|FHWA > Engineering > Pavements > FHWA-HIF-09-020 > Chapter 3|
Development and Implementation of a Performance-Related Specification for SR 9a, Florida: Final Report
Chapter 3: Development of the Performance-Related Specification
In developing the PRS for the SR 9A project, the latest FHWA procedures (Hoerner and Darter, 1999) and software (PaveSpec 3.0) were used. A level 1 (simplified) specification was chosen to minimize deviation from the department's existing specifications and testing practices and thus provide the best chance possible for successful implementation.
To begin the development process, much information about the department's current specifications and design criteria, construction sampling and testing techniques, pavement performance measures, and typical maintenance and rehabilitation strategies and costs was collected and carefully reviewed. This information, along with data specific to the SR 9A project, was used to create the framework for the specification and provide the necessary inputs to the PaveSpec program, which are provided in appendix A.
This chapter discusses in detail the various types of data collected in the study and how the data were used to develop the SR 9A PRS. It also presents the resulting PRS pay factor curves used in compensating the contractor for the level of quality achieved on the project. The final, binding version of the PRS, in the form of a Technical Special Provision, is provided in appendix B.
Selection of Acceptance Quality Characteristics and As-Designed Quality Levels
In the construction of its concrete pavements, the department calls for the inspection and testing of several quality characteristics. Among these characteristics are slump, air content, slab thickness, strength, dowel and tie bar placement, and surface smoothness. For the SR 9A PRS implementation, Florida DOT decided that three of the five AQCs considered by PaveSpec would provide the basis for concrete pavement pay adjustments. These AQCs included slab thickness, 28-day compressive strength, and surface smoothness, as determined using a California-type profilograph with a 0.2-in. (5 mm) blanking band.
To define for each AQC the levels of quality for which the department is willing to pay 100 percent of the bid price (i.e., target values) and the levels it considers unacceptable (minimum and maximum values), the department's applicable concrete specifications (Florida DOT, 2000) and design methodology were examined. In addition, actual AQC data from three nearby concrete paving projects completed in 2000 were obtained and analyzed. The sections below discuss how the gathered information was used to establish as-designed target values (i.e., mean and standard deviation) and corresponding rejectable and maximum quality limits (RQLs and MQLs) for each of the three AQCs included in the PRS.
Section 350-16 of the department's 2000 Standard Specifications discusses how slab thickness is measured and evaluated for acceptance. The specification requires that the contractor take cores at randomly selected locations, with each core representing no more than 2,500 yd2 (2,090 m2) of pavement area. The department determines the average thickness of pavement from the lengths of all cores taken from the entire job. In this computation, cores measuring more than 0.5 in. (12.7 mm) greater than the specified thickness are assigned a thickness equal to the specified thickness plus 0.5 in. (12.7 mm).
Areas of pavement found by the department to be deficient in thickness by more than 0.5 in. (12.7 mm) are handled in one of two ways. The first option allows the contractor to remove and replace the deficient area with concrete of the thickness shown in the plans. No compensation is given for the removal and replacement. The second option allows the contractor to leave the deficient pavement in place, but to receive zero compensation for the subject pavement area.
The final pay quantity is determined by multiplying the area of pavement to be paid for by the ratio of the average thickness to the specified thickness. This prorated amount of pavement is then multiplied by the bid unit price for concrete pavement. The final pay quantity is capped, however, by a maximum average of over-thickness of 0.25 in. (6.4 mm).
As discussed in chapter 2, the specified pavement thickness on the SR 9A project is 12.5 in. (317.5 mm). Because the department will not pay for, and may require replacement for, pavement that is more than 0.5 in. (12.7 mm) below the specified thickness, the department's RQL for slab thickness for the SR 9A project is assumed to be 12.0 in. (304.8 mm) (12.5 in. - 0.5 in. [317.5 - 12.7 mm]). This value was deemed appropriate by the department for use in the PRS.
Department specifications indicate that the MQL for thickness for the SR 9A project is 12.75 in. (323.9 mm) (12.50 in. + 0.25 in. [317.5 + 6.4 mm]). No additional bonus money is paid to the contractor for achieving an average thickness for the project greater than 12.75 in. (323.9 mm). For PRS development and implementation, however, the department determined that the MQL should be increased from 12.75 in. (323.9 mm) to 13.5 in. (342.9 mm) to allow for more incentive opportunity.
