Short-Term Laboratory Conditioning of Asphalt Mixtures (2015)

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.

25 Findings and Applications This chapter provides the mixture test results for the Phase I and Phase II experiments. Mixture volumetrics, stiffness, and HWTT results are summarized and analyzed for simulating plant aging and examining the effects of the selected factors on plant aging. Phase I Mixture Volumetrics Table 3-1 presents the comparison of the volumetrics of LMLC specimens fabricated following the selected labora- tory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA, along with PMPC specimens, in terms of Gmm, percentage of absorbed binder (Pba), percentage of effective binder (Pbe), and effective binder film thickness (FTbe). The volumetrics were calculated using the mix design aggregate gradation and asphalt content per Superpave Mix Design (Asphalt Institute 2001). Figures 3-1 and 3-2 present the volumetric correlations for PMPC specimens versus LMLC specimens in terms of Gmm and Pba values, respectively. As illustrated in Figure 3-1, most of the data points fall on the line of equality, indicating that equivalent Gmm values were achieved by PMPC specimens and LMLC specimens. The exceptions were mixtures from the Iowa test site that were produced as HMA and foamed WMA with highly absorptive aggregates (3.2 percent AC). A reason- able correlation in terms of Pba values was also observed, as shown in Figure 3-2, when comparing the two specimen types, with the exception of the same subset of the Iowa mixtures. It can be seen here that the absorption in the plant for the Iowa high-absorption mixtures was not as great in the laboratory as in the plant. There may be numerous explanations for this but the high-absorption aggregate mixture was produced at a higher-than-planned temperature in the plant as noted in Appendix A. Based on the data summarized in Table 3-1 and Figures 3-1 and 3-2, practically equivalent mixture volumet- rics were observed for PMPC specimens and LMLC specimens for a wide range of asphalt mixtures. Therefore, the selected laboratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA were consid- ered suitable to simulate the asphalt absorption during plant production and construction. Simulation of Plant Aging As mentioned previously, the laboratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA were selected and used in the project to simulate asphalt aging during plant production and con- struction. To explore the correlation in binder or mixture aging induced by the selected laboratory STOA protocols versus that during plant production, MR stiffness, E* stiff- ness, and HWTT RRP and rut depth at 5,000 load cycles for LMLC specimens from all nine field sites were plotted against the corresponding results obtained for PMPC specimens and cores at construction. In addition, continuous PG and FT-IR CA results for extracted and recovered binders from LMLC specimens from three field sites (i.e., Indiana, Florida, and Texas II) were plotted against the corresponding results for the binders extracted and recovered from plant loose mix and cores at construction. A detailed discussion of the results of each test is presented in the following subsections. MR Test Results Figures 3-3 and 3-4 present the MR stiffness correlation of LMLC specimens versus PMPC specimens and cores at con- struction, respectively. In Figure 3-3, most of the data points fall around the line of equality, which indicates that MR stiff- ness for LMLC specimens with the selected laboratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA fairly mimicked the MR stiffness of C H A P T E R 3

26 Field Site Mixture Type PMPC LMLC Gmm Pba Pbe FT Gmm Pba Pbe FT Texas I HMA 2.420 0.53 4.70 9.09 2.397 0.10 5.11 9.88 Evotherm 2.408 0.30 4.91 9.50 2.399 0.13 5.07 9.81 Foaming 2.400 0.15 5.06 9.77 2.407 0.28 4.93 9.53 HMA + RAP/RAS 2.410 0.83 4.42 7.89 2.418 0.98 4.27 7.64 Evotherm + RAP/RAS 2.420 1.02 4.24 7.58 2.417 0.96 4.29 7.67 New Mexico HMA 2.342 0.41 5.01 10.21 2.329 0.16 5.25 10.70 HMA + RAP 2.340 0.66 4.78 9.66 2.339 0.64 4.79 9.70 Evotherm + RAP 2.343 0.72 4.72 9.55 2.333 0.52 4.91 9.93 Foaming + RAP 2.335 0.56 4.87 9.85 2.349 0.84 4.61 9.32 Connecticut HMA + RAP 2.676 1.26 3.71 8.55 2.652 0.90 4.04 9.33 Foaming + RAP 2.675 1.24 3.72 8.59 2.658 0.99 3.96 9.14 Wyoming HMA 2.470 0.76 4.28 8.81 2.491 1.13 3.93 8.09 Evotherm High T 2.479 0.92 4.13 8.50 2.494 1.18 3.88 7.98 Evotherm Ctrl T 2.487 1.06 3.99 8.22 2.501 1.30 3.76 7.75 Foaming High T 2.485 1.03 4.03 8.29 2.497 1.24 3.83 7.88 Foaming Ctrl T 2.470 0.76 4.28 8.81 2.505 1.37 3.70 7.61 South Dakota HMA + RAP 2.441 0.58 4.75 7.28 2.441 0.58 4.75 7.28 Evotherm + RAP 2.440 0.56 4.77 7.31 2.440 0.56 4.77 7.31 Foaming + RAP 2.428 0.35 4.97 7.62 2.440 0.56 4.77 7.31 Advera + RAP 2.432 0.42 4.90 7.51 2.432 0.42 4.90 7.51 Iowa High Abs HMA + RAP High T 2.425 2.35 4.82 10.18 2.373 1.35 5.75 12.14 High Abs HMA + RAP Ctrl T 2.439 2.61 4.57 9.67 2.373 1.35 5.75 12.14 High Abs Foaming + RAP High T 2.435 2.54 4.64 9.81 2.365 1.19 5.89 12.45 High Abs Foaming + RAP Ctrl T 2.437 2.57 4.61 9.74 2.373 1.35 5.75 12.14 Low Abs HMA + RAP High T 2.481 0.61 4.42 9.28 2.482 0.63 4.40 9.25 Low Abs HMA + RAP Ctrl T 2.476 0.52 4.50 9.46 2.479 0.58 4.45 9.35 Low Abs Foaming + RAP High T 2.477 0.54 4.49 9.42 2.488 0.73 4.30 9.04 Low Abs Foaming + RAP Ctrl T 2.474 0.49 4.54 9.53 2.489 0.75 4.29 9.00 Indiana HMA + RAP BMP 2.451 1.31 4.77 8.12 2.458 1.32 4.65 7.92 HMA + RAP DMP 2.446 1.48 5.00 8.55 2.443 1.43 5.05 8.64 Advera + RAP BMP 2.448 1.29 4.84 8.24 2.456 1.43 4.71 8.02 Foaming + RAP DMP 2.455 1.43 4.73 8.06 2.440 1.16 4.98 8.49 Florida High Abs HMA + RAP 2.350 2.03 4.66 6.93 2.341 1.86 4.83 7.17 High Abs Foaming + RAP 2.363 2.18 4.37 6.48 2.365 2.22 4.33 6.43 Low Abs HMA + RAP 2.537 0.79 3.74 5.61 2.540 0.84 3.70 5.54 Low Abs Foaming + RAP 2.548 1.09 3.64 5.46 2.540 0.96 3.76 5.65 Texas II HMA BMP Binder A 2.402 1.39 5.06 7.97 2.393 1.11 5.11 8.10 HMA DMP Binder A 2.415 1.34 4.65 7.28 2.393 1.11 5.11 8.10 HMA BMP Binder V 2.395 1.26 5.19 8.17 2.392 1.16 5.21 8.20 HMA DMP Binder V 2.411 1.26 4.71 7.38 2.392 1.16 5.21 8.20 Table 3-1. Mixture volumetrics for PMPC and LMLC specimens.

27 the PMPC specimens. The biggest explainable exceptions are for the high-absorption mixes from Iowa, which were affected by the laboratory conditioning; the high-RAP mixture from New Mexico; and the rapidly aging asphalt mixture from Texas II. Although a reasonable correlation is observed in Figure 3-4 between the cores at construction and the LMLC specimens, the cores exhibited lower MR stiffness, possibly due to the higher AV content and perhaps different aggre- gate orientation in the construction cores. Previous studies have shown that flatter aggregate orientation in field cores can lead to aniso tropic behavior resulting in lower mixture- stiffness values measured in the MR tests (Yin et al. 2013; Zhang et al. 2011). E* Test Results Figure 3-5 presents the correlation of E* stiffness results at 68°F (20°C) and 10 Hz for LMLC specimens versus PMPC specimens of asphalt mixtures from the Connecticut, Indiana, and Texas II field sites. Consistent with the results shown in Figure 3-3, a good correlation in E* stiffness is observed for LMLC specimens versus PMPC specimens in Figure 3-5. Therefore, the laboratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA were able to produce laboratory asphalt mixtures with an E* stiffness equivalent to that of plant- produced asphalt mixtures. The outlier shown in Figure 3-5 is the BMP PMPC specimen of HMA from the Indiana field 2.3 2.4 2.5 2.6 2.7 2.3 2.4 2.5 2.6 2.7 PM PC - G m m LMLC - Gmm TX I NM CT WY SD IA IN FL TX II Figure 3-1. Theoretical maximum specific gravity correlation for LMLC specimens versus PMPC specimens. 0.0 0.6 1.2 1.8 2.4 3.0 0.0 0.6 1.2 1.8 2.4 3.0 PM PC - P b a(% ) LMLC - Pba(%) TX I NM CT WY SD IA IN FL TX II Figure 3-2. Percentage of absorbed binder correlation for LMLC specimens versus PMPC specimens. 0 200 400 600 800 1000 0 200 400 600 800 1000 PM PC - M R (k si) LMLC - MR(ksi) TX I NM CT WY SD IA IN FL TX II Figure 3-3. Resilient modulus stiffness correlation for LMLC specimens versus PMPC specimens. 0 200 400 600 800 1000 0 200 400 600 800 1000 C on st ru ct io n C or e - M R (k si) LMLC - MR(ksi) TX I NM CT WY SD IA IN FL TX II Figure 3-4. Resilient modulus stiffness correlation for LMLC specimens versus cores at construction.

