Performance Specifications for Asphalt Mixtures (2016)

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

11 chapter two Literature review on Performance SPecificationS for aSPhaLt mixtureS introduction This chapter provides an overview of the specific types of testing used to predict the asphalt mixture performance and on the types of performance specifications used both nation- ally and internationally for plant-produced asphalt mixtures. The information reported will assist in defining the amount to which performance specifications for asphalt have been developed and implemented. This is accomplished through a review of published literature, university and other research reports, and information publicly available on transportation agency websites. Pavement performance was defined by Von Quintus (2009) as “changes in the pavement surface condition over time” and pavements that have excellent performance show little surface distress and have a smooth riding surface over the intended design period. Furthermore, an ideal pavement was defined as consisting of a sustained long-lasting structure, having a smooth surface (both at the time of construction and over time), and requiring low levels of maintenance and rehabilitation over time (Von Quintus 2009). This paper pre- sents the ideal asphalt pavement performance characteristics in terms of the pavement smoothness and measurable lev- els of various distress types including rutting, load-related fatigue cracking, alligator cracking, longitudinal cracking in the wheel path, longitudinal cracking not in the wheel path, and transverse cracking. It also introduced the performance attributes required for designing and constructing long-life flexible pavements. Literature review on Performance teStS for aSPhaLt mixtureS In this section, the test procedures and equipment used for asphalt mixtures are summarized, including information on testing of various types of asphalt mixtures [e.g., WMA and stone matrix asphalt (SMA)] and materials (e.g., binder modi- fiers, RAS, and RAP). This section also summarizes literature findings related to optimum performance for mixtures with recycled materials, optimum performance for mixtures with modifiers, and performance tests and volumetric properties for asphalt mixtures. A summary of the various tests discussed in this section is presented in Table 1, including associated test methods, applicability, and implementation issues. The concept of tying mixture parameters to a PRS was first explored in the WesTrack studies in the late 1990s (Epps et al. 1999). Asphalt mixtures included in the WesTrack experi- ment were tested using the Superpave indirect tensile (IDT) creep and strength test and the thermal stress restrained spec- imen test (TSRST) to predict the mixture propensity to crack at low temperatures and to fail in fatigue. The fatigue tests were performed on beam specimens from both original and reconstructed sections of the test track, whereas probabilistic empirical performance prediction equations were developed for predicting the future deformation and fatigue cracking of asphalt mixes. The study determined that the most important mix parameter for fatigue cracking is compaction, and as the degree of compaction increased, the fatigue cracking poten- tial significantly decreased. The asphalt content, level of com- paction, pavement temperature, and aggregate gradation were all reported to have an impact in predicting each mixture’s future rutting performance. accelerated Pavement testing NCHRP Synthesis Report 433 presented significant findings from full-scale accelerated pavement testing documents and summarized the significant findings from the various experi- mental activities associated with full-scale accelerated pave- ment testing programs (JvdM Steyn 2012). The focus was on activities that took place between 2000 and 2011. The report identified that agencies viewed a major benefit of full- scale accelerated pavement testing programs was to assess improved performance modeling and the development of PRS. A number of agencies noted that they were performing benefit–cost ratio evaluations of their full-scale accelerated pavement testing facilities. dynamic modulus The use of the dynamic modulus as a property for assess- ing asphalt pavement performance, based on its role in the mechanistic–empirical pavement design method, was studied by Jeong et al. (2015). The research builds on the concept of dynamic modulus potentially being one of the major char- acteristics in PRS for QA. The study reported that there is currently no feasible method of estimating the mean and vari- ance of the in-place dynamic modulus for the as-built asphalt

12 TAblE 1 SuMMARy oF KEy FEATuRES oF PERFoRMAncE TESTS FoR ASPhAlT MIxTuRES PERFORMANCE TESTS FOR ASPHALT MIXTURES Asphalt Pavement Analyzer Test Method: AASHTO TP 63: Determining Rutting Susceptibility of Asphalt Paving Mixtures Using the Asphalt Pavement Analyzer (APA) Applicability: The Asphalt Pavement Analyzer (APA) is a second generation device that was originally developed in the mid-1980s as the Georgia Loaded Wheel Tester; a device designed for rut proof testing and field quality control. The APA tracks a loaded aluminum wheel back and forth across a pressurized linear hose over a HMA sample. Although the APA can be used for a number of tests, it is typically used to measure and predict rutting. Implementation: Test time: An 8,000 cycle test takes about 8.5 hours (6 hours to preheat the samples plus about 2.5 hours for the 8,000 cycle test and rut measurements). Creation and preparation of the samples can take upwards of several days depending upon conditioning times. Bending Beam Rheometer Test Method: AASHTO T 313: Determining the Flexural Creep Stiffness of Asphalt Binder Using the Bending Beam Rheometer (BBR) AASHTO PP 42: Determination of Low-Temperature Performance Grade (PG) of Asphalt Binders for the Superpave PG Binder Specification Applicability: Provides a measure of low temperature stiffness and relaxation properties of asphalt binders. These parameters give an indication of an asphalt binder’s ability to resist low temperature cracking. Implementation: Ease of measurement—most labs not equipped to do this Time to process test data—slow Cost of equipment, measurements/tests—Cost ($125–150k) for unit Technician skills required—medium to high

13 Disc-shaped Compact Tension (DSC) Fracture Energy Test Test Method: ASTM D7313-13. Standard Test Method for Determining Fracture Energy of Asphalt– Aggregate Mixtures Using the Disk-Shaped Compact Tension Geometry Applicability: Determines the fracture resistance of asphalt–aggregate mixtures. The fracture resistance can help differentiate mixtures whose service life might be compromised by cracking. The test is used to obtain the fracture energy of asphalt mixture lab or field specimens, which can be used in performance-type specifications to control various forms of cracking such as thermal, reflective, and block cracking of pavements surfaced with asphalt concrete. Implementation: Similar time and cost to perform other mixture and binder performance tests. Researchers at the University of Illinois have determined the average fabrication time per specimen to be in the 10 to 15 minute range for DC(T) testing, which includes the four saw cuts and two cored holes. This is based on mass production of at least a dozen test specimens. The fabrication of fewer test specimens will lead to a longer per-specimen preparation time. Thus, combined with testing time, each DC(T) test will take approximately 30 minutes of technician time for specimen preparation and testing when larger batches of specimens are tested. Material testing labs are currently charging in the neighborhood of $200 per test specimen (replicate) for DC(T) testing, and somewhat less for larger quantities of specimens ($150 per test). Equipment costs can range from $10–50K depending on whether a cooling chamber is required. Dynamic Modulus Test Test Method: AASHTO TP 62-07 (2009) Determining Dynamic Modulus of Hot Mix Asphalt (HMA) Applicability: Dynamic modulus values measured over a range of temperatures and frequencies of loading can be shifted into a master curve for characterizing asphalt concrete for pavement thickness design and performance analysis. The values of dynamic modulus and phase angle can be used as performance criteria for asphalt concrete mix design. Implementation: Equipment cost, manpower, and testing proficiency issues Specimen preparation and testing time Determining specification values TAblE 1 (continued) (continued on next page)

