A Brief Review of Determinism in the Prediction of Localized Corrosion Damage

The great majority of models that have been developed to predict localized corrosion damage are empirical in nature. As such, they lack the ability to effectively predict damage outside of their immediate realm of calibration and generally can only “predict” what is already known. Furthermore, empirical models generally cannot predict new phenomena or relationships. It is shown that it is possible to describe propagation of corrosion damage deterministically,i.e.to describe how the systems evolves from the present state to the future state on the basis of natural laws [conservation of charge, mass-energy, and mass-charge equivalence (Faraday's law),etc.], subject to constraint by the natural laws. It is clear that deterministic models have much more predictive power than do empirical models. However, in the general case, it is possible to describe the development of localized corrosion damage in terms of the propagation of an ensemble of corrosion events, rather than as individual cavities. Accordingly, the prediction can be made in terms of statistical terms, for example, probability that the deepest pit will exceed the critical dimension that defines failure (e.g., thickness of a pipe wall). In doing so, the statistical parameters (mean depth of the deepest pit with its standard deviation,etc.) can be calculated deterministically. In order to perform such calculations, we must possess deterministic models for every stage of cavity development (pit nucleation, propagation and repassivation, transition pit into crack, crack propagation, and so forth). Some of these models are outlined in the current review. The deterministic theory outlined here has been applied for the predation of localized corrosion damage in important, practical systems, including pitting in oil field components, in low pressure steam turbine blades and discs, and in condensing heat exchangers, to name but a few of the current and past applications, some of which are reviewed in this paper.