National Academies Press: OpenBook

Quality Management of Pavement Condition Data Collection (2009)

Chapter: Chapter Three - Data Quality Management Concepts

« Previous: Chapter Two - Pavement Condition Data Collection Overview
Page 20
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 20
Page 21
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 21
Page 22
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 22
Page 23
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 23
Page 24
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 24
Page 25
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 25
Page 26
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 26
Page 27
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 27
Page 28
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 28
Page 29
Suggested Citation:"Chapter Three - Data Quality Management Concepts." National Academies of Sciences, Engineering, and Medicine. 2009. Quality Management of Pavement Condition Data Collection. Washington, DC: The National Academies Press. doi: 10.17226/14325.
×
Page 29

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.

20 A complete pavement condition data collection quality man- agement program provides a comprehensive, systematic approach to data collection and processing. This chapter pre- sents the main data quality management concepts and princi- ples as they apply to pavement condition data collection; the main sources of variability in pavement condition data collec- tion; and the general policies, principles, and guidelines cur- rently being followed by transportation agencies to conduct quality management activities. The concept of quality is reasonably simple to understand; however, the methods, tools, attitudes, and values involved in providing high-quality products and services are not. Approaches to achieve quality, such as Total Quality Manage- ment (TQM), are often difficult to understand and operational- ize because they are not simply toolkits, methods, or manage- ment theories, but rather part of an encompassing quality approach (44). Furthermore, the concept is particularly com- plex to apply to services such as pavement data collection, because the approach was developed for, and has had its great- est successes in, manufacturing industries where the “product” is a clearly defined physical entity. Managing the quality of pavement condition data is particularly challenging because not only the product is not clearly known, but also the ground truth or reference value is often difficult to determine (45). It is important that the quality principles and practices used in the manufacturing industries be adapted to match pavement data collection services. The most efficient way to achieve high-quality services is to adopt a comprehensive quality management approach that includes methods, techniques, tools, and model prob- lem solutions. The development of a quality management sys- tem requires, as any management system, the interaction of three fundamental components: processes, people, and tech- nology. If any of these components is lacking, it is unlikely that the system will be successful (46). Furthermore, adopt- ing a comprehensive quality approach is typically hard to justify with a simple cost-benefit analysis. The costs and ben- efits of a quality “approach” are clear only after the quality processes have been tested and the organization can exploit the benefits from the improved quality. The overall quality approach would include the following elements (47): 1. Identification and documentation of the procedures that cover all key business processes (control of documents); 2. Monitoring processes to ensure these procedures are effective (including audits); 3. Adequate record keeping (control of records); 4. Continuous checking output for defects (control of nonconforming product/service), with appropriate corrective actions; 5. Periodic reviews of individual processes, preventive actions, and the quality system itself to verify its effec- tiveness (often including both internal and external audits); and 6. Fostering continuous improvement. PAVEMENT CONDITION DATA COLLECTION QUALITY MANAGEMENT SYSTEM Independently of the mechanism used to collect the data, in- house or through a service provider, a complete pavement con- dition data quality management system would include a clearly documented quality management plan, detailed quality accep- tance procedures, and established guidelines to monitor the entire process (as summarized in Figure 2). It is important that the quality management plan include the activities to be con- ducted, as well as clear timelines, milestones, and evaluation criteria. For the quality management system to properly work, everything from effective data collection procedures and training to efficient data processing and quality control/quality acceptance reviews needs to be performed in a timely manner. For example, previous studies have suggested that the steps in the quality management process of distress data col- lection include the following activities [adapted from Morian et al. (45)]: • Distress definition; • Rater training (and equipment calibration); • Systematic data collection process management; • Systematic data handing and processing; • Timely, effective quality control system; • Timely, effective quality acceptance check system; • Timely identification and implementation of corrective actions; • Timely report development; and • Delivery of results to the owner agency. BACKGROUND ON QUALITY MANAGEMENT CONCEPTS AND PROCESSES The subjects of quality, quality management systems, qual- ity management, quality control, and quality acceptance have been the focus of a series of documents developed by the CHAPTER THREE DATA QUALITY MANAGEMENT CONCEPTS

