Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios (2020)

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.

79 Chapter 4 presents expanded laboratory performance results for binder blends and mixtures with high RBRs to explore the impact of higher recycling-agent doses than those used in the field projects with respect to the following issues: • Binder blend rheology with aging, • Binder blend aging prediction, • Recycling-agent characterization, • Mixture performance, and • Recycled binder availability. Appendix F provides additional data on binder blend aging prediction, and Appendix G pro- vides additional data on recycling-agent characterization. Additional data and details on mix- ture performance are included in Kaseer et al. (2018a). 4.1 Binder Blend Rheology with Aging The effects of recycling, aging, and rejuvenation on binder blend performance at low and intermediate temperatures are discussed in this section using DTc and the G-R parameter in Black space and respective existing thresholds tied to cracking resistance. Figure 47 illustrates the typical direction of the shifts observed in Black space with the inclusion of recycled materials, recycling agents, and aging considering binders without polymer modification. A new asphalt binder without polymer modification has a relatively low |G*| and high d, so it is located in the lower right corner of the Black space diagram. The inclusion of recycled materials (labeled “Recycling” in Figure 47) is reflected as an increase in |G*| and reduction in d, similar to the effect of laboratory and/or field aging. Conversely, considering rejuvenation as the partial reversal of the impact of aging on asphalt binders from a rheological standpoint, the inclusion of recycling agents is expected to reduce |G*| and increase d as an indication of improved ductility. The selected recycling-agent doses to match continuous PGH for the target climate were evalu- ated using various base binders, recycled materials, RBRs, and recycling-agent types, taking into consideration different target climates: PG 70–22 (TX), PG 64–22 (IN), PG 64–28 (NV and DE), and PG 58–28 (WI). Table 26, Table 27, Table 28, Table 29, and Table 30 summarize the com- ponents and characteristics of the recycled and rejuvenated binder blends evaluated in Black space (and corresponding recycled and rejuvenated asphalt mixtures discussed subsequently). Gray shading indicates field project combinations, and NV binder blends were not evaluated. In Table 26, Table 27, Table 28, Table 29, and Table 30, the DOT control blends refer to the recycled binder blends without recycling agent with a RAP/RAS binder content within the maximum allowable content per the different state DOT specifications (TX, IN, DE, and WI). C H A P T E R 4 Expanded Laboratory Performance of High RBR Binder Blends and Mixtures

Figure 47. Illustration of |G*| and c changing with recycling, aging, and rejuvenation in Black space. Blend/Mixture DOT Control (0.28 RBR) +0.5% WMA Rejuvenated (0.28 RBR) +2.7% T1 Rejuvenated (0.28 RBR) +6% T1 Rejuvenated (0.28 RBR) +6.5% A1 Binder PG 64-22 64-22 64-22 64-22 Binder Contenta 4.9% 4.9% 4.9% 4.9% RAP/RAS Contentb 10% RAP 5% RAS 10% RAP 5% RAS 10% RAP 5% RAS 10% RAP 5% RAS RBR 0.28 (0.1 RAP + 0.18 RAS) 0.28 (0.1 RAP + 0.18 RAS) 0.28 (0.1 RAP + 0.18 RAS) 0.28 (0.1 RAP + 0.18 RAS) Recycling-Agent Type and Dosec — 2.7% T1 6% T1 6.5% A1 Warm-Mix Additive Dosec 0.5% — — — NOTE: — = not applicable. aTotal binder in the mixture (base binder + recycled binders). bBy percentage of total weight of the mixture. cBy percentage of total binder in the mixture to match continuous PGH for target climate of PG 70-XX. Table 26. Characteristics of TX binder blends and asphalt mixtures. Blend/Mixture DOT Control (0.32 RBR) Recycled Control (0.42 RBR) Rejuvenated (0.42 RBR) +3.5% T2 Rejuvenated (0.42 RBR) +8% T2 Rejuvenated (0.5 RBR) +9.5% T2 Binder PG 58-28 58-28 58-28 58-28 58-28 Binder Contenta 5.8% 5.8% 5.8% 5.8% 5.8% RAP/RAS Contentb 28% RAP 2% RAS 16% RAP 8% RAS 16% RAP 8% RAS 31% RAP 4% RAS 40% RAP 4% RAS RBR 0.32 (0.25 RAP + 0.07 RAS) 0.42 (0.14 RAP + 0.28 RAS) 0.42 (0.14 RAP + 0.28 RAS) 0.42 (0.28 RAP + 0.14 RAS) 0.5 (0.36 RAP + 0.14 RAS) Recycling-Agent Type and Dosec — — 3.5% T2 8% T2 9.5% T2 NOTE: — = not applicable. aTotal binder in the mixture (base binder + recycled binders). bBy percentage of total weight of the mixture. cBy percentage of total binder in the mixture to match continuous PGH for target climate of PG 64-XX. Table 27. Characteristics of IN binder blends and asphalt mixtures.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 81 Blend/Mixture DOT Control (0.15 RBR) Rejuvenated (0.3 RBR) +2% T2 Rejuvenated (0.3 RBR) +2% A2 Rejuvenated (0.3 RBR) +3.5% T2 Rejuvenated (0.3 RBR) +5.5% A2 Binder PG 64-28 Pd 64-28 P 64-28 P 64-28 P 64-28 P Binder Contenta 5.0% 4.6% 4.6% 4.6% 4.6% RAP Contentb 15% RAP 33% RAP 33% RAP 33% RAP 33% RAP RBR 0.15 RAP 0.33 RAP 0.33 RAP 0.33 RAP 0.33 RAP Recycling- Agent Type and Dosec — 2% T2 2% A2 3.5% T2 5.5% A2 NOTE: — = not applicable. aTotal binder in the mixture (base binder + recycled binders). bBy percentage of total weight of the mixture. cBy percentage of total binder in the mixture to match continuous PGH for target climate of PG 64-XX. dPolymer-modified binder. Table 28. Characteristics of NV asphalt mixtures. Blend/ Mixture DOT Control (0.22 RBR) (PG 58- 28) Recycled Control (0.31 RBR) (PG 58- 28) Recycled 0.31 RBR (PG 52- 34) Rejuvenated (0.31 RBR) (PG 58-28) +1.2% V2 Rejuvenated (0.31 RBR) (PG 58-28) +5.5% V2 Rejuvenated (0.5 RBR) (PG 58-28) +9% V2 Binder PG 58-28 58-28 52-34 58-28 58-28 58-28 Binder Contenta 5.6% 5.4% 5.4% 5.4% 5.4% 5.4% RAP Contentb 27% RAP 36% RAP 36% RAP 36% RAP 36% RAP 58% RAP RBR 0.22 0.31 0.31 0.31 0.31 0.5 Recycling- Agent Type and Dosec — — — 1.2% V 5.5% V 9% V NOTE: — = not applicable. aTotal binder in the mixture (base binder + recycled binders). bBy percentage of total weight of the mixture. cBy percentage of total binder in the mixture to match continuous PGH for target climate of PG 58-XX. Table 29. Characteristics of WI binder blends and asphalt mixtures. Blend/ Mixture DOT Control (0.34 RBR) Rejuvenated (0.41 RBR) +0.8% T2 Rejuvenated (0.41 RBR) +0.8% T2 +0.25% WMA Rejuvenated (0.41 RBR) +8.5% T2 Rejuvenated (0.5 RBR) +10% T2 Base Binder PG 64-28 64-28 64-28 64-28 64-28 Binder Contenta 5.4% 5.4% 5.4% 5.4% 5.4% RAP/RAS Contentb 20% RAP 4% RAS 29% RAP 4% RAS 29% RAP 4% RAS 29% RAP 4% RAS 40% RAP 4% RAS RBR 0.34 (0.17 RAP + 0.17 RAS) 0.41 (0.24 RAP + 0.17 RAS) 0.41 (0.24 RAP + 0.17 RAS) 0.41 (0.24 RAP + 0.17 RAS) 0.5 (0.33 RAP + 0.17 RAS) Recycling-Agent Type and Dosec — 0.8% T2 0.8% T2 8.5% T2 10% T2 Warm-Mix Additive Dosec 0.4% — 0.25% — — NOTE: — = not applicable. aTotal binder in the mixture (base binder + recycled binders). bBy percentage of total weight of the mixture. cBy percentage of total binder in the mixture to match continuous PGH for target climate of PG 64-XX. Table 30. Characteristics of DE binder blends and asphalt mixtures.

82 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios These blends were regarded as the reference blends and compared to other blends of similar or higher RBR with recycling agent to evaluate the effectiveness of the recycling agents at the selected doses in improving the performance of the DOT control blend, and in facilitating the use of higher RBR than currently allowed by the DOTs. In Table 27, the rejuvenated binder blends (and asphalt mixtures) with the selected dose of recycling agent were designed to have a balanced combination of RBR from RAP and RAS. This was accomplished by maximizing RAP and minimizing RAS contents, while maintaining the same total RBR compared to the rejuve- nated binder blend (and asphalt mixture) with the field dose of recycling agent. Figure 48, Figure 49, Figure 50, and Figure 51 present the Black space diagrams for the recy- cled binder blends for TX, IN, WI, and DE materials with PG 70–22, PG 64–22, PG 64–28, and PG 58–28 target climates, respectively. The three data points represent RTFO, 20-h, and 40-h PAV aging. With aging, all the blends showed the expected increase in |G*| and decrease in d with aging, indicating loss of ductility. For the TX blends, the DOT control blend with 0.28 RBR shown in Figure 48 was located within the block cracking zone after 20-h PAV aging exhibiting very high |G*| and low d. The rejuvenated binder blends with 0.28 RBR and recycling agent had lower |G*| and larger d values compared to the DOT control, indicating restored ductility. However, the blends with the recy- cling agent at the selected dose to match continuous PGH for the target climate (6% T1 and 6.5% A1) showed the best reduction in |G*| and improvement in d. Considering DTc, all blends had poor DTc values (based on existing threshold of −5.0 after 20-h PAV aging), with the blends at the selected recycling-agent dose showing the highest (less negative) DTc values. For the IN blends shown in Figure 49, the DOT control blend with 0.32 RBR and the recycled control blend at higher 0.42 RBR were located within the transition zone after 20-h PAV aging and within the block cracking zone after 40-h PAV aging with high |G*| and low d. The rejuvenated binder blend with 0.42 RBR and recycling-agent dose close to the field dose (3.5%) had similar |G*| and d values, compared to the DOT control, after 40-h PAV aging and was located within the block cracking zone. The two rejuvenated binder blends with balanced recycled materials combinations (0.42 and 0.5 RBR) and the selected recycling-agent dose (8.0% and 9.5% T2) had lower |G*| and larger d values compared to the DOT control, and they were located within the transition zone. Figure 48. |G*| and c in Black space for TX binder blends (target PG 70–22 climate).

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 83 Figure 49. |G*| and c in Black space for IN binder blends (target PG 64–22 climate). Figure 50. |G*| and c in Black space for WI binder blends (target PG 64–28 climate).

