Pavement Management Systems

The American Association of State Highway and Transportation Officials (AASHTO) defines pavement management as “…the effective and efficient directing of the various activities involved in providing and sustaining pavements in a condition acceptable to the traveling public at the least life cycle cost (AASHTO, 1985[1]).” This concept of providing pavements and maintaining them in acceptable condition is as old as the first pavement. As pavement networks grew slowly in the first half of the twentieth century and then quickly in the 1950s and 1960s, simple procedures or experience that had worked previously was no longer able to manage these burgeoning networks. Instead, a more holistic systems approach was needed.

Originally described as “a systems approach to pavement design”, the term “pavement management system (PMS)” came into popular use in the late 1960s and early 1970s to describe decision support tools for the entire range of activities involved in providing and maintaining pavements (OECD, 1987 and Peterson, 1987[2]). Hudson et al. (1979[3]) describe a “total pavement management system” as

“…a coordinated set of activities, all directed toward achieving the best value possible for the available public funds in providing and operating smooth, safe, and economical pavements.”

Haas and Hudson (1978[4]) expand on this by defining “activities” as those actions associated with pavement planning, design, construction, maintenance, evaluation and research.

There are numerous different pavement management systems (PMS) from which to choose, each one with its own level of complexity. For a small town or rural county a simple system based on visual inspection and maintained in a Microsoft Excel or Access database may be more than sufficient. For a state road network a more complex PMS is usually appropriate.

Pavement Management System Components

Most formal definitions of a “pavement management system” agree on five key components (Peterson, 1987[5]):

  1. Pavement condition surveys. Pavement condition surveys were probably the first PMS component to be adopted on a large scale by U.S. transportation agencies. For example, WSDOT, a PMS early adopter, began pavement surveys in about 1965 (Nelson and LeClerc, 1982[6]). Condition survey research is largely concerned with advancing or refining measurement and data collection.
  2. Database containing all related pavement information. Databases have evolved along with the pavement condition survey data they are designed to house. Computer databases gained prevalence in the 1970s and as adequate, cost effective computing power and storage became available. Recent research has concentrated on implementing more robust databases (e.g., Microsoft SQL server, Oracle, etc.) and better user interfaces including GIS-based spatial interfaces. These interfaces are as important as the data itself because they enable users to view and manipulate data in a meaningful way.
  3. Analysis scheme. Analysis schemes are those algorithms used to interpret data in meaningful ways. The late 1960s and early 1970s saw the introduction of computer-based optimization algorithms (Haas et al., 1979[7]). Recent software can combine the database, analysis scheme and decision criteria in one package. Recent research has focused on advancing or refining life-cycle costing analysis, optimization algorithms and performance prediction.
  4. Decision criteria. Decision criteria are those rules developed to guide pavement management decisions. As pavement management systems have evolved, decision criteria have become more complex and now account for items such as user delay, vehicle operating costs and, in limited cases, environmental effects. Research is ongoing to develop and refine appropriate decision criteria and the ability to automatically apply these criteria.
  5. Implementation procedures. Implementation procedures are those methods used to apply management decisions to roadway sections. Implementation is often thought of as a political, budgetary or procedural issue and is not often dealt with in research.

Pavement Management Methods

The structure of a pavement management system can be separated into two general levels; the network (or system) level and project level. The network level deals with the pavement network as a whole and is generally concerned with high-level decisions relating to network-wide planning, policy and budget. For example, managers at this level might compare the benefits and costs for several alternative programs and then identify the program/budget that will have the greatest network benefit-cost ratio over the analysis period.

The project level deals with smaller constituent sections within the network and is generally concerned with lower-level decisions relating to condition; maintenance, reconstruction and rehabilitation (MR&R) assignment; and unit costs. For example, at this level detailed consideration is given to alternative design, construction, maintenance and rehabilitation activities for specific projects. This might be accomplished by comparing benefit-cost ratios of several design alternatives and picking the design alternative that provides the desired benefits for the least total cost over the projected life of the project.

