RELATIONSHIPS BETWEEN LABORATORY MEASURED CHARACTERISTICS OF HMA AND FIELD COMPACTABILITY Except where referenced is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. Fabricio Leiva-Villacorta Certificate of Approval: David Timm Assistant Professor Civil Engineering Elton Ray Brown, Chair Director National Center for Asphalt Technology Randy West Assistant Director National Center for Asphalt Technology Joe F. Pittman Interim Dean Graduate School RELATIONSHIPS BETWEEN LABORATORY MEASURED CHARACTERISTICS OF HMA AND FIELD COMPACTABILITY Fabricio Leiva-Villacorta A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn University December 17, 2007 iii RELATIONSHIPS BETWEEN LABORATORY MEASURED CHARACTERISTICS OF HMA AND FIELD COMPACTABILITY Fabricio Leiva-Villacorta Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights Signature of Author Date of Graduation iv VITA Fabricio Leiva-Villacorta was born August 20, 1979, in Heredia, Costa Rica. He graduated with a Bachelors of Science Degree in Civil Engineering from the University of Costa Rica in 2003. After graduation he worked with the National Laboratory of Materials ? University of Costa Rica for 2 years. He also graduated with a Masters Degree in Business Administration from the State University at Distance in 2005 while worked in Costa Rica. In fall 2005 he began his studies as a graduate student at Auburn University in pursuit of a Masters of Science degree in Civil Engineering. v THESIS ABSTRACT RELATIONSHIPS BETWEEN LABORATORY MEASURED CHARACTERISTICS OF HMA AND FIELD COMPACTABILITY Fabricio Leiva-Villacorta Master of Science, December 17, 2007 (MBA, UNED-Costa Rica, 2005) (B.S., University of Costa Rica, 2003) 159 Typed Pages Directed by E. Ray Brown Compactability of HMA mixtures is often used to describe how easy or difficult a mixture is to compact on a roadway. Several asphalt researchers have proposed laboratory measured parameters of mixtures and/or their components as indicators of HMA compactability and/or resistance to permanent deformation. However, most of these measured characteristics have not been validated with actual field performance. The first part of this study includes a comparison between the laboratory compactability parameters Compaction Energy Index (CEI), number of gyrations to reach 92% of G mm (N@92%G mm ), Slope, Locking Point and Bailey Method ratios. The data used for this stage came from Superpave mixtures placed on the NCAT Test Track in the vi first two cycles (quality control samples). It was found that CEI, N@92%G mm , Slope, Locking Point can be used to represent the applied energy to reach a level of compaction in the SGC. The second part of this study includes the determination of a field compactability indicator based on rolling operation (Accumulated Compaction Pressure ? ACP) and correlation between this indicator and laboratory parameters. When all the combined data were used to correlate ACP and lab compactability parameters, the values of simple linear correlation (R-value) were always near zero. The results showed that t/NMAS and temperature significantly affected the applied compactive effort to reach the post- construction density level. The third part of this project includes compaction of specimens using the SGC at to meet the 8% air voids at thicknesses equal to those in the field. A multiple regression analysis showed that eighty two percent of the variability in the ACP can be explained by four predictors: PCSI, FAc ratio, lift temperature and number of gyrations to reach the post construction density level at lift thickness (N@field-density). The last part of this study involved density testing during the rolling operation. The purpose of this part was to determine the field compaction energy required to produce the same level of density as samples compacted in the laboratory and correlate that energy with laboratory compaction parameters. A multiple regression analysis provided a model with ACP@92%G mm as the response, while ninety two percent of the variability in the response can be explained by the interaction temperature*thickness, % passing No 200 sieve, actual PG grade, slope, locking point/Slope ratio, FAc ratio and PCSI square. vii ACKNOWLEDMENTS The author would like to thank Dr. Randy West and Dr. E. Ray Brown for all of their support in this endeavor. Also the author would like to express his gratitude for all the support he received from the staff at the National Center for Asphalt Technology. Finally, to his wife Adriana, very special thanks are given for all of her patience and hard work. viii Style manual used: Proceedings, Association of Asphalt Paving Technologists Computer software used: Microsoft Word, Microsoft excel, Minitab, Pine Pave. ix TABLE OF CONTENTS LIST OF TABLES...........................................................................................................xiii LIST OF FIGURES .......................................................................................................... xv CHAPTER 1. INTRODUCTION ....................................................................................... 1 1.1 BACKGROUND............................................................................................................ 1 1.2 OBJECTIVES ............................................................................................................... 2 1.3 SCOPE ........................................................................................................................ 3 CHAPTER 2. LITERATURE REVIEW ............................................................................ 5 2.1 FIELD COMPACTION.................................................................................................... 5 2.1.1 Introduction........................................................................................................ 5 2.1.2 Aggregate characteristics................................................................................... 6 2.1.3 Environmental conditions .................................................................................. 7 2.1.4 Binder characteristics......................................................................................... 7 2.1.5 Compaction equipment and roller operation...................................................... 8 2.1.6 Gradation............................................................................................................ 9 2.1.7 Lift thickness.................................................................................................... 11 2.2 LABORATORY COMPACTION .................................................................................... 15 2.2.1 Introduction...................................................................................................... 15 2.2.2 Potential effect of the internal angle of gyration ............................................. 16 x 2.3 LABORATORY MIX PARAMETERS USED TO DESCRIBE COMPACTABILITY ................... 16 2.3.1 Introduction...................................................................................................... 16 2.3.2 The percentage of maximum theoretical specific gravity at N ini (%G mm @N ini ). ................................................................................................................................... 17 2.3.3 Compaction Energy Index (CEI) ..................................................................... 17 2.3.4 Compaction slope determined from compaction in the Superpave Gyratory Compactor (SGC). .................................................................................................... 18 2.3.5 The number of gyrations with the SGC to reach the Locking Point of the mixture. ..................................................................................................................... 20 2.3.6 Other densification indices .............................................................................. 21 2.4 AGGREGATE CHARACTERISTICS RELATED TO COMPACTABILITY .............................. 23 2.4.1 Introduction...................................................................................................... 23 2.4.2 Bailey Method ratios........................................................................................ 24 2.4.3 Primary Control Sieve Index PCSI.................................................................. 28 2.5 STUDIES THAT RELATE LABORATORY CHARACTERISTICS AND FIELD COMPACTABILITY ....................................................................................................................................... 29 2.5.1 C-value method................................................................................................ 29 2.5.2 The modified Mohr method............................................................................. 31 2.5.3 The k-factor method......................................................................................... 32 2.6 SUMMARY OF FINDINGS ........................................................................................... 33 CHAPTER 3. RESEARCH PLAN................................................................................... 36 3.1 OVERVIEW ............................................................................................................... 36 3.2 ANALYSIS OF TEST TRACK MIXES ............................................................................ 39 xi 3.2.1 Material and Mixture Properties ...................................................................... 39 3.2.2 Test Track construction.................................................................................... 39 3.3 ACCUMULATED COMPACTION PRESSURE (ACP) AS A FIELD COMPACTABILITY INDICATOR ..................................................................................................................... 43 3.4 LABORATORY COMPACTION PARAMETERS ............................................................... 45 3.5 CORRELATION OF ACP AND LABORATORY PARAMETERS......................................... 46 3.6 MATERIAL AND MIXTURE PROPERTIES FOR VALIDATION ANALYSIS ....................... 50 CHAPTER 4. RESULTS AND ANALYSES .................................................................. 52 4.1 CONCEPTUAL HYPOTHESIS FOR EXPLAINING EXPECTED TRENDS .............................. 52 4.2 LABORATORY COMPACTION PARAMETERS ............................................................... 53 4.2.1 Parameters obtained from Densification Curve............................................... 54 4.2.2 Gradation parameters as indicators of compactability..................................... 61 4.2.3 Comparison between laboratory measured characteristics of HMA ............... 64 4.2.4 Effect of physical properties on the SGC parameters...................................... 68 4.2.5 Assessment of variability among observations (multivariate statistical analysis) ................................................................................................................................... 72 4.3 FIELD COMPACTION.................................................................................................. 79 4.3.1 Conceptual hypothesis for explaining expected trends.................................... 80 4.3.2 Analysis of the Accumulated compaction Pressure......................................... 82 4.3.3 Analysis of pairs .............................................................................................. 88 4.4 CORRELATIONS BETWEEN ACP AND LABORATORY COMPACTION PARAMETERS ...... 92 4.5 COMPACTION OF SPECIMENS USING THE SGC AT FIELD THICKNESS ......................... 97 4.6 CORRELATIONS BETWEEN ACP@92%GMM AND LAB COMPACTION PARAMETERS 102 xii 4.7 SUMMARY OF FINDINGS ......................................................................................... 105 4.8 APPLICABILITY OF THE ACP CONCEPT FOR VALIDATION PURPOSES ....................... 108 CHAPTER 5. CONCLUSIONS AND RECOMMENDATIONS.................................. 111 REFERENCES ............................................................................................................... 114 APPENDIX A. MATERIAL PROPERTIES................................................................. 118 APPENDIX B. COMPACTION INFORMATION........................................................ 132 APPENDIX C. INFORMATION OF FIELD AND LABORATORY STUDY ............ 136 xiii LIST OF TABLES Table 2.1: Definition of Fine- and Coarse-Graded Mixes (9) ......................................... 10 Table 2.2: Recommended lift thickness for HMA mixes (9)........................................... 11 Table 2.3: Summary of minimum t/NMAS to provide 7 % air voids in laboratory (2) ... 13 Table 2.4: Recommended ranges of aggregate Ratios...................................................... 27 Table 2.5: Recommended ranges of aggregate Ratios for SMA mixes............................ 28 Table 2.6: Control sieves .................................................................................................. 28 Table 3.1: Summary of the experimental plan.................................................................. 38 Table 3.2: Summary of mix types evaluated .................................................................... 40 Table 3.3: Compaction Pressure per Pass (psi)................................................................. 45 Table 3.4: Properties of NCAT Test Track mixtures 2nd and 3rd cycle .......................... 48 Table 3.5: Data from NCHRP 9-27 mixtures ................................................................... 51 Table 4.1: Single correlation among parameters .............................................................. 65 Table 4.2: Principal effects - MANOVA analysis............................................................ 71 Table 4.3: Canonical functions on canonical correlation (*).......................................... 74 Table 4.4: Factorial experiment analysis ......................................................................... 76 Table 4.5: Cluster separation and properties.................................................................... 78 Table 4.6: Post-construction density level (%G mm ) sorted by Gradation Type and Test Track cycle........................................................................................................................ 80 xiv Table 4.7: Description of levels per factor used in analysis of variance .......................... 86 Table 4.8: ANOVA for ACP ............................................................................................ 87 Table 4.9: Comparison of ACP by various subjects......................................................... 89 Table 4.10: Comparison of CEI by various subjects ........................................................ 91 Table 4.11: Single correlation between ACP and some parameters................................. 93 Table 4.12: Correlation and expected trend between ACP and compactability parameters including mix properties (Tangents only, cycle 1) ........................................................... 94 Table 4.13: Correlation and expected trend between ACP@92% of G mm and compactability parameters .............................................................................................. 103 Table 4.14: Summary of models used to correlate field and laboratory compactability 107 Table 4.15: Single correlation among laboratory parameters based on NCHRP 9-27 mixtures........................................................................................................................... 109 xv LIST OF FIGURES Figure 2.1: 9.5 mm NMAS Superpave Gradations (2).....................................................10 Figure 2.2: Relationships of t/NMAS and Air Voids for Superpave Mixes (2)...............12 Figure 2.3: Relationships of Gradations and laboratory Air Voids for Superpave Mixes (2)...................................................................................................................................... 13 Figure 2.4: Relationship of Air Voids and Thickness for 19.0 mm Coarse-Graded with Modified Asphalt (2)......................................................................................................... 14 Figure 2.5: Schematic of the Superpave Gyratory Compactor..........................................15 Figure 2.6: Illustration of CEI and TDI Indices (4)...........................................................18 Figure 2.7: Illustration of Compaction Slope (14).............................................................19 Figure 2.8: Effect of mixture properties on compaction slope (15). .................................20 Figure 2.9: Energy Indices CEI and TDI; CFI and TFI (18).............................................22 Figure 2.10: Correlation between CEI and CFI; TDI and TFI (18)...................................23 Figure 2.11: Illustration of the PCSI for 12.5 mm NMAS gradations ..............................29 Figure 2.12: The compaction process according to the C-value method. .........................30 Figure 3.1: Frequency distribution of test track layer thicknesses. ...................................43 Figure 3.2: Contact arc of the drum on the mat.................................................................45 Figure 3.3: Relationship between nuclear density (pcf), lift temperature, compaction time and roller pattern............................................................................................................... 49 xvi Figure 3.4: Relationship between ACP and density (%G mm ). ...........................................49 Figure 3.5: Illustration of the model used to fit field compaction data. ............................50 Figure 4.1: %G mm @N ini sorted by Gradation Type. ..........................................................55 Figure 4.2: Relationship between %G mm @N ini and asphalt content..................................56 Figure 4.3: Relationship between %G mm @N ini and PCSI. ................................................56 Figure 4.4: Compaction Energy Index (CEI) sorted by Gradation Type. .........................57 Figure 4.5: Relationship between PCSI and %G mm @N ini and PCSI and CEI for SMA mixtures............................................................................................................................. 58 Figure 4.6: Traffic Densification Index (TDI 92-96) sorted by Ndes. ..............................59 Figure 4.7: Compaction Slope sorted by Gradation Type. ................................................60 Figure 4.8: N @ Locking Point sorted by Gradation Type. ..............................................61 Figure 4.9: CA Ratio sorted by Gradation Type................................................................63 Figure 4.10: FAc Ratio sorted by Gradation Type. ...........................................................63 Figure 4.11: Relationship between %Gmm@Nini and CEI..............................................67 Figure 4.12: Relationship between N@92%Gmm and CEI ..............................................67 Figure 4.13: Post-construction density level (%G mm ) sorted by Gradation Type. ............79 Figure 4.14: ACP sorted by Gradation Type.....................................................................83 Figure 4.15: ACP sorted by Gradation Type and cycle.....................................................84 Figure 4.16: Thickness and T/NMAS ratio for each cycle................................................85 Figure 4.17: Temperature measured at different compaction stages.................................85 Figure 4.18: Main Effects Plot (fitted means) for ACP.....................................................87 Figure 4.19: Interaction Plot (fitted means) for ACP. .......................................................88 Figure 4.20: Comparison of ACP by various subjects. .....................................................90 xvii Figure 4.21: Comparison of CEI by various subjects........................................................91 Figure 4.22: Effect of thickness on ACP. ..........................................................................93 Figure 4.23: Relationship between ACP and N@92%G mm ...............................................95 Figure 4.24: Relationship between ACP and number of cycles to reach 92%Gmm and field density....................................................................................................................... 98 Figure 4.25: Compaction of lab specimens at 50 mm. ......................................................99 Figure 4.26: Compaction of lab specimens at 100 mm. ....................................................99 Figure 4.27: Effect of reducing thickness on lab specimens ...........................................100 Figure 4.28: Comparison of predicted ACP for test track sections and NCHRP 9-27 projects using Equation 18.............................................................................................. 110 1 CHAPTER 1. INTRODUCTION 1.1 Background Meeting the specified density of HMA is often cited as a difficult challenge for asphalt pavement construction (1). Since the introduction of Superpave designed mixtures this challenge has become a bigger issue. Overall, Superpave designed mixes have been cited as more difficult to compact than Marshall/Hveem mixes and greater compactive efforts have been needed to achieve similar density levels (1). Data from research projects such as NCHRP 9-27 (2) and NCHRP 9-9 (3) have also shown that density was less than desirable for many field projects. The importance of achieving a well compacted pavement is crucial to avoiding problems with numerous types of distresses including permanent deformation, moisture damage, and cracking. Numerous factors affect the contractor?s ability to achieve the target density for HMA mixtures, including weather, support of underlying layers, layer thickness, compaction equipment, experience of roller operators and mixture characteristics. Compactability of HMA mixtures is often used to describe how easy or difficult a mixture is to compact on a roadway. Several asphalt researchers (4, 5) have proposed laboratory measured parameters of mixtures and/or their components as indicators of 2 HMA compactability and/or resistance to permanent deformation. However, most of these measured characteristics have not been validated with actual field performance. 1.2 Objectives The primary objective of this research was to evaluate a variety of mixture characteristics and determine if they are correlated to compactability in the field. The mixture characteristics included in the evaluation were: 1. The percentage of maximum theoretical specific gravity at N ini (%G mm @ N ini ). 2. Compaction slope determined from compaction in the Superpave Gyratory Compactor (SGC). 3. Number of gyrations to achieve 92% of the maximum theoretical specific gravity (N@92%G mm ). 4. The Compaction Energy Index determined from the SGC compaction process as recommended by Bahia (4). 5. The number of gyrations with the SGC to reach the Locking Point of the mixture. 6. The Coarse and Fine Aggregate Ratios as determined using the Bailey Method recommended by Pine (5). 7. Mix parameters such as gradation, aggregate shape, binder grade, and mix volumetric properties. 8. The Primary Control Sieve Index PCSI, which is the difference in percent passing from gradation to primary control sieve. It represents the relative coarseness or fineness of the gradation. 3 A second objective was to explain why some mixtures are more compactable than others using basic mix parameters such as gradation, aggregate shapes, binder grade, and mix volumetric properties. This analysis included parameters obtained from quality control samples and specimens compacted to the field lift thickness. The underlying goal of this study was to identify a practical method to evaluate the compactability of an HMA mixture in the laboratory for use by mix designers and quality control technicians to help them achieve suitable levels of density in the field. 