Abstract – Building envelopes commonly comprise porous building materials, with their functional performance governed by their microstructure, including the underlying pore structure. In recent years, pore network (PN) modeling has emerged as a powerful tool to investigate flow and transport properties of such porous materials. However, reliable predictions require PNs that capture the complete pore structure, which remains challenging due to the inherently wide and continuous pore size distributions of porous building materials. Building on our earlier work with X-ray computed tomography (XCT) and X-ray microscopy (XRM), this study introduces a full-scale pore network (FSPN) generation approach that statistically represents the pore structure at full scale. Single-scale pore networks (SSPNs) are extracted from the XCT-XRM image chain, and robust full-scale statistics are derived using a bootstrapping-based procedure. These statistics then inform a novel stochastic algorithm that regenerates both the geometry and topology of the pore structure. Using ceramic brick as a case study, the generated FSPN exhibits strong agreement with image-derived full-scale statistics in both parameter distributions and correlations, thereby also accurately reproducing experimentally measured liquid permeability through pore network modeling. This study demonstrates, for the first time, the stochastic generation of FSPNs for porous (building) materials, thereby providing a foundation for predictive simulations of their flow and transport properties.
