass
is determined after patting the sample dry using a damp towel. Procedures for this test
method are outlined in AASHTO T166 and ASTM D2726.
13
5.1.2 Vacuum-Sealing Test Method
Buchanan (1) recently reported on a comparison between the vacuum-sealing device
and other more conventional Gmb methods that included: SSD, parafilm, and dimensional
methods. This comparison indicated that the vacuum-sealing method appeared to be able
to determine Gmb with greater accuracy than the conventional methods when samples
were at low densities (i.e., high air voids). This vacuum-sealing device utilizes an
automatic vacuum chamber (shown in Figure 4a) with a specially designed plastic bag,
which tightly conforms to the sides of the sample (shown in Figure 4b) and prevents
water from infiltrating into the sample.
Figure 4a. Vacuum-Sealing Device Figure 4b. Sealed Sample
The steps involved in sealing and analyzing compacted HMA samples are as follows
(2):
Step 1: Determine the density of the plastic bag (generally provided by the
manufacturer).
14
Step 2: Place the compacted HMA sample into the bag.
Step 3: Place the bag containing the HMA sample inside the vacuum chamber.
Step 4: Close the vacuum chamber door. The vacuum pump starts automatically and
evacuates the chamber.
Step 5: In approximately two minutes, the chamber door will automatically open
with the sample completely sealed within the plastic bag and ready for water
displacement testing.
Step 6: Perform SSD method without obtaining SSD mass. Correct the results for
the bag density and the displaced bag volume.
In addition to Buchanan (1), Hall et al (3) and Cooley et al (4) have also indicated
that the vacuum-sealing method is a viable option for determining the Gmb of compacted
HMA. Hall et al indicated that the within-lab (operator) variability for the vacuum-
sealing method was less than the SSD method. Based on two separate round-robin
studies, Cooley et al (5) and Spellerberg et al (6) both suggested that the vacuum-sealing
method was slightly more variable (both within- and between-laboratory) than AASHTO
T166; however, both round-robin studies noted that a portion of the participating
laboratories had little experience with the equipment and test procedure. A standard
ASTM test method has been developed for the vacuum-sealing test method, ASTM 6752-
02a, “Bulk Specific Gravity and Density of Compacted Bituminous Mixtures Using
Automatic Vacuum Sealing Method.”
15
5.1.3 Gamma Ray Method
A relatively new method of determining the bulk specific gravity of HMA
includes the use of gamma ray technology. This method is based upon the scattering and
absorption characteristics of gamma rays within a material. The instrument works in
transmission mode which means that a sample is placed between the source of gamma
rays and the gamma ray detector. During the test, the device counts the gamma rays that
travel through the sample in order to determine the sample’s volume. Figure 5 illustrates
the device used in this study.
Figure 5: Equipment for Gamma Ray Test
The steps involved in determining the bulk specific gravity of HMA samples
using the gamma ray method include (7):
1. Using calibrated vernier calipers, measure the height of the specimen (in
millimeters) in six locations and determine the average height.
2. Place the sample on the sample tray, return the tray to its home position and
close the door.
16
3. Initiate the program.
4. Input the average sample height determined in step 1.
5. The equipment uses a four-minute count and then displays the measured bulk
specific gravity.
6. Remove sample from the device.
Malpass and Khosla (8) reported on testing conducted at North Carolina State
University to evaluate the gamma ray method for measuring the bulk specific gravity of
HMA. Based upon their work, it was shown that the gamma ray and AASHTO T166
methods provided practically similar bulk specific gravity values at low to medium air
void contents. However, at high air void contents, the gamma ray method provided lower
values of bulk specific gravity.
5.1.4 Dimensional Method
The dimensional method of determining the bulk specific gravity included height
and diameter measurements for each sample to calculate the volume of samples. The dry
mass of a sample was divided by the calculated sample volume to estimate the bulk
specific gravity of the sample.
5.2 Materials
As stated previously, 24 mixes from Task 3, Part 1 were included within this
experiment. Also, results from the 20 field projects visited as part of Task 5 were
included within this analysis. The following sections provide information on the
materials used/encountered.
17
5.2.1 Task 3, Part 1 Laboratory Prepared Materials
Properties of the coarse and fine aggregates from Part 1 are shown in Table 1.
The aggregates were selected because they represented a range of physical properties,
such as absorption (0.3 to 0.9 percent), Los Angeles Abrasion (31 to 37 percent), and fine
aggregate angularity (44.6 to 48.2 percent), and should provide some variability of mix
properties.
Table 2 presents the test results for the asphalt cement utilized in the study. The
binder was graded as PG 64-22 (meeting high temperature requirements above 67C) and
is commonly used for warm climates.
Table 1: Physical Properties of Aggregate

Aggregate Type
Property Test Method
Granite Limestone
Coarse Aggregate
Bulk Specific Gravity AASHTO T-85 2.654 2.725
Apparent Specific Gravity AASHTO T-85 2.704 2.758
Absorption (%) AASHTO T-85 0.7 0.4
Flat and 19.0 mm 14, 0 10, 0

Elongated (%), 12.5 mm ASTM D4791 16, 0 6, 0
3:1, 5:1 9.0 mm 9, 1 16, 3
Los Angeles Abrasion (%) AASHTO T-96 37 35
Percent Crushed (%) 100 100
Fine Aggregate
Bulk Specific Gravity AASHTO T-84 2.678 2.689
Apparent Specific Gravity AASHTO T-84 2.700 2.752
Absorption (%) AASHTO T-84 0.3 0.9
Fine Aggregate AASHTO T-33
45.8 44.6
Angularity (%) (Method A)
Sand Equivalency (%) AASHTO T-176 92 93
18
Table 2 Asphalt Binder Properties
Original RTFOT RTFOT+ PAV residue
DT
Dynamic Shear Dyn. Shear Dynamic Shear Flexural Creep
1mm/mi
10 rad/s 10 rad/s 10 rad/s (at 60 sec)
n
G*/sin d G*/sin d G*/sin d Slope, m Strain
Temp Temp. Temp. Stiffness, S
(kPa) 1.0 kPa (kPa) 1.0 kPa (kPa) 5000 0.30 1.0%
(oC) (oC) (oC) 300 Mpa max
min. min. kPa max. min. min
67 1.078 2.279 25 4752 -12 226 0.325 NA
NA-Results not available
Part 1 included four gradation shapes and three nominal maximum aggregate
sizes (NMAS). Three gradation shapes fell within Superpave gradation control points and
one gradation conformed to SMA specifications. The general mix gradations used are
illustrated in Figures 6 through 9.
Task 3 ~ 9.5 mm NMAS Superpave Gradations

Control Points Resricted Zone BRZ ARZ TRZ
100
90
80
70
Percent Passing, %
60
50
40
30
20
10
0
0.075 0.30 0.60 1.18 2.36 4.75 9.5 12.5
Sieve Size, mm
Figure 6: 9.5 mm NMAS Superpave gradations
19
T a s k 3 ~ 1 9 .0 m m N M A S S u p e rp a v e G ra d a tio n s
C o n tro l P o in ts R e s ric te d Z o n e BRZ ARZ TRZ
100
90
80
70
Percent Passing, %
60
50
40
30
20
10
0
0 .0 7 5 0 .3 0 0 .6 0 1 .1 8 2 .3 6 4 .7 5 9 .5 1 2 .5 1 9 .0 2 5 .0
S ie v e S iz e , m m

Task 3 ~ 37.5 mm NMAS Superpave Gradations
Control Points Resricted Zone BRZ ARZ TRZ
100
90
80
70
Percent Passing, %
60
50
40
30
20
10
0
0.075 0.60 1.18 2.36 4.75 9.5 12.5 19.0 25.0 37.5 50.0
Sieve Size, mm
20
Task 3 ~ SM A G radations
Control Points 9.5 m m 12.5 m m 19.0 m m
100
90
80
70
Percent Passing, %
60
50
40
30
20
10
0
0.075 0.30 0.60 1.18 2.36 4.75 9.5 12.5 19.0 25.0
Sieve Size, m m
Figure 9: SMA gradations
Of the 24 mix designs utilized in this experiment, 18 were Superpave mixes and 6
were SMA mixes. For the Superpave mixes, each sample was designed to 100 gyrations
in the Superpave gyratory compactor (SGC). The 100-gyration level was selected
because it covers the widest range of traffic categories in the Ndesign chart within the
AASHTO PP28-01, “Standard Practice for Superpave Volumetric Design for Hot–Mix
Asphalt (HMA).” For the SMA mixes, each sample was designed at 75 gyrations in the
SGC based on “Standard Practice for Designing SMA”, AASHTO PP 44-01. The reason
for using 75 gyrations was that both aggregate types had Los Angeles Abrasion values
above 30 percent. Designs for both mix types were conducted to determine the asphalt
binder content necessary to produce 4.0 percent air voids at the design number of
gyrations. A summary of the mix designs for the Superpave and SMA mixes is presented
in Table 3 and 4, respectively.
21
Table 3: Summary of Mix Design Information for Superpave Mixes
Opt. Binder Eff. Binder % Gmm
Agg. NMAS Gradation Content, % Content, % VMA VFA @Nini D/AC
9.5 ARZ 6.7 6.2 18.4 76 89.0 0.80
9.5 BRZ 5.3 4.9 15.7 72 86.7 1.02
9.5 TRZ 5.4 5.0 15.6 75 88.9 1.00
19 ARZ 4.7 4.3 14.1 72 89.5 1.17
Granite 19 BRZ 4.4 3.9 13.3 68 86.0 1.00
19 TRZ 4.0 3.6 12.5 68 88.8 1.40
37.5 ARZ 4.2 4.0 13.7 69 89.8 0.75
37.5 BRZ 3.3 3.0 11.3 64 86.8 1.00
37.5 TRZ 3.6 3.3 12.0 65 88.1 0.90
9.5 ARZ 6.0 5.7 17.4 76 87.8 0.70
9.5 BRZ 5.0 4.6 15.3 72 85.5 0.86
9.5 TRZ 4.4 4.2 14.4 70 86.0 1.18
19 ARZ 4.1 3.5 12.6 66 88.3 1.42
Limestone 19 BRZ 4.7 4.4 14.3 71 85.5 0.68
19 TRZ 3.3 2.8 11.0 62 85.7 1.80
37.5 ARZ 3.2 3.1 11.8 64 88.8 0.95
37.5 BRZ 2.7 2.6 10.6 60 86.0 1.15
37.5 TRZ 2.8 2.6 10.6 61 87.7 1.12
Table 4: Summary of Mix Design Information for SMA Mixes

