<%BANNER%>

Evaluation of Hybrid Binder for Dense and Open-Graded Asphalt Mixtures

Permanent Link: http://ufdc.ufl.edu/UFE0041190/00001

Material Information

Title: Evaluation of Hybrid Binder for Dense and Open-Graded Asphalt Mixtures
Physical Description: 1 online resource (136 p.)
Language: english
Creator: Li, Weitao
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: crm, dcse, dsr, healing, hybrid, idt, resilient, sbs
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Binder and mixture tests were performed to evaluate the relative performance of a PG 67-22 base binder and six other commercially available binders produced by modifying the same base binder with the following modifiers: one Styrene Butadiene Styrene (SBS) polymer, three commercially available hybrid binders composed of different percentages of rubber and SBS polymer, and two asphalt rubber binders (5% and 12 % rubber: ARB-5 and ARB-12). Results indicated that hybrid binders (modified with more rubber than SBS) that exceed the cracking performance characteristics of unmodified binder and asphalt rubber binders, and have about the same cracking performance characteristics of SBS polymer modified binder can be produced commercially. Results also indicated that hybrid binder can be suitably specified using existing specification requirements for PG76-22 binder and solubility. Therefore, it appears that hybrid binder has the potential to replace three binders currently used by FDOT in hot mix asphalt: SBS polymer modified asphalt, ARB-5, and ARB-12. It was recommended that FDOT develop a transition plan to accomplish this. The research also showed that existing binder tests do not accurately predict cracking performance at intermediate temperatures, even in a relative sense. A further Healing testing was performed on the dense-graded granite mixtures to evaluate the healing potential of hybrid binders compared to base binders. Different testing procedures were carried out and analyzed to give a better understanding of healing mechanisms. Results from healing test agreed with those from cracking performance evaluation. However, more healing parameters and procedure need to be fully developed to capture healing characteristics more specifically. It was recommended that FDOT pursue development and evaluation of the healing test.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Weitao Li.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Roque, Reynaldo.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041190:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041190/00001

Material Information

Title: Evaluation of Hybrid Binder for Dense and Open-Graded Asphalt Mixtures
Physical Description: 1 online resource (136 p.)
Language: english
Creator: Li, Weitao
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: crm, dcse, dsr, healing, hybrid, idt, resilient, sbs
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Binder and mixture tests were performed to evaluate the relative performance of a PG 67-22 base binder and six other commercially available binders produced by modifying the same base binder with the following modifiers: one Styrene Butadiene Styrene (SBS) polymer, three commercially available hybrid binders composed of different percentages of rubber and SBS polymer, and two asphalt rubber binders (5% and 12 % rubber: ARB-5 and ARB-12). Results indicated that hybrid binders (modified with more rubber than SBS) that exceed the cracking performance characteristics of unmodified binder and asphalt rubber binders, and have about the same cracking performance characteristics of SBS polymer modified binder can be produced commercially. Results also indicated that hybrid binder can be suitably specified using existing specification requirements for PG76-22 binder and solubility. Therefore, it appears that hybrid binder has the potential to replace three binders currently used by FDOT in hot mix asphalt: SBS polymer modified asphalt, ARB-5, and ARB-12. It was recommended that FDOT develop a transition plan to accomplish this. The research also showed that existing binder tests do not accurately predict cracking performance at intermediate temperatures, even in a relative sense. A further Healing testing was performed on the dense-graded granite mixtures to evaluate the healing potential of hybrid binders compared to base binders. Different testing procedures were carried out and analyzed to give a better understanding of healing mechanisms. Results from healing test agreed with those from cracking performance evaluation. However, more healing parameters and procedure need to be fully developed to capture healing characteristics more specifically. It was recommended that FDOT pursue development and evaluation of the healing test.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Weitao Li.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Roque, Reynaldo.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041190:00001


This item has the following downloads:


Full Text

PAGE 1

1 EVALUATION OF HYBRID BINDER FOR DENSEAND OPEN -GRADED ASPHALT MIXTURES By WEITAO LI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE O F DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Weitao Li

PAGE 3

3 To my be loved parents, Shuling Li and Yuying Meng

PAGE 4

4 ACKNOWLEDGMENTS It is a great pleasure to thank those people and organizations who offer ed their self -giving assistance t o this research and finally made this dissertation possible. First of all, I am heartily thankful to my supervisor, Dr. Reynaldo Roque, whose encouragement, supervision and support from the preliminary to the concluding level enabled me to develop a deep understanding of the subject. Without your knowledgeable guidance and mentoring, it would have been impossible for me to finish my doctoral work Also I owe my great gratitude and thankfulness to my committee members: Dr. Mang Tia, Dr. Dennis R. Hiltunen a nd Dr. Bhav ani V. Sankar for their invaluable support and advices on accomplishing my work. Thirdly, special thanks go to Florida Department of Transportation (FDOT) for providing financial support, testing instrumentation, materials that made this resear ch possible. I would like to specifically thank Gale Page, David Webb, Aaron Turner, and Mabel Stickles for their help; their efforts are sincerely appreciated. I would also like to extend my thanks to Frank Fee from C ITGO Petroleum for his assistance in obtaining the control binders for this study, and to the three hybrid binder producers for their time and efforts in producing their different products for this study In addition, I would like to express my appreciation to Mr. George Lopp for his kind help with material preparation, testing setup and many other technical supports for this research I also thank Tianying Niu for his hard work and contribution to this project and Dr. Alvaro Guarin for his assistance in helping organiz e part of th e final forma t of this dissertation. L ast ly I thank all of my colleagues for spending these wonderful years together.

PAGE 5

5 TABLE OF C ONTENTS page ACKNOWLEDGMENTS ...................................................................................................... 4 LIST OF T ABLES ................................................................................................................ 7 LIST OF FIGURES .............................................................................................................. 9 LIST OF ABBREVIATIONS .............................................................................................. 13 ABSTRACT ........................................................................................................................ 15 CHAPTER 1 INTRODUCTION ........................................................................................................ 17 Background ................................................................................................................. 17 Objectives ................................................................................................................... 20 Scope .......................................................................................................................... 21 2 LITERATURE REVIEW .............................................................................................. 23 Crumb Rubber Used a s Asphalt Binder Modifie r ...................................................... 23 Polymers Used a s Asphalt Binder Modifier ............................................................... 25 Development of Hybrid Binders in Recent Years ...................................................... 26 Methodologies of Evaluating Asphalt Mixtures Healing ............................................ 28 Pseudo Stiffness Method ..................................................................................... 28 Ratio of Di ssipated Energy Change (RDEC) Method ......................................... 30 Dissipated Creep Strain Energy (DCSE) Method ............................................... 32 Elastic Deformation and Strain Recovery ........................................................... 34 Summary ..................................................................................................................... 36 3 MATERIALS AND METHODS ................................................................................... 38 Binders ........................................................................................................................ 38 Aggregates .................................................................................................................. 42 Mixtures ....................................................................................................................... 45 Mixture Preparation .................................................................................................... 46 4 BINDER TEST RESULTS AND ANALYSIS .............................................................. 50 Binders Physical Properties ....................................................................................... 50 Specific Gravity of Bi nders ................................................................................... 50 Solubility Analysis ................................................................................................ 51 Mass of Volatiles Loss Analysis .......................................................................... 52 Binders Rheological Properties .................................................................................. 53

PAGE 6

6 DSR Test Results Analysis .................................................................................. 53 BBR Test Results Analysis .................................................................................. 57 Multiple Stress Creep Recovery (MSCR) .................................................................. 59 Elastic Recovery ......................................................................................................... 64 Force Ductility Test ..................................................................................................... 65 5 MIXTURE IDT TEST RESULTS AND ANALYSIS .................................................... 70 Mixture Test Results ................................................................................................... 70 Analysis of IDT Test Results ...................................................................................... 77 DG Mixtures ......................................................................................................... 77 OGFC Mixtures .................................................................................................... 81 Summary ..................................................................................................................... 82 6 DEVELOPMENT AND EVALUATION OF HEALING TEST ..................................... 87 Experimental and Theoretical Background of Healing Test ...................................... 87 Fatigue Test with Static and Cyclic Loading ....................................................... 87 Damage and Healing ........................................................................................... 89 Healing Test D evelopment ......................................................................................... 91 Cyclic Damage Loading (CDL) Mode .................................................................. 91 Loading Amplitude Verification ............................................................................ 97 Healing Test Program ........................................................................................ 101 Data Acquisition ................................................................................................. 101 Materials Prepared for Healing Test .................................................................. 104 Healing Test Results Analysis .................................................................................. 105 Damage Analysis during CDL ........................................................................... 105 Healing Analysis of D G Mixtures ....................................................................... 111 Summary ................................................................................................................... 115 7 CLOSURE AND RECOMMENDATIONS ................................................................ 117 Sum mary ................................................................................................................... 117 Conclusions .............................................................................................................. 119 Recommendations .................................................................................................... 1 20 APPENDIX A BINDER T EST RESULTS ........................................................................................ 122 B HEALING TEST RESULTS ...................................................................................... 129 C CITGO CERTIFICATES OF ANALYSIS .................................................................. 132 LIST OF REFERENCES ................................................................................................. 134 BIOGRAPHICAL SKETCH .............................................................................................. 136

PAGE 7

7 LIST OF TABLES Table page 3 -1 Asphalt binder and the constituents/formulations ................................................. 40 3 -2 Binder tests summary ............................................................................................ 41 3 -3 Aggregate source ................................................................................................... 42 3 -4 DG mixtures IDs for testing .................................................................................... 46 3 -5 OGFC mixtures IDs for testing ............................................................................... 46 3 -6 Dense graded mixture volumetric information ....................................................... 49 3 -7 OGFC mixture volumetric information ................................................................... 49 4 -1 Specific gravity of binders 15.6 C (6 0 F) ............................................................. 51 5 -1 Summary of total mixture tests .............................................................................. 70 5 -2 DG mixtures creep and damage test results ......................................................... 71 5 -3 DG mixtures strength and fracture test results ...................................................... 72 5 -4 DG mixtures energy ratio results ........................................................................... 73 5 -5 OGFC mixtures creep and damage test results .................................................... 74 5 -6 OGFC mixtures strength and fracture test results ................................................ 75 5 -7 OGFC mixtures energy ratio results ...................................................................... 76 6 -1 Failure criteria for fatigue test .............................................................................. 100 6 -2 Materials for healing test ...................................................................................... 105 6 -3 Poissons ratio comparison .................................................................................. 107 6 -4 Comparison of DCSE and MR reduction at 20 minutes CDL ............................. 115 A-1 G*/s C (152.6 F) ......................................................................... 122 A-2 C (158 F) ............................................................................. 122 A-3 C (168.8 F) ......................................................................... 122 A-4 C (179.6 F) ......................................................................... 122

PAGE 8

8 A-5 C (190.4 F) ......................................................................... 123 A-6 C (194 F) ............................................................................ 123 A-7 C (77 F) ................................................................................ 123 A-8 C (71.6 F) ............................................................................ 123 A-9 C (66.2 F) ............................................................................ 124 A-10 C (60.8 F) ............................................................................ 124 A-11 BBR test results at 12 C (10.4 F) ...................................................................... 124 A-12 BBR test results at 18 C (0.4 F) ........................................................................ 125 A-13 Multiple stress creep recover y, %, RTFOT residue ............................................ 125 A-14 Non -recoverable creep compliance, kPa1, RTFOT residue ............................... 125 A-17 Elastic recovery at 25 C (77 F) (R TFOT residue) .............................................. 125 A-18 Force ductility test result ...................................................................................... 126 A-19 Smoke point, flash point and solubility of original binders .................................. 126 A-20 Mass loss after RTFOT at 163 C (325.4 F) ........................................................ 126

PAGE 9

9 LIST OF FIGURES Figure page 1 -1 Waste tires use history in Florida ........................................................................... 18 2 -1 Typical effect of healing due to rest period ........................................................... 30 2 -2 Healing test results ( 1,000 cycles, 75 psi, 15 C ) .................................................. 34 2 -3 Normalized healing ( after 1000 cycles, 75 psi, 15 C ) .......................................... 34 2 -4 Load controlled cyclic loading and rest periods .................................................... 35 2 -5 A sample plot of the strain versus time during the loading/rest period ................ 36 3 -1 DG granite gradation .............................................................................................. 43 3 -2 DG limestone gradation ......................................................................................... 43 3 -3 OGFC granite gradation ......................................................................................... 44 3 -4 OGFC limestone gradation .................................................................................... 44 3 -5 Mixture testing plan for each mixture and aggregate type .................................... 45 3 -6 Pill contained with mesh ........................................................................................ 48 3 -7 CoreLok sample weighing equipments ................................................................. 49 4 -1 Solubility of original binders ................................................................................... 51 4 -2 RTFO T, mass loss at 163 C (325.4 F) ................................................................. 52 4 -3 C (168.8 F) @ 10 rad/s ...................................................................... 55 4 -4 C (168.8 F) @ 10 rad/s .......................................................... 55 4 -5 C (77 F) @ 10 rad/s ............................................................................ 56 4 -6 C (77 F) @ 10 rad/s ............................................................... 56 4 -7 S(t), 12 C (10.4 F) @ 60 second ......................................................................... 58 4 -8 m value, 12 C (10.4 F) @ 60 second .................................................................. 58 4 -9 Typical ten cycles creep and recovery with creep stress of 100 Pa ..................... 60 4 -10 Typical creep and recovery cycle with creep stress of 100 Pa ............................. 60

PAGE 10

10 4 -11 MSC R average recovery at 67.0C (152.6 F) (RTFOT residue) .......................... 62 4 -12 MSCR non -recoverable compliance at 67.0 C (152.6 F) (RTFOT residue) ....... 62 4 -13 MSCR average recovery at 76 C (168.8 F) (RTFOT residue) ............................. 63 4 -14 MSCR non -recoverable compliance at 76 C (168.8 F) (RTFOT residue) .......... 63 4 -15 Elastic recovery at 25 C (77 F) (RTFOT residue) ................................................ 64 4 -16 Force ductility test result, 10 C (50 F) .................................................................. 65 4 -17 Force ductility test result, 25 C (77 F) .................................................................. 66 4 -18 Stress -strain diagram of RTFOT residue at 10 C (50 F) ..................................... 67 5 -1 Ninitiation f or DG granite mixtures ............................................................................. 78 5 -2 Npropagation for DG granite mixtures .......................................................................... 78 5 -3 Ninitiation for DG limestone mixtures ......................................................................... 79 5 -4 Npropagation for DG limestone mixtures ..................................................................... 79 5 -5 ER for DG granite mixtures .................................................................................... 80 5 -6 ER for DG limestone mixtures ................................................................................ 80 5 -7 Ninitiation for OGFC granite mixtures ........................................................................ 83 5 -8 Npropagation for OGFC granite mixtures .................................................................... 83 5 -9 Ninitiation for OGFC limestone mixtures .................................................................... 84 5 -10 Npropagation for OGFC limestone mixtures ................................................................ 84 5 -11 ER for OGFC granite mixtures ............................................................................... 85 5 -12 ER for OGFC limestone mixtures .......................................................................... 85 6 -1 Typical streng th test result for DG mixtures, STOA, 10 C ................................... 89 6 -2 HMA material fatigue curve under cyclic loading .................................................. 90 6 -3 Damage recovery curve ......................................................................................... 91 6 -4 Haversine loading waves ....................................................................................... 93 6 -5 Initial resilient deformation with different rest periods, 10 C ................................ 94

PAGE 11

11 6 -6 Rest period effects on resilient deformation, 1755 lbs, 10 C ............................... 96 6 -7 CDL test results, 0.9 second rest, 10 C ............................................................... 96 6 -8 DG mixtures IDT strength test, 10 C .................................................................... 98 6 -9 DGUS cyclic loading test results, 0.4 second rest, 10 C ..................................... 98 6 -10 DCSE obtained from tensile strength test ............................................................. 99 6 -11 Healing test flow chart .......................................................................................... 102 6 -12 Data acquisition during healin g phase ................................................................ 103 6 -1 3 Data collection illustration .................................................................................... 103 6 -14 Resilient deformation ........................................................................................... 105 6 -15 Poissons ratio comparison, 10 C ....................................................................... 107 6 -16 Damage and healing interpretation with resilient modulus ................................. 109 6 -1 7 DG mixtures normalized resilient modulus at damage phase ............................ 110 6 -1 8 Relationship between modifier contents and damage rate ................................. 110 6 -19 Combi ned regression for healing test .................................................................. 111 6 -2 0 Damage (DCSE) and h ealing r ate c omparison ................................................... 113 6 -2 1 Damage ( r esilient m odulus r eduction) and h ealing r ate c omparison ................. 114 6 -22 Reduced resilient modulus vs reduced DCSE, 20 minutes loading ................... 114 A-1 Original binders stress -strain diagram, 10 C (50 F) ........................................... 1 27 A-2 RTFOT residues stress -strain diagram, 10 C (50 F) ......................................... 127 A-3 PAV residues stress -strain diagram 25 C (77 F) ............................................. 128 B-1 MRDGUS at damage and healing, 10 C ................................................................ 129 B-2 MRDGMS at damage and healing, 10 C ................................................................ 129 B-3 MRDGRS at damage and healing, 10 C ................................................................ 130 B-4 MRDGAS at damage and healing, 10 C ................................................................ 130 B-5 MRDGBS at damage and healing, 10 C ................................................................ 131

PAGE 12

12 B-6 MRDGCS at damage and healing, 10 C ................................................................ 131

PAGE 13

13 LIST OF ABBREVIATION S AASHTO American Association of State Highway and Transportation Officials ARB-12 Asphalt rubber b inder with 12% crumb rubber ARB-5 Asphalt rubber binder with 5% crumb rubber BBR Bending Beam Rheometer CDL Cyclic damage loading CRM Crumb rubber modifier DG Dense graded granite DGUL Long term oven aged dense graded granite mixture with binder PG 6722 DGUS Short term oven aged dense graded g ranite mixture with binder PG 6722 DL Dense graded limestone DLUL Long term oven aged dense graded limestone mixture with binder PG 67 -22 DLUS Short term oven aged dense graded limes tone mixture with binder PG 67 -22 DSR Dynamic Shear Rheometer ER Energy ratio FDOT Florida Department of Transportation HB Hybrid binder HMA Hot Mix Asphalt IDT Indirect tension test MSCR Multiple Stress C reep Recovery NCAT National Center of Asphalt Techn ology OGFC Open graded friction course PG Performance grade

PAGE 14

14 RTFOT Rolling thin film oven test C DL Cyclic damage loading SBS Styrene Butadiene Styrene

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVALUATION OF HYBRID BINDER USE IN SURFACE MIXTURES IN FLORIDA By Weitao Li December 2009 Chair: Reynaldo Roque Major: C ivil Engineering Binder and mixture tests were perfo rmed to evaluate the relative performance of a PG 67 -22 base binder and six other commercially available binders produced by modifying the same base binder with the following modifiers: one Styrene Butadiene Styrene (SBS) polymer, three commercially available hybrid binders composed of different percentages of rubber and SBS polymer, and two asphalt rubber binders (5% and 12 % rubber: ARB -5 and ARB -12). Results indicated that hybrid binders (modified with more rubber than SBS) that exceed the cracking perfo rmance characteristics of unmodified binder and asphalt rubber binders, and have about the same cracking performance characteristics of SBS polymer modified binder can be produced commercially. Results also indicated that hybrid binder can be suitably spec ified using existing specification requirements for PG7622 binder and solubility. Therefore, it appears that hybrid binder has the potential to replace three binders currently used by FDOT in hot mix asphalt: SBS polymer modified asphalt, ARB 5, and ARB -1 2. It was recommended that FDOT develop a transition plan to accomplish this. The research also showed that existing binder tests do not accurately predict cracking performance at intermediate temperatures, even in a relative sense.

