Understanding the relevant behaviour of fire in buildings is critical for the continued provision of fire safety solutions as infrastructure continually evolves. Traditionally, new and improved understanding has helped define more accurate classifications and correspondingly, better prescriptive solutions. Among all the different concepts emerging from research into fire behaviour, the compartment fire is probably the one that has most influenced the evolution of the built environment. Initially, compartmentalization was exploited as a means of reducing the rate of fire spread in buildings. Through the observations acquired in fires, it was concluded that reducing spread rates enabled safe egress and a more effective intervention by the fire service. Thus, different forms of compartmentalization permeated through most prescriptive codes. Once fire behaviour within a compartment was conceptualized on the basis of scientific principles, the compartment fire framework became a means to establish, under certain specific circumstances, temperatures and thermal loads imposed by a fire to a building. This resulted not only in improved codes but also in a scientifically based methodology for establishing the thermal input from which to assess structural performance. The last decades have however seen an evolution of the built environment away from compartmentalization while the classic compartment fire framework has remained. Within this framework, while Regime I corresponds to the idealised experimental setups adopted by many of the researchers, the usually ignored Regime II is characteristic of open spaces and volumes, typical of contemporary architecture. This research project commences, through a review of classic literature by those regarded as the fathers of fire safety engineering, by revisiting the knowledge underpinning this seminal approach, and initiating the discussion of its continued relevance and applicability to an increasingly non-compartmentalised built environment. Compartment fires are extremely complex processes. Nevertheless, when treating the theoretical problem with sufficient accuracy, simple mathematical approaches can be extremely informative and serve as the background to more complex methodologies. In this context, the project introduces the problem of the compartment fire in its full complexity before discussing some simplifications typically assumed when representing the actual problem for design purposes. Further, despite the detailed experimental and theoretical background behind analytical formulations in the classic compartment fire framework, their development is revisited to establish the extent to which they can be applied. In this way, the range of validity of the classic framework is characterized, clarifying the limitations of existing design methods based on this framework, and identifying the areas where further research and extension is necessary. Given the importance of counting on simple analytical formulations at the early design stage when dealing with atypical architectural designs in today’s fire safety practice, an elementary theoretical compartment fire framework is elaborated with the aim of enveloping traditional as well as contemporary architectural layouts. This gave way to the development of a new set of regime of behaviour definitions, in addition to – and falling in-between – the classic Regime I and Regime II fullydeveloped compartment fire behaviours. With the aim of filling this gap of knowledge empirically and characterizing these additional behaviours, a series of small and large-scale tests are presented. The results demonstrate complex behaviours that cannot be described in terms of the classic framework. This evidences the great need to conduct research that provides physical insight into the dynamics of a fire in spaces that deviate from the small quasi-cubic enclosure – the natural consequence of compartmentalization – that was typically adopted throughout the original work that resulted in the data which validated the classic compartment fire framework. Overall, this project aims to inform and encourage the discussion of the existence of a broader compartment fire framework, where the historical Regimes I and II are limiting cases of a vaster fire behaviour which is intimately linked to the geometry of the compartment, the ventilation conditions, and the available fuel. While the classic compartment fire framework is still a robust tool, it is only one piece in the puzzle of approaching and resolving the fire problem in a building in a holistic way. The relevance of this discussion is apparent in face of contemporary architecture and infrastructure.