84 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios This behavior indicated improved performance, with use of the recycling agent at the selected dose facilitating higher recycled materials contents compared to that of the DOT control. Con- sidering DTc, all blends had poor DTc values, with the blend with recycling agent at close to the field dose showing a higher (less negative) DTc value. DTc values for the binder blends with the recycling agent at the selected dose were not available. For the WI blends shown in Figure 50, the DOT control blend with 0.22 RBR and the recycled control blend at higher 0.31 RBR were located within the no block cracking zone after 20-h PAV aging, but within the block cracking zone after 40-h PAV aging. The recycled binder blend with 0.31 RBR with the softer base binder (PG 52–34) or the rejuvenated blend with the recycling agent at the field dose (1.2% V2) had slightly lower |G*| and higher d values, compared to the DOT control, after 40-h PAV aging. The two rejuvenated binder blends with 0.31 and 0.5 RBR and recycling agent at the selected dose (5.5% and 9% V2, respectively) had much lower |G*| and larger d values compared to the DOT control, with both blends located in the no block cracking zone after 40-h PAV aging. This again indicated that using the recycling agent at the selected dose facilitated the use of higher recycled materials contents compared to that of the DOT con- trol, with much better performance than using a softer base binder. Considering DTc, all blends had good DTc values (higher than those in the TX and IN binder blends), with the blends at the selected recycling agent dose and the blend with the softer base binder showing the highest (less negative) DTc values. Finally, the DE binder blends shown in Figure 51 illustrated that the recycled binder blend with 0.41 RBR and the field dose of recycling agent (0.8% T2) had similar |G*| and d values to those of the DOT control blend with 0.34 RBR, and both were located at the onset of the block cracking zone after 40-h PAV aging. The two recycled binder blends with 0.41 and 0.5 RBR and recycling agent at the selected dose (8.5% and 10% T2) had much lower |G*| and larger d values compared to the DOT control, and neither blend reached the block cracking zone after 40-h PAV aging. This again indicated that using the recycling agent at the selected dose facilitated the Figure 51. |G*| and c in Black space for DE binder blends (target PG 58–28 climate).

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 85 use of higher recycled materials contents compared to the DOT control. Considering DTc, all blends had good DTc values (higher than those in the TX, IN, and WI blends), with the blends at the selected recycling-agent dose showing the highest DTc values. 4.2 Binder Blend Aging Prediction An extensive binder aging experiment was conducted to gather oxidative aging kinetics data and resulting rheological changes for materials combinations shown in Table 1 under isother- mal oven aging at 60°C, 85°C, and 100°C over multiple temperature-specific durations. Then a diffusion-based oxidative aging model (Han 2011) was used to predict CA with depth and in-service time at the TX and WI field project locations from the following inputs: • Binder aging kinetics (Ea and A [Equation 18]) and hardening parameters (hardening suscep- tibility [HS] and hardening function constant [m]), • Binder initial CA at the beginning of constant-rate aging, • Predicted hourly pavement temperatures at the selected depth in the HMA layer at a specific location from the Temperature Estimate Model for Pavement Structure (TEMPS) (http:// www.arc.unr.edu/Software.html#TEMPS), and • Average representative AV radius and effective aging distance (e.g., binder film). Some additional data and details are included in Pournoman et al. (2018). These binder blend aging predictions were explored to capture the materials-specific influence of multiple components in each materials combination. Figure 52 provides an overview of how predicted CA [from the model based on changes in low shear viscosity (LSV)] and correspond- ing limiting CA values (with the model based on changes in G-R parameter) when G-R cracking onset and significant cracking thresholds are reached were utilized to generate predicted aging times when cracking is of concern. This procedure permitted the hypothetical relative compari- son of the binder blends for the selected field project locations using measured chemical and rheological binder properties. Appendix F provides additional details. The wide spectra of chemical and rheological measurements made in terms of tempera- tures and durations enabled a robust evaluation of the respective aging path of each materials Figure 52. Overview of binder blend aging prediction for selected field project locations.

86 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios combination with the prediction of both oxidation and rheological characteristics in service. The oxidation rate along this path was measured in terms of CA growth (CAg), and oxidation kinetics were modeled by Equation 18. [ ]( )= * − +−1 Equation 18CA M e k tg k t cf where CAg = carbonyl area growth (CA-CA0), CA = carbonyl area, CA0 = original or tank CA measurement (after RTFO aging for this study), M = initial jump, magnitude of fast rate reaction in terms of CAg, kf = fast rate of CAg, kc = slow or constant rate of CAg, and t = time, days. The duration of time at the fast rate is much shorter than that for the constant rate as tempera- ture increases, consistent with an Arrhenius reaction function as shown in Equation 19, and both oxidation kinetics parameters or rates (kf and kc) are represented as follows: [ ]( ) = α − Equation 19r k ork AP eCA f cg Ea RT where rCag = rate of carbonyl area growth (CAg), either kf or kc. A = pre-exponential factor. P = absolute oxygen pressure during oxidation, atm. α = reaction order with respect to oxidation pressure. Ea = activation energy, kJ/mol. R = ideal gas constant, 8.3144621 L/mol °K. T = temperature, °K. Figure 53 provides an example of constant oxidation rates (kc) as a function of the inverse of the aging temperature multiplied by the gas constant R for binder blends with the TX PG 64–22 base binder. Appendix F contains additional data for binder blends with four other base binders Figure 53. Constant-rate oxidation kinetics for TX PG 64–22 base binder blends.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 87 (NH PG 64–28, NV PG 64–28P, WI PG 58–28, and WI PG 52–34). Evaluation of these kc plots in Appendix F resulted in the following observations: • The addition of T1 to the TX PG 64–22 base binder did not change kc, but the rejuvenated blend with recycled materials and T1 at the low field dose (2.7%) did exhibit increased kc, possibly due to the low field recycling-agent dose that was unable to counteract the addition of RAP and MWAS. • The addition of T1 to the NH PG 64–28 rejuvenated blend at the dose to restore PGL (12.5%) for 0.5 RBR significantly reduced kc when compared to that of the NH PG 64–28 base binder or the rejuvenated blend with the lower dose (2.7%). Conversely, the rejuvenated blend with A1 and only RAP at 0.4 RBR did not show reduced kc relative to the base binder at the dose to restore PGL (6%). Thus, oxidation kinetics depend on both recycling-agent dose and type and recycled materials content and type. • The addition of T1 to the NV PG 64–28P base binder at the lower dose (2.7%) or the addition of T1 to the corresponding rejuvenated blend at the higher dose (11%) generally reduced kc when compared to the base binder and demonstrated that the use of recycling agent can facilitate the use of recycled materials. • The addition of V2 to the WI PG 58–28 rejuvenated blend at 0.31 RBR did not show reduced kc relative to the recycled control at 0.31 RBR and the DOT control at 0.21 RBR, but the softer rejuvenated blend with WI PG 52–34 and 0.31 RBR exhibited noticeably lower kc values, as expected. Figure 54 provides an example of the rheological response of the binder blends to CAg in terms of the G-R parameter for the TX PG 64–22 base binder. The slope of this plot is defined as the G-R/CAg HS. Similar plots are included in Appendix F for binder blends with the NH PG 64–28, NV PG 64–28P, WI PG 58–28, and WI PG 52–34 base binders. For the majority of the binder blends evaluated, the dual-slope model form previously used for fast and constant-rate kinetics provided a better representation of the measurements. Thus, similar to Equation 18 representing the kinetics, G-R/CAg HS is presented in the form of Equation 20 as follows: [ ]( )( ) ( )− = − + ′ + −− ′1 Equation 200ln G R M e k CA ln G Rk CA c gf g Figure 54. Glover-Rowe parameter at 15çC for PG 64–22 base binder blends.

88 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios where G-R = Glover-Row parameter (kPa) at 15°C and 0.005 rad/s. G-R0 = initial Glover-Row parameter (kPa) at 15°C and 0.005 rad/s. CAg = carbonyl area growth. Kf′ = fast rate of G-R growth. Kc′ = constant rate of G-R growth. In this model form with the G-R parameter, the long-term or constant-rate G-R/CAg HS region of the function is of primary interest since it is expected to represent the material late in its service life. Subsequently, evaluation of the plots in Appendix F resulted in the following observations: • The addition of the recycling agent to the TX PG 64–22 base binder increased the flexibility of the binder based on lower G-R parameter values at a given CAg during the early stages of the aging process. However, after a certain level of oxidation, the TX PG 64–22 base binder modified with T1 resulted in higher G-R parameter values, indicating increased stiffness and embrittlement relative to the base binder due to interaction with the recycling agent. The addition of recycled materials (RAP and MWAS) increased the stiffness and reduced the phase angle of the blend, as indicated by the higher G-R parameter. Further, the recycling agent was not effective in fully restoring the rheological properties at the low field dose (2.7%) for at least the early stages of aging. However, as the aging progressed, the T1 in the rejuvenated blend retained more flexibility and eventually resulted in the lowest G-R parameter, below that of either the base binder or the base binder with T1. • The addition of T1 to the NH PG 64–28 base binder at the lower dose (2.7%) increased the flexibility of the binder, resulting in lower G-R parameter values at a given CAg. In contrast to the results for the TX PG 64–22 base binder, the G-R/CAg HS (slope) was relatively unchanged by the addition of T1, but the recycling agent softened the blend. Similar to the results for the TX PG 64–22 base binder, the addition of recycled materials and T1 (even at the higher 12.5% dose to restore PGL) significantly increased the brittleness of the blend at 0.5 RBR, and like- wise reduced the G-R/CAg HS compared to that of the base binder. In a similar comparison, A1 had a similar influence on the NH PG 64–28 base binder at the dose to restore PGL (6%). However, when combined with RAP only at 0.4 RBR, the G-R/CAg HS for the rejuvenated blend with A1 was significantly reduced compared to the T1 blends with either base binder at different RBRs, possibly due to the fact that no RAS was present. • The addition of T1 to the NV PG 64–28P base binder slightly increased the flexibility of the binder at the low dose (2.7%), resulting in lower G-R parameter values at a given CAg. How- ever, the G-R/CAg HS of the two are similar for this combination compared to those for the other two base binders. When considering the addition of the recycled materials with T1 at the higher dose (11%), a similar response to that for the TX PG 64–22 binder blends was noted. In this case, the initial reduction in flexibility observed at higher G-R parameter values was eventually overcome by the reduced G-R/CAg HS with the rejuvenated blend. After some level of aging, the G-R parameter values of the rejuvenated blends were lower than those for either the base binder or the blend with only T1. • The addition of V2 to the WI PG 58–28 rejuvenated blend resulted in an increase in the flexi- bility of the blend at the same 0.31 RBR, as evidenced by lower G-R parameter values at a given CAg during the initial stages of aging. Then after a certain level of oxidation, the rejuvenated blend exhibited higher G-R parameter values, which indicated an increase in embrittlement when compared to the blends without recycling agent. • As a preliminary summary of these G-R/CAg HS relationships, the recycling agents evaluated reduced the overall stiffness in the initial stages of oxidation. However, differential aging