21 International Standards Organization (ISO) known as the ISO 9000 series, or “family,” of standards (48). These stan- dards offer quality management guidance and identify the elements necessary to direct and control an organization with regard to quality. The standards lay out the requirements for an organization (company, public agency, etc.) to deliver products and services that consistently meet customer expec- tations and for an organization to be “certified.” The Deming cycle, plan-do-check-act, is one of the models that can be used to achieve higher levels of quality. One of the main concepts that have influenced the develop- ment of the aforementioned standards is that of TQM. Total Quality is a process by which an organization strives to provide customers with products and services that satisfy its needs. The TQM philosophy seeks to integrate all organizational functions (design, engineering, production, etc.) to focus on meeting customer needs and organizational objectives. The key prin- ciples of TQM include management commitment (plan- do-check-act), employee empowerment, fact-based decision making, continuous improvement, and customer focus (49). Another business management approach that is closely related with quality is Six Sigma. This management philoso- phy, originally developed by Motorola, emphasizes setting extremely high-quality objectives, collecting data, and ana- lyzing results to a fine degree as a way to reduce defects in products and services. The name comes from the Greek let- ter sigma, which is used to denote variation from a standard (e.g., standard deviation). The philosophy behind Six Sigma is that if you measure how many defects are in a process, you can determine ways to systematically eliminate them and get as close to perfection as possible (50). IMPORTANCE OF QUALITY DATA TO SUPPORT PAVEMENT MANAGEMENT “Good” data are very important in providing effective pave- ment management. In particular, adequate quality and quantity of pavement condition data are a very important component in a PMS. For example, accurate and temporally consistent data are critical to develop models to predict smoothness and crack progression (38). These models are necessary for developing effective multi-year preservation plans and work programs. Even when performing network-level analysis, errors in the data can have a significant effect on the recom- mended treatments and budgetary requirements. Systematic errors are considered especially critical at the network level, where a large volume of data is collected and errors can be compounded (51). Less critical are the random errors, because it is expected that they will offset each other if enough data are collected. Adequate quality management can help eliminate sys- tematic errors and minimize random errors. For example, in Virginia, the introduction of a third party to provide inde- pendent validation and verification has been especially use- ful. A third-party contractor was asked to manually check 10% of the data collected and analyzed through automated methods. This process provided a high-level check of the deliverable tables to verify data completeness and data rea- sonableness, as well as a direct pavement distress compar- ison between the service provider automated ratings and manual ratings from experienced pavement raters. The process also helped identify several systematic errors (e.g., erro- neous classification of a particular distress type). The cor- rection of these errors resulted in an 83% reduction in the pavements requiring rehabilitation, and a 22% increase in pavements requiring no maintenance. The overall effect of these changes was a decrease of $18 million in the pavement maintenance recommendation for the Interstate Highway System (51). QUALITY MANAGEMENT PLANS A Quality Management Plan documents how the agency will plan, implement, and assess the effectiveness of its pavement data collection quality control and quality acceptance opera- tions. It describes the quality policies and procedures; areas of application; and roles, responsibilities, and authorities. The Quality Management Plan is a program-specific document that describes the general practices of the program. It may be viewed as the “umbrella” document under which individual quality activities are conducted. Figure 10 summarizes the state of the practice with respect to the use of formal data collection quality man- agement plans among state DOTs and Canadian provinces. Approximately one-third of the agencies (35%) already have a formal plan and an additional 27% are working on developing such a plan. It is interesting to note, however, that a large percentage of the agencies still do not have a for- mal approach for ensuring the quality of the data and 11% of the respondents did not know if a quality management plan existed. Yes 35% Under Develop. 27% No 27% Not Sure 11% Question: Does your agency have a formal pavement data collection quality management plan? FIGURE 10 Use of data collection quality management plans.

