Pyrolysis and Cracking of Nordic Timbers Under External Heat Exposure

Timber as a construction material has been gaining renewed interest in recent times. However, wood is inherently associated with an increased risk to fire safety. To address the fire risks of wood construction, regulators impose prescriptive building codes that define the limits for use of timber in construction. Performance-based fire safety design is the only method to realize timber construction projects that go beyond the boundaries set by prescriptive building codes. Computational fluid dynamics (CFD) -based fire simulation is a common tool in performancebased fire safety engineering. In realistic CFD fire simulations involving wooden surfaces, wood is defined by a material model that predicts char front progress and release of combustible volatiles, thus interacting with flaming in the gaseous phase. This thesis proposes material models for spruce and pine woods for use in performance-based fire safety design. The thesis also provides an experimental dataset of direct observations of cracking on surface of charring timber to help in implementation of cracking effects to material models in later research. During this work, two independent material models were developed for the studied wood species of spruce and pine: one that assumes wood decomposing through pyrolysis as a single component (single reaction), and another that observes pyrolysis as a sum of individual decomposition reactions of wood primary components: hemicellulose, cellulose and lignin (parallel reactions). Material properties and decomposition kinetics of wood were studied by use of various microscale experimental methods, such as thermogravimetry and differential scanning calorimetry. Cone calorimeter experiments were used in estimation of any remaining model parameters, and in material model validation. The single reaction model arises as the preferable option, offering a similar quality of fit to the experimental data as the more complex parallel reactions model, but without the associated increased model uncertainty. Validation using large-scale experiments revealed that the model is highly sensitive to the surrounding oxygen concentration of the surrounding atmosphere, and inclusion of the oxidation reaction is important for correct fire spread prediction. Formation of cracks on charring spruce, pine and birch woods was studied in real time using an infrared camera. The experiments revealed a linear relationship between heat flux and the inverse of the square root of crack formation time, analogously to the time to ignition in ignition model for thermally thick solids. Contrary to expectations, the number of cracks did not grow consistently as a function of heat flux. The analytical formulation of a previously existing cracking model predicts the number of cracks in correct order of magnitude, but not accurately.