Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories (2020)

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.

9 C H A P T E R 2 This chapter presents results of the literature review of practices for fabricating specimens for performance testing in laboratories and covers a discussion of various asphalt performance tests and a review of the impact of specimen fabrication practices on performance test results. Topics addressed as part of the specimen fabrication practices include sampling location, mixture type, conditioning protocol (aging and moisture), laboratory compaction methods, specimen parameters (specimen size and geometry or specimen air voids), shelf life, and anisot- ropy. The information presented in this chapter was gathered from published technical papers and reports as well as several DOT manuals for specimen fabrication. The NCHRP has also funded research projects that cut across some of the specimen fabrication practices; these were extensively reviewed and are presented as part of this chapter. Performance Tests for Asphalt Mixtures In this subsection, the most common asphalt-mixture performance tests currently employed by both agency and partner laboratories are briefly discussed. Stiffness and Viscoelastic Characterization Dynamic (Complex) Modulus Testing (E*) (AASHTO T 342) The dynamic (complex) modulus test is used to characterize the viscoelastic properties of asphalt mixtures using an asphalt-mixture performance tester (AMPT). In recent years, its use is becoming more widespread as it is implemented as a reliable method for character- izing stiffness and load resistance properties of asphalt mixtures and is used as an input in mechanistic-empirical pavement design. The dynamic modulus is determined by applying sinusoidal axial loads to cylindrical test specimens at varying frequencies and temperatures. These test specimens are typically 100 mm (4 in.) in diameter and 150 mm (6 in.) in height. However, more recently there has also been use of smaller cylindrical specimens having a diameter and height of 38 × 135 mm (1.5 × 5.3 in.), 50 × 135 mm (2 × 5.3 in.), 38 × 110 mm (1.5 × 4.3 in.), or 50 × 110 mm (2 × 4.3 in.) (Diefenderfer et al. 2015). The specimens are cored out of laboratory-compacted samples, with a saw used to shave off the ends (or faces) to ensure they are smooth and parallel. The specimen can also be fabricated from field cores, and for this the smaller cylindrical specimens are typically used because pavement lift thicknesses are usu- ally less than 100 mm (4 in.). Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories

10 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories Resilient Modulus Testing The resilient modulus (MR) test (ASTM D7369) is used to characterize the stiffness and load resistance properties of asphalt mixtures. It is the primary test employed by most state DOTs still utilizing the AASHTO 1993 pavement design, evaluation, and analysis. This test is con- ducted by applying a repeated haversine compressive load of a fixed magnitude, load duration, and cyclic duration to the vertical diameter of a cylindrical testing specimen, and the resilient (recoverable) deformation of the specimen is measured in the vertical and horizontal direc- tions. The specimen dimensions are 101.6 ± 3.8 mm (4 in.) or 152.4 ± 9 mm (6 in.) in diameter and between 38.1 mm (1.5 in.) and 63.5 mm (2.5 in.) in thickness. It is conducted at a test temperature of 25 ± 1°C (77 ± 2°F). Cracking Direct Tension Cyclic Fatigue Test The direct tension cyclic fatigue (DTCF) test (AASHTO TP 107) is used to characterize the fatigue behavior of asphalt mixtures by determining the damage characteristic relationship using Simplified Viscoelastic Continuum Damage (S-VECD) theory. The testing is conducted at a temperature based on the binder performance grade (PG) given as Test temperature C high PG temperature low PG temperature 2 3 (1)( )° = + −  The test is conducted by applying repeated tensile loads at 10 Hz under different strain amplitudes to cylindrical specimens typically 100 mm (4 in.) in diameter and 130 mm (5 in.) in height but can also be done on a small specimen geometry (SSG) that is 38 mm (1.5 in.) in diameter and 110 mm (4.3 in.) tall. Test specimen fabrication requires coring the required dimension from laboratory-compacted samples or field cores and gluing of platens to both ends of the specimen. Flexural Beam Fatigue Test The flexural beam fatigue (FBF) test (AASHTO T 321) is used to evaluate the fatigue life of asphalt mixtures under cyclic haversine four-point flexural loading at different strain levels. The frequency of the loading ranges from 5 to 10 Hz. Testing is conducted at a temperature of 20.0 ± 0.5°C (68.0 ± 1°F). Test specimens are beams 380 mm (15 in.) long by 50 mm (2 in.) thick by 63 mm (2.5 in.) wide. These beam specimens are sawed from laboratory or field-compacted asphalt mixtures. Cracking Tolerance Index Test The cracking tolerance index or CTIndex (formerly known as IDEAL-CT) test (ASTM D 8225) is used to characterize the cracking tolerance of asphalt mixtures. It is run at interme- diate temperature [25°C (77°F) or PG intermediate temperature] on a cylindrical specimen of 150 mm (6 in.) diameter and 62 mm (2.5 in.) height at a loading rate of 50 mm/min (2 in./min). Mixtures with nominal maximum aggregate size (NMAS) ≥ 25 mm and a height of 95 mm (3.75 in.) is specified. It is a simple test that requires no instrumentation, cutting, gluing, drilling, or notching of specimens. The standard suggests that all the specimens should be compacted to the same level of air voids (e.g., 7 ± 0.5%). Superpave Indirect Tensile Strength and Creep Test The Superpave indirect tensile (IDT) strength and creep test (AASHTO T 322) is used to determine the tensile creep compliance at different loading times, tensile strengths, and

