LAUSR

Abstract A Bayesian approach was studied to improve the belief for input parameters for the probabilistic safety assessment of high-level waste (HLW) repository. A GoldSim-based module has been developed by which sequential Bayesian updating of parameter distributions and safety assessment of the repository can be carried out. Updating the distributions for four selected parameters associated both with nuclide transport in the fractured rock medium and abnormal earthquake events was demonstrated through probabilistic safety assessment of the conceptual repository. For illustrative purpose, all the prior, sequential posterior distributions and likelihoods were assumed. Although, with insufficient data, there is a limitation to estimation of the real likelihood function, it sufficiently seems to be possible to offer improved results with enhancement of the belief for parameter values, which are usually uncertain, especially for the nuclide behavior in and around the repository system, which typically has both long time span and wide modeling domain.