Dynamic subcomponent testing and finite element simulation of guardrail systems with alternative post installation methodologies

GEORGIA DOT RESEARCH PROJECT 15-08 FINAL REPORT
DYNAMIC SUBCOMPONENT TESTING AND FINITE ELEMENT SIMULATION OF GUARDRAIL
SYSTEMS WITH ALTERNATIVE POST INSTALLATION METHODOLOGIES
OFFICE OF PERFORMANCE-BASED MANAGEMENT AND RESEARCH
15 KENNEDY DRIVE FOREST PARK, GA 30297-2534

TECHNICAL REPORT DOCUMENTATION PAGE

1. Report No.: FHWA-GA-18-1508

2. Government Accession No.:

3. Recipient's Catalog No.:

4. Title and Subtitle:
Dynamic Subcomponent Testing and Finite Element Simulation of Guardrail Systems with Alternative Post Installation Methodologies

5. Report Date: August 2018
6. Performing Organization Code:

7. Author(s): D.W. Scott, L.K. Stewart, D.W. White, E. Bakhtiary, and S.-H. Lee
9. Performing Organization Name and Address: Georgia Institute of Technology School of Civil and Environmental Engineering 790 Atlantic Drive NW Atlanta, GA 30332

8. Performing Organ. Report No.: 15-08
10. Work Unit No.:
11. Contract or Grant No.: 0013698

12. Sponsoring Agency Name and Address: Georgia Department of Transportation Office of Performance-based Management and Research 15 Kennedy Drive Forest Park, GA 30297-2534

13. Type of Report and Period Covered: Final; January 2016 November 2017
14. Sponsoring Agency Code:

15. Supplementary Notes: Prepared in cooperation with the U.S. Department of Transportation, Federal Highway Administration.

16. Abstract: This report details the results of an experimental and numerical investigation into the performance of guardrail posts driven through asphalt layers. Dynamic subcomponent tests were performed on a number of post/asphalt layer configurations. The dynamic test results were used to develop performance criteria to indicate expected levels of restraint for guardrail posts driven through asphalt layers. In addition, subcomponent and full-scale numerical simulations were performed that included a more detailed characterization for the asphalt layer. The tests and simulations indicate that geometric parameters of the asphalt layer are the critical parameters in evaluating the restraint of the asphalt layer on the post. Reducing specific geometric parameters or making targeted alterations of the asphalt layer may reduce restraint in the system to a level similar to that found when using a leave-out around the post.

17. Key Words: Guardrails, mow strip, dynamic testing, finite element analysis

18. Distribution Statement:

19. Security Class (this report):
Unclassified

20. Security Class (this page):
Unclassified

21. Number of Pages: 259

22. Price:

Form DOT 1700.7 (8-69)

GDOT Research Project 15-08
Final Report
DYNAMIC SUBCOMPONENT TESTING AND FINITE ELEMENT SIMULATION OF GUARDRAIL SYSTEMS WITH ALTERNATIVE
POST INSTALLATION METHODOLOGIES
By David Scott, Associate Professor Lauren Stewart, Assistant Professor
Donald White, Professor Esmaeel Bakhtiary, Graduate Research Assistant
Seo-Hun Lee, Graduate Research Assistant
Georgia Tech Research Corporation Atlanta, Georgia Contract with
Georgia Department of Transportation In cooperation with
U.S. Department of Transportation Federal Highway Administration
August 10, 2018
The contents of this report reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Georgia Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation.

TABLE OF CONTENTS

Page

LIST OF TABLES ............................................................................................................. iv

LIST OF FIGURES ............................................................................................................ v

EXECUTIVE SUMMARY ............................................................................................. xiv

ACKNOWLEDGEMENTS............................................................................................ xvii

INTRODUCTION AND BACKGROUND ........................................... 1

1.1

Problem Statement ................................................................................. 1

1.2

Project Objectives................................................................................... 2

1.3

Background ............................................................................................ 2

1.4

Report Organization ............................................................................... 7

SELECTION OF ALTERNATIVES FOR SUBCOMPONENT DYNAMIC TESTS ................................................................................ 9

2.1

GDOT Workshop ................................................................................... 9

2.2

Test Input Parameters ........................................................................... 10

2.3

Dynamic Test Matrix............................................................................ 12

LABORATORY DYNAMIC TESTING ON POST INSTALLATION ALTERNATIVES .................................................. 17

3.1

Development of Dynamic Impact Test Protocol Using High-

Speed Hydraulic Actuator .................................................................... 17

3.2

Dynamic Test Results........................................................................... 33

3.3

Performance Assessment Criteria for Dynamic Tests.......................... 51

3.4

Effect of Mow Strip Design Parameters............................................... 55

DYNAMIC SUBCOMPONENT FINITE ELEMENT SIMULATIONS ................................................................................... 67

4.1

Design of the Impactor ......................................................................... 67

ii

4.2

Mock-up Test Simulation ..................................................................... 69

4.3

Dynamic Mock-up Simulations............................................................ 70

4.4

Dynamic Subcomponent Simulations .................................................. 79

DEVELOPMENT OF EXPERIMENTAL AND NUMERICAL PERFORMANCE CRITERIA ............................................................. 89

5.1

System Assessment Using Quantitative Performance Criteria ............ 89

ASSESSMENT USING MASH FULL-SCALE CRASH SIMULATIONS ................................................................................. 101

6.1

Finite Element Model Description ..................................................... 101

6.2

Evaluation Based on MASH Guidelines ............................................. 103

6.3

Full-Scale Crash Simulation Results.................................................. 106

6.4

Quantitative Comparison between Guardrail Systems with

Different Mow Strips.......................................................................... 121

CONCLUSIONS................................................................................. 125

REFERENCES ................................................................................... 130

APPENDIX A DYNAMIC SUBCOMPONENT TEST SETUP AND DETAILED RESULTS ...................................................................... 135

APPENDIX B MATERIAL CHARACTERIZATION OF ASPHALT ..................... 155

APPENDIX C DETAILED CRASH SIMULATION RESULTS .............................. 171

iii

LIST OF TABLES

Table

Page

1. MASH 3-11 Crash Test Parameters......................................................................... 11 2. Dynamic Test Matrix ............................................................................................... 13 3. Mock-up Test Summary........................................................................................... 22 4. Basic Configuration Tests: Impact Details .............................................................. 39 5. Treatment behind Post Experiments: Impact Details............................................... 45 6. Variation in Dimension Experiments: Impact Details ............................................. 51 7. Performance Assessment Criteria from Dynamic Tests .......................................... 55 8. Performance Ranking for Various Mow Strip Designs ........................................... 65 9. Material Constants Used in the Dynamic Finite Element Subcomponent and
Full-scale Crash Simulations ................................................................................... 72 10. Comparison between FEA and Experimental Results for the Mock-up Tests ........ 76 11. Comparison between Experimental Results and FE Simulations ............................ 90 12. Full-scale Crash Simulations ................................................................................. 102 13. Comparison between the Simulation Results for Guardrail Systems with
Different Mow Strips ............................................................................................. 108 14. Cold Mix Asphalt Compression Test Results ........................................................ 139 15. Accelerometer Specification .................................................................................. 142 16. Data Acquisition System Specification.................................................................. 143 17. High-speed Camera Specification.......................................................................... 144 18. Asphalt Compression Test Results Showing Effect of Aging ............................... 163 19. Asphalt Compression Test Results Showing Effect of Temperature..................... 163 20. CMA Test Results for Various Mixing Ratios....................................................... 169

iv

LIST OF FIGURES

Figure

Page

1. Guardrail Installations: (a) Typical Installation in Georgia; (b) Installation Incorporating Grout Leave-outs as Recommended in the Roadside Design Guide .......................................................................................................................... 1
2. Typical Setup for Asphalt Mow Strip Test Cases by Bligh et al. .............................. 4 3. Installation of Concrete Mow Strip with Alternative Leave-out by Whitesel
et al. ............................................................................................................................ 5 4. Test Pictures of Wood Post with Asphalt Mow Strip by Jowza et al. ....................... 6 5. Test Pictures of Socketed Steel Post with Asphalt Mow Strip by Rosenbaugh
et al. ............................................................................................................................ 7 6. MASH Test 3-11 Crash Test Condition................................................................... 11 7. Representative Dynamic Test Setup Photographs by Test Category....................... 15 8. Tested Mow Strips with Treatment Behind Post: (a) Leave-out; (b) Parallel
Pre-cut; (c) Diagonal Pre-cut ................................................................................... 16 9. Schematic of Dynamic Impact Test Using Hydraulic System................................. 19 10. Mock-up Test on Post with Rigid Connection: (a) Drawing; (b) Experimental
Setup......................................................................................................................... 21 11. Mock-up Test Flyer Raw Accelerations .................................................................. 23 12. Comparison of Programmer Deformation After Impact (Test M4):
(a) Experiment; (b) FEA .......................................................................................... 24 13. Final Programmer Design Used in Dynamic Test Program .................................... 25 14. Schematic Illustration of Dynamic Test Configuration ........................................... 25 15. Locations of Accelerometers for Dynamic Testing ................................................. 26 16. Example of Raw and Filtered Acceleration Signals ................................................ 27 17. Protection and Reinforcement for Accelerometer Cables........................................ 28 18. Removal of Low-frequency Noises from Flyer Acceleration.................................. 29 19. High-speed Camera Setup and Images: (a) Ground Level; (b) Impact Level ......... 30 20. Example of Motion Tracking for Displacements..................................................... 31 21. Example of Target PositionTime History Plots for Multiple Targets.................... 31

v

22. Sequential Photographs for Behavior of Flyer and Post After Impact .................... 32 23. Baseline Test (Test #13): Sequential Photographs for Post Displacements ............ 34 24. Baseline Test (Test #13): Overall Damage After Impact ........................................ 35 25. Typical Mow Strip Test (Test #6): Overall Damage After Impact.......................... 36 26. Typical Mow Strip Test (Test #6): Sequential Photographs for Post
Displacements .......................................................................................................... 37 27. Basic Configuration Tests: AccelerationTime History.......................................... 38 28. Basic Configuration Tests: DisplacementTime History ........................................ 38 29. Sequential Photographs for Leave-out Configuration ............................................. 40 30. Leave-out Test (Test #12): Overall Damage After Impact ...................................... 41 31. Pre-cut Test (Test #8): Overall Damage After Impact............................................. 42 32. Pre-cut Test (Test #8): Sequential Photographs for Post Displacements................. 43 33. Treatment Behind Post Experiments: AccelerationTime History.......................... 44 34. Treatment Behind Post Experiments: DisplacementTime History ........................ 44 35. Thick Mow Strip Test (Test #11): Sequential Photographs for Post
Displacements .......................................................................................................... 46 36. Thick Mow Strip Test (Test #11): Overall Damage After Impact........................... 47 37. Reduced RD Test (Test #7): Overall Damage After Impact.................................... 48 38. Reduced RD Test (Test #7): Sequential Photographs for Post Displacements........ 49 39. Variation in Dimension Experiments: AccelerationTime History......................... 50 40. Variation in Dimension Experiments: DisplacementTime History ....................... 50 41. Sequential Progression of Dynamic Impact Test and Detailed Descriptions .......... 52 42. Assessment Criteria: Effect of Thickness ................................................................ 57 43. ForceDisplacement Curves: Effect of Thickness................................................... 58 44. Assessment Criteria: Effect of Rear Distance.......................................................... 60 45. ForceDisplacement Curves: Effect of Rear Distance ............................................ 61 46. Assessment Criteria: Effect of Pre-cut..................................................................... 63 47. ForceDisplacement Curves: Effect of Pre-cut ....................................................... 64 48. Steel Impactor, Crushable Foam, and Rubber ......................................................... 69
vi

49. Representation of the Mock-up Setup...................................................................... 70 50. StressStrain Curves Used for the Post Flanges ...................................................... 73 51. StressStrain Curves Used for the Post Web ........................................................... 73 52. StressStrain Curve for the Foam Material............................................................. 74 53. Nominal Yield Stress vs. Volumetric Strain for the Foam Material........................ 76 54. Plastic Deformation of the Post at the Connection: (a) FEA; (b) Experiment......... 77 55. Representation of the System After Impact: (a) FEA; (b) Experiment.................... 78 56. Crushed Foam After the Impact: (a) FEA; (b) Experiment ..................................... 78 57. Finite Element Model Used for Dynamic Subcomponent Simulations................... 80 58. Simulation Result for the Model with Only Soil After the Impact .......................... 84 59. Location of the Accelerometers Installed on the Impactor ...................................... 85 60. Location of the Accelerometers Installed on the Impactor ...................................... 85 61. AccelerationTime History Obtained from Experiments and FEA......................... 86 62. Simulation Result for the Model with 3.5-in.-thick Asphalt and Rear Distance
of 24 in. .................................................................................................................... 87 63. Experimental Result for the Case with 3.5-in.-thick Asphalt and Rear
Distance of 24 in. ..................................................................................................... 87 64. AccelerationTime History Obtained from Experiments and the FEA ................... 88 65. Peak Applied Force FEA Contour Plots for Combinations of Thickness and
Rear Distance ........................................................................................................... 92 66. Peak Displacement FEA Contour Plots for Combinations of Thickness and
Rear Distance ........................................................................................................... 93 67. Ground Displacement FEA Contour Plots for Combinations of Thickness and
Rear Distance ........................................................................................................... 94 68. Effective Force FEA Contour Plots for Combinations of Thickness and Rear
Distance.................................................................................................................... 95 69. Combined FEA Contour Plots with Equivalent Restraint to the Leave-out
Setup......................................................................................................................... 96 70. Diagonal Asphalt Pre-cut Used in the Dynamic Tests and Simulations.................. 97 71. Parallel Asphalt Pre-cut Used in the Dynamic Tests and Simulations .................... 98
vii

72. Simulation Result for 3.5-in.-thick Asphalt with 24-in. Rear Distance Using a Stiffer Asphalt .......................................................................................................... 99
73. FE Model with 29 Posts ......................................................................................... 102 74. The Suggested Coordinate System in MASH........................................................ 105 75. Summation of Ground-level Displacement of Posts Compared for Guardrail
Systems with Different Mow Strips Based on Simulation Results........................ 123 76. Dimension of Dynamic Test Bed (6-cubic-yard Dumpster) .................................. 136 77. Steel Container Reinforcement: (a) Drawing; (b) Fabrication............................... 137 78. Asphalt Mow Strip Installation: (a) Shear Studs; (b) Compaction ........................ 138 79. Post Driving for Dynamic Test Bed....................................................................... 140 80. Adjustable Flyer Mass ........................................................................................... 141 81. Safety Chain System .............................................................................................. 141 82. Accelerometer (PCB 356B20) ............................................................................... 143 83. Data Acquisition System (Hi-Techniques Synergy P)........................................... 143 84. High-speed Cameras Used in Dynamic Tests........................................................ 144 85. Summary of Dynamic Test Results: Baseline........................................................ 145 86. Summary of Dynamic Test Results: Typical Mow Strip....................................... 146 87. Summary of Dynamic Test Results: Leave-out Installation .................................. 147 88. Summary of Dynamic Test Results: Parallel Pre-cut............................................. 148 89. Summary of Dynamic Test Results: Diagonal Pre-cut .......................................... 149 90. Summary of Dynamic Test Results: Thin Mow Strip............................................ 150 91. Summary of Dynamic Test Results: Thick Mow Strip.......................................... 151 92. Summary of Dynamic Test Results: Reduced RD................................................. 152 93. Summary of Dynamic Test Results: Thick and Reduced RD................................ 153 94. MohrCoulomb Parameters from Unconfined Compression Test ........................ 157 95. Cohesion Ratio and Angle of Friction ................................................................... 158 96. Difference in Mow Strip Fracture Between Two Tests ......................................... 159 97. Asphalt Test Bed and Representative Cored Specimens ....................................... 160 98. Temperature Conditioning of Cored Specimens.................................................... 161
viii

99. Test Specimen in Compressive Failure.................................................................. 162 100. Empirical TemperatureCompressive Strength Model for Asphalt ...................... 165 101. Empirical Model of Cohesion Versus Age for Asphalt ......................................... 166 102. Cold Mix Asphalt Specimen Preparation............................................................... 168 103. Unconfined Compression Test Pictures of CMA Specimen: (a) Pre-test;
(b) Post-test ............................................................................................................ 169 104. CMA Test Results: Compressive Strength vs. Cement Content............................ 170 105. Simulation Result for the Guardrail System Without Mow Strip Up to
0.3 Sec .................................................................................................................... 171 106. Simulation Result for the Guardrail System Without Mow Strip After
0.3 Sec .................................................................................................................... 172 107. Simulation Result for the Guardrail System Without Mow Strip Up to
0.3 Sec .................................................................................................................... 173 108. Simulation Result for the Guardrail System Without Mow Strip After
0.3 Sec .................................................................................................................... 174 109. Simulation Result for the Guardrail System Without Mow Strip.......................... 175 110. Vehicle Deformation for the Guardrail System with Soil Only............................. 175 111. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Soil Only ..................... 176 112. Relative Longitudinal Velocity of the Occupant (ft/s) Soil Only ....................... 176 113. Relative Longitudinal Displacement of the Occupant (ft) Soil Only.................. 177 114. 33 ft/s Average Vehicle Lateral Acceleration (g) Soil Only............................... 177 115. Relative Lateral Velocity of the Occupant (ft/s) Soil Only ................................ 178 116. Relative Lateral Displacement of the Occupant (ft) Soil Only........................... 178 117. Vehicle Roll for the Setup with Soil Only (deg).................................................... 179 118. Vehicle Pitch for the Setup with Soil Only (deg) .................................................. 179 119. Vehicle Yaw for the Setup with Soil Only (deg) ................................................... 180 120. Simulation Result for Test T2-R24 Up to 0.3 Sec .............................................. 181 121. Simulation Result for Test T2-R24 After 0.3 Sec............................................... 182 122. Simulation Result for Test T2-R24 Up to 0.3 Sec .............................................. 183 123. Simulation Result for Test T2-R24 After 0.3 Sec............................................... 184
ix

124. Simulation Result for Test T2-R24 ........................................................................ 185 125. Vehicle Deformation for Test T2-R24................................................................... 185 126. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T2-R24 ................ 186 127. Relative Longitudinal Velocity of the Occupant (ft/s) Test T2-R24 .................. 186 128. Relative Longitudinal Displacement of the Occupant (ft) Test T2-R24............. 187 129. 33 ft/s Average Vehicle Lateral Acceleration (g) Test T2-R24.......................... 187 130. Relative Lateral Velocity of the Occupant (ft/s) Test T2-R24............................ 188 131. Relative Lateral Displacement of the Occupant (ft) Test T2-R24 ...................... 188 132. Vehicle Roll (deg) Test T2-R24.......................................................................... 189 133. Vehicle Pitch (deg) Test T2-R24 ........................................................................ 189 134. Vehicle Yaw (deg) Test T2-R24 ......................................................................... 190 135. Simulation Result for Test T3.5-R12 Up to 0.3 Sec ........................................... 191 136. Simulation Result for Test T3.5-R12 After 0.3 Sec............................................ 192 137. Simulation Result for Test T3.5-R12 Up to 0.3 Sec ........................................... 193 138. Simulation Result for Test T3.5-R12 After 0.3 Sec............................................ 194 139. Simulation Result for Test T3.5-R12 ..................................................................... 195 140. Vehicle Deformation for Test T3.5-R12................................................................ 195 141. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R12 ............. 196 142. Relative Longitudinal Velocity of the Occupant (f/s) Test T3.5-R12 ................ 196 143. Relative Longitudinal Displacement of the Occupant (ft) Test T3.5-R12.......... 197 144. 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R12....................... 197 145. Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R12......................... 198 146. Relative Lateral Displacement of the Occupant (ft) Test T3.5-R12 ................... 198 147. Vehicle Roll (deg) Test T3.5-R12....................................................................... 199 148. Vehicle Pitch (deg) Test T3.5-R12 ..................................................................... 199 149. Vehicle Yaw (deg) Test T3.5-R12 ...................................................................... 200 150. Simulation Result for Test T3.5-R24 Up to 0.3 Sec ........................................... 201 151. Simulation Result for Test T3.5-R24 After 0.3 Sec............................................ 202
x

152. Simulation Result for Test T3.5-R24 Up to 0.3 Sec ........................................... 203 153. Simulation Result for Test T3.5-R24 After 0.3 Sec............................................ 204 154. Simulation Result for Test T3.5-R24 ..................................................................... 205 155. Vehicle Deformation for Test T3.5-R24................................................................ 205 156. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24 ............. 206 157. Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24 ............... 206 158. Relative Longitudinal Displacement of the Occupant (ft) Test T3.5-R24.......... 207 159. 10 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24....................... 207 160. Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24......................... 208 161. Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24 ................... 208 162. Vehicle Roll (deg) Test T3.5-R24....................................................................... 209 163. Vehicle Pitch (deg) Test T3.5-R24 ..................................................................... 209 164. Vehicle Yaw (deg) Test T3.5-R24 ...................................................................... 210 165. Simulation Result for Test T6-R24 Up to 0.3 Sec .............................................. 211 166. Simulation Result for Test T6-R24 After 0.3 Sec............................................... 212 167. Simulation Result for Test T6-R24 Up to 0.3 Sec .............................................. 213 168. Simulation Result for Test T6-R24 After 0.3 Sec............................................... 214 169. Simulation Result for Test T6-R24 ........................................................................ 215 170. Vehicle Deformation for Test T6-R24................................................................... 215 171. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T6-R24 ................ 216 172. Relative Longitudinal Velocity of the Occupant (ft/s) Test T6-R24 .................. 216 173. Relative Longitudinal Displacement of the Occupant (ft) Test T6-R24............. 217 174. 33 ft/s Average Vehicle Lateral Acceleration (g) Test T6-R24.......................... 217 175. Relative Lateral Velocity of the Occupant (ft/s) Test T6-R24............................ 218 176. Relative Lateral Displacement of the Occupant (ft) Test T6-R24 ...................... 218 177. Vehicle Roll (deg) Test T6-R24.......................................................................... 219 178. Vehicle Pitch (deg) Test T6-R24 ........................................................................ 219 179. Vehicle Yaw (deg) Test T6-R24 ......................................................................... 220
xi

180. Simulation Result for Test T3.5-R24-C Up to 0.3 Sec ....................................... 221 181. Simulation Result for Test T3.5-R24-C After 0.3 Sec........................................ 222 182. Simulation Result for Test T3.5-R24-C Up to 0.3 Sec ....................................... 223 183. Simulation Result for Test T3.5-R24-C After 0.3 Sec........................................ 224 184. Simulation Result for Test T3.5-R24-C ................................................................. 225 185. Vehicle Deformation for Test T3.5-R24-C............................................................ 225 186. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24-C ......... 226 187. Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24-C ........... 226 188. Relative Longitudinal Displacement of Occupant (ft) Test T3.5-R24-C............ 227 189. 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24-C................... 227 190. Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24-C..................... 228 191. Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24-C ............... 228 192. Vehicle Roll (deg) Test T3.5-R24-C................................................................... 229 193. Vehicle Pitch (deg) Test T3.5-R24-C ................................................................. 229 194. Vehicle Yaw (deg) Test T3.5-R24-C .................................................................. 230 195. Simulation Result for Test T3.5-R24-S Up to 0.3 Sec........................................ 231 196. Simulation Result for Test T3.5-R24-S After 0.3 Sec ........................................ 232 197. Simulation Result for Test T3.5-R24-S Up to 0.3 Sec........................................ 233 198. Simulation Result for Test T3.5-R24-S After 0.3 Sec ........................................ 234 199. Simulation Result for Test T3.5-R24-S ................................................................. 235 200. Vehicle Deformation for Test T3.5-R24-S ............................................................ 235 201. 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24-S .......... 236 202. Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24-S............ 236 203. Relative Longitudinal Displacement of Occupant (ft) Test T3.5-R24-S ............ 237 204. 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24-S ................... 237 205. Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24-S ..................... 238 206. Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24-S ............... 238 207. Vehicle Roll (deg) Test T3.5-R24-S ................................................................... 239
xii