AMPT Flow Number Testing Test Method: AASHTO TP 79-10, Standard Method of Test for Determining the Dynamic Modulus and Flow Number for Hot Mix Asphalt (HMA) Using the Asphalt Mixture Performance Tester (AMPT) Applicability: The AMPT is actually a testing machine capable of performing several tests relating to HMA performance. The SPT can perform a flow time test, a flow number test (a repeated load test), and a dynamic modulus test. It can measure fundamental properties related to rutting and cracking susceptibility (Pavement Interactive 2010). Implementation: AMPT cost. Manpower and testing proficiency are issues Specimen preparation and testing time. Further guidance on specimen preparation is needed. The guidance should include details on support equipment requirements and potential sources and best practices for technicians to prepare AMPT specimens (NCAT 2013). Determining specification values AMPT S-VECD Test Method: AASHTO TP 107-14. This test method covers procedures for preparing and testing asphalt concrete mixtures to determine the damage characteristic curve via direct tension cyclic fatigue tests Applicability: Fatigue performance—An advanced testing protocol using the AMPT for the simplified viscoelastic continuum damage (S-VECD) model has been developed in an AASHTO standard. Push-pull cyclical results from the S-VECD testing can be obtained in two days, which is much quicker than traditional beam fatigue testing. Implementation: Speed of measurement (i.e., real time)—medium, faster than beam fatigue Ease of measurement—medium, cutting edge—most labs not equipped to do this Time to process test data—medium Cost of equipment, measurements/tests—Cost of AMPT is high but add-on is modest ($12–15K) if added to an existing AMPT used for rutting. Technician skills required—medium-high Repeatability and accuracy—to be determined Standardized measurement (ASTM, AASHTO, agency, etc.)—Yes—AASHTO draft out Special calibration requirements—Yes Flexural (Bending) Beam Fatigue Test Method: AASHTO T 321: Determining the Fatigue Life of Compacted Hot-Mix Asphalt (HMA) Subjected to Repeated Flexural Bending ASTM D7460-10: Standard Test Method for Determining Fatigue Failure of Compacted Asphalt Concrete Subjected to Repeated Flexural Bending TAblE 1 (continued)

15 Applicability: The flexural fatigue test is used to characterize the fatigue life of HMA at intermediate pavement operating temperatures. The laboratory fatigue life determined by this standard has been used to estimate the fatigue life of asphalt concrete pavement layers under repeated traffic loading. Although the field performance of asphalt concrete is impacted by many factors (traffic variation, speed, and wander; climate variation; rest periods between loads; aging; etc.), it has been more accurately predicted when laboratory properties are known along with an estimate of the strain level induced at the layer depth by the traffic wheel load traveling over the pavement. Hamburg Wheel Tracking Test Test Method: AASHTO T324, Standard Method of Test for Hamburg Wheel-Track Testing of Compacted Hot-Mix Asphalt (HMA) Applicability: Laboratory wheel-tracking devices are used to run simulative tests that measure HMA qualities by rolling a small loaded wheel device repeatedly across a prepared HMA specimen. Performance of the test specimen is then correlated to actual in-service pavement performance. Laboratory wheel-tracking devices can be used to make rutting, fatigue, moisture susceptibility, and stripping predictions. Some of these devices are relatively new and some have been used for upwards of 15 years such as the French Rutting Tester (FRT). The Hamburg Wheel Tracking Device (HWTD), developed in Germany, can be used to evaluate rutting and stripping potential. The HWTD tracks a loaded steel wheel back and forth directly on a HMA sample. Tests are typically conducted on 10.2 x 12.6 x 1.6 inch (260 x 320 x 40 mm) slabs (although the test can be modified to use SGC compacted samples) compacted to 7 percent air voids with a linear kneading compactor. Most commonly, the 1.85 inch (47 mm) wide wheel is tracked across a submerged (underwater) sample for 20,000 cycles (or until 20 mm of deformation occurs) using a 158 pound (705 N) load. Rut depth is measured continuously with a series of LVDTs on the sample. Several modified HWTDs have been produced in the United States with the principal modifications being loading force or wheel type. Implementation: Also called four-point beam bending test, most accepted fatigue test in the United States. It is a standard test method for determining the fatigue life of compacted hot mix asphalt (HMA) subjected to repeated flexural bending. Testing time is dependent on the strain level chosen for the test. High strain (400–800 microstrain) may be completed in a few hours. Low strain tests (200–400 microstrain) can take several days. Even lower strain levels (50–100 microstrain) can take upwards of a month. Typically 8 to 10 samples are used to develop results for any mix. Hence, it may take several days to several weeks to develop sufficient fatigue data to allow analysis of a given mixture. It has not been widely implemented by transportation agencies because of two main issues: specimen preparation and testing time. Compared to a cylindrical specimen, it is more expensive and difficult to prepare a beam specimen in the laboratory or extract a beam from a pavement section for testing. In addition, a BBF test can take up to more than 50 days depending on the selected strain level. Thus, it is not used for routine asphalt mix design or quality assurance/quality control (QA/QC) testing, which often requires a quick turnaround. TAblE 1 (continued) (continued on next page)

16 Implementation: Presence of detailed equipment requirements may limit the use of Hamburg test equipment manufactured by other companies. T 324 details specific equipment requirements that appear to be written around the original Hamburg wheel-track equipment available in the United States from Precision Machine & Welding (PMW). There is an initiative to make the standards more generic (to allow use of other company’s equipment). Users must be careful to establish laboratory conditions (e.g., load, number of wheel passes, temperature) that produce consistent and accurate correlations with field performance. Performance is largely PG driven. The higher the High Temperature Grade, the lower the rut depth. The higher the use of recycled material, the lower the rut depth. HWTD: encourages high RAP and stiff mixes that will result in increased fatigue and low temp cracking. Need companion test(s) to identify mixes that are overly stiff. Shear Tester Test Method: AASHTO T 320: Determining the Permanent Shear Strain and Stiffness of Asphalt Mixtures Using the Superpave Shear Tester (SST) Applicability: The Superpave Shear Tester (SST) is used to characterize a HMA mixture’s resistance to permanent deformation (rut resistance). This characterization can be used as a performance test for HMA mixtures designed using Superpave mix design or other mix design procedures. The most common SST tests, the repeated shear at constant height (RSCH), and the frequency sweep at constant height (FSCH) tests subject a short HMA cylinder to repeated shear in a pulse manner (RSCH) or a range of loading frequencies (FSCH) in a controlled atmosphere. The SST can measure many parameters but the most typical are permanent shear strain, shear dynamic modulus (|G*|), phase angle (ϕ), maximum shear strain, and recovery. Implementation: The SST tests are sensitive to sample compaction method. Samples compacted with the SGC (the standard method) tend to exhibit greater resistance to permanent deformation than cores extracted from the field or samples compacted with the rolling wheel compactor. Rolling wheel compacted samples show about the same resistance to permanent deformation as field cores. This is why some researchers advocate the use of the rolling wheel compactor when making SST and other HMA test samples. Test time is an issue. It takes about 1 week for sample preparation and testing in the SST. About 2 days of this is actual time in the SST. TAblE 1 (continued)

17 Semicircular Bending Test Test Method: Proposed AASHTO Test Method for Determining the Fracture Energy of Asphalt Mixtures Using the Semi Circular Bend Geometry Applicability: SCB test is used to obtain the fracture energy of asphalt mixture lab or field specimens, which can be used in performance-type specifications to control various forms of cracking such as thermal, reflective, and block cracking of pavements surfaced with asphalt concrete. Implementation: Similar time and cost to perform other mixture and binder performance tests Good comparisons with more well-known Indirect Tensile Strength (ITS) test Simpler to perform and quicker than other fatigue cracking tests Equipment costs can range from $10–50K depending on whether a cooling chamber is required. Texas Overlay Test Test Method: Tex-248-F, Overlay Test Applicability: The overlay test (OT) was developed in the 1970s to test an asphalt mixture’s resistance to reflective cracking, but it has also been evaluated to determine the bottom-up fatigue cracking and the thermal reflective cracking resistance of asphalt mixture (Tex-248-F-09). This test method determines the susceptibility of bituminous mixtures to fatigue or reflective cracking. The test is rapid and repeatable, and poor samples fail in minutes. It characterizes both crack initiation and crack propagation properties of asphalt mixtures. Advantages are that field cores or gyratory compacted samples can be tested relatively quickly. Tests are rapid and repeatable. Results found to correlate to flexible and composite pavements Disadvantages in the sample preparation process (cutting and gluing) In the current Texas DOT procedure (Tex-248-F), the maximum opening displacement of 0.635 mm (0.025 in.) is deemed too large for testing stiff asphalt mixtures (e.g., with higher RAP and/or RAS contents) and the mixtures of asphalt overlay placed in different climate conditions (i.e., smaller daily temperature variation). The OT is currently conducted at a frequency of 0.1 Hz according to the current procedure, but the OT can be conducted at a higher frequency to reduce testing time. In the current procedure, the failure point is defined as the number of cycles where 93% reduction of the initial peak load occurs. This method of determining the failure point is not consistent with those used in other cracking tests, such as the BBF test in accordance with ASTM D7460 procedure. Thus, additional work is needed to evaluate the maximum opening displacement, test frequency, and method for determining the failure point specified in the current OT procedure (Ma 2014). TAblE 1 (continued)

