THE EFFECT OF STRESS AMPLITUDE ON THE CAUSE OF FATIGUE CRACK INITIATION IN A SPRING STEEL

Hour-glass-shaped specimens of a hardened spring steel were fatigue tested with fully reversed loading at five different stress amplitudes. The corresponding Weibull diagrams for failure probability were determined and the failure causes were established by fractographic studies. The failure curves in the Weibull diagrams showed a double linear appearance at the lower stress amplitudes and a linear appearance for the higher stress amplitudes. The low-life part of the broken Weibull curve at low stress amplitude was associated with crack-initiating inclusions sitting close to the surface of the specimens. The high-life end of the curve was caused by inclusions sitting in the specimen interior. At high stress amplitudes the specimen failures were caused by inclusions situated at the specimen surface. The inclusion size distribution of the steel was determined with three different techniques based on optical microscopy, scanning electron microscopy and a chemical dissolution technique. The three techniques gave results in fair agreement with one another for large inclusion sizes. Long crack growth data were established with three-point bending. Residual stresses on the specimen surfaces were measured with X-ray techniques. A model for the failure probability of specimens is presented. It considers the growth of cracks from spherical pores. The pores are used to model inclusions. Crack growth is considered from a crack length of one Burgers vector to final failure of the specimen. Crack growth curves for short and long cracks are designed so that the growth can be calculated. The short cracks are assumed to grow at a constant rate until the threshold for long cracks is reached and growth proceeds according to the Paris law. The constant crack growth rate for short cracks is found to be a function of the applied stress amplitudes and different for inclusions sitting at the specimen surface and in the specimen interior. The failure probability is calculated for distributions of pores corresponding to the experimentally determined size distributions of inclusions. Reasonable agreement is obtained between experimental and theoretical failure probabilities.