# Corrosion Fatigue Uncertainty

## Overview

Uncertainty in engineering analysis arises from three types of sources: (1) Physical uncertainty or inherent variability; this is generally quantified by a probability distribution estimated from observed data. (2) Statistical uncertainty; this refers to the uncertainty in the statistical distribution parameters of the random variables identified in the first source, due to the scarcity in the data. (3) Modeling uncertainty; this includes uncertainty in both probabilistic and mechanical models. In each case, the uncertainty exists in model accuracy as well as model selection. In the case of probabilistic models, the accuracy in distribution parameter estimates is already accounted for by statistical uncertainty. In addition, there are approximations in the computational procedures, and the extent of these approximations is uncertain. In the case of mechanical models, model accuracy relates both to mathematical idealization of behavior, as well as approximations in the numerical solution procedure.^{[1]}

## reference

- ↑ R. Zhang, S. Mahadevan, Model uncertainty and Bayesian updating in reliability-based inspection, Structural Safety, 22 (2000) 145-160.