DETERMINING THE OPTIMUM COMPACTION LEVEL FOR DESIGNING STONE MATRIX ASPHALT MIXTURES Except where reference is made to the work of others, the work described in this dissertation is my own or was done in collaboration with my advisory committee. This dissertation does not include proprietary or classified information. ________________________ Hongbin Xie Certificate of Approval: _________________________ _________________________ Mary Stroup-Gardiner E. Ray Brown, Chairman Associate Professor Professor Civil Engineering Civil Engineering _________________________ _________________________ David Timm Stephen L. McFarland Assistant Professor Acting Dean Civil Engineering Graduate School DETERMINING THE OPTIMUM COMPACTION LEVEL FOR DESIGNING STONE MATRIX ASPHALT MIXTURES Hongbin Xie A Dissertation Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Doctorate of Philosophy Auburn, Alabama May 11, 2006 DETERMINING THE OPTIMUM COMPACTION LEVEL FOR DESIGNING STONE MATRIX ASPHALT MIXTURES Hongbin Xie Permission is granted to Auburn University to make copies of this dissertation at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ______________________________ Signature of Author ______________________________ Date of Graduation iii VITA Hongbin Xie, was born on June 7, 1976, in China. He entered Tongji University in September 1993 and gained his Bachelor of Science and Master of Science, both in Civil Engineering, in July 1997 and March 2000, respectively. He entered Royal institute of Technology, Sweden, in May 2000 to pursue his Ph.D. degree, also in Civil Engineering, and transferred to Auburn University in February 2002. He was employed as a part-time on-site quality control engineer during his senior undergraduate year for a new constructed freeway project following the summer intern in the same project. His Bachelor thesis was on polymer-modified asphalt, including lab comparison of several polymer additives and field verification tests. During his graduate study in China, he participated in two SMA research projects in Shanghai and Shenzhen, and was employed as on-site asphalt plant quality control engineer and did the major work on SMA mix designs for the Shenzhen project. His Master thesis was on mix design of Asphalt Treated Permeable Base (ATPB). The thesis work included developing a design method for ATPB mixes and construction of two test roads. His Ph.D. program in Sweden was on unbound granular base materials, he finished the literature review report and some course work before he transferred to Auburn University. Hongbin married Jingna Zhang, on May 16, 2000. They are the parents of one cute boy, Edwin Anthony Xie. iv DISSERTATION ABSTRACT DETERMINING THE OPTIMUM COMPACTION LEVEL FOR DESIGNING STONE MATRIX ASPHALT MIXTURES Hongbin Xie Doctor of Philosophy, May 11, 2006 (M.S., Tongji University, 2000) (B.S., Tongji University, 1997) 348 Typed Pages Directed by E. Ray Brown As Stone matrix asphalt (SMA) becomes more widely used in the United States, there is a need to further refine its mix design procedure. Some states have found that 100 gyrations with the Superpave Gyratory Compactor (SGC) suggested in current design guides are excessive for their materials and have specified the use of lower compaction levels. However, the use of these low compaction levels had little research support. The objective of this study was to determine the optimum compaction level for SMA mixtures that will provide increased durability and acceptable rutting resistance. The study is also needed to determine if the same compaction effort is applicable for SMA mixes of various nominal maximum aggregate sizes (NMAS). The study was carried out by conducting SMA mixture designs using different compaction levels, and comparing these different compaction levels in terms of volumetric properties and rutting performance. Five aggregates with a wide range of Los v angles (L.A.) abrasion values were selected. For each aggregate, three NMAS (19 mm, 12.5 mm, and 9.5 mm) mixtures were designed by at least three compaction efforts (50 blows Marshall, 65 and 100 gyrations with the SGC. A further lower gyration level (40 gyrations) was also used to design two mixtures, to show the effect of further reduction in number of gyrations. A total of 47 mixture designs were conducted in this study. Both vacuum seal (CoreLok) and saturated surface dry (SSD) methods were used for measurement of air voids. Aggregate breakdown was evaluated for all compaction efforts. Permeability, wheel load tracking (Asphalt pavement analyzer, APA), dynamic modulus, static creep, and repeated load tests were conducted on all mixtures designed with the different gyration levels. The CoreLok and SSD method provided a significant difference in air void results for lab compacted SMA mixtures. The correction factor embedded in the CoreGravity TM software is not acceptable for determining the bulk specific gravity of SMA mixtures. The error potentials for both methods were analyzed and suggestions were made to properly use these two methods for determining air voids of SMA mixtures. SMA mixtures designed with 65 gyrations should provide improved durability than those designed with 100 gyrations due to the increased optimum asphalt content. SMA mixtures designed with 65 gyrations were generally had similar or lower permeability than those designed with 100 gyrations at similar air voids. Marshall compaction resulted in significant higher aggregate breakdown than gyratory compaction. The aggregate breakdowns for both 65 and 100 gyrations were very similar to that observed in the field. All designed SMA mixtures achieved stone-on-stone contact as indicated by the VCA ratio, and had acceptable asphalt draindown. vi For the APA rutting test, 13 of 15 SMA mixtures designed with 65 gyrations performed well when 5.0 mm was used as the maximum allowable rut depth. The dynamic modulus test results indicated that reducing the compaction level from 100 gyrations to 65 gyrations only resulted in a small difference, and the ability to use E*/sin? term for predicting rutting resistance is questionable at high temperature for SMA mixtures due to the erroneous trend shown. Due to high test variability and long testing time, static creep test was not recommended to be used for evaluating SMA rutting resistance. Most (14 of 15) mixtures designed with 65 gyrations met the suggested 5 percent cumulative strain criterion after 10,000 cycles in the repeated load test. The rutting resistance indicated by the APA rut depth and cumulative strain from repeated load test becomes marginal when the gyration level reduced to 40 gyrations. The findings of this study indicated that 65 gyrations (the SGC used had an internal gyration angle of 1.23 degrees) can be used to design a more durable SMA mixture, while still maintaining the good rutting resistance that SMA mixtures are noted for. The successful design with 65 gyrations for all five aggregates in this study indicates that a lower design compaction level may allow the use of more aggregate sources for SMA mixtures without adversely affecting the performance. The current requirements for L.A abrasion and F&E content may be too stringent and these two aggregate properties within the range of this study appear not detrimental for the rutting performance. The NMAS did not show as a significant factor for many of the test results, therefore, no difference in compaction level was suggested for different NMAS mixtures. vii ACKNOWLEDGEMENTS The work presented in this document was accomplished as part of the Federal Highway Administration Project AU-334, determining the optimum compaction level for SMA mixtures. The author wishes to thank everyone associated with the project for his or her efforts. Special thanks go to Dr. E. Ray Brown, for his direction for my whole Ph.D. study and research. Special thanks also go to Mr. Donald E. Watson, for his detailed guidance and valuable discussion in the whole research period. Thanks to Tim Vollor and the entire laboratory staff at the National Center for Asphalt Technology, for their consistent support on my laboratory work. The author also extends his appreciation to Dr. Mary-Stroup Gardiner and Dr. David Timm, for their inspiring discussion on data analysis and review of the dissertation. Very special thanks are given to Jingna Zhang, and Edwin Anthony Xie, for their support and understanding during the completion of this research and dissertation. viii Style manual or journal used Transportation Research Board, National Academy of Sciences Computer software used Microsoft Office XP, Minitab version 14.1 ix TABLE OF CONTENTS CHAPTER 1 INTRODUCTION ...................................................................................... 1 1.1 BACKGROUND .................................................................................................... 1 1.2 OBJECTIVES......................................................................................................... 2 1.3 SCOPE OF STUDY................................................................................................ 3 CHAPTER 2 LITERATURE REVIEW ........................................................................... 6 2.1 DEVELOPMENT AND EVALUATION OF THE SUPERPAVE GYRATORY COMPACTOR........................................................................................................ 6 2.1.1 Individual Literature ...................................................................................... 6 2.1.2 Summary of Literature................................................................................. 37 2.2 THE DEVELOPMENT OF SMA MIX DESIGN................................................ 39 2.2.1 Individual Literature .................................................................................... 39 2.2.2 Summary of Literature................................................................................. 74 2.3 FINDINGS ON COMPACTION EFFORT FOR DESIGNING SMA MIXTURES ............................................................................................................................... 76 2.4 OTHER RELEVANT TESTS RELATED TO TOPICS...................................... 78 2.4.1 Vacuum Seal Method (CoreLok)................................................................. 78 2.4.2 Permeability ................................................................................................. 80 2.4.3 Aggregate Breakdown ................................................................................. 81 2.4.4 Locking Point............................................................................................... 83 2.4.5 Asphalt Pavement Analyzer Rutting Test.................................................... 85 2.4.6 Triaxial Tests ............................................................................................... 90 CHAPTER 3 EXPERIMENTAL PROCEDURES......................................................... 98 3.1 RESEARCH PLAN .............................................................................................. 98 x 3.1.1 Materials Selection....................................................................................... 98 3.1.2 Mixture Design and Volumetric Properties ............................................... 101 3.1.3 Performance Testing .................................................................................. 105 3.2 MATERIALS SELECTION............................................................................... 106 3.2.1 Aggregate Tests ......................................................................................... 106 3.2.1.1 Specific gravity test........................................................................ 106 3.2.1.2 L.A abrasion test ............................................................................ 106 3.2.1.3 Flat and elongated content test....................................................... 107 3.2.1.4 Fine aggregate angularity............................................................... 107 3.2.1.5 Uncompacted air voids of coarse aggregate .................................. 108 3.2.1.6 Voids in coarse aggregates............................................................. 109 3.2.2 Fine Mortar Tests....................................................................................... 109 3.2.2.1 Dynamic shear rheometer test........................................................ 109 3.2.2.2 Bending beam rheometer test......................................................... 111 3.3 MIXTURE DESIGN PROCEDURES................................................................ 112 3.3.1 Mixture Design by Marshall Hammer ....................................................... 112 3.3.2 Mixture Design by Superpave Gyratory Compactor ................................. 114 3.3.3 Draindown Test.......................................................................................... 114 3.4 TESTS CONDUCTED ON MIX DESIGN SAMPLES..................................... 115 3.4.1 Air Void Content Determination ............................................................... 115 3.4.2 Flexible Wall Falling Head Permeability .................................................. 117 3.4.3 Ignition Oven Test ..................................................................................... 118 3.5 PERFORMANCE TESTING EQUIPMENT AND METHODS ....................... 119 3.5.1 Asphalt Pavement Analyzer (APA) Wheel Tracking ................................ 119 3.5.2 Dynamic Modulus...................................................................................... 120 3.5.2.1 Testing setup .................................................................................. 120 3.5.2.2 Method of analysis......................................................................... 122 3.5.3 Static Creep................................................................................................ 124 3.5.3.1 Background of Creep Behavior ..................................................... 124 3.5.3.2 Testing Setup ................................................................................. 127 xi 3.5.3.3 Method of analysis......................................................................... 127 3.5.4 Repeated Load Confined Creep................................................................. 130 3.5.4.1 Testing setup .................................................................................. 130 3.5.4.2 Method of analysis......................................................................... 131 3.6 PERFORMANCE TEST SPECIMEN PRODUCTION..................................... 132 3.6.1 APA Specimen Production ........................................................................ 132 3.6.2 Triaxial Test Specimen Production............................................................ 133 CHAPTER 4 TEST RESULTS, ANALYSIS AND DISCUSSION ON MIX DESIGN PROPERTIES......................................................................................... 136 4.1 MATERIAL PROPERTIES ............................................................................... 136 4.1.1 Coarse Aggregate Properties...................................................................... 136 4.1.2 Fine Aggregate Angularity ........................................................................ 138 4.1.3 Mineral Filler Properties............................................................................ 138 4.1.4 Asphalt Binder Properties.......................................................................... 138 4.1.5 Fiber Properties.......................................................................................... 139 4.1.6 Fine Mortar Properties ............................................................................... 139 4.2 MIXTURE DESIGN PROPERTIES .................................................................. 141 4.2.1 Marshall Mix Design ................................................................................. 141 4.2.2 Gyratory Mix Design ................................................................................. 142 4.2.2.1 Locking points ............................................................................... 142 4.2.2.2 Gyratory mix design results........................................................... 148 4.2.3 Effects of Compaction on Volumetric Properties...................................... 150 4.2.4 Draindown Test Results............................................................................. 154 4.3 AIR VOID CONTENT MEASUREMENT........................................................ 155 4.3.1 Concepts of Air Voids ............................................................................... 157 4.3.2 Comparison of Two Test Methods ............................................................ 160 4.3.3 Effect on Mix Design Volumetric Properties ............................................ 170 4.3.4 Triaxial Test Sample Preparation............................................................... 173 4.3.5 Vacuum Seal Method Manual Calculation................................................ 178 xii 4.4 PERMEABILITY TEST..................................................................................... 184 4.4.1 The Relationship between Permeability and Air Voids............................. 184 4.4.2 Effect of Compaction level on Permeability.............................................. 188 4.5 AGGREGATE DEGRADATIONS.................................................................... 193 4.6 SUMMARY........................................................................................................ 208 CHAPTER 5 TEST RESULTS, ANALYSIS AND DISCUSSION ON PERFORMANCE TESTS ...................................................................... 211 5.1 APA RUTTING TEST ....................................................................................... 211 5.1.1 APA Test Results and Analysis ................................................................. 211 5.1.2 Discussion on APA Rut Depth versus Gyration Level.............................. 218 5.2 DYNAMIC MODULUS..................................................................................... 222 5.3 STATIC CREEP ................................................................................................. 245 5.4 REPEATED LOAD TEST ................................................................................. 255 5.4.1 Repeated Load Test Results and Analysis................................................. 255 5.4.2 Discussion on Cumulative Strain Criteria.................................................. 268 5.5 SUMMARY........................................................................................................ 275 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS................................. 278 6.1 CONCLUSIONS................................................................................................. 278 6.2 RECOMMENDATIONS.................................................................................... 281 REFERENCES ............................................................................................................... 283 APPENDIX A AIR VOID CONTENT AND PERMEABILITY TEST RESULTS ...... 293 APPENDIX B COREGRAVITY TM PROGRAM CORRECTION FACTOR ................ 303 APPENDIX C AGGREGATE BREAKDOWN FOR DIFFERENT COMPACTION EFFORTS ............................................................................................... 305 APPENDIX D TRIAXIAL PERFORMANCE TEST RESULTS .................................. 314 xiii LIST OF TABLES TABLE 2.1 Summary of Average Differences between Field Cores and Lab-Compacted Specimens (6)............................................................................................... 10 TABLE 2.2 Mean Squared Error (MSE) Comparison of Compaction Data (6) ............. 11 TABLE 2.3 Evolution of Gyratory Compaction (23)...................................................... 35 TABLE 2.4 Aggregate Gradations of Mixtures Tested (33) ........................................... 51 TABLE 2.5 Volumetric Properties of SMA Mixtures Compacted by the Marshall Hammer (36)................................................................................................ 59 TABLE 2.6 Volumetric Properties of SMA Mixtures Compacted by the SGC (36) ...... 59 TABLE 2.7 APA Testing Specifications Used by Various State Agencies .................... 87 TABLE 2.8 APA Rut Depth Criteria (70-71, 79, 82)...................................................... 89 TABLE 2.9 Criteria for Static Creep Test Results (95)................................................... 93 TABLE 3.1 Trial Gradations Used in Preliminary Mix Designs .................................. 100 TABLE 3.2 Preliminary Mix Design Results................................................................ 101 TABLE 3.3 Gradations Used in This Study .................................................................. 103 TABLE 3.4 SMA Mortar Quality Requirements (43)................................................... 111 TABLE 3.5 Automatic Marshall Hammer Calibration Data......................................... 112 TABLE 3.6 Geometric Requirements for Triaxial Samples (84, 86)............................ 133 TABLE 4.1 Aggregate Properties.................................................................................. 137 TABLE 4.2 Mineral Filler Properties............................................................................ 138 TABLE 4.3 PG 76-22 Asphalt Binder Properties.......................................................... 139 TABLE 4.4 Cellulose Fiber Properties.......................................................................... 139 TABLE 4.5 Mortar Test Results.................................................................................... 140 TABLE 4.6 Marshall Mix Design Volumetric Properties 1 Summary........................... 142 xiv TABLE 4.7 Locking Point Results Summary ............................................................... 143 TABLE 4.8 Forward Stepwise Regression Results for Locking Point.......................... 145 TABLE 4.9 100 Gyrations Mix Design Volumetric Properties 1 ................................... 149 TABLE 4.10 65 Gyrations Mix Design Volumetric Properties 1 ..................................... 149 TABLE 4.11 40 Gyrations Mix Design Volumetric Properties 1 ..................................... 150 TABLE 4.12 Draindown Test Results Summary ............................................................ 155 TABLE 4.13 Paired T-Test for CoreLok and SSD air voids........................................... 161 TABLE 4.14 GLM for Influencing Factors on VTM Ratio ............................................ 162 TABLE 4.15 GLM for Influencing Factors on Water Absorption.................................. 168 TABLE 4.16 Mix Design Volumetric Properties 1 Summary .......................................... 171 TABLE 4.17 Targeting Whole Samples VTM 1 for Triaxial Samples............................. 177 TABLE 4.18 Critical Air Void Content for Permeable SMA Mixtures.......................... 187 TABLE 4.19 Critical Air Void Content for Permeable SMA Mixtures.......................... 191 TABLE 4.20 F-tests for permeability regressions of two compaction levels.................. 192 TABLE 4.21 Marshall Compaction Aggregate Breakdown Results............................... 195 TABLE 4.22 100 Gyrations Aggregate Breakdown Results........................................... 196 TABLE 4.23 65 Gyrations Aggregate Breakdown Results............................................. 197 TABLE 4.24 Paired-T Test on Degradation for Two gradations .................................... 198 TABLE 4.25 ANOVA on Aggregate Breakdown at Critical Sieve Size ........................ 200 TABLE 4.26 Paired T Test on Degradation for 100 Gyrations and Marshall Compaction .. .............................................................................................................. 204 TABLE 4.27 Paired T Test on Degradation for 65 and 100 Gyrations........................... 204 TABLE 4.28 Average Percent Passing Changes at Three Sieve Sizes ........................... 205 TABLE 4.29 Average Test Results for Different Compaction Levels............................ 208 TABLE 5.1 APA Rutting Results.................................................................................. 212 TABLE 5.2 ANOVA for APA Rutting Results............................................................. 215 xv TABLE 5.3 ANOVA for Dynamic Modulus................................................................. 224 TABLE 5.4 ANOVA for Phase Angle .......................................................................... 224 TABLE 5.5 Dynamic Modulus Test Results at Load Frequency of 10 Hz................... 230 TABLE 5.6 Dynamic Modulus Test Results at Load Frequency of 0.1 Hz.................. 231 TABLE 5.7 Dynamic Modulus Test Results at Load Frequency of 10 Hz (Without 6 Outlier Samples) ........................................................................................ 233 TABLE 5.8 Dynamic Modulus Test Results at Load Frequency of 0.1 Hz (Without 6 Outlier Samples) ........................................................................................ 234 TABLE 5.9 GLM Results on E*/sin? at Load Frequency of 10 Hz.............................. 235 TABLE 5.10 Pair T-Test Results on E*/sin? Value of Two Compaction Levels........... 237 TABLE 5.11 GLM Results on E*/sin? at Load Frequency of 0.1 Hz............................. 238 TABLE 5.12 Static Creep Test Results Summary........................................................... 248 TABLE 5.13 ANOVA for Slope of Strain in Secondary Phase of Static Creep Test ..... 249 TABLE 5.14 ANOVA for Log Test Time when 4 Percent Strain Occurred................... 251 TABLE 5.15 Summary of Repeated Load Test Results.................................................. 257 TABLE 5.16 ANOVA for Strain at 10,000 Cycles ......................................................... 260 TABLE 5.17 Paired-T Test Results on Strain at 10,000 Cycles...................................... 263 TABLE 5.18 ANOVA for Strain Slope at Secondary Phase of Repeated Load Test ..... 265 TABLE 5.19 Average Test Results for Different Compaction Levels............................ 275 TABLE A1 Air Voids and Permeability Test Results................................................... 294 TABLE B1 CoreGravity TM Program Correction Factors (110) .................................... 304 TABLE D1 Dynamic Modulus Test Results................................................................. 315 TABLE D2 Static Creep Test Results ........................................................................... 319 TABLE D3 Repeated Load Confining Creep Test Results........................................... 323 xvi LIST OF FIGURES FIGURE 2.1 Effect of increase in compaction effort on limestone SMA mixture (31). 48 FIGURE 2.2 Optimum asphalt content from two compaction efforts (36).................... 58 FIGURE 2.3 Gmb ratio as a function of gyratory level and L.A. abrasion loss (42)..... 71 FIGURE 2.4 Correlation of L.A Abrasion values and aggregate breakdown for field compacted samples (64)............................................................................ 83 FIGURE 2.5 Permanent strains of core samples by triaxial repeated load test (89)...... 96 FIGURE 2.6 Field rut depth versus the lab strain from repeated load test (96)............. 96 FIGURE 3.1 Work plan for phase I: material selection and preliminary test................ 99 FIGURE 3.2 Work plan for phase II: SMA mix designs. ............................................ 102 FIGURE 3.3 Gradations used in this study. ................................................................. 103 FIGURE 3.4 Work plan for phase III: performance tests. ........................................... 104 FIGURE 3.5 Components of complex modulus G*..................................................... 110 FIGURE 3.6 Automatic Marshall hammer calibration. ............................................... 113 FIGURE 3.7 Flexible wall falling head permeameter.................................................. 117 FIGURE 3.8 Asphalt Pavement Analyzer.................................................................... 119 FIGURE 3.9 Haversine loading pattern or stress pulse for the dynamic modulus test.120 FIGURE 3.10 The MTS and environmental chamber used for triaxial testing. ............ 121 FIGURE 3.11 A sample prepared for triaxial testing..................................................... 122 FIGURE 3.12 HMA creep behavior in static creep test................................................. 125 FIGURE 3.13 Krass?s model for creep behavior (107).................................................. 126 FIGURE 3.14 Typical test results between compliance and loading time..................... 128 xvii FIGURE 3.15 Regression constants a and m obtained from the secondary zone of the log compliance?log time plot........................................................................ 129 FIGURE 3.16 Repeated load test schematic graph. ....................................................... 130 FIGURE 3.17 Typical relationship between total cumulative plastic strain and loading cycles....................................................................................................... 131 FIGURE 3.18 Regression constants a and b when plotted on a log?log scale............... 132 FIGURE 3.19 Whole and cored sample prepared for triaxial testing. ........................... 134 FIGURE 4.1 Histogram of locking point results.......................................................... 144 FIGURE 4.2 Average locking point results. ................................................................ 146 FIGURE 4.3 Average locking point values versus asphalt content. ............................ 148 FIGURE 4.4 Comparison of optimum asphalt content. ............................................... 150 FIGURE 4.5 Comparison of VMA for various compaction levels.............................. 152 FIGURE 4.6 Comparison of VCA ratio for various compaction levels. ..................... 153 FIGURE 4.7 Volumes associated with compacted HMA (49). ................................... 157 FIGURE 4.8 Comparison of the CoreLok and SSD air voids...................................... 161 FIGURE 4.9 Surface textures for different NMAS mixtures....................................... 163 FIGURE 4.10 Relationships between the CoreLok and SSD air voids for three NMAS. ... .............................................................................................................. 164 FIGURE 4.11 Air voids difference between the CoreLok and SSD method versus the SSD air voids. ......................................................................................... 165 FIGURE 4.12 Relationships between absorbed water and air void content. ................. 168 FIGURE 4.13 The difference between the CoreLok and SSD air voids versus absorbed water content........................................................................................... 169 FIGURE 4.14 Effect of CoreLok and SSD methods on optimum asphalt content........ 170 FIGURE 4.15 Effect of CoreLok and SSD methods on voids in mineral aggregate..... 172 FIGURE 4.16 Corrected CoreLok and SSD Air Voids for Whole Samples.................. 174 FIGURE 4.17 CoreLok and SSD air voids for core samples......................................... 175 xviii FIGURE 4.18 Air voids relationships between the whole and core samples by the SSD method..................................................................................................... 176 FIGURE 4.19 CoreLok air voids by program versus uncorrected air voids.................. 181 FIGURE 4.20 Air voids difference between program and uncorrected calculation. ..... 182 FIGURE 4.21 Relationship between permeability and VTM for 19 mm NMAS.......... 185 FIGURE 4.22 Relationship between permeability and VTM for 12.5 mm NMAS....... 185 FIGURE 4.23 Relationship between permeability and VTM for 9.5 mm NMAS......... 186 FIGURE 4.24 Permeability results for 65 and 100 gyration levels for 19 mm NMAS. 189 FIGURE 4.25 Permeability results for 65 and 100 gyration levels for 12.5 mm NMAS. ... .............................................................................................................. 189 FIGURE 4.26 Permeability results for 65 and 100 gyration levels for 9.5 mm NMAS.190 FIGURE 4.27 Typical aggregate breakdown results for different compaction efforts (C.GVL 12.5 mm NMAS). ..................................................................... 193 FIGURE 4.28 Comparison of N and F gradation on aggregate breakdown. ................. 198 FIGURE 4.29 Critical sieve changes due to compaction............................................... 199 FIGURE 4.30 Average aggregate breakdown for all main factors. ............................... 201 FIGURE 4.31 Relationship between aggregate breakdown and L.A abrasion value. ... 202 FIGURE 4.32 Relationship between aggregate breakdown and F&E content. ............. 203 FIGURE 4.33 Interaction between aggregate type and compaction level on aggregate breakdown............................................................................................... 206 FIGURE 4.34 Interaction between aggregate type and NMAS on aggregate breakdown... .............................................................................................................. 206 FIGURE 4.35 Interaction between compaction level and NMAS on aggregate breakdown..................................................................................................... 207 FIGURE 5.1 Comparison of APA rutting for three compaction levels. ...................... 213 FIGURE 5.2 Comparison of average APA rut depth................................................... 215 xix FIGURE 5.3 Interaction between aggregate type and NMAS on APA rut depth........ 217 FIGURE 5.4 Interaction between NMAS and compaction level on APA rut depth.... 218 FIGURE 5.5 The relationship between APA rut depth and compaction level............. 219 FIGURE 5.6 Correlation between field and APA rut depth from NCHRP 9-17 project (80).......................................................................................................... 220 FIGURE 5.7 Effects of air voids on APA rut depths (80). .......................................... 221 FIGURE 5.8 Average dynamic modulus versus load frequency. ................................ 226 FIGURE 5.9 Average phase angle versus load frequency. .......................................... 227 FIGURE 5.10 Average storage and loss modulus at different frequencies.................... 228 FIGURE 5.11 Average E*/sin? results at frequency of 10 Hz. ..................................... 236 FIGURE 5.12 Interaction between aggregate type and NMAS on E*/sin? results at frequency of 10 Hz.................................................................................. 237 FIGURE 5.13 Average E*/sin? results at frequency of 0.1 Hz. .................................... 239 FIGURE 5.14 Interaction between aggregate type and NMAS on E*/sin? results at frequency of 0.1 Hz................................................................................. 239 FIGURE 5.15 The comparison of E*/sin? between 10 Hz and 0.1 Hz.......................... 240 FIGURE 5.16 The correlations between field rut depths and E*/sin? values (84)........ 242 FIGURE 5.17 The relationships between E*/sin? and APA rut depths......................... 244 FIGURE 5.18 A typical static creep test result with tertiary flow. ................................ 245 FIGURE 5.19 A typical static creep test result without tertiary flow. ........................... 246 FIGURE 5.20 Average slopes for two gyration levels................................................... 250 FIGURE 5.21 Interaction between aggregate type and NMAS on slope of strain in static creep test. ................................................................................................ 251 FIGURE 5.22 Average log time to reach 4 percent strain in static creep test................ 252 FIGURE 5.23 Interaction between aggregate type and NMAS on average log time to reach 4 percent strain. ............................................................................. 253 xx FIGURE 5.24 The Correlations between flow time and field rut depths (84). .............. 254 FIGURE 5.25 A typical repeated load test result with tertiary flow.............................. 258 FIGURE 5.26 A typical repeated load test result without tertiary flow......................... 259 FIGURE 5.27 Strain level at 10,000 cycles for three gyration levels........................... 260 FIGURE 5.28 Average strains at 10,000 cycles for two main factors. ......................... 261 FIGURE 5.29 Interaction between aggregate type and compaction level on strain at 10,000 cycles........................................................................................... 263 FIGURE 5.30 Strain slope at secondary phase for three compaction levels................. 264 FIGURE 5.31 Average strain slopes for different aggregate types............................... 265 FIGURE 5.32 Relationship between strain slope and uncompacted voids of coarse aggregate................................................................................................. 266 FIGURE 5.33 Interaction between aggregate type and compaction level on strain slope.. .............................................................................................................. 267 FIGURE 5.34 The relationship between ram strain and LVDT strain reading............. 269 FIGURE 5.35 The relationship between strain at 3600 and 10,000 cycles................... 269 FIGURE 5.36 The relationship between strain level and compaction level. ................ 271 FIGURE 5.37 Relationship between repeated load cumulative strain and APA rut depth. .............................................................................................................. 272 FIGURE 5.38 The Correlations between flow number and field rut depths (84)......... 273 FIGURE 5.39 Field rut depth versus the lab strain from repeated load test (96)........... 274 FIGURE C1 Aggregate breakdown for crushed gravel 19 mm NMAS mixture......... 306 FIGURE C2 Aggregate breakdown for crushed gravel 12.5 mm NMAS mixture...... 306 FIGURE C3 Aggregate breakdown for crushed gravel 9.5 mm NMAS mixture........ 307 FIGURE C4 Aggregate breakdown for lab granite 19 mm NMAS mixture. .............. 307 FIGURE C5 Aggregate breakdown for lab granite 12.5 mm NMAS mixture. ........... 308 FIGURE C6 Aggregate breakdown for lab granite 9.5 mm NMAS mixture. ............. 308 xxi FIGURE C7 Aggregate breakdown for limestone 19 mm NMAS mixture................. 309 FIGURE C8 Aggregate breakdown for limestone 12.5 mm NMAS mixture.............. 309 FIGURE C9 Aggregate breakdown for limestone 9.5 mm NMAS mixture................ 310 FIGURE C10 Aggregate breakdown for ruby granite 19 mm NMAS mixture............. 310 FIGURE C11 Aggregate breakdown for ruby granite 12.5 mm NMAS mixture.......... 311 FIGURE C12 Aggregate breakdown for ruby granite 9.5 mm NMAS mixture............ 311 FIGURE C13 Aggregate breakdown for traprock 19 mm NMAS mixture................... 312 FIGURE C14 Aggregate breakdown for traprock 12.5 mm NMAS mixture................ 312 FIGURE C15 Aggregate breakdown for traprock 9.5 mm NMAS mixture.................. 313 xxii CHAPTER 1 INTRODUCTION 1.1 BACKGROUND Stone Matrix Asphalt (SMA) was first introduced into the United States as a result of the European Asphalt Study Tour of 1990 (1). European experience with SMA showed that this mix technology resulted in improved resistance to rutting of hot mix asphalt pavements. As a result of information and recommendations from the tour, several states became interested in the SMA technology and placed test sections or test projects beginning in 1991 to evaluate this mix. A Technical Working Group (TWG) was sponsored by the FHWA to develop guidelines for materials and mix design requirements, and to assist state Departments of Transportation (DOTs) as needed in providing information in regards to mix design, production, and placement of SMA mixtures. The TWG, in a cooperative effort with state and federal agencies and the asphalt paving industry, published ?Guidelines for Materials, Production, and Placement of Stone Matrix Asphalt (SMA)? in 1994 (2). Based on European experience and limited experience in this country, the guidelines recommended a mix design compactive effort of 50 blows for each face of the test specimens using a Marshall hammer. The guidelines were updated in 1999 with a National Asphalt Pavement Association (NAPA) publication, ?Designing and Constructing SMA Mixtures ? State-of-the-Practice,? which described laboratory samples being compacted with either the 50 blow Marshall method or by using 1 100 gyrations of the Superpave gyratory compactor (SGC). However, some states such as Georgia and Texas have found that 100 gyrations with the SGC is excessive for their materials and results in mixtures with lower than desired optimum asphalt contents. The high level of density obtained in the laboratory is also difficult to obtain in the field without excessive fracturing of aggregate particles. Experience from Georgia and Texas indicates that for their materials the optimum SGC compactive effort should be between 50 and 75 gyrations. Georgia, for example, has required 50 gyrations as the standard gyratory compaction level for these mixes. Since there is a renewed interest in SMA technology by state agencies around the country in their search for a more durable, rut-resistant pavement, there is a need to identify a standard compaction effort with the SGC that will provide optimum density and overall good performance. 1.2 OBJECTIVES As states renew their interest in SMA mixture technology, some are finding that previous guidelines of 100 gyrations with the SGC may not be satisfactory for their materials, and may result in lower optimum asphalt contents than needed or desired. The objective of this research study is to evaluate a lower compaction level for SMA mixtures that will provide more durability and satisfactory rutting resistance through stone-on-stone contact without fracturing aggregates due to excessive compactive force. Another objective is to determine if the same compaction effort is applicable for SMA mixes of various nominal maximum aggregate sizes (NMAS). 2 1.3 SCOPE OF STUDY The objectives will be accomplished by executing the following tasks: Task 1: Select aggregates which will be representative of those used in typical SMA mixes, such as granite, limestone, crushed gravel, and traprock etc. Select aggregate from sources that represent a range in Los Angeles (L.A.) abrasion values. Aggregates with L.A. abrasion values ranging from approximately 20 to near 40 (based on the B grading in ASTM C131 (3)) will be used to determine if the compactive effort results in significant aggregate breakdown. Three nominal maximum aggregate sizes (19 mm, 12.5 mm and 9.5 mm) will be used to determine if the same compaction level can be used regardless of nominal maximum aggregate size. Task 2: Conduct SMA mixture designs with Marshall compaction and Superpave gyratory compaction. Marshall compaction uses the compaction effort that is equivalent to 50 blows with the manual hammer. Conduct trial SMA mix designs by the SGC with different gyration levels. For this study two gyration levels will be used: (1) 100 gyrations, (2) the lowest level of gyrations that approach the locking point. Locking point is defined as a gyration level at which sample height remains the same or less than 0.1 mm in difference for two successive gyrations in this study. Locking point is a point in the compaction process where additional gyrations provide very little increase in density (4). A third compaction level will be used to show the effect of further reduction in compaction level. This will result in 3 45 mix designs (5 aggregates x 3 compaction efforts x 3 mix types), plus some extra mix design work for the third gyratory compaction level. Task 3: Determine the air void content by using two methods: the vacuum seal (CoreLok) method and the saturated surface dry (SSD) method. Compare the effects of test method on determining SMA volumetric properties. Task 4: Use the two gyratory compaction levels indicated from Task 3 to prepare SGC specimens for performance testing. Some additional testing will be done at a third gyration level. Laboratory permeability tests will be conducted for all mix design samples to help determine at what point SMA mixtures become permeable, and the effects of compaction level on permeability. The APA will be used to test specimens at 64 ?C for 8,000 cycles after which the amount of rutting will be measured. APA tests will be conducted for all mixtures designed using different gyration levels, to evaluate how sensitive SMA mixtures are to variations in asphalt content and the effect of such variation on rutting resistance. Triaxial performance tests (dynamic modulus test, static creep test, and repeated load creep test) based on research by Arizona State University will be performed for all mixtures designed with different compaction levels. Triaxial samples will be cored and sawn from a SGC specimen with 150 mm in diameter and 170 mm in height to provide a specimen with 100 mm in diameter and 150 mm in height for testing. All the triaxial tests will be conducted at a high temperature of 60?C to 4 evaluate the rutting performance of designed SMA mixtures. A 20 psi confining pressure will be applied for all triaxial tests to simulate a typical in-place confining pressure. The dynamic modulus samples will be tested at a set of frequencies from 25 Hz to 0.1 Hz to simulate different traffic conditions. The load amplitude will be selected to produce a microstrain between 50 and 150 for each mixture. The static creep test will use a deviatoric stress of 100 psi and maintain the load till tertiary flow happens or after 5 hours loading, whichever comes first. Repeated load creep tests will use a peak deviatoric stress of 100 psi and will be conducted up to 10,000 cycles or until tertiary flow happens. A haversine loading of 0.1 second load time and 0.9 second rest will be used for the repeated load test. Task 5: Summarize and analyze all test results, refine current compaction level if needed, to ensure the design compaction level provides stone-on-stone contact and the most rutting resistance possible. Determine if the same gyration level can be used for each NMAS and, if not, establish the gyration levels needed for each NMAS. 5 CHAPTER 2 LITERATURE REVIEW This chapter presents an extensive literature review pertaining to the development and evaluation of the Superpave gyratory compactor (SGC), the development of stone matrix asphalt (SMA) mixture design, and other relevant literature related to topics of this study. The individual literature reviews are conducted for the background of the SGC development and SMA mixture design as two separate sections, and are in chronological sequence in each section. A section summary is given following each section, and a summary on SMA mixture design using different compaction levels is given at the end of this Chapter. In the individual reviews, comments by the author are generally designated by text in square brackets with italic font. Those comments are intended to clarify the original text, sometimes to draw the reader?s attention to the background of study or the premise of conclusions, occasionally to add information that was not contained in the original text but might be of advantage to the reader, or to correlate several papers having similar results or conflicting information. 2.1 DEVELOPMENT AND EVALUATION OF THE SUPERPAVE GYRATORY COMPACTOR 2.1.1 Individual Literature The literature reviews in this section are conducted to specifically answer the following questions: 6 ? How does the gyratory compactor compare to field compaction and other compactive efforts? ? What are the key parameters for gyratory compaction? Ortolani, L. and H.A. Sandberg, Jr. ?The Gyratory-Shear Method of Molding Asphaltic Concrete Test Specimens; Its Development and Correlation with Field Compaction Methods. A Texas Highway Department Standard Procedure?, Journal of Association of Asphalt Paving Technologists, Vol: 21, 1952. Ortolani et al (5) presented the development of a gyratory compaction procedure, and showed how the specimen compacted with the gyratory compactor correlated with field compaction. It is believed that simulating the final pavement density or ultimate density is the goal of any compaction method used in mix design. Several criteria were set up to determine a good compaction procedure: 1. A good laboratory compaction method should be able to be used for field quality control; 2. The compactor should yield essentially the same density as the pavement density after some years of traffic; 3. The compactor should have similar aggregate degradation as under the field condition. Several compaction devices were investigated, including two hydraulic compression test machines with different loading speeds, a standard Proctor Soil Compaction Machine, Public Roads Administration Vibratory Machine, a pneumatic roller-type molding machine, and a compaction device utilizing a conical roller or 7 compaction ram. However, after being tested in the laboratory, all of them were rejected for various reasons. The Gyratory Molding machine was another compactor to be investigated. It had been incorporated in a standard procedure by the Texas Highway Department. It is reported in this paper that the Gyratory Molding machine served its purpose well. The procedure was as fast or faster than most commonly accepted compaction techniques. It only required one operator and this person could easily compact thirty specimens a day by using the Gyratory Molding machine. It was also found that the gyratory compactor, which produced proper orientation of aggregate at low initial pressures, closely simulated the degradation found in field compaction. To correlate the field compaction with gyratory compaction, more than 400 field cores were collected from widely separated asphalt pavement sections in the state of Texas. These pavements comprised many different designs using different aggregates and types of asphalt. In addition, these pavements had been in service from a minimum of one year to twelve years under varied weather conditions. The road densities were also recorded at the time of construction. Results indicated that newly constructed surface courses had an average density 3.8 percent less than laboratory design density, and field cores after several years in service had an average density of only 0.8 percent less than the laboratory design density. It is also reported in this paper that samples prepared with gyratory compaction had good reproducibility in density. Some mixes were prepared with original asphalt and extracted aggregate from cores that were known to contain sound aggregate and were 8 expected to have litter degradation. Results showed that re-compacted samples had a density average of only 0.3 percent less than the density of the cores. Consuegra, A., D.H. Little, H.V. Quintus, and J. Burati, ?Comparative Evaluation of Laboratory Compaction Devices Based on Their Ability to Produce Mixtures with Engineering Properties Similar to Those Produced in the Field?, Transportation Research Record 1228, TRB, National Research Council, Washington, D.C., 1989. Consuegra et al (6) reported a field and laboratory study that evaluated the ability of five compaction devices to simulate field compaction in engineering properties. The compaction devices evaluated in this study included the mobile steel wheel simulator, the Texas gyratory compactor (gyration angle is fixed at 3 degrees), the California kneading compactor, the Marshall impact hammer, and the Arizona vibratory-kneading compactor. The engineering properties used for evaluation included resilient modulus, indirect tensile strength and strain at failure, and tensile creep data. Five projects were selected for this study. These projects were located in Texas, Virginia, Wyoming, Colorado, and Michigan. Field cores from each project were drilled one day after compaction. The sampling of asphalt mixtures for laboratory specimen preparation was performed with great care to ensure the random selection of trucks and to prevent segregation of mixtures. The loose field mix was properly sealed and transported to the laboratory, then reheated to the same compaction temperature as was used in the field. A trial and error method was used to determine the compactive effort to produce similar air void contents to that for field cores. 9 The repeated load indirect tensile test (resilient modulus) was performed in accordance with ASTM D4123-82 on samples from all five field projects. Indirect tensile strength tests were performed for three of five projects in accordance with test methods TEX-226-F of the Texas State Department of Highways and Public Transportation at 41?F, 77?F, and 104?F and at a loading rate of 2.0 in./min. The indirect tensile creep was performed for all five projects in the same way as the resilient modulus except that a static load, instead of a repeated load, was continuously applied for 60 min and then removed. Simple comparison of test results for the field cores and laboratory compacted specimens indicated the Texas gyratory compactor, on the average, simulated the field compaction most closely. The average differences for each of these properties are calculated by absolute difference between test value of field cores and laboratory compacted samples over test value of field cores. The summary of average differences is presented in Table 2.1. TABLE 2.1 Summary of Average Differences between Field Cores and Lab- Compacted Specimens (6) Compaction Device Creep Compliance at 77?F Indirect Tensile Strength Tensile Strain at Failure Resilient Modulus Arizona Compactor 0.77 0.51 0.47 0.41 Marshall Hammer 0.80 0.35 0.45 0.55 California Kneading 0.59 0.21 0.27 0.42 Steel Wheel Simulator 0.51 0.31 0.11 0.26 Texas Gyratory Compactor 0.44 0.14 0.16 0.37 The mean squared error (MSE) using the mean test value from field cores as a target value, was also employed to analyze the difference between the engineering 10 properties of field- and laboratory-compacted specimens. All test results were sorted and analyzed in terms of project, test, and temperature. The MSE ranking results are summarized in Table 2.2. A lower MSE value indicates a smaller difference, therefore a better compaction device to simulate laboratory-compacted specimens to field cores. TABLE 2.2 Mean Squared Error (MSE) Comparison of Compaction Data (6) Average MSE Rankings by Mixture Compaction Device Project Property Temperature Arizona Compactor 5.0 4.8 4.7 Marshall Hammer 4.0 3.5 3.3 California Kneading 2.0 2.0 2.0 Steel Wheel Simulator 1.7 2.8 2.0 Texas Gyratory Compactor 2.0 1.5 1.3 The Texas gyratory method had generally better MSE ranking than the other methods in most of the tests. The evaluation on MSE of all laboratory compaction devices indicated that the engineering properties of asphalt mixture depended on the type of compaction device used. Overall, the Texas gyratory compactor demonstrated the best ability to produce mixtures with similar engineering properties to those determined from field cores. The California kneading compactor and the mobile steel wheel simulator ranked second and third, respectively, but with little difference between the two. The Arizona vibratory- kneading compactor and the Marshall hammer ranked as the least effective in terms of their ability to produce mixtures with engineering properties similar to those from field cores. 11 Button, J.W., D.N. Little, V. Jagadam, and O.J. Pendleton. ?Correlation of Selected Laboratory Compaction Methods with Field Compaction?, In Transportation Research Record 1454, TRB, National Research Council, Washington, D.C., 1994. This study (7) compared the four laboratory compaction methods (Exxon rolling wheel, Texas gyratory, rotating base Marshall hammer, and the Elf linear kneading compactor) with the field compaction. The objective of this study was to recommend a best laboratory compaction method that can simulate the field compaction well and be convenient to use. Six laboratory tests were selected to evaluate samples compacted using the four lab compaction procedures and core samples. The tests were indirect tension at 25?C, resilient modulus at 0?C and 25?C, Marshall stability, Hveem stability, and uniaxial repetitive compressive creep followed by compression to failure. These tests were selected because they can be performed on 100 mm core samples from thin pavement layers. Based on the statistical analysis of the test data, the Texas gyratory compactor simulated pavement cores most often (73 percent of the tests performed). The Exxon rolling wheel and Elf compactor simulated pavement cores with equal frequency (64 percent of the tests performed). The rotating base Marshall hammer simulated pavement cores least often (50 percent of the tests performed). These differences are not statistically significant (at ?=0.05). When compared with the Exxon rolling wheel compactor, the Texas gyratory compactor is more convenient for preparing lab specimens for routine mixture design testing of asphalt concrete. Air voids distribution of gyratory compacted specimens may 12 be less similar to pavement cores than rolling wheel compacted specimens, however, this difference did not adversely affect the mixture properties measured for this study. Based solely on the findings of this study, the Texas gyratory compactor was recommended to SHRP for use in preparing routine lab test specimens. [Testing in this study was limited to dense graded mixtures, SMA was not evaluated.] Blankenship, R.B., K.C. Mahboub, and G.A. Huber. ?Rational Method for Laboratory Compaction of Hot-Mix Asphalt?, In Transportation Research Record 1454, TRB, National Research Council, Washington, D.C., 1994. Blankenship et al (8) commented that the purpose of the N design experiment was to determine the number of gyrations (N design ) required to represent mixture densification in the actual pavement. Gyrations must relate to traffic levels and different high-temperature climates. This relationship is proven to exist and provides a method of choosing a mix design to have the blended aggregate gradation and percent asphalt binder matched to a desired traffic level in a specific climate. [For SMA mixture applications, the traffic levels are always considered to be high and asphalt binder used is usually 1 or 2 grades stiffer than that required for conventional mixes] The objective of this study was to determine the N design required to represent the various traffic levels in different geographical locations and climates. To achieve this objective, two gyration levels were evaluated: one was N construction which represents the initial laydown compaction level, C construction , and the other was N design which represents the compaction in the wheel path of pavement under traffic, C design . The value of C construction was assumed to be 92 percent of G mm due to the lack of information. 13 Eighteen pavements were evaluated, with fifteen being available for final evaluation. It was assumed that all the mixtures were designed to have about 3 to 5 percent air voids in the laboratory and 7 to 9 percent after construction. The field cores from the various pavements were first extracted and then remixed with an unaged AC-20 asphalt cement. The mixed materials were then aged for 4 hours at 135?C and compacted to 230 gyrations using the SHRP gyratory compactor. All mixtures used in this study were fine-graded mixtures. Mixtures with NMAS equal to or less than19 mm were compacted using 100 mm compaction molds while mixtures with NMAS larger than 19 mm used 150 mm molds. Some reasonable relationship trends between gyrations and climate and traffic were made in this study, generally at the high temperature climate and higher traffic level, one should use a higher N design . Analysis of the testing results provided a method of choosing N design for a desired traffic level and an average 7-day high temperature. The authors suggested that the results and conclusions from the experiment were acceptable but more research needed to be completed to increase the precision of N design . Harvey, J., C.L. Monismith, and J. Sousa. ?A Investigation of Field- and Laboratory-Compacted Asphalt ? Rubber, SMA Recycled and Conventional Asphalt ? Concrete Mixes Using SHRP A-003A Equipment?, Journal of Association of Asphalt Paving Technologists, Vol: 63, 1994. Harvey et al (9) evaluated several laboratory compaction methods and field compaction in terms of the permanent deformation resistance of compacted samples. It is believed that different lab compaction methods can produce specimens with different degrees of 14 resistance to permanent deformation as measured in lab testing. The objective of this paper was to compare the performance of lab-compacted specimens to field-compacted specimens, and estimate which lab compaction method produces specimens most similar to those produced by field compaction. The lab methods evaluated include Texas gyratory (7-inch diameter molds), University of California at Berkeley (UCB) rolling wheel, ASTM kneading (7.5-inch diameter molds), SHRP gyratory (6-inch diameter molds), and Marshall hammer (6-inch molds) compaction. Field samples were cored from three sites, with 13 test sections total, and mixtures were collected in the field at about the same locations (with one exception) for laboratory compaction. The mixtures used in the sections included conventional dense- graded mixtures, SMA mixtures, and mixtures with 30 percent of reclaimed asphalt pavement (RAP) material. The laboratory samples were compacted at the same temperatures as the field cores, except the SHRP gyratory and Marshall hammer samples were all cut and cored from larger compacted masses to the same 150-mm diameter, 50- mm high, cylindrical shape as the field cores. All specimens were tested with the Universal Testing System (UTS) and the constant height repetitive shear test for permanent deformation developed as part of SHRP Project A-003A (10). Based on the test results, it was concluded that the ranking of the methods in order of resistance to permanent shear deformation was: samples by SHRP gyratory > samples by Kneading compaction = core samples subject to some age hardening and trafficking > core samples subject to no or little age hardening and trafficking = samples by rolling wheel compaction > samples by Texas gyratory. The authors indicated that the significant 15 difference between the SHRP gyratory and the Texas gyratory was due to the different speed of rotation and gyration angle used in the two gyratory compactors. The Texas gyratory had a slower rotation and larger angle. The Marshall samples were not included in the comparison because the Marshall hammer could not achieve the air void contents obtained in the field no matter the number of blows. But even at the higher air void contents, the Marshall samples had much higher permanent shear deformation resistance than the field cores. It was also concluded that specimens produced by rolling wheel compaction best duplicated the properties of specimens compacted in the field. The rolling wheel compaction is competitive in terms of labor and materials efficiency with the SHRP and Texas gyratory compactors. Cominsky, R., Leahy, R.B., and Harrigan, E.T., ?Level One Mix Design: Materials Selection, Compaction, and Conditioning.? Strategic Highway Research Program Report No. A-408, National Research Council, Washington D.C., 1994. Cominsky et al (11) presented this SHRP report that provided the detailed background of the development of the Superpave mix design system. Specifically, this report provides a detailed description of how the Superpave gyratory compactor was selected for use in mix design and field control in the Superpave system. The major reasons that a gyratory compactor was selected for Superpave system were because gyratory compaction reasonably simulated field compaction and provided quick and economical means for a laboratory compaction procedure. After considerable research and effort, SHRP researchers selected a gyratory compactor operating with a similar protocol as the French 16 LCPC compactor for Superpave mix design system. Summaries of the development of Superpave compaction parameters are provided as following. Revolutions per Minute The French gyratory compactor operates at a speed of 6 revolutions per minute (rpm). SHRP researchers wanted to reduce the compaction time as long as the high speed didn?t adversely affect the volumetric properties of mixtures. An experiment was conducted using crushed granite (SHRP code: RB) aggregate and a PG 64-22 (SHRP code: AAK-1) asphalt from SHRP?s Material Reference Library, to compare the volumetric properties (optimum asphalt content, air void content, VMA, VFA, and density) based on the speed of 6, 15, and 30 rpm. The results showed no statistical difference between these three compaction speeds, therefore, a speed of 30 rpm was selected to reduce the laboratory compaction time. Comparisons of Gyratory Compactors An experiment was conducted to determine if it was sufficient to specify the angle of gyration, speed of rotation (30rpm), and vertical pressure (0.6 Mpa) in order to standardize requirements for the manufactories of gyratory compactors. The experiment compared the SHRP gyratory compactor (manufactured by the Rainhart Company), the modified Texas gyratory compactor, and Corps of Engineering (COE) Gyratory Testing Machine (GTM). Four aggregate blends with nominal maximum aggregate sizes ranging from 9.5 to 25 mm were selected. Two specimen sizes were evaluated: 150 mm and 100 mm. Three asphalt contents were used with one asphalt binder. All specimens were short term aged at 135?C for four hours. Compaction parameters were selected: angle of gyration (1 17 degree), vertical pressure (600 kPa), and rotational speed (30 rpm), except for the compaction angle of the GTM, which changed during the compaction. Conclusions of the experiment are shown below. 