Use of Reclaimed Asphalt Pavement and Recycled Asphalt Shingles in Asphalt Mixtures (2016)

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.

7 materials or the properties of the total asphalt mixture after removing the asphalt from the particulates. Asphalt Content Either the ignition oven (AASHTO T308) or solvent extrac- tion (AASHTO T164) test methods can be used to measure the asphalt content of recycled materials. The ignition oven burns the asphalt off of the aggregate at high temperatures and the percentage of mass loss is measured as the asphalt content. In some cases, a correction factor may be needed for nonasphalt components that burn off along with the asphalt (e.g., some limestones and shingle backing materials). These factors can be established by calculating the difference in the asphalt content between solvent extraction and ignition oven results. Alternatively, historical laboratory results can be used to estimate aggregate correction factors. When testing RAS, AASHTO PP78-14 recommends using 400 grams of RAS so that the ignition oven ventilation system is not overloaded (i.e., clogged). If necessary, the RAS sample can be split and run and tested in two parts. One report noted the ignition oven RAS asphalt content was higher than obtained with solvent extraction (Roque et al. 2015). Centrifuge or reflux solvent extraction methods to deter- mine the asphalt content use one of several solvents [trichloro- ethylene (TCE), n-propyl bromide (nPB), toluene, methylene chloride, or a toluene and ethanol blend]. When the asphalt does not have to be recovered from the solvent for asphalt testing, a vacuum extraction method or simply soaking the recycled materials in solvent can be used to estimate the amount of asphalt in recycled material. Alternatively, an organic solvent such as Bioact™ can be used with all of the solvent extraction methods when the asphalt does not need to be recovered. Any solvent extraction method can have difficulties with removing all of the asphalt from both porous (absorptive) virgin aggregates and from RAP because of the strong bonds of the harder asphalt with the aggre- gate surface. The hard RAS asphalt can be difficult to dis- solve and remove from the other shingle materials with solvents (NCAT 2012). In general, the asphalt content determined with the ignition oven method is slightly higher than determined using solvent extraction methods (Michael 2011). This is attributed to a small percentage of the asphalt being strongly bound to the aggregate, which is not removed by the solvent. Since 2009, the National Asphalt Pavement Association (NAPA) has tracked the use of RAP and RAS in the United States through annual industry surveys, and has determined that the use of both has increased in the United States (Hansen and Copeland 2013). In 2009, contractors in 23 states reported pro- ducing less than 15% of their total amount of asphalt mixtures with RAP (Figure 2). By 2012, contractors in only 12 states used RAP in less than 15% of their total tonnage (i.e., more states using at least 15% RAP). These changes represent an increase in the total tonnage of asphalt mixtures with RAP by 22% from 2009 to 2012 (from 56 to 68.3 million tons). RAP is used in all states and is typically available throughout each state, although the majority of the RAP stockpiles are usually concentrated along major highways (transportation logistics) and near urban areas (more miles of roadways) (Figure 3). RAS was used in almost 1.9 million tons of asphalt mix- tures in 2012. As of 2012, contractors in 17 states reported using RAS in all four of the annual NAPA surveys (Figure 4). However, contractors in 10 states failed to report using RAS in any of the NAPA annual surveys. Contractors in four states (Florida, Georgia, West Virginia, and Massachusetts) reported using RAS prior to 2012, but did not report any usage in 2012 (i.e., fewer states using RAS). Information about RAP and RAS topics that influence the use of these materials in asphalt mixtures are presented in this chapter. This information is organized into the following topics: • Recycled material properties • Asphalt mix designs with recycled materials • Mixture testing • Asphalt plant practices and equipment • Pavement performance • Economics • Research in progress. RECYCLED MATERIAL PROPERTIES The recycled material asphalt content, asphalt properties (after extraction and recovery), aggregate gradation, and aggregate specific gravity are most often determined by testing. Aggregate consensus properties (i.e., various particle shape characteris- tics) and source properties (toughness, durability, clay-sized particulates, and polish value) are only occasionally deter- mined, if at all, at this time. Any requirements are agency- specific and can require the testing of individual recycled chapter two LITERATURE REVIEW

8 Advantages associated with using the ignition oven method for determining the asphalt content are that test results can be obtained quickly for QC testing and aggregate properties can be determined after the asphalt is removed. Disadvan- tages include the need for correction factors to account for the mass loss of materials other than asphalt, which may be burned off during testing. Advantages for using solvent extraction are that the asphalt can be recovered for testing and the aggregate properties can be determined after the asphalt is removed. The disadvantages are the length of test time, the need to use solvents that are costly to purchase and to dispose of after testing, and worker safety concerns. Measuring Asphalt Content—Section Summary • The ignition oven method is used more frequently to determine the asphalt content; however, correction fac- tors may be necessary to account for any aggregate min- eralogy or other nonasphalt material that also burns off. Adjustments to the oven temperatures and sample size for testing RAS may be necessary. • Solvent extractions are used to remove the asphalt from recycled materials when the recycled material asphalt is to be recovered for testing. However, fewer agencies use solvent extraction methods because of the difficul- ties with obtaining and disposing of the solvents (i.e., safety and environmental hazards). RECYCLED MATERIAL ASPHALT PROPERTIES Both the centrifuge or reflux solvent extraction methods gener- ate a solution of solvent and asphalt from which the asphalt can be recovered using either the Abson (AASHTO T170) or FIGURE 2 Use of RAP in the United States in 2009 and 2012 as reported by contractors (Source: Hansen and Copeland 2013). FIGURE 3 Example of locations of RAP stockpiles along transportation routes and around urban areas in Virginia (Source: Hoppe et al. 2015). FIGURE 4 Use of RAS in the United States as reported by contractors (Source: Hansen and Copeland 2013).

9 Rotavapor (ASTM D5404) recovery methods. A range of sol- vents (TCE, nPB, and methylene chloride) can be used with either method. Zhou et al. (2013) compared different extraction and recovery methods for RAS asphalt for testing and found that neither the choice of extraction or recovery method influ- enced RAS asphalt properties. An alternative method for extraction and recovery is detailed in the AASHTO T319 Standard Method of Test for Quantitative Extraction and Recovery of Asphalt Binders from Asphalt Mixtures. This test method uses a combined solvent extraction–Rotavapor recovery process. However, significant difficulties with extracting and recovering when using the AASHTO T319 method were noted by Scholz (2010) when testing RAS samples that included: • RAS asphalt clogged screens and the outlet of the extrac- tion vessel. • Material was described by lab staff as very thick and viscous. • Removing tear-off shingle asphalt with solvent extraction was difficult. • Recovering sufficient RAS asphalt for low temperature binder testing was difficult. The recovered asphalt is used to determine the upper and lower critical PG temperatures using: • A rotational viscometer (Brookfield) at high temperatures; • A dynamic shear rheometer (DSR) at high, intermediate, and low temperatures; and • A bending beam rheometer (BBR) at low temperatures. DSR testing is used to determine the critical high PG temperature and evaluate properties at intermediate and low in-service temperatures. Test results for both the virgin and recycled material asphalt are used to estimate changes in the upper critical PG temperature based on the percentage of each asphalt in the anticipated total asphalt blend. Roque et al. (2015) used DSR testing to evaluate recycled RAP and virgin asphalt and found the shear modulus (G*) as well as the G*/sin d parameter increased with the increasing percentage of RAP at both the high and intermediate test temperatures. Differ- ent RAP sources had different shear moduli and other DSR parameters. Maupin et al. (2008) found that RAP asphalt changed the high and low temperature asphalt grading, with the low temperature grading changing from a PG xx-22 to a PG xx-16 for the recovered asphalts from RAP mixtures and increased the high PG temperature by one to two grades. Scholz (2010) reported difficulties in determining the low temperature DSR properties for RAS asphalt because the stiffness of the asphalt exceeded the DSR equipment limita- tions. Similar difficulties were reported by NCAT (2012) and Zhou et al. (2013). BBR results use measurements of the asphalt stiffness, s, and a rate of change in stiffness with time parameter, m-value, at low temperatures to determine the critical low PG tem- perature. Roque et al. (2015) found the low temperature BBR stiffness increased with increasing RAP, m-value decreased with the increasing percentage of RAP, and the magnitude of the changes were dependent on the RAP source. Several other researchers reported that the critical low PG temperature increased with increased recycled material asphalt (Maupin et al. 2008; Schroer 2009; McGraw 2010; Booshehrian et al. 2013; Scholz 2010; Zhou et al, 2013). Scholz (2010) used the correlation between changes in the critical upper PG and changes in the critical lower PG temper- atures to estimate the critical low temperature for RAS asphalt. Scholz found this approach useful because DSR equipment limitations precluded testing the stiff RAS asphalt at low test temperatures. Limited information about changes in the viscosities of blends of virgin and recycled asphalts were found in the lit- erature, most likely the result of the large sample size that has to be extracted and recovered for this test. Roque et al. (2015) showed that rotational viscosity increased with a com- bination of RAP and polymer-modified virgin asphalt, but the magnitude of the changes was dependent on the RAP source. When between 20% and 40% RAP asphalt was blended with crumb rubber modified asphalt the stiffness of the crumb rub- ber asphalt masked the impact of one source of RAP asphalt (little change). This concept of a linear relationship between changes in the upper and lower critical temperatures was used in this synthesis to compare DSR data from multiple studies. Data generated from six different research projects with different types, percentages, and combinations of recycled materials, as well as different virgin asphalt grades and various rejuvena- tors were used to develop a regression equation (Figure 5) that FIGURE 5 Correlation of changes in upper critical PG temperatures to changes in the lower critical PG temperatures.

10 shows that Scholz’ conclusion of a generally linear relation- ship between changes in critical upper and lower PG tempera- tures can be replicated with data from other researchers. The upper critical temperature changes almost twice as much as the lower critical temperature. Two additional test methods, the multiple stress creep recovery and the binder fracture energy tests, were used by various researchers to evaluate the impact of recycled material asphalt on performance-related asphalt properties. The mul- tiple stress creep recovery test (AASHTO TP70) uses the data to calculate the nonrecoverable creep compliance and the aver- age percent recovery information that is used to indicate asphalt-related rutting characteristics. Roque et al. (2015) found that the nonrecovery creep compliance decreased with the increasing percentage of RAP (i.e., rut resistance increased). The results were dependent on the RAP source and the percent recovery parameter was more dependent on the type of polymer modifier in the asphalt than on the percentage of RAP. Other researchers found the nonreco- verable creep strain, Jnr, also had a good correlation with rut resistance (Anderson and Bukowski 2012; Booshehrian et al. 2013), with lower Jnr values indicating improved rutting resistance. The higher percentage strain recovery, e, values also indicate increased resistance to rutting. The binder fracture energy test was developed to predict the cracking potential of the asphalt at intermediate temperatures (Roque et al. 2015). The geometry is designed to focus the failure location at the center of the specimen (Figure 6) and the area under the stress and strain plot is used to calculate the fracture energy density, which is the area under the stress strain curve up to stress peak. The fracture energy density decreased with increasing RAP and was sensitive to the different RAP (i.e., cracking potential increased) sources used in the study. New Approach to Binder Modification with RAS Recent research explored a different approach to incorporat- ing RAS asphalt into virgin asphalt (Salari 2012). This study adapted a concept for using finely ground tire rubber as an asphalt modifier, which is referred to as the wet process, for incorporating ultra-finely ground RAS. Grinding of the RAS was accomplished using a Pulva-Sizer with a rotor assembly and hammer mill, and operated at 9,600 rpm. A Coulter Par- ticle Size Analyzer, operated in wet mode, showed that the mean RAS particle sizes were 85.5 µm for tear-off RAS and 201.9 µm for manufacturer waste RAS. Asphalt and RAS blends were prepared using a mechanical shear mixer operated at 1,500 rpm for 30 minutes using 0%, 10%, 20%, 30%, and 40% RAS. An HP-GPC analysis was used to evaluate changes in the high and low molecular weight components in the blended asphalts. Results showed the high molecular weight content (3,000 or greater) increased slightly for the RAS blends and there was more of a shift toward the higher molecular weights when blending the RAS with the softer PG 52-28 asphalt, which indicates an increased potential to crack at warmer critical low PG temperatures. Confocal laser scanning microscopy used to detect the for mation of wax crystals can be responsible for asphalt hardening at low temperatures (i.e., increase critical low PG temperature). The results showed that waxy crystals in the virgin asphalt were not evident once RAS was added. The microscopy also showed that ground minerals were uniformly dispersed in the asphalt. Brookfield rotational viscosity testing of the finely ground RAS increased the viscosities from 3% to 130% over a range of temperatures (95°C to 135°C). The viscosity increase was proportional with the increasing percentage of RAS and the viscosities were higher, as expected, when the blends were produced with the tear-off shingles. The Brookfield viscosity measurements were designed to estimate the vis- cosity temperature susceptibility (VTS) using the following equation: VTS n n T T ( ) ( ) ( ) ( ) ( ) = − + − + log log log log log 273.15 log 273.15 1 2 2 1 Where: n1 and T1 = viscosity in Pa.s at T1 = 95°C n2 and T2 = viscosity in Pa.s at T2 = 135°C. Increases in VTS indicate increases in the temperature susceptibility of the asphalt. In general, VTS changes were small and the trends showed that the value of VTS decreased with the increasing percentage of RAS. Superpave asphalt testing revealed that the RAS increased both the critical upper and lower PG temperatures. DSR fre- quency sweeps showed that the only significant differences in the shear modulus and phase angle were seen at 5°C. Blends of virgin asphalt and 10% RAS were less stiff and more elastic (higher phase angle) than the 20% RAS blend. The thixotropy FIGURE 6 Binder fracture energy test (Source: Roque et al. 2015).