Opt. Binder Eff. Binder
Agg. NMAS Gradation Content, % Content, % VMA VFA VCAmix VCAdrc
9.5 SMA 7.2 6.56 18.7 78 30.9 41.9
Granite 12.5 SMA 6.6 6.42 18.8 77 30.3 42.7
19 SMA 6.4 5.91 17.6 77 29.6 42.0
9.5 SMA 6.2 5.76 17.4 76 30.7 38.4
Limestone 12.5 SMA 7.4 6.97 19.6 80 31.1 38.9
19 SMA 6.0 5.59 16.8 77 29.8 40.3
5.2.2 Task 5 Field Projects

A total of 20 field projects were visited as part of the Task 5 field validation
study. Table 5 provides information on each of the 20 projects. This table shows that six
of the projects had a design NMAS of 9.5 mm. Four of these 9.5 mm NMAS mixes had
fine gradations and the other two were coarse-graded. Six projects were designed as 12.5
mm NMAS gradations. Three of these six projects utilized coarse-graded gradation, two
22
were fine-graded, and the sixth project utilized a SMA gradation. A total of six projects
were designed to have 19.0 mm NMAS gradations. Four of the six projects utilized
coarse gradations, while there was one fine-graded mix and one SMA. The remaining
two projects were designed to have 25.0 mm NMAS gradations, both of which were
coarse-graded.
Table 5: Summary Information on Field Projects From Task 5

Project NMAS Fine or Average Lift Lift AC Ndesign
ID Coarse Thickness Thickness/ Performance
Gradation (mm) NMAS Ratio Grade
1 9.5 Fine 38.1 4:1 70-22 65
2 19.0 Coarse 63.5 4:1 64-22 65
3 9.5 Coarse 38.1 4:1 64-22 65
4 9.5 Fine 68.6 5:1 NA 75
5 9.5 Fine 31.8 3:1 70-22 100
6 12.5 Coarse 57.2 4:1 58-28 75
7 12.5 Fine 50.8 4:1 64-28 75
8 19.0 Coarse 50.8 3:1 64-22 100
9 25.0 Coarse 101.6 4:1 64-22 100
10 19.0 Coarse 57.2 3:1 64-34 100
11 19.0 Coarse 38.1 2:1 64-34 125
12 19.0 SMA 61.0 3:1 76-22 50
13 25.0 Coarse 70.0 3:1 67-22 100
14 12.5 SMA 26.8 3:1 76-22 75
15 19.0 Fine 50.4 3:1 76-22 100
16 9.5 Fine 43.8 4:1 67-22 86
17 12.5 Fine 50.8 3:1 64-22 75
18 12.5 Coarse 38.1 3:1 67-22 75
19 9.5 Coarse 31.8 3:1 67-22 75
20 12.5 Coarse 38.1 3:1 67-22 80
6.0 TEST RESULTS AND ANALYSIS
Within this section, test results and analyses are provided for the experiments to
recommend improvements to AASHTO T166. Because of the potential differences in
conclusions between laboratory and field compacted samples, this section is divided into
two primary subsections that describe test results and analyses for each sample type.
23
6.1 Laboratory Compacted Samples
Results of bulk specific gravity measurements on the 24 laboratory mixes at the
three gyration levels are presented in Tables 6 through 13. These tables include results of
bulk specific gravity measurements using the four test methods: water displacement,
vacuum sealing, gamma ray, and dimensional analysis.
Table 6: Results of Part 3 Testing for 9.5 mm NMAS Limestone Mixes

