Methods and Practices on Reduction and Elimination of Asphalt Mix Segregation (2015)

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.

7 Other terms that have been used to mean end-of-truck seg- regation include truck to truck (AASHTO 1997), truckload to truckload (Scherocman 2011), periodic segregation on each side (Brock et al. 1998), and chevrons of rough texture (Murphy 2012b). Random Segregation Random segregation is coarse textured areas at irregular inter- vals and is the least frequent type of segregation seen behind the paver (Scherocman 2011). The coarsely textured areas can be localized transverse or longitudinal in no consistent or continu- ing pattern (Figure 4). Random segregation with a fine texture (i.e., fine aggregate segregation) is more difficult to visually identify and can be seen under certain lighting conditions or with the aid of texture measurements. Random segregation can occur during the formation and handling of aggregate stockpiles, asphalt plant operations, clumping of mix components, and as windrows are formed. Other descriptions for random segregation that are related to the separation of the asphalt mix components include “clumps” (e.g., non-mixed fibers, polymers, or other additives) or “fat spots” (e.g., binder-rich areas) (Figure 5). Clumps of additives occur because of the improper location of addi- tion during mixing or insufficient time in the batch or drum mixer. Fat spots can occur because of the binder draining off of the aggregate surface during silo storage, asphalt mix transfer, or asphalt mix placement. Longitudinal Segregation Longitudinal segregation is described as a stripe, or streak, of coarsely textured asphalt mix behind the paver. Coarse textured stripes can occur in the center of a lane that is usually under the gear box at the back center of the paver. Longitudinal stripes on either or both sides of the center of the lane can correspond to the outside edges of the paver conveyors, at the edges of the screed, and where screed extensions start at the edge of the fixed screed. Centerline segregation is a longitudinal strip of coarsely textured mix under the screed auger gear box located in the center of the screed. The literature review information contained in this chapter is covered in the following sections: • Descriptions of segregation • Detecting segregation • Testing in segregated areas • Segregation specifications • Pavement condition in segregated areas • How and where mix segregates. DESCRIPTIONS OF SEGREGATION Mix segregation is usually described by the location and the pattern of the coarsely textured areas in the finished mat. Although a range of terms and descriptive phrases have been used by various agencies and consultants for specific types of segregation, the following terms will be used for consistency throughout this report: • End-of-truck segregation • Random segregation • Longitudinal segregation • Temperature segregation. Longitudinal joint construction and segregation at the joints is not specifically covered in this synthesis. However, success- ful practices that reduce or eliminate longitudinal segregation are also considered successful practices for minimizing joint segregation. End-of-Truck Segregation End-of-truck segregation (AASHTO 1997; Scherocman 2011; Warren 2013) is described as a separation of coarse and fine aggregate fractions in the asphalt mix and appears behind the paver as two coarser textured areas in a transverse location on either side of the center of the paver (Figure 3). The pattern of the coarser areas is commonly described as a chevron. End-of-truck segregation occurs because of improper load- ing of the silo, improper loading and unloading of haul trucks, running the paver hopper too low or empty, dumping left- over mix in the paver wings (i.e., “flipping” the wings), or not removing spilled mix. chapter two LITERATURE REVIEW

8 Longitudinal segregation (Figures 6 and 7) can occur because of improper loading of the bin batcher, under- or over-filling the silos, running the paver screed augers too slowly (“starving” the augers), and paver designs. When mix is segregated as it leaves the asphalt mix plant and is not adequately remixed before transfer to the paver, the segrega- tion will move through the paver and appear as longitudinal segregation. One-sided longitudinal segregation can also be caused by an imbalance in the volume of mix across the width of the screed augers. The segregation can be seen on the side of the screed with the lowest volume (“starving” the augers on one side). Longitudinal centerline segregation occurs when the coarser aggregate rolls off of the paver hopper conveyors or coarser FIGURE 3 Example of end-of-truck segregation [Source: Stroup- Gardiner and Brown (2000)]. FIGURE 4 Example of random segregation [Source: Stroup- Gardiner (2014)]. FIGURE 5 Example of random segregation (separation of binder and stone) in SMA [Source: Stroup-Gardiner (2014)]. FIGURE 6 Example of longitudinal segregation on one side [Source: Adams et al. (2001)]. FIGURE 7 Example of longitudinal segregation, both sides [Source: Murphy (2012a)].

9 aggregate at the inside edges gets dropped under the gear box as the mix moves toward the screed augers (Figure 8). Temperature Segregation Temperature segregation is localized areas of cold mix sur- rounded by a majority of mix with hotter and more uniform temperatures. Temperature segregation is measured with infra- red temperature “guns,” infrared camera image analysis, or infrared sensor units (e.g., Pave-IR) mounted to the back of paver. Research by Gunter (2012) showed that the four most fre- quent locations to detect cold mix are at the end of each truck load of mix, when the paver stops to change haul trucks, when the paving operation stops while waiting for haul trucks to arrive, and in areas with handwork (Table 1). Hand work is typically done at the stop-start of paving operations and at the joints and edges of the fresh mat, and may or may not indicate segregated mix. DETECTING SEGREGATION The following five methods are currently used or have some potential to detect segregation behind the paver: 1. Visual 2. Surface texture 3. Temperature differences 4. Ground penetrating radar (density mapping) 5. Intelligent compaction. Visual Method Visual detection has been used the longest and is considered the benchmark against which other methods for detecting seg- regation are compared. One study documented the subjectiv- ity (i.e., high variability) in visual detection of segregation. A Texas study in 1999 detailed five expert inspector’s evalu- ations of 5.11 miles of pavement construction (Tahmoressi et al. 1999). This length of paving was divided into nine test sections that investigated the influence of various methods of mix transfer on segregation. In the most efficient case, the number of segregated areas identified by each expert var- ied from 7 to 9 for a specific 1,000-ft sublot of paving. In the worst case, the number of segregated areas identified by each expert varied from three to 11 for the same section of pavement. Because the severity of the segregation significantly impacts the loss of pavement life, there has been some effort to define the levels of segregation by visual examination. Gavin and Heath (2002) reported that slight segregation is described as a “. . . matrix, asphalt cement and fine aggregate, [that] is in place between the coarse aggregate. However, there is more stone in comparison to the surrounding acceptable mix . . .” (Figure 9). Moderate segregation is described as “. . . signi- ficantly more stone than the surrounding mix; moderately FIGURE 8 Example of centerline longitudinal segregation under gear box [Source: Stroup-Gardiner (2014)]. Cold Areas on South Carolina Projects Ranking and Type of Non-Uniformity Reasons for Cold Areas No. of Images Mix Temp. Mix Transfer End of Truck 129 1 Work Stoppages 98 2 Cold Mix Due to Waiting During Long Work Stoppages 40 3 Liquid Spills on Mat 3 9 Paver Operations Wing Dump (“flipping”) 18 7 Streaks in the Mix 25 5 Mechanical Problems 5 8 Start-Up, Cold Joint, Etc. 21 6 6 Miscellaneous Hand Work 31 4 Environmental (Weather) 2 10 Source: Gunter (2012). Ranking: 1= largest number of infrared images with cold areas. 10 = fewest number of infrared images with cold areas. TABLE 1 RANKING OF CAUSE OF TEMPERATURE DIFFERENCES

10 segregated areas usually exhibit a lack of surrounding accept- able matrix . . .” (Figure 10). Severe segregation “. . . appears as an area of very stony mix, stone against stone, with very little or no matrix . . .” (Figure 11). The major disadvantage with visual assessment is that each person assesses the mat texture based on their own experiences and interpretation of segregation (Mahoney et al. 2003). Varia- tions in lighting, angle of view, and shadowing increase variations in the visual appearances of the mat. Differ- ences of opinion between experts performing the visual assessments lead to discussions, arguments, and require- ments for dispute resolutions, any of which delay construc- tion, increase the time needed for lane closures, increase the project cost for the agency, generate additional testing requirements, and may result in lost revenue for the con- tractor (i.e., disincentives). ➢ Experts visually inspecting a project for segregation can frequently disagree on the number and extent of segregation. Surface Texture Quantitative measures of the surface texture are useful for eliminating the subjective nature of visual assessments of texture. Surface texture can be measured using: • Static texture measurements, • Longitudinal texture profiles, and • Photographic image analysis. Static Texture Measurements Static texture measurements are tests that evaluate a limited area of the pavement surface either immediately after con- struction is completed or at some time after the roadway is opened to traffic. Once the roadway is open to traffic, traf- fic control is required during testing for worker safety. The simplest method of estimating the surface texture is with the ASTM E965 Standard Test Method for Measuring Pave- ment Macrotexture Depth Using Volumetric Technique. This method uses a known volume of fine sand that is spread in a circle on the pavement surface. The diameter of the circle and the mass and volume of the sand are used to estimate the depth of the surface voids (Figure 12). The circular texture meter uses a laser sensor to measure the texture profile of an 11.2-in. (284-mm) diameter circle according to ASTM E2157 Standard Test Method for Measuring Pavement Macrotexture Properties Using the Circular Track Meter (Meegoda et al. 2002, 2003; Applied Pavement Technology 2008). Both static measurement methods provide similar, but somewhat offset, texture measurements. ➢ Static texture measurements can quantify texture changes resulting from segregation, but they only provide mea- surements for a small area of the overall pavement. Longitudinal Texture Profiles Longitudinal texture profiles are obtained using vehicle- mounted, high-speed laser distance measurement sensors usually mounted over the right wheel path. The sensor(s) mea- FIGURE 9 Example of slight level of segregation [Source: Gavin and Heath (2002)]. FIGURE 10 Example of moderate level of segregation [Source: Gavin and Heath (2002)].

11 sures the distance from the mounted position to the pavement surface every 2 mm (0.8 in.) of longitudinal distance. The data are used to develop a longitudinal profile of the texture depth (Meegoda et al. 2002, 2003; Williams 2003). End-of-truck segregation is typically seen as cyclic peaks in the surface texture at intervals that correspond with the length of paving completed with each haul truck (Figure 13). The use of a material transfer vehicle (MTV) can reduce seg- regation, but may not completely eliminate it such as when the mix is not fully reblended by the MTV. However, the cyclic peaks, although reduced in height, can still be detected with the longitudinal texture measurements. ➢ Longitudinal texture profiles quantify: – Texture changed as a result of end-of-truck segregation. – Effectiveness of material transfer units for reducing end-of-truck segregation. Photographic Image Analysis Digital photographs can be used to record evidence of visible texture differences and various analysis methods can be used to mathematically quantify the changes in the pixel bright- ness (de Leon Izzepi et al. 2006; de Leon Izzepi and Flintsch 2006a, b; Zelelew and Pagagiannakis 2011). The changes in the pixel color or grey tone values indicate the consistency of the pavement texture seen in the photograph. The simplest representation of the data is a histogram of the brightness val- ues (Figure 14), which indicates a fairly uniform distribution of pixel color and therefore a uniform texture. When the histo- gram is wider and more brightness values are contained in the distribution, the image captures non-uniform texture character- istics. This is still an experimental approach, as non-uniform texture areas not only locate potentially segregated areas they also identify white pavement markings, pockets of moisture, and tire marks on new pavements as having non-uniform tex- ture. On-site visual evaluation for segregation detection is nec- essary for areas having high variability in the image analysis. ➢ Photographic image analysis has the potential to detect textural changes resulting from segregation. – Lighting, environmental conditions (e.g., moisture, clouds, etc.), and other pavement textures such as FIGURE 11 Example of severe segregation [Source: Gavin and Heath (2002)]. Sand Patch Method Circular Texture Meter FIGURE 12 Examples of static circular area texture measurement methods [Source: Hanson and Prowell (2004); APT (2008)].