28 site, which showed a significantly lower E* stiffness compared to its corresponding LMLC counterpart. The authors could not find a reasonable explanation for this occurrence. HWTT Results Figures 3-6 and 3-7 present the HWTT RRP results for LMLC specimens versus PMPC specimens and cores at con- struction, respectively. The asphalt mixtures included in this evaluation did not show early stripping during the tests and had LCSN values greater than 3,000 load cycles. As illus- trated in Figure 3-6, a reasonable correlation, indicated by values scattered about the line of equality, in HWTT RRP values between LMLC specimens and PMPC specimens was obtained, indicating the selected LTOA protocols were able to produce laboratory asphalt mixtures with rutting resistance in the HWTT equivalent to that of the mixture produced in the plant. However, a distinct trend is shown in Figure 3-7, where almost all of the data points are above the line of equal- ity. Thus, cores at construction exhibited a higher rutting sus- ceptibility in the HWTT compared to their corresponding LMLC specimens. The degradation and debonding of the plaster needed to fit the cores into the testing mold was likely a significant contributor to higher rut depths for cores at con- struction and a consequent poor correlation with the LMLC results. Therefore, there is a need for appropriate modifica- tions to the HWTT procedure for testing field cores in the future. It is also possible that anisotropy in the field cores could have contributed to this behavior. In addition to the HWTT RRP results, the traditional rutting resistance parameter of rut depth at 5,000 load cycles was also used to evaluate the simulation of plant aging by the selected laboratory STOA protocols. The results for LMLC specimens versus PMPC specimens and cores at construction are pre- sented in Figures 3-8 and 3-9, respectively. Although a sub- stantial variability in the rut depth measurements is exhibited 5000 6000 7000 8000 9000 10000 5000 6000 7000 8000 9000 10000 PM PC - E* (M Pa ) LMLC - E* (MPa) CT IN TX II Figure 3-5. Dynamic modulus stiffness at 20ºC/10 Hz correlation for LMLC specimens versus PMPC specimens. 0 4 8 12 16 20 0 4 8 12 16 20 PM PC - H W TT (m icr os tra in/ cy cle ) LMLC - HWTT (microstrain/cycle) TX I NM CT WY SD FL Figure 3-6. HWTT RRP correlation for LMLC specimens versus PMPC specimens. 0 4 8 12 16 20 0 4 8 12 16 20 C on st ru ct io n C or e - H W TT (m icr os tra in/ cy cle ) LMLC - HWTT (microstrain/cycle) TX I NM CT WY SD FL Figure 3-7. HWTT RRP correlation for LMLC specimens versus cores at construction. 0 3 6 9 12 15 0 3 6 9 12 15 PM PC - H W TT (m m) LMLC - HWTT (mm) TX I NM CT WY SD IN FL TX II Figure 3-8. HWTT rut depth at 5,000 load cycle correlation of LMLC specimens versus PMPC specimens.

29 in Figure 3-8, there is a reasonable correlation, as indicated by a scattering around the line of equality, in terms of rutting resistance between LMLC specimens and their corresponding PMPC specimens. The reduced correlation observed for speci- mens with high rut-depth values (i.e., 6 to 12 mm at 5,000 load cycles) is possibly due to the rut depth induced by stripping. Similar to the results shown in Figure 3-7, a higher rutting sus- ceptibility as indicated by higher rut depths at 5,000 load cycles for cores at construction versus their corresponding LMLC specimens can also be observed in Figure 3-9. Again, HWTT results for cores at construction were probably affected by the disintegration and debonding of the plaster needed to properly fit the cores into the molds during the test as well as possible anisotropy in the field cores. Continuous PG Results Figure 3-10 presents the continuous high-temperature and low-temperature PG correlations for extracted and recovered binders from plant loose mix versus those from LMLC speci- mens. As illustrated, most of the data points fall on the line of equality, indicating equivalent continuous PG results. This correspondence indicates that the plant production and the laboratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA produced an equivalent effect on binder stiffening. The correlations for the binders extracted from the Indiana and Florida mixtures were slightly lower than those extracted from the Texas II mixtures, possibly due to the incorporation of RAP. Figure 3-11 presents the continuous high-temperature and low-temperature PG correlations of extracted and recovered binders from cores at construction versus those from LMLC specimens, respectively. A similar trend was shown in that most of the data points fall near the line of equality, indi- cating equivalent continuous PG results. The similarity in the correlation results, as shown in Figure 3-10 versus Fig- ure 3-11, indicates that most short-term aging of asphalt binders and mixtures occurs during plant production, while the aging induced by the construction process (i.e., transpor- tation, laydown, and compaction) may be insignificant. FT-IR Test Results Figures 3-12 and 3-13 present the FT-IR CA correlations of extracted and recovered binders from LMLC specimens versus those from plant loose mix and cores at construction, respectively. As illustrated, most of the data points fall near the line of equality, despite a few outliers. This correspon- dence indicates that the extracted and recovered binders from the plant loose mix and cores at construction experienced an 0 3 6 9 12 15 0 3 6 9 12 15 C on st ru ct io n C or e - H W TT (m m) LMLC - HWTT (mm) TX I NM CT WY SD IN FL TX II Figure 3-9. HWTT rut depth at 5,000 load cycle correlation of LMLC specimens versus cores at construction. (a) High-Temperature PG 58 64 70 76 82 58 64 70 76 82 Pl an t L oo se M ix IN FL TX II LMLC Specimen (b) Low-Temperature PG -34 -28 -22 -16 -10 -34-28-22-16-10 Pl an t L oo se M ix LMLC Specimen IN FL TX II Figure 3-10. Continuous PG correlation for extracted and recovered binders from LMLC specimens versus plant loose mix.

30 equivalent level of oxidation during the plant production and construction as those binders extracted and recovered from the LMLC specimens fabricated using the STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA. Summary According to the MR and E* test results, good correlations in mixture stiffness between LMLC specimens with the selected laboratory STOA protocols and PMPC specimens and cores at construction were obtained for a wide range of asphalt mixtures from nine field sites. In addition, an approximately equivalent rutting resistance was observed for LMLC speci- mens and PMPC specimens in terms of HWTT RRP values and rut depths at 5,000 load cycles. A higher rutting suscepti- bility in the HWTT was shown for cores at construction than for the corresponding LMLC specimens, which was possibly caused by the need to plaster the cores to fit the height of the HWTT molds and subsequent disintegration of this plaster. Although it was not an experiment design factor, different binder grades and the use of polymers were included in the study. Across a wide spectrum of grades (PG 58-22 to 76-22), the LMLC MR test results were fairly consistent with the PMPC values. One of the most interesting observations con- cerning polymer modification can be seen in the New Mexico data where a virgin polymer-modified (i.e., PG 76-22) mix- ture had a lower stiffness than a high-RAP content mixture made with a straight PG 64-28 binder. Good correlations in binder stiffening and oxidation were observed for the extracted and recovered binders from LMLC specimens versus plant loose mix and cores at construction, as indicated by the continuous PG and the FT-IR CA results. 58 64 70 76 82 58 64 70 76 82 C on st ru ct io n C or e LMLC Specimen IN FL TX II (a) High-Temperature PG (b) Low-Temperature PG -34 -28 -22 -16 -10 -34-28-22-16-10 C on st ru ct io n C or e LMLC Specimen IN FL TX II Figure 3-11. Continuous PG correlation for extracted and recovered binders from LMLC specimens versus cores at construction. 0.5 0.7 0.9 1.1 1.3 1.5 0.5 0.7 0.9 1.1 1.3 1.5 Pl an t L oo se M ix LMLC Specimen IN FL TX II Figure 3-12. FT-IR carbonyl area correlation for extracted and recovered binders from LMLC specimens versus plant loose mix. 0.5 0.7 0.9 1.1 1.3 1.5 0.5 0.7 0.9 1.1 1.3 1.5 C on st ru ct io n C or e LMLC Specimen IN FL TX II Figure 3-13. FT-IR carbonyl area correlation for extracted and recovered binders from LMLC specimens versus cores at construction.

