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.

157 Chapter 6. Lessons Learned and Conclusion This section discusses some key lessons learned through the SHRP Asphalt Research Program and the implementation of its products. These lessons were gleaned from interviews with those involved and gathered by the research team as they put together all the various inputs. 6.1 CONTINUALLY REFINE ANTICIPATED PRODUCTS AND DELIVERABLES Special Report 202 identified the SHRP Asphalt Research Program as an applied research program with well-defined products and deliverables, viz., “specifications, tests . . . needed to achieve and control the pavement performance desired.” The 1986 Brown Book further defined these deliverables as a performance-related specification for asphalt binder and an asphalt-aggregate mixture analysis system. The 1987 Contracting Plan for SHRP Asphalt Research approved assigned responsibility for the development of the performance-based asphalt binder specification to the A-001 contractor and that for the performance-based specification for AAMAS to the A-006 contractor. (As discussed above, later programmatic changes transferred this responsibility to the A-001 contractor). Finally, as the research progressed from 1988 onward, a dialogue among the program sponsors, the stakeholders in the highway community, and the various oversight bodies led to a consensus that the specification for an AAMAS should be broadened to a specification for asphalt- aggregate mixtures, to include a mixture analysis system, mixing and compaction procedures, and performance tests. Once the research began, these well-defined products and deliverables provided a clear metric against which to measure progress in the various elements of the program. Early on (through about mid-1991) research proceeded, by plan, along many paths and with the conduct of considerable basic research; one example of this is the assessment of the influence of asphalt chemistry on its performance. However, from mid-1991 to the program’s end in 1993, the scope was continually narrowed, with the goal of shifting finite resources to work elements that were judged by the program’s technical management and oversight bodies to be most critical to achieving the defined deliverables. Such decisions were often unpopular, and it is true that promising research activities were terminated or reduced in scope because they were judged— sometimes incorrectly in retrospect—to be unproductive or of peripheral value. 6.2 REALISTICALLY DEFINE TIME AND RESOURCES The highway community is now accustomed to FHWA, NCHRP, SHRP2, and other organizations routinely conducting multi-million-dollar research projects. It is easy to forget how remarkable such projects are as compared to the typical $50,000 projects of the mid-1980s. Aside perhaps from the AASHO Road Test, nothing like SHRP had ever taken place in highway research. This change, of course, was by design—those who planned SHRP and secured its funding successfully argued that the targeted problems were so large and deep-seated that only a program devoting massive funding to their solutions would suffice. The SHRP Asphalt Research program budget was $50 million, still a truly significant amount, even after taking into account 25 years of inflation. Perhaps more noteworthy, though, was the requirement to spend this amount and deliver the two performance-related specifications within a period of only five years, roughly mid-1988 to mid-1993. To meet this requirement, it was necessary for SHRP to solicit proposals and award contracts for the projects shown in

158 Tables 3 and 4 between 1988 and 1990 and for the selected contractors to carry out the research and develop the products between 1988 and 1993. Was this allocation of time and resources (money) realistic? To answer this question, it is necessary to judge how well SHRP delivered the two key products of the asphalt research program. A complete asphalt binder specification and its supporting tests and equipment were delivered in 1993 and, as discussed elsewhere, have been successfully implemented and adopted into U.S. practice. What was adopted looks much like what was delivered in 1993, although it is undeniable that much of the binder specification’s implementation involved further refinement and enhancement of the original SHRP product. So the asphalt binder specification may be considered a qualified success. The situation is not so sanguine for the asphalt-aggregate mixture specification. What appeared to be a complete specification, supporting tests and equipment, and software were also delivered in 1993 (see, for example, report SHRP-A-407). What proved ready to implement and adopt in routine practice was limited, however, to the Superpave volumetric design method, whose hallmarks are the use of the gyratory compactor and well-defined aggregate and mixture volumetric property specifications. While this limited design method has proven robust and extremely useful in improving asphalt pavement performance, it cannot be considered a true performance-related asphalt-aggregate mixture specification. The original performance tests and equipment for permanent deformation and fatigue cracking were too complex and expensive, though they remain in use today for research applications. The Superpave software developed to provide transfer functions and distress prediction models did not function as needed and was not salvageable by a well-funded FHWA research project in the 1990s, though this project did deliver usable performance tests for permanent deformation and fatigue cracking in the 2000s (at a combined cost to FHWA and NCHRP of almost $5 million). Ultimately, it required the development of the MEPDG between 1996 and 2004 (at a cost of more than $9 million in NCHRP funding) to provide the wherewithal for the performance-related asphalt-aggregate mixture specification envisioned by SHRP. This specification has only now become available through NCHRP Projects 9-19, 9-22, 9-22A, and 9-33A. So the SHRP Asphalt Research program delivered perhaps 65% of the key products identified by the program’s planners. Considering everything, this is a solid if not spectacular return for the time and resources committed to the program. Would more time and resources have made a difference? Probably not, for in hindsight, it can be argued that the completion of a viable asphalt-aggregate mixture specification required model development and computing power that were not available in the early 1990s. This technological lack, not time or money, was likely the true limiting factor for SHRP. And it should be noted that the successful development of the MEPDG also required pavement performance data gathered through the SHRP and FHWA LTPP for model calibration and validation. Such data were not available in sufficient quantity and quality until the late 1990s and early 2000s. Even so, the question can be asked whether a different allocation of the available funds during SHRP might have delivered a more finished, viable asphalt-aggregate mixture specification. Interestingly, the authors of both Special Report 202 and the 1987 Contracting Plan for SHRP Asphalt Research provided direction on this issue. To quote the contracting plan: “In the asphalt area, the original report, “America’s Highways: Accelerating the Search for Innovation,” clearly put the dominant focus on asphalt binders. Subsequent discussions by the AASHTO Task Force, the AASHTO Select Committee on Research, and the National Research Council’s SHRP Executive Committee have reinforced this initial vision, placing the primary