12 striping and tire tracks can significantly influence the image analyses. Temperature Differences Temperature differences can be used to locate non-uniformity in the mix, production, and paving process. Areas with suf- ficiently large localized temperature differences can be tested or sampled for laboratory testing to determine if the mix prop- erties meet the specification requirements. Temperature differences can be determined using handheld infrared sensors (“guns”), infrared cameras, or paver-mounted infrared sensors and computer systems. The simplest method for identifying temperature differences in the mix is with the infrared gun. This technology is economical, easy to use, and 0 0.1 0.2 0.3 0.4 0.5 0.6 0 1,000 2,000 3,000 4,000 Te xt ur e, m m Distance, feet 0 0.1 0.2 0.3 0.4 0.5 0.6 0 1,000 2,000 3,000 4,000 Te xt ur e, m m Distance, feet A) Without MTV B) With MTV Average texture Average texture Peaks are End of Truck Texture Variations FIGURE 13 Longitudinal texture profile changes with and without MTV [Source: Williams (2003)]. N um be ro fP ix el s N um be ro fP ix el s Pixel Grey Scale Value Pixel Grey Scale Value A) Uniform Texture B) Segregated Mix Texture FIGURE 14 Example for the analysis of photographic images [Source: de Leon Izeppi et al. (2006b)].

13 numerous measurements can be taken throughout the pav- ing process. However, the area included in the temperature reported on the device’s display can change substantially owing to the distance of the gun to the area being tested and the distance to spot (D:s) characteristic of the device that defines the area averaged in the measurement. The results from these devices may also need to be hand-recorded, as most guns do not have a data collection component. ➢ Temperature measurements obtained with infrared guns can vary substantially depending on the device D:s ratio and the distance of the user to the target. When these factors vary, it is difficult to accurately detect temperature segregation. Infrared cameras are more expensive but provide a more detailed method of “photographing” the pavement during construction. Thermal photographs, such as digital photo- graphs, are dependent on the knowledge and experience of the person operating the camera, as well as the options avail- able on the camera. The following is to be considered when acquiring thermal images so that temperature segregation is not over- or underestimated: • The area of the paving included in each infrared image and analysis is to be standardized as much as possible (Gunter 2012) (Figure 15). It is important that non-paved areas or right-of-way obstructions be excluded from the analysis area • Images are to be in focus so the color analysis of pixels can be accurately analyzed (Figure 16). • Zooming in on one section can result in too small an area of the pavement being represented in the image and complicate subsequent analysis (Figure 17). It is diffi- cult to assess areas of segregation when the image does not show the width of the lane. • Temperature scale increments can be set by the user on some devices to highlight incremental tempera- ture changes of interest. For example, the temperature scale can be set to indicate increments of 18.8°F (10°C), which helps show the larger chevron pattern of end-of- truck segregation (Figure 18). • Newer infrared cameras can also capture a digital image at the same time the thermal image is saved so that visible evidence of segregation can be documented (Figure 19). ➢ Standardized protocol for image collection and camera scale setting are necessary to reliably detect temperature segregation. – It is essential that infrared camera images be in focus and represent a sufficient area of the pave- ment so that non-uniform and uniform areas can be identified in the analyses. Temperature differences have been directly linked to visu- ally identifiable types of segregation (Adams et al. 2001); thermal and digital photographs that show: • End-of-truck segregation (Figure 20), • Longitudinal centerline segregation (Figure 21), and • Longitudinal one-sided segregation (Figure 22). Note that Figure 21 also includes temperature differences resulting from longitudinal ridges and depressions because of the screed settling down during a paver stop. Although some temperature differences directly indicate visibly iden- tifiably segregation, other temperature differences indicate non-segregated textural changes. The Pave-IR™ unit combines paver-mounted infrared sensors, a computer system for collecting and analyzing FIGURE 15 Example of standardizing thermal image for analysis by eliminating non-paving areas from the analysis [Source: Gunter (2012)].

14 the temperature data, and a global positioning system (GPS) to provide a location reference for temperature anomalies and paver stops (Figure 23). A visual display allows the pav- ing crew to identify when paving operations or equipment would be adjusted to improve the uniformity of the mat properties. ➢ Infrared temperature profiles developed using infrared sensor bars such as the Pave-IR keep the distance-to- target constant, standardize the area of the pavement included in each profile, and document the location of any non-uniform areas with the unit’s GPS technology. All of these features help to accurately detect any tem- perature segregation. ➢ Temperature images (infrared camera, infrared sensor bar) can help detect types of segregation that can also be seen as visibly detectable segregation. Ground Penetrating Radar Segregated areas of the pavement typically have lower den- sities and higher air voids; therefore, technologies that can evaluate these properties may have the potential to detect segregation. Several states have active ground penetrating radar (GPR) programs for the measurement of layer thick- nesses, pockets of water, determination of voids, delamination on bridge decks, or moisture in or damage to asphalt pave- FIGURE 19 Example of image from newer infrared camera that simultaneously collects digital and infrared image [Source: Song et al. (2009)]. There is not enough data contained in the image to determine significant temperature changes across the width of the mat. FIGURE 17 Example of a too-close view of a temperature anomaly [Source: Henault et al. (2005)]. Blurry edges in image will influence the data analysis of the image which is based on pixel color values FIGURE 16 Example of an out-of-focus temperature image [Source: Henault et al. (2005)]. 60.0°C 160.0°C 60 80 100 120 140 160 SP01 SP02 AR01 AR02 Each color = 10°C change in temperature FIGURE 18 Example of setting the thermal image colors to represent increments of 10°C [Source: Gunter (2012)].

15 ment (Al-Qadi and Lahouar 2004). GPR measurements of the dielectric constant are correlated with densities or air voids of cores taken from the pavement to be tested. Mix proper- ties can be obtained for a portion of a pavement lane or the entire lane width, depending on the number of GPR units that are mounted on the vehicle (Figure 24) (Sebesta and Scullion 2002, 2012; Sebesta et al. 2006, 2013). ➢ GPR technology can be used to develop a geospatial density map, but it is unclear if the technology can detect localized areas of low density resulting from segregation. ➢ Further evaluations of the GPR technology are impor- tant in this area. Intelligent Compaction The lower densities and higher air voids in segregated areas are responsible for the loss of mix stiffness in those areas. The intelligent compaction (IC) technology uses an instrumented tandem steel wheel vibratory roller to estimate changes in freshly placed asphalt mix stiffness in the mat during break- down rolling. The IC technology is comprised of a GPS unit, a display for the roller operator, a temperature sensor, and an accelerometer sensor mounted to the front steel drum. Changes in vibrations from the eccentric weights inside the drum are used to estimate changes in stiffness (Figure 25). Areas of the pavement with segregation-related stiffness changes may influence changes in the IC roller vibrations. It is possible this technology may be capable of detecting segregation by detect- ing areas with low stiffness; however, no research has yet been conducted to investigate this possibility. Intelligent compaction technology has the potential to: • Provide real-time feedback to the roller operator about the number of passes completed, roller speed, mat tem- perature, and the sensor data needed for analyses. • Document all data collected during compaction. • Statistically evaluate data and report information in geo- spatial format. • Identify underlying “soft spots” that may be detrimental to the compaction of the new mat over these areas. FIGURE 20 Temperature differences that reflect visible changes in texture (end-of-truck segregation) [Source: Adams et al. (2001)]. Screed Extension Ridge Lines FIGURE 21 Temperature differences that reflect visible changes in texture (centerline segregation and screed extension anomalies) [Source: Adams et al. (2001)]. FIGURE 22 Temperature differences that reflect visible changes in texture (longitudinal one-sided segregation) [Source: Adams et al. (2001)].

16 • Potentially identify areas of low stiffness in new layers that may also indicate segregated areas. • Prevent “over-compaction,” which is detrimental to obtaining optimum stiffness. Current barriers include: • Uncertainty of the technology to adequately differentiate between the stiffness of each pavement layer. • Limited availability of equipment that meets the FHWA criteria for IC equipment that requires high precision GPS, IC valves, and temperature measurement. • Need for simplification of data collection, management, and analyses processes. ➢ While the IC technology has been used to develop geo- spatial density, modulus, and underlying support maps, it is unclear if the technology can identify localized areas of low stiffness resulting from mix segregation. ➢ Future research programs could explore this possible use for the IC technology. Summary of Methods for Detecting Segregation Each method for detecting segregation has advantages and disadvantages (Table 2). TESTING Over the last 20 years, a number of research and pilot proj- ects were conducted related to different methods for detect- ing segregation to measureable changes in density, air voids, Temperature Profile Output La ne W id th Paving Direction Infrared Sensors Display Screen “Chevron” Pattern Indicating End of Truck Segregation FIGURE 23 Pave-IR system for collecting and analyzing temperature data during paving [Source: Rand (2012)]. Eccentric Weights Inside Drum Temperature Sensor Display for Driver GPS Accelerometer on Drum FIGURE 25 Schematic of sensors and electronics used for intelligent compactors [Source: Stroup-Gardiner after Haskell (2007); FHWA (2011)]. FIGURE 24 GPR vehicle developed by Texas Transportation Institute [Source: Sebesta et al. (2006)].

