REVISED INTERIM REPORT: CHAPTERS 1 AND 2 NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM (NCHRP)

Transcription

1 REVISED INTERIM REPORT: CHAPTERS 1 AND 2 Submitted to the NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM (NCHRP) For Project NCHRP 1-42: Top-Down Fatigue Cracking of Hot-Mix Asphalt Layers LIMITED USE DOCUMENT This Interim Report is furnished only for review by members of the NCHRP project panel and is regarded as fully privileged. Dissemination of information included herein must be approved by the NCHRP. From Advanced Asphalt Technologies, LLC May 27, 2004

2 ACKNOWLEDGEMENT OF SPONSORSHIP The work was sponsored by the American Association of State Highway and Transportation Officials, in cooperation with the Federal Highway Administration, and was conducted in the National Cooperative Highway Research Program, which is administered by the Transportation Research Board of the National Research council. DISCLAIMER This is an uncorrected draft as submitted by the research agency. The opinions and conclusions expressed or implied in the report are those of the research agency. They are not necessarily those of the Transportation Research board, the National Research Council, the Federal Highway Administration, the American Association of State Highway and Transportation Officials, or the individual states participating in the National Cooperative Highway Research Program. i

3 TABLE OF CONTENTS ACKNOWLEDGMENT OF SPONSORSHIP... DISCLAIMER... LIST OF FIGURES... LIST OF TABLES... ACKNOWLEDGMENTS... ABSTRACT... SUMMARY OF FINDINGS INTRODUCTION... 1 RESEARCH PROBLEM... 1 RESEARCH OBJECTIVES... 2 SCOPE BACKGROUND... 8 REVIEW OF LITERATURE AND CURRENT PRACTICE... 8 PRELIMINARY FINDINGS CONCERNING TOP-DOWN CRACKING OF ASPHALT CONCRETE PAVEMENTS CHAPTER 2 REFERENCES i i iii iv v v v ii

4 LIST OF FIGURES 1. Top Down Cracking in Colorado Transverse Contact Stresses Predicted at Pavement Surface with Pavement Thickness = 10 cm, E 1 = 1,400 MPa, E 2 = 300 MPa Stress Intensity Factors as a Function of Crack Length at Various Distances from Applied Load Octahedral Shear Stress in an Asphalt Concrete Pavement as a Function of Depth, at Different Radial Distances from the Applied Load Shear Stresses as a Function of Depth for a 200 mm-thick Flexible Pavement; Rigid and Flexible Loading, with and without a Temperature Gradient Maximum Surface Tensile Stresses in Flexible Pavements in Michigan, as Calculated Using Michpave, a Layered Elastic Analysis Software Program Effect of Segregation on IDT Strengths of HMAC Mixtures in Michigan Segregation in Hot-Mix Asphalt Pavement Exhibiting Top-Down Cracking Diagram of Relationship between Slat Conveyors and Typical Location of Segregation and Top Down Cracks in Colorado Relationship Between Corrected and Uncorrected IDT Strength, Showing Regression Line, 95 % Prediction Intervals for New Observations, and d2s Error Bars for Measured Data IDT Creep and Strength Test Diagram Illustrating Concept of Dissipated Creep Strain Energy Comparison of Measured E* Values and Those Predicted Using the Hirsch Model; R 2 = 98 % Predicted and Measured G* Values for SHRP Core Asphalt Binders, Using Christensen s Two Point System and an Assumed Glassy Modulus of 1.0 GPa Estimated Viscosities after Age-Hardening as Estimated Using the Mirza- Witczak Global Aging System, and as Estimated Using the Proposed Modification of this System Predicted Aged Binder Complex Modulus Values as a Function of Temperature Predicted Aged Binder Complex Modulus Values as a Function of Depth Comparison of Mixture Compliance Values as Measured Using the IDT Creep Procedure and as Predicted using the Proposed System iii

5 19. Plot of Failure Envelopes Predicted by Maximum Normal Stress (Maximum Principal Stress), Maximum Shearing Stress and Maximum Distortion Energy Theories Plots of Damage And Stress in a 17.8-cm Thick Asphalt Concrete Pavement under a Non-Uniform 40 kn Load Continuum Damage Analysis of Flexural Fatigue Predicted Damage Modulus Values for VA Limestone Mixture/Fine Gradation/Optimum Binder Content, at the Conclusion of Flexural Fatigue Test Predicted Complex Modulus in Tension/Compression Compared with Measured Flexural Complex Modulus for SHRP Mixtures Predicted and Observed Values for Continuum Damage Fatigue Constant C 2 for SHRP Flexural Fatigue Data and NCHRP Uniaxial Fatigue Data Predicted and Measured Log Cycles to Failure for SHRP Flexural Fatigue Data, with d2s Confidence Limits Percent Observed Cracking as a Function of Calculated Damage for WesTrack Fatigue Experiment Healing Rate of Asphalt Mixtures Made with Five Different SHRP Core Binders LIST OF TABLES 1. Summary of Research on Top-Down Cracking Merrill s Summary of Features of Some Layered Elastic Analysis Packages for Application to Flexible Pavement Systems iv