11 (i.e., non-Newtonian behavior) of the blends increased with the percentage of RAS at the intermediate temperatures, but not at the upper or lower temperatures. The shear stress increased with increasing RAS and the samples tended to fail during testing at 6°C. Separation of the RAS and particulate materials during storage was evaluated using the “cigar tube” test (ASTM D7173-05). At 20% or lower percentages of RAS, some blends showed evidence of separation and high levels of separation at the 40% RAS level, possibly the result of the mineral fillers settling to the bottom over 48 hours and sug- gested that a digestion tank with an agitator and heater can be implemented if the wet process is used to produce these types of blends in the field. Recycled Material Asphalt Properties— Section Summary • DSR shear modulus and rotational viscosities increase with increasing recycled asphalt content. Changes in the critical PG temperatures can be dependent on RAP sources. • Recycled materials appear to have more influence on the upper and intermediate critical asphalt temperatures than on the low critical temperatures. Upper critical tem- peratures increase about twice as quickly as the lower critical temperatures. RECYCLED MATERIAL AGGREGATE PROPERTIES Gradations are typically evaluated after the asphalt is removed from the recycled materials using either the ignition oven or solvent extraction methods. Roque et al. (2015) found that gradation analysis of aggregates after using the ignition oven were finer than for the same aggregates recovered from solvent extraction. Consensus properties (i.e., particle shape characteris- tics), other than gradation, are less frequently determined for recycled material particulates. Particle shape characteris- tics include the measurement of flat and elongated particles (ASTM D4791), percent fractured faces of coarse aggre- gates (AASHTO T61 and ASTM D5811), and the fine aggre- gate angularity (AASHTO TP56). Fine aggregate angularity (AASHTO TP56) is only necessary when there is more than about 30% fine RAP aggregate (Newcomb et al. 2007). Source aggregate properties are rarely determined for RAP aggregate because the testing was used to accept the original aggregate source when the asphalt mixtures were originally produced. Source properties include sand equiv- alent (AASHTO T76 and ASTM D2419), organic impuri- ties (AASHTO T21 and ASTM C40), clay lumps and friable particles (AASHTO T112 and ASTM C142), toughness with the Los Angeles abrasion test (AASHTO T96), and soundness (AASHTO T104 and ASTM C88). If toughness is to be eval- uated, the micro-Deval method (AASHTO T58) can be used (Newcomb et al. 2007; Copeland 2011). The sand equivalent may be waived because of changes in aggregate properties after either ignition oven or solvent extraction methods can influence the test results. Aggregate Bulk Specific Gravity The bulk specific gravity (Gsb) of the virgin and recycled material aggregate is required to calculate the voids in min- eral aggregate (VMA), a key mix design volumetric property. Aggregate-specific gravity can be measured for the aggregate remaining after either the ignition oven or solvent extrac- tion. However, measured bulk specific gravities tend to be higher for aggregates obtained from the ignition oven than from solvent extraction (Table 1). Alternatively, the recycled material theoretical maximum specific gravity, Gmm, can be measured and the value is used to estimate the effective specific gravity, Gse: G P G P G se b mm b b( ) ( ) ( ) = − − 100 100 The recycled material asphalt content, Pb, is obtained using either the ignition oven or solvent extraction method and the recycled material asphalt-specific gravity, Gb, can be obtained from historical records or assumed based on TABLE 1 EXAMPLE OF THE PERCENTAGE OF ASPHALT CONTENT OBTAINED BY IGNITION OVEN AND SOLVENT EXTRACTION* AS WELL AS THE BULK SPECIFIC GRAVITIES OF THE AGGREGATES AFTER THE ASPHALT IS REMOVED Property RAP 1 RAP 2 RAP 3 RAP 4 RAP 5 RAP 6 Pb After Ignition Oven 5.43 5.04 5.81 6.27 5.3 5.62 After Solvent Extraction 5.64 4.98 5.11 5.28 4.69 5.18 Gsb After Ignition Oven 2.765 2.689 2.682 2.525 2.632 2.643 After Solvent Extraction 2.719 2.647 2.650 2.481 2.610 2.573 Source: Michael (2011). *Centrifuge extraction with TCE solvent was used.

12 experience. The calculated effective specific gravity is then used to calculate the bulk specific gravity of the aggregate, Gsb, using: G GG P G sb estimated se se ba b( )( )( )= +100 1, The percentage of asphalt absorbed, Pba, by the recycled aggregate is generally assumed based on typical values for local aggregate sources or previous experience when using RAP. The asphalt absorption by the RAS particles is consid- ered to be negligible (AASHTO PP78-14). Adjustments to the theoretical maximum specific gravity test (AASHTO T209, ASTM D2041) may be necessary to keep the RAS particles from floating on top of the water during testing. Misting alcohol onto the surface helps reduce the surface tension and allows the RAS particles to settle (AASHTO PP78-14). It is important that the effective specific gravity not be used as a direct replacement for the bulk specific gravity value of the RAP; however, because of the low absorption of the RAS aggregate, Gse, the RAS effective specific grav- ity can be used until a better method is available (AASHTO PP78-14). Recycled Material Aggregate Properties— Section Summary • Gradations of recycled material aggregates are deter- mined after either ignition oven or solvent extraction to remove the asphalt. The ignition oven may damage the aggregate and gradations tend to be finer than after solvent extraction. • Aggregate specific gravity measured after ignition oven testing is typically higher than values obtained after sol- vent extraction. • Aggregate specific gravities can be calculated by measur- ing the theoretical maximum specific gravity, calculating the effective aggregate specific gravity, and, finally, the bulk specific gravity of the aggregate. • Using the effective specific gravity of the recycled material aggregates as a direct replacement for the bulk specific gravity is not recommended for RAP, but can be acceptable for RAS because of the negligible absorption of asphalt by the RAS particles. • Consensus and source aggregate properties are not typi- cally measured for individual recycled material aggre- gates at this time, although these properties may need to be determined when the percentage of recycled material increases. ASPHALT MIX DESIGNS WITH RECYCLED MATERIALS Total Asphalt Content The total asphalt content (TAC) in the asphalt mixture is a function of the virgin asphalt and the available asphalt from the recycled materials. There are three approaches that can be used to establish the asphalt content available from the recycled materials. The first is to assume the entire asphalt content in the recycled material contributes to the total asphalt content. The second approach is to con- sider that none of the recycled asphalt is useful (i.e., “black rock”). The third approach acknowledges that the reality is somewhere in between, but that the actual percentage is difficult to determine. Regardless of which approach is used, the general equation for calculating the total asphalt content of the asphalt mixture is: F F ( ) ( ) ( ) ( )= + +       TAC RAP AC RAP% RAS AC RAS% Virgin AC% RAP RAS Where: FRAP, FRAS = Asphalt availability factors for RAP and/or RAS asphalt content; RAP AC = Asphalt content of RAP, decimal form; RAP% = Percentage of RAP in mixture, %; RAS AC = Asphalt content of RAS, decimal form; and RAS% = Percentage of RAS in mixture, %. When 100% of the recycled material asphalt is consid- ered to contribute to the total asphalt content the asphalt availability factors, FRAP and FRAS, are 1. If none of the recycled material asphalt is useful, then the asphalt avail- ability factors are 0. AASHTO PP78-14 considers that only a portion of the RAS asphalt is available and recommends using a RAS asphalt availability factor between 0.70 and 0.85. This same standard assumes 100% of the RAP asphalt contributes to the total asphalt content by using a value of 1 for FRAP. The availability factors for both RAS and RAP can vary depending on each agency’s experiences. For example, Georgia uses an asphalt availability factor of 0.75 for RAP (Hines 2015). A Louisiana laboratory study used a volumetric method to estimate the RAS asphalt availability factor for 12.5-mm mixtures with either 5% manufacturer waste RAS or 5% tear-off RAS, and a stone matrix asphalt (SMA) mixture with 5% tear-off RAS and 3% hydrated lime that was used to meet the passing 0.075-mm sieve size SMA requirement (Cooper et al. 2014). The asphalt availability factor measured using this approach ranged from 35% to 50% (0.35 and 0.50 in decimal form).

13 Other information related to asphalt availability factors found in the literature included: • Virgin asphalt content can be reduced by approximately 0.2% for every 1% by weight of RAS (manufacturer’s waste) used in a mixture (Mallick and Mogawer 2000). • Five percent (5%) of RAS in the asphalt mixture con- tributes approximately 1% asphalt to the total binder content (AsphaltPro.com 2012; Jackson 2012). • Mixtures with tear-off shingles require slightly more virgin asphalt than similar mixtures using manufacturer shingle waste (McGraw et al. 2010). Recent research shows that the percentage of the virgin asphalt in mixture is more important to good pavement per- formance than the PG grade of the virgin asphalt (Johnson et al. 2013). The minimum amount of virgin asphalt can be defined by using a ratio of virgin asphalt to the total asphalt content asphalt binder ratio (ABR), which is calculated as: ABR Virgin asphalt, % Total asphalt content, = %     Alternatively, the maximum percent of recycled asphalt that can contribute to the total asphalt content can be defined as a ratio of the maximum percentage of recycled material asphalt to the total asphalt content (i.e., recycled binder ratio, RBR), which is calculated as: ( )( ) ( )( ) = +          RBR RAP AC RAP% RAS AC RAS% Total asphalt content 100 The Minnesota Department of Transportation (MnDOT) established a minimum criterion of 70% for the ABR for its specifications in 2012. A revised version in 2013 defines the ABR based on the type of recycled materials, location in the pavement structure, and the specified virgin asphalt grade (Table 2) (Johnson et al. 2013). The Texas DOT (TxDOT) specification requires a maxi- mum RBR based on the originally specified PG asphalt, the allowable substitution of another PG asphalt, and the location of the mixture in the pavement structure (Table 3) (TxDOT 2014, Item 341). The Bonaquist methodology is used by a number of researchers to evaluate if the recycled material asphalt fully contributes to the total asphalt content of the mixture. This method requires dynamic modulus, E*, data for the com- pacted recycled material asphalt mixture be measured for a range of test temperatures and loading frequencies (Bonaquist 2007). The recycled material asphalt is extracted, recovered, and blended with virgin asphalt at the same percentages used for the mixture. The determined blended asphalt DSR shear modulus, G*, using a range of test temperatures and load- ing frequencies and the G* obtained data is mathematically converted to E* values using the Hirsch model. The recycled asphalt fully contributes to the total asphalt content of the mixture when the dynamic modulus from the mixture testing and the E* values calculated using the Hirsch model overlap. Mixed reports of the usefulness for this approach found in the literature are briefly described here. McDaniel et al. (2012) used the Bonaquist method to evaluate if RAP asphalt fully blended with the virgin asphalt using 24 plant-produced RAP mixtures obtained from five different contractors. Twenty of the mixtures show that most RAP asphalt contributed to the total asphalt. However, one mixture showed that the RAP asphalt only partially con- tributed, and three other mixtures showed little contribution from the RAP asphalt. This study showed that RAP asphalt provided a significant contribution about 80% of the time, but only partial to little contribution 20% of the time. Turner (2013) found that the Hirsch model did not accu- rately estimate asphalt properties of plant or laboratory pro- duced mixtures used in this study. The model was also not sensitive to changes in the asphalt properties resulting from increases in the RAP content. Total Asphalt Content—Section Summary • The TAC of the asphalt mixture is calculated using the sum of the percentage of virgin asphalt and the asphalt contained in the percentage of the recycled materials added to the mixture. The percentage of useful recycled asphalt included in the calculation of the total asphalt content can be considered as 100% useful, 0% useful, or some percentage in between. The asphalt availability factor is used to define the percentage of useful recycled material asphalt. • Recent research shows that the performance of recycled material asphalt mixtures is a function of the percentage of the virgin asphalt in the mixture and either the ABR or the RBR can be used to control the amount of virgin asphalt in the mixture. TABLE 2 CRITERIA FOR MNDOT MINIMUM RATIO VIRGIN ASPHALT TO TOTAL ASPHALT BINDER (ABR) Specified Asphalt Grade Lift Minimum ABR for Recycled Material Asphalt Mixtures RAP only RAS only RAP and RAS PG XX-28 PG 52-34 PG 49-34 PG 64-22 Wear 70 70 70 Non-Wear 70 70 65 PG 58-34 PG 64-34 PG 70-34 Wear 80 80 80 Non-Wear Source: MnDOT (2013; Table 2360-8).

14 • The Bonaquist method can be used to estimate if most or all of the recycled material asphalt contributes to the total asphalt content. At this time, this method is primarily a research tool. SELECTING THE VIRGIN ASPHALT GRADE FOR RECYCLED MATERIAL MIXTURES It is important that the virgin asphalt grade be selected so that combined the virgin and recycled asphalt properties meet the specified requirements. When lower percentages of recycled materials are used, usually less than 15%, no change in the typical virgin asphalt grade is required. When the recycled material content is between 15% and 25%, one grade softer is typically selected for the virgin asphalt. FHWA recommends extracting, recovering, and testing the recycled material con- tent when using content of more than 25%. The test results are used to develop blending charts for selecting the required upper and lower PG temperatures used to specify the virgin asphalt. One approach for using blending charts is to select the per- centage of recycled material to be used in the mixture, deter- mine the critical temperature determined for the recycled material asphalt and the critical temperature for the blend of virgin and recycled asphalt, then calculate the critical tem- perature for the virgin asphalt, Tvirgin. For example, using the percent RAP (RAP%) as the recycled material, the equation for calculating the virgin asphalt critical temperature is: T T T( )( )( )( )= − − RAP% 1 RAP%virgin blend RAP The required time and cost associated with determin- ing all of the different asphalt properties required for this approach can deter agencies from using more than 24% RAP. Other agencies have used research studies and local experi- ence to identify specific virgin asphalt grades to be used for any percentage of recycled materials in asphalt mixtures. For example, recent changes in the Florida specifications still use the three-tiered approach for adjusting the selection of the virgin asphalt, but identify the specific grade for each level of recycled material content (Table 4). TABLE 3 ALLOWABLE SUBSTITUTE PG BINDERS AND MAXIMUM RECYCLED BINDER RATIOS Originally Specified PG Binder Allowable Substitute PG Binder Maximum RBR1 for Recycled Material Asphalt Mixtures, % Surface Intermediate Base HMA 76-222 70-22 or 64-22 20.0 20.0 20.0 70-28 or 64-28 30.0 35.0 40.0 70-222 64-22 20.0 20.0 20.0 64-28 or 58-28 30.0 35.0 40.0 64-222 58-28 30.0 35.0 40.0 76-282 70-28 or 64-28 20.0 20.0 20.0 64-34 30.0 35.0 40.0 70-282 64-28 or 64-28 20.0 20.0 20.0 64-34 or 58-34 30.0 35.0 40.0 64-282 58-28 20.0 20.0 20.0 58-34 30.0 35.0 40.0 WMA3 76-222 70-22 or 64-22 30.0 35.0 40.0 70-222 6-22 or 58-28 30.0 35.0 40.0 64-224 58-28 30.0 35.0 40.0 76-282 70-28 or 64-28 30.0 35.0 40.0 70-282 64-28 or 58-28 30.0 35.0 40.0 64-284 58-28 30.0 35.0 40.0 Texas Section 341, Table 5. 1Combined recycled binder from RAP and RAS. 2Use no more than 20.0% recycled binder when using this originally specified PG binder. 3WMA as defined in Section 341.2.6.2 “Warm Mix Asphalt (WMA).” 4When used with WMA, this originally specified PG binder is allowed for use at the maximum recycled binder ratios shown in this table. TABLE 4 VIRGIN ASPHALT GRADE FOR RAP MIXTURES RAP Content, % PG Grade 0–15 PG 67-22 16–30 PG 58-22 >30 PG 52-28 Source: Florida Department of Transportation Specifications (2015, Table 334-2).