Bulk Specific Gravity, Gmb
Gradation Gyration Replicate Dimensional Gamma Ray Vacuum-Seal AASHTO T166
15 1 2.238 2.227 2.268 2.267
15 2 2.234 2.226 2.265 2.266
15 3 2.235 2.232 2.259 2.263
50 1 2.329 2.341 2.354 2.356
ARZ 50 2 2.333 2.333 2.348 2.354
50 3 2.332 2.333 2.371 2.357
125 1 2.390 2.390 2.405 2.411
125 2 2.391 2.393 2.408 2.413
125 3 2.383 2.392 2.409 2.407
15 1 2.168 2.189 2.217 2.245
15 2 2.166 2.192 2.212 2.254
15 3 2.188 2.204 2.227 2.255
50 1 2.306 2.334 2.343 2.362
BRZ 50 2 2.311 2.344 2.336 2.356
50 3 2.298 2.354 2.343 2.363
125 1 2.383 2.417 2.420 2.431
125 2 2.391 2.422 2.430 2.443
125 3 2.389 2.423 2.421 2.429
15 1 2.225 2.258 2.267 2.292
15 2 2.240 2.280 2.270 2.296
15 3 2.242 2.269 2.279 2.300
50 1 2.369 2.402 2.405 2.416
TRZ 50 2 2.362 2.400 2.405 2.414
50 3 2.367 2.397 2.396 2.409
125 1 2.440 2.461 2.465 2.469
125 2 2.434 2.463 2.465 2.470
125 3 2.445 2.472 2.466 2.473
24
Table 7: Results of Part 3 Testing for 9.5 mm NMAS Granite Mixes
Gradation Gyration Replicate Dimensional Gamma Ray Vacuum Seal AASHTO T166
15 1 2.192 2.181 2.221 2.224
15 2 2.189 2.205 2.218 2.218
15 3 2.192 2.180 2.226 2.222
50 1 2.271 2.281 2.303 2.302
ARZ 50 2 2.268 2.273 2.304 2.300
50 3 2.280 2.279 2.299 2.297
125 1 2.316 2.327 2.345 2.340
125 2 2.315 2.328 2.338 2.345
125 3 2.323 2.326 2.351 2.348
15 1 2.157 2.166 2.198 2.231
15 2 2.169 2.206 2.210 2.236
15 3 2.158 2.208 2.200 2.229
50 1 2.258 2.286 2.302 2.318
BRZ 50 2 2.269 2.305 2.304 2.322
50 3 2.293 2.318 2.309 2.322
125 1 2.348 2.381 2.355 2.393
125 2 2.342 2.367 2.381 2.390
125 3 2.350 2.365 2.389 2.396
15 1 2.207 2.206 2.241 2.255
15 2 2.212 2.206 2.247 2.255
15 3 2.199 2.226 2.235 2.249
50 1 2.302 2.313 2.322 2.330
TRZ 50 2 2.283 2.301 2.316 2.324
50 3 2.284 2.300 2.324 2.329
125 1 2.342 2.359 2.374 2.375
125 2 2.337 2.361 2.366 2.371
125 3 2.348 2.370 2.375 2.379
25
Table 8: Results of Part 3 Testing for 19.0mm NMAS Limestone Mixes
15 1 2.299 2.326 2.335 2.346
15 2 2.296 2.314 2.335 2.345
15 3 2.291 2.306 2.329 2.331
50 1 2.377 2.400 2.413 2.418
ARZ 50 2 2.369 2.394 2.405 2.410
50 3 2.370 2.407 2.409 2.414
125 1 2.449 2.475 2.480 2.480
125 2 2.452 2.490 2.475 2.476
125 3 2.461 2.481 2.483 2.483
15 1 2.187 2.211 2.247 2.297
15 2 2.191 2.203 2.223 2.289
15 3 2.188 2.221 2.229 2.304
50 1 2.319 2.358 2.368 2.393
BRZ 50 2 2.332 2.375 2.380 2.409
50 3 2.331 2.382 2.382 2.413
125 1 2.412 2.456 2.449 2.474
125 2 2.403 2.440 2.447 2.472
125 3 2.432 2.449 2.457 2.469
15 1 2.273 2.391 2.317 2.351
15 2 2.270 2.372 2.321 2.362
15 3 2.267 2.354 2.322 2.363
50 1 2.373 2.423 2.424 2.449
TRZ 50 2 2.373 2.439 2.423 2.456
50 3 2.367 2.430 2.418 2.453
125 1 2.457 2.507 2.497 2.517
125 2 2.463 2.514 2.504 2.516
125 3 2.490 2.526 2.514 2.522
26
Table 9: Results of Part 3 Testing on 19.0 mm NMAS Granite Mixes
15 1 2.230 2.263 2.280 2.298
15 2 2.252 2.275 2.289 2.304
15 3 2.252 2.257 2.285 2.300
50 1 2.330 2.338 2.367 2.370
ARZ 50 2 2.308 2.336 2.357 2.368
50 3 2.321 2.346 2.332 2.366
125 1 2.370 2.396 2.407 2.410
125 2 2.384 2.399 2.397 2.410
125 3 2.364 2.383 2.378 2.407
15 1 2.167 2.223 2.244 2.302
15 2 2.179 2.223 2.236 2.322
15 3 2.191 2.251 2.243 2.320
50 1 2.296 2.372 2.356 2.397
BRZ 50 2 2.290 2.349 2.352 2.388
50 3 2.298 2.357 2.356 2.388
125 1 2.360 2.444 2.415 2.442
125 2 2.362 2.404 2.418 2.441
125 3 2.368 2.415 2.431 2.456
15 1 2.211 2.269 2.285 2.311
15 2 2.230 2.250 2.279 2.283
15 3 2.222 2.297 2.274 2.298
50 1 2.328 2.357 2.381 2.395
TRZ 50 2 2.332 2.359 2.374 2.389
50 3 2.341 2.366 2.374 2.392
125 1 2.379 2.427 2.415 2.421
125 2 2.384 2.419 2.420 2.426
125 3 2.365 2.408 2.414 2.425
27
Table 10: Results of Part 3 Testing for Limestone SMA Mixes
NMAS Gyration Replicate Dimensional Gamma Ray Vacuum-Seal AASHTO T166
15 1 2.123 2.169 2.183 2.251
15 2 2.124 2.194 2.186 2.251
15 3 2.127 2.176 2.166 2.244
50 1 2.281 2.351 2.330 2.355
9.5 mm 50 2 2.263 2.329 2.314 2.342
50 3 2.266 2.307 2.333 2.359
125 1 2.418 2.458 2.444 2.466
125 2 2.412 2.464 2.437 2.465
125 3 2.404 2.433 2.429 2.456
15 1 2.105 2.160 2.179 2.237
15 2 2.113 2.129 2.163 2.232
15 3 2.105 2.119 2.181 2.244
50 1 2.231 2.288 2.291 2.320
12.5 mm 50 2 2.223 2.303 2.278 2.320
50 3 2.235 2.312 2.306 2.331
125 1 2.303 2.349 2.363 2.389
125 2 2.327 2.381 2.368 2.393
125 3 2.317 2.358 2.380 2.398
15 1 2.066 2.096 2.162 2.306
15 2 2.069 2.093 2.172 2.296
15 3 2.050 2.143 2.159 2.293
50 1 2.260 2.382 2.323 2.382
19.0 mm 50 2 2.255 2.382 2.343 2.388
50 3 2.291 2.364 2.376 2.405
125 1 2.384 2.474 2.430 2.457
125 2 2.359 2.501 2.421 2.450
125 3 2.368 2.491 2.432 2.447
28
Table 11: Results of Part 3 Testing for Granite SMA Mixes
NMAS Gyration Replicate Dimensional Gamma Ray Vacuum-Seal AASHTO T166
15 1 2.036 2.119 2.109 2.207
15 2 2.045 2.126 2.115 2.192
15 3 2.055 2.160 2.119 2.191
50 1 2.192 2.313 2.258 2.277
9.5 mm 50 2 2.209 2.323 2.268 2.296
50 3 2.206 2.304 2.271 2.295
125 1 2.282 2.387 2.337 2.355
125 2 2.280 2.385 2.339 2.355
125 3 2.284 2.378 2.337 2.357
15 1 2.041 2.056 2.125 2.259
15 2 2.016 2.045 2.097 2.224
15 3 2.016 2.147 2.110 2.258
50 1 2.167 2.240 2.250 2.311
12.5 mm 50 2 2.179 2.288 2.258 2.310
50 3 2.180 2.309 2.266 2.287
125 1 2.274 2.394 2.333 2.359
125 2 2.276 2.334 2.344 2.372
125 3 2.276 2.388 2.337 2.360
15 1 2.010 2.108 2.141 2.269
15 2 2.064 2.104 2.164 2.284
15 3 2.003 2.106 2.102 2.262
50 1 2.278 2.350 2.380 2.430
19.0 mm 50 2 2.210 2.228 2.308 2.366
50 3 2.156 2.244 2.254 2.336
125 1 2.289 2.360 2.369 2.400
125 2 2.301 2.350 2.368 2.400
125 3 2.278 2.328 2.354 2.386
29
Table 12: Results of Part 3 Testing on 37.5 mm NMAS Limestone Mixes
15 1 2.252 2.290 2.325 2.373
15 2 2.215 2.286 2.308 2.349
15 3 2.299 2.353 2.313 2.381
50 1 2.388 2.446 2.434 2.453
ARZ 50 2 2.365 2.467 2.427 2.448
50 3 2.383 2.451 2.417 2.448
125 1 2.418 2.471 2.460 2.493
125 2 2.431 2.471 2.484 2.498
125 3 2.457 2.505 2.492 2.504
15 1 2.182 2.218 2.268 2.443
15 2 2.192 2.234 2.265 2.442
15 3 2.201 2.268 2.256 2.441
50 1 2.290 2.393 2.423 2.497
BRZ 50 2 2.276 2.389 2.406 2.495
50 3 2.311 2.390 2.390 2.456
125 1 2.421 2.473 2.485 2.521
125 2 2.413 2.489 2.471 2.522
125 3 2.401 2.505 2.475 2.552
15 1 2.193 2.285 2.296 2.366
15 2 2.187 2.294 2.294 2.381
15 3 2.250 2.307 2.320 2.387
50 1 2.322 2.435 2.421 2.466
TRZ 50 2 2.364 2.439 2.427 2.462
50 3 2.360 2.487 2.409 2.491
125 1 2.428 2.481 2.499 2.532
125 2 2.422 2.490 2.470 2.528
125 3 2.422 2.493 2.488 2.523
30
Table 13: Results of Part 3 Testing on 37.5 mm NMAS Granite Mixes
15 1 2.226 2.265 2.283 2.319
15 2 2.224 2.282 2.269 2.304
15 3 2.204 2.324 2.270 2.306
50 1 2.318 2.551 2.361 2.368
ARZ 50 2 2.315 2.547 2.349 2.368
50 3 2.338 2.585 2.356 2.379
125 1 2.374 2.501 2.413 2.427
125 2 2.365 2.501 2.401 2.421
125 3 2.385 2.442 2.410 2.425
15 1 2.066 2.109 2.219 2.390
15 2 2.087 2.155 2.194 2.378
15 3 2.108 2.175 2.218 2.382
50 1 2.271 2.457 2.340 2.399
BRZ 50 2 2.228 2.360 2.338 2.424
50 3 2.232 2.437 2.337 2.408
125 1 2.376 2.462 2.435 2.464
125 2 2.332 2.456 2.409 2.458
125 3 2.330 2.459 2.421 2.464
15 1 2.195 2.320 2.293 2.361
15 2 2.212 2.329 2.340 2.410
15 3 2.260 2.407 2.295 2.403
50 1 2.292 2.436 2.361 2.392
TRZ 50 2 2.236 2.492 2.377 2.405
50 3 2.297 2.461 2.376 2.413
125 1 2.342 2.488 2.381 2.447
125 2 2.349 2.484 2.404 2.448
125 3 2.360 2.475 2.365 2.450
Bulk specific gravity values shown in Tables 6 through 13 were converted to air
void contents for some analyses. This was done because each of the aggregate/gradation
combinations had different bulk specific gravities for the aggregate. An initial analysis of
variance (ANOVA) was conducted on the air voids data to determine whether the
different methods yielded different bulk specific gravity measurements on similar mixes.
If all four methods resulted in similar air void contents, then no modifications to
AASHTO T166 would be required. However, if the different methods produced
31
significantly different air void contents then the next course of action would be to
determine which method provided a better overall estimate of bulk specific gravity. If
AASHTO T166 was not the most accurate method over all values of NMAS, gradation
shapes and air void contents, then analyses were needed to determine when AASHTO
T166 was applicable, identify potential improvements to AASHTO T166 to achieve a
more accurate measure of bulk specific gravity, or recommend alternate methods of
measuring bulk specific gravity.
Two separate ANOVAs were conducted: one for the Superpave mixes and one for
the SMA mixes. The data was separated in this manner because combining the two data
sets into a single data set would result in an unbalanced experimental design. The
Superpave and SMA data sets contained different NMAS.
Results of the ANOVA conducted on the Superpave mix data are presented in
Table 14. All five main factors (gradation shape, aggregate type, NMAS, gyration level,
and method) significantly affected air void contents. There were also a large number of
two- and three-way interactions that were significant. Based upon the F-statistics, the
gyration level (compactive effort) had the greatest effect on resulting air void contents.
The next most significant factor was bulk specific gravity method.
Because of the differences in resulting air voids for the four methods of
measuring bulk specific gravity, a Duncan’s multiple range test (DMRT) was conducted
to determine which methods, if any, provided similar results. This analysis method
provides a ranking comparison between the different methods. The range of sample
means for a given set of data (method) can be compared to a critical valued based on the
32
Table 14: Results of ANOVA Conducted on Superpave Designed Mixes
Source DF Mean FStatistic FCritical Significant
Squares (α=0.05)?
Aggregate Type (Agg.) 1 7.45 26.21 3.84 Yes
Nominal Max. Agg. Size (NMAS) 2 39.20 137.92 3.00 Yes
Gradation Shape (Grad.) 2 132.28 465.33 3.00 Yes
Gyration Level (Gyr.) 2 2304.71 8107.76 3.00 Yes
Bulk Specific Gravity Method (Meth.) 3 282.49 993.76 2.60 Yes
Agg*NMAS 2 10.95 38.52 3.00 Yes
Agg.*Grad 2 0.89 3.15 3.00 Yes
Agg*Gyr 2 18.28 64.32 3.00 Yes
Agg*Meth 3 5.37 18.88 2.60 Yes
NMAS*Grad 4 7.92 27.86 2.37 Yes
NMAS*Gyr 4 2.22 7.81 2.37 Yes
NMAS*Meth 6 39.81 140.04 2.10 Yes
Grad*Gyr 4 21.28 74.88 2.37 Yes
Grad*Meth 6 19.13 67.31 2.10 Yes
Gyr*Meth 6 15.75 55.42 2.10 Yes
Agg*NMAS*Grad 4 11.38 40.04 2.37 Yes
Agg*NMAS*Gyr 4 1.31 4.62 2.37 Yes
Agg*NMAS*Meth 6 8.85 31.14 2.10 Yes
Agg*Grad*Gyr 4 3.66 12.89 2.37 Yes
Agg*Grad*Meth 6 0.63 2.20 2.10 Yes
Agg*Gyr*Meth 6 1.52 5.35 2.10 Yes
NMAS*Grad*Gyr 8 2.68 9.44 1.94 Yes
NMAS*Grad*Meth 12 5.80 20.39 1.75 Yes
NMAS*Gyr*Meth 12 6.02 21.16 1.75 Yes
Grad*Gyr*Meth 12 2.44 8.57 1.75 Yes
Agg*NMAS*Grad*Gyr 8 3.73 13.11 1.94 Yes
Agg*NMAS*Grad*Meth 12 1.05 3.70 1.75 Yes
Agg*NMAS*Gyr*Meth 12 1.48 5.20 1.75 Yes
Agg*Grad*Gyr*Meth 12 0.36 1.28 1.75 No
NMAS*Grad*Gyr*Meth 24 0.88 3.08 1.52 Yes
Agg*NMAS*Grad*Gyr*Meth 24 0.46 1.62 1.52 Yes
Error 432 0.28 --- --- ---
percentiles of the sampling distribution. The critical value is based on the number of
means being compared (four, representing the different methods) and the number of
degrees of freedom at a given level of significance (0.05 for this analysis). Results of the
DMRT analysis for the Superpave mixes are illustrated in Figure 10.
33
Effect of Bulk Specific Test Method on Air Voids
11 Letters represent results of Duncan's

A Multiple Range Test for air voids
10 9.39 resulting from the bulk specific gravity
methods. Methods with the same letter
9 ranking are not significantly different.
Average Air Void Content, %
B C
8 7.50
7.26
D
7
6.22
6
0
Dimensional Vacuum-Sealing Gamma Ray AASHTO T166
Test Method
Figure 10: Average Air Voids and DMRT Results for Superpave Mixes
Statistically, results of the DMRT comparisons showed that all methods produced
differing resulting air void contents. However, vacuum-sealing and gamma ray bulk
specific gravity methods practically provided similar results given a difference of 0.24
percent air voids. On average, the dimensional method resulted in the highest air void
contents, followed by the vacuum-sealing and gamma ray methods, respectively. Air
void contents determined from AASHTO T166 were the lowest. None of the alternative
methods provided similar results to AASHTO T166.
Results of the ANOVA conducted for the SMA mixes are presented in Table 15.
Factors included within the ANOVA were aggregate type, NMAS, gyration level, and
bulk specific gravity method. All of the main factors except NMAS significantly affected
34
the resulting air void contents. The factor having the most affect was gyration level
followed by bulk specific gravity method and aggregate type.
Table 15: Results of ANOVA Conducted on SMA Mixes