1.3 Scope To accomplish the objectives of this study, a literature review was completed to understand the different parameters used to measured compactability in the lab. HMA mixtures placed on the NCAT Test Track in the first two cycles were used to calculate those parameters (Thirty-five different surface mixtures and seven binder mixtures placed on the track in 2000, seventeen surface mixtures and twenty-two binder mixtures placed in 2003 were used in this analysis). The data used to determine the laboratory measured mix characteristics were obtained from quality control samples taken during track construction. Triplicate gyratory samples were compacted for each section. Compaction operations at the track were well documented and provided good information about the compactability of the mixtures in the field. These data were used to determine the total compaction energy applied by the rollers during construction. 4 Each of the mixture parameters listed in the objectives were calculated from the quality control samples taken during construction. Statistical analyses were used to describe the relationships among these parameters. The effect of different mix properties over these parameters was also evaluated. Regression between these parameters and the field compaction energy were analyzed. The laboratory measured parameters which yield the best correlations were analyzed further by performing multiple stepwise regressions with basic mixture properties. 5 CHAPTER 2. LITERATURE REVIEW This chapter presents an overview of the research that has been conducted in the following areas: 1. Factors affecting field compaction, 2. Laboratory compaction, 3. Laboratory mix parameters used to describe compactability, 4. Aggregate characteristics related to compactability, 5. Studies that relates laboratory characteristics and field compactability 2.1 Field compaction 2.1.1 Introduction Compaction is the process by which the volume of air in an HMA mixture is reduced by the application of external forces to reorient the constituent aggregate particles into a more closely spaced arrangement. The reduction of air voids in a mixture produces an increase in HMA unit weight (6). HMA compaction is influenced by many factors; some related to the environment, some determined by mix and structural design and some under contractor and agency control during construction. Some of the most important 6 factors that affect field compaction include aggregate characteristics, environmental conditions, compaction equipment and roller operation, gradation and lift thickness. 2.1.2 Aggregate characteristics Aggregate gradation influences key HMA parameters such as stiffness, stability, durability, permeability, workability, fatigue resistance, frictional resistance and resistance to moisture damage. The maximum aggregate size can be influential in compaction and lift thickness determination (6). Coarse aggregate (aggregate retained in the No. 4 sieve). Surface texture, particle shape and the number of fractured faces can affect compaction. Rough surface texture, cubical or block shaped aggregate and highly angular particles will all increase the required compactive effort to achieve a specific density (6). Midsize fine aggregate (between the 0.60 and 0.30-mm (No. 30 and No. 50) sieves). High amounts of midsize fine, rounded aggregate cause a mix to displace laterally or shove under roller loads. This occurs because the excess midsize fine, rounded aggregate results in a mix with insufficient voids in the mineral aggregate (VMA). This provides only a small void volume available for the binder to fill. Consequently, if the binder content is slightly high, it completely fills the voids and the excess binder serves to resist compaction by forcing the aggregate apart and lubricate the aggregate making it easy for the mix to laterally displace (6). Fines or dust (aggregate passing the 0.075-mm (No. 200) sieve). Generally, a mix with high fines content will be more difficult to compact than a mix with low fines content. Gradations with excessive fines cause distortion because the large amount of fine 7 particles tend to push the larger particles apart and act as lubricating ball-bearings between these larger particles (6). 2.1.3 Environmental conditions HMA temperature has a direct effect on the viscosity of the asphalt binder and thus compaction. As HMA temperature decreases, its asphalt cement binder becomes more viscous and resistant to deformation for a given compactive effort. The major environmental conditions affecting field compactability are (6): ? Initial mat temperature. Higher initial mat temperatures require more time to cool down, which means more time available for compaction, but to high temperatures may damage the binder and make the mix tender during compaction. ? Temperature of the surface on which the mat is placed. Cooler surfaces will remove heat from the mat at a faster rate, decreasing the time available for compaction. ? Ambient temperature. Hotter air temperatures will remove heat from the mat at a slower rate, increasing the time available for compaction. ? Wind speed. Lower wind speeds will decrease mat heat loss by convection, which will increase the time available for compaction. 2.1.4 Binder characteristics Asphalt binders with lower PG grade tend to deform more easily under load. Modified asphalt binders tend to have higher shear stiffness and lower permanent shear strain; in other words, they tend to increase resistance to permanent deformation. 8 The asphalt binder grade affects compaction through its viscosity. A binder that has higher viscosity will generally result in a mix that is more resistant to compaction (6). Mixes with low asphalt content are generally difficult to compact because of inadequate lubrication, whereas mixes with high asphalt content will be easier to compact. Since the viscosity of asphalt is highly temperature dependent, the temperature of the mix therefore affects its compactability. 2.1.5 Compaction equipment and roller operation Compaction is done by any of several types of compactors or rollers, which reduce the volume of air in the mix and increase in HMA unit weight or density. There are three basic pieces of equipment available for HMA compaction: 1) the paver screed, 2) the steel wheeled roller (including vibratory rollers) and 3) the pneumatic tire roller. The type and operational characteristics of rolling equipment can affect the level of density obtained in the asphalt concrete mix. For steel wheel rollers, a greater roller mass will result in more change in the degree of density per roller pass. Vibratory rollers use dynamic force to increase the compaction energy per pass. For pneumatic tire rollers, the compactive effort applied to the mix is a function of the wheel load and the tire pressure (7, 8). In terms of roller operations, a number of variables affect the ability of the compaction equipment to adequately densify the mix. Operating at lower speeds allows the roller to remain in contact with a particular mat location longer than it would at higher speeds. Lowering equipment speed increases the shearing stress. Higher shearing stresses are more capable of rearranging aggregate into more dense configurations (7). 9 Earlier roller passes over hotter (as long as not too hot) HMA will increase density (decrease air voids) more than later passes over cooler HMA. Adding rollers can be used to increase the number of roller passes in a given time. 2.1.6 Gradation Gradation is one of the most influential aggregate characteristics affecting HMA properties and performance. The aggregate size distribution influences almost every important property of asphalt mixes including volumetrics, stiffness, stability, durability, permeability, workability, fatigue resistance, frictional resistance and resistance to moisture damage (1). The simplest definition of fine and coarse gradations establishes a gradation that, when plotted on the 0.45 power gradation graph, falls mostly above (fine) or below (coarse) the 0.45 power maximum density line. These terms generally apply to dense graded aggregate. Dense or well-graded refers to a gradation that is near the 0.45 power curve for maximum density. Many research studies, involving HMA gradations, have identified fine-graded and coarse-graded mixtures based on the definition given by the National Asphalt Pavement Association (NAPA) (9). Percent passing certain sieve sizes for a given Nominal Maximum Aggregate Size (NMAS) is used to define fine-graded and coarse- graded mixes as shown in Table 2.1. Other studies have used definitions based on the location of the gradation curve with respect to the maximum density line and the restricted zone (3). Figure 2.1 illustrates 9.5 mm NMAS Superpave gradations where BRZ, ARZ and TRZ stand for below, above and through the restricted zone, respectively. However, since the restricted zone has been eliminated from AASHTO specifications; coarse, fine and intermediate-graded (medium- graded) mixtures are more commonly used (2). Table 2.1: Definition of Fine- and Coarse-Graded Mixes (9) Mixture NMAS Coarse-Graded Fine-Graded 37.5 mm (1 1/2") < 35% Passing 4.75mm Sieve > 35% Passing 4.75mm Sieve 25.0 mm (1") < 40% Passing 4.75mm Sieve > 40% Passing 4.75mm Sieve 19.0 mm (3/4") < 35% Passing 2.36mm Sieve > 35% Passing 2.36mm Sieve 12.5 mm (1/2") < 40% Passing 2.36mm Sieve > 40% Passing 2.36mm Sieve 9.5 mm (3/8") < 45% Passing 2.36mm Sieve > 45% Passing 2.36mm Sieve 4.75 mm (No. 4 Sieve) N/A (No Standard Superpave Gradation) 9.5 mm NMAS Superpave Gradations 0 10 20 30 40 50 60 70 80 90 100 Sieve Size, mm P e r c e n t P assin g Control Points Resricted Zone BRZ ARZ TRZ 0.075 0.30 0.60 1.18 2.36 4.75 9.5 12.5 Figure 2.1: 9.5 mm NMAS Superpave Gradations (2). 10 11 2.1.7 Lift thickness Prior to Superpave implementation, the rule of thumb for lift thickness was two times the maximum aggregate size which is approximately equivalent to three times the NMAS. Table 2.2, from the ?HMA Pavement Mix Type Selection Guide? (9), presents the recommended minimum lift thickness for various mixes. Overall, for fine-graded mixes the recommended t/NMAS ratio is from 2.4 to 5.0 and for coarse-graded mixes a range from 3.0 to 6.0. Dense-graded is in an intermediate range between fine-graded and coarse-graded mixes. In terms of minimum thickness, thicker layers are recommended for coarse-graded mixes than for fine-graded mixes. Table 2.2: Recommended lift thickness for HMA mixes (9) Mix Minimum t/NMAS Minimum Thickness, mm 4.75 mm dense-graded 2.6 ? 4.0 12.5 ? 19.0 9.5 mm fine-graded 2.6 ? 3.9 25.0 ? 37.5 9.5 mm coarse-graded 3.4 ? 5.3 32.0 ? 50.0 12.5 mm fine-graded 2.4 ? 5.0 30.0 ? 62.5 12.5 mm coarse-graded 3.0 ? 6.0 37.5 ? 75.0 19.0 mm fine-graded 2.6 ? 3.7 50.0 ? 70.0 25.0 mm dense-graded 3.0 ? 4.0 75.0 ? 100.0 37.5 mm dense-graded 2.7 ? 4.0 100.0 ? 150.0 9.5 mm SMA 2.6 ? 3.9 25.0 ? 37.5 12.5 mm SMA 3.0 ? 4.0 37.5 ? 50.0 19.0 mm SMA 2.6 ? 3.9 50.0 ? 75.0 Figure 2.2 was obtained from the NCHRP 9-27 study (2) and shows the impact t/NMAS on the air voids using the gyratory compactor. The figure indicates that as the t/NMAS increases, the air voids decrease for a given NMAS. Figure 2.3 shows a general trend for Superpave mixtures for a given NMAS. ARZ mixes (Above the Restriction Zone mixes ? Fine-Graded mixes) had the lowest air voids compared to the TRZ and BRZ mixes. This result suggested that fine-graded mixes are easier to compact compared to coarse-graded. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 2:1 3:1 4:1 8:1 2:1 3:1 4:1 2:1 2.5:1 3:1 9.5 mm NMAS 19.0 mm NMAS 37.5 mm NMAS t/NMAS A ver age A i r vo i d s, % Figure 2.2: Relationships of t/NMAS and Air Voids for Superpave Mixes (2). Table 2.3 is the summary of minimum t/NMAS to provide 7 % air voids using the SGC for NCHRP 9-27 study. The results show that as the NMAS increases the minimum t/NMAS decreases and fine-graded mixes have lower desired t/NMAS values than the coarse-graded mixes. 12 0.0 2.0 4.0 6.0 8.0 10.0 12.0 ARZ BRZ TRZ ARZ BRZ TRZ ARZ BRZ TRZ 9.5 mm NMAS 19.0 mm NMAS 37.5 mm NMAS Gradation A v e r age A i r V o i d s , ( % ) Figure 2.3: Relationships of Gradations and laboratory Air Voids for Superpave Mixes (2). Table 2.3: Summary of minimum t/NMAS to provide 7 % air voids in laboratory (2) Mix Minimum t/NMAS Minimum Thickness, mm 9.5 mm ARZ 3.9 37 9.5 mm BRZ 5.2 49 9.5 mm TRZ 5.4 51 19.0 mm ARZ 2.4 46 19.0 mm BRZ 3.0 57 19.0 mm TRZ 2.8 53 37.5 mm ARZ 2.0 75 37.5 mm BRZ 2.4 90 37.5 mm TRZ 2.0 75 9.5 mm SMA 7.3 69 12.5 mm SMA 7.5 94 19.0 mm SMA 4.4 84 13 Figure 2.4 represents the general trend observed in NCHRP study for field compaction of a 19.0 mm coarse-graded mixture. The best fit lines indicate that as the thickness increased the air voids decreased until a point where excessive thickness resulted in an increase in air voids. This figure also pointed out the difference in the results due to types of rollers used which may indicate that minimum field t/NMAS criteria should include other factors besides density. 19.0 mm Coarse-Graded with Modified Asphalt Steel/Rubber Tire Roller R 2 = 0.81 Steel Roller R 2 = 0.65 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 110.0 120.0 130.0 Thickness, mm Air Voids, % Figure 2.4: Relationship of Air Voids and Thickness for 19.0 mm Coarse-Graded with Modified Asphalt (2). 14 2.2 Laboratory Compaction 2.2.1 Introduction Superpave mix design method accounts for traffic loading and environmental conditions. One of the biggest differentiating aspects of the Superpave method compared to other methods such as Hveem and Marshall is the use of the gyratory compactor to simulate field compaction. The Superpave gyratory compactor was developed to improve the ability to compact samples for mix design to simulate actual field particle orientation (6). A compaction pressure of 600 kPa (87 psi) is applied to the sample top. The sample is inclined at 1.25? and rotates at 30 revolutions per minute as the load is continuously applied (see Figure 2.5). This helps achieve a sample particle orientation that is somewhat like that achieved in the field after roller compaction. Initially, compactors from different manufactures provided different results (densities). Lack of control of the internal angle was the main reason for the difference between brands of compactors (10). 15 Ram pressure 600 kPa 150 mm diameter mold 30 gyrations per minute 1.25 degrees Figure 2.5: Schematic of the Superpave Gyratory Compactor. 16 2.2.2 Potential effect of the internal angle of gyration One influencing factor that has been identified to explain the differences in sample density produced by different models and units of gyratory compactors is the dynamic internal angle (DIA) of gyration. Prowell et al. (10) measured the DIA on 112 different SGCs in Alabama (seven different models) and it was found that on average a change in 0.1 degrees of internal angle will result in a change of 0.010 G mb units or a difference in air voids of approximately 0.4 percent. FHWA conducted a study to determine the target and tolerance for the DIA of 1.16 ? 0.03 degrees (11). The difference of DIA affects the number of gyrations necessary to reach a required level of density. In theory, the compacted sample density from a compactor has to be adjusted to that which would have been produced if it had been set to a DIA of 1.16 degrees. Another study conducted by Prowell (12) suggested that the locking point of the mixture (the first instance of two consecutive gyrations resulting in the same sample height) was approximately the same number of gyrations for two different gyratory compactors, without any adjustments. However, the density at a given definition of the locking point was higher for the Pine compactor, compared to the Troxler compactor, when the data were not corrected to a DIA of 1.16 degrees. 2.3 Laboratory mix parameters used to describe compactability 2.3.1 Introduction Several asphalt researchers have proposed laboratory measured parameters of mixtures and/or their components as indicators of HMA compactability and/or resistance to 17 permanent deformation. Some of the parameters used to describe lab compactability are the percentage of maximum theoretical specific gravity at N ini (%G mm @ N ini ), the Compaction Energy Index determined from the SGC compaction process as recommended by Bahia (4) and the Coarse and Fine Aggregate Ratios as determined using the Bailey Method recommended by Pine (5). Compaction slope and the number of gyrations to reach the Locking Point of the mixture are related to resistance to permanent deformation. 2.3.2 The percentage of maximum theoretical specific gravity at N ini (%G mm @N ini ) The Superpave mix procedure (13) suggests that the compactability of a mixture can be indicated by its relative density at N initial which is an early point in the gyratory compaction process. According to the Superpave mix procedure (13), mixes that compact too quickly (air voids at N initial are too low) may be tender during construction and unstable when subjected to traffic. This is an indication of aggregate quality. Mixes with excess natural sand will frequently fail the N initial requirement (6). 2.3.3 Compaction Energy Index (CEI) The Compaction Energy Index (CEI) was defined by Bahia (4) as the area beneath the compaction curve from percent of G mm at the 8 th gyration to 92% of G mm as shown in Figure 2.6. Bahia reasoned that this index is analogous to the work applied by the roller to compact the mixture to the required density during construction. It is reasoned that mixtures with lower values of CEI are easier to compact; while a very low value of CEI could be an indication of a tender mixture and should be avoided. Bahia also introduced the Traffic Densification Index (TDI) which is defined as the area beneath the compaction curve from 92% to 98% of G mm (Figure 2.6). This index represents the energy required by traffic to densify the mixture from 92% G mm to a terminal density of 98% of G mm . 98% of G mm is considered a critical density, at which the mixture is approaching the plastic failure zone. Mixtures with lower values of CEI and higher values of TDI will have better constructability and performance (4). 82% 84% 86% 88% 90% 92% 94% 96% 98% 100% 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 Number of Gyrations %G mm Traffic Densification Index (TDI) Compaction Energy Index (CEI) Figure 2.6: Illustration of CEI and TDI Indices (4) 2.3.4 Compaction slope determined from compaction in the Superpave Gyratory Compactor (SGC). Figure 2.7 shows the percentage of maximum theoretical density versus the log of gyrations and the equation used to calculate the compaction slope (14). Figure 2.8 is an 18 illustration of the effect of different mixture properties on compaction slope: a) Higher compaction slopes are associated with higher asphalt content for the same mixture. It can be seen that a mixture asphalt content of 4.7% resulted in a slope of 8.02 while the same mixture with an asphalt content of 6.2% ended with a slope of 8.3. b) Finer gradations tend to have lower compaction slopes (slope of 6.6 for the finest gradation and 9.93 for the coarsest). c) More rounded aggregates, or those with less internal friction (gravel2 with a slope of 6.14), also tend to produce lower compaction slopes than more angular aggregates (limestone2 with a slope of 8.84). d) Higher compaction slope mixtures tend to have higher shear stiffness and lower permanent shear strain. A good correlation was obtained between compaction slope and mixture stiffness (R 2 = 0.69) for a mixture placed on Mississippi US 61 in 1994 (15). 82% 84% 86% 88% 90% 92% 94% 96% 98% 0.1 1 10 100 1000 Number of Gyrations %G mm N ini N des Figure 2.7: Illustration of Compaction Slope (14). 19 a) Effect of asphalt binder content b) Effect of aggregate gradation c) Effect of aggregate type (angularity) d) Relationship between shear stiffness and compaction slope Figure 2.8: Effect of mixture properties on compaction slope (15). 2.3.5 The number of gyrations with the SGC to reach the Locking Point of the mixture. Initially, the locking point was defined as the first of three gyrations at the same height which were preceded by two gyrations of the same height (16). Vavrick and Carpenter (17) refined the definition of the locking point as the first gyration in the first occurrence of three gyrations of the same height proceeded by two sets of two gyrations with the same height. Since its development, other agencies have altered the definition of the locking point. Other values used include: first instance of two consecutive gyrations resulting in the same sample height (locking point 2-1), second instance of two 20 21 consecutive gyrations resulting in the same sample height (locking point 2-2), the third instance of two consecutive gyrations resulting in the same sample height (locking point 2-3). In this study, the first instance of two consecutive gyrations resulting in the same sample height (locking point 2-1) was used as the locking point of the mixture. The trend is similar to the compaction slope; higher locking point mixtures tend to have higher shear stiffness and lower permanent shear strain (14). 2.3.6 Other densification indices Since the CEI and TDI are derived from densification (volume change) only, they could be considered incomplete in representing the resistance of mixtures to distortion under traffic (4). Another method was developed to directly measure the shear resistance of mixtures. Delage (18) proposed analyzing data from SGC testing with the Gyratory Load-Cell Plate Assembly (GLPA) and the introduction of two more indices based on resistive effort curve, which is similar to the densification curve. The GLPA is placed on top of an HMA specimen during compaction in the SGC (19). In this configuration, the GLPA is able to record the resultant force on the sample and the radial eccentricity throughout the compaction. The resultant force and the eccentricity are used to estimate the resistive effort of compaction. To quantify the resistive efforts above and below 92% G mm , Figure 2.9 shows that the area under the resistive effort curve between N ini and 92% G mm is calculated and named the compaction force index (CFI), and the area between 92% and 98% G mm is calculated and named the traffic force index (TFI). Delage also suggested that the CEI relating to the compaction curve be renamed the construction densification index (CDI). This study also showed that CDI and CFI are well correlated (see Figure 2.10). Therefore, hypothetically both indices equally represent the effort applied by the rollers to compact the mix to the required density during construction. Figure 2.9: Energy Indices CEI and TDI; CFI and TFI (18). 22 Construction Indices Correlation Correlation of Traffic Indices (TEI/ TDI and TFI). Figure 2.10: Correlation between CEI and CFI; TDI and TFI (18). 