22 DATA MANAGEMENT ACTIVITIES For data quality management practices to be effective and efficient, quality management methods needs to be employed throughout the entire data collection process. Figure 11 sum- marizes the types of activities used for quality control and acceptance by the agencies that responded to the survey. It can be observed that several of the tools and methods used for quality control and acceptance are basically the same. This is probably one of the reasons why the two processes are often confused. However, the objective of the activities, the way they are conducted, and the personnel responsible for it are typically different in the two quality management phases as discussed in chapter one. The main techniques used by state and provincial DOTs for pavement data quality management are calibration of equipment and/or analysis criteria before the data collec- tion, testing of “control” segments before and during data collection, and software routines for checking the reason- ableness and completeness of the data. Similarly, 100% of the pavement data collection service providers indicated that they use equipment and/or analysis criteria before the data collection, and software routines for checking the rea- sonableness and completeness of the data, and most (86%) reported that they use testing of “control” segments before and during data collection. These tools are briefly intro- duced in the following sections and are discussed in detail in chapter four. Personnel Training and Certification: Continuous train- ing is very important to ensure that the personnel oper- ating the equipment or conducting the visual surveys are properly trained. That the classification of the dis- tresses is somewhat subjective makes training even more critical for the distress surveys. Some agencies require a formal “certification” of the pavement distress raters and equipment operators to verify that they have the required knowledge and skills. Equipment and Method Calibration, Certification, and Verification is to be conducted before the initiation of the data collection activities and periodically thereafter to verify that equipment is functioning according to expectations and that the collection and analysis meth- ods are being followed. Data Verification Procedures by Testing of Control or Verification Sites are used for both quality control and acceptance before and during production. Typical ver- ification techniques include periodic retesting of con- trol or verification pavement segments, oversampling or cross-measurements, and reanalyzing or resurvey- ing a sample of the sections measured by an indepen- dent evaluator. The locations of sections can be known or unknown (blind) to the data collection crews. 4% 24% 26% 38% 42% 47% 55% 57% 81% 94% 94% 12% 21% 27% 50% 50% 48% 61% 71% 71% 73% 80% 0% 20% 40% 60% 80% 100% Verification of sample data by an independent consultant Periodic testing of blind ìcontrol ” segments during production Cross-measurements (i.e., random assignment of repeated segments to different teams or devices) Statistical/software routines that check for inconsistencies in the data Comparison with existing time-series data Verification of the post-survey processing software/ procedures Software routines that check for missing road segments or data elements Software routines that check if the data is within the expected ranges Periodic testing of known “control” segments during production Testing of known “control” segments before data collection Calibration of equipment and/or analysis criteria before the data collection Quality Acceptance Quality Control FIGURE 11 Percentage of state and provincial agencies using each quality control and acceptance activity.

23 Software Data Checks are used during production for quality control, when the data are submitted for qual- ity acceptance, and when the data have been entered into the pavement management database. Typical checks include network-level checks for ratings that are out of expected ranges, checks for detecting missing segments or data elements, and statistical analyzes to check for data inconsistencies. Other Tools: In addition to the test described earlier, some agencies also conduct other tests, such as time- history comparisons, geographic information system (GIS)-based analysis, and verification of sample data by independent third parties. As previously discussed, the tools described can be included in the quality control plans, quality acceptance procedures, and/or independent verification processes. For example, the equipment calibration is a key component of the quality con- trol process, but the verification of this calibration is typically also included as part of the quality acceptance plans. Simi- larly, control or verification test sites are used by most agen- cies for both quality control and acceptance. QUALITY CONTROL Quality control includes actions and considerations necessary to assess and adjust production processes to obtain the desired level of quality of pavement condition data. These activities include checks on the equipment used to collect the data, the personnel responsible for the data collection, and the data col- lection process itself. When data are being collected using auto- matic data collection equipment, quality control may include equipment maintenance, testing, and calibration. Training and supervision of the survey crews is critical when data are col- lected using manual/visual surveys. Examples of quality con- trol activities also include data verification checks using control or verification sections, on-vehicle real-time data checks, peri- odic diagnostics/data checks, submitted data and video checks, distress rating data checks, and database checks. Before data collection, equipment is to be properly cali- brated, procedures clearly defined and documented, and per- sonnel trained. During data collection, it is important that pavement condition data be continuously monitored by a vari- ety of possible methods to ensure equipment calibration and data accuracy and consistency during the collection effort. This monitoring allows for errors to be detected and cor- rected before submission of large batches of unsatisfactory data. After data collection is complete, the data may then be validated before acceptance. QUALITY ACCEPTANCE Quality acceptance activities are those that govern the accep- tance of the pavement condition data; this is often referred to as quality assurance in the pavement data collection terminol- ogy. However, this latter term is not used in this synthesis because quality assurance in the quality management literature encompasses all the activities focusing on increasing the abil- ity to fulfill requirements for the product of service being pro- duced. The definition used in pavement engineering is closer to the definition provided by the National Quality Institute, which defines quality assurance as “actions taken by the buyer or user of the data to ensure that the final product is in compli- ance with the agreements, provisions, or specifications.” Quality acceptance tools are used for testing both the pave- ment condition data that are collected by the agency and those that are collected by a service provider. These tests validate that the data meet the establish requirements before they are used to support pavement management decisions. Examples of commonly used quality activities include control/verification site testing, complete database checks (e.g., to check for rat- ings outside of an expected range), sampling and retesting for quality acceptance, GIS-based quality checks, and time- history comparisons. Software programs used for quality management usually search for data that are missing, misidentified, incorrect with respect to segment size, improperly formatted, and/or outside of expected ranges. These programs ensure data complete- ness and functionality. For example, the state of Oklahoma uses a Microsoft Access-based program with Visual Basic modifications to perform such tasks (52). Use of such pro- grams not only ensures completeness, but standardization as well, reducing problems with time-history updates and allow- ing for better analysis of raw data into higher-level data such as composite indices. INDEPENDENT ASSURANCE Quality engineering practices typically recommend the inclu- sion of at least some degree of external audit in the quality management plan; this is called independent assurance. The purpose of the independent assurance testing is to validate the data for the user agency. Data checks by quality acceptance personnel are intended to ensure the accuracy of the data. Such checks typically involve making sure that distresses are prop- erly identified and severity is properly evaluated. The check can be done by the data collection personnel, by someone else internal to the organization, or by an external third party. For example, a procedure to verify the quality of the pavement data collection during production is the use of a sample of a “con- trol” section that is resurveyed or reanalyzed by an indepen- dent evaluator and the results compared with the production ratings. The reference measurements on these sections are determined using the best available practical technique for that particular pavement condition indicator. The survey showed that only 4% of the agencies use independent verification for quality control and 12% for quality acceptance. REFERENCE VALUES/GROUND TRUTH The determination of the correct value for pavement condi- tion data is a particularly challenging task that has received significant attention. This reference value or “ground truth”