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 89 rates or G-R/CAg HS were observed between blends with only recycling agent added and rejuvenated blends (with recycling agent and recycled materials) for the TX PG 64–22, NH PG 64–28, and NV PG 64–28P base binders. These differences were not consistent with the type of RAS (i.e., MWAS or TOAS). Using the measurements of kinetics and LSV/CAg HS for the selected binder blends facilitated predictions through oxidation modeling of the changes in rheology over simulated in-service time for a given location and subsequent comparisons. Specifically, the model predicted the level of oxidation in terms of CAg expected in the different binder blends as a function of in-service time, which was then directly correlated to rheological properties afforded by the G-R/CAg HS relationships. While the previous comparisons used the G-R parameter determined at the stan- dard temperature of 15°C and G-R/CAg HS, the oxidation prediction models used LSV/CAg HS due to developed correlations between LSV and oxygen diffusivity of asphalt binders. Oxidative aging (CAg) predictions were completed over the analysis period at 0.01 m below the pavement surface using the predicted hourly pavement temperature at the corresponding depth for the TX and WI field project locations and the binder oxidation properties shown in Table 31. Binder ID Eaa (kJ.mol–1. °K–1) Pre- exponential Factorb, APα LSV/CAg HSc (1/CAg) md (ln(poise)) CARTFOe (arbitrary unit) kc Temperature (°C) 60 85 100 TXf Base (PG 64-22) 75 1.129E+09 11.31 46.91 0.323 0.002 0.013 0.035 Base (2.7%) T1 76 1.357E+09 13.28 0.12 0.637 0.002 0.012 0.034 Rejuvenated 0.50 RBR (PG 64-22) (2.7%) T1 64 4.049E+07 11.61 0.29 0.857 0.003 0.016 0.038 NHf Base (PG 64-28) 65 5.189E+07 9.46 201.22 0.131 0.003 0.018 0.042 Base (2.7%) T1 69 2.354E+08 8.92 6.55 0.420 0.004 0.022 0.056 Base (6%) A1 70 3.700E+08 8.16 123.22 0.107 0.004 0.021 0.053 Rejuvenated 0.50 RBR (PG 64-28) (12.5%) T1 70 1.400E+08 8.43 1.00e-05 2.446 0.001 0.009 0.022 Rejuvenated 0.40 RBR (PG 64-28) (6%) A1 73 8.101E+08 5.93 75.43 0.679 0.003 0.020 0.052 NVf Base (PG 64-28P) 73 5.931E+08 6.53 997.61 0.067 0.002 0.015 0.041 Base (2.7%) T1 79 4.709E+09 8.90 22.91 0.383 0.002 0.013 0.038 Rejuvenated 0.50 RBR (PG 64-28P) (11%) T1 86 5.045E+10 6.12 0.01 2.112 0.002 0.013 0.043 WIg DOT Control 0.22 RBR 81 1.0494E+10 9.15 5.453 0.288 0.002 0.014 0.041 Recycled 0.31 RBR 84 3.0189E+10 6.49 5.482 0.285 0.002 0.016 0.050 Recycled 0.31 RBR (PG 52-34) 103 8.4628E+12 9.74 6.260 0.295 0.0005 0.007 0.027 Rejuvenated 0.31 RBR (1.2%) V2 64 3.0880E+07 8.5 6.588 0.227 0.003 0.013 0.031 aEa: activation energy. bAPα: pre-exponential factor. cLSV/CAg HS: hardening susceptibility, based on LSV. dm: hardening function constant, based on LSV. eCARTFO: CA after RTFO aging. fTX: environment simulation. gWI: environment simulation. Table 31. Oxidative aging model parameters for the evaluated asphalt binders.

90 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios As an example, a summary of CAg predictions for the TX PG 64–22 binder blends over the analysis period at 0.01 m below the surface at the TX field project location is provided in Figure 55. Using the CAg predictions from all the selected base binder blends resulted in the estimated time to reach the G-R thresholds as a function of in-service time at the TX and WI field projects. These are hypothetical aging simulations using the selected TX and WI field project environ- mental conditions and the materials-specific binder blend aging characteristics, but only a few of these materials combinations were constructed as part of the TX and WI field projects (Table 8 and Table 11). Consequently, Figure 56, Figure 57, Figure 58, and Figure 59 provide a materials- specific comparison of the base binders from TX, NH, NV, and WI to evaluate the overall abil- ity of the oxidation kinetics and G-R/CAg HS relationships to identify the simulated time to reach the G-R parameter thresholds for cracking onset (i.e., 180 kPa) and significant cracking Figure 55. Carbonyl area prediction in mixture surface layer at TX field project. Figure 56. Simulated time to reach G-R = 180 kPa and G-R = 600 kPa in asphalt pavement for TX PG 64–22 base binder blends (in TX field project location).

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 91 (i.e., 600 kPa). These results and subsequent interpretations highlight the complex nature of the potential interactions between the different components in these binder blends. Considering the binder blends with TX PG 64–22 base binder, as shown in Figure 56, there was a systematic influence of the added components. The addition of T1 at the field dose soft- ened the blend and restored some of the flexibility as demonstrated by longer durations to meet the respective G-R parameter thresholds. Addition of the recycled materials (RAP and MWAS) resulted in a drastic reduction in the softening effect of the T1 and thus a reduction in the simu- lated durations to the G-R parameter thresholds to less than those for the TX PG 64–22 base binder. These data clearly indicate that the T1 field dose was inadequate. Figure 57. Simulated time to reach G-R = 180 kPa and G-R = 600 kPa in asphalt pavement for NV PG 64–28P base binder blends (in TX field project location). Figure 58. Simulated time to reach G-R = 180 kPa and G-R = 600 kPa in asphalt pavement for NH PG 64–28 base binder blends (in TX field project location).

92 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios Consideration of the NV PG 64–28P binder blends shown in Figure 57 resulted in similar findings for the respective influences of the various components. The exception noted with this polymer-modified blend was found with the cracking onset threshold of 180 kPa for the reju- venated blend, which was observed at a time similar to that for the NV PG 64–28P base binder, while the significant cracking threshold of 600 kPa was predicted to occur substantially after that for the base binder. Numerically, this was the result of the slightly lower oxidation rate (kc) combined with the lower G-R/CAg HS or flatter slope noted with the rejuvenated blends. These results provide examples of similar or improved performance for the rejuvenated blend with an effective T1 dose relative to the base binder. A nearly identical discussion for the NH PG 64–28 base binder results presented in Figure 58 with A1 and RAP at 0.4 RBR is also appropriate. An increase in the time to reach the G-R param- eter thresholds was observed when either T1 (at a low 2.7% dose) or A1 (at a higher selected dose to restore PGL) was added to the NH PG 64–28 base binder. The addition of recycled materials at 0.5 RBR to this base binder with T1 at the selected dose to restore PGL resulted in an increase in the time to reach the G-R parameter thresholds, thus demonstrating the efficiency of T1 when used at the higher selected dose. The addition of recycled materials at 0.4 RBR to this base binder with the A1 resulted in similar behavior to that of the base binder with only A1 at the same dose. Considerations of the diffusion limitation potential became even more prominent with the NH PG 64–28 base binder and respective rejuvenated blends with T1. In this case, both the kc and G-R/CAg HS for the rejuvenated blend were lower than those for the base binder with only T1 or even the NH PG 64–28 base binder, thus indicating a restricted oxidation path with these materi- als, both of which combined to simulate much slower oxidation and thus longer time durations to reach the G-R parameter thresholds. It is also important to note the differences in the doses between the blend of the NH PG 64–28 base binder and T1 only (i.e., at the low dose of 2.7%) compared to the selected T1 dose of 12.5% to restore PGL when TX RAP and TX TOAS were included. Despite the similarities in the dose of T1 as well as the TOAS in both the NH PG 64–28 and NV PG 64–28P blends, the influence of those components was not consistent. Figure 59 provides results for the WI binder blends in the WI field project location that is significantly milder in terms of temperature and other climatic parameters required as inputs Figure 59. Simulated time to reach G-R = 180 kPa and G-R = 600 kPa in asphalt pavement for WI binder blends (in WI field project location).

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 93 for the oxidation model relative to the TX field project location. After a 20-year simulation, none of the recycled blends with recycled materials (and either base binder) reached either of the G-R parameter thresholds, but the rejuvenated blend with V2 reached these thresholds after shorter durations. However, these observations are not consistent with the early field distress data col- lected, where all mixtures showed low-severity transverse cracking mainly as a result of cracks reflected from the existing underlying concrete slabs. Several reasons might have contributed to the observed discrepancies between laboratory and field data from the WI field project. For instance, neither the aging prediction model nor the G-R parameter thresholds take into consideration reflective cracking from an existing concrete layer. The G-R parameter thresholds of 180 kPa and 600 kPa were originally developed to con- trol cracking in the asphalt binder due to an increase in stiffness and a reduction in phase angle as a result of oxidative aging. Furthermore, these cracking thresholds were developed based on a specific binder aging to this level and not based on rheological changes due to modifica- tion processes such as polymer modification or the addition of recycled materials. Accordingly, the standard testing conditions of 15°C and 0.005 rad/s for G-R parameter determination may introduce a bias in the results when dealing with binders subjected to different modification processes. In addition, the aging predictions are based on assumed full blending between base and recycled binders, which is likely not representative of field-produced mixtures. Thus, while binder testing can help screen and preclude the use of materials combinations likely to perform inadequately in the field, it is still important to evaluate mixture performance and resistance to critical distresses. In summary, both T1 and A1 showed beneficial effects, with larger doses resulting in a more substantial benefit. However, this consistent trend was not shown for V2. This is not unexpected, considering the highly materials-specific interactions observed previously with these materials combinations. The simulated oxidative aging predictions generally provided the following conclusions: • The field dose of T1 (2.7%) in the TX PG 64–22 rejuvenated blend with 0.28 RBR mixture (0.1 RAP + 0.18 MWAS) was not sufficient to restore the binder blend to be rheologically similar to the base binder. • The NH PG 64–28 rejuvenated blend with a selected dose of 12.5% T1 to restore PGL and 0.5 RBR (0.25 RAP + 0.25 TOAS) almost restored the binder blend to be similar to the base binder. Similarly, the NH PG 64–28 rejuvenated blend with A1 at the selected dose of 6% to restore PGL and 0.4 RBR (0.4 RAP) exhibited slightly better restoration. These results imply that using a recycling agent at the selected dose can have a positive influence in restoring binder rheology. • The NV PG 64–28P rejuvenated blend contained 0.5 RBR (0.25 RAP + 0.25 TOAS) and a higher T1 dose of 11%. This dose not only restored the binder blend life prediction to that of the base binder but also improved it for almost an additional 6 years. • The WI PG 58–28 rejuvenated blend with 0.31 RBR (0.31 RAP) and 1.2% V2 exhibited a reduction in the time to reach the established cracking thresholds. Thus, the field dose of V2 may not be sufficient to restore binder rheology. The resulting durations to reach the G-R parameter thresholds for the base binders stem from the influence of the determined kc rates combined with the relative differences in G-R/ CAg HS. Specifically, there was a minor reduction in the kc values of the NV PG 64–28P base binder relative to the NH PG 64–28 and the TX PG 64–22 base binders (particularly at the lower temperatures, i.e., higher 1/RT values). Similar kc terms for the NH PG 64–28 and TX PG 64–22 base binders yielded modest differences in durations when compounded by the minor discrep- ancies noted in the G-R/CAg HS between the two. These conditions were further exemplified by the lower G-R/CAg HS for the NV PG 64–28P base binder resulting in the longest simulated

