Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures (2022)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

6 Background Fatigue in Asphalt Pavements Traffic-associated fatigue damage is one of the major distresses in which flexible pavements fail. This type of distress is the result of many thousands—or even millions—of wheel loads passing over a pavement. Each load repetition induces stresses and strains that gradually damage the pavement and lead to failure. Figure 1 is a photograph of severe fatigue cracking in an asphalt concrete pavement at the National Center for Asphalt Technology (NCAT) test track. This type of cracking is often called “alligator” cracking because the pavement surface resembles an alligator’s hide. The alligator cracking shown in Figure 1 is typical of pavement failure resulting from bottom-up fatigue cracking, meaning fatigue cracking that originates at the bottom of the asphalt concrete layers and gradually works up toward the surface. Bottom-up fatigue cracking is easy to explain because tensile strains in a flexible pavement will almost always be largest at the bottom of the asphalt structure. Engineers have long focused on bottom-up fatigue cracking in their testing procedures, materials specifications, and pavements design and analysis methods. However, it has become clear over the past 10 to 15 years that fatigue cracking can also origi- nate at the surface and work down toward the center of the pavement. The source of the stresses and strains is more complex in this type of cracking, originating from tire-pavement inter- actions, shear stresses in the pavement surface near the outside edge of the tire, and tensile stresses in the pavement surface some distance from the tire. Thermal gradients within the pavement may also contribute to such fatigue cracking. Top-down fatigue cracking usually has a different appearance from bottom-up cracking, typically occurring as longitudinal cracks near or between the wheel paths. There is no reason to suspect that these two mechanisms for fatigue damage (three, if including thermally induced damage) are mutually exclusive. In fact, bottom-up fatigue, top-down fatigue, and thermally induced fatigue damage likely occur simultaneously in many asphalt pavements. A third type of fatigue cracking, reflective cracking, occurs in asphalt concrete overlays when a crack or joint in the underlying pavement gradually works its way up through the overlay from repeated load applications or movement of the underlying pavement. The underlying pavement can be Portland cement concrete or asphalt concrete. Figure 2 is a photograph of reflective cracks in an asphalt concrete overlay (Miller and Bellinger, 2003). Although reflective cracking is a complex mode of distress that is difficult to analyze mechanistically, measures to improve traffic-associated fatigue and thermal fatigue performance will also likely improve resistance to reflective cracking. Traffic-associated fatigue cracking of any type in flexible pavements is a complex phenomenon that depends on many factors. Perhaps most important is the severity of traffic loading—the C H A P T E R 1

Background 7   heavier the axle loads on a pavement and the higher the number of heavy vehicles, the more quickly fatigue cracking will occur, all else being equal. The pavement structure is also extremely important. For full-depth asphalt pavements, thicker pavements will exhibit lower stresses and strains under a given traffic load; thus, they will resist fatigue cracking signifi- cantly longer than thinner pavements. The subgrade underneath the pavement will also affect the occurrence of fatigue cracking; weak subgrades will result in higher stresses and strains Figure 1. Alligator cracking at the NCAT test track. Figure 2. Reflective cracking in asphalt concrete overlay (Miller and Bellinger, 2003).

