<%BANNER%>

Development of a Binder Fracture Test to Determine Fracture Energy

Permanent Link: http://ufdc.ufl.edu/UFE0043732/00001

Material Information

Title: Development of a Binder Fracture Test to Determine Fracture Energy
Physical Description: 1 online resource (178 p.)
Language: english
Creator: Niu, Tianying
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: asphalt -- fracture
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: It has been found that all existing binder testing methods in current specifications do not accurately predict cracking performance at intermediate temperatures. Fracture energy has been determined to be strongly correlated to fracture resistance of asphalt mixture, so a test to measure fracture energy of binder is expected to provide an excellent tool to evaluate fracture resistance of binder. A new fracture test and interpretation system was successfully developed based on Finite Element Analysis (FEA) and prototype test on MTS machine to consistently measure fracture energy of binder at intermediate temperature. For evaluation, the new fracture test and interpretation system was applied to a range of binders including unmodified binder, SBS modified binder, rubber modified binder, hybrid binder and highly SBS modified binder from PAV residue or recovered from field test sections. Statistical analysis was conducted on test results, which showed that the new fracture test and interpretation system significantly distinguished between different binders by fracture energy. Expected trends in fracture energy between binders were observed, which indicates the test was successful. It was also shown that for the same binder, fracture energy is independent of loading rate evaluated in this study and test temperature from 0 to 15 °C. Thus, fracture energy appears to be a fundamental property of binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate. The results also showed that different types of binder have different characteristic true stress-true strain curves, which can be used to identify the binder type, modifier type and relative content. Basic principles were proposed to identify the presence of modifier from true stress-true strain curves. A detailed testing protocol was recommended. The protocol helps assure the appropriate loading rate range so that the complete true stress vs. true strain curve can be identified for accurate determination of fracture energy. In conclusion, the new fracture test and interpretation system appears to be suitable to measure fracture energy of the broad range of binder used in asphalt paving.
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 Tianying Niu.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Roque, Reynaldo.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043732:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043732/00001

Material Information

Title: Development of a Binder Fracture Test to Determine Fracture Energy
Physical Description: 1 online resource (178 p.)
Language: english
Creator: Niu, Tianying
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: asphalt -- fracture
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: It has been found that all existing binder testing methods in current specifications do not accurately predict cracking performance at intermediate temperatures. Fracture energy has been determined to be strongly correlated to fracture resistance of asphalt mixture, so a test to measure fracture energy of binder is expected to provide an excellent tool to evaluate fracture resistance of binder. A new fracture test and interpretation system was successfully developed based on Finite Element Analysis (FEA) and prototype test on MTS machine to consistently measure fracture energy of binder at intermediate temperature. For evaluation, the new fracture test and interpretation system was applied to a range of binders including unmodified binder, SBS modified binder, rubber modified binder, hybrid binder and highly SBS modified binder from PAV residue or recovered from field test sections. Statistical analysis was conducted on test results, which showed that the new fracture test and interpretation system significantly distinguished between different binders by fracture energy. Expected trends in fracture energy between binders were observed, which indicates the test was successful. It was also shown that for the same binder, fracture energy is independent of loading rate evaluated in this study and test temperature from 0 to 15 °C. Thus, fracture energy appears to be a fundamental property of binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate. The results also showed that different types of binder have different characteristic true stress-true strain curves, which can be used to identify the binder type, modifier type and relative content. Basic principles were proposed to identify the presence of modifier from true stress-true strain curves. A detailed testing protocol was recommended. The protocol helps assure the appropriate loading rate range so that the complete true stress vs. true strain curve can be identified for accurate determination of fracture energy. In conclusion, the new fracture test and interpretation system appears to be suitable to measure fracture energy of the broad range of binder used in asphalt paving.
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 Tianying Niu.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Roque, Reynaldo.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043732:00001


This item has the following downloads:


Full Text

PAGE 1

1 DEVELOPMENT OF A BINDER FRACTURE TEST TO DETERMINE FRACTURE ENERGY By TIANYING NIU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 T ianying N iu

PAGE 3

3 To my p arents

PAGE 4

4 ACKNOWLEDGMENTS I thank my parents and my wife for all the support s they gave to me these years. I thank my advisor, Dr.Reynaldo Roque, for all the valuable instructions he gave to me during the research projects. I also thank all the people who helped me with my research work

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 11 LIST OF ABBREVIATIONS ................................ ................................ ........................... 18 ABSTRACT ................................ ................................ ................................ ................... 19 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 21 1.1 Background ................................ ................................ ................................ ...... 21 1.2 Objectives ................................ ................................ ................................ ........ 22 1.3 Hypothesis ................................ ................................ ................................ ....... 23 1.4 Research Approach ................................ ................................ ......................... 23 2 LITERATURE REVIEW ................................ ................................ .......................... 24 3 MATERIALS AND METHODS ................................ ................................ ................ 27 3 .1 Components of Binders ................................ ................................ ................... 27 3 .1 Preliminary Tests ................................ ................................ ............................. 28 3 .1.1 Binder Types ................................ ................................ .......................... 28 3 .1. 2 Testing Method ................................ ................................ ....................... 28 3 .2 Tests on Binders Recovered from Superpave Sections ................................ ... 28 3 .2.1 Binder Types ................................ ................................ .......................... 29 3.2.2 Testing Metho d ................................ ................................ ....................... 29 3.3 Tests on Hybrid Binders and Highly Polymer Modified Binder ......................... 29 3.3.1 Binder Types ................................ ................................ .......................... 30 3.3.2 Testing Method ................................ ................................ ....................... 30 4 DEVELOPMENT OF A NEW BINDER FRACTURE TEST ................................ ..... 32 4.1 Development of Speci men Geometry by Finite Element Analysis (FEA) ......... 32 4.1.1 Reasons Why to Develop a New Specimen Geometry .......................... 32 4.1.2 Development and Optimization Process ................................ ................. 33 4.1.2.1 Geometry No.1 ................................ ................................ .............. 33 4.1.2.2 Geometry No.2 ................................ ................................ .............. 35 4.1.2.3 Geometry No.3 ................................ ................................ .............. 36 4.1.2. 4 Geometry No. 4 ................................ ................................ .............. 38 4. 2 Data Interpretation ................................ ................................ ........................... 41

PAGE 6

6 4. 2 .1 Premature Failure Identification ................................ .............................. 41 4.2.2 New Calculation Procedure for Fracture Energy ................................ .... 42 4.2.3 Finite Element Analysis (FEA) Results ................................ ................... 44 5 TESTS AND ANALYSES ON A RANGE OF BINDERS ................................ .......... 46 5 1 Preliminary Tests ................................ ................................ ............................. 46 5.1.1 Testing Method ................................ ................................ ....................... 46 5.1.2 Test Result and Analysis ................................ ................................ ........ 46 5.1.2.1 15 C ................................ ................................ ............................. 46 5.1.2.2 10 C ................................ ................................ ............................. 47 5.1.2.3 5 C ................................ ................................ ............................... 48 5.1.2.4 0 C ................................ ................................ ............................... 49 5 .2 Tests on Binders Recovered from Superpave Sections ................................ ... 52 5.2.1 Testing Method ................................ ................................ ....................... 53 5.2.2 Test Result and Analysis ................................ ................................ ........ 54 5.2.2.1 PG 76 22 recovered from field ................................ ..................... 54 5.2.2.2 AC 20 recovered from field ................................ ........................... 56 5.2.2.3 AC 30 recovered from field ................................ ........................... 59 5.2.2.4 PG 64 22 recovered from field ................................ ...................... 60 5 .3 Tests on Hybrid Binders and Highly Polymer Modified Binder ......................... 65 5.3.1 Testing Method ................................ ................................ ....................... 66 5.3.2 Test Result and Analysis ................................ ................................ ........ 67 5.3.2.1 Hybrid binders ................................ ................................ ............... 67 5.3.2.2 Marianni ................................ ................................ ........................ 71 5.3.2.3 H ighly polymer modified binder PG 82 22 ................................ ..... 73 5 .4 Summary ................................ ................................ ................................ ......... 77 6 STATISTICAL ANALYSIS ................................ ................................ ....................... 79 6 1 Consistency of Fracture Energy ................................ ................................ ....... 79 6.1.1 Two way A nalysis of Variance ................................ ................................ 79 6.1.1.1 PG 67 22 P AV residue ................................ ................................ .. 79 6.1.1.2 PG 82 22 PAV residue ................................ ................................ .. 80 6.1.2 One way A nalysis of Variance ................................ ................................ 82 6.1.2.1 AC 20 recovered from field ................................ ........................... 82 6.1.2.2 AC 3 0 recovered from field ................................ ........................... 83 6.1.2.3 PG 64 22 recovered from field ................................ ...................... 84 6.1.2.4 ARB 5 recovered from field ................................ ........................... 85 6.1.2.5 PG 76 22 recovered from field ................................ ...................... 85 6.1.2.6 PG 76 22 at 15 C (both recovered and PAV residue) .................. 86 6 .2 Significance of Difference of Fracture Energy between Binders ...................... 87 6. 2 1 Three way A nalysis of Variance ................................ ............................. 88 6.2.1.1 PG 67 22 and PG 76 22 ................................ ............................... 88 6. 2 2 Two way A n alysis of Variance ................................ ................................ 89 6.2.2.1 Modified binders ................................ ................................ ............ 89 6.2.2.2 Hybrid binders and PG 76 22 ................................ ........................ 91

PAGE 7

7 6.2.2.3 Unmodified binders ................................ ................................ ....... 93 6 .3 Summary ................................ ................................ ................................ ......... 95 7 CHARACTERISTIC TRUE STRESS TRUE STRAIN CURVES ............................. 96 7 1 Typical True Stress True Strain Curve of Each Type of Binder ....................... 96 7.1.1 Unmodified Binders ................................ ................................ ................ 96 7.1.1.1 PG 67 22 PAV residue ................................ ................................ .. 97 7.1.1.2 AC 30 recovered from field ................................ ........................... 97 7.1.1.3 AC 20 recovered from field ................................ ........................... 98 7.1.1.4 PG 64 22 recovered from field ................................ ...................... 99 7.1.1.5 Comparison between unmodified binders ................................ ... 100 7.1.2 SBS Polymer modified binders ................................ ............................. 101 7.1.2.1 PG 76 22 recovered from field ................................ .................... 101 7.1.2 .2 PG 76 22 PAV residue ................................ ................................ 104 7.1.2.3 PG 82 22 PAV residue ................................ .............................. 105 7.1.2.4 Comparison between polymer modified binders ......................... 106 7.1.3 Rubber modified binders ................................ ................................ ...... 106 7.1.3.1 ARB 5 and ARB 12 PAV residue ................................ ................ 107 7.1.3.2 ARB 5 recovered from field ................................ ......................... 108 7.1.3.3 Marianni PAV residue ................................ ................................ 109 7.1.3.4 Comparison between rubber modified binder s ............................ 110 7.1.4 Hybrid binders ................................ ................................ ...................... 110 7.1.4.1 Wright PAV residue ................................ ................................ ..... 111 7.1.4.2 Hudson PAV residue ................................ ................................ ... 111 7.1.4.3 Geotech PAV residue ................................ ................................ .. 112 7.1.4.4 Comparison between hybrid binders ................................ ........... 113 7 .2 Comparison of True Stress True Strain Curve between Binders ................... 114 7 .3 Summary of Characteristic True Stress True Strain Curves .......................... 118 7.4 Fracture Energy Determination ................................ ................................ ...... 119 8 RECOMMENDED TESTING PROTOCOL ................................ ........................... 123 8. 1 Preparation ................................ ................................ ................................ .... 123 8. 2 Testing and Analysis ................................ ................................ ...................... 123 8.3 Simplified Testing Protocol ................................ ................................ ............ 124 9 CONCLUSIONS ................................ ................................ ................................ ... 126 9. 1 Summary ................................ ................................ ................................ ....... 126 9.2 Conclusions ................................ ................................ ................................ ... 129 APPENDIX A FINITE ELEMENT ANALYSIS (FEA) RESULTS ................................ .................. 131 B PRELIMINARY TEST RESULTS ................................ ................................ .......... 133

PAGE 8

8 C TEST RESULTS OF BINDERS RECOVERED FROM SUPERPAVE SECTIONS ................................ ................................ ................................ ........... 140 D TEST RESULTS OF HYBRID BINDERS AND PG 82 22 ................................ ..... 151 E TEST RESULTS SUMMARY ................................ ................................ ................ 156 F STATISTICAL ANALYSIS RESULTS ................................ ................................ ... 157 G CHARACTERISTIC TRUE STRESS TRUE STRAIN CURVES ........................... 162 LIST OF REFERENCES ................................ ................................ ............................. 177 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 178

PAGE 9

9 LIST OF TABLES Table page 3 1 Binders and the constituents/formulations ................................ .......................... 27 5 1 A part of test results of binders recovered from Superpave sections (box 1) ..... 55 5 2 A part of test results of binders recovered from Superpave sections (box 2) ..... 55 5 3 Test results of AC 20 ................................ ................................ .......................... 56 5 3 (Continue d) ................................ ................................ ................................ ......... 57 5 4 PG 82 22 fracture energy at 15 o C ................................ ................................ ..... 75 5 5 PG 82 22 fracture energy at 10 o C ................................ ................................ ..... 75 6 1 Key statistical analysis results of PG 67 22 ................................ ........................ 80 6 2 Key statistical analysis results of PG 82 22 ................................ ........................ 81 6 3 Key statistical analysis result of AC 20 ................................ ............................... 82 6 4 Key statistical analysis result of AC 30 ................................ ............................... 83 6 5 Key statistical analysis res ult of PG 64 22 ................................ .......................... 84 6 6 Key statistical analysis result of ARB 5 recovered from field .............................. 85 6 7 Key statistical analysis result of P G 76 22 recovered from field ......................... 86 6 8 Key statistical analysis result of all PG 76 22 at 15 C ................................ ....... 8 7 6 9 Key statistical analysis res ults of PG 67 22 and PG 76 22 ................................ 88 6 10 Key statistical analysis results of all modified binders except PG 82 22 ............. 90 6 11 Key stati stical analysis results of hybrid binders and PG 76 22 ......................... 92 6 12 Key statistical analysis results of contrast between Hudson and PG 76 22 ....... 93 6 13 Key statistical analysis results of unmodified binders ................................ ......... 94 6 14 Key statistical analysis results of contrast between unmodified binders ............. 95 B 1 Fracture energy density at 15 C (PAV residue) ................................ ............... 133 B 2 Fracture energy density at 10 C (PAV residue) ................................ ............... 134

PAGE 10

10 B 3 Fracture energy density at 5 C (PAV residue) ................................ ................. 135 B 4 Fracture energy density at 0 C (PAV residue) ................................ ................. 136 C 1 Test res ults of binders recovered from Superpave sections (box 1 box 3) ....... 140 C 2 Test results of binders recovered from Superpave sections (box 4 box 6) ....... 141 C 3 Test results of binders recovered from Superpave sections (box 7 box 9) ....... 142 C 4 Test results of binders recovered from Superpave sections (box 10 box 12) ... 143 C 5 Test results of binders recovered from Superpave sections (box 13 box 15) ... 144 C 6 Test results of binders recovered from Superpave sec tions (box 16 box 18) ... 145 C 7 Test results of binders recovered from Superpave sections (new boxes) ........ 146 D 1 PG 82 22 fracture en ergy at 15 C ................................ ................................ ... 152 D 2 PG 82 22 fracture energy at 10 C ................................ ................................ ... 152 F 1 Key statistical analysis result of PG 67 22 ................................ ........................ 157 F 2 Key statistical analysis results of PG 82 22 ................................ ...................... 157 F 3 Key statistical analysis result of AC 20 ................................ ............................. 157 F 4 Key statistical analysis result of AC 30 ................................ ............................. 158 F 5 Key statistical analysis result of PG 64 22 ................................ ........................ 158 F 6 K ey statistical analysis result of ARB 5 recovered from field ............................ 158 F 7 Key statistical analysis result of PG 76 22 recovered from field ....................... 159 F 8 Key statistical analysis result of all PG 76 22 at 15 C ................................ ..... 159 F 9 Key statistical analysis results of PG 67 22 and PG 76 22 ............................... 159 F 10 Key statistical analysis results of all modified binders except PG 82 22 ........... 160 F 11 Key statistical analysis results of hybrid binders and PG 76 22 ....................... 160 F 12 Key statistical analysis results of contrast between Hudson and PG 76 22 ..... 160 F 13 Key statistical analysis results of unmodified binders ................................ ....... 160 F 14 Key statistical analysis results of contrast between unmodified binders ........... 161

PAGE 11

11 LIST OF FIGURES Figure page 1 1 Traditional Superpave DT test ................................ ................................ ............ 21 4 1 2 D FEA model ................................ ................................ ................................ ... 33 4 2 3 D FEA model ................................ ................................ ................................ ... 33 4 3 Geometry N o.1, true stress vs. true strain ................................ .......................... 34 4 4 Geometry No.1, testing on MTS ................................ ................................ ......... 35 4 5 Geometry No.2, FEA model ................................ ................................ ............... 36 4 6 Geometry No.3, FEA model ................................ ................................ ............... 36 4 7 Geometry No.3, asphalt peeled off from load head ................................ ........... 37 4 8 Geom etry No.3, PG 76 22, 5 C, 300 mm/min & 150 mm/min ........................... 38 4 9 Geometry No. 4 ................................ ................................ ................................ ... 38 4 10 Dimensions of Geometry No. 4 ................................ ................................ ............ 39 4 11 Stress distribution on h orizontal c ross s ection by FEA ................................ ....... 39 4 1 2 Preparation of specimen ................................ ................................ ..................... 40 4 13 New DT t esting e quipment on MTS ................................ ................................ .... 40 4 14 Fracture of s pecimens ................................ ................................ ........................ 41 4 15 Identification of premature fracture ................................ ................................ ..... 42 4 16 True stress true strain curve by new calculation procedure ............................... 43 4 17 For t ransforming e xtension to t rue s train ................................ ............................ 44 4 18 For t ransforming l oad to t rue s tress ................................ ................................ .... 44 4 19 For c alculating a rea of c ross s ection at the first stress p eak .............................. 45 4 20 For c alculating l ength of 3mm middle p art at the first stress p eak ...................... 45 5 1 Fracture e nergy d ensity at 15 C ................................ ................................ ........ 47 5 2 Fracture e n ergy d ensity at 1 0 C ................................ ................................ ........ 47

PAGE 12

12 5 3 Fracture e nergy d ensity at 5 C ................................ ................................ .......... 48 5 4 Fracture e nergy d ensity at 5 C (without l oading ................. 48 5 5 Fracture e nergy d ensity at 0 C ................................ ................................ .......... 49 5 6 Fracture e nergy d ensity at 0 C (without load ................. 50 5 7 Average fracture energy density at each temperature ................................ ........ 50 5 8 PG 67 22 at 10 C, def ormation at fracture vs. loading rate ............................... 51 5 9 Before modification ................................ ................................ ............................. 52 5 10 After modification ................................ ................................ ................................ 52 5 1 1 PG 76 22 recovered, fracture energy vs. loading rate ................................ ........ 54 5 1 2 AC 20 recovered, fracture energy vs. loading rate ................................ ............. 58 5 1 3 AC 30 recovered, fracture energy vs. loading rate ................................ ............. 59 5 14 PG 64 22 recovered, fracture energy vs. loading rate ................................ ........ 60 5 15 True stress vs. true strain, PG 76 22, recovered from field (Superpave #19278) ................................ ................................ ................................ .............. 61 5 16 True stress vs. true strain, PG 64 22, recovered from field (Superpave #19312) ................................ ................................ ................................ .............. 62 5 17 True stress vs. true strain, ARB 5, recovered from field (Superpave #19298) ... 62 5 18 True stress vs. true strain, ARB 5, PAV Residue ................................ ............... 62 5 19 PG 76 22, rubber modified and unmodified binder, fracture energy vs. loading rate ................................ ................................ ................................ ......... 63 5 20 ................................ ................................ ................. 67 5 21 Hybrid binders, fracture energy vs. loading rate ................................ ................. 68 5 22 True stress vs. true strain, Geotech hybrid binder at 100 mm/min ..................... 69 5 23 True stress vs. true strain, Geotech hybrid binder at 200 mm/min ..................... 69 5 24 True stress vs. true strain, Wright hybrid binder at 500 mm/min ......................... 70 5 25 True stre ss vs. true strain, Hudson hybrid binder at 500 mm/min ....................... 71

