Uncertainty quantification for Passive Safety System and treatment of Model Uncertainty

초록

As interest in improving the safety of nuclear power plants is increased, there are a lot of research about passive safety system and therefore it is necessary to assess the reliability of passive safety systems. However, the performance of passive safety systems is highly dependent on environmental conditions, making it crucial to assess their reliability. To do so, best estimate simulation models are commonly used to analyze the performance of the system under various input variabilities that represent different environmental conditions. To perform an appropriate reliability analysis, a sufficient number of simulations should be conducted to reflect all expected environmental conditions. However, best-estimate models can be complex and simulation costs can be high, limiting the number of simulations. Therefore, data-driven surrogate models are developed as approximations of best-estimate models to perform reliability analysis. These surrogate models are typically parameterized based on point values derived from pre-performed simulation data and uncertainty analysis is performed only for the variabilities of environmental conditions. However, there is another source of uncertainty that arises from the surrogate model because the model parameters are determined using a limited number of simulations. This type of uncertainty in the model parameters also affects the reliability of a passive safety system and it is necessary to analyze its impact on reliability. In this paper, a surrogate model is developed using a Bayesian approach to reflect the uncertainties in the model parameters. Then, the predicted model output and its uncertainty distribution are derived and compared to the resultant uncertainty distribution for environmental variabilities. © 2023 Proceedings of 18th International Probabilistic Safety Assessment and Analysis, PSA 2023. All Rights Reserved.

키워드