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.
10 Overview In the recently published Pavement Asset Management, the authors documented the progression of pavement management from âearly rudimentary efforts in the 1960s to a comprehensive technology, economic, and business-based processâ today (Haas et al. 2015). Their work references the initial studies of pavement performance that took place at an AASHTO-sponsored Road Test in Ottawa, Illinois, and the researchersâ efforts to turn those measurements into an index representative of field conditions. The early developers of pavement management recognized the importance of managing and designing pavements from a systems perspective, which led to the eventual development of the field of pavement management. The rapid advancements in computer technology since the late 1980s have had a tremendous impact on the availability of computerized pavement management systems for agencies of all sizes and the methodologies being used to collect and analyze data. The organizational, political, and societal changes that have taken place since that period have also significantly shaped the use of pavement management in transportation agencies. Pavement Management Decision Support Pavement management is used to support agency decisions at three different levels (AASHTO 2012). At the highest level, referred to as the strategic level, decisions traditionally focus on investment levels and strategies that enable an agency to achieve its goals and objectives. The ability to forecast future conditions and illustrate the consequences of deferred investment are key to being able to sup- port decisions at this level. Second, at the network level, summary information related to the entire highway network is used to identify the most effective mix of projects and treatments for a multi- year improvement program. At this level, it is important to be able to evaluate the costs and benefits of different combinations of projects and treatments on current and future conditions. At the third project level, decisions are focused on a particular segment of the pavement network. The informa- tion needed from the pavement management system to support project-level analyses is typically more detailed than the data used at the strategic or network levels, and it focuses more on in-place conditions. Investigations into the causes of a particular pavement section that is not performing as expected is an example of a project-level analysis. Pavement Management Components A pavement management system supports these different types of analyses through the data, analysis, and reporting components depicted in Figure 1. These components include various inputs that are stored in a database for use in the analysis and reporting modules. Databases to support pavement management may range from a spreadsheet at the most basic level to a relational database or an agency-wide data warehouse. The sophistication of pavement management analysis parameters varies based on the type of pavement management software used, influencing the extent to which pavement performance prediction models can be customized, establishing rules to define the applicability of different types of treatments under different conditions, and tailoring treatment costs and impacts to agency conditions. These parameters are used in the analysis module to determine the funding level needed to achieve a targeted performance level, identify the most effective combination of treatments under a constrained funding scenario, or predict future conditions under different invest- ment strategies. These and other types of outputs are generated in the reporting module in a variety chapter two Literature review
11 of different formats. The final component of a pavement management system, the feedback loop, is intended to ensure that the projects and performance trends from the field are input back into the pavement management system to keep the database current and to update the analysis parameters. evolution of Pavement Management Although the components of a pavement management system have not changed dramatically over time, the sophistication of those systems and the extent to which pavement management information is used to support decisions has evolved. Initially, pavement management systems were used primar- ily to document pavement conditions and estimate current funding needs. In a 1987 synthesis on pavement management, Pavement Management Practices, the authors documented that the primary outputs from a pavement management system were pavement condition, prioritized listings, deficien- cies, and treatment needs and costs (Peterson 1987). Weaknesses in the systems at that time included the inability to estimate life-cycle costs, predict conditions, and integrate pavement management data with other data systems in the agency (Peterson 1987). A subsequent synthesis, Current Practices in Determining Pavement Condition, focused on pave- ment data (Gramling 1994). At that time, approximately 85% of the agencies responding to a survey were using mainframe computers to store their pavement management data and all agencies either had, or were in the process of developing, a pavement management system (Gramling 1994). Gramling notes a trend toward the use of the International Roughness Index (IRI) in response to Highway Performance Monitoring System (HPMS) reporting requirements that were issued by FHWA. The synthesis reported that in 1989 eight states used the South Dakota Road Profiler to collect roughness data, but 5 years later 24 states were using the device (Gramling 1994). In 1995, another NCHRP synthesis, Pavement Man- agement Methodologies to Select Projects and Recommend Preservation Treatments, documented the use of three different methodologies in pavement management: a pavement condition analysis, priority FIGURE 1 Pavement management components (AASHTO 2012).