31 Thus, this study verified the simulation of binder or mixture aging during plant production and construction by the labo- ratory STOA protocols of 2 hours at 275°F (135°C) for HMA and 240°F (116°C) for WMA for asphalt mixtures with a wide spectrum of materials and production parameters (aggregate type, asphalt source, recycled materials, WMA technology, plant type, and production temperature). Identification of Factors Affecting the Performance of Short-Term Aged Asphalt Mixtures This section presents the results of laboratory experiments to identify mixture components and production parameters— i.e., factors—with significant effects on the performance of short-term aged asphalt mixtures. These factors included (1) WMA technology, (2) production temperature, (3) plant type, (4) recycled material inclusion, (5) aggregate absorption, and (6) binder source. Detailed discussions for each factor are presented in the following subsections. A statistical analysis was also performed to identify which factors had a significant effect on MR stiffness. Separate statis- tical experiments and analyses were performed to assess the effects of each of the six factors of interest while incorporating information on field site, specimen type (i.e., cores at con- struction, PMPC, and LMLC), NMAS (i.e., 9.5 mm, 12.5 mm, and 19 mm), and AV content as variables. Results of the statis- tical analyses are presented in Appendix E. WMA Technology (HMA versus WMA) The MR and HWTT results for LMLC specimens, PMPC specimens, and cores at construction from eight field sites with both HMA and WMA are shown in Figures 3-14 through 3-16; MR stiffness and HWTT RRP and rut depths at 5,000 load cycles for HMA mixtures are plotted against those of corre- sponding WMA mixtures. The x-axis coordinate represents HMA test results and the y-axis coordinate represents the cor- responding WMA test results. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for MR stiffness or rutting resistance parameters in the HWTT. The MR stiffness comparison for HMA versus WMA shown in Figure 3-14 illustrates that most of the data points are below the line of equality, indicating a higher MR stiffness for HMA compared to WMA. According to the shift from the line of equality, the average ratio of WMA MR stiffness over HMA MR stiffness is approximately 0.85. Thus, the inclusion of WMA technology in the asphalt mixture was likely to pro- duce asphalt mixtures with an approximately 15 percent lower stiffness than HMA. Figure 3-15 presents the comparison of HWTT RRP values for HMA versus WMA. As illustrated, most of the data points 0 200 400 600 800 1000 0 200 400 600 800 1000 W M A - M R (k si) HMA - MR(ksi) TX I NM CT WY SD IA IN FL Figure 3-14. MR stiffness comparison for HMA versus WMA. 0 4 8 12 16 20 0 4 8 12 16 20 W M A - H W TT (m icr os tra in/ cy cle ) HMA - HWTT (microstrain/cycle) TX I NM CT WY SD IN FL Figure 3-15. HWTT RRP comparison for HMA versus WMA. 0 3 6 9 12 15 0 3 6 9 12 15 W M A - H W TT (m m) HMA - HWTT (mm) TX I NM CT WY SD IA IN FL Figure 3-16. HWTT rut depth at 5,000 load cycle comparison for HMA versus WMA.

32 were above the line of equality, indicating a better rutting resistance in the HWTT for HMA than WMA. It may also be noticed that the data scatter tends to increase with larger values. Figure 3-16 presents the HWTT traditional rutting resistance parameter of rut depth at 5,000 load cycles. Similar to Figure 3-15, most of the data points in Figure 3-16 aligned above the line of equality. According to the slope of the line that illustrates the shift from the line of equality, the rut depth after the first 5,000 load cycles in the HWTT for WMA mix- tures is approximately 1.26 times higher than that for HMA. Thus, the inclusion of WMA technology is likely to produce early-life asphalt mixtures with a higher rutting susceptibil- ity, though the difference has not been observed in any of the field projects. Figures 3-17 and 3-18 present the E* master curve compar- isons for HMA versus WMA from the Connecticut and Indi- ana field sites, respectively. The comparison was performed for each specimen type available (i.e., PMPC and LMLC for Connecticut and BMP PMPC, DMP PMPC, BMP LMLC, and DMP LMLC for Indiana). For all comparisons in terms 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz PMPC - HMA PMPC - WMA (a) PMPC Specimens (b) LMLC Specimens 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz LMLC - HMA LMLC - WMA Figure 3-17. E* master curve comparison for HMA versus WMA for the Connecticut field site. (a) BMP PMPC Specimens (b) DMP PMPC Specimens (c) BMP LMLC Specimens (d) DMP LMLC Specimens 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz BMP PMPC - HMA BMP PMPC - WMA 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz DMP PMPC - HMA DMP PMPC - WMA 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz BMP LMLC - HMA BMP LMLC - WMA 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz DMP LMLC - HMA DMP LMLC - WMA Figure 3-18. E* master curve comparison for HMA versus WMA for the Indiana field site.

33 of E* stiffness for HMA versus WMA, with the exception of BMP PMPC specimens from the Indiana field site, the E* master curves for HMA mixtures were above or overlapping with those for WMA counterpart mixtures, indicating higher or equivalent E* stiffness values over a wide range of test- ing temperatures and frequencies. For the exceptional case of BMP PMPC specimens from the Indiana field site, slightly lower E* stiffness values were observed for HMA mixtures compared to their WMA counterparts. This was the same data anomaly that was mentioned earlier (Figure 3-5). For the statistical analysis, MR stiffness measurements obtained from eight field sites (Connecticut, Florida, Indiana, Iowa, New Mexico, South Dakota, Texas I, and Wyoming)— with information about specimen type, WMA technology, NMAS, and AV content—were used. The analysis of covari- ance (ANCOVA)—having WMA technology and specimen type as main effects along with a two-way interaction effect between them (specimen type*WMA technology), NMAS and AV content as covariates, and field site as a random effect—was fitted to the data. The analysis outputs are listed in Appendix E. The fixed-effect test results indicated that specimen type, WMA technology, AV content, and specimen type*WMA technology were statistically significant at a = 0.05, while the effect of NMAS was not significant. When there is a signifi- cant interaction effect, the effect of each factor involved in the interaction needs to be assessed against the levels of the other factor because the effect might be different for each level of the other factor. The effect of WMA technology was assessed for each level of specimen type. Except for cores at construc- tion, the predicted MR stiffness was lower for WMA than for HMA. In addition, the ANCOVA model without the two-way interaction term was also fitted to the data for comparison purposes. As before, the effects of specimen type, WMA tech- nology, and AV content were statistically significant, while the effect of NMAS was not. Production Temperature (High vs. Control) The MR test results for LMLC specimens, PMPC specimens, and cores at construction from the Wyoming and Iowa field sites with different production temperatures are presented in Figure 3-19, with MR stiffness for mixtures produced at high temperatures and those at control temperatures plotted against each other. The evaluation of rutting resistance in the HWTT by RRP value and rut depth at 5,000 load cycles was not avail- able for this factor since early stripping was observed for the majority of Iowa mixtures when tested at 122°F (50°C), with LCSN values less than 3,000 load cycles and rut depths greater than the failure criteria of 12.5 mm at 5,000 load cycles. The x-axis coordinate represents the MR stiffness for mixtures produced at control temperatures, and the y-axis coordinate represents MR stiffness for mixtures produced at high tem- peratures. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for MR stiffness. As illustrated in Figure 3-19, most of the data points fell on the line of equality, indicating equivalent MR stiffness for those asphalt mixtures. Therefore, an increase in production temperature during mixing followed by a conditioning pro- tocol that consisted of heating the PMPC to compaction tem- perature had no significant effect on mixture stiffness. For the statistical analysis, the ANCOVA model, including production temperature, WMA technology, and specimen type as main effects along with all possible two-way inter actions, AV content as a covariate, and field site as a random effect, was first fitted to the data, but none of the two-way interaction effects were statistically significant. Thus, the two-way interac- tion effects were removed from the model, and the ANCOVA model was fitted again to the data. The results showed that the effects of specimen type and WMA technology were statisti- cally significant at a = 0.05, while the effect of production temperature and AV content were not (see Appendix E). Plant Type (BMP vs. DMP) The MR test and HWTT results for LMLC specimens, PMPC specimens, and cores at construction from the Indiana and Texas II field sites with different plant types are presented in Figures 3-20 and 3-21 for MR stiffness and the HWTT tradi- tional rutting resistance parameter of rut depth at 5,000 load cycles, respectively, with BMP-produced mixtures and DMP- produced mixtures plotted against each other. The evaluation of rutting resistance in the HWTT by RRP value was not avail- able for this factor since early stripping was observed for the majority of Indiana and Texas II asphalt mixtures. The x-axis coordinate represents the test results for DMP-produced 0 100 200 300 400 500 600 0 100 200 300 400 500 600 M ix tu re a t H ig h Te m pe ra tu re - M R (k si) Mixture at Control Temperature - MR(ksi) WY IA Figure 3-19. MR stiffness comparison for asphalt mixtures produced at different temperatures.

