Verification of Superpave Ndesign compaction levels: final report [July 2007]

VERIFICATION OF SUPERPAVE NDESIGN COMPACTION LEVELS
Final Report
By: Donald E. Watson, Research Engineer Justin Heartsill, Laboratory Engineer
Jason Moore, Laboratory Engineer National Center for Asphalt Technology Auburn University, Auburn, Alabama
And David, Jared, Special Research Engineer Peter Wu, Technical Assistance Bureau Chief Georgia Department of Transportation
Forest Park, Georgia
Sponsored by Georgia Department of Transportation
Atlanta, Georgia
July 2007

DISCLAIMER The contents of this report reflect the views of the authors who are responsible for the facts and accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Georgia Department of Transportation or the National Center for Asphalt Technology. This report does not constitute a standard, specification, or regulation.
i

TABLE OF CONTENTS
Executive Summary .................................................................................................... iii Introduction ................................................................................................................... 1
Problem Statement ............................................................................................... 1 Project Objective .................................................................................................. 2 Scope .................................................................................................................... 2 Summary of Phase 1 Results ......................................................................................... 2 Selection of Field Projects ............................................................................................ 2 Difficulty in Finding Comparable Projects and Project Records ..........................3 Field Evaluation ................................................................................................... 3 Results of Field Analysis .............................................................................................. 4 Rutting .................................................................................................................. 5 Cracking ............................................................................................................... 8 Air Void Analysis .............................................................................................. 11 Phase 2 Results ........................................................................................................... 15 Research Test Plan ...................................................................................................... 15 Test Results and Analysis ........................................................................................... 16 Rutting Susceptibility ......................................................................................... 18 Moisture Susceptibility ...................................................................................... 21 Permeability ....................................................................................................... 24 Fatigue ................................................................................................................ 27 Conclusions ................................................................................................................. 30 Phase 1 ............................................................................................................... 30 Phase 2 ............................................................................................................... 31 Recommendations ....................................................................................................... 32 Acknowledgements ..................................................................................................... 32 References ................................................................................................................... 32 Appendix ..................................................................................................................... 34 Table 1-A. Selected Marshall Projects.................................................................35 Table 2-A. Selected Superpave Projects..............................................................36 Table 3-A. Summary of Field Data for Marshall Projects...................................37 Table 4-A. Summary of Field Data for Superpave Projects ................................38
ii

EXECUTIVE SUMMARY
The Georgia Department of Transportation specifies the Superpave asphalt mix design system for the majority of its dense-graded Hot Mix Asphalt (HMA) mixes. However, there is concern that the number of design gyrations (NDesign) required in Superpave specifications (AASHTO R-35) may be too high for a given traffic level. Mixes designed with too high of an NDesign would be rut resistant, but may be difficult to adequately compact in the field and may lack sufficient durability due to reduced asphalt binder content.
The objective of this project was to evaluate the performance of Georgia's mixes designed using the Superpave gyratory compactor. The comparisons will be used to optimize the Superpave NDesign levels.
The following conclusions and recommendations are made based on an evaluation of 16 Marshall and 16 Superpave projects placed about the same time and serving under approximately the same traffic and other environmental conditions and from laboratory volumetric and performance testing with various aggregate sources and various gyration levels.
1. Both Marshall and Superpave mixtures were generally performing quite well with very little rutting and cracking after a period of about four years.
2. Compared with Marshall mixes, Superpave projects showed a slight trend of increased resistance to cracking at higher asphalt contents and where polymer modifiers were used.
3. It is unlikely that Superpave mixtures will reach the design air voids of 4.0 percent during the life of the pavement. After nearly five years, the average air voids measured in the wheelpaths was 5.7 percent for Superpave projects and 3.8 percent for Marshall projects. Based on volumetric comparisons, 66 gyrations with laboratory produced mix should give approximately the same density as was achieved on the Superpave field projects.
4. The correlation between 4 inch diameter Marshall and 6 inch diameter gyratory samples was not statistically significant enough to assign a relationship between TSR4 and TSR6.
5. An analysis of permeability test methods showed there was no significant difference between uncut, one side cut, and both sides cut for permeability specimens.
6. Aggregate source and strain level were the most significant variables affecting fatigue results. However, samples at 85 and 110 gyrations had only about half of the fatigue life of samples with optimum asphalt content selected at 35 and 60 gyrations. Fatigue life was approximately 25 times longer when the strain level was reduced from 500 to 250 .
Keywords: Marshall, Superpave, gyratory, fatigue, polymer modified asphalt
iii

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
VERIFICATION OF SUPERPAVE NDESIGN COMPACTION LEVELS
Donald E. Watson, Justin Heartsill, and Jason Moore NCAT David Jared and Peter Wu - GDOT
INTRODUCTION
Problem Statement
In order to determine the optimum asphalt content for hot mix asphalt (HMA), the correct laboratory compaction effort needs to be applied during the mix design phase. The laboratory compaction effort for the Superpave mixture design system has been subject to refinement since first being introduced in 1994. When the Superpave gyratory compactor was first developed, a field study was implemented to establish the number of gyrations required for different traffic levels and for different climates. This work was somewhat limited but it did provide information necessary to tentatively establish gyration levels. Three levels of compaction were suggested in the original guidance: Ninitial, Ndesign, and Nmaximum. The test results for samples compacted to Ndesign gyrations were used to determine the optimum asphalt content while the test results of samples compacted to Ninitial and Nmaximum were used to help evaluate potential attributes of the mixture. Initially there were 28 different compaction levels that could be selected based on traffic and climate.
A national study, NCHRP 9-9, "Refinement of the Superpave Gyratory Compaction Procedure," was conducted to consolidate the number of proposed gyration levels. After this study was completed the number of gyration levels was reduced from the original 28 down to 4 (1). This reduction in the number of gyration levels was made in part by concluding that the climate had no significant effect on mixture compactibility since asphalt binder grades are varied to account for the climatic differences. Designers in warmer climates are required to use stiffer asphalts to provide adequate rutting resistance in the hotter geographical environment. This will result in the mixture for two different climatic areas having about the same mixture stiffness at the higher temperatures for which each mixture will need to perform. Data from NCHRP 9-9 also showed that 4 different compaction levels were sufficient to handle all of the traffic levels that needed to be considered (Table 1). Even though this study was a major study, it only consolidated the gyration levels that already existed in the gyration table; it did not verify that the specified number of gyrations was correct.
Georgia Department of Transportation (GDOT) specifies the Superpave asphalt mix design system for the majority of its dense-graded HMA mixes. However, there is concern that the number of design gyrations for Ndesign may be too high for a given traffic level. Mixes designed with too high of an Ndesign gyration level would be rut resistant, but may be difficult to adequately compact in the field, may be more prone to segregation, and may suffer from durability problems such as pre-mature cracking and
1

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
raveling due to inadequate asphalt binder content. The optimization of the Ndesign levels based on the field performance of Marshall and Superpave designed mixes should produce a balance between rut resistance and durability resulting in longer service life.

TABLE 1. Superpave Gyratory Compaction Effort (AASHTO R 35)

Design ESALs (million) <0.3 0.3 to <3 3 to <30 30

Ninitial 6 7 8 9

Ndesign 50 75 100 125

Nmax 75 115 160 205

Application Light traffic, local roads Medium traffic, collector roads Medium to heavy traffic, multilane Heavy traffic, majority of interstates

Other states, such as Alabama and Virginia, have reduced their Ndesign values based on comparisons with Marshall designed mixes that incorporated Superpave performance grade (PG) binders and similar aggregate properties. If the field performance of the Marshall designed mixes indicates that they are rut resistant, and if these mixes contain higher asphalt contents, then it is believed that the Superpave Ndesign levels may be adjusted to provide comparable laboratory compaction of the Marshall mixtures used in the past.

Project Objective

The objective of this project was to evaluate the performance of Georgia's mixes designed using the Superpave mix design procedure and compare them to the performance of Georgia's Marshall designed mixes using Superpave PG Binders and similar aggregate properties. The comparisons will be used to optimize the Superpave Ndesign levels.
Scope

The work was completed in two phases. The first phase includes field distress surveys of pavements constructed using Marshall-designed mixes that included the Superpave aggregate requirements and PG binders with Superpave-designed mixes (design asphalt content determined using the gyratory compactor). The second phase consisted of a laboratory study to evaluate the effects of changing gyration levels in terms of rut resistance, moisture susceptibility and permeability.