6 ACKNOWLEDGMENTS This Revised Interim Report presents the work accomplished on NCHRP Project 1-42 through May 25, Advanced Asphalt Technologies, LLC (AAT), is the prime contractor for this project, and Abatech, Inc. (Abatech) is the only subcontractor. The Principal Investigator for this project is Dr. Donald Christensen of AAT. The effort at Abatech is being performed under the supervision of Dr. Geoff Rowe. This report was primarily compiled by Dr. Christensen, with input and assistance by Dr. Ray Bonaquist of AAT and Dr. Geoff Rowe of Abatech. ABSTRACT This Report is a partial summary of the interim results of NCHRP Project 1-42: Top-Down Fatigue Cracking of Hot-Mix Asphalt Layers. It includes a review of literature; the revised work plan for Phase II of the project is not included in this abbreviated report. The primary findings of the literature review are that top-down cracking in flexible pavements is primarily caused by traffic-associated fatigue and/or thermal stresses. Top-down cracking is significantly affected by the interaction of many factors, including poor compaction and segregation during construction, pavement structure, modulus gradients within the pavement, age hardening of the pavement surface, moisture damage and mixture fracture toughness. SUMMARY OF FINDINGS This Report is a partial summary of the interim results of NCHRP Project 1-42: Top-Down Fatigue Cracking of Hot-Mix Asphalt Layers. It includes a review of literature, but not a revised work plan for Phase II of the project. The literature review consists of two components a review of recent research publications dealing with top-down cracking in asphalt concrete, and the refinement of several models as needed for the proposed Phase II work plan. These models will also be useful to other pavement engineers analyzing top-down cracking and related phenomena in flexible pavements. The following list summarized the findings are made based upon this literature review and associated analysis. For the convenience of the reader, it has been organized according to the four project objectives, each finding being listed under the objective to which it most closely pertains. v

7 Objective 1. Identify the Mechanisms that govern the initiation and propagation of top-down cracking in HMA layers Accumulated damage associated with repeated traffic loading is the primary mechanism of top-down cracking in asphalt concrete pavements. It is also likely that thermal stresses contribute significantly to this form of distress. Both tire-contact surface stresses and shear stresses contribute to top-down cracking. Contact stresses are probably more important in the initiation of top-down cracks, while propagation of these cracks to significant depth in the pavement primarily results from shear stresses. Objective 2. Identify or develop method(s) of laboratory testing HMA mixtures for determining susceptibility of the HMA surface layer to this cracking The most effective means for reducing top-down cracking in asphalt concrete pavements at this time involve for the most part mix design top-down cracking can be reduced by ensuring that HMAC mixtures have adequate resistance to lowtemperature cracking, high fracture toughness, and good resistance to moisture damage and age hardening. For most pavement design and analysis problems, creep compliance and complex modulus values can be estimated with adequate accuracy using the Hirsch model and associated procedures. For evaluating new or unusual material, for forensic studies and for research purposes when laboratory testing is needed creep compliance should be determined using procedures given in AASHTO TP-9, and complex modulus should be determined using dynamic uniaxial compression, as developed in projects NCHRP 9-19 and The dissipated creep strain energy (DCSE) approach to estimating fracture toughness, as developed by Roque et al., appears to be very promising for evaluating the resistance of asphalt concrete mixtures to crack propagation, an important aspect of resistance to top-down cracking. However, further evaluation of this concept is needed, and is proposed in the Revised Phase II Work Plan for NCHRP Project vi

8 Objective 3. Determine the significant factors associated with the occurrence of top-down fatigue cracking. Mixture properties probably affect top-down cracking in many different complex ways. However, it is clear that increasing fracture toughness will reduce the extent of top-down cracking in flexible pavements. Increasing resistance to age hardening and moisture damage will also probably reduce the incidence of top-down cracking. Increased traffic loading, increasing magnitude of tire stresses, and complex tirepavement contact stresses are all likely to increase the incidence of top-down cracking. Traffic loading contributes to top-down cracking not only through contact stresses, but also through shear stresses extending deep into the pavement. Pavement age hardening and moisture damage contribute to top-down cracking in asphalt concrete pavements. Pavement thickness probably affects the occurrence of top-down cracking, but there is not clear agreement among researchers concerning how pavement thickness affects this form of distress. Stiffness gradients in pavements, arising from binder grade selection, mix design, age hardening and/or temperature gradients, also probably contribute to top-down cracking, but the nature and magnitude of this contribution is not yet clear. Segregation during construction, particularly slat-conveyor segregation near the wheel paths, can cause or exacerbate top-down cracking in asphalt concrete pavements. Objective 4. Identify promising models for predicting the initiation and propagation of top-down cracking. In the long term, the best approach to analyzing stresses and strains potentially leading to top-down cracking in flexible pavements is the three-dimensional finite element method. However, for this approach to be practical, it must be possible to perform thousands of analyses in a short period of time as will be done, for example, in the parametric study included in the Revised Phase II Work Plan. Threedimensional finite element analyses cannot yet be performed and interpreted quickly enough to perform such a large number of analyses; to date, most studies using this method have been limited to a few dozen analyses of idealized pavements. The effective evaluation of a pavement design and analysis method must involve the calculation of stresses, strains and damage over a period of at least several years, using realistically variable conditions of temperature, asphalt concrete modulus, sub grade stiffness, age hardening, and other factors; the results must then be compared to measured deflections and observed distress to determine if the design and analysis method is accurate. Such evaluations require thousands of separate analyses, and so vii

