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

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

14 Experimental Plan Introduction As illustrated in Figure 2-1, the research plan included sev- eral tasks generally organized into three parts. The first part included identifying field projects—constructed prior to this study—that used well-documented mixtures containing RAS and WMA. Those existing projects were inspected at roughly 1-year intervals to gather performance information and to take cores for density. Cores from the last inspections were also tested using the Illinois Flexibility Index Test (I-FIT). Table 2-1 provides general information about the three existing projects. The second part of the experimental plan involved identifying new field projects using mixtures con- taining RAS, where the mixtures would be produced at hot-mix temperatures and with a WMA technology at a reduced-production temperature. On a few of the new proj- ects, additional mixtures were produced with other variables. The second part of the plan consumed the majority of the research effort, as described in the next section. The third part of the research plan included a variety of tasks, such as preparation of a document on the best practices for process- ing RAS, an economic analysis of using RAS, and additional laboratory work to evaluate the activation of RAS binders in asphalt mixtures. The economic analysis is provided in Chap- ter 6. Other tasks from the third part are described in the appendices of the report. Production and Construction Information The five projects built and evaluated during this study are summarized in Table 2-2. For each of the projects, a control HMA section was constructed to provide a direct comparison for field performance and materials properties to mixtures containing RAS and using a WMA technology. The materials properties were also used to examine relationships between engineering properties and field performance. The mix designs for the five new projects are included in Appendix A. The research team sampled the mixtures and raw materials and collected construction data for the new projects. Documentation of the construction information for the mixtures included the items listed in Table 2-3. Materials Information. The engineer at the plant col- lected the job mix formula (JMF) and WMA dosage rate and adjustments to the mix designs. Target Mixing Temperature. The target mixing tem- perature for both the HMA and WMA was obtained from the plant operator. Delivery Temperature. Delivery temperatures were recorded every 10 min at the beginning of each paving day until the delivery temperature stabilized. Experience has shown that delivery temperature for both HMA and WMA tends to fluctuate at the beginning of each paving day for the first few truck loads or anytime the plant starts and stops. Once the delivery temperature stabilized, delivery tem- peratures were recorded hourly. Identifying the differences in delivery temperatures between the HMA and WMA was important to compare the two types of mixes. Temperature Behind the Screed. Temperature read- ings were taken with a hand-held infrared gun immediately behind the screed. Lift Thickness. The target lift thickness was obtained by the engineer at the paving site. Lift thickness measurements were obtained from cores. Densities from Cores. Cores were obtained after con- struction to determine the initial density of the pavement test sections. The cores were obtained by the engineer at the paving site, and the densities were determined at the main NCAT laboratory. C H A P T E R 2

15 Figure 2-1. Flow chart illustrating the research plan. Parameter Project 1 Project 2 Project 3 Location Fort Worth, Texas Austin, Texas Aurora, Illinois Project date 10/2012 12/2011–01/2012 06/2012–08/2012 No. of mixtures 2 4 2 Mix variables WMA versus HMA WMA versus HMA HMA HMA with PG 58-28 Two aggregate types WMA technology Cecabase Evotherm 3G na na Evotherm Road US 287 FM 973 I-88, Illinois Tollway Base binder 64-22 64-22 58-28 70-28 NMAS (mm) 9.5 19 19 % RAP 15 15 0 15 13 % MW–RAS 5 3a 5a 3a 0 % PC–RAS 0 5 Note: na = not applicable. aNo information available on type of RAS. Table 2-1. Existing projects and mix information. Mean Texture Depth. The engineer at the site conducted the sand patch test in accordance with ASTM E 965 at three locations on the finished surface. The location of the tests was recorded using a hand-held GPS. The Sand Patch Test provided the mean texture depth of the pavement. Performance Monitoring Field Performance Data Collection To collect field performance data for the projects, a mem- ber of the research team carefully reviewed the entire project length by driving and then randomly selected three evaluation sections per mix placed during construction for the new projects (or during the first field performance inspection for the existing projects). These evaluation sections were 200 ft (61 m) in length and contained the location of the original field cores taken at the time of construction. All of the field performance inspections—regardless of whether the site was a new or existing site—included detailed visual examina- tions and distress mapping of each 200-ft evaluation section to quantify the extent of cracking, rutting, raveling, patching, potholes, shoving, and bleeding. Classification of distresses was in accordance with the Distress Identification Manual for the Long-Term Pavement Performance Program (LTPP) (Miller and Bellinger 2003). Rutting was assessed by string-line

