The growing development of soft mobility, especially electrically power assisted cycles ( EPAC), is leading to more complex frame geometry and greater loads. Therefore, the aim of this study, conducted in collaboration with Decathlon, is to enhance the understanding and simulation of the fatigue life of aluminum bike frames. The study specifically focuses on tube welding, a favored crack initiation area. Scientifically, the challenges are centered on the emergence of a property gradient, multi-axial loading, and the thin tubular nature of the concerned areas. Firstly, to simplify the problem and overcome the difficulties associated to the geometry of bikes frame, micro-tensile tests instrumented by stereoscopic image correlation were carried out on flats shaped specimens to identify the constitutive laws in the different zones of interest close to the welded area. Secondly, to get closer to the real shape of an aluminum cycle frame, combined tensile and torsion tests on butt-welded tubes were carried out. The instrumentation of these tests is based on stereoscopic image correlation, initially allowing access to in plan and out of plan displacement fields on the surface and subsequently, access to information related to the topology of the weld beads. The study of these kinematic fields has enabled the identification of areas of interest and the recalibration of material properties based on a representative digital model. The test load path was designed to be as representative as possible of the stresses experienced by bike frames during normatives tests. Regarding the representative numerical model, a tiling of solutions has been proposed to cover a wide range of properties. This database serves to feed an AI model which allows for the interpolation of solutions. Starting from this hypermodel, the data from the stereo correlation are re-injected to identify the real properties. All studies were performed in quasi-static conditions but will be extended to fatigue cases in future actions.