1. The modified Texas gyratory compactor and the SHRP gyratory compactor did not compact mixtures similarly. A verification of the compaction parameters indicated that the two devices were not compacting at the same angle. The modified Texas gyratory compactor had an angle of 0.97 degrees while the SHRP gyratory compactor had angles of 1.14 and 1.30 degrees for the 150 mm and 100 mm specimens, respectively. 2. A change in the angle of compaction of 0.02 degree resulted in an air voids change of 0.22 percent at 100 gyrations. This resulted in a 0.15 percent change in the optimum asphalt content for the 19 mm NMAS mixture. 3. Specifying the angle of gyration, speed of rotation, and vertical pressure alone is not sufficient to produce similar compactors. [We now know that the internal angle may be significantly different for different compactors even when the external angle is the same.] 4. Based on limited information, the COE GTM did not produce similar results to the SHRP gyratory compactor. This is mainly due to the difference in the method of applying the angle for the two compaction devices. Sousa, J.B., G. Way, J.T. Harvey and M. Hines. ?Comparison of Mix Design Concepts?. In Transportation Research Record 1492, TRB, National Research Council, Washington, D.C., 1995. 18 The goal of a mix design procedure is to combine aggregates and binder into a mix that is able to satisfy desired levels of performance. Sousa et al (12) summarized a set of mix design concepts, based on the level of complexity and ability to predict performance: ? Level 1: under a given set of conditions, mix design specimens are compacted to determine their volumetric characteristics. Aggregate and binder requirements are based on prior experience. Asphalt content is determined by volumetrics information. This is basically the approach followed by Superpave Level I. ? Level 2: mix design specimens are compacted under a given set of conditions, and a reduced set of tests are conducted. Limits of those properties are based on prior experience. Asphalt content is based on limits, ranges, or extreme values of the properties evaluated. This is basically the concept followed by the Marshall method. ? Level 3: some fundamental properties of specimens are determined with some specific preconditioning. Performance is predicted on the basis of statistical correlations between laboratory results and field observations. Asphalt content can be selected based on desired pavement performance, such as fatigue and permanent deformation. This is achievable with current state of knowledge and is basically proposed by some other researchers and used by Superpave Level III. ? Level 4: fundamental properties of the mix (and/or components) and evolution of those properties with time, aging, strain and stress levels, and moisture are determined. Prediction of behavior is made through an elaborate set of computer simulations. This approach is beyond the current state of knowledge. Asphalt 19 content would be selected based on predicted pavement performance, which would be very close to actual performance. In this paper, authors report a study conducted by the Arizona Department of Transportation to evaluate mixes designed using the Marshall, Superpave Level I, and a performance based procedure developed under SHRP-A003A. The mixture was placed in two 1-mile test sections on Interstate 17 near Phoenix, in November 1993. The major objective of this study was to evaluate the HMA component requirements for the Superpave system. This study used a PG 70-10 asphalt binder and a partially crushed river gravel (90 percent of coarse aggregate had two or more fractured faces; all fine aggregate was crushed gravel). One percent Portland cement was added to all mixtures to reduce moisture susceptibility. The gradation used in the study had 19 mm NMAS and passed through the Superpave restricted zone. Marshall stability results of 75-blow Marshall designed field mix and cores were 5044 and 3760 lbs, respectively. Both results are well above the Arizona DOT?s minimum requirement of 3000 lbs. Field mixtures were also compacted with the Superpave gyratory compactor at a compaction level of N initial (9), N design (135), and N maximum (220) using the asphalt content determined by the Marshall procedure. Results indicated that the field mix would not meet the volumetric requirements for a Superpave Level I mix design. In particular, the air void content was too high (7.6 percent and 6.3 percent, with and without parafilm, respectively), and the VFA was too low (53.3 percent). [This indicates that for this mixture, 75-blow Marshall gives more compaction effort than 135 gyrations by SGC, which is not usually observed by other researchers] An optimum asphalt content of 5.2 percent using SGC was estimated and used to produce 20 some new samples for volumetric determinations. The results showed that the mixture marginally failed the VMA and the% G mm at N initial requirements. Field cores were evaluated in the Hamburg wheel tracking device at 55?C. The results indicated that the pavement would perform well and that it would last about 10 to 15 years. The inspections of the pavements in July 1994 showed an average rut depth of 1.5 mm, which is an indication of the good performance of the mixtures, since most of the pavement failures due to rutting in Arizona usually happen during the first summer in service. An evaluation was also conducted to determine which laboratory compaction device yielded the best correlation with field compaction. Laboratory compaction devices evaluated consisted of UC-Berkeley rolling wheel compactor, the California kneading compactor, the Texas gyratory compactor, the Marshall hammer, the SHRP Rainhart gyratory compactor (Asphalt Institute), and the SHRP gyratory compactor (FHWA field trailer). Based on their permanent deformation resistance in the repeated simple shear at constant height test (RSST-CH), it was concluded that the rolling wheel compactor produced specimens that best correlated against field cores. Hafez, I.H. and M.W. Witczak. ?Comparison of Marshall and Superpave Level I Mix Design for Asphalt Mixes?. In Transportation Research Record 1492, TRB, National Research Council, Washington, D.C., 1995. Hafez et al (13) described the differences on choosing compaction level when designed by two procedures: Superpave Level I and Marshall. In the Superpave Level I mix design procedure, there is a table for gyration levels, which is dependent on the anticipated traffic volume and project site climatic conditions. These design gyrations, coupled with 21 the specific mixture gyratory densification curves developed for each mix under different asphalt contents, can be used to determine the design asphalt content. The final design asphalt content depends on traffic level and environmental conditions. In contrast to the Superpave gyratory mix design approach, the Marshall mix design uses an impact hammer to achieve the design level of compaction (air voids) as a basis for establishing the design asphalt content. The majority of agencies using the Marshall specify 35, 50, or 75 blow compaction consistent with the anticipated traffic level (?10 4 , 10 4 -10 6 , >10 6 ESALs, respectively). Thus, the final design asphalt content will only depend on traffic level. This study performed mix designs for 20 different mixes using both the Marshall procedure and the Superpave gyratory compactor Level I procedure. The mixes evaluated included dense graded mixtures and SMA-like [in this paper, it was referred to as open grading Plus Ride mixtures] mixtures. Optimum asphalt contents for all mixes in the study were determined by the Marshall 75 blow and Superpave Level I procedures. The Marshall procedure consisted of preparing three replicates at 1.0 percent asphalt content increments in order to cover an air voids range of 3.0 to 5.0 percent. The Superpave design consisted of compacting 100 mm diameter specimens at three different Ndesign values corresponding to a traffic level less than 10 million ESALs and design air temperatures of ?34?C, 37-39?C, and 43-44?C. The Ndesign values corresponding to these parameters are 67, 96, and 119 gyrations, respectively. It was concluded when the design compaction level for the SGC decreased from 119 to 67 gyrations, asphalt content increased about 1 percent for all the mixes evaluated. There were no consistent trends found between the density obtained using the Superpave 22 procedure and the Marshall procedure. [Explanation of this result and aggregate properties related to breakdown were not reported in the study.] D?Angelo, J. A., Paugh, C., Harman, T. P., and Bukowski, J., ?Comparison of the Superpave Gyratory Compactor to Marshall for Field Quality Control.? Journal of the Association of Asphalt Paving Technologists, Volume 64, 1995, pp. 611-635. In the study conducted by D?Angelo et al (14), the Superpave Level I and the Marshall procedures were compared by using five different asphalt mixes produced at five different asphalt plants. Two mixes were designed using the SGC at Ndesign levels of 86 and 100 gyrations. These two mixes were evaluated with the Marshall hammer using 112 blows (6 inch sample) and 50 blows, respectively. The other three mixes were designed using the Marshall hammer with 112 (6 inch sample), 50, and 75 blows. The SGC was used to evaluate these mixes at Ndesign levels of 100, 126, and 109 gyrations, respectively. Samples of the five mixes were obtained and compacted with both the SGC and the Marshall hammer to compare the results of the SGC and Marshall hammer when used for quality control. The results of the analysis indicated that samples compacted with the SGC had slightly less variability in air voids than did the Marshall samples. Based on air voids alone, the SGC and the Marshall hammer could both be expected to perform well in quality control applications and they would be interchangeable. As an indication of the aggregate structure, the voids in mineral aggregate (VMA) appears to distinguish between the two compaction devices. The results showed that for all mixtures tested, the SGC samples had lower VMA than Marshall samples. The general trend of lower VMA with the SGC indicates that the compaction effort obtained 23 with the SGC is greater than with the Marshall hammer. Although the high variability existed in VMA comparison, it can still be observed that the VMA determined from the SGC specimens did not respond the same as from the Marshall samples to the changes in asphalt content. For three mixtures, the slopes of the VMA to asphalt content for two compaction devices had the same trend. For the other two mixtures, the slopes had the opposite trend, with the increase of asphalt content, VMA of SGC specimens decreased, and VMA of Marshall specimens increased. This indicates that the asphalt contents are on the low and high side of VMA curve for the SGC and the Marshall compacted samples. The different trends and the high degree of variability of the data indicate that the SGC and the Marshall are not interchangeable for quality control. The overall conclusion of the study was that the SGC was better able to track plant production variability than the Marshall hammer. The mixtures designed with the SGC can not be tested and controlled in the field using the Marshall. McGennis, R.B., R.M. Anderson, D. Perdomo and P. Turner. ?Issues Pertaining to Use of Superpave Gyratory Compactor?. In Transportation Research Record 1543, TRB, National Research Council, Washington, D.C., 1996. McGennis et al (15) reported the results of the Superpave gyratory compactor study to determine the effect of various compaction parameters on the mixture volumetric properties. Parameters included mold diameter, short-term aging time, and compaction temperature. Differences between different brands of SGC were also compared. Mold Diameter The compaction characteristics resulting from two mold sizes of 150 mm and 100 mm were compared. For the comparison, five 19 mm and two12.5 mm nominal maximum 24 size aggregate blends were used, with seven gradations ranging from gap-graded to finer gradations. The optimum asphalt content for these mixes was selected to achieve 4 percent air voids at a design gyration level of 172. Three asphalt contents: optimum, optimum plus 0.5 percent, optimum minus 0.5 percent were used for compaction with two mold sizes for all seven mixtures. The volumetric properties at gyration levels of 10, 100, 150 and 250 were observed for the analysis. Two sample t-tests were performed at a level of significance of 5 percent and indicated that for 47 out of 84 comparisons (56 percent), there was a significant difference between the 150 mm and 100 mm diameter specimen. The overall trend is that the 150 mm mold produces the same or, more likely, a higher density than the 100 mm mold. Compaction Temperature Based on only one design aggregate which was predominately crushed limestone and two binders PG 64-28 and PG 76-28, it was found that variation in compaction temperature (120, 135, 150, 165 and 180?C were used) did not seem to substantially affect volumetric properties of a mixture containing an unmodified binder. However, variation in compaction temperature did significantly affect the volumetric properties of the same mixture containing a modified binder. To explain this result, authors used a hypothesis, which is: with the unmodified binder, the aggregate structure dominated the compaction characteristics of the mixture, whereas with the modified binder the binder properties dominated. Short Term Aging The study also compared the variable short-term aging period effect on the volumetric properties of mix. The results indicate the expected trend: as aging time increased, the 25 theoretical maximum specific gravity G mm increased and bulk specific gravity of mixture G mb decreased. This is because increasing the asphalt absorption will lower the effective volume of the aggregate therefore increasing the aggregate effective specific gravity G se , and G mm , and increasing the asphalt absorption will lower the effective asphalt content, thereby decreasing the compactibility of the mix and G mb . The combination of increasing the G mm and decreasing the G mb will result in higher air voids. Gyratory Compactor Comparsion This study also compared the compaction characteristics of several different units, including Pine SGC, Troxler SGC, Rainhart SGC and the modified Texas SGC. The compactor comparison testing program consisted of the preparation and compaction of three mixture specimens at design asphalt content for each of six mixtures. Specimens were prepared to determine differences in the percent G mm at N initial (10 gyrations), at N design (100 gyrations), and at N maximum (152 gyrations). The results showed that there were significant differences in volumetric properties produced among the four SGCs evaluated. At N design , the modified Texas and Pine SGCs tended to produce similar results. The Troxler and Rainhart devices also produced similar results at N design . However, the modified Texas and Pine units tended to produce lower air voids and thus lower optimum asphalt content at N design than the Troxler and Rainhart devices. In addition, the modified Texas and Pine SGC produced flatter compaction slopes than the Troxler and Rainhart SGC. Brown, E.R., D.I. Hanson, and R.B. Mallick. ?Evaluation of Superpave Gyratory Compaction of Hot-Mix Asphalt?. In Transportation Research Record 1543, TRB, National Research Council, Washington, D.C., 1996. 26 The main objective of this study (16) was to compare the density of specimens compacted with the SGC at different gyration numbers, with the density of in-place cores obtained from test sections with different levels of cumulative traffic. Field cores were obtained from six test sections having different combinations of age and traffic levels: two in Alabama (US-280 and AL-86), and one each in Idaho (I-90), New Mexico (I-40), South Carolina (I-385), and Wisconsin (STH-67). For all the sections, the cores were drilled immediately after construction and after one year of traffic. For four of six sections, cores were also obtained after two years of traffic. In the laboratory, two sets of specimens were produced using the Superpave gyratory compactor for the six sections. One set used original plant-produced loose mixture, and another set used the same aggregate and asphalt to produce mixture in the laboratory similar to the plant-produced mixture. The field core densities were compared to the density data obtained from these two sets of laboratory specimens. It was concluded that 100 gyrations of laboratory compaction produced higher density than either the 1 or 2 year in place density for all the mixes. At similar gyration levels, it was found that reheated specimens had an average about 1 percent higher density than laboratory prepared specimens that were not reheated. It was also concluded that the density of the test sections varied linearly with logarithm of cumulative traffic, an average increase of 4.8 percent of theoretical maximum density corresponding with a cumulative traffic application of 1 million ESALs. Forstie, D. A. and Corum, D. K., ?Determination of Key Gyratory Compaction Points for Superpave Mix Design in Arizona.? ASTM Special Technical Publication, Volume 1322, September 1997. ASTM, Philadelphia, PA., pp. 201-209. 27 Forstie et al (17) conducted a study to evaluate the level of Superpave laboratory compaction necessary to equal the in-place field density after various levels of traffic. The further evaluation of SHRP recommendations for the number of design gyrations was necessary because of the following reasons: 1. The angle of gyration used by SHRP researchers to develop the current levels of Ndesign was 1.0 degree, while the angle currently specified in AASHTO TP-4 is 1.25 degrees. 2. The N design experiment was conducted using 100 mm diameter specimens, not the currently used 150 mm specimens. 3. The mixes used in the N design experiment were predominately fine-graded mixes, not the coarse-graded mixes, which are most commonly used today. 4. Only two cores at each project location were obtained for testing and evaluation in the original N design experiment. More specimens may have provided a greater confidence in the field density. Field cores from seven projects on Interstate 10 were obtained within and between the wheel paths. Cores obtained from the field were tested to determine bulk specific gravity and theoretical maximum specific gravity, and then processed through the ignition oven to determine asphalt content and gradation. The salvaged aggregate from each project was re-mixed with the same amount of asphalt cement and compacted in the Troxler SGC to determine its volumetric properties at Ndesign. The Ndesign level of gyrations was determined based on the project traffic and temperature. All of the projects evaluated were from a hot or warm climate location and ranged in age from 5 to 8 years and had Ndesign values ranging from 113 to 135 gyrations. Statistical analysis (t-tests at a 28 level of significance of 5 percent) indicated that average bulk specific gravities from the Superpave gyratory compactor were significantly higher (2.355 to 2.318) than the field cores. Based on test results, it was concluded that the current Ndesign compaction levels table should be revised to account for the 1.0 to 1.25 gyration angle change that occurred after the original SHRP research. Mixes designed at the original Ndesign levels using a gyration angle of 1.25 degrees will likely have higher laboratory densities (lower optimum asphalt content) than mixes designed using a gyration angle of 1.0 degrees, which was the angle used to establish the original Ndesign levels. This over-compaction could lead to unnecessary difficulties in the field compaction. Mallick, R.B., S. Buchanan, E.R. Brown and M. Huner. ?An Evaluation of Superpave Gyratory Compaction of Hot Mix Asphalt?, NCAT Report No. 98-5, Auburn, AL, 1998. Mallick et al (18) conducted a study to evaluate some issues with the Superpave gyratory compactor. One of the objectives of this study was to evaluate the correction factor used to back calculate the bulk specific gravity of a compacted specimen at any gyration. This correction factor was determined at N maximum and usually assumed to be constant at all gyration levels. One aggregate type (traprock) and a PG 64-22 asphalt were used. Two aggregate gradations including a typical SMA and a typical dense graded mixture were used. The samples were compacted with the SGC at different gyration levels and the bulk specific gravities were determined. Back calculations were also conducted for these different gyration levels by using the correction factor determined at N maximum . These two 29 sets of bulk specific gravities were compared. It was found that the correction factor first decreased and then became constant at higher gyration levels. Densities were found to be greater than that obtained by back calculation for lower gyration samples. Also, there was a greater difference between back calculated and actual air voids for coarse textured mixtures. SMA had a greater error in back calculation of air voids compared to a dense graded mix. It was concluded that the correction factor was not constant at different gyration levels, and it was recommended that mixtures be compacted to N design to determine the optimum asphalt content. Anderson, R. M., R.B. McGennis, W. Tam, and T. W. Kennedy, ?Sensitivity of Mixture Performance Properties to Changes in Laboratory Compaction Using the Superpave Gyratory Compactor ?, Journal of Association of Asphalt Paving Technologists, Vol: 69, 2000. Anderson et al (19) presented a study to evaluate and adjust the N design table based on the sensitivity of mechanical properties (other than the volumetric properties in NCHRP 9-9). The purpose of this study was to estimate the consequences, in terms of change in performance properties, of using different numbers of design gyrations. The experimental design included six Superpave mix designs (two aggregates, three design gyrations). The two aggregate types were crushed limestone and crushed gravel. All mixtures had a NMAS of 12.5 mm. Three design gyrations were used: 70, 100, and 130 to represent low, medium, and high traffic level, respectively. 30 The performance properties used in this study included complex shear modulus (G*) from the shear frequency sweep test and permanent shear strain (? p ) from the repeated shear test at constant height. The results indicated that a general trend can be observed of decreasing shear stiffness (G* 10Hz ) with decreasing design gyrations. It was concluded that with other factors (asphalt content, asphalt binder stiffness and volume of air voids) held relatively constant, this trend reflected the change in aggregate structure produced as design gyrations were reduced. This trend was more notable as N design changed from 100 to 70 gyrations. It was observed that a decrease of 30 gyrations could lower the design shear stiffness of an asphalt mixture by as much as 35 percent, and approximately 15 percent for average. It was also observed that limestones have a significantly lower permanent shear strain (? p ) value (average 1.29 percent) at 60?C than the gravel mixtures (average 2.02 percent). [No explanation was given in the literature and aggregate properties were missing] However, for each aggregate, there were no significant differences in permanent shear strain apparent between mixtures designed at 130 gyrations and 70 gyrations. Buchanan, MS and E.R. Brown, ?Effect of Superpave Gyratory Compactor Type on Compacted Hot-Mix Asphalt Density? In Transportation Research Record 1761, TRB, National Research Council, Washington, D.C., 2001 Buchanan et al (20) presented an effort to evaluate the effect of SGC type on compacted HMA density. One major concern was the degree of reproducibility between laboratories having different brands of approved SGC. The objective of this study was to present 31 observed precision values and practical differences between the currently used SGC and to discuss the possible project implications of the differing results. The data used in this study include the Southeast Superpave Center gyratory compactor proficiency sample testing, results from the initial six projects from NCHRP 9-9: verification of the gyratory levels in the N design table, and mix design and quality control (QC) and quality assurance (QA) results from a state DOT. Based on the analysis of the observed data, the following conclusions are drawn: 1. The SGC has higher precision for both the single- and multi-laboratory compared to the past data with the mechanical Marshall hammer. Average values for the gyratory compactor of 0.0094 and 0.0133 were determined for single- and multi- laboratory precision, respectively. For comparison, the mechanical Marshall hammer had an average single-operator precision of 0.012 and a multi-laboratory precision of 0.022 (21). 2. Significant difference in compaction efforts was observed for different SGCs, even after proper calibration. This difference resulted in significant differences in air void content (in some cases up to 2 percent) of compacted samples, significant differences in optimum asphalt content (up to 1.3 percent) between designed mixtures and verified mixtures, and significant differences between QC and QA air void results (0.79 percent). It is indicated that the magnitude of changes in the gyration angle during compaction contributes to the difference of compaction efforts for different SGC. Therefore, it is suggested by the authors that a protocol for an independent gyration angle-measuring device should be developed as soon as possible to ensure the gyration 32 angle of all gyratory compactors remains within the specification range of 1.23 to 1.27 degrees during compaction, and consequently to ensure all gyratory compactors are providing similar compactive efforts. Tashman, L., E. Masad, R. Peterson, and H. Saleg. ?Internal Structure Analysis of Asphalt Mixes to Improve the Simulation of Superpave Gyratory Compaction to Field Conditions?. Journal of the Association of Asphalt Paving Technologists, Vol. 70, 2001. Tashman et al (22) compared laboratory compaction with SGC to field cores that were compacted using three field compaction patterns. The internal structure, i.e. the distribution of aggregates and air voids inside the mix was used to evaluate the similarity of laboratory compacted samples and field cores. Computer automated image analysis techniques and X-ray computed tomography were used to capture and quantify the internal structure distribution. The properties related to internal structure included in this study were air voids distribution, aggregate orientation, aggregate segregation and aggregate contacts. Three test sections were constructed for this study. Each section was about 90 meters in length. A transition section of about 30 meters was placed between each test section for changing directions or stopping. The three sections used the same materials (limestone and a PG 70-22 asphalt with Superpave 12.5mm NMAS gradation) and a similar construction temperature. The variables among these sections were the compaction equipment and the number of passes used in order to achieve target air voids of 7 percent. 33 Field cores were obtained from the test sections directly after construction and before trafficking. Loose mix samples that represent the three sections were obtained from loaded trucks and tested for asphalt content and aggregate gradation, to ensure the laboratory samples had similar material properties as in the field. All mixes were reheated to the field compaction temperature (149?C) before compacting in the gyratory compactor. The gyratory compactor used in this study had various levels of several key parameters, including the angle of gyration (1.25, 1.5 and 2.0 degree), height of specimen (50, 75 and 135 mm), and compaction pressure (400, 600 and 800 kPa). Another temperature (175?C) of base plate and mold was also used at an angle of 1.25 degree, 50 mm height, and 600 kPa pressure for studying air voids distribution. The results indicated that different field compaction patterns didn?t affect the internal structure much. However, the compaction parameters (angle, pressure, height, and temperature) of SGC influenced the internal structure of lab compacted samples and changes of these parameters could make the internal structure of lab samples similar to field cores. The findings suggest the use of an angle of 1.5 degree and a specimen height of 50 mm to 75 mm in the SGC to better simulate the internal structure of field cores. Authors also indicated that this specimen height may be affected by the NMAS of mixture, which is 12.5 mm in this study. Pressure was not a significant influencing factor for the internal structure. However, using 600 kPa gave the closest results to field cores. The stiffness test using the shear frequency sweep test at constant height (FSCH) was also conducted to verify the usefulness of simulation in internal structure. The results showed that improving the simulation of field cores in SGC samples reduced the difference in stiffness between SGC samples and field cores. 34 Harman, T., J.R. Bukowski, F. Moutier, G. Huber, and R. McGennis. ?The History and Future Challenges of Gyratory Compaction 1939 to 2001?. In Transportation Research Record 1789, TRB, National Research Council, Washington, D.C., 2002. Harman et al (23) reviewed the history of gyratory compaction and presented the evolution of gyratory compaction as shown in the Table 2.3. TABLE 2.3 Evolution of Gyratory Compaction (23) Timeline Device/Agency Specimen Size Compaction Effort 1939 Concept TX Highway Department D ? 4? H ? 2? P ? Unknown A ? Manual S ? Manual 1946 TX Highway Department D ? 4 & 6? H ? 2 & 3? P ? Variable A ? Fixed 6? S ? 60 rpm 1957 US Corps Engineers GTM D ? 6? H ? Variable P ? Variable A ? Floating 0 to 3? S ? Variable 12 to18 rpm H ? Heated mold 1960?s First Prototype Texas at LCPC, France D ? ? H ? ? P ? Variable A ? Variable S ? Variable 1968 Second Prototype Texas at LCPC, France D ? 80 or 120 mm H ? Variable P ? Variable A ? Floating 0.5 to 5? S ? Variable H ? Heated mold 1974 to 1985 PCG1, PCG2 at LCPC, France D ? 160 mm H ? Fixed 80 to 300 mm P ? 600 kPa A ? Fixed 1 to 4? S ? Fixed 6 to 30 rpm H ? Heated mold 1991 Modified Gyratory Shear Test Machine, FHWA D ? 4? H ? 2.5? P ? 600 kPa A ? Fixed 0.5 to 3? S ? 30 rpm 1991 Modified TX Highway Department, SHRP D ? 6? H ? 3.75? P ? 600 kPa A ? See History S ? Variable H ? Heated mold 1993 SHRP/Superpave Gyratory Compactor in USA D ? 150 mm H ? 115 mm P ? 600 kPa A ? Fixed 1.25? S ? 30 rpm 1996 PCG3 at LCPC, France D ? 150 mm H ? Fixed 100 to 160 mm P ? Fixed 500 to 800 kPa A ? Fixed 0.5 to 2? S ? Fixed 6 to 30 rpm Key: D ? diameter, H ? height, P ? consolidation pressure, A ? external mold wall angle, S ? speed of gyration, and H ? heated mold. 35 It is reported that over 2000 SGCs manufactured by five companies are currently in use in the United States for design and field control of asphalt mixtures. A total of eight different models from the five companies are using totally different methods of setting, inducing, and maintaining the angle of gyration. A calibration system is required in each device to measure the angle of gyration. All measurements are made externally relative to the mold wall. However, the difference in internal angle of gyration is believed to result in the different compaction effort for different types of SGC even when the external angle is controlled. In response to this issue, FHWA has developed an angle validation kit (AVK) to measure the internal angle of gyration in any SGC. However, before the AVK can be considered in standard practice, the target and tolerances for a standard internal angle must be established. Peterson, R.L., K.C. Mahboub, R.M. Anderson, E. Masad and L. Tashman. ?Superpave Laboratory Compaction versus Field Compaction?. In Transportation Research Record 1832, TRB, National Research Council, Washington, D.C., 2003. Peterson et al (24) presented a follow-up study on evaluation of Superpave compaction versus field compaction for the previous one (19). The mechanical properties were used to evaluate the similarity of laboratory compacted samples and field cores. The field compaction consisted of three test sections with different compaction patterns. The laboratory compaction used the Superpave gyratory compactor with adjustments to several parameters, including sample height, compaction pressure and gyration angle. Performance tests conducted with the Superpave shear tester in accordance with AASHTO TP 7 [updated version is AASHTO T320 (25)] were used to evaluate field and laboratory compaction. Frequency sweep at constant height (FSCH) testing was 36 conducted at 30?C and 40?C. Repeated shear at constant height (RSCH) testing was conducted at 50?C. It was concluded that current gyratory protocol produces specimens with significantly different mechanical properties than those of field cores produced with the same material and compacted to the same air voids. It was indicated that adjustments to certain parameters of the gyratory can produce specimens that better simulate the mechanical properties of pavement cores. It was suggested that the use of 1.5 degree angles and a specimen height of 50 or 75 mm would better simulate mechanical properties of roadway cores tested in the Superpave shear tester. 2.1.2 Summary of Literature Historically, three compaction methods: impact compaction (Marshall), kneading compaction (Hveem) and gyratory compaction (SGC) have been used in routine HMA mix design. Several studies (5-7, 9, 12-14) were conducted to determine the best available laboratory compaction device that could simulate field compaction. The results indicate that the gyratory compaction is a promising compaction procedure in simulating field degradation, density and engineering properties. During the Strategic Highway Research Program (SHRP), intense debate focused on the effectiveness and appropriateness of the gyratory compaction. Some research studies (9, 12) indicated that the rolling wheel compactor produced specimens that best correlated against field cores. However, other studies (7, 11) indicated the impracticality of the rolling wheel compactor as a means of mix design compaction. The equipment proposed was large and required very large batches of mixture. Although the performance test results of the gyratory specimens did not correlate with field cores as 37 well as rolling wheel compacted specimens, the air voids produced during compaction were very repeatable and the compactor was easy to use. This ease of use and repeatability are desirable because mix designs are based on the volumetric properties of the specimens. Two major factors favor using a gyratory compactor for laboratory compaction for mixture design: 1) the close simulation of field compaction (5-7, 14), including volumetric properties, engineering properties, and aggregate breakdown; and 2) the ease of use and good repeatability (5, 7). Gyratory compaction was first developed in the 1930s in Texas (5). The compaction process involved applying a vertical load while gyrating the mold in a back- and-forth motion. This gyratory compactor produced a kneading action on the specimen by gyrating the specimen though a horizontal angle. The gyratory compaction has evolved since its first development, resulting in several unique devices and a variety of methods (23). Compaction using gyratory action was further developed and applied by the Army Corps of Engineers as well as the Central Laboratory for Bridges and Roads (LCPC) in France. The Superpave gyratory compactor (SGC) was developed during the SHRP program. It is similar to the French gyratory but with some modifications. It operates with a vertical load of 600 kPa, a gyration speed of 30 rpm, and a constant angle of gyration of 1.25 degrees. The gyration angle and applied pressure were found to be critical for consistent compaction results. Some studies (22, 24) have shown that adjusting these compaction parameters would make gyratory samples better simulate field compaction. Some studies 38 (11, 15, 20, and 23) have also indicated that different brands of SGCs result in different compaction results, and the difference in internal angle was believed to be one of the primary reasons. Therefore, it is necessary to calibrate the internal angle during compaction in order to get consistent compaction results between different brands of compactors (20, 23). An angle validation kit (AVK) has been developed as a response to this issue (23). 2.2 THE DEVELOPMENT OF SMA MIX DESIGN 2.2.1 Individual Literature The literature reviews in this section are conducted to specifically answer the following questions: ? What are the design criteria for designing SMA mixture? ? What is the compaction effort used for designing SMA mixture? ? What are the conclusions and recommendations of researchers who have evaluated the compaction effort used in SMA mix design? ? What are the performance tests used to evaluate SMA mixtures? AASHTO, ?Report on the 1990 European Asphalt Study Tour?, Washington D.C., June 1991. This report (1) is an outcome of the European Asphalt Study Tour (EAST) in 1990. One of the objectives of this study tour was to review and evaluate European pavements and asphalt technology. A section of this report addresses SMA. The report describes the origin of SMA and its development. Some general design criteria that had been in common practice in Europe are also summarized. 39 SMA was first developed in the 1960?s in Germany as an HMA mixture that was especially resistant to studded tire damage. The term splittmastixasphalt is used in the German ?Supplemental Technical Specifications and Guidelines on Asphalt Surface Course.? SMA continues to evolve in Europe, and was introduced into several European countries including Sweden and Denmark. The report indicates that SMA mix design in Europe generally follows a recipe-type approach from standard designs. In Europe, the Marshall mix design method is usually used for voids analysis and the selection of target bitumen content. For a selected aggregate gradation, the Marshall specimens are compacted at 50 blows per side for several asphalt contents mixed at 135?5?C. The asphalt content that yields an air void content of 3 percent is selected as the target value for the job mix formula. European specifications require a maximum 6 percent of air void content for SMA field construction. Stuart, K.D. ?Stone Mastic Asphalt (SMA) Mixture Design?, FHWA-RD-92-006 Federal Highway Administration, March 1992. Stuart (26) documented SMA mixture design information from Europe and a SMA project conducted by the Georgia Department of Transportation. There are two chapters in this report. Chapter 1 presents information on European SMA mixture design technology, which was obtained primarily from sources in Sweden and Germany. The Marshall method of mixture design is used in Sweden and Germany for designing SMA mixtures. However, stability and flow values are often not used, many designs are based on air void requirements and minimum asphalt content. The compaction temperature in the lab in Sweden is generally between 293 and 302?F (145 and 150?C) and rarely exceeds 311?F (155?C). But when a fiber is used as the stabilizing 40 additive, temperatures up to 338?F (170?C) are allowed. Germany generally uses a temperature of 275?F (135?C). Neither loose mixtures nor compacted specimens are aged in an oven in either country. Based on their experiences, the Swedes and the Germans to date are satisfied with the 50-blow Marshall compaction effort for mix design. Increasing the number of blows is not recommended by them because this may increase the number of fractured aggregates with little to no increase in density. Therefore, all optimal binder contents reported by the Swedes and Germans and minimum binder contents contained in their specifications were obtained through a 50-blow Marshall compactive effort. It is also indicated that SMA design air void levels are often slightly lower than those used for dense-graded mixtures. The high stone-on-stone contact allows the use of lower air void levels. There are no requirements for voids in the mineral aggregate (VMA) or voids filled with asphalt (VFA). However, there is a minimum asphalt content requirement which has the similar overall effect. For a nominal maximum aggregate size (NMAS) of 12.5 mm, these properties are generally above 16.5 percent and 78 percent, respectively. The air voids level requirement in an SMA pavement layer after compaction is lower than for a dense-graded mixture using the same type of aggregate and maximum aggregate size. The Swedes and Germans report that field air void levels are typically 3 to 5 percent and are specified to be less than 6 percent. When the design air voids level is 3 percent, mixtures with fibers often compact to a 3 to 4 percent level. It is also indicated that SMA mixtures at their ultimate or refusal density after traffic will not have air voids 41 levels significantly lower than the design level. This may be because SMA normally uses the 50-blow Marshall hammer which provides adequate compaction. The author reports that very little evaluation testing has been done on SMA in Europe. At the time of writing the report, the author indicated that research was underway in Germany and the Netherlands on the resistance of SMA mixtures to permanent deformation using creep, repeated load, and wheel-tracking tests. Chapter 2 documents the work performed for Georgia Department of Transportation (GDOT). GDOT performed mixture designs and requested assistance from other organizations with designs. The lab mixing temperature was 325?F (163?C) and the compaction temperature was 310 to 315?F (154 to 157?C). These temperatures were based on past experiences of GDOT using Novophalt-modified binders. It was indicated that there was no firm basis for choosing mixing and compaction temperatures in this project. A 50-blow Marshall design was used. The optimum binder content was taken at a 3.5 percent (requires a 3 to 4 percent design air voids level) air voids level. Kennepohl, G.J., and J.K. Davidson. ?Introduction of Stone Mastic Asphalt (SMA) in Ontario?, Journal of Association of Asphalt Paving Technologists, Vol 61, 1992. Kennepohl et al (27) presented the results of SMA mix designs for three projects in Ontario, which were constructed between December 1990 and October 1991. Three projects were Miller Avenue project, Highway #7N project and Highway #404 project. The Marshall mix design method was used in all three projects. The mixes were designed to have an asphalt content that would give a value of 3 percent air voids. The compactive effort to achieve the proper air voids was the equivalent mechanical blow 42 count to provide a density equal to that provided with the 75-blow hand hammer or 50- blow hand hammer. In the Miller Avenue project, 75-blow was chosen because it was felt that the expected heavy traffic conditions would cause over-densification of the mix. SMA mixtures were used in both surface and base courses. The mixing and compacting temperature for these designs were 150?C and 145?C, respectively. For the other two projects, the samples were compacted with 37 blows mechanical (compactor has a rotating base and a beveled foot, equivalent with 50 blows manual) on each side at a temperature of 135?C. Several slabs were removed from the Miller Avenue and Highway #7N projects for rutting tests, using the Ministry of Transportation (MTO) wheel tracking machine. The MTO rutting test is done at controlled temperature of 60?C using a rubber tired wheel rim along the test slab for 4000 cycles. The results indicated that the SMA surfacing and base courses demonstrated significantly better rutting resistance compared with the control section of asphalt concrete. In the discussion on comparing the results between 50 and 75-blow Marshall designs for SMA, the authors stated: ?The 50 blow seems to do what we want it to do. Based on the results we are getting in the field, it is more than adequate?. Brown, E.R. ?Evaluation of SMA used in Michigan (1991)?, NCAT Report No. 93-3, National Center for Asphalt Technology, Auburn, AL, 1993. Brown (28) evaluated the sensitivity of SMA mixture properties to changes in proportions of various mixture components. This study was performed using the same materials and job mix formula as that used in the Michigan project, which is one of the first SMA projects conducted in the U.S. The mixture components that were varied to 43 evaluate sensitivity included amount of cellulose fiber, asphalt content, percent passing No.4 (4.75 mm) sieve, and percent passing No. 200 (0.075 mm) sieve. The properties evaluated in this study included tensile strength, Marshall stability and flow, Gyratory properties (gyratory stability index, gyratory elastic plastic index, and shear stress to produce 1 degree angle), resilient modulus, confined creep, and volumetric properties including voids, voids in mineral aggregate, and voids filled. It was indicated that all the test samples in this study were compacted in the Corps of Engineers Gyratory Testing Machine (COE GTM), which was set to produce a density equivalent to 50 blows with the Marshall hammer. Based on a calibration curve, the gyratory machine was set at 120 psi, 1 degree angle and 120 revolutions. [The calibration information was not given in the report] It was concluded that in most laboratory tests, the HMA performed better than the SMA. However, these tests had not been shown to be closely related with performance. SMA performed reasonably well in the confined creep test and generally performed better than the dense-graded HMA in the procedure for the gyratory shear stress to produce one degree angle. The author notes that these two tests are indicators of rutting resistance and appear to be the best tests conducted for predicating performance of the SMA mixture. Brown, E.R. ?Experience with Stone matrix Asphalt in United States?, NCAT Report No. 93-4, National Center for Asphalt Technology, Auburn, AL, 1993. Brown (29) summarized the construction information of the first five major SMA sections placed in the U.S. in 1991. These were in Georgia, Indiana, Michigan, Missouri, and Wisconsin. Some mix design and construction control data are presented. 44 All mix designs of these five SMA sections were performed with the 50 blows Marshall compaction. It was indicated that SMA mixtures compact quickly, so additional blows would not likely significantly increase the density but only cause excessive breakdown in the aggregate. It was summarized that SMA mixtures have been designed to have as low as 3 percent voids in some cases and as high as 4 percent voids in others. The author commented that hotter climates should probably design closer to 4 percent voids, and cold climates should design closer to 3 percent. The author also indicated that SMA appeared to be able to tolerate lower voids better than dense-graded HMA, but a minimum 3 percent air voids was desirable to prevent the potential of rutting problems. Brown, E.R., and H. Manglorkar. ?Evaluation of Laboratory Properties of SMA Mixtures?, NCAT Report 93-5, National Center for Asphalt Technology, Auburn, AL, 1993. Brown et al (30) evaluated laboratory properties of SMA by using some tests developed for dense graded mixtures. These tests include Marshall stability and flow, gyratory properties (include gyratory shear index, gyratory elasto plastic index and shear stress to produce 1? angle), resilient modulus at various temperatures (40, 77, and 104?F or 5, 25, and 40?C), static and dynamic confined creep at 140?F (60?C), and indirect tensile strength at 77?F (25?C). One major objective of this study was to evaluate the potential of those tests to predict the performance of SMA mixtures. Two typical aggregate types, a granite with a L.A. abrasion value of 35% and a siliceous gravel with a L.A. abrasion value of 46.5%, were used in this study. The asphalt cement was AC-20 grade. Three fibers were used, including U.S. and European cellulose 45 fibers, and European mineral fiber. The filler was obtained by screening a local agricultural lime. The optimum asphalt content was selected at the design air void content of 3.5 percent by 50 blows Marshall compaction. The samples were compacted at optimum asphalt content using the Corps of Engineering Gyratory Testing Machine set at 75 revolutions (the machine had a vertical pressure of 120 psi (0.83 MPa) and 1 degree gyration angle). This compaction level was selected because it provided similar density as 50 blows Marshall hammer compaction [In literature 27, the equivalent compaction level using GTM was 120 revolutions]. It is also reported that dense graded mix samples used to compare with SMA samples were compacted using 300 revolutions by the same gyratory machine. [This indicates the dense graded mix samples were compacted at a much higher compaction effort than SMA mixtures, which is also mentioned in literature 30, 33.] It was concluded that only some of the tests may have the potential to predict the performance of SMA mixtures. These tests included gyratory shear, confined creep, and permanent deformation by means of dynamic creep. [Similar conclusion is drawn in literature 27, by using the Michigan Project material] Carpenter, R.H. ?Mix Design Considerations for SMA Mixes?, Presented in Transportation Research Board Annual Meeting, 1994. Carpenter (31) believed that a performance test was necessary for SMA mixtures to determine at what extent the stone-on-stone skeleton has been developed. One can not assume that two different mixes prepared with the same gradation would develop the same degree of a stone skeleton. There were significant differences in these mixtures 46 even though the gradations were precisely controlled by blending individual sieve sizes. Therefore, gradation alone cannot be relied on when different aggregate types are used. The author commented that SMA should not be significantly densified during compaction if properly designed. However, if the gradation does not allow a stable stone skeleton developed in the mixtures, the sand/asphalt matrix will experience densification during compaction and the volumetric properties will show a distinct relation with asphalt content. A real SMA mixture should not show a great effect of compaction on volumetric properties because there is essentially no densification on the matrix. Thus, the change of asphalt content will not significantly change compaction characteristics. For example, the VMA curve should not change much with the change of asphalt content in an SMA mixture. Because there is little densification, the VMA should be relatively constant, and the void space between the aggregates in the sand sized fraction should remain unchanged as the extra asphalt cement being added would only serve to reduce air voids, and not alter compactability of the aggregate portion of the mixture. The author cites two cases in Illinois, one of them had developed a problem. In the problem mix, the VMA increased with increasing asphalt content, whereas, in the other mix the VMA was virtually stable at between 15.5 and 16% for asphalt contents between 6 and 7.5%. To illustrate the presence of a skeleton, samples at optimum asphalt content (designed at 3 percent air voids) were compacted at various compactive efforts of 35, 50, 75, and 110 blows per side with the Marshall hammer. The change in air void content (VTM) is plotted in Figure 2.1. 47 Based on the test results, it was concluded that compaction at 50 blows should be selected as the minimum level for developing the SMA skeleton structure in the lab. The density increase from 50 to 75 blows may not be significant enough to recommend a higher compaction effort. One should never use compaction effort higher than 75 blows. FIGURE 2.1 Effect of increase in compaction effort on limestone SMA mixture (31). , ct of the SMA Technical Working Group (TWG) and provides general guidelines for the use of Design air void content (VTM) was set at 3 to 4 percent by using 50-blow Marshall compaction. A minimum asphalt content of 6 percent and a minimum VMA content of 17 percent were also specified. ?Guidelines for Materials, Production, and Placement of Stone Matrix Asphalt? National Asphalt Pavement Association, Information Series 118, 8/94, 1994. This publication (2) is a produ SMA paving mixtures, including composition of an SMA mixture, aggregates and additives, production, hauling and paving, and compaction. A set of volumetric requirements of SMA mixtures was set in this guideline. 48 Mixing temperature was also specified in this guideline. Asphalt cement sha mixed at a temperature as required to achieve a viscosity of 170?20 centistokes. Typical plant mixing tempera ll be ture for SMA is 155 to 163?C (310 to 325?F). However, at no time nology, Aubur ratory ). ntent was selected at 3.0 percent air voids. Samples were compac s evolutions in the GTM, set at one degree and 120 psi, produc r. shall the mixing temperature exceed 177?C (350?F). Brown, E.R., and R.B. Mallick. ?Stone Matrix Asphalt- Properties Related to Mixture Design?, NCAT Report No. 94-2, National Center for Asphalt Tech n, AL, 1994. Brown et al (32) conducted a study on SMA volumetric properties. One of objectives of this report was to develop a relationship of laboratory densities of SMA mixes prepared by a gyratory machine to those prepared by the mechanical Marshall hammer. Two aggregates, granite and limestone, were used to do the comparison. Gy compaction was performed by the Corps of Engineers Gyratory Testing Machine (GTM Mix designs were conducted by using 50 blows with the mechanical Marshall hammer, and optimum asphalt co ted with the GTM at various numbers of revolutions by using the optimum asphalt content obtained by the Marshall method of design. The number of revolutions required to produce 3.0 percent air voids was determined from plots of air voids versu number of revolutions. The test results showed that for gravel mixes, 73 revolutions in the GTM correlated with 50 blows with a mechanical Marshall hammer in terms of air voids; for limestone approximately 103 r ed similar air voids as produced by 50 blows with a mechanical Marshall hamme [The authors did not give the possible reason why these two aggregates gave two 49 different numbers, some basic aggregate properties such as L.A abrasion, F&E content were not listed in the report.] It was concluded that 90 revolutions in the GTM is a reasonable estimate o blows mechanical Marshall for SMA mixes. It was also addressed that the 90 revolutio required for SMA is much less than the 300 revolutions that have been used to compa dense graded mixtures to a density equivalent to 75 blows Marshall compaction. Dynamic creep tests were performed on a number of mixes with different percentages passing the #4 sieve. A dense-graded mix with 5.1% asphalt content was used for comparison. [Test temperature and confining condition are not stated.] The results showed that both gravel and limestone SMA mixtures had higher strain valu lower creep modulus values compared with the corresponding dense-graded mixes. Th authors noted that these findings are contrary to observe f 50 ns ct es and e d field performance. It may be d e; at this gradation evaluated mixtures ional study on evaluation of SMA versus dense-graded mixture explained by the significant difference in optimum asphalt content between SMA an dense-graded mixes. The optimum gradation for the two aggregates appears to be approximately 25 percent passing the No.4 siev provided the highest creep modulus and lowest strain. Mogawer, W.S., and K.D. Stuart. ?Evaluation of Stone Matrix Asphalt versus Dense-graded Mixtures?, In Transportation Research Record 1454, TRB, Nat Research Council, Washington, D.C., 1994. Mogawer et al (33) conducted a s. The objectives of this study were to 1) compare SMA with dense-graded mixtures in terms of their resistance to rutting, moisture damage, low-temperature cracking, and aging; 2) determine which mechanical tests can be used to predict the rutting susceptibility of SMA. 50 Dense-graded mixtures and SMA mixtures with NMAS of 12.5 mm and 9.5 mm were used. The aggregate was a crushed diabase and asphalt binder was AC-20. M mix design procedure was used to determine the optimum asphalt contents. The SMA mixtures were compacted using 50 blows and targeting 3 percent air voids. The de graded mixtur arshall nse- es were compacted using 75 blows and targeting 4 percent air voids. The Corps o 20 psi on the SMA gradations were also in a significant increase in the percentage of aggregate passing t 4.75 nd No 6 mm . At these two sieves, Marshall compaction broke slightly more aggregate than did the GTM. The results are shown in the Table TABLE 2.4 Aggregate Gradations of Mixtures Tested (33) SMA 9. NMAS SMA 12 NMAS f Engineers gyratory testing machine (GTM) was used to evaluate rutting resistance. The mixtures were tested at 60?C, and tested with a vertical pressure 1 (0.83MPa) and a 1? (0.0175-rad) gyratory angle. Samples were compacted to 300 revolutions. Extractions were performed on the SMA mixtures before and after the GTM testing. The effect of the Marshall compaction examined. Both compactions fractured the aggregate and resulted he No.4 ( mm) a .8 (2.3 ) sieves 2.4. 5mm .5mm Sieve Size D n M l D n M l mm esig GTM arshal esig GTM arshal 19 100 100 100 12.5 100 100 100 95 95.5 94.6 9.5 95 95.1 94.5 71 74.5 76.1 4.75 46 50.8 52.9 25 33.4 34.4 2.36 25 29.6 31.0 20 24.1 24.2 1.18 20 23.4 23.4 18 20.7 20.5 0.6 16 19.0 18.8 16 18.1 17.9 0.3 13 15.6 15.4 13 15.2 14.9 0.15 12 13.7 13.6 12 13.3 13.3 0.075 10 11.4 11.2 10 10.9 11.1 51 Several testing devices were used to evaluate rutting potential, including LCPC pavement rutting tester, Georgia loaded wheel tester (GLWT), GTM, and unconfined repeate a, A had more permanent deform .6 test will be less susceptible to low temperature cracking. After aging, both 12.5 and 9.5 mm NMAS dense-graded mixtures had significant increases in dynamic modulus and tensile strength results compared with both SMA mixtures, indicating that the dense-graded mixtures might be more susceptible to cracking after aging than the SMA mixtures. d load tests at 40?C on samples compacted to design air voids by kneading compaction. The vertical stress was 0.45 MPa. Two confined repeated load tests were also conducted on the 12.5 mm NMAS SMA. The confining pressure used was 0.14 MP with 0.45 and 0.31 MPa deviator stress. The results indicated that there was no significant difference among SMA and dense-graded mixtures when using the LCPC rut tester, the GLWT, and the GTM. SMA mixtures can pass LCPC rut tester and the GLWT, however, SM ation than dense-graded mixtures by unconfined, compressive, repeated load test, and applying a confined pressure did not improve the results. [No information on whether membrane was used for the confined test.] The authors concluded that a test using 101 mm by 203.2 mm specimens might not be applicable to SMA. Other properties evaluated in this study were moisture damage, low temperature cracking and aging cracking. The visual observation of stripping and tensile strength results indicated that the SMA mixtures were more resistant to moisture damage than dense-graded mixes. 12.5 mm NMAS SMA showed significantly lower diametral modulus than the 12.5 mm NMAS dense-graded mixture, which indicated that SMA 52 Partl, M.N., T.S. Vinison, R.G. Hicks, and K. Younger. ?Performance-Related Testing of Stone Mastic Asphalt? Journal of Association of Asphalt Paving Technologists, Vol 64, 1995. Partl et ures g evaluat oning System (ECS), and the . showed that six of the fourteen SMA specimens tested display micro- d ens al (34) evaluated the influence of several material parameters of SMA mixt by using several selected Strategic Highway Research Program (SHRP) tests and agin conditioning methods. The influence of long-term oven aging (LTOA), low temperature cracking, resilient modulus, rutting, and moisture sensitivity of SMA mixtures were ed. The test methods used to evaluate those properties included thermal stress restrained specimen test (TSRST), the indirect tensile test (IDT), the constant height repetitive simple shear test (CHRSST), the Environmental Conditi Laboratoire Centrale des Ponts et Chaussees (LCPC) wheel tracking device Two types of SMA mixtures were investigated: slabs from a road in Switzerland, and laboratory samples produced with two extreme air void contents using the same aggregate gradation as the Swiss SMA. The laboratory prepared SMA mixtures were compacted with a kneading and a steel wheeled roller compactor. The TSRST results ed a drop in stress without a clear fracture. This is possible due to the continuous change of interlock within the aggregate skeleton combined with local multiple fracture resulting in a successive redistribution and reorientation of stresses. It was note by the authors that this behavior did not occur regularly and was observed on specim with and without LTOA. 53 The CHRSST frequency sweep test results indicated that the shear phase angle decreases as frequency decreases. The shear phase angle also decreases at low frequen se- deform isible shear flow zo permeable. However, water was found in the center of all the specimens after testing and splitting. The specimens with 4.8 percent air voids showed no stripping, whereas those with high air void contents showed some stripping. It is concluded that because of the coarse aggregate skeleton of SMA mixtures, conventional laboratory test procedures will need modifications in order to properly evaluate SMA mixtures. Appropriate confinement and the use of realistic specimen dimensions are needed. cies as temperature increases. The authors concluded that SMA behaves like a viscoelastic solid and thus may show different behavior when compared with den graded HMA due to its different structure. The CHRSST cumulative permanent ation test results indicated that two SMA specimens with low air voids (2.6%) required more cycles (3600 and more than 5000 compared to 600 and 800 cycles) than those with high air voids (7.7%) to reach a strain limit of 5.5 percent. The LCPC tested specimens were found to deform laterally, and had v nes under the wheel track. It indicates that an aggregate skeleton without sufficient lateral confinement and interlocking becomes unstable and tends to shove. The ECS test results showed that SMA specimens with lower air voids were found to be practically im West, R.C. and B.E. Ruth. ?Compaction and Shear Strength Testing of Stone Matrix Asphalt Mixtures in the Gyratory Testing Machine?, Journal of Association of Asphalt Paving Technologists, Vol: 64, 1995. 