Automated Pavement Distress Collection Techniques (2004)

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.

3To meet increasing demands for network-level pavement condition data, state and local highway agencies are faced with the problem of collecting more data, at lower cost, with improved quality, in a safer manner. Moreover, most agencies are attempting to meet those needs with reduced resources. Recent state-of-the-art advancements in computer technol- ogy, digital pavement imaging, and digital image process- ing provide greatly enhanced methods to collect and inter- pret the required information. Currently, highway agencies are employing a wide variety of approaches to collect, store, analyze, and disseminate information on pavement condition. The need to synthesize and incorporate the information on these varied approaches into a single resource document became evident and gave rise to the present synthesis, which, in many ways, is an update of several earlier similar under- takings as will be described later in this chapter. The reader is cautioned that the synthesis attempts to cover what is being done in the field, but it does not constitute a “toolbox” (a how to do it) of automated pavement data collection and process- ing. The synthesis effort ultimately revealed that the develop- ment of such a toolbox is one of several research needs. OBJECTIVE This synthesis focuses on automated pavement condition data collection techniques, specifically for the measurement of pavement cracking, roughness, rutting, and faulting. The major objective was to document how agencies conduct automated data acquisition and processing for network-level pavement management. Other information included contrast- ing state of the art with the state of the practice from the per- spective of user agencies. Furthermore, it was desirable to synthesize the contractual arrangements in use by the various agencies, as well as the costs of automated data collection and processing. Finally, data quality assurance (QA) programs were to be synthesized. SCOPE AND ORGANIZATION The synthesis (1) reviews the literature to document method- ologies and equipment available to highway agencies to auto- mate the collection and processing of pavement condition data and (2) reports the results of a survey of state highway agen- cies to determine the current state of their practice in adopting automated distress collection techniques. The available inter- national information is included. The synthesis is limited to network-level data only. Project-level data are beyond the scope and are not generally discussed, except to the extent that some data elements (e.g., linear referencing) can have both network and project applicability. The automated pavement distress data collection tech- nologies addressed by this synthesis may be grouped into two classes: (1) imaging technologies involving the capture and interpretation of images of the pavement surface and (2) sens- ing technologies involving the use of noncontact sensors to measure deviations of the pavement surface from a horizon- tal plane. Specific information captured in the synthesis includes the following: • A discussion of automated versus semi-automated tech- niques, including the degrees of automation used and human intervention required; • The benefits of automated techniques (including any benefit–cost studies); • Methods of procurement (in-house, contracted, etc.); • Contracting procedures (warranties, penalties, audit pro- cedures, period of performance, etc.); • Quality control (QC) and QA procedures used to vali- date and evaluate data, or to determine data accuracy, variability, and consistency; • Equipment specifications and test protocols that states are using; • Monitoring frequencies, by system; • Degree of adoption of AASHTO, ASTM, and other standards; • Costs of automated techniques (including costing method used); • Additional features being collected by the automated equipment, such as right-of-way, drainage, and signage inventory, excluding devices that monitor only these features; • Case studies of states using different methods; • Limitations of available technologies; and • The contrast between the state of the art and the state of the practice. The synthesis is organized to first present a general discus- sion of background materials, methodology of development, and questionnaire responses. Sections that roughly parallel the questionnaire’s organization follow that general discussion. CHAPTER ONE INTRODUCTION

However, because those items are applied in approximately the same way regardless of the distress under consideration, location reference, distress monitoring frequency, and road- way sampling frequency are discussed before the distresses themselves. The distress sections cover the collection and pro- cessing of surface distress, smoothness and roughness, rutting, and joint-faulting data. The latter chapters of the synthesis address the more general issues of data management, con- tracting procedures, costs of automated data, advantages and disadvantages of automated collection and processing, and QC and QA issues. Finally, Appendices A, B, and C, respectively, present the questionnaire itself, responses to the questionnaire, and identify the agencies that responded. BACKGROUND AASHTO provides the following definition of the subject being studied: “A pavement management system (PMS) is a set of tools or methods that assist decision-makers in finding opti- mum strategies for providing, evaluating, and maintaining pavements in a serviceable condition over a period of time” (1). Although called by other names, programs directed at provid- ing the best possible highways at the lowest practical cost to the taxpayer have been in use since the first state highway agencies were established. Since the time such agencies were created, a need for securing some measure of the condition of the high- way network(s) being managed was recognized (2). Almost all earlier data were collected through visual surveys (3). Even pavement