A probabilistic model to consider scale and gradient effects in the prediction of the fatigue life of Inconel 718 for turbine disk application

Abstract

The aim of this paper is to implement and compare different fatigue post-processing approaches for fatigue life assessment of complex parts. Inconel 718 is taken as an example, as it can exhibit several factors influencing fatigue life, such as mean stress, stress gradient and scale effects. Tests on different specimen geometries to exacerbate these effects were carried out at 550°C. The range of service operating life is between 103 and 106 cycles. A modelling chain was then set up. A structural calculation was performed using an elasto-visco-plastic behavior law to obtain the mechanical fields at cycle stabilize. These values were finally exploited by applying a post-processing treatment approach to predict the fatigue life of the structure. Two main types of post-processing approach were investigated: standard and probabilistic. The way in which the different factors influencing fatigue life are considered, depending on the approach used, was discussed. Finally, the probabilistic volume approach yields better results, thanks to its ability to consider mean stress, stress gradient and scale effects in the proposed formulation.