A topology-regulated stochastic pore reconstruction for graphite foam: Morphological validation and transport property prediction

Abstract

Digital reconstruction of graphite foam requires reconciling the parametric controllability of periodic unit cell models with the morphological fidelity of image derived geometries. Porosity and pore size statistics alone do not uniquely determine the connectivity, window geometry, and transport pathways of a random pore network. This study develops a topology-regulated stochastic pore reconstruction (TRSPR) framework for generating statistically representative, parametrically controllable, and simulation ready graphite foam structures. TRSPR prescribes porosity, pore size distribution, and coordination number distribution while regulating inter-pore intersections and window formation. Pore radii sampled from micro computed tomography (μCT) data are retained, whereas octree assisted stochastic seeding, nearest neighbor identification, and iterative attractive-repulsive adjustment of pore centers jointly regulate network topology. Boolean reconstruction produces continuous pore and solid domains suitable for pore scale simulation. Validation against four μCT characterized samples yielded a mean porosity deviation of 0.28% and a maximum deviation of 0.61%, with no statistically significant differences between paired pore size distributions. Three independent realizations demonstrated the repeatability of the stochastic procedure. Although not independently prescribed, volume specific surface area and hydraulic tortuosity were reproduced with deviations of 7.16–16.09% and 3.88–8.28%, respectively. Pore scale simulations captured the reported decrease in effective thermal conductivity with increasing porosity, while an independent 61% porosity POCO foam benchmark yielded a permeability deviation of 13.25%. The resolved fields further linked macroscopic transport to continuous solid ligaments, preferential flow pathways, window induced acceleration, flow redistribution, and localized pressure losses. The proposed TRSPR therefore extends stochastic reconstruction from geometric randomness to statistically constrained topology control, providing a controllable link between graphite foam morphology, pore network topology, and transport behavior.