94 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios durations among these base binders. For the WI binder blends, different environmental con- ditions yielded longer durations despite the similarities in kc rates when compared with the other blends. 4.3 Recycling-Agent Characterization Throughout this study, results indicated that initial compliance with target PG require- ments is not sufficient to ensure long-term effectiveness of rejuvenated binder blends. PGH increases with the addition of recycled materials (or oxidation) in these blends, and recycling agents can be formulated to restore PGL (after 20-h PAV) for the target climate. Because phase angles of rejuvenated binder blends are relatively low, the PGI requirement is usually also met. However, cracking continues to be a problem for corresponding rejuvenated mixtures, especially at high RBRs. Restoring the rejuvenated binder blend to a predetermined stiffness is necessary, but not sufficient, to preclude cracking. Oxidative aging seriously restricts molecular motion, leading to a rapidly decreasing phase angle. Recycling agents must be evaluated based on their ability to restore rheology both in terms of decreasing stiffness and increasing phase angle at low and intermediate temperatures. In this study, Black space analysis and the calculated G-R parameter were utilized to rank recycling agents in binder blends initially and with aging, with the more effective recycling agent exhibiting a higher phase angle at any given |G*| (data point further to the right). To further explore the rejuvenating mechanism of different types of recycling agents while comparing the evolution of their effectiveness with aging, recycling-agent characterization experiments were conducted. The primary experiment evaluated both rheological and chemi- cal changes in rejuvenated binder blends after short- and long-term aging. Since aging of these blends may result in chemical changes in the recycling agent itself and subsequent reduction in the dispersive power of the maltene phase, a second experiment was completed to examine the effect of oxidation on recycling agents themselves. A third experiment was completed to compare binder blends after two different conditioning sequences: one with rejuvenation prior to aging (RTFO and 20-h PAV) and the other with aging prior to rejuvenation. Finally, corre- sponding mixture tests were completed for comparison with the binder blend results. Rejuvenated binder blends were prepared by combining the DE base binder (PG 64–28), DE RAP at 0.5 RBR, and one of seven recycling agents (aromatic extract A1, reacted bio-based oils B1 and B2, paraffinic oil P, tall oil T1, and modified vegetable oils V2 and V3). The dose of recycling agent in each blend was selected to match the continuous PGH required by climatic and traffic conditions at the DE field project. The rejuvenated binder blend with A1 required 13.5% recycling agent, the highest to match continuous PGH, followed by the binder blend with P at 11%, the blend with B2 at 10.5%, the blend with V2 at 9%, the blend with T1 at 8.5%, and the blends with B1 and V3 at 8%. These results are in agreement with the literature (Zaumanis et al. 2014), with petroleum-based recycling agents (including A and P types in this study) requiring higher doses than their bio-based counterparts (including T, V, and B types in this study). The rejuvenated binder blends and the pure recycling agents were subjected to four aging levels: • Unaged, • RTFO, • RTFO and 20-h PAV, and • RTFO and 40-h PAV. For each binder blend or recycling agent at each aging level, rheological evaluation in Black space and calculation of the G-R parameter and chemical analysis using FT-IR spectra were completed.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 95 The basic rejuvenation mechanism by the selected recycling agents is the addition of strongly polar compounds that help to polarize asphaltene clusters in recycled binders and compatibilize them with maltenes, thus breaking up the large asphaltene clusters. Recycling agents with these strongly polar compounds are classified as rejuvenators, and their addition reduces stiffness (|G*|) and increases phase angle (d), shifting the Black space plot to the bottom right. Recycling agents without these compounds, like paraffinic oil P, are classified as softening agents, which only decrease stiffness (|G*|) without sufficiently reducing phase angle (d). However, this also makes them the least susceptible to aging. The other recycling agents under consideration were rejuvenators that had strong polar groups including aromatics in A1, fatty acids in T1, a mixture of glycerides in V2 and V3, and glycerides stabilized through crosslinking and/or ester and amide bonds in B1 and B2. A Black space diagram for the aged blends is shown in Figure 60, with calculated G-R param- eters provided in Appendix G. Similar in effect to paraffinic-like recycled engine oil bottoms (REOBs), addition of the paraffinic oil P produced blends that are far more brittle with low phase angles compared to blends with the other six recycling agents. The corresponding G-R param- eter value of 197 kPa after 40-h PAV indicated this as the only rejuvenated blend that extended into the transition zone. This poor effectiveness exhibited by the P recycling agent is generally attributed to problematic compatibility between aromatic asphaltenes and the high concentra- tion of non-aromatic non-polar paraffins in the continuous phase of the binder. Paraffinic oils are not generally used as asphalt additives or recycling agents, and this result makes a clear case supporting previous concerns that excess paraffin concentrations may accelerate cracking even when aliphatic molecules might be non-crystalline at low temperatures. The bio-based recycling agents (including T, V, and B types in this study) led to rheological properties before and after aging that are slightly better than the aromatic oil traditionally used for rejuvenation. The two bio-oils (B1, B2) and the two modified vegetable oils had similar G-R values after 40-h PAV Figure 60. Black space diagram for binder blends with different recycling agents.

96 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios ranging from 42 kPa to 45 kPa, whereas the aromatic extract A1 was slightly higher at 55 kPa, and the tall oil exhibited more sensitivity to aging with a G-R value of 84 kPa. A Glover-Rowe long-term aging index was also calculated as the logarithm of the ratio of the G-R parameter for the blend after 40-h PAV to that after RTFO. The blend with T1 exhibited the highest aging index (1.84), followed by the blend with P (1.58); the blend with B1 (1.46); the blends with V2, V3, and B2 (1.38); and the blend with A1 with the smallest aging index (1.31). This aging index for the base binder (1.17) and that for the recycled blend with no recycling agent (0.82) were significantly less than those for the rejuvenated blends, but they were both shifted to the upper left corner in Black space because neither contained a recycling agent. FT-IR spectra provide a fingerprint for asphalt binder chemical functionality, with primary emphasis on the CA around 1710 cm–1 to define oxygen uptake during aging. However, bio- based recycling agents (including T, V, and B types in this study) as used for rejuvenation con- tain high concentrations of fatty acids, either alone or as part of mono-, di-, and triglycerides that create ester functionality in vegetable oils. Fatty acids may be further reacted or derivatized to change their chemical stability and rheological behavior. As the immediate chemical environ- ment changes, the carbonyl IR absorbance bands may shift somewhat, but the carbonyl itself will almost always remain even when the recycling agents have been chemically stabilized in some manner to resist further aging. Thus, most bio-based recycling agents (including T, V, and B types in this study) will contain high concentrations of carbonyl functionality even before any oxidative aging occurs. Depending on the source and any chemical reactions used to stabilize or upgrade the recycling agent, the carbonyl functionality may be in the form of fatty acids, esters, fatty anions, or even amides or imidazolines. The specific chemical functionality surrounding it will cause the energy of maximum IR absorbance for the carbonyl to shift somewhat, but IR absorbance for carbonyl usually remains within the range of 1,650 cm–1 to 1,820 cm–1. When blending 0.50 RBR blends with bio-based recycling agents (including T, V, and B types in this study), FT-IR spectra for unaged rejuvenated binder blends, as shown in Figure 61, reveal significant information even without further RTFO or PAV oxidation. However, it can Figure 61. FT-IR spectra for unaged binder blends with different recycling agents.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 97 be difficult to differentiate and quantify the overlapping peaks, especially for those blends with bio-based recycling agents (including T, V, and B types in this study). The oxidized carbonyl ketones in asphalt absorb IR light near 1,700 cm–1 and fatty acids absorb near 1,710 cm–1, caus- ing peaks to overlap with carbonyl peaks from asphalt oxidation. Ester peaks are formed when the fatty acids from bio-based recycling agents remain attached to glycerin, or when fatty acids are converted to esters through reactions with alcohols. The carbonyl groups in esters should show maximum IR absorbance near 1,750 cm–1, making them easier to quantify in the presence of oxidized asphalt than their fatty acid counterparts. Other bands are also valuable for differ- entiating the presence of bio-based recycling agents from the carbonyls from asphalt oxidation. In particular, carbon-oxygen bonds are present for both esters and fatty acids, but only in small amounts for oxidized asphalt binder. Additional FT-IR spectra for the aged binder blends are provided in Appendix G. In this study, CA and CAg were determined from the FT-IR spectra to capture chemical changes due to oxidative aging. This chemical analysis is complicated by the presence of competing carbonyl peaks from fatty acids and esters as found in bio-based recycling agents, but these could be sub- tracted out if comparable aged and unaged binder blends are available. To combine the chemical analysis and rheological response of the rejuvenated binder blends, G-R/CAg HS was defined and tabulated as shown in Appendix G. The G-R parameter was utilized instead of more traditional LSV to capture rheology in terms of both stiffness and embrittlement as indicated by phase angle. Figure 62 provides log G-R versus CAg with the slopes provided as G-R/CAg HS values for the aged rejuvenated blends, plus the base binder and corresponding recycled binder blend without recycling agent. As expected, the rejuvenated binder blend with P had a significantly higher G-R/CAg HS than the blends with all other recycling agents. This find- ing supports other evidence that the paraffinic oil P was least sensitive to oxidation as measured Figure 62. G-R/CAg HS (provided as slope of trendlines) for binder blends with different recycling agents.

