Advances in Determining the Size Distribution of Blast‑Fragmented Rock Through Numerical Modelling

Predicting fragmentation in blasting remains a significant engineering challenge. Explosive loading induces complex fracture processes that govern the final fragment-size distribution. This study develops, calibrates, validates, and applies an integrated numerical framework to simulate rock fragmentation in small-scale blasting tests. The framework integrates five components: (i) blasting simulation based on the Discrete Element Method (DEM) coupled with the Particle Blasting Method (PBM); (ii) rock characterisation through microparameter calibration with virtual uniaxial compression, Brazilian tensile, and P-wave tests via Bayesian optimisation; (iii) fragment identification using Density-Based Spatial Clustering of Applications with Noise (DBSCAN); (iv) fragment reconstruction by closed-surface meshing; and (v) virtual sieving using size estimation with oriented bounding boxes (OBB) to obtain fragment-size distribution curves. Six published experimental tests were considered: two were used to develop the workflow, and the remaining four were simulated under the reported input condi tions. The numerical dataset comprises these six simulations and an additional densification set. Predictive performance was assessed by deriving from this dataset power-law relationships between the fragment size at percentage passing P and the specific charge q, and comparing them with experimentally reported power laws through statistical error analysis. For lower percentiles, agreement is limited to low specific charges, whereas for higher percentiles, the model maintains predictive capability across the full range, with mean absolute percentage error below 14% and coefficient of determination above 0.94. The framework provides a basis for predicting fragmentation under broader blasting conditions.