Correlation of circumferentially notched and smooth specimen fatigue test results in two nickel base superalloys

Abstract

Prediction of the fatigue lives of components in engineered products is an important area of study; fatigue failure of a component may have product safety implications. Nearly all components contain stress concentration features and fatigue cracks will typically develop from these stress concentrations. Therefore, it is important that fatigue life prediction methods take into account the presence of stress concentrations. This article compares two methods of correlating fatigue test results between smooth and notched specimens in order to be better able to predict the fatigue lives of full-scale components. The first method uses the Neuber analysis approach to predict the strain range and mean strain at the root of the notch; the second uses elastic-plastic finite element analysis of the notched specimens. Two circumferentially notched specimen geometries are considered, with Kt values of 2.19 and 3.43. The analyses show that the Neuber method under-predicts the fatigue lives of the notched specimens, whereas a closer correlation is obtained using the finite element models. This is explained in terms of the different levels of constraint on yielding in a circumferentially notched specimen relative to the pure shear loading that is the basis of the Neuber analysis. Four methods of enhancing the finite element correlations are described. The first uses a method developed by Peterson to define a fatigue stress concentration factor. The second uses a mean stress and stress range approach developed by Smith, Watson and Topper. The third uses a critical distance approach developed by Pluvinage. The final method takes into account the volume of highly stressed material at the root of the notch using an approach first described by Waters and Norris to define a fatigue stress concentration factor. This approach has many parallels with the strain energy density approach.