Using Recycled Asphalt Shingles with Warm Mix Asphalt Technologies (2018)

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.

30 Production, Construction, and Performance of Field Projects Production and construction information were collected from three WMA–RAS projects built before the start of this study and five new WMA–RAS projects that were con- structed and monitored during the course of this study. The projects built before this study are referred to as the “existing projects”; the five projects built and evaluated during this study are referred to as the “new projects.” For each project (existing and new), a control HMA section was constructed to provide a direct comparison for field per- formance and materials properties. The existing and new projects are discussed in the chronological order of their construction. Existing Projects Austin, Texas In December 2011, the Texas DOT set up an experimental field demonstration in Austin, Texas, on FM 973, with J. D. Ramming Paving Company as the contractor. The origi- nal experiment included nine mixes consisting of different amounts of RAP and RAS as both HMA and WMA to evaluate the effects on construction. For this research, the four mixes containing RAS were further evaluated. The asphalt mixtures evaluated were all 12.5-mm NMAS– stone matrix asphalt mixtures that were designed using the Texas Gyratory Compactor. All mixes contained the same vir- gin aggregates: limestone, manufactured sand, and field sand. Table 3-1 shows the differences in the four RAS mixes. The average gradations, asphalt content, and volumetric proper- ties from Texas DOT’s QA testing are shown in Table 3-2a and Table 3-2b. Site Description. The mixes were placed on a 2.9-mi portion of FM 973. The north end of the project is the intersection of FM 969, and the south end is just south of Green Grover Road. This two-lane roadway has entryways to both a concrete plant and aggregate quarry, which yield a significant percentage of heavy truck traffic. At the time of construction, the annual average daily traffic was reported to be 11,000 on the north end and 11,300 on the south end. The percentages of truck traffic were reported as 4.2% and 4.3%, respectively. Figure 3-1 shows the location of the test sections. The mixes were placed in both lanes at a target thickness of 2 in. For the purposes of this research, only the southbound lane was inspected. Table 3-1 shows the length of each mix section in the southbound lane. Figure 3-2 shows the in-place densities reported by Texas DOT during construction. Field Performance. Field performance evaluations were conducted on October 10, 2014, and November 4, 2015, after approximately 35 and 47 months of traffic had been applied to the sections, respectively. Data were collected on each section to document performance with regard to rutting, cracking, and raveling by randomly selecting three 200-ft (61-m) data sections within each mix section. Each data section was then inspected to assess performance. In addition, five 6-in. (150-mm) diameter cores were taken from between the wheelpaths for each mix to determine the in-place density. Rutting. The rut depths were measured at the beginning of each 200-ft section with a straight edge and a wedge. All four test sections had performed well with regard to rutting, exhibiting only small rut depths. Table 3-3 shows the rut depths measured in each section. Cracking. The entirety of each 200-ft section was carefully inspected for visual signs of cracking and rated based on the Distress Identification Manual for the Long-Term Pavement Per- formance Program. Table 3-4 shows the total cracking observed, based on crack type and severity at the time of the 35-month inspection. The WMA mix with 15% RAP and 3% RAS using a PG 64-22 binder performed the best with regard to cracking, with only three total cracks observed in the data sections. C H A P T E R 3

31 Texas DOT Section Number Base Binder WMA RAP (%) RAS (%) Date Paved Section Length (ft) 9 PG 64-22 Evotherm DATa 15 3 12/13/2011 2,080 3 PG 64-22 None 15 3 1/5/2012 1,985 4 PG 64-22 None 0 5 1/6/2012 2,010 6 PG 58-28 None 15 3 1/18/2012 1,352 aDispersed asphalt technology. Table 3-1. Differences in mix design and production in Austin, Texas. Sieve Size Percent Passing HMA PG 64-22a 15%–3% HMA PG 64-22a 0%–5% HMA PG 58-28a 15%–3% WMA PG 64-22 15%–3% (Evotherm DAT) 19.0 mm (3/4 in.) 100.0 99.7 100.0 100.0 9.5 mm (3/8 in.) 84.2 86.1 80.1 82.8 4.75 mm (No. 4) 54.6 55.5 51.6 53.6 2.36 mm (No. 8) 36.3 39.2 37.4 38.2 0.6 mm (No. 30) 22.7 24.8 23.7 23.3 0.3 mm (No. 50) 18.1 17.6 17.3 18.0 0.075 mm (No. 200) 6.0 5.4 5.4 5.8 aNo WMA technology. Table 3-2a. Gradation from Texas DOT QA testing. Variable HMA PG 64-22 a 15%–3% HMA PG 64-22a 0%–5% HMA PG 58-28a 15%–3% WMA PG 64-22 15%–3% (Evotherm DAT) Asphalt content (%) 5.3 5.3 5.3 5.3 Air voids (%) 2.5 2.7 2.6 3.2 VMA (%) 14.6 14.6 14.6 15.0 VFA (%) 82.9 81.5 82.2 78.7 aNo WMA technology. Table 3-2b. Asphalt content and volumetrics from Texas DOT QA testing.

32 Figure 3-1. Test section layout on FM 973 in Austin, Texas.

In -P la ce D en si ty (% ) WMA PG 64-22 15%–3% HMA PG 64-22 15%–3% HMA PG 58-28 15%–3% HMA PG 64-22 0%–5% Figure 3-2. In-place densities based on cores at construction reported by Texas DOT. Test Section 200-ft Section ID Inside Wheelpath Rut Depth (in.) Outside Wheelpath Rut Depth (in.) WMA PG 64-22 15% RAP–3% RAS 1 0 1/16 2 0 1/16 3 1/16 1/16 HMA PG 64-22 15% RAP–3% RAS 1 0 1/16 2 0 1/16 3 0 1/16 HMA PG 64-22 0% RAP–5% RAS 1 0 1/16 2 0 1/16 3 0 0 HMA PG 58-28 15% RAP–3% RAS 1 0 1/16 2 0 1/16 3 0 1/16 Table 3-3. Average rutting in RAS sections on FM 973 in Austin, Texas. Note: Loc. = locations. Section Crack Severity Wheelpath Longitudinal Nonwheelpath Longitudinal Transverse Block No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Loc. Total Area (ft2) WMA PG 64-22 15% RAP 3% RAS Low 3 38 0 0 0 0 0 0 Moderate 0 0 0 0 0 0 0 0 High 0 0 0 0 0 0 0 0 HMA PG 64-22 15% RAP 3% RAS Low 0 0 0 0 9 24 6 4,455 Moderate 0 0 0 0 0 0 0 0 High 0 0 0 0 0 0 0 0 HMA PG 64-22 0% RAP 5% RAS Low 13 136 0 0 0 0 0 0 Moderate 0 0 0 0 0 0 0 0 High 0 0 0 0 0 0 0 0 HMA PG 58-28 15% RAP 3% RAS Low 19 236 1 7 46 158 0 0 Moderate 2 55 0 0 0 0 0 0 High 0 0 0 0 0 0 0 0 Table 3-4. Total cracking in RAS sections on FM 973 in Austin, Texas, after 35 months.

34 The HMA mix with 15% RAP and 3% RAS using a PG 64-22 binder exhibited the most cracking, with block cracking observed in all three data sections. Figure 3-3 shows one of the three cracks observed in the WMA section, while Figure 3-4 shows the block cracking observed through- out the HMA 15% RAP–3% RAS sections. Figure 3-5 and Figure 3-6 show examples of the typical cracking observed in the HMA mix with 0% RAP and 5% RAS and the HMA mix using a PG 58-28 binder, respectively. Some of the longitudinal cracking observed throughout the sec- tions may be attributed to material segregation during construction. At the time of the 47-month inspection, cracks had pro- gressed in all four RAS mix sections. A few new cracks had propagated in the WMA mix with 15% RAP and 3% RAS using a PG 64-22 binder, and other existing cracks had grown slightly. The block cracking in the HMA mix with 15% RAP and 3% RAS using a PG 64-22 binder had become more widespread. An example of this can be seen in Figure 3-7. Several new cracks had also propagated in the HMA mix with 0% RAP and 5% RAS and the HMA mix using a PG 58-28 binder. It should be noted that the last 200-ft data Figure 3-3. Example of low-severity crack observed in WMA 15% RAP–3% RAS mix in Austin, Texas. Figure 3-4. Example of block cracking observed in HMA 15% RAP–3% RAS mix in Austin, Texas. Figure 3-5. Example of low-severity crack observed in HMA 0% RAP–5% RAS mix in Austin, Texas. Figure 3-6. Example of low-severity cracking observed in 15% RAP–3% RAS mix with PG 58-28 in Austin, Texas.

35 section in the HMA mix using a PG 58-28 binder was no longer in service at the time of this inspection. Traffic had been diverted, and construction was underway on a new bridge south of the test sections. The construction had removed part of the pavement area where the third data section was located. Table 3-5 shows the total cracking after 47 months. Raveling and Weathering. The surface textures of the test sections were measured using the Sand Patch Test in accor- dance with ASTM E965. The Sand Patch Test was conducted at the beginning of each 200-ft section in the outside wheel- path. The calculated mean texture depths for all four mixes are shown in Table 3-6. These values represent the average and standard deviation of the three tests conducted on each evaluation section. A smaller mean texture depth indicates a smoother surface or one with less surface texture. These results show that all four mixes had similar and acceptable mean texture depths at the time of both inspections. Core Testing. At the time of each project inspection, five 6-in. (150-mm) cores were taken from each mix sec- tion. The densities of these cores were first measured using AASHTO T 166. However, most of the cores had water absorptions greater than 2.0%, which requires the density to be determined according to AASHTO T 331 instead. A sum- mary of the core densities at the time of both inspections is shown in Figure 3-8. ANOVA was performed to evaluate how the type of mixture, the age of the pavement, and the interaction between these two variables affect the in-place density. Table 3-7 shows the results of ANOVA. As can be observed, only the type of mixture has a significant effect (α = 0.05) on the in-place density for this location with a p-value of 0.001. Table 3-7 also shows the sta- tistical grouping from Tukey’s Test of Multiple Comparisons. The results of Tukey’s Test indicate that means that do not share a letter are significantly different at α = 0.05. Therefore, the PG 64-22 mixture with 5% RAS has a different (higher) density compared to the other three mixtures. Aurora, Illinois Two RAP and RAS mixes were placed on the I-88 Toll- way in the summer of 2012. These were stone matrix asphalt mixes containing either gravel or quartzite aggregates. Both lanes in each direction were paved for this construction. Figure 3-7. Example of block cracking observed in HMA 15% RAP–3% RAS mix at 47 months in Austin, Texas. Mix Section Crack Severity Wheelpath Longitudinal Nonwheelpath Longitudinal Transverse Block No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Loc. Total Area (ft2) WMA PG 64-22 15% RAP 3% RAS Low 6 113.0 0 0.0 0 0.0 0 0.0 Moderate 0 0.0 0 0.0 0 0.0 0 0.0 High 0 0.0 0 0.0 0 0.0 0 0.0 HMA PG 64-22 15% RAP 3% RAS Low 0 0.0 0 0.0 0 0.0 3 7,200.0 Moderate 0 0.0 0 0.0 0 0.0 0 0.0 High 0 0.0 0 0.0 0 0.0 0 0.0 HMA PG 64-22 0% RAP 5% RAS Low 18 219.0 0 0.0 1 1.0 0 0.0 Moderate 0 0.0 0 0.0 0 0.0 0 0.0 High 0 0.0 0 0.0 0 0.0 0 0.0 HMA PG 58-28 15% RAP 3% RAS Low 23 287.0 1 7.0 46 157.8 0 0.0 Moderate 4 86.0 0 0.0 0 0.0 0 0.0 High 0 0.0 0 0.0 0 0.0 0 0.0 Table 3-5. Total cracking in Austin, Texas, after 47 months.

36 Mix Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation 35 Months (November 2014) 47 Months (November 2015) WMA PG 64-22 15% RAP–3% RAS 0.7 0.2 0.6 0.1 HMA PG 64-22 15% RAP–3% RAS 0.6 0.1 0.5 0.1 HMA PG 64-22 0% RAP–5% RAS 0.5 0.1 0.5 0.1 HMA PG 58-28 15% RAP–3% RAS 0.6 0.1 0.5 0.0 Table 3-6. Mean texture depths in Austin, Texas. In -P la ce D en sit y (% ) WMA PG 64-22 15%–3% HMA PG 64-22 15%–3% HMA PG 58-28 15%–3% HMA PG 64-22 0%–5% 35 Months 47 Months Figure 3-8. In-place densities based on cores in Austin, Texas. Source df Adj. SS Adj. MS F-Value P-Value Mixture 3 110.58 36.86 12.56 0.001 Age 1 7.78 7.78 2.65 0.113 Mixture × Age 3 15.15 5.05 1.72 0.182 Error 33 96.85 2.93 Total 40 228.01 Statistical Grouping Mixture N Mean Grouping HMA PG 64-22 0% RAP–5% RAS 11 94.2 A HMA PG 58-28 15% RAP–3% RAS 10 90.7 B HMA PG 64-22 15% RAP–3% RAS 10 90.7 B WMA PG 64-22 15% RAP–3%RAS 10 90.1 B Note: df = degrees of freedom; Adj. SS = adjusted sum of squares; Adj. MS = adjusted mean square; N = number of observations. Table 3-7. Density ANOVA for Austin, Texas.

37 But for the purposes of this research, only the outside west- bound lane from Mileposts 113.0 to 91.5 was investigated. The stone matrix asphalt mixtures on this project con- sisted of a 19-mm NMAS gradation with a compactive effort of 80 gyrations. This is considered a 12.5 NMAS mixture per Illinois DOT specifications. The same volumetric mix design was used for the two mixes. The gravel mix was used in the straightaways, while the quartzite mix was used in the hori- zontal curves to provide better friction. Both mixes contained 13% RAP and 5% RAS and were produced as WMA using Evotherm M1. Companion HMA mixtures were not pro- duced on this project. Table 3-8 shows the material percent- ages used for submittal of the JMF. The mixtures contained an SBS polymer-modified PG 70-28 and 0.4% Evotherm M1. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifica- tions are shown in Table 3-9a and Table 3-9b. According to the Illinois State Toll Highway Authority, the special provi- sion requires a minimum VMA of 16. Since there is no deci- mal on the 16, the VMA can actually be as low as 15.5 and rounded up to meet the specification. Site Description. The two test mixes were placed in the right travel lane of the westbound I-88 Tollway from Mileposts 113.0 to 91.5 at a target thickness of 2.25 in. Figure 3-9 shows an example of the gravel stone matrix asphalt transitioning to the quartzite stone matrix asphalt in a curve. Figure 3-10 shows the location of the test sections. Field Performance. Three separate field performance evaluations were conducted in Aurora, Illinois, on June 18, 2014, June 9, 2015, and June 28, 2016, after approximately 22, 34, and 46 months of traffic had been applied to the test sections, respectively. Data were collected on both sections to document performance with regard to rutting, cracking, and raveling by randomly selecting three 200-ft (61-m) data sections within each test section. Each data section was then inspected to assess performance. In addition, five 6-in. (150-mm-) diameter cores Aggregate Type JMF (%) Coarse material No. 16s 21 Coarse material No. 14s 57 Fine material No. 01s 4 RAP 13 RAS 5 Table 3-8. Aggregate percentages used in mix design and production in Aurora, Illinois. Sieve Size Percent Passing Mix Design AASHTO Specification Illinois DOT Specification 19.0 mm (3/4 in.) 100 90–100 100 maximum 12.5 mm (1/2 in.) 82 50–88 82–100 9.5 mm (3/8 in.) 62 25–60 65 maximum 4.75 mm (No. 4) 30 20–28 20–30 2.36 mm (No. 8) 20 16–24 16–24 1.18 mm (No. 16) 18 NA NA 0.6 mm (No. 30) 13 NA 12–16 0.3 mm (No. 50) 11 NA 10–15 0.15 mm (No. 100) 9 NA NA 0.075 mm (No. 200) 7.7 8–1 8–10 Note: NA = not available. Table 3-9a. Design gradation and specifications for Aurora, Illinois, mix design. Variable Mix Design AASHTO Specification Illinois DOT Specification Asphalt content (%) 6.0 6.0 minimum NA Air voids (%) 3.5 3.0–4.0 3.5 VMA (%) 15.6 17.0 minimum 16 minimuma VFA (%) 77.6 NA 75–80 D/A ratio 1.2 NA NA Note: NA = not available; D/A = dust to asphalt. aIllinois Tollway special provision allows mix with 15.5 to be rounded to 16. Table 3-9b. Asphalt content, volumetrics, and specifications for Aurora, Illinois, mix design.

38 were taken from between the wheelpaths for each mix to deter- mine the in-place density after each time period. Rutting. The rut depths were measured at the beginning of each 200-ft section with a straight edge and a wedge. After 22 months, both test sections had performed well with regard to rutting. Out of the six total data sections for the two mixes, five exhibited no measurable rutting. One of the gravel stone matrix asphalt sections had 1/16-in. rutting in the outside wheelpath. At the time of the 2015 and 2016 inspections, neither mix section exhibited any further signs of rutting. Cracking. Each 200-ft section was carefully inspected for visual signs of cracking and rated based on the Distress Identi- fication Manual for the Long-Term Pavement Performance Pro- gram. Both mix sections exhibited transverse cracking. Most Figure 3-9. Example of transition from gravel to quartzite mix in Aurora, Illinois. Figure 3-10. Layout of test sections in Aurora, Illinois.

39 of these cracks appeared to be reflective cracking from the underlying concrete, since cracks were evident across both lanes and the shoulder at these locations. To verify the type of cracking, two cores were taken on transverse cracks in 2015, and two more were taken in 2016. Three of the four cores exhibited evidence of reflective cracking, as suspected. How- ever, the fourth core did not show any evidence of reflective cracking. Therefore, it is surmised that this crack is thermal in nature. Figure 3-11 shows an example of a core taken on one of the reflective cracks, and Figure 3-12 shows the core that was taken on a thermal crack. Since three of the four cores taken on cracks had reflective cracking, it was assumed that the majority of the transverse cracks were reflective cracks, while a smaller percentage of the cracks may have been thermal cracks. Table 3-10 shows the total cracking observed, based on crack severity at the time of each inspection. Figure 3-13 shows examples of the low- and high-severity transverse cracking observed in the gravel stone matrix asphalt and quartzite stone matrix asphalt sections. From the 2014 and 2016 inspections, transverse cracking progressed slowly in extent and severity for both sections. The quartzite stone matrix asphalt had substantially more high-severity cracking during each inspection. Raveling and Weathering. The surface textures of the test sections were determined using the Sand Patch Test. The calculated mean texture depths for both mixes are shown in Table 3-11. These results show that both mixes had very similar textures at the time of both inspections. As seen in these results, the mean texture depths for stone matrix asphalt mixes are typically higher than for dense-graded mixtures. Figure 3-14 and Figure 3-15 show examples of the surface texture for the gravel and quartzite sections, respectively. Core Testing. At the time of each inspection, five 6-in. (150-mm) cores were obtained from both mix sections. The densities of these cores were measured using AASHTO T 166. A summary of the core densities is shown in Figure 3-16. The in-place densities for the two mixes were similar. Figure 3-11. Example of core hole on reflective crack in Aurora, Illinois. Figure 3-12. Example of core hole on thermal crack in Aurora, Illinois. Transverse Cracks 2014 (22 months) 2015 (34 months) 2016 (46 months) Mix Section Severity No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) Gravel stone matrix asphalt Low 10 93.5 13 126.0 15 140.5 Moderate 7 59.0 6 72.0 7 84.0 High 1 12.0 3 36.0 4 48.0 Quartzite stone matrix asphalt Low 7 69.0 17 160.5 16 162.0 Moderate 5 51.5 3 30.0 11 69.0 High 3 36.0 6 72.0 7 132.0 Table 3-10. Total cracking in Aurora, Illinois.

