Performance-related specifications (PRS) have been developed for jointed plain concrete pavements (JPCP), based on prototype specifications developed under a previous Federal Highway Administration (FHWA) research project.^(1,2,3) Even though many useful and innovative PRS concepts were developed under the earlier project, it was recognized that the prototype specifications had many limitations that made their use impractical. Therefore, much of the research conducted under this project focused upon revising the prototype specifications to make them more practical, acceptable, and implementable.

Under this project, a revised, improved PRS prototype has been developed. In an attempt to bridge the gap between currently used construction specifications and fully implementable PRS, three different specification levels (simply titled Level 1, Level 2 and Level 3) were developed. The revised specifications address two of the three specification levels (Levels 1 and 2) and are presented in appendix A. Level 3 is considered a futuristic specification that includes rapid nondestructive testing for all acceptance quality characteristics (AQC's).

Background

Over the past 25 years, there has been a growing interest in the development of PRS for highway pavement construction. Currently, PRS are defined as specifications that describe the desired levels of key materials and construction AQC's (e.g., concrete strength, slab thickness, and initial smoothness) that have been found to correlate strongly with fundamental engineering properties that predict pavement performance.⁽⁴⁾ True PRS not only describe the desired AQC quality levels but use the measured AQC quality to predict subsequent pavement performance, thus providing the basis for rational acceptance and price adjustments.⁽⁴⁾ Therefore, the PRS methodology can be used to relate quality of pavement construction to the estimated performance and life-cycle costs (LCC's) of a given pavement lot. This ability leads to the identification of optimum levels of AQC construction quality and the rational determination of pay adjustments for specific pavement lots.^(1,2,3)

The basic concepts making up PRS are not new, as they are essentially improved quality assurance (QA) specifications. QA specifications and PRS are similar in that they specify the desired product quality rather than the desired product performance. However, in PRS, when you specify quality, you know what performance you are specifying. Another major difference comes from the methods used to determine the overall lot pay adjustment. Conventional QA acceptance plans use engineering judgment to establish individual AQC pay adjustments (and weighting factors for each) for determining the overall price adjustment for the lot⁽⁵⁾ PRS, however, use mathematical models (taking AQC values into account) to estimate future pavement performance and corresponding LCC's to compute one overall price adjustment.⁽⁵⁾

Another advantage of PRS is their ability to identify desirable levels of AQC quality that provide a desired pavement performance. Most current specifications are unable to identify the true value of the measured AQC quality that is being produced. Using the PRS methodology, for a specified AQC (say, concrete strength), the performance of the pavement can be estimated using established relationships. Therefore, the desired concrete strength quality can be determined based on observing the range of performance coming from specifying different values for concrete strength. A more focused investigation of this kind will allow an agency to identify the optimum levels of AQC quality that provide the best balance between costs and performance for a given project. It has been estimated that such improvements in quality control (QC) procedures could produce long-term benefits to agencies and users measured in billions of dollars.⁽⁶⁾

PRS also provide a rational basis on which pay adjustments can be determined. Using PRS, lot pay adjustments are based on the difference between the LCC's associated with the target (as-designed) pavement and those associated with the actual (as-constructed) pavement. First, the agency chooses AQC target values (desired material and construction AQC quality levels). These AQC targets are used to predict the future performance (using mathematical distress prediction models) and the associated estimated future LCC's defining the as-designed pavement. (Note: The future LCC's include those maintenance and rehabilitation costs expected to be incurred by the agency [maintenance and rehabilitation of the pavement] and potential users [user costs may be included by the agency] over the life of the project, assuming a given rehabilitation policy.) The estimated LCC's corresponding to the as-designed AQC quality are then summarized into one overall LCC (LCC_DES) representing the AQC quality of the as-designed pavement. Next, the as-constructed AQC quality is measured at the time of construction on the in situ pavement. The measured as-constructed AQC values are used to predict the future pavement performance and associated LCC's (using the same distress indicator and cost models used in the determination of LCC_DES) defining the as-constructed pavement. The estimated LCC's corresponding to the measured as-constructed AQC quality are then summarized into one overall LCC (LCC_CON) representing the AQC quality of the as-constructed pavement.