Source DF Mean FStatistic FCritical Significant
Squares (α=0.05)?
Aggregate Type (Agg.) 1 17.31 25.14 3.84 Yes
Nominal Max. Agg. Size (NMAS) 2 11.63 16.89 3.00 Yes
Gyration Level (Gyr.) 2 1624.87 2359.26 3.00 Yes
Bulk Specific Gravity Meth. (Meth.) 3 240.28 348.88 2.60 Yes
Agg*NMAS 2 7.57 10.99 3.00 Yes
Agg*Gyr 2 5.44 7.90 3.00 Yes
Agg*Meth 3 3.99 5.79 2.60 Yes
NMAS*Gyr 4 6.55 9.51 2.37 Yes
NMAS*Meth 6 6.87 9.98 2.10 Yes
Gyr*Meth 6 24.50 35.57 2.10 Yes
Agg*NMAS*Gyr 4 5.08 7.38 2.37 Yes
Agg*NMAS*Meth 6 3.40 4.93 2.10 Yes
Agg*Gyr*Meth 6 0.5 0.70 2.10 No
NMAS*Gyr*Meth 12 1.20 1.74 1.75 No
Agg*NMAS*Gyr*Meth 12 1.65 2.39 1.75 Yes
Error 215 0.69 --- --- ---
Because bulk specific gravity method significantly affected the resulting air voids,
a DMRT analysis was again conducted to determine which methods provided similar
results (if any). Results of the DMRT analysis are presented in Figure 11. Similar to the
Superpave mixes, the vacuum-sealing and gamma ray methods resulted in similar air
void contents. The dimensional method resulted in the highest air voids and the
AASHTO T166 method resulted in the lowest air voids.
35
Effect of Bulk Specific Test Method on Air Voids
12 Letters represent results of Duncan's

A Multiple Range Test for air voids
10.11 resulting from the bulk specific gravity
10 ranking are not significantly different.
Average Air Void Content, %
B
B
8 7.24 7.09
C
6
4.97
0
Dimensional Vacuum-Sealing Gamma Ray AASHTO T166
Test Method
Figure 11: Average Air Voids and DMRT Results for SMA Mixes
Analysis of both the Superpave and SMA data indicated that the four methods of
measuring bulk specific gravity significantly affected resulting air voids. For both mix
types, the vacuum-sealing and gamma ray methods provided similar air voids; however,
the dimensional method provided significantly higher air voids and AASHTO T166
provided significantly lower air void contents.
Theoretically, the dimensional method should provide the highest measured air
void content as this method includes both the internal air voids and the surface texture of
the sample. Therefore, the results in Figures 10 and 11 pass the test of reasonableness for
the vacuum-sealing, gamma ray, and AASHTO T166 methods as all three provided air
void contents less than the dimensional method.
36
Previously within this report, the potential problems with the AASHTO T166
method were discussed, namely the loss of water through surface connected voids during
determination of saturated-surface dry (SSD) mass. When the sample loses water during
determination of the SSD mass, AASHTO T166 would be expected to overestimate
actual density (i.e. air voids are lower). Therefore, if the vacuum-sealing and gamma ray
methods are providing good estimates of bulk specific gravity when all of the data was
compared, then the results of the DMRTs for the Superpave and SMA mixes appear to be
make sense as the AASHTO T166 results provided the lowest air void contents.
However, this finding needs to be verified through further analyses.
The hypothesis for evaluating the vacuum-sealing, gamma ray, and AASHTO
T166 bulk specific gravity methods was that the water displacement method is accurate at
low levels of water absorption. This has been shown (or assumed) by numerous
researchers (1, 4, 9, and 10). Therefore, one method of determining the acceptability of
the vacuum-sealing and gamma ray methods would be to compare these methods to
AASHTO T166 results with samples having low water absorptions.
Based upon the experimental plan for the study, mixes having low water
absorptions would include the fine-graded (ARZ) mixes compacted to 125 gyrations. To
verify which mixtures had low water absorption values, the average water absorption
levels per mix type (a given NMAS, gradation, and gyration level) were plotted in Figure
12. This figure illustrates that mixes meeting the ARZ gradation had the lowest water
absorption level of the gradations studied. Also, samples compacted at 125 gyrations did
in fact have the lowest water absorption levels. Mixes having ARZ gradations and a
NMAS of 9.5 and 19.0 mm had average water absorption levels well below 0.5 percent.
37
Therefore, an ANOVA was conducted on the air void contents results from the 9.5 and
19.0 mm NMAS mixes having an ARZ gradation compacted to 125 gyrations. The
average water absorption value for the 48 samples included in the ANOVA was 0.20
percent with a standard deviation of 0.14 percent. Factors included in the ANOVA were
aggregate type, NMAS, and test method. This ANOVA was conducted to determine if the
different methods of measuring bulk specific gravity provided similar results when water
absorption levels are low and AASHTO T166 is accurate. Based on the ANOVA, the
four methods of measuring bulk specific gravity provided significantly different air void
contents. Therefore, a DMRT ranking was conducted to determine which, if any, of the
methods provided similar results. Results of the DMRT are presented in Figure 13.
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
15
50
125
15
50
125
15
50
125
15
50
125
15
50
125
15
50
125
15
50
125
15
50
125
15
50
125
ARZ BRZ TRZ ARZ BRZ TRZ ARZ BRZ TRZ
9.5 mm 19.0 mm 37.5 mm
Figure 12: Water Absorption Levels by NMAS, Gradation, and Gyration Level
38
Comparison of Test Methods At Low Water Absorption Levels
7.0
Letters represent results of Duncan's
Multiple Range Test for air voids
6.0 resulting from the bulk specific gravity
A
B ranking are not significantly different.
4.9
5.0
Air Void Content, %
4.3 C C
4.0
4.0 3.8
3.0
2.0
1.0
0.0
Dimensional Gamma Ray Vacuum-Sealing AASHTO T166
Test Method
Figure 13: Comparison of Test Methods for Low Water Absorption Level Mixes
Figure 13 shows that the vacuum-sealing and AASHTO T166 methods provided
similar results and both were significantly different than the dimensional and gamma ray
methods. The dimensional method provided the highest air void content, as expected. If
the AASHTO T166 method is accurate for low water absorption mixes, then these results
suggest that the vacuum-sealing method is the only other method that is also accurate.
Figures 10 and 11 suggest that the gamma ray method does an overall adequate job of
estimating bulk specific gravity; however, Figure 13 suggests that it is not as accurate as
AASHTO T166 or the vacuum-sealing methods. Refinements to the gamma ray method
may make this method a viable option in the future.
The next step in analyzing the data was to compare test results from the vacuum-
sealing and AASHTO T166 methods for all combinations of materials utilized in the
39
study. Figures 14 through 21 illustrate the comparisons between the vacuum-sealing and
AASHTO T166 methods. The different figures reflect different aggregate types since the
ANOVA conducted on the overall data set (Tables 14 and 15) indicated that aggregate
type significantly affected the resulting air voids.
Figure 14 illustrates the comparison between the vacuum-sealing and AASHTO
T166 methods for mixes comprised of the limestone aggregate and having a NMAS of
9.5 mm. Figure 15 presents the comparisons for the granite mixes. Four gradation shapes
are illustrated on both figures: ARZ, BRZ, TRZ, and SMA. Paired t-tests were conducted
to determine whether there were differences between the two methods for each of the mix
types (gradation-gyration level).
Comparison Between Vacuum-Sealing and AASHTO T166 Gmb Values,

Limestone 9.5 mm NMAS
2.500
2.450 ARZ BRZ TRZ SMA
2.400
AASHTO T166 Gmb
2.350
2.300
2.250
2.200
2.150
2.100
2.100 2.150 2.200 2.250 2.300 2.350 2.400 2.450 2.500
CoreLokGGmb
Vacuum-Seal mb
Figure 14: Comparison Between Vacuum-Sealing and AASHTO T166 Methods, 9.5
mm Limestone
40
Actual bulk specific gravity values were used in this analysis. No matter the gyration
level, the two methods yielded similar results for the fine-graded (ARZ) mixes. For TRZ
and BRZ mixes, there were significant differences in bulk specific gravity values for all
three gyration levels. For the SMA mixes, there were significant differences at 15 and 50
gyrations, but the two methods provided similar results at 125 gyrations.

Granite 9.5 mm NMAS
2.450
2.350
AASHTO T166 Gmb
2.300
2.250
2.200
2.150
2.100
2.050
2.000
1.950
1.950 2.000 2.050 2.100 2.150 2.200 2.250 2.300 2.350 2.400 2.450
CoreLok Gmb Gmb
Vacuum-Sealing
Figure 15: Comparison Between Vacuum-Sealing and AASHTO T166 Methods,

