When quality assurance specifications came into use in the late 1960s, the goal was to improve them to a level such that they would become performance-related specifications (PRS). Now, more than 30 years later, PRS are at a point where State highway agencies finally can begin to implement them. How did we get there, what are PRS, what benefits are to be gained from them, and what lies in store for the future?

History of Specifications

In the early days of road construction in the United States, little was known about the factors that contribute to the success or failure of pavements. Local jurisdictions responsible for road construction were small, numerous, and independent of each other. If these jurisdictions used specifications at all, they were generally skimpy and inadequate.

Under these circumstances, the best option for governing agencies was to require a guarantee, whereby the contractor promised to do any needed road maintenance and repairs for a specified time period. An 1898 decision by a New Jersey court supported this kind of maintenance guarantee. The court stated: "The quality of the pavement cannot be well ascertained without the test of time. It is therefore reasonable that those who lay such pavement should submit it to this test in order to insure its goodness."

As State highway agencies acquired a better understanding of road building, detailed "method specifications" began to replace guarantees as the preferred approach. Method specifications provide instructions to the contractor on the type of equipment to use, the time of year to pour the concrete, and even a "recipe" specifying the number of parts of aggregate, cement, and other ingredients.

The formation of the American Association of State Highway and Transportation Officials (AASHTO) in 1914 led to some uniformity in the method specifications used by State highway agencies. By 1935, most municipalities and local governments had dropped maintenance guarantees and were using method requirements similar to those used by the States.

In the 1960s, two events forced the highway community to question the desirability of method specifications. One was the advent of the AASHTO Road Test, which demonstrated that many specification requirements were unrealistic because they did not allow for variability in materials and construction. The Road Test provided the impetus for development of statistical quality assurance (QA) specifications that do take variability into account.

A second contributor was the interstate highway system, which encouraged technological advances that increased construction speed. State highway agencies may have been at least partly motivated to implement QA specifications because they had too few inspectors to oversee the rapidly growing interstate system under method specifications.

QA requirements replaced some (but not all) prescribed methods with descriptions of the desirable properties of the completed product, such as air content, strength, and consolidation. Assuming that faster, nondestructive tests of highway performance could be developed, the QA specifications were expected to drop additional prescriptive requirements and become more focused on the end result. The promise of evolving into such performance-related end-result specifications was an appealing selling point for QA specifications.

The transition from method to QA specifications thus generated a demand for new tests that would provide agencies and contractors with fast, accurate, and precise pictures of the quality of the finished product. In the 1960s and 1970s, the Federal Highway Administration (FHWA), the State highway agencies, universities, and private industry funded numerous research activities that resulted in new QA tests. These tests ranged from nuclear density and cement content gauges to improved methods for measuring pavement smoothness. The search for such methods is far from over.

Debut of Performance-Related Specifications

By the early 1980s, most State highway agencies that had moved toward QA specifications had added payment reduction provisions to their specifications. The goal was to penalize contractors when construction did not meet specific targets for quality (most often strength, thickness, and smoothness), assuming that the construction quality was not poor enough to warrant removal and replacement. More recently, agencies have started to add pay increase provisions, i.e., bonuses, to their QA specifications to reward contractors for superior quality.

Penalties varied significantly from one agency to another to the extent that construction that one agency considered acceptable for 100 percent payment might be heavily penalized by another. The reason was apparent: Agencies based their payment reduction plans on "engineering judgment." The judgments varied widely, since few models were available to relate test results to performance.

FHWA and the State-directed, Federally funded National Cooperative Highway Research Program (NCHRP) immediately began efforts to provide rationales for pay increase and decrease provisions, through what were called "performance-related specifications." The goal was to develop specifications built around QA tests, whose results are quantitatively related to long-term pavement performance.

In 1987, Paul Irick and a team of NCHRP researchers introduced a framework for establishing PRS. During the 1990s, researchers used Irick's framework to develop prototype PRS for both portland cement concrete and hot-mix asphalt paving. For concrete, the researchers also created the PaveSpec computer program.

