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.

60 Research Approach Overview Alternatives to loose mixture aging to obtain inputs for the PAM were investigated; these alternatives include aging binders in both an RTFO and PAV before testing as well as using the PG to derive approximations. Obtaining material-specific kinetics parameters [i.e., log |G*|0 and M in Equation (1)] by test- ing extracted and recovered binders is a cumbersome process due to the demands of the extrac- tion and recovery process. Therefore, the possible ability to use standard binder aging methods (i.e., RTFO and PAV) to approximate the material-specific parameters was evaluated. Eight binders were aged in an RTFO, and the residue was subjected to 20-hour and 40-hour PAV aging. An empirical relationship was developed that relates the binder AIPs measured from the RTFO-aged and 20-hour PAV-aged and 40-hour PAV-aged binders to those measured from binder extracted and recovered from the long-term aged loose mixture. This developed relation- ship was then used to estimate the PAM input parameters. In addition, the relationship between the binder PG and the PAM parameters was used to develop an even simpler, albeit less accurate, alternative to obtain the PAM inputs. In practice, performance grading is generally conducted using virgin binders, not extracted and recovered RAP binders. Therefore, to facilitate the consideration of RAP when estimating the PAM input parameters, a means to estimate the properties of the RAP binders using the PAM was also established. To establish this framework, the applicability of blending rules to RAP mixture aging was first investigated using loose mixture and USAT binder aging tests. Based on the findings, a means was established to approximate the PAM parameters of RAP mixtures based on the combination of virgin and RAP binder grades. For cases where the RAP binder grade was unknown, the PAM generated a map to approximate the PGs of 20-year-old RAP binders throughout the United States. Test Materials Eight virgin mixtures were used to develop a framework to estimate the PAM parameters using standard binder aging methods. These eight mixtures are ALF Control, ALF SBS, ALF CRTB (crumb rubber terminal blend), LTPP Washington, LTPP South Dakota, LTPP Texas, LTPP New Mexico, and ARC (Asphalt Research Consortium). These mixtures were aged in the oven at 95°C, and the binders were extracted and recovered at multiple durations and then tested. The same eight virgin binders were aged in the RTFO, followed by 20-hour PAV and C H A P T E R 5 Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 61   40-hour PAV aging. The kinetics of the loose mixture aging of each mixture were compared against the kinetics obtained from RTFO and PAV aging. Table 10 presents details about the eight mixtures and binders. Note that information about all the binders used in this project was employed to develop a relationship between the kinetics parameters (M and log |G*| at short- term aging) and the PGs. Sample Preparation Methods The sample preparation methods used to refine the PAM largely coincide with those described in Chapter 3 but are repeated here for convenience of the reader. Asphalt Mixture Aging All the asphalt mixtures aged in the laboratory were prepared using the component aggre- gate and binder that were used to construct the pavements from which the field cores were obtained. The mixtures were subjected to short-term aging at 135°C for 4 hours in accordance with AASHTO R 30 prior to long-term aging. The long-term oven aging of the loose mixtures was accomplished by separating the mix into several pans such that each pan had a relatively thin layer of loose mix that was approximately equal to the NMAS of the aged mix. The pans of loose mixture were conditioned in an oven at 95°C and systematically rotated to minimize any effects of an oven temperature gradient and/or draft on the degree of aging. After long-term aging, the materials were taken out of the oven and mixed to obtain a uniform mixture. Project ID Section ID Location/Source % RAP NMAS % AC Binder Grade FHWA ALF Control (ACTRL) Virginia 0% 12.5 mm 5.30% PG 70-22 ALF - Lane 1 SBS-LG (ASBS) Virginia 0% 12.5 mm 5.30% PG 70-22 ALF - Lane 4 CRTB (ACRTB) Virginia 0% 12.5 mm 5.30% PG 70-22 ALF - Lane 5 LTPP LWA Washington State 0% 9.5 mm 6.10% AR-4000 (AR_40 by AASHTO designation)LTPP section ID: 53-0801 LSD South Dakota 0% 12.5 mm 5.90% Pen. 120–150LTPP section ID: 46-0804 LTX Texas 0% 9.5 mm 5.40% AC-20LTPP section ID: 48-0802 LNM New Mexico 0% 19.0 mm 6.50% AC-20LTPP section ID: 35-0801 ARC ARC ARC BI-0001 Binder and North Carolina Aggregate 0% 9.5 mm 6.40% PG 67-22 Note: SBS-LG = linear grafted styrene-butadiene-styrene, CRTB = crumb rubber terminal blend, ARC = Asphalt Research Consortium Table 10. Materials used to estimate PAM Parameters using standard binder aging.