The logical target mean for thickness for the SR 9A project is represented by the specified thickness of 12.5 in. (317.5 mm). To determine the appropriate standard deviation target, slab thickness data from three previous SR 9A jobs (Financial Projects 20959315201, 20929615201, and 20929315201) were analyzed. These projects represented approximately 21 lane-miles (34 lane-kilometers) of mainline pavement, extending from MP 24.496 northeasterly to MP 20.917. For each project, only the core thickness measurements taken on mainline pavement (specified thickness of 12.5 in. [317.5 mm]) were evaluated.
Table 1 provides a statistical breakdown of the measured slab thicknesses for each project, while figure 4 shows the corresponding thickness distributions. Because of the unusually high variation in thickness for project 1, only the data from projects 2 and 3 were considered in establishing the target standard deviation. The weighted average thickness (based on number of independent cores) for these two projects was computed to be 12.67 in. (321.8 mm) and the standard deviation of the pooled variances of thickness was computed to be 0.49 in. (12.5 mm) Based on these results, the department recommended establishing the target standard deviation at 0.50 in. (12.7 mm). These are the target mean and standard deviations for which the department is willing to pay 100 percent bid price.
Acceptance sampling and testing protocol and requirements for concrete strength are provided in Sections 347-4 and 347-5 of the department's 2000 Standard Specifications. According to the protocol, at least one representative sample of concrete must be obtained from each day's production of each design mix from each production facility. From that sample, the contractor must cast four concrete cylinders, 6 in. (152.4 mm) in diameter by 12 in. (304.8 mm) long. Two of the cylinders must then be tested for compressive strength 7 days after casting, while the other two cylinders must be tested 28 days after casting. For each pair of cylinders tested, the average compressive strength is determined. Concrete below the 28-day minimum compressive strength requirement of 2,700 lbf/in2 (18.62 MPa) is subject to removal and replacement by the contractor. This strength value represents the department's existing RQL, and the department recommended that it be applied to the SR 9A PRS.
A corresponding MQL for strength was found to not exist. However, based on the department's target for strength and the variability of strength observed in past projects (see discussion below), the department determined that 5,500 lbf/in2 (37.92 MPa) would be a suitable MQL value for the 9A PRS.
The Florida DOT's current procedure for designing JPCPs is based on the 1993 AASHTO Design Guide. The procedure and the standard design input values used by the department are presented in its 1996 Rigid Pavement Design Manual. In this manual, the design concrete strength is represented by the 28-day modulus of rupture determined through third-point loading. The standard design value is given as 4,400 kPa (638 lbf/in2). Using the following equation for converting flexural strength to compressive strength, the corresponding 28-day design compressive strength was computed to be 4,510 lbf/in2 (31.10 MPa):
MR,28-day = 9.5 * (f'C,28-day)0.5 Eq. 2
MR,28-day = Estimated modulus of rupture at 28 days, lbf/in2.
f'C,28-day = Estimated compressive strength at 28 days, lbf/in2.
For PRS purposes, the target mean for compressive strength was set at 4,500 lbf/in2 (31.03 MPa).
Evaluation of 28-day compressive strength data on cylinders tested in the three previous SR 9A projects yielded the strength statistics listed in table 2 and the strength distributions shown in figure 5. Again, because of the unusually high variation in strength for project 1, only the data from projects 2 and 3 were considered in establishing the target standard deviation. The weighted average strength (based on number of pairs of cylinders) for these two projects was computed to be 5,548 lbf/in2 (38.25 MPa), and the standard deviation of strength was computed from pooled variances to be 610 lbf/in2 (4,206 kPa). Based on these results, the department recommended establishing the target standard deviation at 610 lbf/in2 (4,206 kPa). As previously stated, these are the means and standard deviations for which the department is willing to pay 100 percent of bid price.
Sections 350-14 and 352-4c of the department's 2000 Standard Specifications describe how concrete smoothness is tested and evaluated for acceptance. The procedure requires that the contractor furnish and operate an electronic California-type profilograph along each wheel path of each traffic lane longer than 250 ft (76.2 m). The profilograph must be capable of producing profile traces and computing profile index (PI) based on a 0.2-in. (5 mm) blanking band (herein denoted as PI0.2-in).
Profilograph test results are examined by the department's field engineer. Individual high points in excess of 0.3 in. (7.6 mm) per 25-ft (7.6 m) length are identified for grinding, and the average PI0.2-in for each 0.1-mi (0.16 km) section is computed using the left and right wheel path PI0.2-in values. Each 0.1-mi (0.16 km) tangent or slightly curved (centerline radius of curvature ≥ 2,000 ft) (609.6 m) section with an average PI0.2-in greater than 7 in./mi (111 mm/km) must be corrected by the contractor via grinding. Contract unit price adjustments for smoothness, prior to grinding, are made according to the schedule shown in table 3.