208. Vehicle Pitch (deg) Test T3.5-R24-S.................................................................. 239 209. Vehicle Yaw (deg) Test T3.5-R24-S .................................................................. 240
xiii

EXECUTIVE SUMMARY
Steel guardrail is the most common roadside barrier installed along Georgia's 20,000 miles of interstates and state routes. The objective of this multiphase research program is to evaluate the structural behavior of guardrail posts embedded through asphalt layers. Phase I of this research focused on static evaluation and numerical simulation of the structural performance of guardrail posts installed in accordance with current Georgia Department of Transportation (GDOT) procedures that include a mow strip as well as alternative installation options developed in consultation with GDOT. A subset of the most promising alternative installation methods was selected for further evaluation under subcomponent dynamic loading in the Phase II effort, which is the subject of this report. The dynamic tests' results were used to refine and expand results of finite element analysis (FEA) simulations.
Most of the guardrail installed with mow strips in Georgia prior to 2017 employed the configuration shown in GDOT Standard Detail S-4-2002. However, in every GDOT District in Georgia approximately 10% of guardrail/mow strip installations do not conform to GDOT S-4-2002 in terms of mow strip geometry.
Based on the results of the Phase I research program and a series of workshops with selected GDOT Design, Maintenance, and Construction personnel, a test matrix for dynamic subcomponent experiments was developed to better understand the response of guardrail posts embedded in asphalt layers. A novel dynamic subcomponent testing method for testing guardrail systems using a high-speed hydraulic actuator was developed for use in this program.
xiv

Detailed finite element simulations of the subcomponent tests were performed that included a more detailed characterization of material properties for the asphalt layer. Fullscale crash simulations were also conducted to assess the performance of guardrail posts embedded in soil and asphalt mow strips. Numerous parameters based on the AASHTO Manual for Assessing Safety Hardware (MASH) guidelines were used to evaluate the simulations.
The experimental results and simulations were correlated to develop a methodology for the rapid evaluation of alternative mow strip configurations in terms of restraint on the post. The experimental and numerical simulation results indicate that decreasing the mow strip thickness and/or rear distance behind the post can be an effective way to reduce the restraint imparted by a mow strip on a guardrail system. For existing installations with unfavorable mow strip geometry, the installation of targeted pre-cuts into the asphalt layer appears to dramatically reduce the restraint of the mow strip on the posts.
The subcomponent testing, subcomponent finite element analysis, and full-scale crash simulations all indicate that the restraint on the guardrail posts that occurs using the GDOT S-4-2002 configuration typical in Georgia exceeds the restraint on the post that occurs for mow strips that include leave-outs. The results also indicate that reducing specific geometric parameters in the mow strip or using cuts in the asphalt layer may reduce the restraint in the system to a level similar to that found in a mow strip that incorporates a leave-out in accordance with the AASHTO Roadside Design Guide. However, these results cannot be taken as a definitive indicator of the performance of guardrail / mow strip configurations in actual crash conditions. Full-scale crash testing at an approved facility in
xv

accordance with the MASH guidelines is necessary to definitively evaluate the effectiveness of a given mow strip configuration.
xvi

ACKNOWLEDGEMENTS The following individuals at GDOT provided many valuable suggestions throughout this study: Mr. Brent Story, State Design Policy Engineer; Mr. Daniel Pass, Assistant State Design Policy Engineer; Mr. Beau Quarles, Assistant Construction Engineer; Mr. Walter Taylor, Senior Design Engineer; and Mr. David Jared, Assistant Office Head (Research), Office of Performance-based Management and Research. The opinions and conclusions expressed herein are those of the authors and do not represent the opinions, conclusions, policies, standards, or specifications of GDOT or of other cooperating organizations. At the Georgia Institute of Technology, Jeremy Mitchell provided help and expertise with equipment and research tasks at the Georgia Tech SEMM Laboratory. Javaid Anwar, Nan Gao, and Jiuk Shin assisted with the construction of the experimental test specimens. Catherine Sanborn, Marc Sanborn, and Gennieve Pizzola provided invaluable assistance overseeing the loading equipment during the dynamic testing program. The authors express their profound gratitude to all of these individuals for their assistance and support during the completion of this research project.
xvii

INTRODUCTION AND BACKGROUND 1.1 Problem Statement
Prior to March 2017, the preferred procedure for steel guardrail installation in the state of Georgia [1,2] employed a post-installation machine, which is typically hydraulic, to drive the posts through a layer of asphalt (i.e., a "mow strip") placed to retard vegetation growth around the system (Figure 1(a)). This procedure was outlined in Georgia Department of Transportation (GDOT) Standard Detail S-4-2002 (referred to hereafter as GDOT S-4-2002). However, to avoid undesirable restraint at the ground line, the Fourth Edition of the AASHTO Roadside Design Guide [3] recommends a post installed incorporating grout leave-outs (LOs) (Figure 1(b)). This recommendation is based on research performed by the Texas Transportation Institute (TTI) [4,5].
FIGURE 1 Guardrail Installations: (a) Typical Installation in Georgia; (b) Installation Incorporating Grout Leave-outs as Recommended in the Roadside Design Guide [3]
1

1.2 Project Objectives The objective of this research program is to evaluate the structural behavior of
guardrail posts embedded through asphalt layers. Phase I of this research focused on static evaluation and numerical simulation of the structural performance of guardrail posts installed in accordance with current GDOT procedures that include a mow strip [5], as well as alternative installation options developed in consultation with GDOT. A subset of the most promising alternative installation methods was selected for further evaluation under subcomponent dynamic loading in the Phase II effort, which is the subject of this report. The dynamic tests' results are used to refine and expand the results of finite element analysis (FEA) simulations. Phase III of the research program will involve a full-scale crash test of a standard guardrail system installed in accordance with GDOT S-4-2002; the results of this crash test will be presented in a subsequent report.
Steel guardrail is the most common roadside barrier installed along Georgia's 20,000 miles of interstates and state routes [6]. This multiphase research program addresses a specific concern raised by GDOT personnel relating to the safety and efficacy of current state guardrail installation procedures in comparison to guidelines found in the Roadside Design Guide. The safety and effectiveness of the guardrail systems installed using these procedures must be rigorously evaluated to ensure compliance with Federal Highway Administration (FHWA) guidelines.
1.3 Background Relatively few research studies have been performed to address the effect of asphalt
layers (i.e., "mow strips") on the overall behavior of guardrail posts [4,5,79]. In addition,
2

dimensions and material properties of the mow strip have not been evaluated for their parametric impact on the ground-level restraint of guardrail posts.
Research performed by TTI to investigate the impact of mow strips on the performance of guardrail systems [4] formed the basis for the adoption of the guardrail post installation detail incorporating grout leave-outs into the AASHTO Roadside Design Guide. The researchers at TTI examined the performance of guardrail and mow-strip systems using experimental evaluation and numerical simulation. A concrete mow strip of 5-in. thickness and an asphalt mow strip of 8-in. thickness were considered, in addition to the presence of "leave-out" sections around posts. Seventeen configurations of wood and steel guardrail posts embedded in various confinement conditions were subjected to dynamic impact testing with a bogie vehicle (Figure 2). The dynamic impact tests were numerically simulated, and full-scale mow strip system models were assembled using the subcomponent models. Based on predictive numerical simulations, a concrete mow strip with grout-filled leave-outs was selected for full-scale crash testing in accordance with National Cooperative Highway Research Program Report 350 (NCHRP 350) criteria [10]. Crash tests of a steel post guardrail system and wood post guardrail system encased in the selected mow strip configuration were deemed successful.
Further research on the performance of guardrail systems with concrete mow strips was presented by TTI in 2009 [5]. This work focused primarily on alternative materials used in the post leave-outs: a urethane foam, two types of molded rubber mat, and a precast concrete wedge. All posts were placed with a concrete mow strip of 5-in. thickness but with different leave-out materials. These alternative configurations were evaluated using the bogie vehicle employed in the previous study. The authors asserted that three of the
3

four alternative leave-out materials demonstrated satisfactory performance in comparison with a post with no mow strip installed.
FIGURE 2 Typical Setup for Asphalt Mow Strip Test Cases by Bligh et al. [4] In 2011, a study incorporating a low-strength concrete mow strip was conducted by the California Department of Transportation (CALTRANS) to solve the problem of weed growth beneath a metal beam guardrail system [7]. In this study, two guardrail systems installed with low-strength concrete mow strips with expanded polystyrene foam around steel guardrail posts were tested according to NCHRP 350 as shown in Figure 3. The installation of relatively weak foam material around the posts was an alternative to the leave-out method proposed by the TTI studies. From two full-scale crash tests, the CALTRANS researchers recommended the depth (thickness) of the low-strength concrete
4

mow strip should be 2 in. or less to achieve the desired performance of their guardrail system.
FIGURE 3 Installation of Concrete Mow Strip with Alternative Leave-out
by Whitesel et al. [7] In 2012, Jowza et al. [8] conducted a series of dynamic tests using a bogie vehicle to investigate the effect of asphalt mow strips located on a sloped terrain. In most of the tests conducted, the impacted wood posts could break the asphalt mow strips and rotate backward (Figure 4) when 2-in. asphalt mow strips were used. However, the researchers recommended the wood posts not be completely surrounded by asphalt since the postsoil
5

resistance observed from the mow stripincorporated tests was higher than the resistance of tests without a mow strip.
In 2015, a series of dynamic impact tests on weak steel posts (S35.7) embedded in three different surrounding soil conditions (i.e., weak soil, strong soil, and strong soil covered by an asphalt mow strip) was conducted by Rosenbaugh et al. [9]. A total of 10 bogie vehicle tests were run, and one of the tests included the asphalt mow strip of 4-in. thickness and 24-in. rear distance (RD). As shown in Figure 5, the test with a mow strip showed excessive ground-level restraint that prevented the rotation of the post and caused the formation of a plastic hinge at ground level.
FIGURE 4 Test Pictures of Wood Post with Asphalt Mow Strip by Jowza et al. [8]
6

FIGURE 5 Test Pictures of Socketed Steel Post with Asphalt Mow Strip by Rosenbaugh et al. [9] 1.4 Report Organization
Chapter 2 of this report summarizes the selection of alternative installation configurations as candidates for further evaluation through subcomponent dynamic testing and both subcomponent and full-scale dynamic numerical simulation.
Chapter 3 summarizes the development of a dynamic subcomponent test program carried out on the selected alternatives. These results were analyzed to provide a better understanding of the dynamic behavior of a post restrained with an asphalt layer at the ground line in a variety of alternative installations.
Chapter 4 summarizes the development of advanced finite element models to analyze the the dynamic behavior of the guardrail posts to compare to the results of the subcomponent experimental program discussed in Chapter 3.
7

Chapter 5 outlines the use of the experimental and simulation results in the development of performance criteria to identify critical parameters influencing the dynamic performance of the posts embedded in asphalt layers.
Chapter 6 presents the results of finite element simulations of full-scale crash tests on guardrail systems embedded through asphalt layers. The simulations included variations in asphalt layer geometry and layout.
Chapter 7 contains the conclusions for Phase II of the research program. Chapter 8 contains the references cited in this report. The Appendices contain detailed descriptions of the testing procedures and results, along with the computational methods and procedures utilized in Phase II of the research program.
8

SELECTION OF ALTERNATIVES FOR SUBCOMPONENT DYNAMIC TESTS
2.1 GDOT Workshop A workshop with selected GDOT personnel was held on May 18, 2016, to present
the project deliverables from the Phase I research program and to solicit guidance for the Phase II research effort. The specific objectives for the workshop were the following:
Get feedback on actual range of geometric parameters related to mow strip installations in Georgia--how representative is the GDOT standard compared to what is actually in the field?
Solicit input from stakeholders and other individuals with the greatest combination of experience and expertise on the viability of alternate mow strip design and remediation methods.
In particular, focus on constructability/economic issues and viability for recommended design and/or remediation methods to be broadly applied in Georgia.
The takeaways from the workshop discussions were the following: As of March 2016, paving under posts for vegetation control is not really a concern. Paving under posts to reduce erosion/washout is still a consideration, but other techniques exist to help with this problem. Most of the mow strips in Georgia will be very similar to GDOT S-4-2002. There are a few locations (<10% on average) in each District where the geometry is very different than GDOT S-4-2002, but these locations are not well identified.
9

Retrofits for existing mow strip employing saw-cuts is a potential option from a construction standpoint.
Adding leave-outs via retrofit on existing asphalt in all locations is not viable. These discussions were incorporated into the development of the dynamic subcomponent testing program. 2.2 Test Input Parameters
One of the most critical steps for dynamic impact testing is to determine a reference dynamic loading condition with test input parameters. As a first step, the reference kinetic energy level for dynamic tests was determined using the crash test parameters of the MASH Test 3-11 [11]. The MASH Test 3-11 configuration specifies a test section and conditions for full-scale crash tests using a pickup truck impacting a guardrail system as shown in Figure 6 and Table 1. The test guideline specifies a dynamic load in terms of mass (m), impact angle (), and velocity (v) of the testing vehicle for the entire guardrail system. Hence, a dynamic impact load profile (e.g., loadtime history curve) for a single guardrail post is not available from the test guideline or available in test reports from other roadside testing agencies.
10

FIGURE 6 MASH Test 3-11 Crash Test Condition (adapted from [11])

TABLE 1 MASH Test 3-11 Crash Test Parameters [11]

Test Condition
Weight of a pickup truck (m) Impact velocity (v) Impact angle () Number of posts (n) (length of test section)

Values 5000 lb 62 mph
25
10

The total lateral kinetic energy (EK) contributing to the lateral displacement of the

guardrail system can be calculated as shown in Equation (2-1). By assuming the lateral

kinetic energy is distributed over 10 guardrail posts along the length of the test section, the

average kinetic energy input on single post (EK,avg) can be estimated as 137.9 kip-in. as

shown in Equation (2-2). This value can be used as the reference kinetic energy input for

test configurations including a single post.



=

1 2

(

sin

)2

=

1379

kip-in.

, = / = 137.9 kip-in.

(2-1) (2-2)

11

Even though numerous combinations of velocity and mass can yield the reference kinetic energy, both the maximum capacity of the hydraulic actuator and the upper limit of the flyer dimension to avoid potential interference between test components were considered for conducting a dynamic test safely. As such, an impact velocity of 32.4 mph and a flyer mass of 305 lb were selected as the most feasible dynamic test input parameters for the test program. 2.3 Dynamic Test Matrix
The selected dynamic test matrix is given in Table 2. The test program classifies each test into one of three categories: basic configuration, treatment behind post, and variation in dimensions. Test setup photographs for all test categories are shown in Figure 7. An impact velocity of 32.4 mph and a flyer mass of 305 lb were used throughout the entire test program.
12

TABLE 2 Dynamic Test Matrix

Test Category
Basic Configuration
Treatment Behind Post

Test Configuration
Baseline
Typical mow strip Leave-out
Pre-cut (parallel) Pre-cut (diagonal)

Mow Strip Dimension Thick. RD* (in.) (in.)

0

0

3.5 24

3.5 24

Test Date (Test Number)
9/12/16 (#1) 9/29/16 (#2) 1/26/17 (#13) 10/27/16 (#3) 12/19/16 (#6) 10/28/16 (#5) 1/26/17 (#12) 10/27/16 (#4) 12/20/16 (#8)
1/5/17 (#10)

Thin

1.5 24 1/5/17 (#9)

Variation in Dimension

Thick Reduced RD

5.5 24 1/6/17 (#11) 3.5 12 12/19/16 (#7)

Thick and reduced RD

5.5 12 1/27/17 (#14)

* RD = Rear distance of mow strip behind trailing flange of post

Test Condition

Temp. (F) 85 75 71 79 66 82 71 79 68

Asphalt Age
15 days 16 days 15 days 13 days 13 days 13 days

71 16 days

71 16 days 71 16 days 68 13 days

69 11 days

A total of five tests are classified as having a "basic configuration" category in Table 2. These include a baseline configuration with no mow strip and with the least amount of ground-level restraint. The main purpose of this test was to provide a lower bound for system response regarding the ground-level restraint. The typical mow strip configuration given in GDOT S-4-2002 (3.5-in. thick and 24-in. rear distance) was identified from the workshops as the most common mow strip dimensions in U.S. roadways. The typical configuration represents a test configuration with a moderately rigid ground-level restraint that other mow strip configurations can be compared to.
The "treatment behind post" category in Table 2 includes a total of five tests. Thickness and rear distance in these tests were set equal to that in GDOT S-4-2002. As

13

shown in Figure 8, two treatment techniques intended to reduce ground-level restraint were applied at the mow strip area behind the post. Two tests including a leave-out application (based on recommendations in the AASHTO Roadside Design Guide) were conducted to establish a reference performance level for a post struck by the impactor. The 28-day compressive strength of the grout materials used with the leave-out was less than 120 psi, which satisfies the Roadside Design Guide recommendation. Next, three pre-cut mow strip configurations were tested as an alternative to the leave-out treatment.
A total of four tests are classified in the "variation in dimension" category in Table 2. The main purpose of these tests was to provide a quantitative guideline on the variability in ground-level restraint caused by changes in mow strip geometric parameters. Three mow strip thicknesses (1.5 in., 3.5 in., and 5.5 in.) and two rear distances (12 in. and 24 in.) were considered in the development of the test matrix; ultimately, four cases (1.5 in. 24 in., 5.5 in. 24 in., 3.5 in. 12 in., and 5.5 in. 12 in.) were tested under the dynamic test protocol.
14

FIGURE 7 Representative Dynamic Test Setup Photographs by Test Category
15

(a)

(b)

(c)

FIGURE 8
Tested Mow Strips with Treatment Behind Post: (a) Leave-out; (b) Parallel Pre-cut; (c) Diagonal Pre-cut

16

LABORATORY DYNAMIC TESTING ON POST INSTALLATION ALTERNATIVES
A novel dynamic subcomponent test method was developed using a high-speed hydraulic actuator in the performance assessment of guardrail posts in asphalt layers. A total of 14 dynamic impact tests were conducted under various mow strip designs and configurations. Experimental data from accelerometers and high-speed cameras were utilized to quantify the relative ground-level restraint of the guardrail post driven through the asphalt mow strip. A series of quantitative performance criteria was selected to evaluate the structural performance of the subcomponent system. These experimental results were used as one of the assessment tools in this study to aid in the understanding of the effect of mow strip design variables on guardrail post performance.
3.1 Development of Dynamic Impact Test Protocol Using High-Speed Hydraulic Actuator Historically, in lieu of performing costly full-scale vehicle crash tests, the dynamic
performance of guardrail subcomponents has been examined through four different dynamic test methods--the gravitational pendulum, drop mass, reduced-scale models, and bogie vehicles [11]. The most common method employed is the bogie vehicle method, which has been primarily used for tests on guardrail posts of various designs and other roadside safety hardware subcomponents.
However, a dynamic test method using a high-speed hydraulic actuator can be employed in testing various roadside safety hardware subcomponents as an alternative methodology. The test method with the hydraulic actuator features several advantages over
17

the existing bogie vehicle method: (1) the amount of impact energy transferred from the actuator can be readily controlled or estimated; (2) the actuator allows the user to control, measure, and replicate the impact features more precisely and safely; (3) supplementary instrumentation on the test specimen and on the actuator itself is available; and (4) testing under controlled indoor conditions allows a user to minimize the variability in ambient conditions such as the moisture content in the soil base around the post and under the mow strip.
3.1.1 High-speed hydraulic actuator The high-speed hydraulic actuator, located in the Structural Engineering and
Mechanics and Materials (SEMM) Laboratory of Georgia Institute of Technology, was designed to produce an impulse by impacting the test specimen in a controlled manner [12]. The actuator used in the dynamic test program is capable of producing a computercontrolled impact up to 73.5 miles per hour (mph) velocity or 890 kip-in. of kinetic energy with a repeatability in velocity of no greater than 4% [13]. The desired impact condition is achieved through the precise timing of valve opening and pressure control in the actuator system, along with appropriate loading fixture design. The actuator system can be programmed to simulate an equivalent vehicle impact condition by providing a specific amount of kinetic energy and initial impact velocity.
Figure 9 shows a schematic illustration of the hydraulic system, which consists of an actuator, control valves, accumulators, and transducers [14]. The flyer mass is located next to the impactor plate, attached to the piston rod of the hydraulic actuator. To simulate the one-dimensional (1-D) movement of a vehicle with a constant velocity as closely as possible, a rail system securely guides the flyer mass that is accelerated by the impactor
18

plate, released at the desired velocity, and subsequently impacts the target structure at the desired impact height. During the test, the pressure valve openings are precisely controlled by the main control computer to accelerate the piston rod and flyer mass to the desired impact velocity [13].
FIGURE 9 Schematic of Dynamic Impact Test Using Hydraulic System (after [12,13]) 3.1.2 Loading calibration with mock-up experiments The magnitude and duration of the dynamic impact loading can be tailored not only by controlling the hydraulic setting but also by placing a relatively weak and deformable medium (e.g., urethane foam) on the impact side of the flyer mass. This deformable medium acts as a programmer, which is equivalent to a shock absorber in vehicles. A programmer transfers the energy and momentum of the hydraulic force to the test specimen. It also reduces the magnitude of acceleration so that the peak acceleration can be within the measurement range of accelerometers. Equation (3-1) explains the general concept of the dynamic impact testing, including a programmer-attached flyer mass. The net force, (), acting on the flyer mass during the collision is equal to the product of its mass, , and acceleration, (), which is measured by accelerometers located on the non-impact side of the flyer. The actual
19

force on the test specimen (guardrail post), (), can be adjusted by a modification factor, (), related to the programmer (e.g., geometry and/or material), which is primarily a function of velocity, .

F (t)net = ma(t) = (v)F (t)specimen

(3-1)

Thus, as shown in Equation (3-2), the dynamic force on the specimen can be manipulated

by the adjustment and calibration of the mass and the programmer of the impacting module.

F (t)specimen

=

ma(t) (v)

(3-2)

In this study, a series of "mock-up" tests, classified as 1-D momentum transfer

experiments, was conducted to calibrate the forces from the hydraulic system. It was

critical to confirm if a dynamic test with the most restrained condition of the post could be

conducted safely within a reasonable range of accelerations. Hence, the mock-up test was

designed to approximate a post in the most restrained condition. The mock-up test

configuration includes a standard steel post with two supports fixed to the rigid steel frame

using bolted connections as shown in Figure 10.