40 (a) Low-Severity Transverse Crack, Gravel (b) Low-Severity Transverse Crack, Quartzite (c) High-Severity Transverse Crack, Gravel (d) High-Severity Transverse Crack, Quartzite Figure 3-13. Cracking observed in Aurora, Illinois.

41 Mix Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation 2014 (22 months) 2015 (34 months) 2016 (46 months) Gravel stone matrix asphalt 1.1 0.1 1.1 0.1 1.1 0.1 Quartzite stone matrix asphalt 1.0 0.2 1.1 0.2 1.2 0.3 Table 3-11. Mean texture depths in Aurora, Illinois. Figure 3-14. Example of surface texture for gravel mix in Aurora, Illinois. Figure 3-15. Example of surface texture for quartzite mix in Aurora, Illinois. 22 Months 34 Months 46 Months In -P la ce D en si ty (% ) Figure 3-16. In-place densities based on cores in Aurora, Illinois.

42 ANOVA was performed to evaluate how the mix (gravel or quartzite), the age of the pavement, and the interac- tion between these two variables affected the in-place den- sity. Table 3-12 shows the results of ANOVA. As can be observed, neither the variables nor their interactions had a significant effect (α = 0.05) on the in-place density for this project. Fort Worth, Texas In October 2012, a field demonstration was conducted in Fort Worth, Texas, with mixes containing RAP and RAS. Two mixes were placed on US 287 in Fort Worth: one as HMA and one as WMA. The mixes were placed on the evenings of October 2 and 3, with APAC Texas, Inc. as the contrac- tor. NCAT was on site for construction of these sections as part of the Asphalt Research Consortium program funded by FHWA. The asphalt mixture used for this trial was a 9.5-mm NMAS design using the Texas Gyratory Compactor at 75 gyrations. Although separate mix designs were performed for the two mixes, both used the same JMF except for the WMA additive. The aggregate used for the design and construction was a locally available crushed limestone. The mixes consisted of 15% RAP and 5% RAS. The RAP was a multiple-source milling that had been crushed to –½ in., and the RAS was MW RAS that was shredded to –¼ in. The materials percent- ages used for mix design submittal and production are shown in Table 3-13. The mixes used a PG 64-22 virgin liquid asphalt binder supplied by Hartland Asphalt Materials in Saginaw, Texas. The WMA mix was produced using the chemical addi- tive Cecabase RT, which is produced by Arkema Group. The Cecabase RT had been pre-blended with Ad-Here LOF 6500 at the terminal for use as an antistrip. This blend was metered in at the plant at a rate of 0.5% by weight of total asphalt. The HMA used 1.0% Akzo Nobel Kling Beta 2550 HM as the antistripping agent. The JMF, optimum asphalt content, and specifications are shown in Table 3-14a and Table 3-14b. Production. The plant used for this project was located on Cold Springs Road in Fort Worth, Texas, approximately 5 to 8 mi from the paving location, which yielded an approximate travel time of 8 to 20 min. The plant, shown in Figure 3-17, was an Astec parallel-flow drum plant that used a separate mini-drum to mix the liquid with the aggregate, which is shown at the left of the picture under the rightmost silo. Source df Adj. SS Adj. MS F-Value P-Value Mix 1 3.71 3.71 2.6 0.12 Age 2 2.53 1.27 0.89 0.424 Mix × Age 2 0.24 0.12 0.08 0.92 Error 24 34.18 1.42 Total 29 40.65 Table 3-12. Density ANOVA for Aurora, Illinois. Aggregate Type Mix Design (%) Production (%) Type D limestone rock 46.7 51.0 Limestone screenings 28.0 23.0 Sand 6.0 6.0 RAP 15.0 15.0 RAS 4.3 5.0 Table 3-13. Aggregate percentages used in mix design and production. Sieve Size Percent Passing JMF Specification 19.0 mm (3/4 in.) 100.0 100–100 12.5 mm (1/2 in.) 99.9 98–100 9.5 mm (3/8 in.) 95.0 85–100 4.75 mm (No. 4) 63.2 50–70 2.36 mm (No. 8) 40.8 35–46 0.6 mm (No. 30) 20.9 15–29 0.3 mm (No. 50) 15.7 7–20 0.075 mm (No. 200) 4.3 2–7 Table 3-14a. Design gradation for Fort Worth, Texas, mix design. Variable JMF Specification Asphalt content (%) 5.1 4.8–5.4 Air voids (%) 3.0 NA VMA (%) 15.0 VFA (%) 80.0 NA D/A ratio 1.02 NA Note: NA = not available. Table 3-14b. Asphalt content, volumetrics, and specifications for Fort Worth, Texas, mix design.

43 Production temperatures and rates were monitored and recorded throughout production of the two mixes. Table 3-15 shows production temperature information, and Table 3-16 shows the production rates and totals. Volumetric Mix Properties. The two mixtures were sampled each night after a few hundred tons had been produced and the temperature and mix variations stabi- lized, as observed by NCAT and APAC personnel. The mix sample was used to fabricate a variety of specimens for deter- mining volumetric properties and performance tests in the on-site NCAT mobile laboratory and later in NCAT’s main laboratory. One large sample was taken to ensure consistency between the tests. Specimens were compacted using 50 gyrations in the Superpave Gyratory Compactor, which was assumed to be equivalent to 75 gyrations in the Texas Gyratory Com- pactor. The volumetric samples were PMLC on site in the NCAT mobile laboratory to avoid possible effects of reheat- ing. This is often referred to as being hot-compacted. The samples were placed in an oven for a brief time after sam- pling only to get back up to the compaction temperature. Water absorption results were below 2.0%; therefore, the Gmb were determined in accordance with AASHTO T 166. The asphalt content and gradation of each mix was tested according to AASHTO T 309 and AASHTO T 30, respec- tively, at NCAT’s main laboratory. The volumetric prop- erties, asphalt content, and gradation results are shown in Table 3-17a and Table 3-17b. The gradations for both mixes were similar to the JMF on every sieve except the percentage passing the No. 200 sieve. Both mixes had much higher dust content than the JMF. This difference was because NCAT results were based on washed gradations, whereas the Texas DOT mix designs used dry gradations. It should also be noted that Texas DOT calculates VMA using effective spe- cific gravity (Gse) instead of bulk specific gravity (Gsb). Both methods for determining VMA are shown in Table 3-17a and Table 3-17b. When calculating VMA using the Texas DOT method, the production values are similar to the design. However, this is not the case when using the AASHTO method with Gsb. Construction. The two mixes were placed on the inside lane of US 287 southbound, a 5- to 8-mi haul from the asphalt plant, which took approximately 8 to 20 min. The compacted mat had a target thickness of 2.0 in. Figure 3-18 shows the test section layout. The mixtures were delivered to the site using nine to 10 live-bottom Flow Boy trucks and one tandem-axle dump truck. A RoadTec SB2500E Shuttle Buggy was used to trans- fer the mixtures from the trucks to the Terex Cedarapids CR paver. The paver is shown in Figure 3-19. The temperature of the mix was measured behind the paver every 5 to 30 min with a hand-held temperature gun. These temperature data are summarized in Table 3-18. The WMA section was placed approximately 30°F lower than the HMA section. Two rollers were used for compaction of the two mixes to target a minimum in-place density of 91.5%. The breakdown roller was a Sakai SW900 steel wheel roller operated in vibra- tory mode. The rolling pattern for the breakdown roller was two passes down each side followed by one pass down the middle of the lane. The finishing roller was a Sakai GW750-2 rubber tire roller operated in vibratory mode. The finishing roller used the same rolling pattern as the breakdown roller. Figure 3-17. Asphalt plant in Fort Worth, Texas. Statistic HMA Temperature (°F) WMA Temperature (°F) Average 293.0 264.4 Standard deviation 11.8 8.6 Maximum 310.0 295.0 Minimum 275.0 258.0 Table 3-15. Production temperatures in Fort Worth, Texas. Production HMA WMA Average production rate (tph) 213 197 Total tons shipped 1,229 1,400 Note: tph = tons per hour. Table 3-16. Production rates and totals in Fort Worth, Texas.

44 Note: na = not applicable; NA = not available. aNo WMA technology. Sieve Size Percent Passing JMFa HMAa WMA (Cecabase) Tolerance 25.0 mm (1 in.) 100.0 100.0 100.0 na 19.0 mm (3/4 in.) 100.0 100.0 100.0 ± 5.0 12.5 mm (1/2 in.) 99.9 99.7 99.4 ± 5.0 9.5 mm (3/8 in.) 95.0 96.6 95.1 ± 5.0 4.75 mm (No. 4) 63.2 68.5 67.2 ± 5.0 2.36 mm (No. 8) 40.8 40.5 40.5 ± 5.0 1.18 mm (No. 16) NA 28.4 28.6 ± 3.0 0.60 mm (No. 30) 20.9 21.9 22.2 ± 3.0 0.30 mm (No. 50) 15.7 16.7 17.2 ± 3.0 0.15 mm (No. 100) NA 9.5 10.3 ± 3.0 0.075 mm (No. 200) 4.3 6.1 6.9 ± 2.0 Table 3-17a. Gradation for Fort Worth, Texas. Note: NA = not available; Gmm = theoretical maximum specific gravity; Pba = percent binder absorbed; Pbe = percent binder effective; D/B = dust to binder. aNo WMA technology. b4.9 production. Variable JMFa HMAa WMA (Cecabase) Tolerance Asphalt content (%) 5.1b 4.75 4.97 ± 0.3 Air voids (%) 3.0 4.4 3.3 ± 1.0 Gmb @ Ndesign 2.419 2.386 2.403 NA Gmm 2.494 2.495 2.486 NA AASHTO VMA (%) 12.9 13.8 13.4 NA Texas DOT VMA (%) 14.9 15.4 14.9 >14.0 VFA (%) 76.8 68.2 75.0 NA Gsb 2.636 NA NA NA Gse 2.699 2.685 2.684 NA Pba (%) 0.91 0.72 0.71 NA Pbe (%) 4.24 4.06 4.30 NA D/B ratio 1.01 1.51 1.61 NA Table 3-17b. Asphalt content and volumetrics for Fort Worth, Texas.

45 Figure 3-18. Location of test sections in Fort Worth, Texas.

46 Figure 3-20 shows an example of the surface of the mat during construction. Ten cores were obtained from the HMA mix section, and six were obtained from the WMA sections. The densities of these cores are summarized in Figure 3-21. The WMA results averaged about 1.6% higher density compared to the HMA. Field Performance. Field performance evaluations were conducted on the Fort Worth project on September 24, 2014, and October 22, 2015, after approximately 24 and 37 months of traffic had been applied to the two mix sections. Perfor- mance data were collected in the same manner as the other projects. In addition, five 6-in. (150-mm-) diameter cores were taken from between the wheelpaths for each mix to determine the in-place density after 24 months. Rutting. At the time of each inspection, neither test section had any measurable rutting after 24 months or 37 months. Cracking. At the time of the first inspection, one of the three HMA data sections had a single low-severity transverse crack. As shown in Figure 3-22, this crack was across two of the three southbound lanes and partially across the third lane. This crack was not believed to be mix-related but rather caused by reflective cracking from the underlying material. Figure 3-19. Paver used in Fort Worth, Texas. Statistic HMA Temperature (°F) WMA Temperature (°F) Average 284.0 252.2 Standard deviation 9.8 7.6 Maximum 301.5 266.0 Minimum 264.0 225.0 Table 3-18. Temperatures behind the screed in Fort Worth, Texas. Figure 3-20. Example of mat during construction in Fort Worth, Texas. In -P la ce D en sit y (% ) Figure 3-21. In-place densities based on cores at construction in Fort Worth, Texas. Figure 3-22. Transverse crack observed in HMA section in Fort Worth, Texas.

47 (a) 2014 (b) 2015 Figure 3-23. Longitudinal crack observed in WMA section in Fort Worth, Texas. Only one small crack was observed in the three data sec- tions for the Cecabase WMA mix. As shown in Figure 3-23a, this was a low-severity longitudinal crack that measured about 10-ft long. At the time of the second inspection, no new cracks were found. The small longitudinal crack in the Cecabase section had extended from 10 to 21 ft in length. Figure 3-23b shows this longitudinal crack at the time of the second inspection. It can be seen that the crack did not increase in severity; only in length. Overall, both mixes performed well with regard to crack- ing after 37 months in service. However, there are numerous transverse cracks on the existing shoulder along both mix sec- tions. These cracks should be monitored to determine if they will propagate into the test sections. Figure 3-24 shows an example of the cracking observed on the shoulder through- out both mix sections. Raveling and Weathering. The mean texture depths for both mixes—based on Sand Patch Tests—are shown in Table 3-19. These results show that both mixes had simi- lar mean texture depths at the time of both inspections. These values are consistent with a good-performing dense- graded pavement with normal wear having no raveling or segregation. Flushing. Some small spots—hand size and smaller—of binder flushing were observed in both the HMA and Ceca- base mix sections. Although these spots occurred throughout the project, some areas were worse than others. Figure 3-25 and Figure 3-26 show an example of flushing observed at the time of the inspections (both in 2014 and 2015). Core Testing. Five 6-in. (150-mm) cores were taken from both mix sections. The densities of these cores were determined first using AASHTO T 166. However, most of the cores had water absorptions greater than 2.0%, requiring the densities to be determined according to AASHTO T 331 instead. A summary of the core densities from the 24-month Figure 3-24. Example of transverse cracks on shoulder in Fort Worth, Texas.

48 Mix Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation 2014 (24 months) 2015 (37 months) HMA 0.6 0.0 0.6 0.1 Cecabase 0.6 0.0 0.6 0.0 Table 3-19. Mean texture depths in Fort Worth, Texas. Figure 3-25. Example of binder flushing and pooling on surface in Fort Worth, Texas. Figure 3-26. Close up of binder flushing and pooling on surface in Fort Worth, Texas. and 37-month inspections is shown in Figure 3-27, along with the densities measured by NCAT at construction. Dur- ing construction, the cores were obtained throughout the day’s paving operations. At the time of the inspections, the cores were taken near the 200-ft data sections for traffic con- trol and safety reasons. Therefore, some variation in density results was expected. ANOVA was performed to determine whether the mix type (HMA or WMA), age of the pave- ments, and the interaction of mix type and age had a statisti- cal effect on the in-place density. Table 3-20 shows the results of ANOVA. As can be observed, the in-place density results were not significantly affected by mix type, age, or their inter- actions for this project. New Projects The study also included sampling of mixtures from overlay projects in five states. For each project, pave- ment test sections were built with and without the addi- tion of a WMA technology. A couple of the projects also included additional experimental factors. The following sections describe these projects, the sampled mixtures, mix production information, the basic mix properties and as-constructed densities, as well as the performance of the test sections over the first few years of service. For the new projects, the percentage of active binder for the RAS stock- piles was assumed to be 100%, although that is unlikely to be the case. The mix designs for each project were prepared in accor- dance with the specifications of the respective state DOT at the time of the project letting. None of the state DOTs required any kind of mix performance testing as part of mix design or QA. Although the research team was involved in the selection of the projects to meet the research objectives, no additional requirements or recommendations were made to the agencies or the contractors. Larsen, Wisconsin A field project with mixes containing RAP and RAS was constructed on SR-96 near Larsen, Wisconsin, in September 2013. Three mixes were produced as part of this project: Con- trol HMA, Rediset WMA, and Zycotherm WMA. The three mixes were produced and placed over a span of 3 days from September 10 through 12. Each mix was placed in a two-lane portion of SR-96 by Northeast Asphalt.

49 The asphalt mixtures consisted of a fine-graded 12.5-mm NMAS Superpave mix design with a compactive effort of 75 gyrations. The same mix design was used for all three mixes, with the only difference between the HMA and WMA mixes being the addition of the WMA additives. Each of the mixes contained 14% RAP, 3% RAS, and a limestone vir- gin aggregate. The RAP used was a multiple-source–½-in. crushed RAP, while the RAS was a PC–RAS obtained by Northeast Asphalt from several local roofing contractors. Table 3-21 shows the material percentages used for the mix design and production. The asphalt mixtures used a PG 58-28 asphalt binder sup- plied by Construction Resource Management in Green Bay, Wisconsin. While an antistrip was not needed for the control 0 Month 24 Months 37 Months In -P la ce D en si ty (% ) Figure 3-27. In-place densities based on cores in Fort Worth, Texas. Source df Adjusted SS Adjusted MS F-Value P-Value Mix type 1 1.16 1.16 0.37 0.552 Age 1 0.05 0.05 0.02 0.904 Mix type × Age 1 0.01 0.01 0 0.955 Error 16 50.11 3.13 Total 19 51.32 Table 3-20. Density ANOVA for Fort Worth, Texas. Aggregate Type Mix Design (%) Production Control (%) Production Rediset (%) Production Zycotherm (%) 5/8-in. x 1/2-in. chips 11 10 10 10 1/2-in. x 1/4-in. chips 8 9 9 9 Manufactured sand 36 35 35 35 Natural sand 27.5 29 29 29 Baghouse fines 0.5 Return 100 Return 100 Return 100 Fine RAP 14 14 14 14 PC–RAS 3 3 3 3 Table 3-21. Aggregate percentages used in mix design and production in Larsen, Wisconsin.

50 plant operator for monitoring the amount of each material being introduced. Second, the drum was fitted with two separate recycled material collars to better control feed rates for RAP and RAS. The RAP was introduced about 8 ft prior to the RAS, giving the RAP more time to dry and increase in temperature. Figure 3-28 shows a general over- view of the Northeast Asphalt plant used for this project. Figure 3-29 shows a close-up of the two recycled material collars entering the drum. Table 3-23 provides a summary of production tem- peratures, and Table 3-24 shows the production rates and totals. Volumetric Mix Properties. NCAT personnel collected three mix samples during production from each mix. The first sample was taken at the beginning of production each day after a couple hundred tons had been produced. The first sample of each mix was used to fabricate specimens for deter- mining volumetric properties and performance tests. The first samples were taken at one time to ensure mix consistency between the performance tests. This process of sampling and compacting specimens was followed for each of the new proj- ects in this study. Volumetric specimens were compacted on site to 75 gyra- tions in the Superpave Gyratory Compactor. The compac- tion temperature for each mix was determined using the average compaction temperature observed on the test sec- tion through the first couple of hours of construction for each mix. Water absorption levels were below 2.0%, so Gmb results using AASHTO T 166 were used. Asphalt content and gra- dations of each mix were tested at the main NCAT labora- tory per AASHTO T 164 and AASHTO T 30, respectively. Sieve Size Percent Passing Mix Design Specification 19.0 mm (3/4 in.) 100.0 100 maximum 12.5 mm (1/2 in.) 96.9 90–100 9.5 mm (3/8 in.) 89.3 90 maximum 4.75 mm (No. 4) 76.2 NA 2.36 mm (No. 8) 55.9 28–58 1.18 mm (No. 16) 41.7 NA 0.6 mm (No. 30) 30.9 NA 0.3 mm (No. 50) 16.1 NA 0.15 mm (No. 100) 6.8 NA 0.075 mm (No. 200) 4.2 2–10 Note: NA = not available. Table 3-22a. Design aggregate gradation for Larsen, Wisconsin, mix design. Variable Mix Design Specification Asphalt content (%) 5.4 NA Air voids (%) 4.0 NA VMA (%) 14.5 >14.0 VFA (%) 72.4 65–75 D/A ratio 0.90 NA Note: NA = not available. Table 3-22b. Asphalt content, volumetrics, and specifications for Larsen, Wisconsin, mix design. mix, both WMA additives have antistripping properties. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifications are shown in Table 3-22a and Table 3-22b. Production. The production temperature was not low- ered significantly for the two WMA mixes compared to the control HMA mix, since the contractor was uncomfortable decreasing the production temperature below 300°F for mixes containing RAS. Furthermore, the night before pro- duction of the first WMA mix (Rediset), it rained approxi- mately ½ of an inch, which meant the virgin aggregates had to be heated longer to remove the moisture. Rediset additive was pumped into the virgin binder line during production using a portable pump system at a rate of 0.50% by mass of virgin binder. Zycotherm additive was introduced and blended on site in one of the binder tanks at a rate of 0.12% by mass of virgin binder. The mixes were produced using a modified counter-flow drum plant manufactured by Dillman Equipment, Inc. The burner used recycled waste oil. Several modifications had been made to this plant. First, load cells were installed on each cold-feed bin to provide real-time feedback to the Figure 3-28. Dillman-modified counter-flow plant in Larsen, Wisconsin.