2.4 Aggregate characteristics related to compactability 2.4.1 Introduction Changing the aggregate gradation of a mixture alters the particle size distribution which in turn influences the packing of the aggregate skeleton. Pine says that the Bailey Method of Gradation Analysis can be used as an indicator of HMA compactability (20). The Bailey Method involves the following approach: ? Evaluates packing of coarse and fine aggregates separately ? Defines a gradation as either fine-graded, coarse-graded, or an SMA ? Evaluates the ratio of percentages of different size particles (indicators of HMA compactability used in this study) ? Evaluates the individual aggregates and the combined blend by volume 23 24 2.4.2 Bailey Method ratios Four sieves are defined to quantify the shape of the gradation curve and the determination of the ratio of different size particles. ? The primary control sieve (PCS) is designated as the split between coarse aggregate and fine aggregate. ? The half sieve is designated as an intermediate sieve in the coarse aggregate ? The secondary control sieve (SCS) ? The tertiary control sieve (TCS) The primary control sieve is determined as the sieve size closest to the size defined by: PCS = NMPS x 0.22 [1] Where, PCS = primary control sieve for the overall blend NMPS = nominal maximum particle size for the overall blend The value 0.22 is the factor that gives the average size opening between the coarse particles, considering the different shapes of aggregates. The Half sieve is determined as follows to find the closest sized sieve: Half sieve = NMPS x 0.5 [2] The secondary and tertiary control sieves are defined as follows: SCS = 0.22 x PCS [3] TCS = 0.22 x SCS [4] 25 Three ratios define the shape of the gradation curve. One ratio defines the shape of the coarse aggregate portion of the gradation. The second ratio defines the shape of the coarse portion of the fine aggregate, and the third ratio defines the shape of the fine portion of the fine aggregate. The CA ratio is used to represent the packing characteristics of the coarse aggregate fraction of the combined blend. For coarse gradations this ratio is defined as follows: CA Ratio = (% passing half sieve - % PCS) / (100 - % half sieve) [5] Where, % half sieve = percent passing the half sieve % PCS = percent passing the primary control sieve This ratio describes how the coarse aggregate particles pack together and, consequently, how these particles compact the fine aggregate portion of the aggregate blend that fills the voids created by the coarse aggregate. The FA C ratio of the fine aggregate is used to estimate the packing characteristics of the coarse portion of the fine aggregate. For coarse gradations this ratio is defined as follows: FA C = % passing SCS / % PCS [6] This ratio describes how the coarse portion of the fine aggregate packs together and, consequently, how these particles compact the material that fills the voids it creates. 26 The FA F ratio of the fine aggregate is used to estimate the packing characteristics of the fine portion of the fine aggregate. For coarse gradations this ratio is defined as follows: FA F = % passing TCS / % FAIB [7] This ratio describes how the fine portion of the fine aggregate packs together. It also influences the voids that will remain in the overall fine aggregate portion of the blend because it represents the particles that fill the smallest voids created. For fine gradations, new factors are necessary to apply equations 5, 6 and 7 and those factors are given as follows: New NMPS =PCS [8] New Half sieve = 0.5 x New NMPS [9] New PCS = 0.22 x New NMPS [10] New SCS = 0.22 x New PCS [11] New TCS = 0.22 x New SCS [12] Pine describes how the three ratios affect mixture compactability of dense gradations: ? As the CA Ratio increases, the mixes are more difficult to compact in the field. ? In general, as the FA c ratio increases towards 0.55, compactability of the overall fine fraction increases. And as the ratio decreases, compactability of the mixture increases. 27 ? In general, as the FA f Ratio increases towards 0.55, compactability of the overall fine fraction increases. And as the ratio decreases, compactability of the mixture increases. ? For SMA mixtures a value of 0.65 may be used instead of 0.55. Tables 2.4 and 2.5 show the recommended ranges of aggregate ratios for conventional and SMA mixtures, respectively. It can be seen that the CA ratio is the only parameter that is clearly affected by the aggregate size. The CA ratio decreases as the nominal particle size decreases for coarse-graded and SMA mixes and varies from 0.30 to 0.95 for coarse-graded and 0.15 to 0.50 for SMA. The remaining parameters are mostly contained in a reduced range. Table 2.6 shows the controls sieves used to calculate aggregate ratios (5). Table 2.4: Recommended ranges of aggregate Ratios NMPS, mm 37.5 25.0 19.0 12.5 9.5 4.75 CA Ratio Coarse-graded mixes 0.80-0.95 0.70-0.85 0.60-0.75 0.50-0.65 0.40-0.55 0.30-0.45 CA Ratio Fine-graded mixes 0.60-1.0 FA C Ratio 0.35-0.50 FA f Ratio 0.35-0.50 28 Table 2.5: Recommended ranges of aggregate Ratios for SMA mixes NMPS, mm 19.0 12.5 9.5 CA Ratio 0.35-0.50 0.25-0.40 0.15-0.30 FA C Ratio 0.60-0.85 FA f Ratio 0.65-0.90 0.60-0.85 0.60-0.85 Table 2.6: Control sieves Nominal size, mm Primary Control Sieve Half Sieve Initial Break Secondary Break 4.75 1.18 2.36 0.30 0.075 9.5 2.36 4.75 0.6 0.15 12.5 2.36 4.75 0.6 0.15 19.0 4.75 9.5 1.18 0.30 2.4.3 Primary Control Sieve Index PCSI The Primary Control Sieve Index PCSI is defined as the difference of percentage passing from the mixture?s gradation to primary control sieve. It represents the relative coarseness or fineness of the gradation. Figure 2.11 shows an example of the PCSI for 12.5 mm NMAS fine and coarse gradations. Coarse gradations will have negative values and fine gradations will have positive values of PCSI. 2. 36 4. 75 9. 50 12. 50 19. 00 25. 00 0. 0 7 5 0. 15 0. 30 0. 60 1. 18 0 10 20 30 40 50 60 70 80 90 100 Sieve Size (mm) P e r c e n t P assi ng Blend 1 Blend 2 PCSI Fine-Graded = Percent passing PCS - Percent passing PCS at MDL (+) PCSI Coarse-Graded = Percent passing PCS - Percent passing PCS at MDL (-) Figure 2.11: Illustration of the PCSI for 12.5 mm NMAS gradations. 2.5 Studies that relate laboratory characteristics and field compactability In the literature, several methods for measuring compactability can be found and some of them relate laboratory and field compaction. Three methods are discussed here: 2.5.1 C-value method The C-value method is the most frequently used method to measure compactability. It describes the progress in compaction by using an exponential formula. Kezdi and other researchers (21) developed this principle. The model is based on the assumption that the progress of compaction, expressed in terms of increase in density (?), 29 due to an added amount of compaction energy (S), depends on the difference between the current density state and the maximum density (? ? ). Where ? 0 is the density at the start of the compaction process. The concept of Kezdi can be formulated as: ]/[)()( 3 0 mkgeS c s AAAA ???= ?? ???? [13] The C value in the formula is a measure of the rate at which the density approaches the asymptotic value ? ? and thus also for the rate of the compaction progression (see Figure 2.12). Analyzing the equation above, one can see that materials with lower C-values are easier to compact compared to materials with higher C-values. For asphalt concrete mixtures commonly used in Germany during the 1980?s typical values were between 10 and 30, and in the Netherlands C-values ranged from 18 to 38 (21). Figure 2.12: The compaction process according to the C-value method. 30 2.5.2 The modified Mohr method Based on the similarity between HMA and granular materials, Nijboer (21) characterized HMA mixes by means of the parameters ? and ? which follows the Mohr theory to characterized granular materials. In addition, he introduced the parameter ? to capture the viscous component of HMA. Nijboer assumed that the resistance against deformation of ?bituminous mineral aggregate mixtures? could be represented by three physical quantities: ? the angle of internal friction, ?, ? the initial resistance, ?, ? the viscosity of the mass, ?, which denotes the influence of the viscosity on the shear resistance of the bitumen aggregate mixture. Nijboer studied the plastic behavior of bituminous aggregate mixtures for developing his ?rolling theory?. He made an inventory of all parameters of which he thought that they were relevant for the progress of bituminous mixtures compaction. As a result of Nijboer?s investigation is the R f factor which is a parameter that should indicate how far the compaction process of HMA is progressed. 4.0 ? ? ? ? ? ? ?? ? = v h n C ID P R m cb f ? ? [14] Where: P = the weight of the roller drum, N l = the width of the roller drum, mm D = the diameter of the roller drum, mm 31 32 C = factor for the roller type (i.e. 2.5 for static steel rolling) ? cb = the initial resistance of the bituminous mixture, N/mm 2 ? m = the viscosity of the mass of the compacted mixture, poise n = the number of roller passes applied h = thickness of the layer, mm v = speed of the compactor, mm/sec The method is in principle not a ?pure? method to measure compactibility but rather a method to describe the plastic deformation of HMA?s. During material measurements in the tri-axial apparatus, the governing stress conditions cause the material to shear. During compaction, excessive shear stresses must be avoided because when granular materials shear, they dilate, producing lower densities and even cracks can develop. This inconsistency makes the method, in the author?s opinion, not fully suitable for modeling bituminous aggregate mixtures compaction behavior. 2.5.3 The k-factor method This study was conducted in France and this included bituminous concrete layer thicknesses between 30 and 120 mm. The laboratory samples were compacted using a gyratory compactor developed in France called gyratory shear compacting press (PCG). The field samples were compacted using a type of pneumatic compactor in the laboratory. The number of gyratory revolutions corresponding to the number of passes by the pneumatic compactor was found to be reasonably accurate by the formula (22): 33 Ng = k?e?Np [15] Where: Ng = number of PCG revolutions e = layer thickness (mm) Np = number of compactor passes. k = 0.0625 The k-factor depends essentially on the nature of the compactor and increases with compactor efficiency. For a vibratory roller with a linear static load of 3.5 N/mm and a frequency of 25 to 30 Hz the k-factor value was determined to be about 0.25. When the thickness of a layer is known, this expression makes it possible to calculate the number of PCG revolutions for which the void content obtained in the laboratory will be equal to the void content obtained on the job site for a given number of passes of the compactor. Thus, if the intended job site thickness is 100 mm and the number of passes of the roller in the field is 16, the reference number of revolutions is 100. Therefore, it is possible to predict whether the void content in situ will be acceptable and to adjust the mix composition if necessary. 2.6 Summary of findings Literature suggests that field compactability of asphalt mixtures has become a major concern since the adoption of Superpave designed mixes. Many factors have been identified affecting the compaction on the field including material properties, environmental conditions, gradation, lift thickness and roller operations. In general, 34 rough surface texture, cubical or block shaped aggregate and highly angular particles all decrease mix compactability. Mixes with mixes with high asphalt content will be easier to compact. Higher initial mat temperatures require more time to cool down, which means more time available for compaction. The desired density is difficult to obtain on thin lifts (layers less than 50 mm) because of the mix?s rapid decline in temperature (8). With respect to laboratory compaction, literature suggests that the factors mentioned above are under control during the compaction process. Laboratory compaction using the gyratory compactor is characterized by samples with one and only height (thickness), confined in mold, compacted to a relatively high density, not sensitive to temperature and rapid compaction process (<3 min.). The dynamic internal angle (DIA) of gyration is probably the only influencing factor that has been identified to produce differences in sample density (10). Several asphalt researchers (4, 5) have proposed laboratory measured parameters of mixtures and/or their components as indicators of HMA compactability. Bahia (4), for example, proposed the Compaction Energy Index (CEI) which simulates the field compaction process to obtain an air voids content of 8%. Meanwhile, Pine (5) developed a set of parameters, called Bailey Method ratios, based on particle packing in a determined gradation. According to Pine the compactability of a mixture in the field can be predicted by observing how the aggregate particles are packed together. In general, the mixes are more difficult to compact in the field when the Bailey Method ratios increase. Literature also suggests that other researchers (14, 16) have developed laboratory measured parameters of mixtures and/or their components as indicators of resistance to 35 permanent deformation. Because of this resistance to permanent deformation the compaction slope determined by Anderson (14) and the number of gyrations with the SGC to reach the Locking Point of the mixture developed by Pine (16) have a potential used to predict field compactability of a mixture. Literature shows that some studies have been conducted to relate laboratory characteristics and field compactability (21, 22); the modified Mohr method by Nijboer, C-value method by Kezdi and the K-factor method. The applicability of these methods is yet questionable. In summary, literature suggests that there is a significant need to evaluate the factors affecting mix compactability in the laboratory and in the field. Most of the parameters mentioned above that have been used to measure mix compactability have not been validated with actual field performance and have been developed considering only laboratory compaction. 36 CHAPTER 3. RESEARCH PLAN 3.1 Overview The first part of this study included a comparison between laboratory measured characteristics of HMA (%G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios), an evaluation of the effect of physical properties on the SGC parameters and an assessment of variability among observations. The data used for this stage came from the Superpave mixtures placed on the NCAT Test Track in the first two cycles (2000 and 2003). This data set is well suited for this analysis because of the wide variety of mixtures included in the test track and the uniformity in construction operations at the track. This analysis included thirty-five different surface mixtures and seven binder mixtures placed on the track in 2000, and seventeen surface mixtures and twenty-two binder mixtures placed in 2003. The data used to determine the laboratory measured mix characteristics were obtained from quality control samples taken during track construction. Triplicate gyratory samples were compacted for each section. The second part of this study included the determination of a field compactability indicator based on the rolling operation: the Accumulated Compaction Pressure (ACP) that is defined in section 3.3. Compaction operations at the track were well documented and provide good information about the compactability of the mixtures in the field. These data were used to determine the total compaction energy applied by the rollers 37 during construction. Regressions between the laboratory compaction parameters and the field compaction energy were analyzed. The laboratory compaction parameters which yielded the best correlations were analyzed further by performing multiple regression analysis with basic mixture properties. The third part of this project included laboratory compaction of specimens using the SGC at the field lift thickness. The specimens were compacted to determine the number of gyrations to reach 92 percent of G mm (N@92%G mm ). The N@92%G mm for the field lift thickness specimens were compared to the N@92%G mm of normal height (115 ? 5 mm) specimens. Initially 25 mixtures were included but only 23 met the target air void content. For the remaining two mixtures, which have thicknesses below 30 mm, it was impossible to obtained an air voids content of 8% and the target thickness at the same time. The fourth part of this study involved eleven mixtures placed on the Test Track in 2006 and twelve mixes placed in 2003. These mixes were used to evaluate the field compactability indicator by conducting nuclear density tests and surface temperature measurements after each roller pass. Surface temperatures were obtained with an infrared temperature gun. The purpose of this part was to obtain the field compaction energy at 92 percent of G mm and correlate that energy with the laboratory compaction parameters. The last part of this project involved validation of one of the final compaction models using information from the NCHRP 9-27 study. The data set includes a variety of compaction equipment and mixtures. Table 3.1 shows a summary of the experimental plan. 38 Table 3.1: Summary of the experimental plan Section Number of observations (mixtures) Description Analysis of laboratory compaction parameters 81 Variety of mixes including fine, coarse, intermediate- graded and SMA. Nominal maximum size aggregate of 9.5, 12.5 and 19.0 mm. Parameters analyzed: %G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios. Correlation of ACP and laboratory parameters 71 Mix and materials properties used to correlate with ACP: air voids, VMA, VFA, microdeval, FAA, CAA, F&E 3:1, %pass 0.075mm, lift thickness, mix temperature and density level. Laboratory parameters analyzed: %G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios. Correlation of ACP and laboratory parameters for SGC specimens compacted at lift thickness 25 Variety of mixes including fine, coarse, intermediate- graded and SMA. Nominal maximum size aggregate of 9.5, 12.5 and 19.0 mm. A variety of lift thicknesses from 35 to 65 mm. Correlation of ACP obtained at 92% of G mm and laboratory parameters 23 Density testing was conducted on 11 mixes placed in 2006 and formation of 12 mixes was used to complete the analysis. Lift thickness and mix temperature were also used as predictor variables. Analysis of NCHRP mixes for validation purposes 16 A variety of mixes including fine, coarse, intermediate-graded and SMA. Nominal maximum size aggregate of 9.5, 12.5 and 19.0 mm. Variety of lift thicknesses from 30 to 100 mm. Variety of equipment, roller operation and environmental conditions. 39 3.2 Analysis of Test Track mixes 3.2.1 Material and Mixture Properties The HMA mixtures used in the first part of this study are shown in Table 3.2, which includes surface and some binder layer mixtures constructed on the NCAT Test Track in 2000 and 2003. This data set is well suited for this analysis because of the wide variety of mixtures included in the test track and the uniformity in construction operations at the track. 3.2.2 Test Track construction Several aggregate types were used on the track including limestone, granite, Florida limestone, gravel, and slag. Reclaimed asphalt pavement (RAP) was also used in a few sections. Each section on the test track had the same structural pavement foundation and used the same rollers and operator within each cycle. Weather condition was another similitude for each section within each cycle. In order to obtain the target density, each section was rolled with different number of roller passes with a combination of rollers. Each roller type applies a specific compactive effort ? vibratory force, static force and pneumatic tire kneading action. 40 Table 3.2: Summary of mix types evaluated Quad Sec Cycle* Sublot ** Aggregate Type Prod. NMAS Binder Grade Mod. Type Grad. Type N des . E 1 1 S quartzite 9.5 67-22 NEAT Fine 100 E 2 1 S granite 12.5 67-22 NEAT Coarse 100 E 3 1 S granite 12.5 76-22 SBR Coarse 100 E 4 1+2 S granite 12.5 76-22 SBS Coarse 100 E 5 1+2 S granite 12.5 76-22 SBS Intermediate 100 E 6 1+2 S granite 12.5 67-22 NEAT Intermediate 100 E 7 1+2 S granite 12.5 76-22 SBR Intermediate 100 E 8 1+2 S granite 12.5 67-22 NEAT Fine 100 E 9 1+2 S granite 12.5 76-22 SBS Fine 100 E 10 1 S granite 12.5 76-22 SBR Fine 100 N 1 1 S lms/slag 9.5 76-22 SBS Fine 100 N 2 1 S lms/slag 9.5 76-22 SBS Fine 100 N 3 1 S lms/slag 9.5 67-22 NEAT Fine 100 N 4 1 S lms/slag 9.5 67-22 NEAT Fine 100 N 5 1 S lms/slag 12.5 67-22 NEAT Coarse 100 N 6 1 S lms/slag 12.5 67-22 NEAT Coarse 100 N 7 1 S lms/slag 12.5 76-22 SBR Coarse 100 N 8 1 S lms/slag 12.5 76-22 SBR Coarse 100 N 9 1 S lms/slag 12.5 76-22 SBS Coarse 100 N 10 1 S lms/slag 12.5 76-22 SBS Coarse 100 N 11 1+2 S granite 12.5 76-22 SBS Intermediate 100 S 1 1 S granite 12.5 76-22 SBS Coarse 100 S 2 1+2 S gravel 9.5 76-22 SBS Coarse 100 S 3 1+2 S gvl/lms 9.5 76-22 SBS Coarse 100 S 4 1 S limestone 12.5 76-22 SBS Fine 125 S 5 1 S gravel 12.5 76-22 SBS Intermediate 125 S 6 1+2 S lms/RAP 12.5 67-22 NEAT Fine 100 S 7 1+2 S lms/RAP 12.5 67-22 NEAT Coarse 100 S 8 1+2 S mar. schist 9.5 76-22 SBS Coarse 100 S 9 1+2 S granite 12.5 67-22 NEAT Coarse 100 S 10 1+2 S granite 12.5 67-22 NEAT Fine 100 S 11 1+2 S mar. schist 9.5 76-22 SBS Coarse 100 W 6 1 S lms/slag 12.5 67-22 NEAT Intermediate 100 W 9 1 S gravel 12.5 67-22 NEAT Coarse 100 W 10 1+2 S gravel 12.5 76-22 SBR Coarse 100 E 1 2 S lms 12.5 76-22 SBS SMA 50 E 2 2 S marine lms 9.5 67-22 NEAT Fine 100 E 3 2 S marine lms 9.5 76-22 SBS Fine 100 N 1 2 S grn/lms/snd 9.5 76-22 SBS Fine 80 N 3 2 S grn/lms/snd 9.5 70-22 NEAT Fine 80 N 4 2 S grn/lms/snd 9.5 76-22 SBS Fine 80 * Cycle 1 = 2000 Test track experiment, cycle 1+2 = 2000 and 2003, cycle 2 = 2003 only. ** S = surface mixture, B = binder or bottom mixture. 41 Table 3.2 (continued): Summary of mix types evaluated Quad Sec Cycle* Sublot ** Aggregate Type Prod. NMAS Binder Grade Mod. Type Grad. Type N des . N 5 2 S grn/lms/snd 9.5 76-22 SBS Fine 80 N 6 2 S grn/lms/snd 9.5 70-22 NEAT Fine 80 N 9 2 S lms 12.5 70-22 SBS SMA 75 N 10 2 S lms/chert 12.5 70-22 SBS SMA 75 N 13 2 S granite 12.5 76-22 SBS SMA 50 S 1 2 S granite 12.5 76-22 SBS SMA 50 S 5 2 S gvl/lms/snd 12.5 76-22 SBS Intermediate 75 W 2 2 S porph/lms 19.0 70-22 SBS SMA 75 W 3 2 S lms 9.5 67-22 NEAT Fine 50 W 6 2 S lms/gvl/snd 4.75 76-22 SBS Fine 50 W 9 2 S granite 9.5 67-22 NEAT Fine 100 S 1 1 B Granite 19.0 76-22 SBS Coarse 100 S 2 1+2 B Gravel 19.0 76-22 SBS Coarse 100 S 3 1+2 B Limestone 19.0 76-22 SBS Coarse 100 S 4 1 B Lms/RAP 19.0 76-22 SBS Fine 125 S 5 1 B Lms/Grv/RAP 19.0 76-22 SBS Coarse 125 S 11 1+2 B Marble-Schist 19.0 67-22 NEAT Coarse 100 N 11 1+2 B Granite 19.0 67-22 NEAT Coarse 100 N 2 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 3 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 3 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 3 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 3 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 4 2 B grn/lms/snd 19.0 70-22 SBS Fine 80 N 4 2 B grn/lms/snd 19.0 70-22 SBS Fine 80 N 5 2 B grn/lms/snd 19.0 70-22 SBS Fine 80 N 5 2 B grn/lms/snd 19.0 70-22 SBS Fine 80 N 5 2 B grn/lms/snd 19.0 70-22 SBS Fine 80 N 6 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 6 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 6 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 7 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 7 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 7 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 8 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 8 2 B grn/lms/snd 19.0 67-22 NEAT Fine 80 N 9 2 B lms 12.5 70-22 SBS SMA 75 N 10 2 B lms 12.5 70-22 SBS SMA 75 N 13 2 B Granite 19.0 67-22 NEAT Intermediate 100 S 1 2 B Granite 19.0 67-22 NEAT Intermediate 100 * Cycle 1 = 2000 Test track experiment, cycle 1+2 = 2000 and 2003, cycle 2 = 2003 only. ** S = surface mixture, B = binder or bottom mixture. 