Approaches to pavement management tend to attack this two-level system either from the top down by dealing with network-level decisions first, or from the bottom up by dealing with project-level decisions first. Either method can be quite detailed or relatively simple depending upon data amount and quality and desired analytical capabilities.

As an example, the AASHTO 1990 Guidelines for Pavement Management Systems sets forth three basic pavement management methodologies, two are bottom-up and one is top-down. The following lists these methodologies and discusses their basic approach and why they would be considered project-level approaches (bottom-up) or network-level approaches (top-down) (Zimmerman and ERES Consultants, 1995[8]):

  1. Pavement condition analysis (project-level approach). This method, considered the simplest of the three, aggregates pavement condition information at the project level and then selects the most appropriate MR&R strategy. Each project is assigned a priority based on a number of factors including location, traffic, safety, etc. Pavement MR&R projects are then selected by priority based on a network-level budget. This system is simple but limited in effectiveness because it does not consider future pavement condition. Because the first decisions are made at the project-level, this would be considered a project-level approach.
  2. Priority assessment models (project-level approach). This method improves upon pavement condition analysis by incorporating predicted future pavement condition information. Thus, priority assessment models consider future predicted conditions and can possibly perform limited “what if” scenarios based on network level decisions. However, conditional (“what if”) scenario capability is limited because alternate decisions can only be modeled by changing the project-level data – generally a time-consuming task. This is also considered a project-level approach because, although more complex, it still begins at the lowest decision level (determining an individual pavement section’s MR&R strategy) and progresses to the highest decision level (an overall pavement network strategy).
  3. Network optimization models (network-level approach). This method, considered the most sophisticated, simultaneously evaluates an entire pavement network to determine the optimum network management strategy. Specific MR&R projects and locations are then selected to meet this strategy. This method is considered a network-level approach because it begins at the highest decision level and progresses to the lowest decision level.

The most appropriate pavement management approach depends largely on the situation. The top-down network-level approach offers (1) better institutional control, (2) clear advantages in conditional scenario capability and (3) is best able to accommodate the ever-changing political and social issues inherent in our republican form of government. The bottom-up project-level approach provides only basic conditional scenario capabilities but can provide much more detailed and accurate data that is capable of supporting individual project decisions.

Comparison of Approaches

This section discusses the advantages and disadvantages of both network-level and project-level approaches to pavement management by listing the principal advantages of each approach. The disadvantages of each approach correspond to the advantages of the other; for instance, one advantage of the network-level approach is that it can optimize solutions for the entire network. The conjugate disadvantage of the project-level approach is that it may not be able to optimize solutions for the entire network.

Network-Level Approach Advantages

The network-level approach is characterized by top-down logic, system optimization, aggregate data, large data and resource requirements, and sophisticated models. Its chief advantages are that it can:

  1. Optimize solutions for the entire network. By definition, this is what a network-level approach does. For instance, a network-level approach can optimize the cost-benefit ratio for the entire network. This would seem the most logical since the system, rather than an individual project, is the overarching concern. Project-level approaches attempt to replicate this ability by assigning project priorities that are commensurate with network-level programs, decisions or budgets. However, because projects are already planned before high-level decisions are made, project-level decisions and priorities may not be consistent with network-level decisions and priorities. This can lead to a suboptimal system solution that is driven by individual project-level decisions instead of network-level decisions.
  2. Quickly and accurately produce conditional scenarios. Software models using a network-level approach allow the user to adjust top-level budget and policy inputs and then quickly calculate the resulting network-wide effects because these models are driven by top-level (network-level) decisions. For example, a network-level model could calculate the economic and pavement conditions effects of a proposed lower axle load limit law or the long-term network performance under varying levels of funding. Conversely, project-level software is generally driven by low-level inputs and thus, a change in top-level budget or policy would be input into the system by adjusting the multitude of lower-level inputs – a more laborious process.
  3. Prioritize broad areas of MR&R. Since network-level analysis provides target MR&R treatments and costs, these targets can be easily and consistently applied to individual projects. In order to accomplish the same thing with a project-level approach, network-level targets need to be provided in advance such that project-level decision can be made with network-level targets in mind.
  4. Use consistent inputs in scenario comparisons. Using a network-level model, different scenarios can be modeled on the same system. This helps if each scenario is modeled using consistent assumptions models, outcomes may still be able to reflect appropriate qualitative and comparative results. Project-level approaches have more difficulty in this area because basic assumptions must be input on the lower, project level. As this is done agency-wide, communication problems and personal/geographic bias may have a substantial effect on input consistency.
  5. More easily obtain top management attention. At the 1997 New Orleans pavement management workshop, many pavement management practitioners raised the following issues: (1) pavement management had lost the attention of top management because they did not understand its capabilities or importance, and (2) some managers haphazardly overrode pavement management recommendations with little or no though to network implications (Zimmerman, Botelho and Clark, no date given). With its conditional scenario capability, a network-level approach could easily show the fiscal importance of pavement management as well as the implications of various decisions.