12 assessment models, and network optimization models (Zimmerman and ERES 1995). At that time, 29 of 56 responding agencies indicated that projects were prioritized based on a condition-ranking method, and only 12 agencies used a benefitâcost analysis to prioritize projects. The ability to forecast future conditions was again noted as a shortcoming of pavement management systems at that time. Two other syntheses of practice focused on data collection practices and methods to manage data quality (McGhee 2004; Flintsch and McGhee 2009). These documents reflect the transition that was occurring at that time as transportation agencies shifted from manual surveys conducted using in-house staff to more automated pavement condition surveys often conducted by a contractor. NCHRP Synthesis 334: Automated Pavement Distress Collection Techniques documents the incon- sistencies that existed in the automated data collection processes at that time and the attempted to develop standards to address these issues. With the exception of roughness, the synthesis reports that few agencies were using the provisional standards developed by AASHTO (McGhee 2004). The study concluded with a recommendation to address data quality management, which became the focus of the 2009 synthesis Quality Management of Pavement Condition Data Collection. This synthesis pre- sented an example of a quality management plan that could be used to improve quality, even though only one-third of the state and provincial highway agencies that responded to the report survey had that type of plan in place (Flintsch and McGhee 2009). Methods of sharing pavement management data using geographic information systems (GIS) were the focus of NCHRP Synthesis 335: Pavement Management Applications Using Geographic Informa- tion Systems (Flintsch et al. 2004). The practices documented in the synthesis reflect the enhanced spatial referencing technologies that were becoming available to improve data collection and data integration practices. However, most agencies that indicated they were using GIS to support pavement management were only using their GIS to prepare maps and graphic displays (Flintsch et al. 2004). At that time, the use of GIS to integrate data or to serve as an enterprise-wide database was still in its infancy. As transportation agencies have faced growing competition for funding and observed significant growth in the use of their pavement networks, it has become increasingly important to use available funding as cost-effectively as possible. In addition, recent federal legislation has placed a greater empha- sis on performance-based investment decisions that improve agency accountability and transparency. Together, these factors have led to a growing need for sophisticated and reliable pavement management systems. In 2008, recognizing that not all state DOTs have pavement management systems capable of addressing these demands, FHWA began sponsoring a series of regional peer exchanges that provided an opportunity for pavement management practitioners to share experiences and develop the skills nec- essary to enhance their pavement management capabilities. The pavement management peer exchange program continues to this day and has proven to be a popular method of transferring technology. In 2010, FHWA sponsored the development of a Pavement Management Roadmap designed to identify the existing gaps in pavement management practices at the state DOT level and identify research and development initiatives and priorities. The resulting Roadmap set the following vision for pavement management and identified more than $14.5 million in short- and long-term initiatives needed to advance pavement management practices (Zimmerman et al. 2010). The 2020 Vision for Pavement Management Pavement management will make use of a new generation of technology so agencies are less dependent on manual labor for data collection. Pavement management tools will allow agencies to communicate effectively with stakeholders, using clear statements that are tied to agency goals and pavement worth. Within an asset management framework, pavement management will be used for investigating deci- sions and program options in both private and public sectors. A pavement management analysis will consider new materials and construction/design practices, as well as other factors that influence project and treatment selection, including safety, congestion, and sustainability. As a result of these changes, pavement management will be robust, comprehensive, and credible, and will address agency needs at the project, network, and strategic levels.
