Modelling Procedures for Predicting the Lifetimes of Nuclear Waste Containers

Abstract

Approaches to modelling the corrosion of nuclear waste containers are reviewed. The required containment of many thousands of years makes this a daunting task. The process has been simplified for a disposal vault in which redox conditions evolve from initially oxidizing to eventually non-oxidizing. The corrosion behaviour can be divided into two periods: an early hot, oxidizing period when localized corrosion damage is to be expected; and a later cool, non-oxidizing period when localized processes would be stifled, or repassivated, and general corrosion will predominate. At present, deterministic models to predict localized corrosion damage during the early period are unavailable or, at best, preliminary. Generally, the approach to predicting localized penetration of the container has been stochastic in nature and extreme value statistical analyses have been used to predict the expected penetration of carbon steel or copper containers by pitting. Experiments to determine the rate of crevice propagation in titanium are discussed and a model developed to predict failure of titanium waste containers by either crevice corrosion or hydrogen-induced cracking described. General corrosion occurring in the second, less oxidizing, period is more amenable to modelling by deterministic methods. Models based on electrochemical descriptions of the interfacial kinetics are described for carbon steel and copper containers, two materials expected to corrode actively under waste vault conditions. To date, no adequate model exists to predict the slow general corrosion of passivated materials.