15 SAMPLE PREPARATION FOR MIX DESIGNS Laboratory procedures for material preparation, batching, preheating, mixing, and compacting asphalt mixtures for mix designs were originally developed using virgin aggre- gates and asphalts. Batching of materials has to consider what portions of the recycled material mass are included in the solid particulate measurements and what part of the mass is included in the determination of the total asphalt content. Temperatures, mixing times, and order of addition of materials are based on typical asphalt plant operations. However, when recycled material is included in the mixtures, adjustments to conventional procedures may be necessary to account for how, when, where, and at what tempera- tures these materials are added during the asphalt plant production. Calculating Batch Weights The mass of any nonusable recycled asphalt is to be included as a part of the recycled material aggregate mass (AASHTO R-35). Various agencies have developed their own equa- tions for determining material batch weights (masses) for mix design samples. Generic equations, modified from the Oregon DOT Section 2327-CB (calibration batch sheet) spreadsheet example to include asphalt availability factors for both RAP and RAS are shown here (ODOT 2013). Batching calculations start with determining the total mass of the asphalt mixture sample, Masssample, needed to produce the desired sample height after compaction. The total mass of asphalt for one of the mix design asphalt contents, Pb, to be used in the design is calculated as: Pb( )( )( ) =Mass Mass 100total asphalt sample Typical mix designs use from three to five different total asphalt contents to determine the optimum asphalt content to be used with the selected aggregate gradation. Once the total mass of asphalt is determined, the total mass of aggregate, Masstotal aggregate, is calculated: ( ) ( )= −Mass Mass Masstotal aggregate sample total asphalt The mass of RAP asphalt that will be used in calcula- tions of ABR and RBR is calculated using the target total asphalt content, Pb; the percentage of RAP to be used, RAP%; the percentage of recycled asphalt in the RAP, Pbr; and the RAP asphalt availability factor, FRAP. All percent- ages are expressed in whole numbers (i.e., not in decimal form): Mass Mass 1 100 100 RAP% 1 100 100 1 RAP asphalt RAP sample ,RAP ,RAP ( ) ( )( )= − −   + −        F P P P b br br The mass of RAP aggregate is calculated as: Pbr = −  Mass Mass 100 1RAP aggregate RAP asphalt ,RAP The sum of both the calculated RAP asphalt and RAP aggregate is the mass of RAP that is to be batched: = +Mass Mass MassRAP RAP asphalt RAP aggregate If RAS is also included in the asphalt mixture, the same series of calculations are required to calculate the mass of RAS material to be batched. First, calculate the mass of RAS asphalt: Mass Mass 1100 RAS% 1 100 100 1 RAP asphalt RAS sample ,RAS ,RAS ( )( ) ( )= −−   + −        F P P P b br br Next calculate the mass of RAS aggregate: Pbr = −  Mass Mass 100 1RAP aggregate RAS asphalt ,RAS And then calculate the mass of RAS material to be batched: = +Mass Mass MassRAS RAS asphalt RAS aggregate Two additional calculations are used to determine the mass of virgin aggregate to be batched: = − − Mass Mass Mass Mass virgin aggregate total aggregate RAP aggregate RAS aggregate And the mass of virgin asphalt to add during mixing: = − − Mass Mass Mass Mass virgin asphalt total asphalt RAP asphalt RAS asphalt Material Preparation, Mixing, and Compacting Each research study found in the literature uses defined, but laboratory-specific, steps to prepare materials for batching, combine materials for heating, and determine the order of addi- tion of materials into the mixing bowl, short-term aging times and temperatures, and levels of compaction. Two examples of

16 variations in the steps used to prepare mix design samples are shown in Table 5. Molenaar et al. (2011) compared laboratory mixing procedures to those used for two different asphalt plants (parallel flow plant and Astec Double Barrel drum plant) (Table 6). Standard laboratory practices as used by these researchers call for preheating both the virgin aggregate and RAP to 170°C (338°F). The parallel drum mix plant evaluated for comparison superheated the virgin aggregate to above 170°C (338°F) and preheated the RAP to 130°C (266°F). Both the virgin aggregate and RAP were dry mixed before adding the liquid asphalt. The second plant used for comparison was an Astec Double Barrel plant that super- heated the virgin aggregate well above the standard labo- ratory temperature of 170°C (338°F). Higher temperatures were necessary because the RAP was added as stockpiled (at ambient temperatures, moisture contents between 1% and 4%) and the conductive heat transfer from the hot aggregate to the RAP is needed to both dry and preheat the RAP before adding the hot liquid virgin asphalt. Research- ers tried to approximate the parallel plant temperatures in the laboratory but failed to come close to replicating heat- ing conditions in the Astec Double Barrel drum. Sample Preparation for Mix Designs— Section Summary • Batch weights (masses) of the recycled materials are to be adjusted by the mass of the recycled material asphalt that is not considered in the calculation of the total asphalt content. – No standard procedure for batching, preparing, and mixing materials for samples with recycled materials was found in the literature. Laboratory temperatures and procedures for drying and preheating varied widely and do not appear to replicate temperatures and conditions used in typical asphalt plants. TABLE 5 COMPARISON OF DIFFERENT LABORATORY PROCEDURES FOR BATCHING, PREHEATING, AND MIXING Step Minnesota Study (RAP Study) (Source: McGraw 2010) Oregon Study (RAP and RAS Study) (Source: Scholz 2010) Aggregates Fractionate coarse and fines on 2.36-mm (No. 8) sieve; further fractionate coarse on individual sieve sizes Fractionate into individual sizes (full range of sieve sizes) RAP Fractionate on 4.75-mm (No. 4) sieve; further fractionate coarse on individual sieve sizes Fractionate into individual sizes from 9.5-mm (3/8-in.) to passing 0.15-mm (No. 100) sieve sizes RAS Not applicable Fractionate into two sizes: ½-in. to 0.30-mm (No. 30), and passing 0.30-mm (No. 30) Preheating Aggregate: Preheat for 4 to 5 h at 315oF (157oC) Aggregates: Preheat to mixing temperature RAP: Preheat for 4 to 5 h at 315oF (157oC) RAP: Preheat to mixing temperature RAS: Not applicable RAS: Keep at room temperature Mixing Aggregates and RAP dry mixed for 1 to 2 min Dry-mix aggregates, RAP, and RAS Virgin asphalt added Virgin asphalt added Mixed for an additional 2 min Mixed (no time indicated) Short-Term Aging 2 h at 275 oF (135oC) At compaction temperature Compaction Ndesign = 60 Not noted TABLE 6 VARIABLES USED IN STUDY TO SIMULATE PLANT CONDITIONS IN THE LABORATORY MIXING PROCEDURES Production Facility Temperature Variables Virgin Aggregate Preheating Temperature, oC (oF) RAP Preheating Temperature 30% RAP 60% RAP Typical Laboratory Procedure 170 (338) 170 (338) 170 (338) Parallel Flow Plant 240 (464) 330 (626) 130 (266) Astec Double Barrel Plant 290 (554) 430 (806) 25 (77) 345 (653) 515 (959) 25 (77) Source: Molenaar et al. (2011).

17 MIXTURE TESTING The volumetric properties of the compacted samples are used as parameters for determining the optimum total asphalt con- tent using the selected aggregate gradation. Performance-based testing of the compacted mixtures is used to evaluate that the likelihood the mixture, as designed, will achieve the design service life. Volumetrics Examples of recent research that report changes in mix design volumetrics resulting from the percentage and type of recycled materials are summarized in Table 7. There is general agree- ment that the asphalt film thickness decreases and the dust content increases with increasing percentages and/or differ- ent types of recycled materials. Some studies report decreases in air voids, VMA, and voids filled with asphalt (VFA) with increasing percentages of recycled materials or when using different types of recycled materials, whereas other studies have reported opposite trends. Differences in the reported volumetric trends are most likely a function of other factors such as gradations, effective volume of asphalt, and additives, rather than simply the use or increasing percentage of recycled materials. AASHTO PP78-14 notes that although the percentage of RAS typically used in asphalt mixtures is small, the non- asphalt components that include the aggregate particles and backing materials can increase the VMA. At the same time, the dust content can decrease the VMA; however, the net change is usually a net increase in VMA. The dust-to-asphalt ratio can also increase. The AASHTO standard recommends limiting the percentage of RAS to 5% until more is known about the impact of RAS on mixture volumetrics. MnDOT uses the adjusted asphalt film thickness (AFT) in its specification to ensure a minimum effective asphalt volume coverage that is a function of the aggregate surface area. P SA P be s ( ) ( )=AFT 4870 And the AFT is: SA[ ]( )= + −Adj. AFT AFT 0.06 28 AFT = asphalt film thickness, µm; SA = surface area, ft2/lb; Pbe = percentage effective binder; Ps = percentage solids; and Adj. AFT = adjusted asphalt film thickness, µm. The surface area of the aggregate is calculated as: SA a b c d e f g = + + + + + + + 2 0.02 0.04 0.08 0.14 0.30 0.60 1.60 SA = surface area, ft2/lb; a = 4.75-mm (No. 4); b = 2.36-mm (No. 8); TABLE 7 EXAMPLE OF VOLUMETRIC CHANGES WITH INCREASING RECYCLED MATERIAL PERCENTAGES* *Includes both RAP and RAS studies. S = similar results in given research study; M = mixed results in given research study. Indicates a given property or test result decreases with increasing percentage of recycled material in given study. Indicates a given property or test result increases with increasing percentage of recycled material in given study. Testing Influence on Results for Mixtures with Recycled Materials References Volumetrics Air voids 1 1 Roque et al. (2015); Booshehrian et al. (2012) VMA 3 2 Roque et al. (2015); Lee et al. (2015); Daniel and Lachance (2005); West and Willis (2014); Booshehrian et al. (2012); AASHTO PP78-14 VFA S 3 1 Roque et al. (2015); Lee et al. (2015); Daniel and Lachance (2005); Booshehrian et al. (2012); Shannon (2012) Film thickness 3 Shannon (2012); AAT (2011) Dust content S 4 Lee et al. (2015); Newcomb et al. (2007); Booshehrian et al. (2012); Shannon (2012) Mixture Properties Needed to Calculate Volumetrics Theoretical maximum gravity 1 Lee et al. (2015) Percent binder absorbed 2 Lee et al. (2015)

18 c = 1.18-mm (No. 16); d = 0.6-mm (No. 30); e = 0.3-mm (No. 50); f = 0.15-mm (No. 100); and g = 0.075-mm (No. 200). An alternative equation found in the literature for calcu- lating the AFT is (AAT 2011): V S P G be S S mb =  AFT 1,000 Where: AFT = apparent film thickness, µm; Vbe = effective binder content, % by total weight of mixture; Ps = aggregate content, % by total weight of mixture; and Gmb = bulk specific gravity, compacted sample. A simplified equation for calculating the aggregate surface, SS, area is: S P P PS ( )= + +50.30 0.15 0.075 Performance Testing Performance testing used to evaluate key mixture properties related to key pavement distress(es) includes: • Dynamic modulus to evaluate mixture stiffness. • Loaded wheel rut testing. • Cracking (bottom down and/or top-down traffic-related cracking, thermal cracking, reflective cracking) test methods: – Bending beam fatigue – Disk-shaped compact tension (DSC) – Indirect tension (IDT) – Overlay tester (Texas) – Repeated direct tension – Semi-circular bend (SCB) – Simplified viscoelastic continuum damage (S-VECD) – Thermal stress restrained stress test (TSRST) and uniaxial thermal stress and strain (UTSST). Dynamic Modulus (AASHTO TP79) The Asphalt Mixture Performance Tester (AMPT) can be used to determine the dynamic modulus (stiffness), referred to as the complex modulus, E*, over a range of tempera- tures and/or loading frequencies (Figure 7). Stiffer mixtures are more resistant to rutting and, when located in the lower lifts, provide support to minimize longitudinal cracking in the wheel paths. Cylindrical samples are loaded by applying a uniaxial sinu- soidal in compression to the sample in an unconfined or con- fined condition. Test temperatures of 14°F, 39°F, 68°F, 102°F, and 123°F (-10°C, 4°C, 20°C, 38.8°C, and 54.4°C) have been used by some researchers (Michael 2011; Cooper et al. 2014). Loaded Wheel Tracking Device (AASHTO TP63) Loaded wheel devices [Hamburg, Asphalt Pavement Analyzer (APA)] simulate mixture deformation resulting from multiple passes of traffic loads (Figure 8). Mixtures that can sustain a preset number of passes without exceeding a maximum rut depth are considered resistant to rutting. When the load- ing passes are conducted under water, a discernible change (inflection point) in the depth versus number of passes is iden- tified as the stripping inflection point (SIP). A higher number of passes associated with the inflection point indicates a more moisture-resistant mixture. Test Methods for Evaluating Cracking Potential There are eight test methods that can be used to evaluate traffic-related (fatigue, top-down, bottom-up) cracking, thermal cracking, and reflective cracking. Each test method is briefly described here and includes a description of the type(s) of cracking evaluated for the testing condition(s). Bending Beam Fatigue Testing (AASHTO T321) The bending beam fatigue test evaluates the potential for tra- ditional fatigue cracking (i.e., bottom-up cracking). Testing is FIGURE 7 Set up for AMPT (Source: Michael 2011).

