Multiscale Models Predicting Wood Structure Infiltration

Transparent Wood (TW) is an innovative composite material distinguished by its lightweight nature, optical transparency in the visible spectrum, and environmental sustainability. Through a series of treatments, including delignification, resin infiltration, and optional functionalisation, TW emerges as a promising alternative to glass and plastics across various industrial sectors, such as construction, automotive, electronics, and furniture. The AI-TranspWood project, supported by the European Commission, aims to develop AI-driven multiscale modelling approaches for TW composites (https://www.ai-transpwood-project.eu/). Within this framework, simulating resin infiltration into delignified wood is a key step toward understanding and optimising the manufacturing process of TW-based products. This presentation describes a computational framework for predicting infiltration kinetics in delignified wood structures during transparent wood fabrication. The integrated modeling approach comprises three components: (1) molecular dynamics simulations to determine infiltrant viscosity and transport properties under selected conditions, (2) a hierarchical multiscale modeling framework based on micromechanical principles

to predict the effective permeability of delignified wood microstructures, and (3) finite element analysis employing modified Darcy flow equations to investigate the effects of processing parameters and boundary conditions on infiltration time scales.


We generate the molecular structures for an infiltrant liquid starting from the SMILES description of the infiltrant molecule. We use open source tools (ACPYPE, Veloxchem) to generate Gromacs topologies for the GAFF force field, and to obtain the atomic partial charges. Gromacs itself is used to initialize and equilibrate the molecular liquid, and to perform the non-equilibrium shear flow simulations, using the box deformation method, for accurate computation of viscosity while maintaining system homogeneity. The workflow has been automated in the AiiDA environment, facilitating the screening of several infiltrants. The micromechanical framework captures essential qualitative trends in structural and fluid transport transformations during lignin removal and subsequent polymer infiltration, revealing that delignification substantially increases pore volume and permeability while poly(methyl methacrylate) (PMMA) infiltration alters fluid pathways and preserves certain permeable domains. The analysis demonstrates strong sensitivity of longitudinal and transverse permeability components to microstructural features including lumen aspect ratio, lumen volume fraction, and pore shape anisotropy. The modified Darcy approach extends classical Darcy's law to account for permeability variations in different directions, and incorporates surface tension effects, contact angle dependencies, and external pressure gradients in the boundary conditions. This enhanced model is implemented using the FEniCSx finite element framework to solve the coupled flow equations in complex three-dimensional wood geometries. A comprehensive parametric analysis was conducted to evaluate the sensitivity of infiltration kinetics to key processing variables, including infiltrant properties, wood microstructure characteristics, and boundary condition configurations. As a toolchain, the different levels of methodologies enable quantitative prediction of mass transport phenomena in the complex porous network of delignified wood, providing insights for optimizing transparent wood processing conditions.