The Deterministic Prediction of Damage in Light Water Nuclear Reactors

Abstract Localized corrosion phenomena, including pitting corrosion, stress corrosion cracking, and corrosion fatigue, are the principal causes of corrosion-induced damage in electric power generating facilities and typically result in more than 50% of the unscheduled outages. Prediction of damage, so that repairs and inspections can be made during scheduled outages, could have an enormous impact on the economics of electric power generation. To date, prediction of corrosion damage has been made on the basis of empirical/statistical methods that have proven to be insufficiently robust and accurate enough to form the basis for the desired inspection/repair protocol. In this paper, we describe a deterministic method for predicting localized corrosion damage. We have used the method to illustrate how pitting damage functions might be predicted for low pressure steam turbines downstream of the Wilson line, where a thin condensed liquid layer exists on the stainless steel blade surfaces. Our calculations show that the pitting corrosion damage function is very sensitive to the oxygen content of the steam, with increasing oxygen levels resulting in more extensive and deeper pitting for a given time of observation.