PAGE 16

16 A further Healing testi ng was performed on the dense graded granite mixtures to evaluate the healing potential of hybrid binders compare d to base binders. Different testing procedures were carried out and analyzed to give a better understanding o f healing mechanisms. Results fro m healing test agreed with those from cracking performance evaluation. However, more healing parameters and procedure need to be fully developed to capture healing characteristics more specifically. It was recommended that FDOT pursue development and evalu ation of the healing test.

PAGE 17

17 CHAPTER 1 INTRODUCTION Background According to the 2007 estimates of the United States C ensus Bureau, the State of Florida is the fourth most populous state in the union with a population of approximately 18.25 million people and growing by approximately 1000 residents every day. This population growth not only increases the number of vehicles using the states infrastructure, but also adds to the states waste management efforts with respect to the increasing number of waste tires which will eventually accompany the growth in the number of automobiles using Floridas highways. The Florida Department of Environmental Protection (DEP) reports that prior to 1989, almost all waste tires were either land filled (whole carcasses) or stockpiled. That same year, legislation was passed requiring all tires to be cut or shredded into 8 or more pieces prior to disposal thereby, reducing the total volume of the waste product. This effort consequently sparked the development of alternative uses for this waste product; including asphalt and soil modification; playground or sporting area surfacing or covers; the molding of new rubber based consumer products, and other applications. The Florida Department of Transportation (FDOT) utilizes tons o f crumb rubber annually, from local producers, for use in FDOT contracted Asphalt Rubber Membrane Interlayer (ARMI), friction courses and sealants used in roadway const ruction and maintenance. In fact, Florida is the only state which routinely specifies Rubber Modified Asphalts (RMAs) for use in their final surface asphalt mixture (friction courses) on all state highways. The following figure indicates that although both the total number of waste tires and the amount of crumb rubber generated from these was te tires have

PAGE 18

18 remained relatively constant over the period; the usage by FDOT has been decreasing, from approximately 18% to 10% of the total crumb rubber generation. Waste Tires Usage in Florida 19,500,000 20,500,000 20,500,000 16,600,000 16,200,000 18,500,000 4,940,000 4,110,000 4,730,000 900,000 600,000 500,000 000.0E+0 5.0E+6 10.0E+6 15.0E+6 20.0E+6 25.0E+6 30.0E+6 2002 2005 2007 PTE (Passenger Tire Equivalent) Total Waste Tires Waste Tires Used Crumb Rubber Generated Crumb Rubber used by FDOT Figure 11 Waste t ires use history in F lorida Currently Floridas specifications identify asphalt binders incorporating the use of crumb rubber by binder type and application. These include: ARB-5 (5% rubber by weight of asphalt), used in Dense Graded Surface Mixtures ARB-12 (12% rubber by weight of asphalt), used in Open Graded Friction C ourses (OGFCs) ARB-20 (20% rubber by weight of asphalt), used as part of an anti -reflective crack relief layer The use of these binders was not introduced just to consume crumb rubber as a means to an end, that is, to comply with the comprehensive 1988 Florida State Solid Waste Law. Research conducted inhouse by FDOT, the National C enter for Asphalt Technology at Auburn University (NCAT) and the University of Florida has shown the beneficial effects of these materials. OGFCs have benefited from asphalt rubber binders

PAGE 19

19 by exhibiting improved short -term raveling resistance, and improved cracking resistance; and Floridas dense graded friction courses, FC -9.5 and FC -12.5, exhibited small improvements in rut resistance over a conventional binder as determined, in an FDOT accelerated p avement analyzer study (Moseley et al 2003). In addition, it is generally well accepted that rubber reduces the rate of oxidative agehardening, which can have a beneficial effect on cracking. Polyme r Modified Asphalts, or PMAs, have been used in Florida since 2001. PMAs are modified by the reacted addition of Styrene Butadiene (SB) polymer or Styrene Butadiene Styrene (SBS) polymers to a base binder. Based on research performed on Floridas Accelerat ed Pavement Tester (APT) and work performed at NCAT, PMAs have been shown to improve the rutting resistance of good performing asphalt mixtures. C onsequently, Florida now uses polymer modified asphalt mixtures for the top layer, or top two layers, on Inter state high truck volume construction projects. In 2004, Florida decided to include the use of PMAs in Interstate high truck volume OGFC based on data from University of Florida testing which indicated better rutting and cr acking performance of OGFC (Tia, e t al 2002), and as a method to simplify construction by allowing contractors to purchase larger quantities of a single binder. The cost of Hot Mix Asphalt (HMA) tripled from about $35 a ton in 1999 to over $100 a ton in 2007. This is mainly due to the reduction in crude oil supply, which therefore, increased the cost of asphalt as a by product of crude. The increased price of aggregate due to shortages also contributed to the increased cost of HMA. From 1999 to about 2005, asphalt binder prices remained r elatively flat, from $100 to $200 a ton, but spiked to almost $500 a ton by 2008. In 2008, a Florida Department of

PAGE 20

20 Transportation commissioned economic study included information regarding the supply shortage of styrene butadiene polymers for the asphalt industry. This was not new information, just corroboration of well known industry facts. Both reports recommended that alternate asphalt modifiers be considered during supply shortages, including a very interesting alternative: hybrid binders. A hybrid binder, as described here, is a blending of SB or SBS polymer with digested ground tire rubber (GTR) to produce a cross -linked storage stable polymer modified asphalt (in some states called Terminal Blend C rumb Rubber). As a consequence of this type hybrid bin der, the use of waste tire rubber in Florida pavements would continue and possibly increase. PMAs are normally formulated with about 4% SB(S ). If the percent SB(S) was reduced and substituted with equal or more GTR, which is more readily available, a lik ely substitute for the standard PMA could be obtained. We know that both asphalt rubber binders and polymer modified binders can improve the performance of mixtures over the same mixtures produced with unmodified binders. Therefore, it is important to iden tify and evaluate whether different hybrid binders can perform competitively versus other modified asphalts currently used in Floridas highway applications and identify critical specification properties that must be met. Objectives The overall objective o f this work is to determine whether a hybrid binder, composed of tire rubber and polymer, results in an asphalt mixture with improved performance related to a mixture produced with unmodified asphalt. More specifically, project objectives include:

PAGE 21

21 Identify three hybrid binder producers and binders which are currently available or that can be produced for evaluation in this study. Characterize the hybrid binders to verify that they can meet all appropriate specifications for polymer modified binders (PG76 -22 ) and to identify potential issues associated with the specifying and implementing the use of hybrid binders in Florida. Compare the performance of OGFC and dense-graded asphalt mixtures produced with hybrid binders to the performance of the same mixtures produced with an unmodified binder, an SBS polymer modified binder, an ARB -5 binder for dense graded mixtures, and an ARB 12 for OGFCs. Performance will be evaluated in terms of the mixtures resistance to cracking, because one primary concern was that jus t stiffening the binder could result in brittleness and reduced cracking resistance. Develop and further healing test procedure s. Interpret damage and healing to HMA mixtures with resilient modulus changes. Characterize dense graded mixtures damage and healing performances, compare binders effects on mixtu res healing behavior and find relationship between resilient modulus and dissipated creep strain energy. Provide recommendations for future work to further understand the behavior of this type of binder, so that blends can be optimized for enhanced performance and to identify properties that accurately reflect the binders performance in asphalt mixtures and pavement. Scope The primary focus of the work will be on three hybrid binders obtained from differ ent producers. Tests were performed to assess the performance of the binders and their controls; and the performance of the mixtures produced with these binders. Binder performance was characterized using traditional Superpave binder tests (FDOT Standard S pecifications 9161 for PG Superpave asphalt binders) as well as tests for Elastic Recovery (ER) and a newer test called the Multiple Stress C reep Recovery test or MSCR. The MSCR test was primarily developed to identify the presence of polymer in an asphal t binder and to better characterize the high temperature elastic component of polymer modified binders. Force Ductility test was

PAGE 22

22 also adopted because this test method merited further investigation and could be used to characterize the binders. Mixture perf ormance was evaluated for two mixture types: an OGFC and a d ensegraded Superpave mixture. In addition, two different aggregates, limestone and granite, which are extensively used in Florida, were ev aluated with each mixture type. For each of the mixtures, hybrid binder performance was compared to the following: unmodified binder (PG 67 -22), SBSmodified binder (PG 76-22), and crumb-rubber modified binder (ARB-5) for dense-graded mixtures; SBS modified binder (PG 76-22), and crumbrubber modified binder (AR B12) for OGFC mixtures. Healing potentials were evaluated for dense graded granite mixtures: hybrid binders performance was compared with PG 7622, ARB 5 and PG 67-22. Performance evaluation involved the most advanced laboratory tests and interpretation methods available to assess asphalt mixture resistance to cracking in order to ensure that the modified binders did not stiffen the mix to the point that it was brittle and prone to cracking. The primary tools were the Superpave indirect tension test (IDT) along with the HMA fracture mechanics model and energy ratio concept developed at the University of Florida. Also healing test procedures were developed and used to evaluate dense graded granite mixtures healing potentials. Comparison between hybrid binders and control binders was made with respect to their healing performance.

PAGE 23

23 CHAPTER 2 L ITERATURE R EVIEW Crumb Rubber U sed a s A sphalt B inder M odifier Over the last three decades, many different modifiers have been added to asphalt binders to improve both t he rutting and cracking resistance of Hot Mix Asphalt (HMA). Of all the available modifiers, two major categories see extensive use today: Rubber and Polymers. Rubber, as an asphalt binder modifier most normally referred to as crumb rubber modifier or C RM, is composed of natural rubber (latex), synthetic rubber (polymer), and carbon black. It is known that the natural rubber enhances elastic properties, whereas the synthetic rubber improves thermal stability (NCAT, 1996). C RM is obtained from whole tire rec ycling and retreading operations. Heitzman (1992) summarized factors that affect the C RM -binder interaction: production method (ambient versus cryogenic grinding), particle size, specific surface area and chemical composition. Among these, the specific sur face area has been reported as the most influential. This document has become the prime source document for specifications for both the recycled tire rubber and asphalt rubber binders. Putman, (2005) found that the C RM -binder interaction can be described by two essential effects: the Interaction Effect (IE) and the Particle Effect (PE). The IE is related to the absorption of aromatic oils from the binder by the rubber, while the PE considers the rubber acting as filler in the binder. He concluded that the I E is greatly influence by the crude source of the binder and could potentially be used as an indicator of a binders compatibility with C RM. A higher IE value would indicate a more compatible binder.

PAGE 24

24 Currently, there are three methods of incorporating rubb er into HMA: the wet process, the dry process, and the terminal blend process. It should be noticed that wet and dry processes are performed at the plant site rather than at a refinery or terminal. Wet process: the rubber and asphalt binder are mixed toget her prior to addition with the aggregates (by far, the most widely accepted and used method, in Florida, this is primarily done at the asphalt terminals and can cause confusion with the Terminal blend process definition) Dry process: the rubber and the aggregates are mixed together prior to the addition of the asphalt binder. Terminal blend process: the rubber is dissolved in the asphalt binder at the terminal with addition of other additives/modifiers. Generally, a proprietary means using a combination of chemicals, heat and physical processing is used to achieve solubility. In many different regions of the country, pavements using asphalt rubber binders have exhibited better cracking resistance and increased durability over pavements using conventional asp halts. Several State experiences are summarized by Hicks et al (1995): The Arizona Department of Transportation (ADOT) started using rubber in HMA test sections in the 1970s. With the experience gained from these test sections, ADOT used both open-graded and gap -graded mixtures over existing rigid and flexible pavements. Since 1989, over 40 projects have been placed using rubber modified mixtures, and as a result, ADOT has observed a dramatic decrease in their pavement cracking.

PAGE 25

25 California (CalDOT or C alt rans) has experimented with both wet and dry rubber processes for HMA since the 70s, but stopped using the dry process due to erratic pavement performance. C ook et al. (2005), utilized Superpave tests, as well as, the Hamburg wheel tracking device to eval uate the fatigue and rutting performance of rubber modified mixtures in 2005. They concluded that asphalt rubber modified mixtures performed at least as well as, if not better than, the conventional dense graded asphalt mixtures; therefore, they recommended the use of C RM mixtures. The Florida Department of Transportation (FDOT) started using rubber in asphalt mixtures in 1988 and fully implementing its use in 1994. They used an asphalt rubber binder (ARB-5) in dense graded friction courses 25 mm thick to i mprove the resistance to shoving and rutting, particularly at intersections. On Interstate high truck volume highways, they placed a thin 15 mm open graded friction course (using ARB -12) to improve their durability. Polymers U sed a s A sphalt B inder M odifier Polymers are characterized as thermoplastic rubbers or elastomers and examples of these include: Styrene Butadiene Rubber (SBR or SB), Styrene Butadiene Styrene (SBS), Styrene Isoprene Styrene (SIS), Polybutadiene, and Polyisoprene. (NCAT, 1996) These elastomers have an important effect on the temperature susceptibility and stiffness of the asphalt binder. Due to their chemical structure, polymers are generally less susceptible to changes in temperature than standard asphalt binders; therefore, polymer modified asphalt binders (PMAs) offer a great reduction in their temperature susceptibility. A small sampling of PMA experiences is presented here: Kentucky Transportation C enter and Kentucky Department of Transportation (KDOT) tests showed that polymer modif ied binders can improve the rutting (using

PAGE 26

26 wheel tracking tests) and the cracking resistance of asphalt mixtures (Fleckenstein, et al 1992). The Oregon Department of Transportation (ODOT) validates that polymers are a practical way to reduce the temperat ure susceptibility of asphalt pavements. They also found that polymerized asphalt mixtures are more resistant to freezethaw damage (Rogge et al 1992). At the University of Florida, Kim (2003) showed that SBS modified mixtures generally have a lower m v alue than the same unmodified mixture; indicating a reduced rate of damage in the mixture. Development of H ybrid B inders in R ecent Y ears The hybrid binder composed of SBS, rubber and asphalt was a relatively new approach at the beginning of this study. Therefore, there were very few research papers on these materials. Essentially, there is little to no knowledge of the engineering performance of hybrid binder. An FHWA evaluation of modified binders included lab as well as accelerated loading of test sections. The rutting performance of Section 5 Terminal Blend C rumb Rubber (a hybrid binder) performed as well as SBS polymer modified binders (Tia, 2002). According to the SBS Polymer Supply Outlook (by Association of Modified Asphalt Producers 2008 ), there was a shortage of SBS for the asphalt industry and the price of SBS was increasing, which could happen again. Because of this background, hybrid binder provides an attractive alternative.

PAGE 27

27 Most research studies have focused on SBS modified binder or Asphalt Rubber Binder separately. A summary of research is presented below on the fracture resistance of these two systems. As for the SBS modified binder and Asphalt Rubber Binder, most researchers have primarily used traditional test methods including Dynamic S hear Rheometer, Bending Beam Rheometer, Penetration, Brookfield Viscosity, Elastic Recovery, Ductility, Softening Point, thin layer chromatography, etc. C omparisons have generally been based on the traditional test properties such as the complex shear modulus G*, phase angle and other Superpave indices. Some researchers have developed other parameters to evaluate performance of different modified binders. For example, Gilberto et al (2006) used the Binder Aging Ratio (BAR) calculated from G* to differentiate binders, and found that Asphalt Rubber can decrease BAR 40% 50% compared with unmodified asphalt, but its aging level is similar to Polymer Modified Binders. Other researchers used traditional test devices such as the Dynamic Shear Rheometer to evalu ate the creep behavior of binders (e.g., Felice, et al 2006 ). Some researchers noticed the limitations of traditional Superpave indices. For example, Bahia, et al (2008) but neglects the nonlinear viscoelastic behavior that may be more indicative of resistance to fracture and rutting. As an alternative, he performed time sweep tests based on the Dynamic Shear R heometer He found that both Yield Energy and strain at maximum stress obtained from these tests correlated well with field performance. Bahia, et al (2008) also evaluated the Elastic Recovery and Multiple Stress C reep Recovery tests for modified binders and found that Elastic Recovery is a good tool to identify

PAGE 28

28 Polymer Modified Binders, and Jnr from Multiple Stress C reep Recovery tests characterizes nonlinear behavior. In addition, some new test devices have been developed. For instance, the Asphalt Binder C racking Device (ABCD) was used to evaluate the Low Temperature Thermal C racking (SangSoo Kim, 2008) When temperature drops, asphalt shrinks 100 times or more than the ABCD invar ring, so the asphalt compresses the ring, and an Electrical Strain Gaug e measures this compression at cracking, which is related to the tensile fracture resistance of the binders. This device was also found to be able to characterize Polymer Modified Binders but only at low temperatures. Methodologies of Evaluating A sphalt M i xtures Healing Fatigue cracking in asphalt concrete pavement is considered one of the four types of distresses, as well as rutting, low temperature cracking and moisture damage. Both theoretical and empirical models have been proposed which tried to predic t the fatigue life of the pavement, but the reality is most of these models developed in the laboratory have underestimated the fatigue life in the field. While one of the reasons may be contributed to the difference of rest periods between laboratory test and the real road way traffic, healing during the rest periods has been observed and verified by many researchers. Pseudo Stiffness M ethod Throughout the '80s and '90s, Lytton, Little, Kelleher and their followers from Texas Transportation Institute (TTI ) have been dedicated to studying the fracture healing of asphalt concretes. U sing Schapery (1984) correspondence principle, which states that constitutive equations for certain viscoelastic media are identical to those for the elastic cases, but

PAGE 29

29 stresses and strains are not necessarily physical quantities in the viscoelastic body, Lytton, Little, et al. (1997) developed the dissipated pseudo-strain energy method to evaluate the fatigue life and healing for asphalt concretes. Since asphalt concrete is a nonlinear viscoelastic material, a non linear reference modulus is introduced to eliminate the nonlinearity of the material in order to simplify the relationship between linear viscoelastic stress and pseudo strain (straight line). ) ( ) ( ) ( ) ( t t E t E t Eu m u c R R R Wh ere: R E = linear reference modulus (constant for linear viscoelasticity); ) ( t = nonlinear reference modulus correction factor; ) ( tu c = calculated linear viscoelastic stress for undamaged nonlinear vis coelastic asphalt concrete; ) ( tu m = measured stress for undamaged nonlinear viscoelastic asphalt concrete. Therefore, the pseudo strain for nonlinear viscoelastic asphalt concrete (damaged condition) can be calculated in the following equation: ) ( ) ( ) ( t E t tR d c d R Where ) ( td c is the calculated linear viscoelastic stress in the damaged specimen. After application of the correspondence principle using the non-linear correction reference modulus, the corrected typical ps eudo hysteresis loop can be plotted and the slope of the linear regression of the pseudo hysteresis loop is defined as pseudo stiffness, which is an unambiguous indicator of damage and the effect of rest periods on

PAGE 30

30 microcracking and healing. A typical pseu do stiffness recovery due to rest period is shown in Fig. 2 -1. Figure 21. Typical effect of healing due to rest period A healing index was introduced to describe asphalt mixture healing properties: before before afterHI Where: HI =healing index; before =pseudo stiffness before rest period; after =pseudo stiffness after rest period. Ratio of D issipated E nergy C hange (RDEC) M ethod Carpenter (2006) stated that when sustaining cyclic fatigue loading, t he viscoelastic HMA material traces different paths for the unloading and loading cycles and creates a hysteresis loops. The area inside of the loop is called dissipated energy. The difference between these loops during the fatigue test indicates the amount of the dissipated energy that is producing damage.