PAGE 13

13 5 26 Rubber modified binders, true stress vs. true strain ................................ ........... 71 5 27 True stress vs. true strain, Marianni hybrid binder at 100 mm/min ..................... 72 5 28 True stress vs. true strain, Marianni hybrid binder at 225 mm/min ..................... 72 5 29 Extension greater than 1.2 in without fracture, PG 82 22 at 15 C ..................... 73 5 30 True stress vs. true strain, PG 82 22 at 15 C, 900 mm/min .............................. 74 5 31 True stress vs. true strain, PG 82 22 at 15 C, 800 mm/min .............................. 74 5 32 PG 82 22, fracture energy vs. loading rate ................................ ......................... 75 5 33 Comparison of fracture section between PG 82 22 and other binders ............... 76 7 1 PG 67 2 2 PAV residue, true stress vs. true strain, 15 C ................................ ... 97 7 3 AC 20 recovered, true stress vs. true strain ................................ ....................... 99 7 4 PG 64 22 recovered, true stress vs. true strain ................................ .................. 99 7 5 Unmodified binders, true stress vs. true strain ................................ .................. 100 7 6 PG 76 22 recovered, true stress vs. true strain (1) ................................ ........... 101 7 7 PG 76 22 recovered, true stress vs. true strain (2) ................................ ........... 102 7 8 PG 76 22 recovered, true stress vs. true strain (3) ................................ ........... 102 7 9 PG 76 22 recovered, true stress vs. true strain (4) ................................ ........... 103 7 10 PG 76 22 PAV residue, true stress vs. true strain ................................ ............ 104 7 11 PG 82 22 PAV residue, true stress vs. true strain ................................ ............ 105 7 12 Polymer modified binders, true stress vs. true strain ................................ ........ 106 7 13 ARB 5 and ARB 12 PAV residue, true stress vs. true strain ............................ 107 7 14 Overheated ARB 12 PAV residue, true stress vs. true strain ........................... 108 7 15 ARB 5 recovered, true stress vs. true strain ................................ ..................... 108 7 16 Marianni PAV residue, true stress vs. true strain ................................ .............. 109 7 17 Rubber modified binders, true stress vs. true strain ................................ ......... 110 7 18 Wright PAV residue, true stress vs. true strain ................................ ................. 111

PAGE 14

14 7 19 Hudson PAV residue, true stress vs. true strain ................................ ............... 112 7 20 Geotech PAV residue, true stress vs. true strain ................................ .............. 112 7 21 Hybrid binders, true stress vs. true strain ................................ ......................... 113 7 22 PG 76 22 and unmodified binder, true stress vs. true strain ............................. 114 7 23 Comparison with Hudson, true stress vs. true strain ................................ ........ 115 7 24 Comparison with Geotech, true stress vs. true strain ................................ ....... 115 7 25 Comparison with Wright true stress vs. true strain ................................ .......... 116 7 26 Comparison with ARB 12, true stress vs. true strain ................................ ........ 117 7 27 Comparison with Marianni, true stress vs. true strain ................................ ....... 117 7 28 Comparison with PG 82 22, true stress vs. true strain ................................ ..... 118 7 29 True stress vs. true strain, polymer modified (reduced size) and unmodified binders ................................ ................................ ................................ .............. 120 7 30 PG 76 22 and Hudson, true stress vs. true strain ................................ ............. 121 A 1 For t ransforming e xtension to t rue s train ................................ .......................... 131 A 2 For t ransforming l oad to t rue s tress ................................ ................................ .. 131 A 3 For c alculating a rea of c ross s ection at the first stress p eak ............................ 132 A 4 For c alculating l ength of 3mm middle p art at the first stress p eak .................... 132 B 1 Fracture e nergy d ensity at 15 C (PAV residue) ................................ ............... 133 B 2 Fracture e nergy d ensity at 1 0 C (PAV residue) ................................ ............... 134 B 3 Fracture e nergy d ensity at 5 C (PAV residue) ................................ ................. 135 B 4 Fracture e nergy d ensity at 5 C (PAV residue) (without loading 10mm/min) ................................ ................................ ................................ ........ 136 B 5 Fracture e nergy d ensity at 0 C (PAV) ................................ ............................. 137 B 6 Fracture e nergy d ensity at 0 C (PAV 10mm/min) ................................ ................................ ................................ ........ 137 B 7 Average fracture energy density at each temperature ................................ ...... 138

PAGE 15

15 B 8 PG 67 22 at 10 C, deformation at fracture vs. loading rate ............................. 138 B 9 Before modification ................................ ................................ ........................... 139 B 10 After modif ication ................................ ................................ .............................. 139 C 1 PG 76 22 recovered, fracture energy vs. loading rate ................................ ...... 147 C 2 AC 20 recovered, fracture energy vs. loading rate ................................ ........... 147 C 3 AC 30 recovered, fracture energy vs. loading rate ................................ ........... 148 C 4 PG 64 22 recovered, fractur e energy vs. loading rate ................................ ...... 148 C 5 PG 76 22, rubber modified and unmodified binder, fracture energy vs. loading rate ................................ ................................ ................................ ....... 149 C 6 True stress vs. true strain, PG 76 22, recovered from field (Superpave #19278) ................................ ................................ ................................ ............ 149 C 7 True stress vs. true strain, PG 64 22, recovered from field (Superpave #19312) ................................ ................................ ................................ ............ 150 C 8 True stress vs. true strain, ARB 5, recovered from field (Superpave #19298) 150 C 9 True stress vs. true strain, ARB 5, PAV residue ................................ ............... 150 D 1 Hybrid binders, fracture energy vs. loading rate ................................ ............... 151 D 2 Rubber modified bi nders, true stress vs. true strain ................................ ......... 151 D 3 PG 82 22, fracture energy vs. loading rate ................................ ....................... 152 D 4 True stress vs. true strain, Wright hybrid binder at 500 mm/min ....................... 153 D 5 True stress vs. true strain, Geotech hybrid binder at 100 mm/min ................... 153 D 6 True stress vs. true strain, Geotech hybrid binder at 200 mm/min ................... 153 D 7 True stress vs. true strain, Hudson hybrid binder at 500 mm/min ..................... 154 D 8 True stress vs. true strain, Marianni hybrid binder at 100 mm/min ................... 154 D 9 True stress vs. true strain, Marianni hybrid binder at 225 mm/min ................... 154 D 10 True stress vs. true strain, PG 82 22 at 15 C, 900 mm/min ............................ 155 D 11 True stress vs. true s train, PG 82 22 at 15 C, 800 mm/min ............................ 155

PAGE 16

16 E 1 ................................ ................................ ............... 156 G 1 PG 67 22 PAV residue, true s tress vs. true strain, 15 C ................................ 162 G 2 AC 30 recovered, true stress vs. true strain ................................ ..................... 162 G 3 AC 20 recovered, tru e stress vs. true strain ................................ ..................... 163 G 4 PG 64 22 recovered, true stress vs. true strain ................................ ................ 163 G 5 Unmodified binders, t rue stress vs. true strain ................................ .................. 164 G 6 PG 76 22 recovered, true stress vs. true strain (1) ................................ ........... 164 G 7 PG 76 22 recove red, true stress vs. true strain (2) ................................ ........... 165 G 8 PG 76 22 recovered, true stress vs. true strain (3) ................................ ........... 165 G 9 PG 76 22 recovered, true stress vs. true strain (4) ................................ ........... 166 G 10 PG 76 22 PAV residue, true stress vs. true strain ................................ ............ 166 G 11 PG 82 22 PAV residue, true stress vs. true strain ................................ ............ 167 G 12 Polymer modified binders, true stress vs. true strain ................................ ........ 167 G 13 ARB 5 and ARB 12 PAV residue, true stress vs. true strain ............................ 168 G 14 Overheated ARB 12 PAV residue, true stress vs. true strain ........................... 168 G 15 ARB 5 recovered, true stress vs. true strain ................................ ..................... 169 G 16 Marianni PAV residue, true stress vs. true strain ................................ .............. 169 G 17 Rubber modified binders, true stress vs. true strain ................................ ......... 170 G 18 Wright PAV residue, true stress vs. true strain ................................ ................. 170 G 19 Hudson PAV residue, true stress vs. true strain ................................ ............... 171 G 20 Geotech PAV residue, true stress vs. true strain ................................ .............. 171 G 21 Hybrid binders, true stress vs. true strain ................................ ......................... 172 G 22 PG 76 22 and unmodified binder, true stress vs. true strain ............................. 172 G 23 Comparison with Hudson, true stress vs. true strain ................................ ........ 173 G 24 Comparison with Geotech, true stress vs. true strai n ................................ ....... 173

PAGE 17

17 G 25 Comparison with Wright, true stress vs. true strain ................................ .......... 174 G 26 Comparison with ARB 12, true stress vs true strain ................................ ........ 174 G 27 Comparison with Marianni, true stress vs. true strain ................................ ....... 175 G 28 Comparison with PG 82 22 true stress vs. true strain ................................ ..... 175 G 29 True stress vs. true strain, polymer modified (reduced size) and unmodified binders ................................ ................................ ................................ .............. 176 G 30 PG 76 22 and Hudson, true stress vs. true strain ................................ ............. 176

PAGE 18

18 LIST OF ABBREVIATION S ARB Asphalt Rubber Binder. Rubber modified binder. BBR Bending Beam Rheometer A standard binder test. DSR Dynamic Shear Rhe ometer A standard binder test. DT Direct Tension. A standard binder test. ER Elastic Recovery. A standard binder test. FD Forced Ductility. A standard binder test. FDOT Florida Department of Transportation. FEA Finite Element Analysis. A numerical simulat ion method. MSCR Multiple Stress Creep Recovery A standard binder test. MTS Materials Testing System. A materials testing machine. PAV Pressure Aging Vessel. A standard long term aging method for binder. SBS Styrene Butadiene Styrene. A type of polymer u sed to modify binder. SENB Single Edge Notched Beam. XFEM Extended Finite Element Method. A numerical simulation method.

PAGE 19

19 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 DEVELOPMENT OF A BINDER FRACTURE TEST TO DETERMINE FRACTURE ENERGY By T ianying N iu December 2011 Chair: Reynaldo Roque Major: Civil Engineering I t has been found that all existing binder testing metho ds in current specifications do not accurately predict cracking performance at intermediate temperatures F racture energy ha s been determined to be strongly correlated to fracture resistance of asphalt mixture, so a test to measure fracture energy of binde r is expected to provide an excellent tool to evaluate fracture resistance of binder. A new fracture test and interpretation system was successfully developed b ased on Finite Element Analysis (FEA) and prototype test on MTS machine to consistently measure fracture energy of binder at intermediate temperature For evaluation, t he new fracture test and interpretation system was applied to a range of binders including unmodified binder, SBS modified binder, rubber modified binder hybrid binder and highly SBS modified binder from PAV residue or recovered from field test section s S tatistical analys i s w as conducted on test results which showed that the new fracture test and interpretation system significant ly distinguish ed between different binders by fracture energy Expected trends in fracture energy between binders were observed, which indicates the test was successful It was also shown that f or the same binder, fracture energy is independent of loading rate evaluated in this study and

PAGE 20

20 test temperature from 0 to 15 C Thus, fracture energy appears to be a fundamental property of binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate. T he results also showed that d ifferent types of b inder have different characteristic true stress true strain curves, which can be used to identify the binder type, modifier type and relative content. Basic principles were proposed to identify the presence of modifier from true stress true strain curves A detailed test ing pro tocol wa s recommended The protocol helps assure the appropriate loading rate range so that th e complete true stress vs. true strain curve can be identified for accurate determination of fracture energy. I n conclusion, the new fractu re test and interpretation system appears to be suitable to measure fracture energy of the broad range of binder used in asphalt paving.

PAGE 21

21 CHAPTER 1 INTRODUCTION 1.1 Background Fatigue performance of asphalt mixture and pavement is known to be strongly inf luenced by the fatigue resistance of the binder. Also, there is a need for a system to determine mixture fracture energy from constituent properties (i.e., properties of fr Unfortunately, a recentl y completed study for FDOT showed that none of the existing Force Ductility (FD) were able to provide parameters that consistently correlated with the relative cracking performance of mixtures at intermedia te temperatures Fracture energy is an important property that relates to the fatigue resistance of binders. In the same FDOT study mentioned above, an approach to determine cumulative energy to failure from FD results showed improved ability to predict cr acking performance at intermediate temperatures Even though the FD is not optimized to determine fracture energy accurately, the results indicated that a test designed specifically to obtain fracture energy could provide a much better parameter related to fatigue resistance of binder. Figure 1 1. Traditional Superpave DT t est

PAGE 22

22 In concept, the Direct Tension (DT) test (Figure 1 1) is a more suitable approach to measure fracture energy of binder. However, the traditional DT test cannot reflect the real con dition of binder between aggregates which experience cracking, and exhibits high variability in results due to its specimen shape. Within the long middle part with uniform area, where the stress distribution is similar everywhere, the specimen may crack an ywhere, which makes it impossible to determine failure strain accurately and results in high deviation of test results. In other words, it is difficult to predict the deformed shape, which makes it difficult to get accurate stress strain relationships. In addition, the long middle part makes it difficult to apply a high enough strain rate to any inaccuracy in simulation. For some types of binder, the fracture test may e xceed the capacity of testing equipment without failure. These issues made it necessary to develop a new DT test that allows for accurate determination of stress strain relationships of binder, from which fracture energy can be obtained. 1.2 Objectives Th e primary objective of this study is to develop a binder Direct Tension test and associated data interpretation methods that allow for determination of binder fracture energy at intermediate temperatures. Detailed objectives can be summarized as follows: 1. Use 3 D finite element analysis (FEA) and prototype testing to develop and identify an optimized specimen configuration to determine fracture energy accurately. 2. Develop and identify test procedures and appropriate measurement systems from which fracture e nergy can be determined accurately.

PAGE 23

23 3. Identify appropriate data interpretation procedures to calculate true stress and true strain, and to determine the instant of fracture initiation and to calculate fracture energy. 4. Evaluate the system developed by measuri ng fracture energy for a range of binders for which expected trends in fracture energy are known. 5. Evaluate the effect of measured binder fracture energy on the fracture energy of mixtures produced with the binders tested. 1.3 Hypothesis The fracture energ y of asphalt binder can be measured consistently at intermediate temperatures associated with fatigue cracking (0 to 20 o C). The fracture energy is a fundamental asphalt binder property, which has a typical value range regardless of test conditions 1.4 Re search Approach The research process consists of six tasks as follows: 1. Literature Review 2. Identify and Optimize Specimen Geometry 3. Build Prototype System and Modify as Needed 4. I dentify Test Procedures, Measurement Systems and Data Interpretation Methods 5. P er form Tests on a Range of Binders 6. Evaluate System In task 2 and task 3, a new specimen configuration was developed by FEA and prototype test In task 4 and task 5, it was proved that the fracture energy of binder is independent of test temperature and loadi ng rate. The optimal test condition (loading rate and temperature) was also identified. Task 5 test ed and verif ied the new system can accurately predict cracking performance at intermediate temperatures

PAGE 24

24 CHAPTER 2 LITERATURE REVIEW I n order to improve f racture resistance of binder, many types of modifier including polymer and rubber have been applied to binder. R ecently, various combination s of polymer, rubber and binder called hybrid binder were also produced H owever, it has been found very difficult to quantitatively evaluate the fracture resistance of modified binders and differentiat e between them particularly for those with complicated material properties. At present, most relevant researches are still focused on traditional testing methods such as Dynamic Shear Rheometer (DSR) Bending Beam Rheometer (BBR), Elastic Recovery, Ductility, etc., with traditional parameters such as complex shear modulus G* phase angle etc., or some parameters derived from th ese tests such as y ield e nergy and strain at maximum stress ( Bahia et al 2008 ) Some researchers realized th at some traditional testing methods are not suitable for modified binders, and tried to develop new tes ts to improve the accuracy. Rosales (2011) used the single edge notched beam (SENB) as an alternative to Bending Beam Rheometer (BBR) to determine both stiffness and fracture energy of modified binders, and found that the stiffness obtained through SENB is higher than that of BBR. The SENB is a typical fracture mechanics testing method. However, it is not suitable for highly ductile materials. Actually, the SENB is designed for binder at low temperature with brittle fracture, but not applicable to binder at intermediate temperature with ductile fracture. S ome researcher s noticed the limitation of current binder tests. At the University of Florida, in a recently finished research project in hybrid binder Roque et al (2009)

PAGE 25

25 evaluated almost all existing bind er testing methods including Dynamic Shear Rheometer (DSR) Bending Beam Rheometer (BBR) Multiple Stress Creep Recovery (MSCR) Elastic Recovery (ER) and Forced Ductility (FD) test and found that all tests were not able to accurately predict cracking per formance at intermediate temperatures although Multiple Stress Creep Recovery (MSCR) Elastic Recovery (ER) and Forced Ductility (FD) test can identify polymer modifie d binder to some extent A n approach to determine cumulative energy to failure from FD r esults showed improved ability to accurately predict cracking performance at intermediate temperatures Even though the FD is not optimized to determine fracture energy accurately, the results indicated that a test designed specifically to obtain fracture energy could provide a much better parameter related to fatigue resistance of binder. T herefore, t he research recommended to devel o p a new binder fracture test to determine fracture energy, which resulted in this research project. T o test binder s ductile fracture energy at intermediate temperature is difficult and thus pretty new since it is not easy to accurately perform constitutive m odeling of true s tress true strain for a complicated nonlinear viscoelastic or plastic modified binder which is often hi ghly ductile and undergo es large deformation and fracture process This constitutive m odeling of true s tress true strain is based on both Finite Element Analysis (FEA) simulation and prototype test H owever, there is usually a maximum strain level limit t hat FEA can accurately simulate based on large strain formulation due to the very small or negative determinant of Jacob matrix, which results from the too distorted mesh. T he simulation of facture process is similar since the mesh will be changed T he mes hfree method as an alternative to FEA, is good at dealing with

PAGE 26

26 the problem s. H owever, at present, it is not mature and has its own limitations and is pretty time consuming, which prevents it from practical use. T he Extended Finite Element Method (XFEM) by use of additional discontinuous basis functions for crack opening displacement, is good at solving fracture problem because the mesh need not to be updated along with cracking T he Extended Finite Element Method (XFEM) is being incorporated by some FEA softwares. In the future, we can expect a rapid development in nonlinear numerical simulation field. However, considering the different combinations of various materials and the complicated materials properties, to ac curately simulate the large deformation and fracture process will remain a difficult challenge. Just because of this reason, this is an exciting field.