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 11 Poisson’s ratios of asphalt mixtures. The tensile creep and strength have been used to estimate thermal cracking potential. It can also be used to determine potential for moisture damage where moisture-conditioned and unconditioned specimens are evaluated to obtain the tensile strength ratio (TSR). It is conducted using cylindrical specimens that are typically laboratory fabricated or field cores. For laboratory-compacted specimens, a minimum specimen height of 50 mm (2 in.) is required for specimens with a nominal diameter of 100 mm (4 in.), and a mini- mum specimen height of 75 mm (3 in.) is required for specimens with a nominal diameter of 150 mm (6 in.). For field cores, a minimum height of 38 mm (1.5 in.) is specified for specimens with a nominal diameter of 100 mm (4 in.). Texas Overlay Test The Texas Overlay test (TxOT) (TxDOT Designation: TEX-248-F) is used to evaluate the resistance of asphalt mixtures to fatigue or reflective cracking. The goal of the test is to obtain the critical fracture energy and cracking resistance index which are used to characterize crack- ing resistance. This test is conducted by applying repeated direct tension load to cylindrical- laboratory-molded or field-cored specimens that are trimmed perpendicular to the top surface on two parallel sides. The test is performed at a constant temperature of 25 ± 0.5°C (77 ± 1°F). The cylindrical specimen is specified to have diameter of 150 mm (6 in.) and height of 115 ± 5 mm (4.5 ± 0.2 in.) before trimming. The actual test specimens after trimming are required to be 76 ± 0.5 mm (3 ± 0.02 in.) wide and 38 ± 0.5 mm (1.5 ± 0.02 in.) high. The test also requires that test specimens are glued at the bottom to a metal plate before being mounted on the tester. A repeated tensile load is applied until the maximum load is reduced by 93% of the maximum load during the first cycle, indicating that cracking has developed and the specimen is considered failed. Disk-Shaped Compact Tension Test The disk-shaped compact tension (DCT) test (ASTM D 7313) is used to determine the fracture resistance of asphalt mixtures at cold in-service temperatures of 10°C (50°F) or below. The disk-shaped test specimens are prepared from 150 mm (6 in.) diameter samples compacted in the laboratory or from field cores. The test specimen is fabricated to have a thickness of 50 ± 5 mm (2 ± 0.2 in.), a 62.5 ± 5 mm (2.5 ± 0.25 in.) notch along the diameter of the specimen not wider than 1.5 mm (0.06 in.), a flat surface cut at 90 ± 5° to the notch, and two 25 ± 1.0 mm (1 ± 0.04 in.) loading holes on opposite sides of the notch. The test is conducted by loading the specimens in tension to evaluate fracture energy. Semicircular Bend Test The semicircular bend (SCB) test (AASHTO TP 124, ASTM D8044, AASHTO TP 105) has been used to measure cracking resistance of asphalt mixtures both at intermediate tempe ratures following the AASHTO TP 124 or ASTM D8044 procedures and low temperatures following the AASHTO TP 105 procedure. All procedures have similar specimen geometry, fabricated by cutting laboratory-prepared or field core disks in half and notching parallel to the loading and vertical axis. However, the specimen thickness and notch length vary based on the specific procedure. The test methods involve loading the semicircular specimen to failure at a constant rate of deformation under three-point bending. Rutting, Moisture Susceptibility, and Durability Asphalt Pavement Analyzer Test The asphalt pavement analyzer (APA) test (AASHTO T 340) is used to evaluate permanent deformation (rutting), fatigue cracking, and moisture susceptibility of asphalt mixtures. There is

12 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories also an APA junior that is a smaller version of the APA. The APA apparatus applies 8,000 repeti- tive linear wheel loads of 445 ± 422 N (100 ± 5 lb) to test specimens via a hose at a pressure of 690 ± 35 kPa (100 ± 5 psi). It has an environment-controlled chamber that can be used for test- ing in dry or submerged conditions; the test is typically conducted at the PG high temperature. The test is conducted at the same time on three rectangular slabs with dimensions of 75 mm × 125 mm × 300 mm (3 in. × 5 in. × 11.8 in.) or six cylindrical specimens 150 mm (6 in.) diameter and 75 mm (3 in.) height which can be fabricated from field cores or laboratory-compacted samples. Average rut depth is reported. Repeated Load Permanent Deformation Flow Number Test The repeated load permanent deformation (RLPD) flow number test (AASHTO T 378) is conducted using the AMPT and is used to evaluate rutting performance of asphalt mixtures. This test involves application of a specific stress level in a haversine waveform for 0.1 second, followed by a rest or dwell period of 0.9 second to obtain the permanent strain. The test is typi- cally conducted at the high pavement temperature selected by the agency. The flow number is defined as the number of load cycles corresponding to the minimum rate of change of perma- nent axial strain. Test specimens are typically fabricated to be 100 mm (4 in.) in diameter and 150 mm (6 in.) in height, but can also be done on an SSG that is 38 mm (1.5 in.) in diameter and 110 mm (4.3 in.) tall. These test specimens can be produced from roadway cores or laboratory- compacted samples. Hamburg Wheel Tracking Test The Hamburg Wheel Tracking (HWT) test (AASHTO T 324) is used to test asphalt mixtures for rutting and moisture susceptibility. The test is typically conducted on specimens submerged in a water bath at temperatures ranging from 40 to 50°C (104 to 122°F). The apparatus applies repetitive loads (52 ± 2 passes per minute) of 705 ± 4.5 N (158 ± 1.0 lb) to slab or cylindrical specimens using a steel wheel. Slab specimens 320 mm (13 in.) long and 260 mm (10.25 in.) wide and thickness ranging from 38 mm (3 in.) to 100 mm (4 in.) may be used. Alternatively, two cylindrical specimens may be fabricated to have diameter of 150 mm (6 in.) and thickness ranging from 38 mm (3 in.) to 100 mm (4 in.). The cylindrical specimens are required to be cut at the edges along a secant line so that they can be joined together with no space between the cut edges. The test concludes after either 20,000 passes or when a maximum rut depth (established a priori) is reached. Cantabro Test The Cantabro test (Tex-245-F, AASHTO TP 108-14, ASTM D 7064-08), as specified by the Texas DOT, is used to evaluate abrasion loss of asphalt specimens using the Los Angeles Abrasion Machine. The significance of the test is to determine the durability of a mixture and strength of bond between binder and aggregates. It is conducted by rotating the compacted specimen of diameter 150 mm (6 in.) and height 115 ± 5 mm (4.5 ± 0.2 in.) for 300 revolu- tions and measuring the mass of material before and after. This test is conducted at a test temperature of 25 ± 1°C (77 ± 2°F) on specimens fabricated from laboratory-compacted specimens. Research on the Impact of Specimen Fabrication Practices on Performance Test Results In this subsection, literature on the impact of various specimen fabrication practices relating to sampling, specimen fabrication methods, conditioning procedures, compaction methods, test specimen parameters, and shelf life are summarized.

