Analysis of Conventional Dynamic Modulus Predictive Models of Asphalt Mixtures

Dynamic modulus is a fundamental property of asphalt materials that is used as a key parameter in the Mechanistic-Empirical pavement analysis and design (MEPDG). This modulus is a function of loading frequency and temperature that is sensitive to the aggregates, binder content and air voids in the mixture. Hence, the method of measurement and determination of dynamic modulus is important in pavement design and predicting its performance. There are several methods to determine dynamic modulus of asphalt mixtures. These methods include: Directly measuring in laboratory; Empirical and regression-based methods using prediction models; Modeling dynamic modulus using soft computing techniques like Artificial Neural Networks (ANN); Analytical and numerical methods for modeling the dynamic modulus; and, In-situ methods using nondestructive testing (NDT). This paper presents a comprehensive analysis of prediction models for dynamic modulus of asphalt mixtures including Witczak, Modified Witczak, Hirsch, Al-Khateeb, New Regression-based and ANN models. With reference to the literature, performance of prediction, sensitivity analysis with respect to input parameters, the effect of mixture type, and finally a comparison of goodness for these models were evaluated and reported. In addition, calibrated models for different local conditions were presented. The results show developed conventional predictive models are simple and applicable in determination of dynamic modulus of asphalt mixtures using some modifications for different mix types and local conditions.