An incentive pay adjustment is computed if the as-constructed AQC quality is measured to be better than the agency-specified target values (due to an increase in pavement life, resulting in a corresponding decrease in LCC's). Conversely, a disincentive pay adjustment is computed if the as-constructed AQC quality is measured to be poorer than the agency-specified target values (due to a decrease in pavement life, resulting in a corresponding increase in LCC's).^(1,2,3) The amount of the pay adjustment (incentive or disincentive) is based on the direct comparison of LCC_DES and LCC_CON.

The development of PRS requires a thorough understanding of how material and construction AQC's affect the performance and costs of a pavement. Unfortunately, the effects of all material and construction AQC's on performance are not well understood. At present, only one agency is known to use PRS. However, research and development progress has been made, sponsored by FHWA, on the development of PRS for both asphalt concrete (AC) and portland cement concrete (PCC) pavements. This report documents the latest development for PCC pavements and represents a product that is ready for implementation.

Some of the recognized benefits of PRS to a State highway agency (SHA) are the following:

• PRS require the establishment of clear AQC target values (means and standard deviations) that define the pavement quality for which the agency is willing to pay 100 percent of the contractor bid price.
  
  
• PRS provide a straightforward method for determining rational (LCC-driven) pay adjustments (incentives and disincentives) that are applied when a higher or lower level of quality (as compared to the chosen AQC target values) is produced by the contractor.
  
  
• PRS relate the quality of pavement construction to the performance and subsequent LCC's of a given pavement lot. This ability provides the opportunity to identify optimum levels of AQC construction quality that would minimize LCC's while maintaining the desired performance.
  
  
• The clear and rational methodology of PRS (including fair incentives and disincentives during construction) is expected to lead to significantly improved highway construction quality levels, better pavement performance, and reduced pavement LCC.

Project Objectives and Scope

The focus of this study was to conduct research that will continue the advancement toward the development of a fully practical and implementable PRS for PCC pavement construction. Specifically, the contract objectives were the following:

1. Develop a detailed field/laboratory plan and conduct the required testing to establish additional (as well as improve existing) relationships between PCC pavement construction AQC's and pavement performance (distress indicator models).
   
   
2. Develop a detailed field/laboratory plan and conduct required testing to establish the typical variabilities associated with the AQC's chosen for inclusion in the revised prototype PRS. This field/laboratory work is to supplement findings from an extensive literature search.
   
   
3. Develop improved prototype PRS (Level 2) through a critical review and refinement of the initial draft prototype and the addition of performance-related construction criteria.
   
   
4. Develop a simplified PRS (Level 1) methodology that can be immediately implemented by SHA's with minimal changes to their existing sampling and testing plans. Demonstrate this Level 1 methodology for one SHA by developing recommended acceptance plans with price adjustments.
   
   
5. Demonstrate the Level 1 and Level 2 PRS methodologies on an actual construction project to obtain experience, verify/confirm their effectiveness, identify potential problem areas, and assess their reasonableness. This field trial is labeled as a shadow field trial because the PRS are not the governing specifications. SHA acceptance procedures and contractor pay are not in any way affected by the shadow field trial.
   
   
6. Provide training to personnel from FHWA's Office of Technology Applications (OTA), enabling them to conduct three additional shadow PRS field trials during the fiscal year 1997. Provide as needed assistance to OTA by participating in pre-construction meetings, assisting during the conduct of the field trials, and presenting the field trial results at State follow-up meetings.
   
   
7. Obtain data from SHA's in order to perform comparative analyses of actual price adjustments awarded to the contractor versus those that would theoretically have been assessed if PRS Level 1 had governed the projects.

Advisory Panel

An experienced advisory panel was assembled to provide guidance to the research team in addressing the many complex issues of the project. The panel was comprised of representatives from SHA's, industry, PCC construction, and academia who were collectively knowledgeable in the areas of PCC pavement design, PCC pavement construction, PCC materials, statistics/variability, and specification development. The advisory panel provided valuable assistance to the research team by conducting comprehensive reviews of the prototype specifications and approach, providing input into the selection of potential AQC's for investigation in the laboratory/field investigations, and reviewing all pertinent project documentation. Members of the advisory panel included:

• Jim Hall, Illinois Department of Transportation.
  
