Probability approach for prediction of corrosion and corrosion fatigue life

A probability approach for life prediction is developed and illustrated through a simplified model for the pitting corrosion and corrosion fatigue crack growth in aluminum alloys in aqueous environments. A method for estimation of the cumulative distribution function (CDF) for the lifetime is demonstrated by using an assumed CDF for each key random variable (RV). The basic aim of this approach is to make predictions for the lifetime, reliability, and durability beyond the range of typical data by integrating the CDFs of the individual RVs into a mechanistically based model. The contribution of each key RV is considered, and its significance is assessed. Thus, the usefulness of probability-based modeling is demonstrated. It is noted that physically realistic parameters were assumed for the illustrations. As such, the results from analysis of the model qualitatively agree quite well with experimental observations. However, these results should not be construed to represent behavior in actual systems. Because of these assumptions, confidence levels for the predictions are not addressed. 9 refs.