Specifying and Measuring Asphalt Pavement Density to Ensure Pavement Performance (2017)

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.

6This chapter presents the results of the survey conducted to determine the density of asphalt pavements. Each question in the survey is highlighted followed by the background information, survey results, and an analysis of the results. This chapter is presented in the following format: • A survey question is stated. • Background information from the literature review is presented. • The survey results are presented and analyzed. General Issues in the Density of Asphalt Pavements As seen in Pauls and Goode, the measurement of density in asphalt pavements has been well established in literature, at least for the last 80 years (Pauls and Goode, 1939). The process of compaction is to reduce the air voids in the mixture thereby increasing the density of the in-place mixture. This process reorients the aggregate particles to a dense configuration, thereby increasing the strength of the mixture. This reorientation of the aggregate particles occurs only when adequate temperature is available in the mix to allow the particles to move. In order to achieve adequate density in the asphalt pavement, many factors need to be con- sidered. Epps et al. (1969) and Hughes (1989) discuss the factors involved. Table 1 presents a listing of materials properties, initial density (at time of construction), and traffic that are the key factors for technical considerations in achieving density in asphalt pavements. Figure 1 is a Survey to Determine the Density of Asphalt Pavements C h a p t e r 2 Table 1. Factors affecting the density of asphalt pavements. Materials Properties Initial Density TrafficAsphalt Binder Aggregate Temperature susceptibility Surface texture Subgrade support Vehicle weight and type Rheology Shape Lift thickness Axle configuration Maximum size Equipment Lane distribution Gradation Rolling sequence Daily distribution Absorption Rolling procedures Yearly distribution Base temperature Moisture Wind velocity Ambient temperature Pavement temperature Season of the year Source: Hughes, 1989; Epps et al., 1969.

Survey to Determine the Density of asphalt pavements 7 flow chart illustrating how these factors affect the desirable properties of the asphalt concrete mixture in-place. The Guide to Asphalt Paving (2016) provides an excellent overview of the theory and prac- tice of achieving density in an asphalt pavement. The report comments that the relative cost of aggregate and asphalt binder is much higher than the cost of the compaction process, yet the effect of achieving density relative to the pavement life is equal to the properties of the materials being used to produce the mix. The conclusion from the discussion is that achiev- ing density in the in-place mixture is equal in importance to the pavement life as is the quality of the materials being used. Survey Question 1 Please rank the following issues in order of importance to achieving density of asphalt pavements (1 is the most important and 7 is the least important). The selection options identified in the survey were roller operations, mix design, binder content, aggregate properties, paver operations, environmental factors, and binder stiffness. Background It can be argued that all of the factors listed in the survey are important. These factors also have subsets that significantly affect the achievement of density in asphalt pavements. Participants were asked to respond based on what they thought the relative importance of each factor to be in the relationship between density and pavement performance. Survey Analysis Survey participants were asked to rank the following issues in order of relative importance to achieving density of asphalt pavements. Table 2 presents the items in order of importance from the survey. The ranking value is based on a weighted percentage of responses for each factor. Three dis- tinct groups of “importance” were noted in the analysis of data. Roller operations, mix design, and binder content were ranked closely to each other. Aggregate properties and paver operations were also ranked closely together but significantly lower than the first group. The last group of environmental factors and binder stiffness received rankings substantially below the other two groups, almost half of the score of the top category. Survey Question 2 What weaknesses do you see in our ability to measure and control density on the roadway? Figure 1. Pavement analysis flow chart. Materials Properties Initial Density Traffic Final Density Desirable Pavement Properties •Stability •Durability •Strength •Fatigue Resistance •Stiffness •Flexibility Source: Hughes, 1989; Epps et al., 1969.

8 Specifying and Measuring asphalt pavement Density to ensure pavement performance Background With the list of factors in Table 1, it is easy to understand the complexity of the process for achieving density. With so many elements required for success in achieving density, it is easy to understand how things can and do, at times, go wrong. Warren (1996) proposes a method to account for the many variables involved in the paving process. He describes four factors that are key to “balancing production,” namely plant produc- tion, hauling, paver operations, and rolling operations. There are many variables but these four factors must be in balance to achieve a consistent, uniform paving operation. Communication between all parties is also critical to the entire process. If the industry hopes to be able to measure and control the density of the asphalt concrete, the right personnel must be available and they must have the proper tools to both achieve density and measure it. Survey Analysis Survey participants were asked to identify weaknesses in the ability to achieve density. The issues shown in Table 3 were specifically identified and ranked. Respondents believed that the major issue was trained operators. Adequate personnel and the method of measurement were the second grouping of responses, with density measurement equip- ment, “other,” and compaction equipment populating the third group. Laydown equipment did not receive a significant percentage of responses. Among significant factors in the “other” category were: • Effect of underlying layers, • Unplanned schedule interruptions (accidents, weather, breakdowns, etc.), • Low bid contract environment encouraging low binder content, • Specifications requiring overly stiff and dry mixes, • QC/QA people with little experience, and • Too many mixes. Table 2. Important issues to achieving density. Ranking Issue Weighted Score from Survey 1 Roller operations 5.45 2 Mix design 5.17 3 Binder content 5.06 4 Aggregate properties 3.44 5 Paver operations 3.22 6 Environmental factors 2.87 7 Binder stiffness 2.83 Table 3. Weakness in the ability to achieve density. Ranking Issue Percentage of Responses 1 Trained operators 78.8 2 Adequate QC/QA personnel 46.5 3 Method of density measurement 45.5 4 Density measurement equipment 33.3 5 Other 28.3 6 Compaction equipment 27.3 7 Laydown equipment 15.2 Note: QC/QA = quality control/quality assurance.

Survey to Determine the Density of asphalt pavements 9 Survey Question 3 Is not achieving density a routine problem for your organization? Background Concerns are often expressed by agencies about achieving proper density in asphalt pave- ments and, commonly, the concern is with the density being too low. Brown and Cross (1992) reported that in-place density above 97 percent (less than 3 percent air voids) may increase the probability of premature rutting throughout the life of the pavement. Brown (1990) suggests that the specification should be written to ensure that maximum voids do not exceed 8 percent and minimum voids are above 3 percent. Low density conditions may present themselves years later with poor pavement performance. Survey Analysis Participants were asked if achieving specified density was a problem for their projects. More than 80 percent of the respondents indicated that achieving density was generally not a problem for their projects. While the responses indicated that low density might not be routine, many comments regarding difficulty in achieving density were made. Comments included: • Most projects have marginal density. • Projects may run on the high side of the binder content that may compromise other perfor- mance parameters. • Projects may have a minimum binder content that may compromise the durability of the mixture. • Designers may specify too stiff a binder, thereby making achieving density difficult. • Placing the mix at the low side of voids will yield higher density but may compromise the performance of mix. Survey Question 4 Do you believe that your density specification is adequate for successful pavement performance? Background While the importance of density is well established, the value used for acceptance is not uni- versal. There is not even a consensus on the reference density that is used for the calculation of the percent density. This question was the first in a series to determine the type of specification and the acceptance value for density. Survey Analysis The survey revealed that 83 percent of the respondents believed that their density specification was adequate. Comments included: • Consideration is being given to increasing the density requirement both for mainline and joint density. • Many respondents are migrating to a percent within limits (PWL) specification. • Comments were made that the specification is not always rigorously and uniformly applied. • Specifications are met most uniformly when incentives are involved.

10 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Question 5 Do you think that under-compaction of an asphalt mixture (high in-place voids) creates pavement performance problems? How about over-compaction (low in-place air voids)? Background Under-compaction of asphalt concrete results in a lower density and higher in-place voids. Over-compaction results in a higher density and lower in-place voids. An upper and lower limit for air voids has been common in asphalt concrete specifications for many years, has been required by state specifications for many years, and was recommended by Hughes (1989), Cominsky et al. (1998), and The Asphalt Handbook (Asphalt Institute, 2007). Survey Analysis Participants were asked their opinion about under- and over-compaction of asphalt mixtures. As high as 98 percent of the respondents either moderately or strongly believed that under- compaction affected pavement performance. Only 64 percent of the respondents either moder- ately or strongly believed that over-compaction affected pavement performance. And roughly one-third of the respondents were neutral, disagreed, or strongly disagreed with over-compaction being a performance issue for the pavement. Survey Question 6 What life expectancy does your organization assign to an asphalt pavement? Background Similar to any purchased product, the issue of how long an asphalt pavement will last has been a concern from the beginning of using asphalt pavements. Age ranges provided for this question were somewhat arbitrary but useful to understand the expectations of respondents. The response options were for an overlay, a new pavement, or perpetual pavement. With a well-established highway system such as exists in much of North America, overlays using asphalt concrete are by far the most common type of application although, based on conversations with the FHWA, no verifiable data to quantify a percentage exists (Aschenbrener, 2016). The principle issue for the placement of an overlay is the underlying condition of the pavement structure. The Asphalt Institute describes three major categories of evaluation that need to occur during the preparation stages for an asphalt pavement overlay: (1) assess the functional characteristics (ride quality and surface friction), (2) conduct condition and distress surveys, and (3) perform structural testing (non-destructive and destructive) (Asphalt Overlay for Highway and Street Rehabilitation, 2001). The traffic, environment, type of underlying material, thickness and prop- erties of the individual layers, and the condition of the existing pavement all have a significant impact on the ability of the contractor to place the overlay as well as the performance of the overlay. For new construction, there is often a greater likelihood for better control of the underly- ing conditions, given that the contractor is building a complete structure and may not have to handle the maintenance of traffic. In the United States, and likely most of the developed world, the predominance of asphalt concrete placement is for overlays, not new construction. Long-lasting asphalt concrete pavements have been built for decades in the United States (Newcomb et al., 2001). These full-depth asphalt pavements have performed well in many areas.

Survey to Determine the Density of asphalt pavements 11 The Asphalt Pavement Alliance has defined these long-lasting pavements as perpetual pave- ments. The definition is “an asphalt pavement designed and built to last longer than 50 years without requiring major structural rehabilitation or reconstruction, and needing only periodic surface renewal in response to distresses confined to the top of the pavement” (Newcomb, 2002). Fundamentally, the concept of a perpetual pavement is that the pavement structure is built to last for decades while the wearing surface is expected to be replaced at intervals during the life of the pavement. Excellent descriptions of the perpetual pavement concept are presented by Newcomb et al. (2001, 2010) and Newcomb (2002, 2009). Survey Analysis When asked about the life expectancy assigned to asphalt pavements, the respondents provided the information shown in Table 4—50 percent indicated the life expectancy for an overlay to be 5 to 10 years while 43 percent indicated the life expectancy for an overlay to be 11 to 15 years. Approximately 93 percent of respondents indicated overlay life expectancy to be between 5 and 15 years. For new constructions, life expectancy increases, with 38 percent of respondents indicating a life expectancy of 11 to 15 years and 50 percent indicating 16 to 20 years. Interestingly, about 90 percent of the respondents were in the two time ranges. Lastly, for perpetual pavements, the life expectancy increases again with about 35 percent of respondents indicating a 16 to 20 year life expectancy and about 54 percent indicating 21 to 25 year life expectancy. Once again about 90 percent of respondents were within the two higher time ranges. Specification Types for the Control of Density The type of specification has a significant impact on the way the project is bid, managed, and performed. Over the years, a wide variety of specification types have been used, some more suc- cessfully than others. It is important for all parties to understand the type of specification and the potential impact on the project. Several different types of specifications were identified in Transportation Research Circular E-C173. A brief synthesized description is presented in Table 5. Each type of specification has advantages and disadvantages as can be observed in the description. All of these approaches have been used success- fully for specific projects. The overall process for assuring the quality of products being produced and sold to an agency is a critical part of every construction project. Every organization should develop a culture of quality to ensure that work will be completed satisfactorily. The process of developing a quality culture includes the assessment of existing operations, the development of measurement criteria, and the conduction of audits to ensure compliance. These elements, with a system of active, posi- tive communication between all groups involved, will enable the development of a high-quality product (Decker, 2012). Table 4. Life expectancy of asphalt pavements. Percentage of Responses Life Expectancy, years Overlay New Construction Perpetual Pavement 5–10 50.0 2.0 1.7 11–15 42.9 37.8 10.3 16–20 7.1 50.0 34.5 21–25 0 10.2 53.5

