Welding processes imply rapid solidification, in presence of high thermal gradient and strong fluid flows in the weld pool. This work presents, in the case of welding, an analysis of the coupling between solidification mechanisms and fluid flows at the macroscale and the microscale. An experimental setup was designed in order to observe in situ a fully penetrated weld pool generated on a Cu30Ni plate with a GTAW torch. At the macroscale, observations are carried out by three cameras: two cameras recording in visible light the top and the back side of the whole weld pool, and one infrared camera catching the thermal field on solid part at the back side. At the microscale, a high-speed camera, mounted with a microscope lens, is used to observe dendritic growth and fluid flows at the back side trailing edge of the weld pool. The observations provide a lot of data allowing analyses and discussions on the correlations between solidification mechanisms and fluid flows, with respect to welding parameters. Then, the experimental results are compared to theoretical results obtained from analytical modelling, in order to discuss the possible limitations of models and try to better understand the coupling between physical phenomena.

The purpose of this article is to investigate the influence of growth kinetic models and nucleation models on the grain structure predicted by cellular automata (CA) during bead on Cu30Ni thin plate. Temperature field is obtained with the computation of two-dimensional finite element model (FE). Based on these temperatures, a CA model simulates the evolution of the envelope of grains along the solidification front. It is shown that for the same temperature field, the grains structure is modified if the growth model changes. The influence of the nucleation law parameters is also discussed from a modelling point of view. Experimental size of the weld pool are compared with thermal results and EBSD maps are compared with grain structure prediction obtained with the CA algorithm.

Due to its ability to account for shear effects, the Reissner–Mindlin plate model is often preferred in the engineering literature over the Kirchhoff–Love plate model. So, as already done for the Kirchhoff–Love plate model, it is challenging to proceed with a rigorous mathematical derivation of the Reissner–Mindlin plate model by studying the asymptotic behavior of a thin 3-dimensional elastic body when its thickness goes to zero. This was done in a simplified framework by using a second gradient or Cosserat continuum for the body, jointly with constitutive symmetry assumptions ; here – being aware of results on the bonding of thin plates – we prefer to consider a strongly heterogeneous classical linearly elastic body made of a periodic distribution of thin anisotropic plates abutted together. The mathematical study via variational convergence shows that it is not necessary to use a different continuum model nor to make constitutive symmetry hypothesis as starting points to deduce the Reissner–Mindlin plate model.