62 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Microextraction and Recovery Microextraction and recovery of the asphalt binder from the asphalt mixtures and field cores were undertaken following the procedure proposed by Farrar et al. (2015). This procedure uses a solvent mixture of toluene and ethanol (85:15). The mixture sample size is limited to 200 g to produce approximately 10 g of asphalt binder per extraction, which is adequate for both ATR-FTIR spectrometry testing and DSR testing. To prevent further aging of the binder, the distillation flask was subjected to vacuum pressure of 80.0 ± 0.7 kPa (600 ± 5 mm Hg) under nitrogen gas during the recovery procedure. The recovered samples were then placed in a degassing oven and heated to 130°C for 60 minutes under nitrogen to remove any remaining traces of the solvent. In this study, ATR-FTIR spectrometry testing was conducted following extraction and recovery to ensure that no detectable solvent was present prior to DSR testing. RTFO Aging RTFO aging was conducted using selected original asphalt binder samples according to AASHTO T 240 to simulate short-term aging. PAV Asphalt binder residues obtained from the RTFO aging were subjected to PAV aging accord- ing to AASHTO R 28 at 100°C at the standard 20 hours and for an additional total aging time of 40 hours. Determination of Rheological Aging Index Properties Frequency sweep tests of the extracted and recovered binders were conducted using an Anton Paar modular compact rheometer (MCR) 302 at frequencies ranging from 0.1 Hz to 30 Hz at 64°C using the 8 mm or 25 mm parallel plate geometry. A strain amplitude of 1% was applied at all frequencies. The log |G*| at 64°C and 10 rad/s was used as the rheological AIP for all analyses. Findings and Applications The use of the PAM requires the short-term aged log |G*| (i.e., and log |G*|0) and the material- specific parameter M. In all the work presented thus far, these two parameters have been obtained through testing binder that has been extracted and recovered from aged loose mixture (fol- lowing the loose mixture long-term aging procedure recommended by the original NCHRP Proj- ect 09-54). The log |G*| at short-term aging conditions is obtained by aging loose mixture in the oven at 135°C for 4 hours and then extracting, recovering, and testing the binder. The parameter M is obtained by measuring log |G*| at short-term aging conditions and at multiple long-term aging conditions by aging loose mixture in an oven at 95°C, then extracting, recovering, and testing the binder. M in Equation (1) is then optimized such that the predicted log |G*| values match the measured values. Recall that any laboratory, long-term aging duration can be considered in order to obtain log |G*| to calibrate M as long as the duration allows for binder AIPs that are well dis- persed on the oxidation timescale. The aging durations should allow binder AIPs that belong to the constant region in the kinetics plot of |G*| versus aging duration so that the oxidation kinetics data obtained are meaningful and reproducible, as shown in Figure 40. Obtaining material-specific kinetics parameters through testing extracted and recovered binders is a cumbersome process due to the experimental demands of the extraction and recov- ery process. Two alternatives were considered in this study to obtain the material-specific parameters. The first was to use standard binder aging methods, i.e., RTFO and PAV testing, and

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 63   the second was to establish an empirical relationship between the binder PG and the material- specific parameters. The method presented here to equate loose mixture kinetics and binder kinetics obtained from RTFO and PAV tests is empirical in nature. Loose mixture aging has been proven to be kinetics-controlled due to the thin binder film that covers the aggregate particles and allows faster diffusion of oxygen through the film than the oxidation reaction itself. Loose mixture aging, however, also includes the physicochemical effects of filler, which slows down the asphalt binder oxidation rate. The binder film used in RTFO and PAV tests is not thin enough to induce a kinetics-controlled reaction; hence, the binder oxidation rates are also diffusion-controlled. Because the oxidation mechanisms of the two methods (loose mixture aging and RTFO/PAV aging) are not similar, relating the loose mixture rate to the RTFO/PAV rate is accomplished empirically. The original NCHRP Project 09-54 provided an alternative to loose mixture aging to obtain the inputs for the PAM. This alternative was the USAT that uses thin binder film (0.3 mm thick) to induce a kinetics-controlled reaction. The original NCHRP Project 09-54 also developed an empirical model to relate the USAT-based kinetics to loose mixture kinetics by accounting for the filler effects and the presence of hydrated lime. Although the aging mechanisms for the USAT and loose mixture aging are similar in that they are both kinetics-controlled, the draw- back of the USAT is that it is a non-standard, binder aging method and is relatively cumber- some to run in comparison to RTFO and PAV tests, especially in achieving and maintaining the 0.3-mm-thick binder film before and during short-term aging and long-term aging. The research effort herein used the results of RTFO and PAV aging tests instead of deriving loose mixture kinetics information. The PG has been found to relate to the binder log |G*| in terms of both short-term aging conditions and the material-specific parameter M. However, the PG method is regarded as secondary to the RTFO/PAV aging alternative, which in turn is considered secondary to the loose mixture aging methodology in terms of accuracy. Estimation of Material-Specific Parameters Using Standard Binder Aging Methods Eight binders were aged in the RTFO and twice in the PAV (20 hours and 40 hours). Fig- ure 41 shows the relationship between the log |G*| values obtained under short-term aging 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10 20 30 40 lo g |G *| at 6 4° C, 1 0 ra d/ s (k Pa ) Loose Mixture Lab Aging Duration (Days) Extracted and Recovered Fit Figure 40. Example of log |G*| values obtained from testing extracted and recovered binder at STA and multiple LTA durations.

