Major solutions needed in frlcture analysis are (d) simple and accurate matedal characterization and (ô) easy transferring of malerial data lo cracked structures. ln !h€ proposed methodology. constitutive relationships. including cavity growth and coalescence, are used. Material characterization is based on the simple notched tension lest. Several structural steels have been characterized. especially 4508 steel. and the direct transferring of material data hâs been demonstrated wilh tests on circumf€rentially cracked tension specimens.In addition. the extrapolation of material data io different inclusion conl€nls and temperatures was attempted, with favorabl€ rcsults for the first factor through a specific parameter of the model. Th€ temperature dcpendence of 4508 steel ductiliiy is related to an inverle strain rale effect on the flow curv€: the modeling of this effect gives encoùraging results, but it must be refined to producc an effeclive prediction. KEY WORDSI ductile fracture. damâge mechanics. local approach to fracture, crack iniriation, slable crack growth, 4508 steel. inclusion contenls, notched tension test. fracture mechanics. nonlinear fracture mcchanics Some of the major problems engineers have to cope with in fracture analysis are the following:(a) material characterization, that is. the generation of adequate data from specimen testing, and(ô) the transferring of fracture mechanics data (o the struclural analysis of components.As a mattcr of fact, the generation of matedal data can be money and time consuming: for example, the determination ofJ-Ad resistance curves is still a toilsome task, though some progress was gained with partial unloading cornpliance methods. Frequently, existing material data do not correspond to the specific application (in reference to temperature, strain rate, irradiation. aging, and so forth), and hazardous extrapolations are necessary.

No abstract

No abstract

Abstract. On the framework of the French research project SINAPS@, the main goal of this work is to assess the impact of different soil-structure interaction modeling strategies on the Unit 7 Kashiwasaki-Kariwa Nuclear Power Plant reactor building response during the Niigata Chuetsu-Oki earthquake mainshock. his actual nuclear power plant case-study was already investigated during the KARISMA (KAshiwazaki-Kariwa Research Initiative for Seismic Margin Assessment) international benchmark exercise on 2009 organized by the IAEA (International Atomic Energy Agency) and OECD/NEA, which concluded on the difficulty to correctly predict the Unit 7 Reactor Building response during the mainshock by considering available standard soil-structure interaction modeling. The modeling strategies investigated in this work include soil non-linearity, by considering a cyclic critical state elastic-plastic model, and structuresoil-structure interaction influence by directly considering a simplified model of the adjacent Turbine Building.All numerical simulations are performed using code aster Open source software. The comparison of results with recorded signals within the Unit 7 Reactor Building allow to emphasize some conclusions about the role of modeling assumptions. This paper gathers the research work conducted by EDF R&D in the SINAPS@ task 4.3 during the years

No abstract

No abstract

No abstract

Une comparaison entre des méthodes temporelles et des méthodes temps/fréquence est proposée pour l'étude du décollement des fondations en cas de séisme. L'approche temps/ fréquence est fondée sur l'utilisation d'un modèle détaillé du sol sous la fondation faisant appel à une méthode d'éléments finis en fréquence. Le décollement est traité de manière itérative sur l'ensemble de la plage de temps en utilisant un opérateur auxiliaire construit à partir de statistiques sur la réponse, supposée aléatoire. Les méthodes temporelles sont fondées sur des techniques de pénalisation des conditions de contact dans le Code_Aster, le sol ayant été remplacé par un tapis de ressorts équivalents calibré à partir du modèle par éléments de frontière. Ces deux approches sont comparées sur un cas industriel. ABSTRACT. A comparison of time-domain and time-frequency domains algorithms is proposedto study the uplift of building foundations during earthquakes. The time-frequency approach is based on a detailled model of the soil using a BEM in the frequency domain and using the Green's functions of a layered elastic half-space. In this case, uplift is accounted for using an iterative quasi-Newton scheme operating on the entire time range. At each iteration step, the quasi-tangent operator is built by mean of a stochastic linearisation based on statistics of the current solution and is inverted in the frequency domain. The TD approach is based on classical penalty techniques in the FEM representing the soil by an equivalent layer of springs that are calibrated using the results of the BE model. These two approaches are compared on a real case consisting of a reactor-building resting on a layered soil.