is typically determined using the “most appropriate” method- ology available, which is done by using trained and experi- enced raters for pavement surface distress and a calibrated or certified piece of equipment for sensor-based measure- ments. Because the most appropriate methodology is some- times difficult to determine, some agencies prefer to use the term reference value instead of ground truth. Ongoing studies may provide additional insights on this issue. For exam- ple, the Strategic Highway Research Project 2 (SHRP 2), S03: Roadway Measurement System Evaluation, is developing, organizing, and conducting a roadway measurement accuracy evaluation of mobile road and pavement inventory services collected at highway speeds (53). The recent National Work- shop on Highway Asset Inventory and Data Collection (54) compared automatic systems for inventorying roadway geom- etry and roadside element data. For surface distress, the control sections are typically eval- uated using visual surveys or by independently analyzing the images captured by pavement evaluation equipment. Rutting and faulting are typically determined using a straight edge and rut wedge or a static inclinometer, and faulting using a straight edge and a ruler or the Georgia Faultmeter. Statistical meth- ods are typically used in conjunction with control site testing to establish acceptable ranges for various data collection tech- niques. The ground truth smoothness is currently determined based on the profile measured using rod and level, the static inclinometer, or “walking” profilers; however, research is underway to find a more appropriate reference profiler (55). To compare friction measurements obtained with differ- ent types of friction equipment, the Permanent International Association of Road Congresses (PIARC) has devised the IFI (56). ASTM currently uses the Dynamic Friction Tester (DFTester) and Circular Track Meter (CTMeter) as the ref- erence devices for the IFI. The index is composed of two numbers, the friction value at 60 km/h (F60) and the change of friction with speed (sp) (57). SOURCES OF VARIABILITY IN PAVEMENT CONDITION DATA COLLECTION One of the purposes of the quality control, and to some degree quality acceptance, processes is to reduce the variability in the pavement condition measurements. Thus, it is important to understand the various sources of variability before design- ing pavement control and acceptance programs. Because there is inherent variability in the pavement condition data that are collected, it is important that agencies understand the magnitude and sources of variability in the data being col- lected to be able to compare results and establish target control and acceptance level for quality management. For example, Stoffels et al. (58) proposed a process to identify acceptable ranges for comparing results from two independent sources using standard variability control concepts with pavement data collection. 24 In general, sources of variability for pavement condition data collection can be related to the equipment used, equip- ment operation (including rater/operator training and skills), processing of the data collected, environmental conditions, and shape and condition of the pavement surface. All of these potential sources have to be considered and, if possible, con- trolled because they will affect the quality of the data collected. Although the potential sources of variability are many, this sec- tion summarizes the primary sources that can be controlled during the data collection process. Calibration and/or valida- tion before, during, and after data collection are necessary to ensure accuracy, given the possible variations between dif- ferent devices and operators, as discussed in chapter four. Surface Distress Sources of variability for manual and automated methods of data collection are similar and generally involve distress iden- tification and classification, as well as assigning distress sever- ity levels. McGhee (6) classified the automated distress data in distresses collected with sensors (e.g., smoothness, rutting, and faulting) and distresses obtained from processing of pavement images (e.g., cracking). Because some of the sources of vari- ability for sensor-based (rutting) and image-based (cracking) distresses are in general different, they are presented in two separate sections. Typically, both manual and automated crack detection methods display a noticeable bias toward detecting higher-severity distresses, while missing lower-severity dis- tresses. This is usually the result of higher-severity distresses being more readily identifiable than medium- or lower-severity distresses (18). Cracking Cracking is measured from a moving vehicle or “walking” the section. The evaluations from the moving vehicle can be done manually (“windshield” evaluation) or automatically by processing images of the pavement collected by a pavement distress data collection vehicle. The processing of the images is done manually in semi-automated data collection and auto- matically by the processing software in the fully automated process. Sources of variability for automated, semi-automated, and manual distress include (59): • Type of equipment/data collection method – Image quality—for automated and semi-automated surveys.  Type of technology used; for example, analog images, digital images, laser-based images.  Resolution of the imaging equipment (detection of smaller cracks require higher resolution equipment).  Field of view (distresses may be missed if they do not cover entire lane).  Quality of the color contrast of the pavement image (color contrast between the crack and the