12 Specifying and Measuring asphalt pavement Density to ensure pavement performance Table 5. Types of specifications for pavement construction. Type of Specification Brief Description Materials/Methods (also known as recipe or prescriptive) Contractor required to use specified materials in definite proportions and specific types of equipment and methods. Each step is directed by agency. Experience has shown this tends to obligate the agency to accept work regardless of quality. End result Contractor required to take entire responsibility for product. Agency’s responsibility is to either accept, reject, or apply pay adjustment relative to degree of compliance. Affords contractor flexibility in use of new materials, techniques and procedures. Quality assurance Contractor required QC and agency acceptance throughout operations. Final acceptance based on statistical sampling for key quality characteristics. Uses random sampling and lot-by-lot testing to control operations. Statistically based Based on random sampling. Properties of desired product described by statistical parameters. Performance Describes how finished product should perform over time. Testing to determine performance not currently available for asphalt concrete mixtures. Performance-Based Describes desired levels of fundamental engineering properties that predict performance. Most testing is not amenable to timely acceptance testing so this process is not used. Performance-Related Describes desired levels of key materials and construction characteristics that correlate with fundamental engineering properties that predict performance. Timing of testing is amenable to construction process. Provide basis for rational acceptance/pay adjustment decisions. Warranty Guarantees the integrity of product and assigns responsibility for repair or replacement to contractor. Can guarantee either materials and workmanship or product performance. Materials and workmanship warranties Contractor held responsible for correcting defects in work within the contractor’s control during warranty period. Warranty period varies. Agency responsible for pavement structural design. Contractor assumes no responsibility for distresses resulting from design but may be responsible for materials selection and workmanship. Typically 1 to 3 years. Performance warranties Contractor held fully responsible for product performance during warranty period. Contractor guarantees pavement will perform at desired level. Contractor may be responsible for both structural and mix design. Short-term warranty is 5 to 10 years. Long-term warranty is 10 to 20 years. Source: Transportation Research Circular E-C173.

Survey to Determine the Density of asphalt pavements 13 The current definitions for quality assurance elements from TRB are as follows (Transportation Research Circular E-C173): Quality assurance (QA). Actions necessary to provide confidence that a product will per- form satisfactorily in service: i.e., making sure the quality of a product is what it should be. QA addresses the overall process of obtaining quality that includes elements of quality control, independent assurance, acceptance, dispute resolution, laboratory accreditation, and personnel certification. Quality control (QC). System used by the contractor to ensure the product will meet speci- fied level of quality. QC may or may not be specified by the agency. The contractor may elect to perform evaluations beyond QC requirements to ensure specified level of quality. QC may or may not be used as a basis for acceptance. Acceptance. Agency determines degree of compliance with contract requirements and cor- responding value for a given product. QC measurements may or may not be included in agency’s acceptance decision. Survey Question 1 Does your agency have a percent density specification for asphalt mixture placement? Is the density requirement waived for thin lift applications? Background In typical specifications, acceptance quality characteristics are based on materials properties such as gradation, asphalt binder content, and air voids as well as construction properties such as density and smoothness. The requirement for density control is a universal element of asphalt concrete pavement construction. Density specifications for asphalt concrete mixtures have been used for decades. Such is the level of importance of this basic engineering property of the in-place mixture. A key challenge is to get an adequate number of tests to represent the entire pavement. Clearly it is neither reasonable nor practical to test all the mix placed. Samples are usually taken in some random manner in an attempt to represent all the mix that was placed. The application of a thin-lift pavement presents a challenge for an in-place density specification. Thin lifts are being used widely as pavement preservation tools with many advantages identified by Newcomb, and Watson and Heitzman. The industry standard definition for thin lifts is less than 1.5 in. thick (Newcomb, 2009; Watson and Heitzman, 2014). Because of the thin application of asphalt concrete, it is extremely difficult to take a core sample or to use a non-destructive test method to determine the density of the thin application (Wilson et al., 2015). Cores are very fragile and difficult to handle for thin lifts. With a non-destructive testing (NDT) method of density test- ing, it is likely that the density value obtained will be significantly affected by the underlying layers. For applications greater than one inch, a thin-lift gauge may be useful if properly calibrated. For thin lifts, some agencies use a method specification as a QC tool (Newcomb, 2009; Watson and Heitzman, 2014). Since the thin-lift applications use small aggregate particle size and have high asphalt contents, the permeability is quite low as shown by Brown and Heitzman (2013). As has been noted in NCHRP Synthesis 464 (Watson and Heitzman, 2014) and NAPA publi- cations (Newcomb, 2009), density measurement of thin-lift applications is typically not accom- plished with cores and may not be tested for density at all. In this latter case, agencies often use a method specification to ensure proper placement of the thin-lift asphalt concrete.

14 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Analysis Not surprisingly, all 100 respondents indicated that their agency had a density specifica- tion requirement. Approximately 60 percent of the respondents waive the density requirement for thin-lift applications. The definition of thin lifts varied within the respondents. Responses included: • Less than three times the nominal maximum aggregate size (NMAS), • Less than 0.75 in., • Less than 1 in., • Less than 1.25 in., and • Less than 1.5 in. Survey Question 2 What percent density specification type do you believe provides optimum pavement performance? Background The types of specification have been discussed previously and each type has been used for vari- ous projects. Generally for asphalt concrete pavements, the options are not as broad as previously noted. For this question, the participants were given the options of method, end result, both, or neither for responses. The risk is greater for the agency with a method specification and the risk is greater for the contractor with an end result specification. To achieve a true end result specification, an agreed upon performance test is required. McCarthy et al. (2016) reported that significant research is underway to generate data and criteria needed to develop performance-based specifications, but that very few agencies are currently using performance tests for acceptance. Survey Analysis Approximately 80 percent of the respondents indicated that their density requirement was an end result specification. About 15 percent of the respondents employed both end result and method. Comments included: • End result specifications hold the contractor accountable and allow for innovation. • With more risk to the contractor, a better overall product is achieved. • Method specifications can work if the agency has adequate staff but end result specifications are not dependent on staff; however, there may be too many loopholes. • Specifications need to be achievable. • Tell the contractor what you want, pay accordingly, and you will get what you want. • PWL forces the contractor to have better control of the process. Survey Question 3 Is in-place density a pay factor? Is the pay factor an incentive or disincentive? Background Several terms relating to pay factors need to be defined. Transportation Research Circular E-C173 offers the following definitions: Pay Factor. A multiplication factor used to determine the contractor’s payment for a unit of work, based on the estimated quality of the work. Typically applies to only one quality characteristic.

Survey to Determine the Density of asphalt pavements 15 Composite Pay Factor. A factor obtained from two or more quality characteristics, to be multiplied by the bid price to determine the contractor’s final payment for a unit of work. May be a weighted average. Composite pay can also be calculated by adding the sum of individual pay adjustments to the bid price. Incentive/Disincentive. [More commonly known as a bonus/penalty by contractors.] A pay adjustment schedule whose function is to motivate the contractor to provide a high level of quality. Individual pay increases or decreases may not be of sufficient magnitude to motivate the contractor. As the steward of public funds, every agency must ensure that decisions made are in the best interest of the transportation system, the traveling public, and the taxpayer. Arizona DOT reported that the introduction of incentive and disincentive provisions to their specification led to an approximately 1 percent reduction in the average in-place voids on Arizona DOT end product asphalt projects. Arizona DOT believes that this improvement in the density of the in-place pave- ment will result in improved performance and offset any additional costs incurred (Nodes, 2006). The composite pay factor is used to determine payment using multiple quality characteristics. Hand and Epps note that there exists significant inconsistency among specifications across the United States in acceptance sampling. The “same” specification limits may lead to different pay- ments “due to differences in sampling and testing frequency, the number of tests per sample, sampling techniques, and lot and sublot definitions.” In-place density is normally a heavily weighted pay factor component, thereby recognizing the importance given to that factor. The lack of strong relationships between the quality measured and pavement performance is a key weakness in determining equitable pay factors (Hand and Epps, 2006). Survey Analysis Ninety-six percent of the respondents said that in-place density was a pay factor in their specifi- cation. When asked about the composite pay factor, a fairly wide range of responses was received, ranging from 25 percent to 50 percent as the weighting in the composite pay factor for percent density. Of the responses, 90 percent indicated a weighting in the composite pay factor between 35 percent and 50 percent with about 70 percent between 40 percent and 50 percent weighting. About 75 percent of the respondents indicated that their density specification contained both an incentive and a disincentive (bonus and penalty) clause, with 12 agencies using disincentives only. Comments from participants included: • Incentives only offset disincentives as the agency does not actually pay more than 100 percent. • Bonuses awarded too often. • We use only penalties as we expect the contractor to exceed our minimums in all cases. • Agencies should want the bonus to be awarded as this means that best practices are being followed and the agency gets a better product. Survey Question 4 Who is responsible for measurement of density for QC? For acceptance? Background QC and acceptance have been previously defined. The purpose of this question was to ascer- tain where the responsibility for both the operations resided—with the agency or the contractor. Hughes (2005) found that QC and acceptance were not being separated in the project anal- ysis. Decker (2008) presented data showing a statistically significant difference between QC

16 Specifying and Measuring asphalt pavement Density to ensure pavement performance and acceptance databases. The study was based on 5,598 data points (from 2004 and 2005) with 160 unique job mix formulae (JMF) from the Maryland State Highway Administration. The conclu- sion was that QC and acceptance databases should not be combined for acceptance purposes. It was further shown that air voids and Gmm test results do not correlate well between QC and acceptance. Survey Analysis The survey contained two questions relating to the measurement of density—one for QC operations and one for acceptance. For QC, 94 percent of the respondents indicated that the contractor is responsible for the density testing for QC, although five agencies indicated that the agency was responsible for QC testing. For acceptance, 87 percent of the respondents indicted that the agency was the responsible party, although 12 agencies indicated that contractor test results were used for acceptance. Often the caveat with contractor test results is that the results must be within some stated tolerance of agency results. In both QC and acceptance, some agencies used third-party consultants for testing. Survey Question 5 Does the responsibility for density measurement vary for different types of projects? If yes, what are the criteria that prompt a change? Background It is undeniable that the overall importance of a major heavily trafficked interstate highway is greater to a large number of motorists than a rural lightly traveled highway. But for the users of a rural highway, their road is just as important for commerce and safety. The purpose of this question was to understand how, if at all, agencies are differentiating between roadway types. Survey Analysis The response for this question was approximately two-thirds saying that the responsibility for density measurement does not change for different types of projects and one-third saying that it does. Comments discuss the use of a method specification/consultant/contractor data for low volume roads while others indicate that the acceptance specification is blind to project type. Responses for what criteria prompt a change included size of project (60%), type of mixture (60%), and lift thickness (43%). For small projects and/or maintenance projects, the level of testing is often less frequent. Survey Question 6 What is the reference density for calculating percent density? Background When an asphalt concrete mixture is compacted, the in-place density achieved is compared with a reference density to calculate a percent density (some refer to this value as the percent com- paction). This percent density is compared with specification requirements for determination of acceptance and payment for the work performed. Many approaches have been used successfully. It is critical that the user understand the differences in values that are obtained using different reference density values. The Asphalt Handbook (Asphalt Institute, 2007) provides a thorough dis- cussion of the differences obtained when using different reference density values. Huber (1999) also provides a review on the subtle differences that are noted for different methods: • The Gmm for the JMF as the reference density is known before construction begins and never changes. Plant mix tends to have a higher density than laboratory mix so a low target is established.