Today, State highway agencies can use the current PaveSpec 3.0 version to develop PRS for concrete paving prior to construction and compute pay adjustments during and after construction. Contractors can use it to help establish the level of quality they should target under PRS (or under warranty specifications). PaveSpec 3.0 also serves as a technology transfer tool to enable engineers to obtain a better understanding of what it takes to construct high-performance pavements. It is available as a CD or can be downloaded from FHWA's www.fhwa.dot.gov/research/tfhrc/ Web site.

Field Trials

Field trials of PRS for concrete were conducted in four States: Iowa in 1996 and Kansas, Missouri, and New Mexico in 1997. In these field trials, the use of PRS was simulated on projects being constructed under the respective State's current specifications. The field trials showed that the quality control/quality assurance (QC/QA) testing by the contractor and the agency did not take more time or effort than required under current specifications.

"All the States seemed concerned about the amount of testing [turning out to be] more than they usually did," says Todd Hoerner, pavement engineer with Applied Pavement Technology in Illinois, "but once they saw what we were doing, they weren't as concerned."

During the field trials, the researchers developed a methodology to help agencies convert from their specifications to PRS, adjust existing models to local conditions, and add new prediction models. "Our proposed methodology has been well-received around the country at different presentations," says Hoerner.

In 1999, the Indiana Department of Transportation (using the PaveSpec 2.0 version) developed PRS for a paving project on I-465 in Indianapolis, IN. The project, constructed the following year, consisted of 2.43 kilometers (1.50 miles) of six-lane jointed plain concrete pavement.

"The specifications provide the basis for rational acceptance and/or price adjustment decisions," says Tommy E. Nantung, research division section manager for Pavement, Materials, and Accelerated Testing at the Indiana Department of Transportation (IND.). "Indiana is the first State in the country to implement the FHWA Performance Related Specifications Level 1 in pavement. The specification was well received by IND. and the contractors involved in this PRS project."

Jason Weiss, assistant professor at Purdue University, adds, "The use of PRS in the reconstruction of I-465 during the summer of 2000 provided valuable information on the selection of target inputs, construction quality and consistency, applications and limitations of nondestructive testing, and pavement smoothness measurements. Lessons learned from this project have been incorporated into the specification development for the second PRS project in Indiana, the reconstruction of I-65 near Clarksville, which is currently under construction."

Berns Construction Company, Inc., was the first contractor in Indiana to have a project using PRS. Richard M. Newell, quality control supervisor with Berns, says, "PRS are an exciting step forward for our industry since they enable us to provide the traveling public with a high-quality, long-lasting concrete pavement while rewarding the contractors who are capable of providing this material. This results in an overall win-win solution for all those involved."

In 2001, the Florida Department of Transportation followed with the development of PRS for a 0.8-kilometer (0.5-mile), six-lane project on I-295 in Jacksonville using jointed plain concrete pavement. The project is to be completed in 2002. The Tennessee Department of Transportation is making plans for a PRS project, and other agencies have expressed interest as well.

Main Features of PRS

Like other QA specifications, PRS specify requirements for quality. The major difference between conventional QA specifications and PRS is that the latter include performance-related pay increase and decrease provisions. PRS are based on quantified relations (i.e., mathematical models) between key design, materials, and construction variables and pavement performance. The models, based on laboratory and field data, present a clearer and more realistic picture of performance than can be visualized through engineering judgment and intuition alone.

PRS contain performance-prediction models, which use a pavement's design and materials to predict when and to what extent the pavement will exhibit various types of distress, such as transverse cracking or joint spalling (chipping). PRS also contain models that start from the development of distress to estimate a post-construction life-cycle cost of maintenance and rehabilitation.

Inputs to the performance-prediction model are materials, design variables (such as traffic loading, climatic factors, drainage, and roadbed soil factors), and quality characteristics of the materials and construction (such as the water-cement ratio, strength, slab thickness, and concrete stiffness). The output is a prediction of life-cycle cost for the constructed pavement.

