DEVELOPMENT OF MIX
DESIGN PROCEDURES FOR
GAP- GRADED ASPHALT-RUBBER
ASPHALT
CONCRETE
Final Report 524
Prepared by:
Anne Stonex
James M. Carusone
MACTEC Engineering and Consulting
3630 E. Wier Ave.
Phoenix, AZ 85040
November 2007
Prepared for:
Arizona Department of Transportation
206 South 17th Avenue
Phoenix, Arizona 85007
in cooperation with
U. S. Department of Transportation
Federal Highway Administration
The contents of the report reflect the views of the authors who are responsible for the facts and the
accuracy of the data presented herein. The contents do not necessarily reflect the official views or
policies of the Arizona Department of Transportation or the Federal Highway Administration. This
report does not constitute a standard, specification, or regulation. Trade or manufacturers’ names that
may appear herein are cited only because they are considered essential to the objectives of the report.
The U. S. Government and the State of Arizona do not endorse products or manufacturers.
Technical Report Documentation Page
1. Report No.
FHWA- AZ- 06- 524
2. Government Accession No.
3. Recipient's Catalog No.
5. Report Date
November 2007
4. Title and Subtitle
DEVELOPMENT OF MIX DESIGN PROCEDURES FOR GAP-GRADED
ASPHALT- RUBBER ASPHALT CONCRETE
6. Performing Organization Code
7. Author
Anne Stonex and James M. Carusone
8. Performing Organization Report No.
4975- 03- 3008 Final Report
10. Work Unit No.
9. Performing Organization Name and Address
MACTEC Engineering and Consulting, Inc.
3630 East Wier Avenue
Phoenix, Arizona 85040
11. Contract or Grant No.
SPR- PL- 1( 03) 524
13. Type of Report & Period Covered
FINAL REPORT
November 30, 2007
12. Sponsoring Agency Name and Address
Arizona Department of Transportation
206 S. 17th Avenue
Phoenix, Arizona 85007
Project Manger: Christ Dimitroplos
14. Sponsoring Agency Code
15. Supplementary Notes
Prepared in cooperation with the U. S. Department of Transportation, Federal Highway Administration
16. Abstract
A research project was conducted to identify and document current modifications to ARIZONA 815c ( 75- blow
Marshall method) used to develop gap- graded asphalt rubber asphalt concrete ( GG AR AC) mix designs, and to
develop and test improvements to provide a standard mix design method for use by contractors and consultants.
Based on field performance data provided by the Arizona Department of Transportation ( ADOT), the existing mix
design method was successful and should serve as the standard for comparison of proposed improvements.
Best practices were synthesized to develop proposed improvements. Three aggregate sources and two asphalt-rubber
( AR) binders were used for initial testing of the existing ( control) mix design method and of the proposed
changes. Rebound of compacted AR AC specimens was evaluated, as were Rice test results at 6% and 7% AR
binder by weight of mix. The composition of the AR binders ( rubber gradation and content) had more effect on
the results than which mix design method was used. Additional replicate testing was performed by MACTEC and
ADOT to confirm these findings. Changes to the AR AC mix design method consist primarily of making and
curing Rice specimens in the same manner as Marshall specimens, tighter temperature ranges for mixing and
compaction, incorporating Asphalt Institute calculations in a “ User’s Guide”, and improving presentation. An
ADOT construction project was used as an “ acid test” to pilot the proposed mix design method and provide
materials for a four- laboratory round robin to evaluate the precision of testing AR AC materials. The precision of
round robin testing appears very similar to that of conventional asphalt concrete mixtures based on data from
Proficiency Sample Programs of the AASHTO Materials Reference Laboratory and ADOT. The results indicate
that the mix design method developed can be used by qualified laboratories to provide suitable AR AC mix
designs.
17. Key Words
Asphalt- rubber, asphalt- rubber asphalt concrete,
AR AC, Gap- graded asphalt concrete mixtures,
Marshall mixture design, rubber- modified asphalt
concrete
18. Distribution Statement
Document is available to the U. S.
Public through the National
Technical Information Service,
Springfield, Virginia, 22161
19. Security Classification
Unclassified
20. Security Classification
Unclassified
21. No. of Pages
264
22. Price
23. Registrant's Seal
SI* ( MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS APPROXIMATE CONVERSIONS FROM SI UNITS
Symbol When You Know Multiply By To Find Symbol Symbol When You Know Multiply By To Find Symbol
LENGTH LENGTH
in inches 25.4 millimeters mm mm millimeters 0.039 inches in
ft feet 0.305 meters m m meters 3.28 feet ft
yd yards 0.914 meters m m meters 1.09 yards yd
mi miles 1.61 kilometers km km kilometers 0.621 miles mi
AREA AREA
in2 square inches 645.2 square millimeters mm2 mm2 Square millimeters 0.0016 square inches in2
ft2 square feet 0.093 square meters m2 m2 Square meters 10.764 square feet ft2
yd2 square yards 0.836 square meters m2 m2 Square meters 1.195 square yards yd2
ac acres 0.405 hectares ha ha hectares 2.47 acres ac
mi2 square miles 2.59 square kilometers km2 km2 Square kilometers 0.386 square miles mi2
VOLUME VOLUME
fl oz fluid ounces 29.57 milliliters mL mL milliliters 0.034 fluid ounces fl oz
gal gallons 3.785 liters L L liters 0.264 gallons gal
ft3 cubic feet 0.028 cubic meters m3 m3 Cubic meters 35.315 cubic feet ft3
yd3 cubic yards 0.765 cubic meters m3 m3 Cubic meters 1.308 cubic yards yd3
NOTE: Volumes greater than 1000L shall be shown in m3.
MASS MASS
oz ounces 28.35 grams g g grams 0.035 ounces oz
lb pounds 0.454 kilograms kg kg kilograms 2.205 pounds lb
T short tons ( 2000lb) 0.907 megagrams
( or “ metric ton”)
mg
( or “ t”)
mg megagrams
( or “ metric ton”)
1.102 short tons ( 2000lb) T
TEMPERATURE ( exact) TEMPERATURE ( exact)
º F Fahrenheit
temperature
5( F- 32)/ 9
or ( F- 32)/ 1.8
Celsius temperature º C º C Celsius temperature 1.8C + 32 Fahrenheit
temperature
º F
ILLUMINATION ILLUMINATION
fc foot candles 10.76 lux lx lx lux 0.0929 foot- candles fc
fl foot- Lamberts 3.426 candela/ m2 cd/ m2 cd/ m2 candela/ m2 0.2919 foot- Lamberts fl
FORCE AND PRESSURE OR STRESS FORCE AND PRESSURE OR STRESS
lbf poundforce 4.45 newtons N N newtons 0.225 poundforce lbf
lbf/ in2 poundforce per
square inch
6.89 kilopascals kPa kPa kilopascals 0.145 poundforce per
square inch
lbf/ in2
SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380
TABLE OF CONTENTS
Executive Summary ....................................................................................................... 1
1. Introduction................................................................................................................ 5
1.1. Organization of the Report .................................................................................. 7
2. Development of a Mix Design Procedure.................................................................. 9
2.1. Document Existing Modifications to Arizona 817 c ........................................... 9
2.2. Material Selection................................................................................................ 9
2.2.1. Aggregates ..................................................................................................... 9
2.2.2. Rubber............................................................................................................ 9
2.2.3. Asphalt Cement............................................................................................ l0
2.2.4. Asphalt Rubber Binders............................................................................... l0
2.3. Pilot AR- AC Mix Design- Control Mixes.......................................................... 12
2.3.1. Issues With CKC and Grey Mountain AR AC Control Mixes.................... 13
2.3.2. Salt River Control Mixes ............................................................................. 13
2.4. Version 1 to Existing ADOT AR- AC Mix Design Procedure........................... 16
2.4.1. List of Proposed Procedural Changes to AR AC Mix Design Method ....... 16
2.4.2. Mix Designs - Version 1 Mixes ................................................................... 17
2.4.3. Analysis of Rice Results at 6.0% and 7.0% AR Binder Content................. 19
2.4.4. AR- AC Rebound of Compacted Specimens................................................ 20
2.4.5. Round 2 Replicate Testing ADOT Central Lab and MACTEC. ................. 22
3. Round Robin Testing for Verification of Proposed AR- AC Mix Design Method.. 27
3.1. Project and Materials Selection ......................................................................... 27
3.2. Materials Designs .............................................................................................. 28
3.2.1. Asphalt- Rubber Binder Design.................................................................... 28
3.2.2. AR- AC Mix Design ..................................................................................... 28
3.3. Preparation of AR Binder Samples for Round Robin Testing........................... 29
3.4. Instructions and Distribution of Samples for Round Robin Testing ................. 30
3.5. Basics of Estimating Variability of Test Methods and
Acceptable Ranges of Test Results ................................................................... 30
3.5.1. Considerations Regarding Volumetric Calculations and Analysis.............. 32
3.6. Round Robin Test Results ................................................................................. 33
3.7. Additional Considerations ................................................................................. 38
3.7.1. Laboratory Technicians and Equipment ...................................................... 38
3.7.2. Field Performance........................................................................................ 39
3.7.3. Resistance to Moisture Damage................................................................... 40
3.7.4. Marshall Method for AR- AC....................................................................... 40
4. Conclusions.............................................................................................................. 41
Appendix A: Existing Modifications to ARIZ 815c Used for AR- AC Mix Designs until
2006 ( Version 5- 28- 03)............................................................................... 43
Appendix B: Initial Control Mix Design Data................................................................... 57
Appendix C: Initial Version 1 Mix Design Data Round 1............................................... 129
Appendix D: Rebound and Rice Data.............................................................................. 133
Appendix E: Round 2 Mix Design Data.......................................................................... 145
Appendix F: Big Bug Round Robin Preliminary Data and Analyses.............................. 171
Appendix G: Big Bug Round Robin Normalized Data and Analyses ............................. 211
Appendix H: ARIZ 832 Draft September 6, 2007 Marshall Mix Design Method for
AR- AC........................................................................................................ 237
References..................................................................................................................... .. 256
List of Tables
Table 1 Binder 1 Design Profile ................................................................................... 11
Table 2 Binder 1 Rubber Gradation, Percent Passing ( ARIZ 714)............................... 11
Table 3 Binder 2 Design Profile ................................................................................... 11
Table 4 Binder 2 Rubber Gradation, Percent Passing ( ARIZ 714)............................... 12
Table 5 Design Binder and Air Voids Contents, Salt River Control Mixes ................. 15
Table 6 Design Binder and Air Voids Contents, Salt River Version 1 Mixes.............. 17
Table 7 Original Big Bug AR Binder Design Profile ................................................... 28
Table 8 Big Bug Binder Rubber Gradation, Percent Passing ( ARIZ 714) ................... 28
Table 9 Big Bug AR Binder Design Profile – Updated for Round Robin Testing........ 29
Table 10 Compiled Round Robin Results for Aggregate Specific Gravity.................... 34
Table 11 Compiled Round Robin Results for Rice at 6.0% AR Binder Content ........... 35
Table 12 Multilaboratory Proficiency Sample Program Ranges for Gmm Results........ 35
Table 13 Within Laboratory Standard Deviation ( 1s) Ranges of Gsb Results ............... 36
Table 14 Preliminary AR Binder Content Selection....................................................... 37
Table 15 Normalized AR Binder Content Selection....................................................... 38
Table 16 Compiled CKC AR- AC Control Mix Design Data ......................................... 69
Table 17 Compiled Grey Mountain AR- AC Control Mix Design Data......................... 70
Table 18 Compiled Round 1 Salt River AR- AC Control Mix Design Data .................. 71
Table 19 Compiled CKC AR- AC Version 1 Mix Design Data.................................... 130
Table 20 Compiled Grey Mountain AR- AC Version 1 Mix Design Data.................... 131
Table 21 Compiled Round 1 Salt River AR- AC Version 1 Mix Design Data ............. 132
Table 22 Rebound Experiment Using 2000 g Weight First Round of Control
and Version 1 Designs ................................................................................ 134
Table 23 Rebound Experiment Using 2000 g Weight Repeats 1 and 2 of Control
and Version 1 Designs ................................................................................ 135
Table 24 Rebound Experiment Using 2000 g Weight Soufflé Mix.............................. 136
Table 25 Statistical Analysis of MACTEC’s Measured Rice Values ( Gmm)
for Salt River Aggregate at 6.0% and 7.0% AR Binder Contents .............. 137
Table 26 Single Factor ANOVA for Rice Data ........................................................... 138
Table 27 Rice Data Two- Way ANOVA for Relative Effects of Binder
and Design Method..................................................................................... 142
Table 28 Combined ADOT MACTEC Control Mix Data Rounds 1 and 2 Salt River
Aggregate with Binders 1 and 2, Source Data for Plots ............................. 146
Table 29 Combined ADOT MACTEC Version 1 Mix Data Rounds 1 and 2
Salt River Aggregate with Binders 1 and 2, Source Data for Plots ............ 147
Table 30 Legend Key for Plots ..................................................................................... 148
Table 31 One- Way Analysis of Variance Results Matrix............................................. 167
Table 32 Two- Way Analysis of Variance Results Matrix ADOT and MACTEC
( Rounds 1 and 2) B1 vs. B2, Control vs. Version 1 Mixes......................... 169
Table 33 Big Bug Round Robin Compiled Preliminary Round Robin Source Data
for Plots....................................................................................................... 177
Table 34 Duncan’s Multiple Range Test ...................................................................... 185
Table 35 Preliminary Statistical Analysis of Big Bug Round Robin Data
at 6.5% AR Binder Content ........................................................................ 186
Table 36 Preliminary Statistical Analysis of Big Bug Round Robin Data
at 7.5% AR Binder Content ........................................................................ 192
Table 37 Preliminary Statistical Analysis of Big Bug Round Robin Data
at 8.5% AR Binder Content ........................................................................ 198
Table 38 Summary of Duncan’s Multiple Comparisons of Mean Results
( Preliminary Data) ...................................................................................... 204
Table 39 Statistical Analysis of Bulk Specific Gravity of Marshall Specimens .......... 205
Table 40 Big Bug Round Robin Normalized Compiled Round Robin Source Data