PAGE 27

27 CHAPTER 3 MATERIALS AND METHOD S 3 .1 Components of Binders A primary purpose of this research is to accurately predict cracking performanc e of binders at intermediate temperatures Therefore, a wide range of binders were selected to verify the effectiveness of the new Direct Tension test. T otal twelve types of binder including unmodified binders, SBS modified binders, rubber modified binders and hybrid binders were tested and analyzed. Both PAV residues and recovered binders were prepared. Their components are listed in Table 3 1. Table 3 1. Binder s and the c onstituents/ f ormulations Binder Modifying c omponents PG 67 22 None (tested as a PG 69.78 26.50) PG 64 22 None AC 30 None AC 20 None PG 76 22 4.25% SBS (tested as a PG76.7 27.16) PG 82 22 8.5% SBS Geotech 1% SBS (approximately 30 mesh, incorporated dry), 8% of Type B GTR, 1% hydrocarbon Hudson 3.5% crumb rubber, 2.5% SBS, 0.4% plus Link PT 743 cross linking agent Wright GTR and SBS (digested rubber) Unknown contents. Marianni Unclear. Maybe 13% GTR I ts true stress true strain curve shows that it may contain polymer. ARB 5 5% Type B rubber ARB 12 12% Type B rubber

PAGE 28

28 3 .1 Prel iminary Tests I n order to verify the new Direct Tension test, and identify the optimal test conditions and appropriate test procedures, the preliminary tests were performed. 3 .1.1 Binder T ypes All the binders are PAV residues. 1 type of SBS polymer modif ied binder: PG 76 22 (4.25% SBS) 1 type of u nmodified binder: PG 67 22 3 .1. 2 Testing Method Tests were run on MTS machine. Test temperature. T he following t est temperatures were used : 0 C, 5 C, 10 C, 15 C, 20 C A t 20 C, specimens became too soft to obtain fracture energy accurately. Loading rates Various loading rates were used depend ing on test temperature in order to avoid premature fracture. 3 2 Tests on Binders Recovered from Superpave Sections T he new D irect T ension test system was perfor med on a range of binders recovered from Superpave Monitoring Project (BDK 75 977 06) for further evaluation. These 12 Superpave Monitoring Project s each included two layers of asphalt mixture encompassing a broad range of binders. All the recovered binder s were prepared by FDOT. Totally several hundreds of specimens were tested, and more than 200 specimens were tested successfully that is, to avoid premature fracture, while at the same time, to avoid the extra ductile fracture section shape which results in under estimation of fracture energy

PAGE 29

29 3 .2.1 Binder T ypes All binders were recovered from asphalt mixture of 12 Superpave Monitoring Projects each included two layers. 3 types of u nmodified binders: AC 30 AC 20 PG 64 22 1 type of SBS polymer modified binder: PG 76 22 1 type of rubber modified binder: ARB 5 3 .2. 2 Testing Method Tests were run on MTS machine. Test temperature A ll tests were conducted at 15C since prior work indicated that this was the optimal temperature to get consistent fracture en ergies. Loading rates V arious loading rates were used depending on the property of individual binders in order to avoid both premature fracture, which results in erroneous results, and excessively ductile fracture which result s in the und erestima tion of fracture energy. 3 3 Tests on Hybrid Binders and Highly Polymer Modified Binder In order to further verify and evaluate the capability of the new DT test to distinguish between relatively complicated modified binders in terms of fracture resistence, 3 types of hybrid binders: Wright Hudson Geotech, 1 type of highly polymer modified binder PG 82 22 and 1 type of rubber modified binder Marianni were tested and analyzed. The components of Marianni are unclear. It was said that Marianni contains 13% ru bber modifier. However, its true stress true strain curve shows that it may contain True Strain

PAGE 30

30 The original binders were prepared by FDOT. Then the PAV was performed on them. All the DT tests were conducted on PAV residues. 3.3.1 Binder Types All the binders are PAV residues. 3 types of hybrid binders: 1. Wright (GTR and SBS. Unknown contents) 2. Hudson (3.5%crumb rubber+ 2 .5%SBS+0.4% plus Link PT 743 Cross Linking) 3. Geotech (1 % SBS (approximately 30 mesh, incorporated dry) +8% of Type B GTR+1% hydrocarbon) 1 type of rubber modified binder: Marianni (Components are unclear. Maybe 13 % Tire Rubber The true stress true strain curve shows that it may contain polymer.) 1 type of hi ghly polymer modified binder: PG 82 22 ( 8.5% SBS ) 3.3. 2 Testing Method Tests were run on MTS machine. Test temperature. 10 C and 15 C were used. Hybrid binders and Marianni : all tests were conducted at 15 C because it is the optimal temperature for mos t types of binder to get consistent fracture energies. PG 82 22: a PG 82 22 binder was tested at both 15 C and 10 C. Due to the very ductile nature of this highly polymer modified binder PG 82 22, it was impossible to perform the new DT test successfull y at 15 C, where the deformations to failure were excessive even at very fast loading rates. The excessively ductile specimen shape may make the interpretation inaccurate. Loading rates. V arious loading rates were used depending on the property of indiv idual binders in order to avoid both premature fracture, which results in erroneous

PAGE 31

31 results, and excessively ductile fracture which result s in the und erestima tion of fracture energy. For hybrid binders and Marianni, we applied the loading rates up to the possible fastest rate creating mature fracture. For highly polymer modified binder PG 82 22, premature fracture is uncommon and complicated, which will be shown and discussed later. Due to this reason, a broad range of loading rates wer e applied to PG 82 22 in order to find out the method to deal with this problem.

PAGE 32

32 CHAPTER 4 DEVELOPMENT OF A NEW BINDER FRACTURE TEST 4.1 Development of Specimen Geometry by Finite Element Analysis (FEA) 4.1.1 Reason s Why to Develop a New Specimen Geome try According to former research, the fracture energy analysis based on the Direct Tension test can predict cracking performance at intermediate temperatures better compared to other binder tests. However, the traditional Direct Tension test has some cruci al defects in terms of getting accurate fracture energy. T cross section area in a long middle part. This specimen cannot reflect the real condition of asphalt binder between agg regates which experience cracking. Within the long middle part with uniform cross section area, except that close to both ends, the stress distribution is similar everywhere. The specimen may crack anywhere, which results in high deviation of test results (strength). Due to the same reason, it is difficult to predict the deformed shape. In order to simulate the deformed shape during extension process accurately, we would like to reduce the ductility of asphalt specimen. However, the long middle part makes makes the calculated fracture energy sensitive to any inaccuracy in simulation. For some types of binder, because of the long middle part of the specimen, the fracture tes t may exceed the capacity of testing equipment without failure. In conclusion, it is difficult for traditional Direct Tension test to get the accurate stress strain relationship during fracture process, which is required to calculate fracture energy. We ne ed a new specimen geometry to obtain the accurate fracture energy.

PAGE 33

33 4.1. 2 Development and Optimization Process 4.1.2.1 Geometry No.1 In order to simulate the real condition of asphalt binder between aggregates which experience cracking, and get a high str ess concentration factor, we created a 2 D FE A model (Figure 4 1) A B C Figure 4 1. 2 D FEA model A) S implified spheres B) 2 D complex curves C) S tress distribution by FEA Since for the 2 D shape, the possible highest stress concentration factor is 2.05 which is not enough, then we extended it to 3 D shape (Figure 4 2) A B Figure 4 2. 3 D FEA model. A) 3 D specimen shape, B) S tress distribution on horizontal and v ertical cross section by FEA 3 D FE A results indicate that a highly uniform, nearly isotropic stress state exists in its central narrow portion (3mm3mm). Also, the tensile stresses are eleven times higher than tensile stresses near the edge, which helps e nsure the specimen will fail

PAGE 34

34 first within the region of the narrow gap. Allowing for 1 mm to account for end effects at the binder loading head interface, a cross section of 5mm5mm was selected. At first, the test was run on the Direct Tension test machi ne. However, at a relatively high extension speed, the tensile stress of the specimen may exceed the capacity of the DT machine and t hen the testing system would stop. At a rela tively low extension speed, an over large deformation to fracture was encounte red, and the strain rate dependence, a typical viscoelastic property, clearly exhibited after we plotted the stress strain curves (Figure 4 3 ) based on the Large Displacement/Large Strain formula in FE A Figure 4 3 Geometry No.1, t rue s tress vs. t rue s train A fast fracture with a small deformation is preferred in order to avoid th is behavior of asphalt binder Then the test was moved to the MTS machine which can provide a high extension speed as well as a high tensile stress. In order to avoid any ben ding moment or torque caused by inaccurate position of load heads, a special equipment with two guide bars and a movable load head connection was designed for the MTS (Figure 4 4 A)

PAGE 35

35 A B Figure 4 4. Geometry No.1, t esting on MTS. A) Testing equipm ent. B) Asphalt peeled off from load head. However, at 5 C, the asphalt was peeled off from load head (Figure 4 4 B) which showed the adhesion between asphalt and load head is less than the cohesion within asphalt. We have to modify the specimen shape in order to eliminate the stress concentration on the contact surface of load head. 4.1.2.2 Geometry No. 2 In order to make the specimen crack from middle part, and release the stress concentration on the contact surface of load head, concaves with curved ch amfer were made on both sides (Figure 4 5) Although the stress concentration on the corners has been eliminated, this shape is too complicated to make in practice. If we are only interested in the bottleneck shape at the center of the specimen, why not j ust extract this simple shape?

PAGE 36

36 A B Figure 4 5. Geometry No.2, FEA model A) Concave on b oth s ides B) Vertical cross s ection 4.1.2.3 Geometry No. 3 A B Figure 4 6. Geometry No.3, FEA model. A) Bottleneck s hape B) Str ess distribution on vertical and horizontal cross section. For a relatively thick bottleneck specimen, the stress concentration is in the middle part on both sides. For a very thin specimen, the stress concentration is on the edges. Therefore, we prefer a relatively thick specimen. We also made two concaves on both sides close to form a relatively uniform stress concentration area at the center of specimen. The stress concentration factor is greater than 5, which is high enough. From the FE A model (Figure 4 6) we can see there is no stress concentration on load heads.

PAGE 37

37 The test was run on the Direct Tension test machine at 4 loading rates (300mm/min, 150mm/min, 100mm/min and 70mm/min) and at 2 temperatures (5 C, 10 C) for 2 types of binder (unmodified PG 67 22, SBS modified PG 76 22). Figure 4 7 Geometry No.3, asphalt peeled off from load head All specimens cracked from the contact surface of load head (Figure 4 7) which showed the adhesion between asphalt and load head is still not enough to make the specimen crack in the middle. We found that the extension rate of Direct Tension test machine was not accurate above 150mm/min for SBS modified binder PG 76 22. For 150mm/min and 300mm/min, its actual rate was 120mm/min and 258mm/min respectively, whi ch showed the force required for these loading rate s exceeded the capacity of Direct Tension test machine. Therefore, we may need to move tests to MTS machine. We can also see that their response was close (Figure 4 8 ) which means the loading rate is almo st high enough. Just because of the cracking from contact surface of load head, the same type of specimen may crack at different extension distance.

PAGE 38

38 Figure 4 8 Geometry No.3, PG 76 22, 5 C, 300 mm/min & 150 mm/min We have to strengthen the connecti on between asphalt and load head, and eliminate any high stress on the corners of load head, which resulted in the Geometry No.4 (Figure 4 9 ) 4.1.2. 4 Geometry No. 4 The connection is strengthened by extending the asphalt specimen on both sides of load he ad. Compared to Geometry No.3, the load heads are backed away from the previous specimen ends in order to avoid cracking from contact surface. Figure 4 9 Geometry No. 4 The corners of load head are rounded to eliminate any stress concentration and unwan ted cracking. This specimen cracked in the middle. The connection between specimen and load heads was tight. It meets our requirement Figure 4 1 0 shows the final dimensions of Geometry No.4

PAGE 39

39 Figure 4 1 0 Dimensions of Geometry No. 4 Figure 4 1 1 S tress distribution on h orizontal c ross s ection by FEA The stress concentration is still on both sides in the middle. A relatively uniform stress distribution on cross section is formed (Figure 4 1 1 ) The stress concentration on the contact surface of load head has been eliminated. The specimens were prepared by molds shown in Figure 4 1 2

PAGE 40

40 A B Figure 4 1 2 Preparation of specimen. A) mold. B) specimen. In order to avoid any bending moment or torque caused by inaccurate position of load heads, we used a special equipment with two guide bars and a movable load head connection designed for the MTS (Figure 4 1 3 ) Figure 4 1 3 New DT t esting e quipment on MTS As shown in Figure 4 14, compared to PG 67 22, PG 76 22 has a better ductility. This ca n be seen on the fracture section, particularly at lower loading rates. Geometry No.4 was finally determined as our optimal specimen configuration.

PAGE 41

41 SBS Modified Binder PG 76 22 Unmodified Binder PG 67 22 Figure 4 1 4 Fracture of s pecimens 4. 2 Data Interpretation 4. 2 .1 Premature Failure Identification At low temperatures and/or faster loading rates, any imperfect ion of specimen may result in premature failure. These specimens did not always crack at the exact center, which indicated that the fr acture actually resulted from flaws. These erroneous test results must be identified and discarded. T hese results also impl y that there is an optimal combination of temperature and loading rate range to consistent ly obtain fracture energy of binder. In pra ctice, in most cases it is relatively easy to identify premature fracture based on necking, true stress true strain graph and fracture energy (Figure 4 1 5 ) Compared to mature fracture the premature fracture However, b ehind these intuitive evidences, the real nature of premature fracture is the in complete t rue stress true strain curve. We will state the details later.

PAGE 42

42 Mature Fracture Premature Fracture Figure 4 1 5 Identifi cation of premature fracture 4. 2 2 New Calculation Procedure for Fracture Energy The original test result is load versus extension. In order to accurately calculate fracture energy of binder on middle cross section, we need to transform measured load to average stress on middle cross section, and measured extension to average true strain on middle cross section by use of FEA at first. However, all ductile cracks clearly exhibit necking in the middle of specimen. For this research, FEA is not adequate to simulate this type of failure accurately even with large strain formulation because we need an uniform analysis for a standard test, while FEA requires different constitutive modeling for different binders with various complicated properties in different c onditions, which is inconvenient in practice. The calculation only based on FEA with a general material model often results in an underestimated fracture energy, and a false stress stain curve, and hides the real fracture point. The difference between poly mer modified binder and unmodified binder wi ll not be as large as expected. FE=208.3 psi FE=11.6 psi

PAGE 43

43 A new data analysis procedure was developed to account for this necking, and correct the error from limitation of FEA. From the true stress true strain graph obtained based on FEA: 1. B efore the peak of stress, still use FEA 2. A fter the peak of stress, based on observation, assume most strain happens in the middle 3 mm of the specimen, and use the large strain formula (3 1) W here : L 0 A s shown in Figure 4 1 6 a fter applying the new calculation method, the point of initial fracture is very clear. T he post peak energy after the point of initial fracture should not be co nsidered Fracture energy should be calculated from the beginning of true stress true strain curve to the last stress peak which is the point of initial fracture. We True Strain Curves A B Figure 4 1 6 True s tress t rue s train c urve by n ew c alculation p rocedure A) PG 67 22, 10 C, 400mm/min B) PG 76 22, 10 C, 400mm/min fracture fracture

PAGE 44

44 4. 2 3 Finite Element Analysis (FEA) Results For convenience of performing this calculation procedure, based on finite element analysis (FEA), a set of diagrams were plotted (Figure 4 1 7 to Figure 4 2 0 ). Figure 4 1 7 For t ra nsforming e xtension to t rue s train Figure 4 1 8 For t ransforming l oad to t rue s tress

PAGE 45

45 Figure 4 19 For c alculating a rea of c ross s ection at the first stress p eak Figure 4 2 0 For c alculating l ength of 3mm middle p art at the first stress p eak

PAGE 46

46 C HAPTER 5 TESTS AND ANALYSES O N A RANGE OF BINDERS 5 1 Preliminary Tests I n order to verify the new Direct Tension test, and identify the optimal test conditions and appropriate test procedures, the preliminary tests were performed. 5.1.1 Testing Method Tests were run on MTS machine. Test temperature s T he following t est temperatures were used : 0 C, 5 C, 10 C, 15 C, 20 C A t 20 C, specimens became too soft to obtain fracture energy accurately. Loading rates Various loading rates were used depend ing on test temperature in order to avoid premature fracture. Binder types. All the binders are PAV residues. 1 type of SBS polymer modified binder: PG 76 22 (4.25% SBS) 1 type of u nmodified binder: PG 67 22 5.1.2 Test Result and Analysis T est r esults are shown in Figure 5 1 through Figure 5 6 : 5.1.2.1 15 C A s shown in Figure 5 1 the fracture energy density is very consistent for the same binder at different loading rates, and the difference between modified and unmodified binder is very clear. It a ppears that this is the optimal test condition.

PAGE 47

47 Figure 5 1. Fracture e nergy d ensity at 15 C 5.1.2.2 10 C Figure 5 2. Fracture e nergy d ensity at 1 0 C A s shown in Figure 5 2 a t 10 C, some specimens fracture d prematurely whi ch results in high er deviation in results. However for the same binder, the fracture energy density is still relatively consistent at different loading rates, and is also consistent with th ose tested at other temperatures. The difference between modified and unmodified binder remains very clear.

PAGE 48

48 5.1.2.3 5 C Figure 5 3. Fracture e nergy d ensity at 5 C A s shown in Figure 5 3 w hen the loading rate is equal to or lower than 10 mm/min at 5 C the specimen becomes so soft that the fracture section is like a needle, which res ults in underprediction of fracture energy F igure 5 4 shows the results at 5 C without these loading rates. Figure 5 4 Fracture e nergy d ensity at 5 C (without loading rates 10 mm/min )

PAGE 49

49 A t 5 C, some specimens fractured prematurely which results in a high er deviation in results for polymer modified binder A s shown in Figure 5 4 for the same binder, the fracture energy density is still relatively consistent at different loa ding rates, and is also consistent with th ose tested at other temperatures. Again, t he difference between modified and unmodified binder remain s very clear. 5.1.2.4 0 C Figure 5 5. Fracture e nergy d ensity at 0 C At 0 C, all PG 67 22 specimens faile d prematurely T herefore, only PG 76 22 results are presented in Figure 5 5 for 0 C A s shown in Figure 5 5 similar results were observed at loading rates equal to or lower than 10 mm/min at 0 C F igure 5 6 shows the results at 0 C without 10 mm/min. A s shown in Figure 5 6 for PG 76 22, the fracture energy density is consistent at different loading rates, and is also consistent with th ose tested at other temperatures. From Figure 5 6 we can see that the fracture energy density of PG 76 22 at 100 mm/mi n is a little bit higher than at other loading rates result, which can be adjusted by using a length higher than 3mm. This will be explained later.

PAGE 50

50 Figure 5 6. Fracture e nergy d ensity at 0 C (without loading rates 10 mm/min ) To summarize all results, the average fracture energy density of the same binder at each temperature was plotted in Figure 5 7 Figure 5 7. Average fracture energy density at each temperature As shown in Figure 5 7, the average fracture e nergy at each temperature is very consistent for PG 76 22 and PG 67 22, respectively. The difference between PG 76 22 and PG 67 22 is very clear. It shows the fracture energy is independent of temperature,

PAGE 51

51 and the proposed DT test appears to be effective a nd accurate. The more reliable statistical analysis will be presented later. Figure 5 8. PG 67 22 at 10 C, d eformation at f racture vs. l oading r ate Figure 5 8 shows that the lower the loading rate, the higher the deformation at fracture. At low loadin g rates, the center of specimen becomes extremely ductile, sometimes we may take a length < 3 mm as the portion where most strain happens during necking. At high loading rates, due to the same reason, sometimes we may take a length > 3 mm. For PG 67 22 at 10 C (Figure 5 9 3 mm for all specimens), if we use 2 mm for the low loading rate 25mm/min, and 4 mm for the high loading rates 600mm/min and 400mm/min, the fracture energy becomes consistent (Figure 5 1 0 2 mm low, 4 mm high). In summary, the new Dir ect Tension test can differentiate between the polymer modified PG 76 22 and the pure binder PG 67 22 clearly. At 15 C, the measured fracture energy is very consistent for the same binder at different loading rates.

