The Superpave Mix Design System: Anatomy of a Research Program (2012)

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.

51 What’s In a Name? The NCHRP 9-6 research project developed recommendations for a mix design system called the Asphalt-Aggregate Mixture Analysis System (AAMAS). During the early days of SHRP before plans were made for a mix design system, there was often reference to AAMAS. But then there was confusion, “Are you talking about NCHRP AAMAS or SHRP AAMAS?” And so in Denver one night after frustration at the confusion, a new term was coined, Mix Design and Analysis System, MiDAS. This not only differentiated SHRP from NCHRP; it had a marketable ring. But, within a short time, SHRP staff in Washington decided that the system should not be called MiDAS. Reasons given were that Midas was the name of a common muffler shop which would not be a flattering comparison. Of course there was also the fable about King Midas, which was considered to be a sad story. After all, King Midas died of starvation because everything he touched turned to gold. It would not be good to have such a negative image as part of the SHRP Program. There also seemed to be some pride of authorship issues. SHRP staff were in charge of the program, and they would retain naming rights. So a new name was coined, SUperior PERforming AsPHALT, or SuperPhalt. So, during the summer of 1991, the system was officially known as SuperPhalt. The retort became, “What is that? A rough concrete pavement with extreme faulting?” Then, in October 1991, at the AASHTO trade fair in Milwaukee where the SuperPhalt system was on display, the message about the name was communicated to SHRP leadership. The response? Well, no we can’t use MiDAS because of the negative image associated with that. And so, a new name was coined, Superpave, short for SUperior PERforming asphalt PAVEments. And, so it is today.

52 4.5.1 Binder Specification and Supporting Tests The binder specification (AASHTO MP 1), as originally configured at the end of SHRP, is shown in Table 5. The grading system, designated “PG” for performance grade is intended to capture the binder’s contribution to pavement performance as measured by permanent deformation, fatigue cracking, and low-temperature cracking. The specification also contains requirements for safety and constructability. The properties at three temperatures to which the pavement will be exposed – high, intermediate, and low – define the binder grade. The properties include the following:  G*/sin δ measured with the dynamic shear rheometer (DSR) on the unaged binder and residue from the rolling thin film oven (RTFO) test;  G*sin δ on the pressure aging vessel (PAV) residue along with stiffness and m-value from the bending beam rheometer (BBR). The objective of the SHRP Asphalt Research team was to provide, in general, tests that captured fundamental properties yet were reliable, simple, and affordable. Where possible, the team was encouraged to use existing test equipment and protocols. That said, the researchers used several approaches to measure fundamental properties or to “condition” the binder. They developed completely new test devices such as the bending beam rheometer (BBR) and pressure aging vessel (PAV). They built upon existing empirical measurements such as ductility to capture failure properties through the direct tension tester (DTT). Borrowing concepts from the chemical industry, the researchers reconfigured rheometers to capture binder properties over a range of temperatures and frequencies. The equipment and test protocols initially developed to support the binder specification included the following:  Bending Beam Rheometer (AASHTO TP 1) for low-temperature stiffness;  Dynamic Shear Rheometer (AASHTO TP 5) for intermediate and high temperature stiffness and phase angle;  Direct Tension Test (AASHTO TP 3) for low-temperature fracture properties; and  Pressure Aging Vessel (PAV) (AASHTO PP 1) to simulate long-term aging. Additional existing binder tests supporting the specification included the following:  Rolling Thin Film Oven Test (RTFOT, AASHTO T 240, and ASTM D2872) to simulate short-term aging;  Rotational Viscometer (ASTM D4402) for high temperature viscosity and constructability;  Flash Point (Cleveland Open Cup, ASTM D92) for safety;  Mass Loss (AASHTO T 240) for volatile loss, and  Solubility (AASHTO T 44) to assure homogeneity (or to assure no contaminants).

53 PERFORMANCE GRADE PG 46- PG 52- PG 58- PG 64- 34 40 46 10 16 22 28 34 40 46 16 22 28 34 40 10 16 22 28 34 40 Average 7-day Maximum Pavement Design Temp, °C <46 <52 <58 <64 Minimum Pavement Design Temperature, °C >-34 >-40 >-46 >-10 >-16 >-22 >-28 >-34 >-40 >-46 >-16 >-22 >-28 >-34 >-40 >-10 >-16 >-22 >-28 >-34 >-40 ORIGINAL BINDER Flash Point Temp, T48, Min °C 230 Viscosity, ASTM D4402, Max, 3 Pa·s, Test Temp, °C 135 Dynamic Shear, TP 5, G*/sin δ, Min, 1.00 kPa, Test Temp @ 10 rad/s. °C 46 52 58 64 ROLLING THIN FILM OVEN (T240) Mass Loss, Max, percent 1.00 Dynamic Shear, TP 5, G*/sin δ, Min, 2.20 kPa, Test Temp @ 10 rad/s. °C 46 52 58 64 PRESSURE AGING VESSEL RESIDUE (PP1) PAV Aging Temperature, °C 90 90 100 100 Dynamic Shear, TP 5, G*sin δ, Max, 5000 kPa, Test Temp @ 10 rad/s. °C 10 7 4 25 22 19 16 13 10 7 25 22 19 16 13 31 28 25 22 19 16 Physical Hardening Report Creep Stiffness, TP1 S, maximum, 300.0 MPa, m-value, Minimum, 0.300 Test temp @ 60 s, °C -24 -30 -36 0 -6 -12 -18 -24 -30 -36 -6 -12 -18 -24 -30 0 -6 -12 -18 -24 -30 Direct Tension, TP3 Failure Strain, Min, 1.0% Test temp @ 1.0 mm/min, °C -24 -30 -36 0 -6 -12 -18 -24 -30 -36 -6 -12 -18 -24 -30 0 -6 -12 -18 -24 -30 Table 5 Performance Graded Asphalt Binder specification (AASHTO MP 1)

54 PERFORMANCE GRADE PG 64- PG 76- PG 82- 10 16 22 28 34 40 10 16 22 28 34 10 16 22 28 34 Average 7-day Maximum Pavement Design Temp, °C <70 <76 <82 Minimum Pavement Design Temperature, °C >-10 >-16 >-22 >-28 >-34 >-40 >-10 >-16 >-22 >-28 >-34 >-10 >-16 >-22 >-28 >-34 ORIGINAL BINDER Flash Point Temp, T48, Min °C 230 Viscosity, ASTM D4402, Max, 3 Pa·s, Test Temp, °C 135 Dynamic Shear, TP 5, G*/sin δ, Min, 1.00 kPa, Test Temp @ 10 rad/s. °C 70 76 82 ROLLING THIN FILM OVEN (T240) Mass Loss, Max, percent 1.00 Dynamic Shear, TP 5, G*/sin δ, Min, 2.20 kPa, Test Temp @ 10 rad/s. °C 70 76 82 PRESSURE AGING VESSEL RESIDUE (PP1) PAV Aging Temperature, °C 100 (110) 100 (110) 100 (110) Dynamic Shear, TP 5, G*sin δ, Max, 5000 kPa, Test Temp @ 10 rad/s. °C 34 31 28 25 22 19 37 34 31 28 25 40 37 34 31 28 Physical Hardening Report Creep Stiffness, TP1 S, maximum, 300.0 MPa, m-value, Minimum, 0.300 Test temp @ 60 s, °C 0 -6 -12 -18 -24 -30 0 -6 -12 -18 -24 0 -6 -12 -18 -24 Direct Tension, TP3 Failure Strain, Min, 1.0% Test temp @ 1.0 mm/min, °C 0 -6 -12 -18 -24 -30 0 -6 -12 -18 -24 0 -6 -12 -18 -24 Table 5 Performance-Graded Asphalt Binder Specification (AASHTO MP 1) Continued

55 4.5.2 Other Binder-Related Products The A-003B team, studying asphalt-aggregate interactions, produced two major products from their work: models for adhesion and stripping, and the net adsorption test. The net adsorption test provided a method for determining the affinity of an asphalt-aggregate pair and its sensitivity to water. Other products included a limestone reactivity test and a test to determine the reactivity of different asphalt-aggregate systems when anti-stripping agents are used. The team concluded that aggregate properties are more influential in adsorption and stripping potential as compared to asphalt properties. Neither the net adsorption test nor the limestone reactivity test, though effective screening tools, is used routinely today. Probably the most significant product from the A-IIR studies was the pavement core tomography work conducted at the University of Southern California under the guidance of Professor Costas Synolakis. This technology has evolved and is being used today in practice. Other significant efforts from the A-IIR contracts assisted the A-002A contractor in understanding specific chemical and physical characteristics that relate to performance. 4.5.3 Mix Design System and Software The Superpave mix design and analysis system, hierarchical in nature and vertically integrated, is illustrated conceptually in Figures 21 and 22. Three levels of design were defined based on traffic with suggested boundary values at 1 million and 10 million ESALs. As shown, all three design levels included a volumetric mix design phase. In levels 2 and 3, accelerated performance-based tests were recommended to facilitate mix optimization for resistance to permanent deformation, fatigue cracking and low-temperature cracking. For level 1, the laboratory mix design involved only volumetric design, which evaluates aggregates and asphalt binders to select a gradation and asphalt binder content to satisfy specified criteria for air voids, voids in mineral aggregate, and voids filled with asphalt. For levels 2 and 3, performance-based tests would be conducted and estimates of distress with time would be made. This would allow the mix design to be optimized with regard to one or more of three distresses; permanent deformation, low-temperature cracking, and fatigue cracking. It was anticipated that the majority of the mix designs would use the level 1 and level 2 procedures, while level 3 would be used for pavements expected to carry very heavy traffic loads (more than 107 ESALs over the anticipated service life) or roadways of critical importance.