17 Detection Method Advantages Disadvantages Visual Detection • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • Benchmark method for detection of segregation Detects mix segregation (coarse and fine) Locates binder-rich areas No equipment needed Mix can be evaluated for segregation and can be evaluated during or after construction No cost Subjective Only evaluates surface anomalies Requires experienced field staff, inspectors, contractors Can be wide differences in detection of segregation between experts Fails to locate localized areas of poor density not associated with mix segregation Difficult to see texture changes during night- time paving Does not always identify areas with significantly different mix properties Infrared Gun Identifies areas with temperature differences Effective during day or nighttime paving Easy to use Immediate results obtained Economical Only provides spot-specific measurements Only evaluates surface anomalies Measurements need to be recorded manually Area of mat included in the measurement is device-dependent, distance-dependent, and user-dependent Does not always identify areas with significantly different mix properties Infrared Images or Profiles Detects areas with temperature differences Provides record of temperature variability on entire project Software can indicate significant temperature differences in real time so corrections to the construction process can be made immediately Effective during day or nighttime paving Data needs to be collected during construction Only evaluates surface anomalies or those underlying anomalies that influence surface temperature Technician training and standardized data collection method needed Does not differentiate between mix segregation and temperature segregation Moderate cost Does not always identify areas with significantly different mix properties Quantifies texture changes seen with Testing can only be done after compaction is Surface Texture visual detection of segregation Data can be collected during or after construction Data can be collected at highway speeds using an inertial profiler with a high frequency laser Longitudinal texture profiles can evaluate the successful remixing of segregated mixes by material transfer devices complete Only evaluates surface anomalies Vehicle mounted methods only provide longitudinal profile Static texture measurement methods only provide single point measurements and may require traffic control during testing Does not always identify areas with significantly different mix properties Lighting and environmental conditions can influence results TABLE 2 ADVANTAGES AND DISADVANTAGES FOR METHODS FOR DETECTING SEGREGATION (continued on next page)

18 gradations, and asphalt content (Table 3). These mix prop- erties are typically assessed during normal quality control (QC) and quality assurance (QA) testing. When segrega- tion is detected, additional testing can be done to evaluate if the mix properties in these areas meet the specification requirements. Quality Control/Quality Assurance Testing Density and Air Voids In-place densities or the densities of cores can be used to determine if the mix properties exceed the specification lim- its. Air voids are calculated from test results used to calcu- late density and the maximum specific gravity determined during the mix design. Some states specify densities in their specifications, while others specify air voids or the percent of maximum density, which is another form of controlling air voids. One Texas study found that density profiles in visu- ally detected segregated areas only failed to meet the speci- fication criteria 17% of the time, whereas profiles in other segregated areas failed to meet the criteria 83% of the time (Tahmoressi et al. 1999). Most of the research reported in the literature shows that at least a moderate or high level of segregation is required before the properties fail to meet the specification require- ments (Wolff et al. 2000; Adams et al. 2001; Mahoney et al. 2003; Willoughby et al. 2003). Roadway (In-Place) Density Testing Methods Roadway Density testing is accomplished using either a nuclear density gauge or a non-nuclear device such as the Pavement Quality Indicator (PQI) (Figure 26). The nuclear density gauges are commonly used by both agencies and contractors for QC. A few agencies allow nuclear density gauge readings for acceptance in specific cases. Pavement surfaces with a coarse or open texture require the surface voids be filled with fine sand and leveled before determining the density with a nuclear density gauge. This is necessary because the small gap caused by the Ground Penetrating Radar (GPR) Estimates changes in density due to • • • • • • • • • • • • • • • • • • • changes in mix properties, which is a key factor in the pavement service life Can provide information on densities across the full lane width of the project Has potential for density monitoring during compaction when mounted on roller Limited experience with this specific use of technology as a means of detecting segregation Testing can only be done after compaction is complete GPR measurements of dielectric constant need to be correlated with core densities (air voids) for each project Requires significant technician training Complicated analysis High cost Intelligent Compaction Estimates changes in layer stiffness Evaluates the full lane width of layer properties during construction Information during rolling can immediately show operator where coverage is not complete Real time display allows the roller operator to proactively adjust rolling patterns to achieve the most uniform stiffness Existing vibratory compaction equipment can be retrofitted with sensors and data collection/analysis devices Retrofit only moderately expensive Results are dependent on the underlying pavement structure stiffness and layer thicknesses Current use focused on overall project stiffness and evaluation of the uniformity of existing pavement structure stiffness Mixed results in linking results with density, estimated from stiffness information, of individual layers No research has been conducted for use in detecting segregated mix Detection Method Advantages Disadvantages TABLE 2 (continued)

19 State References No. Sect. Gradation Information* Year Constructed Detection Method QC/QA Properties Alabama Cerdas (2012) 28 OGFC (1), SMA (3), WMA (3) 100% passing 25 mm (3) 100% passing 19 mm (9) 100% passing 12.5 mm (7) 100% passing 9.5 mm (2) 2010 to 2011 Temperature Density Mix Properties Alabama Stroup-Gardiner and Brown (2000) 2 100% passing 25 mm (2) 1998 Visual Density Mix Properties Colorado Gilbert (2005) 20 SMA (1) 100% passing 25 mm (11) 100% passing 19 mm (8) 2004 to 2005 Temperature Density Connec- ticut Henault (1999) Henault et al. (2005) 11 100% passing 19 mm (11) 1999 Temperature Density Mix Properties Connec- ticut Mahoney et al. (2003) 38 100% passing 25 mm (26) 100% passing 19 mm (12) 2001 to 2003 Temperature Density Mix Properties Florida Sebesta et al. (2013) 1 100% passing 19 mm (1) 2010 Temperature GPR Density Mix Properties (no gradations) Georgia Stroup-Gardiner and Brown (2000) 4 SMA (1) 100% passing 25 mm (1) 100% passing 19 mm (2) 1998 Visual Temperature Density Texture Mix Properties Maine Sebesta et al. (2013) 1 WMA with 100% passing 19 mm (1) 2011 Temperature GPR Density Mix Properties (no gradations) Michigan Wolff et al. (2000) 22 100% passing 19 mm (22) 1998 Visual Density Mix Properties Minnesota Adams et al. (2001) 63 100% passing 12.5 mm (24) 100% passing 9.5 mm (39) 2000 Temperature Density Minnesota Sebesta et al. (2013) 1 WMA with 100% passing 19 mm (1) 2010 Temperature GPR Density Mix Properties Minnesota Stroup-Gardiner and Brown (2000) 2 100% passing 12.5 mm (2) 1998 Visual Density Mix Properties Nebraska Bode (2012) Cho et al. (2010) 18 Information not available 2007 to 2009 Temperature Density Texas Stroup-Gardiner and Brown (2000) 2 100% passing 37.5 mm (2) 1998 Visual Density Mix Properties TABLE 3 SUMMARY OF FIELD PROJECTS INVESTIGATING SEGREGATION (continued on next page)

20 et al. 2005). Each material has its own temperature-dependent dielectric constant. When dielectric constant measurements are used to estimate the pavement density, the gauge readings represent a combined dielectric value for all of the materials. (Note: This is the same material property measured with the GPR technology.) surface air voids results in lower densities being reported. Small air gaps resulting from debris on the bottom of the gauge can also lead to underestimated densities. The PQI uses measurements of dielectric constant to esti- mate the pavement density (McGhee et al. 2003; Flintsch Texas Sebesta and Scullion (2002) 4 100% passing 19 mm (4) 2001 Visual Temperature GPR Density Mix Properties Texas Sebesta et al. (2013) 1 SMA (1) 2009 Temperature GPR Density Mix Properties Virginia McGhee et al. (2003) 8 100% passing 37.5 mm (2) 100% passing 25 mm (2) 100% passing 19 mm (2) 100% passing 12.5 mm (2) 2002 Texture Density Mix Properties Washing- ton Stroup-Gardiner and Brown (2000) 2 100% passing 25 mm (2) 1998 Visual Density Mix Properties *Gradation information: each state uses different mix designations and mix design methods (e.g., Marshall, Superpave). In order to make comparisons between different state studies, the largest sieve size with 100% passing is used to charac terize the gradations. SMA = stone matrix asphalt; WMA = warm mix asphalt; OGFC = open-graded friction course. State References No. Sect. Gradation Information* Year Constructed Detection Method QC/QA Properties TABLE 3 (continued) Non Nuclear Density Gauge (Pavement Quality Indicator (PQI)) ASTM D2950 Nuclear Density Gauge Surface air voids or small gaps between the gauge bottom and surface result in lower density estimates. PQI uses measurements of dielectric constant to estimate density. Because of the high dielectric value of water, any moisture in or on the pavement can significantly influence density readings. FIGURE 26 In-place density testing devices [Source: (left) Dixon (n.d.) Troxler Electronic Laboritories (right) Pavementinteractive.org (http://www.pavementinteractive.org].

21 General dielectric constant ranges for typical asphalt mix materials are: • Aggregates, 2.5 to 5.0 • Asphalt (unmodified or modified), 2.5 to 3.2 • Air, 1 • Water, 80.4 @ 68°F, 55.3 @ 212°F. The low dielectric value for air, compared with those for asphalt binder and aggregates, helps identify changes in den- sity that are linked to changes in air voids. However, because of the very high dielectric constant of water and the sensitivity of the value to temperature, any moisture content in the mix has the potential for significantly influencing the gauge readings. Because the sources, types, and combinations of materials are unique for each property, it is important to determine the dielectric value in non-segregated areas of the pavement so that any significant changes in the proportions of aggregates and air voids can be detected. ➢ If the rough texture in segregated areas is not sanded prior to determining the density with a nuclear gauge, the density readings can be underestimated. ➢ It is important that non-nuclear density gauges be cali- brated relative to readings in non-segregated areas so that an accurate estimate of density can be obtained. Roadway Density Changes and Temperature Differences Various approaches to evaluate density changes in seg- regated areas such as a single point measurement, two density measurements (one in a segregated area and one in an unsegregated area), or density profiles (longitudinal, trans- verse, or skewed) are shown in Table 4. When density profiles are used, density changes over the profile length are evaluated by a single criterion or set of criteria (Figure 27). For example, Washington and Minnesota adopted criteria developed by Kansas using ten density measurements collected for a 50-ft length. The criteria used were: • Density range: The difference between the high- and low-density values. The range needs to be no more than 6.0 lb/ft3. • Density drop: The difference between the average den- sity for the profile minus the lowest density reading in the profile. The acceptable density drop is no more than 3.0 lb/ft3. Minnesota and Washington studies reported that at least 80% of the density profiles met both density range and den- sity drop criteria in areas with temperature differences of less than 25°F. No more than 49% of the profiles met both require- ments when the temperature differences were more than State Density Requirements Results T < 25oF Washington (Willoughby et al. 2001) Density (high)—Density (low) in profile < 6.0 lb/ft3 Density (average)—Density (low) < 3 lb/ft3 80.5% met both criteria Minnesota (Adams et al. 2001) 93.1% met both criteria T > 25oF Washington (Willoughby et al. 2001) Density (high)—Density (low) in profile < 6.0 lb/ft 3 Density (average)—Density (low) < 3 lb/ft3 10.7% met both criteria Minnesota (Adams et al. 2001) 48.6% met both criteria Texas (Rand 2010) Criteria applied at paver stops, when using non- automated temperature measurements, visually identified mix segregation No statistics in reports Connecticut (Mahoney et al. 2003) 3-ft intervals, starting 15 ft in front of cold area and ending 15 ft after cold area Average decrease in density was 5.73 lb/ft3 Connecticut (Henault et al. 2005) No identifiable density differences due to temperature differences Nebraska (Bode 2012) Single point measurements Temperature differences explains at least 59% of changes in densities TABLE 4 EXAMPLE OF DENSITY CHANGES DUE TO TEMPERATURE DIFFERENCES