19 usually conducted using at least two different stress or strain levels and the data are analyzed to determine the slope of the stress (or strain) versus the number of cycled to failure relationships (log-log relationships). A rectangular beam is cut from a slab of compacted asphalt mixture, clamped into an apparatus, and a sine wave loading generates tensile stresses over the bottom center third of the beam (Figure 9). Loading frequencies can vary from 5 to 10 Hz and failure is typically defined as a 50% reduction of the initial stiffness. The resistance of the mixture to traffic-related flex- ural stresses and strains increases with the increasing number of cycles to failure. Alternatively, the data can be used in math- ematical models to estimate the fatigue life of the mixture. Disc-Shaped Compact Tension (ASTM D7313) The disc-shaped compact tension test determines the frac- ture energy of an asphalt mixture at low temperatures. The low temperature cracking potential decreases as the fracture energy increases. Variations of this test method conducted at intermediate test temperatures can be used to evaluate potential reflective cracking characteristics. Testing is conducted on a 2-in.-thick disc-shaped sample that has been cut from a gyratory compacted cylinder or a core (Figure 10). Two holes are drilled into either side of a thin notch cut into the edge of the sample and pins are inserted into the holes. A constant strain is applied to the notch (i.e., crack) so that it opens at a rate of 1 mm/minute. Failure typi- cally occurs between 1 mm and 6 mm of crack opening. The standard test temperature is 10°C warmer than the lower PG temperature. Indirect Tension (AASHTO T322) The indirect tension test is used to determine the creep com- pliance and tensile strength of the mixture at low temperatures (Figure 11). An increase in the creep compliance indicates a mixture that can better resist low temperature cracking owing FIGURE 8 Types of loaded wheel testers used to evaluate asphalt mixture rutting potential APA (upper left) and Hamburg (lower right) (Source: Willis et al. 2012; Pavement Interactive website 2015).

20 to increased strains as temperatures drop. The tensile strength decreases with increases in creep compliance (i.e., inverse relationship). Testing is conducted on a disc-shaped sample that has been cut from a gyratory compacted cylinder or core and is about 1.5-in. to 2.0-in. thick. Typically, the creep compliance is determined by applying a static load for 100 seconds. Because this portion of the testing is not destructive as long as the strain, e, is kept below 500-µe, several tests can be conducted at different temperatures. Once the creep com- pliance testing is completed, the indirect tensile strength is determined (destructive portion of the test). The sample is loaded at a strain rate of 12.5 mm per minute until failure and the tensile strength is determined when the maximum load is reached. The traffic-related top-down cracking potential can be estimated by using a variation of this test conducted at inter- mediate temperatures and calculations of the energy ratio, ER (Willis et al. 2012). Recommended energy ratio criteria are a minimum of 1.0 for less than 250,000 equivalent single-axle loads (ESALs)/year, a minimum of 1.3 when the traffic is below 500,000 ESALs/year, and a minimum of 1.95 for traffic levels up to 1 million ESALs. Resilient modulus is obtained from stress and strain mea- surements by applying a repeated haversine load for 0.1 sec- ond followed by a 0.9 second rest period and measuring the stress and strain. Next, the creep compliance is performed using AASHTO T322-07 at a test temperature of 50°F (10°C) and a test duration of 1,000 seconds (creep compliance). The indirect tensile strength dissipated creep strain energy, which is a portion of the area under the stress-strain curve. The energy ratio: ER DSCE S m D 7.294 10 6.36 2.46 10f t5 3.1 8 2.98 1 [ ])( )( )( = σ − +− − − Where: s = tensile stress at the bottom of the asphalt layer, 150 psi; D1, m = power function parameters; FIGURE 9 Beam fatigue apparatus (Source: Pavementinteractive [Online]. http://www.pavementinteractive.org/article/ flexural-fatigue/). FIGURE 10 Disc-shaped tension test (Source: NCHRP 9-57 Workshop 2015). FIGURE 11 Indirect tension test (Source: NCHRP 9-57 Workshop 2015).

21 DSCEf = dissipated stress creep energy at failure (portion of area under stress–strain curve from indirect tensile test); and ER = energy ratio. Overlay Tester (TEX-249-F) The Overlay Tester is used to estimate the potential resis- tance of a mixture to reflective cracking and/or traffic-related top-down cracking. Resistance to cracking increases with the number of cycles needed to fail the sample. The Overlay Tester uses a specimen cut from a gyratory compacted sample, adhered to horizontal steel plates sepa- rated by a narrow gap, which are moved back and forth using a saw tooth waveform (Figure 12). The force required to move the plates is recorded and failure is defined as a 93% reduction in the load magnitude recorded for the first cycle. Texas DOT (TxDOT) uses a maximum displacement of 0.025 in. (0.635 mm); however, some research studies indi- cate that this displacement may be too high to evaluate stiff mixtures, such as those containing recycled materials. One study used displacement openings of 0.01, 0.013, and 0.015 in. (0.254, 0.330, and 0.381 mm). Results using the higher displacement were more variable and the lowest displacement level extended the number of cycles to fail- ure for stiff mixtures to more than 2,000. A displacement of 0.013 in. was considered the most effective compromise between lowering the variability and keeping the testing time to a reasonable level. Repeated Direct Tension (Texas A&M Test Method) Information obtained from the repeated direct tension test are used to develop estimates of load-related bottom-up and top-down traffic-related cracking. The test uses a cylindrical sample (heights more than diameter) and applies cyclic ten- sile loads to obtain stress and strain data. The results are used to calculate Paris’ law parameters, endurance limits, healing properties, and average crack size (Figure 13). Semi-Circular Bend (AASHTO TP105) The SCB critical fracture release energy, determined using multiple-notch depths and intermediate test temperatures [e.g., 77°F (25°C)], can be used as an indication for top-down, fatigue, and reflective cracking potential. When a single low temperature and a single notch depth is used, increases in fracture energy, Gf, and fracture toughness, K1C, indicate increases in low temperature cracking resistance. A thin circular disc is cut out of a gyratory compacted sample, or core, then cut in half to produce the semi-circular test specimen (Figure 14). The flat edge of the half circle is notched to the desired notch depth (a) to the specimen radius (rd), typically from 0.5 to 0.75 (for intermediate tempera- ture testing). The specimen is supported at either end of the flat side of the semi-circle (notched side facing down) and a constant load is applied to the top of the sample at a rate of 0.20 in./minute (0.5 mm/minute). The load and deformation with time is recorded and used to calculate the critical energy release rate, Jc value: J b dU dac ( )= 1 Where: Jc = critical strain energy release rate, kJ/m2; b = specimen thickness, m; FIGURE 12 Overlay tester for evaluating reflective cracking resistance (Source: Klutzz and Mogawer 2012). FIGURE 13 Repeated direct tension test (Source: NCHRP 9-57 Workshop 2015).

22 U = strain energy to failure (i.e., area under deformation- stress curve up to maximum stress), kJ; a = notch depth, m; and dU/da = change of strain energy with notch depth, kJ. Test results using different notch depths are used to calcu- late the strain energy that is plotted versus the notch depth, and the slope of the line is the value used for dU/da in the previous equation. Simplified Viscoelastic Continuum Damage (AASHTO TP107) The S-VECD test uses stress and strain measurements acquired under different loading conditions to estimate bottom-up and top-down traffic-related cracking (Figure 15). First, the dynamic modulus or frequency/temperature sweep testing is used to measure the mixture stiffness followed by the application of a constant strain until failure. The data from these tests are used as input into advanced mathematical models (e.g., linear visco- elastic continuum damage and viscoelastic continuum damage models with a public domain finite element program, FEP++). Thermal Stress Restrained Specimen Test and Uniaxial Thermal Stress and Strain (AASHTO TP105) The TSRST is used to measure the critical low cracking tem- perature and the tensile stress at failure (Figure 16). A rectangular beam cut from a compacted asphalt mix- ture slab, or pavement section, is confined at either end so that it cannot contract as the temperature is lowered at 18°F (10°C) per hour. As the temperature drops, the stress essen- tial to maintain the fixed specimen length increases. When the stress level exceeds the tensile strength of the material, the sample fractures (fails). The temperature at which the specimen fails is the critical cracking temperature. FIGURE 14 SCB testing set up and data plots needed for calculations of the energy ratio (Source: NCHRP 9-57 Workshop 2015). FIGURE 15 Simplified viscoelastic continuum damage set up (Source: NCHRP 9-57 Workshop 2015). FIGURE 16 Thermal stress restrained specimen test set up (Source: Western Regional Superpave Center [Online]. http://www.unr.edu/wrsc/research/facilities/asphalt).

23 Another use for this test configuration is measuring the coefficient of thermal contraction (UTSST). Examples of Performance Test Results Examples of recent results for a range of performance tests for RAP asphalt mixtures are summarized in Table 8. In gen- eral, increasing percentages of RAP decreases rutting poten- tial and increases stiffness. Increasing percentages of recycled materials increases low temperature cracking potential (i.e., raises the critical low temperature). Mixed results, both within and between studies, can be found for cracking poten- tial at intermediate temperatures and moisture sensitivity. Fewer test methods (Table 9) have been used to evaluate the performance characteristics of RAS asphalt mixtures and findings tend to show limited significant differences between control and RAS mixture properties, which may be because of the small amount of RAS that is added (typically 3% to 5% typical). A number of recent studies have investigated the use of rejuvenators added to the asphalt to help soften the stiffer recycled asphalt. A variety of materials used in the studies included those defined as rejuvenator or recycling addi- tives in AASHTO R14 or ASTM D4552 standards, as well as waste vegetable oil, waste vegetable grease, organic oil (Hydrogreen S™), distilled tall oil, aromatic extract, waste engine oil (Zaumanis et al. 2014), flux oil, lube stock, slurry oil, lubricating oils, extender oils, Cyclogen-L (Cooper et al. 2014), and other specialty products (Al-Qadi et al. 2009). TABLE 8 SUMMARY OF CHANGES IN RAP MIXTURE PROPERTIES Testing Influence of Increasing RAP References Rutting Rutting (loaded wheel units) S 2 Maupin (2008); Zaumanis et al. (2014); Willis et al. (2012); Watson et al. (2008) Creep flow time M Daniel and Lachance (2005) Creep stiffness 1 Abdulshafi et al. (2002) Cracking (Intermediate Temperatures) Fatigue S 1 2 M Maupin et al. (2008); Abdulshafi et al. (2002); Zaumanis et al. (2014); Vukosavlievic (2006); Watson et al. (2008); McDaniel et al. (2012) Reflective cracking (Overlay Tester) 1 Willis et al. (2012) Dissipated energy 1 Vukosavlievic (2006) Fracture energy 1 Vukosavlievic (2006) SBC Fracture energy 2 1 Lee et al. (2016); Willis et al. (2012); Johnson et al. (2013) Tensile strength 2 Vukosavlievic (2006); Johnson et al. (2013) Fracture work M Lee et al. (2015) Moisture Sensitivity Moisture sensitivity S 1 Olard (2010) Toughness index 1 Vukosavlievic (2006) Differences over Range of Temperatures Stiffness (dynamic modulus) 3 M Daniel and Lachance (2005); Abdulshafi et al. (2002); Roque et al. (2015); Lee et al. (2015); Olard (2010); McDaniel et al. (2012) Phase angle, mix 1 Abdulshafi et al. (2003); Vukosavlievic (2006) Fracture toughness S M Lee et al. (2015); Johnson et al. (2013) Low Temperature Testing Indirect tensile strength M Roque et al. (2015) Indirect tensile creep compliance S 1 2 Zaumanis et al. (2014); Roque et al. (2015); Johnson et al. (2013); Watson et al. (2008) Thermal cracking 2 Zaumanis et al. (2014) Critical cracking temperature 2 Vukosavlievic (2006); McDaniel et al. (2012) Notched fracture energy 1 Swiertz et al. (2011) Failure strain 1 Roque et al. (2015) Energy Ratio 2 Roque et al. (2015); Willis et al. (2012) S = similar results in given research study; M = mixed results in given research study. Indicates a given property or test result decreases with increasing percentage of recycled material in given study. Indicates a given property or test result increases with increasing percentage of recycled material in given study.

24 Results for a range of performance testing for RAS mix- tures with rejuvenators were found in the literature and are summarized in Table 10. Changes in the mixture properties depend on the percentage and type of rejuvenators used in the studies. Rejuvenators can reduce mixture stiffness and lower the critical low temperature when used in sufficient amounts; however, this can also increase the rutting potential. Care is required to select an optimum percentage of rejuvenators to achieve the desired results. Mixture Testing—Section Summary • RAP can either increase or decrease mixture volumet- rics depending on variables such as gradation, effective volume of asphalt, and other additives. • The nonasphalt components in RAS can increase VMA, and the dust content can decrease the VMA; however, the net change is usually a net increase in VMA. AASHTO PP78-14 recommends limiting RAS to 5% TABLE 9 SUMMARY OF CHANGES IN RAS MIXTURE PROPERTIES Testing Influence of Increasing RAS References Rutting Rutting S M Foo et al. (1999); Cooper et al. (2014) Creep flow time S Maupin et al. (2008) Cracking Fatigue S 2 Foo et al. (1999); Boyle and Bonaquist (2005); Maupin et al. (2008) Thermal cracking S Foo et al. (1999) Moisture Sensitivity Moisture sensitivity S M Boyle and Bonaquist (2005); Maupin et al. (2008) Stiffness S McGraw et al. (2010) S = similar results in given research study; M = mixed results in given research study. Indicates a given property or test result decreases with increasing percentage of recycled material in given study. Indicates a given property or test result increases with increasing percentage of recycled material in given study. TABLE 10 SUMMARY OF RECYCLED MATERIAL (RAP AND/OR RAS) MIXTURE PROPERTIES WHEN USING REJUVENATORS Testing Influence of Using Rejuvenators in Mixtures with Recycled Materials References Rutting Rutting S 1 2 Booshehrian et al. (2012); Shen et al. (2007); Tran et al. (2012); Green Asphalt Technologies (2012) Cracking Reflective cracking S 1 Booshehrian et al. (2013) Moisture Sensitivity Moisture sensitivity S Tran et al. (2012); Green Asphalt Technologies (2012) Indirect tensile strength 1 1 Shen et al. (2007); Green Asphalt Technologies (2012) Stiffness over Range of Temperature Stiffness S 1 Booshehrian et al. (2013); Sullivan (2011) Phase angle, mix S Booshehrian et al. (2013) Low Temperature Cracking TSRST M Booshehrian et al. (2013) Critical cracking temperature 2 Zaumanis et al. (2013); Tran et al. (2012) S = similar results in given research study; M = mixed results in given research study. Indicates a given property or test result decreases with increasing percentage of recycled material in given study. Indicates a given property or test result increases with increasing percentage of recycled material in given study.

