Development of Efficient External Multi-Hazard Risk Quantification Methodology for Nuclear Facilities

Abstract

Probabilistic safety assessment (PSA) of nuclear facilities on external multi-hazards has become a major issue after the Fukushima accident in 2011. However, the existing external hazard PSA methodology is for single hazard events and cannot cover the impact of multi-hazards. Therefore, this study proposes a methodology for quantifying multi-hazard risks for nuclear energy plants. Specifically, we developed an efficient multi-hazard PSA methodology based on the probability distribution-based Boolean algebraic approach and sampling-based method, which are currently single-hazard PSA methodologies. The limitations of the probability distribution-based Boolean algebraic approach not being able to handle partial dependencies between the components are solved through this sampling-based method. In addition, we devised an algorithm that was more efficient than the existing algorithm for improving the limits of the current sampling-based method, as it required a significant computational time. The proposed methodology was applied from simple examples to single-and multi-hazard PSA examples of actual nuclear power plants. The results showed that the proposed methodology was verified in terms of accuracy and efficiency perspectives. Regarding the sampling-based method, it was confirmed that the proposed algorithm yielded fragility and risk results that have similar degrees of accuracy, even though it extracted a smaller number of samples than the existing algorithm.

Development of a Two-Stage DQFM to Improve Efficiency of Single- and Multi-Hazard Risk Quantification for Nuclear Facilities

“…Flowchart of the conventional direct quantification of fault tree using Monte Carlo simulation (DQFM) for singleand multi-hazard risk quantification (adapted from Ref. [14] and [23]).…”

“…Since system size cannot be reduced, the computational cost of the conventional DQFM can be reduced by optimizing the number of hazard points or by optimizing the N for each hazard point. Recently, Kwag et al [23] improved the conventional DQFM by reducing the hazard points for a given hazard map by optimizing the interval of the hazard map. Yet, there has been no attempt to reduce the N for each hazard point in the conventional DQFM.…”

“…Accordingly, the final multi-hazard grid has 861 points. While the same system model and seismic fragility model were adopted (See Equations (13)- (19) and Table 1), tsunami fragility information is adopted from Kwag et al [23] and summarized in Table 3. Like the previous example, two groups of components are assumed to be partially correlated because of spatial proximity ( = = 0.7): S11, S12, S13, and S14 are in the same reactor building, and S15 and S16 are located in the same diesel generator building.…”

“…When using the DQFM method, though, many samples (N) are required for each component and each hazard condition [23]. The number of samples required increases proportionally with the number of system components, hazard points, and N at each hazard point.…”

“…The proposed method assigns a small N to hazard intensity points that make small contributions to the final risk value and a large N to those that make large contributions. To validate and investigate the proposed method, the risk to nuclear facili-ties obtained by the two-stage DQFM is compared with those computed with the conventional DQFM, which is known to provide accurate results for a partially correlated system for both single-and multi-hazard risk quantification [14,23]. The risks to a nuclear facility that has partially correlated components are calculated under both single and multiple hazards to demonstrate the two-stage DQFM.…”