Introduction
Due to its wide availability and renewability, wood has applications in many fields. Wood has been used as a heat source, for weaponry, in creating works of art, and as a construction material for centuries. If applied correctly, wood is a highly durable material, able to stand the test of time. In fire risk assessment, it is of high importance to be able to accurately predict the fire behavior of materials comprising the building component. Since moisture plays a major role in many properties of wood, the qualification and quantification of moisture transport mechanisms is essential.The goal of this study is the simultaneous and non-destructive measurement of both moisture content and temperature in pine wood during intensive one-sided heating, simulating the initial period of a fire. Accordingly a custom-built NMR setup is used, which has been introduced before, in combination with inserted thermocouples. Moreover, numerical experiments with a non-isothermal moisture transport model are performed to capture basic characteristic features of moisture transport during these conditions.

    Results
As an example the normalized moisture content and temperature profiles during onesided heating of a pine cylinder, with its axis in the combined radial/tangential direction, are shown in Fig. 1a and b respectively. Time between two consecutive profiles is 10 minutes. Initially, the moisture content and temperature are constant throughout the sample. After switching on the heat source, the temperature near the surface quickly rises,  resulting in steep gradients as extreme as∼500 C/m. Meanwhile, a peak in the moisture content is formed, which travels towards the back of the sample, while the moisture content left from the peak attenuates to low values. A sharp boundary between dry material left from the peak and the peak in the moisture content itself is observed. The emergence of the moisture content peak is related to rapid release of cell wall moisture in regions with high temperatures. The resulting vapor pressure gradient induces vapor transport. In addition, diffusive vapor transport may also occur due to concentration differences. Moisture then travels as vapor towards regions of low vapor pressure or low vapor concentration, i.e. towards the surface and towards the back of the sample. Whereas the temperature near the exposed surface is high, lower temperatures towards the back of the sample cause exchange of the water vapor to cell wall moisture, which results in an increase in measurable moisture content. This increase is perceived as the peak shown in Fig. 1.



Fig 1 a) Normalized moisture content profiles and (b) temperature profiles during one-sided heating of a pine cylinder,
with its axis in the combined radial/tangential direction. The time between consecutive profiles is 10 minutes.
The position of the peak in moisture content is indicated in each temperature profile with marker.



    Simulations
To qualitatively and quantitatively describe the observed transport processes from experiments, numerical simulations with the model are performed. To this end, the balance equations along with initial and boundary conditions.are implemented in Comsol and solved numerically. The input  parameters. Simulated normalized moisture content profiles and temperature profiles for transport in the radial/tangential direction are shown in
Fig. 2a and b respectively. In these simulations we have assumed values for the diffusion coefficient, permeability, and heat conductivity of 10−6m2/s, 10−14m2, and 0.12 W/m K respectively. Experimental profiles at similar exposure times have been added to the figures for comparison. Similar to the experiments, a peak in the moisture content is formed, which travels towards the back of the sample. Although the qualitative evolution of the peak is similar, with an initially increasing peak value which eventually decreases, the moisture peak is less pronounced and will reach values of 1.4 at maximum. The discrepancy between experiments and simulations is probably caused by the overestimation of the peak moisture content in the experiments, due to the T2-correction of the original data. It is therefore likely that we experimentally capture the dynamics correctly, but overvalue the moisture peak. The simulated temperature profiles in Fig. 2b, however, resemble the experimental profiles adequately, despite the slight overestimation of the temperature near the exposed surface. Similar to the experiments, a small kink in the temperature profiles is observed. This kink is associated with the exchange of water from vapor to cell wall, which is accompanied by heat release thus causing distorted profiles near the peak in

 moisture content.



Fig 2: Simulated (a) normalized moisture content profiles and (b) temperature profiles for transport in the radial/tangential direction
 (D=10−6m2/s,k=10−14m2, λ=0.12 W/m K). Time between profiles is 20 minutes, total simulation time is 2 hours.
For illustrative purposes, experimental profiles at matching time stamps are shown as dashed lines.


     


T. Arends, A.J. Barakat, L. Pel, Moisture transport in pine wood during one-sided heating studied by NMR, Exp. Thermal and Fluid Science 99, 259271 (2018)