54 West e dure n, terization study of SMA mixtures. Eleven SMA mixtures were compacted in the laborat r h study, to produce samples having similar initial in-place air void contents, the foll ee t al (35) presented an effort to identify a more appropriate design proce compared to the Marshall procedure for SMA mixture. The authors first stated that there are some existing problems in SMA mix design by 50 blows Marshall hammer, which include 1) the lack of correlation between the Marshall compaction and field compactio resulting in adjustments of the binder content during the field production of SMA, and 2) the lack of basis for evaluating SMA performance in the Marshall procedure. This paper investigated the results from a laboratory compaction and charac ory with a Corps of Engineering gyratory testing machine (GTM) with an air rolle to simulate initial construction and traffic densification. It is believed that with the air roller the strain applied to a mix is an indication of stability of the mix. When applied to a stable mix the strain would decrease with an increase in shear strength of the mix, whereas the strain would increase with loss in shear strength in the case of a low strength mix. One objective of this study was to determine a standard compactive effort which could be used to prepare future SMA mixtures for design and evaluation. An analysis of variance was performed on the available data to determine the number of gyrations whic produced sample air void contents closest to the initial field air voids for each mixture. Based upon this owing compactive effort using the Model 6B/4C GTM appears to provide the best results for SMA mixtures: 9 psi initial air roller pressure, 100 psi ram pressure, 3 degr initial angle of gyration and 12 gyrations. It is noted that the compaction effort for SMA mixtures is less (i.e. fewer gyrations) than typically used for dense-grade mixtures, and 55 this finding reflects the practice of using 50-blow Marshall rather than 75 blows to des SMA mixtures. Using the above settings, three compacted samples from each SMA mixture wer further tested at 60?C to 300 gyrations with height and air roller pressure recorded at periodic intervals of gyrations. Two parameters, percent densification and gyratory sh strength, were determined from recorded information as a measure of mixture quality. To evaluate the rutting susceptibility of mixes, the percent densification and the gyratory shear strength at 200 gyrations were determined. Percent densification is defined as the change in air voids of the sample at any point during densifi ign e ear cation from initial in-place density ents stable t lumetric design and or further strength testing can be accomplished with the Corps of Engineers GTM with an air roller, . The general criteria used for dense-graded mixtures have been: ? 2.5 percent is good, ? 3.5 percent is acceptable, ? 4.0 percent is undesirable [The reference for this criteria was not given]. The critical value of gyratory shear strength at 200 gyrations has been established at 372 kPa (54 psi), which was correlated with a Hveem Stability value of 37.2 and has been subsequently shown to distinguish poor rutting performance from good performance for a variety of dense-graded mixtures. Three of the eleven mixtures showed declining gyratory shear strength and were below the critical value of 372 kPa at 200 gyrations. At 200 gyrations, most mixtures had densified to between 2 and 3 percent air voids. Four of the mixtures had air void cont more than 4.0 percent, however, the shear strength of these mixtures remained very and did not indicate potential for a loss of shear resistance. It was noted as indicated tha the SMA mixtures are less sensitive to low air void contents. It was concluded that compaction of SMA for vo 56 and the GTM is sensitive to the shear strength characteristics of SMA mixtures. The authors expected that if the shear strength proves to be a good indicator of rutting resistance, it may b e used in the future to optimize mix design for SMA mixtures. It is Brown, E.R, and J.E. Haddock, T.A. Lynn and R.B. Mallick. ?Designing SMA Mixtures Volume II- Research Results?, NCHRP 9-8/2 Draft Final Report, September 1996. Brown et al (36) conducted the NCHRP 9-8 project on mix design of SMA mixtures, which was completed in two distinct parts. Part 1 began in April 1994 and had its main goal as the development of a tentative SMA mixture design method. Part 2 began in early oals were to evaluate and finalize the proposed design procedure, Compaction Effort also recommended that an N-design experiment specifically for SMA mixtures is necessary to determine appropriate compactive effort. 1995, and its main g analyze the SMA mixtures produced using the method, and produce a final report detailing the research project. The final report was published in 1999 as a NCHRP report (42). The Volume II presents the detailed laboratory research results of Part 2. Some interesting topics are included and summarized below. One objective of this study was to adapt the Superpave volumetric mixture design procedure for use with SMA. A total of 8 aggregates were used in this study to provide a wide variety of particle shapes, surface textures, absorption and L.A abrasion values. There were two limestone aggregates, two granite aggregates, one traprock, one dolomite, one blast furnace slag, and one gravel aggregate. 57 Two sets of SMA mixtures were designed for each of the 8 aggregate sources. One set was designed using 50 blows Marshall compaction, and the other set was designed using 100 gyrations of SGC. The Marshall hammer in this study was a flat faced, static base, mechanical Marshall ham - mer. The SGC used in this study was Troxler SGC, model 4140. A comparison of the optimum asphalt contents for SMA mixtures compacted with 50-blow Marshall and 100 gyrations of the SGC is shown in the Figure 2.2. [It should be noted that there is a slight difference in aggregate gradation for the SMA mixtures designed with the two compaction efforts.] FIGURE 2.2 Optimum asphalt content from two compaction efforts (36). high but lower optimum asphalt content when the asphalt content was low. Only one The results show that the SGC on the average gave a lower optimum asphalt content than the Marshall hammer. It also appears that the SGC at 100 gyrations gave higher optimum asphalt contents than 50 blows Marshall when the asphalt content was 58 mixture had less than 6 percent optimum asphalt content when compacted with 50 blow of the Marshall hammer while five mixtures had less than 6 percent optimum a s sphalt content f 75 blows with the Marshall hammer is summarized in Table 2.5. [This Table is combined fr .2 .2 .30 e pa n 5, , a 25 at o S is s ed a 6 s m T s 4 a 3 TABLE 2.5 m c Ha ) Ag . Limestone (1) Granite (1) Traprock Granite (2) when compacted with 100 gyrations of the SGC. This indicates that the SGC is forcing the aggregate closer together than the Marshall hammer, or it may be causing more aggregate breakdown. [The comparison of aggregate breakdown results by two compaction efforts didn?t support the higher aggregate breakdown concept.] The effect of compaction level was determined by compacting SMA mixtures for 4 of the aggregates with three levels of blows of the Marshall, and three gyration levels o the SGC. The comparison on SMA mixture design volumetric information of 35, 50 and om Tables 4 8, 4 9, 4 ] Th com riso of 7 100 nd 1 gyr ions f the GC ummariz in T ble 2. . [Thi table is co bined from able 4.33, .34, nd 4. 5]. Volu etri Properties of SMA Mixtures Compacted by the Marshall mmer (36 g Blows 35 50 75 35 50 75 35 50 75 35 50 75 Opt % 7. 8 .6 5.5 4.7 6.5 .2 . AC, 1 6.4 5.8 6. 6.1 5 5.3 6.0 5 VTM, % 3.8 3.8 3.8 3.7 3.8 3.8 3.6 3.6 3.8 3.7 3.6 3.7 VMA, % 19.4 17.9 16.5 18.7 17.3 16.2 17.6 17.3 17.8 16.2 15.2 13.5 VFA, % 80 78.8 77.0 80.1 78.0 76.5 79.5 79.2 78.6 77.5 76.3 72.5 VCA, % 31.9 32.2 31.3 35.2 34.2 33.3 38.6 35.4 37.8 32.0 31.1 30.5 TABLE 2.6 m c x p e Ag . Limestone (1) Granite (1) Traprock Granite (2) Volu etri Properties of SMA Mi tures Com acted by th SGC (36) g Gyrat 125 ions 75 100 125 75 100 125 75 100 125 75 100 Opt. AC 8 , % 6.2 5.8 5.5 6.2 5.5 5.3 5.7 5.6 5.1 5.3 4.9 4. VTM, % 3.7 3.7 3.7 3.6 3.6 3.8 3.8 3.8 3.8 3.7 3.8 3.7 VMA, % 17.6 16.8 16.2 17.3 15.7 15.5 17.8 17.3 16.4 13.3 12.6 12.2 VFA, % 79.0 78.3 76.9 79.5 77.4 75.5 78.2 78.2 76.9 72.0 70.9 69.2 VCA, % 37.0 36.2 35.3 33.3 32.1 32.2 38.2 36.5 34.3 26.9 29.7 26.1 59 The differences in the optimum asphalt contents between 35 and 50 blows averages about 0.5 percent, and between 50 and 75 blows averages about 0.6 percent. Th differences between the various SGC compactive efforts are less than that for the Marshall hammer. The average difference betwe e en 75 and 100 gyrations is 0.4 percent and the that mpactive effort did result in the aggregate being forced closer together and average difference between 100 and 125 gyrations is 0.3 percent. The difference was thought to be due to aggregate breakdown, but the breakdown results indicated the higher co not so much the result of aggregate breakdown. It was concluded that 50 blows Marshall compactive effort appears reasonable. The compactive effort with the SGC should be 100 gyrations [in report page 82]. Breakdown This subtask was developed in an attempt to determine how aggregate breakdown affe SMA characteristics and to correlate the amount of aggregate breakdown experienced during laboratory and field compaction. The amount of aggregate breakdown experi during the mixture design phase was determined for both 50 blows Marshall compaction and 100 gyrations of SGC for each of the 8 aggregates. For 4 of these aggregates, the compaction level was varied to determine how compaction effort affects the aggregate breakdown for both compaction devices. Two additional compaction levels for the Marshall compaction were 35 and 75 blows, the additional cts enced SGC compaction levels were 75 and ts 125 gyrations. The amount of aggregate breakdown produced in dense-graded mixtures using 3 of the aggregates was also determined, and the compaction efforts used were 75 blows Marshall hammer and 128 gyrations of the SGC. A total of 4 field projec were investigated to determine the aggregate breakdown. 60 Breakdown in the laboratory tends to be 5 to 10 percent on the 4.75 mm sieve for harder aggregates and more for softer aggregates. The Marshall hammer (50 blows) ten to breakdown the aggregate more than the gyratory compactor (at 100 gyrations). laboratory compactors provided breakdown approximately equal to in-place compaction Mixture designs com ds Both . pacted with 75, 100, and 125 gyrations of the SGC, and 35, 50, and te or more aggregate properties to the amount d reasonable breakdown. construction projects evaluated in this study, there was no significant ion Permea 75 blows of the Marshall hammer were completed for the limestone, traprock, and two granite aggregates. The results indicate little difference in the amount of breakdown produced by the three compaction levels, but the type of compactor has an effect for some aggregates. Since the aggregate type was found to be significant to the amount of aggrega breakdown, an attempt was made to correlate one of aggregate breakdown. The L.A. abrasion value was shown to have a good correlation with the aggregate breakdown. And it appears that an L.A. abrasion requirement of 30 which is often used is reasonable to separate these aggregate with extreme breakdown an For the difference in aggregate breakdown for static and vibratory rollers. The L.A. Abras appeared to have little effect on the in-place breakdown of aggregate. [Final report Volume III, page 62] bility The objective of this subtask was to determine if SMA mixtures are more or less permeable to water than are conventional dense-graded HMA mixtures. A falling head permeometer was used to determine how permeability varies as a function of air voids. 61 The permeability of the SMA mixture was found to be very sensitive to the air void content and may change by a factor of 10 with very little change in air void conte The results also showed that the percent of absorbed water and the permeability start increase rapidly at approximately nt. to 6 to 6.5 percent air voids. test results, it was concluded that SMA mixes are more permeable Perform Based on the than Superpave mixes with similar void contents and similar nominal maximum size aggregates. Therefore, SMA must be compacted to a lower in-place void content of approximately 6 percent or less. ance Tests The performance tests were used to modify design criteria if necessary, and to further refine the design method. Two tests were conducted in this study, including Wheel Tracking (WT), and Indirect Tensile Creep (ITC). The wheel tracking test was used to estimate the rutting potential of SMA mixtures. The test equipment used in this study was manufactured by Couch Constructio and was very similar to the Hamburg equipment. The loads were applied under water. Four aggregates (limestone, slag, traprock, an n, d granite) and 4 mortars were evaluated in the wh 4 tes. The rutting rates for slag, traprock, and granite were 0.31, 0.22, and 0.21 eel tracking device. Each mixture was tracked at 55?C with up to 20,000 passes unless failure occurred earlier. The ITC test was conducted at -10?C to evaluate the low temperature properties of the mixture. The tensile strength and strain at failure for the aggregates and 4 mortars were determined. The comparison between four aggregate types indicated that limestone aggregate had a significantly higher average rutting rate of 5.02 mm/hr compared to the other three aggrega 62 mm/hr, ll 0 on between wheel tracking results at 55?C and mortar dynamic shear rheome s Based on this study, it was suggested that the wheel tracking test should probably be a part of mix design of SMA mixture to predict rutting performance. The indirect tensile test should also be included in SMA mixture design to evaluate the potential for low temperature cracking if its correlation with performance is further verified. ne lt (SMA) Mixtures in the United States. In Journal of the Association of Asph . summa e ment respectively. The reason for this is not clear. The poor performance of the limestone compared with other three aggregates was also identified by the low Marsha stability (5010 N versus 6100-9700 N) and low tensile strength (526 kPa versus 668-105 kPa). The comparis ter (DSR) results at 58?C indicated a reasonable correlation between DSR and rutting rate based on the limited data. No indication of a good correlation between ITC test results of mixtures and the bending beam rheometer (BBR) test results of mortar wa shown in this study. Brown, E.R., R.B. Mallick, J.E. Haddock, and J. Bukowski. Performance of Sto Matrix Aspha alt Paving Technologists, Volume 66, 1997. Brown et al (37) evaluated a total of 86 SMA projects built from 1991 to 1996 in USA Data on SMA material and mixture properties, and performance were collected and rized. At the time this paper was presented, some SMA mixtures were beginning to b designed using the Superpave gyratory compactor. However, all the SMA pave sections evaluated in this study were designed with 50-blow Marshall compaction. 63 It was summarized that early SMA mixtures (1991-1993) were typically desi to have 3.5 percent air voids or less, and later on from year 1994, the design air voi became close to 4 percent to reduce the possibility of fat spots and permanent deformation. Authors also indicated that the design air voids were generally lowe northern states compared to the southern states. More than 9 gned ds r for the 0 percent of the SMA projects had VMA ranging from 15 to 20 percent indicate m, and about 2 s. have not been a significant problem for SMA. SMA m blem. The possible reasons for this include segregation, draindown, high asphalt content, or improper type or amount of stabilizer. Based on the findings in this study, it was concluded that SMA mixtures should continue to provide good performance in high volume traffic roads. The extra cost for SMA construction should be more than offset by the increased performance. based on the collected information, and 53 percent of the projects met the requirement of minimum 17 percent VMA. In the mean time, about 50 percent of the projects satisfied the 6.0 percent minimum recommended asphalt content. It was d that a minimum asphalt content requirement is not necessary if the minimum VMA is required. Over 90 percent of the projects had rutting measurements of less than 4 m 5 percent of the projects had no measurable rutting. SMA presented excellent rutting resistance considering most of the projects were located in high traffic area Thermal and reflective cracking ixtures appear to be more resistant to cracking than dense mixtures, which is likely due to the relatively high asphalt content resulting in high film thickness. Asphalt flushing (fat spots) appeared to be the biggest performance pro 64 Louw, L., C.J. Semmelink, and B. Verhaeghe. ?Development of a Stone Mastic Asphalt Design Method for South African Conditions?, Eighth International Conference on Asphalt Pavements, Vol. I., Seattle, Washington, August 1997. Louw et al (38) presented a research study performed to develop a volumetric approa for designing SMA mixture. Two compaction methods, the gyrator ch y compaction and the Marsha table olumetric properties (VMA values of seven mixtures were between 17.7 an e uld 0) this report that the gyratory compaction information gives the indicat compactibility index number (C) is determined by carrying out a regression analysis on the gyratory results. A linear regression analysis is performed on the compaction (%Gmm) ll compaction, were compared in this study. The Marshall compaction used 50 blows on each face with standard Marshall hammer and 100 mm molds. The gyratory compactor used was a Troxler Gyratory Compactor (with gyration angle of 1.25?, compaction pressure of 0.6 MPa, and speed of gyration of 30 rpm), with cylindrical molds having diameters of 100 mm, and 200 gyrations were used. Seven mixtures were designed according to the recipe approach with the same asphalt content of 6.5 percent. These mixtures were compacted by the two compaction devices. The really low Marshall stabilites at 60?C indicated these mixtures were uns even though the v d 20.1 percent, air voids were between 2.9 to 5.7 percent) indicated that thes mixtures were suitable. [European experience(1) indicates the Marshall stability sho not be used to evaluate the performance of SMA mixtures, the research in the U.S. (28, 3 also indicates that Marshall stability of SMA mixture is normally lower than dense- graded mixture] It is stated in ion of compactibility, workability and rutting resistance of SMA mixtures. A 65 versus the natural log of the number of gyrations between 20 and 200. An equation is (2.1) K = slope of the regression analysis d cate that the mix is susceptible to deformation, rather y compactor can give an ?Development of a Mixtur (SMA)?, Journal of Ass a Brown the s also been re Some selected conclusions from this study are: given as following: C = C i + k ? Ln (N) Where: C = compactibility index C i = intercept on the y-axis N = number of gyrations The authors believe that a high value of C i indicates a good workability of evaluated mixture, and a high value of k indicates a high tendency of permanent deformation during the design life of the mixture. However, a high C i value combine with a low k-value could also indi especially if the asphalt content is very high and overfilling the air voids. Based on test results, it was recommended that the gyratory compaction than Marshall compaction be used to design SMA. The gyrator indication of the rut resistance of the mixtures and volumetric properties during the design life of the SMA mixtures. Brown, E.R., J.E. Haddock, R.B. Mallick, and T.A. Lynn. e Design Procedure for Stone Matrix Asphalt oci tion of Asphalt Paving Technologists, Vol 66, 1997. et al (39) presented a mixture design procedure for SMA mixtures developed by National Center for Asphalt Technology. Most of the content of this paper ha ported in the Final report of NCHRP project 9-8 (36). 66 1. There is a good correlation between the Los Angeles abrasion loss and aggregate breakdown (Marshall compaction R 2 =0.62, SGC R 2 =0.84), 3. action R 2 =0.89), er paction effort along with the gradation in meeting the required VMA criteria. 5. It was found that 50 blows of Marshall hammer produced about the same density as 100 revolutions of the SGC. The SGC was found to produce less aggregate breakdown than the Marshall hammer. rse aggregate (VCA) for the coarse r A to 2. The 3:1 or 2:1 flat and elongated particles appear to provide much better classification for the various aggregates than a 5:1 ratio, There is an excellent correlation between the flat and elongated particle ratio and aggregate breakdown (Marshall comp 4. It was observed that a VMA significantly lower than specified VMA can be obtained due to aggregate breakdown. Therefore, the mix designer must consid aggregate type, compactor type and com Brown, E.R., and J.E. Haddock. ?Method to Ensure Stone-on-Stone Contact in Stone Matrix Asphalt Paving Mixtures?, In Transportation Research Record 1583, TRB, National Research Council, Washington, D.C., 1997. Brown et al (40) presented a method for determining when stone-on-stone contact exists. The proposed method first determines the voids in the coa aggregate-only fraction of the SMA mixture. Second, the VCA is determined fo the entire SMA mixture. When the two VCA values are compared, the VCA of the SM mixture should be less than or equal to the VCA of the coarse aggregate-only fraction ensure that stone-on-stone contact exists in the mixture. 67 Five different methods for determining the VCA of the coarse aggregate-only fraction were used to see which performed best and was the most practical. The five method the s en compaction method and aggregate gradati as er o t the tests had been conducted]. Test results showed that the breakdown of SMA mixtures compacted by Marshall were similar to the coarse aggregate-only SMA s were Marshall hammer, SGC, dry-rodded test, vibrating table, and vibrating hammer. When using Marshall hammer, SGC, and vibrating hammer, 2 percent AC-20 asphalt by total specimen mass was added to aid in the compaction process. While in dry-rodded method, and the vibrating hammer test, no asphalt cement was added. In general, the Marshall hammer and SGC produced the lowest VCA values, while the vibrating hammer gave the highest. The dry-rodded and vibrating table method produced VCA values that were approximately equal and between the high and low extremes. It was also indicated that for a giv on combination, the coarse aggregate-only fraction produced approximately the same VCA regardless of aggregate type. [Data shown in the paper still have as high 8.7 and 8.9 percent difference between FL limestone and Traprock when compacted by Marshall hammer and SGC, respectively.] It was concluded that the reason for the lower VCA with the Marshall hammer than with the SGC was aggregate breakdown. It was also indicated that the method ultimately selected to measure coarse aggregate stone-on-stone contact should result in breakdown similar to that occurring during construction. Samples of SMA mixtures compacted in the laboratory with the Marshall hamm were extracted to evaluate the amount of breakdown in the total SMA mixtures. [The extraction data from SGC compacted SMA mixtures can not be found in the paper, n indication tha 68 compac e the aggregate breakdown they produce with aggregate breakdown during the SMA mixture design. Also, aggregate breakdown that occurs during construction should be quantified to determine whether it is similar to that produced by laboratory compaction. trix Asphalt designs complement each other. Maryland used the term ?Gap-graded? Superpave to describe SMA mixes designed with Superpave Performance-graded (PG) asphalts, fiber filler, and Superpave testing equipment, including the gyratory compactor. rt. So this report has two parts, the first part contains a summary of research results ted by SGC, lower than the coarse aggregate-only SMA compacted by Marshall, and much higher than the coarse aggregate-only SMA compacted by the other three compactors. It was concluded that the SGC and dry-rodded methods produced the best results, and further study is needed to compar Kuennen, T. ?Gap-Graded Maryland Mixes Meld SMA, Superpave Designs?, HMAT, Vol. 4, No.2, 1999. Kuennen (41) stated that Maryland was defining the area where the Superpave mix design system and high performance Stone Ma Brown, E.R. and L.A. Cooley, Jr. NCHRP Report 425. ?Designing SMA Mixtures for Rut-Resistant Pavements?. Transportation Research Board, Washington, D.C., 1999. Brown et al (42) reported the findings for NCHRP Project 9-8, ?Designing Stone Matrix Asphalt Mixtures?. This report includes Volumes III and IV of the original five-volume final repo in the areas of SMA mixture design and performance evaluation, the second part 69 presents a mix design method and construction guidelines for SMA in AASHTO stand format. Two distinct phases of research were conducted to achieve the overall objec ard tive of develop the s blows with the Marshall hammer. [The results while t paction effort w ith an ing a mixture design procedure for SMA. Phase I included evaluating critical material and mixture properties of SMA and choosing laboratory tests to evaluate selected material and mixture properties. Phase II included the field validation of the proposed mixture design procedure and give the construction guidelines for SMA. In phase I, the volumetric properties for SMA were compared between 50 blow of Marshall hammer and 100 gyrations of the SGC. It was concluded that a good correlation existed between the two compactive efforts, and 100 gyrations of the SGC would provide about the same density as 50 are shown in Figure 2.2 in literature 35] The data indicated that the Marshall hammer tended to give higher optimum asphalt contents at lower asphalt contents, at higher asphalt contents, the SGC gave higher optimum asphalt contents. [The repor didn?t give any explanations for this trend] In phase II, field projects were used to verify that 100 gyrations with SGC was equal to 50 blows of the Marshall hammer. Since the 50 blows Marshall hammer compaction was widely used and proved to have good performance, the com ith SGC was set to provide a density approximately equal to that of 50 blows w the Marshall hammer. The results showed a lot of scatter on the gyrations with SGC to produce the same density as 50-blow Marshall compaction. The range was from less th 60 gyrations to more than 100 gyrations, and averaged about 80 gyrations. 70 The further analysis on the variation of gyrations indicated that aggregate L.A. abrasion loss may be a possible source. The analysis showed that with L.A. abrasion loss 20 to 40 percent, ranges from 68 to 82 gyrations with SGC were needed to produce a ilar density as 50 blows of the Marshall hammer (results are shown in Figure 2.3). sim FIGURE 2.3 Gmb ratio as a function of gyratory level and L.A. abrasion loss (42). Therefore, it was recommended that 70 gyrations should be used with L.A. abrasion loss of 30 or more, while 100 gyrations should be used when harder aggregate with L.A abrasion less than 30 was used. It was also indicated that these two gyration levels will result in a difference in optimum asphalt content of about 0.4 percent. ate- PA publication (43) is usually referred to as the SMA guidelines, which is an update version from 1994 SMA guidelines (2). The updated research results from LA Abrasion, % Loss: 0.985 0.995 1.000 1.005 1.010 1.015 1.020 1.025 Number of Gyrations Gm b R a t i o ( G y r a G m b / M a r s ha l l G m b) 0.990 t or y 0.975 0.980 50 55 60 65 70 75 80 85 90 95 100 20 30 40 Gmb Ratio =0.943+(0.000586*(Number of Gyrations)) + (0.000416* (LA Abrasion, % Loss)) NAPA, ?Designing and Constructing Stone Matrix Asphalt (SMA) Mixtures - St of-the-Practice?, 1999 (Updated edition is published in 2002). This NA 71 NCHR inders ations of SGC) with 150 mm diameter specimens or 50-blows of the f ts that choosing the asp GC. en 30 and 45. A c off P project 9-8 are reflected in these guidelines. Some concerns about the SMA mixture design, especially compaction method and design air void content are listed below. The compaction temperature is determined in accordance with AASHTO T245 (25), section 3.3.2, or that recommended by the producer when polymer-modified b are used. Laboratory samples of SMA should be compacted using either 100 gyr the Superpave Gyratory Compactor ( lat-faced, static base, mechanical Marshall hammer with 100 mm diameter specimens. Higher compactive efforts than these can cause excessive aggregate breakdown and should not be used. The optimum asphalt content is chosen to produce 4 percent air voids in the mixture. The NCAT performance evaluation of SMA pavement sugges halt content to produce an air void content near 4 percent will provide protection against fat spots after laydown and provide better rut resistance, particular in warm climates. Cold climates may use an air void content near 3.5 percent. It is indicated in these guidelines that with the increased emphasis on the use of Superpave mixtures and the SGC, more SMA mixtures are being designed with the S When using the SGC, 100 gyrations is typically used for aggregates with L.A. Abrasion value less than 30, and 75 gyrations may be used for L.A. Abrasion value betwe ompactive effort of 78 gyrations of the SGC was comparable to the 50-blow Marshall compaction based on NCHRP project 9-8, and this number can be rounded to 75 gyrations to correspond to N design gyrations used in AASHTO PP28 (44). 72 [No performance evaluation test is included in this publication. However, it is indicated that triaxial test and wheel tracking test have potential to evaluate the rut Kuennen, T. ?Stone Matrix Asphalt is Catching on in the U.S.?, in Better Roads, Sep. 2003. Kuennen (45) gave a general idea in this newsletter how SMA compared to Superpave designed dense mixtures, and how these two asphalt mixtures integrated together. Maryland is one of the first few states that integrated Superpave design technology into SMA mixes, also called gap-graded Superpave, which is in fact SMA designed using Superpave, commended the following modifications: ? 44). ? ? Design air voids should be 4 percent. There are no recommended N initial or N maximum values. rpave mix design procedures actually resistance of SMA mixtures, and further research is need to develop the necessary test criteria.] including gyratory compaction, PG-binders, etc. In 1998 the Superpave lead states re SMA should use asphalt binders at least one grade higher at high temperature grade than selected in accordance with AASHTO MP-2 ( A fiber modifier should be used to facilitate placement regardless of the binder grade selected. ? Use 150-mm diameter Superpave specimens in SMA mixture design. It was concluded that SMA and the Supe complement each other. In some respects, Superpave mix design methods are making SMA even more durable. 73 Aschenbrener, T. ?Results of Survey on Stone Matrix Asphalt?, Colorado Deportment of Transportation, April 2004. A comp was ral . s. significant experience and answered the questions. paction efforts used for SMA mixture design, 2 states use ny in 1960?s (1) to decrease the excessive wear and damage e more widespread in Europe, various specifications evolved to suit the regional materials. These specifications and empirical methods were not directly rehensive survey on mix design and construction practices of SMA mixtures conducted by the Colorado Department of Transportation (46). The survey mainly included 4 questions, which concerned paving requirements, use of fiber, use of mine filler, and compaction effort for SMA mixture design A total of 43 responses from different states were received and summarized. Sixteen states had no or little experience with SMA and provided no answers to question Eleven states had limited experience and answered the questions. Sixteen states had For the question on com 50 gyrations, 4 states use 75 gyrations , 16 states use 100 gyrations, 2 states use various gyrations based on traffic level, and 4 states continue using 50 blow Marshall compaction. 2.2.2 Summary of Literature SMA was first developed in Germa caused by studded tire use. SMA continues to evolve in Europe, and has been introduced into several European countries as well as other countries outside of Europe. Due to its exceptional rut resistance and good durability, the use of SMA has become widespread throughout the world. The early SMA mixture design in Europe was essentially a recipe-type method, which dictated mixture ingredients and proportions that seemed to work well. As the use of SMA becam 74 applica cted in were built in North America thereaf in rly e from Europe and the United States, and set volumetric requirements for SMA mixture l of 17 lumetric -on-stone contact was validated by using la ble to all situations; therefore, similar satisfactory results could not be expe when different materials were used. Despite these facts, few SMA failures are reported the literature. SMA was first introduced into the United States in 1990 as a result of the European Asphalt Study Tour (1). The first SMA trial in the United States was built in Wisconsin in June 1991, and many subsequent projects ter (26-27, 29). However, there was no standard mix design procedure available the early 1990?s. Consequently, most specifications and design procedures used for ea U.S. trials were a reflection of those of the Europeans. In 1994, guidelines (2) for materials, production and placement of SMA were developed by the SMA Technical Working Group (TWG) and then published by the National Asphalt Pavement Association (NAPA). These guidelines summarized the experienc , including a design air void content at 3 to 4 percent by using 50-blow Marshal compaction, a minimum asphalt content of 6 percent and a minimum VMA content percent. In the middle 1990?s, several studies (31-32, 40) addressed mixture vo properties and establishment of the stone-on-stone contact that provides SMA with excellent rutting resistance. The importance of stone boratory shear strength measurements made with the GTM (35). The measurement of stone-on-stone contact in the SMA mixture design procedure has made a definite improvement in the mix design procedure. 75 The current mix design method for SMA mixtures was developed in the NCHRP project 9-8 (36), and was summarized in the final report (42). One objective of this stu was to adapt the Superpave volumetric mixture design procedure for use with SMA. Th results of this project suggested that the equivalent compaction effort as the 50-blow Marshall hammer depended on the L.A. abrasion value of aggregates used, and a dy e veraged about 8 appropriate for evaluating performance of SMA mixtures, including Marshall stability and flow, resilient modulus, and indirect tensile strength etc. It is generally believed (30, 32-33, 36) that loaded wheel tracking tests and/or the confined repeated load permanent deformation test have the best potential to evaluate rutting resistance of SMA mixtures. oven to be adequate for several projects in Europe and North A 0 gyrations. Therefore, it was recommended that 70 gyrations should be used when the L.A. abrasion loss was 30 or more, while 100 gyrations should be used when the L.A. abrasion was less than 30. The establishment of stone-on-stone contact is required in this mix design method by comparing VCA mix and VCA drc (40, 42). Although SMA mixtures have excellent performance in the field, there is no generally accepted performance test to evaluate it in the laboratory. Several studies (28, 29, 31-33, 35) evaluated various performance tests that could best evaluate SMA mixtures. Several evaluation tests for dense-graded mixture were proven not 2.3 FINDINGS ON COMPACTION EFFORT FOR DESIGNING SMA MIXTURES SMA mixtures have been commonly designed with the 50-blow Marshall compaction, and this compaction effort has pr merica (1, 27, 29, 36). Several studies (30, 32, 35) have also indicated that the compaction effort for SMA mixture was less than typically used for dense-graded 76 mixture, and this finding reflects the practice of using 50-blow Marshall rather than 75 blows to design SMA mixtures. However, with increased emphasis on the use of Superpave mixtures and the SGC, SMA has been incorporated into the Superpave mix design system, and more SMA mixtures are being designed with the SGC. Studies (38, 45) have shown more interest in using SGC instead of the Marshall compactio n procedure. At the time this report was prepare edure to ad of design guidelines (43) list two compac e . Experience from d, only four States in the United States continued to use the Marshall proc do SMA mix design (46). One of the most important reasons for using the SGC inste the Marshall hammer is because the SGC better simulates the degradation and aggregate orientation found in field compaction (5-7). A comparative study on SGC compaction and Marshall compaction was performed during NCHRP 9-8 (36) and indicated that about 78 gyrations with SGC produced a similar density as 50 blows of Marshall compaction. However, 100 gyrations was adopted in the final report (42). Current SMA tion options to design SMA: 50 blow Marshall or 100 Gyrations by SGC. Th design guidelines also note that 75 gyrations should be used if the aggregate has a L.A abrasion value higher than 30 percent loss. Therefore, one of three compaction efforts is typically used in the SMA mix design procedure. Some states such as Georgia have found that 100 gyrations with the SGC is excessive for their materials and results in mixtures with lower than desired optimum asphalt contents. Georgia has many aggregate sources with L.A abrasion value above 30 The high level of density obtained in the laboratory is also difficult to obtain in the field in some cases without excessive fracturing of aggregate particles. 77 Georgi le, g ixture and having rutting problems if the compaction level is too l nge ent NMAS. In this study, several compaction levels will be evaluated by volumetric properties, aggregate degradation, and rutting resistance. Recommendations will be made about adopting a lower compaction level based on the overall performance of the mixtures designed in this study. he etical a indicates that for their materials the optimum SGC compactive effort should be between 50 and 75 gyrations. The recent survey on SMA (46) showed that some states had already used either 50 or 75 gyrations to design SMA mixtures. Georgia, for examp requires 50 gyrations in their SMA mixture design specification. Based on the literature review, the preference of using a compaction level lower than the standard 100 gyrations has little research support. The major reason for choosin a lower compaction level is to produce a more durable mixture by designing a m with higher optimum asphalt content. Using a decreased compaction level may result in the SMA mixture becoming unstable ow. No study is available that evaluates performance of SMA mixture designed with lower compaction levels. Therefore, a study is needed to characterize the performance of SMA mixtures designed with lower compaction levels using a wide ra of aggregates and differ 2.4 OTHER RELEVANT TESTS RELATED TO TOPICS This section includes several other relevant tests that will be performed in this study. T background information of these individual tests is provided for each test. 2.4.1 Vacuum Seal Method (CoreLok) The air voids is determined by comparing the bulk specific gravity and the theor maximum specific gravity of hot mix asphalt (HMA). The standard method currently 78 used to measure the SMA bulk specific gravity is AASHTO T166 (25), commonly known ly ant differen nine l ures 49) xes having coarse gradations. The study also suggested that the CoreLo ill determine if the CoreLok method should be used for SMA e is will affect the volumetric properties and whether this difference depends on NMAS. as the saturated surface dry (SSD) method. Another alternative method recent used is the CoreLok method following ASTM D6752 (47). Both methods have advantages to some other alternative methods such as the parafilm method, and cut and measure method (48). The study conducted by Cooley et al (49) showed there was not a signific ce between the variability of the CoreLok method and SSD method in six of cases, while in another three cases the SSD method was less variable. However, Hall et a (50) indicated that the CoreLok method had smaller multi-operator variability compared to the SSD method in 82 percent of 144 samples with a wide range of air voids. Research by Buchanan (51) has indicated that the CoreLok vacuum-sealing device provides a better measure of internal air void contents of coarse graded mixt than other conventional methods because of the increased repeatability. Cooley et al ( also reported a significant difference in air voids measured by the CoreLok and SSD methods for the mi k procedure was a better measure of density for samples with high air void contents. The primary reason for this is the high potential for overestimating sample density due to the water draining out of the mixture when determining the SSD weight in the SSD method. This study w mixtures. As part of this evaluation the study will look at how much difference ther between the CoreLok and SSD methods for different air void levels, how this difference 79 2.4.2 Permeability It is generally believed (52-59) that permeability is a function of both the air voids conten . f n in Virginia (56) has indicated all NMAS . t and the degree of interconnectivity of those voids. Therefore, besides the total air void contents, permeability also depends on gradation, NMAS, and other factors that may affect the air voids shape and air voids distribution. Many researchers (52-58) have suggested that permeability depends on gradation It is generally believed that mixtures with coarse gradations have larger individual voids which increase the potential for interconnected air voids. Fine-graded mixtures tend to have smaller individual air voids and have less potential for interconnectivity. An early study (52) indicated that permeability of aggregate materials depends more on the size o the voids than the volume of voids. They found that fine aggregates with VMA of 30 to 35 percent were less permeable than well-graded coarse aggregates with VMA of 12 to 15 percent. SMA has been found (53) to be more permeable compared to dense-graded mixtures at a given air voids content. It was especially significant that the 9.5 mm NMAS SMA mixtures were found to be permeable at a 5 to 7 percent void range, while a 9.5 mm Superpave designed mixtures above the maximum density line was not permeable at a air voids content of 9 percent (59). However, experience (25 mm, 19 mm, and 9.5 mm) SMA mixtures were impermeable with in-place air void contents below 6 percent. In current SMA design guides (36), the suggested air voids level after field compaction is 6.0 percent or less. Several studies (53-58) also suggested that NMAS had an effect on permeability As the NMAS increases, the size of the individual air voids also tend to increase which could result in higher potential for interconnected air voids. Therefore, the larger NMAS 80 mixtures would be expected to be more permeable than the smaller ones at a given air voids level. A thin overlay study (54) indicated that at 7 percent air voids, only two 4.75 mm NMAS SMA mixtures complied with permeability requirement, while 19 mm and 12.5 mm NMAS mixtures had permeability values as much as 50 times of that generally specifie less and 25 A mixtures. This study will determine at what air voids the SMA mixtures will become permeable for different NMAS mixtures, and how compaction level affects the permeability. Several decreases. The LA Abrasion value also became more c degrade more. It was suggested that when high abrasion loss aggregate was used, the d. The Virginia experience (56) also showed that 9.5 mm NMAS SMA was permeable at the same air voids level compared to the more commonly used 19 mm NMAS SM 2.4.3 Aggregate Breakdown studies (7, 60-65) have focused on aggregate degradation. Generally speaking, the most influential factors affecting the degradation of aggregates in HMA include aggregate gradation, aggregate toughness, particle shape, and compactive effort. Among all the factors, aggregate gradation was found to be the most important factor controlling degradation (60). As the gradation becomes denser (closer to the maximum density line), degradation ritical when coarse graded mixture were used. For a given aggregate, degradation in SMA mixtures would be expected to be more severe than in dense graded HMA because of the stone-on-stone contact. Aggregate toughness, mostly indicated by LA abrasion, was another important factor (60, 62-64). Collins et al (62) found that aggregates with high abrasion loss 81 gradation should be designed to compensate for the effect of degradation during compaction. Brown et al (64) also found LA abrasion loss at 30 percent resulted in about 10 perc tes. The researchers (63-65) showed that for a given gradation and aggrega low initial pressur egradation in HM e to the ent degradation on the 4.75 mm sieve size, and it should be set as the limit for aggregate used in SMA mixtures. Particle shape, usually indicated by F&E content, was found to affect the degradation of aggrega te type, increasing the percentage of F&E particles would increase the amount of aggregate breakdown. The compaction method was believed to be another influential factor. Button et al. (7) compared field cores with specimens compacted by an Exxon rolling wheel compactor, a gyratory compactor, a linear kneading compactor, and a rotating base Marshall compactor. The study showed that the Marshall compactor fractured more aggregate than the other three compactors. An early study (5) on gyratory compactor concluded that gyratory compaction can closely simulated the degradation found in the field compaction because it allowed proper orientation of aggregate with e. A recent study (22) using x-ray tomography indicated that the gyratory compaction could provide similar aggregate orientation as in the field cores. Obviously, these factors would also interactively affect the aggregate d A mixtures. Aggregates with high L.A abrasion values were more sensitiv effect of F&E particles than aggregates with lower L.A abrasion values (63). One of the purposes of laboratory compaction is to simulate the aggregate breakdown in the field. The aggregate breakdown due to the field construction was reported by Brown et al (64) and is shown in Figure 2.4. From Figure 2.4, one can 82 observe that the aggregate breakdown is not dependent on the L.A abrasion value of aggregate. The aggregate breakdown values in the field were between 3.7 and 6.3 per on the 4.75 mm sieve for all the projects investigated, except for one project that had unexpect cent ed high aggregate breakdown (11.1 percent). If this project is deemed as an outlier, the average aggregate breakdown in th all projects is 4.9 pe e field construction for rcent. y = -0.004x + 5.8007 R 2 = 0.0002 6.0 8.0 10.0 o n 4. 75 m m eve, 0.0 2.0 4.0 12.0 L.A Abrasion of Coarse Aggregate, % loss P t o w n S i % B r eakd 10 15 20 25 30 35 40 45 e r cen FIGURE 2.4 Correlation of L.A Abrasion values and aggregate breakdown for field compacted samples (64). This study will determine the aggregate breakdown values in the laboratory for different compaction efforts. The influence of these factors that affect the aggregate breakd e in resistance to further densification, was first developed by Illinois (4, own will be determined and discussed. 2.4.4 Locking Point The ?locking point?, or the point during compaction at which the mixture exhibits a marked increas 83 66) to p ixture questio two ts are defined and used as maximum allowed gyration levels for two traffic conditions. The ?first? and ?second? locking points are defined as the number of e height has been recorded for the third e for two consecutive gyrations. revent the over-compaction of their designed mixtures and better back-calculate the specimen densities at prior levels of compaction. The locking point is defined by the authors as the first of three consecutive gyrations producing the same sample height [the difference in height is less than 0.1 mm]. Generally, the densification rate of the m is nonlinear at any further gyration levels. The ?Locking Point? has also been found to be related to compaction tendencies in the field (67). Compaction of a mix past the ?locking point? generally results in aggregate degradation that is not representative of field compaction, and thus the benefit of compacting the mix to very high gyration levels, such as 125 or even 100 gyrations is nable. As a result, several states have already included the locking point concept into their mix design specifications. In the Georgia DOT provisional specification (68), locking poin gyrations at which, in the first occurrence, the sam time and the fourth time, respectively. It was also reported (68) that for Georgia materials, typical first and second locking points are around low 60?s and high 80?s, respectively. In the Alabama DOT provisional specification (69), the locking point is defined as the first occurrence at which the sample height remains the sam This study will use 100 gyrations, a compaction level that is above the locking point for most mixtures as the second gyration level, and a third level that is lower than the locking point. The influencing factors that affect the locking point will also be discussed. 84 2.4.5 Asphalt Pavement Analyzer Rutting Test The loaded wheel rut tests are simulative test methods used for evaluating rut resistance of HMA is urrently alt ), tracker), Purdue Univer r) (70). . mixtures. In simulative tests, the load and environmental conditions in the field are simulated in the laboratory, and the response of the mixture in the laboratory recorded and used for field performance prediction. Several loaded wheel testers (LWT) have been used in the past and are c being used to evaluate rutting performance. Some of these methods include the Asph Pavement Analyzer (APA, second generation of Georgia Loaded Wheel Tester, GLWT Hamburg Wheel-Tracking Device, French Rutting Tester (LCPC wheel sity Laboratory Wheel Tracking Device, Model Mobile Load Simulator, Dry Wheel Tracker (Wessex Engineering), and Rotary Loaded Wheel Tester (Rutmete Out of all LWTs, the APA is one of the most commonly used and most readily available Brown et al (71) recommended using the APA as a temporary simple performance tests for evaluating rutting. Their recommendation was based on a comparative assessment for almost all available test methods. The ability to predict permanent deformation was one of the most important characteristics for the comparison. The GLWT, then the APA has been used by numerous researchers in an attempt to evaluate rutting resistance of HMA mixtures. The capability of this device to estimate field rutting performance has been validated by many studies (72-77). Lai (72) evaluated rut-prediction capability of the GLWT using four mixes from Georgia with known field rut performance. Three of the four mixes had shown a tendency to rut in the field. Results of this work showed that the GLWT was capable of ranking mixtures similar to actual field performance. West et al (73) conducted a study to 85 verify that the GLWT can be used to predict actual field performance. Three field projects with known rutting performance were included in the study. They found that there was a good correlation between the GLWT results and the actual rutting performance of the mixtures. Miller et al (74) evaluated the feasibility of the GLWT to predict rutting at the University of Wyoming. Core specimens from 13 pavement sections with known field performance were selected for evaluation. The sections had a wide range of rutting performance. The APA test results had a good correlation with actual field measurement when project elevation and pavement surface type were considered. Choub s ts on the 7 that have the greatest influence on the outcom results. Factors included in the study were air void content of test specime % vs 8%) men preheating time (6-hour vs 24- hou est temp ure (55?C ), wheel 95 lb vs 105 lb), hose pressure (95 psi vs 105 psi), and specimen type nder vs be he speci type factor was actually ane et al (75) from Florida DOT used three mixes of known field performance to evaluate the suitability of APA for predicting pavement rutting. The rutting performance of these three mix types were good, very poor, and moderate. The results indicated that APA ranked the mixes according to their field performance. This ranking is the same using either beam or gyratory specimens. Mohammad et al (76) evaluated three Superpave implementation projects using the APA test. The APA test results were found to correlate well with the field rutting data. The WesTrack Forensic Team (77) conducted a study on the performance of coarse-graded mixes at WesTrack sections. Their resul actual field performance and the predicted performance using the APA showed a good correlation with an R 2 value of 0.797. West ( 8) conducted a ruggedness study to evaluate APA testing factors e of test ns (6 , speci r), t erat vs 60?C load ( (cyli am). T men 86 a confounded factor in that the cylinders were compacted with an SGC and beams were com ted with Asphalt Vi y Comp C). The und that of the six ma ctors we ignificant: id content, test temperatu nd specim pe. Many te nditions ha een used rutting tests. A New Jersey study (79) summarized a range of APA test specifica used by v us state agencies. TABLE 2.7 APA Testing Specifications U y Vario e Agen State Test Temp, Air V , % Co or Seating C C pac an brator actor (AV y fo three in fa re s air vo re a en ty st co ve b for APA tions ario sed b us Stat cies ?C oids mpact Type ycles ycles AL 67 4?1 SGC 25 8000 AR 64 4?1 SGC 25 8000 CN PG 7?1 SGC/AVC 25 8000 DE 67 7?0.5 AVC 25 8000 FL 64 7?0.5 AVC 25 8000 GA 49 6?1 SGC 50 8000 IL 64 7?1 SGC 25 8000 KS (3M ESALs) New Jersey 5 high traffic 3 very high traffic 8 medium traffic South Carolina 5 Utah 5 Virginia 3.5 high traffic(PG-76) 5.5 medium traffic (PG-70) 7 low traffic (PG-64) West Virginia 6 Zhang et al (70) 8.2 Brown et al (71) 6.0 In summary, the APA rutting test has been successfully used to differentiate the good or bad rutting resistance, and ranked the mixtures according to their field perform o ance. Good relationships between APA rut depth and field rut depth have als been developed for several individual studies. In this study, the APA rutting test will be employed to evaluate the rutting resistance for different compaction levels. The test conditions will follow the most commonly used ones, i.e. 6% air voids samples 89 compacted by SGC, at PG temperature required for location, and 100 lb load and hose pressure. 2.4.6 Triaxial Tests In contrast to simulative test, the fundamental tests examine 100 psi the relationship between stress a tial include ing ape of the specimen, also the only relatively uniform state of stress is tension nd strain in the laboratory, and develop prediction models for predicting field performance. The fundamental performance tests for evaluating rutting poten triaxial (or uniaxial) tests, shear loading tests, and diametrical tests. The triaxial (or uniaxial) tests have been widely used to estimate the rutting potential and to provide necessary input for structural analysis (71). The shear load tests are usually conducted by Superpave Shear Tester (SST), which is expensive and complex to run, and has very limited availability (71). The diametrical test has been deemed inappropriate (83) because the state of stress is nonuniform and strongly dependent on the sh . The use of mechanical properties determined by diametrical testing almost always resulted in overestimates of pavement rutting (83). Multiple members of the NCHRP 9-19 project research team (84) ranked the tests and parameters that are used for predicting permanent deformation based on comprehensive evaluation including test feasibility, relationship between field performance, repeatability of the test and the sensitivity of the test parameter to different mixture variables, and the application of test results. The top three parameters for permanent deformation were 1) the dynamic modulus term (E*/sin?), determined from the triaxial dynamic modulus; 2) the flow time (F t ) from the triaxial static creep; and 3) 90 the flow number (F n ) from the triaxial repeated load test. All these three parameters are from triaxial tests. Uniaxial test without a confining pressure is relatively inexpensive and easy to conduct, however, witho ut confining pressure the creep test or repeated load test usually has to b d P dense-graded mixtures, although unconfined testing gave good results (88) C). le failure. However, the confined test cou on e conducted at a relatively low stress (30 psi, or 207 kPa) and low temperature (can?t usually exceed 104?F, or 40?C), otherwise the sample fails prematurely (71). For predicting rutting, laboratory testing was suggested (85) to be conducted at a high temperature as expected in the pavement in service, because the rate at which permanent deformation accumulates increases rapidly with higher temperatures. For the NCHRP 9- 19 project (84), the triaxial tests for evaluating rutting were conducted at 100 to 130?F (37.8 to 54.4?C). The NCHRP 9-29 project (86) conducted both unconfined and confine repeated load test at 45?C. Additional analysis (87) on mixture verification in the NCHR 9-19 project indicated that confinement is needed to capture the performance of SMA mixtures compared with for many tests. Therefore, confinement is recommended for testing of SMA and open-graded mixtures. To better simulate actual traffic and environmental conditions, Brown et al conducted unconfined creep tests at an axial stress of 120 psi, and for confined tests, a confining pressure of 20 psi was used with an axial stress of 120 psi, both at 140?F (60? Brown found that for the ?ideal test? conditions presented above, the unconfined test, in most cases, could not be performed due to rapid samp ld be conducted and, therefore, was recommended for future testing. Gabriels (89) also used 140?F (60?C) in his rutting study to best simulate the average maximum 91 pavement temperature throughout the country, and maintained 120 psi axial stress (100 psi deviator stress with 20 psi confining stress) to best represent in situ HMA pavemen and traffic interaction. A certain minimum height to diameter ratio is necessary for the accuracy of the tests due to the end effects concern. Foo (90) found that there was no significant end effect when using samples with 2.5 inches in height and 4 inches in diameter, if a confining pressure was applied. However, during NCHRP 9-19 project (84), it was found that a minimum height-to-diameter ratio of 1.5 was required in order to ensure that the response of a sample evaluated in either the dynamic modulus or permanent deformation tests represents a fundamental engineering property. There were some debates on the use of dynamic modulus test for predicting rutting resistance recently. The dynamic modulus term (E*/sin?) was selected as th parameter for predicting rut resistance by NCHRP 9-19 project (84, 91). However, the dynamic modulus is usually conducted at r t e top elatively low stress and/or strain level, and reflects ularly n A ot using the visco-elastic behavior of the material. Neither plastic nor visco-plastic behavior exhibited in rutting is measured by the dynamic modulus test. A recent study (92) indicated that this term may not always relate to HMA rutting resistance, partic when polymer-modified asphalts are used. The study also suggested that tests in additio to dynamic modulus should be considered to accurately access rutting resistance of HM mixtures. A research project in Florida (93) also indicated that there was no discernable relationship between complex modulus and rutting for mixtures of varying gradations and aggregate structure. The study concluded that the complex modulus should generally n be used to determine rutting or fracture resistance of mixtures. While the ability of 92 dynamic modulus results to predict rutting resistance is questionable, the dynamic modulus test will still be conducted in this study as a part of the work plan, and prov the information for stiffness of mixtures. The static creep test has been used for evaluating HMA rutting ide potential for many tions. Typically, the total o (the ap tress d by the pe t strain r 1 hour l ing and 1 r recovery) and the slope of strain with tim used as teria for a pting or rejecting mixtures. Sousa et al (94) reported that, under the unconfined test conditions, reasonable consistency was found in the strain g. Failure strain levels of 0.8 a comprehensive literature review, a study in Texas (95) concluded that under typical test conditions, a creep modulu h se in ion of compressive creep test data, as shown in Table 2.9. TABLE years and it was often conducted under unconfined test condi strain after 1 hour of loading, the creep m dulus plied s ivided rmanen afte oad hou e were cri cce level at which a variety of mixtures failed under creep loadin percent was reported for the compressive creep. Based on s greater than 69 MPa (10ksi) indicates a mix has low sensitivity to rutting, a creep modulus between 41.4 to 69 MPa (6 to 10 ksi) represents a moderate to hig sensitivity to rutting. This study (95) also summarized a criteria table suggested for u the evaluat 2.9 Criteria for Static Creep Test Results (95) Slope of Steady State Creep Curve Total strain < 0.40 at 1 hour of loading, % < 0.17 < 0.20 < 0.25 < 0.30 < 0.35 < 0.25 IV IV IV IV IV III < 0.40 IV IV IV III III III < 0.50 IV IV III III III II < 0.80 III III II II II II < 1.0 I I I I I < 1.2 I I I Notes: I ? Low traffic intensity, < 10 ESALs 5 6 5 II ? Moderate traffic intensity, between 10 5 and 5?10 5 ESALs III ? Heavy traffic intensity, between 5?10 and 10 ESALs IV ? Very heavy traffic intensity, > 10 6 ESALs 93 The unconfined condition generally limited the test temperature to under 40?C (104?F) and axial pressure to under 30 psi (207 kPa). As discussed before, these conditions do not represent the field conditions under which the majority of rutting happens. A rutting study (90) used the confined static creep test to evaluate the cores from 42 pavement sites. The confining and deviator stresses were 20 psi and 120 psi, respectively. The test temperature was 60?C. The field rut depth and rut rate were used t validate th o e confined creep test. The permanent strain from the creep test correlated the best wi ches. s); a effects l eep (6 th the field rut depth and rut rate. And it was concluded that a laboratory permanent strain of 1.2% would be expected to result for a field rut depth of 0.5 in However, the low correlation as indicated by the low coefficient of determination, R 2 (0.35 and 0.21 for creep test vs rut depth and rut rate, respectively) may limit the use of this conclusion. This criterion was developed by using shorter samples (2.5 inche taller sample (6 inches) will likely have a higher strain criterion because of the end of the short samples (90). The NCHRP 9-19 report (84) suggested the constant load should be held in confined static creep test until the tertiary flow occurs or the tota axial strain reaches 4 to 5 percent. Although the flow time and strain slope in static cr test showed a fair to good correlation with field rutting of three field sites (84), additional work is needed to establish the criteria for these test parameters using taller samples inches). Several studies (71, 89, 96-99) reported that the confined repeated load test was found to give a better correlation with field rut depths and more responsive to the 94 presence of modified binders in HMA mixtures than the static constant load (creep The greater suitability of the dynamic test for rating the effect of the binder modification is due to the recovery effects of the tests (98). Gabrielson (89) also repor ) tests. ted that variabi ests. e re ercent for unconfined tests, and 4 to 5 percent for confined tests. Hofstra et al (102) reported in-situ strains up to 15 percent. Based on a national study on rutting, Gabrielson (89), Brown and Cross (96) provided information to show that 13% strain was a good pass/fail criteria for triaxial repeated load tests. Pavement cores were tested to validate the confined repeated load test. The cores were from pavements identified as ?good? pavements or ?rutted? pavements based on the rate of rutting with respect to traffic (89). The original test results are plotted in Figure 2.5. Based on the same set of data, a relationship between the field rut depth and permanent strain by repeated load was developed. As shown in Figure 2.6, a laboratory strain less than 10% would help ensure that the rut depth does not exceed 0.5 inches. Achieving high strain levels in the laboratory more clearly shows the difference between rut susceptib strain levels (96). lity of the static creep tests is substantially higher than that for repeated load t Brown et al (100) discussed the failure conditions in the repeated load. The failur point was defined as the point at which the deformation rate starts to increase rapidly. Strains at failure measured in the study were about 2 percent for unconfined tests. In another study, Pell et al (101) reported the failure strains for the repeated load tests we also about 1 to 2 p le mixes and rut-resistant mixes. These differences may be subtle at low 95 FIGURE 2.5 Permanent strains of core samples by triaxial repeated load test (89). FIGURE 2.6 Field rut depth versus the lab strain from repeated load test (96). In summary, triaxial tests are the most commonly used fundamental tests for predicting rutting performance, and have been recently recommended by the NCHRP 9- 96 19 project as the simple performance tests to complement the Superpave Level 1 volumetric mixture design procedure. In this study, all three triaxial tests, including the dynamic modulus test, static creep test, and repeated load test will be used to compare the effects of differen l test will be rature and with confinement to better simulate the field t compaction levels on rutting performance. The triaxia performed at a high tempe conditions. 