Nuclear Technology / Volume 172 / Number 2 / November 2010 / Pages 108-119
Technical Paper / Reactor Safety / dx.doi.org/10.13182/NT10-A10898
We consider the design of a sodium-cooled fast reactor (SFR) in the context of the risk-informed technology neutral framework (TNF) for licensing new reactors that has been proposed by the U.S. Nuclear Regulatory Commission staff. In lieu of design-basis accidents (DBAs), the TNF imposes limits on the frequency and consequences of accident sequences called licensing-basis events (LBEs). We present a method to define LBEs for a SFR using generic functional event trees. Very large consequence events are considered beyond the licensing basis in the TNF as long as their mean frequencies are less than 1 × 10-7 per reactor year.
For SFRs, energetic accidents have historically represented a major regulatory hurdle in the traditional licensing system that is based on DBAs. As a result, key systems that prevent or mitigate these accidents may have been overdesigned. We propose a new importance measure, the Limit Exceedance Factor (LEF). It is the factor by which the failure probability of structures, systems, and components (SSCs) may be multiplied such that the frequency of a risk metric reaches a limit. LEF allows a designer to know how much margin exists to the safety limit for each SSC. Alternatively, in the case where a design does not meet the frequency limit, LEF can reveal which systems are candidates for improvement to satisfy the limit. Within the TNF, using a frequency limit of 1 × 10-7 per reactor year and LEF, we find that for some SSCs a wide margin exists to this limit. Therefore, these SSCs are candidates for simplification resulting in economic benefit. This simplification should be done under the frequency-consequence constraints and the deterministic defense-in-depth requirements described in the TNF.