18 pavement; therefore, a methodology of stochastically evalu- ating the in-place dynamic modulus of the as-built asphalt pavement using a relationship of in-place air voids and a set of single dynamic modulus values measured in the laboratory was presented. The methodology was validated with asphalt mix collected from a construction site and produced a plot to be used for comparing the estimated in-place dynamic modu- lus and the as-designed asphalt mix as a simple approach to evaluating as-built mix quality. The reported outcome of the methodology established the possibility of incorporating the dynamic modulus into PRS. repeated Load Permanent deformation A study by Azari (2012) investigated the characterization of asphalt mixtures using an incremental repeated load perma- nent deformation test. nine different mixtures from differ- ent state DoTs throughout the united States were selected to accurately represent a wide variety of locations, temperatures, mixture types, and traffic loads. Each sample was tested at the same controlled temperature for each test with the same load cycles and stress levels. The temperature was continu- ally increased for each test cycle and the minimum strain rate (MSR) curves were developed to show the resistance of an asphalt mixture to permanent deformation. The MSR curves were generated as a function of temperature * pressure, or TP value, noting that as the TP values increased the MSR value increased exponentially. These charts were used to esti- mate rut depths by multiplying the given MSR at the given TP value and multiplying the MSR by traffic equivalent single axle loads (ESAls) to obtain the total strain. once the total strain was found, it was multiplied by pavement thickness to estimate the rut depth. The study reported that with a given TP value for a particular location and traffic pattern, a mini- mum MSR can be calculated and the mixture can be designed to meet that minimum value. It was reported that a mixture with a lower MSR is more resistant to permanent deforma- tion. In addition, an asphalt mixture with a higher TP value is more resistant to permanent deformation when the MSR is held constant. overlay tester A report by Zhou et al. (2014) discussed the possible use of the Texas overlay Tester as an appropriate test for predict- ing repeated load and cracking in the routine asphalt mixture design process. The report surmised that for standard overlay projects the selection of an appropriate mixture design has been difficult, a result in part of the improvement in a mix- ture’s rutting resistance being offset by a negative impact on its cracking resistance. The Texas overlay Tester could poten- tially eliminate the need for these decisions to be made in that it allows for the calculation of certain parameters, depending on location and loading patterns. A number of changes have been made to the original Texas overlay Tester to modify it for a larger range of pavement samples. Some of the changes include requiring a testing temperature of 77°F (25°c), a loading time of 10 s, a maximum opening size of 0.025 in. (0.625 mm), and a failure point defined as a 93% reduction in load from the maximum load measured at the first cycle. The study proposed that the Texas overlay Tester can balance the rutting and cracking requirements and would be most effective when combined with the hamburg Wheel Tracking Device (hWTD) (for evaluating the potential rutting). Studded tire wear Simulator A study by Wen and bhusal (2013) investigated various approaches to developing asphalt mixtures that best resist studded tire wear. Studded tires are commonly used in Alaska, colorado, Idaho, Montana, nebraska, oregon, South Dakota, and Washington State and have been reported to cause signifi- cant rutting damage. Ruts as deep as 1 in. (25 mm) in 6 years of loading, which exceeds the 0.75 in. (19 mm) maximum allow- able rutting in most states, have been observed. In this study, the results of the Studded Tire Wear Simulator (developed in Washington State) are discussed and the sensitive mixture vari- ables were determined to be aggregate type, asphalt binder grade, gradation type, nominal maximum aggregate size, and air void content. The study reported that the benefit of the Studded Tire Wear Simulator is that in addition to measur- ing studded tire resistance, it also measures standard rutting resistance and standard fatigue cracking resistance of asphalt pavements. disc-Shaped compact tension Minnesota Department of Transportation (MnDoT) currently specifies low temperature binder grades to minimize thermal cracking at low winter temperatures (MnDoT 2014). The disc-shaped compact tension test (DcT) was used to simulate the stresses that develop in an asphalt pavement as it shrinks in low temperatures, which was modified from the ASTM D7313 procedure. The DcT measures the fracture energy of a mix- ture (in J/m2) by loading a specimen to fracture. The test con- sists of a disc-shaped test specimen that has a 6-in. (150-mm) diameter and is 2-in. (50-mm) thick, which is placed on the testing apparatus with two holes cut in it for loading locations and a notch cut to initiate the location of cracking. The crack propagation is measured corresponding to the load applied by providing a value in fracture energy at a minimum accept- able value of 0.035 bTu/ft2 (400 J/m2). In 2013, five different asphalt pavement projects were tested using the DcT test. If certain specimens did not meet the minimum fracture energy of 0.035 bTu/ft2 (400 J/m2), the following recommendations were made to increase fracture resistance: (1) reducing the amount of RAP or RAS in an asphalt mixture; (2) reducing the low-end temperature performance grade, increasing the high-end temperature performance grade; (3) using a smaller nominal aggregate size; (4) increasing binder content; and (5) using harder, crushed quarry rock instead of standard gravel aggregates. The study reported that during the next two

19 upcoming construction seasons that MnDoT plans to finalize implementation of the DcT test. At present, the MnDoT is in the process of implement- ing a low-temperature cracking performance specification for asphalt mixtures. The specification utilizes DcT fracture energy as a performance criteria. A pilot implementation was undertaken in 2013 by the use of performance specifications for five construction projects in Minnesota (Johanneck et al. 2015). The implementation required the mix design speci- mens to be tested as part of mix approval and verification testing conducted on production mix samples. The pilot study helped identify some challenges to full-scale implementation as well as find out certain deviations in DcT fracture energy measurements that can be seen between laboratory-prepared mix design samples and plant-produced production mix. on the basis of the lessons learned through the pilot implementa- tion, current research is underway to modify and finalize the DcT fracture energy performance specifications. MnDoT is presently implementing the use of provisional specifications, intended for use in 2017. This study also reaffirmed traditional viewpoints on asphalt mix design such as increasing levels of binder content and/ or the use of a “colder” performance grade low-temperature binder that is presumed to create a softer mix and result in higher fracture energies. Research is ongoing on the impact of these individual mix design parameters, along with other relevant parameters (VMA, VFA, PG range, percent of recycled materials, etc.) on fracture energy. Furthermore, the pavement sections constructed during the pilot imple- mentation (both with and without adjusted mixes) are being continually observed and their field cracking performance documented to study the effects of fracture energies on crack- ing performance. The chicago Department of Transportation (chicago DoT) adopted a test procedure, entitled Determining Frac- ture Energy of Asphalt-Aggregate Mixtures using the Disk- Shaped compact Tension Geometry (IDoT 2014), which was modified from the ASTM D7313 procedure. In this method, cores must be taken no less than 12 in. (300 mm) from the edge of the pavement surface and a minimum of three cores must be extracted and tested in the DcT. The specification lists requirements such as compaction, vari- ance, and size of samples, as well as adjusted requirements for non-standard samples such as SMA and other pavement types. The minimum requirements for the mixtures tested in the DcT include: 0.035 bTu/ft2 (400 J/m2) for dense-graded asphalt mixtures; 0.031 (350 J/m2) for low ESAl (i.e., lower heavy vehicle volumes) asphalt mixtures; and, 0.017 (200 J/m2) for pervious asphalt mixtures. chicago DoT also requires that all mix designs meet a set of minimum DcT values based on whether the mixture is dense-graded, pervious, or low-volume (low ESAl pavement) when tested using the modified Illinois DoT specification (IDoT 2014). chicago DoT requires that contractors test all mix designs for DcT compliance, regardless of the agency giving the mix design approval. In addition, the contractor will be required to sub- mit two prepared uncut gyratory specimens 4.75 in. (120 mm) in height for DcT verification testing by the chicago DoT’s QA staff. A portion of the specification used by chicago DoT is included in web-only Appendix D. mixture types evaluated with Performance tests Reclaimed Asphalt Pavement A study was conducted in 2013 that served as a synthesis of various types of testing on mixtures designed with RAP using materials from different parts of the united States with different levels of RAP (Marasteanu et al. 2013). In the con- text of the study, a high RAP content was defined as asphalt mixture produced with more than 25% RAP. The first experiment was designed to determine an opti- mal binder content; the results were inconclusive. Dynamic modulus testing was completed to determine critical tem- peratures; however, the results varied from measured critical temperatures and were also deemed inconclusive. Moisture damage susceptibility was then tested. A large portion of the high RAP content mixes did not meet the 0.80 TSR criteria; therefore, an anti-stripping additive was added in these cases. In all cases, the tensile strength of the RAP mixes exceeded the tensile strength of the virgin mixes. Permanent deforma- tion was evaluated by testing each mix with the confined flow number test. none of the samples exhibited deformation using this method and rutting resistance was determined based on strain instead, which was not affected by the RAP content. Resistance to fatigue cracking was tested with IDT strength tests. The results showed that high RAP mixes had signifi- cantly lower fracture energies than the virgin mixes. Fracture energy was improved when a softer grade of virgin binder or a rejuvenating agent was used. Resistance to thermal cracking was tested with both the low temperature Semi-circular bend (Scb) test and the bending beam Rheometer (bbR) test. Typi- cally, the high RAP content mixes had higher fracture tough- ness than the virgin mixes but similar, or even lower, fracture energy results. The bbR results showed that mixes containing RAP had higher stiffness and lower m-values, which would result in a higher potential for thermal cracking. The report con- cluded that based on the critical temperatures, the high RAP content mixes appeared to perform similarly to the mixes pro- duced with virgin aggregates. Wisconsin DoT created special provisions to establish a procedure for the use of RAP. The testing done on asphalt mixtures is described in detail in the latter half of the special provisions document starting under Section 460.2.8.4.1.4.2, Department (bureau of Technical Services) Verification Per- formance Testing Requirements (WisDoT 2014). The appro- priate range was determined for the following material parameters: (1) air voids within a range of 2.2% to 4.8%; (2) VMA within ±0.5 of the minimum requirement; and

