Aggregate representation for mesostructure of stone based materials using a sphere growth model based on realistic aggregate shapes

The objectives of this study are: (1) to create aggregate representations based on their realistic shapes and establish an aggregate shape library by storing aggregate numerical representations of various shapes and angularities; and (2) to prepare mesostructure models for stone based materials using aggregate representations from the library. The numerical representations based on realistic shapes are created through three steps: (1) scan individual aggregate particles using an X-ray CT machine to obtain a series of cross sectional images at different heights; (2) process the sectional images and stack them to obtain the 3D aggregate boundary; and (3) apply a sphere growth model to fill the interior space of the boundary. The sphere growth includes a first growth in 26 divergent directions and a second growth in 8 divergent directions which start from each sphere in the first growth. Four aggregate particles were selected for the numerical representation construction using the sphere growth model. The modeling results showed that the four aggregates had filling ratios of 97.8, 95.3, 91.3 and 98.5 %, respectively. Afterward, the four numerical representations were stored as aggregate templates to establish a preliminary representation library, which can be invoked to generate numerical aggregate particles in the model reconstruction of stone based materials. A discrete element model and a finite element model of an asphalt concrete beam were reconstructed as examples of application of the numerical representation library. The study overall showed that: (1) the sphere growth model is an effective approach to construct aggregate numerical representations with a high accuracy; (2) the sphere growth model can significantly reduce the sphere amount of the numerical representations compared to previous related studies; and (3) these numerical representations can be successfully used to reconstruct the mesostructure models of stone based materials, which can potentially bring time and cost savings while keeping the model precision.