Assessing the sensitivity of DEM-generated pore structures to aggregate morphology and gradation in porous asphalt mixtures

Discrete element method (DEM) is widely used to reconstruct pore structure in porous asphalt (PA) pavement for hydraulic performance investigation. However, the contributions of aggregate morphology and gradation to pore structure formation remain unclear, particularly considering computational cost. This study employed a voxel-based virtual computed tomography (CT) framework to analyse the three-dimensional pore characteristics of DEM-generated PA samples. Seven groups of PA samples were generated using different gradation ranges and aggregate morphology templates, with flatness, elongation, roundness, angularity and the number of templates (25–75) varied. Pore size distribution, air void distribution, tortuosity and pore network model (PNM) metrics were analysed to evaluate the effects of morphology and gradation. Results show that aggregate gradation has a greater influence on pore characteristics than aggregate morphology and template number, affecting pore size distribution, reducing the proportion of small pores, and increasing pore throat size and channel length in PNMs. Increasing the number of morphology templates produces only minor changes in tortuosity (approximately 4.7%), suggesting limited benefits from greater morphological diversity. Overall, relatively few aggregate morphologies are sufficient to reproduce representative pore structures, while gradation selection is more critical. These findings guide efficient DEM-based porous media generation and future pore-scale flow and clogging simulations.