No description available.

I assess the potential of seismic and time-domain controlled-source electromagnetic (CSEM) methods to monitor carbon dioxide (CO2) migration through the application of a monitorability workflow. The monitorability workflow describes a numerical modelling approach to model variations in the synthetic time-lapse response due to CO2 migration. The workflow consists of fluid-flow modelling, rock-physics modelling and synthetic seismic or CSEM forward modelling. I model CO2 injected into a simple, homogeneous reservoir model before applying the workflow to a heterogeneous model of the Bunter Sandstone reservoir, a potential CO2 storage reservoir in the UK sector of the North Sea. The aim of this thesis is to model the ability of seismic and time-domain CSEM methods to detect CO2 plume growth, migration and evolution within a reservoir, as well as the ability to image a migrating front of CO2. The ability to image CO2 plume growth and migration within a reservoir has not been demonstrated in the field of CSEM monitoring. To address this, I conduct a feasibility study, simulating the time-lapse CSEM time-domain response of CO2 injected into a saline reservoir following the multi-transient electromagnetic (MTEM) method. The MTEM method measures the full bandwidth response. First, I model the response to a simple homogeneous 3D CO2 body, gradually increasing the width and depth of the CO2. This is an analogue to vertical and lateral CO2 migration in a reservoir. I then assess the ability of CSEM to detect CO2 plume growth and evolution within the heterogeneous Bunter Sandstone reservoir model. I demonstrate the potential to detect stored and migrating CO2 and present the synthetic results as time-lapse common-offset time sections. The CO2 plume is imaged clearly and in the right coordinates. The ability to image seismically a migrating front of CO2 remains challenging due to uncertainties regarding the pore-scale saturation distribution of fluids within the reservoir and, in turn, the most appropriate rock-physics model to simulate this: uniform or patchy saturation. I account for this by modelling both saturation models, to calculate the possible range of expected seismic velocities prior to generating and interpreting the seismic response. I demonstrate the ability of seismic methods to image CO2 plume growth and evolution in the Bunter Sandstone saline reservoir model and highlight clear differences between the two rock-physics models. I then modify the Bunter Sandstone reservoir to depict a depleted gas field by including 20% residual gas saturation. I assess the importance and implication of patchy saturation and present results which suggest that seismic techniques may be able to detect CO2 injected into depleted hydrocarbon fields.

The quality of parts fabricated using Laser Solid Freeform Fabrication (LSFF) is highly dependent on the physical phenomena and operating parameters which govern the process. For instance, the thermal stress patterns and intensity, induced throughout the process domain due to the layer-by-layer material deposition and the temperature distribution characteristics, contribute significantly to potential delamination and crack formation across the fabricated part. In this research, some of the main features as well as drawbacks of this technique are studied through a multi-physics analysis of the process. For this purpose, a coupled time-dependent 3D model is developed with which the geometry of the deposited material as well as temperature and thermal stress fields across the process domain can be predicted. In the proposed approach, coupled thermal and stress domains are numerically obtained assuming a decoupled interaction between the laser beam and powder stream. To predict the geometry of the deposited material, once the melt pool boundary is obtained, the process domain is discretized in a cross-sectional fashion based on the powder feed rate, elapsed time, and intersection of the melt pool and powder stream projected on the substrate. Layers of additive material are then added onto the non-planar domain. The main process parameters affected by a multilayer deposition due to the formation of non-planar surfaces, such as powder catchment, are incorporated into the modelling approach to enhance the accuracy of the results. To demonstrate the proposed algorithm and to study the main features of the process, a four-layer thin wall of AISI 304L steel on a substrate of the same material is numerically and experimentally fabricated. The numerical analyses along with the experimental results are then used to investigate the correlation between the temperature-thermal stress fields and crack formation across the fabricated parts. The trend of the results reveals that by preheating the substrate prior to the fabrication process, it is possible to substantially reduce the formed micro-cracks. To demonstrate the feasibility of preheating on the reduction of micro-cracks, several simulations and experiments are performed in which a crack-free result is obtained, with a 22 per cent reduction in thermal stresses when the substrate is preheated to 800 K. The numerical and experimental results are also used to study the circumstances of the microstructural formation during the fabrication process. To conclude this research, the developed modelling approach is further extended to briefly discuss the effects of the path patterns and the main operating parameters on the outcomes of the process. The effects of the material properties and their variations on the temperature distributions and thermal stress fields are studied by fabrication of a thin wall of two Stellite 6 layers and two Ti layers on a stainless steel substrate.