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 13 Sampling State DOTs employ various sampling techniques for quality assurance. The goal is to get an uncontaminated, representative sample of the final product in the field. Hassan (2002) performed a forensic study for Colorado DOT and found that differences of no more than 3% existed in asphalt content of samples obtained from different locations (at the plant with a tube sampler, at the point of delivery, and behind the paver) and that none of these differ- ences were statistically significant. Turner and West (2006) evaluated the effect of sampling locations (truck by shovel, truck by robotic device, behind the spreader, and cores) on asphalt content, gradation, and percent air voids. Results showed that there is no significant statisti- cal difference in the gradation and percent air voids due to sampling location. A difference was noted for percent asphalt content; however, segregation was observed in the samples obtained from the truck by using a shovel resulting in finer gradation, higher asphalt con- tent, and lower-percent air voids. The study also indicated that samples obtained using the remote truck sampling device had similar properties to samples obtained behind the paver. Elseifi (2007) evaluated the quality control/quality assurance (QC/QA) sampling practices of six highway agencies and determined that sampling behind the paver provides a sample that is more representative of the final product, and many of the states are able to sample at this location without much difficulty. Production and Compaction Types To conduct performance testing on asphalt mixtures, there are different methods employed to fabricate specimens. The ideal method for obtaining a sample representative of field properties is fabrication of specimens from field cores. However, this method is not gener- ally applicable especially when the purpose of testing precedes construction. It is therefore important to fabricate specimens using other methods. Alternative specimen fabrication methods include laboratory mixed laboratory compacted (LMLC), plant mixed laboratory compacted (PMLC), and reheated plant mixed laboratory compacted (RPMLC). Researchers have attempted to study the effect of the various fabrication methods to identify their effect on performance test results as well as to determine which method most closely represents field mixture properties. Johnson et al. (2010) showed that laboratory-produced mixtures containing reclaimed asphalt pavement (RAP) and reclaimed asphalt shingles (RAS) are gen- erally stiffer than the corresponding plant-produced mixture as a result of a better blending of the recycled and virgin binders occurring in the laboratory. Mogawer et al. (2012) evalu- ated plant-produced mixtures containing RAP and showed that RPMLC were significantly stiffer than PMLC. However, with increase in RAP content, the PMLC specimens had a higher increase in stiffness as compared to RPMLC specimens. Rahbar-Rastegar and Daniel (2019) compared the measured properties and cracking behavior of PMLC and LMLC specimens, observing that either fabrication method could produce higher stiffness depending on the plant where the mixture is produced. There was, however, no distinct trend in the cracking behavior. Jacques (2016) showed that specimens fabricated from field cores have lower air void content compared with plant- and laboratory-produced mixtures. Ranking the fabrica- tion methods in terms of mixture stiffness, it was observed that the field cores and RPMLC had the highest stiffness, followed by the LMLC and then the PMLC. Daniel et al. (2018) also observed that there is a difference in the visco elastic and fatigue properties of LMLC, PMLC, and RPMLC, and the magnitude of the difference is dependent on RAP content and binder performance grade. In the NCHRP Project 09-48, these differences in volumetric and mechanical properties of mixture type were quantified, and conversion factors were recom- mended. Mechanistic-empirical (M-E) design models were further employed to show that the predicted performance is significantly affected by mixture type (Mohammad et al. 2016).

