Optimisation des assemblages par interference soumis a des sollicitations de fatigue

This thesis presents a methodology for designing interference fit assemblies subjected to fatigue load. Due to the assembly, a fretting-fatigue phenomenon was found responsible of reduction of the fatigue life inside the contact. However, there are geometric configurations that contribute to reduce its impact on the life of the assembly. Experimental tests have shown that the addition of a fillet, circular groove or hub overhang improves the fatigue life of interference fit assemblies. It was also determinate that in contact, failure is caused by fretting fatigue while it is caused by normal fatigue in the fillet or groove. As a result of the competing maximum equivalent stresses in contact and in the fillet, a shift in the location of fatigue failure has been observed in both experimental tests and finite elements simulation.;In order to address this problem, a Sequential Approximate Multi-Objective Optimization (SAMOO) algorithm was developed in this work. This method consists of optimizing an interpolation function from design of experiments. Entire algorithm was programmed from modules Hypermath, HyperStudy Sampling of Altair HyperWorks V11 suite. The kriging method was used to interpolate the Optimal Latin Hypercube design of experiments. The multi-objective objective genetic algorithm (MOGA) from Hyperstudy is used to generate the Pareto Front of the problem.;Although it was developed to address the problem of optimizing the interference fit joint, the algorithm can effectively solve structural problems. Thus, it was applied to a well known academic problem of a 10-bar truss. Its computation time is negligible compared to finite element model and therefore allows more flexibility to validate the algorithm. The results of this application demonstrate that the SAMMO algorithm can solve efficiently structural problem with fewer calculations. Moreover, its performances have been compared to an existing algorithm.;An application of the SAMOO on a geometric configuration of interference fits assembly studied in the litterature has been presented as a part of a scientific paper submitted to Structural and Multidisciplinary optimization. The optimization problem consisted in finding a tradeoff between mass and competing safety factors in two zones subjected to fatigue failures. The application of SAMOO algorithm on the interference fit problem produces good results. An ANOVA has generated a sensitivity study applied on the design of experiments. Moreover, optimizing the interference fit assembly with the multi-objective approach adds valuable information concerning the design specific behaviour around the optimum zone. Finally, it provides further arguments to help choosing an optimal design based on different criteria.;A methodology for fatigue strength calculation of interference fit assemblies subjected to fatigue load is presented. A comparison between calculations by finite element analysis and experimental tests has show the existence of fatigue strength reduction factor (kfret). While it's possible to calculate the fatigue strength with precision, there is actually no method to directly obtain the parameters values according to design criteria. Plastic deformation and contact element needs non-linear calculations of finite element model. Therefore, obtaining the optimal solution with a classic procedure may be time consuming, hence inefficient.