Standard Definitions for Common Types of Pavement Cracking (2020)

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.

Standard Definitions for Common Types of Pavement Cracking 37 C H A P T E R 5 Proposed Standard Definitions for Common Cracking Types Introduction Manual surveys based on the ability of human surveyors are to subjectively recognize distresses by identifying crack patterns in various locations on pavement surfaces. Automated technologies rely on high-resolution pavement visual data collected in 2D and 3D formats and process the data for cracking characterization using special software tools. The automated technologies have the potential to produce objective cracking data at higher levels of precision and faster speed. However, such ambitions were not fully realized in the past decades due to the underestimation of substantial difficulties and obstacles in consistently producing repeatable high-quality cracking results in varying pavement conditions and for different pavement types. Based on the work completed in Phase I (Chapters 1 to 4), this chapter proposes a preliminary set of standard definitions for common crack types specifically designed for automated cracking survey technologies to meet various engineering practices and requirements. A Set of Standard Definitions The aim of this research was to (a) exploit strengths of computer processing-based automated procedures for higher levels of precision, (b) develop cracking definitions that can satisfy practical needs for pavement management primarily at the network level, and potentially project level work as well. The following guiding principles were streamlined to achieve the stated goals: Principle #1 – Cracking Definitions are Defined with Mathematical Clarity for Automatic Computer Processing Computers and humans are optimized for different tasks. Many pavement distress classification and quantification criteria are easily understandable and implementable for manual surveys due to the fact that human visual intelligence is extremely good at recognizing patterns in variable conditions. However, human processing of pavement condition data is relatively slow, tedious, expensive, and may increase safety concerns due to the exposure of raters to live traffic, and may not be consistent among different raters. Computer algorithms can be extremely effective at identifying well-defined patterns at high-speed; however, unacceptable levels of errors can occur with distress patterns that cannot be accurately defined with mathematical clarity. For instance, asphalt pavement fatigue cracking in the LTPP Distress Manual requires judgment regarding the interconnectivity of cracks, the existence of crack spalling, crack sealant, and presence of

Standard Definitions for Common Types of Pavement Cracking 38 pumping. As a result, numerous studies have shown that automated crack classification and quantification results usually have poor correlation with manual survey results (McQueen and Timm 2005, Albitres et al. 2007). It is of great importance to define standard cracking that can be quantitatively measured by computers at high levels of precision in a consistent manner regardless of pavement condition or pavement type. Therefore, simple definitions of cracking with mathematical clarity are needed for automated implementations. Principle #2 – Objective Quantification of Cracking is More Desired in Automatic Processing It is desired to separate objective quantification and subjective judgment for cracking data collection and interpretation tasks. As previously noted, for automatic processing, only the cracks that can be objectively quantified will generate repeatable results. One example is a two- level crack classifying and quantifying method developed for Caltrans (Fu et al. 2011). The first level includes capturing images and quantifying geometrical features related to distresses seen in the images. At this level, automated surveys report only objective geometrical measurements of pavement distress and condition features on the raw images, but do not make subjective interpretations about the distress severity and extent. The second level includes interpreting the geometrical features from the pavement engineering or management perspective, which is subjective, highly dependent on the agency’s engineering tradition and management needs, and difficult for automated algorithms to do well. If the agency’s needs evolve because of advancement in pavement management principles, historical cracking data can be reprocessed at a low cost by performing only the tasks in the second level with no need to revisit the raw image data in the first level. Principle #3 – Locating Cracks Is More Feasible than Identifying the Causes of Cracks Naming conventions that imply the cause of cracking should be avoided in automatic processing, simply because a computer driven cracking detection system does not have the intelligence or cognition to identify them with high accuracy. Crack location on the pavement surfaces, however, may indicate load or non-load associated cracks. Identifying crack location is relatively straightforward for automated algorithms; therefore, crack location is a recommended parameter in automated processing. Principle #4 – Clear and Quantitative Criteria Are Needed for Cracking Classification and Measurement The definitions of pavement lanes and wheel paths are important for accurate cracking classification. Comprehensive lane and wheel path definitions should consider various scenarios of pavement features, such as those with and without lane markings, narrow or wide lanes, or no shoulders. These factors would significantly impact the accuracy of crack classification and measurement. In addition, wheel path cracking severity should not be determined via the subjective judgment of features such as crack interconnectivity, but based on quantitatively measurable criteria such as length, or crack width. Elaborate crack width measurements based on digital pavement images are often problematic. A simple mean crack width is recommended and defined as the average width measured at quartile locations, or more frequent intervals as