20

(a)

(b)

FIGURE 10
Mock-up Test on Post with Rigid Connection: (a) Drawing; (b) Experimental Setup

Various impact speeds and programmer designs were tested to determine the appropriate hydraulic control needed to achieve the desired dynamic impact load in the dynamic test program. Table 3 summarizes results from the mock-up tests, and Figure 11 shows the flyer raw acceleration curves. It was determined from the M1 test that the 4-in.-thick neoprene rubber programmer design was not suitable due to its high acceleration response. In the M2 and M3 tests, two different types of medium-density (10 lb/ft3) urethane foam programmers were tested and a significant reduction in accelerations was observed compared to the M1 results. The M4 test indicated that using a low-density (6 lb/ft3) urethane foam is a more effective way to reduce the flyer acceleration and to increase the deformation of the programmer. Based on these results, the research team constructed and calibrated a reference FEA model. The FEA simulation for the M4 test shows reasonable agreement in both the maximum acceleration of the flyer and

21

deformation of the programmer. Figure 12 compares the two deformed shapes of the programmer, one from the experiment and the other from the FEA simulation.

TABLE 3 Mock-up Test Summary

Test Number

Impact Speed (mph)

Programmer Design and Thickness
(in.)

Max. Acceleration of Flyer Mass (g)

Exp.

FEA

Max. Deformation of Programmer (in.)

Exp.

FEA

M1

13.0

(1): 4

500



0



M2

13.0

(2): 4

54



0.2



M3

13.0

(3): 6

46



0



M4

21.6

(4): 5

52

65

1.6

1.46

(1) 4-thick neoprene rubber (2) 3-thick medium-density (10 lb/ft3) urethane foam + 1-thick neoprene rubber
(3) 2-floral foam + 3-thick medium-density urethane foam + 1-thick neoprene rubber (4) 4-thick low-density (6 lb/ft3) urethane foam + 1-thick neoprene rubber

22

FIGURE 11 Mock-up Test Flyer Raw Accelerations
23

(a)

(b)

FIGURE 12
Comparison of Programmer Deformation After Impact (Test M4): (a) Experiment; (b) FEA

The calibrated FEA model guided the selection of programmer design for further dynamic experiments. Since the maximum acceleration of the flyer was still higher than that in typical full-scale crash tests as well as the MASH maximum acceleration requirement of 20.49 g, it was necessary to increase the thickness of the programmer material. Based on the FEA simulation results on combinations of various thicknesses and candidate materials, the researchers selected a programmer that was 9 in. thick, consisting of two 4-in.-thick, low-density urethane foams (6 lb/ft3) and a 1-in.-thick neoprene rubber pad (Figure 13).
Utilizing the test input parameters determined in the previous chapter and the methodology described in the previous section, experiments on guardrail post specimens with various test configurations were planned by using the generalized dynamic test setup shown in Figure 14.

24

FIGURE 13 Final Programmer Design Used in Dynamic Test Program
FIGURE 14 Schematic Illustration of Dynamic Test Configuration
25

3.1.3 Test instrumentation and data processing 3.1.3.1 Accelerometers
Shock accelerometers ranging up to 5000 g acceleration were installed on opposite sides of the impact surface--one on the flyer mass and the other on the post. The locations of the accelerometers are shown in Figure 15. To minimize damage to the instrumentation, accelerometer 1 (flyer mass) was mounted at a secured position on the top of the flyer and accelerometer 2 (guardrail post) was mounted on the opposite side flange. A high-speed portable data acquisition system was used to record all acceleration data at a sampling rate of 100 kHz. A detailed list of equipment used in the experimental program is given in Appendix A.
FIGURE 15 Locations of Accelerometers for Dynamic Testing
26

3.1.3.2 Acceleration data processing Acceleration data were processed using a digital low-pass filter. Specifically, the
CFC 60 filter was used in this research. This filter is one of most widely used filters in automotive engineering for processing impact signals, and it is recommended in MASH. This filtering process is necessary to remove high-frequency noises that can conceal the underlying trend in the signal. This filtering method was used in plotting accelerationtime history curves for all performed tests. Figure 16 gives an example of the raw and filtered acceleration curves.
FIGURE 16 Example of Raw and Filtered Acceleration Signals The first few dynamic tests were conducted with the accelerometer cables provided by the manufacturer with no protection/reinforcement. However, a significant amount of
27

low-frequency noise in the acceleration data was recorded due to the vibration of the accelerometer cables while the flyer was moving, which made the behavior of the flyer difficult to capture. To address this issue, more robust shielded extension cables were used and attached on the flyer with proper adhesive reinforcement, as shown in Figure 17. The flyer acceleration was successfully captured in later tests using this approach. Figure 18 shows the filtered acceleration records from two different tests (before and after the treatment) but from the same test configuration.
FIGURE 17 Protection and Reinforcement for Accelerometer Cables
28

FIGURE 18 Removal of Low-frequency Noises from Flyer Acceleration 3.1.3.3 High-speed cameras Two high-speed cameras were used for both qualitative and quantitative evaluation of the structural behavior of the post. These cameras can provide video recording at both impact height and ground level with an acceptable frame rate (1000 frames/second) and resolution (0.6 mm/pixel) during each test. Figure 19 shows the location of both cameras and their corresponding fields of view. The recorded images can be analyzed with motion tracking software to provide detailed displacement information for the flyer mass, guardrail post, and surrounding asphalt. A detailed description of the high-speed cameras used in the dynamic testing program is given in Appendix A.
29

(a)

(b)

FIGURE 19 High-speed Camera Setup and Images: (a) Ground Level; (b) Impact Level

3.1.3.4 Displacements from high-speed images High-speed images were utilized for plotting displacementtime history curves via
PCC, a software package featuring not only high-speed camera control but also motion tracking on high-speed images. Figure 20 shows an example of motion tracking based on the recorded images focused on targets located around the impact point and the flyer.

30

Displacementtime history curves for three targets marked in this example are plotted in Figure 21.
FIGURE 20 Example of Motion Tracking for Displacements
FIGURE 21 Example of Target PositionTime History Plots for Multiple Targets
31

Figure 22 shows an example of sequential photographs taken every 30 ms after the impact between the flyer and the post. A noticeable deviation occurs when the flyer starts rotating horizontally when the post has nearly reached its maximum dynamic displacement (see photographs taken at 203 and 233 ms). This resulted from the accumulation of small imperfections and asymmetry in the test setup and specimen orientation. Because of this rotational movement of the flyer observed in several tests, the lateral displacement of the flyer via motion tracking and the acceleration of the flyer from the accelerometers were less accurate/reliable after the maximum dynamic displacement of the post occurred.
FIGURE 22 Sequential Photographs for Behavior of Flyer and Post After Impact
32

3.2 Dynamic Test Results 3.2.1 Basic configuration tests
In the baseline tests with no mow strip, the guardrail post could contain the flyer mass and redirect it to a gradual stop in conjunction with lateral deflection of the post. Figure 23 shows sequential photographs of Test #13 from the high-speed camera. Figure 24 shows a representative baseline test setup (Test #13) and the overall displacement of the post and the deformation of the soil foundation. The maximum post displacement at the impact height was 23.11 in. at 109 ms after the initial impact. The flyer started rotating horizontally after maximum displacement and landed on the ground of the impact side. The post displacement at the ground level was measured at 10.83 in. and there was no observed post bending or local yielding in the post section due to ground-level restraint. The acceleration response showed that the peak flyer acceleration of 27.4 g occurred at 3 ms and the acceleration slowly decreased after the peak while the flyer and the post were in contact.
In the typical mow strip tests, the embedded guardrail post successfully redirected the flyer mass to the impact side. Figure 25 shows a representative test setup (Test #6), the overall displacement of the post, and the deformation of the mow strip. The mow strip on the rear side was completely fractured by the impact: two diagonal crack lines propagated from the post flanges and one crack line propagated along the direction of the impact. However, there was no significant bending or local yielding in the post due to ground-level restraint by the mow strip.
33

FIGURE 23 Baseline Test (Test #13): Sequential Photographs for Post Displacements
34

FIGURE 24 Baseline Test (Test #13): Overall Damage After Impact
Figure 26 shows the sequential photographs of Test #6 from the high-speed camera. The maximum post displacement at the impact height was 13.71 in. at 96 ms after the initial impact, while the post displacement at the ground level was measured at 5.12 in. The flyer was decelerated by the post and safely landed on the rails after maximum displacement. The acceleration response showed that the peak flyer acceleration of 42.5 g occurred at 19 ms and the acceleration rapidly decreased after the peak compared to the baseline results.
35

FIGURE 25 Typical Mow Strip Test (Test #6): Overall Damage After Impact Acceleration curves from representative tests in the basic configuration category are shown in Figure 27. Displacement measurements from high-speed images are plotted in Figure 28. Impact details based on the acceleration and displacement responses are summarized in Table 4.
36

FIGURE 26 Typical Mow Strip Test (Test #6): Sequential Photographs for Post Displacements
37

FIGURE 27 Basic Configuration Tests: AccelerationTime History
FIGURE 28 Basic Configuration Tests: DisplacementTime History
38

TABLE 4 Basic Configuration Tests: Impact Details

Test Configuration
Baseline Typical mow strip

Mow Strip Dimension Thick. RD (in.) (in.)
0 0
3.5 24

Peak Accel.
(g)
27.4 42.5

Time at Peak Accel. (ms)
3
19

Max. Displ. (in.)
23.11 13.71

Time at Max. Displ. (ms)
109
96

3.2.2 Tests with treatment behind post In the leave-out mow strip tests, the flyer mass was successfully decelerated by the
post and dropped on the impact side of the mow strip / leave-out zone. Figure 29 shows the sequential photographs of Test #12 from the high-speed camera. Figure 30 shows the leaveout test setup (Test #12), the overall displacement of the post, and the deformation of the grout leave-out with surrounding mow strip. The leave-out zone behind the post was completely crushed and the mow strip on the rear side was also fractured by the impact. Two diagonal cracks propagated from two rear-side corners of the leave-out and another fracture appeared in the mow strip along the direction of impact. No significant local damage in the post was observed. The maximum post displacement at the impact height was 15.91 in. at 92 ms after the initial impact, while the post displacement at the ground level was measured at 6.69 in. The flyer was decelerated by the post and redirected to the front side of the mow strip after maximum displacement. The peak flyer acceleration of 32.4 g was recorded at 21 ms, which was lower than the typical mow strip test (42.5 g) but higher than the baseline test (27.4 g).

39

FIGURE 29 Sequential Photographs for Leave-out Configuration
40

FIGURE 30 Leave-out Test (Test #12): Overall Damage After Impact
In the pre-cut configuration tests, the lateral restraint by the guardrail post was reduced by applying one of the pre-cutting patterns shown in Figure 8. The guardrail posts were able to contain the flyer mass and redirect it to the front side of the ground. Figure 31 shows the pre-cut test setup (Test #8), the overall displacement of the post, and the fracture mode of the mow strip. Figure 32 shows the sequential photographs of Test #8 from the high-speed camera. The maximum post displacement at the impact height was 22.0 in. at 123 ms after the initial impact and the ground-level displacement was measured at 9.49 in.
41

The peak flyer acceleration of 29.6 g was recorded at 21 ms. As expected, the fracture of the mow strip behind the post was guided by the pre-cuts. There was no remarkable damage in the mow strip outside the pre-cut area nor in the guardrail post near the ground level.
FIGURE 31 Pre-cut Test (Test #8): Overall Damage After Impact Acceleration curves from this test category are shown in Figure 33. Displacement measurements from high-speed images are plotted in Figure 34, and impact details based on the acceleration and displacement responses are shown in Table 5.
42

FIGURE 32 Pre-cut Test (Test #8): Sequential Photographs for Post Displacements
43

FIGURE 33 Treatment Behind Post Experiments: AccelerationTime History
FIGURE 34 Treatment Behind Post Experiments: DisplacementTime History
44

TABLE 5 Treatment behind Post Experiments: Impact Details

Test Configuration
Leave-out Pre-cut (parallel) Pre-cut (diagonal)

Mow Strip Dimension Thick. RD (in.) (in.)
3.5 24

Peak Accel.
(g)
32.4
29.6 44.2

Time at Peak Accel. (ms)
21
21 19

Max. Displ. (in.)
15.91
22.00 14.61

Time at Max. Displ. (ms)
92
123 96

3.2.3 Tests with variation in dimension In the variation-in-dimension test category, the thick mow strip test (#11) and the
reduced rear distance test (#7) are addressed as representative cases of each geometric parameter. The thick mow strip test was conducted as the most restrained among all test configurations. The guardrail post contained the flyer mass until its maximum dynamic displacement and pushed the flyer back to the rail. Figure 35 shows the sequential photographs of the thick mow strip test (#11). The maximum post displacement at the impact height was 10.45 in. at 81 ms after the initial impact, and the post displacement at the ground level was measured at 4.21 in. The acceleration response showed that the peak flyer acceleration of 53.5 g occurred at 18 ms and the acceleration rapidly decreased after the peak. Figure 36 shows the test setup (#11), the overall displacement of the post, and the deformation of the mow strip. The mow strip on the rear side was fractured by the impact and two distinct crack lines were formed from the rear side flange of the post. Due to excessive restraint by the mow strip, plastic deformation of the post near the ground level was clearly observed as shown in Figure 36 (d).

45

FIGURE 35 Thick Mow Strip Test (Test #11): Sequential Photographs for Post Displacements
46

FIGURE 36 Thick Mow Strip Test (Test #11): Overall Damage After Impact
In the reduced rear distance test, the post contained the flyer with a significant level of translation. Figure 37 shows the test setup (Test #7), the overall displacement of the post, and the mow strip fracture. Apparently, the mow strip fracture shape in the reduced RD test did not involve major crack lines, which is a common fracture shape observed from the 24-in.-wide mow strip tests. The rear side mow strip was scattered with small pieces by the impact, which may indicate a lower level of ground-level restraint due to the reduction of RD. Figure 38 shows the sequential photographs of Test #7 from the high-
47

speed camera. The maximum post displacement at the impact height was 25.81 in. at 131 ms after the impact and the ground-level displacement was measured at 11.73 in. The peak flyer acceleration of 26.6 g was recorded at 4 ms, and the acceleration slowly decreased after the peak while the flyer and the post were in contact.
FIGURE 37 Reduced RD Test (Test #7): Overall Damage After Impact
48

FIGURE 38 Reduced RD Test (Test #7): Sequential Photographs for Post Displacements Acceleration curves from this test category are shown in Figure 39. Displacement measurements from high-speed images are plotted in Figure 40 and impact details based on the acceleration and displacement responses are shown in Table 6.
49

FIGURE 39 Variation in Dimension Experiments: AccelerationTime History
FIGURE 40 Variation in Dimension Experiments: DisplacementTime History
50

TABLE 6 Variation in Dimension Experiments: Impact Details

Test Configuration

Mow Strip Dimension Thick. RD (in.) (in.)

Test Number

Thin

1.5 24 #9

Thick

5.5 24 #11

Reduced RD 3.5 12 #7

Thick and Reduced RD

5.5 12

#14

Peak Accel.
(g)
26.1 53.5 26.6
33.7

Time at Peak Accel. (ms) 23
18
4
20

Max. Displ. (in.)
21.08 10.45 25.81
17.52

Time at Max. Displ. (ms) 112
81
131
112

3.3 Performance Assessment Criteria for Dynamic Tests A comprehensive analysis related to the desirable performance of guardrail posts
was made based on the observations from prior guardrail studies and the statements in the AASHTO Roadside Design Guide. Guardrail posts in a typical guardrail system are required to rotate and translate under impact conditions. At the same time, the impacting vehicle needs to be contained by the guardrail system and decelerated in a controlled manner.
Based on this performance description, the researchers selected four dynamic assessment criteria to evaluate the influence of varying mow strip installations on the performance of the post: peak dynamic force, ground-level displacement, impact-height displacement, and effective dynamic force. Each criterion was selected to be a quantitative measure of ground-level restraint and to be informative/applicable to full-scale tests. These assessment criteria can be determined using experimental data: accelerations and displacements.
Figure 41 illustrates a sequential progression of the dynamic impact test from the release of the flyer to the maximum displacement of the post. Acceleration, velocity, and
51

displacement of each test component can be used to determine the state of the dynamic behavior and the relative stiffness of different mow strip configurations.
FIGURE 41 Sequential Progression of Dynamic Impact Test and Detailed Descriptions
52

3.3.1 Peak dynamic force

In an impact condition, a dynamic load is generally difficult to measure using

conventional load cells. Instead, the dynamic force, F (t) , can be estimated from the

product of acceleration, a(t) , and mass, m, of the flyer using Newton's second law of

motion:

F (t) = ma(t)

(3-3)

Because the mass of the flyer remains constant, higher dynamic force yields higher

acceleration of the impacting vehicle. Therefore, lower peak dynamic force is a more

desirable response when the same amount of kinetic energy/impulse is used.

3.3.2 Ground-level displacement As a direct measure of ground-level restraint, the ground-level displacement can be
estimated from the high-speed images focused on ground-level targets and their processing through motion tracking software. Since applied kinetic energy in this dynamic test program is controlled to be consistent (137.9 kip-in), a guardrail post that translates farther potentially has a greater chance of stopping an impacting vehicle in a safer fashion by dissipating its kinetic energy. Therefore, larger ground-level displacement indicates lower ground-level restraint. In this test program, residual ground-level displacements were used for comparison among the various mow strip installation methods tested.

3.3.3 Impact height displacement The impact height displacement, usually referred to as "maximum dynamic
displacement" or "dynamic crush," is a widely used criterion for initial evaluation of guardrail systems [15] or vehicles in crash tests [16]. Generally, displacements at impact

53

height have a strong correlation with the flexibility of the system. A stiffer guardrail system yields less dynamic deformation on the guardrail side but results in more deformation of the vehicle. The impact height displacement can be a simple but critical measure of the overall system response regardless of testing dimensions.

3.3.4 Effective dynamic force

An effective dynamic force, , can be defined as the average required force to displace the post up to a reference displacement, . The effective dynamic force criterion

represents a measure of the overall ground-level restraint of a post while being displaced

to the reference point by an impact. This criterion can be written as shown:



=

1

0

()

(3-4)

where () is the dynamic force acting on the impact plane. The lower effective dynamic

force, generally resulting from lower acceleration during the impact, indicates that a system

has less ground-level restraint. When a series of dynamic tests is conducted under the same

test protocol, a reference point can be designated as the maximum (or near maximum)

dynamic displacement of the most restrained test configuration--the minimum of the

maximum dynamic displacement among all tests. This allows a user to evaluate the relative

performance of all performed dynamic tests with the maximum amount of test

measurement data. In this study, the reference point of 10 in. was selected for performance

assessment, which is the closest smaller integer to the minimum of the maximum

displacement of 10.45 in. recorded in the test with the highest ground-level restraint

(Test #9: thick mow strip configuration).

54

3.4 Effect of Mow Strip Design Parameters 3.4.1 Summary of dynamic test results
The mow strip design parameters were analyzed using the dynamic performance assessment criteria identified in the previous section. Table 7 gives a summary of the results from dynamic tests performed in the present work, including: (1) peak dynamic force, (2) effective dynamic force (= 10 in.), (3) impact height displacement, and (4) groundlevel displacement. Results from tests performed using the leave-out configuration are designated as the "target performance value" for checking the relative efficiency of other tested alternative designs.

TABLE 7 Performance Assessment Criteria from Dynamic Tests

Test Configuration

Mow Strip Dimension
Thick. RD (in.) (in.)

Baseline

0

0

Typical mow strip

3.5 24

Leave-out

Pre-cut (parallel) 3.5

24

Pre-cut

(diagonal)

Thin

1.5 24

Thick

5.5 24

Reduced RD

3.5 12

Thick and reduced RD

5.5 12

Dynamic Forces (lb)

Peak
8154 12677 9650 8821 13181 7771 15931 7922 10030

Effective
5066 7269 6419 5760 7066 5624 7566 5122 5517

Displacements (in.)

Impact Height (maximum)
23.11

Ground Level (residual)
10.83

13.71

5.12

15.91

6.69

22.00

9.49

14.61

5.79

21.08 10.45 25.81

10.51 4.21 11.73

17.52

8.23

55

3.4.2 Effect of thickness A total of five test configurations--consisting of the baseline, the leave-out, and
three different mow strip thicknesses (thin: 1.5 in.; typical: 3.5 in.; and thick: 5.5 in.)-- were selected for investigating the effect of mow strip thickness on relative restraint imparted by an asphalt layer. Figure 42 shows the four assessment criteria with a strong linear trend between thickness and ground-level restraint. Under all four criteria, the target performance values from the leave-out configuration are located between the performance values of 1.5-in. and 3.5-in.-thick mow strip configurations. The thin mow strip configuration (1.5 in.) shows lower peak and effective dynamic forces than the target, along with larger displacements at both the impact point and the ground level.
Forcedisplacement curves can efficiently visualize relative ground-level restraint. The curves can be drawn from the combination of the accelerationtime and displacement time history curves. Figure 43 displays the forcedisplacement curves of the five test configurations. It can be seen that the most restrained configuration is the one with 5.5-in. thickness. The results presented here indicate that the leave-out mow strip configuration may not result in significantly less ground-level restraint than a 1.5-in.-thick mow strip configuration under all four assessment criteria. The limited data appear to indicate that a mow strip thickness of approximately 2.7 in. would exhibit roughly equivalent performance to the leave-out; this, of course, cannot be definitively asserted without more test repetitions.
56

18000 16000 14000 12000 10000
8000 6000 4000 2000
0

8154 0"

Peak dynamic force (lb)

15931

12677

7771

9650

1.5"

3.5"

5.5"

Leave-out

8000 7000 6000 5000 4000 3000 2000 1000
0

5066 0"

Effective dynamic force (lb)

5624

7269

7566

6419

1.5"

3.5"

5.5"

Leave-out

Maximum displacement at impact (in.)

25

23.11

21.08

20

15

13.71

10.45

10

5

0 0"

1.5"

3.5"

5.5"

Note: RD=24 constant

FIGURE 42 Assessment Criteria: Effect of Thickness

15.91 Leave-out

57

Residual ground-level displacement (in.)

12

10.83

10

10.51

8

6

5.12

4.21

4

2

0 0"

1.5"

3.5"

5.5"

Note: RD=24 constant

6.69 Leave-out

FIGURE 42 (continued)

FIGURE 43 ForceDisplacement Curves: Effect of Thickness
58

3.4.3 Effect of rear distance A total of six test configurations were chosen for investigating the effect of mow
strip rear distance. Two reduced RD (RD=12 in.) tests are paired with the typical RD tests (RD=24 in.) for two thicknesses (typical: 3.5 in.; thick: 5.5 in.). The baseline and leave-out test configurations are included as the reference values. The effect of rear distance is also very significant in terms of its influence on ground-level restraint. Figure 44 shows four assessment criteria and a strong linear correlation between rear distance and ground-level restraint for the two mow strip thicknesses. The rear distance reduction to 12 in. improves performance under all four criteria: (1) peak dynamic force decreased by 37%, (2) effective dynamic force decreased by 28%, (3) maximum displacement at impact increased by 44%, and (4) residual ground-level displacement increased by 53%. Specifically, the 5.5-in.thick and reduced RD configuration shows lower peak and effective dynamic forces than the target (leave-out) but larger displacements at both the impact point and the ground level. Dynamic forcedisplacement curves shown in Figure 45 also exhibit this correlation. The results presented here indicate that the leave-out mow strip configuration may not result in significantly less ground-level restraint than a 12-in.-wide mow strip configuration under all four assessment criteria. It is remarkable that the pre-installed guardrail posts driven through asphalt mow strip could have significant less ground-level restraint by reducing their rear distance.
59

Peak dynamic force (lb)

18000 16000 14000 12000 10000
8000 6000 4000 2000
0

15931

10030

12677

7922

9650

8154

24" (tk=5.5") 12" (tk=5.5") 24" (tk=3.5") 12" (tk=3.5") 24" (Leave- 0" (baseline) out)

Effective dynamic force (lb)

8000 7000 6000 5000

7566

5517

7269

5122

6419

5066

4000

3000

2000

1000

0

24" (tk=5.5") 12" (tk=5.5") 24" (tk=3.5") 12" (tk=3.5") 24" (Leave- 0" (baseline) out)

Maximum displacement at impact (in.)