9 cannot be performed at this time on techniques involving three-dimensional finite element analysis. Therefore, further evaluations of models for top-down cracking that involve 3-dimenional finite element analysis cannot be performed during Phase II of NCHRP Project Thorough parametric studies of factors effecting top-down cracking including pavement thickness, binder grade, asphalt concrete composition, sub grade stiffness, segregation and age hardening also involve thousands of individual analyses. Therefore, three-dimensional finite element methods cannot at this time be used to perform extensive parametric studies of top-down cracking and similar phenomena. Because three-dimensional finite element methods are not yet practical for performing the thousands of analyses needed for a thorough parametric study of the factors that potentially contribute to top-down cracking in asphalt concrete pavements, layered elastic analysis must be used for this purpose in Phase II of NCHRP Project Pavement engineers and researchers should however continue to develop and refine three-dimensional finite element methods for pavement design and analysis, and associated engineering tools for using the results of these analyses, including continuum damage theory and computational fracture mechanics. These approaches to pavement design and analysis will probably become practical in about 10 years. Layered elastic analysis, properly executed, represents an exact closed form solution involving the rigorous application of mechanics, and is not an approximate method. Some specific implementations of layered elastic theory are simplified, and as a result cannot accurately deal with horizontal surface stresses in flexible pavements; such programs are not suitable for use in evaluating top-down cracking. A thorough comparison of the results of layered elastic analyses and finite element methods has however not been performed and reported on in the literature. Such a study is included in the revised Phase II Work Plan, and should clarify the performance of these methods of pavement analysis. Regardless of the method used for stress-strain analysis of flexible pavements, it is essential that the analysis use accurate models for predicting pavement temperature as a function of time and depth, asphalt concrete modulus, strength and/or fracture toughness, the accumulation of fatigue damage, and age hardening. Accurate approaches for estimating these properties have been identified and will be used in Phase II of NCHRP Project Because the stresses at the surface of a flexible pavement subject to traffic loading are complex, involving both normal and shear stresses, evaluation of surface stresses must involve application of an appropriate failure theory. The most effective such theory appears to be octahedral shear stress theory, because it is suitable for ductile materials, is mathematically direct and has been applied to asphalt concrete pavements in the past. Accurate continuum-damage fatigue equations have been developed and validated by Christensen and Bonaquist, and appear the most effective approach for routine pavement design and analysis, including application to top-down cracking. For cases viii

10 when laboratory testing is needed, uniaxial fatigue testing, analyzed using continuum damage principals, is probably the most efficient and effective procedure for characterizing the fatigue resistance of asphalt concrete mixtures. Ideally, continuum damage theory should be implemented in conjunction with threedimensional finite element methods to provide an accurate estimate of damage caused by traffic loading and thermal stress; however, as discussed above, three-dimensional finite element methods are not yet practical for routine use; until such methods are available, continuum damage approaches can probably be used with layered elastic analysis to provide reasonable estimates of surface damage and top-down cracking. Christensen and Bonaquist have developed a simple, direct approach to applying continuum damage theory to pavement analysis which can be easily implemented and appears very effective in modeling both uniaxial and flexural laboratory fatigue tests, and in modeling fatigue in the WesTrack project. Healing is potentially an important factor in top-down cracking, but no simple and effective models exist for predicting the effects of healing in flexible pavements. It does appear that healing rates can be related to asphalt binder viscosity, a relationship that might be useful in developing practical models for predicting healing rates in asphalt concrete mixtures. Ideally, methods for modeling top-down cracking should probably address plastic deformation, in addition to viscoelastic behavior and fatigue damage. There is evidence that significant plastic deformations can occur in asphalt concrete mixtures, even at low temperatures. As with other types of pavement modeling requiring threedimensional finite element analysis, visco-elastic-plastic models must await further improvements in computer hardware and software before they become a practical tool, or even before they can be thoroughly evaluated and refined. Fracture mechanics is probably useful in modeling some aspects of top-down cracking, especially the relative resistance of different mixtures to crack propagation, as characterized through fracture toughness. However, computational fracture mechanics may not be essential in developing practical approaches to modeling the initiation and propagation of top-down cracking in asphalt concrete pavements. Distortional stresses and continuum damage might be adequate for practical modeling of this form of distress, and is much simpler to implement than computational fracture mechanics, which should be regarded for the time being as a research tool and not a procedure for routine use by pavement engineers. Thermal stresses probably contribute significantly to top-down cracking. The thermoviscoelastic analysis developed by Roque et al. is the most widely used and probably the most effective approach to estimating thermally induced stresses in flexible pavements. Stresses estimated in this way can be combined with stresses resulting from traffic loading, through the use of octahedral shear stress, to evaluate distortional stresses and the resulting damage that are the primary cause of top-down cracking. ix

