Le titane et ses alliages qui sont largement répandus dans l'industrie aéronautique, sont concernés par le fluage à température ambiante ce qui conduit à une réduction de la résistance et provoque le phénomène de rupture différée. Une partie des études montrent que ce comportement viscoplastique inhabituel à température ambiante est lié aux phénomènes d'interactions entre les dislocations et les atomes interstitiels comme l'hydrogène et l'oxygène, aussi appelés vieillissement statique et dynamique. Le but de cette étude à la fois expérimentale et numérique multi-échelle est de mieux comprendre les effets souvent antagonistes et en partie couplés de l'oxygène et de l'hydrogène en solution sur le comportement viscoplastique du titane non-allié de phase alpha. Dans un premier volet, un scénario du vieillissement statique et dynamique dans le titane non-allié de phase alpha est proposé. La présence du pic de traction est attribuée à la ségrégation des atomes interstitiels d'oxygène sur les dislocations coin de vecteur de Burgers <c+a>. Dans le cas du vieillissement dynamique les instabilités observées, typiques de l'effet Portevin-Le Chatelier, sont associées à l'étalement du cœur non planaire des dislocations vis de vecteur de Burgers <a>. Une loi de comportement prenant en compte les effets liés aux interactions entre dislocations et atomes en solution a été développée. Le modèle de Kubin-Estrin-McCormick qui permet de prendre en compte l'effet du vieillissement a été étendu au cas de la plasticité cristalline. Les simulations par éléments finis ont été réalisées sur des agrégats polycristallins avec différents nombres de grains. Ensuite, les essais de fissuration (ténacité et rupture différée) ont été réalisés sur les matériaux bruts, et chargés en hydrogène. Enfin, des simulations numériques de la rupture de ces éprouvettes ont été réalisées pour toutes les conditions expérimentales testées en utilisant le modèle de comportement mécanique macroscopique identifié. Un modèle de zone cohésive a été développé pour la simulation de la propagation des fissures. / Widely used for aircraft or rocket engine manufacturing titanium and its alloys are prone to the room-temperature creep that leads to the phenomenon of sustained load subcritical crack growth. One of the major cause of such unusual viscoplastic behavior of titanium is the phenomena of static and dynamic strain aging which represents an interaction between dislocations and interstitial atoms of oxygen and hydrogen. The aim of the present experimental and numerical multiscale study is to investigate the influence of the interstitial hydrogen and oxygen on the viscoplastic behavior and the resistance to sustained load cracking in commercially pure titanium of phase alpha.In a first step, a scenario of static and dynamic strain aging was proposed. The presence of the stress peak was attributed to the segregation of interstitial atoms of oxygen on the edge <c+a> dislocations. In case of dynamic strain aging, the observed instabilities, typical for the Portevin-Le Chatelier effect, were associated with the non-planar core of screw <a>-type dislocations. The crystal plasticity was introduced into the phenomenological model in order to capture the strain aging phenomena and the anisotropy of the mechanical properties. The modeling approach for strain aging suggested by Kubin-Estrin-McCormick is based on the internal variable called the aging time which corresponds to the waiting time of a dislocation in a pinned state. Finite element simulations were then performed on the polycrystalline aggregates for different number of grains. At the next step, fracture toughness and sustained load cracking tests were performed on the material with different levels of hydrogen. Finally, numerical simulations of toughness and sustained load cracking tests using the identified viscoplastic model were carried out for all experimental conditions. A cohesive zone model was then introduced ahead of the crack tip to simulate crack propagation.