8 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures in the overlying structure, which in turn will reduce the fatigue life of the pavement system. Pavement drainage and moisture damage are also related to the subgrade. When moisture is present in the subgrade and the pavement drainage system is not properly designed or con- structed, water can infiltrate the pavement and accelerate any existing fatigue damage both by weakening the subgrade and increasing the strains in the bound layers, and by causing moisture damage. This can be a severe problem resulting in rapid failure. Moisture damage and fatigue damage often occur together, and the relative contributions of the two failure mechanisms can be difficult to sort out. Another factor affecting the fatigue behavior of asphalt concrete is the modulus of the material. This complex relationship is the result of several related factors. In a flexible pavement system, as the modulus of the asphalt concrete increases, the stiffness of the pavement system will increase, and strain under traffic loading will decrease, which will tend to reduce fatigue damage. At the same time, the amount of strain an asphalt mixture can withstand without failure tends to decrease with increasing modulus, which will increase the fatigue damage caused under a given traffic load—although using polymer-modified binders or special mix designs such as stone matrix asphalt may tend to offset this trend. To further complicate matters, the relative importance of these two offsetting factors depends largely on the pavement structure. In stiff pavement structures—with thick asphalt concrete layers, a substantial base layer, and a good- quality subgrade—increasing the modulus of a mix will generally improve fatigue performance because this will lower strains in the bound material. In thinner pavements, especially with poor-quality subgrades, strains are largely independent of the surface layer stiffness; increasing the modulus of the asphalt mix in this situation will decrease fatigue performance because of decreased strain tolerance in the bound material. The modulus of asphalt concrete depends on several factors: temperature, loading rate, mixture composition, and the asphalt binder used in the mix. Temperature and loading rate greatly affect asphalt concrete modulus but are the result of climate and traffic conditions, which cannot be controlled through specifications or construction practice. Mix composition is measured and controlled through parameters such as air void content and voids in the mineral aggregate (VMA). Although the composition of asphalt concrete is controlled through tests and specifications during mix design, mix production, and pavement construction, the effect of composition on modulus is small. The most important factor affecting the modulus of asphalt mixture that can be controlled by engineers is the modulus of the asphalt binder used in the mix. Another factor affecting the fatigue performance of asphalt pavements, and one critical in addressing the objectives of NCHRP Project 09-59, “Relating Asphalt Binder Fatigue Properties to Asphalt Mixture Fatigue Performance,” is the inherent strain tolerance of the binder. The term inherent strain tolerance as used here means the overall strain tolerance of the binder, compared with average or typical strain tolerance for a wide range of binders. As mentioned previously, as binder modulus increases, binder failure strain decreases, and the relationship between modulus and failure strain tends to be similar but not identical for all asphalt binders. Some binders, however, because of their rheologic type or the presence of polymer modifier, exhibit unusually high failure strain over a wide range of modulus. Similarly, some binders exhibit low failure strains at any given modulus value. This is what is meant by inherent strain tolerance—not the absolute failure strain of a binder under specific testing conditions, but the failure strain relative to other asphalt binders over a wide range of conditions. Binders with good inherent strain tolerance will exhibit better fatigue performance than those exhibiting poor strain tolerance. As discussed later in this chapter, the NCHRP 09-59 research team focused on characterizing the inherent strain tolerance of a wide range of binders, using both mixture and binder tests.

