Improving the determination of the dynamic modulus of asphalt concrete for mechanistic-empirical pavement design

The reliability of asphalt pavement design depends on the accuracy of material characterization. The dynamic modulus (|E*|) of asphalt concrete is the main parameter in both structural and performance models of asphalt layer, and its accuracy contributes significantly to the reliability of pavement analysis. This thesis deals with the determination of |E*| for mechanistic-empirical pavement design. Published studies have clearly identified three issues that may affect the accuracy of |E*|. Firstly, predictive techniques for estimating | E*| from mixture volumetric properties are widely used in pavement design and evaluation when the measured values are not available. The reliability of these techniques is lowest at high service temperatures above 35°C. Rutting is one of the major distresses in asphalt pavements and is considered highly sensitive to |E*| at high temperatures. Secondly, |E*| is measured or predicted as a function of frequency. Since the vehicular pulse loads are represented in the time domain, the correctness of the selected |E*| for the multilayered elastic analysis depends on the accuracy of the conversion from pulse loading time to the testing frequency. The mechanistic-empirical pavement design guide (MEPDG) assumes that the loading frequency is the inverse of the pulse time. This approach overestimates the frequency and leads to unconservative pavement design. Finally, the asphalt mixtures are conveniently characterized with |E*| master curves developed from values measured at a wide range of temperature and frequency. However, stiffness of in-service asphalt concrete is limited to the resilient or backcalculated moduli because the thickness of the asphalt layer, in most cases, is less than the required height of |E*| specimens. The objective of this thesis is to improve the reliability of pavement design and evaluation through the followings: improving the prediction of |E*| at high temperature determining the |E*| master curve from the resilient modulus and proposing a method to determine the frequency of the pulse loading to assign |E*| for the multilayered elastic analysis.A viscoelastic method is proposed to calculate the frequency of a pulse load to determine the applicable |E*| for pavement design. The method can be used to calculate the frequency of any shape of loading pulse and to investigate the effect of the mixture type. Conversion equations from loading time to the frequency are developed for both field and laboratory pulses. The loading frequencies of the MR pulse that has a loading period of 0.1 second and the FWD pulse load are 5 Hz and 15.7 Hz, respectively. The MEPDG approach overestimates the loading frequency up to 173%.A method is proposed for predicting |E*| from MR measured. The method applies a conversion equation to determine the frequency of the loading pulse, the parabolic interpolation function to determine |E*| at various temperatures, and the shift factor of the binder to substitute the shift factor of the mixture. Predicted |E*| values at various temperatures and frequencies are an acceptable substitution for measured |E*| values at these conditions.The relationship between |E*| at high temperature and the aggregate gradation represented with a new gradation parameter is examined. The correlation between |E*| at high temperatures and the gradation parameter improved when it was carried out independently on fine- and coarse-graded mixtures. The new parameter shows the ability to describe the effect of aggregate gradation on the stiffness at high temperatures even for as-constructed asphalt concrete. Reasonable results were predicted for the calibration data and a validation dataset from the literature. For fine-graded mixtures, a linear regression model is developed to predict |E*| at 40°C and it is used to adjust the |E*| master curve obtained with the predictive techniques. The adjustment improved the predicted moduli at high temperatures.