Effect of Laboratory Data Range Selection in Permanent Deformation Characterization of Asphalt Mixtures

As the power model parameters play a critical role in the permanent deformation characterization of asphalt mixtures and their direct use in pavement structural design methodologies becomes standard practice, all known sources of variability in their estimation should be considered. Although there is clear consensus in the ability of the power model to characterize the secondary stage of deformation properly, defining where this stage begins and ends is an arbitrary process, and therefore different parameter estimates could be obtained for the same mixture, potentially generating different design recommendations when they are used in the Guide for Mechanistic–Empirical Design of New and Rehabilitated Pavement Structures (referred to as MEPDG). The effect that laboratory data range selection has in the determination of the power model parameters was evaluated by using data from repeated load tests with short, medium, and long duration times performed with standard and modified asphalt mixtures. The results from the analysis indicate the potential for significant differences in the power model parameters because of data range selection alone during model fitting, even when the data selected are confined to the secondary deformation stage. The magnitude of the observed differences and its potential impact in rut depth prediction supports the need for the establishment of unified criteria for the selection of ranges of data from permanent deformation tests. As the potential implementation and acceptance of MEPDG will rely heavily on local calibration efforts, including permanent deformation laboratory testing, this paper proposes a procedure to unify the way in which the range of data used to estimate the parameters from the power model is selected.