25 until more is known about the impact of RAS on mix- ture volumetrics. • Increasing percentages of RAP may: – Increase stiffness and tensile strength, and decrease rutting potential. – Increase the thermal cracking potential (i.e., raise the cracking temperature). – Show mixed results for cracking potential at inter- mediate temperatures. • Asphalt mixtures with or without RAS tend to show similar or mixed results. – Most rejuvenators increase rutting potential, decrease stiffness, and lower the critical low temperatures. Care is required to select the optimum amount of rejuvena- tor used. ASPHALT PLANT PRACTICES AND EQUIPMENT When higher percentages of RAP are used in asphalt mix- tures, more attention to the RAP processing, stockpiling, and how RAP is added to the plant as needed (Udelhofen 2007). The age, type, and characteristics of the asphalt plant can limit the percentage and type of recycled materials that can be used. RAS material properties and sources of contami- nates vary significantly among manufacture waste and tear- off shingles; therefore, it is required that they be processed and stockpiled separately. This section summarizes the key factors that can influence the use of recycled materials in asphalt mixtures. Stockpiling and Processing Recycled Materials Both RAP and RAS recycled material properties vary by source. RAP aggregate gradations and asphalt contents vary by the type of mixture (e.g., large stone base asphalt mixtures, dense-graded mixtures, and open-graded friction course), depth of pavement layer milled, type of milling equipment, and depth of milling. RAS aggregate and asphalt properties vary significantly between manufacturing waste shingles and old roofing materials (tear-off shingles). The variability of either recycled material can be minimized by keeping different types and sources in separate stockpiles. Incoming recycled materials are to be documented (e.g., by source, mix type, aggregate properties, asphalt content, and shingle type), materials tracked (process auditing), and equipment and asphalt plant or facility operators trained on how to appropriately manage recycled material stockpiles. Storage Areas A major factor influencing asphalt plant production rates and drying costs is the moisture content of the recycled materi- als. Sources of moisture in the recycled material stockpiles include rain, water used during processing, water sprayed on conveyor belts to prevent sticking, or water misted on stock- piles for fugitive dust control. Moisture from rain can be minimized by covering the stock- piles (Figure 17). When the recycled material stockpiles are covered, an open-sided shed or building works most efficiently for access for loaders (Ontario Hot Mix Producers Association 2007). An opening at either end of the cover allows the loader operator to use the material stored in the shed to be the first used when producing the mix. The next most effective option is to use a conical-shaped stockpile to help naturally protect it from rain or snow, place stockpiles on a paved slope surface to drain any excess water, limit the stockpile height to reduce potential for self- consolidation, and limit use of heavy equipment on top of the stockpiles to avoid compaction (West 2010; Garrett 2012; Jackson 2012; Cleaver 2013). General estimates of typical RAP moisture contents by contractors are from 0.8% to 2% (Howard et al. 2009). Deter- mining RAP stockpile moisture prior to asphalt mixture pro- duction is a function of the sampling depth into stockpile, the size of the stockpile, whether the RAP stockpile has been fractionated or unfractionated (finer RAP holds more water), time since milling, and recent rainfall. Moisture and dust in the recycled materials can contribute to clogging screens if in-line processing is used to size RAP as it is fed into the asphalt plant. Besides moisture, a major challenge noted by Texas con- tractors when stockpiling RAS is workability. Hot weather and heating from solar radiation tends to stick RAS particles together, which makes it difficult to feed through cold feed bins and to obtain a uniform distribution in the mix. Covering RAS stockpiles not only limits additional moisture but helps with workability by limiting heating from solar gain. FIGURE 17 Covering stockpiles helps control the moisture content (Source: Jackson 2012).

26 RAS clumping in the stockpile can also be minimized by blending with an acceptable source of fine aggregate or with RAP. A ratio of RAP to RAS of either 75:25 or 80:20 can minimize clumping; however, the RAP/RAS blend must be consistent throughout the stockpile to prevent variations in the material properties of the total mixture (Carolina Asphalt Pavement Association 2011; NCAT 2012). RAP Stockpiles RAP is obtained from pavement demolition, milling, and asphalt plant waste. Demolition is done with bulldozers or backhoes, is usually limited to small areas, and produces large blocks of old asphalt pavement that are to be crushed. Milling (grinding) removes one or more layers of an exist- ing pavement surface. Milled materials tend to be finer and contain appreciable amounts of minus 0.075 mm than the gradation determined from cores and, typically, between 10% and 20% (Christman and Dunn 2013). Aggregate breakdown during milling is a function of the hardness and brittleness (impact resistance) of the aggregate, stiffness of asphalt that depends on the pavement temperature at the time of milling, milling machine speed, and depth of cut. Materials obtained from paved shoulders or lane widening projects, either by demolition or milling, may have differ- ent asphalt contents, aggregate gradations, and qualities of aggregates than those obtained from removing the main line roadways. Plant waste is what is left over at the asphalt plant when the plant starts up, shuts down, or mixtures are rejected by the agency. When fresh asphalt mixtures are added to RAP stockpiles, the fresh, unaged asphalt and the gradations with significantly fewer fines can increase the variability of the RAP stockpile asphalt content, asphalt properties, and grada- tion. These are all reasons why unused fresh mixtures, RAP from different sources, and RAP from different processes should be stockpiled separately to minimize RAP variability (Figure 18). Agency terminology used to identify RAP stockpile char- acteristics varies substantially among agencies (Table 11). Examples of terms used to indicate that no new material can be added to a RAP stockpile once the QC testing is completed include “designated,” “captive,” “non-continuous,” and “cer- tified.” Terms such as “active” and “continuous” are used to indicate RAP stockpiles that can be continuously replenished as the RAP is used. The continuous process of adding new material as the RAP is used can work well, but requires an established RAP QC plan that includes frequent, regular test- ing and analysis of the stockpile variability. This method is particularly helpful when the asphalt plant has limited space for multiple stockpiles. The consistency of the RAP stockpiles can be evaluated by monitoring the coefficient of variability (COV) for mul- tiple test results by taking samples from at least 10 different locations throughout the stockpile (AAT 2011). Alterna- tively, samples may be taken from haul trucks as the stockpile is built. Each sample is split so that one sample from each location can be used to determine the variability of the material properties stockpile (i.e., average, standard deviation) (Table 12). It is important that higher variability (higher COV) suggests stockpiles be reblended or the maximum percentage of RAP in mixture has to be reduced. The second set of split samples can be combined and split so that one “representa- tive” sample is tested for use in mix design calculations. Deleterious materials can be incorporated into the RAP when multiple lifts are milled (i.e., deep milling). This is because other materials such as crack fillers, soil from the underlying unbound layers, base materials, and paving geo- textiles used between layers to reduce reflective cracking are removed along with the old pavement (Cleaver 2013). Geo- textiles are a problem in RAP crushing operations because they tend to build up in crusher, wrap around moving parts, and lock up the crushing equipment. Geotextiles and crack fillers, which tend to be “ropes” of rubbery material, would be removed as the RAP is stockpiled. It is important that con- tamination of existing stockpiles be controlled, which means keeping out dirt, rubbish, vegetation, and trash. These con- taminates are to be removed as soon as they are noticed so they are not covered up as the stockpile is built. Usually the plant QC personnel and loader operators are responsible for continuously monitoring unprocessed and processed RAP (West 2010). Processing RAP RAP stockpiles are most often built by using a vibrating grizzly with a single screen to control the top size of the RAP in the stockpile. The 12.5-mm or 9.5-mm (½-in. or ³⁄8-in.) screen is a typical size used for scalping as about 75% of as-milled RAP passes through ½-in. sieve. Any material not passing through the top screen is fed into a crusher or lump breaker before being fed back in the grizzly (McDaniel and Anderson FIGURE 18 Example of RAP from variations in RAP materials from various sources [demolition (top, center) and millings from different projects (bottom)] (Source: West 2010).

27 2001; West 2010). This results in a single stockpile with a wide range of particle sizes that are, like virgin aggregate stockpiles, prone to segregation. When lower percentages of RAP are used, variations in the RAP gradations gener- ally have a low impact on the gradation and asphalt content of the final mixture. As the percentage of RAP increases, it becomes more difficult to maintain consistent gradation and asphalt content. Better control of the RAP gradation and asphalt content in the stockpile can be accomplished by using two screens, such as a slotted 5⁄8 in. by 6 in. screen and a ¼ in. by 6 in. screen State Specification Section (Source) Terminology Characteristics Iowa SS-0139, 2006 Classified Documented source, defined quality of materials Iowa 2303 Unclassified Unknown source; visual inspection for uniformity; tested for gradation and asphalt content Designated RAP Obtained from project; used on same project Active stockpiles Term used but not defined Certified RAP Sources known and no more than two sources in the same stockpile; stockpiles separated by aggregate quality and gradation, asphalt type, and content; no additional RAP added once tested Ohio 401.04 Standard RAP 100% passing 2-in. screen (nonsurface mixtures) 100% passing ¾-in. screen (surface mixtures) Extended RAP Fractionated or additional in-line processing of already approved stockpile; quality control plan. In-line processing: Double deck screen between cold feed bin and mixer with 9/16-in. screen for surface mixtures; 1.5-in. screen for base mixtures. Florida 334-2.3.3 Continuous RAP from one or more sources; processed, blended, or fractionated and stockpiled in a continuous manner; QC plan for monitoring gradation and asphalt content; visual inspection and review of data for suitability assessment Noncontinuous Individual (single) stockpile with known gradation and asphalt content; QC plan; no additional material added once approved Homogenous Material from Class I mixtures; requirements for aggregate quality, level of crushing, aggregate type (e.g., type of slag), and gradation; quality of RAP defined by lowest coarse aggregate quality; RAP from sources with similar asphalt content Conglomerate Class I mixtures; 100% passing 5/8-in. screen (or smaller) crushed coarse aggregate, but more than one aggregate type or quality; inconsistent gradation and asphalt content prior to processing; no steel slag or expansive materials Conglomerate “D” Quality Inconsistent gradation and asphalt content; no steel slag or expansive materials; coarse aggregate “D” quality or better TABLE 11 EXAMPLES OF AGENCY TERMINOLOGY USED TO IDENTIFY TYPES OF RAP STOCKPILES RAP Material Property Test Methods Frequency Minimum Number of Tests per Stockpile Maximum Standard Deviation Asphalt Content AASHTO T164 or AASHTO 308 1 per 1,000 tons 10 0.5% Recovered Aggregate Gradation AASHTO T30 1 per 1,000 tons 10 5.0%, 0.014 mm or larger and 1.5%, 0.075-mm sieve Recovered Aggregate Bulk Specific Gravity AASHTO T84 and AASHTO T85 1 per 3,000 tons 3 0.030* Consensus, source, or other aggregate properties Samples may be obtained by retaining and combining aggregates used for gradation analysis Binder Recovery and PG Grading AASHTO T319 or ASTM D5404 and AASHTO R29 1 per 5,000 tons 1 Not applicable Source: After West and Willis (2014). *Value recommended based on limited data and potential impact to mixture volumetrics (e.g., VMA). TABLE 12 SUGGESTED PRELIMINARY VALUES FOR CONTROLLING RAP STOCKPILE VARIABILITY

28 to fractionate the RAP into coarse and fine RAP stockpiles. Because the asphalt content of finer RAP particles is gener- ally higher than for the coarse RAP, fractionating the RAP also helps control the RAP asphalt content. The sizes used for fractionating will depend on mix designs being produced by the asphalt plant. Examples of commonly used sizes include: • Passing the ½ in. (12.5 mm) and retained on ¼ in. (6.35 mm) (coarse RAP). • Passing the ¼ in. (6.35 mm) screen (fine RAP). Fine RAP fractions are most useful when producing smaller maximum size aggregate mixtures typically used in wear courses or thin lifts (Brock and Richmond 2007; Cleaver 2013). Processing the RAP just in time for a full day’s production prevents the stockpiles from crusting in hot climate. The RAP can be screened and stockpiled for future use or processing can be completed during asphalt mixture produc- tion using in-line sizing and crushing operations. In-line RAP crushers or crusher circuits use roller crushers (lump breakers) or reduced speed impact crushers to break up agglomerations (clumps) of RAP and/or RAS. These crushers typically have a minimal influence on gradations and samples obtained dur- ing production can be tested to monitor gradations before and after the in-line processing (Ontario Hot Mix Producers Association 2007). In-line processing is most useful when using lower percentages of recycled material. Fractionating RAP stockpiles can also help manage the total dust content in the final asphalt mixture. A recent Iowa research study evaluated options for fractionating RAP to control and/or minimize dust content (Shannon 2012). The contractor participating in the study used an Astec ProSizer with a high frequency vibration screen to scalp the RAP on various screens to determine which size reduced the dust to useful levels so that higher percentages of RAP could be used and still meet dust to asphalt specification limits. Fractionat- ing on the 4.75-mm screen was selected as a useful size for managing the total dust content (Table 13). The dust content as well as the percentage of coarse and fine RAP fractions varied between the RAP sources. The percentage of fine frac- tions in the RAP stockpiles ranged from approximately 35% to 56% and the dust content of the fine RAP fractions from 13% to 19%. RAS Stockpiles AASHTO MP23 requires that separate stockpiles be main- tained for manufacturer waste shingles and tear-off shingles, because the RAS asphalt, particulates, and backing materials are significantly different between the two sources of RAS. The asphalt availability factors are also expected to be differ- ent for the different types of RAS. Because manufacturing waste and tear-offs come obtained from very different points in the product life cycle, the types and quantities of deleterious materials will also be very dif- ferent. Deleterious materials in manufacturing waste include packaging materials, scraps of unused or partially coated backing materials, and miscellaneous trash. Tear-off RAS can contain roofing underlayment materials, plywood from roof sheathing, roofing nails, scraps of flashing (aluminum scraps), and other construction demolition-related debris. AASHTO MP23-15 identifies deleterious materials as glass, rubber, soil, brick, paper, wood, and plastics, and is limited to no more than 1.5% of the material retained on and above the 4.75-mm (No. 4) sieve. The nonmetallic deleterious materials cannot exceed 0.5% of the total. Cleaning tear-off shingles before grinding helps limit deleterious materials (Figure 19) (Carolina Asphalt Pavement Association 2011; Jackson 2012). TxDOT requires less than 1.5% deleterious materials using the Tex-217-F that utilizes: • 1,000 g of RAS poured over a specially designed pan fitted with a magnet across the middle that removes most metals (Figure 20). • Metal contaminates are weighed and the percentage of metal in the RAS is calculated. • Remaining RAS material is sieved over the ³⁄8 in., No. 4, No. 8, and No. 30 sieves. The minus No. 30 material is discarded. Iowa RAP Source Passing 0.075-mm Sieve for Each RAP Fraction, % Coarse (12.5 mm to 4.75 mm) Fine (Minus 4.75 mm) Percent in total RAP stockpile, % Passing 0.075- mm sieve, % Percent in total RAP stockpile, % Passing 0.075- mm sieve, % RAP A 44.0 9.1 56.0 18.4 RAP B 50.6 11.1 49.4 19.1 RAP C 64.2 7.2 34.8 13.1 Source: After Shannon (2012). TABLE 13 EXAMPLE OF DUST CONTENT IN FRACTIONATED RAP