22 25°F. A Connecticut study reported mixed results for density changes in areas with colder temperatures and the Nebraska research found lower densities were likely when the tempera- ture difference was more than 25°F. ➢ Densities that do not meet the specification require- ments are more likely to be found in areas with tem- perature differences of more than 25°F lower than the surrounding areas. Laboratory Density Testing When segregation is detected after the paving is completed, additional cores can be taken so that densities can be determined with laboratory test methods. When segregation is detected at some point before the mix is compacted, loose mix samples can be collected and compacted in the laboratory before testing. Laboratory density tests that may be specified by agencies include methods using: • Saturated surface dry samples (uncoated; AASHTO T166) • Hot paraffin wax-coated samples (AASHTO T275) • Parafilm™ stretchable wax wrap coated samples (ASTM D1188) • Samples vacuum sealed in a heavy duty plastic bag (AASHTO T331). Samples prepared using typical dense mix that is not seg- regated absorbs only a minimal amount of water into the sur- face voids during density testing (AASHTO T166). Samples with segregated mix are likely to have large surface voids; interconnected air voids that allow water to be absorbed dur- ing testing (Figure 28). When this happens, the density is significantly overestimated and the severity of the segrega- tion is significantly underestimated. ➢ Potentially segregated mix samples are to be coated prior to laboratory density testing to prevent signifi- cantly overestimating the density and underestimating the severity of the segregation. Aggregate Gradations and Asphalt Content Testing Cores or loose mix samples of segregated mix during con- struction are used to determine if gradations and asphalt con- tents are within the specification limits. Aggregates and asphalt are separated using either solvent extraction methods or with an ignition oven. Solvent extraction methods have a longer history. The asphalt is extracted from the mix using one of several approved solvents. The mix and recovered aggregate are measured before and after soaking to determine the asphalt content. When cores are tested, the cut faces on the edges of the core are to be removed before testing so that the after-extraction gradations are not skewed by the percent of cut aggregates. The ignition oven method burns the asphalt off of the mix sample and the difference in the weight before and after burning is recorded as the asphalt content. However, some aggregates also have individual components that can burn off along with the asphalt and calibration information is needed for the specific aggregates used on the project. Gradations determined after burning may also be influenced by the high heat as some aggregates fracture at the higher temperatures. If there is any water trapped in the internal aggregate voids, the water turns to steam as the oven temperature increases and fractures the aggregate. These artificial aggregate size reduc- tions may mask the extent of coarse aggregate segregation. ➢ When the testing is done to determine changes in gra- dations resulting from segregation, the selection of the Segregated mix Weight determined using scale. Dense mix Volume is determined by weighing under water. Dense mix cores absorb very little water during testing. Cores with segregated mix absorb water during testing. The volume used for the density calculations is reduced by the volume of the water absorbed. Result: Calculated density is reasonably accurate. Result: Calculated density is over estimated. Successful laboratory density testing of segregated mix uses samples which are coated on the outside so water can’t be absorbed. FIGURE 27 Explanation of the impact of segregation on laboratory density measurements [Source: Stroup-Gardiner (2014)].

23 test method and/or test equipment may unintentionally underestimate gradation changes. ➢ Laboratory methods for separating the aggregates and asphalt for cores or loose mix can mask the extent of coarse aggregate segregation. It is important to understand the limitations of gradation and asphalt content estimates once the components have been mixed together. ➢ It is important that agency field and laboratory staff understand the impact of test methods when testing segregated mixes. It is important that this information be included in any agency training programs. PERFORMANCE-RELATED TESTING The literature review found various research studies that investigated the impact of segregation on performance-based mix properties such as permeability, rutting, fatigue, tensile strength, and mix stiffness. The findings from these studies are summarized by each material property. • Permeability – Increases quickly when air voids are greater than 8% (Williams et al. 1996). – Predictive equation showed a 300% increase in permeability with every 1% increase in air voids (Willoughby et al. 2001). • Rutting Potential – Severely segregated mix resulted in increases of rut depths from 69% to more than 100% (Williams et al. 1996). – Rut depths increased because of the increase in air voids owing to segregation (Brock and Jakob 1999). • Fatigue Life – Fatigue life decreases substantially with increasing segregation. Moderate levels of segregation can reduce the fatigue life from 20% to 40% and severe segrega- tion can reduce fatigue life from about 80% to 95% (Williams et al. 1996; Brock and Jakob 1999). • Tensile Strength – Tensile strength can be reduced by 20% to 70% when the mix is moderately to severely segregated (Stroup- Gardiner and Brown 2000). – Decreases in tensile strength are a function of increases in air voids in segregated mix (Sebesta et al. 2013). • Mix Stiffness – Resilient or dynamic modulus can be reduced by 30% to 50% when the mix is moderately to severely segre- gated (Stroup-Gardiner and Brown 2000). ➢ Research studies show that moderate to high levels of segregation significantly increase permeability, increase rut depths, decrease fatigue life, and decrease tensile strengths. AASHTO T166 (Uncoated) AASHTO T331 (Vacuum Sealed) ASTM D1188 (ParafilmWrapped) Regardless of coating (or not), samples are weighed under water to determine volume for density calculations. FIGURE 28 Example of methods for coating samples for density testing [Source: Stroup-Gardiner (2014)].

24 HOW AND WHERE MIX SEGREGATES Segregation can start as early in the paving process as aggregate production and can continue through the asphalt mix design phase, mix production, and paving operations (AASHTO Joint Task Force 1997; Cleaver 2012). The fol- lowing sections identify equipment and practices that can reduce segregation at each step of the design and produc- tion of asphalt mix. Aggregate Production Aggregate particle sizes can be separated (segregated) by: • Stockpile construction methods, • Conveyor systems, and/or • Loader operators. Large stockpiles with a range of particle sizes usually have coarser aggregates that roll down the sides of the stockpile and collect around the lower edges (Figure 29). When it is necessary to form conical stockpiles, it is effec- tive to use short drop distances (close enough so aggre- gates do not roll), limited ranges of particle sizes in each stockpile, and adequate separation between the stockpiles to avoid cross-contamination (Figure 30). Sufficient space between stockpiles is desirable but sometimes difficult to achieve because of space limitations, especially in urban environments. Telescoping and/or radial conveyors (Figure 31) that drop the aggregate onto the top of the stockpile from a short dis- tance help maintain uniformity in the aggregate gradation (Quality in California 2011). Layered stockpiles can be constructed with radial stackers that minimize the segregation-prone cone shape. Larger maximum size aggregate gradations benefit the most from the use of radial stackers (Cleaver 2012; Zettler 2012). Aggregate conveyors are sometimes called slingers because they “throw” the material off the end of the belt. Vulcan Materials developed a laboratory demonstration to show how conveyors distribute different particle sizes in the stockpile. Because larger particles weigh more per particle (i.e., contain more mass), they will be “flung” the furthest and the smallest particles with the least mass per particle will tend to just drop off the end of the belt. The colors of the particles are uniformly and randomly dis- tributed on the moving conveyor (Figure 32). Once the particles reach the end of the conveyor, the smaller white particles are concentrated at the edge nearest the conveyor and mostly covered by the larger green particles rolling down from the top of the pile. The orange particles are the FIGURE 31 Telescoping, radial stacker used to build large scale stockpiles [Source: Sam Johnson, Vulcan Materials (n.d.)]. Coarse aggregate around bottom outside edges FIGURE 29 Stockpiles separated to avoid overlap of materials [Source: Scherocman (2011)]. Stockpiles separated to avoid overlap of materials FIGURE 30 Course aggregates collect around the outside bottom edges of the stockpile [Source: Warren (2012)].

25 largest and are thrown around the outer circumference of the pile. Loader operators can help minimize aggregate segrega- tion by blending the fine and coarse areas of the stockpile. Random segregation can be avoided by reworking the stock- pile pile to reblend the coarser aggregate at the bottom with the rest of the aggregate or by filling the loader bucket with non-segregated aggregate in the pile (Scherocman 2011). When stockpiles are built with dozer operations, movement of the equipment over the layers should be minimized so that the heavy equipment does not crush (degrade) the aggregate. This is especially important when working with light weight aggregates that are easier to crush. ➢ Segregation is reduced when aggregate stockpiles are produced with narrow ranges of particle sizes, short drop distances, and skilled loader operators. Mix Designs The segregation potential of mixes can be minimized during the design phase by the appropriate selection of the (AASHTO 1997; Brock et al. 1998): • Aggregate gradation, • Maximum aggregate size, • Asphalt content (asphalt film thickness), and • Recycled materials that contribute asphalt content. The literature review found two potential test methods and one European standard for estimating the segregation poten- tial of mixes during the mix design phase. Aggregate Gradations Aggregate gradations are to be evaluated for segregation poten- tial using all of the Superpave sieve sizes for the gradation analysis. Gradations with the least segregation potential have percentages of aggregates about evenly distributed on each of the sieves (i.e., typical well-graded dense mixes). Gradations with low or no percentages of aggregates on one or more of the standard Superpave sieve sizes are more likely to segre- gate. Gap-graded or open-graded friction course gradations are much more prone to segregation during mixing, truck load out, transportation, or paving processes (Brock et al. 1998; Quality in California 2011; VDOT 2012). The Bailey method for aggregate gradation design can be a useful tool for predicting how changes in the gradation influence aggregate structure (i.e., packing). A well-packed gradation reduces the segregation potential (Marais and Pretorius 2007). ➢ Segregation potential can be reduced by avoiding gaps in the aggregate gradation. If gapped gradations are used, more attention is required during construction to control segregation. – Using the full range of Superpave sieve sizes when constructing the gradation curves can help identify any gaps in the gradation. Maximum Aggregate Size The segregation potential of the mix increases with the maxi- mum size aggregate of the gradation (Brock et al. 1998; VDOT 2012). Larger aggregates are more likely to: • Get “thrown” farther than smaller particles by conveyor systems. Fines (Green, White) Coarse (Orange) Particles (colors) well distributed on conveyor FIGURE 32 Demonstration of how different aggregate sizes are “slung” off conveyor [Source: Sam Johnson, Vulcan Materials (n.d.)].