64 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results (loose mixture aging) conditions and log |G*| values obtained from RTFO aging. Some agree- ment between the log |G*| obtained from loose mix aging and RTFO aging can be found. Short- term aging of loose mix is conducted at 135°C for 4 hours, whereas RTFO aging is conducted at 163°C for 85 minutes. Roughly the same log |G*| value is obtained at the higher temperature with less time as at the lower temperature for a longer time. This finding is in concert with the filler physicochemical effects that slow down the oxidation reaction in loose mixture and the thicker binder film used in RTFO aging. Figure 42 (a) shows the ratio of log |G*| for 2 days of aging with respect to short-term aging log |G*| values versus the ratio of log |G*| for 20 hours of PAV + RTFO aging with respect to RTFO log |G*| values. Figure 42 (b) shows the ratio of log |G*| for 6 days of aging with respect to short-term aging log |G*| values versus the ratio of log |G*| for 40 hours of PAV + RTFO aging with respect to RTFO log |G*| values. The log |G*| values after RTFO aging and 20 hours in the PAV seem to agree with 2 days of oven aging at 95°C after 4 hours of short-term aging at 135°C, as shown in Figure 42 (a). Also, the log |G*| values after RTFO aging and 40 hours in the PAV seem to agree with 6 days of oven aging, as shown in Figure 42 (b). Because the kinetics plot consists of a secondary constant region, estimating the slope should be possible, albeit with some error. Hence, the log |G*| of that region should also be possible to estimate, if knowing log |G*| at 6 days of aging. Figure 43 demonstrates the changes in slope for the kinetics model fit of each of the eight binders. A little after the 6-day mark, the slope converges to a constant value that varies slightly among the binders, depending on their sus- ceptibility to aging. ST A lo g |G *| at 6 4° C, 1 0 ra d/ s (k Pa ) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 RTFO log |G*| at 64°C, 10 rad/s (kPa) R2 = 0.8884 RMSE = 0.1309 Figure 41. Agreement between log |G*| obtained from RTFO aging of binder and STA of loose mixture. 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 2- da y lo g |G *| / S TA lo g |G *| (k Pa /k Pa ) 20-hr PAV + RTFO log |G*| / RTFO log |G*| (kPa/kPa) (a) R2 = 0.8611 RMSE = 0.3286 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 6- da y lo g |G *| / S TA lo g |G *| (k Pa /k Pa ) 40-hr PAV + RTFO log |G*| / RTFO log |G*| (kPa/kPa) (b) R2 = 0.9232 RMSE = 0.4613 Figure 42. Agreement between (a) 2-day log |G*| and 20-hour PAV + RTFO log |G*| and (b) 6-day log |G*| and 40-hour PAV + RTFO log |G*|.