for Plots....................................................................................................... 212
Table 41 Statistical Analysis of Normalized Big Bug Round Robin Data
at 6.5% AR Binder Content ........................................................................ 218
Table 42 Statistical Analysis of Normalized Big Bug Round Robin Data
at7.5% AR Binder Content ......................................................................... 224
Table 43 Statistical Analysis of Normalized Big Bug Round Robin Data
at 8.5% AR Binder Content ........................................................................ 228
Table 44 Summary of Duncan’s Multiple Comparisons
of Mean Normalized Results ...................................................................... 232
Table 45 Precision Calculations for Results of Big Bug Round Robin ........................ 233
Table 46 Comparison of Multilaboratory Precision of Test Results ........................... 236
List of Figures
Figure 1 2,000 gram Rebound “ Puck” and Dial Indicator ............................................. 21
Figure 2 Instructions for Big Bug Round Robin ............................................................ 31
Figure 3 CKC B1 Control Trial A Mix Design.............................................................. 58
Figure 4 CKC B2 Trial A Mod Mix Design .................................................................. 64
Figure 5 GM B1 Control Trial A Mix Design................................................................ 72
Figure 6 GM Trial B Crusher Fines Paramount Mix Design......................................... 78
Figure 7 GM B2 Control A Mix Design ........................................................................ 83
Figure 8 GM B2 Control Trial B Crusher Fines Mix Design ........................................ 88
Figure 9 Salt River B1C1 Mix Design ........................................................................... 93
Figure 10 Salt River B1C2 Mix Design ........................................................................... 99
Figure 11 Salt River B1C3 Mix Design ......................................................................... 105
Figure 12 Salt River B2C1 Mix Design ........................................................................ 111
Figure 13 Salt River B2C2 Mix Design ........................................................................ 117
Figure 14 Salt River B2C3 Mix Design ......................................................................... 123
Figure 15 MACTEC Rounds 1 & 2 and ADOT Round 2 Effective Binder Volume
Salt River B1 Control and Version 1 .......................................................... 149
Figure 16 MACTEC Rounds 1 & 2 and ADOT Round 2 Effective Binder Volume
Salt River Controls B1 & B2 ...................................................................... 150
Figure 17 MACTEC Rounds 1 & 2 and ADOT Round 2 VMA
Salt River Controls B1& B2 ....................................................................... 151
Figure 18 MACTEC Rounds 1 & 2 and ADOT Round 2 VMA
Salt River Version 1 B1 & B2 .................................................................... 152
Figure 19 MACTEC Rounds 1 & 2 and ADOT Round 2 VMA
Salt River B1 Control & Version 1............................................................. 153
Figure 20 MACTEC Rounds 1 & 2 and ADOT Round 2 VMA
Salt River B2 Control & Version 1............................................................. 154
Figure 21 MACTEC Rounds 1 & 2 and ADOT Round 2 VFA
Salt River B2 Control & Version 1............................................................. 155
Figure 22 MACTEC Rounds 1 & 2 and ADOT Round 2 VFA
Salt River Version 1 B1 & B2 .................................................................... 156
Figure 23 MACTEC Rounds 1 & 2 and ADOT Round 2
VFA Salt River Controls B1 & B2 ............................................................. 157
Figure 24 MACTEC Rounds 1 & 2 and ADOT Round 2
VFA Salt River B1 Control & Version 1.................................................... 158
Figure 25 MACTEC Rounds 1 & 2 and ADOT Round 2 Air Voids
Salt River B1 Control & Version 1............................................................. 159
Figure 26 MACTEC Rounds 1 & 2 and ADOT Round 2 Air Voids
Salt River B2 Control & Version 1............................................................. 160
Figure 27 MACTEC Rounds 1 & 2 and ADOT Round 2 Air Voids
Salt River Controls B1 & B2 ...................................................................... 161
Figure 28 MACTEC Rounds 1 & 2 and ADOT Round 2 Air Voids
Salt River Version 1 B1 & B2 .................................................................... 162
Figure 29 MACTEC Rounds 1 & 2 and ADOT Round 2 Stability
Salt River Version 1 B1 & B2 .................................................................... 163
Figure 30 MACTEC Rounds 1 & 2 and ADOT Round 2 Stability
Salt River Controls B1 & B2 ...................................................................... 164
Figure 31 MACTEC Rounds 1 & 2 and ADOT Round 2 Stability
Salt River B2 Control & Version 1............................................................. 165
Figure 32 MACTEC Rounds 1 & 2 and ADOT Round 2 Stability
Salt River B1 Control & Version 1............................................................. 166
Figure 33 Big Bug Version 2 Mix Design ..................................................................... 176
Figure 34 Preliminary Big Bug Effective Binder Volume............................................. 179
Figure 35 Preliminary Big Bug VMA............................................................................ 180
Figure 36 Preliminary Big Bug VFA ............................................................................. 181
Figure 37 Preliminary Big Bug Air Voids ..................................................................... 182
Figure 38 Preliminary Big Bug Stability........................................................................ 183
Figure 39 Preliminary Big Bug Flow ............................................................................. 184
Figure 40 Normalized Big Bug Effective Binder Volume............................................. 214
Figure 41 Normalized Big Bug VMA............................................................................ 215
Figure 42 Normalized Big Bug VFA ............................................................................. 216
Figure 43 Normalized Big Bug Air Voids ..................................................................... 217
1
EXECUTIVE SUMMARY
The purpose of this study was to develop a standard mix design method for the Arizona
Department of Transportation ( ADOT) gap- graded asphalt- rubber asphaltic concrete
( AR- AC) mixtures that can be used by contractors and consultants. The Department is
seeking to transfer AR- AC mix design responsibilities to industry, similar to the current
practice for standard Marshall and Superpave asphaltic concrete mixtures.
The first task was to review and document ADOT’s existing Marshall- based mix design
procedure for AR- AC, based on interviews with ADOT personnel, and a review of
ADOT’s AR- AC performance data. Field performance data provided by ADOT indicated
that more than 96% of AR- AC pavements provided generally good performance.
Therefore, the ADOT mix design method was considered a successful standard for
comparison of proposed improvements.
Methods and practices for AR- AC mix design used by industry and other agencies were
reviewed and synthesized to develop proposed improvements to the existing ADOT
procedure. Rice testing according to ARIZ 8061 was evaluated at two asphalt- rubber
( AR) binder contents, 6% and 7% by total weight of AR- AC mixture, to determine
whether the binder content should be increased to 7% for testing. Findings indicated that
results for samples at both binder contents fall within the precision of the test procedure;
either may be used, as the level of precision is equivalent.
Rebound of mix specimens after compaction was also measured and evaluated, with and
without constraining weights. Rebound has been a concern for AR- AC mix designers,
but no documentation of actual measurements of this anecdotal phenomenon could be
found. This may be the only study to address specimen rebound. Findings indicated that
most mixes exhibit some slight shrinkage as they cool which appears to be normal
volume change. Few mixes rebound. A failed mix design trial for another project
provided a mix which did visibly rebound, but the measurements were small. It was
decided that mixes that rebound should be discarded and redesigned.
As directed by the Technical Advisory Committee ( TAC), MACTEC developed new mix
designs for initial testing, using three different sources of aggregate and two different AR
binders. The same source of rubber was used in both AR binders. Gradation was varied
so that one binder used a rubber gradation on the coarse side, and the other used a
gradation on the fine side of the allowable rubber gradation limits. Rubber content was
varied to meet the required AR properties. The quantity of rubber required is a function
of the rubber gradation and the source and grade of the base asphalt cement.
ADOT’s original mix design procedure ( newly documented) was used to develop
“ control” AR- AC mix designs, which established aggregate gradation targets. The
Version 1 modified mix design procedure was then applied to the established aggregate
gradations for the respective sources. These initial designs performed for Task 1 are
referred to as “ Round 1” in this report.
1 Arizona Department of Transportation. ( ADOT) Materials Testing Manual. 1985. Section 806.
2
The Version 1 designs seemed to highlight differences in the effects of the two AR
binders on resulting volumetric properties. It appeared that the different binders had more
effect on the results than the choice of mix design procedure. However, two of the
aggregate sources had relatively high water absorption and yielded more variable test
results than the third. The question arose as to whether the Version 1 method better
distinguished AR- related differences in volumetric results or was the cause of these
differences. Therefore to better distinguish the effects of binder and mix design method,
additional testing was focused on mixes using less absorptive and less variable Salt River
aggregates.
As work with the Version 1 Marshall mix design method proceeded and the need for
additional replicate testing was identified, the project TAC decided to waive
the planned gyratory portion of the study to allow full evaluation of the Marshall
approach. It appeared that appropriate modifications to the Marshall method could be
established to provide a readily useable standard mix design procedure. It also appeared
that more resources would be required to thoroughly research the application of gyratory
compaction to AR- AC materials, while it was not clear if it would be possible to develop
a gyratory mix design method.
The next step was to further explore the relative effects of binder versus mix design
method using the relatively consistent Salt River aggregate source, and whether these
effects could be reproduced by other laboratories. MACTEC batched aggregate and
provided prepared binder to ADOT for “ shadow” or replicate testing of control and
Version 1 mixes, which is referred to as “ Round 2” in this report. Extensive analysis of
the results of Round 2 testing supported the initial findings that the AR binders had more
effect on volumetric results than the differences between the control and Version 1 mix
design procedures. ADOT’s results generally fell within the range of MACTEC’s results
for Rounds 1 and 2. The relatively close conformance of the results indicated that both
methods ( control and Version 1) could be reproduced by another laboratory.
Presentations of preliminary results were delivered at meetings of the Pacific Coast
Conference on Asphalt Specifications and at the Arizona State University Paving and
Materials Conference, rather than in workshop format. Comments were solicited. In
addition, the test results and the proposed Version 1 mix design procedure were
distributed for review and comment among the project team ( which also included Speedie
& Associates ( Speedie) and Rinker Materials Corporation Arizona ( Rinker) and two
others experienced with these materials including Western Technologies Inc. ( WTI)).
Results indicated that any of the modifications could be adopted but some were not
needed; Version 2 incorporated selected changes to clarify and streamline lab procedures.
ADOT offered an opportunity to use a 2004 AR- AC construction project to pilot the
proposed Version 2 AR- AC Marshall mix design method and provide materials for round
robin testing by the project team. The project selected provided an “ acid test” as the
subject “ Big Bug” aggregate materials have high water absorption and corresponding
increased testing variability. MACTEC performed the original mix design, and
developed an alternate AR binder for subsequent round robin testing. ADOT personnel
3
sampled the aggregate stockpiles and delivered these materials to MACTEC for
distribution among the participating laboratories.
Round robin testing was performed by four laboratories: ADOT, Speedie, Rinker and
MACTEC. These labs batched the aggregates and used prepared AR binder as would
normally be done for a new mix design or a verification of an existing design.
MACTEC compiled and analyzed the test results, which consist of a limited number of
physical tests ( which are also possible sources of variability) and calculated the
volumetric properties of interest. One of the participating laboratories experienced some
equipment problems that affected its results. To remove inaccuracies contributed by
variability of other tests, results were normalized by using overall averages of aggregate
specific gravity and Rice results to recalculate volumetric properties for each laboratory.
MACTEC performed statistical analyses to determine whether the mean results of the
respective laboratories for the properties of interest were statistically similar, and to
group and rank statistically different means. Precision of the proposed Version 2 mix
design procedure was evaluated with respect to results of Marshall asphaltic concrete
proficiency sample programs of the AASHTO Materials Reference Laboratory ( AMRL)
and ADOT, and ASTM precision statements for bulk and maximum theoretical specific
gravities. Although the normalized round robin results for some of the volumetric
properties did show significant differences among the respective laboratories, the
precision of the round robin testing performed by the individual laboratories is generally
within the ranges established for conventional asphaltic concrete materials.
The results of this study indicate that the proposed Version 2 AR- AC mix design
procedure is generally as repeatable and reproducible as a 75- blow Marshall mix design
for conventional asphalt concrete. Version 2 is presented in Appendix H as ARIZ 8322,
Marshall Mix Design for Asphaltic Concrete ( Asphalt- Rubber) [ AR- AC]. It has been
used for ADOT AR- AC projects in 2006. Some refinements may be made with
continuing use, but major procedural changes are not expected.
2 ADOT Materials Testing Manual. 1985. Section 832.
4
5
1. INTRODUCTION
The purpose of this study was to develop standard mix design methods for gap- graded
asphalt- rubber asphaltic concrete ( AR- AC) mixtures that can be used by contractors and
consultants. The AR- AC aggregate gradation is gapped on the coarse side of the
maximum density line to provide sufficient void space to accommodate the rubber
particles in the asphalt- rubber ( AR) and high AR binder contents. To date, ADOT’s
Central Laboratory has been responsible for performing the mix designs for these
materials which has at times been a strain on ADOT’s limited resources. The Department
is seeking to transfer AR- AC mix design responsibilities to industry, similar to the
current practice for standard Marshall and Superpave asphaltic concrete mixtures.