Survey to Determine the Density of asphalt pavements 17 • Percentage of the JMF Gmm can change from mix design to production due to, among other factors, variations in specific gravity or gradation of the aggregates and absorption of the aggregate that may result in higher air voids. • Lab-compacted plant mix can be variable as mix properties change, thereby changing the target density. • Specifying the percentage of Gmm means that the target in-place voids are always the same during the production of the mix. Variations that occur are accounted for in the production operation. • Percentage of test strip are dependent on the conditions at the specific location of the test strip. The Asphalt Handbook (2007) and Brown (1990) recommend the use of Gmm as the refer- ence density criterion. The advantage of using Gmm is that the value represents zero air voids (or maximum density) of the mixture. This zero void condition is obtained by applying a vacuum to an uncompacted sample of the mixture in accordance with established ASTM and AASHTO test methods. Because it is the maximum density obtainable, using this value as the reference density provides an engineering value for the percent density. Gmm does vary with asphalt content and gradation and should, therefore, be performed on a daily basis during production and used accordingly. This process ensures that an accurate value is obtained as the reference density. Use of Gmm from the mix design or JMF that may have been developed months or years prior to the mix production is a risky proposition. Brown (1990) recommends that the Gmm from the mix design should not be used as the reference density during production. The Gmm should be measured on the plant-produced mixture, should be verified either before or early in the produc- tion of mix for the project, and should be performed regularly during mix production (Cominsky et al., 1998). In the past, performing the testing to determine Gmm was cumbersome. But the equipment and procedures available today have made the testing easy, repeatable, and relatively quick (Buchanan, 2000). Survey Analysis Reference density is used as a comparison for the in-place density on the project. However, there is no universal agreement about what that reference density should be. Table 6 presents the responses from the survey. Approximately 92 percent of the respondents indicated using some variation of Gmm with approximately 74 percent using Gmm from plant-produced mixtures. There are at least three different approaches to the use of Gmm, all relating to where the sample is taken. Approximately 18 percent use Gmm from the JMF; approximately 29 percent use Gmm from an individual test from plant-produced mixtures; and approximately 45 percent use Gmm from a running average from plant-produced mixtures. Two percent of the respondents used each of the following methods: control strip density, lab density from the Marshall mix design, Table 6. Reference density used for the calculation of percent density. Reference Density Used Percentage of Responses Gmm of plant-produced mix based on average/running average 44.9 Gmm of plant-produced mix based on individual test 28.6 Gmm from JMF 18.4 Control strip density 2.0 Lab density from JMF with Marshall mix design 2.0 Lab density from JMF with gyratory mix design 2.0 Density of lab-compacted plant mix with gyratory mix design 2.0 Density of lab-compacted plant mix with Marshall mix design 0

18 Specifying and Measuring asphalt pavement Density to ensure pavement performance lab density from the gyratory mix design, and the density of the lab-compacted plant mix from the gyratory mix design. Survey Question 7 Where is the reference density sample obtained? Background In the overall scheme of asphalt operations, all acceptance decisions are based on getting a reli- able sample for testing. Since production decisions are based on the reference density, this test becomes particularly critical in the evaluation of field-placed mixtures. McMahon et al. (1990) discuss the importance of materials evaluation. Figure 2 illustrates the relationship between test- ing variability, materials variability, and sampling variability. The McMahon report shows that 43 percent of the variability is the result of variability of the testing, 23 percent of the variability is due to the sampling procedure, and only 34 percent of the variability is due to the actual materials variability experienced by the asphalt concrete mixture. ASTM D979 and AASHTO T168 present the same procedures for sampling of asphalt mixtures. Sampling locations specified are: • Sampling from a conveyor belt, • Sampling from truck transports, • Sampling from the roadway prior to compaction, • Sampling from a skip conveyor delivering mixture to bin storage, • Sampling from a funnel device feeding a conveyor for mixture delivery to storage, and • Sampling from bituminous cold mix stockpiles. Survey Analysis The responses were roughly evenly distributed between sampling from a truck at the plant and uncompacted from behind the paver, each about 40 percent as shown in Table 7. About Testing (43%) Sampling (23%) Materials (34%) Figure 2. Testing variability, sampling variability, and materials variability. Table 7. Reference density sample location. Sample Location Percentage of Responses At the plant discharge location 12.4 From a truck at the plant 42.7 From a truck at the paver 5.6 Uncompacted from behind the paver 39.3

Survey to Determine the Density of asphalt pavements 19 12 percent of respondents indicated the sample was taken at the plant discharge location while 6 percent reported sampling from a truck at the paver. Two respondents indicated taking a sample either in front of the screed or from the paver/material transfer vehicle hopper. While relatively easy to sample at either of these latter two locations, these procedures do not meet the require- ments of ASTM and AASHTO. The safety of the person doing the sampling is also compromised because of the proximity to operating equipment. One respondent commented that the best way to determine if the mix was being produced properly was to do plant sampling. The respondent further stated that sampling from other loca- tions introduces variability from the additional steps as well as the sampling procedure. The respondent continued that he understood that the material behind the paver is what is being purchased but thought that the inspector should be watching for truck loading techniques and mix management best practices. The issue raised by this respondent has been an ongoing concern in the industry for decades. The counter argument from the agency perspective is that the mate- rial on the roadway is actually what is being purchased so the sample should be taken behind the paver. This discussion will undoubtedly continue as long as asphalt pavements are produced and placed. Acceptance tolerances should be based on the uniformity of the sampling location. Survey Question 8 How is maximum theoretical specific gravity determined? Background AASHTO T209 and ASTM D2041 have standard procedures for determining Gmm. A weak- ness in the procedures is that multiple types of sample containers are allowed within the proce- dure. Therefore it is important for labs that are comparing results to verify that they are using the same equipment or to develop a correlation between results. A concern expressed by the Asphalt Institute (1983) is in regard to the amount of asphalt absorption that will ultimately occur in the roadway. In general, the longer an asphalt concrete mixture is held at high temperature, the more absorption will occur. As absorption occurs, the overall volume of the mixture decreases, resulting in a higher Gmm value. Clearly this is more of an issue with aggregates that have high absorption. This is why standard curing times are recom- mended in the Superpave system. Survey Analysis While there are some minor state variations for the standard procedures, 95 percent of the respondents indicated that AASHTO T209 and ASTM D2041 were used for the determination of Gmm. If working in multiple states, it is incumbent on the user to understand specifications for the jurisdiction in which the project is located. Only four respondents calculated the Gmm based on aggregate and asphalt specific gravities. Survey Question 9 What is the frequency for determination of the maximum theoretical specific gravity? Background Brown (1990) states that Gmm should be measured “routinely during construction.” Cominsky et al. (1998) indicate that the Gmm should be determined for every sample. Both agree that the results should be concurrent with the mix placement, not from a mix design value that

20 Specifying and Measuring asphalt pavement Density to ensure pavement performance may have been done months or even years in advance of the project. Root (1997) provides a framework for determining and using the running average value in order to better track variability in the testing process. Survey Analysis Eighty-seven percent of the respondents determined the Gmm daily or for every sample, with 67 percent performing the test on every sample. One respondent indicated that the test was done weekly, one that the test was done monthly, and nine that the test was only performed at mix design. Two respondents reported that the Gmm was never performed. For those running the tests on a daily frequency, the use of a running average was a common response. Frequency of the test varied from one each sublot to various increments of mixture production (e.g., 600T, 750T, 1000T). Some verified Gmm in early production and then use that value throughout the remainder of the job. Survey Question 10 Are percent density specification requirements the same for each of the following categories of pavement? Background The ability to achieve density in an asphalt concrete pavement is directly related to the under- lying structure. It is critical that the agency and contractor go into the project cognizant of condi- tions that might impact the ability to achieve proper percent density. Asphalt mixtures are used in a wide variety of applications. The intent of this question was to determine potential variability in testing requirements for different applications. Survey Analysis As shown in Table 8, 75 percent to 88 percent of the respondents replied that density speci- fication requirements did not vary for most categories. It is interesting to note that 37 percent of respondents indicated that different density specification limits or, in some cases, no density requirement may apply to shoulders. Many commented that the density requirement varies by mix type. It was noted that the lower lifts and shoulders may have same target values but differ- ent pay adjustments, thereby recognizing a difference in the importance of the location. It was commented that the French standard bases the density requirement on the type of product not the type of road. Comments were also made that density requirements may vary if the pavement structure was poor. Table 8. Density specification requirements for various pavement types. Pavement Categories Percentage of Yes, by Responses Interstate 86.5 Arterials 88.0 Urban 84.4 Rural 78.3 High traffic 88.3 Lower lifts 75.3 Mainline 87.2 Shoulders 63.0

Survey to Determine the Density of asphalt pavements 21 Survey Question 11 Do your percent density specification requirements vary for different material products? Background Participants were asked if percent density requirements varied by mix type, neat versus polymer-modified binder, large stone versus small stone mixes, and mix design procedure. Comparison of Marshall versus Superpave mix design would potentially yield varying density, depending on the compaction effort applied to the mix. Survey Analysis Ninety-four percent of the respondents indicated that the density requirements varied for different mix types. However, 23 respondents indicated no variation for different material prod- ucts. For some products, such as open graded friction course or thin-lift mixes, method type specifications were employed because of the difficulty of using conventional density measure- ment techniques. Variations were also noted for stone matrix asphalt (SMA) mixtures as well as for bridge deck applications. Survey Question 12 Does your organization use Engineering Limits/Specification Limits/Acceptance Limits? What are the upper and lower limits for each? Please reference to Gmm. Background There were several approaches to the setting of limits to determine acceptability of materials. The questionnaire contained a series of questions regarding the use of engineering/specification/ acceptance limits for the percent density specification. The responses have been combined. It should be noted that the response rate was low. This could indicate confusion in the distinctions made between the types of limits described in Transportation Research Circular E-C173. The limits described in Transportation Research Circular E-C173 have been paraphrased here: • Engineering limits: The absolute limiting value (either high or low) placed on a quality char- acteristic beyond which the result for an individual sample is considered to be unacceptable. Engineering limits are established to identify material that does not provide the required engineering properties. The limit may be the same as the specification limits. • Specification limits: Statistically based limiting values on specific properties that usually have either upper or lower limits or both. Individual test results may fall beyond limits and still be included in acceptance determination. • Acceptance limit: The limiting upper or lower value that will permit acceptance of a lot. Survey Responses Table 9 presents the survey results regarding engineering, specification, and acceptance limits. Based on Transportation Research Circular E-C173 definitions, 62 percent of the respondents indicated that they do not use engineering limits. Engineering limits are clearly not in widespread use, with only 36 of the 100 total respondents answering the question. For those that are using engineering limits, the most commonly used lower and upper limits are 92 and 97 percent density, respectively. More than 80 percent of the respondents used a lower limit between 90 and 92 percent density and roughly the same percentage used an upper limit between 96 and 98 percent density. Approximately 27 percent of the respondents indicated that there was no upper engineering limits in their specification.