42 The breakdown roller utilized in the 2000 experiment was a 10-ton steel double drum roller HYPAC C778B with 78 inch drum width, which could operate in vibratory or static mode. The rubber tire roller was a HYPAC C560B with a tire pressure of 83 psi. For most segments, the initial rolling was performed using vibratory mode at low amplitude and high frequency (3400 vpm). This was followed by a variable number of passes of static mode. In a few cases the rubber tire roller was incorporated at the end of the compaction process. In the 2003 experiment, a 13-ton double drum roller HYPAC C784 with 84 inch drum width, which could operate in vibratory or static mode, was used as the breakdown roller using vibratory mode at low amplitude and high frequency (4000 vpm). This was followed by a variable number of passes of static mode using either, double drum roller HYPAC C784 or HYPAC C778B. In some cases a rubber tire roller (HYPAC C560B with a tire pressure of 83 psi) was used at the end of the compaction process. Figure 3.1 shows the distribution of thicknesses placed in the test track in both cycles for either binder or surface mixes. It can be observed that the mean thickness was around 50 mm and the majority of the layers (47 layers) were constructed with the same thickness (50 mm). Thickness, mm F r e q ue nc y 80706050403020 50 40 30 20 10 0 Mean 49.96 StDev 10.62 N10 Histogram of Thickness, mm Normal Figure 3.1: Frequency distribution of test track layer thicknesses. 3.3 Accumulated Compaction Pressure (ACP) as a field compactability indicator Accumulated Compaction Pressure (ACP) was defined to quantify the total applied compactive effort to the HMA mat. It is the summation of the force applied by each pass of each roller in the field compaction process. For any roller type and any number of passes made by each roller type, the ACP is calculated using Equation 16. 43 For steel wheel rollers in vibratory mode, the total applied load in pounds per linear inch (pli) is the sum of centrifugal force plus the unsprung drum weight of both vibratory drums, divided by drum width. The compaction pressure was calculated using pli divided by the small contact arc of the drum on the mat (Figure 3.2). The contact arc decreases with each drum pass as mix densifies. According to the roller manufacturer, the contact arc for vibratory mode after the third pass will be nearly constant. For steel wheel rollers in static mode, each pass will produce similar pressure and for pneumatic rollers the contact pressure is similar to tire pressure. ?? == = m r n p rp CPACP 11 [16] Where, r = roller type p = pass number CP = compaction pressure Equation 17 shows an example of the model if vibratory mode is used for initial breakdown rolling followed by static mode. In this model was assumed that each pass made with the roller in vibratory mode provides a different compactive energy to the mat while each pass made with the roller in static mode provides the same amount of energy to the mat. SS n p V CPnCPACP V ?+= ? =1 [17] Where, CP V = compaction pressure for vibratory mode CP S = compaction pressure for static mode n V = total number of passes in vibratory mode n S = total number of passes in static mode 44 Contact Arc Figure 3.2: Contact arc of the drum on the mat. Table 3.3 shows the data used to calculate the compaction pressure per roller pass for the test track in the first two cycles. The roller manufacturer provided the contact arc for vibratory mode. Table 3.3: Compaction Pressure per Pass (psi) C778 C784 Passes Contact arc (in) Low amp. High amp. Static Low amp. High amp. Static 1 6.92 100 112 109 120 2 4.90 141 159 NA* 153 169 NA* 3 3.46 200 225 217 240 >3 3.46 200 225 87 217 240 97 * If used as breakdown roller it would be necessary to calculate pressure for the first two passes 3.4 Laboratory compaction parameters Quality control data were used to calculate the different laboratory compaction parameters: %G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios. The results for eighty one mixtures were then compared and several statistical 45 46 analyses were conducted to identify correlations and trends among lab compaction parameters. Overall, it was expected that all these parameters would follow the same trend that coarser mixes are more difficult to compact than finer mixes (4). The effect of PCSI, as a measurement of the relative coarseness or fineness of the gradation, on the compactability of the mix with the gyratory compactor was a key factor to identify the expected trend. 3.5 Correlation of ACP and laboratory parameters Seventy one mixtures placed on the test track were used for this analysis. These mixes corresponded to surface and binder layers placed on top of HMA. The remaining ten mixes were excluded from this analysis because they were placed directly on top of unbound material. The ACP represents the total compactive effort during the compaction process. However, the resulting density from the compaction operations varied somewhat from section to section. Nevertheless, single and multiple linear regression analyses were conducted to evaluate possible relationships between ACP and laboratory compaction parameters. For the multiple regression analysis, the following parameters were used as predictor variables: asphalt content, actual PG grade, compaction slope, Compaction Energy Index (CEI), %G mm @ N ini , N@92%G mm , locking point (2-1) as defined in section 2.3.4, coarse and fine aggregate ratios, lift temperature, t/NMAS ratio, PCSI, fine aggregate angularity (FAA), the effective asphalt content (Vbe), and Micro Deval abrasion loss which is a measure of aggregate toughness and abrasion resistance. 47 Specimens were compacted to meet 8% air voids and obtain the number of gyrations at 92 percent of G mm (N@92%G mm -lift thickness). The original objective of this part of the study was to compare the N@92%G mm obtained from these specimens with those calculated in the first part and to evaluate the effect of using field thickness. The correlation between ACP and this new parameter (N@92%G mm -lift thickness) was also analyzed. Due to limited material availability, only twenty five mixtures were used in this stage and included a variety of mat thicknesses from 35 mm to 65 mm for coarse, fine and intermediate gradations. Finally, twenty three mixes (Table 3.4) were used to compare the ACP at 92% of G mm with different laboratory compaction parameters, mix properties and field conditions to try to improve any model obtained in previous stages. Density testing was performed on eleven surface mixtures placed on the 2006 Test Track to determine the compaction curve. Figure 3.3 is an illustration of one of the studied mixes that shows the relationship among density (pcf), lift temperature (?F), compaction time and roller pattern (V- vibratory, R-pneumatic, S-static). Figure 3.3 is another illustration that shows the relationship between corrected density (nuclear density expressed as %G mm once corrected by core density) and ACP for mixtures with different gradations. A Weibull model was used to fit these curves and obtain ACP at 92% of G mm . Figure 3.5 is an example of the Weibull model used to fit field compaction data. The remaining mixtures used in this analysis were taken from the 2003 experiment and only those mixes which had complete information were selected. 48 Table 3.4: Properties of NCAT Test Track mixtures 2nd and 3rd cycle Cycle Quad. Sec. NMAS AC% Temp 1 st pass, F Thickness (mm) ACP@92% G mm E 1 12.5 6.3 251 46 144 N 1 19 4.5 228 53 353 N 4 19 4.3 230 43 350 N 5 9.5 6.1 168 23 1200 N 6 9.5 6.2 162 28 750 S 1 12.5 5.1 226 43 314 N 9 12.5 6.6 249 46 328 N 10 12.5 6.2 245 51 353 W 3 9.5 6.2 221 33 840 W 9 9.5 5.8 158 25 750 E 3 9.5 7.9 168 56 353 2003 E 2 9.5 7.8 217 50 216 N 1 12.5 5.7 198 47 406 N 2 12.5 5.3 246 45 445 N 5 12.5 6.2 212 50 125 N 10 19 4.4 263 44 410 E 5 12.5 5.2 195 54 118 E 6 12.5 5.1 183 51 465 E 7 12.5 5.2 199 54 545 S 2 9.5 7.0 215 41 765 W 3 12.5 6.1 240 50 48 W 4 12.5 6.0 225 56 47 2006 W 5 12.5 5.1 205 52 99 Section N2 100 105 110 115 120 125 130 135 140 145 150 9: 3 0 9 :3 4 9 :3 6 9 :3 8 1 0: 0 1 1 0: 0 3 1 0 :0 5 10 : 0 7 1 0: 1 2 1 0: 1 5 10 : 2 4 10 : 3 0 1 0: 3 3 1 0 :3 6 10 : 3 7 Time D e n s it y , P C F 100 120 140 160 180 200 220 240 260 280 T e m p er at u r e, F Density Temperrature V V S RRRR S S S RR R S Figure 3.3: Relationship between nuclear density (pcf), lift temperature, compaction time and roller pattern. 70% 75% 80% 85% 90% 95% 100% 0 500 1000 1500 2000 ACP, psi D e n s i t y , %G mm N2, Fine-Graded W3, Fine-w/Rap S2, Intermediate-Graded N10, Coarse-Graded 92% of G mm Line Figure 3.4: Relationship between ACP and density (%G mm ). 49 S = 0.92104575 r = 0.98586999 ACP D e n s it y , % G m m 10.1 264.5 518.9 773.3 1027.7 1282.1 1536.5 7 3 .2 2 7 7 .0 7 8 0 . 9 3 8 4 .7 9 8 8 .6 4 9 2 .5 0 9 6 . 3 5 Weibull Model: d Xc ebaY * * ? ?= Coefficient Data: a = 93.840728 b = 3539.7666 c = 1.9191867 d = 0.21762349 Figure 3.5: Illustration of the model used to fit field compaction data. 3.6 Material and Mixture Properties for Validation Analysis The HMA mixtures used for this part of the study are shown in Table 3.5. This includes surface mixtures placed in different states as part of the NCHRP 9-27 study. These data were used to compare the results obtained from the analyses of the Test Track mixes. NCHRP 9-27 projects are characterized by a variety of equipment, roller operations and locations with a variety of environmental conditions. 50 51 Table 3.5: Data from NCHRP 9-27 mixtures Section Prod NMAS PG Grade Grad Type N des . Average field thickness, mm VA-1 9.5 70-22 Fine 65 38.1 VA-2 19.0 64-22 Coarse 65 63.5 VA-3 9.5 64-22 Coarse 65 38.1 NC-1 9.5 70-22 Fine 100 31.8 CO-1 12.5 58-28 Coarse 75 57.2 MO-1 19.0 64-22 Coarse 100 50.8 MO-2 19.0 64-22 Coarse 100 101.6 UT-2 19.0 64-34 Coarse 125 38.1 AL-1 25.0 76-22 SMA 50 60.9 AL-2 25.0 67-22 Fine 100 69.9 AL-4 19.0 76-22 Coarse 100 57.2 AL-5 12.5 67-22 Coarse 86 38.1 FL-2 12.5 64-22 Fine 75 37.5 GA-1 12.5 67-22 Coarse 75 38.1 GA-2 9.5 67-22 Fine 75 31.8 MS-1 12.5 67-22 Fine 80 38.1 52 CHAPTER 4. RESULTS AND ANALYSES 4.1 Conceptual hypothesis for explaining expected trends The following concepts were used to explain expected trends in terms of laboratory compactability using the Superpave gyratory compactor: ? Rough surface texture, cubical or block shaped aggregate and highly angular particles will all increase the required compactive effort to achieve a specific density (8). ? Strength of the aggregate particles directly affects the amount of degradation that occurs in the SGC. Softer aggregates typically degrade more than strong aggregates and allow denser aggregate packing to be achieved. ? Modified asphalt binders tend to have higher shear stiffness and lower permanent shear strain; in other words, they tend to increase resistance to permanent deformation and decrease compactability (8). ? Asphalt binder lubricates the aggregate during compaction and therefore, mixes with low asphalt content are generally difficult to compact because of inadequate lubrication, whereas mixes with high asphalt content will be easier to compact (6). ? According to Bahia (4), it is possible that during the initial stage of compaction (from N initial to N@92%G mm ), the initial density of coarse mixes is relatively low because of the larger voids that can be entrapped under the initial compaction. 53 This can explain the higher compaction energy indices expected for coarser mixes. ? The amount of compactive energy applied to a mix (number of gyrations) indirectly affects compactability. It is expected that mixes compacted at low N des (i.e. 50 gyrations) require higher asphalt content to reach 4% air voids than mixes compacted at high N des (i.e. 125 gyrations). Therefore, mixes with low N des (higher asphalt content) are expected to be easy to densify (6). 4.2 Laboratory compaction parameters The parameters used to describe laboratory compactability are the percentage of maximum theoretical specific gravity at N ini (%G mm @N ini ), the Compaction Energy Index and the Coarse and Fine Aggregate Ratios as determined by the Bailey Method. The first two parameters are obtained from a densification curve using the SGC, the Bailey Method ratios are computed from the gradation. Compaction slope and the number of gyrations to reach the Locking Point of the mixture are also obtained from a Superpave densification curve and because these parameters are related to resistance to permanent deformation they were also used to describe laboratory compactability. When analyzing the data, presenting the information using box plots was the best approach to study and compare the characteristics of a different batch of observations (24). A box plot allows identifying the center and how spread out the data are about this central value. A box plot also allows investigating extreme values (referred to as outliers) or study the distribution of the data values (the pattern of the data values along the 54 measurement axis). Box plots are useful for assessing symmetry, presence of outliers, general equality of location, and equality of variation and usually are a better way to visualize the results of comparisons using analysis of variance (ANOVA) or t-tests. A box plot is made up of a box with various lines and points added to it. The top and bottom of the box are the 25th and 75th percentiles. The length of the box is the interquartile range (IQR). Thus, the box represents the middle 50% of the data. Values outside the upper and lower adjacent values are called outside or extreme values. Values that are under three IQRs from the 25th and 75th percentiles are called mild outliers (shown as whiskers). Those outside three IQRs are called severe outliers (shown as asterisks). Mild outliers are not unusual, but severe outliers are. 4.2.1 Parameters obtained from Densification Curve Figure 4.1 shows a comparison of the percentage of maximum theoretical density at the number of initial gyrations (%G mm @N ini ) in terms of four types of gradations (fine, intermediate, coarse and SMA) using box plots. It can be seen that the coarser the gradation the lower the %G mm @N ini . It was found a strong evidence to conclude that the mean values of %G mm @N ini for the four groups are different (p-value < 0.0001; analysis of variance F-test) for a significance level alpha = 5%. This result suggests that coarser mixes are tougher to compact, at least in the gyratory compactor. When using a two- sample t-test to compare two groups, only the fine-graded and the intermediate-graded groups resulted not significant (p-value = 0.52) which means that there is no evidence that intermediate-graded mixes have lower values of %G mm @N ini than fine-graded mixes. Table A.3 of Appendix A contains the complete data set. It was found that higher values of %G mm @N ini for each category (fine, intermediate, coarse and SMA) are associated to mixes with lower asphalt contents as shown in Figure 4.2. It was determined that the reason why these results went contrary to the expected trend is the fact that gradations relatively finer tended to have lower initial air voids and therefore, higher values of %G mm @N ini . On the other hand, gradations relatively coarser tended to have lower values of %G mm @N ini because of the larger voids that can be entrapped under the initial compaction, as shown in Figure 4.3. Gr a d Ty pe %Gm m @ N i n i d-SMAc-Coarseb-Intermediatea-Fine 92 90 88 86 84 82 80 Figure 4.1: %G mm @N ini sorted by Gradation Type. 55 AC% %G m m @ N i n i 109876543 92 90 88 86 84 82 80 Grad Type c-Coarse d-SMA a-Fine b-Intermediate Figure 4.2: Relationship between %G mm @N ini and asphalt content. R 2 = 0.48 80 82 84 86 88 90 92 94 -30 -20 -10 0 10 20 PCSI %G mm@ N i n i Figure 4.3: Relationship between %G mm @N ini and PCSI. 56 The Compaction Energy Index is another indicator of mixture compactability. According to Bahia (4) low values of CEI represent mixtures easy to compact, and in this case, fine and intermediate mixtures require less energy to achieve the desired density, while SMA mixtures cover a wide range of required energy (see Figure 4.4). It was found that there is a strong evidence to conclude that the mean CEIs in the four groups are different (p-value < 0.0001; analysis of variance F-test) for a significance level alpha = 5%. Once again only the fine-graded and the intermediate-graded groups resulted not significant (p-value = 0.99, two sample t-test) which means that there is no evidence that intermediate-graded mixes have lower values of CEI. Gr a d Ty pe CE I d-SMAc-Coarseb-Intermediatea-Fine 300 250 200 150 100 50 0 Figure 4.4: Compaction Energy Index (CEI) sorted by Gradation Type. 57 The greater spread observed for SMA mixtures can be explained by a difference in compactive effort (N des ) and the relative coarseness expressed by the PCSI. It was found that SMA mixes compacted at N des = 50 gyrations presented CEI values below 70, while SMA mixes compacted at N des = 75 presented CEI values up to 280. Figure 4.5 shows a strong correlation between PCSI and %G mm @N ini and PCSI and CEI which indicates that coarser SMA mixtures also tended to be difficult to compact in the laboratory and also explains the greater spread observed for SMA in Figure 4.1. This suggests that SMA mixtures are very sensitive to the relative coarseness of the gradation and the compactive effort. R 2 = 0.62 80 81 82 83 84 85 86 87 88 -30 -25 -20 -15 -10 PCSI %G mm@ N i n i R 2 = 0.74 0 50 100 150 200 250 300 -30 -25 -20 -15 -10 PCSI CE I N des = 50 Figure 4.5: Relationship between PCSI and %G mm @N ini and PCSI and CEI for SMA mixtures. Figure 4.6 shows the traffic densification index measured from the number of gyrations to compact samples from 92 to 96% G mm (TDI 92-96 ). TDI has been described as an indicator of the potential of the mixture to consolidate under traffic. It can be noticed how this parameter tends to increase for mixtures designed with higher number of gyrations. This trend was expected because mixes compacted at higher N des are designed 58 to resist higher permanent deformations which also requires increased compaction effort to obtain a desired density. Nde s . T D I 92- 96 125100807550 200 150 100 50 Figure 4.6: Traffic Densification Index (TDI 92-96) sorted by Ndes. It can be seen in Figure 4.7 that the parameter compaction slope increases as the mixture becomes coarser. There is significant evidence that the mean slopes for the four groups are different (p-value < 0.0001; analysis of variance F-test) for a significance level of alpha = 5%. Since high slope values have been associated with mixtures resistant to permanent shear strain (15) therefore, the coarser the gradation the more resistance to deformation the mixture is. In this case it was found that there is no evidence that intermediate-graded and coarse-graded mixes have different mean values of slope (p- value = 0.065, two sample t-test) at significance level of 5%. 59 A group of possible outliers can be seen for fine-graded mixes. That group corresponds to mixes with high asphalt contents (above 7%) as part of the 2000 test track study to compare low versus high asphalt contents. Gr a d Ty pe Sl o p e d-SMAc-Coarseb-Intermediatea-Fine 13 12 11 10 9 8 7 6 5 Figure 4.7: Compaction Slope sorted by Gradation Type. The number of gyrations to achieve the locking point (LP 2-1) of the mixture indicates that fine aggregate mixtures tend to compact easily while SMA mixes present the opposite trend (Figure 4.8). It can be seen that the plots of slope and locking point have similar shapes. It was found that there is significant evidence that the mean values of LP 2-1 for the four groups are different (p-value < 0.0001; analysis of variance F-test) for a significance level of alpha = 5%. In this case it was found that there is no evidence that fine-graded and intermediate-graded mixes have different mean values of LP 2-1 (p- 60 value = 0.17, two sample t-test). The same conclusion was obtained for coarse-graded and SMA mixes (p-value = 0.18, two sample t-test) at significance level of 5%. Gr a d Ty pe Lo c k P t . 2 - 1 d-SMAc-Coarseb-Intermediatea-Fine 80 70 60 50 40 30 20 Figure 4.8: N @ Locking Point sorted by Gradation Type. 4.2.2 Gradation parameters as indicators of compactability According to the Bailey Method Coarse Aggregate Ratio, Fine Aggregate (coarse portion) Ratio and Fine Aggregate (fine portion) Ratio are indicators of compactability. As the ratios increase the overall compactability of mixture decreases. When evaluating the compactability using the CA ratio, the results indicate that an opposite trend from CEI and %G mm @N ini as gradations go from fine to coarse (Figure 61 62 4.9). This is indicating that the fine portion of the mix predominates over the coarse portion. On the other hand, FA c ratio has the same trend as CEI and %G mm @N ini for gradation type (Figure 4.10). It was found strong evidence that the mean values of CA ratio are different for the four groups (p-value < 0.0001; analysis of variance F-test) and even for pair comparisons using a t-test for a significance level of alpha = 5%. According to Pine (5), altering the FA c ratio normally involves a change in particle shape, strength and texture, as well as a change in gradation. It was found that this ratio has the most influence on altering VMA or air voids in a coarse-graded mix. Therefore, the assumption that packing characteristics of the coarse particles of the fine portion of the mix predominates over the packing of coarse portion may be valid at least for coarse-graded mixes. It was found that there is significant evidence to conclude that the four groups have different mean FA c ratios (p-value < 0.0001; analysis of variance F- test) for a significance level of alpha = 5%. In this case it was found that there is no evidence that intermediate-graded and coarse-graded mixes have different mean values of FA C ratio (p-value = 0.065, two sample t-test) at significance level of 5%. Only the CA Ratio and the FA C Ratio were included in this study. FA f Ratios could only be calculated for coarse gradations which produced too few observations with respect to the other two parameters. Table A.3 of Appendix A contains the complete data set of the computed parameters. Gr ad Ty pe CA R a t i o d-SMAc-Coarseb-Intermediatea-Fine 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Figure 4.9: CA Ratio sorted by Gradation Type. Gr ad Ty pe FA c R a t i o d-SMAc-Coarseb-Intermediatea-Fine 0.8 0.7 0.6 0.5 0.4 0.3 0.2 Figure 4.10: FAc Ratio sorted by Gradation Type. 63 64 4.2.3 Comparison between laboratory measured characteristics of HMA Correlations were performed in order to compare mixture parameters. Table 4.1 shows the single correlation (R-value) calculated for each pair of variables, the shaded cells indicate a better correlation (above 0.6), and a letter ?C? indicates that the expected trend was correct. Compaction Energy Index correlates well with %G mm @N ini (R = -0.92) and indicates how as the compaction energy increases the initial air voids content of the mixture increases. Mixtures with low CEI and high %G mm @N ini tend to compact easily in the SGC. Superpave compaction slope and locking point of the mixture are strongly correlated (R = 0.92), and both are correlated with %G mm @N ini (R = -0.90 and R = -0.91 respectively). These results indicate that mixes with higher values of slope and locking point are difficult to compact in the lab. It can be noticed how the PCSI has good correlations with the mix compaction parameters as well. The relationship between PCSI and %G mm @N ini (R = -0.63) indicates that fine gradations tend to have low air void content at N initial and are mixes that compact rapidly in the SGC. 65 66 Previously, the potential effect of the internal angle of gyration due to the use of different gyratory compactors was discussed. For the first group of mixes (Test Track cycle 1) a Troxler gyratory compactor was used for design and quality control; and for the second group (cycle 2) a Pine SGC was used. The internal angle of gyration of each device was not determined at the time the data was collected. This factor can be a source of significant error, especially for those parameters which are computed based on specific gravity of specimens. For example, Figure 4.11 shows the difference in comparisons of CEI and %G mm @N ini resulting from the use of two different gyratory compactors. On the other hand, Figure 4.12 shows the relationship between N@92%G mm and CEI, which indicates that the results are not affected much by compactor type. These results follow the findings made by Prowell (12), where the data suggest that parameters expressed in terms of number of gyrations seem to be less affected by the dynamic internal angle (DIA) of gyration than parameters expressed in terms of density. Troxler - R 2 = 0.9481 Pine - R 2 = 0.7497 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200.0 80.0 82.0 84.0 86.0 88.0 90.0 92.0 94.0 %Gmm@Nini CEI Cycle 1 - Troxler Cycle 2 - Pine Poly. (Cycle 1 - Troxler) Poly. (Cycle 2 - Pine) Figure 4.11: Relationship between %Gm m @N ini and CEI. Troxler - R 2 = 0.9728 Pine - R 2 = 0.9473 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 200.0 0 5 10 15 20 25 30 35 40 45 N@92%Gmm CEI Cycle 1 - Troxler Cycle 2 - Pine Poly. (Cycle 1 - Troxler) Poly. (Cycle 2 - Pine) Figure 4.12: Relationship between N@92%G mm and CEI 67 68 4.2.4 Effect of physical properties on the SGC parameters A multiple analysis of variance (MANOVA) was used to evaluate the effect of physical characteristics (primary aggregate type, NMAS, PG grade, gradation type, and gradation ratios) on SGC parameters (%G mm @N ini , N@92%G mm , CEI, slope, locking point). Because of the limited number of observations (less than 100) this analysis included principal-effects only, and all interactions were omitted. A MANOVA analysis has one or more factors each one with two or more levels and two or more dependent variables. The calculations are extensions of the general linear model approach used for ANOVA and significant terms at alpha = 0.05 are shown in Table 4.2. The asphalt content was included in the analysis as a factor. For the first response variable, %G mm @N ini , mixtures containing limestone as primary aggregate source tended to be difficult to compact (%G mm @N ini = 85.5). The main differences between limestone and another aggregate type are: tougher aggregate (low values of micro deval below 9%), more angular (F&E 3:1 = 0%, FAA > 45%) and higher content of fine material (% passing 0.075mm sieve above 7%). This confirms that all the properties mentioned above increase the required compactive effort to achieve a specific density. The same effect was produced by 9.5 mm NMAS mixtures with the lowest value of %G mm @N ini (86.4). In terms of gradation type, fine-graded mixes were the easiest to compact and SMA mixtures were the most difficult to compact. SMA mixtures are designed to improve resistance to permanent deformation which also requires compaction energy. 