The aim of this thesis is to propose a more general understanding of the influence of the thickness of the plate and of the microstructural and mechanical properties of the material on the resistance to ductile tearing in thin metallic plates. The objective is to attempt unifying different observations made in the literature together with the results of a new extensive experimental campaign. The final goal is to develop predictive simulation tools with a micromechanics-based foundation. In order to reach this objective, a detailed experimental campaign has been performed concerning the fracture behavior of the aluminium alloy 6082, complemented by experiments on a stainless steel A316L and on a set of 14 other materials. In a first modelling effort, we propose very simple closed-form models in order to separate the different contributions to the total work of fracture in thin plates: the work of necking and the work of damage and material separation. The respective contributions are compared and an unique explanation of the different behaviors observed experimentally is proposed. In a second modelling step, we develop a full 3D numerical tool based on cohesive elements for simulating crack propagation in thin ductile plates. Three different methods are proposed to calibrate the parameters of the model in order to reproduce the experimental data and to extrapolate the results to other material properties or geometric conditions. Finally, the parameters of the cohesive zone model are justified using micromechanics-based arguments. / Le but de cette thèse est de proposer un modèle général à base micromécanique permettant de comprendre l’influence de l’épaisseur de la tôle ainsi que de la microstructure et des propriétés mécaniques du matériau sur la résistance à la rupture ductile de plaques minces métalliques. L’objectif est d’essayer d’unifier les différentes observations de la littérature ainsi que les résultats d’une nouvelle campagne expérimentale afin d’aboutir au développement d’outils numériques prédictifs. Pour atteindre cet objectif, nous avons réalisé une campagne d’essais concernant le comportement à la rupture de différents matériaux. Cette campagne a été menée en profondeur sur l’alliage d’aluminium 6082 et de manière moins approfondie sur un acier inoxydable A316L ainsi que sur 14 autres matériaux. Dans un premier temps, nous présentons une série de modèles semi-analytiques simples dont le but est de séparer les différentes contributions au travail de rupture total dans les tôles minces : le travail de striction et le travail d’endommagement du matériau. Ces deux contributions sont ensuite comparées et nous proposons une explication qui reprend les différents comportements observés expérimentalement. Dans un deuxième temps, nous développons un outil numérique 3D complet destiné à simuler la propagation de fissures dans les tôles minces ductiles et qui utilise des éléments cohésifs. Trois méthodes différentes sont proposées pour calibrer les paramètres du modèle de manière à reproduire les données expérimentales et à permettre l’extrapolation des résultats à d’autres matériaux ou d’autres épaisseurs de tôles. Finalement, les paramètres du modèle de zone cohésive sont justifiés grâce à des arguments à fondement micromécanique.

No description available.

Nuclear Reactor Pressure Vessels (RPV) are manufactured from medium strength low alloy ferritic steel, specifically selected for its high toughness and good weldability. The ability of the RPV material to resist crack growth is crucial given that it is one of the fundamental containment safety systems of nuclear power plants. For most of their lifetime, the RPV operates at sufficiently elevated temperatures to ensure the material is ductile. However, the development of ductile damage, in the form of voids, and the ability to predict ductile tearing in RPV materials using a mechanistically-based model remains difficult. The Gurson-Tvergaard-Needleman (GTN) model of ductile tearing provides one such tool for predicting ductile damage development in RPV materials. The difficulty in using the GTN model lies in the ability to calibrate the model parameters in a robust manner. The parameters are typically calibrated data, derived from fracture tests and relying on an iterative “trial and error” procedure of numerical simulations and comparison with test data until the model reproduces the experimental behaviour with sufficient accuracy. This research has addressed the development of a mechanistically-based approach to the calibration of the GTN model by developing a new understanding of the ductile fracture mechanism in RPV material through conventional metallography and 3D X-ray computed tomography to image the initiation, growth and coalescence of ductile voids. The metallographic and tomographic data were analysed in a quantitative manner to establish a direct link between the microstructural features and void evolution and the key parameters of the GTN model. This approach has established a more robust mechanistically based method for the calibration of the GTN model that will enhance the conventional iterative calibration procedure. The calibrated model was applied to predict ductile tearing behaviour in compact-tension and notched-tensile specimens. The results showed good agreement with test data and also reproduced the morphology and branching of crack extension observed in practise. Whilst these observations were due, in part, to the numerical solving procedure, they enabled new insights to be gained regarding the development of non-uniform void volume fraction distributions in tested specimensThe results from this research will strengthen the guidance provided to structural integrity engineers in industry regarding the calibration and application of ductile damage mechanics models such as the GTN model for predicting ductile initiation and growth in RPV materials.