Large open-plan compartments have permeated the modern tall building landscape at staggering rates. These types of spaces fall outside the bounds of applicability of existing scientific frameworks defining compartment fire behaviour, and despite their ubiquity, no consolidated efforts at characterising the fire dynamics of large compartments have been made hitherto. Considering these limitations, a theoretical and experimental study is carried out to elucidate the physical phenomena defining large compartment fire dynamics. The compartment fluid mechanics, fire spread mechanisms, and fuel-burning behaviours are investigated. An experimental analysis of data collected from several highly instrumented large-scale compartment fire experiments, spanning several campaigns, is conducted to support the theoretical analysis. The experiments analysed within this thesis are geometrically similar but parameterise the fuel type, ventilation conditions, and lining materials.

A novel model is developed to estimate the global energy balance and the spatial distribution of energy of large compartment fire experiments. It is shown that the spatiotemporal distribution of energy and the fluid mechanics within the compartment are dependent on the ratio of the flame front to burnout front velocities V_S/V_BO of the fire and the imposed ventilation condition. It is found that the fire dynamics for large compartments can be classified into three distinct modes of behaviour;​ being a travelling fire (V_S/V_BO ≈ 1), a growing fire (V_S/V_BO > 1) and a fully-developed fire (V_S/V_BO → ∞). The physical mechanisms controlling V_S/V_BO are studied by analysing four large-scale compartment fire experiments comprising continuous wood crib fuel-beds. It is shown that the transient evolution in the thermal feedback from the smoke layer and compartment boundaries onto the fuel surface and the strength of the flame controls VS. These mechanisms are mainly dependent on the compartment geometry and ventilation conditions. Moreover, it is shown that the fire mode can transition during the evolution of the fire. A phenomenological model is developed using theoretical arguments to help identify the conditions leading to each mode.

The role of the porous wood crib fuel-bed on the compartment fire dynamics is examined through a detailed theoretical analysis of the heat and mass transport processes within and external to the crib. The burning crib is analogised into a reacting forced boundary layer problem (“Emmons problem”) and used to formulate a total mass transfer number that explores the burning dynamics of the crib and its coupling to the compartment. It is found that the wood crib approximates steady-state burning in two regimes, and their occurrence corresponds to the travelling fire (V_S/V_BO ≈ 1) and the fully-developed fire (V_S/V_BO → ∞). The porous and charring nature of the wood crib induces particularly unique burning mech- anisms that are not comparable with real fuels, such as non-charring plastics.

Knowledge of each physical phenomenon studied is then combined to interpret and statistically analyse data collected from almost every single large-scale compartment fire experiment reported in the literature (n = 37). Ceiling heat fluxes, which represent the most onerous heating scenario, are estimated for each experiment and classified in terms of V_S/V_BO. A Large Compartment Fire Framework is constructed based on the statistical analysis and amalgamation of the entire available experimental data, with quantitative bounds established to define the thermal exposure for each distinct fire mode. The framework enables the user to simplify the fire dynamics into a set of design fire scenarios for large compartment fires with an appropriate spatial and temporal treatment of the thermal boundary condition based on a comparison of the characteristic heating times for each fire mode (gas) and the structural material (solid). Expressions for each characteristic heating time are provided along with guidance for the required inputs necessary to utilise the framework. The framework is benchmarked to the state-of-the-art design methods for open-plan compartment fires (travelling fire methods). This benchmarking exercise is undertaken using a canonical example of a large-scale compartment fire experiment. The comparison shows that the magnitude and time-history of the imposed heat fluxes and energy from the existing travelling fire methods are vastly inconsistent with the experimental data from real fire experiments. The underlying assumptions driving these methods do not conserve energy, and thus cannot adequately describe the open-plan compartment fire dynamics. The heat fluxes obtained from the framework have good agreement with the experimental data, and the time-history is captured well given that the mode transition times are user-specified. This characteristic allows for parameterisation of fire scenarios by the user. The underpinning experimental data driving the framework can be easily updated with data from future large compartment fire experiments. Thus, the Large Compartment Fire Framework presents a robust and physically-based approach to quantifying thermal exposure for the entire continuum of possible fire scenarios in large compartments.

This thesis lays the groundwork for a fundamental characterisation of various physical phenomena that defines the fire dynamics in large open-plan compartments. Scientific and statistical arguments are used to analyse the fire dynamics and the available experimental data in the literature to propose an engineering tool that imposes quantitative and realistic thermal boundary conditions for structural design.