When the target values of quality characteristics called for in the specifications are used as inputs, the output produced is the "as-designed life-cycle cost." When the actual measured values of a constructed pavement's quality characteristics are used as inputs, the output produced is the estimated "as-constructed life-cycle cost." The difference between the as-designed and as-constructed life-cycle cost is the basis for any pay adjustment, either positive or negative.

The ability of PRS to predict life-cycle costs addresses many limitations of conventional QA specifications. PRS offer the following advantages:

• Contractors working under PRS focus on minimizing the as-constructed life-cycle costs. No other current specifications seek to minimize these costs.
• Virtually an unlimited number of quality characteristics can be considered in the development of pay adjustments, provided that a prediction model relates the quality characteristic to pavement performance.
• PRS directly consider the within-lot variability of the quality characteristic and account for it in the development of pay adjustments. Many QA specifications specify only the mean value, while ignoring the variability.
• PRS present a procedure for computing pay adjustments, in accordance with the legal principle of liquidated damages.
• The procedure provides an incentive for the contractor to deliver high-quality work by allowing positive as well as negative pay adjustments.
• Some PRS require testing of the in situ pavement, thereby providing a true assessment of its as-constructed properties.

Guidance on PRS

FHWA has made considerable guidance available for highway agencies interested in developing PRS for concrete paving. For agencies having no previous experience with these specifications, FHWA recommends Level 1 PRS that provide the agency with experience in establishing life-cycle costs while allowing use of current tests. Level 2 PRS are more sophisticated and can offer greater advantages, particularly when they call for in situ testing and project-specific pay adjustments. Under Level 2 PRS, one overall pay adjustment is calculated that reflects the interactions among quality characteristics.

The Indiana field trial described earlier was the first project in the United States with PRS developed according to the guidelines. The PRS developed for the Indianapolis and Jacksonville projects are Level 1 PRS from FHWA's guide specifications and accompanying PaveSpec software. They are intended for use on a single project, as they relied on project-specific input values. Had the agency instead identified typical PaveSpec input values associated with different classifications of projects, the resulting PRS could have been used on projects throughout Indiana and Florida. However, the life-cycle cost estimates obtained upon project completion would not have been as accurate.

FHWA's current guide PRS are aimed at controlling distresses such as transverse joint spalling, transverse joint faulting, transverse slab cracking, and development of pavement roughness over time. As the use of PRS grows, FHWA expects to add distresses to meet special needs of highway agencies.

Under the current guide PRS, the agency bases the acceptance of a pavement on any or all of the following quality characteristics:

• Concrete strength
• Entrained air content
• Slab thickness
• Initial surface smoothness
• Percent consolidation around dowels

FHWA may add other quality characteristics, such as surface friction and degree of dowel bar misalignment. The only restrictions are that quality characteristics must be under the control of the contractor, measurable (preferably quickly in situ and nondestructively), and appear in a performance/distress prediction model.

Once the PRS are developed, PaveSpec 3.0 can perform sensitivity analyses to investigate the effects of variations in construction quality on project life-cycle cost, and hence on pay factors. Contractors having an understanding of the costs associated with delivery of various quality levels can use the results of the sensitivity analyses as a basis for bid strategy.

The software uses a Monte Carlo simulation to compute pay adjustments. At the PRS development stage, mean and standard deviation values for the as-designed acceptance quality characteristics are entered into the software along with the other necessary information (for a total of 126 inputs). These as-designed means and standard deviations are based on the design or on agency policy and past experience. Hundreds (or thousands) of iterations are run. For each iteration, a value of each quality characteristic is selected from a normal distribution, defined by an as-designed mean and standard deviation, and a life-cycle cost is computed. The average of the computed life-cycle costs is the as-designed life-cycle cost. The program then uses test results from acceptance sampling to calculate the as-constructed life-cycle cost estimate for each constructed lot and then assigns pay factors.

Current Status of PRS

For portland cement concrete, PRS products have been developed only for jointed concrete pavements. Research is still needed for other types of pavements. For asphalt paving, products are starting to become available.