69 For the parameter N@92%G mm , it was found in terms of primary aggregate type that marble-schist mixtures tended to achieve 92% of G mm at the lowest number of gyrations followed by granite mixtures which matched the results obtained for the parameter %G mm @N ini . Marble-schist aggregate is characterized by having flat and elongated particles (F&E 3:1 = 10%) and higher values of micro deval (17%). These properties are related to aggregate degradation and allow denser aggregate packing to be achieved. On the other hand, granite mixtures contain some flat and elongated particles (F&E 3:1 = 7%) and intermediate values of micro deval (8%), but most of these mixtures are fine-graded which may explain their ability to compact easily. Stiffness of the binder was a significant factor with an interesting result. PG 76 mixtures presented the lowest number of gyrations to reach 92% of G mm . It was followed by PG 67 mixes and finally PG 70 mixes. It was expected that stiffer binders should produce stiffer mixes more difficult to compact. The same behavior was observed for air voids content. On average, PG 76 mixtures presented the lowest air voids content (3.26%) while the remaining mixtures were close to 4% air voids. In addition, it was found that PG 76 mixtures had higher asphalt contents and were designed with that intention which may be the main reason of this behavior. According to Bahia (4), low values of Compaction Energy Index (CEI) indicate that a mixture is easy to compact. In general, marble-schist and granite mixtures were easier to compact than limestone mixes. PG 76 mixtures and fine-graded mixtures tended to be easy to densify as well, according to CEI parameter. This follows the same trend observed for %G mm @N ini . 70 As mentioned before, higher compaction slope mixtures tend to have higher shear stiffness and lower permanent shear strain. It can be noticed how mixtures with lms/slag as a primary aggregate source had the highest slope (10.6), which means that those mixtures are stiffer and more difficult to compact. The combination lms/slag has more angular particles (CAA = 50%) that makes the mixture difficult to compact and more material passing the 0.075 mm sieve (above 8%) which makes the mixture stiffer. As expected, finer gradations presented lower compaction slopes (below 7) and coarser gradations presented higher slopes (SMA = 11.6). The highest values obtained for SMA mixtures are due to a combined effect of gradation (coarser gradation) and higher asphalt contents. In terms of number of gyrations to reach the locking point of the mixture, the difference between coarse-graded mixture and SMA mixtures is minor (54.9 and 55.4 respectively) which indicates that the locking point is affected less by asphalt content than the slope. Limestone/slag mixtures had higher locking point (60) which means that this mixes tend to have higher shear stiffness and lower permanent shear strain, while granite/limestone/sand mixtures, as predicted by other parameters such as CEI, %G mm @N ini and Superpave slope, had a propensity to compact easily. 71 Table 4.2: Principal effects - MANOVA analysis. Response variable Significant terms Levels Least-square means (count) Primary Aggregate Type Granite Lms Marble-Schist gravel grn/lms/snd lms/slag marine lms 87.7 (21) 85.5 (10) 88.1 (3) 87.4 (7) 87.6 (24) 86.3 (11) 87.6 (4) NMAS 9.5 mm 12.5 mm 19.0 mm 86.4 (18) 87.5 (34) 87.6 (28) %G mm @N ini Gradation Type Fine Intermediate Coarse SMA 89.4 (39) 88.0 (9) 86.3 (24) 85.0 (8) Primary Aggregate Type Granite Lms Marble-Schist gravel grn/lms/snd lms/slag marine lms 21.3 (21) 37.5 (10) 17.6 (3) 25.4 (7) 22.5 (24) 28.3 (11) 24.8 (4) PG grade PG 67 PG 70 PG 76 24.6 (34) 28.2 (12) 22.5 (34) N@92%G mm Gradation Type Fine Intermediate Coarse SMA 18.6 (39) 23.6 (9) 31.9 (24) 26.2 (8) Primary Aggregate Type Granite Lms Marble-Schist gravel grn/lms/snd lms/slag marine lms 44.2 (21) 123.2 (10) 38.7 (3) 66.9 (7) 56.1 (24) 81.5 (11) 55.6 (4) PG grade PG 67 PG 70 PG 76 64.9 (34) 88.6 (12) 46.0 (34) Compaction Energy Index (CEI) Gradation Type Fine Intermediate Coarse SMA 25.8 (39) 52.9 (9) 92.1 (24) 95.3 (8) 72 Table 4.2 (continued): Principal effects - MANOVA analysis. Response variable Significant terms Levels Least-square means (count) Primary Aggregate Type Granite Lms Marble-Schist gravel grn/lms/snd lms/slag marine lms 8.27 (21) 9.80 (10) 9.00 (3) 7.92 (7) 7.69 (24) 10.6 (11) 8.43 (4) NMAS 9.5 mm 12.5 mm 19.0 mm 9.51 (18) 8.11 (34) 8.82 (28) SGC Slope Gradation Type Fine Intermediate Coarse SMA 6.42 (39) 8.19 (9) 9.04 (24) 11.6 (8) Primary Aggregate Type Granite Lms Marble-Schist gravel grn/lms/snd lms/slag marine lms 45.8 (21) 57.6 (10) 50.1 (3) 45.9 (7) 42.4 (24) 60.2 (11) 47.0 (4) NMAS 9.5 mm 12.5 mm 19.0 mm 52.8 (18) 48.0 (34) 48.8 (28) PG grade PG 67 PG 70 PG 76 50.8 (34) 51.7 (12) 47.1 (34) Locking Point Gradation Type Fine Intermediate Coarse SMA 40.6 (39) 48.6 (9) 54.9 (24) 55.4 (8) 4.2.5 Assessment of variability among observations (multivariate statistical analysis) Since mix properties are related to each other, multivariate statistical techniques not only apply, but also they are indeed needed. Multivariate techniques allow consideration of any existent correlation among variables (both response and predictor), so that their 73 unique and common effects can be evaluated (25). The following multivariate techniques are considered: (A) Canonical correlation. In order to gather information on how well composition percentages explain variability on physical properties. Canonical correlation simultaneously correlates several independent variables and several dependent variables. (B) Factor analysis. In order to evaluate how different mix properties are related to each other, as well as to extract and interpret hidden effects (properties that influence other properties). Principal component analysis extract factors based on the total variance of the factors and it is used to find the fewest number of variables that explain the most variance. (C) Hierarchical cluster analysis. In order to evaluate how observations can be separated into groups with similar properties, as well as which variables allow for such separation. The purpose of cluster analysis is to reduce a large data set to meaningful subgroups of individuals or objects. The division is accomplished on the basis of similarity of the objects across a set of specific characteristics. A. Effect of gradation properties and asphalt content on the SGC parameters Canonical correlation analysis is the study of the linear relations between two sets of variables. It is the multivariate extension of correlation analysis. A canonical correlation analysis was developed in order to evaluate the effect of the gradation parameters (Primary Control Sieve Index, and CA and FA C ratios) and the asphalt content (AC%) on 74 the SGC measures parameters of the HMA (%G mm @N ini , N@92%G mm , CEI, N@96%G mm , Slope, and Locking Point). Table 4.3 shows the most significant canonical functions on canonical correlation. The meaning of the canonical pairs of variables (response variables ? independent variables) is evaluated in the following form: ? Canonical set 1; the independent variable correlates with the relative coarseness or fineness of the gradation (PCSI), the asphalt content, the packing characteristics of the coarse portion of the fine aggregate (FA C Ratio) and the coarse aggregate ratio. The response variable correlates with %G mm @N ini , N@92%G mm , CEI, Slope and Locking Point. Canonical set 1 explains 64.3 % of the total variability. This analysis indicates that coarse mixtures with low CA ratio require stronger fine aggregate structure (higher FA C ratio) and higher asphalt content to meet the target volumetric properties. Mixtures with these characteristics tend to be difficult to compact in the laboratory. Table 4.3: Canonical functions on canonical correlation (*). Higher loadings (**) Predictor variables Response variables Canonical pairs (***) Variable Loading Variable Loading Accumulated variance among observations explained by response variables, which is reproduced by predictor canonical functions. 1 PCSI CA Ratio FA C Ratio AC, % -0.89 -0.65 0.78 0.53 %Gmm@Nini N@92%Gmm CEI Slope Locking Point -0.93 0.76 0.86 0.96 0.87 64.3 % (*) Even tough there is strong evidence the data corresponds to a non-normal multivariate distribution, how significant the canonical correlation model is gives confidence on results and interpretation. Canonical correlation is highly significant, at 99 % confidence. (**) Loading refers to the correlation between a common effect (canonical function) and an actual variable. (***) Common effects on predictor variables correlated to common effects on response variables. Consider every set has a response canonical function, as well as a predictor canonical function. 75 B. Identification of common effects among parameters Factor analysis (FA) is an exploratory technique applied to a set of observed variables that seeks to find underlying factors (subsets of variables) from which the observed variables were generated. The factor analyst hopes to identify each factor as representing a specific theoretical factor. Therefore, many of the reports from factor analysis are designed to aid in the interpretation of the factors. An FA analysis was performed in order to identify common effects. In this case all variables used in canonical correlation were used, including asphalt content. Table 4.4 shows the results of the factorial experiment analysis. It was found that the considered variables do have some common effect; when considering both physical properties and composition percentages, 96% of variance explained by twelve variables can be explained by only two factorials. From these extracted factors, their common effects can be interpreted, allowing establishing hypotheses on hidden effects, as follows: ? Factor 1 has to do with parameters obtained from Superpave densification curve related to shear stiffness and resistance to deformation, and gradation properties which are used to measure compactability. In this case the PCSI, as a measurement of the relative coarseness or fineness of the gradation, affects the shear stiffness of a mixture and its ability to resist permanent deformation. A positive sign of the loading indicates an increase of a value. A positive value of PCSI indicates finer gradations. In general, finer gradations tend to have lower shear stiffness and lower resistance to deformation. 76 ? Factor 2 explains how some parameters used to measure compactability in the laboratory are highly correlated. Table 4.4: Factorial experiment analysis Factor Variables showing higher loading on factorials (*) Loading (**) Variance among variables Accumulated variance among variables 1 PCSI Slope Locking Point Ca Ratio FA c Ratio 0.88 -0.83 -0.68 -0.69 -0.88 55.8% 55.8% 2 %Gmm@Nini N@92%Gmm CEI -0.72 0.91 0.76 41.8% 96.6% (*) Loading refers to the correlation between a common effect (factorial in this case) and an actual variable. (**) Analysis considering V-rotation. Significant factors are shown. C. Effect of material properties on the studied parameters Hierarchical cluster analysis was performed in order to separate observations by similar groups. Cluster separation was performed taking into account physical properties that were not very well explained by canonical correlation analysis and interpretation allowed identifying some hidden effects (see Table 4.5). Hierarchical clustering pointed out that, in explaining variability among clusters, gradation has a high effect in mixture properties and parameters used to measure compactability. This statement is explained by: ? Cluster 1 includes, basically, un-modified fine-graded mixtures. On average, this group has low values of CEI, slope, locking point and N@92%G mm . This cluster is also characterized by coarse aggregate particles well packed (high CA ratios). 77 ? Cluster 2 is composed by coarse mixtures with intermediate values of CEI, slope, locking point and gradation ratios. In this case, the coarse portion of the fine aggregate controls the particles packing. ? Cluster 3 contains fine-graded and intermediate-graded mixtures. This cluster has the lowest values of CEI and N@92%G mm but intermediate values of slope and locking point. On average, Bailey Method ratios are the same as those for cluster 1. The use of modified asphalt and NMAS of 19.0 mm may be the principal factor of difference. ? Cluster 4 has higher values of CEI, slope, locking point than cluster 2, which indicates that coarse gradations close to the maximum density line and intermediate gradations produce mixes difficult to compact with the SGC. ? Cluster 5 has the highest values of CEI, slope, locking point and air voids, which indicates that this group presents the most difficult mix configuration to compact. Once more, the coarse portion of the fine aggregate controls the particles packing with the highest FAc ratios. Overall, finer gradations and gradations with PCSI values close to zero tend to increase mixture compactability in the laboratory. In terms of gradation components as indicators of HMA compactability, FA c ratio has a strong effect in clustering and CA ratio clearly defines SMA mixtures (average CA ratio = 0.2). CEI, N@92%Gmm, Slope, Locking Point, Primary Control Sieve Index and FA C ratio are the most important response variables, in order to account for cluster separation. The first four variables are related to mixture performance during densification and represent the applied energy to reach a level of compaction and mixture resistance to 78 deform. PCSI and FA C ratio describe gradation properties and how particles are packed together. Both groups of variables are very well related to aggregate geological properties, since every one depends on the aggregate shape and texture. Table 4.5: Cluster separation and properties 1 2 3 4 5 Cluster* Mean St. Dev. Mean St. Dev. Mean St. Dev. Mean St. Dev. Mean St. Dev. PCSI 8.3 4.9 -4.5 5.2 3.2 5.5 -0.9 3.8 -20.3 3.9 Lab Voids 4.4 0.7 3.5 0.4 2.8 0.5 3.6 0.7 4.5 1.4 %G mm @N ini 89.5 0.6 88.4 0.9 89.7 1.7 85.9 0.7 83.9 2.3 N@92 %G mm 18.5 4.3 22.5 4.7 13.7 6.0 33.3 4.1 34.6 15.2 CEI 16.9 11.6 35.6 20.1 12.7 17.2 98.9 25.5 146.5 108.4 TDI 92-96 146.3 34.0 141.0 24.7 93.6 19.6 118.2 16.6 53.7 24.2 TDI 92-Ndes 174.8 61.5 235.3 40.1 250.9 62.6 181.3 40.9 74.6 41.0 N@96 %G mm 84.5 12.1 81.7 12.5 51.6 9.0 87.0 10.8 62.0 20.1 Slope 5.7 0.5 7.4 0.8 7.1 2.0 9.6 1.0 11.6 0.8 LockPt. 34.9 3.8 45.1 3.9 41.0 11.3 58.9 4.7 60.1 10.5 CA Ratio 0.8 0.1 0.4 0.2 0.8 0.2 0.6 0.3 0.2 0.1 FAc Ratio 0.4 0.1 0.6 0.1 0.4 0.1 0.5 0.1 0.7 0.1 Properties AC% 5.4 1.1 4.6 0.4 5.2 1.2 6.2 0.8 6.5 1.4 Primary Agg. Type grn/lms/snd, granite granite grn/lms/snd lms/slag, gravel Lms, granite Prod NMAS 9.5, 12.5, 19.0 12.5, 19.0 19.0 9.5, 12.5 12.5 Asphalt PG Grade 67 67, 76 67, 76 67, 76 69, 81 Mod. Type Neat NEAT, SBS NEAT, SBS NEAT, SBS SBS Grad. Type Fine Coarse Fine, Intermediate Coarse, intermediate SMA Dif f e rences among clusters** Ndes. 80, 100 100 80, 100 100 50, 75 * Hierarchical cluster separation by Ward?s method. ** Those are the most significant differences. 4.3 Field compaction Due to the well controlled construction operations at the test track, there were not significant differences in as-constructed density for intermediate, fine and coarse-graded mixes (p-value > 0.05, t-test). The only difference can be seen in Figure 4.13 for SMA mixtures (p-value < 0.0.001, t-test), which had the highest density (Test Track cycles 1 and 2). It can be seen in Table 4.6 that the mean densities for fine-graded and intermediate-graded mixes were similar with almost all the sections compacted over 92% of G mm. There were not significant differences in as-constructed density for intermediate, fine, coarse-graded and SMA mixes when comparing them by year of construction (p- value > 0.05, t-test). Gr a d Ty pe I n pl a c e de n s i t y , % o f G m m d-SMAc-Coarseb-Intermediatea-Fine 98 97 96 95 94 93 92 91 Figure 4.13: Post-construction density level (%G mm ) sorted by Gradation Type. 79 80 For SMA mixtures placed in 2000, it was found that a few passes of the static roller (2 to 3) produced density levels close to 94% of G mm . This suggests that SMA mixtures were easy to compact in 2000. On the other hand, much more roller passes (5 to 7 static and 3 to 5 pneumatic) were required to achieve density levels near 96% in 2003. This was explained by an increase in the specified density level for SMA mixtures placed in 2003 combined with the reduction in lift thickness compared with mixes placed in 2000. Table 4.6: Post-construction density level (%G mm ) sorted by Gradation Type and Test Track cycle Gradation Type Test Track Cycle Mean density Minimum Maximum Observations 2000 93.7 92.7 95.1 14 Fine 2003 93.5 92.1 96.0 28 2000 92.9 91.5 94.9 5 Intermediate 2003 93.5 93.1 93.9 2 2000 93.7 91.8 94.8 20 Coarse 2003 NA NA NA 0 2000 93.8 92.0 95.0 6 SMA 2003 95.8 93.1 97.5 9 4.3.1 Conceptual hypothesis for explaining expected trends The following concepts were used to explain expected trends in terms of field compactability: ? Rough surface texture, cubical or block shaped aggregate and highly angular particles will all increase the required compactive effort to achieve a specific density (8). 81 ? Strength of the aggregate particles directly affects the amount of degradation that occurs in the field. Softer aggregates typically degrade more than strong aggregates and allow denser aggregate packing to be achieved (8). ? A continuously graded (dense-graded) aggregate is generally easy to compact (8). ? A mix designed with high dust content is generally more difficult to compact (6, 8). ? Modified asphalt binders tend to have higher shear stiffness and lower permanent shear strain; in other words, they tend to increase resistance to permanent deformation and decrease compactability (15). ? Asphalt binder lubricates the aggregate during compaction and therefore, mixes with low asphalt content are generally difficult to compact because of inadequate lubrication, whereas mixes with high asphalt content will be easier to compact (6). ? According to Bahia (4), it is expected that that mixes with higher CEI tend to be difficult to compact in the field. In addition, coarser mixes are also expected to require more energy applied by the rollers. ? According to Pine (5) mixes with higher CA ratios (coarse portion of the gradation highly packed) are more difficult to compact in the field. And as the FAc ratio decreases, compactability of the mixture increases. ? Higher initial mat temperatures require more time to cool down, which means more time available for compaction. On the other hand, if the initial mix temperature is too high, the mix may be tender and difficult to compact until the temperature decreases and the viscosity of the asphalt binder increases (6, 8). 82 ? The desired density is difficult to obtain on thin lifts (layers less than 50 mm) because of the mix?s rapid decline in temperature (8). ? Mixes with properties that improve resistance to fatigue and permanent deformation (i.e. higher SGC slopes) require increased compaction effort to obtain a desired density (8, 15). 4.3.2 Analysis of the Accumulated compaction Pressure Compaction operations at the track were well documented and provide good information about the compactability of the mixtures in the field. These data were used to determine the total compaction energy applied by the rollers during construction. The total accumulated compaction pressures applied on each mixture were analyzed using some factors that affect field compaction: gradation type (fine, coarse, intermediate, SMA), lift thickness and/or t/NMAS, mix temperature, aggregate size (NMAS) and asphalt grade. Different approaches were used for explaining the variability observed in ACP. The analyses included single comparison using box plots, analysis of pairs using t-test and analysis of variance. When comparing the total energy applied by the rollers (ACP) in terms of gradation type for the first two cycles, it can be observed that there is not a clear trend as shown in Figure 4.14. This result was confirmed by an ANOVA F-test that showed poor evidence that the ACP differs for gradation type (p-value > 0.05; F-test). For coarse- graded mixes ACP ranges from 300 to 1800 psi, for fine-graded from 400 to 2400 psi, intermediate-graded from 300 to 1400 psi and SMA mixes from 300 to 2000 psi. However, when the data were subdivided also by cycle, differences were observed for fine-graded and SMA mixes (Figure 4.15). Figure 4.15 shows that greater compaction energy was required for the sections constructed in cycle 2, whereas for cycle 1, coarse- graded mixes required the highest compaction energy followed by fine-graded and intermediate-graded, and SMA mixes required the least (see Appendix B). Observe in Figure 4.15 that comparisons were limited to fine-graded mixes and SMA, and only two values for intermediate-graded mixes found in 2003 were used in the analysis. Gr a d Ty pe AC P SMAc-Coarseb-Intermediatea-Fine 3000 2500 2000 1500 1000 500 0 Figure 4.14: ACP sorted by Gradation Type. 83 AC P Grad Type Experiment SMAc-Coarseb-Intermediatea-Fine 20032000200320002003200020032000 3000 2500 2000 1500 1000 500 0 Figure 4.15: ACP sorted by Gradation Type and cycle. 