The scope of the study was originally divided into three tasks as follows:
• Task One: Review and Documentation of Current Methods
o Review Marshall mix design criteria
o Interview ADOT personnel
o Review industry standards and practices
o Compare various methods and procedures
o Synthesize best practices
o Look for correlations with field performance
o Develop and test proposed mix design improvements
Select three AR- AC mixes
Apply recommended improvements to the same materials
Check for rebound
Evaluate the effects of recommended changes to the mix design procedure
• Task Two: Development of Superpave Gyratory Methods
o Development of mix design procedures using the Standard Highway Research
Program ( SHRP) gyratory compactor
• Task Three: Testing Round Robins, Validation, and Presentation of Work
o Compare results of minimum of 3 mixes ( Round 1)
o Analyze results and conduct workshop
o Prepare formatted Arizona Test Method
o Preparation of Final Report, Technical and Project Presentations
The Technical Advisory Committee ( TAC) redirected some efforts as deemed
appropriate based on ADOT’s needs and on the results of each phase of testing. The
original work plan was to focus on the mixture properties of the material, and not on the
properties of the asphalt- rubber binder. However at ADOT’s request, the effects of rubber
gradation and rubber content of the AR binder on AR- AC mixture volumetrics were
incorporated. The impacts on mixture volumetrics were found to be significant.
6
The Executive Summary summarizes the work performed. ADOT provided AR- AC
performance data, the original formatted mix design method ARIZ 815c3, and ADOT’s
Proficiency Sample Program data for 75- blow Marshall testing performed over the last
ten years. The performance data showed the original ADOT mix design method was a
successful standard for comparison of proposed improvements.
Task One also included a review of various industry methods and practices for AR- AC
mix design, synthesis of best practices to develop proposed improvements, and laboratory
evaluation of the proposed improvements. As one of the proposed improvements, Rice
testing ARIZ 8064 was evaluated at two AR contents, 6% and 7% by total weight of AR-AC
mixture, to determine whether the AR content should be increased to 7% for Rice
testing. Rebound of mix specimens after compaction was also measured and evaluated,
with and without constraining weights.
For Task One, instead of using three existing AR- AC mix designs as planned, the TAC
tasked MACTEC to develop new mix designs using three different sources of aggregate
and two different AR binders. This created some overlap between Tasks One and Three.
The second planned task was to develop AR- AC mix design procedures using the SHRP
( Superpave) Gyratory Compactor. As work with the Marshall- based method proceeded
and the need for additional replicate testing was identified, the project TAC decided to
waive the gyratory work to allow full evaluation of the Marshall approach. It appeared
that appropriate modifications to the Marshall- based method could be established to
provide a readily useable standard mix design procedure. It also appeared that more
resources would be required to thoroughly research application of gyratory compaction to
AR- AC materials, while it was not clear if the desired result could be achieved.
Task Three was redirected by the TAC to further explore the relative effects of AR binder
versus mix design method using the relatively consistent Salt River aggregate source, and
whether these effects could be reproduced by other laboratories.
Workshop presentations were deferred and will likely be used to present the results of
this study along with the proposed AR- AC mix design method and new end result
specifications being implemented for AR- AC in accordance with ADOT 4155.
For Task Three, ADOT offered an opportunity to use a 2004 ADOT AR- AC construction
project to pilot the proposed standard ADOT mix design method and to provide materials
for round robin testing by the project team. The parties involved believed this would be a
superior way to conclude this study. The project selected provided an “ acid test” as the
subject aggregate materials have high water absorption and corresponding increased
testing variability.
3 Ibid. Section 815c.
4 Ibid. Section 806.
5 Arizona Department of Transportation ( ADOT). Standard Specifications for Road and Bridge
Construction. 2000. Section 415.
7
Round robin testing was performed by four laboratories: ADOT, Speedie, Rinker, and
MACTEC. MACTEC compiled and analyzed the results. The precision of the round
robin testing performed by the individual laboratories is generally within the ranges
established for conventional asphaltic concrete materials.
The results of this study indicate that the proposed AR- AC mix design procedure is
generally as repeatable and reproducible as a 75- blow Marshall mix design for
conventional asphaltic concrete.
1.1 ORGANIZATION OF THE REPORT
Chapter 1 is this Introduction.
Chapter 2 presents the development of the AR- AC mix design procedure from
documentation of the existing ADOT Marshall- based AR- AC method to development
and testing of the proposed Version 1modifications. It includes discussions of the
respective specifications and materials, findings of the analyses of Rounds 1 and 2 test
data, and the list of changes included in Versions 1 and 2 of the proposed AR- AC mix
design procedure. Test results and corresponding compilations, plots, and statistical
analyses are presented in Appendices A through E.
Chapter 3 covers the round robin testing of the Version 2 mix design method and
analyses in detail, including materials selection, AR binder preparation, instructions for
handling and testing, data reported, considerations regarding volumetric calculations, and
findings of the analyses. Test results and corresponding compilations, plots, and
statistical analyses are presented in Appendices F and G.
Chapter 4 presents the conclusions of this study.
The current version of the mix design procedure is in Appendix H.
8
9
2. DEVELOPMENT OF A MIX DESIGN PROCEDURE
2.1 DOCUMENT EXISTING MODIFICATIONS TO ARIZONA 815c
The first task of this study was to determine and document any modifications to the ARIZ
815c6 Marshall Mix Design Method that ADOT has been using to design mixes to meet
the requirements of Section 4137 Asphaltic Concrete ( Asphalt- Rubber). A meeting was
held with ADOT materials managers and laboratory personnel to go through the ARIZ
815c procedure line by line to identify and describe in detail the modifications used for
designing gap- graded AR- AC mixes. ADOT provided an electronic copy of ARIZ 815c
for a technical review of drafts. ARIZ 815c Modified for Asphaltic Concrete ( Asphalt-
Rubber) Version 5- 28- 03 was submitted as the first scheduled deliverable for this project,
and is presented in Appendix A.
2.2 MATERIALS SELECTION
Materials selection was a critical part of the experimental plan. The mix design method
to be developed must be applicable to the full range of aggregate, asphalt, and asphalt-rubber
materials available throughout Arizona that are suitable for use in AR- AC mix-tures.
The project TAC took an active role in determining what materials should be
included in the study.
2.2.1 Aggregates
The TAC identified three sources of aggregate for the bulk of the mix design testing that
represented a wide range of physical properties such as specific gravity and water
absorption. The aggregate sources designated were:
• Salt River ( Rinker 19th Avenue plant, Phoenix metropolitan area)
• Grey Mountain ( US 189 Milepost 454, northern Arizona)
• CKC Construction ( 1234 E. Airport Rd. Safford, Arizona)
Details of properties of aggregates from these respective sources are included in the
corresponding mix design summaries presented in Appendix B.
2.2.2 Rubber
The project proposal excluded evaluation of the effects of rubber gradation and content
on the resulting AR binders due to funding constraints. However, ADOT expressed great
interest in the effects of these factors on mixture volumetrics. It was thus decided to
deviate from the project proposal and develop and use AR binders that incorporated,
respectively, relatively coarse or fine rubber gradations within the relatively broad
gradation limits for Type B rubber in ADOT Section 10098, Asphalt- Rubber Material.
Type B rubber is used in AR binders for gap- and open- graded asphaltic concrete mixes,
and the specified gradation limits are shown in Table 2.
6 ADOT. Materials Testing Manual. 1985. Section 815c.
7 ADOT Standard Specifications for Road and Bridge Construction 2000 Section 413
8 Ibid. Section 1009
10
ADOT’s and MACTEC’s experience with AR materials indicated that rubber gradation
would affect the rubber content of the binder and volumetric properties of AR- AC,
particularly the arrangement of the mixture voids. For example, coarsening the rubber
gradation would typically increase the amount of rubber required to achieve the specified
AR binder properties, and would tend to increase Voids in the Mineral Aggregate ( VMA)
of AR- AC mixes.
2.2.3 Asphalt Cement
Most of the AR binders used by ADOT are classified as Type 2, which requires a
Performance Grade ( PG) binder 58- 22 ( ideal for climates with temperatures ranging from
58° Celsius down to - 22° Celsius) for the base asphalt cement. 9 Type 1 AR binders
require a stiffer grade of base asphalt cement, PG 64- 16, for areas with higher pavement
operating temperatures and heavy traffic. Type 3 AR binders require a softer PG 58- 28
and are used where enhanced resistance to low temperature cracking is needed.
2.2.4 Asphalt- Rubber Binders
MACTEC compiled a number of existing AR binder design profiles for consideration by
the TAC, and TAC members also suggested specific AR binders for use in this study.
Two Type 2 AR formulations were selected and designated Binder 1 and Binder 2. The
selected binders were produced and tested by MACTEC using the designated component
sources and grades. However, due to variations in the physical properties of the asphalt
and rubber materials since design, some of the selected formulations required adjustments
in rubber content, or a different source or grade of asphalt to meet specifications. Binder
1 used Para- mount PG 58- 22. The source of the base asphalt cement for Binder 1 was
changed from Chevron to Paramount. Binder 2 used Ergon Snowflake PG 58- 22. The
Ergon Snowflake asphalt cement available at that time for use in Binder 2 actually graded
as a PG 58- 28 rather than PG 58- 22, but since the resulting AR binder properties met
requirements for and conformed to the original Type 2 design, it was used as a Type 2.
The design profiles, components, and rubber gradations for Binder 1 and Binder 2 are
presented in Tables 1 through 4. Crumb Rubber Manufacturers ( CRM) was the source of
rubber for both AR binders.
9 Ibid
11
Table 1 Binder 1 Design Profile
Minutes of Reaction Specified
Test Performed 60 90 240 360 1440 Limits
Viscosity, Haake at 177° C, cP 2000 2300 2800 2900 2700 1500- 4000
Resilience at 25° C, % Rebound
( ASTM D5329) 37 37 37 20 Minimum
Ring & Ball Softening Point, ° F
( ASTM D36) 135.5 137 140 140 138 130 Minimum
Needle Penetration at 4° C, 200g, 60
sec., 1/ 10mm ( ASTM D5) 32 30 31 15 Minimum
Rubber source and type: CRM Type B ( coarse gradation)
Rubber content: 24.2% by weight of asphalt cement, 19.5 % by weight of total binder
Asphalt cement source and grade: Paramount PG 58- 22
Table 2 Binder 1 Rubber Gradation, Percent Passing ( ARIZ 71410)
Sieve Size Result (%) Specified Limits (%)
No. 8 100
No. 10 100 100
No. 16 69.5 65 – 100
No. 30 30.4 20 – 100
No. 50 10.7 0 – 45
No. 200 0.4 0 – 5
Table 3 Binder 2 Design Profile
Minutes of Reaction Specified
Test Performed 60 90 240 360 1440 Limits
Viscosity, Haake at 177° C, cP 2000 2100 2600 2400 2300 1500- 4000
Resilience at 25° C, % Rebound
( ASTM D5329) 39 42 42 20 Minimum
Ring & Ball Softening Point, ° F
( ASTM D36) 143 140 145 144.5 139.5 130
Minimum
Needle Penetration at 4° C, 200g,
60 sec., 1/ 10mm ( ASTM D5) 29 30 34 15 Minimum
Rubber source and type: CRM Type B ( fine gradation)
Rubber content: 22.7 % by weight of asphalt cement, 18.5 % by weight of total binder
Asphalt cement source and grade: Ergon Snowflake PG 58- 28
10 ADOT. Materials Testing Manual. 1985. Section 714
12
Table 4 Binder 2 Rubber Gradation, Percent Passing ( ARIZ 71411)
Sieve Size Result Specified Limits
No. 8 100
No. 10 100 100
No. 16 93.7 65 – 100
No. 30 40.6 20 – 100
No. 50 9.6 0 – 45
No. 200 0.7 0 – 5
Binder 1 did require a somewhat higher content of the coarser- graded rubber ( 24.2% vs.
22.7%) to provide properties similar to Binder 2 made with the finer- graded rubber.
2.3 PILOT AR- AC MIX DESIGNS – CONTROL MIXES
Field performance data provided by ADOT indicated that approximately 104 AR- AC
mixes were designed and placed from August 1989 through March 2001. Of these AR-AC
mixes, bleeding was reported for three that were used as urban arterial pavements in
the Phoenix metropolitan area, and rutting ( believed to be due to structural issues)
occurred in one mix placed on I- 8 near Yuma. Based on this information, as of April
2001, less than four percent of ADOT’s AR- AC pavements had exhibited severe distress
during a time period of over eleven years. Based on the historically good performance of
AR- AC mixes placed throughout Arizona, the existing mix design method was
considered to be successful. Therefore it was designated as the control method for this
study, the standard to which the results of the proposed improvements would be
compared. The method to be developed needs to provide at least the same quality AR-AC
material as the existing method, including adequate AR binder content to promote
long term durability and compliance with specifications.
ADOT AR- AC specifications at the time of this research were limited to requirements for
physical properties of aggregate ( gradation, sand equivalent, fractured faces and
abrasion); effective voids content ( 5.5 ± 1.0%); minimum VMA ( 19.0%); maximum
binder absorption ( 1.0%); and use of 1.0% portland cement or hydrated lime by aggregate
weight as a mineral admixture.
The testing plan allowed for a total of six mix designs to be performed according to the
newly documented existing ADOT AR- AC mix design method to serve as the controls
for this part of the study. AR Binder 1 was used to establish AR- AC control mix designs
with aggregates from each of the three designated sources. In some cases, appropriate
mix designs that met volumetric requirements could not be developed using Binder 1; the
related data for these are identified as “ Trial Summaries.” Design binder contents were
then determined for Binder 2 using similar gradations. The control AR- AC mix design
summaries and trial summaries are presented in Appendix B, along with compilations of
the properties of interest ( effective binder volume, VMA, voids filled with asphalt
( VFA), effective air voids, Marshall stability and flow) for each.
11 Ibid
13
2.3.1 Issues with CKC and Grey Mountain AR- AC Control Mixes
The TAC members selected the CKC and Grey Mountain aggregates to represent types of
aggregate materials present in the respective southern and northern parts of Arizona that
may present challenges to mix designers.