SUMMARY OF PHASE 1 RESULTS

Selection Of Field Projects

Projects for the Phase I field evaluations were selected with the help of the Assistant Materials and Research Engineer of the Georgia Department of Transportation. Efforts

2

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
were made to find sixteen paired sites, with the same Nominal Maximum Aggregate Size (NMAS) and produced with the same aggregates for field distress surveys.
A list of projects let to contract beginning in 1996 to the year 2000 was obtained and used to identify potential Superpave and Marshall projects for this study. Those years were chosen because the first Superpave projects were let to contract in 1996, and by 2000 the use of Superpave was common across the state. It was anticipated that projects constructed during this time period may be old enough to begin showing signs of distress. A total of 95 Marshall projects and 122 Superpave projects were initially identified as potential candidates for this study. The large number of initial projects was needed in order to match as closely as possible projects for each design method with similar materials, traffic levels and age.
Difficulty in Finding Comparable Projects and Project Records
While searching for comparable field projects, it became evident that comparisons for each desired variable may not be possible. During the transition of implementing Superpave, those mixtures were typically used on the higher traffic loading facilities while Marshall mixtures were used for the lower traffic volume routes. As a result there were very few Superpave projects placed on low traffic routes for the time period evaluated. A special effort was made to try to find projects where Superpave was placed on low traffic volume routes so that performance could be evaluated over a wide range of traffic conditions. However, it was more feasible to find Marshall mixes placed on high traffic volume projects around the same time period. While Superpave mixtures were designed at various gyration levels, all Marshall mixtures evaluated were designed at 50 blows.
An attempt was made to match paired sites including both unmodified and modified binders. However, polymer modified binders were not typically used in Marshall mixes. With the implementation of Superpave, a decision was made to require polymer modified binder in the surface course of projects that had more than 25,000 ADT based on twoway traffic. Six of the sixteen Superpave projects used polymer-modified asphalt.
QC/QA test data from the projects selected was obtained from archived project records so that an analysis of project data could be made. The information was available for all but one project. This allowed a comparison to be made between roadway density at the time of construction to the current density obtained from cores taken in the outside wheel path of each project. Some variation between as-built density and current density was expected since the density immediately after construction was usually determined with a calibrated nuclear density gauge rather than cores.
Field Evaluation
From the potential list of projects, 16 Marshall and 16 Superpave projects were selected
3

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
that matched as closely as possible the same age, traffic conditions, mix type, aggregate source, and geographical area. Of the 32 projects evaluated, only 12.5 mm mixes were used for the surface course. The list of Marshall projects included in the final selection is shown in Table 1-A of the Appendix. The Superpave projects included in the final selection are shown in Table 2-A in the Appendix. For each project selected, the sites were visited and a representative area selected for evaluation. Rut depths were manually determined for each wheel path by use of a string line and the results averaged for each project. In addition, cracking and surface condition was noted. AASHTO PP44-01, "Standard Practice for Quantifying Cracks in Asphalt Pavement Surface," was used to quantify the intensity and severity level of pavement cracking.
Results of Field Analysis
Tables 3-A and 4-A shown in the Appendix contain the results of test data for the field evaluation. The average age of Marshall projects selected was 6.1 years and the average age of Superpave projects was 4.7 years. The Marshall projects are typically older because no more Marshall projects were let to contract once the decision to implement Superpave was made. The Average Daily Traffic (ADT) for Marshall projects was 21,245 and the average traffic volume for Superpave projects was 21,420. The average asphalt content based on analysis of "as-built" project Quality Acceptance (QA) data was 5.28 percent for Marshall mixes and 4.94 percent for Superpave mixes. A comparison of asphalt content between the two methods is shown in Figure 1.

Asphalt Content (%)

Asphalt Content Comparison
(Based on Project QA Data)

Marshall Superpave

6.50

6.00 5.50

5.00 4.50 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Project No.

FIGURE 1. Asphalt Content Comparison

4

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Cores from the projects showed the surface layer thickness and total depth of pavement structure for Marshall and Superpave projects were very similar. The surface layer of cores from Marshall projects averaged 1.62 inches thick and the surface of Superpave projects averaged 1.57 inches thick. The total asphalt pavement structure averaged 9.77 inches for Marshall projects and 9.72 inches for Superpave projects. It was important that the paired projects have similar pavement structure. Otherwise, any difference in rutting or cracking may have been attributed to structural variation rather than mixture performance.
Rutting
Of the 32 projects selected for this research study, 11 Marshall and 8 Superpave projects exhibited measurable deformation. The measured rut depths in each wheelpath ranged from 0.0 to 0.19 inches for both mix design methods. Rut depths in each wheel path were then averaged for each project and those results are shown in Figure 2. Marshall projects had an average rut depth of 0.06 inches while Superpave projects had an average rut depth of 0.05 inches. Therefore, there is no significant difference in average rutting resistance between Marshall and Superpave mixtures.

Comparison of Rut Depths

0.25 Marshall

0.20

Superpave

Rut Depth (in)

0.15

0.10

0.05

0.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Project

FIGURE 2. Comparison of Rut Depths
Figure 2 shows that Marshall projects 6 and 15 had the highest degree of rutting. Marshall project 6 (M-6) is a project on State Route (SR) 1/U.S. 27 in Summerville. The project was six years old and had almost 25,000 ADT. Cores from the wheelpath showed

5

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
the current pavement has 3.2 percent in-place air voids. Project M-15 was also six years old and is located on SR 7/U.S. 341 between Perry and Fort Valley. Roadway cores from the project averaged 2.8 percent air voids. The mix on this project may be slightly unstable due to the very low air void level, but the project also has more than 5,000 ADT.
Some of the rutting in Superpave mixes may have been a result of low air voids as well. For example, Superpave project 9 (S-9) had the largest average rutting of all projects evaluated. The project is located on Peachtree Industrial Blvd. off the state route system in Gwinnett County. This was a four year old project with over 50,000 ADT. The project cores showed the mix had only 2.0 percent air voids in the wheelpath. The optimum asphalt content for the initial mix placed on this project was 4.09 percent. The asphalt content was raised to 4.7 percent after Lot 7 due to compaction problems early in the project. (A Lot is defined as the amount of mixture produced and placed in a production day; but, if less than 500 tons is produced, the tonnage may be added to the next production day.) A new mix design was used beginning with Lot 21 due to a change in RAP source. The gradation was the same as for the previous mix except that the No. 200 (0.075m) sieve was reduced from 5.0 to 4.0 percent passing. The new mix had an asphalt content of 5.3 percent. The average asphalt content used throughout the project was 4.85 percent. The high asphalt content used on this project in order to obtain satisfactory compaction results during construction may explain why the project now has low air void levels.
However, the Peachtree Industrial Blvd. project gives evidence that Superpave mixtures may be placed with higher asphalt contents. Even though the average asphalt content for the project was about 0.75 percent higher than the initial mix design, and the project has over 50,000 ADT, the average rutting is less than one-quarter inch. Furthermore, a comparison of asphalt content to rut depth for Superpave mixtures shows there is very little relationship between the two. The low R2 value (0.0215) shown in Figure 3 indicates that the variation in rut depth cannot be explained by changes in asphalt content. However, this information is based on results from different mixes used on 16 different projects. For a particular mix, one would reasonably expect increases in asphalt content to result in increases in rutting.
Figure 4 compares the effect of traffic volume on pavement deformation. For Marshall projects, increases in traffic volume do not impact the amount of rutting. For Superpave projects, there is also little effect of traffic volume on the amount of pavement rutting although there is a slight trend that rutting increases as traffic increases. It is expected that traffic volume would have little effect on rutting of Superpave projects since the compaction level increases and asphalt content decreases with incremental increases in traffic. Both Figures 3 and 4 indicate that the current method of increasing compaction effort (and decreasing asphalt content) are effective in accounting for differences in traffic volume.
6

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

Rut Depth, in

AC vs Rut Depth - Superpave

0.20

0.18

0.16

0.14

0.12

0.10

0.08

0.06 0.04

R2 = 0.0215

0.02

0.00

4.30 4.50 4.70 4.90 5.10 5.30 5.50 5.70 5.90

AC Content, %

FIGURE 3. Comparison of Rut Depth to Asphalt Content

Rut Depth (in)

0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00
0

Traffic vs Rutting
Superpave R2 = 0.1181

Superpave Marshall
Marshall R2 = 0.0154

20,000

40,000

60,000

Traffic Volume (ADT)

80,000

FIGURE 4. Effect of Traffic on Pavement Rutting

7

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
A rutting comparison was made between Superpave projects that used PG 76-22 polymer modified asphalt to Superpave projects that used the standard paving grade, PG 67-22. This comparison was done to evaluate the effectiveness of polymer modifiers in helping to control rutting. The average rutting for projects that used polymer modified asphalt was 0.06 in. while the projects with unmodified asphalt averaged 0.04 in.
These results indicate there is no significant difference in performance of modified and unmodified asphalt as related to rutting resistance. However, one must remember that the polymer modified asphalt mixtures were used on the highest traffic volume projects. Average ADT for projects with polymer modified asphalt was 38,323 while the traffic volume for standard paving grade projects was 10,152. When the order of magnitude (nearly four times) for traffic volume is considered, it is impressive that the polymer modified projects had basically no more rutting than the lower traffic volume projects. Based on the difference in rutting as compared to the difference in traffic volume, it appears that polymer modified asphalt does improve rutting resistance for mixtures placed under high stress conditions.
Cracking
Only 4 of the 16 Marshall projects and 5 of the Superpave projects showed any cracking distress. The cracking intensity, measured in linear feet per square foot, ranged from 0.0 to 0.188 for Marshall projects and from 0.0 to 0.138 for Superpave projects. Two Superpave projects, S-3 and S-4, were overlays of Portland Cement Concrete (PCC) pavement. Cracking intensity for these two projects was 0.173 and 0.122, respectively. The reflective cracking on these projects was not included in the summary of project data because the cracking was more related to the effect of underlying PCC than the mix design method. Therefore, the average cracking intensity for all projects was 0.065 for Marshall and 0.022 for Superpave projects.
The trendline for the data shown in Figure 5 indicates a slight tendency for reduced cracking with increases in asphalt content but the low R2 value (0.19) indicates the correlation is not significant. This is somewhat reasonable since all the mixes were designed at optimum asphalt content, but it does at least indicate that cracking resistance of Superpave mixes may be improved somewhat by increasing asphalt content. A comparison was also made of the difference in asphalt content from one Georgia DOT district to another, and the results, shown in Figure 6, reveal that the average asphalt content for Superpave mixes was relatively consistent from district to district.
It was suspected that cracking may be related to structural depth of the pavement. If pavement depth has not kept pace with the increase in traffic volume since the projects were originally constructed, it would be reasonable to expect more cracking for the thinner pavement sections. However, Figure 7 shows that there is no correlation in the data for cracking as related to pavement depth.
8

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

Cracking Intensity, lf/sf (%)

Cracking vs Asphalt Content
16

14

12

10

8

6

4

2

R2 = 0.19

0 4.25 4.50 4.75 5.00 5.25 5.50 5.75

AC (%)

FIGURE 5. Effect of Asphalt Content on Cracking Intensity

AC (%)

5.75 5.50 5.25 5.00 4.75 4.50 4.25
0

Superpave AC by District
R2 = 0.0777

1

2

3

4

5

6

7

GDOT District

FIGURE 6. Superpave Asphalt Content by District

9

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

Crack Intensity, lf/sf (%)

16 14 12 10
8 6 4 2 0
4

Cracking vs Pavement Depth

R2 = 0.0123

6

8

10

12

14

16

18

Pavement Depth (in)

FIGURE 7. Effect of Pavement Depth on Crack Intensity
The ability of polymer modified asphalt to resist cracking was also considered. It was expected that the PG 76 grade asphalt would be more elastic than the standard PG 67 paving grade asphalt and that it might improve resistance to cracking. Figure 8 shows that polymer modified mixtures had a tendency toward less cracking, but again, there was a poor correlation.

