Effects of Shape and Size Features of Individual Non-Spherical Particles in Randomly Organized Packed Bed Configurations

This work investigates how particle geometry and boundary conditions govern packing density and accessible surface area in randomly organized packed beds. We employ a superellipsoid parameterization to represent non-spherical shapes by aspect ratio (α), oblateness (β), and corner/edge sharpness (p₀, p₁). Rigid-body simulations, validated via computed tomography, were conducted under varying container-to-particle diameter ratios and particle counts. Four key metrics were systematically analyzed: packing density (PD), the total surface-to-volume ratio (SVtotal), the effective (available) surface-to-volume ratio (SVeff), and the fraction of surface area lost to inter-particle contact (SR). Results indicate that PD is primarily controlled by boundary effects, with smaller particles and higher particle counts producing denser configurations. In contrast, α and β strongly influence SVeff, as flatter, elongated particles afford higher surface area but reduce packing efficiency. Sharper edges amplify SVtotal while also raising SR, underscoring the trade-off between maximizing surface exposure and limiting contact-induced surface loss. An automated machine learning approach significantly outperforms linear regression in capturing these non-linear interactions, offering improved predictive accuracy. These findings clarify fundamental trade-offs for optimizing packed bed systems, balancing shape-driven surface benefits with boundary-driven packing constraints, and provide actionable design suggestions for applications such as thermal energy storage, catalytic reactors, and filtration.