2.3.1.1 CKC Aggregates
The CKC source was selected specifically because ADOT’s Central Lab had experienced
problems in developing acceptable volumetric AR- AC mix designs when combining
these aggregate materials with an AR binder made with relatively coarse- graded rubber,
like Binder 1. It was necessary for ADOT to request an alternate AR binder made with a
finer gradation of rubber to obtain an appropriate mix design. The CKC aggregate
exhibited high water absorption which historically increases variability in laboratory mix
testing.
As shown on the CKC AR- AC design and trial summaries, MACTEC experienced the
same problems as ADOT when mixing the CKC aggregate with Binder 1. Increasing the
content of Binder 1 increased the mix VMA, and the mixture voids remained excessive
( 7.9%) even with 8.5% binder by total mix weight. It seemed as if the coarser rubber
particles in the binder were not allowing the aggregate matrix to consolidate and
interlock.
The aggregate blend was modified to provide a slightly denser matrix, but the gradations
of the available stockpiled materials did not allow a significant change in the composite
gradation. None of the stockpiles provided sufficient fines to close up the mix voids
while remaining within ADOT 41312 aggregate gradation limits. Therefore a suitable
mix design could not be developed for the combination of Binder 1 and the available
CKC aggregate materials.
However, when Binder 2 was substituted for Binder 1 the mixture voids dropped into an
acceptable range of 6.1% at 7.5% AR binder content, and 5.4% at 8.5% AR binder. This
also mirrored ADOT’s experience.
2.3.1.2 Grey Mountain Aggregates
The combination of Grey Mountain aggregates and Binder 1 exhibited a trend of
increased VMA with increased AR binder content similar to that of the CKC materials,
but less pronounced. It was possible to develop an AR- AC mix design with Gradation
Trial A and Binder 1. However the resulting combination of high VMA and high binder
content caused decreased Marshall stability and increased Marshall flow, which indicated
that properties were somewhat marginal. Such a design would not be recommended.
A wider range of stockpile gradations was available from the Grey Mountain source
which made it possible to evaluate the effects on the voids structure of either substituting
12 ADOT. Standard Specifications for Road and Bridge Construction. 2000. Section 413
14
or blending in a “ dirtier,” i. e., finer, crusher fines material with the clean crusher fines.
The change in gradation due to blending these two fine aggregate materials was small
enough to fall within production tolerances from Gradation A mix design targets ( see
Appendix B). Limited trials indicated that this small change in gradation resulted in a
drop from 7.5% to 6.9% effective air voids at 7.5% Binder 1 content by mix weight.
Substituting the finer crusher fines to further densify the gradation ( Gradation B with
crusher fines) had a profound effect on the voids content, dropping it down to 4.0% at
7.5% Binder 1 by mix weight.
No difficulties were encountered with developing suitable AR- AC mix designs using
Binder 2 with trial aggregate Gradation A. The finer rubber gradation produced an
acceptable mix design.
2.3.1.3 Discussion
The voids structure of asphaltic concrete and AR- AC mixtures depends on a number of
factors including, but not limited to:
• Aggregate particle size – gradation.
• Aggregate particle shape – examples include cubical, flat, angular.
• Aggregate surface texture – fine or coarse grains, glassy or rough, size and
number of surface voids, etc.
These factors affect how aggregates pack together when compacted. The Uncompacted
Void Content ( ARIZ 24713) used for Superpave mixes may be considered as an index of
such factors.
In AR- AC mixes, the discrete swollen rubber particles that remain in the AR binder after
interaction with the asphalt cement may also affect how aggregates pack together. The
rubber particles must also be accommodated within the aggregate matrix and may fill
some voids. However if the voids are too small to accommodate them, the rubber
particles may interfere with stone- to- stone contact and force the aggregate particles apart,
which increases VMA and mixture voids. In such cases, increasing the AR binder
content increases the number of interfering rubber particles and consequently increases
VMA and mixture voids. Finer rubber particles do not take up as much space as coarser
rubber and are more likely to fit within the aggregate matrix.
ADOT AR- AC mixes are limited to very low fines content in order to promote stone- on-stone
contact in the aggregate matrix and to provide sufficient void space to
accommodate a relatively high content of AR binder that includes discrete rubber
particles. ADOT specifications limit the amount of minus No. 200 material in any of the
component stockpiles to a maximum of 6.0%. Although design AR binder contents are
high compared to conventional mixes, AR- AC mixes do not require high contents of fine
aggregate particles in the mix to avoid drain down or minimize potential for bleeding.
13 ADOT. Materials Testing Manual. 1985. Section 247
15
The lack of allowable fines leaves the mix designer with few options for closing up high
voids AR- AC mixes. If changing the aggregate stockpile or bin blend proportions and
AR binder content cannot reduce the voids enough, then it may not be possible to develop
a suitable mix design with a specific AR binder that fully complies with binder
specification requirements and includes relatively coarse- graded rubber. This situation is
both illustrated in Appendix B in MACTEC’s control mix design trials with CKC
aggregate and Binder 1, and supported by ADOT’s experience with this source.
The control mix design trials performed with the Grey Mountain aggregate ( also
presented in Appendix B) indicate that adding a relatively small proportion of fines can
have major impacts on reducing effective voids contents of gap- graded mixes. However
the crusher fines material used to adjust the Grey Mountain mixes with Binder 1 does not
meet ADOT limits for maximum 6% minus No. 200 material and could not be used
without waiving these requirements.
Although the relative impact of adding fines would be material- specific, mix designers
must have some means to adjust mixture voids. The first option would be to seek a finer
crumb rubber material to use in the AR binder. In cases where finer rubber is not
available and an acceptable AR- AC mix design cannot be developed otherwise,
consideration should be given to allowing use of aggregate stockpiles that include more
than 6.0% passing the No. 200 sieve, raising the upper gradation limit for the composite
aggregate blend including admixture to three or four percent passing the No. 200, or both.
2.3.2 Salt River Control Mixes
No problems were encountered in developing control mixes for the Salt River aggregates.
The mix design data for the control mixes with Binder 1 and Binder 2 are included in
Appendix B. As requested by the project TAC, MACTEC performed two additional
replicate designs for the Salt River control mixes with each binder using the established
target gradation. Results were relatively consistent and are summarized in Table 5. The
limited replicate data show design contents of Binder 2 ( finer rubber) are slightly lower
than those for Binder 1 ( coarser gradation) at corresponding air voids contents.
Table 5 Design Binder and Air Voids Contents, Salt River Aggregate Control
Mixes
Mix ID* Binder 1
% by mix weight Air Voids, % Binder 2
% by mix weight Air Voids, %
B1C1 7.5 5.6 --
B1C2 7.3 5.5 --
B1C3 7.3 5.4 --
B2C1 7.1 5.6
B2C2 7.1 5.5
B2C3 6.8 5.4
Average 7.37% 5.5% 7.0% 5.5%
* Mix ID Example: B1 C1 = Binder 1 Control Mix Trial 1
16
2.4 MODIFICATIONS TO EXISTING ADOT AR- AC MIX DESIGN
PROCEDURE
Development of the Version 1 modifications to the mix design procedure began during
initial documentation of the existing AR- AC mix design method. MACTEC solicited
input from the ADOT Materials staff, the project team and TAC, and other local
consultants who design AR- AC mixes for counties and municipalities.
The primary procedural changes considered included making and treating the Rice
specimens in the same manner as the loose Marshall specimens, and adding weights to
the surface of compacted Marshall specimens to prevent rebound while cooling prior to
extrusion from the molds. Rice tests of AR- AC mixes have customarily been performed
at 6.0% AR binder content, although AR binder content is rarely less than 7.0% by
weight of mix. Thus, a comparison of results of Rice testing at 6.0% and at 7.0% AR
binder was deemed necessary. A complete list of the modifications proposed is presented
in Section 2.4.1.
ARIZ 815c14 includes considerable explanation and exposition of calculations which
makes its presentation lengthy and cumbersome. ADOT Materials staff requested
changes in the presentation format to clarify the method and make it easier to use, and
modification of the volumetric calculations to conform to those used by the Asphalt
Institute for design of Marshall and Superpave mixes. 15,16
2.4.1 List of Considered Procedural Changes to AR- AC Mix Design Method
1. Include mineral admixture in the mix as part of the aggregate.
2. Use “ Wet Prep” method of admixture addition – mix dry admixture
thoroughly with dry aggregate to distribute uniformly throughout, then blend,
then add 3% water by aggregate weight and mix thoroughly to wet.
3. Batch aggregates in oven dry condition.
4. Fabricate Rice specimens at 7.0 % AR binder by total mix weight instead of
6.0 %, and include the required 1% admixture by dry aggregate weight ( added
and wet prepped as in step 2 above) but omit liquid anti- strip.
5. Cure Rice specimens at the same temperature ( 325 º F ± 10 º F) and for the same
amount of time ( 2 hours) as for the loose mixture for Marshall specimens.
6. Mixing temperature: AR binder at 350 º F, aggregate at 325 º F
7. Compaction temperature: 325 º F to 335 º F
8. Cool the compacted AR- AC specimens vertically in the molds ( with base
plate underneath and 2000grams ± 10 gram steel disc on top of specimen) to
less than or equal to 90 º F before extruding them.
14 Ibid, Section 815c
15 The Asphalt Institute. “ Mix Design Methods for Asphalt Concrete and Other Hot- Mix Types”, Chapter 4
16 The Asphalt Institute. “ Superpave Mix Design”, Chapter 4
17
The changes listed were incorporated to develop “ Version 1” mix designs for each
aggregate source, using the composite aggregate gradations developed for the
respective control mix designs with Binder 1 and Binder 2.
2.4.2 Mix Designs – Version 1 Mixes
2.4.2.1 Salt River Aggregate Version 1 Mixes
No problems were encountered in developing Version 1 mix designs for the Salt River
aggregates. As requested by the project TAC, MACTEC performed two additional
replicate designs for the Salt River aggregate Version 1 mixes with each binder using the
established control gradation. The Version 1 mix designs with AR Binders 1 and 2 are
included in Appendix C. Results were relatively consistent and are summarized in Table 6.
Table 6 Design Binder and Air Voids Contents for Salt River Aggregate
Version 1 Mixes
Mix ID* Binder 1
% by mix weight Air Voids, % Binder 2
% by mix weight Air Voids, %
B1PC1 8.0 5.6
B1PC2 8.1 5.6
B1PC3 8.2 5.6
B2PC1 6.9 5.4
B2PC2 6.7 5.5
B2PC3 6.7 5.4
Average 8.10 5.60 6.77 5.43
* Mix ID Example: B1PC1 = Binder 1, Version 1 Mix Design Trial 1
The limited data show Version 1 mix design contents of Binder 2 ( finer rubber) are 1.1%
to 1.5% lower than those for Binder 1 ( coarser gradation) at similar air voids contents.
Compared to the results listed in Table 5, design contents for Binder 1 Version 1 mix
designs increased by 0.5% to 0.9% ( average content 8.1%) over the range of Binder 1
contents determined for the control mix designs ( range 7.3%- 7.5%, average 7.37%).
However the Version 1 design contents of Binder 2 showed very little difference from the
control mix design value range of 6.8%- 7.1% with average of 7.0%. The effects of the
difference in AR binder composition, rubber gradation, and content, appeared to be
accentuated by the Version 1 method.
2.4.2.2 CKC Aggregate Version 1Mix Designs
Work on Version 1 designs was limited to a trial using 7.5% and 8.5% Binder 2 by
weight of the modified composite aggregate gradation used in the control mix. The data
are summarized in Appendix C. Effective air voids of the Version 1 mix were higher
than the control, but no conclusions can be drawn from the limited data.
18
2.4.2.3 Grey Mountain Aggregate Version 1 Mix Designs
Work on Version 1 designs was limited to a trial using 7.5% and 8.5% Binder 2 by
weight of the original aggregate gradation ( A) used in the control mix design. The data
are summarized in Appendix C. Effective air voids of the Version 1 mix were lower than
the control design with Binder 2, but no conclusions can be drawn from the limited data.
2.4.2.4 Discussion of Results
The purpose of the additional mix testing with the Salt River aggregates was to permit
evaluation of the variability of both the control and Version 1 design methods and of the
materials being used. The Salt River aggregate has proved to be a good, sound, durable
material for use in asphaltic concrete, with low water absorption and relatively consistent
physical properties. It has historically proven to be less variable than the CKC or Grey
Mountain aggregates and thus was the best choice for replicate testing to evaluate the
effects of binder and mix design method on the results. Volumetric properties evaluated
included effective binder volume, VMA, VFA, and effective air voids content.
Some volumetric differences due to binder composition were expected and occurred. In
plots of the control mix data, the data tend to group by binder but there is some overlap.
However the plots of the Version 1 mixes show very distinct differences between
volumetric properties of mixes made with Binder 1 and those made with Binder 2 at
corresponding binder contents. 17 The magnitudes of these differences are greater than
would be expected for the relatively minor changes to the mix design procedure and
represent significant practical differences in the results as follows:
• Air Voids – more than 2% difference between Binder 1 and Binder 2 mixes
• VMA – up to 2% difference
• Voids Filled – up to 10% difference
These large differences do follow expected trends for the rubber gradations and relative
contents, but raised the following questions:
1. Did the changes to the mix design method cause these differences in
volumetric results, or simply better distinguish binder related differences in
mixture properties that had been occurring but had not been recognized?
2. Are the differences repeatable and reproducible?
a. With these same materials?
b. With other materials?
17 Referenced plots are included in compiled data plotted for MACTEC- ADOT Rounds 1 and 2 that is
presented in Appendix E, but are presented with other results and not alone due to the large number of plots
included with this report.
19
A program of replicate testing by ADOT and MACTEC was implemented as Round 2 of
this study to answer these questions. Repeatability typically refers to the precision of
testing expected, i. e., the acceptable range of results, for a single test operator or
laboratory. Reproducibility typically refers to the precision of testing expected for two or
more different laboratories. Round 2 activities and findings are discussed in Section 2.5
of this report.