PAGE 52

52 Figure 5 9. Before m odification Fi gure 5 1 0. After m odification Thus, 15 C is the optimal test temperature for these types of binder. The appropriate loading rate should not make specimens too ductile or fracture prematurely which may result in an inaccurate fracture energy. Further res ults will be shown later in a ranger of binder s test. 5 2 Tests on Binders Recovered from Superpave Sections The new Direct Tension test system is supposed to be able to distinguish between different binders clearly in terms of fracture resistance It ha d been applied to PAV residue of SBS modified PG 76 22 and unmodified PG 67 22 successfully. Then the

PAGE 53

53 new DT test system was performed on a range of binders recovered from Superpave Monitoring Project (BDK 75 977 06) for further evaluation. These 12 Superp ave Monitoring Project s each included two layers of asphalt mixture encompassing a broad range of binders. All the recovered binders were prepared by FDOT. Totally s everal hundreds of specimens were tested, and more than 200 specimens were tested successfu lly that is, to get the ductile fracture, and avoid the extra ductile fracture section shape which results in under estimation of fracture energy These test results and analyses further verif ied the new DT test and interpretation system, and help to more clearly de fine appropriate loading rates and test procedures. 5.2 .1 Testing Method Tests were run on MTS machine. Test temperature A ll tests were conducted at 15C since prior work indicated that this was the optimal temperature to get consistent fractu re energies Loading rates. Various loading rates were used depending on the property of individual binders in order to avoid both premature fracture, which results in erroneous results, and excessively ductile fracture, which results in the underestima tion of fracture energy. Binder types All binders were recovered from asphalt mixture of 12 Superpave Monitoring Projects each includ ed two layers. 3 types of u nmodified binders: AC 30 AC 20 PG 64 22 1 type of SBS polymer modified binder: PG 76 22 1 ty pe of rubber modified binder: ARB 5

PAGE 54

54 5. 2 .2 Test Result and Analysis A part of test results are shown in Table 5 1. The complete test results are vast and are shown in appendix. Each type of binder was tested at multiple loading rates. In order to show th e influence of loading rate on fracture energy, for the binders with numerous specimens, the tested fracture energy of the same type of binder was averaged at the same loading rate, then the fracture energy vs. loading rate was plotted for each type of bin der. F or the convenience of comparison, all these figures including hybrid binders which will be test ed and discuss ed later, use the same y axis scale, i.e. fracture energy density. 5.2.2.1 PG 76 22 recovered from field Figure 5 1 1 PG 76 22 recovered, f racture energy vs. loading rate S ince recovered PG 76 22 has numerous specimens, its averaged fracture energy density at each loading rate was shown in Figure 5 11. F rom Figure 5 11, we

PAGE 55

55 Table 5 1. A part of test results of binder s recovered from Superpave sections (box 1) Box No.1 SMO Lab No. Binder t ype Fracture e nergy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 1 5 A WP 19234 AC 30 300 354.82 0.1827 AC 30 500 3 73.63 0.1693 BWP 19235 AC 30 500 374.21 0.1530 AC 30 500 345.66 0.1507 B WP 19236 AC 30 500 313.02 0.1877 AC 30 500 316.01 0.1828 BWP 19237 AC 30 500 341.04 0.1871 AC 30 500 324.73 0.1856 Table 5 2. A part of t est results of binders recovered from Superpave sections (box 2) Box No.2 SMO Lab No. Binder t ype Fracture e nergy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 1 15 A WP 19238 AC 30 500 406.01 0.1 615 AC 30 500 400.87 0.1499 BWP 19239 AC 30 500 371.94 0.1658 AC 30 500 376.52 0.1745 B WP 19240 AC 30 500 321.51 0.1908 AC 30 500 283.12 0.1762 BWP 19241 AC 30 500 348.64 0.1815 AC 30 500 331.02 0.1903

PAGE 56

56 c an see that generally, the fracture energy of PG 76 22 is consistent at various loading rates. T he loading rate didn t influence the fracture energy significantly. A further statistical analysis will be conducted later to prove this point. On the other han d, we can see that f or binders recovered from Superpave sections, the variance of tested fracture energy of the same binder in different conditions (location, layer, etc.) is greater than that of PAV residues. This makes sense since age hardening of binder in mixtures in the field is a much more complex phenomenon than age hardening in the laboratory i.e., the same binder type will age differently in different mixtures and at different depths within the pavement 5.2.2.2 AC 20 recovered from field The t est results of AC 20 are listed in Table 5 3 Table 5 3 Test results of AC 20 SMO Lab No. Binder Loading r ate (mm/min) Fracture energy d ensity (psi) Extension to f racture (in) 19242 AC 20 500 275.85 0.1686 AC 20 700 306.48 0.1712 19243 AC 20 500 249. 98 0.1896 AC 20 500 271.33 0.2074 19244 AC 20 500 268.67 0.2014 AC 20 500 272.02 0.1999 19245 AC 20 500 288.01 0.2084 AC 20 500 284.45 0.2001 19246 AC 20 500 337.11 0.1978 AC 20 500 316.71 0.1819 19247 AC 20 500 266.64 0.1952 AC 20 500 297.24 0.1994 19248 AC 20 500 256.31 0.2171 AC 20 500 234.13 0.2153 19249 AC 20 500 231.44 0.2223 AC 20 500 243.92 0.2134 19250 AC 20 500 249.17 0.2214 AC 20 500 253.2 0.2169

PAGE 57

57 Table 5 3. (Continued) SMO Lab No. Binder Loading r ate (mm/min) Fracture e n ergy d ensity (psi) Extension to f racture (in) 19251 AC 20 500 274.9 0.2184 AC 20 500 244.38 0.2175 19252 AC 20 500 220.76 0.2131 AC 20 500 228.14 0.2122 19253 AC 20 500 258.22 0.2138 AC 20 500 233.79 0.2123 20368 AC 20 800 284.93 0.1488 AC 20 8 00 300.4 0.1507 20369 AC 20 800 283.07 0.1318 AC 20 800 302.73 0.1332 20370 AC 20 800 276.92 0.1483 AC 20 1000 263.34 0.1424 20371 AC 20 1000 256.69 0.1457 AC 20 20372 AC 20 1000 290.47 0.1293 AC 20 1000 288.45 0.1289 20373 AC 20 1000 27 7.09 0.1133 AC 20 1000 275.36 0.1137 20374 AC 20 1000 263.55 0.1436 AC 20 1000 253.29 0.1482 20375 AC 20 1000 330.95 0.095 AC 20 1000 301.99 0.0966 20468 AC 20 600 265.22 0.1311 AC 20 500 264.7 0.147 20469 AC 20 500 282.02 0.1184 AC 20 20470 AC 20 600 240.69 0.1529 AC 20 800 252.1 0.1411 20471 AC 20 900 267.07 0.1374 AC 20 1200 247.42 0.1273 B ecause of the various aging levels resulting from the complicated field conditions such as different locations, mixtures and depths, etc., the property of AC 20 specimens is quite different. Some are soft and ductile, while some are hard. Therefore, the appropriate loading rate for different AC 20 specimens may vary greatly (Table 5 3 ). For the same reason, t he extension to fracture is also messy. For different AC 20

PAGE 58

58 ily result in a longer extension to fracture. However, the fracture energy of different AC 20 specimens remains close regardless of the loading rate and extension to fracture. This can be seen in both Table 5 3 and Figure 5 12. Figure 5 1 2 AC 20 recovered, fracture energy vs. loading rate T he recovered AC 20 also has numerous specimens. T he averaged fracture energy density at each loading rate level wa s plotted in Figure 5 12. F rom Figure 5 12, we can see that the fracture energy of AC 20 is very consistent at different loading rates. T he loading rate has no influence on the fracture energy. Later, a further statistical analysis will also be performed t o test the effect of loading rate on fracture energy for AC 20. The variance of AC 20 is clearly less than that of recovered PG 76 22. Actually, we will repeatedly see the relatively low variance of unmodified binders compared to some modified binders. T hi s makes sense since the environment in field is complicated,

PAGE 59

59 and may have much influence on modifiers such as polymer and rubber by aging or other effect s, which will affect the fracture resistance performance of modified binders. I n different locations an d layers, this influence on modifiers may be different, which results in the variance in fracture energy. I t is clear that the fracture energy of AC 20 is lower than that of recovered PG 76 22, which means the new Direct Tension test effectively distinguis hed between these two types of binder. T he significance of this difference will be proved by statistical analysis later. 5.2.2.3 AC 30 recovered from field The appropriate loading rate for different AC 30 specimens vary as greatly as that for AC 20 Only three loading rate levels were used for numerous AC 30 specimens. Figure 5 1 3 AC 3 0 recovered, fracture energy vs. loading rate T he average fracture energy density at each loading rate level was plotte d in Figure 5 13. F rom Figure 5 13, again, we can see that the fracture energy of AC 30 is

PAGE 60

60 very consistent at different loading rates. A further statistical analysis will also prove that the loading rate has no effect on the fracture energy for AC 30 5.2.2.4 PG 64 22 re covered from field The appropriate loading rate range for PG 64 22 is from 50 mm/min to 600 mm/min, which is relatively large due to the same reason for AC 20 However, the fracture energy of different PG 64 22 specimens is very close regardless of loading rate. This can be seen in Figure 5 14. Figure 5 14. PG 64 22 recovered, fracture energy vs. loading rate T he fracture energy density at each loading rate level was average d and plotte d in Figure 5 1 4 W e can see that the frac ture energy of PG 64 22 is very consistent at different loading rates. A fur ther statistical analysis will be performed later to prove it. From above figures and Figure 5 19 w e can also see that the fracture energy of recovered PG 76 22 is clearly higher than that of unmodified binders, which is similar to the previous test results of PAV residue of PG 76 22 and PG 67 22. U nmodified

PAGE 61

61 S tatistical analys e s conducted later will test the difference between them. B oth tests of binder made in lab (PAV residue) and recovered in field (Superpave) indicated the same results: 1. The new D irect T ension test system clear ly distinguished between unmodified binders and SBS modified binder 2. The tested fracture energy was consistent at differ ent loading rate s From the test results we can see that the fracture energy of recovered rubber modified binder ARB 5 was close to that of recovered unmodified binders (Figure 5 19) However, i t is very interesting that t he true stress vs. true strain cu rve of recovered rubber modified binder ARB 5 (Figure 5 1 7 ) was also similar to that of recovered unmodified binders AC 20, AC 30, PG 64 22 (Figure 5 1 6 ) where only one stress peak occurred In contrast, the true stress vs. true strain curve of SBS polym er modified PG 76 22 has the second stress peak (Figure 5 1 5 ) Figure 5 1 5 True stress vs. true strain, PG 76 22, recovered from field (Superpave #192 78 )

PAGE 62

6 2 Figure 5 1 6 True stress vs. true strain, PG 64 22, recovered from field (Superpave #19312) Figure 5 1 7 True stress vs. true strain, ARB 5, recovered from field (Superpave #19298) Figure 5 1 8 True stress vs. true strain, ARB 5, PAV Residue

PAGE 63

63 It has been found that during the recovery process the rubber particles are caught in the filter pap er and the refore there is no rubber in binder after recovered This recovered rubber modified binder is actually a pure binder. This is the reason why the fracture energy of recovered rubber modified binder ARB 5 was so close to that of unmodified binders and the ir true stress vs. true strain curve was also so similar. In order to clarify this issue, the PAV residue of rubber modified ARB 5 and ARB 12 was made and tested. Because of the existence of rubber, the true stress vs. true strain curve has a n in flection (Figure 5 1 8 ) instead of the second stress peak of SBS modified PG 76 22. From Figure 5 1 9, we can also see that the fracture energy of PAV residue of rubber modified binders is higher than that of recovered rubber modified binders and unmodified binders Both the fracture energy and the shape of true stress vs. true strain curve of PAV residue of rubber modified binders are between th ose of polymer modified binder and unmodified binders. Figure 5 19. PG 76 22, rubber modified and unmodified binder, fracture energy vs. loading rate

PAGE 64

64 The clear difference of fracture energy and true stress vs. true strain curve between PAV residue of rubber modified binders and recovered rubber modified binders shows the capability of the new D irect T ension test to identify the component of binders. F or polymer modified binders, the polymer is not caught in the filter paper in recovery process. The true stress vs. true strain curve of recovered polymer modified binder PG 76 22 was similar t o that of PAV residue made in lab, where a second stress peak was present except the recovered PG 76 22 has more variance, which will be stated later Both recovered polymer modified binder PG 76 energy was high compared to unmodified binders and rubber modified binders (Figure 5 19) The second stress peak and high fracture energy identify the existence of polymer in binder. Even for the same type of recovered binder, in different conditions (location, mixture, layer, etc.) its propert y is different. Therefore, the appropriate loading rate range is also different. This difference sometimes is so huge that i t is impossible to define a n appropriate loading rate range for a certain type of recovered binder. On the other hand, f or e ach type of binder in a specific condition, there is a n appropriate loading rate range which results in consistent fracture energies It is usually a range which is lower than the fastest loading rate creating mature fracture and often includes this f astest loading rate In order to properly determine fracture energy, it is necessary to find out this appropriate loading rate range The detailed protocol will be discussed later.

PAGE 65

65 5 3 Tests on Hybrid Binders and Highly Polymer Modified Binder The previou s tests showed that the new DT test and interpretation system is able to clearly differentiate between unmodified binders, polymer modified binder and rubber modified binders. In order to further verify and evaluate the capability of the new DT test to di stinguish between relatively complicated modified binders in terms of fracture resistence, 3 types of hybrid binders : Wright Hudson Geotech 1 type of highly polymer modified binder PG 82 22 and 1 type of rubber modified binder Marianni were tested and analyzed. The components of Marianni are unclear. It was said that Marianni contains 13% rubber modifier. However, its true stress true strain curve shows that it may contain polymer modifier Stress The original binders were prepared by FDOT. Then the PAV was performed on them. All the DT tests were conducted on PAV residues. T he properties of hybrid binders and highly polymer modified binder are different from those of or dinary modified binders I t has been found that it is difficult for all test methods in current specifications to effectively distinguish between hybrid binders and other modified binders in terms of fracture resistance, particularly when the difference is small. Just because of this reason, whether or not the new DT test can differentiate between them is the convincing evidence of its capability to identify fracture resistance The test results showed that the difference of fracture energy between them is significant. A statistical analysis will further prove it later in Chapter 6.

PAGE 66

66 5.3.1 Testing Method Tests were run on MTS machine. Test temperature. 10 C and 15 C were used. Hybrid binders and Marianni : a ll tests were c onducted at 15 C because it is th e optimal temperature for most types of binder to get consistent fracture energies. PG 82 22: a PG 82 22 binder was tested at both 15 C and 10 C Due to the very ductile nature of this highly polymer modified binder PG 82 22, it was impossible to perfor m the new DT test successfully at 15 C, where the deformations to failure were excessive even at very fast loading rates. The excessively d uctile specimen shape may ma k e the interpretation inaccurate Loading rates. V arious loading rates were used depen ding on the property of individual binders in order to avoid both premature fracture, which results in erroneous results, and excessively ductile fracture which result s in the und erestima tion of fracture energy. For hybrid binders and Marianni we applied the loading rates up to the possible fastest rate creating mature fracture. For highly polymer modified binder PG 82 22, the situation of premature fracture is uncommon and complicated, which will be shown and discussed later. Due to this reason, a broad range of loading rates were applied to PG 82 22 in order to find out the method to deal with this p r oblem Binder types. All the binders are PAV residues. 3 types of hybrid binders: 1. Wright (GTR and SBS. Unknown contents) 2. Hudson (3.5%crumb rubber+ 2 .5%SBS+ 0.4% plus Link PT 743 Cross Linking)

PAGE 67

67 3. Geotech (1% SBS (approximately 30 mesh, incorporated dry) +8% of Type B GTR+1% hydrocarbon) 1 type of rubber modified binder: Marianni (Components are unclear. Maybe 13 % Tire Rubber The true stress true strain curve sh ows that it may contain polymer.) 1 type of highly polymer modified binder: PG 82 22 ( 8.5% SBS ) 5.3. 2 Test Result and Analysis For the convenience of comparison, the average of test results of all types of binder was calculated and plotted in Figure 5 20 We will also analyze binders individually. Figure 5 fracture energy 5. 3 .2 .1 Hybrid binders From Figure 5 20, we can see that a ll hybrid binders had fracture energy higher than that of unmodified binders and comparable to SBS polyme r modified binder PG 76 22. Two hybrid binders, Wright and Hudson, exhibited higher fracture energy than that of SBS modified binder PG 76 22.

PAGE 68

68 The f racture e nergy versus loading rate is shown in Figure 5 2 1 Please see the appendix for the complete result s of fracture energy loading rate extension. From Figure 5 21, we can see that for each type of hybrid binder, the fracture energy is consistent at different loading rates. The difference between different hybrid binders is clear. Later, statistical analys es will be performed to further prove the points. Figure 5 2 1 Hybrid binders, fracture energy vs. loading rate It is interesting to analyze the true stress vs. true strain curve of hybrid binders because they have both polyme r and rubber. Although the true stress vs. true strain curve of some types of hybrid binder is special and sometimes complicated, we still can find trend in it In previous tests, we have found that the second stress peak and high fracture energy identify the existence of polymer in binder. This can also be seen in hybrid bin ders From Figure 5 23, we can see that the second stress peak of Geotech is very clear.

PAGE 69

69 Figure 5 22. True stress vs. true strain, Geotech hybrid binder at 100 mm/min Figure 5 23. True stress vs. true strain, Geotech hybrid binder at 200 mm/min Some types of hybrid binder may not exhibit a second stress peak at low loading rates. But when we increase the loading rate, the second stress peak will be pr esent. Figure 5 22 shows the true stress vs. true strain curve of Geotech hybrid binder at 100 mm/min There is only an inflection instead of the second stress peak, which looks similar to that of rubber modified binders. When we increased loading rate to 200 mm/min, the fastest loading rate creating mature fracture, the second stress peak became clear (Figure 5 23), which indicated the existence of polymer.

PAGE 70

70 Some types of hybrid binder may exhibit a more complicated true stress vs. true strain curve than t hat of polymer modified binders since they also include rubber Figure 5 24 shows the true stress vs. true strain curve of Wright hybrid binder at 500 mm/min. There are some tiny waves on the curve, which is different from other binders. However, the seco nd stress peak is significant, which reveals the presence of polymer in binder. Figure 5 2 4 True stress vs. true strain, Wright hybrid binder at 500 mm/min Figure 5 2 5 shows the true stress vs. true strain curve of Hudson hybrid binder at 500 mm/mi n. It exhibits a significantly lower second stress peak than the first peak, but the second stress peak is still clear.

PAGE 71

71 Figure 5 2 5 True stress vs. true strain, Hudson hybrid binder at 500 mm/min 5. 3 .2 .2 Marianni Figure 5 26 shows the test result s of Marianni. In order to compare between rubber modified binders, ARB 5 and ARB 12 PAV residues are also plotted here. Figure 5 26. Rubber modified binders, true stress vs. true strain From Figure 5 26, we can see that the f racture energy of Marianni is similar to that of ARB 5 and ARB 12, which is lower than SBS modified and hybrid binder, but higher than unmodified binders. The fracture energy is consistent at different loading rates.

PAGE 72

72 Figure 5 2 7 shows the clear second stre ss peak of Marianni hybrid binder at 100 mm/min. At the fastest loading rate of 225 mm/min which create mature fracture, the first stress peak was stretched straight, however, the second stress peak is more evident (Figure 5 2 8 ). The clear second stress p eak usually indicates the existence of polymer. Thus, we doubt the component information provided to us, and it is possible that there is polymer modifier in it. Figure 5 2 7 True stress vs. true strain, Marianni hybrid binder at 100 mm/min (PAV residu e) Figure 5 2 8 True stress vs. true strain, Marianni hybrid binder at 225 mm/min (PAV residue)

PAGE 73

73 5. 3 .2 .3 H ighly polymer modified binder PG 82 22 The properties of h ighly polymer modified binder PG 82 22 are very special and quite different from other bi nders. Just because of this reason, t he analysis of PG 82 22 helped us clarify some important concepts such as for viscoelastic polymer materials determine the appropriate test pro tocol and improve the interpretation method. In the be ginning, PG 82 22 binder was tested at 15 C which is same to other binders However, the extens ions to failure were excessive even at very fast loading rates. For all the three specimens (at 500 mm/min, 700 mm/min, 900 mm/min) with complete true stress vs. true strain curve the extension exceeded the upper limit of 1.2 inch without fracture (Figure 5 2 9 ). However, the fracture point i.e. the second stress peak was reached. Thus the true stress vs. true strain curve is complete (Figure 5 30 ). For all othe r specimens (at 800 mm/min, 1000 mm/min through 1400 mm/min), although the fracture failure happened, the second stress peak was not reached. Thus the true stress vs. true strain curve is incomplete (Figure 5 31 ). Figure 5 2 9 Extension greater than 1.2 in without fracture, PG 82 22 at 15 C

PAGE 74

74 Figure 5 30 True stress vs. true strain, PG 82 22 at 15 C, 900 mm/min Figure 5 31 True stress vs. true strain, PG 82 22 at 15 C, 8 00 mm/min In Table 5 4 o nly the t est results with complete true stress vs. true strain curve at 15 C are present We can see that the extensions to fracture point are all over 1 inch, which is too long. On the other hand, their fracture energy is consistent (Figure 5 32) The excessively ductile s pecimen shape may make the interpretation inaccurate. Due to the very ductile nature of this highly polymer modified binder PG 82 22, test temperature was then reduced to 10 C In Table 5 5 only the test results with complete true stress vs. true strain curve at 10 C are present The fracture energy is still independent of loading rate (Figure 5 32)

PAGE 75

75 The extension to fracture point is more reasonable compared to that at 15 C. Therefore, 10 C is the optimal test temperature for highly polymer modified b inder PG 82 22 Table 5 4 PG 82 22 f racture e nergy at 15 o C Table 5 5 PG 82 22 f racture e nergy at 1 0 o C Figure 5 32. PG 82 22, fracture energy vs. loading rate Loading r ate (mm/min) Fracture e nergy (psi) Extension (in) 500 1620.86 1.0912 700 1696.07 1.0698 900 1574.74 1.0600 Loading r ate (mm/min) Fr a cture e nergy (psi) Extension (in) 100 1670.18 1.1140 200 1621.59 0.9118 300 1602.03 0.7564 400 1641.40 0.7509 500 1714.18 0.7421 700 1665.19 0.6684

PAGE 76

76 From Figure 5 20, w e can clearly see that the highly polymer modified binder PG 82 22 had significantly higher fracture energy than unmodified, rubber modified, SBS polymer modified, and hybrid binders. It is interesting that even at a very low loading rate, with a c omplete true stress vs. true strain curve, the fracture section of PG 82 22 always looks lik e a premature fracture (Figure 5 33 A, B D ). It never exhibits a typical mature fracture section shape of other binders (Figure 5 33 C).This phenomenon helped to c larify the concept of so called premature A B C D Figure 5 33. Comparison of fracture section between PG 82 22 and other binders. A) PG 82 22, mature fracture. B) F racture section of PG 82 22, mature fracture. C) A typical mat ure fracture section of other binders (PG 76 22 in picture). D) A typical premature fracture section of other binders (PG 76 22 in picture).

