Enhancing Rutting Prediction of the Mechanistic–Empirical Pavement Design Guide by Using Data from a Field Test Section in Oklahoma

Abstract

An instrumented test section was constructed on Interstate 35 near Purcell, Oklahoma. Mechanistic–Empirical Pavement Design Guide (MEPDG) input parameters for traffic, climate, and materials were developed from the data for this test section. The site included a weigh-in-motion station and lateral positioning sensors to obtain input parameters for traffic. In addition, laboratory tests—dynamic modulus, resilient modulus, dynamic shear rheometer, and other pertinent tests—were conducted with materials from the test section. Rut measurements were taken at the site on a quarterly basis for 5 years. Differences in traffic input parameters between Level 3 and Level 1 were identified and are discussed in this paper. At first, significant differences were observed between measured and predicted ruts. For example, the average error between the MEPDG-predicted rut and the measured rut was approximately 18% when default input parameters were used. Although the average error was reduced to approximately 10% when the Level 1 inputs were used, differences between the predicted and measured ruts still existed, which necessitated the development of local calibration factors. After the calibration process, the average error was reduced to less than 5%. The optimized calibration coefficients were found to be β_r1 = 2, β_r2 = 1, and β_r3 = 0.9 for asphalt layers; β_GB = 1 for an aggregate base layer; and β_SG = 0.5 for a natural subgrade layer. There is a need for the development of a database and local calibration of models for successful implementation of the MEPDG in Oklahoma.