Pour des raisons d'économie d'énergie, les façades des bâtiments deviennent de plus en plus sophistiquées à la fois par leurs configurations et leurs compositions. Mais la quantité de combustible de ces nouvelles façades est bien supérieure à celle des façades traditionnelles. Par conséquent, le risque de propagation du feu via la façade est plus important. Ainsi, l'objectif de ce travail est de modéliser le développement du feu à l'intérieur et à l'extérieur d'un bâtiment en prenant en compte différentes configurations et compositions de façade. Cette recherche expérimentale et numérique a permis d'identifier les paramètres qui augmentent ou diminuent le risque de propagation du feu via une façade. Dans un premier temps, après avoir vérifié la capacité du code de calcul à modéliser les flammes pariétales, une étude numérique qui étudie l’influence de la géométrie d’une façade sur la propagation du feu via la façade a été réalisée. En effet, les différents phénomènes liés aux dimensions des ouvertures et/ou à la configuration de la façade ont été identifiés. Il a donc été possible d’analyser leur influence sur le risque de propagation du feu en façade à travers des grandeurs telles que la puissance libérée à l’extérieur du bâtiment ou encore la hauteur de flamme et les actions thermiques engendrées. Parmi les configurations étudiées figurent des géométries plus ou moins complexes pouvant être rencontrées sur les bâtiments. Par exemple, les ouvertures multiples ou encore des configurations en « U » afin d’étudier l’influence de l’effet cheminée sur l’extension des flammes. En effet, ce type de configurations a déjà été la cause d’une propagation rapide d'incendies via des façades quelles que soient leurs compositions. Par la suite, une étude expérimentale sur la propagation du feu sur une paroi combustible a été réalisée avec deux objectifs. Tout d’abord afin d’étudier les phénomènes de propagation sur une façade combustible (température de flamme, hauteur de propagation, contribution énergétique de la façade), mais également pour récolter des données expérimentales permettant la validation de modèles de propagation et les simulations numériques dans cette situation. Dans un deuxième temps, une nouvelle campagne expérimentale a permis d’étudier l’influence de la présence d’une lame d’air de ventilation entre le bardage et le mur sur la propagation du feu. Cette dernière configuration est largement utilisée dans la construction des façades comportant généralement une couche d’isolation dans la lame d’air. Cette recherche, à la fois académique et applicative, a permis de fournir des informations originales sur le développement et le comportement du feu le long d’une façade, qu’elle soit combustible ou non. Les résultats numériques présentés mettent en évidence les différents paramètres gouvernant le développement d’un feu le long d’une façade, ce qui facilite la compréhension des phénomènes liés à cette problématique. De plus, les différents essais réalisés pourront servir de base de données à la modélisation de la propagation d’un incendie le long d’une paroi combustible, ainsi qu’à la mise au point des modèles de développement et de propagation. / In order to enhance the energy efficiency of buildings, the facades are becoming more sophisticated in both their configurations and compositions. However, the amount of fuel of these new facades is much higher than that of traditional facades. Therefore, the risk of fire spread through the facade is more important. Thus, the objective of this work is to model the fire development inside and outside of a building, taking into account different configurations and facade compositions. This experimental and numerical research has identified the parameters that increase or decrease the risk of fire spread via the façade. First, after verifying the capacity of the FDS code to model the parietal flames, a numerical study that examines the influence of the geometry of a facade fire spread was completed. Indeed, the various phenomena related to openings dimensions and / or configurations of the façade have been identified. It was therefore possible to analyze their influence on the risk of fire spread along the façade using quantities such as the heat released outside the building, the flame height and thermal actions (temperature, fluxes). Among the configurations studied are contained more or less complex geometries that can be encountered on the buildings. For example, multiple openings or "U" configurations were investigated in order to study the influence of the chimney effect on the extension of flames. Indeed, this type of configuration has already been the cause of the rapid spread fire through walls regardless of their compositions. Subsequently, an experimental study on fire spread along a combustible wall was realized with two goals. First, a series of tests was performed in order to observe propagation phenomena on a combustible façade and to collect experimental data to validate propagation models and numerical simulations in this situation. Secondly, another experimental campaign was used to study the influence of the presence of a ventilation air gap between the cladding and the wall on the spread of fire. This latter is widely used in the construction of facades. This research, both academic and applicative, has provided new information on the fire development and fire behavior along a façade, combustible or not. The numerical results demonstrate the various parameters governing the development of a fire along a façade, which facilitates the understanding of phenomena related to this issue. In addition, various tests can be used as a database for the modeling of fire spread along a combustible wall. Thus, this work contributes to the development of models of fire development and spread on buildings via the façade.

