Sample Size Effects on Morphological and Particle Size Distribution Characteristics of Granular Materials: 2D and 3D Analyses

This paper investigates the statistical reliability of quantification of 2D and 3D morphological and grading characteristics of irregular granular soils across varying sample sizes. Given the inherent heterogeneity in particle shape and size, limited sampling may lead to significant errors in estimating representative values. To address this, a coupled Bayesian–Monte Carlo framework is proposed for determining the sufficient sample size required to meet user-defined accuracy thresholds. The method is applied to nine natural granular materials with irregular particle shapes, using both 2D image analysis and 3D micro-computed tomography (Micro-CT), from which material-specific sample size requirements are derived. Empirical correlations are then established to estimate required sample sizes for similar materials. The influence of sample size on selected mechanical properties is also examined through established empirical models, highlighting the broader implications of statistical representativeness in geotechnical characterisation and design.