Cracking Intensity (%)

Cracking vs PG Grade
16

14

12

10

8

6

4

2

R2 = 0.1212

0

64

67

70

73

76

79

PG Grade

FIGURE 8. Effect of PG Grade on Cracking Intensity

10

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Air Void Analysis
Cores were taken from each project to determine in-place density and percent air voids, thickness of the surface layer, and thickness of the overall pavement structure. Four cores were typically taken from the outside wheelpath of each project. The percent air voids from samples taken in the wheelpath represents the attainable air void levels after several years of traffic and should compare closely with the design air void level of the mixture.
As seen in Figure 9 the average roadway air voids for Marshall projects was 3.8 percent and for Superpave projects was 5.7 percent. Superpave mixtures are designed at 4.0 percent air voids. The design air voids (4.0 percent at Ndesign) represents the air void level that the pavement can be expected to reach after initial compaction during construction and additional densification after several years of traffic. Generally, most of the increase in pavement density after construction occurs during the first three months after placement (2) and, to a much less extent, for up to 3 years afterward under the repeated loading action of traffic. After about 3 years, the mixture is typically aged sufficiently to make it stiff enough to resist further densification. Based on this field data, Superpave mixes are not likely to reach the average of 4.0 percent air voids they were designed at. The high air void levels in the wheelpath indicates that the Superpave gyratory compaction level used for the project was higher than necessary for the additional densification under traffic loading encountered on the project. This also indicates that higher asphalt contents may be needed in Superpave mixtures.

Va (%)

Field Air Voids (From Wheelpath)

10.0 Marshall
9.0 Superpave 8.0

7.0

6.0

Avg. Superpave = 5.7

5.0

4.0

Avg. Marshall = 3.8

3.0

2.0

1.0

0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Project No.

FIGURE 9. Roadway Air Voids
On the other hand, Marshall mixtures were typically designed at 4.5 percent air voids. Figure 9 shows that the Marshall projects averaged less than the 4.5 percent design air
11

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
voids. This means that Marshall mixtures may have been placed with a slightly higher asphalt content than needed at the time of construction, but the average results are very close to the currently recommended 4.0 percent air voids for mix designs.
Based on Superpave criteria, it is recommended that asphalt mixtures be designed so that they still maintain a minimum of 2.0 percent air voids toward the end of their design life. This air void level is needed to resist rutting and to provide room for normal thermal expansion and contraction to take place without fear of asphalt cement flushing to the surface. None of the Marshall projects had air void levels less than 2.0 percent and only one Superpave project had air voids less than 2.0 percent. The standard deviation of field air voids between the 16 Marshall projects was 1.4 percent while the standard deviation for air voids between the 16 Superpave projects was 2.6 percent. The wider range in field air voids for Superpave projects may indicate that compaction was not controlled as consistently during placement as was done for Marshall mixtures and/or that more compaction effort is required to achieve the desired density for Superpave mixes. This may be due to relative inexperience with Superpave mixtures at the time these early projects were placed.
For comparison, project Quality Acceptance tests performed at the time of construction were retrieved from archive records and evaluated (Figure 10).

Field Air Voids (%)

As-Constructed Field Air Voids
(Based on Project QA Data)

10
Marshall

9

Superpave

8
Avg. Superpave = 7.3
7

6
Avg. Marshall = 6.1
5

4 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Project No.

FIGURE 10. As-Built Roadway Air Voids The as-built test results showed that the average field air voids was 6.1 percent for
12

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Marshall projects with values as high as 8.3 percent and a standard deviation in test results of 0.9 percent. The average as-built air void level for Superpave projects was 7.3 percent with values as high as 9.7 percent and a standard deviation of 1.2 percent. These results indicate that Superpave mixtures could have been placed with a higher asphalt content. Since there was very little rutting on the projects reviewed, additional asphalt content could have been used to improve field density and further extend the service life of pavements.
For Superpave mixtures, an increase in asphalt content may be accomplished in part by reducing the gyratory compaction level. However, if only the gyration level is reduced, the mix designer may modify the gradation to meet mixture volumetric requirements without increasing asphalt content. For that reason, other mixture properties such as Voids in Mineral Aggregate (VMA) are increased accordingly to insure asphalt content would be increased. The asphalt mix designs were recovered from archive files for 24 of the 32 projects evaluated. The VMA relationship to design method for mixtures used on these projects is shown in Figure 11. It is important to note that Georgia uses the effective specific gravity of aggregate to determine VMA properties. The average VMA for Marshall mixes was 16.8 percent with a standard deviation of 0.5. In contrast, the average VMA for Superpave mixes was 14.9 with a standard deviation of 0.8. This data shows that Marshall mixtures were designed with almost two percent higher VMA values than Superpave mixes and therefore had greater capacity for increased asphalt content. (As a result of this study, GDOT increased minimum VMA requirements for all Superpave mixes beginning in February, 2006.)

VMA (%)

VMA Comparison

18 Marshall

Superpave 17

Marshall Avg. = 16.8

16
Superpave Avg. = 14.9 15

14 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Project No.

FIGURE 11: Relationship of VMA to Mix Design Method

13

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
As a general rule of thumb, 25 gyrations will result in about 1.0 percent difference in VMA. Therefore, Superpave mixtures would have to be designed at 25-50 gyrations lower in order to have the same asphalt content and VMA properties as the Marshall mixes. This is impractical, however, since the gyration level for lower traffic volume projects is only 50 gyrations.
Therefore, simply reducing the gyratory compaction effort alone in order to increase asphalt binder content without changing other mix properties for Superpave mixtures is not a viable option. A better design approach may be to combine the effects of revising the gyratory compaction effort with the use of a finer gradation than the coarse-graded Superpave mixtures used in the past.
All 12.5 mm Marshall mixtures were designed to have 45 percent passing the No.8 (2.36 mm) sieve. On the other hand, 12.5 mm Superpave mixtures for the 16 projects evaluated had an average of 35 percent passing the No. 8 (2.36 mm) sieve with a range from 29 to 39 percent passing based on project quality acceptance test data. Performance of the Marshall mixtures shows that finer-graded mixes can be used without sacrificing rutting resistance. The finer mixes will likely result in higher VMA values that would increase capacity for more asphalt cement and at the same time would potentially reduce aggregate particle segregation.
One of the primary concerns for this research study was to consider the effect of mix design method on pavement rutting. If mixes designed with the Marshall method were resistant to rutting at a higher asphalt content, then it would be reasonable for Superpave mixes placed under equivalent conditions to be able to sustain higher asphalt contents as well. Therefore, a Two-Sample T-Test was conducted in which rut depth was compared to different mix design methods. The statistical results had a P-value of 0.401 as shown in Table 6 and indicated there is no significant difference in rutting between the two mix design methods.
TABLE 6. Two-Sample T-Test and CI: Marshall Rut vs Superpave Rut
Two-sample T for Marshall vs. Superpave
N Mean StDev SE Mean Marshall 16 0.0650 0.0589 0.015 Superpave 16 0.0469 0.0613 0.015

Difference = mu (Marshall) - mu (Superpave) Estimate for difference: 0.018125 95% CI for difference: (-0.025330, 0.061580) T-Test of difference = 0 (vs. not =): T-Value = 0.85

P-Value = 0.401

DF = 29

14

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
The amount of rutting is not significantly different when the average rut depths for Marshall projects (0.06 in.) is compared to that of Superpave projects (0.05 in.). Since Marshall projects averaged 1.4 years older than Superpave projects, it is reasonable to assume Superpave projects will experience a slight increase in rutting and cracking by the time they reach the same service life as Marshall projects.

PHASE 2 RESULTS

RESEARCH TEST PLAN

This research was planned as a 4 8 test matrix with 4 gyration levels and 8 aggregate sources. However, there were four performance tests with multiple replicates for each test and various conditions for each performance test as well so that a total of 2016 samples would be needed. In order to reduce the amount of lab work necessary and reduce the time for completion of the research a 1/2 factorial experiment was used. An example of how the initial design experiment was set up is shown in Table 7 based on aggregate source and gyration level. Although testing for each cell is not conducted, there are enough replicates (a minimum of three for each test) at each gyration level with different aggregate sources that statistical differences can be determined.