Background 9   Previous Research Related to Asphalt Binder Fatigue Performance Asphalt Binders’ Effect on the Fatigue Performance of Asphalt Mixtures The recently developed AASHTO mechanistic-empirical design guide (ARA Inc., 2004) for flexible pavements uses the following equation for estimating cycles to failure for bottom-up traffic-associated fatigue cracking, calibrated for general use in the United States: = β ε        0.00432 1 1 (1)1 3.9492 1.281 N C E f f t HMA where Nf = number of cycles to failure, βf1 = variable that depends on the asphalt concrete thickness, C = variable that depends on the asphalt concrete composition, εt = maximum tensile strain at the bottom of the asphalt concrete layer, and EHMA = modulus of the asphalt concrete. This equation only includes binder properties indirectly—the mixture modulus EHMA depends strongly on the binder modulus. The strong relationship between binder modulus and mixture modulus is demonstrated by several equations for estimating mixture modulus, such as the Hirsch model and various versions of Witczak’s equation, which include binder modulus in some form as a primary predictor of mix modulus (Christensen and Anderson, 1992; Andrei et al., 1999). Equation 1 suggests that all else being equal, mixture fatigue life will increase with decreasing mixture and binder modulus. The current binder fatigue specification limits the binder loss modulus, |G*| sin δ, to a maximum of 5.0 kilopascal (kPa). This limit was partially based on the observation that for several test roads constructed in the 1950s and 1960s, fatigue cracking increased dramatically when the binder modulus at typical pavement temperatures exceeded this value (University of California, Berkeley, 1994). Thus, the current specification is largely based on the relationship between binder modulus and mixture fatigue. Equation 1 was based on a fatigue equation originally proposed by the Asphalt Institute (Monismith et al., 1972; Shook et al., 1982). An alternative approach to addressing traffic-associated fatigue in flexible pavements based on the Shell fatigue equations was considered but later abandoned. There are two such equations, one for stress-controlled fatigue and one for strain-controlled fatigue, as shown here (Bonnaure et al., 1980): [ ]( )= − + − ε− −0.17 0.0085 0.0454 0.112 (2)5 5 1.8N A PI PI V V Ef f b b t HMA where Nf = number of cycles to failure, Af = correction factor to account for the typical difference between fatigue life in the laboratory and in an actual pavement, PI = penetration index of the asphalt binder, Vb = binder content by volume for the mix, εt = maximum tensile strain at the bottom of the asphalt concrete layer, and EHMA = modulus of the asphalt concrete. Equation 2 indicates that asphalt binders affect mix fatigue in two ways: through their effect on mixture modulus and through the penetration index PI. The penetration index is no longer

10 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures widely used in the United States to characterize asphalt binders but was commonly used until the 1990s. The PI is considered to indicate the rheologic type of an asphalt binder, which is related to the shape of the modulus versus time function for the binder as well as its chemistry. Although PI is no longer commonly used to characterize asphalt binders, a more rational and related parameter—the R-value, also called the rheological index—can be easily calculated from data produced in the bending beam rheometer (BBR) test. The R-value in this case refers to a variable in the Christensen-Anderson model for asphalt binder modulus (Christensen and Anderson, 1992). Figure 3 shows the relationship between PI and R-value for the Strategic Highway Research Program (SHRP) core asphalts (University of California, Berkeley, 1994). Although R has not been widely used in developing fatigue models for asphalt concrete mixes, there should be a relationship between the binder R-value and asphalt mixture fatigue perfor- mance because of its close relationship to PI and the use of PI in the Shell fatigue equations. In addition to modulus and R-value, various binder fatigue and fracture properties have been related to mixture fatigue performance. A recently completed study at the FHWA Accelerated Loading Facility (ALF) did show some relationships between binder fracture properties and the fatigue performance of mixtures made with a wide variety of asphalt binders (Gibson et al., 2012). The experiment was designed so that most of the evaluated mixes used identical (or nearly identical) aggregates and volumetric designs, although this was not possible with one of the crumb rubber mixes. Table 1 summarizes the correlations between mix fatigue performance—as measured both in laboratory testing and in the observed fatigue performance for the ALF lanes—and a variety of binder tests. Many of the tests showed at least moderate correlations, including several tensile fracture tests and shear fatigue tests. This supports the intuitive idea that binder fracture and fatigue properties should correlate to mixture fatigue performance and might also serve as the basis for a binder fatigue specification. Many test methods listed in Table 1 are discussed in more detail later in the work plan for this project. Low-Temperature Cracking and Thermal Fatigue Damage Over approximately the past 15 years, an increase in premature failures of asphalt concrete pavements has been observed in the United States and Canada, especially in colder regions Figure 3. Relationship between penetration index and rheological index (R-value) for SHRP core asphalts (University of California, Berkeley, 1994).