Maintaining structural stability is an integral component of building fire safety. Stability must be ensured to provide adequate time for safe egress of the buildings occupants, fire fighting operations and property protection. Structural fire engineering endeavours to design structures to withstand the effects of fire in order to achieve this objective. The behaviour of reinforced concrete in fire is not as well understood as other construction materials, such as steel. This is in part due to the complexity of concrete material behaviour and also due to concrete’s reputation of superior fire performance. Concrete technology is, however, continually evolving; structures are increasingly slender, more highly stressed and have higher compressive strengths. A more robust understanding of concrete’s behaviour in fire will enable predictions of the implications of changing concrete technology and also help to properly quantify the fire safety risk associated with concrete structures. A fundamental key to understanding structural fire performance is the relationship between the thermal environment induced by the fire and the structure. Significant thermal variation has been found experimentally to exist within fire compartments. Despite this the design of structures for fire almost universally assumes the compartment thermal environment to be homogeneous. In this thesis the implications of compartment fire non-uniformity for concrete structural behaviour is investigated to assess the validity of the uniform compartment temperature assumption. The investigation is conducted using numerical tools; a detailed review of the necessary background knowledge, material modelling of reinforced concrete, finite element modelling of reinforced concrete structures and compartment fire thermal variation is included. The behaviour of a two-way spanning reinforced concrete slab is used as a structural benchmark. The membrane behaviour exhibited by two-way spanning RC slabs at high temperatures has been previously studied under uniform thermal conditions. They therefore are an ideal benchmark for identifying the influence of non-uniform thermal environments for behaviour. The relationship between gas phase temperature variation and concrete thermal expansion behaviour, which is fundamental to understanding concrete high temperature structural behaviour, is first investigated. These preliminary studies provide the necessary fundamental understanding to identify the influence of gas phase temperature variation upon the membrane behaviour of reinforced concrete slabs. The individual influences of spatial and temporal variation upon slab membrane behaviour are investigated and the behaviour under non-uniform thermal variation contrasted with uniform thermal exposure behaviour. The influence of spatial variation of temperature is found to be strongly dependent upon the structural slenderness ratio. The tensile membrane action of slender slabs is particularly susceptible to the distorted slab deflection profiles induced by spatial variation of gas temperature. Conversely the compressive membrane behaviour of stocky slabs is found to be insensitive to the deformation effects induced by spatial variation of temperature. The influence upon slender slabs is demonstrated under a range of temporal variations indicating that the thermal response of concrete is sufficiently fast to be sensitive to realistically varying distributions of temperature. Contrasting behaviour induced by uniform and non-uniform thermal exposures indicates that uniform temperature assumptions provide both conservative and unconservative predictions of behaviour. The accuracy of the uniform temperature assumptions was also found to be dependent upon the type of fire, for example, fast hot and short cool fires. Additionally, the sensitivity of structural performance to deformations caused by spatial variation of temperature demonstrated in this thesis challenges the purely strength based focus of traditional structural fire engineering. Spalling is an important feature of concrete’s high temperature behaviour which is not currently explicitly addressed in design. The incorporation of spalling into structural analysis is not, however, straightforward. The influence of spalling upon behaviour has therefore been dealt with separately. A spalling design framework is developed to incorporate the effects of spalling into a structural analysis. Application of the framework to case studies demonstrates the potential for spalling to critically undermine the structural performance of concrete in fire. It also demonstrates how the framework can be used to quantify the effects of spalling and therefore account for these in the structural fire design addressing spalling risk in a rational manner.

This thesis reports on the procedure, results and analysis of four full scale fire tests that were performed at the University of Waterloo's Live Fire Research Facility. The purpose of these tests was to investigate the thermal characteristics of one room of the Burn House structure. Comparisons were made of Burn House experimental data to previous residential fire studies undertaken by researchers from the University of Waterloo. This analysis showed similarities in growth rate characteristics, illustrating that fire behaviour in the Burn House is typical of residential structure fire behaviour. The Burn House experimental data was also compared to predictions from a fire model, CFAST. Recommendations were made for future work in relation to further investigation of the fire characteristics of the Burn House.