The information in table 3 indicates that the department's RQL and MQL values for smoothness are 7 in./mi (111 mm/km) and 3 in./mi, respectively. These values were deemed appropriate for use in the SR 9A PRS. Table 3 also shows that the DOT's target mean smoothness (prior to grinding) is 5.5 in./mi (87 mm/km), which is the midpoint of the range (5.0 < PI0.2-in ≤ 6.0) that corresponds to 100 percent payment. However, because the SR 9A contract was let with the requirement that all concrete pavement be diamond ground and that contractor bid prices for concrete pavement include the cost of grinding, a different target mean was sought for the PRS.
After-grinding smoothness data for the three previous SR 9A jobs were examined for this purpose. Table 4 shows a statistical breakdown of the measured PI0.2-in values for several 0.1-mi (0.16 km) test segments from each project, while figure 6 shows the corresponding PI0.2-in distributions. The weighted average PI0.2-in (based on number of 0.1-mi (0.16 km) test segments) for these three projects was computed to be 2.7 in./mi (42 mm/km), and the pooled standard deviation was computed to be 1.2 in./mi (19 mm/km). Based on these results, the department recommended establishing the PRS target mean at 3.0 in./mi (47 mm/km) and the target standard deviation at 1.0 in./mi (16 mm/km).
Summary of Acceptance Quality Characteristics Target Values, Rejectable Quality Levels, and Maximum Quality Levels
Table 5 summarizes the target means and standard deviations established for each AQC for the SR 9A PRS. It also lists the established RQLs and MQLs for each AQC. These values apply to each lot of concrete pavement.
Pavement Performance Indicators and Models
The Florida DOT monitors JPCP performance through annual visual distress surveys and ride quality tests. The distress surveys identify the amount and severity level of up to 10 different surface distress types, including slab cracking, joint faulting, and joint spalling, that have developed over time and through the loss of smoothness over time. Smoothness is measured with an inertial profiler and is reported in terms of the International Roughness Index (IRI). The collected distress and smoothness data are entered into the department's pavement management system, which is used to track deterioration rates and predict future conditions and corresponding rehabilitation needs.
For the SR 9A PRS, all four performance indicators-slab cracking, joint spalling, joint faulting, and smoothness-available in PaveSpec 3.0 were selected for predicting pavement service life. In addition, the PaveSpec default performance models linking the three AQCs (thickness, strength, and smoothness) with the four performance indicators were selected for developing the PRS pay factor equations.
Constant Input Values
Constant inputs represent those PaveSpec parameters that do not differ between as-designed and as-constructed pavements. They include various design, traffic, and climatic parameters, as well as the maintenance and rehabilitation strategies and costs used to compute LCCs and corresponding pay factor amounts.
Table 6 lists the constant input values established for the SR 9A PRS. Many of these values were defined in the contract plans, while others represent standard values given in the department's rigid design manual.
Climatic data were derived from two sources: the NOAA 1983 Climatic Atlas of the United States, which includes statistics based on roughly 30 years of U.S. weather data, and the FHWA LTPP database, which includes weather statistics for thousands of test pavements in the United States and Canada. For this latter source, climatic data from three LTPP test sections in the Jacksonville area and covering the last 15 to 20 years were analyzed. The climatic values shown in table 6 represent the best estimates for the SR 9A project.
Maintenance and Rehabilitation Strategies and Costs
The Florida DOT exercises several different options for maintaining and rehabilitating concrete pavements. They include various concrete pavement restoration activities, such as joint resealing, slab replacement, edge drain installation, and diamond grinding, and more extensive measures, such as conventional asphalt concrete (AC) overlays and AC overlays over cracked-and-seated PCC.
Based on discussions with key DOT staff, the following maintenance and rehabilitation activities were established for use in the SR 9A PRS:
Maintenance Plan Summary
Localized Rehabilitation Plan Summary
Sublot Failure Thresholds
If 25 percent of the sublots have failed, apply the global rehabilitation procedures listed in table 7.
Unit cost data, shown in table 8, were provided by Florida DOT in 2001 dollars. Definitions for the cost items are shown below.