97 CHAPTER 3 EXPERIMENTAL PROCEDURES 3.1 RESEARCH PLAN A three-phase test plan was developed to accomplish the project objectives. The first phase consisted of selecting aggregates, asphalt and fiber, and determining their properties. This phase was used to choose a wide range of aggregates and determine proper filler content and gradation. The second phase involved the SMA mixture designs, and evaluation of volumetric properties, permeability and aggregate degradation of several compaction efforts. The third phase involved the performance evaluation of designed SMA mixtures, including wheel track APA testing, repeated load confined creep, dynamic modulus and static creep testing. The general description of each work plan phase is discussed below. Detailed descriptions of tests conducted in each phase are given in later sections. All the test results are shown in chapters 4 and 5. 3.1.1 Materials Selection his phase included the selection of materials and determination of their properties. The test abrasion loss values were selected for the study. L.A abrasion value, flat and elongated (F&E) content, fine aggregate angularity (FAA), and specific gravity of all five aggregates were determined and are shown in chapter 4. T plan for this phase is shown in Figure 3.1. Five aggregates with a range of L.A. 98 y test. Material Selection and Preliminary Test FIGURE 3.1 Work plan for phase I: material selection and preliminar Determine aggregate properties (5 Aggregates) 1. L.A Abrasion (ASTM C 131-96) longated (AST Gravity (A 2. Flat-E M D 4791-95) 3. Specific ASH T8 8 dry (AASHTO T19M-00) TO 4-00, T 5-91) 4. VCA Asphalt Cement (PG76-22) Trial Marshall Mix D oids, es 50 1. Check: Air V VMA and mix ign ( blows) VCA . 1 Agg ? 3 NMAS ? 3 Gradations (upper limit, lower limit, and middle of gradation band) = 9 trial mix designs. 2. Determine design gradations 5 Aggregates: Gravel (LA=31), Limestone (LA=26) 2 Granites (LA=36, 21), Traprock (LA=17), 3 NMAS: 19mm, 12.5mm, 9.5mm Step 1: Obtain Materials and Determine Properties Marble Dust Mortar Test?(determin 3 AC ?3 dust ?3 test ?2 e percent #200 sieve) rep.= 54 samples a FO R? a V Un-aged DSR ? 5 Kp RT DS 11 Kp PA BBR? 1500 Mpa Cellulose Fiber Step 2: Prelim T g ermine Pe t n evinary estin (Det rcen Passi g #4 Si e) 99 The asphalt grade used in this study was a PG 76-22. This PG grade is typically used for SMA mixtures in the southeastern U.S., as well as other areas in the U.S. Also 0.3 percent stabilizing fiber by weight of total mixture as a common practice was used to help eliminate any potential for draindown. Superpave binder tes ts were employed to ortar properties. Trial mix designs were conducted to determine the proper gradation based on volumetric properties resulting from 50 blow Marshall compaction. A total of 9 trial mix des ere con ted for t lowing ination ggrega NMAS, and 3 trial gradations (upper limit er limit and middle value of specified SMA mix gradation band). The trial gradations are shown in Table 3.1. ent Passing, % help determine acceptable filler contents based on fine m igns w duc he fol comb s: 1 a te (crushed gravel), 3 , low TABLE 3.1 Trial Gradations Used in Preliminary Mix Designs Perc NMAS Gradation 3/4" 1/2" 3/8" #4 Codes 1? #8 #16 #30 #50 #100 #200 U 100 88 60 28 24 19.8 16.2 14 12.5 11 19 M 100 95 69 42.5 24 20 16.7 13.9 12 11 10 L 100 90 50 25 20 16 13.6 11.6 10 9 8 U 100 99 85 40 28 22.2 18 15 13 11 12.5 M 100 94.5 67.5 30 22 17.9 14.7 12.5 11 10 L 100 90 50 20 16 13.6 11.4 10 9 8 U 12 100 95 50 30 21 18 15 13 9.5 M 100 82.5 40 25 18.5 15.5 13 11 10 L 100 70 30 20 16 13 11 10 8 The preliminary mix design results are shown in Table 3.2. As shown in Table 3.2, the VCA ratio (VCA / VCA ) for U gradation for all NMAS are either higher than 1 or close to 1. The M gradation for 12.5 mm NMAS is close to 1. With a lower compaction level, the VCA ratio will be expected to be higher indicating that stone on stone contact does not exist. Therefore the gradation should be coarser than the middle of the gradation band as shown in Table 3.1 to ensure stone on stone contact. A set of normal (named N) gradations that are located between the middle of the gradation band mix drc 100 and lower limit were suggested to be used for designing the SMA mixtures for this study (N gradations are shown in Table 3.3. Later results indicated that finer gradations, named F gradations, were necessary for ruby granite and traprock to keep reasonable optimum asphalt content. The F gradations are also shown in Table 3.3). TABLE 3.2 Preliminary Mix Design Results NMAS Gradation VCA drc , % Opt. AC, % VMA, % VCA mix , % VCA ratio U 42.2 6.3 17.0 40.2 0.953 M 40.3 6.4 17.1 36.8 0.913 19 L 40.7 6.7 17.5 33.2 0.816 U 42.4 6.7 17.5 48.0 1.132 M 42.6 6.5 17.1 41.9 0.984 12.5 L 42.4 6.7 18.0 34.2 0.807 U 43.0 6.2 17.0 41.8 0.972 M 41.4 6.4 17.1 37.7 0.911 9.5 L 42.0 7.0 17.7 34.1 0.812 VCA drc - Void in coarse aggregates by dry rodded method VCA mix ? Void in coarse aggregates in compacted mixture Opt. AC ? Optimum asphalt content, at 4 percent air voids VMA ? Void in mineral aggregates 3.1.2 Mixture Design and Volumetric Properties The test plan for the second phase is shown in Figure 3.2. This phase included conducting mix designs and tests to evaluate mix design samples. Three NMAS mixtures were selected: 19 mm, 12.5 mm and 9.5 mm. SMA mixtures were designed with three different compaction efforts, including 50 blows with Marshall hammer, 100 gyrations with the SGC and a lower gyration level near the locking point with SGC for comparison. A third yration level which is below the locking point was also used to design 12.5 mm NMAS MA mixtures for two aggregates. This lower gyration level is lower than that presently eing used but was included to provide some idea of the performance at this lower level. Thus, a total of 47 mix designs were conducted. g S b 101 FIGURE 3.2 Work plan for phase II: SMA mix designs. Step 3: Mix Design Sample Compaction Marshall mix design = 15 mix designs Stone-on-stone contact 5 Agg. ? 3 NMAS 100 Gyrations by SGC 5 Agg. ? 3 NMAS = 15 mix designs Gyration Level near locking point by SGC, = 15 mix designs 5 Agg. ? 3 NMAS Step 4: Mix Design Properties Evaluation 47 mix designs SMA Mix Designs Gyration level below locking point by SGC, 2 = 2 mix designs Air void ontents c D 66 re 67 VMA 1. SS 2. Co , AASHTO T1 Lo Dk, ASTM 52 VCA mix < VCA dry Step 5: Oth val oner E uati s Agg. ? 1 NMAS Aggregate breakdown evaluation NCAT Ignition oven 5 Aggs? 3 NMAS ? 3 comp. efforts ? 2 replicates = 90 tests Permeability Test Falling Head Permeameter, SGC samples 5 Aggs ? 3 NMAS? 2 Comp. levels ? (2-4) AC ? 3 replicates = about 270 tests Drai n P rties B wit m sh, 177?C s ? MA hi ? 2 at 30 t ndow rope asket h 6.3 m me 5 Agg 3 N S? 1 ghest OAC replic es = ests 102 For easy comparison of the effects of aggregate properties, the same gradation (the N gradations) was initially used for all five aggregates. However, for ruby granite and traprock, the optimum asphalt contents for a lower gyration level were higher than would realistically be used (as high as 8.5 percent) when the N gradations were used. This is likely due to two aggregate characteristics: L.A abrasion and F&E content. These two aggregates have the lowest L.A abrasion loss and lowest F&E content (as shown in Table 4.1). Less aggregate breakdown during compaction and more cube-shaped aggregate will result in higher VMA values and higher optimum asphalt content. Therefore, finer gradations (named F gradations) were used for these two aggregates. The gradations used in this study are summarized in Table 3.3 and shown in Figure 3.3. TABLE 3.3 Gradations Used in This Study Percent Passing, % Gradation Code Sieve Size (mm) 1? 3/4" 1/2" 3/8" #4 #8 #16 #30 #50 #100 #200 19 100 93 59 33 20 16 14 12 10 9 8 N 12.5 100 93 59 24 18 15 12 10 9 8 9.5 100 82 35 22 17 14 11 9 8 19 100 95 69 42 24 19 15 12 10 9 8 F 12.5 100 95 65 26 22 18 15 13 12 11 9.5 100 82 35 22 18 16 14 12 11 0 10 20 30 40 50 60 70 80 90 100 0.45 Pow er Sieve Size, m m P e rcen t P a s s i n g , % 19N 12.5N 9.5N 19F 12.5F 9.5F 0.075 1.18 2.36 4.75 9.5 12.5 19 25 FIGURE 3.3 Gradations used in this study. 103 FIGURE 3.4 Work plan for phase III: performance tests. APA 64?C, 100psi hose, 100 LB load, 8000 clycles, 32 mixtures Step 7: Performance Test 32 mixtures APA Samples 150?115 mm 6?0.5% air voids Triaxial Tests: 32 mixtures Repeated Load Confined Creep 60?C Haversine load control-stress 0.1s loading, 0.9s rest 20 psi confining stress 96 tests 120 psi peak normal stress Test continue until 10,000 cycles or tertiary flow. 32 mixtures ? 3 replicates = Static Creep ? Use same samples as Dynamic Modulus 60?C Normal stress 120 psi, Confining stress 20 psi. H terti 96 tests old load constant until ary flow or reach LVDT limits or 5 hours 32 mixtures ? 3 replicates = Step 8: Data Analysis and Discussion, Prepare Draft Report. Step 9: Prepare Final Report Dynamic Modulus 60?C 20 psi confining pressure 25 Hz, 10 HZ, 5 Hz, 1 Hz, 0.5 Hz, and 0.1 Hz. Load to produce 50-150 microstrain 32 mixtures ? 3 replicates = 96 tests Performance Tests Step 6: Performance Test Sample Preparation All mixtures designed with SGC: 32 mixtures Triaxial testing samples 100?150 mm, 4? core from 6? SGC sample, 4?0.5% air voids 104 The SMA mixtures with higher asphalt content are more susceptible to draindo problems if all other influencing factors are identical. Draindown tests were conducted on wn urces and NMAS with the highest optimum asphalt ompaction efforts, to ensure all the designed SMA mixtures can irement. Two replicated tests were conducted for each l ted Ignition oven tes voids close to 4 percent, or that were near optimum asphalt content. This test was used termine the aggregate breakdown due to pac on aft own due to the ignition process. The test plan for the third phase is shown in Figure 3.4. One of the most remarkable characteristics of SMA mixtures compared to the conventional dense-graded mixtures is its excellent rutting resistance. Also, the most critical property that must be evaluated when increasing the asphalt content is rutting potential. Therefore, the performance tests all 15 combinations of aggregate so content from the three c meet the draindown test requ mixture. The volumetric properties of SMA mixtures are very important to ensure the durability and stone-on-stone contact of SMA mixtures. Volumetric properties were determined for the mix design samples, including air void content, voids in minera aggregate (VMA), and voids in coarse aggregate in mixture (VCA mix ). Based on the literature review, SMA mixtures tend to become permeable at lower air voids content than that for dense-graded mixtures. Permeability tests were conduc on all mix design samples, to determine the threshold air voids value for SMA mixtures becoming permeable. ts were conducted for all the mix design samples that had air for removal of asphalt binder, and to de com ti er correcting the aggregate breakd 3.1.3 Performance Testing 105 conducted in this study mainly focused on the evaluation of rutting resistance, wh included APA test, repeated load confined creep, dynamic modulus and static creep tes 3.2 MATERIALS SELECTION ich ts. defined as the portion that remains on the 4.75 mm sieve. F The ic te 3.2.1 Aggregate Tests 3.2.1.1 Specific gravity test In this study, the bulk specific gravity of coarse aggregate and fine aggregate were conducted following the AASHTO T85 and T84 test methods (25), respectively. The coarse aggregate portion is generally or the coarse aggregate portion, the samples were tested in individual size fractions. The fine aggregate portion was tested together for the specific gradation. bulk specific gravity of the mineral filler is hard to test and therefore the apparent specif gravity was used as a substitution to calculate the combined bulk specific gravity of aggregate. The equation used to calculate the combined specific gravity of the aggrega sample is shown as follows: n GGG 21 n +++ ... 21 (3.1) G = combined specific gravity; G 1 , G 2 ,? G n = specific gravity values for fraction 1, 2, ?, n; and n PPP PPP G +++ = ... 21 where, P 1 , P 2 ,? P n = weight percentages of fraction 1, 2, ?, n. 3.2.1.2 L.A abrasion test The Los Angeles (L.A.) abrasion test is most often used to evaluate the toughness and abrasion of the aggregates. When the L.A abrasion is too high, excessive aggregate breakdown may occur during handling, compaction, and traffic, resulting in potential 106 bleeding, rutting, or raveling. The L.A. abrasion test was conducted in this study following ASTM C131 (3). The gradation used in this study was the B grading (25 gram 12.5 mm aggregates and 2500 gram 9.5 mm aggregates). 3.2.1.3 Flat and elongated content test Flat and elongated particles tend to break during compaction and under traffic, thus t may adversely affect the stability and d 00 hey urability of the compacted HMA. The flat and elongated particles in using a proportional caliper device. ted s e aggregate, approximately 100 particles were e percent by weight of particles that had a ension to the shortest dimension greater than 3:1 or 5:1 were n system specifies a minimum angularity for the fine aggregate portion e fine aggregate can be quantified by the use of AASHTO T304 (25), ?Uncompacted Void Content of Fine aggregate?. In this method, coarse aggregates were determined following ASTM D4791 (47), The individual fractions of each sieve size equal to or greater than 4.75 mm were tes eparately. For each sieve siz obtained following AASHTO T248 (25). Th ratio of the longest dim recorded. The flat and elongated content of certain blended aggregate depends on the gradation and can be calculated based on the weight fraction of individual sizes and the flat and elongated content of each size. 3.2.1.4 Fine aggregate angularity The Superpave mix desig of asphalt mixtures to increase internal friction (shear strength) and reduce the rutting potential of the mix. However, there have been many controversies on whether fine aggregate angularity (FAA) related to rut resistance (103). In this study, The FAA values for each aggregate type were determined, and used as an input in the data analysis. The particle shape and surface texture of th 107 a 100 cm 3 cylinder is filled with fine aggregate of a certain gradation by allowing the sample to flow through the orifice of a funnel into the calibrated cylinder. Excess material is struck off and the cylinder with aggregate is weighed. Uncompacted void content of the sample is then calculated using the weight and the bulk specific gravity of the aggregate. A high uncompacted void content is an indication of good aggregate angularity and coarse surface texture. 100 )/( ? ? = GFV U (3.2) where: V = volume of cylinder, mL; F = net mass of fine aggregate, g; and G = bulk dry specific gravity of fine aggregate. V U = uncompacted voids in the material, percent; 3.2.1.5 Uncompacted air v Coarse aggregate angularity is believed to have a significant effect on mixture rutting ended by the The uncompacted air voids of coarse aggregate test followed AASHTO TP56 gradation for Method A. For 12.5 mm NMAS, the aggregate ggregates and 2020 grams of 4.75 mm aggregates. oids of coarse aggregate performance. The uncompacted air voids of coarse aggregate was recomm NCHRP 4-19 project (113) to specify the combination effects of aggregate shape, angularity, and texture. (106), and use the standard combination is 1970 grams of 9.5 mm a The test procedure and calulation is similar to that of the FAA test, but in a large scale to accommodate the larger aggregate sizes. 108 3.2.1.6 Voids in coarse aggregates The stone on stone contact is one of the key criteria for designing SMA mixtures. The existence of stone on stone contact can be determined by comparing VCA mix and VCA drc . The VCA mix can be calculated (104) from the bulk specific gravity G mb of a compacted SMA sample, aggregate gradation, and bulk specific gravity o f total aggregate G sb . The dry rodding test for VCA drc followed AASHTO T 19 (25), ?Unit weight and Voids in Aggregate?. In this method, the coarse aggregate fraction (retained on 4.75 mm sieve for 19 mm ry- he and 12.5 mm NMAS gradation, and retained on 2.36 mm sieve for 9.5 mm NMAS gradation) are filled and dry rodded in three layers into a metal cylinder container. Excess material is struck off and the cylinder with aggregate is weighed. When the d rodded density of the coarse aggregate fraction has been determined, the VCA drc for t fraction can be calculated using the following equation: 100)1( ??= wca s G ? (3.3) where: ? drc VCA ? VCA drc = voids in coarse aggregate in the dry-rodded condition, percent; G = bulk dry specific gravity of the coarse aggregate. ic shear rheometer test lowing the AASHTO T315 procedure (25). Althoug R test is testi r, it i method to characterize fine mortars (42). The fine mortar is a mix of asphalt binder and s = unit weight of the coarse aggregate fraction in the dry-rodded condition; ? w = unit weight of water (998kg/m 3 ), and ca 3.2.2 Fine Mortar Tests 3.2.2.1 Dynam The dynamic shear rheometer (DSR) test was conducted fol h the DS not designed for ng morta s a good 109 m at p eve. In this study, the DS as con f . 9 c binder con and 3 filler c tents. T used for the DSR test were both original PG 76-22 asphalt and the rolling thin-film oven (RTFO) aged asphalt binder. The test temperature was set at 76?C in order to evaluate fine mortar properties at high pavement service temperature. FIGURE 3.5 Components of complex modulus G*. deformation when repeatedly sheared. As shown in Figure 3.5, complex modulus G* consists of two components: storage modulus G? or the elastic part and loss modulus G? or the viscous part. For rutting resistance, a high complex modulus G* value and low phase angle ? are both desirable. A higher G* value indicates the fine mortar is stiffer and thus more resistant to rutting. A lower ? value indicates a more elastic fine mortar thus more resistance to permanent deformation. In the Superpave asphalt binder specification, the G*/sin? parameter was chosen as the rutting parameter. And for fine mortar, a ineral filler th asses 0.075 mm si R test w ducted on 9 ine mortars, i.e ombinations of 3 tents on he binders The DSR measures the complex shear modulus G* and phase angle ? of fine mortar at the desired temperature and frequency of loading. Complex modulus G* can be considered as the total resistance of the fine mortar to G? Viscous Part ? G* G? Elastic Part 110 recomm ended specification of the rutting parameter is listed in Table 3.4. The minimum requirements for G*/sin? are 5 and 11 kPa for original and RTFO aged mortar, respectively. TABLE 3.4 SMA Mortar Quality Requirements (43) Test Material AASHTO Method Property Specification Dynamic Shear Original Mortar TP 5 G*/sin? ? 5.0kPa Dynamic Shear RTFO Aged Mortar TP 5 G*/sin? ? 11.0 kPa Bending Beam PAV Aged Mortar TP 1 S ? 1500 MPa 3.2.2.2 Bending beam rheometer test Th g b ome onduc ing the AASHTO T313 (2 dure. BBR tes as condu -12?C in this study to evaluate the fine mortar properties at low pave nt servic erature. The asphalt binder in the fine mortar for this test was press aging v AV) aged to simulate the long term aging afte ral yea service. Two parameters were recorded during the BBR tests. One is creep stiffness, S(t), which is a measure of how th sphalt bi sists the constant loading. The other is m- value, which is a measure of the rate at which the creep stiffness changes with loading time. As S(t) increases, the thermal stresses developed in a pavement due to thermal shrinking increases proportionally to temperature change. The thermal cracking becomes more likely with higher S(t) values. Therefore, a recommended maximum limit for PAV aged fine mortar is 1500 MPa as shown in Table 3.4. e bendin eam rhe ter (BBR) test was c ted follow 5) proce The t w cted at me e temp ure essel (P r seve rs in e a nder re 111 3.3 MIXTURE DESIGN PROCEDURES 3.3.1 Mixture Design by Marshall Hammer The majority of the early SMA projects in the United States were designed with the Marshall procedure. However, the Marshall stability and flow were generally not included in the acceptance criteria. The compaction effort used in the Marshall procedure to design SMA mixtures was typically 50 blows with a manual hammer. Since the mechanical hammer is normally used and available in the laboratory, a calibration was conducted to give an equivalent blow number by mechanical hammer that results in a similar density as the manual hammer. The M er with a static base. The calibration used lab granite and 12.5 mm NMAS with 6.5 percent asphalt content to represent a typical SMA mixture. The equivalent blow number was calibrated to be 59 blows. The calibration data is shown in Table 3.5 and Figure 3.6. TABLE 3.5 Automatic Marshall Hammer Calibration Data No. Average arshall compactor used in this study was a mechanical hamm Method Blows Sample T166 1 2.328 2 2.334 Manual 50 3 2.335 2.332 1 2.318 2 2.321 50 3 2.316 2.318 1 2.329 2 2.338 60 3 2.333 2.332 1 Auto 2.342 2 2.344 70 3 2.350 2.345 112 y = 0.0014x + 2.251 R 2 = 0.9959 2.31 2.33 45 50 55 Gm b 2.32 2.35 2.36 60 65 70 75 2.34 T1 6 6 Manual Marshall Blow s w ith automatic hammer A 50 A 60 A 70 Average Average results by manual hammer 59 Blow s FIGURE 3.6 Automatic Marshall hammer calibration. A few trial asphalt contents were used to determine the optimum asphalt content that produced the SMA mixture with 4 percent air void content under the designated compaction effort. The number of trial asphalt contents used was as few as possible, as ing long as the 4 percent air void content was achieved. The compaction temperature was controlled within 149 and 154?C (300 and 310?F). This temperature range was selected based on supplier information. The loose mixture was packed into the Marshall mold and placed in an oven at this compaction temperature until the temperature was satisfactory for compaction. No short term ag was conducted before the compaction. 113 3.3.2 M al was 1.27?) and contact pressure of 600 kPa. 44) procedure. Short term aging of the loose mix was performed in the oven set at 149?C (300?F) for two hours in accordance with the AASHTO PP2 (44). The compaction temperature was set at 149?C (300?F). Several trial asphalt contents were used to determine the optimum asphalt content that could produce the SMA specimens with 4 percent air void content under the designated compaction levels. The number of trial asphalt contents used was as few as possible, as long as the aindown Test SHTO T-305 (25) test imately 1200 grams was prepare ioned own as recorded as a percent of the total mixture, by subtracting the initial plate mass fr ixture Design by Superpave Gyratory Compactor The Superpave gyratory compactor used in this study was manufactured by the Pine Instrument Company. It had an average internal gyration angle of 1.23? (average extern gyration angle The compaction of SMA samples with the SGC followed the AASHTO PP41 ( 4 percent air void content was achieved. 3.3.3 Dr The draindown test conducted in this study followed the AA procedure. A sample of the SMA mixture with a mass of approx d in the laboratory. The sample was placed in a wire basket, which was posit on a plate of known mass. The sample, basket, and plate were placed in a forced draft oven for one hour at a pre-selected temperature. At the end of one hour, the basket containing the sample was removed from the oven along with the plate and the mass of the plate was determined. The amount of draindown was then calculated. The draind test result w om the final plate mass and dividing this by the initial total sample mass. 114 For each mixture, two replicates were tested. The test temperature used in this study was 177?C (350?F), which is approximately 15?C above the anticipated plant production temperature for the SMA mixtures. The sieve cloth size of the wire basket was 6.3 mm. 3.4 TESTS CONDUCTED ON MIX DESIGN SAMPLES 3.4.1 Air Void Content Determination The air void content is defined as the total volume of the small pockets of air between the coated aggregate particles throughout a compacted paving mixture, expressed as a percent of the bulk volume of the compacted paving mixture. The formula used for calculating the percent air voids is shown below: 100)1( ??= mm mb G G VTM (3.4) M pecimen; and f mixture. ulk Where, VT = voids in total mix; G mb = bulk specific gravity of compacted s G mm = the theoretical maximum specific gravity o Since G mm can be obtained by following AASHTO T 209 (25), Theoretical Maximum Specific Gravity and Density of Bituminous Paving Mixtures, the value of VTM depends primarily on how G mb is measured. There are several different methods to measure the bulk specific gravity. The difference in the various methods is primarily due to the difference in the volume measured for each sample since the sample mass is identical. Two methods were used in this study to determine the air void content of compacted SMA mixtures. One was the standard method used to measure the SMA b 115 specific gravity by following AASHTO T166 (25), Bulk Specific Gravity of Compac Bituminous Mixtures Using Saturated Surface-Dry Specimens, commonly known a saturated surface dry (SSD) method. The other method used was the vacuum seal method by following ASTM D6752 (47), Standard Test Method for Bulk Specific Gravity and Density of C ted s the , ompacted Bituminous Mixtures Using Automatic Vacuum Sealing Method. The SSD method consists of first weighing a dry sample in air, then obtaining a submerged mass after the sample has been placed in a water bath for a specified time interval. Upon removal from the water bath, the SSD mass is determined after patting the sample dry using a damp towel. The bulk specific gravity of the sample can then be calculated by using the following formula: subSSD dry mb WW G ? = W (3.5) Where, G mb = Bulk specific gravity of compacted specimen by SSD method W dry = Dry weight of compacted sample W SSD = SSD weight of compacted sample W sub = Submerged weight of compacted sample The major error of this method typically comes from the SSD weight. During Gmb testing for SMA with the SSD method, water can quickly infiltrate into the sample. However, after removing the sample from the water bath to obtain the SSD condition, the water can also d om the sample results in an incorrect SSD weight, and usually results in the measured G mb being higher than the actu mb rain from the sample quickly. This draining of the water fr al G . The vacuum seal method uses a vacuum-sealing device (manufactured by CoreLok) to keep water from entering the sample and thus avoid the water-draining 116 problem. This vacuum-sealing device utilizes an automatic vacuum chamber with a specially designed, puncture resistant, plastic bag. Under vacuum, the bag tightly conforms to the sides of the sample and prevents water from infiltrating into the sample. The volume of the specim the sample. ility nducted by following ASTM PS 129 (105) test procedure. ix design asphalt contents resulted in a range of air void tent ontent was used to examine the effect of aggregate type, air void content, and NMAS on permeability. en encapsulated by the bag is considered as the bulk volume of 3.4.2 Flexible Wall Falling Head Permeab The permeability test was co Specimens used for permeability testing were produced in the SGC in the m phase. For each mixture, several trial con s. This variability in air void c FIGURE 3.7 Flexible wall falling head permeameter. d sample. The sides of the sample were sealed by a confining rubber sleeve to prevent the A falling head permeability test was used in this study. The test device is shown in Figure 3.7. Water from an upright standpipe was allowed to flow through a saturate 117 possible leaking from sample sides. The time interval to reach a known change in head was recorded. The coefficient of permeability was then calculated from Darcy?s law as follows: )ln()( 2 hAt K = (3.6) 1 aL h = initial water head, cm; and 2 3.4.3 Ignition Oven Test The asphalt removal procedure used in this study was performed by the ignition oven test, which was developed at the National Center for Asphalt Technology (NCAT). The ignition oven test followed ASTM D6307 test procedure (47). This method determines the asphalt content and provides a clean aggregate sample. The gradation analysis on the samples after ignition was compared with the original gradations used in the SMA mixtures. The changes in each sieve were rec rded indicate the total de o additional breakd n h Where, K = coefficient of permeability, cm/s; a = area of standpipe; L = specimen height, cm; A = cross section area of specimen, cm 2 ; t = the measured time of flow; 1 h = final water head, cm. o to gradation due to compaction and possibly due t own in the ignition oven. The loose mixtures were also put into the ignition ove and followed the same procedure to get the possible gradation changes due to the high temperature in the ignition oven. Hence, the degradation due to compaction could be 118 obtained and used in the data analysis after the correction of possible breakdown in the ignition oven. 3.5 PERFORMANCE TESTING EQUIPMENT AND METHODS 3.5.1 Asphalt Pavement Analyzer (APA) Wheel Tracking new generation of Georgia Loaded Wheel Tester (GLWT). Figure 3.8. The device uses a wheel to apply a load to a rubber can rocess. The APA is an automated, The APA is shown in hose, which is in contact with the test specimens. The hose is air pressurized to the desired pressure. The device is configured with three loading stations so that three tests can be conducted simultaneously. Both rectangular slabs and cylindrical specimens be tested in the device. This test has been used by many States to measure the rutting potential of HMA mixture during the mix design and construction p FIGURE 3.8 Asphalt Pavement Analyzer. d st. The air void In this study, APA testing was used to evaluate the rutting potential for SMA mixtures designed with different compaction levels. Testing with the APA was conducte at 64?C. This temperature corresponds to the standard high temperature performance grade of asphalt cement for most project locations within the southea 119 content ere 690 cles 3.5.2 Dynamic Modulus 3.5.2.1 Testing setup The dynamic modulus test is one of the oldest and best documented of the triaxial compression tests. It was standardized in 1979 as ASTM D3497 (47), ?Standard Test Method for Dynamic Modulus of Asphalt Concrete Mixtures.? The test consists of applying a uniaxial sinusoidal (i.e. haversine) compressive stress to an unconfined or confined HMA cylindrical test specimen, as shown in Figure 3.9. FIGURE 3.9 e angle ? as shown in Figure 3.9, where ? 0 is the amplitude of strain, and ? is the angular frequency. of test specimens was 6.0?0.5 percent. Hose pressure and wheel load w kPa and 445 N (100 psi and 100 lb), respectively. Testing was carried out to 8,000 cy and rut depths were measured continuously. Rut depths were also measured manually after 8,000 cycles as recommended in AASHTO TP 63-03 (106). Time, t ?, ? ? 0 sin?t ?/? ? 0 ? 0 ? 0 sin(?t-?) Haversine loading pattern or stress pulse for the dynamic modulus test. Within a small stress and strain range, the asphalt mixture tested is deemed as viscoelastic material. If a cyclic stress with a constant amplitude ? 0 is applied to a specimen made of viscoelastic material, the strain response will be an oscillation at the same frequency as the stress but lagging behind a phas 120 In this stu at a temper de f 50 d environmental chamber used for this study are shown in Figure 3.10. The specimens were tested in an order of decreasing load frequency including 25Hz, 10Hz, 5Hz, 1Hz, 0.5Hz, and 0.1Hz. This frequency sequence was carried dy, the test was conducted in an environmental chamber ature of 60?C (140?F) with a 20 psi confining pressure. The vertical load was applied by a servo-hydraulic system: the Material Testing System (MTS). The magnitu of the applied load was decided by controlling the responding strain within a range o to 150 microstrain. The MTS an out to cause minimum damage to each specimen before the next sequential test. FIGURE 3.10 The MTS and environmental chamber used for triaxial testing. The deformations were measured through three spring-loaded linear variable Ts). The LVDTs were evenly and vertically placed on the esulted in the LVDTs being approximately 120? apart. apart and located approximately 25 mm from ple differential transformers (LVD side of the specimens, which r Parallel brass studs that were glued 100 mm the top and bottom of the specimens were used to secure the LVDTs in place. A sam ready for the triaxial testing is shown in Figure 3.11. 121 FIGURE 3.11 A sample prepared for triaxial testing. 0.025 inch thi extende In the raw data analysis phase, two equations: equations 3.7 and 3.8 were used to simulate the stress and strain m data file to p square of the error were set up. The Microsoft Excel solver program then was employed to minimize the sum of squares error cell by changing the cells with the equation easured data to ensure the lver may give a solution that is wrong, depending on ed. a= Since it was a confined test, a flexible membrane with the 4 inches diameter and ckness was used to cover the tested sample. The flexible membrane was d over the top and bottom platens and confined with two O-rings to seal the sample from the confining air. 3.5.2.2 Method of analysis easured data, respectively. The columns were set up in a redict stress and strain using these equations, and then columns for the coefficients. The fitted data was plotted against the m simulation was correct, since the so the initial values assum )sin( 11 cttb o +?++ 1 ??? (3.7) )sin( 2022 cttba +?++= ??? (3.8) 122 where, ? = predicted strain at time t; a 1 , a 2 , b 1 , b 2 , c 1 , c 2 = other coefficients of simulative equations. Once the fitted equations were determined, the coefficients of the equations could urther analysis. The stress-to-strain relationship under a continuous aterials is defined by a complex number called defined . ? = predicted stress at time t; ? o = predicted peak stress; ? o = predicted peak axial strain; ? = angular frequency of dynamic stress and strain; and be used to conduct f sinusoidal loading for linear viscoelastic m the ?complex modulus? (E*). The absolute value of the complex modulus, |E*|, is as the dynamic modulus. The dynamic modulus is mathematically defined as the maximum (i.e., peak) dynamic stress (? o ) divided by the peak recoverable axial strain (? o ) 0 ? ? The real and imaginary portions of the complex modulus (E*) can be wri 0 * =E (3.9) tten as =* (3.10) E? is generally r modulus; E? is referred to as the loss or viscous modulus. The phase angle, ?, is ? lags behind ? . It is an indicator of the viscous properties of the sed as Ei ??+? EE eferred to as the storage or elastic modulus component of the complex the angle by which o o material being evaluated. Mathematically, this is expres ?? sin*cos** EiEE += (3.11) 360?= p i t t ? (3.12) 123 or ? ? 180 )( ??= cc (3.13) where t 21 or a pure elastic material, ? is equal to 0, and the complex modulus (E*) is equal to the absolute value, or dynamic modulus. For pure viscous materials, ? is equal to 90?. t high temperature, a higher complex modulus E* indicates a stiffer asphalt mixture, therefore more resistance to deformation. A lower phase angle ? indicates a more elastic asphalt mixture resulting in quicker recovery and less permanent deform ion. The dynamic complex modulus term E*/sin? under high temperature has been suggested (84) to be used as an indication of rutting resistance. The dynamic modulus is also 3.5.3 S p for a mixture is measured in the laboratory under unco , using either one load-unload cycle or incremental load-unload cycles, provides sufficient rmine the instantaneous elastic (i.e., recoverable) and plastic (i.e., are time independent), as well as the viscoelastic and endent) of the material?s response. i = time lag between a cycle of stress and strain (s); t p = time for a stress cycle (s); and i = imaginary number. F A at very important for mechanic pavement design. tatic Creep 3.5.3.1 Background of Creep Behavior In a static compressive creep test, a total strain-time relationshi nfined or confined conditions. The static creep test information to dete irrecoverable) components (which viscoplastic components (which are time dep These four components of creep strain can be expressed as: vpvept e ????? +++ = (3.14) 124 wh ere ? t = tot ain; ? e = ela train, recoverable and time independent; ? p = plastic strain, irrecoverable and time independent; ?ve = viscoelastic strain, recoverable and time dependent and ?vp = viscoplastic strain, irrecoverable and time dependent. igure 3.12 illustrates the creep behavior of HMA mixtures. The load duration is t 1 and the rebound time is t 2 -t 1 . When the load is applied at t=t 0 , a strain ? 0 containing the elastic and plastic components appears instantaneously. During the load duration, viscoelastic and viscoplastic strain occur. Once the load is removed (t = t 1 ), the elastic rain is recovered instantaneously. In the rebound period, the viscoelastic strain is recovered. At th rain consists of the plastic and viscoplastic strains. re in on for Krass?s model is given as al str stic s F st e end of the rebound period (t = t 2 ), the permanent creep st FIGURE 3.12 HMA creep behavior in static creep test. This creep behavior can be modeled by Krass?s model (107). As shown in Figu 3.13, Krass?s model consists of a skidding block, a Maxwell model and a Kelvin model series. The constitutive equati Strain Time, t ? ? vp + ? ve ? e ? v e ? p ? v ? p ? 0 t 0 t t 2 125 )1( 212 2 0 1 10 1 0 ? ? ? ?? ?? pskt EE = ? psk = plastic strain due to skidding block; llel; ? 1 = viscous constant of dashpot in series; dashpot in parallel; and 1 oved. FIGURE 3.13 Krass?s model for creep behavior (107). Comparing the equations 3.14 and 3.15, elastic strain ? e = tE e t ? ?+++ (3.15) where ?0 = applied constant stress; E 1 = elastic constant of spring in series; E 2 = elastic constant of spring in para ? 2 = viscous constant of t = time when constant stress was rem Deformation 126 10 E? , plastic strain ? p ? psk , viscoelastic strain ? ve = = )1)(/( 212 ?tE? 20 ? eE ? , and viscoplastic strain ? vp = 110 /?? t . Load Time, t t 1 Time, tt 1 ? 1 E 1 E 2 ? 2 3.5.3.2 modulus test is generally recognized as a non-destructive test because of the l static r testing for dynamic modulus. The equipment setup and condition were the sam were he it. The valid range for the LVDTs used in this study was 5 mm. While measur the (usually less than 1 hour). 3.5.3.3 Method of analysis A typical relationship between the calculated total compliance and loading time is shown in Figure 3.14. Testing Setup Since the dynamic ow stress/strain applied, the same specimens and setup were used to conduct creep tests afte e as those used in dynamic modulus test. The test temperature for the static creep test was 60?C. The confining pressure was 20 psi. The static compressive load, or the continuous applied load used in this study was 1257 lb, which resulted in the contact pressure of 100 psi. Therefore, the major and minor principle stresses for this test 120 and 20 psi, respectively. The program was set to automatically stop the test when any LVDT reached t maximum lim ing the deformation between two points at 100 mm apart, the LVDT displayed individual strain up to 5 percent. The test was also stopped when a test time was longer than 5 hours and no indication was shown that the test would be completed within a short period of time 127 Compliance Volume change ?V>0 Volume ?V FIGURE 3.14 Typical test results between compliance and loading time. ng secondary zone: the portion in which the strain rate is constant with loading time; and 3. The tertiary flow zone: the portion in which the strain rate increases with loading time. Ideally, the large increase in compliance occurs at a constant volume within the tertiary zone (84). The starting point of tertiary deformation is defined as the flow time, which has been found to be a significant parameter in evaluating an HMA mixture?s rutting resistance (84). The flow time also is viewed as the minimum point in the relationship of the rate of change of compliance to loading time. The flow time, FT, is therefore defined as the time at which the shear deformation under constant volume begins (84). Details on compliance models and regression parameters are available in the NCHRP reports on simple performance test (84, 86). In general, power models are used As shown, the total compliance can be divided into three major zones: 1. The primary zone: the portion in which the strain rate decreases with loadi time; 2. The Secondary Tertiary Flow Time When Shear Deformation Begins Primary D(t) change =0 Time, t 128 to model the secondary (i.e., linear) phase of the creep compliance curve, as illustrated in Figure 3.15. FIGURE 3.15 Regression constants a and m obtained from the secondary zone of the log compliance?log time plot. (3.16) where D(t) = total compliance at any time; neral, the larger the value of a, the larger the compliance value, D(t), the lower the modulus, and the larger the permanent deformation. For a constant a-value, an increase in the slope parameter m means higher rate of permanent Log Time ?t? Log D(t) Slope ?m? Intercept ?a? Strain m atDtDD =?=? 0 )( D? = viscoelastic compliance component at any time; D o = instantaneous compliance; T = loading time; and a, m = materials regression coefficients. The regression coefficients, a and m, are generally referred to as the compliance parameters. These parameters are general indicators of the permanent deformation behavior of the material. In ge deformation. 129 3.5.4 Repeated Load Confined Creep 3.5.4.1 Testing setup Another promising approach to measuring th permanent deformation characteristics of an HMA mixture is to conduct several thousand repetitions of a repeated load test and to record the cumulative permanent deformation as a function of the number of load cycles (i.e., repetitions). A schematic graph of stress and strain relationship in repeated load test is shown in Figure 3.16. A load cycle consisting of a 0.1 second haversine pulse load and a 0.9 second dwell (i.e., rest) time is applied for the test duration?10,000 loading cycles or about 3 hours in this study. The sample setup and test environmental conditions were the same as those used in the dynamic modulus and the static creep test. The test temperature for the static creep test was 60?C and the confining pressure was 20 psi. The load magnitude (peak value) used in this study was 1257 lb, which resulted in a peak contact pressure of e FIGURE 3.16 Repeated load test schematic graph. Stress ? Loading Cycles Loading Cycles Cumulative Strain 130 100 psi 3.5.4.2 Method of analysis Results e loading cycles. g secondary) portion of the cumulative plastic strain?repetitions relationship. . Therefore, the major and minor principle stresses at the peak for this test were 120 and 20 psi, respectively. from repeated load tests are typically presented in terms of the cumulative permanent strain versus the number of loading cycles. Figure 3.17 illustrates such a relationship. Similar to the creep test, the cumulative permanent strain (? p ) curve can b divided into three zones: primary, secondary, and tertiary. The cycle number at which tertiary flow starts is referred to as the ?flow number?. FIGURE 3.17 Typical relationship between total cumulative plastic strain and Figure 3.18 illustrates the same relationship plotted on a log?log scale. The intercept a represents the permanent strain at N = 1 whereas the slope b represents the rate of change of the permanent strain as a function of the change in loading cycles (lo [N]). These two permanent deformation parameters are derived from the linear (i.e., N FN(Flow Number) Tertiary Permanent Strain (in/in) Primary Secondary Loading Cycles 131 FIGURE 3.18 Regression constants a and b when plotted on a log?log scale. d b is typically used to analyze th The regression c zone of material deformability (Figure 3.17) and ar ion estimation of param n ana ear portion of the permanent strain v cycles. The flow ded where the minimum slope occurs (84), or just before the slope begins to increase. ed The classic power-law model, mathematically expresse e test results: y Equation 3.17, b p aN=? (3.17) onstants a and b ignore the tertiary e dependent on the material?test combinat conditions. The eters a and b are obtained from a regressio ersus number of lysis of the lin number is recor 3.6 PERFORMANCE TEST SPECIMEN PRODUCTION 3.6.1 APA Specimen Production The test specimens prepared for APA rut testing had a diameter of 150 mm, a height of 115? 5mm, and air void content of 6.0 ? 0.5 percent. For each mixture, a set of six specimens was prepared. The specimens were compacted at optimum asphalt content with a reduc gyration level based on the compaction curve obtained from the mix design phase to Intercept ?a? Slope ?b? ? p = a N b log (N) log ? p (N) 132 obtain the appropriate air voids. Accordingly, the amount of loose mixture was also reduced in order to get the 6 percent air void content and the designated height range. Th intent of these procedures was to produce specimens having densities representative of those present in-place after construction and to produce specimens having similar air voids in order to make comparisons between the different design compaction levels. e 3.6.2 Triaxial Test Specimen Production The specimens prepared for triaxial tests had average dimensions of 100 mm in diameter and 150 mm in height. The test specimens were cored and sawn from gyratory compacted mixtures. The target air void content for the test specimens was 4 ? 0.5 percent. The detailed requirements on the geometric properties of triaxial samples follow NCHRP report 465 and 513 (84, 86), and are listed in Table 3.6. TABLE 3.6 Geometric Requirements for Triaxial Samples (84, 86) Item Test Procedure Requirements Diameter Measure at the mid height and third points along axes that are 90 degrees apart, a total of six Within 102? 2 mm Standard deviation ? measurement for each sample 2.5 mm Height At least three measurements at approximately 120? intervals for each sample Within 150? 2.5 mm Air Void Content AASHTO T269 Within target ? 0.5 percent End Smoothness Check a minimum of three positions at approximately 120? intervals using a straight edge and feeler gauges approximately 8-12.5 mm wide or an optical comparator Within ? 0.05 mm across any diameter Perpendicularity axis of the specimen by using a machinists square perpendicular by more Measure angle between the specimen end and the and feeler gauges Not depart from than 0.5 degrees This preparation can provide more homogeneous test specimens (108), also provide a height-to-diameter ratio of 1.5, which is recommended as the minimum ratio to 133 ensure the response of a sample evaluated represents a fundamental engineering prop (84). The smooth, parallel specimen ends were needed to eliminate the end friction and violation of the theoretical boundary effects of the specimen during th erty e test. The smooth surface also allows better mounting for the LVDTs. In order to get the test specimens described above, SMA mixtures were compacted in the SGC with the final height of 170 mm. About 10 mm from both ends of the compacted specimen was removed, and a 100 mm (4 inches) core was drilled from each 150 mm (6 inches) specimen. The gyratory specimen, sawing and drilling devices should be adequately supported to ensure the resulting test specimen is cylindrical with parallel ends, and with sides that are smooth, parallel, and free from steps, ridges, and grooves. Figure 3.19 shows a comparison of a 150 mm (6 inches) sample and a 100 mm (4 inches) cored sample side by side. FIGURE 3.19 Whole and cored sample prepared for triaxial testing. The air void content of the whole specimen is generally higher than that of the test specimen after coring and sawing because of the distribution of the air voids (108). The 134 surface of the lab compacted sample is likely ave higher air voids because the surface friction and other edge effects between the m old provide resistance to densification during compaction. Therefore, a higher target air void content for the whole gyra specimens. The specific target a mixtures with different ht and hance of over to h ixture and m tory specimens were used in order to get a 4 percent air void content for test ir voids varied for different aggregates and NMAS. The gyratory compaction was controlled by sample heig adjusting the sample mass of the mixtures, therefore, the gyration number varied for individual specimen. A maximum gyration number was also set to eliminate the c compaction due to errors in estimation for various mixtures. 135 TE TEST SULT ALYSIS AND DISCUSSION ON MIX DESIGN PROPER This chapter presents the m p ix design results using different ethods on air void content, permeability test results, Discussion is emphasized on the volumetric properties comparison for different compaction efforts, air voids measurement effects on volumetric properties, influencing factors on permeability of SMA mixtures, and degradation comparison for different compaction efforts. Five ag ts aterials used in the preparation of each mixture. 4.1.1 Coarse Aggregate Properties CHAP R 4 RE S, AN TIES aterial pro erties, m compaction efforts, effects of test m and aggregate breakdown results due to different laboratory compaction efforts. 4.1 MATERIAL PROPERTIES gregates with a range of LA abrasion loss values were selected for this study. They were crushed gravel, lab stock granite, ruby granite, limestone, and traprock. All aggregates were used to design three NMAS mixtures: 19 mm, 12.5 mm, and 9.5 mm. In addition, each SMA mixture incorporated the same mineral filler ?marble dust?, cellulose fiber stabilizer and a polymer modified PG 76-22 asphalt binder. This section documen the properties of each of these m The specific gravity, L.A. abrasion, flat and elongated (F&E) content, and the uncompacted air voids of coarse aggregates are shown in Table 4.1. 136 TABLE 4.1 Aggregate Properties Bulk Gravity LA Loss , % F&E ratio, % F&E ratio, % Uncompacted 3 Fine Aggregate Type Specific 1 Abrasion 2 Content 1 3:1 Content 1 5:1 Air Voids of Coarse aggregate , % Aggregate Angularity, % C. Gravel 2.600 30.7 35.2 9.4 48.4 50.0 L. Granite 2.666 36.4 28.1 2.4 47.8 49.2 Limestone 2.730 26.4 25.5 3.6 46.6 47.1 R. Granite 2.702 20.6 23.4 4.4 47.1 48.9 Traprock 2.927 16.6 17.7 3.9 48.5 48.7 1. The bulk specific gravity and F&E co shown is for 12.5mm NMAS gradation. ntent depends on the combined gradation and NMAS; the value .A aggregates ranged from 16.6 to 36.4 percent, and two of five aggregates exc high limit of 30 percent for SMA mixture (43). The reasons for choosing the range of L.A asion were to represent the various aggregates that have been used in SMA mixtures and to determine the effect of aggregate L.A. abrasion on the design compaction level The F&E content (3:1 ratio) of aggregates ranged from 17.7 to 35.2 percent. Only one of five aggregates has the F&E content below the suggested maximum limit of 20 perce ). However, two of the oth gregate sources (lab granite ts, and it was believed the range ity to evaluate the effect of F&E on ance properties. The uncompacted air voids of coarse aggregate were within a narrow range from e e 2. LA abrasion values are based on B grading in ASTM C131. 3. Uncompacted air voids of coarse aggregate results are based on AASHTO TP56, method A with 12.5 mm NMAS. The L . abrasion value of the five eeded the suggested . abr . nt for SMA mixture (43 er ag and ruby granite) have been used on Georgia SMA projec in F&E content values would provide an opportun SMA perform 46.6 to 48.5 percent. There is no specific requirement for this aggregate property in th SMA mix design guides (43). However, this aggregate property was tested because som studies (103, 113) have shown it to have a good correlation with rutting performance. 137 4.1.2 Fine Aggregate Angularity The fine aggregate angularity (FAA) which is determined from the uncompacted air void content ontents higher than the suggested minimum limit of 45 percent perties neral filler for all SMA mixtures in this study. The properties shown in Table 4.2. Mineral Filler Properties Particle Size Analysis of fine aggregate is also shown in Table 4.1. All of the fine aggregates had uncompacted air void c (43), and ranged from 47.1 to 50.0 percent. 4.1.3 Mineral Filler Pro Marble dust was used as mi of the mineral filler are TABLE 4.2 Sieve Size, mm Cumulative Percent Passing 1 , % 1.18 100 0.6 100 0.3 100 0.15 10 0.075 99.6 0.045 90.7 0.02 61.1 0.01 40.1 P operty r Value Apparent Spe ity 2 2.566 cific Grav Dry 37.3 -Compacted Voids 3 , % 1. rticle siz er. 2. Determined by AASHTO T-100. 3. D ethod (109). The asphalt binder used in this study was tested using the Superpave binder tests and binder was selected since it is the most common grade appropriate for SMA mixture in Determined S-200 laser pausing a Coulter L e analyz etermined by modified Rigden voids m 4.1.4 Asphalt Binder Properties graded according to the Superpave binder grading specification. The PG 76-22 asphalt 138 the southeaste n United States. The asr phalt was modified with a styrene-butadiene- styr S) po e binder racteriza and perfo e grading ary for the PG 76-22 is shown in Table 4.3. TABLE 4.3 PG 76-22 Asphalt Binder Properties Test Temper C) T lts R ment ene (SB lymer. Th cha tion rmanc summ ature (? est Resu equire Specific Gravity 1.0277 Orig , G*/sin? a) 7 in inal DSR (kP 6 1.650 1.00 m RTFO Aged DSR, G*/sin? (kPa) 76 3.304 2.20 min PAV Aged DSR, G*/sin? (kPa) 25 2831 5000 max PAV Aged BBR, Stiffness (MPa) -12 137 300 max PAV Aged BBR, m-value -12 0.370 0.300 min 4.1.5 Fiber Properties Cellulose fiber was selected for inclusion in the SMA mixtures since it had been used extensively in SMA in the United States and Europe. The properties of the cellulose fibers are shown in Table 4.4. The ash content, Ph value, average fiber length and mesh sieve analysis shown were provided by the manufacturer. TABLE 4.4 Cellulose Fiber Properties Property Value Requirements Fiber length (Method A), mm 6 6 max Passing through 0.15 mm (No. 100) sieve, % 66.71 70 ? 10 Ash Content, % 18.76 18 ? 5 Ph value 7.5 7.5 ? 1 Oil Absorption, times of fiber weight 5.15 5 ? 1 Moisture Content, % 2.97 5 max 4.1.6 Fine Mortar Properties The Superpave binder tests including the DSR test and BBR test were conducted on fine mortars that consist of various asphalt binder contents, various mineral filler contents and 0.3 percent cellulose fiber by total mix weight. The test results are shown in Table 4.5. 139 TABLE 4.5 Mortar Test Results content, % content, % Dust / Asphalt ratio G*/sin?, 76 o C, kPa G*/sin?, 76 o C, kPa Stiffness, S -12 o C, MPa Asphalt Dust DSR (Orig.), DSR (RTFO) BBR (PAV) 8 1.37 9.29 18.64 710 10 1.71 11.08 22.71 9995.5 12 2.06 13.86 30.26 1045 8 1.25 7.86 17.13 637 10 1.56 9.53 19.87 837 6.0 .46 958 12 1.87 11.20 27 8 1.15 7.36 16.17 607 10 1.43 8.70 19.33 766 6.5 10.59 23.55 891 12 1.72 Criteria -- -- ? 5 ? 11 ? 1500 The mortar was evaluated at three levels of asphalt content: 5.5, 6.0, and 6.5 percent, which covered the most commonly used range of asphalt content for SMA mixtures. The mineral filler content also had three levels: 8, 10, and 12 percent by total weight of aggregates (the aggregate type was not specified), which covered the gradation band limit on the No. 200 (0.075 mm) sieve for SMA mixtures. For all the combinations of asphalt content and mineral filler content, the DSR test results on the fine mortar with original and RTFO aged asphalt binder ranged from 7.36 to 13.86 kPa and 16.17 to 30.26 kPa, respectively. These results were higher than the minimum recommended criteria of 5 kPa and 11 kPa (43) recommended for SMA fine mortar with original and RTFO aged asphalt binder, respectively. For the same combinations of asphalt binder and mineral filler content, the BBR h PAV aged asphalt binder ranged from 607 to 1045 MPa. test results on fine mortar wit All the results were less than the recommended maximum limit of 1500 MPa (43) set for SMA fine mortar. The results of mortar tests indicated that the fine mortars with all the combinations of asphalt content and mineral filler content could meet the recommended 140 requirements for SMA fine mortar. Therefore, with the PG 76-22 asphalt binder and marble dust mineral filler used in this study, striction o halt cont nd mineral filler content were urred fro ortar property requi ts. All o asphalt contents and mine ller con evaluated his study c used if lumetric and performance p erties of mixture atisfactory 4.2 MIXTURE DESIGN PROPERTIES In t y, SM gned using both a Marshall compactor and a gyratory compactor. The Mars ompactio as conduct th a stati e, automatic Marshall hammer. The number of blows with the automatic ham automa 4.2.1 Marshall Mix Design The volumetric properties of SMA mixtures designed with the Marshall method are optimum asphalt content in a reasonable range. no re n asp ent a inc m fine m remen f the ral fi tents in t an be the vo rop SMA are s . his stud A mixtures were desi hall c n w ed wi c bas mer was calibrated to 50 blows with a manual hammer. It was determined that 59 blows with the tic hammer were needed to correlate with the 50 blow density of the manual hammer. The gyratory compactor used in this study had an average internal gyration angle of 1.23? and contact pressure of 600 kPa. A standard compaction level of 100 gyrations, a compaction level near the locking point and a compaction level below the locking point were used. This section includes the mix design volumetric properties from different compaction efforts, the draindown test results, and the discussion on these results. shown in Table 4.6. Crushed gravel (C.GVL), limestone (LMS), and lab granite (L.GRN) used a normal (N) gradation in the middle to coarse side of the gradation band (Table 3.1), while traprock (TRAP) and ruby granite (R.GRN) used a finer (F) gradation to keep the 141 TABLE 4.6 Marshall Mix Design Volumetric Properties 1 Summary Agg. NMAS VCA drc , % Opt. AC, % VMA, % VCA mix , % 19 40.7 6.5 17.7 34.1 12.5 41.2 6.6 18.4 38.0 C. GVL 9.5 41.5 6.6 18.0 35.9 19 39.4 6.3 18.0 34.5 12.5 41.0 6.3 18.1 37.8 L. GRN 9.5 41.4 6.3 17.7 35.8 19 41.7 6.1 17.8 34.6 12.5 41.6 5.9 17.4 37.5 LMS 9.5 41.1 5.8 17.1 35.6 19 41.1 6.7 18.4 36.4 12.5 42.1 6.4 18.0 39.0 R. GRN 9.5 42.0 6.8 18.7 37.8 19 42.4 6.5 19.0 37.6 12.5 42.1 6.3 18.4 40.0 TRAP 9.5 42.1 6.6 19.1 39.2 1. Volumetric properties are based on 4.0 percent air voids by AASHTO T166 method. ix opti were in a practical range of 5.8 to 6.8 percent based on 50 blow Marshall designs. Thirteen of fifteen mixtures had the optimum sphalt content greater the gene quirement of 6.0 percent minim sphalt co t. All VMA values were greater th inimum value of 17.0 perce nd ranged om 17.1 t 9.1 percen tone on st contact was achieved for all de ed SMA mix values being less than their corresponding VCA drc valu The volu tric prope results con ed the gra chosen ll aggreg types we easonable 4.2.2 Gyratory Mix Design 4.2.2.1 Locking points Th ard gyra evel used as 100 gy ions since this is the level typically ecified. The lower gyration level was selected based on the locking point concept (4, 66). Based on the compaction information obtained from the 100 gyrations mix design, For all fifteen designed SMA m tures, the mum asphalt contents a than ral re um a nten an the m nt a fr o 1 t. S one sign mixtures, as indicated by VCA es. me rty firm dations for a ate re r . e stand tion l w rat sp 142 the firs ompacted as replicates at each trial asphalt content, and up to four trial asphalt contents were used in the SMA mix design. For each mixture, the average locking point is provided in Table 4.7. The average locking point for each mix ranged from 51 gyrations to 63 gyrations. o equal or exceed the locking points for all mixtures, 65 gyrations was selected as the comparative lower compaction level. For comparison, a third level of 40 gyrations that represents a level below the locking point was used to design two 12.5 mm NMAS SMA m xtures, with lab granite and ruby granite aggregate sources. TABLE 4.7 Locking Point Results Summary Locking Point, gyrations t occurrence of a gyration number that gave a change in height less than 0.1 mm for two successive gyrations was recorded as the locking point. For each mixture, three samples were c T i Aggreg e NMAS Gradation Number of Samples Average St. Dev at 19 N 12 57 4.8 12 0 .5 N 9 57 4.C.GVL 1 9.5 N 9 51 5. 19 N 12 61 6.4 12.5 N 12 61 2.9 L.GRN 9.5 N 6 59 2.4 19 N 9 62 4.7 12.5 N 12 63 3.7 LMS 9.5 N 9 59 5.3 19 N 9 59 2.5 12.5 N 12 56 3.5 R.GRN N 12 53 4.8 9.5 19 N 12 54 4.1 12.5 N 12 53 5.0 TRA 9.5 N 12 52 3.5 P 19 F 9 55 3.8 12.5 F 9 56 4.8 R.GRN 9.5 F 12 57 2.7 19 F 9 55 3.5 12.5 F 6 54 5.1 TRAP 9.5 F 6 55 3.2 143 A histogram of locking point results for all 210 mix design samples (including 69 sampl gr nd k w N g on) 0 s is shown in Figu F E 4. istog f lock point results. d, the 65 gyrations provided about 95 percent confidence level to cover the locking point of designed SMA mixtures. es for ruby anite a traproc ith the radati with 10 gyration re 4.1. IGUR 1 H ram o ing Lo int q cking Po F r e ue nc y 68645248 605644 40 30 20 10 0 65 Mean 56.68 5.319 20 al togr Lock int StDev N Norm His am of ing Po As shown in Figure 4.1, the locking point results had an average of 57 gyrations and a standard deviation of 5.3 gyrations. If a normal distribution is assume A forward stepwise regression was employed to evaluate the influencing factors on locking point results. The factors evaluated include aggregate properties (L.A. abrasion, uncompacted air voids for coarse aggregates -- represents coarse aggregate angularity or CAA, FAA, F&E content), NMAS, and asphalt content. The regression results are shown in Table 4.8. 