16 Parameter Project 1 Project 2 Project 3 Project 4 Project 5 Location Larsen, Wisconsin Enterprise, Alabama Oak Ridge, Tennessee Wilson, North Carolina La Porte, Indiana Project date 09/2013 06/2014 10/2014 06/2015 9/2015 No. of mixtures 3 4 2 4 2 Mix variable WMA versus HMA WMA versus HMA and Design Air Voids WMA versus HMA PC versus MW WMA versus HMA WMA technology Zycotherm and Rediset Gencor Foamer Evotherm M1 Evotherm 3G Foaming Design air voids 4.0 4.0 and 3.5 4.0 4.0 4.0 Road STH 96 US 84 TN 170 SC 42 SR 39 ESAL range (million) 1–3 1.5 NA 0.3–3 NA Base binder 58-28 67-22 64-22 58-28 64-22 NMAS (mm) 12.5 12.5 12.5 9.5 9.5 Percentage binder assumed available from RAS 75 100 100 100 100 % RAP 14 15 10 20 15 % MW–RAS 3 0 0 5a 2 % PC–RAS 0 5 3.5 5b 0 Percentage binder from RAP 12.1 13.5 10.9 18.5 NA Percentage binder from RAS 11.2 19.6 11.7 16.7 NA Note: NA = not available. a5% MW used. b5% PC used. Table 2-2. New projects and mix information. Data Collected Frequency Equipment Materials information One time na Target mixing temperature Hourly na Mix moisture content Twice per production day Oven and can Delivery temperature Hourly Temperature gun and temperature probe Temperature behind screed Hourly Temperature gun and temperature probe Lift thickness Once per day and then checked by cores na Densities from cores Seven per day Contractor or agency coring rig Mean texture depth Three locations per mix Sand and hockey puck Note: na = not applicable. Table 2-3. Field data for new projects. measurements or 6-ft (1.8 m) straight edge. Raveling was quantified by assessing changes in surface macrotexture using the Sand Patch Test (ASTM E 965). Cores were obtained from one of the randomly selected evaluation sections per mix to assess in-place densification, changes in binder absorption (calculated from maximum specific gravity tests), changes in tensile strength with time, and changes in binder properties based on recovered binder testing. Three cores were taken between wheelpaths, and three cores were taken in the right wheelpath to assess changes in density and strength. An additional core was taken between the wheelpaths to determine the change in binder properties. Table 2-4 summarizes the field monitoring activities per mix placed. Laboratory Testing of Field Mixes from New Projects Materials and Mixture Properties One of the objectives of the laboratory portion of this study was to determine materials and engineering proper- ties of WMA–RAS and control HMA–RAS. This objective was accomplished by compiling laboratory test results from

17 materials obtained from new projects where the same mix- ture was produced as HMA and WMA. Engineering properties of plant-produced WMA and HMA were used for paired statistical comparisons. The laboratory testing program evaluated recovered binder PG, mixture stiff- ness over a wide temperature range, moisture susceptibility, fatigue cracking, thermal cracking, and permanent deforma- tion as follows: • PG of extracted and recovered binder, • Mixture stiffness–dynamic modulus (AASHTO TP 79-13), • Hamburg Wheel-Tracking Test (AASHTO T 324-14), • Flow number (AASHTO TP 79-13), • Bending beam fatigue (AASHTO T 321-14), • Energy ratio, • Indirect tensile creep compliance and strength (AASHTO T 322-07), • Overlay tester (Texas DOT 248-F), • Semi-Circular Bend Test (ASTM D8044-16), and • Fracture energy and flexibility index (FI) (AASHTO TP 124-16). The following summarizes the purpose of each test selected for this study. Recovered Binder PG Binders were extracted using ASTM D2172 Method B (Centrifuge) and recovered using ASTM D5404 (Roto-vap). Binders were then tested to assess the PG of the material using AASHTO M 320. Table 2-5 shows a summary of the binder tests, output, and criteria. The following tests were conducted to determine the PG of the recovered binders according to AASHTO M 320 Stan- dard Performance Graded Asphalt Binder and AASHTO R 29 Standard Grading or Verifying the Performance Grade (PG) of an Asphalt Binder: • AASHTO T 316 Standard Viscosity Determination of Asphalt at Elevated Temperatures Using a Rotational Viscometer, • AASHTO R 28 Standard Accelerated Aging of Asphalt Binder Using a Pressurized Aging Vessel (PAV), • AASHTO T 315 Standard Determining the Rheologi- cal Properties of Asphalt Binder Using a Dynamic Shear Rheometer (DSR), and • AASHTO T 313 Standard Determining the Flexural Creep Stiffness of Asphalt Binder Using the Bending Beam Rheometer (BBR). The parameter DTc was also calculated for the recovered asphalt binders and the virgin binders. This parameter has received a great deal of interest in recent years as a potential indicator of susceptibility to age-related block cracking, based on the work by Anderson et al. (2011). DTc was calculated as the numerical difference between the low continuous-grade temperatures determined from the bending beam rheometer stiffness criterion and the m-value criterion. Throughout the Activity Section 1 Section 2 Section 3 Map cracking Measure rutting Map potholes and patches Map bleeding Measure surface texture Map shoving Obtain cores in right wheelpath 3 cores Obtain cores between wheelpaths 4 cores Evaluate windshields 1 pass Table 2-4. Field inspection activities per mix placed. Test AASHTO Method Output Criteria Rotational Viscosity T316 Viscosity (Pa-S) Viscosity ≤ 3.0 Pa-S Dynamic Shear Rheometer T315 G* (kPa a (degrees) RTFO-aged binder: G*/sin( ) ≥ 2.20 kPa PAV-aged binder: G*sin(   ) ≤ 5,000 kPa Bending Beam Rheometer T313 Stiffness (MPa) and m-value (no units) S ≤ 300 MPa m-value ≥ 0.300 PAV R28 Aged asphalt binder for further testing No criteria Note: RTFO = rolling thin-film oven. a = fitting parameter. Table 2-5. Recovered binder tests and criteria.