La propagation des fissures courtes de fatigue dans un matériau polycristallin dépend fortement de la microstructure. Bien que de nombreuses études de caractérisation et de modélisation existent sur le sujet, la prédiction du chemin et de la vitesse de propagation de ce type de fissure n'est pas encore possible aujourd'hui.Afin de bien comprendre les mécanismes de propagation, la caractérisation in-situ d'un échantillon par la tomographie aux rayons X a été réalisée à l'ESRF en combinant deux techniques de caractérisation. La tomographie par Contraste de Diffraction (DCT) qui est une méthode non destructive permettant de caractériser en 3D la morphologie et l'orientation des grains constitutifs de la microstructure, à l'état non-déformé, et la tomographie par Contraste de Phase (PCT) qui permet d'obtenir la forme de fissure à divers étapes de la vie de l'éprouvette. Grâce à ces informations, il est possible de simuler la propagation de fissure en utilisant un maillage réaliste reconstruit à partir des images tomographiques. Dans ce travail, une étude de l'anisotropie de comportement élastique est effectuée dans un maillage microstructural 3D reconstruit à partir des images tomographiques. Cette étude permet de comparer les tenseurs de déformation élastique moyennés à chaque grain avec les mesures expérimentale. Ensuite, une nouvelle méthodologie est proposée pour simuler la propagation de fissure. Issue d'une simulation en plasticité cristalline, la direction et la vitesse de la propagation de fissure est déterminée par un post-traitement, ce qui permet de propager la fissure par remaillage. Cette méthode est appliquée dans un premier temps à un monocristal pré-fissuré pour prédire le trajet de fissuration en fonction des systèmes de glissement activés. L'ensemble de la démarche est enfin appliqué au polycristal complet imagé par tomographie. Le rôle du joint de grains et la vitesse de propagation sont également analysés. En comparant les résultats de simulation avec les mesures expérimentales, le critère de la propagation de fissure est discuté. / The short fatigue crack propagation in polycrystal materials depends strongly on microstructure. Although numerous studies of characterisation and of simulation, the prediction of the short fatigue crack propagation remains a challenge.In order to understand the mechanisms of short fatigue crack propagation, an in-situ characterisation by X-ray tomography was carried out at ESRF, using two techniques of tomography. Diffraction Contrast Tomography (DCT) that is a non-destructive method can be used to obtain 3D morphology and grain orientations in an undeformed state of polycrystal materials. Couple with Phase Contrast Tomography (PCT), it allows to characterise the short fatigue crack propagation at different loading stages. Access to this information, it is possible to simulate the short fatigue crack propagation using a 3D reel microstructural mesh reconstructed from the tomographic images.In this work, the elastic anisotropic behaviour in a 3D microstructural mesh is performed. The elastic strain tensors averaged in grains are also compared to the experimental measurements. Then, a new numerical approach is proposed to simulate crack propagation. From a crystal plasticity FE simulation, the crack growth direction is determined by a post processing. Next, the crack is propagated through remeshing. This approach is firstly applied to the single crystals, then to the polycrystal mesh reconstructed from the tomographic images. The grain boundary effects and the crack growth rate are also analysed. By comparing between simulation and experimental crack, the damage indicator is discussed at the end.

