Corrosion Damage Function—Interface Between Corrosion Science and Engineering

The use of empirical and deterministic models for predicting damage functions for localized corrosion are reviewed and compared. In general, empirical models are successful because the distribution functions employed display great mathematical flexibility. However, they require a very significant database to achieve reliability. In a sense, they are only really successful when the answer is known in advance. Deterministic models are much more powerful and efficient, in that they yield analytical distribution functions and analytical relationships between the model parameters and the system variables. Accordingly, they require a much smaller calibrating database, but they rely heavily on the validity of the model. Nevertheless, as has occurred in other fields of science and engineering (e.g, in physics and mechanics), the trend in predictive corrosion science and engineering must be toward determinism because of the need for more reliable and robust algorithms. In this paper, the authors outline a deterministic approach for calculating damage functions for pitting corrosion. The damage functions that are expressed as plots of the number of pits vs pit depth at any given time of observation are calculated using deterministic models for pit nucleation and pit growth. To the authors’ knowledge, this is the first attempt to devise a full deterministic algorithm for predicting the extent of damage to a surface caused by localized attack.