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.

1 Background New pavement technologies are continuously being devel- oped and evaluated by highway agencies and the asphalt pavement industry to reduce the consumption of natural resources and costs of asphalt paving mixtures. Two technol- ogies that have received substantial attention in the last few years are the use of recycled asphalt shingles (RAS) and warm mix asphalt (WMA). These new technologies address impor- tant issues that face the asphalt industry in different ways. RAS contains a high percentage of asphalt binder, which can be used to reduce the amount of virgin asphalt binder needed. Since binder is the most expensive component of an asphalt mixture, the use of RAS can significantly reduce mixture costs. On the other hand, WMA uses additives or other means of decreasing the viscosity of asphalt binders to allow lower production and compaction temperatures, com- pared to conventional hot-mix asphalt (HMA). Lowering the temperature reduces the amount of energy required for the production of asphalt paving mixtures, thereby reducing emissions and production costs. Since the first WMA technologies in the United States were introduced about a decade ago, WMA has quickly become widely accepted among highway agencies and contractors. Although RAS has had limited use in asphalt paving mix- tures for several decades, its use increased significantly over the past 10 years primarily because of spikes in virgin asphalt prices between 2008 and 2015. The convergence of these two technologies has raised some concerns. Some experts in the industry have questioned if the lower mix temperatures with WMA are sufficient to soften and activate hard RAS binders. Hence, research was needed to determine the amount of mix- ing between RAS binders and virgin binders when WMA is used. Additional guidance was also needed for designing, producing, and constructing asphalt mixtures that use both RAS and WMA to address several gaps in the state-of-the- knowledge on how these two technologies work, or perhaps, do not work together. Therefore, this study was initiated to address three needs with regard to the use of RAS and WMA: • Evaluate the short-term performance of asphalt mixtures that use RAS in conjunction with WMA. • Quantify the effect of RAS on asphalt mixture properties. • Develop guidance for designing and constructing WMA– RAS mixtures. Introduction In the early 1980s, asphalt paving technologists first con- ceived replacing virgin asphalt binder with the asphalt binder from RAS (Davis 2009). In the 1990s, the introduction of Superpave performance-grade (PG) binder specifications brought about an increase in the use of higher cost polymer- modified asphalt binders. Other market forces also drove up the price of unmodified asphalt binders. In the decade between 2003 and 2013, unmodified asphalt prices increased from around $200 per ton to around $600 per ton (Alabama Department of Transportation [DOT] 2014). These cost increases motivated practitioners and researchers to explore alternative means to provide adequate resistance to perma- nent deformation. The most common approach to address pressures of rising prices was to increase the percentages of reclaimed asphalt pavement (RAP) in asphalt mix designs. Another approach that has been used more recently was to incorporate a small percentage of RAS (typically 5% of the mixture or less) to replace 10% to 30% of the virgin asphalt binder (Hansen and Newcomb 2011). It has been estimated that 11 million tons of waste roofing shingles are generated each year in the United States (McGraw et al. 2007, Davis 2009). Ten million tons of these waste shin- gles are generated from re-roofing houses. This source of waste shingles is referred to as tear-off or post-consumer (PC) shingles. For example, in the state of Massachusetts alone, C H A P T E R 1

2 it is estimated that 210,344 tons of PC shingles are generated each year (Bauman 2005). The properties of the PC shin- gles vary depending on their original composition and the amount of time that they have been exposed to weathering, especially exposure to sunlight. The other one million tons of shingles come from manu- facturer’s waste (MW) or factory rejects that may have some minor deficiency preventing them from meeting the speci- fications required for the roofing industry. Since these MW asphalt shingles have not been exposed to the sun, the asphalt in these shingles is not as oxidized and is less hard than PC– RAS. The MW shingles are also less likely to have contamina- tion from other roofing components such as nails, paper, and pieces of wood (Hansen and Newcomb 2011). As the testing standards and specifications for RAS use have constantly evolved at the national and state levels, it is chal- lenging to define the state of practice. Table 1-1 provides the latest information that the authors could find on current state practices. Most specifications require the contractor to choose either MW or PC shingles for an individual mix—if a state does not disallow one RAS type—and discourage mixing the two materials (Dan Krivit Associates 2007, Mallick et al. 2000). At the time of this report, RAS had been effectively used in projects throughout Texas, Missouri, North Carolina, Georgia, and Minnesota (Davis 2009, Krivit 2007, Kandhal 1992, Carolina Asphalt Pavement Association 2011, Watson et al. 1998), as well as other states shown in Table 1-1. According to Hansen and Copeland (2015), the use of RAS in asphalt pavement mixtures has increased from 701,000 tons in 2009 to an estimated 1.93 million tons in 2015, which is a slight (1.6%) decline from RAS tonnage in 2014. A moderate increase in the tons of RAS used by state DOTs was reported from the 2014 to 2015 construction season because of an increase in total state DOT mix tonnage and the percent- age of RAS used in state DOT mixtures. In addition, most states responding to the annual survey on recycled materials and WMA usage reported no limits on the use of RAS in the Commercial and Residential sectors, and some states do not allow RAS in agency mixtures (Hansen and Copeland 2015). While many pavement materials engineers are using RAS for the economic benefits, others focus on the sustainability aspects of using recycled materials. Technical developments continue to be aimed at the reduction of energy consump- tion and greenhouse gas emissions, the usage of renewable resources, and recycling. The concepts associated with sus- tainability are becoming factors in the decision-making pro- cess of roadway projects. Sustainability rating and labeling systems are now being put into practice to benchmark envi- ronmental benefits and recognize best practices while still addressing the functionality of materials and construction practices (Croteau et al. 2009). Project Objectives and Scope The objective of this research was to develop an evaluation procedure that provides acceptable performance of asphalt mixtures incorporating WMA technologies and RAS—with and without RAP—for project-specific service conditions. This research was envisioned to address the following issues: (1) minimizing the risk of premature failure caused by design- ing and producing mixes containing WMA technologies and RAS with poor constructability and durability; (2) evaluating the type, source, quality, and characteristics of RAS, with and without RAP; (3) selecting the virgin binder and evaluating Specification State Allows <3% PC or MW–RAS Alabama, Missouri Allows <5% PC–RAS Georgia, Illinois, Iowa, Nebraska, North Carolina, Oregon, Texas, Virginia, California Allows <5% MW–RAS Alabama, Georgia, Illinois, Iowa, Maryland, Massachusetts, Nebraska, New Jersey, North Carolina, Oregon, Pennsylvania, Texas, Virginia, California Allows <8% RAS South Carolina Allows 25% maximum binder replacement from RAS Indiana, Wisconsin Allows <50% binder replacement Michigan Follows AASHTO MP 15 New Hampshire, Kentucky Making efforts with RAS Arkansas, Mississippi, Montana, North Dakota, Colorado Little to no use of RAS Alaska, Arizona, Connecticut, Hawaii, Idaho, Florida, Kansas, Louisiana, Maine, Nevada, New Mexico, New York, Ohio, Oklahoma, Rhode Island, South Dakota, Tennessee, Utah, Vermont, Washington, West Virginia, Wyoming Not allowed in state DOT mixtures California, Colorado, Connecticut, Florida, Mississippi, New Hampshire, Oklahoma, Vermont Limits the total amount of recycled material Delaware, Minnesota Table 1-1. Current specifications.