Background 11   (Ahearn, 2015; Marks, 2015; Reinke et al., 2015). The rate and severity of these failures seem to be increasing and have been the topic of significant research in NCHRP Project 09-60, “Addressing Impacts of Changes in Asphalt Binder Formulation and Manufacture on Pavement Perfor- mance through Changes in Asphalt Binder Specifications.” Although not specifically mentioned in the request, this increase in premature failures should certainly be considered in conducting NCHRP 09-59 research. Low-temperature cracking—also called transverse cracking—occurs when an extreme cold weather event causes the temperature of a pavement to drop quickly—in Minnesota, for exam- ple, pavement temperatures can drop to −30°C or lower. This rapid decrease in temperature causes thermal stresses to build up in the pavement, which can eventually exceed the pavement’s tensile strength, leading to failure through cracks that typically occur every 6 to 12 meters and run transversely across the pavement. Such cracking may also occur from thermal fatigue, that is, from subcritical cooling events that are not severe enough to cause failure during a single event but can cause enough damage that failure can occur over time. Thermal fatigue is not as well understood as single-event thermal cracking, but the poten- tial for a thermal fatigue component to low-temperature cracking suggests that this type of distress should also be considered within NCHRP 09-59 if feasible. Because traffic-induced stresses should be expected to superimpose on thermally induced stresses, inter action between these failure modes should be expected. Again, this suggests that to maximize fatigue performance of asphalt pavements, some consideration must also be given to low-temperature cracking. The recent increase in premature failures largely linked to nonload-associated cracking suggests that the current binder specification is not adequately addressing binder resistance to low-temperature cracking. One potential problem is the physical hardening that occurs in the BBR test. Canadian research has clearly documented substantial errors in BBR grading caused by physical hardening (Kanabar, 2010; Marks, 2015). The errors often result in grading a binder to a lower temperature than it should be, significantly increasing the chances for premature failure. As a result, Ontario has implemented a BBR specification incorporating extended conditioning to address the errors associated with physical hardening during the BBR test (Kanabar, 2010). Other researchers have linked these premature failures to large differences in BBR grading for stiffness and m-value. The parameter ΔTc = Tc(S) – Tc(m), the critical temperature based on stiffness minus the critical temperature based on m-value, has been correlated to failure in Binder Test Method Type of Test Coefficient of Multiple Correlation (r), with Uniaxial Fatigue Tests Fatigue Performance in ALF Test Lanes Double-edge notched tension (DENT), crack tip opening displacement Tension/fracture 0.95 0.98 Binder yield energy Shear fracture 0.87 0.80 Time-sweep test Shear fatigue 0.79 0.88 Direct tension failure strain Tension/fracture 0.83 0.85 |G*| sin δ (loss modulus) Modulus −0.73 −0.66 Large strain time-sweep surrogate Shear fatigue −0.74 −0.67 DENT elastic work of fracture Tension/fracture 0.43 0.50 Bending beam rheometer m-value Flexural creep 0.52 0.38 Stress sweep Shear fatigue −0.79 −0.73 Table 1. Correlation of binder tests with mix fatigue performance for lanes with 100-mm asphalt concrete pavement thickness from FHWA ALF study (Gibson et al., 2012).

12 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures numerous studies. Specifically, when ΔTc becomes too negative, a failure through nonload- associated cracking is more likely to occur. Because ΔTc is directly related to the Christensen- Anderson R-value, this suggests that the rheologic type of the binder—that is, the overall flow characteristics or shape of the modulus master curve—has a significant effect on thermal cracking independent of low-temperature stiffness. As discussed later in this report, these phenomena both appear to be significant problems potentially increasing the likelihood of low-temperature cracking in asphalt pavements, but fortunately both can be addressed with simple changes to current binder specification. Another parameter recently correlated to nonload-associated cracking is the Glover-Rowe parameter, or GRP (Glover et al., 2005; Anderson et al., 2011; Rowe, 2011). This parameter was originally developed as a surrogate for ductility in evaluating the durability of asphalt binders (Glover et al., 2005). It was then slightly modified by Rowe to make its application more straight- forward. GRP has been correlated to the surface cracking potential of pavements in several studies (Glover et al., 2005; Anderson et al., 2011). Its correlation to ductility also ties it to many old studies linking pavement durability to ductility and the relationship between ductility and penetration. In fact, it will be shown later in this report that the GRP relates well to the results of the DENT test and, perhaps more importantly, to binder fatigue strain capacity (FSC). The GRP apparently addresses the effects of both modulus and rheologic type on binder strain