Citation
Evaluation of Asphalt Layer Interface Resistance to Bond Degradation through Repeated Load

Material Information

Title:
Evaluation of Asphalt Layer Interface Resistance to Bond Degradation through Repeated Load
Creator:
Waisome, Jeremy Alexis Magruder
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (151 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Civil Engineering
Civil and Coastal Engineering
Committee Chair:
ROQUE,REYNALDO
Committee Co-Chair:
TIA,MANG
Committee Members:
HILTUNEN,DENNIS R
LINDNER,ANGELA S

Subjects

Subjects / Keywords:
asphalt -- bond -- breakdown -- interface -- pavement
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Civil Engineering thesis, Ph.D.

Notes

Abstract:
One of the most prevalent forms of pavement distress is top-down cracking. There is incomplete understanding of how this type of near-surface cracking occurs. Recent attention to the importance of pavement interface bond suggests that loss of bond may be a primary contributor to this problem. The vast majority of previous research focused on evaluating near-surface cracking through bond strength tests, which predominately relates to debonding of the uppermost surface layers. However, research suggested the potential for a different mechanism of debonding that relates to repeated shear along an interface located lower in the pavement system. The primary objective of this study was to identify a method to compare relative performance of various interface conditions with the results of an existing bond strength device. Using the newly proposed method, this research evaluated the effect tack coats at the interface when subjected to repeated application of shear stress. An existing interface shear strength test device developed by the Louisiana State University was modified to allow for repeated load testing. A testing scheme called the repeated shear test (RST) method and data collection and interpretation methods were developed to properly analyze the results. Three tack coats were evaluated including a trackless tack coat, a conventional tack coat and a polymer modified asphalt emulsion (PMAE). As a reference, no tack at the interface was also considered. Two application rates of the conventional tack coat were investigated to evaluate the importance of application rate. An elastic layer analysis indicated the location of critical stress states within the pavement system that could lead to debonding at more structural interfaces. Strength test data indicated trackless tack, no tack, and low application rate of the conventional tack perform similarly, reaching on average 126.5, 126.6 and 126.4 psi, respectively. The results of the shear strength tests were consistent with the literature suggesting that brittle systems perform best. However, when these materials were tested using the RST method, trackless tack and no tack specimens were among the worst performers. Using an energy analysis, the results of the two test methods were compared. Generally, the rankings provided by each analysis were reversed, with exception to the conventional tack coats. Overall the results support the hypothesis of this study, which is that monotonic testing does not capture the resistance to loss of bond over time. This type of repeated shear testing is needed to identify the best material for use at structural interfaces. The results show that non-brittle systems may perform best in the conditions found in these critical locations, deeper in the pavement system. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2017.
Local:
Adviser: ROQUE,REYNALDO.
Local:
Co-adviser: TIA,MANG.
Statement of Responsibility:
by Jeremy Alexis Magruder Waisome.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
LD1780 2017 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

EVALUATION OF ASPHALT LAYER INTERFACE RESISTANCE TO BOND DEGRADATION THROUGH REPEATED LOAD By JEREMY ALEXIS MAGRUDER WAISOME A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLM ENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2017

PAGE 2

201 7 Jeremy Alexis Magruder Waisome

PAGE 3

To God be the glory forever and ever. Amen.

PAGE 4

4 ACKNOWLEDGMENTS I would lik e to thank my parents who inspired me to reach further and higher regard less of the obstacles in my way They kept me grounded in my faith in Jesus To my brother who made me laugh, even when I did not want to, I appreciate you. I would also like to thank my husband who always believed in me, and sacrificed his time and energy to ensure I would succeed. Thank you also to my mentor, Dr. Jonathan F. K. Earle, for his encouragement since high school. Thank you to my friends who encouraged me. Thank you from the depths of my soul. For anyone who has ever had a childhood dream and was told it could not be done and persisted in spite of adversity, this is for you.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 1.1 Background ................................ ................................ ................................ ....... 15 1.2 Hypothesis ................................ ................................ ................................ ........ 17 1.3 Objecti ves ................................ ................................ ................................ ......... 17 1.4 Scope ................................ ................................ ................................ ................ 18 1.5 Research Approach ................................ ................................ .......................... 19 2 LITERATURE REVIEW ................................ ................................ .......................... 21 2.1 Background ................................ ................................ ................................ ....... 21 2.2 Interface Bond Testing ................................ ................................ ...................... 23 2.2.1 Interf ace Bond Strength Tests ................................ ................................ 23 2.2.2 Interface Bond Fatigue Tests ................................ ................................ .. 34 2.3 Summary ................................ ................................ ................................ .......... 41 3 STRESS ANALYSIS ................................ ................................ ............................... 44 3.1 Identification of Critical Stress States ................................ ................................ 44 3.2 Elastic Layer Analysis ................................ ................................ ....................... 44 3.2.1 KENLAYER Data Analysis ................................ ................................ ...... 45 3.2.2 Critical Zone for the Onset of Debonding ................................ ................ 49 3.2.3 Traffic Wander, Tire Size and the Critical Zone ................................ ....... 53 3.3 Summary ................................ ................................ ................................ .......... 54 4 MATERIALS AND SPECIMEN PREPARATION ................................ ..................... 56 4.1 Materials ................................ ................................ ................................ ........... 56 4.2 Specimen Preparation ................................ ................................ ...................... 58 4.2.1 Batching and Mixing ................................ ................................ ................ 58 4.2.2 Initial Compaction ................................ ................................ .................... 60 4.2.3 Cutting ................................ ................................ ................................ ..... 60 4.2.4 Tack Coat Appl ication ................................ ................................ .............. 61 4.2.5 Final Compaction ................................ ................................ .................... 64

PAGE 6

6 4.2.6 Determining Air Void Content ................................ ................................ .. 65 5 TESTING DEVELOPMENT AND DATA INTERPRETATION ................................ 66 5.1 Changes in Florida ................................ ................................ ............................ 66 5.2 Florida Interface Shear Tester ................................ ................................ .......... 68 5.3 Energy Losses and Their Solutions ................................ ................................ .. 72 5.3.1 Gap Width ................................ ................................ ............................... 73 5.3.2 Bo nd Strength Testing ................................ ................................ ............. 77 5.3.3 Complications with Repeated Testing Modes ................................ .......... 79 5.3.3.1 Data Interpretation ................................ ................................ ...... 81 5.3.3.2 Progressive LC RST ................................ ................................ ... 87 5.3.3.3 Displacement controlled RST ................................ ..................... 90 5.3.4. Discussi on on the Solutions for Energy Loss ................................ ......... 94 5.4 Final Repeated Shear Stress Testing ................................ ............................... 97 5.4.1 Energy Analysis ................................ ................................ ..................... 101 5.4.2 Statistical Analysis ................................ ................................ ................. 105 5.5 Device Limitations ................................ ................................ ........................... 109 6 RESULTS AND CONCLUSIONS ................................ ................................ ......... 110 6.1 Summary and Findings ................................ ................................ ................... 110 6.2 Recommendations ................................ ................................ .......................... 112 APPENDIX A ASPHALT MIXTURE INFORMATION ................................ ................................ .. 114 B OVERLAY MATERIAL CALCULATION ................................ ................................ 116 C APPLICATION RATE CALCULATION ................................ ................................ .. 117 D SHEAR STRENGTH TEST DATA ................................ ................................ ........ 118 E REPEATED SHEAR STRESS TEST DATA ................................ ......................... 122 F FINAL LC RST DA TA ................................ ................................ ........................... 127 G ENERGY ANALYSIS ................................ ................................ ............................ 132 H STATISTICAL ANALYSIS ................................ ................................ ..................... 138 LIST OF REFERENCES ................................ ................................ ............................. 146 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 151

PAGE 7

7 LIST OF TABLES Table page 2 1 Recommended re sidual application rates of tack coat. ................................ ...... 34 3 1 Layer thickness and material properties. ................................ ............................ 47 5 1 Typical deformation obtained from Microsoft Excel analysis .............................. 84 5 2 Typical stiffness obtained from Microsoft Excel analysis ................................ .... 85 5 3 Percent of measured bond strengt h reached with applied RST stress. .............. 88 5 4 Stiffness reduction of LC RST specimens. ................................ ......................... 98 5 5 Confidence interval data for LC RST en ergy analysis. ................................ ..... 108 A 1 Dense gradation for base job mix formula (JMF). ................................ ............. 114 A 2 Dense gradation for base (cumulative). ................................ ............................ 114 A 3 Dense gradation for overlay (cumulative). ................................ ........................ 115 A 4 Volumetrics for GA I2 granite. ................................ ................................ .......... 115 A 5 Specific gravity for GA I2 materials. ................................ ................................ 115 G 1 Inflection points for each LC RST specimen. ................................ .................... 137 H 1 LC RST ANOVA summary. ................................ ................................ .............. 139 H 2 LC RST ANOVA ................................ ................................ ............................... 139 H 3 F Test two sample for variances: NT to TT. ................................ ..................... 139 H 4 F Test two sample for variances: NT to CL. ................................ ..................... 140 H 5 F Test two sample for variances: NT to CH. ................................ ..................... 140 H 6 F Test two sample for variances: NT to PT. ................................ ..................... 140 H 7 F Test two sample for variances: TT to CL. ................................ ...................... 140 H 8 F Test two sample for variances: TT to CH. ................................ ..................... 141 H 9 F Test two sample for variances: TT to PT. ................................ ...................... 141 H 10 F Test two sample for variances: CL to CH. ................................ ..................... 141

PAGE 8

8 H 11 F Test two sample for variances: CL to PT. ................................ ..................... 141 H 12 F Test two sample for variances: CH to PT. ................................ ..................... 142 H 13 T Test two sample assuming equal variances: NT to TT. ................................ 142 H 14 T Test two sample assuming equal variances: NT to CL. ................................ 142 H 15 T Test two sample assuming unequal variances: NT to CH. ............................ 143 H 16 T Test two sample assuming equal variances: NT to PT. ................................ 143 H 17 T Test two sample assuming equal variances: TT to CL. ................................ 143 H 18 T Test two sample assuming unequal variances: TT to CH. ............................ 144 H 19 T Test two sample assuming equal variances: TT to PT. ................................ 144 H 20 T Test two sample assuming unequal variances: CL to CH. ............................ 144 H 21 T Test two sample assuming equal variances: CL to PT. ................................ 145 H 22 T Test two sample assuming unequal variances: CH to PT. ............................ 145

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Surface failure due to poor interface bond in asphalt pavements ....................... 16 1 2 Research approach flow chart. ................................ ................................ ........... 20 2 1 Direct shear test with normal load schematic ................................ ..................... 25 2 2 FDOT shearing apparatus ................................ ................................ .................. 27 2 3 Schematics of the a) ASTRA test device and b) LPDS test device .................... 29 2 4 Schematic of the NCAT Bond Stre ngth device. ................................ .................. 30 2 5 Schema tic of the LISST device ................................ ................................ ........... 31 2 6 Schematic of shear fatigue test ................................ ................................ .......... 35 2 7 Specimen in the MCS device. ................................ ................................ ............. 36 2 8 Schematic (left) and photograph (right) of the DST device ................................ 38 2 9 Image of the SISTM device. ................................ ................................ ............... 39 2 10 Schematic of the SDSTM ................................ ................................ .................. 41 3 1 Typical cross section of a conventional flexible pavement ................................ 46 3 2 Structural characteristics of evaluated pavement sections (layer thickness in parenthesis) ................................ ................................ ................................ ........ 49 3 3 Horizontal shear stress through the depth of the 4.0 inch AC layer cases. ........ 50 3 4 Horizontal shear stress and average vertical stress for the 4.0 inch AC layer cases at a depth of 2.0 inches. ................................ ................................ ........... 50 3 5 Horizontal shear stress through the depth of the 8.0 inch AC layer cases located under the edge of the tire. ................................ ................................ ...... 52 3 6 Horizontal shear stress and a verage vertical stress for the 8.0 inch AC layer cases at a depth of 2.0 inches. ................................ ................................ ........... 52 3 7 Schematic of debonded zone around tire and debonding strip along the tire wheel path within a traffic la ne. ................................ ................................ ........... 53 4 1 GA I2 dense graded aggregate mixture. ................................ ............................ 56

PAGE 10

10 4 2 Materials and mixing tools inside of the oven ................................ ..................... 59 4 3 Mechanical mixer ................................ ................................ ................................ 59 4 4 Device used to produce SGC pills ................................ ................................ ...... 60 4 5 Water cooled masonr y saw prior to cutting a specimen ................................ .... 61 4 6 Heater with graduate d cylinder containing tack coat ................................ .......... 6 2 4 7 Brush used to apply tack co at ................................ ................................ ............. 63 4 8 Tack coat being applied to the bottom half of a specimen on a scale. ................ 63 4 9 SGC mold with base insertion in progress ................................ .......................... 64 4 10 Determination of number of gyrations to compact the overlay. ........................... 65 5 1 Image of the FIST ................................ ................................ ............................... 69 5 2 Image of FIST mounted on the MTS system ................................ ...................... 70 5 3 Close up view of a specimen loaded in the FIST. ................................ ............... 70 5 4 Two specimens failed with different gap widths on the FIST .............................. 73 5 5 Collar insert that shows the beveled edge on the FIST. ................................ ..... 74 5 6 Comparison of shear stress response of a pill and specimen with no tack. ........ 76 5 7 Average of the maximum shear stress for various tack coats. ........................... 78 5 8 Repeated loading period. ................................ ................................ ................... 80 5 9 Typical raw data for repeated loading versus time. ................................ ............ 82 5 10 Typical raw 6 cycle data of axial displacement versus time. .............................. 82 5 11 Progression of displacement for five acquisitions of 6 cycle data versus time. .. 83 5 12 Typical full data set of axial displacement data. ................................ .................. 83 5 13 Normalized stiffness for 1500 lbf LC RST NT specimen. ................................ ... 85 5 14 Loads used for progressive loading method. ................................ ...................... 88 5 15 Repeated applied deformation schematic. ................................ ......................... 91 5 16 Load reduction for typical displacement controlled RST evaluation. ................... 92

PAGE 11

11 5 17 Stiffness reduction for typical displacement controlled RST evaluation. ............. 92 5 18 Displacement controlled RST normalized stiffness results for all specimens. .... 93 5 19 Curved disc spring washers stacked in series. ................................ ................... 95 5 20 Typical resilient deformation data for final LC RST evaluation. ......................... 98 5 21 Typical stiffness data for final LC RST evaluation. ................................ ............. 99 5 22 Typical load data from LC RST evaluation. ................................ ........................ 99 5 23 Typical total deformation data from LC RST evaluation. ................................ .. 100 5 24 Typical permanent deformation data for final LC RST evaluation. ................... 100 5 25 Identifying permanent deformation data from a typical curve. .......................... 102 5 26 Graphical representation of energy calculation of typical strength test data. .... 105 D 1 Shear stress of specimens with no tack coat. ................................ ................... 119 D 2 Shear stress of specimens with trackless tack coat. ................................ ......... 119 D 3 Shear stress of specimens with PMAE tack coat. ................................ ............. 120 D 4 Shear stress of specimens with conventional tack coat applied at a low rate. .. 120 D 5 Shear stress of specimens with conventional tack coat applied at a high rate. 121 E 1 Trial LC RST resilient deformation data at 1500.0 lbf. ................................ ...... 122 E 2 Trial LC RST stiffness data at 1500.0 lbf. ................................ ......................... 122 E 3 Trial LC RST normalized stiffness data at 1500.0 lbf. ................................ ...... 123 E 4 8 hour LC RST progressive loading trial resilient deformation data. ................. 123 E 5 8 hour LC RST progressive loading trial stiffness data. ................................ .... 124 E 6 8 hour LC RST progressive loading trial normalized stiffness data. ................. 124 E 7 12 hour LC RST progressive loading trial resilient deformation data. ............... 125 E 8 12 hour LC RST progressive loading trial s tiffness data. ................................ .. 125 F 1 Permanent deformation data from LC RST specimen NT 1. ............................ 127 F 2 Permanent deformation data from LC RST spec imen NT 2. ............................ 127

PAGE 12

12 F 3 Permanent deformation data from LC RST specimen TT 1. ............................ 128 F 4 Permanent deformation data from LC RST specimen TT 2. ............................ 128 F 5 Permanent deformation data from LC RST specimen CTL 1. .......................... 129 F 6 Permanent deformation data from LC RST specimen CT L 2. .......................... 129 F 7 Permanent deformation data from LC RST specimen CTH 1. ......................... 130 F 8 Permanent deformation data from LC RST specimen CTH 2. ......................... 130 F 9 Permanent deformation data from LC RST specimen PT 1. ............................ 131 F 10 Permanent deformation data from LC RST specimen PT 2. ............................ 131 G 1 Typical slope of LC RST permanent deformation data for initial point. ............. 132 G 2 Typical slope of LC RST permanent deformat ion data for final point. .............. 132 G 3 Cubic function fit to steady state of NT 1 permanent deformation data. ........... 133 G 4 Linear regressi on line fit to NT 1 permanent deformation data. ........................ 133 G 5 Energy of LC RST specimen results. ................................ ............................... 134 G 6 Average energy of LC RST spec imen results. ................................ .................. 134 G 7 Energy of strength test specimen results. ................................ ......................... 135 G 8 Average energy of strength specimen results. ................................ ................. 135 G 9 Line of equality for energy data. ................................ ................................ ....... 136 H 1 Standard deviation of averaged LC RST results. ................................ ............. 138 H 2 90 percent confidence interval of averaged LC RST results. ........................... 138 H 3 85 percent confidence interval of averaged LC RST results. ........................... 139

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVALUATION OF ASPHALT LAYER INTERFACE RESISTANCE TO BOND DEGRADATIO N THROUGH REPEATED LOAD By Jeremy Alexis Magruder Waisome December 2017 Chair: Reynaldo Roque Cochair: Mang Tia Major: Civil Engineering One of the most prevalent forms of pavement distress is top down cracking. There is incomplete understanding of how this type of near surface cracking occurs. Recent attention to the importance of pavement interface bond suggests that loss of bond may be a primary contributor to this problem. The vast majority of previous research focused on evaluating near surface cra cking through bond strength tests, which predominately relates to de bonding of the uppermost surface layer s However, research suggest ed the potential for a different mechanism of de bonding that relates to repeated shear along an interface located lower in the pavement system The primary objective of this study was to identify a method to compare relative performance of various interface conditions with the results of an existing bond strength device Using the newly proposed method, this research evaluate d the effect of repeated application of shear stress An existing interface shear strength test device developed by the Louisiana State University was significantly modifi ed to allow for repeated load testing. A testing scheme called the repeated shear test (RST) method

PAGE 14

14 and data collection and interpretation methods were developed to properly analyze the results Three tack coats were evaluated including a trackless tack c oat, a conventional tack coat and a polymer modified a sphalt emulsion (PMAE) As a reference, no tack at the interface was also considered. Two application rates of the conventional tack coat were investigated to evaluate the importance of application rate An elastic layer analysis indicated the location of critical stress states within the pavement system that could lead to debonding at more structural interfaces. Strength test d ata indicated trackless tack no tack and low application rate of the conve ntional tack perform similarly, reaching on average 126.5, 126.6 and 126.4 psi, respectively. The results of the shear strength tests were consistent with the literature suggesting that brittle systems perform best. However, when these materials wer e teste d using the RST method, trackless tack and no tack specimens were among the worst performers Using an energy analysis, the results of the two test methods were compared. Generally, the rankings provided by each analysis were reversed, with exception to th e conventional tack coats. O verall the results support the hypothesis of this study, which is that monotonic testing does not capture the resistance to loss of bond over time. This type of repeated shear testing is needed to identify the best materi al f or use at interfaces located deeper in the pavement system The results show that non brittle systems may perform best in the conditions found in these critical locations, deeper in the pavement system.