9.5 mm Granite
Figure 15 presents the comparisons between the two bulk specific gravity
methods for the 9.5 mm NMAS granite mixes. Paired t-tests comparing bulk specific
gravity results from the two methods were conducted. For the granite mixes, bulk
specific gravity measurements on Superpave mix samples prepared at 125 gyrations were
similar for both methods. For the BRZ and TRZ gradation shapes, the two methods
41
provided significantly different bulk specific gravity values at 15 and 50 gyrations. The
two methods resulted in significantly different bulk specific gravity measurements at all
three gyration levels for the SMA mixes.
Figures 16 and 17 illustrate comparisons between the two bulk specific gravity
methods for the 12.5 mm SMA mixes. Paired t-tests were conducted to determine if the
two methods provided significantly different results for the 12.5 mm NMAS SMA mixes.
Figure 16 illustrates that for the limestone 12.5 mm NMAS SMA mixes, the two methods
provide closer results at higher densities, but not similar. However, results of the paired
t-tests showed that there were significant differences in bulk specific gravities at all three
gyration levels. Figure 17 shows that the two methods also provided somewhat similar
results at 50 and 125 gyrations for the granite mixes. Statistically, the results were
similar; however, from a practical standpoint the two methods were different as the
average difference in bulk specific gravities were 0.045 and 0.042 for the 50 and 125
gyration mixes, respectively. Both of these average differences in bulk specific gravity
would have resulted in differences in air void contents of approximately 0.9 percent.
Based upon the statistical and practical analysis of the 12.5 mm NMAS SMA mixes (both
aggregate types), the two methods of measuring bulk specific gravity provided different
results.
Comparisons between the vacuum-sealing and AASHTO T166 bulk specific
gravity methods for the 19.0 mm NMAS mixes utilizing the limestone and granite
aggregates are illustrated in Figures 18 and 19, respectively.
42
2.500
2.450 SMA
2.400
AASHTO T166 Gmb
2.350
2.300
2.250
2.200
2.150
2.100
2.050
2.000
2.000 2.050 2.100 2.150 2.200 2.250 2.300 2.350 2.400 2.450 2.500
CoreLok Gmb Gmb
Vacuum-Sealing
Figure 16: Comparison Between Vacuum-Sealing and AASHTO T166 Methods,
Granite 12.5 mm NMAS
2.500
2.400
2.300
AASHTO T166 Gmb
2.200
2.100
2.000
1.900
SMA
1.800
1.700
1.700 1.800 1.900 2.000 2.100 2.200 2.300 2.400 2.500
CoreLok Gmb Gmb
Vacuum-Sealing
Figure 17: Comparison Between Vacuum-Sealing and AASHTO T166,
43
2.550
2.450
2.400
AASHTO T166 Gmb
2.350
2.300
2.250
2.200
2.150
2.100
2.050
2.000
2.000 2.050 2.100 2.150 2.200 2.250 2.300 2.350 2.400 2.450 2.500 2.550
CoreLok Gmb Gmb
Vacuum-Sealing
2.500
2.450
2.400
AASHTO T166 Gmb
2.350
2.300
2.250
2.150
2.100
2.050
2.050 2.100 2.150 2.200 2.250 2.300 2.350 2.400 2.450 2.500
CoreLok Gmb Gmb
Vacuum-Sealing
Figure 19: Comparison of Vacuum-Seal and AASHTO T166, Granite 19mm NMAS
44
Figure 18 illustrates that limestone mixes having ARZ and TRZ gradations
compacted to 125 gyrations provided similar bulk specific gravity values when the two
methods were compared. The two methods yielded similar bulk specific gravity values at
all three gyration levels for the ARZ gradation mixes. For the BRZ and SMA mixes, the
two methods provided significantly different bulk specific gravity values at all three
gyration levels.
Figure 19 illustrates the comparisons between the vacuum-sealing and AASHTO
T166 methods for the 19.0 mm NMAS granite mixes. Data are presented for the four
gradation shapes evaluated for the 19.0 mm NMAS mixes: ARZ, BRZ, TRZ, and SMA.
Figure 19 shows that for the two methods the ARZ mixes compacted to 50 and 125
gyrations provided similar results. Two other mixes, TRZ-15 gyrations and TRZ-125
gyrations also had strong comparisons between the two methods.
Figures 20 and 21 illustrate the comparisons between the vacuum-sealing and
AASHTO T166 methods for the mixes containing the limestone and granite aggregates,
respectively, having gradations with 37.5 mm NMAS. Results in Figures 20 and 21 show
that the two bulk specific gravity methods did not compare well for the 37.5 mm NMAS
mixes. In all cases, the AASHTO T166 method provided higher bulk specific gravity
values.
45
2.600
2.550
2.500
AASHTO T166 Gmb
2.450
2.400
2.350
2.300
ARZ BRZ TRZ
2.250
2.200
2.150
2.150 2.200 2.250 2.300 2.350 2.400 2.450 2.500 2.550 2.600
CoreLok Gmb G
Vacuum-Sealing mb


2.55
2.5
2.45
AASHTO T166 Gmb
2.4
2.35
2.3 ARZ BRZ TRZ
2.25
2.2
2.15
2.15 2.2 2.25 2.3 2.35 2.4 2.45 2.5 2.55
CoreLok Gmb G
Vacuum-Sealing mb

46
The preceding comparisons between the vacuum-sealing and AASHTO T166
bulk specific gravity methods showed that the two methods provided similar results (from
a statistical and practical standpoint) only about 24 percent of the time. The NMAS
which provided similar results the most often was 9.5 mm, while the gradation shape that
provided similar results the most was the ARZ gradation. The gyration level which
provided similar results the most often was 125. As noted previously, these mixes (9.5
mm, ARZ and 125 gyrations), are the mixes with the lowest water absorption levels
(Figure 12) which is the condition for which the AASHTO T166 method is most
accurate.
There are two possible reasons for the two methods providing significantly
different results for larger NMAS mixes, coarser gradations, and lower gyration levels.
The first possibility causing differences between the vacuum-sealing and AASHTO T166
methods would be that of surface texture on the sample. The vacuum-sealing method
uses a vacuum to conform a plastic bag around the sample. If the bag does not tightly
conform within the texture of the sample, then a portion of the sample’s texture can be
counted as part of the sample’s volume. If this occurs, the sample’s bulk specific gravity
would be lower resulting in higher air voids. Figures 14 to 21 showed that for every
instance where the two methods provided significantly different results, the vacuum-
sealing method provided higher air voids.
Secondly, the combinations of larger NMAS, coarse gradations, and lower
gyration levels provide for large voids within samples (though the overall volume would
be typical). These large voids can be interconnected within the sample and therefore
increase the potential for air voids that are interconnected to the sample’s surface. As
47
stated previously, these large interconnected voids could provide avenues for water to
quickly infiltrate the sample while submerged during AASHTO T166 testing. If the
water can quickly infiltrate then it can also quickly exit the sample after it is taken out of
the water. This would lead to errors in the AASHTO T166 method. The errors would
lead to higher measured densities for samples (lower air voids).
A method for comparing the amount of surface texture accounted for with the
vacuum-sealing method would be to compare the results from the vacuum-sealing and
AASHTO T166 to the dimensional method on mixes with low water absorption. The
difference between the dimensional and the other two bulk specific gravity methods
should be the surface texture of the sample. Figure 22 illustrates the results of bulk
specific gravity tests using the vacuum-sealing and AASHTO T166 methods versus
results based upon the dimensional method for the 9.5 mm NMAS-ARZ-granite mixes.
The amount of measured surface texture would be the difference between the line of
equality and the results from the two methods. This figure shows that both the vacuum-
sealing and AASHTO T166 methods provided a consistent measure of surface texture for
these small NMAS fine-graded mixes. Both regression lines had slopes near 1.0 which
would indicate little change in surface texture between the 15 and 125 gyration mixes.
This lack of change in surface texture was confirmed using a modified sand patch
test to measure the macrotexture of samples. The test was similar to ASTM E965,
“Measuring the Surface Macrotexture Depth Using a Volumetric Technique,” except that
the laboratory compacted samples were utilized. For this test, Ottawa sand was spread
evenly over the face of a laboratory prepared sample. Knowing the mass of the sample
before the sand was added and the mass of the sample with the evenly spread sand
48
allowed the amount of sand remaining within the surface texture (or macrotexture) of the
sample to be determined. The amount of sand within the surface texture could then be
converted to a volume using the bulk specific gravity of the Ottawa sand to represent the
amount of macrotexture on the sample’s surface.
Granite, 9.5 mm NMAS ARZ

14.0
Vacuum-Sealing
12.0
AASHTOT166
10.0 y = 0.9663x - 0.9229

Vacuum-Sealing
Air Voids, %
8.0 y = 0.9626x - 0.8539

AASHTO T166
6.0
4.0
2.0
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Dimensional Air Voids, %
Figure 22: Evaluation of Surface Texture (Granite, 9.5 mm NMAS, ARZ)
The modified sand patch test was conducted three times on each of the 9 samples
(three replicate samples at three gyration levels) meeting the 9.5 mm NMAS-ARZ-
granite combination. Based upon the results, the surface texture measurements at 15, 50,
and 125 gyrations were 26, 29, and 27 cm3, respectively, which would indicate minimal
differences in the amount of macrotexture at the three gyration levels.
49
At the other extreme, with respect to surface texture, would be a larger NMAS
SMA mixture. Figure 23 shows a comparison between the dimensional results and
results from the vacuum-sealing and AASHTO T166 methods for the 19.0 mm NMAS-
SMA-granite combination. Samples of this combination compacted to 15 gyrations
would also represent a worst case for the AASHTO T166 method because they should
have large interconnected voids. Figure 23 shows a much larger difference between the
dimensional results and the other two test methods than was observed for the finer 9.5
mm NMAS mixes. Neither of the regression lines have a slope near 1.0. Again, the
difference between the two regression lines and the line of equality would be the amount
Granite, 19.0 mm NMAS, SMA

20.0
18.0 Vacuum-Sealing
AASHTOT166 y = 0.888x - 2.573
16.0 R2 = 0.9871
Vacuum-Sealing
14.0
12.0
Air Voids, %
10.0
8.0
6.0 y = 0.5022x - 1.323

R2 = 0.9577
AASHTO T166
4.0
2.0
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0
Dimensional Air Voids, %
Figure 23: Evaluation of Surface Texture (Granite, 19.0 mm NMAS, SMA)
50
of surface texture. Based upon the regression lines, the AASHTO T166 method indicates
more overall surface texture and a larger difference in surface texture between the 15 and
125 gyration samples (smaller slope than vacuum-sealing method).
The modified sand patch test was again conducted on the nine 19.0 mm NMAS-
SMA-granite samples. Results of this testing for the 15, 50, and 125 gyration samples
were 129, 100, and 83 cm3, respectively. These results show that the macrotexture was
affected by the amount of compactive effort used to compact the samples (or air voids).
However, the question still remains, which of the two methods (vacuum-sealing or
AASHTO T166) better predicted the amount of surface texture and, hence, bulk specific
gravity?
The modified sand patch results were used to estimate the total amount of
macrotexture for each of the nine 19.0 mm NMAS-SMA-granite samples. This was
accomplished by determining the amount of macrotexture per unit area based upon the
sand patch test on a single face of a sample. Next, the total surface area of the sample
was estimated by adding the area of the two sample faces and the total perimeter of the
sample. The macrotexture per unit area was then applied to the total surface area of the
sample to estimate the total amount of macrotexture on the sample. Results of these
calculations are presented in Table 16. Also included within this table are the volumes
calculated from the bulk specific gravity results (all three methods) and differences in
volumes between the dimensional method and the other two methods. Again, these
differences should be estimates of surface texture on a sample. The method for which the
estimated surface texture better approximates the measured surface texture (depending
upon the method) would be a better predictor of bulk specific gravity.
51
Table 16: Evaluation of Surface Texture for 19.0 mm NMAS-SMA-Granite Samples
Vacuum- Est.
Sealing T166 Total
Dim. Volume, Volume, Macrotexture Surface
Volume, cc cc cc on Single Dvol- Dvol- Texture,
Gyration Rep. (Dvol) (VSvol) (Tvol) Face, cc VSVol Tvol cc
15 1 2196.6 2062.3 1945.9 27.1 134.2 250.6 144.2
15 2 2142.6 2044.2 1936.1 22.6 98.5 206.5 118.2
15 3 2195.6 2092.2 1944.6 23.6 103.4 251.0 125.4
50 1 2026.8 1939.7 1899.7 19.2 87.1 127.0 96.9
50 2 2087.3 1999.1 1950.1 21.7 88.1 137.2 111.8
50 3 2115.0 2022.9 1952.0 17.9 92.0 163.0 93.1
125 1 2079.3 2009.5 1983.6 16.2 69.8 95.8 83.4
125 2 2066.4 2008.1 1981.0 16.5 58.4 85.4 84.7
125 3 2088.9 2020.8 1994.0 15.7 68.0 94.9 81.1
Figure 24 illustrates the relationships between the measured surface texture
(differences in volumes) and the estimated total surface texture from the modified sand
patch test for the nine samples of the 19.0 mm NMAS-SMA-granite combination. This
figure shows that the results of the vacuum-sealing method closely tracked the estimated
amount of surface texture from the modified sand patch test. Results from the vacuum-
sealing test method (difference in sample volume between dimensional and vacuum-
sealing) fell almost on the line of equality and had a slope of approximately 1.0. Results
from the AASHTO T166 method were near the line of equality at the lower values of
surface texture (125 gyration samples) but deviated from the line of equality at higher
levels of surface texture (15 gyration samples). This would indicate that the vacuum-
sealing method does an adequate job of taking into account the surface texture of a
sample when air voids are low. Also, results from AASHTO T166 greatly overestimated
the amount of surface texture at higher overall air void contents (15 gyration samples).
Consequently, if AASHTO T166 is overestimating the amount of surface texture, then it
must also be underestimating the amount of internal air voids. Therefore, based upon this
52
analysis of the surface texture data, the vacuum-sealing method does a better job of
estimating the amount of surface texture on a sample and, hence, does a better job of
estimating the internal air voids for samples at low densities.
Granite, 19.0 mm, SMA