  
• Steve Horton, Colorado Department of Highways.
  
  
• Ken McGhee, Consultant.
  
  
• Lee Powell, Ballenger Paving.
  
  
• Doug Schwartz, Minnesota Department of Transportation.
  
  
• Clint Solberg, American Concrete Pavement Association.
  
  
• Garland Steele, Consultant, Steele Engineering, Inc.
  

Research Approach

The research team began by conducting a comprehensive literature search to identify previous studies where variations of concrete pavement AQC's had been measured and documented in an unbiased manner. Eleven different potential AQC's were targeted in the search. These included the following:

• Concrete strength.
  
• Slab thickness.
  
• Entrained air content.
  
• Initial smoothness.
  
• Percent consolidation around dowels.
  
• Tie bar depth.
  
• Sawcut depth.
  
• Sawcut timing.
  
• Dowel bar misalignment.
  
• Effectiveness of curing.
  
• Depth of tining.

The results of the literature search (specifically, the assessment of the amount of available variability data for each of the investigated AQC's) were used to identify and prioritize needed supplemental laboratory/field investigations. As a result of the data collected in the literature search, laboratory/field variability investigations were selected to collect additional data for the following six AQC's only:

• Concrete strength.
  
• Slab thickness.
  
• Entrained air content.
  
• Percent consolidation around dowels.
  
• Tie bar depth.
  
• Sawcut depth.
  

• Establishment of guidelines for levels of as-designed variability, based on typical values of as-constructed variability.
  
  
• Development of new construction-related performance prediction (distress indicator) models, or improvement of existing models.

Based on preliminary variability-related data collection results, it was determined that there was insufficient sawcut depth and tie bar depth data to develop construction-related performance models for these AQC's. Therefore, the model development/improvement tasks under this contract consisted of the following:

• Laboratory analysis of concrete strength, including relationships between flexural and compressive strength, maturity methods, and the prediction of 28-day strengths from early age strengths (3 to 5 days).
  
  
• Modification/calibration of an existing transverse joint spalling model to include air content and concrete strength.
  
  
• Modification of an existing transverse joint faulting model to incorporate the effects of the percent consolidation at doweled joints.

Capabilities of Developed PRS

Drawing upon the comprehensive reviews of the prototype PRS by the advisory panel members, the results of the literature search, and the results of the extensive laboratory/field investigations, the prototype PRS were revised into more practical and accurate specifications. Some of the improvements to the prototype PRS include the following:

• Inclusion of percent consolidation around dowels as a construction-related AQC.
  
  
• Inclusion of improved distress indicator models for transverse slab cracking, transverse joint faulting, transverse joint spalling, and present serviceability rating (PSR) (an indicator of smoothness), as well as the inclusion of an international roughness index (IRI) model for predicting smoothness over time.
  
  
• More available maintenance and rehabilitation (M & R) activities available for inclusion in a defined M & R plan.
  
  
• Clearer definition of lots and sublots.
  
  
• Recommendations for typical variabilities of the included AQC's.
  
  
• Improved sampling and testing plan recommendations, including the ability to have different numbers of samples per sublot for each AQC.
  
  
• Improved strategies for measuring and accepting concrete strength.
  
  
• Development of a practical and readily implementable PRS (Level 1).

A simplified version of the PRS was developed and included as part of the revised PRS. This simplified methodology (Level 1) uses the basic concepts of the original prototype PRS; however, because of a few important characteristics, it is believed to be more practical and readily implementable by a proactive SHA. One important characteristic of the Level 1 approach is that a SHA can most likely use its current sampling and testing methods with only slight modifications. The overall Level 1 lot pay factor is also determined in a more straightforward manner than the method used in the prototype. Individual pay factors are determined for each AQC using predetermined curves or equations. The overall lot pay factor is then computed using a simple mathematical function (composite pay factor [CPF] equation) of the individual AQC pay factors. The computed Level 1 CPF is actually an estimate of the pay factor computed using the Level 2 method. Because the simplified Level 1 PRS procedure is a practical approach that is immediately implementable, it is a very useful first step in introducing general PRS concepts to a SHA.