PAGE 15

15 CHAPTER 1 INTRODUCTION 1.1 Background Top down cracking has been repo r t ed a s a mod e of pavement distress since the late 1990s I nitial understanding of this mechanism of cracking originated as an observation of surface cracking in the late 1980s (Dauzats and Rampal, 1987), and has been increasingly studied in a concerted effort to reduc e its occurrence Top down cracking manifests itself as near surface, longitudinal cracks along the wheel path. Numerous reports of this mechanism of failure exist from across the United States (US) and other countries. In the past, the Florida Department of Transportation (FDOT) reported that top down cracking was the primary distress mode for 90% of the roads scheduled for resurfacing ( Myers and Roque 200 2 ). Even today, the state of Florida considers top down cracking as the most common cause of pavement rehabilitation Recent attention to the importance of interface bond suggests that loss of bond may be a primary contributor to near surface cracking R esearchers have focu sed on assess ing interface bond primarily by way of bond strength testing These te sts focus on evaluating the bond between the upper most layer of pavement and the pav ement beneath. At these surface level interfaces, shear stresses induced by braking are considered the most critical. At the surface, vehicle braking can ultimately lead to slippage sliding and delamination Slippage is characterized by a crescent shaped crack along the wheel path as shown in Figure 1 1 D elamination occurs when sections of the overlay are completely separate d from the structure below Th ese failure modes a re often caused

PAGE 16

16 by a single load excursion to failure that monotonic bond strength test s may capture well. Figure 1 1. Surface failure due to poor interface bond in asphalt pavements Source: West, R. C., J. Zhang, and J. Moore. Evaluation of Bon d Strength between P avement L ayers National Center for Asphalt Technology 2005, pp. 1 58. Recent field observations revealed that near surface longitudinal cracking is present when there is poor bo nd between the pavement layers The debonding identified was found at structural interfaces located deeper in the pavement system, and was connected to the failure at the surface by vertical cracks (Willis and Timm, 2007) Identifying which of these failures at the surfac e or structural level manifest itself fir st is nearly impossible. However, there is the potential for this type of debonding to contribute to the aforementioned surface failure s, by introducing discontinuities within the pavement structure. P revious research efforts have not consider ed interfaces located deeper in the pavement structure, where the shear stress es are possibly lower than values measured by bond strength test s I t is necessary to determine whether or not a high shear stress

PAGE 17

17 value, such as those resulting from monotonic bond strength tests, is representative of the failure loads that cause debonding in these deeper pavement interface locations. In these area s potentially low levels of vertical confinement and the repeated application of lower magnitude shear stresses m ight contribute to accelerated failure at the interface by breaking down the bond between pavement layers over time As a result, a new testing methodology is necessary to study this type of mechanism. This research addresses th is potential pavement distress mechanism for interface bond breakdown 1.2 Hypothesis Monotonic bond strength tests are unable to capture interface bond resistance to breakdown caused by the repetitive application of lower magnitude shear stre sses in areas of low confinement. As a result, performanc e of the materials used at the interface is improperly assess ed when based primarily on the outcomes of bond strength test ing 1.3 Objectives The primary objective of this study was to determine whether the results from monotonic bond strength tests provid e an accurate reflection of interfacial bond resistance to breakdown under repeated load. Detailed objectives to identify and evaluat e interface bond breakdown are summarized in the following statements: Determine the stress states (magnitude and extent) a t critical l ocations where pavement interfaces are typically located Develop a test that can allow for repeated application of shear stress to assess the relative performance of interface bond conditions using stress states identified in the previous objec tive Evaluate the effects of tack coat type and application rate on the resistance to interface bond breakdown

PAGE 18

18 Compare the relative performance of each interface condition resulting from monotonic and repeated shear stress tests 1.4 Scope In order to dev elop testing protocols for the Repeated Shear Test (RST) one Superpave dense graded mixture was selected A conventional asphalt binder PG 6 7 22 was used for the mixture i n addition to a commonly util ized Georgia granite gradation. Four types of interfa ce bonding conditions were examined: a polymer modified asphalt emulsion (PMAE ) developed by Road Science; a conventional SS 1, created by Ergon Asphalt & Emulsions, Inc.; a widely used trackless tack NTSS 1 HM, produced by Blacklidge Emulsions; and no ta ck (i.e. no tack coat was applied). For PMAE and trackless tack, one application rate was examined, as recommended by FDOT specification. For the conventional tack material two applic ation rates were selected (low and hi gh) based on previous research. A modified version of the Louisiana Interlayer Shear Strength Test was chosen to evaluate interface bond strength and resistance to bond breakdown The device was chosen due to its extensive use in the National Cooperative Highway Research Program (NCHRP) Re port 712, Optimization of Tack Coat for HMA Placement, which provides a baseline of measurements that were initially used to evaluate the accuracy of the bond strength test results. It is important to note that surface conditions, i.e ., milled vs. compacte d, clean vs. dirty, wet vs. dry, can influence interface bond. Only clean, dry, laboratory compacted surfaces were evaluated in this work. All tests we re conducted at room temperature, w hich was measured and recorded.

PAGE 19

19 1.5 Research Approach Thi s research p rimarily focused on the comparison of results of various interface bond types and application rates ( described in s ection 1.4 ) from an existing interface bond strength test method to those obtained from a newly developed method designed to evaluate the res istance of interface bond breakdown The overall approach taken to meet the objectives of this study includes : Literature review: investigate existing and potential mechanisms of near surface cracking evaluate factors that impact the performance of the as phalt pavement interface and identify existing test methods / devices for evaluating both shear strength and interface bond breakdown. Choose an appropriate device that can accommodate both strength and repeated load testing or identify the characteristics of one for modification to do both testing modes St ress state analysis : u tilize an elastic layer analysis program to determine the critical locations and stress states in areas where structural interfaces can occur within asphalt pavement S hear strength testing : evaluate the effect of application rate and tack coat on interface bond and compare the results to those found in existing literature for quality control purposes. Determine whether or not additional adjustments to the device are necessary for te sting. Repeated shear test method development: d evelop or modify an existing strength test to allow for repeate d application of shear stress at the interface This includes determination of the appropriate (1) loading mode (load controlled vs. deformation controlled), (2) loading procedure which includes the magnitude, duration and load shape and (3) device configuration Determine an appropriate associated data interpretation method for evaluating the effect of repeated shear stress on interfacial bond. T his requires not only assuring reasonable testing and data interpretation time, but also the adequate frequency of data acquisition. It will also include the modification of the device to eliminate any errors that affect the results of the chosen test meth od. This iterative process is indicated in Figure 1 2 Repeated shear testing: evaluate the various interface bond types and application rates using the method determined from the previous step. Data analysis: rank the results of each test to compare relat ive results E valuate the effect of application rate and tack coat on interface bond.

PAGE 20

20 Figure 1 2 Research approach flow chart Perform trial shear strength tests. Compare results with literature and the effects of gap width Perform trial repeated load tests with load level based on the shear stress values found in pavement analysis No Yes Perform final testing for both shear strength and repeated load tests Evaluate results and compare bond strength test results to repeated load test results Use optimum load level(s) in future analysis Adjust loading sche me Does selected load level allow the specimen to reach the failure limit within the selected time frame? Is the relative performance the same? No Yes Strength test is suitable Strength test is not suitable

PAGE 21

21 CHAPTER 2 LITERATURE REVIEW 2.1 Background Roadways ar e layered systems designed to work as a monolithic structure to carry traffic loads. The uppermost portion of this system is typically composed of layers of asphalt concrete (AC). Where two adjacent AC layers meet is referred to as an interface. Interfaces are inherent components of our pavement systems, created during both new construction and pavement rehabilitation. Without proper bond between adjacent pavement layers, the magnitude and location of the critical responses (stress and strain) varies from a fully bonded system ( Willis and Timm, 2007) A number of factors influence the quality of the bond between pavement layers including tack coat type, application rate, temperature, surface conditions, moisture, pavement structure, construction methods and more. This makes predicting the performance and longevity of the pavement structure challenging. However, previous research indicates that ensuring proper bond at these potential d iscontinuities is paramount to good pavement performance. Tack coats are app lied to the underlying pavement surface (new or milled) prior to the placement of a new layer of hot mix asphalt (HMA) to enhance interface bond and improve the structural performance of pavements. This material help s to fully integrate the adjacent layers into one system. In current practice, tack coats are often virgin asphalt binder, or an asphalt emulsion (emulsified asphalt). According to a recent worldwide study, the most widely used method to bond adjacent pavement layers is by applying emulsified as phalt tack coat (Mohammad et al. 2012) Emulsified asphalt is asphalt binder suspended in water by way of an emulsifying agent, like soap. When

PAGE 22

22 combined, the asphalt binder disperses into small droplets within the water. After being applied to the pavemen t surface, the emulsion breaks, leaving behind an asphalt binder residue. Tack coats are applied in a thin layer by a sprayer attached to a distributor vehicle. In the US tack coat application rates are determined by state Department s of Transportation (D OTs) based on characteristics of the existing pavement surface Application rates are often described by the percent of asphalt binder residue required to remain on the surface Tack coats can be applied at higher rates to allow the material to infiltrate the void systems of the adjacent pavement layers. However, applying tack coats at higher rates can lead to construction issues like tracking or bleeding According to Mohammad et al. 2012, it is best to apply tack coat on a dry, clean pavement surface, to achieve optimal bond. Contractors may also choose not to use a bonding agent, which is often done in new construction if it is not required by the state R esearchers have used several methods to evaluate pavement interface bond. Torque tests, pull off test s, shear strength tests, direct tension tests, etc. are all met hods used in previous research (Mohammad et al. 2012 ; Tran et al. 2012) Most researchers use a unique device specifically designed to apply one of the aforementioned loading schemes to assess bond strength A review of methods used for the evaluation of pavement interface bond in bo th field and laboratory studies is presented in the following sections Though many test devices and loading methods exist in the litera ture, the focus of this revi ew wa s on interface bond strength and fatigue testing. Emphasis was placed on identifying

PAGE 23

23 potential fact or s that cause premature bond failure, including material properties, application rates, construction conditions and environmental conditions 2. 2 Inter face Bond Testing T he evaluation of pavement interface bond is typically c onducted using bond strength test ing devices. These tests are typically designed to assess whether adequate shear strength is present to resist stresses induced by traffic loading. T he principal technique used in these tests involves the application of a single load excursion applied at a designated rate until the specimen fails. However, each device is unique in construction and the loading technique employed. While most bond tests u se a single load excursion of shear stress, there are several variations between tests including specimen size and geometry, loading rate, and more. In addition to shear bond tests, pull off tests and torsional shear tests are utilized to assess interface bond. R esearchers believe that a variety of failure mechanisms, including tension, torsion and/or shear modes occur due to traffic loading. 2. 2 .1 Interface Bond Strength Tests Uzan et al. (1978) conducted one of the first reported studies to better under stand pavement interface bond. This study used a shear test to assess the bond strength variations due to temperature, tack coat application rate, vertical pressures and asphalt binder types Laboratory compacted specimens were tested at 25 .0 C, and 55 .0 C (77 .0 F and 131 .0 F) in constant deformation at 2.5 mm/min (0.1 in/min) T he apparatus used measured deformation from deflectometers. T he study concluded that optimum tack coat rates are influenced by both the binder and the testing temperature. It al so showed that decreasing temperature and increasing vertical pressure enhance d interface bond strength.

PAGE 24

24 In 1997, Hac hiy a and S a to evaluated the bond strength of surface courses typically used at airports to address debonding caused by the horizontal for ce of aircrafts as t hey brake and turn on the pavement. Through the computer analysis program BISAR stresses were calculated at the interface between the wearing and binder courses. Results of the BISAR analysis indicated that failure of the surface cours e is a result of horizontal shear stress and insufficient adhesion at the pavement interface. The authors suggest increased thickness of the wearing course and increased bond strength help to reduce the shear stress at the interface from BISAR results L a boratory tension and flexure tests were conducted to obtain the bond strength for dense graded asphalt concrete specimens cut from laboratory roller compacted specimens with and without tack coat. Tack coats included cationic asphalt emulsions which were developed to counteract the impact of dirt on the surface. Three of the four developed emulsions included rubber. In addition to determining bond strength at the interface for the various tack materials, tack coat application rate, curing time, low and hig h temperature, surface preparation (clean or with sand), and lift thickness were also evaluated by way of shear and tensile strength tests. Findings suggest ed the introduction of a thick lift as the surface course or the utilization of tack coat to increas e bond strength by improving adhesion B ond strength w as found to be similar for both emulsion types at higher temperatures, while at low temperatures, the rubber modified asphalt emulsions provided more strength at the interface than cationic asphalt emul sions. In addition, t he importance of adequate curing time and the implications it has on bond strength was emphasiz ed

PAGE 25

25 In 2001, Romanoschi and Metcalf sought to characterize asphalt concrete interfaces particularly those at the surface layer The resear ch consisted of two major components : the development of a constitutive model from direct shear testing data and t he development of a shear fatigue test which will be d iscussed in section 2.2.2. Specimens with asphalt to asphalt interfaces cored from the Louisiana Pavement Research Facility site were tested using a direct shear test at a constant normal load Some specimens had 0.1 l/m 2 (0.32 gal/yd 2 ) of tack coat, while others had no tack coat. A schematic of the device used for testing is shown in F igur e 2 1 Figure 2 1. Direct shear test with normal load schematic Source: Romanoschi, S. A., and J. B. Metcalf. Characterization of Asphalt Concrete Layer Interfaces. Transportation Research Record: Journal of the Transportation Research Board No. 1778, 2001, pp. 132 139. A constant displacement rate of 0.2 mm/s (0.008 in/s) was used until 12 .0 mm (0.472 in) of total shear displacement was achieved. Normal stress levels varied fro m 138 .0 276 .0 414 .0 and 522 .0 kPa ( 20 .0 40 .0 60 .0 and 80 .0 psi, respect ively). The gap width over the interface of the specimen was 5 .0 mm (0.197 in) which according to the authors allowed shear force to be applied to the interface and not across the asphalt concrete mixture. Testing was conducted at 15 .0 C, 25 .0 C and 35 .0 C (59 .0

PAGE 26

26 F, 77 .0 F and 95 .0 F) Five cores were teste d for each combination of the variables indicated Using th is test configuration, all specimens failed along the plane of the interface, regardless of the variables involved in testing. Line a r vari able differential transducer (LVDT) sensors measured the displacement at the interface. Shear and normal force and displacement were measured and controlled by an MTS Systems Corporation (MTS) loading frame device. Using a typical shear stress displaceme nt curve, three distinguishable sections of the curve were characterized for the derivation of the interface constitutive model. The curve was described by an initial linear shear section a post failure section and ultimately a fri ction section Parameter s for the post failure section were not computed, and specimens were considered to be completely separated after failure. Thus, the model proposed was a two stage model where in the first stage the shear stress is smaller than the shear strength of the int erface, and once the interface fails the second stage begins. An important finding was that shear stress and displacement were proportional until the shear stress equaled the shear strength, causing the interface to fail. The authors also suggest ed that te sting be conducted at several temperatures to identify temperature dependency of parameters used in interface modeling. Sholar et al. ( 2004 ) investigated the effects of moisture, application rate and the interaction of aggregate at the interface using a d irect shear bond test (Figure 2 2) and procedure they developed The device was a modified version of the Iowa Department of Transportation shearing device for Portland cem ent concrete. The device had a 0.125 inch gap width and was designed to house 4 .0 in ch diameter cylindrical specimens. Specimens were cored from three Florida Department of Transportation (FDOT) field

PAGE 27

27 projects with rates of tack coat application varying from no tack to 0.362 l/m 2 (0.08 gal/ y d 2 ) Figure 2 2. FDOT shearing apparatus So urce: Sholar, G. A., G. C. Page, J. A. Musselman, P. B. Upshaw, and H. L. Moseley. Preliminary Investigation of a Test Method to Evaluate Bond Strength of Bituminous Tack Coats. Journal of the Association of Asphalt Paving Technologists Vol. 73, 2004 Res ults indicated that the presence of water on the surface prior to tack coat application can significantly decrease interface bond strength. This finding substantiated that tack coats should be applied to dry surfaces, and at times when there is no potentia l for rain. Aggregate gradation of the mixtures was also found to directly impact the magnitude of the shear strength at the interface. Coarse gradations produced higher shear strengths than fine gradations. In addition, a milled surface at the interface p roduced the highest shear strengths of the specimens tested. In 2005, Canestrari and Santagata utilized the Ancona shear testing research analysis (ASTRA) device (Figure 2 3 a ) to analyze the effect of temperature and normal stress on the interface ASTRA shear tests were conducted at a constant displacement rate of 2.5 mm/min (0.10 in/min) on 9.2 mm (0.36 in) thick specimens. Specimens either

PAGE 28

28 received no tack coat or a conventional cationic emulsion with 60.0% bitumen. The specimens with tack coat were coa ted with 0.5 kg/m 2 ( 0.11 gal/yd 2 ) of emulsion. Various combinations of normal stress at 1.27, 3.00 or 4.73 kg/cm 2 (18.1, 42.7, 67.3 psi) and testing temperature at +20 .0 C +12.5 C +5 .0 C 2.5 C or 10 C (+68 .0 F +54.5 F +41 .0 F 36.5 F or 50 .0 F ) were evaluated and for each combination, both treated (with tack coat) and untreated (without tack coat) interfaces were tested. The results indicate that as the test temperature was decreased, shear resistance increased. This was expected as the stiffness of the material increased with temperature, regardless of whether or not tack coat was applied at the interface. In addition, the authors suggest that the effectiveness of the tack coat on specimens increases with temperature. Another finding was that the increase of normal stress causes an increase of the maximum shear stress, regardless of whether or not tack coat is applied for all evaluated test temperatures. An additional study conducted by Canestrari et al. ( 2005 ) focused on evaluating vari ous interface behaviors Two test methods were used including the ASTRA device in Figure 2 3 a, as well as the layer parallel direct shear (LPDS) tester Figure 2 3b which was developed by the Swiss Federal Laboratories for Material Testing and Research. Thr ee interface treatments were utilized including a polymer modified cationic emulsion, a conventional cationic emulsion and no tack The specimens with tack coat were coated with 300 .0 g/m 2 (0.0948 gal/ y d 2 ) of residual binder Specimens were cored from doub le layered laboratory compacted slabs produced with two types of dense graded mixtures. The loading rates for each test were 2.5 mm/min (0. 1 in/min) and 50.8 mm/min (2.0 in/min) for the ASTRA and LPDS respectively.

PAGE 29

29 Figure 2 3 Schematics of the a ) ASTRA test device and b ) LPDS test device Source: Canestrari, F., G. Ferrotti, M. Partl, and E. Santagata. Advanced Testing and Characterization of Interlayer Shear Resistance. Transportation Research Record: Journal of the Transportation Research Board No. 1 929, 2005, pp. 69 78. The peak shear stress for the ASTRA device was previously determined to be an idealized linear superposition of stress contributions from the residual friction, inner cohesion, dilatancy and tack coat adhesion Peak shear values from specimens tested using the ASTRA device were compared to peak shear stress from the LPDS device (which is computed from a shear force versus relative displacement diagram). The tests

PAGE 30

30 resulted in similar and comparable ranking s of the shear resistance for the various interface conditions. Results showed that modified emulsions provide d greater interlayer shear stress resistance for each applied normal stress at 20 .0 C (68 .0 F) It also showed that when temperatures reach 40 .0 C (104 .0 F) mixture charac teristics played a more dominate role in peak shear stress than the presence of or type of tack coat. Also in 2005, a study was performed by West et al. at the National C enter for Asphalt Technologies (NCAT) to d evelop a test that measured the bond streng th between pavement layers. This study included both a laboratory and field phase that evaluated a variety of tack coats and application rates for use by the Alabama Department of Transportation (ALDOT). The authors chose to develop a simple shear device ( Figure 2 4 ) that is comparable to devices used by other researchers. S pecimens were tested at 50 .0 F, 77 .0 F and 140 .0 F ( 10 .0 C, 25 .0 C and 60 .0 C ) at a loading rate of 50 .8 mm/min ( 2.0 in/min ) Figure 2 4 Schematic of the NCAT Bond Strength d evice Source: West, R. C., J. Zhang, and J. Moore. Evaluation of Bond Strength between Pavement Layers. National Center for Asphalt Technology 2005, pp. 1 58.