56 Level 1 Level 2 Level 3 Volumetrics • Volumetrics • Volumetrics • Performance Tests • Performance Tests • Performance Predictions • Performance Predictions 106 107 ESALs Level 3 also included an optional proof testing scheme that would allow the mix to be subjected to tests simulating the actual traffic and environmental conditions to confirm that the mix actually would perform at the desired level. In level 2 mix design, fewer tests were to be performed at fewer temperatures than for level 3 mix design. Performance-based tests for permanent deformation were to be done at a single effective temperature for permanent deformation. Likewise, tests used to predict fatigue cracking were to be performed at a single effective temperature for fatigue cracking. Low- temperature tensile strength was measured at a single temperature in level 2 design. Level 3 mix design simulated the entire year by breaking it into representative seasons. Performance-based tests for permanent deformation and fatigue cracking were performed over a range of temperatures. A larger slate of tests was proposed to more rigorously evaluate mix response across a greater range of stress. Permanent deformation and fatigue cracking were predicted using mix properties in each of the representative seasons. A summary comparison of level 2 and level 3 is shown in Table 6. Figure 21 Hierarchical Organization of the Superpave Mix Design and Analysis

57 Aggregate Selection Asphalt Selection Volumetric Mixture Design include Moisture Susceptibility Measurement of Performance Based Material Properties Level 1 Permanent Deformation Fatigue Low Temperature Cracking Estimate of Pavement Performance Final Mix Design for Production Level 3*Level 2 Permanent Deformation Fatigue Low Temperature Cracking Estimate of Pavement Performance Field Mixture Control Tests Proof Testing Optional * Level 3 provides the highest reliable estimate of pavement performance V ol um et ri c D es ig n M ec ha ni ca l P ro pe rt ie s F ie ld C on tr ol Figure 22 Flowchart for Superpave Mix Design and Analysis

58 Table 6 Comparison of Level 2 and Level 3 Mix Design Methods Permanent Deformation/ Fatigue Cracking Low-Temperature Cracking Test Types Level 3 considers more states of stress and requires two additional test methods No difference between level 2 and level 3 Test Temperatures Level 3 considers range of temperatures from 4 to 40°C Level 3 considers three temperatures Level 2 uses one effective temperature for fatigue cracking and one for permanent deformation Level 2 considers tensile strength at one temperature only Performance Prediction Level 3 breaks the year into seasons No difference between level 2 and level 3 Level 2 considers the entire year as a single season As originally configured, equipment and test protocols supporting the Superpave mix design system included the following:  Gyratory or Rolling Wheel Compaction.  Short and Long-Term Aging (Forced Draft Oven).  Simple Shear Test for permanent deformation and fatigue cracking.  Indirect Tensile Creep and Strength for low-temperature cracking.  AASHTO T 283 or Environmental Conditioning System for moisture sensitivity. An optional test, the net adsorption test, was available to screen for asphalt-aggregate compatibility. The Superpave software was intended to integrate the specification, mix design and support routines into one program. It was designed to guide the mix design process from beginning to end and provide an orderly, self-contained means for the recording of all test data and analysis results, performance predictions, and other information required for a complete mix design at levels 1, 2 and 3. 4.5.4 Modifier Evaluation Protocol The Superpave practice for modifier evaluation, as originally described in AASHTO Provisional Practice PP 5, provided a framework for identifying the need for a modifier and estimating its performance. Additionally, it facilitated a simple cost comparison of modified vs. unmodified mixes. Finally, it provided guidance on other aspects of modifiers such as purity, toxicity, storage stability and compatibility. The standard was not widely used and was eventually dropped. Some features of PP 5 are incorporated in AASHTO R 15, Standard Practice for Asphalt Additives and Modifiers.

59 Research addressing modified binders was conducted under NCHRP 9-10, Superpave Protocols for Modified Binders, and other research. 4.5.5 The Gyratory Story One of the most visible differences in the current Superpave method of mix design is the gyratory compactor. The story of its selection starts before the SHRP Program began. In January 1987, NCHRP initiated a contract (NCHRP 9-6) called the Asphalt-Aggregate Mixture Analysis System. This work was intended to be a precursor to and in support of the SHRP Program. A major part of the research effort was to look at different methods of laboratory compaction and make a recommendation to be followed in SHRP. Methods investigated included (11): • Marshall compaction (mechanical, static-base, flat face), • Marshall compaction (mechanical, rotating base, slanted face), • Marshall compaction (hand compaction), • Kneading compactor, • Vibratory hammer, • Simulated rolling wheel (quarter circle), • Vibrating, kneading compactor and • Gyratory compactor (Texas 4-inch gyratory). The key method of evaluating each type of compaction was to compare laboratory-compacted specimens to field-compacted specimens. The comparison was based on Marshall stability, resilient modulus, tensile strength and aggregate orientation. One outcome of the NCHRP 9-6 research was the recommendation for gyratory compaction to be used as a part of the preliminary mix design method developed at the direction of the NCHRP project panel. This method became known as the Asphalt-Aggregate Mix Analysis System (AAMAS). In the final report, the Corps of Engineers gyratory test machine was specifically mentioned although a Texas gyratory had been used in the research (11). The later part of this NCHRP study overlapped with the commencement of the SHRP research. So, one of the early questions for the A-006 contract to investigate was which gyratory compactor to specify. 4.5.5.1 History of Gyratory Compaction Gyratory compaction can be traced back to the Texas Department of Highways in 1939. The Department began a study for the design and control of hot-mix asphalt. A key part of that work was the investigation of a laboratory compactor. Two criteria were used: first, the compactor should achieve the final density of the pavement after being subjected to traffic, and second, aggregate break down should approximate break down in the field (12). A total of nine potential compactors were tried including various types of shearing or kneading compactors, impact compaction, static compaction, vibratory compaction, pneumatic tire compaction, and miniature rolling wheel compaction. In the end a shearing compactor was selected. Phillippi, Raines and Love, all of the Texas Department of Highways, developed the manual compactor shown in Figures 23 and 24.

60 The mold was a piece of pipe with an inside diameter of 4 inches and a wall thickness of ½ inch. It was placed between two horizontal plates with an opening ½ inch greater than the height of the mold. Handles were attached to supports clamped to the outside of the mold, and Figure 23 Compacting a Sample using a Manual Texas Gyratory Compactor (circa 1950) Figure 24 Close Up of Mold in Manual Texas Gyratory Compactor

61 the mold was twisted so opposite corners of the mold contacted the top and the bottom plates. By chance, the angle happened to be approximately 5 degrees and 40 minutes. In the early 1950s, the Texas Department of Highways developed a mechanized compactor that faithfully matched results obtained from the manual method. The compaction protocol consisted of groups of three gyrations applied at one gyration per second. At the beginning of each group, the vertical pressure was adjusted to 50 psi. Groups of gyrations continued until one pump of the hydraulic ram created a vertical pressure of 150 psi. If the pressure was less than 150 psi, then another set of gyrations was applied. This became the standard laboratory compaction method. In the 1960s, a large-scale version was developed for base mixtures containing larger aggregate size. The compaction protocol was entirely different. The gyrations were applied continuously at a rate of 30 per minute instead of in groups of three. The angle was an even five degrees. Specimen height was monitored and the compaction was stopped when the specimen height changed less than a specified amount per gyration. One other variation of the Texas 4-inch gyratory compactor was the Oklahoma gyratory compactor. Oklahoma decided to adopt gyratory compaction. They purchased compactors from the only commercial manufacturer. After a period of time, it was discovered that the angle was about one degree less than the Texas version. Rather than invalidate existing mix designs, the Oklahoma Department of Highways decided to keep the angle that had been used rather than adjust to match the Texas compactors. In the late 1950s, John McRae of the U.S. Army Corps of Engineers started development of the Corps of Engineers gyratory test machine (GTM), shown in Figure 25. Intrigued by the principle of gyratory compaction, he developed a compactor that measured changes in mixture response during compaction. Unlike the Texas machine that used three points to hold the angle constant during compaction, the GTM used two points across the diameter of the specimen. The mold was free to rotate about the two points allowing the angle to float. McRae developed parameters to evaluate the mix based on the change of angle. Figure 25 Inventor John McRae and Gerry Huber stand before a Corps of Engineers Gyratory Test Machine, circa 1990

62 In 1959 a delegation from France visited the United States and observed both the Texas gyratory compactor and the Corps of Engineers compactor. Curious about parameters of gyratory compaction, the Laboratoire Central des Ponts et Chausees (LCPC) undertook studies throughout the 1960s. LCPC was in the midst of developing a new method of mix design, and the gyratory compactor was incorporated as part of the new method. The French gyratory compactor, shown in Figure 26, applied a constant angle of one degree. Francis Moutier of the LCPC performed studies on the compactor. By monitoring specimen height, density was tracked during compaction. The relationship of density to the log of the number of gyrations was found to be nearly linear. In the fall of 1990, based on the NCHRP 9-6 findings that gyratory compaction most closely simulated field compaction, the discussion within the SHRP research team focused on which gyratory compactor should be used. This led to an investigation of the history of gyratory compaction. Also in the fall of 1990, the Texas DOT arranged for loan of a 6-inch Texas gyratory to the Asphalt Institute for the research effort. Interestingly, the engineer who arranged for the loan made the assumption that the Texas protocol would be part of Superpave. Post-SHRP he expressed his disappointment that SHRP had not adopted the Texas method. Figure 26 Francis Moutier with Second Generation LCPC Gyratory Compactor, circa 1998

63 4.5.5.2 Selection of SHRP Gyratory Compactor In May 1991, Gerry Huber was part of a panel that travelled to France to review French highway technology. The LCPC laboratories in Nantes, where Francis Moutier was located, was one of the stops. Moutier provided French technical articles that discussed development of the French method of mix design, including developmental studies that had been done. By the summer of 1991 a picture of the gyratory state of knowledge had been developed. It can be summarized as follows.  For the Texas 4-inch compactor there was no information about selection of operating parameters. An AAPT paper in 1952 (12) documented the development of the mechanized compactor, but its development was based solely on matching density results obtained with the manual method.  The Texas 6-inch compactor was interesting in that specimen height was monitored and the end point of compaction was based on a specified rate of change.  The GTM was more of a testing machine than a compactor. In the summer of 1991, the vision of using mixture tests to predict performance still permeated the research vision. As a result, the various parameters that had been developed for the GTM were not interesting to the research effort because they were empirical.  The LCPC gyratory had several interesting components. Work had been done on compaction vs. gyrations for a number of mixtures and had been related to aggregate properties as well as rutting performance in the field. Studies had been done on specimen size parameters and relation to maximum aggregate size. Also, studies had been done on the angle of gyration. So, the technical direction was selected. SHRP would use the principles of the LCPC compactor, but needed to evaluate and make changes to the protocol. 4.5.5.3 Development of the Superpave Gyratory Compactor Decisions on the operating parameters of the proposed compactor were set as follows:  The one degree angle of LCPC was accepted;  The vertical pressure of 600 kPa of LCPC was accepted, but  The LCPC speed of gyration of 6 gyrations per minute was not accepted. LCPC had addressed concern about specimen cooling during the 10 to 20 minute compaction time by installing an electrical heating system around the mold. The slow speed of gyration was based on a 1960s version of the compactor that included a load cell to measure the eccentric force. And, although the eccentric force was not part of the final protocol, the speed remained fixed at 6 gyrations per minute. For SHRP Tom Kennedy decided to investigate higher speeds of gyration. The first step was to obtain a gyratory compactor that could be modified for performing experiments. A Texas 6-inch compactor was obtained on loan from the Texas Department of Transportation. As shown in Figures 27 through 29 it was modified as follows:  To reduce the speed of rotation to 6 gyrations per minute, a frequency modulator was added to the power supply. The Texas standard was 30 gyrations per minute, while the LCPC compacted at 6 gyrations per minute.  The vertical pressure was already adjustable, so it could be matched to LCPC.

