Addressing Multi-Hazards Risk Aggregation for Nuclear Power Plants through Response Surfaces and Risk Visualization Tools
Traditional approaches to multi-hazards risk aggregation (MHRA) in the context of nuclear power plants probabilistic risk assessment (PRA) are limited by a number of heterogeneities in uncertainties and levels of conservatism between PRA models for various hazard groups. In this study, we propose a new, consistent approach to the quantifiable aspects of MHRA with a focus on relative rather than absolute risk metrics. Using response surfaces based on arbitrary polynomial chaos expansion in combination with radar chart visualization of overall risk and associated uncertainties, impacts of a large number of uncertain input parameters on plant response (e.g., core damage frequency, or CDF) are systematically assessed, and cliff edge regions where system response gradients exceed certain thresholds are identified. This integrated tool complements traditional PRA methods to facilitate understanding of plant risk status and provide insight for risk-informed decision-making.
- Investigated and implemented response surface methodology applied to probabilistic risk assessment. This involved adapting advanced response surface algorithms, only used by the aerospace and petroleum engineering industries so far, to nuclear power plant risk analysis.
- Created scripts to automatically generate surrogate response functions and detect cliff edge effects in plant risk, based on actual plant PRA models from U.S. utilities.
- Developed visualization methods using radar charts to represent and compare risks to regulatory metrics in support of high-level, risk-informed decision-making.