Standard Definitions for Common Types of Pavement Cracking 39 required. Additional research is needed to further evaluate how crack width can be used to define crack severity levels. Principle #5 – Cracking Definitions Eliminate or Reduce the Impact of Non-Critical Influencing Factors An important cause of past failures in evaluating automated technologies for pavement cracking surveys was that many influencing factors had been considered in the evaluations or equipment rodeos, due to the needs of traditional pavement engineering practices. However, such considerations substantially added complexity in data analysis beyond the capabilities of software technology. Entire evaluations were frequently meaningless due to the fact that ambiguity or uncertainty of certain definitions or factors could not be quantified with automated systems. An example is block cracking: “Interconnected cracks that divide the pavement up into rectangular pieces.” The sizes of blocks vary greatly, resulting in substantial challenges for computers to properly classify block cracks. Further, most automated systems save image data into frames with a fixed longitudinal distance. When one block cracking spans in multiple image frames, it introduces an additional factor or analysis requirement resulting in even more inconsistent results. Therefore, only the most critical factors or requirements should be considered in the new cracking definitions to reduce variations and ambiguity in implementation. Principle #6 – Cracking Definitions Support Different Levels of Pavement Management Activities As discussed in Chapter 4, various applications of cracking data include PMS, HPMS reporting, MEPDG local calibration, reporting as required in the final rulemaking in the MAP-21 or FAST Act, and other asset management activities. Different applications require different levels of cracking data. For example, the cracking percentage is the only indicator required for the HPMS reporting, whereas detailed cracking data (e.g., longitudinal cracking, transverse cracking, alligator cracking) are required for project level decision-making. Therefore, the proposed cracking definitions should consider such needs and support several key reporting and pavement engineering applications when possible. However, it should be acknowledged that the current applications are primarily based on manual cracking protocols, meeting all requirements without substantial institutional changes in existing practices will be impossible for the shift to a new generation of definitions for automatic processing. Based on the six principles, the research team recommends a unique and novel set of definitions tailored specifically for automated pavement cracking surveys. Recognizing that many cracking patterns deemed to be “simple” in manual surveys may not be fully automated by modern computers due to cognition limitations of software algorithms. Common terminologies for cracking definitions are adopted in the project from the AASHTO Standard Practice for Quantifying Cracks in Asphalt Pavement Surfaces from Collected Pavement Images Utilizing Automated Methods (AASHTO R85-18):  Crack: a fissure of the pavement material at the surface with minimum dimensions of 0.04 in. (1 mm) width and 1 in. (25 mm) length.

Standard Definitions for Common Types of Pavement Cracking 40  Crack orientation: the angular measurement in degrees between the direction of travel and a line drawn between the crack termini as measured within the measurement zone of interest.  Crack width: the average gap in inches (mm) between the two edges of a crack measured at points along the crack with a spacing between 0.04 in. (1 mm) and 1 in. (25 mm).  Crack length: the length measured along the crack path using all available data points.  Wheel path definition: since locations of cracking are increasingly used in both pavement management and design practices, it is recommended that the definitions of wheel paths dimensions in AASHTO R85 be directly used in the proposed cracking definitions as shown in Figure 2. It is also understood that both the wheel path widths and positions in Figure 2 were determined in accordance with tire spacing and position of typical trucks on U.S. roads. In this project, it is further recommended that cracking on areas outside of and on lane markings be excluded from the cracking definitions. Three levels of cracking data are therefore recommended for asphalt and concrete pavements:  Level 3 cracking data are primarily used for vendor technology verification and qualification.  Level 2 data are used for network level pavement management.  Level 1 data are used for project level pavement engineering. For both asphalt and concrete pavements, the data analysis for cracking definitions begins with Level 3, followed by Level 2, and then Level 1. The analysis process is accumulative from Level 3 to Level 1. Asphalt Pavements Level 3: Cracking Percent for Extent Level 3 definitions only compute the percent of cracking between the inside edge of the lane markings; classification, crack severity, and other details are not included. The concept of the Level 3 cracking definition is based on the UK SCANNER method which computes the percent of each 528 ft. (161 m) roadway section affected by cracking. The roadway surface is divided into 10 in.  10 in. (250 mm 250 mm) grid squares, and the index for Level 3 cracking is calculated using (HGL 2005): %100Index 3 Level  N nc (4) where, cn = Number of 10 in.  10 in. (250 mm 250 mm) grids containing cracks. N = Total number of 10 in.  10 in. (250 mm 250 mm) grids for a 528 ft. (161 m) pavement section. Per AASHTO R 85 and the HPMS Field Manual, wheel paths are defined as a longitudinal pavement strip having 39 in. (1 m) width and two wheel paths separated by a 30 in. (0.75 m)