3 the composite binder; (4) assessing the current range of asphalt mixture production temperatures; and (5) evaluat- ing the mixing efficiency of RAS and virgin binders. This research was divided into two phases. The first phase involved literature reviews on RAS and WMA and the devel- opment of experimental plans to accomplish the research objectives. An interim report that included the literature reviews, the current state-of-knowledge report, and the ampli- fied experimental plan was submitted to NCHRP in 2013. The second phase of the project involved executing the approved experimental plans to gather materials from field projects, evaluate the engineering properties of WMA and HMA mixtures containing RAS, compare the early life field performance of selected pavement sections constructed with WMA–RAS and HMA–RAS mixtures, develop protocol revi- sions, perform cost–benefit analysis, provide best practices documents, and prepare this final report. Report Organization This report includes the experiments related to the analysis of engineering properties of WMA–RAS mixtures compared to HMA–RAS mixtures and the early field performance of test sections built across the United States. Chapter 1 presents the report introduction, objectives of the project and scope of work, and a brief literature review covering a summary of RAS characterization recommendations, RAS mixture per- formance information, and a summary of field performance of WMA–RAS and HMA–RAS experimental pavement test sections. The experimental plans for field construction docu- mentation and monitoring, as well as laboratory testing, are presented in Chapter 2. Chapter 3 presents information on the field projects and outcome of the laboratory test results. Chapter 4 presents analysis of the engineering properties with an emphasis on comparing results of HMA–RAS mixtures and WMA–RAS mixtures. Chapter 5 covers mix design veri- fications in accordance with the revised AASHTO PP 53-09 Provisional Standard. Chapter 6 provides a brief economic analysis of using RAS in asphalt mixtures. And Chapter 7 summarizes the project findings and presents suggestions for modifying current practice. Literature Review Composition of Asphalt Shingles Asphalt roofing materials include composition shingles, built-up roofing, and torch-down roofing (a polymer-modified asphalt membrane that is strengthened with fabrics com- monly used on flat roofs). The major components of asphalt roofing waste include asphalt, mineral filler and granules, glass fiber matting, organic paper felt, and nails. There are a number of potential end usages for asphalt roofing waste including asphalt mixtures, which is currently the largest market for RAS. When RAS is used in asphalt mixtures, the percentage of virgin asphalt is typically reduced. The mineral aggregate in the shingles provides a source of fine aggregate, reducing the demand for virgin aggregate. Numerous studies have shown that mixtures containing RAS have improved rutting resis- tance. Some studies have also shown mixtures containing RAS to have good cracking resistance, but other studies have found conflicting results (Zhang 2011). While the composition of shingles varies depending on manufacturer and roofing application, most RAS is composed of four basic materials: asphalt cement, felt or fiber, mineral or ceramic aggregate granules, and mineral filler. Organic or fiberglass felt backings provide the basic structure for shingles. The organic felt is typically composed of cellulose fibers and is designed to support the asphalt and aggregate granules. Fiber- glass backings are manufactured by mixing fine glass fibers with water in the form of a glass pulp, which is then formed into a fiberglass sheet (Blachford and Gale 2002, Grodinsky et al. 2002). In the manufacture of shingles, the backing is saturated with asphalt binder. This asphalt binder has been air blown to increase its stiffness compared to conventional paving asphalt. The asphalt is further stabilized with lime dust (70% passing the No. 200 sieve) (3M Corporation 2007, Townsend et al. 2007). A second application of air-blown asphalt is applied, and then the surface of the shingle is covered with granules. These granules are designed to protect the asphalt from the sun’s ultraviolet rays and physical damage caused by abrasion on rooftops. Most shingle manufacturers use crushed fine aggregate coated with ceramic metal oxides as granules. Addi- tional headlap granules can be used in this application. Both aggregate granules are ideal for roofing shingles because of their uniform size, toughness, and angular shape (Grodinsky et al. 2002). In some cases, chemicals are added to the aggre- gate to prevent algae growth (3M Corporation 2007). Shingles are finished with a dusting of fine mineral material on the back surface to prevent the shingles from stick- ing together prior to installation. A schematic of the final prod- uct is shown in Figure 1-1. Table 1-2 presents estimates of the percentage of each material in organic and fiberglass shingles. It can be seen that fiberglass mat shingles have lower asphalt content than organic felt shingles. In addition to differences in the stiffness of the asphalt for MW and PC shingles caused by aging, loss of aggregate particles on roofs over time and during removal generally causes PC shingles to have higher asphalt content than the MW shingles. PC shingles typically contain more deleterious materials—such as paper, wood, and nails—than do MW shingles. While many of these contaminants may be removed

4 during the grinding process, minimizing such contaminants prior to processing is recommended (Grodinsky et al. 2002). PC shingle stockpiles also tend to exhibit more variability in size, aggregate gradation, and asphalt content—as well as material properties such as specific gravity—than do MW shingles. Shingle type, manufacturer, and age all can signifi- cantly influence these factors (Foo et al. 1999). RAS Asphalt Quantifying the asphalt content of RAS is a critical part of material characterization for asphalt mixture design. PC shingles can average 30% to 40% asphalt binder, while MW shingles typically have 15% to 25% asphalt binder (Krivit 2007). Two methods may be used for determining the asphalt content of RAS: the ignition method and solvent extraction. The Texas DOT completed a study comparing the asphalt content of RAS determined by the ignition method and solvent extraction (Zhou et al. 2012). The ignition method consistently resulted in higher asphalt content than from sol- vent extractions. This higher asphalt content of the ignition method may be caused by burned-off fibers and the loss of mineral fines in the furnace. However, this increase in asphalt content was relatively small when the error was compared to the RAS binder content. RAS Aggregate RAS contains aggregate granules, which contribute to the fine aggregate portion of asphalt mixtures (Tighe et al. 2008). The gradation and bulk specific gravity (Gmb) of the RAS aggregate must be properly determined before the RAS is used in an asphalt mixture. RAS aggregate properties are determined by testing the aggregate granules that have been separated from the other RAS components. Either the solvent extraction method or the ignition method is appropriate for recovering the RAS aggregate. AASHTO TP 2 provides guid- ance for conducting solvent extraction and recovery of the asphalt binder and aggregate if recovering the asphalt binder for further testing is required. However, if asphalt binder recovery is not necessary, the ignition method defined in ASTM D228 Section 13 and Section 14 can be used to recover the RAS aggregate. RAS Fibers Fibers (either in the form of cellulose or fiberglass) are com- monly present in RAS. Past research has shown that MW shingles can average 1.7% fibers by weight of the shingle; however, this value can be much higher (McGraw et al. 2007). While the fibers are an integral part of RAS, very little work has been done in quantifying the effect of RAS fibers in asphalt mixtures. Some have suggested that the fibers might increase mixture durability; however, no definitive work has been com- pleted to confirm this conclusion. It has been suggested that the potential presence of asbes- tos fibers may be one of the greatest challenges in dealing with recycling PC shingles (Zickell 2002). Asbestos was used in the manufacture of asphalt shingles and asphalt-containing roofing materials as early as the late 1800s and continued to the early 1980s. While asbestos fibers were present in asphalt shingles, they were typically in low percentages. The aver- age asbestos content in 1964 was 0.02%, while the amount decreased in 1973 to 0.00016% (U.S. Geological Survey 2007). Asbestos fibers have not been used in asphalt shingles since the 1980s (Townsend et al. 2007). However, it is considered a good practice for contractors using MW shingles to request documentation from shingle manufacturers stating that their shingles are free of asbestos before they are processed for use in asphalt mixtures (Dan Krivit Associates 2007). Source: Townsend et al. 2007. Figure 1-1. Schematic of asphalt shingle composition. Component Organic Felt (%) Fiberglass Mat (%) Asphalt binder 30–36 19–22 Felt (fiber) 2–15 2–15 Mineral aggregate 20–38 20–38 Mineral filler 8–40 8–40 Source: 3M Corporation 2007, Lee 2009, and National Association of Home Builders 1998. Table 1-2. Composition of shingles.