159 emphasis on research to improve asphalt binders.” Knowing what we know now, it is clear that diverting funds from the asphalt binder specification to that of the asphalt-aggregate mixture specification during SHRP would have jeopardized the former without significantly improving the later. The planners’ direction was prescient. 6.3 CONTINUALLY CHALLENGE BASIC HYPOTHESES As stated in Chapter 3, the overall objective of the SHRP Asphalt Research program, as articulated in Special Report 202, was to improve pavement performance through an increased understanding of the chemical and physical properties of asphalt cement in the context of its use in pavement. From the beginning, then, investigation of the relationship of asphalt pavement performance and asphalt chemistry was given a co-equal place in the research program with the relationship to asphalt physical behavior. Indeed, the earliest “strawman” specification for asphalt binder developed in 1989 included criteria for nitrogen and acid factor contents to help select materials with adequate resistance to moisture sensitivity. However, succeeding versions of the SHRP strawman specification and the present AASHTO specification M 320, Performance-Graded Asphalt Binders, which developed from the strawmen, rely exclusively on rheology to define the expected performance of asphalt binders. Volume 1 of Report SHRP-A-367 describes the reasons for this ultimate shift from the co-equal status of chemical and physical properties as the basis for the asphalt binder specification: “The original goal of the physical-chemical correlations was to relate asphalt chemistry to pavement performance. This goal was found to be exceedingly optimistic and unrealistic. Physical properties determine the response of a pavement to traffic loading, and there are endless combinations of chemistries that can result in a given value for any of the performance-related binder physical properties. Thus, although it may be possible to define the physical properties needed to provide a certain level of performance, there are innumerable asphalt chemistries that can produce the desired asphalt physical properties. Relationships between asphalt chemistry and pavement performance could undoubtedly be developed empirically by simply correlating chemical properties with percent cracking and other performance-related properties, but this would provide little basic understanding of the real role of asphalt chemistry in determining binder performance.” That asphalt chemistry, which encompasses topics such as crude oil sources and refining methods, could be correlated with asphalt pavement performance was a basic hypothesis of the SHRP Asphalt Research program—remember Bob Farris’s assertion that asphalt “wasn’t as sticky as it used to be.” Given the emphasis on this hypothesis in Special Report 202 and the Brown Book, SHRP might well have pursued development of a solely or predominately chemical-based specification that would have ultimately yielded an unworkable product. Instead, a working consensus developed in SHRP that challenged this hypothesis. This consensus led to development of the rheology-based binder specification, which has proven both workable and practical. 6.4 CLEARLY APPRECIATE PROBLEM SCOPE, SIZE AND COMPLEXITY For any research program, the size of the problem to be solved must be understood. Careful forethought and planning are essential to a successful outcome of the research. Resources, both financial and human, are tied to the estimation of the problem size. A problem that is “discovered” to be of a larger size than anticipated can lead to insufficient resources and

