Seismic-HRA Methods Implementation in a Realistic-NPP Model: Comparative Analysis on a Plant-Level Risk Contribution

Existing human reliability analysis (HRA) methods were developed mainly for internal events (IEs) probabilistic safety assessment (PSA). These methods often contain underlying assumptions, such as the availability of procedures and level of basic training, which may or may not be applicable to the challenging and new environment created by an external event. For the evaluation of the operator actions (OAs) after an earthquake, the Swiss Regulator (ENSI) in its guideline ENSI-A05 proposes the refined basic model. The advantages of the method are its relative simplicity and its thoroughness. Naturally, this method has some disadvantages as well. Namely, for simple OAs with low values of the IEs human error probability - HEP_{\text int} , the increase factor is already very high for medium-strength earthquakes. Moreover, after a certain PGA-level, the human error probability as a response to an external hazard - HEP_{\text ext} becomes practically indifferent on its corresponding HEP_{\text int} . On the other hand, in the U.S., HEP_{\text ext} are calculated using increase factors. One such approach is the EPRI approach to HRA for external events with a focus on seismic. The HEP_{\text int} are being multiplied by these increase factors, which in turn are a function of a range of parameters, e.g. action location, plant damage state, time margin, memorized OAs, etc. Although more time-consuming for implementation, the HEP_{\text ext} becomes not that steep and not indifferent on its HEP_{\text int} . For the NPP Gösgen (KKG) PSA, a calculus of variations is carried out, using an approach that combines both the above methods. The so-called hybrid-method - the novelty of the work presented - makes distinction between four earthquake strengths and three time ranges. A distinction is also made between whether the action takes place in the command room or outside the command room. This method has the advantage over the refined basis model approach that the complexity of the human failure event (HFE) is directly incorporated into the HEP. Due to the large increase factors, especially for short-term HFE, sufficient conservatism is also guaranteed. The objective of this manuscript is to obtain and present a quantitative assessment of the relative change in the computed plant risk if one swaps between the hybrid approach and the EPRI approach. A comparative analysis is conducted and the results are discussed.

Keywords: PSA, HRA, Seismic-HRA, Realistic-NPP PSA.