roughness was seldom measured until about the 1950s (3), with mechanical measures of pavement condition in use for only about the last half-century. Over that time, there evolved a recognition that safety issues and issues relating to data volume and quality dictated that high-speed and objective means of automated pavement condition data collection and processing were needed. The following discussion provides the background for the technologies involved in pavement condi- tion evaluation and how those technologies have evolved. Three earlier syntheses summarized much of the background material, and they are referenced frequently in this report. One of the earlier documents to catalog methods of pave- ment condition evaluation was NCHRP Synthesis of Highway Practice 76, published in 1981 (3). The focus of that synthesis was on the collection and use of pavement condition data, and the scope included discussions of the need for condition data, the types of data collected, the sampling programs employed, and the costs of various data elements. Interestingly, even in 1981, only a relatively small number of agencies were involved in pavement data collection on a large scale. For that reason, only nine agencies (including the U.S. Army Con- struction Engineering Research Laboratory) were surveyed. All those agencies were collecting data on pavement distress, ride quality, structural integrity, and skid resistance. In 1981, the method of choice for pavement distress eval- uation was the visual survey, because even photographic 4 methods were only in the research phase at that time. Most agencies doing roughness monitoring were using response- type road roughness measuring systems (RTRRMS), such as Mays or Cox meters mounted on passenger cars or on trail- ers. Although widely used at that time, this technology was severely limited because the equipment measured only vehi- cle response to the longitudinal road profile rather than the profile itself. Furthermore, RTRRMS roughness values were subject to all kinds of vehicle suspension and other varia- tions, so that frequent calibrations were required to achieve any consistency of results. Of the agencies surveyed, only the U.S. Army Construction Engineering Research Laboratory reported using a laser profilometer to measure actual profile. No automated methods of collecting rut-depth or joint-faulting measurements were reported in that synthesis. The next related synthesis was the 1986 NCHRP Synthe- sis of Highway Practice 126: Equipment for Obtaining Pave- ment Condition and Traffic Loading Data (4). Again, the pavement-related information gathered was on surface dis- tress, roughness, structural capacity, and friction. Weigh- in-motion equipment development was in its infancy, but it was included. A questionnaire was circulated in 1983 to all 50 states, and responses were received from 44. By that time, a few agencies were using photologging tech- niques to capture pavement surface condition, although most still used visual surveys. The study reported that transfer of distress data from the photographs proved to be both time con- suming and expensive. Most road roughness monitoring still was done by RTRRMS, although profilometers and profilers were being used by a number of research units. At the same time, five states had started to use a new version where road meters using ultrasonic waves measured vehicle displace- ments relating to road roughness. Furthermore, the surface dynamics profilometer, employing two spring-loaded follow- ing wheels, was in production use in four states. Also, by then K.J. Law Engineers had developed a noncontact profilometer using an optical reflectivity system to measure profiles in both wheel paths. No agency reported routine use of this version of profilometer in 1983. Apparently, a compelling reason that profiling equipment was slow in coming into widespread use was the huge gap in investment cost between those devices and the RTRRMS. The synthesis reported a cost of some $15,000 for a ride meter-equipped automobile and approximately 15 times that much for a profilometer. By the time a third synthesis effort was undertaken in 1994, several federal initiatives had provided special impe- tus to pavement data collection activities through mandated or semi-mandated programs. One of these, the Highway Per- formance Monitoring System (HPMS), requires that states provide pavement condition information to support the func- tions and responsibilities of the FHWA. The Highway Per- formance Monitoring System Field Manual has been pub- lished in numerous editions, the latest of which was issued in December 2000 (5). That manual requires a periodic section-

5by-section report of pavement condition parameters for high- ways designated as HPMS sections. Those sections generally number in the hundreds to thousands of sections per state, depending on the size of the system administered. The major pavement condition data requirement is either the Interna- tional Roughness Index (IRI), described later, or the present serviceability rating, described in the HPMS manual, on a biennial basis. The HPMS manual also provides for a stan- dardized linear reference system that is now used by many agencies. The Intermodal Surface Transportation Efficiency Act of 1991 provided another federal incentive through a mandate that all roads eligible to receive federal-aid monies were to be covered by a PMS by January 1, 1995. Although ensuing directives lifted the mandate, by that time many of the states had implemented PMSs and were collecting the supporting data for their own purposes. The third related synthesis is NCHRP Synthesis of High- way Practice 203: Current Practices in Determining Pave- ment Condition (6 ), published in 1994. The scope of that report addressed the measurement or collection, reporting, and use of pavement condition data, including roughness, surface friction, distress, and structural evaluation. Location- reference methods and data management approaches also were addressed in that synthesis. Questionnaires