PERFORMANCE ASSESSMENTS OF GEOLOGIC REPOSITORIES FOR HIGH-LEVEL NUCLEAR WASTE: ARE THEY NECESSARY OR SUFFICIENT?

The global inventory of used or spent nuclear fuel (SNF) is approximately 150,000 metric tonnes. Geologic disposal is the generally accepted solution for the disposition of SNF, without reprocessing, and the high-level nuclear waste that result from reprocessing. The challenge, however, has been to develop methods for evaluating, with a “reasonable expectation” of success, the actual performance of a geologic repository over periods of tens to hundreds of thousands of years. Probabilistic risk analysis (PRA), originally developed for the safety analysis of nuclear reactors, has now been applied to the evaluation of the long-term performance of geologic repositories. A typical PRA for a geologic repository will have thousands of input parameters and hundreds of linked models that are meant to describe all of the physical, chemical and biological processes that affect repository performance. These models are necessarily “abstracted” in order to reduce the computational complexity of the total system analysis. Although PRA is now used in a wide variety of fields, e.g., reactor safety and health care, the application to geologic systems poses unique challenges because of the long periods of projected performance and the complexity of the system; most importantly, there is limited, on-going operating experience that can be used to refine and improve the analysis. Despite these limitations, PRAs are used to establish research priorities, make repository design choices and evaluate compliance with regulations. This presentation will discuss the different methodologies embedded in a typical PRA, the sources and propogation of uncertainty, and the limitations of modeling highly-coupled systems. The essential question is whether a PRA provides a sufficient basis for a safety assessment. If the answer is “No”, what are the alternatives?