18 course of this project, most discussions with regard to the DTc parameter among asphalt researchers were based on tests conducted after the standard 20-h pressure aging vessel (PAV) aging for PG testing. That was how the DTc parameter was determined in this study; except for the work done as part of the mix design verifications covered in Chapter 5, which was completed in 2017 after the revised AASHTO PP 78-17 publication in which the 40-h PAV DTc was added as a criterion. Multiple Stress Creep Recovery The Multiple Stress Creep Recovery Test was conducted according to AASHTO TP 70 Standard Method of Test for Multiple Stress Creep Recovery (MSCR) Test of Asphalt Binder Using a Dynamic Shear Rheometer (DSR). This method was used to assess the binder’s resistance to perma- nent deformation. Multiple stress creep recovery is designed to identify the elastic response in a binder and the change in the elastic response at two stress levels while being subjected to 10 cycles of creep stress and recovery. The stress levels used are 0.1 kPa and 3.2 kPa. The creep portion of the test lasts for 1 s, which is followed by a 9-s recovery. Two performance parameters are used to characterize the material’s response after 10 cycles at each stress level: nonrecoverable creep com- pliance Jnr and percent recovery er. Two replicate specimens were tested at the high-temperature grade of the base binder for each binder blend. All of the binder samples were first short-term aged using the RTFO. Linear Amplitude Sweep The Linear Amplitude Sweep (LAS) Test was conducted on recovered binders to assess their resistance to fatigue cracking. This test uses cyclic loading (in shear) at linearly increasing amplitudes to accelerate fatigue damage. The rate of damage accumulation in the binder is used as an indica- tor of fatigue performance of the asphalt binder. Tests were performed at the intermediate temperature of 25°C. Two rep- licate specimens were tested for each extracted and recovered binder. All recovered binders were short-term aged using the RTFO and then tested using a frequency sweep to determine their linear rheological properties. The frequency sweep con- sists of applying a load of 0.1% ± 0.01% strain over a range of frequencies from 0.2 Hz to 30 Hz. Following the frequency sweep, the samples were subjected to a series of oscillatory load cycles in strain-controlled mode at a frequency of 10 Hz. Strain was increased linearly from 0% to 30% over the course of 3,100 cycles of loading. The binder fatigue parameter Nf was calculated using the following equation, where A and B are coefficients based on the materials properties. The analy- sis of results is based on the viscoelastic continuum damage approach used to model the fatigue behavior of asphalt binders and mixtures. LAS testing and analysis followed AASHTO TP 101. (2-1)maxN Af B( )= γ Deleterious Materials While MW–RAS is typically free of deleterious materials, PC–RAS may be contaminated with unwanted waste such as plastics, paper, wood, and metal. Texas DOT developed a method for determining the deleterious material content in processed RAS. In this method, a 1,000-g sample of RAS is poured over a specially designed pan containing a magnet across the middle of the pan designed to catch ferrous material in the RAS. The metal pieces are weighed to determine the amount of metal in the RAS. The remaining RAS is sieved over a 3⁄8-in. No. 4, No. 8, and No. 30 sieve. Material passing the No. 30 sieve is discarded. The deleterious materials retained on other sieves are then determined by manual separation and weighed by size fraction. The total percentage of deleterious material in the RAS sample is quantified using Equation 2-2. Texas requires less than 1.5% deleterious material in their processed RAS. 100 (2-2) 3 8 4 8 30 P M N N N N Wt = + + + + ∗ where P = percentage of deleterious matter by weight; M = weight of material retained by magnet (grams); N# = weight of deleterious material on sieve number (grams); and Wt = total weight of sample (grams). Mixture Stiffness Dynamic modulus testing was conducted to assess differ- ences in mix stiffness between WMA and HMA. Dynamic modulus testing was performed using AASHTO TP 79-13. The samples were prepared to 7% ± 0.5% air voids and were compacted to a height of 175 mm and a diameter of 150 mm, in accordance with AASHTO P 60-09 tolerances. Dynamic modulus testing was performed using an IPC Global Asphalt Mixture Performance Tester (AMPT), shown in Figure 2-2. Dynamic modulus testing is performed to quantify the stiffness of the asphalt mixture over a wide range of testing temperatures and frequencies. The temperatures and frequencies used for testing the mixes are those recom- mended in AASHTO PP 61-13. For this methodology, the high-test temperature is dependent on the high-PG of the base binder used in the mix that is being tested.

