VERIFICATION, VALIDATION, AND RELIABILITY OF PREDICTIONS
T.H. Pigford, P.L. Chambre · University of North Texas Digital Library (University of North Texas) · 1987
The objective of predicting long-term performance should be to make reliable determinations of whether the prediction falls within the criteria for acceptable performance. Establishing reliable predictions of long-term performance of a waste repository requires emphasis on valid theories to predict performance. The validation process must establish the validity of the theory, the parameters used in applying the theory, the arithmetic of calculations, and the interpretation of results; but validation of such performance predictions is not possible unless there are clear criteria for acceptable performance. Validation programs should emphasize identification of the substantive issues of prediction that need to be resolved. Examples relevant to waste package performance are predicting the life of waste containers and the time distribution of container failures, establishing the criteria for defining container failure, validating theories for time-dependent waste dissolution that depend on details of the repository environment, and determining the extent of congruent dissolution of radionuclides in the UO/sub 2/ matrix of spent fuel. Prediction and validation should go hand in hand and should be done and reviewed frequently, as essential tools for the programs to design and develop repositories. 29 refs.