84 The main reason why 2003 mixes required more compactive effort is observed in Figure 4.16. In 2000 the mean layer thickness was 53 mm, whereas in 2003 the mean thickness was 44 mm. The effect of placing thinner layers (<50 mm) in 2003 decreased the lift temperature and compaction time requiring higher compactive efforts to achieve similar density levels. Figure 4.17 shows the comparison of temperature at different compaction stages for the two cycles (2000 and 2003). It can be seen that the mean laydown temperatures (T1) of the mix were similar for the two cycles. When comparing the temperature at the beginning of the compaction (T2) a significant drop in temperature can be observed for mixes placed in 2003. On average, the compaction process for 2000 mixes started at 250 ?F, while the compaction process for 2000 mixes started at 220 ?F allowing less time to achieve the desired density. Finally, it can be seen that the rolling operation ended with temperature (T3) below 175 ?F for 2003 mixes. According to some authors (8), below 175 ?F little or no gain can be achieved with the application of additional compactive effort. Cycle T h i c k n ess, m m 20032000 56 54 52 50 48 46 44 42 40 Interval Plot of Thickness, mm vs Cycle 95% CI for the Mean Cycle T/ N M A S 20032000 7 6 5 4 3 2 1 Boxplot of T/NMAS vs Cycle Figure 4.16: Thickness and T/NMAS ratio for each cycle. T e m p erat u r e, F Experiment Final Pass (T3)First Pass (T2)Mix Laydown (T1) 200320002003200020032000 275 250 225 200 175 150 95% CI for the Mean Figure 4.17: Temperature measured at different compaction stages. 85 86 An analysis of variance (ANOVA) was performed to determine which factors (NMAS, gradation type, t/NMAS, temperature of mix, PG grade and post construction density level) significantly affected the resulting ACP. Table 4.7 shows the levels for each factors used in this analysis and Table 4.8 provides the ANOVA of ACP. The results show that only t/NMAS and temperature have a significant effect on ACP at a level of significance alpha = 0.05. T/NMAS is the most impacting factor (F-statistic = 7.19). Figure 4.18 shows the effect of t/NMAS on the total effort applied to the mix; t/NMAS ratios below 3:1 required much more compaction energy. Figure 4.18 also indicates that mixes with temperatures at the first pass of the breakdown roller below 225 ?F required more total compaction energy (ACP). Notice how the ACP increases as the post-construction density level increases. This may suggest that more energy was applied to reach a desired density level and may explain some variability observed on the ACP. Table 4.7: Description of levels per factor used in analysis of variance Factor Gradation type NMAS, mm t/NMAS Post construction density level, %G mm Temperature at first roller pass ?F PG grade Level Fine Intermediate Coarse SMA 9.5 12.5 19.0 Low < 3:1 Medium 3:1 ? 4:1 High > 4:1 Low < 93 Medium 93 ? 94 High > 94 Low < 225 High > 225 67 70 76 Figure 4.19 shows the interaction plot of ACP for the factors t/NMAS and temperature. Notice that for high t/NMAS ratios (above 4:1) the temperature has minimum effect on the compaction energy. Low t/NMAS ratios (below 3:1) with lower temperature require a substantial increment in compaction energy. Table 4.8: ANOVA for ACP Source Reduced DF Sum of Squares Mean Squares F- Statistic P-value Significant at ? = 5% Gradation Type 3 112861.6 37620.55 0.18 0.906 No NMAS 2 76625.79 38312.89 0.19 0.829 No Thickness/NMAS (t/NMAS) 2 2086232 1043116 7.19 0.008 Yes Density level, %Gmm 2 496946.1 248473 1.22 0.302 No Temperature (T2) 1 1464502 1464502 5.12 0.009 Yes PG grade 2 681993.8 340996.9 1.67 0.196 No Error 58 Total 70 Me a n o f A CP , p s i S M A c - C o a r s e b - I n t e r m e d ia t e a - F i n e 1600 1400 1200 1000 1 9 . 0 1 2 . 5 9 . 5 > 2 2 5 < 2 2 5 > 4 : 1 3 : 1 - 4 : 1 < 3 : 1 1600 1400 1200 1000 > 9 4 9 3 - 9 4 < 9 3 7 6 7 0 6 7 Grad Type NMAS T1 t/NMAS Density PG Grade Temperature Figure 4.18: Main Effects Plot (fitted means) for ACP. 87 t/NMAS Me a n A C P, p s i c-highb-mediuma-low 2500 2000 1500 1000 500 T1 a-low b-high Temperature <225 >225 <3:1 3:1 - 4:1 >4:1 Figure 4.19: Interaction Plot (fitted means) for ACP. 4.3.3 Analysis of pairs One of the main objectives of the first cycle of the test track was the evaluation of performance for different types of mixtures (23). Several mini experiments were conducted and the evaluation included: fine graded vs. coarse graded mixes, effect of asphalt grade, asphalt content and polymer type and effect of aggregate type. A two- sample t-test for ACP was performed using mixtures from the 2000 experiment. The following subjects were used as comparison factors: ? Binder Grade with two levels: Low (PG 68/70) and High (PG 78/80) ? Asphalt content with levels Low (optimum) and High (optimum + 0.5%) 88 ? Gradation with Coarse and Fine 89 ? Polymer type SBR and SBS ? Aggregate type with granite and others. Table 4.9 shows the results of the 2-sample t-test. The results show that even though there was no significant difference in ACP at ? = 5% for any factor, the most significant difference (lowest P-value) was obtained for aggregate type followed by gradation. Figure 4.20 shows a better form to understand the trends found in this analysis. A p-value of 0.125 provides suggestive evidence that granite mixes required lower compaction effort than mixes with another aggregate source and fine-graded mixes also required lower compaction effort than coarse-graded mixes. Finally, it can be seen that an increase in asphalt content slightly decreases the required effort. Table 4.9: Comparison of ACP by various subjects Factor Level Mean Difference T-statistic P-value Low 824 Binder Grade High 796 27.7 0.22 0.831 Low 969 Asphalt content High 780 189.4 0.79 0.457 Coarse 835 Gradation Fine 620 215 1.38 0.196 SBR 932 Polymer type SBS 997 -65.5 -0.34 0.75 Granite 606 Aggregate type Other 931 -224.7 -1.65 0.125 0 200 400 600 800 1000 1200 Low High Low High Coarse Fine SBR SBS Granite Other Binder Grade Asphalt content Gradation Polymer type Aggregate type Subjects and Levels of comparison A v er age ACP Figure 4.20: Comparison of ACP by various subjects. A similar analysis was conducted to evaluate the effect of the same subjects on one of the laboratory compactability parameters. In this case, the Compaction Energy Index (CEI) was selected. Table 4.10 shows that gradation was significant at ? = 5% and similar trends as those observed for ACP were found for CEI in terms of aggregate type and asphalt content. Once again a p-value of 0.178 provides suggestive evidence that mixtures contained granite as aggregate source required lower energy to reach 92% of G mm in the SGC than other mixes. Notice that an increase in asphalt content resulted in a slight decrease in CEI (see also Figure 4.21). 90 Table 4.10: Comparison of CEI by various subjects Factor Level Mean Difference T-statistic P-value Low 65 Binder Grade High 52 12.1 0.53 0.607 Low 96.3 Asphalt content High 75.6 20.7 0.73 0.491 Coarse 81.7 Gradation Fine 28.5 53.2 3.29 0.006 SBR 59.4 Polymer type SBS 54.7 4.66 0.15 0.885 Granite 19.4 Aggregate type Other 68 -48.6 -1.75 0.178 0 20 40 60 80 100 120 Low High Low High Coarse Fine SBR SBS Granite Other Binder Grade Asphalt content Gradation Polymer type Aggregate type Subjects and Levels of comparison C o m p act i o n E n er gy I n dex ( C E I ) Figure 4.21: Comparison of CEI by various subjects. 91 92 4.4 Correlations between ACP and laboratory compaction parameters Regressions between the field compaction energy (ACP) and laboratory parameters (%G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios), mixture properties (air voids, VMA, VFA, microdeval, FAA, CAA, F&E 3:1, %pass 0.075mm), lift thickness, mix temperature and density level were analyzed. The laboratory measured parameters which yield the best correlations were analyzed further by performing multiple regression analysis with basic mixture properties. Based on the conceptual hypothesis explained before, it was expected that the total accumulated compaction pressure would increase as the following parameters increased: %G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and CA ratio. In terms of mixture properties, it was expected that the total accumulated compaction pressure would increase as VMA increases, VFA decreases, PG grade increases, asphalt content decreases, micro deval decreases, FAA and CAA increase, F&E 3:1 decreases and the % passing 0.075mm increase. In addition, for low mix temperatures (<225 ?F), thin layers (<50 mm) and low t/NMAS ratios (<3:1), an increase in ACP was expected. As shown in the preceding discussion, there is an important difference in the compactive effort applied to the mixes of the first two test track experiments. When all the combined data were used to correlate ACP and lab compactability parameters, the values of simple linear correlation (R-value) were always near zero (Table 4.11). The factors which have the greatest influence of ACP were grade of the asphalt, temperature, thickness and t/NMAS. Most of the layers placed on the two test track experiments had a thickness near 50 mm as indicated by the oval in Figure 4.22. The best fit line between ACP and thickness (quadratic model) shows a minimum point of ACP near 65 mm, or in terms of t/NMAS ratio, near 5.0, which may vary depending on gradation type and aggregate size (according to NCHRP 9-27 results). Table 4.11: Single correlation between ACP and some parameters PG Grade 0.37 NMAS 0.12 PCSI -0.06 CA 0.24 FA c -0.20 % pass PCS 0.14 % pass 2.36 0.12 % pass 0.075 -0.14 AC% -0.08 Thickness (mm) -0.31 Actual PG 0.09 t/NMAS -0.34 Density 0.08 %G mm @Ni 0.13 N@ 92%G mm -0.14 CEI -0.10 R 2 = 0.1617 0 500 1000 1500 2000 2500 3000 0 2040608010 Thickness, mm ACP R 2 = 0.2267 0 500 1000 1500 2000 2500 3000 0.0 2.0 4.0 6.0 8.0 t/NMAS AC P Figure 4.22: Effect of thickness on ACP. 93 94 When the data obtained from the 2000 cycle were used to correlate field and lab compactability (see Table 4.12), the majority of the results followed the expected trend and some of them correlated well. Parameters such as N@92%G mm and CEI showed good correlation with ACP (R = 0.68 and 0.67, respectively). As expected, an increment of those parameters produced an increase of the field compaction energy (ACP). Most of these mixes were placed at a thickness of about 50 mm. When the data were limited to mixes placed in one 50 mm lift presented the highest correlation (Figure 4.23, R 2 = 0.45). These results indicate that mixes compacted in field with similar thicknesses can be easily correlated to ACP. Table 4.12: Correlation and expected trend between ACP and compactability parameters including mix properties (Tangents only, cycle 1) Parameter Correlation Expected trend Density 0.17 correct % Pass PCS -0.35 correct PCSI 0.38 correct %G mm @N ini -0.63 correct N@92%G mm 0.68 correct CEI 0.67 correct N@96%G mm 0.38 correct Slope 0.46 correct LockPt 0.51 correct CA Ratio 0.06 inconclusive FA c Ratio -0.11 inconclusive VMA 0.19 inconclusive VFA -0.49 incorrect CAA -0.01 inconclusive F&E 3:1 -0.21 incorrect FAA -0.58 incorrect Micro Deval 0.63 incorrect AC % 0.21 incorrect Actual PG 0.22 correct Temperature -0.20 correct R 2 = 0.4523 0 200 400 600 800 1000 1200 1400 1600 1800 2000 10 15 20 25 30 35 40 45 N@92%Gmm ACP Figure 4.23: Relationship between ACP and N@92%G mm . Despite these results, the relationship between lab and field compactability is complicated by many other variables which are known to affect compaction. In other words, a simple linear correlation does not adequately describe the relationship. The use of multiple regression analysis was necessary to incorporate the wide variety of mix properties and factors affecting field compactability. For the multiple regression analysis, the following parameters were used as predictor variables: asphalt content, actual PG grade, compaction slope, Compaction Energy Index (CEI), %G mm @N ini , N@92%G mm , locking point, coarse and fine aggregate ratios, lift temperature, t/NMAS ratio, PCSI, fine aggregate angularity (FAA), VFA and Micro Deval. The best model was selected by choosing the model with the least number of predictor variables, the highest adjusted R 2 , Mallow?s Cp statistic less than the number 95 96 of predictor variables and minimum standard error of the regression. This procedure provides a model with almost all the variables significant at a pre-defined significance level (in this study 5%). Once the best predictors were determined, a multiple linear regression was performed to determine the coefficients for each predictor. The final model also needed additional residual treatments to account for normality and linearity and only those predictors which followed a logical trend were selected. The final model is shown as follows: SQRTACP = 57.2 - 0.0765 T3 + 1.55 NMAS + 11.5 CA + 2.19 %pass200 - 0.767 VFA + 0.0755 N@field [18] S = 5.55 R-Sq = 54.9% R-Sq(adj) = 47.6% Predictor Coef SE Coef T P Constant 57.25 17.64 3.24 0.002 T3 -0.07652 0.02908 -2.63 0.012 NMAS 1.5546 0.3105 5.01 0.000 CA 11.455 4.498 2.55 0.015 % pass 200 2.1943 0.6330 3.47 0.001 VFA -0.7668 0.2018 -3.80 0.001 N@field 0.07549 0.05925 2.17 0.051 Where, SQRTACP = square root of ACP NMAS = nominal maximum size aggregate size CA = CA ratio %pass200 = percent passing 0.075mm sieve T3 = temperature at the last roller pass N@field = number of gyrations to reach the post construction density level 97 For this final analysis the square root of ACP was the response, while NMAS, CA ratio, % passing No 200 sieve, VFA, temperature and the number of gyrations to reach the post construction density level (N@field) were the predictors. The negative sign for temperature shows that the total compactive effort applied to the mixture decreases as the lift temperature increases. Cooler surfaces remove heat from the mat at a faster rate, decreasing the time available for compaction and eventually increasing the compactive effort. It can be observed in this model that, as expected, mixes with higher voids filled with asphalt tend to be more compactable. Equation 18 follows the Bailey Method theory that mixes with higher CA ratios are difficult to compact in the field. The positive sign of %pass200 indicates that higher amounts of material passing the 0.075mm sieve are related to stiffer mixes which tend to be difficult to compact in the field. Finally, the number of gyrations to reach the post construction density level was proportional to the accumulated compaction pressure. 4.5 Compaction of specimens using the SGC at field thickness The third part of this project included compaction of specimens using the SGC at thicknesses equal to those in the field. The specimens were compacted to meet 92 percent of G mm . One of the objectives of this part of the study was to compare the number of gyrations to reach the post construction density level at lift thickness (N@92%G mm-field ) obtained from these specimens to normal size specimens (115 ? 5 mm) and evaluate the effect of thickness reduction. Initially 25 mixtures were included but only 23 met the required air voids content. Mixtures used in this analysis included a variety of mat thicknesses from 35 mm to 65 mm and included SMA, coarse, fine, and intermediate gradations (see Table C1 of Appendix C). A quadratic model was used to fit the relationship between ACP and the number of gyrations to reach 92% of G mm (Figure 4.24) resulting in a poorer correlation (R 2 = 0.21) than was achieved with the standard height specimens, as shown in Figure 4.23. However, when the number of gyrations to reach the as-constructed field density was used, the correlation improved (R 2 = 0.62). Thus, the actual density level achieved with the compaction process is an important factor to take into account. R 2 = 0.21 0 500 1000 1500 2000 2500 0 50 100 150 N@92%Gmm-field thickness ACP, p s i R 2 = 0.62 0 500 1000 1500 2000 2500 0 100 200 300 400 N@field thickness/density AC P , p s i Figure 4.24: Relationship between ACP and number of cycles to reach 92%Gmm and field density. The majority of the mixes were compacted with a thickness near 50 mm. The effect of reducing the sample high was evaluating by the compaction of samples with a thickness of 100 mm. Figures 4.25 and 4.26 show typical results of the compaction process for 50 mm and 100 mm samples using the Superpave Gyratory Compactor. In Figure 4.25, it can be observed that SMA and fine-graded mixtures required very few gyrations to reach the target of 8% air voids. Meanwhile, coarse and intermediate-graded 98 mixtures required as much as twice the number of gyrations to reach 92% of G mm . On the other hand, Figure 4.26 shows a smaller difference in the required number of gyrations to reach 8% air voids. 0 20 40 60 80 100 120 140 160 180 24681012 Air voids, % N u m b e r of G y r a t i ons E10-2000 N8-2000 W6-2000 N10-2003 Coarse-Graded Fine-Graded SMA Intermediate-Graded Figure 4.25: Compaction of lab specimens at 50 mm. 0 5 10 15 20 25 30 35 40 456789101 Air voids, % N u mb e r of G y r a t i on s S3-2000 S4-2000 S5-2000 Coarse-Graded Fine-Graded Intermediate-Graded Figure 4.26: Compaction of lab specimens at 100 mm. 99 As final product of this analysis, by using a quadratic best fit Figure 4.27 shows the effect of reducing the thickness of lab specimens on the number of gyrations to reach 92% G mm . For specimens compacted at almost the design thickness the number of cycles is similar (N@92%Gmm-field / N@92%Gmm-design ratio = 1). Additionally, as the Thickness-design / Thickness-field ratio increases, extra energy is necessary (gyrations) to reach the target density. The best fit line establishes a minimum thickness point near 80 mm where fewer gyrations are needed than the ones obtained from design specimens, but further investigation is required in order to fill the gaps and confirm the results. R 2 = 0.6212 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 01234 Thickness-design / Thickness-field N @ 9 2 %G mm-fi e l d / N @ 9 2 %G mm-d e s i g n R 2 = 0.5065 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0 20 40 60 80 100 120 Field thickness, mm N @ 9 2 % G mm- fi e l d / N @ 9 2 %G mm -d e s i g n Figure 4.27: Effect of reducing thickness on lab specimens The use of multiple regression analysis was used to incorporate the wide variety of mix properties and factors affecting field compactability. For this analysis, several parameters obtained from the specimens compacted at lift thickness using the SGC were used as predictor variables: number of gyrations to reach 92% of G mm at lift thickness 100 101 (N@92%G mm-field ), number of gyrations to reach the post construction density at lift thickness (N@field-density) and locking point at lift thickness. Mix properties such as asphalt content, actual PG grade, coarse and fine aggregate ratios, lift temperature, t/NMAS ratio, PCSI were included in the analysis as well. The best model was a combination of PCSI, FAc ratio, lift temperature and number of gyrations to reach the post construction density level at lift thickness. These four identified factors were then regressed versus the ACP and the following regression equation was obtained: ACP = 2560 - 19.4 PCSI - 1018 FAc - 6.18 Temp + 4.02 N@density [19] S = 181.858 R-Sq = 82.2% R-Sq(adj) = 78.2% Predictor Coef SE Coef T P Constant 2560.1 574.5 4.46 0.000 PCSI -19.416 6.931 -2.80 0.012 FAc -1018.5 428.8 -2.37 0.029 Temp -6.178 1.941 -3.18 0.005 N@density 4.0250 0.9290 4.33 0.000 Where, ACP = accumulated compaction pressure PCSI = primary control sieve index FAc = FAc ratio (Bailey Method) Temp = temperature at the first roller pass, ?F N@field-density = number of gyrations to reach the post construction density level at lift thickness. An R 2 of 82.2 indicates that eighty two percent of the variability of ACP can be explained by these four factors. The positive coefficient for N@field-density indicates that ACP increases as the N@field-density increases. The negative coefficient for PCSI 102 indicates that finer mixes are easy to compact in the field (lower ACP). As the Bailey Method FAc ratio decreases (negative sign) compactability of the mixture in the field increases. As expected, higher mat temperatures required less compactive energy to reach the desired density level. 4.6 Correlations between ACP@92%G mm and lab compaction parameters The fourth part of this study involved eleven mixtures placed on the 2006 Test Track experiment and twelve mixes placed on the 2003 experiment. These mixes were used to evaluate the field compactability indicator by conducting nuclear density testing and taking temperature readings after each roller pass on each section and obtaining the plot of density versus roller pass. Surface temperatures were obtained with an infrared temperature gun. The purpose of this part was to obtain the field compaction energy at the same level of compaction of laboratory samples and correlate that energy with lab compaction parameters. The twenty three mixes were used to evaluate the relationship between field compaction energy at 92% of G mm and laboratory compaction parameters. Once again, based on the conceptual hypothesis explained previously, it was expected that the ACP@92%G mm would increase as the following parameters increase: %G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and CA ratio. In terms of mixture properties, it was expected an increase in ACP@92%G mm as the PG grade increases, asphalt content decreases and the % passing 0.075mm increase. In addition, for low mix temperatures 103 (<225 ?F), thin layers (<50 mm) and low t/NMAS ratios (<3:1), an increase in ACP@92%G mm was expected. As can be seen in Table 4.13, none of the individual laboratory compactability parameters had a strong correlation with ACP@92%G mm . The compactability of a mix in the field can be explained by a strong correlation with lift thickness (R = -0.81) and the negative sign indicates that the compaction energy applied to reach 92% of G mm decreases as the thickness increases for a range of thicknesses between 25 to 80 mm (see Table C2 of Appendix C). Table 4.13: Correlation and expected trend between ACP@92% of G mm and compactability parameters Parameter ACP@92% Expected trend PCSI 0.11 Incorrect CA ratio -0.08 Inconclusive FAc ratio -0.02 Inconclusive % pass 0.075 mm -0.07 Inconclusive AC% 0.08 Inconclusive Thickness (mm) -0.81 Correct Actual PG 0.21 Correct %G mm @N ini -0.19 Correct N@92%G mm 0.23 Correct CEI 0.04 Inconclusive Slope -0.02 Inconclusive Locking Point 0.11 Correct NMAS -0.38 Incorrect t/NMAS -0.45 Correct Temperature -0.19 Correct A multiple regression analysis was performed using the following parameters as predictor variables: asphalt content, actual PG grade, compaction slope, Compaction Energy Index (CEI), %G mm @N ini , N@92%G mm , locking point, coarse and fine aggregate ratios, lift temperature, t/NMAS ratio and PCSI. The best model was selected by 104 choosing the model with the least number of predictor variables, the highest adjusted R 2 , Cp less than the number of predictor variables and minimum standard error of the regression. The best model included the ACP@92% G mm as the response, while the interaction temperature*thickness, % passing No 200 sieve, actual PG grade, slope, locking point/Slope ratio, FAc ratio and PCSI square were the predictors (see Equation 20). Note that locking point and compaction slope are determined from normal height SGC specimens. ACP@92% = - 1884 - 730 FAc + 62.1 % pass 200 + 17.8 Actual PG + 57.3 Slope + 249 Lp/Slope - 0.0943 T*H - 0.504 PCSI^2 [20] S = 95.6737 R-Sq = 92.6% R-Sq(adj) = 89.1% Predictor Coef SE Coef T P Constant -1883.6 482.6 -3.90 0.001 FAc -730.1 315.1 -2.32 0.035 % pass 200 62.08 21.46 2.89 0.011 Actual PG 17.768 4.283 4.15 0.001 Slope 57.34 14.68 3.91 0.001 Lp/Slope 249.45 46.19 5.40 0.000 T*H -0.094266 0.008095 -11.64 0.000 PCSI^2 -0.5038 0.2084 -2.42 0.029 Where, ACP@92% = accumulated compaction pressure at 92% of G mm FAc = Fine aggregate coarse ratio (Bailey Method) %pass200 = percent passing 0.075mm sieve Actual PG = high temperature grade at failure Slope = Compaction slope for design specimens LP/Slope = Locking point / slope ratio for design specimens PCSI = primary control sieve index T2*H = lift thickness (mm) multiplied by temperature at the first roller pass, ?F 105 Equation 20 shows that the compactive effort applied to the mixture to reach 92% of G mm decreases as the variable temperature*thickness increases, which is the most significant variable. The positive sign of %pass200 indicates that higher amounts of material passing the 0.075mm sieve are related to stiffer mixes which tend to be difficult to compact in the field. This model also shows that the use of stiffer binders also increases the required compactive effort. It can be seen that the ACP@92% of G mm increases as the locking point/Slope ratio increases. The locking point of the mix (LP 2-1) and the slope are strongly correlated and as these parameters increase the compactability of the mixture decreases. Both variables, slope and locking point, resulted significant when included individually in the model but a multicollinearity problem was detected due to their strong correlation. As described by the Bailey Method, as the FAc ratio decreases (negative sign) compactability of the mixture in the field increases. Finally, the negative sign of PCSI square indicates that fine-graded mixes and coarse-graded mixes with gradations highly deviated from the maximum density line tend to reach 92% of G mm easily (i.e. SMA and finer mixes). 