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 65   The log |G*| beyond 6 days of aging was observed to be proportional to log |G*| at 6 days of aging by some factor. Equation (25) was developed after averaging the factors obtained from the eight binders. = × ≥−log * log * 1.02635 6 (25)6 6G G for iiD D i Knowing log |G*| at 0 days, 2 days, and 6 days and beyond, M can be obtained by optimizing Equation (1) such that the predicted log |G*| values match the estimated ones obtained from RTFO and PAV aging. Table 11 shows the M values obtained from loose mix aging and RTFO/PAV aging. Generally, the M values obtained from RTFO/PAV aging are fairly close to the measured loose mix aging values with percentage errors less than 10%, except for ACTRL and ARC that have percentage errors up to 20%. The ARC binder is highly structured. Struc- tured binders tend to be more sensitive to parameters such as temperature and pressure. The high difference observed between the M values could be due to the ARC binder being sensitive to the aging method. The aforementioned approach requires both 20 hours and 40 hours of PAV aging to estimate M. However, the estimation of M using a single PAV aging duration also merits con- sideration due to the associated time savings. The process of using a single PAV to estimate M would be like that presented earlier. For example, with RTFO and 40-hour PAV, log |G*| at 0 and 6 days will be known since RTFO |G*| can be considered equivalent to loose mixture STA |G*|. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 5 10 15 20 25 Sl op e of K in et ic s C ur ve Duration (Days) ARC ACTRL ASBS ACRTB LWA LSD LTX LNM Figure 43. Convergence of slope of kinetics curve to a constant value that varies among binders. Binder ID M from Loose Mix Aging M from RTFO- PAV Aging ARC 1.075 0.848 ACTRL 0.772 0.668 ASBS 0.598 0.620 ACRTB 0.681 0.698 LWA 0.865 0.899 LSD 0.684 0.654 LTX 0.852 0.838 LNM 0.546 0.583 Table 11. Material-specific M parameters obtained from loose mix aging and RTFO-PAV aging.

66 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Similarly, 40-hour PAV |G*| can be considered equivalent to 6-day loose mixture aging |G*|. Equation (25) can then be used to estimate |G*| beyond 6 days. Afterward, M can be obtained by optimizing Equation (1) such that the predicted log |G*| values match the estimated ones obtained from RTFO and 40-hour PAV aging. Figure 44 shows an example of the fit of the PAM using both the RTFO and 40-hour PAV aging and the loose mixture aging as well as the measured data point from both aging methods that were used to obtain the fit. Figure 45 demonstrates that the M values obtained from only 40 hours of PAV [Figure 42 (b)] match well along the LOE. However, significant scatter is observed along the LOE if only 20 hours of PAV is used [Figure 42 (c)]. Therefore, M can be estimated without sacrificing the prediction accuracy significantly using only 40 hours of PAV. However, 20 hours of PAV should not be used exclusively to estimate M. Validation or recalibration of this methodology using local materials is encouraged. Also, a similar methodology can be applied to estimate M and log |G*| at STA using other aging methods of interest (e.g., modified RTFO/PAV, USAT, etc.). The estimated M and log |G*| using RTFO/PAV were used along with the PAM to compare the predicted results of log |G*| at a certain time and depth of the pavement with those measured from field cores. The results are shown in Figure 46 and demonstrate that using RTFO/PAV can 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.4 0.6 0.8 1 1.2 M R TF O /P AV M Loose Mix (a) R2 = 0.3211 RMSE = 0.0914 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.4 0.6 0.8 1 1.2 M R TF O /P AV M Loose Mix (b) R2 = 0.3190 RMSE = 0.0912 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0.4 0.6 0.8 1 1.2 M R TF O /P AV M Loose Mix (c) R 2= – 0.2411 RMSE = 0.1465 Figure 45. M from loose mix versus M from RTFO/PAV using: (a) 20 hours and 40 hours of PAV, (b) only 40 hours of PAV, and (c) only 20 hours of PAV. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10 20 30 40 lo g |G *| at 6 4° C, 1 0 ra d/ s (k Pa ) Loose Mixture Lab Aging Duration (Days) Extracted and Recovered Fit Fit RTFO/PAV RTFO 40-hour PAV Figure 44. Example of the log |G*| values obtained from testing RTFO and 40-hour PAV-aged binder and the corresponding fit.