30 25.81

25

23.11

20
15 10.45
10

17.52

13.71

15.91

5

0 24" (tk=5.5") 12" (tk=5.5") 24" (tk=3.5") 12" (tk=3.5") 24" (Leave-out) 0" (baseline)

FIGURE 44 Assessment Criteria: Effect of Rear Distance

60

Residual ground-level displacement (in.)

14

11.73

12

10.83

10

8.23

8

6.69

6

4.21

5.12

4

2

0 24" (tk=5.5") 12" (tk=5.5") 24" (tk=3.5") 12" (tk=3.5") 24" (Leave-out) 0" (baseline)

FIGURE 44 (continued)

FIGURE 45 ForceDisplacement Curves: Effect of Rear Distance
61

3.4.4 Effect of pre-cutting Five test configurations were selected for investigating the effect of pre-cutting.
The mow strip corresponding to GDOT S-4-2002 (thickness = 3.5 in.; RD=24 in.), the leave-out configuration, and two pre-cut configurations were compared under the performance assessment criteria to evaluate if pre-cutting can be an effective solution for existing mow strips with excessive ground-level restraint. Figure 46 shows four assessment criteria, and Figure 47 shows the dynamic forcedisplacement curves. The parallel pre-cut application significantly improves the performance under all four criteria: (1) peak dynamic force decreased by 30%, (2) effective dynamic force decreased by 21%, (3) maximum displacement at impact increased by 38%, and (4) residual ground-level displacement increased by 46%. The diagonal pre-cut application also improves the performance but not as effectively as the parallel pre-cut: (1) peak dynamic force increased by 4%, (2) effective dynamic force decreased by 3%, (3) maximum displacement at impact increased by 6%, and (4) residual ground-level displacement increased by 12%. As mentioned previously, these results, of course, cannot be definitively asserted without more test repetitions and more test condition variables (e.g., direction of dynamic load).
62

14000 12000 10000
8000 6000 4000 2000
0

Peak dynamic force (lb)

13181

12677

8821

9650

Pre-cut (P) Pre-cut (D) Typical Leave-out

8154 Baseline

Effective dynamic force (lb)

8000 7000 6000 5000 4000 3000 2000 1000
0

5760

7066

Pre-cut (P) Pre-cut (D)

7269 Typical

6419 Leave-out

5066 Baseline

Maximum displacement at impact (in.)

25

22

23.11

20

15

14.61

13.71

15.91

10

5

0 Pre-cut (P) Pre-cut (D)

Typical

Leave-out Baseline

Note: Mow strip thickness 3.5 / rear distance 24 (constant)

FIGURE 46 Assessment Criteria: Effect of Pre-cut

63

Residual ground-level displacement (in.)

12

10

9.49

10.83

8

6.69

6

5.79

5.12

4

2

0 Pre-cut (P) Pre-cut (D)

Typical

Leave-out Baseline

Note: Mow strip thickness 3.5 / rear distance 24 (constant)

FIGURE 46 (continued)

FIGURE 47 ForceDisplacement Curves: Effect of Pre-cut
64

3.4.5 Performance ranking for various mow strip designs Each dynamic performance criterion represents a quantitative measure of ground-
level restraint, and it is possible to rank the various mow strip designs under these criteria. Because all dynamic tests were conducted with more controlled soil and asphalt conditions compared to those in the static tests, a performance ranking based on the dynamic test results could be informative for planning full-scale crash tests. Table 8 shows the individual and overall performance ranking of various mow strip designs. The reduced rear-distance configuration, followed by the thin, the parallel pre-cut, and the thick and reduced reardistance configurations were determined to have less ground-level restraint than the leaveout configuration. However, as mentioned earlier, these results cannot be definitively asserted without more test repetitions.

TABLE 8 Performance Ranking for Various Mow Strip Designs

Test Configuration Baseline

Peak Dynamic
Force 3

Typical mow strip

7

Leave-out

5

Pre-cut (parallel)

4

Pre-cut (diagonal)

8

Thin

1

Thick

9

Reduced RD

2

Thick and reduced RD

6

Note: 1=lowest ground-level restraint 9=highest ground-level restraint

Individual Ranking

Effective Max. Displ.

Dynamic at Impact

Force

Height

1

2

8

8

6

6

5

3

7

7

4

4

9

9

2

1

Residual Groundlevel Displ.
2 8 6 4 7 3 9 1

3

5

5

Overall Ranking
2 8 6 4 7 3 9 1
5

65

DYNAMIC SUBCOMPONENT FINITE ELEMENT SIMULATIONS
A series of finite element models was constructed to perform mock-up dynamic simulations and subcomponent dynamic simulations in parallel with the subcomponent dynamic testing program described in the previous chapter. The simulations were initially performed to determine the appropriate material properties and dimensions for the foam programmer used in the impactor for the subcomponent testing. This model was then used to determine various parameters needed in the test setup (i.e., impact mass, impact speed, instrumentation fixtures, etc.). Thereafter, dynamic subcomponent tests and simulations were conducted to assess the performance of guardrail posts embedded in soil and asphalt. These tests are less expensive than full-scale crash tests and are used to evaluate additional alternative designs for asphalt mow strips. The material models used for the soil, asphalt, and steel in the static tests in Phase I of the research project were updated for use in the dynamic loading simulations.
4.1 Design of the Impactor The impact loading was applied to the guardrail posts with a steel impactor that had
a specified initial velocity. A dynamic actuator was used to provide the initial velocity to the steel mass. A critical step in the development of the test program was to determine the appropriate impact speed and mass of the impactor to use. Based on the testing conditions, the safe maximum impact speed was limited to 50 ft/s. Using this speed, the mass of the impactor could be specified for a given kinetic energy. The MASH guidelines [11] specify the mass and speed of vehicles used for full-scale crash testing. MASH Tests 3-11 and 3-10
67

were used to evaluate the guardrail system discussed in this research. Vehicles with an

impact angle equal to 25 degrees and with masses of 5000 lb and 2425 lb are used for

Tests 3-11 and 3-10, respectively. The kinetic energy for Test 3-11 is used here as a

reference in the subcomponent dynamic testing. It was assumed that the energy dissipated

by guardrail posts and associated with the component of velocity perpendicular to

guardrails is divided equally between 10 posts. Therefore, the energy dissipated by one

guardrail post can be obtained as:



=

0.5 v

252 10

=

0.5



5000





91 252 10

=

11.5

kip. -ft

(4-1)

The mass required for a steel impactor to reach the above energy with a velocity of 50 ft/s

was approximately 300 lb. This mass was used in the experimental subcomponent tests. In

the initial mock-up tests, a velocity of 33 ft/s was used instead of 50 ft/s in order to increase

the safety of the tests. Also, 4-in.-thick foam was used instead of 8-in.-thick foam because

the kinetic energy was less. However, the velocity of 50 ft/s and foam thickness of 8 in.

were used in the dynamic subcomponent tests with the guardrail post embedded in soil and

asphalt.

When the steel impactor directly impacted the steel guardrail posts, the resultant

acceleration due to the impact and based on the FEA simulations was more than 1000 g.

Vehicle ride-down accelerations in acceptable full-scale crash tests on guardrails are

typically less than 20 g based on the MASH criteria. Therefore, a crushable foam layer was

attached to the front of the impactor to reduce impact accelerations in the subcomponent

tests. An 8-in. crushable foam with a density of 6 lb/ft3 was chosen for use in the tests. A

neoprene rubber layer was placed between the crushable foam layer and the steel mass to

protect the mass when the crushable foam was destroyed in an impact. An illustration of

68

the steel impactor with crushable foam and rubber layers is shown in Figure 48. The steel impactor has four wings that allow it to sit and slide on guidance rails.
FIGURE 48 Steel Impactor, Crushable Foam, and Rubber 4.2 Mock-up Test Simulation A simple dynamic mock-up test was constructed by attaching a 5.9-ft-long guardrail post to a steel beam as discussed in the previous chapter and shown in Figure 10. A finite element model of this mock-up test was developed, and the results were compared by measuring the deformation of the foam layer and the displacement of the post at the point of impact. After verifying the finite element model of the steel post, foam, and rubber layer, the model for the impactor and the steel post was used in the subcomponent testing of guardrail posts embedded in soil and asphalt. The mock-up FE model is presented in Figure 49.
69

FIGURE 49 Representation of the Mock-up Setup
4.3 Dynamic Mock-up Simulations 4.3.1 Finite element model description
LS-DYNA V971 R9.1.0 [15] was used in this research. The quasi-static problem was solved with an explicit algorithm, instead of implicit. The model developed in this chapter was used for full-scale crash simulations presented in Chapter 6. 4.3.1.1 Geometry
The length of the post used was 6 ft. The impact point was 5.3 ft from the bottom of the post as shown in Figure 49. The steel mass length, width, and depth were 13.4 in., 10 in., and 6 in., respectively. The width of the mass wings was 36 in. The foam and rubber layer thicknesses were 4 in. and 1 in., respectively.
70

4.3.1.2 Simulation of initial velocity and impact The velocity was modeled by assigning an initial velocity equal to 3.3 ft/s to all
nodes of the impactor. The impactor flew for 0.1 in. and then impacted the post. The impactor stayed in contact with the post until its horizontal speed reached zero, and then the impactor bounced back; after a fraction of a second, the impactor lost contact with the post and flew freely away from the setup. 4.3.1.3 Mesh sensitivity analysis
The finite element mesh density for the guardrail post, foam, and rubber varied from 1 in. to 0.125 in. The post and foam deflection were monitored during the analysis. The results showed that using a mesh size of 0.125 in. gave results with less than 5% difference compared to using a mesh size of 0.25 in. Therefore, 0.125 in. was used as the mesh size for these parts. Further refining the mesh size was not feasible, as reducing the size of elements by a factor of 2.0 increased the simulation time by more than 100%. 4.3.1.4 Hourglass controls and energy checks
Because fully integrated shell and solid elements were used in the model, the hourglass energy was zero, and no hourglass control was needed. 4.3.1.5 Steel
A piecewise linear metal plasticity model was used in LS-DYNA for the steel post. The static yield strength of the steel, modulus of elasticity, and Poisson's ratio were given as inputs. Common steel parameters, presented in Table 9, were employed in this model. Strain rate effects were accounted for by using a set of stressstrain curves associated with different strain rates [16]. These sets are shown in Figure 50 and Figure 51. The curves were extended to very high strains (up to 1.0) to account for large local strains that occur
71

in the post during simulations. Shell element formulation number 16 was selected for the steel elements; this element did not exhibit hourglass modes.

TABLE 9
Material Constants Used in the Dynamic Finite Element Subcomponent and Full-scale Crash Simulations

Material

Constitutive Parameter

Value

Determined from

Steel

Density, Young modulus, E Poisson's ratio, Yield stress for flanges, yf Yield stress for the web, yw
Density,

495 lb/ft3 29000 ksi
0.3 50 ksi 58 ksi 144 lb/ft3

Material test [4] [4]
Material test Material test Material test

Cohesion, C

0.145 psi

Material test and via system test calibrationa

Peak friction angle, p Soil
Critical friction angle, cr

45o

Material test and via system test calibrationa

15o

[17] and via system test calibrationa

Shear modulus, G

7.25 ksi

[18] and via system test calibrationa

Poisson's ratio, Density, Cohesion, C

0.25 144 lb/ft3 72.5 psi

[19] Material test Material test

Asphalt

Friction angle, Shear modulus, G

35 7.25 ksi

[20]
Via system test calibrationa

Rubber Foam

Poisson's ratio,
Failure volumetric strain, v-failure
Density, Shear modulus, G
Density,
Shear modulus, G

0.35
%2 in./in. 85 lb/ft3 58 ksi 6 lb/ft3
14.5 ksi

[21] Via system test
calibrationa
Material test [22]
Material test
Material test and via system test calibrationa

aThe term "system test calibration" refers to the selection of particular material constants based on one selected system test as described above.

72

4.3.1.6 Rubber The rubber layer used in the impactor was modeled using a simple Blatz-Ko rubber
material with a shear modulus of 58 ksi [22]. The density of the rubber was estimated, using experimental results, as 85 lb/ft3.
FIGURE 50 StressStrain Curves Used for the Post Flanges (Strain rates are given for each curve.)
FIGURE 51 StressStrain Curves Used for the Post Web (Strain rates are given for each curve.)
73

4.3.1.7 Foam The foam layer was modeled using a crushable foam material in LS-DYNA
(material #63). The density of the foam was estimated, using experimental results, as 6 lb/ft3. The stressstrain curve for compression loading of the foam under static loading was obtained experimentally. This curve is presented in Figure 52.
FIGURE 52 StressStrain Curve for the Foam Material After examination of the specimen during the compression test, no deformation was observed in the transverse direction, which implies that the Poisson's ratio of the foam material is zero. The initial and final cross-section areas of the crushable foam in compression remain constant. Therefore, the engineering and true stressstrain diagrams are identical. The volume of the material is not conserved during compression and the density increases while the material is compressed. The volumetric strain is equal to the
74

compressive strain. The crushable foam material model requires nominal yield stress versus the volumetric strain curve as an input. This curve was obtained using Figure 52 and is presented in Figure 53. This curve is based on static loading and was used initially in the model. Thereafter, in the calibration process of the model, a common dynamic rate factor equal to 2 was used to scale the stress values at a given strain on the curve. This scale factor resulted in the same foam deformation in the FE results as observed in the experiments. The results of the experiments and the simulations are compared in Table 10.
In the experiments, it was observed that the foam cracks and crushes during the impact. It is also known that foams are weak in tension. Therefore, an LS-DYNA element erosion criterion based on the maximum principal strain was added to the foam model in order to capture the cracking and crushing of the foam. The elements were eroded when the principal strain reached a specified value. The value of the erosion principal strain equal to 0.1 in./in. was found using model calibration by comparing foam deformation in the experiment and the finite element model at various times during the impact. This value is mesh-dependent, and lower values of maximum principal strain are needed for larger mesh sizes. This occurs because the average strain is smaller for a larger mesh than for a smaller mesh size.
75

FIGURE 53 Nominal Yield Stress vs. Volumetric Strain for the Foam Material

TABLE 10 Comparison between FEA and Experimental Results for the Mock-up Tests

FEA

FEA

Experiment

Foam Deformation

1.46 in. 1.38 in.

Post Disp. at Impact Point 5.71 in. 5.98 in.

Impactor Max. Acceleration 65 g

52 g and 71 ga

a Results from two separate accelerometers installed on the mass.

% Difference 6% 5% 25% and 8% a

4.3.2 Comparison between experiments and FE simulations The deflection of the post at the impact point and the deformation of the foam were
measured in the simulations and the experiments. The results are compared in Table 10. The accelerations were higher than common full-scale crash test accelerations, which are typically less than 20 g; these higher accelerations are due to the lighter mass used in this

76

research, compared with the mass of vehicles. However, the accelerations, velocities, and strain rates in the mock-up tests are still of the same order of magnitude as those in the fullscale crash tests. This indicates that the material response will not be considerably different between these tests and full-scale crash tests and these tests can be used to assess different components of the guardrail systems separately.
The plastic deformation of the post at the connection and the overall shape of the post after impact are shown in Figure 54 and Figure 55 for both the FEA and experimental results. These figures show that the FE model can accurately capture the plastic deformation of the post due to the impact. The crushed foam after impact is shown in Figure 56 for both the FEA and experimental results.
FIGURE 54 Plastic Deformation of the Post at the Connection: (a) FEA; (b) Experiment
77

FIGURE 55 Representation of the System After Impact: (a) FEA; (b) Experiment
FIGURE 56 Crushed Foam After the Impact: (a) FEA; (b) Experiment
78

4.4 Dynamic Subcomponent Simulations 4.4.1 Finite element model description 4.4.1.1 Model domain and geometry
The soil was installed in a container with depth, width, and length of 62 in., 65 in., and 71 in., respectively. The container was filled with soil, which was compacted; then, asphalt layers with various thicknesses were installed and compacted over the soil. A guardrail post was driven through the asphalt and soil to the depth of 40 in., positioned 48 in. from the back of the container and 36 in. from the front of the container. The impactor was located on two parallel rails at an elevation of 25 in. from ground level. The dynamic actuator pushed the impactor until it reached the speed of 50 ft/s and then the impactor hit the guardrail post.
The finite element model had the same geometry as the test setup, as shown in Figure 57. The steel mass length, width, and depth were 13.4 in., 9.8 in., and 5.9 in., respectively. The width of the mass wings was 35.4 in. The foam and rubber layer thicknesses were 8 in. and 1 in., respectively. 4.4.1.2 Element formulations and mesh sensitivity analysis
The same mesh sizes used in the mock-up simulations were repeated in this set of simulations. Constant stress solid elements (element formulation #1 in LS-DYNA) were used to model the soil and the steel part of the impactor. The fully integrated formulation intended for solid elements with a poor aspect ratio (element formulation #-2 in LS-DYNA) was used for the foam and rubber layer. These types of elements show better stability in simulations with highly distorted foam elements with the possibility of high hourglass energy with constant stress elements. One-point integration tetrahedron elements
79

(element formulation #10 in LS-DYNA) were used for the asphalt, which captures the asphalt rupture better than the element erosion formulation.
FIGURE 57 Finite Element Model Used for Dynamic Subcomponent Simulations 4.4.1.3 Hourglass controls and energy checks The foam, rubber, steel post, and steel container were modeled using full integration elements. Therefore, there is no hourglass energy for these parts. Hourglass energy for the soil part and the steel part of the impactor was estimated as less than 1% of the internal energy, which is acceptable. 4.4.1.4 Steel, foam, and rubber The same material properties used in the mock-up simulations were used for the steel, foam, and rubber.
80

4.4.1.5 Soil The soil type that was used in the road base has never been tested experimentally
to investigate strain rate effects, and no data are available to adjust soil material properties for higher strain rates. Therefore, strain rate effects on the soil material were ignored, and the same soil material properties that were used in the static simulations were used in the dynamic simulations as initial inputs; then, the model was calibrated using experimental test results. The ground-level displacement of the post in the FEA simulation was compared to the experimental results for the dynamic tests with only soil and without an asphalt layer. The ground-level displacement for the FEA simulation was considerably lower than the experimental tests, which showed that the soil material properties in the simulations resulted in a stiffer response than the soil properties in the experiments. Therefore, the shear modulus was lowered to 3.6 ksi from 7.3 ksi, which was used in the static tests, and the cohesion value was set equal to zero. This difference occurred because the soil that was used in the dynamic tests was not exactly the same soil that was used in the static tests. Moreover, the dynamic tests were conducted in the laboratory with a controlled environment in contrast with the static tests that were performed in an open area exposed to environmental factors. 4.4.1.6 Asphalt
Strain rate effects were initially ignored for the asphalt. The same material properties that were used as initial inputs in the static simulations were employed in the dynamic simulations; the model was then calibrated using experimental results. Material constants used in the dynamic finite element simulations are provided in Table 9.
81

After examining the dynamic simulation results, it was noted that the elements behind the post were eroding, which created a gap between the post and the asphalt layer until the post moved more and filled the gap. In the dynamic simulations, the asphalt elements behind the post and on the surface did not have overburden pressure. This caused the elements to expand upward under the sudden impact load. This expansion caused large positive principal strains although the element was compressed in the other directions. When the principal strain reached a high value based on the erosion criterion used in the static tests, the element was eroded and a gap occurred behind the post. To avoid this problem, the erosion criterion was changed in the dynamic simulations. A volumetric strain criterion was used to account for the average of principal strains in all directions.
Using this criterion did not allow the elements behind the post to be eroded suddenly and made the model more stable under impact loads. Using erosion criterion of volumetric strain equal to 2% gave comparable results between FEA simulations and experimental results. Although the model was not very sensitive to this parameter, it was obtained based on the mesh size used in this model and is element-formulation and mesh-size dependent. Smaller meshes or higher order elements captured higher peaks in strains and stresses, and a higher value for erosion criterion was needed. Larger elements, with one point integration, resulted in an average value of strain and stress throughout the element. For these elements, a lower value for the erosion criterion was required. However, slight changes in the element size (up to two times smaller or larger) or mesh did not affect the erosion criterion, and the same value for the erosion criterion could be used. The other material properties, including cohesion, friction angle, shear modulus, and Poisson's ratio,
82

remained unchanged after model calibration, and the same values used in the static simulations were employed in the dynamic simulations.
4.4.2 Comparisons between experiments and FE simulations A comparison between the results from the experiments and the FE simulations is
given in Figure 58 and Figure 59. As can be seen in these figures, the finite element model demonstrates similar behavior to the experimental results. The model effectively captured the foam deformation, movement of the post in the soil, and soil plowing behind the post. Element erosion for the soil material in tension was not included in the model. Including element erosion for the soil caused erosion of the elements behind the post due to soil plowing. When the elements were eroded from behind the post, the post could move freely, and the applied force to the post suddenly dropped, which caused instability in the model and did not match the behavior of the posts in the experiments. Therefore, element erosion was not included for the soil material, and the model was not able to predict the gap in front of the post that occurred during experimental tests, as shown in Figure 59. Although the elements in front of the post experienced large tensile strains, because the cohesion value was given equal to zero, the tensile stresses in the elements in front of the post were approximately zero. Therefore, although there was no gap in the soil in front of the post, because tensile stresses were approximately zero, the tension force applied to the post by the soil was approximately zero, which is similar to a gap in the soil material.
The acceleration of the impactor mass was recorded using an accelerometer installed on the center top of the mass, as shown in Figure 60. Also, the accelerations were obtained using the FEA results. The accelerationtime history comparison is presented in Figure 61. The FEA model can predict the peak acceleration and late response of the
83

system. However, in the experiments, the acceleration had a sharp peak versus a flat peak in the FEA. This difference occurred because in the experiments the foam crushed into separate pieces and its strength decreased more rapidly than in the FEA. However, the peak acceleration was captured properly and the response of the system, after the foam was crushed, was similar in the FEA and the experiments.
FIGURE 58 Simulation Result for the Model with Only Soil After the Impact
84

FIGURE 59 Experimental Result for the Model with Only Soil After the Impact
FIGURE 60 Location of the Accelerometers Installed on the Impactor
85

FIGURE 61 AccelerationTime History Obtained from Experiments and FEA
The asphalt thickness and rear distance were varied in the FEA simulations and experiments. In one of the cases, an asphalt thickness of 3.5 in. with a rear distance of 24 in. was used to examine the mow strip setup configuration defined in GDOT S-4-2002. The after-impact result for this case is presented in Figure 62 and Figure 63 for the FEA and the experiments, respectively.
The acceleration of the impactor mass was recorded using an accelerometer installed on the center top of the mass, as shown in Figure 60. The accelerations were also obtained using the FEA results. The accelerationtime history comparison is presented in Figure 64. The FEA model predicted a peak acceleration close to the one measured in the experiments. Moreover, the shape of the accelerationtime history is similar for both cases before and after the peak force.
86

FIGURE 62 Simulation Result for the Model with 3.5-in.-thick Asphalt
and Rear Distance of 24 in.
FIGURE 63 Experimental Result for the Case with 3.5-in.-thick Asphalt
and Rear Distance of 24 in. 87

FIGURE 64 AccelerationTime History Obtained from Experiments and the FEA
88

DEVELOPMENT OF EXPERIMENTAL AND NUMERICAL PERFORMANCE CRITERIA
5.1 System Assessment Using Quantitative Performance Criteria 5.1.1 Parametric studies on asphalt mow strip geometry
As discussed in the previous chapters, changing mow strip geometry influences the behavior of guardrail posts embedded in asphalt layers. As the mow strip thickness and rear distance behind the post increase, the ground-level restraint of the asphalt layer on the post increases. Parametric studies on the combination of different thicknesses and rear distances are a critical step to explore the impact of each of these parameters. The asphalt thickness used for a mow strip conforming to GDOT S-4-2002 is 3.5 in., and the minimum feasible asphalt thickness based on constructability is estimated to be 2 in. To show the system response for thicker mow strips, 6-in.-thick asphalt was included in the simulations. Rear distance values of 0 in., 12 in., 24 in., and 48 in. were used. The time history of the impactor acceleration, post ground-level displacement, post displacement at the impact point, and post velocity at ground level were measured experimentally and with FE simulations. The results are compared in Table 11. The effective force parameter shown in Table 11 is obtained by dividing the kinetic energy of the impactor (11.5 kip-ft) by the peak displacement at the impact point. This parameter is representative of the effective force applied to the post during the impact.
89

TABLE 11 Comparison between Experimental Results and FE Simulations

90

Test Description

Mow Strip Size (in.)