34 mixtures, and the y-axis coordinate represents corresponding test results for BMP-produced mixtures. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for MR stiffness or rut depth measure- ments in the HWTT. The MR test results in Figure 3-20 show equivalent mixture stiffness was achieved by asphalt mixtures produced in a BMP and the corresponding mixtures produced in a DMP. Similar to the MR test results, most of the data points shown in Fig- ure 3-21 fall on the line of equality. Therefore, equivalent mix- ture stiffness and rutting resistance was observed for asphalt mixtures produced in a BMP and a DMP. For the statistical analysis, the ANCOVA model includ- ing plant type and specimen type as main effects, plant type* specimen type as a two-way interaction effect, AV content as a covariate, and field site as a random effect was first fitted to the data. However, the two-way interaction effect was not statisti- cally significant, and the ANCOVA model without the two-way interaction effect was fitted again to the data. The results (see Appendix E) show that the effects of specimen type and AV content were statistically significant at a = 0.05, while the effect of plant type was not. Inclusion of Recycled Material (RAP/RAS vs. No RAP/RAS) The MR test and HWTT results for LMLC specimens, PMPC specimens, and cores at construction from the Texas I and New Mexico field sites with mixtures with and without RAP/RAS are presented in Figures 3-22 through 3-24; MR stiffness and HWTT RRP and rut depth at 5,000 load cycles for RAP/RAS mixtures and control mixtures without recycled materials are plotted against each other. The control mixtures from the Texas I field site were produced using a PG 70-22 binder, while 0 200 400 600 800 1000 0 200 400 600 800 1000 BM P M ix tu re - M R (k si) DMP Mixture - MR(ksi) IN TX II Figure 3-20. MR stiffness comparison for asphalt mixtures produced at BMP versus DMP. 0 3 6 9 12 15 0 3 6 9 12 15 BM P M ix tu re - H W TT (m m) DMP Mixture - HWTT (mm) IN TX II Figure 3-21. HWTT rut depth at 5,000 load cycle comparison for asphalt mixtures produced at BMP versus DMP. 0 200 400 600 800 1000 0 200 400 600 800 1000 R A P/ R A S M ix tu re - M R (k si) No RAP/RAS Mixture - MR(ksi) TX I NM Figure 3-22. Resilient modulus stiffness comparison for asphalt mixtures with and without RAP and RAS. 0 4 8 12 16 20 0 4 8 12 16 20 R A P/ R A S M ix tu re - H W TT (m icr os tra in/ cy cle ) No RAP/RAS Mixture -HWTT (microstrain/cycle) TX I NM Figure 3-23. HWTT RRP comparison for asphalt mixtures with and without RAP and RAS.

35 the RAP/RAS mixtures were produced using a softer PG 64-22 binder in conjunction with 15 percent RAP and 3 percent RAS. The RAP came from a stockpile at Ramming Paving in Austin, Texas, and the RAS was from tear-off shingles ground to 100 percent passing the 0.5-in. (12.5 mm) sieve. The control mixtures from the New Mexico mixtures were produced using a PG 76-28 binder, while the RAP mixtures were produced using a softer PG 64-28 binder in conjunc- tion with 35 percent RAP. The x-axis coordinate represents test results for the control mixtures, and the y-axis coordinate represents corresponding test results for RAP/RAS mixtures. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for MR stiff- ness or rutting resistance parameters in the HWTT. The MR stiffness results in Figure 3-22 indicate, based on the shift from the line of equality, that the RAP/RAS mixtures have higher stiffness as compared to the control mixtures. However, the scatter around the red dashed line indicates that the increase in mixture stiffness induced by adding recycled materials was inconsistent. Recycled materials have a pro- found effect on the properties of WMA and HMA as they tend to stiffen mixtures appreciably. A large difference in behavior of RAS-bearing mixtures will occur depending upon the ori- gin of the RAS, whether it comes from manufacturing waste or tear-offs. Likewise, the character of RAP will vary depending upon its age and climatic exposure. This suggests that recycled materials (i.e., RAP and RAS) from different sources utilized in different field sites should be treated as unique materials whose properties are related to the original asphalt mixtures and in-service times and climatic conditions. Figure 3-23 presents the comparison of HWTT RRP val- ues for RAP/RAS mixtures versus control mixtures. As illus- trated, most points were above the line of equality, indicating decreased rutting resistance in the HWTT for RAP/RAS mix- tures compared to the control counterpart mixtures. This was possibly due to the softer virgin asphalt binders used in the RAP/RAS mixtures and the degree of blending with the recy- cled binders. According to the shift from the line of equal- ity, the average ratio of the RRP value for RAP/RAS mixtures over the RRP value for the control mixtures was approxi- mately 1.44. Figure 3-24 presents the traditional HWTT rut- ting resistance parameter of rut depth at 5,000 load cycles. No consistent trend in the comparison of RAP/RAS mixtures versus control mixtures was observed for this parameter. For the statistical analysis, the ANCOVA model includ- ing recycled materials, specimen type, and WMA technology as main effects along with all possible two-way interaction effects—among them, AV content as a covariate and field site as a random effect—was first fitted to the data. Because the WMA technology*recycled materials interaction effect was not statistically significant at a = 0.05, it was removed, and the ANCOVA model was fitted again to the data. The results (see Appendix E) show that the only effect that was not sta- tistically significant at a = 0.05 was AV content. A multiple comparison procedure (Tukey’s honestly significant differ- ence) carried out to test which of those factor levels were statistically different showed that the difference between no RAP/RAS and RAP/RAS mixtures was statistically significant for cores at construction, PMPC, and LMLC, although the amount of difference varied with specimen type. The conclu- sion from the statistical analysis is that, in general, mixtures with RAP/RAS had higher MR stiffness than mixtures with no RAP/RAS although there is considerable variability due to the origin, age, and nature of the recycled materials. Aggregate Absorption (High- vs. Low-Absorptive Aggregate) The MR test and HWTT results (rut depth at 5,000 load cycles) for LMLC specimens, PMPC specimens, and cores at construction from the Iowa and Florida field sites with aggre- gates of different absorption are presented in Figures 3-25 and 3-26, respectively. In the figures, results from asphalt mix- tures using high-absorptive aggregates versus low-absorptive aggregates are plotted against each other. The evaluation of rutting resistance using the HWTT RRP value was not avail- able for this factor since early stripping was observed for the majority of Iowa and Florida asphalt mixtures, with LCSN val- ues less than 3,000 load cycles, most likely due to the use of PG 58-28 binder at both locations. The x-axis coordinate rep- resents test results for mixtures using high-absorptive aggre- gates, and the y-axis coordinate represents corresponding test results for mixtures using low-absorptive aggregates. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for the MR stiff- ness or rut depth measurements in the HWTT. 0 3 6 9 12 15 0 3 6 9 12 15 R A P/ R A S M ix tu re - H W TT (m m) No RAP/RAS Mixture - HWTT (mm) TX I NM Figure 3-24. HWTT rut depth at 5,000 load cycle comparison for asphalt mixtures with and without RAP and RAS.

36 The MR stiffness comparison for mixtures using high- versus low-absorptive aggregates in Figure 3-25 shows that most of the data points are above the line of equality, indicat- ing a higher MR stiffness for mixtures using low-absorptive aggregates compared to those with high-absorptive aggre- gates. As evident in Figure 3-25, mixtures with highly absorp- tive aggregates may produce results that are variable when they are subjected to the recommended STOA. Figure 3-26 presents the HWTT results in terms of rut depth at 5,000 load cycles. Consistent with Figure 3-25, most of the data points in Figure 3-26 are below the line of equality, indicat- ing better rutting resistance in the HWTT for mixtures using low-absorptive aggregates. The reduced stiffness and rutting resistance observed for the mixtures with high-absorptive aggregates could be due to the thicker FTbe that resulted from incorporating higher binder contents, as indicated by the higher Pbe and FT values listed in Table 3-1. For the statistical analysis, the ANCOVA model—including aggregate absorption, specimen type, and WMA technology as main effects; AV content as a covariate; and field site as a random effect—was fitted to the MR stiffness measurements obtained from the Iowa and Florida field sites. The results show that the effects of specimen type, WMA technology, and aggregate absorption were statistically significant at a = 0.05, while the effect of AV was not (see Appendix E for details). Binder Source (Binder A vs. Binder V) The MR test results for LMLC specimens, PMPC specimens, and cores at construction from the Texas II field site with different binder sources are presented in Figure 3-27, with MR stiffness for mixtures using Binder A plotted against those of corresponding mixtures using Binder V. The evaluation of rutting resistance in the HWTT by RRP value and rut depth at 5,000 load cycles was not available for this factor since early stripping in the HWTT was observed for all Texas II mixtures. The x-axis coordinate represents MR stiffness for mixtures using Binder A, and the y-axis coordinate represents MR stiff- ness for mixtures using Binder V. The black solid line is the line of equality, and the red dashed line illustrates the shift from the line of equality for MR stiffness. As shown in Figure 3-27, most of the data points are below the line of equality, indicating a higher MR stiffness for asphalt mixtures using Binder A than the mixtures using Binder V. Figure 3-28 presents the E* master curve comparisons for PMPC specimens and LMLC specimens of asphalt mixtures with Binder A versus Binder V. The comparison was performed for each specimen type (i.e., BMP PMPC, DMP PMPC, and 0 200 400 600 800 1000 0 200 400 600 800 1000 Lo w -A bs or pt io n M ix tu re - M R (k si) High-Absorption Mixture - MR(ksi) IA FL Figure 3-25. Resilient modulus stiffness comparison for asphalt mixtures using high-absorptive versus low-absorptive aggregates. 0 3 6 9 12 15 0 3 6 9 12 15 Lo w -A bs or pt io n M ix tu re - H W TT (m m) High-Absorption Mixture - HWTT (mm) IA FL Figure 3-26. HWTT rut depth at 5,000 load cycle comparison for asphalt mixtures using high- absorptive versus low-absorptive aggregates. 0 200 400 600 800 1000 0 200 400 600 800 1000 M ix tu re w ith B in de r V - M R (k si) Mixture with Binder A - MR(ksi) TX II Figure 3-27. Resilient modulus stiffness comparison for asphalt mixtures using different binder sources.