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 67   provide an adequate estimation of the material-specific kinetics parameters that yields long- term aged log |G*| values close to those measured from field cores. Estimation of Material-Specific Parameters Using Performance Grade A less accurate alternative for determining the material-specific parameters, compared to the methods described in the previous section, is to use a relationship that allows conversion of the binder PG to log |G*| for both short-term aging and M. The log |G*| values at short-term aging and M are correlated with each other and with the high PG (HPG) of the binder, as shown in Figure 47. A total of 42 points were used to construct the graphs shown in Figure 47. The points were obtained through either loose mixture aging or USAT aging of the different binder sources and RAP sources used in this project. A functional form was fitted to each of these three relationships as shown in Equation (26), Equation (27), and Equation (28). Equation (26) is fitted such that a log |G*| value of 4.5 kPa at 64°C, 10 rad/s is obtained at HPG of 130. Note that more data are needed, especially with HPG greater than 82, in order to validate the form of this equation. It is not directly intuitive that such a relationship exists since it is well known that binders with the same PG but from different sources can have vastly different aging sus- ceptibilities. However, with testing of local materials and RAP sources, such a relationship can gain more credibility. At this stage, this relationship represents the least accurate approach to determine the material-specific parameters but provides considerable time savings. = +     −log * 6.552 1 107.167 (26)4.068G HPG = +     0.94 1 82.68 (27)7.96M HPG ( ) ( )= − × +M Glog 0.1814 log * 0.0417 (28) 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 M ea su re d lo g |G *| af te r Sm oo th in g at 6 4° C , 1 0 ra d/ s (k Pa ) Predicted log |G*| at 64°C, 10 rad/s (kPa) All Depths R2 = 0.5051 RMSE = 0.3122 Figure 46. Comparison between smoothed field core measurements and field-calibrated PAM predictions using inputs estimated from RTFO and 40-hour PAV at all depths.

68 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Estimation of the Pavement Aging Model Parameters for RAP Mixtures Using Virgin Binder Properties For the mixtures that contain RAP, obtaining the material-specific kinetics parameters becomes more complex because of the presence of both virgin and RAP binder and the degree of blending between the two. The blending of the virgin and RAP binders is intricate and is governed by the RAP content, RAP properties, virgin binder properties, mixture volumetrics, and other factors. Ideally, the material-specific kinetics parameters of RAP-containing mixtures are best obtained by aging the loose mixture followed by extraction and recovery and testing of the binder. During loose mixture aging, the virgin binder, RAP binder, and/or any blend of the two, are aged simultaneously, and the inclusion of the physiochemical effects of the aggregate must be considered also as part of this process. Although complete blending of the binders will occur following extraction and recovery, this approach is considered to yield the most accurate characterization of the kinetics of the binder in RAP-containing mixtures. However, the long-term aging of loose mixture and the extraction and recovery process can be cumbersome. Thus, other approaches already described for estimating the material-specific kinetics parameters of RAP-containing mixtures should be investigated. First, the following sec- tion demonstrates that the material-specific kinetics parameters of a RAP-containing mixture can be obtained if individual virgin and RAP kinetics parameters are known. Next, a method for obtaining the RAP kinetics parameters using the PG is presented. Because RAP generally is not characterized per se, PAM predictions are utilized to estimate the RAP properties after a certain field aging duration. Recall that the PAM predicts the binder log |G*| after long-term aging given the initial binder properties, its kinetics parameters, and the pavement temperature history. 0 0.2 0.4 0.6 0.8 1 1.2 50 60 70 80 90 100 110 M High PG (b) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 50 60 70 80 90 100 110 lo g |G *| a t S TA High PG (a) log |G* | at STA -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.0 1.0 2.0 3.0 4.0 lo g M (c) log (M) = -0.1814 x log |G*| + 0.0417 R2 = 0.6943 Figure 47. Relationship between (a) log |G*| at short-term aging and HPG, (b) M and HPG, and (c) log M and log |G*| at short-term aging.

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 69   The RAP kinetics parameters can then be combined with the virgin kinetics parameters to esti- mate the RAP-containing mixture kinetics parameters. Estimation of the Material-Specific Parameters of RAP-Containing Mixtures Given Individual Virgin and RAP Parameters For virgin binder, the material-specific parameters, log |G*| at short-term aging and M, can be estimated using any of the three approaches discussed in the previous section, although loose mixture aging might not be as straightforward as the other two approaches because the mixture is designed to have RAP, so using aggregate with only virgin binder would have to be carefully considered. For RAP mixtures, the two parameters that need to be characterized are log |G*| at short-term aging and M. A method to estimate these two parameters is discussed later in this section. Obtaining log |G*| of a RAP-containing mixture at short-term aging conditions requires combining the log |G*| values of both the virgin and RAP parameters at short-term aging con- ditions using the Arrhenius rule. Similarly, obtaining M of RAP-containing mixtures can be obtained using Glaser’s model. 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 Equation (29). Yousefi Rad, Roohi Sefidmazgi, and Bahia (2014) validated the Arrhenius rule for the binder shear modulus values of two binders in the laboratory, as presented in Equation (30). η = η × η η α α (29)mix A B B = ×α α* * * * (30)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. Considering A to be RAP in Equation (30) and B to be virgin binder, α becomes the asphalt binder replacement. Taking the log of both sides of Equation (30) yields Equation (31). ( )= − + ×log * 1 log * log * (31)0, 0, 0,G ABR G ABR GBlend Binder RAP where |G*|0,Blend = shear modulus of short-term aged RAP blended binder, |G*|0,Binder = shear modulus of short-term aged virgin binder, |G*|0,RAP = shear modulus of short-term aged 100% RAP binder, and ABR = asphalt binder replacement. Glaser et al. (2015) developed a model to account for the effects of RAP content on long- term aging rates. In Glaser’s study, the C+ S 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 C+ S. The reactive material in the blend was predicted using a mass fraction