20 (3) adequate lift thickness. In Wisconsin, pay factors were pri- marily determined based on meeting the minimum required density depending on the type of roadway, although other parameters can be considered as well. WisDoT uses ASTM D 7313-07 (ASTM 2013) using the DcT geometry as the stan- dard test method for determining fracture energy of asphalt aggregate mixtures. Random samples are selected for test- ing and the procedure includes creating a valid test specimen with appropriate material qualities, placing the specimens in a standard freezer for 8 to 12 hours at 10°F (-12°c), and then placing the specimens in a DcT chamber for 1.5 hours at the standard testing temperature, which varies based on the PG binder grade. The minimum allowable fracture energy for all test specimens is 0.035 bTu/ft2 (400 J/m2). WisDoT also reported that it uses the AAShTo T 324-11 hamburg Wheel Track Test (AAShTo 2011) as the standard test for determining allowable rutting levels for hMA. Depending on the asphalt binder grade, a range of 5,000 to 20,000 load passes are completed with a maximum rut depth found to be consistently measuring 0.50 in. (12.5 mm). WisDoT uses the Scb test for the evaluation of crack propagation in asphalt mixtures by computing the critical strain energy rate (Jc) for mixtures containing RAP. The university of oklahoma is investigating the fatigue testing of RAP in six different mixes that range in terms of the amount of RAP and the types of binder grades (Zaman 2014). The mixture samples were tested for beam fatigue, IDT strength, dynamic modulus, creep compliance, and resilient modulus. In addition, the overlay Tester and the Scb test are planned for the study. Following AAShTo TP 107-14 (AAShTo 2014b), the fatigue testing is being conducted three times with each sample, using short, medium, and large fail- ure cycles (i.e., the number of cycles at which the phase angle reaches its peak or the modulus decreases to 10% of the initial modulus, whichever occurs first). Prelimi- nary results were reported and revealed good correlation between the fatigue lives of the binder and the mixture. High-Performance Thin Overlays To evaluate the potential for performance-based specifica- tions of high-performance thin asphalt overlay mixtures, Mogawer et al. (2014) investigated the laboratory perfor- mance of asphalt mixtures from three state DoTs: Minnesota, new hampshire and Vermont. These asphalt mixtures were used to develop a pilot specification for pavement overlays. A guide was developed to address reflective cracking, ther- mal cracking, fatigue cracking, and rutting of asphalt overlay mixtures, and in each of the three pavement mixtures that were tested up to 25% RAP was used in the mix. criteria were developed for both types of cracking and for rutting. For thermal cracking, it was reported that the mixture must be within ±10°F (6°c) from the low temperature PG of the binder. For other types of cracking, the mixture is expected to exhibit average overlay testing of greater than 300 cycles to failure. For fatigue life, the mixture is intended to meet flexural beam testing of greater than 100,000 cycles. Finally, for rutting the average rut depth for six specimens is intended to be less than 0.16 in. (4 mm) at 8,000 cycles. The study also found that when RAP is included in the asphalt mixtures, it shall exhibit overlay testing cycles to failure within ±10% of those in the control specimens without RAP. based on the results of laboratory testing and field observations, the pilot specification was slightly refined; however, the impact of RAP on the mixture performance was not determined as part of this study. Warm Mix Asphalt In a study by Jones et al. (2010), a series of tests were com- pleted to assess the differences in performance between hMA and WMA when the WMA additive Rediset® WMx was used. The tests were conducted to determine rutting potential, fatigue cracking performance, and moisture sensitivity of both mixture types. The tests conducted included shear testing, fatigue testing, hWTD test, cantabro test, and TSRST test. It was determined that in the TSRST test, the mixtures with the Rediset WMx additive exhibited significantly better moisture resistance than the control mixes. In each of the other tests, similar results with regard to performance were displayed. The performance of WMA was assessed in a study by Sargand et al. (2009) in which WMA was constructed out- doors and subjected to standard vehicular loading. There were no obvious visual differences in the WMA compared with hMA after 20 months of service life; therefore, the investiga- tion of WMA was then observed under laboratory conditions in an instrumented section in order to measure the tempera- ture, deflection, subgrade pressure, and longitudinal and trans- verse strains. There were four different mixes: a control hMA and three WMA manufactured in three different approaches. The laboratory specimens were subjected to rolling wheel loads at temperatures of 40°F, 70°F, and 104°F (4°c, 21°c, and 40°c, respectively). All three of the WMA mixes experi- enced more initial consolidation than the hMA mix, and the WMA made with emulsion consolidated about twice as much as the other WMA mixes. After initial consolidation, differ- ences in further consolidation were negligible. The transverse and longitudinal strains under falling weight deflectometer (FWD) loading were reported to be consistent among all of the mixes and the study concluded that WMA performed at least as well, if not better, than the hMA mix. Research conducted on WMA in both the laboratory and field using the hWTD test demonstrated the difference in optimum asphalt content for WMA as compared with that of hMA (Alvarez et al. 2010). The study also used the hWTD test to determine the rutting potential of WMA compared with hMA and found that, if given sufficient time to cure, the WMA can achieve the same strength as the hMA. other tests were also run to compare the performance of WMA with

