Calibration of Pavement Rutting Prediction in Colorado Using Layer-Specific Rutting Model Coefficients for Hot-Mix Asphalt in AASHTOWare Pavement ME Design

Abstract

The Colorado Department of Transportation (DOT) recently concluded a study to adopt the AASHTO Mechanistic–Empirical Pavement Design Guide (MEPDG) and the accompanying software into its routine pavement design practice. Implementation of the MEPDG in Colorado required local calibration of the MEPDG prediction models for distress and smoothness. This paper presents a discussion of the calibration of the “global” rutting model to account for Colorado’s unique climate, traffic, and soil conditions as well as the various asphalt concrete (AC) mix types used in the state. A key challenge during the local calibration effort was to produce a rutting coefficient for each AC mix type: Marshall, Superpave®, and polymer-modified asphalt (PMA). Local calibration of the MEPDG global models by using Colorado DOT input data was performed with a nonlinear model optimization tool available in the SAS statistical software. Global model coefficients were adjusted through local calibration for AC rutting, unbound aggregate base rutting, and subgrade rutting. Four of the 10 model coefficients were adjusted. The local calibration coefficient βr1turned out to be different for each of the three primary Colorado DOT ACs. As expected, of the three asphalt mixes, the PMA had the coefficient with the lowest value for βr1, and as a result this mix had the lowest AC rutting. The goodness of fit and bias test results indicated an adequate goodness of fit with minimal bias. This local calibration effort improved the overall prediction accuracy of the rutting model and its standard error.