13 The short- and long-term needs included in the Pavement Management Roadmap were organized into the following four themes: ⢠Theme 1âUse of Existing Technology and Tools: The problem statements in this area include recommendations for technology and tools to support traditional applications of pavement management. ⢠Theme 2âInstitutional and Organizational Issues: The recommendations in this theme address issues related to workforce development, communication, contracting, and organiza- tional structure. ⢠Theme 3âThe Broad Role of Pavement Management: The problem statements in this area focus on the expanded application of pavement management for purposes related to pavement design, asset management, and load impact studies. ⢠Theme 4âNew Tools, Methodologies, and Technologies: The recommendations in this theme are intended to lead to the development of new tools, methods, and technology to support the evolving role of pavement management. Since the development of the Roadmap, pavement management conferences and publications have continued to document advances that have taken place in the types of data being collected, the method of collection, and how pavement management information is being used to support agency decisions. The program for the most recent International Conference on Managing Pavement Assets, held in Alexandria, Virginia, from May 18 to 21, 2015, reflects these changes. Conference sessions included innovative topics such as multi-objective analysis approaches, accelerated testing and instru- mentation, asset valuation, integration of sustainability-rating tools in pavement management, improved performance modeling techniques, use of pavement management data to calibrate pavement design programs, the availability of web-based software tools, and the development of performance mea- sures and compliance specifications for use in highway concessions and long-term performance- based contracts. Plenary sessions focused on using pavement management data in response to natural disasters and the design and construction of more sustainable and durable pavements. Based on the information presented at the conference, the field of pavement management has advanced consider- ably over the past decade. Current Legislation Supporting Pavement Management Today, legislation is in place that requires state DOTs to use a pavement management system to sup- port the development of their Transportation Asset Management Plan. According to the Notice of Proposed Rulemaking (NPRM) published in the Federal Register (Vol. 80, No. 34, Feb. 20, 2015), a pavement management system should include, at a minimum, formal procedures for: ⢠Collecting, processing, storing, and updating pavement inventory and condition data. ⢠Predicting changes in pavement condition over time. ⢠Evaluating the costs and benefits of alternative investment strategies. ⢠Estimating short- and long-term budget needs. ⢠Determining optimal improvement programs. ⢠Recommending strategies to manage pavements under constrained conditions. In addition, the NPRM for Assessing Pavement Condition for the National Highway Performance Program (Federal Register, Vol. 80, No. 2, Jan. 5, 2015) requires each state DOT to develop and use an FHWA-approved Data Quality Management Program to assess the quality of all data collected to report pavement condition metrics. The Data Quality Management Program, at a minimum, is to include methods and processes for: ⢠Data collection equipment calibration and certification. ⢠Certification processes for persons performing manual data collection. ⢠Data quality control measures to be conducted before data collection begins and periodically during the data collection program. ⢠Data sampling, review, and checking processes. ⢠Error resolution procedures and data acceptance criteria.
14 At the time this synthesis was written, the final rules had not been issued for either of the NPRMs. Therefore, some of the minimum requirements listed previously may change; however, the overall intent of the rules is not expected to change significantly. Benefits to the use of Pavement Management Agencies that have utilized pavement management systems have recognized a variety of different types of benefits, including those listed here (AASHTO 2012): ⢠The effective use of available resources to improve pavement performance. ⢠The ability to justify funding needs. ⢠An understanding of current and projected pavement conditions and needs. ⢠Improved access to pavement information throughout the agency. ⢠Increased accountability and transparency in the decision process. ⢠Objective decision making based on data. The actual benefits that are realized are influenced by factors such as the comprehensiveness and quality of the data, the degree to which agency decisions are affected by the pavement management recommendations, and the capabilities of the software tools. At least two studies have attempted to quantify the benefits to using pavement management strategies. In one study, Hudson et al. (2000) reported that the Arizona DOT realized a savings of at least $30 in agency costs for every $1 spent on the development, implementation, and operation of their pavement management software by using an optimized set of treatments. The authors reported that if user costs had been considered, the sav- ings would have approached $250 for each dollar spent. The second study, conducted by the Ministry of Transportation in Alberta, found that over a 5-year period the return from pavement management was 100 to 1 as a result of changes from fixing the worst roads first to a strategy that included a mix of more cost-effective treatments (Cowe Falls et al. 1994). aPPrOaCheS tO PaveMent ManageMent Each agency must decide for itself the most effective approach to use for pavement management based on its organizational needs, the size of the organization, the available resources to support data collection and analysis, the minimum requirements established by FHWA, and the level of support available from executive leadership. Once these types of factors have been evaluated, a transporta- tion agency can develop a plan for collecting pavement inventory and condition information, select- ing pavement management software, and gathering the information needed to develop treatment rules and costs. In addition to pavement management systems that are developed in house for an agencyâs own use, there are at least two different categories of pavement management systems that are publicly available. These two categories of software vary significantly in terms of cost, complexity, and flexibility. The simplest software programs have been developed using public funds so that the soft- ware is available at little to no cost. These programs are classified as public domain software. The programs typically provide relatively simple, but effective, database and analysis tools that are used primarily at the local level. Examples of public domain software programs include Wolters et al. (2011) and AASHTO (2012): ⢠PAVERâdeveloped by the U.S. Army Corps of Engineers and distributed by the American Public Works Administration. ⢠StreetSaverâdeveloped by the Metropolitan Transportation Commission in the San Francisco Bay area. ⢠RoadSoftâdeveloped by the Center for Technology and Training at Michigan Technological University. The second category of pavement management software is referred to as proprietary software because the programs are developed, maintained, and licensed by private corporations. These programs