160 cause serious compromises to the research. If the scope of the research is clearly understood prior to commencement, then either sufficient resources can be dedicated or a revised scope can be developed to match resources. In the lead-up to the Special Report 202 there was a meeting held in Dallas, Texas, at which a minority of participants expressed concern that asphalt mixtures were being disregarded in favor of asphalt chemistry, to the detriment of the proposed research effort. The majority remained convinced that mixture design, within the context of then current parameters (aggregate properties, asphalt content, air voids, VMA, etc.), would be supplanted by discoveries to be made. When Special Report 202 was issued in 1984 there was no mention of asphalt mixtures. It was not until the Brown Book was published in May 1986 with proposed research contracts that asphalt mixtures were added. The Brown Book talked about development of models to predict rutting, fatigue cracking and low-temperature cracking of HMA but had only a limited vision about how these might be used. The synopsis of the proposed asphalt mixture specification was only one sentence long. “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.” In reality, the need for the classic HMA parameters remained. And, today they are part of the Superpave mixture design system. Midway through the research program, the realization occurred that the engineering properties being identified in the research were sufficiently complicated to be impractical for low traffic volume pavements where the risk of poor performance was low. Ultimately, the engineering properties from the research were impractical to implement. As a result, the vision that engineering properties of HMA would be measured and matched to the demand of the application never came to fruition. The size of the problem was much larger than had been envisioned. Today the Superpave mix design system is an improved method of volumetric mixture design that falls short of the vision, however brief, presented in the “Brown Book.” In part this is because of low priority originally given to mixture design, which was later “added in” to the research program. In addition, this is because the problem of identifying and measuring engineering properties was much more difficult than had been anticipated. Even today (2012), after an additional 19 years of continued research, the industry is just at the point where implementation of a system of specifying and measuring engineering properties related to pavement performance may be actually realized. Additionally, a greater breadth of expertise was needed. Although statisticians were instrumental in the development of the experiment designs, their talents were not fully employed in the data analysis, likely because of budget constraints and the then unfamiliarity with a statistical approach that is now widely accepted. From the equipment perspective, the absence of mechanical and electrical engineering expertise was evident in problems with the binder direct tension test and mixture indirect tension and shear tests. Another example of too narrow a breadth of expertise was the lack of computer programming expertise, which left much of the heavy lifting to civil engineers with a “gift” or “passion” for programming to develop the Superpave software. There were two major

161 components to the Superpave software. The core software was designed to evaluate performance test results, extract pertinent asphalt mixture material properties and predict performance. The core software was to interface with shell software that would collect project-specific data, manage input of test data files and display performance predictions. More than any other area, the interface between the shell and the core software was the most unworkable. The shell software was developed by a professional software development company. Development of the core software was added to the activities of a graduate student. Documentation of the core software code, not surprisingly, was very limited. Considerable resources were dedicated to development of the core and shell software which today have been discarded. Understanding of the size of the computer programming requirement and using professional software developers could have yielded a workable software program. How much the lack of workable software caused the performance-based tests to be judged too complex for implementations is open to speculation. 6.5 BASE DECISIONS ON OBJECTIVE DATA In any large research program there will be differences in technical opinion. On the one hand, it is important to encourage different thought processes which are crucial to the discovery of new ideas. At the same time, the research program must coordinate numerous competing fields of study that, if left uncoordinated, lead only to new ideas and not to a final product. A great challenge of research programs that are required to develop an implementable product is to determine the balance between allowing additional effort for the study of ideas and the decision to shelve an idea and move on. The challenge for the technical director of the research effort is to ensure that the researchers do not feel unfairly treated and withdraw emotionally from supporting the goal of the research program. If left unchecked, such feelings can at the very least create a negative drag on the program and could cause its disintegration. Two examples arise from the research program. One worked out well; the other did not. In the area of low-temperature cracking the A-003A team investigated the use of the Thermal Stress-Restrained Tensile Test (TSRST) for the measurement of low-temperature cracking susceptibility. The A-005 team was concerned that the test was a torture test and would not yield engineering properties that could be used to predict cracking. Instead, the A-005 investigator wanted to use a creep test measured with indirect tensile creep plus tensile strength. To resolve this difference, the A-003A and A-005 investigators were asked to collaboratively design an experiment to evaluate mixtures with different low-temperature cracking potentials and to each test them using their respective approaches. A week later an experimental plan was developed and several weeks later the results were presented. When the results were presented, the A-003A team compared the cracking temperature for each mixture with the expected performance and demonstrated that the TSRST ranked the mixtures according to expected performance. The A-005A team presented the results of the tensile creep and tensile strength tests and predicted the performance for a specific geographic region. The results concurred with the expected performance. Further, the investigators predicted the cracking temperatures of the mixtures in the TSRST test, which had a much higher cooling rate than that which occurred in nature. The predicted cracking temperature for each mix matched well with the measured cracking temperature. On the basis of these experimental results, the decision was made to use indirect tensile creep and strength in the Superpave mixture specification. The TSRST was designated as a

