Study on the Compaction State of Asphalt Mixtures Based on Three-Dimensional Modeling Using a Mesoscale Element Equivalence Method

As a multiscale material, it is difficult for a single compaction degree index to fully reveal the mesostructure state and change process of the multiphase and multiscale components of asphalt mixture. In this paper, computed tomography (CT) scanning and image processing model based on a convolutional neural network were employed to accurately obtain the mesostructure of the asphalt mixture at different compaction stages. Based on the representative volume element method (RVE) and mesoscale element equivalence method (MEEM), a three-dimensional (3D) simulation model of local homogenization was constructed to analyze the mechanical behavior of the asphalt mixture under different factors and the evolution of the equilibrium state during compaction. The results indicated that the MEEM can effectively reflect the compaction state of the asphalt mixture. Furthermore, basalt aggregates exhibited greater stability in both horizontal and shear directions compared to limestone aggregates. The incorporation of hard asphalt resulted in a more uniform mixture than matrix asphalt, but the mixture containing hard asphalt was more sensitive to variations in compaction times. This study provides a scientific basis for the design and construction control of asphalt mixtures.