144 TABL Step 1 23456 E 4.8 Forward Stepwise Regression Results for Locking Point Constant 79.96 160.95 155.27 172.2 181.37 180.62 AC -3.74 -3.15 -2.33 -1.84 -1.68 -1.67 T-Value -7.95 -6.57 -4.43 -3.4 -3.03 -3.02 P-Value 0.000 0.000 0.000 0.001 0.003 0.003 CAA -1.78 -1.85 -2.23 -1.66 -1.5 T-Value -3.92 -4.18 -4.92 -2.72 -2.36 P-Value 0.000 0.000 0.000 0.007 0.019 L.A 0.17 0.284 0.303 0.278 T-Value 3.42 4.56 4.76 4.03 P-Value 0.001 0.000 0.000 0.000 F&E -0.197 -0.16 -0.1 T-Value -2.95 -2.22 -1.05 P-Value 0.004 0.028 0.296 FAA -0.79 -0.98 T-Value -1.37 -1.6 P-Value 0.173 0.111 NMAS 0.1 T-Value 0.95 P-Value 0.341 S 4.67 4.52 4.4 4.32 4.31 4.32 R-Sq 23.31 28.62 32.46 35.2 35.79 36.08 R-Sq(adj) 22.95 27.93 31.47 33.94 34.22 34.19 Mallows C-p 37.5 22.7 12.5 5.8 5.9 7 As shown in T nificant level of 95 percent (?=0.05) and the regression equations are shown in Equation 4.1: has a relatively low coefficient of determination (R 2 =0.352). Any predict o able 4.8, four factors are significant with the sig )352.0(&197.023.2..284.084.1172 2 =??+?= REFCAAALACLP (4.1) From regression equation 4.1, the locking point value is associated with four factors, while it still ion based on an equation with such a low coefficient of determination is likely t have significant error. However, this equation did show some interesting trends about how these factors affect the locking point. 145 With the increase of asphalt content, the locking point tends to decrease. This is likely due to the increased amount of asphalt providing a better lubrication of aggregates during compaction, resulting in quicker compaction. With the increase of L.A abrasion loss value, the locking point tends to increase. This is likely due to the breaking of aggregate associated with high L.A abrasion causing additional densification thus a higher locking point. The aggregate breakdown during compaction is found to have good correla n tion with L.A abrasion value, and will be shown in a later section. With the increase of uncompacted air voids for coarse aggregates (CAA) and F&E content, the locking point tends to decrease. This is not as expected. It is believed that more angular aggregates are more difficult to be compacted and therefore should have higher locking points. This misleading result in equation 4.1 is likely due to the limited range of CAA values (from 46.6 to 48.5%), and high variability or low coefficient of determinatio value in the regression model. 40 45 50 55 60 65 70 C.GVL L.GRN LMS R.GRN TRAP 4.5 5.0 5.5 6.0 6.5 7.0 7.5 Aggregates Asphalt Content, % A ver ag e L o cki n g P o in t s , G y r s N 1 = 42, N 2 = 30 FIGURE 4.2 Average locking point results. 146 The average locking points for different aggregate types and different asphalt 62 re ranite had high locking points, hich is likely due to combination effects of low asphalt contents (average of 5.3 percent ersus 6.3 percent for the other three aggregates, as shown in Table 4.9) and high L.A ates, as shown in Table 4.1). hown e right f Figur , with the increase of asphalt content from 4.5 to 7.5 percent, the average locking point dropped from close look at the effects of asphalt content on locking point is shown in Figure 4.3. There is a strong tendenc at the l g point eases w increase of asphalt content. The regression in Figure 4.3 showed a significant higher coefficien etermination, R 2 , than the Equation 4.1, because the data used in Figure 4.3 are average values thus showing reduced v ability. contents are shown in Figure 4.2. N 1 and N 2 represent the number of samples used for average for each group. Error bars that stand for one standard deviation of results are also shown in the graph. For different aggregates, the average locking point varied from about 54 to gyrations. Limestone and lab granite had high average locking points which were mo than 60 gyrations, while crushed gravel and traprock had low average locking points which were equal or less than 55 gyrations. The difference in average locking point for different aggregates is likely due to the difference in aggregate L.A. abrasion value and asphalt content range. For example, the limestone and lab g w v abrasion value (average of 32.3 percent versus 22.6 percent for the other three aggreg As s in th side o e 4.2 about 61 to 52 gyrations. A y th ockin decr ith an t of d ari 147 y = -3.3044x 89 R 2 = 0. 40 45 50 55 60 65 4.0 .5 5.5 6.0 7.0 .5 8.0 Asp Content, r ag e L o g P o in y r s + 76.9 953 70 4 5.0 6.5 7 halt % A v e cki n t s , G N = 0 FIGURE 4.3 Average locking point values versus asphalt content. Th ric es o d S ures gyr els ar wn in T s 4.9, 4. nd 4.11 ectively expected, a lower values. SMA mixtures designed with 100 gyrations had optimum asphalt contents ranging nt. SMA mixtures designed with 40 gyrations had optimum asphalt contents of 7.2 and 7.5 percent, and VMA of 20.0 and 3 4.2.2.2 Gyratory mix design results e volumet pr rtiope f de gnesi MA ixt m u g the SGC at 100, 65, and 40 sin ation lev e sho able 10 a , resp . As design compaction level resulted in higher optimum asphalt contents and higher VMA from 4.8 to 6.7 percent, and VMA ranging from 15.0 to 19.3 percent. SMA mixtures designed with 65 gyrations had optimum asphalt contents ranging from 6.0 to 7.2 percent, and VMA ranging from 17.4 to 20.0 perce 20.2 percent for lab granite and ruby granite, respectively. 148 TABLE 4.9 100 Gyrations Mix Design Volumetric Properties 1 Type mm Aggregate NMAS, VCA Opt. AC, VMA, VCA mix , % drc, % % % 19 40.7 5.8 16.4 33.0 12.5 41.2 6.4 17.7 37.4 C.GVL 9.5 41.5 6.2 17.3 35.5 19 39.4 4.8 15.0 32.1 12.5 41.0 5.4 16.2 36.4 L.G 9.5 41.4 5.7 16.6 34.9 RN 19 41.7 5.1 15.4 32.6 12.5 41.6 5.5 16.6 36.9 LMS 9.5 41.1 5.3 16.1 34.9 19 41.1 6.2 17.5 37.6 12.5 42.1 6.0 17.1 38.9 R.GRN 9.5 42.0 6.7 18.9 37.0 19 42.4 6.7 19.3 39.5 12.5 42.1 6.1 17.9 40.5 TRAP 9.5 42.1 6.5 18.7 37.7 1. Volumetric properties are based on 4.0 percent air voids by AASHTO T166 method. TABLE 4.10 65 Gyrations Mix Design Volumetric Properties 1 Aggregate Type NMAS, mm VCA drc, % Opt. AC, % VMA, % VCA mix , % 19 40.7 6.6 18.0 34.3 12.5 41.2 7.1 19.1 38.5 C.GVL 9.5 41.5 6.5 18.0 35.9 19 39.4 6.0 17.4 34.0 12.5 41.0 6.5 18.6 38.2 L.GRN 9.5 41.4 6.6 18.5 36.4 19 41.7 6.0 17.6 34.4 12.5 41.6 6.5 18.7 38.5 LMS 9.5 41.1 6.1 17.9 36.2 19 41.1 6.6 18.3 38.0 12.5 42.1 6.7 18.6 40.1 R.GRN 9.5 42.0 7.2 19.8 37.7 19 42.4 7.0 20.0 40.0 12.5 42.1 6.5 18.9 40.9 TRAP 9.5 42.1 7.0 20.0 38.7 1. Volumetric properties are based on 4.0 percent air voids by AASHTO T166 method. 149 TABLE 4.11 40 Gyrations Mix Design Volumetric Properties 1 Type mm Aggregate NMAS, VCA Opt. AC, VMA, VCA , drc, % % % mix % L. GRN 12.5 41.0 7.2 20.0 39.2 R. GRN 12.5 42.1 7.5 20.2 41.3 1. Volumetric properties are based on 4.0 percent air voids by AASHTO T166 method. As shown in Table 4.9, all mixtures using lab granite and limestone can not meet the volumetric property requirements for SMA when designed with 100 gyrations. The discussion for this result is shown in the next section. 4.2.3 Effects of Compaction on Volumetric Properties The volumetric properties of SMA mixture are critical to ensure its structural properties and durability. The requirement of minimum asphalt content and minimum VMA is to ensure the durability of SMA mixtures. The requirement of VCA ratio is to ensure the stone-on-stone contact and eliminate possible unstable mixtures. 4 4.5 5 5.5 6 6.5 7 7.5 8 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 C.GV L L.GRN LMS R. GRN TRA P O p t i m u m As p h al t C o n t en t , % 50-blow Marshall 100 Gyrs 65 Gyrs 40 Gyrs FIGURE 4.4 Comparison of optimum asphalt content. 150 The side by side comparison of different compaction efforts on optimum asphalt content is shown in Figure 4.4. As shown in the Figure, going from 100 to 65 gyrations resulted in an average 0.7 percent increase in optimum asphalt content. For individual comparison the increase ranged from 0.3 percent to 1.2 percent. Going from 65 to 40 gyrations for lab and ruby granite 12.5 mm NMAS mixtures resulted in 0.7 and 0.8 percent increase in optimum asphalt content, respectively. SMA mixtures designed with 65 gyrations had an average of 0.2 percent higher optimum asphalt content than those designed with Marshall compaction. Higher asphalt content is believed to provide more durability for SMA mixtures as long as the mixtures maintain stability and rutting resistance. he tone . The t here was no space t l not Also shown in Figure 4.4, all SMA mixtures designed with 65 gyrations met t general requirement for minimum asphalt content of 6.0 percent. However, only 8 of 15 (53 percent) mixtures designed with 100 gyrations met this requirement. Even with L.A abrasion values less than 30, the three mixtures using limestone designed with 100 gyrations all failed this minimum asphalt content requirement. For example, limes with 19 mm NMAS by 100 gyrations had an optimum asphalt content of 5.1 percent gradation used for this mixture was N gradation, which is near the lower limit of the gradation band in SMA design guides (N gradation with 19 mm NMAS has 20 percen passing the 4.75 mm sieve, and 8 percent passing the 0.075 mm sieve). So t o adjust the gradation in order to increase the optimum asphalt content, which means if 100 gyrations is used, SMA mixtures designed with limestone aggregate wil meet the design requirements. For the same aggregate and same gradation, when the 151 compaction level dropped to 65 gyrations, the optimum asphalt content increased to 6.0 percent and was able to meet the minimum asphalt content requirement. From Figure 4.4, one can also observe that ruby granite and traprock were less sensitive to compaction level than the other three aggregates. For ruby granite and traprock, going from 100 to 65 gyrations resulted in an average 0.5 percent increase of optimum asphalt content, while the other three aggregates had an average 0.9 percent increase. This may be explained by the fact that these two aggregates had lower L.A. abrasion values, and used finer gradations than the other three aggregates. For these two aggregates, the additional 35 gyrations would be expected to break less aggregate, therefore, resulting in less change in optimum asphalt content. 14 15 21 19 9.5 19 9.5 19 19 12.5 12.5 9.5 16 18 19 20 17 12.5 12.5 12.5 9.5 9.5 19 C. N R. GRN TRA PGV L L.GR LMS V a r s l A g g e g a t e , % 50-blow Marshall 100 Gyrs 65 Gyrs 40 Gyrs o ids in M i ne r FIGURE 4.5 Comp on of VM arious com n levels. parison of differ paction efforts on VMA value is shown in Figure 4. MA com ison result ed very similar observations as optim phalt c t results. Going from 65 gyrations d in an average aris A for v pactio The side by side com ent com 5. V par s show um as onten 100 to resulte 152 1.5 percent increase in VMA. All SMA mixtures designed with 65 gyrations met the minimum VMA requirement of 17 percent for SMA mixture, while only 8 of 15 (53 percent) of mixtures designed with 100 gyrations met this requirement. All mixtures designed with the Marshall method also met this minimum VMA requirement, and had an average 0.5 percent less than those designed with 65 gyrations. 1.00 r (V C A C % 0.70 0.75 0.80 0.85 0.90 0.95 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 mix / V A dr c ), 50-blow Marshall 100 Gyrs 65 Gyrs 40 Gyrs a ti o C.GV L L.GRN LMS R. GRN TRA P V C A lso result FIGURE 4.6 Comparison of VCA ratio for various compaction levels. The side by side comparison of different compaction efforts on VCA ratio is shown in Figure 4.6. As shown in Figure 4.6, all designed SMA mixtures have the VCA ratio less than one, which indicates that stone-on-stone contact existed for all designed mixtures. The VCA ratios for Marshall compaction are similar to those compacted with 65 gyrations. The VCA ratio decreases with the increase of gyration level, which indicates the coarse aggregate skeleton gets tighter with the additional compaction. However, additional aggregate breakdown with the high compaction level will a 153 in the a the original gradation will be smaller than the actual VCA ratio. verse ranking of durabil esults 4.2.4 Draindown Test Results A lower compaction level tends to result in higher asphalt content; therefore draindown may be a concern for a low compaction level. The asphalt contents selected for draindown test were the highest optimum asphalt contents resulting from all compaction efforts. Two replicate tests were conducted for each mixture. During testing of the different mixes, it was noted that aggregate particles were able to fall through the wire mesh onto the plate when using the standard ? inch (6.3 mm) ate particles that fell through the openings of the basket , and ctual aggregate gradation getting finer and therefore the calculated VCA ratio based on In summary, in consideration of durability, the order from the best to the worst is 40 gyrations, 65 gyrations, 50 blow Marshall, and 100 gyrations. In consideration of rutting resistance, the order from the best to the worst may be the re ity. The present recommended gyration level is 100. If the performance test r support a lower level, this will be lowered to improve durability without affecting rutting potential. mesh basket. Some fine aggreg were counted as draindown following the test procedure AASHTO T305 (25). This phenomenon became more common when smaller NMAS mixtures were tested resulted in relatively high draindown results for 9.5 mm NMAS mixtures. The draindown test results are summarized in Table 4.12. 154 TABL * Aggregate particles were observed in the plate, and counted as draindown. As shown in Table 4.12, the draindown results ranged from 0.02 to 0.28 percent. The results showed high variability because some aggregate particles were included as draindown. Even without the correction, all draindown results met the maximum limit of 0.3 percent as required in the SMA design specifications (44). Therefore, in terms of the draindown test requirement, all compaction levels including 40 and 65 gyrations are satisfac , voids were measured in two commonly used methods: the SSD method following E 4.12 Draindown Test Results Summary Aggregate Type NMAS, mm Asphalt Content, % Percent Draindown, % St. Dev of Draindown, % 19 6.6 0.028 0.017 12.5 7.1 0.126* 0.023 C.GVL 9.5 6.5 0.170* 0.106 19 6.0 0.029 0.041 12.5 7.2 0.047 0.001 L.GRN 9.5 6.6 0.283* 0.014 19 6.0 0.073 0.057 12.5 6.5 0.084 0.039 LMS 9.5 6.1 0.212* 0.061 19 6.6 0.167* 0.135 12.5 7.5 0.118* 0.022 R.GRN 9.5 7.2 0.059 0.005 19 7.0 0.020 0.006 12.5 6.5 0.020 0.005 TRAP 9.5 7.0 0.060 0.028 tory for all the mixtures evaluated in this study. The use of modified asphalt binder and 0.3 percent cellulose fiber effectively prevents the draindown problem. If fiber was not used, the draindown amount would almost certainly be a problem. 4.3 AIR VOID CONTENT MEASUREMENT The amount of air voids is one of the most important mixture properties evaluated in this study. Accurate determination of air void content is critical to ensure proper mix design and to provide a pavement with good performance and durability. For this research, air 155 AASHTO T166 (25) and the vacuum seal method following ASTM D 6752 (47). The comparison of the test results by using these two test methods and the effects on mix design volumetric properties were analyzed and evaluated. In addition, the air void results of whole samples and cored-and-sawn triaxial samples are also included in this section. An error with the software for the CoreLok method was identified and a further study on direct calculation based on the original concept of vacuum sealing method for results was also conducted to provide a better method of making correction factors. Some guidance was provided to let the user know when each method should be used. Four different air voids were used in this study relating to the vacuum sealing method: CoreLok D 1) CoreLok air voids, which is calculated by the CoreGravity TM program created by IntroTEK, Inc (110). This air voids reflects the embedded correction by the program, and it is generally known as the air voids by the CoreLok method. 2) Corrected CoreLok air voids, which is CoreLok air voids with additional correction factor. This correction factor is determined by the difference between the SSD and CoreLok air voids using a sample with low air voids. 3) Uncorrected vacuum sealing air voids, which is a calculation based on the concept of vacuum sealing test method without any corrections. 4) Corrected vacuum sealing air voids, which is a vacuum sealing air voids with a correction factor. The correction factor is determined by the difference between the SS and uncorrected vacuum sealing air voids using a sample with low air voids and similar surface condition. These air voids will be discussed in detail later and used throughout this section. 156 4.3.1 Concepts of Air Voids As discussed in the literature review and the test procedure, air voids is defined as the total volume of the small pockets of air between the coated aggregate particles throughout a compacted paving mixture, expressed as a percent of the bulk volume of th compacted paving mixture. The value of air voids depends o e n the bulk specific gravity G mb an A. ith coarse and fine gradations. d the measured theoretical maximum specific gravity G mm . Hence accurate measurement of G mb and G mm is very important in the calculation of air voids in a mixture. This report will provide a detailed discussion concerning measurement of G mb . There are several different methods to measure the bulk specific gravity G mb . The difference between various methods is primarily due to the different methods of measuring sample volume since the sample mass can be measured very accurately. Figure 4.7 (49) illustrates volumes and air voids that are associated with compacted HM Each of the diagrams within Figure 4.7 are divided into halves with each half representing the volumes and air voids of mixes w a) Dimensional Volume b) Apparent Volume c) Bulk Volume FIGURE 4.7 Volumes associated with compacted HMA (49). 157 The dark black line in Figure 4.7a shows the volume that is associated with the dimensional procedure. This volume includes any surface irregularities on the outside of the sample and thus overestimates the internal air void content. Figure 4.7b illustrates apparent volume of compacted HMA samples. This volume can be calculated by th difference of dry weight and submerged weight of sample based on Archimedes? Principle. This calculated volume does not include any of the surface irregularities on the sample or the air voids that are interconnected to the surface. Water that infiltrates the sample through the interconnected surface voids is not considered a portion of the samp volume. Therefore, the apparent volume underestimates the sample?s true internal voids. This problem is more prevalent with mixes having coarser gradations, as there are more voids interconnected to the surface of the sample. Figure 4.7c illustrates the bulk volume determined from the SSD method. The difference between the bulk and apparent volume is that the bulk volume includes the voids that are interconnected to the surface. T accomplished by using the SSD weight instead of dry weight when using the the e le his is Archim r t to determine the bulk volume of the fficult to differentiate between mixture air void urfa um sealing method, because much of the surface texture in some HMA samples, especially in SMA samples, lies uter geometrical area. The texture does not increase edes? Principle to calculate the sample volumes. Therefore, the volume of wate retained in the sample at the SSD condition is included as internal voids. The bulk volume lies between the dimensional and apparent volumes and is desired for HMA volumetric calculations. The CoreLok and SSD methods both attemp HMA sample. A high amount of surface texture makes it more di s and s ce texture for the vacu within the pore space rather than the o 158 the inte and A mixture during the SSD test due to large voids with a higher percentage of coarse aggregate particles. Water draining from the large interco or lower calculated sam void content. However, the most accurate, repeatable, and feasible method to measure the lower calculated sample volume and higher calculated bulk specific gravity. Therefore, the SSD m by mixture type. For the CoreLok test method, the bridging effect of high surface texture volume, therefore a proper correction factor is needed to reduce this measured volume bulk specific gravity. rnal air voids, however, it may result in a higher measured air voids. The rate extent of water penetration into SMA mixtures during the SSD test depend not only on the total air voids content but also on the actual void size inside the mixture. It is easier for water to enter into and drain out from SM nnected voids within SMA mixtures leads to a lower SSD weight, ple volume. Consequently it becomes difficult to define the volumetric properties of compacted SMA mixtures if the voids are high. If the voids are low one can get an accurate measure of air voids by weighing the sample in air and water. Since it is not possible to know the true sample volume, one can never accurately know the true air true volume should be determined and adopted. The volumetric requirements should be specified based on this most accurately measured volume. For the SSD test procedure, the water draining out of compacted samples with high air voids could result in a lower SSD weight of the sample, therefore resulting in a ethod may only be accurate for a certain air void range, which is also affected (the plastic bag can?t tightly conform to rough sample surface due to bag stiffness and not unlimited vacuum pressure) may result in higher enclosed volume than the true sample and increase the calculated 159 4.3.2 Comparison of Two Test Methods The com SM tu les includ arshall samples, SGC sign sam les, APA samples, triaxi g a r coring) samples, w by ing t hod SS od he seal (CoreLok) method. However, the SMA les d i analy s consist of SGC mix design samp the data is because of he wide distribution of air void content results, identical sample size, and all of these samples had permeability information, which is presented in section 4.4. The permeability information can be used as an indication of error potential for the SSD method. The air void results for SMA mixtures are shown in Appendix Table A1 and in Figure 4.8. As shown in Figure 4.8, the air void content measured had a range from 1.1 to 8.0 percent by the SSD method, and a range from 1.6 to 10.9 percent by the CoreLok method. The Corelok air void results were calculated using the computer program CoreGravity TM created by IntroTEK, Inc. It is noteworthy that correction factors are applied in the program based on bag size and the ratio of sample weight and bag weight (110). For all the test results, the difference of these two methods ranged from 0.4 to 4.6 percent. The CoreLok method provided an average of 1.2 percent higher air voids than the SSD method. This is as expected, because the SSD method is likely to underestimate the air voids due to the water draining out problem while the CoreLok method is likely to overestimate the air voids due to high surface textures for the SMA samples. A true air void content is likely between the results from these two methods. For samples with low air voids, the error potential for the SSD method is limited and should provide a measurement of air voids close to the true value. air void content of pacted A mix re samp ing the M mix de p al testin (before nd afte ere determined us wo met s, the D meth and t vacuum samp include n this si les only. The reason for using this part of t 160 A pa ng the CoreLo E 4.13 Paired T-Test for CoreLok and SSD air voids ired t-test was employed to compare air void content results by usi k and SSD methods. Statistical results are shown in Table 4.13. There is a significant difference at the 95 percent confidence level between these two methods as indicated by a P-value equal to 0.000. Keep in mind that these 372 samples are lab compacted SMA samples with variable air voids. TABL N Mean StDev SE Mean CoreLok 372 5.391 1.555 0.0806 SSD 372 4.200 1.199 0.0622 Difference 372 1.190 0.559 0.0290 95% CI for mean difference: (1.133, 1.247) T-Test of mean difference = 0 (vs not = 0): T-Value = 41.08, P-Value = 0.000 A linear regression between the results of the two methods along with the R 2 value is shown in Figure 4.8. It is noticed that all the data points are above the equality line, which indicates the CoreLok air void content is higher than the air void content by the SSD method for all the tested samples. y = 1.2349x + 0.2017 2 2.0 3.0 5.0 6.0 8.0 9.0 11.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 r e l o k r V o d s , % R = 0.9017 Equality line y = x 1.0 4.0 7.0 10.0 SSD Air Voids, % Co Ai i FIGURE 4.8 Comparison of the CoreLok and SSD air voids. 161 Also, the slope of the trendline is higher than one, which indicates that the difference between these two methods increases with an increase in air void content. This is likel t r voids. This water draining problem is difficult to correct and resulted in higher test variability with the increase of air voids as shown in Figure 4.8. A ratio of the air void contents by the CoreLok and SSD method is used for representing the difference between the two methods. A general linear model (GLM) analysis on this ratio was conducted to evaluate the effect of the main factors, such as aggregate type, asphalt content, NMAS and air void content by SSD method. GLM is used because it can be used to analyze unbalanced test results. Statistical results are shown in Table 4.14. TABLE 4.14 GLM for Influencing Factors on VTM Ratio Source DF Seq SS Adj SS Adj MS F P Value y due to the increasing error potential for the SSD method with an increase of air voids. For SMA samples with high air void contents, the water was observed to drain ou of the samples quickly and a significant amount of water was retained on the scale after the SSD weight was measured. The SSD weight depends on how quick a sample can be patted to surface dry and be weighed, and it usually results in decreased calculated air void contents when compared to the actual ai AGG. Type 4 0.112 0.025 0.006 1.05 0.384 NMAS 2 1.775 0.482 0.241 39.82 0.000 Asphalt Content 6 0.140 0.039 0.006 1.07 0.387 VTM by SSD 253 2.390 1.56 0.005 2.390 0.009 Error 106 0.641 0.641 0.006 Total 371 5.059 162 From Table 4.14, the significant factors were NMAS and air void content by the SSD method based on a level of significance of 95 percent. The fact that NMAS was a significant factor is likely due to the different amount of air trapped in the surface textur of compacted samples, and the larger NMAS likely results in more connected internal air voids therefore resulting in higher permeability and higher error potential for the SSD method. The surface texture for three samples with different NMAS and similar air vo are shown in Figure 4.9. The average surface texture depths for these 19, 12.5, and 9.5 mm NMAS samples are 1.80, 1.55, and 1.36 mm following the sand patch method ASTM E965 (3), respectively. The fact that air void content was a significant factor indicated that water draining out of samples during the SSD test became a e ids problem for the SMA samples with high air voids. FIGURE 4.9 Surface textures for different NMAS mixtures. Since the NMAS is a significant factor affecting the ratio of air void cont the CoreLok and SSD methods, the air void content data was separated for three subsets based on the three NMAS. Linear regressions were conducted for these subsets of data respectively. The regressions along with the respective R-squares are shown in Figure ents by 4.10. 163 19 m m N M A S 12.5 mm NM A S 9.5 mm NM AS Linear (19 mm NM AS) Linear 12.0 19 mm NMAS: y = 1.3666x + 0.0346 R 2 = 0.8971 12.5 mm NMAS: y = 1.2588x + 0.064 R 2 = 0.9551 9.5 mm NMAS: y = 1.148x + 0.228 R 2 = 0.9713 0.0 2.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Co lo k Air V o id s % 4.0 6.0 8.0 10.0 7.0 8.0 9.0 r e , SSD Air Voids, % (12.5 mm NM AS) Linear (9.5 mm NM AS) FIGURE 4.10 Relationships between the CoreLok and SSD air voids for three NMAS. One can observe from Figure 4.10 that slopes of these three trendlines are larger than one, and the magnitude of these three slopes are in order of NMAS. As expected, with the increase of NMAS, the r all atio of the slope of the two air void methods increas more Lok es. This is likely explained in that with the increase of NMAS, the surface macro- texture depth tends to get larger. In the meanwhile, the larger NMAS results in a greater chance of internal air voids becoming connected to the surface at high air voids, and bridging of surface texture by CoreLok bag. Therefore, the SSD method tends to underestimate the air voids because water drains out of samples quicker for larger NMAS mixtures at a high air void level when determining the SSD weight, and the Core method tends to overestimate the air voids because of more bridging for larger NMAS mixtures with high coarse aggregate content. As a combined result, there is a bigger 164 difference between the CoreLok and SSD air voids for larger NMAS mixtures. For samples with high air voids, the error in the SSD method due to the water draining out problem is hard to correct, however, an appropriate correction factor can be determined for the CoreLok method if one assumes the surface texture for the mixture with various ifferen etw C an ethod versus air void content by SSD for thr NMA ow gu . It at at the lowest air voids level (1.1 percent by the SSD od) the th ence between the two m s was s hig .5 p It ved t at at low air void content, when m eable, these two methods should give sim SSD method is believed to be more reliable. air voids is similar. The d ce b een the oreLok d SSD m method ee S is sh n in Fi re 4.11 is noticeable th meth of all data, e differ ethod till as h as 0 ercent. is belie h ixtures are imperm ilar results. If not, the 19 mm NMAS 12.5 mm NMAS 9.5 mm NMA S Expon. (19 mm NMAS) Expon. (12.5 mm NMAS) Expon. (9.5 mm NMAS) 19 mm NMAS: y = 0.5678e 0.2246x R 2 = 0.399 12.5 mm NMAS: y = 0.4619e 0.2004x R 2 = 0.4539 9.5 mm NMAS: y = 0.3929e 0.17x R 2 = 0.3694 0.0 0.5 1.0 1.5 2.0 2.5 3.5 5.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 SSD Air Voids, % A i r V o i d s d i ffe r e n c e (C o D ), % 3.0 4.0 4.5 o r e l k -S S FIGURE 4.11 Air voids difference between the CoreLok and SSD method versus the SSD air voids. 165 To estimate the difference between the SSD and CoreLok method, a comparison was made between the two methods by using a steel cylinder 100 mm (4 in) in diameter ? 63 mm (2.5 in) in height. The steel cylinder had a smooth surface and no internal voids connected to the surface. The G mb result by the SSD method was considered to be very accurate since there was no voids in the cylinder. This resulted in G mb and G mm being equal for the cylinder. The calculated air void content by the CoreLok method was 0.6 percent based on the average of two replicates. This result indicates that there is not only an error for the CoreLok method for samples with rough surface textures (lab compacted SMA mixtures), but also for samples with very smooth surface and no internal voids (steel sample). The CoreLok air voids calculated by CoreGravity TM are not accurate for all types of tested sam standardized by the manufacture bedded correlation factors (110), t s considered excessive by the CoreGravity TM program and a new test was run after drying the sample. Based on the trendlines and regression equations shown in Figure 4.11, the average difference between the two methods for three NMAS SMA mixtures is approximately 0.5 percent when the air void content by the SSD method approaches zero. ples. Since bag stiffness, vacuum pressure and bag sizes are r and considered into the em he user should use caution when a new type of material is tested. Another error may be introduced into the procedure if samples are not tested immediately or within a reasonable time frame. After vacuum sealing, the bags may relax and allow air to leak into the bags. Samples for this study were tested immediately after sealing so that this potential for error was minimized. Also, the sample weight before vacuum sealing and after submerging was compared. A weight gain more than 5 grams due to water penetration wa 166 Therefore, an additional correction factor of 0.5 percent was suggested for the CoreLok ir er program CoreGravity TM seem insufficient for laboratory compacted SMA samples. A study on co (CoreLok) air voids is necessary to provide a set of better correlation factors. The results and discussion are shown in section 4.3.5. The CoreLok device was designed to aid in density determination of asphalt cores or laboratory specimens that are porous (water absorption during the SSD test is greater than 2 percent). However, even though all the samples used in this study had water absorption less than 2 percent, these two methods still showed a significant difference. This indicated that the allowable water absorption for using the SSD method should be a level less than 2 percent. A GLM analysis was conducted to determine the most device used in this study since it was assumed the system error for all the NMAS and air voids levels was similar. The air voids with this additional correction factor are defined as corrected CoreLok air voids in this study. From Figure 4.11, if the additional correction factor is applied to the CoreLok a voids, a difference between the two methods of more than one percent occurs for 19 mm NMAS samples with air voids greater than 4.3 percent, for 12.5 mm NMAS samples with air voids greater than 5.9 percent, and for 9.5 mm NMAS samples with air voids greater than 7.9 percent by the SSD method. The threshold air voids increase with the decrease of NMAS is likely due to the decreased permeability and error potential for the SSD method with lower NMAS mixture at the same air void level. The difference between the SSD and CoreLok methods for SMA samples with low air voids indicate that the correlation factors embedded into the comput mparing uncorrected air voids by vacuum seal method and program corrected 167 significant influencing factors on water absorption during the SSD test. As shown in Table 4.15, the NMAS and air void conten factors. This is as expected because the water rption during the SSD test depends not only on the total air voids, but also on the actual void size inside the mixture. As shown in Figure 4.11, the m ure w high MA gre water rptio cause the larger vo e. TABLE 4.15 GLM for In encin ctors on Water Absorption urc Seq SS SS MS F lue t are the two most significant influencing abso ixt ith er N S has ater abso n be of id siz flu g Fa So e DF Adj Adj P Va AGG. Type 75 19 .57 2 4 1.181 0.0 0.0 2 0.04 NMAS 16 58 .99 1 2 0.353 0.1 0.0 7 0.00 Asphalt C 57 09 .30 4 ontent 6 2.974 0.0 0.0 1 0.26 VTM by SSD 2 05 35 .79 0 53 8.805 8.8 0.0 4 0.00 Error 106 70 07 0.770 0.7 0.0 Tota 14.083 l 371 19 mm NMAS: y = 0.0753e 0.3797x R 2 = 0.7299 12.5 mm NMAS: y = 0.0678e 0.3668x R 2 = 0.7911 9.5 mm NMAS: y = 0.0483e 0.4148x R 024 0.0 0.2 0.6 0.8 1.2 1.4 1.8 3 4.05.0 .07 8.09.0 SSD A ds, % W a t e r A b s o r p ti on dur i ng S S D Te s t , % 2 = 0.8 0.4 1.0 1.6 0. 1.02.0 .0 6 .0 ir Voi 19 mm NMAS 12.5 M A S mm N 9.5 mm NMAS Expon. (19 M AS)mm N Ex m Npon. (12.5 m M AS) Expon. (9.5 mm S) NMA FIGURE 4.12 Relationships between absorbed water and air void content. 168 Figure 4.12 demonstrates the absorbed water during the SSD test procedure versus the air void content by the SSD method. As expected, the water absorption increased with an increase of air voids. 19 mm: y = 0.7721e 1.5475x R 2 = 0.4918 12.5 mm: y = 0.6945e 1.2127x R = 0.4603 2 9.5 mm: y = 0.5664e 1.0811x R 2 = 0.368 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 Water Absorption, % A i r V o id s D i f f e r e n ce ( C o r eL o k - % 4.50 5.00 5.50 S S D ) , 19 mm 12.5 mm 9.5 mm Expon. (19 mm) Expon. (12.5 mm) Expon. (9.5 mm) FIGURE 4.13 The difference between the CoreLok and SSD air voids versus absorbed water content. Figure 4.13 combines some information from Figure 4.11 and 4.12. As shown in F is set as 1 percent (difference between the CoreLok and SSD air voids is 1.5 percent), the threshold water absorption values are 0.4, n igure 4.13, if the tolerance between Corrected CoreLok air voids and the SSD air voids 0.6 and 0.9 for 19 mm, 12.5 mm and 9.5 mm NMAS mixtures, respectively. This concept is important because the absorbed water ca quickly be determined by the SSD test, and used to indicate if a substantial error for the SSD method is occurring. 169 4.3.3 Effect on Mix Design Volumetric Properties The measured air void content is the primary property used in the mix design process to select the optimum asphalt content. The difference in mix design volumetric properties by these two methods is summarized in Table 4.16. Table 4.16 shows that there was as much as 0.8 percent difference in optimum asphalt content between the SSD and CoreLok method when the design air voids was set at the s e s of determining G mb . This amount of difference is significant because many State agents limit asphalt e California DOT requires the asphalt content mb ame level of 4 percent. The average difference in optimum asphalt content for th two methods was 0.45 percent. Figure 4.14 shows graphically the difference in optimum asphalt content by aggregate type and NMAS for each of the two method content to a range of 1 percent, for exampl within ? 0.5 percent of the target value. This study indicates that most of the asphalt content tolerance may be used just by the difference in G measurement. 4.0 4.5 5.0 6.0 6.5 7.5 8.0 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 19 12 . 5 9. 5 5.5 7.0 C.GV L L.GRN LMS R. GRN TRA P C.GV L L.GRN LMS R. GRN TRA P O p t u m s pha l C ont n t , i m A t e % SSD Corelok 100 Gyrations 65 Gyrations on optimum asphalt content. FIGURE 4.14 Effect of CoreLok and SSD methods 170 TABL 1. Optimum asphalt content and VMA are based on 4.0 percent air voids. E 4.16 Mix Design Volumetric Properties 1 Summary AASHTO T166 CoreLok Difference Comp. AGG. NMAS Grad. Level Opt. AC VMA Opt. AC VMA Opt. AC VMA 19 N 5.8 16.4 6.6 18.0 0.8 1.6 12.5 N 6.4 .7 6.8 18.5 0.4 0.8 17 C.GVL 9.5 N 6.2 17.3 6.6 18.0 0.4 0.7 19 N 4.8 15.0 5.6 16.6 0.8 1.6 12.5 N 5.4 16.2 5.9 17.3 0.5 1.1 L.GRN 9.5 N 5.7 16.6 5.9 17.2 0.2 0.6 19 N 5.1 15.4 5.5 16.6 0.4 1.2 12.5 N 5.5 16.6 6.0 17.7 0.5 1.1 LMS 9.5 N 5.3 16.1 5.7 16.8 0.4 0.7 19 F 6.2 17.5 6.7 18.6 0.5 1.1 12.5 F 6.0 17.1 6.5 18.3 0.5 1.2 R. GRN 9.5 F 6.7 18.9 7.0 19.5 0.3 0.6 19 F 6.7 19.3 7.2 20.6 0.5 1.3 12.5 F 6.1 17.9 6.8 19.5 0.7 1.6 100 TRAP 9.5 F 6.5 18.7 7.0 19.9 0.5 1 Gyrs .2 19 N 6.6 18.0 7.2 19.2 0.6 1.2 12.5 N 7.1 19.1 7.5 19.9 0.4 0.8 C.GVL 9.5 N 6.5 18.0 6.9 18.7 0.4 0.7 19 N 6.0 17.4 6.5 18.6 0.5 1.2 12.5 N 6.5 18.6 6.9 19.2 0.4 0.6 L.GRN 9.5 N 6.6 18.5 6.8 18.9 0.2 0.4 19 N 6.0 17.6 6.4 18.5 0.4 0.9 12.5 N 6.5 18.7 6.8 19.3 0.3 0.6 LMS 9.5 N 6.1 17.9 6.5 18.6 0.4 0.7 19 F 6.6 18.3 7.1 19.4 0.5 1.1 12.5 F 6.7 18.6 7.0 19.4 0.3 0.8 R. GRN 9.5 F 7.2 19.8 7.5 20.4 0.3 0.6 19 F 7.0 20.0 7.5 21.1 0.5 1.1 12.5 F 6.5 18.9 7.0 19.8 0.5 0.9 65 Gyrs TRAP 9.5 F 7.0 20.0 7.4 20.7 0.4 0.7 171 Voids in mineral aggregate (VMA) is an important volumetric property specified in most mix design procedures. This property is defined as ?the volume of intergranular void space between the aggregate particles of a compacted paving mixture that includes the air voids and volume of the asphalt not absorbed into the aggregates? (104). When the target air voids is chosen, VMA is an indication of the void space to be partially filled with asphalt binder. Therefore, a minimum VMA is needed to ensure the long-term durability and an upper limit is also desirable to prevent an unstable asphalt mixture. 14.0 17.0 8.0 19.0 20.0 21.0 22.0 1 15.0 16.0 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 19 12. 5 9. 5 V o i d C.GVL L.GRN LMS R. GRN TRAP C.GVL L.GRN LMS R. GRN TRAP 100 Gyrations 65 Gyrations s i n M i n e ral A g g r eg at e s , % SSD Corelok As shown in Figure 4.15, there is as much as 1.6 percent difference in calculated VMA depending on whether SSD or CoreLok method is used to determine G of the percent. In this study most of the VMA tolerance may be nullified just by the difference between G mb measurements. Therefore, in determining VMA values, specifying agencies FIGURE 4.15 Effect of CoreLok and SSD methods on voids in mineral aggregate. mb specimens. This difference is critical because some agencies limit VMA to a range of 2 172 m method should be followed throughou n and quality control in construction. Howev alysis. 4.3.4 Triaxial Test Sample Preparation As expected, the cored samples for the triaxial test had different air void contents from the whole samples compacted by the SGC because of the significant difference in air voids in the middle of the compacted sample and that around the top, bottom and sides. The target air void content for the cored samples was 4.0?0.5 percent by the SSD method. However, the target air void content for the whole samples had to be higher to provide 4.0 percent air voids in the cored sample, therefore a trial-and-error method was used. Based on some literature (86, 108), an initial air void level for the whole sample was ith the feedback of air voids of sawn-and-cored samples, the target air voids for the whole sample were adjusted. In order to get enough cored s s with oids in r about 5 rcent more sa pacted. id cont to 4.0 percent conduct the triaxial tests. A total of 277 samples were compacted and prepared for triaxial tests. ust clearly define the method that is to be used for determining G mb . The same test t the mix desig er, one test method may be more appropriate than the other for different types of mix or with different air void levels. The correction for CoreLok method will help to solve some of these problems. This study will recommend when a specific test method should be used based on the air voids, water absorption, and permeability data an As shown in Figures 4.14 and 4.15, the effects of the two test methods on these volumetric properties are greater for larger NMAS mixtures. This is consistent with the greater difference in air void measurements for larger NMAS mixtures. targeted at around 5.5 percent. W ample air v ange of 4.0?0.5 percent, generally 0 pe mples were com The samples with the closest air vo ents and in good physical shape were selected to 173 The air void content for all whole samples and cored samples were measured by both SSD and CoreLok method. The comparison of these two test methods for samples is demonstrated in Figure 4.16. the whole 19 m m N M A S 12.5 mm NMAS 9.5 mm NM AS Linear (19 mm NM AS) Linear (12.5 mm NM AS) Linear (9.5 mm NM AS) 19mm NMAS: y = 1.1664x - 0.3984 R 2 = 0.6987 12.5mm NMAS: y = 0.9177x + 0.5994 R 2 = 0.6842 4.0 5.0 6.0 e lo k r V o id f o r W o l e S a p l e , % 9.5 mm NMAS y = 1.0099x - 0.0766 R 2 = 0.8102 3.0 7.0 8.0 SSD Air Voids for Whole Sample , % r C o r A i s h m 2.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 C o r ect ed As shown in Figure 4.16, after applying the 0.5 percent correction factor from the pervious discussion for CoreLok air voids, the difference between these two methods for 12.5 mm and 9.5 mm NMAS SMA samples and most of 19 mm NMAS SMA samples were less than 1 percent. This verifies that the 0.5% correction factor, which is developed using standard SGC samples of 115 mm in height, is applicable to the taller samples with about 170 mm in height. This also indicates that the correction factor mainly depends on the surface texture, and not the amount of air voids in the mixture. For 19 mm SMA samples, a higher correction factor is needed to bring all the data points within the 1 percent difference. FIGURE 4.16 Corrected CoreLok and SSD Air Voids for Whole Samples. 174 The comparison of the CoreLok and SSD method for the core samples is demonstrated in Figure 4.17. The air void results for the core samples showed a different trend fr t of differen om the whole samples prior to coring. This again verifies the significant effec surface condition on the test results of the CoreLok method. The CoreLok results for the core samples were very similar to the results measured by the SSD methods. The ce between these two methods was within 1 percent for all the core samples. 19mm NMAS y = 1.033x - 0.3227 R 2 = 0.8547 12.5mm NMAS y = 0.9886x - 0.2454 6.0 p l e s , R 2 = 0.8834 5.0S a m 9.5mm NMAS y = 1.0646x - 0.5865 7.0 reL i r V s f 19 mm NMAS 12.5 mm NM A S 9.5 mm NM AS % 4.0 o r C o r e 3.0 o i d R 2 = 0.8779 1.0 2.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 SSD Air Voids for Core Samples, % C o o k A Linear (19 mm NM AS) Linear (12.5 mm NM AS) Linear (9.5 mm NM AS) FIGURE 4.17 CoreLok and SSD air voids for core samples. TM For samples with 4 percent air voids by the SSD method, the Coregravity program provided the air voids of 3.8, 3.7, and 3.7 percent for 19 mm, 12.5 mm, and 9.5 mm cored SMA mixtures. It is important to say that all lines plotted on top of each other and the SSD and corelok voids were very similar. The cored sample is smooth therefore the effect of surface texture is removed. This indicates that the accuracy of CoreLok method mainly depends on the surface condition regardless of NMAS. The significance 175 of NMAS for lab compacted samples is likely due to difference in surface texture provided by different NMAS. The correlation factor embedded in the CoreGravity TM program (Appendix B) seems to work well on core samples; therefore no additional correction factor was needed for the air voids results calculated by the CoreGravity T program. M 19 m m N M A S 12.5 mm NM A S 9.5 mm NM AS Linear (19 mm NM AS) Linear (12.5 mm NM AS) Linear (9.5 mm NM A S) 19 mm NMAS y = 1.115x - 1.5001 12.5 mm NMAS y = 1.1282x - 1.6938 R 2 = 0.8113 5.0 6.0 f o r C r e Sa p l e s , R 2 = 0.7493 3.0 4.0 7.0 .0 A i r d s o m % 9.5 mm NMAS y = 0.9911x - 1.1295 R 2 = 0.8153 2.0 2.0 3.0 4.0 5.0 6.0 7 SSD Air Voids for Whole Samples , % SSD Vo i FIGURE 4.18 Air voids relationships between the whole and core samples b t SSD method. and cored samples are demonstrated in Figure 4.18. There is approximately a 1 percent distribution of the air voids. These relationships were developed separately based on the y he Based on air void content results of 277 samples, the air void content relationships between whole and cored samples were developed. The relationships between the whole difference between the whole and cored samples because of the inhomogeneous different NMAS. However, the results showed the difference between these three NMAS 176 was not large (Table 4.17). In order to get the cored samples with 4.0 percent air void content (using the SSD method), the whole samples should be compacted to 4.9, 5.0, and 5.2 percent air void content (using the SSD m or the 19 mm, 12.5 mm and 9.5 mm NMA r air voids for the whole sample in order to get the core samples with 4 percent air voids. This ma Regression Equations ethod) f S mixtures, respectively. The larger NMAS SMA mixtures require slightly lowe y be due to the greater error potential when measuring air voids by the SSD method for the whole samples with larger NMAS. The threshold air voids value for the 19 mm NMAS whole samples may be higher than 4.9 percent if the water draining problem is prevented. TABLE 4.17 Targeting Whole Samples VTM 1 for Triaxial Samples NMAS, mm Core VTM V c , % Whole VTM V w , % 19 4.0 V c = 1.1150V w - 1.5001 R =0.7493 4.9 2 12.5 4.0 V c = 1.1282V w - 1.6938 R 2 =0.8113 5.0 9.5 4.0 V c = 0.9911V w - 1.1295 R 2 =0.8153 5.2 1. VTM measured by the SSD method. The difference in air void content between the whole and core samples is likely due to the edge effects and surface friction around the side of the mold during compaction (108). The surface of lab compacted samples is likely to have higher air voids because the loss in freedom for re-orientation for the surface portion of samples and the surface friction during compaction provide resistance to densification. The amount of difference depends on the degree of non-uniformity in the air voids and aggregate structure between the surface and the interior of the samples. In comparison with core samples, the whole samples are less homogeneous in their distribution of air voids. Many 177 s the cor will have nt for the high surface texture. The steel sample will have higher air void h situatio tudies have shown the air voids around the outside of a core are higher than the inside of e (84, 86, 108). 4.3.5 Uncorrected Vacuum Seal Method Calculation From the test results shown in previous sections, the CoreGravity TM program provided similar air voids for some type of samples (cored-and-sawn samples) with the SSD method, however significantly different results were provided for other type of samples (lab compacted SMA sample or steel sample). The lab compacted SMA sample higher air voids by CoreGravity TM because the embedded correction factor in the program is not sufficie s by CoreGravity TM because there is no internal air voids therefore less pat comparing to cored sample for air trapped on the surface to go out during vacuuming, which resulted in some air remaining trapped after vacuum. A CoreLok test on a solid steel sample showed about 0.6 percent air voids. Therefore, a study on how to appropriately apply the correlation factor for the vacuum sealing method is needed because the correlation factor embedded into the Coregravity TM is not applicable for all ns. In the vacuum seal method, a plastic bag tightly conforms to the sides of the sample by the vacuum and prevents water from infiltrating into the sample. The volume of the specimen encapsulated by the bag is considered as the bulk volume of the sample. The equation used to calculate the bulk specific gravity is shown as follows: T mb F EABC A G ???+ = )]([ (4.2) AB ? where: A = initial mass of dry specimen in air, g, 178 B = mass of dry, sealed specimen, g, C = final m E = mass o ass of specimen after removal from sealed bag, g, f sealed specimen underwater, g, and F T n ic gravity of plastic bags in the Coregravity TM program, or part of the air trapped in the bag was deemed as part of the bag. As shown in Appendix varied from 0.142 to 0.843 depending on the test situations. This correlation approach seems reasonable, but depends on an empirical and-sawn samples were within one percent for all tested samples as shown in Figure 4.17, provided only 0.2 to 0.3 percent difference, which indicates this correlation approach ples with high surface i re = apparent specific gravity of plastic sealing materials at 25?C. The automatic calculation of CoreLok test results was provided by the Coregravity TM program. A set of correction factors was embedded into the program by the manufacturer (110) and shown in the Appendix Table B1. The correction factors are given based on three regression equations (110) for three test situations and are functions of ratio of sample mass and bag mass. The three test situations are using different bags, i.e. small bags (10?14 inch), big bags (14.75?18 inch) and double bags. These correlatio factors are used as specific gravity of plastic bags in the Coregravity TM program following Equation 4.2. In other words, the correlation on CoreLok test results was conducted by adjusting the specif Table B1, the specific gravity of bags can be regression relationship. The difference between the CoreLok and SSD results for cored- and for samples with 4 percent air voids by the SSD method, the CoreLok program works very well for cored samples. For lab compacted SMA sam texture, however, this correlation seems insufficient and results n higher air voids than those using the SSD method, as shown in Figure 4.10. For impermeable samples with low air voids, this difference is mainly due to the bridging effect of high surface textu 179 for the CoreLok method. Based on the previous discussion, this difference is abou percent. To clarify the effect of these embedded correlation factors on air voids calculatio the air voids calculated by Equation 4.2 t 0.5 n, without adjusting the specific gravity of the plastic bags needs to be investigated. The specific gravity of the plastic bags was det gravity value is higher than all the specific gravity values used in the Coregravity TM program. This indicates that the specific gravity value used in the program is reduced for correcting the tendency of overestimating the air voids in the vacuum sealing method. Calculations were conducted by using Equation 4.2 with actual bag specific gravity for all tested samples with different sizes, including the mix design samples (150?115 mm), whole samples for triaxial tests (150?170 mm), and cored samples (100?150 mm) for triaxial tests. The uncorrected vacuum sealing air voids based on Equation 4.2 were compared to those from the Coregravity TM program (CoreLok air voids) and are shown in Figure 4.19. ermined as 0.916 following the ASTM D 792 (47). It is noteworthy that this specific 180 y = 0.9743x + 0.928 R 2 = 0.9991 12.00 14.00 0.00 2.00 4.00 6.00 8.00 10.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 U n r e ct e A i r V o CoreLok Air Voids by Program, % co r d i d s , % Mix design samples Whole samples Cored samples Equality Line URE 4.19 eLok air voids by progr shown in re 4.19, the different sam not make a significant difference and basically overlap each other. The re uation sho very high coefficient of det n R 2 of 0.999. The prog er air void an the ravity of plast FIG Cor am versus uncorrected air voids. As Figu ple sizes do gression eq wed erminatio ram had low s th uncorrected results because the embedded correlation factors reduced the specific g ic bags. This resulted in higher volume for the bags and thus lower volume of tested samples, and consequently higher bulk specific gravity of tested samples resulting in lower air voids. The differences between the uncorrected calculation (uncorrected vacuum seal air voids)and the program calculation (CoreLok air voids) are shown in Figure 4.20. 181 y = -0.0254x + 0.9471 R 2 = 0.4089 0.8 0.90 0.95 e c t ed pr og r m ) , % 0.60 0.75 0.80 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 Uncorrected Air Voids, % A i r V i d s D i er enc 0.65 0.70 5 1.00 o ff e, ( u n c or r - a FIGURE 4.20 Air voids difference between program and uncorrected calculation. From Figure 4.20, one can observe that the difference between the two calculation approaches ranged from 0.63 to 0.95 percent, and had an average of 0.80 percent with a standard deviation of 0.05 percent. The correction in air voids by the program decreases with the increase of air voids. This trend is not desirable since it is believed that for the amples should be at least similar or even higher. Therefore, the CoreLok bag tends to overestimate the sample volume more, as a result a higher or at least an equal amount of correction is needed. In consideration of the difference between the uncorrected air voids using TM the SSD method and the CoreLok method (0.5 percent), the combined difference between the uncorrected air voids for vacuum seal method and the SSD method is around 1.4 same type of mixture at higher air void content, the surface texture of compacted s Equation 4.2 and the Coregravity program (0.9 percent), and the difference between 182 percent for SMA with low air voids. This difference should be the average correla factor used for the vacuum seal method when lab compacted tion SMA samples are tested. The correlation factor for vacuum seal method can be determined by the sample with similar surface condition and low air voids. The similar surface condition is critical because a different correlation between the SSD and the CoreLok test results had been found for the sawn-and-cored samples and untreated lab compacted samples. Mixtures with different NMAS, which resulted in different surface conditions, also showed different relationships between the SSD and CoreLok test results. The low air void content is necessary because the SSD method results are believed to be accurate at low air voids. In summary, for impermeable samples with low air voids, the SSD method has little error potential therefore it should be used for air voids measurement. For permeable samples with high air voids, the water draining problem associated with the SSD method is difficult to correct, theref be used The correlation factor can be de by comparing the SSD air voids and uncorrected vacuum sealing air voids using test samples with similar surface conditions and low air voids. The water absorption can be easily used to determine if a significant error potential exists for the SSD method, a threshold value is approximately 0.4 to 0.9 percent difference between the SSD air voids and uncorrected vacuum seal air voids using a test ore the vacuum sealing method with correction factor should termined . for different NMAS mixtures. The threshold air voids for permeable samples will be discussed in the next section. 183 4.4 PE C tween permeability and air voids for SMA mixtures, and effect of compac e hown in Figures 4.21, 4.22 and 4.23, respectively. The impermeable (perme here RMEABILITY TEST Permeability tests were conducted on all the mix design samples compacted by the SG with 100 gyrations and 65 gyrations. The samples were tested as compacted without sawn or cored treatment. The permeability test results are shown in Appendix Table A1. The relationship be tion level on permeability are evaluated in this section. 4.4.1 The Relationship between Permeability and Air Voids Permeability test results had a range from 0.00 to 6343.9 (1?10 -5 cm/s). Based on the literature study (53-58), the NMAS affects the permeability at similar air void contents because of different size of voids, therefore the data was separated into three subsets based on NMAS. The relationships between permeability and total air voids for thre NMAS are s ability less than 0.01?10 -5 cm/s) samples were not included in the analysis. T were 56 impermeable samples (out of 372 tested samples), with an air voids range from 1.8 to 4.5 percent by the SSD method. 184 SSD: y = 0.0431x 4.8899 R 2 = 0.3642 Corelok: y = 0.0037x 5.4086 1.00E-02 1.00E-01 1.00E+00 1.00E+01 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Air Voids Content, % P e r eab i l i y, 10 -5 cm /s R 2 = 0.4498 1.00E+02 1.00E+03 1.00E+04 m t SSD Corelok Pow er (SSD ) Pow er (Corelok) 125 5.1 6.9 FIGURE 4.21 Relationship between permeability and VTM for 19 mm NMAS. SSD: y = 0.004x 5.8609 2 R = 0.5575 CoreLok: y = 0.0006x 6.1449 2 00E-02 00E- 1.00E+ it R = 0.5796 eab il 1.00E+01 1.00E+02 1.00E+03 1.00E+04 y, 10- 5 cm / s 1. 1. 01 00 0.0 2.0 4.0 6.0 8.0 10.0 12.0 Air Voids Content, % P e r m SSD CoreLok Pow er (SSD) Pow er (CoreLok ) 125 5.8 7.3 FIGURE 4.22 Relationship between permeability and VTM for 12.5 mm NMAS. 185 SSD: y = 0.0015x 5.8968 R 2 = 0.544 CoreLok: y = 0.0002x 6.5912 1.00E-02 6.0 7.0 8.0 9.0 P e bi l i ty , 10 - 5 R 2 = 0.6088 1.00E+02 1.00E+03 1.00E+04 c m / 1.00E-01 1.00E+00 1.00E+01 0.0 1.0 2.0 3.0 4.0 5.0 Air Voids Content, % r m ea s SSD Corelok Pow er (SSD ) Pow er (Corelok) 125 6.8 7.6 ermeability and VTM for 9.5 mm NMAS. It is noteworthy that the coefficients of determination R 2 for the regressions in all Figures are not high (from 0.36 to 0.61). This is because the data for each regression includes variation in aggregate type and AC content, and these factors will affect the size, shape, and connectivity of the air voids. Therefore the amount of interconnected air voids will vary even if the total air voids are the same. The amount of interconnected air voids is believed to have a better correlation with permeability but it is difficult to measure. If the threshold value for permeable SMA mixtures is set at 125?10 -5 cm/s (111), rom Table 4.18, one can observe that the cr ntent increases with the decreas FIGURE 4.23 Relationship between p the critical air voids values by both SSD and CoreLok methods are shown in Table 4.18. itical air void coF e of NMAS. In other words, SMA mixtures become permeable at higher air void content for lower NMAS. The critical air void contents by the SSD method are 5.1, 5.8, and 6.8 percent for 19 mm, 12.5 mm, and 9.5 mm NMAS mixtures, respectively. 186 TABLE 4.18 Critical Air Void Content for Permeable SMA Mixtures NMAS Test Method Regression Equation Critical 1 VTM, % SSD 8899.4 0431.0 a Vk = , R 2 =0.3642 5.1 19 mm CoreLok 4086.5 0037.0 a Vk = , R 2 =0.4498 6.9 SSD Vk = , R 2 =0.5575 5.8 8609.5 004.0 a 12.5 mm CoreLok 1449.6 0006.0 Vk = , R a 2 =0.5796 7.3 SSD 8968.5 0015.0 a Vk = , R 2 =0.5440 6.8 9.5 mm CoreLok a Vk = , R =0.6088 7.6 5912.6 0002.0 2 1. Critical VTM calculation based on threshold permeability 125?10 -5 cm/s The critical air void contents were consistent with the discussion in air voids comparison for the CoreLok and SSD method. As shown in Figures 4.21 to 4.23, for 19 mm NMAS, the critical air voids for a permeable SMA mixture was 5.1 percent by the mm NMAS, the critical air void content by the SSD method was 5.8 percent, which was the similar air voids level (5.9 percent, as shown in Figure 4.11) at which the two methods began to have more than one percent difference. For 9.5 mm NMAS, most of the tested samples were not considered permeable, and the difference between the corrected 4.11). For air voids higher than the critical value, the SSD method was considered inaccurate because of the problem of water draining out of the specimen during the SSD procedure. Based on the air voids and permeability comparison information (Figure 4.11, NMAS mixtures with similar results to be expected. This is because the relatively low surface texture and relatively high threshold air voids to become permeable for 9.5 mm mixtures. For 12.5 mm NMAS mixtures with about 6.0 percent or more air voids and for SSD method, which corresponds to a corrected CoreLok air voids of 6.4 percent. For 12.5 CoreLok and SSD results for most of the tested samples were within one percent (Figure Figures 4.21 to 4.23), both the SSD and CoreLok methods can be used for SMA 9.5 mm 187 19 mm NMAS mixtures with more than about 5.0 percent air voids, the SMA mixtures be p s a er p al f r b SD d than when the vacuu ealing me is u 4.4.2 Effect of Compaction el erm ility At similar air voids, different com on l s may lt in fferen voids distribution and a different degree of interconnectivity of those voids, therefore different permeability. The eff compaction level on perm was determ co ng eab est ult two pact evels The comparisons of pe bility st results for th MA xtures are shown in Figures 4.24, 4.25, and 4.26, respectively. The coefficients of determination R 2 for the regressions in all Figur are compatible with r permeability stud (55, 58-59). Th rrela would be he tiv nte nn air v s cou e used s disc d in th literature review, perm y depend on the total air voids, but also depended on the size, shape, and distribution of these voids. come ermeable and therefore th iere great otenti or erro y the S metho m s thod sed. lev on P eab pacti evel resu a di t air ect of eability ined by mpari perm ility t res s for com ion l . the rmea te ree N S mi es are not high (from 0.39 to 0.66). However, these values othe ies e co tion be tter if t effec e or i rco ected oid ld b . A usse e eability did not onl 188 100 Gyrs 0.0 307 R 4779 : y = 2 = 0. 444x 5. 65 G 0.01 R = 0.3882 1.00E-02 1.00E-01 1. 1.00E+01 1.00E+02 1.00E+03 1.00 4 0.00 1.00 3. 4 5 6. 7.00 Air Voids by SSD, % P e r i l i t y , 1x 10 - 5 yrs: y = 2 41x 5.4372 00E+00 E+0 2.00 00 .00 .00 00 m e a b c m /s 100 Gyrs 65 Gyrs Power (100 Gyrs) Power (65 Gyrs) FIGURE 4.24 Permeability results for 65 and 100 gyration levels for 19 mm NMAS. 10 10 0 G y r s 65 Gyrs P 00 Gyrsower (1 ) Power (6 0 Gyrs: y .009 R 2 = 332 = 0 0.6 2x 5.4583 65 Gyr 000 .4385 1.00E-02 1.0 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.0 6.00 7.00 8.00 9.00 Voids by SSD, % r m eab i l i t y, - 5 c m /s s: y = 0. R 2 = 0 3x 7.3986 0E-01 0E+04 P e 1x10 0.00 1.00 2.00 3.00 4.00 5.00 Air 5 Gyrs) FIGURE 4.25 Permeability results for 65 and 100 gyration levels for 12.5 mm NMAS. 189 100 Gyrs: y = 0.0 5298 R 2 = 0.436 113x 4. 65 Gyrs: y E- R 2 = 0.6596 1.00 1.00E-01 00E 1.00E+01 1.00E+02 1.00E+03 1. 0.00 2. 