On développe dans ces travaux un outil numérique permettant de simuler le procédé Friction Stir Spot Welding. Le modèle est basé sur un couplage fluide/solide et permet une description correcte des parties fortement malaxée et solide de la structure. Une approche ALE est utilisée avec un mouvement arbitraire défini de façon que le maillage suive la matière dans la partie solide mais pas dans la partie pâteuse. Ceci permet la simulation de plusieurs tours de l'outil tout en suivant les bords des tôles soudées durant le procédé. Ce modèle numérique s'appuie sur l'élément fini mixte P1+/P1. Ce dernier a été développé avec une formulation température/vitesse/pression en mécanique des fluides (dans le cas d'un écoulement laminaire incompressible et transitoire) et en mécanique des solides dans le cadre des grandes transformations. La transition fluide/solide est effectuée au moyen d'un test explicite sur une température moyenne par élément, l'interface passant alors entre les éléments du maillage. Une procédure d'actualisation de la géométrie associée à l'approche ALE est effectuée à convergence. Ce couplage a été intégré au sein d'une nouvelle option du code SYSWELD. On présente ici une première simulation du procédé Friction Stir Spot Welding. D'autre part, deux montages spécifiques sont proposés pour l'investigation du procédé Friction Stir Spot Welding. Ces deux dispositifs intègrent une démarche de validation globale visant à calibrer la modélisation proposée du procédé. La stratégie expérimentale suivie est détaillée, et des premiers résultats obtenus sur un alliage d'aluminium basique sont présentés.

Ce travail présente le développement d'un outil numérique. Une formulation arbitrairement lagrangienne-eulérienne (ALE) est implémentée dans le logiciel 3D éléments finis FORGE3® pour simuler les différentes étapes du procédé de soudage par frottement malaxage (FSW). Une méthode découplée est utilisée : a) les champs de vitesses, pressions et températures du matériau sont calculés, b) la vitesse de maillage est calculée à partir de l'évolution des frontières du domaine et d'un critère de raffinement adaptatif procuré via une estimation d'erreur, c) les variables nodales et P0 sont transportées. Différentes techniques de calcul de la vitesse de maillage et de transport des variables sont étudiées, apportant des avantages significatifs par rapport à des approches plus standard. L'algorithme de contact a également été enrichi par une procédure de lissage d'outil. Ces améliorations ont été testées et appliquées sur des cas industriels.L'état stationnaire de soudage, tout comme les phases transitoires, sont simulés, montrant une bonne robustesse et une bonne précision de la formulation ALE développée. Dans un premier temps, la simulation de la phase de soudage stationnaire permet d'identifier, par comparaison avec des résultats expérimentaux, les paramètres de frottement. Dans un second temps, un des intérêts majeurs du modèle ALE étant la possibilité de simuler la formation de vide à l'interface outil/matière, la phase de plongée et des phases transitoires sont modélisées. Leurs simulations peuvent ainsi aider à mieux appréhender les mécanismes du phénomène complexe de déposition de matière qui doit avoir lieu à l'arrière du pion de façon à obtenir un joint sans défaut.