19 The collected data were used to generate a master curve for each individual mix. The master curve uses the princi- ple of time–temperature superposition to horizontally shift data at multiple temperatures and frequencies to a refer- ence temperature so that the stiffness data can be viewed without temperature as a variable. This method of analysis allows for visual relative comparisons to be made between multiple mixes. An example of using the time–temperature superposition principle to generate a master curve is shown in Figure 2-3. The master curve provides continuous modulus infor- mation over the entire range of temperatures and frequen- cies required for mechanistic–empirical pavement design. An equation for the master curve describing the stiffness of the asphalt mix using both interpolated and extrapolated data points at any point along the curve can be calculated. The general form of the master curve equation is shown as Equation 2-3. As mentioned, the dynamic modulus data are shifted to a reference temperature by converting testing frequency to a reduced frequency using the Arrhenius Equa- tion (Equation 2-4). Substituting Equation 2-4 into Equa- tion 2-3 yields the final form of the master curve equation, shown as Equation 2-5. The shift factors required at each temperature are given in Equation 2-6. A reference temper- ature of 20oC was used for this analysis. The limiting maxi- mum modulus in Equation 2-6 is calculated using the Hirsch Model, shown as Equation 2-7. The Pc term, Equation 2-8, is simply a variable required for Equation 2-7. A limiting binder modulus of 1 GPa is assumed for this equation. Nonlinear regression is then conducted using the Master- solver.exe program to develop the coefficients for the master curve equation. Typically, these curves have an Se/Sy term of less than 0.05 and an R2 value of greater than 0.99. Defi- nitions for the variables in Equations 2-3 to 2-8 are given in Table 2-6. 1 (2-3)Log E Max e log fr ( )∗ = ∂ + − ∂ + β+γ Figure 2-2. IPC Global Asphalt Mixture Performance Tester. 10,000.0 1,000.0 100.0 10.0 D yn am ic M od ul us (k si ) 21.1°C 4.4°C 37.8°C 54.4°C Master Curve -5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 Frequency/Reduced Frequency (Hz) Figure 2-3. Example master curve generation.

20 19.14714 1 1 (2-4)log f log f E T T r a r = + ∆ −    1 (2-5) 19.14714 1 1log E Max e log f Ea T Tr ( )∗ = ∂ + − ∂ + { }β+γ + ∆ −    19.14714 1 1 (2-6)log a T E T T a r [ ]( ) = ∆ −    4,200,000 1 100 435,000 10,000 1 1 100 4,200,000 435,000 (2-7) E P VMA VFA VMA P VMA VMA VFA max c c ( ) ( ) ( ) ∗ = − + ∗   + − − +                   20 435,000 650 435,000 (2-8) 0.58 0.58P VFA VMA VFA VMA c ( ) ( ) ( ) ( ) = + + Permanent Deformation Each of the plant-produced mixtures from the new proj- ects was evaluated for resistance to permanent deformation using the Hamburg Wheel-Tracking Test and the Flow Num- ber Test. NCHRP Project 9-43 recommended flow number testing for evaluating the permanent deformation potential of WMA during mix design. Hamburg Wheel-Tracking Test. The Hamburg Wheel- Tracking Test was performed in accordance with AASHTO T 324-14 at 50°C. Figure 2-4 shows the Cox & Sons Hamburg wheel-tracking device at NCAT used for this study. Rut depths are automatically measured for each load cycle. Testing is terminated if rutting exceeds ½ in. Figure 2-5 illustrates typical data output from the Hamburg Wheel- Tracking Test. These data show the progression of rut depth with number of cycles. From this plot, two tangents are evident: the steady-state rutting portion of the plot and the portion of the plot after stripping. The intersection of these two tangents defines the stripping inflection point of the mixture. Many state DOTs that require the Hamburg Wheel-Tracking Test have criteria for both total rutting and the stripping inflection point (Table 2-7). NCAT established a good correlation between Hamburg Wheel-Tracking Test results and rutting on the test track from which a maxi- mum Hamburg rut depth criterion of 8 mm at 20,000 passes was recommended for heavy traffic pavements (West et al. 2012). A stripping inflection point (SIP) of greater than 10,000 passes has been shown to be a good indicator of a moisture-resistant mix (Kvasnak et al. 2010). However, there is scant published research to validate SIP criteria for assessing moisture-damage susceptibility. Flow Number. The determination of the flow number for the mixtures was performed using an AMPT, in accor- dance with AASHTO TP 79-13. Flow Number tests were conducted on dynamic modulus specimens at the 50% reli- ability temperature for the project location at a depth of 20 mm, based on LTPPBind Version 3.1. Additionally, the specimens were tested using a deviator stress of 87 psi with- out confinement, as recommended by NCHRP Report 673 Variable Definition |E*| Dynamic modulus, psi Fitting parameters Max Limiting maximum modulus, psi fr Reduced frequency at the reference temperature (Hz) f Loading frequency at the test temperature (Hz) Ea Activation energy (treated as a fitting parameter) T Test temperature (°K) Tr Reference temperature (°K) a(T) Shift factor at temperature, T |E*|max Limiting maximum HMA dynamic modulus (psi) VMA Voids in mineral aggregate (%) VFA Voids filled with asphalt (%) Table 2-6. Master curve equation variable descriptions. Figure 2-4. Hamburg wheel-tracking device.