162 research tool to validate the asphalt binder specification. Once this decision was made, work progressed on using indirect tensile testing for low-temperature cracking. The success of this approach for low-temperature cracking should have encouraged a similar approach for rutting and fatigue cracking. However, because each of the respective researchers had a long history with and was deeply committed to their respective approaches, such a cooperative experiment was never developed. Thus, objective, empirical evidence was not available with which to reach a clear-cut decision. A decision was postponed while considerable capital resources and emotional capital were expended. Finally, under the pressure of deadlines, a decision was made that proved less than satisfactory—adoption of the difficult-to-implement performance tests for rutting and fatigue cracking based on the Superpave Shear Test device. 6.6 PROVIDE STRONG TECHNICAL LEADERSHIP The research program must have strong technical leadership as well as good administrative leadership. Technical competence enhances the legitimacy of the leader in the eyes of the researchers, especially when unfavorable decisions must be made. The technical leader can and should make use of expert panels to review research and recommend action. But the leader should have sufficient technical ability to justify decisions that are made. One challenge a technical leader faces is trying to make judgment of ideas that are in the process of being “discovered” or developed. Sometimes there is not an obvious answer as to whether an idea should be pursued further or set aside. The pressures of time and budget ultimately force decisions to be made, sometimes to the dissatisfaction of the researchers. A strong leader with competent technical skills enhances the chance of success. The SHRP Overview and Integration Report (commonly known as the Brown Book) that laid out objectives and plans for the research program recognized the leadership challenge in stating that “The technical and administrative management of SHRP will be a complex endeavor requiring the most effective management and communication tools available.” These words proved to be very true. Some of the technical controversies which occurred among the team members, such as compaction method, tests and models, pushed the technical leader and the administrative leader to the limit of their abilities. Many of the issues were successfully decided. Others were not. 6.7 ANTICIPATE THE POLITICS OF IDEAS One of the challenges for a research project is to determine which ideas to advance and which to leave behind. The leaders must reconcile the personalities of the lead researchers. Generally, the lead researchers in the Asphalt Research Program had years and decades of research experience. Ideas developed during that time became the basis of moving forward. Typically such research had been done in distinct geographic or academic environments. And generally the researchers felt constrained by the lack of resources (money) to develop their ideas further. The SHRP Asphalt Research Program was viewed as a vehicle to at last provide adequate funding to allow development that was national in scope to occur. As a result, competition for the research contracts was intense among the well known researchers of the day. It has often been stated that contract A-003A, which was tasked with developing asphalt mixture tests to measure engineering properties, and contract A-005, which was tasked with developing material response and performance models, were awarded in the wrong order as is

163 obvious from Figure 42. Indeed, if the goal of the asphalt mixture research program was to select engineering properties to predict mixture performance, then the first order of business should have been to select the properties and the models for material behavior and mixture performance. Then, the contract to develop tests would be constrained to finding appropriate tests to measure the identified properties. Figure 42 Sequencing of A003A and A005 Projects Instead, because the testing contract led and the modeling contract followed, the stage was inevitably set for conflict. As part of testing a mixture there was a need for material response models. So independent of the modeling contract, material property models were selected. A short discussion of models is appropriate for understanding the situation. In simplistic terms a material property model would predict the stress and strain response for a material. As part of this model, a basic type of behavior must be encoded. The behavior of hot-mix asphalt is complex. Hot-mix asphalt is a composite material of compacted granules of aggregate glued together by asphalt binder. The response of systems composed of compacted particles is by itself complex. In addition the properties of the added asphalt binder change with temperature and time of loading. The resulting hot-mix asphalt has numerous types of behavior depending on temperature, time of loading and age. As a result, hot- mix asphalt can react as a linear elastic material, a non-linear elastic material, a visco-elastic material, a visco-plastic material and a plastic material. For any given combination of temperature and time of loading, the HMA can react with one behavior or a combination of the above behaviors. Material models that capture all of this behavior will be complex indeed. From the material response model, stresses and strains can be predicted as load and temperature change. After determining the stress and strain response, performance models are used to predict behavior of the asphalt mix. How much unrecoverable strain (rutting) will occur? Will it crack from fatigue behavior or from low temperature? During the SHRP Asphalt Research program, the modeling contract selected material property models and performance models that did not use the properties coming from the test develop lab tests to characterize material behavior; ie, generate properties identify models to predict performance input material properties into models to predict performance select lab tests to generate material properties needed for performance prediction models A003A Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec Ja n- M ar Ap r- Ju n Ju l-S ep O ct -D ec 1987 1988 1989 1990 1991 1992 1993 A005 A003A A005

