Quantifying the Differential Phase Angle Behaviour of Asphalt Concrete Mixtures Using Artificial Neural Networks

Abstract

Phase angle is an important property of asphalt concrete (AC) mixtures that can aid in proper material selection and thereby assist in accurate design of flexible pavements. In particular, it is imperative to quantify differential phase angle behaviour for varying mixture characteristics to minimise premature failure of flexible pavements. To this end, this study aims to provide insights into the differential phase angle behaviour of wearing versus base course mixtures, field versus laboratory prepared mixtures, and for a full spectrum of binder grades. To achieve this research aim, this study employs a two-step framework. First, an artificial neural network (ANN) model is developed to predict phase angle using laboratory test data as input. Twenty-three AC mixtures consisting of different penetration grade binders, varying mix proportions, and mix types are used for phase angle testing performed at four testing temperatures (4.4, 21.1, 37.8, and 54.4 °C) and six loading frequencies (25, 10, 5, 1, 0.5, and 0.1 Hz) using the asphalt mix performance tester equipment. In the second step, a sensitivity/parametric analysis is performed for the phase angle model revealing the differences in phase angle characteristics of wearing and base course, field and laboratory mixtures as well as mixtures prepared with different binder grades. Furthermore, a comparison of the proposed ANN model with linear and non-linear regression models is performed, and the ANN model outperforms the competing models. The developed ANN model can be used as a surrogate to laboratory testing and utilised by transport departments and pavement analysts to characterise the phase angle behaviour of heterogenous AC mixtures.