37 (a) BMP PMPC Specimens (b) DMP PMPC Specimens (c) LMLC Specimens 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz BMP PMPC - Binder A BMP PMPC - Binder V 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz DMP PMPC - Binder A DMP PMPC - Binder V 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E-04 1.0E-02 1.0E+00 1.0E+02 1.0E+04 D yn am ic M od ul us , M Pa Reduced Frequency, Hz LMLC - Binder A LMLC - Binder V Figure 3-28. E* master curve comparison for asphalt mixtures using different binder sources for the Texas II field site.

38 LMLC). As illustrated, the E* master curves for mixtures with Binder A were consistently above the curves obtained for mix- tures with Binder V. Therefore, binder source exhibited a sig- nificant effect on the performance of short-term aged asphalt mixtures and, consequently, asphalt mixtures using the same performance-graded binders from different sources could exhibit substantially different mixture performance in terms of stiffness and rutting resistance. For the statistical analysis, the ANCOVA model included binder source and specimen type as main effects and AV con- tent as a covariate; originally, a model including a two-way interaction—binder source*specimen type—was used, but the interaction was not statistically significant. The results detailed in Appendix E show that the effects of binder source, specimen type, and AV content were all statistically signifi- cant at a = 0.05. Specifically, Binder A yielded significantly higher MR stiffness than Binder V. Summary In this section, the effects of various mixture components and production parameters including WMA technology, pro- duction temperature, plant type, recycled material inclusion, aggregate absorption, and binder source on the performance of short-term aged asphalt mixtures were evaluated. The cor- relations in terms of MR stiffness and HWTT rutting resis- tance parameters including RRP and rut depth at 5,000 load cycles were performed for each factor, and the results are summarized in Table 3-2. Mixture components and produc- tion parameters with significant effects were identified based on the magnitude of the slope of the shifted line with respect to the line of equality being greater than 1.05 or smaller than 0.95 (i.e., 5 percent off from the line of equality) and corrobo- rated via statistical analysis. As shown in Table 3-2, binder source, aggregate absorption, WMA technology, and inclusion of recycled materials had significant effects on stiffness and rutting resistance of short- term aged asphalt mixtures. However, no significant effect from plant type and production temperature was observed. Based on the results of Phase I, guidance for conducting short-term aging studies for different factors affecting asphalt mixtures was prepared. This guidance is found in Appendix F, Proposed AASHTO Recommended Practice for Conducting Plant Aging Studies. The guidance is applicable to the experi- ment design, mix design, asphalt mix sampling, plant sample compaction, performance testing, and data analysis. Phase II Quantification of Field Aging CDD (32°F [0°C] base) is proposed to provide a basis for field aging and to account for the differences in construction dates and climates for various field sites. The CDD values for the seven field sites included in the Phase II experiment are presented in Figure 3-29; they are noticeably different and are able to provide a distinct indication of the individual climatic characteristics. Specifically, the average slopes (i.e., secant slopes) of the curves for the Texas I, New Mexico, and Florida field sites are significantly steeper than those located Factor MR Stiffness HWTT RRP HWTT Rut Depth Slope Magnitude Significant Effect Statistically Significant Slope Magnitude Significant Effect Slope Magnitude Significant Effect WMA Technology 0.836 Yes Yes 1.259 Yes 1.251 Yes Production Temperature 0.985 No No N/A Plant Type 1.008 No No N/A 0.968 No Inclusion of Recycled Materials 1.779 Yes Yes 1.44 Yes 0.981 No Aggregate Absorption 1.271 Yes Yes N/A 0.675 Yes Binder Source 0.818 Yes Yes N/A Table 3-2. Summary of the effects of mixture components and production parameters on the performance of short-term aged asphalt mixtures.

39 in colder climatic zones, including Wyoming, South Dakota, Iowa, and Indiana, due to differences in pavement in-service temperatures. Additionally, the construction date has a significant effect on the CDD value and a subsequent effect on field aging of asphalt mixtures. For example, the South Dakota field site shown in Figure 3-29 was constructed in October 2012, and the pave- ment went through the 2012 winter prior to the 2013 sum- mer. Consequently, the CDD curve was flat for the first several months (corresponding to the winter season); afterwards, the slope of the curve increased due to the high in-service tem- peratures during the summer. On the basis of these consider- ations, field sites with different construction dates and climates will have different CDD values for a given pavement in-service time. Specifically, field sites located in warmer climates and constructed in the spring or summer are likely to experience more severe initial field aging due to higher CDD values com- pared to those located in colder climates and constructed in the fall or winter. In Figure 3-29, data points highlighted in black represent the time when field cores were acquired. It was the intent to have, at a minimum, construction cores and, depending upon the time of construction, one or more subsequent cores acquired 8 to 12 months apart. The mixture test results of these cores were used to evaluate the mixture stiffness and rutting resistance evolution with field aging. Figures 3-30 and 3-31 present the plots of the CDD values for post-construction cores versus their associated MR ratios and HWTT RRP ratios, respectively; the data points repre- sent the average property ratio values for each field site, and the adjusted line represents the exponential function from Equation (2-6). As illustrated, both MR ratio and HWTT RRP ratio exhibit a significant increase with CDD values. Accord- ing to the coefficient of determination (R2) values shown in 0 5000 10000 15000 20000 25000 30000 35000 40000 Dec-11 Jul-12 Jan-13 Aug-13 Mar-14 Sep-14 Apr-15 C um ul at iv e D eg re e- D ay s ( °F -d ay s) Coring Date Texas I New Mexico Wyoming South Dakota Iowa Indiana Florida Figure 3-29. Cumulative degree-days for various field sites. 1.0 1.5 2.0 2.5 3.0 0 10000 20000 30000 40000 M R St iff ne ss R at io Cumulative Degree-Days (°F-days) Predicted Measured R2 = 0.831 Figure 3-30. Resilient modulus ratio versus cumulative degree-days. 1.0 3.0 5.0 7.0 9.0 0 10000 20000 30000 40000 H W TT R R P R at io Cumulative Degree-Days (°F-days) Predicted Measured R2 = 0.753 Figure 3-31. HWTT RRP ratio versus cumulative degree-days.

40 Figures 3-30 and 3-31, it is reasonable to use the MR ratio or HWTT RRP ratio as a function of CDD values to quantify mixture aging in the field. Figures 3-32 and 3-33 present the continuous PG results for extracted and recovered binders from cores at construc- tion and post-construction from the Indiana and Florida field sites, respectively. As illustrated, for most of the extracted and recovered binders, post-construction cores (i.e., 9-month cores from the Indiana field site, and 9-month and 15-month cores from the Florida field site) show higher continuous performance grades than the corresponding cores at con- struction, indicating increased binder stiffness but a reduced ductility with field aging. Figures 3-34 and 3-35 present the plots of the CDD values for Indiana and Florida post-construction cores versus their associated DSR G* ratios and FT-IR CA ratios, respectively. The data points represent the average DSR G* ratio and FT-IR CA ratio for each field site and pavement in-service time, and the predicted lines represent the exponential function from Equation (2-6). As illustrated, both the DSR G* ratio and the FT-IR CA ratio exhibit an increase with CDD values, indi- cating that a significant level of binder stiffening and oxi- dation occurred during pavement in-service life. Although limited binder DSR G* at 77°F (25°C) and FT-IR CA results were used to determine these correlations, the results are promising. Correlation of Field Aging with Laboratory LTOA Protocols As previously introduced, binder or mixture property ratios (MR ratio, HWTT RRP ratio, DSR G* ratio, and FT-IR CA ratio) were proposed to quantify the evolution of mixture stiffness, rutting resistance, and binder oxidation with field Figure 3-32. Continuous PG results for Indiana post-construction cores versus cores at construction. Figure 3-33. Continuous PG results for Florida post-construction cores versus cores at construction. 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10000 20000 30000 40000 D SR G * R at io Cumulative Degree-Days (°F-days) Predicted Measured R2 = 1.00 Figure 3-34. DSR complex modulus ratio versus cumulative degree-days.