4.7 Summary of Findings The main objective of the first part of this study was an evaluation of the SGC parameters (%G mm @N ini , N@92%G mm , CEI, Slope, Locking Point and Bailey Method ratios) to determine compactability of a mix in the laboratory. The results indicated that %G mm @N ini , N@92%G mm , CEI, Slope and Locking Point are highly correlated, as shown 106 in Table 4.1. Further analysis also indicated that fine-graded mixes are easier to compact compared to coarse-graded. In addition, it was found that the use of highly angular particles, tougher aggregate and mixtures with low asphalt contents tend to increase the required compactive effort to achieve a specific density in the laboratory. The primary objective of this research was to evaluate a variety of mixture characteristics and determine if they are correlated to compactability in the field. The Accumulated Compaction Pressure (ACP) was introduced as a field compactability measure based on the rolling operation. A strong correlation between ACP and any of the individual laboratory parameters mentioned above was not obtained. It was determined that the ACP was significantly affected by temperature of the mix, lift thickness and the field density level. Several models were developed to correlate laboratory and field compactability. These models took into account temperature, thickness and density to minimize differences between laboratory and field compaction. Table 4.14 shows a summary of models used to correlate field and laboratory compactability. The first model included Test Track tangent sections placed in 2000. Most of these mixes are characterized for having thicknesses of 50 mm. The parameter that better correlated with ACP at post-construction density level was the number of gyrations to reach 92% of G mm (R 2 = 0.45). Multiple linear regression analyses were used to include all the mixes placed in 2000 and 2003 and try to find a better correlation between field and laboratory compactability. From Table 4.14, it can be seen that the multiple regression did not improve the correlation between field and laboratory compactability (R 2 = 0.55). Further 107 investigation showed that lift thickness and the temperature of the mix measured at the beginning of the compaction process were significant factors that explained most of the variability observed in ACP. This effect can be clearly seen for the second model shown in Table 4.14. The third model shown in Table 4.14 included the parameter N@field-density obtained from specimens compacted in the SGC at thicknesses similar to those in the field. A better correlation was obtained for this analysis (R 2 = 0.62) and it indicates that the actual density level achieved with the compaction process is an important factor to take into account. Table 4.14: Summary of models used to correlate field and laboratory compactability Response variable Variables included in the model R 2 Characteristics ACP at field density level N@92%Gmm 0.45 Test Track tangent sections, first cycle (Figure 4.23) Square root of ACP at field density level Temperature NMAS CA ratio %pass 0.075 mm VFA N@field 0.55 Parameters obtained from QC specimens (Equation 18) ACP at field density level N@field - thickness/density 0.62 Parameter obtained from specimens compacted in the SGC at thicknesses similar to those in the field (Figure 4.24) ACP at field density level PCSI FAc ratio Temperature N@field-density 0.82 Parameters obtained from specimens compacted in the SGC at thicknesses similar to those in the field (Equation 19) ACP@92%Gmm FAc ratio PCSI % passing 0.075mm Actual PG Slope Locking point Thickness Temperature 0.92 Parameters obtained from QC specimens (Equation 20) 108 The fourth model shown in Table 4.14 also included the parameter N@field- density obtained from specimens compacted in the SGC at thicknesses similar to those in the field. A combination of N@field-density, PCSI, FAc ratio and temperature did improve the relationship between field and laboratory compactability (R 2 = 0.82). This model suggests that taking an extra step during the mix design process by compacting specimens at thicknesses similar to those to be placed in the field may help predict the required field compactive effort to achieve the desired density. The first three models were characterized for having a response variable clearly affected by the post-construction density level which differs for each mixture. The last model shown in Table 4.14 used a field compaction energy calculated at a reference density level of 92% of G mm (ACP@92%G mm ). The best correlation between field and laboratory compactability was obtained with this model (R 2 = 0.92). This model suggests that a combination of laboratory parameters obtained from the original QC specimens (FAc ratio, slope and locking point), mix properties (% passing 0.075mm and actual PG) and factors affecting field compaction (thickness and temperature) may help predict the compactive effort applied by the rollers to achieve a minimum density level of 92%G mm . 4.8 Applicability of the ACP concept for validation purposes Sixteen surface mixtures placed in different U.S. states as part of the NCHRP 9-27 study were used to compare the results obtained from the analyses of the Test Track mixes. Table 4.15 shows the single correlation value (R-value) calculated for each pair of laboratory parameters used to describe compactability. The shaded cells indicate a better 109 correlation (R-value > 0.60) between parameters. CEI and %G mm @N ini have a strong correlation (R = -0.87), slope and %G mm @N ini have a very strong correlation (R = -0.99), and in this case a poor correlation was found between slope and locking point, contrary to the results obtained for test track mixes that is consistence with the use of much more variable data (see Appendix C). Table 4.15: Single correlation among laboratory parameters based on NCHRP 9-27 mixtures PCSI %G mm @N ini N@92%G mm CEI Slope LockPt1 PCSI 1 %G mm @N ini -0.41 1 N@92%G mm -0.21 -0.29 1 CEI 0.20 -0.87 0.68 1 Slope 0.39 -0.99 0.16 0.80 1 LockPt1 -0.59 -0.10 0.80 0.40 0.03 1 FAc Ratio -0.71 0.19 -0.11 -0.19 -0.11 0.27 Accumulated compaction pressure ACP was computed for the sixteen projects and specimens were also compacted in the SGC to meet the 92 percent of G mm and lift thickness. These data were used to compare the results obtained for Equation 19. Figure 4.28 shows the relationship between the actual ACP and predicted ACP for test track mixes and NCHRP 9-27 projects using Equation 18. Notice that this equation provides a relatively good relationship (R 2 = 0.77) between actual and predicted ACP for NCHRP projects which present higher construction variability. A slight deviation from the line of equality indicates that Equation 19 underestimated the accumulated compaction pressure. A calibration factor of 1.15 was needed to account for local conditions and make the model applicable to these observations. R 2 = 0.82 0 400 800 1200 1600 2000 0 400 800 1200 1600 2000 Predicted ACP for Test Track sections Ac t u a l A C P Line of Equality R 2 = 0.77 0 400 800 1200 1600 2000 0 400 800 1200 1600 2000 Predicted ACP for NCHRP 9-27 projects Ac t u a l A C P Line of Equality Figure 4.28: Comparison of predicted ACP for test track sections and NCHRP 9-27 projects using Equation 18 110 111 CHAPTER 5. CONCLUSIONS AND RECOMMENDATIONS In terms of parameters used to measure laboratory compactability, it was found that coarse-graded mixtures with low CA ratios require stronger fine aggregate structure (higher FA C ratio) and higher asphalt content to meet the target volumetric properties. Mixtures with these characteristics tend to be difficult to compact in the laboratory. Overall, finer gradations and gradations with PCSI values close to zero tended to increase mixture compactability in the laboratory. CEI, N@92%G mm , Slope and Locking Point can be used to represent the applied energy to reach a level of compaction and mixture resistance to deformation. PCSI and FA C ratio describe gradation properties and how particles are packed together and also can be used as laboratory compactability parameters. This points out that from the selection of an optimum gradation, compactability of a mixture in the SGC may be predicted. The relationship between laboratory and field compactability is complicated by many other variables which are known to affect compaction. A simple linear correlation was not found to describe that relationship. The use of multiple regression analysis was necessary to incorporate the wide variety of mix properties and factors affecting field compactability. Results showed that field compaction is more affected by t/NMAS ratio and mat temperature. 112 In general, it was found that mixes placed in 2003 required more compactive effort than mixes placed in 2000. Thinner layers were placed in 2003 (< 50 mm), which led to a reduction in temperature and finally an increase in compaction energy and a reduction in compaction time to achieve the desired density. The results also suggested that more energy was applied to reach a higher density level and that may explain some variability observed on the ACP. The interaction between temperature and t/NMAS ratio showed that for high t/NMAS ratios (above 4:1) the temperature has minimum effect on the compaction energy and as the t/NMAS ratio decreases lower temperatures required substantially higher compaction energy. The relationship between compaction effort and thickness or t/NMAS ratio for specimens compacted in the SGC to reach lift thickness showed similar trend to the relationship between ACP and thickness or t/NMAS ratio. In other words, specimens compacted in the SGC are significantly affected by thickness and this is consistent with the results observed in the field. Future investigation should be addressed to reduce the number of variables and increase the number of observations. The results of this study showed that the data can be grouped mainly by thickness. Keeping the same aggregate source and same rolling pattern may improve the relationship between ACP and lab parameters. Further investigation should also include refinement of the ACP concept. 113 Current mix design procedures do not provide any specific criteria to estimate achievable roller compacted density. This study can be used as baseline to develop procedures and criteria to identify compaction equipment characteristics and rolling patterns relative to lift thicknesses, mix temperature, and asphalt/aggregate/mixture characteristics that will produce optimum achievable density. 114 REFERENCES 1. Brown, Decker, Mallick and Bukowski. Superpave Construction Issues and Early Performance Evaluation. Journal of the Association of Asphalt Paving Technologists, Vol. 68, pp. 613-623, 1999. 2. Brown, E. Ray, Hainin, M. Rosli, Cooley, Allen, Hurley, Graham. NCHRP Report 531. Relationships of HMA in-place air voids, lift thickness, and permeability. National Cooperative Highway Research Program. Transportation Research Board, National Research Council, Washington, D. C., 2004. 3. Prowell, B. D., E. R. Brown. NCHRP Project 9-09(1) Verification of Gyration Levels in the Ndesign Table. National Cooperative Highway Research Program. Transportation Research Board, National Research Council, Washington, D. C., 2006. 4. Bahia, Hussain U., Frie mel Timothy P., Peterson, Pehr A., Russell, Jeffery S., and Poehnelt, Brian. Optimization of Constructibility and Resistance to Traffic: A New Design Approach for HMA Using the Superpave Compactor. Asphalt Paving Technology, Journal of the Association of Asphalt Paving Technologists, Vol. 67, pp. 189-232, 1998. 5. Pine, Bill, ?The Bailey Method. Achieving volumetrics and HMA Compactability.? Heritage Research Group. November 2004. 115 6. Roberts, F. L. , P. S. Kandhal, E. R. Brown, D-Y, Lee, and T. W. Kennedy, Hot Mix Asphalt Materials, Mixture Design and Construction. Second Edition, NAPA Education Foundation, Lanham, MD, 1996. 7. Scherocman, James A. and Acott, Mike. Hot mix asphalt production, placement, and compaction related to pavement performance. Journal of the Association of Asphalt Paving Technologists, Vol. 58, Nashville, TN, 1989. 8. US Army Corps of Engineers. Hot-Mix Asphalt Paving Handbook 2000. AC 150/5370-14A Appendix 1, 2nd Edition, 2000. 9. Garcia, J., K. Hansen. ?HMA Pavement Mix Type Selection Guide.? National Asphalt Pavement Association. Information Series 128, 2001. 10. Prowell, B. D., E. R. Brown and M. Huner, Evaluation of the Internal Angle of Gyration of Superpave Gyratory Compactors in Alabama, NCAT 03-04. National Center for Asphalt Technology, 2003. 11. Al-Khaateeb, G., C. Paugh, K. Stuart, T. Harman, and J. D?Angelo. ?Target and Tolerance for the Angle of Gyration Used in the Superpave Gyratory Compactor (SGC).? In Transportation Research Record 1789, TRB, National Research Council, Washington, DC. 2002. Pp. 208-215. 12. Prowell, Brian. Verification of the Superpave Gyratory Ndesign Compaction Levels. Doctoral Dissertation. Auburn University, 2006. 116 13. The Asphalt Institute. Superpave Mix Design (SP-2), Third Edition, 2000. 14. Anderson, R. Michael, Turner, Pamela A., Peterson Robert L. NCHRP Report 478. Relationship of superpave gyratory compaction properties to HMA rutting behavior. National Cooperative Highway Research Program. Transportation Research Board, National Research Council, Washington, D. C., 2002. 15. Anderson, R. M., and H. U. Bahia. ?Evaluation and Selection of Aggregate Gradations for Asphalt Mixtures Using Superpave,? Transportation Research Record 1583, Transportation Research Board, National Research Council, Washington, DC, 1997. 16. Pine, W. J., Superpave Gyratory Compaction and the Ndesign Table, Internal Report to Illinois Department of Transportation, Illinois, 1997. 17. Vavrik, W. R., and S. H. Carpenter, ?Calculating Air Voids at Specified Numbers of Gyrations in Superpave Gyratory Compactor,? In Transportation Research Record 1630, Transportation Research Board, National Research Council, Washington, DC, 1998, Pp 117-125. 18. Delage, Kenneth P. The Effect of Fine Aggregate Angularity on Hot Mixture Asphalt Performance. Master Degree Thesis, Department of Civil and Environmental Engineering, University of Wisconsin-Madison, November 2000. 19. M. Guler and H. U. Bahia ?Development of a Device for Measuring Shear Resistance of HMA in the Gyratory Compactor.? Presented at the meeting of the Transportation Research Board, Paper No. 00-1318, Washington, D.C., January 2000. 117 20. Vavrik, Pine, Huber, Carpenter and Bailey, ?The Bailey Method of Gradation Evaluation,? Proceedings of the Association of Asphalt Paving Technologists, Vol.70, 2001. 21. Grootenboer, H..J. Compaction of Asphalt Road Pavements. University of Twente, Netherlands. H.L. ter Huerne, Enschede, 2004. 22. Bonnot, J., ?Asphalt analysis, sulfur, mixes, and seal coats?, Transportation Research Record, no.1096, 1986. 23. Brown, Cooley, Hanson, Lynn, Powell, Prowell and Watson. NCAT Test Track Design, Construction, and Performance. NCAT 02-12. National Center for Asphalt Technology, 2002. 24. Devore, Jay L., ?Probability and statistics for Engineering and the Sciences.? Fifth Edition, 2000. 25. Johnson, Richard and Wichern, Dean. ?Applied Multivariate Statistical Analysis?, Fifth Edition, 2002. 26. Minitab Reference Manual, Release 10 for Windows, Minitab Inc., State College, PA, 1994. 118 APPENDIX A MATERIAL PROPERTIES 119 Table A.1: Mixture properties Quad Sec* Cycle** Sublot Initial field density %G mm Lab. Air Voids % VMA % VFA % Asphalt Content % % Pass PCS*** E 1 1 S 94.0 3.41 16 79 5.3 54 E 2 1 S 94.7 2.78 12 77 4.7 29 E 3 1 S 93.5 3.89 12 68 4.8 29 E 4 1+2 S 93.8 3.73 12 70 4.7 29 E 5 1+2 S 92.7 3.63 13 73 5.1 40 E 6 1+2 S 92.9 4.09 13 69 5 37 E 7 1+2 S 93.2 3.53 13 72 4.8 38 E 8 1+2 S 92.7 4.25 16 73 5.6 51 E 9 1+2 S 92.9 4.66 16 70 5.4 49 E 10 1 S 93.0 3.58 15 77 5.8 51 N 1 1 S 95.1 2.38 13 81 7.4 52 N 2 1 S 94.7 2.11 13 84 7.8 50 N 3 1 S 94.1 2.41 13 81 7.6 51 N 4 1 S 93.4 3.75 12 69 6.8 52 N 5 1 S 93.8 3.86 13 71 6.9 38 N 6 1 S 94.4 3.61 13 72 6.8 37 N 7 1 S 93.9 2.45 13 81 6.9 36 N 8 1 S 94.7 4.15 13 72 6.6 37 N 9 1 S 94.5 3.19 12 74 6.7 40 N 10 1 S 94.7 3.53 13 74 6.8 34 N 11 1+2 S 93.1 3.48 11 70 4.3 37 S 1 1 S 94.8 2.71 14 80 5 36 S 2 1+2 S 93.8 4.90 14 65 6 41 S 3 1+2 S 92.7 3.29 13 75 5.6 43 S 4 1 S 94.3 2.04 13 84 5.3 46 S 5 1 S 94.9 3.56 14 76 5.6 45 S 6 1+2 S 92.9 4.50 x x 6.2 53 S 7 1+2 S 93.2 3.30 x x 6.6 34 S 8 1+2 S 91.8 2.48 12 80 4.2 38 S 9 1+2 S 93.4 3.93 14 72 4.7 36 S 10 1+2 S 93.7 3.07 15 79 5.2 52 S 11 1+2 S 93.2 3.23 13 75 3.9 47 * Cycle 1 = 2000 Test track experiment, cycle 1+2 = 2000 and 2003, cycle 2 = 2003 only. ** S = surface mixture, B = binder or bottom mixture. *** PCS = Primary Control Sieve 120 Table A.1 (continued): Mixture properties Quad Sec* Cycle** Sublot Initial field density %G mm Lab. Air Voids % VMA % VFA % Asphalt Content % % Pass PCS*** W 6 1 S 92.1 2.59 x x 6.2 45 W 9 1 S 93.6 3.91 x x 5 33 W 10 1+2 S 93.3 3.77 x x 5 33 E 1 2 S 96.4 4.62 17 73 6.3 23 E 2 2 S 94.8 3.73 14 73 7.8 50 E 3 2 S 93.2 3.55 13 73 8.2 54 N 1 2 S 92.8 4.39 18 76 6.2 63 N 3 2 S 92.8 5.65 17 67 6.1 63 N 4 2 S 93.4 5.53 19 71 6.1 61 N 5 2 S 93.3 5.43 19 71 6.1 61 N 6 2 S 93.7 5.00 19 74 6.2 62 N 9 2 S 95.1 5.03 17 70 6.6 17 N 10 2 S 95.6 4.28 17 75 6.2 17 N 13 2 S 94.6 2.90 16 82 5.9 21 S 1 2 S 95.6 2.05 16 87 5.1 25 S 5 2 S 93.1 2.88 14 79 5.6 43 W 2 2 S 96.8 5.03 19 74 9.7 22 W 3 2 S 92.1 2.75 14 80 6.2 51 W 6 2 S 92.2 4.01 16 75 6.1 50 W 9 2 S 93.4 3.87 18 78 5.8 61 S 1 1 B 93.7 3.26 x x 5 32 S 2 1+2 B 93.0 4.76 x x 4.9 46 S 3 1+2 B 92.8 3.82 x x 4.2 47 S 4 1 B 93.6 4.43 x x 4.1 48 S 5 1 B 91.5 3.26 x x 4 53 S 11 1+2 B 94.6 2.28 x x 3.6 38 N 11 1+2 B 92.7 3.23 x x 4.1 46 * Cycle 1 = 2000 Test track experiment, cycle 1+2 = 2000 and 2003, cycle 2 = 2003 only. ** S = surface mixture, B = binder or bottom mixture. *** PCS = Primary Control Sieve 121 Table A.1 (continued): Mixture properties Quad Sec* Cycle** Sublot Initial field density %G mm Lab. Air Voids % VMA % VFA % Asphalt Content % % Pass PCS*** N 2 2 B 93.9 4.79 14 66 4.3 51 N 3 2 B 93.3 4.66 14 67 4.3 51 N 3 2 B 93.7 3.06 13 76 4.5 53 N 3 2 B 93.0 5.06 16 68 4.3 57 N 3 2 B 94.6 4.02 15 73 4.6 50 N 4 2 B 92.9 4.67 15 69 4.3 52 N 4 2 B 93.2 3.35 14 76 4.4 51 N 5 2 B 92.9 4.29 15 71 4.3 52 N 5 2 B 92.8 3.02 14 78 4.4 51 N 5 2 B 93.2 3.27 15 78 4.7 49 N 6 2 B 94.1 4.88 15 67 4.6 52 N 6 2 B 93.4 3.12 14 78 4.5 52 N 6 2 B 96.0 2.92 14 79 5 53 N 7 2 B 94.3 4.59 15 69 4.6 52 N 7 2 B 93.3 3.07 14 78 4.5 52 N 7 2 B 95.0 2.88 14 79 5 53 N 8 2 B 93.0 4.80 15 68 4.6 52 N 8 2 B 93.0 2.59 13 80 4.5 52 N 9 2 B 95.2 5.85 18 68 6.2 15 N 10 2 B 97.5 6.20 19 67 6.3 18 N 13 2 B 93.9 3.09 11 72 4.3 42 S 1 2 B 95.7 1.81 12 85 4.9 43 * Cycle 1 = 2000 Test track experiment, cycle 1+2 = 2000 and 2003, cycle 2 = 2003 only. ** S = surface mixture, B = binder or bottom mixture. *** PCS = Primary Control Sieve 122 Table A.2: Gradations of 2000 and 2003 test track experiment Percentage Passing (%) Quad Sec Cycle Sub lot 1" 3/4" 1/2" 3/8" No. 4 No. 8 No. 16 No. 30 No. 50 No. 100 No. 200 E 1 1 S 100 100 99 92 73 54 38 25 14 9 7.4 E 2 1 S 100 100 96 74 41 29 22 18 12 7 4.1 E 3 1 S 100 100 94 73 41 29 23 18 12 7 4.2 E 4 1+2 S 100 100 95 75 42 29 23 18 13 8 4.6 E 5 1+2 S 100 100 98 83 54 40 30 24 16 9 5.1 E 6 1+2 S 100 100 96 81 52 37 28 22 15 8 4.3 E 7 1+2 S 100 100 97 83 53 38 29 22 16 9 5.2 E 8 1+2 S 100 100 98 86 66 51 38 28 18 10 5.2 E 9 1+2 S 100 100 97 85 64 49 36 27 18 10 5.2 E 10 1 S 100 100 97 87 67 51 38 29 19 10 5.6 N 1 1 S 100 100 100 92 69 52 33 22 15 10 6.7 N 2 1 S 100 100 99 90 66 50 33 22 16 11 7.6 N 3 1 S 100 100 99 91 68 51 33 22 15 10 6.5 N 4 1 S 100 100 99 91 68 52 35 23 15 9 6.0 N 5 1 S 100 100 99 84 52 38 26 18 14 11 8.3 N 6 1 S 100 100 99 85 54 37 25 17 13 10 8.2 N 7 1 S 100 100 98 83 52 36 24 17 13 10 7.8 N 8 1 S 100 100 99 85 55 37 24 17 13 10 7.5 N 9 1 S 100 100 99 87 57 40 26 19 14 11 8.8 N 10 1 S 100 100 98 84 51 34 23 17 13 10 7.7 N 11 1+2 S 100 100 97 80 52 37 30 24 18 11 7.2 N 12 1+2 S 100 100 96 73 32 23 21 19 17 14 11.8 N 13 1 S 100 100 99 74 30 25 23 21 17 13 11.5 W 1 1+2 S 100 100 95 68 28 20 18 16 14 12 9.7 W 2 1 S 100 100 98 77 35 24 17 15 13 12 10.7 W 3 1 S 100 100 98 68 19 13 11 10 9 8 6.8 W 4 1+2 S 100 100 95 66 23 14 13 12 11 10 8.6 W 5 1+2 S 100 100 95 67 22 15 12 11 11 10 8.5 W 6 1 S 100 100 99 89 65 45 28 18 13 10 7.8 W 7 1+2 S 100 100 95 74 32 23 18 15 12 9 5.9 W 8 1 S 100 100 99 80 33 25 22 20 18 15 12.9 W 9 1 S 100 100 96 80 51 34 22 16 12 9 6.7 W 10 1+2 S 100 100 96 81 51 33 22 16 12 9 6.5 123 Table A.2 (continued): Gradations of 2000 and 2003 test track experiment Percentage Passing (%) Quad Sec Cycle Sub lot 1" 3/4" 1/2" 3/8" No. 4 No. 8 No. 16 No. 30 No. 50 No. 100 No. 200 S 1 1 S 100 100 95 86 54 36 28 21 15 9 5.5 S 2 1+2 S 100 100 100 96 67 41 29 22 15 10 8.4 S 3 1+2 S 100 100 100 100 70 43 29 21 15 11 8.9 S 4 1 S 100 100 98 88 63 46 33 23 13 9 7.8 S 5 1 S 100 100 95 82 61 45 33 22 10 7 5.0 S 6 1+2 S 100 100 95 87 74 53 41 33 24 12 5.9 S 7 1+2 S 100 100 96 88 71 34 25 20 16 10 6.2 S 8 1+2 S 100 100 100 93 58 38 25 19 15 12 7.8 S 9 1+2 S 100 100 93 82 53 36 27 20 14 9 5.7 S 10 1+2 S 100 100 95 88 69 52 38 27 19 11 6.6 S 11 1+2 S 100 100 100 92 62 47 30 22 17 13 7.5 S 12 1+2 S 100 100 97 82 63 46 32 23 16 10 7.0 S 13 1+2 S 100 100 93 80 68 50 37 27 19 11 6.6 E 1 2 S 100 100 91 69 35 23 17 14 12 11 10.0 E 2 2 S 100 100 96 93 73 55 44 37 24 10 5.1 E 3 2 S 100 100 96 92 73 54 43 36 24 10 5.3 N 1 2 S 100 100 100 100 81 63 51 38 20 12 7.0 N 2 2 S 100 100 100 100 80 63 51 38 21 12 6.6 N 3 2 S 100 100 100 100 80 63 51 38 21 12 6.6 N 4 2 S 100 100 100 100 81 61 49 37 21 12 6.7 N 5 2 S 100 100 100 100 81 61 49 37 21 12 6.7 N 6 2 S 100 100 100 100 81 62 50 37 21 12 6.8 N 7 2 S 100 100 100 100 49 24 20 17 14 12 9.2 N 8 2 S 100 100 100 100 49 24 20 17 14 12 9.2 N 9 2 S 100 100 97 83 37 17 13 12 11 10 8.6 N 10 2 S 100 100 95 87 30 21 17 15 14 13 11.5 N 13 2 S 100 100 95 71 32 21 18 16 