Dong Feng
A controllable algorithm is developed to generate simulation-ready virtual aggregates using a domain-warped multifractal implicit-radius field. Fractional Brownian motion (FBM), ridged multifractals, Worley-type chipping, and layered anisotropy are integrated into a unified formulation, enabling decoupled and intuitive control of multiscale roughness, ridge sharpness, chipping intensity, and directional stratification within a single generator. The workflow automatically produces watertight triangular meshes and supports dimensional calibration based on principal component analysis (PCA) or an axis-aligned bounding box (AABB) to match prescribed particle proportions. Optional operations, including angular twisting, light smoothing, and bottom capping, are provided to enhance numerical stability and facilitate robust placement and contact handling in downstream simulations. A descriptor-driven calibration is further established to map generator inputs to roundness, boxiness, 3D solidity, and true sphericity, yielding interpretable parameter sensitivities and enabling targeted tuning toward the desired morphology. Morphological fidelity is validated against a 3D-scanned aggregate database using distribution-similarity measures and hypothesis testing, demonstrating close agreement at the population level. Discrete element method (DEM) simulations of Superpave gyratory compaction (SGC) are then conducted to assess engineering relevance, showing that the generated particles reproduce packing density evolution and gyration–height responses comparable to those of scan-based models. Overall, the proposed algorithm provides a practical bridge from controllable multifractal synthesis to validated, watertight, simulation-ready aggregate meshes for high-throughput numerical studies.