51 Table 3-25a and Table 3-25b show the results from NCAT’s testing on the three Wisconsin mixes. The values shown in these tables are based on NCAT’s work with the large sample taken once the mix production was stable. The contractor’s QC data is shown in Appendix A. The asphalt content of each mix was determined both by the ignition method (AASHTO T 308) and solvent extrac- tion (AASHTO T 164) using trichloroethylene (TCE). The binder was recovered and graded after the solvent extraction. The average asphalt content from both methods are shown in Table 3-26. There are substantial differences between the ignition and extraction results, likely caused by the dolo- mitic limestone used in this area, which has a high and some- what erratic correction factor in the ignition furnace. For the RAP and RAS samples, asphalt content results from sol- vent extraction were lower than from ignition method. For the RAP, the difference can be attributed to an uncorrected aggregate mass loss. For the PC–RAS, cellulose fibers were burned in the ignition method, yielding a false high asphalt content result. Table 3-27 shows the binder grade test results for the virgin binder, as well as the binders recovered from the recycled materials and the mixtures. The true grades and DTc (20-h pressure aging vessel [PAV]) results for the bind- ers recovered from the control mix and the Rediset mix were similar; the true grade of the Zycotherm mix was slightly higher, and its DTc (20-h PAV) was slightly lower. Construction. The three test sections are located on SR-96 near Larsen, Wisconsin. The portion of SR-96 paved during this field demonstration was approximately 8 mi from the asphalt plant, with a haul time of about 10 to 15 min. The existing roadway consisted of 8 to 10 in. of concrete that was rubblized. For the new construction, a 2.25-in. Figure 3-29. Recycled collar modifications in Larsen, Wisconsin. Statistic Control Temperature (°F) Rediset Temperature (°F) Zycotherm Temperature (°F) Average 323.9 316.7 321.0 Standard deviation 10.0 8.6 18.1 Maximum 343.0 339.0 351.0 Minimum 295.0 303.0 252.0 Table 3-23. Production temperatures in Larsen, Wisconsin. Production Control Rediset Zycotherm Average production rate (tph) 295 308 292 Total tons shipped 2,527 2,543 2,580 Table 3-24. Production rates and totals in Larsen, Wisconsin. Sieve Size Percent Passing JMFa Controla Rediset Zycotherm Tolerance 19.0 mm (3/4 in.) 100.0 100.0 100.0 100.0 100 12.5 mm (1/2 in.) 96.9 96.8 97.0 97.2 91.4–100 9.5 mm (3/8 in.) 89.3 88.7 89.4 89.4 83.8–94.8 4.75 mm (No. 4) 76.2 74.7 75.1 75.3 NA 2.36 mm (No. 8) 55.9 55.0 55.5 55.7 50.9–60.9 1.18 mm (No. 16) 41.7 41.0 41.5 41.5 NA 0.60 mm (No. 30) 30.9 30.9 31.1 31.0 NA 0.30 mm (No. 50) 16.1 16.3 16.4 15.9 NA 0.15 mm (No. 100) 6.8 6.9 6.7 6.4 NA 0.075 mm (No. 200) 4.2 4.3 4.0 3.8 2.2–6.2 Note: NA = not available. aNo WMA technology. Table 3-25a. Gradation for Larsen, Wisconsin.

52 intermediate course was placed, followed by a 1.75-in. surface course. However, in the portion of SR-96 near the intersec- tion of County Road M, there are several utilities embedded in the original concrete roadway. For this portion of roadway, the concrete was not rubblized but only patched, which could lead to reflective cracking in the new surface mixes. Since the intermedi- ate course had been placed prior to NCAT’s arrival, a condition survey of the underlying concrete layer could not be performed. NCAT was on site for the construction of the three surface mixes. The control mix was placed in the eastbound lane, while the Rediset and Zycotherm mixes were placed in the westbound lane. The mixes were placed as surface mixes Variable JMFa Controla Rediset Zycotherm Tolerance Asphalt content (%) 5.40 5.46 5.31 5.42 5.0–5.8 Air voids (%) 4.0 4.4 4.4 4.4 2.7–5.3 Gmb @ Ndesign 2.410 2.404 2.417 2.396 NA Gmm 2.510 2.515 2.527 2.507 NA VMA (%) 14.5 15.5 15.0 15.8 >13.5 VFA (%) 72.4 71.7 70.8 72.0 NA Gsb 2.668 2.691 2.691 2.691 NA Gse 2.732 2.741 2.749 2.729 NA Pba (%) 0.91 0.70 0.81 0.54 NA Pbe (%) 4.54 4.79 4.54 4.91 NA D/B Ratio 0.90 0.90 0.89 0.77 NA Note: NA = not available. aNo WMA technology. Table 3-25b. Asphalt content and volumetrics for Larsen, Wisconsin. Material Corrected Ignition Oven (Average) TCE Extraction (Average) Difference (Ignition– Extraction) Control mix 5.98 5.46 0.52 Rediset mix 5.76 5.31 0.46 Zycotherm mix 5.55 5.42 0.13 RAP 6.21 4.32 1.90 RAS 30.75 25.77 4.98 Table 3-26. Asphalt content test results from Larsen, Wisconsin. Material Tc High Tc Int. Tc Low True Grade PG Tc (20-h PAV) Virgin PG 58-28 59.1 16.6 –30.9 59.1-30.9 58-28 1.4 Control mix 76.6 20.9 –25.2 76.6-25.2 76-22 –3.5 Rediset mix 79.0 23.5 –24.3 79.0-24.3 76-22 –3.8 Zycotherm mix 83.5 23.7 –21.1 83.5-21.1 82-16 –5.0 RAP 86.3 30.2 –18.8 86.3-18.8 82-16 –1.7 RAS 145.3 36.0 +12.0 145.3+12.0 142+14 –45.6 Note: Tc = Tcritical; Int. = intermediate. Table 3-27. Performance-grade test results from Larsen, Wisconsin. at a target thickness of 1.75 in. A CSS-1H emulsion was used as the tack coat at a total diluted rate of 0.025 gal/yd2. Figure 3-30 shows the layout of the test sections. The asphalt mixes were delivered to the paving site using a combination of 10 live-bottom beds and semidump trucks. A RoadTec SB-25 Shuttle Buggy was used to trans- fer the mixes to the Blaw–Knox PF-R200 paver, as shown in Figure 3-31. The temperature of the mix behind the paver was mea- sured every 10 to 30 min using a hand-held temperature gun. The temperatures measured behind the screed for each mix are summarized in Table 3-28.

53 Figure 3-30. Location of test sections near Larsen, Wisconsin. Figure 3-31. Material transfer vehicle (MTV) transferring mix to paver in Larsen, Wisconsin. Statistic Control Temperature (°F) Rediset Temperature (°F) Zycotherm Temperature (°F) Average 290.2 288.9 287.7 Standard deviation 6.3 4.5 8.7 Maximum 302.5 298.0 300.0 Minimum 273.0 280.0 273.0 Table 3-28. Temperatures behind the screed in Larsen, Wisconsin.

54 Three rollers were used for compaction to target a mini- mum in-place density of 91.5%. The breakdown roller— a Sakai SW850—was operated in vibratory mode for five passes. The intermediate roller was a Bomag BW 27RH pneumatic roller. There was no set rolling pattern, and the tire pressures were set to 87 psi. The finishing roller was a Hypac C778A, operated in static mode for seven passes. Figure 3-32 shows the breakdown roller compacting the mat. Three cores were taken for NCAT from each mix sec- tion the day after construction of each mix. Figure 3-33 shows the densities of these cores. A two-sample t-test was used to determine if density results were significantly dif- ferent for the three test sections. At a confidence level of 95%, there was no statistical difference in density between the control section and either of the two WMA sections (p-values 0.131 and 0.223). Field Performance at 13-Month and 24-Month Project Inspections. The first field performance evaluation was conducted on October 10, 2014, approximately 13 months after construction. A second evaluation was performed on October 8, 2015, after 25 months. For this project, the data sections were marked at the time of construction based on the location of the three mix samples NCAT took during production. Cores were obtained from between the wheel- paths for each mix to determine the in-place densities after 13 and 25 months. Rutting. At the time of both inspections, none of the three mix sections had any measurable rutting. Cracking. All three mix sections performed very well with regard to cracking. For both the control and Zycotherm mixes, only one of the three data sections had crack- ing, but these were where the underlying concrete pave- ment had not been rubblized. The rest of the sections, including all three Rediset sections, were placed over rub- blized concrete. There were no cracks in any of these sec- tions. The cracks observed in the control and Zycotherm sections were reflective cracks directly over concrete joints and were not considered to be an indication of mix stiff- ness or brittleness. Table 3-29 summarizes the cracking observed at the 13-month inspection. Figure 3-34 and Figure 3-35 show an example of the transverse reflec- tive cracks seen in the control and Zycotherm sections, respectively. At the 25-month inspection, no new cracks were observed. However, in the control section, the single longitudinal crack Figure 3-32. Breakdown roller compacting mat in Larsen, Wisconsin. In -P la ce D en sit y (% ) Figure 3-33. In-place densities based on cores at construction in Larsen, Wisconsin.

55 they can break and cause surface pop-outs. These pop-outs were not observed to have worsened in severity between the two inspections. An example of these pop-outs is shown in Figure 3-37. Field Cores. During the inspections, cores were taken from each mix section. The densities of these cores were measured using AASHTO T 166, since all water absorp- tions were below 2.0%. A summary of the core densi- ties from construction and the inspections is shown in Figure 3-38. Mix Section Severity Wheelpath Longitudinal Nonwheelpath Longitudinal Transverse No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) No. of Cracks Total Length (ft) Control Low 0 0 1 11 9 111 Moderate 0 0 0 0 1 12.3 High 0 0 0 0 0 0 Rediset Low 0 0 0 0 0 0 Moderate 0 0 0 0 0 0 High 0 0 0 0 0 0 Zycotherm Low 0 0 0 0 1 3 Moderate 0 0 0 0 4 49.3 High 0 0 0 0 0 0 Table 3-29. Total cracking in Larsen, Wisconsin, after 13 months. Figure 3-34. Low-severity transverse crack observed in control section in Larsen, Wisconsin, after 13 months. Figure 3-35. Medium-severity transverse crack observed in Zycotherm section in Larsen, Wisconsin, after 13 months. had extended in length from 11 to 19 ft, and two of the trans- verse cracks had progressed from low severity to medium severity. Figure 3-36 shows one of the medium-severity cracks from the 25-month inspection. Raveling and Weathering. The texture results for the three sections, shown in Table 3-30, were very similar and had acceptable mean texture depths at the time of the inspection. It was not possible to measure mean texture depths during the second inspection, since the sections were wet from rain- fall before and during the evaluation. Some small aggregate pop-outs were observed through- out the sections. The contractor stated that these types of pop-outs are common in the area and that they were likely caused by highly porous chert particles commonly found in the area. When these particles absorb water and freeze,

56 While the control section had a slightly higher den- sity compared to the two sections with WMA additives at the time of construction, the densities of all sections increased with traffic and were similar by the first inspec- tion. ANOVA was performed to determine how the mix type (HMA or WMA), the age of the pavement, and the interaction between mix type and age affected the density. As can be seen in Table 3-31, only the age had a significant effect on in-place density for this project with a p-value of 0.001. Table 3-31 also shows the statistical groupings from Tukey’s Test of Multiple Comparisons. Factors with means that do not share a letter are significantly different at α = 0.05. Therefore, there was no statistical difference between den- sities measured at 13 and 24 months; however, the density obtained right after construction was statistically different (lower) than from the two inspections. Enterprise, Alabama In June 2014, a field project with four test sections was constructed on US-84 near Enterprise, Alabama. Four mixes containing RAP and RAS were included in this project: low air void HMA, low air void WMA, adjusted air void HMA, and adjusted air void WMA. These four mixes were produced and placed by Wiregrass Construction on June 13 and June 16 to 18 on a four-lane portion of US-84 just north of Enterprise, Alabama. All of the mixes were based on a single HMA mix design. The design was a fine-graded 12.5-mm NMAS Superpave mixture using a 60 gyration compactive effort and a target air void content of 4.1%. WMA mixes for this project used water-injected foamed asphalt. The mix design was not altered for the WMA mixes. The original plan was to produce this mix as designed—both as HMA and WMA—and then, on subsequent days, increase the asphalt content to achieve 3.5% air voids produced as both HMA and WMA. The same aggregate blend was used for all four mixes. However, on the first day of paving, the laboratory air void results from NCAT’s mobile laboratory were much Figure 3-36. Medium-severity transverse crack observed in control section in Larsen, Wisconsin, after 25 months. Mix Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation Construction (September 2013) 13 Months (October 2014) Control 0.4 0.0 0.4 0.0 Rediset 0.5 0.0 0.4 0.0 Zycotherm 0.5 0.0 0.4 0.0 Table 3-30. Mean texture depths in Larsen, Wisconsin, after 13 months. Figure 3-37. Example of aggregate pop-outs observed throughout mixes in Larsen, Wisconsin.

57 lower than the QC laboratory’s results. Upon investigation, it was learned that the QC laboratory was not heating their gyratory molds but instead used cold molds for compaction, which caused higher air voids for their QC samples. Therefore, the true air voids were lower. The decision was made to desig- nate this mix as the Low Air Void WMA and keep everything the same for the companion HMA (Low Air Void HMA). The other two mixes were then adjusted to yield at least 1% higher air voids. These mixes were designated Adjusted Air Void WMA and Adjusted Air Void HMA. The actual air voids results from NCAT are provided later in this report. All four mixes contained 15% RAP and 5% RAS, with a gravel and limestone blend used as the virgin aggregate. The RAP used was a single-source–½ in. crushed RAP. The RAS was a PC–RAS procured from local roofing contractors, which was blended with the RAP using a loader on site so that both materials could be added using one cold-feed bin. Table 3-32 shows the material percentages used for the mix design and production. The mixtures used a PG 67-22 asphalt binder supplied by Ergon, Inc. in Bainbridge, Georgia. All mixes contained 0.5% Ad-Here LOF by mass of virgin binder as an antistrip- ping agent. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifications are shown in Table 3-33a and Table 3-33b. Production. The WMA mixes were produced using a Gencor Green Machine foamer that injected water at a rate of 1.5% by mass of virgin binder. All mixes were produced using a counter-flow drum plant manufactured by CMI. The burner fuel was recycled waste oil. Figure 3-39 shows the asphalt plant in Ariton, Alabama, used by Wiregrass Construction, Inc. for this project. Figure 3-40 shows the Gencor Green Machine used for foaming the virgin binder. 0 Month 13 Months 25 Months In -P la ce D en si ty (% ) Figure 3-38. In-place densities based on cores in Larsen, Wisconsin. Source df Adj. SS Adj. MS F-Value P-Value Mix type 2 0.39 0.19 0.30 0.740 Age 2 31.19 15.59 24.46 0.001 Mix type × Age 4 2.20 0.55 0.86 0.498 Error 30 19.13 0.64 Total 38 52.78 Statistical Grouping Age (months) N Mean Grouping 13 15 93.4 A 24 15 92.8 A 0 9 91.1 B Table 3-31. Density ANOVA for Wisconsin.

58 Aggregate Type Mix Design (%) Production Low Air Void WMA (%) Production Low Air Void HMA (%) Production Adjusted Air Void WMA (%) Production Adjusted Air Void HMA (%) ½ in. crushed gravel 39 36 36 36 36 Shot gravel 8 8 8 8 8 No. 78s limestone 7 10 10 10 10 Limestone screenings 7 7 7 7 7 Sand 18 19 19 18 18 Baghouse fines 1 Return 100 Return 100 Return 100 Return 100 RAP 15 15 15 15 15 RAS 5 5 5 5 5 Table 3-32. Aggregate percentages used in mix design and production in Enterprise, Alabama. Sieve Size Percent Passing Mix Design Specification 19.0 mm (3/4 in.) 100 100 maximum 12.5 mm (1/2 in.) 97 90–100 9.5 mm (3/8 in.) 90 90 maximum 4.75 mm (No. 4) 64 NA 2.36 mm (No. 8) 48 28–58 1.18 mm (No. 16) 39 NA 0.6 mm (No. 30) 29 NA 0.3 mm (No. 50) 16 NA 0.15 mm (No. 100) 10 NA 0.075 mm (No. 200) 6.5 2–10 Note: NA = not available. Table 3-33a. Design aggregate gradation for Enterprise, Alabama, mix design. Variable Mix Design Specification Asphalt content (%) 5.1 5.1 minimum Air voids (%) 4.1 NA VMA (%) 15.5 15.5 VFA (%) 73.8 65–75 D/A ratio 1.29 0.6–1.60 Note: NA = not available. Table 3-33b. Asphalt content, volumetrics, and specifications for Enterprise, Alabama, mix design. Figure 3-39. CMI counter-flow plant in Ariton, Alabama. Figure 3-40. Gencor Green Machine in Ariton, Alabama.

59 Production temperatures and rates recorded throughout production of the four mixes are summarized in Table 3-34. Production temperatures for the WMA mixtures were gener- ally 40°F to 50°F lower than the HMA mixtures. Volumetric Mix Properties. Volumetric specimens were compacted using 60 gyrations in a Superpave Gyratory Com- pactor. The Gmb of the compacted specimens was determined in accordance with AASHTO T 166, since water absorption values were below 2.0%. Asphalt content and gradation of each mix were tested per AASHTO T 164 and AASHTO T 30, respectively. Table 3-35a and Table 3-35b show the results from NCAT’s testing on the four mixes. These results are based on the large initial sample obtained once the mix production was stable. The contractor’s QC data is shown in Appendix A. As previously noted, the initial production of HMA and WMA with RAS was intended to target the JMF. Because of QC problems, these mixes had very low air void content and were consequently referred to as the low air void samples. The gradations of these low air void mixes were substantially coarser than the JMF, which likely contributed to the prob- lem. Also, the asphalt content of the WMA mixture was 0.39% above the JMF target, whereas the asphalt content of the HMA sample was close to the target. Rather than increase the asphalt content as planned for the additional test sections, the con- tractor adjusted the mix gradation and reduced the asphalt content to get the air void content into an acceptable range. Thus, the third and fourth sections are referred to as adjusted air void sections or mixtures. Calculated asphalt absorptions for all of the plant-produced mixtures were higher than that of the mix design, but there was not a substantial difference in asphalt absorption between companion WMA and HMA samples. Except for the low air void WMA mixture, the effec- tive asphalt content of the plant-produced mixtures were about 0.3% to 0.6% lower than the mix design. The asphalt content of each mix was determined both by the ignition method and TCE extraction. The binder was recovered and graded after the extractions. The average Statistic Low Air Void WMA (°F) Low Air Void HMA (°F) Adjusted Air Void WMA (°F) Adjusted Air Void HMA (°F) Average 304.2 350.0 311.5 351.1 Standard deviation 33.8 10.6 12.5 5.4 Maximum 348.0 366.0 330.0 357.0 Minimum 225.0 324.0 270.0 340.0 Average production rate (tph) 230 247 264 262 Total tons shipped 678 1,730 2,167 1,805 Table 3-34. Production temperatures and rates in Ariton, Alabama. Sieve Size Percent Passing JMFa Low Va WMA (Gencor foam) Low Va HMAa Adjusted Va WMA (Gencor foam) Adjusted Va HMAa Tolerance 19.0 mm (3/4 in. ) 100.0 100.0 100.0 100.0 99.8 ± 7 12.5 mm (1/2 in. ) 97.0 93.5 94.1 97.3 94.6 ± 7 9.5 mm (3/8 in.) 90.0 83.6 84.8 88.2 85.2 ± 7 4.75 mm (No. 4) 64.0 59.9 60.6 64.2 61.6 ± 7 2.36 mm (No. 8) 48.0 43.8 43.8 46.2 44.0 ± 4 1.18 mm (No. 16) 39.0 34.3 34.1 34.9 33.9 ± 4 0.60 mm (No. 30) 29.0 25.4 25.2 25.2 25.3 ± 4 0.30 mm (No. 50) 16.0 14.4 14.4 14.7 14.3 ± 4 0.15 mm (No. 100) 10.0 7.6 7.9 8.2 7.5 ± 4 0.075 mm (No. 200) 6.5 4.9 5.2 5.4 4.7 ± 2 aNo WMA technology. Table 3-35a. Gradation for Ariton, Alabama.