14 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories Conditioning Procedures Aging The effects of aging on asphalt mixtures and its implication on cracking performance have been established. Earlier efforts to simulate mixture aging in the laboratory led to the devel- opment of the AASHTO R30 standard specification. Researchers have studied the effect of the standard short-term aging protocols of 2 hours at compaction temperature or 4 hours at 135°C (275°F) for HMA and 2 hours at 116°C (240°F) for warm-mix materials, in an attempt to compare with short-term aging that occurs during production in plant. In a study by Yin et al. (2015), the effect was evaluated through volumetric analysis and performance evaluation by MR, E*, and HWT tests. It was observed that the short-term aging protocols were able to simu- late the asphalt aging and absorption that occur during plant production and construction. NCHRP Report 815: Short-Term Laboratory Conditioning of Asphalt Mixtures also reviewed and confirmed that the current short-term aging protocols are able to simulate plant conditions (Newcomb et al. 2015). Lolly (2013) showed that elevated short-term aging temperature (by 25°F) and time (2 and 4 hours) increase the stiffness of the asphalt mixtures measured in terms of dynamic modulus and IDT strength, with increased aging time having more effect. Over the years, researchers have also attempted to study several existing laboratory long-term aging protocols. The current standard protocol is performed on compacted short- term aged specimens by placing them in a forced draft oven at 85°C ± 3°C (185°F ± 5°F) for 120 ± 0.5 h to represent 7 to 10 years of aging in the field. There have, however, been issues observed regarding this standard procedure related to distortion in the specimen geometry and volumetrics as well as an aging gradient in the specimen (Elwardany et al. 2017). As an alterna- tive, researchers have recommended the long-term aging of loose mixtures (Arega et al. 2013, Partl et al. 2013). In the extensive study by Elwardany et al. (2017), the possibility of employing a pressure aging vessel in place of oven aging was explored. It was observed that the application of pressure deformed the compacted specimens. The study also compared the practical impli- cation of aging loose mixtures as opposed to compacted specimen; results showed that there is an advantage in terms of efficiency and specimen integrity. The Asphalt Institute recommends long-term aging loose mixture in the oven for 24 h at 135°C (275°F). However, recent stud- ies have shown that aging asphalt mixtures at temperatures higher than 100°C (212°F) causes changes in the chemistry of the binder and therefore does not adequately reflect field aging (Rad et al. 2017). Findings from the NCHRP Project 9-54 study led to the recommendation for loose mix aging in the oven at 95°C (203°F) (Kim et al. 2018). Rahbar-Rastegar et al. (2018) investigated how mixture properties change with four different long-term aging protocols on loose mixtures and 5 days at 85°C (185°F) on compacted samples. Longer aging protocols of 12 days at 95°C (203°F) and 24 h at 135°C (275°F) produced mixtures with similar rheological properties but different fracture properties. Zhang et al. (2019) evaluated the three loose-mix long-term aging protocols; the study showed that the protocols of 24 h at 135°C (275°F) and 95°C (203°F) for 12 days resulted in similar changes in both rheological and fatigue properties. Moisture Conditioning Moisture-induced damage has been identified as one the challenging distresses that lead to premature failure in asphalt pavements. To determine the susceptibility of mixtures to moisture-related damage, laboratory conditioning and testing typically follows the established and accepted modified Lottman procedure (AASHTO T 283). However, as a result of incon- sistencies in identifying moisture-susceptible mixtures, there are other applicable moisture conditioning procedures. Earlier research to study different moisture conditioning proce- dures was done by Coplantz and Newcomb (1988), in which four procedures on compacted specimens were evaluated: full vacuum saturation only, full vacuum saturation with one freeze-thaw cycle, 55 to 80% vacuum saturation with one freeze-thaw cycle, and 55 to 80%

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 15 vacuum saturation with multiple freeze–thaw cycles. The study reported that there was no evidence of moisture damage to mixtures subjected to only vacuum saturation; however, damage became more severe with an increase in number of freeze–thaw cycles even with lower saturation levels. Alam et al. (1998), in an attempt to modify the Strategic Highway Research Program Environmental Conditioning System (ECS) (AASHTO TP 34) procedure, evaluated various parameters that could affect the severity of conditioning. It was observed that there was no difference between the ECS saturation and static immersion saturation and that the severity of the conditioning process could be controlled by changing the confining pressure or the chamber temperature. NCHRP Project 9-34 showed that the ECS conditioning procedure is the most promising when evaluated with E* test as compared with the TSR or HWT, when compared with field performance (Solaimanian et al. 2007). In NCHRP Project 9-13, the AASHTO T 283 conditioning procedure was evaluated for its compatibility with Superpave® mix design; specimens were prepared using the Superpave gyratory compactor (SGC) (at 100 mm (4 in.) and 150 mm (6 in.) diameter), Marshall hammer, and Hveem kneading compactor. Results showed that there was no difference in TSR after conditioning with or without freeze–thaw. The study also showed that saturation level (55, 75, or 90%) has a minor effect on the moisture-induced damage (Epps et al. 2000). Sebaaly et al. (2001) reported similar findings that the AASHTO T 283 procedure with or without freeze– thaw had no significant difference in terms of induced damage. Moaveni and Abuawad (2012) compared the modified Illinois DOT and AASHTO T 283 moisture conditioning methods; the study showed that the AASHTO T 283 conditioning resulted in more severe damage as a result of the freeze and thaw cycle included in the procedure. Amelian et al. (2014) employed digital image analysis on specimens subject to the boiling water test and demonstrated a good correlation between these results and TSR and dynamic modulus (|E*|) ratio after AASHTO T 283 moisture conditioning but not retained Marshall stability. Figueroa and Reyes (2016) showed that moisture-induced sensitivity test (MIST) condi- tioning resulted in greater strength reduction as compared with AASHTO T 283 because of vacuum pressure included in the procedure to simulate realistic dynamic load effect. Vishal et al. (2018) also compared the AASHTO T 283 to MIST conditioning for 3,500 cycles at two different temperatures (40°C (104°F) and 60°C (14°F)) and two different pressures (40 psi and 70 psi). In this study, it was observed that the moisture damage done using the AASHTO T 283 conditioning process was similar to the MIST conditioning process at a temperature of 60°C, 40 psi pressure, and 3,500 conditioning cycles. Laboratory Compaction Method In order to simulate field compaction in the laboratory, several laboratory compaction methods and devices have been employed over the years. Several researchers have investigated the existing compaction methods and reported that they produced specimens with significantly different properties and varying trends when correlated with field core properties (Consuegra et al. 1989; Sousa et al. 1991; Button et al. 1994; Khan et al. 1998; Khosla and Sadasivam, 2002; Mbarki et al. 2012; Azari 2014). Additionally, efforts have been made to evaluate the SGC, which is currently the most popular laboratory compaction method. Harvey et al. (2000) showed that laboratory specimens compacted from reheated field mix using the SGC have much greater resistance to permanent shear deformation than do field cores taken from the locations where the field mix was sampled. Epps et al. (2000) evaluated Hveem, Marshall, and SGC compaction methods for specimens of 100 mm (4 in.) in diameter subject to dry and conditioned IDT tests. They found that there were significant differences between the compaction methods for some of the study mixtures.

