Viscosity of Asphalt Binder through Equilibrium and Non-Equilibrium Molecular Dynamics Simulations

Abstract

Viscosity is a curial indicator for evaluating asphalt performance, representing its ability to resist deformation under external forces. The Green–Kubo integral in equilibrium molecular dynamics simulations and the Muller-Plathe algorithm in reverse non-equilibrium molecular dynamics simulations were used to calculate the asphalt viscosity. Meanwhile, the key parameters of both methods were rationalized. The results show that in equilibrium calculations, using a 1/t weighting for the viscosity integral curve results in a well-fitted curve that closely matches the original data. The isotropy of the asphalt model improves for atomic counts exceeding 260,000, rendering viscosity calculations more reasonable. When the viscosity did not converge, it increased linearly with the number of atoms. In non-equilibrium calculations, the number of region divisions had almost no effect on the viscosity value. A momentum exchange period of 20 timesteps exhibits a favorable linear trend in velocity gradients, and an ideal momentum exchange period was found to be between 10 and 20 timesteps. As the model size increased, the linear relationship with the shear rate became more pronounced, and the isotropy of the asphalt system improved. Using an orthogonal simulation box with a side length of 75 Å effectively meets the computational requirements.