No. of Gyrations
35 60 85 110

TABLE 7. One-Half Factorial Design Experiment

BarinColumbus
X
X

Kennesaw
X X

Lithia Springs
X X

Aggregate Source

Norcross
X X

Stockbridge X
X

Mt. View X
X

Camak X
X

Ruby
X X

A variety of test conditions were used for conducting the performance tests. Moisture susceptibility was performed on both 4 inch and 6 inch diameter samples. A Marshall hammer was used to compact the 4 inch specimens and a gyratory compactor was used to prepare the 6 inch diameter samples. Six replicates were made for each test. For fatigue testing, two strain levels (250 and 500 ) were used with three replicates per test. For rutting susceptibility 25 mm Superpave mix was tested at both 64 and 50C while 12.5 mm mix was tested at only 64C. Six APA replicates are used to represent one test. Permeability tests were conducted at target air void levels of 7, 9, and 11 percent and each sample was also tested for three conditions (uncut, top cut, both ends cut) and three replicates were made for each test.

A 12.5 mm Superpave mix was used for performance testing. However, the rutting susceptibility test using an Asphalt Pavement Analyzer (APA) was also conducted on a 25 mm mix used for base courses. Since base courses do not reach the elevated

15

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
temperatures of surface courses, there was a need to evaluate performance of the 25 mm mix at 50C which more closely represents the in-place high service temperature of the base mix.
TEST RESULTS AND ANALYSIS
This research concentrated on the effect gyration level had on mixture performance with 35, 60, 85, and 110 being the gyration variables. Another concept that was evaluated was that of the aggregate locking point. The "locking point" is the point at which the aggregate structure begins to lock together so that additional gyrations result in greater potential for fractured aggregate and increased difficulty in obtaining the desired field compaction. As the number of gyrations is increased after the locking point is reached, the density of samples will continue to increase because fractured aggregate fills the void structure. However, it is often very difficult to obtain the same level of density in the field because the mixture is not confined as it is in the laboratory molds. It is also not desirable to have fractured aggregate under field conditions because it creates uncoated surfaces that tend to make the mixture more susceptible to moisture damage and accelerates pavement deterioration.
One of the more difficult aspects of using the "locking point" concept has been to define when the locking point is actually reached. Typically this is done by examining the sample height during compaction and identifying the locking point as when the height remains the same for successive gyrations. But, there is no particular agreement between agencies as to whether the first, second, or other occurrence of having the same height with successive gyrations is the actual locking point. For this study, the "locking point" at which the sample height remained the same for two, three and four consecutive gyrations was considered. In Table 8, for example, LP 2 indicates the first occurrence when two successive gyrations resulted in the same sample height, LP 3 is the first occurrence where the sample height remained the same for three successive gyrations, and LP 4 is the first occurrence where the sample height remained the same for four successive gyrations.
For this study, all 12.5 mm mix designs were compacted to 110 gyrations and the locking point determined for those samples. Although there is no agreement between agencies on the definition of the locking point, an analysis in NCHRP 9-9 (1) research showed a trend that LP 3 most closely represented the level of density obtained in the field (2).
16

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

TABLE 8. Locking Point Gyration Level

Aggregate Source LP 2 Gyrations LP 3 Gyrations LP 4 Gyrations

Kennesaw

35

Lithia Springs

39

Camak

39

Stockbridge

41

Ruby

43

Norcross

37

Mt. View

36

Barin-Columbus

42

Average

39

68

92

68

90

64

88

69

91

83

100

67

84

62

83

73

102

69

91

Based on Equation 1, the average LP 3 locking point value of 69 gyrations resulted in samples with an average of 94.6 percent of Gmm (or 5.4 percent air voids). This compares very closely to the average field air voids of 5.7 percent (Figure 10) for Superpave mixtures that were an average of 4.7 years old. This indicates that the NDesign levels previously specified for Superpave mixes was higher than necessary, and that the LP 3 locking point gyration level may result in laboratory density that is closer to the ultimate density obtained in the field.

%Gmm xx

=

100



Gmb Gmm



hd hxx

(Equation 1)

Where,

%Gmmxx = Percent of maximum mix specific gravity at gyration level xx

Gmb = Bulk specific gravity of compacted mixture

hd = Sample height at the design level of gyrations

hxx = Sample height at gyration level xx

During the course of this research study, Georgia DOT changed from using gyration levels related to traffic conditions to currently require that samples be compacted to 65 gyrations regardless of traffic condition. The compaction level of 65 gyrations was chosen to simulate the locking point of typical aggregates used in Georgia based on the LP 3 definition for locking point. The 65 gyration value was found to be very close to the LP 3 average of 69 gyrations obtained for the eight aggregate sources used in this study. Based on Figure 12 it would take 66 gyrations for the mixes tested in this study to have the same average density as the roadway cores from field projects.

17

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

% Gmm

98 96 94.3 94 92 90 88 86 84 82 80 78
0

y = 2.7439Ln(x) + 82.8 R2 = 0.8845
10 20 30 40 50 60 66 70 80 90 100 110 120 No. of Gyrations

FIGURE 12. Gyrations to Match Lab Density to Field Density

Rutting Susceptibility

There was also a need to evaluate the temperature used for rut testing of asphalt base courses with the Asphalt Pavement Analyzer (APA). In SHRP-A-648A, "Weather Database for the Superpave Mix Design System," (3) an equation was described that related air temperature to pavement surface temperature based on the latitude of the location. Work by Solaimanian and Kennedy (4) also used this equation for predicting pavement temperatures.

The relationship between air temperature and pavement temperature is as follows:

Tsurf = Tair - 0.00618 lat2 + 0.2289 lat + 24.4

(Equation 2)

Where,

T is expressed in EC and the latitude is in degrees.

Below the pavement surface, the temperature is predicted using heat flow models contained in the Federal Highway Administration's Environmental Effects Model (3). During the hottest 7-day period, in the heat of the day pavement surface temperature is increasing and heat flows downward into the pavement. Using data from the SHRP temperature data base, an equation was developed that expresses the change in temperature with depth:

T(d) = T(surf) (1 - 0.063 d + 0.007 d2 - 0.0004 d3)

(Equation 3)

18

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Where, T(d) and T(surf) are in EF and the depth, d, is in inches.
This information is useful because rut testing with the Asphalt Pavement Analyzer (APA) requires the typical 7-day average high temperature of the surface course (64C) be used for testing. Since base courses do not reach the high temperature of a surface course, APA testing could be performed at lower temperatures and would result in base courses with a higher asphalt content. The increase in asphalt content would improve fatigue life of Georgia pavements and reduce the tendency of these coarse mixes to segregate.
As a general rule, most of the pavement rutting occurs within the top four inches of the pavement surface. This is where the asphalt binder is heated during the hot summer to temperatures that exceed the softening point of the standard paving grade asphalt cement. The softening point of typical asphalt binders used in the southeast is around 120F. At the NCAT Test Track in 2001, the average 7-day maximum surface temperature was 136.8F and in 2002 it was 142.2F. However, the average high temperature at a depth 4 inches below the surface was only 115.3F and 121.8F, respectively (5). Based on this information the pavement cooled approximately 21F within the top 4 inches.
Analysis of the temperature data shows that the pavement only cooled 10.2F in 2001 and 11.3F in 2002 when going from a 4 inch depth to a 10 inch depth. Therefore, based on the 7-day high temperatures measured within the pavement structure at the NCAT Test Track, the maximum pavement temperature of typical base courses for the same geographical environment would be no more than 122F (50C). This means that APA testing for base courses should probably be conducted at 50C rather than the 64C now used. This change in test method would be in keeping with Superpave binder testing in which the test parameter is kept constant but the test temperature is changed to match field temperature conditions.
Rutting susceptibility was performed according to GDT-115 in which samples are prepared at target air voids of 5.0 1.0 percent. Testing of the 12.5 mm mixes was performed at 64C, but for this study 25 mm NMAS mixtures were tested at both 64C and 50C. The different temperature was used because a study at the NCAT research test track showed that base mixtures do not reach the same elevated temperatures of surface mixtures. For Georgia's geographical location, the base temperatures are generally around 49-50C. therefore a comparison was needed to see the effect of reducing test temperature from the standard of 64C to the more reasonable test temperature of 50C for base mixtures. A load of 100 pounds and a hose pressure of 100 psi was used for all tests.
Based on Figure 13, to get the equivalent of 4 mm rutting at 64C, but perform the test at 50C, it would take an increase of 0.75% asphalt (from 4.15 to 4.9). From Figure 14, it
19

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
would take 45 gyrations (from 40 to 85) to get the same difference in asphalt content.

Rut Depth (mm)

Effect of AC & Temp on Rutting

7 50C
6 5 64C 4 3

y = 0.2082x2.0726 64 C R2 = 0.9454

y = 0.0164x3.4573 50 C R2 = 0.7478

2

1

0

3.5

4.0

4.5

5.0

5.5

AC (% )

FIGURE 13. Effect of AC and Test Temperature on Rutting

AC, %

Effect of Gyration Level on AC

6.0

y = 11.181x-0.1875

25mm

5.5

12.5 mm R2 = 0.9998 12.5mm

5.0

4.5

4.0

y = 11x-0.2282

25 mm R2 = 0.9999

3.5

3.0 10

35

60

85

110 135

No. of Gyrations

FIGURE 14. Effect of Gyration Level on AC
This means that if 25 mm mixes are to be tested at 50C, those mixes could have 0.75 percent higher asphalt content or be designed at 45 gyrations lower than the Ndesign value used for surface mixes on the same project in order to get the same rutting results
20