The quality of parts fabricated using Laser Solid Freeform Fabrication (LSFF) is highly dependent on the physical phenomena and operating parameters which govern the process. For instance, the thermal stress patterns and intensity, induced throughout the process domain due to the layer-by-layer material deposition and the temperature distribution characteristics, contribute significantly to potential delamination and crack formation across the fabricated part. In this research, some of the main features as well as drawbacks of this technique are studied through a multi-physics analysis of the process. For this purpose, a coupled time-dependent 3D model is developed with which the geometry of the deposited material as well as temperature and thermal stress fields across the process domain can be predicted. In the proposed approach, coupled thermal and stress domains are numerically obtained assuming a decoupled interaction between the laser beam and powder stream. To predict the geometry of the deposited material, once the melt pool boundary is obtained, the process domain is discretized in a cross-sectional fashion based on the powder feed rate, elapsed time, and intersection of the melt pool and powder stream projected on the substrate. Layers of additive material are then added onto the non-planar domain. The main process parameters affected by a multilayer deposition due to the formation of non-planar surfaces, such as powder catchment, are incorporated into the modelling approach to enhance the accuracy of the results. To demonstrate the proposed algorithm and to study the main features of the process, a four-layer thin wall of AISI 304L steel on a substrate of the same material is numerically and experimentally fabricated. The numerical analyses along with the experimental results are then used to investigate the correlation between the temperature-thermal stress fields and crack formation across the fabricated parts. The trend of the results reveals that by preheating the substrate prior to the fabrication process, it is possible to substantially reduce the formed micro-cracks. To demonstrate the feasibility of preheating on the reduction of micro-cracks, several simulations and experiments are performed in which a crack-free result is obtained, with a 22 per cent reduction in thermal stresses when the substrate is preheated to 800 K. The numerical and experimental results are also used to study the circumstances of the microstructural formation during the fabrication process. To conclude this research, the developed modelling approach is further extended to briefly discuss the effects of the path patterns and the main operating parameters on the outcomes of the process. The effects of the material properties and their variations on the temperature distributions and thermal stress fields are studied by fabrication of a thin wall of two Stellite 6 layers and two Ti layers on a stainless steel substrate.

The University of Manchester Greg BlackThesis submitted for the degree of Doctor of EngineeringIrradiated Graphite Waste: Analysis and Modelling of Radionuclide Production with a View to Long Term Disposal23rd June 2014The UK has predominantly used graphite moderator reactor designs in both its research and civil nuclear programmes. This material will become activated during operation and, once all reactors are shutdown, will represent a waste legacy of 96,000 tonnes [1]. The safe and effective management of this material will require a full understanding of the final radiological inventory. The activity is known to arise from impurities present in the graphite at start of life as well as from contamination products transported from other components in the reactor circuit. The process is further complicated by radiolytic oxidation which leads to considerable weightloss of the graphite components. A comprehensive modelling methodology has been developed and validated to estimate the activity of the principle radionuclides of concern, 3H, 14C, 36Cl and 60Co. This methodology involves the simulation of neutron flux using the reactor physics code WIMS, and radiation transport code MCBEND. Activation calculations have been performed using the neutron activation software FISPACT. The final methodology developed allows full consideration of all processes which may contribute to the final radiological inventory of the material. The final activity and production pathway of each radionuclide has been researched in depth, as well as operational parameters such as the effect of changes in flux, fuel burnup, graphite weightloss and irradiation time. Methods to experimentally determine the activity, and distribution of key radionuclides within irradiated graphite samples have been developed in this research using a combination of both gamma spectroscopy and autoradiography. This work has been externally validated and provides confidence in the accuracy of the final modelling predictions. This work has been undertaken as part of the EU FP7 EURATOM Project: CARBOWASTE, and was funded by the Office for Nuclear Regulation.