164 development contract. And, the modeling contract was developing tests to measure their properties because the tests from the other contract did not work as well for those properties. Why did this occur? Each group had different ideas about how best to model the behavior of HMA. Each approach made sense, but the two approaches were different. Attempts to understand the advantages of each and to compare the differences led to a defensive posture with each camp defending their ideas. Finally, when reconciliation seemed impossible, the decision had to be made. Time and budget no longer allowed the two independent approaches to continue. The A-003A contract was responsible for developing tests. The A-005 contract was responsible for the material response and performance models. And so, the tests developed by A-003A would be used. The A-005 contract would extract material properties from these tests and use them in the models. The result was a less than optimal approach that ended up being non-implementable. The research leader of each contract had defended their ideas and each felt their approach had been discriminated against in the final decision. Also, each leader still believes that their approach is the better one. Today, 19 years after the end of SHRP, neither approach has penetrated the market. Most asphalt industry engineers (both contractor and state DOT) are unaware of the findings of either researcher and almost none have used them. During the time of trying to reconcile the two approaches, a third researcher was used as an expert to understand if the work of the two could be used in a single approach. The net effect of this effort was that during additional NCHRP work to complete the models and performance prediction part of Superpave the third researcher rejected the approach of either of the two SHRP contractors and promoted his own ideas for testing and modeling. Hence the work of the A-003A and A-005 contracts remains largely unused to this day. In the view of the third researcher, the approach he had been developing since the 1960s was the best approach for the prediction of rutting and fatigue cracking. When the third researcher performed post-SHRP research on the “Simple Performance Test,” the SHRP methods were evaluated and rejected in favor of the researcher’s own approach. So, what does all this mean? All these researchers were convinced that their approach was the best approach. The main lesson to be learned from this is that all researchers tend to be vested in their own ideas and it is difficult, perhaps impossible, for them to be objective regarding other possibilities. As a result, special care and attention should be given to fostering cooperative research and subduing defensiveness to challenges of established concepts. 6.8 DEVELOP A TEAM PHILOSOPHY Beyond the sheer technical difficulties confronting the SHRP researchers, there was also the challenge of developing a team philosophy that mediates the disparate personalities of the researchers involved. While different thinking can be the source of new discoveries, if not managed well, differences in philosophy can be destructive to the overall team. Just as in sport or business, strong teams can produce great accomplishments. Teams that break down, that show a lack of respect among the team members, can be very ineffectual.

165 One of the key challenges for the technical and administrative directors of a large research program is to develop and nurture a team mentality. During the asphalt research program there were examples of both good and bad team behavior. From the beginning of the research program, research teams from each of the contracts assembled for a multiday meeting several times a year. These were large meetings attended by 30 to 50 people. Each research team would present plans for experiments and through discussion would receive feedback from other investigators. Therefore the other investigators acted as a sounding board. During and after the experiments, results would be presented to the group. Discussion would help with evaluation of the test results as well as determining if those experiments impacted experiments others were running. On one such occasion an experiment to investigate the mechanism of moisture damage was being discussed. Asphalt binder molecules were known to adhere only to specific sites on the aggregate surface that had suitable chemistry. The hypothesis was that aggregates exhibiting stripping behavior had a lower density of active sites than aggregates known to be resistant to stripping. But the experimental results were puzzling. The granite aggregate was shown to have approximately the same number of active sites as the limestone aggregate, yet the granite was known to be a stripping aggregate. When water was added to the system, the asphalt molecules were detached from the granite but not from the limestone. One of the other investigators started writing equations on an overhead transparency and asked to share his thoughts with the group. Gibbs Free Energy (essentially the energy given off during adsorption of a molecule to a surface) just might explain the results. At the next meeting the investigator showed results to confirm that energy of adsorption did indeed explain the phenomena of stripping. This is an example of team philosophy. Despite working for different organizations and being involved with different research contracts, the exchange of ideas helped solve a problem. These two individuals worked together because there was no competition between them. There are many other examples of team work to solve a problem within the asphalt research program. Strong differences in philosophy, however, can lead to a non-supportive environment. For example, one researcher believed the team was being asked to push the bounds of knowledge and that the search for truth was a pure and admirable goal. This belief fostered the view that the researchers should tell state DOTs and industry what is required to get the truth. In fact, the leaders of the research program considered implementation issues identified by industry and agencies and modified research recommendations for implementation based not only on the findings of the research but also on political necessities. The leaders saw this as being pragmatic and increasing the chances of successful implementation. The researchers saw this as a sign of weakness. Ultimately, this generated a lack of respect between the individuals and created a disruptive force on the team. A team’s performance is enhanced when its members respect one another and work together. This does not connote a “Pollyanna” atmosphere where only sweetness and smiles are present. It does mean that team members can disagree but still respect each other because the team is more important than the individuals of which it is composed.