PAGE 31

31 In 2012 the most comprehensive study on the influence of tack coat on pavement was conducted b y the National Cooperative Highway Research Program (Mohammad et al. 2012). A total of 432 specimens were evaluated in this study. Project 9 40 involved identifying the best asphalt binder material and method of tack coat application, application rates, e quipment and procedures used in the asphalt industry. From this study, recommendations for revision were made to the American Association of State Highway and Transportation Officials (AASHTO) to improve existing tack coat related practices. The work consi sted of two phases in which Phase I involved a literature review of current standard practices and Phase II involved laboratory testing and field experiments. Figure 2 5. Schematic of the LISST device Source: Mohammad, L. N., M. A. Elseifi, A. Bae, N. B. Patel, J. Button, and J. A. Scherocman. Optimization of Tack Coat for HMA Placement. National Cooperative Highway Research Program Vol. 712, 2012. Phase II was separated into two distinct studies. One focused on the characterization of the quality of t ack coat using the Louisiana Ta ck Coat Quality Tester (LTCQT). The second involved the development of a new test method to evaluate bond

PAGE 32

32 strength called the Louisiana Interface Shear Strength Test (LISST). This device was comparable to the Superpave Shear Tester, but e asy to use, portable, and reasonable in cost and could be adopted for universal testing machines Both laboratory prepared specimens (used to develop the device) and field cores were produced for this project. Field cores were obtained from pr ojects across the US in a variety of climates and under varied traffic conditions w ere evaluated with this device, w hile researchers at the Louisiana Transportation Research Center (LTRC) Pavement Research Facility constructed full scale pavements for anal ysis. The primary focus was to evaluate the effects of a number of variables on the bonding characteristics of the interface including: Trackless, CRS 1 and SS 1h tack coats; 0.031, 0.062 and 0.155 gal/yd 2 applications rates ; old, new and milled HMA, and P ortland cement concrete (PCC) pavement surfaces ; and swept, not swept, wet and dry pavement surface conditions. Issues with coring specimens without tack coat prevented this condition from being analyzed. A confinement condition was also evaluated at 20 ps i. Testing was conducted at 25 C with three replicates of each specimen. Using a preliminary analysis, a loading rate of 0.1 in/min was chosen to simulate the rate of loading at the interface that would be encountered in the field. Results were presented in terms of the interface shear stress ( ISS ) which is computed as follows in Equation 2 1: ( 2 1 ) Where, ISS=Interface Shear Strength (ksi) ; P ULT =Ultimate load applied to the specimen (lb); and

PAGE 33

33 A=Cross sectional area of the specimen (in 2 ) Statistical analysis indicated that test results had less than a 10% coefficient of variation At the highest application rate (0.155 gal/yd 2 ), all tack coats tested in this study exhibited the highest strength. For the SS 1 and Trackless tack coats, strength increased as application rate increased. Yet for the CRS 1 tack coat, the res ults remained stable at and above the 0.062 gal/yd 2 rate. Results were consistent with or without confinement but dusty and dry conditions made the effect of confinement more distinctive Full mixtures (no interface) were also analyzed and shown to be the best performers in the LISST evaluation with an ISS of 105 psi. Trackless specimens proved to be the best overall performer in comparison, reaching 60% of ISS of the full mixture. In terms of evaluating clean and dusty surfaces, 13 of the 24 cases studied indicated a signi ficant effect on the ISS values with dusty conditions displaying higher ISS values. However, there was no significant effect on the ISS response of the tested materials between wet and dry conditions. Milled HMA surfaces gave the highest ISS values, followed by grooved PCC surfaces. In most cases, old HMA surfaces provided higher ISS values than new HMA surfaces. Additionally, the findings from the surface type analysis were more pronounced with low application rates than higher ones. The authors suspect that microstructure contributes to the surface texture/roughness which is less pronounced with the application of tack coat. Using the results from a finite element analysis, a minimum ISS of 40 psi at 25 C was determined to be acceptable for performance. A clean and dry surface was recommended as well to avoid any negative effects on the bonding at the interface due

PAGE 34

34 to the presence of water. Ultimately the authors provided recommendations for residual application rates to use with various surface types as presented in Table 2 1. Table 2 1 Recommended residual application rates of tack coat Surface Type Residual Application Rate (gal/yd 2 ) New Asphalt 0.035 Old Asphalt 0.055 Milled Asphalt 0.055 Portland Cement Concrete 0.045 These f indings indicate that the LISST was successful at distinguishing between the conditions at the interface. 2. 2 .2 Interface Bond Fatigue Tests Romanoschi and Metcalf were some of the first to document the use of fatigue testing of the AC layer inte rface. The primary focus of their work was to evaluate a constitutive model that relates an interface reaction modulus, K which is the slope of the shear stress displacement curve; shear strength S max; and the friction coefficient (Romanoschi and Metcalf 2001) u sing the results from strength testing In addition they developed a shear fatigue test to determine pavement fatigue resistance to obtain a parameter for validation of their constitutive model As shown in Figure 2 6 the longitudinal axis of the specimen was held at an angle of 25.5 degrees from vertical so the shear stress at the interface would be half of the normal stress. This was to simulate interface conditions beneath the wearing course directly under a wheel load. Thus, results could be used to pr ovide relative comparisons between bond strength and fatigue resistance as they were found for an interface in a

PAGE 35

35 similar location. Four vertical loads were applied in the form of a haversine load using at a frequency of 5.0 Hz. Figure 2 6 Schematic of shear fatigue test Source: Romanoschi, S. A., and J. B. Metcalf. Characterization of Asphalt Concrete Layer Interfaces. Transportation Research Record: Journal of the Transportation Research Board, No. 1778, 2001, pp. 132 139. Testing end ed when permanent shear deformation (PSD) reached 6 .0 mm (0.24 in) at the interface or when a number of cycles of PSD of 6 .0 mm (0.24 in) could be extrapolated from the data. Ultimately their findings indicated that the increase in permanent shear deformation with the numb er of load deformations is linear and the rate of increase is higher for higher stresses. They also showed that all parameters used for the constitutive model were temperature dependent. In 2006 Diakhat et al. used the principles from a s hear fatigue tes ting device called the Modified Compact Shearing (MCS) test developed by the Groupe d'Etudes des Matriaux Htrognes ( GEMH ) Gnie Civil et Durabilit ( GCD ) laboratory at the University of Limoges, France to investigate shear fatigue on tack coats The device developed by Diakhate et al shown in Figure 2 7 requires a three layer specimen

PAGE 36

36 bonded specimen but can also work without tack coat. The device operated by subjecting the specimens to displacement in a repetitive manner. The two outside layers re main fixed, while the displacement is applied to the center layer, resulting in cyclic shear loading at each interface. Resistors measured the response at the interface to determine failure. Tests wer e run at a constant frequency of 1 Hz and at a constant temperature of 5 .0 C, which according to the authors, results in bituminous material moduli equivalent to similar tests run at 10 Hz and 15 .0 C. Compacting to 93 .0 %, materials typically utilized in France pavements were used according to French design sp ecification. Figure 2 7 Specimen in the MCS device Source: Diakhat, M., A. Phelipot, A. Millien, and C. Petit. Shear Fatigue Behavior of Tack Coats in Pavements. Road Materials and Pavement Design, Vol. 7, No. 2, 2006, pp. 201 222. Results indicated that there are two stages to interface failure resulting from this shear fatigue test. The first involves progressive failure, where the force amplitude decreases in a linear fashion. The second stage, however, involves more rapid failure, which the author s believe is the result of a loss of shear stiffness in the tack coat

PAGE 37

37 material and aggregate interlock becoming the prevailing factor of the response. This study also included the use of direct shear testing for comparison purposes. It is believed that fat igue testing can be an asset to evaluating the durability of various interfaces. The authors suggest that further evaluation of fatigue testing is necessary to determine more interface parameters that can be assess ed to ultimately improve modeling of the a sphalt concrete interface. In 2008, Diakhat et al showed the relationship between monotonic and cyclical testing of pavement interfaces. The correlation was developed to allow contractors to determine fatigue responses of tack coats from strength testing reducing testing time. The relationship could be used to help contractors and pavement designers expeditiously predict pavement life. M odeling efforts towards evaluating longitudinal top down cracking were developed by Petit et al. using data from the i r previous work (Diakhat et al. 2008). This research focused on introducing new failure modes, related to the repeated application of shear stress on thin surface layers into pavement design. The results indicated the necessity for evaluating interface f atigue in the pavement design process for pavements with thin surface layers to ensure the proper selection of tack coat and application rate. Results also suggested that this is research is of particular importance for curved roads, where horizontal loadi ng is often more critical. In 2011, Diakhat et al further investigated shear fatigue behavior using the D ouble S hear T est (DST) (Figure 2 8 ) to perf orm both an experimental and numerical stud y of interface fatigue behavior to enhance life cycle assessmen ts of asphalt pavement. Sp ecimen production for the test wa s very involv ed, and included several

PAGE 38

38 steps from double layered (one thick, one thin) slab compaction, applying tack coat, sawing the specimen and gluing steel plates on to the specimen to secure it to the loading frame. The frequency used for each test was 10 Hz, and specimens were tested at 10 .0 C and 20 .0 C. Figure 2 8 Schematic (left) and photograph (right) of the DST device Source: Diakhat, M., A. Millien, C. Petit, A. Phelipot Mardel, and B. Pouteau. Experimental Investigation of Tack Coat Fatigue Performance: Towards an Improved Lifetime Assessment of Pavement Structure Interfaces. Construction and Building Materials Vol. 25, No. 2, 2011, pp. 1123 1133. Results of this force controlle d test were presented as a plot of normalized shear stiffness modulus versus loading cycles. There were two observed st ages within the resulting trend curve. Stage one shows a n initial slow decrease in stiffness modulus was determined to result from the de velopment of micro cracks. The second stage was characterized by a rapid decrease in the stiffness modulus, which the authors believed macro cracks propagated quickly at the interface This finding directly supports previous findings by the author in 2008, as discussed earlier in this section.

PAGE 39

39 A new apparatus developed at the Sapienza University in Rome, called the Sapienza Inclined Shear Test Machine (SISTM) (Figure 2 9 ) was util ized for dynamic shear testing (Tozzo et al. 2014). The test involved using both normal and shear stress to evaluate pavement interfaces. Using the CIRCLY software program, various stress levels were determined for evaluation of the uppermost pavement interface (wearing course) that would represent actual stress levels found in r eal pavements at the edge of the wheel path Double layer specimens were constructed and the most common application rate of 0.4 kg/m 2 (0.088 gal/yd 2 ) residual binder was used to improve adhesion. Figure 2 9 Image of the SISTM device Source: Tozzo, C., A. D'Andrea, D. Cozzani, and A. Meo. Fatigue Investigation of the Interface Shear Performance in Asphalt Pavement. Modern Applied Science Vol. 8, No. 2, 2014, pp. 1 11. Using results from the CIRCLY analysis, a triangular load was selected for testing. S everal normal and shear stress values were analyzed. Lower load states were selected to directly correlate with values determined from the CIRCLY analysis, while

PAGE 40

40 higher loading states were incorporated to r educe testing time and potentially be used to esti mate tack coat responses. Testing was conducted on double layer cylindrical specimens at 20 .0 0.5 C. Loading times were held constant at 0.1 s ec but total cycle time varied from 0.2, 0.5 and 1.0 seconds to evaluate the effect of relaxation time (0.1 s e c 0.4 s ec or 0.9 s ec respectively). The device was set to a 60 degree incline in the longitudinal direction of the specimen. This was done so that the normal and shear stress response measured would relate to 30 .0 mm (1.299 in) from the edge of a tire wh eel load. Using this set up, a fatigue law was developed to relate load level to the number of cycles to failure of a specimen. Results indicated that further analysis was required to capture the ratio between shear stress and normal stress for more accura te predictions. The Sapienza Direct Shear Testing Machine (SDSTM) was developed next, to allow for the application of both shear and normal stress to a double layered cylindrical specimen This guillotine style test involved one stationary and one dynamic side of the test device. This testing configuration has been used frequently to evaluate several pavement responses. No rmal stress was applied using a pneumatic actuator. Interface displacement was captured from an LVDT. Shown in Figure 2 10 this novelty of this device was that fatigue response could be captured under dynamic loading conditions. A rapid setting cutback emulsion was used as a tack coat, and was applied at 0.46 kg/m 2 (0.101 gal/yd 2 ) residual binder. Testing was conducted at 21.5 1.0 C. A haversine shear load was applied until failure of the specimens. The load was applied in 0.1 sec for each specimen, but the relaxation time (rest period) varied from either 0.1 sec or 0.9 sec. Though preliminary results indicated that the shorter rest per iod would

PAGE 41

41 be sufficient for viscoelastic recovery in low shear stress levels, for higher shear stress values, the longer period of 0.9 sec was preferred. Figure 2 10 Schematic of the SDSTM Source: Dynamic S hear T e sts for the E valuation of the E ffect of the N ormal L oad on the I nterface F atigue R esistance Construction and Building Materials Vol. 61, 2014, pp. 200 205. According to the authors, displacement data indicated that there were three distinct periods durin g the progression of the test. The first stage involved aggregate interlocking, leading to a quick increase in displacement. The second stage involved a constant, and slow growth stage as sliding continues with an almost constant trend and ending with a sl ight inflection. The final stage showed a steep incline in displacement, indicating failure propagation. Ultimately, a 50% reduction in shear stiffness and the number of cycles to failure were used to categorize responses of the specimens. The authors also developed an interface regression law to relate shear stress, normal stress and number of cycles to failure. Monotonic testing results were compared in an attempt to show a correlation between the two tests, but it was not found in this study. 2. 3 Summar y The importance of incorporating cyclical loading into the design process is clearly indicated through the efforts of the research described in s ection 2.2.2. This could have

PAGE 42

42 a significant impact on the pavement design process. Better prediction of paveme nt life could be obtained through the introduction of a practical new method of analysis. I t is important to note that there are many existing bond strength tests currently in practice, as indicated above. However, c yclical tests often take long periods to complete, and can require new equipment. Though more recent efforts have gone towards modifying existing bond strength testing devices to allow for fatigue testing, providing more functionality from a singular device. T hough several tests to evaluate int erface bond exist, there is no consensus on which type of test provides the best prediction of pavement performance. Despite the fact that bond strength tests are a commonly accepted method of evaluati on, they primarily focus on the uppermost layers of the pavement system. As a result, most bond strength tests evaluate the potential for slippage (high temperature) and/or delamination (low temperature). However, the literature suggests that there may be an additional mechanism of failure that is deeper withi n the pavement system wh ich has not yet been fully evaluated. It is common knowledge that the loading experienced by pavement is not the result of a single loading incident. Therefore, bond strength tests, which use monotonic loading modes, may not provide the best representation of what actually occurs in the field. Further, e valuating the quality of the bond at pavement layer interface is paramount to fully predicting pavement life. Therefore a proper evaluation of the materials used for bond should be in corporated into research evaluation, which to date has not been a major component fatigue or cyclical testing research efforts Poor bond leads to an increase in the magnitude of critical strain and a change in its location

PAGE 43

43 Properly predicting pavement re sponse at these inherent discontinuities (interfaces) during the pavement design process can enhance roadway performance

PAGE 44

44 CHAPTER 3 STRESS ANALYSIS 3.1 Identification of Critical Stress States Based on the existing literature the need for a test method t hat captures the resistance to breakdown response of the bond at the pavement interface became apparent. As previously discussed, f ield observations of trench ed sections of pavement structures indicated near surface cracking along the wheel path occurs in areas of deeper interface de bonding (Willis and Timm, 2007) Identifying the location and magnitude of the critical stresses that promote this type of de bonding was therefore necessary for the development of a new test method. T raditional bond strength te st s have been used to evaluate interfaces located 0. 5 to 1 .5 inches beneath the pavement surface. This is a typical thickness of a surface course At these depths in the pavement system, bond strength is considered to be the driving factor for predicting p avement distress However, t here is a lack of knowledge about what occurs at interfaces deeper in the pavement system. As a result, a parametric study was conducted using an elastic layer analysis program to locate critical stress states at a number of dep ths within the pavement system. 3.2 Elastic Layer Analysis I t is common to utilize computer modeling to provide relative predictions of pavement performance. These models are often coupled with labor atory testing to provide input data t o improve prediction accuracy. There are numerous modeling techniques that exist to predict pavement response. T he KENPAVE program was selected to provide a linear elastic a nalysis of the pavement structures considered in

PAGE 45

45 this research. Within the KENPAVE program is KENLAYER, a tool specifically developed for the evaluation of flexible pavement systems. 3.2. 1 KENLAYER Data Analysis KENLAYER was used to identify potentially critical stress states in typical flexible pavement systems. T he following hypotheses were made: Pavement s with higher stiffness ratios between the asphalt concrete (AC) layer and the base layer produce higher shear and confining stresses There are critical locations in the pavement sys tem where interfaces can be located at which shear stresses are high and confining stresses are low In order to evaluate these hypothes e s the following objectives were employed: Model various stiffness ratios to determine how temperature affects shear and confining stresses felt by the pavement system Identify critical locat ions in the pavement system where shear stresses are highest and confining stresses are lowest Data analysis focused on identifying the locations of critical stress states within the pavement system. These critical locations were defined by having a high shear stress, combined with low levels of confinement. In these areas, it is suspected that the onset of debonding in the pavement is most likely to occur. The pavement layer was assumed to be fully bonded during this analysis. A typical asphal t pavement s ection was analyzed to replicate systems that are traditionally found in the field. Figure 3 1 provides a characteristic cross section of a flexible pavement as it would be modeled using KENLAYER It consists of a surface layer (friction course) an aspha lt concrete layer (binder course) a base layer (base course) and the subgrade. In order to identify critical stress states, the surface (friction) and binder course s were considered fully bonded

PAGE 46

46 Initial analysis was conducted on an 8.0 inch binder cour se with a 12.0 inch base. The 8.0 inch structure, which is often found on interstate pavements and major highways, relates to pavement systems with relatively long service life. Further analysis was conducted on a 4.0 inch binder course layer and 12.0 inch base course as indicated in Table 3 1 which is more commonly found in the field The evaluation of the 4.0 inch pavement structure generally represents a thin asphalt layer. Figure 3 1 Typical cross section of a conventional flexible pavement A circular loading condition was chosen to represent a single wheel An 18 .0 kip single axle load with uniform tire pressure of 100 .0 psi was used for evaluation. The tire was defined to have a 5.35 inch contact radius, which is typical for a single tire ( Hwang, 2000). Appropriate elastic moduli (stiffness) were selected based on typical pavement temperatures. The values selected of both the AC and the base layers are also typical values seen in pavement analysis. However, AC layer elastic moduli varied fro m .0 psi) to a late fall/earl y spring evening ( 1 2 00 000 .0 psi). Friction course < 1 inch Binder course(s) 2 8 inches Base course 4 12 inches Subgrade Infinite R adial distance (in) Tire load (psi)

PAGE 47

47 Table 3 1 Layer t hickness and m aterial p roperties Stiffness ratio Pavement layer Thickness (in) Elastic modulus (psi) ratio SR=5 Asphalt concrete 4.0 200000 .0 0.35 Base 12.0 40000 .0 0.40 Subgrade Infinite 15000 .0 0.45 SR=8 Asphalt concrete 4.0 200000 .0 0.35 Base 12.0 25000 .0 0.40 Subgrade Infinite 15000 .0 0.45 SR=20 As phalt concrete 4.0 800 000 .0 0.35 Base 12.0 40000 .0 0.40 Subgrade Infinite 15 000 .0 0.45 SR=32 Asphalt concrete 4.0 800000 .0 0.35 Base 12.0 25000 .0 0.40 Subgrade Infinite 15000 .0 0.45 SR=Grad Asphal t concrete 0.5 1 200000 .0 0.35 Asphalt concrete 0.5 1100000 .0 0.35 Asphalt concrete 1.0 950000 .0 0.35 Asphalt concrete 1.0 780000 .0 0.35 Asphalt concrete 1.0 640 000 .0 0.35 Base 12.0 40000 .0 0.40 Subgrade I nfinite 15 000 .0 0.45 AC layers are often idealized to possess a uniform stiffness throughout the cross section of the pavement for evaluation. However, in the field, temperature change causes variations in stiffness throughout the structure of t he pavement. Comparatively, high temperature (low stiffness) is associated with the viscous response where rutting is the primary form of pavement distress, while low temperatures (high stiffness) are associated with a brittle response, where cracking is t he major form of pavement distress. Asphalt concrete exhibits viscoelastic properties due to the presence of asphalt binder, consequently, evaluating how the pavement responds to temperature is important. Thus, a stiffness gradient was used to evaluate the effect of temperature on the pavement response.