300.0
y = 2.7511x - 130.13
R2 = 0.8503
AASHTO T166
Difference From Dimensional Volume, cc
250.0
200.0
150.0
Vacuum-Sealing
AASHTO T166
100.0
y = 0.9669x - 12.012
R2 = 0.8732
50.0 Vacuum-Sealing
0.0
0.0 50.0 100.0 150.0 200.0 250.0 300.0
Sand Patch Method Estimated Surface Texture, cc
Figure 24: Comparisons of Measured and Estimated Surface Textures
The second possibility discussed for the difference between the vacuum-sealing
and AASHTO T166 methods was that of excessive amounts of water entering and exiting
a sample. A method of determining whether an excessive amount of water enters a
sample and leads to errors in AASHTO T166 testing would be to evaluate the amount of
water absorption. This analysis would not provide an exact measure of the volume of
water that enters and exits a sample during AASHTO T166, but rather provides a
measure of the potential. As water absorption increases, the number and size of air voids
53
interconnected to the surface would also increase. For this analysis, air void contents
were utilized instead of bulk specific gravities because of the differences in aggregate
specific gravities between the mixes. Also, mixtures were categorized by only gradation
shape instead of NMAS, gradation shape, and aggregate type.
Figure 25 illustrates the relationship between air voids and water absorption for
both bulk specific gravity methods for mixes having ARZ gradations. Data shown in this
figure represent NMASs of 9.5, 19.0 and 37.5 mm. This figure shows that the two
methods provided similar air void contents at low water absorptions (as was shown
previously). However, at higher levels of water absorption the two methods begin to
diverge. The standard method for AASHTO T166 indicates that the method is only
applicable at water absorptions of 2.0 percent and below. Figure 25 shows that at 2.0
percent water absorption, the two methods resulted in a difference of 1.0 percent air voids
(8.7 percent for the vacuum-sealing method and 7.7 percent for AASHTO T166). This
difference is greater than would be desired.
Based upon the discussion of the problems with the AASHTO T166 method
earlier within this report, Figure 25 is logical. At high densities (low air voids), there are
very low water absorptions and the two methods provide similar results. However, at
lower densities there are higher water absorption values and differences between the two
methods. Because of the vacuum-sealing test method, conforming the plastic bag to the
sample, it can be surmised that if the method works at high densities, it also works at
lower densities.
54
Gradations Above the Restricted Zone
18.0
16.0
14.0
Air Void Content, %
12.0 0.2579
y = 7.3023x
2
R = 0.5733
10.0 Vacuum-Sealing
8.0 0.212
y = 6.6292x
2
R = 0.4118
6.0 AASHTO T166
4.0
Vacuum-Sealing
2.0 AASHTO T166
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Water Absorption (volume), %
Figure 25: Relationship Between Water Absorption and Air Voids, ARZ mixes
Now the question that must be answered is whether the two methods provide
similar results over the range of water absorption levels encountered. Figure 25 shows
that the two methods provided similar results for mixes having ARZ gradations.
The relationship between air voids and water absorption for mixes having BRZ
gradations is illustrated in Figure 26. Once again, this figure shows that at very low
levels of water absorption, the two methods resulted in similar air void contents.
However, as the level of water absorption increased the air void contents from the two
methods diverged. This figure shows that the two methods diverged at approximately 0.4
percent water absorption, which indicates the two methods provided significantly
55
different results at water absorption values above 0.4 percent. This level of water
absorption at which the two methods diverged was also found by Cooley et al (4) for
coarse-graded mixes. For the BRZ mixes, the 0.4 percent water absorption level
corresponded to approximately 3.5 percent air voids. This level of air voids is below the
typically used design air void level of 4.0 percent. This would indicate that the use of the
vacuum-sealing method during mix design would have resulted in a higher asphalt binder
content. At 4 percent air voids determined by the vacuum-sealing method, the difference
in air voids determined by the two methods was 0.7 percent. Based upon this difference
in air voids, the use of the vacuum-sealing method would have resulted in approximately
0.3 percent more asphalt binder than when using AASHTO T166.
Gradations Below the Restricted Zone

18.0
16.0
Vacuum-Sealing
y = 3.4364Ln(x) + 6.8484
14.0 AASHTO T166 2
R = 0.8243
12.0
Air Void Content, %
10.0
8.0 y = 2.2592Ln(x) + 5.2876

2
R = 0.6788
6.0
4.0
2.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Figure 26: Relationship Between Water Absorption and Air Voids, BRZ mixes
56
Figure 27 illustrates the relationship between air voids and water absorption for
the mixes having TRZ gradations. Similar to the ARZ and BRZ relationships, this figure
shows that at low levels of water absorption (high densities) the vacuum-sealing and
AASHTO T166 methods yielded similar results. However, at higher levels of water
absorption the two methods diverge with the vacuum-sealing method providing a higher
air void content. Figure 27 illustrates that the two methods yielded similar results up to
about 0.4 percent water absorption. Above this level of water absorption, the vacuum-
sealing method resulted in higher air void contents. 0.4 percent water absorption level
corresponded to approximately 3.5 percent air voids for both methods. This level of air
voids is below the typically used design air void level of 4.0 percent. This would indicate
that the use of the vacuum-sealing method during mix design would have resulted in a
higher asphalt binder content.
Gradations Through the Restricted Zone

18.0
16.0
Vacuum-Sealing
AASHTO T166
14.0
Air Void Content, %
12.0 y = 2.9704Ln(x) + 6.347

2
R = 0.7796
10.0
y = 2.2677Ln(x) + 5.4299
8.0 2
R = 0.6041
6.0
4.0
2.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Figure 27: Relationship Between Water Absorption and Air Voids, TRZ mixes
57
Figure 28 illustrates the relationship between water absorption and air voids for
mixes having a SMA gradation. Both relationships had strong R2 values as both were
above 0.80. This figure shows that the two bulk specific gravity methods yielded similar
results at very low water absorption values. This figure shows that for the SMA mixes,
the two methods diverged at approximately 0.2 percent water absorption. This level of
water absorption corresponded to an air void content of approximately 2 percent which is
less than typical design air voids. At 4 percent air voids for the vacuum-sealing method,
the average difference in air voids between the two methods was 1.0 percent. This
difference in air voids at the design level would have resulted in about 0.4 percent more
asphalt binder if the vacuum-sealing method was used instead of AASHTO T166.
Stone Matrix Asphalt Gradations

16.0
y = 3.3377Ln(x) + 8.166
2
R = 0.8875
14.0 Vacuum-Sealing
AASHTO T166
12.0
Air Void Content, %
10.0
8.0
y = 2.1458Ln(x) + 5.6251
2
6.0 R = 0.816
4.0
2.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Figure 28: Relationship Between Water Absorption and Air Voids, SMA mixes
58
6.1.1 Summary of Comparisons Between Bulk Specific Gravity Methods for
Laboratory Specimens
The analyses comparing the vacuum-sealing and AASHTO T166 methods
suggest that the vacuum-sealing method is more accurate for low density samples. The
relationships between water absorption and air void contents suggested that the AASHTO
T166 method was accurate for all water absorption levels encountered for mixes that
were fine-graded (ARZ). However, for mixes having gradations near the maximum
density line (TRZ) or coarser (BRZ and SMA), the level of water absorption that
AASHTO T166 no longer becomes accurate was between 0.2 and 0.4 depending upon
the gradation. These values were much lower than anticipated; therefore, an additional
study was conducted to further evaluate the water absorption level at which AASHTO
T166 becomes inaccurate.
This experiment involved evaluating the effect of time to achieve the saturated
surface dry (SSD) condition on water absorption. Times taken to achieve the SSD
condition were 6 (as fast as could be achieved), 10, 15, 20, 30, and 60 seconds. The
premise of the experiment was that samples having a large proportion of interconnected
voids to the surface of a sample should show decreasing levels of water absorption as the
time to achieve the SSD condition increases. Results of this side experiment should
provide a level of water absorption that leads to increased potential for errors during
AASHTO T166 testing. If a level of water absorption was identified at which the
potential for errors during AASHTO T166 testing increased, the standard could be
changed to indicate a new critical value of water absorption (instead of the current 2.0
percent).
59
A total of 40 samples were selected from the 216 used in the primary experiment.
These samples included the all of the NMASs, gradation shapes, aggregate types, and
gyration levels evaluated. Water absorption values ranged from 0.06 percent to 5.23
percent. Information on the 40 selected mixes is provided in Appendix A. For analysis,
the different mixtures were grouped by gradation shape (ARZ, BRZ, TRZ, and SMA) and
level of water absorption. Categories for water absorption were less than 0.5 percent, 0.5
percent to 1.0 percent, 1.0 percent to 1.5 percent, 1.5 percent to 2.0 percent, 2.0 percent to
2.5 percent, and above 3.0 percent.
Results of this side experiment for mixes having an ARZ gradation are illustrated
in Figure 29. The y-axis for this figure is the difference in air voids from the initial (6
seconds) determination of SSD. The x-axis represents the time taken to achieve the SSD
condition. Also shown on the figure is a horizontal line at 0.25 percent air voids. This air
void content was deemed a critical air void content by Al-Khaateeb et al (11) because it
resulted in an increase of 0.1 percent asphalt binder content during design.
Figure 29 illustrates that at water absorption levels less than 1.5 percent, the air
void content level remained basically unchanged. Only the absorption category of 1.5 to
2.0 percent was above the critical change in air voids of 0.25 percent. However, this
category did not reach the critical level until a time of 30 seconds to achieve the SSD
condition. Results illustrated in Figure 29 suggest that at water absorption levels less
than 2.0 percent the AASHTO T166 method is accurate for fine-graded mixes. This is
similar to the results shown in Figure 25.
60
Time, seconds
0 10 20 30 40 50 60 70 80
0.10
Difference in Air Voids From Initial Observation
Abs. < 0.5
0.00
0.5 < Abs. < 1.5
-0.10
-0.20 1.5 < Abs.
-0.30
-0.40
-0.50
< 0.5 Percent Absorption
-0.60
0.5 < Percent Absorption < 1.0
-0.70 1.0 < Percent Absorption < 1.5