De nombreuses structures aéronautiques telles que les contacts aube/disque, rivet/tôle, assemblages frettés ou boulonnés subissent des endommagements de fretting fatigue dus à des sollicitations vibratoires. Ces sollicitations, qui combinent des chargements de contact de fretting (micro déplacements alternés) et de fatigue, peuvent induire des phénomènes de fissuration. L’objectif de cette thèse est d’identifier les mécanismes de propagation et d’arrêt de fissuration pour différentes configurations, impliquant des chargements de fretting et de fatigue. Différents aspects seront étudiés : - Les essais de fretting simple, fretting fatigue et fatigue seront corrélés de manière à mesurer certaines propriétés mécaniques de matériaux. Dans cette démarche, une méthode inverse permettant l’identification du seuil d’arrêt de propagation des fissures longues et courtes en fatigue à partir d’essais de fretting simple et fretting fatigue est introduite. - L’effet du rapport de charge est pris en compte afin de quantifier la fermeture de fissure pour les essais de fretting et fretting fatigue. Un nouveau modèle de fermeture sera proposé. - L’effet d’échelle est étudié grâce au développement d’une nouvelle machine de fretting fatigue permettant le suivi in situ des propagations de fissures sur des éprouvettes millimétriques. Le comportement des fissures courtes en fretting pendant un essai de fretting fatigue a été observé pour la première fois. L’idée principale de ce travail est de considérer que lors d’un essai en fretting simple, une condition d’arrêt de propagation est systématiquement atteinte. Les facteurs d’intensités de contraintes seuils en fatigue peuvent ainsi être extraits par une méthode inverse à partir des essais de fretting. La méthodologie adoptée consiste à mesurer la longueur de fissure maximale obtenue lors d’un essai de fretting correspondant à une condition d’arrêt de propagation. En jouant sur les chargements de fretting, les domaine des fissures courtes et longues sont étudiés. Pour chaque condition d’arrêt, le facteur d’intensité de contraintes seuil correspondant est calculé. Enfin, en traçant l’évolution de ce dernier en fonction de la longueur de fissure, une description complète des seuils d’arrêt de fissuration en fatigue peut être obtenue. Cette méthodologie est appliquée sur plusieurs configurations de fretting simple et est étendue aux essais de fretting fatigue. Ainsi une large gamme de rapports de charge peut être étudiée. Des essais conventionnels de fatigue propagation sur éprouvettes C(T) ont été réalisés de manière à valider complètement la démarche d’identification inverse des conditions d’arrêt de fissuration. Le manuscrit est divisé en six chapitres adoptant une évolution linéaire. Le chapitre 1 traite d’une revue bibliographique concernant la propagation des fissures courtes et longues en fatigue dans les matériaux métalliques, introduisant les phénomènes de fermeture. L’analyse est étendue à la propagation des fissures en fretting et fretting fatigue. Le chapitre 2 présente l’ensemble des techniques et méthodes expérimentales utilisées au cours de ce travail. Une analyse numérique et la méthode inverse d’identification des conditions d’arrêt de propagation sont introduites dans le chapitre 3. Les chapitres suivants traitent les différents résultats obtenus. Le chapitre 4 présente les essais de fretting simples menés sur différents matériaux métalliques. Une corrélation entre les conditions d’arrêt de fretting simple avec celles de fissures longues en fatigue est obtenue pour rapports de charge négatifs. Cette analyse est étendue aux rapports de charge positifs au chapitre 5 en couplant les essais de fretting simple, fretting précontraints et essais de fatigue C(T). Enfin, les effets d’échelle et la cinétique de propagation des fissures de fretting fatigue sont étudiés au chapitre 6. / Aircraft structures such as blade/disk, rivet/sheet and fretted or bolted assembly contacts are subjected to fatigue fretting damage caused by vibratory loads. These loads combining fretting (alternating micro displacements) and fatigue contact loading can induce cracks. The objective of this thesis is to identify the propagation and crack arrest mechanisms for different configurations involving fretting and fatigue loads. Various aspects will be studied : - Simple fretting, fatigue fretting and fatigue tests will be correlated in a way to measure some material mechanical properties. A reverse approach allowing to identify the long and short fatigue crack propagation thresholds from simple fretting and fretting fatigue will be introduced. - A new closure model taking into account the effect of loading ratio on crack closure in fretting and fretting fatigue tests will be proposed. - The scaling effect will be studied thanks to the development of a