15 typically provide more flexibility in terms of configuring the database, the analysis parameters, and the reporting features. The additional flexibility is often accompanied by more sophisticated modeling approaches; however, these additional features come at a higher cost than the public domain software programs. The proprietary software programs are more likely to be found in larger DOTs. In a Pave- ment Management Catalog published by FHWA in 2008, there were 12 different proprietary software programs listed and four public domain systems (FHWA 2008a). The type of pavement management software used and the methods selected for determining pave- ment conditions has a significant impact on the extent to which pavement management is used to support agency decisions. The remainder of this chapter highlights some of the different approaches being used in pavement management, focusing primarily on: ⢠Methods used to ensure data quality. ⢠Methods used to develop pavement management prediction models and decision trees. ⢠Methods of assessing pavement performance trends and treatment cost-effectiveness. ⢠Uses of pavement management data for estimating funding needs, predicting network perfor- mance, setting performance targets, developing work plans, and allocating funding. ⢠Processes for developing work programs, including comparisons between planned and actual work activities. ⢠Strategies for presenting pavement management data and analysis results. Methods used to ensure Data Quality The quality of the pavement condition data contained in the pavement management system is criti- cal for producing informed decisions because it serves as the basis for all recommendations that are generated (Flintsch and McGhee 2009). Recent efforts have defined quality as âthe degree to which data conforms with a given requirementâ (AASHTO 2011) and focused on improving pavement condition data quality through the development of a Quality Management Plan that documents an agencyâs process for managing data quality (Pierce et al. 2013). There are a number of factors that influence the quality of pavement condition data, including both the survey methodology being used and the manner in which the data are collected. The manner in which the survey results are used influences the level of quality that must be attained and current trends at outsourcing data collection activities have introduced new issues related to the consistency in data when vendors or equipment change. Errors in data can have a significant impact on the rec- ommended treatments and budgetary requirements generated by the pavement management system. At the network level, systematic errors are considered to be especially critical to address because of the large volume of data collected and the potential for these errors to be compounded (Shekharan et al. 2007). One study quantified the impact of systematic errors in pavement condition on system outputs, reporting that a 10% error in the distress score can over- or underestimate the annual budget needs by as much as 85% (Saliminejad and Gharaibeh 2013). The study further confirmed that sys- tematic errors have a higher impact on pavement management outputs because the entire network is impacted (Saliminejad and Gharaibeh 2013). The typical types of pavement condition data collected by state and provincial transportation agencies in 2009 are shown in Figure 2 (Flintsch and McGhee 2009). This information is based on the responses from 46 state DOTs and nine Canadian provinces. At that time, most agencies were collecting surface distress and smoothness data at the network level. Fewer agencies were collecting structural capacity and surface friction properties at the network level. Correspondingly, at that time more than 50% of the agencies that responded to the survey were collecting the majority of their pavement condition data using in-house staff; however, agencies were increasingly using contractors to provide sensor-measured data for smoothness, rut depth, and joint faulting (Flintsch and McGhee 2009). Today, the use of contractors for collecting pavement condi- tion data is much more widespread. As evidenced from information provided during workshops con- ducted by FHWA to promote the Practical Guide for Quality Management of Pavement Condition Data Collection, 37 of the 50 states that attended the workshops between 2014 and 2015 indicated
16 that they are currently using automated and semi-automated processes for collecting and reporting pavement condition information (excluding friction), as shown in Figure 3 (http://www.fhwa.dot. gov/pavement/mana.cfm). As shown in Figure 4, the data continues to be collected primarily using in-house personnel, but a large number of states use contractors to collect the data. Figure 5 shows the frequency of network-level pavement condition surveys. Most states reported that they collect data on a portion of the network annually, but the rest of the system is surveyed every 2 or 3 years. Twenty states reported that they annually collect pavement condition information on their entire sys- tem. The practices used by state agencies are largely influenced by the size of the state-maintained network as well as the availability of in-house personnel to collect the data. In recent years, some states have extended their data collection activities to include roads maintained by local agencies to promote statewide consistency in the way pavement conditions are reported. In the absence of FIGURE 2 Types of pavement condition data collected (Flintsch and McGhee 2009). 0 5 10 15 20 25 30 35 40 Automated Automated Sensor Data & Manual Distress Manual Number of States FIGURE 3 Data collection methods used (out of 50 states) (http://www.fhwa.dot.gov/pavement/mana.cfm).
