The aim of this paper is to build more reasonable three-dimensional (3D) mesostructure numerical specimens and to study relationships between 3D mesostructure numerical specimens and their two-dimensional (2D) slices using an improved algorithm called Embedded-Zoom Aggregate Algorithm. According to experiments used by Walraven, equivalent 3D numerical specimens containing cube aggregates were built. All cut planes parallel to each surface of above specimens were identified respectively to count the number of different aggregate cross section areas. Comparison results between 2D identification method and Walraven function can prove the consistency of 2D identification method. For asphalt concrete, by analyzing 2D slice identification method, Walraven function and real specimen cut planes, results of the first two methods are both similar to the last one. Grading curves in ASTM were selected to further judge the applicability of Walraven function and 2D identification method. Results indicated that Walraven function is unsuitable for ASTM. Meanwhile, a relationship between an aggregate gradation and a special section of aggregate cross section areas under descending spectrum in cut planes was found by 2D slice identification method. Embedded-Zoom Aggregate Algorithm presented in this paper is proved to be valid. Moreover, the average area of aggregate cross sections in a special section can effectively distinguish various 3D grading curves.
- Aggregate packing algorithm
- Image identification
- Three-dimensional numerical specimen
- Two-dimensional slice identification