new fretting fatigue machine allowing for in-situ monitoring of the propagation of small scale cracks. The behavior of fretting short cracks has been observed for the first time in a fretting fatigue test. The main idea behind the current work is the consideration that for a simple fretting test, the crack will always lead to a crack propagation arrest condition. Threshold stress intensity factors can thus be extracted from fretting tests using a reverse method. The methodology used consists in measuring the longest fretting crack corresponding to a crack propagation arrest condition. Short and long crack regimes are studied by varying fretting loads and geometries. The corresponding threshold stress intensity factor is calculated for each arrest condition. A complete description of fatigue crack arrest thresholds can be obtained from a plot depicting the evolution of the stress intensity factor range as a function of crack length. This methodology has been applied on multiple simple fretting configurations and was extended to fretting fatigue tests allowing to study a large range of loading ratios. Conventional fatigue crack propagation tests were carried out on C(T) specimens in order to validate the reverse method used for crack arrest conditions. The present thesis is divided into six main chapters. Chapter 1 provides a complete literature review on short and long crack propagation in metallic materials introducing crack closure. It provides a treatment for both fretting and fretting fatigue including a comparative analysis. Chapter 2 provides an overview of the different techniques and methodologies used as part of this research project. The reverse method used for the identification of crack propagation arrest conditions as well as a numerical analysis are presented in Chapter 3. The following Chapters provide a treatment of the various results obtained. Chapter 4 focuses on simple fretting tests carried out on different metallic materials. A correlation between simple fretting and long fatigue crack arrest conditions was obtained for negative loading ratios. This analysis is extended to positive loading ratios in Chapter 5 combining simple fretting, prestressed fretting and C(T) fatigue tests. Finally, scaling effects and fretting fatigue crack propagation kinetics are studied in Chapter 6.

L'objectif de ce travail est d'étudier l'influence de l'état de dispersion de la silice sur les propriétés viscoélastiques et mécaniques des élastomères chargés silice pour l'application pneumatique. En effet, il est montré qu'une grande partie de la résistance au roulement dépend de la dissipation d'énergie de l'élastomère chargé. De plus, les propriétés mécaniques pourraient intervenir dans les propriétés ultimes telles la propagation de fissure en fatigue et l'usure : l'étude de ces différentes propriétés est donc primordiale. La première partie s'attèle donc à l'identification des leviers pouvant moduler la dispersion de la silice dans les élastomères. La dispersion pouvant être vue comme la compétition entre les forces de cohésion des charges et les forces appliquées au système pour les rompre, ces paramètres ont été étudiés de manière systématique. Enfin, l'influence des interactions silice-matrice est abordée (modification de l'état de surface de la silice, nature de l'élastomère) : le nombre d'interactions silice-élastomère par chaîne de polymère pourrait intervenir dans les mécanismes de dispersion. Dans la seconde partie, l'influence de ces différents états de dispersion sur les propriétés viscoélastiques et mécaniques est discutée. L'amélioration de la dispersion à iso-formulation permet de diminuer le module élastique linéaire dans le domaine linéaire et entraîne une augmentation des modules aux grandes déformations en traction. Concernant l'étude des propriétés ultimes, notre dispositif expérimental n'a pas permis de mettre en évidence, sur les formulations testées, d'influence de la dispersion sur la dynamique de propagation de fissure en fatigue. En revanche, l'amélioration de la dispersion entraîne une amélioration de la résistance à l'usure et ce malgré une diminution de la dureté / Filled elastomers are used in tread tires. It has been demonstrated that most of rolling resistance of tires is due to filled elastomer energy dissipation. In that way, understanding viscoelastic properties of these materials is a key point. Then, filled elastomer behavior at high deformations may be involved in ultimate properties of tire application such as fatigue crack propagation and wear. The aim of this work is to study the impact of silica dispersion on viscoelastic and mechanical properties of filled elastomers. First, levers impacting silica dispersion are evaluated. Dispersion of fillers can be considered as a competition between fillers cohesion forces and