25 surrounding pavement is an important factor when distress software programs are evaluating the sever- ity of the distress).  Lighting method. – Rater’s vision—in case of windshield surveys. • Raters/equipment operator training—The experience and understanding of rating protocols is paramount to reduce the variability of the collected data. • Processing software—The algorithm used to detect and quantify the various types of cracks is critical in the case of automated surveys. For example, one common prob- lem with an automatic cracking detection algorithm is the classification of the pavement shoulder joint as a longitudinal crack. • Measurement environment—The conditions under which the distress surveys are conducted will affect the detec- tion of the cracks. For example, cracks are typically more visible soon after rain because they may be filled with water. The severity of the extent detected under such circumstances may be different. Rutting Most agencies are currently measuring rut depth (transverse profile) with some type of automatic data collection equip- ment, including rut bars with multiple sensors or continuous measurement systems. In general, the sources of variation include the following: • Type of equipment – Type of sensors—Common sensors include point laser, ultrasonic, and continuous scanning lasers. “Point” lasers are currently the most commonly used sensors. Continuous scanning lasers can cover a total width of up to 13 ft with a resolution of 1,280 points. These sensors are very accurate; however, the user needs to input the width of the lane to allow the soft- ware to exclude curbs or edge drop-off. Ultrasonic sensors are still used in some systems, but they can be affected by temperature and moisture. – Rut bar width (and lane coverage)—Even with a large number of sensors, most older (and still used) rut bars will only cover a certain width of the lane. Further- more, even the extendible rut bars will almost never be fully extended for safety reasons. Typically, the rut bar is only extended to 10 ft. Some rut bars have the two most lateral sensors angled out to increase the width of coverage. – Number of sensors—The number of sensors can range from 3 to 37, to continuous, with continuous systems becoming more common. – Distance measuring system. • Equipment operation – Wheel path wander affects noncontinuous measure- ment because it affects the data points collected. – Edge drop-off and/or narrow lanes—The far right sensor may pick up a drop-off or grass on the right shoulder, which will affect the measurements. – Equipment driver and/or operator—The experience, training, and driving skills of the data collection per- sonnel will affect the measurements and thus the qual- ity of the data collected. • Rut depth calculation method is to be controlled to ensure consistency from year to year or from service provider to service provider because there are different algorithms for processing the transverse profile and calculating rut depth. The main methods currently in use are the string- line or “wire” method [AASHTO PP-38-00 Standard Practice for Determining Maximum Rut Depth in Asphalt Pavements (2005)] and the straight edge. The stringline method allows an imaginary line to bend at the hump between the wheel paths if the hump is higher than the outside and centerline of the road. The straight edge method projects a straight line across from the inside to the outside of the lane and results in lower rut depth cal- culations than the stringline method. • Measurement environment – Temperature, wind, humidity, and surface moisture affect the various types of sensor differently and can add to the variability of measurements. – Presence of pavement contaminants, such as sand, gravel, etc. – Lighting conditions affect optical sensors. • Surface texture—High-textured surfaces, such as open- graded friction courses and chip seals, can affect the sensor readings It is generally accepted that rut bars with a greater number of sensors (or transversal measurements points in the contin- uous systems) yield more accurate and consistent measure- ments. When changing from the older style rut bar to a new scanning laser, the Oklahoma DOT (ODOT) found that the rut calculations were usually deeper but closer to manual measurements. Older rut bars could under-report rut depth because of a lack of full-lane-width coverage. Smoothness Given the large variety of devices commercially available, smoothness measurements of the same pavement segment can show significant variation from device to device (effecting reproducibility). However, measurements by the same device are generally repeatable. The main factors that affect variabil- ity of smoothness measurements include the following (60): • Type of profiler—The various profilers commercially available use different technologies, sensors, and signal processing techniques. – Height sensor—Most current profilers use a laser sensor; however, some agencies still operate profilers with ultrasonic or infrared sensors. Important sensor

