Application of Bayesian networks to the decommissioning of Fukushima Daiichi Nuclear Power Plant for rational decision-making

Many real-world problems involve significant uncertainty, and require the treatment of information from multiple sources, both subjective and objective. Bayesian networks allow for a more rigorous process, and provide results with quantified uncertainty. Four possible scenarios that may occur during the fuel debris retrieval process of the Fukushima Daiichi decommissioning works have been modeled in a Bayesian network in this paper, together with numerous sources of information that may be used to identify the scenario. The model has been shown to demonstrate remarkable flexibility in its usage. It has also provided some interesting results that demonstrate its usefulness in cases where data is scarce or in conflict, when human intuition is likely to fail. (author)