Improving asphalt discrete numerical modelling with realistic particle shapes

  • Micaelo, Rui (Universidade NOVA de Lisboa)
  • Monteiro Azevedo, Nuno (National Laboratory for Civil Engineering)
  • Câmara, Gustavo (Universidade NOVA de Lisboa)

Please login to view abstract download link

Micromechanical modelling based on the Discrete Element Method (DEM) has been widely used to investigate asphalt behaviour due to its ability to represent an irregular microstructure with variable-sized aggregates, bitumen and voids. The 3D rigid particle models with randomly distributed spherical particles and adopting elastic and/or simple viscoelastic models at the contacts are the standard approach [1], however, in recent years, a significant research effort is noted to incorporate real particle morphologies in the numerical models [2]. Asphalt behaviour is dependent on the material’s microstructure that is largely defined by the aggregates’ shape and size. In this study, a previously developed 3D DEM model of asphalt mixture employing a generalised Kelvin contact model formulation for the viscoelastic contacts [3] is further improved with realistic particle shapes representing the coarse aggregates. A digital library of aggregate shapes was created from the X-ray computed tomography (CT) scan of an asphalt specimen, using an adaptive image processing method to separate the aggregates in the CT images and the Delaunay triangulation method to define the aggregate 3D surface model. Several virtual aggregates with different sizes were selected from the library to represent the coarse aggregate gradation of the 3D DEM asphalt model used in [3]. Each virtual aggregate is discretized with smaller spherical particles and its deformability is taken into account through the inner particle contacts. The numerical asphalt specimens with realistic particle shapes were submitted to uniaxial tension-compression cyclic tests at various frequencies to determine the stiffness properties, and the results were compared with those of the numerical specimens with all constituents represented by single spherical particles. As shown, the proposed methodology greatly enhances the 3D DEM model ability to simulate the asphalt behaviour.