22 Specifying and Measuring asphalt pavement Density to ensure pavement performance As shown in Table 9, 89 percent of respondents had a lower specification limit between 91 and 93 percent density, with 57 percent of the respondents indicating a lower specification limit of 92 percent. The overall average of responses was a lower limit of 91.3 percent density. This response represents 72 percent of all respondents to the survey. About 77 percent of the respon- dents indicated an upper specification limit between 97 and 98 percent density, with 58 percent of the respondents indicating an upper specification limit of 97 percent. It is interesting to note that 21 percent of the respondents indicated an upper limit of 100 percent density and approximately 35 percent indicated no upper limit for percent density in the upper specification limit. Wide variations existed in approaches and opinions espoused by the respondents. As shown in Table 9, 88 percent of the respondents have a lower acceptance limit between 90 and 92 percent density, with 46 percent indicating a lower acceptance limit of 92 percent. About 71 percent of the respondents indicated an upper acceptance limit between 97 and 98 percent density, with 46 percent indicating an upper acceptance limit of 97 percent. About 29 percent of the respondents indicated an upper limit of 100 percent density and approximately 35 percent indicated no upper limit for percent density. Density Measurement Techniques Quality cannot be tested or inspected into the mix; it must be “built in” (Cominsky et al., 1998). With percent density requirement as part of the specification, it is critical that a method for measuring the in-place density be easily performed, is fair to both the agency and contractor, and is based on sound scientific principles. The density requirement should be established based on optimal in-place density, not on what has been achievable in the past. This section of the survey investigates the approaches that have been taken to determine the density of the asphalt concrete in the pavement. Survey Question 1 How is in-place density measured for QC and for acceptance? Background Core testing has been the historical standard for determining the density of in-place asphalt concrete for many decades. It can easily be argued that the core provides a fundamental Percentage of Responses Engineering Limits Specification Limits Acceptance LimitsPercent Density Lower Limit Upper Limit Lower Limit Upper Limit Lower Limit Upper Limit 90 25.0 8.3 24.6 91 16.7 16.7 17.5 92 41.7 56.9 45.6 93 13.9 15.3 8.8 94 0 2.8 3.5 95 2.8 96 18.2 97 42.4 58.3 45.8 98 21.2 18.8 25.0 99 3.0 2.1 0 100 15.2 20.8 29.2 Note: Number of responses for engineering limits: lower limit = 36, upper limit = 33; number of responses for specification limits: lower limit = 72, upper limit = 48; and number of responses for acceptance limits: lower limit = 57, upper limit = 48. Table 9. Use of engineering, specification, and acceptance limits.

Survey to Determine the Density of asphalt pavements 23 engineering material property of the in-situ mixture. However, as with any procedure, there can be variations in the coring operation that can lead to erroneous density results, such as mis- handling of the sample, poor sample locations, or mislabeling of samples. One significant drawback to cutting a core is that the result is a hole in the pavement. Unfortunately, there is no standard protocol for patching the hole to ensure durability of the replacement mixture to ensure that the performance of the pavement at the core location is not compromised. Buchanan (2000) reports that a density gradient is likely within a core sample. He reports less variability for trimmed samples. Beainy et al. (2014) reports variations in density of up to 1.9 percent within 200 mm of laboratory produced slab samples. NDT for asphalt density has been in use since the 1970s with the nuclear density gauge (Troxler, 2016). Alexander and Doty (1984) reported that variability between cores and the nuclear density gauge was caused by significant irregularities in the mix surface. Prowell and Dudley (2002) reported that nuclear densities correlate well with core densities if measured prop- erly. He further commented that the premise of using an offset to correlate the nuclear density gauge to core densities appeared to be valid, concluding that a properly calibrated gauge can be used to measure in-place density. In 1998 the electromagnetic gauge was introduced as NDT for asphalt density with the principle advantage of not requiring a nuclear regulatory license (Apkarian, 2016). Smith and Diefenderfer (2008) reported good correlation between electromagnetic gauge and cores as long as moisture correction is done. Prowell and Dudley (2002) reported good repeatability with the electromagnetic device but noticed that the results were not well correlated to the core densities. Williams (2008) reported similar results as Prowell and further noted that small amounts of moisture significantly affected the electromagnetic gauge and that the orientation of the gauge on the pavement was significant. Both NDT devices have been used successfully across the country. Survey Analysis Clearly the nuclear density gauge and the thin-lift density gauges are the most commonly used devices for QC, as shown in Table 10, and the electromagnetic gauge is gaining in use. The cores referenced to in the “Quality Control” column under “Commonly Used” were generally used to calibrate the gauge based on comments received. For bridge deck testing, NDT methods are used almost exclusively. For acceptance, cores remain the dominant procedure for determining the percent density in the roadway. As shown in Table 10, 78 percent of the respondents indicated that cores were com- monly used for acceptance decisions. Of the 60 agencies, 20 indicated that the nuclear gauge was Percentage of Respondents Number of Agencies Using the Measurement Tool Measurement Tool Quality Control Acceptance Allowed Commonly Used Allowed Commonly Used Cores 75 46 87 78 51 Nuclear density gauge 80 71 33 32 20 Thin-lift nuclear gauge 58 42 22 20 13 Electromagnetic gauge 48 38 12 6 5 Table 10. Density measurement tools.

24 Specifying and Measuring asphalt pavement Density to ensure pavement performance commonly used, 13 indicated that the thin-lift gauge was commonly used, and 5 indicated that the electromagnetic gauge was commonly used for acceptance testing. Survey Question 2 How does your organization ensure the core is dry before testing? Background Specimen dry weight is a critical measure in the determination of the accurate density of core samples. If the specimen is not dry at the time of testing, incorrect values of specific gravity will result. Buchanan (2000) evaluated several methods for measuring specific gravity, namely water dis- placement, dimensional analysis, parafilm, and vacuum sealing. Buchanan found that if a sample absorbs 0.5 percent moisture during submersion, the resulting air void difference is approximately 1 percent. At a specification limit used by some agencies of 2 percent absorbed water, Buchanan observed a difference in air voids of as high as 6 percent between vacuum sealing and water dis- placement methods. Buchanan further reported that for almost all SMA and open graded friction course (OGFC) testing, the vacuum sealing method resulted in higher void content than the water displacement method, due to absorption in the coarse and/or open mixture. He concludes that the vacuum sealing method provides the most accurate measurement of bulk specific gravity (Gmb). Howard and Doyle (2014) concur that vacuum sealing is the most accurate and versatile procedure for determining the Gmb of the core sample. King et al. (2009) report on a project to evaluate the Thermolyne SSDetect (for fine aggregate Gmb and fine aggregate absorption determination), Instrotek CoreLok (to determine the specific gravity of core samples), and Instrotek CoreDry (to dry the core sample prior to testing). The report concludes that all three methods compared favorably with the existing ASTM and AASHTO proce- dures, with a major advantage of time saving for both SSDetect and CoreDry. These comments are based on 15 core samples taken in each of the nine Louisiana Department of Transportation and Development districts. The cores were from ongoing projects. The survey question gave participants the option of answering oven dry, vacuum drying, or other. As described in ASTM D7227 and AASHTO R79, drying specimens at room temperature is desired for some tests and it also provides the advantage of ensuring the integrity and preserv- ing the characteristics of the specimens. However, the process is time consuming. The vacuum drying procedure was developed in 2004 and is a relatively recent development (Regimand, 2016). Survey Analysis Almost 70 percent of the respondents indicated that the cores were oven dried to constant weight prior to testing, in spite of the lengthy time required. The vacuum drying procedure is used by 44 percent of the respondents. Even though vacuum drying is a relatively recent development, the use of vacuum drying has grown significantly. But 20 percent of the respondents used some other method of drying the core prior to testing—generally using some method of air drying either with or without a fan for air circulation. Survey Question 3 What is the limiting value for water absorption that is used to determine the need for a drying operation? Background As previously discussed, the core sample must be dry in order to get a correct specific gravity value. One of the limiting issues is how much absorption can be accommodated before a drying

Survey to Determine the Density of asphalt pavements 25 process needs to be implemented. Historically, this value has been a limiting value of 2 percent (ASTM D2726, AASHTO T166). Some older test methods prescribe coating the core sample with either a paraffin film or hot paraffin coating. The paraffin film procedure is very sensitive to the manner in which the film is applied and, therefore, operator sensitive. The hot paraffin coating is not used often primarily because of the safety issues of hot paraffin. Survey Analysis Of the 83 responses, 18 percent indicated a limiting absorption of 1 percent and 30 percent reported a limiting absorption of 2 percent although some of these respondents are considering an absorption limit of 1 percent (the current ASTM requirement). The remaining 52 percent of the respondents indicated “other,” which includes the following: • 15 percent always use the dry back procedure, • 21 percent indicated they don’t use any absorption evaluation, and • 1 respondent indicated using the paraffin coating procedure. Neither the paraffin film nor the hot paraffin coating was widely used by respondents to this survey. Survey Question 4 How are the test locations for cores identified? Are any exceptions allowed? Background Where the sample is taken is as important as how the sample is taken. It is obvious that sam- pling the entire pavement is not possible. The use of random sampling procedures has spread throughout the industry in the last decades. The concept of random sampling is that every loca- tion on the pavement has an equal opportunity to be tested. Clearly, in order to minimize risk to both the agency and the contractor, it is valuable to evaluate as much of the pavement as possible. Based on a review of the formula to calculate standard deviation, it is known that increasing the number of samples decreases standard deviation (the equation has n—the number of samples— in the denominator). In any sampling process, it is imperative to minimize potential sources for statistical bias in the evaluation. If, for example, every sample was taken at the edge of the pavement (for ease of sampling), there would be significant bias in the test results and the natural tendency for the contractor would be to compact more along that edge. Random sampling throughout the pavement eliminates that potential source of bias in the dataset. The downside from the agency’s and contractor’s perspective is that when relatively few tests are taken, the potential risk to both parties is high. Survey Analysis Ninety-four percent of the respondents use a random sampling procedure for the determina- tion of the core location. Most commonly, agency representatives perform the calculations for determining the core locations. These procedures are well defined by Burati and Hughes (1993) and Burati et al. (2003 and 2004). Table 11 lists some of the possible exceptions allowed for determining core locations when random sampling procedures are used. An overwhelming 91 percent of the responders thought it was important to relocate a core that was too close to the edge of the pavement (with the exception of joint density testing, which will be discussed later) and about 66 percent wanted to make adjustments to core locations that were too close to roadway hardware. As can be seen from the results in Table 11, there is not a

26 Specifying and Measuring asphalt pavement Density to ensure pavement performance strong consensus on potential other exceptions. Comments included a concern about core loca- tions over known underlying pavement defects such as milling scabs, PCC (Portland cement concrete) punch-out or severe cracking. A common comment was to keep the core 1 foot away from an unsupported edge. Survey Question 5 How are test locations for NDT identified? Background One of the major advantages for NDT is that many tests can be quickly performed to evaluate the pavement whereas with coring, only a few locations are actually tested due to the destructive nature of the test procedures and the time need to perform the procedures. As with any statistical evaluation, the variability of the process is reduced with additional samples. Survey Analysis Approximately 44 percent of the respondents indicated that NDT locations were selected by the random sampling procedure and approximately 31 percent indicating that the locations were chosen by the contractor (Table 12). In reality, often, the technician performing NDT walks down the pavement, testing at frequent locations. The location is often not statistically random but rather sporadic. Survey Question 6 What is the validation procedure for the use of NDT gauges? Background As previously discussed, there are two primary types of NDT procedures—ASTM D2950 and AASHTO T31 for nuclear density gauges and ASTM D7113 and AASHTO R79 for electro- magnetic gauges. A summary of the two NDT procedures is presented in Table 13, based on ASTM test methods. The ASTM and AASHTO test methods are clear about the need for the proper use of NDT gauges. Although some of the procedures are different for nuclear and electromagnetic gauges, one procedure is universal and necessary—correlation of the gauge with the cores. Table 11. Exceptions to determining core locations by randomly selected procedures. Exceptions to Determining Core Locations Percentage of Responses Proximity to edge of pavement 91.1 Proximity to road hardware (grates, drop inlets, valves, etc.) 66.7 Shoulders 38.9 Proximity to concrete (curb, slab, etc.) 36.7 Proximity to underground utilities 18.9 Table 12. Selection of NDT locations. NDT Locations Selected by Percentage of Responses Random sampling procedure 43.6 Inspector 5.3 Contractor 30.9