64  On this gyratory the angle was induced by a cam on a lever. A new cam was made to change the angle from 5 degrees to one degree. Figure 27 Modified Texas Six-Inch Gyratory Compactor at the Asphalt Institute. Note pressure gauges at top of machine, spring 1991 Figure 28 Turntable and mold. Molds had temporary insulation applied to the outside. Note dial gauge for measuring height of the ram inside the mold, spring 1991.

65 Speeds greater than 30 gyrations per minute were not investigated because the maximum speed of the Texas gyratory was 30 gyrations per minute. Changing the mechanism of the Texas 6-inch gyratory to increase the speed was too difficult. Besides, the time savings obtained by increasing to 45 or 60 gyrations per minute were not as great as increasing the speed from 6 to 30. So the decision was made to use 30 gyrations per minute as the standard. In July 1992 the Rainhart Company delivered the first prototype Superpave compactor, shown in Figure 30. It was available just in time for an open house in Tomah, Wisconsin, the first trial SPS-9 project to produce and place Superpave hot mix. Figure 29 Angle is applied by rotating the handle and raising the side of the mold. Handle is removed before rotation (gyrations) start, spring 1991

66 The mix design for the project had been done on the modified Texas gyratory at the Asphalt Institute but the prototype was to be used for quality control in the FHWA field laboratory. Immediately there were problems. The bearings were undersized for the loads being applied and failed rapidly (see Figure 31). Compliance of the compactor frame was also an issue. The frame visibly flexed during compaction. Figure 30 First Prototype Superpave Compactor. Machine designer, Ed Hamilton, on left. Company owner Larry Hart on right. Taken at Rainhart Company, Austin, Texas, July 1, 1992 Figure 31 Bearing problems on first prototype Superpave gyratory compactor, Tomah, Wisconsin, July 8, 1992

67 The next field trial occurred three weeks later in Waukesha, Wisconsin, on I-43. Again, the mix designs were done at the Asphalt Institute on the modified Texas 6-inch gyratory. For quality control testing that modified Texas gyratory was hauled to Wisconsin in a rented box truck. Jack Weigel, Quality Control Manager of Payne and Dolan, graciously purchased a transformer so that the researchers could get the necessary 220 volt power in the lab trailer at the plant from the 440 volt generator system used to run the hot-mix plant. The next trial field section was scheduled for September 24, 1992, on I-65 at West Lafayette, Indiana. The Rainhart prototype compactor had been redesigned, and it was being used for the design. A week before construction, it was discovered that the Rainhart and the modified Texas 6-inch compactors produced different mix designs. An investigation was launched, and it was quickly discovered that the angle of gyration (externally measured) was different between the two machines. Studies were done and it was determined that the prototype compactor that was supposed to have an angle of 1 degree actually had an angle of 1.27°. In the end it was decided that the angle should be standardized at 1.25°. (See So, How Did the Angle Change?)

68 So, How Did the Angle Change? It was based on a mistake. In 1991 a Texas 6-inch gyratory compactor was modified to change the angle from five degrees to one degree in preparation for laboratory studies about gyratory compaction. The angle was applied with a cam, which raised the side of the mold when rotated. A new cam was made to reduce the lift, but the cam was incorrectly made. In September 1992, the Texas modified-gyratory compactor was discovered to have an angle of 1.27°. Work on the design compaction levels had been done on this compactor, and the decision was made to complete the work with the angle as is. The July 1993 Pacific Rim Conference in Seattle was selected for roll out of Superpave. On July 26th , at the conference, a meeting was held with the SHRP Asphalt Advisory Committee, SHRP executives, FHWA and the researchers to finalize the standard. (See Figure 32.) After review it was agreed that the larger angle should be used, but it was decided to use 1.25° as the angle, not 1.27°. Figure 32 Figure 32 SHRP Asphalt Advisory Committee meeting, July 26, 1993. It was at this meeting that the final decision about the angle for the gyratory compactor was made. Back row L-R Gerry Huber (SHRP A-001 Contract), unidentified, unidentified, Tony Kriech (Asphalt Materials), unidentified, Haleem Tahir (Maryland Department of Transportation), Dale Decker (NAPA), and John d’Angelo (FHWA). Front row L-R Chuck Hughes (former Virginia Research Council), Damien Kulash (SHRP Manager), Gale Page (Florida Department of Transportation), Ed Harrigan (SHRP Asphalt Program Manager), Eric Harm (Illinois Department of Transportation) and Peter Bellin (SHRP loaned staff from Germany)

69 0.25 0.5 0.75 1.0 2.0 Height to Diameter Ratio 4.5.5.4 Density Gradient in Gyratory Specimens The LCPC compactor could be used with 80, 120 and 160 mm diameter molds. LCPC had done some experiments looking at these three mold sizes using 10 mm maximum size and 20 mm maximum size mixtures. They found that the compaction characteristics remained the same as long as the ratio of diameter to maximum particle size remained above 6 (13). The Texas compactor had a six inch mold, so the decision was made to do all SHRP testing with the six inch mold. When specifications were put together for the Superpave gyratory compactor, one concern of the industry was the amount of material that would be required to switch from 4-inch Marshall specimens to 6-inch Superpave specimens. As a result, the specifications were written to include a 100 mm diameter mold in addition to a 150 mm diameter mold. This was done without testing to confirm that the compaction characteristics reported by LCPC held true for North American mixtures. As an aside, the Colorado Department of Transportation is the only agency to adopt use of the 100 mm molds. A second experiment was done to investigate density gradients in the compacted samples. LCPC had standardized a height to diameter ratio of approximately one. The final height of a 160 mm diameter specimen is targeted to be 150 mm. A zone of disturbance approximately equal to the radius of the mold was known to exist within the mold. The zone of disturbance is shown in Figure 33 for different height to diameter ratios. It was known from the LCPC literature that density gradients exist in gyratory-compacted specimens. Tall specimens have a portion at the center that receives less shearing action and hence less compaction. If the specimen is significantly shorter than the diameter, then the compaction effect is reduced because the zone of disturbance cannot form properly. The compaction plate on the other side of the mold interferes with it. Therefore, a decision was made to investigate density gradients in gyratory-compacted specimens in order to select a height to diameter ratio that produced density gradients similar to the density gradient of field-compacted mix. The NCHRP 9-6(1) AAMAS study that was a precursor to SHRP had already investigated density gradients in laboratory-compacted and in field-compacted mixtures. The Figure 33 Effect of Height to Diameter Ratio on Compaction in Gyratory Compactor

70 AAMAS study evaluated several different laboratory compactors. NCHRP 9-6(1) did field studies in Colorado, Michigan, Texas, Virginia and Wyoming in which mixture compaction on the road was monitored and loose mix samples were collected and compacted in several different laboratory compactors. Air voids were measured on cores taken from the road and on laboratory- compacted specimens. Next, the cores or lab-compacted specimens were sawn horizontally into three disks. Air voids were measured in the top, middle and bottom. The difference between the lowest air void piece and the highest air void piece was defined as the density gradient. Table 7 shows the difference in air voids of the middle slice compared to the highest air void slice. The same information is also shown for the Texas Gyratory Shear Compactor (the standard TxDOT design, not the modified version used during SHRP). For both the laboratory specimens and the field cores, the middle third always had lower air voids. About half the time the top slice had the highest air voids. The rest of the time, the bottom slice had the highest air voids. Table 7 NCHRP 9-6(1) AAMAS Comparison of Air Void Gradient (Difference between Middle of Specimen and Lowest Third) (11) Air Void Difference between Lowest and Highest Third of Specimen Project Location Field Core Texas Gyratory Colorado 1.79 1.35 Michigan 0.94 .89 Texas 2.15 1.71 Virginia 2.13 1.52 Wyoming 2.04 1.88 Average 1.81 1.53 Knowing that density gradients occur in both field-compacted and laboratory-compacted specimens, a study was done during SHRP to evaluate density gradients in the Superpave gyratory compactor. Specimens were compacted in the prototype Superpave gyratory compactor then were sawed and cored. Figure 34 shows a core with a height to diameter ratio of 0.75. It is a mix design done with one of the SHRP aggregates, RL, compacted to approximately 7% air voids. Air void measurements of the different pieces are shown in Table 8. Air voids of the uncut specimen were 6.8%. The top slice has the highest air voids at 7.4%, which is 1.5% higher than air voids of the middle slice. Note that air voids of the outside ring are 7.5%, 2% higher than the inside ring or the center. Together the inside ring and the center have a consistent air void content. The maximum difference in density from the top outside ring to the center is 4%. Based on this experiment, the height to diameter ratio was selected to be 0.75, which for 150 mm diameter means the height would be 112.5 mm. Ultimately the specimen height in AASHTO TP 4 (later AASHTO M 312) was set at 115 ± 5 mm.

