An assessment was performed of approaches to life prediction for high-level nuclear waste containers. It was concluded that the commonly used remaining life assessment techniques are inadequate for the prediction of performance of the waste canisters primarily because of the reliance on empirical damage accumulation laws. The best hope for prediction of performance of the repository relies on utilizing mechanistic models based on first principles. It was further concluded that prediction of performance of the waste containers should rely primarily on propagation phenomenon, because of the uncertainties in proving that initiation will not occur.
The above philosophy imposes significant restrictions on the selection of container materials, and selection of the repository site. The materials selection for the waste canister should focus on corrosion allowance materials because the general corrosion rate can be bounded and the rates of pitting corrosion are expected to be much lower than for corrosion resistant materials. Corrosion resistant materials should be avoided because the probability of pit initiation is high for the range of possible environments and the rates of pitting corrosion are expected to be high, as a result of hydrolysis reactions. In order to reduce possible rates of localized corrosion, an anoxic repository should be selected. Because of the uncertainties in the predictive capability of the mechanistic models, corrosion monitoring, and a multiple barrier design for the waste package should be used.