Application of probabilistic methods to fuel rod design evaluation

Full text: The safety concept of nuclear power plants fuel rods relies on a system of multiple barriers to preclude the release of fuel containing uranium and/or plutonium or highly radioactive solid and gaseous fission products generated during operation of the reactor. The gas-tight welded fuel rods (FRs) represent a first and important barrier of this concept. As a consequence, FR design must demonstrate that systematic failures are precluded by applying appropriate codes and methods. In particular, this means that critical physical quantities (e.g. temperatures, strains, internal pressure) must not exceed their design respectively defect limits. Traditional proofs in the field of nuclear engineering are based on a conservative-deterministic approach which makes use of an unfavorable combination of input parameters with respect to interesting quantities. However, there have been significant drivers during the last years which triggered the advancement of more realistic methodologies. For instance, power uprates of plants, increased enrichments and burnups, and the pertaining challenging experience feedback from PIE and pool measurements push forward to regimes which were not preconceived when the traditional concepts in fuel rod design evaluation were established. Thus, all stakeholders of the licensing process start to reflect the potential of the traditional approaches in the light of current and future demands, i.e. in particular to improve fuel utilization and reduce the amount of radioactive waste: - What is the degree of conservatism of the conservative-deterministic approach? Is it over-conservative, are sufficient margins left at more challenging demands? Or might there even be a lack of conservatism possible? - Is the selection of input parameters within the deterministic approach still justified? Are there - dependent on the situation - other combinations conceivable with more unfavorable results? Is it possible to map the possible nonlinear response of state-of-the-art codes? To answer these questions it was proposed to apply an advanced approach based on the combination of probabilistic methodologies, best-estimate codes, and a comprehensive, reliable database what allows for a controlled way to approach the technical limits by a realistic assessment ('best-estimate scenario') as well as by a conservative result with a well defined degree of conservatism in terms of an appropriate quantile. Thus, the new approach is able to fulfill all requirements of licensing scenarios and, thus, can be used as a direct licensing tool or as a benchmark for the traditional methods. (author)