Standard Definitions for Common Types of Pavement Cracking 41 median zone. The proposed 10 in.  10 in. (250 mm 250 mm) square grids for Level 3 computation is recommended so that the layout for the wheel paths and the median zone results in no partial grids (Figure 16). The grid layout for areas outside the wheel paths, for pavement widths other than 12 ft (3.6 m), may result in incomplete grids. Figure 16 Level 3 with 10 in. 10 in. (250 mm x 250 mm) grids with wheel paths. A sensitivity analysis was conducted to determine the effect of different grid sizes on the Level 3 definition. As shown in Figure 17, for two pavement samples (one with low extent cracks, the other with severe extent cracks), the index values increase nearly linearly as grid sizes increase from grid size 4 in. (100 mm) to grid size 39 in. (1000 mm). At this time, the grid size of 10 in.  10 in. (250 mm 250 mm) is used to simply divide the zones into an integer number of squares for now, which may be subject to further validation and sensitivity analysis in the future. Level 3 defines the extent of cracking and is useful for network level pavement management activities at the macro level. It also has the potential to assist with performance evaluation of various automated cracking data collection devices, such as a prequalification of potential technology suppliers. The described approach is easy to apply in software algorithms due to lack of ambiguity in implementation; ambiguity was a primary cause of different interpretations of cracking definitions in the past evaluation efforts. The proposed Level 3 cracking definition is conceptually similar to the MAP-21/FAST Act cracking performance measure in terms of cracking percent. It should be pointed out that results from Level 3 analysis would not be sufficient to conduct many pavement engineering tasks, which would require more detailed cracking data from Level 2 and Level 1 analyses.

Standard Definitions for Common Types of Pavement Cracking 42 Figure 17 Sensitivity of cracking percent at Level 3 to grid size. Level 2: Cracking Width for Severity The majority of SHAs use crack width to define the severity levels. The most common severity thresholds include low: ≤ 0.25 in. (6 mm), medium: 0.25 in. (6 mm) to 0.50 in. (12 mm), and high: ≥ 0.50 in (12 mm). For Level 2, the average width of cracking is computed as a single value for each 10 in.  10 in. (250 mm 250 mm) grid. The severity index is based on the percent of grids within each of the three severities for the 528 ft. (161 m) lane section. For Level 2, three statistics are generated representing the percent of cracked grids at each severity level over the entire 528 ft. (161 m) lane section. As a numerical example: for a 528 ft. (161 m) pavement lane section, the Level 3 Index is 15% as a percentage of grids with cracking; within this Level 3 15% cracked area, 25%, 35%, and 40% of them are at high severity (H), at moderate severity (M), and at low severity (L) respectively for the Level 2 statistics. Level 2 also calculates the percent cracking, by severity level, for each of the five zones, resulting in an additional 15 statistics per every 528 ft. (161 m) lane section. It is further recommended that cracking percent as defined in Level 3 is computed for each of the five zones. For instance, the overall cracking percent (Level 3) numbers are 15%, 35%, 5%, 35%, and 10% for Zones 1 to 5, respectively, indicating the pavement structure has experienced load related deterioration. For Zone 2 in particular, the cracking percent at low, medium, and high severity levels (Level 2) are 10%, 20%, and 70%, indicating the cracking in Zone 2 is poor in terms of severity. Therefore, a total of 23 (3+15+5) statistics or indices are included in the Level 2 calculations. Level 1: Cracking Positions and Orientations for Project Level Analysis Level 1 cracking definition focuses on crack position and orientation to meet the application requirements of project level PMS and pavement design. Implementation feasibility for fully automated cracking survey technologies is a particular consideration in the proposed Level 1 definitions. Cumulative results from the three levels of analysis would meet the demands of pavement engineering practices. For Level 1, cracking is defined as: 0 20 40 60 80 100 0 10 20 30 40 Ca ck  P er ce nt ag e  (% ) Grid Size (Inch) Severe Extent Crack Low Extent Crack

