Modeling the interface of manual fire protection actions with fire progression in fire probabilistic risk assessment of nuclear power plants

This paper develops a methodology for �explcit� modeling of the interface between manual fire protection (i.e., manual fire detection and suppression) and a Computational Fluid Dynamics (CFD) fire progression model, utilizing Fire Dynamics Simulator (FDS), in Fire Probabilistic Risk Assessment (PRA) of nuclear power plants (NPPs). A literature review revealed that there had been no research on developing an �explicit� interface between a CFD-based fire model and manual fire protection until very recently, when Kloos et al.^1,2 integrated FDS with dynamic event trees and Human Reliability Analysis (HRA). The research demonstrated in this paper has been conducted in an Integrated PRA (I-PRA) framework, i.e., an integration of classical PRA of the plant and a simulation-based module, and therefore, using dynamic event trees is not applicable. However, to obtain a more accurate and realistic estimation of fire-induced NPP risk, there is a need to account for (i) the performance of the plant's crew in manual detection and suppression, and (ii) the interactions of the crew with the fire progression. In the existing Fire PRA methodology (NUREG/CR-6850),³ manual suppression is addressed by a data-driven approach, where the time to manual suppression is estimated by a non-suppression curve - a statistical probability model derived from historical fire event data. Meanwhile, the interactions between manual suppression and fire progression are addressed through an implicit method based on the competition between two separately computed time quantities for �time to target damage� and �time to fire suppression�. In the methodology introduced in this paper, the explcit interface between FDS and manual fire protection is developed using a data-driven model for manual suppression. To build this interface, the Heat Release Rate (HRR) curve, which is an input to FDS, is modified based on data-driven probability models of three timings associated with manual fire protection: time to fire detection, time to fire brigade response, and time duration of fire suppression. A case study, using a typical NPP fire scenario, is conducted to demonstrate the implementation of the explicit interface and to illustrate the impact that the interface can have on the results of Fire PRA. The results show that the fire-induced damage probabilities computed by the I-PRA framework are smaller than those computed by the existing Fire PRA of NPPs (i.e., NUREG/CR-6850 methodology).