Peak Disp. At Impact Point
(in.)

Ground-level Disp. (in.)

Peak Ground-
level Velocity
(ft/s)

Peak Force Effective Force

(kip)

(kip)

Thick.

Rear Dis.

FEA

Exp.

FEA

Exp.

FEA

FEA Exp. FEA Exp.

Thick Extended Rear Distance 6

48 7.0



0.8



11.5 17.8 19.6

Thick Regular Rear Distance

6

24 10.2 10.6 2.4 4.3 14.9 15.3 16.0 13.3 12.8

Thick Reduced Rear Distance

6

12 17.5 17.8 7.4 8.4 15.9 12.4 10.1 7.9 7.7

Thick Zero Rear Distance

6

0 35.0

18



19.5 7.2 3.8

GDOT Extended Rear Distance 3.5 48

8



1.9



12.3 16.7 17.1

GDOT

3.5 24 12.7 13.9 6.2 5.2 15.0 14.6 13.7 10.8 9.9

GDOT Reduced Rear Distance 3.5 12 17.9 26.2 9.7 11.9 16.4 11.0 6.5 7.7 5.2

GDOT Zero Rear Distance

3.5

0 30.3

16



18.9 8.3 4.5

Thin Extended Rear Distance

2

48 14.1

7.1



14.6 14.0 9.7

Thin Regular Rear Distance

2

24 18.2 21.4 10 10.7 14.9 11.9 7.9 7.4 6.3

Thin Reduced Rear Distance

2

12 20.7

11



17.6 10.6 6.5

Thin Zero Rear Distance

2

0 26.3 14.0

18.9 9.5 5.2

Only Soil

0

0

24 23.5 13.1 11 17.1 9.2 4.5 5.6 5.9

GDOT with Leave-out

3.5 24

16.2 -

6.8



9.7 8.6

GDOT with Parallel Pre-Cut

3.5 24 23.0 22.4 12.0 9.6 17.8 11.7 8.8 5.9 6.1

GDOT with Diagonal Pre-Cut

3.5 24 18.6 14.8 9.5 5.9 16.2 13.1 13.3 7.4 9.2

GDOT with Stiffer Asphalt

3.5 24 6.8



1.4



9.4 18.0 20.0

Experimental tests or finite element simulations were not performed for all setups.

A set of quantitative performance criteria can be used to evaluate alternative mow strip designs. Ground-level displacement, displacement at the impact point, peak force applied to the post, and the effective applied force to the post were chosen to evaluate guardrail post performance. These criteria can be used to compare setups with various mow strip thicknesses and rear distances with setups incorporating a leave-out. Setups that show more restraint than the target setup with a leave-out are considered less likely to pass fullscale crash tests. Setups that demonstrate comparable or less restraint than the target setup with a leave-out are considered more likely to pass the crash tests.
5.1.2 Assessment using peak force criterion The first criterion considered is peak force. This criterion is based on the fact that
higher peak force applied to the post before its failure shows higher ground restraint of the post. Figure 65 shows an FE simulation contour plot of peak force applied to the post. Experimental data are marked on the plot with parentheses. Mow strip setups with the thicknesses and rear distances associated with the dashed lines have less ground-level restraint than setups with leave-out. The dotted lines have more restraint than the leave-out setup, which is shown with a solid line. For example, as shown in Figure 65, guardrail posts installed in mow strips with 2-in. thickness and 12-in. rear distance would result in less restraint than the target leave-out configuration.
5.1.3 Assessment using peak displacement criterion Figure 66 shows an FE simulation contour plot of the peak displacement of the post
at the impact point. Larger displacement is an indication of less restraint on the posts. Experimental data are marked on the plot with parentheses. For example, as shown in
91

Figure 66, guardrail posts installed in mow strips with 2-in. thickness and 24-in. rear distance would result in less restraint than the target leave-out configuration.
FIGURE 65 Peak Applied Force FEA Contour Plots for Combinations
of Thickness and Rear Distance
92

FIGURE 66 Peak Displacement FEA Contour Plots for Combinations
of Thickness and Rear Distance 5.1.4 Assessment using ground-level displacement criterion
Figure 67 shows an FE simulation contour plot of residual ground-level displacement of the post after the impact. Larger ground-level displacement is an indication of less ground restraint of guardrail posts. Experimental data are marked on the plot with parentheses. For example, as shown in Figure 67, guardrail posts installed in mow strips with 2-in. thickness and 18-in. rear distance would result in less restraint than the target leave-out configuration.
93

FIGURE 67 Ground Displacement FEA Contour Plots for Combinations
of Thickness and Rear Distance 5.1.5 Assessment using effective force criterion
Figure 68 shows an FE simulation contour plot of effective force, which is obtained by dividing the kinetic energy of the impactor (11.5 kip. -ft) by the displacement of the impact point. Larger effective applied force is an indication of higher ground restraint. Experimental data are marked on the plot with parentheses. For example, as shown in Figure 68, guardrail posts installed in mow strips with 2-in. thickness and 12-in. rear distance would result in less restraint than the target leave-out configuration.
94

FIGURE 68 Effective Force FEA Contour Plots for Combinations
of Thickness and Rear Distance
5.1.6 Combined assessment using the four criteria The four contour plots are combined in Figure 69. Because the peak displacement
criterion and the effective force criterion have the same shape, they are shown with only one curve. Any design below all criteria lines is expected to show less ground restraint than the typical leave-out design. Additionally, a setup that is above the criteria lines is expected to exhibit a higher level of ground restraint than setups with a leave-out. These contours are used to better understand the behavior of guardrail post setups with mow strip and their relative performance. However, these curves cannot be used solely to reject or approve a setup to be used for road safety. The dynamic testing results give insight into the anticipated behavior in full-scale crash simulations of various mow strip designs, which are discussed in the next chapter.
95

FIGURE 69 Combined FEA Contour Plots with Equivalent Restraint to the Leave-out Setup
5.1.7 Effect of asphalt pre-cutting Based on the experimental results, rupture is the primary mechanism of the asphalt
failure around the guardrail post. As the rupture propagates, the strength of the asphalt layer decreases up to the point that a portion of the asphalt detaches from the rest of the mow strip. After this occurs, the asphalt has a negligible impact on the system and the soil is the only source of ground restraint. Therefore, one potentially effective way to decrease mow strip restraint is to add pre-cuts to the asphalt layer. Based on the results discussed in Phase I of the research project, the two most effective cuts were selected for the dynamic tests: diagonal (Figure 70) and parallel (Figure 71) cuts. The cuts were selected based on the experimental and numerical investigations of various rupture patterns. These designs shorten the distance that the asphalt rupture needs to propagate until one part detaches from
96

the rest of the layer. These two cut patterns were tested experimentally and using numerical simulations. The results are presented in Table 11. Comparing the values of various parameters for the asphalt layer with pre-cuts and for the leave-out setup shows that the peak force is higher for setups with pre-cuts because the peak force occurs in early system response, which is more affected by the asphalt layer. Pre-cut setups provide less restraint based on the comparison of ground-level displacement, displacement at the impact point, and the effective applied force. Considering these three criteria, guardrail systems utilizing these pre-cuts show less ground restraint than the leave-out. However, based on the peak force criterion, they show higher ground-level stiffness for the early system response.
FIGURE 70 Diagonal Asphalt Pre-cut Used in the Dynamic Tests and Simulations
97

FIGURE 71 Parallel Asphalt Pre-cut Used in the Dynamic Tests and Simulations 5.1.8 Effect of asphalt material properties In some cases, the asphalt layer could be extremely cold, or an asphalt mix other than the one classified as PG 76-22 binder and 0.75-in. aggregate size could be used, leading to stiffer and stronger asphalt than the asphalt used in this research, which had 73-psi cohesion and 35-degree friction angle as determined via material testing. Therefore, the GDOT S-4-2002 setup with 3.5-in. thickness and 24-in. rear distance was simulated with a stiffer and stronger asphalt layer to show how the system would respond when the asphalt layer could potentially be stiffer than normal. The value of cohesion, shear modulus, and the volumetric failure strain were increased to 135 psi, 13.5 ksi, and 0.04 in./in. based on the specimen tests done on the stiffer asphalt. The simulation result is presented in Table 11 and Figure 72.
98

FIGURE 72 Simulation Result for 3.5-in.-thick Asphalt with 24-in. Rear Distance
Using a Stiffer Asphalt Only a limited asphalt rupture occurred, and the results show that this change in the material properties makes the asphalt layer significantly stronger than the asphalt used in this research. This decreases the likelihood that a guardrail system with these asphalt properties will pass a full-scale crash test.
99

ASSESSMENT USING MASH FULL-SCALE CRASH SIMULATIONS
6.1 Finite Element Model Description The vehicle model used for the simulation of full-scale crash tests was the 2007
Chevrolet Silverado, which was developed and verified by the National Crash Analysis Center (NCAC) [23]. This vehicle model represents the MASH 2270P test vehicle and has approximately one million elements. A validated model for the guardrail system was also obtained from NCAC. The model was updated for this research by increasing the rail height to match the Midwest Guardrail System (MGS) W-beam guardrail system with a 31-in. rail height. The guardrail splices were originally located at the posts. The splices were moved between the guardrail posts in the model to match current MGS designs. An asphalt layer was added to the model, and the material properties of the soil and steel were updated. The hourglass formulation, element size, and material properties used in the previous chapter for modeling of one guardrail post system were used here for the full-scale crash model. The model has 29 posts. Initial vehicle impact was to occur 13 ft 3.5 in. upstream of the middle post in the model (post #15), as shown in Figure 73, which was selected using the critical impact point (CIP) plots given in the MASH guidelines. The complete model has approximately 1.3 million elements.
101

FIGURE 73 FE Model with 29 Posts
First, the model was validated for the case without mow strips by comparing the simulation results with available experimental test results [24]. Thereafter, the validated model was used to simulate guardrail systems with various mow strip designs (Table 12) to assess the effect of mow strip geometric and material properties on the system performance.

TABLE 12 Full-scale Crash Simulations

Test Number
Soil Only T2-R24 T3.5-R12 R3.5-R24 T6-R24 T3.5-R24-C T3.5-R24-S

Mow Strip Thickness
(in.)
No Mow Strip 2 3.5 3.5 6 3.5 3.5

Mow Strip Rear Distance (in.)
No Mow Strip 24 12 24 24 24 24

Mow Strip Modification
No Mow Strip None None None None
Diagonal Pre-cuts Stiffer Asphalt

102

6.2 Evaluation Based on MASH Guidelines The simulations presented here are based on MASH Test 3-11 with a 2270P vehicle
on an MGS 31-in. guardrail system embedded in asphalt mow strips. In this test, a 5000-lb vehicle (a 2007 Chevy Silverado in the present work) impacts the guardrails with an angle of 25 degrees and a speed of 62 mph. Several parameters must be obtained from full-scale crash simulations and tests based on the MASH guidelines. These parameters are used in this research and are briefly explained in the following.
Based on the MASH guidelines, occupant risk is assessed by the response of a hypothetical, unrestrained front-seat occupant whose motion relative to the occupant compartment is dependent on vehicle accelerations. The occupant, as an assumed point mass, moves through space until striking a hypothetical instrument panel, windshield, or side structure and subsequently is assumed to experience the remainder of the vehicle acceleration pulse by remaining in contact with the interior surface. There are two performance factors:

1. Lateral and longitudinal occupant impact velocities (OIVs) at the time of initial contact with interior surfaces of the vehicle. The following expression is used for the occupant impact velocity:

OIV, = 0 ,

(6-1)

where OIVx,y is occupant-to-car interior impact velocity in x and y directions; ax,y is vehicle acceleration in x and y directions; and t* is the smaller of the time when the occupant has traveled 24 in. forward or 12 in. laterally. Time t* is determined by incremental integration as follows:

103

= 0 0 2 = 0 0 2 = ,

(6-2) (6-3) (6-4)

where X is 24 in. and Y is 12 in. The preferred limit for the longitudinal and lateral OIVs is 30 ft/s, and the maximum allowed limit is 40 ft/s. 2. Largest lateral and longitudinal vehicular or occupant ride-down acceleration, ORA, averaged over every 10-ms interval for the collision pulse subsequent to occupant impact time, t*. After the occupant impact with the vehicle interior, the occupant, vehicle acceleration, and velocity are assumed to be equal. The interval of 0.010 sec is chosen because spikes of minimum duration to produce injury range from 0.007 sec to 0.04 sec. ORA can be obtained using accelerometers at the center of mass of the vehicle. The preferred limit for the longitudinal and lateral ORAs is 15 g, and the maximum allowed limit is 20.49 g.

Moreover, the vehicle should remain upright during and after the collision. The maximum roll and pitch angles are not to exceed 75 degrees. All accelerations and rotations of the vehicle are measured using a local coordinate system located at the center of mass of the vehicle. The standard coordinate system suggested by MASH is shown in Figure 74.

104

FIGURE 74 The Suggested Coordinate System in MASH
The deformation of the guardrail system can be assessed by measuring the maximum deflection of the rails during the collision (dynamic deflection) and the permanent deflection at the end of the collision. These two parameters were obtained from the simulations and are reported in this research. Although these parameters give a good measure of guardrail system deflection, they are not completely representative of the relative ground stiffness of systems with different mow strips. The guardrail posts close to the impact point usually fail, and their tops move considerably even with very stiff asphalt mow strips. However, if the mow strip is extremely strong, then the ground-level displacement of the posts is very limited. Therefore, the summation of ground-level displacement of all posts is used as an additional parameter to better assess the relative ground stiffness of the various mow strips considered in this research. This parameter is a
105

direct measure of ground-level stiffness and ground-level translation of the guardrail system, which is critical. If the guardrail posts do not translate sufficiently at the ground level, then the likelihood of wheel snagging and vehicle pocketing increases. Pocketing failure causes the vehicle to get stuck in the guardrail and not get redirected back to the road. A sudden stop of the vehicle during pocketing failure causes large deformations of the vehicle and extremely high vehicle deceleration, leading to a crash test failure.
6.3 Full-Scale Crash Simulation Results 6.3.1 Guardrail system without mow strip (soil only)
First, a baseline model without added asphalt mow strips was considered for validation and to establish a baseline for comparison with cases with asphalt mow strips. The simulation results are graphically presented in Appendix C in Figure 105 through Figure 110. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec (see Figure 105 and Figure 107 at 0.1 sec). The wheel failed partially but stayed connected to the vehicle and continued to rotate throughout the remainder of the simulation, as shown in Figure 110. At approximately 0.2 sec the vehicle hit the second post (see Figure 105 and Figure 107 at 0.2 sec) and passed over this post. The vehicle moved forward and hit the third post at approximately 0.3 sec (see Figure 105 and Figure 107 at 0.3 sec).
However, the wheel did not completely go over the post, as with the first and second posts. At this point, the vehicle was parallel to the guardrail. The vehicle was redirected back to the road and lost contact with the guardrail at approximately 0.65 sec (see Figure 106 and Figure 108). Vehicle pocketing did not occur during this simulation.
106

The longitudinal and lateral accelerations of the vehicle were obtained and are shown in Figure 111 and Figure 114. The integration of these accelerations over time produced the velocity of the occupant relative to the vehicle in longitudinal and lateral directions that are presented in Figure 112 and Figure 115. Integration of the velocity resulted in the displacement of the occupant relative to the vehicle, as presented in Figure 113 and Figure 116. The times when the lateral displacement became equal to 12 in. and when the longitudinal displacement became equal to 24 in. were found using the relative displacement curves (see Figure 113 and Figure 116). The smaller of these two was for the lateral displacement, equal to 0.142 sec. This indicates that the occupant first hit the interior of the vehicle on the sides before impacting somewhere in the front of the vehicle. After the occupant struck the interior, it is assumed that the occupant velocity and acceleration were the same as the vehicle. This is why the horizontal axis is extended only up to 0.2 sec, which was slightly after the time when the occupant hit the interior of the vehicle, in Figure 112 and Figure 113 where relative displacement or velocity of the occupant is presented. The longitudinal and lateral OIVs were found using the relative velocity curves (see Figure 112 and Figure 115) by finding the relative velocities at 0.142 sec. Thereafter, the peak accelerations from the acceleration curves (see Figure 111 and Figure 114) were found for the period of time after 0.142 sec until the end of the collision. The OIVs and ORAs were smaller than the maximum limit in the MASH guidelines. The results are reported in Table 13.
107

TABLE 13 Comparison between the Simulation Results for Guardrail Systems with Different Mow Strips

108

Soil Only Exp. [24]

Soil Only T3.5-R24-C T3.5-R12

Speed (mph)

40

36

36

37

Exit conditions

Trajectory angle, deg 14

19

13

15

ORA, g < 20.5

Longitudinal Lateral

8.2

-7.3

7.4

-7.8

6.9

7.4

8.5

7.4

OIV, ft/s < 40

Longitudinal Lateral

15.4

18.0

15.8

16.0

15.4

16.1

14.8

16.1

Test article
deflection, ft

Dynamic

3.61

Permanent

2.62

Impact time for the occupant t*, sec

N/A

Sum. of ground-level displacement of posts (ft)

N/A

Max. yaw angle, deg

46

3.32 2.67 0.142
5.53
41

3.35 2.66 0.143
4.84
44

2.98 2.32 0.136
4.83
40

Max. roll angle, deg Max. pitch angle, deg Posts detached from rails during impact

-5 -2 4 posts

-3.5 -2.9 3 posts

-4 -4 3 posts

-2 1 3 posts

Posts hit by leading tire (wheel snag) 3 posts 3 posts 3 posts 3 posts

Posts pulled out of ground

None

None

None

None

Leading tire/wheel disengaged

Mostly Mostly Mostly Mostly

T2-R24
35 13 -9.3 8.7 19.0 15.1 3.15 2.66 0.135
3.42
41 -3 -1 3 posts 3 posts None Mostly

T3.5-R24
36 18 -11.2 11.7 18.4 15.4 2.76 2.47 0.140
2.41
44 -6 -5 3 posts 3 posts None Mostly

T6-R24 T3.5-R24-S

35 18 -7.4 7.3 16.7 15.7 2.66 2.14 0.136
1.09
44 -7 -4 3 posts 3 posts None Mostly

37 15 -7.8 7.4 16.1 16.1 2.98 2.32 0.140
0.63
44 -4 -4 3 posts 3 posts None Mostly

The roll, pitch, and yaw angles of the vehicle were recorded during the simulation. The time histories for these parameters are shown in Figure 117 through Figure 119. The peak values of roll and pitch were less than the MASH limit, which is equal to 75 degrees. The initial impact angle of 25 degrees was subtracted fom the yaw angle at the time when the vehicle lost contact with the guardrail to determine the exit angle. The peak values of yaw, roll, and pitch angles, the exit angle, and the exit speed of the vehicle are reported in Table 13.
The maximum deflection of the guardrail, the permanent deflection of the guardrail, and the summation of ground-level displacement of all posts were measured and are reported in Table 13. None of the posts was pulled out of the ground during the simulation, and three posts were detached from the guardrail. Table 13 compares the results obtained from the simulations and an experimental MASH test on 31-in. MGS [24]. The comparison shows that all the parameters predicted by the finite element model were close to those measured in the experimental test. Therefore, the model was used for other simulations with added asphalt mow strips. The simulation results for the cases with asphalt mow strips will be compared with the results for the case with soil only to show how much ground restraint is added when the guardrail posts are encased in asphalt mow strips. The FEA timehistory simulations may be found in figures located in Appendix C of this report.
6.3.2 Guardrail mow strip with 2-in. asphalt thickness and 24-in. rear distance The model with soil only was updated by including an asphalt layer with a thickness
of 2 in. and rear distance of 24 in. (Test T2-R24). The simulation results are presented graphically in Figure 120 through Figure 125 in Appendix C. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec (see
109

Figure 120 and Figure 122 at 0.1 sec). The suspension system failed partially but stayed connected to the vehicle, and the wheel continued to rotate throughout the rest of the simulation as shown in Figure 125. At approximately 0.2 sec the vehicle hit the second post and passed over this post (see Figure 120 and Figure 122 at 0.2 sec). The vehicle moved forward and hit the third post at approximately 0.3 sec (see Figure 120 and Figure 122 at 0.3 sec). The leading wheel did not completely go over the post, as with the first and second posts. The wheel snagging for this case was similar to the simulation of the case without an asphalt layer, which was discussed in the previous section. The vehicle moved parallel to the guardrail at approximately 0.3 sec. The vehicle was redirected back to the road and lost contact with the guardrail at approximately 0.63 sec (see Figure 121 and Figure 123). Vehicle pocketing did not occur during this simulation. Asphalt rupture propagated for each post and was connected to the ruptures around other posts. This created a continuous rupture that ran parallel to the guardrail as shown in Figure 121 and Figure 123. Moreover, asphalt rupture occurred at the edge of the mow strip and behind each post, with a substantial ground-level displacement.
The four posts that had the largest ground-level displacement caused large asphalt ruptures, and one large part of the asphalt mow strip behind these four posts was nearly detached from the rest of the mow strip. This asphalt rupture limited the effect of the mow strip on the ground-level restraint.
The longitudinal and lateral accelerations of the vehicle were obtained and are shown in Figure 126 and Figure 129. The integration of these accelerations over time produced the velocity of the occupant relative to the vehicle in longitudinal and lateral directions as presented in Figure 127 and Figure 130. Another integration over time
110

resulted in the displacement of the occupant relative to the vehicle, as presented in Figure 128 and Figure 131. The times when the lateral displacement was equal to 12 in. and when the longitudinal displacement was equal to 24 in. were found using relative displacement curves (see Figure 128 and Figure 131). The smaller of these two was for the lateral displacement equal to 0.135 sec, meaning that the occupant first hit the interior of the vehicle on the sides before hitting somewhere in the front of the vehicle. After the occupant hit the interior, it is assumed that the occupant velocity and acceleration were the same as the vehicle for the remainder of the collision. The longitudinal and lateral occupant impact velocities were then found using the relative velocity curves (see Figure 127 and Figure 130) by finding the relative velocities at 0.135 sec. Thereafter, the peak accelerations from the acceleration curves (see Figure 126 and Figure 129) were found for the period of time after 0.135 sec until the end of the collision. The OIVs and ORAs were less than the maximum limit in the MASH guidelines. The results are reported in Table 13.
The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 132 to Figure 134. The peak values of roll and pitch were less than the MASH limit, which is equal to 75 degrees. The initial impact angle of 25 degrees was subtracted from the yaw angle at the time when the vehicle lost contact with the guardrail to determine the exit angle. The peak values of yaw, roll, and pitch angles, the exit angle, and the exit speed of the vehicle are reported in Table 13. The maximum deflection of the guardrail, permanent deflection of the guardrail, and summation of ground-level displacement of all posts are reported in Table 13. None of the posts were pulled out of the ground during the simulation; three posts were detached from the guardrail.
111