16 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories Solaimanian et al. (1999) evaluated six models of SGC (Rainhart, Troxler, Updated ITC, Test Quip, Pine, and ITC) based on the AASHTO PP 35 procedure and found that all the compactors produce similar results within the 1% air void tolerance level at the design gyration level. As part of the NCHRP Project 9-29 study, five Superpave gyratory compac- tors (Interlaken, Pine AFGC125X, Pine AFG1, Servopac, and Pine AFGB1A) were compared to determine their effect on |E*| and flow number test. Results showed that the effect was insig- nificant (Bonaquist 2011). Prowell et al. (2003) observed a strong trend between the internal angle of gyration measured by the Dynamic Angle Validation Kit (DAVK) and the result- ing compacted sample density for a wide range of SGCs. For all the SGC brands and models evaluated, results showed on average that a change in 0.1 degree of internal angle will result in a change of 0.010 in bulk specific gravity of the mi (Gmb) or a difference in air voids of 0.4%. DeVol et al. (2007), in a similar study, compared calibration of SGCs using external angle versus internal angle of gyration; average air voids (target of 4.0%) of 3.81% with a standard deviation of 0.54% were observed via external angle, while the internal angle resulted in aver- age air voids of 3.98% with a standard deviation of 0.50%. Specimen Parameters Specimen parameters (e.g., geometry, size, and air void content) with some tolerance level are typically specified in performance tests. Practically, in fabricating specimens, failure in meeting the specified tolerance level has resulted in rejected specimens. Several studies have been done to determine the necessity of adhering to these specified parameter tolerance levels in different performance tests for consistency in obtained results. Specimen Size and Geometry Harvey et al. (2000) compared variance of the repeated simple shear test at constant height (RSST-CH) results for cylindrical specimens 150 mm (6 in.) and 200 mm (8 in.) in diameter versus cylinders 150 mm (6 in.) and 200 mm (8 in.) in diameters that were trimmed along the length in the direction of shearing to obtain specimens that were almost prismatic in shape. It was observed that neither increase in diameter nor trimming reduced the variance of the results. Tandon et al. (2006) showed that measured |E*| values on specimens 150 mm (6 in.) in diameter were more consistent than those measured on specimens 100 mm (4 in.) in diam- eter, especially when the length-to-diameter ratios is less than two. The study also showed that end-friction-reducing layers affect the accuracy and precision of measured dynamic modulus. Bowers et al. (2015), in a study of the use of smaller cylindrical specimens for complex modulus testing, showed that specimens having a diameter and height of 38 × 135 mm (1.5 × 5.3 in.), 50 × 135 mm (2 × 5.3 in.), 38 × 110 mm (1.5 × 4.3 in.), or 50 × 110 mm (2 × 4.3 in.) are suitable alternatives to the full-size specimen for 9.5-mm (3/8-in.) and 12.5-mm (0.5-in.) NMAS mixtures. Only the specimens 50 mm (2 in.) in diameter are suitable alter- natives for 19-mm (0.75-in.) and 25-mm (1-in.) NMAS mixtures. Bonaquist (2008), in his ruggedness study for the dynamic modulus and flow number test, indicated that there was no significant difference in the dynamic modulus measurement between milled specimen ends and sawed specimen ends; however, the end conditions have significant effect on flow- number permanent strain. Epps et al. (2000) found that the dry tensile strengths and TSRs were different between 100 mm (4 in.) and 150 mm (6 in.) SGC specimens. The study also reported differences between SGC and Hveem compaction methods. Saleh (2008) found that resilient modulus is affected by size and geometry; smaller-sized specimens tend to have higher moduli than larger- size specimens. Ahmed et al. (2014) also found that the resilient modulus and indirect tensile strength values of specimens prepared with 100 mm (4 in.) diameter and compacted with a

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 17 Marshall hammer were greater than those of specimens prepared with 150 mm (6 in.) diameter and compacted with a gyratory compactor. Li (2013) studied the effect of specimen size on different types of asphalt-mixture fatigue tests [uniaxial tension and compression (UT/C), four-point bending (4PB), and IDT fatigue tests]. Results showed that the UT/C and IDT fatigue results are not significantly influenced by the specimen size. However, the 4PB test results depend on the dimension of the specimen, because the stress–strain field of the beam specimen varies along the length and cross section. Nsengiyumva et al. (2015) observed that a notch length from 5 mm (0.2 in.) to 25 mm (1 in.) and a specimen thickness of 40 mm (1.6 in.) to 60 mm (2.4 in.) showed good repeatability of SCB fracture energy with small coefficients of variation (≤ 15%). Porter (2016) studied the effects of three different types of notch geometry (a typical rect- angular notch, a semicircular notch, and a fatigue-cracked notch) on the SCB fracture energy. The fracture energy measured with the semicircular notch was greater than that of the standard rectangular notch and is a measure of the crack propagation. The fracture energy measured with fatigue-cracked samples was less than that of the standard rectangular notch and is a more representative measure of both crack initiation and propagation. Barry (2016) observed that the SCB flexibility index increases with decreases in specimen thickness and increases in air voids. Based on the results of the study, correction factors were proposed. In a similar study by Rivera-Perez (2017), the SCB flexibility index was observed to be affected by notch length, and a correction factor was also proposed to account for these differences. Lee et al. (2017) evaluated test specimen diameters of 75 mm (3 in.) and 100 mm (4 in.) and heights of 130 mm (5.1 in.) and 150 mm (6 in.) to determine an appropriate geometry for the DTCF test specimen; the study showed that specimen height and diameter do not affect the damage characteristic curve but affect the propensity of failure inside the gauge length during testing. Based on the results, a specimen of 100 mm (4 in.) in diameter and 130 mm (5.1 in.) in height with a 70 mm (2.75 in.) gauge length was recommended for testing. Air Voids Bonaquist (2011) reported that a wider tolerance level of 2% was not a significant factor affecting the reproducibility of either the E* or flow number test. Dave et al. (2015) observed a weak positive and negative correlation of effect of air voids levels ranging from 1 to 6% on indirect tensile strength (ITS) and TSR, respectively. Marasteanu et al. (2010) found that there is minimal effect of specimen air void content on DCT, but it significantly affects SCB and IDT test results. Zhao (2011) showed that a decrease of 3% (7% to 4%) in air void content significantly increases rutting performance (using APA) and IDT (strength and fracture energy) but has no significant effect on moisture susceptibility (TSR). Azari (2014) compared the two standard methods of measuring of air voids [AASHTO T 166 (saturated surface dry, SSD) and AASHTO T 331 (CoreLok)], observing that the CoreLok tended to measure higher air void content and is less variable when evaluated inter-laboratory as compared to the SSD measure- ment. It was highlighted that this could be attributed to the subjectivity in SSD determination. Specimen Age or Shelf Life Minimal research has been conducted to evaluate the effect of specimen age or shelf life (including storage method) on performance test results; however, as a nonstandardized aspect of specimen fabrication, it may contribute to the variability of test results. Bonaquist (2011) showed that specimen age did not affect the dynamic modulus and flow number over the range of 5 to 200 days included in his study. The study included specimens with 7.0% ± 1.0% air void content, and no specific storage preparation was done between fabrication and testing.

