The prediction of adhesive failure between aggregates and asphalt mastic based on aggregate features

The interface stripping between aggregates and asphalt mastic deteriorates the strength of asphalt mixture, causing disease such as cracks and pits of asphalt pavement. This paper aim to quantify the effect of aggregate on interface, and develop a method to predict the adhesive failure. At first, microstructural models of aggregate particles were reconstructed based on computed tomography (CT) images, then aggregate features and energy consumption of interface during fracture were determined by digital image processing and numerical simulation respectively; dissipation damage energy (DDE) quantified the adhesive failure, and an effective method was established by artificial neural network based on back-propagation (ANN-BP) to present the relationship between DDE and aggregate features. Results show that the influence of aggregate on interface damage can be evaluated by multiple parameters of aggregate features; ANN-BP is an effective tool to synthesize the coupling influences of different aggregate features and determine a prediction model; ANN-BP model presents a prediction for adhesive failure according to the states of aggregates, the prediction results are acceptable to show the damage of asphalt mixture.