Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results (2021)

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.

134 Material-specific parameters for virgin mixtures, specifically log |G*| at short-term aging (STA) and M, can be estimated using standard binder aging methods as discussed previously. However, since standard aging methods are generally used to age virgin binders, the material- specific parameters obtained would not be representative of the actual material-specific param- eters of mixtures containing RAP. RAP is generally significantly more aged compared to virgin binders and would thus have a higher log |G*| at STA and lower M. The material-specific kinetics parameters of a mixture with a specific RAP content can be obtained directly from loose mix aging at a single temperature (95°C). However, the devel- opment of a model framework that can estimate the material-specific kinetics parameters of a mixture with a specific RAP content from virgin and 100% RAP material-specific kinetics parameters is needed. Arrhenius (1887) introduced a blending rule to estimate the overall rheological properties of a blend of two viscous components given their viscosities and concentrations. Davison et al. (1994) validated the Arrhenius blending rule for asphalt binder blends, as shown in Equa- tion (1). Yousefi Rad, Roohi Sefidmazgi, and Bahia (2014) validated the Arrhenius rule for the binder shear modulus |G*| values of two binders in the laboratory, as presented in Equation (2). η = η × η η α α (1)mix A B B = ×α α* * * * (2)G G G Gmix A B B where ηmix = viscosity of the overall blend, ηA = viscosity of component A, ηB = viscosity of component B, |G*|mix = shear modulus of the overall blend, |G*|A = shear modulus of component A, |G*|B = shear modulus of component B, and α = concentration of component A. Glaser et al. (2015) developed a model to account for the effects of RAP content on long- term aging rates. The carbonyl + sulfoxide data obtained from virgin binders and 100% RAP extracted binders were fitted to the oxidation model by adjusting only the reactive material and initial time zero carbonyl + sulfoxide. The reactive material in the blend was predicted using a A P P E N D I X B Estimation of the Material-Specific Parameters of RAP-Containing Mixtures Given Individual Virgin and RAP Parameters

Estimation of the Material-Specific Parameters of RAP-Containing Mixtures Given Individual Virgin and RAP Parameters 135   mass fraction weighted average of the virgin binder’s reactive material and 100% RAP extracted binder’s reactive material, as shown in Equation (3). ( )= − +1 (3)RM X RM X RMBlend RAP Binder RAP RAP where RMBlend = reactive material in the blend, RMBinder = reactive material in the binder, RMRAP = reactive material in the RAP, XRAP = mass fraction of the RAP, and (1 − XRAP) = mass fraction of the binder. Glaser et al. (2015) successfully predicted long-term aging rates of 15% and 50% RAP given the aging rates of virgin and 100% RAP binders, as presented in Figure 6. Two Strategic Highway (a) (b) (c) (d) Figure 6. Oxidation behavior of virgin binder, RAP binder, and various blends along with predicted behavior based on a mass averaged amount of reactive material in the blends: (a) AAC-1 and Manitoba RAP blends, (b) AAC-1 and South Carolina RAP blends, (c) AAA-1 and Manitoba RAP blends, and (d) AAA-1 and South Carolina RAP blends (Glaser et al. 2015).

