Seite - 1044 - in Book of Full Papers - Symposium Hydro Engineering

discharge and water surface level). However, the existing reservoir risk approach showed a limitation in assessing the interdependencies across the variables. In this context, this study proposed a Bayesian network based reservoir risk analysis. Based on the basic information of reservoir, various failure modes was first identified and the failure modes are then translated into nodes in the Bayesian network framework. The failure probabilities of the each node were estimated quantitatively by integrating limit state equations, which are composed of a set of random variables. Moreover, we investigated an integrated Bayesian network model to estimate overtopping risk from water surface level rise informed by climate change scenarios. A further discussion on the role of the uncertainty for overall risk is provided. This proposed procedure will help to effectively introduce the risk analysis for reservoirs safety in Korea. ACKNOWLEDGEMENTS This research was supported by a grant(17AWMP-B127568-01) from the Water Management Research Program funded by Ministry of Land, Infrastructure and Transport of Korean government. 1044