Aggregate Representation Approach in 3D Discrete-Element Modeling Supporting Adaptive Shape and Mass Property Fitting of Realistic Aggregates

Aggregate based composites are a common material in civil engineering. Examples of such composites include: asphalt concrete, filled plastic and reinforced concrete. More often than not, it is essential for one to determine the properties of the composite. Presently, micromechanical analysis of such composites using the discrete element method (DEM) has become an effective way to investigate their behaviors. Several approaches have been developed and can be categorized to either 2D or 3D. Existing 3D discrete-element (DE) models fall into two categories: models of realistic microstructure and models of virtual microstructure. Aggregate representations in the DE method are actually sphere clumps. It is very important that the mass properties of aggregate models be the same as those of real aggregate when sphere clumps are directly used in the numerical simulation of stone-based material. The mass and inertia tensor of an aggregate can also be calculated based on its shape and density, which are then used to represent the rotational internality of the sphere clump of the aggregate in simulation. However, a composite usually contains thousands of aggregates, which makes the manual input of inertia properties a very heavy task.

On this account, researchers from the Hefei University of Technology: Professor Can Jin and Ma. Shouguo Li, in collaboration with Professor Xu Yang at the Chang’an University, Dr.  Xiaodong Zhou and Professor Zhanping You at the Michigan Technological University developed a novel DE modeling method for realistic aggregates through which the obtained sphere clumps could accurately simulate the shape and mass properties of the particles with a far lower number of spheres than anything reported before. The mass properties herein referred to the shape, mass, barycenter and inertia tensor. Their work is currently published in the Journal of Engineering Mechanics.

In their approach, a 3D solid model of the particle was accurately reconstructed from X-ray CT imaging, based on which the shape of the model was accurately represented with surface spheres. Consequently, the internal space of the model was fully filled with inner spheres. Finally, the mass properties of the obtained spheres, namely the sphere clump, were calibrated with that of the realistic particle.

The authors reported that the obtained DE representation could occupy the volume of the aggregate more than 99.5% with a properly defined fitting accuracy and fit the aggregate in inertia very well. In addition, numerical results showed that the aggregate models generated using the proposed method could be successfully used to generate stone-based materials such as asphalt mixtures.

In summary, the study developed a new adaptive representation method for aggregates to obtain corresponding clumps of spheres that closely fit realistic shapes and the inertia of the aggregates. In other words, the study proposed a novel method to generate aggregate models that mimic the real aggregate shapes via overlapped spheres. The results presented showed that the aggregate models were in very good agreement with the real aggregates. In a statement to Advances in Engineering, the authors highlighted that the proposed methodology was robust and more effective than the currently used ones.