5 Mix Design One of the most contentious parts of the RAS mixture design procedure is determining how much the RAS binder actually contributes to the overall binder content of the asphalt mixture. It is now generally accepted that less than 100% of the RAS binder is effective in the mix. Therefore, if it is assumed that 100% of the RAS binder contributed to the mixture but it was not actually incorporated, the mixture would have lower effective binder content (LaPlante 2011). One reason some engineers believe 100% of the binder is not effective is because of the high stiffness of the RAS asphalt. Since the RAS binder is stiffer than a virgin binder, it requires more energy to melt and blend with the other binders in an asphalt mixture. This would require a higher mixing tempera- ture and more time in the asphalt plant. This causes many engineers to believe that a lower percentage of the RAS binder is activated (Williams et al. 2013). In 2006, AASHTO introduced the AASHTO MP 15 Stan- dard Specification for Use of Reclaimed Asphalt Shingles as an Additive in Hot Mix Asphalt (HMA). One year after the introduction of MP 15, AASHTO PP 53 was introduced as the Standard Practice for Design Considerations When Using Reclaimed Asphalt Shingles (RAS) in New Hot Mix Asphalt (HMA). This standard provided guidance on design considerations such as shingle grinding, assessment of what happens to RAS asphalt that is not activated, and effects of fibrous material. This standard practice gave practitioners a method for determining the shingle aggregate gradation and estimating the contribution of RAS binder to the total binder content. The latest AASHTO PP 78-17 Provisional Standard Practice for Design Considerations When Using Reclaimed Asphalt Shingles (RAS) in Asphalt Mixtures provides updated guid- ance for designing asphalt mixtures that contain RAS from manufactured waste or tear-off sources. Specific guidance includes how to determine the shingle aggregate gradation and specific gravity, how to adjust voids in mineral aggregate (VMA) requirements, how to test the composite binder for embrittlement using the critical low-temperature difference parameter DTc, as well as notes for the mix designer to con- sider. One note states that “[a] mixture performance test for cracking implemented by the agency is acceptable in lieu of the binder testing for DTc.” An ongoing study conducted by Zhou et al. (2013A) includes a balanced mix design procedure for RAP–RAS mixtures. The study also included a performance evaluation system in which the Hamburg Wheel-Tracking Test (HWTT) and associated criteria were used to control rutting/moisture damage. The Overlay Test was also used to evaluate cracking potential [with the required Overlay Test cycles determined from the Texas Asphalt Concrete Overlay Design and Analysis System that predicts pavement cracking], given inputs of climate, traffic, pavement structure, and existing pavement conditions. Laboratory Testing of Asphalt Mixtures Containing RAS One of the earliest laboratory studies on the use of RAS in asphalt mixtures was conducted by Newcomb et al. (1993) for the Minnesota DOT. At the time, more than 36,000 tons of MW–RAS was available in the Twin Cities area. The objective of the study was to evaluate how RAS affected asphalt mixture properties for both dense-graded mixtures and stone matrix asphalt. The dense-graded mixtures were evaluated using one aggregate gradation, three proportions of RAS, two binder grades (85/100 and 120/150 Penetration Grade), and both PC and MW–RAS. The stone matrix asphalt experiment only used one binder grade (85/100), one gradation, three fibers, and a singular RAS content. Cold temperature properties of the mixtures were evalu- ated using indirect tensile testing at a slow rate of loading, and the mixtures were evaluated based on tensile strength and strain at peak stress. It was concluded that increasing the RAS content decreased the tensile strength of the asphalt mixture; however, when felt from RAS was used in the mix, the peak strain of the mix increased. The PC–RAS mixtures were more brittle. No effects of RAS were noticed with the stone matrix asphalt mixtures. Overall, the research suggested that Minnesota DOT could use up to 5% of MW–RAS in their mixtures without adversely affecting the mixture properties. However, at the time there were no facilities in Minnesota that could process RAS, which proved a difficult issue for implementing the practice. In 1996, Texas DOT sponsored a limited study to gather information needed to specify, design, produce, place, and evaluate paving mixtures containing roofing waste (Button et al. 1996). A dense-graded mixture and a coarse matrix– high binder mixture were chosen for the evaluation. The roof- ing waste materials were added to both mixtures at 5% and 10%, in conjunction with an AC-20 binder. The engineering properties of the resulting HMA mixtures were compared to virgin asphalt mixtures. One coarse matrix–high binder mixture with 5% PC–RAS incorporated an AC-10 binder to assess how the softer binder affected the asphalt mixture. The results of this study indicated that the addition of RAS always caused a reduction in the tensile strength of the asphalt mixture for the dense-graded mixtures; however, it was noted that the MW shingles had a larger reduction in strength compared to the PC shingles. Foo et al. (1999) completed a study designed to evaluate the engineering properties of asphalt mixtures with RAS. This project included blending three gradations with one

6 source of RAS at 0%, 5%, and 10% by weight of aggregates. An AC-20 asphalt binder was used in the study, meeting the requirements for a PG 64-22. The Dynamic Creep Test was used to evaluate the permanent deformation performance of the asphalt mixtures. For each of the mixes, the addition of RAS (by 5% or 10%) decreased the permanent strain of the HMA mixes. This improvement was attributed to increased stiffness of the RAS binder and/or the presence of hard sharp granules of the shingle aggregate. McGraw et al. (2007) conducted a study investigating the use of both PC and MW–RAS in combination with RAP for the state of Minnesota using the standard PG 58-28 binder for a 20% RAP mixture and two 15% RAP and 5% shingle mix- tures with different shingle sources. To determine variability in the processing of waste shingles for use in HMA, 10 random samples were taken for each type of recycled product used in the mixes in the project. Each of these recycled products was tested for asphalt content, PG on recovered binder, grada- tion, percentage of glass fiber, and paper content in extracted aggregate. Asphalt extractions were performed using AASHTO T-164 Method A (centrifuge using toluene as the solvent). Fines were removed from the extracted material by the high-speed cen- trifuge. The binder recovery method used was ASTM D 5404, and the PGs of the extracted asphalt binders were determined by AASHTO R-29. Bending Beam Rheometer tests and Direct Tension tests were also performed on the extracted asphalt specimens. It was observed from master curves of the bend- ing beam rheometer creep stiffness that the two binders with shingles were softer at the shorter times or low test tem- peratures but also had flatter curves (lower m-values) than the RAP-only binder, which makes them stiffer at the higher loading times or higher test temperatures. This indicates that the binders with shingles may behave worse with respect to fatigue cracking rather than low temperature. The binders containing MW–RAS accumulated less thermal stress than either the RAP or MW binders; however, the thermal stress gradients for all