26 • Roll to the outer edges of the mix in the silos and the edges of the mix in the haul truck bed. • Drop off of the paver hopper conveyors and be deposited under the paver gear box. • Move to the outer edges of the screed augers. Typical dense-graded mixes with a maximum aggregate size of 25 mm are more segregation-prone than 12.5-mm mixes (Mahoney et al. 2003; Buncher and Rosenberg 2012). Wise (2007) recommends that the ratio between the maxi- mum aggregate size and lift thickness be less than 1:2 to help with the compactability of the mix and decrease the potential for mix segregation. Others recommend ratios of 1:3 or 1:4 to help with compaction (USCOE 2001). ➢ Segregation can be reduced by using mixes with maxi- mum aggregate sizes of 12.5 mm or less. ➢ Segregation potential increases with the maximum aggregate size. Asphalt Content (Asphalt Film Thickness) Asphalt content was identified by the AASHTO Joint Task Force (1997) as the single most important mix design cri- terion for segregation potential. Adequate asphalt content is necessary to provide a sufficient film thickness to hold the range of aggregate particles together and provide workability of the mix (Advanced Asphalt Technologies 2011). Low film thicknesses result in mixtures that segregate more easily and are more difficult to place. A small increase of 0.2% in the asphalt content can help reduce segregation in gapped grada- tions (Brock et al. 1998). The potential for segregation can be reduced by avoiding the following: • Fluctuations in the aggregate gradations in general and dust in particular: – Decreasing the finer particles decreases the aggregate surface area that needs to be, or can be, coated. • Aggregates with high moisture capacities: – Any retained moisture makes it difficult to uniformly coat the aggregates. Absorptive aggregates tend to hold moisture that may not be adequately removed during mixing. – Absorptive aggregates also absorb more asphalt binder that can reduce the film thickness over extended hot storage times (i.e., time in silo). ➢ Segregation potential is reduced when there is suffi- cient asphalt film thickness to hold the different aggre- gate sizes together. Recycled Materials More additives and recycled materials are added to today’s mixes so that there is greater potential for additional forms of material segregation. Individual materials are to be added at the correct time and point in the mixing process so that they are fully blended with all of the other materials. Flexibil- ity in the point-of-introduction is important so that different materials can be added for optimum distribution throughout the mix and long mixing zones provide adequate mixing time (Cleaver 2012). Proper handling, preparation, and sampling of asphalt- containing recycled stockpiles is particularly important because of the potential for wide ranges of material prop- erties in recycled materials (Figure 33). When recycled asphalt pavement (RAP) is separated into coarse and fine fractions, the asphalt content of the coarse fraction can be the most variable (Valdes et al. 2011). It is likely this will be a more significant factor in mix uniformity as agencies increase the allowable percentages of recycled materials in asphalt mixes. Correct sampling procedures of RAP and recycled asphalt shingles (RAS) stockpiles are especially important because Grinding Partially Processed Recycled Asphalt Shingles Ready to Use FIGURE 33 Several steps may be needed to process recycled material so a uniform, consistent size is obtained [Source: National Asphalt Pavement Association (n.d.)].

27 improper sampling can lead to incorrectly calculating the asphalt content in the RAP (Advanced Asphalt Technolo- gies 2011). Variability in the asphalt content of the recy- cled materials leads to variability in the asphalt content and asphalt film thickness. This makes it difficult to control the segregation potential of the mix. ➢ Segregation potential can be reduced with a sufficient and consistent asphalt film thickness. – Uniform preparation and proper sampling of recy- cled material stockpiles is required to ensure that all of the individual components are uniformly distrib- uted throughout the mix. – Any component added to the mix that increases the variability in asphalt content has the potential for pro- ducing mixes that may vary between over- or under- asphalted mixes (i.e., variability in film thickness), which makes it difficult to control segregation. – Variability in gradations has the potential for creat- ing more of a gap gradation that can be more prone to mix segregation. Determining Segregation Potential During Mix Designs Three methods for estimating the segregation potential dur- ing mix design were found in the literature (Murphy 2012a, b; Feng et al. 2013). All of the methods evaluate how hot, freshly blended asphalt mix segregates as it is dropped from a given height (Figures 34 and 35). A sample report of results from an asphalt mix known to segregate showed a significant change in the 4.75-mm (No. 4) sieve, asphalt content, and air voids (Table 5). The European Standard BS EN 12697-15 2003 Hot Mix Asphalt—Segregation Sensitivity describes a test method that can be used during mix design, which is similar to the two concepts described previously. ➢ It may be possible to estimate the segregation potential of a mix during the design phase of a project. Asphalt Plants Cold Feed Bins Cold feed bins are a source of aggregate segregation in both batch and drum mix plants. The skill of the loader opera- tor is a major factor in minimizing any potential segregation (Brock et al. 1998; Scherocman 2011; Cleaver 2012). The loader operator is responsible for: • Keeping the cold feed bins as full as possible. Bins that are allowed to get too low form a reverse cone, which allows the coarser particles to roll down the slope and onto the cold feed belt (Figure 36). • Avoiding scooping the material under the stockpile along with a load from the stockpile to avoid contamination. • Carefully reblending any coarser particles around the lower portions of the stockpiles with upper portions. Keep reworking of the stockpile to a minimum to prevent degradation of the particles. • Filling the loader bucket with non-segregated aggregate taken from several feet above the ground level. Each cold feed bin is loaded from a single stockpile; that is, when an asphalt plant only has three cold feed bins, the 1 meter (about 3 ft) Material inside the bucket diameter is tested and compared to the results for the material outside of the bucket diameter FIGURE 34 Laboratory test for estimating segregation potential [Source: Murphy (2012a)].

28 80 cm (31 in) 25 cm (10 in) 40 cm (15.5 in) Storage Hopper 30 cm (12 in) Each Side Mix Hopper Valve for opening hopper gates FIGURE 35 Feng’s concept for a laboratory test for evaluating the segregation potential of asphalt mixes [Source: Feng et al. (2013)]. Property Mix Inside Bucket Diameter Mix Outside Bucket Diameter Difference 12.5 mm (½ in.) 100% passing 100% passing 0% 4.75 mm (No. 4) 59% 45% 14% (coarser) Asphalt Content 5.8% 5.5% 0.3% (lower) Air Voids (Marshall Compaction) 3.6% 5.2% 1.6% (higher) Source: Murphy (2012a). TABLE 5 EXAMPLE OF RESULTS FOR A LABORATORY TEST TO ESTIMATE THE POTENTIAL OF A GIVEN MIX TO SEGREGATE FIGURE 36 Reverse “cone” formed in center of cold feed bin which has been run too low [Source: Stroup- Gardiner (2014)]. Reverse Cone Formed

29 plant only needs to keep three stockpiles on site. The limited number of stockpiles implies that there is a wider range of particle sizes in each stockpile, which makes it easier for the coarse and fine aggregates to segregate. Asphalt plants with a larger number of cold feed bins allows for better control of segregation in the stockpiles. ➢ Segregation can be reduced by increasing the number of cold feed bins used to produce the mix. ➢ Well-trained and skilled loader operators responsible for loading the cold feed bins are important for reduc- ing segregation at the beginning of mix production. Batch Plants Segregation occurs most frequently in a batch plant in the hot aggregate bins (Figure 37). Segregation is generated when the aggregate is loaded from the screens into the #1 bin, which receives the fine aggregate fraction (4.75-mm to 0.075-mm sizes) (AASHTO 1997; Brock et al. 1998; Figure 38). The dust tends to collect on the sloping bin wall nearest the aggre- gate feed drop, while the larger particles end up collecting on the opposite side. The fines pack together then drop all at once when the bin is almost empty. This produces mix with a high percentage of fines that are typically uncoated. The buildup of fines can be minimized by vibrating the side of the bin to keep the fines from accumulating in one spot. ➢ The No. 1 hot aggregate bin in a batch plant is the source of most segregation because of the buildup of dust or coarse aggregates on opposite sides of the bin. ➢ Segregation can be reduced by vibrating the sides of the bin to prevent buildup. Drum Plants Segregation can be generated at six locations during mix pro- duction (Figure 39): 1. Drum 2. Exit from the drum 3. Drag slat conveyor 4. Bin charging batcher 5. Silos 6. Load out. FIGURE 37 Components of a batch plant [Source: Stroup- Gardiner (2014)]. Hot Screens Hot Bins Bucket Elevator from Aggregate Cold Bins Batch Plant Mixer FIGURE 38 Schematic of aggregate bins in a batch plant [Source: Brock et al. (2003); Stroup-Gardiner after Murphy (2012a)]. Bin #1 Dust accumulates along this side Coarser portion of fines accumulates along this side 4.75 mm to dust Screen Decks Cold Aggregate Feed Bin #2 Bin #3 Bin #4 Aggregate segregation happens when the build up of either dust or coarse aggregate gets too large and drops all at once into the mixer

30 A more recent concern for mix segregation is in how and where materials other than aggregates and asphalt are added into the mix. RAP has been used in mix for years; however, the recent trend is to increase the amount allowed in the mix as well as comingling RAP with RAS. New additives and pro- cesses for producing warm mix asphalt may also influence the segregation potential of the mix. In some cases, new additives may reduce segregation, or in others, increase segregation. No research was found in the literature that evaluated these factors in segregation. Drum Configurations An inadequate asphalt film thickness can increase the segregation potential of the mix. Insufficient or non-uniform asphalt film thicknesses can be the result of short mixing times or damp aggregates. Of the two general types of drum mix plants, parallel flow and counter flow (Figure 40), the parallel flow has less time for the aggregates to dry and blend in the drum before the asphalt is introduced. Counter flow drum mix plants are better at mixing the aggregate gradation in the aggregate drying portion of the drum length. Mixing and drying times can be increased by installing kick-back flights or dams inside the drum and/or decreasing the drum slope to slow the movement of the mix through the drum. However, decreasing the slope can decrease the produc- tion rate if the drum drive motor is too small. The preferred options for minimizing segregation during mix production are options that do not impact the production rates. FIGURE 39 Locations in a drum mix plant where segregation can occur [Source: Stroup- Gardiner (2014)]. Aggregate from Cold Feed Bins Drag Slat Conveyor Transverse Conveyor Truck Load Out Drum Silos Point of Entry to Drum Batcher FIGURE 40 Schematics of parallel flow and counter flow drums [Source: Stroup-Gardiner after Brock et al. (2003)]. RAP Collar RAP Collar Counter Flow Drum Mix Plant Aggregate Enters at Burner End Aggregate Flow Towards Burner End