This thesis reports on the procedure, results and analysis of four full scale fire tests that were performed at the University of Waterloo's Live Fire Research Facility. The purpose of these tests was to investigate the thermal characteristics of one room of the Burn House structure. Comparisons were made of Burn House experimental data to previous residential fire studies undertaken by researchers from the University of Waterloo. This analysis showed similarities in growth rate characteristics, illustrating that fire behaviour in the Burn House is typical of residential structure fire behaviour. The Burn House experimental data was also compared to predictions from a fire model, CFAST. Recommendations were made for future work in relation to further investigation of the fire characteristics of the Burn House.

Engineered timber products are becoming increasingly popular in the construction industry due to their attractive aesthetic and sustainability credentials. Cross-laminated timber (CLT) is one such engineered timber product, formed of multiple layers of timber planks glued together with adjacent layers perpendicular to each other. Unlike traditional building materials such as steel and concrete, the timber structural elements can ignite and burn when exposed to fire, and thus this risk must be explicitly addressed during design. Current design guidance focusses on the structural response of engineered timber, with the flammability risk typically addressed by encapsulation of any structural timber elements with the intention of preventing their involvement in a fire. Exposed structural timber elements may act as an additional fuel load, and this risk must be adequately quantified to satisfy the intent of the building regulations in that the structure does not continue burning. This can be achieved through timber’s natural capacity to auto-extinguish when the external heat source is removed or sufficiently reduced. To address these issues, a fundamental understanding of auto-extinction and the conditions necessary to achieve it in real fire scenarios is needed. Bench-scale flammability studies were undertaken in the Fire Propagation Apparatus to explore the conditions under which auto-extinction will occur. Critical conditions were determined experimentally as a mass loss rate of 3.48 ± 0.31 g/m2s, or an incident heat flux of ~30 kW/m2. Mass loss rate was identified as the better criterion, as critical heat flux was shown by comparison with literature data to be heavily dependent on apparatus. Subsequently, full-scale compartment fire experiments with exposed timber surfaces were performed to determine if auto-extinction could be achieved in real fire scenarios. It was demonstrated that auto-extinction could be achieved in a compartment fire scenario, but only if significant delamination of the engineered timber product could be prevented. A full-scale compartment fire experiment with an exposed back wall and ceiling achieved auto-extinction after around 21 minutes, at which point no significant delamination of the first lamella had been observed. Experiments with an exposed back and side wall, and experiments with an exposed back wall, side wall, and ceiling underwent sustained burning due to repeated delamination, and an increased quantity of exposed timber respectively. Firepoint theory was used to predict the mass loss rate as a function of external heat flux and heat losses, and was successfully applied to the bench-scale experiments. This approach was then extended to the full-scale compartment fire experiment which achieved auto-extinction. A simplified approach based on experimentally obtained internal temperature fields was able to predict auto-extinction if delamination had not occurred – predicting an extinction time of 20-21 minutes. This demonstrates that the critical mass loss rate of 3.48 ± 0.31 g/m2s determined from bench-scale experiments was valid for application to full-scale compartment fire experiments. This was further explored through a series of reduced-scale compartment fire experiments, demonstrating that auto-extinction can only reliably be achieved if burnout of the compartment fuel load is achieved before significant delamination of the outer lamella takes place. The quantification of the auto-extinction phenomena and their applicability to full-scale compartment fires explored herein thus allows greater understanding of the effects of exposed timber surfaces on compartment fire dynamics.