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
as if the test had been conducted at 64C. These values are based on average results of mixes from 8 different aggregate sources so individual results may vary. The data indicate that basically all 25 mm mixes could be designed at a maximum of 50 gyrations if the APA rutting test is conducted at 50C.
Moisture Susceptibility
There has also been some dissatisfaction throughout the industry with the moisture susceptibility test developed for implementation of the Superpave mix design system. AASHTO T 283 has been criticized since its inception as not being a good predictor of field performance it that aggregates with a past history of poor performance may pass the test while aggregates with a good performance history may not yield acceptable results. The moisture sensitivity test procedure used by GDOT is believed to be more severe (and more representative of actual field conditions) than the current procedure recommended by AASHTO. For that reason, GDOT kept the modified Lottman procedure as described in GDT-66 rather than switching to AASHTO T 283 when the Superpave mix criteria were implemented.
There is a problem, however, with the current GDT procedure. For the Superpave mix design system, six-inch (150 mm) samples are used so that larger aggregate used in base courses could be added to the aggregate batch. Due to the limitations of the size of Marshall specimens, any aggregate larger than 3/4 inch was removed and replaced with 3/4 inch aggregate. Therefore, the size of Superpave samples more nearly reflects the actual materials that will be used in mixture production and construction. The problem with using Superpave samples for GDT-66 is that the test was developed using 4 inch diameter samples and there has been no correlation of the difference in test results between the 4 inch and 6 inch samples. This has resulted in the current mixes being designed with the Superpave gyratory compactor, but samples for moisture susceptibility testing have been compacted with the Marshall hammer. This not only requires two separate pieces of equipment for mixture compaction, but encourages inconsistency by combining two separate mix design systems that are not readily compatible. For that reason there was a need to see if there was a correlation between the 4 inch diameter samples compacted with the Marshall hammer and the 6 inch diameter samples compacted with the gyratory compactor.
Moisture susceptibility testing was performed according to GDT-66 which is the moisture susceptibility test procedure Georgia has been using since 1980. The procedure was originally based on Marshall mixtures and sample sizes (2.5 inches high x 4 inches in diameter) and differs from the procedure described in AASHTO T 283 as follows:
Air voids are controlled at 7.0 1.0 percent. Vacuum saturation is for 30 minutes. The percent saturation is determined for
information only and is not used to determine how long to saturate the samples. The loading rate is 0.065 inches/minute rather than 2 inches /minute used in the
AASHTO procedure.
21

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
A minimum average tensile strength of 60 psi is required. The minimum acceptable TSR is 0.80, except that a TSR as low as 0.70 may be
acceptable if all individual strength values exceed 100 psi.
The original test data for developing the tensile strength criteria was based on Marshall samples 4 inches in diameter. Superpave specifications in AASHTO R 35 apparently use the same criteria that were developed for Marshall mixtures in the early 1980s. There has not been adequate research to determine how well the 4 inch Marshall samples correlate to the 6 inch Superpave samples that are being used in the current Superpave mix design method. For that reason samples in this study were made using both specimen diameters and were subjected to the test conditions described in GDT-66. The sample heights were also varied to match standard practice such that the 4 inch diameter Superpave specimens were compacted with a Marshall hammer to 2.5 inches in height and the 6 inch diameter gyratory samples were compacted to 95 mm high. The air voids in both sets of samples was controlled within 7.0 1.0 percent.
Figure 15 shows a scatterplot of the test results and the best fit line based on a regression analysis. The regression shows that a minimum tensile strength retained (TSR) of 80 percent that is currently used would require a minimum TSR of 98 percent for the 6 inch Superpave gyratory samples. However, there was a lot of variability in the test data and no significant correlation was found at the 5 percent significance level.

Scatterplot of TSR6 vs TSR4
110

100

90

TSR6

80
Regression Fit, TSR6 = 108.8 0.2505TSR4
70

60

50

60

70

80

90 100 110 120 130 140

TSR4

FIGURE 15. Comparison of 4" and 6" TSR Results

A regression analysis showed that the TSR4 results can only explain about 19 percent of the variation in TSR6 results. The relationship of TSR4 and TSR6 becomes significant at
22

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
the 10 percent significance level, but is not statistically significant enough to assign a relationship between TSR4 and TSR6. Based on these results, one cannot directly specify parameters for 6 inch diameter gyratory samples based on 4 inch Marshall-sized samples.Table 9 shows that aggregate source is the most significant factor affecting TSR results. This was not surprising because it has long been known that some aggregates are hydrophobic while other aggregate sources are hydrophyllic. The water absorption tendency often has a direct relationship to the moisture susceptibility of asphalt mixtures from these aggregate sources.

An Analysis of Variance (ANOVA) of the 4 inch and 6 inch diameter test results (Table 10) also showed that aggregate source is the most significant factor in the experiment. The aggregate source factors in Table 2, S1, S2, and S3 represent two sources each so that S1S2S3 = 23 or 8 sources. Likewise the gyration factors G1 and G2 represent two levels each so that G1G2 = 22 or 4 gyration levels. Aggregate source contributes about 44 percent of the total variability in test results for the 6 inch diameter samples. Aggregate source was also found to contribute 67 percent of the variability in the 4 inch diameter Marshall samples. However, the ANOVA also shows that the interaction of aggregate source and gyration level for both sample sizes is significant and cannot be separated. The interaction of these two variables contributes over 40 percent of the total variability for the 6 inch gyratory samples and 27 percent of the total variability in the 4 inch diameter samples.

TABLE 9. Regression Analysis for 6" Diameter TSR Results
General Linear Model: TSR6 versus S1, S2, S3, G1, G2

Factor S1 S2 S3 G1 G2

Type fixed fixed fixed fixed fixed

Levels 2 2 2 2 2

Values -1, 1 -1, 1 -1, 1 -1, 1 -1, 1

Analysis of Variance for TSR6, using Adjusted SS for Tests

Source DF

S1

1

S2

1

S3

1

S1*S2

1

S1*S3

1

S2*S3

1

S1*S2*S3 1

G1

1

G2

1

S1*G1

1

S2*G1

1

Error

Seq SS 93.61
405.02 363.86 354.38
0.00 256.80
1345.06
49.35 91.68 188.38 86.03
4 Total

Adj SS Adj MS

93.61 93.61

405.02 405.02

363.86 363.86

354.38 354.38

0.00

0.00

256.80 256.80

1345.06 1345.06

49.35 49.35

91.68 91.68

188.38 188.38

86.03 86.03

88.53 88.53

15 3322.68

F 4.23 18.30 16.44 16.01 0.00 11.60
60.77
2.23 4.14 8.51 3.89 22.13

P 0.109 0.013 0.015 0.016 0.996 0.027
0.001 = G1*G2
0.210 0.112 0.043 0.120

S = 4.70459 R-Sq = 97.34% R-Sq(adj) = 90.01%

23

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

Source S1 S2 S3
S1*S2 S1*S3 S2*S3 Subtotal
S1*S2*S3=G1*G2
G1 G2 Subtotal
S1*G1 S2*G1 Error Total

TABLE 10. ANOVA Table for TSR6

DF

SeqSS

F

P-value

1.0

93.6

4.23

0.109

1.0

405.0

18.3

0.013

1.0

363.9

16.44

0.015

1.0

354.4

16.01

0.016

1.0

0.0

0

0.996

1.0

256.8

11.6

0.027

6.0

1473.7

Source =

%Contribution 44.4%

1.0

1345.1

60.77

0.001

40.5%

1.0

49.4

2.23

0.210

1.0

91.7

4.14

0.112

2.0

141.0

Gyration=

4.2%

1.0

188.4

8.51

0.043

1.0

86.0

3.89

0.120

4.0

88.5

15.0

3322.7

R-Squared= 97.34%

In addition to TSR results, the individual conditioned tensile strength values were compared for both 4 inch and 6 inch specimens to see if a relationship existed. The low R-square value and the P-value of 0.084 (Table 11) indicated there is not a significant relationship.

TABLE 11. Tensile Strength Relationship of 4" vs 6" Test Results
Regression Analysis: TS6 versus TS4
The regression equation is TS6 = 118 - 0.335 TS4

Predictor

Coef SE Coef

T

P

Constant 117.59 12.27 9.58 0.000

TS4

-0.3353 0.1894 -1.77 0.084

S = 21.0358 R-Sq = 6.7% R-Sq(adj) = 4.5%
Permeability
Permeability testing was performed on 120 samples (40 sets 3 replicates). One-third of the samples had target air voids of 7 percent, one-third had target air voids of 9 percent, and one-third had target air voids of 11 percent. Each specimen was tested three times (for a total of 360 samples) and the results for each set averaged. Three test conditions

24

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
were also used to see if there is a significant difference in permeability results based on the way samples are prepared. The first time, the samples were tested uncut, the second time one face was sawed as per the ASTM PS 129 provisional method, and the third time both faces were cut as described in Florida DOT FM 5-565 permeability test procedure.

The density of compacted specimens is usually higher in the middle portion of the compacted sample since the mixture begins to cool from the top and bottom during the compaction process and there are some edge effects around the sides of the sample due to mold confinement during compaction. The three methods described above were performed to evaluate an assumption that sawing of specimens was unnecessary. Since water has to drain through the center portion of the compacted sample in each case, it would seem immaterial whether the ends were removed. A regression analysis of test results supported this assumption by showing that the test method was not significant with a P-value of 0.533. There was a lot of variability in the test results as indicated in the R2 value of 39%.

The interaction between source and gyration level shown in Table 12 indicates that 60 gyrations resulted in the lowest average permeability when all aggregate sources was considered. An ANOVA, Table 13, with air voids as a covariate shows that the interactions of source and gyration level was significant in affecting permeability results.

TABLE 12. Permeability Results Based on Interaction Between Aggregate Source and Gyration Level

Source\Gyration 1-Barin 2-Kennesaw 3-Lithia Sp 4-Norcross 5-Stockbridge 6- Mt View 7- Camak 8- Ruby
Avg.

