Benchmarking Permeability Predictions for an Engineered Reference Medium: Toward Calibration of Permeability Rigs

ABSTRACT Accurate permeability prediction in fibrous media is essential for modeling liquid composite manufacturing processes, yet numerical predictions are often inconsistent owing to modeling assumptions and numerical implementation choices. This work introduces a 3D‐printable anisotropic reference porous medium designed to reproduce the main flow pathways and anisotropy trends observed in textile reinforcements, while avoiding the inherent variability of real fabrics. The idealized geometry provides a well‐controlled, reproducible benchmark for permeability prediction, acknowledging that it does not capture fine‐scale features such as intra‐tow pores or fiber surface curvature, and may have higher absolute permeability values than real textiles. Fourteen research groups independently simulated fully saturated, incompressible, laminar, and steady‐state flow through a given CAD‐based medium using their own numerical setups. While simple analytical flows (e.g., laminar flow in a pipe or slit) can validate individual codes, they are insufficient to capture the complexity and anisotropy of textile‐like porous media. Benchmarking is therefore necessary to reveal real‐world variability in permeability predictions. Results show good consistency for in‐plane permeability ( K xx : mean 2.75 × 10 −9 m 2 , CoV = 0.105; K zz : mean 6.8 × 10 −10 m 2 , CoV = 0.205), while out‐of‐plane permeability ( K yy ) shows much higher variability (unit‐cell mean 1.40 × 10 −10 m 2 , CoV = 0.790; periodic‐assembly mean 1.27 × 10 −10 m 2 , CoV = 0.470). These findings (i) demonstrate the feasibility of using idealized additive‐manufactured porous media as reproducible calibration benchmarks and (ii) highlight the value of cross‐validation across multiple numerical platforms to bolster confidence in permeability prediction for composite manufacturing.