21 hMA: (1) the Texas overlay Tester was conducted to pre- dict reflective cracking resistance, (2) a dynamic mechanical analysis was completed to observe fatigue cracking resis- tance, and (3) surface energy measurements were completed for determining moisture susceptibility. The investigation demonstrated that the suite of tests included in the study were effective for describing the fatigue life, rutting resistance, and moisture susceptibility of WMA mixtures. NCHRP Research Results Digest 374 (2012) provided a recommended testing to define a WMA technology evaluation program that would be compatible with a centralized system of testing, evaluation, and data reporting of engineering materials for the state DoTs, AAShTo national Transportation Product Evaluation Program (nTPEP). The suite of mixture perfor- mance tests recommended for the qualification of WMA, as part of the nTPEP program, is shown in Table 2. Recycled Asphalt Shingles A study conducted by MnDoT considered the incorporation of RAS in asphalt mixtures of varying composition: (1) a mixture containing 20% RAP, (2) a mixture containing 15% RAP with 5% tear-off shingles, and (3) a mixture contain- ing 15% RAP with 5% manufactured shingles (McGraw et al. 2010). The mixtures were tested in the laboratory to determine stiffness, rutting potential, and moisture sensitivity using dynamic modulus testing, asphalt pavement analyzer, and the lottman test, respectively. The dynamic modulus test- ing showed that mixes with tear-off shingles were stiffer than mixes with manufactured waste scrap shingles. At higher tem- peratures, there was a large stiffness difference between RAP mixes and virgin mixes. The manufactured shingles mix exhib- ited higher rutting potential than the tear-off shingle mix, and both indicated less rutting potential than virgin mixes. Tear-off shingle mixes were found to be more susceptible to moisture damage than manufactured shingle mixes, with most tear-off shingle mixes failing to meet MnDoT specifications. The results of IDT testing indicated that the tensile strength was not affected by substituting shingles for a percentage of the RAP material. Another study by McGraw et al. (2007) investigated the use of both tear-off shingles and manufactured shingles, com- bined with traditional RAP materials, in both Minnesota and Missouri. Tensile strength tests using the IDT were conducted along with the mixture bbR and direct tension tests on both conventional asphalt produced with virgin binder and on mix- tures with various RAS contents. The research results indicated that the addition of shingles lowered the temperature suscepti- bility to moisture damage of the binders, rendering them stiffer than conventional and RAP-modified binders, at intermediate temperatures more characteristic of fatigue cracking distress. Asphalt Mixtures Produced with Recycled Tire Rubber In a study by bennert et al. (2004), traditional asphalt mix- tures were modified using crumb rubber at 20% of the total weight of the asphalt binder. The crumb rubber was blended with the PG64-22 for 1 hour before mixing with the aggre- gates. Four hMA mixes were used with different PG grades along with one asphalt rubber mix, and the mixtures were analyzed by the following performance tests: (1) APA, (2) repeated load permanent deformation, (3) dynamic modu- lus, (4) repeated shear, (5) frequency shear, and (6) simple shear test. The simple shear test results indicated that the rubber- modified hMA mixture experienced creep development at lower test temperatures, but limited creep at high tempera- tures when compared with the standard mix. Frequency shear and simple shear testing revealed that the rubber-modified asphalt mix had significantly lower stiffness at higher loading frequencies, indicating that the mix is less prone to fatigue cracking at lower temperatures. At elevated test temperatures, the rubber-modified asphalt mix had the highest shear modu- lus, which was found to be indicative of improved rutting resistance. The dynamic modulus testing indicated that by adding crumb rubber to an asphalt mixture both the high tem- perature and the low temperature grade should be increased. overall, the performance testing indicated that the rubber- modified asphalt mix will perform well in both temperature extremes along with exhibiting improved fatigue resistance and the potential for longer service lives. tooLS for deveLoPment and imPLementation of Performance SPecificationS In this section, a summary is presented of findings from the literature related to the development and implementation of tools that support the performance specifications for asphalt mixtures. one of the first efforts to initiate the concept of a PRS for asphalt mixtures was developed by Epps et al. (2002b) as part of the WesTrack project. A Microsoft Windows-based software package, hMA Spec, was developed to generate a construction specification and provide equations that define TAblE 2 SuMMARy oF lAboRAToRy TESTS: MIxTuRE PERFoRMAncE Source: Marzougui et al. (2012). Test Mixture design verification with 150-mm diameter Rutting Dynamic modulus AASHTO TP 79, T 324, and T 340 AASHTO TP 79 and PP 61 Compactability Durability AASHTO R35 draft appendix section 8.3 AASHTO T 283 and T 324 AASHTO T 320 Specification

22 how pay will be either increased or decreased for meeting, or failing to meet, established performance target values. The software allowed users to input asphalt mixture variables and predicted changes in performance based on adjustments to the inputs. In addition, the software determined how pay is adjusted based on target values selected by the user. The per- formance prediction models were developed based on data from the WesTrack project, published data, and laboratory- determined data for the specific location in which the asphalt mix was intended to be constructed. Stiffness, permanent deformation, and fatigue cracking were the three primary variables that the software used to predict performance. The first in a number of national studies in recent years focusing on PRS for asphalt is presented in NCHRP Research Results Digest 291 (2004). The document presents a sum- mary of the key findings of nchRP Project 9-15, “Quality characteristics and Test Methods for use in Performance- Related Specifications for hot Mix Asphalt Pavements,” which investigated simple and rapid nDT procedures for eval- uating the properties of as-constructed hMA pavements by measuring mixture quality characteristics. The study included performance indicators for segregation, initial ride quality, in-place mat density, longitudinal joint density, and in-place permeability. based on this study, initial specification crite- ria and threshold values for these five parameters for a PRS are presented and recommendations for further evaluation and validation of these test methods and suggested values are provided. one of the more recent national efforts for advancing the development and implementation of performance specifi- cations was the development of the Quality Related Speci- fication Software (QRSS) through nchRP Project 9-22, A Performance-Related Specification for Hot-Mixed Asphalt (Fugro consultants and Arizona State university 2011). Through this effort, a Microsoft Windows-based program entitled QRSS was created that uses the distress perfor- mance models, originally developed as part of nchRP Proj- ect 1-37A (Applied Research Associates 2004), to predict how an asphalt mixture will perform over time. The QRSS was created to predict asphalt pavement distresses such as rutting, fatigue cracking, thermal cracking, and rideability based on the International Roughness Index (IRI). The per- formance measures are a function of air voids, asphalt con- tent, aggregate gradation, volumetric properties, and binder viscosity of the Ac layer, among others. In addition, the soft- ware takes into account pavement structure, traffic loading, and climate at a given location of interest. The software uses the volumetric and material properties of the as-constructed asphalt pavement to predict future performance by estimat- ing the dynamic modulus of each asphalt layer. The perfor- mance results are then compared with the predictions of the as-designed asphalt mixture. These predictions form the basis of a PRS, in which predetermined parameters selected for pay adjustments will not be based strictly on volumetric properties, but on predicted differences in the service lives (or long-term performance) of the flexible pavements. For example, selections for performance indicators in asphalt mixtures include an allowable amount of rutting, thermal cracking, and/or service life, all determined by each indi- vidual state DoT. nchRP Project 9-22A, Evaluation of the Quality Related Specification Software (QRSS) Version 1.0, was then con- ducted to beta test the QRSS Version 1.0 through the analy- sis of samples obtained from multiple paving projects in Texas, Rhode Island, and utah (Moulthrop et al. 2012). The samples were taken from each project and evaluated for perfor- mance, service life, and life expectancy difference between as- designed and as-built pavements. Volumetric-based dynamic modulus values were calculated and compared with the val- ues measured originally in the laboratory and then input to the QRSS. The predicted performance, service life, and life expectancy were determined and compared with outputs from the QRSS through thermal fracture calculations, fatigue crack- ing, and permanent deformation in the asphalt pavements. The report indicated that the dynamic modulus values measured in the laboratory corresponded well with the dynamic modulus values calculated using the QRSS. It also indicated that when using highly modified asphalt mixtures, the dynamic modulus predictive equation in the QRSS gives less comparable results in terms of rutting predictions. overall, the research showed that the QRSS accurately predicted life expectancy for each sample. nchRP Project 9-22b, Comparing HMA Dynamic Mod- ulus Measured by Axial Compression and IDT Methods, studied the impacts of using different asphalt pavement spec- imens and configurations for distress model predictions from the QRSS Version 1.0 (Mccarthy and bennert 2012). There are various specimen configurations used for dynamic modu- lus testing along with multiple options for pavement predic- tion models. Two types of specimens were used for testing: field cores and laboratory compacted. The specimens were tested using different test configurations under the uniaxial compression test and the indirect tension test. The rutting in the asphalt surface layer and asphalt binder layer were pre- dicted, along with bottom-up fatigue cracking in the binder layer. Dynamic modulus values were inputted into three analy- sis programs: the Mechanistic Empirical Pavement Design Guide (MEPDG), Arizona State university’s SPT program, and QRSS. The types of materials explored in this study included traditional dense-graded asphalt mixtures, SMA, WMA, and a high and low percentage RAP mixture. A few key conclusions were derived from this study. It was concluded that indirect tension testing is an appropriate alternative to uniaxial compression testing. It was reported that laboratory-compacted samples, plant-compacted samples, and field-compacted samples all displayed similar results in terms of rutting predicted by the various tools, but that the plant- compacted samples produced the most accurate results in terms of predicting the appropriate level of fatigue crack-