11 CHAPTER 1. INTRODUCTION RESEARCH PROBLEM The research problem being attacked in NCHRP Project 1-42 is summarized by NCHRP as follows: Until Recently, load-associated fatigue cracking of hot-mix asphalt (HMA) concrete-surfaced pavement that occur in the wheel path have been thought to always initiate at the bottom of the HMA layer and propagate to the surface... However, recent studies have determined that load-related HMA fatigue cracks can also be initiated at the surface of the pavement and propagate downward through the HMA layer. The penetration of water and other foreign debris into these cracks can further accelerate the propagation of the crack through the HMA surface layer. These studies indicated that environmental conditions, tire-pavement interaction, mixture characteristics, pavement structure, and construction practices are among the factors that influence the occurrence of this cracking. Hypotheses regarding the top-down cracking mechanisms have been suggested; test methods for evaluating HMA mixture susceptibility to cracking have been proposed; and preliminary models for predicting crack initiation and propagation have been developed. However, only limited research has been performed to evaluate and validate these hypotheses, test methods, and models. Research is needed to address the issues associated with top-down fatigue cracking and to develop guidance for pavement engineers in selecting HMA mixtures and designing flexible pavements. From Project 1-42, FY 2003: Top-Down Fatigue Cracking of Hot-Mix Asphalt Layers, at 1

12 RESEARCH OBJECTIVE There are four primary objectives to the research discussed in this Interim report. As described in the Research Project Statement (RPS) for NCHRP Project 1-42: The objectives of this research are to (1) identify the mechanisms that govern the initiation and propagation of top-down fatigue cracking in HMA layers, (2) identify or develop method(s) of laboratory testing of HMA mixtures for determining susceptibility of the HMA surface layer to this cracking, (3) determine the significant factors associated with the occurrence of top-down fatigue cracking, and (4) identify promising models for predicting the initiation and propagation of top-down cracking. From Research Project Statement for NCHRP Project SCOPE The work involved in completing NCHRP Project 1-42 has been broken down into seven tasks: Task 1: Review Literature Task 2: Recommend Conceptual Model Task 3: Refine Phase II Work Plan Task 4: Prepare Interim Report Task 5: Perform Testing and Analysis Task 6: Develop AASHTO Model Standards Task 7: Develop Validation Plans Task 8: Prepare Final Report At this point, Tasks 1 through 4 have been completed. An initial Interim Report was submitted several months ago and reviewed by the NCHRP Project 1-42 Panel. A panel meeting was held in April 2004, in which the results summarized in the Interim Report were presented and discussed. Several shortcomings in the Interim Report and plans for Phase II of the project were identified by the panel, and are discussed later in this introduction. This Revised Interim Report has been modified in view of this meeting and panel comments received after submission of the initial Interim Report. This introduction is relatively long and detailed, because it is important that several fundamental and important issues 2

13 concerning the overall approach to this project be addressed early in this report, so that the reader has some understanding of these issues while reading the literature review. This Interim Report is an abbreviated version of the report as initiall submitted to the NCHRP, in that it does not contain a Revised Phase II Work Plan. It contains only two chanpters: (1) Introduction, and (2) Background. The technical background presented in Chapter 2 includes a review of recent research pertaining top-town cracking in asphalt concrete pavements, and an in-depth technical description of the analytical methods (and major alternatives) proposed for use in Phase II of the project. Details of specific models for estimating the modulus of asphalt binders and asphalt concrete, and for predicting the accumulation of damage during fatigue loading of asphalt concrete mixtures might seem tedious, but these are essential features of any model for predicting top-down cracking and other forms of pavement distress. Unfortunately, the importance of this aspect of pavement design and analysis are often minimized when researchers focus on developing and evaluating advanced methods of mechanical analysis. It should be common sense among practicing engineers that such techniques will be useless if the fundamental properties of the pavement are not properly modeled. Conversely, layered elastic analysis, although much maligned by pavement researchers, may prove to be much more effective than generally accepted if combined with accurate models for asphalt concrete modulus, accumulation of fatigue damage and age hardening. 1 In understanding the literature review and Revised Phase II Work Plan contained in this report, it is essential to acknowledge that the fourth project objective in the RPS is to identify promising models for predicting the initiation and propagation of top-down cracking. This objective does not require the development or refinement of models, or even the thorough evaluation of existing models only the identification of promising models. Furthermore, the objective uses the plural models rather than the singular. Therefore, it is not necessary to identify a single best model, but only one or more promising models. Acknowledging these facts is important, because the most promising models for evaluating top-down cracking are at this time in an early stage of development, and thus cannot be 3