98 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios by CAg, but the blend experienced substantial increases in G-R due to a rapid deterioration in compatibility between a saturate-rich maltene phase and increasingly larger and more polar asphaltenes. This parameter would likely show the same deficiencies for REOB modified bind- ers, given previous reports from Reinke (2015) and Reinke et al. (2015) noting dramatic losses in DTc when REOB was blended with asphalt binder at similar concentrations. The rejuvenated blends with various bio-based recycling agents (including T, V, and B types in this study) contain double bonds and therefore exhibit considerably more CAg than the blend with P. However, those same double bonds encourage molecular motion, leading to higher phase angles and lower G-R parameter values. For the bio-based recycling agents, the blend with tall oil T1 had a rela- tively high G-R/CAg HS due to a higher G-R parameter value. The other four bio-based recycling agents (V2, V3, B1, and B2) had almost identical G-R parameter values, but G-R/CAg HS values differed with varying CAg values. G-R/CAg HS provides a necessary, but not sufficient, parameter for evaluating recycling agents as it indicates the rate of rheological change with chemical oxidative aging. Location in Black space is also important with respect to cracking resistance in corresponding mixtures. For exam- ple, the blends with aromatic extract A1 and reacted bio-based oil B2 were almost equivalent to the base binder in terms of G-R/CAg HS, but their locations in Black space and G-R parameter value were different (Figure 60). In addition, the lowest G-R/CAg HS value was exhibited by the recycled blend without recycling agent due to significant previous aging, which is also shown by the highest G-R parameter values. All seven recycling agents were also evaluated individually for their sensitivity to aging. Com- plex viscosity (η*) was measured at 15°C and 10 rad/s by 50-mm DSR for all four aging levels (unaged, RTFO, RTFO and 20-hour PAV, RTFO and 40-h PAV) and then tabulated as shown in Appendix G, and an aging index was calculated by dividing the complex viscosity after RTFO and 40-h PAV by that of the unaged recycling agent. The chemically stable paraffinic oil P and the aromatic extract A1 had low aging indices of 1.09 and 1.15, respectively, indicating little oxidation. As expected, given many reactive double bonds that may oxidize or crosslink, the aging indices for the bio-based recycling agents were higher: 1.85 for V2, 2.60 for B2, and 2.88 for B1. The remaining two recycling agents (V3 and T1) had extremely high complex viscosi- ties after RTFO and 40-h PAV, suggesting these recycling agents had almost completely cross- linked to form harder resin-like materials. Furthermore, as these two recycling agents aged in the PAV, FT-IR indicated strong growth in a broad spectral region between 900–1,250 cm–1, with maximum changes noted near 1160 cm–1. Such changes were minor to nonexistent for the other recycling agents. Curiously, the corresponding binder blend with V3 showed no significant anomalies that would suggest similar crosslinking. Although the rejuvenated blend with T1 exhibited the high- est G-R parameter value of the blends with the five bio-based rejuvenating agents, there was no indication of any gel formation or unusual behavior for the blend with V3 during aging. Other bio-based recycling agents, most particularly those that had been chemically modified, did not show this tendency to gel when aged alone. One possible explanation for the observed behavior is that the asphalt binder contains high concentrations of natural antioxidants, such as phenols (Branthaver et al. 1993). Although these antioxidants do not inhibit oxidation mecha- nisms known to create benzylic carbonyls in asphalt binder, they should help to protect simple olefinic double bonds in the bio-based recycling agents. Further work is needed to understand the functionality responsible for these high molecular weight products formed when some pure bio-based recycling agents are aged, but these results suggest that recycling agents should be evaluated in binder blends rather than by themselves. After dose selection to match continuous PGH for the target climate and traffic conditions, the rejuvenated blend must be evaluated with respect to cracking resistance initially and with

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 99 aging. A third experiment was completed to compare the conditioning sequence for rejuve- nation and aging for the following two separate options in terms of Black space analysis and calculated G-R parameter: • Conditioning sequence #1: The rejuvenated binder blend was prepared by combining DE base binder (PG 64–28), DE RAP at 0.5 RBR, and one of seven recycling agents at doses to match continuous PGH. Each blend was then subjected to 20-h PAV aging. • Conditioning sequence #2: The recycled binder blend was prepared by combining DE base binder (PG 64–28) and DE RAP at 0.5 RBR. This blend was then subjected to 20-h PAV and then back-blended with one of the same seven unaged recycling agents at the same doses used in Conditioning sequence #1. A Black space diagram for the blends from both conditioning sequences is shown in Figure 63, with calculated G-R parameters shown in Figure 64. In Figure 63, the lines on the bottom right represent Conditioning sequence #1, where the unaged recycled blend was combined with each recycling agent and then aged. The lines on the upper left represent Conditioning sequence #2, where the aged 0.50 RBR blend was then rejuvenated with the same dose of each of the seven unaged recycling agents. These results were surprising, with all binder blends in which the recy- cling agent was aged with the binder blend exhibiting lower stiffnesses and higher phase angles than corresponding blends with unaged recycling agent. Figure 64 graphically shows the scale of this difference by comparing the G-R parameter values between the two conditioning sequences for all seven recycling agents. In spite of the fact that the recycling agents were not aged for binder blends subjected to Conditioning sequence #2, Figure 63. Black space diagram for different blends with different conditioning sequences.

100 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios their G-R parameter values were much higher (typically double) than the G-R parameter values of their counterpart blends in which the recycling agents were aged as part of the conditioning. One explanation for this unexpected behavior could be that the recycling agent is consuming some of the available oxygen during PAV aging, but the rheological consequences of these oxidation products were not as great as forming benzylic carbonyl on highly aromatic asphalt molecules, as discussed subsequently. The effectiveness of each recycling agent with aging is represented by low G-R parameter values after Conditioning sequence #1 with 20-h PAV, and these results are in agreement with the effectiveness rankings after 40-hour PAV discussed previously. After Conditioning sequence #1, binder blends with all five bio-based recycling agents had lower G-R parameter values within the range of 11 kPa–16 kPa, while the blend with aromatic extract A1 had a higher value (22 kPa), and the blend with paraffinic oil P exhibited an even higher value (53 kPa). To better understand the unexpected reversal in G-R parameter values for the two condition- ing sequences, CAg was monitored for the rejuvenated binder blends subjected to the two differ- ent conditioning sequences, with results shown in Figure 65. These results contradicted those for the G-R parameter with CAg indicating more oxygen uptake when the recycling agent was aged with the blend in Conditioning sequence #1 and the G-R parameter suggesting less embrittle- ment. This was true even for the aromatic extract A1, which should oxidize via mechanisms similar to those for asphalt binder, and the paraffinic oil P, which takes up almost no oxygen on its own. For this disparity to occur, the G-R/CAg HS had to be much lower when the recycling agent was aged with the blend in Conditioning sequence #1 (more oxygen uptake, less damage) as opposed to when adding the recycling agent after the recycled blend was aged in Conditioning sequence #2. That is, when the recycling agent was present during aging, the damaging impact of each carbonyl-based oxygen atom on the G-R parameter was greatly reduced. In addition, rank- ings for carbonyl uptake in the rejuvenated blends were generally consistent with rankings for the recycling agents characterized by themselves. Some of the disparity in oxygen uptake might be attributed to the fact that the rejuvenated binder blend was much softer and less brittle before aging, so oxygen diffusion into the binder blend could be higher. However, oxygen diffusion during aging cannot explain the unexpected consequences on rheology. Although CAg is typically tied to increases in binder stiffness, results in this study suggest that the oxygen uptake versus binder embrittlement in terms of G-R/CAg HS may change significantly Figure 64. G-R parameter for different blends with different conditioning sequences.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 101 when a recycling agent is added to a recycled binder. If dissolved oxygen in the asphalt really can be diverted to other reactions, rather than forming the highly damaging benzylic carbonyl, there could be additional benefits from recycling agents beyond the initial impact on rheology. The results presented in this section focused on providing a better understanding of concepts linking chemical changes during oxidation to resulting rheological response of binder blends with aging. Findings include the following: • Because recycling agents may be rich in carbonyl content before oxidation, FT-IR analysis for aging must focus on growth in the carbonyl peak area (CAg) between two aging conditions, not on CA at the final aging state. • During PAV aging, recycling agents seem to change the pathway for at least some oxidation reactions within the recycled binder blend. That is, in the presence of a recycling agent, more carbonyls are formed in the aging blend, but the oxidative impact on key rheology indica- tors tied to cracking seem to be ameliorated. Such trends can be quantified by evaluating the change in G-R/CAg HS when recycling agents are blended with high RBR binders before aging. Recycling agents that lower G-R/CAg HS without significantly increasing total oxygen uptake should have better cracking resistance. • The rejuvenation mechanism of recycling agents varies by recycling-agent type. The paraffinic oil P was included in this study as a presumed poor recycling agent. Blends with P had no problem satisfying target PG grades, but all evidence places this potential recycling agent in a category by itself as being the poorest performer that serves only as a softening agent, not a rejuvenator. Even though the paraffinic oil P showed almost no oxygen uptake on aging, a lack of compatibility with increasingly structured oxidized aromatics killed molecular motion, leading to extraordinarily low phase angles, high G-R parameter values, and marginal to failing performance predictions. These same problematic issues pervade indus- try concerns that high concentrations of REOB in binder blends lead to premature cracking. REOB is likewise a highly aliphatic oil that appears unable to stabilize asphaltenes in highly aged (40-h PAV) binders. Figure 65. Carbonyl area growth in different blends with different conditioning sequences.

102 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios The aromatic extract A1 has historically been the gold standard for petroleum-based recy- cling agents. Often promoted as replacing the aromatics and resins lost to oxidation, these recycling agents can maintain compatibility with oxidized asphaltenes under much more rig- orous aging conditions than their paraffinic counterparts. For this study, A1 served as a known control. Within appropriate limits for base binder quality and RAP quality and proportion, aromatic extracts can work well as recycling agents. However, based on the results from the TX field project, these recycling agents cannot restore cracking resistance when combined at high RBRs with low-quality binder blends and highly aged recycled materials including RAP and RAS. When viewed in the context of restoring binder rheology in Black space, A1 was much better than P. When used with appropriate constraints as proposed subsequently, A1 can be an effective recycling agent. Nonvolatile vegetable oils and reacted bio-based oils offer a very different path to asphalt compatibilization and rejuvenation. There are no aromatics to compatibilize the asphaltenes in aged recycled binders. Instead, these recycling agents act almost like an emulsifier, with the highly polar carbonyl groups on the molecules interacting with polar sites on asphaltene agglomerates, while the less polar olefinic chain remains in the binder’s mobile phase. The double bonds on the olefin chain increase molecular motion and lower the glass transition of the mobile phase. The result is an increased phase angle and better cracking resistance. With aging, the benefits of the double bond may be taken away through oxidative aging of the recycling agent itself. However, this oxidation does not always have the expected negative impact on rheological parameters tied to cracking. When vegetable or reacted bio-based oils are present during aging, much of the consumed oxygen goes to reducing the G-R/CAg HS of the rejuvenated blend, rather than reducing the phase angle. That is, each oxygen atom react- ing with a bio-based molecule has much less impact on G-R because these carbonyls do not lead to the type of asphaltene agglomerations that inhibit stress relaxation. Some oxidation of the double-bond sites on the recycling agent molecule could even be helpful by creating more compatibility with asphaltenes through polar interactions. Certain vegetable oil and reacted bio-based oil recycling agents have been chemically stabilized to further reduce the impact of long-term aging. Among the seven recycling agents included in this study, the modified vegetable oils V2 and V3 and the reacted bio-based oils B1 and B2 consistently showed the lowest G-R parameter values after PAV aging of rejuvenated blends. Even though each of these recycling agents elevates the carbonyl content as measured by FT-IR, the extra oxygen seems to have a positive impact on ultimate cracking resistance. Pure vegetable oils are usually edible, and flash points are high, so no safety concerns are reported. Although not included in this study, saturated fatty acid chains with no double bonds, such as steric or palmitic acid, have high melting points and behave like waxes. Even when substituted on larger triglyceride molecules, saturated fatty acids will crystallize and thereby damage rheo- logical properties needed for cracking resistance. Thus, not all reacted bio-based oils should be used as recycling agents. Tall oil T1 is a pine tree by-product from paper production. Given its moderate olefin content, the cracking resistance of aged blends with T1 was generally better than that for blends with the aromatic extract A1, but typically not as good as the vegetable oils and reacted bio-based oils. In addition, the molecular weight is lower, so T1 may also be slightly volatile. • The current classification system for recycling agents is best described by ASTM D4552. This specification is based on kinematic viscosity at 60°C, flash point, saturates content, and viscos- ity ratio with short-term aging (TFO or RTFO). As discussed previously, the aging index based on complex viscosity for the recycling agent itself could be highly misleading, and this speci- fication for recycling agents does not include critical aged rheological properties for the reju- venated blend. Based on these results, evaluation tools are proposed for rejuvenated blends with long-term aging in Black space instead to capture the different rejuvenating mechanisms