70 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results weighted average of the virgin binder’s reactive material and 100% RAP-extracted binder’s reactive material, as shown in Equation (32). ( )= − +1 (32)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. Rewriting Equation (32) using the variables of PAM would yield Equation (33). RM is replaced with M for each virgin binder, RAP, and the blend. ( )= − +M X M X MBlend RAP Binder RAP RAP1 (33) Glaser et al. (2015) then successfully predicted the long-term aging rates of 15% RAP and 50% RAP given the aging rates of the virgin and 100% RAP binders. Two SHRP binders (AAA-1 and AAC-1) were blended with two RAP sources (Manitoba and South Carolina) in the Glaser et al. study (2015). The proposed model was able to predict the long-term aging rates of the dif- ferent RAP blends. In a further investigation conducted under this project but presented in Appendix B, the Arrhenius rule for blending is used to predict the log |G*| values of the RAP-containing mixtures at short-term aging conditions, and the Glaser model is used to predict M of the RAP-containing mixtures of the materials used in this project. The Glaser model was developed originally based on binder thin film aging data and the use of chemical AIPs. However, in this investigation, the Glaser model is applied to log |G*| at 64°C and 10 rad/s data obtained from binder extracted and recovered from the oven-aged loose mixture and from USAT binder aging. In addition, predicted and measured binder log |G*| values obtained from short-term aging and long-term aging are compared to evaluate whether the accuracy of the proposed framework is accept- able. Finally, the predicted material-specific kinetics parameters of the 50% RAP-containing mixture are used as inputs in the calibrated PAM to predict field aging. The predicted field aging is compared against measured field aging of field cores to evaluate the end results of the proposed framework. The predictions were found to agree reasonably well with the field core measurements. It was 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. Estimation of the RAP Kinetics Parameters Using Performance Grade According to Figure 47, if the HPG of the RAP is known, the log |G*| and M of RAP can be estimated. If the HPG of the RAP is unknown, the PAM can be used to predict the log |G*| for given initial kinetics parameters at long-term aging conditions. Table 12 shows the kinetics parameters M and log |G*| at short-term aging for binders with an HPG ranging from 52 to 76 obtained using Equation (26) and Equation (27). After predicting log |G*| using the PAM, which can be considered the log |G*| at short-term aging conditions for RAP, Equation (28) can be used to estimate M of RAP. For each HPG listed in Table 12, the PAM was used to predict log |G*| under different climatic conditions. The predicted log |G*| was then used to estimate the HPG of the RAP using Equation (26). This process was iterated for PAM predictions at different years for different pavement depths to obtain RAP HPG at different climates that is close to RAP