Survey to Determine the Density of asphalt pavements 27 Survey Analysis As shown in Table 14, the survey indicated that only 84 percent of respondents perform the critical step of correlation to a core density. About 54 percent and 40 percent perform calibration and standardization procedures, respectively. For nuclear gauges, the calibration and standardization procedures for the gauges are equally important to ensure validity of the test results. For the electromagnetic gauges, only the stan- dardization process is required. Survey Question 7 What is the frequency of coring for determination of in-place density? Background Typically cores are taken at some frequency based on a lot size determined by the agency. The challenge is to get an adequate number of tests to fairly evaluate the material placed. Survey Analysis More than 88 percent of the respondents indicated that cores were taken either by a tonnage produced or by a statistical sample lot, as shown in Table 15. There were 86 responses to the question and 69 comments from respondents. The tonnage and/or lot size frequency varied dramatically as noted in the following partial list of comments: • Every 0.2 miles, • 1 for every 100 tons on minor projects, • 4 for every 1,000 tons, Table 13. Comparison of nuclear density gauge and electro magnetic gauge. Nuclear Density Gauge Testing Electromagnetic Gauge Testing Operations Requires calibration, standardization, and correlation to cores Requires standardization and correlation to cores Nuclear license is required No license is required Interferences Spatial bias exists Spatial bias exists Chemical composition of mix may affect readings Chemical composition of mix may affect readings Correction factor may be required for overlay Electromagnetic fields may impact reading Surface roughness may impact reading Surface roughness may impact reading Oversize aggregate may impact reading Oversize aggregate may impact reading Sample volume is small Sample volume is small Table 14. Validation of NDT gauges. Validation Procedure Percentage of Responses Calibration of the gauge 53.5 Standardization of the gauge 40.0 Correlation of the gauge with the cores 83.5

28 Specifying and Measuring asphalt pavement Density to ensure pavement performance • Minimum 3 cores per day, • 4, 5, or 6 sublots per lot, • Daily/one shift lot, • 500, 600, or 750 ton sublots, and • No coring except to correlate gauge. Survey Question 8 What is the frequency for NDT for determination of in-place density? Does the frequency vary by job size? Background It is most common for the person doing NDT for density to assist with establishing the roller pattern at the beginning of the job and then to take routine tests during placement of the asphalt concrete. The technician performing the NDT may have duties on the job site other than testing (some- times even flagging). In addition, the technician may be reporting to an operations person rather than to the senior management of the company. Both of these situations compromise the effi- ciency and impartiality of the testing. Survey Analysis There is a relatively uniform distribution of responses between tonnage, lot, and as instructed by contractor (Table 16). Since the NDT operation is generally a QC function, it is rational that the contractor would be directing that work. With regard to frequency, 71 percent of the respon- dents indicated that the frequency of NDT testing does not vary by job site. Most commonly, the testing technician will perform tests almost continuously during the paving operation. Survey Question 9 What is the typical sample size for in-situ sampling? Frequency of Coring Percentage of Responses Daily 8.1 By tonnage 38.4 By lot 50.0 As instructed by inspector 3.5 Table 15. Frequency of coring. Table 16. Frequency of NDT testing for density. NDT Frequency Percentage of Responses By station number 5.3 By tonnage 25.3 By lot 30.7 As instructed by contractor 34.7 As instructed by inspector 4.0 Frequency does not vary by job size 71.0

Survey to Determine the Density of asphalt pavements 29 Background A larger sample inherently produces a more representative test result. The volume of mix for a 6 in. core is approximately 2.25 times as large as for a 4 in. core. This, along with the ability to use 6 in. cores for other testing, is why many agencies now use 6 in. cores. Survey Analysis Of the 96 responses to this question, 52 percent replied that they used 4 in. cores whereas 58 percent replied that they used 6 in. cores. Only 2 percent used a slab sample. Some agen- cies allow both 4 in. and 6 in. core samples. Comments indicated that the 6 in. sample size was increasingly common because the sample could be used for further testing. The larger sample also provides a more representative sample of the mixture. Survey Question 10 Does the density measurement test procedure vary by job type and size? Does the density measurement specification vary by job type and size? Background Both ASTM and AASHTO have standard test procedures for density measurement. Although, it is common for agencies to modify the test method from the standard test protocols, the standard test procedures are recommended. Survey Analysis Eighty-five percent of the respondents indicated that changes were not made to the testing procedure for different job types or job sizes. Sixty-eight percent of the respondents indicated that the density specification did not vary by job type or job size. Comments on both questions included: • Variations may exist for minor quantity projects. • Cores are required for PWL projects but NDT is used for other projects. • Evaluation of the data may vary but the procedures remain the same. • Method specifications are used for thin lifts. Survey Question 11 How is the measured percent density compared to the acceptance criteria? Background Statistical concepts were introduced into the pavement construction industry many years ago. The approach was developed to measure the quality of bullets being produced during World War II (McMahon et al., 1990). A detailed description of the statistical approach to materials evaluation can be found in reports by Burati and Hughes (1993), Burati et al. (2003 and 2004), and Root (1997). The first step in the statistical approach is to establish a “lot” as a unit of measure of the material to be evaluated. The lot is designed to be a discrete portion of the material that can be easily managed. For example, a lot for an asphalt project may be a day’s production or perhaps some increment of tonnage of mix. That lot of material is most commonly broken down into stratified sublots to ensure even distribution of the testing throughout the lot as described by Burati and Hughes (1993).

30 Specifying and Measuring asphalt pavement Density to ensure pavement performance Quality measures are any one of several mathematical tools that are used to quantify the level of quality of an individual characteristic. Typical quality measures used in the asphalt industry are average, standard deviation and PWL or percent defective (PD). PWL is used more commonly than PD, most likely because of the positive connotation rather than a negative. PWL is the percentage of a lot falling above the lower specification limit, beneath the upper specification limit, or between the lower and upper specification limit, depending on how the specification is written. PD is the percentage of the lot falling outside specification limits (Trans- portation Research Circular E-C173). The PWL procedure can be used for any test parameter desired. According to McMahon et al. (1990) the advantages of using statistical concepts are that they: • Provide a statement of concise quality requirements, • Develop valid tolerances, • Delineate the responsibility for QC and acceptance, • Develop valid sampling plans, • Establish precise decision criteria, and • Develop valid proportional payment schedules. The major concern with using PWL as a quality measure is understanding the ramifications of how the specification limits are set. Simply tightening the specification bands may not necessarily improve the quality of the mix produced. Hand and Epps (2006) raise a concern about variabil- ity, stating that two specifications with the same specification limit using the same PWL model and the same pay factor tables may provide different payments to the contractor. This situation may occur due to differences in the sampling and testing frequency, the number of tests per lot, sampling technique, and lot or sublot definitions. Hand and Epps (2006) further acknowledge that risk is inherent because of the small number of samples that are used to estimate the quality of a large amount of material. They offer sugges- tions to minimize risk to both agency and contractor: • Require AASHTO accreditation for all laboratories, • Require all technicians to be certified, • Select sample locations and sampling/splitting methods that result in the lowest amount of variability, • Select test methods that result in lowest variability, • Eliminate options within test methods to reduce variability, and • Use only QC rather than pooled QC and QA data. McMahon et al. (1990) state that test measurements that are portrayed graphically are more easily understood. Further control charts show cumulative trends in properties and are effective tools to visually forewarn both agency and contractor that undesirable trends may be develop- ing, thereby helping to make decisions about when action may be necessary to adjust the process. Running average is a graphical method of data analysis that is often used. The concept of run- ning average is that an average value of the parameter is determined as tests are performed. The running average approach will “even out” variations in the test procedure and will also account for modifications made in the production process. Running average can quickly show the drift of the process average if it occurs. When several tests are done, a running average of the value is determined, usually with either four or five tests. For example, if a running average of five is used, when the sixth test result is obtained, the first result is dropped from the average and the sixth result is included. This process continues for the duration of the project. Such a procedure can be used for any test parameter desired (Root, 1997).

Survey to Determine the Density of asphalt pavements 31 Survey Analysis As shown in Table 17, 54 percent of respondents indicated that PWL (or PD) quality measures were employed for acceptance. Forty-six percent indicated that an arithmetic average, often a run- ning average, was used. Sixty percent of the DOT respondents indicated that their agency used PWL specifications. Others commented that they were moving to PWL specifications in the near future. Construction Parameters Affecting Density This section addresses the issues of construction procedures necessary to achieve density in the finished pavement. The process of achieving density in an asphalt concrete pavement is a complicated series of many activities that are significantly affected by the elements of mix design, paver, and roller operations. From the time the asphalt concrete mixture leaves the plant, is loaded into the paver, is placed on the prepared surface to be paved, and compacted by the rollers, there are numerous opportuni- ties for things to go wrong. With proper planning, an attitude to maintain quality, and personnel trained to do their job correctly, a long-lasting pavement can be obtained. With the implementation of Superpave, there was a learning curve for the new, sometimes coarser, mixes. However, over time the mixes have reduced in NMAS and laboratory compactive efforts have decreased in many jurisdictions. Roadway compaction of Superpave mixes has not proven to be significantly different from conventional mixes although, in the first few years, there was a learning curve for both agencies and contractors. Brown (1998) discusses issues affecting compaction, including total fluids content (asphalt binder and moisture), coarse graded mixes, and challenges with NDT of coarse graded mixes. Survey Question 1 What is the minimum lift thickness to nominal maximum aggregate size ratio requirement? Background Aggregate constitutes approximately 95 percent of the asphalt concrete mix by weight. As such, the quality properties of the aggregate must be evaluated during the mix design process. From the construction perspective, NMAS is an important parameter that must be considered in the mix design and construction phases of the project. The concern is that if the aggregate size is too large for the lift thickness, achieving density in the pavement might be extremely difficult if not impos- sible due to the inability of the aggregate particles to be reoriented into a dense configuration. A fundamental design concept for asphalt concrete mixtures is that every aggregate particle should be coated with asphalt binder. If the aggregate is too large for the lift thickness and is fractured during the compaction process, uncoated aggregate faces will be present, which violates the coat- ing requirement and results in potential performance issues. The lift thickness to NMAS ratio is a parameter that is evaluated to avoid the issue of the aggregate being too large for the lift thickness. With Marshall mixes, the ratio that was considered acceptable was 2:1. However, in the Marshall mix design procedure, the maximum aggregate size was generally Table 17. Comparison of percent density to acceptance criteria. Approach Percentage of Responses PWL or PD 54.0 Average 46.0 Lowest single measurement 11.0