41 aging. They were defined as the ratios of binder or mixture properties of short-term aged specimens to those properties of long-term aged specimens. For specimens fabricated in the laboratory, short-term aged mixtures include LMLC speci- mens with STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA, and long-term aged mixtures refer to HMA and WMA LMLC specimens with STOA protocol of 2 hours at 275°F (135°C) followed by LTOA protocols of 5 days at 185°F (85°C) or 2 weeks at 140°F (60°C). For field specimens, short-term aged mixtures corre- spond to construction cores, while long-term aged specimens refer to cores acquired several months after construction. Findings from Phase I of this project indicated that the selected STOA protocols of 2 hours at 275°F (135°C) for HMA and 2 hours at 240°F (116°C) for WMA were repre- sentative of PMPC specimens as well as construction cores in terms of mixture stiffness and rutting resistance, and binder stiffening and oxidation. Thus, the correlation between the long-term field aging and the selected laboratory LTOA pro- tocols in terms of mixture stiffness and binder oxidation could also be made on the basis of MR ratio, DSR G* ratio, and FT-IR CA ratio values for long-term aged cores and LMLC specimens and their corresponding extracted and recovered binders. Mixture test results showed that the average MR ratio val- ues for all LMLC specimens with STOA protocol of 2 hours at 275°F (135°C) plus LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) from the selected field sites were approximately 1.46 and 1.76, respectively. The higher MR ratio value indicated that the LTOA protocol at 185°F (85°C) produced a greater level of mixture aging compared to that at 140°F (60°C), though associated with a shorter aging time period (i.e., 5 days versus 2 weeks). This suggests that mixture aging was more sensitive to aging temperature than to aging time in the LTOA protocol, which is consistent with the findings from NCHRP 9-49 (Epps Martin et al. 2014). Figure 3-36 illustrates the correlation of field aging and lab- oratory LTOA protocols on MR stiffness. The average MR ratio values for LMLC specimens with STOA protocol of 2 hours at 275°F (135°C) plus LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were plotted as markers by crossing the curve of the MR ratio versus CDD values, as presented in Figure 3-30. The vertical and horizontal error bars represent one standard deviation from the average MR ratio values and corresponding CDD values, respectively. As illustrated, labo- ratory LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were able to produce mixture aging equiva- lent to an average of 9,100 and 16,000 CDD, respectively, in the field. A subset of LMLC specimens from the Indiana and Florida field sites was aged using an additional LTOA protocol of 3 days at 185°F (85°C), because the aging induced by the standard lab- oratory LTOA protocol of 5 days at 185°F (85°C) per AASHTO R35 was more significant than that of 2 weeks at 140°F (60°C). The property ratios for the Indiana and Florida LMLC speci- mens with different LTOA protocols were compared against each other to determine if an equivalent level of laboratory aging could be produced by LTOA protocols of 3 days at 185°F (85°C) and 2 weeks at 140°F (60°C). Figure 3-37 presents the compari- son of different LTOA protocols in terms of CDD after equating their MR ratio values to the curve presented in Figure 3-30. For both the Indiana and Florida mixtures, equivalent CDD values were achieved for LTOA protocols of 3 days at 185°F (85°C) and 2 weeks at 140°F (60°C), which were significantly lower than those from 5 days at 185°F (85°C). Therefore, laboratory STOA protocol of 2 hours at 275°F (135°C) plus LTOA protocols of 3 days at 185°F (85°C) and 2 weeks at 140°F (60°C) were able to produce an equivalent level of mixture aging, which was less than the 5 days at 185°F (85°C) aging protocol. 1.0 1.1 1.2 1.3 1.4 1.5 0 10000 20000 30000 40000 FT -I R C A R at io Cumulative Degree-Days (°F-days) Predicted Measured R2 = 1.00 Figure 3-35. FT-IR carbonyl area ratio versus cumulative degree-days. 1.0 1.5 2.0 2.5 3.0 0 10000 20000 30000 40000 M R St iff ne ss R at io Cumulative Degree-Days (°F-days) 2h@135C + 2w@60C/3d@85C 2h@135C + 5d@85C Figure 3-36. Resilient modulus ratio correlation between field aging and laboratory LTOA protocols (based upon Figure 3-30).

42 The DSR G* results measured at 77°F (25°C) indicate that the average DSR G* ratio values for the Indiana extracted and recovered binders and Florida recovered binders and LMLC specimens with LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were approximately 1.24 and 1.47, respectively. In addition, the binder oxidation results from the FT-IR CA ratio values for the two LTOA protocols were approximately 1.11 (2 weeks at 140°F [60°C]) and 1.15 (5 days at 185°F [85°C]), which may not be practically different. Con- sistent with the mixture-stiffness results discussed previously, the LTOA protocol of 5 days at 185°F (85°C) produced a greater level of binder stiffening and oxidation compared to that of 2 weeks at 140°F, indicating a greater sensitivity of binder aging characteristics to LTOA temperature than to LTOA time. Figure 3-38 presents the correlation of field aging and laboratory LTOA protocols in terms of the binder stiffening effect. The average DSR G* ratio values for extracted and recovered binders from LMLC specimens with the two LTOA protocols were plotted as markers by crossing the curve of the DSR G* ratio versus CDD values, as shown in Figure 3-34. The vertical and horizontal error bars represent one standard deviation from the average DSR G* ratio values and the cor- responding CDD values for six different mixtures (i.e., four Indiana mixtures and two Florida mixtures). As illustrated by the DSR G* results, laboratory LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were able to pro- duce binder stiffening equivalent to an average of 10,200 and 14,300 CDD, respectively. These CDD values were close to those determined using the MR ratio results shown in Fig- ure 3-36. Figure 3-39 presents the correlation of binder G* stiffness at 77°F (25°C) and mixture MR stiffness at 77°F (25°C) for 32 Indiana and Florida field cores and LMLC speci- mens, as well as the corresponding extracted and recovered binders. As illustrated, similarity between binder stiffness and mixture stiffness was observed. Note that MR is influenced by more factors than just binder aging, such as air voids and aggregate absorption. Figure 3-40 illustrates the correlation of field aging and laboratory LTOA protocols in terms of the effect on binder oxidation (i.e., FT-IR CA). The average FT-IR CA ratio values for extracted and recovered binders from LMLC specimens with STOA and LTOA protocols were plotted as markers by crossing the curve of the FT-IR CA ratio versus CDD values, as presented in Figure 3-35. The vertical and horizontal error bars represent one standard deviation from the average FT-IR CA ratio values and the corresponding CDD values for six dif- ferent mixtures (i.e., four Indiana mixtures and two Florida mixtures). As illustrated by the binder oxidation results, labo- ratory LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were able to produce binder oxidation equiv- alent to an average of 14,100 and 16,900 CDD, respectively, 0 3000 6000 9000 12000 IN FL C um ul at iv e D eg re e- D ay s Field Site 2w@60C 3d@85C 5d@85C Figure 3-37. Comparison of cumulative degree-days achieved by different LTOA protocols for the Indiana and Florida field sites. 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 10000 20000 30000 40000 D SR G * R at io Cumulative Degree-Days (°F-days) 2h@135C + 2w@60C 2h@135C + 5d@85C Figure 3-38. DSR complex modulus ratio correlation between field aging and laboratory LTOA protocols. 2000 4000 6000 8000 10000 2000 3000 4000 5000 6000 7000 M R a t 2 5° C (M Pa ) G* at 25°C (kPa) Indiana Florida Figure 3-39. Correlation between binder complex modulus stiffness and LMLC mixture resilient modulus stiffness.

43 in the field. Note that the correlation illustrated in Figure 3-40 was determined based on a limited amount of binder FT-IR CA results and therefore the cross points for LTOA protocols of 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) in Figure 3-40 would likely change if more binder results were included. Based on the mixture MR stiffness results, LTOA protocols of 3 days at 185°F (85°C) or 2 weeks at 140°F (60°C) and 5 days at 185°F (85°C) were representative of field aging of approximately 9,100 and 16,000 CDD, respectively. Using the information shown in Figure 3-29, the in-service time for each field site corresponding to 9,100 and 16,000 CDD was deter- mined and is summarized in Table 3-3. As shown, the labora- tory LTOA protocols of 3 days at 185°F (85°C) or 2 weeks at 140°F (60°C) were equivalent to approximately 7 months in service in warmer climates and 12 months in service in colder climates. As for the aging induced by the laboratory LTOA protocol of 5 days at 185°F (85°C), approximately 11 months and 22 months in service were required for warmer climates and colder climates, respectively. Identification of Factors with Significant Effects on Mixture Aging Characteristics This section presents the laboratory test results for iden- tifying factors with significant effects on long-term mixture aging characteristics. These factors include WMA technology, production temperature, plant type, recycled material inclu- sion, and aggregate absorption. Detailed discussions for each factor are presented in the following subsections. WMA Technology (HMA vs. WMA) The MR test and HWTT results for long-term aged mixtures including cores and LMLC specimens with LTOA protocols from the seven Phase II field sites are shown in Figures 3-41 and 3-42. The x-axis coordinate represents HMA test results and the y-axis coordinate represents the corresponding WMA test results. The solid line is the line of equality, and the dashed line illustrates the shift from the line of equality for the MR ratio or HWTT RRP ratio. The MR ratio comparison for HMA versus WMA shown in Figure 3-41 illustrates that most of the data points aligned above the line of equality, indicating a greater increase in MR stiffness after long-term aging for WMA compared to HMA. According to the shift from the line of equality, the average dif- ference in MR ratio for WMA versus HMA is approximately 11 percent greater. Although it exhibited more data scatter, a similar trend in the HWTT RRP ratio comparison for WMA versus HMA is observed in Figure 3-42, with most of the data points falling above the line of equality. This indicates that the increase in mixture rutting resistance in the HWTT after 1.0 1.1 1.2 1.3 1.4 1.5 0 10000 20000 30000 40000 FT -I R C A R at io Cumulative Degree-Days (°F-days) 2h@135C + 2w@60C 2h@135C + 5d@85C Figure 3-40. FT-IR carbonyl area ratio correlation between field aging and laboratory LTOA protocols. Field Site Climate MR Ratio 2 weeks at 140°F (60°C) or 3 days at 185°F (85°C) 5 days at 185°F (85°C) Texas I Warmer Climate 6 months 10 months New Mexico 8 months 12 months Florida 6 months 11 months Average 7 months 11 months Wyoming Colder Climate 12 months 22 months South Dakota 12 months 22 months Iowa 12 months 22 months* Indiana 11 months 21 months* Average 12 months 22 months * Projected in-service time based on historical climatic information. Table 3-3. Correlation of field aging in terms of in-service time and laboratory LTOA protocols.

