High-packing-rate assembly method for mesoscopic modeling of asphalt mixtures based on genetic algorithm

Abstract

The mesoscopic structure of asphalt mixtures significantly influences their macro-mechanical properties. An artificial random generation technology has been extensively implemented for mesoscopic modeling of asphalt mixtures. However, during the aggregate assembly process, conventional random packing method struggles to achieve dense aggregate spatial arrangements, especially in 3D modeling. To address this limitation, a high-density packing method that optimizes the geometric centers and three-axis rotation angles of aggregates using a genetic algorithm was developed, based on rigid body motion principles. This approach was applied to construct a mesoscopic numerical model of an AC-13 asphalt mixture with an 82 % aggregate placement rate. Furthermore, uniaxial compression simulations using the generated model demonstrated strong agreement with experiments, predicting macro-dynamic modulus values within 20 % deviation from the physical test. The proposed method provides an efficient tool for generating a more reliable mesoscopic structure in asphalt mixture modeling, offering improved accuracy for numerical simulations in pavement engineering.