Evaluation of stress relaxation process of wood based on the eigenvalue distribution of near infrared spectra

Abstract

Wood is a highly heterogeneous composite material consisting of three major natural polymers featuring properties that can vary significantly among samples. We examined the stress relaxation of wood as a stochastic process. The eigenvalue distribution at each time was calculated from the near-infrared spectral matrix during the stress relaxation test. The eigenvalue motion was regarded as Dyson’s Brownian motion. The first eigenvalue, which is equivalent to the Helmholtz free energy, gradually increased with time, indicating that the relaxation of wood exposed to change necessitates a more organised molecular configuration. The time evolutions of the Shannon entropy calculated from the probability associated with each energy eigenstate did not overlap between the inverse temperatures. Thus, the curves were smoothly shifted until continuous, thereby generating a master curve. The proposed method addresses the limitation of sample differences by assessing the behaviour of the statistical ensemble rather than that of the individual samples.