15 14 12.1 W 2 2 S 100 100 88 54 22 17 14 13 12 11 9.7 W 3 2 S 100 100 100 100 79 51 39 29 21 14 8.7 W 6 2 S 100 100 100 100 98 75 50 35 22 15 11.3 W 8 2 S 100 100 100 96 40 25 19 15 13 10 7.5 W 9 2 S 100 100 100 98 83 61 43 32 23 15 7.5 S 1 2 S 100 99 92 74 33 25 24 22 19 16 13.0 S 4 2 S 100 100 95 78 19 5 3 3 2 2 1.6 S 5 2 S 100 100 96 87 66 43 30 21 10 7 5.5 124 Table A.2 (continued): Gradations of 2000 and 2003 test track experiment Percentage Passing (%) Quad Sec Cycle Sub lot 1" 3/4" 1/2" 3/8" No. 4 No. 8 No. 16 No. 30 No. 50 No. 100 No. 200 S 1 1 B 100 97 66 48 32 24 20 16 11 7 4.1 S 2 1 B 100 100 86 69 46 30 23 19 11 7 5.5 S 3 1 B 100 97 86 80 47 27 20 16 12 9 7.3 S 4 1 B 100 99 88 69 48 38 30 24 15 9 6.5 S 5 1 B 100 95 83 73 53 36 27 21 15 12 8.7 S 11 1 B 100 100 86 70 38 26 18 14 12 10 7.2 N 11 1 B 100 100 81 70 46 34 27 21 15 10 6.3 N 5 2 B 100 92 82 72 52 44 37 28 15 9 5.5 N 5 2 B 100 92 82 71 51 42 34 24 13 7 5.1 N 5 2 B 100 92 79 66 49 43 36 26 14 8 5.5 N 6 2 B 100 93 82 71 52 45 39 30 16 9 5.7 N 6 2 B 100 96 85 74 52 43 35 24 14 9 5.6 N 6 2 B 100 90 78 71 53 44 36 27 15 9 5.7 N 7 2 B 100 93 82 71 52 45 39 30 16 9 5.7 N 7 2 B 100 96 85 74 52 43 35 24 14 9 5.6 N 7 2 B 100 90 78 71 53 44 36 27 15 9 5.7 N 8 2 B 100 93 82 71 52 45 39 30 16 9 5.7 N 8 2 B 100 96 85 74 52 43 35 24 14 9 5.6 N 9 2 B 100 100 96 85 32 15 11 10 10 9 8.2 N 10 2 B 100 100 94 84 27 18 15 13 12 11 10.2 N 13 2 B 100 100 80 68 42 29 24 20 14 9 5.3 N 2 2 B 100 92 82 72 51 43 37 29 16 9 5.6 S 1 2 B 100 100 81 68 43 31 25 21 15 10 5.9 N 3 2 B 100 92 82 72 51 43 37 29 16 9 5.6 N 3 2 B 100 93 84 74 53 43 35 24 14 9 5.5 N 3 2 B 100 100 84 75 57 48 42 33 20 11 6.7 N 3 2 B 100 90 79 68 50 44 39 30 16 9 5.6 N 4 2 B 100 92 82 72 52 44 37 28 15 9 5.5 N 4 2 B 100 92 82 71 51 42 34 24 13 7 5.1 125 Table A.3: Parameters obtained from Densification Curve Quad Sec Cycle %G mm @N ini N@92 %G mm CEI TDI 92-96 TDI N@96 %G mm Slope Lock. Pt. E 1 1 89.2 20 20.0 147 237.9 79 6.78 43 E 2 1 88.8 19 20.4 107.3 282.5 63 7.66 48 E 3 1 87.3 28 56.3 128 183.9 96 7.99 49 E 4 1+2 87.9 25 41.7 159.6 203.9 89 7.64 47 E 5 1+2 89.4 18 15.4 165.4 238.1 84 6.33 40 E 6 1+2 89.0 21 23.5 201.5 201.5 100 6.27 42 E 7 1+2 89.5 17 13.7 156.5 245.6 81 6.32 38 E 8 1+2 89.8 18 12.8 205.1 205.1 100 5.44 33 E 9 1+2 89.2 21 22.3 172.2 172.2 100 5.55 37 E 10 1 90.4 14 5.5 167 262.7 80 5.48 35 N 1 1 88.1 21 31.3 90.7 285.9 61 8.68 51 N 2 1 88.2 20 27.1 84.3 306.6 57 8.84 56 N 3 1 87.9 23 37.2 91.2 270.5 65 8.85 60 N 4 1 86.5 32 82.7 140 174.3 92 8.90 57 N 5 1 85.0 38 130.9 133.8 153.1 95 10.17 63 N 6 1 84.8 38 134.3 115.2 162.8 91 10.56 64 N 7 1 86.1 28 74.5 88.3 247.9 69 10.43 61 N 8 1 84.9 40 145.3 136 136.0 98 9.99 62 N 9 1 85.9 31 87.9 114.3 201.7 81 9.95 61 N 10 1 85.5 35 108.1 120.6 174.3 89 10.03 64 N 11 1+2 89.9 16 10.6 147 258.4 77 6.08 37 S 1 1 89.4 17 15.2 104.2 300.5 59 7.23 42 S 2 1+2 86.4 37 103.1 119 119.0 100 7.92 51 S 3 1+2 86.3 29 76.8 119.9 207.2 81 9.51 56 S 4 1 89.5 18 17.0 100.3 403.2 60 7.39 46 S 5 1 89.5 20 21.3 201.2 287.8 102 6.06 38 S 6 1+2 88.9 24 29.8 170.8 170.8 100 6.04 40 S 7 1+2 85.4 35 113.5 111.8 182.4 86 10.27 59 S 8 1+2 87.1 23 45.9 90.2 276.6 62 9.53 54 S 9 1+2 88.1 25 41.2 156.7 193.9 93 7.30 44 S 10 1+2 90.1 15 7.6 128 282.8 68 6.20 40 S 11 1+2 88.5 21 27.8 127.5 249.6 75 7.51 47 W 6 1 86.5 28 67.6 94 239.9 71 9.98 61 W 9 1 86.8 31 77.9 137.3 178.4 88 8.45 47 W 10 1+2 87.4 27 52.8 156.2 197.8 91 8.02 47 E 1 2 84.2 26 70.7 49.4 49.4 49 12.14 50 E 2 2 89.5 18 16.5 158.5 231.0 84 6.18 39 E 3 2 89.6 17 13.1 158.5 245.6 82 6.27 36 N 1 2 89.3 17 12.7 149.1 149.1 80 5.95 38 N 3 2 88.3 26 38.9 76.8 76.8 80 5.75 37 N 4 2 88.4 25 33.4 84.4 84.4 80 5.73 33 126 Table A.3 (continued): Parameters obtained from Densification Curve Quad Sec Cycle %Gmm @N ini N@92 %G mm CEI TDI 92-96 TDI N@96 %G mm Slope Lock. Pt. N 5 2 88.5 24 31.7 90.3 90.3 80 5.78 36 N 6 2 88.7 21 24.3 111.3 111.3 80 5.96 34 N 9 2 82.0 43 208.1 54.6 54.6 75 12.60 74 N 10 2 84.1 33 111.4 91.3 91.3 75 11.29 66 N 13 2 87.2 15 14.5 49.5 112.8 37 10.72 49 S 1 2 87.4 13 8.8 38.9 142.7 30 11.42 47 S 5 2 89.7 13 5.5 85.3 212.5 48 7.21 43 W 2 2 83.0 49 239.8 88 88.0 80 10.89 59 W 3 2 85.8 28 77.2 95.7 238.6 71 10.43 58 W 6 2 86.0 31 86.6 171.5 171.5 80 9.06 53 W 9 2 89.2 13 5.4 97.3 97.3 47 7.56 43 S 1 1 87.3 25 51.97 123.4 226.46 78 8.63 50 S 2 1+2 87.3 32 73.43 134.6 136.37 100 7.23 48 S 3 1+2 85.6 35 109.93 132.2 164.24 94 9.65 60 S 4 1 88.7 29 46.88 208.7 208.69 125 6.03 34 S 5 1 87.8 27 57.04 156 288.17 94 7.83 49 S 11 1+2 86.9 24 48.24 83.3 282.42 61 9.85 58 N 11 1+2 89.0 19 19.95 129.5 259.85 71 7.08 45 N 2 2 89.8 16 9.28 133.5 135.52 80 5.08 32 N 3 2 90.0 15 7.96 144.4 144.40 80 5.08 31 N 3 2 91.0 9 0.86 104.6 244.66 49 5.59 31 N 3 2 89.6 18 12.71 120.5 120.49 80 5.02 31 N 3 2 90.6 12 3.45 140 188.57 69 5.11 30 N 4 2 90.0 15 6.83 144.1 144.10 80 5.07 29 N 4 2 90.8 10 1.24 117.1 224.51 57 5.50 32 N 5 2 90.3 13 5.00 167.9 167.87 79 5.15 32 N 5 2 91.0 10 0.83 99.7 238.47 50 5.63 34 N 5 2 91.0 9 0.54 133.1 231.56 54 5.41 32 N 6 2 89.8 16 8.48 132.3 132.33 80 5.02 31 N 6 2 91.4 8 0.10 103.2 248.60 49 5.19 27 N 6 2 91.4 8 0.34 91.5 258.60 45 5.34 33 N 7 2 90.0 15 6.96 146.4 146.42 80 5.14 32 N 7 2 91.3 9 0.49 107.1 245.64 50 5.31 26 N 7 2 91.5 8 0.21 92.2 261.75 45 5.28 32 N 8 2 89.8 16 9.31 134.7 134.65 80 5.07 32 N 8 2 91.5 8 0.21 78.6 275.86 40 5.56 32 N 9 2 80.9 49 280.10 30.3 30.30 75 12.87 71 N 10 2 82.5 49 238.91 27.3 27.32 75 11.00 65 N 13 2 88.6 19 23.31 124 258.60 71 7.61 44 S 1 2 87.8 12 6.69 35.3 150.92 29 11.24 48 127 Table A.4: Gradation parameters as indicators of compactability Quad Sec Cycle Sublot PCSI* CA Ratio FAc Ratio E 1 1 S 7 0.81 0.36 E 2 1 S -10 0.20 0.62 E 3 1 S -10 0.20 0.62 E 4 1+2 S -10 0.22 0.62 E 5 1+2 S 1 0.60 0.38 E 6 1+2 S -2 0.31 0.59 E 7 1+2 S -1 0.32 0.58 E 8 1+2 S 12 0.77 0.36 E 9 1+2 S 10 0.69 0.37 E 10 1 S 12 0.69 0.34 N 1 1 S 5 0.58 0.45 N 2 1 S 3 0.65 0.50 N 3 1 S 4 0.61 0.45 N 4 1 S 5 0.71 0.39 N 5 1 S -1 0.29 0.47 N 6 1 S -2 0.37 0.46 N 7 1 S -3 0.33 0.47 N 8 1 S -2 0.40 0.46 N 9 1 S 1 0.50 0.58 N 10 1 S -5 0.35 0.50 N 11 1+2 S -2 0.31 0.65 N 12 1+2 S -16 0.13 0.83 N 13 1 S -14 0.07 0.84 S 1 1 S -3 0.39 0.58 S 2 1+2 S -6 0.79 0.54 S 3 1+2 S -4 0.90 0.49 S 4 1 S 7 0.77 0.39 S 5 1 S 6 0.92 0.32 S 6 1+2 S 14 0.67 0.36 S 8 1+2 S -9 0.48 0.50 S 9 1+2 S -3 0.36 0.56 S 10 1+2 S 13 0.79 0.41 S 11 1+2 S 0 0.47 0.59 S 12 1+2 S 7 0.64 0.43 S 13 1+2 S 11 0.77 0.41 * PCSI = Primary control Sieve Index 128 Table A.4 (continued): Gradation parameters as indicators of compactability Quad Sec Cycle Sublot PCSI* CA Ratio FAc Ratio W 1 1+2 S -19 0.11 0.80 W 2 1 S -15 0.17 0.63 W 6 1 S 6 0.59 0.56 W 7 1+2 S -16 0.13 0.65 W 8 1 S -14 0.12 0.80 W 9 1 S -5 0.35 0.47 W 10 1+2 S -6 0.37 0.48 E 1 2 S -16 0.18 0.61 E 2 2 S 8 0.64 0.27 E 3 2 S 7 0.64 0.28 N 1 2 S 16 1.08 0.32 N 2 2 S 16 1.08 0.32 N 3 2 S 16 1.08 0.32 N 4 2 S 14 1.00 0.32 N 5 2 S 14 1.00 0.32 N 6 2 S 15 1.08 0.32 N 7 2 S -23 0.49 0.71 N 8 2 S -23 0.49 0.71 N 9 2 S -22 0.32 0.71 N 10 2 S -18 0.13 0.71 N 13 2 S -18 0.16 0.76 S 1 2 S -14 0.12 0.88 S 5 2 S 4 0.69 0.33 W 3 2 S 4 0.83 0.48 W 8 2 S -22 0.25 0.60 W 9 2 S 14 0.61 0.47 S 1 2 B -15 0.31 0.63 S 2 2 B -1 0.74 0.50 S 3 2 B 0 0.57 0.56 S 4 2 B 1 0.80 0.50 S 5 2 B 6 0.53 0.56 S 11 2 B -9 1.07 0.47 N 11 2 B -1 0.80 0.59 N 12 2 B 2 0.62 0.57 S 8 2 B -8 1.03 0.46 * PCSI = Primary control Sieve Index 129 Table A.4 (continued): Gradation parameters as indicators of compactability Quad Sec Cycle Sublot PCSI* CA Ratio FAc Ratio W 3 2 B -10 0.20 0.62 W 4 2 B -21 0.08 0.72 W 5 2 B -20 0.08 0.74 W 7 2 B -22 0.06 0.71 S 1 2 B -4 0.78 0.58 N 2 2 B 4 0.75 0.43 N 3 2 B 4 0.75 0.43 N 3 3 B 6 0.80 0.40 N 3 4 B 10 0.67 0.48 N 3 5 B 3 0.83 0.41 N 4 2 B 5 0.88 0.41 N 4 3 B 4 0.89 0.38 N 5 2 B 5 0.88 0.41 N 5 3 B 4 0.89 0.38 N 5 4 B 2 1.17 0.39 N 6 2 B 5 0.86 0.41 N 6 3 B 5 0.89 0.40 N 6 4 B 6 0.89 0.42 N 7 2 B 5 0.86 0.41 N 7 3 B 5 0.89 0.40 N 7 4 B 6 0.89 0.42 N 8 2 B 5 0.86 0.41 N 8 3 B 5 0.89 0.40 N 9 2 B -24 0.25 0.67 N 10 2 B -21 0.12 0.72 N 13 2 B -5 0.81 0.57 E 1 2 B -17 0.22 0.59 E 3 2 B 8 0.73 0.31 N 1 2 B 5 1.13 0.46 N 1 3 B 2 0.78 0.39 N 2 3 B 6 0.80 0.40 N 4 4 B 2 1.17 0.39 N 4 5 B 2 0.86 0.44 N 8 4 B 7 0.89 0.38 w 2 2 B -21 0.06 0.72 * PCSI = Primary control Sieve Index 130 Table A.5: Aggregate properties, Cycles I and II Quad Sec Cycle Primary Agg. Type CAA LA Abrasion Micro Deval FAA E 1 1 quartzite 35.4 26.0 4.92 46.1 E 2 1 granite 46.8 26.7 8.00 47.7 E 3 1 granite 46.8 26.9 8.00 47.6 E 4 1 granite 46.8 26.9 8.00 47.6 E 5 1 granite 47.0 27.8 8.00 47.7 E 6 1 granite 47.0 27.8 8.00 47.7 E 7 1 granite 47.0 27.8 8.00 47.7 E 8 1 granite 46.8 26.6 8.00 48.2 E 9 1 granite 46.8 26.6 8.00 48.2 E 10 1 granite 46.8 26.6 8.00 48.2 N 1 1 lms/slag 50.1 39.0 15.00 45.0 N 2 1 lms/slag 50.1 39.0 15.00 45.0 N 3 1 lms/slag 50.1 39.0 15.00 45.0 N 4 1 lms/slag 50.1 39.0 15.00 45.0 N 5 1 lms/slag 50.1 39.0 15.00 44.2 N 6 1 lms/slag 50.1 39.0 15.00 44.2 N 7 1 lms/slag 50.1 39.0 15.00 44.2 N 8 1 lms/slag 50.1 39.0 15.00 44.2 N 9 1 lms/slag 50.1 39.0 15.00 44.2 N 10 1 lms/slag 50.1 39.0 15.00 44.2 N 11 1 granite 46.5 25.0 8.00 47.6 N 12 1 granite 47.0 27.8 8.00 47.7 N 13 1 gravel 38.6 13.4 4.00 39.7 W 1 1 granite NA NA NA NA W 2 1 lms/slag NA NA NA NA W 3 1 lms/slag NA NA NA NA W 4 1 granite NA NA NA NA W 5 1 granite NA NA NA NA W 6 1 lms/slag 50.1 39.0 15.00 44.3 W 7 1 granite NA NA NA NA W 8 1 sndstn/lms/slag NA NA NA NA W 9 1 gravel 35.3 28.5 5.60 44.6 W 10 1 gravel 35.3 28.5 5.60 44.6 S 1 1 granite 45.5 45.8 12.98 48.3 131 Table A.5 (continued): Aggregate properties, Cycles I and II Quad Sec Cycle Primary Agg. Type CAA LA Abrasion Micro Deval FAA S 2 1 gravel 38.6 13.4 4.00 39.7 S 3 1 gvl/lms 37.3 15.3 5.29 43.8 S 4 1 limestone 47.2 12.0 4.00 42.2 S 5 1 gravel 37.9 14.0 2.00 44.6 S 6 1 lms/RAP 23.6 14.2 0.49 44.1 S 7 1 lms/RAP 13.2 7.9 0.27 44.0 S 8 1 mar. schist 47.9 22.0 17.00 50.1 S 9 1 granite 47.5 7.8 1.34 48.3 S 10 1 granite 47.5 9.7 1.67 48.2 S 11 1 mar. schist 47.9 22.0 17.00 50.1 S 12 1 lms NA NA NA NA S 13 1 granite 49.6 21.1 0.75 45.1 E 1 2 lms 17.6 8.5 45.3 49.0 E 2 2 marine lms 38.3 36.5 36.4 47.6 E 3 2 marine lms 38.3 36.5 36.4 47.6 N 1 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 2 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 3 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 4 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 5 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 6 2 grn/lms/snd 30.3 10.8 45.9 44.1 N 7 2 granite 33.5 9.9 47.4 46.1 N 8 2 granite 33.5 9.9 47.4 46.1 N 9 2 lms 37.3 17.5 47.5 43.8 N 10 2 lms/chert 23.9 8.8 47.3 44.4 N 13 2 granite 32.3 4.2 48.1 51.9 W 2 2 porph/lms 23.9 7.2 46.6 47.0 W 3 2 lms 36.6 29.3 48.2 48.5 W 6 2 lms/gvl/snd NA NA NA 46.7 W 8 2 granite 26.8 8.6 47.6 43.4 W 9 2 granite 28.4 11.6 51.1 42.3 S 1 2 granite 27.9 12.1 47.3 49.3 S 4 2 lms 18.9 9.1 47.9 44.2 S 5 2 gvl/lms/snd 15.8 3.1 43.9 42.5 132 APPENDIX B COMPACTION INFORMATION 133 Table B.1: ACP for 2000 sections Number of Roller Passes Quad Sec Sublot Vibratory Static Rubber T1* T2** T3*** Accumulated Compaction Pressure E 1 4 2 4 2 279 242 152 786 E 8 4 4 0 0 203 203 179 720 E 9 4 4 2 0 253 246 159 895 E 10 4 4 0 0 259 233 204 720 N 1 4 2 1 0 223 215 111 358 N 2 4 2 1 0 258 258 128 358 N 3 4 2 3 0 229 218 140 533 N 4 4 4 1 0 222 219 160 807 N 5 4 3 2 0 239 234 220 670 N 6 4 6 1 0 238 228 174 1256 N 7 4 6 0 0 240 240 220 1169 N 8 4 6 1 0 241 231 140 1256 N 9 4 6 0 0 227 214 180 1169 N 10 4 6 0 0 234 231 198 1169 N 11 4 2 2 0 295 280 174 445 N 12 4 2 0 0 320 293 285 271 N 13 4 2 1 0 248 235 156 358 S 1 4 4 1 0 270 270 156 807 S 2 4 7 5 0 285 262 184 1830 S 3 4 3 2 4 249 244 124 1002 S 4 4 3 1 0 275 269 221 583 S 5 4 5 0 0 275 228 187 945 S 6 4 3 0 0 278 253 233 495 S 7 4 4 1 0 279 237 163 807 S 8 4 3 2 0 281 266 173 670 S 9 4 2 0 0 274 258 246 271 S 10 4 4 2 0 243 230 137 895 S 11 4 2 4 2 294 277 151 786 S 12 4 2 4 6 308 265 148 1118 S 13 4 2 4 6 297 292 219 1118 W 1 4 3 1 0 282 278 172 583 W 2 4 4 0 0 248 224 196 720 W 3 4 4 3 0 249 233 200 982 W 4 4 4 0 0 280 246 215 720 W 5 4 3 3 0 276 205 156 757 W 6 4 3 1 0 268 255 215 583 W 7 4 2 1 0 295 276 238 358 W 8 4 1 3 0 304 288 269 374 *T1 = Temperature in ?F degress behind the paver **T2 = Temperature at first pass ***T3 = Temperature at the end of compaction 134 Table B.1 (continued): ACP for 2000 sections Number of Roller Passes Quad Sec Sublot Vibratory Static Rubber T1* T2** T3*** Accumulated Compaction Pressure W 9 4 4 2 0 284 263 153 895 W 10 4 4 2 0 276 258 160 895 E 1 2 4 2 0 290 269 148 895 N 1 2 3 2 0 237 237 166 670 N 2 2 2 4 0 241 232 224 620 N 3 2 4 3 0 236 213 139 982 N 4 2 4 1 0 227 209 144 807 N 5 2 3 3 0 225 224 149 757 N 6 2 3 5 0 199 199 160 932 N 7 2 3 5 0 221 221 133 932 N 8 2 5 0 0 251 251 216 945 N 9 2 6 4 0 218 218 132 1518 N 10 2 6 1 0 232 218 164 1256 N 11 2 3 1 0 243 237 220 583 N 12 2 3 1 5 244 209 138 998 S 1 2 1 3 0 290 279 186 374 S 2 2 1 2 0 302 288 264 287 S 3 2 4 0 0 287 281 256 720 S 4 2 2 5 12 312 276 179 1703 S 5 2 2 1 0 293 293 219 358 S 6 2 2 0 8 279 279 173 935 S 7 2 4 3 0 288 261 144 982 S 8 2 5 0 0 300 280 249 945 S 9 2 2 1 0 278 220 200 358 S 10 2 3 1 0 275 264 160 583 S 11 2 5 1 0 288 270 176 1032 S 12 2 2 4 3 271 251 178 869 S 13 2 1 3 0 284 284 267 374 W 1 2 6 1 0 298 270 185 1256 W 2 2 2 5 0 281 263 214 707 W 3 2 6 2 0 298 273 175 1344 W 4 2 2 3 0 290 279 235 533 W 5 2 2 2 0 286 284 254 445 W 6 2 2 0 0 268 266 254 271 W 7 2 4 0 0 280 272 252 720 W 8 2 2 0 0 271 249 244 271 W 9 2 2 1 0 263 263 201 358 W 10 2 3 2 0 260 256 189 670 *T1 = Temperature in ?F degress behind the paver **T2 = Temperature at first pass ***T3 = Temperature at the end of compaction 135 Table B.2: ACP for 2003 sections Number of Roller Passes Quad Sec Sublot Vibratory Static Rubber T1* T2** T3*** Accumulated Compaction Pressure E 1 1 2 3 0 281 251 187 529 E 1 2 1 6 0 253 209 176 659 E 2 1 3 2 17 248 217 130 2056 E 3 2 6 1 0 244 168 143 1178 N 1 1 2 6 0 162 150 129 806 N 1 2 4 3 8 275 228 141 1610 N 1 3 1 4 6 280 196 166 972 N 2 1 5 10 6 238 191 109 2300 N 2 3 1 5 10 298 242 173 1397 N 3 2 4 4 5 167 150 132 1453 N 3 3 5 3 10 237 236 175 1984 N 3 4 1 4 2 263 260 165 640 N 3 5 2 7 14 212 212 125 2061 N 4 1 8 0 4 295 214 120 1834 N 4 2 5 4 5 258 230 133 1662 N 4 3 3 4 5 285 194 165 1245 N 4 4 1 2 3 297 200 151 538 N 4 5 1 8 12 275 206 115 1840 N 5 1 5 4 9 194 168 117 1994 N 5 3 5 2 6 230 216 154 1560 N 5 4 3 6 5 318 318 117 1430 N 6 1 4 4 12 216 162 96 2034 N 6 2 2 4 6 251 226 166 1119 N 6 3 3 1 10 262 226 151 1382 N 6 4 2 2 2 248 248 183 602 N 7 1 9 3 10 223 205 113 2818 N 7 3 4 0 7 196 153 128 1249 N 7 4 1 4 7 260 236 184 1055 N 8 4 1 5 0 278 275 155 567 N 9 1 6 3 0 278 249 169 1363 N 10 1 5 3 0 280 245 159 1154 N 10 2 9 3 0 266 223 142 278 N 13 1 2 3 0 290 222 173 1988 N 13 2 2 6 6 212 212 123 1304 S 5 1 1 6 0 273 208 156 659 S 1 1 4 5 0 265 226 147 1131 W 2 1 14 14 0 272 247 121 4047 W 2 2 10 10 0 252 236 115 2844 W 3 1 0 5 4 286 207 138 795 W 9 1 1 4 3 230 158 109 723 *T1 = Temperature in ?F degress behind the paver **T2 = Temperature at first pass ***T3 = Temperature at the end of compaction 136 APPENDIX C INFORMATION OF FIELD AND LABORATORY STUDY 137 Table C.1: Specimens compacted at field thickness Quad Prod NMAS Initial density PCSI Field Thickness, mm ACP t/NMAS N@density N@92 t/field E1-2000 9.5 94.0 7 52 786.0 5.48 66 33 E8-2000 12.5 92.7 12 53 720.0 4.27 50 36 E10-2000 12.5 93.0 12 56 720.0 4.47 27 18 N1-2000 9.5 95.1 5 50 358.1 5.21 44 28 N3-2000 9.5 94.1 4 53 532.7 5.61 50 30 N4-2000 9.5 93.4 5 53 807.3 5.61 79 35 N6-2000 12.5 94.4 -2 52 1256.4 4.17 71 36 N7-2000 12.5 93.9 -3 50 1169.1 3.96 62 36 N8-2000 12.5 94.7 -2 50 1256.4 3.96 105 37 N10-2000 12.5 94.7 -5 53 1169.1 4.27 85 18 N11-2000 12.5 93.1 -2 104 445.4 8.33 15 10 S1-2000 12.5 94.8 -3 99 807.3 7.92 29 11 S3-2000 9.5 92.7 -4 102 1002.0 10.69 26 23 S4-2000 12.5 94.3 7 102 582.7 8.13 28 15 S5-2000 12.5 94.9 6 104 944.6 8.33 47 18 S6-2000 12.5 92.9 14 52 495.4 4.17 71 46 S7-2000 12.5 93.2 -5 51 807.3 4.06 71 45 S8-2000 9.5 91.8 -1 97 670.0 10.16 13 14 S11-2000 9.5 93.2 0 91 786.0 9.63 21 15 W6-2000 12.5 92.1 6 52 582.7 4.17 47 46 W9-2000 12.5 93.6 -5 51 894.6 4.06 59 38 W10-2000 12.5 93.3 -6 50 894.6 3.96 35 24 N11B-2000 19 92.7 -1 64 582.7 3.37 23 16 S1B-2000 19 93.7 -15 64 374.1 3.37 49 30 S2B-2000 19 93 -1 64 286.8 3.37 44 28 S5B-2000 19 91.5 6 64 358.1 3.37 30 36 S11B-2000 19 94.6 -9 53 1031.8 2.81 136 49 N10-2003 12.5 95.6 -18 51 1154.1 4.06 160 11 N13-2003 12.5 94.6 -18 43 1987.7 3.46 290 34 S1-2003 12.5 95.6 -14 43 1130.7 3.46 145 15 138 Table C.2: Properties of mixes for field study Cycle Qua. Sec. CA Ratio FAc Ratio % pass 200 AC% Thick . (mm) Actua l PG %G m m @Nin i N @ 92% G mm CEI E 1 0.18 0.61 10 6.3 46 78 84.2 26 70.7 N 1 1.13 0.46 5.8 4.5 53 81 89.3 17 12.7 N 4 0.88 0.41 5.5 4.3 43 81 90.0 15 6.83 N 5 1.00 0.32 6.7 6.1 23 81 88.5 24 31.7 N 6 1.08 0.32 6.8 6.2 28 70 88.7 21 24.3 S 1 0.12 0.88 13.0 5.1 43 78 87.4 13 8.8 N 9 0.32 0.71 8.6 6.6 46 74 82.0 43 208 N 10 0.13 0.71 11.5 6.2 51 74 84.1 33 111.4 W 3 0.83 0.48 8.7 6.2 33 69 85.8 28 77.2 W 9 0.25 0.60 7.5 5.8 25 69 89.2 13 5.4 E 3 0.73 0.31 6 7.9 56 78 89.6 17 13.1 2003 E 2 0.64 0.27 5.1 7.8 50 69 89.5 18 16.5 N 1 0.81 0.46 8.7 5.7 47 67 91.4 9 0.36 N 2 0.81 0.49 9.6 5.3 45 76 91.1 10 0.99 N 5 1.02 0.35 6.8 6.2 50 67 91.3 7 0.77 N 10 0.83 0.42 5.6 4.4 44 70 84.9 48 187 E 5 1.07 0.39 6.2 5.2 54 67 91.0 8 1.39 E 6 1.08 0.40 7.3 5.1 51 76 91.0 8 1.47 E 7 1.09 0.39 7.2 5.2 54 76 90.8 8 2.25 S 2 0.58 0.32 6.0 7.0 41 76 89.5 13 9.46 W 3 1.01 0.38 7.5 6.1 50 67 92.5 5 0.03 W 4 1.05 0.38 7.6 6.0 56 76 92.1 5 0.02 2006 W 5 1.00 0.44 8.3 5.1 52 70 91.8 6 0.03 139 Table C.2 (continued): Properties of mixes for field study Cycle Quad Sec. NMAS PCSI Slope LockP ACP ACP@92 %G mm Temp 1 st pass, F Temp. Behind Paver, F E 1 12.5 -16 12.14 50 553 144 251 281 N 1 19 5 5.95 38 1606 353 228 275 N 4 19 5 5.07 29 1707 350 230 258 N 5 9.5 14 5.78 36 2067 1200 168 NA N 6 9.5 15 5.96 34 1373 750 162 216 S 1 12.5 -14 11.42 47 1126 314 226 265 N 9 12.5 -22 12.60 74 1357 328 249 278 N 10 12.5 -18 11.29 66 923 353 245 280 W 3 9.5 4 10.43 58 934 840 221 286 W 9 9.5 14 7.56 43 843 750 158 230 E 3 9.5 8 6.27 36 1217 353 168 244 2003 E 2 9.5 8 6.18 39 1965 216 217 248 N 1 12.5 12 5.95 37 1805 406 198 215 N 2 12.5 11 5.93 33 1306 445 246 254 N 5 12.5 15 6.27 33 770 125 212 264 N 10 19 -6 9.01 57 875 410 263 277 E 5 12.5 3 6.17 34 1060 118 195 240 E 6 12.5 8 6.17 35 1570 465 183 276 E 7 12.5 9 6.26 35 1570 545 199 266 S 2 9.5 1 7.40 43 1850 765 215 230 W 3 12.5 15 5.82 33 705 48 240 247 W 4 12.5 15 5.99 34 595 47 225 260 2006 W 5 12.5 7 6.28 32 595 99 205 275 70% 72% 74% 76% 78% 80% 82% 84% 86% 88% 90% 92% 94% 96% 98% 100% 0 500 1000 1500 2000 ACP C o r r ect ed d e n s it y, % G m m N2 W3 W4 W5 E5 E6 E7 N5 N1 S2 N10 Figure C1: Change in density level as function of compaction pressure for mixes placed in 2006 140 141 Table C.3a: Laboratory and field properties for NCHRP 9-27 projects Project Prod NMAS PG Grade Grad Type Ndes Lab Voids AC % %Gmm @Ni N@ 92% Gmm CEI Slope LockPt VA-1 9.5 70-22 Fine 65 4.55 5.5 89.1 18 17.3 6.6 39 VA-2 19.0 64-22 Coarse 65 4.36 4.7 86.1 27 64.0 9.8 55 VA-3 9.5 64-22 Coarse 65 3.10 5.5 87.8 18 21.7 9.4 60 NC-1 9.5 70-22 Fine 100 5.08 6.9 87.9 31 60.3 6.5 41 CO-1 12.5 58-28 Coarse 75 2.15 6.2 88.1 16 15.1 9.5 51 MO-1 19.0 64-22 Coarse 100 5.53 4.2 82.7 62 339.6 10.3 65 MO-2 19.0 64-22 Coarse 100 4.35 4.5 82.8 49 250.0 11.7 69 UT-2 19.0 64-34 Coarse 125 3.18 4.6 85.9 36 128.4 9.5 56 AL-1 25.0 76-22 SMA 50 1.82 5 89.3 10 2.5 9.7 41 AL-2 25.0 67-22 Fine 100 4.70 3.5 52.5 26 521.8 39.1 40 AL-4 19.0 76-22 Coarse 100 3.55 4.1 87.1 27 57.9 8.5 51 AL-5 12.5 67-22 Coarse 86 2.77 5.5 91.6 8 0.1 5.2 36 FL-2 12.5 64-22 Fine 75 3.37 4.4 91.5 8 0.5 5.0 27 GA-1 12.5 67-22 Coarse 75 3.07 4.7 87.7 19 26.2 9.0 48 GA-2 9.5 67-22 Fine 75 2.38 5.4 89.8 12 4.3 7.6 42 MS-1 12.5 67-22 Fine 80 2.93 4.8 87.7 20 27.6 8.6 50 Table C.3b: Laboratory and field properties for NCHRP 9-27 projects Project PCSI CA FAc Initial density Temp Field Thickness, mm ACP t/NMAS N@density VA-1 5 0.64 0.29 91.7 250 38.10 1031 4.0 91 VA-2 -6 1.06 0.47 93.5 265 63.50 991 3.3 121 VA-3 -14 0.41 0.58 91 265 38.10 901 4.0 110 NC-1 14 0.93 0.24 90.6 230 31.80 1407 3.3 150 CO-1 0 0.61 0.55 94.4 230 57.20 1103 4.6 55 MO-1 -10 0.66 0.40 90.5 250 50.80 1673 2.7 109 MO-2 -3 0.87 0.32 93.4 250 101.60 1490 5.3 65 UT-2 -6 1.46 0.46 92.8 250 38.10 1718 2.0 175 AL-1 -10 0.00 0.57 94.4 260 60.90 832 2.4 24 AL-2 17 0.85 0.32 90.7 260 69.90 468 2.8 36 AL-4 1 0.60 0.40 88.5 250 57.20 450 3.0 14 AL-5 -2 0.44 0.56 91.3 230 38.10 970 3.0 80 FL-2 7 1.40 0.24 90 230 37.50 720 3.0 11 GA-1 -3 0.43 0.49 91.3 230 38.10 1475 3.0 61 GA-2 3 0.66 0.41 91.2 230 31.80 1628 3.3 154 MS-1 4 0.86 0.23 92 265 38.10 979 3.0 60 142 Table C.4: Field compaction information for NCHRP 9-27 projects TAC: time available for compaction, minutes Project Nearest City Pavement Date Weather Mix Temp behind paver thickness/ NMAS Asphalt Grade TAC Breakdown Roller Breakdown Passes Intermediate Roller Intermediate Passes Finish Roller Finish Passes T166 %Gmm Achieved AL-4 Troy over PCC 10/3/2002 90, little to no wind 315 3 76-22 120+ IR DD110 1, hi amp., hi freq., 2 static IR DD90 2, static 88.5% AL-3 Opelika interstate hwy 8/9/2002 60-65, night, slight breeze 290 2 76-22 8 Dynapac 5 to 6, vib. Dynapac 4, vib. Cowin, ST105 1, static 89.8% FL-2 Marianna county road 7/9/2002 90, humid, cloudy 255 3 64-22 23 IR DD110 4 static 90.0% MO-1 Kansas City over PCC 8/20/2002 70, clear, night 2.7 64-22 IR DD130 in echelon 5, hi amp., hi freq. IR PT220R pneumatic 5 IR DD90 2 static 90.5% NC-1 New Bern 2-lane state highway 5/29/2002 80, sunny, no wind 285 3.3 70-22 19 IR DD110HF 4 to 5, hi amp., hi freq. CAT CB634C 3, hi amp., hi freq. Hamm HD12 2 static 90.6% AL-2 Prattville new lane 8/29/2002 80, overcast 295 2.8 67-22 50 IR DD110 2, hi amp., hi freq., one static IR DD90 2 static 90.7% VA-3 Floyd 2-lane county hwy 5/23/2002 70-75, sunny 265 4 64-22 27 IR DD110HF 2 to 3, hi amp., hi freq. IR DD110HF 2 to 3, hi amp., hi freq. Dynapac 2 static 91.0% GA-2 Macon state hwy 6/23/2002 90, clear 3.3 67-22 IR DD 90 3, med. amp., med. freq., 2 static pneumatic 15 IR DD90 6 91.2% GA-1 Junction City state hwy 6/19/2002 85, humid, mostly cloudy 3 67-22 IR DD130 2, med. amp, med. freq., 4 static IR PT-125 pneumatic 7 IR DD90 7, static 91.3% AL-5 Banks US hwy 6/26/2002 95, partly cloudy 282 3 64-22 28 IR DD90 2, med. amp, med. freq., 3 static Dynapac 6 91.3% VA-1 Roanoke 2-lane county road 5/21/2002 70, overcast drizzle 310 4.0 70-22 19 IR DD90 2 to 3, hi amp., hi freq. 2 to 3 static IR DD90 3 static 91.7% MS-1 Starkville new highway ######## 65, clear 3 67-22 CAT CB 634C in echelon 4 vib, 1 static PS-150B pneumatic Dynapac CC42 static 92.0% UT-2 Fillmore interstate hwy 8/8/2002 90-95, sunny, windy 300 3 64-34 24 IR DD130 4, hi amp., hi freq. IR Propac 100DA 3 static, 2 vib. IR DD103 4 to 5, med. amp., med. freq. 92.8% MO-2 Joplin new highway 8/23/2002 95, sunny, no wind 315 4.1 64-22 120+ IR DD130 3 to 4, hi amp., hi freq. IR PT240R pneumatic 8 2 lo vib, 1 static 93.4% VA-2 Blacksburg new highway 5/22/2002 60, sunny 300 3.3 64-22 63 IR DD110HF 4 to 5, hi amp., hi freq. IR DD90 HF 4 to 5 static 93.5% UT-1 North Glendale granular base 8/5/2002 90-95, mostly sunny, windy 3 64-34 CAT CB 634C vib. CAT PS360B pneumatic CAT CB634C static 93.6% FL-1 Jacksonville agg base interstate shoulder 5/16/2002 90, sunny, windy 300 5.1 RA295 58 CAT CB 634C 4 to 5, static IR DD125 six static IR DD110 4 to 5 static 93.9% CO-1 Pagosa Springs unbound base 8/13/2002 80, sunny, light wind 255 4.6 58-28 10 IR DD130 3, hi amp., hi freq. CAT PS360B pneumatic 4 to 5 IR DD130 2 static 94.4% CO-2 Silverthorne existing hwy 8/14/2002 75, sunny, 15-20 mph winds 285 5.3 64-28 24 CAT CB 634C 3, med. amp., hi freq. CAT CB634C 3 to 4, static 94.4% AL-1 Opelika interstate hwy 7/23/2002 80-85, mostly sunny, slight breeze 315 2.4 76-22 64 Dynapac CC522 Dynapac CC522 IR ST105 3 to 4, static 94.4%