18 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories Anisotropy Over the years, the effect of anisotropy in fabrication of test specimens has been neglected. A few studies have been conducted to evaluate whether the anisotropic effect is significant. Mamlouk et al. (2002) showed that the effects can be ignored for asphalt mixtures fabricated using the SGC. Liang et al. (2006) evaluated specimens fabricated using the roller compactor and cored in the vertical, horizontal, and diagonal (at 45°) directions; the anisotropic behavior of HMA was found to be significant in this study. Kongkitkul et al. (2014), using special molds to compact the mix parallel and horizontal to loading direction, found that the anisotropic effect is significant in the compressive strength and the elastic stiffness of normal HMA and polymer- modified asphalt. Hofko (2013) found that the low-temperature (thermal stress-restrained specimen test), intermediate-temperature (cyclic fatigue), and high-temperature performance (triaxial compression) of HMA fabricated using roller compactors is sensitive to the anisotropy of the material as a result of the coring direction from compacted slabs. Castorena et al. (2017) conducted dynamic modulus and uniaxial cyclic fatigue on small geometry specimens that were cored both horizontally and vertically from SGC specimens. They found that the effect of anisot- ropy was minimal on measured dynamic modulus and fatigue properties. Summary of the Findings from the Literature: National and International Practices for Fabrication of Asphalt Specimens for Performance Tests in Laboratories A summary of the findings from the literature regarding the impact of specimen fabrication practices on performance test results is presented in Table 1 through Table 9, organized by the various topics covered. Table 1 summarizes findings on the potential effect of sampling location on performance test results. Based on volumetric evaluation, research has shown that there is little to no statistical difference resulting from choice of sampling location. Table 2 summarizes the effect of mixture type on performance test results. The literature shows there is an expected difference in mechanical properties based on the mixture type, and the degree of difference is dependent on the plant or laboratory in which the mixture is produced. The findings on the effect of aging conditioning protocols on performance test results are summarized in Table 3. Based on reported findings, different aging protocols can result in significantly different properties that can be captured by different performance tests. Age condi- tioning protocols that best simulate plant and field aging are recommended by various research studies. References Major Findings Performance Test Employed Hassan (2002) No statistical difference in asphalt content of samples obtained from the different locations (at the plant with a tube sampler, at the point of delivery, and behind the paver). NA Turner and West (2006) Little statistical difference in the laboratory properties (volumetrics) as a result of sampling location (except for percent asphalt content). NA Elseifi (2007) Sampling behind the paver provides a sample that is more representative of the final product and is being conducted by states without much difficulty. NA Note: NA = No performance tests were conducted in the study. Table 1. Effect of sampling location.

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 19 Table 2. Effect of mixture type. References Major Findings Performance Test Employed Johnson et al. (2010) Laboratory-produced recycled mixtures are generally stiffer than the corresponding plant-produced mixture as a result of a better blending of the recycled and virgin binders occurring in the laboratory. E* Mogawer et al. (2012) RPMLC materials were significantly stiffer than PMLC. With increase in RAP content, the PMLC had a higher increase in stiffness as compared to RPMLC. E* Jacques (2016) Field cores have lower air void content compared to plant and laboratory-produced mixtures. Field cores and RPMLC had the highest stiffness followed by the LMLC and then the PMLC. E* Mohammad et al. (2016) Recommended conversion factors for the difference in volumetric and mechanical properties of mixture type. HWT, E*, IDT Daniel et al. (2018) There is a difference in the viscoelastic and fatigue properties of LMLC, PMLC, and RPMLC, and the magnitude of the difference is dependent on RAP content and binder performance grade. E*, DTCF Rahbar-Rastegar and Daniel (2019) Either PMLC or LMLC could produce higher stiffness depending on the plant where the PMLC is produced. No distinct trend in fatigue cracking behavior. E*, DTCF Table 3. Effect of aging conditioning protocols. References Major Findings Performance Test Employed Arega et al. (2013) and Partl et al. (2013) Recommended the long-term aging of loose mixtures. Not applicable Lolly (2013) Elevated short-term aging temperature (by 25°F) and time (2 and 4 hours) increase the stiffness of the asphalt mixtures measured in terms of dynamic modulus and IDT, with increased aging time having more effect. E*, IDT Elwardany et al. (2017) Current standard long-term aging protocol of compacted specimen results in issues related to distortion in the specimen geometry and volumetrics as well as an aging gradient in the specimen. Use of a pressure aging vessel in place of oven aging to expedite aging deforms the compacted specimens. There is an advantage in terms of efficiency and specimen integrity on aging loose mixtures as opposed to compacted specimens. E*, DTCF Yin et al. (2015) and Newcomb et al. (2015) Current standard short-term aging protocols are able to simulate the asphalt aging and absorption that occur during plant production and construction. MR, E*, HWT Kim et al. (2018) Recommendation for loose mix aging in the oven at 95°C. Various times depending on climate location and pavement E*, DTCF layer depth. Rad et al. (2017) Aging asphalt mixtures at a temperature over 100°C cause changes in the chemistry of the binder and therefore do not adequately reflect field aging. E*, DTCF Rahbar-Rastegar et al. (2018) Aging loose mix for 5 days at 95°C simulated more aged properties than compacted specimen for 5 days at 85°C did. Longer aging protocols of 12 days at 95°C and 24 h at 135°C produced mixtures with similar rheological properties but different fracture properties. E*, DCT, SCB Zhang et al. (2019) Aging protocols of 24 h at 135°C and 95°C for 12 days resulted in similar changes in both rheological and fatigue properties. E*, DTCF, DCT, SCB