71 Table 8 Density Gradient in Specimen Compacted on Prototype Gyratory Compactor (14) 4.5.5.5 Field Compaction of Superpave Mixes Early in Superpave implementation there were some issues with achieving density, especially in areas where the Marshall mixes had been easy to compact. Some people investigated Superpave mix compactability by compacting gyratory specimens equal to the lift thickness. So, for example, a 12.5-mm mixture would be compacted 40 mm (1½ inches) tall and the density was found to be much lower than a specimen compacted 115 mm tall. This led to a debate at the Asphalt Mixture Expert Task Group. The point being debated was whether specimen height should be the same as lift thickness. The issue is that compaction efficiency in the mold is more sensitive than in the field. There is a wider range of lift thicknesses that can be used successfully in the field as long as the lift thickness is not too thin. Outside Ring Inside Ring Center Inside Ring Outside Ring Average Top Slice 8% 6.5% 6% 6.5% 8% 7.4% Middle Slice 7% 4% 4% 4% 7% 5.9% Bottom Slice 7.5% 6% 6½% 6% 7.5% 7.0% Average 7.5% 5.5% 5.5% 5.5% 7.5% 6.8% Figure 34 Gyratory Specimens Sliced and Cored for Density Gradient Analysis

72 In the gyratory mold, a change in specimen height means a change in the zone of disturbance and a different compaction efficiency. As a result, the decision was made to keep the compaction efficiency in the mold the same for all mixtures (i.e., to keep the height to diameter ratio the same). During early implementation, field compaction problems became apparent when lifts were too thin. As a result of Florida DOT experience, the recommended minimum lift thickness was set based on the nominal maximum size of the mixture. The desirable lift thickness for coarse-graded mixtures was set at 4 to 6 times the nominal maximum aggregate size (NMAS). For fine-graded mixtures, the desirable lift thickness was set at 3 to 6 times the NMAS. 4.5.5.6 The Corps of Engineers Gyratory Test Machine A significant debate occurred during the latter part of the SHRP Program and continued on for several years afterward. The manufacturer of the gyratory test machine (GTM) believed the SHRP researchers were speaking of the GTM when they talked about gyratory compaction. When they discovered that it was not the GTM being considered, a campaign was started to change the recommendation. This raises the question of what is behind the story and why the gyratory test machine was not selected. As discussed earlier a study of laboratory compaction occurred during the NCHRP 9-6 project. For gyratory compaction, the investigators used the Texas 4-inch Gyratory Shear Compactor in that study. The Gyratory Shear Compactor was identified as best simulating field compaction based on mechanical properties of laboratory-compacted specimens and aggregate orientation in the specimens. It was on the basis of this study that gyratory compaction was selected as the preferred method and a Superpave gyratory compaction protocol was developed. In the conclusions of the NCHRP 9-6 report, the Texas Gyratory Shear Compactor was recommended as being the best for field compaction simulation (11). Also as part of the NCHRP 9-6 study, cores were taken from the field projects for a period of two years and the change in density from traffic loading was monitored. The GTM was used to evaluate densification under traffic. The final NCHRP 9-6 report contained recommendations about the need for continued research into the GTM. The following paragraph is from Section 4.3 (11). Uncommon Tests Gyratory shear strength or the use of the Corps of Engineers GTM was found to provide a reasonable evaluation of asphalt concrete mixtures that were known to be “sensitive” mixtures or mixtures that are susceptible to a reduction in shear strength with traffic. However, this parameter is not used in any mechanistic model nor is it commonly used to evaluate mixtures. Thus, additional mixtures should be evaluated and designed with the GTM and then monitored to gain the critical performance data to validate its results. The GTM manufacturer believed that the 9-6 report recommended use of rather than further research on the Corps of Engineers machine. In the summer of 1991, as the SHRP work was progressing with the modified Texas 6-inch gyratory compactor, the GTM inventor, John McRae, realized that his machine was no longer being considered as a compactor for Superpave, which led to debates paraphrased as follows:  “My machine can do everything you are trying to do in SHRP.”

73  “You mean it can predict rutting and cracking?”  “Well, you are trying to design pavements that don’t rut. It doesn’t matter about predicting how much rutting there is. You just don’t want any rutting to occur and my machine can design mixes that don’t rut.”  “No. Performance prediction is our objective. Other contracts are evaluating mixture properties and performance prediction.”  “My machine measures fundamental engineering properties.”  “What about compaction? Would you make your machine a compactor? What would you change? And how much less would it cost?”  “No, the GTM cannot be changed. The price would stay the same.”  “But currently it is a laboratory machine. It should be made smaller since it has to be used in the field for QC.”  “No it doesn’t need to be smaller; I can make it portable.” And so the conversation went over the course of several months. In response to the interaction, McRae outfitted a GTM with a computer to capture the information as a replacement to the x-y paper plotter that had been used until then. Also, a GTM was mounted in a trailer to demonstrate that the machine was portable. On March 9, 1992, an open house was hosted by the Corps of Engineers to demonstrate the updated equipment. The open house, not held specifically for SHRP, was attended by about 25 representatives from different parts of the asphalt industry. In discussions about the updated equipment it was clear that the inventor strongly believed in the ability of his machine to design an asphalt mixture that would not rut. But, it also was clear that he did not understand the goals and objectives of the SHRP Asphalt Program nor how his compactor would fit into those objectives. He was adamant that the researchers were ignoring the results of the NCHRP 9-6 study and that he was being discriminated against. After review of the situation, Damian Kulash, SHRP Director, asked for a meeting to review the pertinent facts and to hear discussion regarding applicability of the GTM to the goals of the Asphalt Research Program. On April 16, 1992, a meeting was held at the SHRP offices in Washington, DC. It was attended by John McRae and his son John McRae Jr. representing the Engineering Development Company. SHRP was represented by Kulash and Ed Harrigan, the Asphalt Program Manager. The researchers were represented by Tom Kennedy, Technical Director of the Asphalt Program, Gerry Huber of the A-001 contract and Carl Monismith (by phone) of the A-003A contract. The discussion was primarily technical with Professor Monismith debating the technical analysis provided by McRae. In the end, the SHRP staff considered the points being made by the researchers and the GTM inventor. Kulash agreed that the case for the GTM was not compelling and it was removed from further consideration. The rejection only served to strengthen the resolve of the inventor. The campaign became grass roots with packages of technical data and arguments – the same ones aired at the meeting with SHRP – being mailed to various researchers and DOT people in the country along with requests for support in helping SHRP realize the error they were making. At one point after the end of SHRP, a senator from Mississippi intervened on behalf of the GTM inventor. FHWA was asked to support its decision to implement Superpave without the GTM and defend against the accusation that a federal agency was acting in restraint of domestic trade in preference to foreign (French) technology. Ultimately, that defense was made.

74 Discussion of the GTM waned and implementation of the Superpave gyratory compactor continued. 4.5.5.7 Rolling Wheel Compaction A significant debate occurred within the SHRP Asphalt Research team about using a rolling wheel compactor, such as that shown in Figure 35, as the laboratory compaction method. This discussion originated from the A-003A contract and overshadowed research activities for an extended period of time. The A-003A contract was awarded in the fall of 1988, and work started shortly thereafter. One of the early studies was an experiment to investigate the effect of compaction method on mechanical properties of HMA. The NCHRP 9-6 study had looked at several laboratory compaction methods and had made recommendations that the gyratory compactor appeared to be the best as compared to field-compacted mixtures. Rolling wheel simulation and kneading compaction also compared quite well. The A-003A study focused on Texas gyratory compaction, kneading compaction (used in the Hveem mix design method) and rolling wheel compaction. A large experiment was done that included two mix designs with two different asphalt binders. Specimens were tested at design air voids and high air voids. A low and a high asphalt content were used; the low value was obtained by Hveem design and the high value by Marshall design. The main effects studied were rutting and fatigue resistance. Generally the experiment showed that gyratory specimens had a weaker aggregate skeleton leading to poorer rut resistance and better fatigue properties. At the other end of the scale were kneading compacted specimens, which had the strongest aggregate structure. Kneading compacted specimens had the best rut resistance and the poorest fatigue resistance. Rolling wheel compacted specimens were somewhere between. Figure 35 Students Compacting a Specimen by Rolling Wheel at University of California Berkeley, April 1991 (All persons are unidentified)

75 Before continuing discussion about the type of compaction, it is important to understand the thought process in place in 1991 when the compaction experiment was completed. At the beginning of SHRP, the vision for mix design was a new method where properties of candidate mixtures would be measured and performance would be predicted. Today, in the current Mechanistic-Empirical Pavement Design Guide, this is the process that is used to design structural thickness. In 1991, the thought was to design mixtures based on predicted performance. And so, gone would be the days of air voids and VMA and other such empirical properties. In their place would be performance-based properties that would be used in determining an acceptable mix design. In January 1991, the A-001 team conceived the idea of different levels of mix designs where the base level design would be a volumetric design. However, the idea of predicting performance was still considered to be the ultimate goal. Based on results of the compaction experiment the A-003A team began a concerted campaign to have rolling wheel compaction adopted as the compaction method for the new mix design method. The A-003A team developed a draft specification that called for compacting 7 kg of mix in a mold 24 inches by 24 inches by 3 inches deep. The mixture was to be mixed then cured for 15 hours overnight at 60ºC, heated for an hour and a half to compaction temperature, then compacted. After compaction the mix was to remain in the mold and be cooled overnight. The next day it would be cored and cut for testing, as diagrammed in Figure 36. The net effect was a three day process to mix and compact specimens. Figure 36 A-003A Proposed Sawing and Coring of Slab for Rolling Wheel Compacted Specimen. (Slab is in the center of the square) (12)

76 The view of the A-001 team was that rolling wheel compaction was impractical for use in mix design. Although the gyratory-compacted specimens might not have exactly the same properties as rolling wheel compacted specimens, evaluation of compacted mixture gave the same trends for both. To demonstrate that their results could be applied beyond their laboratory experiment the A-003A team compared the properties of lab-compacted mix from two projects in California. Both projects had properties that best correlated with rolling wheel compacted specimens. This further enhanced the case for rolling wheel compaction. Seeking support outside the SHRP community, the A-003A team hosted an open house with representatives from NAPA. The case for using rolling wheel compaction was presented. To allay fears of the difficulty of using rolling wheel compaction, the method of specimen preparation was presented. The argument was presented that instead of producing many specimens for the different types of testing, one compacted slab could provide all the necessary specimens for rut resistance, fatigue resistance and low-temperature cracking. Eight fatigue beams, seven 4-inch cores and two 6-inch cores could be obtained from the single slab, as shown in Figure 36. The reaction of the contracting community was negative. Contractors were used to compacting three 4-inch Marshall specimens on a $1,500 compactor and were already balking at the idea of a $20,000 gyratory compactor to produce 6-inch specimens. The rolling wheel proposal looked even less attractive. In response, the A-001 contractor commissioned the A-005 team at Texas A&M University to perform a study of compaction. The focus of this study was narrower than the A- 003A study. Cores were obtained from five different pavement sections. Some of the cores were tested directly and some were re-heated and compacted using Texas gyratory, rotating base Marshall hammer, Exxon rolling wheel simulator and the ELF linear kneading compactor. Measured properties of the cores and the compacted specimens included stiffness at two temperatures, repeated loading creep and stability (by Marshall and Hveem methods). The properties of specimens prepared with the gyratory compactor were found to be equivalent to those of the cores in 24 of 33 comparisons. The Exxon rolling wheel simulator was equivalent in 19 of 32 comparisons. This data supported the A-001 position that gyratory compaction was at least as good as, maybe better than, a rolling wheel compactor. This study was completed in the spring of 1992 (14). In effect, the thought of the A-001 contractor was that NCHRP 9-6 had answered the question of which compactor should be used. This Texas A&M study was a supplement that supported the NCHRP 9-6 study. On the other hand, the A-003A contractor viewed the report as an indicator of doubt about the goodness of the data developed at UC-Berkeley. The debate about the compaction method went on for a long time. In June 1992, more than a year after the discussion started, there was a proposal to construct a 1000 foot long test section using SHRP aggregate RB (one of the same used in the A-003A experiment) to again compare the different laboratory compaction methods with field compaction. The proposal was not acted upon. All of this contributed to a strained relationship between the A-001 and A-003A research teams. One thought the other was refusing to acknowledge reasonable engineering results. The other thought the first was fixated on technical details without considering practicality. And so, the debate continued. Considerable energy and expense were expended as a result. Perhaps, in retrospect, it should have been recognized at the beginning of SHRP that