17 state-provided data, local agencies may elect to collect pavement condition data for their own road network independently. Transportation agencies have taken a variety of approaches to address data quality efforts. The primary techniques used by state and provincial transportation agencies include calibration of equip- ment and/or analysis criteria before data collection starts, testing of control sections before and dur- ing data collection, and software routines for checking the reasonableness and completeness of the data (Flintsch and McGhee 2009). Training is also used extensively, especially for pavement distress surveys. Some agencies require a formal certification of raters and/or equipment operators as a way of verifying that field crews have the skills and knowledge required to help ensure data quality (Flintsch and McGhee 2009). Virginia DOT hired a third-party, independent contractor to manually check 10% of the data collected and analyze using automated methods. The process identified systematic errors that included misclassifications of a particular distress type. Once these errors were addressed, the number of pave- ments requiring rehabilitation decreased by 83% and an additional 22% of the pavements were found to require no maintenance, resulting in an $18 million reduction in treatment needs for the Interstate 0 5 10 15 20 25 Inhouse Contractor Combination Number of States FIGURE 4 Responsibility for data collection (out of 50 states) (http://www.fhwa.dot.gov/ pavement/mana.cfm). 0 5 10 15 20 25 Annually Some Annually and Some Every 2 to 3 Years Every 2 Years Number of States FIGURE 5 Data collection frequency (out of 50 states) (http://www.fhwa.dot.gov/pavement/ mana.cfm).
18 FIGURE 6 Quality management activities suggested during each stage of data collection (Flintsch and McGhee 2009). system (Shekharan et al. 2007). The same study estimated that without a quality plan agencies may be over- or underestimating maintenance and rehabilitation needs by 25% or more (Shekharan et al. 2007). NCHRP Synthesis 401 found that approximately one-third of the state and provincial transporta- tion agencies had formal Quality Management Plans in place to document how the agency evaluates and manages data quality (Flintsch and McGhee 2009). The study found that a comprehensive Qual- ity Management Plan addresses all three phases of the data collection process: prior to the start of data collection, during the production stage, and as data are submitted. The activities to be completed during each of these phases are shown in Figure 6 (Flintsch and McGhee 2009). Methods used to Develop Pavement Management Prediction Models and Decision trees The project and treatment recommendations generated from a pavement management system are based on analysis parameters defined in the software. At the most basic level, this involves develop- ing deterioration models that predict pavement conditions over time and treatment rules or decision trees that identify the conditions under which each treatment is considered a suitable alternative.
19 trends in Prediction Modeling The AASHTO Pavement Management Guide describes several characteristics associated with pavement performance modeling that reflect general trends in the industry. These trends indicate that the following practices are used commonly in pavement management (AASHTO 2012): ⢠Pavement performance models may be developed for individual distress types and/or pavement condition indices (such as a cracking index or roughness index). ⢠At least four different approaches have been used to develop models, including: â subjective models based on agency expertise, â deterministic models that predict a single dependent variable from one or more independent variables, â probabilistic models that estimate the likelihood that a pavement will change from one condition state to another, and â Bayesian models that combine both objective and subjective data in terms of a probability distribution. ⢠Individual models can be developed for each pavement section in the database or models can be developed for groups of pavements with similar characteristics, often referred to as a âfamily.â Modeling the performance of a family simplifies the modeling approach by reducing the number of independent variables used in the equation. As some states update their pavement management processes, new pavement performance models are being developed. For instance, the Maryland State Highway Administration is updating its pave- ment management system and used the opportunity to move away from the hundreds of performance models that were in their previous system. The update process included a statistical analysis on IRI values to define groups with similar characteristics (such as region and traffic levels) by pavement and treatment types (Arambula et al. 2011). A histogram-based approach was used to develop the performance models to circumvent the use of pavement age as an independent variable. The vali- dation process confirmed that the resulting models provided satisfactory network-level predictions (Arambula et al. 2011). The California DOT (Caltrans) also recently developed new performance models using a mechanistic-empirical design system and an incremental recursive approach to modeling (Lea et al. 2014). During the development of their models, Caltrans