PAGE 48

48 Based on previous research the stiffness gradient selected was chosen to This can be caused by rapid cooling of the pavement due rai ns, age hardening or temperature differentials due to climate. The addition of the gradient allowed for the evaluation of the effect of field oxidative aging. Since the AC layers analyzed in this study were 4.0 inch layers, only the top 4.0 inches of the s tiffness gradient from Myers et al. were used as indicated in Table 3 1 The stiffness ratio (SR) is the relationship between the stiffness of the asphalt concrete (AC) layer to the stiffness of the base layer described below in Equation 3 1 ( 3 1 ) Where, E asphalt =Elastic modulus of the asphalt layer (psi); and E base =Elastic modulus of the base layer (psi). Ea ch stiffness ratio is denoted above in Table 3 1 along with the structural and material properties chosen for evaluation. A l s o noted in Table 3 1 are the structural and material properties associated with the pavement, which were held constant. Figure 3 2 provides a summary of the pavement structures, and various ch aracteristics evaluated. P avements are labeled according to either SR or gradient. As previously discussed, a broad range of AC elastic moduli were included in the analysis. Figure 3 2 summarizes each pavement structure and its corresponding characteristics, which are labeled acco rding to its stiffness ratio or gradient

PAGE 49

49 SR=5 SR=8 SR=20 SR=32 Gradient SR=5 SR=8 SR=20 SR=32 Gradient Figure 3 2. Structural characteristics of evaluated pavement sections (layer thick ness in parenthesis) 3.2. 2 Critical Zone for the Onset of Debonding The results of the analysis indicate areas within the pavement structure where high shear stress is observed over a broad range of depths around the center of the AC layer. Figures 3 3 and 3 4 show the stress distribution results of the 4.0 inch pavement structures. From Figure 3 3, high shear stress is observed from 1.0 to 3.0 inches, with the maximum value occurring slightly above approximately 2.0 inches in depth. This depth in the pave ment agrees with previous research field observations (Muench and Moomaw 2008 ; Roque et al. 2011). In addition, as SR increases, shear stress also increases. This is expected, as shear stress is a function of bending, which increases with SR. The maximum s hear stress occurred for SR 32, with the gradient having the second highest shear stress response. 25 ksi 200 ksi ) 40 ksi 800 ksi ) 25 ksi 800 ksi ) 40 ksi 200 ksi ) 40 ksi 200 ksi (8 40 ksi 2 5 ksi 200 ksi ( 40 ksi 800 ksi ( 25 ksi 800 ksi ( 40 ksi

PAGE 50

50 Figure 3 3. Horizontal shear stress through the depth of the 4.0 inch AC layer cases Figure 3 4. Horizontal shear stress and average vertical str ess for the 4.0 inch AC layer cases at a depth of 2.0 inches. Figure 3 4 shows the horizontal shear stress distribution at a depth of 2.0 inches in the AC layer. This figure indicates that the maximum shear stress for each case 0.0 1.0 2.0 3.0 4.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Pavement depth (in) Horizontal shear stress (psi) SR=5 SR=8 SR=20 SR=32 Gradient 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 0.0 5.0 10.0 15.0 20.0 Average vertical stress (psi) Horizontal shear stress ( psi) Distance from load center (in) SR=5 SR=8 SR=20 SR=32 Gradient Vertical stress Confined Zone 30 v Critical Zone 4.0 inches

PAGE 51

51 occurred under the edge of t he tire The is located directly under the tire, where normal stress reaches nearly 75 .0 psi. Once outside of the edge of the tire and the confined zone, normal stress decreases drastically below 30 .0 ps i, located outside of the tire edge In this approximately 4 .0 inch rapidly drops below 30 .0 psi, yet shear stress values are relatively high for each SR ( greater than 30 .0 psi). Converting the 100.0 psi s tress applied into an equivalent load indicates the pavement was subjected to 8992.0 lbf. Therefore, estimated load levels at the critical depth of 2.0 inches were 6294.4 lbf (equivalent to 70.0 psi of shear stress ). Simila r results were found for the 8.0 inch AC layer structures as indicated in Figures 3 5 and 3 6 As expected, the magnitude of the horizontal shear stress through the depth of the AC layer and located under the edge of the tire is lower than the 4.0 inch AC layer cases. This is due to the r eduction of bending as a result of a thicker AC layer, while maintaining the same structure underneath. An area of high shear stress e xists between 1.0 to 3.0 inches beneath the surface, with a maximum at the depth of 2.0 inches, comparable to the results for the 4.0 inch AC layer structures. This indicates that at 2.0 inches beneath the pavement surface, horizontal shear stress reaches its maximum value, regardless of AC thickness and the characteristics of the pavement structures analyzed. Figure 3 6 also indicates the existence of a confined zone within the pavement structure located beneath the tire, and ending at the edge of the tire, as in the 4.0 inch AC layer case. Normal stress reached closer to 90.0 psi in this case. This area was then followed by an area in the pavement where confinement was relatively low (reducing drastically from 40.0 to 10.0 psi), but shear stress values were relatively

PAGE 52

52 high (45.0 to 30.0 psi). Since the maximum horizontal shear stresses were much lower than in the 8.0 inch cas e, the critical zone was approximately 2.0 inches from the edge of the tire. Figure 3 5. Horizontal shear stress through the depth of the 8.0 inch AC layer cases located under the edge of the tire Figure 3 6. Horizontal shear stress and average vert ical stress for the 8.0 inch AC layer cases at a depth of 2.0 inches. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 Pavement depth (in) Horizontal shear stress (psi) SR=5 SR=8 SR=20 SR=32 Gradient 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 0 5 10 15 20 Average vertical stress (psi) Horizontal shear stress ( psi) Distance from load center (in) SR=5 SR=8 SR=20 SR=32 Gradient Vertical stress Confined Zone 3 5 v 90 Critical Zone 2.0 inches

PAGE 53

53 3.2.3 Traffic Wander Tire Size and the Critical Zone The results of s ection 3.2.2 indicate that there is the existence of a critical zone of high shear and low confinement at a depth of 2.0 inches in the AC layer extending from the edge of the tire for up to 5 .0 inches. This critical zone can potentially promote the development of a debonded area referred to as a debonding strip, along the tire wheel path. However, traffic does not stay in a singular path along the highway, as drivers in their vehicles tend to display lateral movement within a lane. This lateral movement is referred to as traffic wander and occurs in the transverse direction of the pavement Though traffic wander is oft en unaccounted for in structural design, it has an effect on the pavement response. Previous studies indicate that the lateral position of wheel loads follow a normal distribution with a standard deviation of approximately 4.0 to 11.0 inches within a lane depending on the method used for calculation (Erli n g sson et al. 2012). Figure 3 7. Schematic of debonded zone around tire and debonding strip along the tire wheel path within a traffic lane. Traffic Travel Direction Debonded zone Tire Debonding strips along wheel path

PAGE 54

54 The impact of tire wander varies with the type of tire used by truck drivers As the trucking companies move to increase the use wide based radial tires due to their fuel efficiency, the impact on effect of wander can increase the debonding strip. T ire contact size can range from a conventional radial tire with a c ontact width of 8.0 inches, to wide base tires ranging from 12.0 18.0 inches of contact (Roque et al. 2011) In combination with wander, the debonding strip width can vary from 20.0 up to 35 .0 inches (Figure 3 7), as the trucking industry moves towards u sing wide based tires to reduce emissions, improve fuel efficiency and increase load capacity. 3. 3 Summary The elastic layer analysis clearly indicates the existence of a critical zone within the AC layer. In this zone with an applied load of 100.0 psi, h igh horizontal shear stress up to 70.0 psi can be coupled with low vertical stress (confinement) at a depth of 1.0 to 3.0 inches below the pavement surface where the most critical depth is approximately 2.0 inches When pavement interfaces are located wit hin this range of depth, a debonding strip may occur along the interface even in the case of thick AC layers. In addition, b oth wander and tire size can impact the debond ing strip significantly. Depending on the tire size and brea d th of traffic wander con ditions, the debonding strip can vary from 20.0 to 35 .0 inches in width. This indicates the need for further analysis of the bond at pavement interfaces, particularly those located near 2.0 inch depths. As indicated in section 2.2.1, shear strength tests are often used to evaluate thin surface AC layers, typically of 0.5 to 1. 0 inches in depth. The most critical areas located deeper within the pavement surface, have low confinement which may lead to the onset of debonding. In these areas, pavement bond she ar strength may not be the best method of evaluation, as it may not

PAGE 55

55 capture the gradual loss of bond within the pavement s tructure. This loss of bond may not result from the pavement reaching its ultimate shear s trength but as a result from the repeated a pplication of high horizontal shear stress where confinement is low In order to assess this concept that may lead to near surface cracking, a new bond test method is necessary.

PAGE 56

56 CHAPTER 4 MATERIALS AND SPECIMEN PREPARATION 4.1 Materials The materials in t his research represent those typically used in the state of Florida. In order to replicate conditions typically found at interfaces between structural layers a dense graded mixture was chosen. All mixtures were made with the same aggregate and binder. A Su perpave mix designed fine dense graded asphalt mixture (GA I2) was selected and the full gradation and mix design are located in Appendix A. The aggregate in the mixture was Georgia granite and local sand with a 12.5 mm (0.492 in) nominal maximum aggregate size. The GA I 2 particle size distribution (gradation) is shown in Figure 4 1 A conventional asphalt binder, PG 67 22, was selected for this mix design with a 4.8 % optimal asphalt content. This mixture is commonly used in FDOT projects and serves as a go od performing granite mix design. Figure 4 1. GA I 2 dense graded aggregate mixture

PAGE 57

57 The aggregate used in this research was obtained from several mines used in the North Florida region. Table 4 1 shows a detailed breakdown of the locations for each mate rial Table 4 1 Aggregate sources Material FDOT Code Producer Pit No. #78 Stone 43 Junction City Mining GA 553 #89 Stone 51 Junction City Mining GA 553 W 10 Screenings 20 Junction City Mining GA 553 Local Sand V.E. Whitehurst & Sons Starvation Hill Three different tack coat materials were selected for evaluation in this research. All materials were supplied by companies that produce tack coat. As the state of Florida has chosen to almost exclusively use trackless tack (TT) coat in construction, on e was selected for use. A conventional tack (SS 1) and a PMAE (PT) were selected for comparison. Since the trackless tack and the PT are proprietary, application rates provided by the manufacturer were chosen. A low (0.02 gal/ yd 2 ) and high application rate (0.155 gal/ yd 2 ) for the conventional tack coat w as chosen from the NCHRP 9 40 report. The low rate came as a result of the international survey discussed in the report, whereas the high rate was selected as the rate chosen for their testing ( Mohammad 201 2) This would allow for the evaluation of the effect of tack coat material and application rates. Each tack coat had manufacturer specific storage and application temperatures which were followed to ensure optimal use of the materials. Tack materials wer e stored

PAGE 58

58 in their original containers and were occasionally stirred according to their storage instructions. Materials for tack coat were obtained from three vendors across the US 4.2 Specimen Preparation All specimens were prepared in the laboratory for testing using the device described in s ection 5.2 D ouble layered specime ns were prepared for testing by combining half of a full Superpave pill, and an over lay mixture through compaction. A total of 3 0 specimens we re made for trial testing, and 30 specim ens were used in final testing, in accordance with the testing plan below in Table 4 2 For final testing, triplicate specimens were produced for the evaluation of each testing method, tack coat, and application rate. Table 4 2 Testing p lan Tack c oat Su rface t ype Application r ate Temp C Replicates Tests Total s pecimens None (NT) Compacted 0.000 gal / yd 2 25.0 2.0 C 3 2 6 Trackless (TT) Compacted 0.060 gal / yd 2 25.0 2.0 C 3 2 6 PMAE (PT) Compacted 0.120 gal / yd 2 25.0 2.0 C 3 2 6 Conventional ( CTL) Compacted 0.020 gal / yd 2 25.0 2.0 C 3 2 6 Conventional (CTH) Compacted 0.155 gal / yd 2 25.0 2.0 C 3 2 6 Total n umber of t ested s pecimens 30 4.2.1 Batching and Mixing A 4500 .0 g ram batch was produced to create two bottom halves using the gra dation found in Appendix A A n overlay batch of 2254.8 gram s was produced for

PAGE 59

59 each b ottom half This mass was determined based on a proportional design method described i n Appendix B After batching, a sphalt binder and aggregates were placed in an industri al laboratory oven and heated for two hours prior to mixing at a temperature of 315 .0 F (157.2 C) ( Figure 4 2 ) Figure 4 2 Materials and mixing tools inside of the oven. Photo courtesy of author. After being heated, the appropriate combination of agg regate and binder were mixed in a mechanical mixer (Figure 4 3 ) until the asphalt covered the aggregate well Figure 4 3 Mechanical mixer. Photo courtesy of author.

PAGE 60

60 Mixtures were short term oven aged also at 315 .0 F. After an hour in the oven, the mixt ures were stirred to ensure uniformity in aging. 4.2.2 Initial Compaction Specimens were produced by compacting a full Superpave pill using the Superpave Gyratory Compactor (SGC) to 4 .0 1.0% air voids where N design =2 7 ( Figure 4 4 ) Figure 4 4 Device used to produce SGC pills Photo courtesy of author. 4.2.3 Cutting Onc e the pill cooled for 24 hours after compaction the pills were measured and the halfway point of the cylinder was measured. U sing a metallic permanent marker a line was drawn on the pi ll to identify the midpoint A water cooled masonry saw ( Figure 4 5 ) with a diamond tipped blade was used to cut each pill. The blade was aligned with the halfway point of each specimen. This process allowed for each pill to create two

PAGE 61

61 b ottom halves for fu ture specimens. After cutting, each b ottom half was labeled and air dried overnight before further preparation. A fan blew ambient air to help with the drying process S pecimens were elevated on wire mesh stands to allow the air to move underneath as well Figure 4 5 W ater cooled masonry saw prior to cutting a specimen. Photo courtesy of author. 4.2. 4 Tack Coat Application Tack coats were removed from their storage containers and slowly transferred into a graduated glass flask. Glass was chosen because i t does not interact with the chemicals used in the tack coats, as suggested by a manufacturer. A glass thermometer was inserted into the flask to measure the temperature. Using an insulated portable heater ( Figure 4 6 ) each tack coat was then heated to th e appropriate temperature for application

PAGE 62

62 Figure 4 6 Heater with graduated cylinder containing tack coat. Photo courtesy of author. Prior to the application of the tack coat, the average of three height measurements was also recorded to provide a refer ence for future testing for the location of the interface. In addition, the initial weight of the specimen was recorded to ensure the proper amount of tack coat was applied. The scale was then zeroed for the application of the tack coat. The appropriate am ount of tack coat was evenly applied with a brush (Figure 4 7 ) to the compacted side of the base (Figure 4 8 ) It is important to note that the compacted side of the b ottom half was selected to represent a typical new pavement surface. The total mass of th e combined initial weight of the specimen and the a pplied amount of tack coat was record ed.

PAGE 63

63 Figure 4 7 Brush used to apply tack coat. Photo courtesy of author. Figure 4 8 Tack coat being applied to the b ottom half of a specimen on a scale Photo cou rtesy of author.

PAGE 64

64 An equivalent mass for each tack coat application rate w as calculated based on the percentage of binder using the equations found in Appendix C After the correct amount of tack coat was applied to the b ottom half the specimens were place d aside to allow for the tack coat to set. The base with tack coat was periodically pl aced back on the scale to determine if constant weight had been reached Once the weight stabilized, the b ottom half could move to the final stage of specimen preparation final compaction 4.2.5 Final Compaction For the overlay, a batch of 24 5 4.8 grams was prepared to be compacted on top of each bottom half Once surface preparation of the base and mixing and aging of the overlay were complete, compaction of the base coul d take place. T he b ase of the SGC mold was removed to insert the base The base was placed in the SGC mold with the cut surface facing down ( Figure 4 9 ) Figure 4 9 SGC mold with base insertion in progress Source: Chen, Y. Composite Specimen Testing to Evaluate the Effects of Pavement Layer Interface Characteristics on Cracking Performance. University of Florida 2011.

PAGE 65

65 Loose mix was subsequently placed on top of the surface of the b ottom half and then the specimen was compacted to 7 .0 1.0% air voids. 4.2 6 Determining Air Void Content During the production of trial specimens, the overlay was compact ed to a height of 2.2 inches instead of choosing a gyration number This resulted in very high compaction effort, where the number of gyrations was between 180 and 280, well over what would be expected in the field This was due to the variation in height of each bottom half. An analysis of the results for t rial specimens can be seen in Figure 4 10 From this analysis, a conservative gyration number was sele cted for future trial tests and final testing where N design = 2 5 Figure 4 10 Determ ination of number of gyrations to compact the overlay. Once the specimen preparation process was complete, specimens were transferred to the Advanced Materials and Ch aracterization lab until testing. Specimens remained at room temperature prior to testing.

PAGE 66

66 CHAPTER 5 TEST ING DEVELOPMENT AND DATA INTERPRETATION A new testing system was developed to analyze a new mechanism of interface bond failure due to repeated load application. Results from this new device were used to compare with the relative results of a typical bond strength test. Ideally the new system would allow for repeate d application of shear stress at the inte rface with little to no confinement simulating what occurs in real pavement systems as discussed in chapter 3 A deformation controlled loading mode might provide the most relevant comparison to field conditions by bringing the interface back to zero deformation after each passing of the load Howev er, there is no existing knowledge of how to effectiv ely conduc t repeated shear test ing using a deformation controlled loading mode. The results of the repeated shear testing configuration could provide a measurement of the reduction of effective stiffness at the interface obtained from by the change in load over a constant deformation Ultimately an initial configuration for a new system was based on the information obtained from the literature review in chapter 2. As mentioned in s ection 2.2.2, fatigue type tests can be plagued with complications associated with the testing configuration, loading mode, and time for testing. This chapter primarily focuses on device selection and issues associated with testing and obtaining usable data for interpretation 5. 1 Changes in Florida A s the s tate of Florida moves toward s the sole utilization of non tracking or trackless tack coats as its bonding method, it was important to evaluate all potential uses for this material. In January 2014, FDOT changed its specifica tions to implement

PAGE 67

67 trackless tack materials as the only bonding agent for pavement interfaces. However, research in 2011 indicated that brittle systems like those created by using trackless tack coats, perform poorly by reducing the cracking resistance of materials near the interface (Chen, 2011). However, the work conducted by Chen focused primarily on issues of cracking performance, and not interfacial bond. This project began as an effort to answer the question of whether or not moving to a stiffer inte rface system, defined by the implementation of trackless tack systems, was beneficial to interface s located deeper in the pavement system. The scarcity of information on the potential impacts beyond interfaces located directly beneath thin friction course layers became apparent in the literature review. Moreover, the consequence of only allowing for brittle interface systems in areas where strength may not be the primary con t ributor to failure is also not addressed. The FDOT requires the materials chosen by its contractors to reach a minimum bond strength value of 100 .0 psi for utilization on its roadways. This value is obtained using bond strength testing, which is the primary method used by m any state agencies in the US. Bond strength tests help researcher s understand n ear surface failures, like delamination However, bond strength testing may not necessarily provide insight on issues of pavement bond occurring deeper in the pavement system. As mentioned in section 1.1, this type of testing is often conduct ed to evaluate friction courses and loads located directly beneath the tire where levels of confinement are highest. These are very high stress, single excursion to failure episodes which bond strength tests are able to replicate. However, it represents just one way in which the bond between pavement layers can fail.

