ABSTRACT
A method based on the concept of solving inverse heat and mass transport problems was proposed to identify wood physical properties, with the use of empirical data, a mathematical model corresponding to the direct problem, and an optimization procedure. The computer software was developed to solve transient, three-dimensional, quasi-linear direct and inverse problems of heat and mass transport in wood as an anisotropic body together with initial and boundary conditions of any kind. The software was adapted to data parallel processing environment of high performance computers. Identification of thermal conductivity was performed for beech wood in three principal anatomic directions, and accuracy of temperature distribution predictions was significantly increased (the global relative error of prediction was reduced to 1.0 - 1.9%).
Keywords and Phrases: