Fire Risk Analysis

Background

Recently, fire protection programs at nuclear power plants (NPP) have been under a gradual transition from a deterministic, prescriptive approach to a Risk-Informed, Performance-Based (RIPB) approach based on NFPA 805. In the NFPA 805 transition, as a basis for Fire Risk Evaluation (FRE), a plant-specific Fire Probabilistic Risk Assessment (Fire PRA) is utilized by each NPP. Experts have pointed out that the main gap in the “current Fire PRA methodology” (based on NUREG/CR-6850 and subsequent NUREGs) is the overestimation of plant risk due to the lack of realism introduced in the input parameters and modeling assumptions. A literature review [1] identified five areas of the current Fire PRA methodology where realism could be improved: (1) Fire ignition frequency, (2) Fire progression and damage modeling, (3) Interaction between fire progression and manual suppression, (4) Circuit failure analysis, and (5) Post-fire Human Reliability Analysis (HRA).

Integrated PRA (I-PRA) Methodology

In order to improve the realism associated with areas #2 and #3, the SoTeRiA Research Laboratory has developed the I-PRA methodology for fire risk analysis in NPPs (Figure 1) [2, 3]. Fire I-PRA, when compared with the current Fire PRA methodology, is advanced in three aspects: (i) probabilistic connection between plant PRA and the fire physics model, (ii) uncertainty propagation and computation of fire-induced cable damage probability, and (iii) fire detection and suppression analysis focusing on treatment of manual suppression [1].

Figure 1: Integrated Probabilistic Risk Assessment (I-PRA) Methodology for Fire PRA of Nuclear Power Plants.

The current Fire PRA methodology uses a data-driven approach for modeling human performance in manual suppression and utilizes an implicit time-based coupling between fire propagation and human performance (i.e., time-to-damage and time-to-suppression are computed by separate models without consideration of their interactions). Fire I-PRA has improved the coupling between fire progression and manual suppression in three steps [4]. In the first step, the explicit unidirectional coupling between the data-driven human performance model and a Computational Fluid Dynamics (CFD)-based fire model (Fire Dynamics Simulator [FDS]) was created by modifying the Heat Release Rate (HRR) curve based on the key timings of manual suppression activities [2]. In the second step, an HRA-based approach was developed by generating an explicit bidirectional coupling between an HRA model for the onsite fire crew and the FDS fire propagation model [5]. In the third step, a spatiotemporal human performance model was developed using an Agent-Based Modeling (ABM) technique and bidirectionally coupling it with a fire propagation model utilizing a Geographic Information System (GIS)-based spatial environment [6].

Although this new spatiotemporal coupling can be useful for fully dynamic PRAs, the widespread use of classical PRA logic by the nuclear industry and regulatory agency means that a transition to a fully dynamic PRA would require a significant investment of resources. To avoid this, the I-PRA methodology [2, 3] integrates the underlying simulation models with the existing static PRAs of NPPs by generating a probabilistic interface (‘d’ in Figure 1) that is equipped with advanced uncertainty analysis, dependency treatment, and Bayesian updating. Uncertainty propagation in Fire I-PRA is performed by conducting Monte Carlo simulation using the replicated Latin Hypercube Sampling (LHS) method on the Fire Dynamic Simulator (FDS). In I-PRA, the existing plant PRA model is integrated with underlying simulations in a “unified” computational platform. In the current Fire PRA methodology, the connection between underlying fire simulation models and the plant PRA is “passive,” i.e., the cable damage probabilities are computed as outputs from the fire model, which are then used as inputs to the PRA software. In contrast, in Fire I-PRA, the spatial and temporal information on input-output relationships between the plant PRA model and the underlying simulations (i.e., the coupling of the fire model and the human performance model) are recorded and analyzed in a unified platform. This unified connection allows a Global Importance Measure analysis to be conducted to generate a risk-importance ranking based on the impact of the fire protection parameters on the plant risk metrics [7-9]. The Fire I-PRA methodological framework was applied to a critical fire-induced scenario that has improved the realism of PRA, leading to a 50% reduction in core damage frequency [1].

Academia-Industry Project to Enhance Fire PRA

In order to enhance the methodological and practical values of Fire I-PRA, a Department of Energy (DOE)-funded academic-industry project was started in 2019 to expand the Fire I-PRA research to a full-scope NPP. The project is executed in three phases.

In the first phase of this project [10], an advanced screening process for Fire PRA is developed and implemented in order to identify the critical fire scenarios that could benefit the most from further detailed analysis using the Fire I-PRA methodology. The Risk-Informed over Deterministic (RoverD) screening methodology is introduced to gradually improve the Degree of Realism (DoR) and guide the screening process based on the estimation of risk and the associated cost (e.g., cost related to the required data collection for advancing DoR). The RoverD methodology utilizes a blend of deterministic and risk-based elements to demonstrate compliance with the regulatory requirements. This study discusses a spectrum of DoR in the screening process of risk-informed applications where the DoR is gradually and efficiently increased. Although the RoverD methodology is applicable for diverse risk-informed applications, this research covers its implementation for Fire PRA. In the case study, two fire modeling approaches (i.e., engineering correlations and zone models) are used to investigate the impact of improving the DoR in the screening process of Fire PRA for NPPs, specifically advancing the DoR in the Zone of Influence (ZOI) of fixed ignition sources. Ongoing research is directed toward theorizing and computationalizing the decision-making processes involved in the RoverD and its linkage the I-PRA approach.