This thesis is devoted to heat transfer and fire dynamics in enclosures. It consists of a main part which summarizes and discusses the theory of heat transfer, conservation of energy, fire dynamics and specific fire scenarios that have been studied. In the second part of this thesis, the reader will find an Appendix containing seven scientific publications in this field. In particular, one- and two-zone compartment fire models have been studied. A new way of calculating fire temperatures of pre- and post-flashover compartment fires is presented. Three levels of solution techniques are presented including closed form analytical expressions, spread-sheet calculations and solutions involving general finite element temperature calculations. Validations with experiments have shown good accuracy of the calculation models and that the thermal properties of the surrounding structures have a great impact on the fire temperature development. In addition, the importance of the choice of measurement techniques in fire engineering has been studied. Based on the conclusions from these studies, the best techniques have been used in further experimental studies of different fire scenarios. / Denna avhandling behandlar problem kopplade till värmeöverföring och branddynamik i slutna utrymmen med tonvikt på värmeöverföring mellan gaser och utsatta konstruktioner. Avhandlingen består av en huvuddel och ett appendix innehållande sju vetenskapliga artiklar. I huvuddelen sammanfattas och diskuteras grundläggande teorier och principer inom värmeöverföring och branddynamik samt studier av ett antal specialfall av brandscenarion som baseras på dessa teorier. I de avslutande bilagorna (Artiklar A1-A3 och Artiklar B1-B2) finns sju vetenskapliga artiklar som grundligare beskriver de ovan nämnda specialfallen. Huvudfokus i avhandlingen ligger på temperaturutveckling vid brand i slutna utrymmen. I avhandlingen studeras i synnerhet en- och två-zonsmodeller för brand i slutna utrymmen, och en ny metod för att beräkna brandgastemperaturer före och efter övertändning i rumsbränder är framtagen. Validering av dessa modeller med experiment visar att deras noggrannhet är bra. Modellerna visar också att de termiska egenskaperna hos de omgivande ytorna har stor inverkan på brandtemperatursutvecklingen. I tillägg studeras i denna avhandling betydelsen av val av mätmetoder i brandtekniska tillämpningar. På grundval av slutsatserna från dessa studier har de främsta mätteknikerna använts i ytterligare experimentella studier av olika brandscenarier.

Dans les feux de compartiment, les fumées représentent un danger majeur lors de l’intervention des sapeurs-pompiers. Elles sont souvent à très hautes températures et sont surtout très mobiles. De ce fait, elles provoquent la propagation du feu et peuvent engendrer des phénomènes thermiques, imprévisibles et incontrôlables, tels que les explosions de fumées ou l’embrasement généralisé éclair. L’étude présentée se propose de contribuer à construire des méthodes et des moyens pour pallier aux manques d’informations lors de l’intervention des sapeurs-pompiers sur feux de compartiments. Il s’agit de proposer une méthode de diagnostic en temps réel afin, en particulier de détecter l’occurrence de dangers potentiels spécifiques à ces situations.Pour une approche plus rationnelle, l’incendie a été étudié comme un phénomène physique, ce qu’il est, et analysé à partir des relevés de température au cours de l’évolution du feu. La particularité de l’étude a été d’examiner ces relevés de températures comme des séries temporelles et d’étudier leurs caractéristiques à partir de deux moyennes mobiles dans le passé, soit sur un même point de mesure soit sur deux points de mesures positionnés en des endroits clés. Cette méthode a permis de montrer que l’entrecroisement des moyennes mobiles donne des informations pertinentes sur l’évolution à venir du feu et sur la possibilité de détecter l’apparition possible de phénomènes thermiques catastrophiques. / In compartment fires smokes are a major danger during firemen intervention. Most of the time, they are at high temperature and they flow everywhere through many kinds of ducts, which leads to the propagation of the combustion by the creation other fires in places which may be far from the initial fire. Besides, since the appearance of the new building materials and with the heat insulations which lead to the concentration of heat during a fire, the compartments fire became even more dangerous for firefighters during their phases of attack.In the presente study we will introduce a new approach of the problem, which allows to better follow the fire behavior and especially to detect the dangers that may appear and endanger firefighters. This approach consists in a mathematical analysis based on the comparison of moving averages, calculated on the temperature recordings of the smokes. As a consequence, this method may allow improving decision support in real time and therefore to improve the security and the efficiency of firefighters in their operations against that kind of fires.