Sampling and Testing Methods
As discussed previously, existing department specifications require the following:
Under the PRS concept, pay adjustments are made on a lot-by-lot basis, with a lot being defined as a discrete quantity of constructed pavement having the same mix design, material sources, and design characteristics (e.g., joint spacing, drainage, dowel bar size) and subjected to the same climatic, traffic, and support conditions. The size of a lot is one lane in width and between 0.1 and 1.0 mi (0.160 and 1.61 km) long. Each lot is divided into sublots of approximately equal surface area, and all sampling and testing of concrete AQCs is performed at the sublot level.
For the SR 9A PRS, a minimum sublot length of 250 ft (76.2 m) was established, corresponding to the department's existing procedure for testing smoothness. In each sublot, it was determined that (a) two core borings be taken at random locations after 3 days for slab thickness measurement, (b) two cylinders be cast from one truck within the sublot and be tested for compressive strength after 28 days, and (c) profilograph traces be taken for each wheel path. This defined sampling frequency is illustrated in figure 7, along with the layout of lots and sublots.
It can be seen that the proposed PRS requires minimal changes to the department's existing sampling and testing procedures. The main requirement is that a complete set of AQCs be taken from each sublot to facilitate PRS performance projection.
Table 9 shows the testing methods associated within the PRS and FDOT's existing construction specifications for concrete strength, slab thickness, and initial smoothness. The testing methods for these AQCs are discussed further in the following sections
Concrete strength-The cylindrical specimens shall be molded and cured in accordance with FM 1-T 023 (Making and Curing Test Cylinders) and tested in accordance with FM 1-T 022 (Testing Cylinders), standard Florida Test Methods. Improper sampling, molding, handling, and curing will be handled according to FDOT's existing specifications.
Slab thickness-Thickness cores shall be a minimum diameter of 2 in. The slab thickness at a cored location shall be recorded to the nearest 0.1 in. (25.4 mm) as the average of three caliper measurements of the core length. The three measurements shall be obtained and marked at locations spaced at approximately equal distances around the circumference of the core.
Initial smoothness-The pavement surface smoothness shall be tested using an electronic model of the California profilograph with 0.2-in. (5.1 mm) blanking band. The smoothness testing shall be conducted after the concrete cures and grinding have been completed. Pavement profiles shall be taken at the traffic wheel paths (3 ft [0.9 m] from and parallel to each edge of pavement placed at 12-ft [3.66 m] width, or less). When pavement is placed at a greater width than 12 ft (3.7 m), the profile will be taken 3 ft (0.9 m) from and parallel to each edge and each side of the planned longitudinal joint. When the pavement being constructed is contiguous with an existing parallel pavement that was not constructed as a part of this contract, the profile parallel with the edge of pavement contiguous with the existing pavement shall not be taken. The profile shall be started and terminated 15 ft (4.8 m) from each bridge approach or existing pavement that is being joined.
Development of Pay Factors for the SR 9A Project
Using the PaveSpec 3.0 software program and the various inputs discussed throughout this chapter, a set of concrete thickness, strength, and smoothness pay factors were developed for use in the SR 9A project. These resultant pay factors for slab thickness are shown in table 10. These factors are also illustrated in figure 8. The lowest noted pay factor is 93.67 percent for the RQL (12.0 in. [304.8 mm]) with a high lot standard deviation (2.0 in. [51.8 mm]). When the mean slab thickness reaches the MQL of 13.5 in. (342.9 mm), with an ideal standard deviation of 0.0 in., the pay factor is 104.26 percent. For the target standard deviation, the pay factor between the RQL and the MQL varies 9.09 percent. There is little increase in pay factor for variability less than the target value. Pay factors for standard deviations below the target value decrease at about twice the rate of pay factor increases for standard deviations above the target. The slab thickness pay factor curves are fairly flat due to the conservative design of 12.5 in. (317.5 mm) resulting from the AASHTO design procedures.
Pay factors for strength are shown in table 11 and figure 9. Obviously, PCC strength plays an important part in long-term pavement performance, particularly on the low side of the target. As a result, the pay factors at the RQL and MQL with target standard deviations range 50.7 percent, from 57.4 to 108.1 percent. For each incremental change in standard deviation from the target value, the pay factor changes about two times as fast for higher standard deviations compared with lower standard deviations.
Computed surface smoothness PI pay factors are shown in table 12 and figure 10. The range of pay factors between the RQL and the MQL for the target standard deviation is 9.89 percent (93.59 to 103.48). Variability within the range of 0 to 3 in./mi (0 to 47 mm/km) has greater effect on the pay factors. These curves were developed with 5 percent user costs. If a greater amount had been used, the curves would have been steeper.