23 ing. The report suggested that it is not necessary to require plant-compacted or field-compacted samples to get accurate pavement performance predictions and that the MEPDG soft- ware was a relatively appropriate tool for determining the life expectancy of a pavement. In the 1990s, california DoT (caltrans) determined that the method and material specifications that were being used at the time were not adequate for producing long-lasting asphalt mixtures, due to variable performance over time (harvey et al. 2014). In 2000, caltrans developed the calME flexible pave- ment design software (california Department of Transporta- tion 2014), which was calibrated using accelerated pavement testing from numerous locations. over a period of 10 years of testing and calibration, caltrans implemented calME on three northern california Interstate highway rehabilitation projects. The study reported that material properties in dif- ferent regions of california are not well established and that testing is required to determine performance metrics for local materials in particular regions. It was determined that pay scale factors be appropriate for each of the different regions of the state, particularly considering the wide range of aggre- gate bases used that have varying material properties and strengths. The use of repeated load testing was suggested and that different combinations of stiffness and fatigue behavior will allow for mixture designs to be within the acceptable parameters for a certain project. The study recommended that it would be ideal to provide designers with more flexibil- ity and consider alternative combinations to achieve certain asphalt mixture properties. Pay adjuStment factorS in Performance SPecificationS Any information used to set pay adjustment factors for performance-based specifications (PbS) or PRS for asphalt mixtures will be summarized in this section. NCHRP Research Results Digest 371 reported an evalu- ation of common approaches for pay adjustment factors for asphalt pavement (hanna 2013). The most common approaches reported included: engineering-based (complex) methods, empirical methods, and experience-based meth- ods. Engineering-based methods are methods that have been developed based on relationships, and mathematical data and empirical methods are similar but derived from experience rather than engineering principles. Experienced-based meth- ods do not use mathematical principles and do not predict performance; however, the pay factor adjustments are deter- mined based on whether the pavement conforms to certain mix design standards. A survey to state agencies included in the document reported that incentives typically range from 1% to 15%, with the most common incentive reported to be set at 5%, and smoothness is the most frequently used quality measure, followed by the percent within limits associated with performance parameters such as density or air voids. In addi- tion, the survey revealed that most agencies have maximum disincentives and many agencies use a remove-and-replace provision, but only a few agencies use a shutdown provision. The details of triggers for disincentives, remove-and-replace, and shutdown provisions varied among different agencies. NCHRP Report 704 (Fugro consultants and Arizona State university 2011) identified that pay factors are a bonus (incentive) or penalty (disincentive) applied to the contract, depending on the predicted service life of as-constructed asphalt mixes. The report listed IRI, rutting, and fatigue cracking as common characteristics used in determin- ing pay factors. Rutting in the surface asphalt layer was considered in the QRSS, based on a database of more than 800 samples under a variety of conditions. An empir- ical model was developed to predict fatigue cracking, in which more than 7,500 simulations were run to develop a fatigue model. A sample of pay factors determined for IRI are shown in Table 3. TAblE 3 SAMPlE PAy FAcToRS DETERMInED by ThE nchRP QRSS Tool Source: Fugro Consultants and Arizona State University (2011). Schedule 2 (Intermediate Traffic Volume) Pay Adjustment ($/0.1 mile of Traffic Lane) Schedule 1 (High Traffic Volume) Schedule 3 (Low Traffic Volume) Average IRI for each 0.1 mile of Traffic Lane (ln/mile) ≤ 30 600 580 580 29031 32 560 560 280 27054054033 74 75 76 77 78 -260 -60 0 -240 -40 0 -220 -20 0 -200 0 0 … … … … -180 0 0 600 300

24 A study by Monismith et al. (2000) developed perfor- mance models that can be used for PRS for flexible pave- ments based on the results of the WesTrack experiment. The models that were developed account for permanent deformation and fatigue cracking and can be used to develop pay factors that can be used for PRS and hMA pavements. The pay factors consider the quality characteristics of air void content, asphalt content, hMA thickness, and aggregate gradation. The cost model presented considers the present worth of rehabilitation costs due to the as-designed versus as-constructed quality of the pavement. A study by Epps et al. (2002a,b) used a life-cycle cost model that outputs distresses such as fatigue cracking, rut- ting, and serviceability loss. The performance inputs were reported to include target pavement thickness, smoothness, asphalt content, air void content, and aggregate gradation among other site factors. An iterative process involving thou- sands of individual life-cycle cost analyses was conducted to determine the differences between the as-designed and as-constructed asphalt pavement and, depending on values such as the standard deviation and mean of the mixes, pay adjustment factors were determined based on the differ- ences between the performance life predictions. Weed (2003) proposed a simplified procedure for devel- oping PRS for hMA pavements, which directly considers the effects of as-constructed quality characteristics on expected pavement life-cycle costs in the selection of pay adjust- ment factors for these quality characteristics. The procedure consisted of using in-place air voids, thickness, and initial smoothness of a constructed flexible pavement as the pri- mary as-constructed quality characteristics that affect pave- ment performance and expected pavement life. A generic exponential model for computing expected pavement life was developed based on acceptable and unacceptable levels of each quality characteristic. A separate model can then be used to convert expected pavement life to a pay adjustment and pay schedule or incentive/disincentive for the different quality characteristics. More recent research by Weed (2006) presented a more general model that allows for greater flexibility in devel- oping multicharacteristic relationships. The refined model designated high and low failures as two-sided requirements for parameters, such as high and low limits for air voids in flexible pavements, because conditions considered either too high or too low can negatively affect pavement performance. The model provides a rational approach for tying expected pavement life back to pay adjustments for as-constructed quality. A study for the nebraska Department of Roads (nebraska DoR) investigated the creation of a system intended to assign incentives or disincentives for pavement construc- tion, based on long-term performance characteristics (Peruri 2007). because the nebraska DoR already has an incen- tive program in place, only certain additional characteristics were investigated such as IRI, rutting index, and bleeding (flushing) of paved surface based on data from a number of roadways across the state. These characteristics were sug- gested for inclusion in the existing pay scale due to their potential to provide better long-term performance predic- tions than the current approach. tyPeS of Performance SPecificationS uSed nationaLLy and internationaLLy This subsection summarizes literature related to types of performance specifications for asphalt mixtures documented in the united States and internationally. The information in the following sections was obtained from published litera- ture and is presented in subsections organized by state or country. Louisiana The louisiana Department of Transportation and Develop- ment (louisiana DoTD) published an update on its expe- rience with performance-based and PRS and achieving a balanced asphalt mixture design through the modification of specifications by cooper et al. (2014a). The research reported that ensuring performance by balancing the amount of rutting and fatigue cracking has been an issue. In 2013, louisiana DoTD proposed specification modifications for balancing the mixtures, using the testing done on 11 mixtures produced in 2013 to compare with the performance of 40 different mixtures produced in 2006. The testing done on the samples included the hWTD with its loaded wheel test (lWT) and the semicircular bending (Scb) fatigue test. michigan Williams (2004) used characterization of materials, perfor- mance testing of asphalt specimens, and statistical analysis in an attempt to develop a performance-based specification for the Michigan DoT. The objectives of the study were to obtain and characterize asphalt field samples through- out Michigan, develop performance testing criteria, and, ultimately, develop field specifications for acceptance. The main reason for the study was to move forward in testing and acceptance procedures and to facilitate the eventual implementation of PRS. The research divided Michigan into six different regions to address the various climatic properties and levels of material availability, and testing was done on asphalt sampled from each of the six regions. The research primarily used IDT, Super- pave shear tester, beam fatigue, uniaxial strain test, and