applied forces to the system to break them, these parameters have been studied methodically. Finally, the impact of silica-matrix interactions is studied, changing silica surface treatments or elastomer natures. The quantity of interactions possible per polymer chain between the silica and the elastomer may play a role in silica dispersion. Secondly, the impact of silica dispersion on viscoelastic and mechanical properties is discussed. It is shown than increasing silica dispersion leads to a decrease of linear elastic modulus and an increase of reinforcement in tensile at high deformations. Finally, regarding ultimate properties, our experimental device on the selected formulations has not shown any impact of silica dispersion on fatigue crack propagation. Nevertheless, we observe a better wear resistance with increasing dispersion, despite the lower materials hardness

The problem of vibration induced structural faults has been a real one in engineering over the years. If left unchecked it has led to the unexpected failures of so many structures. Needless to say, this has caused both economic and human life losses. Therefore for over forty years, structural damage identification has been one of the important research areas for engineers. There has been a thrust to develop global structural damage identification techniques to complement and/or supplement the long-practised local experimental techniques. In that respect, studies have shown that vibration-based techniques prove to be more potent. Most of the existing vibration-based techniques monitor changes in modal properties like natural frequencies, damping factors and mode shapes of the structural system to infer the presence of structural damage. Literature also reports other techniques which monitor changes in other vibration quantities like the frequency response functions, transmissibility functions and time-domain responses. However, none of these techniques provide a complete identification of structural damage. This study presents a damage detection technique based on operational response monitoring, which can identify all the four levels of structural damage and be implemented as a continuous structural health monitoring technique. The technique is based on monitoring changes in internal data variability measured by a test statistic <font face="symbol">c</font>2Ovalue. Structural normality is assumed when the <font face="symbol">c</font>2Om value calculated from a fresh set of measured data is within the limits prescribed by a threshold <font face="symbol">c</font>2OTH value . On the other hand, abnormality is assumed when this threshold value has been exceeded. The quantity of damage is determined by matching the <font face="symbol">c</font>2Om value with the <font face="symbol">c</font>2Op values predicted using a benchmark finite element model. The use of <font face="symbol">c</font>2O values is noted to provide better sensitivity to structural damage than the natural frequency shift technique. The analysis carried out on a numerical study showed that the sensitivity of the proposed technique ranged from three to thousand times as much as the sensitivity of the natural frequencies. The results from a laboratory structure showed that accurate estimates of damage quantity and remaining service life could be achieved for crack lengths of less than 0.55 the structural thickness. This was due to the fact that linear elastic fracture mechanics theory was applicable up to this value. Therefore, the study achieved its main objective of identifying all four levels of structural damage using operational response changes. / Dissertation (MSc (Mechanics))--University of Pretoria, 2007. / Mechanical and Aeronautical Engineering / unrestricted

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Titanium alloys are used for several applications of greater responsibility, for example in the biomedical, aeronautical, energy, security, among others sectors. Several studies have shown that TiNbSn alloys have advantages compared to other titanium alloys with respect to mechanical properties. Studies indicate a possibility of obtaining a tensile strength value as high as that of alloys traditionally used as Ti6Al4V, through cold deformation and aging. In addition, cold deformed TiNbSn alloys may have modulus of elasticity less than half the modulus presented by traditional alloys. These indications point to the possibility of intensifying the production and commercialization of TiNbSn alloys for the manufacture of highly responsible components. Therefore, for these applications, studies on the crack propagation behavior are important and even mandatory in certain circumstances, for example for the manufacture of components that may suffer catastrophic failures. The purpose of this study was to evaluate the fatigue