PAGE 31

31 By definition, RDEC is the ratio of dissipated energy change between two loading cycles divided by the number between the two cycles, that is, the average ratio of dissipated energy change per loading c ycle. In practical usage, the RDEC value at the 50% stiffness reduction point is defined as plateau value, PV. According to the findings by Shen and C arpenter (2005), there is a unique relationship between PV and Nf50 (fatigue life at 50% stiffness reducti on point) for different mixtures, loading modes, loading levels, and testing conditions (frequency, rest periods, etc.). The PV is a comprehensive damage index that contains the effect of both material property and loading conditions. ) ( a b DE DE DE RDECa b a a Where: aRDEC is the average ratio of dissipated energy change at cycle a, comparing to next cycle b; a, b = load cycle a and b, respectively, The typical cycle count between cycle a and b for RDEC calculation is 100, i.e., ba=100; DEa, DEb = the dissipated energy produced in load cycle a, and b, respectively. And: 100 100 1 150 fNf PV The PV recovery per second of rest period is an indication of healing capacity. At normal damage levels, it can require a very long rest time to fully recover damage. This is why the healing effect is not observed in normal laboratory testing. Testing results

PAGE 32

32 indicate the healing/recovery rate of the polymer modified binders is significantly greater than the neat binder tested, which may account for their extended fatigue life observed in the field. Carpenter (2006) stated that healing is a continuous physical -chemical reaction that may occur continuously as applied load damage develops, not just between load applications. C onsidering the energy behavior of the viscoelastic HMA material, the actual fatigue behavior can be explained as energy equilibrium between surface energy and dissipated energy, generally expressed as (Freund, et al 2003): Chemical potential (healing potential) = Surface energy Diss ipated energy If surface energy is smaller than the dissipated energy, the chemical potential (healing potential) is negative, thus the material has the tendency to increase surface energy through creating more surfaces. This is the process of crack initia tion and propagation (damage). On the other hand, if the dissipated energy is at a very low level and the healing potential is positive, the energy equilibrium leads to a decrease of surface energy, that is, some open crack surfaces will close through a healing process. Dissipated C reep Strain Energy (DCSE) M ethod A fundamental crack growth law was developed at the University of Florida that allows for the prediction of crack initiation and crack growth in asphalt mixture subjected to any specified loading history. This law, which is based on the principles of viscoelastic fracture mechanics, was incorporated into a cracking model which is called the HMA Fracture Mechanics Model (Roque 2002). The HMA Fracture Mechanics Model is driven by the fact that asphalt mixture has been determined to have fundamental dissipated creep strain energy (DCSE) thresho ld above which cracking will

PAGE 33

33 initiate, or propagate, if the crack is already present This threshold has been found to be independent of loading mode or loading history. DCSE is calculated based on a particular tensile stress level with a haversine load of 0.1s followed by 0.9s rest period, which is commonly used to represent an applied wheel load. Therefore dt t t cycle DCSEP AVE) 10 sin( ) 10 sin( /1 0 0 max Since 1 1 1 1 1 0 max) 1000 ( ) ( m AVE m AVE m AVE AVE PmD t mD dt t D D d dt t dD So 1 1 2 1 0 0 2 1 1 2) 1000 ( 20 1 ) 10 sin( ) 1000 ( / m AVE m AVEmD dt t mD cycle DCSE Where: AVE is the average stress within the portion of the asphalt mixture in interest ; D1 and m are the creep compliance power law parameters. Based on the above calculations, Kim (2003) found relationship betw een DCSE recovery and time. Fig. 2 -2 giv es a clear evidence of healing by the continuous reduction in DCSE. Regression analysis was performed on these curves and logarithmic functions were the best -fit. Healing rate was defined as the slope of the logarithmic functions and was used to evaluate the healing property of asphalt mixtures. Also a normalized damage parameter (DCSE/DCSEapplied) was defined to evaluate the healing property independent of the amount of damage incurred in the asphalt mixture ( Fig. 2 3 ).

PAGE 34

34 Figure 2 2 Healing test results (1,000 cycles 75 psi 15 C) Figure 2 3 Normalized healing (after 1000 cycles 75 psi 15 C) Elastic D eformation and Strain R ecovery Chowdary et al. (2005) from Indian Institute T echnology (I.I.T.) carried out repeated triaxial tests on sand asphalt mixtures with varied confinement conditions and rest periods to quantify healing in the laboratory.

PAGE 35

35 35 mm diameter and 70 mm height cylindrical cored samples with 8% and 10% air voids from Mrshall sized samples were prepared for triaxial test. Repeated triaxial tests were carried out on san asphalt mixtures with varied confinement conditions. All tests were conducted in load controlled mode. Two loading/unloading cycles of 7 and 14 sec onds were conducted. Two sets of lateral pressure/ vertical pressure (0.5kg/cm2 / 2.5 kg/cm2 ; 0.875kg/cm2 / 4.375 kg/cm2) corresponding to a ratio of 1:5 were applied for each specimen. One of the main considerations for choosing this specific ratio is related to subjecting the specimen to load levels that will engender deformation capable of healing during the rest periods and yet not physically deform the specimen. Figure 2 4 Load controlled cyclic loading and rest periods The material was allowed to rest for one hour and the same loading and rest cycles of equal duration were applied again to observe the deformation response. The deformation of the material with time during loading and rest periods was measured. Two parameters were selected for chara cterizing the healing of sand asphalt mixtures investigated in this study. The first parameter corresponds to the change in the instantaneous elastic deformation at time t=0. For the same load application, a reduction in the instantaneous elastic deformati on after rest period signifies improved material property (the material modulus value increases resulting in decreased elastic deformation). Results showed this parameter depends on the magnitude of rest period.

PAGE 36

36 Figure 2 5 A sample plot of the strain v ersus time during the loading/rest period The second parameter chosen to characterize healing is related to improved strain recovery or springiness resulting due to the rest period. The percentage of strain recovered at the end of three seconds after load removal when compared with the strain at the end of the loading cycle was selected as the second parameter. This parameter depends on the loading and rest period. Summary Generally speaking, it has been found that traditional Superpave tests and indices cannot clearly differentiate between modified binders Also, although the Multiple Stress C reep Recovery Elastic Recovery and Force Ductility test are able to identify polymer like behavior to some extent, they may not differentiate between different modi fied binders: SBS, hybrid binder and rubber modified binder. These and other limitations with the current binder test methods need to be explored to determine whether development of new test methods which can accurately reflect the different properties of various modified binders, and reflect their relative cracking or fatigue performance at ambient temperatures is needed. The goal would be to obtain as accurate as possible

PAGE 37

37 stress, strain, time and fracture energy relationships and other crucial properties, s o reliable relationship between asphalt binder and mixture properties can be established. Different approaches have been practiced by many researchers, and all of the results have shown the existence of healing characteristics of asphalt mixtures. Among these methods, apparently the dissipated energy method is the most acceptable and convenient one to detect healing potentials of the mixtures. However, evaluation methods for healing as of today are still inadequate and not mature enough to quantify it. Th is is because of the lack of an appropriate testing and interpreting system to measure damage recovery rates of asphalt mixtures. Therefore, there is a significant necessity to develop an appropriate and practical method for detecting and quantifying heali ng potentials of asphalt mixtures.

PAGE 38

38 CHAPTER 3 MATERIALS AND METHOD S Since this is the first research project focused on the evaluation of hybrid binder in Florida, two commonly used aggregate types in the State were chosen (limestone and granite). Followi ng FDOT instructions, typical gradations currently used in Florida were selected to quantify the effect of C RM and hybrid binder on mixture cracking performance. Two mixture types frequently utilized in Florida were considered for this study: dense -graded (DG) and opengraded friction course (OGFC). DG mixtures are widely used for structural purposes; whereas OGFCs are used for their outstanding capacity for providing and maintaining good pavement frictional characteristics to reduce hydroplaning and improv e safety in wet weather. Binders A search was conducted to gather information regarding possible sources or producers for hybrid binders as defined by this research. At first, seven vendors or companies were identified as possible participants or sources o f binder for this study. When available, an assessment was made regarding the current products these companies produced and whether any of their binders would qualify for this research. Of the original producers list, it was determined that two of them wer e actually working in concert and could produce a viable product, and that another company already had an existing product and had been producing it for some time. Of the remaining companies, one had extensive experience in polymer modification of asphalt and showed great interest in the research but, did not currently have a product to offer. They speculated that development of such a product would take between six months to

PAGE 39

39 one year to complete. Lastly, a fourth company was developing some similar interes ting product ideas but, was looking for someone to help them bring it to fruition, i.e., no product available. The remaining suppliers were either out of business, or produced a deadend lead. Therefore, the initial search for hybrid binder producers ident ified only two existing viable sources for these materials. The study was to contain three hybrid binders obtained from different producers, and this was proving to be a difficult task. After much due diligence, a third producer was identified, who produced a hybrid binder for use as a bonding agent, but had no experience using this product to produce hot mix asphalt. This was not deemed important and since it met the requirements for a hybrid binder, it was added as our third and final binder. The researc h originally intended to establish guidelines for the design of the hybrid binders; controlling the amount of rubber and polymer, and the ratio between the two components. More importantly, specifying that the amount of ground tire rubber must exceed that of polymer. The least requirement to which the producers would be subject to: that their final product must be formulated to meet and pass the Superpave PG 7622 binder specifications. Upon further reflection, this decision would cause the research and res earchers to relinquish considerable control over any aspect of the binder production, including the source of the original binder prior to modification. Therefore, it was decided to establish a baseline for the modification, that is, that all the hybrid bi nder producers should start with the same base binder. The three binder producers were informed of this decision and all concurred with the rationale, and agreed to modify any supplied base binder.

PAGE 40

40 It was agreed to use C ITGO Petroleum products, PG 6722 and PG 7622, as the control binders. C ITGO Petroleum delivered, to each of the three hybrid binder participants, a minimum of 10 gallons of their PG 67 22 binder for modification. The University of Florida received enough PG 6722 binder for binder testing, for mixture production, and as a base binder, to produce the rubber modified binders (ARB -5, and ARB-12) needed for the project. Each of the hybrid binder participants was asked to disclose as much about the formulation of their product as they were willi ng, without infringing on proprietary products or processes. More specifically, the researchers were interested in the SBS and ground tire rubber content for comparison between producers, and for possible explanations in binder and mixture performance. In total, seven different binders were used in this project. These are outlined in the table 3-1: Table 3 1 Asphalt b inder and the c onstituents/ f ormulations Binder Modifying C omponents PG 67 -22 None PG 76 -22 4.25% SBS HB A 1% SBS (approximately 30 mesh, incorporated dry), 8% of Type B GTR, 1% hydrocarbon HB B 3.5% crumb rubber, 2.5% SBS, 0.4% plus Link PT 743 cross linking agent HB C 10% rubber, 3% 0.1% radial SBS ARB-5 5% Type B rubber ARB-12 12% Type B rub ber Binde r testing w as performed by the Florida Department of Transportation State Materials Office The tests performed were all those required by FDOT Standard Specifications 916 -1 for PG Superpave asphalt binders. In addition, DSR and creep

PAGE 41

41 stiffness were perfor med after PAV at 110C, in addition to the standard 100C. The basic binder testing program is summarized in table 3 2. Table 3 2 Binder t ests s ummary Binder Type Number Number of Tests* Number of Replicates Total Number of Binder Tests Base 1 12 2 2 4 Hybrid 3 12 2 72 SBS modified 1 12 2 24 ARB 12 1 12 2 24 ARB 5 1 12 2 24 Totals 7 12 2 168 Binder tests are as follows (FDOT Specifications 916 -1; Superpave PG Asphalt Binder) : Original Binder: Spot Test, Solubility, Smoke Point, Flash Point, Rotational Viscosity, Absolute Viscosity, Dynamic Shear Rheometer (DSR) Rolling Thin Film Oven Test Residue: Mass Loss, Dynamic Shear Rheometer Pressure Aging Vessel Residue: Dynamic Shear Rheometer (2 temperatures), C reep Stiffness The test results were used to verify that all binders met appropriate specifications for a PG 76 -22 Superpave asphalt binder. In addition, test results were evaluated to identify binder properties or parameters that may be suitable to uniquely characterize these hybrid binders and t o identify potential issues associated with specifying and implementing the use of hybrid binders in Florida. Several non-routine tests were performed on these binders: 1) binders were PAV ag ed at 110 C, which may possibly be used to identify potential aging issues of concern to Florida, 2) binders were subjected to the Elastic Recovery test, which according to Bahia (2008) will identify the presence of polymer modification, 3) binders were subj ected to the Multiple Stress C reep Recovery test (AASHTO TP70 -08), which

PAGE 42

42 according to Bahia (2008) can be used to characterize a binders nonlinear behavior, and 4) binders were tested using the Force Ductility test, which is unique in that it loads the sp ecimen to failure. This last test may be used to calculate energy to failure, which may be correlated to binder and possibly mixture cracking performance. This is essentially the standard ductility test with an added load cell to measure the load applied t o the sample throughout its elongation. Aggregates Aggregates sources were chosen based on previous research work and FDOT directions; detailed information is presented in the Table 33. Both dense -graded (DG) and opengraded friction course (OGFC) mixtur es were designed for each aggregate type (limestone and granite). The particle size distribution of DG mixes is presented in Table 3 3 Aggregate s ource Source Type FDOT C ode Pit No. Producer Nova Scotia Granite # 7 Stone 44 NS -315 Martin Mariette Aggr egates # 789 Stone 51 NS -315 Martin Mariette Aggregates Stone Screenings 22 NS -315 Martin Mariette Aggregates South FL Limestone S-1 -A Stone 41 87339 White Rock Quarries S 1 B Stone 53 87 339 White Rock Quarries Asphalt Screenings 22 87339 Whit e Rock Quarries Georgia Granite # 78 Stone 43 GA 553 Junction C ity Mining # 89 Stone 51 GA -553 Junction C ity Mining W -10 Screenings 20 GA -553 Junction C ity Mining Rinker South FL Limestone # 67 Stone 42 87090 Rinker Materials C orp. S 1 B 55 87 090 Rinker Materials C orp. Med. Screenings 21 87090 Rinker Materials C orp. Local Sand Local Sand Starvation Hill V. E. Whitehurst & Sons

PAGE 43

43 0 10 20 30 40 50 60 70 80 90 100 Sieve size, ^0.45 % passing MDL JMF # 78 Stone # 89 Stone W-10 Screenings Local Sand #30 #16 #8 #4 #100 Figure 3 1 DG g ranite g radation 0 10 20 30 40 50 60 70 80 90 100 Sieve size, ^0.45 % passing MDL JMF # 67 Stone S-1-B Med. Screenings Local Sand #30 #16 #8 #4 #100 Figure 32 DG l imestone g radation

PAGE 44

44 0 10 20 30 40 50 60 70 80 90 100 Sieve size, ^0.45 % passing MDL JMF # 7 Stone # 789 Stone Stone Screenings #30 #16 #8 #4 #100 Figure 3 3 OGFC g ranite g radation 0 10 20 30 40 50 60 70 80 90 100 Sieve size, ^0.45 % passing MDL JMF S-1-A Stone S-1-B Stone Ashpalt Screenings #30 #16 #8 #4 #100 Figure 34 OGFC l imestone g radation

PAGE 45

45 the figures 3 -1 and 3 -2 and the OGFC gradation curves are shown in the figures 3 -3 and 3 -4: the granite blend was added with hydr ated lime (1% by weight) to prevent stripping Mixtures All dense graded mixtures were d esigned to be 12.5 mm nominal maximum aggregate size mixes and to meet specification requirements for a traffic level C which corresponds to 3 to 10 million Equivalent Single Axle Loads (ESA Ls ) over a 20 year period. A summary of the mixture testing plan for this project is presented in the Fig. 3 5. A total of 88 gyratory specimens were prepared. Figure 35 Mixture t esting p lan for e ach m ixture and a ggregate t ype Each mixture in the test plan was designed with a particular binder type while the aggreg ate gradation was kept constant in order to evaluate binder effect on mixture cracking performance. In total, 12 DG (6 binders and 2 aggregate types) and 10 OGFC (5 binders and 2 aggregate types, 0.4% fiber by weight of the mix was added to granite OGFCs t o prevent drain-down) mixtures were evaluated and have identifications (IDs) shown in Tables 3 -4 and 35 (next page)

PAGE 46

46 Initially, all mixtures (conventional and modified) with the same aggregate type and gradation were prepared in the laboratory with the same percentage of binder by weight. Theoretically, all mixes should have had the same effective asphalt volume, and consequently the same volumetric properties. However, during the laboratory work, the effective asphalt volume was found to be about the same for OFGC mixtures but different for DG mixtures. Two factors were thought to have caused this difference: specific gravity of binder (Gb) and aggregate absorption. As mentioned previously, Gb was measured in the laboratory and also aggregate absorption te sts conducted on the different binders indicated definite differences in absorption. C onsequently, asphalt contents were adjusted to ensure that all mixtures had the same effective asphalt by volume. Table 3 4 DG m ixtures IDs for t esting Binder PG 67 -22 PG 76 -22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Limestone DL U DLM DLA DLB DLC DLR Granite DG U DGM DGA DGB DGC DGR Table 3 5 OGFC m ixtures IDs for t esting Binder PG 76 -22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Limestone OLM OLA OLB OLC OLR Granite OGM OGA OGB OGC OGR Mixture Preparation Aggregates and binders were preheated in the oven for 3 hours before mixing; mixing temperature was set to 310 5 F for unmodified and ARB -5 binder mixes and 3 3 0 5 F for PMA and hybrid binder mixes After preheating the hybrid binders, in some containers for all hybrid binders, undissolved modifiers (rubber particles) were

PAGE 47

47 found accumulated on the surface of the binder resulting in about a 2 mm thick film; thus, before pouring the binder into the mixing bucket with the aggregates, a clean steel stick was used to stir the binder evenly to dissolve the film into the binder. T he a ggregates and binder were then mixed in a rotating bucket until the aggregates were well coated with the binder. Before the DG and OGFC samples were compacted, they were placed in a pan and heated in an oven for about 2 hours at the mixing temperature, which is the Short T erm O ven A ging (STOA). The mix was stirred after one hour of heating to obtain a more uni formly aged sample. DG and OGFC mixtures were compacted at 310 5 F and 3 3 0 5 F respectively. Even though the DG mixes were designed to have 4% air void content at Ndesign, they were compacted in the Servopac Gyratory C ompactor to the number of gyrati ons needed to get 7% air voids. The number of g yrations obtained from mix design to get 7% air voids for DG mixtures was 20 for limestone and 24 for granite mixes. F or OGFC mixtures, 50 gyrations were used to achieve compaction level similar to fie ld after traffic consolidation (Varadhan, 2004) S pecimens we re allowed to cool fo r 30 minutes before extrud ing from the molds and for at least 24 hours before cutting or preparation for testing LTOA is meant to represent 15 years of field aging in a Wet -No -Free ze climate and 7 years in a Dry Freeze climate. LTOA re quires a compacted sample (after STOA) be placed in a force draft oven at 185 5F for 5 days (Harrigan, et al., 1994). The same aging procedure was used for both DG and OGFC mixtures.