20 Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories References Major Findings Performance Test Employed Coplantz and Newcomb (1988) There is no evidence of moisture damage to mixtures subjected to only vacuum saturation; however, damage becomes more severe with increase in the number of freeze–thaw cycles even with lower saturation levels. MR, IDT Alam et al. (1998) There was no difference between the ECS saturation and static immersion saturation. Severity of the ECS conditioning process could be controlled by changing the confining pressure or the chamber temperature. MR Epps et al. (2000) and Sebaaly et al. (2001) There is no difference in TSR after conditioning with or without freeze–thaw. Saturation level has minor effect on the moisture- induced damage. TSR, MR Solaimanian et al. (2007) ECS conditioning procedure is the most promising when evaluated with dynamic modulus test to simulate field performance as compared to the TSR or HWT. E*, TSR, HWT Moaveni and Abuawad (2012) AASHTO T 283 conditioning results in more severe moisture- induced damage as a result of the freeze–thaw cycle included in the procedure compared with the modified Illinois DOT, which does not include the freeze–thaw cycle. TSR Amelian et al. (2014) There is a good correlation of the results of boiling water test to TSR and |E*| stiffness ratio but not retained Marshall stability after AASHTO T 283 moisture conditioning. TSR, E*, Marshall Stability Figueroa and Reyes (2016) MIST conditioning results in more strength reduction as compared to AASHTO T 283 because of vacuum pressure included in the procedure to simulate realistic dynamic load effect. TSR, Trapezoidal Fatigue Vishal et al. (2018) Moisture-induced damage using AASHTO T 283 conditioning process was similar to the MIST conditioning process at 60°C temperature, 40 psi pressure, and 3,500 conditioning cycles. TSR, Marshall Stability Table 4. Effect of moisture conditioning protocols. Table 4 summarizes the effect of moisture conditioning protocols on performance test results. The research indicates that discrepancies in results depend on the performance test and/or conditioning protocol employed. The effect of compaction methods on performance test results is summarized in Table 5. The literature indicates that the different existing compaction methods result in varying mechanical properties. The literature also recommends to routinely calibrate the compaction devices for better precision and repeatability. Table 6 summarizes the effect of specimen size and geometry on performance test results. The findings suggest that test result precision, typically related to stiffness characterization and flow and fracture properties, may be affected by specimen geometry and size. However, performance tests used to characterize fatigue behavior are not affected. Table 7 summarizes the effect of specimen air voids on performance test results. Some tests results may be significantly affected by a minor change in air voids, whereas other tests can accommodate a wider air void tolerance level. As listed in Table 8, the little research that has been published to evaluate the effect of storage or shelf life on performance test results indicates that it has no significant effect. The literature related to effect of anisotropy on performance test results is summarized in Table 9. There are conflicting results from the literature as to whether anisotropy has a signifi- cant impact on measured mechanical properties.

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 21 References Major Findings Performance Test Employed Consuegra et al. (1989) Five compaction devices were evaluated for mixtures from five projects; the Texas gyratory compactor was found to most closely correlate with field cores. MR, ITS, and static creep Sousa et al. (1991) The compaction method was found to significantly affect measured properties; various methods were more or less sensitive to aggregate and asphalt characteristics. FBF, IDT fatigue, creep in axial compression and shear Button et al. (1994) Four compaction methods were evaluated and compared to field cores; the Texas gyratory compactor was found to most closely simulate the field cores, but there were no statistically significant differences among the compaction types. IDT, MR, Marshall stability, Hveem stability, repeated creep Khan et al. (1998) Five compaction methods were evaluated for four mixtures; the SGC with 1.25o angle of gyration was found to best represent field cores. MR, static creep Solaimanian et al. (1999) Six different models of SGC (Rainhart, Troxler, Updated ITC, Test Quip, Pine, and ITC) produce similar results within the 1% air void tolerance level at the design gyration level. NA Harvey et al. (2000) RPMLC specimens using the SGC have much greater resistance to permanent shear deformation than do field cores taken from the locations where the field mix was sampled. RSST-CH Khosla and Sadasivam (2002) Mixtures from four field sites were evaluated using SGC and rolling wheel compactor (RWC); the RWC was found to most closely simulate field cores. Frequency sweep test at constant height, repeated shear test at constant height, APA, North Carolina State University wheel tracking device Epps et al. (2000) There are differences between 100 mm SGC-, Marshall-, and Hveem-compacted specimens. IDT, TSR Prowell et al. (2003) There is a strong trend between the internal angle of gyration measured by the DAVK and the resulting compacted sample density for a wide range of SGCs. NA DeVol et al. (2007) Calibration of SGCs using internal-angle-of-gyration results in a more accurate and precise (closer to target and less variation) density as compared to use of external angle of gyration. NA Bonaquist (2011) There is an insignificant difference in the result of dynamic modulus and flow number test conducted on specimens compacted using five different Superpave gyratory compactors (Interlaken, Pine AFGC125X, Pine AFG1, Servopac, Pine AFGB1A). E*, Flow number Mbarki et al. (2012) There are significant differences between specimens cored horizontally from field cores and laboratory-compacted specimens. DTCF Azari (2014) Comparison of SGC and kneading slab compactor showed differences depending on the measurements used for evaluation and the quality of the mixture. HWT Note: NA = No performance tests were conducted in the study. Table 5. Effect of compaction methods.