were widely distributed and responses were received from all 50 states, the District of Columbia, and 9 Canadian provinces. By 1994, essentially all agencies were collecting pavement condition data in one form or another. There was still no con- sensus on how distress surveys were conducted, although many agencies still were using windshield or walking surveys. Six agencies had adopted video distress collection techniques and one was using photologging. None reported automated distress data processing techniques (automated reduction of distress types, frequency, and severity from images) at that time. The synthesis did report that about one-half of the agen- cies were using automated means to collect rutting data. There was no elaboration on those automated methods. In the late 1980s and early 1990s, roughness data collec- tion had undergone an enormous evolution. By 1994, nearly all agencies were collecting and reporting IRI data to the HPMS program. Although an FHWA initiative, the HPMS program had a special advantage for the states, because it pro- vided a standard making it possible to compare roughness data between states. NCHRP Synthesis 203 (6) describes the IRI as a standardized roughness measurement that is calcu- lated by mathematically applying a reference quarter car sim- ulation to a measured profile. Based on extensive research, the World Bank established those simulation parameter values that best represented roughness-related measuring equipment being used worldwide (7,8). The IRI is measured in units of meters and kilometer or inches and miles, and it can easily be related to those measurements obtained by RTRRMS. This index is very useful for relating a roughness measure to over- all ride quality, which is obtained at highway speeds. However, pavement profiling is still an evolving science, as demonstrated by the trials of the Long-Term Pavement Per- formance (LTPP) program as highlighted in the LTPP Man- ual for Profile Measurements (9) and LTPP Profile Variabil- ity (10). These documents address the “how to” and the variability issues of profile measurement, respectively, for the program. At least three national efforts have contributed to develop- ments in pavement condition evaluation in general and to automated surveys in particular over the past decade: the development and implementation of the Strategic Highway Research Program (SHRP) Distress Identification Manual for the Long-Term Pavement Performance Project (11), updating of the AASHTO Pavement Management Guide (1), and pub- lication of AASHTO provisional standards on pavement data collection (12). Starting in 1989, the implementation of the distress identi- fication manual on LTPP general and specific pavement stud- ies gave credibility to a set of pavement distress definitions for both asphaltic concrete and portland cement concrete (PCC) pavements. These definitions were adopted by many agencies that had lacked an agency-specific distress manual. Other agencies made modifications to their procedures to more closely adhere to what was to become somewhat of a national standard. Later work by LTPP contractors assessed the vari- ability of distress data collected for the program (13). Still later, work addressed concerns about possible differences in manual versus photographic distress survey results (14). That study addressed a proposed consolidation of manual and pho- tographic databases consisting of a reconciliation of differ- ences between the two survey methods. Results showed that such consolidation was unnecessary, for data collected by the two methods could be combined for analysis purposes. It should be noted that LTPP applies a rigorous QA process to both types of data and much of both types is rejected for fail- ure to meet quality standards. Similarly, the AASHTO pavement management guide- lines published in 1990 and revised in 2001 provided agen- cies with information on what types of data to collect, the importance of various data elements, and the management of much of that data. Perhaps one of the most important guide- lines cautioned against gathering “huge amounts of data sim- ply because such collection is automated or available” (6). The guidelines make specific recommendations that agencies should make sure that data would be used before resources are expended in their collection and storage. The development and implementation of the AASHTO provisional standards on pavement data collection have con- tinued over approximately the last decade. The standards relative to this synthesis are those on quantifying cracks in

asphalt pavements, quantifying pavement roughness, and determining rut depth and joint faulting of concrete pave- ments. Because two-thirds of the AASHTO member agencies must approve the provisional standards before they become “standards,” revisions and re-balloting may occur for an indef- inite time. The most recent versions are dated 2003, but they still carry the designation “provisional.” The provisional stan- dards are discussed in detail in later chapters. At this point, it is sufficient to note that they provide a beginning for stan- dardization of the procedures. In 1999, Wang and Elliott (15) studied pavement imaging and distress identification from those images for the Arkansas State Highway and Transportation Department. Although the focus of their work was an automated crack identification and classification system, the study also provided an excellent overview of technology in the area of automated pavement distress surveys and looked at possible future directions. Even in 1999, almost all pavement imaging was through very high speed (VHS) videotaping, because digital technologies were just emerging in the pavement evaluation area. The report concludes that “there still exist limitations in accuracy, speed, and degree of automation with the existing systems” (15). Clearly, the adoption of methods and technologies associ- ated with pavement distress data collection and analysis has evolved rapidly over about the past decade and even more