PAGE 68

68 R oadways are subject to repeated applications of various stresses. As indicated in chapter 3, critical stress states where pavements experience low levels of confinement and relatively low le vels of shear stress exist in the pavement system. No existing test evaluates the resistance of the bond to breakdown over time due to the repeated application of lower magnitude shear stress es which may be another mechanism by which bond at the interface fails With a focus on addressing the changes to practice in Florida, the test method s discussed in the remaining sections of this chapter w ere developed. 5. 2 Florida Interface Shear Tester Given there was no existing device to assess pavement bond resist ance to breakdown at the interface several devices previously used in asphalt research were identified with the initial goal of identifying one that could work in both monotonic and repeated loading modes. This would allow for consistent specimen preparat ion and evaluation after they were subjected to each loading method After reviewing the literature, several guillotine type strength test devi ces that could be mounted on a universal testing system (ex. MTS) were chosen for comparison This style of devic e is very common for evaluation of bond strength and would be familiar to potential future practitioners of the new method proposed by this work (Mohammad et al., 2012). In addition, the specimens produced for this type of test are commonly used. They were cylindrical, and c ould either cored from the field, cored from laboratory prepared slabs or produced in the laboratory using a SGC. S chematic drawings of three potential devices were obtained to determine whether or not they could withstand repeated appli cation of shear stress or be modified to do so. Ultimately the LISST ( Figure 2 5) and the FDOT Shear Tester ( Figure 2 2)

PAGE 69

69 served as inspiration for the new device (Figure 5 1). These devices were discussed in more detail in c hapter 2. The Florida Interface Shear Tester (FIST) was produced by Associated Technologies & Manufacturing (ATM) in Lo uisiana (Figure 5 1) Th e FIST was a modification of the LISST device designed to our specification to allow for repeated load. The FIST was installed on a MTS servo h ydraulic load frame to allow for accurate measuring of the material response (Figure 5 2). The FIST included the addition of a thicker and elongated bottom plate to reduce the eccentricity of the load, increase flexural strength of the device, and provide a better l ocation to attach it to the MTS system. Its primary components are a loading frame that is allowed to move vertically, and a reaction frame that is stationary, each with a removable top collar. F igure 5 1. Image of the FIST. Photo courtesy of a uthor.

PAGE 70

70 Figure 5 2. Image of FIST mounted on the MTS system Photo courtesy of author. Figure 5 3. Close up view of a specimen loaded in the FIST Photo courtesy of author.

PAGE 71

71 The frames were designed with a 0. 5 inch gap The two collars are removed in or der to properly place the specimen allowing for the interface to be align ed within the gap Two locking pins were provided to attach each collar to the base of the loading and reaction frames. The pins are tightened a r ound threaded rod coupling s (original ly designed with a finger grip ) which allows the device to tighten the around the specimen. As shown in Figure 5 3 the top of the loading frame collar comes in contact with the MTS actuator. A modification of a dditional steel was added to the top of the l oading frame collar to provide a more even distribution of the load where the collar contacts the specimen. In addi tion, the added height prevents the actuator from coming into contact with the reaction frame The add itional material also provides space fo r a threaded bolt to be attached which can fully link the loading frame collar to the MTS system actuator, allowing them to act as one ( if desired ) Multiple collar inserts were provided by ATM to allow for the testing of 150.0 mm (5.905 inches) and 4.0 i nch diameter specimens. O nly 150.0 mm diameter specimens were used in this research. The following sections discuss several iterations of testing and their corresponding results. For each loading method (monotonic and repeated), several trial tests were ru n to identify the optimal testing conditions. In some cases, the optimal testing c onditions required device modifications As issues that contributed to loss of energy applied at the interface became apparent, their sources were eliminated. These energy lo sses were either due to the device itself or the chosen testing mode. Ultimately a final test method was decided upon for both monotonic and repeated testing, of which the findings will be presented and discussed.

PAGE 72

72 5. 3 Energy Losses and Their Solutions Sin ce the FIST was a modification of an existing bond strength test, monotonic loading was chosen for initial testing. Using information provided by researchers at the LTRC facility, a program was developed in the MTS MultiPurpose TestWare (MPT) software to a pply a strain rate of 50.8 mm/min (2.0 in/min). Specimens chosen for initial testing had interfaces, but no tack coat applied. Initial values were in the 300 .0 psi range, nearly double the anticipated results based on previous tests with specimens of the s ame geometry (Mohammad et al., 2012). I t is important to note that there was no reference made to the types of materials used for the mix design of the specimens in the NCHRP 9 40 report, though the authors discuss using a variety of mix designs Researche rs at the LTRC facility were contacted to obtain a more detailed explanation of their operating procedures for their device. According to researchers at LTRC, bond strength testing procedures adopted for the FIST were consistent; however there was little c onfidence in this response, as it remained unclear what led to the increase in stress. A secondary set of tests were performed to compare with another loading rate. The loading rate was adjusted in an attempt to achieve results similar those obtained by t he FDOT (Sholar et al., 2004). This was also useful for addressing the changes to the materials used in this this research are consistent with materials used in the state Thus, the loading rate was reduced from 50.8 mm/min (2. 0 in/min) to 2.54 mm/min (0.1 in/min). Though the results became more similar to those found by Sholar et al. upon further inspection the specimens were not failing directly through the interface. This indicated that the measured response was not only tha t of

PAGE 73

73 the interface, but also t he mixture itself. This led to the investigation of the influence of gap width on shear strength test results. 5. 3 .1 Gap Width After observing several tests with the 0.5 inch gap width failure was seen propagating from a cont act point where the inside c orners of the c ollars touched to the specimen and through the mixture From this point, cracks moved through the mixture and ultimately reached the interface. The results of these tests were suspected to be of in layer shear beh avior and not necessarily representative of the interface. The right side of Figure 5 4 shows specimen T S2B that failed through the mixture during trial testing instead of through the interface as shown on the left side with specimen CL S3. Figure 5 4 Two specimens failed with differe nt gap widths on the FIST Photo courtesy of author.

PAGE 74

74 The device specific ation called for a 0.5 inch gap between the collars on the device. Upon a more critical investigation, a chamfered edge on the interior side of each collar added 0.125 inches to each side. This meant that the gap width was effectively 0.75 inch es, much larger than what has been previously found in the literature. Figure 5 5 shows the full extent of the beveled edge on a collar insert, which i s difficul t to show on the FIST due to its location Figure 5 5 Collar insert that shows the beveled edge on the FIST. Photo courtesy of author. Despite the existence of several modified shear bond testing devices, there remains limited evidence relating to why a specific gap width was chosen. A s tudy on gap width that investigated interlayer versus in layer shear behavior also noted a lack of previous research on gap widths (Raab et al., 2010). This work was in response to concerns raised in response to the mod ifications made by the United Kingd om (UK) for standardization of European testing. The UK wanted to ensure that devices would allow for specimens to be properly aligned between shearing rings in direct shear tests. Raab

PAGE 75

75 et al. investigated 0.0, 2.5 and 5. 0 mm gap widths and compared bond strength between the pavement layers. Specimens were comprised of four layers, with the top layer being stone m astic asphalt (SMA), and all subsequent layers being asphalt concrete in accordance with Swiss standards. Five to seven 150.0 mm diameter cores were tested at 20.0 C on a Leutner Shear T est (Raab et al., 2010), which is a guillotine style device for each gap width with l oading rates of 2.5 mm/min and 50 .0 mm/min being evaluated. Results from this work indicated t hat eccentricity is increased as gap width increase s This eccentricity leads to a bending and shear stress state. Additionally, the authors mentioned that zero gap width would be ideal, but not practical Their research indicated that the gap width on exi sting direct shear devices should not be greater than 2.5 mm. The results also showed that in layer shear forces were higher than shear forces at the interface. Though t here is limited research on the effect of gap width on interface shear bond results th ese results confirmed that adjusting the gap width was important, particularly since the gap width of the FIST was nearly an order of magnitude larger than the largest gap width used by previous researchers L arge gap widths can be associated with lack of a defined failure plane, resulting in failure occurring at the weakest point in the materia l (Raab et al., 2010) This was visually confirmed in Figure 5 4 Previous research also indicates that eccentricity is increased as gap width increases (Raab et al. 2010) The eccentricity caused by the larger gap width allows for bending in the specimen thus the results are a combination of bending and shear stress. Thus, reducing the gap width of the FIST device became necessary to improve how the load was being transferred to the specimen.

PAGE 76

76 The FIST gap width was reduced to allow for the comparison of results between those obtained from previous FDOT Shear Bond testing (Sholar et al., 2012). It was ultimately reduced to 0.25 inches (6.35 mm) which is larger than the width suggested by Raab et al. This width allows for some irregularity in the interface surface, so that the interface can be aligned within the shear plane, but also reduced eccentricity in the specimen during loading. This reduction also involved th e removal of the chamfered edge dimension on either side of the device. SGC pills (specimens without an interface) were subsequently tested to obtain an in layer shear stress value. The comparison of the results from a typical pill and a typical no tack sp ecimen with an interface is shown in Figure 5 6. Although the response is similar to what was seen for no tack specimens, it is believed that the slight increase in peak stress was due to the specimen failing through the mixture first. Figure 5 6 Compar ison of shear stress response of a pill and specimen with no tack. 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear stress (psi) Displacement (in) Pill NT

PAGE 77

77 After the gap width was reduced, specimens failed along the interface, as originally anticipated. The right side of Figure 5 4 shows specimen CL S3 which failed through the interface after the gap width was adjusted for final testing. 5. 3 2 Bond S trength Test ing Strength testing was conducted using the FIST device for evaluation of the various interface conditions indicated in chapter 1. Specimens were prepared according to the steps in sect ion 4.2, and for each interface condition, three replicates were evaluated. The following steps were taken prior to testing: Bring the loading and reaction frames to the same level and use the pin to secure them in place Place the specimen in the FIST, ali gning the interface marking with the gap. The bottom half of the specimen should be on the reaction frame side, and the top half on the loading frame side Draw the direction of loading on the specimen Lock the specimen in the collar by tightening the locki ng pins using the knurling until finger tight and ensure cams are completely tight After checking the specimen to ensure it is still properly aligned, remove the pin and begin testing. The program starts by the actuator applying a 10.0 lbf seating load af ter which the operator can begin testin g. The actuator then moves down at a constant rate of displacement 2.54 mm/min (0.1 in/min) until the p redetermined displacement limit of 0.1 inches is reached Once the displacement limit is reached, the actuator ret racts and the specimen can be removed from the device. Each test took four minutes on average to complete The resulting shear displacement curves for each interface condition are shown in Figures D 1 to D 5 During testing, operator error caused the resul ts of a replicat e to be invalid (Figure D 3 ). The pin used to align the loading frame with the

PAGE 78

78 reaction frame was left in place during testing. The results are therefor e not included in the analysis. Figure 5 7 Average of the maximum shear stress for va rious tack coats. Figure 5 7 shows the average maximum shear stress for specimen s with different tack coat. The results of the strength tests indicate that the specimens with no tack coat (NT) trackless tack (TT) and a low application rate of conventional tack coat (CTL) exhibited the highest shear strengths This is consistent with the results of previous studies on bond strength (Sholar, 2004; Mohammad et al., 2012). It is important to note that application rate played a role in the results. From Figure 5 7, it is clear that brittle, low application rate materials exhibited higher strength. The NT TT and CTL shear strengths were 25.7% greater than the PMAE tack coat ( PT ) and the high application rate of conventional tack coat ( CTH ) specimens on average r eaching 126.6, 126.5 and 126.4 psi respectively The high application rates of the PT and CTH yielded very similar results of 100.9 and 100.7 psi respectively indicating that NT TT CTL CTH PT Strength (psi) 126.6 126.5 126.4 100.7 100.9 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 Shear strength (psi) Specimen type

PAGE 79

79 higher rates of tack coat can diminish the performance of the interface in singl e action, high stress failures This also supports previous research conducted on bond strength testing (Sholar, 2004; Mohammad et al., 2012). 5. 3 .3 Complications with Repeated Testing Modes For repeated shear stress testing (RST) determining the appropr i ate testi ng mode was a major topic to address Previous research described in section 2.2.2 indicates practitioners of similar fatigue type analysis have found difficulty with effectively obtaining data for evaluation Many researchers limit testing time, and extrapolate data for analysis. These considerations went into the process of developing the repeated testing method that would give insight into the breakdown of bond at the interface However, it is important to note that this previous work was not in an effort to characterize the resistance to the breakdown of the bond at the interface. The UF MTS servo hydraulic system allows for both load and deformation controlled repeated testing schemes to be programmed within the MPT software With any testing m ode, it i s important to identify an appropriate level of load Concerns with the ability to estimate and control the amount of load applied to the interface under deformation controlled conditions led to the decision to attempt test ing using a load control led mode. Applied load levels were determined from the elastic layer analysis in chapter 3 in tandem with the results of the strength data presented in section 5.3. 2 Several attempts of load controlled testing were made, primarily due to difficulties ide ntifying an appropriate load level that would result in failure of the specimen within a reasonable time frame.

PAGE 80

80 For each variation of the L oad C ontrolled RST (LC RST) the full loading sequence was one second. The loading period was 0.1 seconds (0.05 sec l oad, and 0.05 sec unloading) followed by a r est period of 0.9 seconds. Figure 5 8 Repeated load ing period The loading period was selected to allow for the relaxation of the materials at the interface, unlike in monotonic testing. A half sine load was the shape of the load applied as illustrated in Figure 5 8. All tests were performed at room temperature (25.0 2.0 C ) The initial design o f the LC RST program applied a 10.0 lbf seating load. Then the chosen load was applied. The actual applied load was varied with each attempt of the test from 1000.0 lbf up to 2250.0 lbf, which will be discussed more in detail in this section. T he LC RST program ran until the displacement limit of 0.50 inches was reached or the operator terminated the program. At this p oint, the actuator was retracted and the raw data was saved. -0.2 0 0.2 0.4 0.6 0.8 1 1.2 -1 4 9 14 Load (lbf) Time (sec) Peak load 0.1 sec load 0.9 sec rest

PAGE 81

81 5.3.3.1 Data Interpretation This section provides a general overview of the techniques used to analyze the data evaluated in section 5.3.3 as well as section 5.4. Previous research indicates tha t at least 500 points per second is required to capture the effects of repeated load (Chen, 2011). For RST evaluation, d ata was acquired every three minutes, with 512 points acquired every second for 6 seconds (cycles) The measurements obtained were load (applied), deformation (measured from loading head movement ) and time These measurements were sent to a .DAT file created by the MPT software. Data was subsequently exported into Microsoft Excel for post processing and further interpretation. Typical plot s of the repeated loading and axial displacement are shown in Figures 5 9 to 5 1 2 As shown in Figures 5 11 and 5 12, several 6 cycle data made up the full data set that was ultimately used for interpretation. The number of 6 cycle data that were acquired depended on how many cycles it took to reach the established displacement limit of 0.5 inches which terminated the test It is important to note that similar to as shown in Figure 5 12, the point at which the displacement limit was reached was not acquired by the software, as data was only acquired for 6.0 seconds every 3.0 minutes. Data reduction was necessary to analyze the results provided from the MPT software. Depending on the length of the test the number of data points could range from 400000.0 to 95 0000.0 for each of the outputs. The complexities of handling this size of data required the development of a routine in Microsoft Excel for data reduction. Axial displacement data provided by the MPT software was used to identify the total, permanent, and resilient deformation information. The routine identified the maxima (total deformation) and minima (permanent deformation) of the 6 cycle data.

PAGE 82

82 Figure 5 9. Typical raw data for repeated loading versus time. Figure 5 10. Typical raw 6 cycle data of axi al displacement versus time. 0.0 450.0 900.0 1350.0 1800.0 2250.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Load (lbf) Time (sec) 0.417 0.418 0.419 0.420 0.421 0.422 0.423 0.424 0.425 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Axial displacement (in) Time (sec)

PAGE 83

83 Figure 5 11. Progression of displacement for five acquisitions of 6 cycle d ata versus time. Figure 5 12. Typical full data set of axial displacement data. F our full cycles (i.e. the first and final cycles were removed) fo r each acquisition time was averaged together. This value would represent the deformation (total or 0.415 0.420 0.425 0.430 0.435 0.440 0.445 0.450 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Axial displacement (in) Time (sec) Note: Three minutes were between each 6 cycle data acq uisition

PAGE 84

84 permanent) for the time at which the data was acquired. The difference between the total and permanent deformation, referred to as resilient deformation, wa s also calculated. A similar calculation was made to confirm the applied load was consistent throughout testing. Table 5 1 shows typical values for a specimen tested using the load controlled a nalysis. Table 5 1. Typical deformation obtained from Microsoft Excel analysis Deformation (in) Cycle 1 Cycle 2 Cycle 3 Cycle 4 Average Total (max ) 0.4242 0.4243 0.4244 0.4245 0.4244 Permanent (min) 0.4178 0.4178 0.4179 0.4180 0.4179 Resilient (max min) 0.0064 0.0065 0.0065 0.0065 0.0065 Resilient deformation and load data were used to determine the stiffness (lbf/in) of the material. Normalized values of the stiffness were calculated by dividing the stiffness values by the initial stiffness value. Stiffness reduction has been well recognized as a way to measure d amage induced in a specimen. Often, a stiffness reduction of 50.0% serves as a measure of failure in asphalt testing. Typical calculations of stiffness and normalized stiffness are shown in Table 5 2. Figure 5 13 shows a typical normalized stiffness for t he no tack specimen show a significant decrease until it was beginning to fail. This information was insufficient for observing the resistance of breakdown in this an d other specimens tested at this load level.

PAGE 85

85 Figure 5 13. Normalized stiffness for 1500 lbf LC RST NT specimen. Table 5 2. Typical stiffness obtained from Microsoft Excel analysis Time (sec) Load (lbf) Resilient deformation (in) Stiffness (lbf/in) Nor malized stiffness (dimensionless) 0.0 1854.1 0.0065 285246.15 1.000 3.0 1857.1 0.0066 281378.79 0.986 The results of the elastic layer analysis in chapter 3 indicated that for a typical pavement structure shear stress at structural interfaces was 30.0 psi to 60.0 psi with an applied stress of 100.0 psi. This led to the selection of 1000 .0 lbf (36.5 psi) for initial testing. This load represented 28.8% of the average shear strength for lowest performing specimens and 36.3% for the highest performing spe cimens in the bond strength tests as described in section 5. 3 .2. After testing at the 1000 .0 lbf ( 990 .0 lbf and 10.0 lbf seating load) load for two hours, the specimens remained intact with minimal 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Normalized Stiffness (dimensionless) Time (seconds)

PAGE 86

86 external sign s of damage and very limited reduction in sti ffness. This indicated that the load, and/or the amount of time for testing would need to be increased to induce failure Subsequently, the load was increased to 1500 .0 lbf which represented 43.3% of the average shear strength for lowest performing specim ens and 54.4% for the highest performing specimens. After conducting several trial test s for four hours th e specimen exhibited minimal signs of damage The damage that was visible was located where the specimen and the collar were in contact. It was deter mined that the locking pins were loosening as a result of the compaction of the exterior of the specimen The locking pins were subsequently tightened every two hours to ensure the collars remained in constant contact with the specimen. The decisi on was ma de to proceed with the load controlled method by maintaining the load level but increasing the time of loading. The next set of LC RST trials were run with an applied compressive load of 1500.0 lbf (1490.0 lbf and 10.0 lbf seating load). The time of load a pplication fo r these tests was increased to eight hours, in an attempt to induce more damage at the interface. The results of these trials are presented in Figure s E 1 to E 3 Stiffness reduction for all of the specimens was less than 20.0% prior to specim ens failing. The TT specimen exhibited the highest reduction in stiffness (19.9%), but did not fail during testing. The results of this attempted LC RST evaluation were inconsistent potentially due to a number of factors including the movement of the spec imen in the device and compaction of the mixture at the collars In summary, u sin g 1000.0 and 1500.0 lbf load controlled method could not achieve the typically desired 50.0% reduction in stiffness that was initially chosen to be a measure of failure within a reasonable time frame. The data indicated that specimens