32 Specifying and Measuring asphalt pavement Density to ensure pavement performance the first sieve that had 100 percent of the material passing. The definition of NMAS was changed with the implementation of Superpave and now includes aggregate particles at the NMAS. So a mixture that was previously known as ¾ in. (19 mm) would likely now be a ½ in. (12.5 mm) prod- uct. As a result, the minimum thickness to NMAS ratio has been increased to 3:1 in the Superpave requirements (Brown, 1998). Some agencies use a higher ratio for coarse graded mixtures. Survey Analysis As shown in Table 18, almost 70 percent of the respondents indicated the lift thickness to NMAS ratio requirement for their agency was a minimum of 3:1, as recommended by the Super- pave requirements. Comments received included: • Requirement is primarily for dense graded mixtures. • Differences in design and construction may lead to less than 3:1. For example, if a ¾ in. mix is designed at 4 in. thickness but the contract/phasing/maintenance of traffic require 2-in. lifts, the 3:1 requirement will be violated. • Agencies using Marshall mix design use 2:1. Table 18. Minimum lift thickness to NMAS ratio. Ratio Responses 2:1 12.8 3:1 69.2 4:1 18.1 Percentage of Table 19. Use of vibrating screed. Response Percentage of Responses Vibrating Screed Required Vibrating Screed Used Yes 39.4 51.0 No 50.5 30.6 Survey Question 2 Are contractors required to use vibrating screeds on the paver? Are the vibrating screeds routinely turned on during paving? Background Throughout North America, the most common type of paver is one that has a vibrating screed. In Europe, the tamping bar screed is more common than it is in North America. The key purpose of the vibrating screed is to assist in the extrusion of the asphalt mixture under the screed. In addition, compactive effort is applied to the mixture, thereby modestly improving the density of the mix being placed (Guide to Asphalt Paving, 2016). There are no data from a scientifically designed experiment, but the rule-of-thumb generally accepted is that the vibrating screed can increase the density of the mixture up to about 2 percent. This potential increase is dependent on the mix type, thickness, and environmental conditions. Although the increase may seem to be nominal, an additional 2 percent behind the paver leaves less work for the roller operators to achieve specified density requirements. Survey Analysis As shown in Table 19, about 51 percent of the respondents indicated that the vibrating screed was not required. One of the challenges in using the vibrating screed was that the screed operator

Survey to Determine the Density of asphalt pavements 33 is essentially standing on a vibrating platform. The operator often finds this situation to be uncomfortable and may even turn the vibrator off. But 51 percent of the respondents also indicated that the vibrating screed was routinely used. Twenty of the 33 contractor representatives (61 percent) responding to the survey indicated that the vibrating screed was commonly used. There were many comments that mentioned that the vibrating screed was not always used. Survey Question 3 How is the rolling pattern established? Is there a formal procedure for establishing the rolling pattern? Background The compaction process is a procedure to “grow” density in the asphalt concrete mixture. With each successive pass of the roller, the density increases up to some refusal point at which the density decreases—a process similar to a Proctor test on soil. Linden and Van der Heide (1987), Hedderich (1996), Starry (2006), Kearney (2006), and the Guide to Asphalt Paving (2016) provide excellent discussions about the process of compaction and the importance of achieving a proper roller pattern. Establishing the roller pattern to achieve proper density in the pavement is a fun- damental matter of physics and the environment. The roller operator must know the number of passes that have been applied in order to achieve the necessary rolling effort. As passes are made, NDT is performed to inform the roller operator of the “growth” in density. The operation requires attention to detail on the part of the roller operator and cooperation with the QC technician operating the NDT gauge. For each mix and set of project circumstances, there is a parameter called time available for compaction. Since an asphalt mixture stiffens as it cools, the roller operator must also be cogni- zant of the temperature of the mix as the rolling is performed. As previously discussed, the tem- perature of the mix and environmental conditions are critical to evaluate. Both have a significant impact on the time available for compaction. Rolling when the mix is too hot may cause the mix to move both in the longitudinal and transverse directions. Rolling when the mix is too cold may fracture the pavement. Starry (2006) recommends: • Operating the vibratory roller at the highest frequency (vibrations per minute) available for the roller. • Using the lowest amplitude consistent with achieving target density. Amplitude is dependent on the conditions of the underlying pavement structure. • Vibrate unless there is a compelling reason not to vibrate. Starry comments that the contractor has an interest in achieving density because of the specifi- cation requirements for both density and smoothness, which need to be accomplished while maintaining a high level of production. A free software named MultiCool is available to determine the time available for compaction for the specific conditions being encountered. The software can be obtained at www.eng.auburn. edu. To use the software, the user needs to enter data on the construction project and the environ- mental conditions. The software then generates a cooling curve. The cooling curve is quite useful to contractors needing to make a decision about the paving conditions, paver speed, number and type of rollers needed, and trucking requirements. The software can be used as an application on a smart phone or tablet.

34 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Analysis According to the survey, 84 percent of the respondents indicated that the roller pattern is estab- lished by using one of the NDT technologies. Many respondents used the “comment” box to say that the roller pattern establishment was the choice of the contractor, which would likely increase that percentage. Only about 8 percent of the responders indicated that either the roller pattern was a specification requirement or that cores were used to establish the roller pattern. Some respondents indicated the use of a test strip for establishing the roller pattern. On some projects, a formal test strip is not performed. However, for every job the initial portion of the paving functions as a test strip whether it is called that or not. The bottom line is that the contractor must “tweak” the rolling operations at the beginning of the paving operation in order to achieve the desired results. According to the survey, 64 percent of the respondents indicated that there was no formal procedure for establishing the roller pattern while the remaining 36 percent indicated that there was a formal procedure. For agency-only respondents, 22 percent have a formal procedure for establishing the rolling pattern. The most common approach was the requirement for a test strip to establish the roller pattern. Survey Question 4 Who establishes the roller pattern? Who ensures the roller pattern is followed? Background As previously discussed, development of the rolling pattern is a critical element in the asphalt pavement construction process. All parties must know who is responsible for establishing and maintaining the roller pattern. Hedderich (1996) describes a good roller pattern as a process that provides “the uniformity and efficiency needed to meet density and smoothness requirements and still keep up with the paver’s production.” The contractor must establish (1) how many passes are required to cover the width of the mat being placed, (2) how many repeat passes are needed to achieve the specified density, and (3) how close to the paver the roller needs to operate to achieve the density in the time available for compaction. One pass of the roller is defined as one movement of the roller from Point A to Point B. A round trip from Point A to Point B is two passes. Survey Analysis As shown in Table 20, 94 percent of the respondents indicated that the contractor or consultant established the roller pattern. Only 6 percent indicated that the agency performed this task. According to the survey, 79 percent of the respondents indicated that the contractor ensured that the roller pattern was followed. However, 20 agencies (40 percent) indicated that an agency Table 20. Responsibility for establishing and following the roller pattern. Percentage of Responses Personnel Establishes Rolling Pattern Ensures Pattern is Followed Agency Inspector 6.3 21.0 Contractor Personnel 91.7 79.0 Consultant 2.1 0

Survey to Determine the Density of asphalt pavements 35 inspector was present to ensure that the roller pattern was followed. Typically, the inspector did not direct the process but verified and reported the process. Survey Question 5 Is the roller pattern adjusted for the following? (Options were mix type, air temperature, wind speed, base temperature, inspector on site) Background Mix type, air temperature, wind speed, and base temperature are all factors that can signifi- cantly impact the compaction process. The establishment of a rolling pattern is not a one-time event as conditions can change even during the progress of a day. The roller operator needs to be aware of these elements and the potential impact of these elements on the ability to achieve den- sity. The software, MultiCool, which has been previously discussed, can be very useful to under- stand the impact of site conditions on the compaction process. On some projects, an inspector on the site directs the rolling pattern. Survey Analysis As seen in Table 21, that the principle factor that calls for roller pattern adjustment is mix type, which includes the mixing and compaction temperature requirements from the mix design. The most common comment is that the decision to adjust roller pattern is the responsibility of the contractor. Survey Question 6 Does your organization have training requirements for roller operators? Background Operating the mechanical systems on the roller and understanding the concepts of rolling are not the same. Efforts need to be taken to ensure that the roller operator understands what is being asked relative to achieving asphalt density and why. An operator with experience in compacting soils may not fully understand the temperature sensitivity of asphalt concrete. It is incumbent on the contractor to ensure that all personnel are properly trained. It is further incum- bent for the operator to seek guidance to perform the rolling task properly. Survey Analysis As high as 83 percent of the respondents indicated that they had no training requirements for roller operators. The training that existed was mostly classroom training sessions with no exami- nation to gauge what participants learned. None of the respondents indicated a roller opera- tor certification program. Many indicated that training was available but not required. Of the 97 respondents, 17 respondents indicated that training requirements existed. Some of the train- ing requirements were specific to a project, perhaps demonstrating a new product or new piece of equipment. In order for training to be effective, there must be follow-up enforcement of the best practices to improve quality. Table 21. Variables that impact roller patterns. Variables Percentage of Responses Mix type 89.2 Air temperature 67.5 Wind speed 60.2 Base temperature 60.2 Inspector on site 42.2

36 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Question 7 How many rollers are in your typical operation? What type of rollers are typically used? Background The process of rolling is usually divided into three approximate temperature zones— breakdown, intermediate, and finish. Different rollers may be used for each zone. The break- down zone is located closest to the paver and is the first roller to impact the pavement. This zone is generally considered the primary compaction zone because the temperature of the mixture is at its highest. The intermediate zone is located behind the breakdown zone and represents a region where the temperature drops rapidly. Density can still be achieved in this zone until the mixture reaches a lower temperature (dependent on the mix type and environmental conditions). The finish zone is where minor surface imperfections are smoothed out. This zone is usually not used to achieve the specified density. The number and type of rollers is an integral element of the process for achieving density. A very brief discussion of the roller types follows: • Double drum vibratory (DDV) rollers consist of two full width steel drums that provide vibra- tory compaction to the pavement. The weight of the roller, the frequency of impact on the pavement, and the amplitude (vertical travel distance) of the drums are key elements of the process. The amplitude is achieved by using an eccentric weight inside the drum. As the weight spins, the drum lifts and drops creating the amplitude of the roller. The frequency is measured by how many times per linear foot the drum impacts the pavement. • Pneumatic tire rollers (PTRs) consist of two sets of pneumatic tires (front and back). The tires are offset from front to back to ensure complete coverage of the mat. The tire pressure and the weight of the roller have an impact on its rolling efficiency. Density is achieved through the weight of the machine, the tire pressure, and the “kneading” effect on the asphalt concrete mixture. A subset of the PTR is the Vibratory PTR that is used in some areas. • Double drum static (DDS) rollers consist of two full width steel drums similar to DDV rollers but without the vibratory capability. DDV rollers can be used as a static roller simply by not using the vibratory function. • Three Wheel Rollers (TWRs) consist of one full width drum and two narrow width drums on the front and back respectively. These rollers are currently not used widely, although some agencies use them for achieving joint density. • Oscillatory (OSC) rollers have the same configuration of DDV rollers and can operate in the same manner. The distinction is that oscillatory rollers use dual, opposed weights rotating in the same direction around the roller drum axis to produce a rocking motion. OSC rollers can vary the direction of the compactive force applied to the pavement and are used widely on bridge decks and in areas with structures close to the road (Kearney, 2006). Survey Analysis About 76 percent of the respondents indicated that the typical rolling train consisted of three rollers and 15 percent indicated that it consisted of two rollers. A few respondents indicated four or more rollers, depending on the conditions at the time of paving. Table 22 presents the responses on the types of rollers used. An overwhelming response (almost 94 percent) was that DDV rollers were used for breakdown rolling. For intermediate rolling, there was a relatively even distribution of use between the DDV roller (65 percent) and the PTR (57 percent). For finish rolling, the most common type of roller used was the DDS roller. Almost a quarter of the respondents indicated that a DDV roller was used for finish rolling. Although the respondents mentioned the DDV roller, it was most likely used without vibration so effectively

Survey to Determine the Density of asphalt pavements 37 performing the same function as the DDS roller. A few respondents indicated the use of a combi- nation roller with front drum and rear pneumatic either as an intermediate or finish roller. Some agencies required the use of a pneumatic tire roller. Survey Question 8 Who makes the decision on roller type? What drives the roller type decision-making process? Background Unfortunately, there is no table or chart to indicate what type of rollers should be used in spe- cific pavement construction conditions. Roller manufacturers typically have a table that provides comparisons between the different frequencies for the roller and the impact spacing as a function of speed of the roller. The decision is typically made based on experience in a specific area with specific materials and with the specific equipment available to the contractor. Table 23 provides insight into the roller speed issue. The horizontal axis shows the speed of the roller while the vertical axis presents the frequency for the roller. The dark gray cells show com- binations of roller speed and frequency where the required 10 impacts per foot can be achieved. The light gray cells demonstrate combinations of roller speed and frequency where the required 10 impacts per foot cannot be achieved. The table clearly shows that the requisite 10 impacts per foot can be achieved with any of the frequencies shown if rolling is done at an appropriate speed. The difference between 2,000 vibrations per minute and 3,800 vibrations per minute is that the roller can move at a faster pace and still achieve the required density. Rolling Location Percentage of Responses DDV PTR DDS TWR OSC Breakdown 93.5 12.0 14.1 6.5 23.9 Intermediate 65.2 57.3 25.8 4.5 21.4 Finish 23.3 6.7 84.4 6.7 16.7 Table 22. Type of rollers used. Frequency 2 MPH 3 MPH 4 MPH 5 MPH 2,000 vpm 11.36 7.58 5.68 4.55 2,200 vpm 12.50 8.33 6.25 5.00 2,400 vpm 13.64 9.09 6.82 5.45 2,600 vpm 14.77 9.84 7.39 5.91 2,800 vpm 15.91 10.61 7.95 6.36 3,000 vpm 17.05 11.36 8.52 6.82 3,200 vpm 18.18 12.12 9.09 7.27 3,400 vpm 19.32 12.88 9.66 7.72 3,600 vpm 20.45 13.64 10.22 8.18 3,800 vpm 21.59 14.39 10.80 8.63 Note: vpm = vibrations per minute. Table 23. Select DDV roller speed and frequency to achieve impacts/foot.

