The existence of flammable aerosols presents a high potential for fire hazards in the process industry. Various industrial fluids, most of which operate at elevated temperatures and pressures, can be atomized when released under high pressure through a small orifice. Because of the complexity in the process of aerosol formation and combustion, the availability of data on aerosol flammability and flame propagation behaviors is still quite limited, making it difficult to evaluate the potential fire and explosion risks from released aerosols in the process industry and develop safety measures for preventing and/or mitigating aerosol hazards. A study is needed to investigate the relationship between aerosol combustion behaviors and the properties of the aerosols. This dissertation presents research on the combustion behaviors of flammable aerosols. Monodisperse aerosols created by industrial heat transfer fluids were generated using electrospray. The characteristics of flame propagations in aerosols and the influence of the presence of fuel droplets in the system are studied in the aerosol ignition tests. Flames in aerosols are characterized by non-uniform shapes and discrete flame fronts. Flames were observed in different burning modes. Droplet evaporation was found to play an important role in aerosol burning modes. Droplet evaporation behaviors and fuel vapor distributions are further related to aerosol droplet size, droplet spacing, movement velocity, and liquid volatility. The burning mode of a global flame with rapid size expansion is considered the most hazardous aerosol combustion scenario. This burning mode requires a smaller droplet size and smaller space between droplets. Larger droplet sizes and spacing may hinder the appearance of global flames. But when the liquid fuel has a certain level of volatility, there is an uneven distribution of fuel vapor in the system and this may cause the unique phenomenon of burning mode variations combined with enhanced flame propagation speed. Using an integrated model, the minimum ignition energy values of aerosols were predicted. The aerosol minimum ignition energy is influenced by the fuel-air equivalence ratio and the droplet size. Higher equivalence ratios, up to 1.0, significantly reduce the minimum ignition energy, while larger droplet sizes result in a higher minimum ignition energy.

Energetic materials encompass a wide range of chemical compounds all associated with a significant risk of fire and explosion. They include explosives, fireworks, pyrotechnics, powders, propellants and other unsteady chemicals. These materials store a high level of chemical energy and are able to release it rapidly without external contribution of oxygen or any other oxidizer. The behaviour of these materials in case of explosive detonations is relatively wellknown from empirical and practical points of view. However, fundamental scientific questions remain unanswered related to the mechanisms of heat release. The current understanding of these mechanisms lacks appropriate thermochemical characterisation. The aim of the study is the analysis of thermal and chemical characteristics of energetic materials under conditions that exclude detonations. Detonation is excluded in order to better isolate the thermal and chemical mechanisms involved in the burning process. The experimental work has been conducted using the FM Global Fire Propagation Apparatus (FPA) [ASTM E2058‐03]. One of the benefits of using this experimental apparatus rather than the Cone Calorimeter is that it allows controlling the feed of heat and oxidizer to the reaction zone. The material chosen to conduct experiments on is a ternary smoke powder based on a mixture of starch and lactose as fuel components and potassium nitrate as oxidizer. This product is currently used by fire brigades to assess smoke venting systems efficiency of buildings. The kinetics associated with the combustion of the material was assessed slow enough to allow measuring instruments to capture the thermal and chemical evolution during combustion reaction. Thermal analysis has first been carried out by means of DSC, TGA, DTA, MS and FTIR data in order to understand the decomposition of the material and its energetic evolution when undergoing heating. However, if the latter methods help defining the decomposing path of the material, they do not provide an integral view of its combustion behaviour, in particular, the emissions of toxics which are kinetic path dependent. Subsequently, combustion tests have been carried out using the FPA. Its ability to capture the evolution of gases emissions formed during the reaction has been proved. The influence of two configuration parameters on the combustion behaviour and on the gaseous emissions of the material has been investigated. The proportion fuel/oxidizer has been varied as well as the composition of the reacting atmosphere. Results shows that the quantity of oxidizer in the material affects the kinetics of the reactions taking place in the condense phase. Increasing the concentration of potassium nitrate in the mixture enhanced the reaction rate of the smouldering combustion. Higher quantity of volatiles is released which favoured the initiation of a diffusion flame regime in the gaseous phase, above the sample. While the kinetics of the condense phase is governed by the oxidizer concentration, experiments show that the flaming regime is influenced by the concentration of oxygen (O2) in the reacting atmosphere. A transition from diffusion to premixed flame is found when the concentration of O2 surrounding the sample is reduced below 18%. An analytical model has been used to explain the existence of a transition for a critical O2 concentration. Finally, thermal and combustion analyses have allowed to characterise the behaviour of the material under critical conditions, in terms of decomposition taking place in the condense phase but also potential toxic emissions that can be released. Toxicity, kinetics, temperature evolution do not provide a complete view of the combustion phenomenon. Beside these elements that characterise the behaviour of a material for given conditions as well as also the degree of fire hazard encountered, the energetic issue holds as an essential feature that cannot be neglected. The heat release rate (HRR) is a critical parameter that defines a fire. It does not constitute an intrinsic material property but it describes the energetic response of the couple formed by the material and its environment. Oxygen Consumption calorimetry (OC) and Carbon Dioxide Generation calorimetry (CDG) are widespread methods to calculate the HRR resulting from a combustion reaction. Apparatuses such as the FPA or the cone calorimeter have already proved their potential to qualify the burning behaviour of common fuels in addition to polymers when their data are combined with an adapted calorimetric procedure. The same approach has been applied to energetic materials. However, prior to using these techniques, it is fundamental to have identified their restrictions. These techniques provide approximate estimations of the HRR. Results are affected by the propagation of uncertainties. Several sources of uncertainties can be found. One can cite: 1. Uncertainties associated with the sample material; 2. Uncertainties associated with the test conditions; 3. Uncertainties associated with the measurements; 4. Uncertainties associated with calculation assumptions. If uncertainties cannot always be estimated, the three first sources cited have received attention in the past from the scientific community, alike the last one. The restrictions associated with the assumptions developed for using the OC and CDG principles have to be clarified. The limits of validity of the hypotheses have to be clearly defined. In particular, the present dissertation questions the relevance of the energy constants that have been specified for OC and CDG as well as their related uncertainties. One of the purposes of the research deals with the ability to estimate accurate error bars for the calculation of the HRR. Once uncertainties related to the calorimetric methods are assessed, a method adapted from the basic OC and CDG principles is introduced that allows estimating the HRR of energetic materials. The approach is based on considering the chemical decomposition of the burning compound and defining a fictitious molecule for which energy coefficients can be calculated. Nevertheless, it requires the material to be known. Finally, the question of the advantage brought by these techniques over others, in terms of accuracy, is discussed within the framework of unconventional products, such as energetic materials or compounds whose composition is ignored. The results from this work will contribute to the development of fireanalysis methodologies and validate their use with energetic materials.