Le friction stir welding (FSW) est un procédé de soudage inventé en 1991 par l'institut de soudure anglais, le TWI. Celui-ci suscite un vif intérêt de la part de l'industrie aéronautique par sa capacité de souder les alliages d'aluminium de la série 2XXX et 7XXX, à durcissement structural, réputés pratiquement insoudables. Ce procédé étant relativement récent, il fait encore sujet de recherches actives. Ce travail a pour objectif de prévoir le profil de dureté d'un joint soudé par FSW d'un alliage d'aluminium, le 2024 T3. Cet alliage étant à durcissement structural, il est nécessaire de prévoir l'influence de la température sur l'évolution de la précipitation au cours du procédé pour en déduire sa limite d'élasticité. L'estimation du champ de température durant le régime stationnaire du procédé s'appuie sur des travaux internes au centre SMS. La prévision de la précipitation au cours du soudage est effectuée à l'aide de deux modèles. Le premier modèle, à base d'équivalence temps–températures, est une proposition d'extension aux alliages d'aluminium sous-revenu du modèle de Myhr & Grong (1991) établi dans le cas des alliages d'aluminium sur-revenu. Le deuxième modèle s'appuie sur une discrétisation de la distribution des rayons des précipités, suivant le schéma numérique de Kampmann et Wagner (1983), pour calculer ensuite son évolution. Bien que le premier modèle permette de prévoir l'évolution de la dureté au cours de recuits isothermes, les profils de dureté simulés ne sont pas en accord avec les profils expérimentaux. Seul le deuxième modèle permet une prévision raisonnable de la microstructure, en accord avec les mesures réalisées dans la thèse de Genevois (2004), et des profils de dureté proches des résultats expérimentaux. Finalement, une expression analytique en fonction des paramètres microstructuraux du flux de chaleur lors d'un essai de calorimétrie différentielle (DSC) a été établie. Celle-ci donne la possibilité de simuler un essai de DSC, et de vérifier ainsi la cohérence entre les grandeurs thermodynamiques et cinétiques introduites dans le deuxième modèle de précipitation.

In this thesis, the relationship among friction stir processing (FSP) parameters, microstructure evolution, texture development, and mechanical hehavior of AZ31B Mg alloy was investigated. First of all, in order to reveal the correlation among the deformation conditions, dynamic recrystallization (DRX) mechanisms, and microstructure evolution in the Mg alloy, hot compression tests at a wide range of Zener-Hollomon parameter (Z) values were conducted. Through optical microscopic examination, it was found out that above a critical Z value, twinning influences the DRX process resulting in a more effective grain refinement, which is manifested in a significant change in the slope of the Z-drec relationship, where drec is the recrystallized grain size. Moreover, EBSD examination revealed that the twinning also contributed to a distinct change in the recrystallization texture. Compression tests were performed along both through-thickness and in-rolling-plane directions of the plate to study the orientation dependency of twinning activities and its influence on the DRX process. X-ray line profile analysis (XLPA) provides further insights by highlighting the differences in the dislocation density/types, subgrain sizes, and twin densities during the DRX processes operating with or without the twinning. Secondly, the constitutive behaviour study was applied to the investigation of microstructure evolution during FSP. By varying the key FSP parameters systematically, i.e. rotation and travel rates of the tool, a series of FSP specimens were prepared with a wide range of thermo-mechanical inputs in terms of Z. The resulting tensile behavior in the stir zone (SZ) showed a dramatic change as a function of Z, caused by a systematic change in the texture within SZ measured by neutron diffraction. A three-dimensional transient model was developed to investigate the detailed deformation history including the temperature and strain rate profiles and material flow pattern during FSP of the Mg alloy. Such deformation history can be combined with the constitutive study from the compression tests in order to analyze the developments of micro-texture and DRX grains during FSP, which will, in turn, dominate the mechanical properties. Based on the studies above, new fundamental understandings were gained on the governing mechanisms for the deformation and recrystallization processes during FSP and the influence of thermo-mechanical input during FSP on ductility enhancement in the Mg alloy.

Friction stir welding (FSW) is a low heat input solid state welding technology. It is often used for fabrication of aluminium alloys in transportation applications including railway, shipbuilding, bridge structures and automotive components. In these applications the material is frequently subject to varying load conditions and fatigue failure is a critical issue. In most cases standard codes and fatigue guidelines for aluminium welded joints address only welded structures with conventional welding methods but not those with FSW procedure. In the scope of this thesis fatigue life assessment of friction stir welded components was performed using theoretical approaches along with finite element method (FEM). The further aim of this study was to generate a basis for standardization of fatigue assessment of friction stir welded joints. Friction stir welded hollow aluminium panels of alloy 6005A are investigated. The panels are used for train wall sides, train floors, deck and bridges. Each panel is made of several profiles that are joined with the friction stir welding method. Fatigue bending tests were performed for profiles in these panels. Fatigue cracks and failure appeared at notches in the profiles. With FEM simulations critical positions for crack initiation and failure were identified. The method of critical distance was used to analyse and estimate the fatigue life. It was shown that the failure location and fatigue limit could be predicted for both base metal and weld location. Choice of welding procedure (clamping condition) can significantly influence the fatigue life. It was shown that for some panels the critical distance method was not able to explain the failure in the weld. In this case fracture mechanics together with residual stress analysis were used successfully to predict the failure. Assuming homogeneous material properties throughout the weld and the base material, FEM analysis for T and overlap joints as well can provide a reasonable fatigue prediction. This suggests that the same assumption can be extended to complex components for failure analysis of the friction stir welded joints when using the critical distance method. Fatigue assessment of friction stir welded joints was also performed using standard codes Eurocode 9 and IIW. Fatigue curves of traditional fusion welded joints were used. The results are in reasonable agreement with experimental data and FEM predictions. / QC 20120330