29 • Deleterious material retained on each sieve is visually determined and the total percentage of deleterious material is calculated. AASHTO MP23 requires that RAS be certified as conform- ing to local requirements concerning asbestos when using tear- off shingles. If testing for asbestos is required, either polarized light microscopy (PLM) and transmission electron micros- copy (TEM) test methods can be used. The TEM method is the most sophisticated for quantifying asbestos fibers in RAS. A list of accredited laboratories for asbestos testing can be found at: http://ts.nlst.gov/standards/scopes/programs.htm. Once the RAS has been tested, no more RAS can be removed to the stockpile (i.e., captive stockpile). A Missouri DOT report noted that in 2008 its Department of Natural Resources allowed RAS to be processed under the National Emission Standards for Hazardous Air Pollut- ants guidelines that do not require testing for asbestos when the tear-off shingles come from small residential buildings (Schroer 2009). Processing RAS AASHTO MP23-15 requires that RAS be ground so that 100% of the particles pass the 9.5-mm (³⁄8-in.) sieve before the RAS asphalt extraction. The finer grinding size helps the uniformity of the RAS in the asphalt mixture, which reduces the occur- rence of shingle clumps or pop-ups on the roadway (ForPros. com 2013). TxDOT grinds RAS so that it passes a ¼-in. sieve for better heat transfer and verifies the RAS gradations on a daily basis. Ground RAS can be uniformly blended with fine aggregate, powdered zeolites, or RAP to prevent the clump- ing (agglomeration) of the RAS (AASHTO MP23-14). Any of these materials that blended with the RAS would be included in the RAS aggregate calculations for mix design batch weights. Asphalt Mixture Production Various considerations and/or modifications are required to the asphalt plant when more than approximately 25% RAP is added to the mixture. Design changes to the cold feed bins and the conveyors can improve the uniformity of the amount of material that is fed into the plant. The plant type and char- acteristics can also limit the percentage of recycled materials that can be used. Batch plant characteristics are often more restrictive than drum mix plants. Feeding Recycled Materials into Asphalt Plants Wet RAP or hot summer temperatures make the recycled materials stickier and more likely to clump in the cold feed bins, stick to conveyor belts, and accumulate under convey- ors (West 2010; Jackson 2012). Cold feed bin features that are useful allow the moisture content to be monitored for plant control, prevent recycled materials from sticking to the sides of the bins, aid in the flow of material out of the bottom of the bin, and provide easy access for plant personnel to maintain and clean the bins (Garrett 2012). Newer feed bin features that are also useful include heat recovery bins that help dry material and reduce emissions by reducing the need for higher temperatures for drying. Cold feed bins with steep side walls also generally help pre- vent materials from sticking to the inside of the bin (Ontario Hot Mix Producers Association 2007). Increasing the RAS cold feed bin side slope by 70% improves the flow of the RAS, but decreases bin capacity. Because the RAS percent- age is approximately 5%, the reduced bin capacity is still acceptable for typical production rates (AsphaltPro.com 2012). Adding small amounts of material such as RAS can be more easily controlled by adding the RAS with RAP through a cold fed bin. It is important that RAS cold feed bins be cleaned out nightly to prevent clumping. A conveyor belt should have the proper slope, support, and optimum belt tension to keep the belt from sagging. Covering the conveyors to protect materials from the environment, vulca- nized belts, and the addition of good belt scrapers minimize the FIGURE 19 RAS stockpiles need to be free of debris prior to grinding (Source: Jackson 2012). FIGURE 20 TxDOT tray for removing metal from RAS during testing for the percentage of deleterious materials (Source: TxDOT test procedure designation Tex-217-F).

30 amount of recycled materials that stick to the belts. When the conveyors are weigh belt scales, metering difficulties through cold feed bins can lead to nonuniform amounts of recycled material being fed into the plant (West and Willis 2014). Add- ing RAS on top of the RAP on the conveyor helps prevent the RAS from sticking to the conveyor belt. Types of Asphalt Plants The percentage of RAP that can be added during production depends on the age, type, and characteristics of the asphalt plant (Brock and Richmond 2007; AAT 2011). Plant differ- ences directly impact the ability of the plant to add, dry, heat, and effectively mix materials. An Ohio survey in 2013 found that three states limited the percentage of RAP to 25% or less when producing asphalt mixtures with a batch plant and another three states limited the percentage of RAP to 40% when using a drum mix plant (ODOT 2013). Ohio DOT recently revised their specifications to allow a 5% higher percentage of RAP when either a counterflow drum mix plant or “mini-drum” batch plant (i.e., batch plant converted to continuous production) than when using a standard batch or parallel flow plant. Regardless of the plant age or type, when higher percentages of recycled materials are used it is critical to have sufficient amounts of the processed material on hand to provide a con- tinuous supply to the plant. Stopping the flow of RAP can cause the virgin asphalt to come into direct contact with the super- heated virgin aggregate, which is not only a fire hazard and causes smoking, but can damage the asphalt. Higher tempera- tures needed for superheating high RAP contents also increase the wear and tear on plant equipment. Additional equipment inspections and maintenance for drum shells, flights, and any other area exposed to higher temperatures are required. Higher RAP contents often require a softer asphalt, which means the plant has to have a second asphalt tank available. This can be a problem if the plant normally produces mix- tures with a single grade of asphalt. Batch Plants Batch plants use conductive heat transfer from the heated vir- gin aggregate to preheat the recycled materials in the weigh bucket and pugmill throughout the dry-mix cycle (Banasik 2000). When the moisture content recycled material is too high the water flashes off as steam, which leads to poten- tial emission problems. Plant operations or modifications are often necessary to facilitate drying and preheating higher percentages of recycled materials or recycled materials with elevated moisture contents (Table 14). Drum Mix Plants Older drums and newer single drum (either parallel or counter- flow) mix plant operations or characteristics can be modi- General Percentages of Recycled Materials Options Benefits Under 25% Use separate belt scale Add scalping screen for oversized materials (RAP) Improves uniformity of mixture Slow down how fast RAP is fed into pugmill Allows more time for steam to vent Keep recycled materials dry Add additional venting capacity Increase size of baghouse Add separate baghouse unit for venting extra steam Minimizes emission problems Keeps oily steam from clogging baghouse Convert to combination or continuous batch plant facility Diverts superheated aggregate from bucket elevator directly into pugmill Allows steam to be continuously vented into baghouse 25% to 40% Combine aggregate and RAP in bucket elevator, bypass main vibrating screen and discharge into No. 1 bin. Add additional scale adjacent to tower Minimizes blinding screens Better control of percent added Preheat RAP prior to entering tower Helps manage venting of steam 40% or more Additional feed bins Add parallel flow drum for recycled material drying Increase mixing times Improves gradation control Provides separate system for drying and venting Improves uniformity of mixtures Source: After Brock and Richmond (2007); AspahltPro.com (2012). TABLE 14 SUGGESTIONS FOR USING BATCH PLANTS FOR PRODUCING MIXTURES WITH INCREASING PERCENTAGES OF RECYCLED MATERIALS

31 fied to increase the percentage of recycled materials added to the mixtures. Counterflow and double drum mix plants are newer designs that accommodate a wider range and higher percentages of recycled materials (Table 15). Moisture Content and Higher Plant Temperatures Unless the plant design provides a separate system for drying and preheating the recycled materials, the temperature required for superheating the virgin aggregate is dependent on the amount of moisture in the RAP and the desired final mixture temperature. The most common equipment problem resulting from higher plant temperatures is caused by the elevated temperatures going from drum mixer to baghouse, which can increase from 10°F to 100°F higher than temperatures at the discharge point of the drum (Garrett 2012). This can damage the baghouse or carry liquid asphalt and fines into the bag- house and leads to increased maintenance and increased wear on drum shells, tires, and trunnions (Cleaver 2013). Elevated temperatures require significantly more energy. For example, when 20% RAP is added, a change in moisture content from 0% to 5% only requires an increase in the aggre- gate temperature of less than approximately 45°F (Figure 21; temperatures measured at the stack). However, at 50% RAP, Type of Plant Characteristics/Options Benefits Older drum plants Enlarge opening from RAP chute into drum Helps keep recycled materials from clogging opening Slow production rate to compensate for shorter drum lengths in older plants Allows more time for drying and for recycled asphalt transfer to virgin aggregate Avoid returning the dust from the baghouse near where the recycled material enters the drum Keeps dust from adhering to damp recycled material Parallel flow drum plants Longer drum lengths in newer plants allows for more drying and mixing time Helps with conductive heat transfer from superheated virgin aggregate to recycled material Helps remove moisture Allows more time for steam to vent Change flighting (plant staff needs training and experience for selecting proper flighting) Improves uniform mixing and drying of virgin aggregate and recycled materials Relocate RAP collar further down the drum toward the discharge point and shorten the liquid asphalt pipe lines Lengthens the time for superheating the virgin aggregate Add second dryer drum to replace RAP collar feed Improves ability to dry recycled materials by extending the dwell time (i.e., time in dryer drum) Counterflow drum plant Heats virgin aggregates and recycled materials in different areas of the drum Tend to have longer drum lengths Helps minimize emission problems Allows more time for drying and heat transfer Double drum plant Virgin aggregate superheated in inner drum Outer drum dries and preheats RAP before it enters the inner drum; asphalt is added in the inner drum Moisture flashes off in outer drum; keeps steam and asphalt separated that minimizes emissions Design allows higher RAP percentages (>40%) to be added to the mixture Sources: Banasik (2000); After ForConstructionPros.com (2005); Olard (2010); Garrett (2012); Astec (2014). TABLE 15 GENERAL PLANT CHARACTERISTICS THAT HELP OR LIMIT THE PERCENTAGE OF RECYCLED MATERIALS THAT CAN BE ADDED TO THE MIXTURE DURING PRODUCTION Aggregate Temperature Needed to Achieve a Mix Temperature of 260°F Su pe rh ea te d Te m pe ra tu re N ee de d fo r A gg re ga te , ° F RAP Moisture Content, % FIGURE 21 Impact of RAP moisture content and the percent of RAP on temperature (Source: After Brock and Richmond 2007).

32 the aggregate temperature needs to be increased from about 460°F (no moisture) to almost 700°F if the RAP contains 5% moisture (Brock and Richmond 2007). Separating dry- ing and preheating the recycled materials process from the aggregate drying and heating can help keep temperatures to a reasonable level and limit damage to the plant. Storage Times Storage time in silos, particularly large silos, likely facilitates diffusion of the virgin asphalt into the layer of aged and/or stiff recycled material asphalt. Longer times at elevated tem- peratures accelerate the rate of diffusion (D’Angelo et al. 2014; Rowe 2014). Although diffusion occurs, information on the impact of silo storage times and temperatures on the blending (diffusion) of virgin and recycled material asphalt was not found in the literature. Warm Mix Asphalt Used with Recycled Materials Warm mix additives have been used by some agencies and con- tractors to keep mixture temperatures down to acceptable levels when producing mixtures with recycled materials. Warm mix asphalt (WMA) helps lower temperatures necessary to super- heat aggregate, minimizes heat hardening of virgin binders, and limits overheating of RAP (Jackson 2012). The use of WMA in the production of asphalt mixtures increased approximately 26% from 2011 to 2012 (Hansen and Copeland 2013). A 2009 survey conducted by South Carolina DOT (24 respondents) showed that two states (8%) used WMA as a way to increase the percentage of RAP used in mix- tures, eight states (33%) did not specify WMA at that time, 12 states (50%) used WMA technology in conjunction with RAP mixtures, and 14 states (58%) had adopted specifications to allow for the use of WMA in general (Copeland 2011). West and Willis (2014) noted that no change in binder grade is needed if the percentage of RAP is kept between 26% and 40% and a foamed warm mix technology is used to keep the mixture temperature below 275°F. Although WMA technologies can be useful in keeping mixture temperatures at acceptable levels, no information was found in the literature about how reducing temperatures with WMA addresses the reason for the higher temperatures, which is to dry moist (or wet) recycled materials. Asphalt Plant Practices and Production— Section Summary Stockpiling Recycled Material • Covering recycled material stockpiles minimizes addi- tional moisture from rain events and heating from solar gain. – Damp, sticky, recycled material clumps adheres to belts, blinds screens, and clogs crushers, all of which make it difficult to uniformly process and feed materials into the asphalt plant. • Plant quality control (QC) personnel and loader oper- ators are critical for keeping contaminates such as dirt, rubbish, vegetation, etc., out of the stockpiles. Contaminates should be removed as soon as they are noticed. • RAP stockpiles: – Some agencies use agency-specific terms to desig- nate RAP materials from designated sources, have similar material properties, use documented QC test- ing plans, and indicate how the stockpile is built and/or maintained. – Fractionating RAP helps control RAP stockpile gra- dations and ranges of RAP asphalt content. • RAS stockpiles: – AASHTO MP78-14 requires a maximum RAS size of ³⁄8-in. (9.5-mm) sieve size. n Some agencies specify a finer grind maximum RAS size of passing ¼-in. (6.35-mm) sieve. – Ground RAS can be uniformly blended with fine aggre- gates, zeolites, or RAP to help minimize clumping. n Any material added to the ground RAS has to be accounted for in the mix design batch weights. Asphalt Plants • Additional cold feed bins or bins with improvements such as steeper side slopes, self-relieving bottoms, and moisture sensors help uniformly feed recycled materi- als into the asphalt mixture. • Conveyor belts with appropriate slopes, covered, equipped with good belt scrapers, and supported so as not to sag help keep the recycled materials from clumping, sticking, and rolling backwards or off the conveyors. • Batch plants can add higher percentages of RAP when the aggregate and RAP are combined in the bucket elevator and bypass the screens, preheat RAP prior to entering the tower, or converting to a continuous batch plant facility. • Parallel flow drum mix plants can handle higher RAP percentages with proper flighting inside the drum, mov- ing the RAP collar farther down the drum toward the discharge point, or adding a second drum for drying and preheating the recycled materials. • Counterflow and double-barrel drum designs are newer designs that can handle higher percentages of recycled materials. • Higher plant temperatures are required to superheat the virgin aggregate so that the conductive heat transfer is sufficient to dry and preheat the recycled materials when there is no separate system added to the plant for drying and preheating recycled materials.

