Validation of inverse stereology generation of two dimensional area gradations for computational modelling of asphalt mixtures

Over the last few years, several researchers have employed computational methods to model and investigate the overall behaviour of asphalt mixture composites and different factors that influence this behaviour. Two-dimensional models, particularly those with computationally generated geometries, are commonly used in lieu of three-dimensional models due to their computational efficiency. When building these types of models, it is critical to properly determine a two-dimensional area distribution of aggregates in a given cross section, which needs to be generated from a representative three-dimensional volumetric gradation. This can be typically achieved using some form of mathematical transformation; one type of transformation often used for asphalt mixture gradations is inverse stereology. The goal of this study was to determine the effectiveness of the inverse stereology approach when compared with the true two-dimensional area gradation observed in laboratory compacted HMA specimens. The results from this study show that an inverse stereology approach based on a polyhedron shape was effective in replicating the two-dimensional area gradation created by cutting a laboratory specimen. In addition, the aspect ratio of particles was considered based on cutting the specimens vertically and horizontally; it was determined that no substantial difference existed between the vertical and horizontal cut specimens. However, it was also confirmed that the aspect ratios obtained using aggregate imaging are not representative of aspect ratios in a two dimensional cross-section of an asphalt mixture. An inverse stereology approach to converting aspect ratios from 2D to 3D provided better results, but was still not accurately able to represent the distribution of aspect ratios of aggregates.