PAGE 48

48 Because of the very coarse and open structure of OGFC; there was a possibility of these mixes falling apart at the high temperature used for LTOA Hence, a procedure was developed to protect the pills A wire mesh with openings of 0.125 in and steel clamps were used. The mesh size was chosen in order to ensure that there is good air circulation within the sample for oxidation and to prevent the smaller aggregate particles from falling through the mesh. The specimen wa s wrapped twice with the mesh cloth and two clamps were used to contain the specimen without applying excessive pressure on it. The system is shown in the Figure 3 -6. Figure 36 Pill c ontained with m esh After cooling the specimens at room temperature, they were cut to the required thickness for testing. Th e bulk specific gravity for DG mixes was determined in accordance with AASHTO T166 to ensure that the air voids of the specimens were within the required range of 7.0 0.5 % The DG mixture volumet ric information is shown in Table 3 6

PAGE 49

49 For OGFC mixtures, physical parameters were obtained from the CoreLok test. The procedure is described in the Appendix D. After the sample was sealed, it was weighed in the water tank. The OGFC and DG mixture volumet ric information is shown in Table 3 7. Table 3 6 Dense g r aded m ixture v olumetric i nformation Mixture DGU DGM DGA DGB DGC DGR P b 4.80% 4.82% 4.90% 4.89% 4.89% 4.84% G mm 2.578 2.579 2.581 2.58 0 2.58 0 2.579 G mb 2.390 2.380 2.388 2.408 2.399 2.386 Mixture DLU DLM DLA DLB DLC DLR P b 6.60% 6.49% 6.33% 6.18% 6.42% 6.60% G mm 2.319 2.316 2.312 2.309 2.314 2.319 G mb 2.165 2.145 2.153 2.155 2.150 2.148 Table 3 7 OGFC m ixture v olumetric i nformation Mixture Type Aging C ondition Gmm Gmb AV % OGFC Granite STOA 2.441 1.995 18.28 LTOA 1.996 18.23 OGFC Limestone ST OA 2.309 1.990 13.80 LTOA 1.978 14.33 Figure 37. CoreLok sample weighing equipments

PAGE 50

50 CHAPTER 4 BINDER TEST RESULTS AND ANALYSIS It is stated (NCAT, 1991) that the physical properties of asphalt measured by Superpave binder tests are directly related to field performance by engineering properties so the binder tests were performed first and then mixtures tests (fatigue cracking and healing). Conclusions will be made on how significantly binder properties will affect their performances in mixtures. Physical property tests including specific gravity, s olubility smoke point, flash point, rolling thin film oven mass change and spot tests were performed. A summary of test results and findings of binder tests is presented in the sections below. Addition al binder test results are presented in Appendix A. Conventional Superpave binder tests were performed using the Dynamic Shear Rheometer and Bending Beam Rheometer The following tests, which have been specifically developed and identified to evaluate modi fied binders, were also performed: Multiple Stress C reep Recovery (AASHTO TP70 08)) Elastic Recovery (AASHTO T301 -99(2003)) Force Ductility (AASHTO T300 -00) Binders Physical Properties Specific Gravity of Binders Results of specific gravity of binders b ased on the Standard Test Method for Density of Semi Solid Bituminous Materials (ASTM Designation: D 70 -03, Pycnometer Method) are presented in Table 4-1. As expected, all of the modified binders had a higher specific gravity than that of the base binder.

PAGE 51

51 Table 4 1 Specific g ravity of b inders 15.6 C (60 F) Binder s Specific Gravity Gravity (kg/m3) PG 67 -22 1.031 1027.907 PG 76 -22 1.033 1031.389 HB A 1.044 1040.918 HB B 1.036 1032.892 HB C 1.043 1040.356 ARB-5 1.036 1033.004 ARB-12 1.042 1038.824 Solubility Analysis To determine the purity of asphalt cement, solubility test was conducted to compare modified binders and control binders. The solubility of hybrid binder A (92.76%), hybrid binder B (96.905%), ARB 5 (93.835%) and ARB 12 (88.765%) did n ot meet the specification requirement (minimum 99% ). As illustrated in Fig. 4 1, the 86.0 88.0 90.0 92.0 94.0 96.0 98.0 100.0 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Binders Solubility (%) PG minimum 99% Figure 41 Solubility of o riginal b inders

PAGE 52

52 solubility was lower for binders with higher coarse rubber content ( hybrid binder A ( 8 %), hybrid binder B (3. 5%), ARB-5 (5%) and ARB -12 ( 12 %) ), indicating that the rubber may not have been fully digested in the base binder. Hybrid binder C, which was produced with finer grained rubber, did meet FDOTs solubility specification, indicating that the rubber was fully digested in the base binder thereby making it more suitable for DSR testing. B ased on these results, it appears that solubility may be a good way to distinguish binders that may have excessively coarse particle s (e.g. undigested rubber particles) that would make them unsuitable for DSR testing. A lso, results of hybrid binder C show that hybrid binder can meet the solubility requirement. T herefore, solubility appears to be a good way to distinguish hybrid binder from asphalt rubber binder. Mass of Volatiles Loss Analysis 0.0 0.1 0.2 0.3 0.4 0.5 0.6 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Binders Mass Loss (%) Figure 42 RTFOT, m ass l oss at 163 C (325.4 F)

PAGE 53

53 The RTFOT allows the determination of the mass of volatiles lost from the binder during the test. The amount of volatiles lost indicates the amount of aging that may occur during HMA production and cons truction. As indicated in Fig. 4 2, all binders except hybrid binder C which had a Mass Loss of 0.524% met the specification requirement f or Mass Loss after RTFOT ( 0.5%) The M ass Loss of hybrid binder A, B wa s the smallest. Binders Rheological Propert ies Dynamic Shear Rheometer (DSR) and Bending Beam Rheometer (BBR) tests were performed and analyzed at different testing temperature s i.e. DSR at high ( 76 C, 168.8 F ) and intermediate temperatures (25 C, 77 F ), and BBR at low temperature (12C, 10.4 F ) DSR Test Results Analysis The DSR is used to characterize the viscous and elastic behavior of asphalt binders at high and intermediate service temperatures. It measures the complex shear modulus G* and phase angle at the desired temperature and frequency of loading. According to NCAT, original and RTFOT aged binder samples were tested at high temperature to determine the binder s ability to resist rutting, while PAV aged samples were tested at the intermediate temperature to determine binders ability to resist fatigue cracking. G* and results with 5% testing error range at high temperature were shown through Fig. 4-3 and Fig. 4-4. Fig. 45 and Fig. 46 show the results at low temperature. As indicated in Fig. 4 3, all modified binders resulted in (indicator of of all modified binders was above the minimum requirements for PG 76-22 binder. A significant difference was observed in the magnitude of for the different modified binders in

PAGE 54

54 both original and RTFOT conditions. The were observed for binders with the highest concentration of coarse rubber (hybrid binder A, hybrid binder B and ARB 12) and may be suspect Fig. 4.4 illustrates that all modified binders exhibited a lower phase angle than the base binder. The SBS modified binder and hybrid binder A and B resulted in the greatest reduction. Lower phase angle is associated with lower energy loss or more elastic behavior, which would indicate better rutting and crack ing resistance. S olubility results indicated that the coarser rubber in hybrid binder A and B as well as the ARB binders were not fully digested in the base binder made th e test results from DSR suspect because the presence of particulates in the binder is well known to affect DSR results The binders produced with the coarser grained rubber met, and even far exceeded requirements for PG7622 binder, resulting in binder performance parameters that indicated better performance characteristics than all other binders evaluated, including the SBS polymer modified binder. These results were not consistent with relative cracking performance characteristics determined from mixture tests. C onversely, solubility results indicated that the finer rubber in Hybrid binder C was fully digested in the base binder which made it suitable for DSR testing. This binder also met requirements for PG7622 binder with the exception of the maximum phase angle (which is an FDOT requirement). Fig 4 5 shows that all binders, including the base binder, mee t the specification requirement for a maximum G*sin of 5000 k P a for both the 100 C and 110 C PAV residue. All modified binders, except hybrid binder C, exhibited lower G*sin than the base binder. G*sin was intended to be an indicat or of resistance to fatigue cracking

PAGE 55

55 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Original binders RTFOT residue Binders PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 PG minimum 1.00 kPa PG minimum 2.20 kPa Figure 43 76 C (168.8 F) @ 10 rad/s 020 40 60 80 100 120 Original bindersRTFOT residue Binders o) PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 44 Phase a ngle 76 C (168.8 F) @ 10 rad/s

PAGE 56

56 0 1000 2000 3000 4000 5000 6000 7000 8000 100C PAV residue 110C PAV residue Binders PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 PG maximum 5000 kPa Figure 45 25 C (77 F) @ 10 rad/s 0 10 20 30 40 50 60 100C PAV residue 110C PAV residue Binders o) PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 46 Phase a 25 C (77 F) @ 10 rad/s

PAGE 57

57 because it represents a measure of energy loss (higher G*sin higher energy loss). However, post -SHRP research has revealed that this parameter may not relate very well to fatigue cracking resistance because a large part of the energy loss associated with G*sin is not related to damage. Fig. 4 6 shows that all modified binders result in phase angles lower than the base binder. Lower phase angles imply lower energy loss, but as with G*sin the energy loss associated with lower is not necessarily related to damag e. B BR Test Results Analysis The BBR tests asphalt binders at low pavement service temperatures to determine the binders propensity to thermal cracking. Thermal cracking of HMA pavements results when the temperature drops rapidly at cold temperatures. As the pavement contracts, stresses begin to build up within the HMA pavement layers. If the contraction occurs very rapidly the stresses can build and eventually exceed the stress relaxation ability of the HMA pavement. The BBR uses a transient creep load, applied in the bending mode, to load an asphalt beam specimen held at a constant low temperature. T he stiffness, S(t), is a measure of the thermal stresses developed in the HMA pavement as a result of thermal contraction. The slope of the stiffness curve, m is a measure of the rate of stress relaxation by asphalt binder flow. Accordingly, the Superpave binder specification requires a maximum limit of S(t) at 60 seconds and a minimum m value. Figure 4-7 and 4-8 show the results of S(t) and m value with 5% te sting error range for all tested binders. These two figures imply all binders meet specification requirement for both creep stiffness (S) and m value at 60 seconds loading time

PAGE 58

58 0 50 100 150 200 250 300 350 400 100C PAV residue 110C PAV residue Binders S(t) (MPa) PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 PG maximum 300 MPa Figure 47 S(t), 12 C (10.4 F) @ 60 second 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 100C PAV residue 110C PAV residue Binders m-value PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 PG minimum 0.300 Figure 48 m v alue, -12 C (10.4 F) @ 60 second

PAGE 59

5 9 Multiple Stress C reep Recovery (MSCR) MSCR (AASHTO TP70 -07, ASTM D7405) is used to identify the presence of elastic response in a binder and the change in elastic response at two different stress levels. The percent recovery in the MSCR test of asphalt binders is affected by the type and amount of polymer used in the polymer modified asphalt binder so it is intended to provide a means for determining if the polymer used in modification will provide an elastomeric response. Non -recov erable creep compliance (Jnr) has been shown to be an indicator of the resistance of an asphalt binder to permanent deformation (rutting) under repeated load. DAngelo et al. (2009) found that reducing Jnr by half typically reduced rutting by half. The fol lowing Fig. 4 9 and 410 show typical MSCR test results Two parameters were calculated and evaluated: average percent recovery and non -recoverable compliance at two different stresses 100 Pa and 3200 Pa. Percent recovery ) 100 ( Nr for N=1 to 10: 1 1 10100 ) ( ) 100 ( Nr Where 10 is the adjusted strain value at the end of recovery portion of each cycle. 1 is the adjusted strain value at the end of creep portion of each cycle. Non -recoverable compliance ) ( N Jnr for N=1 to 10: 10) ( N Jnr Where is the applied stress.

PAGE 60

60 Figure 49 Typical ten cycles creep and recovery with creep stress of 100 Pa Figure 410 Typical creep and recovery cycle with cr eep stress of 100 Pa

PAGE 61

61 Fig. 4 11 through 4-1 4 provide MSCR results with 5% testing error range in terms of percent recovery and non-recoverable compliance at different stress level s at two different test temperatures 67 .0 C (Fig 4 -11 and 4 -1 2 ), and 76.0 C ( Fig 4 1 3 and 41 4 ). In general, modified binders p ercent recovery is greater than control binder (PG 6722) at both test temperatures. Accordingly creep compliance i s much lower f or all modified binders than that of the control binder. However, among al l modified binders, hybrid binder C does not appear as competitive as others. Considered percent recovery aspect, hybrid binder C exhibits even lower recovery than ARB 5 Accordingly, hybrid binder C shows higher non-recoverable compliance at both test tem peratures. Since the percent recovery reflects the elastic response of the materials, the results seem to show that PG 76-22, hybrid binder A and B and ARB 12 exhibit the best elastic response among tested binders, and they are less sensitive to stress ch anges. According to DAngelo et al. (2009) and some other sources, these binders will exhibit less rutting fatigue in the field under the same loading conditions compared to other tested binders. Although it is believed that MSCR can identify the presence of polymers in asphalt binders, this is not true for hybrid binder C as shown in the results. Whether it is because the 3% radial SBS and 10% fine rubber in hybrid binder C have been completely dissolved when produced needs to be further tested and verifie d. All in all, the MSCR test results are strongly related the presence and concentration of coarse rubber (hybrid binder A, hybrid binder B and ARB -12) not just SBS polymer. P arameters obtained from the MSCR test distinguished most of the SBS polymer modi fied binder s from the control binder.

PAGE 62

62 0 20 40 60 80 100 0.1 kPa 3.2 kPa Binders Average Recovery (%) 1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 411 MSCR average r ecovery at 67 .0 C (152.6 F) (RTFOT residue) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.1 kPa 3.2 kPa Binders Jnr, kPa-1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 41 2 MSCR non -recoverable compliance at 67 .0 C (152.6 F) (RTFOT residue)

PAGE 63

63 0 20 40 60 80 100 0.1 kPa 3.2 kPa Binders Average Recovery (%) 1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 41 3 MSCR average recovery at 76 C (168.8 F) (RTFOT r esidue) 0.0 2.0 4.0 6.0 8.0 10.0 0.1 kPa 3.2 kPa Binders Jnr, kPa-1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 41 4 MSCR non -recoverable compliance at 76 C (168.8 F) (RTFOT residue)

PAGE 64

64 Elastic Recovery The elastic recovery is a measure of the tensile properties of the polymer modified asphalt cement RTFOT residue. It is measured by the per centage to which the asphalt cement residue will recover its original length after it has been elongated to a specific distance at a specified rate of speed and then cut in half Fig. 4 13 illustrates that the SBS modified binder and the hybrid binders exh ibited greater elastic recovery at 25 C than the base binder. Both rubber modified binders broke before the specified elongation of 20cm was reached, indicating that the rubber appears to make the binder more brittle at this temperature. Also, it appears t hat the presence of SBS made the binder less brittle (even when combined with rubber). H ybrid binder C which used rubber with the finest gradation, did not increase the elastic recovery as much as the SBS modified binder or the other two hybrid binders. 0.0 20.0 40.0 60.0 80.0 100.0 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Binders Elastic Recovery %) Figure 41 5 Elastic r ecovery at 25 C (77 F) (RTFOT r esidue)

PAGE 65

65 Force Ductility Test Force ductility test (AASHTO T 300) is used as a means of characterizing polymer modified binder. During the test, the specimen is elongated in a mold at a test temperature and loading rate, the results can be constructed as a stress and strain curve. An important parameter calculated from this test is the force ratio between first loading peak and second loading peak. The test was performed at elongation rate of 5 cm/minute at temperature of 4 C in water bath until the len gth reaches 100 cm or ruptures. Peak force ratios with 5% testing error range for all tested binders were plotted through Fig. 4 -16 and Fig. 417. Fig. 4 1 6 shows that all modified binders exhibit higher ratio of residual to peak force ( 1 2/ f f ) than the control binder which is similar to MSCR test results 0.0 0.2 0.4 0.6 0.8 1.0 Original binders RTFOT residue Binders f2/f1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 41 6 Force d uctility t est r esult 10 C (50 F)

PAGE 66

66 0.0 0.2 0.4 0.6 0.8 1.0 PAV residue Binders f2/f1 PG 67-22 PG 76-22 HB A HB B HB C ARB-5 ARB-12 Figure 4 17 Force ductility t est r esult 25 C (77 F) Binders performed slightly different at different test temperatures. At the test temperature 10 C, control binder (PG67-22) apparently shows less value than any other modified binders However, except for binders after RTFOT, hybrid binder C along with ARB 5 and ARB 12, d o not perform as good as PG 76 -22, hybrid binder A and B, which is consistent with MSCR test results. A s tress -Strain curve was constructed to observe binders response s to loading. The s train may be calculated as follows: A A L L L dLL L t 0 0ln ln0 Where, L0 Original length of specimen L Length of specimen after elongation

PAGE 67

67 A0 Original cross -sectional area of specimen A C ross -sectional area of specimen after elongation 0 50 100 150 200 250 300 350 400 450 0.0 0.5 1.0 1.5 2.0 2.5 Strain Stress (psi) 67-22 76-22 Hybrid_Binder_A Hybrid_Binder_B Hybrid_Binder_C ARB-5 ARB-12 Fig ure 4 1 8 Stress -s train d iagram of RTFOT r esidue at 10 C (50 F) Fig. 4 18 show s two characteristic primary and secondary loading regions From this figure it can be seen that both modified binders and control binder appear similar left half stress -strain curve, or asphalt modulus. However, as unloading occurs after peak stress, co ntrol binder continues to unload to approximately zero stress, following the modified binders unloading curve to the point where modified binders demonstrates secondary loading. At this point, the curves deviate, the control binder continuing to unload whi le most modified binders begin to increase in load. This secondary increase in load phenomenon can be regarded as an indicator of the presence of modifiers in binders, and the modifier, at this point is thought to begin carrying the load. Apparently there is some difference for different modifiers in carrying secondary load capability.

PAGE 68

68 ARB-5 and ARB -12 broke they reach certain strains while hybrid binders and PG 7622 keep increasing secondary half until reach either higher stress or higher strain, which can be attributed to the either the presence of SBS polymers or the cross -link interaction between crumb rubber and SBS polymer. In a word, force ductility parameters f2/f1 is helpful to differentiate most SBS polymer modified binders from control binder. In addition, the stress -strain curve constructed from force ductility test, especially the secondary half, seems to be a good indicator of the presence of SBS polymers and the cross -link of SBS polymer and crumb rubbers. Summary The physical property results reveal the information that control binders (PG 7622) can be differentiated from polymer and rubber modified asphalt, but purely polymer modified binder PG 76-22 can not be clearly distinguished from hybrid binders or sometimes even ARB -12 simply by t hese physical properties. The solubility test results show that some coarse crumb rubber particles were not completely digested in asphalt, which affected binders rheology test results. Since the DSR test requires asphalt specimen with specific shape, the DSR results were influenced the most by the undigested particles in the binder. Therefore, DSR test results provide little meaningful information with respect to binders ability to resist rutting or cracking. Combined with Elastic Recovery test and Forc e Ductility test, MSCR test can not only distinguish modified binders from control binders, but differentiate crumb rubber modified binders ARB 5 and ARB -12 from hybrid binders: ARB 5 and ARB -12 broke at some strain level where PG 76-22 and hybrid binders still exhibited smooth extension. It

PAGE 69

69 is believed that it is the cross link of CRM and SBS polymer which improved binders ductility. Although test results showed the hybrid binder A and B behave as well as PG 7622, hybrid binder C did not exhibit the sam e competency in some properties such as the elastic recovery and non-recoverable compliance. Why this is the case implies further binder test needs to be developed to verify the existence of polymers in binders. Also, whether it is true that binders showin g improved performance in binder test will exhibit the same in mixture test needs to be verified in the following mixture tests.