PAGE 77

77 Actually, when we determine whether a fracture is premature the only evidence should be the true stress vs. true s train curve, but not the fracture section shape. A complete true stress vs. true strain curve indicates a mature fracture, otherwise it is a premature fracture. A flat fracture section does not necessarily mean it is a premature fracture. Therefore, we sho uld only use data with complete true stress vs. true strain curve. The determination of complete true stress vs. true strain curve is below. More True 1. For unmod ified binder, a single stress peak should be reached prior to fracture failure. 2. For rubber modified binder, a single stress peak followed by an inflection (instead of the second stress peak of polymer modified binder) should be reached prior to fracture f ailure. 3. For polymer modified or hybrid binder, the second stress peak should be reached prior to fracture failure. Sometimes the first stress peak is stretched straight, or tiny waves may occur, but the second stress peak always exists. Sometimes at low l oading rate, the second stress peak is hidden like the inflection of rubber modified binder, and a higher loading rate can make it present. The second stress peak and high fracture energy identify the existence of polymer in binder The accurate complete true stress vs. true strain curve usually results from an acceptable range of extension and loading rate. The reasonable extension range is between 0.05 to 1 inch. In any conditions, the loading rate should not be greater than 900 mm/min. This may have to be achieved by reducing the test temperature. 5 4 Summary To sum up, the new Direct Tension test distinguished between various modified binders and unmodified binders clearly. F rom Figure 5 2 0 we can see that t he highly

PAGE 78

78 polymer modified binder PG 82 2 2 had significantly greater fracture energy than unmodified, SBS modified, rubber modified and hybrid binders. All hybrid binders had fracture energy higher than that of unmodified binders and comparable to SBS modified binder PG 76 22. Two hybrid binders, Wright and Hudson, exhibited higher fracture energy than that of SBS modified binder PG 76 22. The rubber modified binders had fracture energy greater than that of unmodified binders, but lower than that of other modified binders. For each type of binder, the fracture energy is consistent regardless of different loading rates. The statistical analyses will be conducted to further test the points.

PAGE 79

79 CHAPTER 6 STATISTICAL ANALYSIS I n order to further test whether fracture energy is independent of loading rate and temperature for a binder and whether the new Direct Tension test significantly differentiate d between binders, statistical analyses were conducted. T he data distribution of fracture energy is unbalanced because we only need ductile fracture thus the appropriate loading rates for different binders at different temperatures are quite different. For this reason, the general linear model was used for analysis of variance In addition, the number of data at each combination of temperature and loading rate sometimes is also different Therefore the Type III sums of squares were selected for the interpret ation of output in this unbalanced case 6 1 Consistency of Fracture Energy W e have seen the consistency of fracture energy in the previous chapter. In orde r to further test whether the fracture energy is independent of temperature and loading rate, the two way analysis of variance w as performed for the binders which were tested at different temperatures and loading rates. 6 1 1 Two way A nalysis of Variance 6.1.1.1 PG 67 22 PAV residue The statistical model used is a quadratic equation as shown below. Actually, a cubic regression model was also conduct ed, and the conclusion was the same. (6 1) Where: e fracture energy d ensity t temperature l loading rate

PAGE 80

80 The key statistical analysis result s are in Table 6 1: Table 6 1. Key statistical analysis result s of PG 67 22 Source DF F v alue Pr > F Model 5 1.65 0. 2068 R Square 0.355116 Type III sums of squares Source F v alue Pr > F t 1 0.24 0.6330 l 1 0.08 0.7747 t*t 1 0.00 0.9936 l*l 1 0.31 0.5886 t*l 1 0.56 0.4660 From Table 6 1, we can see that according to a significance level of 5% ( =0.0 5) the overall F test is not significant ( F =1.65, p =0.2068), which means the whole m odel amount of variation in e, i.e. fracture energy density. R 2 =0.36). The t*l interaction in the Type III sums of squares is not significant ( F = 0.56 p = 0.466 ) indicating that the effect s of t and l are independent from each other, which means the tests for the individual effects are valid The effects of t, l, t 2 and l 2 in the Type III sums of squares are not significant ( t: F = 0.24 p = 0.633; l: F = 0.08 p = 0.7747; l 2 : F = 0.24 p = 0.633; t 2 : F = 0.24 p = 0.633 ) In conclusion, for PG 67 22 PAV residue, the fracture energy is independent of temperature and loading rate in a certain range. 6.1.1.2 PG 82 22 PAV residue T he statistical model used is a quadratic equation as shown below. Because t has only two levels, the degree of freedom of t 2 is 0. Therefore, the quadratic term of t is not

PAGE 81

81 included in this regression model. Actually, a cubic regression model was also cond ucted, and the conclusion was the same. (6 2 ) Where: e fracture energy density t temperature l loading rate The key statistical a nalysis results are in Table 6 2 : Table 6 2 Key stat istical analysis result s of PG 82 22 Source DF F v alue Pr > F Model 4 0.24 0.8999 R Square 0.195954 Type III sums of squares Source F v alue Pr > F t 1 0.04 0.8602 l 1 0.44 0.5413 l*l 1 0.03 0.8672 t*l 1 0.08 0.7909 From Table 6 2 we can see that according to a significance level of 5% ( =0.05) the overall F t est is not significant ( F =0.24, p =0.8999), which means the whole model R 2 =0.196). The t*l interaction in the Type III sums of squares is not significant ( F = 0.08 p = 0.7909 ) indicating that the effects of t and l are independent from each other, which means the tests for the individual effects are valid The effects of t, l, t 2 and l 2 in the Type III sums of squares are not significant ( t: F = 0.04 p = 0.8602; l: F = 0.44 p = 0.5413; l 2 : F = 0.03 p = 0.8672 )

PAGE 82

82 In conclusion, for PG 82 22 PAV residue, the fracture energy is independent of temperature and loading rate in a certain range. 6 1 2 One way A nalysis of Variance For the binders which were test ed at only one temperature level 15 C, but have many specimens tested at multiple loading rates, we performed one way analysis of variance to test the effect of loading rate on fracture energy. 6.1.2.1 AC 20 recovered from f ield The statistical regressio n model used is shown as below. Actually, a cubic regression model was also conducted, and the conclusion was the same. (6 3 ) Where: e fracture energy density l loading rate The key statistical analysis results are in Table 6 3 : Table 6 3 Key statistical analysis result of AC 20 Source DF F v alue Pr > F Model 2 2.70 0. 0783 R Square 0. 111741 Type III sums of squares Source F v alue Pr > F l 1 3.44 0. 0704 l*l 1 2.94 0.0 934 From Table 6 3 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F = 2.7 p =0. 0783 ), which means the whole model

PAGE 83

83 R 2 =0. 11 ). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 3.44 p = 0 0704 ; l 2 : F = 2.94 p = 0. 0934 ) In conclusion, for AC 20 recovered from field, the fracture energy is independent of loading rate in a certain range. 6.1.2.2 AC 3 0 recovered from field The statistical regression model used is shown as below. Actually, a cub ic regression model was also conducted, and the conclusion was the same. (6 4 ) Where: e fracture energy density l loading rate The key statistical analysis results are in Table 6 4 : Table 6 4 Key statistical analysis result of AC 3 0 Source DF F v alue Pr > F Model 2 0.75 0. 4773 R Square 0. 023574 Type III sums of squares Source F v alue Pr > F l 1 1.17 0. 2833 l*l 1 0.99 0. 3238 From Table 6 4 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F = 0 .7 5 p =0 4773 ), which means the whole model R 2 =0. 023574 ). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 1.17 p = 0. 2833 ; l 2 : F = 0.99 p = 0. 32 38 )

PAGE 84

84 In conclusion, for AC 3 0 recovered from field, the fracture energy is independent of loading rate in a certain range. 6.1.2.3 PG 64 22 recovered from field The statistical regression model used is shown as below. (6 5 ) Where: e fracture energy density l loading rate The key statistical analysis results are in Table 6 5 : Table 6 5 Key statistical analysis result of PG 64 22 Source DF F v alue Pr > F Model 2 0.19 0.8243 R Square 0.009161 Type III sums of squares Source F v alue Pr > F l 1 0.29 0.5906 l*l 1 0.37 0.5474 From Table 6 5 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F =0.19, p =0.8243), which mea ns the whole model R 2 =0.009). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 0.29 p = 0. 5906 ; l 2 : F = 0. 37 p = 0. 5474 ) In conclusion, for PG 64 22 recovered from field, the fracture energy is independent of loading rate in a certain range.

PAGE 85

85 6.1.2.4 ARB 5 recovered from field The statistical regression model used is shown as below. (6 6 ) Where: e fracture energy density l loading rate The key statistical analysis results are in Table 6 6 : Table 6 6 Key statistical analysis result of ARB 5 recovered from field Source DF F v alue Pr > F Model 2 1.8 0 0.1820 R Square 0.107384 Type III sums of squares Source F v alue Pr > F l 1 0.63 0.4335 l*l 1 1.31 0.2618 From Table 6 6 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F =1.8, p =0. 182), which means the whole model R 2 =0.107). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 0.63 p = 0 .4335; l 2 : F = 1.31 p = 0.2618 ) In conclusion, for ARB 5 recovered from field, the fracture energy is independent of loading rate in a certain range. 6.1.2.5 PG 76 22 recovered from field The statistical regression model used is shown as below. (6 7 ) Where:

PAGE 86

86 e fracture energy density l loading rate The key statistical analysis results are in Table 6 7 : Table 6 7 Key statistical analysis result of PG 76 22 recovered from field Source DF F v alue P r > F Model 2 0.89 0.4223 R Square 0.052458 Type III sums of squares Source F v alue Pr > F l 1 0.05 0.8214 l*l 1 0.01 0.9112 From Table 6 7 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F = 0 .8 9 p =0 4223 ), which means the whole model ( R 2 =0. 05 ). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 0. 05 p = 0. 8214 ; l 2 : F = 0.0 1 p = 0. 9112 ) I n conclusion, for PG 76 22 recovered from field, the fracture energy is independent of loading rate in a certain range. 6.1.2.6 PG 76 22 at 15 C (both recovered and PAV residue) The statistical regression model used is shown as below. (6 8 ) Where: e fracture energy density l loading rate The key statistical analysis results are in Table 6 8 :

PAGE 87

87 Table 6 8 Key statistical analysis result of all PG 76 22 at 15 C Source DF F v alue Pr > F Model 2 1.49 0.2375 R Square 0.066167 Type III sums of squares Source F v alue Pr > F l 1 0.16 0.6877 l*l 1 0.00 0.9944 From Table 6 8 we can see that according to a significance level of 5% ( =0.05) the overall F test is not significant ( F = 1.49 p =0 2375 ), which means the whole model ( R 2 =0. 066 ). The effects of l and l 2 in the Type III sums of squares are not significant ( l: F = 0. 16, p = 0. 6877 ; l 2 : F =0.00, p = 0. 9944) In conclusion, for all PG 76 22 at 15 C the fracture energy is independent of loading rate in a certain range. 6 2 Significance of Difference of Fracture Energy between Binders I n the previous chapter w e have seen th at the new Direct Tension test diffe rentiated between binders clearly in terms of fracture energy. In order to further test it by statistical analysis, the three way and two way analys e s of variance w ere performed for various types of binder.

PAGE 88

88 6 2 1 Three way A nalysis of Variance 6.2.1.1 PG 67 22 and PG 76 22 A ccording to our experiences, we have already known that t he fracture resistance of PG 67 22 and PG 76 22 is quite different. T herefore, t he difference between their test results should be statistically significant. B oth PG 67 22 a nd PG 76 22 specimens were tested at multiple temperatures and loading rates. Thus, a three way (materials, temperatures and loading rates) analysis of variance was conducted for the two types of binder. The statistical regression model used is shown as be low. (6 9 ) Where: e fracture energy density i th material j th temperature k th loading rat e interaction between material and temperature interaction between temperature and loading rate interaction between material and loading rate interaction between material, temperature and loading rate i=1, number of levels of material j=1, number of levels of temperature k=1, number of levels of loading rate The key statistical analysi s results are in Table 6 9 : Table 6 9 Key statistical analysis results of PG 67 22 and PG 76 22 Source DF F v alue Pr > F Model 7 9 0.9 5 <.0001 R Square 0. 867791

PAGE 89

89 Table 6 9 (Continued) Type III sums of squares Source F v alue Pr > F m 1 46.89 <.0001 t 1 0.12 0.7322 l 1 1.54 0.2172 t*l 1 0.78 0.3786 l*m 1 0.29 0.5914 t*m 1 0.01 0.9290 t*l*m 1 0.26 0.6139 From Table 6 9 we can see that according to a significance level of 5% ( =0.05) the overall F test is significant ( F =9 0 9 5, p < 0 .0001 ), which means the whole model accounts for a significant amount of variation in e. With R 2 =0. 87 the model fits well and accounts for 87 % of variation in e. The m, l, t interaction s in the Type III sums of squares are not significant ( t*l: F = 0. 78 p = 0. 3786 ; l*m: F = 0.29 p = 0. 5914 ; t*m: F = 0.0 1 p = 0. 929 ; t*l*m: F = 0.26 p = 0. 6139 ) indicating that the effects of m l and t are independent from each other, which means the tests for the individual effects are valid The effect of m in the Type III sum s of squares is significant ( F = 46.89 p < 0.00 0 1 ) The effect s of l and t in the Type III sums of squares are not significant ( l: F = 1.54 p = 0. 2172 ; t: F = 0.12 p = 0. 7322 ) In conclusion, the new Direct Tension test can effectively distinguish between PG 67 22 and PG 76 22 in terms of fracture energy. For these binders, the fracture energy is independent of temperatures and loading rate s evaluated 6 2 2 Two way A nalysis of Variance 6.2.2.1 M odified binders From our former research project, we kn e w that curre nt tests do not differentiate between some modified binders. T he statistical analyses were performed to test whether the new Direct Tension test distinguish ed between them.

PAGE 90

90 M ost modified binders were tested at 15 C and multiple loading rates Therefore, a two way (materials and loading rates) analysis of variance was performed for all modified binders to test the significance of difference between modified binders fracture energy density. PG 82 22 was excluded because its very high fracture energy density will obviously make the difference statistically significant. The statistical regression model used is shown as below. (6 10 ) Where: e fracture energy density i th material j th loading rate interaction between material and loading rate i=1, number of levels of material j=1, number of levels of loading rate The key statistical ana lysis result s are in Table 6 10 : Table 6 10 Key statistical analysis result s of all modified binders exc ept PG 82 22 Source DF F v alue Pr > F Model 1 9 3.8 2 0.000 1 R Square 0.6 22610 Type III sums of squares Source DF F v alue Pr > F m 6 4.78 0.0008 l 7 1.01 0.4375 m*l 6 0.08 0.9978 From Table 6 10 we can see that according to a significance level of 5% ( =0.05) the overall F test is significant ( F =3.8 2 p =0.00 0 1 ), which means the whole model accounts for a significant amount of v ariation in e. With R 2 =0.6 2 the model accounts for

PAGE 91

91 6 2 % of variation in e. The m*l interaction in the Type III sums of squares is not significant ( F = 0.0 8 p = 0.99 78 ) indicating that the effects of m and l are independent from each other, which means the tes ts for the individual effects are valid The effect of m in the Type III sums of squares is significant ( F = 4.7 8 p = 0. 00 08 ) The effect of l in the Type III sums of squares is not significant ( F = 1.01 p = 0. 4375 ) In conclusion, the new Direct Te nsion test distinguish ed between modified binders in terms of fracture energy. For these binders, t he fracture energy i s independent of loading rate s evaluated 6.2.2.2 Hybrid binders and PG 76 22 W e have seen a limited difference of fracture energy between hybrid b inders and SBS modified binder PG 76 22. A two way analysis of variance was further performed for Wright, Hudson, Geotech and SBS modified binder PG 76 22 to test the significance of difference between their fracture energy density The statistical regres sion model used is shown as below. (6 11 ) Where: e fracture energy density i th material j th loading rate interaction between material and loading rate i=1, number of levels of material j=1, number of levels of loading rate The key statistical analysis results are in Table 6 11 :

PAGE 92

92 Table 6 11 Key statistical analysis result s of hyb rid binders and PG 76 22 Source DF F v alue Pr > F Model 1 3 2. 87 0.00 4 6 Type III sums of squares Source DF F v alue Pr > F m 3 4.46 0.0082 l 7 0.96 0.4697 m*l 3 0.13 0.9398 From Table 6 11 we can see that according to a significance level of 5% ( =0.05) the overall F test is significant ( F =2. 87 p =0.00 46 ), which means the whole model accounts for a significant amount of variation in e. The m*l interaction in the Type III sums of squares is not significant ( F = 0. 1 3 p = 0.9 398 ) indicating that the e ffects of m and l are independent from each other, which means the tests for the individual effects are valid The effect of m in the Type III sums of squares is significant ( F = 4.46 p = 0.00 82 ) The effect of l in the Type III sums of squares is not signif icant ( F = 0. 96 p = 0. 4697 ) In conclusion, the new Direct Tension test distinguish ed between various hybrid binders and SBS modified binder PG 76 22 in terms of fracture energy. For these binders, the fracture energy is independent of loading rate s evaluate d As for the contrast, d ue to the limited data of some types of hybrid binder, it is impossible to compare between every two types of modified binder s H owever, both Hudson and PG 76 22 have adequate data to perform contrast. The statistical model is stil l two way, but without the interaction since it is not significant. The key statistical analysis results of the contrast are in Table 6 1 2

PAGE 93

93 Table 6 1 2 Key statistical analysis results of contrast between Hudson and PG 76 22 Contrast DF F v alue Pr > F H udson vs. PG 76 22 1 5.17 0.027 5 From Table 6 1 2 we can see that according to a significance level of 5% ( =0.05) the F test is significant ( F = 5.1 7 p =0. 027 5 ), which means the difference of e between Hudson and PG 76 22 is significant. Furthermore it is clear that compared to other hybrid binders, Hudson is the closest to PG 76 22. T herefore, the difference between each type of hybrid binder and PG 76 22 is clear. 6.2.2. 3 Unmodified binders All unmodified binders fracture energy density looks ver y close. Are they actually different? I n other words, did the new Direct Tension test d ifferentiate between them? F or this purpose a two way analysis of variance was performed for all unmodified binders The statistical regression model used is shown as below. (6 1 2 ) Where: e fracture energy density i th material j th loading rate interaction between mater ial and loading rate i=1, number of levels of material j=1, number of levels of loading rate The key statistical analysis results are in Table 6 1 3 :

PAGE 94

94 Table 6 1 3 Key statistical analysis results of unmodified binders Source DF F v alue Pr > F Model 21 4.09 < 0.00 01 Type III sums of squares Source F v alue Pr > F m 3 9.33 < 0.00 01 l 11 0. 77 0. 6730 m*l 7 0. 70 0. 6684 From Table 6 1 3 we can see that according to a significance level of 5% ( =0.05) the overall F test is sig nificant ( F = 4.09 p < 0.00 01 ), which means the whole model accounts for a significant amount of variation in e. The m*l interaction in the Type III sums of squares is not significant ( F = 0. 7 p = 0. 6684 ) indicating that the effects of m and l are independent f rom each other, which means the tests for the individual effects are valid The effect of m in the Type III sums of squares is significant ( F = 9.33 p < 0.00 01 ) The effect of l in the Type III sums of squares is not significant ( F = 0. 77 p = 0. 673 ) In conclu sion, the new Direct Tension test di fferentiated between unmodified binders in terms of fracture energy. For these binders, the fracture energy is independent of loading rate s evaluated A s for the contrast, t he statistical model is still two way, but with out the interaction since it is not significant. The key statistical analysis results of the contrast are in Table 6 1 4 From Table 6 14, we can see that according to a significance level of 5% ( =0.05) the F test is significant for AC 20 vs. AC 30 ( F = 35.3 p < 0.0 001 ), AC 20 vs. PG 67 22 ( F = 6.12 p = 0.0 145 ), AC 30 vs. PG 64 22 ( F = 5.4 p = 0.0 215 ), AC 30 vs. PG 67 22 ( F = 26.74 p <0.0001 ), PG 64 22 vs. PG 67 22 ( F = 9.94 p = 0.0 02 ), which means the di fference of e between them is significant. T he F test is not significant for AC 20 vs. PG

PAGE 95

95 64 22 ( F = 0.09 p =0.7593 ) which means the difference of e between them is not significant. Table 6 1 4 Key statistical analysis results of contrast between unmodifi ed binders Contrast DF F v alue Pr > F AC 20 vs. AC 30 1 35 .30 <.0001 AC 20 vs. PG 64 22 1 0.09 0.75 93 AC 20 vs. PG 67 22 1 6.12 0.01 45 AC 30 vs. PG 64 22 1 5. 40 0.02 1 5 AC 30 vs. PG 67 22 1 2 6.74 <.0001 PG 64 22 vs. PG 67 22 1 9. 94 0.002 0 The statis tical analysis clearly exhibits that the new Direct Tension test identif ied the tiny difference of fracture energy between unmodified binders. 6 3 Summary All the statistical analyses showed that the new Direct Tension test significantly differentiated be tween binders in terms of fracture energy. Most statistical analyses showed that for the same binder, the fracture energy is independent of loading rate s and temperature evaluated This is an important finding. It indicates that fracture energy is a funda mental property of binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate.