found that âpavement response is highly sensitive to the thickness and material properties of each layer, especially the type of asphalt material and the source of the asphalt binderâ (Lea et al. 2014). They also found that certain variables that were not contained in their pavement management database, such as subgrade type, asphalt source, pavement condition before treatment, and layer thickness, were statistically significant in influencing pavement performance (Lea et al. 2014). The literature indicates that several agencies are exploring the use of pavement condition survey data for calibrating Mechanistic-Empirical Pavement Design Guide (MEPDG) models. For instance, in a study for the Texas DOT, researchers used pavement data from the Texas Specific Pavement Study (SPS)-1 and SPS-3 experiments conducted under the FHWAâs Long-Term Pavement Performance study to calibrate the asphalt concrete pavement deformation performance model (Banerjee et al. 2009). Washington State DOT (WSDOT) documented its work in calibrating the MEPDG models, which was greatly facilitated because their historical pavement condition database included more than 30 years of longitudinal cracking, transverse cracking, alligator cracking, rutting, and roughness data for their asphalt roads (Li et al. 2009). The Washington State pavement management system also provided detailed structural information such as layer thickness, material, and asphalt binder type that was instrumental in the calibration process. FHWA illustrated the feasibility of using pavement management data to support the local calibra- tion of the MEPDG models using actual data provided by the North Carolina DOT (FHWA 2010). Although the study was successful at using pavement management data to perform the calibration, there were several identified challenges, including the time required to match the design and con- struction records to the pavement management sections (FHWA 2010). One more challenge resulted
20 from the differences in the stateâs pavement condition survey procedures and the distress definitions that served as the basis for the MEPDG models. Another trend in pavement performance modeling involves the incorporation of uncertainty into pavement management performance modeling. In one study on this topic, the authors address the dis- advantages to characterizing pavement sections using average condition ratings, including the loss of valuable information and the increased likelihood of developing inaccurate or misleading answers (Kadar et al. 2015). Their research indicates that by using the full data set, and treating each data set as a distribution, the probability of the outcome can be estimated with the predicted value (Kadar et al. 2015). The literature also addresses the use of Bayesian approaches to update expert-based Markov tran- sition probability matrices as historical data becomes available in a pavement management system (Tabatabaee and Ziyadi 2013). The approach developed by the researchers incorporates the uncer- tainty in both the initial transition probability matrices and the pavement condition survey method. The approach was tested using data from the Minnesota DOTâs MnROAD test facility and verified the importance of taking the variability of both factors into account. Decision tree trends To develop recommendations for the optimal use of available funding, a pavement management system includes decision trees or treatment rules that identify when each treatment is considered to be a feasible option. These treatment rules recognize that different types of strategies are appropriate at different times in a pavement life cycle, as depicted in Figure 7 (AASHTO 2012). In this exam- ple, preventive maintenance and minor rehabilitation, which are both considered to be preservation activities, are feasible strategies when a pavement is in relatively good condition, whereas major rehabilitation and reconstruction are more appropriate when a pavement is in fair or poor condi- tion. Other considerations, such as pavement type, distress type, road functional classification, and previous treatment history are also used to identify appropriate treatments, as shown in the decision tree example used by the Minnesota DOT for its asphalt and asphalt over concrete roads (Figure 8). Over the last decade, there have been an increasing number of publications addressing the integra- tion of preventive maintenance treatments into a pavement management system. One study identified typical gaps in a pavement management system that limit the ability to successfully model preventive maintenance treatments. The paper suggested that current pavement condition surveys make it dif- ficult to trigger treatments designed to address bleeding or raveling because those distresses are not commonly found in a network-level survey (Zimmerman and Peshkin 2004). It also recognized that the lack of integration between pavement management and maintenance databases makes it difficult to develop performance models and treatment impact rules for preventive maintenance treatments. FIGURE 7 Relationship between pavement condition and treatment strategies (AASHTO 2012).
FIGURE 8 Decision tree used by the Minnesota DOT for asphalt and asphalt over concrete pavements (http://www.dot.state.mn.us/materials/pvmtmgmtdocs/Bituminous_ Decision_Tree_07-01-12.pdf).