21 and NCHRP Report 691 for HMA and WMA, respectively (Table 2-8). In the Flow Number Test, the AMPT applies a repeated compressive loading to an asphalt specimen and records the deformation of the specimen at each cycle. The user defines the temperature, applied stress state (deviator stress and confining stress), and the number of cycles at which the test is performed. The loading is applied for 0.1 s, fol- lowed by a 0.9-s rest period every 1-s cycle. The tests were conducted until either the samples reached 5% axial strain (7.5 mm of deformation on a 150-mm sample) or the test completed 20,000 cycles. Flow number data are commonly fit with the Francken Model, shown as Equation 2-9. An example of unconfined Flow Number Test data is shown in Figure 2-6. 1 (2-9)N aN c ep b dN( )( )ε = + − where ep(N) = permanent strain at N cycles, N = number of cycles, and a, b, c, d = regression coefficients. Figure 2-5. Example of Hamburg Wheel-Tracking Test data analysis. High-Temperature Binder Grade Minimum Passes to 0.5-in. Rut Depth PG 64 or lower 10,000 PG 70 15,000 PG 76 or higher 20,000 Source: Advanced Asphalt Technologies, LLC 2011, Bonaquist 2011. Table 2-7. Texas requirements for Hamburg Wheel-Tracking Test. Traffic Level (million ESALs) NCHRP Report 673 (HMA) NCHRP Report 691 (WMA) <3 na na 3 to <10 53 30 10 to <30 190 105 740 415 Note: na = not applicable. Table 2-8. Recommended minimum flow number criteria.

22 Flow number is defined as the number of cycles at which the sample begins to rapidly deform and coincides with the minimum rate of strain accumulation measured during the test. This is more properly defined as the breakpoint between steady-state rutting (secondary rutting) and the more rapid failure of the specimen (tertiary flow). Figure 2-6 illustrates this concept graphically. If a sample does not exhibit tertiary flow (common for confined samples), then the amount of deforma- tion at a specified loading cycle can be used to give a relative ranking of tested mixes with respect to rutting susceptibility. Cracking Resistance Six test procedures were conducted to evaluate and com- pare cracking resistance of RAS mixtures with and without WMA: the Bending Beam Fatigue Test, the Energy Ratio Test, the Texas Overlay Test, the I-FIT, the intermediate tempera- ture Semi-Circular Bend Test, and the Indirect Tensile Creep Compliance and Strength Test. A summary of these tests is presented below. Bending Beam Fatigue Test. Bending Beam Fatigue test- ing was performed in accordance with AASHTO T 321-14 to determine the fatigue resistance and endurance limits of each mixture. Six beam specimens were tested for each mix. Within each set of six, three beams were tested at strain levels of 250 and 600 microstrain to assess the fatigue resistance at lower and higher strain levels. The specimens were compacted in a knead- ing beam compactor and were then trimmed to the dimensions of 380 ± 6 mm in length, 63 ± 2 mm in width, and 50 ± 2 mm in height. The beam fatigue apparatus, shown in Figure 2-7, applies haversine loading at a frequency of 10 Hz. During each cycle, a constant level of strain is applied to the bottom of the speci- men. The loading device consists of four-point loading and reaction positions that allow for the application of the target strain to the bottom of the test specimen. Testing was per- formed at 20°C ± 0.5°C. The data acquisition software was 100,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 M ic ro st ra in Figure 2-6. Typical Flow Number Test data. Figure 2-7. IPC Global beam fatigue testing apparatus.