166 6.9 EXPECT RESEARCHERS TO BE SOLELY DEDICATED TO THE RESEARCH EFFORT Usually researchers are involved in a multiplicity of activities. A university professor may have multiple research projects, teach classes and be involved in university administration. A consultant may have multiple customers and performing research for a particular client may be only one of the activities with which they will be involved. In normal research, multi-tasking is routine. The main difference between the SHRP research and much other research that had gone before it was the imposition of a deadline and the demand for implementable products. Many of the researchers devoted large percentages of their time to the SHRP research. Several worked solely on SHRP research. For them, the asphalt research program became a sole focus. Others who devoted a lower percentage of time found the SHRP research competing with other demands. As a result, their view was not solely focused on the asphalt research program. For a focused research project such as SHRP there are benefits to a sole focus. The technical leader of the A-001 contract would compare the mission of the SHRP Asphalt Research Program to that of the development of the atomic bomb. Technically, it was not known if the mission to develop performance-based properties for asphalt binder and asphalt mixture could be accomplished. It was not clear whether the technology existed to accomplish this mission or if the technology could be developed. As discussed earlier, the lack of performance-based properties and performance prediction for Superpave mixtures is partially based on the inability to completely conquer the technical challenges. Tom Kennedy hypothesized that if the researchers were co-located to a location devoid of outside interferences that the possibility for success would have increased. Given the size of the Asphalt Research Program and the high profile in the technical community, it may have been possible to require that researchers be given a leave of absence from their regular position for the duration of the SHRP research. The lesson to be learned is that if the problem is sufficiently large, and the outcome uncertain, it may be advantageous to dedicate sole effort of the participants to the research effort. 6.10 BUILD A COOPERATIVE COMMUNITY With a new technology as far-reaching as Superpave, it is important to build a community that shares information to help everyone progress up the learning curve. Some of the most memorable legacies of Superpave stem from building these communities or peer groups. The Expert Task Groups are the best example of this. The ETGs have become so important and so influential that their recommendations are carefully considered by AASHTO as changes are made in the standards and guidelines. The ETG concept has now been expanded beyond the Binder and Mix ETGs to also include groups looking into Models, Recycling and Warm Mix Asphalt (though that is called a technical working group (TWG), its aim is the same). The WMA TWG in particular has been very successful in obtaining research funding in support

167 of its highest rated research needs. It is anticipated that ETGs and TWGs will continue to be used in the future to oversee and coordinate efforts in a variety of fields. The User-Producer Groups have also been quite successful. The UPGs are among a very few venues where agencies and industry can gather and share information on a regional basis. This provides a way to transmit a common message to a large group of people and get feedback from a variety of perspectives. The UPGs have evolved, tailoring their activities and programs to meet the needs of their own regions. Recent economic problems and severely reduced opportunities for agency personnel to travel have hampered the ability of some states to fully participate in the UPGs, but it is hoped that this will be a short-lived barrier. As long as the UPGs can put together compelling programs and share worthwhile information, it is anticipated that they will continue to function as a beneficial communication tool. The Lead State Team was another example of a cooperative community. Although it had a limited lifespan, it is a good model to consider for future implementation efforts. Not every state or every contractor needed assistance from the Lead State Team, but for those who did need a little advice, getting it from their peers (state to state, contractor to contractor) was very effective. The all-too prevalent distrust between agencies and industry can be mitigated by such peer exchange. The early days of the Superpave Centers also fostered this cooperative exchange. Frequent meetings between the Centers definitely helped to coordinate research, training and shakedown testing of the new equipment. Lasting friendships and cooperation were established between the Centers and, in most cases, between the Centers and their clients. 6.11 FIND A CHAMPION FOR THE RESEARCH RESULTS As indicated in the report, the main short-coming of the SHRP Asphalt Program was the non-implementation of mixture performance prediction models. Various reasons can be offered for an explanation, including the facts that the task was more difficult than anticipated; inadequate resources were directed to the task, especially the computer software, and others. At the end of the SHRP research effort the products included performance-based mixture tests for rutting, fatigue cracking and low-temperature cracking. The products also included performance prediction models. It would seem the research project goals were met, but none of these products ever saw implementation. Why? For lack of a champion. The rutting and fatigue cracking prediction models were a composite of performance tests developed by the A-003A team that were considered by the A-005 team to be less than desirable. The A-005 team had been directed to extract material properties from the A-003A tests and use them in the performance prediction models. This situation occurred as a compromise since both the A-003A and A-005 teams each developed a set of performance-based tests and performance prediction models. Each team had hoped to convince the A-001 technical coordination team that their set of tests plus prediction models was superior and should be adopted. An outside expert from the University of Maryland helped mediate the technical debates and provided input on what was possible to do. So at the end of SHRP, as implementation began, a contract for continued development of the models was won by the University of Maryland. Consider the situation: • It was clear to the AASHTO Lead States group and FHWA that the performance models were not ready for implementation.