Standard Definitions for Common Types of Pavement Cracking 43  Wheel path crack: any crack at least 12 in. (0.3 m) long within the wheel path zones.  Longitudinal crack: a crack, outside the wheel paths, at least 12 in. (0.3 m) long, with a crack orientation ±20 degrees relative to the centerline.  Transverse crack: a crack, spanning more than two zones, with a crack orientation between 70 and 110 degrees relative to the centerline. A transverse crack is recorded as 6 ft. (1.8 m) long if it spans two zones and 12 ft. (3.6 m) long if it spans three or more zones.  Sealed crack: the placement of a sealant material into cracks of an existing pavement surface to prevent excess water and moisture from penetrating the pavement.  Other cracks: all other cracks outside of the wheel path zones, not identified as longitudinal, transverse, or sealed cracks. In this category, special crack types can be defined or customized by individual SHAs to fit different agency needs. The obtained crack data shall be calculated and reported by each of the pavement zones in Figure 16 for each crack category with the following Level 1 quantification procedure:  Longitudinal crack is computed in length in feet (meters) along the path of the crack separately for each severity level for each non-wheel path zone.  Transverse cracking includes the length, in feet (meters), for each severity level.  Wheel path cracking includes the length, in feet (meters), for each severity level. Further, the grid area of wheel path cracking, all severity levels, divided by the total area is calculated as the percent of wheel path crack.  Sealed cracking includes the length, in feet (meters), without consideration of severity levels. Severity Levels The average width of cracking is computed to define low, medium, and high severity levels of pavement cracking:  Low: mean crack width ≤ 0.25 in. (6 mm).  Medium: mean crack width > 0.25 in. (6 mm) and ≤0.50 in. (12 mm).  High: mean crack width > 0.50 in. (12 mm). Applications Per the review of desired cracking data by SHAs in Chapter 4, potential applications of the obtained from the three levels of cracking data of asphalt pavements are as follows:  HPMS and MAP-21 reporting: percent wheel path cracking.  MEPDG calibration: length of transverse cracking can be associated with MEPDG transverse cracking (ft/mi), and the percent of wheel path cracking can be associated with MEPDG alligator cracking (% lane area).  PMS: the severity, extent, and cracking percentage for all the defined cracks will be used for PMS including cracking performance prediction and treatment timing. The applications may vary among different SHAs. In addition, the analysis results can be integrated with non-cracking distresses to form condition-based indices.

Standard Definitions for Common Types of Pavement Cracking 44 Concrete Pavements It is recommended that Levels 3 and Level 2 cracking definitions for asphalt pavements are directly used for concrete pavements without the consideration of wheel path definitions. Similar to the grid patterns for asphalt pavements shown in Figure 16, the grid generation for concrete pavements starts at mid-lane and propagates to both lane markings. If lane markings are visible in the pavement images, the middle is determined as the equidistance between lane markings. If lane markings are not visible, the middle is equivalent to the middle of the image. Currently, the AASHTO R 85 only includes cracking on asphalt pavements. However, the AASHTO R 85 minimum crack length and orientation for longitudinal and transverse cracking are recommended for use with concrete pavements. Cracking definitions and application for JPCP/JRCP and CRCP for Level 1 are summarized below: JPCP/JRCP For JPCP/JRCP surfaces, four crack types are defined for pavement evaluation as follows for Level 1:  Longitudinal crack: a crack, at least 12 in. (0.3 m) long, with a crack orientation of ± 20 degrees relative to the centerline.  Transverse crack: a crack, at least 12 in. (0.3 m) long, spanning more than two zones, with a crack orientation between 70 and 110 degrees relative to the centerline.  Corner break: a crack, which intersects the adjacent transverse and longitudinal joints, describing approximately a 45-degree angle with the direction of traffic. The length of the sides is from 12 in. (0.3 m) to half the width of the slab on each side of the corner.  Other cracks: all cracks not identified as longitudinal crack, transverse crack, or corner breaks. The obtained crack data shall be calculated and reported for each crack category at Level 1. Cracking is quantified as the sum of the lengths, in feet (meters), of longitudinal, transverse, and other cracks for each user-defined reporting section. The number of corner breaks is summed over the user-defined reporting section. The number of slabs containing one or more transverse cracks and the number of slabs with cracks are recorded separately. The average width of cracking is computed to define low, medium, and high severity levels of cracking on JPCP/JRCP. Cracking severity levels include:  Low: mean crack width ≤ 0.25 in. (6 mm).  Medium: mean crack width > 0.25 in. (6 mm) and ≤0.50 in. (12 mm).  High: mean crack width > 0.50 in. (12 mm). Applications Per the review of desired cracking data by SHAs in Chapter 4, potential applications of the obtained Level 1 cracking data of JPCP/JRCP are as follows:  HPMS and MAP-21 reporting: percent of transverse cracked slabs.

