Three-dimensional discrete element modeling of asphalt concrete: Size effects of elements

This paper investigated effects of element sizes in three-dimensional discrete element modeling of asphalt concrete. Asphalt concrete was considered as a two-phase composite of mineral aggregates larger than 2.36 mm and asphalt mastic. Even though asphalt mastic was a composite of fines, fine aggregates, asphalt, and air voids in the reality, it was considered as a homogenous media to interact with larger-size aggregates. A computer-generated discrete element model was employed to simulate the geometry of asphalt concrete. In the model, mineral aggregate particles larger than 2.36 mm were simulated with randomly-created ellipsoids, while asphalt mastic occupied the remaining volume of the asphalt concrete. Each ellipsoid particle in the model was a cluster of discrete elements (balls) which were bonded with a built-in contact-bond model. With the computer-generated model, four digital samples were generated, whose element sizes (radius of balls) were 0.75 mm, 0.65 mm, 0.50 mm, and 0.35 mm. The mechanical interactions within the digital samples were simulated through four contact models, namely the linear contact stiffness model, the Burger’s model, the slip model, and the contact-bond model. The first two models were used to simulate force–displacement relations at contacts, while the last two models were employed to model the strength properties at the contacts. Creep compliance tests were simulated on the four digital samples to investigate size effects of discrete elements on creep stiffness. Through this research, it was found that (1) discrete element sizes gave insignificant effects on creep stiffness (the maximum difference among different samples were within 5%) and (2) discrete element sizes could impact time-steps, computational time, aggregate shape representation, etc.

Highlight
► Element size effects were investigated on three-dimensional discrete element modeling of asphalt concrete.
► Mineral aggregate particles were simulated with the randomly created ellipsoids.
► Element sizes gave insignificant impacts on the predicted creep stiffness.
► Element size could affect time-steps, computational time, aggregate shape representations, etc.
► The keywords are as follows: 1. Pavement engineering. 2. Asphalt concrete mixture. 3. Discrete element method. 4. Creep stiffness. 5. Discrete element size.