No description available.

Die räumliche und zeitliche Organisation von Zellen während Embryonalentwicklung, Regeneration oder Erneuerung von Geweben ist eine faszinierende Fähigkeit lebender Organismen. Dazu benötigen die Zellen Informationen über ihre Position im Organismus. Diese Informationen werden oft in Form von Signal- oder Morphogengradienten bereitgestellt, also von Signalmolekülen, die Konzentrationsprofile im Raum bilden. Plattwürmer (Planarien) sind ein sehr geeigneter Modellorganismus, um solche Gewebeorganisationsprozesse zu erforschen, weil sie kontinuierlich alle Zellen ihres Körpers erneuern und aus kleinsten Gewebestücken regenerieren können. Bei einer Körperlänge von bis zu 2 cm muss Gewebe auf größeren Längenskalen organisiert werden, als es für die Embryonalentwicklung in anderen Spezies nötig ist. Trotzdem treten auch in Planarien Signalgradienten auf. Ihre Hauptkörperachse wird, wie bei anderen Tieren auch, von einem Wnt-Signalgradienten organisiert. Experimentelle Beobachtungen legen nahe, dass ein positiver Feedbackmechanismus, in dem ein Wnt-Signal zur Erzeugung von mehr Wnt-Molekülen führt, wesentlich zur Bildung dieses Gradienten beiträgt. Inspiriert durch diese Beobachtungen stellen wir in dieser Arbeit einen Mechanismus zur Ausbildung von Signalgradienten vor, der auf positivem Feedback basiert. Um die besondere Bedeutung der Zellen für dieses Feedback berücksichtigen zu können, ist das hier präsentierte Modell diskret und besteht aus Zellen und Extrazellularräumen. Das positive Feedback sorgt für eine Signalübertragung von Zelle zu Zelle, wobei die Konzentration der extrazellulären Signalmoleküle die Konzentration des intrazellulären Effektors positiv reguliert, was wiederum zur Bildung von mehr Signalmolekülen führt. Wir zeigen, dass dieser Signalübertragungsmechanismus langreichweitige Signalgradienten mit einer Längenskala von mehreren hundert Zellen, also in der Größenordnung von Millimetern, ausbildet. Die Längenskala wird durch die Stärke des positiven Feedbacks reguliert. Eine entsprechende Regulation der Feedbackstärke ermöglicht es, die Längenskala des Signalgradienten an die Größe des Systems anzupassen. Erfolgt die Sekretion der Signalmoleküle, die die Zellen als Antwort auf das Feedback produzieren, gerichtet, führt das zu einer gerichteten Ausbreitung der Signalmolekülkonzentration im System, also zu Drift. Auf diese Weise können bei biologisch relevanten Werten des Diffusionskoeffizienten und der Degradationsrate der Signalmoleküle Signalgradienten mit einer Längenskala von mehreren zehn bis hundert Zellen in Stunden bis Tagen gebildet werden. Im Unterschied zum Diffusions/Degradations-Mechanismus, der häufig zur Erklärung von Gradientenbildung im Kontext von Embryonalentwicklung herangezogen wird, benötigt der in dieser Arbeit präsentierte Signalübertragungsmechanismus also weder sehr schnell diffundierende noch sehr langlebige Moleküle, um die Bildung von langreichweitigen Signalgradienten auf biologisch relevanten Zeitskalen zu erklären. Da viele Morphogene langsam diffundieren, macht das den Zell-zu-Zell-Signalübertragungsmechanismus zu einem attraktiven Konzept, um die Bildung von langreichweitigen Morphogengradienten zu erklären. / Embryonic development, regeneration, and tissue renewal are spectacular tissue-patterning events. Tissue patterning requires information. This information is often provided by signalling molecules that form graded concentration profiles in space, referred to as signalling gradients or morphogen gradients. Planarian flatworms are an ideal model organism to study tissue patterning as they constantly turn over all of their tissues and are able to regenerate from arbitrary amputation fragments. At a body length of up to 2 cm, planarians are orders of magnitudes larger than tissues organised during embryonic development in other species. Yet, flatworms employ signalling gradients for tissue patterning. Like in other organisms throughout the animal kingdom, their main body axis is patterned by a Wnt signalling gradient. Experiments have suggested a positive feedback mechanism of Wnt-mediated Wnt expression to be implicated in the formation of this Wnt signalling gradient in planarians. Inspired by these observations, in this thesis we present a cell-to-cell relay mechanism based on positive feedback to explain long-ranged signalling gradient formation. To account for the cellular nature of the relay, we built a discrete model, that considers individual cells and extracellular spaces. The relay is generated by a positive feedback loop in which extracellular signalling levels positively regulate intracellular effector concentrations which in turn leads to production of