25 asphalt pavement analyzer (APA). With this information, the accuracy of empirical models used in the past and the effect that asphalt content and air voids have on long-term performance were determined. new jersey currently, the new Jersey DoT is in the implementation stages of an asphalt mix design and acceptance procedure that includes: (1) performance of volumetric design and allowance for verification by the new Jersey DoT; (2) sup- plying of materials and asphalt mixes to an external labo- ratory for performance testing; (3) production of mixes through a plant and paving of test strips offsite; (4) sam pling during production and provision of samples to an external laboratory for further performance testing; and (5) sampling and testing every other lot (bennert et al. 2011). A rutting check, flexural cracking check, and a pavement cracking check are the tests performed by an external laboratory using an APA, a flexural beam fatigue test, and an overlay Tester, respectively. The APA tests for rutting susceptibil- ity, whereas the flexural beam fatigue test determines the fatigue life of asphalt mixtures. The overlay Tester was used to simulate horizontal movement at the Pcc joint to capture reflective cracking in the asphalt overlay, owing to environmental temperature cycling of the underlying rigid pavement. The type of test used in each case is dependent on whether the mode of cracking is dependent on the flex- ural properties or the expansive properties of the asphalt pavement. currently, there are five different performance-based asphalt mixtures that are required to undergo the testing pro- cedure previously mentioned (bennert et al. 2014). These mixtures are the high-performance thin overlay (hPTo), binder rich intermediate course (bRIc), bridge deck water- proofing surface course (bDWSc), bottom rich base course (bRbc), and asphalt mixtures containing high percentages of RAP. Each of these asphalt mixtures requires different volumetric properties for acceptance. Although the new Jersey DoT only implemented these performance-based specifications in 2008, field performance has shown that these mixtures are performing exceptionally well, particu- larly in comparison with previously designed mixtures that were only reaching about half of the expected service life. These performance-based specifications allow for both con- tractors and engineers to be more accurate in accounting for performance in each of these distress categories. virginia hughes and Maupin (2000) reported that since the mid- 1960s, the Virginia DoT has been working to develop end result specifications, ideally in the form of PRS, where the mixture quality is directly related to performance. The dif- ficulty comes in developing tests for quality characteristics that accurately represent a multitude of different approaches that contractors may have to generate mixture design. The identified Ac quality characteristics deemed necessary for predicting future performance in the study were degree of compaction, thickness, smoothness, segregation, strength, and durability. The research proposed required actions to move forward with development such as creating a density specification that measures air voids and implementation of an improved smoothness specification. The report also identi- fied that a test that measures strength accurately be developed along with a measure of durability to address segregation and other durability-related distresses. international efforts: australia The state of Queensland in Australia developed a tech- nical specification to assist in the construction of dense- graded asphalt layers for heavily trafficked roads (State of Queensland Department of Transport and Main Roads 2013). The specification includes that standard testing shall be done on the asphalt mixtures in order to test their sensitivity to water (Test #Q315) and the compacted den- sity by the amount of air voids at 250 cycles (Test #Q322). Asphalt mixtures intended for use as a binder layer under the surface layer must have a high level of rutting resistance, greater than 12 years of service life, texture depth greater than 0.016 in. (0.4 mm), and an average permeability of less than 15 µm/s. Asphalt mixtures used in structural layers as part of a heavy-duty flexible pavement must have a relatively high level of rutting resistance and an average permeabil- ity of less than 15 µm/s. The specification defines that asphalt mixtures may be comprised of coarse aggregate, fine aggregate, filler, addi- tives, RAP, and/or binder. The maximum RAP content is 15% and RAP may not be used in surface layers. The following performance requirements shall be met: • The sensitivity to water must be at least 80%; • Air voids at 250 cycles must be at least 2.5%, 2.6%, or 3.0%, depending on the mix type; and • The final rut depth shall not exceed 0.16 in. (4 mm). The specification also requires that asphalt mixtures shall be tested for binder content, gradation, density, binder vol- ume, binder fraction, stability, flow, stiffness, air voids, VMA, voids in the binder, resilient modulus, wheel tracking, and sensitivity to water. During construction, the minimum lift thickness shall be 2 in. (50 mm). The maximum thickness shall be either 2.4 in. (60 mm) or 3.2 in. (80 mm) depending on mixture type. The compaction shall be at least 92.5% or 93%, depending on mixture type.

26 international efforts: new Zealand A performance-based specification was implemented in 2000 in new Zealand for the design, maintenance, and perfor- mance requirements for flexible unbound pavement layers for the construction of new pavements and reconstruction of existing pavements (Transit new Zealand 2000). The con- tractor is responsible for the pavement design, including selection of materials, layer thicknesses, drainage, and the binder type. The contractor is also responsible for main- taining the pavement and seal, including the shape and structural integrity of the pavement, for 12 months after construction. The compliance Assessment requirements are provided in this document for the following param- eters, among others: • Pavement layer compaction; • Pavement stiffness (moduli) or strength; • Surface shape; • Rut depth; • Roughness; • Surface texture (minimum texture depth from “sand patch” test); • chip retention; • Surface waterproofness; and • Saturation before sealing (i.e., moisture content of pave- ment surface prior to sealing). While this specification does provide certain performance parameters for the pavement, it stipulates a 12-month main- tenance period and the document noted that the parameters would be analyzed and considered for potential performance specifications for hMA pavements in the future. international efforts: South africa In 2004, a pavement investigation was completed on the cape Town International Airport, which combined Mar- shall mixture design principles with a variety of reliable performance tests and criteria. The outcomes of the air- port pavement project were reported to be successful and subsequently led to development of similar approaches for highway pavements (Grobler et al. 2004). In addition to standard volumetric specifications, performance properties have been included for design verification in the asphalt mixture design process in South Africa, such as rutting at 100,000 wheel load applications with the accelerated wheel rut tester and the application of the council for Scientific and Industrial Research Wheel Tracking Device until the development of rut depths measuring 0.4 in. (10 mm). In addition, Superpave gyratory compaction is now used along with measurement of the average deformation from 2,000 to 3,000 repetitions (tied to traffic levels) in the Dynamic creep Test. Fatigue resistance was reported to be measured using the four-point bending beam Fatigue test, IDT test, and the Repetitive Strain test. The Modified lottman test and wet Model Mobile load Simulator test were consid- ered to determine the durability of asphalt mixtures in terms of the amount of stripping of binder. In addition, Grobler et al. (2004) developed two performance-based parameters: the comprehensive Rut Resistance Index and the compaction Effort Index. The comprehensive Rut Resistance Index is the a function of (1) the Model Mobile load Simulator rut depths measured at 100,000 load repetitions; (2) number of repetitions of the council for Scientific and Industrial Research wheel tracking to 0.4 in. (10 mm) rutting; and (3) the percentage of voids in the mix (VIM). The compaction Effort Index was reported to be a function of (1) the Marshall VIM, (2) filler-to-bitumen ratio, and (3) number of gyrations in the Superpave gyratory compactor that achieve 93% density. It was reported that the intention was to further develop these parameters for use in establishing pay factors. international efforts: united Kingdom The guide or standard most commonly referenced in rela- tion to asphalt mixtures in the united Kingdom is PD 6691 (Mineral Products Association 2009). To meet these stan- dards, the united Kingdom uses “type testing,” a method in which the appropriate ingredients are selected for the mixture and a target grading and binder content are selected for the use and application. Whenever performance speci- fications are used, trial strips are created and tested. These test strips are placed using conventional paving and com- paction equipment according to bS 594987, and then tested in accordance with the PD 6691 (which includes volumet- ric testing on binder content and gradation, as well as other potential performance requirements such as compaction or wheel tracking) and the bS 594987 (which includes a speci- fication for transport, laying and compaction, and type test- ing protocols) standards. An article by Ellis et al. (2002) discussed the features of the united Kingdom’s performance specifications for sur- facing and base layers of pavements. The benefits of using performance specifications for the road user, infrastructure owner, and construction industry in the united Kingdom were discussed. The benefits were listed in the areas of quality (improved safety and travel time, smoother ride and improved pavement performance, and reduced contractual risk), economy (better use of available funds), innovation (new solutions to benefit all stakeholders), and environ- ment (less congestion near work zones, encouragement of sustainable development, and reduced material waste). A PRS for asphalt materials was implemented into the united Kingdom specifications in 1996 and the clause was primar- ily implemented to: (1) ensure materials reach the standard assumed in the pavement design; (2) allow more scope for contractors to produce the most economic mix design; and (3) ease the introduction of alternative materials into asphalt mixture designs.