35 103.98 75.39 464.55 118.48 171.30 253.77 154.37 494.81 229.58

60 136.81 42.74 169.71 77.40 76.49 44.79 74.20 264.34 110.81

85 391.42 49.04 180.20 83.76 307.74 30.47 56.95 460.51 195.01

110 205.26 49.62 202.92 66.99 268.47 33.64
17.58**
128.00 136.41

** Per Table 12 the best combination to minimize permeability is Camak aggregate at 110 gyrations. The combinations (6, 85), (6, 110), (2, 60) and (6, 60) are nearly as optimal in the order listed.

Although the test method was found to be not significant, an analysis based on the least squares (Table 14) shows a trend that Bottom Cut gives the least variability followed by Uncut.

25

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

TABLE 13. ANOVA for Permeability Using Adjusted SS for Tests

Source

DF Seq SS Adj SS Adj MS

F

P

S1 S2 S3 TM S1*S2 S1*S3 S1*TM S2*S3 S2*TM S3*TM S1*S2*S3 S1*S2*TM S1*S3*TM S2*S3*TM S1*S2*S3*TM G1 G2 TM*G1 TM*G2 S1*G1 S1*G2 S2*G1 S2*G2 S3*G1 S3*G2 S1*TM*G1 S1*TM*G2 S2*TM*G1 S2*TM*G2 S3*TM*G1 S3*TM*G2
Error

1 15370 15452 15452

1 23966 24043 24043

1 405712 405035 405035

2 17325 13832 6916

1 44899 48769 48769

1 356079 352275 352275

2

7731 7611 3805

1 42316 42584 42584

2

612

902

451

2

4603 4729 2365

1 19050 19298 19298

2 28018 25388 12694

2

2102 2273 1136

2 10408 10633 5317

2

2715 2598 1299

1 33998 31960 31960

1 137663 133769 133769

2

1119 1880

940

2 14863 16170 8085

1 165894 158022 158022

1 34350 37784 37784

1 186152 190338 190338

1 21728 20689 20689

1 50407 51520 51520

1

6708 5996 5996

2 30582 31149 15575

2

1592 1696

848

2

136

85

43

2 12534 13150 6575

2 14455 14810 7405

2

9924 9924 4962

96 394371 394371

Total

143 2097382

3.76 5.85 98.60 1.68 11.87 85.75 0.93 10.37 0.11 0.58 4.70 3.09 0.28 1.29 0.32 7.78 32.56 0.23 1.97 38.47 9.20 46.33 5.04 12.54 1.46 3.79 0.21 0.01 1.60 1.80 1.21 4108

0.055 0.017 0.000 0.191 0.001 0.000 0.400 0.002 0.896 0.564 0.033 0.050 0.759 0.279 0.730 0.006 0.000 0.796 0.145 0.000 0.003 0.000 0.027 0.001 0.230 0.026 0.814 0.990 0.207 0.170 0.303

S = 64.0940 R-Sq = 81.20% R-Sq(adj) = 71.99%

TABLE 14. Analysis of Test Methods Using Least Squares

Test Method

Seq SS

Rank

Uncut

25259

2

Bottom Cut

22734

1

Both Ends Cut

27468

3

An analysis of the importance of air voids to permeability results was conducted and the Pearson correlation of air voids and permeability resulted in a P-Value = 0.000 which shows the air void level in samples is highly significant in determining permeability

26

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
results. From this analysis, the air void level in samples should be closely controlled since percent air voids has a significant effect on permeability results. However, the test method does not have a significant effect and the practice of sawing samples on one, or both, sides is unnecessary.

Fatigue

Fatigue testing was performed according to AASHTO TP 8 on samples that had been compacted to 6.0 1.0 percent air voids. Strain levels of 250 m and 500 m were used and three replicates were made for each strain level per mix design. Fatigue failure was defined as the number of cycles at which the strain level decreases to 50 percent of the initial strain value.

As shown in Table 15, the gyration factor was not found to be significant at the 5 percent significance level (P-value = 0.095). However, aggregate source and strain level, as well as the interaction between source and strain, were found to significantly affect the test results. Gyration level does appear to be significant when the average cycles to failure are considered.

TABLE 15. ANOVA for Fatigue Evaluation
General Linear Model: y versus Source, Gyr, Strain

Factor Source
Gyr Strain

Type fixed
fixed fixed

Levels 8
4 2

Values Barin-Columbus, Camak, Kennesaw, Lithia Springs, Mt. View, Norcross, Ruby, Stockbridge 35, 60, 85, 110 250, 500

Analysis of Variance for y, using Adjusted SS for Tests

Source Source Gyr Strain Source*Strain

DF

Seq SS

Adj SS

Adj MS

F

P

7 2.81994E+13 2.78869E+13 3.98385E+12 8.68 0.000

3 7.99972E+12 3.01278E+12 1.00426E+12 2.19 0.095

1 5.01490E+13 5.24607E+13 5.24607E+13 114.26 0.000

7 1.70510E+13 1.70510E+13 2.43586E+12 5.31 0.000

Error

92 4.22414E+13 4.22414E+13 4.59145E+11

Total

110 1.45641E+14

S = 677603 R-Sq = 71.00% R-Sq(adj) = 65.32%
Although the experiment was conducted with gyration level as one of the main factors, the end result of changing the gyration level is that the optimum asphalt content is changed inversely. The effect of gyration level on average optimum asphalt content for each mix type is given in Table 16.

27

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

TABLE 16 Effect of gyration level on optimum asphalt content

No. of Gyrations
35 60 85 110

Avg. Opt. AC (%) 25 mm Mix
4.9 4.3 4.0 3.8

Avg. Opt. AC (%)
12.5 mm Mix 5.7 5.2 4.9 4.6

In Table 17 one can see that the test data can be grouped into two sets with the results for 35 and 60 gyrations being similar and the results for 85 and 110 gyrations being similar. Since all specimens were compacted to 6.0 1.0 % air voids, the difference in gyration level reflects higher optimum asphalt contents for lower gyration levels. Therefore, one would expect the lower gyrations to result in higher fatigue life. The average cycles to failure for 85 and 110 gyrations were only about half of the average values for 35 and 60 gyrations. Based on this analysis, there is a definite trend that lower gyration levels (and higher asphalt contents) will increase the fatigue life of the pavement.

TABLE 17. Comparison of Average Fatigue Cycles to Failure

Gyrations 35 60 85 110

250 2,305,467 1,514,725 792,721 847,496

500 79,564 57,101 40,562 30,104

Based on the apparent grouping in Table 17, the fatigue results for all mixes were grouped accordingly as shown in Table 18. One can see that reducing the strain at the bottom of the asphalt structural layer is critical to extending the life of asphalt pavements. This is the concept of "perpetual pavements" where the asphalt structural layers may last indefinitely and only periodic maintenance of the surface course is needed due to wear and environmental effects. The average fatigue life of various mixes at 250 was approximately 25 times that for mixes tested at 500 .
The average fatigue life of various mixes at the same strain level for 35 and 60 gyrations was approximately 200 percent of the average fatigue life of mixes compacted to 85 and 110 gyrations. Although the average results were based on various aggregate sources rather than the same mix, and a one-half factorial experimental design was used, the results indicate a trend that mixes with 35-60 gyrations could possibly last twice as long as mixes compacted with 85-110 gyrations.

28

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

TABLE 18. Fatigue Life - Cycles to Failure

Aggregate Source Barin-Columbus
Kennesaw

35 & 60
250 1,101,450 931,030 1,356,470 903,450 917,840 578,370 1,935,520 675,240 2,315,300 105,230 1,400,500 586,260

Lithia Springs

Norcross

Stockbridge

4,956,860 1,507,350 1,417,970 2,762,800 995,440 3,417,210

Mt. View

Camak 6,857,480

Gyrations & Strain Level

35 & 60 500

85 & 110 250

58,390 35,670 46,190 24,610 46,890 58,820 32,930 18,120 23,990 28,200 6,430 11,650

933,070 1,247,770
441,250 534,080 784,530 499,310 336,870 807,410 213,730 199,060 281,290

35,830 65,530
86,470 27,630 50,510 148,630 245,900 164,020 155,290 95,490 93,220

499,310 336,870 807,410 213,730 199,060 281,290 933,010 1,247,770 1,071,070 5,036,150 993,480 992,300

Ruby

1,967,260 3,027,620

85 & 110 500 37,320 42,860 41,240 39,140 33,210 16,000 33,350 23,270 22,150 16,770 47,260 25,220
35,680 19,710 28,170 16,000 44,680 33,840
36,820 91,110 75,060 35,020 24,000

Avg.