characteristics include sampling rate, resolution, foot- print, and range. – Sensor footprint—Although traditional laser profil- ers use point lasers, agencies are starting to require sensors that have a wide footprint. – Accelerometer—The type (range) and location of the accelerometers may affect the measured profile and processed indices. – Distance measurement system. – Number of sensors, sensor location, and spacing— when multiple sensors are used. • Profiler operation—the manner in which the profiler is driven. – Profiler driver and operator—experience, training, and driving skills. – Lateral position—Wheel path wander is a big source of variation; thus, operators are trained to stay in the center of the lane. – Longitudinal positioning/triggering. – Measurement speed—Although it has been hypothe- sized that measurement speed has an influence on the measured smoothness (61), some of the latest research indicates that most profilers produce measurements that are stable with respect to the measuring speed (55). – Lane measured—Although most agencies measure only the outermost lane, others are starting to measure the profile on all lanes. – Tire inflation pressure affects longitudinal distance measurements. – Calibration of the various components of the equipment. • Profile data interpretation and processing – Filters—Most profilers use filters to eliminate un- wanted high and low frequencies in the measured profile; although some allow the user to select the fil- ters, others do not. – Profiler computation algorithm—The algorithm used to combine the output from the key component sen- sors and determine the profile. – IRI calculation algorithm and procedure—for exam- ple, some profiler manufacturers automatically apply nonstandard filters to the profile. Other manufactur- ers and states may choose to average the profiles from the left and right wheel paths before applying the IRI algorithm, thus generating a half-car rough- ness index instead of an IRI from a single wheel path or an average of two wheel paths of IRI. – Integration interval—The length of the segment over which the smoothness is reported is important because the profile elevation data are aggregated. A relatively large sum of elevation values can indicate a pavement that is moderately rough over the entire segment or very rough over a small section of the entire segment. Measurements over smaller segments tend to yield more useful results because short, rough areas are detected and might be unnoticed if the segments were larger. 26 – Wheel path measured—Although some agencies report one of the wheel paths, others compute the average of both wheel paths. This is significant because measurements from the outer (right) wheel path are generally rougher than those in the inner wheel path (20). – Bridges—There is lack of agreement on how to deal with bridges included within the considered road seg- ment. These bridges are often localized areas of high roughness. Some agencies include bridges as part of the road segment because this better reflects the actual user’s perception for the overall road segments. Other agencies do not include bridges to avoid artificially high estimates of the pavement IRI in the segments. • Measurement environment – Temperature, wind, humidity, and surface moisture affect the various types of sensor differently and can add to the variability of measurements. – The presence of pavement contaminants, such as sand, gravel, etc. – Lighting conditions affect optical sensors. • Surface shape—The condition and texture of the sur- face affect the accuracy and repeatability of profilers. – Surface distresses have a major influence on trans- versal variations of the profile. – Daily and seasonal profile variations are caused by curling of PCC slabs, moisture changes in the sub- grade, freeze and thaw cycles, etc. Differences as high as 0.4 m/km have been observed in some sites. – Road geometrics—Cross slope, curves, hill, and grades can affect the output of accelerometers, which are key components of inertial profilers. Surface Friction Properties Typical sources of variation in friction measurements include the following. • Equipment used—Most state DOTs use the locked- wheel skid trailer (ASTM E274) for high-speed friction measurements (62). The system includes a truck with a water tank and a trailer system that can lock one of the wheels for measuring the friction coefficient between tire and wet pavement. The wet pavement friction (or skid) number is reported as 100 times the coefficient of friction (63). Some of the equipment has incorporated laser- based systems to determine the surface macrotexture and use it to estimate the gradient of friction with speed. • Operation – Type of tire—Some states (e.g., Virginia) use smooth tires and others (e.g., Florida) still use ribbed tires. – Testing speed—Some states have equations for cal- culating the friction number from skid testing at other speeds, but those correlations have been only locally verified (61).