PAGE 87

87 were exhibiting a limited loss of stiffness prior to failure. Additionally, some specimens were unable to fail within the allotted testing time. T o encourage failure of the specimen, higher load l evels were required. 5. 3 .3. 2 Progressive LC RST In an attempt to improve upon the previous method a progressive loading scheme was conceived where the applied load went from 1500 .0 1750 .0 2000 .0 and 2250 .0 lbf sequentially ( 10 .0 lbf seating load and 14 90.0 lbf 1740.0 lbf, 1990.0 lbf and 2240.0 lbf) with each load being applied for two hours (Figure 5 1 4 ) NT, TT and PT we re all tested using this method During data interpretation, the effect of the load changes was visible as seen in Figure E 4 to E 6 To aid in analysis, lines were placed on each graph to indicate where the load level changes occurred. No changes were made to the data interpretation as the Microsoft Excel routine accounted for the load changes. Results showed that the NT, TT and PT s pecimens had 16.5%, 17.3% and 23.7% reduction in stiffness, respectively. While the NT and PT specimens failed before the end of testing, the TT specimen remained intact It was speculated that the specimens might be failing due to high load levels approac relative strength as the PT specimen failed within the load 3 Not apparent from the graphical data shown in Figures E 4 through E 6 is that the TT specimen remained intact after testing. Table 5 3 provides the comparison between the m easured strength of the surface treatments obtained from section 5.3.2 with the chosen applied loads for the RST analysis

PAGE 88

88 Figure 5 1 4 Loads used for progressive loading method. Table 5 3 Percent of measured bond strength reached with applied RST str ess. Applied load (lbf) Applied stress (psi) Specimen type (average shear strength) NT (127.0 psi) TT (133.6 psi) PT (106.6 psi) 1500.0 54.8 43.1% 41.0% 51.3% 1750.0 63.9 50.3% 47.8% 59.9% 2000.0 73.0 57.5% 54.6% 68.5% 2250.0 82.2 64.7% 61.5% 77.0% As the applied loads reached the measured shear loads of the specimens tested, specimens began to fail due to lack of bond strength This was particularly visible while evaluating the PT specimen, since it failed much earlier than the NT and TT specimens The PT specimens exhibited the lowest shear strength, and thus failed in an earlier loading period than the TT and NT specimens. During the third load level, the PT sp ecimen experienced reached 68.5 % of its bond strength, while the TT and NT specimens we re at 54.6% and 57.5%, respectively. 0.0 500.0 1000.0 1500.0 2000.0 2500.0 Load 1 Load 2 Load 3 Load 4 Load (lbf)

PAGE 89

89 E valuating the mechanism of bond strength failure was never the intention of the repeated shear testing. As discussed in the elastic layer analysis conducted in chapter 3, interfaces deeper in the pavement can experien ce load levels up to 70.0% of applied stress It was important to take this into account when determining whether or not the loads were characteristic of what might occur at deeper pavement interfaces. However, the majority of the pavement systems evaluate d experienced stress levels closer to 30.0 % to 60.0% of the applied stress Thus, the method was altered by applying each load for three hours to approximate the 6 0.0% maximum stress The resulting test could then be up to nine hours if the specimens were able to last until then. Also, t he final load (load 4) was removed from the analysis to be certain that the s pecimen shear s trength or load levels beyond the predicted values from the elastic l ayer analysis would not be reached. With the new progressive l oading scheme the LC RST was able to induce failure Each of the specimens evaluated were also able to reach the final loading stage, unlike the previous attempt with four load levels. This allowed for a better comparison between the results, but also inc reased the length of testing. The results of these tests are shown in Appendix E, Figures E 7 to E 9 Ultimately the r esults indicated that the NT, TT and PT sp ecimens evaluated exhibited 19.2 %, 1 5 2 % and 28 1 % reduction in stiffness respectively However this data was determined to not be fully representative of the breakdown of the bond at the interface. D uring a visual inspection of FIST device with the progressive loading method the locking pins were again found loose after testing In addition, s eve ral specimens failed through the mixture, instead of through the interface. This was attributed to the specimens moving during testing as a result of the

PAGE 90

90 looseness of the collars. While a negligible amount of dilation could be expected due to aggregate shi fting during testing as specimens were not confined, the observed movement of the entire specimen was un desired T he attempts at the LC RST were not producing results that were representative of the breakdown of the interface. With specimens failing with a shear strength response, the progressive method was also abandoned. A re evaluation of the methodology for the test was conducted. It was determined that the options moving forward could be a higher single load controlled option or a displacement controll ed RST could be implemented. The displacement controlled method was subsequently pursued. 5. 3 .3. 3 Displacement controlled RST T he results of the LC RST evaluation s led to the implementation of a deformation controlled method. The benefit of having the LC R ST results meant that average deformation associated with each load level could be obtained I t was assumed that this deformation controlled method would better represent field conditions where the elastic energy within the real pavement system would bring the deformation back to zero. The value of 0.0085 in was identified from the average total deformations from the 1500.0 lbf RST data. It was a middle ground representation of the 1500.0 lbf applied load. A conditioning period of one hour was put in place in an attempt to reduce the loosening from the specimen compressing due to the collars, which was identified in the progressive LC RST analysis. For conditioning a LC RST program was run at 1500 .0 lb f The device was then fully tightened before starting t he deformation controlled analysis removing the space caused from specimen compression at the collars Because of this conditioning, tightening every two hours during testing was eliminated

PAGE 91

91 as it was assumed the system would remain fully tight A seating load of 10.0 lbf was applied before the deformation controlled program could proceed. Again a load pulse of 0.1 seconds and a rest period of 0.9 seconds were chosen, as shown in Figure 5 1 5 Three trial tests including an SGC pill (no interface), TT and N T specimens were evaluated. Displacement controlled data analysis was conducted using the same Microsoft Excel routine as discussed in section 5.3.3.2. The difference between the analyses is found in the load data. Over time, the load acquired during the d eformation controlled analysis reduced. A typical representation of this is shown in Figure 5 16. The figure also clearly shows the conditioning period. This data was removed for the calculation of deformation, stiffness and normalized stiffness. Figure 5 17 shows a typical stiffness plot with the conditioning period data removed Figure 5 1 5 Repeated applied deformation schematic. -0.2 0 0.2 0.4 0.6 0.8 1 1.2 -1 1 3 5 7 9 11 13 Deformation (in) Time (sec) 0.0085 in 0.1 sec deformation 0.9 sec rest

PAGE 92

92 Figure 5 1 6 Load reduction for typical displacement controlled RST evaluation. Figure 5 1 7 Stiffness reduction for typ ical displacement controlled RST evaluation. 0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 1800.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Load (lbf) Time (sec) 0.0 50000.0 100000.0 150000.0 200000.0 250000.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Stiffness (lbf/in) Time (sec)

PAGE 93

93 Figure 5 1 8 D isplacement controlled RST normalized stiffness results for all specimens Results indicated a 33.0% higher reduction in stiffness than the load controlled tests, which was consistent across all of the initial specimen types evaluated during trial testing. With these apparently promising results, analysis of the other specimen types was conducted. Normalized stiffness results for all specimens are shown in Figure 5 18, and the data shows a consist ent trend across the specimens, with each dropping beyond the 50.0% reduction in stiffness just after 4.0 hours of testing. Normalized stiffness results show little variation between the conditions, with exception of the CTH specimen, which experienced a s econdary drop in stiffness at around 6.5 hours into testing. Though the data appeared to be more consistent, the loosening of the device controlled the response obtained from the MPT software. Thus, the results could not be evaluated. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Normalized Stiffness (dimensionless) Time (sec) NT TT CTL CTH PT

PAGE 94

94 After thoroughly insp ecting the specimens tested under deformation controlled RST compression of the mixture caused by the collars w as found to be more pronounced than in the load controlled test These indent at ions allowed the specimen to work its way loose during testing co ntributing to even more movement than specimens evaluated during LC RST tests It was determined that t he indentations were caused by the free motion of the specimen and not by loading Data also indicated most of the displacement occurred under little to no load This was consistent across all tests. Over time the load felt by the specimen had effectively reduced to zero and the play in the device increased Ideas on how to improve the deformation controlled method were extremely complex to implement with the existing FIST design It might have require d more intuitive software for identifying a zero reference to return the device to after a displacement pulse For example, applying a seating load, then a deformation and finally releasing the load until it w ould reach zero could successfully work as a method, but this design would also increase testing time Ultimately the deformation controlled RST was abandoned as a methodology because of the inability to control the device with the existing setup 5. 3 .4. D iscussion on the S olutions for E nergy L oss Throughout all of the testing conducted with the device, it became clear that energy was not being fully transferred from the loading head t o the location of interest to evaluate interface bond performance. During strength testing, the gap width of the device led to an abnormally high shear stress response from the materials. After reducing the gap, data became more consistent with the anticipated results from the literature. Six attempts were made at developing a test to evaluate resistance to

PAGE 95

95 breakdown of the interface during this testing phase. With each attempt, information on how energy was lost during testing gave insight in to the complexities of cyclical / fatigue ty pe testing in asphalt pavement. Of particular note was the issue of the loosening of the collars of the FIST device. Since previous tests were plagued with loosening of the locking pins, new threaded hex nuts were purchased to improve tightening the device, in addition to circular curved disc spring washers (Figure 5 1 9 ) to help mitigate excess vibration These washers compress as load is applied as the round threaded rod coupling on the pins is tightened actively maintaining a desired load within the pin shaft. Sixteen washers were placed on the dev ice, four per pin s haft. As they are compressed, a light load the washers exert prevent ed the pins from loosening due to the vibration of the FIST Washers were stacked in series, to further stiffen the washers allowing for the hex nut s to remain tightene d for longer periods Additional washers were added to adjust for the height difference between the shorter Allen nuts and the taller original locking pins. Figure 5 1 9 C urved disc spring washers stacked in series. Photo courtesy of author. Additionally capture screws were installed on each of the cams to prevent their movement. The capture screws effectively locked the cams in place during testing,

PAGE 96

96 preventing them from loosening. This was identified as a problem during trials prior to final testing whi ch will be discussed in section 5. 4 Each of the aforementioned changes also culminated in a discussion on the need for the RST method to induce failure in specimens. In several of the trial testing methods used, specimens were able to fail. The failure o f these specimens was often not related to interfacial bond breakdown, as previously reported in section 5.4.3. In an ideal test, f ailure of the specimen would mean that the geometry of the interface changed due to the surface detaching. However in previou s attempts, data acquired with failure of the specimens were often not related to the interface None of these failures were relevant for answering how the bond at interfaces breakdown due to repeated applicat ions of shear stress. Thus, visible failure of the specimen was no longer a requirement for or indicator of a successful test. Reduced gap width, curved disc spring washers, and the capture screw were all device modifications that resulted in the collection of more consistent data for final RST method. Determining what could correlate with RST stiffness from the strength test data was also not apparent. Thus, continuing to use stiffness as a measurement would be problematic moving forward, as one of the goals of this study was to compare relative rankin gs of tack coat performance. Load controlled testing was chosen as the final RST method with the changes presented in this section as data could be consistently obtained All of these solutions were implemented in the final LC RST tes ting discussed in sec tion 5. 4 Ultimately the findings of the trial RST methods led to an improved method of the LC R ST evaluation at a single load.

PAGE 97

97 5. 4 Final Repeated Shear Stress Testing The new testing methodology for the LC RST is described as follows. An 18 5 0.0 lbf compre ssive load ( 50 .0 lbf seating and 1 80 0 .0 lbf load pulse) was selected which represents 60.0% of the average strength of the specimens assesse d in section 5.3.1 using monotonic load Every hour, the hex nuts were tightened to 35.0 inch pounds using a drive click torque wrench to ensure a constant grip on each specimen. The loosening associated with this testing could be attributed to a loss of compression in the springs over time. This new method prevented movement, which allowed for the majority of the load exerted on the specimen to be targeted to the interface. A set of trial tests were run on specimens without tack coat and with trackless tack coat to see how the results would improve based on the se and other changes described in section 5. 3 During thes e tests, the cams became loose, allowing the collar to move minimally. A capture screw was installed on the cams as a preventive measure to mitigate loosening. This was the final change to the FIST design. This design, while still requiring frequent tighte ning of the hex nuts, allowed for consistent data acquisition between tests Several factors were considered for data analysis approaches for th is LC RST data. Though the original intent was to use reduction in effective stiffness as a measurement of the b reakdown of the bond at the interface the response from tightening the device appeared to be quite visible within the resilient deformation data ( which was used for the stiffness calculation ) The influence of the tightening is shown in Figures 5 20 and 5 2 1 which depict the resilient deformation and stiffness data for a typical specimen The average reduction in stiffness for all of the tests was 9.5 % with the maximum reduction of 23.4%. The standard deviation of the stiffness reduction was

PAGE 98

98 7.2%. The dat a was inconsistent between tests for each specimen type as shown in Table 5 4. S tiffness could not be effectively implemented for the comparison of the relative results from the RST analysis to the shear strength analysis. The same can be said for number o f cycles to failure, which could only be obtained for the RST analysis. Table 5 4. Stiffness reduction of LC RST specimens Specimen ID Stiffness reduction (%) NT 1 14.0 NT 2 2.0 TT 1 6.9 TT 2 13.9 CTL 1 0.0 CTL 2 23.4 CTH 1 0.0 CTH 2 11.8 PT 1 15 .6 PT 2 7.4 Figure 5 20 Typical resilient deformation data for final LC RST evaluation 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 35000.0 40000.0 Resilient deformation (in) Time (sec)

PAGE 99

99 Figure 5 2 1 Typical stiffness data for final LC RST evaluation Figures 5 2 2 to 5 2 4 show typical load, total and permanent deformation data acquired for the L C RST final analysis. The average applied load throughout testing varied only 0.21% As shown in Figure 5 2 2 the load obtained for the test was consistently near 1850 lbf. Figure 5 2 2 Typical load data from LC RST evaluation. 0.0 50000.0 100000.0 150000.0 200000.0 250000.0 300000.0 350000.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 35000.0 40000.0 Stiffness (lbf/in) Time (sec) 0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 1800.0 2000.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 35000.0 40000.0 Load (lbf) Time (sec)

PAGE 100

100 Figure 5 23 Typical tot al deformation data from LC RST evaluation. Figure 5 2 4 Typical permanent deformation data for final LC RST evaluation. With the load controlled RST method, the deformation data were the only changing measurement that could provide insight into bond de gradation occurring at the 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 35000.0 40000.0 Total deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 35000.0 40000.0 Permanent deformation (in) Time (sec)

PAGE 101

101 interface. With the variations found in the resilient deformation, both total and permanent deformation data was more closely evaluated. Permanent deformation data was chosen as it provided the most insight into the breakdown of the bond at the in terface. As shown in Figure 5 24 tightening the device did not show up in the permanent deformation data. Ultimately an energy analysis approach, presented in section 5.4.1, was chosen as the method for evaluating the results of RST anal ysis The energy analysis also provided the best comparative analysis between RST data and the strength test data, as energy could also be calculated from shear strength test data. Graphical representation of the permanent deformation results from the fin al RST analysis is presented in Figures F 1 to F 10. 5. 4 1 Energy Analysis In order to compare the relative values from the strength analysis (section 5. 3.2 ) to the LC RST analysis energy from each test was calculated and compared To make this calculatio n, the permanent deformation curve s of each specimen (Fig ure s F 1 to F 10) wer e used to identify the cumulative permanent deformation found in the st eady state region of the curve This value is multiplied by the load to calculate the energy. A typical per manent deformation curve where the steady state portion is identified is presented in Figure 5 2 5 The permanent deformation curve can be divided into three stages: the initial stage, which involves local damage due to the collars, changes in temperature a nd elastic and delayed elastic responses ; the second stage, wh ere relevant permanent deformation associated with the breakdown of the interface is located ; and the final stage, where debonding propagate s through the specimen causing a detachment of the two surfaces and the specimen breaks. The steady state

PAGE 102

102 portion of the curve provides the insight into the breakdown of the bond at the interface. The circle in Figure 5 24 indicates the location of the y intercept for the steady state portion of the curve, wh ich is used in the energy calculations Figure 5 2 5 Identifying permanent deformation data from a typical curve. To identify the initial point of the steady s tate region of the permanent deformation curve the rate of change (slope) between two cons ecutive data points until the midpoint of the data. Using the first hour of t his slope data, the initial point of analysis for the steady state region was selected by visual inspection. A typical slope plot is shown in Figure G 1, where the chosen initial point is circled. Similarly, the slopes between last full hour of data points and any points thereafter until the end of the test were plotted to identify the final point of steady state, as shown in Figure G 2. Once 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 Permanent deformation (in) Time (sec) Permanent deformation

PAGE 103

103 these points were identified, a new per manent deformation plot of the steady state data could be analyzed. A third order polynomial was fit to the steady state permanent deformation curve Using the equation for this curve, the inflection point of this data was identified using Equation 5 1. ( 5 1 ) Where, The cubic equation is defined as ax 3 +bx 2 +cx+d; and The coefficients a, b, c and d are real numbers. An example graph of the steady state data is shown in Figure G 3. A tabl e of the inflection point data is found in Table G 1. The permanent deformation between the initial point and the inflection point were subsequently plotted, and a linear regression was fit to the points to identify a y intercept value. Using this point, t he energy can be calculated. The energy is described by Equation 5 2. ( 5 2 ) Where, y final = final point of the steady state portion of permanent deformation data; and y intercept = the y intercept linear regression fit from the initial point to the of the i nflection point of the permanent deformation data. The results of this analysis are presented in Figures G 5 and G 6. The results indicate that PT specimens performed best with an average of 363.4 in lbf of energy, while the CTH specimens performed worst with 260.6 in lbf of energy on average. Interestingly the C TL specimens were the second best performer with

PAGE 104

104 335.5 in lbf of energy. This indicates that application rate of the tack coat had little impact on the energy of the specimens tested using the LC RST method. TT specimens were the second worst performers in the test, with 274.3 in lbf of energy, while NT specimens fell in the middle with 316.1 in lbf of energy. S tatistical analysis was conducted to determine statistical significance, if any. Those r esults can be found in section 5.4.2. To compare the results of the LC RST data to the strength data, energy needed to be calculated for the strength test results. A simple approximation o f the energy from the strength test data was chosen for this evaluat ion. Using the maximum recorded load and its corresponding displacement from the data obtained for each strength test, a triangular calculation (Equation 5 3) for the energy was made for each specimen and averaged for each specimen type. ( 5 3 ) Where, d peak = the displacement at the maximum load recorded; and P peak = the maximum load (lbf) recorded from the strength test data. Graphical representation of this equation can b e found in Figure 5 26. This novel approach to calculating energy provided a comparable value to assess the energy obtained for reach material analyzed in the bond strength testing. The results of this analysis show a different trend than the energy result s of the LC RST data. The energy calculated from the strength data indicate that the TT specimens show higher energy than any other specimen, with 249.4 in lbf of energy on average. The CTH specimens exhibited the lowest average energy from the strength da ta, followed by the PT

PAGE 105

105 specimens with 128.8 and 144.9 in lbf of energy respectively. The full results of the energy analysis of the strength data is reported in Figures G 7 and G 8. Figure 5 2 6 Graphical representation of energy calculation of typica l strength test data. Figure G 9 shows a scatterplot of the data from both energy analyses with a line of equality. With the data gathered away from a line of equality, results indicate these tests do not provide the same comparison of the tack coat types T he PT specimens are furthest away from the line of equality, indicating that the strength test is the worst indicator of its performance. This supports the hypothesis of this research as it was hypothesized that the performance of the materials used at t he interface is improperly evaluated when based primarily on the outcomes from bond strength tests. 5. 4 .2 Statistical Analysis Statistical data was calculate d for the purpose of comparing the data found in the energy analysis of the LC RST data Due to the limited data evaluated in the LC RST analysis, the initial statistical analysis was done manually, employing a normalization technique commonly used in previous asphalt research conducted at the University of 0.0 500.0 1000.0 1500.0 2000.0 2500.0 3000.0 3500.0 4000.0 0 0.1 0.2 0.3 0.4 0.5 0.6 Load (lbf) Displacement (in) P peak d peak

PAGE 106

106 Florida. In order to construct confidence int ervals for the LC RST data, this normalization approach was used. The energy of replicates of a specimen type was averaged (Equation 5 4) and then the energy of the individual specimen was normalized to that average. ( 5 4 ) Where, x=Energy of the specimens (in lb.); n=Number of specimens evaluated. The VAR.P function was used to find the variance of all of the normalized energy values. The population variance of the energy data from the LC RST tests were calculated using Equation 5 5 using the VAR.P fun ction. The population variance is a statistical measurement often used to identify the amount values vary from t he average within a population. ( 5 5 ) Where, x=Energy of the specimens (in lb.); =Average energy of all of the specimens evaluated (in lb.); and n=Number of specimens evaluated. The population standard deviation was also calculated using t he STDEV.P function. The population standard deviation is the measure of how widely values are distributed from the average (mean) within a population (Equation 5 6)

