Evaluation of the MEPDG Using Asphalt Material Inputs Obtained From Plant Mix

摘要

The Virginia Department of Transportation (VDOT) is expecting torntransition to using the methodology described in the Guide for the MechanisticEmpiricalrnDesign of New & Rehabilitated Pavement Structures (MEPDG) in thernnear future, and decisions must be made regarding the hierarchical level ofrndesired asphalt material inputs and the extent to which asphalt mixturerncharacterization must be performed. Thus a study was undertaken to investigaternthe effects of these variables using input values measured from 11 typical plantrnmixtures (i.e., three surface, four intermediate, and four base mixtures). Thernpredicted fatigue and rutting distresses were used to evaluate the response of thernMEPDG to differences in the mixture properties and to assess the future needsrnfor implementation. Two example pavement sections were modeled: a primaryrnand an interstate roadway section. Pavement distress data were compiled for anrninterstate and primary route corresponding to the modeled sections and wererncompared to the MEPDG-predicted distresses.rnPredicted distress quantities for fatigue cracking and rutting wererncompared to the calculated distress model predictive errors to determine if thernmaterial property input levels were statistically different. There were differencesrnbetween all rutting and fatigue predictions using level 1, 2, and 3 asphaltrnmaterial inputs, but they were not statistically significant. Various combinationsrnof level 3 inputs showed expected trends in rutting predictions when increasedrnbinder grades were used, but the differences were not statistically significantrnwhen the calibration model error was considered. Fatigue distress predictionsrnwere approximately comparable to the pavement distress data, but the modelrnpredictive errors were greater than the distress predictions.rnAlthough this study was limited, it indicated several steps VDOT should takernprior to implementing the MEPDG. Additional work to identify the source ofrnpredictive insensitivity to changes in both asphalt mixture properties and the levelrnof input values is necessary. The comparisons of predicted fatigue and rutting distressesrnto pavement distress data for the primary and interstate pavement indicaternthe need to consider local calibration and verification of the predictive models.