Considering the hypothetical core melt down scenario for a light water reactor (LWR) a possible failure mode of the reactor pressure vessel (RPV) and its failure time has to be investigated for a determination of the loadings on the containment. Numerous experiments have been performed accompanied with material properties evaluation, theoretical, and numerical work /REM 1993/, /THF 1997/, /CHU 1999/. For pre- and post-test calculations of Lower Head Failure experiments like OLHF or FOREVER it is necessary to model creep and plasticity processes. Therefore a Fi-nite Element Model is developed at the FZR using a numerical approach which avoids the use of a single creep law employing constants derived from the data for a limited stress and temperature range. Instead of this a numerical creep data base (CDB) is developed where the creep strain rate is evaluated in dependence on the current total strain, temperature and equivalent stress. A main task for this approach is the generation and validation of the CDB. Additionally the implementation of all relevant temperature dependent material properties has been performed. For an evaluation of the failure times a damage model according to an approach of Lemaitre is applied. The validation of the numerical model is performed by the simulation of and com-parison with experiments. This is done in 3 levels: starting with the simulation of sin-gle uniaxial creep tests, which is considered as a 1D-problem. In the next level so called "tube-failure-experiments" are modeled: the RUPTHER-14 and the "MPA-Meppen"-experiment. These experiments are considered as 2D-problems. Finally the numerical model is applied to scaled 3D-experiments, where the lower head of a PWR is represented in its hemispherical shape, like in the FOREVER-experiments. This report deals with the 1D- and 2D-simulations. An interesting question to be solved in this frame is the comparability of the French 16MND5 and the German 20MnMoNi55 RPV-steels, which are chemically nearly identical. Since these 2 steels show a similar behavior, it should be allowed on a lim-ited scale to transfer experimental and numerical data from one to the other.

Considering the hypothetical core melt down scenario for a light water reactor (LWR) a possible failure mode of the reactor pressure vessel (RPV) and its failure time has to be investigated for a determination of the loadings on the containment. Numerous experiments have been performed accompanied with material properties evaluation, theoretical, and numerical work /REM 1993/, /THF 1997/, /CHU 1999/. For pre- and post-test calculations of Lower Head Failure experiments like OLHF or FOREVER it is necessary to model creep and plasticity processes. Therefore a Fi-nite Element Model is developed at the FZR using a numerical approach which avoids the use of a single creep law employing constants derived from the data for a limited stress and temperature range. Instead of this a numerical creep data base (CDB) is developed where the creep strain rate is evaluated in dependence on the current total strain, temperature and equivalent stress. A main task for this approach is the generation and validation of the CDB. Additionally the implementation of all relevant temperature dependent material properties has been performed. For an evaluation of the failure times a damage model according to an approach of Lemaitre is applied. The validation of the numerical model is performed by the simulation of and com-parison with experiments. This is done in 3 levels: starting with the simulation of sin-gle uniaxial creep tests, which is considered as a 1D-problem. In the next level so called "tube-failure-experiments" are modeled: the RUPTHER-14 and the "MPA-Meppen"-experiment. These experiments are considered as 2D-problems. Finally the numerical model is applied to scaled 3D-experiments, where the lower head of a PWR is represented in its hemispherical shape, like in the FOREVER-experiments. This report deals with the 1D- and 2D-simulations. An interesting question to be solved in this frame is the comparability of the French 16MND5 and the German 20MnMoNi55 RPV-steels, which are chemically nearly identical. Since these 2 steels show a similar behavior, it should be allowed on a lim-ited scale to transfer experimental and numerical data from one to the other.

Cette thèse vise à étudier le comportement mécanique de l’acier de cuve 16MND5 (ou A508cl3 pour la norme anglaise) à l’échelle de la microstructure en croisant des approches expérimentale et numérique. Plusieurs contributions au développement de l’essai de traction in-situ à l’intérieur de MEB ont été apportées. En premier, les biais de mesure de différentes modalités (BSE, EBSD et SE) d’acquisition d’images sous MEB ont été caractérisés et corrigés. Les images MEB de différentes modalités ont été corrélées de façon précise afin de décrire la topographie de l’éprouvette. Les images d’orientation cristallographique (EBSD) ont été corrélées afin de révéler la rotation cristalline et les champs de déplacement de surface au long de la traction. La déformation élastique de l’éprouvette a été mesurée par corrélation intégrée des images de diffraction électronique à haute-résolution. Les microstructures fines de l’éprouvette à trois dimensions après déformation ont été mesurées par FIB-EBSD. L’essai a également été simulé par calcul de plasticité cristalline sur un maillage 3D, basé sur les microstructures mesurées dans la configuration déformée. Un algorithme a été proposé pour estimer la configuration initiale de l’éprouvette et identifier les paramètres de loi de plasticité en procédant par itérations. Un cas test synthétique 2D a été employé pour valider la faisabilité de l’algorithme. Deux lois de plasticité cristalline ont été testées sur le maillage 3D: dynamique des dislocations des cristaux cubiques centrés, et une version modifiée de la loi Méric-Cailletaud. Pour cette dernière loi, deux jeux de paramètres ont été identifiés pour les ferrites et bainites par recalage des éléments finis. / The PhD project is devoted to the study of the mechanical response of the reactor pressure vessel steel A508cl3 (or 16MND5 in French nomenclature) at the microscopic scale by experimental analyses and numerical simulations. Different aspects of in-situ tests inside an SEM chamber have been considered. First, the characterization and corrections of bias and uncertainties of different SEM imaging modalities (SE, BSE, and EBSD) have been performed. Precise registrations of SEM images in different modalities have been developed in order to give a comprehensive description of the sample surface topographies. Crystallographic orientation maps (from EBSD analyses) are registered to measure the crystal rotation and displacement fields along the tensile test. The elastic deformations of the surface are assessed by integrated correlation of high-resolution electron diffraction images. The 3D microstructure of the analyzed sample is revealed a posteriori by combining FIB milling andEBSD images.The experimental test is also simulated by crystal plasticity calculations on a 3D mesh created according to the 3D microstructure observed in the deformed configuration. An algorithm has been proposed to estimate its initial configuration and to identify the plastic parameters iteratively. A synthetic 2D model has been used to prove its feasibility. Two crystal plasticity laws have been validated on the 3D mesh, namely dislocation dynamics for body-centered cubic crystals and a modified version of Méric-Cailletaud model. In thepresent work finite element model updating was used to provide two sets of parameters (for ferrite and bainite) for the latter law.