more extracellular signalling molecules. We show that a cell-to-cell relay gives rise to steady-state gradients reaching length scales of the order of hundreds of cells, corresponding to millimetres. The length scale is regulated by the strength of the feedback, which allows scaling the steady-state gradient to tissue size by adapting the feedback strength. Polarised secretion of signalling molecules in response to the positive feedback leads to an effective drift of signalling molecule concentration through the system. This allows the formation of signalling gradients with a length scale of tens to hundreds of cells (millimetres) within hours to days for a physiologically relevant diffusion coefficient and degradation rate of the signalling molecules. Thus, in contrast to a diffusion/degradation-based mechanism that is widely used to explain signalling gradient formation during embryonic development, the relay mechanism requires neither extraordinarily quickly-diffusing nor very long-lived signalling molecules to explain the formation of long-ranged signalling gradients on biologically relevant time scales. The cell-to-cell relay mechanism is therefore an attractive concept to explain the long-ranged patterning effects of poorly diffusive morphogens.

In the past years, more concerns are focused on the nuclear waste management due to the very long half-lives of various actinides produced in Light Water Reactors (LWRs). Sodium Fast Reactors (SFRs) are thus becoming more attractive since they are known to be very efficient to transmute long-lived radionuclides present in spent fuel. However, the current simulation tools (thermal-hydraulics code with point kinetics) and safety assessment methods are not as mature as for LWR applications and need to be enhanced. This thesis aims at filling the gap in safety analysis of SFR cores to reach a standard similar to LWR applications by applying multi-physics modelling. In contrast to LWRs, the reactivity in SFRs is affected by three main feedback: the Doppler broadening reactivity effect, the sodium density change reactivity effect and the thermal expansion of several mechanical components of the reactor. In this thesis, the thermal-hydraulic system code ATHLET is coupled with the three-dimensional neutron-physics code PARCS for transient analysis. Developed at GRS, ATHLET was recently upgraded for sodium coolant properties. The nodal diffusion codes PARCS, developed at the University of Michigan, can solve the multi-group diffusion equation in hexagonal geometry. While both codes already have the main features to simulate SFRs, the development of models dedicated to the thermal expansion effect of reactivity is necessary. The latter has three main origins i.e. the core axial thermal expansion effect (caused by the fuel and the cladding axial thermal expansion), the core radial thermal expansion effect (caused by the diagrid thermal expansion), the control rod displacement due to the thermal expansion of the Control Rod Drive Lines (CRDLs), the strongback and the reactor vessel. Thus, the three main new developments achieved in the scope of this work are: - Development of a method to generate homogenized multi-energy-group neutron macroscopic cross sections (needed by PARCS) for SFR applications which consider not only the Doppler temperature and sodium density but also the core axial and radial thermal expansion. - Development of a three-dimensional core radial thermal expansion model and its implementation in PARCS. A core axial thermal expansion model has already been developed for PARCS prior to this work. - Development of a module in ATHLET for modelling the control rod displacement as a result of the influence of the reactor structures thermal expansion. The parametrized homogenized multi-energy-group neutron macroscopic cross section libraries for PARCS applications are generated with the Monte Carlo reactor-physics code Serpent. For all materials contained in fuel assemblies, a three-dimensional model is used while the SPH method is applied to materials contained in non-fuel assemblies (e.g. control rods, etc.). The cross section libraries are collapsed into a 12-energy-group structure. Furthermore, a dedicated module was successfully developed and implemented within the core simulator KMACS (developed at GRS). The core radial thermal expansion effect is implemented in PARCS using a coordinate transformation of the diffusion equation from the expanded state to the nominal geometry. The core radial thermal expansion depends on the diagrid temperature. It is calculated by ATHLET and transferred to PARCS by the extended interface between both codes. The modelling of the control rod displacement as