Project-Level Approach Advantages

The project-level approach is characterized by simpler models, less data aggregation, fewer data and resource requirements, less reliance on feedback for success and better understanding. Its chief advantages are:

  1. Relies less on aggregate data. The network-level approach relies on aggregate data to drive its models. There is an inherent risk involved when using aggregate data to drive network-level decisions. First, the aggregate data, if not carefully chosen, may not be representative of actual conditions, which could lead to incorrect decisions at the highest level. Second, aggregate data is sometimes difficult to translate into specific project-level results (e.g., overlaying all pavement above a certain IRI on a particular route may be costly if these high-IRI sections are small isolated segments).
  2. Able to be used with little data. Project-level systems can be used in situations where data availability/requirements are small making them ideal for smaller agencies. Network-level systems require large amounts of data and resources (such as computers, trained personnel, advanced algorithms), which smaller municipal agencies may not be able to afford. Additionally, the networks managed by these smaller municipal agencies may be simple enough that they do not require a network-level analysis. For instance, a small city with perhaps 40 km (25 miles) of roadway could probably use a simple pavement condition analysis method to meet all their information needs. Project-level systems can, however, require a significant amount of data depending upon their complexity and how they are modeled.
  3. Better link between network-level and project-level management decisions. Because decisions flow from the bottom up, the high-level network decisions, although somewhat limited in scope, in essence must be based on low-level project decisions. However, in a network-level approach it is sometimes difficult to translate broad network-level decisions into specific project actions. For instance, a decision to improve all IRI > 3.0 m/km roadways by overlay may be made at the network level, but these segments of roadway may be located in small sections over a wide area making for inefficient construction.
  4. Less dependent upon feedback for success. Political sentiment, budgets, pavement condition and MR&R strategies are highly dependent upon the local environment. Because of this, network-level models will invariably need calibration, which can only occur through continual feedback and updating. If this feedback/update process is interrupted or halted (e.g., through budget cuts, personnel transfers) a network-level model’s utility can quickly degrade. Additionally, uncalibrated network-level models that are used will produce erroneous results.
  5. Easier to obtain buy-in from others. Project-level approaches can be simpler and more easily understood. Network-level approaches typically use sophisticated models that make many generalizations and assumptions. Those who are unfamiliar with the model may be unwilling to use its results because (1) they do not understand how it works or (2) they do not agree with its generalizations and assumptions.

Footnotes    (↵ returns to text)
  1. AASHTO (1985). Guidelines on Pavement Management, American Association of State Highway and Transportation Officials, Washington D.C., 9p.
  2. Pavement management systems.  A report prepared by the OECD scientific expert group.  OECD, Paris, 1987.
  3. National Cooperative Highway Research Program Report 215: Pavement Management System Development.  NCHRP, TRB, National Research Council. Washington, D.C.
  4. Pavement Management Systems.  McGraw-Hill Book Company.  New York, NY.
  5. National Cooperative Highway Research Program Synthesis of Highway Practice 135: Pavement Management Practices.  NCHRP, TRB, National Research Council. Washington, D.C.
  6. National Cooperative Highway Research Program Synthesis 222: Pavement Management Methodologies to Select Projects and Recommend Preservation Treatments.  Transportation Research Board, National Research Council.  Washington, D.C.