23 used to record load cycles, applied loads, and beam deflec- tions. The software also computed and recorded the maxi- mum tensile stress, maximum tensile strain, phase angle, beam stiffness, dissipated energy, and cumulative dissipated energy at user-specified load cycle intervals. At the beginning of each test, initial beam stiffness was cal- culated by the data acquisition software after 50 conditioning cycles. Based on the collected data, the value of normalized modulus × cycles was calculated using Equation 2-10 to help interpret the point of failure. The failure point of the beam occurs at the maximum point on a plot of normalized modu- lus × cycles versus number of testing cycles. An example of this type of plot is shown in Figure 2-8. This also corresponds to a sudden reduction in stiffness of the specimen. Given the cycles to failure for two different strain levels, the fatigue limit was then calculated for each mixture. (2-10)NM S N S N i i o o = × × where NM = normalized modulus × cycles, Si = flexural beam stiffness at cycle i, Ni = cycle i, So = initial flexural beam stiffness (estimated at 50 cycles), and No = actual cycle number where initial flexural beam stiff- ness is estimated. Using the procedure developed in NCHRP Report 646 (Proj- ect 09-38), the endurance limit for each mixture was estimated using Equation 2-11, based on a 95% lower prediction limit of a linear relationship between the log–log transformation of the strain levels and cycles to failure. All of the calculations were conducted using a spreadsheet developed in NCHRP 09-38. _ ˆ 1 1 (2-11)0 0 2 Endurance Limit y t s n x x Sxx ( ) = − + + −α where ŷo = log of the predicted strain level (microstrain), tα = value of t distribution for n-2 degrees of freedom = 2.131847 for n = 6 with α = 0.05, s = standard error from the regression analysis, n = number of samples = 6, Sxx = ∑( )− = 2 1 x xi i n (Note: log of fatigue lives), xo = log (50,000,000) = 7.69897, and x– = log of average of the fatigue life results. Energy Ratio. The Energy Ratio Test procedure was developed to assess an asphalt mixture’s resistance to top- down cracking (Roque et al. 2004). The energy ratio is deter- mined using a combination of three tests: Resilient Modulus, Creep Compliance, and Indirect Tensile Strength. These tests were performed at 10°C using an MTS Servo-Hydraulic Test system. The tests were conducted on three specimens of 150-mm diameter by approximately 38-mm thick and cut from gyratory compacted samples (Figure 2-9). The target air voids for the specimens was 7.0% ± 0.5%. The resilient modulus is obtained by applying a repeated haversine waveform load in load control mode. The load is 0 20 40 60 80 100 120 140 160 0 5,000 10,000 15,000 20,000 25,000 Cycles N o rm al iz ed S ti ff n es s * C yc le s Fatigue Life Figure 2-8. Sample plot of normalized modulus ë cycles versus number of cycles.

24 applied for 0.1 s and then followed by a 0.9-s rest. The resilient modulus is calculated using the stress–strain curve, as shown in Figure 2-10a. The Creep Compliance Test was performed as described in AASHTO T 322-07; however, the temperature of the test was 10°C with test duration of 1,000 s. The power- function properties of the Creep Compliance Test can be deter- mined by curve-fitting the results obtained during constant load control mode (Figure 2-10b). Finally, the tensile strength and dissipated creep strain energy at failure (DCSEf) are deter- mined from the stress–strain curve of the given mixture during the Indirect Tensile Strength Test (Figure 2-10c). Detailed testing procedures and data interpretation meth- ods for the three tests are described elsewhere (Roque and Buttlar 1992, Roque et al. 2004). Data analysis was performed using a software package developed at the University of Florida. The results from these tests were used to calculate the energy ratio using Equation 2-12. According to the devel- opers of the method, a higher energy ratio indicates that the mixture is more resistant to load-related top-down cracking. Table 2-9 lists the recommended thresholds for the energy ratio as a function of rate of traffic. 7.294 10 6.36 2.46 10 (2-12) 5 3.1 8 2.98 1 ER DSCE S m D f t[ ]( )= × × σ − + × − − − where s = tensile stress at the bottom of the asphalt layer, 150 psi; D1, m = power function parameters; St = tensile strength; DSCEf = dissipated stress creep energy at failure; and ER = energy ratio. Overlay Tester. The Overlay Test was performed in accordance with Tex-248-F using the original Overlay Test Figure 2-9. Energy Ratio Test specimen setup. (a) (b) (c) Figure 2-10. Stress–strain graphs for the calculation of energy ratio from (a) Resilient Modulus, (b) Creep Compliance, and (c) Indirect Tensile Strength testing. Traffic (ESALs/year) Minimum Energy Ratio <250,000 1 <500,000 1.3 <1,000,000 1.95 Source: Roque et al. 2004. Table 2-9. Recommended energy ratio criteria.