2.4.3 Analysis of Rice Results at 6.0% and 7.0% AR Binder Content
While performing the control and Version 1 mix designs with aggregate materials from
the respective sources, MACTEC prepared and tested corresponding sets of Rice
specimens at AR binder contents of 6.0% and 7.0% by total mix weight. Additional
replicate Rice testing of control and Version 1 mixes was also performed during Round 2.
The dry back procedure was used because it is the referee method, although it
incorporates more possible sources of variation. The increased variability is reflected in
the precision and bias statements for the corresponding ASTM D 2041, Standard Test
Method for Theoretical Maximum Specific Gravity and Density of Bituminous Paving
Mixtures, developed from AMRL Proficiency Sample Program data with and without dry
back. 18 Results and statistical analyses of Rice testing are presented in Appendix D.
To validate the data, the measured Rice value at one binder content was used to calculate
the effective specific gravity of the aggregate, Gse, using Equation 1. The calculated Gse
value was used in Equation 2 to calculate the Rice value at the other binder content.
b
b
mm
mm
mm b
se
G
P
G
P
P P
G
−
−
= Equation 1
b
b
se
s
mm
mm
G
P
G
P
P
G
+
= Equation 2
Where
Gse = Effective specific gravity of the aggregate- admixture blend
Gmm = Maximum theoretical specific gravity of the AR- AC at AR binder content Pb
Pb = AR binder content at which the Rice test was performed
Gb = Specific gravity of the AR binder
Ps = Aggregate content, percent by total weight of mix ( 100- Pb)
Pmm = Percent by weight of total loose mixture = 100%
Results of the measured and calculated Rice values were then compared. The differences
between measured and calculated Rice values at 6.0% and 7.0% AR binder contents are
no greater than 0.012, which is at the limit of the acceptable range of two results obtained
18ASTM. “ ASTM D 2041- 03a, Standard Test Method for Theoretical Maximum Specific Gravity and
Density of Bituminous Paving Mixtures.” ASTM Book of Standards 2005, Volume 4.03, pp. 177- 180.
20
on the same material by a single operator according to ARIZ 417b. 19 The maximum
difference was obtained for a control mix made with the Grey Mountain aggregate. Only
one of the mixes made with the Salt River aggregate yielded a difference of greater than
0.004 between measured and calculated Rice values at 6% and 7% AR binder contents.
Thus the variability of the results for both the control and Version 1 mixes appears to fall
within the acceptable range for this test.
Analysis of variance ( ANOVA) was also used to evaluate the relative effects on Rice
results of AR binder ( Binder 1 or Binder 2) and design method ( control or Version 1).
The results of the analysis indicate negligible effects of these factors on the Rice results.
The effects of interaction of binder and method were stronger than either factor alone but
were still negligible. The analysis indicates that including mineral admixture does not
measurably increase variability of Rice test results and is feasible. Including the
admixture in the Rice specimens also simplifies calculations.
2.4.3.1 Summary
Rice testing for AR- AC mix design may be performed at either 6.0% or 7.0% AR binder
content on mix specimens that include lime as a mineral admixture. Although no testing
was done with cement as a mineral admixture, it is expected that these results would
apply to cement. Although samples fabricated with 7.0% AR binder were reportedly
more difficult to work with, the quality of the results of this study did not appear to be
affected. Asphalt- rubber is very sticky, so increasing the binder content can make it more
difficult to break up any clumps of fine aggregate particles as required by the test
procedure.
The TAC decided to continue using the lower 6.0% AR binder content for AR- AC mix
design to facilitate handling and breakup of the Rice specimens, as the analysis of results
indicated no need to change. The same type and proportion of mineral admixture
included in the Marshall specimens should be included in the Rice specimens.
2.4.4 AR- AC Rebound of Compacted Specimens
For purposes of this study, rebound is defined as a measurable increase in the height of a
compacted AR- AC specimen after completion of compaction and prior to extrusion. This
phenomenon has been observed occasionally and reported anecdotally during the last 20
years or so, but MACTEC was not able to find any indication that rebound of AR- AC
mixes has ever been formally documented. 20
In the early 1990s, AR- AC mixes were developed for demonstration projects throughout
the U. S. in response to the legislative mandate of the 1991 Intermodal Surface Trans-portation
Efficiency Act ( ISTEA) to include scrap tire rubber in asphalt pavements.
Rebound was occasionally reported during attempts at mix design verification by
19 ARIZ 417b Maximum Theoretical Specific Gravity of Field Produced Bituminous Mixtures ( Rice Test),
December 1987.
20 “ Use of Scrap Tire Rubber – State of the Technology and Best Practices.” Caltrans, 2005
21
laboratories that had little if any experience in working with asphalt- rubber materials.
The Principal Investigator has personal knowledge of four such cases, of which all but
one seemed to be generally resolved by substituting hand Marshall compaction ( the
referee method) for mechanical compaction and improving temperature control during
mixing and compaction. In those three cases, it was found that the mechanical Marshall
hammers had not been calibrated to the referee hand method; some states did not require
it. The exception was a dense- graded mix which exhibited some volumetric issues and
likely did not have enough void space to accommodate the rubber particles in the binder.
Although AR- AC specimen rebound is not often observed, most of the local consultants
informally surveyed by MACTEC indicate that they routinely take some action to prevent
specimen rebound during AR- AC mix design. Several of the laboratories keep base
plates on top of the specimen in the Marshall mold during cooling, and others place
weights of up to 5,000 grams directly on the top surface of the compacted Marshall
specimen. Base plates do not assure uniform contact with the specimen and thus were
not considered appropriate for this study.
MACTEC had steel weights with handles (“ pucks”) fabricated to fit on top of 4- inch
diameter AR- AC Marshall specimens inside the compaction mold. Puck weight was
2,000 ± 10 grams. Figure 1 shows a picture of the puck and of the dial indicator that was
used to measure vertical displacement of the puck over time.
Figure 1: 2,000 gram Rebound “ Puck” and Dial Indicator
Results of rebound testing are presented in Appendix D. The results for the Round 1 and
Round 2 control and Version 1 mixes show that height change was negligible for most of
the specimens tested with or without the 2,000 gram weight. The data indicate that most
of the specimens experienced some minor shrinkage upon cooling. The 2,000 gram
weight did not appear to make a practical difference in height of compacted specimens of
mixes that did not swell.
22
By chance, a mix design trial for a different project yielded specimens that were observed
to puff up like a soufflé in the Marshall molds after compaction. This mixture was
duplicated and tested for rebound with and without the 2,000 gram puck. Results for the
“ soufflé mix” are also included in Appendix D. Although un- weighted specimens did
exhibit rebound, increases in height measured no more than 0.014 inch. The pucks did
succeed in preventing rebound of the soufflé mix.
2.4.4.1 Summary of Rebound Evaluation
This rebound evaluation may be the first to be documented. Results indicated that
changes in AR- AC specimen height after compaction are generally negligible, and that
most specimens exhibit minor shrinkage while cooling in the molds. Although weights
may be used to prevent rebound, there is no compelling reason to require their use.
It was the consensus of the project team and TAC that AR- AC specimens that exhibit
noticeable rebound after compaction should be considered as indicators of mixture
volumetric issues. Such specimens should be discarded and the composite aggregate
gradation should be adjusted to better accommodate the AR binder.
2.4.5 Round 2 Replicate Testing – ADOT’s Central Lab and MACTEC
Review with the TAC of MACTEC’s results of replicate tests of control and Version 1
mixes made with Binder 1 and Binder 2, respectively, indicated that more testing was
needed to evaluate the effects of the Version 1 modifications, as well as their
repeatability.
A focused test plan and handling instructions were developed for both ADOT and
MACTEC to evaluate MACTEC’s Round 1 results, and Round 2 of testing was initiated.
MACTEC presented the instructions for making specimens of Version 1 mixes in the
format of the proposed revised mix design procedure as Version 9- 26- 03, updated 10- 29-
03. This was an intermediate draft to be applied only to this replicate testing phase of this
study and was not intended to be the final version. The control mix replicates were to be
made according to the existing ADOT mix design method.
MACTEC batched the Salt River aggregate materials for ADOT to use for “ Round 2” rep-licate
testing for control and Version 1 AR- AC mixes. The aggregate samples were de-livered
to ADOT’s Central Laboratory along with lime admixture, batch sheets, six gallons
each of Binder 1 and Binder 2, and a 2000- gram rebound “ puck” as a template for ADOT to
duplicate. MACTEC also prepared and tested three more replicates each of the Salt River
control and Version 1 mixtures with Binders 1 and 2, respectively, for Round 2.
When ADOT personnel began to fabricate specimens for the Version 1 mixes, it became
apparent that there had been a misunderstanding as to how MACTEC had incorporated
the lime admixture in these mixes during the Round 1 testing. MACTEC had reported
that the lime was substituted for 1% of the crusher fines in the composite blend, and
viewed this simply as a modification of the existing laboratory procedure. However
23
ADOT was concerned that this approach could be construed as a policy change regarding
admixture addition, which was not intended. ADOT therefore instructed MACTEC to
incorporate lime in the Version 1 mixes the same as for the control mixes, by determining
the composite aggregate blend and then adding 1% lime by total dry weight of aggregate.
MACTEC batched new specimens for the Version 1 mixes for Round 2 testing.
MACTEC compiled and plotted test results of Rounds 1 and 2. Microsoft Excel was used
to calculate means, standard deviations, and outlier limits ( according to the ADOT
method for dispute resolution) for the respective data sets. The one- way analysis of
variance ( ANOVA) feature of the Excel Data Analysis package was used to evaluate the
statistical validity of combining MACTEC’s data from Rounds 1 and 2, for respective
binders and content levels. MACTEC considered this particularly important due to the
difference in batching aggregates and admixture for the Version 1 method between
rounds. Results of these analyses indicate that MACTEC’s data from Rounds 1 and 2
may be combined at levels of confidence ranging from 95% to 99%. Printouts of the
ANOVA analysis are included in Appendix E. The results are summarized in the One-way
ANOVA Results Matrix also in Appendix E.
Two- way ANOVA was used to evaluate the relative effects of both Binders 1 and 2 as
well as the mix design method ( existing ADOT versus Version 1) on the results. The
results are also presented in Appendix E. These ANOVAs indicate that although there
are some effects of mix design method, binder is clearly the primary source of differences
among the control and Version 1 mixtures tested by MACTEC.
The ADOT results were provided in two compilations, with voids analyses performed
based on Rice values at 6.0% and 7.0% AR binder content, respectively. MACTEC had
based voids analyses for the control mixes on Rice at 6.0%, and used Rice at 7.0% for
volumetric calculations for the Version 1 mixes. The corresponding ADOT data
compilations were used for comparison in the various plots and analyses of variance
which are presented in Appendix E.
A full set of 24 plots of MACTEC’s and ADOT’s combined Rounds 1 and 2 test results
for control and Version 1 mixes made with Salt River aggregates and Binder 1 and
Binder 2 were generated and are presented in Appendix E of this report. A detailed
legend is provided to facilitate review of the plots. Differences between Rounds 1 and 2
in batching and gradation of the Version 1 mixes appear to be reflected in the plots of
MACTEC’s results, which typically bracket the ADOT Round 2 results.
The plots of VMA, VFA, and effective air voids results versus AR binder content for the
replicates from both Rounds 1 and 2 illustrate that the distinctions between binders
highlighted in the Round 1 Version 1 mix results still exist. However the differences are
smaller. Since one of the Version 1 modifications ( approach to adding lime) was elimi-nated
along with the related minor difference in composite gradation, this shift toward the
control mix results makes sense. The remaining differences seem most likely to be bin-der
related. The plots also illustrate the two- way ANOVA results. For each binder,
24
results of control and Version 1 mixes tend to overlap. However the volumetric results of
Binder 1 mixes generally differ from those of Binder 2 mixes.
After visual examination of the plots with ADOT Round 2 data added indicated similar
results, MACTEC performed numerous ANOVAs to evaluate and compare results with
respect to design method, binder, and laboratory. It was necessary to tabulate the
ANOVA results to look for patterns and correlations.
Two- way ANOVA of the ADOT results were performed to evaluate the relative effects
of binder and mix design method. The individual ANOVAs are presented in Appendix E.
To facilitate review, these ANOVA results are summarized in the Two- Way ANOVA
Results Matrix included in Appendix E along with the results of the corresponding
analysis of MACTEC data. The statistical analysis indicates that binder had a very strong
effect on test results from both laboratories, and that the design method used ( control
versus Version 1) had relatively little impact. This finding validates the mix design
procedure that ADOT has been using and indicates that only the most useful and practical
of Version 1 mix design modifications should be adopted. It also validates a considerable
body of experience and anecdotal data that has long indicated that the AR binder is a key
factor in AR- AC mixture volumetrics.
The findings of the analyses of Round 1 and 2 results are summarized as follows:
• Review of plots of VMA, VFA, and effective air voids results indicate that
both the control ( existing ADOT) and Version 1 mix design methods
generally distinguish between Binder 1 and Binder 2 for these properties.
• The respective averages of MACTEC and ADOT Round 2 test results are in
substantial agreement for both binders and design methods, except for
Marshall stability.
• ADOT’s stability results were systematically higher than MACTEC’s.
• Results of Marshall stability and flow tests do not reliably distinguish among
binders.
• Effective binder volume appears relatively insensitive to binder type or design
method used in this study.
• Analysis of variance indicates that the mixes made with Binder 1 ( Paramount
PG 58- 22 with 24.4% coarse CRM rubber by weight of AC) exhibited greater
variability than mixes made with Binder 2 ( Ergon PG 58- 28 with 22.7% fine
CRM rubber by weight of AC). This is best illustrated by comparison and
ANOVA of MACTEC’s Round 1 and Round 2 test results for control mixes
made with the respective binders.
