| dc.description.abstract | This paper gives a complete account of recent work on the prediction of power restoration in severe accidents and events, and quantifies the benefits and impact of deploying back-up or emergency power systems in nuclear reactors. The overall outage data for all types of major events and disasters follow the same fundamental trends based on statistical learning theory, and are correlated by simple theoretically based exponential equations that include the degree of difficulty. This trend is shown to be completely independent of severe event type (e.g., hurricanes, ice storms, flooding, earthquakes, cyclones, and fires) but the rate systematically depends on the degree of difficulty. The paper emphasizes that the physics of learning, analytical methodology, technical statistical theory, and extensive event database already implicitly and fully include all human errors, actions, and decisions made during power restoration for the extremely adverse conditions prevalent in severe events and actual disasters. The theory shows flexible coping strategies/emergency power system FLEX/EPS reliability, deployment timescale, and severe event power restoration rate are intrinsically coupled together. The analytical results can be used to define the EPS design and reliability requirements and the potential risk benefit and deployment timescales in terms of the probability change in the risk of extended power loss. | |