PAGE 70

70 CHAPTER 5 MIXTURE IDT TEST RESULTS AND ANALYSIS Mixture Test Results In accordance with AASHTO T 322, s tan dard Superpave Indirect Tension Test (IDT) was performed at 10C on all mixtures to determine resi lient modulus (Mr), creep compliance (m value and D1), tensile strength (St), failure strain ( f), fracture energy (FE) and dissipated creep st rain energy (DCSE) (Roque, 1997) to failure. Results were combined and analyzed using Hot Mix-Asphalt (HMA) Fracture Mechanics Model (Zhang, 2001) and Energy Ratio Theories (Roque, 2004), to evaluate the mix tures resistance to cracking. The number of specimens and testing cycles are listed in Table 5-1. A total number of 132 IDT specimens were tested for this project. For each specific type of mixture, three specimens were tested and the variability of the s pecimens was considered and treated by using a trimmed mean approach. Table 5 1 Summary of t otal m ixture t ests All test results and calculated parameters are listed in Table 5-2 through Table 57. Mixture Type Aggregate Type Conditions Types of Binders Number of Replicat e s Total No. of Mixture Tests OGFC Limestone LTOA/STOA 5 3 90 Gran ite LTOA/STOA 5 3 90 Superpave Dense Limestone LTOA/STOA 6 3 108 Granite LTOA/STOA 6 3 108 Totals 4 2 7 132 396

PAGE 71

71 Table 5 2 DG m ixtures c reep and d amage t est r esult s Aggregate Binder Type Aging Conditions m value D 1 (1/psi) D(1000 sec) (1/GPa) d(D)/ dt(1000 sec) Granite PG 67 -22 STOA 0.668 4.77E 07 7.055 3.20E 08 LTOA 0.532 4.48E 07 2.619 9.43E 09 PG 76 -22 STOA 0.534 7.54E 07 4.414 1.61E 08 LTOA 0.413 5 .43E 07 1.414 3.88E 09 HB A STOA 0.446 5.93E 07 1.926 5.76E 09 LTOA 0.411 4.35E 07 1.128 3.05E 09 HB B STOA 0.455 9.17E 07 3.110 9.64E 09 LTOA 0.438 5.18E 07 1.584 4.66E 09 HB C STOA 0.521 7.52E 07 4.074 1.43E 08 LTOA 0.402 6.73E 07 1.602 4. 33E 09 ARB-5 STOA 0.600 3.841E 07 3.575 1.45E 08 LTOA 0.576 3.05E 07 2.444 9.44E 09 Limestone PG 67 -22 STOA 0.477 5.42E 07 2.176 6.99E 09 LTOA 0.385 4.892E 07 1.062 2.69E 09 PG 76 -22 STOA 0.436 5.44E 07 1.665 4.83E 09 LTOA 0.308 6.60E 07 0.83 1.70E 09 HB A STOA 0.376 6.24E 07 1.291 3.15E 09 LTOA 0.327 4.12E 07 0.628 1.29E 09 HB B STOA 0.386 4.26E 07 0.948 2.38E 09 LTOA 0.300 5.30E 07 0.652 1.27E 09 HB C STOA 0.406 5.38E 07 1.353 3.63E 09 LTOA 0.348 3.44E 07 0.592 1.32E 09 ARB-5 STOA 0.506 6.08E 07 3.019 1.02E 08 LTOA 0.392 4.72E 07 1.069 2.78E 09

PAGE 72

72 Table 5 3 DG m ixtures s trength and f racture t est r esults Aggregate Binder Type Aging Conditions S t (MPa) M R (GPa) e f (micro) Ninitiation N propagation (2in) FE (kJ/m 3 ) DCSE H MA (kJ/m 3 ) Granite PG 6722 STOA 2.14 10.85 2566.05 1.63E+04 5.58E+03 4.2 4.0 LTOA 2.25 11.99 1336.78 2.02E+04 6.92E+03 2.2 2.0 PG 76 22 STOA 2.23 10.55 3326.20 3.15E+04 1.08E+04 5.5 5.3 LTOA 2.59 11.37 1824.64 6.01E+04 2.06E+04 3.5 3.2 HB A STOA 1.90 11.55 1272.15 2.24E+04 7.68E+03 1.8 1.6 LTOA 2.26 14.13 940.13 3.14E+04 1.07E+04 1.5 1.3 HB B STOA 1.92 10.12 2426.19 2.84E+04 9.73E+03 3.6 3.4 LTOA 2.08 11.96 1537.91 3.51E+04 1.20E+04 2.3 2.1 HB C STOA 2.02 11.35 2285.38 2.17E+04 7. 42E+03 3.5 3.3 LTOA 2.44 13.23 1423.10 3.73E+04 1.28E+04 2.5 2.3 ARB5 STOA 2.12 13.26 1470.04 1.64E+04 5.62E+03 2.3 2.1 LTOA 2.12 13.85 1100.17 1.62E+04 5.53E+03 1.6 1.4 Limestone PG 67 22 STOA 2.17 11.88 1167.65 1.69E+04 5.80E+03 1.6 1.4 LTO A 2.2 13.62 1066.45 1.69E+04 5.80E+03 1.5 1.3 PG 76 22 STOA 2.41 11.36 1431.47 3.25E+04 1.11E+04 2.3 2.0 LTOA 2.71 11.97 1294.71 7.37E+04 2.52E+04 2.5 2.2 HB A STOA 2.04 11.16 1000.95 2.57E+04 8.81E+03 1.4 1.2 LTOA 2.02 12.00 707.20 3.38E+04 1 .16E+04 0.9 0.7 HB B STOA 2.40 11.87 1116.24 4.49E+04 1.54E+04 1.8 1.6 LTOA 2.33 11.94 864.94 4.76E+04 1.63E+04 1.3 1.1 HB C STOA 2.32 12.56 1116.28 3.14E+04 1.07E+04 1.8 1.6 LTOA 2.62 12.88 962.87 6.80E+04 2.33E+04 1.7 1.4 ARB5 STOA 1.9 10.81 1185.45 1.18E+04 4.05E+03 1.5 1.3 LTOA 2.38 13.53 999.93 3.48E+04 1.19E+04 1.6 1.4

PAGE 73

73 Table 5 4 DG m ixtures e nergy r atio r esults Aggregate Binder Type Aging Conditions DCSEMIN, (kJ/m3) ER@ stress, 150 psi Granite PG 67 -22 STOA 2.971 1.34 LTOA 1. 440 1.38 PG 76 -22 STOA 2.440 2.16 LTOA 0.852 3.76 HB A STOA 1.081 1.52 LTOA 0.646 2.04 HB B STOA 1.773 1.93 LTOA 0.910 2.33 HB C STOA 2.206 1.51 LTOA 0.956 2.38 ARB-5 STOA 1.738 1.23 LTOA 1.226 1.17 Limestone PG 67 -22 STOA 1 .247 1.12 LTOA 0.595 2.22 PG 76 -22 STOA 0.984 2.08 LTOA 0.438 5.01 HB A STOA 0.695 1.75 LTOA 0.302 2.42 HB B STOA 0.537 2.90 LTOA 0.312 3.43 HB C STOA 0.781 2.03 LTOA 0.325 4.41 ARB-5 STOA 1.617 0.82 LTOA 0.619 2.25

PAGE 74

74 Ta ble 5 5 OGFC m ixtures c reep and d amage t est r esults Aggregate Binder Type Aging Conditions m value D1 (1/psi) D(1000 sec) (1/G P a) d(D)/dt(1000 sec) Granite PG 76 -22 STOA 0.599 1.49E 06 13.601 5.59E 08 LTOA 0.577 8.68E 07 6.851 2.70E 08 H B A STOA 0. 487 1.15E 06 4.929 1.63E 08 LTOA 0.459 6.88E 07 2.496 7.52E 09 H B B STOA 0.478 1.64E 06 6.491 2.13E 08 LTOA 0.439 1.65E 06 5.035 1.50E 08 H B C STOA 0.537 1.31E 06 7.932 2.87E 08 LTOA 0.570 6.29E 07 4.804 1.84E 08 ARB-12 STOA 0.557 8.38E 07 5 .828 2.19E 08 LTOA 0.555 7.47E 07 5.118 1.91E 08 Limestone PG 76 -22 STOA 0.434 8.83E 07 2.657 7.65E 09 LTOA 0.365 9.02E 07 1.741 4.11E 09 H B A STOA 0.458 6.35E 07 2.254 6.86E 09 LTOA 0.366 5.12E 07 0.994 2.36E 09 Hybrid Binder B STOA 0.451 9. 50E 07 3.199 9.62E 09 LTOA 0.416 4.89E 07 1.310 3.61E 09 Hybrid Binder C STOA 0.521 6.53E 07 3.522 1.24E 08 LTOA 0.408 9.95E 07 2.484 6.80E 09 ARB-12 STOA 0.533 5.87E 07 3.500 1.25E 08 LTOA 0.427 6.26E 07 1.824 5.13E 09

PAGE 75

75 Table 5 6 OGFC m ixt ures s trength and f racture t est r esults Aggregate Binder Type Aging Conditions S t (MPa) M R (G p a) e f (micro) Ninitiation N propagation (2in) FE (kJ/m 3 ) DCSE HMA (kJ/m 3 ) Granite PG 76 22 STOA 1.61 5.29 3601.16 2.14E+04 7.33E+03 4.5 4.3 LTOA 1.44 6.46 1454. 68 1.39E+04 4.77E+03 1.5 1.3 Hybrid Binder A STOA 1.35 6.13 1538.19 2.51E+04 8.58E+03 1.6 1.5 LTOA 1.38 8.92 674.36 1.84E+04 6.31E+03 0.6 0.5 Hybrid Binder B STOA 1.33 5.47 1966.58 2.43E+04 8.33E+03 2.0 1.8 LTOA 1.54 4.92 2638.98 5.35E+04 1.83E+0 4 3.1 2.9 Hybrid Binder C STOA 1.07 5.81 1018.97 5.91E+03 2.02E+03 0.7 0.6 LTOA 1.43 6.59 1136.02 1.60E+04 5.46E+03 1.2 1.0 ARB-12 STOA 1.17 6.93 1499.10 1.54E+04 5.28E+03 1.3 1.2 LTOA 1.27 7.29 1215.67 1.46E+04 4.98E+03 1.1 1.0 Limestone PG 76 22 STOA 1.58 7.83 1107.59 3.83E+04 1.31E+04 1.2 1.0 LTOA 1.50 8.53 732.86 3.89E+04 1.33E+04 0.7 0.6 Hybrid Binder A STOA 1.59 7.42 1175.16 5.04E+04 1.73E+04 1.4 1.2 LTOA 1.82 9.71 916.91 1.11E+05 3.80E+04 1.1 0.9 Hybrid Binder B STOA 1.64 7.28 12 11.57 3.55E+04 1.22E+04 1.4 1.2 LTOA 1.77 8.23 1220.33 1.02E+05 3.49E+04 1.5 1.3 Hybrid Binder C STOA 1.56 7.99 1073.92 2.15E+04 7.34E+03 1.1 0.9 LTOA 1.62 7.03 975.14 3.78E+04 1.29E+04 1.1 0.9 ARB-12 STOA 1.45 9.10 1058.80 2.45E+04 8.38E+03 1.2 1.1 LTOA 1.57 10.16 1013.60 5.37E+04 1.84E+04 1.1 1.0

PAGE 76

76 Table 5 7 OGFC m ixtures e nergy r atio r esults Aggregate Binder Type Aging Conditions DCSE MIN (kJ/m 3 ) ER @ stress 150 psi Granite PG 76 -22 STOA 6.326 0.7 LTOA 3.246 0.41 Hybrid Binder A STOA 2.578 0.56 LTOA 1.290 0.38 Hybrid Binder B STOA 3.449 0.53 LTOA 2.758 1.04 Hybrid Binder C STOA 3.793 0.16 LTOA 2.265 0.46 ARB-12 STOA 2.740 0.44 LTOA 2.436 0.41 Limestone PG 76 -22 STOA 1.427 0.73 LTOA 0.868 0.65 Hybrid Binder A STO A 1.208 1.02 LTOA 0.515 1.80 Hybrid Binder B STOA 1.735 0.70 LTOA 0.715 1.83 Hybrid Binder C STOA 1.821 0.52 LTOA 1.348 0.68 ARB-12 STOA 1.735 0.62 LTOA 0.969 1.01

PAGE 77

77 Analysis of IDT Test Results Since currently there is no single mixture property or characteristic that can reliably predict top-down cracking performance of HMA (Roque, 2004), a number of mixture parameters obtained from the IDT were evaluated by using HMA fracture mechanics and DCSE theory to determine the mixtures potenti al to cracking. In addition, some observations regarding mixture preparation were cited as they helped to explain some of the findings. Since the relative cracking performance was different in the two types of mixtures evaluated, the analysis was categoriz ed into two parts: dense-graded (DG) mixtures and opengraded friction course (OGFC) mixtures. DG Mixtures The number of loading cycles for crack initiation (Ninitiation) and to 5-mm of propagation (Npropagation) were calculated from Dissipated C reep Strain Energy to failure (DCSEf) and the DCSE/cycle concepts based on resilient modulus, creep test and tensile strength test results (Appendix B and C ). Energy Ratio, defined as the dissipated creep strain energy threshold of the mixture divided by the minimu m dissipated creep strain energy required, is a criterion recently developed by Roque, et al.(2004) to evaluate top-down cracking performance of mixtures. These three parameters: Ninitiation, Npropagation and ER were used as the principal basis to evaluate the mixtures cracking performance in this research. Fig 5 1 through 56 show that hybrid binder mixtures generally performed better than both PG 67-22 and ARB -5 mixtures regardless of aggregate types and aging conditions. These figures also show that SBS polymer modified binder mixtures exhibited superior performance among all mixtures regardless of aggregate type or aging condition.

PAGE 78

78 IDT: 10 C (50 F), 100 psi Loading 0.00E+00 2.00E+04 4.00E+04 6.00E+04 8.00E+04 STOA LTOA Aging Conditions Ninitiation PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 51 Ninitiation for DG g ranite m ixtures IDT: 10 C (50 F), 100 psi Loading 0.00E+00 7.50E+03 1.50E+04 2.25E+04 3.00E+04 STOA LTOA Aging Conditions Npropagation PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 52 Npropagation for DG g ranite m ixtures

PAGE 79

79 IDT: 10 C (50 F), 100 psi Loading 0.00E+00 2.00E+04 4.00E+04 6.00E+04 8.00E+04 STOA LTOA Aging Conditions Ninitiation PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 53 Ninit iation for DG l imestone m ixtures IDT: 10 C (50 F), 100 psi Loading 0.00E+00 7.50E+03 1.50E+04 2.25E+04 3.00E+04 STOA LTOA Aging Conditions Npropagation PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 54 Npropagation for DG l imestone m ixtures

PAGE 80

80 IDT: 10 C (50 F) 0.00 1.50 3.00 4.50 6.00 STOA LTOA Aging Conditions ER @ 10 0C PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 55 ER for DG g ranite m ixtures IDT: 10 C (50 F) 0.00 1.50 3.00 4.50 6.00 STOA LTOA Aging Conditions ER @ 10 0C PG 67-22 PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-5 Figure 56 ER for DG l imestone m ixtures

PAGE 81

81 If considered by STOA and LTOA separately, all three hybrid binders were found exhi biting similar cracking resistance trends for both granite and limestone mixtures. However, if compared for the same mixtures with different aging conditions, different cracking performance trends were observed: the LTOA apparently increased the cracking r esistance of hybrid binder mixtures. A larger increase in cracking resistance was observed for limestone mixtures, which could be explained by the fact that limestone has a much rougher surface texture and greater absorption than granite. Therefore, it is hypothesized that laboratory aging at 85C (LTOA) results in more binder being absorbed by the limestone, which in these mixtures appeared to increase resistance to damage with little or no reduction in fracture energy limit. The ARB 5 mixtures did not exh ibit improvements in cracking resistance to the PG 6722 mixtures. This result is consistent with previous research which indicated that rubber alone did not improve cracking resistance of mixtures. As for the other mixtures, aging effects were found to be particularly acute in the limestone mixtures. Once again it is hypothesized that these effects may be somewhat artificially caused by increased absorption in these aggregates during LTOA. OGFC Mixtures Although the relative performance of hybrid binders in OGFC mixtures was somewhat different from that observed in DG mixtures, Fig. 5 7 through 512 show that hybrid binders exhibited similar or better cracking resistance than both SBS polymer modified binder and ARB -12 in OGFC mixtures, except for one spec ial case (hybrid binder C STOA in granite mixture). This result was true for all parameters evaluated (Ninitiation Npropagation and ER) for both aggregate types and aging levels. Hybrid binders A and B resulted in OGFC mixtures with particularly high re sistance to cracking,

PAGE 82

82 especially for the LTOA condition and limestone aggregate. These effects are likely responsible: the coarse rubber binders may be more resistant to agehardening and the limestone aggregate absorbs more asphalt during LTOA, therefore making the mixture more resistant to damage. It is interesting to note that the hybrid binders exhibited greater cracking resistance than ARB 12, indicating that the addition of SBS polymer provided an added benefit. The relatively low fracture resistance exhibited by hybrid binder C with the fine rubber, and granite aggregate was probably a result of binder redistribution (partial draindown), rather than the quality of the binder itself. The smoother texture and lower absorption of the granite, combined wi th the lower viscosity of the finer rubber binder provide an explanation for this phenomenon. These factors may have contributed to the binders inability to maintain a uniform distribution within the granite OGFC, therefore creating areas of relative weak ness within the mixture. This effect was minimized or eliminated where the rougher, more absorptive limestone aggregate was used. In summary, it appears that the hybrid binders evaluated in this study can be used as a substitute for either SBS modified (PG 76 -22) or ARB-12 in OGFC mixtures. However, there may be a need to check on draindown potential of hybrid binder produced with finer rubber when used in smooth textured, nonabsorptive aggregate OGFC mixtures. Summary In general, the IDT test results sho wed that all mixtures with hybrid binders, regardless of aggregate types and aging conditions, performed comparatively better than PG 67-22 and ARB -5 mixtures in terms of cracking resistance. Better cracking

PAGE 83

83 IDT: 10 C (50 F), 100 psi Loading 0.00E+00 1.50E+04 3.00E+04 4.50E+04 6.00E+04 STOA LTOA Aging Conditions Ninitiation PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 57 Ninitiation for OGFC g ranite m ixt ures IDT: 10 C (50 F), 100 psi Loading 0.00E+00 7.50E+03 1.50E+04 2.25E+04 3.00E+04 STOA LTOA Aging Conditions Npropagation PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 58 Npropagation for OGFC g ranite m ixtures

PAGE 84

84 IDT: 10 C (50 F), 100 psi Loading 0.00E+00 3.00E+04 6.00E+04 9.00E+04 1.20E+05 STOA LTOA Aging Conditions Ninitiation PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 59 Ninitiation for OGFC l imestone m ixtures IDT: 10 C (50 F), 100 psi Loading 0.00E+00 1.50E+04 3.00E+04 4.50E+04 6.00E+04 STOA LTOA Aging Conditions Npropagation PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 510 Npropagation for OGFC l imestone m ixtures

PAGE 85

85 IDT: 10 C (50 F) 0.00 0.50 1.00 1.50 2.00 STOA LTOA Aging Conditions ER PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 511 ER for OGFC g ranite m ixtures IDT: 10 C (50 F) 0.00 0.50 1.00 1.50 2.00 STOA LTOA Aging Conditions ER PG 76-22 Hybrid Binder A Hybrid Binder B Hybrid Binder C ARB-12 Figure 512 ER for OGFC l imestone m i xtures

PAGE 86

86 response observed in hybrid binder mixtures compared to both unmodified and asphalt rubber modified binders offer the promise of using tire rubber while providing similar performance benefit as polymer modified asphalts. If STOA and LTOA were consid ered separately, all three hybrid binders exhibited similar cracking resistance trends for both granite and limestone mixtures. However, the same mixtures showed different cracking performance trends at different aging conditions: the LTOA apparently increased the cracking resistance of hybrid binder mixtures. A larger increase in cracking resistance was observed for limestone mixtures, which could be explained by the fact that limestone has a much rougher surface texture and greater absorption than granite. In summary, it appears that the hybrid binders evaluated in this study can be used as a substitute for either SBS modified (PG 76-22) or ARB-12 in OGFC mixtures. However, there maybe a need to check the draindown potential of hybrid binder produced with finer rubber when used in smooth textured, nonabsorptive aggregate OGFC mixtures.

