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.

1   Asphalt mixture fatigue response is a complex function of various parameters. As modulus increases, strain capacity decreases, which decreases fatigue life. At the same time, increasing modulus will result in a decreasing phase angle, which will increase the fatigue exponent, increasing fatigue life. Healing is probably a significant factor in asphalt mixture fatigue performance, with increasing phase angles correlating to better healing and improved fatigue life for binders that are not polymer modified; for polymer-modified binders, the relationship between rheology and healing is unclear. To further complicate matters, in real pavement systems softer binders and mixes will generally exhibit higher failure strains but will tend to result in higher strains in situ, at least partly offsetting the beneficial effects of lower modulus values. The failure strain of asphalt binders tends to follow a well-defined failure envelope with respect to modulus; NCHRP Project 09-59, “Relating Asphalt Binder Fatigue Properties to Asphalt Mixture Fatigue Performance,” used a power law model to define this relationship. However, this standard failure envelope varies significantly among binders. A fundamental question addressed in NCHRP 09-59 is which test or tests are good indicators of this inherent strain tolerance. Laboratory testing conducted during NCHRP 09-59 suggests that binder R-value, or the rheological index, predicts overall fatigue strain capacity well. At a given modulus level, binders with lower R-values will exhibit higher failure strains than binders with higher R-values, such as those that have been heavily oxidized. Although the simplified double-edge notched tension (SDENT) test is also a reasonable indicator of inherent fatigue strain capacity, the correlation is not as good as for R-value. Considering the difficulty and cost of implementing a specification that includes the SDENT test, this procedure cannot be recommended based on research conducted as part of NCHRP 09-59. NCHRP 09-59 used heavily aged binders and mixtures in its test program. There is a concern that the extended loose-mix aging used in NCHRP 09-59 might have resulted in an unrealistic degradation of the performance of some or all of the polymer- modified binders studied. If the materials had not been as thoroughly aged, the findings concerning polymer-modified binders might therefore have been different. This is an important topic for further research. Although R-value is a good indicator of overall strain tolerance, modulus has an even stronger effect on failure strain and fatigue life. The Glover-Rowe parameter (GRP) appears to relate well to binder failure strain, accounting for the effects of both modulus and R-value on failure strain. The current binder fatigue parameter, |G*| sin δ, does not relate as well to binder failure strain, and in fact can allow binders with poorer strain tolerance to be used in S U M M A R Y Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures

2 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures pavement applications. The correlation between SDENT extension and binder failure strain is also not as strong as for GRP. Because numerous researchers and engineers have used the GRP over the past 5 years to indicate nonload-associated cracking and fatigue cracking, and because validation testing suggests the GRP correlates reasonably well to fatigue perfor- mance in the field, this parameter is proposed here as the best overall choice for a binder fatigue specification parameter to replace |G*| sin δ. A serious problem—probably even more significant than that with |G*| sin δ—is the ineffective way the current specification addresses binder fatigue test temperature. Ideally, the fatigue test temperature should be closely tied to average pavement temperature, but the current specification does not appear to do this well. This is partly because the test temperature is tied to both the low and the high binder temperature grades and unless an agency carefully applies the specification, the fatigue test temperature can be significantly higher than intended. Even when temperature is applied correctly (to base binder grades before adjustments for traffic and vehicle speed), the relationship between the current binder fatigue test temperature and average pavement temperature is not ideal. A different approach is suggested in this report, in which the fatigue test temperature is tied to the low performance grade (PG), rather than to the average of the low and high PG grades. Furthermore, this report suggests several notes to make the intent of the specification clear. For example, some areas commonly use binders rated as much as two grades higher than the low PG grade the FHWA LTPPBind software recommends. In such cases, the pavement will not only be prone to increased thermal cracking but will also be subject to significantly greater fatigue cracking unless the pavement thickness is increased. Including a note to this effect would help ensure that highway agencies are aware of this potential problem. The problem associated with binder fatigue test temperature and grade bumping does not occur when AASHTO M 332 is used in place of AASHTO M 320, because M 332 does not involve changing the higher binder test temperature for higher traffic levels. Replacing |G*| sin δ with GRP and using an improved protocol for specifying binder fatigue test temperature would solve many problems with the current binder fatigue speci- fication. However, in addition to these changes, an allowable range for R-value (or an equivalent parameter such as ΔTc) should be established. There are two reasons for this limitation: 1. In thin pavements at low temperatures, binders with high R-values can result in rapid accumulation of fatigue damage. This is likely a major reason for recently observed pre- mature failure of pavements in Ontario and the Northern United States made with binders containing REOB (recycled engine oil bottoms), which tend to have high R-values. 2. In thick pavements, binders with low R-values can show poor fatigue performance. By limiting both high and low R-values, both problems are addressed. Tentative allowable ranges for R-value are from 1.5 to 2.5 for binders aged with a rolling thin film oven test (RTFOT) followed by 20 hours of aging in a pressure-aging vessel (PAV), and from 2.0 to 3.2 for binders aged with RTFOT followed by 40-hour PAV. The more critical limit for R-value is the upper one, as the upper limit may be associated with rapid failure of thin pavements and severe distress under other high-strain applications. NCHRP Project 09-60, “Addressing Impacts of Changes in Asphalt Binder Formulation and Manu- facture on Pavement Performance through Changes in Asphalt Binder Specifications,” is collecting and analyzing additional information concerning the relationships among rheological parameters and properties related to asphalt binder performance and should provide more insight into the most effective ways of incorporating R-value or related parameters into a revised specification. Data collected during NCHRP 09-60 should also