the recycled binders were less steep than the PG 58-28 binder, as calculated by AASHTO MP 1a. Direct Tension Test results exhibited similar trends for the 20% RAP and the MW–RAS, while the PC–RAS binders were more brittle than the other binders at the higher temperatures. The mixture stiffness from Indirect Tensile tests (IDT) at 100 s and 500 s—and 0°C, −10°C, and −20°C—showed that adding PC–RAS always increased mixture stiffness. The most significant increase was at the lowest temperature. MW–RAS only increased the stiffness of the mixture at 0°C and −10°C. It was the softest material at −20°C. These data indicated that the tensile strength of the mixtures was not significantly affected by RAS, which contradicted the binder strength data for the extracted PC binders. The study concluded that adding RAS to asphalt mixtures could decrease the temperature susceptibility of the binders and mixtures. The authors noted that using a softer grade of virgin binder might increase the mixture’s resistance to cracking; however, it was not part of the experimental plan. Tighe et al. (2008) also conducted a laboratory study that included five mix designs: • Mix 1 (control): Superpave 19C, virgin material; • Mix 2: Superpave 19C, 20% RAP material; • Mix 3: Superpave 19C, 20% RAP material, 1.4% shingles; • Mix 4: Superpave 19C, 20% RAP material, 3.0% shingles; and • Mix 5: Superpave 19C, 3.0% shingles. The mixtures were compared using the Dynamic Modulus Test, Resilient Modulus Test, Thermal Stress Restrained Spec- imen Tensile Strength Test, and French Wheel-Rutting Test. Dynamic modulus testing was performed following AASHTO TP 62-03. At low temperatures, Mix 1 (control) and Mix 2 (20% RAP) were the stiffest, which is indicative of lower fatigue susceptibility. At high temperatures, Mix 3 (20% RAP and 1.4% shingles), Mix 4 (20% RAP and 3.0% shingles), and Mix 5 (3.0% shingles) each had lower stiffness. Thus, the inclusion of RAS reduced the dynamic modulus, which is counter to many other studies. The Resilient Mod- ulus Test was performed following AASHTO TP 31. Mix 1 (control) had the highest resilient modulus, while Mix 5 (3.0% shingles) had the lowest. The remaining mixes fell sequentially between Mix 1 and Mix 5. Rutting susceptibility was assessed using the French Wheel- Rutting Test at 100, 300, 1,000, 3,000, 10,000, and 30,000 cycles (Figure 1-2). The 20% RAP mixture had the greatest amount of rutting. Adding 3.0% shingles to the 20% RAP mixture sig- nificantly improved the rutting resistance. McGraw (2010) investigated the effect of asphalt binder grade and content, RAP source and content, and shingle type and content on asphalt mixtures. Laboratory-produced mix- tures that incorporated both PC and MW–RAS and RAP—in conjunction with a PG 58-28—were tested for both asphalt binder and mixture properties. Recovered asphalt binder from HMA and RAS was tested for high- and low-temperature properties. Additionally, the moisture susceptibility and ther- mal cracking characteristics of the mixtures were tested on laboratory and field samples and compared to verified field performance. The researchers observed that mixtures contain- ing finer-grind PC–RAS appeared to be more homogenous but demanded slightly more asphalt binder than MW–RAS mixtures. Test results for extracted and performance-graded binders from the mixtures were strongly influenced by the virgin binder content and its high and low PG-critical temperatures. Higher reclaimed binder percentages were associated with higher critical temperatures of the blended binder. Mixture

7 testing also showed a correlation between virgin binder con- tent and dynamic modulus at the high-temperature portion of the master curve. The authors stated that the results provided justification for the current 70% minimum virgin binder cri- terion. Note that the materials in this study met this criterion with 19% recycled materials content. Mixture and binder test- ing indicated that increasing RAP in RAS mixtures increased the total stiffness of the mixture. The use of different RAP sources in the mix design did not have a significant effect on the stiffness of the mixture. The asphalt binder from PC–RAS was stiffer than that from MW–RAS, and its effect was most pronounced at the 5% level at the lower frequencies. The differences in binder stiffness resulted in high-mixture moduli for the PC mixes. Decreas- ing the shingle content to 3% reduced differences between mixes containing MW and PC–RAS. Results also showed that a softer virgin binder could decrease the mixture’s stiffness. The stiffening of the mixture from the PC–RAS was evident in Asphalt Pavement Analyzer (APA) results, as well. The mix- tures that used PC–RAS had the least amount of rutting. A 2011 University of Minnesota study assessed how RAS affected the low-temperature properties of asphalt mixtures using small mix beam samples in the bending beam rheom- eter (Austin 2011). Creep stiffness [Log S(60)] and m-value at 60 s were dependent upon the amount of MW or PC–RAS and the temperature of the test. The interaction of tempera- ture and percentage of RAS type were also considered in the statistical analysis. Two analysis of variance (ANOVA) test- ing groups were defined based on the quantity of RAP in the mixture design. Group 1 contained approximately 15% RAP, while Group 2 contained approximately 25% RAP. The con- clusions of this study were: 1. The creep stiffness of a 15% RAP mixture was increased by adding 5% MW–RAS; however, it did not affect the m-value. Using 3% MW–RAS decreased the creep stiff- ness compared to the control mixture. Therefore, using up to 3% MW shingles will not significantly change the low-temperature properties of the blended binder. 2. When 25% RAP was used, no amount of MW shingles changed the S(60) and m(60) values. 3. The creep stiffness of the bending beam rheometer sam- ples was not changed by the addition of up to 5% PC–RAS. The m-value was not affected for the 15% RAP mixtures when adding PC–RAS; however, there was a change for the 25% RAP mixtures using PC–RAS. These results showed that RAS has a more prominent effect on the blended binder when RAP is not used or when RAP is used in low percentages in the mixture design. However, using more than 20% RAP in a mixture will decrease the over- all impact of the RAS on the asphalt binder low-temperature properties. In 2012, the Florida DOT sponsored a set of test sections on the National Center for Asphalt Technology (NCAT) Test Track to study the performance of two virgin asphalt mix- tures: one containing a polymer-modified binder and the other using a hybrid ground tire rubber (GTR) and polymer- modified binder. Another test section contained 25% RAP, and a fourth section contained 20% RAP and 5% RAS. The RAP mix and the RAP–RAS mix both contained a styrene- butadiene-styrene (SBS)–modified virgin binder (Willis et al. 2016). All four 12.5 mm nominal maximum aggregate size (NMAS) Superpave mixtures were constructed as a 1.5-in. overlay on a very thick existing pavement. A variety of labora- tory rutting and cracking tests were conducted on each mix- ture. Field performance of the test sections was assessed for rutting, cracking, and ride quality. After 2 years and 10 million equivalent single-axle loads (ESALs) of traffic, approximately 75% of the RAP–RAS test section lane area had low-severity cracking. The 25% RAP mixture had approximately 23% lane area cracking, the virgin mix test section with the GTR-hybrid Source: Tighe et al. 2008. 0 5,000 10,000 15,000 20,000 25,000 30,000 Number of Cycles P er ce n ta g e R u t D ep th Figure 1-2. Percentage of rut depth versus number of cycles.