PAGE 87

87 CHAPTER 6 DEVELOPMENT AND EVALUATION OF HEALING TEST Experimental and Theoretical Background of Healing Test As stated in Chapter 2, healing properties of HMA mixtures hav e been regarded as an important parameter to predict HMA pavement life. This chapter explains about develop ing IDT programs to evaluate the healing potentials of dense graded granite mixtures with different binder types. The healing test basically comprise s of two parts: damage phase and healing phase. Results were analyzed and compared to show how hybrid binders behave differently from control binders (PG 67 22, ARB -5) and SBS modified binders with respect to damage and healing. Fatigue Test with Static and Cyclic Loading Typically, there are two different types of loading modes to test fatigue life of IDT specimens, static loading with constant loading rate or displacement rate and cyclic loading with rest period or without rest period. The first loading condition is usually used to obtain the tensile strength of the HMA IDT samples while the second one is to simulate the real traffic loading conditions on the road. During the standard Superpave IDT strength test, gyratory compressed samples with 7% air vo ids were loaded at a constant displacement rate. For OGFC samples, the rate is set as 100 mm/min, whereas for DG mixtures, 50 mm/min. The reason for using these loading rates during the strength test is to allow little or no time for stress relaxat ion or c reep to develop prior the specimen failure This characteristic of static loading mode makes its elf impractical for healing, as we know the purpose of healing test is trying to measure how much damage will be recovered after loading is removed from the mat erial. If the static loading were used to evaluate the damage phase, it would

PAGE 88

88 be almost impossible to differentiate delayed elasticity recovery or stress relaxation from damage recovery. Therefore, static loading mode will not be considered to damage the m aterial for healing test. If cycl ic loading mode were decided for the damage phase of healing test, three other aspects need to be considered: loading shape, loading amplitude and rest period. T he relationship between static loading and cyclic loading only is disc ussed, whereas more discussion can be found in the following subchapter, IDT strength test results, obtained under static loading mode, should be analyzed as a reference to cyclic loading. It is believed that if the HMA specimen subjected to loading less than certain amplitude in a r elatively short period (0.1 second, for example) it behaves as an elastic material which means the stress and strain relationship will be linear Therefore it is necessary to know in advance in what loading amplitude r ange the HMA material will behave as an elastic material. A typical horizontal stress-strain curve from IDT strength test for dense graded granite mixture is shown in Fig. 61. From Fig. 6 -1 it can be seen that the stress -strain relationship c ould be rega rded as linear at the beginning of loading curve (for loading value less than 150 psi). Since the resilient modulus is computed as the tangent of this linear relationship, it can be regarded as a constant parameter during instant loading period (0.1 second). Also loading magnitude which will be used in healing test should be in this linear range (less than 150 psi). Based on this point of view, IDT strength test results for DG mixtures will be used as a reference to decide loading magnitude for the cyclic l oading during healing test. If the IDT sample experienced cyclic loading with certain amplitude, it will deteriorate and fail after a certain number of N cycles and the fatigue life can be

PAGE 89

89 0 50 100 150 200 250 300 0 200 400 600 800 1000 1200 1400 1600 1800 Horizontal Strain (10-6) Applied Stress (psi) Figure 61 Typical strength test result for DG mixtures, STOA 10 C obtained through the stress and strain curve. The typical loading waves used in IDT test is the haversine wave with 0.1 second loading followed by 0.9 second rest period However, loading period and rest period may change for different testing purp oses, which will be discussed in the following subchapters. Damage and Healing In this research, there will be two different ways to describe damages made to HMA materials: dissipated energy and resilient modulus reduction. The dissipated energy can be ob tained through work from cyclic loading, whereas modulus reduction needs to be measured through resilient deformation changes. Because for the same material in the same testing environment, resilient deformation is inverse to resilient modulus of the tested HMA material, the fatigue curve could be expressed by 1/MR versus loading cycles as shown in Fig. 6 -2. T he fatigue process could be divided into

PAGE 90

90 three parts: at the beginning, the mixture experienced microdamage, heat ing and reversal of steric hardening. After that, the mixture will experience a steady -state damage process where 1/MR is linear to number of loading cycles After a certain amount of loading cycles the rate of DCSE loading curve rapidly increases with loading cycles, at which point macrocr ack occurs. 1/MRN (cycle) t (second) Load Heat + reversal of steric hardening + damage Steady state damage Crack and damage Figure 62 HMA material fatigue curve under cyclic loading As shown in Fig. 6 2 the HMA material will experience a steady state damage period before it cracks under cyclic loading. Once cracks ccur in the mater ial, it is no longer considered healable. Therefore, i n order to observe healing potentials of HMA mat erials, the accumulated damage should be controlled in such a manner that it wont cause any macrocrack s in the material. If the reduction of resilient modulus is considered as a parameter to describe damage and healing, the total resilient modulus should decrease with loading. As soon as the cyclic loading removed, healing starts in the HMA material.

PAGE 91

91 What is ha ppening to the asphalt mixture after loading was removed is a complicated process: the mixture is not only experiencing healing, but also cooling and long term steric hardening as shown in Fig. 63 However, the steric hardening and cooling have such little effects on resilient modulus recovery, they were considered negligible. Therefore, the resilient deformation measured in this healing test will not count in cooling and steric hardening effects. The healing procedure itself also consists of reconstruction of chemical bonds in asphalt and particle f low in microscopic scale (Little et al. 1997) Fig 6 3 shows the sketch of the damage recovery process. 1/MRT (second) Steric Hardening Heat release (cooling) Healing t (second) Load 0.1s 0.9s Ti Heat and reversal of steric hardening Damage Figure 6 3 Damage recovery curve Healing Test Development Cyclic Damage Loading (C DL ) Mode There are t hree most importa nt elements need to be considered when choosing CDL mode: loading amplitude, shape and rest period. Among these three elements, loading amplitude and shape is directly related to damage a pplied to the material whereas rest period determines how much delay ed elasticity could be recovered. In this

PAGE 92

92 section, only loading shape and rest periods were discussed, whereas loading amplitude will be discussed in the following section. Major concerns about loading shapes are maneuverability and side effects. B ecause of the limitation of available testing equipments, only haversine and square shapes of loading were analyzed Compared to haversine loading, it is more difficult to control square loading in MTS machine which applies loading by hydraulic transmission. As w e know square loading is applied by adjusting frequencies and amplitude of harmonic loading, it is always hard to minimize the harmonic loading effects at the edge of square load ing, which causes work instability to the tested specimen. Another issue need to be addressed about square loading is that u pon the command of loading removal from the control program loading in the MTS machine can not be instantaneously removed, which causes additional damage to the tested specimen and it is almost impossible to m easure that. In contrast haversine wave loading does not have this problem and it is convenient to compute dissipated energy per loading cycle. In addition to the problem s stated above, Haversine load assure s longer res t period for del ayed elasticity; Sq uare load may need much longer time than haversine load to assure delayed elasticity resumption. Therefore, haversine w ave loading shape was considered for the damage phase during healing test As long as the loading shape was decided, rest period needs t o be studied because various rest periods may cause vast difference in results if not chosen reasonably. The main effects that rest period may cause are summarized in the following: Effect on healing Delayed elastic recovery

PAGE 93

93 Effect on measured MR t (seconds) Load 0.1s 0.1s, 0.2s,0.4s, 0.9s Load 0.1s 0.4s, 0.9s,1.9s Figure 64 Haversine loading waves In order to study how different rest periods affect the material responses to loading several rest periods: 0.1, 0.2, 0.4 and 0.9 second (Fig. 64) following 0.1 second haversine wave loading were tried during damage phase Since it is believed that healing will occur as soon as loading i s removed from the specimen, t he major consideration of different res t period s should be minimization of healing during the rest period while allow ing delayed elas ticity fully recovered. All four tests were applied on the same IDT specimen and the final resilient deformations were normalized for a direct comparison of the results. The ever first cycle of loading curves obtained during fatigue test was picked and tr immed Results were plotted in Fig. 6-5. From Fig. 65 it can be seen that if the material experienced only 0.1 second rest period 10.4% of resilient deformation would be missed compared to 0.9 second rest period. This would def initely result in overestim ation of resilient modulus of the material. Another effect should be addressed here is that cyclic loading with 0.1 second rest period will lead to large permanent deformation in a relatively short loading time which will result in an earlier specimen fai lure. The same problem will happen to loading with 0.2 second rest period as well

PAGE 94

94 0 0.0001 0.0002 0.0003 0.0004 0 0.2 0.4 0.6 0.8 1 Relative Time (second) Trimmed Initial Resilient Deformation (in) 10.4% Permanent Deformation Total Resilient Deformation 2.6% 6.9% End of loading 0.2 second rest 0.4 second rest 0.9 second rest 0.1 second rest Figure 65 Initial resilient deformation with different rest periods, 10 C Further time of loading was performed to evaluate rest periods effects on materials resili ent deformation Fig. 6 -6 shows the normalized resilient deformation of the dense graded granite IDT specimen which was loaded with the same loading amplitude of 1755 lbs for 0.1 second per cycle, but with different rest periods From Fig. 6 -6 it can be seen that rest period has a significant inf luence on HMA material resilient deformation changes Since short term rest period allows less time for delayed elasticity recovery, the tested material deforms in a much faster rate compared to long term rest period s. However, short term rest period show s some advantages that long term rest periods can not compete with: certain amount of damage to the material could be reached at a relatively short time As shown in Fig. 6 -6, 10 minutes loading could result in up to 50% reduction in resilient modulus The shortage of short term rest period is not

PAGE 95

95 allow ing enough time for full delayed elasticity recovery, which is practically difficult to be differentiated from healing at the end of damage phase during healing test Co nsequently, it is not convenient to evaluate the healing potentials of the material if short term rest period was applied to cyclic loading. Moreover, F ig. 6-6 discloses another piece of important information: r esilient deformation almost remains the same during the whole damage phase for 0.9 second rest period loading This result implies that 0.9 second rest period allows enough time not only for delayed elasticity recovery, but also for complete healing in the material. Compared to other short term rest periods, 0.9 second rest is such a long term period that microdamage in the HMA material has been fully healed A further proof of this conclusion is shown in Fig. 67 From Fig. 6 -7 it can be seen that even if the loading amplitude has been almost doubled (1100 lbs to 1974 lbs) the reduction in resilient deformation did not show any significant difference. If Fig 6-7 was interpreted by dissipated energy, t he DCSE/cycle of loading 1974 lbs is about 4 times of that of loading 1100 lbs. However, Fig. 6-7 sho ws that all damage has been fully recovered in 0.9 second rest period. In Fig. 6 -6, r esults from 0.2 and 0.4 second rest period during damage phase have shown that there is an ideal rest period which not only allows enough time for the material to fully r ecover delayed elasticity, but also limit the time for healing. For 0.2 second rest period loading case in Fig. 6-6 most of the damage occurred during the initial 2 minutes. As it is known that during the initial minutes of damage phase, most of the defor mation is due to materials heating and reverse of steric hardening. Apparently the 0.2 second rest period loading curve has shown that most of the damage has

PAGE 96

96 0 0.5 1 1.5 2 0 10 20 30 40 50 60 Time (minute) Normalized Horizontal Resilent Deformation 0.1 second rest 0.2 second rest 0.4 second rest 0.9 second rest Figure 66 Rest period effects on resilient deformation, 1755 lbs, 10 C 0 0.5 1 1.5 2 0 10 20 30 40 50 60 Time (minute) Normalized Horizontal Resilent Deformation 1100 lbs 1316 lbs 1535 lbs 1755 lbs 1974 lbs Figure 67 CDL t est results, 0.9 second rest, 10 C

PAGE 97

97 occurred in the beginning minutes, which makes it impossible to separate heat and reverse of steric hardening from damages. However, the 0.4 second rest period loading curve didnt show this problem. The 0.4 second rest period could be considered as a balance between delayed elasticity and damage healing. Therefore, 0.4 second rest period was chosen for the damage phase during healing test. Loading Amplitude Verification After the rest period was decided, loading amplitud e for CDL needs to be finalized There are three issues need to be considered for loading amplitude : first, the amplitude should be in such a range that the induced stress is linear to resilient deformations; second, the materials stiffness and ductility e ffects and third, number of CDL cycles should be in a reasonable testing period. Although in IDT tensile strength test the HMA specimen was loaded with static loading mode, r esults presented a clear relationship between loading stress and horizontal resili ent deformation for STOA dense graded mixtures. As shown in Fig. 68, at the beginning of loading (load amplitude less than 2000 lbs), the applied force could be regarded as linear to horizontal deformation. This would be true especially when the material undergoes 0.1 second quick loading. The linear relationship guarantees that if material is loaded within this range, the error of resultant resilient modulus calculated through stress -strain relationship can be negligible. And this is true for all dense gr aded granite mixtures with all types of binders. Another problem with loading more than 2000 lbs is that it cause s more temperature changes and reversal of steric hardening at the beginning of damage phase Also because of the IDT specimen dimension, high level loading can cause stress concentration at the loading edge, which results in bulging problems at an early

PAGE 98

98 -5000 -4000 -3000 -2000 -1000 0 0.000 0.001 0.002 0.003 0.004 0.005 Horizontal Deformation (in) Applied Force (lbf) DGUS DGRS DGAS DGBS DGCS DGMS Linear relationship P=k*h Figure 68 DG mixtures IDT strength test, 10 C 0 0.5 1 1.5 2 0 10 20 30 40 50 60 Time (minute) Normalized Horizontal Resilent Deformation 1600 lbs 1755 lbs Figure 69 DGUS cyclic loading test results, 0.4 second rest, 10 C

PAGE 99

99 loading stage. Ulti mately, the specimen fails at the loading edge rather than at the center of the specimen. In contrast, if the loading amplitude is too low, the temperature changes and steric hardening may not be problems any more, but it would cost too much time to finish the test, which is not desirable for this research (Fig. 69). Two criteria were chosen to terminate CDL for the healing test: percentage of DCSEf (50%) (Fig. 6 -10) or reduction in resilient modulus (30% reduction) which means if either of these two criteria is reached, the damage phase will be stopped and loading removed. Both of these two criteria were monitored during CDL in such a manner that the accumulated microdamage in the specimen would not lead to macrodamage which is considered not healable T he loading amplitude also affects the amount of dissipated creep strain energy done to the specimen per loading cycle. Figure 6 10 DCSE obtained from tensile strength test The DCSEf could be obtained from regular IDT test and DCSE/cycle can be expresse d as :

PAGE 100

100 1 1 2 1 0 0 2 2 2 1 0 0 2 1 0 01000 20 1 ) 10 ( sin ) ( ) 10 ( sin ) ( ) 10 sin( ) 10 sin( m AVE AVE AVE pMAX AVEm D dt t t D dt t t D dt t t cycles DCSE Where D1 and m are creep parameters and can be obtained from IDT creep test. T herefore if the stress is known cycles needed for obtaining some percentage of DCSEf can be calculated by cycle DCSE DCSE cyclesf/ % The HMA material properties especially stiffness and ductility play an important role in CDL amplitude decision. If the mixture is very brittle at given testing temperature, th e strain at failure will be lower than less brittle materials although the 50% of DCSEf criter ion may has not been reached after a certain cycles of loading. In this case, observation of reduction in resilient modulus of mixture is very important. 10 20% Table 6 1 Failure criteria for fatigue test Mixture Stiffness FE and/or cr Reduction in MR Brittle Ductile Low 1020% Medium 50% High >50% reduction in modulus can be a substitute of 50% of DCSEf criterion. Reversely, if the resilient modulus of the mixture is very low, which means the mat erial is ductile, more than 50% reduction in modulus may be obtained when the 50% of DCSEf criterion is

PAGE 101

101 reached. A brief summary of the fatigue loading criteria to materials with different properties is listed in Table 61. Healing Test Program A program c hart for running healing test was summarized in Fig. 6 11. From this chart it can be seen that before running healing test, some basic information need to be obtained from IDT test, such as resilient modulus, tensile strength, DCSEf, etc. Of the IDT test r esults, the tensile strength is of importance because it would be referred as to decide the loading amplitude for fatigue test. Also the failure strain and fracture energy will be referred as well to estimate starting loading amplitude and cycles to end th e fatigue test. This test continues until it is manually stopped. This allows the user to test a specimen for a certain amount of time. On screen, a constantly updated graph of the deformations shows the user the current state of the gyratory specimen and a rough estimate of when macrocrack propagation begins. Data Acquisition The data acquisition is similar to that used for the MR test, but different in some specific program steps. The healing test requires achieving enough microdamage in the mixture in a reasonable period (ideally about 2 hours for both loading and unloading process. The data acquisition system was divided into two parts: damage phase and healing phase. There are some minor differences between data acquisition systems in these two phases For the damage phase, loading amplitude is around 1800 lbs and loading frequency is 2 Hz. In order to record readings from strain gages as much

PAGE 102

102 IDT Test (MR, Creep and St) % DCSEf Trial Loading Amplitude Trial Loading Rest Period (t seconds) Loading Magnitude (psi) 40~80% of StYes Start End Computed Loading Cycles (N) Is healing affacted? Delayed Elasticity? Healing? No Yes No Reduction in Resilient Modulus Figure 6 11 Healing test flow chart

PAGE 103

103 HRT (minutes) t (second) Load 0.1s 0.9s Ti=15 2 0 4 8 30 600 Figure 612 Data acquisition during healing phase -0.014 -0.012 -0.01 -0.008 -0.006 -0.004 -0.002 0 0 600 1200 1800 2400 3000 3600 4200 Time (second) Strain Gage Reading (in) Figure 6 1 3 Data collection illustration

PAGE 104

104 accurate as possible within the capability of computer ram, 512 sets of reading per loading cycle were saved. However, it would be redundant and unnecessary to record ever y single loading cycle. So the program was set to pick up to 5 cycles of data each time it was commanded to do so. These times were set as 0.1, 2, 4, 6, 8, 10, 15, 20, 25, 30, 40 minutes until the damage phase was manually terminated. If the required dam age was observed on the program during the damage phase loading process, the operator needs to press a button manually to record the last set of data and immediately after the program finish recording the loading was removed and healing phase started. Duri ng the healing process, t he user manually controlled the data collection at the interval times (0.1, 2, 4, 8, 15, 30 and 60 minutes) (Fig. 612). This denser data acquisition frequency at the beginning is because based on trial healing test results, up to 80% of the healing will happen at the first several minutes of loading removal. Every time the data collection button was pushed during the test, the program takes an MR reading over five cycles of loading at an arbitrary time. When the test was terminated, the testing and seating loads were removed from the test specimen. Fig. 613 shows a data acquisition sample from dense graded mixture healing test. Materials Prepared for Healing Test Georgia granite aggregates were prepared and batched. The batching fo rmula, binder content and other volumetric properties could be found in chapter 3. The mixtures were gyratory compacted to 7% air voids and then processed with STOA. After the pills cooled down, they were cut to IDT specimens with about 1.5 in thickness. A summary of number of samples and tests is shown in Table. 68. All healing test s will be done at 10 C in the MTS temperature controlled chamber after specimens were dehumidified up to 8 hours.