crack propagation behavior of TiNbSn alloys system cold deformed and aged. In this study, the cold-rolled Ti35Nb2Sn, Ti42Nb2Sn and Ti42Nb alloys were produced with 80% true deformation and aged at 400 ° C for 48 hours. da/dN x ΔK tests were carried out and it was observed that the Ti422 and Ti420 alloys, both aged, had the best results of crack propagation resistance in the Paris regime and in the crack propagation threshold. That is, the increase of Niobium and the aging treatment produces alloy of greater crack propagation resistance. These results were attributed to the great cracks tortuosity. This tortuosity leads to a greater roughness induced crack closure effect. The threshold values increased with the aging process due to the increase in the modulus of elasticity values. / Ligas de titânio são utilizadas para várias aplicações de maior responsabilidade, como por exemplo, nos setores biomédico, aeronáutico, energia, segurança, entre outros. Diversos estudos vem demonstrando que as ligas TiNbSn apresentam vantagens se comparadas a outras ligas de titânio com respeito a propriedades mecânicas. Estudos indicam a possibilidade de obter valor de resistência a tração tão elevado quanto de ligas tradicionalmente usadas como Ti6Al4V, através de deformação a frio e envelhecimento. Além disso, as ligas TiNbSn deformadas a frio podem apresentar módulo de elasticidade inferior a metade do módulo apresentado por ligas tradicionais. Esses indícios apontam para a possibilidade de intensificar a produção e comercialização de ligas TiNbSn para a fabricação de componentes de alta responsabilidade. Portanto, para essas aplicações, estudos sobre o comportamento de propagação de trincas são importantes e até mandatórios em certas circunstâncias, como por exemplo para a fabricação de componenetes que possam sofrer falhas catastróficas. O objetivo do presente estudo foi avaliar o comportamento de propagação de trincas por fadiga das ligas do sistema TiNbSn deformadas a frio e envelhecidas. Neste estudo, foram produzidas as ligas Ti35Nb2Sn, Ti42Nb2Sn e Ti42Nb laminadas a frio com 80% de deformação verdadeira e envelhecidas a 400 °C por 48 horas. Ensaios da/dN x ΔK foram realizados e foi visto que as ligas Ti422 e Ti420, ambas envelhecidas, apresentaram os melhores resultados de resistência à propagação de trincas no regime de Paris e no limiar de propogação de trinca. Ou seja, o aumento de Nióbio e o tratamento de envelhecimento produzem liga de maior resitência a propagação de trinca. Esses resultados foram atribuídos à grande tortuosidade das trincas. Essa tortuosidade leva a um maior efeito de fechamento de trinca induzido pela rugosidade da fratura. Os valores de limiar aumentaram com o processo de envelhecimento devido ao aumento dos valores de módulo de elasticidade.

The strength of materials such as glass and ceramics can be influenced by the environment (stress corrosion). Under applied stress defects (cracks) can grow sub-critically below fracture toughness K_Ic. The aim of this work was to develop a three-dimensional finite-element model to analyze the subcritical crack growth behavior of ceramic-based particulate composites. The maximum tangential stress criterion (MTS) was used to predict the direction of the crack propagation in the framework of linear elastic fracture mechanics. The modeled material was a Low Temperature Co-fired Ceramics (LTCC), containing alumina particles embedded in a glass matrix. The main aim of this work was to develop a 3D model describing the crack growth. Conclusions from this work can contribute to a better understanding of subcritical crack propagation in particle composites.

This work is focused on the realization of experiment allowing simultaneous loading under mode II and III in a single circular specimen. Proposed experiment allowed to minimize crack closure during the cyclic loading and obtained values of thresholds of stress intensity range can be considered to be very close to effective values. This was attained by means of an unique experimental devices and procedure of preparation of pre-crack of specimen with circumferential notch, which was made of stainless austenitic steel. The obtained values were compared with theoretical models with the support of molecular dynamics and ab-anitio calculation. Based on observation was found out, that fatigue crack propagation is controlled by decohesion model in austenitic steel. The morphology of fracture surfaces was studied by means of optical chromatographie and 3D stereophotogrammetry, which allowed a comparison of created morphology under shear modes II and III. Morphology of fracture surface formed (static and cyclic loading) by pre-crack was also studied by means of selected roughness parameters. The mechanism of deflection (kink) of crack growth under mode II was defined.