capacity. That it should correlate to nonload-associated cracking in pavements is therefore not surprising. Recent research on nonload-associated cracking suggests that rheologic type, as indicated most commonly by ΔTc, and binder FSC, as indicated by the GRP, both have a significant effect on nonload-associated cracking. Because these parameters (or other closely related parameters) can be measured relatively easily using existing equipment and test methods, these are promis- ing approaches to improving binder specification to reduce nonload-associated cracking. Binder Rheology, Adhesion, and Healing Some recently observed premature pavement failures have been characterized by severe raveling and associated surface distress (Ahearn, 2015). This would suggest a significant loss of adhesion between the asphalt binder and the aggregate. Standard laboratory moisture resistance tests did not indicate that these mixes were unusually susceptible to moisture damage, which in turn suggests that the cause is mechanical. One possible explanation for this distress is that the asphalt binder, because of physical changes from aging, loses significant adhesiveness at intermediate temperature. Rheologically, tackiness and adhesion are related to the phase angle—as the phase angle decreases, the tackiness and adhesion of an asphalt binder (or any similar material) will in general decrease. For instance, an elastic solid with a phase angle near zero will have essentially no tackiness or adhesion. An example of this relationship, for a polymer gel, is shown in Figure 4 (Grillet et al., 2012). Similar findings were made by another group of researchers studying hot melt adhesives (Tse, 1995). It is also well established in polymer science that adhesive tack is directly related to the storage modulus G′(ω). In fact, a rule called the Dahlquist criterion states that adhesive tack for pressure-sensitive polymers will be essentially lost when the storage modulus is above 100 kPa (Dahlquist, 1959). Pressure-sensitive adhesives are simply materials that bond to other materials primarily through the application of pressure and as a result of their tackiness. It seems possible that the Dahlquist criterion could apply to asphalt concrete mixtures. As discussed later in this report, the hypothesis that mixture healing can be related to binder rheology was difficult to evaluate but appears to be true, although the relationship is perhaps not as strong as hoped and is somewhat complicated. Mixture healing appears to be related more

Background 13   strongly to the binder phase angle than to the storage modulus, with healing increasing with increasing phase angles above about 35 degrees. At lower phase angles, there appears to be little or no healing. Because the phase angle at any given modulus value will be lower for binders with higher R-values, this leads to the important finding that overall healing potential will decrease with increasing R-values. Laboratory Aging of Asphalt Binders and Mixtures Many researchers have concluded that in characterizing asphalt binder and mixture prop- erties when studying recent premature failures, significantly more age hardening is needed than standard laboratory aging procedures provide. Much of the recent research presented at FHWA Expert Task Group (ETG) meetings over the past several years has used extended aging of binders in a pressure-aging vessel (PAV), using 40 hours or more instead of the standard 20 hours (Bennert, 2015; Reinke et al., 2015). The 40-hour PAV protocol is probably the most widely used method of extended binder aging and can be completed in a reasonable amount of time. Therefore, the research team decided to use 40-hour PAV aging, after aging with the rolling thin film oven test (RTFOT), for most of the binder tests performed as part of NCHRP 09-59. An important question for NCHRP 09-59 was what sort of mixture con- ditioning would provide age hardening similar to the RTFOT/40-hour PAV. After reviewing several studies, including preliminary data from NCHRP Project 09-54, “Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction,” the NCHRP 09-59 research team selected loose-mix aging, using a temperature of 95°C for 5 days (120 hours). As presented later in this report, this mixture-aging protocol seems to closely match the RTFOT/40-hour PAV binder aging. As results from NCHRP 09-59 and related research are implemented, it is essential to consider the ramifications of field aging and laboratory aging on any binder test specification. Any specification limits established directly using data from RTFOT/40-hour PAV-aged binders can only be applied to binders aged using this same protocol. Using different laboratory aging protocols will mean that any suggested specification limits will have to be adjusted. In the same way, other engineers and researchers evaluating the results of NCHRP 09-59 should carefully consider any differences in binder and mixture-aging protocols while comparing data from different sources. Figure 4. Relationship between adhesion and loss tangent for a polymer gel (Grillet et al., 2012).

