New modelling approaches to predict wood properties from its cellular structure: image-based representation and meshless methods

Key messageThe real tissue structure, including local anisotropy directions, is defined from anatomical images of wood. Using this digital representation, thermal/mass diffusivity and mechanical properties (stiffness, large deformation, rupture) are successfully predicted for any anatomical pattern using suitable meshless methods.IntroductionWood, an engineering material of biological origin, presents a huge variability among and within species. Understanding structure/property relationships in wood would allow engineers to control and benefit from this variability. Several decades of studies in this domain have emphasised the need to account simultaneously for the phase properties and the phase morphology in order to be able to predict wood properties from its anatomical features. This work is focused on the possibilities offered by meshless computational methods to perform upscaling in wood using actual tissue morphologies obtained by microscopic images.MethodsAfter a section devoted to the representation step, the digital representation of wood anatomy by image processing and grid generation, the papers focuses on three meshless methods applied to predict different macroscopic properties in the transverse plane of wood (spruce earlywood, spruce latewood and poplar): Lattice Boltzmann Method (LBM) allows thermal conductivity and mass diffusivity to be predicted, Material Point Method (MPM) deals with rigidity and compression at large deformations and peridynamic method is used to predict the fracture pathway in the cellular arrangement.ResultsThis work proves that the macroscopic properties can be predicted with quite good accuracy using only the cellular structure and published data regarding the cell wall properties. A whole set of results is presented and commented, including the anisotropic ratios between radial and tangential directions.