Prediction of the Fatigue Life of Cast Steel Containing Shrinkage Porosity

Abstract

A simulation methodology for predicting the fatigue life of cast steel components with shrinkage porosity is developed and validated through comparison with previously performed measurements. A X-ray tomography technique is used to reconstruct the porosity distribution in 25 test specimens with average porosities ranging from 8 to 21 pct. The porosity field is imported into finite element analysis (FEA) software to determine the complex stress field resulting from the porosity. In the stress simulation, the elastic mechanical properties are made a function of the local porosity volume fraction. A multiaxial strain-life simulation is then performed to determine the fatigue life. An adaptive subgrid model is developed to reduce the dependence of the fatigue life predictions on the numerical mesh chosen and to account for the effects of porosity that is too small to be resolved in the simulations. The subgrid model employs a spatially variable fatigue notch factor that is dependent on the local pore radius relative to the finite element node spacing. A probabilistic pore size distribution model is used to estimate the radius of the largest pore as a function of the local pore volume fraction. It is found that, with the adaptive subgrid model and the addition of a uniform background microporosity field with a maximum pore radius of 100 μm, the measured and predicted fatigue lives for nearly all 25 test specimens fall within one decade. Because the fatigue lives of the specimens vary by more than four orders of magnitude for the same nominal stress amplitude and for similar average porosity fractions, the results demonstrate the importance of taking into account in the simulations the distribution of the porosity in the specimens.