PAGE 105

105 Table 6 2 Materials for healin g test Mixture Type Aggregate Type Conditions Number of Binders Number of Pills No. of Tests/Specimen Superpave Dense Granite STOA 6 12 18 Healing Test Results Analysis Healing test results will be analyzed through damage phase and healing phase separat ely. For each phase, recorded data will be trimmed and interpreted via resilient modulus. After that, resilient modulus will be normalized and compared to evaluate dense graded granite mixtures damage and healing potentials. Dissipated creep strain energy will also be included in the final analysis. D amage Analysis during CDL t (seconds) Deformation T Maximum deformation point Unloading Loading Cycle starting point Cycle starting point (next cycle) Figure 614 Resilient deformation Fig. 6 14 shows an illustration of resilient deformation for one loading cycle. The resilient modulus calculation is based on the formula below:

PAGE 106

106 r r rM Roque and Buttlar (1994, 1997) developed IDT data analysis program, of which resilient modulus calculation program was applied here to compute resilient modulus of the material during both damage and healing phases. The final equation is described below and more details could be found from Roque (1997) Evaluation of SHRP indirect tension tester to mitigate cracking in asphalt concrete pavements and overlays (page 108-128): CPMLT AVG AVG i TTRIM AVG i RTC t D H P GL M * Where: i RTM : total resilient modulus of each cycle ( i = 1~3, the number of cycle) GL : gage length AVGP : average peak load of three replicate specimens i TTRIMH : total trimmed mean horizont al deformation array of each cycle (i = 1~3, the number of cycle) AVGD : average diameter of three replicate specimens AVGt : average thickness of three replicate specimens CPMLTC : nondimensional factor Roque (1994) And i M Mi RT RT

PAGE 107

107 Another important parameter for calculating resilient modulus is the poissons ratio: 2 2 2778 0 480 1 100 0 i TTRIM i TTRIM AVG AVG i TTRIM i TTRIM i TV H D t V H v Th e variables in this equation refer to the same meaning as those in resilient modulus equation. The poissons ratios calculated from CDL results and IDT results for all tested dense graded granite mixtures are summarized and plotted in Table 6-3 and Fig. 6-15 separately, which shows a good consistency between these two tests. Table 6 3 Poissons ratio comparison Mixtures DGUS DGMS DGRS DGAS DGBS DGCS CDL 0.35 0.30 0.33 0.25 0.32 0.42 IDT strength test 0.37 0.31 0.35 0.27 0.32 0.44 DGUS DGMS DGRS DGAS DGBS DGCS 0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Poisson' ratio from RDL Poisson' ratio from IDT strength test Figure 615 Poissons ratio comparison 10 C

PAGE 108

108 The determination of original resilient m odulus of HMA materials form healing test is different from that obtained through IDT resilient modulus test. As of IDT specimens, thanks to the existence of air voids, HMA materials thermal properties and cyclic loading effects, the material will first e xperience reversal of steric hardening and heat transformed from loading work, as shown in Fig. 616. This work induced a jump reduction in the resilient modulus as shown at the beginning of damage phase in Fig. 616. After that, the materials resilient m odulus exhibited a linear relationship with loading cycles. If this linear relationship was fitted by regression line, the interception of this line and the Y axis will be the original resilient modulus of the material, which is ER0 in Fig. 6 16. A summary of original resilient modulus of the materials from IDT and healing test is shown in Table 6 4 and Fig. 617. Although the healing test results underestimated the resilient modulus compared to IDT strength test results, it would not affect the damage and healing rates analysis in the following. Fig. 6 16 also shows the sketch how damage will affect the materials and how the materials recover from damage. During the damage phase, damage is presented by resilient modulus reduction and the rate of this reduct ion will be compared for mixtures with different binders. Typically the damage phase takes around 30 minutes to have a nearly 12% reduction in resilient modulus: % 1000 0 R Rd RE E E The damage healing phase will also be described by resilient modulus and the results will be shown in next subchapter. After that, DCSE will be analyzed associated with resilient modulus. Plotting of resilient modulus changes with loading time were summarized in appendix B. In order to have a direct comparison of the CDL damage effects on

PAGE 109

109 Time or loading cycles Resilient Modulus Heat and reversal of steric hardening ER0 ERdERh Damage recovery Damage phase Healing phase Linear regression Figure 61 6 Damage and healing interpretation with resilient modulus mixtures with different binders, the resilient modulus was normalized to original resi lient modulus and the results were plotted in Fig. 6-1 7 Regression was made on the linear parts of these normalized curves and the resultant fitting equations and R squarevalues were also shown in the figure. Damage rate was defined as percentage of original resilient modulus reduction per minute. The mean ing of this parameter to the surface mixtures in the field is related to how fast cracks can be initiated. D amage rate s in Fig. 6 -17 show that mixtures with different binders exhibit different responses to damages. Apparently mixtures with control binder PG 67 22 appears the worst case: material deteriorates at the fastest rate under CDL, which reflects that the surface dense graded granite mixtures designed with control binder will exhibit cracks at an earlier stage compared to the same mixtures designed w ith other binders under the same traffic load There are no apparent differences among mixtures with PG 76-22, hybrid binder B and C, which means some hybrid binders can perform as well as polymer modified binder PG 76 -22 with respect to resistance to RDL damage.

PAGE 110

110 DGMS: y = -0.0033x + 0.915, R2 = 0.9946 DGCS: y = -0.0028x + 0.8864, R2 = 0.9952 DGRS: y = -0.0017x + 0.871, R2 = 0.9873 DGBS: y = -0.0033x + 0.8419, R2 = 0.9889 DGAS: y = -0.005x + 0.8647, R2 = 0.9975 DGUS: y = -0.006x + 0.8229, R2 = 0.9981 0.0 0.4 0.8 1.2 0 5 10 15 20 25 30 35 Time (minute) Normalized Resilient Modulus Figure 61 7 DG mixtures normalized resilient modulus at damage phase 0.000 0.002 0.004 0.006 0.008 0.010 PG67-22 PG76-22, 4.25% HBA, 10% HBB, 6% HBC, 13% ARB-5, 5% Binder Types and Modifier Contents Damage Rate %MR/minute Figure 61 8 Relationship between modifier contents and damage rate

PAGE 111

111 Furthermore, damage rates are plotted with modifiers content in binders in order to check whether there is a relationship between these two parameters. As shown in Fig. 6 -19, modifiers (no matter it is CRM or SBS polymers) do decrease damage rate applied to dense graded granite mixtures (STOA). However, whether the CRM or SBS polymer was leading a major rol e in this case can not be told from this figure. Healing Analysis of DG Mixtures T he analysis of dense graded granite mixtures healing performance is similar to that of mixtures damage performance. One of the few differences is that the modulus in healing phase was normalized to the modulus at the beginning of healing, which is ERd shown in Fig. 616. ERh is the resilient modulus at the end of healing phase. 0.0 0.3 0.6 0.9 1.2 1.5 0 5 10 15 20 25 30 35 Relative Time (minute) Normalized Resilient Modulus Recovery HDGUS HDGMS HDGRS HDGAS HDGBS HDGCS DGAS: y = 0.0342Ln(x) + 1.1225 DGBS: y = 0.0313Ln(x) + 1.1102 DGCS: y = 0.0291Ln(x) + 1.0851 DGMS: y = 0.0264Ln(x) + 1.0709 DGRS: y = 0.0261Ln(x) + 1.0695 DGUS: y = 0.0247Ln(x) + 1.0652 80% recovery Figure 619 Combined regression for healing test Fig. 6 19 shows the trimmed normalized resil ient modulus recovery during the healing phase. From this figure it can be seen that there is no much difference with

PAGE 112

112 respect to healing rates for dense graded mixtures with different binders, no matter the binder was modified or not. An important point sh ould be addressed is that all mixtures recovered 80% of the resilient modulus in 10 minutes after the CDL was removed. This property is independent of damages accumulated from CDL. This figure appears that binders will not significantly affect mixtures hea ling performance. A further study was conducted on the relationship between healing rates and damage (DCSE and resilient modulus reduction separately) accumulated at the end of damage phase The DCSEapplied/DCSEf is the ratio of the dissipated creep strain energy to the total dissipated creep strain energy at failure during CDL. The calculation of DCSEapplied is shown in the following equation: cycles m D DCSEm AVE applied 1 1 21000 20 1 Where: AVE : applied stress 1D m: parameters f rom power model Parameters 1D m, and DCSEf can be obtained from IDT test. Fig. 6 20 shows that the amount of accumulated DCSE (at 20 minutes loading) in dense graded granite mixtures does not apparently affect healing rates. In Fig. 620, the DCSE accumulated in DGRS and DG US at the end of damage phase almost doubles that of mixtures with other binders, but the healing rates for DGRS and DGUS do not vary much accordingly. This implies that healing rates are independent of accumulated DCSE during damage phase. But if compared with the same accumulated DCSE,

PAGE 113

113 DGAS, followed by DGBS and DGCS, apparently recovers faster than mixture with PG 7622. We have to notice that although the healing rates are different for mixtures with different bind ers, more than 80% of damage will be healed in around 10 minutes, which seems to minimize the effects from healing rates. DGUS DGMS DGAS DGBS DGCS DGRS 0.00 0.01 0.02 0.03 0.04 0.05 0 5 10 15 20 % DCSEHMA at End of Damage Phase Healing Rate log(%MR)/minute Figure 6 2 0 Damage (DCSE) and h ealing r ate c omparison Fig. 6 21 shows the relationship between accumulated resilient modulus reduction (at 20 minutes loading) and healing rates for DG mixtures. The approximate trend curves seem to imply that DG mixtures with modified binders heal faster than control binder mixtures. However, the conclusion from this figure needs to be verified with more data at both lower or higher damage zones. The relationship between accumulated DCSE and reduction in resilient modulus is shown in Table 6 11 and plotted in Fig. 622. From Fig. 622, it can be seen that mixture

PAGE 114

114 DGUS DGMS DGAS DGBS DGCS DGRS 0.00 0.01 0.02 0.03 0.04 0.05 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% Reduced % MR at End of Damage Phase Healing Rate log(%MR)/minute Figure 621. Damage ( r esilient m od ulus r eduction) and h ealing r ate c omparison 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% Reduced % MR % DCSEf DGUS DGMS DGAS DGBS DGCS DGRS Figure 62 2 Reduced resilient modulus vs reduced DCSE 20 minutes loading

PAGE 115

115 points below the midline will exhibit more reduction in resilient modulus with less dissipated creep strain energy absorption compared to mixtures above the midline. However, except for DGBS, dense graded granite mixtures appear to have a liner relationship between DCSEapplied and reduction in resilient modulus. Table 6 4 Comparison of DCSE and MR reduction at 20 minutes CDL Mixutures % DCSE HMA Reduc tion % M RI DGUS 11 5 6 14 5 5 DGMS 3 8 6 7 5 9 DGAS 5 30 4 1 1 DGBS 4 1 7 11 78 DGCS 5 55 8 0 5 DGRS 10 4 8 6 4 5 Summary In general, HMA mixtures damage and healing performance can be evaluated through their resi lient modulus changes. Cyclic damage loading mode provides not only a convenient way to interpret damages through resilient modulus, but also allows delayed elasticity recovery, which removes difficult y of differentiating delayed elasticity and permanent d eformation. Furthermore, CDL makes it easier to control dissipated creep strain energy by monitoring loading cycles. Considered damage phase during healing test, the damage rate showed that dense graded mixtures with hybrid binders after STOA exhibited sim ilar behavior with SBS polymer modified binders (PG76 22). All mixtures with modified binders presented better performances compared to unmodified base binder (PG 6722). The modified binders apparently slower down damages to dense graded granite mixtures. If considered from healing phase during healing test, binder types turned out no effects on healing performances of dense graded granite mixtures. Most of all, 80% of

PAGE 116

116 damages will be recovered in about 10 minutes in this healing test, which minimized the importance of healing rates. In a word, modified binders, especially SBS modified and hybrid binders will definitely decrease damage rates and performs better than unmodified binders in dense graded granite mixtures. Healing rates could not differentiate binder components effects on healing potentials of dense graded granite mixtures after STOA.

PAGE 117

117 CHAPTER 7 CLOSURE AND RECOMMENDATIONS Summary Binder and mixture tests were performed to evaluate the relative performance of a PG 67 -22 base binder and six oth er binders produced by modifying the same base binder with the following modifiers: one SBS polymer, three commercially available hybrid binders composed of different percentages of rubber and SBS polymer, and two asphalt rubber binders (5% and 12 % rubber : ARB5 and ARB 12). The primary goal was to evaluate whether commercially available hybrid binder could exceed the performance characteristics of the base and asphalt rubber binders, as well as approach, meet or exceed the performance characteristics of t he SBS polymer modified binder. Secondary goals were to determine whether available binder tests and characterization methods are suitable for specifying hybrid binder. Key findings from the study are summarized below: Mixture tests indicated that cracking performance characteristics of densegraded mixtures (granite and limestone) produced with the commercially available hybrid binders used in this study exceeded the cracking performance characteristics of mixtures produced with the base binder and the ARB 5 binder, and were about the same as the cracking performance characteristics of the SBS polymer modified binder. Results of tests on open -graded friction course (OGFC) mixtures (granite and limestone) indicated that except for one special case (granite O GFC mixture with hybrid binder C ), the commercially available hybrid binders used in this study exhibited cracking performance characteristics that were about the same as those exhibited by mixtures produced with SBS polymer modified binder and ARB 12. It was concluded that hybrid binder C which included the finer grained rubber, may not have maintained appropriate consistency to achieve and maintain uniform distribution within the smoother textured and less absorptive granite OGFC during mixing and compac tion. The resulting nonuniformity is the most probable cause of the anomalous result (lower cracking performance characteristics). Addition of fibers or mixing and compaction at lower temperatures would likely have resulted in better distribution and crac king performance characteristics.

PAGE 118

118 The two hybrid binders produced with coarser grained rubber (hybrid binders A and B), as well as the two asphalt rubber binders (ARB -5 and ARB -12) did not meet FDOTs solubility specification, indicating that the rubber may not have been fully digested in the base binder. C onsequently, test results on these binders and parameters derived from the newly proposed MSCR test, were considered suspect, because the presence of particulates in the binder i s well known to affect DSR results. The binders produced with the coarser grained rubber met, and in most cases far exceeded requirements for PG7622 binder, resulting in binder performance parameters that indicated better performance characteristics than all other binders evaluated, including the SBS polymer modified binder. These results were not consistent with relative cracking performance characteristics determined from mixture tests. Hybrid binders A and B were also found to result in significantly lo wer absorption than all other binders in OGFC mixtures This indicated that the combination of coarser rubber particles and polymer affected absorption into the aggregate Differences in absorption were taken into account when determining the effective asp halt content, which was the same for all binder mixture combinations. Hybrid binder C which was produced with finer grained rubber, did meet FDOTs solubility specification, indicating that the rubber was fully digested in the base binder, thereby making it suitable for DSR testing. This binder also met all requirements for PG7622 binder with the exception of maximum phase angle (an additional FDOT requirement). None of the existing or currently proposed intermediate temperature binder tests, including DS -Ductility (FD) were found to provide parameters that consistently correlated with the relative cracking performance of mixtures. Parameters obtained from the new multiple stress creep recovery (MSCR) test and fr om Elastic Recovery (ER) distinguished the SBS polymer modified binder, but not hybrid binder C from the base binder. Therefore, it appears questionable whether either of these tests is suitable in their present form to specify hybrid binder. Only the elo ngation at failure from either the ER or FD tests was able to clearly distinguish the observed relative cracking performance of the SBS polymer modified and hybrid binders from that of the asphalt rubber binders. The asphalt rubber binders were more brittl e (less elongation to failure) than the SBS and hybrid binders. Consider ed damage phase during healing test, the damage rate showed that dense graded mixtures with hybrid binders after STOA exhibited similar behavior with SBS polymer modified binders (PG76 22). All mixtures with modified binders presented better performances compared to unmodified base binder (PG 67-22).

PAGE 119

119 The modified binders apparently slower down damages to dense graded granite mixtures. If considered from healing phase during healing tes t, binder types turned out no effects on healing performances of dense graded granite mixtures. Most of all, 80% of damages will be recovered in about 10 minutes in this healing test, which minimized the importance of healing rates. Conclusions The followi ng conclusions may be drawn on the basis of the research findings: Hybrid binders produced commercially, consisting of crumb rubber and SBS polymer (more rubber than SBS), can approach, meet or exceed the cracking performance characteristics of the SBS pol ymer modified binder. Although all the hybrid binders in this study did not meet all the Superpave binder tests, it appears that hybrid binder can be suitably specified using existing specification requirements for PG7622 binder and solubility (to disting uish it from asphalt rubber binder and to assure the validity of DSR test results). Hybrid binder specified in this manner has the potential to replace three binders currently used by FDOT in hot mix asphalt: SBS polymer modified asphalt, ARB 5, and ARB-12 This would result in the following benefits: Continued and probably increased use of tire rubber in asphalt. The ground tire rubber will not settle out like asphalt rubber binders. Eliminate a method recipe specification asphalt rubber for performance re lated hybrid binder. Simplify storage of binders at the hot mix plant by replacing three currently used asphalt binders. Improved cracking, and probably rutting, resistance of dense graded friction courses (FC9.5 and FC12.5)

PAGE 120

120 Recommendations As indicated above, hybrid binder specified in a proper manner, has the potential to replace three binders currently used by FDOT in hot mix asphalt: SBS polymer modified asphalt, ARB -5, and ARB 12. It also appears that a benefit may be derived by taking this course of action (i.e. eventually specifying hybrid binder exclusively for use in FDOT hot mix asphalt). Therefore, it is recommended that FDOT develop a transition plan to accomplish this. This should involve an assessment of impact and cost, development of a draft specification and strategy for implementation. C onsideration should be given to first allowing the use of hybrid binder as an alternate binder, then eventually requiring its use. Hybrid Binders have never been used on an actual project in Florida. The imp lementation process should include a number of demonstration projects where the hybrid binder is specifically specified in addition to the polymer modified binder for the project. The asphalt suppliers timeline to supply hybrid binder to Florida will have to be taken into account, and suppliers will need to know the level of Floridas commitment to this product before making the necessary investments. Finally, it is recommended that FDOT pursue development and evaluation of the new binder direct tension te st configuration for eventual use in performance based specification of hybrid binder, particularly since not even the newest MSCR test was successful in identifying its benefits Further healing test is also recommended to find aging effects on healing c haracteristics on both dense graded and open graded mixtures. Resilient modulus is a good parameter to interpret damage and healing performance of HMA mixtures, but

PAGE 121

121 further relationship of resilient modulus and dissipated creep strain energy needs to be ve rified and analyzed.