a result of the reactor structures thermal expansion is performed by a module linked to ATHLET. The strongback, the reactor vessel and the CRDLs are modelled as heated structures in ATHLET, which calculates their respective temperature. The module can compute the thermal expansion of each structure as well as the total control rod banks displacement. The new techniques are verfied on a selected case study, the ASTRID core design. First, full core criticality simulations are performed with the Monte Carlo reactor-physics code Serpent (considered as reference calculations) and with PARCS. Good agreement between the two codes is achieved in terms of multiplication factors and power distribution. This allows to conclude that the developed method for neutron cross section libraries can be used for SFR applications. The newly implemented core radial expansion model in PARCS is successfully verified on the ASTRID core with the standalone version of PARCS. Then, various transient simulations are performed in order to separately analyse the different contributions to the reactivity by: the Doppler broadening effects, the sodium density change effect, the core radial and axial thermal expansion effect and the control rod displacement effect. It is demonstrated that the core power responses are plausible which allows the conclusion that all the different thermal expansion models are properly implemented. Furthermore, the presented simulations show very different core power responses. It appears that the effect of the sodium density change on reactivity is a parameter that is strongly heterogeneous (depending on the core location). This shows the importance of using a three-dimensional neutron kinetics model rather than a point-kinetic model for transient simulations with thermal-hydraulic codes. Moreover, the time-scale of the various effects are ranging from few seconds to several hundred seconds. While the Doppler broadening, the sodium density change, as well as the core axial and radial thermal expansion effects on reactivity are fast, the thermal expansion of the strongback and the vessel only appears after several hundred seconds. This emphasizes the importance of considering all thermal expansion effects in addition to the usual thermal-hydraulic feedback parameters (e.g. fuel temperature, coolant density etc.) to be able to compute the core behavior realistically.:Contents Abstract II List of Figures VII List of Tables X List of Acronyms XI Acknowledgments XIII 1 Introduction 1.1 Sodium cooled fast reactors 1.1.1 Fast reactor development 1.1.2 Comparison of sodium fast reactor and pressurized water reactor designs 1.1.2.1 Neutron spectrum 1.1.2.2 Breeding 1.1.2.3 Partitioning and Transmutation 1.1.2.4 Control of the reactivity in the core 1.1.2.5 Coolant properties 1.1.2.6 Reactivity feedback 1.1.2.7 Comparison summary 1.2 Objectives and structure of the thesis 1.2.1 Objectives 1.2.2 Structure of the thesis 2 State of the art of Sodium Fast Reactor safety assessment 2.1 Relevant safety events to consider for Sodium Fast Reactors 2.2 Major reactivity feedback mechanisms 2.3 State of the art of safety analysis methods for Sodium Fast Reactor 3 Methods and codes for safety assessment of sodium cooled fast reactors 3.1 Neutronics core calculations 3.1.1 Core calculations with the diffusion code PARCS 3.1.2 Generation of nodal few-group cross sections with the Monte Carlo code Serpent 3.1.3 Core simulator KMACS 3.2 Thermal-hydraulics simulations with the system code ATHLET 3.3 Coupled three-dimensional thermal-hydraulics / neutronics calculations 4 Development of three-dimensional thermal expansion models 4.1 General calculation approach proposed for safety assessment 4.2 Thermal expansion in solids 4.3 Model for generating nodal few-energy-group cross sections for deterministic core analysis 4.3.1 Energy group structure 4.3.2 Full-scale three-dimensional fuel assembly models in Serpent 4.3.3 Two-dimensional non-fuel assembly models in Serpent 4.3.4 Super homogenization method for non-multiplying media 4.3.5 Automated creation of Serpent models for parametrized cross section generation with KMACS 4.4 Core radial thermal expansion effect 4.4.1 Description of the core radial thermal expansion phenomenon 4.4.2 Coordinate transformation of the diffusion equation 4.4.3 Implementation of the coordinates transformation in PARCS 4.4.4 Adapted cross section parametrization scheme for the core radial expansion model 4.4.5 Diagrid model in ATHLET and temperature transfer 4.5 Core axial thermal expansion effect 4.5.1 Description of the core axial thermal expansion phenomenon 4.5.2 Implementation of a core axial thermal expansion model in PARCS 4.5.3 Appropriate