14 thoroughly evaluated. Furthermore, such a thorough evaluation would not only require that these models be in a much greater state of refinement than is the case, it would also require a thorough field validation, which is entirely outside the scope of the NCHRP Project 1-42 Research Project Statement (RPS) and the contract documents. It should be clearly stated at this point that the ideal model for predicting top-down cracking should include the following features: Computation of stresses and strains in three dimensions using a finite element analysis with a properly defined element types and mesh fineness, with a simple and effective user interface incorporated automated pre- and post-processing True viscoelastic response (as oppose to the quasi-elastic approach often used in pavement analysis and design) Anisotropic material properties Plastic behavior Accumulation of microdamage, most likely through application of continuum damage theory, including an indication of when and where crack initiation is likely to occur Modeling of the initial stages of crack propagation (assuming that being able to model the precise behavior of a pavement with substantial, deep cracking is of no value to the practicing engineer) Although various researchers and engineers have implemented separately all of these features in a variety of studies, the commercial development of such a pavement design and analysis program is still years away. Even proponents of finite element methods (FEM) admit to its shortcomings. Merrill, in his dissertation (which recommended the use of FEM in studying surface cracking) stated Realistic three-dimensional models can require vast amounts of computation time and storage. In evaluating his three-dimensional FEM model, Merrill performed a total of 21 analyses using the FEM model, which allowed only limited variation of a few factors (1). In contrast, the more substantial analytical study proposed later in this revised Interim Report includes over 6,000 analyses, and even then does not include as many factors over as wide a range as is ideally needed by engineers studying pavement behavior. Recent research by Myers et al. relied heavily on FEM, but using for the most part a simplified two-dimensional approach, and not three-dimensional analyses, and only 4

15 involved several hundred analyses (2,3). Application of realistic non-linear material models faces similar obstacles. In a recent publication summarizing results of visco-plastic continuum damage modeling, Chehab and his co-authors presented the results of the characterization and modeling of a single asphalt concrete mixture (4). This approach could be considered ready for use by practicing engineers if most highway agencies in the United States would agree to use only this mixture in constructing and rehabilitating their pavements. The authors of this Interim Report believe that this is unlikely to occur. Computational fracture mechanics involves both three-dimensional FEM and fracture mechanics, with very stringent mesh requirements since accurate modeling of stresses and strains at a crack tip are required. Anderson, in his text on fracture mechanics, makes the following statements in his concluding remarks on computational fracture mechanics (Anderson, 1995): A numerical fracture simulation of a cracked body can compute crack tip parameters, but such an analysis alone cannot predict when fracture will occur Fracture can be modeled, but a separate failure criterion is required Computer simulation of processes such as microcrack nucleation, void growth, and interface fracture should lead to new insights into fracture and damage mechanisms. Such research may then lead to rational failure criteria that can be incorporated into global continuum models of cracked bodies. T.L. Anderson in Fracture Mechanics: Fundamentals and Applications In other words, computational fracture mechanics is of limited use in modeling realistic failure processes, but probably can provide useful information for refining more conventional engineering design methods involving continuum mechanics and strength of materials. This is a research tool and not a design method for use by practicing engineers. These observations concerning the difficulties of three-dimensional finite element analyses, viscoplastic continuum damage modeling, and computational fracture mechanics are not meant to discourage their use and further development by pavement researchers. Indeed, as stated above, a mechanistic approach incorporating most or all of these features 5

16 should prove most accurate in predicting the occurrence and extent of top-down cracking, and will no doubt at some time be incorporated into a nearly completely mechanistic pavement design and analysis approach. However, at this time, such an approach is simply not realistic as a standard method for pavement design and analysis. In fact, because these methods are currently so difficult and time consuming to apply, they cannot be used to evaluate a wide range of realistic pavement systems and make predictions that can be compared with observations. Therefore, these approaches cannot even be effectively compared or evaluated, beyond the simple conclusion that they are theoretically sound features of an ideal model, that must await further development of computer hardware, software and engineering models prior to further evaluation and implementation. In the meantime, pavement researchers should be strongly encouraged to further develop and refine various approaches to three-dimensional FEM, viscoplastic continuum models, continuum damage theory, and computational fracture mechanics. Although there are significant limitations to these advanced models, Objective 4 of Project 1-42 identification of promising models for predicting the initiation and propagation of top-down cracking can still be achieved. In fact, in the statements above, it has already been achieved. What cannot be accomplished is the refinement of these models, or a realistic evaluation and comparison of what they contribute to the effectiveness of a fully mechanistic flexible pavement design and analysis program addressing top-down cracking. Fortunately, this is not included as an objective in the Project 1-42 RPS, and was not contemplated in the proposal submitted by AAT. However, Objectives 1 and 3 of the NCHRP Project 1-42 RPS involve identification of mechanisms governing top-down cracking, and determination of significant factors associated with the occurrence of top-down cracking in asphalt concrete pavements. Achieving these objectives requires analysis of pavement behavior under a wide range of conditions, including different average temperatures, different temperature gradients, varying subgrade stiffness values, a range of asphalt concrete compositions, etc. As mentioned above, the analytical experiment included in this Revised Interim Report addresses these issues, and includes over 6,000 separate analyses. The only approach that can be used to perform such a large number of analyses is layered elastic analysis (LEA). There are of course shortcomings to LEA, although a careful 6