60 asphalt content from both methods are shown in Table 3-36. For each of the mix samples, the results from the solvent extractions were higher than for the ignition method. It is believed that the ignition method correction factor deter- mined by NCAT for this mix design (0.44) may have been too high and resulted in corrected asphalt content that was low by about 0.2%. For the RAP and RAS samples, results from the solvent extraction method were lower than for the igni- tion method. For the RAP, the difference in the results can likely be attributed to aggregate mass loss; for the RAS sam- ples, the difference can be attributed to organic fibers being burned by the ignition method. Table 3-37 shows the Binder Grade Test results for the mixes, recycled materials, and virgin binder. The true grades of the binders recovered from each of Variable JMF a Low Va WMA (Gencor foam) Low Va HMAa Adjusted Va WMA (Gencor foam) Adjusted Va HMAa Tolerance Asphalt content (%) 5.10 5.49 5.07 5.17 4.80 ± 0.31 b Air voids (%) 4.1 1.3 1.8 2.5 3.0 ± 1.25b Gmb @ Ndesign 2.359 2.416 2.423 2.414 2.405 NA Gmm 2.459 2.447 2.467 2.476 2.480 NA VMA (%) 15.5 13.5 12.9 13.3 13.3 >15.0 VFA (%) 73.8 90.6 86.2 81.1 77.2 NA Gsb 2.650 2.640 2.640 2.640 2.640 NA Gse 2.654 2.656 2.663 2.678 2.666 NA Pba (%) 0.05 0.23 0.34 0.55 0.39 NA Pbe (%) 5.05 5.27 4.77 4.65 4.43 NA D/A ratio 1.29 0.92 1.10 1.16 1.05 NA Note: NA = not available. aNo WMA technology. bBased on results of four or more tests to achieve 100% pay. Table 3-35b. Asphalt content and volumetrics for Ariton, Alabama. Material Corrected Ignition Method (Average) Solvent Extraction (Average) Difference (Ignition–Extraction) Low air void WMA 5.30 5.49 -0.20 Low air void HMA 4.90 5.07 -0.18 Adjusted air void WMA 5.03 5.17 -0.14 Adjusted air void HMA 4.57 4.80 -0.23 RAP 4.50 4.27 0.23 RAS 22.77 18.92 3.85 Table 3-36. Asphalt content test results from Ariton, Alabama. Material Tc High Tc Int. Tc Low True Grade PG Tc (20-h PAV) Virgin PG 67-22 69.4 23.6 -24.3 69.4-24.3 67-22 -1.9 Low air void WMA 84.6 29.1 -15.5 84.6-15.5 82-10 -8.1 Low air void HMA 91.7 28.4 -16.0 91.7-16.0 88-16 -7.7 Adjusted air void WMA 90.9 29.5 -15.1 90.9-15.1 88-10 -8.6 Adjusted air void HMA 91.2 30.4 -12.3 91.2-12.3 88-10 -10.8 RAP 92.8 32.3 -15.3 92.8-15.3 88-10 -3.3 RAS 206.5 38.0 -8.8 206.5-8.8 202-4 -20.0 Table 3-37. Performance-grade test results from Ariton, Alabama.

61 the mixtures were similar. Also, the DTc (20-h PAV) results for each of the recovered binders were below –5°C. Construction. The test sections were located on US-84 north of Enterprise, Alabama, approximately 20 mi from the plant in Ariton. The haul time was approximately 30 to 40 min. Construction consisted of widening the roadway from two lanes to four lanes. The westbound lanes were new construction consisting of three asphalt layers: base, intermediate, and surface. The intermediate layer had been placed a few weeks before NCAT arrived on site. The target thickness for the surface lift was 1.5 in. The first three test sections were placed in the westbound portions of US-84 in both the outside and inside lanes, all of which were placed on the new intermediate and base courses. The adjusted air void HMA mix was placed in both the westbound inside lane and eastbound outside lane. The underlying condition of this mix section was noted as being milled in the west- bound lane, since this portion of roadway tied back into the existing pavement. The milled portions showed signs of scabbing, as shown in Figure 3-41. However, no other major distresses were noted on the milled surface. The portion of the mix placed in the eastbound lane was done as an overlay to the existing pavement with no milling. No major crack- ing was observed on this portion of the test section prior to the overlay. A CQS-1HP emulsion was used as the tack coat for all four mixes at a total diluted rate of 0.05 gal/yd2. Figure 3-42 shows the layout of the test sections. The tack coat application for this project was not well distributed on the sections. Rather, it was distributed in rows, as shown in Figure 3-43. This may have contributed to some of the observed pickup of the tack coat by the paver tires. The mixes were delivered to the paving site using 18 tri- axle dump-bed trucks. A RoadTec SB-2500-C Shuttle Buggy was used to transfer mixtures to the Caterpillar AP1000D paver. The paver used screed extensions with a total width of 16 ft. The shuttle buggy and paver used are shown in Fig- ure 3-44. The temperatures measured behind the screed for each mix are summarized in Table 3-38. Two rollers were used for compaction to target a mini- mum in-place density of 94%. The breakdown roller was a Volvo DD138HF operated in vibratory mode for two passes on each side followed by one pass back up the middle. The finishing roller was a Volvo DD118HF operated in vibra- tory mode for two passes on each side. A final static pass was then performed back up the middle of the mat to remove any roller marks. Figure 3-45 shows the breakdown roller compacting the mat. Three cores were taken for NCAT from each mix the day each test section was constructed. Figure 3-46 summarizes the densities of the cores from the four sections. ANOVA was performed to evaluate how the mix type (HMA or WMA) and air void content (low or adjusted) and the interaction between these two factors affected the initial in-place density results. The results of ANOVA, shown in Table 3-39, indi- cate that only the laboratory-compacted air void content factor had a significant effect on the in-place density for this location with a p-value of 0.044. Mix type had a p-value of 0.081, so it was not quite statistically significant at α = 0.05, but the HMA sections were more than 1.0% higher in den- sity than the companion WMA sections, which is significant from a practical point of view. Table 3-39 also shows the statistical grouping from Tukey’s Test of Multiple Com- parisons. This indicates that the adjustments made to the mixes to increase the QC air voids (i.e., reduced asphalt con- tent by about 0.3%) had a negative effect on the in-place density results. Field Performance at 17-Month and 29-Month Project Inspections. Field-performance evaluations were con- ducted on November 19, 2015 and November 16, 2016, after approximately 17 and 29 months of traffic had been applied to the four mix sections. Rutting. None of the four test sections exhibited any signs of rutting at the time of either inspection. Cracking. At the 17-month inspection, two low-severity longitudinal cracks were observed in the low air void HMA section. One crack was 1 ft in length; the other was 4 ft. Figure 3-41. Scabbing in milled section prior to placement of adjusted air void HMA surface.

62 Figure 3-42. Location of test sections in Enterprise, Alabama. Figure 3-43. Tack coat application in Enterprise, Alabama. Figure 3-44. MTV transferring mix to paver in Enterprise, Alabama.

63 Statistic Low Air Void WMA Temp. (°F) Low Air Void HMA Temp. (°F) Adjusted Air Void WMA Temp. (°F) Adjusted Air Void HMA Temp. (°F) Average 257.7 303.8 275.7 304.3 Standard deviation 15.3 7.3 6.1 8.4 Maximum 273.5 317.5 291.0 324.0 Minimum 231.0 286.5 259.5 290.0 Note: Temp. = temperature. Table 3-38. Temperatures behind the screed in Enterprise, Alabama. Figure 3-45. Breakdown roller compacting mat in Enterprise, Alabama. In -P la ce D en si ty (% ) Figure 3-46. In-place densities based on cores at construction in Enterprise, Alabama. Figure 3-47 shows the 1-ft crack in the low air void HMA section. At the 29-month inspection, the 1-ft and 4-ft cracks had connected with one another and had grown to a total crack length of 27 ft. Four new low-severity longitudinal cracks were also observed, measuring 3 ft, 3 ft, 7 ft, and 13 ft in length. Figure 3-48 shows an example of the new longitudinal cracks that developed between the 2015 and 2016 inspections. At the 29-month inspection, a total of 53 ft of low-severity longitudinal cracking was measured in the low air void HMA section. No cracking was observed in the other test sections. Raveling and Weathering. The texture depths for the Enterprise, Alabama, sections are summarized in Table 3-40. All four test sections had similar and acceptable mean texture depths. The surface texture of all four sections increased sub- stantially between the 2015 and 2016 inspections, but they

64 Source df Adjusted SS Adjusted MS F-Value P-Value Air voids 1 9.35 9.35 5.73 0.044 Mix type 1 6.45 6.45 3.95 0.082 Mix type × Air voids 1 0.03 0.03 0.02 0.897 Error 8 13.07 1.63 Total 11 28.89 Statistical Grouping Air voids N Mean Grouping Low 6 93.3 A Adjusted 6 91.5 B Table 3-39. Initial density ANOVA for Enterprise, Alabama. Figure 3-47. Low-severity longitudinal crack in low air void HMA section in Enterprise, Alabama (2015). Figure 3-48. Low-severity longitudinal crack in low air void HMA section in Enterprise, Alabama (2016). Mix Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation Construction (June 2014) 17 Months (November 2015) 29 Months (November 2016) Low air void WMA 0.4 0.0 0.4 0.0 0.5 0.1 Low air void HMA 0.5 0.0 0.4 0.0 0.5 0.0 Adjusted air void WMA 0.4 0.0 0.5 0.0 0.6 0.1 Adjusted air void HMA 0.4 0.0 0.5 0.0 0.6 0.1 Table 3-40. Mean texture depths in Enterprise, Alabama.

65 were still within the normal range for dense-graded mixes. Figure 3-49 shows examples of the surface texture for the low air void WMA and the low air void HMA and for the adjusted air void WMA and the adjusted air void HMA, respectively. Each of the test sections had some palm-sized or smaller spots of flushed binder, and the low air void HMA mix appeared to have the most instances. Figure 3-50 shows an example of one of the larger spots. These spots were believed to be created by binder-rich fines that collect in the paver and eventually drop off in the mix or on the mat. These spots are not evident at the time of construction since the entire mat is dark. Once traffic has been applied for a year or more and the color fades, these spots become evident. Also, ambient heat and traffic would tend to make these spots spread out with time. Field Cores. Five cores were taken from each section at both inspections. The densities were measured using AASHTO T 166, since all water absorptions were below 2.0%. A summary of the core densities at the time of the inspections compared to construction is shown in Figure 3-51. Table 3-41 shows the results of ANOVA performed to evalu- ate how mix type (HMA, WMA, low, and adjusted), age of the (a) Low Air Void WMA (b) Low Air Void HMA (c) Adjusted Air Void WMA (d) Adjusted Air Void HMA Figure 3-49. Examples of surface texture in Enterprise, Alabama.

66 Figure 3-50. Example of binder pooling on surface in Enterprise, Alabama. 0 Month 17 Months 29 Months In -P la ce D en si ty (% ) Low Va WMA Adj. Va WMA Adj. Va HMALow Va HMA Figure 3-51. In-place densities based on cores in Enterprise, Alabama. pavement, and the interaction between these factors affected the in-place density. As can be seen, both factors and their interaction had a significant effect (α = 0.05) on the in-place density for this project. Table 3-41 also shows the statistical grouping from Tukey’s Test of Multiple Comparisons. From this analysis, it can be seen that the density obtained during construction was statistically different (lower) than the two measurements. The section with the lowest density was the adjusted air void WMA; the section with the highest density was low air void WMA. Oak Ridge, Tennessee On October 8 and 9, 2014, a project was constructed by Roger’s Group on Raccoon Valley Drive near Oak Ridge, Tennessee. This project included test sections with two mixes containing RAP and RAS; one mix was produced as HMA, and the other was produced as WMA. The same mix design was used for both mixes with the difference being the addition of Evotherm 3G and lower production temperatures for the WMA mixture. The mixes were fine-graded 12.5-mm NMAS mixes designed by the Marshall mix design method with a compactive effort of 75 blows. Both mixes contained 10% RAP and 3% RAS with a limestone and slag virgin aggregate blend. The RAP used was a multiple-source−½ in. crushed RAP, and the RAS material was PC tear-off RAS acquired from a local landfill. Table 3-42 shows the material percentages used for mix design and production. Both mixes contained a PG 64-22 asphalt binder supplied by Marathon Petroleum in Knoxville, Tennessee. The HMA mix used 0.5% ArrMaz Ad-Here by mass of virgin binder as an antistripping agent. The WMA mix used 0.5% Evotherm by mass of virgin binder as the antistrip and WMA additive. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifications are shown in Table 3-43a and Table 3-43b. Production. These mixes were produced using a modi- fied drum plant that had once been a batch plant. The batch tower was replaced with a second drum for mixing, while the material drying was done in the original counter-flow drum

67 Source df Adjusted SS Adjusted MS F-Value P-Value Mix type 3 12.19 4.06 3.01 0.041 Age 2 27.43 13.72 10.16 0.001 Mix type × Age 6 21.5 3.58 2.66 0.029 Error 40 53.99 1.35 Total 51 113.14 Statistical Grouping Age (months) N Mean Grouping 29 20 94.3 A 17 20 93.8 A 0 12 92.4 B Mix Type N Mean Grouping Low air void WMA 13 94.1 A Low air void HMA 13 93.9 A Adjusted air void HMA 13 93.3 A B Adjusted air void WMA 13 92.8 B Mix type × Age (months) N Mean Grouping Low air void WMA × 29 5 95.1 A Low air void HMA × 29 5 94.8 A Low air void WMA × 17 5 94.5 A Adjusted air void HMA × 17 5 94.1 A Low air void HMA × 0 3 94.1 A B Adjusted air void WMA × 29 5 93.8 A Adjusted air void WMA × 17 5 93.7 A B Adjusted air void HMA × 29 5 93.6 A B Low air void HMA × 17 5 92.8 A B Low air void WMA × 0 3 92.5 A B Adjusted air void HMA × 0 3 92.2 A B Adjusted air void WMA × 0 3 90.9 B Table 3-41. Density ANOVA for Alabama. Aggregate Type Mix Design (%) Production HMA (%) Production WMA (%) D-Rock hard limestone 42 42 42 Coarse slag 10 11 11 No. 10 soft limestone 10 9 9 Natural sand 25 24 24 RAP 10 11 11 RAS 3 3 3 Table 3-42. Aggregate percentages used in mix design and production in Oak Ridge, Tennessee. that used a natural gas burner. The plant was located in Oak Ridge, Tennessee. Figure 3-52 shows a photograph of the plant. Evotherm 3G M1 was added at a rate of 0.5% by mass of virgin binder and was metered into the virgin binder line from a portable tote at the plant. The project start was delayed 2 days because of heavy rains in the area. On the third day, production began for the HMA mix early in the morning. But there were issues on the roadway that required the mix to be stored in the silo for about 4 h before paving could begin. The next day, construc- tion of the WMA went more smoothly, and the storage time was significantly less (less than 30 min on average). Volumetric Mix Properties. Since the original designs were completed using the Marshall mix design method and NCAT fabricated samples using a Superpave Gyratory

68 Compactor, a density correlation was conducted on site to determine the number of gyrations that provided the same density as 75 blows with the Marshall hammer. This yielded a seemingly low gyration level of 25. Later, it was found that the sample used for the Superpave Gyratory Compactor compac- tion tests may have had a significantly higher asphalt content than the actual test mix, which would have contributed to the low gyration correlation. Volumetric samples were hot-compacted at the average compaction temperature observed on the test section through the first couple of hours of construction for each mix. The Gmb tests were conducted in accordance with AASHTO T 166; water absorption values were less than 2.0%. Table 3-46a and Table 3-46b show the results from NCAT’s testing on the large sample taken once the mix production was considered stable. These results show that the grada- tions of the two mix types were similar, but the WMA mix had 0.41% more asphalt than the HMA. Calculated asphalt absorptions were similar despite the higher production temperature and longer storage time for the HMA. How- ever, the WMA had a significantly higher effective asphalt content than the HMA. The contractor’s QC data is shown in Appendix A. The asphalt content of both mixtures was determined by the ignition method and solvent extraction. The binder was also recovered and graded after extraction. The average asphalt content from both methods are shown in Table 3-47. The solvent extraction results for the HMA sample were sub- stantially lower than for the ignition method. In this case, the low extracted asphalt content result is suspected to be affected by sampling or testing error. For the RAP, the higher result for the ignition method is attributed to aggregate mass loss (i.e., unknown correction factor) rather than a real difference in asphalt content. For the RAS, the higher result for the igni- tion method is attributed to organic fibers in the RAS that are burned off in this test. Table 3-48 shows the Binder Grade Test results for the mixes, recycled materials, and virgin binder. The true grade of the HMA recovered binder was higher than that from the WMA, and the DTc (20-h PAV) of the HMA was much lower. These differences may have been affected by the higher mixing temperature and storage time of the HMA. Sieve Size Percent Passing Mix Design Specification 15.9 mm (5/8 in.) 100.0 100 maximum 12.5 mm (1/2 in.) 98.0 95–100 9.5 mm (3/8 in.) 88.0 80–93 4.75 mm (No. 4) 59.0 54–76 2.36 mm (No. 8) 41.0 35–57 0.6 mm (No. 30) 23.0 17–29 0.3 mm (No. 50) 14.0 10–18 0.15 mm (No. 100) 7.9 3–10 0.075 mm (No. 200) 5.1 0–6.5 Table 3-43a. Design aggregate gradation for Oak Ridge, Tennessee, mix design. Variable Mix Design Specification Asphalt content (%) 5.7 5.7–7.0 Air voids (%) 3.8 3.8–4.2 VMA (%) 17.4 >14.0 VFA (%) 78.3 NA D/A ratio 0.90 0.6–1.2 Note: NA = not available. Table 3-43b. Asphalt content, volumetrics, and specifications for Oak Ridge, Tennessee, mix design. Figure 3-52. Asphalt plant in Oak Ridge, Tennessee. Statistic HMA Temperatures (°F) WMA Temperatures (°F) Average 314.6 267.1 Standard deviation 11.8 15.3 Maximum 338 302 Minimum 298 248 Table 3-44. Production temperatures in Oak Ridge, Tennessee. Production HMA WMA Average production rate (tph) 268 279 Total tons shipped 708 1,724 Table 3-45. Production rates and totals in Oak Ridge, Tennessee.