22 Utah DOT is an example of an agency that has developed decision trees in its pavement manage- ment system to guide the selection of preventive maintenance treatments. Figure 9 illustrates the factors that differentiate the type of seal coat recommended on asphalt roads maintained by Utah DOT (FHWA 2008b). In this example, treatment selection is differentiated by the type of facility (rural versus urban), functional classification, and traffic volumes. Ohio DOT also has developed treatment rules to guide the selection of preservation treatments, as shown in Figure 10 (Peshkin et al. 2011). In addition to the condition data shown in the figure, Ohio DOT also uses information about the types of distress present and traffic volumes to make final decisions about which treatment is most appropriate. There is also an increasing emphasis on developing decision trees and treatment rules for using pres- ervation treatments, including preventive maintenance, on high-volume roads (Peshkin et al. 2011). A research study for the SHRP 2 program developed guidelines for using preservation treatments on these types of facilities that emphasized the importance of design and quality construction, condition of the existing pavement, level of traffic under which the treatment must function, and climatic conditions to which the treatment is exposed as being key factors to treatment performance (Peshkin et al. 2011). Methods of assessing Pavement Performance trends and treatment Cost-effectiveness Pavement performance data in a pavement management system can be used to evaluate the cost- effectiveness of various treatments. In recent years, the literature has revealed that agencies are very interested in determining the cost-effectiveness of pavement preservation activities. One such study evaluated several common methods of evaluating pavement preservation interventions including effectiveness (benefit) only, cost only, cost-effectiveness, and economic efficiency (Khurshid et al. 2009). The study found that short-term effectiveness for a treatment does not necessarily translate FIGURE 9 Decision tree used by the Utah DOT for seal coats (FHWA 2008b).
23 into long-term effectiveness. It also found that evaluations focused only on cost or effectiveness yield biased results, so that methods that consider both factors (such as cost-effectiveness and economic efficiency) are preferred (Khurshid et al. 2009). WSDOT conducted an evaluation of the cost-effectiveness of its pavement management decisions using economic evaluation techniques such as Cost-Effectiveness Analysis, Replacement Analysis, and Break-Even Analysis (Luhr and Rydholm 2015). The study found that even a one-year extension in pavement life from the use of cost-effective preservation treatments can have significant benefits to an agency (Luhr and Rydholm 2015). Another study evaluated the impact that data errors can have on the analysis results, using data from the Quebec Ministry of Transport and Virginia DOT (Saliminejad 2016). The researcherâs risk assess- ment approach demonstrates how an agency can estimate the magnitude of different types of data errors to better focus their data quality management activities on those risks that have the highest impact. Pavement management condition data has also been used to evaluate the effectiveness of the pavement warranty program initiated by Mississippi DOT in 2000 (Qi et al. 2013). The study evalu- ated the programâs effectiveness using information on roughness, rutting, cracking, and other surface distress information stored in the agencyâs pavement management system. By comparing the distress data for warranted and nonwarranted projects using statistical analysis, the study found that the pave- ments covered under a warranty were deteriorating at a slower rate than the rest of the pavement sections and the overall performance of the warranty sections was better than the rest of the sections over the same length of service (Qi et al. 2013). trends in the use of Pavement Management Data Traditional uses for a pavement management system include activities associated with identifying treat- ment recommendations for developing a multi-year work plan and displaying the impacts of different investment strategies on system conditions with time. The literature indicates that as agencies increas- ingly move toward making performance-based investment decisions, pavement management will have an important role in providing analysis results that convey the consequences of different options. The use of performance-based investment decisions is common internationally, where many coun- tries are using asset management principles. In 2012, an international scan was conducted to explore how certain transportation agencies were managing and monitoring their pavements. The scan participants FIGURE 10 Preservation treatment rules used by the Ohio DOT (Peshkin et al. 2011).