136 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Research Program (SHRP) binders (AAA-1 and AAC-1) were blended with two RAP sources (Manitoba and South Carolina) in the Glaser et al. study (2015). As shown in Figure 6, the pro- posed model was able to predict the long-term aging rates of different RAP blends. In this investigation, the Arrhenius rule for blending was used to predict the binder shear modulus values of RAP blends under short-term aged conditions, and the Glaser model was used to account for the effects of RAP content on the long-term aging rate to predict the material- specific kinetics of the RAP mixtures. Glaser et al.’s (2015) model was developed originally based on binder thin-film aging data and the use of chemical AIPs. However, here, the Glaser model was applied to log |G*| at 64°C and 10 rad/s data obtained from binder extracted and recovered from oven-aged loose mixture and from USAT binder aging. In addition, predicted and measured binder log |G*| from short-term aging and long-term aging were compared to evaluate the accuracy of the proposed framework. Finally, the predicted material-specific kinetics parameters of 50% RAP were used as inputs in the calibrated pavement aging model to predict field aging. The predicted field aging was compared against measured field aging from field cores to evaluate the end results of the proposed framework. Considering A to be RAP in Equation (2) and B to be virgin binder, α becomes the asphalt binder replacement (ABR). Taking the log of both sides of Equation (2) yields Equation (4). ( )= − + ×log * 1 log * log * (4)0, 0, 0,G ABR G ABR GBlend Binder RAP where |G*|0,Binder = short-term binder shear modulus of virgin binder, |G*|0,RAP = short-term binder shear modulus of 100% RAP binder, |G*|0,Blend = short-term binder shear modulus of RAP blend, and ABR = asphalt binder replacement. As shown in Table 1, three sources of virgin binders and RAP were considered. The material- specific kinetics parameters of the NCAT virgin binder and RAP binder were obtained through loose mix aging of a mixture with no RAP and through loose mix aging RAP itself at multiple durations, followed by extraction, recovery, and testing of the binder. Similarly, the NCAT materials were used to create 20% RAP, 30% RAP, 50% RAP, and 70% RAP mixtures. These mixtures were aged at multiple durations, and the binder was extracted, recovered, and tested. The material-specific kinetics parameters of the virgin and RAP binder were used along with the Arrhenius and Glaser models to predict the material-specific kinetics parameters of the 20% RAP, 30% RAP, 50% RAP, and 70% RAP mixtures. Table 2 shows the measured material- specific parameters of the NCAT virgin and RAP binders as well as the predicted material- specific parameters of the 20% RAP, 30% RAP, 50% RAP, and 70% RAP mixtures. Mixture/Site ID Material Source % RAP Binder Grade Date Built Date Core Extracted NCAT N10 Alabama 50% PG 67-22 2009 2013 NCAT Alabama 20%, 30%, 70% PG 67-22 N/A N/A MnRd C.21 Minnesota 20% PG 58-34 2016 N/A RS9.5B North Carolina 30%, 50% PG 58-28 N/A N/A * NCAT is National Center for Asphalt Technology; MnRd stands for MnROAD, which is a pavement test track operated by Minnesota DOT. Table 1. Information about materials with RAP used to verify Arrhenius rule and Glaser’s model.

Estimation of the Material-Specific Parameters of RAP-Containing Mixtures Given Individual Virgin and RAP Parameters 137   Figure 7 presents the predicted and measured binder shear modulus values for the short- term aged and long-term aged mixtures. As shown, the proposed framework can adequately predict the log |G*| of long-term aged RAP mixtures. However, the proposed framework under- predicted the log |G*| of the short-term aged RAP mixtures. The material-specific kinetics parameters of the virgin and RAP binders of MnRd C.21 and RS9.5B were obtained through aging the binders using USAT. Recall that USAT is a binder aging method that uses thin binder film (0.3  mm thick) to induce a kinetics-controlled reaction (Farrar et al. 2014). The material-specific kinetics parameters of the virgin and RAP binder were used along with the Arrhenius and Glaser models to predict the material-specific kinetics parameters of the 20% RAP MnRd blended binder, 30% RAP RS9.5B blended binder, and the Mix ID Asphalt Binder Replacement (ABR) log |G*| at STA M 0% RAP (Measured) 0 0.851 0.883 100% RAP (Measured) 1 3.520 0.267 20% RAP 0.152 1.257 0.789 30% RAP 0.229 1.462 0.742 50% RAP 0.385 1.879 0.646 70% RAP 0.545 2.306 0.547 Table 2. Summary of the material-specific kinetics parameters for NCAT. 1 1.5 2 2.5 3 3.5 4 0 10 20 30 40 lo g G * Duration (Days) Measured LTA Aging at 95°C Measured STA Prediction LTA Aging at 95°C 1 1.5 2 2.5 3 3.5 4 0 10 20 30 40 lo g G * Duration (Days) Measured LTA Aging at 95°C Measured STA Prediction LTA Aging at 95°C 1 1.5 2 2.5 3 3.5 4 0 10 20 30 40 lo g G * Duration (Days) Measured LTA Aging at 95°C Measured STA Prediction LTA Aging at 95°C 1 1.5 2 2.5 3 3.5 4 0 10 20 30 40 lo g G * Duration (Days) Measured LTA Aging at 95°C Measured STA Prediction LTA Aging at 95°C (a) (b) (c) (d) Figure 7. Predicted versus measured short-term aged (STA) and long-term aged (LTA) RAP mixtures: (a) NCAT 20% RAP, (b) NCAT 30% RAP, (c) NCAT 50% RAP, and (d) NCAT 70% RAP.