7. Air oids by S , % P e r m it y, 1x10 - 5 = 8 05x 7.7515 E-02 1.eab il +00 00E+04 1.00 00 3.00 V 4.00 SD 5.00 6.00 00 cm / s 10 0 G y r s 65 Gyrs Power (100 Gyrs) Power (65 Gyrs) FI 4.2 rme ty r u r 65 d 100 atio els f 5 mm AS. As shown in Figure 4.24, 19 mm NMAS SMA mixtur esign ith 65 gyr gen y ha er rm ility similar air voids. This indicates one of the advantage of using 65 gyrations for designing For 12.5 and 9.5 mm NMAS mixtures as shown in Figure 4.25 and 4.26, the permeability test results for the two compaction levels are mixed at similar air voids. SMA mixtures designed with 65 gyrations had a slightly higher slope than those designed with 100 gyrations. The two best-fitted regression lines crossed at an intermediate air void content. SMA mixtures designed with 65 gyrations had higher asphalt content than those designed with 100 gyrations at similar air voids (Figure 4.4). The high asphalt content will help seal the air voids and prevent the connectivity of internal air voids. At low air GURE 6 Pe abili es lts fo an gyr n lev or 9. NM es d ed w ations erall d low pe eab than those designed with 100 gyrations at SMA mixtures. 190 void levels, this sealing effect of the asphalt binder becomes more significant and makes those designed with 65 gyrations more impermeable. At high air void levels, the asphalt binder seems not enough to seal the interconnected voids and the mixture becomes permeable. The critical air voids calculated from the best-fitted regression lines for two compaction levels are shown in Table 4.19. As shown in Table 4.19, the critical VTM is lower for larger NMAS mixture. This is because the size of air voids tends to become larger and more interconnected with larger NMAS mixture , resulting in higher permeability at the same level of total air voids. TABLE 4.19 Critical Air Void Content for Permeable SMA Mixtures NMAS Gyration Level Regression Equation Critical* VTM, % 100 , R 2 =0.4779 4.5 307.5 0444.0 a Vk = 19 mm 65 , R 2 =0.3882 5.3 4372.5 0141.0 a Vk = 100 , R 2 =0.6332 5.7 4583..5 0092.0 a Vk = 12.5 mm 65 , R 2 =0.4385 5.7 3986.7 0003.0 a Vk = 100 , R 2 =0.4360 7.8** 5298.4 0113.0 a Vk = 9.5 mm 65 , R 2 =0.6596 6.3** 7515.7 00008.0 a Vk = * Critical VTM calculation based on threshold permeability 125?10 -5 cm/s, by SSD method. ** The critical VTM is extrapolated from regression equation or at the end of data range. For 19 mm NMAS, the SMA mixture designed with 65 gyrations become permeable at higher air voids than those designed with 100 gyrations. For 12.5 mm MAS, the critical air voids for the two compaction levels are the same at 5.7 percent. For 9.5 mm NMAS e cri r v those designed with 65 gyrations are lower than those designed with 100 gyrations. It is notable that for 9.5 mm NMAS, the critical air vo ults are extrapolated at the high end of the sample air voids range. As shown th 65 gyrations resulted N , th tical ai oids for ids res in Figure 4.26, the low permeability results for those designed wi 191 in a hig ult in a ge. sum error (SSE) for each individual regression of one compaction level and an SSE for a new regression of combined data were calculated. Then an F statistics was formulated by t ual regressions and the total SSD (SSD (T)) of two individual regressions. The results for the F tests are summarized in Table 4.20. TABLE 4.20 F-tests for permeability regressions of two compaction levels her slope than those designed with 100 gyrations. This high slope may res misleading extrapolated critical air voids at the very high end of the air voids ran Three F tests were employed to examine the difference between the regression equations of the two compaction levels for the three NMAS. For each of the F test, a of square he SSD difference (SSD (D)) between combined and individ SSE (D) DF (D) SSE(T) DF(T) F stat P value 19 mm 7060603 2 54394245 89 5.776 0.004 12.5 mm 1263962 2 3100691 93 18.955 0.000 9.5 mm 9294 2 241115 81 1.561 0.216 As shown in Table 4.20, the difference between the regression for 100 and 65 gyrations are significant for 19 mm and 12.5 mm SMA mixture results. For 9.5 mm SMA mixtures, the difference between the permeability results for the two compaction levels is not significant even though the two regressions showed big difference. This is logical because as shown in Figure 4.26, the two sets of data are mixed together and the two regression lines are crossed in the middle. In the overall range of tested air voids, there i no significant difference between two compaction levels. In summary, the SMA mixtures designed with 65 gyrations generally had simila or lower permeability than those designed with 100 gyrations at a given level of air v For 19 mm NMAS mixtures, 65 gyrations resulted in a lower permeability than 100 s r oids. 192 gyratio 4.5 AGGREGATE DEGRADATIONS Samples were tested in the NCAT ignition oven for each aggregate and NMAS combination (loose samples, Marshall compacted samples, and the SGC samples) to determine the aggregate breakdown. Washed gradation analyses were conducted on the remaining aggregates to determine the aggregate breakdown. The aggregate breakdown for the combinations of five aggregates and three NMAS using different compactive efforts are shown in Appendix C, from Figure C1 to C15. A typical result of aggregate break nder diff co e sh Figu .27. ns at similar air voids. For 12.5 and 9.5 mm NMAS mixtures, the effects of two compaction levels on permeability were not significant. down u erent mpactive fforts is own in re 4 C.GVL N 0 40 50 60 0 0.45 Power Sieve Size, mm c e n t 12.5 mm MAS 10 Orig Loos 90 10 20 30 Pe r 70 g , 80 % e 100 Gyrs P a s s i n 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 9.5 19 FIGURE 4.27 Typical aggregate breakdown results for different compaction efforts (C.GVL 12.5 mm NMAS). As shown in Figure 4.27, the gradation of the loose mixture is basically the same as the original gradation, which indicates there was virtually no breakdown as a result of 193 mixing and testing in the ignition oven. However, there was a big difference between compacted samples to loose samples. The Marshall compactor produced more breakdown than the SGC, and therefore produced gradations that were denser and closer to the maximum density line. The 100 gyrations compaction effort generally resulted in some additional aggregate breakdown when compared to the 65 gyrations compaction. However, the difference between these two compaction levels was generally not large. The gradation changes by using different compaction efforts, including the Marshall compaction, 100 gyrations by the SGC, and 65 gyrations by the SGC are summarized in Tables 4.21, 4.22, and 4.23, respectively. Each result shown in the tables is the average value from two replicates. The information in these tables includes the aggrega e es due to the c ile e e t type, LA abrasion, NMAS, F&E content, and selected sieve size chang ompaction. The deviation in results was shown as the difference in gradation between compacted samples and loose mixtures. The results therefore present the compaction effect only. As shown in Tables 4.21 to 4.23, lab granite had the highest breakdown wh traprock had the lowest breakdown. The lab granite is a major aggregate type used in Georgia and the traprock is the primary aggregate used in Maryland. These two States probably are the two biggest producers of SMA. The State of Georgia has specified th 50 gyration level for designing SMA mixture, while the State of Maryland has always insisted on 100 gyrations. The reason for the different specification is likely due to th different availability of aggregate sources. The lab granite has high breakdown therefore a lower compaction level is desired, while the traprock produces very little breakdown and therefore can use the higher compaction level. 194 TABLE 4.21 Marshall Compaction Aggregate Breakdown Results Type % mm % change, % change, % change, % change, % change, % Agg. L.A. Abrasion, NMAS, Grad. F&E, 12.5mm sieve 9.5 mm sieve 4.75 mm sieve 2.36 mm sieve 0.075 mm sieve 19 N* 23.8 11.1 16.8 9.2 5.8 0.4 12.5 N 35.2 -- 8.4 11.9 6.9 1.1 C.GVL 30.7 9.5 N 37.7 -- 4.1 10.1 6.9 0.4 19 N 18.6 9.3 15.2 9.9 6.3 0.6 12.5 N 28.1 -- 7.8 11.7 6.1 0.1 L.GRN 36.4 9.5 N 30.3 -- 2.7 12.6 7.7 0.6 19 N 20.7 15.1 15.0 8.6 4.8 1.2 12.5 N 25.5 -- 8.7 11.4 5.7 1.4 LMS 26.4 9.5 N 26.8 -- 2.7 10.1 5.7 1.5 19 N 24.1 4.5 11.7 7.3 3.7 0.5 12.5 N 23.4 -- 6.2 9.9 4.5 0.6 R. GRN 20.6 9.5 N 24.7 -- 1.1 9.1 4.3 0.5 19 N 12.8 3.7 6.8 3.0 1.0 0.0 12.5 N 17.7 -- 5.7 6.3 3.1 0.4 TRAP 16.6 9.5 N 19.5 -- 0.9 7.0 3.2 0.3 19 F* 24.0 2.9 8.8 5.5 2.5 1.1 12.5 F 23.8 -- 4.9 9.5 3.4 0.5 R. GRN 20.6 9.5 F 24.7 -- 2.7 9.6 4.7 1.3 19 F 13.9 3. 9 2.5 0.3 0.5 1 3. 12.5 F 18.2 -- 2.6 2.8 0.1 0.4 TRAP 16.6 9.5 F 19.5 -- 2.6 2.4 1.3 0.2 Note: -- Test result is not available for this sieve. * N and F stand for normal and fine gradation, respectively. 195 TABLE 4.22 100 Gyrations Aggregate Breakdown Results Agg. Type Abrasion, NMAS, mm Grad. F&E, % sieve sieve sieve sieve sieve L.A. % 12.5mm change, % 9.5 mm change, % 4.75 mm change, % 2.36 mm change, % 0.075 mm change, % 19 N* 23.8 5.3 10.4 6.1 4.1 0.5 12.5 N 35.2 -- 7.6 8.7 4.7 0.9 C.GVL 30.7 9.5 N 37.7 -- 3.0 8.1 5.2 0.7 19 N 18.6 4.3 9.4 8.0 5.7 1.5 12.5 N 28.1 -- 2.2 9.7 5.2 0.8 L.GRN 36.4 9.5 N 30.3 -- 0.4 9.7 6.0 1.1 19 N 20.7 5.9 8.1 5.4 3.8 0.7 12.5 N 25.5 -- 3.2 5.8 3.8 0.6 LMS 26.4 9.5 N 26.8 -- 1.5 5.2 4.8 1.3 19 N 24.1 1.4 6.7 5.5 3.2 0.7 12.5 N 23.4 -- 2.3 5.6 3.0 0.7 R. GRN 20.6 9.5 N 24.7 -- 0.1 6.3 3.8 0.3 19 N 12.8 0.6 2.0 -0.1 -0.6 -0.3 12.5 N 17.7 -- 2.4 1.2 1.4 0.3 TRAP 16.6 0.5 9.5 N 19.5 -- 1.1 0.9 2.0 19 F* 24 2.1 5.3 4.6 2.7 1.0 .0 12.5 F 23 -- 3.0 6.2 2.5 0.7 .8 R 20.6 .5 F 24 -- 1.9 6.5 3.8 0.8 . GRN 9 .7 19 F 13.9 1.6 0.6 1.2 -0.2 0.4 12.5 F 18.2 -- 0.9 1.7 -0.1 0.6 TRAP 16.6 9.5 F 19 -- 2.0 0.3 0.3 0.1 .5 Note: * N and F sta r norm fine on, resp . result is not avail r thi nd fo al and gradati ectively -- Test able fo s sieve. 196 TABLE 4.23 65 Gyrations Aggregate Breakdown Results Agg. Type Abrasion, NMAS, mm Grad. F&E, % sieve sieve sieve sieve sieve L.A. % 12.5mm change, % 9.5 mm change, % 4.75 mm change, % 2.36 mm change, % 0.075 mm change, % 19 N* 23.8 4.7 8.2 5.4 3.8 0.7 12.5 N 35.2 -- 4.5 7.2 4.4 1.0 C.GVL 30.7 9.5 N 37.7 -- 1.6 5.3 4.2 0.7 19 N 1 5.6 8.6 7.3 5.5 1.7 8.6 12.5 N 2 -- 3.7 8.0 4.5 0.8 8.1L.GRN 36.4 8.1 5.4 1.1 9.5 N 30.3 -- 1.8 19 N 20.7 5.1 7.5 5.1 3.2 1.2 12.5 N 25.5 -- 4.2 5.9 3.4 0.8 LMS 26.4 9.5 N 26.8 -- 1.4 5.2 3.4 1.1 19 F* 24.1 2.9 5.4 4.5 2.5 0.9 12.5 F 23.4 -- 3.2 5.7 2.4 0.6 R. GRN 20.6 9.5 F 24.7 -- 1.7 6.4 3.5 1.3 19 F 12.8 2.4 0.3 1.1 -0.1 0.5 12.5 F 17.7 -- 1.4 1.4 0.0 0.7 TRAP 16.6 9.5 F 19.5 -- 0.9 -0.2 0.0 0.1 Note: * N and F stand for normal and fine gradation, respectively. The aggregate breakdown results for N gradation of the traprock and ruby granite are also included in Tables 4.21 and 4.22. Th is shown in Figure 4.28. One standard deviation is shown as the error bar for each result. The pooled standard deviations for N and F gradations are 0.35 and 0.31 percent, percent passing the critical sieves. The critical sieve chosen was the 4.75 mm sieve for 19 ixtures. -- Test result is not available for this sieve. e comparison of N gradation and F gradation respectively. The aggregate breakdown results were represented by the changes in mm and 12.5 mm NMAS mixtures, and the 2.36 mm sieve for 9.5 mm NMAS m 197 These sieve sizes are most critical for SMA gradations and are used to determine the nd F gradations is generally not large and doesn?t show a consistent trend. This is not as expected but not surprising. The F gradation is finer than the N gradation and closer to the maximum density line, therefore it was expected to have less degradation. However, the difference between the F and N gradations is not large, both gradations fall within the gradation band for SMA mixtures (Tables 3.1 and 3.3). Therefore the test variability overwhelmed the effect of small differences in gradation. existence of stone-on-stone contact. As shown in Figure 4.28, the difference between N a -2.0 2.0 0.0 4.0 8.0 10.0 12.0 19 12 . 5 9 . 5 19 12 . 5 9 . 5 19 12 . 5 9 . 5 19 12 . 5 9 . 5 6.0 R.GRN TRAP R.GRN TRAP C r i t i c al S i ev n g e s , % e Ch a 100 Gyrations M arshall N gradation F gradation N = 2 . TA FIGURE 4.28 Comparison of N and F gradation on aggregate breakdown BLE 4.24 Paired-T Test on Degradation for Two gradations N Mean StDev SE Mean N gradation 12 4.233 2.820 0.814 F grad 12 3.675 2.607 0.753 ation Difference 12 0.558 1.458 0.421 95% CI for mean difference: (-0.367754, 1.484420) T-Test of mean difference = 0 (vs not = 0): T-Value = 1.33, P-Value = 0.211 198 A paired T-test (Table 4.24) was used to compare breakdown of N and F gradations and the results indicated that these two gradations had no significant difference in terms of aggregate breakdown under compaction. Therefore, the follow analysis on aggregate breakdown is limited to F gradation for the trap ing rock and ruby granite. -2.0 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 0.0 4.0 10.0 2.0 6.0 8.0 12.0 14.0 C.GVL L.GRN LMS R.GRN TRAP C es i r cen i t i i t P assi n g C r cal S eve, % h an g n P e Marshall 100 Gyrs 65 Gyrs N = 2 FIGURE 4.29 Critical sieve changes due to compaction. Figure 4.29 is a schematic summary of the breakdown results due to compaction. in percent passing the critical sieves, fro lic ac . O rd n is shown as the error bar for each result. (The oled standard deviations are 0.65, 0.29, and 0.22 percent for 50 blow Marshall, 100 ions an gyratio spect Fr 4.29, one c serve t arshal paction e the aggregate breakdown of the thr 0 gyrations generally broke more ag ared to 65 tions, bu pically n y a sig nt amount. The The aggregate breakdown results were represented by the average changes m two rep ates for e h mixture ne standa deviatio po gyrat d 65 ns, re ively). om Figure an ob hat M l com gav highest ee compaction efforts. The 10 gregate comp gyra t ty ot b nifica 199 cha the critical sieve has a range from 0.3 to 9.7 percent for 100 gyration action, from 0.0 to 8.0 percent for 65 gyrations SGC compaction, ults are as expected. The compactor and allows little reorientation of ggregates during compaction, therefore higher aggregate breakdown was expected in order to e Even though the effects of interact or as Source DF Seq SS Adj SS Adj MS F P nge in percent passing s SGC comp and from 1.3 to 11.9 percent for Marshall compaction. The res Marshall hammer is considered as an impact a achieve similar air voids to the gyratory compactor. On the other hand, the gyratory compactor produces a kneading action which allows the proper reorientation during compaction. An ANOVA was conducted on the critical sieve breakdown results to evaluate th effect of main factors (aggregate type, NMAS, and compaction level) and any interaction between these main factors. The ANOVA results shown in Table 4.25 indicate that all the main factors and all the interactions are significant. This result is due to the surprising low variability in error (about 1%) in the ANOVA result. ion are much lower than those of the main factors, all the interactions are shown as significant. This small error term makes the ANOVA results very sensitive to min effects. The effects of interactions may not be practically significant but will be shown statistically significant. TABLE 4.25 ANOVA on Aggregate Breakdown at Critical Sieve Size Agg. 4 492.35 492.35 123.09 665.74 0.000 Comp 2 124.84 124.84 62.42 337.60 0.000 NMAS 2 132.92 132.92 66.46 359.46 0.000 Agg.*Comp 8 14.62 14.62 1.83 9.88 0.000 Agg.*NMAS 8 14.55 14.55 1.82 9.84 0.000 Comp*NMAS 4 12.29 12.29 3.07 16.62 0.000 Agg.*Comp*NMAS 16 9.33 9.33 0.58 3.15 .001 0 Error 45 8.32 8.32 0.19 Total 89 809.22 200 The average breakdown for different aggregate types, compaction efforts and NMAS are shown in Figure 4.30. One standard deviation is shown as error bar. The numbers of results used for calculating the average values and standard deviations are also included in the Figure. The high variability as shown in the Figure 4.30 is as expecte he d because only a single factor is considered to draw the average bar while all t other factors are significant. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 C.GVL L.GRN LMS R.GRN TRAP 65 100 Mars 19 12.5 9.5 Aggregates Compaction Level NMASA v er ag e C h a n n ssin e C r i t i c al % n g e s i P e r c e t P a g t h S i ev e, N = 18, N = 30, N = 30 1 2 3 FIGURE 4.30 Average aggregate breakdown for all main factors. The aggregate breakdown depended on the aggregate type. The traprock had the lowest aggregate breakdown of 1.4 percent on average, while the lab granite had the highest aggregate breakdown of 8.2 percent on average. Aggregate properties such as L.A. abrasion and F&E content are believed to correlate the most to the aggregate breakdown value (36). The regressions for average breakdown value versus aggregate L.A. abrasion value and F&E content (3:1 ratio) are shown in Figure 4.31 and Figure 201 4.32, respectively. The correlation between aggregate breakdown under laboratory compaction and the L.A. abrasion value is very strong (R 2 larger than 0.84 for all NMAS ixtures). The results indicate that with the increase of aggregate L.A. abrasion, the aggrega indication of aggregate hardne gh and hard aggregate would m te breakdown is greater. This result is logical because the L.A abrasion is an ss and impact resistance, a tou be expected to have less breakdown during compaction. 19 m m N M A S 12.5 mm NMAS 9.5 mm NM AS 10.0 12.0 n , % 19 mm: y = 7.9841Ln(x) - 20.087 R 2 = 0.9454 12.5 mm: y = 9.0918Ln(x) - 22.148 R 2 = 0.8434 9.5 mm: y = 6.6883Ln(x) - 17.374 0.0 10 15 20 25 30 35 40 L.A. Abrasion Value, % A v eg r e w R 2 = 0.8836 2.0 4.0 6.0 8.0 er ag e A g g r at e B akd o 0 5 Log. (19mm NM AS) Log. (12.5 mm NM AS) Log. (9.5 mm NM AS) FIGURE 4.31 Relationship between aggregate breakdown and L.A abrasion value. 202 19 mm: y = 6.660 R 2 = 5Ln(x) - 14.2 0.3392 28 12.5 mm 483Ln(x) - 30.0 0.7998 : y = 11. R 2 = 34 9.5 mm 1Ln(x) - 21.42 7266 0 2.0 4.0 6 8.0 10.0 5 10 20 30 40 Content (3 Aver ag e A g g r e g at e B r eak d o w n , : y = 7.76 R 2 = 0. .0 0 .0 12.0 15 25 35 F&E :1), % % 19 m m N M A S 12.5 mm NM A S 9.5 mm NM AS Lo AS)g. (19 mm NM Log. (12.5 m )m NM AS Lo g. (9.5 mm NM A S) FIGURE 4.32 een aggregate breakdown and F&E content. The aggregate breakdown under laboratory co ction didn? a strong correlation (R 2 =0.34) with aggregate F&E content for 19 mm NMAS m ures. However, for 12.5 mm and 9.5 mm NMAS mixtur rrelation between aggregate F&E content and aggregate breakdown is good (R 2 ?0.73). The general trends showed an increase in F&E content created more aggregate brea n. This is l l because the flat or elonga icles will tend to break quicker than more cube-s d particles. As expected, the co action eff as also s ig ant factor. The Marshall compaction, 100 gyrations with the SGC, and 65 gyrations with the SGC gave an average change of 7.3, 5.2 and 4.6 per t in percen ing the critical sieves, respectively. As discussed before, the Marshall compactor is an impact compactor which allows om stable (average 4.9 percent) regardless the aggregate types (64). Both 65 gyrations and Relationship betw mpa t show ixt es, the co kdow ogica ted part hape mp ort w hown to be a s nific cen t pass little aggregate re-orientation and therefore tends to break more aggregates. Fr the literature review, aggregate breakdown due to the field construction seems relative 203 100 gyrations resulted in similar (average difference is within 0.3 percent) aggregate breakdown as in the field construction. Two paired t-tests were employed to compare the Marshall compaction with 1 gyrations, and the 100 gyrations with 65 gyrations for critical sieve changes. The results are shown in Tables 4.26 and 4.27, respectively. Both paired t-tests showed there were significant differences between the two compaction efforts compared. For the paired t t between 100 and 65 gyrations, 00 est even though the difference is not practically significant (0.6 percent), the statistical analysis showed a significant difference because of the consistent difference between the two compaction levels. This is logical because additional gyrations have to provide at least as much and probably more breakdown. TABLE 4.26 Paired T Test on Degradation for 100 Gyrations and Marshall Compaction N Mean StDev SE Mean Marshall 15 7.287 3.448 0.890 100 Gyrs 15 5.180 2.646 0.683 Difference 15 2.107 1.316 0.340 95% CI for mean difference: (1.37784, 2.83549) T-Test of mean difference = 0 (vs not = 0): T-Value = 6.20, P-Value = 0.000 TABLE 4.27 Paired T Test on Degradation for 65 and 100 Gyrations N Mean StDev SE Mean 100 Gyrations 15 5.18 2.646 0.683 65 Gyrations 15 4.54 2.337 0.603 Difference 15 0.64 0.541 0.140 95% CI for mean difference: (0.340460, 0.939540) T-Test of mean difference = 0 (vs not = 0): T-Value = 4.58, P-Value = 0.000 The NMAS was also shown as a significant factor in the breakdown of the aggregates. The 12.5 mm NMAS mixtures had the highest aggregate breakdown of 7.2 percent on average for the critical sieve, while the 9.5 mm NMAS mixtures had the lowest aggregate breakdown of 4.2 percent on average. The average breakdown values for the three compaction efforts at different sieve sizes are shown in Table 4.28. 204 TABLE 4.28 Average Percent Passing Changes at Three Sieve Sizes Average Breakdown Value at Sieve Size, % NMAS 9.5 mm 4.75 mm 2.36 mm 19 mm 8.2 5.6* 3.4 12.5 mm 4.4 7.2* 3.5 9.5 mm 2.1 6.6 4.2* * Selected as representative value as the choosing of critical sieve size. As shown in Table 4.28, for 12.5 mm NMAS mixtures, the critical sieve selected was the same sieve size (4.75 mm) at which the maximum breakdown happened. However, for 19 mm and 9.5 mm NMAS mixtures, the maximum breakdown due to compaction generally happened at 9.5 mm and 4.75 mm sieve size, respectively. The critical sieves selected for 19 mm and 9.5 mm NMAS mixtures were one size lower than where the maximum breakdown happened. As shown in Table 4.25, all the interactions between these main factors were significant. This indicates that aggregate breakdown depends on specific combinations of aggregate type, NMAS, and compaction level. Any factor alone can not be used to estimate the aggregate breakdown. The interaction between aggregate type and compaction levels is shown in Figure 4.33. One standard deviation is shown as error bar. For each aggregate, the three compaction efforts followed the same order for aggregate breakdown levels. However, the amount of breakdown was slightly different for different aggregate types. Considering the test precision of the sieve analysis test (single operator precision is 3.7%, from AASHTO T27), these differences are not significant. 205 0.0 2. ag e C 0 0 8.0 10.0 12.0 C.GVL L.GR LM R. T A ver h an n ssin e al 4. 6.0 g es i P e r cen t P a g t h C r i t ic S i eves, % 65 Gyrs 100 Gyrs Marshal N S GRN RAP l N = 6 FIG teraction etwee ga an act vel regate akd URE 4.33 In b n aggre te type d comp ion le on agg bre own. -2.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 e e P e r cen t P assin g t h i t i c a l i eves, % e C r 19 C.GVL L.GRN LMS R.GRN TRAP A ver ag C h an g s i n S mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 breakdown. FIGURE 4.34 Interaction between aggregate type and NMAS on aggregate The interaction between aggregate type and NMAS is shown in Figure 4.34. One standard deviation is shown as error bar. For each aggregate type, the three NMAS 206 followed the same order for aggregate breakdown results. However, the differences among the three NMAS varied a little for different aggregate types. Considering precision of the sieve analysis test, these differences are not practically significant. the test 0.0 2.0 4.0 6.0 8.0 10.0 12.0 A ver e C h g es in P e r cen t P assi g t h S i eves, % 14.0 19 mm NMA S 12.5 mm NMA S 9.5 mm NMA S ag an n e C r i t ical 65 Gyrs 100 Gyrs Mars hall N = 10 FIGURE 4.35 Interaction between compaction level and NMAS on aggregate breakdown. rts , , aggregate breakdown, while 65 gyrations provided the least aggregate breakdown. The The interaction between compaction level and NMAS is shown in Figure 4.35. One standard deviation is shown as error bar. For each NMAS, the compaction effo followed the same order for aggregate breakdown results. However, for different NMAS the amounts of change varied for different compaction levels. For example, for 19 mm NMAS mixtures, the difference between 100 gyrations and 65 gyrations was smaller when compared to that of mixtures with the other two NMAS. Considering the test precision of sieve analysis test, these differences are not significant. In summary, three compaction efforts that were evaluated included 65 gyrations 100 gyrations, and 50 blow Marshall. The Marshall compaction provided the most 207 difference between 65 and 100 gyrations in terms of aggregate breakdown values at t critical sieves are sho he wn to be statistically significant, but not large. Both 65 and 100 gyratio 4.6 SUMMARY The effects of different compaction levels on volumetric properties, permeability and aggregate breakdown were evaluated. The two test methods for determining air voids was compared and suggestions on how to properly determine air voids for SMA mixtures at different air void levels were made. The test results for different compaction levels are summarized in Table 4.29. TABLE 4.29 Average Test Results for Different Compaction Levels Properties\Compaction level 50 blows Marshall 100 Gyrations 65 Gyrations 40 Gyrations 1 Suggested Criteria ns provided similar aggregate breakdown as in the field construction. Asphalt content, % 6.4 5.9 6.6 7.4 6.0 min VMA, % 18.1 17.1 18.6 20.1 17 min VCA ratio 0.89 0.88 0.90 0.97 1.0 m x a Draindown, % N/A 2 N/A 0.10 0.08 0.3 max Breakdown, % 7.3 5 6 N/A 4.9 3 .2 4. Permeability for 19 mm NMAS mix at 6% air voids 4 , N/A 190 74 N/A 125 max 1?10 -5 cm/s 1. Only two mixtures were designed with 40 gyrations. 4. Air voids is the corrected vacuum sealing air voids. . 2. N/A?the tests were not conducted for this compaction effort. 3. Average value from field observation (64). The volumetric properties comparison on four compaction levels indicated that with the decrease of compaction level, the optimum asphalt content and VMA increased 208 All mixtures designed with 65 and 40 gyrations met the minimum optimum asphalt content of 6.0 percent and minimum VMA of 17 percent, while only 8 out 15 mixtu designed with 100 gyrations met these requirements. All designed SMA mixtures in th study met the VCA requirement and draindown requirement. Th res is e decrease of compaction level will result in more durable mixtures and will allow the use of more aggregate types for designing SMA mixtures if rutting is not a problem. For determination of air voids, both the SSD and CoreLok methods can be used for SMA 9.5 mm NMAS mixtures with similar results to be expected. For 12.5 mm NMAS SMA mixtures with more than 6.0 percent air voids or more than 0.6 percent water absorption during the SSD test; and for 19 mm NMAS SMA mixtures with more than about 5.0 percent air voids or more than 0.4 percent water absorption during the SSD test, there is a greater potential for error by the SSD method than when the CoreLok method is used. Since this error is difficult to correct with the water draining problem for the SSD method, the CoreLok method is recommended for SMA mixtures with high air voids. However, the correlation factor embedded into the CoreGravity TM program is not sufficient for SMA laboratory compacted samples, an additional 0.5 percent correlation factor was recommended for use in this study. A further study of the correction factor indicated that an average 1.4 percent should be used on top of the uncorrected vacuum sealing air voids for lab compacted SMA mixtures when the vacuum sealing method was used. From the comparison of two compaction levels on permeability, 65 gyrations resulted in a lower permeability than 100 gyrations at similar air voids for 19 mm NMAS 209 mixtures. For 12.5 and 9.5 mm NMAS mixtures, the effects of compaction level on difference in permeability between the two compaction levels were not significant. Marshall compaction resulted in significantly higher aggregate breakdown than gyratory compaction. The increase of compaction level from 65 gyrations to 100 gyrations resulted in some additional aggregate degradation. But both 65 and 100 gyrations resulted in similar aggregate breakdown as that observed during field compaction. 210 CHAPTER 5 TEST RESULTS, ANALYSIS AND DISCUSSION OF PERFORMANCE TESTS The rutting resistance for the SMA mixtures was evaluated by several laboratory tests, including the APA rutting test as a simulative test, and more fundamental tests such as dynamic modulus, static creep and repeated load tests. This chapter presents the results, data analysis, and discussions of these performance tests. The di on these perform eld scussions emphasize the effects of different compaction levels ance test results. The discussion of the test results will give the basis for recommending a compaction level that provides a more durable mix with satisfactory rutting resistance. 5.1 APA RUTTING TEST 5.1.1 APA Test Results and Analysis As shown in the literature (72-77), the APA rutting test has been validated by many fi projects, and can be used for differentiating rut susceptible mixtures. Many state DOTs have begun to use this test to test the rutting resistance of HMA during mix design and quality control/quality assurance. The APA rut depth results for the 32 mixes designed with the SGC at the three levels of compaction are summarized in Table 5.1. 211 TABL Level mm Content, % Rutting, mm rutting, mm E 5.1 APA Rutting Results Gyration Agg. Type NMAS, Asphalt APA St. Dev. Of APA 19 5.8 4.10 0.62 12.5 6.4 2.80 0.50 C.GVL 9.5 6.2 2.63 0.41 19 4.8 2.37 0.30 12.5 5.4 2.29 0.65 L.GRN 9.5 5.7 3.61 1.01 19 5.1 2.47 0.78 12.5 5.5 4.42 1.22 LMS 9.5 5.3 3.92 0.13 19 6.2 2.55 0.42 12.5 6.0 2.41 0.61 R.GRN 9.5 6. 100 7 2.59 0.21 19 6.7 4.46 0.87 12.5 6.1 2.67 0.35 TRAP 9.5 6.5 3.14 0.33 19 6.6 4.17 0.58 12.5 7.1 5.04 1.54 C.GVL 9.5 6.5 4.43 1.02 19 6.0 3.72 1.07 12.5 6.5 4.68 1.37 L.GRN 9.5 6.6 3.94 1.19 19 6.0 2.97 1.22 12.5 6.5 6.14 0.91 LMS 9.5 6.1 4.73 1.39 19 6.6 2.65 0.43 12.5 6.7 3.01 0.64 R.GRN 9.5 7.2 2.42 0.20 19 7.0 4.15 1.08 12.5 6.5 3.22 0.99 65 9.5 7.0 3.93 1.53 TRAP L.GRN 12.5 7.2 5.05 1.71 40 R.GRN 12.5 7.5 4.56 0.86 A side by side comparison of the APA rut depth results for all three gyration levels (40 gyrations was used only for 12.5 mm NMAS mixture with two aggregate types: 212 lab granite and ruby granite) are shown in Figure 5.1. One standard deviation is shown as error bar. The pooled standard deviation values for rut depth results are 0.63, 1.08, an 1.35 mm with the compaction level of 100, 65, and 40 gyrations, respectively. Thes variability values are less than the maximum allowable variability (2.0 mm) specified in the AASHTO TP-63 (105). The increased test variation with low d e er compaction level may be due to the increased rut depth results and higher variability in aggregate orientation with lower compaction level. 0.00 2.00 3.00 5.00 6.00 8.00 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 1.00 4.00 7.00 C.GV L L.GRN LMS R.GRN TRA P AP A Ru t n g , m t i m 100 Gyrs 65 Gyrs 40 Gyrs N = 3 FIGURE 5.1 Comparison of APA rutting for three compaction levels. ad d with 100 gyrations because of the higher asphalt content. Even with higher asphalt contents, the mixtures designed with 65 gyrations were still rut resistant except for the crushed gravel and limestone 12.5 mm mixtures. Thirteen of fifteen From Figure 5.1, one can observe that the APA rut depth generally increased with a decrease of compaction level. The two SMA mixtures designed with 40 gyrations h higher APA rut depths than the same mixture designed with higher compaction levels. SMA mixtures designed with 65 gyrations generally had higher APA rut depths than those designe 213 mixtures (87 ) compacted with 65 gyrations stil% l performed well when 5.0 mm was used as the maximum rut pth al ). For 40 gyrations, one o m s still met this requirement and d a 4.6 t dep epth of t SMA mixtures us gra tr k a ates wer ve to com ction an th res e o thr regates. For these two aggregates, going from 100 to 65 gyrations resulted in an average 0.26 mm in t dept hile th three gates n av e of 1.25 mm in kely d o the s change in optimu phalt ten age 0.5 percent) and aggregate degradation (average t) for these two aggregates when compared to the other three aggregates (average 0.9 percent change in asphalt content, and 0.8 o e er analysis o hese A depth data was performed by conducting an ANOV valuate the ect of t com l) and an nteract tween the main s on ept ce only two re design with 4 tions mparison of m al PA rut depth between 40 gyrations with the other two action ls co ot de. Th ere ded The APA results of these two mixtures were only used for comparison to the rut depth results e te t sam AS. de lowed (68 of tw ixture ha mm ru th. The rut d he ing ruby nite and aproc ggreg e less sensiti pa level th e mixtu using th ther ee agg crease in APA ru h w e other aggre had a erag crease. This is li ue t maller m as con t (aver 0.3 percen percent change in degradation) between the two compaction levels. The effects of compaction level on mix properties are essentially caused by the optimum asphalt content and aggregate gradation changes due to compaction. When the differences in these tw properties are smaller, the difference in performance for two mixtures is likely to b smaller. Furth f t PA rut A to e eff he main factors (aggregate type, NMAS, and paction leve y i ions be factor rut d hs. Sin mixtures we ed 0 gyra , o the c ean v ue of A comp leve uld n be ma erefore, two mixtures designed with 40 gyrations w exclu from this analysis. of mixtures using the sam aggrega ype and e NM 214 Results of the ANOVA are presented in Table 5.2. From Table 5.2, aggregate type and compaction level were significant influencing factors. The interaction between aggregate type and NMAS, and the interaction between NMAS and compaction level were also significant. Side by side comparison for the effects of these significant fact and interactions were shown in the Figures 5.2, 5.3 and 5.4 to better visualize the discussion. TABLE 5.2 ANOVA for APA Rutting Results Source DF Seq SS Adj SS Adj MS F P ors Agg. 4 23.7189 23.7189 5.9297 7.54 0.000 NMAS 2 1.4509 1.4509 0.7254 0.92 0.403 Gyrs 1 16.2053 16.2053 16.2053 20.59 0.000 Agg.*NMAS 8 27.0070 27.0070 3.3759 4.29 0.000 Agg.*Gyrs 4 5.6395 5.6395 1.4099 1.79 0.142 NMAS*Gyrs 2 5.3331 5.3331 2.6666 3.39 0.040 Agg.*NMAS*Gyrs 8 4.5286 4.5286 0.5661 0.72 0.674 Error 60 47.2138 47.2138 0.7869 Total 89 131.097 0.00 2.00 4.00 5.00 6.00 3.00 1.00 C.GV L L.GRN LMS R.GRN TRA P 100 65 Aggregates Gyrations A v e r a g e De m AP A Ru t t i n g p t h , m N = 18, N = 45 1 2 FIGURE 5.2 Comparison of average APA rut depth. 215 Figure 5.2 shows the comparison of the average APA rut depth for different aggregate types and two compaction levels. One standard deviation for each group shown as error bar. N is each estone had the lowest coarse aggregate uncompacted voids of 46.6 percent as shown in Table 4.1. The uncompacted air voids was recommended by the NCHRP project 4-19 (113) as an indication of effects of aggregate shape, angularity and texture. Full-scale rutting tests (114) at Indiana DOT APT Facility indicated a good correlation between the rutting resistance and uncompacted air voids for coarse aggregate. Aggregates with high uncompacted air voids are likely to produce more rut-resistant pavement. The differences between all different aggregate types are not large considering the test precision (allowable standard deviation in AASHTO TP63 is 2.0 mm). The low value of the APA results for all aggregate types indicates the requirements of aggregat propert s suc ent maxim with 1 and N 2 represent the number of samples used for average for group. The ruby granite mixtures had the lowest average APA rut depth of 2.6 mm, while the limestone mixtures had the highest average APA rut depth of 4.1 mm. This is likely due to the lim e ie h as L.A abrasion (30 percent maximum), F&E content (20 perc um for 3:1 ratio) for SMA mixtures seems too stringent. Two aggregates (crushed gravel and lab granite) have L.A abrasion values higher than 30 percent, and only one aggregate (traprock) has the F&E content (3:1) less than 20 percent. However, two aggregates: lab granite and ruby granite have been used on Georgia SMA projects proven good rutting performance. The average rut depth for mixtures designed with 100 gyrations was 3.1 mm, while the average rut depth for mixtures designed with 65 gyrations was 3.9 mm. A 216 higher compaction level will result in lower optimum asphalt content and a tighter aggregate structure, therefore a better rutting resistance. However, considering the test precision of the APA rutting test, these is no practical difference for rutting between the SMA mixtures with the two compaction levels. Other studies (37, 115) have shown tha the rutting for SMA mixtures is not sensitive to asphalt content. This is because of t stone on stone contact and high voids in aggregates result in less build-up of pore pressure. Based on the literature review (68, 79), mixtures that have an APA rut depth less than 5.0 mm are deemed as rutting resistant. Most (29 of 32 mixtures) of the APA r depths measured in this study are below the 5.0 mm criteria and therefore should be resistant to rutting. The interaction between aggregate type and NMAS is shown in Figure 5.3. One standard deviation value is shown as error bar. For different aggregate types, the ord the three NMAS in term t he ut ers of s of average APA rut depth are different. However, this appears to be due to normal variations considering the precision of APA rutting test. 0.00 1.00 2.00 3.00 6.00 C.GV L L.GRN LMS R.GRN TRA P Av ag A r d e p t , m 4.00 5.00 7.00 er e AP u t h m 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 FIGURE 5.3 Interaction between aggregate type and NMAS on APA rut depth. 217 1.00 7.00 ag A e m 2.00 3.00 4.00 5.00 6.00 Av er e AP r u t d p t h , m 100 Gyrs 0.00 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS 65 Gyrs N = 15 FIGURE 5.4 Interaction between NMAS and compaction level on APA rut dept The interaction between NMAS and compaction level are shown in Figure 5. One standard deviation is shown as error bar. For different NMAS, a decrease of compaction level from 100 to 65 gyrations resulted in a different amount of increase i average APA rut depth. The highest increase in APA rut depth was 1.5 mm for 12.5 m NMAS h. 4. n m mixtures (from 2.9 to 4.4 mm), while the lowest increase in APA rut depth was 0.3 mm for 19 mm NMAS mixtures (from 3.2 to 3.5 mm). This indicates that 12.5 mm NMAS mixtures were more sensitive to compaction level than the other two NMAS mixtures in terms of APA rut depth. However, this difference is not considered as practically significant since the highest difference is only 1.5 mm, and the same trend with compaction level are shown for all three NMAS mixtures. 5.1.2 Discussion on APA Rut Depth versus Gyration Level The discussions in this section will focus on the effect of compaction level on the APA rut depth test results which will provide a basis to recommend a gyration level that provides as much asphalt as possible without causing potential rutting problems. All the 218 APA rutting results versus gyration levels are shown in Figure 5.5. Since only two m s ( m S it w ne t at evels while a est o tures desi wit rati l o ll an s on ect o tion on th A r sult d to uct ixture 12.5 m NMA SMA with lab gran e and ruby granite) ere desig d with hree gyr ions l ll the r f mix were gned h two gy on evels, an vera alysi the eff f gyra level e AP utting re is ifficult cond . y = R 79.55 509 2 = 0. 0 1 2 4 6 7 8 9 60 70 0 11 ation 4x -0.7 7949 .0 30 .0 .0 3.0 .0 5.0 .0 .0 .0 .0 40 50 8 90 100 0 Gyr s A P A r u t d e p t h , m m All Thre se Level Po Threw er ( e Levels) FIGUR 5 T elati bet APA dept d co on l re for SMA mix ign h 65 and 100 gyrations varied within similar ranges. Even with various aggregate types and v M the ranges a rge considering the test variability. The APA results for a ed wit yrat re less than the suggested criteria of 5 mm. The APA rut depths for most (13 out 15) of the mixtures designed with 65 gyrations are less than this criteria. E 5. he r onship ween rut h an mpacti evel. As shown in Figure 5.5, the APA rutting sults tures des ed wit arious N AS, re not la ll mixtures design h 100 g ions a 219 The two mixtures with three gyration levels are highlighted using different data sy . A fitt gre esults of these two mi lso included. A n ure the A t de f the mix dec ith increase of gyration level. However, the decrease A ru th i e w the c on l in es f to 100 gyratio or th o m , th rut de co arginal when the gyrat evel to 40 tion C 9-1 ject rovi corr n betw the laboratory rut depth and normalized field rut depth, as shown in Figure 5.6. The field rut depth data were from the MnRoad and W ck p . Th ut d were ined duc PA rutting tests on cylinder samp th 4 nt ai s at ade atur c on ween nd 7 percent ai s for rutti t is in F 5.7 f sa esea The ind air is only slightly larger than those for 4 per her the c atio b r ly icab r the ent ids s s use this mbols best ed re ssion line for the r xtures is a s show in Fig 5.5, PA ru pth o se two tures reases w an in AP t dep s not larg hen ompacti leve creas rom 65 ns. F ese tw ixtures e APA pth be mes m ion l drops gyra s. N HRP 7 pro (80) p ded a elatio een estra rojects e lab r epths obta by con ting A les wi perce r void PG gr temper e. A omparis bet 4 a r void APA ng tes shown igure rom the me r rch. results icate the APA rut depth for 7 percent voids cent. T efore, orrel n should e easonab appl le fo 6 perc air vo ample d in study. Without Section 24: y = 0.0007x 0.9278 All = 0.0011x 39 2 0.3647 0.00600 0.01000 0.012 0.014 0.016 0.018 e l d R p t h , m m 0.00800u t D e 00 00 00 00 R 2 = 0.7907 Data: y 0.81 R = 0.00000 0.00200 0.00400 0.0 5.0 10.0 15.0 20.0 Lab Rut Depths, mm Fi SQ ESA R T ( L s ) Secti om W mayb r on 24 fr e outlie estrack, FIGURE 5.6 Correlation between field and APA rut depth from NCHRP 9-17 project (80). 220 FIGURE 5.7 Effects of air voids on APA rut depths (80). 2 strong evidence to identify this point is an outlier, therefore the researchers (80) still used the regression from all data to set up the APA rutting criteria. If we use the regression equation with all data to predict the field rut depth, the average field rut depth will be expected as 8.7 and 10.5 mm after 10 million ESALs for SMA mixture designed with 100 and 65 gyrations, respectively. This level of rutting is acceptable if we assume the maximum acceptable field rut depth is 12.5 mm, which has been used for many studies (71, 80, 89). For SMA mixtures designed with 40 gyrations, the predicted field rut depth is 12.5 mm based on the average APA rut depth of 4.8 mm. This is the maximum acceptable field rut depth and therefore SMA mixtures designed with 40 gyrations may not have enough rutting resistance. As shown in Figure 5.6, the normalized field rut depths show a positive trend with lab rut depth. If a possible outlier from Westrack project is removed from the regression, the R value of the regression equation increased from 0.36 to 0.79. However, there is no 221 In summary, the APA rut depth appears to be affected by the aggregate types and co tio . S is not s n as a significan actor. However, the APA rutting r th udy n a ely n w ran .3 to m re t p of test ma sign ce o e inf cing factors. The good A in lts ll te ty ic sug irements aggregate properties such as 20 percen mu F&E ) an rce m f .A a ion too ect c on ate t 6 ions ow c ctio l th an de sa ory ng anc 5 A MO LU A sse the atu , t nam dulus is one of the to s s by CH 9-1 rch 4) for predicting rutting resistance although m sea s (92-93) h stio he ac y of g th r rutting p n. sect wil t th amic ulus esu xam he i ng ors o e t ts. T ffect mpac lev dy c modulus test results will be ted e ef enes e d modulus test f ti tting st l be ussed odul was ucte a tota 32 m , in h 15 mixtures were designed at the 100 gyration level, 15 mixtures were designed at the 65 gyration level, and 2 mixtures were designed with 40 gyrations. Three replicates were conducted for each mixture. The individual dynamic modulus test results are shown in Appendix Table D1. Table D1 includes the dynamic modulus and phase angle results under the dynamic stress at a series of load frequencies of 25Hz, 10Hz, 5Hz, 1Hz, 0.5Hz mpac n levels NMA how t f esults in is st are i relativ arro ge (2 6.1 m ) therefo he recision the may sk the ifican f thes luen PA rutt g resu for a aggrega pes ind ate the gested requ for t maxi m for (3:1 d 30 pe nt aximum or L bras may be stringent. The discussion on the eff of ompacti icind s tha 5 gyrat is a l ompa n e lev at still c provi tisfact rutti resist e. .2 DYN MIC DU S s discu d in liter re review he dy ic mo test p test elected the N RP 9 resea ers (8 any re rcher ave que ned t curac usin is test fo redictio This ion l presen e dyn mod test r lts and e ine t nfluenci fact n th est resul he e of co tion el on the nami evalua a hnd t fectiv s h of t ynamic or predic ng ru resi ance wil disc . The dynamic m us test cond d for l of ixtures whic 222 and 0.1Hz. At a high test temperature of 60?C, the higher complex modulus E* indicates the stiffer m s, theref higher deformation resistance; the lower phase a ndi s m last alt re, th re qu r re nd p nt d ma Th oun x E under high t ure een suggested ( o be as icat f rutt sista The gr 12 NM ixtu igne 40 ns d u ed dy c m val he a e dy c m alu r 12.5 mm ruby granite m s a 785 119 a at for an gyrations, respectively. The high value for the 4 tions igne re w ticed during the t and ca take test c ition ing t ure loa con s we e sam requ r t f th . A t xtu as d d an luat supp tal mixtures after testing all the othe ture s c ed, mple preparati llowed the procedure, and samp ize voi tents is m e w in t allowable range. To examin ossi tchi blem as ten g n w eter ed b duct e ig oven and nal he i tio ults wed the correct asphalt content and gradations were used. The asphalt content was within 0.1 percent fro e designed value, and the values of percent passing each sieve were within a reasonable range of 1-2 percent from the gradation sults for 65 gyration samples. Since only two mixtures were designed with 40 gyrations, test results for these two mixtures were not included in the following ANOVAs and discussions. asphalt ixture ore ngle ? i cate ore e ic asph mixtu erefo icke covery a less ermane efor tion. e comp d inde */sin? emperat has b 84) t used an ind ion o ing re nce. ruby anite .5 mm AS m re des d with gyratio showe nexpect high nami odulus ues. T verag nami odulus v es fo ixture re 910, , and 9 MP 10Hz 100, 65 d 40 0 gyra des d mixtu as no est, extra re was n to ensure the ond s includ test emperat and ding dition re th e as ired fo he rest o e test lthough his mi re w esigne d eva ed as lemen r mix s wa omplet the sa on fo same le s and air d con for th ixtur ere with he e the p ble ba ng pro , the phalt con t and radatio ere d min y con ing th nition test sieve a ysis. T nvestiga n res sho m th re 223 TABL Source DF Seq SS Adj SS Adj MS F P E 5.3 ANOVA for Dynamic Modulus Agg. 4 5153812 5153812 1288453 24.2 0.000 NMAS 2 893965 893965 446982 8.4 0.000 Gyrs 1 782476 782476 782476 14.7 0.000 Hz 5 16003526 16003526 3200705 60.12 0.000 Agg.*NMAS 8 1961960 1961960 245245 4.61 0.000 Agg.*Gyrs 4 948451 948451 237113 4.45 0.002 Agg.*Hz 20 454733 454733 22737 0.43 0.987 NMAS*Gyrs 2 237551 237551 118775 2.23 0.109 NMAS*Hz 10 94580 94580 9458 0.18 0.998 Gyrs*Hz 5 51347 51347 10269 0.19 0.965 Agg.*NMAS*Gyrs 8 2760676 2760676 345084 6.48 0.000 Agg.*NMAS*Hz 40 68749 68749 1719 0.03 1.000 Agg.*Gyrs*Hz 20 16681 16681 834 0.02 1.000 NMAS*Gyrs*Hz 10 52411 52411 5241 0.1 1.000 Agg.*NMAS*Gyrs*Hz 40 166726 166726 4168 0.08 1.000 Error 360 19166711 19166711 53241 Total 539 48814354 Initial analysis of the dynamic modulus test results was performed by conducting two ANOVAs to evaluate the effect of the main factors (aggregate type, NMAS, gyration level, and test frequency) and any interactions between the main factors on dynamic modulus and phase angle. These two ANOVAs a in T les 5.3 d 5.4. TAB ANOVA for n e DF q d dj M F P re shown ab an LE 5.4 Phase A gle Sourc Se SS A j SS A S Agg. 4 4. 4 8. 18. .000 71 12 71 .12 17 53 16 0 NMAS 2 .1 28.09 2.86 .059 56 8 56.18 0 Gyrs 1 3. 103. 10.52 0.001 10 4 103.4 4 Hz 5 86 717 683 .000 335 .33 33586.33 6 .27 .25 0 Agg.*NMAS 8 5. 5 68. 6.94 .000 54 77 54 .77 22 0 Agg.*Gyrs 4 1. 1 80.39 8.18 0.000 32 55 32 .55 Agg.*Hz 20 5.73 165.73 8.29 0.84 0.661 16 NMAS*Gyrs 2 4.24 4.24 2.12 0.22 0.806 NMAS*Hz 10 21.31 21.31 2.13 0.22 0.995 Gyrs*Hz 5 1.25 1.25 0.25 0.03 1.000 Agg.*NMAS*Gyrs 8 325.1 325.1 40.64 4.13 0.000 Agg.*NMAS*Hz 40 101.11 101.11 2.53 0.26 1.000 Agg.*Gyrs*Hz 20 29.49 29.49 1.47 0.15 1.000 NMAS*Gyrs*Hz 10 21.51 21.51 2.15 0.22 0.995 Agg.*NMAS*Gyrs*Hz 40 68.27 68.27 1.71 0.17 1.000 Error 360 3539.26 3539.26 9.83 Total 539 39604.63 224 For dynamic modulus, all the main factors were significant and the interaction between aggregate type and NMAS, aggregate type and compaction level, and interaction among aggregate types, NMAS and compaction level were also si gnificant. As shown in Tables 5.3 and 5.4, the load frequency is the most significant factor indicated by the highest F statistic value of 60 and 683, respectively. The effect of load frequency will be discussed first and all the other significant factors will be discussed later. The average dynamic modulus results with different test frequencies are shown in Figure 5.8. One standard deviation is shown as error bar. As shown in Figure 5.8, the standard deviation values for each load frequency depended on the average modulus values, an average COV of 19.8 percent was determined. This average COV value is not surprising because the test variability is expected to be higher for SMA mixture than dense-graded mixture. SMA is a gap-graded mix with a high percentage of coarse aggregates. With the sa e sample size as dense-graded m ause the high po f ture er m ix, SMA mixture is likely to be more variable bec tential for segregation within a sample. As expected, the dynamic modulus increased with an increase of loading frequency for all tested samples. With a decrease o load frequency, the mixtures tend to have less stiffness according to the time-tempera superposition mechanism, therefore lower dynamic modulus values occur at these low frequencies. 225 y = 729.43x 0.1075 R 2 = 0.9407 0.0 0.01 0.1 1 10 100 Load Frequency, Hz 200.0 ver 400.0 ag e 600.0 .0 1400.0 A D y n a m i c M o d u P 800 1000.0 l u s, M 1200.0a Temp = 60 ? C N = 96 or type, NMAS and compaction level. The average phase angles with different test frequencies are shown in Figure 5.9. One s frequen FIGURE 5.8 Average dynamic modulus versus load frequency. The ANOVA results (Table 5.4) for phase angle were similar to those results f dynamic modulus. All the main factors except NMAS were significant and there were significant interactions between aggregate type and NMAS, aggregate type and compaction level, and between aggregate tandard deviation is shown as error bar. The pooled standard deviation for all load cies is 2.2 degrees. This standard deviation is compatible to the result from other studies (84, 86), where the standard deviation for phase angle was reported from 1.8 to 2.3 degrees. 226 y = 20.905x 0. R 0.9549 5.0 10.0 5.0 20.0 25.0 30.0 35.0 40.0 45.0 A ver ag e g r ee 1605 2 = 0.0 1 e P h 0.01 0.1 1 10 100 Load Frequency, Hz ase A n g l e , d Temp = 60 ? C N = 96 As expected, the phase angle increased with the increase of loading frequency for all tested samples. Similar phenomenon had also been reported by other researchers (87, 112). However, this trend is not usually observed at the low or normal test temperature, such as 5?C and 25?C. This phenomenon can be explained by the mechanism of composite material. The SMA mixture which had a good aggregate skeleton structure can be seen as a composite material that is composed of aggregate structure and asphalt mortar. The decrease of loading frequency, which gives the same effect as an increase in test temperature, will cause the asphalt mortar to have a small elastic component and become more viscous. At a high test temperature, when the load frequency is lowered to a certain level, the aggregate structure becomes dominant within the whole mix structure. With the confining test condition, the aggregate structure is close to becoming elastic. Therefore, with a decrease of loading frequency, the aggregate structure within the mixtures becomes more dominant, and increases the elastic component of the whole FIGURE 5.9 Average phase angle versus load frequency. 227 structure, which resulted in the lower phase angle. The decrease of phase angle with the decrease of loading frequency, to some extent, indicates that a strong internal structure existed. To better explain this, the elastic and viscous parts of complex modulus were calculated based on Equation 3.11, and are shown in Figure 5.10. One standard deviation is shown as error bar. Since the standard deviation values depended on modulus values, average COV values of 20.4 and 19.3 percent were determined for storage and loss modulus, respectively. 0 200 Load Frequency, Hz 100 300 400 700 800 1100 25 10 5 1 0.5 0.1 M odu , M p a 500 600 lu s 900 1000 Storage Modulus, E' Loss Modulus E" Temp = 60 ? C N = 96 FIGURE 5.10 Average storage and loss modulus at different frequencies. loss nd quickly approached zero with the decrease of load frequency. The storage modulus maintained at a certain level indicates the aggregate skeleton became dominant and this structure is less dependent of load frequency. The fact As shown in Figure 5.10, both the storage and loss modulus decreased with the decrease of load frequency. However, the decrease rate of storage modulus slowed down with the decrease of load frequency and maintained nearly a constant level, while the modulus had a higher decrease rate a 228 that loss modulus approached zero indicates the role of asphalt binder becomes negligible at low frequency and high temperature. As shown in Tables 5.3 and 5.4, test frequency is the only testing condition variable in the ANOVA analysis and the most significant factor as indicated by the highest F statistics and lowest P value. Therefore, the test results are demonstrated and discussed separately regarding different frequencies. Since the 10 Hz and 0.1 Hz are typically used to simulate the high speed and low speed vehicle loading (84, 87), these two frequencies will be analyzed in more detail. The test results for all designed mixtures are presented in Tables 5.5 and 5.6 for the load tion (COV) frequency of 10 Hz and 0.1 Hz. The average value and coefficient of varia of dynamic modulus E*, phase angle ? and E*/sin? are included in the tables. Comparing these two tables, the COV values for test results of 0.1 Hz are generally higher than those of 10 Hz. This might be due to the lower value for both dynamic modulus E* and phase angle ? in lower frequency. The same variation in test results measurement and data processing procedure gave the lower test value a higher COV value. 229 TABL E*, MPa ? , degree E*/sin? , MPa E 5.5 Dynamic Modulus Test Results at Load Frequency of 10 Hz Agg. Type NMAS Gyrs Average COV,% Average COV,% Average COV,% C.GVL 19 100 773.3 4.4 26.8 5.9 1720.1 9.5 C.GVL 12.5 100 834.2 24.4 29.4 2.6 1691.8 22.3 C.GVL 9.5 100 983.3 17.6 27.6 9.4 2147.5 24.4 L.GRN 19 100 894.0 17.6 28.1 7.6 1900.6 17.3 L.GRN 12.5 100 1248.1 10.1 27.4 5.3 2724.1 14.5 L.GRN 9.5 100 1120.0 19.7 30.9 8.7 2189.9 21.6 LMS 19 100 968.7 17.0 29.3 15.9 2038.3 28.8 LMS 12.5 100 1206.6 18.6 30.4 4.1 2384.2 17.3 LMS 9.5 100 1084.5 50.3 28.8 10.6 2339.1 55.2 R.GRN 19 100 865.9 22.4 32.0 5.1 1639.9 23.0 R.GRN 12.5 100 910.3 11.1 31.7 11.0 1732.4 5.5 R.GRN 9.5 100 893.0 31.9 32.9 5.8 1631.1 27.1 TRAP 19 100 820.8 19.7 27.5 3.6 1784.3 21.5 TRAP 12.5 100 900.3 4.1 25.8 1.7 2066.7 4.7 TRAP 9.5 100 1098.7 33.3 27.1 11.3 2492.2 45.3 C.GVL 19 65 615.8 9.7 29.6 16.5 1283.2 26.2 C.GVL 12.5 65 717.8 59.6 31.0 17.8 1498.1 66.9 C.GVL 9.5 65 824.7 10.7 31.6 2.0 1573.4 11.0 L.GRN 19 65 1117.9 25.5 27.5 7.6 2457.6 33.4 L.GRN 12.5 65 1024.9 29.0 30.9 7.8 2022.2 34.9 L.GRN 9.5 65 761.5 31.6 32.5 8.9 1446.5 37.2 LMS 19 65 1112.7 3.7 28.8 2.2 2310.1 4.5 LMS 12.5 65 1009.4 51.5 29.6 21.2 2262.8 72.7 LMS 9.5 65 1214.4 23.1 28.9 9.3 2534.2 27.6 R.GRN 19 65 852.8 22.3 30.3 5.1 1686.7 19.7 R.GRN 12.5 65 785.0 14.0 32.1 6.9 1489.3 19.4 R.GRN 9.5 65 1020.7 20.0 28.4 9.5 2146.0 17.6 TRAP 19 65 573.5 15.5 29.9 13.1 1159.8 16.3 TRAP 12.5 65 976.9 21.4 26.7 10.3 2169.5 15.4 TRAP 9.5 65 656.7 40.1 30.0 11.8 1357.7 46.7 L.GRN 12.5 40 1126.5 9.4 25.9 0.8 2582.6 10.1 R.GRN 12.5 40 1198.8 9.1 22.5 11.4 3167.2 17.8 Overall Average: 943.5 21.8 29.1 8.5 1988.4 25.5 230 TABL E*, MPa ? , degree E*/sin? , MPa E 5.6 Dynamic Modulus Test Results at Load Frequency of 0.1 Hz Agg. Type NMAS Gyrs Average COV,% Average COV,% Average COV,% C.GVL 19 100 561.0 10.3 11.8 4.1 2757.8 13.2 C.GVL 12.5 100 537.9 30.4 12.6 11.5 2444.2 24.2 C.GVL 9.5 100 647.4 21.1 12.0 5.7 3147.5 26.2 L.GRN 19 100 601.2 28.8 15.6 15.1 2308.0 37.7 L.GRN 12.5 100 893.7 19.0 14.2 13.7 3747.5 30.1 L.GRN 9.5 100 666.2 21.2 16.9 20.5 2352.0 29.4 LMS 19 100 707.7 32.5 14.4 15.6 2940.9 40.9 LMS 12.5 100 683.0 27.9 18.3 21.5 2251.1 34.9 LMS 9.5 100 642.1 58.2 16.2 23.6 2573.1 68.6 R.GRN 19 100 527.4 28.9 17.7 9.1 1759.3 31.8 R.GRN 12.5 100 529.6 4.7 17.2 9.9 1795.4 5.9 R.GRN 9.5 100 596.5 40.0 19.4 16.4 1760.7 26.1 TRAP 19 100 539.2 34.7 14.4 3.4 2161.0 33.0 TRAP 12.5 100 602.3 6.8 13.2 5.5 2656.8 11.5 TRAP 9.5 100 803.1 36.1 13.3 19.5 3782.2 58.2 C.GVL 19 65 422.1 27.1 14.1 43.4 2271.4 82.5 C.GVL 12.5 65 519.8 66.4 12.8 43.9 2932.3 74.4 C.GVL 9.5 65 547.9 7.2 15.4 10.1 2092.5 17.8 L.GRN 19 65 728.0 31.0 14.9 14.5 2964.9 46.9 L.GRN 12.5 65 713.2 29.4 16.5 10.7 2568.7 39.0 L.GRN 9.5 65 401.5 45.7 19.4 3.7 1212.1 45.4 LMS 19 65 717.3 21.2 16.8 4.6 2487.7 21.5 LMS 12.5 65 657.5 71.1 18.0 31.9 2731.6 104.3 LMS 9.5 65 866.9 27.1 14.5 3.1 3481.7 29.8 R.GRN 19 65 558.6 17.1 15.2 9.9 2115.8 8.4 R.GRN 12.5 65 485.6 34.6 20.0 17.3 1507.2 49.9 R.GRN 9.5 65 638.9 27.6 15.7 23.7 2512.8 40.8 TRAP 19 65 321.7 23.0 18.2 14.8 1037.7 22.9 TRAP 12.5 65 716.5 28.8 12.5 7.1 3291.6 25.2 TRAP 9.5 65 406.1 47.4 17.0 24.9 1540.7 59.0 L.GRN 12.5 40 798.7 0.5 13.2 2.1 3496.4 1.8 R.GRN 12.5 40 906.7 12.1 10.5 18.9 5142.3 26.5 Overall Average: 623.3 28.7 15.4 15.0 2557.0 36.5 and E*/sin are 21.8, 8.5 and 25.5 percent, respectively. At 0.1 Hz load frequency, the average COV for dynamic modulus, phase angle and E*/sin? are 28.7, 15.0 and 36.5 percent, respectively. At 10 Hz load frequency, the average COV for dynamic modulus, phase angle ? 231 Some extremely high or low values might adversely affect the analysis of the results, and are seen as outliers. The outliers were determined based on the E*/sin? data set wit test h COV higher than 50 percent. This criterion is a round-up value from overall average COV plus one standard deviation of COV values. Within these data sets, the test results that had the biggest difference from the average were considered as potentially erroneous and were excluded from the analysis data set. The average test values for these mixtures were from the average of two remaining individual test results instead of all three replicates. A total of 6 samples (out of total 96 samples) were determined as outliers following the criteria described above. After removing these outliers, the average COV decreased. The new average and COV values are shown in Tables 5.7 and 5.8. The following data analyses are based on the data set after the removal of the outliers. At 10 Hz load ? are 16.5, 6 . re since n er studies (84, 86). For loa frequency, the average COV for dynamic modulus, phase angle and E*/sin .4 and 18.6 percent, respectively. At 0.1 Hz load frequency, the average COV for dynamic modulus, phase angle and E*/sin? are 22.1, 10.5 and 26.1 percent, respectively The test variability is expected higher for SMA mixture than dense-graded mixtu SMA is a gap-graded mix with high percentage of coarse aggregates. With the same sample size as dense-graded mix, SMA mixture is likely to be more variable because the high potential for segregation. The test precision for dynamic modulus has not yet bee developed, however, the COV in this study is similar to that from oth d frequency of 10 Hz, Witczak et al (84) and Bonaquist et al (86) reported the average COV values for E* were 15.2 and 13.0 percent, respectively, and the standard deviation values for phase angle were 2.3 and 1.8 degrees, respectively. 232 TABLE 5.7 Dynamic Modulus Test Results at Load Frequency of 10 Hz (Without E* ? E*/sin? 6 Outlier Samples) Agg. NMAS Gyrs Type Average COV,% Average COV,% Average COV,% C.GVL 19 100 773.3 4.4 26.8 5.9 1720.1 9.5 C.GVL 12.5 100 834.2 24.4 29.4 2.6 1691.8 22.3 C.GVL 9.5 100 983.3 17.6 27.6 9.4 2147.5 24.4 L.GRN 19 100 894.0 17.6 28.1 7.6 1900.6 17.3 L.GRN 12.5 100 1248.1 10.1 27.4 5.3 2724.1 14.5 L.GRN 9.5 100 1120.0 19.7 30.9 8.7 2189.9 21.6 LMS 19 100 968.7 17.0 29.3 15.9 2038.3 28.8 LMS 12.5 100 1206.6 18.6 30.4 4.1 2384.2 17.3 LMS 9.5 100 1398.1 4.9 27.0 2.3 3079.7 7.0 R.GRN 19 100 865.9 22.4 32.0 5.1 1639.9 23.0 R.GRN 12.5 100 910.3 11.1 31.7 11.0 1732.4 5.5 R.GRN 9.5 100 893.0 31.9 32.9 5.8 1631.1 27.1 TRAP 19 100 820.8 19.7 27.5 3.6 1784.3 21.5 TRAP 12.5 100 900.3 4.1 25.8 1.7 2066.7 4.7 TRAP 9.5 100 889.6 8.5 28.8 2.0 1843.1 6.7 C.GVL 19 65 582.9 4.3 32.3 6.5 1094.7 10.1 C.GVL 12.5 65 939.4 28.5 27.8 1.9 2020.7 30.1 C.GVL 9.5 65 824.7 10.7 31.6 2.0 1573.4 11.0 L.GRN 19 65 1117.9 25.5 27.5 7.6 2457.6 33.4 L.GRN 12.5 65 1024.9 29.0 30.9 7.8 2022.2 34.9 L.GRN 9.5 65 761.5 31.6 32.5 8.9 1446.5 37.2 LMS 19 65 1112.7 3.7 28.8 2.2 2310.1 4.5 LMS 12.5 65 718.6 25.3 33.1 6.9 1327.5 31.1 LMS 9.5 65 1214.4 23.1 28.9 9.3 2534.2 27.6 R.GRN .7 19 65 852.8 22.3 30.3 5.1 1686.7 19 R.GRN 12.5 65 785.0 14.0 32.1 6.9 1489.3 19.4 R.GRN 9.5 65 1020.7 20.0 28.4 9.5 2146.0 17.6 TRAP 19 65 573.5 15.5 29.9 13.1 1159.8 16.3 TRAP 12.5 65 976.9 21.4 26.7 10.3 2169.5 15.4 TRAP 9.5 65 808.0 3.8 28.1 4.1 1720.2 7.6 L.GRN 12.5 40 1126.5 9.4 25.9 0.8 2582.6 10.1 R.GRN 12.5 40 1198.8 9.1 22.5 11.4 3167.2 17.8 Overall Average: 948.3 16.5 29.2 6.4 1983.8 18.6 233 TABLE 5.8 Dynamic Modulus Test Results at Load Frequency of 0.1 Hz (Without 6 Outlier Samples) E* ? E*/sin? Agg. Type NMAS Gyrs Average COV,% Average COV,% Average COV,% C.GVL 19 100 561.0 10.3 11.8 4.1 2757.8 13.2 C.GVL 12.5 100 537.9 30.4 12.6 11.5 2444.2 24.2 C.GVL 9.5 100 647.4 21.1 12.0 5.7 3147.5 26.2 L.GRN 19 100 601.2 28.8 15.6 15.1 2308.0 37.7 L.GRN 12.5 100 893.7 19.0 14.2 13.7 3747.5 30.1 L.GRN 9.5 100 666.2 21.2 16.9 20.5 2352.0 29.4 LMS 19 100 707.7 32.5 14.4 15.6 2940.9 40.9 LMS 12.5 100 683.0 27.9 18.3 21.5 2251.1 34.9 LMS 9.5 100 856.9 5.3 14.2 14.2 3553.1 19.2 R.GRN 19 100 527.4 28.9 17.7 9.1 1759.3 31.8 R.GRN 12.5 100 529.6 4.7 17.2 9.9 1795.4 5.9 R.GRN 9.5 100 596.5 40.0 19.4 16.4 1760.7 26.1 TRAP 19 100 539.2 34.7 14.4 3.4 2161.0 33.0 TRAP 12.5 100 602.3 6.8 13.2 5.5 2656.8 11.5 TRAP 9.5 100 637.1 7.6 14.7 6.1 2518.9 13.6 C.GVL 19 65 361.0 17.0 17.7 3.4 1193.4 20.2 C.GVL 12.5 65 699.6 30.1 9.6 12.8 4154.6 17.7 C.GVL 9.5 65 547.9 7.2 15.4 10.1 2092.5 17.8 L.GRN 19 65 728.0 31.0 14.9 14.5 2964.9 46.9 L.GRN 12.5 65 713.2 29.4 16.5 10.7 2568.7 39.0 L.GRN 9.5 65 401.5 45.7 19.4 3.7 1212.1 45.4 LMS 19 65 717.3 21.2 16.8 4.6 2487.7 21.5 LMS 12.5 65 392.9 33.7 21.2 5.3 1095.2 38.4 LMS 9.5 65 866.9 27.1 14.5 3.1 3481.7 29.8 R.GRN 19 65 558.6 17.1 15.2 9.9 2115.8 8.4 R.GRN 12.5 65 485.6 34.6 20.0 17.3 1507.2 49.9 R.GRN 9.5 65 638.9 27.6 15.7 23.7 2512.8 40.8 TRAP 19 65 321.7 23.0 18.2 14.8 1037.7 22.9 TRAP 12.5 65 716.5 28.8 12.5 7.1 3291.6 25.2 TRAP 9.5 65 517.0 2.7 14.5 3.3 2063.2 5.9 L.GRN 12.5 40 798.7 0.5 13.2 2.1 3496.4 1.8 R.GRN 12.5 40 906.7 12.1 10.5 18.9 5142.3 26.5 Overall Average: 623.7 22.1 15.4 10.5 2517.9 26.1 234 The following data analyses are based on the data set after the removal of the outliers. General linear model analysis was used instead of ANOVA since the data set is not bal gle ? n actors (aggregate type, NMAS, and compaction level) and any ults Source DF Seq SS Adj SS Adj MS F P anced after removing the outliers. Since the dynamic modulus E* or phase an alone is not sufficient as an indication of rutting performance, the following analyses o the dynamic modulus test results will focus on the compound index E*/sin? under load frequencies of 10 Hz and 0.1 Hz. The test parameter E*/sin? was recommended as an indication for rutting resistance from the NCHRP study (84). A higher E*/sin? value indicates a stiffer and more elastic mixture, therefore more rutting resistance. A GLM analysis was conducted to evaluate the effect of the main f interactions between the main factors on 10 Hz E*/sin? results. The GLM res are shown in Table 5.9. TABLE 5.9 GLM Results on E*/sin? at Load Frequency of 10 Hz Agg 4 4482858 4397433 1099358 6.27 0.000 NMAS 2 811976 938666 469333 2.68 0.078 Gyrs 1 794780 1060275 1060275 6.05 0.017 Agg*NMAS 8 4001692 4522435 565304 3.22 0.005 Agg*Gyrs 4 648364 716546 179136 1.02 0.405 NMAS*Gyrs 2 311698 246147 123073 0.70 0.500 Agg*NMAS*Gyrs 8 4109750 4109750 513719 2.93 0.009 Error 54 9470151 9470151 175373 Total 83 24631269 As shown in Table 5.9, the aggregate type and compaction level are significant factors. The interaction between aggregate type and NMAS, and the interaction among all three main factors are significant. 