In the friction stir welding (FSW) process, a high-speed rotating tool, which consists of the probe and the shoulder, are employed to plunge into the faying surfaces. By using the friction heating and the stirring action of the material, the solid-state welding is accomplished to joint two pieces of metal by material diffusion to form a densification structure in the weld. According to the experimental results, the mechanism of friction stir welding is as follows: the probe plunge into the sample and the shoulder is in contact with the sample to generate a large amount of friction heat, which causes the materials soft. When the probe moves forward, the soft materials in front of the probe are scratched. The scratched materials are subjected to the rotational and squeeze actions of the shoulder so that they are refilled into the welded surface behind the probe. For the dissimilar metals joint (6061-T6 aluminum and C1100 copper plates), results show that when C1100 copper is located at the advancing side, the measured feed force appears drastic changes and it is also seen that the components of the force for the friction-stir welding of dissimilar metals become more unstable than those for the same metals joint, so that the structure which has been observed by optical microscopy appears to be open with pores and defects so that the welded quality becomes poor. According to the three components of the measured force during FSW process, the friction between the probe and the work piece can be computed. By using the friction theory, the hardness and the yield strength of the materials in front of the probe can be calculated, and then the faying surface temperature is approximately predicted to be 565.5 oC.

In this study, a novel inserted type of friction welding tool was proposed, where the circular rod was embedded in its central axis using the material same as the workpiece, so that it could effectively promote the friction heat quickly and enhance the welding quality. The welding tool was made of the high-speed steel, the workpiece with its embedded material 6061-T6 aluminum alloy. A vertical milling machine equipped with dynamometer, which could measure the power during the friction stir welding, was employed as the experimental apparatus. During the welding process, the K-type thermocouple was used simultaneously in measuring the welding temperature at the interface of joint. The operating conditions of welding were as followings: the welding speed of 800 rpm, the tool inclination of 1¢X and the clamping force 2kN, the tool with 12mm in diameter and 0.21mm in depth under the downward force about 2 kN. The experiment was conducted into two stages. The first stage was a spot welding to investigate the effect of the ratio of the diameter of embedded material (d) to the diameter of welding tool (D) on the temperature of the interface of joint, the thickness of plastic flow, and the failure load of weld. Experimental results revealed that the interface temperature, the plastic flow thickness, and the failure load of weld are directly proportional to d/D. In comparison with the welding tool without insert (d/D = 0), the maximum interface temperature increased about 1.12 times at d/D = 0.83, the plastic flow thickness increased about 1.52 times, and the failure load of weld increased about 1.45 times. In the second stage, the feeding process was included to investigate the influence of the diameter and the thickness of embedded material on the interface temperature, the plastic flow thickness, and the failure load of weld. Experimental results revealed that the plastic flow thickness was less than 2 mm when the thickness of embedded material was less than 3 mm. However, when the thickness of embedded material was larger than 5 mm, the plastic flow thickness could achieve to 3 mm. Hence, the thickness of embedded material should be larger than 5mm. Moreover, the effect of the diameter of embedded material on the interface temperature and the plastic flow thickness using the feeding process was almost the same as the spot welding. However, in comparison with the welding tool without insert, the failure load of weld increased about two times.