1,891,269

67,844

821,297 35,560

29

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
CONCLUSIONS
The following conclusions are made based on Phase 1 and Phase 2 of the research study. Phase 1 consisted of an evaluation of field performance of Marshall and Superpave mixtures and Phase 2 was a laboratory study.
Phase 1
For Phase 1, an evaluation was made of 16 Marshall projects and 16 Superpave projects placed about the same time and serving under approximately the same traffic conditions. Conclusions were as follows:
1. Both Marshall and Superpave mixtures were generally performing quite well with very little rutting and cracking after a period of about 4-6 years.
2. Rutting resistance as well as cracking resistance appears to be similar for both Marshall and Superpave mixtures. However, Superpave projects were an average of 1.4 years younger than Marshall projects. Superpave projects may experience slightly more rutting and cracking by the time they reach the same service life as Marshall projects.
3. Geographical location had little influence on rutting and cracking. 4. None of the Marshall projects had polymer-modified asphalt while five (31
percent) of the Superpave projects used modified asphalt. Superpave projects showed a slight trend of reduced cracking with the use of polymer modified asphalt. 5. Compared to Marshall projects, Superpave projects showed a slight trend of reduced cracking with an increase in asphalt content. 6. Rutting results were very similar for projects that used polymer-modified asphalt as compared to the standard paving grade asphalt. However, polymer-modified asphalt projects had almost four times as much traffic volume. 7. The average asphalt content for Marshall projects was 0.34 percent higher than the average for Superpave projects. 8. Roadway air voids determined immediately after construction averaged 7.3 percent for Superpave projects and values were as high as 9.7 percent. Therefore, Superpave mixtures may have been more permeable than the Marshall mixtures which averaged only 6.1 percent air voids with values as high as 8.3 percent. 9. It is likely that Superpave mixtures will not reach the design air voids of 4.0 percent during the life of the pavement. After nearly five years, the average air voids measured in the wheelpaths was 5.7 percent for Superpave projects while the air voids of comparable Marshall projects averaged 3.8 percent. 10. Based on field comparisons, the Superpave mixture specifications could be changed to increase asphalt content and VMA properties without adversely affecting rutting resistance. (As a result of these Phase 1 findings, GDOT increased the minimum VMA of all Superpave mixtures by 1% effective February, 2006.)
30

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
11. Reducing the gyratory compaction effort alone in order to increase asphalt binder content without changing other mixture properties for Superpave mixtures is not a viable option. A better design approach may be to combine the effects of revising the gyratory compaction effort with revisions to volumetric requirements such as percent VMA. These changes may require the use of a finer gradation in order to improve mixture volumetric properties.
Phase 2
Phase 2 represents results of laboratory testing using eight Georgia aggregate sources. For each source, mixture volumetric and performance testing was conducted to examine the effect of different gyration levels. Samples were compacted to standard test air void criteria so that the increase in gyration level amounted to a reduction in optimum asphalt content. Phase 2 conclusions were as follows:
1. For the laboratory mixes used in this study, 66 gyrations gave density results similar to those found in field cores. This gyration level also closely matched the locking point (LP 3) average of 69 gyrations and the current GDOT requirement of 65 gyrations.
2. When the APA test temperature for HMA Base mixtures was reduced to 50C, it resulted in asphalt contents as much as 0.75 percent higher than traditionally used in the past without increasing the rut depth.
3. Moisture susceptibility (TSR) results for 4 inch diameter samples could only explain 19 percent of the variation in 6 inch diameter gyratory samples. The comparison was not practically good enough to establish a correlation factor.
4. Aggregate source was the most significant factor affecting moisture susceptibility results. However, there was some interaction between aggregate source and gyration level that could not be separated. Aggregate source contributes 44 percent of the total variability for 6 inch samples and 67 percent of the variability for 4 inch samples.
5. Air voids contribute significantly to permeability results. An analysis with air voids as a covariate, shows that aggregate source and gyration level become significant in affecting permeability results. When aggregate source and gyration level was considered, 60 gyrations resulted in the lowest permeability.
6. An analysis of test methods showed there was no significant difference between uncut, one side cut, and both sides cut for permeability specimens.
7. Aggregate source and strain level were the most significant variables affecting fatigue results. However, samples at 85 and 110 gyrations had only about half of the fatigue life of samples with optimum asphalt content selected at 35 and 60 gyrations.
8. Fatigue life was approximately 25 times longer when the strain level was reduced from 500 to 250 .
31

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
RECOMMENDATIONS
As a result of this study, the following recommendations are made:
1. Since lab densities with an average of 66 gyrations gave density results similar to those found in field cores, GDOT should continue to use 65 gyrations as the NDesign level for Superpave mixtures. The average locking point for mixes in this study was found to be 69 gyrations. This also closely matches GDOT specification requirements and supports the use of 65 gyrations. Field adjustments for asphalt content may need to be made based on traffic level and seasonal effects (such as high summer temperatures).
2. Use polymer modified asphalt for high traffic volume projects to improve resistance to rutting and cracking susceptibility.
3. The APA test temperature for HMA Base (25 mm) mixtures should represent the maximum pavement temperature at the depth of the base layer. (As a result of this study, GDOT began in February, 2006 to require 25 mm Superpave mixtures to be tested at 120F (49C.))
4. The practice of cutting off one (or both) ends of specimens for permeability testing is unnecessary since the method is not significantly different from uncut specimens. Therefore, sawing of specimens should be eliminated.
5. In order to extend the fatigue life of asphalt pavements, the strain level at the bottom of the structural asphalt layer should be reduced by increasing the thickness of the asphalt pavement structure.
ACKNOWLEDGEMENTS
The authors wish to thank the Georgia Department of Transportation for its support in sponsoring this study. GDOT District Maintenance personnel were instrumental in helping provide traffic control for coring operations. GDOT Materials personnel helped in locating some of the projects selected for this study as well as in coordinating project coring. Thanks are also extended to Dr. Saeed Maghsoodloo, Statistician, for conducting the factorial design experiment and the statistical analysis of test results.
REFERENCES
1. Brown, E. R.,and M.S. Buchanan, "Superpave Gyratory Compaction Guidelines," Research Results Digest 237, (NCHRP 9-9), NCHRP/TRB/NRC, Washington, DC, 1999.
2. Prowell, Brian D., and E. Ray Brown, "Verification of Gyration Levels in the NDesign Table," NCHRP 9-9 (1) Final Report, TRB, Washington, DC, 2006.
32

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
3. Huber, Gerald A., "Weather Database for the Superpave Mix Design System." Strategic Highway Research Program, Report No. SHRP-A-648A. Washington, DC. 1994.
4. Solaimanian, Mansour, and Thomas W. Kennedy, "Predicting Maximum Pavement Surface Temperature Using Maximum Air Temperature and Hourly Solar Radiation." TRR 1417, TRB/NRC. Washington, DC. 1993, pp. 1-11.
5. Watson, Donald E., Jingna Zhang, and R. Buzz Powell, "Analysis of Temperature Data for the National Center for Asphalt Technology Test Track," TRR 1891. TRB/NAS, Washington, DC. 2004. pp. 68-75.
33

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
APPENDIX
34

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Table 1-A. Selected Marshall Projects

No.

District Age, Yrs.

Project

County

Description

Traffic, Contractor Mix Type AADT

AC Type

1

7

1.8 km on SR 14 from SR 74 to

6

SAMA-14(58)01

Fulton

N. of Strickland

APAC

12.5

28,400

67-22

2

7

3

3

4

3

5

6

6

6

8.9 km on SR 314 from Fayette

6

STP-9009(13)01

Clayton

County line to SR 139

APAC

12.5

21,020

67-22

15.4 km on SR 109 from Troup

County line to SR 18 in

6

STP-5-1(28)01 Meriwether

Greenville

APAC

12.5

3,047

67-22

Harris-

on SR 1/US 27 from I-185 to

7

STP-11-1(50)01 Muscogee

Moon Rd.

APAC

12.5

25,651

67-22

0.8 mi on SR 1/US 27 from S. of

Kingston St. to N. of Gordon

6

STPN-17-3(60)01 Walker

Lee Rd.

Dalton Paving 12.5

13,352

67-22

on SR 1/US 27 from Maple Dr.

to N of Penn Bridge Rd. (CR

6

SAMA-1(225)01 Chattooga

170)

C.W. Matthews 12.5

24,896

67-22

7

6

8

7

9

7

10

1

11

3

12

7

13

2

14

2

15

3

16

3

Avg. Age

1.4 mi on SR 293 from Kingston

7

MLP-293(26)01

Floyd

Ave. to E. Rome Connector Spriggs Paving 12.5

6,386

67-22

9.1 km on SR 92 from Fulton

Riverdale

6

MASTP-MS(245)01 Douglas

County line to Flowers Dr.

Paving

12.5

15,000

67-22

1.9 km on SR 3/US 41

(Northside Parkway) from N. of

Beechwood Dr. to Cobb County

7

STP-1-5(57)01

Fulton

line

APAC

12.5

85,400

67-22

8.2 km on SR 8/US 29 from

Dekalb County line to SR 378 in

6

STP-3-2(77)01

Gwinnett

Lilburn

E.R. Snell

12.5

31,882

67-22

7.3 km on SR 138 from E. of SR

42/US 23 to Rockdale County

7

STP-35-1(37)01

Henry

line

Couch

12.5

27,214

67-22

5.8 km on SR 138 from SR 85 Riverdale

7

STP-168-1(19)01 Clayton

in Riverdale to SR 3

Paving

12.5

28,286

67-22

on SR 24 from Burke Co. line to

5

STP-937(9)01

Jefferson

Mulberry St. in Louisville

Knox-Rivers

12.5

1,844

67-22

3.1 km on SR 10/US 78 from

6

SAMA-10(114)01 McDuffie

I-20 to SR 17 in Thomson

Knox-Rivers

12.5

15,000

67-22

19.8 km on SR 7/US 341 from

Houston- US 41 in Perry to SR 49 in Ft.

Douglas

6

STP-1-3(32)01

Peach

Valley

Asphalt

12.5

7,500

67-22

6.8 mi on SR 57 from W. of

Jones-

Jones County line to E. of

Douglas

4

FLF-540(4)01

Twiggs

Wilkinson County line

Asphalt

12.5

5047

67-22

6.1

Avg. ADT 21,245

35

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Table 2-A. Selected Superpave Projects

No. District Age, Yrs.

Project

County

Description

Contractor

1

7

6

SAMA-279(11)01 Fulton

3.3 mi. on SR 279 from S. of Flat Shoals Rd. to SR 14/US 29

APAC

2

7

5.6 mi. of intersection improvements on

SR 3/US 19/41 in Jonesboro, Lovejoy

3

STP-1-4(64)01 Clayton

and Riverdale

APAC

3

3

2.8 km on SR 22 Spur from SR 1/US 27

5

NH-215-1(3)01 Muscogee

to E. of Tate Rd. in Columbus

APAC

4

3

1.9 km on SR 1/US 27 from SR 520 to

5

NH-11-1(54)01 Muscogee

SR 22 Spur (13th St.)