235 The average E*/sin? values for different aggregate types and two compaction levels are shown in Figure 5.11. One standa 2 represent th be sam ed erag ach rd deviation is shown as error bar. N 1 and N e num r of ples us for av e for e group. 1000.0 .0 .0 .0 .0 .0 VL R TRA 65 3000 1400 1800 2200 2600 C.G L.GRN LMS .GRN P 100 Agg Type Gregate yrations A v e r a g e E * / s i n ? a t 1 0 H z , M P a N 1 2 = UR .11 e E ? results at frequency of 10 Hz. s s n in ure */si pend aggre pe. estone had the highest average E*/sin? f 2 Pa, crus vel gran d tr ad ilar average E*/sin? v of a 750 M n th nd. T li an by ite her n? v hic be e ed i these two aggregates had the lowest VMA values (average 17.1 percent versus average 19 ent oth ee tes the metho d th tigh aggregate structures. comp ve 100 gyra esu an i e in E*/sin? value from 1837 to 2019 MPa (9.9%). A paired t-test was conducted to compare = 17, N 42 FIG E 5 Averag */sin how Fig 5.11, E n? de s on gate ty The limA value o 288 M while hed gra , ruby ite an alues bout 1 Pa i e low e he aprock h sim E*/si alues w h may xplain n thatmestone d ru gran had hig .5 perc for er thr aggrega , with SSD d) an erefore ter A decrease in action le l from to 65 tions r lted in ncreas 236 the me TABLE an E*/sin? value for the mixtures designed with 65 and 100 gyrations, and the results are shown in Table 5.10. 5.10 Pair T-Test Results on E*/sin? Value of Two Compaction Levels N Mean StDev SE Mean 100 Gyrs 15 2038.25 420.14 108.48 65 Gyrs 15 1810.56 463.45 119.66 Difference 15 227.69 509.69 131.60 95% CI for mean difference: (-54.570, 509.943) T-Test of mean difference = 0 (vs not = 0): T-Value = 1.73, P-Value = 0.106 As shown in Table 5.10, there is no significant difference between E*/sin? values of mixtures designed with these two compaction levels at a significant level of 95 percent. The mean E*/sin? values shown in Table 5.10 are slightly different from the values shown in Figure 5.11. That is due to m ue in T le 5.10 the mean values of 15 mixture nd e u a re es a oving 6 outliers, while the average values sho u re era ge of ean val s shown ab are s, a ach mixt re may h ve 2 or 3 plicat fter rem wn in Fig re 5.11 a the ov ll avera individual samples. 0.0 500.0 1500.0 2000.0 2500.0 3000.0 A ag e E /si n ? 10 H , MP a 1000.0 3500.0 4000.0 C.GV L L.GRN LMS R.GRN TRA P ver * at z 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 FIGURE 5.12 Interaction between aggregate type and NMAS on E*/sin? results at frequency of 10 Hz. 237 The interaction between aggregate type and NMAS on E*/sin? at 10 Hz is shown in Figure 5.12. One standard deviation is shown as error bar. The effect of NMAS varied for different aggregate types. Again, for most aggregate the differences within th ree NMAS are not large. A GLM a actors (aggregate type, NMAS, and compaction level) and any interactions between the main factors on 0.1 Hz E*/sin results. The GLM results are shown in Table 5.11. TABLE 5.11 GLM Results on E*/sin? at Load Frequency of 0.1 Hz nalysis was conducted to evaluate the effect of the main f ? Source DF Seq SS Adj SS Adj MS F P Agg 4 6850333 6488918 1622229 2.98 0.027 NMAS 2 1803832 2222765 1111382 2.04 0.140 Gyrs 1 1252325 1740256 1740256 3.20 0.079 Agg*NMAS 8 20764608 22422769 2802846 5.15 0.000 Agg*Gyrs 4 2029754 2025696 506424 0.93 0.453 NMAS*Gyrs 2 417676 576286 288143 0.53 0.592 Agg*NMAS*Gyrs 8 14399518 14399518 1799940 3.31 0.004 Error 54 29392371 29392371 544303 Total 83 76910416 As shown in Table 5.11, only aggregate type is a significant main factor for E*/sin? at load frequency of 0.1 Hz at a significant level of 95 percent. The interaction between aggregate type and NMAS, and the interaction among three main factors are also significant. The average E*/sin? values for different aggregate types are shown in Figure 5.13. One standard deviation is shown as error bar. As shown in Figure 5.13, the ruby granite had the lowest average E*/sin? value of 1900 MPa, while crushed gravel, lab granite, and limestone had the similar average E*/sin? values of about 2600 MPa. The low E*/sin? value for traprock and ruby granite is probably due to the relatively high asphalt content for these two aggregates (average 6.6 percent versus average 6.0 percent for the other 238 three aggregates). The high asphalt content increased the effect of asphalt binder under high temperature and low load frequency. 0.0 .GRN LMS 500.0 1000.0 1500.0 2000.0 2500.0 3000.0 3500.0 4000.0 C.GVL L R.GRN TRAP A v e r ag e E * / s i n ? at 0. 1 H z , M P a N = 17 FIGURE 5.13 Average E*/sin? results at frequency of 0.1 Hz. 0.0 500 A .0 0.0 500.0 2000.0 00.0 5000.0 L 1 H a 100ver 1 ag 25? at 3000.0 0 . 3500.0 4000.0 4500.0 z , MP C.GV L.GRN LMS R.GRN T PRA e E * /s in 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 FIGURE 5.14 Interaction between aggregate type and NMAS on E*/sin? results at shown frequency of 0.1 Hz. The interaction between aggregate type and NMAS on E*/sin? at 0.1 Hz is in Figure 5.14. One standard deviation is shown as the error bar. For different aggregate 239 types, the average E*/sin? followed the different orders for the three NMAS. 12.5 mm NMAS mixtures for limestone and ruby granite had the lowest E*/sin? value for the three NMAS the MAS on E*/sin? value. It is believed that showing that these two propertie have an interaction is simply due to normal variation of the test results. mixtures, while for the other three aggregates, 12.5 mm NMAS mixtures had highest E*/sin? value for the three NMAS mixtures. There appears to be no scientific reason for the combination effects of aggregate type and N s y = 1.6316x - 738.07 R = 0.7249 0.0 1000.0 2000.0 0.0 500.0 1000.0 1500.0 2000.0 2500.0 3000.0 3500.0 4000.0 E*/sin (?) at 10 Hz, Mpa E * /s ( ? ) at 0. 1 H M p a 2 3000.0 4000.0 5000.0 6000.0 in z , Equality line y=x FIGURE 5.15 The comparison of E*/sin? between 10 Hz and 0.1 Hz. A comparison on E*/sin? value between 10 Hz and 0.1 Hz is shown in Figure 5.15. The E*/sin? value at 0.1 Hz is generally larger than those at 10 Hz. This is due to the combination effects of decreasing E* and the phase angle ? value with the decrease of load frequency. However, this indicates that at the test temperature of 60?C, the E*/sin? may give an erroneous prediction about the rutting resistance because it is believed that mixture will have lower rutting resistance at lower load frequency. 240 It is also observed that the difference between two load frequencies increases with the increase of test values. This is likely due to the enlarged effect of phase angle changes when the phase angle is smaller. For example, when the phase angle changes from 10 to 11 degrees, the E*/sin? value changes about 9 percent, however when the phase angle changes from 30 to 31 degrees, the E*/sin? value changes only 3 percent if the E* stays constant. The theoretical basis of using the E*/sin? value to predict the rutting resistance relates to the assumption that it gives better protection against rutting at high temperature minimized in two ways: by having a stiffer binder or by having a lower phase angle value in the binder, i.e. more elastic behavior. However, the phase angle decreased with the the theoretical formulation of minimizing dissipated energy, as it is used for the stiffness factor G*/sin? in the binder specification, is not valid for the asphalt mixtures throughout the entire range of mixture performance. A study (87) on use of stiffness as a simple performance test concluded that the E*/sin? value correlated with field rut depth best at high temperature (54.4?C) and relatively high load frequency (5 Hz). When the load frequency goes lower, the correlation between E*/sin? and permanent deformation decreases. than the modulus alone by minimizing dissipated energy (87). Dissipated energy can be decrease of load frequency at high temperatures because the elastic response from the aggregate skeleton overpowered the viscous influence of the binder in the mixtures. Thus, 241 FIGURE 5.16 The correlations between field rut depths and E*/sin? values (84). NCHRP project 9-19 (84) presented several correlations between E*/sin? and field rut depths based on the field data from MnRoad, ALF and Westrack test sections. These correlations are shown in Figure 5.16. It is noteworthy that there are some significant differences in test conditions between the NCHRP 9-19 project and this study. The correlations in Figure 5.16 were developed under unconfined and 130?F test temperature conditions. The unconfined test condition at high temperature is not practical 242 for SMA mixtures because the test samples with high asphalt content tend to slum therefore have high test variability. Also, the unconfined test condition does not represent the real situation in the field. p and The E*/sin? results for SMA mixtures designed with 100 and 65 gyrations under 6 psi) and 1800 MPa (0.26 ?10 psi). These values are higher than the maximum value for prediction if we directly put these values in the correlations developed by NCHRP 9-19 researchers. As we discussed above, the lower temperature (130?F versus 140?F) and lower load frequency (5 Hz versus 10 Hz) are likely to get even higher E*/sin? values based on the results from this study. Therefore the SMA mixtures designed with both 100 and 65 gyrations are likely to be rutting resistant and have less than 10 mm field rut depth if the difference between confined and unconfined test conditions is not considered. The correlation between E*/sin? results at the two load frequencies and APA rutting results are shown in Figure 5.17. The regressions shown in Figure 5.17 indicate there is no correlation between E*/sin? tribute to this close to zero correlation. The APA rutting results confined test conditions of 140?F and 10 Hz are about 2000 MPa (0.29 ?10 6 at either load frequency and the APA rut depths. This result is not surprising because as discussed above, the E*/sin? is not a good indication for rutting resistance at high temperature and low load frequency. The narrow range of data may also con indicate only 3 out of 32 mixtures showed significant rutting, which is more than 5 mm. 243 10 Hz: y = 56.011x + 1804.2 R 2 = 0.0136 0.1 Hz: y = 141.9x + 2024.2 0.0 / s ), R 2 = 0.0237 1000.0 2000.0 3000.0 4000.0 5000.0 6000.0 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 APA Rut Depth, mm E * in ( ? M p a 10 H z 0.1 Hz Linear (10 Hz) Linear (0.1 Hz) FIGURE 5.17 The relationships between E*/sin? and APA rut depths. In summary, compaction level was not shown as a significantly influencing factor for the E*/sin? value at 10 Hz (Table 5.10) and 0.1 Hz (Table 5.11). The E*/sin? value seems to depend on VMA values, or the tightness of aggregate structure. If this speculation is true, a dense-graded mixture will likely have a higher dynamic modulus value than a SMA mixture. The effect of stone-on-stone contact and the good resistance to high shearing force within a SMA mixture can?t be appropriately evaluated by the relatively low applied stress associated with dynamic modulus test. The E*/sin? value at high temperature showed contrary results when the load frequency changed, i.e. had higher value under lower load frequency. The E*/sin? value appears no longer appropriate as the indication for rutting resistance of SMA mixtures. Therefore, the effectiveness to evaluate rutting potential with the dynamic modulus test is indeed questionable. Many tests (30, 32) in the past have shown that in the laboratory dense-graded mixes have higher dynamic modulus values than SMA mixtures. 244 5.3 STATIC CREEP test has been used to evaluate HMA rutting potential for m s. However, it was often conducted under the unconfined test condition. Only a few studies have used confined ndi nd a l is ne d to sh c to differenti tin ptib t ne ic c test ect l pr the eep test results and ex th uen fact the lts eff m lev the s ep results will b ated and effectiveness c st fo predicting rutting resistance will be discussed. As discussed in the literature review (94-95), the static creep any year test co tions a dditiona work ede establi riteria ate a rut g susce le mix for he confi d stat reep . This s ion wil esent static cr amine e infl cing ors on test resu . The ect of co paction el on tatic cre test e evalu of static reep te r 0.001 0.01 0.1 1 10 100 10.0 Time, seconds LVDT A LVDT B LVDT C LVDT A S t r a in, % 0.1 1.0 100.0 1000.0 vg Ram FIGURE 5.18 A typical static creep test result with tertiary flow. The individual static creep test results are shown in Appendix Table D2. Table D2 includes the information about the slope and intercept within the secondary phase, the time to reach 1%, 2%, 3%, and 4% strain level, and the flow time if available. Tertiary 245 flow w , he as observed for only 7 samples out of 96 samples tested. For most of the samples no significant increase of strain slope was observed during the test. A typical test result showing tertiary flow is shown in Figure 5.18, and a typical test result showing no tertiary flow is shown in Figure 5.19. Tertiary flow is considered to exist when the displacement of the sample begins to increase quickly as shown on the right side of t curves in Figure 5.18. 0.00 0.01 0.10 1.00 100.00 in, % 10.00 0.1 1.0 10.0 100.0 1000.0 10000.0 Time, Seconds S t r a LVDT A LVDT B LVDT C LVDT Avg Ram FIGURE 5.19 A typical static creep test result without tertiary flow. The average slope in the secondary phase and the time to reach 4 percent strain are listed in Table 5.12. The time to reach 4 percent strain was converted from seconds to hours for easy reading. The static creep test results showed a very high variability, the overall COV for the time to reach 4 percent strain is 101.8 percent. Within three replicates finished, while another sample had less than 4 percent strain and a very flat slope of , one sample reached the LVDT limits in a few seconds and the test was 246 strain a vel at a certai t if te and estimate the strain level at 3600 se e strain n 4 nt. fter more than 4 hours loading and had to be stopped manually. The strain le n time (for example 3600 seconds) and creep modulus at a certain time are not included in the data analysis because high variability in test time makes it very difficul not impossible to normalize the strain level. For example, if a test reached the LVDT limits in 10 seconds, it is extremely hard to extrapola conds since a linear relationship does not likely exist. Instead of using th level at a certain time, the time to reach a specific strain was used. The high variability of the test time to reach a certain strain still resulted in difficulty in analyzing the test data, and much of the information had to be estimated. Since only a few samples were observed having tertiary flow, all the missing informatio was extrapolated based on the secondary slope and intercept. As shown in Table 5.12, the static creep test showed a very high variability as indicated by the overall average COV value of 101.8 percent for test time to reach percent strain. Due to the high COV, no additional analysis was directly conducted on the data for time to 4 percent strain. The slope of the secondary phase and the logarithmic value of the test time to reach 4 percent strain, however, showed more manageable COV values of 24.2 perce Therefore, data analysis was conducted on these two properties. 247 TABLE 5.12 Static Creep Test Results Summary Log Slope, 1/log(sec) Time at 4% strain T 0.04 , hrs Log T 0.04 , log(sec) Agg. Type Gyrs NMAS Average COV, % Average COV, % Average COV, % C.GVL 100 19 0.1314 29.8 126.1 167.7 4.734 26.5 C.GVL 100 12.5 0.1419 12.0 5.8 27.1 4.309 3.0 C.GVL 100 9.5 0.1114 21.7 3454.8 95.9 6.279 26.9 L.GRN 100 19 0.1767 71.3 2142.0 133.4 5.783 34.9 L.GRN 100 12.5 0.0847 23.5 4911.4 72.3 7.141 5.7 L.GRN 100 9.5 0.1208 16.9 15.1 4.3 4.736 0.4 LMS 100 19 0.0856 27.8 10.9 73.8 4.478 9.5 LMS 100 12.5 0.1294 27.0 2.8 58.9 3.918 9.2 LMS 100 9.5 0.1873 81.0 1214.0 173.0 4.399 58.2 R.GRN 100 19 0.0909 5.9 175.7 162.7 4.654 37.2 R.GRN 100 12.5 0.1188 25.2 20.5 23.6 4.860 2.3 R.GRN 100 9.5 0.1191 21.9 415.0 171.4 4.966 29.8 TRAP 100 19 0.2021 95.3 6.2 89.8 3.555 47.0 TRAP 100 12.5 0.1114 19.5 9.4 49.5 4.490 5.2 TRAP 100 9.5 0.1305 12.3 2.9 88.6 3.600 27.9 C.GVL 65 19 0.1358 38.2 56.7 171.2 4.135 35.1 C.GVL 65 12.5 0.3786 101.4 2.4 132.0 2.955 61.9 C.GVL 65 9.5 0.1436 1 24 1 9 8 6.8 .2 1 8.4 4.72 10. L.GRN 65 19 0.1553 2 2. 77.6 3.722 15.7 3.8 2 L.GRN 65 12.5 0.1333 2 4. 32.5 .152 7 1.9 1 4 3. L.GRN 65 9.5 0.3245 55 1.2 170.2 .597 .0 .4 2 50 LMS 65 19 0.1070 3 309 172.6 .567 .2 3.5 .1 4 37 LMS 65 12.5 0.2349 58.6 5.1 172.6 .760 .1 2 62 LMS 65 9.5 0.0979 19.3 18. 114.6 .447 .4 6 4 19 R.GRN 65 19 0.1097 13 10.9 27.3 .581 6 .9 4 2. R.GRN 65 12.5 0.1106 3 5.2 80.4 4.017 18.2 7.7 R.GRN 65 9.5 43.1 512.2 170.8 4.775 40.4 0.1393 TRAP 65 19 0.3704 8.9 0.01 20.6 1.546 6.1 TRAP 65 12.5 0.1104 14.5 9.5 96.2 4.374 11.0 TRAP 65 9.5 0.1918 58.8 0.6 84.5 2.950 34.5 L.GRN 40 12.5 0.2071 5.1 0.08 51.2 2.405 8.6 R.GRN 40 12.5 0.1280 7.8 4545.8 172.9 5.533 34.0 Overall Average: 0.1569 32.8 563.1 101.8 4.255 24.2 248 Since only two mixtures were designed with 40 gyrations, test results for these two mixtures were not included in the following ANOVA and discussions. nducted to evaluate the effect of the main factors (aggregate type, NMAS, and gyration level) and any interact able 5.13. C Since most of the tested samples were still in the secondary phase when the tests were finished, the slope of strain at the secondary phase is the most useful information available without the need for extrapolation. An ANOVA was first co ions between the main factors on slope of strain. The results are shown in T ompaction level and interaction between aggregate type and NMAS are significant factors at 95 percent significance level. TABLE 5.13 ANOVA for Slope of Strain in Secondary Phase of Static Creep Test Source DF Seq SS Adj SS Adj MS F P Agg 4 0.05882 0.05882 0.0147 1.43 0.235 Gyrs 1 0.06415 0.06415 0.06415 6.24 0.015 NMAS 2 0.00003 0.00003 0.00001 0.00 0.999 Agg*Gyrs 4 0.02714 0.02714 0.00678 0.66 0.623 Agg*NMAS 8 0.22856 0.22856 0.02857 2.78 0.011 Gyrs*NMAS 2 0.00611 0.00611 0.00306 0.30 0.744 Agg*Gyrs*NMAS 8 0.13341 0.13341 0.01668 1.62 0.138 Error 60 0.61723 0.61723 0.01029 Total 89 1.13545 As shown in Figure 5.20, the average slopes of strain for the mixtures designed with 65 and 100 gyrations were 0.183 and 0.129, respectively. One standard deviation shown as an error bar. The higher slope of mixtures designed with 65 gyrations indicate less deformation resistance than those designed with 100 gyrations, which likely results from the higher optimum asphalt content with the lower compaction level. Howeve criteria for the secondary slope in the confined static creep tests using high samples are available at this time to distinguish if the mixtures designed with 65 gyrations can still is r, no provide satisfactory rutting resistance. The SMA mixtures designed with 65 and 100 249 gyrations appear to have good rutting resistance if the criteria of slope with unconfined test conditions (The best and the second best categories for good rutting resistance is less than 0.17 and 0.20, respectively as shown in Table 2.9) are used as references. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 S p e o f eco n ar y P h ase, 1/l g ( sec) 0.35 100 65 Gyration Level l o S d o N = 45 FIGURE 5.20 Average slopes for two gyration levels. The interaction between aggregate type and NMAS on slope of strain in the standard deviation is shown as error bar. e and NMAS (e.g. 12.5 mm crushed gravel, 9.5 mm lab granite, and 19 mm traprock), the test variability is significant higher than others. This is likely due to one or two samples of these mixtures should be considered as outliers. These samples reached the limit of LVDTs in very short time and therefore showed a very high slope of the strain without tertiary flow. For different aggregate types, the average slope had no consistent trend with the changes of three NMAS. There appears to be no scientific reason for the combination effects of aggregate type and NMAS on slope results. It is believed that showing that these two properties have an interaction is simply due to high variation of the test results. secondary phase is shown in Figure 5.21. One For some combinations of aggregate typ 250 -0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 C.GVL L.GRN LMS R. GRN TRAP S l o p e o f S e co n d a ry P h a s e , 1/ l o g ( se c) 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 FIGURE 5.21 Interaction between and NMAS on slope of strain in static creep test. ic aggregate type Since the test times of the static creep test showed high variability, the logarithm (log) test time was used for the analysis. An ANOVA was conducted to evaluate the effect of the main factors (aggregate type, NMAS, and gyration level) and any interactions between the main factors on log test time at 4 percent strain. TABLE 5.14 ANOVA for Log Test Time when 4 Percent Strain Occurred Source DF Seq SS Adj SS Adj MS F P Agg. 4 20.374 20.374 5.094 3.43 0.014 Gyrs 1 24.336 24.336 24.336 16.41 0.000 NMAS 2 0.478 0.478 0.239 0.16 0.851 Agg.*Gyrs 4 12.638 12.637 3.159 2.13 0.088 Agg.*NMAS 8 38.487 38.487 4.811 3.24 0.004 Gyrs*NMAS 2 0.715 0.715 0.357 0.24 0.787 Agg.*Gyrs*NMAS 8 5.72 5.72 0.715 0.48 0.864 Error 60 88.976 88.976 1.483 Total 89 191.725 The results are shown in Table 5.14. The ANOVA results indicate that aggregate type and gyration level are tw between aggregate type and NMAS are also significant. Side by side comparison for the o significant influencing factors, and the interaction 251 effects sent of these significant factors and interactions were shown in the following Figures to better visualize the discussion. The average log time for five aggregate types and two compaction levels are shown in Figure 5.22. One standard deviation is shown as error bar. N 1 and N 2 repre the number of samples used for the average in each group. 0.00 1.00 2.00 3.00 5.00 C.GVL L.GRN LMS R. GRN TRAP 65 100 4.00 7.00 t ra i n 6.00 Av erag e L g ( T ) 4% o at s Aggregates Gyrations N 1 = 18, N 2 =45 FIGURE 5.22 A rag t p ain igure 5 , S r d yrat enera uired les 674 se nd 1 s urs 17.3 to ain n n . Th her co ion level w ti alt co ent and tighter aggregate structure therefore higher rutting resistance. The limestone and traprock aggregates required less time to reach 4 percent strain than the other three aggregates. For traprock, a possible reason is the high optimum asphalt contents (average 6.6 percent). For limestone, it is likely due to its relatively low uncompacted air voids for coarse aggregate (46.6 percent) and fine ve e log time o reach 4 ercent str in static creep test. From F .22 MA mixtu es designe with 65 g ions g lly req s test time (5 co s versus 62 66 second , or 1.6 ho versus hours) reach 4 percent str tha those desig ed with 100 gyrations e hig mpact ill result in less op mum asph nt 252 aggregate angularity (47.1 percent). The lower angularity and surface texture of the aggregate is likely to produce less rutting resistant mixtures. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 C.GVL L.GRN LMS R. GRN TRAP Av er ag e L o g ( T ) at 4% st r a i n 19 mm NMAS 12.5 mm NMAS 9.5 mm NMAS N = 6 FIGUR e to e d d that showing that these two properties have an interaction is simply due to low preci on of the test results. NCHRP 9-19 project (84) recommended using the flow time as an indicator of rutting resistance, and presented severa between flow time and field rut depths based on the field data from MnRoad, ALF and Westrack test sections. These E 5.23 Interaction between aggregate type and NMAS on average log tim reach 4 percent strain. The interaction between aggregate type and NMAS is shown in Figure 5.23. On standard deviation is shown as error bar. For lab granite and traprock, 12.5 mm NMAS mixtures required more time to reach 4 percent strain than 19 mm and 9.5 mm NMAS mixtures. However, for the other three aggregates, 12.5 mm NMAS generally required less time to reach 4 percent strain than the 19 mm and 9.5 mm NMAS mixtures. There appears to be no scientific reason for the combination effects of aggregate type an NMAS on average log (T) results. It is believe si l correlations 253 correlations are shown in Figure 5.24. As expected, the field rut depth show negative trend with the flow time. FIGURE 5.24 The Correlations between flow time and field rut depths (84). There were only 7 out of 96 tested SMA samples that showed tertiary flow even with the higher test temperature (140?F) used in this study. This is partly due to the f that most tests were manually stopped after 4 hours loading based on the test plan. However, there were a few samples were tested for more than 2 days during the act 254 weekends, which is up to 200,000 seconds. Most of these tests were manually stopped Monday since it appeared no tertiary flow would happen. Due to the lack of flow time data in on this study, it is difficult to make use of these correlations to predict the field rut epth. As shown in Figure 5.24, a 10 mm field rut depth correlates with a flow time of g time for each sample. This long tatic alysis As disc of sample d about 150,000 seconds, which is about 2 days testin testing time makes static creep test impractical for predicting rutting performance. In summary, although the data analysis on static creep test results showed a significant difference between 100 and 65 gyrations, the high variability of the s creep test results limited the use of test results for predicting rutting resistance of the SMA mixtures. Also, no criteria of strain level at certain time or time to reach a certain strain level for confined static creep tests are available at this time to distinguish if the SMA mixtures designed with 65 gyrations can still provide a satisfactory rutting resistance. Due to the high variability of results and long testing time for the static creep test, the use of the static creep test for predicting rutting potential is not recommended. 5.4 REPEATED LOAD TEST 5.4.1 Repeated Load Test Results and An ussed in the literature review (89, 96-99), the repeated load test has been found to give a better correlation with the field rut depth and more responsive to the presence modified binder in HMA mixtures than static creep test. The repeated load test has been used for predicting rutting potential of HMA mixtures for many years, and some acceptance criteria has also been established. However, these criteria were often established under unconfined test conditions, using short samples, or using whole deformation. Only a few studies used confined test conditions and relatively high (up to 5 255 inches) samples. Therefore, additional work was needed to establish an acceptance crit This section will present the repeated load test results and examine the influencing factors on the test results. An acceptance criterion for repeated load test under current test conditions will be established. The effect of compaction level on the repeated load test results will be evaluated. The repeated load confined creep test was conducted for a total of 32 mixtures, in which 15 mixtures were designed with 100 gyrations, 15 mixtures were designed with 65 gyrations, and 2 mixtures were designed with 40 gyrations. Three replicated samples were tested for each mixture. The repeated load test results for individual samples are shown in Appendix Table D3. Table D3 include secondary phase, the microstrain at 100 cycles, 1000 cycles, 5000 cycles, and 10,000 cycles. The average test results of three replicate samples for each mixture are summarized in Table 5.15. Table 5.15 includes information on the slope of the linear secondary phase and the average strain at 10, 000 cycles. Since the ideal full range of LVDTs used in this study was limited to 5 percent strain, and there were unequal readings among the three LVDTs, the effective average readings from the three LVDTs were often less than 4 percent. When any LVDT reached the full range of measurement before the 10,000 cycles, the reading at 10,000 cycles was extrapolated from the last point at which all LVDTs were functional using the tangent slope. erion to differentiate a rutting susceptible mix under current test conditions. s the information on the intercept a and slope b for the linear 256 TABLE 5.15 Summary of Repeated Load Test Results Log Slope b, 1/log(sec) Strain @ 10, 000 Cycles, % Agg. Type Gyrs NMAS, mm Average St. Dev COV, % Average St. Dev COV, % GVL 100 19 0.130 0.029 22.5 1.2693 0.189 14.9 GVL 100 12.5 0.173 0.082 47.5 1.7506 0.657 37.6 GVL 100 9.5 0.157 0.058 36.7 1.3696 0.630 46.0 L.GRN 100 19 0.120 0.004 3.4 1.1672 0.181 15.5 L.GRN 100 12.5 0.137 0.014 10.5 1.1794 0.178 15.1 L.GRN 100 9.5 0.158 0.017 10.8 1.2706 0.199 15.7 LMS 100 19 0.207 0.011 5.2 2.7016 0.195 7.2 LMS 100 12.5 0.184 0.033 17.9 2.6660 0.081 3.0 LMS 100 9.5 0.236 0.069 29.1 3.4665 0.958 27.6 R.GRN 100 19 0.199 0.058 29.2 2.8104 0.374 13.3 R.GRN 100 12.5 0.197 0.039 19.8 2.0961 0.358 17.1 R.GRN 100 9.5 0.195 0.042 21.6 1.9882 0.334 16.8 TRAP 100 19 0.170 0.070 40.9 3.0269 1.374 45.4 TRAP 100 12.5 0.197 0.034 17.4 2.6278 0.148 5.6 TRAP 100 9.5 0.190 0.049 25.9 3.4426 3.541 102.9 GVL 65 19 0.133 0.024 17.7 2.0486 0.287 14.0 GVL 65 12.5 0.144 0.012 8.1 1.7777 0.500 28.1 GVL 65 9.5 0.143 0.013 8.9 1.3883 0.285 20.5 L.GRN 65 19 0.191 0.033 17.1 2.8100 1.137 40.5 L.GRN 65 12.5 0.255 0.010 4.1 3.4775 0.433 12.5 L.GRN 65 9.5 0.261 0.089 34.2 4.2354 2.374 56.0 LMS 65 19 0.238 0.035 14.5 5.7501 0.306 5.3 LMS 65 12.5 0.221 0.056 25.5 4.3918 1.897 43.2 LMS 65 9.5 0.192 0.014 7.5 3.5698 2.336 65.4 R.GRN 65 19 0.211 0.046 21.6 3.1657 0.399 12.6 R.GRN 65 12.5 0.227 0.028 12.3 3.0269 0.388 12.8 R.GRN 65 9.5 0.218 0.018 8.0 2.6745 0.828 31.0 TRAP 65 19 0.120 0.054 45.3 2.3247 0.269 11.6 TRAP 65 12.5 0.154 0.034 22.1 1.9580 0.177 9.0 TRAP 65 9.5 0.165 0.021 12.6 2.6186 0.850 32.5 L.GRN 40 12.5 0.310 0.067 21.5 5.7059 0.775 13.6 R.GRN 40 12.5 0.259 0.028 11.0 4.8806 1.015 20.8 257 Tertiary flow was observed for only 4 samples out of 96 tested samples. One of these four samples was designed with 100 gyrations compaction level (traprock 9.5 mm NMAS t most of the designe No.1 sample). The other three were designed with 65 gyrations (limestone 12.5 mm NMAS No.5 sample, limestone 9.5 mm NMAS No.2 sample, and lab granite 9.5 mm NMAS No. 6 sample). A typical repeated load test result that shows tertiary flow is shown in Figure 5.25. For the 92 samples with no tertiary flow, the slope of cumulative strain gradually decreased during the test, and became relatively stable after a certain number of cycles. A typical repeated load test result without tertiary flow is shown in Figure 5.26. The fact that only a few samples have the tertiary flow developed after 10,000 cycles with 120/20 psi loading at 60?C test temperature indicated tha d mixtures had good resistance to deformation. 0.10 1.00 10.00 100.00 , 1 10 100 1000 10000 Cycle Number C u m u l a t i v e S t r a in % LVDT A LVDT B LVDT C LVDT Avg Ram FIGURE 5.25 A typical repeated load test result with tertiary flow. 258 0.10 1.00 10.00 100.00 1 10 100 1000 10000 100000 Cycle Number Cu u la t i S t r a % m v e in , LVDT A LVDT B LVDT C LVDT Avg Ram FIGURE 5.26 A typical repeated load test result without tertiary flow. Further statistical analyses were conducted on the accumulated strain at 10,000 cycles and the strain slope during the secondary phase. Since only two mixtures were designed with 40 gyrations, test results for these two mixtures were not included in the fo The cumulative strain levels at 10,000 cycles for all mixtures tested are demonstrated in Figure 5.27. One standard deviation is shown as error bar. All 4 mixtures that had samples showing tertiary flow had higher standard deviations for the results. The pooled standard deviation values are 0.90, 1.11, and 1.06 percent for mixtures designed with 40, 65, and 100 gyrations, respectively. For lab granite and ruby granite 12.5 mm NMAS mixtures with three gyration levels, as expected, the strain level at 10,000 cycles decreased as compaction level increased due to the lower optimum asphalt content. A different trend is shown for traprock. This is likely due to two reasons. One is the llowing ANOVAs and some of the comparisons. 259 difference between 100 gyrations and 65 gyrations for traprock is smaller than that o other aggregates due to the smaller changes i f the n asphalt content (average 0.4 percent versus 0.8 per t ed cent) and aggregate breakdown (average 1.4 percent versus 6.8 percent changes a critical sieves). The other reason is that one of the 9.5 mm traprock samples show tertiary flow and had a significantly higher strain level at 10,000 cycles. 0.00 1.00 2.00 4.00 5.00 6.00 7.00 8.00 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 3.00 C.GV L L.GRN LMS R.GRN TRA P S t r a L evel @ % in 10,000 C ycles, 40 Gyrs 65 Gyrs 100 Gyrs N = 3 FIGURE 5.27 Strain level at 10,000 cycles for three gyration levels. ANOVA for Strain at 10,000 Cycles DF Seq SS Adj SS Adj MS F statistics P value TABLE 5.16 Source Agg. 4 43.096 43.096 10.774 9.17 0.000 Gyrs 1 15.273 15.273 15.273 13.00 0.001 NMAS 2 0.676 0.676 0.338 0.29 0.751 Agg.*Gyrs 4 25.142 25.142 6.286 5.35 0.001 Agg.*NMAS 8 6.464 6.464 0.808 0.69 0.700 Gyrs*NMAS 2 0.744 0.744 0.372 0.32 0.730 Agg.*Gyrs*NMAS 8 7.958 7.958 0.995 0.85 0.566 Error 60 70.470 70.470 1.174 Total 89 169.823 260 An ANOVA was conducted to evaluate the effect of the main factors (aggregate type, NMAS level) and any interactions between the main factors on cu s cles. The ANOVA result is shown in Table 5.16. Table 5.16 indicates that agg d compaction level are significant influencing factors, and the intera hese two factors is also significant. Side by side comparisons for the effects o nificant factors and interactions were shown in Figures 5.28 a b e dis ion. The a el a 00 c leve a is shown as error bar. N 1 and N 2 p . , and compaction mulative train at 10,000 cy regate type an ction between t f these sig nd 5.29 to etter visualize th cuss verage strain lev t 10,0 ycles for five aggregates and two compaction ls re shown in Figure 5.28. One standard deviation re resent the number of samples used for the average for each group 1.00 2.00 3.00 4.00 5.00 6.00 ver t r 10 , ycl 0.00 C.GV L L.GRN LMS R.GRN TRA P 65 100 Aggregates Gyrations A ag e S a i n L e vel @ 000 C e s , % N 1 = 18, N 2 =45 1.6 percent at 10,000 cycles. The high strain results for limestone are consistent with its APA rutting test results, and are likely due to its low angularity and surface texture. The FIGURE 5.28 Average strains at 10,000 cycles for two main factors. As shown in Figure 5.28, limestone has the highest average strain of about 3.8 percent at 10,000 cycles, while the crushed gravel has the lowest average strain of about 261 lowest strain results for crushed gravel are unexpected, because crushed gravel has the highest F&E content (35.2 percent for 3:1 and 9.4 percent for 5:1) and the second highest L.A ab d voids for coarse aggregates (48.4 percent) and high FA 9, 94), therefo tion level provides a lower asphalt paction levels in terms of strain level at 10,000 cycles. rasion value of 30.7 percent in all five aggregates. The high F&E content and L.A abrasion are generally considered undesirable for good performance. However, some recent studies (103) have shown that there did not appear to be a relationship between the F&E content (3:1) - in a range of about 10 to 40 percent- and performance. The good performance of crushed gravel is probably due to its good angularity and surface texture. This is indicated by the high uncompacte A value (50.0 percent). The good repeated load results for crushed gravel indicated that these two aggregate properties (F&E content and L.A abrasion) may not be related to the rutting performance within the range used in this study. Also, the strain results for all aggregate types are considered low based on the literature review (8 re the difference between all aggregate types is not practically significant. The average strain at 10,000 cycles for mixtures designed with 100 gyrations was about 2.2 percent, which is lower than the 3.0 percent for those designed with 65 gyrations. The result is logical because the higher compac content and tighter aggregate structure, therefore likely to be more rutting resistant. A paired-t test was employed to compare these two compaction levels on cumulative strain at 10,000 cycles. The result is shown in Table 5.17 and indicates that there is a significant difference between these two com 262 TABLE 5.17 Paired-T Test Results on Strain at 10,000 Cycles Gyrs N Mean StDev SE Mean 65 15 3.0145 1.1474 0.2963 100 15 2.1888 0.8298 0.2143 Difference 15 0.8257 1.2683 0.3275 95% CI for m T-Test of mean difference = 0 (vs not = 0): T-Value = 2.52, P-Value = 0.024 between aggregate types and compaction levels for cumulative r ean difference: (0.1233, 1.5280) The interaction strain at 10,000 cycles is shown in Figure 5.29. One standard deviation is shown as erro bar. 0.00 1.00 2.00 C. V L L.GRN LMS R.GRN TRA P ver a g t r a in 3.00 4.00 L evel 10, cycles, % 5.00 6.00 G A e S at 000 65 Gyrs 100 Gyrs N = 9 10,000 cycles. As shown in Figure 5.29, for limestone and lab granite, the changes in design tion level made a gre FIGURE 5.29 Interaction between aggregate type and compaction level on strain at compac ater difference in strain results than other aggregates. This can be two agg compac in at 10,0 hose designed with 100 explained in that there was a greater change in optimum asphalt content for these regate types than for the other three aggregate types (1.0 versus 0.5 percent) when tion level changed from 100 to 65 gyrations. Traprock had a lower average stra 00 cycles for mixtures designed with 65 gyrations than t 263 gyr designe n level at 10, 000 cycles. If this traprock sample is seen as an outlier and excluded from the analysis, the average strain at 10,000 cycles for traprock mixtures designed with 100 gyrations will be slightly lower than those designed with 65 gyrations. ations. As explained before, this is mainly due to the fact that one traprock sample d with 100 gyrations had tertiary flow, which resulted in an extremely high strai 0.000 0.050 0.100 0.200 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 19 12.5 9.5 0.150 0.300 19 12.5 9.5 0.250 0.350 0.400 C.GVL L.GRN LMS R.GRN TRAP S a in S l p e o n t r o L cal o g S e 40 Gyrs 65 Gyrs 100 Gyrs N = 3 vels. 5.30. O es are 0.05 respect gyratio few oth likely d FIGURE 5.30 Strain slope at secondary phase for three compaction le The strain slopes on a log scale for all mixtures tested are demonstrated in Figure ne standard deviation is shown as error bar. The pooled standard deviation valu 1, 0.039, and 0.041 for SMA mixtures designed with 40, 65, and 100 gyrations, ively. For lab granite and ruby granite 12.5 mm NMAS mixtures with three n levels, the strain slope decreased with the increase of compaction level. For a er mixtures, the trend of strain slope with gyration level is opposite, and this is ue to the high test variability. 264 An ANOVA was conducted to evaluate the effect of the main factors (aggregate MAS, and compaction level) and any interactions between the main fa lope at secondar type, N ctors on strain s y phase. The ANOVA results are shown in Table 5.18. Table 5.18 ind e levels a TABLE est Source DF Seq SS Adj SS Adj MS F statistics P value icat s that aggregate type and the interaction between aggregate type and gyration re significant. The compaction level is not a significant influencing factor. 5.18 ANOVA for Strain Slope at Secondary Phase of Repeated Load T Agg. 4 0.056893 0.056893 0.014223 7.79 0.000 Gyrs 1 0.00488 0.00488 0.00488 2.67 0.107 N 84 0.168 MAS 2 0.006718 0.006718 0.003359 1. Agg.*Gyrs 4 0.047979 0.047979 0.011995 6.57 0.000 Agg.*NMAS 8 0.010498 0.010498 0.001312 0.72 0.674 Gyrs*NMAS 2 0.000814 0.000814 0.000407 0.22 0.801 Agg.*Gyrs*NMAS 8 0.008794 0.008794 0.001099 0.60 0.772 Error 60 0.109516 0.109516 0.001825 Total 89 0.246093 0.050 0.100 0.150 A v e r e S t i n S l e o n 0.000 0.200 0.250 C.GVL L.GRN LMS R.GRN TRAP a g r a op Lo a 0.300 g S c l e N = 18 FIGURE 5.31 Average strain slopes for different aggregate types. The average strain slopes on log scale for five aggregate types are shown in Figure 5.31. One standard deviation is shown as error bar. The lab granite showed 265 signific differe , as show higher crushed accumu h acc ant higher test variation than other aggregates. This is likely due to the greater nce between two compaction levels for lab granite mixture in terms of strain slope n in Figure 5.33. Combining two sets of data with greater difference will result in standard deviation. The limestone had the highest strain slope of 0.213 and the gravel had the lowest strain slope of 0.139. The results are consistent with the lated strain results shown in Figure 5.28, i.e. the aggregate that has hig umulated strain also has high strain slope. y = -0.0341x + 1.8084 R 2 = 0.8075 0.0000 0.0500 0.1000 46.0 46.5 47.0 47.5 48.0 48.5 49.0 ver ag S t r a in 0.1500 0.2000 0.2500 Uncompacted Voids of Coarse Aggregate, % A e S S l o p e o n L o g cale FIGURE 5.32 a good correlation with the uncom cted voids of coarse aggregate. As shown in Figure 5.32, the average strain slope decreases with an increase in the uncompacted voids. This strong internal friction caused by the increased aggregate Relationship between strain slope and uncompacted voids of coarse aggregate. It is noteworthy that the average strain slope for different aggregate types showed pa trend is as expected because the angularity and surface texture will increase the rutting resistance. Also, the SMA has a 266 hig importa h coarse aggregate content which makes the coarse aggregate properties more nt than fine aggregate properties. 0.000 0.050 C.GVL L.GRN LMS R.GRN TRAP Ave r 0.100 0.150 0.200 e S t i n S l e o n o g S 0.250 0.300 ag ra o p L c al e 65 Gyrs 100 Gyrs N = 9 FIGURE 5.33 The interaction between aggregate types and compaction levels for strain slope is shown in Figure 5.33. One standard deviation is shown as error bar. For lab granite, the changes in design compaction level resulted in a greater difference in slope of strain than any other aggregates. This may be due to the significant change in optimum asphalt content of 1.1 percent between the two compaction levels. Traprock had lower average slope of strain for mixtures designed with 65 gyrations than those designed with 100 gyrations. This is likely due to the lowest change in optimum asphalt content and one outlier sample for traprock designed with 100 gyrations. That sample showed tertiary flow. Interaction between aggregate type and compaction level on strain slope. 267 5.4.2 Discussion on Cumulative Strain Criteria At the time this report was written, there was no proven criterion for cumulative strain level at 10,000 cycles to distinguish the acceptable and unacceptable rutting resistance based on the new test procedure. Some studies (71, 89) recommended a critical cu la resistance. This range was given based on the correlation between the laboratory repeated load te te er si a e study ( ed on some sh fi in reading e st u transdu cycles and 5.31 based o this study. Four samples that showed tertiary flow were e d mbols. The LV ng the se d ened. mu tive strain level range of 10 to 13 percent to determine the good or poor rutting st results and field rutting observations. The sample diameter of 4 inches, test mp ature of 60?C, and axial load pressure of 120 psi and confining pressure of 20 p re th same as used in this study. However, some laboratory test conditions used in that 89) are different from this study. The laboratory tests were conduct ort eld core samples (The height of samples varied from 1.8 to 5.4 inches), the stra was for the whole sample height from the ram transducer, and the cumulativ rain sed was after 3600 cycles loading. In order to use the critical strain level for this study, a relationship between ram cer and LVDT?s reading and a relationship between 3600 cycles and 10,000 need to be developed. These two relationships are shown in Figures 5.30 n the 96 samples tested in xclu ed from the analysis, but are still shown in the Figures using different data sy DT strains larger than 5 percent in this study were extrapolated usi con ary phase slope. This extrapolation became invalid once the tertiary flow happ 268 35.00 Samples showed tertiary flow y = 1.5149x - 0.0917 R 2 = 0.7082 10.00 15.00 20.00 R a m St ra i n 25.00 30.00 , % 0.00 5.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 LVDT Strain, % FIGURE 5.34 The relationship between ram strain and LVDT strain reading. y = 1.3931x - 0.2952 4.00 5.00 , 0 00 Cy c l es R 2 = 0.9596 6.00 7.00 8.00 , % Samples showed tertiary flow 1.00 2.00 3.00 S t r a i n at 10 0.00 0.00 1.00 2.00 3.00 4.00 5.00 Stain at 3600 Cycles, % FIGURE 5.35 The relationship between strain at 3600 and 10,000 cycles. strain i , the cor al strain level at 10,000 cycles with LVDT reading will be 9.0 Based on the two relationships shown in Figures 5.34 and 5.35, when 10.0 percent s selected as the critical strain level at 3600 cycles with the ram transducer reading responding critic 269 p t since th height). The criterion for the higher sam lower b ef o lower a th i to sample e co p showing tertiary flow had cum all sam te y criterion for differentiating between good and poor performance mixture. As shown in Figure of 3600 cycles and 1 e is a poten 60 percent and still get good information. If r 00 cycles s ure 5.36. Since only two mixtures (12.5 mm NMAS SMA with lab granite and ruby granite) were d signed ercen . This transformed criteria should be correct only for relatively short samples e original criteria was developed using short samples (1.5 to 5.4 inches in ples (6 inches in height) in this study is believed to be ecause of less effect due to end friction and confinement. A study (84) on the fect f ratio of sample height to diameter showed that the failing strain level becomes nd less variable with the increase of this ratio and is close to a constant level after e rat o exceeds 1.5:1. As shown in Figure 5.34, a cumulative strain of 6 percent seems be a threshold value when tertiary flow begins. However, as shown in Figure 5.35 a showed tertiary flow when the strain was less than 4 percent at 3600 cycles. Th rres onding strain level at 10,000 cycles for this result is about 5 percent. All samples ulative strains higher than 6 percent at 10,000 cycles while ples having strain levels lower than 5 percent at 10, 000 cycles did not show rtiar flow. Therefore a strain level of 5 percent at 10,000 cycles appears to be a 5.35, a good linear correlation exists between strain level 0,000 cycles for these mixtures that do not show the tertiary flow. This indicates ther tial to reduce the test time by more than 5 pe cent is selected for the criteria at 10,000 cycles, a corresponding criterion for 36 hould be about 3.8 percent. All the strain results at 10,000 cycles versus gyration levels are shown in Fig esigned with three gyrations levels while all the rest of mixtures were de 270 w cumula ult to conduct. ith two gyration levels, an overall analysis on the effect of gyration level on the tive strain is diffic All Three Levels Pow er (Three Levels) 7.00 y = 707.57x -1.3147 R 2 = 0.8689 5.00 6.00 0.00 1.00 2.00 Gyrations S t a i n l e v e l a t 1 000 C y c l e s , % 3.00 4.00 0 , 30 40 50 60 70 80 90 100 110 F el. w ions. This is c c gyrations and most (14 out of 15) mixtures designed with 65 gyrations are less than the sugges symbol of these two mixtures is also included. As sho increase of gyration level. However, the decrease in cumulative strain is not large when IGURE 5.36 The relationship between strain level and compaction lev As shown in Figure 5.36, the cumulative strain results for SMA mixtures designed ith 65 gyrations varied over a wider range than those designed with 100 gyrat likely due to the higher variability of mix properties associated with a lower ompa tion level. The cumulative strain results for all mixtures designed with 100 ted criteria of 5 percent. The two mixtures with three gyration levels are highlighted using different data s. A best fitted regression line for the results wn in Figure 5.36, the cumulative strain of these two mixtures decreases with an 271 th m e cumula other words, 65 gyrations appear to be the lo rutting on the evaluation of both APA and repeated load test results, the SMA m re in cumula igure 5.37. e co paction level increases from 65 to 100 gyrations. For these two mixtures, th tive strain becomes marginal when the gyration level drops to 40 gyrations. In west compaction level that still can ensure a resistant mixture based on the suggested repeated load test criteria. Based ixtu s designed with 65 gyrations have a high probability being rutting resistant. It is teresting to see the correlation between these two tests. The relationship between tive strain from repeated load test and APA rut depth results is shown in F y = 0.4804x + 1.0454 R 2 = 0.1525 2.00 3.00 4.00 a in @ 10, 0 00 0.00 1.00 S t 5.00 6.00 7.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 r cycles, % APA Rut Depth, mm RE 5.37 RelaFIGU tionship between repeated load cumulative strain and APA rut depth. elation (R 2 =0.15) between APA ru is not s A mixtures designed and tested in this study are considered to be resistant to rutting with low APA rut depths and low cumulative strain As shown in Figure 5.37, there is only a poor corr t depth and strain level at 10,000 cycles of the repeated load test. This poor correlation urprising, because most SM 272 v , gure 5.37, o res failed either of the two criteria. alues and these test results are only within a narrow data range. As shown in Fi nly 4 out of 32 designed SMA mixtu FIGURE 5.38 The Correlations between flow number and field rut depths (84). NCHRP 9-19 project (84) recommended using the flow number as an indicator of rutting d are sho ples showed tertiary flow before 10,000 n this study. In other w , therefo d as resistant to rutting. The predicted field rut depths for resistance, and presented several correlations between flow number and field rut epths based on the field data from MnRoad, and ALF test sections. These correlations wn in Figure 5.38. There were only 4 out of 96 tested SMA sam cycles even with higher test temperature (140?F) used i ords most of tested samples had flow numbers greater than 10,000 cycles, and re should be considere 273 m correlations shown in Figure 5.38 are used. and cu a cumula 0 The tes er (1.4 to 5.4 inches in height) and strain results were from ram transducer and recorded at 3600 cycles. As discussed above, a new criterion of 5 p t for SMA m 100 and 65 gyrations are 8.3 and 9.8 mm, respectively, if w u the n a designe h is m ost SMA mixtures designed with 65 and 100 gyrations are less than 10 mm if the A national rutting study (96) presented a correlation between field rut depth mul tive strain from repeated load, as shown in Figure 5.39. The criterion for tive strain is about 10 percent if the maximum allowable field rut depth is set as .5 inch, or 12.5 mm. However, the lab test conditions were different from this study. t samples were short ercen is recommended based on the data in this study. The predicted field rut depth ixtures designed with e ass me a linear relationship for the cumulative strain between this SMA study and ation l study, and used the correlation shown in Figure 5.39. For SMA mixtures d with 40 gyrations, the average field rut depth is expected as 12.7 mm, whic arginal if we assume the maximum allowable field rut depth is 12.5 mm. FIGURE 5.39 Field rut depth versus the lab strain from repeated load test (96). 274 5. The rut by APA rutting test, dynamic m u levels a .19. TABLE ificant rence 40 Suggested iteria 5 SUMMARY ting resistance for the SMA mixtures was evaluated odul s, static creep and repeated load tests. The test results for different compaction re summarized in Table 5 5.19 Average Test Results for Different Compaction Levels 100 65 Sign Diffe Test Properties Gyrations Gyrations between 100 and 65 gyrations? Gyrations 1 Cr APA Test Rut depth, mm 3.1 3.9 Yes 4.8 5 max E*/sin? at 10 Hz, 2038 1811 No 2875 N/A MPa Dynamic modu us E*/sin? at 0.1 Hz, MPa 2576 2265 No 4236 N/A l test Time to reach 4% strain, hrs 17.3 1.6 Yes 1.0 N/A 2 Static est Slope of Strain in secondary phase, 0.129 0.183 Yes 0.168 creep t 1/log (sec) N/A Cumulative strain at 10,000 cycles, % 2.2 3.0 Yes 5.3 5 max Repea load te A ted st Slope of Strain in secondary phase, 1/log (sec) 0.174 0.191 No 0.285 N/ 1. lted in a sm SMA m were still satisfactory for APA rut depth criterio ns become s also able to m Only two mixtures were designed with 40 gyrations. 2. Average value from the antilog of the average logarithmic value. As shown in Table 5.19, a decrease in compaction level generally resu all decrease in rutting resistance of designed SMA mixtures. However, 13 out of 15 ixtures designed with 65 gyrations n of 5.0 mm. The rutting resistance for mixtures designed with 40 gyratio s marginal, but one of two SMA mixtures designed with 40 gyrations wa eet this criterion. 275 The dynamic modulus test results showed that there is no significant difference between SMA m of the d se the relatively low stress and strain associated with the test do not reflect the stress-strain situation in the pavement. The E*/sin? value at high temperature showed contrary results w tightne le for SMA m strain l in indicated that there were significant differences between 65 and 100 gyrations. However, additio high va , the static creep test was not recommended to be used to test was developed based on literature review and test data in this study. Most (14 out of 15 e of the two requirem ) are be too stringent. There appears to be no correlation between these aggregate properties with rutting performance within the range of this study. ixtures designed with 65 gyrations and 100 gyrations. The effectiveness ynamic modulus test for predicting rutting resistance is questionable becau hen the load frequency changed, and the dynamic modulus results depended on the ss of aggregate structure or VMA value, which may not be applicab ixtures. The static creep test showed very high variability on test time to reach a certain evel. The data analysis on slope of strain and log time to reach 4 percent stra nal work is needed to determine criteria for these static creep results. With the riability and lack of criteria draw any conclusions for this study. A cumulative strain criterion of 5 percent after 10,000 cycles for repeated load ) SMA mixtures designed with 65 gyrations were satisfactory for this criterion. On SMA mixtures designed with 40 gyrations marginally met this criterion. The successful design for SMA mixtures using all aggregate types indicates the ents for F&E content and L.A abrasion in SMA mixture design guides (43 276 In summary, a SMA mixture designed with 65 gyrations level will be resistant to with a good confidence based on the discussion on APrutting A and repeated load test re potenti ation. sults. The feasibility of dynamic modulus and static creep tests for predicting rutting al is questionable and need further evalu 277 R 6 6.1 CONCLUSIONS Based on the data analysis and discussion presented in chapters 4 and 5, the following general conclusions were drawn for the design and testing of SMA mixtures. 1. The vacuum seal method (CoreLok) appeared to be more accurate than the SSD method at higher air void levels where the SMA mixtures become permeable. The SSD method should only be used when the water absorptions are lower than 0.4 to 0.9 percent depending on NMAS. The corrected vacuum seal air voids should be used when water absorption higher than these limits. 2. The vacuum seal method overestimated air void content when measuring laboratory compacted samples. The correction factor embedded into the Coregravity TM software program by the manufacturer is appropriate for cored- and-sawn samples, however, is appears not sufficient for laboratory compacted SMA mixtures. An additional correction of 0.5 percent should be used when the software is used. A correction factor of 1.4 percent should be used when the software is not used. 3. SMA mixtures become permeable at higher air void contents for lower NMAS. If the threshold value for permeable mixtures is set at 125?10 -5 cm/s, the critical air voids values by the SSD method are approximately 5, 6, and 7 percent for 19, 12.5 and 9.5 mm NMAS SMA mixtures, respectively. CHAPTE CONCLUSIONS AND RECOMMENDATIONS 278 4. To produce a cored-and-sawn triaxial SMA sample with 4 percent air void content s s t s o a rce ir v co ggr te d ada by or ry c tion h od cor ons w aggregate L.A. abrasion value and fair ations with F&E content (3:1) 6. Under the test tem ur 60 , both dynamic lus and e angl sul ecre d w he decrease of l g freq . The se of p gle icated the aggregate structure becomes m minan the h m re lo d re ency e ssf SM ix designs with all five aggregates indicate the qu nts r ag at op ies, s L.A ion an E cont re o ent hes a ga prop with ranges n in th ud y not be detrimental for the m perfo ce. ase the ta a si d ussi esente hapter d 5, th following conclusions wer de with respect state ect ob es. lo com ctio e 5 ratio ovides re durable SMA m e ith eased opt a lt nten allows the use of aggreg pes pa to y ns MA ures d ed with yratio av ge o .7 p t er timu phalt t than design ith gyr ons 0 rc t hig ptimum asphalt co than th sig wit 0 b w he arsh mmer mixtures designed with g ons d a ra f 1 perc igher than t esigne h 0 ion All m re esig ith 65 tions m e mini u ing the SSD method require that he whole ample be c mpacted to bout 5 pe nt a oid ntent. 5. A ega egr tion lab ato ompac ad go relati ith the correl . perat e of ?C modu phas e re ts d ase ith t oadin uency decrea hase an ind ore do t with igh te peratu and w loa ing f qu . 7. Th succe ul A m re ireme fo greg e pr ert such a abras d F& ent, a to string . T e two ggre te erties in the show is st y ma ixture rman B d on da naly s and isc on pr d in c s 4 an e e ma to the d proj jectiv 1. A wer pa n lev l of 6 gy ns pr a mo ixtur w incr imum spha co t and more ate ty com red 100 g ratio . S mixt esign 65 g ns had an era f 0 ercen high op m as conten those ed w 100 ati , and .2 pe en her o ntent ose de ned h 5 lows ith t M all ha . SMA 65 yrati ha n ave ge o .5 ent h VMA hose d d wit 10 gyrat s. SMA ixtu s d ned w gyra et th mum 279 optimum asphalt content and VMA requirements for SMA mixture, while only 8 of 15 (53 percent) of mixtures designed with 100 gyrations met these qu nts tur des w 65 yrati ere ge r or lo rm lity n those designed with 100 gyratio ilar air voids. For 19 m S tur 5 tio resu n a lo ermeab than 10 r at s ilar air voids. For 12.5 .5 mm S mix , the ef of o paction levels on eability are not sign nt. oth gyra ns 0 rat s re in sim aggreg eakdown as in o ro 65 t gyrat ompaction resulted in an erc d na ggre reakd t the critical sieve mpacted samples had more fractured aggregates than the SGC p ed s ple e g hang percent passing the critical sieve for ns were 7.3, 5.2, and 4.6 percent, esp ly. ll ed A u it iffer ompaction levels were satisfac or ent. The draindown test re or all m res we than the maximum t o p ent the us olymer ified a ind d c los er ixtures designed with 65 gyrations is ene low tha s ig d wi 0 gyra but is atisfac for he A rut th rep d d cu tive st riteria een ou ixtu co te ith gy s still rmed well when 5.0 mm re ireme . 2. SMA mix es igned ith g ons w nerally had simila wer pe eabi tha ns at sim m NMA mix es, 6 gyra ns lted i wer p ility 0 gy ations im and 9 NMA tures fects tw com perm ifica 3. B 65 tio and 1 0 gy ion sulted ilar ate br the field construction. G ing f m o 100 ions c average of 0.6 p ent a ditio l a gate b own a s. Marshall co com act am s. Th avera e c es in 50 blow Marshall, 100 gyrations and 65 gyratio r ective 4. A design SM mixt res w h d ent c tory f the draindown requirem sults f ixtu re less limi f 0.3 erc with e of p mod sphalt b er an ellu e fib . 5. The rutting resistance of the SMA m g rally er n tho e des ne th 10 tions, still s tory t PA dep and eate loa mula rain c . Thirt t of fifteen m res mpac d w 65 ration perfo 280 was used as the maximum APA rut depth allowed. A cumulative strain criterion of 5 percent for repeated load test was developed based on literature review and e t esults in this study. Fourteen out of fifteen SMA mixt signe gy ons t th ite . B h A d repe load re indicated the A xtur des w 40 yrati ad ma rutting resistance, ly out tw a inall A tur des w 65 yrati id not a sign t diffe th e d ne h gy tions rms of dynamic modulus test results. h mp tur E*/sin? alue ased w e decr f load q fo MA re, the effectiveness of us */sin? ed th utti r SMA is questionable. e ana is o ti ep st re showe gnifica ferenc tw 100 d 6 at . H eve high variability o est re d tes tim i he e of est fo icting rutting resistance of e S mi res e ber of gyration should be the s egard f NMA o signi id wa bse th ffe nt co tion l are necessary for si SM mi w di rent N S. Th OVA o A rutti atic ep, re d results did not show the NMAS was a gni nt in enc to 6.2 RECOMMENDATIONS Based on the data analysis and discussion pres in Ch s 4 and follow re a th est r ure de d with 65 rati me is cr rion ot PA an ated sults SM mi es igned ith g ons h rginal and on one of o mixtures m rg y met these suggested criteria. 6. SM mix es igned ith g ons d show ifican rence wi thos esig d wit 100 ra in te At igh te era e, the v incre ith th ease o fre uency r S mixture. Therefo ing E to pr icting e r ng fo 7. Th data lys n sta c cre te sults d a si nt dif e be een an 5 gyr ions ow r, the f the t sults an long ting e lim ted t us this t r pred th MA xtu . 8. Th num ame r less o S. N ficant ev ence s o rved at di re mpac evels de gning A xture ith ffe MA e AN n AP ng, st cre and peate load test si fica flu e fac r. ented apter 5, the ing commendations were m de: 281 1. When using the vacuum seal method to determine the air void content, different correction factors are necessary for different combinations of surface conditions, M d void range te d sa . A sa with a er su igher NMA nd er o generally requires a higher correlation factor. he lat fac r vacuu m od can alibrate the dif e a s b ee S me od a correc lculat r the v al od. e s shou e low ids an m rfa con n e ple tested 5 g ons rec ended as the op com n level in terms o ver erformance in m ra ity a tting resistance. val n o eld or ce SM igh sira to y ab ator result N AS an air s of ste mples mple rough rface, h S a high air v ids 2. T corre ion tor fo m seal eth be c d by ferenc in ir void etw n the SD th nd un ted ca ion fo acuum se meth Th ample used for calibration ld hav air vo d si ilar su ce ditio as th sam to be . 3. 