-0.80
-0.90
Figure 29: Results of Time To Reach SSD Condition, ARZ Mixes
Figure 30 illustrates the relationship between the change in air voids and time to
achieve SSD for mixes having coarse gradations (BRZ). Based upon these results, mixes
having water absorption levels greater than 1.0 percent exceeded the critical change in air
voids while mixes having water absorptions less than 1.0 percent did not. Most of the
mixes having water absorption levels above 1.0 percent resulted in changes in air voids
greater than the 0.25 percent at times less than 15 seconds.
The effect of time to achieve the SSD condition on the change in air voids for
mixes having gradations passing near the maximum density line (TRZ) is illustrated in
Figure 31. Similar to the BRZ gradation mixes, this figure shows that once the level of
water absorption increased above 1.5 percent the critical air void level was exceeded.
61
Time, seconds
0 10 20 30 40 50 60 70 80
0.1
0.0
-0.1
0.5 < Abs. < 1.0
-0.2 Abs. < 1.0
-0.3
1.0 < Abs. < 2.5
-0.4
3.0 < Abs.
-0.5

2.0 < Percent Absorption < 2.5 2.5 < Abs. < 3.0
Percent Absorption > 3.0
-0.9
Figure 30: Results of Time To Reach SSD Condition, BRZ Mixes
Time, seconds
0 10 20 30 40 50 60 70 80
0.1
Abs. < 1.0

0.0
-0.1
1.0 < Abs. < 1.5

-0.2
-0.3
3.0 < Abs.
-0.4
-0.5
0.5 < Percent Absorption < 1.0 1.5 < Abs. < 2.5
-0.6

-0.8
-0.9
Figure 31: Results of Time To Reach SSD Condition, TRZ Mixes
62
These results would suggest that a critical level of water absorption for mixes having
gradations passing near the maximum density line (TRZ) would be 1.5 percent.
Figure 32 illustrates the results of the time to achieve the SSD condition for the
SMA mixes. This figure shows that there were large differences in air voids at water
absorption levels above 1.0 percent. However, at water absorption levels below 1.0
percent, the data did not pass the critical air void content of 0.25 percent.
Time, seconds
0 10 20 30 40 50 60 70 80
0.00
Abs. < 1.0

-0.25
1.5 < Abs. < 2.0

-0.50
-0.75
-1.00
2.0 < Abs. < 2.5
-1.25 < 0.5 Percent Absorption Abs. < 3.0
-1.75
-2.00
Figure 32: Results of Time To Reach SSD Condition, SMA Mixes
Results of the primary experiment and the side experiment suggest that the
AASHTO T166 method is not accurate for mixes having gradations near the maximum
density line and coarser when water absorption levels are at 2.0 percent. From the main
experiment, the results suggested that the critical level of water absorption was 0.4
percent for mixes having BRZ and TRZ gradations, and 0.2 percent for SMA mixes.
63
These are all relatively small amounts of water absorption. From the side experiment, the
critical levels of water absorption were 1.5, 1.0, and 1.0 percent for the TRZ, BRZ, and
SMA gradation mixes. These values seem more reasonable; however, the problem still
remains that the critical levels are different for each of the gradation shapes.
6.2 Analysis of Field Compacted Samples
Included within this portion of the study were the cores obtained during the Task
5 field validation experiment. Only the vacuum-sealing and AASHTO T166 test
methods were analyzed, as they were shown most accurate during the laboratory phase of
this experiment. Figure 33 illustrates the relationship between air voids determined from
the two methods for all cores obtained during Task 5. This figure illustrates that at air
void contents less than about 5 percent, the two methods provided approximately similar
results. Above 5 percent air voids, the vacuum-sealing method resulted in higher air void
contents. As air voids increased, the two methods diverged.
Initial analysis of the data compared the vacuum-sealing device versus that of the
AASHTO T166 for these field compacted samples. Similar to the analysis conducted for
the laboratory prepared samples, mixes having low water absorption levels and small
NMAS were utilized in this analysis. These mixes were selected because the AASHTO
T166 method is believed to be accurate at low water absorption levels.
64
Comparison of AASHTO T166 and Vacuum-Sealing Method Field Projects
24.0
22.0
20.0
18.0
Air Voids (AASHTO T166), %
16.0
14.0 0.842
y = 1.2486x
12.0 2
R = 0.8676
10.0
8.0
6.0
4.0
2.0 Line of Equality

0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0
Air Voids (Vacuum-Sealing), %
Figure 33: Comparison Between AASHTO T166 and Vacuum-Sealing Methods,

Field Projects
A paired t-test was conducted on data from mixes having a 9.5 or 12.5 mm
NMAS and water absorption levels less than 0.5 percent. A total of 44 cores were
included in the data set. Results of this analysis indicated that the two methods provided
similar results for these low absorption mixes (t-statistic=0.486, t-critical=2.017, and p-
value=0.629).
Figure 34 illustrates the comparison between the two methods at water absorption
levels below 0.5 percent. This figure confirms that the two methods resulted in similar air
void contents. The average air void content using the AASHTO T166 method was 6.89
percent and the average air void content with the vacuum-sealing method was 6.82
65
Comparison at Water Absorption Levels Below 0.5 percent
14.0
12.0
Air Voids (Vacuum-Sealing), %
10.0
8.0
6.0
4.0
2.0
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Air Voids (AASHTO T166), %
Figure 34: Comparison of Two Methods at Water Absorption Levels Less than 0.5
Percent
percent. Therefore, based upon this analysis the vacuum-sealing method does provide an
accurate measure of bulk specific gravity.
The next analysis was to categorize all of the mixes according to their water
absorption level and conduct paired t-tests to compare the vacuum-sealing and AASHTO
T166 methods. The first category was all mixes having a water absorption level less than
0.5 percent. A total of 58 cores met this requirement. Results of the paired t-test for this
category are presented in Table 17. The results show that the two bulk specific gravity
methods resulted in similar air void contents when water absorption levels were less than
0.5 percent. Average air voids resulting from the AASHTO T166 method were 6.43
percent, while the average air voids resulting from the vacuum-sealing method were 6.66
66
percent. Therefore, the two methods produced an average difference in air voids of 0.23
percent.
Table 17: Comparison of Field Samples by Water Absorption Level

Water Air Voids Air Voids Average
Absorption (AASHTO (Vacuum- Difference, t- p-
Level, % T166), % Sealing), % % statistic value Different?
<0.5 6.43 6.66 0.23 -1.988 0.052 No
0.5 < > 0.75 7.29 7.66 0.37 -1.651 0.110 No
0.75 < > 1.00 7.88 8.00 0.13 -0.346 0.732 No
1.00 < > 1.25 7.75 8.71 0.97 -7.324 0.000 Yes
1.25 < > 1.50 7.20 8.76 1.56 -5.261 0.000 Yes
1.50 < > 1.75 8.16 9.26 1.10 -2.772 0.015 Yes
1.75 < > 2.00 9.03 10.65 1.63 -4.055 0.027 Yes
2.00 < > 2.25 9.86 11.28 1.42 -5.155 0.001 Yes
> 2.25 10.88 13.37 2.50 -8.538 0.000 Yes
The next water absorption level category was from 0.5 to 0.75 percent. A total of
28 cores met this requirement. Results of the paired t-tests showed that the two bulk
specific gravity methods again produced similar air void contents (Table 17). The
average difference in air voids between the two methods was 0.37 percent.
Water absorption levels from 0.75 to 1.00 percent were the next category. Table
17 shows that the two methods again resulted in similar air void contents. The average
difference in air voids for this water absorption category was 0.13 percent.
A number of other water absorption level categories were developed on 0.25
percent water absorption increments above 1.00 percent. However, Table 17 shows that
the two bulk specific gravity methods resulted in significantly different air void contents
once the water absorption level increased above 1.00 percent. This would indicate that
the vacuum-sealing method should be utilized when water absorption levels are above
1.00 percent.
67
An interesting observation about Table 17 is that even at relatively low levels of
water absorption (less than 1.00 percent) the vacuum-sealing method provided slightly
higher air void contents for each of the water absorption level categories where the two
methods were statistically similar. This observation suggests that there could potentially
be a small correction factor required in order for the vacuum-sealing method to result in
the same air void content as AASHTO T166 for a given sample. Figure 35 illustrates a
histogram of the differences in air voids resulting from the vacuum-sealing and AASHTO
T166 methods for mixes having water absorption levels less than 0.5 percent water
absorption. This figure illustrates that the average difference is above 0.0 and
approximately 0.3 percent (as shown in Table 17).
Histogram of Differences (Water Absorptions Less than 0.5 percent)

7 100%
90%
6
Frequency 80%
Cumulative %
5 70%
60%
4
Frequency
50%
3
40%
2 30%
20%
1
10%
0 0%
00
90
80
70
60
50
40
30
20
10
00
10
20
30
40
50
60
70
80
90
00
.
.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
1.
-1
-0
-0
-0
-0
-0
-0
-0
-0
-0
Difference in Air Voids Between Vacuum-Sealing and AASHTO T166, %
Figure 35: Histogram of Differences in Air Voids for Mixes with Water Absorptions
Less than 0.5 Percent
68
A method of determining a correction factor, if needed, would be to regress water
absorption versus the difference in air voids. Figure 36 illustrates the relationship
between the difference in air void contents by the vacuum-sealing and AASHTO T166
methods versus water absorption. Based upon the regression equation, the intercept for
the data was 0.20 percent air voids. Therefore, 0.20 percent air voids would be subtracted
from the vacuum-sealing results in order to match the results from the AASHTO T166
test method.
7.0
6.0
5.0
Difference in Air Voids, %
4.0 y = 0.6503x + 0.2089