168 • The A-003A researchers agreed with the tests included in the system but disagreed with the performance prediction models and their use in a mix design system. In fact, they continued independent development of their system for several years following SHRP. • The A-005 researchers disagreed with the tests being used to measure performance properties and so continued to develop their tests and the performance prediction models. • The A-001 principal investigator had serious health issues and became a non-participant. • The FHWA was committed to implement SHRP and set up a pooled fund for states to purchase equipment. When problems with the performance prediction became known they put a hold on that equipment and supported continued development with a large research contract. • NCHRP funded a project to develop a simplified version of the Shear Tester to reduce the equipment cost and complexity, major stumbling blocks to implementation. Although successful, the principal investigator of the project had no desire to push for implementation. • The FHWA contract to continue development of the SHRP performance prediction models was won by the University of Maryland. This researcher was strongly invested in dynamic modulus research since the 1960s. He strongly viewed the shear tester as being technically flawed. He became involved in the AASHTO pavement design project (now known as DARWin ME) and used dynamic modulus as the basis for asphalt mixture properties and performance prediction. • Ultimately, he pushed for adoption of the dynamic modulus test in lieu of the A-003A or A-005 tests. He also developed his own modeling and performance prediction which became part of the Mechanistic-Empirical Design Guide, now DARWin ME. • A stand-alone performance evaluation for mix designs was not developed. • Only now (2012), 19 years after the end of SHRP, are states beginning to implement a performance-based test (the Asphalt Mix Performance Test) and performance prediction models (in the MEPDG/DARWin ME), though both differ from those envisioned during SHRP. In the interim, some states have adopted various wheel track testers and other empirical tests. Others continue to design using only the classic volumetric properties. What's the lesson in this? There was not a ready champion for adoption of performance- based tests for asphalt mix. As a result, nothing was implemented. 6.12 RECOGNIZE SIZE OF THE IMPLEMENTATION EFFORT As implementation of Superpave progressed, it became apparent that the scale of the implementation effort would far exceed that of the research phase itself in terms of time, resources and funding. While the SHRP research period lasted for a nominal five years, the implementation efforts have been going on for 18 years and counting. Thousands of pieces of equipment have been purchased for laboratories across the country, and thousands of people have been trained to perform the testing, analyze the results and employ the new technologies. The costs associated with those equipment purchases and training efforts are virtually impossible to quantify. Follow-up research conducted through the NCHRP program alone, however, amounts to over $16 million. Research funded by individual states, industry and other groups likely exceeds that amount.

169 It is safe to say that although many recognized that significant investments would be required to implement the Superpave technology, the total magnitude of that investment was not initially anticipated. Future implementers of large-scale research efforts should be aware of the magnitude of the task ahead of them. As noted before, in some areas, the research under SHRP was not complete when time ran out, leaving a significant amount of follow-up work to be undertaken. Even if the research had been completed to the point of having purchase specifications for equipment, validated testing and analysis methods, etc., the effort required to implement a new system on the scale of Superpave is enormous. The benefits are equally great, however, so the magnitude of the effort should not be a deterrent. Knowing the size of the task ahead, however, should allow for more realistic time frames and budgets to be developed. It should provide an opportunity to put in place measures to ensure that adequate training is available and that resources can be made available when needed. These needs should be anticipated and not underestimated. 6.13 CULTIVATE CONTINUED SUPPORT FOR PROGRAMS (BOTH FINANCIAL AND INTELLECTUAL) Along with recognizing the size of the research and implementation efforts comes the need to cultivate continued support for the programs. The budgetary needs perhaps come first to mind, but there is also a need to support the program intellectually or through the personnel involved. As a long-term research program or implementation effort progresses, the personnel involved will grow and change. It is essential to bring new people “into the fold” and keep the interest level high. The Long-Term Pavement Performance program is one example of this effort. Most of the original state champions (state coordinators) for the program have advanced in their careers or even retired. People occupying those positions now are frequently not familiar with LTPP, its history, its goals, its needs and its benefits. FHWA and the regional LTPP contractors have had to make outreach efforts to inform the new personnel. This is especially true, of course, for the chief executive officers (commissioners, chairmen, etc.), who tend to change about every two years or so. Another aspect of the continued support is planning ahead to ensure that the resources needed for follow-up research and validation efforts are available. The Materials Reference Library has proven to be an excellent resource that allows the work under SHRP to be expanded through numerous projects undertaken since SHRP. Future large-scale materials related projects would be well served by following this model and providing for accessibility of material samples. Archeologists sometimes choose to document the location of an archeological site but leave it in place, undisturbed, for future archeologists to investigate with improved technologies. Engineers and other researchers could learn from their example and preserve samples that can be tested or analyzed in the future. All projects would benefit from ensuring the accessibility of the raw data that was collected during the research. Data maintenance and management is an ongoing commitment, not a onetime effort. As computer technology changes, data may need to be transferred or translated to new formats. Data from the AASHO Road Test, stored at the time on magnetic tapes, has been lost. The LTPP program is currently grappling with data storage and accessibility issues and