The PRS program in the United States now is beginning to transition from research to development and implementation. In 1999, in anticipation of PRS implementation, FHWA formed a team to develop a comprehensive, coordinated action plan to advance the understanding and use of PRS as a viable contracting option.

Guiding execution of the National PRS Action Plan is the PRS Technical Working Group, an 18-member steering committee that includes representation from States, industry, academia, and FHWA. Expert Task Groups will handle specific areas such as asphalt and concrete paving, structures, and pavement preservation.

The groups developed a specification matrix, which acknowledges that the various aspects of pavement performance (e.g., rideability, skid resistance, and noise) can best be controlled through method specifications, QA specifications, PRS, or performance specifications (e.g., warranties). Which specifications to employ to control a particular aspect of pavement performance depends on the availability of suitable test procedures and appropriate performance models, as well as understanding what it takes to achieve the desired performance. Which specifications (or combination of specifications) to employ on a given project, or on a statewide basis, depends on additional issues that must be deliberated carefully.

Ironically, warranties were seen in the early days of road construction as the logical specifications due to lack of knowledge and technology (i.e., there was nothing better). They were later discontinued primarily for being a "source of endless litigation," as Roger L. Morrison put it in a report for The Asphalt Institute. Now, warranties are once again emerging, but this time as specifications whose success depends on obtaining more knowledge and technology. The National PRS Action Plan recognizes that PRS and warranties have much in common. They are actually dependent on each other and are both aimed at evolving into cost-effective, end-result performance specifications.

A Look at the Future

The evolution of highway construction specifications always has been dependent on advances in technology and understanding of the product. Political and economic influences (such as staffing shortages and contractor demographics) sometimes inhibited, or even set back, their evolution. Ideally, agencies should be specifying pavement performance requirements, rather than materials and construction quality requirements, or method requirements. The implementation of PRS is in keeping with the ultimate goal of developing end-result specifications whereby expected performance is promptly and accurately assessed upon product delivery.

What then might PRS be like in the future? Initially, PRS will continue to be Level 1 as agencies and contractors obtain an understanding of and experience with life-cycle costs. Some agencies will try projects governed by Level 2 PRS within the traditional (cost of construction) low-bid award system, or by Level 2 PRS within an innovative multiparameter low-bid award system.

As experience with PRS projects is gained, agencies will want their PRS to control additional aspects of performance through distress/performance prediction models. Also, PRS will be developed for the entire pavement structure, rather than just the concrete layer, making the life-cycle cost estimates more accurate.

The need to collect and maintain comprehensive data on quality, cost, time, and performance will become apparent. As more PRS projects are constructed, they will generate more data leading to more useful databases to improve the models.

As the performance models improve, they will benefit contractors working under warranty projects as well. Contractors will refine the models and develop their own specific models. Both contractors and agencies will understand the product better; thus the risks inherent in contracting will decrease for both parties. Greater innovation will follow, and new construction procedures and equipment will be introduced. Improved quality control and (nondestructive, end-result) acceptance procedures and equipment will be developed.

Increased product understanding will lead to increased use of innovative contracting procedures. Agencies will allow contractors to decide individually whether to bid on a project as a PRS or warranty project (or both). With parallel advances in asphalt paving, the potential for bid competition between hot-mix asphalt and portland cement concrete will exist.

The increased use of PRS and the gained product understanding will result in agencies establishing more cost-effective requirements that specify an optimum quality (or performance) level. Contractors will deliver a project-specific quality level further optimized with respect to their locus of control. Nationally, tremendous potential savings will be realized as project life-cycle costs are minimized. Just as QA specifications 30 years ago had the potential for future improvement and development, today's PRS present an even greater opportunity.

Table 1: Key Differences between Level 1 and Level 2 PRS

References

Peter A. Kopac is a research highway engineer on the Portland Cement Concrete Pavement Team of FHWA's Office of Infrastructure R&D. He has more than 30 years of highway-related experience, including 25 years with FHWA. His field experience with the enforcement of specifications led to his focus on research in materials and construction quality management. Kopac has assisted numerous agencies in developing, reviewing, and analyzing their quality assurance specifications.