77 laboratory compaction still had several unanswered questions and that a study was required with a larger scope than the A-003A experiment. In the end, it probably doesn’t matter who was right, A-001 or A-003A (or both, or neither). Performance prediction as envisioned for Superpave never came to fruition. Only after many more years of research sponsored by the FHWA and NCHRP – and in particular the development of the Mechanistic-Empirical Pavement Design Guide (MEPDG), the Asphalt Mix Performance Tester and supporting test methods – did a systematic method for performance prediction come into being. In retrospect, the approach envisioned for Superpave might have been successful had there been a strong champion for the approach and resources to fund more development of the models and revision of the software. Because of the lack of agreement between the researchers, there was no champion for this effort. The Superpave version implemented at the end of SHRP was a volumetric mix design method rooted in empirical properties of the past. Gyratory compaction works just fine for that. Also, A-001 selected a different gyratory angle. The Texas gyratory used in the A-003A experiment and the NCHRP 9-6 study had a nominal 6° angle. SHRP ended up selecting 1.25° as the angle of gyration. The effect of this change in angle on mechanical properties of the mixture was not investigated as part of SHRP. A few months after SHRP ended, in June 1993, the Asphalt Institute conducted a mix design for an SPS-9 section in Arizona. The new tests proposed by A-003A were performed on the gyratory-compacted specimens and something unexpected was discovered. In the A-003A work, gyratory-compacted specimens were shown to have lower stiffness and less rutting resistance than roller compacted specimens. In the Arizona SPS-9 section, specimens compacted by the Asphalt Institute with the new SHRP gyratory compactor were found to have higher stiffness and rut resistance. Properties of gyratory specimens compacted with the 1.25° angle were different than ones compacted with an angle of 6°. This fact remained undiscovered during SHRP. The reason for the change in mechanical properties with the change in angle was never fully answered. Subsequent research by FHWA and NCHRP found that even small changes in the angle—on the order of several tenths of degree—can have substantial effects on both compaction and mechanical properties. Various hypotheses have been put forward, but the cause of these effects is not well understood, even to this day. 4.5.6 The Delphi Story Early on in the Asphalt Research Program, the emphasis was on asphalt binder research. The A-002 projects were concerned with asphalt binder properties and what could be done differently than in the past. Much of the emphasis was on asphalt chemistry, and it was envisioned that the new specification would be a chemical specification. At the same time, many exotic technologies were being explored, such as acoustic emissions from the poker chip test. The first asphalt binder contract (A-002A), let in the second quarter of 1988, was tasked with identifying binder properties that influenced mix behavior. In the last quarter of 1988, the first asphalt mixture contract was let. This contract was tasked with validating that asphalt binder properties did influence mixture behavior. Also, it was tasked with developing mixture evaluation tests that could be used to measure fundamental engineering properties. A fundamental engineering property was defined as a property that could be used to predict a material’s response to stresses or strains. For example, if asphalt mixture was an elastic material, then modulus could be measured and strain could be calculated for any imposed load.

78 Modulus would be the fundamental engineering property. Unfortunately, the behavior of HMA is more complicated, and, depending on the temperature and the time of load, its behavior may be linear elastic, non-linear elastic, visco-plastic, or plastic. To predict stress or strain in HMA required a material property model that encompassed all of the above. In the last quarter of 1989, Contract A-005 was let. This contract was tasked with developing performance prediction models that would convert stress and strain imposed on the asphalt mixture to expected performance such as rutting, fatigue cracking and low-temperature cracking. So, for example, for an amount of stress or strain imposed on the HMA, the amount of fatigue damage or the amount of non-recoverable deformation (rutting) would be calculated. Throughout 1989 and 1990, work progressed on the validation of properties and development of tests. In the third quarter of 1990, contract A-006 was awarded. This contract was tasked with developing a mix design system using the asphalt binder specification, asphalt mixture tests and asphalt mixture performance models. The first official meeting of the A-006 contract occurred in September 1990. Up until that time, the vision of the new mix design system was very different than what we have today. Then it was envisioned that the mixture tests and performance models that were being developed would give performance predictions and candidate mixtures would be selected based on their predicted performance. Such things as gradation, air voids, VMA, even asphalt binder content, were considered to be concepts of the past. True, it was recognized, as had been in the past, that engineering properties of the mixture were influenced by these volumetric properties. But the vision was to use the new properties to predict performance directly. Perhaps the goal was too large. Perhaps the complexity was underestimated. Perhaps the focus on asphalt binder had reduced proper consideration being given to asphalt mixture. But in the fall of 1990 when the A-006 project commenced, there was some discussion of the old properties. By January 1991, it became clear that the mixture tests and the mixture performance prediction models were going to be much more complicated than anticipated. Work on the tests and models continued, and the final outcome was not yet known, but as the research continued, one thing was clear; it would not make sense to use this system for all mix designs. A graduated mix design system was needed. The full-blown performance prediction made sense for high volume, high-cost projects. For others it did not. And so, the concept of Level 1, Level 2 and Level 3 mix designs was formed. Level 3 mix design would use the entire spectrum of tests to measure mixture properties and mixture performance. Level 2 would be either a simplified version of Level 3 with a performance prediction or it would be a torture test. One such torture test discussed was wheel track rut-testers. Following the European Asphalt Study Tour (EAST), a delegation of senior FHWA, state DOT and industry association personnel returned with information on the French LCPC wheel track tester and proposed it or similar wheel-tracking equipment as a performance test (15). Level 1 mix design was to be based on the old volumetric mixture properties. Such was the necessity of developing a workable system. However, the problem facing A-001 was there was no time or funding available for research on:  the proper level of air voids,  how VMA should be calculated, and  how gradation should be specified.

79 The old properties were needed. The question was “How to get them?” A study of state DOT specifications indicated that there was no consensus, at least in what was being done. And so, the decision was made to use the Delphi method. 4.5.6.1 The Delphi Method The Delphi method is a process for developing consensus among a group of experts. It requires that the experts have a working knowledge of the area of study. The method does not use debate; that is, experts do not debate who is right and who is wrong regarding some property. Instead, the experts are given a series of questionnaires. The first questionnaire defines the area to be investigated. So, on a scale of very strongly disagree, strongly disagree, disagree, neutral, agree, strongly agree and very strongly agree, the participants indicated the relative importance of a property in the mix design system in the first questionnaire. For example, they would react to the statement “Design air voids should be part of the mix design”. There were seven aggregate properties and three mixture properties considered:  gradation limits  crushed faces  natural sand content  LA abrasion  soundness  deleterious content  sand equivalent  air voids  VMA  Voids filled with asphalt Also on Questionnaire 1 there was a set of follow-up questions for each property:  What is the best way to measure the property (say air voids, for example)?  Are there any external influences that would change the level of air voids used?  How does that factor affect air voids? Fourteen experts were selected who represented a balance of state DOTs, industry representatives (NAPA, NCAT, National Aggregates Association and contractors), and university researchers. They received the questionnaire by mail. After the first questionnaire, all of the remaining questionnaires and decision-making happened in a two-day meeting attended by the fourteen experts. After the first questionnaire results were received, the panel members were brought together and the results were presented and discussed. Questionnaire 2 included a shortened version of the first questionnaire, including only the strongly disagree to strongly agree scale about the properties being included in a mix design specification. This was done to see if the discussion influenced their opinion. The second part of Questionnaire 2 included a series of eight scenarios that had been statistically designed. The scenarios described highways that had different traffic, pavement thickness, climate, etc. The participants were asked to give their best judgment of what the specification limit should be for the property. The highway locations included different:

80  precipitation levels,  July temperature,  coldest winter temperature,  traffic level and  depth from the pavement surface. The results from the second questionnaire were tabulated and discussed. A third questionnaire was designed overnight and given to the group the next morning. By the end of the third questionnaire, it was clear which properties the group felt were important. Some of the properties changed during the discussion. For example, percent natural sand was dropped and fine aggregate angularity took its place. In the fourth questionnaire, participants were asked to rank each property as to its importance for performance. The fifth and final questionnaire was given. It built upon the previous scenarios and was used to estimate specification limits. In the end, it was possible to determine what properties should be used, under what circumstances they should be adjusted and what the specification limits should be. Some properties did not change with changes in condition. For example, for low traffic pavements, air voids were suggested to be between to be between 3.4 and 4.9%. For high traffic pavements the range was 3.5 to 4.9%. On the other hand, the average for crushed faces was 67% for low traffic and 84% for high traffic. Some of the properties had a low probability of error, less than one percent, meaning that the experts were in close agreement. The properties with higher error, 5%, tended to be properties on which the group was polarized and could not reach as good a consensus. In the end, it was decided to call toughness, soundness and deleterious content agency or source properties. The other properties became known as consensus properties because a consensus had been reached. These included:  gradation,  coarse aggregate angularity,  fine aggregate angularity,  flat and elongated particle content  clay content (sand equivalent),  air voids,  voids in mineral aggregate and  voids filled with asphalt. Although consensus was reached on these properties, it does not mean that everyone agreed. For example, one of the most contentious discussions was the calculation of VMA. One part of the group felt that aggregate bulk specific gravity (Gsb) should be used. The rest felt that effective specific gravity (Gse) was the correct way because “it represented the true volume of rock in asphalt” and was more easily measured than Gsb, especially for absorptive aggregates. The group remained polarized. Finally the discussion was settled with a discussion of absorption. For states that used Gse, the question of how to account for absorbed asphalt was discussed. For aggregates with significant absorption (typically considered to be greater than 2% water absorption), the amount of absorbed asphalt must be accounted for. This correction required Gsb to be measured – a worst case situation since measurement on absorptive aggregate

