|FHWA > Engineering > Pavements > FHWA-HIF-09-020 > Chapter 5|
Development and Implementation of a Performance-Related Specification for SR 9a, Florida: Final Report
Chapter 5: Evaluation of the Performance-Related Specification
The success and effectiveness of the PRS project was discussed with the contractor and the FDOT staff who participated in the PRS implementation. The discussion included questions assessing the functionality of the PRS, any related problems encountered in the process, and changes that were made in response to the PRS. Results of general questions indicate that the PRS documents were adequate, the PRS concept was desirable, and PRS implementation was not difficult.
Interviews were completed with a representative of the paving contractor (McCarthy Improvement Company). The following comments and recommendations were received from the contractors.
Nick Wolf described the revised methods, advantages, and disadvantages of the PRS project in a recent article (Feingold, 2004): "We're using a little more quality control and we were a lot more careful before starting. On this project we tried to exceed our normal high standards for workmanship. To do this took a little extra planning to make sure none of the little things were missed." He indicated that the management staff met to specifically consider the PRS project and schedule prior to construction.
Wolf indicated some benefits and difficulties that the contractor experienced when following the PRS (Feingold, 2004). "It can be good for the contractor, because it allows the contractor to do whatever they want as long as they meet the specs provided. There's no prescribed way on how to do things, so that can make getting started a little harder." "They're tough specs," he said. "They strive for the best. To meet them, you have to do your best work. They don't give you the leniency some other specs give you, but they provide you with a bonus for doing better work."
Tony Dimaggio, Manager of Technical Services for Florida Businesses at concrete supplier Tarmac Titan America, described the concrete mix supplier's perspective on the PRS project (Feingold, 2004):
Instead of giving a spec saying "you're going to put in this much cement and get this strength," [FDOT] said, "What we really want is a smooth pavement and durability." Our ready mix people put together a concrete mix that would meet the specs for minimum strength, but looked for something that was very workable for the smoothness they needed and the early strength they wanted on the job. We were really excited when we heard part of this job was a PRS, but the parameters they wanted resulted in a concrete mix that wasn't that different from the rest of the job. A real performance-based specification would be when the owner says, "This is what I want and you figure out how to do it." That gives the ready mix people and the contractors an opportunity to come up with a concrete that will work and save them some money.
Construction Management Assessment
Interviews were conducted and surveys were received from representatives of FDOT's consultant CEI team (PTGcsc/JEAces). Greg Graden of JEAces served as the roadway project engineer on the pavement construction. In an article in the June 2004 issue of Florida Concrete, Graden provided several comments regarding PRS (Feingold, 2004):
The main objective of these PRS is to provide the agency with a methodology to assure that the design assumptions are being fulfilled, promote high quality construction, and to protect the agency from poor workmanship…. At the same time it allows the contractor the maximum freedom in deciding how to perform the construction. PRS provide rational methods for contract price adjustment based on the difference between the as-designed and as-constructed life-cycle costs of the pavement.
In the same article, Brett Pielstick, senior project engineer with PTGcsc, indicated, "It [PRS] does provide an opportunity for the contractor to be rewarded for doing good work or get penalties for poor workmanship or poor quality of material" (Feingold, 2004).
Graden and Ted Worthington (PTGcsc) agreed on the following conclusions and suggestions:
Florida Department of Transportation Assessment
FDOT engineers who had participated in the design and implementation of the PRS project responded as follows:
Bouzid Choubane, FDOT: "The results look promising. However, because of the project constraints, the performance data may be too limited to provide for enough quantitative information on conclusively assessing the subject PRS methodology. Further validation and refinement are needed on a larger scale."
Carrie A. Stanbridge, FDOT resident engineer / D2: "Glad to see the final product. It was also good to finally see the reason for the selection of the compressive strength target value."
Mike Bergin, FDOT:
I suggest that a compressive strength window be required as opposed to a minimum required compressive strength. By this I mean that the contractor should be required to meet the compressive strength but not exceed it by too much. When the 28-day strength is substantially higher than the minimum required it normally means that the concrete was batched to produce a high early strength. The high early strength allows the contractor to get on the new concrete quickly but may result in shrinkage cracking which shortens the service life of the pavements. For instance, I suggest a target 28-day compressive strength of say 5,500 lbf/in2 [37.92 MPa] (plus or minus 500 lbf/in2 [3,447 kPa] at 28 days. If the contractor can control this, it will lower initial heat and provide a concrete with a slightly lower MOE [measure of effectiveness]. The combination of these will provide a more durable and longer lasting pavement.
Quantitative assessment of the effectiveness and usefulness of the PRS process can be accomplished by comparing the final PRS pay factors and payments against the factors that would have been implemented under the standard Florida DOT specification. In addition, the cost effectiveness of PRS specification can be summarized and compared against an independent analysis method.
Comparison of Performance-Related Specification and Florida Department of Transportation Standard Specification Results
The quality levels used by the PRS and FDOT standard specifications are summarized in table 19. They include only slight differences in the specified quality requirements. Differences between the specifications are more evident in figures 17, 18, and 19, which summarize the pay factors associated with each specification. While the FDOT standard and PRS strength pay factors are similar, the FDOT thickness standard does not provide additional incentive for thicknesses above 12.8 in. (325.1 mm). FDOT smoothness pay factors are much more lenient with un-ground surfaces than the PRS pay factors that are based on fully ground surface measurements.