R2 = 0.5103
3.0
2.0
1.0
0.0
-1.0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
Water Absorption, %
Figure 36: Relationship Between Differences in Air Voids and Water Absorption
There are several factors that could affect a correction factor between the two
bulk specific gravity methods; namely, gradation shape and NMAS. The most probable
reasons for differences in air void levels (at low water absorption levels) are the surface
69
texture of samples and the size of individual air voids. Therefore, factors such as
gradation shape and NMAS that affect the surface texture and size of air voids would
likely affect the correction factor. Thus, if a correction factor is needed, then it would
likely be different for each mixture.
7.0 CONCLUSIONS AND RECOMMENDATIONS
The objectives of Task 3, Part 3 were to: (1) compare AASHTO T166 with other
methods of measuring bulk specific gravity to determine under what conditions
AASHTO T166 is accurate; (2) identify potential improvements to AASHTO T166 to
achieve a more accurate measure of bulk specific gravity (if needed); and (3) recommend
alternate methods of measuring bulk specific gravity (if needed). Separate sets of data
for laboratory prepared samples and field compacted samples were evaluated to
accomplish these objectives. Based upon these results the following conclusions are
made:
• When laboratory prepared samples having low levels of water absorption were
evaluated, the dimensional method resulted in the highest air void contents
followed by the gamma ray method. The vacuum-sealing and water displacement
(AASHTO T166) methods resulted in similar air void contents when the water
absorption level was low.
• At low levels of water absorption, the water displacement method is an accurate
measure for bulk specific gravity. The error develops when removing the sample
from water to determine the SSD weight. If water flows out of the sample an
error occurs. The allowable absorption level to use the displacement test method
70
is 2% in AASHTO T166 but this level of absorption can create accuracy problems
as shown in this report. This number should be reduced to 1% or lower for better
accuracy. If the allowable water absorption is reduced much below 1%, many
field compacted mixes will exceed this absorption resulting in a need for an
alternate test method. It is recommended that the absorption limit for the
displacement test method be reduced to 1%. If the vacuum-seal method is
adopted on a project the contractor should realize that the compactive effort may
now have to be increased over what has been used in the past since the voids
measured with the vacuum-seal method will be higher than that measured with the
AASHTO T166 method in the 7-10 percent void range or higher.
• The water displacement method was accurate for all water absorption levels
encountered for mixes that were fine-graded (ARZ gradations). For mixes having
gradations near the maximum density line (TRZ) or coarser (BRZ and SMA), the
level of water absorption at which AASHTO T166 was no longer accurate was
between 0.2 and 0.4 percent.
• For mix design samples and other laboratory samples that are compacted to
relatively low voids, the displacement method will provide reasonable accurate
answers. However, for field samples where the void levels will typically be 6%
or higher it is important to evaluate absorption to determine if the vacuum-seal
method needs to be used.
• Care must be used when using the vacuum sealing method to measure density.
Many times the plastic bag developed a leak during the test, leading to an error in
the result. Weighing the sample in air after measuring the submerged weight will
71
indicate if a leak has developed. If a leak is identified the test must be repeated
until an acceptable test is achieved.
• There appears to be a need for a correction factor for the vacuum-sealing and
water displacement methods to provide equal measured air void contents even
when the air void level is low. The correction factor for the mixtures evaluated in
this report should be approximately 0.2% air voids. A better determination of the
correction factor can be made for specific dense graded mixes by compacting
samples in the Superpave gyratory compactor to approximately 4% air voids
(design air void content) and testing using the two test methods. The difference
between these two tests will be the correction factor for the mix.
8.0 REFERENCES
1. M. S. Buchanan. An Evaluation of Selected Methods for Measuring the Bulk
Specific Gravity of Compacted Hot Mix Asphalt (HMA) Mixes. In Journal of the
Association of Asphalt Paving Technologists, Vol. 69. Reno, NV. 2000. pp. 608-
634.
2. Instrotek Incorporated. Corelok Operator’s Guide. Version 10, Raleigh, NC.
2001.
3. K. D. Hall, F. T. Griffith, and S. G. Williams. Examination of Operator
Variability for Selected Methods for Measuring Bulk Specific Gravity of Hot-Mix
Asphalt Concrete. In Transportation Research Record 1761, TRB, National
Research Council, Washington D. C. 2001.
72
4. L.A. Cooley, Jr., B.D. Prowell, and M.R. Hainin. “Comparison of the Saturated
Surface-Dry and Vacuum Sealing Methods for Determining the Bulk Specific
Gravity of Compacted HMA.” In Journal of the Association of Asphalt Paving
Technologists, Vol. 72. Lexington, KY 2003. pp. 56-96.
5. L.A. Cooley, Jr., B.D. Prowell, M.R. Hainin, M.S. Buchanan, and J. Harrington.
“Bulk Specific Gravity Round-Robin Using the Corelok Vacuum Sealing
Device.” FHWA Report No. FHWA-IF-02-044. National Center For Asphalt
Technology Report No. 02-11. November 2002.
6. P. Spellerberg, D. Savage, J.Pielert, “Precision Estimates of Selected Volumetric
Properties of HMA Using Non-Absorptive Aggregate”. Prepared for NCHRP,
February 2003.
7. Troxler Electronics Laboratories, Inc. Manual of Operation and Instruction
Model 3660 CoreReader, Research Triangle Park, NC, 2002.
8. G.A. Malpass and N.P. Khosla,`Evaluation of Gamma Ray Technology for the
Direct Measurement of Bulk Specific Gravity of Compacted Asphalt Concrete
Mixtures.` In Journal of the Association of Asphalt Paving Technologists, Vol.
70. Clearwater Beach, FL 2001. pp. 352-367.
9. J.T. Harvey, Mills, C.Scheffy, J. Sousa and C.L. Monosmith, ”An Evaluation of
Several Techniques for Measuring Air-Void Content in Asphalt Concrete
Specimens.” Journal of Testing and Evaluation, JTEVA, Vol. 22, No. 5,
September 1994, pp. 424-430.
73
10. J.E. Stephens, “Bituminous Mix Density By Coated Specimen” University of
Connecticut Report No: JHR 73-59, 1973
11. Al-Khaateeb, G., C. Paugh, K. Stuart, T. Harman, and J. D’Angelo. “Target and
Tolerance for the Angle of Gyration Used in the Superpave Gyratory Compactor
(SGC).” In Transportation Research Record 1789, TRB, National Research
Council, Washington, DC. 2002. Pp. 208-215.
74

NCHRP w68v2

Uploaded by

Copyright:

Available Formats

NCHRP w68v2

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NCHRP w68v2

Uploaded by

Copyright:

Available Formats

NCHRP Web Document 68 (Project 9-27)

Relationships of HMA In-Place Air

This work was sponsored by the American

The opinion and conclusions expressed or implied in

LIST OF TABLES…………………………………………………………………………......... iii

LIST OF FIGURES ..……………………………………………………………………… ........ iv

1.0 INTRODUCTION AND PROBLEM STATEMENT……………………………... ..........1

2.0 OBJECTIVE……………………………………………………………………….. ..........8

3.0 SCOPE ……………………………………………………………………………... .........8

4.0 RESEARCH APPROACH…………………………………………………………. .........9

4.1 Laboratory Compacted Samples…………………………………………… ........10

4.2 Field Compacted Samples…………………………………………………..........13

5.0 TEST METHODS AND MATERIALS……………………………………… ........……13

5.1 Test Methods………………………………………………………………..........13

5.1.1 Saturated Surface-Dry Method (AASHTO T166).………………………13

5.1.2 Vacuum-Sealing Test Method……………………………………………14

5.1.3 Gamma Ray Method…………………………………………………... ...16

5.1.4 Dimensional Method ………………………………………………….....17

5.2 Materials ……………………………………………………………………… ...17

5.2.1 Tasks 3, Part 1 Laboratory Prepared Materials …………………….....…18

5.2.2 Task 5 Field Projects ………………………………………………….…22

6.1 Laboratory Compacted Samples ………………………………………………..24

6.1.1 Summary of Comparison Between Bulk Specific Gravity

Method for Laboratory Specimens ……………………………………..59

6.2 Analysis of Field Compacted Samples …………………………………………64

7.0 CONCLUSIONS AND RECOMMENDATIONS………………………….……….… 70

8.0 REFERENCES ……………………………………………………………………….... 72

Table 1: Physical Properties of Aggregtae….…………………………………….. 18

Table 2: Asphalt Binder Properties ………………………………………………. 19

Table 3: Summary of Mix Design Information for Superpave Mixes ……………. 22

Table 4: Summary of Mix Design Information for SMA Mixes ……….………… 22

Table 5: Summary Information on Field Projects from Task 5 ………………….. 23

Table 6: Results of Part 3 Testing for 9.5 mm NMAS Limestone Mixes………. 24

Table 14: Results of ANOVA Conducted on Superpave Designed Mixes ……… 33

Table 15: Results of ANOVA Conducted on SMA Designed Mixes …………….35

Table 16: Evaluation of Surface Texture for 19.0 mm NMAS-SMA-Granite

Table 17: Comparison of Field Samples by Water Absorption Level.…………… 67

Figure 2: Volumes Associated with Compacted HMA………………………… 6

Figure 3: Research Approach for Part 3 of Task 3………………………………. 10

Figure 4a: Vacuum-Sealing Device…………….………………………………… 14

Figure 4b: Sealed Sample…………………………………………………………. 14

Figure 5: Equipment for Gamma Ray Test……………………………………….. 16

Figure 6: 9.5 mm NMAS Superpave Gradations..………………………………… 19

Figure 7: 19.0 mm NMAS Superpave Gradations………………………………… 20

Figure 8: 37.5 mm NMAS Superpave Gradations ………………………………… 20

Figure 9: SMA Gradations………………………………………………………… 21

Figure 12: Water Absorption Levels by NMAS, Gradation, and

Figure 13: Comparison of Test Methods for Low Water Absorption

Figure 14: Comparison Between Vacuum-Sealing and

AASHTO T166 Methods, 9.5 mm Limestone…………………….….. 40

Figure 15: Comparison Between Vacuum-Sealing and

AASHTO T166 Methods, 9.5 mm Granite…………………….. …….. 41

Figure 16: Comparison Between Vacuum-Sealing and

Figure 17: Comparison Between Vacuum-Sealing and

AASHTO T166 Methods, Granite 12.5 mm NMAS ……………….…..43

Figure 18: Comparison Between Vacuum-Sealing and

AASHTO T166 Methods, Limestone 19.0 mm NMAS……………….. 44

Figure 19: Comparison Between Vacuum-Sealing and