Volume 3, Issue 10 (November 2006)
A Comparison of Predictions From Probabilistic Crack Growth Models Inferred From Virkler's Data
The significant variability observed in fatigue crack growth experiments requires application of probabilistic modeling techniques to fatigue analyses. In this work, philosophical and practical concerns for probabilistic modeling are discussed. Several probabilistic fatigue models are derived from the Virkler data and then used to make predictions under a different loading condition. This comparison of predictions demonstrates their sensitivity to the choice of model. In this effort a method for inference of random process parameters and corresponding confidence limits from highly statistically dependent observations is developed.