Devant le besoin de nouveaux outils d’aide à l’investigation post-incendie, cette thèse se propose d’évaluer le recours à des modèles numériques pour apprécier un scénario proposé par un expert. Pour cela, les conditions d’utilisation de la version 4 de Fire Dynamics Simulator (FDS) ont été déterminées dans une optique de reconstitution de sinistres. Une méthode a ensuite été développée afin de confronter les résultats des simulations aux observations faites lors de l’investigation : cette méthode se base sur l’utilisation de multiples points de comparaison qui sont des effets thermiques ou mécaniques remarquables sur un matériau. Les résultats obtenus ont ensuite été utilisés pour la simulation de deux cas réels. Le premier s’attache à comparer, à des simulations, des mesures de propagation de flammes à la surface d’un matelas dans deux configurations différentes. Cela est réalisé afin d’évaluer le modèle de combustion solide de FDS dans différentes conditions de ventilation. La seconde application est la reconstitution d’un incendie ayant fait une victime dans un appartement. Ce second cas a été choisi dans le but d’évaluer les méthodes de simulation et d’utilisation des points de comparaison établies dans les premiers chapitres. / This work is intended to evaluate the use of numerical models to assess fire scenarios proposed by investigators. First chapter of this thesis is about theoretical fire phenomenon useful in the investigation frame. The second chapter is devoted to the validation of a CFD tool named Fire Dynamics Simulator (FDS) for fire reconstruction purposes. Guidance is developed in the third chapter in order to compare modelling results to degradations observed on fire scenes. It is explained how the accuracy of a scenario could be assessed using several comparison points, the thermal or mechanical effects on materials after a fire. Results obtained in previous chapters are finally applied on two cases. The first one is the comparison of fire spread velocities measured on two different experiments with simulations. This study is carried out to test FDS combustion model under ventilated and underventilated conditions. The second case is the reconstruction of a fire which killed the occupant of an apartment, the purpose of this work being to apply modelling and investigation guidance to a real case.

Lors d’essais de feux d’hydrocarbures liquides dans le dispositif DIVA de l’IRSN, un phénomène oscillatoire basse-fréquence (BF), a été observé. Ce phénomène se manifeste par des fluctuations importantes de la pression dans le local, qui peuvent conduire à une perte de confinement et ainsi favoriser la propagation du feu et le rejet de polluants au-delà du local. Il s’accompagne de déplacements intermittents de la flamme hors du bac. L’étude fine de ce phénomène oscillatoire a tout d’abord consisté à concevoir une maquette à l’échelle 1:4 du dispositif DIVA dans lequel nous avons fait varier différents paramètres. L’analyse des résultats obtenus nous a permis d’identifier différents régimes de combustion, de décrire les mécanismes responsables de l’apparition des oscillations BF et de caractériser les propriétés de ces oscillations (fréquence et amplitude). L’occurrence et la persistance des oscillations BF dépendent essentiellement de l’équilibre, plus ou moins précaire, entre la quantité d’air disponible pour la combustion et le débit d’évaporation du combustible résultant des flux thermiques reçus à sa surface. Une étude numérique exploratoire utilisant le code CFD SAFIR a été ensuite conduite en utilisant le débit d’évaporation mesuré expérimentalement, puis en le calculant à l’aide d’un modèle d’évaporation. Si le code ne permet pas de décrire correctement le déplacement de la flamme hors du bac, il reproduit de façon satisfaisante le comportement oscillatoire BF du feu, en particulier sa fréquence dominante. / During liquid hydrocarbon fire tests in the DIVA device of IRSN, a low-frequency (LF) oscillatory phenomenon, was observed. This phenomenon manifests itself by large variations of the average pressure in the room, which can lead to a loss of confinement and thus promote the spread of fire and the release of pollutants beyond the local. It is accompanied by intermittent displacements of the flame outside the fuel pan. The fine study of this phenomenon consisted in designing a 1:4 scale model of the DIVA device, allowing us to carry out a very large number of tests, varying some parameters. The analysis of the results obtained allowed us to identify different combustion regimes, to describe the mechanisms responsible for the appearance of the LF oscillations, and to characterize the properties of these oscillations (frequency and amplitude). The occurrence and persistence of LF oscillations essentially depend on the precarious equilibrium between the supply of fresh air and the supply of fuel vapors which results from the heat flux received at its surface. An exploratory numerical study using the CFD code SAFIR was then conducted using both the experimentally measured evaporation rate and that calculated using an evaporation model. The model does not correctly describe the displacements of the flame outside the fuel pan. However, it satisfactorily reproduces the LF oscillatory fire behavior, especially its dominant frequency.