25
• In spite of the variations in individual mix property values, the agreement
between averages of ADOT and MACTEC Round 2 test results remains very
good for the binders and procedures used. This indicates that the overall AR-AC
mix design results can be reproduced by other laboratories.
• The ANOVA results matrix shows relatively good agreement between
MACTEC Round 1 and ADOT Round 2 results, in spite of differences in
binder storage time and Version 1 aggregate gradation. This further supports
MACTEC’s conclusion that the AR- AC design results are reproducible.
• ANOVA of the ADOT and MACTEC data indicates that the effects of the
binder are consistently very strong, while mix design method within this study
has relatively little if any effect.
• Based on the findings to date, it is not necessary to adopt each of the changes
to the existing ADOT mix design method for AR- AC that MACTEC
originally proposed. Recommended changes are limited to the following:
o Use oven- dry batching only when aggregates can not be air- dried to a
moisture content of less than 3%.
o Use “ Wet Prep” method of admixture addition – add 1% admixture by
aggregate weight and mix thoroughly to distribute, and then thoroughly
mix in 3% water by aggregate weight.
o Fabricate Rice specimens with 1% admixture by weight of aggregate
( added by wet prep) and 6% AR binder by total mix weight.
o Cure Rice specimens at the same temperature ( 330 º F ± 5 º F) for the same
amount of time ( 2 hours) as the loose AR- AC mixture used to make
Marshall specimens.
o Set mixing temperature: AR binder at 350 º F, aggregate at 325 º F.
o Set compaction temperature: 330 º F ± 5 º F.
o Cool the compacted specimens upright in the molds to less than or equal
to 90 º F before extruding them. Specimens should not be extruded until
just prior to testing.
o Do not place weights on top of compacted AR- AC specimens while
cooling in the mold. Mixes that exhibit rebound in the mold should be
discarded and redesigned.
26
The TAC concurred with the findings of the analyses and the recommended changes to
the mix design method, which are relatively minor. These changes were incorporated as
Version 2 of the AR- AC mix design procedure.
The results of the Round 2 replicate testing indicated that the control and Version 1
methods were relatively repeatable within a single laboratory and that the resulting mix
designs could be substantially reproduced by another laboratory. However the replicate
testing was performed on mixes made with a single source of relatively consistent high
quality aggregate materials, batched by a single laboratory under tightly controlled
conditions, so more evaluation would be useful.
The next task was to use round robin testing to evaluate whether the proposed Version 2
mix design method was robust enough to be used by other qualified laboratories to design
AR- AC mixes, using aggregate materials of varying quality that are more challenging to
work with than the Salt River materials.
27
3. ROUND ROBIN TESTING FOR VERIFICATION OF
PROPOSED AR- AC MIX DESIGN METHOD
The purpose of the round robin testing was to provide an “ acid test” for the proposed mix
design procedure. The round robin was intended to simulate real world mix design
and/ or verification operations. Participants would start with bulk samples of respective
aggregate stockpile materials, mineral admixture and prepared AR binder. Each
participating laboratory would measure aggregate specific gravity and absorption
properties; batch aggregates to meet composite gradation targets and mix with the
prepared AR binder; compact, condition, and test mixture specimens fabricated with a
range of AR binder contents; and calculate volumetric properties. The results would be
used to select a design AR binder content for each of three sets of replicate results.
3.1 PROJECT AND MATERIALS SELECTION
ADOT provided the opportunity to use a 2004 ADOT AR- AC construction project to
pilot the proposed standard ADOT AR- AC mix design method and provide materials for
round robin testing by the project team ( Speedie and Associates, Rinker, ADOT’s Central
Lab, and MACTEC). In addition, ADOT planned to obtain samples for acceptance
testing during construction to characterize the mix as produced and placed ( including
compaction results) so that the performance of the resulting pavement can be monitored
over time by periodic surveys. The parties involved believed this would be the best way
to conclude this study.
ADOT selected the following ARAC construction project to pilot the proposed mix
design method.
Project Name: Badger Springs – Big Bug
Project No.: IM- 017- B( 005) A
TRACS No.: 017 YV 256 H611501C
Project Location: I- 17 NB and SB MP 263- 255
The project was called “ Big Bug” and the source of the aggregate was the Dugas Pit.
ADOT personnel obtained bulk samples of the designated project aggregate materials
from the Dugas Pit, including clean crusher fines, 3/ 8” and 3/ 4” stockpile materials, for
use in the mix design and round robin testing. ADOT delivered the aggregate samples to
MACTEC in late June, 2004.
The Dugas aggregate has relatively high water absorption: more than 1.5% for the coarse
fraction, and more than 2% for the fine fraction.
28
3.2 MATERIALS DESIGNS
3.2.1 Asphalt- Rubber Binder Design
A Type 2 AR binder was designed and produced by Speedie and Associates ( Speedie) in
June 2004 for use in the AR- AC mix design. The AR binder design profile is presented
in Table 7. The rubber, CRM, which came from the same source, was included with
Binders 1 and 2 for Rounds 1 and 2 of this study. The PG 58- 22 asphalt was from
Chevron ( a different source than used in Rounds 1 and 2). Sieve analysis results in Table
8 show that the rubber gradation was coarse and very similar to that used in Binder 1.
ADOT provided samples of this AR binder to MACTEC for use in the mix design.
Table 7 Original Big Bug AR Binder Design Profile
Test Performed Minutes of Reaction
60 120 240 1440
Specified
Limits
Viscosity, Haake at 177° C, cP 2100 1900 2300 2700 1500- 4000
Resilience at 25° C, % Rebound
( ASTM D3407) 31 33 35 34 20 Minimum
Ring & Ball Softening Point, ° F
( ASTM D36) 139 138 140 143 130 Minimum
Needle Penetration at 4° C, 200g,
60 sec., 1/ 10mm ( ASTM D5) 23 22 30 25 15 Minimum
Rubber source and type: CRM Type B ( coarse gradation)
Rubber content: 25.8% by weight of asphalt cement, 20.5 % by weight of total binder
Asphalt cement source and grade: Chevron PG 58- 22
Table 8 Big Bug AR Binder Rubber Gradation, Percent Passing ( ARIZ 71421)
Sieve Size Results
( percent passing
Specified Limits
( percent passing)
No. 8 100
No. 10 100 100
No. 16 78 65 - 100
No. 30 28 20 - 100
No. 50 4 0 - 45
No. 200 0 0 - 5
3.2.2 AR- AC Mix Design
MACTEC performed the AR- AC mix design according to the procedure described. The
mix design summary and detailed test results are presented in Appendix F. The design
AR binder content of 7.8% yielded a target air voids content of 5.7%.
21 ADOT. Materials Testing Manual. 1985. Section 714
29
3.3 PREPARATION OF ASPHALT- RUBBER BINDER SAMPLES FOR
ROUND ROBIN TESTING
It was discovered that the amount of AR binder originally prepared and submitted for use
in the mix design was not sufficient to complete the planned round robin testing.
Therefore MACTEC prepared and tested AR specimens using the source and grade of
respective asphalt cement and rubber materials used in the original binder design
developed by Speedie and Associates. However, differences in the properties of
PG 58- 22 asphalt cement samples received by MACTEC’s laboratory three months after
completion of the original AR binder design required some adjustments to the AR blend.
It was necessary to increase the rubber content from 25.8% to 26.6% by weight of asphalt
cement to provide an AR binder that fully complied with specifications throughout the
24- hour laboratory interaction period. The updated binder design data is presented in
Table 9. MACTEC does not know if any similar adjustments to rubber content were
required during field blending of the AR binder for AR- AC construction on the Big Bug
project in September 2004.
Table 9 AR Binder Design Profile for Round Robin Testing Version 2 Mix Design
Minutes of Reaction Specified
Test Performed 60 90 240 360 1440 Limits
Viscosity, Haake at 177° C, cP 1600 2100 2000 1900 1500- 4000
Resilience at 25° C, % Rebound
( ASTM D5329) 35 37 35 20 Minimum
Ring & Ball Softening Point, ° F
( ASTM D36) 152 152 153 147 130
Minimum
Needle Penetration at 4° C, 200g,
60 sec., 1/ 10mm ( ASTM D5) 20 22 23 15 Minimum
Rubber source and type: CRM Type B ( coarse gradation)
Rubber content: 26.6 % by weight of asphalt cement, 21.0 % by weight of total AR
binder
Asphalt cement source and grade: Chevron PG 58- 22
Since the AR binder is a major factor in mix volumetrics, it was important to assure that
there was a sufficient amount of the updated binder for the participating laboratories to
complete their testing. MACTEC was tasked to prepare 20 gallons of the AR binder
represented by Table 9 in order to provide sufficient material. The change in the binder
was expected to cause some changes in volumetric properties compared to the original
mix design, but comparisons to the original design were not necessary. Since each of the
round robin participants was using the new AR binder material, the conduct and analysis
of the round robin testing would not be affected, although the individual test results were
expected to differ from the original design parameters.
30
3.4 INSTRUCTIONS AND DISTRIBUTION OF SAMPLES FOR ROUND
ROBIN TESTING
MACTEC prepared instructions for conduct of the round robin testing for the Version 2
mix design method to promote procedural uniformity among the participants, to highlight
differences between the revised ADOT AR- AC mix design procedure and current
practice, and to list the data items required to complete the round robin. A copy of the
sheet of instructions is presented in Figure 2. MACTEC also provided an electronic
spreadsheet file for data entry and corresponding hard copy, which clearly showed what
test results and data items were required for MACTEC’s analysis of the results.
MACTEC delivered copies of these documents, individual and target composite aggre-gate
gradation data, and the revised ADOT AR- AC mix design procedure along with bulk
samples of the individual aggregate and admixture materials and five one- gallon cans of
asphalt- rubber binder to the participating laboratories during the last week of October and
first week of November 2004. Each lab was instructed to determine aggregate specific
gravities ( bulk oven dry, saturated surface dry ( SSD), and apparent) and absorption of the
composited coarse and fine fractions, to fabricate and test three replicates of the mix
design using the updated AR binder, including one set of Rice tests per replicate, and to
report their test results to MACTEC. Each replicate included three AR binder contents.
To provide a better simulation of the entire mix design process, the aggregates for the
round robin were not pre- batched as they were in Rounds 1 and 2. Two of the partici-pating
laboratories reported some minor departures in their aggregate blends from the
target composite gradation due to variations from the overall average gradation within the
stockpile samples. They were not instructed to do any artificial blending. The largest
difference from the target gradation was a 2% increase on the percentage passing the No.
8 sieve ( 23% vs. 21%); a few screens showed a plus or minus 1% difference, but percent-age
passing No. 200 was within 0.4% or less from the target. Such minor departures
remain well within production tolerances and make this simulation more realistic,
particularly for mix design verification.
3.5 BASICS OF ESTIMATING VARIABILITY OF TEST METHODS AND
ACCEPTABLE RANGES OF TEST RESULTS
To facilitate review of the round robin results and analyses presented herein, this section
includes a brief summary of how testing variability is estimated, and how acceptable
ranges for various numbers of individual test results are established.
The basic statistic for evaluating precision of tests of construction materials is the stan-dard
deviation of the population of measurements ( test results), which is typically ex-pressed
in terms of the one- sigma limit ( 1s). 22 The one- sigma limit may be established
for single- operator precision or multilaboratory precision. Limits for multilaboratory
precision are larger due to different test operators, equipment, and laboratory
environments that provide more sources of variability or error.
22 ASTM. “ ASTM C 670- 03, Standard Practice for Preparing Precision and Bias Statements for Test
Methods for Construction Materials” ASTM Book of Standards 2006
31
Round Robin testing is required to verify the proposed Marshall mix design procedure for ADOT 413
Asphalt Rubber Asphaltic Concrete ( ARAC). To assure that sufficient AR binder is available to complete
the testing, MACTEC has prepared 5 one- gallon cans of AR binder for each participating laboratory. These
will be distributed with along bulk samples of the respective component aggregate materials and hydrated
lime mineral admixture, and copies of these instructions, the mix design procedure, pertinent information
from MACTEC’s original mix design, and blank Mix Design Data Report Form.
PLEASE READ THESE INSTRUCTIONS BEFORE PROCEEDING
1. Read the entire mix design procedure first and follow it exactly – there are some important differences
from the previous procedure for Rice specimens, and temperature control. If you have any questions,
contact Anne Stonex immediately at 602- 437- 0250 ( MACTEC), or Scott Thompson if Anne is not
available.
2. Each lab shall complete three replicates of the mix design, with one set of Rices per replicate. Please
present the results for each replicate ( 3 plugs each at 3 AR binder contents and 1 set of Rices)
separately for inclusion in the statistical analysis. A blank Mix Design Data Report Form is attached
and an electronic copy will be provided.
3. Check aggregate gradations with washed sieve analysis. Batch aggregates in oven dry condition to
meet mix design gradation targets for the respective sieve sizes.
4. Determine specific gravities ( bulk oven dry, SSD, and apparent) and absorption of the composited
coarse and fine aggregate fractions.
5. Use “ Wet Prep” method of admixture addition – mix the designated proportion of lime with the dry
aggregate, then add 3% water by aggregate weight and mix thoroughly
6. Include admixture ( added by wet prep) in the Rice specimens, and 6% AR binder by total mix weight.
7. Cure Rice specimens at the same temperature ( 330 ± 5 º F) for the same amount of time ( 2 hours) as the
loose GG AR AC mixture.