Summary 3   allow initial estimates of the precision of R-value, ΔTc, and related parameters, which is critical in establishing realistic specification limits. The current laboratory aging method for binder fatigue testing uses RTFOT condition- ing followed by PAV aging for 20 hours. In NCHRP 09-59, the aging protocol was more severe—RTFOT conditioning and 40 hours of PAV aging. For any binder fatigue test to be effective, it must include a laboratory aging method that mimics aging in the field with reasonable accuracy. Much research has recently been conducted to address this problem, including NCHRP Project 09-61, “Short- and Long-Term Binder Aging Methods to Accu- rately Reflect Aging in Asphalt Mixtures,” which directly addresses laboratory aging of binders (Bonaquist et al., 2021). Because future changes in the binder-aging protocol are probable, and because problems in the binder fatigue specification should be addressed as quickly as possible, the findings of NCHRP 09-59 should be implemented with the current aging protocol. A final critical issue is what value the chosen specification parameter should have. The current binder fatigue parameter, |G*| sin δ, has a maximum value of 5,000 kilopascal (kPa). A simple approach—and one that would be easy to implement—would be to use an equivalent limit on the new specification parameter. For GRP, this equivalent value is estimated to be 5,300 kPa; for practical purposes, the value could probably be rounded down to 5,000 kPa—identical to the current maximum for |G*| sin δ. Using this equivalent value would be to assume that the aging protocol would initially be unchanged (RTFOT followed by 20-hour PAV). Using this equivalent GRP value and similar, but more effective, test temperatures would ensure that the new specification is not overly restrictive or lax. Furthermore, these analyses suggest that the primary problems with the current binder fatigue specification do not involve the exact specification value, but instead result from the weak relationship between |G*| sin δ and strain capacity, poorly defined and implemented test temperatures, and an aging protocol that may not accurately reflect in situ aging. Maintaining similarity between the current specification and the proposed binder fatigue specification should allow changes addressing two problems with the current binder fatigue specification to be implemented rapidly. The third issue—binder laboratory aging—has been addressed in NCHRP 09-61 (Bonaquist et al., 2021). On the basis of research conducted for NCHRP 09-59, the following 11 conclusions concerning the relationship between asphalt mixture fatigue performance and potential binder fatigue specification parameters can be made: 1. Asphalt binder fatigue performance increases with increasing binder failure strain and increasing fatigue exponent. 2. Modulus is the most important factor affecting binder failure strain; as modulus increases, failure strain decreases dramatically. 3. The fatigue exponent for an asphalt mixture is inversely related to the binder phase angle. 4. A binder’s failure strain during fatigue loading at a given temperature and loading rate can be experimentally measured; it is called the fatigue strain capacity (FSC) and is close in value to direct measurements of binder failure strain. 5. The current binder fatigue specification parameter, |G*| sin δ, and SDENT extension are only moderately correlated to FSC. The GRP correlates better to FSC and is a good indi- cator of binder failure strain under a given set of conditions. For this reason, GRP is a better choice for a binder fatigue parameter than |G*| sin δ or SDENT. 6. For thin pavements—those subject to higher strains and those for which pavement deflections are largely controlled by the subgrade—high R-values can result in poor fatigue performance at low temperatures. For this reason, an improved binder fatigue