27 – Equipment calibration—Calibration is usually per- formed once at the beginning of the data collection season (or every two years), with verification testing taken at control segments during data production. • Measurement environment—Annual variations in pave- ment friction have been detected; therefore, annual test- ing of the network is recommended (61). Temperature has been shown to be one of the contributing factors (64). Some states use seasonal (typically monthly) correction factors for the measured friction. Structural Evaluation Sources of variability on FWD measurements include the following: • Equipment used—Different FWDs use slightly differ- ent configurations and sensing technologies. In addi- tion, available devices use a variable number of sensors. • Equipment operation/testing protocol – Load—Most agencies use 9,000 lb, whereas some add additional tests at variable load levels to assess the nonlinearity of some of the materials. – Sensor spacing—Several devices and testing pro- tocols use different sensor spacing; however, the LTPP data collection guidelines have helped stan- dardize them. – Sitting errors—Some protocols call for one or two initial drops to improve the contact between the plate and the pavement and the repeatability of the measurements. – Equipment calibration is important to correct sys- tematic errors in the sensor measurements and there are regional sites that have been established for this purpose, as discussed in chapter four. – Testing spatial frequency—Although closely spaced tests are preferred for accurately assessing the struc- tural capacity of the pavements, this reduces produc- tivity and increases cost. – Lane tested—Testing in Indiana showed that structural capacity in one direction of an Interstate highway was nearly identical to the capacity in the opposing direc- tion. However, non-interstate roads showed more vari- ability and more variable results can be expected in highways where opposite lanes were constructed at dif- ferent times and/or with different structural materials and designs. • Measurement environment—Environmental factors have substantial influence on the pavement response and thus on the measured deflections – Temperature affects the stiffness of the asphalt-based materials. Deflection adjustment factors have been developed at the national (LTPP) and state level to account for temperature variations. – Moisture—The presence of water affects the bearing capacity of soils and unbounded pavement materials. • Interpretation of the results – Type of analysis conducted—the information col- lected can be used for assessing the overall structural capacity of the pavement (e.g., computing a surface modulus of an effective Structural Number or deter- mining the moduli of the various pavement layers using backcalculation). – Software used—the backcalculation software used typically affects the resulting layer properties. Ground Penetrating Radar Some agencies have used GPR to determine layer thickness data. This device operates by using electromagnetic waves to identify and locate interfaces between layers within the pave- ment, which in turn allows for determination of layer thick- nesses. For GPR to distinguish layer separations, the pavement layers must have different dielectric properties. Additionally, higher frequency waves yield better resolution, whereas lower frequencies allow for further penetration into the pavement, resulting in upper layer profiles being more accurate than those of lower layers (40). Testing has shown that GPR provides accurate layer thicknesses if calibrated with just a few cores, and the GPR measurements can be used to influence future coring requirements (65). EFFECTS OF NETWORK SIZE ON QUALITY MANAGEMENT The size of the network appears to have an effect on the quality management procedures. For example, it takes considerably more effort to conduct quality assurance checks for large net- works than for smaller ones. Because large agencies have to review large amounts of data they appear to be more motivated to develop formal quality management plans. Figure 12 shows that large agencies (e.g., with more than 25,000 lane-miles) were noticeably more likely to have a formalized quality man- agement plan, or have one under development, than were the small agencies. This is logical given that agencies with larger networks would receive higher quantities of data, and the occurrence of systematic errors could result in large quantities of poor data. Larger networks are also more costly to maintain on an annual basis, making development of new and improved quality management methods more cost-effective. Network size also appears to play a minor role in the condition data quality management process. Agencies with larger networks (e.g., more than 25,000 lane-miles) under their management collected data less frequently than agen- cies with fewer than 5,000 lane-miles. This trend held not only for highways, but arterials, collector, and local roads as well. The same was true for smoothness data collection; agencies with larger networks generally collected data less often than those with smaller networks. This is illustrated in Figure 13, which summarizes the average (considering the various functional categories) data collection frequency