8 binder had 20% cracking, and the virgin mix with SBS-modified binder had approximately 8% of the lane area cracked. The majority of the cracking in the test sections was low severity. The overlay laboratory test matched the field performance best. The Energy Ratio and Semi-Circular Bend (SCB)–Fracture Resistance (Jc) parameter tests did not correctly rank the cracking of the four mixtures. Wang et al. (2014) studied RAS in open-graded friction course mixtures for rutting resistance and moisture suscepti- bility. Open-graded friction course mixtures with PG 64-22 and PG 76-22 asphalt binder were tested with and without RAS for rutting resistance using APA. The PG 64-22 mix- ture without RAS quickly failed, whereas the PG 64-22 mix- ture with RAS had similar rutting results as the PG 76-22 mixture without RAS. The PG 76-22 mixture with RAS had the lowest APA rutting results. Dynamic modulus testing was performed on the PG 76-22 mixture with and without RAS at 4°C, 25°C, and 45°C. The mixture with RAS was consistently stiffer than the mixture without RAS. The mixture without RAS had a tensile strength ratio of 0.81, and the mixture with RAS had a tensile strength ratio of 0.69. But one of the con- ditioned samples with RAS had much higher air voids than the rest of the samples (25% compared to 21%) and a much lower tensile strength. When the outlier was removed, the tensile strength ratio of the mixture with RAS was 0.83, which indicates the addition of RAS did not have a significant impact on the moisture susceptibility. Wu et al. (2016) performed a study on field cores obtained from a Washington State project with four sections: two sections with 15% RAP and two sections with 3% RAS and 15% RAP. The study compared extracted binder testing for rutting, fatigue, and thermal cracking resistance to mixture test results for the same distresses. They found that the inclu- sion of RAS had little to no effect on the PG of the recovered asphalts, but it did contribute to the significantly reduced non recoverable creep compliance (Jnr) and increased percent- age of recovery (R3.2), which corresponds with better rutting resistance. The Monotonic Binder Fatigue Test was used to assess fatigue and thermal cracking resistance at 20°C and 5°C, respectively, and was reported to correlate well with field performance (Wen and Bhusal 2013). The recovered binders containing RAS showed comparable fatigue resistance but also showed lower thermal cracking resistance because of the lower failure strains at 5°C. For mixture testing, the Hamburg wheel-tracking device was used to evaluate rutting resistance. Results showed that the mixtures containing RAS had better rutting resistance. The low-frequency (high temperature) dynamic modulus and the IDT creep compliance values were higher for the mixtures containing RAS, also indicating increased rut resistance. Fatigue and thermal cracking resis- tance were tested using IDT and measuring the fracture work density, vertical deformation, and horizontal fracture strain (Wen 2013). This testing showed no significant difference in the mixtures with or without RAS for both fatigue and thermal cracking resistance. The mixture test results and binder test results were consistent with each other (better rutting resis- tance and no difference in fatigue resistance), except for the thermal cracking resistance. The binder testing indicated a reduction in the thermal cracking resistance, and the mixture testing showed no significant difference in the thermal crack- ing with the addition of RAS. The authors explained this con- flict as the binder behavior being negatively affected by the shingle asphalt, while the mixture thermal cracking resistance was improved by the RAS fibers (Wu et al. 2016). Warm Mix Asphalt WMA is a broad category of technologies used to reduce the mixing and placement temperatures of asphalt for the con- struction of pavements. Although the temperature reduction varies by technology, WMA is generally produced at tempera- tures in the range of 30°F to 100°F lower than typical HMA. However, there is not a standard definition of WMA. Some states define WMA by a maximum production temperature of 275°F. WMA originated in Europe in the late 1990s to achieve target reductions of greenhouse gases in the 1997 Kyoto Treaty on Climate Change. In 2002, representatives from the United States asphalt paving industry traveled to Europe to learn about WMA. In 2005, FHWA and the National Asphalt Pavement Association (NAPA) formed the WMA Technical Working Group to guide proper implementation through data collection and analysis to develop guidelines/specifications. The first documented WMA pavement in the United States was constructed in 2004. In 2007, FHWA, AASHTO, and NCHRP conducted a scan tour to European countries to collect information on WMA technologies that could help its implementation in the United States. The use of WMA has grown faster than any other new asphalt technology in the past several decades. From 2009 to 2015, NAPA conducted surveys to monitor the usage of WMA from four stakeholder groups: (1) state DOTs, (2) asphalt con- tractors, (3) state asphalt paving associations, and (4) WMA technology providers (Hansen and Copeland 2015). The NAPA surveys of WMA usage indicated that in 2009, WMA accounted for 5.4% of all asphalt plant mix. By 2011, WMA was 19% of the mix produced. By 2015, WMA production accounted for 30.4% of all plant mix. This suggests that across the United States, implementation of WMA has continued to grow. The survey also indicates that WMA usage has grown among all segments of owners. Most new pavement technolo- gies are typically implemented first by state agencies, followed some years later by local agencies (i.e., cities and counties) and nongovernment clients (i.e., commercial developers).