cross section parametrization scheme 4.6 Control rod displacement due to reactor structures thermal expansion effects 4.6.1 Modelling scheme 4.6.2 Strongback model in ATHLET 4.6.3 Vessel model in ATHLET 4.6.4 Control rods drive lines ATHLET model 5 Verification on a case study 5.1 Description of the ASTRID reactor 5.2 Full core models 5.2.1 Full core Serpent reference models of the ASTRID core 5.2.2 Three-dimensional neutron kinetics model of ASTRID core in PARCS 5.2.3 Generation of appropriate few-group cross sections 5.2.4 Thermal-hydraulic model in ATHLET and ATHLET-PARCS feedback mapping 5.3 Verfications of the radial core expansion model 5.4 Assessment of the Doppler and sodium density effects 5.4.1 Assessment of the Doppler effect 5.4.2 Assessment of the sodium density effect 6 Coupled three-dimensional thermal-hydraulics/neutron-physics transient simulations with ATHLET-PARCS 6.1 Description of the models and transient simulations 6.2 Simulation 1: Doppler effect 6.2.1 Description 6.2.2 Results 6.3 Simulation 2: Sodium density effect 6.3.1 Description 6.3.2 Results 6.4 Simulation 3: Doppler and sodium density effects 6.4.1 Description 6.4.2 Results 6.5 Simulation 4: Core radial thermal expansion effect 6.5.1 Description 6.5.2 Results 6.6 Simulation 5: Doppler, Sodium density and core radial thermal expansion effects 6.6.1 Description 6.6.2 Results 6.7 Simulation 6: Core axial thermal expansion effect 6.7.1 Description 6.7.2 Results 6.8 Simulation 7: Doppler, Sodium density and core axial thermal expansion effects 6.8.1 Description 6.8.2 Results 6.9 Simulation 8: Doppler effect, Sodium density effect, core radial thermal expansion effect and core axial thermal expansion effect 6.9.1 Description 6.9.2 Results 6.10 Simulation 9: Doppler effect, Sodium density effect, core radial thermal expansion effect, core axial thermal expansion effect and control rod displacement due to reactor structures thermal expansion effect 6.10.1 Description 6.10.2 Results 6.11 Preliminary conclusions of the test calculations 7 Conclusion and outlook for future developments 7.1 Summary and conclusions 7.2 Suggestions for future work Appendices A The Boltzmann equation B Macro-group structure Bibliography

Les modèles aux moments angulaires constituent des descriptions intermédiaires entre les modèles cinétiques et les modèles fluides. Dans ce manuscrit, les modèles aux moments angulaires basés sur un principe de minimisation d'entropie sont étudiés pour des applications en physique des plasmas. Ce mémoire se découpe en trois parties. La première est une contribution à la modélisation en physique des plasmas à travers le formalisme des modèles aux moments angulaires. Dans celle-ci, le domaine de validité de ces modèles est étudié en régimes non-collisionels. Il est également montré que les opérateurs de collisions proposés pour le modèle M1 permettent de retrouver des coefficients de transport plasma précis. La deuxième partie de ce document concerne la dérivation de méthodes numériques pour l'étude du transport de particules en temps long. Dans ce cadre, des schémas numériques appropriés pour le modèle M1, préservant l'asymptotique, sont construits et validés numériquement. La troisième partie représente un premier pas significatif vers la modélisation multi-espèces. Ici, le modèle aux moments angulaire M1, construit dans un référentiel mobile, est appliqué à la dynamique des gaz raréfiés. Les propriétés de ce modèle sont détaillées, un schéma numérique est proposé et une validation numérique est menée. / Angular moments models represent alternative descriptions situated in between the kinetic and the fluid models. In this work, angular moments models based on an entropy minimisation principle are considered for plasma physics applications. This manuscript is organised in three parts. The first one is a contribution to plasma physics modelling within the formalism of angular moments models. The validity domain of angular moments models in collisionless regimes is studied. It is also shown that the collisional operators proposed for the M1 angular moments model enable to recover accurate plasma transport coefficients. The second part of this document deals with the derivation of numerical methods for the long timescales particle transport. Appropriate asymptotic-preserving numerical schemes are designed for the M1 angular moments model and numerical validations are performed. The third part represents a first important step toward multi-species modelling. The M1 angular moments model in a moving frame is introduced and applied to rarefied gas dynamics. The model properties are highlighted, a numerical scheme is proposed and a numerical validation is carried out.