80% 56% 73% 39% 62% 20% 36% 27% 61% 34% 0% 4% 0% 0% 2% 0% 20% 40% 60% 80% 100% Less than 5,000 miles Between 5,000 and 10,000 miles Between 10,000 and 25,000 miles Between 25,000 and 50,000 miles More than 50,000 miles Less than 5,000 miles Between 5,000 and 10,000 miles Between 10,000 and 25,000 miles Between 25,000 and 50,000 miles More than 50,000 miles Every 4 years or more Every 2 to 3 years Once a year (a) Surface Distress 75% 63% 70% 41% 56% 25% 25% 30% 59% 36% 0% 13% 0% 0% 8% 0% 20% 40% 60% 80% 100% Every 4 years or more Every 2 to 3 years Once a year (b) Smoothness FIGURE 13 Temporal data collection frequency for (a) surface distress and (b) smoothness as a percent of agencies collecting the pavement condition indicator. 28 25% 0% 29% 55% 58%0% 50% 35% 36% 17%75% 50% 35% 9% 25% 0% 20% 40% 60% 80% 100% Less than 5,000 miles Between 5,000 and 10,000 miles Between 25,000 and 50,000 miles Between 10,000 and 25,000 miles More than 50,000 miles No No; but under development Yes FIGURE 12 Percentage of agencies having a quality management plan as a function of network size.

29 as a percentage of agencies collecting surface distress and smoothness. SUMMARY “Good” pavement condition data are very important in pro- viding effective pavement management. Pavement condition data collection quality management is necessary to ensuring that the collected data meet the requirements of the pavement management system. This chapter presented the main data quality management concepts and principles as they apply to pavement condition data collection. Effective data collection quality management programs provide a comprehensive, systematic approach to data collec- tion and processing. It is important that a complete pavement condition data quality management system include a clearly documented quality control plan, detailed quality acceptance procedures, and established guidelines to monitor the entire process, with timelines, milestones, and evaluation criteria. The Quality Management Plan is the “umbrella” document under which individual quality activities are conducted. Approximately one-third of the DOTs (35%) already have a formal plan and an additional 27% are in the process of developing such a plan. The main techniques used for pavement data quality man- agement are calibration of equipment and/or analysis criteria before the data collection, testing of “control” segments before and during data collection, and software routines for checking the reasonableness and completeness of the data. These tools can be included in the quality control plans, quality acceptance procedures, or for independent assurance. Quality control includes those activities necessary to assess and adjust production processes to obtain the desired level of quality of pavement condition data. Included are checks on the equipment used to collect the data, the per- sonnel responsible for the data collection, and the data col- lection process itself conducted before, during, and after the data collection. Quality acceptance includes those activities conducted to verify that the collected pavement condition data meet the quality requirements. Quality acceptance tools are used for testing both the pavement condition data that are col- lected by the agency and those that are collected by a ser- vice provider. Common methods include testing of controls or verification sites, use of software to check for errors such as incorrect asset data or ratings outside of an expected range, and checking a certain percentage of data by quality assurance personnel. The independent assurance testing aims at validating the data for the user agency. For example, a procedure to verify the quality of the pavement data collection during production is the use of a sample or “control” section that is resurveyed or reanalyzed by an independent evaluator and the results compared with the production ratings. In general, sources of variability for pavement condition data collection can be related to equipment used, operation (including rater/operator training and skills), processing of the data collected, environmental conditions, and shape and condition of the pavement surface. All these potential sources have to be controlled (or at least accounted for) because they will affect the quality of the data collected.

Next: Chapter Four - Quality Management Practices »
Quality Management of Pavement Condition Data Collection Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Synthesis 401: Quality Management of Pavement Condition Data Collection explores the quality management practices being employed by public highway agencies for automated, semi-automated, and manual pavement data collection and delivery.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!