4 Relationships Between the Fatigue Properties of Asphalt Binders and the Fatigue Performance of Asphalt Mixtures specification should include a maximum value for R or some equivalent control such as a limit on ΔTc. NCHRP 09-59 also found evidence, though not as compelling, that low R-values can result in poor fatigue performance for thick pavements, where the strain is controlled by characteristics of the bound layers. 7. The current protocol for determining binder fatigue test temperature is not consistently tied to average pavement temperatures for a range of climates. An improved protocol is described in this report, in which binder fatigue test temperature is tied to the low- temperature PG grade, which avoids many problems with how the binder fatigue test temperature is currently determined. 8. Several field validation sites and FHWA Accelerated Loading Facility (ALF) fatigue data show reasonably good correlations between GRP and fatigue performance. Although the correlations with |G*| sin δ are also good in some cases, GRP appears to relate better overall to fatigue performance. 9. When using the proposed binder fatigue test temperatures, a reasonable and effective maximum value for GRP after RTFOT/20-hour PAV aging is 5,000 kPa at 10 radian per second (rad/s). For RTFOT/40-hour PAV aging, this limit should be raised to 8,000 kPa at 10 rad/s. 10. A reasonable and effective maximum R-value for binders aged using RTFOT/20-hour PAV aging is 2.50. For RTFOT/40-hour PAV aging, this limit should be increased to 3.20. 11. The extended loose-mix aging procedure used in NCHRP 09-59 may have resulted in an unrealistic loss of fatigue and fracture performance for some or all polymer-modified binders included in the study. If so, this aging procedure would affect the applicability of some study findings to polymer-modified binders and would also call into question the advisability of adopting extended loose-mix aging for widespread use in asphalt con- crete mix design and analysis. Of particular concern in this context is whether the same maximum R-value should apply to both non-modified and polymer-modified binders, or whether the maximum value of R should be increased or even eliminated for polymer- modified binders. Extended loose-mix aging is an important topic for follow-up research. On the basis of research conducted for NCHRP 09-59, the following proposals can be made concerning the relationship between asphalt mixture fatigue performance and potential binder fatigue specification parameters: 1. The current binder fatigue test temperatures should be replaced with the values shown in Table S-1. 2. The current binder fatigue specification parameter, |G*| sin δ, should be replaced by the GRP, |G*| (cos δ)2/sin δ. As in the current specification, GRP should be determined at a frequency of 10 rad/s. The maximum allowable value for GRP after RTFOT/20-hour PAV aging should be 5,000 kPa. 3. The binder fatigue specification should include an allowable range for the Christensen- Anderson R-value of from 1.5 to 2.5, after RTFOT/20-hour PAV aging. The R-value should be calculated using the following equation: ( ) ( ) ( ) = − R log log S log m 2 3,000 1 where R = Christensen-Anderson R-value (rheologic index), S = bending beam rheometer (BBR) creep stiffness at 60 seconds, Megapascal (MPa), and m = BBR m-value at 60 seconds. Low PG Grade °C Proposed Binder Fatigue Test Temp. °C −46 15 −40 17 −34 19 −28 22 −22 25 −16 27 −10 29 Table S-1. Proposed binder fatigue test temperatures.

Summary 5   4. This proposed maximum value for GRP and range for R-value should be considered tentative. Final values should be based on review and comment about the proposed speci- fication by pavement engineers and researchers and on collection of additional data for a wide range of binders. One consideration is whether the precision of R is adequate for a specification involving both a minimum and a maximum value. If the R-value is not precise enough, a specification using only a maximum value of R should be implemented. These initial proposed values are for binders aged using RTFOT/20-hour PAV. If in the future the binder-aging protocol is changed, any specification limits adopted as a result of this research should be thoroughly reviewed and adjusted to accommodate changes in the binder-aging protocol. 5. Suitable alternatives to R-value may be used in an improved binder fatigue specification. These include ΔTc and BBR stiffness at an m-value of 0.3. Some additional work would be needed. At least some of this research is being addressed in NCHRP 09-60. 6. Additional data concerning the relationship between binder rheology, fracture proper- ties, and other performance-related parameters are being generated for a wide range of binders, including polymer-modified binders, as part of NCHRP 09-60. The findings, conclusions, and proposals of NCHRP 09-59 should be reevaluated after NCHRP 09-60 concludes.