PAGE 107

107 ( 5 6 ) Where, x=Energy of the speci mens (in lb.); =Average energy of all of the specimens evaluated (in lb.); and n=Number of specimens evaluated. From the standard deviation and mean (for normalized data the mean is 1.0), the coefficient of variation for the population was calculated. The coefficient of variation (CV) is described by Equation 5 7. ( 5 7 ) Where, =Standard deviation of the specimens (in lb.); and =Average energy of all of the specimens evaluated (in lb.). The calcula ted population CV for the LC RST energy test data was 6.95%. Using the mean for each specimen type, a normalized standard deviation was calculated by multiplying it by the population CV. Confidence intervals were then constructed using the appropriate Z va lue. Table 5 5 shows the data for the confidence interval calculations. Graphs of the confidence intervals of the data are also shown in Figures H 2 to H 3. It appears from Figure H 2 that there is significance between some, but not all of the data at a 90 % confidence interval. At an 85% confidence interval, the data does not

PAGE 108

108 overlap indicating overall statistical significance. Additional post hoc analysis was conducted to determine statistical significance between the specimen types. Table 5 5. Confidence interval data for LC RST energy analysis. Specimen Type Mean (in lbf) Normalized Standard Deviation (in lbf) Confidence Intervals 95% Z=1.96 90% Z=1.645 85% Z=1.44 NT 316.12 21.96 13.61 11.42 10.00 TT 274.29 19.05 11.81 9.91 8.68 CTL 335.52 23.31 1 4.45 12.12 10.61 CTH 260.63 18.11 11.22 9.42 8.25 PT 363.44 25.25 15.65 13.13 11.50 Due to the limited number of specimens available to run trial testing, additional statistical analysis was conducted to evaluate the consistency of the results from the final LC RST. Data was analyzed using Microsoft Excel Data Analysis Tool s, which are part of the Analysis Toolpak Add In The ANOVA function was used to determine if there was any statistical significance within the overall population. The alpha c hosen fo r this analysis was 0.10 Using the data from the energy analysis, the resulting P value from the ANOVA was 0.11 (see Tables H 1 and H 2 in Appendix H) Th is confirms that the data overall is not significantly significant at a 90% confidence interval ANOV A analysis does not identify whether or not there is any significance between the sample groups (specimen types) A post hoc analysis was subsequently conducted at an alpha of 0.10 to determine if there was significance between the sample groups When a sa mple size is less than 30 (n < 30), the t distribution is appropriate for determining significance between sample groups To identify the correct t test to run using Microsoft Excel, an f test for two sample variances was conducted to determine

PAGE 109

109 whether or not the variances of the compared data sets were equal or unequal. The null hypothesis assumes that the variances are equal. An alpha of 0.10 was assumed for this analysis. If the resulting P one tail fr om the f test results was greater than alpha (the nul l hypothesis is accepted ), a two sample t test assuming equal variances was conducted. If the resulting P one tail was less than alpha (the null hypothesis is rejected) a two sample t test assuming unequal var iances was run. The alpha used for the t test analysis was also 0.10. T test data results indicate that the TT and PT specimens show significant difference between the means (Figure H 19) The resultant data from each of the sample gr oups is reported in Figures H 3 to H 22 5. 5 Device Limitations It i s also important to mention that the FIST has limitations that cannot be avoided due to its design. The major issue as described in section 5. 3 wa s the difficulty associated with ensuring the applied load is transferred directly to the interface. The secon d is that since the focus of this research was to provide a simple comparison of results for existing bond strength test data, specimens traditionally evaluated using those tests should be accommodated by the FIST. This work only used laboratory prepared s pecimens, eliminating some of the variability of the location of the interface. Cores from real pavements may have interfaces that are not perfectly in plane with the surface of the road. With a gap of only 0.25 inches, ensuring the interface is aligned wi thin the gap could be a challenge with the FIST device

PAGE 110

110 CHAPTER 6 RESULTS AND CONCLUSIONS 6.1 Summary and Findings In this research, the Florida interface shear tester (FIST) was developed to evaluate the resistance to breakdown of the bond for various b onding agents and application rates used for asphalt pavement layer interfaces. This type of failure is believed to occur deeper in the pavement system, where the mechanism of interface bond breakdown is difficult to evaluate. In order to pro perly assess this potential mechanism of failure, an elastic layer analysis was conducted to determine stress states deeper in the pavement system than previous research efforts. The results of the elastic layer analysis indicated the existence of critical stress states supporting the need for this research. The FIST device was used in both monotonic and repeated loading modes to provide insight into how various interface conditions performed under each mode Using the monotonic loading mode the shear stren gth of the bond at the interface was determined D ouble layered cylindrical specimens made with a dense graded mix design were prepared with and without tack coat for evaluation. Three tack coat materials were tested including a conventional tack coat, tra ckless t ack coat and a polymer modified asphalt emulsion (PMAE) The conventional tack coat was applied at two rates (low and high), while the PMAE and trackless tack coats were applied at rates suggested by their manufacturers. Initial attempts to determ ine the shear strength of the specimens indicated that specimens were not failing through the interface .This led to the reconfiguration of the gap width of the FIST device allowing for consistency in results The findings from the

PAGE 111

111 elastic layer analysis and the shear strength tests were used to determine the initial level of loading required for failure due to repeated application of shear stress. Much of the focus of this research ultimately shifted towards the development of the new RST testing method as s everal iterations of analysis were conducted to attempt to reduce the lo ss of energy associated with mixture compaction and loosening of the device A load controlled repeated shear application method was developed that resulted in an energy associated with the breakdown of the bond at the interface. This energy provided insight into the various tack coat types and application rates evaluated in this study. These results were able to show the relative performance of these interface conditions and their resistance to the breakdown of the bond over time. Ultimately, using an energy calculation, the results were compared to the existing monotonic loading method traditionally used to evaluate interface bond strength The f indings associated with this researc h are summarized as follows: Critical stress states conducive to debonding exist 1.0 to 3.0 inches depth beneath the pavement surface, with the most critical depth being 2.0 inches. Bond strength test results grouped the stiffer interface systems together with the NT, TT and CTL specimens exhibiting the highest shear strength. The PMAE tack coat and the CTH had the lowest shear strength. Using the FIST device, t ack coats applied at rates of 0.06 gal/yd 2 or less ha d a negligible effect on interface shear st rength, whereas higher application rates, greater than 0.12 gal/yd 2 reduced shear strength 2 5 7 % Monotonic testing is not relevant for assessing the bond degradation deeper in the pavement system. Repeated loading systems should be used to identify how m aterials at the interface will perform at depths of greater than 1.0 inches. RST analysis indicates PMAE tack coat exhibits the highest energy, while trackless tack displays the lowest energy. Strength test energy indicates that trackless tack displays the highest energy, while the low application rate of the conventional tack exhibits the lowest energy.

PAGE 112

112 Materials with stress relaxation properties (high application rates of PMAE) appear to resist breakdown of bond at the interface better than stiffer system s (little to no tack coat). This is directly counter to the results obtained from monotonic shear strength testing in this research and found in the literature. Overall the results indicate that shear strength cannot be used as the sole method of evaluation of i nterface performance. In conclusion, it is clear that bond strength testing cannot capture how pavement bond degrades due to repeated applications of lower magnitude shear stresses. Thus, monotonic load tests improperly predict interface bond performance of various tack coats applied in location s deeper in the pavement system where strength may not be the primary failure mode Ultimately this work indicates the FIST device and RST method developed for this research can be used to evaluate the relative performance of pavement layer interface conditions. 6.2 Recommendations Based on the work completed in th is research, the following is recommended for investigation : As this was the first study developed to specifically evaluate characteristics associated with the interface between structural layers of asphalt concrete, additional surface characteristics shou ld be investigated to determine the effects they have on the breakdown of the bond including: milled vs. new, wet vs. dry, and additional tack coat types. In addition, variations of the mixture types should be evaluated. This could lead to the determinatio n of optimum materials used for structural interface condition. The gap width for guillotine type shear stress testing devices is an important factor to consider in test development. Further investigation on the influence of gap width on the results from shear bond testing should be conducted. There is limited data and understanding of effect of gap width on various types of tack coat materials and application rates. Optimizing the gap in the FIST or other shear bond devices could lead to better prediction of pavement performance. A new iteration of th e design of this device that could restrict vibration of the FIST device would provide more accurate data for analysis. Curved disc spring washers helped dampen the vibration on the system. However, a better system that reduced vibration altoget her could potentially results. This design could

PAGE 113

113 incorporate a method for fully encasing the specimen in a rigid tube, leaving only the interface free to move. Due to the difficulty experienced with managing the data ou tput files from this type of analysis in Microsoft Excel, a n algorithm that could reduce the large data sets would improve the analysis portion of this work for future research.

PAGE 114

114 APPENDIX A ASPHALT MIXTURE INFORMATION Table A 1. Dense gradation for base job mix formula (JMF) Sieve Size Percent Passing 33% 7% 50% 10% 100% # 78 Stone #89 Stone W 10 Screenings Sand JMF 100.0 100.0 100.0 100.0 100 97.0 100.0 100.0 100.0 99.0 58.5 100.0 100.0 100.0 86.4 #4 8.1 30.1 100.0 100.0 65.1 #8 3.0 4.0 70.2 100.0 46.6 #16 1.0 2.0 42.4 94 .0 31.2 #30 1.0 1.0 25.5 53 .0 18.5 #50 1.0 1.0 15.9 11 .0 9.5 #100 1.0 1.0 9.9 3 .0 5.7 #200 1.0 1.0 7.0 0 .0 3.9 Gsb 2.809 2.799 2.770 2.626 2.770 Table A 2. Dense g radation for base (cumulative) Sieve Siz e Retained Weight, g # 78 Stone #89 Stone W 10 Screenings Sand 0.0 1471.0 1785.1 4050.0 44.6 1471.0 1785.1 4050.0 609.8 1471.0 1785.1 4050.0 #4 1352.3 1690.6 1785.1 4050.0 #8 1426.5 1772.5 2460.1 4050.0 #16 1456.2 1778.8 3090.1 40 50 0 #30 1456.2 1782.0 3472.6 4077.0 #50 1456.2 1782.0 3690.0 4261.5 #100 1456.2 1782.0 3825.0 4450.5 #200 1456.2 1782.0 3892.5 4486.5 Pan 1471.0 1785.1 4050.0 4500.0

PAGE 115

115 Table A 3. Dense gradation for overlay (cumulative) Table A 4. Volumetrics for GA I2 g ranite Table A 5. Specific gravity for GA I2 material s. Sieve Size Retained Weig ht, g rams # 78 Stone #89 Stone W 10 Screenings Sand 0.0 799.2 969.8 2200.4 24.2 799.2 969.8 2200.4 331.3 799.2 969.8 2200.4 #4 734.7 918.5 969.8 2200.4 #8 775.0 963.0 1336.6 2200.4 #16 791.2 966.4 1678.8 2200.4 #30 791.2 968.2 1886. 7 2215.0 #50 791.2 968.2 2004.8 2315.3 #100 791.2 968.2 2078.1 2418.0 #200 791.2 968.2 2114.8 2437.5 Pan 799.2 969.8 2200.4 2444.8 G mm G b P b G sb G se P ba P be VMA (%) VFA (%) V a DP 2.579 1.03 4.8 2.770 2.791 0 .3 4.5 14.9 73.2 4.0 0.93 Stone G sb #78 Stone 2.809 #89 Stone 2.799 W 10 Screenings 2.770 Local Sand 2.626

PAGE 116

116 APPENDIX B OVERLAY MATERIAL CALCULATION To calculate the amount of overlay material needed to reach 2 .2 inches the following equation was used. Where, It is important to note that all lengths are converted to cm for calculation. Therefore, To determine the mass of the aggregate required for the overlay, the following calculation is required:

PAGE 117

117 APPENDIX C APPLICATION RATE CALCULATION To calculate the mass equivalency of the application rate of tack coat used for a Superpave gyratory compacted specimen the followin g formulas were used: Calculate the surface area of the pill where the tack coat will be applied: Convert the application rate from gal/yd 2 to centimeters: The equivalent mass of tack coat applied is then, Where G b is the specific gravity of the tack coat. This value can be found from the fact sheets provided by the source of the tack coat material. A typical value for G b is around factored into the equation to determine the appropriate mass. T herefore the final equation used for calculation is as follows: For the trackless specimen, with a 0.06 application rate, G b =1.03 and 50% binder residue, the mass of emulsion applied to the base is:

PAGE 118

118 APPENDIX D SHEAR STRENGTH TEST DATA Table D 1. Maximum s hear s trength d ata Specimen ID Maximum Shear Strength (lbf) Maximum Shear Stress (psi) NT S1 3400.4 124.1 NT S2 3563.1 130.1 NT S3 3436.9 125.5 TT S1 3110.0 113.5 TT S2 3687.4 134.6 TT S3 3596.1 131.3 PT S1 2669.5 97.5 PT S2 2856.6 104.3 PT S3 3209.3 117.2 CTL S1 3567.7 130.3 CTL S2 3435.1 125.4 CTL S3 3385.4 123.6 CTH S1 2721.0 99.3 CTH S2 2642.9 96.5 CTH S3 2907.0 106.1

PAGE 119

119 Figure D 1. Shear stress of specimens with no tack coat. Figure D 2. Shear stress of specimens with t rackless tack coat. 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear Stress (psi) Displacement (in) NT-S1 NT-S2 NT-S3 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear Stress (psi) Displacement (in) TT-S1 TT-S2 TT-S3

PAGE 120

120 Figure D 3 Shear stress of specimens with PMAE tack coat Figure D 4. S hear stress of specimens with conventional tack coat applied at a low rate. 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear Stress (psi) Displacement (in) PT-S1 PT-S2 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear Stress (psi) Displacement (in) CTL-S1 CTL-S2 CTL-S3

PAGE 121

121 Figure D 5. Shear stress of specimens with conventional tack coat applied at a high rate. 0 20 40 60 80 100 120 140 160 0 0.1 0.2 0.3 0.4 0.5 0.6 Shear Stress (psi) Displacement (in) CTH-S1 CTH-S2 CTH-S3

PAGE 122

1 22 APPENDIX E REPEATED SHEAR STRESS TEST DATA Figure E 1. Trial LC RST resilient deformation data at 1500.0 lbf. Figure E 2. Trial LC RST stiffness data at 1500.0 lbf. 0.006 0.007 0.008 0.009 0.010 0.011 0.012 0.013 0.014 0.015 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Resilient Deformation (in) Time (seconds) NT TT PT 100000.0 120000.0 140000.0 160000.0 180000.0 200000.0 220000.0 240000.0 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Stiffness (lbf/in) Time (seconds) NT TT PT

PAGE 123

123 Figure E 3 Trial LC RST normalized stiffness data at 1500.0 lbf. Figure E 4 8 hour LC RST progressive loading trial resilient deformation data. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 5000.0 10000.0 15000.0 20000.0 25000.0 30000.0 Normalized Stiffness (dimensionless) Time (seconds) NT TT PT 0.005 0.007 0.009 0.011 0.013 0.015 0.0 7200.0 14400.0 21600.0 28800.0 Resilient deformation (in) Time (sec) NT TT PT

PAGE 124

124 Figure E 5 8 hour LC RST progressive loading trial stiffness data. Figure E 6 8 hour LC RST progressive loading trial normalized stiffness data. 150000 170000 190000 210000 230000 0.0 7200.0 14400.0 21600.0 28800.0 Stiffness (lbf/in) Time (sec) NT TT PT 0.600 0.700 0.800 0.900 1.000 1.100 0.0 7200.0 14400.0 21600.0 28800.0 Normalized stiffness (dimensionless) Time (sec) NT TT PT

PAGE 125

125 Figure E 7 12 hour LC RST progressive loading trial resilient deformation data. Figure E 8 12 hour LC RST progr essive loading trial stiffness data. 0.005 0.007 0.009 0.011 0.013 0.015 0.0 7200.0 14400.0 21600.0 28800.0 Resilient deformation (in) Time (sec) NT TT PT 140000 180000 220000 260000 300000 0.0 7200.0 14400.0 21600.0 28800.0 Stiffness (lbf/in) Time (sec) NT TT PT

PAGE 126

126 Figure E 9 12 hour LC RST progressive loading trial normalized stiffness data. 0.600 0.700 0.800 0.900 1.000 1.100 0.0 7200.0 14400.0 21600.0 28800.0 Normalized stiffness (dimensionless) Time (sec) NT TT PT

PAGE 127

127 APPENDIX F FINAL LC RST DATA Figure F 1. Permanent d eformation data from LC RST specimen NT 1. Figure F 2 Permanent d eformation data from LC RST specimen NT 2. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec)

PAGE 128

128 Figure F 3. Permanent d eformation data from LC RST specimen TT 1. Figure F 4. Permanent d eformation data from LC RST specimen TT 2. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec)

PAGE 129

129 Figure F 5. Permanent d eformation data from LC RST specimen C T L 1. Figure F 6. P ermanent d eformation data from LC RST specimen C T L 2. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec)

PAGE 130

130 Figure F 7. Permanent d eformation data from LC RST specimen C T H 1. Figure F 8. Permanent d eformation data from LC RST specimen C T H 2. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec)

PAGE 131

131 Figure F 9. Permanent d eformation data from LC RST specimen PT 1. Figure F 10. Permanent d eformation data from LC RST specimen PT 2. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec)

PAGE 132

132 APPENDIX G ENERGY ANALYSIS Figure G 1. Typical slope of LC RST permanent deformation data for initial poin t Figure G 2. Typical slope of LC RST permanent deformation dat a for final point. 0.000000 0.000005 0.000010 0.000015 0.000020 0.000025 0.000030 0.000035 0.000040 0.000045 0.000050 0.0 720.0 1440.0 2160.0 2880.0 3600.0 Slope (in/sec) Time (sec) 0.000000 0.000005 0.000010 0.000015 0.000020 0.000025 0.000030 0.000035 0.000040 0.000045 0.000050 26280.0 27000.0 27720.0 28440.0 29160.0 29880.0 30600.0 Slope (in/sec) Time (sec)

PAGE 133

133 Figure G 3. Cubic function fit to steady state of NT 1 permanent deformation data. Figure G 4. Linear regression line fit to NT 1 permanent deformation data. y = 9E 15x 3 3E 10x 2 + 8E 06x + 0.0495 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.0 10000.0 20000.0 30000.0 40000.0 50000.0 60000.0 Permanent deformation (in) Time (sec) y = 5E 06x + 0.0558 0.00 0.03 0.05 0.08 0.10 0.13 0.15 0.0 2000.0 4000.0 6000.0 8000.0 10000.0 12000.0 Permanent deformation (in) Time (sec)

PAGE 134

134 Figure G 5 Energy of LC RST specimen results Figure G 6. Average e nergy of LC RST specimen results. 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 450.0 NT 1 NT 2 TT 1 TT 2 CTL 1 CTL 2 CTH 1 CTH 2 PT 1 PT 2 Energy (in lbf) Specimen ID 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 NT TT CTL CTH PT Energy (in lbf) Specimen Type

PAGE 135

135 Figure G 7. Energy of strength test specimen results. Figure G 8. Average energy of strength specimen results. 0.0 50.0 100.0 150.0 200.0 250.0 300.0 NT TT CTL CTH PT Energy (in lbf) Specimen Type 0.0 50.0 100.0 150.0 200.0 250.0 300.0 NT TT CTL CTH PT Energy (in lbf) Specimen Type

PAGE 136

136 Figure G 9. Line of equality for energy data. 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 Energy from strength tests (in lbf) Energy from LC RST analysis (in lbf) NT TT CTL PT CTH

PAGE 137

137 Table G 1. Inflection points for each LC RST specimen. Specimen ID Inflection Point (in) NT 1 11111.1 NT 2 10000.0 TT 1 13333.3 TT 2 16666.7 C T L 1 11111.1 C T L 2 16666.7 C T H 1 11666.7 C T H 2 22222.2 PT 1 16666.7 PT 2 16666.7