44 the long-term aging for WMA was more significant than that for HMA. As the results of Phase I showed, in general, WMA mix- tures begin their service lives with lower stiffness and tend to age in place more rapidly, as shown in Figures 3-41 and 3-42, but ultimately their stiffness and rutting resistance are com- parable to HMA. The lower stiffness in WMA was possibly due to the use of WMA technologies as well as lower produc- tion and short-term aging temperatures. It is reasonable to hypothesize that there is a critical in-service time for WMA at which an equivalent mixture property as HMA is achieved. To determine the critical in-service time when WMA equals HMA, CDD values for WMA and HMA post-construction cores and their associated MR ratio values were fitted sepa- rately using Equation (2-6), as shown in Figure 3-43. In the figure, the data points represent the average HMA and WMA MR ratio values for each field site, and the curves represent the exponential functions as expressed in Equation (2-6) for the MR ratio versus CDD values. As illustrated, the WMA curve was above the HMA curve, verifying a greater increase in mix- ture stiffness after long-term aging for WMA versus HMA. Based on the definition of the mixture property ratio, the MR stiffness of WMA and HMA cores at a given in-service time or CDD value can be determined using Equations (3-1) and (3-2), respectively. The fitted exponential functions for HMA and WMA post-construction cores shown in Figure 3-43 are denoted as fHMA and fWMA. M M M 1 exp CDD Eq. (3-1) R WMA R WMA0 WMA R WMA0 WMA f= = + α − β        γ p p p Where: MR WMA = MR stiffness of WMA post-construction cores at a given CDD value; and MR WMA0 = MR stiffness of WMA construction cores. M M M 1 exp CDD Eq. (3-2) R HMA R HMA0 HMA R HMA0 HMA f= = + α − β        γ p p p Where: MR HMA = MR stiffness of HMA post-construction cores at a given CDD value; and MR HMA0 = MR stiffness of HMA construction cores. By making Equations (3-1) and (3-2) equal, the critical CDD value for achieving equivalent MR stiffness by WMA 0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0 W M A M R R at io HMA MR Ratio TX I NM WY SD IA IN FL y = 1.107x Figure 3-41. Resilient modulus ratio comparison for HMA versus WMA. 0.0 2.0 4.0 6.0 8.0 0.0 2.0 4.0 6.0 8.0 W M A H W TT R R P R at io HMA HWTT RRP Ratio TX I NM WY SD FL y = 1.3767x Figure 3-42. HWTT RRP ratio comparison for HMA versus WMA. 1.0 1.5 2.0 2.5 3.0 3.5 0 10000 20000 30000 40000 M R St iff ne ss R at io Cumulative Degree-Days (°F-days) HMA Predicted WMA Predicted HMA Measured WMA Measured Figure 3-43. Resilient modulus ratio versus cumulative degree-days for HMA and WMA post-construction cores.

45 and HMA (CDDWMA=HMA) can be determined, as expressed in Equation (3-3). M 1 exp CDD M 1 exp CDD Eq. (3-3) R WMA0 WMA R HMA0 HMA + α − β        = + α − β        γ γ p p p p In addition to CDDWMA=HMA, the determination of the CDD value at which the MR stiffness of WMA post-construction cores equaled that of HMA construction cores (CDDWMA=HMA0) was also essential in order to understand the performance evo- lution of WMA in the field compared to HMA. CDDWMA=HMA0 can be calculated in accordance with Equation (3-4). M 1 exp CDD M Eq. (3-4) R WMA0 WMA R HMA0+ α − β        = γ p p Depending on the difference in the magnitude of MR stiff- ness for WMA and HMA construction cores, the mixture stiffness evolution of WMA and HMA with field aging can be categorized into three different scenarios, as shown in Figure 3-44. Scenario I in Figure 3-44(a) illustrates the case where the MR stiffness of the HMA cores was always higher than their WMA counterparts, but the difference in stiff- ness between these two mixtures decreased with field aging. Scenario II in Figure 3-44(b) indicates the case where HMA had higher mixture stiffness compared to WMA at the ini- tial aging stage (i.e., construction cores). It should be noted that beyond the catch-up point shown in Figure 3-44(b) is a projection from a regression beyond the limit of most of the data. Hence, it should not be concluded that WMA will age at a faster rate than HMA beyond this point. Scenario III repre- sents the case where equivalent mixture stiffness was shown for HMA and WMA construction cores, but higher stiffness for post-construction cores was observed for WMA versus HMA, as shown in Figure 3-44(c). Table 3-4 summarizes the CDDWMA=HMA and CDDWMA=HMA0 values for each field site. For the majority of the field sites (four out of seven), the MR stiffness evolution with field aging followed the trend illustrated in Scenario II, indicating that the stiffness of WMA was initially lower than that of HMA but it was able to catch-up to the stiffness of HMA after a certain amount of time in the field. The average CDDWMA=HMA and CDDWMA=HMA0 values for those four field sites were approxi- mately 23,000 and 3,000 CDD, respectively. Thus, field aging of approximately 3,000 CDD is necessary for the stiffness of WMA to equal the initial stiffness of HMA, and equivalent WMA and HMA mixture stiffness is likely to be achieved after 23,000 CDD of field aging. (a) 0.5 1.0 1.5 2.0 2.5 3.0 0 10000 20000 30000 40000 N or m al iz ed M R St iff ne ss Cumulative Degree-Days (°F-days) HMA Predicted WMA Predicted Scenario I (b) 0.5 1.0 1.5 2.0 2.5 3.0 0 10000 20000 30000 40000 N or m al iz ed M R St iff ne ss Cumulative Degree-Days (°F-days) HMA Predicted WMA Predicted Scenario II (c) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 10000 20000 30000 40000 N or m al iz ed M R St iff ne ss Cumulative Degree-Days (°F-days) HMA Predicted WMA Predicted Scenario III Figure 3-44. Evolution of mixture resilient modulus stiffness with field aging for HMA versus WMA. Referring to Figure 3-29, the in-service time for each field site corresponding to 23,000 and 3,000 CDD was determined and is summarized in Table 3-5. As shown, approximately 17 months in service in warmer climates and 30 months in service in colder climates were needed in order to achieve equivalent mixture stiffness for WMA versus HMA. As for the

46 in-service time corresponding to CDDWMA=HMA0 of 3,000 CDD, approximately 2 months and 3 months were required in warmer climates and colder climates, respectively. For the statistical analysis (see Appendix E), the analysis of variance (ANOVA) having WMA technology and aging level as main effects along with field site as a random effect was fitted to the data. The results showed that the effects of WMA technology and aging level were statistically significant at a = 0.05. Also, WMA mixtures showed a higher predicted MR ratio value than HMA mixtures. As expected, the pre- dicted MR ratio seemed to increase as aging level increased. Production Temperature (High vs. Control) The MR test results for long-term aged mixtures includ- ing cores after certain in-service times and LMLC specimens with STOA protocol of 2 hours at 275°F (135°C) plus LTOA protocols from the Wyoming and Iowa field sites are shown in Figure 3-45, with the MR ratio for mixtures produced at high temperatures and those at control temperatures plot- ted against each other. The evaluation of mixture rutting resistance evolution with long-term aging on the basis of the HWTT RRP ratio was not available for this factor since early stripping during the HWTT was observed for the majority of Wyoming and Iowa mixtures, with LCSN values less than 3,000 load cycles, most likely due to the high test temperature and the softer binders from these sites. The x-axis coordinate represents the test results for control temperature mixtures, and the y-axis coordinate represents the results for mixtures at high temperature. The solid line is the line of equality, and the dashed line illustrates the shift from the line of equality for the MR ratio. The MR ratio results shown in Figure 3-45 illustrate that most of the data points aligned along the line of equality, indicating an equivalent increase in MR stiffness induced by long-term aging of mixtures produced at high versus control temperatures. Therefore, production temperature differences for these two sites had no significant effect on the sensitivity of mixture stiffness to long-term aging. For the statistical analysis, an ANOVA model including production temperature, WMA technology, and aging level as fixed effects and field site as a random effect was fitted to the data. Results showed that the effect of the factor of interest— production temperature—was not statistically significant at a = 0.05. Details of the analysis can be found in Appendix E. Plant Type (BMP vs. DMP) The MR test results for long-term aged mixtures including cores after 10 months in service and LMLC specimens with Field Site Scenario WMA0/HMA0 CDD Values WMA = HMA WMA = HMA0 Texas I I 0.717 N/A 7,672 New Mexico II 0.876 22,421 2,929 Wyoming II 0.867 25,486 3,152 South Dakota II 0.897 16,543 2,426 Iowa II 0.860 28,156 3,328 Indiana III 1.002 0 0 Florida III 0.999 0 0 Table 3-4. CDDWMA=HMA and CDDWMA=HMA0 values for each field site. Field Site Climate CDD Values WMA = HMA WMA = HMA0 Texas I Warmer Climate 16 months 2 months New Mexico 19 months 3 months Florida 15 months 1 months Average 17 months 2 months Wyoming Colder Climate 32 months* 2 months South Dakota 32 months* 7 months Iowa 28 months* 2 months Indiana 26 months* 2 months Average 30 months 3 months * Projected in-service time based on historical climatic information. Table 3-5. Field in-service time corresponding to CDDWMA=HMA and CDDWMA=HMA0 values for each field site.