81 is more variable. Therefore, the group consensus was to use Gsb for all aggregates, absorptive and non-absorptive alike. Interestingly, this exercise occurred during SHRP. Years later, NCHRP research projects were conducted looking at air voids and VMA. The findings of the NCHRP researchers were based on laboratory test data, not opinions. They found that that the values from the Delphi group in AASHTO M 323 were reasonable and no changes were recommended. As the recommendations from SHRP became known, state DOT engineers considered how to reconcile the findings with their existing specifications. In response to an inquiry from Dave Esch, of the Alaska DOT, the following letter was drafted on February 24, 1993, to explain the basis of the recommendations on gradation. This letter illustrates the types of discussions that were occurring at the time. February 24, 1993 Dear Dave, You raise some interesting points which will be expressed again by others in the future. First of all, let me reassure you that the SHRP team gave careful consideration to the selection of gradation controls. Second, it is not always possible to exchange existing gradation controls one for one with the new SHRP gradation bands. Such is your case in Alaska. Let me explain the rationale used to set SHRP gradation controls. Maximum density lines are not specifically part of the SHRP specs although the controls are built around them and maximum density lines are shown on figures. The most definitive work done on maximum density lines and nominal maximum size is contained in an ASTM paper from the Asphalt Institute. The Asphalt Institute built upon the work of Goode and Lufsey from the Bureau of Public Roads published in AAPT in 1962. The new proposed definition of nominal maximum size and drawing of maximum density lines provided an explanation for some “anomalies” identified by Goode and Lufsey. Incidentally, the definition of nominal maximum size is a more specific interpretation of the ASTM definition. SHRP’s adoption of nominal maximum size and maximum density line is supported by independent adoption by the FHWA. An FHWA Expert Task Group on Volumetric Properties debated the issues at length and considered all available information. Opinions of the ETG were not unanimous; in particular, some views wanted the “max density line” drawn to the actual percent minus 200. A review by the ETG supported adoption of the Asphalt Institute (method). Next, consider the restricted zone. SHRP had opted to specify mixtures with distinct coarse aggregate skeletons in line with the philosophy of European mix designs. European porous asphalt has a coarse aggregate skeleton with 20% air voids. European SMA mixtures use the same coarse aggregate skeleton filled with a bitumen:filler mastic. SHRP desired to specify dense-graded mixtures with a coarse aggregate skeleton and sand asphalt occupying the space within the skeleton. Examination of the attached SHRP gradation controls illustrates the role of the restricted zone in accomplishing the desired mixture. A gradation below the restricted zone must pass above the minimum % passing 2.36 mm (#8) sieve and not enter the restricted zone. To meet VMA requirements several coarse aggregate gradations can be evaluated. Regardless of coarse aggregate gradations, fine aggregate gradation cannot change significantly. In other words, VMA is obtained in SHRP mixtures by changing the coarse aggregate skeleton. The “filling” in the skeleton is sand asphalt. Now let me consider the case you present in your letter. First, notice the wide boundaries in Alaska’s specification. The fine aggregate portion can vary greatly producing very “sandy” mixtures or very coarse mixtures. Indeed the specification band is so wide as to allow significantly different mixtures to be produced, mixtures which would be categorized as two different SHRP mixtures. The broadness of Alaska’s specification is specifically the reason why it cannot be replaced with a single SHRP gradation control. To combat the problems of very wide gradation bands some other states have tried to narrow the range of acceptable gradations making for “tight” gradation specs. Unfortunately, it can be difficult or impossible to achieve adequate VMA. The SHRP gradation controls solve the problem of wide gradation bands without the dilemma of narrow bands. SHRP gradation controls are “narrow” enough to specify mixtures which will meet only one nominal maximum size and at the same time are “wide” enough achieve required VMA levels by building a coarse aggregate skeleton.

82 I hope the above explanation gives you some insight into the logic of SHRP gradation bands. I know that the SHRP gradation controls are significantly different than some existing specifications as you point out and states will be faced with change during the implementation process. We will be faced with many changes originating from the SHRP research. It is my belief that many changes will be justified by benefits received. Gradation controls are a change which is justified. Sincerely, Ed Harrigan cc: H. Tahir 4.5.7 Products from Studies of Moisture Damage: NAT and AASHTO T 283 In addition to the mix design procedure, the SHRP Asphalt Research program led to the development of some additional mix related products. Some of the most notable of these dealt with the question of moisture damage in asphalt pavements. Although many factors contribute to the degradation of asphalt concrete pavement, damage caused by moisture was considered (and remains) a key element in the deterioration of asphalt mixes. Most of the early research, in the late 1930s and early 1940s focused on adhesive failure or stripping rather than cohesive failure. In the 1950s, the immersion-compression test, later adopted as an ASTM standard, was developed to evaluate the moisture sensitivity of compacted asphalt mixes. Also in this decade, researchers concluded that the rate at which stripping occurs depends upon the surface energy of the materials involved. A sonic test to evaluate stripping resistance was introduced. In the early 1960s, Caltrans engineers observed “serious pavement failures … with little evidence of internal stripping,” thus recognizing cohesive failure in the binder as a moisture damage issue. In 1970s, resilient modulus testing was used to show the cycling effect moisture has on asphalt mix stiffness. Other research in the 1970s captured the detrimental effects of water and freeze-thaw cycling and led to the development of a test which measures the retained strength of asphalt compacted cores subjected to defined exposure conditions. This test, commonly known as the “modified Lottman,” was standardized and adopted as AASHTO T 283. It is still widely used today to measure the resistance of compacted asphalt mixes to moisture induced damage. In fact, AASHTO T 283 was integrated in the Superpave mix design system. Research in the 1980s focused on the evaluation of anti-strip additives. Also in this decade, additional tests were introduced: the boiling water, freeze-thaw pedestal and bonding energy tests. The next major contribution to research on moisture sensitivity of asphalt mixes emerged as a result of the SHRP Asphalt Program. As noted previously, a major goal of SHRP was to relate asphalt binder properties to field performance of asphalt concrete mixes. There are three mechanisms by which moisture can degrade the integrity of an asphalt concrete matrix: (1) loss of cohesion (strength) and stiffness of the asphalt film that may be due to several mechanisms; (2) failure of the adhesion (bond) between the aggregate and asphalt (often called stripping), and (3) degradation or fracture of the aggregate, particularly when the mix is subjected to freezing. Research on these mechanisms was addressed in two major contracts: A003A and A003B. The focus of the former was to define water sensitivity of asphalt concrete mixes with respect to performance, including fatigue, rutting, and thermal cracking, and (2) to develop laboratory testing procedures that would predict field performance. The latter examined the chemistry and physics of the asphalt-aggregate bond with emphasis on adhesion

83 and absorption properties. Key products of these studies, the net adsorption test (NAT) and Environmental Conditioning System (ECS), are addressed in the following narrative. 4.5.7.1 Net Adsorption Test The overall objective of the A-003B contractor was to investigate and understand fundamental aspects of asphalt-aggregate interactions including both chemical and physical processes. The critical phenomena explored were adhesion and absorption. A key product that emerged from this research was the net adsorption test (NAT). The NAT provides a method for selecting asphalt-aggregate pairs and determining their compatibility. The test is composed of two steps. First, asphalt is adsorbed onto aggregate from a toluene solution, the amount of asphalt remaining in solution is measured, and the amount of asphalt adsorbed to the aggregate is calculated by difference. Second, water is introduced into the system, asphalt is desorbed from the aggregate surface, the asphalt present in the solution is measured, and the amount remaining on the aggregate surface is calculated. The amount of asphalt remaining on the surface after the desorption step is termed net adsorption. The development of the test occurred in two steps: first, an initial screening of MRL aggregates (both siliceous and calcareous) was performed with three different MRL asphalts. The initial testing used 5 grams of -40 to +80 (passing the #40 sieve and retained on the #80 sieve) mesh washed aggregates. Second, a scaled-up version of the test was developed by the University of Nevada Reno. This somewhat refined procedure, as outlined in SHRPT Test M- 001, employed a sample size of 50 grams of -#4-fraction of unwashed aggregate and commercially available equipment. The researchers recommended the criteria shown in Table 9 and concluded that the interactions between asphalt and aggregate were dominated by aggregate chemistry. Asphalt chemistry also had an influence, though much smaller than that of the aggregate. Table 9 Recommended NAT Criteria Net Adsorption(%) Moisture Sensitivity of Asphalt-Aggregate Pair >70 Acceptable 55 – 69 Marginal <55 Poor Although the NAT was considered an effective screening test (i.e., an indicator of adhesion), it did not consider the cohesion-loss mechanism of water damage. Hence, even acceptable test results did not obviate the need for testing the compacted mix. Other likely reasons for its lack of use include the following: The test procedure requires about 8.5 hours. The fact that it was “optional” in the originally configured Superpave mix design system, and hence, not part of the FHWA’s pooled-fund equipment purchase may also have contributed to its premature demise. 4.5.7.2 ECS vs. AASHTO T283 As noted in Section 4.5.7, the development of tests to determine the water sensitivity of asphalt concrete mixes began in the 1930s. Since then, numerous tests had been (and continue to

84 be) developed in an attempt to identify asphalt concrete mixes susceptible to water damage. However, at the onset of SHRP, none had emerged as acceptable over a wide range of conditions or materials, and none were performance-related. A-003A contract researchers had hypothesized that much of the water damage in pavements was caused by water in the void system; i.e., that most of the water damaged occurred when the void content was in the range typically usually used in construction of dense-graded mixes, about 5 to 12 percent. Furthermore, the researchers noted that although the terms moisture and water were often used interchangeably, there appeared to be a difference between the actions of moisture vapor and liquid water in distress mechanisms such as stripping. To evaluate this hypothesis and the numerous variables affecting water sensitivity, the Environmental Conditioning System (ECS) was designed and fabricated. The ECS consisted of three subsystems: (1) fluid conditioning, in which the specimen is subjected to predetermined levels of water, air, or vapor, and permeability is measured; (2) an environmental chamber that controls the temperature and humidity and encloses the entire loading frame; and (3) the loading system that determines resilient modulus at various times during environmental cycling and also provides continuous repeated loading as needed. The ECS procedure that emerged at the conclusion of the research required testing of cylindrical specimens (4 inches in diameter by 4 inches in height) for six-hour cycles of wet-hot (140ºF) and wet-freeze (-40ºF) while resilient modulus was monitored before and between cycling at 77ºF. The researchers demonstrated that the ECS was capable of discerning the relative differences in performance, as measured by resilient modulus. In addition, the temperature cycling and repeated loading used in the ECS provided a good indicator of long-term mix performance. The ECS test method provided a number of parameters from the tested specimen (e.g., retained resilient modulus, permeability, stripping rate), and stress-strain information at different temperatures during conditioning, through the data-acquisition capability of the system. Finally, the test results suggested that the ECS had better repeatability and required fewer specimens than the widely used AASHTO T 283. Advantages of the ECS cited by the researchers included the following:  permeability monitoring after each conditioning cycle;  reduced variability because of only one specimen set-up; and  application of repeated load throughout the test. Despite the fact that the ECS test results showed reasonable correlation with field performance, it was not included in the Superpave mix design system. Why? The reasons were twofold: equipment cost and specimen configuration. The benefits of the ECS were not deemed sufficient to outweigh the additional cost, at that time estimated to be about $70,000. Decision- makers were concerned that state DOTs, already burdened with purchasing the required binder equipment and the gyratory compactor – the minimum needed to transition from Marshall or Hveem to Superpave – would be reluctant to expend additional funds. Furthermore, most state DOTs were familiar with T 283 or some variation thereof. Finally, for volumetric mix design (Superpave Level 1), gyratory-compacted specimens would be 150mm in diameter. An obvious and expensive question was, how could this mesh with the ECS recommended specimen configuration? Despite its advantages – better repeatability and fewer specimens than the widely used AASHTO T 283 – and the fact that it yielded an indicator of performance, as measures by resilient modulus, it was not adopted.