6.3.3 Guardrail mow strip with 3.5-in. asphalt thickness and 12-in. rear distance The model was updated by changing the asphalt layer geometry, increasing the
thickness to 3.5 in., and decreasing the rear distance to 12 in. (Test T3.5-R12). The simulation results are presented graphically in Figure 135 through Figure 140. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec (see Figure 135 and Figure 137 at 0.1 sec). Similar to the previous simulations, the suspension system failed partially but stayed connected to the vehicle and the wheel continued to rotate throughout the rest of the simulation as shown in Figure 140. At approximately 0.2 sec the vehicle hit and then passed over the second post (see Figure 135 and Figure 137 at 0.2 sec). The vehicle moved forward and hit the third post at approximately 0.3 sec. The leading wheel did not completely go over the third post as it did with the first and second posts (see Figure 135 and Figure 137 at 0.3 sec). The wheel snagging for this case was similar to the case with 2-in. thickness and 24-in. rear distance, Test T2-R24. The vehicle ran parallel to the guardrail at approximately 0.3 sec. The vehicle was redirected back to the road and lost contact with the guardrail at approximately 0.64 sec. Vehicle pocketing did not occur during this simulation. Asphalt rupture propagated around each post; however, unlike Test T2-R24, the ruptures remained local to each post and did not connect with the ruptures around other posts. This occurred because the rear distance here was half of the rear distance in Test T2-R24. Therefore, the rupture for each post reached the edge of the mow strip before joining with ruptures around the posts. When the rupture around each post reached the edge of the mow strip, the part of the mow strip behind each post was locally detached from the rest of the mow strip. This local
112

asphalt rupture around the posts considerably decreased the ground-level restraint caused by the asphalt mow strip.
The longitudinal and lateral accelerations of the vehicle were obtained and are shown in Figure 141 and Figure 144. The integration of these accelerations over time produces the velocity of the occupant relative to the vehicle in longitudinal and lateral directions as presented in Figure 142 and Figure 145. Another integration over time resulted in the displacement of the occupant relative to the vehicle, as presented in Figure 143 and Figure 146. The times when the lateral displacement equaled 12 in. and when the longitudinal displacement equaled 24 in. were found using the relative displacement curves (Figure 143 and Figure 146). The smaller of these two was for the lateral displacement equal to 0.136 sec, which is close to the occupant impact time for Test T2-R24, which was 0.135 sec. After the occupant hit the interior, it is assumed that the occupant velocity and acceleration were the same as the vehicle for the remainder of the collision. The longitudinal and lateral occupant impact velocities were then found using relative velocity curves (see Figure 142 and Figure 145) by determining the relative velocities at 0.136 sec. Thereafter, the peak accelerations from the acceleration curves (see Figure 141 and Figure 144) were found for the period of time after 0.136 sec until the end of the collision. The OIVs and ORAs were lower than the maximum limit in the MASH guidelines. The results are reported in Table 13.
The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 147 through Figure 149. The peak values of roll and pitch are lower than the MASH limit, which is equal to 75 degrees. The peak values of yaw, roll, and pitch angles, the exit angle, and the exit speed of the
113

vehicle are reported in Table 13. The maximum deflection of the guardrail, the permanent deflection of the guardrail, and the summation of ground-level displacement of all posts are reported in Table 13. None of the posts was pulled out of the ground during the simulation, and three posts were detached from the guardrail.
6.3.4 Guardrail mow strip with 3.5 in. asphalt thickness and 24 in. rear distance The model was updated by increasing the rear distance to 24 in. while keeping the
thickness unchanged at 3.5 in. (Test T3.5-R24). This asphalt mow strip geometry is the GDOT setup in the past. The simulation results are graphically presented in Figure 150 through Figure 155. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec (see Figure 150 and Figure 152 at 0.1 sec), which is the same as in previous simulations. However, this time the wheel passed behind the post instead of going over the post. This is due to the fact that the ground-level displacement of the post for this case was noticeably lower than in previous cases. Again, the suspension system failed partially but stayed connected to the vehicle; the wheel continued to rotate throughout the rest of the simulation, as shown in Figure 155. At approximately 0.2 sec the vehicle hit and passed over the second post (see Figure 150 and Figure 152 at 0.2 sec). The vehicle moved forward and hit the third post at approximately 0.3 sec. Unlike with the first and second posts, the leading wheel did not completely go over the third post. The wheel snagging for this case was worse than for Tests T2-R24 and T3.5-R12. The vehicle ran parallel to the guardrail at approximately 0.3 sec, and was then redirected back to the road, losing contact with the guardrail at approximately 0.65 sec. Vehicle pocketing did not occur during this simulation. Asphalt rupture propagated around each post; however, it was much more limited than in Tests T2-R24 and T3.5-R12. The ruptures for each post connected to
114

the ruptures around other posts. This is similar to the ruptures that occurred in Test T2-R24, which had the same rear distance. Ruptures behind only two posts reached the edge of the mow strip. Those segments of asphalt became detached from the rest of the asphalt. This considerably decreased the ground-level restraint for these two posts. However, the other posts remained encased in asphalt through the end of the simulation and experienced much higher ground-level restraint.
The longitudinal and lateral accelerations of the vehicle were obtained and are shown in Figure 156 and Figure 159. The integration of these accelerations over time produced the velocity of the occupant relative to the vehicle in the longitudinal and lateral directions as presented in Figure 157 and Figure 160. Another integration over time resulted in the displacement of the occupant relative to the vehicle as presented in Figure 158 and Figure 161. The times when the lateral displacement equalled 12 in. and when the longitudinal displacement equalled 24 in. were found using the relative displacement curves (see Figure 158 and Figure 161). The smaller of these two was for the lateral displacement, equal to 0.140 sec, which was close to the occupant impact time for previous test simulations. After hitting the interior, it is assumed that the occupant moved with the vehicle. The longitudinal and lateral occupant impact velocities were then determined using relative velocity curves (see Figure 157 and Figure 160) by finding the relative velocities at 0.140 sec. Thereafter, the peak accelerations from the acceleration curves (Figure 156 and Figure 159) were found for the period of time after 0.140 sec until the end of the collision. The OIVs and ORAs were lower than the maximum limit in the MASH guidelines. The results are reported in Table 13.
115

The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 162 to Figure 164. The peak values of roll and pitch were lower than the MASH limit of 75 degrees. The peak values of yaw, roll, pitch, the exit angle, the exit speed of the vehicle, the maximum deflection of the guardrail, the permanent deflection of the guardrail, and summation of ground-level displacement of all posts are reported in Table 13. None of the posts was pulled out of the ground during the simulation, and three posts were detached from the guardrail, which is the same as for all previously discussed test simulations.
6.3.5 Guardrail mow strip with 6-in. asphalt thickness and 24-in. rear distance The model was updated by increasing the asphalt layer thickness to 6 in.; the rear
distance remained unchanged at 24 in. (Test T6-R24). This asphalt mow strip geometry is the thickest and has the longest rear distance among all the setups investigated in this project. The simulation results are graphically presented in Figure 165 through Figure 170. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec, which is the same as in previous simulations. However, this time the wheel passed far behind the post instead of going over the post as in previous simulations (see Figure 165 and Figure 167 at 0.1 sec). This shows that the ground-level displacement of the post for this case was considerably lower than for the previous cases. In all the simulations discussed so far, the suspension system failed partially but remained connected to the vehicle, and the wheel continued to rotate throughout the rest of the simulation as shown in Figure 170. At approximately 0.2 sec, the vehicle hit the second post and passed behind it (see Figure 165 and Figure 167 at 0.2 sec). This did not happen in Tests T2-R24, T3.5-R12, T3.5-R24, or in the test with soil only. The vehicle moved
116

forward and hit the third post at approximately 0.3 sec (see Figure 165 and Figure 167 at 0.3 sec). The leading wheel passed over the post, which did not happen in Tests T2-R24, T3.5-R12, and T3.5-R24. The wheel snagging for this case was much more pronounced than in Tests T2-R24 and T3.5-R12. The vehicle ran parallel to the guardrail at approximately 0.3 sec and was redirected back to the road, losing contact with the guardrail at approximately 0.65 sec. Vehicle pocketing did not occur during this simulation. There were very limited asphalt ruptures around each post. These ruptures did not connect to any of the ruptures from other posts. This means all posts remained encased in asphalt through the end of the simulation and experienced extreme ground-level restraint.
The longitudinal and lateral acceleration, velocity, and displacement of the vehicle were obtained and are shown in Figure 171 to Figure 176. The occupant impact time is governed by the lateral impact of the occupant and was determined to be equal to 0.136 sec. The OIVs and ORAs were obtained based on the occupant impact time and are reported in Table 13. They are lower than the maximum limit in the MASH guidelines. The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 177 to Figure 179. The peak values of roll and pitch were lower than the MASH limit of 75 degrees. The peak values of yaw, roll, pitch, the exit angle, the exit speed of the vehicle, the maximum deflection of the guardrail, the permanent deflection of the guardrail, and summation of ground-level displacement of all posts are reported in Table 13. None of the posts was pulled out of the ground during the simulation, and three posts detached from the guardrail, which is the same as all previously discussed test simulations.
117

6.3.6 Guardrail mow strip diagonal pre-cuts The model for Test T3.5-R24 was modified by adding diagonal pre-cuts to the
asphalt layer behind all posts (Test T3.5-R24-C). The simulation results are graphically presented in Figure 180 through Figure 185. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec (see Figure 180 and Figure 182 at 0.1 sec). The suspension system failed partially but remained connected to the vehicle, and the wheel continued to rotate throughout the rest of the simulation as shown in Figure 185. At approximately 0.2 sec the vehicle hit and passed over the second post. The vehicle moved forward and hit the third post at approximately 0.3 sec. The leading wheel did not completely go over the post, as on the first and second posts. The wheel snagging for this case was much less than Test T3.5-R24; it was very similar to the baseline test without any asphalt mow strips. The vehicle moved parallel to the guardrail at approximately 0.3 sec. The vehicle was redirected back to the road and lost contact with the guardrail at approximately 0.63 sec. Vehicle pocketing did not occur during this simulation.
Asphalt rupture for each post started to propagate from the edge of each post's flanges and became connected to the pre-cuts. After the ruptures on the sides of a post connected with the pre-cuts behind that post, the asphalt resistance became zero for that post, and the post behaved as if there were no asphalt mow strip around it. The detached section of mow strip behind the posts slid freely and translated with little to no resistance. This occurred for five posts.
The longitudinal and lateral acceleration, velocity, and displacement of the vehicle were obtained and are shown in Figure 186 to Figure 191. The occupant impact time was
118

governed by the lateral impact of the occupant and was determined to be equal to 0.143 sec. The OIVs and ORAs were obtained based on the occupant impact time and are reported in Table 13. They were less than the maximum limit in the MASH guidelines. The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 192 to Figure 194. The peak values of roll and pitch were lower than the MASH limit, which is equal to 75 degrees. The peak values of yaw, roll, pitch, the exit angle, the exit speed of the vehicle, the maximum deflection of the guardrail, the permanent deflection of the guardrail, and summation of ground-level displacement of all posts are reported in Table 13. None of the posts were pulled out of the ground during the simulation and three posts were detached from the guardrail, which is the same as all previously discussed simulations.
6.3.7 Guardrail mow strip constructed using a stiffer asphalt The model for Test T3.5-R24 was updated by increasing the asphalt thickness and
stiffness while using the same material properties as in the previous sections (Test T3.5-R24-S). This asphalt was approximately two times stronger and stiffer than the asphalt type used in this research for all other test simulations. The simulation results are graphically presented in Figure 195 through Figure 200. During the collision, the leading front wheel of the vehicle hit a guardrail post at approximately 0.1 sec, which was the same as in previous simulations. However, in this case, the wheel passed far behind the post instead of going over the post's web. This shows that the ground-level displacement of the post in this scenario is extremely limited. As in all the other simulations in this research, the suspension system failed partially but stayed connected to the vehicle and the wheel continued to rotate throughout the rest of the simulation. At approximately 0.2 sec, the
119

vehicle hit the second post and passed far behind this post (Figure 195 and Figure 197 at 0.2 sec). This did not happen in Tests T2-R24, T3.5-R12, T3.5-R24, T3.5-R24-C, or the test with soil only. The vehicle moved forward and hit the third post at approximately 0.3 sec. The leading wheel passed over this post, which did not happen in Tests T2-R24, T3.5-R12, T3.5-R24, T3.5-R24-C, or the test with soil only. The wheel snagging for this case was the worst among all the test simulations. The vehicle ran parallel to the guardrail at approximately 0.3 sec, was redirected back to the road, and lost contact with the guardrail at approximately 0.61 sec. The vehicle lost contact with the guardrail earlier than any other case discussed in this research. Vehicle pocketing did not occur during this simulation. Very limited asphalt rupture occurred around the posts. This means all of the posts remained fully encased in asphalt through the end of the simulation and experienced extreme ground-level restraint that was close to having a rigid mow strip.
The longitudinal and lateral acceleration, velocity, and displacement of the vehicle were obtained and are shown in Figure 201 to Figure 206. The occupant impact time was governed by the lateral impact of the occupant and was determined to be equal to 0.140. The OIVs and ORAs were obtained based on the occupant impact time and are reported in Table 13. They were lower than the maximum limit in the MASH guidelines. The roll, pitch, and yaw angles of the vehicle were recorded during simulations. The time histories for these parameters are shown in Figure 207 to Figure 209. The peak values of roll and pitch were lower than the MASH limit, which is equal to 75 degrees. The peak values of yaw, roll, pitch, the exit angle, the exit speed of the vehicle, the maximum deflection of the guardrail, the permanent deflection of the guardrail, and summation of ground-level displacement of all posts are reported in Table 13. None of the posts were pulled out of the
120

ground during the simulation and three posts were detached from the guardrail, which is the same as in all previously discussed test simulations.
6.4 Quantitative Comparison between Guardrail Systems with Different Mow Strips Finite element crash simulations have limitations that prohibit them from being
used to determine if a particular safety structure passes the MASH requirements. The MASH guidelines specify that experimental tests are the only acceptable method for deciding whether a safety structure passes or fails its requirements. However, finite element simulations can be used to test additional alternatives, help design future experimental crash tests, and optimize the design of safety structures. The limitations of the finite element model used in this research include the following:
Steel rupture was not included. Therefore, rupture could not occur in guardrails and posts.
The vehicle wheels did not fail and detach as easily as they failed in experimental tests.
Rupture was not modeled for the bolted connections in the guardrails. Material failure was not modeled for the blockouts between the posts and
guardrails. The friction coefficient between the vehicle and the guardrail system is an
important parameter in crash simulations. In this research, the friction coefficient was assumed as the same between all surfaces of the vehicle and all parts of the guardrail system. Through calibration of the model for the case with soil only and by comparing the simulation results with experimental crash test results, the friction
121

coefficient was determined to be equal to 0.15. Higher values of friction angle resulted in too much loss of energy and lower vehicle exit speeds than was observed in the experimental test. Considerably higher values of friction coefficient (0.3 and higher) caused vehicle pocketing because the vehicle got stuck in the guardrail system. Friction values less than 0.15 resulted in too little loss of energy and higher vehicle exit speeds than was observed in the experimental test. Because of these limitations and based on the MASH guidelines, the crash simulation results presented here cannot be used to decide whether a guardrail system with a specific mow strip specification passes the MASH requirements. Instead, the results are used to determine the relative performance of guardrail systems with different asphalt mow strips. The guardrail mow strip configurations were compared based on wheel snagging, ground-level displacement of the guardrail posts, and rupture patterns of asphalt mow strips. This comparison shows that Tests T2-R24, T3.5-R12, and T3.5-R24-C show much less ground restraint than the other tests with asphalt mow strips. A quantitative comparison is presented in this section. The simulation results for all cases considered are summarized in Table 13. The summation of ground-level displacement of all posts in each simulation is used as the primary parameter. The cases presented in Table 13 are sorted in such a way that summation of ground-level displacement of posts decreases from left to right. The results are also presented in Figure 75. As shown in the table, the dynamic and permanent test article deflections did not have the same pattern as the ground-level displacement. This is because the rails detached from three of the posts in all simulations; therefore, the guardrail displacement was less dependent on the posts' ground-level displacements, which is critical. This table also shows that, with the exception of the summation of ground-
122

level displacement of posts, other parameters in the table do not show a consistent pattern as a function of mow strip thickness, rear distance, and material properties. Therefore, only the ground-level displacement of posts is used here as a parameter to compare the different setups. This parameter clearly shows that Tests T3.5-R12 and T3.5-R24-C behaved similarly to the case without an asphalt mow strip. The other setups had considerably less ground-level displacement of the posts when compared to Tests T3.5-R12, T3.5-R24-C, and the test with soil only. This indicates that these two setups are more likely to pass an experimental full-scale crash test based on the quantitative comparison of full-scale crash simulation results.
FIGURE 75 Summation of Ground-level Displacement of Posts Compared for Guardrail Systems with
Different Mow Strips Based on Simulation Results
123

CONCLUSIONS
The following conclusions can be drawn from this phase of the research project:
1. Most of the guardrail installed with mow strips in Georgia prior to 2017 employed the configuration shown in GDOT Standard Detail S-4-2002. This configuration includes an asphalt layer 3.5 in. thick. The posts for the guardrail system are driven through the asphalt layer, which extends 24 in. behind the post location. Where necessary, a curb is also installed to assist in drainage/runoff control.
2. In every GDOT district, a certain percentage of guardrail/mow strip installations do not conform to GDOT S-4-2002 in terms of mow strip geometry. GDOT staff from the Offices of Maintenance and Construction estimate these non-conforming sites make up approximately 10% of all guardrail/mow strip installations. The locations for these non-conforming sites vary throughout the state and are typically identified when roads are resurfaced or otherwise repaired.
3. Based on the results of the Phase I research program and a series of workshops with selected GDOT Design, Maintenance, and Construction staff, a test matrix for dynamic subcomponent experiments was developed to better understand the response of guardrail posts embedded in asphalt layers. The test matrix consisted of 14 tests that included three baseline setups (only soil), two benchmark setups using GDOT S-4-2002, and nine alternative mow strip setups to assess various configurations of asphalt layers.
4. A novel dynamic subcomponent testing method for testing guardrail systems using a high-speed hydraulic actuator was developed for use in this program. The test
125

method has benefits over other methods of subcomponent testing in terms of calibration of the impacting load, control of the mass, and development of a suitable programmer. The test setup was developed using specifications and guidelines taken from relevant MASH full-scale test conditions. 5. To evaluate the level of restraint on guardrail embedded in asphalt mow strips, four quantitative performance criteria were suggested: peak dynamic applied force, permanent ground-level displacement, peak displacement of the post at the impact point, and effective applied load. The performance assessment using relevant quantitative criteria and dynamic test results indicates that the restraint on the guardrail posts that occurs using the GDOT S-4-2002 configuration typical in Georgia exceeds the restraint on the post that occurs for mow strips that include leave-outs. The test results and analysis of performance criteria also indicate that either reducing specific geometric parameters of the mow strip or applying precutting can reduce the restraint on the guardrail post to a level similar to that demonstrated by the leave-out configuration. 6. Detailed finite element simulations of the subcomponent tests were performed that included a more detailed characterization of material properties for the asphalt layer. Prior FEA studies where the asphalt layer was assumed to be rigid will obviously cause the asphalt layer to produce excessive levels of restraint on the post. The use of a MohrCoulomb material model for the soil and asphalt provides an effective representation of the load-deflection response of the guardrail post, soil, and asphalt layer system over a broad range of material and geometric parameters. Moreover, erosion based on the stress and volumetric strain to capture
126

the rupture of the asphalt layer and the modeling of the contact conditions between the post and the soil are also key attributes of the FE simulation models. To ensure proper performance of the MohrCoulomb material model and the contact definitions, gravity loading must be applied. Dynamic relaxation should be employed in applying the gravity load to avoid waves caused by the sudden application of the gravity loading to the model. 7. The subcomponent tests and simulations are much more economical than full-scale crash tests and simulations. Therefore, the experimental results and simulations were correlated to develop a methodology for the rapid evaluation of alternative mow strip configurations in terms of restraint on the post. As seen in the standalone evaluation of the experimental results, decreasing the mow strip thickness and/or rear distance behind the post appears to be an effective way to reduce the restraint imparted by a mow strip on a guardrail system. For existing installations with unfavorable mow strip geometry, the installation of targeted pre-cuts into the asphalt layer appears to dramatically reduce the restraint of the mow strip on the posts. 8. Full-scale crash simulations were conducted to assess the performance of guardrail posts embedded in soil and asphalt mow strips. The baseline model without mow strip was first validated using available crash test results in the literature. Then, the model was updated to include an asphalt mow strip. Five different mow strip setups with various rear distances and thicknesses were studied. Two additional setups, one with stiffer asphalt than normal and one with diagonal mow strip pre-cuts, were studied. Numerous parameters based on the MASH guidelines were used to
127

evaluate the simulations. Based on this analysis, ground-level displacement was chosen as the primary parameter to evaluate the various guardrail / mow strip configurations used in the full-scale simulations. The results indicate that reducing the mow strip rear distance behind the post or using targeted pre-cuts in an existing asphalt layer have significant potential for reducing the level of restraint on a guardrail post. 9. The subcomponent testing, subcomponent finite element analysis, and full-scale crash simulations all indicate that reducing specific geometric parameters in the mow strip or using cuts in the asphalt layer may reduce the restraint in the system to a level similar to that found in a mow strip that incorporates a leave-out in accordance with the AASHTO Roadside Design Guide. However, these results cannot be taken as a definitive indicator of the performance of guardrail / mow strip configurations in actual crash conditions. Full-scale crash testing at an approved facility in accordance with the AASHTO Manual for Assessing Safety Hardware is necessary to definitively evaluate the effectiveness of a given mow strip configuration. 10. The GDOT S-4-2002 mow strip configuration is no longer in use by GDOT. Beginning March 15, 2017 all new GDOT guardrail construction projects on Georgia roadways were directed to use asphalt layers that were paved up to the face of the post, leaving the post itself and the area behind unrestrained. As such, new guardrail post installations will not be subject to additional restraint by asphalt layers. For existing guardrail systems with embedded posts where the mow strip geometry indicates the likelihood of excessive restraint, targeted pre-cuts in the
128

asphalt layer adjacent to and behind the post (as shown in Figures 70 and 71) may provide an economical retrofit methodology to reduce ground-level restraint and allow the system to function as intended in a crash situation.
129