138 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results 50% RAP RS9.5B blended binder. The original NCHRP Project 09-54 developed an empirical model to relate the USAT binder aging to loose mixture aging. Using this empirical model, the material-specific loose mixture kinetics parameters were obtained for the 20% RAP MnRd blended binder, 30% RAP RS9.5B blended binder, and the 50% RAP RS9.5B blended binder. Table 3 shows the material-specific parameters obtained at each step in going from USAT kinetics of virgin and RAP binder to loose mixture kinetics of the blended binder. The original MnRd C.21 mixture containing 20% RAP, the original RS9.5B mixture con- taining 30% RAP, and the fabricated RS9.5B mixture containing 50% RAP were also long-term aged in the oven at multiple durations, and the binder was extracted, recovered, and tested. Figure 8 presents the predicted and measured binder shear modulus values for the short-term aged and long-term aged mixtures. The predicted binder shear modulus values agree well with the measured values for the MnRd C.21 mixture. Some deviations are noticeable, though, for the RS9.5B 30% RAP and 50% RAP mixtures, especially at the STA condition. In general, it can be concluded that if the kinetics parameters of the individual virgin and RAP materials are known, the kinetics parameters of a mixture with a specific RAP content can be determined with reasonable accuracy. Mixture Material Determination Method log |G*| at STA M ABR MnRd C.21 Virgin Binder USAT 0.903 0.976 0 RAP Binder USAT 2.260 0.653 1 Blended Binder (20% RAP) USAT, Arrhenius/Glaser Models 1.214 0.902 0.229Blended Binder (20% RAP) USAT, Arrhenius/Glaser Models, Empirical Model 0.731 0.816 RS9.5B Virgin Binder USAT 1.138 0.927 0 RAP Binder USAT 2.137 0.628 1 Blended Binder (30% RAP) USAT, Arrhenius/Glaser Models 1.381 0.855 0.243 Blended Binder (30% RAP) USAT, Arrhenius/Glaser Models, Empirical Model 0.897 0.804 Blended Binder (50% RAP) USAT, Arrhenius/Glaser Models 1.590 0.792 0.452 Blended Binder (50% RAP) USAT, Arrhenius/Glaser Models, Empirical Model 1.105 0.745 Table 3. Summary of the material-specific kinetics parameters for MnRd C.21 and RS9.5B.

Estimation of the Material-Specific Parameters of RAP-Containing Mixtures Given Individual Virgin and RAP Parameters 139   0.0 1.0 2.0 3.0 4.0 0 5 10 15 20 lo g G * Duration (Days) Measured STA Measured LTA at 95°C Predicted Loose Mix LTA (b) 0.0 1.0 2.0 3.0 4.0 0 5 10 15 20 lo g G * Duration (Days) Measured STA Measured LTA at 95°C Predicted Loose Mix LTA (c) 0.0 1.0 2.0 3.0 4.0 0 5 10 15 20 lo g G * Duration (Days) Measured STA Measured LTA at 95°C Predicted Loose Mix LTA (a) Figure 8. Predicted versus measured short-term aged (STA) and long-term aged (LTA) RAP mixtures: (a) MnRd C.21 20% RAP, (b) RS9.5B 30% RAP, and (c) RS9.5B 50% RAP.