A Nonlinear Regression Technique For Calculating the Average Diffusion Coefficient of Wood During Drying

A nonlinear regression technique for determining the optimum drying diffusion coefficient of wood is described. This technique uses the least squares principle to estimate the diffusion coefficient in the infinite Fourier series solution to the one-dimensional unsteady-state form of Fick's law. Five different nonlinear iteration methods for solving the normal equations were evaluated in terms of their ability to find a solution, the rate of convergence, and sensitivity to the parameter's starting value. Application of this technique to a series of experimental isothermal drying runs involving red oak (Quercus rubra) resulted in a better fit between the actual and predicted drying curves, as compared to the logarithmic, square-root, and half-E techniques. The technique yielded diffusion coefficients and residual sum of squares that were almost identical to those obtained from a Fortran-based optimization method reported by Chen et al. (1994). However, the technique described in this paper is more computationally efficient by converging to a solution in a fewer number of iterations than the optimization method