PAGE 96

96 CHAPTER 7 CHARACTERISTIC TRUE STRESS TRUE STRAIN CURVES Another important finding in this research pr oject is the characteristic true stress true strain curve of binders. We found that the shape of true stress true strain curve is closely related to the modifier type and content. Each type of binder has its own characteristic true stress true strain curve Therefore, i n many cases the true stress true strain curve can be used to identify the binder type, existence of modifier s modifier s type and their relative content. In Chapter 5, we have introduced the typical characteristic true stress true strain c urve of various types of binder. Now we will completely investigate all true stress true strain curve of each type of binder, and compare between them to find out the detailed trends which are related to the modifier types and relative contents. 7 1 Typic al True Stress True Strain Curve of Each Type of Binder I n order to find out the typical true stress true strain curve, for each type of binder, the true stress true strain curves at different loading rates were plotted F rom these figures we can see that for a certain type of binder, its true stress true strain curves exhibit similar pattern and shape. Most types of binder were tested at 15 C. So if not mentioned, the figures refer to the tests at 15 C. 7.1.1 Unmodified Binders T o tally four types of unmodified binder were tested, including AC 20, AC 30 and PG 64 22 recovered from field, and PG 67 22 PAV residue stress true strain curves exhibit only one stress peak. Actually, it is the first stress peak

PAGE 97

97 different iated fro m the second stress peak of true stress true strain curve of binders with polymer, which will be stated later. 7.1.1.1 PG 67 22 PAV residue Figure 7 1 PG 6 7 22 PAV residue true stress vs. true strain 15 C PG 67 22 was tes ted at multiple temperatures. Due to the similar shape, only the true stress true strain curves at 15 C were plotted in Figure 7 1 s ince 15 C is the appropriate test temperature for PG 67 22 We can see that all these curves have only one stress peak, whi ch is characteristic of unmodified binder. A higher loading rate makes the curve a little bit higher and shorter. H owever, fracture energy is very consistent, and t he curves shape re mains similar. 7.1.1.2 AC 30 recovered from field A lot of AC 30 specim ens were tested. D ue to the similar shape only one true stress true strain curve at each loading rate w as plotted in Figure 7 2. Again, all these curves exhibit only one stress peak, which is characteristic of unmodified binder.

PAGE 98

98 Figure 7 2 AC 30 recovered true stress vs. true strain T he curves shape is similar for AC 30 specimens in different locations and at different loading rates Actually, we will repeatedly see the consistent shape of true stress true strain curve s o f each type of unmodified binder regardless of the complex conditions in field Their fracture energy is very consistent although there is variance in the length and height of curves. 7.1.1.3 AC 20 recovered from field Numerous AC 20 specimens were tested Because of the similar shape, only one true stress true strain curve at each loading rate was plotted in Figure 7 3. Again, all these curves have only one stress peak, which is characteristic of unmodified binder. The curves shape is similar for AC 20 s pecimens in different locations and at different loading rates. Their fracture energy is very consistent although there is variance in the length and height of curves.

PAGE 99

99 Figure 7 3 AC 2 0 recovered true stress vs. true strain 7.1.1.4 PG 64 22 recovered from field Figure 7 4 PG 64 22 recovered true stress vs. true strain

PAGE 100

100 Similarly, only one true stress true strain curve at each loading rate was plotted in Figure 7 4 Again, all these curves exh ibit only one stress peak, which is characteristic of unmodified binder. T here is variance in shape due to different loading rate s and ductility of individual binder However, t he curves pattern remains similar and the fracture energy remains consistent 7.1.1.5 Comparison between unmodified binders Figure 7 5 Unmodified binders true stress vs. true strain The typical true stress true strain curve of each type of unmodified binder was plotted together in Figure 7 5 W e ca n see that the characteristic true stress true strain curve of different unmodified binders is similar where o nly the first stress peak is present A single first stress peak and a low fracture energy identify the unmodified binders.

PAGE 101

101 7.1.2 SBS Polymer mo dified b inders We tested three types of SBS modified binder including PG 76 22 recovered from field, PG 76 22 and PG 82 22 PAV residue. The typical true stress true strain curve of SBS modified binder exhibits a second stress peak with a high fracture ener gy. 7.1.2.1 PG 76 22 recovered from field Due to the complicated field conditions such as various materials used different locations and aging levels, the true stress true strain curves of PG 76 22 recovered from field exhibit various shapes. However, mo st of them have t he second stress peak The typical true stress true strain curves of PG 76 22 recovered from field are shown in Figure 7 6. Due to the second stress peak, t heir fracture energy is significantly high compared to unmodified binders. Figure 7 6. PG 76 22 recovered true stress vs. true strain (1)

PAGE 102

102 Figure 7 7. PG 76 22 recovered true stress vs. true strain (2) Sometimes the second stress peak may be low ( Figure 7 7 ) However, the frac ture energy is similar. Figure 7 8. PG 76 22 recovered true stress vs. true strain (3) A few specimens of recovered PG 76 22 exhibit a more significant second stress peak and a longer strain to fracture (Figure 7 8). Conseque ntly, their fracture energy is higher.

PAGE 103

103 Due to the complicated aging effect in field, various materials used and other influence factors, some specimens of recovered PG 76 22 only exhibit an inflection instead of the second stress peak (Figure 7 9) A few specimens even only have the first stress peak like unmodified binder If we take this inflection as fracture point, the fracture energy is still consistent with the average fracture energy of those with a second stress peak. A relatively high fracture ene rgy differentiate these specimens from rubber modified binders which also have an inflection instead of the second stress peak, but with a relatively low fracture energy. On the other hand obviously the actual fracture resistance of inflection is not as g ood as a second stress peak. Figure 7 9. PG 76 22 recovered true stress vs. true strain (4)

PAGE 104

104 7.1.2.2 PG 76 22 PAV residue Figure 7 10 PG 76 22 PAV residue true stress vs. true strain PG 76 22 was tested at multiple temperatures. Due to the similar shape, only the true stress true strain curves at 15 C were plotted in Figure 7 1 0 s ince 15 C is the appropriate test temperature for PG 76 22 PAV residue Compared to PG 76 22 recovered from field PG 76 true strain curve shape is very similar Its fracture energy is very consistent. From Figure 7 10, w e can see that the first stress peaks are stretched straight and become an upward slope, and the second stress peaks ar e very clear. Their fracture energy is high and consistent with the average fracture energy of PG 76 22 recovered from field Although there is only one stress peak here, it is very easy to identify that it is a second stress peak. The difference between the first and second stress peak is that the first stress peak is within strain of 1 (usually around 0.5), but the second stress peak is beyond strain of 1 (usually between 1.2 and 1.8)

PAGE 105

105 7.1.2.3 PG 82 22 PAV residue Figure 7 11 PG 82 22 PAV residue true stre ss vs. true strain The SBS content of PG 82 22 is double that of PG 76 22. From Figure 7 11 we can see that the true stress true strain curve of PG 82 22 is much higher and longer than that of PG 76 22. Therefore, its fracture energy is much greater than that of PG 76 22. Like PG 76 22 PAV residue, the true stress true strain curves of PG 82 22 exhibit a very clear second stress peak, and the first stress peak is stretched and becomes an upward slope. At 10 C, the true stress true strain curves of PG 82 2 2 are a little bit higher and shorter than those at 15 C. However, their shape is similar, and their fracture energy is very consistent.

PAGE 106

106 7.1.2.4 Comparison between polymer modified binders Figure 7 12. Polymer modified bin ders true stress vs. true strain The typical true stress true strain curve of each type of polymer modified binder was plotted together in Figure 7 12. We can see that the true stress true strain curve of PG 82 22 is significantly higher and longer than P G 76 22, and its second stress peak is much more evident. A second stress peak with a high fracture energy identify the polymer modifier in binder. 7.1.3 Rubber modified binders For rubber modified binders, we tested ARB 5 recovered from field, and ARB 5, ARB 12, Marianni PAV residue. As stated previously in Chapter 5, ARB 5 recovered from field lost its rubber when filter ed during recovery process and is actually pure binder. The component of Marianni is doubtable because its true stress true strain curv es exhibit clear second stress peaks, which indicates there may be polymer modifier in Marianni.

PAGE 107

107 The characteristic true stress true strain curves of rubber modified binders have the first stress peak and an inflection instead of the second stress peak. 7.1.3.1 ARB 5 and ARB 12 PAV residue Figure 7 13 ARB 5 and ARB 12 PAV residue true stress vs. true strain From Figure 7 13 we can see that ARB 5 and ARB 12 have the first stress peak and an inflection instead of the second stress peak. This is the typical shape of true stress true strain curve of rubber modified binder. Although we took the inflection as the fracture point to calculate fracture energy, obviously its fracture resistance is not as strong as those with a clear second stress peak. This helps us understand the reason ified binder, but it did not improve the fracture energy of mixtures in previous FDOT/UF research Contrarily, the sig nificant second stress peak of SBS modified binders determines its high fracture resistance. Figure 7 13 also shows that fracture energy is consistent although the curve is higher and shorter at a higher loading rate.

PAGE 108

108 It is interesting that a specimen of A RB 12 which was heated many times also exhibits a second peak (Figure 7 14 ) However, its height is pretty low, and its fracture energy is far smaller than that of normal ARB 12 specimens. Therefore, it is very easy with binders comprising SBS polymer. Figure 7 14 Overheated ARB 12 PAV residue true stress vs. true strain 7.1.3.2 ARB 5 recovered from field Figure 7 15. ARB 5 recovered true stress vs. true strain

PAGE 109

109 Many A RB 5 specimen s were tested. Due to the similar shape, only one true stress true strain curve at each loading rate was plotted in Figure 7 15 Because actually there is no rubber in ARB 5 recovered from field all these curves have only one stress peak, which is charact eristic of unmodified binder. 7.1.3. 3 Marianni PAV residue Figure 7 1 6 Marianni PAV residue true stress vs. true strain We are not sure of the exact components of Marianni. It was said that Marianni was modified with 13% rubber. However, its true stress true strain curves with a second stress peak (Figure 7 1 6 ) showed that it may contain polymer modifier. From Figure 7 1 6 we can also see that a higher loading rate can make the second stress peak more evident.

PAGE 110

110 7.1.3. 4 Comparison between rubber modified binders Figure 7 17. Rubber modified binders true stress vs. true strain The typical true stress true strain curve of each type of rubber modified binder was plotted together in Figure 7 17. To sum up for rubber modified binders like ARB 5 and ARB 12 PAV residues there is a first stress peak and an inflection instead of the second stress peak, while when losing rubber like ARB 5 recovered from field there is only the first stress peak. Thi s result showed that t he new Direct Tension test is able to identify the rubber modifier by characteristic true stress true strain curves 7.1.4 Hybrid binders Hybrid binders are modified with both polymer and rubber. Consequently, their true stress true strain curve has the second stress peak due to the existence of polymer, while at the same time, exhibits diversity depending on the content of polymer and rubber. On the other hand, each type of hybrid binder has similar true stress true strain curve shap es.

PAGE 111

111 7.1.4.1 Wright PAV residue Figure 7 1 8 Wright PAV residue true stress vs. true strain We are not clear about the content of polymer and rubber of Wright. However, its second stress peak is very significant (Figure 7 1 8 ) and its fracture energy is pretty high, which implies that it may contain relatively high content polymer. Its true stress true strain curve shapes are consistent. At 500 mm/min, the curve is a little bit longer and shorter than that at 400 mm/min. Howev er, their fracture energy is very consistent. 7.1.4.2 Hudson PAV residue As shown in Figure 7 19, Hudson exhibited a clear second stress peak due to the existence of polymer. On the other hand, Hudson has 3.5% rubber and 2.5% SBS. Its second stress peak is not as significant as that of PG 76 22 (4.25% SBS). However, the combination of rubber and SBS makes its fracture energy even higher than that of PG 76 22 according to the statistical analysis in Chapter 6. In this sense, Hudson is a

PAGE 112

112 successful hybrid b inder. Its true stress true strain curve shapes are similar, and f racture energy is very consistent. Figure 7 1 9 Hudson PAV residue true stress vs. true strain 7.1.4.3 Geotech PAV residue Figure 7 20 Geotech PAV residue true stress vs. true strain

PAGE 113

113 Geotech has 8% rubber and 1% SBS. I t s second stress peak (Figure 7 20 ) is not as significant as that of PG 76 22 (4.25% SBS). Its fracture energy is less than that of other hybrid binders and SBS mod ified binders. From Figure 7 13, we can also see that sometimes at a low loading rate, its second stress peak became an inflection, and a higher loading rate can make the second stress peak present 7.1.4.4 Comparison between hybrid binders Figure 7 21. Hybrid binders true stress vs. true strain From the typical true stress true strain curves of hybrid binders shown in Figure 7 21, we can see that all hybrid binders exhibited a clear second stress peak, which is characteris tic of polymer modifier. Due to different polymer and rubber contents, the true stress true strain curves of hybrid binders showed different shapes. To sum up, each type of binder has its own characteristic true stress true strain curve, which reflects its modifier type and content. A high content of polymer modifier makes a significant second stress peak a long strain to fracture and a high fracture

PAGE 114

114 energy, while the rubber modifier only makes an inflection instead of the second stress peak and a not much high fracture energy All unmodified binders have only the first stress peak, but not the second stress peak. 7 2 Comparison of True Stress True Strain Curve between Binder s Since each type of binder has its own characteristic true stress true strain cu rve we can compare between them to reveal valuable difference s. First we compared SBS modified PG 76 22 to unmodified binders. To simplify the comparison, we only use PG 76 22 PAV residue at here since its true stress true strain curves are more consis tent than those of recovered PG 76 22 and PG 64 22 at here representing unmodified binders whose curve shape is similar From Figure 7 22 we can see that the true stress true strain curve of PG 76 22 is much longer than that of unmodified binders, and it s second stress peak is clear due to the existence of SBS polymer. These differences result in a much higher fracture energy of PG 76 22 Figure 7 22 PG 76 22 and unmodified binder true stress vs. true strain

PAGE 115

115 As shown in Fig ure 7 23, now we add Hudson. The true stress true strain curve of Hudson (2.5% SBS, 3.5% rubber) is shorter than that of PG 76 22 (4.25% SBS) due to the less SBS, but is higher than that of unmodified binder due to modifiers. Because of this curve shape, t he fracture energy of Hudson is significantly higher than that of unmodified binder, and is even a little higher than that of PG 76 22. Figure 7 23 Comparison with Hudson true stress vs. true strain Figure 7 24 Comparison with Geotech true stress vs. true strain

PAGE 116

116 As shown in Figure 7 24 now Geotech (1% SBS, 8% rubber) is added for comparison. Due to the less SBS, the second stress peak of Geotech is not as significant as that of PG 76 22 and H udson. Its height is also lower. Therefore, its fracture energy is less than that of PG 76 22 and Hudson. Figure 7 25 Comparison with Wright, true stress vs. true strain Now we add Wright for comparison (Figure 7 25 ). Its sec ond stress peak is more significant and its curve is longer than that of PG 76 22. We can estimate that Wright has a relatively high polymer content. Since the true stress true strain curves of ARB 5 and ARB 12 are very similar, we only use ARB 12 at here As shown in Figure 7 26 the rubber in ARB 12 only makes an inflection instead of the second stress peak. Its curve is not high compared to hybrid binders and PG 76 22 and is shorter than PG 76 22 for lack of polymer, which results in a relatively low f racture energy of rubber modified binder. Plus, the fracture resistance of inflection obviously is not as strong as the significant second stress peak such as that

PAGE 117

117 of PG 76 22 and Wright. It is clear that only rubber modifier cannot increase fracture energ y too much It has to be with polymer modifier. Figure 7 26 Comparison with ARB 12 true stress vs. true strain Figure 7 2 7 Comparison with Marianni, true stress vs. true strain

PAGE 118

118 Although it was said that Marianni is rubber modified binder, from its clear second stress peak (Figure 7 27), we suspect that it has polymer modifier Its curve is shorter and lower than that of PG 76 22, which may imply a lower polymer content. Figure 7 2 8 Comparison with PG 82 22, true stress vs. true strain Since the true stress true strain curves of PG 82 22 at 10 C and 15 C are similar, we can only use the curve at 15 C (Figure 7 2 8 ). The comparison is very clear. Due to the h igh SB S content (8.5%), PG 82 22 has a far more significant second stress peak and a much longer curve, which are the typical features of polymer modifier. Its fracture energy is significantly greater than all other binders. 7 3 Summary of Characteristic True S tress True Strain Curves The principles listed below can be used to identify the binder type, modifier type and relative content by characteristic true stress true strain curve 1. Each type of binder has its own characteristic true stress true strain curve. 2. Generally, there are two stress peaks on the true stress true strain curve. T he first stress peak is within strain of 1 (usually around 0.5), while the second stress peak is beyond strain of 1 (usually between 1.2 and 1.8)

PAGE 119

119 3. Unmodified binders only have the first stress peak, but do not have the second stress peak. 4. The binders with polymer modifier ( e.g. SBS modified or hybrid binder) have the second stress peak Sometimes at low loading rates, the second stress peak may become an inflection. A higher loading rate can make the second stress peak present 5. T he binders with polymer modifier may or may not have the first stress peak. S ometimes the first stress peak is stretched straight and becomes an upward slope. 6. true strai n curve is usually longer than that of unmodified and rubber modified binders. A higher polymer content makes the curve longer, and the second stress peak more significant, usually with a high fracture energy 7. The second stress peak and high fracture energ y identify the existence of polymer in binder. 8. The rubber modified binders only have an inflection instead of the second stress peak The first stress peak usually exist s. 9. polyme r modifier The polymer modifier is needed to get a high fracture energy 10. Some hybrid binders (e.g. Hudson) exhibit a higher fracture energy with a lower polymer content compared to polymer modified binders (e.g. PG 76 22) 11. Sometimes the specimen of rubbe r modified binder which i s heated many times also exhibits a second peak. However, its height is pretty low, and its fracture energy is far smaller than that of normal specimens. Therefore, it is very easy to ers comprising SBS polymer. 12. Each type of binder has its own normal range of fracture energy. This range help us identify binders in case the curve shape is abnormal or confusing such as last case 10 7.4 Fracture Energy Determination from the beginning of true stress true strain curve to the last stress peak which is the point of initial fracture. Now with characteristic true stress true strain curves, we will discuss this iss ue further.

PAGE 120

120 Some researchers calculate fracture energy based on the area under the whole true stress true strain curve from beginning to end. Conceptually, this is correct. However, in engineering sense, this may not be appropriate. During extension to fai lure process, after a material loses its major connection, even if the true stress true strain curve is still dropping down, it is meaningless to count in the rest fracture energy. The material structure at the micro level should be involved in to investig ate the final connection within material. This final connection within material can also be reflected by the shape of the last stress peak or inflection of true stress true strain curve. Figure 7 29 Tru e stress vs. true strai n, polymer modified (reduced size) and unmodified binders Figure 7 29 shows the typical shape of true stress true strain curve of polymer and lower than the actual size). Now we assume the area below these two curves is exactly the same. Can we say their fracture resistance is the same? Absolutely the answer should be no. For the unmodified binder, a large part of fracture energy is after stress peak, where the major resist ance has already been lost. For the polymer

PAGE 121

121 modified binder, most fracture energy is before the significant second stress peak, where a strong resistance (connection) still exists. Therefore, going through to the last stress peak to calculate fracture ener gy is more reasonable in engineering sense. Actually, when we talk about fracture resistance, we have to look at both fracture energy and the shape of the last stress peak. We have mentioned previously that although we take the inflection of rubber modifi calculate fracture energy, obviously the resistance of an inflection is not as strong as a second stress peak. Therefore, even if the fracture energy is the same, we can expect the fracture resistance of rubber mo dified binder is lower than those with a significant second stress peak in practice. Figure 7 30 PG 76 22 and Hudson, true stress vs. true strain As shown in Figure 7 30, we assume the fracture energy (to the last stress peak ) of PG 76 22 and Hudson is exactly the same. Although both binders have a second stress peak, the shape of their second stress peak is quite different. Since the shape of

PAGE 122

122 the second stress peak reflects the final connection within material, we can expect their actual fracture resistance is not exactly the same. To sum up, in engineering sense, it is more reasonable to calculate fracture energy through the last stress peak. The fracture resistance depends on both fracture energy and the second stress peak shape.