rapidly over the few years since the aforementioned work was reported. Now, pavement management personnel can interact through the Internet and other platforms, such as the annual FHWA-promoted regional pavement management confer- ences, to exchange information and advance the rapid matur- ing of the relevant technologies. Further contributing to this rapid evolution is a parallel rapid evolution in computing and imaging technologies. Today’s computers can handle the enormous volumes of data necessary to support pavement management decisions, where- as the digital technology makes it possible to capture pavement images in a much more user-friendly format than was avail- able with photographic and video technologies. METHODOLOGY Agency Questionnaire Information for synthesis development was gathered by means of an extensive literature search and through a ques- tionnaire, which was distributed to prospective user agencies in both the United States and Canada. In addition, information was gained from the Pavement Evaluation 2002 Conference held in Roanoke, Virginia, during late October 2002. This con- ference provided an opportunity for one-on-one contact with many of the user states and with the vendors furnishing data collection services. In addition, vendors were invited by con- ference organizers to demonstrate their equipment and proce- 6 dures on a specially designated test site near the conference center. The project questionnaire is contained in Appendix A and provides for feedback on these points: • Whether or not an agency was collecting automated pavement condition information; • If not, over what time frame, if any, the agency ex- pected to automate some of its pavement data collection activities; • For surface distress (cracking, patching, etc.), smooth- ness (IRI), rut-depth measurement, and joint-faulting measurements the following items: – Methodologies of data collection, – Location referencing used, – Monitoring frequencies, – Sampling techniques, – Data management, – Costs, and – Protocols used; • Peripheral data (right-of-way, signs, drainage inventory, other) collected and how; • Contracting procedures; • QC/QA procedures; • Benefits and advantages of automation; and • Other issues, problems identified, and changes foreseen. Completed questionnaires were received from 42 states and the District of Columbia, from two FHWA offices, from 10 Canadian provinces and territories, and from Transport Canada (on airfields). Although several prospective ven- dors requested copies of the questionnaire, none provided a response. All except two of those responding were using auto- mated collection on at least one pavement condition data ele- ment. Detailed response tabulations are given in Appendix B, Tables B1 through B9, and are discussed in detail in later chapters. Literature Search An extensive literature search made use of the Library of Con- gress, National Technical Information Service, and Trans- portation Research Information System (TRIS) databases, as well as those of many of the state transportation agencies. TRIS search parameters of “automated and pavement and distress or automated and pavement and condition” returned 130 documents, approximately one-half of which were directly applicable to the current synthesis. Other sources of informa- tion were the FHWA, the Permanent International Associa- tion of Road Conferences (PIARC), and the Australian Road Research Board. The literature surveyed provided general background for the synthesis and an understanding of the technologies involved. In addition, the literature search provided the basis for assess-

7ment of the pavement distress data collection state of the art. One project objective was to contrast the state of the art and the state of the practice as conveyed by questionnaire responses. The relevant discussion is found in chapter nine. DEFINITION OF TERMS Most of the terms used in this synthesis are those common to the pavement engineering community and for convenience most are defined in the Glossary. However, at least two terms are considered specific enough to this document to warrant special definition. These are “automated collection” and “auto- mated processing” in the context of pavement distress data. For the purposes of this synthesis, automated collection was defined, in the project scope, by the topic panel as “data collected by imaging or by the use of noncontact sensor equip- ment.” Thus, data collected through manual procedures, even if recorded by laptop computers or by other methods, are not considered to have been collected through automated means. Manual collection is understood to mean that data are collected through processes where people are directly involved in the observation or measurement of pavement surface properties without the benefit of automated equipment (e.g., visual sur- veys and faultmeters). The processing of data collected through noncontact sen- sors is almost always to some degree automated, for the vol- umes of data relating to transverse and longitudinal profiles are such that manual processing would not be practical. Therefore, for the purposes of this synthesis, automated pro- cessing pertains to the reduction of pavement surface dis- tresses, such as cracking and patching, from images. The def- inition of fully automated is that such distresses are identified and quantified through techniques that require either no or a very minimum of human intervention. Typically, automated in the context of pavement cracking involves the use of digi- tal recognition software capable of recognizing and quantify- ing variations in grayscale that relate to striations (sometimes cracks) on a pavement surface. DISCLAIMER Throughout this synthesis report there are references to vendor reports and websites. These references are used in the course of supporting or explaining technical concepts that arise with many of the issues addressed. Vendor claims about processes and products should be independently ver- ified in the event that a referenced vendor’s services are desired.