31 The uniformity of the film thickness also is improved by limiting the amount of the dust (0.075 mm) material to the lower specification limit. This reduces the aggregate surface area, which is to be coated during mixing. ➢ Segregation is reduced by: – Kick-back flights or dams inside of the drum to help increase the aggregate drying time, which helps pro- mote uniform asphalt coating. – Decreasing the amount of fines to the lower specifi- cation limit to help improve the uniformity of the film thickness by reducing the aggregate surface that is to be covered with asphalt. Point of Discharge Segregation can occur in the drum and result as segregated mix if the configuration or features of the point of discharge fail to help reblend the mix (AASHTO 1997; Brock et al. 1998). Coarser aggregates discharge to one side, while the finer aggregates collect on the opposite side at the end of the drum. Segregation can be minimized by: • Setting the drag slat conveyor at 90° to the end of the drum (Figure 41). • Restricting the mix as it leaves the drum so the mix is forced into the center of the conveyor (Figure 42). ➢ Segregation can be reduced as the mix exits the drum by orienting the drag slat conveyor at 90° to the drum exit or restricting the mix as it exits the drum onto the drag slat. Drag Slat Conveyor Segregation can happen on the drag slat conveyor when there is too much or too little mix on the conveyor. When there is an excess of mix, coarser particles roll backward over the top of the slat and down to the next conveyor section (hydroplaning; Brock et al. 1998). Too little mix allows coarser particles to roll to the outsides of each conveyor section. Keeping the drag conveyor full, but not overly full, helps reduce segregation at this point in the process. ➢ Segregation can be reduced by keeping the drag con- veyor full, but not overly full. Bin Loading Batchers Segregation at the asphalt plant is a function of the surge and storage bin characteristics (Astec 2010; Scherocman 2011). A bin loading batcher, also referred to as a gob hopper, is one of the most common choices for eliminating segregation in the silos. The batcher, which typically holds about three tons of mix, collects a batch of mix from the drag slat conveyor at the top of the silo and drops it as a single large mass into the silo. The impact of the mix landing in the silo distributes the mix uniformly over the surface of the mix in the silo. Like all construction equipment, the bin loading batcher needs to be operated properly or it will promote rather than reduce segregation. Segregation can occur when the loading chute to the batcher is not positioned over the center of the silo, the batcher is not completely emptied, or the batcher gates are too slow in opening and closing. It is important that the chute be positioned in the center of the batcher; other- wise, the coarser particles will roll to the lower, outer edges of the mix (Figure 43). Insufficient emptying of the batcher lets coarser particles build up along the outside edges over time (Brock et al. 1998). The batcher gates are to be opened Drag Slat Turned 90o to Drum Discharge Uniform Mix Coarse Fine Drum Rotation Direction of Mix Drum discharge at 90o to drag slat helps reblend mix when it drops onto conveyor. FIGURE 41 Locating the drag slat conveyor at 90° to the drum discharge helps reduce segregation [Source: Stroup-Gardiner after Murphy (2012a)].

32 quickly so the mix is dropped into the silo in a single mass. It is also important that the gates be closed quickly so the mix does not slowly dribble into the silo. Newer storage bins equipped with batchers and steeper- sided cone-shaped bottoms can be completely emptied with little or no segregating of asphalt mixes. ➢ Batchers reduce segregation when the chute is centered over the batcher, filled before dropping the mix, and the mix is dropped into the silo all at once. Silos Segregation can be reduced by keeping an optimum amount of mix in the silo and by proper operation of the gates at the bottom of the silo (Figure 44). When the level of the mix is too low, any coarser particles at the outer edges collect at the bottom of the silo and are dropped all together into the next truck load. When the silo is too full there is not a sufficient drop height to flatten out the mix and a cone of mix is formed at the top of the silo that allows the coarser particles to roll down the outer edges of the silo. Proper centering of chute over center of batcher Minimizes Segregation Chute not properly aligned over center of silo Causes Segregation Coarse Aggregate Fine Aggregate FIGURE 43 Proper use of bin charging batcher over silo [Source: Stroup-Gardiner after Murphy (2012a)]. Fixed Plow Close up of Fixed Plow Direction of Mix Drum Rotation Fixed plow at drum discharge forces the mix to reblend as it exits. Fine Coarse FIGURE 42 Fixed plow at the discharge helps reduce mix segregation [Source: Stroup- Gardiner after Murphy (2012a)].

33 ➢ Segregation is reduced when: – The level of mix in the silo is kept between 25% and 75% full. – Silo gates are quickly opened and closed when load- ing the haul trucks. Load Out Segregation occurs when the haul truck is loaded with a large single drop of mix, because this allows the coarser particles to roll to the outside edges of the truck bed (AASHTO 1997; Brock et al. 1998; Cleaver 2012). Proper loading of a 10 to 12 cubic yard capacity end dump haul truck can be accom- plished in three drops (Figure 45), with the front third of the truck loaded first, the back loaded second, and the last drop into the center (AASHTO 1997; Scherocman 2011). Larger end dump haul trucks (up to 35 cubic yards), belly dumps (18 to 20 cubic yards), and live bottom trucks (31 to 50 cubic yards) also will be loaded using multiple individual drops (Figure 46). It is important that continual loading of the truck while it slowly moves forward be avoided. ➢ Load Out: – Segregation is reduced when haul trucks are loaded in multiple drops. – Avoid loading with a large single drop or continu- ously as the truck moves forward. Mix Transfer from Haul Truck to Paver Transferring the mix from the haul truck to the paver hopper can be a significant source of segregation. The type of trucks FIGURE 44 Schematic of storage silos with different volumes of mix [Source: Brock et al. (2003); Stroup-Gardiner after Murphy (2012a)]. 75% 25% Optimum Asphalt Mix Level in Silo Too Low Will segregate when mix is added to silo Too Full Mix segregates with coarser particles to the outside of the silo Just Right Mix in silo kept between 25% to 75% full to minimize segregation Coarse Aggregate Fine Aggregate FIGURE 45 Loading with multiple drops [Source: Scherocman (2011)]. 1st Drop towards cab 2nd Drop towards tailgate 3rd Drop in center

34 used to haul the mix determines how the mix is transferred to the paver. End Dump Truck Mix Transfer Segregation can be reduced when the end dump haul truck bed is raised before opening the back gate. When the gate is finally opened, the mix is discharged from the truck bed in mass into the paver hopper (FHWA 2002; Brock et al. 1998). This process rapidly fills the paver hopper and prevents the coarser particles from collecting in the paver wings. The haul truck should not “bump” the paver as it gets ready to transfer the mix, but rather wait for the paver to make contact. The paver is equipped to push the truck forward so the transfer from the truck to the hopper is smooth and con- tinuous (Figure 47). Any segregation in the haul truck can be minimized by reblending the mix as it exits the bed. The end dump bed can be fitted with baffles that form a funnel at the back (Figure 48). This funnel forces any potentially segregated mix along the outside of the truck bed to reblend with the mix in the center when the mix is transferred into the paver hopper. A baffle around the hydraulic lift at the front of the bed (not pictured) can also help keep the mix from segregating as it flows around lift box. Spills can happen when the mix is dribbled out of the end dump gate either at the start of transferring the load or when “shaking” the last of the mix out of the truck bed (Figure 49). Spills can also happen when the hopper wings are raised (i.e., flipped), and if not removed a segregated area of mix and bumps in the pavement surface can be produced (Scherocman 2011). ➢ Segregation is reduced when the – Truck bed is lifted before opening the gate so the mix is rapidly transferred to the paver hopper. – Baffles that form a funnel at the back of the truck bed are added to the back of the truck bed to help reblend the mix as it exits. – Any spills are cleaned up. Belly Dump Truck Mix Transfer by Windrow Belly dump haul trucks deposit mix onto the roadway from the bottom of the truck bed and a pick-up unit (i.e., windrow elevator) is used to move the mix in the windrow into the paver hopper (Figure 50). The simplest pick-up devices use a rotating drum with paddles to push the mix onto a conveyor FIGURE 46 Loading large haul trucks with multiple drops [Source: Warren (2012)]. FIGURE 47 Transferring asphalt mix to the paver hopper [Source: Stroup-Gardiner (2014)]. Haul truck stops and waits for paver to start pushing the truck Mix moves from truck bed to hopper once paver starts to push truck forward FIGURE 48 End dump bed fitted with a funnel at the back to help reblend asphalt mix [Source: Murphy (2012a); Stroup-Gardiner (2014)]. Baffles which create a funnel at the tailgate helps reblendmix as it exits.

35 that deposits the mix into the paver hopper (Figure 51). Other designs include an auger on either side of the rotating pad- dles to help reblend the mix and move it toward the paddles (Gilbert 2005). As with all construction processes, when effective con- struction practices are not followed, a segregated mix placed in a windrow is likely to remain segregated when placed in the paver hopper (Figure 52). The segregated mix visible in the windrow likely occurred at the asphalt plant somewhere between the mixer discharge point and loading of the haul truck. The amount of mix in the windrow needs to be sufficient to keep the paver hopper full and the paver moving. The con- sistency of the mix in the windrow also is to be homogeneous; clumps and areas of cold mix are to be avoided (Figure 53). ➢ Segregation in the windrow can be reduced when the windrow elevator (pick-up unit) helps reblend the mix. – Mix in the windrow is fairly uniform.  Avoid mix that is obviously segregated, contains clumps of mix, or is “dribbled” out of the haul truck. – Sufficient mix is necessary in the windrow to keep a constant supply of mix to the screed augers. ➢ The effectiveness of paddles only and augers with paddles for reducing segregation could be explored in future studies. Live Bottom Truck Transfer Segregation can be reduced by the funnel-shaped inside of a live bottom haul truck. The mix slides down the sides and a conveyor at the bottom of the truck bed moves the mix horizontally out the back (Figure 54). Live bottom trucks can deposit the mix directly into the paver hopper or can place the mix in a windrow. If left, spills create a bump in lane and leave mix to get cold. Cold mix can result in low densities. FIGURE 49 Temperature segregation can result from mix spilled in front of the paver and not cleaned up [Source: Stroup-Gardiner (2014)]. Gates on Bottom Belly dumps drop mix when the gates at the bottom of the truck bed deposit mix into a windrow on the exiting surface FIGURE 50 Windrow formed as belly dump deposits mix [Source: Pavementinteractive.org (http://www.pavementinteractive.org)].

36 Augers move mix from the outside of the windrow towards the paddles. The paddles rotate downward and push the mix onto the internal conveyor which deposits the mix into the paver hopper. Paddles Auger Auger Pick-updevices may have only paddles to move the mix onto the internal conveyor or have a combination of augers and paddles. FIGURE 51 How windrow elevators (pick-up devices) work [Source: Stroup-Gardiner (2014)]. Coarse aggregate segregation deposited in the windrow shows up in paver hopper. Windrow elevators can’t completely reblend segregated mix. FIGURE 52 Segregated mix in the windrow can still remain segregated when it gets to the paver hopper [Source: Stroup-Gardiner (2014)].

37 ➢ Live bottom truck bed designs help reduce segrega- tion. The funnel-shaped sides helps reblend the mix as it is pulled down and moved out the back of the bed. Material Transfer Vehicles and Devices Material transfer units reduce segregation by reblending mix from the haul trucks before it is transferred to the paver. Another benefit is the ability to hold larger quantities of mix, which helps minimize paver stoppages and keeps the screed augers adequately supplied with mix (i.e., prevents “starving” the augers). The two general types of material transfer units are MTV and material transfer devices (MTD), although the terms are frequently interchanged. An MTV is a self-powered unit that is designed to receive mix from more than one haul truck. Mix is transferred from the haul truck into a surge bin with remixing augers at the bottom. As the mix is reblended, it is transferred to a surge bin that sits in the paver hopper (Figure 55). Mix “dribbled” out of haul truck and mix clumps are visible in the windrow (circled areas). Infrared image shows areas of cold mix. All of these non uniformities will result in variations in density and ride quality of the finishedmat. FIGURE 53 Only a limited amount of mix is in the windrow; cold mix and clumps of material can be seen [Source: Adams (2001)]. *<125.0°C 125.1 136.3 146.0 154.8 163.3 171.6 180.0 188.7 196.7 203.6 209.5 214.7 219.3 223.5 A conveyor belt in the bottom of the truck bed to move the mix out of the back of the truck. Mix can be transferred to the paver hopper or deposited into a windrow. FIGURE 54 Live bottom haul trucks move mix horizontally and out the back of the truck bed [Sources: ABS Trailers (http://www.abstrailers.com) (upper left) and Pavementinteractive.org (http://www.pavementinteractive.org) (below)].

