The primary purpose of this study was to measure experimentally the uniaxial (tangential direction) equilibrium moisture profiles in moisture-sealed wood samples subjected to constant but different temperatures T on opposite faces, and to compare these profiles with those predicted by each of several theoretical models. Each test assembly consisted of eight end matched wood laminae, each 0.2 cm thick, for a total thickness of 1.6 cm in the tangential direction. Opposite faces of each moisture-sealed assembly were exposed continuously for approximately five weeks to different but constant temperatures until the original uniform moisture content M redistributed itself to a new constant but non-uniform moisture profile. At moisture equilibrium, the individual wood laminae were removed from the assembly and their moisture contents measured gravimetrically. Both temperature, T vs x, and moisture profiles, M vs x, where x is the distance from the cold face were plotted for each test. The experimental variables considered were wood species (yellow poplar and hard maple), initial moisture content (9%, 12%, and 15%), and temperature range (15°C to 35°C, and 25°C to 45°C). Each condition was replicated 3 or 4 times, giving a total number of 44 different tests.

At the steady state, a moisture content profile opposite to the temperature gradient was established. The temperature gradient dT/dx was constant in all cases, with the moisture content profile increasing almost exponentially with decreasing temperature. The absolute magnitude of the calculated ratio dM/dT was found to vary with initial moisture content, temperature range, and species. The Soret coefficient, defined as - ( 1/M )( dM/dT ), was also calculated and was found to vary with the above variables as the dM/dT ratio. These two quantities always increased with wood moisture content. They were also slightly higher at the higher temperature range. There was only a small difference between species, with yellow poplar giving somewhat higher mean values. The Soret coefficient was generally in the range of 0.03 to 0.05 per degree Kelvin, except for the extremely high moisture contents near and above fiber saturation, where the calculated values approached 0.5 per degree Kelvin.

The ratio dM/dT was analyzed in terms of five different theoretical models, two of which are based on nonequilibrium thermodynamics (NET) and three on classical thermodynamics. All models require sorption isotherm and heat of sorption data; and in some cases, the activation energy for moisture transport through wood. Adsorption and desorption isotherm data were obtained at 30°C for both species. Heat of sorption and moisture transport activation energy data used in testing the models were taken from the literature.

The two NET models provided the best agreement with the experimental values of dM/dT. The Siau model gave the next best prediction, followed by the Stanish model, with the Skaar- Siau model giving the poorest agreement with the experimental results.

The heat of transfer was also computed using the two NET models as well as those of Stanish and Siau. Values ranged from 5500 to 17000 cal/mol and from 8100 to 9900 cal/mol based on the Nelson model. The corresponding values for the Stanish for the general NET model and from model range from 15200 to 16500; while those of the Siau model varied between 11700 to 13600 cal/mol.

Calculations of other quantities from the information generated in these experiments revealed that at the steady state, vapor pressure and spreading pressure were not constant across the thickness of the material; the chemical potential of water vapor and sorbed water were equal at all points indicating local equilibrium; and that the sorbed water and water vapor entropies were more or less constant across the thickness of the sample.