Efficient Prediction of the Thermal Conductivity of Wood from Its Microscopic Morphology

In this work, we combined the homogenization and finite volume methods to predict the solid fraction and the effective thermal conductivity from 3D real morphologies of wood, namely spruce and poplar. High resolution scans performed by a nano-tomograph, together with image processing are two steps of great importance to obtain the digital representation of the real morphology suitable for computation. These tools allow the generation of the 3D mesh of the thresholded sample. The stationary diffusion model is directly considered to gain in performance. Numerical results revealed that several minutes of CPU time are enough to predict the values of the thermal conductivity on the representative volumes. Compared to some of our recent works, the present methodology is not only efficient, but also more accurate.