9 There are three categories of WMA technologies: foaming, organic additives, and chemical additives. The NAPA surveys used a modified classification as follows: (1) chemical additives, (2) organic additives, (3) foaming additives, and (4) water- injection foaming systems (Hansen and Copeland 2013). The foaming technologies create foamed asphalt using water-injection, damp aggregate, or a hydrophilic material such as zeolite. The water turns to steam, disperses through- out the asphalt binder, and expands the binder, providing a corresponding temporary increase in fluid content similar in effect to increasing the binder content. Some examples of technologies that use water to foam the binder include Aspha-min (zeolite), low-energy asphalt (LEA) (foams from a portion of aggregate fraction), Advera (zeolite), Astec Double Barrel Green (DBG), Terex, Gencor Ultrafoam TX, and Maxam AQUABlack. When the water turns to steam from contact with the hot asphalt, it expands the binder phase and increases the mixture’s workability. The amount of expansion depends on a number of factors, including characteristics of the binder, the amount of water added, and the temperature of the binder (Jenkins 2000, Arega et al. 2013, Newcomb et al. 2015). Chemical additives are typically specially formulated surfactants. These chemical additives include Cecabase RT, Evotherm ET (emulsion technology), Evotherm Dispersed Asphalt Technology (DAT), Evotherm 3G, and Rediset. Organic additives are typically special types of waxes that decrease the binder’s viscosity above the melting point of the wax. Therefore, wax properties are carefully selected based on the planned in-service temperatures. Examples of organic additives include Sasobit and SonneWarmix. In the United States, water-injection foaming systems dominate the WMA market, accounting for about 72% of all WMA in 2015 (Hansen and Copeland 2015). The market share of chemical additive WMA has grown significantly in recent years, increasing from 12.1% in 2011 to 25.2% in 2015. The potential benefits of WMA include: • Reduced plant emissions, including greenhouse gases; • Reduced fuel usage; • Reduced binder oxidation; • Ability to increase haul distances and still have sufficient workability; • Compaction aid; • Ability to incorporate higher percentages of recycled materials; • Ability to pave in cool weather; • Better working conditions for paving crews; and • Increased in-place densities. Despite the potential benefits from the use of WMA, proper construction practices are needed. Although most aspects of designing and constructing WMA are similar to HMA, some have expressed concerns that lower production temperatures associated with WMA could result in differ- ences in pavement performance relative to HMA. However, several studies have proven that pavements built with WMA perform no differently than HMA (West et al. 2014). Asphalt Mixtures Using RAS and a WMA Technology Literature on the combination of WMA and RAS is limited. Laboratory preparation of WMA with a water-injection foaming process—by far the most popular method of WMA production—is expensive and challenging. Consequently, laboratory-produced WMA with water-foaming is rare. There have been numerous field projects with WMA and RAS, some of which are reported in trade magazines. But relatively few studies have collected mix samples for advanced char- acterization, and even fewer have had companion sections without WMA or without RAS for comparison. Very few follow-up inspections of the pavements have been reported to document how they have performed over time. Laboratory Properties of WMA–RAS Mixtures Middleton and Forfylow (2008) evaluated a number of WMA–RAS mixtures that were produced at an asphalt plant in Vancouver, British Columbia, Canada. The WMA technol- ogy for all of these mixtures was foamed asphalt. Mixtures contained virgin HMA, virgin WMA, and three WMA mix- tures with varying amounts of RAP and RAS. Samples of the as-produced mixtures were obtained, and the asphalt binder was extracted and recovered. The recovered PG for each of the mixtures is shown in Table 1-3. The results in Table 1-3 follow the expected trends for the low-temperature grades. The HMA virgin, WMA virgin, and WMA with 15% RAP had the same low-temperature grade. The WMA with 15% RAP + 5% RAS and the WMA with 50% RAP graded as one grade higher on the low end. The higher amount of aged binder in the high RAP and the even harder RAS binder reduced the PG of the binder blend. Mixture PG Virgin HMA 64-22 Virgin WMA 70-22 WMA 15% RAP 70-22 WMA 15% RAP + 5% RAS 76-16 WMA 50% RAP 70-16 Source: Middleton and Forfylow 2008. Table 1-3. Recovered binder properties.

10 The high-temperature grade of the mix containing RAS was the highest. Surprisingly, the recovered binder from the mix con- taining 50% RAP was the same grade as the mix with 15% RAP, and the high PG of the WMA virgin was higher than for the HMA virgin. APA test results for the WMA mixtures are shown in Table 1-4. All of the results were 8.0 mm or below, which is considered satisfactory. Moisture susceptibility test results are shown in Table 1-5. The only mixture that failed to meet the minimum 80% retained strength criteria was the virgin WMA mixture, which had a retained strength of 77.5%. This virgin WMA mixture would be expected to have the binder with lowest stiffness, since it has no RAP or RAS. It is normal for mixes with stiffer asphalt binders to have better retained strengths. Mogawer et al. (2011) conducted Hamburg Wheel-Tracking tests, Overlay tests, and the Asphalt Concrete Cracking Device (ACCD) Test on several mixtures, including some with RAS and WMA technology. Results are summarized in Table 1-6. The Overlay Test results for the mix with 5% RAS were about 70% lower than the control mix. Adding WMA did not have a substantial effect on the control mix or the mix with 5% RAS. The mix with 35% RAP + 5% RAS had lower cycles to failure than the mix with just 5% RAS. Adding WMA to the 35% RAP + 5% RAS mix improved the results but not by a statistically significant amount, given the typical variability of this test. The ACCD Test uses thermal contraction of notched ring- shaped specimens compacted around an Invar ring to deter- mine the temperature at which the specimen cracks. The results shown in Table 1-6 indicate that the cracking tem- peratures for the eight mixes were similar. The highest failure temperature was −34.6°F for the 40% RAP mixture and the 35% RAP + 5% RAS mixture. These additions of WMA improved (lowered) the cracking temperature in each case compared to its HMA pair. Moisture susceptibility was evaluated using the Hamburg Wheel-Tracking Test. The control HMA mix, the control WMA mix, and the WMA mix with 5% RAS were the only mixes with stripping inflection points. The results in Table 1-6 are consistent with the results by Middleton and Forfylow (2008) in that the mixtures with the stiffer binders perform best in moisture-susceptibility testing. The recovered asphalt binders from the mixes were graded to determine their continuous grade and PG. The results are shown in Table 1-7. The addition of 5% RAS increased the high and low continuous-grade temperatures slightly over the control. The addition of WMA lowered the high and low continuous-grade temperatures in each case, including the mixtures with RAS. Cascione et al. (2015) reported on the field and labora- tory performance of RAS mixtures from seven state agencies. The mixtures were designed to evaluate multiple properties of RAS, as well as to evaluate the use of RAS with GTR and WMA technologies. Flow number and dynamic modulus testing of the mixtures showed that the addition of RAS or RAS and RAP improved rutting resistance. The improvement was confirmed by the field projects, as the researchers did not observe any measurable amount of wheelpath deformation over multiple years of evaluation. The Four-Point Bending Beam Test was used to analyze the fatigue resistance of the mixtures, and the results showed that the mixtures with RAS had similar results to the mixtures without RAS. The out- come was also consistent with the field trials, since most of the mixtures with RAS performed as well as or better than the mixtures without RAS with regard to fatigue cracking. Low-temperature cracking was evaluated using the low- temperature Semi-Circular Bend Test. For most projects, there were no significant differences in the fracture energies of mixtures with or without RAS. Results for one state showed a significant increase in the low-temperature cracking resis- tance with the addition of RAS, but results from another state had a significant decrease in the low-temperature cracking resistance with the addition of 15% RAP to a 5% RAS mixture. APA Rut Depth After 8,000 Cycles (mm) Mix Air Voids (%) Dry Rutting Wet Rutting Virgin 7.1 4.8 8.0 15% RAP 7.3 5.2 5.2 15% RAP + 5% RAS 7.0 4.1 7.1 50% RAP 7.0 4.1 5.6 Source: Middleton and Forfylow 2008. Table 1-4. APA rutting results of WMA mixtures. Mix Air Voids (%) Unconditioned Strength (psi) Conditioned Strength (psi) Retained Strength (%) Virgin 7.2 806.7 625.2 77.5 15% RAP 6.5 878.1 769.5 87.9 15% RAP + 5% RAS 6.8 985.1 818.6 83.1 50% RAP 7.2 1166.2 1124.7 96.4 Source: Middleton and Forfylow 2008. Table 1-5. Moisture susceptibility of WMA mixtures.