27 reSearch on the advancement of Performance SPecificationS for aSPhaLt mixtureS In this section, the past and current research related to the advancement of performance specifications for asphalt mix- tures is summarized. regional research Much of the effort in the initial development of modern-day PRS for asphalt pavements originated with the WesTrack study. In 1999, a forensic review was conducted by huber and Scherocman to assess why some of the WesTrack test sections did not perform as expected. The underlying sig- nificance of this investigation indicated that due to the experimental design and the pavement sections that did not perform as expected, the development of PRS based solely on the WesTrack experiment may not be entirely appropriate. Williams (2004) researched the use of regression models in order to predict rutting depth based on 10 different hMA material properties designated to be performance-predicting variables. The 10 parameters used to determine future per- formance included Superpave mixture design level; aggre- gate gradation; whether or not the asphalt was graded at or above the required performance grade; fine aggregate angularity; complex modulus (to determine susceptibility to rutting); asphalt film thickness; fines-to-binder ratio; asphalt content; and the air voids and VMA of field mixes. In addi- tion, appropriate ranges were determined for each parameter. The findings reported that the three main parameters that can be directly related to rutting and fatigue performance are air void content, asphalt content, and VMA, and reported satisfactory results for predicting fatigue life and rutting potential based on these three parameters. Dave and Koktan (2011) conducted a study in Minnesota in order to predict the performance of certain asphalt mix- tures in that state and determine which mechanical tests were necessary for developing a PRS. one finding of the research identified that the absence of a global performance indicator, or even a few tests that can encompass the over- all performance of the asphalt pavement, is a critical issue. The tests reported as having the highest potential for deter- mining performance of asphalt mixtures included indirect tensile tests, fracture energy tests, the Texas overlay Tester, and the four point bending beam fatigue test. Suggestions were provided for future studies necessary for determin- ing the practicality of a performance test in Minnesota and for determining the feasibility of using performance tests in surrounding states. Research is currently underway in louisiana to develop a PbS for asphalt mixtures (louisiana Transportation Research center 2011). The report proposed that a minimum of 10 rehabilitation projects with well-known traffic data serve as test locations from which field core samples will be tested with the hWTD loaded wheel test, dynamic modulus test, Scb test, and IDT strength test. The use of a number of nondestructive in situ tests, including the Falling Weight Deflectometer (FWD), light Weight Falling Deflectometer, and Portable Seismic Pavement Analyzer will be included to test in situ pavements at the 10 locations. The research is intended to provide accurate information about asphalt pavement performance over a number of years, depending on the mixture design. A study was done by north carolina State university to develop reliable PRS and considered a fatigue model and a rutting model (Kim 2012). The fatigue model is analyzed using a simplified viscoelastic continuum damage (S-VEcD) model, which is a fatigue test dependent on time and temper- ature, and can determine the effects of aggregate size, asphalt content, PG, and recycled asphalt materials on the fatigue endurance limit of mixtures. Findings from the fatigue model reported that larger aggregate size and the presence of RAP will reduce the endurance limit, while increased asphalt con- tent and PG grades will increase the endurance limit. The rutting model that was analyzed using the triaxial repeated load permanent deformation test indicated that the perma- nent strain increased both with the number of loading cycles and as the testing temperature was increased. national research FhWA advertised a request for proposal (Solicitation DTFh61-13-R-00030) in 2013 for a project to develop a PRS for pavement construction. The scope of the study planned is to advance the pavement-related portion of the Strategic highway Research Program (ShRP 2) Report S2-R07-RR-1 (Scott et al. 2014) project, “Performance Specifications for Rapid Renewal,” by further developing and demonstrat- ing the products of the two FhWA projects and to fill any remaining gaps such that PRS becomes a viable option for use during pavement construction. ShRP 2 recently wrapped up this project. Specifically for pavement PRS, the R07 report incorporates the work of two ongoing and concurrent research activities funded by the FhWA Turner–Fairbanks high- way Research center, specifically: (1) DTFh61-08-c-00029 (awarded to Applied Research Associates, Inc., and entitled “Implementation of Jointed Plain concrete Pavement Perfor- mance Related Specification by State highway Agencies”); and (2) DTFh61-08-h-00005 [awarded to north carolina State university for “hot Mix Asphalt (hMA) Performance- Related Specifications based on Viscoelastoplastic con- tinuum Damage (VEPcD) Models”]. There are also a number of current research projects funded through nchRP whose results will impact, to a cer- tain degree, the development of performance-based specifica- tions. A list of some of these projects and a concise description of their scopes are presented in Table 4.

28 NCHRP Project Number NCHRP Project Title Brief Description of Project Scope 9-46 Improved Mix Design, Evaluation, and Materials Management Practices for Hot Mix Asphalt with High Reclaimed Asphalt Pavement Content Develop a mix design and analysis procedure for HMA containing high RAP contents that provide satisfactory long-term performance. Propose changes to existing specifications to account for HMA containing high RAP contents. High RAP content is defined as greater than 25% and may exceed 50%. 9-48 Field versus Laboratory Volumetrics and Mechanical Properties Determine causes of variability and the precision and bias for volumetric and mechanical properties of dense-graded asphalt mixtures measured within and among these three specimen types: (a) laboratory mixed and compacted, (b) plant mixed and laboratory compacted, and (c) plant mixed and field compacted. Prepare a recommended practice for state DOTs to incorporate these results in specifications and criteria for (a) quality assurance, (b) mix design verification or validation, and (c) structural design and forensic studies. 9-54 Long-Term Aging of Asphalt Mixtures for Performance Testing and Prediction Develop a procedure calibrated and validated with field data to simulate long-term aging of asphalt mixtures for performance testing and prediction to establish a methodology for integrating the effects of long-term aging in Pavement ME Design and other mechanistic design and analysis systems. 9-57 Experimental Design for Field Validation of Laboratory Tests to Assess Cracking Resistance of Asphalt Mixtures Select candidate laboratory tests for load- and environment-associated cracking applicable for routine use through a literature review and workshop. Develop experimental design for a series of coordinated field experiments to establish, verify, and validate (a) laboratory-to-field relationships for the candidate tests and (b) criteria for assessing the cracking potential of asphalt mixtures. 20-07 Task 361 Hamburg Wheel-Track Test Equipment Requirements and Improvements to AASHTO T 324 Document capabilities of available commercial Hamburg test equipment. Determine Hamburg test equipment capabilities, components, or design features that ensure proper testing and accurate, reproducible results. Provide proposed revisions to AASHTO T 324 to enable the use of a performance-type specification for Hamburg test equipment. TAblE 4 lIST oF REcEnT nchRP RESEARch PRoJEcTS RElATED To ASPEcTS oF PERFoRMAncE TESTS