Standard Definitions for Common Types of Pavement Cracking 45  MEPDG calibration: percent of transverse cracked slabs.  PMS: the severity, extent, cracking percentage, and percent of slabs for all the defined cracks will be used for PMS. The applications may vary among different SHAs. In addition, the analysis results can be integrated with non-cracking distresses to form condition-based indices. CRCP For CRCP surfaces, four crack types are defined for Level 1:  Longitudinal crack: a crack, at least 12 in. (0.3 m) long, with a crack orientation of ± 20 degrees relative to the centerline.  Transverse crack: a crack, at least 12 in. (0.3 m) long, spanning more than two zones and with a crack orientation between 70 and 110 degrees relative to the centerline.  Punchouts: the area enclosed by two closely spaced (usually < 2 ft. (0.6 m)) transverse cracks, a longitudinal crack, and the pavement edge or the longitudinal joint.  Other cracks: all cracks not identified as longitudinal crack, transverse crack, or punchouts. The sum of the lengths, in feet (meters), of longitudinal, transverse, and other cracks are summarized in each user-defined reporting section. The number of punchouts is summed over the user-defined reporting section. The average width of cracking is computed to define low, medium, and high severity levels of cracking on CRCP. Cracking severity levels include:  Low: mean crack width ≤ 0.25 in. (6 mm).  Medium: mean crack width > 0.25 in. (6 mm) and ≤0.50 in. (12 mm).  High: mean crack width > 0.50 in. (12 mm). For punchouts, the severity is measured by the punchout area ratio, which is defined as the sum of the punchout area divided by the total area of the reporting section. Applications Per the review of desired cracking data by SHAs in Chapter 4, potential applications of the obtained Level 1 cracking data of CRCP are as follows:  HPMS and MAP-21 reporting: percent area of longitudinal cracking and punchouts.  MEPDG calibration: number of punchouts.  PMS: the severity, extent, and cracking percentage for all the defined cracks will be used for PMS. The applications may vary among different SHAs. In addition, the analysis results can be integrated with non-cracking distresses to form condition-based indices. AASHTO R85 and the Proposed Definitions The AASHTO R85 is a well-known set of cracking definitions outlining the procedures for quantifying cracking at the network level for asphalt pavement surfaces utilizing automated

Standard Definitions for Common Types of Pavement Cracking 46 methods. Some definitions in AASHTO R85 (AASHTO 2018) were borrowed in the proposed definitions, including: (1) Crack, crack orientation, crack width, and crack length. (2) Wheel path definition. (3) The minimum crack length is 12 in. (0.3 m). (4) The angles between crack and lane centerline is +20 and -20 degrees for longitudinal crack and 70 and 110 degrees for transverse crack. In addition, to develop standard definitions for common cracking types for asphalt and concrete pavements utilizing automated systems, some novel considerations are added in the proposed definitions based on the extensive literature review, online survey results, and various engineering practice requirements. Deviations of the proposed protocols from AASHTO R85 include the following: (1) Three levels of cracking data are proposed: a) Level 3 cracking data are primarily used for vendor technology verification and qualification; b) Level 2 data are used for network level pavement management; c) Level 1 data are used for project level pavement engineering. (2) Image area outside of and under lane markings is excluded from the cracking definitions. (3) Wheel path crack is important: a) the literature review and online survey indicate some SHAs started to use wheel path crack to define load related cracks, b) wheel path cracking is easier to identify and quantify than the pattern crack as defined in AASHTO R85 which is difficult for fully automated implementation. (4) Sealed crack is included: a) 74% of the 35 SHA respondents require reporting of sealed cracks, b) it is possible to identify sealed crack automatically from images. (5) Cracking severity levels are defined as low, medium, and high. (6) Cracking definitions for concrete pavements are included for JPCP/JRCP and CRCP. Summary Based on the surveys of SHAs, and the team’s long-term experience in developing automated distress survey technologies, a set of six principles were presented as the technological basis to introduce new cracking definitions. A total of three levels of cracking definitions are proposed. Level 3 provides the measurement of the extent to reveal the general cracking condition of pavements, which can be used as the initial qualification and evaluation of potential technology suppliers, and meet reporting needs. Level 2 addresses cracking severity for network level PMS, and Level 1 meets the requirements of project level PMS, and pavement design. Cracking definitions for concrete pavements are also proposed for JPCP/JRCP and CRCP pavements. A comparison between the proposed cracking definitions and the definitions in the current AASHTO R85 was made at the end of the chapter.