Contraintes résiduelles et leurs impacts sur l'amorçage de fissures en fatigue de flexion dans des engrenages aéronautiques durcis superficiellement par induction

Abstract

The optimization of gearing for aeronautical engines depends on the development of surface hardening processes to significantly improve in-service durability. Induction heating followed by quenching is a treatment increasingly used in this field to improve the fatigue resistance of critical components such as gears. In this context, this thesis studies the impact of the induction process parameters on residual stresses and the effects of those stresses on the bending fatigue of induction hardened gears. Two existing residual stress measurement techniques, X-ray diffraction and the contour method, were adapted for induction hardened components. The residual stress measurement by X-ray diffraction technique has been significantly improved in this thesis by the consideration of the variation of the X-ray elastic constant at different locations in the hardened layer and the development of a finite element based layer removal correction method that enabled more accurate in-depth residual stress measurements in all sort of geometries. The contour method is clearly a powerful tool for residual stress mapping at the core of parts but proved to be inaccurate for near-surface measurements in the case of thin hardened layers. These methods were used to show the effects of several parameters (initial hardness, preheating, final heating, tempering) with discs and aeronautical spur gears. The results indicate that two induction treatments can sometimes produce two different residual stress distributions (amplitude, severity of gradient in the transition zone, etc.) for a similar hardened depth. The bending fatigue of those gears was studied experimentally with a single tooth bending test rig and numerically with the proposition of a fatigue model for the calculation of the bending endurance limit. The calibration of the model was based on fatigue tests results on traction and torsion specimens. Bending fatigue testing results on gears confirmed the overall accuracy of the proposed model predictions. The Crossland criterion without the gradient effect gave better prédictions when compared with the experimental results for that particular case. This 3D multiaxial fatigue prediction model represents a significant improvement over previous approaches by the simultaneous consideration of the so-called gradient effect, residual stresses, surface roughness and the variation of properties caused by the surface hardening. The induction treatments used in this study improved the bending fatigue resistance of spur gears by 45 to 71 % depending on the case.

Date	28 Jan 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Philippe Bocher (Supervisor) & Florent Bridier (Co-supervisor)