Three‐stage weight‐based aggregate generation framework for <scp>3D</scp> concrete mesostructure modeling

Abstract Aggregates determine the durability, workability, and strength of concrete, influenced by their size, shape, and distribution. Existing 3D modeling methods oversimplify aggregate shapes to reduce computational complexity, sacrificing realism. Thus, this study proposes a three‐stage weight‐based framework to generate realistic 3D concrete mesostructures. First, an additively weighted Voronoi diagram creates aggregates with diverse realistic shapes, including nonconvex forms and smooth surfaces. Next, aggregates undergo shape filtering to exclude rarely occurring isotropic shapes based on aspect ratio statistics from cross‐sectional images. Finally, aggregates are packed according to a target grading curve and volume fraction. The proposed model is validated against real specimens by comparing filling ratios, particle size distributions, and aspect ratios. Relative errors in the filling ratio, particle count, and aspect ratio distribution are within 3%, confirming that the framework accurately replicates mesostructural features. The approach provides a reliable basis for finite element simulations and fracture analyses in concrete modeling.