PAGE 123

123 CHAPTER 8 RECOMMENDED TESTING PRO TOCOL Based on all the binders we tested in this research project, a general testing pro tocol is recommended as below. 8 1 Preparation 1. Prepare at least six specimens for each type of binder according to Method of Test for Determining the Fracture Properties of Asphalt Binder in Direct Tension (D T ), AASHTO Designation: T 31 4 0 2 But don t demold and don t put them in bath liquid. 2. Put specimens with molds in the chamber of MTS to completely cool down to 15 C. 3. The binder specimen is very sensitive to the noise of MTS due to its low strength. Therefore, a small load cell is necessary to reduce the noise. Furthermore, for the original output of force versus displacement, the force has to be increased or decre ased by the amount that the force deviates from zero in the beginning of test These factors may have a great influence on the accuracy of results. 4. Turn on MTS. Make sure it is ready to conduct the Direct Tension test. 5. Quickly take out specimens and carefu lly demold. If the binder specimen is deformed even a little bit during demolding, or the demolding is not smooth due to the lack of release agent somewhere on the mold, discard the specimen. 6. U se only two fingers to loosely hold a load head of specimen to make sure the specimen is vertically and loosely suspended. A ny tiny deviation from this holding manner will deform the specimen easily and the specimen should be discarded. I nsert the load head you hold into the slot of upper load head of testing equipmen t on MTS when keeping the holding manner. If the specimen is deformed even a little bit during this process, discard it. I nsert the small steel bar through the holes of load heads. 7. L ower down the load head of MTS slowly and carefully make the lower load h ead of specimen smoothly enter into the slot of the lower load head of testing equipment on MTS I f this process is unsmooth, it is very possible that the specimen is disturbed and should be discarded. I nsert the small steel bar through the holes of load h eads. 8 2 Testing and Analysis 1. After the specimen cool down to 15 C start test ing at 500 mm/min.

PAGE 124

124 2. I f the fracture is ductile, accept it. And r epeat the procedures above to test specimens at 500 mm/min, 600 mm/min and 700 mm/min. 3. At each loading rate lev el, test at least two replicates. If the difference of fracture energy density between two replicates is within 15% (based on the lower fracture energy density), accept the results and take the average value as the fracture energy density at this loading rate level Otherwise, test more replicates until succeeded. 4. W henever a premature fracture happens, stop testing at current loading rate, then test specimens at 400 mm/min, 300 mm/min, 200 mm/min, 150 mm/min, 100 mm/min, 50 mm/min in turn until getting at least three consecutive loading rate levels at which fracture is not premature If at only two ( 100 mm/min, 50 mm/min ) or one ( 50 mm/min ) loading rates fracture is not premature also accept. If at 50 mm/min, it is still premature fracture, increase test temperature to 20 C then perform step 8 through step 14 5. T he identification of premature fracture is based on the true stres s true strain curve. A n incomplete true stress true strain curve indicate s premature fracture. Please refer to the characteristic true stress true strain curve of each type of binder in this paper for the complete true stress true strain curve. 6. Amon g the three consecutive loading rate levels i f the difference of fracture energy density between every two loading rate levels is within 15% (based on the lower fracture energy density), accept the results, and take the average value as the final fracture energy density. Otherwise, test at more loading rate levels in turn until succeeded. (If at only one loading rate ( 50 mm/min ) fracture i s not premature no need to do this step, and just take its fracture energy density as the final result.) 7. If still failed, test more replicates One may need to find the close results, and identify and discard the deviated results until the difference of f racture energy density between every two replicates is within 15% (based on the lower fracture energy density), then accept the results, and take the average value as the fracture energy density at this loading rate level. Then repeat step 13 until succeed ed. 8. need to be reduced to 10 C, then perform step 8 through step 14. 8 3 Simplified Testing Pro tocol 1. If a quick test is needed, from step 8, the test can be conducted at 500 mm/min 400 mm/min, 300 mm/min, 200 mm/min, 100 mm/min, 50 mm/min in turn. Once two replicates of mature fracture at a single loading rate level are obtained, stop testing. 2. If the difference of fracture energy density between two replicates is within 15% (based on the lower fracture energy density), accept the results, and take the

PAGE 125

125 average value as the final fracture energy density. Otherwise, test more replicates, find the close results, and identify and discard the deviated results until the difference of fractur e energy density between every two replicates is within 15% (based on the lower fracture energy density), then accept the results, and take the average value as the final fracture energy density. The simplified testing pro tocol requires less specimens sin ce it only needs to determine one loading rate level at which the fracture is not premature However, its accuracy is not as good as normal testing pro tocol because we are not 100% sure t his loading rate level is within the appropriate loading rate range f or this binder although in most cases it is. Therefore, the simplified testing pro tocol should only be used when the amount of specimens are limited, or in some special conditions.

PAGE 126

126 CHAPTER 9 CONCLUSIONS 9 1 Summary Based on Finite Element Analysis (FEA) and prototype test on MTS machine, a new D irect T ension test and interpretation system was successfully developed to consistently measure fracture energy of binder. The new Direct Tension test was conducted on a range of binders including unmodified binde rs, SBS modified binder, rubber modified binders, hybrid binders and highly SBS modified binder. The statistical analysis showed that i t significantly distinguish ed between different binders in terms of fracture energy. Expected trends in fracture energy b etween binders were observed, which indicates the test was successful. For the same binder, the fracture energy is independent of loading rate and temperature for the ranges of temperature (0 15 C) and loading rates evaluated in this study This is an imp ortant finding. It indicates that fracture energy is a fundamental property for a certain binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate. Each type of binder has its own ch aracteristic true stress true strain curve, which can be used to identify the binder type, modifier type and relative content. Basic principles are listed as below. 1. Generally, there are two stress peaks on the true stress true strain curve. The first stres s peak is within strain of 1 (usually around 0.5), while the second stress peak is beyond strain of 1 (usually between 1.2 and 1.8) 2. Unmodified binders only have the first stress peak, but do not have the second stress peak.

PAGE 127

127 3. The binders with polymer modifie r (e.g. SBS modified or hybrid binder) have the second stress peak. Sometimes at low loading rates, the second stress peak may become an inflection. A higher loading rate can make the second stress peak present. 4. The binders with polymer modifier may or ma y not have the first stress peak. Sometimes the first stress peak is stretched straight and becomes an upward slope. 5. true strain curve is usually longer than that of unmodified and rubber modified binders. A higher pol ymer content makes the curve longer, and the second stress peak more significant, usually with a high fracture energy. 6. The second stress peak and high fracture energy identify the existence of polymer in binder. 7. The rubber modified binders only have an inf lection instead of the second stress peak. The first stress peak usually exists. 8. polymer modifier. 9. Sometimes the specimen of rubber modified binder which is heated many times al so exhibits a second peak. However, its height is pretty low, and its fracture energy is far smaller than that of normal specimens. Therefore, it is very easy to 10. Each type of binder has its own normal range of fracture energy. This range help us identify binders in case the curve shape is abnormal or confusing such as last case. F racture resistance depends on both fracture energy and the second stress peak shape. Rubber modifier alon e did not appear to improve fracture energy of binder much. The polymer modifier is necessary to get a high fracture energy. Some hybrid binders (e.g. Hudson) exhibit a higher fracture energy with a lower polymer content compared to polymer modified binder s (e.g. PG 76 22). For most binders including unmodified binders, SBS modified binder, rubber modified binders and hybrid binders, 15 C was the optimal temperature to get

PAGE 128

128 consistent fracture energies for the same binder. 10 C is the optimal test tempera ture for highly polymer modified binder PG 82 22. The appropriate loading rate depends on the property of individual binders in order to avoid premature fracture which results in erroneous results, while at the same time, avoid excessively ductile fractur e, which results in the underestimation of fracture energy. When we determine whether a fracture is premature the only evidence is the true stress vs. true strain curve, but not the fracture section shape. A n in complete true stress vs. true strain curve i ndicates a premature fracture. A flat fracture section does not necessarily mean it is a brittle fracture. The premature fracture usually results from flaws in specimen. Therefore, w e should only use data with complete true stress vs. true strain curve. To determin e the complete true stress vs. true strain curve of binders, refer to the characteristic true stress true strain curve The accurate complete true stress vs. true strain curve results from the acceptable test conditions including extension and loa ding rate. The reasonable extension range is between 0.05 to 1 inch. In any conditions, the loading rate should not be fast er than 900 mm/min. This may have to be achieved by reducing the test temperature. Even for the same type of recovered binder, in d ifferent conditions (location, mixture, layer, etc.), its property is different. Therefore, the appropriate loading rate range is also different. This difference sometimes is so huge that it is impossible to define an appropriate loading rate range for a c ertain type of recovered binder. On the other hand, for each type of binder in a specific condition, there is an appropriate loading rate range which results in consistent fracture energies. It is usually a range

PAGE 129

129 which is lower than the fastest loading rat e creating mature fracture, and often includes this fastest loading rate. In order to properly determine fracture energy, it is necessary to find out this appropriate loading rate range. A detailed test pro tocol wa s recommended The highly polymer modifie d binder PG 82 22 had significantly greater fracture energy than unmodified, SBS modified, rubber modified and hybrid binders. All hybrid binders had fracture energy higher than that of unmodified binders and comparable to SBS modified binder PG 76 22. Two hybrid binders, Wright and Hudson, exhibited higher fracture energy than that of SBS modified binder PG 76 22. The rubber modified binders had fracture energy greater than that of unmodified binders, but lower than that of other modified binders. 9 2 Co nclusions Conclusions of this research may be drawn as follows The new binder Direct Tension test and interpretation system was able to consistently test fracture energy of binder Expected trends in fracture energy between binders were observed, which in dicates the test was successful. For the same binder, the fracture energy appears to be independent of loading rate and temperature for the range s of temperature (0 15 C) and loading rates evaluated in this study This indicates that fracture energy is a fundamental property of binder, which does not depend on test condition, and can be determined by tests performed at a single temperature and loading rate. Different types of binder have different characteristic true stress true strain curve s which can be used to identify the binder type, modifier type and relative content

PAGE 130

130 I t is important to determine th e appropriate loading rate range and to i dentif y th e complete true stress vs. true strain curve for accurate determination of fracture energy. A detaile d test protocol was recommended.

PAGE 131

131 APPENDIX A FINITE ELEMENT ANALY SIS (FEA) RESULTS Figure A 1 For t ransforming e xtension to t rue s train Figure A 2 For t ransforming l oad to t rue s tress

PAGE 132

132 Figure A 3 For c alculating a rea of c ross s ection at the fir st stress p eak Figure A 4 For c alculating l ength of 3mm middle p art at the first stress p eak

PAGE 133

133 APPENDIX B PRELI MINARY TEST RESULTS Table B 1. Fracture energy density at 15 C (PAV residue) Binder Loading r ate (mm/min) Fracture energy d ensity (psi) E xtension to f racture (in) PG 76 22 700 660.88 0.4590 PG 76 22 700 591.68 0.3984 PG 76 22 600 629.77 0.4433 PG 76 22 600 657.43 0.4616 PG 76 22 500 676.20 0.4873 PG 76 22 500 617.37 0.4654 PG 76 22 400 609.91 0.4776 PG 76 22 400 563.09 0.466 3 PG 76 22 300 584.18 0.4663 PG 76 22 300 576.03 0.4823 PG 67 22 700 244.10 0.1489 PG 67 22 600 235.05 0.1489 PG 67 22 600 220.70 0.1449 PG 67 22 500 251.31 0.1579 PG 67 22 400 233.65 0.1706 PG 67 22 400 244.32 0.1632 PG 67 22 300 195.30 0.1733 PG 67 22 300 205.61 0.1736 Figure B 1. Fracture e nergy d ensity at 15 C (PAV residue)

PAGE 134

134 Table B 2 Fracture energy density at 1 0 C (PAV residue) Binder Loading r ate (mm/min) Fracture energy d ensity (psi) Extension to f racture (in) PG 76 2 2 1480 598.17 0.2025 PG 76 22 1448 714.78 0.2530 PG 76 22 1445 780.47 0.2448 PG 76 22 985 722.72 0.2900 PG 76 22 600 650.26 0.2565 PG 76 22 600 713.35 0.3028 PG 76 22 500 680.86 0.3214 PG 76 22 500 696.24 0.3302 PG 76 22 400 770.25 0.3579 PG 76 22 400 707.09 0.3284 PG 76 22 200 667.98 0.3911 PG 76 22 200 616.38 0.3454 PG 76 22 100 506.69 0.3976 PG 76 22 100 572.01 0.4408 PG 76 22 50 518.84 0.5033 PG 76 22 50 478.63 0.4659 PG 67 22 600 235.13 0.0981 PG 67 22 400 208.26 0.106 5 PG 67 22 150 246.97 0.1522 PG 67 22 100 210.34 0.1552 PG 67 22 100 206.89 0.1679 PG 67 22 50 168.93 0.1679 PG 67 22 50 168.41 0.1737 PG 67 22 25 221.93 0.2138 PG 67 22 25 238.98 0.2134 Figure B 2. Fracture e nergy d ensity at 1 0 C (PAV residue)

PAGE 135

135 Table B 3 Fracture energy density at 5 C (PAV residue) Binder Loading r ate (mm/min) Fracture e nergy d ensity (psi) Extension to f racture (in) PG 76 22 300 745.88 0.2345 PG 76 22 250 666.28 0.1893 PG 76 22 250 707.36 0.2410 PG 76 22 20 0 770.40 0.2601 PG 76 22 150 757.41 0.2625 PG 76 22 100 579.69 0.3040 PG 76 22 50 651.57 0.3543 PG 76 22 50 608.14 0.3195 PG 76 22 25 546.49 0.3628 PG 76 22 25 560.17 0.3743 PG 67 22 35 238.61 0.1358 PG 67 22 35 234.91 0.1456 PG 67 22 25 243.48 0.1367 PG 67 22 25 238.41 0.1509 PG 67 22 10 171.07 0.1844 PG 67 22 10 147.88 0.1598 PG 67 22 5 138.76 0.1823 PG 67 22 5 131.05 0.1920 Figure B 3. Fracture e nergy d ensity at 5 C (PAV residue)

PAGE 136

136 Figure B 4. Fracture e nergy d ensity a t 5 C (PAV residue) (without loading rates 10 mm/min ) Table B 4 Fracture energy density at 0 C (PAV residue) Binder Loading r ate (mm/min) Fracture energy d ensity (psi) Extension to f racture (in) PG 76 22 100 716.12 0.1884 PG 76 22 30 594.14 0.24 61 PG 76 22 30 612.28 0.2455 PG 76 22 25 605.56 0.2535 PG 76 22 25 612.14 0.2408 PG 76 22 20 606.05 0.2619 PG 76 22 15 600.74 0.2699 PG 76 22 10 424.06 0.2852 PG 76 22 10 397.60 0.2747

PAGE 137

137 Figure B 5. Fracture e nergy d ensity at 0 C (PAV) Figure B 6. Fracture e nergy d ensity at 0 C (PAV residue) (without loading rates 10 mm/min )

PAGE 138

138 Figure B 7. Average fracture energy density at each temperature Figure B 8. PG 67 22 at 10 C, deformation at fracture vs. loading rate

PAGE 139

139 Figure B 9. Before modification Figure B 10. After modification

PAGE 140

140 APPENDIX C TEST RESULTS OF BIND ERS RECOVERED FROM SUPERPAVE SECTIONS Table C 1. Test results of binders recovered from Superpave sections (box 1 box 3) Box No.1 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ens ity Extension (in) 1 5 A WP 19234 AC 30 300 354.82 0.1827 AC 30 500 373.63 0.1693 BWP 19235 AC 30 500 374.21 0.1530 AC 30 500 345.66 0.1507 B WP 19236 AC 30 500 313.02 0.1877 AC 30 500 316.01 0.1828 BWP 19237 AC 30 500 341.04 0.1871 AC 30 500 324.73 0.1856 Box No.2 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 1 15 A WP 19238 AC 30 500 406.01 0.1615 AC 30 500 400.87 0.1499 BWP 19239 AC 30 500 371.94 0.1658 AC 30 500 376.52 0.1745 B WP 19240 AC 30 500 321.51 0.1908 AC 30 500 283.12 0.1762 BWP 19241 AC 30 500 348.64 0.1815 AC 30 500 331.02 0.1903 Box No.3 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 2 5 A WP 19242 AC 20 500 275.85 0.1686 AC 20 700 306.48 0.1712 BWP 19243 AC 20 500 249.98 0.1896 AC 20 500 271.33 0.2074 B WP 19244 AC 20 500 268.67 0.2014 AC 20 500 272.02 0.1999 BWP 19245 AC 20 500 288.01 0.2084 AC 20 500 284.45 0.2001 15 A WP 19246 AC 20 500 337.11 0.1978 AC 20 500 316.71 0.1819 BWP 19247 AC 20 500 266.64 0.1952 AC 20 500 297.24 0.1994

PAGE 141

141 Table C 2. Test results of binders recovered from Superpave sections (box 4 box 6) Box No.4 SMO Lab No. Binder t ype Fracture e nergy d ensity @ 15 C (psi) Project Location Layer Source Lo ading r ate (mm/min) FE d ensity Extension (in) 2 15 B WP 19248 AC 20 500 256.31 0.2171 AC 20 500 234.13 0.2153 BWP 19249 AC 20 500 231.44 0.2223 AC 20 500 243.92 0.2134 25 A WP 19250 AC 20 500 249.17 0.2214 AC 20 500 253.2 0 0.2169 BWP 19251 AC 20 500 274.90 0.2184 AC 20 500 244.38 0.2175 B WP 19252 AC 20 500 220.76 0.2131 AC 20 500 228.14 0.2122 BWP 19253 AC 20 500 258.22 0.2138 AC 20 500 233.79 0.2123 Box No.5 SMO Lab No. Binde r t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 3 5 A WP 19254 AC 30 500 348.12 0.1547 AC 30 500 313.90 0.1443 BWP 19255 AC 30 500 342.96 0.1540 AC 30 500 3 50.69 0.1613 B WP 19256 AC 30 500 324.03 0.1460 AC 30 500 341.37 0.1502 BWP 19257 AC 30 500 264.98 0.1941 AC 30 500 238.36 0.1953 Box No.6 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 3 25 A WP 19258 AC 30 500 332.63 0.1599 AC 30 500 324.34 0.1673 BWP 19259 AC 30 500 292.97 0.1623 AC 30 500 340.23 0.1643 B WP 19260 AC 30 500 317.27 0.1708 AC 30 500 312.13 0.1718 BWP 19261 AC 30 500 354.75 0.1740 AC 30 500 346.96 0.1726 4 5 A WP 19262 AC 30 500 273.75 0.1629 AC 30 500 260.33 0.1458 BWP 19263 AC 30 500 275.65 0.1416 AC 30 500 264.24 0.1407