PRS Computer Software

The original PRS computer software, PaveSpec was also greatly improved under the current contract. The revised software, PaveSpec 2.0, offers much more functionality than its predecessor and offers a much more user-friendly environment. Specific improvements in PaveSpec 2.0 include:

• The ability to demonstrate both Level 1 and Level 2 specifications.
  
  
• The ability to enter actual field data and determine acceptance pay factors.
  
  
• The ability to have a different number of samples per sublot for each AQC.
  
  
• Automatic determination of random sample locations.
  
  
• A more robust M & R plan.
  
  
• The inclusion of improved distress indicator models.
  
  
• The ability to investigate a specification prior to going to the field (using simulated data).
  
  
• The ability to create and save groups of variables as named modules. The following modules are used in PaveSpec 2.0: Design Traffic, Climatic Data, M & R Plan, Pavement Design, and Unit Costs.
  
  
• The inclusion of a sensitivity analysis section.
  
  
• Improved tabular and graphical output.
  
  
• Improved user's guide.

Field Demonstration Trials

The Level 1 and revised Level 2 specifications were both demonstrated under this project using three different investigation methods.

The first method consisted of demonstrating both specification levels on four actual construction projects (one conducted by the project research team and three by OTA personnel). The purpose of these field trials was to obtain experience, verify/confirm each specification's effectiveness, identify potential problem areas, and assess the overall reasonableness of both specification levels (Level 1 and Level 2).

Second, Level 1 pay factor curves (and corresponding equations) were developed for three typical PCC pavement designs for a chosen SHA. The three typical designs were developed to represent medium, heavy, and very heavy traffic conditions. An analysis of the observed pay factor trends (within and between typical designs) was completed and presented.

Third, historical data were collected from SHA's in order to perform comparative analyses of the actual price adjustments awarded to the contractor versus those that would theoretically have been assessed if the Level 1 PRS had governed the projects. Data were collected for a total of 41 lots (7 PCC projects) from 3 different SHA's.

The results of these demonstrations are summarized and presented in volume II (appendix B).

Sequence of Report

This volume is designed as a stand-alone document highlighting the practical output of this work. It will assist a pavement owner in using the PRS (included in appendix A in this volume) for the acceptance of as-constructed pavement lots.

• Chapter 2 provides a brief definition and summary of the general evolution of PRS.
  
  
• Chapter 3 contains a summary of, and quick reference to, many of the important PRS-related terms.
  
  
• Chapter 4 introduces the key PRS concepts used in the development and application of the current revised PRS prototype.
  
  
• Chapter 5 provides guidelines and recommendations for assisting the agency in developing a project-specific PRS (this includes guidance in making the many general PRS decisions, as well as in selecting appropriate values for the numerous required PRS inputs).
  
  
• Chapter 6 provides guidelines for making general pay adjustment decisions.
  
  
• Chapter 7 outlines the step-by-step procedures required to generate the preconstruction output for both Level 1 and Level 2 specification types.
  
  
• Chapter 8 presents the step-by-step procedures used to determine contractor pay adjustments for as-constructed pavement lots.
  
  
• Chapter 9 provides a summary of the research conducted and recommendations for future research.

Several appendixes are included in support of the final report. Appendix A (presented in this volume) contains the complete stand-alone revised PRS prototype (for both Levels 1 and 2). Appendix B (volume II) contains a complete description of all demonstrations of the revised PRS prototype, including:

• Documentation of the four shadow field trials.
  
  
• Development of the Level 1 pay factor curves for three typical designs in a chosen SHA.
  
  
• Comparison of actual pay vs. estimated PRS pay conducted using historical data.
  
  

  Appendix C (volume III) provides a summary of the literature search. Appendix D (volume III) contains details of the laboratory and field studies conducted to investigate typical AQC variability. Appendix E (volume III) presents details of the distress prediction models (including improvements made under this project) used in the PaveSpec 2.0 computer software. Appendix F (volume III) contains an annotated bibliography of the sources reviewed during the conduct of the literature search. Appendix G (volume IV) is the user guide for the PaveSpec 2.0 computer software that was revised under this project.