25 kit designed for the IPC Global Asphalt Mixture Performance Tester. For this test, samples with a 150-mm diameter and a 125-mm target height were produced using the Superpave Gyratory Compactor. Two test specimens from each gyratory sample were trimmed to the following dimensions: 150-mm long by 76-mm wide by 38-mm tall. The trimmed specimens were glued to two steel platens with a 2.0-mm gap between the platens. Four replicates with air voids between 6% and 8% after trimming were tested at 25°C in controlled displace- ment mode. Loading occurs when one platen translates away from the other platen. Tex-248-F specifies a maximum dis- placement of 0.025 in. Loading occurs at a rate of one cycle every 10 s with a sawtooth waveform. The maximum load the specimen resists in controlled displacement mode was recorded for each cycle. The test continues until the sample fails. Failure is defined as 93% reduction in load magnitude from the first cycle. There is no national standard pass–fail criterion for the Overlay Test, with minimum recommended cycles to failure at the above parameters ranging between 150 and 700 cycles to failure, depending on the type of mixture tested (Sheehy 2013, Chen 2008, and Zhou et al. 2007). Fracture Energy and Flexibility Index. I-FIT was devel- oped in 2015 by the Illinois Center for Transportation at the University of Illinois at Urbana–Champaign (Al-Qadi et al. 2015). The test was developed for the Illinois DOT for use in mix design to assess the cracking resistance of asphalt mix- tures. The test method was first established as Illinois Test Procedure 405. A similar procedure was balloted by AASHTO as a provisional standard in 2016 (AASHTO TP 124). In the laboratory experiment part of this study, semi- circular bend specimens were prepared from 160-mm tall by 150-mm diameter Superpave Gyratory Compactor specimen compacted to 7.0% ± 0.5%. Four replicate semi-circular bend specimens were obtained per large Superpave Gyra- tory Compactor specimen. A 15-mm-deep by 1.5-mm-wide notch was cut into each semi-circular bend specimen along the center axis of the specimen. The specimens were tested at 25.0°C ± 0.5°C after being conditioned in an environmental chamber for 2 h. Specimens were loaded monotonically at a rate of 50 mm/min until the load dropped below 0.1 kN after the peak was recorded. Force and actuator displacement were recorded at a rate of 50 Hz by the system. Figure 2-11 shows the Test Quip I-FIT device at NCAT. Cores obtained at the last field inspection for the new projects were also tested using the Illinois Test Procedure 405. The collected data are used to calculate the fracture energy and the flexibility index for each specimen. The fracture energy (Equation 2-13) represents the area under the stress– strain curve normalized for the specimen dimensions. It is calculated by integrating the area under the load–displacement curve and dividing it by the ligament area (the area of the semicircular specimen through which the crack will propa- gate). To calculate the flexibility index (FI in Equation 2-14), the slope of the post–peak portion of the curve must be determined. This is the maximum slope of the curve imme- diately after the peak. The flexibility index is calculated by dividing the fracture energy by the post–peak slope and then multiplying that quotient by a scaling factor. In general, a higher fracture energy is indicative of a mix with better cracking resistance. A higher flexibility index is indicative of a mix exhibiting a more ductile failure, while a lower flex- ibility index indicates a more brittle failure. Data analysis can be performed using software developed by Illinois Center for Transportation. An example of processed I-FIT data from this software is shown in Figure 2-12. (2-13)G w a f f lig = (2-14)FI G m A f= × where Gf = fracture energy (J/m2), wf = work of fracture (J), alig = ligament area (mm2) = (specimen radius – notch length) × specimen width, FI = flexibility index, m = post–peak slope (kN/mm), and A = scaling factor (0.01 for gyratory specimens). Semi-Circular Bend Test (ASTM D8044-16). The Loui- siana Transportation Research Center (LTRC) developed the Semi-Circular Bend Test (Wu et al. 2005). Gyratory specimens of 150-mm diameter, were compacted to a height of 57 mm and an air void content of 7% ± 0.5%. The circular specimens Figure 2-11. Test Quip I-FIT device.

26 were then cut along the center diameter of the specimen, yielding two semicircular halves. To reduce the effect of mix- ture variation between specimens, each half of the specimen was designated with the same notch depth (i.e., 25.4 mm, 31.8 mm, or 38.1 mm). Four semicircular specimens were tested at each notch depth. The tolerance on the notch depth is ±1.0 mm, while the notch width is 3.0 ± 0.5 mm. As shown in Figure 2-13, semi-circular bend samples were symmetrically supported by two fixed rollers with a span of 120 mm. Teflon tape was used to minimize friction between the specimen and the rollers. Load was applied on the top of the specimens at a very slow constant vertical displacement rate of 0.5 mm per minute. The plot of the load versus the displacement was used to compute the area under the curve to the peak load. Figure 2-14 shows typical load–vertical deflection curves obtained in the Semi-Circular Bend Test at three nominal notch depths of 25.4 mm, 31.8 mm, and 38.0 mm. The area under the loading portion of the load–deflection curves—up Displacement (mm) Figure 2-12. Example of processed I-FIT data using University of Illinois at Urbana– Champaign Illinois Center for Transportation IL–Semi-Circular Bend analysis tool. to the maximum load—is determined for each notch depth of each mixture. This area represents the strain energy to failure U. The average values of U at each notch depth are plotted versus notch depth to obtain the slope of U versus notch depth, as illustrated in Figure 2-15. This slope is the value of dU/dA in Equation 2-15. The Jc parameter is computed by dividing dU/dA by the average specimen width b. In a recent publication, the LTRC proposed performance-based specifications for perma- nent deformation and cracking (Mohammad et al. 2016). The proposed performance test criteria are based on two traf- fic levels, as indicated by Mohammad et al. (2016). For Traffic Level 1, the minimum Jc value is 0.5; and for Traffic Level 2, the minimum Jc value is 0.6. In this study, an MTS Servo-Hydraulic testing system equipped with an environmental chamber was used to perform the Semi-Circular Bend tests at 25°C. 1 (2-15)J b dU dA c ( )= −