14 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures Problems with the Existing Binder Fatigue Specification Test Before the results of NCHRP 09-59 can be discussed, perceived shortcomings in the current approach to ensuring adequate fatigue performance must be understood. The existing specifi- cation was developed during SHRP and was largely based on the observation that a dramatic increase in fatigue cracking was observed when the estimated loss modulus of the binder exceeded a value of about 3 Megapascal (MPa) for the Zaca-Wigmore test road, as shown in Figure 5 (University of California, Berkeley, 1994). This value was later raised to 5.0 MPa during implementation of the binder specification. Several researchers have pointed out shortcomings in the Superpave® binder specification (Deacon et al., 1997; Bahia et al., 2001; Gibson et al., 2012). The asphalt binder specification parameter, |G*| sin δ, does not seem to correlate well to fatigue performance in the field (Deacon et al., 1997). Part of the underlying problem is likely because the test is largely empirical. It was based on a correlation between observed fatigue cracking and modulus values estimated from penetration values for a single test road. This problem is especially relevant with respect to polymer-modified binders, which can show significantly enhanced fatigue and fracture properties compared with a non-modified binder. Several other tests appear more strongly related to mixture fatigue performance (refer to Table 1 from the 2012 FHWA ALF study). There is another potential source of problems for the current binder fatigue specification. The specification might be valid for thinner pavements for which strains are affected more by the stiffness of the underlying structure than by the modulus of the pavement. However, the specification may be inaccurate when applied to thicker pavements, for which strains can be significantly reduced by an increase in the pavement modulus. In evaluating the current binder fatigue specification it is essential to understand two aspects that often seem to be misunderstood, or simply ignored: (1) as discussed previously, the specification is meant to address the field performance of actual pavement systems and not the performance of mixes as tested in the laboratory; and (2) the specification was originally based on the loss modulus (G″ = |G*| sin δ) at some intermediate temperature characteristic of a particular climate, not on an intermediate temperature based on binder grades. In other words, the temperature at which G″ is evaluated should be based on the local climate (or the base binder grade used in a given locale), not on binder grades “bumped” for extreme traffic levels. The latter approach can lead to maximum G″ values that are excessively high for a given climate and violate the original intent of the specification. Because of these factors, it is difficult to evaluate the current binder fatigue specification in laboratory tests. Figure 5. Plot of G” = |G*| sin c at 25çC and 10 rad/s for Zaca-Wigmore test road, estimated from penetration data (University of California, Berkeley, 1994).

Background 15   An additional potential problem in the current binder specification was discussed previously in connection with low-temperature cracking—physical hardening can cause significant errors in BBR test data. Such errors will in general result in a binder grade lower than it should be, increasing the potential for transverse cracking. Ontario has already implemented a BBR speci- fication using extended aging to address this problem (Marks, 2015). Other, simpler approaches to addressing physical hardening may be possible, as discussed later in this report. Review and Selection of Binder Fatigue Tests for In-Depth Evaluation in NCHRP 09-59 Appendix A reviews and evaluates the binder fatigue tests the research team identified at the start of NCHRP 09-59. This appendix includes a literature review of potential tests and a detailed numerical ranking of the most promising. The following section summarizes the information given in this appendix. Through the literature review, six primary binder tests and parameters were identified as candidates for further evaluation: 1. The linear amplitude sweep (LAS) 2. The double-edge notched tension (DENT) test 3. The GRP and other similar rheological parameters that can be calculated from dynamic shear rheometer (DSR) data and have been related to performance, such as the R-value 4. The direct tension (DT) test 5. The single-edge notched bending (SENB) test 6. The ductility test The ductility test was included in the ratings because of its historical significance, because it is still widely used in other countries, and because substantial data in the literature link ductility to asphalt concrete pavement durability and fatigue performance. This test was not expected to be selected for further evaluation. Five criteria were selected for evaluating the tests: 1. Additional cost to run the test, primarily purchase of equipment that a commercial testing lab is unlikely to possess 2. Active time requirement, that is, the actual working time required to run the test 3. Correlation with performance based on published research 4. Engineering soundness, that is, a subjective evaluation of how strongly the test is based on engineering principles, as opposed to being empirical 5. Technical difficulty and ease of implementation, also a subjective evaluation of how difficult the test is to perform and the hurdles that may be encountered in implementing the test These criteria are explained in detail in Appendix A. The final criteria and weights used in this ranking were developed in consultation with the NCHRP 09-59 panel. Table 2 summarizes Criterion Weight LAS DENT GRP DT SENB Ductility Equipment cost (additional/new) 10 5 3 5 2 4 2 Active time requirement 20 3 1 3 2 3 2 Correlation with performance 50 5 5 5 2 1 4 Engineering soundness 5 3 3 5 5 5 1 Technical difficulty/ease of implementation 15 5 1 5 1 3 1 Total 100 4.5 3.3 4.6 2.0 2.2 2.8 Table 2. Ratings of candidate binder fatigue tests.

