Use of mechanistic-empirical pavement design principles to assign asphalt pavement pay factor adjustments

This paper promotes the quality analysis of flexible pavements by exploring the role of analysis programs as laboratory support tools to assess the quality of constructed pavements. Specifically, the research explored the significance of laboratory dynamic modulus testing on assigning pay factor adjustments in these tools. In the United States, the current methods of pavement acceptance are based more on construction craftsmanship (i.e., density and smoothness, percent within limits) as opposed to how the pavement will actually perform once constructed. The use of performance-based specification models would enable a state transportation agency to quantify the impacts of varying as-built mixture quality in terms of (1) a projection of pavement service life, and (2) presenting life-cycle costs in a way that both agencies and contractors can understand. Two publicly available performance-related specification (PRS) quality assurance (QA) analysis tools have been developed based on mechanistic-empirical pavement design principles and were assessed for a small set of mixtures. The quality related specification software (QRSS) utilizes a predictive equation and traffic, binder, and mixture volumetric data to predict dynamic modulus (|E*|) values. The asphalt mixture performance tester (AMPT) QA program varies from the QRSS in that it uses the laboratory-measured |E*| as an input. Pavement distresses (i.e., rutting and fatigue cracking) are calculated based on both the as-designed job-mix formula and as-produced construction mixture properties. Then, a predicted service life difference (PLD) and the resultant pay factor adjustments are calculated based on the variance between the as-designed and as-produced mixtures. The distress and service life predictions from the two QA programs were similar for the conventional mixes, which may indicate that requiring a laboratory-tested dynamic modulus does not add significant benefit in the analysis of conventional mixtures. Furthermore, an analysis was done in the QRSS program to compare the impacts of the preloaded default binder values to the laboratory-measured binder properties. The results indicated that laboratory testing of binders is recommended for providing inputs to the QRSS since the in situ characteristics were not fully reflected by the default binder values in the program. The differences in the distress levels predicted for the unconventional mixture types indicated that the use of laboratory-measured |E*| is necessary for setting pay factor adjustments, particularly since only conventional dense-graded mixtures were used to calibrate the models in the programs. The conditions under which the program-based analyses will have limitations are presented. Recommendations are provided on how agencies and contractors can improve the construction process by preliminary monitoring of pavement quality using the analysis tools.