38 An MTD is an attachment to the paver designed to receive and hold larger loads of mix in a hopper, then convey it up and into a surge bin in the paver hopper (Figure 56). The surge bin in the paver hopper has augers in the bottom for reblending the mix as it moves toward the back of the paver. For the greatest reduction in segregation to be achieved, it is important that material transfer units be operated properly. These units would have: • Surge bins sized appropriately for the size of the paver hopper. • Sufficient mix supplied to the job site so that the amount of mix in the surge bin keeps the paver operating without stopping. • Sufficient mix supplied to the screed augers. ➢ Material transfer units are very effective in reducing segregation when they: – Keep sufficient mix supplied to the screed augers. – Have enough mix to keep the paving operations moving at a continuous speed without stopping. Paver Paver Hopper Mix that is still segregated by the time it is transferred to the paver hopper will show up as segregated mix behind the paver. Any coarser mix along the edges of the end dump haul truck bed moves into the hopper and is deposited at the outer edges of the paver hopper (Figure 57). If the paver hopper is run too low before the wings are flipped, a concentrated quantity of segregated or cold mix is dropped directly onto the hopper conveyors. The segregated mix is conveyed to the screed augers as the paver moves forward and segregated mix starts to show up in the mat at the center of the screed. The end-of-truck chevron (“V” shaped) coarse texture pat- tern is created as the segregated mix moves along the screed auger to the outside edges of the lane. Any segregation that is deposited in the paver hopper can be reduced by raising (flipping) the wings when the hopper is at least half full (Figure 58). This helps reblend any segregated mix in the wings with more uniform mix in the center of the Mix is loaded into hopper on MTV Conveyor moves mix to remixer After remixing, mix is conveyed to surge bin in paver hopper Surge bin in paver hopper FIGURE 55 Example of remixer as a separate unit [Source: Stroup-Gardiner (2014)]. Haul truck Mixing augers in the surge bin FIGURE 56 Example of a remixer in the surge bin in paver hopper [Source: Stroup-Gardiner (2014)].

39 hopper as the mix is pulled down and to the back of the paver. The general understanding is that flipping the wings is a bad practice, which is generally true because most paver operators run the hopper almost empty before changing out haul trucks. When haul trucks are changed while the paver hopper is at least half full, any cold mix at the outer edges can be blended with a larger volume of mix before moving to the screed. As with all construction processes, the use of effective practices significantly reduces segregation. Longitudinal segregation can be produced by equipment that moves the mix from the hopper to the screed augers. Older pavers move mix with drag slat conveyors, which allow coarser mix particles to roll off of the edges of the conveyors. In some cases, the aggregate can be fractured by the convey- ors that can also leave a coarse texture, but not necessarily a segregated, longitudinal streak (ForConstructionPros.com 2012). Newer paver designs replace conveyor motors with outboard motors that allow the conveyors to be placed closer together, and the motors can be adjusted to balance the vol- ume of mix delivered to each side of the screed (Figure 59). It is especially important to be able to balance the volume of mix when screeds are extended to a different width on each side. Although the reduction in longitudinal segregation is consid- ered a benefit when outboard motors are used, no research was found that formally documents this reduction in segregation. Segregation may be reduced by newer paver designs that replace the traditional conveyors with a pair of counter- rotating augers (Figure 60). Counter-rotating augers pull the mix from the entire hopper and the variable auger pitch aids in reblending the mix as it moves toward the screed augers. This design helps reduce aggregate fracturing that can hap- pen with conveyors. However, these benefits have not yet been documented. ➢ Segregation is minimized when the hopper is kept at least half full. ➢ Longitudinal segregation may be minimized with the use of outboard motors to move hopper conveyors or by replacing the conveyors with a pair of twin augers. However, these benefits could be explored in future research studies. Hopper Haul Truck Screed Haul Truck Hopper Wings Flipped End of Truck Segregation FIGURE 57 End-of-truck segregation: How segregated mix moves from the outer edges of the truck bed to the pavement [Source: Stroup-Gardiner (2014)]. Hopper Half Full When the hopper is at least half full, the screed augers are not starved for mix. If the paver wings are flipped with the hopper at least half full, any cold or segregated mix is blended with enough uniform mix so segregation behind the paver is minimized. Hopper Almost Empty FIGURE 58 Examples of hopper half full (optimum lowest level) and a too empty hopper [Source: Murphy (2012b); Warren (2013)].

40 Paver Screed Longitudinal segregation on one or both sides of center occurs when the screed augers are operated either too slowly or too fast (Brock et al. 1998; FHWA 2002; Cleaver 2012). If the augers rotate too slowly, the larger aggregate sizes can drop off; running the augers too fast results in too little mix, and coarser mix gets delivered to the outside edges of the screed (USCOE 2001). The most effective option for the screed auger speed is moderate and continuous. Some newer designs pro- vide options for reversing augers to force material under the housing or relying on outboard drive motors for the auger drives (Cleaver 2012). Centerline mix segregation can occur under the auger chain drive. Kicker paddles can be installed next to the gear box to push the mix under the auger gear box (Figure 61). In some cases, the mix is gravity fed into the augers instead of using the conveyors, which may also help minimize segregation (Brock et al. 1998). ➢ Centerline segregation is reduced when kicker paddles or reverse flow options are used to push mix under the gear box. Longitudinal segregation on one or both sides or either side of the centerline can also be caused by screed and auger extensions. The extensions are to be adjusted so that the same amount of mix is pulled from the center to the edges (Cleaver 2012). The depth of mix across the screed augers would cover about 75% of the augers across the entire width of paving (Figure 62). Auger extensions are to be used when screed extensions are used. Extending the augers ensures that the mix is pulled from the center to the outside edge of paving. If auger extensions are not used, the mix has to be pushed into the extension areas that segregate the mix in the extension areas. If a large amount of mix is allowed to accumulate in front of the screed and beyond the end of the augers, the mix gets cold (temperature segregation). Options for variable flow help balance the amount of mix across the screed when the screed extensions are not evenly Independent outboard motors drive the conveyors which allows a balanced flow of mix to each side of the screed. Outboard motor placement also allows for a closer spacing between the conveyors. FIGURE 59 Example of outboard mounted motors and variable flow options for balancing mix to the screed augers [Source: Stroup-Gardiner (2014)]. Pair of twin augers help pull mix from entire hopper and reblend mix as it is moved towards the screed. Conventional drag slat configuration can fracture aggregate which shows up behind the screed as a coarser texture. FIGURE 60 Examples of different methods of moving mix from hopper [Source: (left) Murphy (2012a); (right) Bateman (2009)].

41 extended on both sides of the screed. Longitudinal segregation on one or both sides of the centerline occurs when the augers are “starved” for mix. A constant head of mix keeps the angle of the screed (i.e., angle of attack) constant, so a uniformly textured non-segregated mix is placed by the paver. Keeping the paver hopper at least half full helps maintain a sufficient amount of mix in front of the screed (i.e., head of mix). ➢ Longitudinal segregation (either or both sides) is reduced when: – A balanced volume of mix is supplied to each side of the screed. – Auger extensions are used with screed extensions so the mix is pulled, rather than pushed, into the exten- sion areas. When a uniform amount of mix is not supplied consistently across the full width of the screed, the angle of the screed plate (i.e., angle of attack) varies, which results in varying mat thick- nesses. If the mat thickness is too thin, the screed drags larger aggregate particles and tears the mat surface (Figure 63). The coarse texture left by dragging aggregates may be interpreted as segregation; however, in reality, it is only a change in the surface texture. ➢ Longitudinal coarse textured areas may be the result of the screed dragging aggregates over the top of the mat. – Keep the angle of attack consistent and uniform to prevent tearing the mat. PAVEMENT DISTRESSES AND PAVEMENT CONDITION IN SEGREGATED AREAS Two studies reported on the typical pavement distresses observed in segregated areas of paving projects from 1.5 to 10 years old; raveling and potholes in various stages of Kicker paddles, placed next to the gear box, pushes mix in the opposite direction of the augers. FIGURE 61 Kicker paddles added to the screed augers help move mix under the gear box [Source: Combined from Murphy (2012a)]. Extensions need to be supplied with sufficient mix all the way to the end of the extension. “Starving” the screed augers and extensions tends to segregate the mix. Head of Material Optimum amount ofmix should cover about 75% of the height of the augers across the full width of the screed FIGURE 62 Example of depth of mix at auger [Source: Stroup-Gardiner (2014)].

42 formation (Table 6). On older pavements, the raveling and potholes had evolved into longitudinal cracking, with more severe raveling around the edges of the cracking (Figure 64). Top-down Cracking Three forensic studies linked intermittent longitudinal top- down cracking very early in pavement life to areas with longi- tudinal segregation (Abu-Hassan 2002; De Freitas et al. 2005; Harmelink et al. 2008). Fifty percent of the cores taken from 28 Colorado projects had gradations that were at least 4% coarser on two sieve sizes than gradations from uncracked areas (Harmelink et al. 2008). Harmelink et al. (2008) compared the cracking patterns with the configurations of the pavers used for each project (Figure 65). The longitudinal cracking occurred along the outside edges of the two slat conveyors and under the gearbox between the conveyors. Discussions between the Colorado Department of Transportation (DOT) and the paver manufac- turers resulted in the development of a retrofit system of chains and deflectors to prevent the coarser fractions of the mix from dropping off the outer edges of the conveyors or getting con- centrated under the gear box. ➢ Top-down longitudinal cracking can occur in locations corresponding with the gear box (longitudinal center- line segregation) and with the outside edges of the paver conveyors (longitudinal cracking on one or both sides of the centerline). Reflective Cracking Several studies documented temperature differences during the construction of a number of overlay projects and periodi- cally monitored the pavement condition over 1.5 to 5 years Texture from Screed FIGURE 63 Example of large aggregates leaving tears in mat from being dragged by screed [Source: (top) Warren (2012); (bottom) Stroup-Gardiner (2014)]. TABLE 6 SUMMARY OF PAVEMENT DISTRESSES IN SEGREGATED AREAS REPORTED IN LITERATURE Type of Pavement Distress Stroup-Gardiner and Brown (2000) Bode (2012) 12 Projects/5 States 4 to 10 Years Old 18 Projects/1 State 1.5 to 2 Years Old Raveling 100% 32% Potholes Starting to Form 57% 37% Intermittent Longitudinal Cracking 25% — Intermittent Rutting 20% (1 state) — Transverse Cracking — 25% Reflective Cracking — — Intersection of Multiple Cracks — 7% — = no information provided.