The FDOT contractor constructed the PRS sections with strength, thickness, and smoothness levels at a higher quality than the target values, as shown in table 20. Following the PRS specification, FDOT awarded an incentive for the PRS sections of 110 percent. This was based on the conclusion from the PRS models that the future LCC of the sections will be improved (reduced) by about 115 percent. Under the FDOT standard specifications, the incentive award for these sections would also have been 110 percent of the bid price.
Closer inspection of the quality levels and pay factors provides additional information. For example, figure 20 shows the lot mean and standard deviation PCC strength range for each lot. The average strength of lot 1 is higher than that of lot 3, and their standard deviation levels are similar. Because of the higher strength, the pay factor for lot 1 is greater than that of lot 3. Variability also has an effect on pay factors, as can be seen by comparing the data from lots 1 and 4. The mean strength for these lots is very similar, but the standard deviation of lot 4 is about four times less than that of lot 1. This difference in variability resulted in the pay factor for lot 4 being about 1 percent greater than that for lot 1. Thus, both the mean and standard deviation of the AQC affect the incentive pay factors.
Similar observations from the PRS thickness data shown in figure 21 can be determined from the data for lots 1 and 2 (average value changes) and lots 3 and 5 (standard deviation changes). The effect of the reducing the standard deviation from 0.7 (lot 5) to 0.2 (lot 3) only increased the pay factor by 0.13. At these lower levels of variability, the effect of such changes is minimal. Figure 8 also helps to illustrate this effect.
Figure 22 shows the PI0.2in smoothness values and pay factors for each lot. The large effect of smoothness level on pay factor is evident, especially when comparing lot 1 with the other lots. Variability in lot smoothness data has only a small effect on pay factors, unless the standard deviation is closer to 0 in./mi.
High PCC strength levels played the largest role in increasing the overall pay factor levels, as shown in figure 23. With the exception of lot 1, which was low in all three AQCs, the unlimited overall pay factors were consistent. Because the average unlimited pay factor was 114.8 percent, this exceeded the specified 110 percent pay factor limit. Therefore, the improved pavement characteristics provided an estimated 14.8 percent improvement in LCC for which FDOT reimbursed the contractor with a 10 percent incentive bonus.
Independent Assessment of Effect on Pavement Life
An independent method was also used to compare the pavement life for both the target lot and the as-built lot AQCs. The NCHRP 1-37A mechanistic-empirical design and analysis software was used to predict the performance of the target (or as-designed) and the as-built JPCP (ARA, 2004). The distress and smoothness models in this software have been nationally calibrated under NCHRP 1-37A, which included several concrete pavement sections from Florida and should be reasonably applicable to the Jacksonville area. All inputs were held constant, while the three AQCs (thickness, smoothness, strength) were changed to the average achieved during construction, and the performance predictions of slab cracking, joint faulting, and IRI were estimated for the target and as-built pavement. Results indicated that the JPCP with target AQCs had a life of over 60 years due to the conservatism built into the project. The as-constructed pavement had an increased life of 16 percent. This independent method confirms that an increase in initial cost of 110 percent can be expected to achieve an approximate increase in pavement life of at least 16 percent.
Information was developed that shows the risks of using the sampling and testing plan to both the agency and the contractor. The PaveSpec software provides expected pay charts that are graphical representations of an acceptance plan that show the relation between the actual quality of a given lot and the pay the contractor can expect to receive (on average) for submitted lots of that quality. Figures 24, 25, and 26 are provided in this section to show the expected pay for strength, thickness, and smoothness or PI, using the target standard deviations.
For example, the strength data in figure 24 provide useful information. If the contractor produces a lot with exactly the target mean strength of 4,500 lbf/in2 (31.03 MPa) and standard deviation of 610 lbf/in2 (4,206 kPa), the probability of acceptance with, say, 100 percent pay or better is 50 percent. If the contractor desires a higher probability to achieve an incentive, the mean strength of the lot could be increased to, say, 4,850 lbf/in2 (33.44 MPa). The probability of acceptance with at least 100 percent is then 95 percent. Similarly, figure 25 indicates that the contractor must increase the mean concrete thickness to 12.75 in. (323.9 mm) to achieve a 95 percent probability of receiving a pay increase. Likewise, the contractor must achieve an average PI of about 2.5 in./mi (39.4 mm/km) to expect a pay increase with a 95 percent probability, as shown in figure 26.
The contractor also can affect the probability of receiving a pay increase by reducing or increasing the variability (standard deviation) of the quality levels in the project. For example, in figure 24, the contractor had a 50 percent probability of receiving incentive payments if the average slab thickness was 12.5 in. (317.5 mm) and the standard deviation was 0.5 in. (12.7 mm). If the contractor constructs a lot with the same average slab thickness and a standard deviation of 1.0 in. (25.4 mm), the probability of incentive award is reduced to 45 percent. Obviously many other statements could be created to analyze the risks using the acceptance plan. Also, changing the number of samples per sublot would change the slope of these curves, reducing the risk involved in sampling and testing.
On this project, the contractor chose to minimize the risk by constructing the pavement lots with properties near the maximum quality level and standard deviations below the target levels. This risk minimization also resulted in increased incentive pay for all lots and quality factors. It is not known how this strategy affected contractor profitability.