REFERENCES
1. Georgia Department of Transportation. Standard Specifications: Section 641 Guardrail. http://www.dot.ga.gov/PartnerSmart/Business/Documents/GDOT_SpecBook_2013. pdf. (Accessed Aug. 25, 2015)
2. Georgia Department of Transportation. Construction Standards and Details: No. 4381 and S-4, 2002. http://mydocs.dot.ga.gov/info/gdotpubs/ConstructionStandardsAndDetails/Forms/A llItems.aspx. (Accessed Aug. 25, 2015)
3. American Association of State Highway and Transportation Officials (AASHTO). Roadside Design Guide, 4th Edition. Washington, D.C., 2011.
4. Bligh, R.P., N.R. Seckinger, A.Y. Abu-Odeh, et al. Dynamic Response of Guardrail Systems Encased in Pavement Mow Strips. Research Report FHWA/TX-04/0-41622, Texas Transportation Institute, College Station, Texas, 2004.
5. Arrington, D.R., R.P. Bligh, and W.L. Menges. Alternative Design of Guardrail Posts in Asphalt or Concrete Mowing Pads. Research Report 405160-14-1, Texas Transportation Institute, College Station, Texas, 2009.
6. Pass, D. Office of Design Policy and Support, Georgia Department of Transportation, 2013 (personal communication).
7. Whitesel, D., J. Jewell, and R. Meline. Development of Weed Control Barrier Beneath Metal Beam Guardrail. Technical Report FHWA/CA10-0515, California Department of Transportation, Sacramento, Calif., 2011.
8. Jowza, E.R., R.K. Faller, S.K. Rosenbaugh, et al. Safety Investigation and Guidance for Retrofitting Existing Approach Guardrail Transitions. MwRSF Research Report No. TRP-03-266-12, Midwest Roadside Safety Facility, Nebraska, 2012.
130

9. Rosenbaugh, S.K., T.L. Schmidt, R.K. Faller, and J.D. Reid. Development of Socked Foundations for S35.7 Posts. Research Report TRP-03-293-15, Midwest Roadside Safety Facility, Lincoln, Nebraska, 2015.
10. Ross, Jr., H.E., D.L. Sicking, and R.A. Zimmer. Recommended Procedures for the Safety Performance Evaluation of Highway Features. NCHRP Report 350, National Cooperative Highway Research Program, 1993.
11. American Association of State Highway and Transportation Officials (AASHTO). Manual for Assessing Safety Hardware (MASH), First Edition. AASHTO, Washington, D.C., 2009.
12. Stewart, L.K., N. Gao, G. Pezzola, et al. "Georgia Institute of Technology Laboratory for Blast, Shock, and Impact." In 7th International Conference on Advances in Experimental Structural Engineering (7AESE), Pavia, Italy, 2017.
13. Gram, M.M., A.J. Clark, G.A. Hegemier, and F. Seible. "Laboratory Simulation of Blast Loading on Building and Bridge Structures." WIT Transactions on The Built Environment, 2006, 87 (Structures Under Shock and Impact IX): pp. 3344.
14. Stewart, L.K., A. Freidenberg, T. Rodriguez-Nikl, and G.A. Hegemier. "Methodology and Validation for Blast and Shock Testing of Structures Using High-speed Hydraulic Actuators." Engineering Structures, 2014, 70: pp. 168180.
15. LS-DYNA V971 R9.1.0, Livermore Software Technology Corporation (LSTC), Livermore, Calif., 2017.
16. Malvar, L.J. "Review of Static and Dynamic Properties of Steel Reinforcing Bars." Am. Concr. Inst. Mater. J., 1997, 95:5, pp. 609616.
17. Budhu, M. Soil Mechanics and Foundations, 3rd ed. Hoboken, New Jersey: John Wiley & Sons, Inc., 2010.
18. USACE. Bearing Capacity of Soils: Engineering Manual. United States Army Corps of Engineers, 1982.
19. Bowles, J.E. Foundation Analysis and Design. 5th ed. New York: McGraw-Hill, 2001.
131

20. Christensen, R.F., and D.W. Bonaquist. Evaluation of Indirect Tensile Test (IDT) Procedures for Low-temperature Performance of Hot Mix Asphalt. No. 530, 2004.
21. Pellinen, T., J. Song, and S. Xiao. Characterization of Hot Mix Asphalt With Varying Air Voids Content Using Triaxial Shear Strength Test. 8th Conference on Asphalt Pavements for Southern Africa, 2004.
22. Allen, D. Stiffness Evaluation of Neoprene Bearing Pads under Long-Term Loads. University of Florida, 2008.
23. National Crash Analysis Center (NCAC). Development and Validation of a Finite Element Model for a 2007 Chevy Silverado. NCAC 2009-T-005, 5, prepared for FHWA, 2009.
24. Polivka, K.A., D.L. Sicking, B.W. Bielenberg, et al. Performance Evaluation of the Midwest Guardrail System Update to NCHRP 350 Test No. 3-11 with 28 C.G. Height (2214MG-2). Final Report to the National Cooperative Highway Research Program, MwRSF Research Report No. TRP-03-171-06, Lincoln, Nebraska, 2006.
25. Scott, D.W., D.W. White, L.K. Stewart, et al. Evaluating the Performance of Guardrail Posts Installed by Driving Through Asphalt Layers. Report 13-21, Georgia Department of Transportation, 2015.
26. American Association of State Highway and Transportation Officials (AASHTO). Standard Method of Test for Moisture-Density Relations of Soils Using a 4.54-kg (10-lb) Rammer and a 457-mm (18-in.) Drop (AASHTO T 180). 2010.
27. McLeod, N.W. A Rational Approach to the Design of Bituminous Paving Mixtures. Association of Asphalt Paving Technologists, St. Paul, Minn., 1950.
28. Smith, V.R. Application of the Triaxial Test to Bituminous Mixtures. ASTM Special Technical Publication, 1951, 106: pp. 5578.
29. Fwa, T.F., S.A. Tan, and L.Y. Zhu. "Rutting Prediction of Asphalt Pavement Layer Using C- Model." Journal of Transportation Engineering, ASCE, 2004, 140: pp. 675683.
132

30. Fwa, T.F., S.A. Tan, and L.Y. Zhu. "Reexamining C- Concept for Asphalt Paving Mix Design." Journal of Transportation Engineering, 2001, 127(1): pp. 6773.
31. Zhang, Y., R. Luo, and R.L. Lytton. "Characterization of Viscoplastic Yielding of Asphalt Concrete, Construction and Building Materials." Construction and Building Materials, 2013, 47: pp. 671679.
32. Bell, C.A. Summary Report on Aging of Asphalt-Aggregate Systems. Strategic Highway Research Program (SHRP) Report SR-OSU-A-003A-89-2, 1989.
33. Farrar, M.J., T.F. Turner, J-P. Planche, et al. "Evolution of the Crossover Modulus with Oxidative Aging." Transportation Research Record, 2013, 2370(1): pp. 76 83.
34. Pellinen, T.K., and S. Xiao. Stiffness of Hot-Mix Asphalt. Joint Transportation Research Program (JTRP), Technical Report No. FHWA/IN/JTRP-2005/20, 2006.
35. American Association of State Highway and Transportation Officials (AASHTO). Standard Specification for Performance-Graded Asphalt Binder (AASHTO T320). 2007.
36. American Society for Testing and Materials (ASTM). Standard Test Method for Compressive Strength of Bituminous Mixtures (ASTM D1074-09). 2009.
37. American Society for Testing and Materials (ASTM). Test Method for Effect of Water on Compressive Strength of Compacted Bituminous Mixtures (ASTM D1075). ASTM International, 2011.
38. Niazi, Y., and M. Jalili. "Effect of Portland Cement and Lime Additives on Properties of Cold In-place Recycled Mixtures with Asphalt Emulsion." Construction and Building Materials, 2009, 23: pp. 13381343.
39. Xu, J., S. Huang, Y. Qin, and F. Li. "The Impact of Cement Contents on the Properties of Asphalt Emulsion Stabilized Cold Recycling Mixtures." Int. J. Pavement Res. Technol., 2011, 4(1): pp. 4855.
133

APPENDIX A

DYNAMIC SUBCOMPONENT TEST SETUP AND DETAILED RESULTS

A.1 Dynamic Test Bed Construction To provide the most flexibility for specimen preparation as well as the testing
schedule, a moveable test bed was constructed that could be relocated in and out of the SEMM Laboratory testing area as needed. Three test beds were fabricated using three steel front-load sanitation containers. The containers, shown in Figure 76, were selected after a soil influence zone analysis using visual observation and finite element simulation in a previous study [25] so that the container boundary effect could be lessened in the dynamic tests.
Prior to soil placement, these containers were reinforced by welding additional steel sections around and under the container, as shown in Figure 77. This reinforcement was necessary for multiple uses of the test bed in the dynamic test program because the original unreinforced dumpster was not designed to carry lateral impact loads on the sidewall along with the dead weight of the compacted soil. After being reinforced, the container could be towed into position, laterally supported by a supplemental frame system, and then anchored down to the SEMM Laboratory Strong Floor for dynamic testing. The movable test bed concept allows a test user to significantly reduce the preparation time between dynamic tests in the Laboratory.

135

FIGURE 76 Dimension of Dynamic Test Bed (6-cubic-yard Dumpster)
136

(a)
(b) FIGURE 77 Steel Container Reinforcement: (a) Drawing; (b) Fabrication After container fabrication was completed, the same type of base soil used in the Phase I research program was placed and compacted to meet the grading requirements in
137

accordance with the MASH guidelines [11]. Then, the asphalt mow strip was installed over the compacted soil. In order to provide shear resistance along the side boundary of the asphalt mow strip, metal shear studs were welded on the sidewalls of the container (Figure 78 (a)). These shear studs were located at the middle of the mow strip thickness every 5 in. along the sidewalls. Both soil and asphalt layers were compacted using a plate vibratory compactor (Figure 78 (b)).

(a)

(b)

FIGURE 78 Asphalt Mow Strip Installation: (a) Shear Studs; (b) Compaction

Due to practical limitations in the research program schedule, the hot mix asphalt (HMA) used in the static test program could not be used for the dynamic testing program--there was not sufficient time to allow the hot mix material to age between test sessions. As such, a cold mix asphalt (CMA) was used for the mow strips in the dynamic test program. To provide a strength and stiffness in the restraint layer similar to that seen in the static test program, Portland cement and water were added to the CMA and a Portland cement ratio of 10% by weight and 0.33 w/c (water/cement) ratio was selected as an acceptable mix design for the mow strips used in the dynamic test program.

138

For each mow strip test bed construction, 4-in.-by-4-in. cylindrical specimens were prepared at the same time based on the method specified in AASHTO T 180 [26]. Table 14 shows uniaxial compression test results of the specimens made from four different cold mix batches. The average compressive strength at approximately 2 weeks was 239 psi and exceeded the target strength of 232 psi, which was determined from the Phase I research results. Hence, it can be asserted that the mow strip strength and stiffness were similar to structural properties of asphalt used in the Phase I program under the same temperature.

TABLE 14 Cold Mix Asphalt Compression Test Results

No. of Batch
1 2 3 4 Average

Average Compressive Strength (psi)
273.6 229.7 248.0 193.8 239.3

No. of Specimens
Tested
3 3 2 2

Test Temperature
(F)
66 68 71 69 68.2

Asphalt Age (day)
14 13 13 11 12.9

The hydraulic post-rammer vehicle owned by the Georgia Department of Transportation used in the static test program was not suitable for driving the posts safely into the test container due to the elevation of the container above ground level. Instead, a vertical static load was applied using a hydraulic loading system in the laboratory as shown in Figure 79. The maximum loading capacity of 100 kips and the quasi-static loading rate of approximately 0.5 in./sec enabled both safe application of the loading and continuous monitoring of the slope and orientation of the post while driving. The slope and orientation of the post were carefully controlled so that the post could be driven in a perpendicular

139

fashion. The maximum driving angle of the post did not exceed 1 degree from the perpendicular direction relative to the ground surface.
FIGURE 79 Post Driving for Dynamic Test Bed A flyer mass unit was precisely designed and fabricated for applying the impact to the post as desired. As shown in Figure 80, the flyer mass consists of: (1) two wing steel plates (6361: Height Width Thickness) to use as guides on the impact rail system, (2) middle steel plates (6101) located between the wing plates to allow for adjustment of the mass, (3) front and rear steel plates (6102) with threaded holes and recessed slots, and (4) four threaded rods and nuts (0.5 diameter) for connecting all the parts into one rigid body. By changing the number of middle plates, the mass could be adjusted from 175 to 350 lb. One of the middle plates has a mounting slot on the top side so that an accelerometer can be safely mounted. Additionally, a steel safety chain system (Figure 81)
140

was employed to restrain the motion of the flyer subsequent to post-flyer impact. The safety chain was effective in preventing damage to the test setup (e.g., hydraulic actuator), as well as for allowing the reuse of the test components for multiple experiments.
FIGURE 80 Adjustable Flyer Mass
FIGURE 81 Safety Chain System
141

A.2 Test Instrumentation A.2.1 Accelerometers

TABLE 15 Accelerometer Specification

Manufacturer Model designation Sensitivity (20%) Measurement range Overload limit (shock) Operating temperature range
Frequency range (5%)
Resonant frequency Non-linearity Transverse sensitivity Weight Dimensions

PCB Piezotronics 356B20 1.0 mV/g 5000 g pk 7000 g pk -65 to +250F 2 to 7000 Hz (x axis) 2 to 10000 Hz (y or z axis) 55 kHz 2.5% 5% 0.14 oz 0.4 0.4 0.4 in.

FIGURE 82 Accelerometer (PCB 356B20)
142

A.2.1 Data acquisition system

TABLE 16 Data Acquisition System Specification

Manufacturer Model designation Number of input modules Number of input channels Maximum data rate Data streaming rate Maximum sampling rate with 16-bit resolution

Hi-Techniques Synergy P 4 16 2 MS/s (for each channel) 500 kS/s (for all channels)
100 MS/s (for each channel)

FIGURE 83 Data Acquisition System (Hi-Techniques Synergy P)
143

A.2.3 High-speed cameras

TABLE 17 High-speed Camera Specification

Manufacturer
Model designation
Maximum resolution
Pixel size Maximum FPS at maximum resolution Maximum FPS at 720p HD resolution Throughput
Minimum exposure

Vision Research Phantom Miro M310 1280 800 20 m 3200 fps
3600 fps 3.2 Gpx/s 1 s

Vision Research Phantom Miro C110 1280 1024 5.6 m 915 fps
1295 fps 1.2 Gpx/s 5 s

(a) Phantom Miro M310

(b) Phantom Miro C110

(aaaaa

FIGURE 84 High-speed Cameras Used in Dynamic Tests

144

A.3 Test Record Sheets

Dynamic Test #13

Baseline

* Baseline configuration

* No mow strip

*

Mow strip Thickness

0 in.

configuration Rear Distance

0 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Impact
load 32 in.
25 in.

40 in.

Soil base

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/26/2017

Temperature

71 F

Asphalt age N.A.

21.7 C

Sequential test pictures

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time
0

-5

-10

-15

-20

-25

-30 0

20

40

60

80 100 120

Time after impact (ms)

Post displacement vs. Time
25

20

15

10

5

0

0

20

40

60

80 100 120

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

9000 8000 7000 6000 5000 4000 3000 2000 1000
0 0

5

10

15

20

25

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-27.36 109 3 23.11 8154

FIGURE 85 Summary of Dynamic Test Results: Baseline

145

Dynamic Test #6

Typical mow strip

* 3.5" thickness and 24" rear distance behind the post

*

*

Mow strip Thickness

3.5 in.

configuration Rear Distance

24 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Impact load 32 in. 25 in.
40 in.

3.5 in.
24 in. Asphalt mow strip

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

12/19/2016

Temperature Asphalt age

66 F

18.9 C

14 days from placement

Sequential test pictures

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time

0 -5 -10 -15 -20 -25 -30 -35 -40 -45
0

20

40

60

80 100 120

Time after impact (ms)

Post displacement vs. Time

16

14

12

10

8

6

4

2

0

0

20

40

60

80 100 120

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

14000 12000 10000
8000 6000 4000 2000
0 0

5

10

15

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-42.53 96 19
13.71 12677

FIGURE 86 Summary of Dynamic Test Results: Typical Mow Strip

146

Dynamic Test #12 Leave-out

* Typical leave-out application

* 18"x18" low strength grout as leave-out material

* Recommended by the AASHTO Roadside Design Guide

Mow strip Thickness

3.5 in.

configuration Rear Distance

24 in.

Modification Typical grout leave-out

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/26/2017

Temperature Asphalt age

71 F

21.7 C

13 days from placement

Sequential test pictures

Impact load
25 in. 4 in.

Asphalt Mow

8 in.

Strip

(3.5 in.)

Time after impact

0-ms

40-ms

3.5 in. Soil base

Leave-out Grout fill (3.5 in.)

Acceleration (g)

18 in.

16 in. 8 in.

18 in. 4 in.

End of test pictures

80-ms

120-ms

Flyer acceleration vs. Time

0

-5

-10

-15

-20

-25

-30

-35 0

20

40

60

80

100

Time after impact (ms)

Post displacement vs. Time

18 16 14 12 10
8 6 4 2 0
0

20

40

60

80

100

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

12000

10000

8000

6000 4000

2000

0

0

5

10

15

20

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) Time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-32.38 92 21
15.91 9650

FIGURE 87 Summary of Dynamic Test Results: Leave-out Installation

Displacement at impact height (in)

147

Dynamic Test #8

Pre-cut (parallel)

* Pre-cut mow strip design

* 3.5" thickness and 24" rear distance behind the post

* Parallel pre-cut pattern

Mow strip Thickness

3.5 in.

configuration Rear Distance

24 in.

Modification Pre-cut application (parallel)

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

12/20/2016

Temperature Asphalt age

68 F

20 C

13 days from placement

Sequential test pictures

Impact load 32 in. 25 in.
40 in.

3.5 in.
24 in. Asphalt mow strip

Time after impact

0-ms

40-ms

Thi ckness 3.5" (Rea r distance) 24"

6" (mi n. distance to post) 16"

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time

0

-5

-10

-15

-20

-25

-30

-35 0

50

100

150

Time after impact (ms)

Post displacement vs. Time
25

20

15

10

5

0

0

50

100

150

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

10000 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0

5

10

15

20

25

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-29.60 123 21 22.00 8821

FIGURE 88 Summary of Dynamic Test Results: Parallel Pre-cut

148

Dynamic Test #10

Pre-cut (diagonal)

* Pre-cut mow strip design

* 3.5" thickness and 24" rear distance behind the post

* Diagonal pre-cut pattern

Mow strip Thickness

3.5 in.

configuration Rear Distance

24 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Impact load
32 in. 25 in.
40 in.

3.5 in.
24 in. Asphalt mow strip

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/5/2017

Temperature Asphalt age

71 F

21.7 C

16 days from placement

Sequential test pictures

Time after impact

0-ms

40-ms

Acceleration (g)

(Rear distance) 24"

6" (min. distance to post)

45

End of test pictures

80-ms

120-ms

Flyer acceleration vs. Time
10

0

-10

-20

-30

-40

-50 0

20

40

60

80 100 120

Time after impact (ms)

Post displacement vs. Time

16 14 12 10
8 6 4 2 0
0

20

40

60

80 100 120

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

14000 12000 10000
8000 6000 4000 2000
0 0

5

10

15

20

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-44.22 96 19
14.61 13180

FIGURE 89 Summary of Dynamic Test Results: Diagonal Pre-cut

149

Displacement at impact height (in)

Dynamic Test #9

Thin mow strip

* 1.5" thickness and 24" rear distance behind the post

*

*

Mow strip Thickness

1.5 in.

configuration Rear Distance

24 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/5/2017

Temperature Asphalt age

71 F

21.7 C

16 days from placement

Sequential test pictures

Impact load 32 in. 25 in.
40 in.

1.5 in.
24 in. Asphalt mow strip

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time
0

-5

-10

-15

-20

-25

-30 0

20

40

60

80 100 120

Time after impact (ms)

Post displacement vs. Time
25

20

15

10

5

0

0

20

40

60

80 100 120

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

9000 8000 7000 6000 5000 4000 3000 2000 1000
0 0

5

10

15

20

25

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-26.07 112 23 21.08 7771

FIGURE 90 Summary of Dynamic Test Results: Thin Mow Strip

150

Dynamic Test #11 Thick mow strip

* 5.5" thickness and 24" rear distance behind the post

*

*

Mow strip Thickness

5.5 in.

configuration Rear Distance

24 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/6/2017

Temperature Asphalt age

71 F

21.7 C

16 days from placement

Sequential test pictures

Impact load 32 in. 25 in.
40 in.

5.5 in.
24 in. Asphalt mow strip

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time

10

0

-10

-20

-30

-40

-50

-60 0

20

40

60

80

100

Time after impact (ms)

Post displacement vs. Time
12

10

8

6

4

2

0

0

20

40

60

80

100

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

18000 16000 14000 12000 10000
8000 6000 4000 2000
0 -2000 0

2

4

6

8

10

12

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-53.45 81 18
10.45 15931

FIGURE 91 Summary of Dynamic Test Results: Thick Mow Strip

151

Dynamic Test #7

Reduced RD

* 3.5" thickness and 12" rear distance behind the post

*

*

Mow strip Thickness

3.5 in.

configuration Rear Distance

12 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Impact load 32 in. 25 in.
40 in.

3.5 in.
12 in. Asphalt mow strip

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

12/19/2016

Temperature Asphalt age

68 F

20 C

13 days from placement

Sequential test pictures

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time
0

-5

-10

-15

-20

-25

-30 0

50

100

150

Time after impact (ms)

Post displacement vs. Time
30

25

20

15

10

5

0

0

50

100

150

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

9000 8000 7000 6000 5000 4000 3000 2000 1000
0 0

5

10

15

20

25

30

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-26.58 131 4 25.81 7922

FIGURE 92 Summary of Dynamic Test Results: Reduced RD

152

Dynamic Test #14

Thick & Reduced RD

* 5.5" thickness and 12" rear distance behind the post

*

*

Mow strip Thickness

5.5 in.

configuration Rear Distance

12 in.

Modification N.A.

Impact conditions

Speed

15.0 m/s

32.4 mph

Mass

139 kg

305.8 lb.

KE

15638 J

137.65 kip-in.

Test configuration drawings

Test location SEMM Lab, Georgia Tech, Atlanta, GA

Test date

1/27/2017

Temperature Asphalt age

69 F

20.6 C

11 days from placement

Sequential test pictures

Impact load 32 in. 25 in.
40 in.