A friction stir welding equipment with high rotation speed and constant load is successfully developed in this study to weld the multi-laminated silicon steel sheets widely used on regular transformers. This equipment consists of a spinning unit, a loading unit, and a feeding unit. A WC round rod with 3 mm diameter is used as welding tool. Under different operating conditions, such as the normal load(140~480 N), the spindle speed (12000~24000rpm), the feeding rate (0~1.58 mm/s), the welding characteristics and the welding mechanism of multi-laminated silicon steel sheets, and the welding feasibility of the transformer are investigated. Firstly, the contour map of welding depth in terms of spindle speed,normal load, and depth of point welding is established for dwell welding time 15 seconds. Secondly, based on this contour map, two experimental conditions of the long-pass welding tests are selected to investigate the effect of normal load (Fd), the spindle speed (Ns), and the feeding rate (f) on the failure load of weld under the shear. According to the experimental results, the empirical formula is obtained as Ff =40.6(Fd¡DNs)1.123(f)-0.791. In this formula,(Fd¡DNs)1.123(f)-0.791 is proportional to the frictional work per unit moving distance. With the larger frictional work, this represents the heat generation of the workpiece material is higher with more uniform friction stir, so that the bonding strength of the material increases and the failure load of weld is larger. According to the micrograph observations, the thermo-mechanically affected zone is significantly influenced by high heat action generated from the friction between the tool and the weld surface region, so that the plastic flow of the workpiece material occurs to cause the multi-laminated silicon steel sheets bonding together. Finally, the transformer is successfully welded under the experimental conditions of the long-pass welding tests with the smaller welding depth and the better failure load.

This study aims to design a novel inserted welding tool to friction stir spot weld SS400 low carbon steel sheets of 4mm thickness. In order to enhance the efficiency of frictional heat generation and to enhance the quality of the welding spot, the welding tool based on a cylindrical tungsten carbide and is inserted by a SS400 low carbon cylinder. The welding apparatus composed of a vertical milling machine and a welding platform that can keep the load between tool and workpiece constant. The plunge load is 8kN and there's no inclination angle on the tools. Welding temperature and the tool plunge depth are measured by thermelcouples and a displacement meter. ¡@¡@At the tool rotational speed of spindle of 900rpm and welding for 60 seconds, the temperature rising rate of the tools with 5 mm and 10 mm inserted material are 5.28 times and 6.31 times greater than the one without insert. While they are 1.36 and 1.42 times greater than at 1200rpm.At the tool rotational speed of spindle of 900rpm and welding for 300 seconds, themaximun welding temperature the tools with 5 mm and 10 mm inserted material can reach are 59¢J and 412¢J higher than the one without insert. While they can reach 35.6¢J and 197.6¢J greater than at 1200rpm. According to the tensile test, the shear failure loads of clad steel plates increase 11.3kN and 15.5kN by using tools with 5 mm and 10 mm inserted material at 900rpm for 60 seconds, and increase 7.6kN and 18.3kN by using tools with 5 mm and 10 mm inserted material at 1200rpm.

In order to simulate the histories of temperature distributions and plastic flow of the dwell phase during a friction stir welding process, the Newton-Raphson method is used to solve the simultaneous equations of energy and momentum in the cylindrical-coordinate system. Comparing the simulation with the results of experiment, results show that the contact condition between the tool and the workpiece is at pure sliding without plastic flow at the beginning of the dwell phase until the temperature rises to about 300¢XC at the depth of 1.5 mm. In this period, the heat generation comes from the sliding friction between two surfaces. After the plastic flow occurs, the heat generation rises rapidly, and then decreases to a saturated value so that the temperature rise also achieves a constant value. Thermal expansion of the workpiece will increase the plunge force, so that the heat generation and the temperature raise increase. At the steady state condition, with increasing sticking proportion, the heat generation and the temperature quickly achieve a saturated value. For the steady-state condition, results show that the speed of plastic flow and shear strain rate increase with increasing rotational speed. The control of the contact state variable can effectively describe the heat generation and the distribution of plastic flow in different contact conditions. Comparing the simulation with the results of experiment, the contact condition can be identified.