27 Figure 2-13. Semi-Circular Bend Test. Figure 2-14. Typical plot of load versus load line displacement. where b = sample thickness, A = notch depth, U = strain energy to failure, and dU/dA = slope of the linear regression between area to peak load versus notch depth. Indirect Tension Creep Compliance and Strength. For a given mix, a critical cracking temperature exists where the estimated thermal stress from pavement contraction exceeds the tested indirect tensile strength of the mixture. This type of analysis is referred to as a “critical temperature analysis.” A mixture exhibiting a critical cracking temperature below the expected low temperature for the project environment is desired. In this study, the IDT system was used to collect the necessary data for the critical cracking temperature analy- sis. The testing was conducted using an MTS load frame equipped with an environmental chamber capable of maintaining the low temperatures required for this test. Creep compliance was measured at 0°C, −10°C, and −20°C and tensile strength at −10°C in accordance with AASHTO T 322. Four samples were prepared for each mix. The first sample was used to find a suitable creep load for that particular mix at each testing temperature. The remain- ing three samples were tested at this load. Specimens used for the creep and strength tests were prepared to 7% ± 0.5% air voids. The IDT data were analyzed using the low- temperature stress spreadsheet developed by Christenson (2013). Creep compliance master curves, estimated thermal stress models, and the critical low-temperature cracking temperatures were compared between mixtures. The critical low-temperature cracking temperatures are determined as the minimum temperature at which the asphalt mixture is flexible enough to relieve the accumulating thermal stresses caused by thermal shrinkage. Below this critical tempera- ture, pavement is expected to begin to develop thermal cracking.

28 Figure 2-15. Typical plot specimen notch versus area to peak load. Summary of Laboratory Performance Testing A comprehensive suite of laboratory tests was conducted to evaluate the mix properties of plant-produced RAS mix- tures with and without WMA. The results of all tests were used to compare the engineering properties of WMA to those of HMA. Table 2-10 provides a summary of the testing for each of the new projects. Mix Design Verifications Another aspect of the work plan for this study included verifications of the mix designs from the five new projects and assessments of whether the verified mix designs satisfied the criteria in the recently revised AASHTO PP 78-17 Standard Practice for Design Considerations When Using Reclaimed Asphalt Shingles (RAS) in Asphalt Mixtures. Over the course of this project, the AASHTO standards regarding the use of RAS in asphalt mixtures changed substantially. It is expected that the standards and criteria will continue to evolve as more information is brought forward through additional research and field performance assessments. For the mix design verifications, the target asphalt content and gradations were based on the field-produced mixtures. Since changes in gradations during plant production affect the volumetric properties, the average measured field grada- tion for a project was used as the target for the laboratory verification instead of the target gradation from the JMF. The binder content were based on extraction tests according to AASHTO T 164. Laboratory trial samples were batched, and their gradation was determined according to AASHTO T 11 and T 27. Adjustments were made as necessary to match field production results. Table 2-10. Summary of mix performance tests. Test Equipment Replicates Dynamic Modulus (AASHTO TP 79-13) AMPT 3 specimens per mix Bending Beam Fatigue (AASHTO T 324-14) BBF device 6 specimens per mix Energy Ratio MTS 4 specimens per mix Overlay (Texas DOT 248-F) AMPT 4 specimens per mix Illinois Flexibility Index (AASHTO TP 124-16) I-FIT device 4 specimens per mix Semi-Circular Bend (ASTM D8044-16) MTS 12 specimens per mix Flow Number (AASHTO TP 79-13) AMPT 4 specimens per mix Hamburg Wheel-Tracking (AASHTO T 324-14) Hamburg wheel- tracking device 2 twin sets per mix IDT Creep Compliance and Strength for Thermal Cracking (AASHTO T 322-07) MTS 3 specimens per mix

29 Additional Testing and Analysis of Mixtures Containing RAS In addition to the primary experiments focused on the production, field performance, and engineering properties of plant-produced HMA and WMA mixtures containing RAS, this project also explored research along three other areas to gain a better understanding of the behavior of RAS in asphalt mixtures. The first study explored the use of differential scan- ning calorimetry to determine thermal energy required to iden- tify the phase change (melting point) of RAS binder so that an appropriate mixing temperature could be established. The sec- ond study used scanning electron microscopy (SEM) to detect how well RAS was dispersed in an asphalt mixture. These two studies and their findings are reported in Appendix B. The third area of additional research included a series of experiments to evaluate the activation of shingle asphalt and assess the contribution of the individual components within RAS on the properties of asphalt mixtures. In the first experiment, the effect of mixing temperature on the activa- tion of shingle asphalt was investigated by evaluating a mix- ture containing 5% RAS at six mixing temperatures ranging from 225°F to 350°F. The effects of temperature on volumet- ric properties and mechanical properties were examined. In the second experiment, the components of RAS (RAS binder, granules, and fibers) were separated and added individually to a control mixture to make four separate mixtures. Volu- metric properties and selected laboratory performance tests were conducted on the four mixtures to assess the impacts of the different RAS components. In the third and final experi- ment, mixture performance tests were conducted on three plant-mixed–laboratory-compacted (PMLC) mixtures placed on Lee County Road 159 in Alabama in 2012 as part of a pavement preservation study. The performance tests included dynamic modulus, a modified version of the Texas overlay, indirect tension creep compliance and strength, and energy ratio. A summary of this research is provided in Appendix C.