85 4.5.8 Analysis of a Meeting The SHRP Asphalt Research Program was a high pressure experience for the participants. A meeting held on July 25, 1991 at the Green Building (National Academy of Sciences at 2001 Wisconsin Ave NW) provides an insight into the pressure and its effect on the people involved. This meeting was one of the regularly scheduled contractor meetings at which the researchers from various projects met to discuss experiments and data. These meetings also provided the A-001 team with an opportunity to review the direction of research and make decisions about what should be done next. Relations between Professor Carl Monismith, head of the A-003A contract, Dr. Bob Lytton, head of the A-005 contract, and Dr. Tom Kennedy, head of the A-001 contract, had been growing more strained over the past few months. One issue was the competing ideas from the A- 003A and A-005 teams over models to be used. The main purpose of this meeting was to have the groups present their ideas and make a decision about which direction to go. Bob Lytton had made a presentation about the models proposed by the A-005 team and experiments proposed to investigate them. Dr. Jim Rosenberger, a statistician from Penn State University on the A-001 team, was critiquing the proposal, and the conversation was becoming heated. The emotion and intensity in the room had grown to a breaking point. It culminated with Bob Lytton giving a treatise on Sir Isaac Newton’s development of his model of F=ma. The lecture continued into the lack of innovation in America and the resultant growing trade imbalance with Japan. At this point Mr. Jim Moulthrop of the A-001 team suggested the group take a 10 minute break. As the discussion had degenerated Tom Kennedy sat at the table and made notes on a folded piece of copy paper. During the break he made copies of it and handed it out to members of the A-001 team charged with pulling together a mix design system from the research being done. A copy of the notes is shown in Figure 1. First he laid out the status of the development of a mix analysis method. Then he laid out a plan for what needed to be done.

86 Page 1 1. Gyratory • 1º • 6 rpms Hughes (Chuck Hughes A-006 team) will work Harman (Tom Harman, FHWA) will do study on speed 2. Dynamic creep • Shear can be substituted if the bugs are worked out • Hire Rowe on A-001 to develop theory, etc for Dissipated Energy Master Curve • Ed (Ed Harrigan, SHRP staff) has agreed 3. Run Dynamic creep Master Curve (fatigue) • Same as above • Continue work on bending beam if desired 4. Indirect Creep with fracture, cold temperature 5. Forced draft aging • Moisture? Next page Page 2 1. Letter • Lytton – Monismith together • Quickly evaluate model analysis being done at Berkeley • Identify mechanics types who can sort out • Witczak has volunteered • Again Negotiate Head review group 2. We build MiDAS around the tests on the other page. Others can be substituted if possible, when ready. 3. We need protocols for all tests being conducted. FIGURE XX Notes Made by Tom Kennedy, Technical Leader Asphalt Research Program, July 25, 1991

87 Below are some comments on the points in the Kennedy note. The first page of notes is an outline of what the mix design system would be. Laboratory Compaction The first point (gyratory) deals with the laboratory compactory to be used. It was to be gyratory compaction with a 1° angle and 6 gyrations per minute. The A-003A team had been pressing for rolling wheel compaction to be used in the laboratory. NCHRP Project 9-5 that preceded SHRP had recommended gyratory compaction, and work was progressing on what the gyratory protocol should be. Texas four-inch gyratory compactors had been used in the NCHRP 9-5 project. The Corps of Engineers gyratory had also been discussed. The Kennedy note indicates the gyratory compactor should operate similarly to the LCPC compactor (described in 4.5.3) except for the speed. The slow speed of rotation of the French gyratory had been identified as problematic, so studies were planned to determine a suitable speed. Tom Harman of FHWA was identified by Kennedy to do the work but it was later done at the Asphalt Institute by Huber. Rutting Test The next point on the Kennedy note, “dynamic creep” speaks to a rutting test. The A- 003A team was recommending a repeated-shear-constant-height test to evaluate rutting, but there were still a lot of questions, and it was not clear if it would be possible to work them out. As a fallback, Tom Kennedy had decided that dynamic creep (repeated load axial deformation) could be used. This was a test that was being promoted by Ken Kandhal at the National Center for Asphalt Technology. Tom Kennedy was relatively confident that the repeated shear test could be sorted out and used. As the note says, “(repeated) shear can be substituted if the bugs are worked out”. It was proposed to bring Geoff Rowe on to the A-001 team to evaluate the dynamic creep approach. Tom Kennedy had visited Professor Steve Brown at the University of Nottingham where Geoff Rowe was a graduate student at the time. Ed Harrigan, Program Manager for the Asphalt Research Program, had apparently already agreed. Fatigue Test The A-003A team argued that beam fatigue was the only valid way to measure fatigue properties. However, there were a host of issues that came with the test. - The test had high variability. - It required compaction of a slab and sawing to produce a specimen. - There was an uncertain correlation between laboratory and field results. The A-003A team members were focusing efforts to reduce variability and address ease of sample preparation and testing. The Kennedy note indicates that dynamic creep master curves should be used for fatigue using Geoff Rowe to evaluate the test. Kennedy did not believe that issues with beam fatigue testing can be resolved but, as a consideration to the A-003A team, allows that beam fatigue can proceed. Low-Temperature Cracking The note indicates that indirect tensile creep with fracture will be used. There had been a debate whether the Thermal Stress-Restrained Tensile Test (TSRST) should be used. A

88 cooperative experiment by Dr. Ted Vinson and Dr. Rey Roque had been done, and the A-001 team had already decided to proceed with the indirect tensile creep and tensile strength. Laboratory Aging Protocol Sufficient work had been done to decide that forced draft oven aging would be used. Moisture Damage Moisture damage was still up in the air. The net adsorption test had been developed by Dr. Christine Curtis of Auburn University. Work was continuing on the Environmental Conditioning System but the decision to use Tensile Strength Ratio would not be made until the fall of 1992. Hence the question mark about moisture damage on the note. Kennedy Note, Page 2 The second page of notes speaks to things to be done to finalize the mix design system. The pressure of time is reflected in the comments. The SHRP Program is three years, ten months old. There remains only one year, eight months until the program ends. Decisions must be made if anything is going to be ready. Kennedy is still hopeful that Monismith and Lytton can work together to jointly recommend a materials model and indicates that he will write a letter to that effect. Still, Kennedy is uncertain if the A-005 material model that uses k1 to k5 is better, worse or the same as the C1 to C9 model proposed by A-003A. His need for an outside expert is evident, and he has talked to Dr. Matt Witczak of the University of Maryland about filling that role. Witczak was the main author of the “Brown Book”. He had the technical ability to evaluate the two approaches and was not already involved in the research projects. Interestingly enough, this debate would continue for another year. The final decision regarding which tests and which models to use was made at a meeting at TRB headquarters on August 7, 1992. It was decided that the A-005 team was to use the set of tests developed by the A-003A team and extract information to determine the k1 to k5 used in their models. Ensuring Development of Mixture Design System The risk of failure was weighing heavily on Kennedy. The situation between Monismith and Lytton was becoming intractable and threatened to jeopardize the ability of the Asphalt Research Program to deliver a mixture analysis system. Therefore, Kennedy proposed the mix design system to be a modular system, whereby various components could be plugged or unplugged. Alternate tests, although not as rigorously developed as had been envisaged, were necessary in case no consensus could be reached in the A-003A / A005 debate.

89 4.6 PEOPLE AND ORGANIZATIONS In addition to the program staff, committee members and researchers, there are a couple of other groups or people that deserve special mention. This section describes the spin-off Canadian Strategic Highway Research Program, C-SHRP, and those involved with that parallel effort. It also discusses the important role of graduate students and loaned staff in and after the research phase. 4.6.1 Canadian Strategic Highway Research Program The Canadian Strategic Highway Research Program (C-SHRP) is an example of leveraging international research for local application. C-SHRP was operated by the Transportation Association of Canada (TAC). TAC is similar to AASHTO as an association of highway agencies representing all the provinces in Canada. Unlike AASHTO, which controls the National Cooperative Highway Research Program (NCHRP), TAC had a less direct role in Canadian highway research. By circumstance, plans for SHRP were developing at the end of a Canadian Weights and Dimensions Study sponsored by TAC. Having just experienced the benefit of strong cooperation among the highway agencies, TAC members saw SHRP as another opportunity to achieve benefits through cooperation. The keys to successful Canadian participation in SHRP include  SHRP’s invitation for international participation.  Success of the weights and dimensions study.  Ontario’s willingness to under-write early participation in the SHRP planning process.  Willingness of other provinces to be involved in a high profile project. The Ontario Ministry of Transportation and Communications was the main conduit for Canadian participation in SHRP. In 1985 plans were developing for SHRP, and in September 1985 a workshop was held in Dallas, Texas, to gain input from the research community. Six research areas of study had already been delineated, and a team of technical experts had been appointed to develop a plan. A Canadian participant was included on each team as listed in Table 10. During the same period, Ataur Bacchus of the Ontario Ministry of Transportation and Communications (OMTC) was loaned to TAC to write the original plan for C-SHRP. John Hill of TAC, who had just completed the Canadian Weights and Dimensions Study, was the manager of C-SHRP but left after the first few months. Greg Williams of Alberta Transportation filled in and then joined TAC to become the program manager. The C-SHRP plan defined four roles of TAC within the SHRP effort.  Monitor research and research plans in the U.S. SHRP Program. This led to inclusion of Canadian asphalt binders in the SHRP Materials Reference Library.  Participate in the research activities. This was particularly evident in the complimentary research program for low-temperature cracking and test sections for Long-Term Pavement Performance (LTPP) studies. Canadian sites were included in the SPS5 LTPP Maintenance Treatment study (chip seals, crack sealing, slurry seals, etc.) and SPS-9A LTPP Superpave field site study.