8. Batch Marshall specimens at 6.5%, 7.5%, and 8.5% AR binder content by total mix weight.
9. Mixing temperature for Marshall and Rice specimens is: AR binder @ 350 º F, aggregate @ 325 º F
10. Compaction temperature for Marshall specimens is 330 ± 5 º F
11. DO NOT place any weights on the compacted Marshall specimens.
12. Cool the compacted specimens in the molds to ≤ 90 º F before extruding them. Specimens shall be
cooled, extruded, and bulk specific gravity determined within 8 hours from the time of compaction.
13. Measure and report Marshall stability and flow.
14. For each replicate of the mix design, and for each binder content, use Asphalt Institute formulas in the
User’s Guide to calculate mixture volumetrics including: effective binder volume, VMA, VFA,
effective air voids, effective specific gravity of aggregate– admixture blend, binder absorption and
effective binder content.
15. Report results to MACTEC by no later than Monday, November 15, 2004 on the provided Mix Design
Data Report Form ( e- mail transmittal to astonex@ mactec. com is preferred).
Figure 2 Instructions For Round Robin Mix Design Testing
32
The commonly used term coefficient of variation ( COV) refers to the one- sigma limit in
percent ( 1s%) and is sometimes used as the basis of precision statements for physical
tests. The COV is calculated by dividing the standard deviation ( 1s) by the average of
the test results and multiplying by 100%.
The acceptable difference between two test results for construction materials has been
standardized as the difference two sigma limit ( d2s), which is calculated by multiplying
1s by 2√ 2 rounded to 2.83. The acceptable difference expressed in percent ( d2s%) is
simply 1s% multiplied by 2.83. The level of confidence for d2s is 95%, which means
that this difference would be exceeded on average no more than once in 20 correctly
performed tests.
ASTM C 67023 includes a table of multiplier factors to use for numbers of test results
ranging from 2 through 10; the multiplier increases as the number of test results increase.
Therefore, this ASTM procedure cautions that an index of precision ( d2s) based on the
difference of two results should not be applied to cases where more than two results are
compared. However if differences among more than two results fall within the narrower
acceptable range for two results, the resulting testing precision is well within the
acceptable range.
ADOT supplied multilaboratory statistics ( 1s, d2s, 1s%, d2s%) from the last 10 years of
their asphaltic concrete proficiency sample program for information. MACTEC also
reviewed multilaboratory and single operator Marshall Proficiency Sample Program
( PSP) statistics presented on the AASHTO Materials Reference Laboratory ( AMRL)
website and in the study “ Effects of Test Variability on Mixture Volumetrics and Mix
Design Verification” by Hand and Epps24 to evaluate the quality of the testing performed.
Analyses of precision of test results obtained for this study are primarily concerned with
acceptable differences between two or more laboratories, rather than for a single operator.
However to evaluate possible problems with test performance, replicate results from the
respective participating laboratories for bulk and maximum theoretical specific gravities
were reviewed with respect to single operator precision information. The ranges of
results were within acceptable limits compared to precision statements and ranges of
available Marshall proficiency sample program results, and no problems were identified.
3.5.1 Considerations Regarding Volumetric Calculations and Analysis
The ultimate products of the mix design procedure are loose mix specimens for Rice
determination and a series of compacted Marshall specimens at designated binder
contents, for which bulk density, stability and flow are measured. Each activity involved
in making and testing these mix specimens is a possible source of variation or error
which may be reflected in the final test results. These activities include materials
sampling, sieve analysis and batching, mixing aggregates with admixture and AR binder,
23 Ibid
24 Hand, Adam J. and Amy Epps. “ Effects of Test Variability on Mixture Volumetrics and Mix Design
Verification.” Journal of the Association of Asphalt Paving Technologists, Vol. 69, pages 635- 674, 2000.
33
and conditioning, compacting, and testing the resulting mix specimens. The AR binder
may introduce additional variability.
Volumetric properties including effective binder volume, air voids content, VMA, and VFA,
are calculated rather than measured. Marshall stability and flow are not volumetric
properties and are of limited interest for AR- AC materials. AR binder content is
controlled in the laboratory along with aggregate gradation. As pointed out by Hand and
Epps, 25 direct property measurements are limited to the following tests, of which each has
its own range of variability:
• Asphalt cement specific gravity ( Gb).
• Combined aggregate specific gravity ( Gsb).
• Bulk specific gravity of compacted Marshall specimens ( Gmb).
• Maximum theoretical specific gravity of the mix ( Gmm).
Because of these considerations, two approaches were used to evaluate the round robin
data. For preliminary evaluation, AR- AC mixture volumetric properties were calculated
for each laboratory’s replicates based on the corresponding aggregate specific gravities
and absorption, and respective Rice and Gmb results supplied. The compiled results are
listed and plotted in Appendix F, which also includes the statistical analysis using
ANOVA, and groups and ranks mean results for the volumetric properties, Marshall
stability and flow.
The second approach was to normalize the data for analysis by using single values for
Gsb, absorption, and Gmm for volumetric calculations for each laboratory’s data. It was
decided that the most representative values would be the overall averages of the values
for Gsb, absorption, and Gmm measured by the laboratories.
3.6 ROUND ROBIN TEST RESULTS
The results of round robin testing and analyses are presented in Appendices F
( preliminary) and G ( normalized). As customary for round robin exercises, the names of
the laboratories have been coded as A, B, C, and D. Each laboratory determined specific
gravities ( bulk oven dry, SSD, and apparent) and absorption of the composited coarse and
fine aggregate fractions. These results are compiled and presented in Table 10. Labs A
and C submitted the aggregate and Rice results, along with Marshall specimen results for
bulk specific gravity, stability and flow, but did not perform the requested volumetric
calculations.
The non- normalized volumetric results for each laboratory were calculated based on the
individual laboratory’s aggregate results, Rice results, and the calculations in the User’s
Guide. These are compiled and plotted in Appendix F. The overall values in the
25 Ibid
34
rightmost column of Table 10 were used to normalize the aggregate results, except that
the numerical overall average for water absorption ( 2.08%) was slightly lower than, and
thus replaced with, the corresponding calculated value of 2.14%.
Table 10 Compiled Round Robin Results for Aggregate Specific Gravity
Laboratory MACTEC D B A C
Source of Data
Original
Mix
Design
Round
Robin
Round
Robin
Round
Robin
Round
Robin
Overall
Round
Robin
" Average"
Coarse Aggregate
Bulk OD Specific
Gravity 2.744 2.731 2.750 2.765 2.743 2.747
SSD Sp. Gravity 2.786 2.783 2.798 2.811 2.794 2.797
Apparent Specific
Gravity 2.886 2.879 2.888 2.897 2.89 2.889
Water Absorption 1.55% 1.88% 1.74% 1.66% 1.85% 1.78%
Fine Aggregate
Bulk OD Specific
Gravity 2.719 2.682 2.722 2.695 2.708 2.702
SSD Specific
Gravity 2.778 2.761 2.782 2.765 2.79 2.775
Apparent Specific
Gravity 2.889 2.912 2.896 2.900 2.951 2.915
Water Absorption 2.17% 2.94% 2.21% 2.63% 3.05% 2.71%
Combined Coarse & Fine without Mineral Admixture
Bulk OD Specific
Gravity 2.735 2.713 2.739 2.740 2.731 2.731
SSD Specific
Gravity 2.783 2.775 2.792 2.794 2.793 2.789
Apparent Specific
Gravity 2.874 2.891 2.891 2.898 2.911 2.898
Water Absorption 1.77% 2.29% 1.89% 2.00% 2.14% 2.08%
Compiled Rice results are presented in Table 11, along with related precision calculations
for the round robin testing. The precision statement for ASTM D 2041 for single
operator, dry back procedure cites a “ 1s” value of 0.0064 for the bowl method. Although
the ADOT method uses flasks, this is the only available comparison for a single operator.
Based on this value, the allowable difference among three results would be 3.3( 0.0064) =
0.0211, and the allowable difference among six results ( Lab A) would be 4.0( 0.0064) =
0.0256. The results in Table 11 are within these ranges. The overall average Rice value
of 2.512 was used to normalize volumetric calculations.
35
Table 11 Compiled Round Robin Results for Rice at 6.0% AR Binder Content
Laboratory MACTEC D B A C
Rice Results
Original
Mix
Design*
Round
Robin
Round
Robin
Round
Robin
Round
Robin
Rice 1 2.516 2.507 2.505 2.522 2.533
Rice 2 2.519 2.499 2.509 2.517 2.520
Rice 3 2.523 2.497 2.499 2.497 2.525
Rice 4 2.515
Rice 5 2.507
Rice 6 2.509
Rice Precision Calculations Overall
Average 2.519 2.501 2.504 2.511 2.526 2.512
Standard Deviation
( 1s) 0.0035 0.0053 0.0050 0.0088 0.0066 0.0106
d2s 0.0099 0.0150 0.0142 0.0250 0.0186 0.0299
COV ( 1s%) 0.139 0.212 0.201 0.351 0.260 0.421
d2s% 0.394 0.599 0.569 0.994 0.735 1.190
* Original mix design used different AR binder than Round Robin
Table 12 presents additional comparisons for Rice testing, including ranges of average
Rice results gleaned from AMRL and ADOT Proficiency Sample Program ( PSP)
multilaboratory statistics, along with the corresponding precision statistics from
ASTM D 2041- 03a, with and without dry back. The multilaboratory ASTM statistics
may include results from bowls and flasks, which may account for some of the
differences from ADOT PSP data.
Table 12 Multilaboratory Proficiency Sample Program Ranges for Rice Results
Range of Results
AMRL
Gmm
Results
ADOT
Gmm Results
ADOT
MAX
Density
ASTM D 2041- 03a
Precision for
2 results
Average 2.417- 2.591 2.420- 2.460
Dryback
( Bowl
only)
No
Dryback
1 Standard
Deviation 0.011- 0.020 0.012- 0.0243 0.0193 0.016
2 Standard
Deviations 0.031- 0.057 0.033- 0.069 0.055 0.044
Coefficient Of
Variation ( 1s%) 0.43- 0.84 0.477- 0.988 0.38- 0.99
Coefficient Of
Variation ( 2s%) 1.27- 2.37 1.349- 2.795 1.08- 2.80
36
Laboratory A experienced problems with their Marshall hammer during round robin
testing. It is not clear if these problems were resolved before round robin testing was
completed, but their Marshall compaction equipment was subsequently replaced. Lab A
asked for additional samples of materials to make and test additional replicates, and
submitted data for eight sets of replicates. These results were checked for outliers
according to ADOT methods. No outliers were identified, although one data point was
right at the upper outlier limit. Thus results for each of the 8 replicates were included in
the statistical analysis. This unbalanced the experimental design, but it does not appear to
have interfered with the One- Way ANOVA analysis.
For each laboratory, results of aggregate bulk specific gravity ( Gsb) testing were also
reviewed. Standard deviations were calculated for combined sets of replicate plugs at
each of the three AR binder contents, and are shown on the compiled data sheets in
Appendix F for each participating laboratory. Gsb is directly measured, so these values
were not affected by normalizing the data for volumetric calculations. Because these
specimens were to be tested for stability and flow, no paraffin or parafilm could be used.
This factor would be expected to increase variability of Gsb measurement, particularly
for specimens with relatively high air voids contents. The ranges of standard deviations
within each laboratory are compiled in Table 13. The within laboratory results are
considered equivalent to single operator precision for this comparison, although in some
cases more than one person performed the testing. Comparisons of within laboratory
standard deviations with AASHTO Materials Reference Library ( AMRL) statistics for
ASTM D 2726- 00 do not indicate any serious or systematic problems with the precision
of the round robin Gsb testing.
Table 13 Within Laboratory Standard Deviation ( 1s) Ranges of Gsb Results
Lab ID Number of
Replicates
Round Robin
Range of 1s values
ASTM
D 2726- 00
A 9
6
0.007- 0.015
0.025
B 9 0.007- 0.011
C 9 0.006- 0.008
D 9 0.009- 0.020
Single Operator
1s limit= 0.0124
2 sample
d2s limit = 0.035
ASTM D 2726- 04 provides precision data only for mixes made with aggregates with
water absorption less than 1.5%, which does not apply to the highly absorptive Dugas
aggregate used in the round robin. Although the single operator precision limits for
nominal ¾ - inch mixes are very similar to those listed in Table 13, the multilaboratory
limits are much tighter for low absorption aggregates. A multilaboratory comparison of
precision of test results is included in Appendix F which supports that Gsb testing among
the respective laboratories was generally performed within acceptable limits.
Preliminary analysis of this round robin experiment indicated that at least two of the
means differed for each property of interest at each AR binder content, except for
Marshall stability at 6.5 and 7.5% AR content. When at least two means were found to
differ, Duncan’s Multiple Range Test was used to compare and rank the respective
37
means, to identify which means were statistically similar and which differed. The
Duncan test can be applied to unequal sample sizes. 26 The Summary of Duncan’s
Multiple Range Comparisons is presented graphically in Appendix F. Lines are used to
group like means and distinguish among groups. Results for Labs A and C were often
similar to each other, while Labs B and D often grouped with each other.
To evaluate the practical differences among the results, design AR contents were
determined for the respective AR- AC mix design replicates and are presented in Table
14. Labs C and D would have selected AR contents of 8.5% to meet mix design air voids
criteria of 5.5% ± 1%, while Lab B’s data would allow slightly lower AR contents of
8.0% to 8.3%. Lab A did not achieve the design air voids requirements within the given
range of AR contents, which may be related to the previously noted equipment problems.
