Uncertainty and Conservatism in Safety Evaluations Based on a BEPU Approach

Atomic Energy Society of Japan has published “Standard Method for Safety Evaluation using Best Estimate Code Based on Uncertainty and Scaling Analyses with Statistical Approach” to be applied to accidents and AOOs in the safety evaluation of LWRs. In this method, hereafter named as the AESJ-SSE (Statistical Safety Evaluation) method, identification and quantification of uncertainties will be performed and then a combination of the best estimate code and the evaluation of uncertainty propagation will be performed. Uncertainties are categorized into bias and variability. In general, bias is related to our state-of-knowledge on uncertainty objects (modeling, scaling, input data, etc.) while variability reflects stochastic features involved in these objects. Considering many kinds of uncertainties in thermalhydraulics models and experimental databases show variabilities that will be strongly influenced by our state of knowledge, it seems reasonable that these variabilities are also related to state-of-knowledge. The design basis events (DBEs) that are employed for licensing analyses form a main part of the given or prior conservatism. The regulatory acceptance criterion is also regarded as the prior conservatism. In addition to these prior conservatisms, a certain amount of the posterior conservatism is added with maintaining intimate relationships with state-of-knowledge. In the AESJ-SSE method, this posterior conservatism can be incorporated into the safety evaluation in a combination of the following three ways, (1) broadening ranges of variability relevant to uncertainty objects, (2) employing more disadvantageous biases relevant to uncertainty objects and (3) adding an extra bias to the safety evaluation results. Knowing implemented quantitative bases of uncertainties and conservatism, the AESJ-SSE method provides a useful ground for rational decision-making. In order to seek for “the best estimation” as well as reasonably setting the analytical margin, a degree of the posterior conservatism shall be lowered by constant efforts of enhancing our state-of-knowledge, i.e. The 13 International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH-13) N13P1341 Kanazawa City, Ishikawa Prefecture, Japan, September 27-October 2, 2009. 2 / 17 reducing uncertainties in inputs, improvement in modeling, enhancement of experimental database, etc. Pursuing this direction will lead to higher accountability of the analytical margin.