6 yrati is omm t mimu pactio f the o all p ix du bil nd ru 4. E uatio f fi perf man of A mixtures designed with 65 gyrations is h ly de ble verif the l or y test s. 282 RE S S , ? por he 0 rop sphalt y Tour shington D.C., ne 1. u nes Ma ls du ion, lacem Stone ix Asp , ti Asp lt P e ss atio rmati ries 11 4, 199 merican Society for Testing M rials (ASTM) Standards on , Volum .02 oncr an gr es 004. avr .R nd C nt ?Ca ing A ds at S ied Num of yra s in m ?. In Transportation Research Record 630 B, tion s C nci hingt .C., 19 rto , L. H an g, . ?Th ratory r Meth Moldi sp Co rete t Specim s; Its elopm d Cor n with d om on tho s hw partm andard edure? u f A cia f ha on a, A D.H tl V uint d J. B ?Com ve evaluation l ory mp n ce ased eir ability to produce mixtures with g ng per im to ose ced in ield?, portati ese Re d 1 T N nal rch C l, Was on, D.C 89. utt .W. .N. e, , .J. Pe n. ?Co tion of ele Lab tor m io ethods with Field Compaction?, In ran tatio Res R rd 54, National Research Council, as ton, C., . an hip .B., . M o , and G.A. Huber. ?Rational Method for bo ry C paction of Hot-Mix Asphalt?, In Transportation Research Record 54 B, tional Res C nci hingt .C., 19 arv J., C.L. Mo mith, and J. Sousa. ?A Investigation of Field- and bo ry-C pa A lt ubb MA R ed and ention sphalt ? Co rete es ng HRP 03A E ent?, Journal of sso ion sp Pa T hno , Vol 994. Sousa, J. Harvey Pa , J eac d C.L. ismith luatio b y Procedures fo m ting alt-A ate Mi ?, SH c Memorandum No. TM-UCS-A-003A-9 nivers Califo er 19 . REFE NCE 1. AA HTO 199 Re t on t 199 Eu ean A Stud ?, Wa Ju 2. ?G ideli for teria , Pro ct and P ent of Matr halt? Na onal ha avem nt A oci n, Info on Se 8, 8/9 4. 3. A ate Disc e 04 , C ete d Ag egat , 2 4. V ik, W ., a S.H. arpe er. lculat ir Voi pecif ber G tion Superpave Gyratory Co pactor 1 , TR Na al Re earch ou l, Was on, D 98. 5. O lani and .A. S dber Jr e Gy -Shea od of ng A haltic nc Tes en Dev ent an relatio Fiel C pacti Me ds. A Texa Hig ay De ent St Proc , Jo rnal o sso tion o Asp lt Paving Technologists, Vol: 21, 1952. 6. C suegr ., . Lit e, H. . Q us, an urati, parati of aborat in i co actio devi s b on th en eer pro ties s ilar th produ the f Trans on R arch cor 228, RB, atio Resea ounci hingt ., 19 7. B on, J , D Littl V. Jagadam and O ndleto rrela S cted ora y Co pact n M T spor n earch eco 14 TRB, W hing D. 1994 8. Bl kens , R K.C ahb ub La rato om 14 , TR Na earch ou l, Was on, D 94. 9. H ey, nis La rato om cted spha ? R er, S ecycl Conv al A nc Mix Usi S A-0 quipm A ciat of A halt ving ec logists : 63, 1 10. J. , L. inter . D on an Mon . ?Eva n of La orator r Co pac Asph ggreg xtures RP Te hnical 0-5, U ity of rnia, B keley, 91 283 11. Cominsky, R., Leahy, R.B., and Harrigan, E.T., ?Level One Mix Design: Materials Selection, Compaction, and Conditioning.? Strategic Highway Research Program Report No. A-408, National Research Council, Washington D.C., 1994. u ., Way, J.T. e nd M es. ?C arison x Desi on ?. I ransportation Research Record 1492, TRB, National Research Council, Wa ngto .C 99 afe H. a M.W itczak. ?Comp of Marshall and Superpave Level I ix ign Asp t M s? Tra tation arch R 1492, , ati Res rch C nc as ngto ., 19 ?A o, J ., Pa , C ar T nd Bu ki, J., ? parison of the pe e Gyratory Com or Mar for Field Quality Control.? Journal of e A ciat of A a vi Technologists, Volume 64, 1995, pp. 611-635. cG is, R ., R An o . P o and P. Turner. ?Issues Pertaining U f Su pav ra C pac n Tra tation Research Re 43 B, tion es C nci hingt .C., 19 row .R. .I. H on R . M . ?Evaluation of Superpave Gyratory om ion Hot x A al In T ortati search rd 154 B, National Research Council, Was ngto ., 19 rs . A nd C m K. Dete tion of Key Gyra ompac in r Su rpav ix ig n Arizona.? AS pecial ical b n, lum 22 pt ber AST ladelp A., p 9 al .B . B na .R row M. H ?An E tion o p e G tory m io f Ho Asph NCAT rt No. ub L 98. 19. Anderson, R. M., R.B. McGennis, W. on Tam, and T. W. Kennedy, ?Sens M e P orm e P ert to Changes in Laboratory C action Using e pav yra C actor ?, Journal of Association of Asphalt Paving ch gis Vol 2 uc n, M and . B n, ffec uperp yrator pactor Type C acted Hot-Mix Asphalt Density? In Transportation R h Reco 61 B, tion es C nci hingt .C., 20 ow .R. Gab so d S. Adettiwar. ?Va in Ho Asph esi rop es? rna Association of Asphalt Paving Technologists, Vol: , 1 . 2 ash . Ma , R te n, an Saleg rnal S re An f sph Mix to Im v S ulation of Superpave Gyratory Compaction to ield ndit s?. na th ssoc of A t Pavin hnolog ol. 200 12. So sa, J.B G. Harv y a . Hin omp of Mi gn C cepts n T shi n, D ., 1 5. 13. H z, I. nd . W arison M Des for ea hal ou ixe . In nspor Rese ecord TRB N onal il, W hi n, D.C 95. 14. D ngel . A ugh ., H man, . P., a kows Com Su rpav pact to shall th sso ion sph lt Pa ng 15. M enn .B .M. ders n, D erdom to se o per e Gy tory om tor?. I nspor cord 15 , TR Na al R earch ou l, Was on, D 96. 16. B n, E , D ans , and .B allick C pact of -Mi sph t?. ransp on Re Reco 3, TR hi n, D.C 96. 17. Fo tie, D . a oru , D. , ? rmina tory C tion Po ts fo pe e M Des n i TM S Techn Pu licatio Vo e 13 , Se em 1997. M, Phi hia, P p. 201- 20 . 18. M lick, R ., S ucha n, E . B n and uner. valua f Su erpav yra Co pact n o t Mix alt?, Repo 98-5, A urn, A , 19 itivity of ixtur erf anc rop omp ies omp th Super e G tory Te nolo ts, : 69, 000. 20. B hana S E.R row ?E t of S ave G y Com on omp esearc rd 17 , TR Na al R earch ou l, Was on, D 01 21. Br n, E , J. riel n, an riation t Mix alt D gn P erti Jou l of 62 993 2. T man, L., E sad . Pe rso d H. . ?Inte tructu alysis o A alt es pro e the im F Co ion Jour l of e A iation sphal g Tec ists, V 70, 1. 284 23. Harman, T., J.R. Bukowski, F. Moutier, G. Huber, and R. McGennis. ?The History and Future Challenges of Gyratory Compaction 1939 to 2001?. In Transportation Research Record 1789, TRB, National Research Council, 2 eter , R. K.C ahboub, R.M. Anderson, E. Masad and L. Tashman Sup ve ora C ac n ve ield C action?. In Transportation es Re d 18 T National rch C l, Was on, D.C 03. 2 A ciat f State Highwa O s (AA ), tion Materials and Methods of Samp nd ng, rt 2A d 2B: Tests, 24th Edition, 2004 2 tu D. tone sti ph ed igh y A ni ion, Marc 92. 2 en l, ., an K. id n. ?I uction one M Aspha A) A lt Pav echnol , Vol: 992 2 row .R. valu n M used ichiga 91)?, N Repor 3-3 tion en or ha ech y, Au AL, 1 2 row .R. xpe ce S ne m Asphalt in United States?, NC ep o. 9 , N na nte or A t Tech y, Au AL, 19 row .R. d H an ka Eva n of L tory Properties of SMA ixt ?, N T Report 5, tion ter fo halt Te logy, A n, , . rp r, R . ?M e C iderations for SMA Mixes? ented ansportati Res h B d nual ting, 1 ow .R. d R M k. tone ix Asphalt- Properties Related ixture Des ?, N T or o. 94-2, National Center for Asphalt Technology, bur L 4. o , W , an .D ar Eva n of S atrix alt ve n ade ixt ?, ra t Res rch Counc as ngto ., 19 rt ., . V n, . o ce-Re s f S e M A alt ourn Assoc of As Paving c gis Vol 1 e . a B.E th om ctio Shea gth T of Sto at sph Mix tory Testing Machine?, Journal of so ion sp Pa T hno , Vol 995. Washington, D.C., 2002. 4. P son L., . M . ? erpa Lab tory omp tio rsus F omp R earch cor 32, RB, Resea ounci hingt ., 20 5. American sso ion o y and Transportation fficial SHTO Standard Specifications for Transporta ling a Testi Pa an Part 6. S art, K. ?S Ma c As alt (SMA) Mixture Design?, FHWA-RD-92-006 F eral H wa dmi strat h 19 7. K nepoh G.J d J. Dav so ntrod of St astic lt (SM in Ontario?, Journal of Association of spha ing T ogists 61, 1 . 8. B n, E ?E atio of S A in M n (19 CAT t No. 9 , Na al C ter f Asp lt T nolog burn, 993. 9. B n, E ?E rien with to atrix AT R ort N 3-4 atio l Ce r f sphal nolog burn, 93. 3 M 0. B n, E , an . M glor r. ? luatio abora ures CA 93- Na al Cen r Asp chno ubur AL 1993 31. Ca ente .H ix D sign ons , Pres in Tr on earc oar An Mee 994. 32. Br n, E , an .B. allic ?S Matr to M ign CA Rep t N Au n, A , 199 33. M gawer .S. d K . Stu t. ? luatio tone M Asph rsus De se-gr d M ures In T nsportation Research Record 1454, TRB, Na ional ea il, W hi n, D.C 94. 34. Pa l, M.N T.S iniso R.G Hicks, and K. Younger. ?Perf rman lated Te ting o ton astic sph ? J al of iation phalt Te hnolo ts, : 64, 995. 35. W st, R.C nd . Ru . ?C pa n and r Stren esting ne M rix A alt tures in the Gyra As ciat of A halt ving ec logists : 64, 1 285 36. Brown, E.R, and J.E. Haddock, T.A. Lynn and R.B. Mallick. ?Designing SMA Mixtures Volume II- Research Results?, NCHRP 9-8/2 Draft Final Report, September 1996. ow .R. .B. lic E. ddo d J. B ski. Performance of Stone at sph (SM M res in the United Sta Journ he so ion sp Pa T hnologists, Volume 66, 199 uw ., C. em ink d JA aeghe velopm f a Sto ast sph Des M d f r South African Conditions?, Eighth ter onal nfe e sp lt Pa nts, V Seattl shingt ugu 997 ow .R. E. H oc .B alli d T.A n. ?De ment o ixt Des Pro ur S e Matrix Asphalt (SMA)?, Journal of so ion sp Pa T hno , Vol 997. own, E.R. d J a k. eth Ensur ne-on- Conta o trix sph av M ures ransp on Research Record 1583, B ion Res h C cil, Washington, D.C., 1997. e T. ap- e ry d M Meld , Super Design ol No 99 o .R. d L. o , Jr. NCHRP Report Desig MA ix for t-R tan ve nts?. Transportation Research Board, as ton, C., 9. P Des ing C ru ng S atrix halt (S Mixtur ate the- ctic 1999 (Updated edition is published in 2002). S , AASHTO ro n e , T. tone tr sp t is ing on e U.S.? etter R , p. 3. sch rene . ? lt Su ey on Stone Matrix Asphalt?, Colorado epo ent Tran rta , A il 20 merican Society for Testing M rials (ASTM) Standards on , Volu 4.0 oad Pa M ria Vehicle-Pavem ystems, 2004. rou L.K t al. er ng ir V onten ompac ot-mix sph Mix es. T sp io ese ecord , TRB onal R ch oun Wa ngto .C 00 pp39 oo .A al, S fi ravi und-R Using oreLo ac ea g D e. T epor 02-11 onal C for A ec gy, bur la a, 2002. 37. Br n, E , R Mal k, J. Ha ck, an ukow M rix A alt A) ixtu tes. In al of t As ciat of A halt ving ec 7. 38. Lo , L J. S mel , an BM . Verh . ?De ent o ne M ic A alt ign etho o In A nati Co renc on A ha veme ol. I., e, Wa on, st 1 . 39. Br n, E , J. add k, R . M ck, an . Lyn velop f a M ure ign ced e for ton As ciat of A halt ving ec logists : 66, 1 40. Br , an .E. H ddoc ?M od to e Sto Stone ct in St ne Ma A alt P ing ixt ?, In T ortati TR , Nat al earc oun 41. Ku nnen, ?G Grad d Ma lan ixes SMA pave s?, HMAT, V . 4, .2, 1 9. 42. Br wn, E an A. C oley 425. ? ning S M tures Ru esis t Pa me W hing D. 199 43. NA A, ? ign and onst cti tone M Asp MA) es - St -of- Pra e?, 44. AA HTO P visio al Standards, May 2002 Edition, 2002. 45. Ku nnen ?S Ma ix A hal Catch in th , in B oads Se 200 46. A enb r, T Resu s of rv D rtm of spo tion pr 04. 47. A ate Disc me 0 3, R and ving ate ls; ent S 48. C ch, ., e Det mini A oid C t of C ted H A alt tur ran ortat n R arch R 1813 , Nati esear C cil, shi n, D ., 2 2, -46. 49. C ley. L . et Bulk peci c G ty Ro obin the C k V uum S lin evic NCA R t No. . Nati enter sphalt T hnolo Au n, A bam 286 50. Hall, K.D., F.T. Griffith, and Stacy G. Williams. Examination of Operator Variability for Selected Methods for Measuring Bulk Specific Gravity of Hot-Mix Asphalt Concrete. Transportation Research Record 1761, TRB, National Research ou Wa ngto .C 00 pp81 uc , M . An alu n Sele etho Measu the Bu pe ra y of p d t Mi halt ( ) Mixe rnal o sso ion sp Pa T hno , Vol eno, N 000, pp608- 34. ud S.B nd R D . ? lationship of Aggregate Voidage to Gradation? P din Asphalt Pavi yn A, E . Br , and L.A. Cooley, Jr. ?Evaluation of Aggregate Si ha risti in S M phalt and Superpave Mixtures?, In ran tatio Res h R rd 81, National Research Council, as ton, C., 9. 54. L.A. Cooley, Jr. and E.R. Brown. ?Po Thin Overlays?, In Transportation Research Record 1749, TRB, National Research Council, Washington, D.C., 2001. 55. Kanitpong, K, C.H. Benson, and H.U. Bahia. ?Hydraulic Conductivity (Permeability) of Laboratory-Compacted Asphalt Mixtures?, In Transportation Research Record 1767, TRB, National Research Council, Washington, D.C., 2001. 56. Prowell, B.D, L.A. Cooley, Jr, and R.J. Schreck. ?Virginia's Experience with A 9.5-mm Nominal-Maximum-Aggregate-Size Stone Matrix Asphalt?, In Transportation Research Record 1813, TRB, National Research Council, Washington, D.C., 2002. 57. Mallick, R.B, L.A. Cooley, Jr., M.R. Teto, R.L. Bradbury, and D. Peabody. ?An Evaluation of Factors Affecting Permeability of Superpave Designed Pavements?, NCAT Report No. 03-02, National Center for Asphalt Technology, Auburn, AL, 2003. 58. Hainin, MR, and L.A. Cooley, JR. ?An Investigation of Factors Influencing Permeability of Superpave Mixes? in International Journal of Pavements, Vol.2, No.2, 2003. 59. Choubane, B., G.C. Page, and J.A. Musselman. ?Investigation of Water Permeability of Coarse Graded Superpave Pavements.? Journal of the Association of Asphalt Paving Technologists, Vol. 67, 1998. 60. Moavenzadeh, F. and W.H. Geotz. ?Aggregate Degradation in Bituminous Mixtures.? In Highway Research Record 24, HRB, National Research Council, Washington, D.C., 1963. C ncil, shi n, D ., 2 2, -85. 51. B hanan .S Ev atio of cted M ds for ring lk S cific G vit Com acte Ho x Asp HMA s. Jou f the A ciat of A halt ving ec logists . 69. R V, 2 6 52. H son, ., a gs of the Association of .L. avis Re In rocee ng Technologists, Vol. 34, 1965. 53. L n, T. .R own ze C racte cs tone atrix As T spor n earc eco 16 TRB, W hing D. 199 tential of Using Stone Matrix Asphalt for 287 61. Subramanyam, B. and M.P. Pratapa. ?Degradation of Dense Aggregate Gradings.? In Transportation Research Record 1228, TRB, National Research Council, Washington, D.C., 1989. 62. Collins, R., D. Watson, A. Johnson, and Y. Wu. ?Effect of Aggregate Degradation on Specimens Compacted by Superpave Gyratory Compactor.? In Washington, D.C., 1997. 63. Aho, B.D., W.R. Vavrik, and S.H. Carpenter. ?Effect of Flat and Elongated Coarse Aggregate on Field Compaction of Hot-Mix Asphalt.? In Transportation Research Record 1761, TRB, National Research Council, Washington, D.C., 2001. 64. Brown, E.R., L.A. Cooley, Jr., and T.A. Lynn, ?Designing Stone Matrix Asphalt Mixtures Volume II (c) ? Research Results for Phase II?, Final Report for NCHRP 9-8, Transportation Research Board, 1998. 65. Buchanan, M.S. ?Evaluation of the Effect of Flat and Elongated Particles on Performance of Hot Mix Asphalt Mixtures.? NCAT Report No. 00-03. National Center for Asphalt Technology, Auburn, Alabama, 2000. 66. Vavrik, W.R., R.J. Fries, and S.H. Carpenter. ?Effect of Flat and Elongated Coarse Aggregate on Characteristics of Gyratory Compacted Samples?. In Transportation Research Record 1681, TRB, National Research Council, Washington, D.C., 1999. 67. Buttlar, W.G. and M. Harrell. ?Development of End-Result and Performance- Related Specifications for Asphalt Pavement Construction in Illinois? Crossroads 2000 Transportation Conference Proceedings, Ames, Iowa, 1998. 68. Georgia Department of Transportation (GDOT), Standard Specification for Construction of Transportation Systems, 2001 Edition with Special Provisions, State of Georgia, 2001. 69. Alabama Department of Transportation (ALDOT), Alabama Standard Specifications for Highway Construction, 2002 Edition with Special Provisions, Alabama, 2002 70. Zhang, J; L.A.Cooley, Jr, and P.S. Kandhal, ?Comparison of Fundamental and Simulative Test Methods for Evaluating Permanent Deformation of Hot-Mix Asphalt?, in Journal of Transportation Research Record 1789, TRB, National Research Council, Washington D.C., 2002. 71. Brown, E.R., P.S. Kandhal, and J. Zhang. ?Performance Testing for Hot Mix Asphalt?, NCAT Report No.01-05, Auburn, AL 2001. 72. Lai, J.S. ?Development of a Simplified Test Method to Predict Rutting Characteristics of Asphalt Mixtures? Final Report, Research Project No.8503, Georgia DOT, 1986. Transportation Research Record 1590, TRB, National Research Council, 288 73. West, R.C, G.C. Page, and K.H. Murphy. ?Evaluation of the Loaded Wheel FL/DOT/SMO/91-391, 1991. 7 ille . L search Record 1454, TRB, National ashington D.C., 1 75 houbane, B. age, and J.A. Mu , ?Suitability of As avement or P ng Pavement Rutt Journal of Transpo Research ashington D.C., 76 ad, L.N, Z. Wu, L. Wang, and C. Abadie, ?A Rut Susceptibility Study f Superpave t Mixtures? in Inte al Journal of Pavem JP 2002, ol. 1, No.2, 7 esTrack Forensic Team. ?Performanc oarse-Graded Mixe sTrack- ature Ru Final Report, 1998 78 est, R.C. ?A edness Study of the Asphalt Pavement Analyzer Rutting orandum to the Asphalt Pav Analyzer User Group and New 7 ennert, T., A r, and I. Marukic. ?Characterization of NJ HMA?, Final 8 ?Accelerated Laboratory Rutting Tests: ent Analyzer?, NCHRP Report 5 B, 8 illims, C.R. and B.R. Prowell. ?Com f Laboratory Wheel-Tracking est Results t rack Performance ansportation Resear ord 1681, ouncil, Washington DC, 1999. 8 uchanan, M. . White, and B.J. S Use of the Asphalt ent lt Mixture Performance? FHWA/MS-DOT- ssissippi State, MS, 2004. racterization of Asphaltic Mixtures for Prediction of Permanent Deformation? in Journal of Transportation Research Record 1034, TRB, Washington DC, 1985. 84. Witczak, M.W., K. Kaloush, T. Pellinen, M.E. Basyouny, and H.V. Quintus. ?Simple Performance Test for Superpave Mix Design? NCHRP Report 465. TRB, Washington DC, 2002. 85. Sousa, J.B. and S.L. Weissman. ?Modeling Permanent Deformation of Asphalt Aggregate Mixes? in Journal of the Association of Asphalt Paving Technologist, Vol. 63, St. Paul, Minnesota, 1994. Tester? Research Report 4. M r, T., K. Ksaibati, and M Farrar. ?Using Georgia oaded-Wheel Tester to Predict Rutting? In Transportation Re Research Council, W 994. . C , G.C. P sselman phalt P Analyzer f redicti ing?, in rtation Record 1723, TRB, W 2000. . Mohamm o Asphal rnation ents, I V 2002. 7. W e of C s at We Prem tting.? . . W Rugg Test.? Mem ement APA Owners, 1999. 9. B . Mahe Report, FHWA-NJ-2002-027, 2003. 0. Kandhal P.S., and L.A. Cooley, Jr. Evaluation of the Asphalt Pavem 08, TR Washington DC, 2003. 1. W parsion o T o WesT ? In Tr ch Rec TRB, National Research C 2. B S., T.D mith ? Pavem Analyzer to Study In-Service Aspha RD-04-155, Final Report, Mi 83. Khosla, N.P., and M.S. Omer. ?Cha 289 86. Bonaquist, R.F., D.W. Christensen, and W. Stump, III. ?Simple Performance Tester for Superpave Mix Design: First-Article Development and Evaluation? NCHRP report 513, TRB, Washington DC, 2003. 87. Pellinen, T.K., and M.W. Witczak. ?Use of Stiffness of Hot-Mix Asphalt as A Simple Performance Test?. In Transportation Research Record 1789, TRB, 88. Brown, S.F. and J.M. for Permanent Deformation Testing of Bituminous Mixtures?, in Journal of the Association of Asphalt Paving Technologist, Vol. 65, St. Paul, Minnesota, 1996. 89. Gabrielson, J. R. ?Evaluation of Hot Mix Asphalt (HMA) Static Creep and Repeated Load Tests? Ph.D. Dissertation, Auburn University, 1992. 90. Foo, K.Y. ?Predicting Rutting in Hot Mix Asphalt? Ph.D. Dissertation, Auburn University, 1994. 91. Kaloush, K.E and M.W. Witczak, ?Tertiary Flow Characteristics of Asphalt Mixtures (With Discussion and Closure)? in Journal of the Association of Asphalt Paving Technologists, Vol. 71, Colorado Springs, Colorado, 2002. 92. Bhasin, A, J.W. Button, and A. Chowdhury. ?Evaluation of Simple Performance Tests on Hot-Mix Asphalt Mixtures from South Central United States? in Transportation Research Record 1891, TRB, Washington D.C. 2004. 93. Birgisson, B, R. Roque, J. Kim, L.V. Pham. ?The Use of Complex Modulus to Characterize the Performance of Asphalt Mixtures and Pavements in Florida?, UF Project 4910-4504-784-12, University of Florida, Gainesville, FL, 2004. 94. Sousa, J.B., J. Craus, and C.L. Monismith. ?Summary Report on Permanent Deformation in Asphalt Concrete? Strategic Highway Research Program, Report No. SHRP-A/IR-91-104, Washington DC, 1991. 95. Little, D. N., J.W. Button, and H. Youssef. ?Development of Criteria to Evaluate Uniaxial Creep Data and Asphalt Concrete Permanent Deformation potential?. in Transportation Research Record 1417, TRB, Washington D.C. 1993. 96. Brown, E.R., and S.A. Cross. ?A National Study of Rutting in Hot Mix Asphalt (HMA) Pavements?, In Journal of the Association of Asphalt Paving Technologists, Vol. 61, 1992. 97. Harvey, J, S. Weissman, F.Long, and C. Monismith, ?Tests to Evaluate the Stiffness and Permanent Deformation Characteristics of Asphalt/Binder- Aggregate Mixes, and Their Use in Mix Design and Analysis (With Discussion)? in Journal of the Association of Asphalt Paving Technologists, Vol. 70, Clearwater Beach, Florida, 2001. 98. Valkering, C.P., J.L. Lancon, E. de Hiltseter, and D.A. Stoker. ?Rutting Resistance of Asphalt Mixes Containing Non-Conventional and Polymer- Washington DC, 2002. Gibb. ?Validation Experiments 290 Modified Binders? In Journal of the Association of Asphalt Paving Technologists, Vol. 59, 1990. 99. Tanco, A.J. ?Permanent Deformation Response of Conventional and Modified Asphalt-Aggregate Mixes under Simple and Compound Shear Loading Conditions?, Ph.D. Dissertation, University of California, Berkeley, 1992. 100. Brown, S.F. and M.S. Snaith. ?The Permanent Deformation Characteristics of a Dense Bitumen Macadem Subjected to Repeated Loading? in Journal of the Association of Asphalt Paving Technologist, Vol. 43, St. Paul, Minnesota, 1974. 101. Pell, P.S. and S.F. Brown. ?The Characteristics of Materials for the Design of Flexible Pavements Structures? Proceedings, Third International Conference on the Structural Design of Asphalt Pavements, London, U.K. 1977. 102. Hofastra, A. and A.J. Klomp. ?Permanent Deformation of Flexible Pavements Under Simulated Road Traffic Conditions? Proceedings, Third International Conference on the Structural Design of Asphalt Pavements, London, U.K. 1977. 103. Prowell, B.D., J. Zhang, and E.R. Brown. ?Aggregate Properties and Their Relationship to the Performance of Superpave-Designed HMA: A Critical 104. Roberts. F.L., P.S. Kandhal, E.R. Brown, D-Y, Lee, and T.W. Kennedy. ?Hot Mix Asphalt Materials, Mixture Design, and Construction?. Second Edition. National Asphalt Pavement Association, Research and Education Foundation. Lanham, Maryland, 1996. 106. AASHTO, AASHTO Provisional Standards, June 2004 Edition, 2004. 107. Ferry, J. D., ?Viscoelastic Properties of Polymers?, 2 nd edition, Wiley, New York, 1970. 108. Harvey, J., J.B. Sousa, J.A. Deacon, C.L Monismith. ?Effects of Sample Preparation and Air-Void Measurement on Asphalt Concrete Properties?. In Transportation Research Record 1317, TRB, National Research Council, Washington, D.C., 1991. 109. Anderson, D.A. ?Guidelines on Use of Baghouse Fines?. NAPA, Information Services 101, 1987. 110. InstroTek, Incorporated. ?CoreLok Operator?s Guide?, Raleigh, NC, 2003. 111. Florida DOT, ?Section 334-Superpave Asphalt Concrete?, Standard Specifications for Road and Bridge Construction, State of Florida, 2004. 112. Clyne, T.R., X. Li, M.O. Marasteanu, and E.L. Skok. ?Dynamic and Resilient Modulus of MN/DOT Asphalt Mixtures?. MN/DOT Final Report, St. Paul, Minnesota, 2003. Review?, NCHRP 9-35 Final Report, TRB, 2004. 105. ASTM, ASTM Provisional Specifications, 2001 291 113. Kandhal, P.S., and F. Parker, Jr. NCHRP Report 405: Aggregate Tests Related to Asphalt Concrete Performance in Pavements, Transportation Research Board, National Research Council, Washington, DC, 1998. 114. Rismantojo, E. ?Permanent Deformation and Moisture Susceptibility Related aggregate Tests for Use in Hot-Mix Asphalt Pavements,? Doctoral Thesis, Purdue University, West Lafayette, IN, 2002. 115. Mallick, R.B., M.R. Teto, and J.E. Haddock. ?Use of the Concept of Pore Pressure in Unsaturated Soils for Evaluation of Rutting Potential of Asphalt Paving Mixtures?. In Proceeding of Ninth International Conference on Asphalt Pavements, ISAP, Copenhagen, Denmark 2002. 292 APPENDIX A AIR VOID CONTENT AND PERMEABILITY TEST RESULTS 293 TABLE A1 Air Voids and Permeability Test Results Agg. Type NMAS, mm Grad. Gyrs AC, % ID Air Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s C.GVL 19 N 100 5.5 1 4.13 5.58 0.41 119.1 C.GVL 19 N 100 5.5 2 4.09 5.94 0.35 68.4 C.GVL 19 N 100 5.5 3 3.90 5.47 0.33 109.6 C.GVL 19 N 100 6.0 1 3.53 5.58 0.45 318.2 C.GVL 19 N 100 6.0 2 3.78 5.15 0.37 99.2 C.GVL 19 N 100 6.0 3 3.75 5.34 0.43 2.27 C.GVL 19 N 100 6.5 1 3.57 4.87 0.28 11.4 C.GVL 19 N 100 6.5 2 3.12 4.35 0.20 0.00 C.GVL 19 N 100 6.5 3 2.87 3.79 0.21 62.5 C.GVL 19 N 100 7.0 1 1.99 3.20 0.14 0.00 C.GVL 19 N 100 7.0 2 2.37 3.60 0.23 0.00 C.GVL 19 N 100 7.0 3 2.00 2.76 0.11 0.00 C.GVL 12.5 N 100 6.0 1 4.74 5.63 0.34 1.63 C.GVL 12.5 N 100 6.0 2 4.89 6.36 0.64 52.5 C.GVL 12.5 N 100 6.0 3 4.54 5.89 0.51 21.12 C.GVL 12.5 N 100 6.5 1 4.12 5.16 0.30 0.00 C.GVL 12.5 N 100 6.5 2 4.16 5.51 0.30 54.8 C.G VL 12.5 N 100 6.5 3 4.08 5.12 0.32 10.3 C.GVL 12.5 N 100 7.0 1 2.81 3.37 0.15 0.00 C.GVL 12.5 N 100 7.0 2 2.73 3.43 0.21 0.00 C.GVL 12.5 N 100 7.0 3 2.64 3.53 0.17 3.6 C.GVL 9.5 N 100 6.0 1 4.51 5.34 0.35 22.6 C.GVL 9.5 N 100 6.0 2 4.50 5.16 0.43 3.48 C.GVL 9.5 N 100 6.0 3 4.55 5.44 0.49 1.27 C.GVL 9.5 N 100 6.5 1 3.50 4.25 0.24 0.67 C.GVL 9.5 N 100 6.5 2 3.73 4.12 0.18 0.29 C.GVL 9.5 N 100 6.5 3 3.85 4.59 0.18 3.3 C.GVL 9.5 N 100 7.0 1 2.05 2.95 0.12 0.00 C.GVL 9.5 N 100 7.0 2 2.30 2.84 0.14 0.00 C.GVL 9.5 N 100 7.0 3 2.01 3.47 0.11 0.39 L.GRN 19 N 100 4.5 1 4.63 6.27 0.57 503.4 L.GRN 19 N 100 4.5 2 4.06 5.57 0.42 447.4 L.GRN 19 N 100 4.5 3 4.5 6.31 0.64 203.2 L.GRN 19 N 100 5.0 1 2.9 4.27 0.24 2.6 L.GRN 19 N 100 5.0 2 3.08 4.69 0.30 45.3 L.GRN 19 N 100 5.0 3 3.09 4.3 0.31 5.0 L.GRN 19 N 100 5.5 1 4.1 5.88 0.51 30.6 L.GRN 19 N 100 5.5 2 2.85 4.33 0.22 28.4 L.GRN 19 N 100 5.5 3 3.14 4.32 0.33 8.8 L.GRN 19 N 100 6.0 1 2.41 3.32 0.17 6.2 L.GRN 19 N 100 6.0 2 2.32 3.06 0.17 13.8 L.GRN 19 N 100 6.0 3 1.98 2.98 0.16 3.6 L.GRN 12.5 N 100 5.0 1 4.31 5.46 0.43 55.4 L.GRN 12.5 N 100 5.0 2 4.39 5.45 0.48 42.2 294 Agg. Type NMAS, mm Grad. Gyrs AC, % ID Air Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s L.GRN 12.5 N 100 5.0 3 4.83 6.17 0.54 51.6 L.GRN 12.5 N 100 5.5 1 3.86 5.08 0.36 48.1 L.GRN 12.5 N 100 5.5 2 3.76 4.97 0.43 18.4 L.GRN 12.5 N 100 5.5 3 3.67 4.81 0.27 21.2 L.GRN 12.5 N 100 6.0 1 3.05 3.87 0.23 1.8 L.GRN 12.5 N 100 6.0 2 3.21 4.28 0.23 17.2 L.GRN 12.5 N 100 6.0 3 3.18 4.25 0.29 15.9 L.GRN 12.5 N 100 6.5 1 1.81 2.58 0.10 0.00 L.GRN 12.5 N 100 6.5 2 1.14 1.64 0.06 0.19 L.GRN 12.5 N 100 6.5 3 1.53 2.14 0.07 0.19 L.GRN 9.5 N 100 5.5 1 4.09 5.22 0.30 5.7 L.GRN 9.5 N 100 5.5 2 4.41 5.08 0.31 13.3 L.GRN 9.5 N 100 5.5 3 4.47 5.21 0.21 18.4 L.GRN 9.5 N 100 6.0 1 3.12 3.88 0.20 1.9 L.GRN 9.5 N 100 6.0 2 3.46 3.94 0.23 0.68 L.GRN 9.5 N 100 6.0 3 3.27 3.82 0.16 1.3 LMS 19 N 100 5.0 1 4.76 6.22 0.54 544.4 LMS 19 N 100 5.0 2 4.39 5.52 0.40 33.1 LMS 19 N 100 5.0 3 3.93 5.13 0.34 31.1 LMS 19 N 100 5.5 1 3.09 4.15 0.23 0.28 LMS 19 N 100 5.5 2 2.71 3.83 0.21 1.3 LMS 19 N 100 5.5 3 3.06 4.06 0.25 44.0 LMS 19 N 100 6.0 1 3.17 0.12 0.27 2.16 LMS 19 N 100 6.0 2 2.14 3.14 0.13 0.00 LMS 19 N 100 6.0 3 2.25 2.8 0.11 0.00 LMS 12.5 N 100 5.0 1 4.92 5.79 0.37 53.8 LMS 12.5 N 100 5.0 2 5.48 6.52 0.54 83.7 LMS 12.5 N 100 5.0 3 5.12 5.91 0.47 114.8 LMS 12.5 N 100 5.5 1 3.8 4.63 0.25 5.4 LMS 12.5 N 100 5.5 2 3.8 4.52 0.25 0.00 LMS 12.5 N 100 5.5 3 4.05 4.79 0.27 17.3 LMS 12.5 N 100 6.0 1 3.17 4 0.17 0.29 LMS 12.5 N 100 6.0 2 2.92 4.72 0.16 0.39 LMS 12.5 N 100 6.0 3 2.88 3.62 0.13 8.4 LMS 12.5 N 100 6.5 1 2.51 3.1 0.17 0.67 LMS 12.5 N 100 6.5 2 2.69 3.88 0.26 0.1 LMS 12.5 N 100 6.5 3 2.11 2.8 0.15 0.57 LMS 9.5 N 100 5.0 1 4.98 5.83 0.37 17.9 LMS 9.5 N 100 5.0 2 5.11 5.79 0.43 5.9 LMS 9.5 N 100 5.0 3 4.71 5.32 0.23 0.5 LMS 9.5 N 100 5.5 1 3.62 4.22 0.14 7.5 LMS 9.5 N 100 5.5 2 3.48 3.9 0.21 0.00 LMS 9.5 N 100 5.5 3 3.64 4.28 0.18 1.7 LMS 9.5 N 100 6.0 1 2.41 2.97 0.10 0.00 LMS 9.5 N 100 6.0 2 2.14 2.57 0.09 0.00 295 Agg. Type NMAS, mm Grad. Gyrs AC, % ID Air Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s LMS 9.5 N 100 6.0 3 2.46 3.08 0.10 1.8 R.GRN 19 N 100 6.0 1 5.75 8.32 0.66 644.0 R.GRN 19 N 100 6.0 2 5.53 7.56 0.55 85.5 R.GRN 19 N 100 6.0 3 5.37 8.02 0.65 360.7 R.GRN 19 N 100 6.5 1 4.57 6.97 0.44 285.7 R.GRN 19 N 100 6.5 2 3.51 5.46 0.33 1.6 R.GRN 19 N 100 6.5 3 3.12 4.85 0.27 5.2 R.GRN 19 N 100 7.0 1 2.14 3.17 0.20 1.6 R.GRN 19 N 100 7.0 2 2.57 4.04 0.23 160.6 R.GRN 19 N 100 7.0 3 2.24 3.9 0.19 68.4 R.GRN 12.5 N 100 6.0 1 6.05 8.22 0.56 494.5 R.GRN 12.5 N 100 6.0 2 6.25 7.8 0.44 319.1 R.GRN 12.5 N 100 6.0 3 5.92 7.67 0.46 193.7 R.GRN 12.5 N 100 6.5 1 4.71 6.05 0.38 76.3 R.GRN 12.5 N 100 6.5 2 4.97 6.16 0.39 8.3 R.GRN 12.5 N 100 6.5 3 4.14 5.53 0.34 67.2 R.GRN 12.5 N 100 7.0 1 4.08 5.16 0.30 2.1 R.GRN 12.5 N 100 7.0 2 3.54 4.78 0.25 0.1 R.GRN 12.5 N 100 7.0 3 3.7 4.99 0.32 104.6 R.GRN 12.5 N 100 7.5 1 3.19 3.84 0.19 34.1 R.GRN 12.5 N 100 7.5 2 2.95 4.00 0.28 19.9 R.GRN 12.5 N 100 7.5 3 2.58 3.74 0.21 1.1 R.GRN 9.5 N 100 6.0 1 4.99 5.99 0.44 159.2 R.GRN 9.5 N 100 6.0 2 5.03 5.87 0.34 35.2 R.GRN 9.5 N 100 6.0 3 5.37 6.63 0.55 85.6 R.GRN 9.5 N 100 6.5 1 4.41 5.45 0.38 6.0 R.GRN 9.5 N 100 6.5 2 4.19 5.13 0.32 17.0 R.GRN 9.5 N 100 6.5 3 4.55 5.44 0.34 32.9 R.GRN 9.5 N 100 7.0 1 3.45 4.49 0.28 26.3 R.GRN 9.5 N 100 7.0 2 3.51 4.24 0.21 0.00 R.GRN 9.5 N 100 7.0 3 3.22 4.11 0.19 0.5 R.GRN 9.5 N 100 7.5 1 2.59 3.62 0.19 3.9 R.GRN 9.5 N 100 7.5 2 2.9 3.55 0.24 0.7 R.GRN 9.5 N 100 7.5 3 2.6 3.58 0.23 4.4 TRAP 19 N 100 6.0 1 6.15 9.41 0.77 2181.7 TRAP 19 N 100 6.0 2 6.26 10.86 0.86 6343.9 TRAP 19 N 100 6.0 3 5.77 9.28 0.86 4569.4 TRAP 19 N 100 6.5 1 5.61 7.94 0.66 20.0 TRAP 19 N 100 6.5 2 5.92 8.31 0.44 535.1 TRAP 19 N 100 6.5 3 5.38 7.88 0.39 432.2 TRAP 19 N 100 7.0 1 4.97 7.3 0.41 266.1 TRAP 19 N 100 7.0 2 5.19 7.89 0.51 1333.6 TRAP 19 N 100 7.0 3 5.47 7.2 0.34 6.8 TRAP 19 N 100 7.5 1 3.77 5.95 0.24 41.08 TRAP 19 N 100 7.5 2 4.73 6.57 0.22 179.84 296 Agg. Type NMAS, mm Grad. Gyrs AC, % ID Air Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s TRAP 19 N 100 7.5 3 4.1 6.35 0.39 1243.73 TRAP 12.5 N 100 6.0 1 7.02 9.68 0.81 836.2 TRAP 12.5 N 100 6.0 2 6.35 8.76 0.85 469.3 TRAP 12.5 N 100 6.0 3 7.13 9.43 0.77 983.8 TRAP 12.5 N 100 6.5 1 5.98 7.99 0.62 987.3 TRAP 12.5 N 100 6.5 2 6.1 7.7 0.34 107.4 TRAP 12.5 N 100 6.5 3 5.68 7.48 0.48 407.4 TRAP 12.5 N 100 7.0 1 5.55 6.57 0.26 130.9 TRAP 12.5 N 100 7.0 2 6.06 7.44 0.40 196.8 TRAP 12.5 N 100 7.0 3 5.46 6.67 0.27 75.2 TRAP 12.5 N 100 7.5 1 3.59 4.72 0.23 53.13 TRAP 12.5 N 100 7.5 2 3.92 5.65 0.33 0.28 TRAP 12.5 N 100 7.5 3 4.27 5.09 0.18 32.39 TRAP 9.5 N 100 6.0 1 6.46 7.74 0.57 128.0 TRAP 9.5 N 100 6.0 2 6.5 8.31 0.72 271.2 TRAP 9.5 N 100 6.0 3 6.34 7.69 0.52 133.5 TRAP 9.5 N 100 6.5 1 4.82 5.95 0.21 32.4 TRAP 9.5 N 100 6.5 2 4.65 5.63 0.20 9.8 TRA 1 P 9.5 N 100 6.5 3 4.45 5.4 0.19 16. TRAP 9.5 N 100 7.0 1 4.52 5.37 0.19 0.3 TRAP 9.5 N 100 7.0 2 4.18 5.07 0.22 3.6 TRAP 9.5 N 100 7.0 3 4.76 5.66 0.22 47.3 TRAP 9.5 N 100 7.5 1 2.97 0.16 2.7 2.41 TRAP 9.5 N 100 7.5 2 3.32 3.78 0.13 3.3 TRAP 9.5 N 100 7.5 3 3.17 3.83 0.16 1.3 R.GRN 19 F 100 5.5 1 5.25 7.18 0.63 133.2 R.GRN 19 F 100 5.5 2 5.31 6.69 0.69 226.3 R.GRN 19 F 100 5.5 3 5.70 6.95 0.52 165.1 R.GRN 19 F 100 6.0 1 4.48 5.89 0.45 3.9 R.GRN 19 F 100 6.0 2 4.09 5.42 0.53 0.0 R.GRN 19 F 100 6.0 3 4.55 5.74 0.44 0.0 R.GRN 19 F 100 6.5 1 3.36 4.43 0.16 0.0 R.GRN 19 F 100 6.5 2 3.87 4.81 0.24 0.2 R.GRN 19 F 100 6.5 3 3.09 4.23 0.21 0.0 R.GRN 12.5 F 100 5.5 1 5.46 6.35 0.49 131.0 R.GRN 12.5 F 100 5.5 2 4.96 6.25 0.54 80.2 R.GRN 12.5 F 100 5.5 3 5.09 6.10 0.54 61.2 R.GRN 12.5 F 100 6.0 1 3.84 4.72 0.25 1.7 R.GRN 12.5 F 100 6.0 2 3.85 4.74 0.30 0.89 R.GRN 12.5 F 100 6.0 3 3.84 4.54 0.35 7.6 R.GRN 12.5 F 100 6.5 1 3.20 3.95 0.19 0.0 R.GRN 12.5 F 100 6.5 2 3.24 3.96 0.17 0.6 R.GRN 12.5 F 100 6.5 3 3.52 4.24 0.22 0.0 R.GRN 9.5 F 100 5.5 1 6.59 7.79 0.84 187.3 R.GRN 9.5 F 100 5.5 2 6.81 7.68 0.71 132.7 297 Agg. Type NMAS, mm Grad. Gyrs AC, % ID Air Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s R.GRN 9.5 F 100 5.5 3 6.56 7.68 0.77 267.2 R.GRN 9.5 F 100 6.0 1 4.93 5.64 0.27 6.6 R.GRN 9.5 F 100 6.0 2 5.50 6.13 0.34 34.1 R.GRN 9.5 F 100 6.0 3 4.99 5.85 0.30 18.9 R.GRN 9.5 F 100 6.5 1 4.93 5.72 0.27 18.2 R.GRN 9.5 F 100 6.5 2 4.28 5.26 0.30 9.60 R.GRN 9.5 F 100 6.5 3 4.62 5.66 0.30 1.0 R.GRN 9.5 F 100 7.0 1 3.05 4.30 0.15 0.0 R.GRN 9.5 F 100 7.0 2 3.39 4.10 0.14 0.0 R.GRN 9.5 F 100 7.0 3 3.34 4.01 0.19 16.2 TRAP 19 F 100 6.0 1 5.21 6.93 0.31 38.5 TRAP 19 F 100 6.0 2 4.59 6.40 0.40 51.4 TRAP 19 F 100 6.0 3 5.12 6.55 0.46 47.0 TRAP 19 F 100 6.5 1 4.63 6.22 0.51 83.0 TRAP 19 F 100 6.5 2 4.34 5.77 0.46 9.1 TRAP 19 F 100 6.5 3 3.83 4.95 0.32 34.2 TRAP 19 F 100 7.0 1 3.35 4.69 0.33 44.5 TRAP 19 F 100 7.0 2 3.43 4.74 0.31 53.6 TR AP 19 F 100 7.0 3 3.40 4.13 0.19 1.0 TRAP 12.5 F 100 6.0 1 4.38 5.59 0.28 1.8 TRAP 12.5 F 100 6.0 2 4.04 5.16 0.35 3.1 TRAP 12.5 F 100 6.0 3 4.28 5.49 0.31 30.1 TRAP 12.5 F 100 6.5 1 3.18 4.10 0.23 0.00 TRAP 12.5 F 100 6.5 2 3.78 4.87 0.32 0.38 TRAP 12.5 F 100 6.5 3 3.34 4.77 0.30 0.00 TRAP 9.5 F 100 6.0 1 4.76 5.73 0.30 17.5 TRAP 9.5 F 100 6.0 2 4.63 5.60 0.35 35.6 TRAP 9.5 F 100 6.0 3 4.67 5.45 0.31 29.8 TRAP 9.5 F 100 6.5 1 4.06 5.02 0.28 0.59 TRAP 9.5 F 100 6.5 2 3.57 4.44 0.21 1.3 TRAP 9.5 F 100 6.5 3 3.94 4.79 0.22 1.4 C.GVL 19 N 65 6.0 1 5.29 6.97 0.54 159.8 C.GVL 19 N 65 6.0 2 4.48 6.35 0.36 60.7 C.GVL 19 N 65 6.0 3 5.29 7.04 0.62 212.4 C.GVL 19 N 65 6.5 1 4.53 5.91 0.36 222.5 C.GVL 19 N 65 6.5 2 4.47 6.20 0.42 1.45 C.GVL 19 N 65 6.5 3 4.01 5.22 0.42 142.7 C.GVL 19 N 65 7.0 4 3.52 4.79 0.41 0.00 C.GVL 19 N 65 7.0 5 3.42 4.73 0.22 0.38 C.GVL 19 N 65 7.0 6 2.74 4.13 0.23 0.00 C.GVL 12.5 N 65 6.0 1 5.59 7.00 0.58 83.7 C.GVL 12.5 N 65 6.0 2 6.03 7.19 0.59 350.5 C.GVL 12.5 N 65 6.0 3 5.78 6.97 0.72 175.3 C.GVL 12.5 N 65 6.5 1 4.87 6.18 0.56 45.0 C.GVL 12.5 N 65 6.5 2 4.82 5.74 0.48 118.4 298 Agg. Type NMAS, mm Grad. Gyrs AC, % ID ir Voids (SSD), % Air Voids (Corelok)% Water Absorp. in SSD, % CORR. K 20, 1x10 -5 cm/s A C.GVL 12.5 N 65 6.5 3 4.69 6.21 0.52 29.5 C.GVL 12.5 N 65 7.0 1 4.03 4.83 0.36 10.3 C.GVL 65 7.5 12.5 N 7.0 2 4.17 4.91 0.42 8 C.GVL 65 7.9 12.5 N 7.0 3 4.43 5.21 0.35 1 C.GVL 9.5 N 65 6.0 1 5.47 6.40 0.40 9.3 C.GVL 9.5 N 65 6.0 2 5.39 6.09 0.34 11.7 C.GVL 9.5 N 65 6.0 3 5.43 6.18 0.51 13.1 C.GVL 9.5 N 65 6.5 1 4.15 4.93 0.34 1.89 C.GVL 9.5 N 65 6.5 2 4.08 4.67 0.30 4.90 C.GVL 9.5 N 65 6.5 3 4.15 4.85 0.29 1.34 C.GVL 9.5 N 65 7.0 1 3.58 3.98 0.23 2.22 C.GVL 9.5 N 65 7.0 2 3.02 3.60 0.22 1.24 C.GVL 9.5 N 65 7.0 3 2.91 3.88 0.18 0.00 L.GRN 19 N 65 5.5 1 4.34 5.05 0.18 0.00 L.GRN 19 N 65 5.5 2 4.80 6.25 0.66 54.2 L.GRN 19 N 65 5.5 3 4.79 6.95 0.76 58.3 L.GRN 19 N 65 6.0 1 4.02 4.96 0.44 150.6 L.GRN 19 N 65 6.0 2 4.10 5.26 0.48 18.7 L.GRN 19 N 65 6.0 3 3.49 4.16 0.32 7.2 L.GRN 19 N 65 6.5 1 2.80 3.82 0.24 8.3 L.GRN 19 N 65 6.5 2 3.17 4.03 0.22 0.00 L.GRN 19 N 65 6.5 3 3.12 4.19 0.27 10.8 L.GRN 12.5 N 65 5.5 1 7.95 10.48 1.66 1741.7 L.GRN 12.5 N 65 5.5 2 6.48 8.58 0.95 286.7 L.GRN 12.5 N 65 5.5 3 5.99 7.48 0.86 394.0 L.GRN 12.5 N 65 6.0 1 5.17 6.01 0.55 9.8 L.GRN 12.5 N 65 6.0 2 4.68 5.96 0.50 99.8 L.GRN 12.5 N 65 6.0 3 4.99 6.06 0.54 2.6 L.GRN 12.5 N 65 6.5 1 4.22 4.94 0.36 10.4 L.GRN 12.5 N 65 6.5 2 3.87 4.48 0.35 2.3 L.GRN 12.5 N 65 6.5 3 4.01 4.94 0.41 1.6 L.GRN 9.5 N 65 6.0 1 5.60 6.41 0.43 37.2 L.GRN 9.5 N 65 6.0 2 5.56 6.52 0.50 87.7 L.GRN 9.5 N 65 6.0 3 5.25 5.95 0.51 33.1 L.GRN 9.5 N 65 6.5 1 4.34 4.97 0.36 4.2 L.GRN 9.5 N 65 6.5 2 3.94 4.43 0.35 0.09 L.GRN 9.5 N 65 6.5 3 4.26 4.94 0.38 4.6 L.GRN 9.5 N 65 7.0 1 2.66 3.16 0.18 0.19 L.GRN 9.5 N 65 7.0 2 2.77 3.35 0.17 0.00 L.GRN 9.5 N 65 7.0 3 2.84 3.32 0.25 0.09 LMS 19 N 65 5.5 1 4.81 6.10 0.38 132.7 LMS 19 N 65 5.5 2 5.71 6.90 0.54 159.6 LMS 19 N 65 5.5 3 5.08 6.16 0.51 134.7 LMS 19 N 65 6.0 1 3.99 4.84 0.30 0.00 LMS 19 N 65 6.0 2 4.31 5.39 0.36 26.5 299 300 Agg. e SSD, % cm/s Typ NM m AS m , Grad. Gyrs AC, % ID Ai (S r Vo SD ids ), % ( Air Cor Vo elo ids k)% Wate orp. in r Abs CORR. K 20, 1x10 -5 LMS 19 N 65 6.0 3 3.72 4.83 0.31 27.3 LMS 19 N 65 6.5 1 2.88 3.64 0.21 4.06 LMS 119 N 65 6.5 2 3.10 3.89 0.21 1 .9 LMS .19 N 65 6.5 3 3.05 3.95 0.17 0 00 LMS 12.5 N 65 6.0 1 4.64 5.32 0.35 26.1 LMS 312.5 N 65 6.0 2 4.75 5.62 0.31 2 .2 LMS 12.5 N 65 6.0 3 5.10 5.92 0.36 0.87 LMS 12.5 N 65 6.5 1 4.26 4.67 0.22 0.29 LMS 12.5 N 65 6.5 2 3.66 4.19 0.24 0.00 LMS 12.5 N 65 6.5 3 3.73 4.53 0.26 0.19 LMS 12.5 N 65 7.0 1 2.79 3.81 0.23 0.00 LMS 12.5 N 65 7.0 2 2.95 3.86 0.19 0.00 LMS 12.5 N 65 7.0 3 2.58 3.44 0.22 0.00 LMS 9.5 N 65 6.0 1 4.29 5.23 0.19 3.8 LMS 9.5 N 65 6.0 2 4.28 5.34 0.18 6.7 LMS 9.5 N 65 6.0 3 4.37 5.34 0.25 1.4 LMS 9.5 N 65 6.5 1 3.33 4.13 0.14 0.00 LMS 9.5 N 65 6.5 2 3.26 4.08 0.16 0.00 LMS 9.5 N 65 6.5 3 3.19 3.77 0.13 0.09 LMS 9.5 N 65 7.0 1 2.23 2.68 0.10 0.00 LMS 9.5 N 65 7.0 2 2.09 2.54 0.08 0.00 LMS 9.5 65 7.0 3 2.30 2.74 0.10 0.00 N R.GRN 19 N 65 6.5 1 5.66 7.93 0.63 22.9 R.GRN 19 N 65 6.5 2 4.65 7.27 0.70 251.1 R N 19 N 65 6.5 3 5.37 7.87 0.73 113.4 .GR R N .GR 19 N 65 7.0 1 4.93 7.58 0.63 417.6 R.GRN 19 N 65 7.0 2 4.13 5.89 0.49 8.0 R.GRN 19 N 65 7.0 3 3.65 5.07 0.39 0.55 R.GRN 19 N 65 7.5 1 4.03 6.39 0.41 29.0 R.GRN 19 N 65 7.5 2 2.72 3.88 0.25 8.2 R.GRN 19 N 65 7.5 3 4.17 6.52 0.43 192.8 R.GRN 12.5 N 65 6.5 1 7.08 9.39 0.73 391.4 R.GRN 12.5 N 65 6.5 2 6.39 8.78 0.75 981.6 R.GRN 12.5 N 65 6.5 3 7.17 9.14 0.70 1174.2 R.GRN 12.5 N 65 7.0 1 5.98 7.95 0.69 165.0 R.GRN 12.5 N 65 7.0 2 5.71 8.06 0.68 132.8 R.GRN 12.5 N 65 7.0 3 5.40 7.10 0.49 5.7 R.GRN 12.5 N 65 7.5 1 5.25 7.71 0.45 22.6 R.GRN 12.5 N 65 7.5 2 4.99 6.37 0.41 216.4 R.GRN 12.5 N 65 7.5 3 4.74 6.06 0.39 51.7 R.G RN 9.5 N 65 6.5 1 6.56 8.33 0.75 134.5 R.G 4RN 9.5 N 65 6.5 2 6.66 8.16 0.80 3 .0 R.GRN 9.5 N 65 6.5 3 6.63 8.48 0.82 232.3 R.GRN 9.5 N 65 7.0 1 5.61 6.67 0.62 102.6 R.GRN 9.5 N 65 7.0 2 6.00 6.84 0.64 46.2 301 yp ids , % SSD, % . -5 cm/s Agg. T e NM m AS m , Grad. Gyrs A % C, ID Ai (S r Vo SD) (C Air or Voi elok ds )% Wate orp. in r Abs CORR K 20, 1x10 R.GRN 9.5 N 65 7.0 3 5.21 6.11 0.59 35.4 R.GRN 9.5 N 65 7.5 1 4.41 5.06 0.37 7.0 R.GRN 9.5 N 65 7.5 2 4.22 5.31 0.23 9.7 R.GRN 9.5 N 65 7.5 3 4.16 5.20 0.46 116.4 R.GRN 19 F 65 6.0 1 5.83 6.86 0.44 48.0 R.GR N 19 F 65 6.0 2 5.77 6.85 0.61 304.0 R.GRN 19 F 65 6.0 3 5.27 6.72 0.59 397.2 R.GRN 19 F 65 6.5 1 4.48 6.06 0.40 17.1 R.GRN 19 F 65 6.5 2 4.03 5.40 0.40 21.4 R.GRN 19 F 65 6.5 3 3.98 5.07 0.34 3.10 R.GRN 19 F 65 7.0 1 3.62 4.78 0.26 0.00 R.GR .3N 19 F 65 7.0 2 3 6 4.19 0.17 11.9 R.GRN 19 F 5 7.0 3 3.0 0.19 5.40 6 9 4.00 R.GRN 12.5 0.67 F 65 6.0 1 6.46 7.71 97.5 R 0.GRN 12.5 F 65 6.0 2 5.54 6.82 .57 104.1 R.GRN 12.5 F 65 6.0 3 6.02 7.33 0.56 315.5 R.GRN 12.5 F 65 6.5 1 4.24 5.46 0.35 55.5 R.GRN 12.5 F 65 6.5 2 4.27 5.13 0.27 10.9 R.GRN 12.5 F 65 6.5 3 4.66 5.56 0.36 39.7 R.GRN 12.5 F 6 5 7.0 1 3.36 4.07 0.14 16.6 R.GRN 12.5 F 65 7.0 2 3.61 4.62 0.24 0.00 R.GRN 12.5 F 6 7.0 2.7 .66 17 0.00 5 3 0 3 0. R.GR 6N 9.5 F 5 6.5 1 6.39 8.03 0.94 330.6 R 6.GRN 9.5 F 5 6.5 2 5.83 6.97 0.82 18.4 R.GRN 9.5 F 65 6.5 3 5.44 6.57 0.69 53.2 R.GRN 9.5 F 65 7.0 1 4.17 4.94 0.35 11.2 R.GRN 9.5 F 65 7.0 2 4.52 5.22 0.44 7.90 R.GRN 9.5 F 65 7.0 3 4.52 5.06 0.37 82.1 R.GRN 9.5 F 65 7.5 1 3.08 4.10 0.18 0.00 R.GRN 9.5 F 65 7.5 2 3.43 4.01 0.23 0.19 R.GRN 9.5 F 65 7.5 3 3.01 3.63 0.14 0.00 TRAP 19 F 65 6.5 1 5.02 7.26 0.55 362.4 TRAP 19 F 65 6.5 2 5.01 6.65 0.54 447.7 TRAP 19 F 65 6.5 3 4.94 6.66 0.52 446.5 TRAP 19 F 65 7.0 1 3.72 4.88 0.41 56.4 TRAP 19 F 65 7.0 2 4.34 5.60 0.47 2.51 TRAP 19 F 65 7.0 3 3.62 4.62 0.38 8.52 TRAP 19 F 65 7.5 1 2.95 4.07 0.22 34.1 TRAP 19 F 65 7.5 2 3.05 3.91 0.25 0.00 TRAP 19 F 65 7.5 3 2.78 3.71 0.20 0.00 TRAP 12.5 F 65 6.0 1 5.36 7.44 0.60 347.0 TRAP 12.5 F 65 6.0 2 5.72 7.43 0.51 541.3 TRA 2.5 3 5.10 6.51 0.37 396.9 P 1 F 65 6.0 TRAP 12.5 F 65 6.5 1 3.88 5.34 0.35 242.3 TRAP 12.5 F 65 6.5 2 4.00 5.69 0.30 25.4 302 yp s SSD, % . -5 cm/s Agg. T e NM m AS m , Grad. Gyrs A % C, ID Ai (S r Vo SD) ids , % (C Air or Void elok)% Wate orp. in r Abs CORR K 20, 1x10 TRAP 12.5 F 65 6.5 3 4.51 5.77 0.38 93.5 TRAP 12.5 F 65 7.0 1 3.43 4.43 0.20 0.00 TRAP 12.5 F 65 7.0 2 2.91 3.59 0.14 0.00 TRAP 12.5 F 65 7.0 3 2.64 3.83 0.15 0.00 TRAP 9.5 F 65 6.5 1 5.44 6.92 0.56 217.0 TRA P 9.5 F 65 6.5 2 5.75 7.00 0.49 168.9 TRAP 9.5 F 65 6.5 3 5.52 6.55 0.49 155.3 TRAP 9.5 F 65 7.0 1 3.87 5.01 0.35 68.2 TRAP 9.5 F 65 7.0 2 4.18 4.95 0.32 0.00 TRAP 9.5 F 65 7.0 3 4.17 5.03 0.37 11.2 TRAP 9.5 F 65 7.5 1 2.96 3.43 0.12 0.00 TRAP 9.5 F 65 7.5 2 3.19 3.81 0.22 0.39 TRAP .5 F 65 7.5 3 3.13 3.74 0.19 3.20 9 303 APPENDIX B COREGRAVITY TM PROGRAM CORRECTION FACTOR TABLE B1 304 CoreGravity TM Program Correction Factors (110) Correction Factor, CF Small Bag Large Bag Double Bags R= Ratio M c /M b CF=-0.000566*R+0.8121 CF=-0.00166*R+0.8596 CF=-0.0022448*R+0.81518 10 0.806 0.843 0.793 20 0.801 0.826 0.770 30 0.795 0.810 0.748 40 0.789 0.793 0.725 50 0.784 0.777 0.703 60 0.778 0.760 0.680 70 0.772 0.6580.743 80 0.767 0.727 0.636 90 0.761 0.710 0.613 100 0.756 0.694 0.591 110 0.750 0.677 0.568 120 0.744 0.660 0.546 130 0.739 0.644 0.523 140 0.733 0.627 0.501 150 0.727 0.611 0.478 160 0.722 0.594 0.456 170 0.716 0.577 0.434 180 0.710 0.561 0.411 190 0.705 0.544 0.389 200 0.699 0.528 0.366 210 0.693 0.511 0.344 220 0.688 0.49 0.4 321 230 0.682 0.478 0.299 240 0.676 0.461 0.276 250 0.671 0.445 0.254 260 0.665 320.428 0.2 270 0.659 0.411 0.209 280 0.654 0.395 0.187 290 0.648 0.378 0.164 300 0.642 0.362 0.142 Note: M c is mass of dry sample, M b is mass of bag 305 APPENDI K W X C AG COMPACTION EFF GREGATE BREA DO RT N FOR DIFFERENT O S 306 C.GVL 19 mm NMAS 10 20 40 50 80 90 0.45 Power Sieve Size, mm e n t Pa s s in g , % 0 30 60 70 100 Pe r c Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 9.5 19 25 F IGURE C1 Aggregate breakdown for crushed gravel 19 mm NMAS mixture. C.GVL 12.5 mm NMAS 30 90 P e rc 0 10 20 40 50 60 70 80 100 0.45 Power Sieve Size, mm en t P assi n g , % Orig Loose 100 Gyrs 65 Gyrs Mar s hall 0.075 1.18 2.36 4.75 9.5 19 UFIG RE C2 Aggregate breakdown for crushed gravel 12.5 mm NMAS mixture. 307 C.GVL 9.5 mm NMAS 10 40 50 80 90 1 0.45 Power Sieve Size, mm P en t P a ssi n g , % 0 20 30 60 70 00 e rc Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4 9.5 12.5 F o .75 GRN 19 mm NMAS 0 20 30 ieve Size, mm r c s IGU Pe e n t Pa s i n g , % RE C3 Aggregate bre 10 40 50 60 70 80 90 001 0.45 Power S akd wn for crushed gravel 9.5 mm NMAS mixture. Orig Loose 100 Gyrs 65 Gyrs Marshall 0. . 3 9.5 19 FIGURE 075 1 18 2. 6 4.75 25 C4 Aggregate breakdown fo r lab granite 19 mm NMAS mixture. GRN 12.5 mm NMAS 30 60 90 100 e S , m s i n g 0 308 10 20 4 5 0 0 7 80 0 0.45 Power Siev ize m P e r c e n t P a s , % Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 9.5 FIGURE C5 Aggregate breakdown for lab granite 12.5 mm NMAS mixture. 19 GRN 9.5 mm NMAS 0 20 30 50 60 70 80 P e ass , % 10 40 90 100 0.45 Power Sieve Size, mm r cen t P i n g Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 9.5 12.5 FIGURE C6 Aggregate breakdown for lab granite 9.5 mm NMAS mixture. LMS 12.5 mm NMAS 0 20 30 50 60 70 80 P e r c P a ss i n , % 309 LMS 19 mm NMAS 20 30 60 90 100 Pe rc e s s i n g , 0 10 40 50 70 80 0.45 Pow er Sieve Size, mm n t Pa % Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9.5 FIGURE C7 Aggregate breakdown for limestone 19 19 25 mm NMAS mixture. 10 40 90 100 0.45 Pow er Sieve Size, m m en t g Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9.5 19 FIGURE C8 Aggregate breakdown for limestone 12.5 mm NMAS mixture. 310 LMS 9.5 mm NMAS 20 30 60 90 100 Pe r c s i n g , 0 10 40 50 70 80 0.45 Pow er Sieve Size, m m e s n t Pa % Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 FIGURE C9 Aggregate breakdown fo 9.5 12.5 r limestone 9.5 mm NMAS mixture. Ruby 19 mm NMAS 0 20 30 50 70 80 Pe rc Pa s s i , % 10 40 60 90 100 0.45 Power Sieve Size, mm e n t n g Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 FIGURE C10 Aggregate breakdown for ruby granite 19 mm NMAS mixture. 2.36 4.75 9.5 19 25 311 Ruby 12.5 mm NMAS 20 30 60 90 100 Pe r c s s i n 0 10 40 50 70 80 0.45 Power Sieve Size, mm e n t Pa g , % Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9. FIGURE C11 Aggregate breakdown for 5 19 ruby granite 12.5 mm NMAS mixture. Ruby 9.5 mm NMAS 0 20 30 50 60 70 80 P e r c e n t P ass i n g , % 10 40 90 100 0.45 Power Sieve Size, mm Orig Loose 100 Gyrs 65 Gyrs Mars hall 0.075 1.18 2.36 4.75 12.5 FIGURE C12 Aggregate breakdown for ruby granite 9.5 mm NMAS mixture. 9.5 312 Trap 19 mm NMAS 0 10 20 30 40 50 60 70 80 90 100 0.45 Power Sieve Size, mm P e r c e n t P a s s i ng, % Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9.5 19 25 FIGURE C13 Aggregate breakdown for traprock 19 mm NMAS mixture. Trap 12.5 mm NMAS 0 10 20 30 40 50 60 70 80 90 100 0.45 Power Sieve Size, mm P e rc en t P a ssi n g , % Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9.5 19 FIGURE C14 Aggregate breakdown for traprock 12.5 mm NMAS mixture. 313 Trap 9.5 mm NMAS 0 10 20 30 40 50 60 70 80 90 100 0.45 Power Sieve Size, mm P e rcen t P assi n g , % Orig Loose 100 Gyrs 65 Gyrs Marshall 0.075 1.18 2.36 4.75 9.5 12.5 FIGURE C15 Aggregate breakdown for traprock 9.5 mm NMAS mixture. 314 APPENDIX D TRIAXIAL PERFORMANCE TEST RESULTS TABLE D1 Dynamic Modulus Test Results 25Hz 10Hz 5Hz 1Hz 0.5Hz 0.1Hz Agg. Type Gyrs NMAS ID E* ? E* ? E* ? E* ? E* ? E* ? C.GVL 65 19 7 693.8 34.4 600.5 30.8 518.6 28.9 403.9 23.4 365.2 20.4 317.6 18.1 C.GVL 65 19 8 774.7 39.0 681.8 24.2 630.4 19.9 574.7 12.9 564.5 10.0 544.3 7.1 C.GVL 65 19 9 697.9 41.8 565.2 33.8 507.0 29.5 424.5 21.4 409.7 19.8 404.4 17.2 C.GVL 65 12.5 1 1297.7 35.1 1128.6 27.4 1049.7 21.0 931.6 15.3 885.9 13.2 848.4 10.5 C.GVL 65 12.5 8 825.1 36.1 750.2 28.2 682.7 20.9 598.4 14.2 572.9 12.2 550.7 8.7 C.GVL 65 12.5 10 347.7 49.8 274.5 37.3 248.3 32.9 194.2 26.8 176.8 24.9 160.2 19.2 C.GVL 65 9.5 6 1069.6 40.5 921.4 31.8 804.4 25.3 653.4 19.0 624.3 17.1 593.3 13.6 C.GVL 65 9.5 8 991.2 44.6 804.4 30.9 694.9 27.6 573.5 20.6 543.1 19.1 526.5 15.9 C.GVL 65 9.5 9 872.4 40.7 748.4 32.2 673.0 27.5 560.0 20.8 539.0 19.2 523.9 16.6 C.GVL 100 19 7 891.6 36.8 790.2 26.9 726.1 22.5 636.9 16.8 625.4 15.0 612.0 11.7 C.GVL 100 19 9 870.4 39.2 734.3 28.4 658.6 23.6 553.7 17.0 528.1 15.8 498.3 12.3 C.GVL 100 19 10 826.7 30.7 795.4 25.2 718.6 21.0 617.3 15.8 592.3 13.7 572.8 11.3 C.GVL 100 12.5 2 1015.0 41.8 827.4 29.1 714.0 24.9 580.8 18.1 552.2 16.5 507.6 13.7 C.GVL 100 12.5 3 774.7 41.4 634.2 29.0 558.3 24.3 449.7 17.4 426.8 15.0 391.8 11.0 C.GVL 100 12.5 7 1217.2 39.6 1041.0 30.3 917.8 25.0 766.6 18.9 742.7 16.9 714.5 13.2 C.GVL 100 9.5 5 1330.8 38.0 1056.3 29.1 926.2 23.7 765.9 17.2 723.4 15.3 668.3 12.3 C.GVL 100 9.5 6 1273.1 34.8 1107.7 24.6 997.1 20.8 846.7 15.5 812.8 14.1 772.3 11.2 C.GVL 100 9.5 7 940.5 42.2 785.8 29.2 691.7 24.9 566.9 18.1 534.2 16.0 501.6 12.4 L.GRN 65 19 R2 1712.5 34.4 1441.9 25.1 1286.9 20.5 1131.2 15.4 1054.5 14.9 975.2 12.4 L.GRN 65 19 R3 1085.4 38.2 908.0 28.9 769.0 26.3 634.3 20.2 590.6 19.0 533.3 16.1 L.GRN 65 19 W 1114.2 38.8 1003.7 28.6 911.2 24.4 752.2 19.2 702.4 18.8 675.6 16.1 L.GRN 65 12.5 R2 1544.6 36.5 1310.9 28.2 1165.5 23.5 1005.7 17.8 963.1 16.0 917.1 14.5 L.GRN 65 12.5 R3 1215.5 39.3 1045.3 32.9 962.3 27.1 796.9 21.9 744.4 21.2 723.7 17.9 L.GRN 65 12.5 R4 818.8 40.2 718.3 31.6 643.5 25.6 545.8 19.6 526.5 17.4 498.7 17.2 L.GRN 65 9.5 R1 1145.5 37.5 916.2 30.3 758.2 27.5 548.6 23.6 498.0 21.6 430.6 18.6 L.GRN 65 9.5 R3 609.1 41.2 484.7 35.8 409.3 29.8 295.0 24.1 251.4 23.3 205.0 19.8 315 25Hz 10Hz 5Hz 1Hz 0.5Hz 0.1Hz Agg. Type Gyrs NMAS ID E* ? E* ? E* ? E* ? E* ? E* ? L.GRN 65 9.5 7 1096.0 39.1 883.6 31.4 795.8 28.4 656.8 22.3 606.0 21.2 568.9 19.8 L.GRN 100 19 6 1197.6 42.4 979.2 30.5 865.8 26.4 739.7 20.7 699.1 20.0 704.5 17.2 L.GRN 100 19 7 818.6 36.9 712.7 27.2 637.0 23.7 511.1 18.6 463.7 18.3 401.2 16.7 L.GRN 100 19 10 1180.9 35.5 990.0 26.6 890.1 22.2 757.6 16.4 714.6 15.3 697.8 12.9 L.GRN 100 12.5 8 1336.1 39.7 1116.2 29.1 1002.9 25.4 818.9 19.7 778.0 18.1 734.8 16.4 L.GRN 100 12.5 9 1505.6 36.3 1261.2 26.7 1130.6 22.5 965.9 16.9 925.4 15.7 873.9 13.4 L.GRN 100 12.5 10 1371.4 30.5 1366.9 26.4 1275.6 21.5 1103.2 16.8 1087.1 16.2 1072.4 12.7 L.GRN 100 9.5 9 1596.8 38.9 1325.3 29.2 1165.4 25.1 926.2 20.2 877.8 18.9 824.3 15.4 L.GRN 100 9.5 11 1408.8 41.5 1148.6 34.0 965.2 30.7 760.6 23.5 703.3 22.8 622.0 20.9 L.GRN 100 9.5 12 1037.9 40.7 886.3 29.5 788.1 24.3 641.7 18.7 602.0 17.5 552.3 14.5 LMS 65 19 6 1420.5 38.1 1150.5 29.1 1038.3 26.4 896.4 20.9 862.5 19.8 855.4 16.3 LMS 65 19 7 1366.5 37.6 1119.6 28.1 953.7 24.4 727.7 19.9 625.5 19.2 554.7 16.3 LMS 65 19 9 1212.8 40.2 1068.1 29.2 953.4 26.2 836.4 20.3 769.4 19.2 741.7 17.7 LMS 65 12.5 3 997.3 39.0 847.0 31.5 713.5 29.2 613.4 24.5 576.8 22.5 486.6 20.5 LMS 65 12.5 6 1807.8 37.6 1590.9 22.6 1465.9 19.3 1259.8 15.0 1224.5 14.5 1186.7 11.4 LMS 65 12.5 8 744.1 41.4 590.3 34.8 500.9 31.2 382.7 26.7 344.8 25.9 299.2 22.0 LMS 65 9.5 4 1720.7 33.4 1422.2 26.1 1285.1 23.1 1129.9 17.4 1089.8 16.5 1073.7 14.1 LMS 65 9.5 6 1016.4 37.6 895.1 29.2 807.7 25.4 665.9 19.5 636.6 17.4 611.3 15.0 LMS 65 9.5 7 1599.1 40.3 1325.8 31.5 1193.2 25.8 987.7 19.7 938.4 18.3 915.8 14.5 LMS 100 19 10 1055.7 46.8 829.5 34.6 703.2 30.2 551.1 22.0 528.3 20.2 503.1 17.0 LMS 100 19 11 1248.8 39.0 1150.5 25.9 1089.7 22.4 977.8 17.5 969.4 16.0 957.1 13.4 LMS 100 19 13 1098.8 36.8 926.0 27.3 838.8 23.2 729.2 16.0 696.7 15.4 663.0 12.9 LMS 100 12.5 5 1640.4 38.3 1308.8 31.8 1116.6 29.0 805.1 25.4 704.8 24.7 583.0 22.7 LMS 100 12.5 6 1571.7 33.4 1361.5 29.8 1226.7 26.2 1011.0 20.8 958.1 19.8 902.9 17.1 LMS 100 12.5 7 1132.7 38.6 949.5 29.4 839.9 26.0 675.4 19.9 618.4 18.1 563.0 15.1 LMS 100 9.5 3 1728.6 35.1 1446.4 26.6 1312.6 22.7 1074.8 17.3 988.2 16.9 889.1 12.7 316 25Hz 10Hz 5Hz 1Hz 0.5Hz 0.1Hz Agg. Type Gyrs NMAS ID E* ? E* ? E* ? E* ? E* ? E* ? LMS 100 9.5 5 577.3 42.7 457.4 32.2 384.8 31.5 294.5 24.9 252.1 23.9 212.6 20.3 LMS 100 9.5 8 1626.6 37.7 1349.7 27.5 1204.9 24.2 1002.8 19.2 924.4 17.2 824.6 15.6 R.GRN 65 19 7 1260.9 41.0 1050.3 30.8 933.8 25.6 769.2 20.5 719.2 19.2 636.6 16.1 R.GRN 65 19 8 799.6 39.9 670.1 28.5 596.1 20.9 503.7 17.3 477.3 15.9 452.2 13.5 R.GRN 65 19 10 961.4 42.1 837.9 31.5 766.2 27.1 630.4 21.0 596.8 19.5 587.1 16.1 R.GRN 65 12.5 7 844.9 41.8 663.8 33.5 551.8 32.0 397.8 27.0 358.9 25.7 303.4 23.0 R.GRN 65 12.5 9 983.8 40.7 877.1 29.5 802.4 24.8 675.8 18.8 662.3 17.1 635.0 16.2 R.GRN 65 12.5 10 985.0 44.1 814.2 33.2 735.9 29.1 601.3 25.0 550.3 24.5 518.4 20.7 R.GRN 65 9.5 6 1037.9 40.2 840.0 29.2 741.5 27.9 585.4 23.1 515.8 21.0 446.3 19.3 R.GRN 65 9.5 7 1520.9 40.4 1242.5 30.7 1097.8 27.4 873.2 21.7 832.1 19.2 792.0 15.9 R.GRN 65 9.5 9 1154.2 31.9 979.4 25.4 885.3 20.6 775.7 14.9 732.4 13.2 678.4 11.9 R.GRN 100 19 3 1057.7 39.4 837.0 30.2 715.4 24.7 597.1 18.5 565.9 17.2 540.9 15.9 R.GRN 100 19 6 886.5 42.6 687.7 33.3 579.3 30.4 449.6 23.2 410.1 21.9 368.7 19.1 R.GRN 100 19 7 1327.3 42.8 1072.9 32.4 967.7 29.0 790.8 23.5 735.2 20.8 672.5 17.9 R.GRN 100 12.5 4 1224.0 38.8 975.8 32.0 835.3 25.8 647.2 22.8 601.2 20.5 542.4 18.9 R.GRN 100 12.5 10 1167.1 44.3 961.6 35.1 831.2 31.1 645.6 24.1 599.7 22.7 545.5 17.3 R.GRN 100 12.5 11 990.7 39.4 793.6 28.1 697.1 25.2 577.0 18.6 549.4 17.3 501.0 15.5 R.GRN 100 9.5 6 786.4 42.9 630.4 30.9 548.8 25.8 459.7 19.0 423.0 17.8 389.9 15.9 R.GRN 100 9.5 7 1329.2 44.6 1195.8 34.6 1078.4 31.7 923.6 26.8 865.3 25.1 857.6 22.1 R.GRN 100 9.5 9 1058.3 43.0 852.9 33.1 780.7 29.6 622.8 24.0 589.4 22.5 542.0 20.2 TRAP 65 19 8 538.9 41.9 473.8 29.0 415.9 25.5 322.3 22.4 285.1 21.8 242.9 18.2 TRAP 65 19 9 626.0 38.2 602.4 26.4 540.9 22.2 440.6 17.2 402.7 16.9 332.1 15.5 TRAP 65 19 10 655.7 47.3 644.3 34.1 584.4 30.6 479.7 25.0 434.6 25.3 390.0 20.9 TRAP 65 12.5 6 953.7 36.0 835.6 27.2 749.5 22.4 636.5 16.3 595.9 14.7 577.2 13.0 TRAP 65 12.5 9 1364.3 38.6 1216.5 29.2 1098.9 22.8 956.3 18.7 943.3 17.2 953.7 13.1 TRAP 65 12.5 10 1013.9 34.8 878.7 23.7 797.0 19.8 697.1 14.8 663.8 13.4 618.5 11.5 317 25Hz 10Hz 5Hz 1Hz 0.5Hz 0.1Hz Agg. Type Gyrs NMAS ID E* ? E* ? E* ? E* ? E* ? E* ? TRAP 65 9.5 7 952.4 37.2 829.8 27.3 745.4 23.3 621.9 18.4 579.0 17.5 526.8 14.2 TRAP 65 9.5 8 942.5 38.5 786.2 28.9 706.1 24.9 588.7 18.8 547.6 17.3 507.2 14.9 TRAP 65 9.5 9 438.5 44.6 354.0 34.0 299.4 30.9 227.5 24.9 202.4 23.3 184.1 21.8 TRAP 100 19 3 766.1 36.7 637.2 27.8 564.4 22.9 437.7 17.4 387.2 16.9 331.7 13.9 TRAP 100 19 7 1070.0 40.3 885.6 28.3 788.7 23.5 674.0 17.6 632.7 16.9 591.4 14.8 TRAP 100 19 8 1048.9 34.9 939.8 26.4 851.4 22.4 749.1 16.6 713.8 15.6 694.3 14.4 TRAP 100 12.5 1 974.8 31.0 879.5 26.3 796.0 23.0 656.7 18.0 648.5 15.6 615.3 13.6 TRAP 100 12.5 6 1100.3 35.6 943.3 25.7 846.0 20.9 718.2 16.1 688.3 14.8 635.4 12.3 TRAP 100 12.5 7 1023.0 30.3 878.0 25.5 786.6 22.5 647.0 17.1 609.9 16.2 556.2 13.6 TRAP 100 9.5 6 942.5 36.5 836.0 28.4 752.0 24.1 649.9 18.7 619.7 18.0 602.8 15.3 TRAP 100 9.5 7 1717.1 33.3 1517.0 23.6 1376.7 19.2 1204.6 14.0 1182.3 12.5 1135.1 10.4 TRAP 100 9.5 8 1100.8 38.3 943.1 29.2 846.9 24.7 732.5 19.2 686.5 18.0 671.5 14.1 L.GRN 40 12.5 6 1183.6 36.0 1054.9 26.1 960.4 22.0 845.1 16.5 816.6 15.0 803.2 13.4 L.GRN 40 12.5 7 1183.9 35.4 1076.2 25.8 991.6 22.1 871.0 16.5 839.0 13.6 797.2 12.9 L.GRN 40 12.5 9 1447.9 35.1 1248.4 25.7 1116.2 22.5 949.3 16.7 872.1 15.0 795.7 13.3 R.GRN 40 12.5 6 1290.2 34.3 1101.0 25.5 1016.9 20.7 885.0 15.2 864.2 13.6 798.1 12.8 R.GRN 40 12.5 8 1321.4 31.2 1178.3 21.0 1114.3 17.1 981.6 12.1 951.7 10.8 905.1 9.3 R.GRN 40 12.5 10 1521.5 32.2 1317.1 21.1 1219.0 16.9 1099.9 11.9 1063.6 10.9 1016.8 9.4 Note: The shady rows were determined as outliers and were excluded from the statistical analysis. 318 TABLE D2 Static Creep Test Results Secondary Time at strain level, sec Agg.Type Gyrs NMAS ID Slope Intercept Max time, sec Flow time, sec 1% 2% 3% 4% Log (TB 0.04 B) C.GVL 65 19 7 0.17042 12741.2 598 2.4 12.3 134.5 1133.0 3.05 C.GVL 65 19 8 0.07611 15150.0 17974 2.2 18.6 5153.8 608000.0 5.78 C.GVL 65 19 9 0.16100 11749.0 1195 2.7 18.4 159.6 3682.0 3.57 C.GVL 65 12.5 1 0.15473 6449.5 14500 15.5 1864.7 6574.8 21840.0 4.34 C.GVL 65 12.5 8 0.15924 10957.8 2013 2.9 25.4 319.3 4404.0 3.64 C.GVL 65 12.5 10 0.82184 7869.5 6.5 1.6 2.7 4.9 7.6 0.88 C.GVL 65 9.5 6 0.16627 6533.8 18734 11.7 819.2 11214.3 23813.0 4.38 C.GVL 65 9.5 8 0.14623 7438.9 22500 8.2 848.9 9694.0 31400.0 4.50 C.GVL 65 9.5 9 0.11817 8550.0 20404 5.9 1493.4 90400.0 206000.0 5.31 C.GVL 100 19 8 0.10951 6284.7 61565 42.6 269000.0 801000.0 1333000.0 6.12 C.GVL 100 19 9 0.17656 9442.3 1797 3.3 35.5 624.2 4939.0 3.69 C.GVL 100 19 10 0.10817 7917.3 22555 5500 9.2 9404.7 20434.9 24229.0 4.38 C.GVL 100 12.5 2 0.12498 7699.6 18875 8.8 2564.3 13874.6 25326.0 4.40 C.GVL 100 12.5 3 0.14161 7220.5 20105 7150 9.9 1544.6 14144.9 22955.0 4.36 C.GVL 100 12.5 7 0.15898 7215.7 13174 8.5 544.2 4104.1 14533.0 4.16 C.GVL 100 9.5 5 0.11625 4872.1 101006 359.4 6314000.0 15060000.0 23810000.0 7.38 C.GVL 100 9.5 6 0.08520 4777.8 53436 352000.0 4726000.0 9100000.0 13480000.0 7.13 C.GVL 100 9.5 7 0.13287 6611.9 17604 16.6 4635.1 12755.2 21398.0 4.33 L.GRN 65 19 R2 0.18522 9588.7 1114 3.4 34.8 443.9 1126.0 3.05 L.GRN 65 19 R3 0.11389 11473.7 789 2.9 78.7 4169.0 9793.0 3.99 L.GRN 65 19 W 0.16686 8654.4 13209 4.8 78.4 2924.4 13284.6 4.12 L.GRN 65 12.5 R2 0.16523 6904.9 9914 2900 9.6 594.7 5704.4 9584.4 3.98 L.GRN 65 12.5 R3 0.12635 6786.0 18505 17.4 4094.6 11684.8 19080.0 4.28 L.GRN 65 12.5 R4 0.10823 10361.8 14770 3.4 329.3 8824.2 15670.0 4.20 L.GRN 65 9.5 7 0.11746 10531.6 13671 3.2 139.8 6144.6 12414.7 4.09 319 Secondary Time at strain level, sec Agg.Type Gyrs NMAS ID Slope Intercept Max time, sec Flow time, sec 1% 2% 3% 4% Log (TB 0.04 B) L.GRN 65 9.5 R1 0.41597 6282.2 95 3.3 15.2 38.2 83.2 1.92 L.GRN 65 9.5 R3 0.43997 6639.2 38 2.9 11.5 30.3 60.0 1.78 L.GRN 100 19 6 0.11553 7280.6 50436 12.7 1894000.0 10643000.0 19390000.0 7.29 L.GRN 100 19 7 0.32162 3883.9 3129 12.9 42.3 229.6 3084.4 3.49 L.GRN 100 19 10 0.09307 5633.4 16765 215.0 1121000.0 2433000.0 3740000.0 6.57 L.GRN 100 12.5 8 0.10711 5515.2 101007 145.2 8170000.0 19420000.0 30470000.0 7.48 L.GRN 100 12.5 9 0.07791 4765.7 47875 799000.0 6423000.0 12048000.0 17670000.0 7.25 L.GRN 100 12.5 10 0.06906 5184.7 16795 178600.0 1753000.0 3328000.0 4903000.0 6.69 L.GRN 100 9.5 9 0.10112 3099.9 52826 11444.9 30855.6 45696.2 53998.0 4.73 L.GRN 100 9.5 11 0.14190 5063.5 64587 9500 78.1 20434.9 43005.8 56976.3 4.76 L.GRN 100 9.5 12 0.11926 7057.6 52696 12.3 7704.3 32675.2 52416.0 4.72 LMS 65 19 6 0.06804 20637.8 15564 1.9 5.2 53.3 3330000.0 6.52 LMS 65 19 7 0.13867 11116.8 3894 3.0 40.3 1284.7 2964.8 3.47 LMS 65 19 9 0.11416 14794.7 7955 2.5 11.3 199.8 5095.1 3.71 LMS 65 12.5 3 0.33223 10985.9 34 2.2 5.6 17.4 54.0 1.73 LMS 65 12.5 6 0.07743 10531.1 19275 4.3 5924.6 25066.0 54700.0 4.74 LMS 65 12.5 8 0.29495 12100.2 88 1.9 5.3 17.4 64.4 1.81 LMS 65 9.5 4 0.08482 7942.9 19635 12.3 30935.0 88600.0 152000.0 5.18 LMS 65 9.5 6 0.08930 13307.5 70912 2.5 36.1 15434.3 45604.9 4.66 LMS 65 9.5 7 0.11955 11171.0 2841 3.2 92.3 1445.5 3159.0 3.50 LMS 100 19 10 0.11309 13547.5 4164 2.6 16.3 1294.3 10078.0 4.00 LMS 100 19 11 0.07028 15917.7 51816 2.4 13.0 16145.5 39726.0 4.60 LMS 100 19 13 0.07356 15698.4 18155 2.2 13.1 17615.1 68035.0 4.83 LMS 100 12.5 5 0.15825 9818.8 2845 2.9 45.3 1165.2 3191.0 3.50 LMS 100 12.5 6 0.13941 8496.1 7675 5.9 184.8 6214.7 13640.0 4.13 LMS 100 12.5 7 0.09063 14202.8 11824 2.4 21.2 4124.2 13038.0 4.12 320 Secondary Time at strain level, sec Agg.Type Gyrs NMAS ID Slope Intercept Max time, sec Flow time, sec 1% 2% 3% 4% Log (TB 0.04 B) LMS 100 9.5 3 0.10430 10494.6 9504 3.5 494.5 4664.6 11036.0 4.04 LMS 100 9.5 5 0.36228 7337.8 125 2.9 14.7 45.7 109.1 2.04 LMS 100 9.5 8 0.09523 12057.5 40815 2.8 75.4 5227000.0 13100000.0 7.12 R.GRN 65 19 7 0.11278 10496.1 29390 3.3 229.3 24943.9 29536.0 4.47 R.GRN 65 19 8 0.12312 11295.8 18745 2.7 60.5 24400.0 50600.0 4.70 R.GRN 65 19 10 0.09312 11486.2 32893 3.1 194.6 18045.0 37084.0 4.57 R.GRN 65 12.5 7 0.15867 12584.3 2955 2.1 15.3 189.6 1484.7 3.17 R.GRN 65 12.5 9 0.08395 10864.3 20235 3.3 2014.4 20234.8 29406.0 4.47 R.GRN 65 12.5 10 0.08916 14309.6 37769 2.1 25.4 6244.6 25795.1 4.41 R.GRN 65 9.5 6 0.20762 10399.4 1094 2.8 17.6 95.6 764.0 2.88 R.GRN 65 9.5 7 0.09487 9742.3 20205 4.6 3214.8 24258.0 50500.0 4.70 R.GRN 65 9.5 9 0.11548 11538.7 14523 1.6 561000.0 3020000.0 5480000.0 6.74 R.GRN 100 19 3 0.09692 5852.0 662 390 258.6 608.6 638.6 658.6 2.82 R.GRN 100 19 6 0.08660 17145.9 16000 1.9 8.0 258.9 76700.0 4.88 R.GRN 100 19 7 0.08921 8767.5 49805 6.9 195166.0 1007000.0 1820000.0 6.26 R.GRN 100 12.5 4 0.13148 7172.6 47336 11.5 6224.8 52593.0 80700.0 4.91 R.GRN 100 12.5 10 0.14027 7728.4 21325 7.2 864.5 30611.0 87000.0 4.94 R.GRN 100 12.5 11 0.08454 9010.8 53804 45500 6.4 11154.6 47815.4 54129.0 4.73 R.GRN 100 9.5 6 0.13455 10316.9 26082 3.1 78.9 8233.5 24523.8 4.39 R.GRN 100 9.5 7 0.13370 8815.3 8383 4.3 429.0 3523.7 7263.8 3.86 R.GRN 100 9.5 9 0.08901 9303.4 17164 5.4 515000.0 2483000.0 4450000.0 6.65 TRAP 65 19 8 0.34132 11225.1 36.4 1.9 5.4 16.4 42.0 1.62 TRAP 65 19 9 0.40609 9427.2 35.1 2.0 6.1 15.1 37.5 1.57 TRAP 65 19 10 0.36366 11120.2 9.3 1.8 5.0 13.3 27.6 1.44 TRAP 65 12.5 6 0.10406 13237.9 22679 2.3 30.0 3724.1 23975.0 4.38 TRAP 65 12.5 9 0.09856 10213.9 4374 3.9 799.1 35600.0 71400.0 4.85 321 Secondary Time at strain level, sec Agg.Type Gyrs NMAS ID Slope Intercept Max time, sec Flow time, sec 1% 2% 3% 4% Log (TB 0.04 B) TRAP 65 12.5 10 0.12854 11409.1 5186 2.6 70.8 2644.3 7744.0 3.89 TRAP 65 9.5 7 0.13443 13305.0 4739 2.2 15.0 319.4 3374.4 3.53 TRAP 65 9.5 8 0.11923 14017.4 1495 2.0 15.8 654.2 3534.0 3.55 TRAP 65 9.5 9 0.32169 10970.8 57.5 1.9 6.2 18.5 59.5 1.77 TRAP 100 19 3 0.42441 7720.1 11.7 2.6 9.3 25.8 42.6 1.63 TRAP 100 19 7 0.09236 13945.9 10660 2.0 27.6 6604.0 27947.0 4.45 TRAP 100 19 8 0.08947 10287.2 38305 3.5 1594.6 22425.0 38880.0 4.59 TRAP 100 12.5 1 0.13483 8137.6 20945 6.1 654.6 17055.2 33449.0 4.52 TRAP 100 12.5 6 0.10729 10813.3 32237 3.5 124.1 20744.2 50843.0 4.71 TRAP 100 12.5 7 0.09203 9503.2 13064 4.5 2204.1 11004.3 17330.0 4.24 TRAP 100 9.5 6 0.11202 11271.8 19130 2.8 88.2 6165.3 18195.5 4.26 TRAP 100 9.5 7 0.14086 5149.3 274 60 101.8 224.0 259.0 277.0 2.44 TRAP 100 9.5 8 0.13866 8337.0 12648 5.5 479.3 6074.7 12524.9 4.10 L.GRN 40 12.5 6 0.21417 12649.0 285 2.1 8.6 44.1 209.8 2.32 L.GRN 40 12.5 7 0.21221 9787.9 199 2.8 27.6 211.0 437.0 2.64 L.GRN 40 12.5 9 0.19496 14394.1 289 2.0 6.9 28.6 179.2 2.25 R.GRN 40 12.5 6 0.13520 9549.2 23175 5.3 56.6 8283.9 16754.2 4.22 R.GRN 40 12.5 8 0.13234 12899.7 20845 2.5 19.2 469.4 48345.0 4.68 R.GRN 40 12.5 10 0.11655 9248.0 101009 5.5 114.9 22780000.0 49030000.0 7.69 Note: The shady cells are filled with estimated value. 322 TABLE D3 Repeated Load Confining Creep Test Results Microstrain at Cycles Agg. Type Gyrs NMAS ID log intercept, a log slope, b 100 1000 5000 10,000 GVL 65 19 1 7830 0.1240 13858 19420 22511 23381 GVL 65 19 3 4265 0.1597 8897 13670 16616 17645 GVL 65 19 5 7144 0.1152 12145 16498 19064 20431 GVL 65 12.5 3 4061 0.1564 8346 12700 15391 16356 GVL 65 12.5 4 4171 0.1342 7736 11133 13076 13638 GVL 65 12.5 5 6456 0.1399 12296 17596 21256 23337 GVL 65 9.5 1 2820 0.1478 5569 8312 9927 10606 GVL 65 9.5 3 4082 0.1518 8214 12144 14876 15796 GVL 65 9.5 4 4874 0.1282 8795 12283 14521 15246 GVL 100 19 2 4264 0.1207 7435 10718 11925 12673 GVL 100 19 3 3469 0.1621 7318 11379 13797 14592 GVL 100 19 5 4314 0.1059 7027 9801 10636 10814 GVL 100 12.5 1 5051 0.1217 8847 12387 14241 14646 GVL 100 12.5 4 4053 0.1296 7360 10664 12218 12845 GVL 100 12.5 5 2125 0.2678 8053 14178 20787 25026 GVL 100 9.5 1 3556 0.1255 6338 9052 10358 10886 GVL 100 9.5 3 2906 0.2236 8136 13689 19510 20972 GVL 100 9.5 4 3330 0.1219 5837 7916 9403 10029 L.GRN 65 19 6 3578 0.1604 7487 11315 14028 15866 L.GRN 65 19 7 7293 0.1866 17221 27576 35754 38350 L.GRN 65 19 8 3913 0.2251 11035 20140 26622 30084 L.GRN 65 12.5 2 2854 0.2665 9738 17323 27621 38626 L.GRN 65 12.5 6 3100 0.2467 9890 17046 25356 30085 L.GRN 65 12.5 8 3374 0.2505 10693 18368 28511 35614 L.GRN 65 9.5 1 4590 0.1930 11163 17177 23769 28106 L.GRN 65 9.5 2 3478 0.2289 9980 16705 24456 29198 323 Microstrain at Cycles Agg. Type Gyrs NMAS ID log intercept, a log slope, b 100 1000 5000 10,000 L.GRN 65 9.5 6 2474 0.3625 12876 29011 54256 69757 L.GRN 100 19 1 4375 0.1157 7455 10120 11724 12553 L.GRN 100 19 2 3201 0.1215 5603 7565 9014 9595 L.GRN 100 19 3 4279 0.1236 7561 10457 12262 12868 L.GRN 100 12.5 1 2895 0.1363 5424 7730 9247 9873 L.GRN 100 12.5 4 3292 0.1509 6598 9483 11909 13381 L.GRN 100 12.5 5 4000 0.1223 7025 9709 11334 12128 L.GRN 100 9.5 5 2341 0.1662 5033 7765 9649 10486 L.GRN 100 9.5 6 2891 0.1696 6312 9906 12263 13294 L.GRN 100 9.5 8 3990 0.1385 7549 10704 12983 14337 LMS 65 19 1 7801 0.2216 21643 37613 51497 60046 LMS 65 19 4 7480 0.2148 21451 33032 46616 54101 LMS 65 19 5 4517 0.2778 16236 29684 48134 58355 LMS 65 12.5 1 5971 0.1670 12885 20320 24781 26539 LMS 65 12.5 2 5755 0.2169 15627 25062 36534 41054 LMS 65 12.5 5 4886 0.2796 17705 31202 52857 64160 LMS 65 9.5 1 4380 0.1752 9813 15376 19484 22984 LMS 65 9.5 2 4887 0.1979 12157 19194 45728 62660 LMS 65 9.5 3 3387 0.2020 8569 13971 18938 21451 LMS 100 19 1 3204 0.2183 8755 14313 20579 25258 LMS 100 19 2 4678 0.1966 11545 17766 24978 29120 LMS 100 19 4 3970 0.2075 10321 17354 23255 26671 LMS 100 12.5 1 5487 0.1697 11988 18153 23297 26577 LMS 100 12.5 3 5572 0.1599 11618 16760 21766 25898 LMS 100 12.5 4 3542 0.2210 9800 15659 23277 27506 LMS 100 9.5 1 5684 0.1809 13076 19775 26553 32127 324 Microstrain at Cycles Agg. Type Gyrs NMAS ID log intercept, a log slope, b 100 1000 5000 10,000 LMS 100 9.5 2 3513 0.2150 9455 15367 21940 26611 LMS 100 9.5 4 2273 0.3134 9594 16511 32827 45256 R.GRN 65 19 2 6353 0.1599 13224 19359 24810 29453 R.GRN 65 19 3 3784 0.2470 11803 22365 31047 36262 R.GRN 65 19 5 3739 0.2264 10055 18508 25741 29257 R.GRN 65 12.5 2 2517 0.2512 8005 14095 21404 26731 R.GRN 65 12.5 4 4688 0.1966 11572 18514 25041 29660 R.GRN 65 12.5 6 3803 0.2340 11174 19278 27937 34416 R.GRN 65 9.5 2 2938 0.2005 7397 11822 16219 19273 R.GRN 65 9.5 3 3341 0.2174 9091 14792 21293 25311 R.GRN 65 9.5 5 3895 0.2355 11522 19667 28973 35651 R.GRN 100 19 1 2947 0.2626 9877 16562 27619 31035 R.GRN 100 19 2 6148 0.1480 12046 17431 21704 23896 R.GRN 100 19 4 5458 0.1878 12961 20510 27038 29380 R.GRN 100 12.5 1 2827 0.2175 7697 13050 18036 21321 R.GRN 100 12.5 2 4393 0.1519 8803 13063 16032 17211 R.GRN 100 12.5 9 3174 0.2210 8782 14654 20862 24351 R.GRN 100 9.5 1 4292 0.1488 8514 12431 15245 16208 R.GRN 100 9.5 2 2438 0.2311 7065 12074 17462 22730 R.GRN 100 9.5 3 3199 0.2051 8226 13342 18361 20707 TRAP 65 19 1 12049 0.0788 17322 21738 23584 24529 TRAP 65 19 2 4021 0.1813 9266 14861 18841 20161 TRAP 65 19 3 9906 0.0991 15634 23042 23042 25052 TRAP 65 12.5 2 5537 0.1296 10060 14329 16712 17702 TRAP 65 12.5 3 3524 0.1931 8572 13817 18253 21219 TRAP 65 12.5 5 5709 0.1398 10871 15976 18795 19820 325 Microstrain at Cycles Agg. Type Gyrs NMAS ID log intercept, a log slope, b 100 1000 5000 10,000 TRAP 65 9.5 2 3228 0.1802 7401 11546 14983 17345 TRAP 65 9.5 4 7595 0.1410 20747 20747 25252 26915 TRAP 65 9.5 5 7279 0.1724 23753 23753 31621 34298 TRAP 100 19 2 7138 0.1319 13104 19573 21954 22582 TRAP 100 19 4 7061 0.1279 12723 17779 20980 22094 TRAP 100 19 5 5130 0.2505 16256 28925 43340 46130 TRAP 100 12.5 2 2894 0.2348 8533 14479 21377 24820 TRAP 100 12.5 3 6041 0.1676 13070 19875 25176 27788 TRAP 100 12.5 4 4910 0.1889 11721 16979 24542 26226 TRAP 100 9.5 1 2205 0.2470 6878 12653 43867 74934 TRAP 100 9.5 2 4255 0.1624 8989 13642 16969 18990 TRAP 100 9.5 5 2213 0.1611 4647 7172 8729 9355 L.GRN 40 12.5 3 1724 0.3870 10242 24969 46547 60868 L.GRN 40 12.5 4 5420 0.2649 18356 33782 51742 62171 L.GRN 40 12.5 5 3682 0.2791 13314 25696 39672 48139 R.GRN 40 12.5 3 6303 0.2432 19321 34239 50038 59227 R.GRN 40 12.5 4 3629 0.2917 13901 30759 43509 48236 R.GRN 40 12.5 5 4715 0.2416 14345 29664 36914 38954 Note: The shady cells are filled with estimated value. 3 2 6