This Ph.D. thesis deals with methodology for determination of residual fatigue lifetime of railway axles based on damage tolerance approach. This approach accepts an existence of potential defect, which could lead to fatigue failure of whole axle. The behavior of crack in railway axle is described by approaches of linear elastic fracture mechanics. There are plenty of factors, which more or less influence determined residual fatigue lifetime. The aim of this thesis is to quantify effects of these factors. The first part of Ph.D. thesis represents overview of studied problems relating to fatigue damage of railway axles. This part is focused on parameters, which influence fatigue crack growth in railway axle materials. The second part of thesis shows procedure for determination of residual fatigue lifetime, which was developed at the Institute of Physics of Materials of the Academy of Sciences of the Czech Republic. The main aim of this thesis was to improve current procedure for more precise estimation of residual fatigue lifetime. Significant part of this work is determination of significance of studied factors, which influencing calculated residual fatigue lifetime of railway axles (e.g. effect of threshold value, load spectrum, retardation effects, residual stress, axle geometry, material of axle etc.). The procedures described and results obtained can be also used for determination of residual fatigue lifetime of general mechanical component (not only railway axles). Therefore, results obtained in this Ph.D. thesis can be used e.g. for assessment of regular inspection intervals of cyclically loaded general mechanical parts.

Les résines époxy renforcées par des nodules thermoplastiques sont largement utilisées dans les matériaux composites à matrice organique de l’industrie aéronautique. Dans le cas particulier des stratifiés, les résines époxy déterminent la résistance du milieu inter-laminaire au délaminage. Ce processus de décohésion macroscopique apparaît classiquement lors d’un impact dynamique et compromet fortement l’intégrité des structures composites légères. Pourtant, les modèles cohésifs actuels ne sont pas prévus pour la simulation du délaminage sous sollicitation rapide. Notamment, l’influence présumée de la dynamique sur la rupture du milieu inter-laminaire riche en résine époxy n’est pas prise en compte. Par conséquent, cette recherche consiste à caractériser l’initiation et la propagation de fissure dans une résine époxy sous chargement dynamique. Dans cet objectif, des éprouvettes entaillées sont fabriquées à partir de la résine époxy Hexply RM21 spécifique du domaine aéronautique. Un protocole expérimental est développé et validé afin d’estimer les ténacités d’initiation et de propagation d’une éprouvette chargée en flexion 3-points par un vérin hydraulique. Les essais réalisés démontrent que la ténacité dynamique augmente fortement avec la vitesse de propagation de fissure, à l’inverse de la ténacité d’initiation qui diminue avec la vitesse d’impact. Ces évolutions sont cohérentes avec les micrographies réalisées post-mortem qui présentent une augmentation des mécanismes de rupture. Finalement, une loi cohésive est définie et identifiée pour prendre en compte l’influence de la dynamique sur l’initiation et la propagation d’une fissure dans la résine époxy étudiée. / Epoxy resins toughened with embedded thermoplastic particles are classically used in the aeronautical composites manufacturing. As matrix into laminated composites, epoxy resins determine the delamination resistance of the interlaminar field. The delamination process is a large decohesion that often occurs during impact loadings and it critically compromises the integrity of the light weight composite structures. However, most of the cohesive zone models are not intended to simulate delamination under high rate loadings. In particular, these models do not consider the expected effects of loading rate on fracture behaviour of the resin-rich interlaminar field. Therefore, this research aims at characterizing crack initiation and propagation in epoxy resins under dynamic loadings. For that purpose, we manufacture notched specimens made of pure Hexply RM21 epoxy resin which is mainly used in aeronautics. An experimental procedure is developed and validated in order to estimate initiation and propagation fracture toughness on a 3-points bending test associated to a servo-hydraulic testing machine. Experimental results demonstrate that the dynamic fracture toughness highly increases with the crack propagation speed while the initiation toughness decreases with the impact velocity. These evolutions are consistent with post-mortem micrographic observations that show intensifying fracture mechanisms. Finally, a cohesive law is defined and identified to take into account the dynamic dependency of crack initiation and propagation in the considered epoxy resin.