69 Sieve Size Percent Passing JMFa HMAa WMA (Evotherm 3G) Tolerance 19.0 mm (3/4 in.) 100.0 100.0 100.0 ± 5.7 12.5 mm (1/2 in.) 98.0 96.9 98.6 ± 5.7 9.5 mm (3/8 in.) 88.0 83.7 86.7 ± 5.7 4.75 mm (No. 4) 59.0 52.4 55.8 ± 4.0 2.36 mm (No. 8) 41.0 38.2 39.5 ± 3.3 1.18 mm (No. 16) NA 29.7 30.1 ± 3.3 0.60 mm (No. 30) 23.0 20.8 20.6 ± 3.3 0.30 mm (No. 50) 14.0 12.7 12.3 ± 3.3 0.15 mm (No. 100) 7.9 7.7 7.3 ± 1.6 0.075 mm (No. 200) 5.1 5.2 5.0 ± 1.6 Note: NA = not available. aNo WMA technology. Table 3-46a. Gradation for Oak Ridge, Tennessee. Variable JMFa HMAa WMA (Evotherm 3G) Tolerance Asphalt content (%) 5.70 5.48 5.79 ± 0.25b Air voids (%) 3.8 6.2 5.3 3–5.5b Gmb @ Ndesign 2.447 2.434 2.433 NA Gmm 2.543 2.596 2.570 NA VMA (%) 17.4 16.1 16.7 >14.0 VFA (%) 78.3 61.3 68.1 NA Gsb 2.752 2.754 2.754 NA Gse 2.792 2.826 2.829 NA Pba (%) 0.53 0.95 0.99 NA Pbe (%) 5.20 4.15 4.80 NA D/B ratio 0.90 1.26 1.03 0.6–1.2 Note: NA = not available. aNo WMA technology. bBased on results of two or more tests to achieve 100% pay. Table 3-46b. Asphalt content and volumetrics for Oak Ridge, Tennessee. Material Corrected Ignition (Average) Solvent Extraction (Average) Difference (Ignition–Extraction) HMA 5.48 5.05 0.44 WMA 5.79 5.74 0.05 RAP 6.11 5.17 0.95 RAS 18.77 17.73 1.04 Table 3-47. Asphalt content test results from Oak Ridge, Tennessee.

70 Construction. The two test sections were placed in the westbound lane of Raccoon Valley Drive. This was 10 to 15 mi from the plant, depending on the test section. The haul time ranged between 15 and 25 min. The mixes were paved as a mill and overlaid with spot leveling. The existing pavement showed raveling and low- to medium-severity cracking before milling. After milling, the test sections appeared to be in fair condition before paving of the surface course, with only a few residual transverse cracks still evident. The target thickness was 1.5 in. An NTSS-1HM trackless tack was used at a total diluted rate of 0.08 gal/yd2. Figure 3-53 shows the layout of the test sections. The mixes were delivered to the paving site using 14 to 17 tri-axle dump-bed trucks. The mix was transferred to the Vögele paver using a RoadTec SB2500D Shuttle Buggy, as shown in Figure 3-54. The temperature of the mix behind the paver was mea- sured every 10 to 30 min using a hand-held temperature gun. These temperatures are summarized in Table 3-49. Material Tc High Tc Int. Tc Low True Grade PG Tc (20-h PAV) Virgin PG 64-22 67.5 22.2 -24.8 67.5-24.8 64-22 -1.9 HMA 82.0 30.9 -10.2 82.0-10.2 82-10 -11.7 WMA 76.9 25.0 -18.6 76.9-18.6 76-16 -5.5 RAP 97.7 36.4 -13.8 97.7-13.8 94-10 -3.5 RAS 169.4 49.5 18.1 169.4+18.1 166+20 -54.9 Note: Int. = intermediate. Table 3-48. Performance-grade test results from Oak Ridge, Tennessee. Figure 3-53. Location of test sections in Oak Ridge, Tennessee.

71 The WMA section averaged about 50°F lower than the HMA section. Three rollers were used for compaction of both mixes. The minimum average density target for each lot was 91%, with no single test less than 90%. The breakdown roller was a Hamm HD120 operated in vibratory mode for two passes on each side, followed by two more passes on each side in static mode. The intermediate roller used a Sakai SW850 operated in static mode for two passes on each side, followed by one pass on each side in vibratory mode. The finishing roller, another Hamm HD120, compacted in static mode for two passes on each side. Figure 3-55 shows the breakdown roller compacting the mat. Three cores were obtained from each mix the day of con- struction. Figure 3-56 shows the density results from these cores. Although the WMA cores had an average density nearly 2% lower than the HMA cores, a two-sample t-test indicated that there was no statistical difference (p-value = 0.134) in density between the WMA and HMA results. Field Performance at 13-Month and 25-Month Project Inspections. Field performance evaluations were con- ducted on November 11, 2015, and November 11, 2016, after approximately 13 and 25 months of traffic had been applied to the two sections, respectively. Rutting. Neither of the sections had any measurable rut- ting at the time of either inspection. Cracking. Neither of the sections had any cracking after 13 months. At the time of the 25-month inspection, one small transverse crack was observed in one of the WMA data sections. This crack was a low-severity transverse crack that measured 4-ft long. A photograph of that crack is shown in Figure 3-57. Raveling and Weathering. The calculated mean texture depths for both mixes are shown in Table 3-50. At the time of both inspections, the mean texture depths for the two sec- tions were similar and stable. Figure 3-58 and Figure 3-59 show examples of the surface texture at the time of the 2015 inspection for the HMA and WMA, respectively. The HMA section appears darker than the WMA section. This is more than likely because of significant tree canopy cover over the HMA section, which places it in the shade much more than the WMA section. Figure 3-54. MTV transferring mix to the paver in Oak Ridge, Tennessee. Statistic HMA Temperatures (°F) WMA Temperatures (°F) Average 298.9 245.8 Standard deviation 2.0 10.0 Maximum 303 259 Minimum 297 218 Table 3-49. Temperatures behind the screed in Oak Ridge, Tennessee. Figure 3-55. Breakdown roller compacting mat in Oak Ridge, Tennessee. In -P la ce D en si ty (% ) Figure 3-56. In-place densities based on cores at construction in Oak Ridge, Tennessee.

72 Field Cores. At the time of each inspection, five cores were obtained from both sections. The density tests on these cores revealed that most of the cores had water absorptions greater than 2%. Therefore, the Gmb of the cores was determined by the CoreLok method (AASHTO T 331). Figure 3-60 shows a sum- mary of the densities at the time of the inspections, along with the densities measured by NCAT at construction. The densi- ties of both mixes were low at the time of construction. The densities have increased with traffic, with the WMA densifying much more than the HMA. Table 3-51 shows the results of ANOVA to evaluate how the mix type (HMA or WMA), the age of the pavement, and the interaction between mix type and age affected the in-place density. Only the age of the pavement had a significant effect (p-value = 0.001) on the in-place density for this project. Table 3-51 also shows results of the Tukey’s Test of Multiple Comparisons. This analysis indicated that there was no statisti- cal difference between densities at 13 months and 25 months; however, the density obtained right after construction was sta- tistically lower than the results after 13 months and 25 months. Wilson, North Carolina Four test sections were constructed as part of a proj- ect on SR-58 in Wilson, North Carolina: HMA mix using Figure 3-57. Low-severity transverse crack in Oak Ridge, Tennessee (2016). Mix 2015 Inspection 2016 Inspection Mean Texture Depth (mm) Standard Deviation Mean Texture Depth (mm) Standard Deviation HMA 0.5 0.0 0.5 0.0 WMA 0.5 0.1 0.5 0.1 Table 3-50. Mean texture depths in Oak Ridge, Tennessee. MW–RAS, WMA using MW–RAS, HMA using PC–RAS, and WMA using PC–RAS. The chemical additive Evotherm 3G M1 was used in the WMA sections. The four mixes were produced and placed over a span of 4 days from June 15 to June 18, 2015. Each mix was placed in a four-lane por- tion of SR-58 in Wilson, North Carolina, by S.T. Wooten Corporation. The asphalt mixtures consisted of a fine-graded 9.5-mm NMAS Superpave mix design with a compactive effort of 65 gyrations. Mix designs for both the MW–RAS and PC–RAS mixes were conducted by S.T. Wooten with the intention of having similar volumetrics and gradations for all mixes. All four mixes contained 20% RAP and 5% RAS with a granite virgin aggregate. A multiple-source crushed RAP was used. The PC–RAS used was obtained from local landfills, while the MW–RAS was obtained from Saint-Gobain in Oxford, North Carolina. The asphalt content of the PC–RAS was determined to be 16.8% using AASHTO T 164, while the MW–RAS was determined to have an asphalt content of 18.0%. Table 3-52 and Table 3-53 show the material percentages used for mix designs and production for the MW–RAS and PC–RAS mixes, respectively. The asphalt mixtures used a PG 58-28 asphalt binder sup- plied by NuStar in Wilmington, North Carolina. All four mixes contained terminally blended Evotherm 3G M1 at a rate of 0.25% by mass of virgin binder. Evotherm served as an antistrip agent and a WMA additive. Therefore, the only difference in the HMA and WMA mixes was the production and compaction temperatures, since all mixes contained Evotherm. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifications are shown in Table 3-54a and Table 3-54b. Production. The mixes were produced using an Astec Double Barrel drum-mix plant located in Simms, North Carolina. The dryer used natural gas. Figure 3-61 shows a photograph of the asphalt plant. Table 3-55 provides a summary of the production tem- peratures for the four mixes, and Table 3-56 shows the production rates and totals. The WMA mixtures were produced 23°F to 27°F cooler than the companion HMA mixtures.

73 Figure 3-58. Example of surface texture of HMA in Oak Ridge, Tennessee (2015). Figure 3-59. Example of surface texture of WMA in Oak Ridge, Tennessee (2015). 0 Month 13 Months 25 Months In -P la ce D en si ty (% ) Figure 3-60. In-place densities based on cores in Oak Ridge, Tennessee. Source df Adjusted SS Adjusted MS F-Value P-Value Mix type 1 2.20 2.20 0.58 0.454 Age 2 80.76 40.38 10.66 0.001 Mix type × Age 2 4.44 2.22 0.59 0.565 Error 22 83.33 3.79 Total 27 169.37 Statistical Grouping Age (months) N Mean Grouping 25 10 92.5 A 13 12 90.7 A 0 6 87.9 B Table 3-51. Density ANOVA for Tennessee.

74 Aggregate Type Mix Design MW–RAS Mixes (%) Production MW–RAS HMA (%) Production MW–RAS WMA (%) No. 78s granite 29 25 26 Dry screenings 13 19 19 Coarse sand 33 31 30 RAP 20 20 20 RAS 5 5 5 Table 3-52. Aggregate percentages used in mix design and production for MW–RAS mixes in Wilson, North Carolina. Aggregate Type Mix Design PC–RAS Mixes (%) Production PC–RAS HMA (%) Production PC–RAS WMA (%) No. 78s granite 29 26 27 Dry screenings 19 19 19 Coarse sand 27 30 29 RAP 20 20 20 RAS 5 5 5 Table 3-53. Aggregate percentages used in mix design and production for PC–RAS mixes in Wilson, North Carolina. Sieve Size Percent Passing MW–RAS Mixes PC–RAS Mixes Specification 12.5 mm (1/2 in.) 100.0 100.0 100 maximum 9.5 mm (3/8 in.) 96.0 96.0 90–100 4.75 mm (No. 4) 72.0 72.0 <90 2.36 mm (No. 8) 57.0 57.0 32–67 1.18 mm (No. 16) 42.0 42.0 NA 0.6 mm (No. 30) 29.0 29.0 NA 0.3 mm (No. 50) 16.0 17.0 NA 0.15 mm (No. 100) 10.0 10.0 NA 0.075 mm (No. 200) 6.2 6.2 4–8 Note: NA = not available. Table 3-54a. Design aggregate gradation for Wilson, North Carolina, mix design. Variable MW–RAS Mixes PC–RAS Mixes Specification Asphalt content (%) 5.4 5.4 NA Air voids (%) 4.0 4.0 NA VMA (%) 16.0 16.1 >16 VFA (%) 75.0 74.9 73–76 D/A ratio 1.16 1.16 0.6–1.2 Note: NA = not available. Table 3-54b. Asphalt content, volumetrics, and specifications for Wilson, North Carolina, mix design.

75 Figure 3-61. Astec Double Barrel plant used in Simms, North Carolina. Statistic MW–RAS HMA Temp. (°F) MW–RAS WMA Temp. (°F) PC–RAS HMA Temp. (°F) PC–RAS WMA Temp. (°F) Average 297.1 276.2 304.8 277.0 Standard deviation 5.9 5.7 5.8 9.7 Maximum 307.0 287.0 318.0 302.0 Minimum 284.0 262.0 290.0 260.0 Table 3-55. Production temperatures in Simms, North Carolina. Production MW–RAS HMA MW–RAS WMA PC–RAS HMA PC–RAS WMA Average production rate (tph) 182 209 209 219 Total tons shipped 1,666 1,724 1,847 1,580 Table 3-56. Production rates and totals in Simms, North Carolina. Sieve Size Percent Passing MW–RAS JMFa MW–RAS HMAa MW–RAS WMA (Evotherm 3G) Tolerance 19.0 mm (¾ in.) 100.0 100.0 100.0 NA 12.5 mm (½ in.) 100.0 99.1 99.3 NA 9.5 mm ( in.) 96.0 93.1 93.6 NA 4.75 mm (No. 4) 72.0 69.7 70.5 NA 2.36 mm (No. 8) 57.0 54.3 54.4 ± 8.0 1.18 mm (No. 16) 42.0 41.8 41.5 NA 0.60 mm (No. 30) 29.0 28.9 28.6 NA 0.30 mm (No. 50) 16.0 15.5 16.2 NA 0.15 mm (No. 100) 10.0 7.7 7.9 NA 0.075 mm (No. 200) 6.2 4.6 4.9 ± 2.5 Note: NA = not available. aNo WMA technology. 83⁄ Table 3-57a. Gradation for MW–RAS mixes in Wilson, North Carolina. Volumetric Mix Properties. Volumetric specimens were compacted using 65 gyrations in a Superpave Gyratory Com- pactor. The Gmb of the compacted specimens was determined in accordance with AASHTO T 166. Water absorptions of the specimens were less than 2.0%. Table 3-57a, Table 3-57b, Table 3-58a, and Table 3-58b show the gradation, asphalt content, and volumetrics for the MW–RAS and PC–RAS mixes. For the MW–RAS mixtures, the gradations of the HMA and WMA were similar. The average effective asphalt content of the WMA samples was 0.28% higher than for the HMA. Calculated asphalt absorption results were similar for the two mix types. The contractor’s QC data is shown in Appendix A. As shown in Table 3-58a and Table 3-58b, the WMA con- taining PC–RAS was slightly finer than its HMA companion. The calculated asphalt absorption for the WMA was slightly higher than for the HMA, resulting in a slightly lower effec- tive asphalt content for the WMA samples.

76 Variable MW–RAS JMFa MW–RAS HMAa MW–RAS WMA (Evotherm 3G) Tolerance Asphalt content (%) 5.4 5.0 5.2 ± 0.7 Air voids (%) 4.0 6.4 4.9 ± 2.0 Gmb @ Ndesign 2.350 2.301 2.329 NA Gmm 2.448 2.459 2.448 NA VMA (%) 16.0 16.5 15.8 >15.0 VFA (%) 75.0 61.3 69.2 NA Gsb 2.648 2.620 2.620 NA Gse 2.653 2.649 2.646 NA Pba (%) 0.10 0.44 0.40 NA Pbe (%) 5.32 4.58 4.86 NA D/B ratio 1.16 1.00 1.00 NA Note: NA = not available. aNo WMA technology. Table 3-57b. Asphalt content and volumetrics for Wilson, North Carolina, MW–RAS mixes. Sieve Size Percent Passing PC–RAS JMFa PC–RAS HMAa PC–RAS WMA (Evotherm 3G) Tolerance 19.0 mm (3/4 in.) 100.0 100.0 100.0 NA 12.5 mm (1/2 in.) 100.0 98.8 99.5 NA 9.5 mm (3/8 in.) 96.0 92.5 93.8 NA 4.75 mm (No. 4) 72.0 69.3 71.5 NA 2.36 mm (No. 8) 57.0 53.7 55.8 ± 8.0 1.18 mm (No. 16) 42.0 41.7 43.3 NA 0.60 mm (No. 30) 29.0 29.0 30.5 NA 0.30 mm (No. 50) 16.0 16.6 18.1 NA 0.15 mm (No. 100) 10.0 8.5 9.2 NA 0.075 mm (No. 200) 6.2 5.3 5.7 ± 2.5 Note: NA = not available. aNo WMA technology. Table 3-58a. Gradation for Wilson, North Carolina, PC–RAS mixes. Variable PC–RAS JMFa PC–RAS HMAa PC–RAS WMA (Evotherm 3G) Tolerance Asphalt content (%) 5.4 5.4 5.4 ± 0.7 Air voids (%) 4.0 4.2 4.2 ± 2.0 Gmb @ Ndesign 2.349 2.333 2.340 NA Gmm 2.447 2.436 2.443 NA VMA (%) 16.1 15.8 15.6 >15.0 VFA (%) 74.9 73.2 73.0 NA Gsb 2.647 2.622 2.622 NA Gse 2.652 2.637 2.647 NA Pba (%) 0.10 0.22 0.38 NA Pbe (%) 5.33 5.15 5.04 NA D/B ratio 1.16 1.04 1.13 NA Note: NA = not available. aNo WMA technology. Table 3-58b. Asphalt content and volumetrics for Wilson, North Carolina, PC–RAS mixes.

77 The asphalt content of each mix was determined both by ignition method and by solvent extraction. The binders were recovered and graded after extractions. The average asphalt content for the mixture samples, shown in Table 3-59, were similar for both methods. For the RAP, the ignition method yielded 0.56% higher asphalt content, but this is likely because of mass loss for the RAP aggregate rather than a true differ- ence in asphalt content. For the PC–RAS samples, the larger difference in results from the two methods is likely caused by burning off cellulose fibers in the ignition oven. The fibers currently used in shingles are fiberglass, which would not be affected by the ignition method for the MW–RAS samples. Table 3-60 shows the Binder Grade Test results for both the mixes and the recycled materials. The true grades and DTc (20-h PAV) of the recovered binders from the HMA and WMA mixes containing MW–RAS were similar, as were the results for the two mixes containing PC–RAS. The binder properties of the RAP and RAS materials follow the expected trends. The DTc (unaged) results for the RAS binders were very low. Construction. The test sections placed on SR-58 were approximately 18 mi from the asphalt plant, with a haul time of about 20 to 30 min. The project consisted of paving a dif- ferent test mix on each of the four lanes from US-264 Alter- nate on the north end to US-264 on the south end. All mixes were placed as surface mixes at a target thickness of 1.5 in. after milling. A CRS-1H emulsion was used as the tack coat at a total diluted rate of 0.06 gal/yd2. Figure 3-62 shows the layout of the test sections. Prior to milling, the existing pavement exhibited block crack- ing throughout the test sections—as shown in Figure 3-63— but was most evident in the outside lanes. During the milling operation, some scabbing was observed. Even after milling, the existing pavement showed signs of fatigue cracking as shown in Figure 3-64. The mixes were delivered to the project using 15 to 22 tarped dump trucks. A RoadTec MTV1000D Shuttle Buggy was used to transfer the mixes to the Caterpillar AP1000E paver, which is shown in Figure 3-65. The temperatures measured behind the screed for each mix are summarized in Table 3-61. Two Caterpillar CB-634D model rollers were used for compaction. The target minimum density was 92%. For the two HMA mixes, the breakdown roller operated in vibra- tory mode for two passes and then in the static mode for two passes. This was then repeated on the other side of the mat followed by a final static pass back up the middle. The finish- ing roller used the same rolling pattern. The rolling pattern was changed slightly for the two WMA mixes. The break- down roller operated in vibratory mode for three passes on one side of the mat followed by one static pass back. Material Corrected Ignition Method Solvent Extraction Difference (Ignition–Extraction)Average Average MW–RAS HMA mix 5.16 4.99 0.17 MW–RAS WMA mix 5.34 5.24 0.10 PC–RAS HMA mix 5.45 5.36 0.09 PC–RAS WMA mix 5.40 5.40 0.00 RAP 5.81 5.25 0.56 MW–RAS 18.27 17.99 0.28 PC–RAS 18.64 16.84 1.80 Table 3-59. Asphalt content test results from Wilson, North Carolina. Material Tc High Tc Int. Tc Low True Grade PG Tc (20-h PAV) Virgin PH 58-28 60.7 16.1 -30.9 60.7-30.9 58-28 +1.9 MW–RAS HMA mix 85.2 26.2 -27.4 85.2-24.6 82-22 -2.7 MW–RAS WMA mix 80.2 24.1 -26.8 80.2-24.7 76-22 -2.0 PC–RAS HMA mix 90.4 26.4 -24.4 90.4-21.3 88-16 -3.2 PC–RAS WMA mix 90.4 28.5 -24.4 90.4-21.5 88-15 -2.9 RAP 110.4 43.0 -9.3 110.4-9.3 106-4 +0.4 MW–RAS 151.2 33.5 -39.5 151.2-3.5 148+2 -36.0 PC–RAS 207.0 55.5 -15.4 207.0+19.5 202+20 -34.9 Table 3-60. Performance-grade test results from Wilson, North Carolina.