17 reading of the pertinent literature shows that for well-designed LEA programs, these concerns are exaggerated. The primary shortcoming appears to be that LEA programs often do not accurately predict stresses and strains in the top 1 mm (approximately) of the pavement surface, especially directly under the tire edge. Furthermore, this shortcoming is apparently not a function of the limitations of LEA per se, but instead is the result of characterizing complex surface loads through a series of discrete circular loads, rather than through a continuous function of horizontal and vertical loads (1). It should be concluded that LEA, properly executed, can be effectively used to study many of the mechanisms and factors affecting top-down cracking, and in any case, is the only available approach that can be used to analyze a realistically broad range of conditions. However, a thorough and careful comparison of LEA and FEM has not yet been conducted the available comparisons have focused on the very surface of the pavement and have been performed over a limited range of conditions. Readers should keep in mind that this is an interim report and not a final project report. This is a working document, the purpose of which is to provide clear direction for successful completion of the project, and not to present the final results of the project in a detailed and thoroughly edited format suitable for publication. Although several were weeks were spent addressing as many of the panel members comments as was practical, as with any document, this Interim Report could be improved further by adding more content, performing further revisions and a thorough re-editing. The most important issue to consider at this time, however, is not whether or not this report is perfect (it is not), or whether it supports the ideas and approaches favored by each panel member (it does not and cannot). This report has been compiled to present the results of a limited review conducted with limited funds and time, and to present an effective plan for completing the balance of NCHRP Project 1-42 and achieving the project objectives, within constraints resulting from the current state of technology in pavement design and analysis, and the limited time and budget allowed for project completion. 7

18 CHAPTER 2. BACKGROUND REVIEW OF LITERATURE AND CURRENT PRACTICE The literature review below has been compiled to provide the necessary technical background to develop an effective plan for Phase II of NCHRP Project The initial part of the review consists of an overview of recent research on top-down cracking in bituminous pavements. This is followed by four sections addressing the four primary objectives of NCHRP Project 1-42: (1) Mechanisms Governing Top-Down Cracking; (2) Factors Affecting Top-Down Cracking; (3) Laboratory Tests for Evaluating Resistance to Top-Down Cracking; and (4) Modeling Top-Down Cracking in Asphalt Concrete Pavements. To help emphasize the most important aspects of the literature review, each of these sections is followed by a summary of findings. The section on modeling is particularly long and detailed, in order to document procedures proposed for use in Phase II of Project NHCRP Chapter 2 concludes with reiteration of the findings concerning top-down cracking in asphalt concrete pavements as presented throughout the literature review. Figure 1 is a photograph of a typical top-down crack, reproduced from a report by Harmelink and Aschenbrenner on top-down cracking in Colorado (5). Overview of Recent Research There have been several recent studies on top-down cracking, some of them more-or-less ongoing and encompassing a wide range of issues related to this problem. Most notably Roque and his associates (and former associates) at the University of Florida have published numerous papers and reports on various aspects of top-down cracking in asphalt concrete pavements (6,7,8,9,10,11,12,13,14,15). Their research emphasizes the complexity and importance of stresses generated at the tire-pavement interface and tends to be highly analytical, incorporating relatively advanced methods such as finite element analyses and fracture mechanics (7,8). Figure 2 is a plot of transverse stresses at the tire-pavement interface as estimated by Roque et al. (16). Like many researchers, Roque and associates 8

19 suggest that thermal stresses contribute significantly to the incidence of top-down cracking, though the recent thrust of their research emphasizes the importance of fracture toughness of asphalt concrete in resisting top down cracking (5). Myers et al. have used fracture mechanics to evaluate the potential effect of various factors on the rate of top-down crack propagation (11,2,3). They found that the driving force in top-down cracks (stress intensity or K 1 ) increases significantly as stiffness gradients increase (decreasing with depth) in asphalt pavements. Not surprisingly, they also found that the position of the wheel load relative to the crack and the crack depth also affected stress intensity in top-down cracks. Their research indicated that for Superpave mixtures, fine-graded aggregates will provide pavements better able to resist cracking compared to coarse-graded aggregate blends. These studies are discussed and evaluated in more detail later in this section. Figure 1. Top Down Cracking in Colorado (5). 9

20 Figure 2. Transverse Contact Stresses Predicted at Pavement Surface with Pavement Thickness = 10 cm, E 1 = 1,400 MPa, E 2 = 300 MPa (16). Svasdisant et al., and Schorsch and Balaldi (researchers at Michigan State University) have published a number of papers and reports on top-down cracking in Michigan (17,18). Their research has been more applied in nature compared to that of Roque s group. They have used layered elastic analysis and some finite element analysis and have not considered tire-pavement interfacial stresses (17). The Michigan State researchers have stressed the importance of segregation and moisture damage in exacerbating top-down cracking and have relied mostly upon empirical models to characterize this problem (18). In some ways, the efforts of the Michigan State group have been similar to those in Colorado, where a recently published report indicated that top-down cracking in that state is largely the result of segregation (5). The Colorado study included no pavement analysis or analysis of mixture materials properties, but focused almost entirely on segregation as caused by specific design flaws in most paving machines. 10