PAGE 122

122 APPENDIX A BINDER TEST RESULTS Table A -1 G*/sin and at 67 C (152.6 F) Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) PG 67 22 1.65 84.05 3.95 78.55 PG 76 22 n/a n/a n/a n/a HB A n/a n/a n/a n/a HB B n/a n/a n/a n/a HB C n/a n/a n/a n/a ARB 5 3.36 76.60 n/a n/ a ARB 12 5.98 75.40 n/a n/a Table A -2 and at 70 C (15 8 F) Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) PG 67 22 1.14 84.80 2.73 79.80 PG 76 22 n/a n/a n/a n/a HB A n/a n/a n/a n/a HB B n/a n/a n/a n/a HB C n/a n/a n/a n/a ARB 5 2.40 78.40 6.14 67.55 ARB 12 4.46 77.05 12.27 59.35 Table A -3 and at 76 C (1 68.8 F) Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) PG 67 22 0.59 86.60 1.39 82.30 PG 76 22 1.5 2 71.95 3.19 65.80 HB A 3.03 71.65 5.83 65.45 HB B 2.25 75.90 4.28 69.10 HB C 1.15 82.55 2.83 77.20 ARB 5 1.34 81.15 3.52 70.60 ARB 12 2.30 80.65 6.91 63.00 Table A -4 and at 82 C (1 79.6 F) Binders Original binders RTFOT r esidue G*/sin ( o ) (kPa) ( o ) PG 67 22 n/a n/a n/a n/a PG 76 22 0.91 74.25 1.88 68.15 HB A 1.70 74.95 3.34 68.60 HB B 1.26 79.25 2.44 72.40

PAGE 123

123 Table A -4. Continued Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) H B C 0.64 83.55 1.49 80.20 RB 12 1.27 82.90 4.10 66.40 Table A -5 and at 88 C (1 90.4 F) Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) PG 67 22 n/a n/a n/a n/a PG 76 22 n/a n/a n/a n/a HB A 1.03 77.30 1.99 70.90 HB B 0.77 81.60 1.39 76.10 HB C n/a n/a n/a n/a ARB 5 n/a n/a n/a n/a ARB 12 1.27 84.85 4.10 70.60 Table A -6 and at 90 C (1 94 F) Binders Original binders RTFOT r esidue ( o ) (kPa) ( o ) PG 67 22 n/a n/a n/ a n/a PG 76 22 n/a n/a n/a n/a HB A 0.86 78.20 n/a n/a HB B n/a n/a n/a n/a HB C n/a n/a n/a n/a ARB 5 n/a n/a n/a n/a ARB 12 n/a n/a n/a n/a Table A -7 and at 25 C (77 F) Binders 100 o C PAV r esidue 110 o C PAV r esidue G*sin (kPa) ( o ) G*sin (kPa) ( o ) PG 67 22 3255.5 49.8 4508.0 44.3 PG 76 22 3192.0 48.2 3633.0 44.0 HB A 2969.0 43.5 3626.5 38.9 HB B 2828.5 45.3 3372.0 40.8 HB C 3693.0 46.3 4692.5 42.1 ARB 5 2770.5 46.6 3750.0 41.8 ARB 12 2139.5 44.9 2604.5 40.5 Table A -8 G* and at 22 C (71.6 F) Binders 100 o C PAV r esidue 110 o C PAV residue G*sin (kPa) ( o ) G*sin (kPa) ( o )

PAGE 124

124 Table A -8 Continued PG 67 22 4901.5 46.9 6446.0 41.7 PG 76 22 4812.5 46.0 5238.0 41.8 HB A 4193.5 41.2 4976.5 36.8 HB B 4122.5 42.9 4749.0 38.8 HB C 5475.5 43.8 6655.5 39.7 ARB 5 4074.0 44.1 5226.5 39.7 ARB 12 3047.5 42.8 3566.5 38.7 Table A -9 and at 19 C (66.2 F) Binders 100 o C PAV r esidue 110 o C PAV residue G*sin (kPa) ( o ) G*sin (kPa) ( o ) PG 67 22 7053.0 44.2 n/a n/ a PG 76 22 6962.0 43.2 n/a n/a HB A 5921.0 38.9 6705.0 34.8 HB B 5877.0 40.7 6542.0 36.8 HB C n/a n/a n/a n/a ARB 5 5946.0 41.6 n/a n/a ARB 12 4246.5 40.6 4868.0 37.0 Table A -10 and at 16 C (60.8 F) Binders 100 o C PAV r esidue 110 o C PAV r esidue G*sin (kPa) ( o ) G*sin (kPa) ( o ) PG 67 22 n/a n/a n/a n/a PG 76 22 n/a n/a n/a n/a HB A n/a n/a n/a n/a HB B n/a n/a n/a n/a HB C n/a n/a n/a n/a ARB 5 n/a n/a n/a n/a ARB 12 5867.5 35.1 6459.5 34.9 Table A -11 BBR test results at 12 C (10.4 F) Binders 100 o C PAV residue 110 o C PAV residue creep stiffness S (M Pa) m v alue creep stiffness S (M P a) m v alue PG 67 22 159.5 0.365 182.5 0.339 PG 76 22 144.0 0.362 170.0 0.334 HB A 137.5 0.322 154.5 0.301 HB B 147.0 0.336 155.5 0.318 HB C 166.5 0.337 185.0 0.315 ARB 5 138.0 0.345 155.5 0.318 ARB 12 109.0 0.337 127.5 0.316

PAGE 125

125 Table A -12 BBR test results at 18 C (0.4 F) Binders 100 o C PAV residue 110 o C PAV residue creep stiffness S (M Pa) m v alue creep stiffness S (M Pa) m v alue P G 67 22 341.5 0.291 400.5 0.276 PG 76 22 331.0 0.295 356.5 0.279 HB A 298.0 0.262 313.5 0.252 HB B 303.0 0.279 303.5 0.269 HB C 358.5 0.274 373.5 0.265 ARB 5 281.0 0.287 302.0 0.270 ARB 12 231.0 0.288 241.5 0.274 Table A -13 Multiple stress creep recovery %, RTFOT r esidue Binders 67 C (152.6 F) 76 C (168.8 F) 3.2 kPa 0.1 kPa Rdiff 3.2 kPa 0.1 kPa Rdiff PG 67 22 3.73 13.27 71.88 0.68 6.16 88.93 PG 76 22 64.25 71.79 10.50 31.87 54.24 41.25 HB A 51.11 67.38 24.14 23.08 53.05 56.46 HB B 40.52 54.15 25.15 16.85 38.75 56.58 HB C 13.13 27.23 51.71 3.05 13.84 78.01 ARB 5 25.03 46.02 45.61 6.81 32.27 78.86 ARB 12 56.64 74.97 24.52 20.30 58.37 65.21 Table A-14 N o n -recoverable creep compliance, k Pa1, RTFOT r esidue 67 C (152.6 F) 76 C (168.8 F) Binders J nr 3.2 J nr 0.1 Diff. % J nr 3.2 J nr 0.1 Diff. % PG 67 22 2.06 1.66 24.51 7.05 5.65 24.84 PG 76 22 0.24 0.19 29.30 1.34 0.81 65.54 HB A 0.21 0.13 63.20 1.02 0.51 103.42 HB B 0.34 0.25 36.17 1.51 0.92 63.76 HB C 0.78 0.61 28.85 3.02 2.25 34.46 ARB 5 0.58 0.38 0.53 2.42 1.35 0.7919 ARB 12 0.15 0.08 0.8 7 0.87 0.36 1.4319 Table A-17 Elastic r ecovery at 25 C (77 F ) (RTFOT r esidue) Binders Replicate A (%) Replicate B (%) Average (%) PG 67 22 7.41 4.94 6.18 PG 76 22 75.00 75.00 75.00 HB A 66 .25 67.50 66.88 HB B 72.50 72.50 72.50 HB C 23.75 25.00 24.38

PAGE 126

126 Table A -18 Force d uctility t est r esult Binders f2/f1 (10 C ) f2/f1 (RTFOT residue, 10 C ) f2/f1 (PAV residue, 25 C ) PG 67 22 0.04 0.04 0.03 PG 76 22 0.53 0.43 0.26 HB A 0.46 0.36 0.40 HB B 0.42 0.40 0.40 HB C 0.17 0.20 0.13 ARB 5 0.20 0.32 0.24 ARB 12 0.24 0.51 0.18 Table A -19 S moke p oint, flash point and solubility of o riginal b inders Binders Smoke Point (F) Flash Point (F) Solubility (%) PG 67 22 322.5 545.0 99.995 PG 76 22 330.0 552.5 99.975 HB A 325.0 557.5 92.760 HB B 320.0 550.0 96.905 HB C 320.0 495.0 99.860 ARB 5 315.0 545.0 93.835 ARB 12 320.0 547.5 88.765 Table A -20 M ass l oss after RTFOT at 163 C (325.4 F) Binders Replicate A (%) Replicate B (%) Average (%) PG 67 22 0.423 0.412 0.418 PG 76 22 0.370 0.369 0.370 HB A 0.341 0.340 0.341 HB B 0.359 0.319 0.339 HB C 0.525 0.522 0.524 ARB 5 0.429 0.433 0.431 ARB 12 0.463 0.472 0.468

PAGE 127

127 0 50 100 150 200 250 300 350 0.0 0.5 1.0 1.5 2.0 2.5 Strain Stress (psi) 67-22 76-22 Hybrid_Binder_A Hybrid_Binder_B Hybrid_Binder_C ARB-5 ARB-12 Figure A 1 Original binders stress-strain diagram 10 C (50 F) 0 50 100 150 200 250 300 350 400 450 0.0 0.5 1.0 1.5 2.0 2.5 Strain Stress (psi) 67-22 76-22 Hybrid_Binder_A Hybrid_Binder_B Hybrid_Binder_C ARB-5 ARB-12 Figure A 2 RTFOT residues stress-strain diagram 10 C (50 F)

PAGE 128

128 0 50 100 150 0.0 0.5 1.0 1.5 2.0 2.5 Strain Stress (psi) 67-22 76-22 Hybrid_Binder_A Hybrid_Binder_B Hybrid_Binder_C ARB-5 ARB-12 Figure A 3 PAV residues stress -strain diagram 25 C (77 F)

PAGE 129

129 APPENDIX B HEALING TEST RESULTS Damge and healing 0.0 0.4 0.8 1.2 1.6 0 10 20 30 40 50 60 70 Time (minute) Resilient Modulus (Mpsi) 05HDGUSB 05HDGUSE 05HDGUSH Figure B 1 MR DGUS at damage and healing, 10 C Damage and healing 0.0 0.4 0.8 1.2 1.6 0 10 20 30 40 50 60 70 80 Time (minute) Resilient Modulus (Mpsi) 05HDGMSA 05HDGMSB 05HDGMSC Figure B 2 MR DGMS at damage and healing, 10 C

PAGE 130

130 Damage and healing 0.0 0.4 0.8 1.2 1.6 0 20 40 60 80 100 Time (minute) Resilient Modulus (Mpsi) 05HDGRSB 05HDGRSC 05HDGRSD Figure B 3 MR DGRS at damage and healing, 10 C Damage and healing 0.0 0.5 1.0 1.5 2.0 0 10 20 30 40 50 60 70 80 Time (minute) Resilient Modulus (Mpsi) 05HDGASA 05HDGASB 05HDGASC Figure B 4 MR DGAS at damage and healing, 10 C

PAGE 131

131 Damage and healing 0.0 0.5 1.0 1.5 2.0 0 20 40 60 80 100 Time (minute) Resilient Modulus (Mpsi) 05HDGBSA 05HDGBSB 05HDGBSD Figure B 5 MRDG B S at damage and healing, 10 C Damage and healing 0.0 0.5 1.0 1.5 2.0 0 10 20 30 40 50 60 70 80 Time (minute) Resilient Modulus (Mpsi) 05HDGCSA 05HDGCSB 05HDGCSC Figure B 6 MRDG C S at damage and healing, 10 C

PAGE 132

132 APPENDIX C CITGO C ERTIFICATES OF ANALYSIS

PAGE 133

133

PAGE 134

134 LIST OF REFERENCES Abdelrahman, M. (2006). Controlling performance of crumb rubber -modified binders through addition of polymer modifiers. Transportation research recor d Washington, D C. 6470. Bouldin, M.G. and C ollins, J.H. (1992). Influence of b inder r heology on rut resistance of p olymer m odified and u nmodified h ot m ix a sphalt, Polymer Modified Asphalt binders ASTM STP 1108, 5060. Buttlar, W. G. and Roque, R. (1 994 ). Experimental d evelopment and e valuation of the n ew SHRP m easurement and a nalysis s ystem for i ndirect t ensile t esting of a sphalt m ixtures at l ow temperatures. Association of Asphalt Paving Technologists 1994. Choubane, B Sholar, G A ; Musselman, J A ; and Page, G C (1999). Ten y ear p erformance e valuation of a sphalt rubber s urface m ixes, Transportation Research R ecord (TRB), No.1681, 1018. Cook, M C ., Bressette, T Holikatti, S Zhou, H and Hicks, R. G. (2006). Laboratory e valuation of a sphalt rubber m odified m ixes, Proceedings of the Asphalt Rubber 2006 Conference, Palm Springs, USA 599618. Cui, Z. (2003). Use of binder rheology to predict the cracking performance of SBS -modified mixture, Doctoral thesis University of Florida, Florida D'Angelo, J., Dongre, R. N. (2009). Practical Use of Multiple Stress Creep Recovery Test: Characterization of Styrene-ButadieneStyrene Dispersion and Other Additives in PMA Binders Transportation Research Board 88th Annual Meeting, Washington DC, USA. Hicks, R.G., Lundy, J.R., Leahy, R.B., Hanson, D., and Epps, J. (1995). Crumb rubber modifiers (CRM) in asphalt pavements: Summary of practices in Arizona, C alifornia, and Florida, Transportation Research Institute, Oregon State Un iversity, Report No. FHWA SA 95056. Fleckenstein, L.J ., Mahboub, K., and Allen, D.L. (1992). Performance of p olymer m odified a sphalt m ixes in k entucky, Polymer Modified Asphalt Binders, ASTM STP 1108, American Society for Testing and Materials Philadel phia USA Kim, B Roque, R and Birgisson, B. (2003) Laboratory evaluation of the effect of modifier on cracking resistance of asphalt mixture, Annual Meeting of the Transportation Research Board, Washington D.C. Moseley, H L. P age G., Musselman J. A. Scholar G. A. Upshaw, P B. (2003). Laboratory m ixture and b inder rutting s tudy, Research Report FL/DOT/SMO/03 465.

PAGE 135

135 Page, G C; Ruth, B E ; West, R C. (1992) Floridas a pproach u sing g round t ire rubber in a sphalt c oncrete mixtures. Transportation Research R ecord (TRB), No.1339, 1622 Page, G C. (1992). Florida's i nitial e xperience u tilizing g round t ire rubber in a sph alt c oncrete m ixes, Association of Asphalt Paving Technologists (AAPT), Vol 61, 446. Roberts, F L., Kandhal, P S., Brown, E. R. Lee, D and Kennedy, T W. (1996). Hot m ix a sphalt m aterials, m ixture d esign and c onstruction National Asphalt Pavement Association Research and Education Foundation, Lanham, Maryland, Second edition. Romagosa, H Corun, R and Berkley, R .(2008). SBS p olymer s upply o utlook, Association of Modified Asphalt Producers (AMAP)s Updated White Paper on the SBS Supply Outlook St. Louis, MO. Rogge, D.F., Terrel, R.L. and George A.J. (1992). Polymer m odified h ot m ix a sphalt Oregon e xperience Polymer Modified Asphalt Binder ASTM STP 1108, Kenneth R. Testing and Materials Philadelphia Roque, R., Birgisson, B., Drakos, C. and Dietrich, B. (2004). Development and f ield e valuation of e nergy -b ased c riteria for t opdown c racking p erformance of Hot Mix Asphalt, Journal of the Association of Asphalt Paving Technologists Vol. 73, 229260. Roque, R., and Buttlar, W. G.(1992). The d e velopment of a m easurement and a nalysis s ystem to a ccurately d etermine a sphalt c oncrete p roperties u sing the i ndirect t ensile test. Association of Asphalt Paving Technologists (AAPT), Vol. 73, 395. Shuler, T. S., Collins, J. H., and Kirkpatrick, J. P., P olymer modified asphalt properties related to asphalt concrete performance. Asphalt Rheology: Relationship to Mixture, ASTM STP 941, O. E. Briscoe, American Society for Testing and Materials, Philadelphia, 1987, pp. 179193. Sousa, J., Way, G. B. and Zare h, A. (2006). Asphalt rubber g ap g raded m ix d esign c oncepts Proceedings of the Asphalt Rubber 2006 Conference, Palm Springs, USA 523543. The Balmoral Group. (2008). 2008 Strategic resource e valuation u pdate: h ighway c onstruction materials. the Balmo ral Group, Maitland, FL. Tia, M ; Roque, R ; Sirin, O and Kim, H J (2002). Evaluation of s uperpave m ixtures with and without p olymer modification by m eans of Accelerated Pavement Testing, Report to FDOT UF PN 4910450480112. Xiao, F Putman, B J. and Amirkhaniam, S N.(2006). Laboratory i nvestigation of d imensional c hanges of c rumb rubber reacting with a sphalt b inder, Proceedings Asphalt Rubber 2006 Conference, Palm Springs, USA, 693713.

PAGE 136

136 BIOGRAPHICAL SKETCH Weitao Li wa s born in a small town, Julu, which lies in south part of Hebei p rovince in north C hina. He grew up there and finished all his pr eliminary school, middle school and high school years in that town. In 1997, he was admitted by Jilin University C hangchun and awarded Bachelor Degree of Drilling and Exploration Engineering in the year of 2001. Thanks to his exceptional overall performa nce in school, he was titled Outstanding Graduate by the university on the Graduation C eremony and in the same year, he was recommended to pursue the m aster degree of g eological e ngineering in the same college with waiver of entrance exams. Three years lat er, he finished his graduate study and was awarded the d egree Master of Geological Engineering in 2004. In August of the year 2005, he came to the U.S. to pursue a Ph.D. degree in the Department of C ivil and C oastal Engineering at UF under the supervision of professors Dr. Reynaldo Roque and Dr. Bjorn Birgisson. Except for the academia, he also actively took part in volunteering activities in the University of Florida to help other students. He worked as the President of Friendship Association of C hinese S tudents and Scholars at the University of Florida during the school year of 2008 to 2009. After completing his Ph.D., he plans to work in industrial companies in C ivil Engineering to continue his dedication to this field.