Fatigue life prediction from inclusion initiated cracks

Abstract

Fatigue cracks have often been observed to nucleate at inclusions or constituent particles in thin sheet 2024-T3 aluminum. Large amounts of scatter are commonly observed in the experimental fatigue lives and crack nucleation site data. This research attempts to address some of the stochastic issues associated with the fatigue process. An experimental program was conducted to investigate the influence of inclusions on fatigue crack formation. Inclusions were found to be the dominant crack nucleation site and the inclusion sizes that nucleated fatigue cracks were from the large particle tail of the overall inclusion size distribution. A probabilistic model has been developed to predict the observed variability in the experimental fatigue lives from the measured distribution of inclusion sizes. In order to model crack growth from a microstructurally sized defect (such as an inclusion), a total fatigue life prediction model accounting for the small crack effect must be utilized. The probabilistic model uses a Monte Carlo simulation and Newman's FASTRAN II model to predict fatigue life. The fatigue life distributions, that were generated with the probabilistic model, predicted the critical, lower experimental fatigue lives and the overall variability in fatigue lives under several loading conditions.