43 (Henault et al. 2005; Gunter 2012; Sebesta and Scullion 2012). The most obvious pavement distress seen on these projects was reflective cracking, which is a form of cracking that is generated by cracks in the original surface propagating up to the surface of the overlay. Other types of distresses appeared to be starting; however, the researchers believed the surfaces were not old enough for the distresses to be significant. ➢ Regardless of segregation, reflective cracking from the underlying layers can be the first pavement distresses seen within the first five years of traffic. Ride Quality Three studies evaluated the impact of temperature segrega- tion on ride quality. A Texas demonstration project in 1999 determined the ability of various MTVs to minimize visibly detectable segregation and improve the ride quality of the sec- tions (Tahmoressi et al. 1999). The profilograph and inertial profiler units showed similar trends in the ride quality for each test section. However, the average number of visibly identifi- able segregated areas did not correlate with the ride quality. The Present Serviceability Index, a calculation of the pave- ment condition, was not correlated with the segregation seen in the test sections. Four projects (eight test sections) constructed with and without MTVs in Alabama in 2001 showed that the Inter- national Roughness Index (IRI, a measure of ride quality) was FIGURE 64 Example of how pavement distresses in segregated areas of the pavement progress with exposure to traffic and environmental conditions [Source: Stroup-Gardiner (2014)]. Raveling in wheel paths Longitudinal cracking in raveled areas; more raveling along length of cracking. Pavement less than 5 years old Distresses occurring at about 150 ft intervals FIGURE 65 Overlay of paver conveyor slat location and the occurrence of longitudinal cracking in the center of lane and on either side of the centerline [Source: Stroup-Gardiner after Harmelink et al. (2008)]. Hopper ScreedDirection of Paving

44 lower (smoother) by 18 in./mi when an MTV was used and no temperature segregation was documented (Harris 2002). The ride quality in the extended screed area tended to be rougher than the ride in the right wheel path for four of the seven test sections. Five Connecticut projects also had lower IRI values in areas without temperature segregation (Nener- Plante and Zofka 2009). ➢ Ride quality can be significantly improved on sections of pavement with no temperature segregation. SPECIFICATIONS Alberta, Canada Gavin and Heath (2002) reported that a Segregation Tri-Party Task Group was formed to adjust Alberta, Canada, Standard Specification 3.50, Asphalt Concrete Pavement-EPS. The main revisions to the existing edition included: • Identifying segregation quickly so the contractor can mitigate the problem to prevent further segregation. • Eliminating the requirement for the repair of “slight” levels of segregation. • Repairing “moderate” to “severe” segregation only repaired during construction. This allows repairs to be made while the asphalt plant and contractor is still on site. • Reinstating the opportunity for substantial bonuses. Pen- alties were increased for moderate-to-severe segregation and penalties for slight segregation and center-of-paver segregation were added. • The contractor provides written documentation of daily changes or modifications to equipment and operations if segregation is identified at the beginning of a project, which is expected to eliminate or minimize any occur- rence of segregation. Section 3.50.4.7.2—Classifying Pavement Segregation defines segregation as an area of pavement with visually identified texture differences that are more than 0.1 m2 or are center-of-paver streaks longer than 1 m (Table 7). The specific definitions of the levels of segregation are: • Slight segregation: The matrix, asphalt cement, and fine aggregate are in place between the coarse aggregate. How ever, there is more stone in comparison with the surrounding acceptable mix. • Moderate segregation: significantly more stone than the surrounding mix; moderately segregated areas usu- ally exhibit a lack of a surrounding matrix. • Severe segregation: Appears as an area of very stony mix, stone against stone, with very little or no matrix. • Center-of-paver: Appears as a continuous or semi- continuous longitudinal “streak” typically located in the middle of the paver mat. The contractor is required to perform a daily inspection for segregation of all lifts paved. If segregation is seen, the con- tractor takes immediate corrective actions. A consultant(s) per- forms inspections during construction of all lifts. If segregation is discovered, the consultant immediately requests that the contractor take corrective action. The consultant(s) conducts a second inspection following construction, normally within 2 weeks after completion of paving, and a written assessment of the location and severity of segregation is submitted. Slight segregation in any lift does not require repairs. Mod- erate segregation in lower lifts does not require repair and severe segregation in the lower lifts requires repairs only when the consultant believes the segregation will reduce the life of the wear course. Assessed penalties are based on the total num- ber of areas with each level of segregation and only moderate, severe, or center-of-paver segregation on the top lift requires repairs. • Moderate segregation can be repaired using slurry or a hot mix patch. • Severe segregation can be repaired by removing and replacing or with an overlay. • Spray seals, applied either with a distributor, by hand, by squeegeeing, etc., is not an acceptable repair method. TABLE 7 INCENTIVES/DISINCENTIVES FOR SEGREGATION IN ALBERTA, CANADA, SPECIFICATION Level of Segregation Frequency of Segregation Incentive/Disincentive Slight 0 Bonus of $1,000 per lane-km only if also no segregation of any type 1 or 2 Bonus pay of $500 per lane-km only if also no moderate, severe, or center-of-pavement segregation More than 2 Reduce pay by no. × $100 Moderate and Severe 0 Bonus of $1,000 per lane-km only if also no segregation of any type 1 or more Reduce pay by no. × $500 Center-of-Paver More than 1-m Reduce pay by length × $1.50/linear meter After Gavin and Heath (2002). Maximum penalty is limited to $2,000 per lane-km.

45 Texas Over the last 10 years, Texas DOT (TxDOT) has implemented thermal measurements using handheld infrared sensors, density measurements, and requirements for using MTVs to reduce segregation (Rand 2010, 2012). TxDOT, in con- junction with the Texas Transportation Institute (TTI) at Texas A&M University, developed a new specification and test method for the identification of segregation. Three levels of segregation are based on ranges of temper- ature differences: no or limited segregation, <25°F; moderate segregation, from 25°F to 50°F; severe segregation, >50°F. Hand-operated or automated infrared sensors can be used to obtain the temperature information (TxDOT 2011): • Infrared guns (D:s minimum 6:1; accuracy of ±2°F). • Thermal imaging camera (D:s minimum 6:1; accuracy ±4°F) and/or paver-mounted. • Infrared sensor bar (i.e., Pave-IR) (10 sensors; spacing no more than 13 in. apart; maximum 3 ft above surface; D:s minimum 6:2; accuracy ±2°F). Hand-held infrared sensor temperature data are taken approximately five feet behind the paver and no more than 20 feet away from the operator while the paver is moving. Collect and record the maximum temperature in the base line area of the profile length (150 ft; Figure 66). The permissible lower temperature for the next 130 ft is calculated by sub- tracting 25°F from the baseline temperature reading. When the temperature of the mat behind the paver drops below this value the edge of the roadway is marked, the temperature is recorded, and a density profile is required. If the lowest temperature recorded for the temperature profile is more than 50°F lower than the baseline temperature, no QC/QA bonus is paid. A minimum of one temperature and density profile is required for each sublot. Density profiles are also required when the paver stops, at locations where the temperatures are lower than 25°F of the base line temperature, and when visible segregation is identified. Density profiles consist of nuclear density testing at intervals of 5 ft over a 50-ft longi- tudinal section using an offset of 2 ft or more from the edge of the pavement. 5 ft 2 ft 2 ft About 20 ft About 150 ft Lane Width A re a to de te rm in e m ax .b as el in e te m p. Area for continuous scanning with infrared gun or capturing infrared image Excluded areas When Using Infrared Gun or Infrared Camera 2 ft 2 ft About 150 ft Lane Width Area used for one thermal profile to determine maximum and minimum Excluded areas When Using Infrared Sensor Bar temperature FIGURE 66 Area specified for one thermal profile [Source: Tex-244-F (TxDOT) (2014)].

46 The automated data collection system continuously col- lects and records temperature profiles and paver stops. When the contractor uses the automated system, no density profiles are required. Research Development of Segregation Specifications McGhee et al. (2003) and McGhee (2004) evaluated the use of longitudinal texture profiles for improving the unifor- mity of Virginia’s roadways. Texture was estimated for every 2 ft of project length for profiles collected with the left wheel path and between the wheel paths. Specific projects with evi- dence of segregation were identified, surface mixes selected from the Virginia Department of Transportation (VDOT) active maintenance schedule, and new projects were used to evaluate intermediate and base mix testing. Eight projects with both uniform and non-uniform areas were selected for the field evaluations. The variability in the longitudinal texture profiles rather than the surface texture measurements were used to develop a possible specification framework (Table 8). Texture variabil- ity fluctuates proportionally and consistently with the maxi- mum aggregate size. The main advantage to this approach is that information about the original mix characteristics job mix formula is not needed. One disadvantage is that any flushed Standard Deviation of Texture, mm Contract Unit Price Adjustment (% of pavement unit price) 9.5 mm 12.5 mm 19.0 mm 25.0 mm 0.5 and under 0.10 and under 0.15 and under 0.20 and under 105 0.06 to 0.10 0.11 to 0.20 0.16 to 0.25 0.21 to 0.30 103 0.11 to 0.15 0.21 to 0.25 0.26 to 0.35 0.31 to 0.45 100 0.16 to 0.20 0.26 to 0.30 0.36 to 0.45 0.46 to 0.75 90 0.20 to 0.25 0.31 to 0.35 0.46 to 0.55 0.76 to 1.0 80 Over 0.25 Over 0.35 Over 0.55 Over 1.0 Corrective action required Source: McGhee et al. (2003). TABLE 8 QUALITY ACCEPTANCE UNIFORMITY RATING SCALE Range of AREA Index Pay Adjustment Factor 0.5–5.0 105 5.0–15.0 95 15.0–25.0 85 25.0–35.0 65 35.0–45.0 25 Source: After Rowe et al. (2004). TABLE 9 SUGGESTED PAY ADJUSTMENT FACTOR BASED ON TEXTURE CALCULATIONS areas with unusually slick surfaces or texture anomalies in between the laser texture profiles are not captured. Rowe et al. (2004) developed a methodology that estimates the area [i.e., volume of surface (voids) determined from lon- gitudinal texture profiles that were divided into increments (base length)] and the area between the profile and a hori- zontal line extended from the maximum particle height in the increment. The areas calculated for each of the increments are summed and used to determine the level and quantity of segregation. A methodology for the acceptance or non- acceptance of a pavement section and a suggested pay scale were proposed (Table 9). ( ) ( ) ( )= + +A A AAREA 1.0 1.42 2.50low medium High