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 71   HPG reported in the literature (shown in Table 13 and Figure 48). The RAP HPG were best matched with PAM simulations run for 14 years at 25.4 mm depth. Afterwards, PAM predictions of 14 years at 25.4 mm depth were made for all regions in the United States. The climatic data were obtained from MERRA-2 stations located across the United States and used to calculate the pavement temperature history using the EICM. The pre- dicted log |G*| for each station was then used to estimate the HPG of the RAP. Maps were created to show the HPGs of the RAP across the United States for each initial HPG listed in Table 12. Figure 49 (a) shows the HPGs of RAP throughout the United States that were estimated from 20-year log |G*| values derived from the PAM using virgin binder with an HPG of 52 as an input kinetics parameter. Figure 49 (b), (c), (d), and (e) also show the HPGs of the RAP but with different input kinetics parameters for binders with HPG 58, 64, 70, and 76, respectively. The RAP grades obtained from the maps in Figure 49 were compared to those reported in the literature (shown in Table 13 and Figure 48). It can generally be said that the maps match the reported RAP grades (with unknowns such as the virgin binder PG) except in hot climates (e.g., Arizona and Alabama) where the reported RAP grade was beyond 100 (i.e., higher than the grades shown in the maps) and in very cold climates (e.g., Minnesota, Vermont, and New Hampshire) where the reported RAP grades were less than 76 (i.e., lower than the grades shown in the maps). It is recommended that Equation (26) be locally or regionally calibrated with the time and pavement depth that represent the local RAP materials to better predict the RAP HPG locally, given differences in materials, timing of rehabilitation, and RAP processing methods. HPG log |G*| at STA M 52 0.328 0.917 58 0.498 0.887 64 0.717 0.832 70 0.984 0.743 76 1.298 0.622 Table 12. Estimation of initial kinetics parameters using HPG. State Number of RAP Sources RAP HPG (Count) Reference Alabama 2 PG 94 (1), PG 100 (1) Castorena (2016) Arizona 5 PG 112 - PG 130 Arredondo (2018) California 1 PG 88 (1) Alavi et al. (2015) Florida 20 Average PG 90 West and Willis (2014) Illinois 2 PG 82 (1), PG 88 (1) Al-Qadi et al. (2012) Kansas 2 PG 86 (1), PG 90 (1) Tavakol (2016) Minnesota 3 PG 70 (1), PG 76 (1), PG 88 (1) Johnson et al. (2010) Nevada 4 PG 82 (4) Hajj (2007) New Hampshire 3 PG 76 (2), PG 82 (1) Mensching et al. (2014) New Mexico 3 PG 82 (2), PG 88 (1) Mannan et al. (2018) New York 1 PG 82 (1) Mensching et al. (2014) North Carolina 28 PG 82 (1), PG 88 (10), PG 94 (11), PG > 100 (6), Average: PG 94 Sree Ramoju (2017) Vermont 1 PG 70 (1) Mensching et al. (2014) Wisconsin 2 PG 82 (1), PG 88 (1) Bonaquist (2011) Table 13. List of RAP source and their HPG from literature.

72 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results 2 Sources: PG 82, PG 88 3 Sources: PG 70, PG 76, PG 88 4 Sources: PG 82 2 Sources: PG 94, PG 100 20 Sources: AVG PG 90 28 Sources: 1 : PG 82 10 : PG 88 11 : PG 94 6 : PG >100 AVG: PG 94 2 Sources: PG 86, PG 90 3 Sources: PG 82, PG 88 5 Sources: PG 112 - PG 130 1 Source: PG 88 1 Source: PG 82 3 Sources: PG 82, PG 76 1 Source: PG 702 Sources: PG 82, PG 88 Figure 48. Distribution of RAP sources from literature with their HPG across the map of the United States. Virgin Binder HPG 52 (a) Figure 49. High-performance grades of RAP obtained using kinetics parameters of virgin binder with (a) HPG 52.

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 73   Virgin Binder HPG 58 (b) Virgin Binder HPG 64 (c) Figure 49. (Continued). High-performance grades of RAP obtained using kinetics parameters of virgin binder with (b) HPG 58 and (c) HPG 64. (continued on next page)

74 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Virgin Binder HPG 70 (d) Virgin Binder HPG 76 (e) Figure 49. (Continued). High-performance grades of RAP obtained using kinetics parameters of virgin binder with (d) HPG 70 and (e) HPG 76.