References Major Findings Performance Test Employed Marasteanu et al. (2010) Minimal effect of specimen air void content on DCT, but significantly affects SCB and IDT test results. DCT, IDT, SCB Bonaquist (2011) A wider tolerance air void level of 2% was not a significant factor affecting the reproducibility of either the dynamic modulus or flow number test. E*, Flow number Zhao (2011) A difference of 3% (7% to 4%) in air void content significantly increases rutting performance and IDT (strength and fracture energy) but has no significant effect on moisture susceptibility (TSR). APA, IDT, TSR Azari (2014) The AASHTO T 331 (CoreLok) method tends to measure higher air void contents and is less variable when evaluated inter-laboratory as compared to the AASHTO T 166 (SSD) method. This could be attributed to the subjectivity in SSD determination. NA Dave et al. (2015) There is a weak positive and negative correlation of effect of air void level of range of 1 to 6% on ITS and TSR, respectively. ITS, TSR Note: NA = No performance tests were conducted in the study. Table 7. Effect of specimen air voids. References Major Findings Performance Test Employed Harvey et al. (2000) An increase in diameter or trimming along the length of specimen does not appear to reduce the variance of the RSST- CH results. RSST-CH Epps et al. (2000) Dry tensile strengths and tensile strength ratios were different between 100-mm and 150-mm SGC specimens. IDT, TSR Tandon et al. (2006) Specimen geometry and the end-friction-reducing layers affect the accuracy and precision of dynamic modulus. E* Bonaquist (2008) There is no significant difference in the dynamic modulus measurement between milled specimen ends and sawed specimen ends; however, the end conditions have significant effect on flow-number permanent strain. E*, Flow number Saleh (2008) Resilient modulus is affected by size and geometry. Smaller- size specimens tend to have higher moduli than larger-size specimens. MR Li (2013) UT/C and IDT fatigue results are not significantly influenced by the specimen size. However, the 4PB test results depend on the dimension of the used specimen because the stress–strain field of the beam specimen varies along the length and cross section. UT/C fatigue, IDT fatigue, 4PB fatigue Ahmed et al. (2014) The resilient modulus and indirect tensile strength values of specimens prepared with 100 mm diameter and compacted with Marshall hammer are greater than those of specimens prepared with 150 mm diameter and compacted with gyratory compactor. MR, IDT Nsengiyumva et al. (2015) A notch length from 5 mm to 25 mm and a specimen thickness of 40 mm to 60 mm shows good repeatability of SCB fracture energy with small COVs (≤15%). SCB Bowers et al. (2015) Small-scale cylindrical specimens having a diameter and height of 38 × 135 mm, 50 × 135 mm, 38 × 110 mm, and 50 × 110 mm are suitable alternatives to the full-size specimen for 9.5- and 12.5-mm NMAS mixtures. Only the 50-mm diameter specimens are suitable alternatives for 19- and 25-mm NMAS mixtures. E* Porter (2016) Fracture energy measured with the semicircular notch was greater than that of the standard rectangular notch and is recommended to be a measure of solely the crack propagation while that measured with fatigue-cracked samples was less than that of the standard rectangular notch and is a more representative measure of both crack initiation and propagation. SCB Barry (2016) SCB Flexibility Index increases with decrease in specimen thickness and increase in air voids. Correction factors are proposed. SCB Lee et al. (2017) Specimen height and diameter do not affect the damage characteristic curve but affect the propensity of failure inside the gauge length during testing. DTCF Rivera-Perez (2017) Proposed a new correction factor to account for difference in SCB test specimen notch length. SCB Table 6. Effect of specimen size and geometry.

Literature Review of Practices for Fabricating Asphalt Specimens for Performance Testing in Laboratories 23 References Major Findings Performance Test Employed Bonaquist (2011) Specimen age does not affect the dynamic modulus and flow number test (over the range of 5 to 200 days included in this study). E*, Flow number Table 8. Effect of shelf life. References Major Findings Performance Test Employed Mamlouk et al. (2002) Anisotropic effects can be ignored for specimens fabricated using the Superpave gyratory compactor. Compression and tensile strength test Liang et al. (2006) The anisotropic behavior of HMA was found to be significant. Confined static and dynamic triaxial compression test Kongkitkul et al. (2014) The anisotropic effect on the compressive strength and the elastic stiffness is significant for normal HMA and polymer- modified asphalt. Triaxial compression test Hofko (2013) Low-, intermediate-, and high-temperature performance of HMA fabricated using roller compactors is sensitive to the anisotropy of the material as the result of compaction direction. Thermal Stress Restrained Specimen Test (TSRST), cyclic IDT fatigue, triaxial cyclic compression test Castorena et al. (2017) The effects of anisotropy on dynamic modulus and uniaxial cyclic fatigue results obtained from vertically and horizontally cored small specimens are minimal. SSG E*, SSG cyclic fatigue Table 9. Effect of anisotropy.