16 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures the resulting ranking of the candidate tests based on these criteria and their weights. It should be emphasized that the rating in Table 2 was not meant to be the only means for selecting the final tests, including for evaluation in the laboratory testing phase of NCHRP 09-59. Instead, this table was meant to provide guidance to the researchers and panel in making the final selection. The highest-rated tests were the LAS, the GRP and related rheological parameters, and the DENT test. Although the ductility test scored relatively highly in this rating, the research team feels that there would be little support and significant opposition to reconsideration of this procedure. As mentioned, it has been included because of historical reasons and because it is still widely used in other parts of the world. The two remaining tests, DT and SENB, are similar in that they both provide information on the strain capacity of asphalt binders at low temperature, yet they use different geometries. Selection of Final Asphalt Binder Tests The NCHRP 09-59 research team has emphasized from the beginning of the project, even in writing the initial proposal, that the funding and time allotted for the project was not adequate for the development, refinement, and validation of a new test procedure for evaluating the fatigue performance of asphalt binders. Therefore, the approach used in selecting tests for laboratory evaluation emphasized identifying and evaluating tests that had already been developed and could be implemented quickly and easily. Potential tests also had to have shown promise by exhibiting high correlation to mixture fatigue performance, in either the laboratory or the field, preferably both. On the basis of these considerations and the ratings summarized in Table 2, the following tests were selected for detailed evaluation in Phase II of NCHRP 09-59: • The LAS test, • The simplified DENT test, and • Various rheological parameters, including the GRP, loss modulus, storage modulus, and phase angle. These tests largely meet the criteria the research team developed for candidate binder tests to be included in Phase II of NCHRP 09-59: they have gone through initial development, they have been correlated to field performance, and they can be realistically implemented as specification tests. Details concerning these test procedures are presented later in this report. Characterization of Mixture Fatigue Performance in the Laboratory Probably the most important aspect of NCHRP 09-59 was relating selected binder test prop- erties to mixture fatigue performance as measured in the laboratory. For this reason, selection of laboratory tests for evaluating mixture fatigue performance was a critical activity. Three test methods were considered for use in NCHRP 09-59: bending beam fatigue, uniaxial fatigue testing, and the overlay test. Appendix B reviews these methods. The most important criterion for laboratory fatigue tests in NCHRP 09-59 was having a good record of correlation to field fatigue performance. Also important was evidence of correlation to binder properties. Unfortu- nately, no currently used laboratory fatigue test has exhibited consistently good correlation with either field performance or binder properties. As discussed later in this report, this is perhaps not because of shortcomings in the test methods, but because of problems in traditional methods of analyzing asphalt mixture fatigue data. After evaluating the three candidate mixture tests and in consultation with the project panel, the NCHRP 09-59 research team initially decided to rely primarily on uniaxial testing but to also use bending beam flexural testing and overlay testing in characterizing the fatigue resistance

Background 17   of mixtures. However, testing plans were modified early in the project and additional overly testing was abandoned. All 16 of the NCHRP 09-59 mixtures were tested using uniaxial fatigue at two or three temperatures, while 9 of 16 of these mixtures were tested using bending beam flexural tests at 10°C and 20°C. Details of the mixture tests are discussed later in this report. Objectives of NCHRP 09-59 The objectives of NCHRP 09-59 are given in the request for proposal: 1. Determine asphalt binder properties that are significant indicators of the fatigue performance of asphalt mixtures. 2. Identify or develop a practical, implementable binder test (or tests) to measure properties that are significant indicators of mixture fatigue performance for use in a performance-related binder purchase specification such as AASHTO M 320 and M 332. 3. Propose necessary changes to existing AASHTO specifications to incorporate the identified binder properties and their specification limits. 