PAGE 142

142 Table C 3. Test results of binders recovered from Superpave sections (box 7 box 9) Box No.7 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 4 5 B WP 19264 AC 30 500 315.70 0. 1601 AC 30 500 290.19 0.1567 BWP 19265 AC 30 500 274.25 0.1842 AC 30 500 283.35 0.1757 15 A WP 19266 AC 30 500 276.64 0.1487 AC 30 500 277.76 0.1630 BWP 19267 AC 30 500 258.77 0.1741 AC 30 500 220.85 0.1651 B WP 19268 AC 30 500 362.03 0.1584 AC 30 500 340.02 0.1434 BWP 19269 AC 30 500 315.42 0.1677 AC 30 500 290.62 0.1610 Box No.8 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loadin g r ate (mm/min) FE d ensity Extension (in) 4 18 A WP 19270 AC 30 500 275.21 0.1632 AC 30 500 263.98 0.1652 BWP 19271 AC 30 500 286.17 0.1594 AC 30 500 279.80 0.1559 B WP 19272 AC 30 500 307.07 0.1472 AC 30 600 332.70 0.1466 BWP 19273 AC 30 500 212.03 0.1863 AC 30 500 247.83 0.1820 AC 30 500 237.40 0.1898 25 A WP 19274 AC 30 500 306.78 0.1362 AC 30 500 312.64 0.1331 BWP 19275 AC 30 500 267.82 0.1430 AC 30 500 279.98 0 .1460 Box No.9 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 4 25 B WP 19276 AC 30 500 320.08 0.1683 AC 30 500 254.72 0.1587 BWP 19277 AC 30 500 297.74 0.1762 AC 30 500 303.20 0.1757 8 15 A WP 19278 PG 76 22 500 525.79 0.5902 PG 76 22 500 550.19 0.5981 PG 76 22 800 683.74 0.5089 BWP 19279 PG 76 22 500 511.65 0.3847 PG 76 22 500 525.85 0.3632 PG 76 22 800 657.45 0.3321 B WP 19280 PG 76 22 500 649.47 0.5277 PG 76 22 500 659.97 0.5014 PG 76 22 800 679.76 0.4836 BWP 19281 PG 76 22 500 601.64 0.5232 PG 76 22 500 590.34 0.5081

PAGE 143

143 Table C 4. Test results of binders recovered from Superpave sections (box 10 box 12) Box No.10 SMO Lab No. Binder t ype Fracture e nergy d ensity @ 15 C (psi) Project L ocation Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 8 22 A WP 19282 PG 76 22 500 575.93 0.3438 PG 76 22 500 629.64 0.3316 BWP 19283 PG 76 22 500 571.15 0.3358 PG 76 22 500 544.56 0.4006 B WP 19284 PG 76 22 500 850.44 0.8202 PG 76 22 500 799.50 0.8268 BWP 19285 PG 76 22 500 641.59 0.4433 PG 76 22 500 608.92 0.4611 25 A WP 19286 PG 76 22 700 594.14 0.2696 PG 76 22 500 520.96 0.2764 BWP 19287 PG 76 22 500 600.92 0.3161 PG 76 22 500 476.88 0.3786 Box No.11 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 8 25 B WP 19288 PG 76 22 500 527.96 0.3245 PG 76 22 700 613. 15 0.4026 BWP 19289 PG 76 22 500 561.89 0.3315 PG 76 22 500 581.31 0.3326 9 15 A WP 19290 ARB 5 500 296.03 0.0908 ARB 5 300 285.54 0.1002 BWP 19291 ARB 5 ARB 5 B WP 19292 PG 64 22 300 245.60 0.1631 PG 64 22 300 254.12 0.1510 BWP 19293 PG 64 22 300 265.58 0.1390 PG 64 22 300 278.33 0.1508 Box No.12 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extensi on (in) 9 25 A WP 19294 ARB 5 300 266.84 0.1038 ARB 5 300 265.40 0.1028 BWP 19295 ARB 5 300 219.77 0.1103 ARB 5 300 253.30 0.1107 B WP 19296 PG 64 22 300 233.65 0.1471 PG 64 22 300 237.91 0.1464 BWP 19297 PG 64 22 300 247.32 0.1371 PG 64 22 300 257.89 0.1346 10 5 A WP 19298 ARB 5 100 282.47 0.1246 ARB 5 100 288.71 0.1097 ARB 5 100 283.52 0.1176 BWP 19299 ARB 5 50 202.48 0.1431 ARB 5 100 248.56 0.1310

PAGE 144

144 Table C 5. Test resul ts of binders recovered from Superpave sections (box 13 box 15) Box No.13 SMO Lab No. Binder t ype Fracture e nergy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 10 5 B WP 19300 PG 64 22 100 314.91 0.1045 PG 64 22 100 279.50 0.1049 PG 64 22 100 339.88 0.0944 BWP 19301 PG 64 22 100 268.13 0.1383 PG 64 22 100 232.96 0.1234 15 A WP 19302 ARB 5 50 266.60 0.1396 ARB 5 50 310.91 0.1372 BWP 19303 ARB 5 100 265. 36 0.1219 ARB 5 100 251.41 0.1280 B WP 19304 PG 64 22 50 240.04 0.1762 PG 64 22 100 274.72 0.1578 BWP 19305 PG 64 22 100 277.08 0.1104 PG 64 22 100 321.08 0.1272 Box No.14 SMO Lab No. Binder t ype Fracture energy d ensit y @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 10 25 A WP 19306 ARB 5 50 284.18 0.1241 ARB 5 50 256.14 0.1311 BWP 19307 ARB 5 50 233.58 0.1382 ARB 5 50 239.78 0.1381 B WP 19308 PG 64 22 50 226.02 0.1778 PG 64 22 100 306.25 0.1573 BWP 19309 PG 64 22 100 282.12 0.1386 PG 64 22 100 289.48 0.1362 11 15 A WP 19310 PG 76 22 100 506.85 0.4490 PG 76 22 500 695.70 0.3551 BWP 19311 PG 76 22 500 692.23 0.3731 PG 76 22 300 596.53 0.4216 Box No.15 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 11 15 B WP 19312 PG 64 22 300 281.03 0.1609 PG 64 22 300 264.93 0.1633 BWP 19313 PG 64 22 300 263.84 0.1714 PG 64 22 400 256.81 0.1584 25 A WP 19314 PG 76 22 300 525.92 0.5359 PG 76 22 600 727.17 0.4494 BWP 19315 PG 76 22 300 633.70 0.3661 PG 76 22 300 750.84 0. 3456 B WP 19316 PG 64 22 300 302.24 0.1599 PG 64 22 300 282.96 0.1589 BWP 19317 PG 64 22 300 253.51 0.1546 PG 64 22 300 264.61 0.1600

PAGE 145

145 Table C 6. Test results of binders recovered from Superpave sections (box 16 box 18) Box No. 16 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 12 5 A WP 19318 ARB 5 300 266.95 0.1529 ARB 5 300 302.66 0.1524 BWP 19319 ARB 5 100 235.51 0.1591 ARB 5 100 243.76 0.1680 B WP 19320 PG 64 22 100 243.51 0.1410 PG 64 22 200 294.75 0.1293 BWP 19321 PG 64 22 200 277.57 0.1379 PG 64 22 200 292.85 0.1314 10 A WP 19322 ARB 5 200 258.34 0.1742 ARB 5 300 281.65 0.1506 BWP 19323 ARB 5 200 241.28 0.1624 ARB 5 200 260.06 0.1462 Box No.17 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 12 10 B WP 19324 PG 64 22 200 233.55 0.1666 PG 64 22 300 277.37 0.1564 BWP 19325 PG 64 22 200 261.04 0.1422 PG 64 22 200 262.64 0.1437 15 A WP 19326 ARB 5 200 234.44 0.1726 ARB 5 300 289.72 0.1665 BWP 19327 ARB 5 200 223.25 0.1976 ARB 5 400 213.27 0.1740 B WP 19328 PG 64 22 500 292.82 0.1163 PG 64 22 400 302.22 0.1255 BWP 19329 PG 64 22 400 299.07 0.1145 PG 64 22 400 319.25 0.1118 Box No.18 SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 12 25 A WP 19330 ARB 5 400 301.93 0.1633 ARB 5 5 00 330.76 0.1526 BWP 19331 ARB 5 400 298.96 0.1361 ARB 5 400 318.14 0.1333 B WP 19332 PG 64 22 400 256.08 0.1480 PG 64 22 600 267.07 0.1323 BWP 19333 PG 64 22 400 298.16 0.1208 PG 64 22 400 276.36 0.1173

PAGE 146

146 Table C 7. Test results of binders recovered from Superpave sections (new boxes) New boxes SMO Lab No. Binder t ype Fracture energy d ensity @ 15 C (psi) Project Location Layer Source Loading r ate (mm/min) FE d ensity Extension (in) 6 5 A WP 20366 AC 20 AC 20 BWP 20367 AC 20 AC 20 15 A WP 20368 AC 20 800 284.93 0.1488 AC 20 800 300.40 0.1507 BWP 20369 AC 20 800 283.07 0.1318 AC 20 800 302.73 0.1332 25 A WP 20370 AC 20 800 276.92 0.1483 AC 20 1000 263.34 0.1424 BWP 20371 AC 20 1000 256.69 0.1457 AC 20 7 5 A WP 20372 AC 20 1000 290.47 0.1293 AC 20 1000 288.45 0.1289 BWP 20373 AC 20 1000 277.09 0.1133 AC 20 1000 275.36 0.1137 B WP 20374 AC 20 1000 263.55 0.1436 AC 20 1000 253.29 0.1482 BWP 20375 AC 20 1000 330.95 0.0950 AC 20 1000 301.99 0.0966 15 A WP 20468 AC 20 600 265.22 0.1311 AC 20 500 264.70 0.1470 BWP 20469 AC 20 500 282.02 0.1184 AC 20 B WP 20470 AC 20 600 24 0.69 0.1529 AC 20 800 252.10 0.1411 BWP 20471 AC 20 900 267.07 0.1374 AC 20 1200 247.42 0.1273

PAGE 147

147 Figure C 1 PG 76 22 recovered, fracture energy vs. loading rate Fig ure C 2 AC 20 recovered, fracture energy vs. loading rate

PAGE 148

148 Figure C 3 AC 30 recovered, fracture energy vs. loading rate Figure C 4. PG 64 22 recovered, fracture energy vs. loading rate

PAGE 149

149 Figure C 5 PG 76 22, rubber modified and unmodified binder, fracture energy vs. loading rate Figure C 6 True stress vs. true strain, PG 76 22, recovered from field (Superpave #19278)

PAGE 150

150 Figure C 7 True stress vs. tru e strain, PG 64 22, recovered from field (Superpave #19312) Figure C 8 True stress vs. true strain, ARB 5, recovered from field (Superpave #19298) Figure C 9 True stress vs. true strain, ARB 5, PAV r esidue

PAGE 151

151 APPENDIX D TEST RESULTS OF HYBRID BINDERS AND PG 82 22 Figure D 1 Hybrid binders, fracture energy vs. loading rate Figure D 2. Rubber modified binders, true stress vs. true strain

PAGE 152

152 Figure D 3. PG 82 22, fra cture energy vs. loading rate Table D 1 PG 82 22 fracture e nergy at 15 C Table D 2 PG 82 22 f racture e nergy at 1 0 C Loading r ate (mm/min) Fracture e nergy (psi) Extension (in) 500 1620.86 1.0912 700 1696.07 1.0698 900 1574.74 1.0600 Loading r a te (mm/min) Fracture e nergy (psi) Extension (in) 100 1670.18 1.1140 200 1621.59 0.9118 300 1602.03 0.7564 400 1641.40 0.7509 500 1714.18 0.7421 700 1665.19 0.6684

PAGE 153

153 Figure D 4 True stress vs. true strain, Wright hybrid binder at 500 mm/min Figure D 5 True stress vs. true strain, Geotech hybrid binder at 100 mm/min Figure D 6 True stress vs. true strain, Geotech hybrid binder at 200 mm/min

PAGE 154

154 Figure D 7 True stress vs. true strain, Hudson hybrid binder at 500 m m/min Figure D 8 True stress vs. true strain, Maria nni hybrid binder at 100 mm/min Figure D 9 True stress vs. true strain, Maria nni hybrid binder at 225 mm/min

PAGE 155

155 Figure D 1 0. True stress vs. true strain, PG 82 22 at 15 C, 900 mm/min Figure D 1 1. True stress vs. true strain, PG 82 22 at 15 C, 800 mm/min

PAGE 156

156 APPENDIX E TEST RESULTS SUMMARY Figure E 1 fracture energy

PAGE 157

157 APPENDIX F STATISTICAL ANALYSIS RESULTS Table F 1. Key statistical analysis result of PG 67 22 Source DF F v alue Pr > F Model 5 1.65 0. 2068 R Square 0.355116 Type III sums of squares Source F v alue Pr > F t 1 0.24 0.6330 l 1 0.08 0.7747 t*t 1 0.00 0.9936 l*l 1 0.31 0.5886 t*l 1 0.56 0.4660 Table F 2 Key stati stical analysis results of PG 82 22 Source DF F v alue Pr > F Model 4 0.24 0.8999 R Square 0.195954 Type III sums of squares Source F v alue Pr > F t 1 0.04 0.8602 l 1 0.44 0.5413 l*l 1 0.03 0.8672 t*l 1 0.08 0.7909 Table F 3 Ke y statistical analysis result of AC 20 Source DF F v alue Pr > F Model 2 2.70 0. 0783 R Square 0. 111741 Type III sums of squares Source F v alue Pr > F l 1 3.44 0. 0704 l*l 1 2.94 0.0 934

PAGE 158

158 Table F 4 Key statistical analysis result of A C 30 Source DF F v alue Pr > F Model 2 0.75 0. 4773 R Square 0. 023574 Type III sums of squares Source F v alue Pr > F l 1 1.17 0. 2833 l*l 1 0.99 0. 3238 Table F 5 Key statistical analysis result of PG 64 22 Source DF F v alue Pr > F Model 2 0.19 0.8243 R Square 0.009161 Type III sums of squares Source F v alue Pr > F l 1 0.29 0.5906 l*l 1 0.37 0.5474 Table F 6 Key statistical analysis result of ARB 5 recovered from field Source DF F v alue Pr > F Model 2 1.8 0 0.1820 R Square 0.107384 Type III sums of squares Source F v alue Pr > F l 1 0.63 0.4335 l*l 1 1.31 0.2618

PAGE 159

159 Table F 7 Key statistical analysis result of PG 76 22 recovered from field Source DF F v alue Pr > F Model 2 0.89 0.4223 R Square 0.052458 Type III sums of squares Source F v alue Pr > F l 1 0.05 0.8214 l*l 1 0.01 0.9112 Table F 8 Key statistical analysis result of all PG 76 22 at 15 C Source DF F v alue Pr > F Model 2 1.49 0.2375 R Square 0.066167 Type III sums of squares Source F v alue Pr > F l 1 0.16 0.6877 l*l 1 0.00 0.9944 Table F 9 Key statistical analysis results of PG 67 22 and PG 76 22 Source DF F v alue Pr > F Model 7 9 0.9 5 <.0001 R Square 0. 867791 Type III sums of squares Source F v alue Pr > F m 1 46.89 <.0001 t 1 0.12 0.7322 l 1 1.54 0.2172 t*l 1 0.78 0.3786 l*m 1 0.29 0.5914 t*m 1 0.01 0.9290 t*l*m 1 0.26 0.6139

PAGE 160

1 60 Table F 10 Key statistical analysis results of all modified binders e xcept PG 82 22 Source DF F v alue Pr > F Model 1 9 3.8 2 0.000 1 R Square 0.6 22610 Type III sums of squares Source DF F v alue Pr > F m 6 4.78 0.0008 l 7 1.01 0.4375 m*l 6 0.08 0.9978 Table F 11 Key statistical analysis results of hyb rid binders and PG 76 22 Source DF F v alue Pr > F Model 1 3 2. 87 0.00 4 6 Type III sums of squares Source DF F v alue Pr > F m 3 4.46 0.0082 l 7 0.96 0.4697 m*l 3 0.13 0.9398 Table F 12 Key statistical analysis results of contrast between Hudson a nd PG 76 22 Contrast DF F v alue Pr > F H udson vs. PG 76 22 1 5.17 0.027 5 Table F 13 Key statistical analysis results of unmodified binders Source DF F v alue Pr > F Model 21 4.09 < 0.00 01 Type III sums of squares Source F v alue Pr > F m 3 9.33 < 0.00 01 l 11 0. 77 0. 6730 m*l 7 0. 70 0. 6684

PAGE 161

161 Table F 14 Key statistical analysis results of contrast between unmodified binders Contrast DF F v alue Pr > F AC 20 vs. AC 30 1 35 .30 <.0001 AC 20 vs. PG 64 22 1 0.09 0.75 93 AC 20 vs. PG 67 22 1 6.12 0.0 1 45 AC 30 vs. PG 64 22 1 5. 40 0.02 1 5 AC 30 vs. PG 67 22 1 2 6.74 <.0001 PG 64 22 vs. PG 67 22 1 9. 94 0.002 0

PAGE 162

162 APPENDIX G CHARACTERISTIC TRUE STRESS TRUE STRAIN CURVES Figure G 1 PG 6 7 22 PAV residue true stress vs. true strain 15 C Figure G 2 AC 30 recovered true stress vs. true strain

PAGE 163

163 Figure G 3 AC 20 recovered true stress vs. true strain Figure G 4 PG 64 22 recovere d true stress vs. true strain

PAGE 164

164 Figure G 5 Unmodified bi nders true stress vs. true strain Figure G 6. PG 76 22 recove red true stress vs. true strain (1)

PAGE 165

165 Figure G 7. PG 76 2 2 recove red true stress vs. true strain (2) Figure G 8. PG 76 22 recover ed true stress vs. true strain (3)

PAGE 166

166 Figure G 9. PG 76 22 reco vered true stress vs. true strain (4) Figure G 10 PG 76 22 PAV r esidue true stress vs. true strain

PAGE 167

167 Figure G 11 PG 82 22 PAV resid ue true stress vs. true strain Figure G 12. Polymer modified bi nders true stress vs. true strain

PAGE 168

168 Figure G 13 ARB 5 and ARB 12 PAV resi due true stress vs. true strain Figure G 14 Overheated ARB 12 PAV resi due true stress vs. true strain

PAGE 169

169 Figure G 15. ARB 5 recover ed true stress vs. true strain Figure G 1 6 Marianni PAV res idue true stress vs. true strain

PAGE 170

170 Figure G 17. Rubber modified bind ers true stress vs. true strain Figure G 1 8 Wright PAV resid ue true stress vs. true strain

PAGE 171

171 Figure G 1 9 Hudson PAV resid ue true stress vs. true strain Figure G 20 Geotech PAV residu e true stress vs. true strain

PAGE 172

172 Figure G 21. Hybrid binde rs t rue stress vs. true strain Figure G 22. PG 76 22 an d unmodified binder true stress vs. true strain

PAGE 173

173 Figure G 23. Comparison with Hudson true stress vs. true strain Fi gure G 24. Comparison with Geotech, true stress vs. true strain

PAGE 174

174 Figure G 25. Comparison with Wright, true stress vs. true strain Figure G 26. Comparison with ARB 12, true stress vs. true strain

PAGE 175

175 Figure G 27. Comparison with Marianni, true stress vs. true strain Figure G 28. Comparison with PG 82 22, true stress vs. true strain

PAGE 176

176 Figure G 29 Tru e stress vs. tru e strain, polymer modified (reduced size) and unmodified binders Figure G 30 PG 76 22 and Hudson, true stress vs. true strain

PAGE 177

177 LIST OF REFERENCES Anderson, T.L. (2005). Fracture mechanics: fundamentals and applications 3 rd Ed., CRC Press. Annual Book of ASTM Standards 0 4.03 ASTM International, West Conshohocken, PA. ASTM D6723 (2001) Standard m ethod o f t est for d etermining the f racture p roperties of a sphalt b inder in d irect t ension (D T ) Annual Book of ASTM Standards 0 4.03 ASTM International, West Conshohocken, PA. temperatu Transportation Research Circular E C147, December, 25 33. Janssen, M., Zuidema, J., and Wanhill, R.J.H. (2002 2006). Fracture Mechanics 2 nd Ed., VSSD, The Netherlands. Roque, R., Lopp G., Li valuati on of hybrid binder use in surface mixtures 545 Contract, University of Florida, Gainesville, FL. Rosales, Alejandro (2011). "Fracture energy method for determining stiffness in polymer modified asphalt binders using t he single edge notched beam," McNair Scholars Research Journal : Vol. 7: Iss. 1, Article 15. SAS Institute Inc. (2002 2011). SAS 9.3 Help and Documentation Cary, NC, USA.

PAGE 178

178 BIOGRAPHICAL SKETCH in 1972. His father is a professional violin player in a s ymphony o rchestra. So he grew up under the nurture of symphony, and loves music all the time After g raduating from a College Diploma in civil engineering in 1994, Tianying Niu worked as an enginee University in 2004, and his Master graduate research project Tianyi ng Niu attended the University of Florida in August, 2006 for Ph.D. study and research. Now he is doing research in binder fracture failure testing method and interpretation system. His favorite fields are fracture mechanics and Finite Element A n alysis (FE A).