24 found that, internationally, agency cultures supported a long-term view for managing pavements in which agency priorities were known and agency personnel were held accountable for their actions (Zimmerman et al. 2013). Most of the agencies visited during the scan were moving toward a service- based approach for managing their road networks rather than a condition-based approach. Under this new approach, customer-driven priorities such as safety, reliability of travel, comfort, and livability are becoming the primary drivers for triggering road maintenance and renewal actions (Zimmerman et al. 2013). This has led to changes in the types of data that are collected and the performance targets that are driving the maintenance and renewal programs. In the United States, there is evidence that transportation agencies are also exploring the applica- bility of pavement management systems to address customer-driven priorities, through the consid- eration of safety and environmental impacts. For instance, one study recommends the use of friction data, combined with crash data, to conduct a network-level analysis intended to better predict loca- tions of vehicular crashes (de León Izeppi et al. 2016). The authors suggest that improved pavement management systems that can predict crash locations, together with proactive maintenance treat- ments to address areas with inadequate friction numbers, could significantly reduce the number of crashes that occur (de León Izeppi et al. 2016). Texas DOT recently completed a study to establish threshold values for skid resistance that could be used in network-level planning and programming decisions (Wu et al. 2014). The framework developed under the study (1) established a quantitative relationship between pavement skid resistance and crash rates and (2) set threshold values for trig- gering maintenance decisions based on skid resistance (Wu et al. 2014). The use of pavement management data to support the identification and assessment of risks is another recent trend. To a certain degree, the consideration of risks using pavement management data has been influenced by the federal legislation, requiring state DOTs to develop risk-based asset management plans. Managing risks or uncertainty typically includes the following fundamental ele- ments (Cambridge Systematics Inc. et al. 2009): ⢠Establishing risk tolerances, ⢠Identifying threats or hazards, ⢠Assessing impacts or consequences, ⢠Identifying potential mitigation strategies, ⢠Developing a mitigation or management plan, and ⢠Implementing the plan. A pavement management system can support several of these elements; for example, providing pavement performance data to assess risk impacts and suggest potential mitigation strategies. The pavement management system can also be used to prioritize risk mitigation investments in areas considered critical to the system. Risk can also be incorporated into pavement management models to better understand the uncertainty associated with recommended investment programs. Another recent trend is the consideration of environmental impacts from maintenance and rehabili- tation treatments recommended in a pavement management system in the project-selection process. Researchers have found that the consideration of these factors reduced energy use and greenhouse gas emissions by 19% to 24%, with just a small sacrifice in pavement performance (i.e., 98.5% of the optimal solution) (Faghih-Imani and Amador-Jimenez 2013). Muench and Van Dam (2015) conducted a study that summarizes how climate change impacts pavement systems and used the results to identify several pavement adaptation strategies that can be incorporated into pavement management systems. These strategies include the use of higher-temperature asphalt binders, increased use of rut-resistant designs, greater consideration of the concrete coefficient of thermal expansion, and shorter joint spac- ing in concrete pavements (Muench and Van Dam 2015). Methods of incorporating environmental impacts into pavement management systems were also addressed by Pellecuer et al. (2016). Putting Pavement Management recommendations into Practice The recommendations from a pavement management system serve as a foundation for develop- ing a network improvement program; however, there are many factors that can influence the final
25 selection of funded projects. In some instances, especially where an agencyâs districts or regions are fairly autonomous, there may be significant differences between the pavement management recommendations and the agencyâs improvement program. In an effort to reduce the differences between the pavement management system recommen- dations and the final work plans, Colorado DOT established a goal to have 70% of a regionâs construction plan match the treatment recommendations generated by its pavement management software (CDOT 2005). The match is based on a common location, the level of treatment, and the treatment timing (±4 years) (CDOT 2002). To help improve consistency in the work plans developed by Ohio DOT district personnel, and to ensure that the DOT meets its performance targets, Ohio DOT recently implemented new business processes to support its asset management activities. As part of these changes, district work plans now combine maintenance and capital improvements into a single plan and a percentage of the work activities must match the treatment strategies recommended by the pavement management system (ODOT 2016). In the first year of the new business processes, districts are expected to match at least 25% of the pavement management recommendations; however, in later years the match will increase to at least 75% (ODOT 2016). Strategies for Presenting Pavement Management Data and analysis results Pavement management information is presented in a number of different formats, including printed documents, GIS maps, and online dashboards. A few examples of typical report outputs from a pave- ment management system are provided, illustrating historical pavement conditions in Figure 11 and expected conditions under three different funding levels in Figure 12. In 2004, an NCHRP synthesis reported that GIS systems have been particularly helpful to pave- ment management practitioners for integrating, managing, analyzing, and presenting data from mul- tiple data sets (Flintsch et al. 2004). An example of the type of pavement management information commonly displayed on maps is provided in Figure 13. In an effort to improve agency accountability, report cards and dashboards showing targeted and actual conditions are becoming increasingly common. These items typically include a range of performance measures such as safety, mobility, and pavement condition measures. An example of a portion of a Performance Measure Report Card published by North Dakota DOT is presented in Figure 14. FIGURE 11 Historical pavement condition trends (Haas, Hudson, and Cowe Falls 2015).
FIGURE 12 Estimated backlog and pavement condition index under three different funding scenarios (Romell and Tan 2010). FIGURE 13 Percent change in âGoodâ or better condition by county for FY 2011â2014 (TxDOT 2014).
27 FIGURE 14 A portion of the Performance Measurement Report Card developed by the North Dakota DOT (http://www.dot.nd.gov/ divisions/exec/docs/pm-rpt-cd.pdf).