In the second phase of the project, a computational platform, “SoTeRiA Fire,” is being developed that will significantly reduce the burden on the plant operator to estimate the level of fire risk at various locations within a nuclear power plant location. The developed code leverages fire simulation software approved by the Nuclear Regulatory Commission (NRC) to assess fire progression and spread in single- and multi-compartment fire scenarios. The SoTeRiA Fire code provides multiple options for initial and boundary conditions, representing different levels of realism and data requirements. Using this “risk-informed” technology, the commercial nuclear power plant operators can most efficiently apply resources to minimize the fire-related risk while reducing costs of operation and maintenance.

In the third phase of the project, the research team has been working on experimental validation of the advanced fire crew model developed for the Manual Fire Suppression Module (‘c’ in Figure 1) in I-PRA. For the regulatory acceptance of advanced simulation modeling used in PRA, validation is a crucial step. In this phase, the human performance model for manual detection and suppression undergoes experimental validation using limited-scope live-fire testing. This research is conducted in collaboration with the Illinois Fire Service Institute (IFSI). Based on results from the first and second phases, the most critical fire scenario is selected as a case study for this phase, and the fire test room will be designed based on the plant design information for that fire scenario. The scenario follows NUREG guidance for fire initiation and progression conditions, and manual suppression procedures follows plant specifications. A Probabilistic Validation (PV) methodology [8, 11, 12], will be implemented to connect the experimental data with the I-PRA framework and characterize and propagate the uncertainty associated with the experimental data. In the nuclear power domain, the historical fire tests focused on physical phenomena, while no existing fire test investigated the fire-human interactions. This research will be the first to validate the probabilistic simulation of fire-human interactions using the fire-human test data and to connect the fire-human test data to the system-level PRA scenarios.

Acknowledgement

This research is supported by the U.S. Department of Energy, Office of Science, Office of Nuclear Energy, under Award Number DE-NE0008856 (2019-2021).

References

[1] T. Sakurahara, Z. Mohaghegh, S. Reihani, E. Kee, Methodological and practical comparison of integrated probabilistic risk assessment (I-PRA) with the existing fire PRA of nuclear power plants, Nuclear Technology, Vol. 204, Is. 3, pp. 354-377, (2018).
[2] T. Sakurahara, Z. Mohaghegh, S. Reihani, E. Kee, M. Brandyberry, S. Rodgers, An integrated methodology for spatio-temporal incorporation of underlying failure mechanisms into fire probabilistic risk assessment of nuclear power plants, Reliability Engineering & System Safety, Vol. 169, pp. 242-257, (2018).
[3] H. Bui, T. Sakurahara, J. Pence, S. Reihani, E. Kee, Z. Mohaghegh, An algorithm for enhancing spatiotemporal resolution of probabilistic risk assessment to address emergent safety concerns in nuclear power plants, Reliability Engineering & System Safety, Vol. 185, pp. 405-428, (2019).
[4] T. Sakurahara, H. Bui, S. Reihani, E. Kee, Z. Mohaghegh, Enhancing realism in fire probabilistic risk assessment of nuclear power plants, Proceeding of the 30th European Safety and Reliability Conference (ESREL) and the 15th Probabilistic Safety Assessment and Management (PSAM) Conference, Venice, Italy, (2020).
[5] T. Sakurahara, Z. Mohaghegh, E. Kee, Human reliability analysis (HRA)-based method for manual fire suppression analysis in an integrated probabilistic risk assessment, SCE-ASME Journal of Risk and Uncertainty in Engineering Systems: Part B. Mechanical Engineering, Special Issue on Human Performance & Decision-making in Complex Industrial Environments, Vol. 6, Is. 1, (2020).
[6] H. Bui, T. Sakurahara, S. Reihani, E. Kee, Z. Mohaghegh, Spatiotemporal integration of an agent-based first responder performance model with a fire hazard propagation model for probabilistic risk assessment of nuclear power plants, SCE-ASME Journal of Risk and Uncertainty in Engineering Systems: Part B. Mechanical Engineering, Special Issue on Human Performance & Decision-making in Complex Industrial Environments, Vol. 6, Is. 1, (2020).
[7] T. Sakurahara, S. Reihani, E. Kee, Z. Mohaghegh, Global importance measure methodology for integrated probabilistic risk assessment, Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, (2019).
[8] H. Bui, T. Sakurahara, S. Reihani, J. Biersdorf, Z. Mohaghegh, Integrated Probabilistic Risk Assessment (I-PRA) Importance Ranking for Fire PRA of Nuclear Power Plants, Transactions of the American Nuclear Society (ANS) Winter Meeting and Technology Expo, Washington DC, USA, (2019).
[9] H. Bui, T. Sakurahara, S. Reihani, J. Biersdorf, Z. Mohaghegh, I-PRA uncertainty importance ranking to enhance fire pra realism for nuclear power plants, Transactions of American Nuclear Society (ANS) Winter Meeting and Technology Expo, Chicago, IL, (2020).
[10] S. Alkhatib, T. Sakurahara, S. Reihani, E. Kee, B. Ratte, Z. Mohaghegh, Academia-industry collaboration to advance screening processes in fire pra of nuclear power plants, Proceeding of the 30th European Safety and Reliability Conference (ESREL) and the 15th Probabilistic Safety Assessment and Management (PSAM) Conference, Venice, Italy, (2020).
[11] T. Sakurahara, “Integrated Probabilistic Risk Assessment (I-PRA) Methodology and Computational Platform for Fire PRA of Nuclear Power Plants (Doctoral dissertation),” Doctoral, Department of Nuclear, Plasma, and Radiological Engineering, University of Illinois at Urbana-Champaign, (2018).
[12] H. Bui, T. Sakurahara, S. Reihani, J. Biersdorf, and Z. Mohaghegh, “I-PRA Uncertainty Importance Ranking to Enhance Fire PRA Realism for Nuclear Power Plants,” 2020 American Nuclear Society (ANS) Winter Meeting, Virtual Conference, (2020).

CookieSettings CookieSettings