33 – Higher moisture contents and percentages of RAP require significantly higher temperatures, which can damage both the plant and the asphalt material properties. – Higher plant temperatures use significantly more fuel (energy), which increases the cost of the asphalt mixture. PAVEMENT PERFORMANCE Six RAP studies, eight RAS demonstration projects, and two combination RAP/RAS studies that reported pavement per- formance findings, were found in the literature. These studies are briefly summarized here. RAP Pavement Performance This section summarizes performance information reported for high RAP mixtures placed in Florida, Ohio, Minnesota, Ala- bama (NCAT test track), Manitoba (Canada), and Long Term Pavement Performance (LTPP) sections around the country. Florida DOT Projects using Marshall mix designs with 30% to 50% RAP were constructed from 1991 to 1999 (Nash et al. 2011). Infor- mation for evaluating the pavement performance of these mix- tures was collected from construction reports (mix design, type of friction courses, structural layer with RAP), financial project management databases (project location, dates for start and completion), pavement management office (mix designs, tonnage), and pavement condition survey data (distress data, previous work on sections of interest, percent trucks, aver- age annual daily traffic). Cracking is the top major distress in Florida and is measured based on the visual evaluation of the pavement surface. The pavement life span was defined as the first year a defi- cient crack rating was documented. Similar mixtures without RAP were identified and used to establish a baseline for com- parisons (i.e., control sections). RAP was typically used in a lower structural layer and a non-RAP friction course upper layer section placed on the surface. The performance of the RAP mixtures was inferred based on an evaluation of the distresses and the ride quality of the surface. The final data set was separated by 30%, 35%, 40%, and 45% RAP for the initial analysis. The results showed that the performance of RAP mixtures generally decreased with the increasing percentage RAP when using the unfiltered database. When the analysis accounted for traffic volumes and only evaluated projects constructed with more than 5,000 tons of mixture with between 30% and 50% RAP, the RAP mixtures tended to perform better than the roadways without RAP mixtures. The same conclusion was reached for the projects, regardless of the type of non-RAP friction course placed over the RAP mixture. Ohio DOT In 1981, an experimental project with 25% RAP in the base course and 45% RAP in the intermediate course was built in Ohio, and after 24 years of service (2005) the RAP section compares favorably with the control section (West and Willis 2014). Minnesota DOT MnDOT performance evaluation was conducted using pave- ments with RAP in the wear courses and found that rutting was reduced when RAP mixtures were used (Johnson and Olson 2009). Approximately 32% of the projects had early cracking and 39% raveling. Most of these projects also noted construction problems that included: • Problems with RAP chunks, debris, foreign materials, crack filling materials, and spalling from shale and other soft aggregate. • “Globs” of oil and fines in the new mat. • Evidence of stiffer mixtures causing workability issues. • Asphalt content and gradation that were too variable. • Oversized material problems when mixture was used in the wear course. • Mixtures that looked grey, dry, and may require a seal coat sooner (i.e., signs of too low asphalt contents). Performance and laboratory testing of cores from eight projects showed moderate correlations between the perfor- mance ranking for the project and: • % RAP • % passing 0.15-mm sieve • % passing 0.075 mm • Dust-to-binder ratio • PG high temperature. Correlations were obtained between cracking and both the dynamic modulus master curve (middle of the frequency range) and the percentage of RAP. However, stronger cor- relations were obtained between performance and both the percentage of virgin asphalt and the PG low temperature. Alabama (NCAT Test Track) The National Center for Asphalt Technology (NCAT), located in Opelika, Alabama, operates a 1.7-mile oval track as an accel- erated loading testing facility. In Alabama, PG 67-22 is the standard grade of virgin asphalt for traffic levels of less than 10 million ESALs, and PG 76-22 for higher traffic levels is specified. Two of the 2006 NCAT test track sections evaluated

34 pavement performance differences when using a PG 76-22 and a PG 67-22 virgin asphalt in mixtures with 20% RAP. Four test sections were also constructed in 2006, each with a different PG virgin asphalt grade and 45% RAP. After more than 20 million ESALs, none of the sections had more than 5 mm of rutting; however, some traffic-related cracking was documented (Willis et al. 2009; West et al. 2011). The total length of cracking decreased with each drop in the upper PG temperature grade (Table 16 and Figure 22). Manitoba, Canada In 2009, pavement sections with 0%, 15%, and 50% RAP, with and without changing the virgin asphalt grade (Pen 150–200; Pen 200–300), were placed in two 2-in. (50-mm) lifts on a Provincial Trunk Highway 8 miles from Gimi to Hnausa in Manitoba, Canada (Hajj et al. 2011). The distresses of concern for these sections were thermal cracking and moisture dam- age. Pavement condition surveys were conducted in October 2010 and, after 13 months of service, no distresses were seen in any of the sections. Researchers believed more time was needed to determine the impact of the variables on pavement performance. LTPP SPS-5 Sections LTPP special pavement sections (SPS)-5 have 18 sites, each consisting of nine overlay test sections to compare virgin Test Section RAP Content* RAP Asphalt, % Virgin Asphalt Grade Date of First Crack ESALs at First Crack Total Length of Cracking Impact of Reducing Critical PG High Temperature (< 25% RAP) W4 20% 17.6 PG 67-22 No Cracking W3 20% 18.2 PG 76-22 4/7/2008 6,522,440 34.0 Impact of Reducing Critical PG High Temperature (>25% RAP) W5 45% 42.7 PG 58-28 8/22/2011 19,677,699 3.5 E5 45% 41.0 PG 67-22 5/17/2010 13,360,016 13.9 E6 45% 41.9 PG 76-22 2/15/2010 12,182,331 53.9 E7 45% 42.7 PG 76-22S 1/28/2008 5,587,906 145.5 Source: West et al. (2011). *RAP asphalt content as a percentage of total aggregate. **Percentage of RAP asphalt as a percentage of the total asphalt content. S = 1.5% Sasobit in virgin asphalt. TABLE 16 LOAD-RELATED CRACKING OF RAP MIXTURES PLACED AT THE NCAT TEST TRACK IN 2006 FIGURE 22 Influence of changes in virgin asphalt PG grade on traffic-related cracking (Source: After West et al. 2011).

35 asphalt mixtures with mixtures with up to 30% RAP. Over- lays were either less than 2-in. (50-mm) thick or at least 5-in. (125-mm) thick. Pavement condition survey results for ride quality, rutting, and fatigue cracking were grouped into short-term performance (0 to 5 years in service) and long-term performance (more than 5 to 10 years) for statisti- cal analyses (Wiser 2011). Over the short term (≤5 years), there were no statistical differences between virgin and RAP mixtures for 61% to 72% of the sections (Figure 23). LTPP (5 to 10 years) shows an increase in the statistical differences between the virgin and RAP mixtures. The percentage of sites with no statistical differences decreased from between 61% and 72% to between 33% and 44%. The virgin mixtures performed better than the RAP mixtures for between 33% and 50% of the sections. However, over time, 17% to 22% of the RAP mixtures showed better performance than the virgin mixtures. There were only limited statistical differences for the thin overlays compared with thick overlays (Figure 24). (a) (b) FIGURE 23 Statistical evaluation of LTPP SPS-5 sections of performance: (a) short term and (b) long term (Source: Wiser 2011).

36 RAS Pavement Performance The performance of RAS and combinations of RAP and RAS demonstration project test sections placed in eight states was recently evaluated. Information about the gen- eral performance of RAS mixtures across the country was discussed in a 2015 FHWA memorandum. This informa- tion about the performance of RAS mixtures is briefly summarized here. Pooled Fund Study The goal of the Transportation Pooled Fund (TPF) study pro- gram TPF 5(213) was to evaluate RAS grind size, percentage of RAS, RAS source (manufacturer waste, tear-offs), RAS with WMA, RAS as a fiber replacement in SMA, and RAS with ground tire rubber as an asphalt modifier on pavement performance (Williams et al. 2013). Demonstration projects were placed in Missouri (lead state), California, Colorado, (a) (b) FIGURE 24 Statistical evaluation of LTPP SPS-5 sections based on overlay thickness of mixtures with and without RAP (Source: Wiser 2011).

37 Illinois, Indiana, Iowa, Minnesota, and Wisconsin. Each demonstration project evaluated variables of importance to each state agency and the pavement condition survey results are summarized in Table 17. Pavement performance results 3 years after placement can be summarized as: • Missouri: Fine grind RAS had less transverse crack- ing than coarse grind RAS. Missouri DOT routinely uses asphalts modified with ground tire rubber and transpolyoctenamer polymer to raise upper PG temperature. • Iowa: The 0% to 5% RAS sections showed similar reflec- tive (transverse) cracking. • Minnesota: Slightly more transverse cracking in man- ufacturer waste RAS than in tear-off RAS sections; other test sections (all on shoulders) showed trans- verse cracking next to portland cement concrete (PCC) joints. • Indiana: RAS asphalt mixtures had somewhat less trans- verse cracking than either RAP or WMA–RAS mixtures. Source: After Williams et al. (2013). *Indicates values estimated from graphs in report. State Variable Transverse (thermal) Cracking (low) Transverse (thermal) Cracking (moderate to severe) Transverse Cracking (reflective) Longitudinal Cracking Raveling Comments Missouri 3 years 15% RAP 28 ln ft 15 ln ft — — — PG 64-22 blended with 10% ground tire rubber and 4.5% transpolyoctenamer rubber (TOR) 10% RAP/5% Fine RAS Post- Consumer 94 ln ft 4 ln ft — — Some 10% RAP/5% Coarse RAS Post- Consumer 123 ln ft 16 ln ft — Some — Iowa 3 years 0% RAS — — 155 ln ft* 165 ln ft — Reflective cracking at PCC joints and edge of driving lane 4% RAS — — 142 ln ft* — Some 5% RAS — — 153 ln ft* — Some 6% RAS — — 147 ln ft* — Some Minnesota 3 years 5% Post-Manufacture RAS 199 ln ft — — Transition section at MnROAD 5% Post-Manufacture RAS — — 28 ln ft* — — Shoulder, next to PCC; cracking at joint 5% Post-Manufacture RAS — — 0 ln ft — — Shoulder, next to PCC; cracking at joint 5% Post-Consumer RAS 173 ln ft — Low to High Transition at MnROAD 5% Post-Consumer RAS — — 141 ln ft — Low to High Shoulder, next to PCC; cracking at joint 5% Post-Consumer RAS — — 4 ln ft — Low to High Shoulder, next to PCC; cracking at joint 30% RAP 0 ln ft — — Shoulder Indiana 3 years HMA–RAP 112 ln ft* 78 ln ft* — 4%* Some Overlay over thick HMA over PCC HMA–RAS 85 ln ft* 55 ln ft* — 29%* Some Overlay over thick HMA over PCC WMA–RAS 198 ln ft* 77 ln ft* — 43%* Some Overlay over thick HMA over PCC Wisconsin 1 year 13% RAP/3% RAS Post-Consumer, WMA No Distresses Overlay over 4-in. HMA over PCC 13% RAP/3% RAS Post-Consumer No Distresses Colorado 1.5 years 20% RAP — — 0 — Some — 15% RAP/3% RAS Post-Manufacture — — 25 ln ft — — — Illinois 1 year PG 70-28, Polymer, 5% Post- Consumer RAS No Distresses — PG 70-28L 5% Post-Consumer RAS,SMA No Distresses — PG 58-28 Ground Tire Rubber (12%), 5% Post-Consumer RAS, SMA No Distresses — PG 70-28, Polymer, 11% RAP/3% Post-Consumer RAS, SMA No Distresses — PG 70-28L, 11% RAP/3% Post- Consumer RAS, SMA No Distresses — PG 58-28 Ground Tire Rubber (12%), 11% RAP/3% Post-Consumer RAS, SMA No Distresses — TABLE 17 SUMMARY OF PAVEMENT PERFORMANCE OF TEST SECTIONS PLACED FOR TRANSPORTATION POOLED FUND (TPF) PROGRAM TPF-5-(213)

38 Pavement performance results 1 and 1.5 years after place- ment can be summarized as: • Wisconsin: No distresses seen in test sections (13% RAP/3% RAS). • Colorado: Limited reflective cracking in the 15% RAP with 3% RAS section. • Illinois: No distresses seen in test sections. FHWA Memorandum A memorandum was issued by the FHWA Administrator for Infrastructure on December 11, 2014, about the use of RAS in new asphalt pavements (FHWA 2014). The results from the November 2014 survey of the AASHTO Subcommittee on Materials showed that at least 14 states have maximum limit for RAS of 5% by total weight of the asphalt mixtures. Most states have various other limitations based on the location of the mixture in the pavement structure, traffic levels, binder availability factors, and the type of RAS used (manufacturer waste RAS preferred). The survey results also indicated that there were only a few states with a limited number of projects that can be used for pavement performance surveys. This memorandum notes previous communications that reported increases in the number of agencies that were see- ing premature cracking in relatively new asphalt pavements using 5% RAS mixtures and requests that each division office ensure that the AASHTO PP78-14 recommendations for binder availability factors of 0.70 to 0.85 be used when there are concerns about premature cracking. The lower vul- nerability to cracking from brittleness in warmer climates was acknowledged. Combination of RAP and RAS Texas DOT In 2009, 2-inch-thick overlay test sections were placed in Houston and Austin, Texas, using mixtures with 15% RAP and 3% RAS and a PG 64-22 virgin asphalt, and the per- formance of the test sections is currently being monitored every 6 months (Zhou et al. 2013). To date, the combination RAP and RAS test sections show no signs of distress as do the control sections. Missouri DOT Field projects, each with a control and two test sections, were constructed by Missouri DOT using mixtures with a (Schroer 2009): • PG 58-22 virgin asphalt (required softer, but more expen- sive PG grade) with 20% RAP, and a combination of 5% RAS and 15% RAP. • PG 65-22 (typical grade, lower cost) with 20% RAP, and a combination of 5% RAS and 15% RAP. A wear course was placed over all of the mixtures and after 2 years of service no rutting or cracking was observed in any of the sections. After 3 years of service there was still no rutting, but cracking was starting to occur in the control section. Two cracks were seen in the 15% RAP and 5% RAS section and the standard PG 64-22. The crack- ing was attributed to an area where pavement geometry changed because of lane widening and was at the end of the concrete shoulder. Transverse cracking was seen in the center and passing lanes, but stopped at the joint adjacent to the driving lane, which contained the 15% RAP and 5% RAS layer. Pavement Performance—Section Summary High RAP asphalt pavement performance studies show the following: • Minnesota DOT study: – Performance of Minnesota roadways is related to the PG critical low temperature and the percentage of virgin asphalt in the mixture. – Projects that showed early cracking also had construc- tion problems associated with the nonuniformity of the mixture (i.e., visible deleterious materials, asphalt- fine balls, dry-looking mixtures, too-variable asphalt content, and gradation). • NCAT test track: – Decreasing the upper PG temperature reduced the impact of high percentages of RAP on traffic-related cracking without a detrimental impact on rutting. • Manitoba, Canada: – After 13 months (one winter), no thermal cracking was seen in any of the sections (0%, 15%, and 50% RAP); however, the researcher believed more time was needed for the assessment of pavement performance. • LTPP SPS-5 sections: – There were only limited differences in ride quality, rutting, and fatigue cracking between virgin and RAP mixtures (30% or less RAP) within the first 5 years of performance. – Time periods of more than 5 years are required to see statistical differences in specific pavement distresses or quality. – After between 5 and 10 years of performance, mix- tures with up to 30% RAP had similar performances compared with control sections almost half of the time (LTPP SPS-5 sections). n When there was a difference in the pavement per- formance the control sections (no RAP) performed better than the RAP sections approximately 30% of the time. n RAP sections performed better than the control sections approximately 20% of the time.