90  Technology transfer. As research results from the SHRP Program became available for implementation, the Canadian participants provided guidance for suitability and implementation in Canada. They also participated in evaluation and training activities during and following SHRP.  Complimentary research. Additional research for Canadian conditions was identified. For example, the need for low-temperature cracking validation led to three controlled field experiments at Lamont, Alberta; Sherbrooke, Quebec; and Hearst, Ontario. Table 10 Canadian Participants in SHRP Pre-Implementation Studies Study Area Name Position and Affiliation Over view and Integration B.J. Hamm Deputy Minister Nova Scotia Department of Transportation Asphalt Richard Langlois Laboratoire Central Ministère de Transports Québec Pavement Performance Tom Christison Senior Research Engineer Alberta Transportation Council Maintenance Lyle Smith Assistant Deputy Minister New Brunswick Department of Transportation Bridge Protection Dave Marett Chief Structural Engineer Municipality of Ottawa-Carleton Cement and Concrete Peter Smith Director of Research and Development Ontario Ministry of Transportation and Communications Snow and Ice Control J.E. Gruspier Research Engineer Ontario Ministry of Transportation and Communications C-SHRP provided several loaned staff to SHRP headquarters in Washington, DC. As well as advancing the activities of SHRP, they helped fulfill the objectives of C-SHRP. Loaned staff included: • Floyd Dukatch of Manitoba Highways and Transportation • Guy Dore, Laval University • Andy Horosko, Saskatchewan Highways and Transportation • Ataur Bacchus, Ontario Ministry of Transportation and Communications • Stella White, Saskatchewan Highways and Transportation • Leonnie Kavanaugh, Manitoba Highways and Transportation Towards the end of SHRP, Sarah Wells joined C-SHRP and worked with Greg Williams in Canadian implementation activities. Later, Steve Goodman joined Williams and Wells. TAC and C-SHRP took the position that benefits of Superpave asphalt technology would be sufficient for agencies to adopt the new technology. Unlike AASHTO, which maintains a set of test methods and standards used as the basis for state DOT specifications, TAC has no such role. Therefore, the mission of C-SHRP was one of demonstration and dissemination of information.

91 The Canadian Technical Asphalt Association (CTAA), which is a technical association holding annual meetings at various locations throughout Canada, became a key tool for dissemination of research results and coordination of implementation activities. C-SHRP worked with CTAA to form the Canadian User-Producer Group for Asphalt (CUPGA) in 1994, a parallel activity to formation of User-Producer Groups in the U.S. CUPGA provided a sounding board for Superpave implementation and acted in a role similar to the AASHTO Lead States Program in the U.S. Presentations at CUPGA included implementation experiences of Canadian agencies and industry as well as updates on implementation activities in the U.S. In addition, C-SHRP displayed new Superpave equipment. To facilitate experience with the new test equipment, C-SHRP purchased the binder equipment and the Ministère de Transport Québec (MTQ) set up a laboratory. MTQ provided testing to other provincial highway agencies as well as training on the equipment. Today, Superpave binder specifications have been widely adopted in Eastern Canada and to a lesser extent in Western Canada. Superpave mix design is commonly used in New Brunswick and Ontario. Quebec uses an alternate form of mix design based on the Superpave gyratory compactor, but using elements of French mix design technology. Newfoundland, Prince Edward Island and Nova Scotia use aggregate specifications from the Superpave specification but continue to design using the Marshall hammer as the compactor. The move toward implementation of Superpave asphalt binder specifications and asphalt mixture design continues. Generally, the participation of C-SHRP in the SHRP Program is viewed as a success. Improved cooperation among the provincial highway agencies in Canada is seen as a key benefit of the experience. 4.6.2 Graduate Students All of the research efforts conducted at universities around the country relied heavily on graduate research assistants. The principal investigators were the driving forces leading the research, but the grad students were in the labs actually performing most of the hands-on work. Graduate students assisted with training efforts and technology transfer through papers and presentations during both the research and implementation phases. Many of the current leaders in the asphalt industry, especially in academic positions, were involved in the SHRP asphalt research and implementation. Many of these graduate students are now professors with their own graduate students, continuing to work on refinements to Superpave and advances in asphalt technology in general. Some have left academia to enter industry as consultants or technical staff at paving associations, contractors, material suppliers, etc. It may not be widely recognized how important the role of the graduate students was during and after the research phase. It is, perhaps, an unintended benefit that the SHRP research developed a strong core of researchers and educators who moved on to influence the industry and future generations of engineers. 4.6.3 Loaned Staff Similar to the graduate students, the loaned staff members of SHRP developed expertise through their association with the program that they then took back to their home agencies.

92 While many SHRP loaned staff members worked in a variety of areas, a few focused on the asphalt research area or later became predominantly asphalt researchers. The experience of working on an unprecedented research effort like SHRP could be very heady stuff. Loaned staff members were sometimes young, lower level employees, though some were also older and more established. For the young members in particular, working alongside internationally recognized researchers and engineers was educational and inspiring. When they returned to their agencies, they had more expertise than their colleagues and could champion the research products. They spread the word about SHRP through their organizations, states and regions. Though the loaned staff was not large in numbers, it proved to be very successful and beneficial to the loaned staff members, their organizations and the SHRP Program as a whole. Loaned staff members came from state DOTs and internationally, spreading the word about SHRP far and wide. 4.7 REFLECTIONS ON THE RESEARCH PROCESS – HINDSIGHT IS 20-20 The following topics are but brief hints of coming attractions in the concluding chapter on “lessons learned.” Asphalt Program Success It was anticipated (if not readily acknowledged) that the success of SHRP would be measured by the results of the asphalt program. By that metric alone, most would agree that SHRP was wildly successful ‒ if for no other reason than the fact that the asphalt paving community (agency and industry personnel) “engaged” (and still do!) in a frequently boisterous, sometimes painful and disarmingly honest debate as to how this new technology could improve pavement performance. The Lead States Program, Superpave Centers, regionally-based User- Producer Groups and the FOCUS newsletter are but four examples of strategic engagement that were initiated during or shortly after SHRP and continue to foster the ongoing dialogue. Still, overselling any product can lead to frustration. Better to acknowledge some of the issues, imperfections, and shortcomings and note the plans for improvement. Consider Microsoft and Apple. We no longer use DOS despite the “bugs” in the numerous versions of the Windows operating system. Apple’s iPhones are wildly popular despite the antenna issues with version 4. Asphalt Program Scope, Objectives, Size and Complexity The 1984 Blue Book made no mention of asphalt mixes. It was not until the Brown Book was published in May 1986 with proposed research contracts that asphalt mixes were mentioned, and then only briefly, as evidenced by the following: “The asphalt research program will culminate in the preparation of performance-based specifications for asphalt and recommendations for an asphalt-aggregate mixture analysis system using modified or unmodified asphalts.” Was this the classic private-sector nightmare of budget-busting, schedule-sabotaging scope creep? Today the Superpave mix design system imposes more stringent requirements on component materials (binder and aggregate) and volumetric properties of the composite, yet falls

93 far short of the initial vision of a mixture analysis system. In part this is because of the initial obsession with binder chemistry and only superficial focus on asphalt-aggregate mixes. Admittedly, the problem of identifying and measuring key engineering properties of mixtures was much more difficult than had been anticipated, and the actual effort required to develop calibrated and validated performance prediction models—as evidenced by the $9 million expended by NCHRP between 1995 and 2004 to develop the Mechanistic-Empirical Pavement Design Guide—far exceeded the resources available to SHRP. Even today, in 2011, 18 years after the conclusion of SHRP, the paving community has just reached the threshold of implementing a system of specifying and measuring engineering properties for a pavement. Asphalt Program Structure and Schedule The Brown Book recognized that the “technical and administrative management of SHRP will be a complex endeavor requiring the most effective management and communication tools available” (1). Although numerous institutional arrangements were considered in the Blue and Brown Books, SHRP was eventually administered as a new operating unit of the National Research Council. Like the Second World War Manhattan project, the asphalt program goal was clear, the direction less so; the schedule tight and the budget finite. Unlike the Manhattan project and the TRB/ AASHO Road Test, the researchers were not co-located and sequestered in Los Alamos, New Mexico, or Ottawa, Illinois, respectively. Furthermore, the researchers were not “hand- picked” but selected through a competitive process. Expert Task Groups (ETGs) were assembled to provide external technical review of each major contract. Despite the SHRP mantra to “attract new blood,” understandably, most of the key researchers were well established and highly regarded for their long-term efforts in particular areas of the asphalt arena. Because of the accelerated schedule, contracts with intersecting or overlapping goals had to be awarded in parallel rather than in series. Did these factors alter the management team’s ability to corral, both literally and figuratively, the researchers’ ideas? Accommodate the personalities? Appease the egos? Establish esprit d’corps? Was it naïve to think that a renowned researcher would readily abandon his approach (upon which his career was built) in deference to his “competition?” Did the geographic hubs of asphalt research bear an eerie resemblance to the industry fragmentation referred to in the Blue Book? Though great technical expertise was assembled, were effective management skills more critical to success? Ability and willingness to coax and cajole the personalities and egos into a cohesive unit? Clear and concise communication? Transparent, timely and well-documented decision-making? Despite the personal, practical, and logistical challenges associated with co-location of the research team, the anticipated benefits were numerous:  daily and frequent interaction to reinforce the “team” goal, develop and nurture the team mentality (esprit d’corps), and foster cooperation rather than competition;  clear statement of and recurring emphasis on the research goal: product-oriented, readily-implementable ‒ perhaps a somewhat foreign concept to most academicians, for whom the common goal is publication of research results in a peer-reviewed journal;  a sole focus which eliminates the competing interests of the academic environment (teaching, other research and “service” to the department/college/university/community); and

94  regular reinforcement of the evolutionary nature of the research: only those ideas/concepts/products that furthered the goal would be pursued; all others would be jettisoned ‒ another concept with which most academic researchers were unfamiliar. The quarterly, pro bono reviews of voluminous technical reports by the ETGs, were, not surprisingly, uneven at best – superficial at worst. Rather than a large group (eight to ten) of unpaid “expert advisors,” in future SHRP-type research endeavors, external, paid peer review should be considered because it might be more effective. Asphalt Program Expertise With few exceptions, namely chemists and statisticians, the key researchers were civil engineers. Given the keen focus on binder chemistry, the chemists played a prominent role; the statisticians a less visible but critical role, especially in terms of experimental design and data analysis. Despite the stated goal of “using the research results to develop standard tests” and the readily acknowledged importance of data management and implementation, the expertise needed to support these critical elements was assembled ad hoc or very “late in the game.” The breadth of expertise needed might have included the following:  mechanical and electrical engineers to assist with test equipment development;  computer programmers for database structure, software development, and documentation; and  marketers for implementation (early involvement with researchers and outreach to industry and stakeholders, political support and long-term, stable source of funding). Obviously, provision of these staff would have required additional resources or diversion of resources away from the key objectives established for the program.