11 Mixture Continuous Grade (°C) PG Control 62.2–31.2 58-28 40% RAP 72.4–27.9 70-22 5% RAS 65.6–32.2 64-28 35% RAP + 5% RAS 77.5–25.9 76-22 Control + 1% WMA 56.4–32.6 52-28 40% RAP + 1% WMA 64.2–30.9 64-28 5% RAS + 1% WMA 60.9–32.7 58-28 35% RAP + 5% RAS + 1% WMA 71.1–27.9 70-22 Source: Mogawer et al. 2011. Table 1-7. Extracted binder grading results. Mixture Average Overlay Tester (Cycles to Failure) ACCD Cracking Temperature [°C (°F)] Hamburg Wheel-Tracking Device Stripping Inflection Point Control 1,004 -38.5 (-37.3) 16,800 40% RAP 3 -37.0 (-34.6) None 5% RAS 308 -38.8 (-37.8) None 35% RAP + 5% RAS 22 -37.0 (-34.6) None Control + 1% WMA 936 -39.3 (-38.7) 6,200 40% RAP + 1% WMA 143 -39.8 (-39.6) None 5% RAS + 1% WMA 297 -40.5 (-40.9) 9,800 35% RAP + 5% RAS + 1% WMA 63 -39.3 (-38.7) None Source: Mogawer et al. 2011. Table 1-6. Performance test results. Evaluation of the field trials in Missouri found that a mix with coarse-ground RAS exhibited more transverse cracking than a similar mix with fine-ground RAS after two winters. The mixtures containing RAS were observed to have slightly more cracking than the mixtures without RAS in Missouri and Colorado one winter after construction. In contrast, mix- tures with RAS were observed to have similar or less cracking than the mixtures without RAS in Iowa and Indiana after two and three winters, respectively. In Minnesota, a mixture with MW–RAS displayed slightly more cracking than a mix with PC–RAS four winters after construction. The authors caution against drawing conclusions from results of the field evalua- tions, since the extent of cracking prior to being overlaid was highly variable for the field projects. Summary of Performance of RAS Mixture Experimental Sections This section summarizes field projects with mixtures con- taining RAS that have been documented to date in literature. Minnesota has conducted numerous field trials investigat- ing the use of RAS in asphalt mixtures. Minnesota’s first test section containing RAS was completed on a recreational trail in Saint Paul in 1990. The subbase was an old railroad track bed covered by 4 in. of crushed concrete base. A 2.5-in.-thick wearing course containing 6% and 9% MW shingles was placed 12-ft wide. In 2003, after 13 years in service, the mix- tures were still performing well (Shively 2011). In 1991, Minnesota DOT completed a trial section in Mayer containing RAS. In 1995 and in 2003, the mixture perfor- mance of the RAS asphalt mixture was equivalent to the con- trol mixture. Transverse reflective cracking had been noticed in both sections; however, no other distresses were noticed (Shively 2011). The New Jersey DOT also conducted early experiments with an asphalt cold-patch material using RAS. After 22 months in service, only minor signs of distress were noted. The conven- tional patch for New Jersey DOT only lasted approximately 6 months; therefore, the use of RAS more than tripled the life expectancy of the patch (Schroeder 1994). In a 1994 survey, only three state DOTs responded to using RAS in asphalt mixtures. Illinois evaluated the use of shingles in asphalt paving mixtures and determined that roofing shin- gles could be used in both dense-graded and stone matrix asphalt mixtures. In 1993, a Minnesota DOT pavement con- taining 5% to 7% shingles (by weight) reported good per- formance after 2 years (U.S. Army Corps of Engineers 1999). Canada Highway 86 near Waterloo, Ontario, was expanded from a two-lane road to a four-lane highway in 1996. The lower binder layer was a 1.5-in. layer containing no RAS; however, the 2-in. upper binder layer and the 1.5-in. wearing course contained 3% RAS. A control mixture was placed along with the RAS mixtures for comparison. Three years after construc- tion, the pavement with the control mixture had more ravel- ing, longitudinal joint openings, and fatigue cracking than the pavement using shingle mixtures (Shively 2011). In 1997, Texas DOT constructed test sections using both PC and MW–RAS in asphalt surface mixtures. In addition, a control section was also constructed. The mix designs for test sections containing roofing shingles (MW and PC) were performed according to Texas DOT Standard Specification

12 Item 340. The control section mix design was based on the Texas DOT Special Specification Item 3000 for [Quality Control/Quality Assurance] QC/QA mixes. The asphalt concrete mix was also tested for Hveem stability, moisture susceptibility, static creep, and voids in VMA. In addition, the boil test (Tex-530-C) was completed to determine the stripping susceptibility of the mix (Rana 2004). After 2 years, the performance of the test sections containing roofing shin- gles was comparable to the control section. The area engineer noted that the RAS mixtures had some reflective cracking, but the time at which the cracks developed did not appear to be any different than conventional asphalt mixtures. Minnesota DOT completed five field projects between 2005 and 2008 that used both MW and PC–RAS. In each of the five projects, 500-ft performance sections were set up to monitor cracking, rutting, and surface characteristics (McGraw 2010). The study was designed to assess the new binder-to-total binder ratio criteria (70% minimum) that Minnesota DOT was considering as a specification. The research indicated that the 70% new binder ratio criteria worked for some projects and not for others. The projects also seemed to confirm that using a softer binder grade could improve the cracking performance of the mixtures. Unlike previous laboratory testing, this research also suggested that little difference was noticed in performance between sections with PC and MW–RAS. In 2009, a field project conducted by the Washington State DOT and King County Department of Transportation (KCDOT) was designed to assess the viability of using RAS and RAP in asphalt mixtures. Two miles of roadway were divided into half-mile test sections containing two over- lay asphalt mixtures: (1) 15% RAP–HMA and (2) 3% RAS and 15% RAP–HMA (Caulfield 2010). The RAS was tested for gradation, deleterious materials, moisture content, and asbestos before it was used to ensure that a high-quality prod- uct could be constructed that would meet the current stan- dards