38 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Analysis As high as 92 percent of the respondents indicated that the contractor made the decision about the roller type, although many agencies have a list of approved equipment that is accept- able for use, generally based on past experience. One exception noted by several respondents is that the type of roller used on a bridge deck was often specified by the agency (not a DDV roller). As shown in Table 24, the decision on the type of roller to be used was predominately made based on the perceived ability to achieve the proper percent density in the pavement with the type of mix (83 percent and 63 percent, respectively). The combined number of responses for smoothness requirements, subgrade conditions, and weather conditions was approximately equal to the number of responses for the ability to achieve density. Survey Question 9 Is the weight of the roller verified? Is the frequency of the DDV roller verified? Background The weight of the roller is an important factor in the equation that results in achieving the density in the pavement. Comparing the density obtainable with a light versus heavy roller would probably reveal very little difference in the final density that could be achieved. However, using a heavier roller will achieve the same density with fewer passes and before the mix cools exces- sively. This is why the frequency settings on the DDV roller have increased in the last few years— to improve compaction efficiency and, therefore, achieve more dense asphalt for each day of the placement. So the decision about the weight of the roller is a tradeoff between the weight of the roller and number of passes required to achieve density. And, of course, the availability of the equipment is an issue that must be considered. The frequency of the vibratory roller is the number of times the roller impacts the pavement. The general recommendation is that the frequency of the roller should be set to achieve 10 to 12 impacts per foot as the roller moves (Starry, 2006). Verifying the frequency of the roller is a relatively straightforward procedure using a device call a Reed tachometer. The device is based on the concept of a tuning fork and a readout of the frequency is produced. According to the manufacturers of the equipment, frequency adjust- ment is relatively simple but the amplitude is set from the factory and cannot be adjusted. As the equipment ages, changes may be noted in the effective amplitude and the frequency achieved. The rolling pattern may need to be adjusted accordingly (Mansell, 2016). Table 24. Drivers for the roller type decision-making process. Ranking Decision Drivers Percentage of Responses 1 Ability to achieve density 83.4 2 Type of mix 62.5 3 Availability of equipment 45.8 4 Size of job 41.7 5 Smoothness requirements 34.4 6 Subgrade conditions 28.1 7 Weather conditions 22.9 8 Unsure 14.6 9 Availability of personnel 12.5 10 Traffic level 6.3

Survey to Determine the Density of asphalt pavements 39 Survey Analysis Of the 97 respondents, 76 percent do not verify the weight of the roller. A total of 11 agencies (18 percent) verified the weight of the roller, varying from annually to per job. Generally, verifica- tion is based on information provided by the manufacturer. This is a reasonable position to take as the objective is to get density—how the contractor achieves density is not of critical importance to the specification writer. Seventy-nine percent of the respondents do not verify the frequency of the DDV roller. Nine agencies and eight contractors indicated that a verification of the frequency of the DDV is per- formed, ranging from daily to occasionally. It is possible that a mechanic performs the verification of the frequency and the respondent of the survey was not aware that it had occurred. In any case, if someone is troubleshooting variations in density, looking at the frequency of the equipment can be a useful point to start a troubleshooting process. Survey Question 10 Are additives used for enhancing compaction efficiency? Background Terrell and Epps (1989) provide a broad overview of the purpose and use of additives and/or modifiers in asphalt concrete. Over the years, many different materials have been proposed for use. The following categories were identified by the authors: • Filler, • Extender, • Polymers, • Fiber, • Oxidant, • Antioxidant, • Hydrocarbon, • Anti-strip, and • Chemical. Terrell and Epps concluded that the user needs to be knowledgeable about the product being used and the product must improve performance of the mixture. The primary compaction aid currently used in the United States is warm mix asphalt (WMA). The concept of WMA was introduced into North America as a result of a European scanning tour that occurred in 2002 (Prowell, 2007). The tour was sponsored by NAPA and was focused on understanding the technology and potential application of WMA to the U.S. market. There are three distinct approaches to achieving the WMA concept: foaming, chemical addi- tives, and waxes. Details on each of the approaches can be found in Prowell (2007), Newcomb et al. (2015), and West et al. (2014). The potential benefits of WMA technology noted by Prowell (2007) and Newcomb et al. (2015) include: • Compaction aid for stiff mixes, • Ability to pave in cooler weather, • Increase haul distances, • Potential for higher percentages of RAP (recycled asphalt pavement), • Reduced energy consumption, • Reduced plant emissions and odors, and • Better working conditions.

40 Specifying and Measuring asphalt pavement Density to ensure pavement performance Since its relatively recent beginning in 2002, WMA has grown to occupy approximately one- third of the total U.S. market. Approximately 75 percent of all the WMA in the United States is produced by the foaming approach (Hansen, 2016). Over the last decade, many reports have been written on WMA. Many of these reports can be found at the TRB, FHWA, and NAPA websites and will not be discussed here (www.trb.org, fhwa.dot.gov, and www.asphaltpavement.org). Survey Analysis In the survey, 82 percent of the respondents indicated that additives (either water, chemical, or wax) were used to enhance compaction efficiency. Many respondents indicated that warm mix technologies were being used at hot mix temperatures to improve compaction efficiency. Survey Question 11 What is the typical tonnage placed per day per job for mainline operations? Background As with any business, contractors try to accomplish as much work in a day as possible. Con- sistency in the operation will result in consistency in the density achieved. It is always better to focus on the achievement of the proper quality of the construction, including density, than to focus on getting a specific number of tons in a day. Survey Analysis As shown in Table 25, about 54 percent of the responses indicated the tonnage to be 2,000 tons or less per day. In the “other” category, 11 respondents indicated that they placed less than 2,000 tons per day. Combined, approximately 66 percent of the respondents indicated 2,000 tons or less per day. Adding in the next level, approximately 73 percent place 3,000 tons or less per day. Comments indicated that significant variability existed depending on the job, mix type, contrac- tor, weather, and other factors. Lane closure time, the number of lanes to be paved, maintenance of traffic, and project staging also significantly affect production capabilities. Survey Question 12 What is the typical paver speed for operations in your jurisdiction? Would it be advanta- geous to use a slower speed to optimize density? Should specifications include a maximum paver speed? Background In any paving operation, it is critical to balance the production rates of the plant, trucking, paving, and rolling in order to achieve consistency not only in the density achieved but also in Typical Tonnage Percentage of Responses 1,000 9.4 2,000 44.8 3,000 18.8 4,000 6.3 5,000 2.1 Other 18.8 Table 25. Typical tonnage placed per day per job for mainline paving.

Survey to Determine the Density of asphalt pavements 41 all mix properties. The major difficulty in producing asphalt concrete is that it must be accom- plished under extremely variable conditions (Warren, 1996). The paver speed is a critical part of that equation. It is not rational for the paver to be operated at maximum speed with the rollers unable to keep up with the paver. This situation creates significant issues for the roller to achieve density while the mix is at an appropriate temperature. Survey Analysis Table 26 presents the typical paver speed responses from the survey. It is interesting to note that the most common response was “unknown.” Forty of the 42 “unknown” responses came from organizations that were not contractors and so this response is understandable—they just don’t know. Of the contractor responses, 88 percent indicated paver speed between 20 and 40 feet per minute. Responses from Europe indicated a paver speed of 10 to 15 feet per minute using tamper bar screeds on the paver—much slower than in the United States. A simple calculation reveals that to achieve the typical 2,000 tons per day, the paver needs to operate at less than 30 feet per minute for an 8-hour day for a 2-in. mat 12 feet wide. Comments indicated that the paver speed was typically slower for PWL specification projects. This seems to imply that where quality carries with it a significant risk, the paver speed is typi- cally slower. In the survey, 72 percent of the respondents indicated that a slower paver speed would be advantageous to optimize density. Some important comments are as follows: • Balancing the rate of plant production, trucking, paver speed, and roller speed is the key to a successful operation. • Trucking is the main challenge to maintaining a consistent paver speed. • Slower speed allows the roller to keep up with paver. • Why should the contractor be handcuffed if the density is being achieved? • Project estimates drive daily production. Also, 72 percent of the respondents indicated that a maximum paver speed should not be included in the paving specification. The major concern expressed was that the maximum paver speed was a contractor means and methods responsibility and should not be part of the specifica- tion. Most of the respondents who believed a maximum speed should be specified were agency personnel with a few contractor QC personnel. Consistency of the paver speed was also a com- mon comment. Longitudinal Joint Construction The longitudinal joint is the seam between two passes of the paver during the construction of an asphalt concrete pavement. The location of joints are critical in pavement construction because, if not constructed properly, they allow for the ingress of water into the pavement structure and can lead to early deterioration of the asphalt concrete pavement. Joint density Table 26. Typical paver speed. Typical Paver Speed, feet per minute Percentage of Responses 10 < 20 5.6 > 20 < 30 17.8 > 30 < 40 21.1 > 40 < 50 5.6 > 50 < 60 2.2 Unknown 46.7

42 Specifying and Measuring asphalt pavement Density to ensure pavement performance requirements are a relatively new type of specification but are seen to be increasing in usage throughout the country. Despite extensive research and training activities, Buncher and Rosenberger (2012) conclude that poor performance of longitudinal joints continues to be one of the highest listed reasons for the premature failure of asphalt concrete pavements. Buncher and Rosenberger report that states that have implemented joint density specifications have also reported improved pavement performance. The report by Buncher and Rosenberger (2012) provides useful definitions applicable to longitudinal joint construction: • Cold lane is the first lane paved and will have one or two unsupported edges. • Hot lane will have at least one edge that is placed against an existing lane or shoulder. • Hot longitudinal joint is formed when two pavers are used in echelon and the longitudinal joint is completed before the material in the cold lanes has had a chance to cool. • Cold longitudinal joint is formed when the first lane was paved previously or when the time between the first and second passes of the paver is such that the first pass has already cooled. Based on their study, Buncher and Rosenberger conclude that joints should be overbanded or have a joint sealer applied if densities are below 92 percent of Gmm. Ohio is an example of a state that requires a hot applied asphalt or joint sealer on the vertical face of the mat prior to the second pass being placed for all cold longitudinal joints on interstate pavements (Ohio DOT Standard Specifications, Paragraph 401.17). Survey Question 1 Does your organization have a longitudinal joint density specification? What type of specification is the longitudinal joint density requirement? Background There is not a “one size fits all” approach to achieving longitudinal joint density. The same can be said about the longitudinal joint density specification. Survey Analysis The responses for having longitudinal joint density specification were approximately evenly split between yes (41 percent) and no (44 percent). Of the respondents, 15 percent indicated that they required the application of a longitudinal joint density specification on some jobs only. Twenty-two agencies responded in the affirmative (42 percent) while 30 responded in the negative (58 percent). Typical comments included: • We are considering a joint density specification but have not yet implemented it. • All jobs have a joint density specification of 90 percent, minimum. • Some question the efficacy of the longitudinal joint specification. • Some states require a joint sealant on the final layer for all pavements. In the survey, 87 percent of the respondents indicated that the longitudinal joint density speci- fication was a percent density requirement while 13 percent indicated that their specification was a method specification. These same percentages were reflected by responses from agencies only. An exception to the determination of percent density at the joint is the Maryland method for longitudinal joint construction. In this method, material at the joint is overlaid from the hot side

