Modelisation des contraintes residuelles thermiques et etude de leur effet sur la vie en fatigue de l'acier inoxydable austenitique 304L.

摘要

The general interest of this study was to examine the fatigue behaviour of components subjected to complex manufacturing processes such as welding and hammer peening. To better understand the influence of the residual stresses induced by these processes on the fatigue life, the effect of this parameter had to be isolated from other influences.; The first objective of this study was to develop an experimental method to induce residual stresses in a steel sample without producing microstructural changes. The second objective was to study the influence of these residual stresses on the fatigue strength of this steel.; An experimental set-up, based on high frequency (360 kHz) induction heating, was designed to introduce the proper amount of residual stresses on cylindrical specimens. The 304L austenitic steel was chosen because it does not undergo microstructural changes during heating and cooling in the temperature range of 20-1000°C. Also, Argon gas was used to prevent oxidation at the surface of the specimen during heating.; Surface temperature measurements allowed to set the parameters of the induction heating. The use of a coolant at the centre of the specimen was necessary to obtain a high tensile residual stress field beneath the surface of the samples. In order to validate the efficiency of the heating system, residual stresses were measured using the X ray diffraction method. The axial component of the residual stress induced with a heating time of 1,4 second at full power was 250 MPa in tension.; The multiphysical simulation of the induction heating process allowed to calculate the residual stress field induced all over the specimen cross-section. This simulation required the simultaneous resolution of electromagnetic, thermal, and elasto-plastic mechanical equation sets, which was done by sequential coupling.; Calculation results confirmed that a tensile residual stress field was induced beneath the surface. The numerical model was validated by experimental X ray diffraction measurements done at a maximum depth of 250 mum. The stress distribution obtained from the simulation has been useful for the analysis of experimental fatigue life results.; A first fatigue test series at constant amplitude and at a stress ratio R = -1 was realized at room temperature in order to establish the 304L steel reference curve. Another test series was done with 1,4 second heat induction samples, subsequently cooled.; Comparing the fatigue curve obtained from the conditioned samples to the reference curve showed that the 304L fatigue life was improved by tensile residual stresses. The lifetime of conditioned samples was improved by a factor of 35 at a stress amplitude of 190 MPa, by a factor of 5,79 at 210 MPa and by a factor of 2,1 at 230 MPa; no noticeable improvement was observed when the stress level was above 230 MPa. This result was unexpected. Indeed, it is generally accepted that this type of stress is deleterious to fatigue. Three statistical methods were then used to test the validity of the results (ASTM E739 standard, Student distribution, and probability function). All three confirmed that fatigue life of heated specimens was longer than that of the reference samples.; The decrease of the influence of the residual stresses with increased applied stress amplitude corresponded to an increase in the residual stress relaxation during cycling. Some X ray measurements on pre-heated and cycled samples at a stress amplitude of 210 MPa showed residual stress redistribution and relaxation phenomena. The initial residual axial stress of 250 MPa relaxed to around 200 MPa during the first 50 cycles and further to -60 MPa after 50 000 cycles, which corresponds to the sample half-life. This mean compression stress explains the increase in fatigue strength of the samples subjected to the residual stress conditioning.; The Morrow model was used to compute fatigue life, based on the residual stresses measured at mid-life of heated samples. T