of the state. Wu et al. (2016) cut cores from the project after 3 years and conducted a variety of tests on the cores and recovered binders. Laboratory test results and field perfor- mance indicated that the RAS had no negative impacts on the pavement’s performance. Examples of Projects That Used RAS and WMA Although there are not a lot of data available for laboratory mix properties and in-place performance, there have been a number of field projects with mixes containing both RAS and WMA. Example projects constructed using WMA–RAS are described below. A section of Interstate 30 in Dallas, Texas, was scheduled for a mill and overlay project. A stone matrix asphalt with PG 76-22 and fiber was specified for the project. At the con- tractor’s request, the Texas DOT allowed the placement of a WMA–RAS section in the stone matrix asphalt. If its observed performance through the winter was good, the Texas DOT would agree to allow more of the WMA–RAS mix to be used on the project. The stone matrix asphalt mixture contained 5% RAS and Advera as the warm-mix additive. Advera— added at a rate of 20 to 40 pounds per ton of asphalt mix— was mixed with the RAS stockpile to facilitate handling. The mixture was placed at 315°F, which is considerably higher than many WMA projects. But the higher temperature was needed to provide good workability (Kopp 2012). The stone matrix asphalt trial was constructed in July 2011, and the ini- tial performance was evaluated in February 2012. At the time of inspection, the RAS section had remained darker than the other sections and was performing well. Because of the good performance during the first winter, the Texas DOT allowed the contractor to continue to use the WMA–RAS mixture. In October 2009, the Virginia Transportation Research Council worked with a contractor to place an SM-12.5 sur- face mix containing 5% RAS on the shoulder of Route 522 south of Winchester (Maupin 2010). An HMA section was placed at approximately 300°F, and a WMA section (water- injection foam) was placed at approximately 270°F. Both mixes were produced in the same double-drum plant. The shingles were blended with fine aggregate before being added into the asphalt plant at the RAP collar. No significant prob- lems were experienced during the production or placement of the mix containing RAS and WMA. The Asphalt Paving Association of Iowa (2010) held an open house to demonstrate construction of an asphalt mixture con- taining WMA and RAS. The project was constructed in 2010 on U.S. Hwy 61 between Muscatine and Blue Grass. The proj- ect consisted of approximately 21,000 tons of asphalt mixture. Some of the work included 20% RAP; additional work con- sisted of 5% RAS and 13% RAP; and finally, 7% RAS with 6% RAP. The WMA technology used was Evotherm 3G. Although the constructed pavements looked good, no follow-up perfor- mance evaluations have been reported. N. B. West Contracting constructed a 9-mi section of pave- ment on Route 72 south of Rolla, Missouri, between March and April 2011 (Lender 2012). This project consisted of WMA, RAP, and RAS. The percentages of RAP and RAS were not reported but it was estimated that 1.8% of the binder came from recycled materials, including 0.6% of the binder from RAS yielding greater than 30% total binder replacement. The WMA technology was Evotherm 3G. The mix was produced at a temperature of approximately 265°F. The project size was more than 25,000 tons with a haul distance of more than 1 h. Remixing with a transfer vehicle on the job site was thought to be essential for this project. The job was inspected in 2012, and it still provided a smooth ride.

13 The 2012 cycle of the NCAT test track included the Green Group experiment. One of the four test sections in the Green Group experiment included mixtures containing RAS in the surface and intermediate layers. All of the mixes were pro- duced at warm-mix temperatures using either a foaming process or a WMA additive. The top lift was a stone matrix asphalt containing 5% RAS and using foamed asphalt WMA. The intermediate lift used a dense-graded mix produced with 25% RAP and 5% RAS using a chemical additive WMA tech- nology. The bottom lift was produced with 25% RAP and foamed asphalt WMA. This section carried over 2 million ESALs before the first crack appeared and approximately 3.5 million ESALs when it reached the cracking distress threshold (20% of the total lane area). A forensic investiga- tion at the end of 2013 indicated that this section failed as a result of a bond failure between the base and intermediate lift (Vrtis et al. 2015). From 2013 to 2015, FHWA’s Turner–Fairbank Highway Research Center conducted a laboratory and field fatigue cracking study using their Accelerated Loading Facility (ALF). The objective of the experiment was to establish realistic limits for high-RAP and RAS mixtures using WMA technolo- gies and softer virgin binder grades (Li and Gibson 2016). Ten mixtures were produced and constructed in test lanes for the ALF. The primary mixture variables included RAP binder ratio (0, 0.20, and 0.40), RAS binder ratio (0 and 0.20), grade of virgin binder (PG 58-28 and PG 64-22), and WMA tech- nologies (foamed asphalt by water injection and the chemical additive Evotherm). All mixes were 12.5-mm NMAS gra- dations designed using the Superpave mix design method with 65 gyrations. For the mix designs containing RAP or RAS, the recycled binders were assumed to be 100% effec- tive. The fatigue cracking in the test lanes was mapped after each 25,000-load application by the ALF. The two test lanes containing RAS and the mixture containing the higher RAP content with a PG 64-22 binder had the shortest fatigue life under the ALF loading. Variations on thicknesses of the test lanes and estimated base moduli that confounded the fatigue results were reported by West et al. (2017). However, Li and Gibson (2016) concluded that a softer virgin binder improved the fatigue life for 40% RAP binder ratio mixes. However, a softer binder was ineffective at improving the fatigue perfor- mance of the mixture containing about 5% PC–RAS (20% RAS binder ratio). The use of WMA slightly improved the fatigue life of 20% RAP mixes, but it was less effective in the 40% RAP mixes.