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 103 used to restore rheology by different types of recycling agents that may be listed on a qualified products list. Mixture characterization tests including I-FIT and HWTT were also conducted for com- parison to the corresponding binder blends. Results shown in Figure 66 and Figure 67 illus- trate similar rankings, with the reacted bio-based oils B1 and B2 and the modified vegetable oil V3 exhibiting superior performance in terms of FI, and the reacted bio-based oil B1 and the tall oil T1 exhibiting superior performance in terms of rutting resistance. For both mix- ture tests, the paraffinic oil P exhibited relatively poor performance, especially in terms of rutting resistance. 4.4 Mixture Performance The effects of recycling, aging, and rejuvenation were evaluated on asphalt mixture perfor- mance in terms of stiffness/rheology by MR and mixture Black space analysis with |E*| and G-Rm, intermediate-temperature cracking resistance by FI and CRI, low-temperature rheology and cracking resistance by BBRm results and CRIEnv, and rutting resistance by N12.5. Similar to binder blends, the selected recycling-agent doses to match continuous PGH for the target climate were evaluated using various base binders, recycled materials, RBRs, and recycling- agent types, taking into consideration different target climates: PG 70–22 (TX), PG 64–22 (IN), PG 64–28 (NV and DE), and PG 58–28 (WI). Table 26, Table 27, Table 28, Table 29, and Table 30 summarize the components and characteristics of the asphalt mixtures evaluated. Gray shading indicates field project combinations. Similar to the evaluation of binder blends, the DOT control mixtures were regarded as the reference mixtures. Therefore, DOT control mixtures were compared to other mixtures of simi- lar or higher RBR with recycling agent to evaluate the effectiveness of the recycling agents at the Figure 66. FI results for mixtures with different recycling agents.

104 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios selected doses in improving the performance of the DOT control mixtures, and in facilitating the use of higher RBR than currently allowed by the DOTs. Results presented in this section are for LMLC specimens after STOA and LTOA. For MR, FI, and CRI, the darker shade stacked column represents the results after STOA, and the hatched lighter-shade stacked column represents the results after LTOA. The error bars in each column represent ± one standard deviation from the average value based on the replicate measurements, and the letters inside each column represent Tukey’s honestly significant difference (HSD), in which mixtures not connected with the same letter are considered significantly different. For the |E*| and BBRm tests, Black space diagrams are presented, where the unfilled symbols represent the STOA specimens and the filled symbols represent the LTOA specimens. 4.4.1 Stiffness/Rheology (MR, Intermediate-Temperature Mixture Black Space with |E*| and G-Rm) As Figure 68 indicates, MR results show that adding the selected dose of recycling agent to match continuous PGH for the target climate is more effective than adding the field dose in producing asphalt mixtures with statistically lower stiffness to that of the DOT control mixture with similar RBR (TX), or producing asphalt mixtures with statistically similar or lower stiffness to that of the DOT control mixtures with lower RBR (IN, NV, WI, and DE). To explore beyond mixture stiffness, mixture Black space (log of |E*| versus phase angle) diagrams were used to qualitatively examine the relative location of mixtures with recycling, aging, and rejuvenation similar to the binder/binder blend analysis. In these diagrams, unfilled symbols represent the STOA condition, while filled symbols represent the LTOA condition. As for binders, with aging or the addition of recycled materials, the mixture rheological state moves from the lower right corner to the upper left corner of the plot, which indicates an increase in stiffness with a corresponding reduction in phase angle. With the addition of recycling agent for Figure 67. HWTT results for mixtures with different recycling agents.

Figure 68. MR test results.

106 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios rejuvenation, mixtures are expected to move toward the lower right corner, which indicates a reduction in stiffness with a corresponding increase in phase angle. The magnitude and slope between two points in mixture Black space were also determined to quantitatively evaluate the change in rheological properties with aging, as defined in Equation 21 and Equation 22, respectively: ( ) [ ]( )= φ − φ + * − * Equation 211 2 2 1 2 2Magnitude Log E Log Econd condo cond cond [ ]= * − * d − d Equation 221 2 1 2 Slope Log E Log Econd cond cond cond where φcond 1 = Condition 1 phase angle, φcond 2 = Condition 2 phase angle, |E*|cond 1 = Condition 1 dynamic modulus, and |E*|cond 2 = Condition 2 dynamic modulus. Generally, a smaller magnitude and steeper slope (less loss of phase angle) is desirable with aging. G-Rm was also calculated and plotted in bar charts, with comparisons made against the refer- ence DOT control mixture with a lower value desirable for better cracking resistance. In the bar charts, error bars represent + one standard deviation of average calculated values, and the letters inside each column represent Tukey’s HSD statistical analysis results, with mixtures connected with the same letter statistically similar. The Black space plot for the IN mixtures is shown in Figure 69, and the corresponding G-Rm results are shown in Figure 70. Both of these analyses indicate that the use of recycling agent at the dose to match continuous PGH for the target climate facilitates increasing RBR (to 0.4 or 0.5), even with LTOA, with statistically equivalent performance (and similar slope) to that of the DOT control mixture with lower RBR despite longer aging paths (larger magnitude). After STOA, a lower dose of recycling agent was also effective. The Black space plot for the WI mixtures is shown in Figure 71, and the corresponding G-Rm results are presented in Figure 72. Both of these analyses show that for both aging conditions, the use of the lower field dose of recycling agent (1.2% V2) with higher 0.31 RBR resulted in rheological performance similar to that of the DOT control mixture at lower 0.22 RBR, but the use of the softer base binder (PG 52–34) resulted in improved performance with a shorter aging path (smaller magnitude) and steeper slope. At the recycling-agent doses to match continuous PGH for the target climate (5.5% for 0.31 RBR, 9% for 0.5 RBR), the resulting mixtures were even softer and less brittle (higher phase angle) with longer aging paths (larger magnitude) and steeper slopes compared to those of the DOT control mixture. Again, the use of recycling agents at the dose to match continuous PGH facilitated the use of higher RBRs. The Black space plot for the DE mixtures shown in Figure 73 and the corresponding G-Rm results presented in Figure 74 concur with those from the other field projects. Both of these analyses show that for both aging conditions and with or without the WMA additive, the use of the lower field dose of recycling agent (0.8% T2) with higher 0.41 RBR resulted in rheo- logical performance similar to that of the DOT control mixture at lower 0.33 RBR. At the recycling-agent doses to match continuous PGH for the target climate (8% for 0.41 RBR, 10% for 0.5 RBR), the resulting mixtures with similar slopes were significantly softer and less

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 107 brittle (higher phase angle) with longer aging paths (larger magnitude) compared to those of the DOT control mixture. 4.4.2 Intermediate-Temperature Cracking Resistance (FI) Figure 75 indicates that adding the selected dose of recycling agent is more effective than the field dose in producing asphalt mixtures with statistically higher FI than that of the DOT Figure 69. |E*| test results in mixture Black space for IN mixtures. Figure 70. G-Rm results for IN mixtures.

108 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios control mixture with similar RBR (TX), or producing high RBR asphalt mixtures with sta- tistically similar or higher FI to that of the DOT control mixtures with lower RBR (IN, NV, WI, and DE). Similarly, Figure 76 for the CRI values demonstrates that adding the selected dose of recy- cling agent was adequate in producing asphalt mixtures with statistically higher CRI values than those of the DOT control mixture with similar RBR (TX), or producing asphalt mixtures with statistically similar or higher CRI values than those of the DOT control mixtures with lower RBR (IN, NV, WI, and DE). Figure 72. G-Rm results for WI mixtures. Figure 71. |E*| test results in Black space for WI mixtures.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 109 4.4.3 Low-Temperature Cracking Resistance (Low-Temperature Mixture Black Space with Sm and m-valuem, CRIEnv) The BBRm test and the UTSST were used to explore low-temperature mixture cracking resis- tance as data were available for different materials combinations from the five field projects. Low-temperature Black space is shown in Figure 77, along with the cracking thresholds devel- oped by Romero (2016) based on field performance in Utah. These BBRm results consistently indicate that adding the selected dose of recycling agent was adequate in producing asphalt mixtures with similar or lower creep stiffness (S) and similar or higher relaxation (m-values) to the DOT control mixture with similar or lower RBR. Figure 73. |E*| test results in Black space for DE mixtures. Figure 74. G-Rm results for DE mixtures.

Figure 75. FI results.

Figure 76. CRI results.

Figure 77. BBRm test results (unfilled symbols represent STOA specimens while filled symbols represent LTOA specimens).

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 113 UTSST modulus and CRIEnv results for NV LMLC mixtures at different levels of aging are pre- sented in Figure 78 and Figure 79, respectively. The moduli variation with decreasing tempera- ture shows the low-temperature behavior of the mixture with a focus on viscous softening, crack initiation, and fracture for this study, while the CRIEnv provides an overall index that simultane- ously takes these different aspects of behavior into account. For all mixtures at high 0.33 RBR, the effect of aging was seen by a shortening of the viscous softening plateau and a shift to warmer crack initiation and fracture temperatures. Similar effects of recycling were also shown when comparing the DOT control mixture to the recycled mixture at the higher 0.33 RBR at the same aging level. The effect of rejuvenation is opposite to that of aging, with a rotation counterclock- wise and resulting decreased stiffness and colder crack initiation and fracture temperatures, and for both aging levels, this effect was illustrated for the mixture with A2 with a more pronounced Figure 79. CRIEnv results for NV LMLC mixtures. Figure 78. UTSST modulus curves for NV LMLC mixtures.

114 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios effect after LTOA. These results were confirmed by CRIEnv values, although all three mixtures at high 0.3 RBR (both the recycled mixture and both rejuvenated mixtures) exhibited inadequate low-temperature cracking resistance based on the proposed threshold after LTOA. UTSST modulus and CRIEnv results for WI LMLC mixtures at different levels of aging are pre- sented in Figure 80 and Figure 81, respectively. For all mixtures, the effect of aging was seen by a shortening of the viscous softening plateau and a shift to warmer crack initiation and fracture temperatures. Similar effects of recycling were also shown when comparing the DOT control mixture to the recycled mixture at higher 0.31 RBR at the same aging level. The effect of reju- venation is opposite to that of aging, with a rotation counterclockwise and resulting decreased stiffness and colder crack initiation and fracture temperatures, and for both aging levels, this effect was illustrated for the rejuvenated mixtures and the mixture with a softer base binder with Figure 80. UTSST modulus curves for WI LMLC mixtures. Figure 81. CRIEnv results for WI LMLC mixtures.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 115 a more pronounced effect after LTOA. These results were confirmed by CRIEnv values, although the higher recycling-agent dose provided sufficient low-temperature cracking resistance even after LTOA. 4.4.4 Rutting Resistance Figure 82 and Figure 83 show results for the HWTT tests (in wet condition), which dem- onstrated that WI asphalt mixtures (with the soft PG 58–28 base binder) and DE asphalt mixtures Figure 82. HWTT and APA test results for WI mixtures. Figure 83. HWTT and APA test results for DE mixtures.