Table 14 Preliminary AR Content Selection
Lab Set B C D A
No. % AR,
% Air voids
% AR,
% Air voids
% AR,
% Air voids
% AR,
% Air voids
1 8.2% AR,
5.5% AV
8.5% AR,
6.5% AV
8.5% AR,
5.6% AV
8.5% AR,
7.1% AV
2 8.3% AR,
5.6% AV
8.5% AR,
6.2% AV
8.5% AR,
5.6% AV
At 7.5 and 8.5% AR,
6.8% AV
3 8.0% AR,
5.4% AV
8.5% AR,
6.2% AV
8.5% AR,
5.7% AV
8.5% AR,
6.8% AV
1R 8.5% AR,
9.1% AV
2R 8.5% AR,
8.6% AV
3R 8.5% AR,
9.0% AV
4 At 7.5 and 8.5% AR,
8.6% AV
5 8.5% AR,
9.0% AV
Normalizing the results removed some of the noise from the data, and results converged
so that statistical differences were eliminated from VMA at 6.5 and 7.5% AR content,
from VFA at 6.5% AR, and effective air voids at 6.5% AR. The normalized results are
compiled and plotted in Appendix G, along with ANOVA and the Summary of Duncan’s
Multiple Range Comparison tests. When there was a difference in means, results from
Labs A and C still tended to group together and results from Labs D and B generally
continued to form a second group. However normalizing had no effect on the measured
values for Marshall Gsb, stability, or flow.
26 Montgomery, Douglas C. Design and Analysis of Experiments Second Edition. John Wiley & Sons,
1984. pp 66- 68
38
To evaluate the practical effects of normalizing the data, AR contents were selected based
on the normalized results and determinations are presented in Table 15.
Table 15 Normalized AR Content Selection
Lab Set B C D A
No. % AR,
% Air voids
% AR,
% Air voids
% AR,
% Air voids
% AR,
% Air voids
1 8.4% AR,
5.4% AV
8.5% AR,
5.8 % AV
8.5% AR,
5.8 % AV
8.5% AR,
6.8% AV
2 8.5% AR,
5.4% AV
8.5% AR,
5.9 % AV
8.5% AR,
6.1 % AV
At 7.5 and 8.5% AR,
6.6% AV
3 8.3% AR,
5.5% AV
8.5% AR,
6.0 % AV
8.5% AR,
6.2 % AV
8.5% AR,
7.4 % AV
1R 8.5% AR,
9.1% AV
2R 8.5% AR,
8.7 % AV
3R 8.5% AR,
9.3 % AV
4 At 7.5 and 8.5% AR,
8.6% AV
5 8.5% AR,
9.1% AV
For Labs C and D, the range of voids at 8.5% AR converged; the voids for Lab C drop-ped
and those for Lab D increased. For Lab B, the selected AR content shifted from 8.0-
8.3% to 8.3- 8.5% to correspond more closely with results from Labs C and D. Lab A
results were based on values that were close to the overall averages selected for normal-izing
the data so little change was achieved. Lab A results did not meet the ADOT
design criterion for effective air voids, which may be related to the compactor problems
encountered. However results of the other 3 participating labs are in close agreement.
3.7 ADDITIONAL CONSIDERATIONS
3.7.1 Laboratory Technicians and Equipment
Although the round robin results reported herein have been coded as customary to protect
the participants, there is some additional information that should not be omitted from the
analysis. Technician experience with the highly modified AR- AC materials appears to be
a factor in repeatability ( within lab) and reproducibility ( between laboratories) in the
design procedure.
During the round robin phase of this study, Lab A not only had major problems with
Marshall hammer calibration, but also lost the technicians who had the most experience
with working with AR- AC mixtures. Lab C, whose results often grouped closely with
those of Lab A, routinely performed conventional mix design testing but had relatively
39
limited experience in designing AR- AC mixes. Labs B and D, whose results also tended
to group closely together and often differed from the other two labs, had fairly extensive
experience in designing AR binders and AR- AC mixes.
3.7.2 Field Performance
Although ADOT AR- AC mixes have historically performed well, sections of the subject
AR- AC mixture and several others constructed in 2004 experienced significant failures.
AMEC Earth & Environmental, Inc. evaluated three of these AR- AC projects including
Big Bug for ADOT and determined that the primary cause was moisture susceptibility
due to high in- place air voids. 27
The subject AR- AC mixture for the Big Bug project was placed on the north and south
bound lanes of I- 17 between mileposts 263 and 256 at night from September 1 to October
4, 2004. The AR- AC was placed at a nominal compacted thickness of two inches on a
new replacement layer in accordance with ADOT 417. Results of acceptance tests
indicated that AR binder content and aggregate gradation were generally within limits. 28
In- place compaction was not an acceptance requirement for AR- AC mixes at that time.
The AR- AC was surfaced with a nominal 2/ 3- inch thick layer in accordance with ADOT
414 Asphaltic Concrete Friction Course ( Asphalt- Rubber), which failed rapidly by
raveling during the winter and was replaced in spring 2005. Additional distress, including
rutting and potholes, developed during summer 2005 that was related to the AR- AC
rather than the friction course. Areas of the AR- AC mix stripped severely, particularly in
the southbound lanes. Although it is clear that water entered the AR- AC layer, questions
remain as to why the water did not drain out.
Forensic data from the failure investigation by AMEC included air voids contents of 31
cores obtained from this project that ranged from 4.9 to 10.8%, with an average of 8.1%.
Four cores had 6.0% air voids or less; three had 10.0% air voids or more.
At this time, ADOT agrees with AMEC that the observed moisture damage in the
projects reviewed is most likely due to inadequate compaction. Marginally low ambient
temperatures during and immediately after construction are considered to be a primary
reason that compaction was not achieved. Night paving at higher elevations conflicts with
the need for relatively high placement and compaction temperatures.
In an effort to avoid such failures in the future, ADOT has implemented a new
specification for AR- AC: in ADOT 41529 Asphaltic Concrete ( Asphalt- Rubber)- End
Product. ADOT 415 adds compaction requirements, including a target of 7.0% in- place
air voids, with Upper Limit of 9.0% and Lower Limit of 4.0% in- place air voids. AMEC
applied these requirements in its forensic analysis and found that the failing materials
were not in compliance, which supports the value of the density requirements.
27 Hanson, Douglas I. and Joseph Phillips. “ Forensic Analysis Asphalt Rubber Asphalt Concrete ( ARAC)”
Report No. 1, AMEC Earth & Environmental, Inc., Phoenix, AZ, May 18, 2006.
28 Ibid
29 ADOT. Standard Specifications for Road and Bridge Construction 2000. Section 415
40
3.7.3 Resistance to Moisture Damage
Neither the ADOT 415 AR- AC End Product specification nor the proposed laboratory
mix design procedure addresses testing to evaluate resistance to moisture damage. There
are some issues to be addressed in determining what method and limits to use for such
testing. The standard immersion- compression test is not appropriate for AR- AC
materials, as the unconfined AR- AC specimens slump and deform during conditioning.
AMEC and others have suggested consideration of tensile strength ratio as a criterion for
evaluating resistance to moisture damage. However, further research is needed to assess
whether this approach will do a better job of predicting AR- AC resistance to moisture
damage than it did when ADOT evaluated use of such tests for predicting susceptibility
of conventional asphaltic concrete mixes to moisture damage.
3.7.4 Draft ARIZ 832 ( October 17, 2006) Marshall Method for AR- AC
The proposed mix design method is currently designated as Draft ARIZ 832 ( October 17,
2006) Marshall Mix Design Method for Asphaltic Concrete ( Asphalt- Rubber) [ AR- AC].
It is presented in Appendix H. Technical changes from Version 2 used in the round robin
primarily consist of reducing temperatures for mixing ( aggregate at 325 ± 3° F instead of
330 ± 5° F), and for curing and compaction ( 300 ± 5° F instead of 330 ± 5° F). Other
changes were made to improve clarity and presentation of the text and calculations. The
October 17 draft is currently under review by ADOT and industry and may be revised
during the approval process. Further refinements may be suggested as the AR- AC mix
design procedure is implemented and used, and may include addition of some method of
evaluating resistance to moisture damage.
Decreasing the mixing and compaction temperatures from that used in the Big Bug round
robin may have some related effects on mixture volumetrics. The increased AR binder
stiffness at lower temperatures is likely to increase the air voids contents measured in the
mix design, which would increase design AR binder content. High AR binder contents
are intrinsic to the performance properties of the desired product, as long as they are not
excessively high.
What is most important is that future AR- AC mixes designed according to this procedure
are able to provide the same enhanced performance properties that ADOT has grown to
expect from the pre- 2004 mixes.
41
4. CONCLUSIONS
Based on the results of testing performed in Rounds 1 and 2, and results of the Round
Robin, Draft ARIZ 832 ( October 17, 2006) appears to be an acceptable and appropriate
procedure for the intended purpose. Although mix design results are somewhat variable,
evaluation of the statistics for the same tests applied to conventional asphaltic concrete
materials indicates the measured variability is very similar.
It does not appear that using asphalt- rubber binder makes the testing of the AR- AC
mixtures significantly more variable than the testing of conventional or polymer modified
asphaltic concrete materials. This was a major concern during this study. No extra
laboratory equipment will be required to perform ARIZ 832. However, as for any
bituminous material, experience, properly operating equipment, and good practices are
required to achieve representative results. Additional training may be appropriate for
technicians who are not experienced in working with AR- AC materials.
The most substantial changes from the previous modified ADOT 815c30 AR- AC mix
design procedure are in the preparation and treatment of the Rice specimens. AR- AC
Rice specimens will include mineral admixture and no liquid antistrip will be added. Rice
specimens will be prepared at 6.0% AR binder content and cured at the same time and
temperature as the loose Marshall specimens. Temperatures for mixing, and for curing
and compacting AR- AC specimens have been modified and the allowable ranges are now
tighter to reduce variability. Volumetric calculations are performed according to national
standards. Rebound is now addressed: no confining weights will be used to prevent
specimen rebound, and if rebound is observed after compaction, the specimens will be
discarded and the target aggregate gradation will be adjusted to better accommodate the
AR binder.
Implementation of ARIZ 832 and ADOT 415 began on a limited basis during the 2006
construction season. It appears that there is a “ learning curve” involved in meeting AR-AC
compaction requirements. A combination of favorable ambient temperatures, proper
equipment, and good practices for materials handling and equipment operation are
needed to meet the requirements.
This study has documented that the asphalt- rubber binder is a major factor in AR- AC
volumetrics. This supports experience and practical observations by ADOT personnel
and others who have been involved in AR- AC mix design. Finer rubber gradations in the
AR binder are likely to facilitate AR- AC mix design. Coarse rubber gradations in the AR
binder may interfere with establishing an appropriate aggregate matrix ( target gradation)
and may not permit development of a suitable AR- AC mix design. If this occurs, the first
alternate should be to try using a binder made with a finer rubber gradation. However in
cases where suitably fine crumb rubber is not available, adjustment of the aggregate
gradation may be necessary.
30 Ibid. Section 815c
42
43
APPENDIX A
EXISTING MODIFICATIONS TO ARIZ 815C31
USED FOR AR- AC MIX DESIGNS UNTIL 2006
( VERSION 5- 28- 03)
31 Ibid
Existing Modifications to ARIZ 815c Currently Used for Asphalt- Rubber Mix Designs Version 5- 28- 03
44
Note: This document describes the existing modifications to the ARIZ 815 mix design
procedure that ADOT currently uses in design of Section 413 Asphaltic Concrete
( Asphalt- Rubber) mixes. No changes were made to Figures 1 through 11 that remain in
current use but are not attached to this version for ease of transmittal. MACTEC’s
recommended revisions to ARIZ 815c for use in the proposed mix design procedure
being developed for GAP- Graded Asphalt Rubber Concrete will be presented in a
separate document.
ARIZ 815c
Modified for Asphaltic Concrete ( Asphalt- Rubber)
May 2003
( 23 Pages including Figures 1 through 11)
MARSHALL MIX DESIGN METHOD
FOR ASPHALTIC CONCRETE ( ASPHALT- RUBBER)
( A Modification of AASHTO T 245)
Scope
1. This method is used to design Section 413 Asphaltic Concrete ( Asphalt- Rubber)
mixes using four- inch Marshall apparatus.
Apparatus
2. The apparatus necessary includes all items required to perform the individual test
methods referred to in this procedure as follows:
ARIZ 201c Sieving of Coarse and Fine Graded Soils and Aggregates
ARIZ 210b Specific Gravity and Absorption of Coarse Aggregate
ARIZ 211c Specific Gravity and Absorption of Fine Aggregate
ARIZ 410c Compaction and Testing of Bituminous Mixtures Utilizing Four- Inch
Marshall Apparatus ( see AASHTO T 245 for required equipment)
ARIZ 415b Bulk Specific Gravity of Compacted Bituminous Mixes
ARIZ 806e Maximum Theoretical Specific Gravity of Laboratory Prepared
Bituminous Mixtures ( Rice Test).
Materials
3. ( a) Mineral Aggregate - The mineral aggregate for the asphaltic concrete
shall be produced material from the source( s) for the project. Use of natural sand is not
permitted in asphalt- rubber mixtures.
1) Mineral aggregate from each source shall be tested for compliance to the
project requirements for Abrasion ( AASHTO T 96).
Existing Modifications to ARIZ 815c Currently Used for Asphalt- Rubber Mix Designs Version 5- 28- 03
45
2) The mineral aggregate shall be combined using the desired percentages of
the different produced materials.
3) The composite blend of mineral aggregate shall be tested for compliance to
the grading limits in Table 413- 2 of the specifications according to ( ARIZ
201) Gradation, modified so that the No. 8 sieve is the smallest coarse
sieve.
4) The composite blend of mineral aggregate shall conform to the requirements
of Table 413- 3 of the specifications for Sand Equivalent ( AASHTO T 176)
and for Crushed Faces ( ARIZ 212)
( b) Bituminous Material - The bituminous material used in the design shall be the
asphalt- rubber conforming to the requirements of Section 1009 of the specifications,
which is to be used in the production of the asphaltic concrete. No dilution with extender
oil, kerosene, or other solvents is allowed. The specific gravity of the bituminous
material sha