Study and Development of Dimensioning Methods for Fastening Elements Using Finite Element Method in an Industrial Context Integration internship

Abstract

This thesis, carried out in collaboration with GDTech, focuses on the ana- lytical and numerical evaluation of bolted and screwed joints, with the aim of assessing the accuracy and limitations of simplified models compared to finite element simulations. Bolted and screwed joints are widely used in mechanical and aeronautical engineering, where their reliability is essential to ensure the overall performance and safety of struc- tures. Despite their apparent simplicity, their behaviour is complex, as it depends on numerous phenomena such as preload, load transfer, bending effects, thread contact and fatigue. An analytical model was developed in Python to calculate the preload, the stiffness dis- tribution, the von Mises stresses, as well as the different potential modes of failure of threaded assemblies. Several approaches from the literature were integrated, allowing the analysis to be adapted to different cases. In addition, finite element simulations were car- ried out using Siemens NX, including the modelling of contacts and boundary conditions, the introduction of preload and the application of external loads up to failure. The comparison between the two approaches shows a good overall consistency, particularly in the prediction of stresses induced by preload and in the general trend of stress evolution under external loading. However, some differences appear. At low loads, the analytical model predicts higher stresses due to the explicit consideration of the tightening torque. At higher loads, NX highlights local stress concentrations near geometric transition zones, potentially overestimating stresses since the fillet radius is not modeled. The analytical model was then applied to industrial cases studied at GDTech, confirming its relevance for stress analysis and for the assessment of thread stripping. Overall, this work shows that the analytical model developed provides reliable estimates and allows the detection of potential failures. Nevertheless, it is important to remain criti- cal of the simplifying assumptions that affect its accuracy and to complement the analysis with numerical simulations.