L’utilisation des matériaux composites devient de plus en plus importante dans les structures aéronautiques de nouvelle génération. Le gain de masse engendré, et donc de carburant, pousse les constructeurs aéronautiques à les employer de façon optimale. Néanmoins, ces matériaux se dégradent rapidement lorsqu’ils sont soumis à des flux de chaleur importants, entraînant une perte de leur résistance mécanique. Ce problème peut être dramatique pour la sécurité des passagers car la tenue de ces nouvelles structures peut ne plus être assurée dans le cas d’un incendie. Les méthodes actuelles de certification de la tenue au feu des matériaux composites aéronautiques reposent principalement sur l’utilisation de moyens expérimentaux, dont les résultats ne sont représentatifs que des conditions particulières dans lesquelles les essais ont été réalisés. La compréhension des différents phénomènes thermiques, chimiques et mécaniques intervenant lors de la dégradation de ces matériaux, avec l’appui de simulations numériques et d’expériences, peut permettre d’améliorer les méthodes existantes et donc d’optimiser les futures structures aéronautiques dès la phase de conception.Cette étude s’est attachée à développer et valider un modèle thermo-chimique de dégradation des matériaux composites multi-dimensionnel et multi-constituants. Ce modèle permet de traiter des cinétiques de dégradation complexes suivant plusieurs réactions de décompositions et de prendre en compte le transport des gaz produits depuis leur formation jusqu’à leur évacuation hors du matériau. L’utilisation de lois d’homogénéisation avancées est proposée afin de rendre compte des effets des transformations sur les transferts de chaleur et de masse se produisant au sein du matériau. L’application du modèle thermo-chimique à un cas de dégradation sous flux thermique connu mais non-uniforme dans un environnement contrôlé permet de confronter les résultats de simulation aux mesures expérimentales et ainsi de valider l’approche multi-constituants adoptée. Enfin, l’étude numérique de la dégradation d’un composite soumis à une flamme met en avant l’effet des gaz de décomposition éjectés à l’interface sur le flux thermique pariétal échangé. / Composite materials are increasingly used in new generation aircraft structures. Mass and as a consequence fuel savingsencourage aircraft manufacturers to use them optimally. However, these materials can degrade quickly when exposed tosignificant heat fluxes, resulting in a loss of mechanical strength. This problem can be dramatic for passenger safety asmechanical resistance of such innovative structures can not be ensured in case of fire events. Current certification methodsof fire resistance of aeronautical composite materials are mainly based on experiments, that are only representative of thespecific conditions under which they were carried out. The understanding of thermal, chemical and mechanical phenomenaoccurring during the decomposition of these materials, with the support of numerical simulations and experiments, can helpimproving existing methods and optimizing the future aeronautical structures from the design chain. This study deals withthe development and validation of a multi-components and multi-dimensional thermo-chemical model of decomposing compositematerials. It can deal with complex degradations following several decomposition reactions as well as transport ofpyrolysis gases from their formation up to their ejection out of the material. The use of advanced homogenization laws isproposed to account for the chemical transformations on heat and mass transfers occurring in the material. The applicationof the thermo-chemical model to a laser degradation study under known but non-uniform heat flux in a controlled environmentallows to confront the simulation results with experimental measurements and thus validate the multi-componentsapproach. Finally, the numerical analysis of a decomposing composite material submitted to a flame highlights the effectof emitted decomposition gases on the exchanged parietal heat flux.