116 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios (with PG 64–28 base binder), all with the recycling agent at the selected dose, passed the mini- mum rutting requirements. In similar climates, mixtures with a target PG 58-XX and PG 64-XX climate are required to sustain at least 5,000 and 7,500 load cycles, respectively, before achieving 12.5 mm rut depth. As expected, APA results (in dry condition) demonstrated improved rutting resistance when water was not present. HWTT and APA test results confirmed that the doses to match continuous PGH of the target climate were not excessive in terms of being detrimental to the rutting performance of the asphalt mixtures. 4.5 Recycled Binder Availability The amount of RAP binder in the mixture is typically represented as RBR. However, the quantity of effective RAP binder in the mixture is usually unknown, which raises concerns due to its ultimate effect on performance. The term effective RAP binder refers to the binder that is released from the RAP, becomes fluid, and blends with the base binder under typical mixing temperatures. Other terms used in the literature include RAP binder contribution, RAP binder activation, degree of RAP activation, RAP working binder, and RAP binder availability, as used in this report. RAP binder availability is typically addressed through one of three assumptions: • 0% availability, where the RAP acts as a black rock; • 100% availability, where all the RAP binder becomes fluid and is available to blend with the base binder; and • Partial availability, where a portion of the RAP binder becomes fluid and is available to blend with the base binder. Although rarely measured, it is generally accepted that the third assumption is more realistic (McDaniel and Anderson 2001; D’Angelo et al. 2011). However, in a recent survey discussed in NCHRP Synthesis 495: Use of Reclaimed Asphalt Pavement and Recycled Asphalt Shingles in Asphalt Mixtures (Stroup-Gardiner 2016), 77% of the responding state highway agencies use the second assumption and consider 100% RAP binder availability; thus, they reduce the base binder content in the mixture by the RAP binder content. About 6% of the respondents in this same survey use the first assumption and consider 0% RAP binder availability, and approxi- mately 17% use the third assumption and consider partial RAP binder availability, assuming around 75% of the RAP binder is available. Designing mixtures with the assumption of 100% availability could result in mixtures with less total binder content than the selected optimum from the mix design. In this case, coatability issues may arise, resulting in a dry mixture with a high AV content with increased potential for cracking, raveling, or moisture damage. Con- versely, designing mixtures with the assumption of 0% availability could result in mixtures with potential rutting problems, due to possible excessive total binder content. 4.5.1 Methodology D’Angelo et al. (2011) investigated the extent of RAP binder availability using the aggregate size exclusion method. In this method, the RAP occupies a designated size in the mixture dif- ferent from that of the virgin aggregates. After mixing with the base binder, the RAP can be separated from the virgin aggregate, which facilitates evaluation of the binder contents for both materials. If the RAP has a higher binder content than the virgin aggregate, then the RAP binder is not fully available to blend with the base binder. In this case, most of the RAP acts like a black rock and the base binder coats the RAP as it does any other aggregate particle.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 117 In a similar manner, the following methodology was developed and demonstrated in Phase 2 to estimate the RAP binder availability based on an evaluation of mixtures with specific sizes of virgin aggregate and RAP: 1. Prepare the virgin mixture using: a. Base binder. b. Virgin aggregate with three distinct fractions: a coarse size (passing the ½-in. sieve and retained on the ⅜-in. sieve), an intermediate size (passing the ⅜-in. sieve and retained on the No. 4 sieve), and fine sizes (a combination of material passing the No. 4 sieve and retained on the No. 8 sieve and passing the No. 8 and retained on the No. 30 sieve). 2. Condition the loose mixture in the oven for 2 h at 135°C to simulate short-term aging. 3. Sieve the loose mixture to separate the coated particles into the different sizes while both the mixture and the sieves are hot enough to allow separation. 4. Determine the binder content of each fraction using the ignition oven per AASHTO T 308 and label the binder content of the intermediate size fraction (passing the 3/8-inch sieve and retained on the No. 4 sieve) as Reference Pb. 5. Prepare the RAP mixture using: a. Base binder. b. Virgin aggregate with two distinct fractions: a coarse size (passing the ½-in. sieve and retained on the ⅜-in. sieve) and fine sizes (a combination of material passing the No. 4 sieve and retained on the No. 8 sieve and passing the No. 8 and retained on the No. 30 sieve). c. RAP of intermediate size (passing the ⅜-in. sieve and retained on the No. 4 sieve). 6. Repeat Steps 2 through 4 and label the binder content of the intermediate size fraction of RAP (passing the ⅜-inch sieve and retained on the No. 4 sieve) as RAP′ Pb. Figure 84 provides an illustration of the proposed methodology. The binder contents of the coated RAP (RAP′ Pb) and coated virgin aggregate (Reference Pb) provide significant insight into the amount of RAP binder that is active and available. To illustrate the methodology, consider an example of a virgin mixture consisting of base binder and virgin aggregate with distinct fractions with the percent retained for each fraction (by weight of total aggregate) of 28% (⅜ in.), 30% (No. 4), 28% (No. 8), and 14% (No. 30). The total binder content of this mixture is 4.5%. The measured binder contents for each sieve size by ignition oven are 2.7%, 4.0% (Reference Pb), and 6.1% for sieves ⅜ in., No. 4, and No. 8 + No. 30, respectively, with the coarse aggregate size absorbing less binder than the intermediate and fine aggregate sizes due to smaller surface area (Brown et al. 2009). The Reference Pb value is only valid for this particular mixture, with its specific total binder content and virgin aggregate type and gradation. When using RAP (with a 4.5% binder content) to prepare a RAP mixture with 0.3 RBR (i.e., 30% RAP binder and 70% base binder) and a total binder content the same as in the virgin mixture (4.5%), the total binder content consists of 3.15% base binder (70%) plus 1.35% RAP binder (30%). Therefore, the base binder contents in each sieve size of aggregate should be close to 70% of the values measured in the virgin mixture with 100% base binder content (i.e., 1.9% [3/8 in.], 2.8% [No. 4], and 4.3% [No. 8 + No. 30]). These values were confirmed by preparing the same virgin mixture but with 3.15% binder content and determining the binder content for each sieve using the ignition oven. The addition of the RAP binder should bring the total binder content for each sieve size of aggregate to 2.7% (3/8 in.), 4.0% (No. 4), and 6.1% (No. 8 + No. 30). In this RAP mixture, the RAP′ Pb (binder content of RAP retained on the No. 4 sieve) is mea- sured by ignition oven, and the following three outcomes are plausible depending on how much RAP binder is active or available: Scenario 1: RAP′ Pb = Reference Pb = 4.0% in this example.

118 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios (a) (b) Figure 84. (a) Summary of the proposed method and (b) possible scenarios for RAP binder availability.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 119 The coated RAP particles in the RAP mixture have the same binder content as the coated vir- gin aggregate particles on the No. 4 sieve in the virgin mixture. This would imply that the RAP binder is fully released, and completely active and available in the mixture, and the total binder blend (base and RAP binders) was evenly distributed within the mixture. This scenario would represent 100% RAP binder availability, as illustrated in Figure 84b. Scenario 2: RAP′ Pb = [1-RBR]*Reference Pb + RAP binder content = 7.3% in this example. The coated RAP particles in the RAP mixture have significantly more binder content than the coated virgin aggregate particles on the No. 4 sieve in the virgin mixture, and this differ- ence is equal to the RAP binder content. This would imply that the RAP binder is acting as a black rock and the RAP binder is not available in the mixture. In other words, only the base binder was evenly distributed within the mixture (between the virgin aggregate and the RAP). This scenario would represent 0% RAP binder availability, as illustrated in Figure 84b. In this example, since the contribution from the base binder equals 2.8% (at 70% of the total binder, as calculated and verified previously when only the base binder is available), the RAP′ Pb will approach 7.3% (2.8% + 4.5%). Again, this value is only valid for these particular mixtures, with their specific virgin aggregate type and gradation, RAP binder content, and the total binder content in the mixture. Scenario 3: Reference Pb < RAP′ Pb < ([1-RBR]*Reference Pb + RAP binder content) The coated RAP particles in the RAP mixture have more binder content than the coated virgin aggregate particles on the No. 4 sieve in the virgin mixture, but this difference is less than the RAP binder content. This represents partial binder availability, as illustrated in Figure 84b. Therefore, the concept behind this methodology is that if there is no difference in binder con- tents between the coated RAP particles and the coated virgin aggregate particles (both retained on the No. 4 sieve), there is 100% RAP binder availability since the RAP binder is fully released and completely active and available in the mixture. However, if the coated RAP particles have a higher binder content than the coated virgin aggregate particles, then the binder in the RAP is not fully released and not fully active and available in the mixture. Depending on the difference between the binder contents of these particles, the RAP binder availability can be calculated. To calculate the percent RAP binder availability, a linear relationship, as shown in Figure 85 and Equation 23, can be used between the following two extremes: Scenario 1 when RAP′ Pb y = -30.3x + 221.2 0 20 40 60 80 100 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 R A P B A F ( % ) RAPʹ Pb (%) Figure 85. Example relationship between BAF and RAP  Pb.