78 Figure 3-62. Location of test sections in Wilson, North Carolina. Figure 3-63. Block cracking before milling in Wilson, North Carolina. Figure 3-64. Cracking after milling in Wilson, North Carolina.

79 This was repeated on the other side of the mat, and was then followed by one last static pass back up the middle of the mat. The same pattern was used for the breakdown roller. Fig- ure 3-66 shows both rollers compacting the mat. Three cores were taken for NCAT from each mix section the day after construction of each mix. Figure 3-67 shows the density results for these cores. The results show that the PC–RAS mixes had slightly higher densities compared to the MW–RAS. Table 3-62 shows the results of ANOVA performed to evaluate how the mix type (HMA and WMA), RAS type (MW–RAS and PC–RAS), and the interaction between the two affected the initial in-place density. Only the type of RAS had a significant effect (p-value = 0.013) on the in-place den- sity for this project. Table 3-62 also shows the results of the Tukey’s Test of Multiple Comparisons. Field Performance at 14-Month Project Inspection. A field performance evaluation was conducted on August 23, 2016, approximately 14 months after the test sections were paved. Rutting. After 14 months, none of the sections exhibited any measurable rutting. Cracking. Only the HMA PC–RAS section exhibited cracking. Four total feet of low-severity transverse cracking was observed in the section, as shown in Figure 3-68. How- ever, cracking was observed in the adjacent lane at this loca- tion as well, indicating that the cracking was caused by an underlying issue. Raveling and Weathering. The mean texture depths for the four North Carolina sections are summarized in Table 3-63. These results show that all four sections had very similar mean texture depths at the time of inspection. Fig- ure 3-69 shows an example of the surface texture of both the MW–RAS WMA on the left and the PC–RAS WMA on the right. Figure 3-70 shows an example of the surface texture of the MW–RAS HMA on the left and the PC–RAS HMA on the right. Cores. Cores were randomly taken throughout the sections. The densities of the cores were determined using AASHTO T 166. A summary of the core densities at the time of the inspections and the densities after construction is shown in Figure 3-71. Figure 3-65. MTV transferring mix to paver in Wilson, North Carolina. Statistic MW–RAS HMA Temp. (°F) MW–RAS WMA Temp. (°F) PC–RAS HMA Temp. (°F) PC–RAS WMA Temp. (°F) Average 281.8 254.4 279.9 249.0 Standard deviation 9.5 6.0 9.1 7.9 Maximum 300.0 264.5 293.5 268.0 Minimum 259.5 243.5 249.5 229.0 Table 3-61. Temperatures behind the screed in Wilson, North Carolina. Figure 3-66. Breakdown and finishing rollers compacting mat in Wilson, North Carolina.

80 Table 3-64 shows the results of ANOVA to evaluate how the mix type (HMA, WMA, MW–RAS, and PC–RAS), the age of the pavement, and the interaction between mix type and age affected the in-place density. Only mix type had a significant effect (p-value = 0.012) on the in-place density for this project. Table 3-64 also shows the results of the Tukey’s Test of Multiple Comparisons. The MW–RAS WMA had a statistically lower density than the two PC–RAS sections. La Porte, Indiana The contractor completed two test sections on SR-39 near La Porte, Indiana, on October 1, 2015. Both sections contained RAP and RAS; one was produced at normal hot- mix temperatures, and the other was a WMA mixture pro- duced by foaming the asphalt using the Maxam AQUABlack water injection system. The same mix design was used for both mixes. The mixes were coarse-graded 9.5-mm NMAS In -P la ce D en si ty (% ) MW–RAS HMA MW–RAS WMA PC–RAS WMAPC–RAS HMA Figure 3-67. In-place densities based on cores at construction in Wilson, North Carolina. Source df Adjusted SS Adjusted MS F-Value P-Value RAS type 1 3.61 3.61 10.16 0.013 Mix type 1 0.94 0.94 2.64 0.143 Mix type × RAS type 1 0.16 0.16 0.44 0.527 Error 8 2.84 0.36 Total 11 7.54 Statistical Grouping RAS type N Mean Grouping PC–RAS 6 93.4 A MW–RAS 6 92.3 B Table 3-62. Initial density ANOVA for Wilson, North Carolina. Figure 3-68. Transverse cracking in PC–RAS HMA section in Wilson, North Carolina.

81 Statistic MW–RAS HMA MW–RAS WMA PC–RAS HMA PC–RAS WMA Mean texture depth (mm) 0.4 0.3 0.4 0.4 Standard deviation (mm) 0.0 0.0 0.0 0.0 Table 3-63. Mean texture depths at 14 months in Wilson, North Carolina. Figure 3-69. MW–RAS WMA on the left and PC–RAS WMA on the right at 14-month inspection. Figure 3-70. MW–RAS HMA on the left and PC–RAS HMA on the right at 14-month inspection. In -P la ce D en si ty (% ) MW–RAS HMA MW–RAS WMA PC–RAS WMAPC–RAS HMA 0 Month 14 Months Figure 3-71. In-place densities based on cores in Wilson, North Carolina.

82 Superpave mix design mixes with a compactive effort of 75 gyrations. The WMA mix contained 15% RAP and 3% MW–RAS with a limestone and slag virgin aggregate blend. The HMA mix used the same materials, but the RAS content was decreased to 2% to bring the QC air voids into an accept- able range. The WMA was produced on the first day, and QC air voids were low. To open the mix up a bit, the contractor decided to reduce the RAS content to 2% for the HMA. The RAP was a multiple-source−½ in. crushed RAP, and the RAS material was MW shingle material acquired from the GAF plant in Michigan City, Indiana. Table 3-65 shows the material percentages used for the mix design and production. Both mixes contained a PG 70-22 asphalt binder sup- plied by Interstate Asphalt’s Ameropan terminal in southern Chicago. Neither mix contained an antistripping agent. The design aggregate gradation, optimum asphalt content, design volumetrics, and specifications are shown in Table 3-66a and Table 3-66b. Production. The mixes were produced using an Astec Double Barrel plant that used natural gas for the burner fuel. Figure 3-72 shows a photograph of the contractor’s plant in La Porte, Indiana. The WMA mix was produced using the Maxam AQUABlack foaming system at a rate of 2% water by mass of virgin binder. Source df Adjusted SS Adjusted MS F-Value P-Value Mix type 3 11.52 3.84 4.52 0.012 Age 1 0.94 0.94 1.11 0.303 Mix type × Age 3 0.81 0.27 0.32 0.813 Error 24 20.38 0.85 Total 31 34.62 Statistical Grouping Mix type N Mean Grouping PC–RAS HMA 8 93.4 A PC–RAS WMA 8 93.1 A MW–RAS HMA 8 92.4 A B MW–RAS WMA 8 91.8 B Table 3-64. Density ANOVA for Wilson, North Carolina. Aggregate Type Mix Design (%) Production HMA (%) Production WMA (%) No. 11 limestone 25 25 24 No. 11 slag 10 13 14 Coarse screenings 27 35 32 Slag sand 13 4 4 Natural sand 7 6 7 MW–RAS 3 2 3 RAP 15 15 16 Table 3-65. Aggregate percentages used in mix design and production in La Porte, Indiana. Sieve Size Percent Passing Mix Design Specification 19.0 mm (3/4 in.) 100.0 NA 12.5 mm (1/2 in.) 100.0 NA 9.5 mm (3/8 in.) 93.0 90–100 4.75 mm (No. 4) 67.1 <90 2.36 mm (No. 8) 43.6 32–67 1.18 mm (No. 16) 27.5 NA 0.6 mm (No. 30) 18.4 NA 0.3 mm (No. 50) 12.1 NA 0.15 mm (No. 100) 7.9 NA 0.075 mm (No. 200) 5.9 2–10 Note: NA = not available. Table 3-66a. Design aggregate gradation and specifications for La Porte, Indiana, mix design. Variable Mix Design Specification Asphalt content (%) 6.0 NA Air voids (%) 4.0 NA VMA (%) 15.6 >15.0 VFA (%) 74.4 73–76 D/A ratio 1.30 0.8–1.6 Note: NA = not available. Table 3-66b. Asphalt content, volumetrics, and specifications for La Porte, Indiana, mix design.

83 Figure 3-72. Asphalt plant in La Porte, Indiana. Statistic HMA Temperatures (°F) WMA Temperatures (°F) Average 317.7 303.3 Standard deviation 7.6 10.6 Maximum 336 317 Minimum 305 280 Table 3-67. Production temperatures in La Porte, Indiana. Production HMA WMA Average production rate (tph) 291 284 Total tons shipped 1,684 2,515 Table 3-68. Production rates and totals in La Porte, Indiana. Sieve Size Percent Passing JMFa HMAa WMA (Maxam Aquablack Foam) Tolerance 19.0 mm (3/4 in.) 100.0 100.0 100.0 NA 12.5 mm (1/2 in.) 100.0 100.0 100.0 100 maximum 9.5 mm (3/8 in.) 93.0 93.8 94.0 90–100 4.75 mm (No. 4) 67.1 68.7 68.7 62.1–72.1 2.36 mm (No. 8) 43.6 46.3 46.9 35.6–51.6 1.18 mm (No. 16) 27.5 30.6 31.3 21.5–33.5 0.60 mm (No. 30) 18.4 20.6 21.0 14.4–22.4 0.30 mm (No. 50) 12.1 12.6 12.4 8.1–16.1 0.15 mm (No. 100) 7.9 8.6 8.1 4.9–10.9 0.075 mm (No. 200) 5.9 6.7 6.1 4.4–7.4 Note: NA = not available. aNo WMA technology. Table 3-69a. Gradations for La Porte, Indiana. Table 3-67 summarizes the production temperature information. The average WMA temperature was only about 14°F lower than the HMA temperature. As shown in Table 3-68, the production rates for the two mixtures were similar. Volumetric Mix Properties. Table 3-69a and Table 3-69b show the results of NCAT’s routine mix tests on these mixes. The asphalt content for the WMA was 0.2% higher than the HMA. The calculated asphalt absorption value was slightly higher for the WMA compared to the HMA. The contractor’s QC data is shown in Appendix A. The average asphalt content from the ignition method and solvent extractions are shown in Table 3-70. Minor differences between the two methods were obtained for the mixture samples. The asphalt content differences for the RAP can likely be attributed to aggregate mass loss in the ignition method. For the RAS, the higher asphalt content from the ignition method is likely caused by organic fibers that would have burned during the test. Table 3-71 shows the Binder Grade Test results for the mixes, recycled materials, and virgin binder. The true grade and the performance grade of the recovered binders from the HMA and WMA were similar and both had DTc (20-h PAV) values below –5°C. The MW–RAS used on this project had the lowest PG and highest DTc (no aging) of all RAS materials used in this study. Construction. The two test sections were placed on SR-39 south of La Porte near Hanna, Indiana. This is 17 to

84 20 mi from the plant, which yielded an average haul time of 25 to 35 min. The test sections were paved to a target thickness of 1.5 in. over a new binder course. Since the binder course was placed prior to NCAT arriving on site, the underlying pavement could not be assessed. NTSS-1HM trackless tack was used at a total diluted rate of 0.08 gal/yd2. The HMA was placed in the northbound lane, while the WMA was placed in the south- bound lane. Figure 3-73 shows the layout of the test sections. The mixes were delivered to the paving site using a cycle of 17 tri-axle dump-bed trucks. The mix was transferred to the Caterpillar AP1055E paver using a Weiler E2850 mix transfer vehicle, as shown in Figure 3-74. The temperatures measured behind the screed for each mix are summarized in Table 3-72. Three rollers were used for compaction of both mixes. The lower specification limit for in-place density was 91.5%, while there was no upper specification limit. The breakdown roller was a Dynapac CC722 operated in vibratory mode for five passes. The intermediate roller used was a Bomag BW190 operated in vibratory mode for four passes, followed by one pass in static mode. The finishing roller, a Caterpillar CB534D, compacted in static mode for five passes. Figure 3-75 shows the breakdown and intermediate rollers compacting the mat. Three cores were taken for NCAT from each mix the day each was constructed. Figure 3-76 shows the density results for these cores. In this case, significant differences in density were obtained at α = 0.05 for the as-constructed density results. Field Performance at 16-Month Project Inspection. A field performance evaluation was conducted on March 15, 2017, approximately 16 months after construction. Heavy snow fell the night before the evaluation, so it was not possible for traffic control to divert traffic and take cores. Variable JMFa HMAa WMA (Maxam Aquablack Foam) Tolerance Asphalt content (%) 6.00 5.53 5.73 ± 0.5 Air voids (%) 4.0 5.3 4.2 3.0–5.0 Gmb @ Ndesign 2.401 2.363 2.386 NA Gmm 2.500 2.494 2.490 NA VMA (%) 15.6 16.0 15.2 14.5–17.0 VFA (%) 74.4 67.1 72.5 NA Gsb 2.675 2.657 2.652 NA Gse 2.752 2.720 2.725 NA Pba (%) 1.06 0.90 1.04 NA Pbe (%) 5.01 4.68 4.75 NA D/B ratio 1.30 1.44 1.29 NA Note: NA = not available. aNo WMA technology. Table 3-69b. Asphalt content and volumetrics for La Porte, Indiana. Material Corrected Ignition Method (Average) Solvent Extraction (Average) Difference (Ignition–Extraction) HMA 5.45 5.53 -0.08 WMA 5.64 5.73 -0.09 RAP 7.08 5.53 1.55 MW–RAS 20.67 19.37 1.30 Table 3-70. Asphalt content test results from La Porte, Indiana. Material Tc High Tc Int. Tc Low True Grade PG Tc (20-h PAV) Virgin PG 70-22 71.7 23.6 -26.2 71.7-24.0 70-22 -2.2 HMA 85.4 26.9 -24.2 85.4-18.6 82-16 -5.6 WMA 85.9 31.1 -21.5 85.9-15.4 82-10 -6.1 RAP 101.4 29.7 -17.8 101.4-12.3 100-10 -5.4 MW–RAS 135.3 26.3 -28.9 135.3-12.0 130-10 -16.9 Table 3-71. Performance-grade test results from La Porte, Indiana.

Figure 3-73. Location of test sections in La Porte, Indiana. Figure 3-74. MTV transferring mix to paver in La Porte, Indiana. Statistic HMA Temperatures (°F) WMA Temperatures (°F) Average 277.7 266.2 Standard deviation 5.3 7.6 Maximum 285.0 277.5 Minimum 263.5 249.0 Table 3-72. Temperatures behind the screed in La Porte, Indiana.

86 problems were reported during production and no issues during construction were found. For the Larsen, Wiscon- sin, project, the production temperatures for the HMA and WMA mixtures were similar. Table 3-74 contains a summary of as-constructed density results of all five new projects. Only the La Porte, Indiana, project had significant differences in the as-constructed density results with 1.0% higher density for the WMA mixture. Field Performance Summary of Existing and New Projects Table 3-75 and Table 3-76 summarize the field performance for the existing and new projects, respectively. Thus far, all projects are characterized by low severity to no cracking. No other types of distress are evident at this point. No significant field performance differences were evident between HMA and WMA sections containing RAS. No meaningful differ- ences were observed in the field performance of the North Carolina test sections containing PC–RAS and MW–RAS. Properties of HMA and WMA Mixtures of New Projects Asphalt Content Ignition Oven tests (AASHTO T 308) and chemical extrac- tions using TCE (AASHTO T 164 Method A) were used to determine the asphalt content of the RAP, RAS, and the sam- pled mixtures. The test results for the RAP and RAS samples from the new projects are summarized in Table 3-77. The dif- ferences in asphalt content of RAP samples between the two Figure 3-75. Breakdown and intermediate rollers compacting mat in La Porte, Indiana. In -P la ce D en si ty (% ) Figure 3-76. In-place densities based on cores at construction in La Porte, Indiana. However, the sections were visually inspected for signs of dis- tress from the shoulder. Both sections appeared to be perform- ing well. No signs of cracking or raveling were observed, and there was no measurable rutting. Figure 3-77 shows an example of both sections at the time of the 16-month inspection, with the HMA on the left and WMA on the right. The dark area to the left of the centerline is caused by moisture that had not evaporated. Production and Construction Summary of New Projects Table 3-73 shows a summary of the most relevant con- struction and production observations. In most cases, no Figure 3-77. HMA on the left and WMA on the right in La Porte, Indiana.

87 Location Construction Date Mix Variables Production/Construction Notes Larsen, Wisconsin September 2013 HMA and two WMA additives No troubles. WMA and HMA production temperatures were within 10°F. Enterprise, Alabama June 2014 HMA and WMA, two air void contents During production, an issue was found with contractor’s QC air voids. On average, WMA temperatures were 40°F lower than HMA. Oak Ridge, Tennessee October 2014 HMA and WMA additive Because of rain and equipment issues on the roadway, HMA had to be stored in silo approximately 4 h. On average, WMA temperatures were 50°F lower than HMA. Wilson, North Carolina June 2015 HMA and WMA, two RAS types No troubles. On average, WMA temperatures were 21°F lower than HMA. La Porte, Indiana October 2015 HMA and WMA, water injection No troubles. On average, WMA temperatures were 14°F lower than HMA. Table 3-73. Summary of construction and production observations. Site Mix ID Average Density (% Gmm) Standard Deviation Sign. Diff.? Larsen, Wisconsin Control 91.6 0.6 NRediset 90.7 1.0 Zycotherm 90.8 1.5 Enterprise, Alabama Low Va HMA 94.1 0.6 N Low Va Gencor Foam 92.5 1.6 Adj. Va HMA 92.2 0.9 N Adj. Va Gencor Foam 90.9 1.7 Oak Ridge, Tennessee HMA 88.8 2.0 N Evotherm 3G 87.0 1.2 Wilson, North Carolina MW–RAS HMA 92.4 0.1 N MW–RAS Evotherm 3G 92.1 0.9 PC–RAS HMA 93.8 0.5 N PC–RAS Evotherm 3G 93.0 0.6 La Porte, Indiana HMA 91.4 0.6 Y AQUABlack WMA 92.4 0.1 Note: Va= air void; Sign. Diff. = significant difference. Table 3-74. Summary of as-constructed density results. Location Mix Variable Age Field Performance US-287 Fort Worth, Texas HMA 37 months Low-severity transverse cracking (reflective) Flushing (small spots) WMA (chemical additive) 37 months Low-severity transverse cracking (reflective) Flushing (small spots) Low-severity longitudinal (edge) cracking FM 973 Austin, Texas HMA PG 64-22, 15% RAP, 3% RAS 47 months Low-severity transverse cracking Low-severity block cracking WMA (chemical additive) PG 64-22 15% RAP, 3% RAS 47 months Low-severity longitudinal cracking HMA PG 64-22 0% RAP, 5% RAS 47 months Low-severity longitudinal cracking HMA PG 58-28 15% RAP, 3% RAS 47 months Low- and medium-severity longitudinal cracking Low-severity transverse cracking I-88, Illinois Tollway Aurora, Illinois WMA (chemical additive), two aggregate types 46 months Low- , medium- , and high-severity transverse cracking (mostly reflective) Table 3-75. Summary of field performance of existing projects.

