A theoretical and experimental study of the autoignition of wood is performed. In the experiment, a wood sample (redwood) of 4 by 4 cm surface area with 4 cm thickness is exposed vertically to a heater panel in a cone calorimeter. The surface temperature is continuously measured by an infrared thermocouple and mass loss is monitored by a load cell. Incident heat flux is varied until glowing ignition could not occur. Times to glowing ignition and flaming autoignition are measured. It is found experimentally that the critical heat flux for flaming autoignition is 20 kW/m2 and for glowing ignition is 10 kW/m2.

A theoretical model for autoignition of wood is developed. The model considers the processes occurring in both solid and gas phases. In the solid phase, a one-dimensional heat conduction model is employed. Char surface oxidation, which can lead to glowing ignition, is taken into account at the solid-gas interface surface. By "glowing ignition", it means the onset of surface combustion. Criteria for glowing ignition are developed based on a surface energy balance. A numerical result shows that according to the present glowing ignition criteria, an inflection point of the surface temperature history can indicate glowing ignition. In the gas phase, a transient two-dimensional laminar boundary layer approximation for gas phase transport equations is constructed. The gas phase model is coupled with the solid phase model via the solid-gas interface surface. Flaming autoignition occurs when the maximum gas reaction rate exceeds a critical value. A numerical result from the coupled gas phase and solid phase models shows that autoignition of the combustible gases behaves in two fashions as autoignition type I at high heat flux and autoignition type II at low heat flux. In the type I autoignition, the flaming occurs just an instant after glowing ignition is initiated, while in the type II autoignition, the solid undergoes glowing ignition long before the flaming is achieved.

Comparisons between the theoretical and experimental results are presented to demonstrate capabilities and limitations of the proposed model.