4. Validate the binder fatigue properties, test(s), and changes to existing and/or proposed AASHTO test methods and specifications with data from field projects, accelerated loading facilities, or both, supplemented, as necessary, with data from additional laboratory-prepared specimens. This research shall emphasize the potential use of binder test equipment used currently in AASHTO binder test methods and consider a range of asphalt binder types, unmodified and modified, appropriately conditioned (i.e., aged), and with fatigue performance expected to vary widely. Field performance data for validation shall be drawn from existing sources such as accelerated loading facilities, LTPP Specific Pavement Studies (including SPS-10), Asphalt Research Consortium (ARC) field sections, and NCHRP Projects 09-47A, 09-49, 09-49A, 09-52, 09-53, 09-54, 09-55, and 09-58. Scope of NCHRP 09-59 NCHRP 09-59, like other NCHRP projects, had a limited time frame and budget. Therefore, the testing of binders and mixtures had to be limited to what could be reasonably accomplished given these constraints. Not all aspects of this problem could be addressed—not every potential binder test evaluated, not every specific binder type tested. The research team assumed that because of the limited resources, the best chance for success was to focus on binder fatigue and fracture tests that had already gone through most of the development process and had shown correlation to field performance. Ideally, candidate binder tests would already be widely used and adopted by one or more highway agencies in a specification. Three binder tests were selected for in-depth evaluation during NCHRP 09-59: the LAS test, a simplified version of the DENT test, and an array of rheological parameters such as the GRP. The research team similarly felt that the laboratory mixture fatigue tests used in NCHRP 09-59 should be those widely used by pavement engineers, with a reasonable history of correlation to field performance, although realistically, as mentioned, it has historically been difficult to link mixture fatigue tests to field performance. As discussed in more detail later in this report, the main mixture tests used in NCHRP 09-59 were uniaxial fatigue and bending beam (flexural) fatigue. Sixteen binders were selected for inclusion in the NCHRP 09-59 primary test program. Binders were selected in close consultation with the panel to represent a wide range of binder types and potential fatigue performance. To expand the NCHRP 09-59 data set beyond these 16 binders, data from previous research projects and asphalt mixture fatigue were used. This includes the fatigue data collected during the SHRP (University of California, Berkeley, 1994) and the first two FHWA ALF fatigue experiments (Stuart et al., 2002; Gibson et al., 2012). Binders from these projects were collected and tested as part of NCHRP 09-59, and the results compared with the observed fatigue performance of the mixtures.

18 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures The primary experiment in NCHRP 09-59 involved performing mixture fatigue tests—both uniaxial and flexural—and relating these results to the results of the selected binder tests to determine which showed the best correlation. This comparison was made by calculating the fatigue/fracture performance ratio (FFPR) value for each binder from the two mixture tests, then comparing these with binder test data, which were in some cases also FFPR values and in some cases other binder test parameters. Binder test data and field performance were also compared to validate project findings. The research team believes that the chosen approach was successful, has led to a better understanding of asphalt mixture fatigue, and will provide a good basis for a few simple modifications in current binder specifications that will significantly improve mixture fatigue performance. Two NCHRP 09-59 products are (1) proposed modifications to existing specifications and (2) a plan for implementing these specification changes. An asphalt mixture healing experi- ment was also performed as part of NCHRP 09-59. Because of budget and time constraints, this experiment was less extensive than those involving mixture fatigue, but it did provide useful information on the relationship between binder properties and mixture healing. Early discussions between the NCHRP 09-59 research team and the panel addressed the elimination of reclaimed asphalt pavement (RAP) and reclaimed asphalt shingles (RAS) from the scope of the project. However, Dr. Walaa Mogawer and Dr. Ahmed Soliman of the University of Massachusetts Dartmouth offered to research the relationship between the binder parameters evaluated in NCHRP 09-59 and the mixture semicircular bend flexibility index. This work was conducted without support from NCHRP 09-59, other than the limited time Dr. Christensen required to coordinate efforts with NCHRP 09-59 and to incorporate an appendix into the NCHRP 09-59 final report. This information should help engineers use the semicircular bend and other mixture tests to evaluate the fatigue performance of mixtures containing RAP and RAS in a way consistent with the findings, conclusions, and proposals of NCHRP 09-59. Organization of This Report Following this background chapter is Chapter 2, which describes the research approach, explains the concepts in more detail, and provides more information on the binders, aggregates, and mixtures NCHRP 09-59 used. Methods of analysis are also described in Chapter 2. Chapter 3 describes findings and applications, presenting the results of testing and analysis performed during the project. It also discusses the resulting findings and how these can be applied to the problem of developing an improved binder fatigue specification. The final chapter, Chapter 4, describes the most important conclusions from NCHRP 09-59 and offers suggestions for future research. This report includes five appendices, in which various details on information collected during NCHRP 09-59 can be found. In most cases, the body of this report only summarizes information presented more thoroughly in the appendices.