88 Location Mix Variable Age Field Performance SR-96 Larsen, Wisconsin Control, Rediset, Zycotherm 24 months Minor reflective cracking over unrubblized portland cement concrete pavement US-84 Enterprise, Alabama HMA and WMA (low air void) HMA and WMA (adjusted air void) 29 months Low-severity transverse cracking Significant increase in texture depth Union Valley Road Oak Ridge, Tennessee WMA and HMA 25 months Low-severity transverse cracking No other distresses SR-58 Wilson, North Carolina HMA and WMA with PC–RAS HMA and WMA with MW–RAS 14 months Low-severity transverse cracking No other distresses SR-39 La Porte, Indiana WMA and HMA 16 months No cracking or other distresses Table 3-76. Summary of field performance of new projects. Location Material Ignition (%) Extraction (%) Diff. (%) Larsen, Wisconsin RAP 6.21 4.32 1.89 RAS 30.75 25.77 4.98 Enterprise, Alabama RAP 4.50 4.27 0.23 RAS 22.77 18.92 3.85 Oak Ridge, Tennessee RAP 6.11 5.17 0.94 RAS 18.77 17.73 1.04 Wilson, North Carolina RAP 5.81 5.25 0.56 MW–RAS 18.27 17.99 0.28 PC–RAS 18.64 16.84 1.80 La Porte, Indiana RAP 7.08 5.53 1.55 RAS 20.67 19.37 1.30 Table 3-77. Comparison of asphalt content for RAP and RAS materials. methods may partly be attributed to aggregate mass loss in the ignition method. The ignition method correction factors for the mixtures were Wisconsin, 1.75%; Alabama, 0.44%; Ten- nessee, 0.48%; North Carolina mix with MW–RAS, 0.15%; North Carolina mix with PC–RAS, 0.20%; and Indiana, 0.89%. The differences in asphalt content for RAS samples may be explained by a number of factors, including organic fibers that are incinerated in the ignition method; aggregate mass loss of shingle granules and loss of fines during the ignition method; and difficulty of dissolving all of the hard asphalt in some RAS samples using the extraction method. The true asphalt content of the RAS materials is likely to be between the ignition method and extraction method results. Average asphalt content of the sampled mixtures using both methods is summarized in Table 3-78. The ignition method results include the correction factors determined by NCAT using the raw materials for each mix design. On average, the ignition method results were 0.1% higher than the solvent extraction method. Differences in results between the two Location Project Variable Ignition (%) Extraction (%) Average Diff. (%)HMA WMA Diff. HMA WMA Diff. Wisconsin Rediset 5.98 5.76 -0.22 5.46 5.31 -0.15 -0.19 Zycotherm 5.98 5.55 -0.43 5.46 5.42 -0.04 -0.24 Alabama Low air void 4.90 5.30 +0.40 5.07 5.49 +0.42 +0.41 Adjusted air void 4.57 5.03 +0.46 4.80 5.17 +0.37 +0.42 Tennessee HMA versus WMA 5.48 5.79 +0.31 5.05 5.74 +0.69 +0.50 North Carolina MW–RAS 5.16 5.34 +0.18 4.99 5.24 +0.25 +0.22 PC–RAS 5.45 5.40 -0.05 5.36 5.40 +0.04 0.0 Indiana HMA versus WMA 5.45 5.64 +0.19 5.53 5.73 +0.20 +0.20 Table 3-78. Comparison of asphalt content for project mixtures.

89 methods on any project can be influenced by inaccurate cor- rection factors caused by recycled materials and natural varia- tions in virgin aggregates. This table also compares the asphalt content of the HMA and WMA samples for each project loca- tion and additional project variables for both test methods. Although each set of mixtures was tested by a single operator in a single laboratory, the values shown are the average of two tests; so the differences should be less than the single-operator d2S limits of the respective tests. For AASHTO T 308, the accept- able range of two results by a single-operator is 0.196. Thus, it can be inferred that the differences in asphalt content shown for the HMA and WMA mixtures are not simply caused by testing variability. WMA mixtures of the two sets from Alabama and the set from Tennessee appear to have signifi- cantly higher asphalt content than their HMA companions. For AASHTO T 164, the acceptable range of two results by a single operator is 0.52%. Only the Tennessee mixture set exceeds this range. However, in most cases, the differences are reasonably consistent for the two methods, which fur- ther indicates that the Alabama and Tennessee WMA mix- tures had significantly higher asphalt content than their respective HMA mixtures. Binder Grade Binders were extracted using ASTM D2172, Method B (centrifuge) and recovered using ASTM D5404 (rotovap). Recovered binders were tested to determine the PG of the material using AASHTO M 320. A summary of the test results is given in Table 3-79. RAS binders were tested with- out RTFO or PAV because RAS binders are too stiff to flow in RTFO bottles. Recovered RAP binders were not aged in the PAV, since RAP is presumed to be thoroughly aged in the field. A dynamic shear rheometer (DSR) equipped with an environmental chamber to control the high temperatures beyond the capacity of standard DSRs that use a water bath Location Material Sample Type Tc High Tc Int. Tc Low PG Tc Wisconsin Control Mix 76.6 20.9 -25.2 76-22 -3.5 Rediset Mix 79.0 23.5 -24.3 76-22 -3.8 Zycotherm Mix 83.5 23.7 -21.1 82-16 -5.0 RAP RAP 86.3 30.2 -18.8 82-16 -1.7 RAS RAS 145.3 36.0 +12.0 142+14 -45.6 Virgin binder Tank 59.1 16.6 -30.9 58-28 +1.4 Alabama Low air void HMA Mix 91.7 28.4 -16.0 88-16 -7.7 Low air void WMA Mix 84.6 29.1 -15.5 82-10 -8.1 Adjusted air void HMA Mix 91.2 30.4 -12.3 88-10 -10.8 Adjusted air void WMA Mix 90.9 29.5 -15.1 88-10 -8.6 RAP RAP 92.8 32.3 -15.3 88-10 -3.3 RAS RAS 206.5 38.0 -8.8 202-4 -20.0 RAP–RAS blend RAP–RAS 97.8 34.8 -13.6 94-10 -5.7 Virgin binder Tank 69.4 23.6 -24.3 67-22 -1.9 Tennessee HMA Mix 82.0 30.9 -10.2 82-10 -11.7 WMA Mix 76.9 25.0 -18.6 76-16 -5.5 RAP RAP 97.7 36.4 -16.8 94-10 -3.5 RAS RAS 169.4 49.5 +18.1 166+20 -54.9 Virgin binder Tank 67.5 22.2 -24.8 64-22 -1.9 North Carolina MW–HMA Mix 85.2 26.2 -24.6 82-22 -2.7 MW–WMA Mix 80.2 24.1 -24.7 76-22 -2.0 PC–HMA Mix 90.4 26.4 -21.3 88-16 -3.2 PC–WMA Mix 90.4 28.5 -21.5 88-16 -2.9 RAP RAP 110.4 43.0 -9.3 106-4 0.4 RAS MW–RAS 151.2 33.5 -3.5 148+2 -36.0 RAS PC–RAS 207.0 55.5 +19.5 202+20 -34.9 Virgin binder Tank 60.7 16.1 -30.9 58-28 +1.9 Indiana HMA Mix 85.4 26.9 -18.6 82-16 -6.1 WMA Mix 85.9 31.1 -15.4 82-10 -5.6 RAP RAP 101.4 29.7 -12.3 100-10 -5.4 RAS RAS 135.3 26.3 -12.0 130-10 -16.9 Virgin binder Tank 71.7 23.6 -26.2 70-22 -2.2 Table 3-79. Performance-grade test results.

90 for temperature control was used for grading all recovered binders. The bending beam rheometer samples of unaged RAS were tested at 0°C and +6°C, the two highest tempera- tures at which the bending beam rheometer bath can main- tain temperature control. As with most RAS materials, the stiffness (S) easily passed the criteria at both temperatures. However, the relaxation value (m-value) often failed to meet the requirement. Positive low temperature grades were extrapolated to the temperature at which the m-value equals 0.300. The DTc results shown in this table are based on 20-h PAV of base binders and binders recovered from the mixtures. RAS binders from Wisconsin, Tennessee, and North Caro- lina (PC–RAS) had Tc-low results well above zero, indicat- ing that those materials are very stiff. Two RAS binders also had Tc-high results above 200°C (RAS from Alabama and PC–RAS from North Carolina), which could have negatively affected compactability of the mixtures. The Tc-low results of the binders recovered from the mixtures were compared to the predicted low critical temperature from LTPPBind Online using the MERRA data for each location. For Larsen, Wisconsin, the 98% reliability low critical temperature is –30.6°C; so all of the mixtures on this project would eventu- ally be expected to have thermal cracking. For Enterprise, Ala- bama, the 98% reliability low critical temperature is –7.7°C; so, despite the relatively high low critical temperatures for the recovered binders from the Alabama mixtures, the test sec- tions are not likely to have thermal cracking. For Oak Ridge, Tennessee, the 98% reliability low critical temperature is –15.5°C; so the HMA mixture on this project would eventu- ally be expected to have thermal cracking, but the WMA mix- ture would not. For the Wilson, North Carolina, project, the 98% reliability low critical temperature is –13.4°C; so none of the North Carolina mixtures are expected to have thermal cracking. For La Porte, Indiana, the 98% reliability low criti- cal temperature is –23.2°C; so both of the Indiana mixtures are likely to eventually have thermal cracking. As previously noted, the DTc results are based on 20-h PAV. Of the 14 mixtures sampled in this study, only six recov- ered binders had a DTc above −5.0, even considering that the 14 binders were only aged for 20 h in the PAV. Those six results were from Wisconsin and North Carolina mixtures, which both used PG 58-28 virgin binders that had positive DTc values. In addition to determining the PG of the binder, multiple stress creep recovery was conducted on the binders recovered from all mixtures, in accordance with AASHTO MP 19. These results are provided in Table 3-80. All of the recovered binders graded as “E”, the highest level of trafficking, which indicates a high resistance to rutting under extremely heavy traffic. In general, mixtures containing RAS should be highly resistant to permanent deformation. Location Mixture Average Recovery (%) Average Jnr (k/Pa) Diff. Recovery (%) Diff. Jnr (%) MP 19 Grade 100 Pa 3,200 Pa 100 Pa 3,200 Pa Wisconsin Control 35.8 32.9 0.141 0.148 8.10 5.24 E Rediset 36.7 36.5 0.120 0.121 0.64 0.59 E Zycotherm 46.1 45.9 0.064 0.064 0.49 0.22 E Alabama Low air void HMA 65.0 60.8 0.020 0.022 6.41 11.01 E Low air void WMA 60.5 56.4 0.025 0.028 6.76 10.22 E Adjusted air void HMA 80.7 75.5 0.004 0.005 6.40 30.92 E Adjusted air void WMA 63.1 59.4 0.023 0.025 5.86 9.37 E Tennessee HMA 35.9 29.3 0.194 0.216 18.27 11.14 E WMA 45.2 39.0 0.079 0.088 13.76 11.68 E North Carolina HMA MW–RAS 35.9 29.3 0.194 0.216 18.27 11.14 E WMA MW–RAS 45.2 39.0 0.079 0.088 13.76 11.68 E HMA PC–RAS 69.0 65.5 0.020 0.022 5.11 11.55 E WMA PC–RAS 65.6 61.8 0.024 0.027 5.82 12.74 E Indiana HMA 37.1 29.9 0.149 0.167 19.22 12.52 E WMA 38.5 31.9 0.153 0.171 17.04 11.80 E Table 3-80. Multiple Stress Creep Recovery Test results.

91 The LAS Test, which is based on viscoelastic continuum damage mechanics, was conducted at 2.5% and 5.0% strain. Table 3-81 shows the results of LAS binder testing. Overall, there was a wide range in LAS results from project to proj- ect. In about half of the HMA–WMA comparisons, the HMA binders had higher cycles to failure than the WMA binders. The reverse was true for the other comparison pairs. Surpris- ingly, the PC–RAS from North Carolina had much higher LAS fatigue results than the MW–RAS from North Carolina at both strain levels. Deleterious Materials and Fiber Table 3-82 shows amounts of unwanted materials in RAS per project and quantities of fiber from chemical extraction and the ignition oven. Texas requires less than 1.5% deleterious material in processed RAS. All RAS samples in this study met that criterion. The fiber content of the RAS was determined during the sieve analysis of the post-extraction aggregates by removing clumps of fiber from the sieves and weighing them separately. Fiber content from ignition tests is presumed to be only fiberglass, whereas fiber content from solvent extrac- tion tests is presumed to include both fiberglass and cel- lulose fibers. Except for the results for the PC–RAS from Wilson, North Carolina, the fiber content obtained from sol- vent extraction is more consistent with data from the litera- ture (3M Corporation 2007; Lee 2009; National Association of Home Builders 1998). Dynamic Modulus (E*) Dynamic modulus testing was performed to quantify the stiffness of the asphalt mixtures over a range of tem- peratures and frequencies. The dynamic modulus tests were conducted on the field-produced mixes (hot-compacted samples) using an AMPT. The dynamic modulus samples were prepared in accordance with AASHTO PP 60. Tripli- cate samples were tested from each mix. The temperatures and frequencies used for testing these mixes were those rec- ommended in AASHTO PP 61-10. For this methodology, the high test temperature is selected based on the high PG of the base binder used in the mix. Data analyses for the dynamic modulus tests were conducted per the methodol- ogy in AASHTO PP 61-10. Dynamic modulus master curves were generated for each of the mixes by project (WMA tech- nologies and HMA control). The reference temperature for the master curves was 68°F (20°C). The following figures present the master curves for each project on a logarithmic scale. Figure 3-78 presents the master curves of the three Wisconsin mixtures. As can be seen, the control mixture was slightly stiffer than the Rediset and Zycotherm mixtures at all temperature–frequency combinations. Figure 3-79 presents the master curves of the Alabama mixtures. As can be seen, the mix with the lowest stiffness was the low air void WMA mix. The lower stiffness of this mix is attributed to the higher asphalt content in this mix compared to the others from this project. Location Mixture Nf at Applied Strain 2.5% 5.0% Wisconsin Control (HMA) 1,167,551 13,616 Rediset 611,916 6,847 Zycotherm 786,562 7,113 Alabama Low air void HMA 392,046 5,926 Low air void WMA 204,333 3,241 Adjusted air void HMA 491,303 5,435 Adjusted air void WMA 376,769 4,978 Tennessee HMA 79,088 1,835 WMA 118,705 2,512 North Carolina HMA MW–RAS 335,630 6,902 WMA MW–RAS 573,039 6,705 HMA PC–RAS 1,873,906 20,188 WMA PC–RAS 1,243,507 10,963 Indiana HMA 66,063 1,263 WMA 192,642 3,806 Table 3-81. Linear Amplitude Sweep (LAS) Test results. Project RAS Type Deleterious Materials (%) Fiber Content from Extraction (%) Fiber Content from Ignition (%) Larsen, Wisconsin PC 0.08 3.80 0.38 Enterprise, Alabama PC 0.22 3.72 0.19 Oak Ridge, Tennessee PC 0.12 0.73 0.47 Wilson, North Carolina MW 0.02 1.85 0.99 PC 0.06 0.43 0.54 La Porte, Indiana MW 0.02 1.72 0.19 Table 3-82. Deleterious materials and fiber content.

92 1.0 10.0 100.0 1,000.0 10,000.0 1.00E-06 1.00E-04 1.00E-02 1.00E+00 1.00E+02 1.00E+04 D yn am ic M od ul us (k si ) Reduced Frequency (Hz) WI Control WI Rediset WI Zycotherm Figure 3-78. Wisconsin dynamic modulus master curves. 1.0 10.0 100.0 1,000.0 10,000.0 D yn am ic M od ul us (k si ) Figure 3-79. Alabama dynamic modulus master curves.

93 The master curves for the Tennessee mixtures are shown in Figure 3-80. Despite the WMA mixture having a substantially higher asphalt content, the HMA mixture was stiffer than the WMA mixture throughout most of the master curve. Figure 3-81 shows the master curves for the four mixtures from North Carolina. The dynamic modulus results for the two HMA mixtures were similar, which indicates that the type of RAS did not have an effect on mix stiffness. The WMA mixtures each had a much lower stiffness compared to the HMA mixtures, and—contrary to the two HMA mixtures— the WMA mixture containing MW–RAS was less stiff than the WMA mixture containing PC–RAS. Figure 3-82 shows the master curves for the two Indiana mixtures. There appears to be no difference between the stiffness of the HMA and WMA mixtures. Table 3-83 provides a summary of the master curve regres- sion coefficients generated using the modified Mechanistic– Empirical Pavement Design Guide master curve model. Goodness of fit parameters are also shown. Hamburg Wheel-Tracking Test The Hamburg Wheel-Tracking Tests were conducted in accordance with AASHTO T 324 at 50°C using samples compacted on site without reheating. NCHRP Report 673 recommends a maximum rut depth of 0.5 in. (12.5 mm) at 20,000 passes and a stripping inflection point (SIP) of not less than 10,000 passes for PG 64-XX or softer binders as cri- teria for the Hamburg Wheel-Tracking Test. All of the mixes met both criteria, as shown in Table 3-84. Figure 3-83 shows a chart of the Hamburg Wheel-Tracking Test results for all mixtures, ordered from lowest to highest rutting. The five mixtures with lowest rut depths were HMA mixtures, and the two mixtures with highest rut depths were WMA mix- tures. This is consistent with the dynamic modulus results, which showed that HMA mixtures are typically stiffer than corresponding WMA mixtures. Flow Number As requested by the project panel, Flow Number tests were conducted on dynamic modulus samples, following AASHTO TP 79-11. Four samples of each mixture were tested in an unconfined state using 87 psi of deviator stress. The test temperatures were chosen as the 50% reliability temperature from LTPPBind at 20 mm below the surface of the pavement at the paving location. These temperatures were 48.5°C for Wisconsin, 60.5°C for Alabama, 56.5°C for Tennessee, 58°C for North Carolina, and 51°C for Indiana, as shown in Table 3-85. Flow number results for each mix- ture were classified for the reported traffic for each proj- ect, according to the recommended criteria from AASHTO TP 79. Overall, flow number results exceeded the minimum criteria recommended for the respective design traffic. 1.0 10.0 100.0 1,000.0 10,000.0 D yn am ic M od ul us (k si ) Figure 3-80. Tennessee dynamic modulus master curves.

94 1.0 10.0 100.0 1,000.0 10,000.0 D yn am ic M od ul us (k si ) HMA MW–RAS HMA PC–RAS WMA PC–RAS WMA MW–RAS Figure 3-81. North Carolina dynamic modulus master curves. 1.0 10.0 100.0 1,000.0 10,000.0 D yn am ic M od ul us (k si ) Figure 3-82. Indiana dynamic modulus master curves.