5.5 in.
12 in. Asphalt mow strip

Time after impact

0-ms

40-ms

End of test pictures

80-ms

120-ms

Acceleration (g)

Displacement at impact height (in)

Flyer acceleration vs. Time

5 0 -5 -10 -15 -20 -25 -30 -35 -40
0

20

40

60

80 100 120

Time after impact (ms)

Post displacement vs. Time

20 18 16 14 12 10
8 6 4 2 0
0

20

40

60

80 100 120

Time after impact (ms)

Force (lb)

Dynamic force vs. displacement at impact

12000

10000

8000

6000

4000

2000

0

0

5

10

15

20

-2000

Displacement at impact height (in)

Peak flyer acceleration (g) Impact duration (ms) time at peak acceleration (ms) Max. displacement at impact (in) Peak dynamic force (lb)

-33.65 112 20 17.52
10030

FIGURE 93 Summary of Dynamic Test Results: Thick and Reduced RD

153

APPENDIX B

MATERIAL CHARACTERIZATION OF ASPHALT

This appendix presents the results of an investigation into the variability of asphalt strength used to evaluate the structural performance of the mow strip, focusing on two environmental parameters: temperature and age. Existing empirical design equations for pavement thickness are evaluated for suitability to be implemented in this study. Subsequently, an empirical procedure is proposed for estimating the failure of the asphalt material including the effects of temperature and aging. Finally, the development of a cold mix asphalt supplemented with Portland cement is discussed for use in the dynamic test program outlined in Chapter 3.
B.1 Specimen Testing for Asphalt Strength Characterization The expected load on the mow strip imparted by a vehicle impact will often lead to
the failure of the mow strip in a single event. Thus, it is necessary to select a suitable failure criterion of asphalt for mow strip performance evaluation. A series of unconfined compression tests was conducted with various temperature and age conditions, with the results used to develop a simplified method to determine a failure criterion for the characterization of asphalt strength.
B.1.1 MohrCoulomb failure model One of the most widely used material models for asphalt concrete is the
MohrCoulomb (MC) failure criterion model. The MC model was first adopted by pavement researchers in the early 1950s to evaluate the performance of asphalt mix designs [27,28]. Later, studies conducted by Fwa et al. [29,30] proposed a modified triaxial

155

test method to determine the C relationship at various temperature conditions. The

cylindrical specimens were of 4-in. diameter and 8-in. height and were mixed and

compacted in the laboratory, and then tested less than 24 hours after asphalt compaction.

However, this method is not adequate to evaluate the C relationship for in-situ asphalt

samples taken from the mow strip since it is nearly impossible to retrieve a triaxial test

specimen with the necessary 8-in. height from roadside asphalt layers whose thicknesses

typically range from 2 to 5 in. Additionally, experimental results on short-term aged

laboratory specimens are unlikely to be indicative of aged asphalt in actual roadway

conditions.

The MC failure criterion model can be expressed as shown in Equation (B-1):

- tan - = 0

(B-1)

where is the yield (failure) shear stress, is the normal stress, C is the cohesion stress,

and is the internal friction angle. The stress condition of a cylindrical specimen under

uniaxial compression with no circumferential confinement can be graphically illustrated

with a Mohr's circle as shown in Figure 94. The axial stress, 1, is the compressive strength (1 = ); the lateral stresses 2 = 3 are zero; and both the radius and the normal coordinate of the center of the circle are half of the compressive strength. The MC failure

envelope can be drawn as a line tangent to the circle as shown.

156

FIGURE 94 MohrCoulomb Parameters from Unconfined Compression Test

The tensile strength of the asphalt, , is not easily estimated from experiments because of the difficulty in applying uniaxial tension to the specimen. Instead, a series of

equations can be written from trigonometric ratios from Figure 94 as follows:



=

tan

sin



=

2+

Equation (B-2) can be written as shown in Equation (B-4):

(B-2) (B-3)



=

2

sin1

-

1

(B-4)

Finally, the cohesion, C, can be written as a function of the internal friction angle, , and

compressive strength, , by substituting in Equation (B-4) with Equation (B-2):



=

2

co1s

-

tan



(B-5)

157

The cohesion is proportional to the compressive strength but varies with the internal friction angle. The cohesion ratio, defined as the cohesion value divided by the compressive strength (/), is plotted in Figure 95.

FIGURE 95 Cohesion Ratio and Angle of Friction

To estimate the cohesion value of the MC failure envelope from the uniaxial

compressive tests on cylindrical specimens, the internal friction angle of asphalt, , is

assumed to be 35 degrees. This value is taken from a representative value of satisfactory

asphalt mix design criteria proposed by Smith [28]. Therefore, the cohesion value of

asphalt, C, is estimated at approximately 26% of the compression strength as shown in

Equation (B-6):

= 0.2603

(B-6)

158

B.1.2 Experimental plan for asphalt material characterization It is known that asphalt strength can vary significantly based on a number of
different factors. As prior studies have reported, asphalt strength is highly sensitive to both temperature [29,31] and age [3234]. Specifically, the performance grade (PG) [35] in an asphalt mix design specifies the maximum and minimum pavement design temperatures to prevent rutting in high temperature and thermal cracking in low temperature. Observations from the static test program showed there is a significant difference in the character of mow strip fracture between the summer and winter test conditions with the same test configuration (thickness: 2 in.; rear distance: 24 in.) as shown in Figure 96. Asphalt rupture in the summer test was accompanied by buckling and more deformation of the asphalt behind the post. However, the asphalt failure mode in the winter test was mostly a lateral rupture. The formation and propagation of cracks were more visible and distinct.
FIGURE 96 Difference in Mow Strip Fracture Between Two Tests
159

A series of compression tests was performed to attempt to estimate the effect of temperature and physical aging on asphalt strength for the specific material used in this research program. The hot mix asphalt used in this research program was designed with a performance grade of PG 76-22 binder with -in. maximum aggregate size. This asphalt mix type is one of the most commonly used in road construction projects in Georgia. Test samples were cored from the asphalt pavement layer and were trimmed to approximately 4-in. diameter and 4-in. height as shown in Figure 97.
FIGURE 97 Asphalt Test Bed and Representative Cored Specimens It is well understood that both temperature and age effects on asphalt strength are not monotonic. For this reason, three or more levels are required to analyze any curvilinear relationship between factors and their responses. In this study, three levels of temperature and eight levels of age condition were evaluated based on the practical limitations of specimen test settings and time constraints. Three test temperature levels were investigated: 32, 68, and 104F. Each level represents a common cold, moderate, and elevated temperature condition, respectively. A total of 36 compression tests were performed at various age conditions of 26, 46, 67, 94, 105, 124, 159, and 182 days from the initial
160

placement of the asphalt mow strip. For each age level, a minimum of three replicate specimens was tested.
Loading speed and moisture are two important control variables in this experimental setup as specified in the relevant testing specifications (ASTM D1074 and D1075 [36,37]). To avoid dynamic or strain rate effects while testing, a loading rate of 0.2 in./min was used; this rate is specified in ASTM D1074 Standard Test Method for Compressive Strength of Bituminous Mixtures [36]. As shown in Figure 98, all specimens were prepared at the same time and moved to an oven (for high temperature conditioning), a refrigerator (for low temperature conditioning), or to an environmental chamber whose temperature was kept constant at 68F.
FIGURE 98 Temperature Conditioning of Cored Specimens Test specimens were loaded to failure in compression using a universal test machine as shown in Figure 99. Due to imperfections in the sample trimming process, not all specimens had completely horizontal top and bottom loading surfaces. A high-strength
161

steel ball was placed on top of the test specimen to minimize effects of stress concentration due to misalignment of the specimen in the testing machine.
FIGURE 99 Test Specimen in Compressive Failure B.1.3 Test results The compressive strength from each test was calculated from the maximum recorded load divided by the original cross-sectional area of the specimen. The cohesion value is estimated using Equation (B-6). Table 18 and Table 19 show the unconfined compression test results for all specimens. The effect of temperature was notable: the compressive strength at the lowest temperature was approximately 10 times higher than the strength at the highest temperature. This result indicates the asphalt mow strip would behave in a more rigid fashion under extremely low temperature conditions. The effect of aging was not as
162

significant as for temperature, but there was a slight positive correlation between the age and the strength/cohesion of the asphalt.

TABLE 18 Asphalt Compression Test Results Showing Effect of Aging

Test Run

Age of

Test

Specimen Temperature

(day)

(F)

Number of Specimens
Tested

Average Compressive
Strength (psi)

1

26

68

2

46

68

3

46

68

4

67

68

5

67

68

6

94

68

7

105

68

8

124

68

9

124

68

10

124

68

11

159

68

12

182

68

3

156.8

3

185.6

3

191.6

3

217.5

3

251.0

3

225.1

3

224.3

3

236.5

3

214.7

3

270.2

3

204.5

3

255.6

Estimated Cohesion
Value (psi)
40.82 48.31 49.87 56.62 65.34 58.59 58.39 61.56 55.89 70.33 53.23 66.53

TABLE 19 Asphalt Compression Test Results Showing Effect of Temperature

Test Run

Age of

Test

Number of

Specimen Temperature Specimens

(day)

(F)

Tested

Average Compressive
Strength (psi)

1

67

32

3

718.2

2

67

68

3

217.5

3

67

68

3

251.0

4

67

104

3

74.03

5

182

32

2

876.0

6

182

68

3

255.7

7

182

104

2

45.45

Estimated Cohesion
Value (psi)
187.0 56.63 65.34 19.27 228.0 66.55 11.83

163

B.2 Empirical Models for the Effect of Temperature and Age

Using the results from the compression tests on the specific asphalt material used

in this study, the effect of temperature and age can be estimated empirically. General curve-

fitting techniques were used to determine the two empirical equations: a

temperaturecompressive strength relationship and an agecompressive strength

relationship.

For the temperature model, a decaying power function was selected for maximizing

the goodness of fit. The function showed a strong correlation with the test data--the regression coefficient, R2, was close to 1.0. The relationship between the temperature and

compressive strength was estimated as shown in Equation (B-7) and Figure 100.

() = 218.4 (/68.0)-1.695

(R2 = 0.9612)

where is the compressive strength (psi) and T is the temperature (F).

(B-7)

164

FIGURE 100 Empirical TemperatureCompressive Strength Model for Asphalt

For the aging model, various types of functions were tested, as the compressive

strength had a relatively weak correlation with the age compared to the temperature model.

After numerous iterations, a power function was selected to maximize the goodness of fit.

The relationship between age and compressive strength is estimated using Equation (B-8)

and graphically shown in Figure 101:

() = 98.77 0.1788

(R2 = 0.5011)

(B-8)

where is the compressive strength (psi) and t is the age of the asphalt material (day).

165

FIGURE 101 Empirical Age-Compressive Strength Model for Asphalt
B.3 Selection of Equivalent Asphalt Material for Dynamic Testing Due to practical limitations in the research program schedule, the hot mix asphalt
mow strip used in the static test program could not be used for the dynamic testing program--there was not sufficient time in the program schedule to allow the hot mix material to age between test sessions. Hence, it was necessary to select an alternative type of asphalt material with equivalent strength developed in a shorter amount of curing time (e.g., 14 days).
Recent experimental studies have proposed using modified CMA as an alternative to conventional HMA. Niazi et al. [38] tested three additives (Portland cement, lime slurry, and hydrated lime) for increasing the strength of CMA; they found that adding Portland cement at a level of 2% by weight increased the indirect tensile strength by approximately
166

70%. Xu et al. [39] tested CMA with Portland cement added; they found a strong linear relationship between the amount of cement added and the resulting strengths (indirect tensile and flexural).
Based on this previous work, CMA supplemented with Portland cement was selected as an alternative asphalt mix for mow strip construction in the dynamic test program. A series of compression tests on CMA specimens was performed to determine the mixing ratio of Portland cement that would result in compressive-strength values approximately equivalent to that seen in the static test program in a much shorter curing period. Using Equation (B-8), the target compressive strength was set to 232 psi, which represents the value of the stiffest mow strip among all the performed static tests (the compressive strength of 118-day-old HMA). In this study, four levels of cement content (4, 6, 8, and 10% by total weight) and two levels of aging (7 and 14 days) were evaluated.
Cylindrical CMA specimens with approximately 4-in. diameter and 4-in. height were prepared according to AASHTO T 180 [26]. The CMA aggregates, Portland cement (Type I), and the minimum required amount of water were mixed together using a mechanical mixer. After a series of trial mix iterations, a w/c ratio of 0.33, which showed a decent workability in both the mixing and compaction processes, was selected as an optimal value. The mixed material was then moved to metal cylinder molds, which were filled to approximately one-fifth of the cylinder height and compacted using a Proctor compaction hammer. These steps were repeated until the mold was filled completely (Figure 102). The molds were removed 24 hours after the final compaction to mitigate premature cracking or fracture of the CMA specimen.
167

FIGURE 102 Cold Mix Asphalt Specimen Preparation The specimen test protocol throughout the CMA test program was the same as the HMA test program. Figure 103 shows representative pre- and post-test pictures. The compressive strength was calculated from the maximum recorded load divided by the original cross-sectional area of the specimen. Table 20 and Figure 104 present the test results of various CMA mix types.
168

(a)

(b)

FIGURE 103 Unconfined Compression Test Pictures of CMA Specimen: (a) Pre-test; (b) Post-test

TABLE 20 CMA Test Results for Various Mixing Ratios

Asphalt Age (day)
7
14

Cement Content
(%)
4 6 8 10 4 6 8 10

Average Compressive
Strength (psi) 79.1 105.5 142.3 182.6 83.7 126.7 184.3 251.7

No. of Specimens
Tested
2 2 2 2 3 2 2 6

Test Temperature
(F)
72 72 72 69 71 71 71 67

169

FIGURE 104 CMA Test Results: Compressive Strength vs. Cement Content As found by previous researchers [39], a strong linear trend between the strength and the cement content was observed both in the 7- and 14-day results. Specifically, the 14-day compressive strength of CMA supplemented with 10% Portland cement by weight slightly exceeded (252 psi) the target compressive strength of 232 psi. Therefore, this material mixture was selected for use in the construction of mow strips for the dynamic test program as a conservative approximation of the HMA asphalt used in the static test program.
170

APPENDIX C

DETAILED CRASH SIMULATION RESULTS

C.1 Guardrail System without Mow Strip (Soil Only)

FIGURE 105 Simulation Result for the Guardrail System Without Mow Strip Up to 0.3 Sec
171

FIGURE 106 Simulation Result for the Guardrail System Without Mow Strip After 0.3 Sec
172

FIGURE 107 Simulation Result for the Guardrail System Without Mow Strip Up to 0.3 Sec
173

FIGURE 108 Simulation Result for the Guardrail System Without Mow Strip After 0.3 Sec
174

FIGURE 109 Simulation Result for the Guardrail System Without Mow Strip
FIGURE 110 Vehicle Deformation for the Guardrail System with Soil Only
175

FIGURE 111 33 ft/s Average Vehicle Longitudinal Acceleration (g) Soil Only
FIGURE 112 Relative Longitudinal Velocity of the Occupant (ft/s) Soil Only
176

FIGURE 113 Relative Longitudinal Displacement of the Occupant (ft) Soil Only
FIGURE 114 33 ft/s Average Vehicle Lateral Acceleration (g) Soil Only
177

FIGURE 115 Relative Lateral Velocity of the Occupant (ft/s) Soil Only
FIGURE 116 Relative Lateral Displacement of the Occupant (ft) Soil Only
178

FIGURE 117 Vehicle Roll for the Setup with Soil Only (deg)
FIGURE 118 Vehicle Pitch for the Setup with Soil Only (deg)
179

FIGURE 119 Vehicle Yaw for the Setup with Soil Only (deg)
180

C.2 Guardrail System with Asphalt Mow Strip with 2-in. Thickness and 24-in. Rear Distance (Test T2-R24)
FIGURE 120 Simulation Result for Test T2-R24 Up to 0.3 Sec
181

FIGURE 121 Simulation Result for Test T2-R24 After 0.3 Sec
182

FIGURE 122 Simulation Result for Test T2-R24 Up to 0.3 Sec
183

FIGURE123 Simulation Result for Test T2-R24 After 0.3 Sec
184

FIGURE 124 Simulation Result for Test T2-R24
FIGURE 125 Vehicle Deformation for Test T2-R24
185

FIGURE 126 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T2-R24
FIGURE 127 Relative Longitudinal Velocity of the Occupant (ft/s) Test T2-R24
186

FIGURE 128 Relative Longitudinal Displacement of the Occupant (ft) Test T2-R24
FIGURE 129 33 ft/s Average Vehicle Lateral Acceleration (g) Test T2-R24
187

FIGURE 130 Relative Lateral Velocity of the Occupant (ft/s) Test T2-R24
FIGURE 131 Relative Lateral Displacement of the Occupant (ft) Test T2-R24
188

FIGURE 132 Vehicle Roll (deg) Test T2-R24
FIGURE 133 Vehicle Pitch (deg) Test T2-R24
189

FIGURE 134 Vehicle Yaw (deg) Test T2-R24
190

C.3 Guardrail System with Asphalt Mow Strip with 3.5-in. Thickness and 12-in. Rear Distance (Test T3.5-R12)
FIGURE 135 Simulation Result for Test T3.5-R12 Up to 0.3 Sec
191

FIGURE 136 Simulation Result for Test T3.5-R12 After 0.3 Sec
192

FIGURE 137 Simulation Result for Test T3.5-R12 Up to 0.3 Sec
193

FIGURE 138 Simulation Result for Test T3.5-R12 After 0.3 Sec
194

FIGURE 139 Simulation Result for Test T3.5-R12
FIGURE 140 Vehicle Deformation for Test T3.5-R12
195

FIGURE 141 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R12
FIGURE 142 Relative Longitudinal Velocity of the Occupant (f/s) Test T3.5-R12
196

FIGURE 143 Relative Longitudinal Displacement of the Occupant (ft) Test T3.5-R12
FIGURE 144 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R12
197

FIGURE 145 Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R12
FIGURE 146 Relative Lateral Displacement of the Occupant (ft) Test T3.5-R12
198

FIGURE 147 Vehicle Roll (deg) Test T3.5-R12
FIGURE 148 Vehicle Pitch (deg) Test T3.5-R12
199

FIGURE 149 Vehicle Yaw (deg) Test T3.5-R12
200

C.4 Guardrail System with Asphalt Mow Strip with 3.5-in. Thickness and 24-in. Rear Distance (Test T3.5-R24)
FIGURE 150 Simulation Result for Test T3.5-R24 Up to 0.3 Sec
201

FIGURE 151 Simulation Result for Test T3.5-R24 After 0.3 Sec
202

FIGURE 152 Simulation Result for Test T3.5-R24 Up to 0.3 Sec
203

FIGURE 153 Simulation Result for Test T3.5-R24 After 0.3 Sec
204

FIGURE 154 Simulation Result for Test T3.5-R24
FIGURE 155 Vehicle Deformation for Test T3.5-R24
205

FIGURE 156 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24
FIGURE 157 Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24
206

FIGURE 158 Relative Longitudinal Displacement of the Occupant (ft) Test T3.5-R24
FIGURE 159 10 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24
207

FIGURE 160 Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24
FIGURE 161 Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24
208

FIGURE 162 Vehicle Roll (deg) Test T3.5-R24
FIGURE 163 Vehicle Pitch (deg) Test T3.5-R24
209

FIGURE 164 Vehicle Yaw (deg) Test T3.5-R24
210

C.5 Guardrail System with Asphalt Mow Strip with 6-in. Thickness and 24-in. Rear Distance (Test T6-R24)
FIGURE 165 Simulation Result for Test T6-R24 Up to 0.3 Sec
211

FIGURE 166 Simulation Result for Test T6-R24 After 0.3 Sec
212

FIGURE 167 Simulation Result for Test T6-R24 Up to 0.3 Sec
213

FIGURE 168 Simulation Result for Test T6-R24 After 0.3 Sec
214

FIGURE 169 Simulation Result for Test T6-R24
FIGURE 170 Vehicle Deformation for Test T6-R24
215

FIGURE 171 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T6-R24
FIGURE 172 Relative Longitudinal Velocity of the Occupant (ft/s) Test T6-R24
216

FIGURE 173 Relative Longitudinal Displacement of the Occupant (ft) Test T6-R24
FIGURE 174 33 ft/s Average Vehicle Lateral Acceleration (g) Test T6-R24
217

FIGURE 175 Relative Lateral Velocity of the Occupant (ft/s) Test T6-R24
FIGURE 176 Relative Lateral Displacement of the Occupant (ft) Test T6-R24
218

FIGURE 177 Vehicle Roll (deg) Test T6-R24
FIGURE 178 Vehicle Pitch (deg) Test T6-R24
219

FIGURE 179 Vehicle Yaw (deg) Test T6-R24
220

C.6 Guardrail System with Asphalt Mow Strip with 3.5-in. Thickness and 24-in. Rear Distance and Asphalt Diagonal Pre-Cuts (Test T3.5-R24-C)
FIGURE 180 Simulation Result for Test T3.5-R24-C Up to 0.3 Sec
221

FIGURE 181 Simulation Result for Test T3.5-R24-C After 0.3 Sec
222

FIGURE 182 Simulation Result for Test T3.5-R24-C Up to 0.3 Sec
223

FIGURE 183 Simulation Result for Test T3.5-R24-C After 0.3 Sec
224

FIGURE 184 Simulation Result for Test T3.5-R24-C
FIGURE 185 Vehicle Deformation for Test T3.5-R24-C
225

FIGURE 186 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24-C
FIGURE 187 Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24-C
226

FIGURE 188 Relative Longitudinal Displacement of Occupant (ft) Test T3.5-R24-C
FIGURE 189 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24-C
227

FIGURE 190 Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24-C
FIGURE 191 Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24-C
228

FIGURE 192 Vehicle Roll (deg) Test T3.5-R24-C
FIGURE 193 Vehicle Pitch (deg) Test T3.5-R24-C
229

FIGURE 194 Vehicle Yaw (deg) Test T3.5-R24-C
230

C.7 Guardrail System with Asphalt Mow Strip with 3.5-in. Thickness and 24-in. Rear Distance and a Stiffer Asphalt (Test T3.5-R24-S)
FIGURE 195 Simulation Result for Test T3.5-R24-S Up to 0.3 Sec
231

FIGURE 196 Simulation Result for Test T3.5-R24-S After 0.3 Sec
232

FIGURE 197 Simulation Result for Test T3.5-R24-S Up to 0.3 Sec
233

FIGURE 198 Simulation Result for Test T3.5-R24-S After 0.3 Sec
234

FIGURE 199 Simulation Result for Test T3.5-R24-S
FIGURE 200 Vehicle Deformation for Test T3.5-R24-S
235

FIGURE 201 33 ft/s Average Vehicle Longitudinal Acceleration (g) Test T3.5-R24-S
FIGURE 202 Relative Longitudinal Velocity of the Occupant (ft/s) Test T3.5-R24-S
236

FIGURE 203 Relative Longitudinal Displacement of Occupant (ft) Test T3.5-R24-S
FIGURE 204 33 ft/s Average Vehicle Lateral Acceleration (g) Test T3.5-R24-S
237

FIGURE 205 Relative Lateral Velocity of the Occupant (ft/s) Test T3.5-R24-S
FIGURE 206 Relative Lateral Displacement of the Occupant (ft) Test T3.5-R24-S
238

FIGURE 207 Vehicle Roll (deg) Test T3.5-R24-S
FIGURE 208 Vehicle Pitch (deg) Test T3.5-R24-S
239

FIGURE 209 Vehicle Yaw (deg) Test T3.5-R24-S
240