120 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios equals 4.0% in this example, which represents 100% availability, and Scenario 2 when RAP′ Pb equals 7.3% in this example, which represents 0% availability. From this relationship, a binder availability factor (BAF) for a given Reference Pb and RAP′ Pb can be calculated. The RAP BAF is the percentage of available (effective) RAP binder in the mixture, and this factor can be used to adjust the base binder content in mixtures with RAP, to ensure that the total optimum (active) binder content as prescribed in the mix design is achieved. [ ]( ) = × ′ +% Equation 23RAP BAF m RAP Pb b where RAP BAF (%) = the RAP binder availability factor, M = the slope (−30.3 in this example), RAP ′ Pb = the binder content of RAP particles retained on the No. 4 sieve, and B = the intercept (221.2 in this example). The slope and intercept values are dependent on both the virgin and the RAP asphalt mixtures (total binder content and aggregate type and gradation), while RAP′ Pb is dependent on the RAP binder availability. Therefore, as long as the virgin and RAP asphalt mixtures have the same total binder content and aggregate type and gradation, Equation 15 can be used to calculate the BAF. Notably, the value of the slope and intercept will proportionally change with the RAP binder content (i.e., using a different RAP source), but that will have no effect on the BAF. In the 0% availability case (Scenario 2 with RAP′ Pb equal to 7.3% in this example), RAP′ Pb will always equal [1-RBR]*Reference Pb + RAP binder content. 4.5.2 Verification This methodology was initially verified in Phase 2 using artificial RAP (i.e., laboratory aged). The artificial RAP was produced by mixing a PG 64–22 base binder with virgin aggregate frac- tions retained on the No. 4 sieve at a binder content of 4.5% to simulate RAP particles retained on the No. 4 sieve. This artificial RAP was then aged in the laboratory according to the following protocols: • No aging: labeled as RAP 1 and representing a soft RAP, • 5 days at 110°C (230°F): labeled as RAP 2 and representing a stiff RAP, • 10 days at 110°C (230°F): labeled as RAP 3 and representing a very stiff RAP, and • 10 days at 110°C (230°F) plus 3 days at 150°C (302°F): labeled as RAP 4 and representing an extremely stiff RAP. The BAF of each artificial RAP was calculated using the method described previously, by pre- paring virgin and RAP (artificial) mixtures with virgin aggregate from TX (limestone) with the percent retained for each fraction (by weight of total aggregate) of 28% (3/8 in.), 30% (No. 4), 28% (No. 8), and 14% (No. 30). The RBR in the RAP mixtures was 0.3, and the total binder content in both mixtures was 4.5%. In the virgin mixture, the Reference Pb was 4.0% by ignition oven. In the artificial RAP mixtures, the RAP′ Pb values for each different artificial RAP were also determined by ignition oven. Figure 86a shows the RAP′ Pb values for the artificial RAPs. As expected, the soft RAP (RAP 1) had a slightly higher binder content (RAP′ Pb) than the Reference Pb (4.27% versus 4.0%), while the extremely stiff RAP (RAP 4) had a much higher binder content (RAP′ Pb) than the Reference Pb (6.01% versus 4.0%). This resulted in higher BAF values for RAP 1 compared to RAP 4, as shown in Figure 86b. As expected, the BAF value has a negative correlation with RAP stiffness (or extent of aging): the softer the RAP binder, the higher the BAF.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 121 4.5.3 Factors Affecting RAP BAF The proposed methodology was also used to estimate the RAP BAF of actual RAP materials from seven different sources in the United States: TX, Florida (FL), IN, New Hampshire (NH), NV, DE, and WI. These materials were used to evaluate the impact of the following variables on the RAP BAF: • Mixing temperature and short-term conditioning period, • RAP source and RAP binder PG, • Recycling-agent addition and method of addition, and • Base binder source (quality). Mixtures were prepared at two mixing temperatures: 140°C and 150°C. Figure 87 a shows the results of RAP BAF versus mixing temperature. The error bars in each column represent ± one standard deviation from the average BAF value of the two replicates. It is clear that mixing temperature plays a dominant role in increasing the RAP BAF: the higher the mixing (a) (b) 4.27% 4.50% 5.11% 6.01% 3% 4% 5% 6% 7% 8% RAP 1 Soft RAP 1 Stiff RAP 1 Very stiff RAP 1 Extremely stiff R A P P b 0% Availability 100% Availability 91.9% 85.0% 66.4% 39.1% 0% 20% 40% 60% 80% 100% RAP 1 Soft RAP 1 Stiff RAP 1 Very stiff RAP 1 Extremely stiff R A P B A F Figure 86. (a) RAP Pb values, and (b) BAF values for asphalt mixtures with artificial RAPs.

122 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios temperature, the higher the BAF. This is expected since higher mixing temperatures help soften the RAP binder, making it more fluid and facilitating its blending with the base binder. Figure 87b shows the estimated RAP BAF of two different short-term conditioning periods (2 h versus 4 h); in both cases, mixing and conditioning temperatures were 150°C and 135°C, respectively. Extending the short-term conditioning to 4 h slightly increased the RAP BAF of FL, IN, and DE RAP sources, but statistically, there was no difference between 2 h versus 4 h of short-term conditioning time. Figure 88 shows the results of RAP BAF versus RAP binder PGH at 140°C and 150°C mixing temperatures. A clear trend is observed in both cases: the lower the RAP binder PGH, the higher the BAF. Therefore, when mixing at 140°C, for instance, it is estimated that only 50% of the TX (a) (b) 49 .6 % 51 .2 % 55 .8 % 74 .4 % 59 .9 % 78 .2 % 80 .6 % 67 .4 % 77 .4 % 75 .4 % 90 .6 % 77 .6 % 91 .3 % 93 .8 % 0% 20% 40% 60% 80% 100% 120% Texas RAP Florida RAP Indiana RAP New Hampshire RAP Nevada RAP Delaware RAP Wisconsin RAP R A P B A F 140°C 150°C 67 .4 % 77 .4 % 75 .4 % 90 .6 % 77 .6 % 91 .3 % 93 .8 % 68 .6 % 80 .5 % 82 .6 % 92 .0 % 77 .3 % 95 .3 % 93 .8 % 0% 20% 40% 60% 80% 100% 120% Texas RAP Florida RAP Indiana RAP New Hampshire RAP Nevada RAP Delaware RAP Wisconsin RAP R A P B A F 2 hours 4 hours Figure 87. (a) Effect of mixing temperature on RAP BAF and (b) effect of short-term conditioning period on RAP BAF.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 123 RAP binder will be active and available in the mixture, compared to 80% for the WI RAP. How- ever, if the mixing temperature is increased to 150°C, the availability of the RAP binder from TX and WI will increase to about 70% and 95%, respectively. To evaluate the effect of recycling-agent addition on the RAP BAF, a modified vegetable oil (V2) was added to the RAP mixtures at a dose of 5%. To evaluate the method of recycling- agent addition, the recycling agent was added to the base binder prior to mixing with the virgin aggregate and RAP in one set of RAP mixtures, while in another set, the recycling agent was added directly to the RAP (at room temperature for about 5 min) before mixing with the virgin aggregate and base binder. Figure 89a shows that including the recycling agent in the mixture clearly increased the RAP BAF for most RAP sources at 140°C mixing temperature. However, the method of adding the recycling agent to the RAP directly, as opposed to mixing it with the base binder, did not show any significant effect on the RAP BAF. This could be due to the fact that the recycling agent was added to the RAP just 5 min before mixing and at room temperature, and thus, there was neither sufficient time nor an elevated temperature to aid the recycling agent in diffusing into the RAP binder. (a) (b) Figure 88. RAP BAF versus RAP PGH at (a) 140çC and (b) 150çC mixing temperature.

124 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios (a) (b) 49 .6 % 51 .2 % 55 .8 % 74 .4 % 59 .9 % 78 .2 % 80 .6 % 61 .0 % 61 .8 % 62 .7 % 82 .0 % 67 .4 % 87 .1 % 86 .4 % 59 .7 % 60 .3 % 62 .3 % 83 .2 % 67 .8 % 85 .9 % 85 .1 % 0% 20% 40% 60% 80% 100% 120% Texas RAP Florida RAP Indiana RAP New Hampshire RAP Nevada RAP Delaware RAP Wisconsin RAP R A P B A F No Recycling Agent Recycling agent Added to the Binder Recycling agent Added to the RAP 67 .4 % 7 7. 4% 75 .4 % 90 .6 % 77 .6 % 91 .3 % 93 .8 % 76 .2 % 84 .1 % 80 .2 % 96 .6 % 80 .1 % 96 .8 % 97 .3 % 75 .6 % 80 .3 % 77 .4 % 93 .3 % 78 .3 % 95 .1 % 96 .8 % 0% 20% 40% 60% 80% 100% 120% Texas RAP Florida RAP Indiana RAP New Hampshire RAP Nevada RAP Delaware RAP Wisconsin RAP R A P B A F No Recycling Agent Recycling agent Added to the Binder Recycling agent Added to the RAP Figure 89. Effect of recycling agent addition and method of addition on RAP BAF at (a) 140çC and (b) 150çC mixing temperature.

Expanded Laboratory Performance of High RBR Binder Blends and Mixtures 125 Figure 89b shows, however, that adding the recycling agent slightly increased the RAP BAF at 150°C mixing temperature but did not show any statistical difference except for the TX and FL RAP sources. This would indicate that the recycling agent helps increase the RAP BAF only at low mixing temperatures and that increasing the mixing temperature has an equivalent effect to adding a recycling agent. The method of adding the recycling agent to the RAP directly, as opposed to mixing it with the base binder at 150°C mixing temperature, also did not show any significant effect on RAP BAF. Again, the limited time and room temperature conditions likely contributed to this result. To evaluate the effect of base binder source, an additional set of virgin and RAP mixtures was prepared with a PG 64–28 base binder from NH, with the exact same composition as the mix- tures with the PG 58–28 base binder from WI. The NH PG 64–28 base binder had a DTc of +1.2, compared to the WI PG 58–28 with a DTc of −3.4. Figure 90 shows that using a high-quality base binder with a high (positive) DTc value, such as the NH PG 64–28 binder, slightly increased the RAP BAF for all RAP sources compared to using the lower-quality WI PG 58–28 binder at the same 140°C mixing temperature. This is despite the fact that the WI PG 58–28 binder is softer on the high-temperature end but the same at the low-temperature end, compared to the NH PG 64–28 binder. 4.6 Key Findings The key findings presented in this chapter are based on expanded laboratory performance results for both binder blends and mixtures with high RBRs to explore the impact of higher recycling agent doses than those used in the field projects and include the following: • Recycling-agent effectiveness must be characterized in high RBR binder blends or mixtures initially and with long-term aging to capture initial compatibility and rheological response to oxidation. • A recycling-agent dose to match continuous PGH for the target climate is required for high RBR binder blends and mixtures to maintain durability with long-term aging. 49 .6 % 51 .2 % 55 .8 % 74 .4 % 59 .9 % 78 .2 % 80 .6 % 55 .6 % 56 .6 % 63 .6 % 86 .0 % 66 .4 % 87 .1 % 88 .7 % 0% 20% 40% 60% 80% 100% 120% Texas RAP Florida RAP Indiana RAP New Hampshire RAP Nevada RAP Delaware RAP Wisconsin RAP R A P B A F WI 58-28 (∆Tc -3.4) NH 64-28 (∆Tc +1.4) Figure 90. Effect of base binder source on RAP BAF.

126 Evaluating the Effects of Recycling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios • Use of high-quality base binders improves performance of high RBR binder blends and mix- tures with recycling agents. • Rejuvenation mechanisms differ by recycling-agent type. • Recycling agents are more effective in rejuvenating less-aged recycled materials (RAP more than RAS and MWAS more than TOAS) in balanced, limited proportions. RAS contents should be limited because at typical production temperatures, RAS likely acts as a filler with none of the stiff, brittle recycled binder available for blending. • Adequate performance for high RBR binder blends with recycling agents can be evaluated by PGH, G-R parameter, and DTc. • Adequate performance for high RBR mixtures with recycling agents can be evaluated by N12.5, G-Rm, FI, Sm and m-valuem, and CRIEnv. • Some high RBR mixtures with recycling agent may be moisture susceptible.