170 must plan for the safe storage of backups to preserve the otherwise irreplaceable database amassed over 20 years. So, there is a need to build grassroots support for the program and continue to promote the program. Data should be maintained in a format accessible by future researchers who may have insights into new methods to analyze the data; samples should be preserved, if possible, for future improved testing and evaluation. This is not a task that can ever be completed and checked off the list. 6.14 INVOLVE RESEARCHERS IN IMPLEMENTATION EFFORT In applied research, the researcher should articulate a clear vision as to how the results can be effectively (technically and economically) used by the client. As part of this vision, it is imperative that the research consider the background, expertise and operating environment of the end user. This is an area that was not used fully. During the research period early implementation fell most strongly on the A-001 contract team. Other researchers were involved to a limited extent. It is likely that increased involvement would have fostered a greater sense of team. 6.15 COMMUNICATE WITH THE INTENDED AUDIENCE Communication with the intended audience – the end users – is vitally important to keep them abreast of progress, to help them prepare for what is coming, and to dispel rumors. This communication should be done before the research starts to outline plans and objectives, during the research to keep the program on the radar screen, and after the research during implementation. It is also important to clearly communicate the status of the research. People need to know that findings are preliminary or tentative and that changes are likely, so that they do not expend resources preparing to implement something that changes dramatically later. 6.16 GET THE TECHNOLOGY OUT TO THE AUDIENCE Past experience in some endeavors has shown that rolling a product out before it is ready can be its death knell. If users try something and it does not work as promised, they are unlikely to try again. On the other hand, letting users know what is coming can help them plan ahead. Users can help in the refinement of a product by giving feedback on potential problems or considerations. The important thing is to make sure the users know they are looking at a draft or prototype, not a finished product. In the SHRP Asphalt Research program, the use of the strawman specifications was very useful. Potential users could begin to see what they would be dealing with in the future. Such specifications also served to focus the researchers on the ultimate product they were expected to produce and drive home the need for practical, workable specifications. The provisional standards were also very successful. They showed people that products were coming, but their provisional status made it clear that future refinements could be expected. Users could decide when to dip their toe in the water and when to dive in head first. It is anticipated that AASHTO will continue using the provisional standards for years to come.

171 Other successful examples of getting the technology out to the audience include the loaned binder equipment and the pooled-fund equipment buy. FHWA has continued to offer to lend some equipment to agencies for trial when feasible; examples include outflow meters, safety edge molds, etc. The pooled-fund equipment buy was highly successful and is a model that is being followed in other applications, most notably the AMPT procurement. 6.17 BENCHMARK Benchmarking is an excellent tool for charting and encouraging progress. The Lead State Team implementation surveys benchmarked where states and industry were in regards to Superpave implementation. The annual surveys showed how the technology was growing in acceptance. There are numerous examples where an agency executive would see the survey results and ask the technical staff, “Why are we lagging behind our neighbors? Why aren’t we taking advantage of this new technology?” 6.18 CONCLUDING OBSERVATIONS The Strategic Highway Research Program was an unprecedented, large-scale highway research effort. The asphalt research effort under SHRP led to substantial changes in the entire asphalt industry in the United States, Canada and overseas. The SHRP Asphalt Research Program resulted in the Superpave mix design system, Performance-Graded binder specification, and many supporting new test protocols and equipment. The implementation of the products of this research program stimulated change in every facet of the asphalt industry. Superpave pavements have been shown to perform better, in general, than previous mixes. Overall, Superpave is recognized as one of the major success stories of SHRP. Not all of the technical efforts under the Asphalt Research Program were entirely successful, however. The planned performance prediction models are still being sought through other research efforts. Moisture damage still occurs in pavements, and there is no widely accepted test method to prevent its occurrence. From a non-technical viewpoint, there were ancillary benefits of the SHRP Program. Many young engineers and researchers were brought into the research and/or implementation efforts at an early stage and have gone on to have illustrious careers. More established researchers were able to make a mark and solidify their reputations. Others were scarred by the clash of egos and disputes that arose. A review of the research and implementation efforts from both a technical and a programmatic perspective documents the evolution of the Superpave system. In addition, the review revealed a number of lessons learned. These lessons may benefit future large-scale research programs. • Make decisions transparent and firm. • Document the decisions made. • Ensure strong technical leadership with management skills • Have a clear vision of the scope, size and complexity of the problem.

172 • Recognize the “politics of ideas” and that researchers will defend their positions. • Develop an atmosphere fostering teamwork and cooperation rather than competition. • Try to ensure researchers have the time and resources to be dedicated to the effort. • Build a cooperative community to help others adopt the new technology. • Recognize the size of the implementation effort – it may be even greater than the size of the research endeavor. • Ensure continued support for the implementation process. • Involve researchers in the implementation effort, and users in the research effort. • Communicate clearly with the eventual users of the research results– give them an idea of what is coming but make it clear what is preliminary and what is ready to implement. • Get the technology out to the users – strawman specifications, first article procurements and pooled-fund equipment buys are very effective strategies to get people to try new technologies and get feedback to refine the products. • Benchmark the status before, during and after implementation to document the success – or lack thereof – of the research and implementation effort.