PAGE 138

138 APPENDIX H STATISTICAL ANALYSIS Figure H 1. Standard deviation of averag ed LC RST results. Figure H 2. 90 percent confidence interval of averaged LC RST results. 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 450.0 NT TT CTL CTH PT Energy (in lbf) Specimen Type 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 NT TT CTL CTH PT Energy (in lbf) Specimen Type

PAGE 139

139 Figure H 3. 85 percent confidence interval of averaged LC RST results. Table H 1. LC RST ANOVA s ummary Groups Count Sum Average Variance NT 2.0 632.241 316.121 569.026 TT 2.0 548.576 274.288 199.976 C T L 2.0 671.034 335.517 3151.023 C T H 2.0 521.273 260.636 0.192 PT 2.0 726.884 363.442 1573.771 Table H 2. LC RST ANOVA Variation Source SS df MS F P value F crit Between Groups 14513.492 4 3628.373 3.302 0.111 5.192 Within Groups 5493.988 5 1098.798 Total 20007.480 9 Table H 3. F Test two sample for variances: NT to TT. Variable 1 Variable 2 Mean 316.1207 274.2879 Variance 569.026 199.9759 Observations 2 2 df 1 1 F 2.845473 P(F<=f ) one tail 0.34067 F Critical one tail 39.86346 0.0 50.0 100.0 150.0 200.0 250.0 300.0 350.0 400.0 NT TT CTL CTH PT Energy (in lbf) Specimen type

PAGE 140

140 Table H 4. F Test two sample for variances: NT to CL. Variable 1 Variable 2 Mean 316.1207 335.5171 Variance 569.026 3151.023 Observations 2 2 df 1 1 F 0.180585 P(F<=f) one tail 0.255813 F Cri tical one tail 0.025086 Table H 5. F Test two sample for variances: NT to CH. Variable 1 Variable 2 Mean 316.1207 260.6364 Variance 569.026 0.192309 Observations 2 2 df 1 1 F 2958.918 P(F<=f) one tail 0.011702 F Critical one tail 39.86346 Table H 6. F Test two sample for variances: NT to PT. Variable 1 Variable 2 Mean 316.1207 363.4419 Variance 569.026 1573.771 Observations 2 2 df 1 1 F 0.361569 P(F<=f) one tail 0.344653 F Critical one tail 0.025086 Table H 7. F Test two sa mple for variances: TT to CL. Variable 1 Variable 2 Mean 274.2879 335.5171 Variance 199.9759 3151.023 Observations 2 2 df 1 1 F 0.063464 P(F<=f) one tail 0.157108 F Critical one tail 0.025086

PAGE 141

141 Table H 8. F Test two sample for variances: TT to CH. Variable 1 Variable 2 Mean 274.2879 260.6364 Variance 199.9759 0.192309 Observations 2 2 df 1 1 F 1039.869 P(F<=f) one tail 0.019736 F Critical one tail 39.86346 Table H 9. F Test two sample for variances: TT to PT. Variable 1 Vari able 2 Mean 274.2879 363.4419 Variance 199.9759 1573.771 Observations 2 2 df 1 1 F 0.127068 P(F<=f) one tail 0.217994 F Critical one tail 0.025086 Table H 10. F Test two sample for variances: CL to CH. Variable 1 Variable 2 Mean 335.5171 26 0.6364 Variance 3151.023 0.192309 Observations 2 2 df 1 1 F 16385.22 P(F<=f) one tail 0.004973 F Critical one tail 39.86346 Table H 11. F Test two sample for variances: CL to PT. Variable 1 Variable 2 Mean 335.5171 363.4419 Variance 3151.02 3 1573.771 Observations 2 2 df 1 1 F 2.002212 P(F<=f) one tail 0.391661 F Critical one tail 39.86346

PAGE 142

142 Table H 12. F Test two sample for variances: CH to PT. Variable 1 Variable 2 Mean 260.6364 363.4419 Variance 0.192309 1573.771 Observati ons 2 2 df 1 1 F 0.000122 P(F<=f) one tail 0.007037 F Critical one tail 0.025086 Table H 13. T Test two sample assuming equal variances : NT to TT. Variable 1 Variable 2 Mean 316.1207 274.2879 Variance 569.026 199.9759 Observations 2 2 Poole d Variance 384.501 Hypothesized Mean Difference 0 df 2 t Stat 2.133377 P(T<=t) one tail 0.083252 t Critical one tail 1.885618 P(T<=t) two tail 0.166503 t Critical two tail 2.919986 Table H 14. T Test two sample assuming equal variances: N T to CL. Variable 1 Variable 2 Mean 316.1207 335.5171 Variance 569.026 3151.023 Observations 2 2 Pooled Variance 1860.024 Hypothesized Mean Difference 0 df 2 t Stat 0.44974 P(T<=t) one tail 0.348471 t Critical one tail 1.885618 P(T<=t) t wo tail 0.696941 t Critical two tail 2.919986

PAGE 143

143 Table H 15. T Test two sample assuming unequal variances: NT to CH. Variable 1 Variable 2 Mean 316.1207 260.6364 Variance 569.026 0.192309 Observations 2 2 Hypothesized Mean Difference 0 df 1 t Stat 3.288864 P(T<=t) one tail 0.093957 t Critical one tail 3.077684 P(T<=t) two tail 0.187914 t Critical two tail 6.313752 Table H 16. T Test two sample assuming equal variances: NT to PT. Variable 1 Variable 2 Mean 316.1207 363.4419 Variance 569.026 1573.771 Observations 2 2 Pooled Variance 1071.398 Hypothesized Mean Difference 0 df 2 t Stat 1.44571 P(T<=t) one tail 0.142575 t Critical one tail 1.885618 P(T<=t) two tail 0.28515 t Critical two tail 2.919986 Table H 17. T Test two sample assuming equal variances: TT to CL. t Test: Two Sample Assuming Equal Variances Variable 1 Variable 2 Mean 274.2879 335.5171 Variance 199.9759 3151.023 Observations 2 2 Pooled Variance 1675.499 Hypothesized Mean Difference 0 df 2 t Stat 1.49584 P(T<=t) one tail 0.136672 t Critical one tail 1.885618 P(T<=t) two tail 0.273344 t Critical two tail 2.919986

PAGE 144

144 Table H 18. T Test two sample assuming unequal variances: TT to CH. Variable 1 Variable 2 Mean 274.2879 2 60.6364 Variance 199.9759 0.192309 Observations 2 2 Hypothesized Mean Difference 0 df 1 t Stat 1.364585 P(T<=t) one tail 0.201305 t Critical one tail 3.077684 P(T<=t) two tail 0.402609 t Critical two tail 6.313752 Table H 19. T Test two sample assuming equal variances: TT to PT. Variable 1 Variable 2 Mean 274.2879 363.4419 Variance 199.9759 1573.771 Observations 2 2 Pooled Variance 886.8733 Hypothesized Mean Difference 0 df 2 t Stat 2.99371 P(T<=t) one tail 0.047906 t Cri tical one tail 1.885618 P(T<=t) two tail 0.095812 t Critical two tail 2.919986 Table H 20. T Test two sample assuming unequal variances: CL to CH. Variable 1 Variable 2 Mean 335.5171 260.6364 Variance 3151.023 0.192309 Observations 2 2 Hypoth esized Mean Difference 0 df 1 t Stat 1.886453 P(T<=t) one tail 0.155155 t Critical one tail 3.077684 P(T<=t) two tail 0.310309 t Critical two tail 6.313752

PAGE 145

145 Table H 21. T Test two sample assuming equal variances: CL to PT. Variable 1 Var iable 2 Mean 335.5171 363.4419 Variance 3151.023 1573.771 Observations 2 2 Pooled Variance 2362.397 Hypothesized Mean Difference 0 df 2 t Stat 0.57453 P(T<=t) one tail 0.311809 t Critical one tail 1.885618 P(T<=t) two tail 0.623619 t Cri tical two tail 2.919986 Table H 22. T Test two sample assuming unequal variances: CH to PT. Variable 1 Variable 2 Mean 260.6364 363.4419 Variance 0.192309 1573.771 Observations 2 2 Hypothesized Mean Difference 0 df 1 t Stat 3.66467 P(T<=t) one tail 0.084795 t Critical one tail 3.077684 P(T<=t) two tail 0.16959 t Critical two tail 6.313752

PAGE 146

146 LIST OF REFERENCES Al Qadi, I. L., K. I. Hasiba, A. S. Cortina, H. Ozer, and Z. Leng. Best P ractices for I mplementation of T ack C oat: Part I l aboratory s tudy Illinois Department of Transportation 2012, pp. 1 37. Al Qadi, I. L., S. H. Carpenter, Z. Leng, H. Ozer, and J. Trepanier. Tack C oat O ptimization for HMA O verlays: Accelerated P avement Test R eport Illinois Center for Transportation (IC T) 2009. Birgisson, B., J. Wang, and R. Roque. Implementation of the Florida C racking M odel into the Mecha n istic Empirical Pavement Design. Florida Department of Transportation Final Report, 2006, pp. 1 36. Canestrari, F., and E. Santagata. T emperature E ffects on the Shear Behaviour of Tack Coat Emulsions used in Flexible Pavements International Journal of Pavement Engineering Vol. 6, No. 1, 2005, pp. 39 46. Canestrari, F., G. Ferrotti, X. Lu, A. Millien, M. Partl, C. Petit, A. Phelipot Mardele, H. Pibe r, and C. Raab. Mechanical T esting of Interlayer B onding in A sphalt P avement Advances in Interlaboratory Testing and Evaluation of Bituminous Materials Vol. 9, No. RILEM State of the Art Reports, 2013, pp. 303 360. Canestrari, F., G. Ferrotti, M. Partl, and E. Santagata. Advanced T esting and Characterization of I nterlayer S hear R esistance Transportation Research Record: Journal of the Transportation Research Board No. 1929, 2005, pp. 69 78. Caro, S., E. Masad, M. Snchez Silva, and D. Little Stochastic M icromechanical M odel of the D eterioration of A sphalt M ixtures S ubject to M oisture D iffusion P rocesses International Journal for Numerical and Analytical Methods in Geomechanics Vol. 35, No. 10, 2011, pp. 1079 1097. Ch en, Y. Composite S pecimen T esting t o E valuate the E ffects of P avement L ayer I nterface C haracteristics on C racking P erformance University of Florida 2011. Dauzats, M., and A. Rampal. Mechanism of Surface C racking in W earin g C ourses Proc,. Sixth International Conference, Structural Design o f Asphalt Pavements Vol. 1, 1987, pp. 232 247. Diakhat, M., A. Millien, C. Petit, A. Phelipot Mardel, and B. Pouteau. Experimental I nvestigation of T ack Coat F atigue P erformance: Towards a n I mproved L ifetime A ssessment of P avement S tructure In terfaces Construction and Building Materials Vol. 25, No. 2, 2011, pp. 1123 1133. Diakhat, M., C. Petit, A. Millien, A. Phelipot Mardel, B. Pouteau and H. Goacolou Interface F atigue C racking in M ultilayered P avements: Experimental

PAGE 147

147 A nalysis Proc. 6 th RILEM Int ernational Conference on Cracking in Pavements Chicago, IL, 2008, pp. 649 659 Diakhat, M. A. Phelipot, A. Millien, and C. Petit. Shear Fatigue Behavior of T ack C oats in P avements Road Materials and Pavement Design Vol. 7, No. 2, 2006 pp. 201 222 Do novan, E. P., I. L. Al Qadi, a n d A. Loulizi. Optimization of T ack C oat A pplication R ate for G eocomposite M embrane on B ridge D ecks Transportation Research Record: Journal of the Transportation Research Board Vol. 1740, 2000, pp. 143 150. Erlingsson S ., S Said a nd T. McGarvey. Influence of Heavy Traffic Lateral Wander on Pavement Deterioration Proc. 4 th European Pavement and Asset Management Conference Malm Sweden, 2012 Hachiya, Y., K. Sato. Effect of T ack Coat on B onding C haracteristics at I nterfa ce between A sphalt Concrete L ayers. Proc. Eighth International Conference on Asphalt Pavements Vol. 1, 1997, pp. 349 362. Hakimzadeh, S., W. Buttlar, and R. Santarromana. Evaluation of B onding between HMA L ayers P roduced with D ifferent T ack C oat A pplicati on R ates U sing S hear type and T ension type T ests Journal of the Transportation Research Board 2012. Hakimzadeh, S., W. Buttlar, and R. Santarromana. Shear and t ension t ype t ests to e valuate b onding of Hot Mix Asphalt l ayers with d ifferent t ack c oat a ppli cation r ates Transportation Research Record: Journal of the Transportation Research Board No. 2295, 2012, pp. 54 62. Hakimzadeh, S., N. A. Kebede, and W. G. Buttlar. Comparison between O ptimum T ack C oat A pplication R ates as O btained from T ension and T or sional S hear t ype t ests Proc. 7th RILEM International Conference on Cracking in Pavements Vol. 4, No.1, 2012, pp. 287 297. Hakimzadeh, S., N. A. Kebede, W. G. Buttlar, S. Ahmed, and M. Exline. Development of F racture e nergy B ased I nterface B ond T est for A sphalt C oncrete Road Materials and Pavement Design Vol. 13, No. 1, 2012, pp. 76 87. Huang, Y.H., Pavement Analysis and Design Prentice Hall Vol. 2, 2003. Jia, X., B. Huang, and L. Li. A S implified A p proach for E valuating I nterlayer S hear R esistance i n A sphalt P avement Proc., Transportation Research Board 92nd Annual Meeting No. 13 3943, 2013. Kruntcheva, M. R., A. C. Collop, and N. H. Thom. Properties of A sphalt C oncrete Layer I nterfaces Journal of Materials in Civil Engineering Vol. 18, No. 3, 20 06, pp. 467 471.

PAGE 148

148 Masad, E., B. Muhunthan, N. Shashidhar, and T. Harman. Quantifying L aboratory C ompaction E ffects on the I nternal S tructure of A sphalt C oncrete Transportation Research Record: Journal of the Transportation Research Board No. 1681, 1999, p p. 179 185. Mehta, Y., and N. Siraj. Evaluation of I nterlayer B onding in HMA P avements Wisconsin Department of Transportation 2007. Mohammad, L. N., M. A. Elseifi, A. Bae, N. B. Patel, J. Button, and J. A. Scherocman. Optimization of T ack C oat for HMA P l acement National Cooperative Highway Research Program Vol. 712, 2012. Mohammad, L., M. Raqib, and B. Huang. Influence of A sphalt T ack C oat M aterials on I nterface S hear S trength Transportation Research Record: Journal of the Transportation Research Board No. 1789, 2002, pp. 56 65. Mohammad, L., Z. Wu, and M. A. Raqib. Investigation of the B ehavior of A sphalt T ack I nterface L ayer Federal Highway Administration 2005, pp. 1 107. M uench, S.T., and T. Moomaw De bonding of Hot Mix Asphalt P avements in Washi ngton State Washington State Department of Transportation Report No. WA RD 712.1, 2008 Myers, L. Development and P ropagation of S urface I nitiated L ongitudinal W heel P ath C racks in F lexible H ighway P avements University of Florida 2000. Myers, L., and R Roque. Top Down C rack P ropagation in B ituminous P avements and I mplications for P avement M anagement Association of Asphalt Paving Technologists Vol. 71, 2002, pp. 651 670. Patel, N. B. Factors a ffecting the I nterface S hear S trength of P avement L ayers L ouisiana State University 2010. Petit, C., A. Millien, F. Canestrari, V. Pannunzio, and A. Virgili. Experimental S tudy on S hear F atigue B ehavior and S tiffness P erformance of Warm Mix Asphalt by A dding S ynthetic W ax Construction and Building Materials Vol. 34, 2012, pp. 537 544. Piber, H., F. Canestrari, G. Ferrotti, X. Lu, A. Millien, M. Partl, C. Petit, A. Phelipot Mardelle, and C. Raab. RILEM : Interlaboratory T est on I nterlayer B onding of A sphalt P avements Proc. 7th International RILEM Symposium on Advanced Testing and Characterisation of Bituminous Materials 2009, pp. 1181 1189. Raab, C. Development of a F ramework for St andardisation of I nterlayer B ond of A sphalt P avements Technical University of Darmstadt 2010.

PAGE 149

149 Raab, C., M. Partl, and A. O A bd El Halim. Evaluation of I nterlayer S hear B ond D evices for A sphalt P avements The Baltic Journal of Road and Bridge Engineering Vol. 4, No. 4, 2009, pp. 186 195. Raab, C., M. Partl, and A. O Abd El Halim. Effect of Gap W idth on I nterlayer S hear B ond R e sults International Journal of Pavement Research and Technology Vol. 3, No. 2 20 10 pp. 79 8 5. Romanoschi, S. A., and J. B. Metcalf. Characterization of Asphalt C oncrete L ayer I nterfaces Transportation Research Record: Journal of the Transportation Res earch Board No. 1778, 2001, pp. 132 139. Ro que R. S. Chun, J. Zou, G. Lopp, and C. Villiers P rojects M onitoring Florida Department of Transportation Tallahassee, FL, Final Report BDK 75 996 06 2011 Sholar, G. A., G. C. Pa ge, J. A. Musselman, P. B. Upshaw, and H. L. Moseley. Preliminary I nvestigation of a T est M ethod to E valuate B ond S trength of B ituminous T ack C oats Journal of the Association of Asphalt Paving Technologists Vol. 73, 2004. Tashman, L., K. Nam, and T. Papa giannakis. Evaluation of the Influence of T ack C oat C onstruction F actors on the B ond S trength between P avement L ayers Washington State Department of Transportation 2006, pp. 1 83. Tozzo, C., A. D'Andrea, D. Cozzani, and A. Meo. Fatigue I nvestigation of t he I nterface S hear P erformance in A sphalt P avement Modern Applied Science Vol. 8, No. 2, 2014, pp. 1 11. S hear T ests for the E valuation of the E ffect of the N ormal L oad on the I nterface F atigue R esistance Co nstruction and Building Materials Vol. 61, 2014, pp. 200 205. Tran, N. H., J. R. Willis, and G. Julian. Refinement of the B ond S trength P rocedure and I nvestigation of a S pecification National Center for Asphalt Technology 2012, pp. 1 83. Uhlmeyer, J., K Willoughby, L. Pierce, and J. Mahoney. Top Down C racking in Washington State A sphalt Concrete Wearing Courses. Transportation Research Record: Journal of the Transportation Research Board No. 1730, 2000, pp. 110 116. Uzan J., M Livneh and Y Eshed I nvestigation of A dhesion P roperties between A sphaltic C oncrete L ayers Asphalt Paving Technology No. 47 1978 pp. 495 521 West, R. C., J. Zhang, and J. Moore. Evaluation of Bond S trength between P avement L ayers National Center for Asphalt Technology 2 005, pp. 1 58.

PAGE 150

150 Wheat, M. Evaluation of B o nd S trength at A sphalt I nterfaces Kansas State University 2007. Willis, J. R., and D. H. Timm. Forensic I nvestigation of a R ich B ottom P avement National Center for Asphalt Technology 2006, pp. 1 62. Willis, J. R ., and D. H. Timm. Forensic I nvestigation of D ebonding in R ich B ottom P avement Transportation Research Record: Journal of the Transportation Research Board, No. 2040, 200 7 pp. 1 07 114

PAGE 151

151 BIOGRAPHICAL SKETCH Jeremy Alexis Magruder Waisome was born in Orla ndo, Florida in 1987. She became passionate about pursing her doctoral degree in middle school, while standing in a nuclear reactor at Oak Ridge National Laboratory. After graduating from Edgewater High School in 2005, she attended the University of Florid a where she received a Bachelor of Science degree in civil engineering in May 2010 and was inducted into the Hall of Fame. Upon graduating, she immediately enrolled in the doctoral program at the University of Florida. She received her Master of Science de gree in civil engineering in May 2012, in addition to a graduate certificate in s ustainable e ngineering in May 2013. She has been co advised throughout her graduate studies by Drs. Reynaldo Roque (chair) and Mang Tia (co chair). Upon completing her Ph.D., she plans to pursue a career in the academy and work towards improving the diversity of the engineering pipeline. T