Couch

5

7

11.2 km on SR 360 from Paulding

5

STP-9023(6)01

Cobb

County line to SR 176

Shepherd

Mix Type 12.5
12.5 12.5 12.5 12.5

6

6

7

7

8

6

9

1

16.3 km on SR 101 from N. of CR 313

5

STP-167-1(15)01 Floyd-Polk

to SR 20

C.W.Matthews 12.5

6.2 mi. on New Wooten Rd. from Capps

4

STP-212-1(3)01

Fulton

Ferry Rd. to Cochran Mill Rd.

E.R. Snell

12.5

9.2 mi. on SR 1/US 27 from SR 20 in

4

SAMA-1(227)01

Floyd

Rome to N. of SR 156 in Armuchee Spriggs Paving 12.5

8.3 mi. on Peachtree Ind. Blvd. (SR

141) from W. of Rogers Bridge Rd. to

4 STP-190-1(9)01/02 Gwinnett

S. of Pinecrest Rd.

E.R. Snell

12.5

10

7

11

3

12

3

13

2

14

2

15

2

16

2

1.0 mi. on SR 120 (Old Milton

Parkway/State Bridge Rd.) from E. of

Park Bridge Pkwy. To W. of Amy

5

STP-189-1(25)01 Fulton

Francis Lane

APAC

12.5

8.1 km on SR 42 from S. of SR 138 to

5

STP-37-2(68)01

Henry

Clayton Co. line

Couch

12.5

Henry-

on SR 155 from SR 42/US 23 in

7

STP-165-1(63)01 Spalding

McDonough to SR 16 in Griffin

Couch

12.5

on SR 12 from W. of CR 29 (Reid

4

STP-46-2(20)01 Greene Duvall Rd.) to E. of Apalachee Ave. Knox-Rivers 12.5

12.6 km on SR 22 from SR 12/US 278

5

STP-785(16)01 Taliaferro

in Crawfordville to SR 44

APAC

12.5

On SR 11 from S. of Williams St.to N.

Douglas

4

SAMA-11(225)01 Baldwin

of Honeysuckle Dr.

Asphalt

12.5

On SR 212 from SR 49/US 129 to

Douglas

4

SAMA-212(31)01 Putnam

Baldwin County line

Asphalt

12.5

Avg. Age 4.7

Design Level
C
D B C B B B B
C
C B B B A B B Avg. AADT

Traffic, AADT 34,197
60,639 22,144 19,554 11,185 7,212
16,924
55,986
48,376 18,135 15,000 1,932 1,131 7,251 1,635 21,420

No. of Gyrations
96
96 75 100 75 75 86 75
100
100 75 86 86 50 75 75

AC Type 76-22
76-22 67-22 76-22 67-22 67-22 67-22 67-22
76-22
76-22 67-22 67-22 67-22 67-22 67-22 67-22

36

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu
Table 3-A. Summary of Field Data for Marshall Projects

No. District

Project

Age

1

7

SAMA-14(58)01

6

2

7

STP-9009(13)01

6

3

3

STP-5-1(28)01

6

4

3

STP-11-1(50)01

7

5

6

STPN-17-3(60)01

6

6

6

SAMA-1(225)01

6

7

6

MLP-293(26)01

7

8

7 MASTP-MS(245)01

6

9

7

STP-1-5(57)01

7

10

1

STP-3-2(77)01

6

11

3

STP-35-1(37)01

7

12

7

STP-168-1(19)01

7

13

2

STP-937(9)01

5

14

2

SAMA-10(114)01

6

15

3

STP-1-3(32)01

6

16

3

FLF-540(4)01

4

Averages 6.1

% AC 5.22 5.12 5.21 5.18 5.23 5.09 5.05 5.38 5.41 5.41 5.21 5.26 5.46 5.25 5.46 5.49

Pavement Depth (in.)

Surface Total

1.50

11.75

1.75

10.00

2.38

9.00

1.75

10.50

1.84

9.00

1.56

9.00

1.44

6.72

1.75

10.50

1.33

9.13

1.56

9.75

1.50

10.00

1.53

11.00

1.56

13.50

1.29

8.46

1.44

11.00

1.69

7.00

Rut Depth (in.) IWP OWP Avg. 0.00 0.00 0.00 0.06 0.06 0.06 0.00 0.00 0.00 0.00 0.00 0.00 0.12 0.06 0.09 0.19 0.12 0.16 0.12 0.06 0.09 0.00 0.00 0.00 0.06 0.00 0.03 0.06 0.00 0.03 0.12 0.12 0.12 0.12 0.12 0.12 0.00 0.00 0.00 0.06 0.06 0.06 0.19 0.12 0.16 0.12 0.12 0.12

Cracking

Severity Intensity, Level LF/SF (%)

2.00

3.40

1.00

11.20

2.00

12.30

1.00

0.70

1.00

3.20

0.00

0.00

1.00

1.40

1.00

7.30

1.00

14.30

1.00

18.80

1.00

5.50

0.00

0.00

1.00

9.40

1.00

16.60

0.00

0.00

0.00

0.00

Wheelpath (After Traffic)

Gmm Gmb Air Voids (%)

2.485 2.355

5.2

2.470 2.336

5.4

2.460 2.352

4.4

2.448 2.385

2.6

2.514 2.453

2.4

2.556 2.475

3.2

2.446 2.349

4.0

2.438 2.358

3.3

2.519 2.457

2.5

2.509 2.449

2.4

2.418 2.334

3.5

2.484 2.307

7.1

2.445 2.320

5.1

2.458 2.394

2.6

2.499 2.429

2.8

2.486 2.389

3.9

As-Built Air Voids (%)
8.3 5.1 6.7 6.8 6.4 6.2 4.4 6.1 5.2 6.6 5.7 5.4 5.8 6.1 6.9 6.3

Design VMA 16.3 16.3
16.8 17.4
17 16.8 16.1
17.4 17.4

5.28 1.62

9.77

0.08 0.05 0.06 0.88

6.51 2.477 2.384

3.8

=

1.37

6.13

16.83

0.9

0.51

37

Watson, D. E., J. Heartsill, J. Moore, D. Jared, and P. Wu

TABLE 4-A. Summary of Field Data for Superpave Projects

No. District

Project

Cracking

Design

Pavement Depth (in.) Rut Depth (in.) Severity Intensity,

Age Level Ndesign % AC Surface Total IWP OWP Avg. Level LF/SF (%)

Wheelpath (After Traffic) Gmm Gmb Air Voids (%)

As-Built Air Design Voids (%) VMA

1

7

SAMA-279(11)01 6

C

96 4.99 1.50

9.50 0.00 0.00 0.00 1.00

0.70

2.498 2.302

7.8

7.9

15.4

2

7

STP-1-4(64)01

3

D

96 4.89 1.50

12.00 0.13 0.13 0.13 0.00

0.00

2.465 2.304

6.5

6.3

14.8

3

3

NH-215-1(3)01

5

B

75 5.14 2.25

8.50 0.00 0.00 0.00 0.00

0.00

2.471 2.376

3.8

7.1

15.7

4

3

NH-11-1(54)01

5

C

100 4.98 1.75

10.50 0.00 0.00 0.00 0.00

0.00

2.525 2.302

8.8

9.7

14.9

5

7

STP-9023(6)01

5

B

75 4.68 1.50

12.00 0.06 0.06 0.06 2.00

3.70

2.597 2.423

6.7

9.1

14.7

6

6

STP-167-1(15)01 5

B

75 4.70 1.22

9.80 0.13 0.13 0.13 1.00

5.20

2.579 2.428

5.9

5.8

14.7

7

7

STP-212-1(3)01

4

B

86 4.62 1.84

8.25 0.00 0.00 0.00 0.00

0.00

2.452 2.259

7.9

8.8

14.2

8

6

SAMA-1(227)01

4

B

75 4.86 1.77

9.30 0.00 0.00 0.00 1.00

1.50

2.433 2.332

4.2

6.8

14.9

9

1

STP-190-1(9)01/02 4

C

100 4.79 1.72

11.50 0.19 0.19 0.19 0.00

0.00

2.452 2.403

2.0

6.9

13.7

10

7

STP-189-1(25)01 5

C

100 5.15 1.50

10.50 0.00 0.00 0.00 1.00

1.00

2.491 2.372

4.8

8.7

15.8

11

3

STP-37-2(68)01

5

B

75 4.63 1.69

16.00 0.00 0.00 0.00 1.00

5.75

2.451 2.389

2.5

5.9

14.1

12

3

STP-165-1(63)01 7

B

86 4.69 1.72

7.50 0.06 0.13 0.09 1.00

13.80 2.475 2.427

1.9

13

2

STP-46-2(20)01

4

B

86 4.64 1.20

11.75 0.00 0.00 0.00 1.00

2.20

2.500 2.297

8.1

5.4

14.3

14

2

STP-785(16)01

5

A

50 5.86 1.38

5.19 0.00 0.06 0.03 0.00

0.00

2.454 2.383

2.9

6.4

17

15

2

SAMA-11(225)01 4

B

75 5.17 1.25

6.00 0.06 0.13 0.09 1.00

0.70

2.512 2.284

9.1

7.8

15

16

2

SAMA-212(31)01 4

B

75 5.17 1.38

7.19 0.06 0.00 0.03 0.00

0.00

2.521 2.311

8.3

7.8

15

Averages 4.7

4.94

1.57

9.72

0.05 0.63

2.16

2.492 2.350

5.7

=

2.55

7.4

14.9

1.3

0.8

38