Numerical Investigations of Wooden Fires Across Various Scales Using Different Fire Modelling Methodologies

The primary objective of this thesis is to deepen the understanding of wooden fires through a computational fire modelling approach. The focus is on the application of various methodologies to model wood pyrolysis and investigate the influence of kinetic parameters on this process. Furthermore, this thesis aims to validate the pyrolysis model for simulating wood fires by comparing its predictions with empirically observed fire behaviour, conducting sensitivity analyses, and refining the model based on the insights gained. To achieve this goal, a computational fluid dynamics (CFD)-based approach, employing the large-eddy simulation (LES) method, was adopted across a spectrum of experiments spanning from small-scale to large-scale scenarios with diverse fire loads. The focus on pyrolysis modelling is specifically aimed at predicting the heat release rate (HRR) and subsequent structural and material responses, enabling the implementation of more realistic fire scenarios and the study of flame spread. This thesis is structured into four meticulously designed tasks, each incorporating different modelling methodologies. Initially, small-scale tests and kinetic model validation were performed using thermogravimetric analysis (TGA) and cone calorimeter experiments. A detailed component mechanism, including single and multi-components for wood, was used to simulate wood pyrolysis. A comprehensive parametric study assessed the influence of kinetic parameters on model performance and accuracy. Subsequently, the numerical investigation of wooden crib fires under various fuel load conditions was explored to evaluate burning behaviour using three methodologies: prescribed heat release rate, Arrhenius rate law, and a simplified reaction scheme. Simulations were performed either by specifying the burning rate of the fuel directly from physical experiments or by modelling the pyrolysis of wood by defining kinetic parameters. This phase included a detailed parametric study incorporating grid sensitivity analysis to determine mesh size requirements, the ratio between different mesh blocks, and the influence of the percentage formation of mass fraction products on output quantities during the thermal degradation of wood. Following this, the pyrolysis model was applied to investigate complex fuel load arrangements in large-scale room fires, considering different fuel sources and evaluating the Fire Dynamics Simulator (FDS) model's limitations in predicting the inside wall temperature (IWT) of the ceiling. Additionally, fire behaviour of single walls, both non-combustible and combustible, at various fire exposures was simulated to assess the contribution of combustible walls to the fire, enhancing the understanding of the performance of FDS model in determining IWT. The thesis findings contribute to the foundational understanding necessary for enhancing fire safety in wooden buildings, highlighting the need for precise modelling assumptions and parameters. Future studies should aim to refine these methodologies further and conduct additional experimental validations to improve the predictive capabilities of CFD-based fire simulations.