95 Location Mix ID Max. E* (ksi) Min. E* (ksi) Beta Gamma Ea R 2 Se/Sy Wisconsin Control 3,128.7 8.69 -0.749 -0.524 186,531 0.997 0.037 Rediset 3,132.6 8.31 -0.625 -0.526 170,765 0.995 0.050 Zycotherm 3,121.6 7.99 -0.680 -0.504 182,310 0.996 0.042 Alabama Low Va HMA 3,116.6 3.59 -1.329 -0.451 203,433 0.999 0.019 Low Va WMA 3,122.3 3.42 -1.131 -0.482 202,939 0.999 0.025 Adj. Va HMA 3,162.9 5.85 -1.340 -0.491 197,753 0.998 0.032 Adj. Va WMA 3,158.4 3.62 -1.351 -0.463 198,600 0.999 0.024 Tennessee HMA 3,220.7 5.63 -1.247 -0.528 203,066 0.998 0.028 WMA 3,140.4 8.78 -0.860 -0.593 200,477 0.996 0.043 North Carolina HMA MW–RAS 3,181.5 5.52 -1.074 -0.485 198,723 0.999 0.021 WMA MW–RAS 3,128.2 4.81 -0.733 -0.561 182,240 0.997 0.040 HMA PC–RAS 3,120.5 3.23 -1.182 -0.452 195,575 0.999 0.018 WMA PC–RAS 3,119.7 3.77 -0.918 -0.501 191,667 0.999 0.025 Indiana HMA 3,250.7 8.15 -1.351 -0.526 185,979 0.999 0.019 WMA 3,216.8 7.90 -1.358 -0.512 191,034 0.999 0.020 Table 3-83. Master curve coefficients. Location Mix ID Average Rut Depth at 20,000 passes (mm) Standard Deviation (mm) Stripping Inflection Point (passes) Wisconsin Control 1.87 0.08 20,000+ Rediset 2.49 0.75 20,000+ Zycotherm 2.31 0.32 20,000+ Alabama Low Va HMA 1.63 0.17 20,000+ Low Va WMA 4.02 0.30 20,000+ Adj. Va HMA 1.35 0.22 20,000+ Adj. Va WMA 2.08 0.05 20,000+ Tennessee HMA 2.52 0.42 20,000+ WMA 4.98 1.48 18,100+ North Carolina HMA MW–RAS 1.68 0.22 20,000+ WMA MW–RAS 2.90 0.22 20,000+ HMA PC–RAS 1.62 0.06 20,000+ WMA PC–RAS 2.54 0.40 20,000+ Indiana HMA 2.96 0.81 20,000+ WMA 2.50 0.36 20,000+ Table 3-84. Hamburg Wheel-Tracking Test results.

96 AL Ad j. V a H M A AL Ad j. V a W M A IN W M A TN H M A IN H M A TN W M A AL Lo w Va H M A NC H M A P C– RA S W I C on tro l W I R ed ise t W I Z yc ot he rm NC H M A M W –R AS AL Lo w Va W M A NC W M A P C– RA S NC W M A M W –R AS Figure 3-83. Hamburg Wheel-Tracking Test performance ranking. Location Mix ID Temp. (°C) Flow Number (cycles) Design ESAL Range (× 106 ESALs) Actual Design Traffic (× 106 ESALs) Average Standard Deviation Wisconsin Control 48.5 163 51.5 10 to <30 1–3Rediset 120 100.9 Zycotherm 117 62.2 Alabama Low Va HMA 60.5 123 28.3 10 to <30 10 to <30 10 to <30 10 to <30 1.5 Low Va WMA 28 1.5 <3 Adj. Va HMA 119 30.1 Adj. Va WMA 106 14.1 Tennessee HMA 56.5 195 55.7 NA WMA 46 5.7 North Carolina HMA MW–RAS 58.0 150 49.0 0.3–3 WMA MW–RAS 18 2.4 <3 HMA PC–RAS 124 6.6 WMA PC–RAS 33 1.3 Indiana HMA 51.0 593 90.3 0.3–3 WMA 530 40.8 Note: NA = not available. 10 to <30 3 to <10 3 to <10 3 to <10 3 to <10 Table 3-85. Flow Number Test results.

97 The last column in Table 3-86 includes an interpolated number of cycles at a strain level of 400 µ to have a common baseline for comparisons. The ranking arrangement showed mixed results with no performance tendency between HMA and WMA (Figure 3-84). Energy Ratio Energy ratio is determined using a combination of three tests: Resilient Modulus, Creep Compliance, and Indirect Tensile Strength. Roque et al. (2004) recommend a minimum DSCEHMA of 0.75 and a minimum energy ratio of 1.95. It is important to remember that these criteria were developed Location Mix ID Strain 1 Average Nf Strain 2 Average Nf Endurance Limit Million Nf @ 400 Wisconsin Control 500 287,530 250 126,510,069 258 3.28 Rediset 500 258,840 250 81,180,843 241 2.58 Zycotherm 500 339,997 250 92,643,767 228 3.20 Alabama Low Va HMA 600 53,803 300 4,074,857 169 0.96 Low Va WMA 600 107,263 300 7,841,577 197 1.88 Adj. Va HMA 600 91,153 300 1,451,193 83 0.58 Adj. Va WMA 600 76,497 300 4,453,407 140 1.15 Tennessee HMA 600 66,908 300 2,842,008 149 0.81 WMA 600 59,745 300 1,105,585 94 0.42 North Carolina MW–HMA 700 45,565 350 1,830,803 165 1.10 MW–WMA 700 38,264 350 861,843 113 0.55 PC–HMA 700 17,427 350 954,994 139 0.54 PC–WMA 700 25,221 350 820,699 118 0.50 Indiana HMA 600 59,437 300 2,245,642 143 0.67 WMA 600 89,964 300 2,473,384 112 0.82 Table 3-86. Bending Beam Fatigue Test results. AL Ad j. V a H MA AL Ad j. V a W MA IN W MA TN HM A IN HM A TN W MA AL Lo w Va HM A NC HM A P C– RA S WI Co ntr ol WI Re dis et WI Zy co the rm NC HM A M W– RA S AL Lo w Va W MA NC W MA M W– RA S NC W MA PC –R AS Figure 3-84. Cycles at 400  ranking. Bending Beam Fatigue Bending Beam Fatigue tests were conducted using re-heated plant mix samples, in accordance with AASHTO T 321-14. Three beams each were tested at two strain magnitudes. For the Wisconsin project, several beams did not fail at 250 µ after 12 million cycles, so the strain levels were adjusted for the other projects. Table 3-86 shows the results of the beam fatigue tests. For the Wisconsin mixture results, failure was based on the old AASHTO criteria of a 50% reduction in beam stiff- ness. For the other projects, the new method for determining failure based on peak modulus × cycles was used. The fatigue endurance limits were based on the old failure criterion.

98 for Florida’s climate and materials. Different criteria may be required for other regions. The results for each test are summarized in Table 3-87. All of the Wisconsin and Tennessee mixtures met the recom- mended Florida DSCEHMA and energy ratio criteria. For the Alabama mixtures, only the low air voids WMA mixture had a DSCEHMA greater than 0.75, which is considered a mini- mum threshold value to discriminate brittle mixtures. The Alabama adjusted air void HMA mixture had the lowest DSCEHMA value of 0.08, and the lowest fracture energy of 0.2, indicating a mixture highly susceptible to cracking. Indirect Tension Creep Compliance and Strength The critical thermal cracking temperature for a mixture is the temperature at which the estimated thermal stress exceeds the mixture’s tensile strength. The Indirect Tensile Creep Compliance and Strength tests were conducted for three replicates of each mix, as specified in AASHTO T 322-07. A thermal coefficient of each mixture was estimated based on its volumetric properties and typical values for the thermal coefficient of asphalt and aggregate. The result of the thermal cracking analysis is a single critical low pavement tempera- ture, each of which is given in Table 3-88. Results were mixed in regard to critical temperatures. In some cases (Wisconsin, North Carolina with MW–RAS, and Indiana), the use of the WMA technology produced slightly warmer critical temperatures, but the opposite effect can be observed in the remaining cases. As can be seen, the results of the WMA for each project were within 2°C of the results for the HMA mixtures. No trends were evident between WMA and HMA results. However, the critical mixture temperatures were at least 5°C lower than the respective bending beam rheometer critical temperatures for the recovered binders in all cases. Overlay Test Overlay tests were performed using the first generation Over- lay Test jig in the IPC Global Asphalt Mixture Performance Tes- ter in accordance with the Texas DOT 248-F parameters, which specify a cyclic maximum opening displacement of 0.025 in. and a test temperature of 25°C. In addition to the standard proce- dure, samples from Wisconsin were also tested at 10°C using a smaller (0.015 in.) displacement to represent the colder climate (Table 3-89). Figure 3-85 shows a tendency for WMA mixtures to perform better than HMA mixtures, with WMA mixtures as the top five Overlay Test performers. In addition, each WMA mixture outperformed its corresponding HMA pair. Fracture Energy and Flexibility Index I-FIT was performed on reheated plant-produced mix in accordance with AASHTO T 324-16. The test is performed on notched semicircular specimens that are loaded monotonically to failure at a rate of 50 mm/min at a temperature of 25°C. The load versus displacement curve for each specimen is used to determine a flexibility index value. A higher flexibility index is generally indicative of a mixture with greater resistance to cracking. The Illinois DOT has set a preliminary minimum Location Mix ID m-value St (MPa) Mr (GPa) Fracture Energy (kJ/m3) DSCEHMA (kJ/m3) Energy Ratio Wisconsin Control 0.386 2.14 10.14 3.4 3.17 3.23 Rediset 0.447 2.13 9.34 4.6 4.36 3.73 Zycotherm 0.441 2.03 9.81 4.1 3.89 2.75 Alabama Low Va HMA 0.328 2.13 13.97 0.6 0.44 1.70 Low Va WMA 0.419 2.44 10.72 2.0 1.72 1.92 Adj. Va HMA 0.266 1.9 14.94 0.2 0.08 0.59 Adj. Va WMA 0.385 2.16 14.23 0.9 0.74 2.03 Tennessee HMA 0.420 2.75 12.12 3.4 3.09 4.48 WMA 0.521 2.4 11.12 5.0 4.74 3.06 North Carolina HMA MW–RAS 0.426 1.77 12.26 0.3 0.17 0.29 WMA MW–RAS 0.586 2.26 9.01 6.2 5.92 2.12 HMA PC–RAS 0.406 2.59 11.43 2.8 2.51 3.89 WMA PC–RAS 0.479 2.39 8.69 5.0 4.67 2.44 Indiana HMA 0.424 2.70 13.50 3.20 2.93 4.39 WMA 0.429 2.45 13.18 2.60 2.37 3.51 Table 3-87. Energy ratio results.

99 Location Mix ID Average Uncorrected Strength (psi) Average Corrected Strength (psi) Critical Pavement Temperature (°C) Critical Air Temperature (°C) Wisconsin Control 437 379 -20 -25 Rediset 428 372 -19 -24 Zycotherm 423 368 -18 -23 Alabama Low Va HMA 470 405 -17 -21 Low Va WMA 439 380 -18 -23 Adj. Va HMA 516 440 -21 -26 Adj. Va WMA 460 397 -18 -23 HMA 498 427 -21 -26 WMA 459 396 -21 -26 North Carolina HMA MW–RAS 496 425 -20 -26 WMA MW–RAS 468 403 -20 -25 HMA PC–RAS 479 411 -17 -22 WMA PC–RAS 490 420 -20 -25 Indiana HMA 510 436 -22 -28 WMA 478 411 -22 -28 Tennessee Table 3-88. Indirect Tensile Test critical temperature analysis. Location Mix ID Temperature (°C) Displacement (in.) Cycles Until Failure Average Standard Deviation Wisconsin Control 10 0.015 792 752.1 Rediset 1,320 – Zycotherm 1,903 705.6 Control 25 0.025 241 83.8 Rediset 285 51.1 Zycotherm 436 96.4 Alabama Low Va H MA 25 0.025 19 0.6 Low Va W MA 214 69.1 Adj. Va H MA 24 8.4 Adj. Va WMA 44 5.6 Tennessee HMA 25 0.025 226 55.4 WMA 807 148.2 North Carolina HMA MW–RAS 25 0.025 125 78.6 WMA MW–RAS 619 88.4 HMA PC–RAS 215 54.9 WMA PC–RAS 333 142.2 Indiana HMA 25 0.025 109 30.3 WMA 158 71.1 Note: – = only two data points available. Table 3-89. Overlay Test results.

100 AL Ad j. V a H M A AL Ad j. V a W M A IN W M A TN H M A IN H M A TN W M A AL Lo w V a H M A NC H M A P C– RA S W I C on tro l W I R ed ise t W I Z yc ot he rm NC H M A M W –R AS AL Lo w V a W M A NC W M A M W –R AS NC W M A P C– RA S O ve rl ay T es t C yc le s Figure 3-85. Overlay Test cycles to failure ranking. criterion for the flexibility index at 8.0. None of the tested mix- tures met the preliminary criterion of 8.0 (Table 3-90). The North Carolina WMA MW–RAS mixture had the highest flex- ibility index of 7.31. Three of the Alabama mixes had flexibility index results below 1.0; those mixes also had low Overlay Test cycles to failure. In most cases (except the Wisconsin Zycotherm mixture), the WMA mixture performed better than its corre- sponding HMA pair. Semi-Circular Bend–LTRC Test Table 3-91 shows the results of the Semi-Circular Bend tests conducted on reheated mixture samples. For each mixture, four specimens at three notch depths (25.4 mm, 31.8 mm, and 38.0 mm) were loaded monotonically at a rate of 0.5 mm/min at a 25°C test temperature until failure. The area under the load–displacement curve to the peak for each specimen is measured and then plotted against the specimen notch depth. This model allows for calculation of the Jc parameter, which is used as a measure of fracture resistance. The Semi-Circular Bend–LTRC method yields a singular Jc result with a minimum specified value of 0.5 or 0.6, depend- ing on the traffic level. Little differences were obtained with regard to slope and Jc values for the Wisconsin, Tennessee, and Indiana projects. However, greater differences can be observed between WMA and HMA mixtures for the Alabama and North Location Mix ID Fracture Energy Average (J/m2) Fracture Energy SD (J/m2) Flexibility Index Average Flexibility Index Standard Deviation Wisconsin Control 1,478.7 122.2 3.33 0.52 Rediset 1,696.2 149.0 5.77 1.76 Zycotherm 1,605.1 157.8 2.90 0.47 Alabama Low Va HMA 1,180.2 222.6 0.66 0.40 Low Va WMA 1,577.6 77.1 2.93 0.65 Adj. Va HMA 839.9 113.0 0.15 0.05 Adj. Va WMA 1,287.5 186.7 0.97 0.16 Tennessee HMA 1,860.9 209.1 3.34 0.90 WMA 2,112.1 211.5 4.92 0.73 North Carolina HMA MW–RAS 1,488.4 79.0 1.77 0.56 WMA MW–RAS 2,042.1 139.9 7.31 0.56 HMA PC–RAS 1,758.6 60.6 3.69 0.81 WMA PC–RAS 1,679.7 146.7 4.67 0.52 Indiana HMA 1,507.2 128.7 1.05 0.39 WMA 1,591.5 90.9 1.72 0.17 Table 3-90. Fracture energy and flexibility index results.

101 Carolina projects. For the Alabama mixtures, higher Jc values indicate an improvement in fracture resistance because of the use of WMA technology. For the North Carolina mixtures, only the HMA PC–RAS mixture appears to have a practical difference in Jc results. Overall, there was no apparent trend between HMA and WMA mixtures for this test. Cracking Susceptibility of Cores Obtained from Field Inspections Cores obtained from the final site revisits for all proj- ects were tested using the I-FIT. A thickness correction was applied to the flexibility index results, since most cores were not 50 ± 1 mm-thick after trimming. Corrected values were obtained by multiplying the measured flexibility index result by the core thickness and dividing by 50 mm. Corrected values for all existing and new projects except Indiana are shown in Table 3-92. For the existing projects, the pavements were 3 to 4 years old; for the new projects, the pavements were 1 to 2 years old. For both existing projects in Texas, the flexibility index results on the cores from all of the sections were less than 1.0, indicating that the pavements are highly susceptible to cracking. Cores from the existing project in Illinois had flex- ibility index values of 7.9 and 10.1. These results are con- sidered very good for nearly 4-year-old pavements. Illinois DOT’s preliminary flexibility index criterion is 8.0. The Illi- nois project was the only project in this study to use stone matrix asphalt mixtures. Of the new projects, flexibility index results ranged from 0.11 to 5.92. Like the cores from Texas, all of the Alabama pavements had flexibility index results less than 1.0. Overall, results on field cores indicate that WMA and HMA mixtures are generally not substantially different. The variability of these flexibility index results was higher than the results for the laboratory-compacted plant-mix samples. This is caused by two factors: • Coefficient of variation (CV) (%) values tend to be higher for low flexibility index values. CV is not a particularly good variability metric when measuring very low magni- tudes of a quantity. For three of the seven projects, this was a big driver of high CV values. • It would be expected that field cores would have greater variability in air voids than specimens produced in the lab- oratory because of production and construction variability. A comparison of flexibility index for laboratory-produced and field samples is shown in Figure 3-86. The flexibility index of the laboratory-prepared samples and cores for the Wisconsin project were similar. All three mixtures from the Wisconsin project were produced at HMA temperatures. The cores from the Alabama project had much lower flex- ibility index than the corresponding laboratory samples. For the North Carolina project, two of the four core sets had lower flexibility index results compared to the laboratory- compacted samples. For the Tennessee project, the core flexibility index was slightly higher than the laboratory- prepared flexibility index for one mix and much lower than the laboratory-prepared flexibility index for the other mix. The flexibility index results for HMA mixtures were closer to the line of equality than the WMA mixtures. Perhaps this indi- cates that using WMA may provide a false sense of improved cracking resistance in plant-produced mixtures containing RAS if reheated laboratory samples are used to evaluate mix properties. Field performance and field core properties indicate that WMA and HMA mixtures are generally not substantially different. Location Mixture Slope Intercept R2 Jc (kJ/m 2) Wisconsin Control -0.021 1.093 0.845 0.37 Rediset -0.023 1.177 0.898 0.41 Zycotherm -0.020 1.063 0.678 0.36 Alabama Low Va HMA -0.023 1.105 0.985 0.41 Low Va WMA -0.039 1.749 0.909 0.69 Adj. Va HMA -0.027 1.233 0.844 0.47 Adj. Va WMA -0.037 1.675 0.898 0.65 Tennessee HMA -0.036 1.712 0.854 0.64 WMA -0.036 1.809 0.840 0.64 North Carolina HMA MW–RAS -0.018 1.002 0.725 0.32 WMA MW–RAS -0.022 1.154 0.833 0.38 HMA PC–RAS -0.032 1.531 0.927 0.57 WMA PC–RAS -0.023 1.204 0.759 0.40 Indiana HMA -0.029 1.379 0.745 0.50 WMA -0.031 1.558 0.800 0.55 Table 3-91. Semi-Circular Bend–LTRC Test results.

102 Location and Age Mix ID Sample Air Voids (%) Flexibility Index Thickness Corrected Average Average Standard Deviation CV (%) Austin, Texas 47 months HMA, PG 64-22 15% RAP, 3% RAS 8.8 0.28 0.10 33.5 HMA, PG 64-22, 5% RAS 5.2 0.09 0.05 58.4 HMA, PG 58-28 15% RAP, 3% RAS 8.8 0.38 0.25 65.8 Evotherm 3G, PG 64-22 15% RAP, 3% RAS 8.5 0.64 0.33 51.3 Illinois 46 months Gravel 4.6 7.93 3.72 46.9 Quartz 4.5 10.13 2.83 27.9 Fort Worth, Texas 37 months HMA 7.3 0.11 0.08 66.9 WMA 8.1 0.10 0.07 69.6 Wisconsin 24 months Control 7.7 3.86 0.54 14.0 Rediset 7.5 5.92 1.77 29.8 Zycotherm 6.7 4.12 0.59 14.4 Alabama 29 months Low Va HMA 5.2 0.11 0.04 41.0 Low Va WMA 5.6 0.36 0.27 76.3 Adj. Va HMA 6.2 0.14 0.20 141.3 Adj. Va WMA 5.6 0.20 0.21 102.4 North Carolina 14 months HMA MW-RAS 8.4 2.03 0.88 43.3 HMA PC–RAS 6.9 1.48 0.49 33.3 WMA MW–RAS 7.0 2.51 0.85 33.8 WMA PC–RAS 7.3 1.70 0.94 55.2 Tennessee 25 months HMA 6.8 4.61 3.22 69.8 WMA 6.7 1.87 1.11 59.6 Table 3-92. Flexibility index of cores from last site inspections. I-F IT –F ie ld C or es I-FIT–Laboratory Mixtures Figure 3-86. Flexibility index values of laboratory compacted samples versus field samples.