21 There has been a wide range of smaller, more theoretical studies dealing with issues related to top-down cracking. Stolarski and his associates at the University of Minnesota have performed relatively advanced analyses on the stresses developed at the pavement surface in both intact and cracked structures (19,20). The results are unfortunately difficult to interpret in practical ways meaningful to pavement engineers. Wang and his co-authors have performed an interesting analysis using micro mechanics, which produced somewhat unusual results in that it demonstrated the importance of shear stresses in top-down cracking, whereas most other studies have focused exclusively on tensile stresses (21). Top-down cracking has also been observed in Washington State and in the well-known MnROAD study (22,23,24,25). Research in Washington has been along the lines of a survey of the frequency and nature of top-down cracking, without developing new theories or analyses to better understand the problem. One important finding in Washington was that both segregation and moisture damage appear to contribute to top-down cracking (22). No research reports or papers have been published on the top-down cracking at MnROAD, though several brief presentations have been made. In this test road facility, the cracking appears to be partly related to pavement age hardening and partly related to traffic loading. Cracking is more severe for sections made using a harder binder (24,25). An interesting research project in Kenya demonstrated the importance of extreme age hardening at the pavement surface in the occurrence of top-down cracking. Such age hardening has been found in several other research projects and can cause very pronounced stiffness gradients in the pavements, which can increase the magnitude of stresses at the pavement surface (26). Mechanisms Governing Top-Down Cracking The literature reviewed to date provides significant insight into the nature of top-down cracking. It is not surprising that asphalt concrete pavements exhibit top-down cracking. The surfaces of flexible pavements are subject to heavy, moving traffic loads that create significant distortional (shear and tensile) stresses at and near the pavement surface. The surface is also subject to solar radiation, age hardening through oxidation, and moisture damage. What is surprising is that this mode of distress has not been acknowledged earlier. 11

22 This may be in part do the emphasis placed in this country on traditional pavement design methods that address fatigue damage exclusively through maximum tensile strains at the bottom of the bound layers in asphalt concrete pavements. It is also quite possible that recent changes in the binder grades and aggregate gradations have increased the frequency and severity of surface initiated cracking in asphalt concrete pavements. Because top-down cracking usually occurs in or near the wheel paths in a pavement, traffic-associated fatigue must play an important role in the process. A reasonable hypothesis is that repeated loading degrades the stiffness and/or strength of the HMAC within and adjacent to the wheel paths. Stresses and strains near the pavement surface are very irregular and are concentrated in localized areas because of the nature of tire-pavement stresses; this can focus and accelerate damage in certain areas of the wheel path. Furthermore, recent laboratory studies of the manner in which asphalt concrete modulus decreases under fatigue loads suggests that this degradation probably occurs very rapid during initial loading (27). This initial rapid fatigue damage is not apparent in bottom-up cracking simply because bottom-initiated fatigue damage isn t visible until it has progressed the entire way through the pavement. It should also be kept in mind that the temperature and temperature gradients at the underside of a flexible pavement are often much different than those at the top, and this might result in an increased potential for damage at the pavement surface even when distortional stresses are lower at this location. During construction of an asphalt concrete pavement, the top several millimeters of the wearing course cannot normally be thoroughly compacted, even under the best conditions and with the most careful workmanship. In many cases, conditions and workmanship are not optimal and so the surface of the pavement suffers even more. This results, in many cases, in a relatively weak, porous pavement surface, prone to age hardening and moisture damage. Segregation during construction, often along longitudinal lines defined by details of the paver design, creates weak spots in the pavement where age hardening and moisture damage can reach more quickly and deeply into the pavement. Another source of distress at the pavement surface is thermal fatigue repeated thermally-induced tensile stresses at the pavement surface that are not large enough to cause sudden thermal cracks, but which can gradually 12

23 damage the pavement and contribute to surface induced cracking. Thus, over a period of several years, traffic-associated fatigue accelerated by stress concentrations at the tirepavement interface, along with age hardening, moisture damage and thermal stresses gradually, but significantly weaken the material in and near the wheel paths of flexible pavements. Ultimately, top-down cracking must occur when the stresses at or near the pavement surface exceed the strength of the material. It is almost common sense that the same source of stresses causing degradation of the pavement surface can at some point cause critical stresses initiating cracks at the pavement surface. Surface cracks in asphalt concrete pavements must be initiated by stresses caused by traffic loading, temperature changes, or by some combination of these factors. In evaluating thermal cracking and traditional bottom-up cracking, failure is determined when maximum tensile stresses exceed the tensile strength of the asphalt concrete. However, the situation is more complicated where top-down cracking is being analyzed, since the stresses induced at the surface of the pavement involve both normal and shearing components. A failure theory suitable for complex states of stress must be applied in this case. The von Mises failure criteria, as implemented through octahedral shear stress, is such a failure theory and has been successfully applied to asphalt concrete (28,29). The octahedral shear stress theory can be thought of as a different but equivalent formulation of the distortion energy failure theory (28). This failure theory and the manner in which it is proposed for use in Phase II of NCHRP Project 1-42 is discussed in more detail later in this chapter. Although failure criteria such as the von Mises theory are useful in predicting crack initiation, they will not necessarily relate to crack propagation rates under fatigue loading. Recently Myers et al. have applied finite element analyses to calculate stress intensity factors for surface cracks in flexible pavements under various conditions (11,2,3). The stress intensity factor, K is a parameter that describes the driving force for a specific crack, and is a complex function of the crack geometry (position, length), the magnitude and geometry of the applied load, and the geometry of the loaded object. Although not a direct indicator of crack propagation rate, for a given material propagation rates should be directly related to K: 13

