In recent years, several advancements in nuclear power plant (NPP) probabilistic risk assessment (PRA) have been driven by increased understanding of external hazards, plant response, and uncertainties. However, major sources of uncertainty associated with external hazard PRA remain. One source discussed in this study is the close coupling of physical impacts on plants and overall plant risk under hazard events due to the significant human actions that are carried out to enable plant response and recovery from natural hazards events. This makes human reliability and human-plant interactions important elements in to consider in enhancing PRA to address external hazards.
One of the challenges in considering human responses is that most existing human reliability analysis (HRA) models, such as SPAR-H and THERP, were not developed for assessing ex-control room actions and hazard response. To support this new scope for HRA, HRA models will need to be developed or modified to support identification of human activities, causal factors, and uncertainties inherent in external hazard response, thereby providing insights regarding event timing and physical event conditions as they relate to human performance. In this study, the first step of such work is performed by identifying human failure events (HFEs) for human response to flooding hazards. These HFEs are human actions or inactions that are involved in human response to flooding hazards and could contribute to the loss of a critical function for the plant in the scenario being examined. Several resources are used to identify these HFEs, including flooding reports from the Nuclear Regulatory Commission (e.g. NUREG/CR-7256: Effects of Environmental Conditions on Manual Actions for Flood Protection and Mitigation), interviews with experienced PRA and HRA analysts, and tabletop walkdowns of flooding scenarios with a project team. Also, task decomposition analyses using the cognitive-based Phoenix HRA model are also used to identify HFEs. This paper will discuss early results of these analyses.