47 LTOA protocols of 5 days at 185°F (85°C) and 2 weeks at 140°F (60°C) from the Indiana field site are shown in Figure 3-46, with the MR ratio for BMP-produced mixtures and DMP- produced mixtures plotted against each other. The evaluation of mixture rutting resistance evolution with long-term aging on the basis of the HWTT RRP ratio was not available for this factor since early stripping was observed for the majority of Indiana mixtures, with LCSN values less than 3,000 load cycles. The x-axis coordinate represents the test results for BMP- produced mixtures, and the y-axis coordinate represents corresponding results for DMP-produced mixtures. The solid line is the line of equality, and the dashed line illus- trates the shift from the line of equality for the MR ratio. The MR ratio results shown in Figure 3-46 illustrate that most of the data points align along the line of equality, indicating an equivalent increase in MR stiffness induced by long-term aging for the BMP- and DMP-produced mixtures. Therefore, plant type had no significant effect on the sensitivity of mixture stiff- ness to long-term aging. For the statistical analysis, the ANOVA model including plant type, aging level, and WMA technology as main effects (since all possible two-way interactions were statistically insig- nificant) showed that none of the factor effects (as well as the factor of interest—plant type) was statistically significant at a = 0.05. Additional details can be found in Appendix E. Inclusion of Recycled Material (RAP/RAS vs. No RAP/RAS) The MR test results for long-term aged mixtures including cores after certain in-service times and LMLC specimens with LTOA protocols of 5 days at 185°F (85°C) and 2 weeks at 140°F (60°C) from the Texas I and New Mexico field sites are shown in Figure 3-47, with the MR ratio for control mix- tures without recycled materials and RAP/RAS mixtures plotted against each other. The evaluation of mixture rutting resistance evolution with long-term aging on the basis of the HWTT RRP ratio was not available for this factor since early stripping was observed for the majority of Texas I and New Mexico mixtures, with LCSN values less than 3,000 load cycles. The control mixtures from the Texas I field site were pro- duced using a PG 70-22 binder, while the RAP/RAS mixtures were produced using a softer PG 64-22 binder in conjunction with 15 percent RAP and 3 percent RAS; the control mixtures from the New Mexico site were produced using a PG 76-28 binder, while the RAP mixtures were produced using a softer PG 64-28 binder in conjunction with 35 percent RAP. The x-axis coordinate represents the test results for the control mixtures, and the y-axis coordinate represents correspond- ing results for RAP/RAS mixtures. The solid line is the line of equality, and the dashed line illustrates the shift from the line of equality for the MR ratio. 0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0 M ix tu re a t H ig h- Te m p M R R at io Mixture at Control-Temp MR Ratio WY IA y = 1.0603x Figure 3-45. Resilient modulus ratio comparison for mixtures produced at high versus control temperature. 0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0 D M P M ix tu re M R R at io BMP Mixture MR Ratio IN y = 1.0063x Figure 3-46. Resilient modulus ratio comparison for mixtures produced at BMP versus DMP. 0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0 R A P/ R A S M ix tu re M R R at io No RAP/RAS Mixture MR Ratio TX I NM Figure 3-47. Resilient modulus ratio comparison for mixtures produced with RAP/RAS versus no RAP/RAS.

48 The MR ratio results shown in Figure 3-47 illustrate that the data points align below the line of equality, indicating a signifi- cantly higher increase in MR stiffness after long-term aging for the control mixtures compared to the RAP/RAS mixtures. The greater sensitivity to aging exhibited by the control mixtures can possibly be attributed to the larger amount of virgin binder in the mixture, which is likely more susceptible to aging. There- fore, the inclusion of recycled materials had a significant effect on mixture aging characteristics in this study. For the statistical analysis (see Appendix E), the ANOVA model included recycled materials, aging level, and WMA technology as main effects (since all possible two-way inter- actions were statistically insignificant at a = 0.05). Because field site was confounded with aging level for this dataset, field site could not be included as a random effect in the ANOVA model. The results showed that the effects of recycled materials, aging level, and WMA technology were all statistically significant at a = 0.05. The conclusion of the statistical analysis was that mix- tures with no RAP/RAS had a higher MR ratio compared to mixtures with RAP/RAS. Aggregate Absorption (High- vs. Low-Absorptive Aggregate) The MR test and HWTT results for long-term aged mix- tures including cores after certain in-service times and LMLC specimens with LTOA protocols of 5 days at 185°F (85°C) and 2 weeks at 140°F (60°C) from the Iowa and Florida field sites are shown in Figures 3-48 and 3-49; the MR ratio and HWTT RRP ratio for mixtures using high-absorptive aggregates versus low-absorptive aggregates are plotted against each other. The evaluation of rutting resistance evolution with long-term aging on the basis of the HWTT RRP ratio was not available for Iowa mixtures since early stripping was observed, with LCSN values less than 3,000 load cycles. The x-axis coordi- nate represents test results for mixtures using high-absorptive aggregates, and the y-axis coordinate represents correspond- ing test results for mixtures using low-absorptive aggregates. The solid line represents the line of equality, and the dashed line illustrates the shift from the line of equality for the MR ratio or HWTT RRP ratio. The MR ratio comparison for mixtures using high-absorptive versus low-absorptive aggregates shown in Figure 3-48 illus- trates that most of the data points align below the line of equality, indicating a greater increase in MR stiffness induced by long-term aging for mixtures using high-absorptive aggre- gates compared to the mixtures using low-absorptive aggre- gates. A similar trend for the HWTT RRP ratio is shown in Figure 3-49, indicating that mixtures using high-absorptive aggregates exhibited a greater increase in rutting resistance than those using low-absorptive aggregates. The greater sen- sitivity of mixture stiffness and rutting resistance to aging for mixtures using high-absorptive aggregates was likely due to the higher volume of effective binder in these mixtures that were available for aging (as indicated by higher Pbe values in Table 3-1), the continuous asphalt absorption by the aggregates with time, or both (West et al. 2014). Therefore, aggregate absorption and, more specifically, the amount of effective binder had a significant effect on mixture aging characteristics in this study. For the statistical analysis, an ANOVA model including aggregate absorption, aging level, and WMA technology as main effects, aging level*aggregate absorption and WMA technology*aggregate absorption as two-way interaction effects (WMA technology*aging level interaction was not statistically significant), and field site as a random effect was fitted to the data. The results showed that the effects of aggregate absorp- tion, aging level, aging level*aggregate absorption, and WMA 0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0 Lo w -A bs or pt io n M ix tu re M R R at io High-Absorption Mixture MR Ratio IA FL y = 0.8509x Figure 3-48. Resilient modulus ratio comparison for mixtures produced using high- versus low-absorptive aggregates. 0.0 2.0 4.0 6.0 8.0 0.0 2.0 4.0 6.0 8.0 Lo w -A bs or pt io n M ix tu re R R P R at io High-Absorption Mixture RRP Ratio FL y = 0.2451x Figure 3-49. HWTT RRP ratio comparison for mixtures produced using high- versus low-absorptive aggregates.

49 technology*aggregate absorption were statistically significant at a = 0.05. The difference between mixtures using high-absorptive and low-absorptive aggregates was statistically significant for WMA but not for HMA. Also, although mixtures with high- absorptive aggregates had a higher MR ratio for each level of aging, in general (except for Field Aging 1_IA [10-month field aging for the Iowa field project]), the difference between mixtures with high-absorptive and low-absorptive aggre- gates was statistically significant only for laboratory LTOA 2 (i.e., 5 days at 185°F [85°C]). Further details of the statistical analy sis can be found in Appendix E. Summary In this subsection, the effects of various mixture compo- nents and production parameters including WMA technol- ogy, production temperature, plant type, recycled material inclusion, and aggregate absorption on the mixture aging characteristics were evaluated based on the change in mix- ture stiffness and rutting resistance after long-term aging. The correlations in terms of MR ratio and HWTT RRP ratio were performed for each factor, and the results are summarized in Table 3-6. Factors with a significant effect on mixture aging characteristics were identified based on the magnitude of the slope of the shifted line with respect to the line of equality being greater than 1.05 or smaller than 0.95 (i.e., 5 percent off the line of equality) and were corroborated via statistical analysis. According to Table 3-6, WMA technology, recycled material inclusion, and aggregate absorption showed signifi- cant effects on mixture aging characteristics, while no signifi- cant effects from production temperature and plant type were observed. For aggregate absorption, the results from Phase I indicated that mixtures with high-absorptive aggregates had a reduced MR stiffness and rutting resistance compared to those with low-absorptive aggregates, likely due to the higher binder content and thus thicker FTbe used in the mixtures with high-absorptive aggregates. However, after long-term aging, a greater increase in MR and rutting resistance was observed for mixtures using high-absorptive aggregates. Factor MR Ratio HWTT RRP Ratio Slope Magnitude Significant Effect Statistically Significant Slope Magnitude Significant Effect WMA Technology 1.107 Yes Yes 1.377 Yes Production Temperature 1.060 No No N/A Plant Type 1.006 No No Inclusion of Recycled Materials 0.713 Yes Yes Aggregate Absorption 0.851 Yes Yes 0.245 Yes Table 3-6. Summary of the effects of mixture components and production parameters on mixture aging characteristics.