Development of Procedures to Estimate the PAM Inputs Using Standard Binder Aging Methods and PG 75   A user would thus refer to Figure 49 to determine the estimated HPG of the RAP, knowing the virgin binder HPG that is commonly used in the area of interest. To be conservative, the HPG of the RAP should be rounded up to the higher end of the scale shown in the legends presented in Figure 49 or even bumped up another grade depending on individual local experience in handling RAP in the different states. In summary, 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 using the Arrhenius rule, Equation (31), and Glaser’s model [Equa- tion (32)]. The virgin binder kinetics parameters can be estimated using any of the three approaches previously discussed, although loose mixture aging might not be as straight- forward as the other two approaches because the mixture is designed to have RAP, so using aggregate with only the virgin binder would have to be carefully considered. If the HPG of the RAP is known, the RAP kinetics parameters can be obtained through Equation (26) and Equation (27). If the HPG of the RAP is unknown, one can refer to Figure 49 to estimate the RAP HPG for a given initial virgin binder HPG and region of interest or otherwise follow the steps shown in Table 14. Step # Step Description Equation(s) Eqn(s) Ref. Definition of Terms 1 Virgin HPG 2 Material Specific Parameters: M & |G*|0 at 64°C, 10 rad/s 4.068 0log | * | 6.552 / 1 107.167 HPGG 7.96 0.94 / 1 82.68 HPGM (26) (27) |G*|0 = short-term aged binder shear modulus at 64°C, 10 rad/s (kPa), M = parameter related to fast reaction reactive material to be used in the kinetics model, and HPG = virgin binder high performance grade. 3 Pavement Aging Model 0log | * | log | * | 11kinetics cf k tc f G G M e k t k k exp aff f E k A RT exp acc c Ek A RT max , 0 0.916 log | * | log | * | 1 1 tN AP t z z AP G G M e z , 0log | * | log | * |kinetics t t G G AP M 0 max 4.5 log | * |G AP M 100.2260.477 tAPN e (1) (2) (3) (22) |G*|kinetics = long-term aged binder shear modulus at 64°C, 10 rad/s (kPa), kf = rate of fast reaction, kc = rate of constant reaction, Af = regression parameter equal to 1.25×103, Ac = regression parameter equal to 3.68×107, Eaf = regression parameter equal to 95.04, Eac = regression parameter equal to 62.21, R = universal gas constant (0.008314 (kJ/mol K), T = pavement temperature (Kelvin), t = reaction time (days), |G*| t,z = long-term aged binder shear modulus after depth-dependent calibration at 64°C and 10 rad/s at time t (kPa), |G*| kinetics,t = long-term aged binder shear modulus calculated at the pavement surface at time t (kPa), APt = Aging Parameter at the pavement surface at time t (kPa), APt=10 = Aging Parameter at the pavement surface calculated at 10 years of aging (kPa), z = pavement depth (cm), and N = material-dependent parameter. 4 |G*|t=10,z=2.54 at 64°C, 10 rad/s 5 HPG of RAP 4.068 10, 2.54 6.552107.167 1 log | * |t z HPG G (26) Table 14. Summary of the calculations needed to estimate the RAP HPG.

76 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction: Phase III Results Summary The PAM requires two primary material inputs, log |G*| at short-term aging and M, and the pavement temperature history. The pavement temperature history can be obtained from the EICM. Table 15 summarizes the different approaches by which the material inputs can be obtained. Virgin Mixtures log |G*| at Short-Term Aging Level 1 Aging loose mixture for 4 hours at 135°C in the oven followed by extraction, recovery, and testing of the binder. Level 2 Standard RTFO aging followed by testing of the binder. Level 3 Use of Equation (26). M Level 1 Loose mixture short-term aging and multiple long-term aging durations at 95°C in oven followed by extraction, recovery, and testing of the binder. M in Equation (1) is then optimized such that the predicted log |G*| values match the measured values. Recall that any laboratory, long-term aging duration can be considered to calibrate M as long as the duration provides binder log |G*| values that are well dispersed on the oxidation timescale. The aging durations should provide binder log |G*| values that belong to the constant region of the kinetics plot so that the oxidation kinetics obtained are meaningful and reproducible. Level 2 40-hour PAV aging of RTFO residue followed by binder testing. 40-hour PAV aging durations are equivalent to 6 days of loose mixture aging. Equation (25) can be used to generate more data beyond 6 days. M can be obtained by optimizing Equation (1) such that the predicted log |G*| values match the ones estimated from RTFO and PAV aging. Level 3 Use of Equation (27). RAP-Containing Mixtures Virgin parameters can be obtained as described above for virgin mixtures. RAP parameters can be obtained as described below. Equation (31) and Equation (32) should be used to obtain RAP-containing mixture parameters. log |G*| at Short-Term Aging of RAP Only Level 1 Aging loose mixture for 4 hours at 135°C in oven followed by extraction, recovery, and testing of the binder. Level 3 Use of Equation (26) if PG is known. If PG is unknown, refer to Figure 49 followed by Equation (26). M of RAP Only Level 1 Loose mixture short-term aging and multiple long-term aging durations at 95°C in oven followed by extraction, recovery, and testing of the binder. M in Equation (1) is then optimized such that the predicted log |G*| values match the measured values. Recall that any laboratory, long-term aging duration can be considered to calibrate M as long as the duration provides binder log |G*| values that are well dispersed on the oxidation timescale. The aging durations should provide binder log |G*| values that belong to the constant region of the kinetics plot so that the oxidation kinetics obtained are meaningful and reproducible. Level 3 Use of Equation (27) if PG is known. If PG is unknown, refer to Figure 49 followed by Equation (27). Table 15. Summary of methods to obtain material-specific inputs (log |G*| and M) for the PAM.