39 RAS pavement performance studies show: • Pavement condition surveys conducted fewer than about 1.5 years after construction typically show no or very limited distresses. • Most of the significant distresses witnessed in the limited RAS test sections reported in the literature at this time are related to PCC joints (reflective cracking). ECONOMICS The cost of asphalt mixtures is a function of materials, plant production, transportation, and placement. Of these four categories, the cost of materials accounts for approximately 70% of the asphalt mixture cost, and the most expensive single material is the asphalt cement (Copeland 2011; data from pre-2000 time period). Cost savings can potentially be achieved by using the asphalt in the recycled materials as a portion of the total asphalt content, because the asphalt is the single most expensive component. Material cost savings are calculated by evaluating the amount of virgin material that is saved by replacing it with recycled materials. Examples of reported material costs are shown in Table 18. Some of the cost findings found in the literature are shown in Table 19. Zhou et al. (2013) noted that the economics associated with the recycling of tear-off shingles are driven by landfill tipping fees, cost of RAS production, and the differences between virgin and recycled materials. Tipping fees can range from less than $10/ton to approximately $45/ton (Krivit 2007). The cost of processing RAS includes the manual labor costs for sorting and cleaning the raw construction debris, capital costs for processing equipment, and shingle transportation costs. Material costs and potential savings can be calculated using the following equations (Willis et al. 2012): [ ]( ) ( ) ( ) ( ) = + + = − = − −     = Cost Cost Cost Cost Cost Price AC% AC RAP% Cost Price Agg% RAP% 1 AC Cost Price RAP% mix virgin asphalt virgin aggregate RAP virgin asphalt virgin asphalt mix RAP virgin aggregate virgin aggregate virgin RAP RAP RAP Where: Costmix = Material cost for total asphalt mixture, $/ton; Materials Source: Howard et al. (2009) Source: Willis et al. (2012) Low High Aggregates Gravel $14 $26 $15 Limestone $15 $38 Coarse sand $3 $14 RAP aggregate: — — $9 Asphalt (2009 costs) $400 $500 $500 to $550 RAP Value Processed and stockpiled $15 $40 — TABLE 18 EXAMPLE OF POTENTIAL MATERIAL COST SAVINGS Time Period of Study Findings Source Pre-2000 Using 20% to 50% RAP may provide cost savings of 20% to 50% when materials and construction costs were considered. This is a potential savings of 1% of mixture cost for every 1% of RAP used. Kandhal and Mallick (1997) 2004 and 2006 Savings of about 7% to 8% with 10% RAP, 15% with 20% RAP, and 20% to 22% with 30% RAP. Vukosavlievic (2006) 2006 Using 20% RAP had the potential to save about $42 million worth of asphalt cement a year. Ontario Hot Mix Producers Association (2007) 2007 Evaluated bid costs for three projects, but found mixed results and noted more data were needed. Maupin et al. (2008) 2010 Reported Florida DOT estimates recycling program saved over $38 million in materials costs in 2010. About 78% of all Florida mixtures contained RAP (average about 20%). West and Willis (2014) 2011 Estimated savings to state of $3 to $5 a ton of mix when using between 5% and 7% of RAS (Missouri). 2012 About 5% RAS can reduce mix cost by about 13% (Texas). Combination of RAS/RAP may reduce cost by up to 20%. 2012 Material cost savings calculated as between 15% and 20% when using 30% RAP, and between 31% and 35% with 50% RAP. Willis et al. (2012) TABLE 19 EXAMPLE OF REPORTED COST SAVINGS WHEN USING RECYCLED MATERIALS

40 Costvirgin asphalt = Cost of virgin asphalt in mixture, $/ton; CostRAP = Cost of virgin aggregate in mixture, $/ton; Pricevirgin asphalt = Price of virgin asphalt, $/ton of asphalt; Pricevirgin aggregate = Price of virgin aggregate, $/ton of aggre- gate; PriceRAP = Price of RAP, $/ton of RAP; AC%mix = Total asphalt content of mix, %; AC%RAP = Asphalt content of RAP, %; RAP% = Percentage of RAP in mixture, %; and Agg% = Percentage of aggregate, %. Wet materials can increase production costs because higher temperatures are needed to dry recycled materials. One source estimates that the cost increases 13% for every 1% of moisture in the total mix (IBuildRoads™ 2012). Howard et al. (2009) documented that moisture content has a significant impact on the asphalt plant energy consump- tion (Table 20). Higher RAP moisture contents combined with higher percentages of RAP in the mixture are likely to increase plant energy usage in order to meet maximum mix- ture moisture content limits. The costs associated with milling asphalt pavement, as reported by Christman and Dunn (2013), were calculated by a North Dakota district that typically uses 20% to 24% RAP in its asphalt mixtures. Between 2008 and 2012, the milling costs were about $1,458,865. A total of 530,857 tons of asphalt mixtures with RAP were placed, with an average additional virgin asphalt content of 4.31%. Non-RAP asphalt mixtures had an average virgin asphalt content of 6.1% for the same time period. An estimated reduction of 9,521 tons of vir- gin asphalt saved was estimated, providing a net savings of $2,778,630 (net savings = cost virgin asphalt saved - milling costs). The average cost of the virgin asphalt for this time period was $445 per ton. The value engineering project option was used by three contractors in Virginia (Maupin et al. 2008). The cost savings were divided between the contractor and the Virginia DOT. Cost savings were obtained by increasing RAP from 20% to 21% (one project), and from 20% to 25% (two projects). The cost savings came from replacing virgin aggregate with RAP and from using a less costly asphalt because of an increased percentage of RAP (i.e., PG grade bump). Maupin et al. (2008) used a database with 120 projects to conduct a statistical analysis using various economic models. These models showed significant relationships between the number of tons in a plant mix line item and the number of bids received; that is, more competition results in lower bid prices. Most of the information on expected savings to the agen- cies by using recycled materials is based on simple calcula- tions for material costs. Different PG grades have different costs and using a percentage of RAP and/or RAS that requires a change of the asphalt grade can impact the material costs. For example, an increase of only 2% of RAP, from 23% RAP to 25% RAP, can change the PG grade to a lower cost asphalt and help with the mixture cost (Willis et al. 2012). Although material costs were found to be the pri- mary contributor to the asphalt mixture costs for informa- tion collected before 2000, the impact of the other three factors (plant production, transportation, and placement) on cost need to be re-evaluated (Copeland 2011). Factors such as increased costs associated with additional QC/quality assurance (QA) testing when using higher RAP contents, additional RAP processing, higher plant energy costs for superheating virgin aggregate, longer drying times (slower production rates), increased plant maintenance and equipment wear resulting from higher plant tempera- tures, and baghouse clogging, wear, and tear may shift the Moisture, % Total Energy BTU/ton Savings, % 310oF (154oC) 240oF (116oC) Change 1.0 123,769 92,874 30,895 25.0 2.0 145,991 114,708 31,283 21.4 3.0 168,212 136,541 31,671 18.8 4.0 190,433 158,375 32,058 16.8 5.0 212,655 180,209 32,446 15.3 6.0 234,876 202,043 32,833 14.0 7.0 257,098 223,877 33,221 12.9 8.0 279,319 245,711 33,608 12.0 9.0 301,540 267,545 33,995 11.3 Source: Howard et al. (2009). TABLE 20 ESTIMATED ENERGY SAVINGS BECAUSE OF A REDUCTION IN MIXTURE MOISTURE CONTENT AND/OR PLANT TEMPERATURE

41 impact on costs from materials to production (Brock and Richmond 2007). RESEARCH IN PROGRESS There is currently one set of FHWA test sections and three NCHRP studies (NCHRP 9-55, NCHRP 9-57, NCHRP 9-58) under way with research topics related to the types and percentages of RAP and/or RAS in asphalt mixtures. The Turner–Fairbank Highway Research Center is currently evaluating several test sections placed at the Accelerated Loading Facilities (ALF) to establish realistic boundaries for high RAP mixtures using WMA technologies and RAS based on the percentage of asphalt replacement and virgin asphalt grade changes (TFHRC 2014). Testing at the ALF facility should be completed by 2016. The objectives of NCHRP Project 9-55, Recycled Asphalt Shingles in Asphalt Mixtures with Warm Mix Asphalt Tech- nologies, are to, at a minimum, address: • Minimizing the risk of designing and producing mix- tures containing WMA technologies and RAS with poor constructability and durability. • Minimizing the risk of designing and producing mix- tures containing WMA technologies and RAS that are susceptible to premature failure. • Evaluating type, source, quality, and characteristics of RAS with and without RAP. • Investigating binder design and selection, including evaluation of the composite binder. • Exploring the current range of asphalt mixture produc- tion temperatures. The objectives of NCHRP Project 9-57, Experimental Design for Field Validation of Laboratory Tests to Assess Cracking Resistance of Asphalt Mixtures (2015), are to select candidate laboratory tests for load- and environment- associated cracking applicable for routine use through a literature review and workshop, and to develop an exper- imental design for field experiments to establish, verify, and validate the laboratory tests. Test methods to evaluate top-down and/or bottom-up load-related cracking, thermal cracking, and reflective cracking that are being considered include (Table 21): • Bending beam fatigue, • DSC, • IDT, • Texas overlay tester, • Repeated direct tension, • SCB at low and intermediate temperatures, • S-VECD, • TSRST, and • UTSST. Name Standard Cracking Type Specimen Geometry Cracking Parameter Criteria COV Bending Beam AASHTO T321 Bottom-up fatigue Rectangle, 15-in. length, 2.5-in. width, 2-in. thickness Number of cycles to failure; fatigue equation Pass/Fail >50% DSC ASTM D7313 Low temperature and reflective Disc, 6-in. diameter, 2-in. thickness, 2 holes (diameter 1 in.), notch depth of 2.45 in. Fracture energy Pass/Fail 10% to 15% IDT AASHTO T322 Low temperature Disc, 6-in. diameter, 1.5-in. to 2.0-in. thickness Creep compliance, tensile strength — <11% AASHTO T322 Top-down fatigue Energy ratio Pass/Fail Not reported Repeated Direct Tension Texas A&M Bottom-up and top-down fatigue Cylinder, 4-in. diameter, 6 in. tall Paris law parameters, endurance limit, healing properties, average crack size Models Low, but more work needed SCB AASHTO TP105 Low temperature Semi-circle, 6-in. diameter, 1-in. thickness, 0.6-in. notch depth Fracture energy Pass/Fail 20% SBC at Intermediate Temperatures LTRC Top-down fatigue cracking; reflective Semi-circle, 6-in. diameter, 2.5- in. thickness, 1-in., 1.25-in., and 1.5-in. notch depth Critical energy release rate Pass/Fail 20% S-VECD AASHTO TP107 Bottom-up and top- down fatigue Cylinder, 4-in. diameter, 5.1-in. tall Number of cycled; predicted number of cycles — Low, but more work needed For E*: 4-in. diameter, 6-in. tall TSRST/UTSST AASHTO TP105 (Monotonic) Low temperature Rectangle, 10-in. length, 2-in. width, 2-in. thickness Fracture temperature; coefficient of thermal contraction Pass/Fail About 10% Texas Overlay Tester Tex-249-F Reflection cracking; bottom-up fatigue Rectangle cut from gyratory; 6-in. maximum length, 3-in. width, 1.5-in. thickness Number of cycled to failure; fracture parameters A and n Pass/Fail 30% to 50% Source: NCHRP 9-58 (2015). — No information provided. TABLE 21 SUMMARY OF POSSIBLE TEST METHODS THAT WILL BE EVALUATED UNDER NCHRP 9-57 CONTRACT

42 The objectives of NCHRP Project 9-58, Effects of Recy- cling Agents on Asphalt Mixtures with High RAS and RAP Binder Ratios, are to: • Evaluate the effectiveness of recycling agents in HMA and WMA mixtures with high RAS, RAP, or combined RAS and RAP binder ratios through a coordinated pro- gram of laboratory and field experiments. • Propose revisions to several relevant AASHTO specifi- cations and test methods. • Develop training and workshop materials and deliver one workshop. The scope covers the investigation of asphalt mixtures prepared with recycling agents and RAS, RAP, or com- bined RAS and RAP at recycled asphalt binder ratios of between 0.3 and 0.5, and the performance of the bind- ers and mixtures will be evaluated. This research will be conducted on plant-mixed, laboratory-compacted speci- mens obtained from trial batches or production runs prepared in asphalt mix plants. Consistent laboratory conditioning procedures will be applied to all specimens and changes in mixture properties with aging in the field will be quantified.