Survey to Determine the Density of asphalt pavements 43 to the cold side and not “bumped back” as is often done. When rolled, there is a white stripe of crushed aggregate that occurs on the cold side. This method has proven to be very effective both from a performance perspective and from a cost perspective. Survey Question 2 How is the longitudinal joint density measured? What is the frequency for joint density measurement? Background The options for measuring the longitudinal joint density are the same as the options for mea- suring the overall mat density—cores or NDT. Because the joint is a discontinuity in the asphalt pavement (unless paving in echelon), it is reasonable that a difference in density may occur. In addi- tion, the distinction between the hot and cold side of the mat can also cause a difference in density from one side to the other. The first or cold side had no lateral support at the time of placement but the second or hot side does have lateral support. Survey Analysis Seventy-eight percent of the respondents indicated that joint density is measured by core testing. Nine respondents indicated the use of a nuclear gauge for joint density measurement while four respondents indicated the use of an electromagnetic gauge. Seventy-seven percent of the respondents said that joint density tests were determined by statis- tical lots while 20 percent said that they were done daily. Comments included: • Four sublots per lot, • Three joint cores per daily lot, • One core per 5,000 linear feet of joint, • One core per 3,000 linear feet of joint, • One core per 1,500 linear feet of joint, and • Five cores per 2,500 liner foot lot. It is seen from the comments that there is significant variation in the frequency of longitudinal joint density testing. Survey Question 3 What is the lower limit for the longitudinal joint density specification? Is the joint density requirement the same for all projects? Background Buncher and Rosenberger (2012) recommend that the minimum specification for joint density be 2 percent lower than the mat density and/or a minimum of 90 percent of Gmm. Survey Analysis The responses to the lower limit question were relatively uniform. As shown in Table 27, 52 per- cent responded that the lower limit was 90 percent and the remaining 48 percent were split evenly between below 90 percent and above 90 percent. The overwhelming comment was that longitu- dinal joint density should be within about 2 percent of the mainline density.

44 Specifying and Measuring asphalt pavement Density to ensure pavement performance In the survey, 81 percent of the respondents (i.e., 89 percent of agency respondents) indicated that their joint density requirement was the same for all projects. Survey Question 4 What is the joint density pay based on? Background TRB defines the pay factor as a multiplication factor, often expressed as a percentage, used to determine the contractor’s payment for a unit of work based on the estimated quality of work (Transportation Research Circular E-C173). As previously discussed, the specification is often written with both incentive and disincentive elements. Survey Analysis As shown in Table 28, 41 percent of the respondents indicated that both incentive and dis- incentive pay factors were used to assess joint density. While 20 percent use a disincentive-only policy, 14 percent don’t have any pay adjustment for longitudinal joint density. Comments from the respondents who don’t have pay adjustment are that joint density is not a pay item. Others commented that joint density carries a shutdown clause. Survey Question 5 Who makes the decision whether to use a joint density specification or not? Background Since longitudinal joint density specifications are not currently required for all projects, the purpose of this question was to ascertain how a decision to include the specification is made. Survey Analysis As shown in Table 29, the responses were approximately equally divided between the agency specification writer and the joint density requirement being part of the standard specifications Table 27. Lower limit for longitudinal joint density specifications. Lower Limit, Percentage Gmm Percentage of Responses 88 10.0 89 14.0 90 52.0 91 14.0 92 2.0 Same as Mainline 8.0 Table 28. Basis for joint density pay. Pay Basis Percentage of Responses Incentive only 1.4 Disincentive only 20.0 Both incentive and disincentive 41.3 Neither incentive nor disincentive 14.3

Survey to Determine the Density of asphalt pavements 45 for the agency. For agency respondents only, 58 percent responded that joint density specifica- tions were in the standard specifications of their organization. Emerging Technologies for Achieving Density The concept for obtaining density in asphalt pavements has not changed much since the early 1900s. The paving equipment and the rolling equipment have become more sophisticated with electronic control, comfort for operators, and better management tools for the process. But fun- damentally both equipment perform the same function—the paver places the asphalt concrete mix uniformly and the roller compacts the mixture to a specified level of density and smoothness (Epps et al., 2000). The significant change in the last 50 years or so for both the paver and the roller has been the free-floating screed and the vibratory roller, respectively. Kilpatrick and McQuate (1967) state: Normal rolling procedures used by roller operators result in wide lateral variations in compactive effort. The number of roller passes applied in the center of the lane is usually 3 to 6 times greater than at the lane edges. While the comment was made 50 years ago, the idea is still germane. The center of the mat will always have more compaction than the edges because of the overlap of the rolling operation. However, the challenge is to use technology to improve the consistency of density laterally across the mat. In the last decade or so, the concept of intelligent compaction (IC) for asphalt concrete pave- ments has evolved at least partially driven by a need to add competent personnel into the asphalt pavement workforce. Gallivan et al. (2011) offers the following definition for IC: IC is defined as a process that uses vibratory rollers equipped with a measurement/documentation system that automatically records various critical compaction parameters in real time during the compaction pro- cess. The recorded information is then displayed for the roller operator and project personnel to improve the compaction process. The original concept focused on two different elements of the process. The first element (den- sity control) uses a system to determine the density of the mat during the compaction process. This consists of equipment to monitor the stiffness and temperature of the mix in order to relate to the density being achieved in the asphalt concrete. Since the stiffness of the mix is constantly changing with decreasing mix temperature, determination of the density using IC has proven to be unreliable (Chang et al., 2014). The second element (roller management) equips the roller with GPS sensors to track the horizontal position of the roller on the pavement as it is being rolled. This roller management approach permits the roller operator to know how many passes have been made on a specific location on the pavement through the installation of a GPS unit on the roller and a monitor in the operator station of the roller. The number of passes on the mat can be accurately controlled by using this technique. Table 29. Who decides to use a joint density specification? Decision Maker Percentage of Responses Agency specification writer 32.9 Agency construction manager 9.6 In standard specifications so it is used on all projects 35.6

46 Specifying and Measuring asphalt pavement Density to ensure pavement performance Survey Question 1 Is Intelligent Compaction (IC) being used in your jurisdiction? If so, does your agency have a specification for IC? Background Whether using IC for density control or roller management, the agency needs to prepare specifications that are reasonable, fair, and achievable. As with any new technology, it is desirable to have trial projects to confirm the process. At the time this document was written, IC was in its infancy. Undoubtedly, it will continue to grow and mature. Survey Analysis Sixty-five percent of the respondents indicated that IC was not being used in their area but that they were reviewing the technologies. Confirming the overall response, 36 of the 56 agencies (64 percent) responding indicated that IC was not being used. As with any new technology, some agencies are performing pilot or demonstration projects to evaluate the use of IC. As shown in Table 30, 72 percent of the respondents indicated that they do not have an agency specification for IC. A common comment was that an IC specification, if used, was project specific. This further illustrates the idea of trying out the technology for evaluation purposes. Understandably, agencies are risk averse in trying out new technologies. Survey Question 2 Is IC being used for density control or for roller pattern management? Background Chang et al. (2014) conducted an extensive study on the use of IC. For density control, the process measures stiffness of the mat. The stiffness is obviously based on the temperature of the mix that decreases constantly during the paving operations. Chang et al. concluded that IC is not recommended to replace cores for density control for acceptance purposes. The roller management element of IC is accomplished by the mounting of a GPS unit on the roller to track its movement. This technology is relatively straightforward, useful, and currently available for roller management. Survey Analysis Table 31 presents the survey results for IC usage. About 54 percent of the respondents are not currently using IC. About 20 percent use IC for roller pattern management and about 16 percent use both the density control and roller management aspects of IC. These results are reflective of a technology in its infancy. Survey Question 3 How effective is IC for density control of asphalt mixtures? For roller pattern management? Agency Specification Percentage of Responses No specification 71.6 Yes, required in the specifications 21.6 Yes, permitted in the specifications 6.8 Table 30. Agencies with IC specifications.

Survey to Determine the Density of asphalt pavements 47 Background Reports by Gallivan et al. (2011), Gallivan (2012), and Chang and Gallivan (2014) have dem- onstrated that the density control element of IC is not currently a practical, reliable approach. Additional development is required to be able to effectively use the technology. However, the roller pattern management element of IC is currently available and usable. The current challenge for implementing the roller pattern management technology is the cost. It is possible to retrofit existing rollers with the technology as well as to buy equipment with the IC package already in it. As with the implementation of any new technology, there is a steep learning curve for acceptance. Survey Analysis Approximately 60 percent of all respondents have no experience with either density control or roller pattern management using IC technologies, as shown in Table 32. Seventy-three percent of the agencies indicated no experience with IC while about 30 percent of the contractors indicated no experience. One thought-provoking comment was whether IC was a solution looking for a problem. The comment came from a person who indicated that after poor mix practices were properly addressed in their jurisdiction, most density penalties were eliminated. This person believed that IC would have masked the fact that the mixes were under-asphalted. From the roller pattern management perspective, comments were positive as a means to ensure proper coverage of the mat by the roller. It is interesting to note that about 36 percent of the respondents indicated that the roller pattern management was either moderately effective or very effective. This is likely indicative of the implementation of any new technology. Seventy percent of agencies indicated no experience with IC for roller pattern management. Survey Question 4 Are tamping bar screeds used in your jurisdiction? Is it being considered by specifying agencies? Background A tamping bar screed is defined as “one or more tamper bars to deliver extra compaction energy to the asphalt layer and produce higher density in the asphalt layer prior to the compaction IC Uses Percentage of Responses Density control 1.2 Roller pattern management 19.3 Both density control and roller management 15.7 Neither density control nor roller management 54.2 Table 31. Current applications for IC. Table 32. Effectiveness of IC for density and roller pattern management. Percentage of Responses Effectiveness Density Control Roller Pattern Management Very effective 3.4 18.1 Moderately effective 9.1 18.1 Doesn’t work at this time 17.1 4.8 No experience at this time 60.2 59.0

48 Specifying and Measuring asphalt pavement Density to ensure pavement performance process” (Guide to Asphalt Paving, 2016). This device is part of the paver and provides additional compaction to the mat prior to the breakdown roller. While used extensively in Europe, the tamping bar screed has not been widely used in the United States. Survey Analysis In the survey, 83 percent of all the respondents (91 percent of agencies) indicated that a tamp- ing bar screed is not used in their jurisdiction. Some agencies allow their use but do not require the tamping bar screed. Only 36 percent of the contractors indicated that they have used tamp- ing bar screed pavers, primarily for specialty mixes, but noted that the tamping bar screed is not commonly used. As high as 96 percent of all the respondents (98 percent of agencies) indicated that the tamping bar screed was not being considered by the specifying agency in their area.