24 as stress intensity increases, rate of propagation should also increase. Myers and Roque provided rough estimates of crack propagation rates based upon these calculated stress intensity factors and the results of fatigue/crack growth tests performed the IDT geometry. These results have not been verified, and the rates seem very high, ranging from 9 to 390 mm/10,000 loading cycles (11). To further complicate matters, there are three modes of cracking: tensile opening (mode I), in-plane shear (mode II), and out-of-plane shear (mode III). Myers found that mode I stress intensity factors (K I values) were greater than mode II (K II ) values at a 625 mm distance from the wheel load for a variety of simple pavement structures, suggesting that tensile opening is the dominant fracture mode in surface cracking of flexible pavements. It should however be expected that K II values would be greatest directly under the edge of the wheel load (and still large in adjacent areas), where shearing stresses are the greatest. Additional efforts should be made to evaluate the magnitude of both K II and K III values near the edge of a wheel load to fully understand the relative importance of the three models of fracture in surface cracking. Furthermore, subsequent work should recognize that the driving force in a propagating crack under mixed mode fracture is a function of all three stress intensity factors not simply the largest. The initial work of Myers et al. indicated that the greatest stress intensity factors for short cracks exist under the widest tire rib near the pavement surface (2). This suggests that surface cracks are often initiated in this location. However, K I values nearly as large as these were found to exist for longer cracks at some distance from the applied wheel load. This initial work was however performed for uniform modulus values through the bound layer in the pavement. Later work include several modulus gradients, and showed much higher K I values for cracks mm in length at a distance of from 625 to 750 mm from the applied load (3). Stress intensity values for surface cracks in flexible pavements were found to be a complex function of crack length, load position and modulus gradients. Myers et al. make several important conclusions based on this research. Temperature and modulus gradients appear to be critical to the initiation and propagation of surface cracks. Furthermore, the driving force in surface cracks will vary depending upon the position of the applied load relative to the crack and the crack length, and these factors should be considered in pavement design and management. For example, a surface crack might initiate and grow relatively quickly to a 14

25 certain depth, and then stabilize for a time. Maintenance to prevent further growth at this point would be unnecessary; crack repair would be more efficiently delayed until the crack reached a length at which more rapid propagation is expected (3). Figure 3 is a plot taken from the initial research publication (2). Figure 3. Stress Intensity Factors as a Function of Crack Length at Various Distances from Applied Load (2). Myers et al. stated in their initial study that distortion energy theory (or the equivalent octahedral shear stress theory) could not be applied to surface cracking because it does not predict significant distortional stresses deeper than about 12.5 mm into the pavement. To evaluate this hypothesis, octahedral shear stresses were calculated for a pavement identical in structure to that used by Myers et al. in this study: 200-mm in thickness, with an HMAC modulus of 5,500 MPa and a subgrade modulus of 140 MPa; a tire loaded under 24 kn at 690 kpa pressure was assumed. Layered elastic analysis was used. The octahedral shear stress values were calculated at depths equivalent to the crack lengths used by Myers et al. 15

26 (6.2, 12.5, 19.1, 25.4, and 37.5 mm), and at distances from the tire center of 0, 0.8, 1, 1.2, 1.5, 2 and 4 times the tire radius of 10.6 cm. The results are shown graphically in Figure 4. Contrary to the hypothesis by Myers et al., the octahedral shear stress directly under the edge of the tire increases rapidly to a value of about 280 kpa and remains nearly constant down to the maximum depth evaluated of 37.5 mm. Furthermore, the magnitude of this stress is substantial; Anderson et al. reported an average IDT strength of 1,100 kpa (160 lb/in 2 ) for mixtures made with 12 widely differing binders at temperatures of 4.4 and 15.5 C. (30) This corresponds to an allowable octahedral shear stress of 520 kpa (75 lb/in 2 ), so the maximum octahedral shear stress, which extends undiminished through much of the pavement, is over 50 % of the allowable stress. It should also be noted that shear failure under the edge of the tire is entirely consistent with the observation that top-down cracking tends to occur near the edges of wheel paths. Octahedral Shear Stress, kpa r 1.2r 4.0r Depth from Surface, mm 0.8r 1.5r 0.0r 2.0r Figure 4. Octahedral Shear Stress in an Asphalt Concrete Pavement as a Function of Depth, at Different Radial Distances from the Applied Load (thickness = 200 mm, E 1 = 5,500 MPa, E 2 = 140 MPa, P = 24 kn, p = 690 kpa). The source of the disagreement between this analysis and the findings by Myers et al. is not clear. It may be the result of their assumption that surface cracks are initiated and 16