Approach to and Insights from Detailed Fire Simulation Studies at Leibstadt NPP ^†

Abstract

The Leibstadt Nuclear Power Plant (KKL) recently completed a comprehensive full-scope Fire Probabilistic Safety Assessment (Fire PSA) to fulfill the updated Swiss regulatory requirements (ENSI-A05) and align with international standards. The study was conducted using the NUREG/CR-6850 framework, incorporating state-of-the-art methodologies across different areas of the study, advanced fire modeling tools (CFAST and FDS), and the latest plant-specific data. As part of detailed fire modeling, a bespoke methodology was developed, tailored to KKL’s plant-specific characteristics, to ensure a systematic and standardized approach to fire scenario analysis while maintaining quality, consistency, and traceability. The analysis focused on evaluating fire risks in critical plant areas, such as the drywell, containment, main control room, remote shutdown areas, and cable spreading room. For each scenario, the fire-generated conditions, such as the extent of fire propagation and the time to damage targets, were analyzed using plant-specific heat release rate (HRR) and calorific potential (CALPOT) values. The study also addressed aspects such as multi-compartment analysis, fire-induced cable impacts, and treatment of multiple spurious operations. This paper highlights the methodological enhancements achieved by integrating international best practices and KKL-specific adaptations into a unified fire modeling framework. The results provide critical insights into fire propagation dynamics, validate the effectiveness of safety features, and support risk-informed decision-making for enhanced fire safety and regulatory compliance. The outcomes of fire modeling were utilized to develop fire event trees and refine the consequences of fire scenarios, thereby enabling a more realistic estimation of fire risk in the KKL Fire PSA study. Overall, the KKL PSA aims to serve as a benchmark for future fire risk assessments in the nuclear industry.

1. Introduction

From 2019 to 2021, KKL undertook a comprehensive update of its internal Fire PSA study to meet the updated ENSI-A05 requirements [1] and align with international best practices, including the IAEA [2] and ASME PRA guidelines [3]. This extensive effort aimed to address evolving regulatory expectations, incorporate advancements in fire modeling methodologies, and leverage the latest plant-specific data to enhance the accuracy and reliability of fire risk assessments.

This project utilized up-to-date KKL databases, covering component lists, cable routing, fire loads, ventilation connections between compartments, fire suppression features, plant-specific walkdown inputs, and the latest internal events PSA model.

The revision of the Fire PSA study was based on the NUREG/CR-6850 framework [4,5], integrating state-of-the-art guidance and tools with plant-specific adaptations. The study utilized industry-recognized fire modeling software, such as consolidated fire and smoke transport (CFAST) and fire dynamics simulator (FDS), to simulate fire growth and propagation dynamics in critical plant areas.

A key component of this study was the development of a tailored fire modeling methodology, designed to integrate international best practices and guidelines into a unified, plant-specific framework [6,7,8,9,10,11,12,13,14,15,16,17]. The methodology enabled consistency, traceability, and quality assurance across all fire scenarios, ensuring a robust and reproducible approach to fire risk assessment. The outcomes of fire modeling were utilized to develop fire event trees and refine the consequences of fire scenarios, which enabled a more realistic estimation of fire risk in the Fire PSA study. This paper presents the methodological framework, key findings, and practical applications derived from KKL Fire PSA, highlighting its contribution to the continuous improvement of fire safety practices in the nuclear industry.

2. Past Experience

The internal fire analysis of KKL was re-evaluated between 2012 and 2015 to incorporate the enhancements made in the updated KKL PSA internal events study and to address recommendations from the IAEA International Probabilistic Safety Assessment Review Team (IPSART) mission conducted in 2014. This re-evaluation focused on revisiting the fire impact assessments and re-quantifying a specific set of scenarios retained after quantitative screening.

Following the 2016 Periodic Safety Review (PSR), the Swiss regulatory authority ENSI issued action points requiring verification of the validity of existing detailed analyses and the accuracy of previous fire simulation studies.

A comprehensive update of the KKL Fire PSA was deemed essential to achieve the following:

• Address regulatory comments from the PSR;
• Comply with enhanced regulatory guidelines;
• Integrate international best practices and standards;
• Leverage recent enhancements in KKL databases;
• Incorporate the latest generic data and plant-specific fire-related events;
• Reflect the ‘as-is’ status of the plant (e.g., plant retrofits and modifications);
• Consider the latest modeling enhancements carried out in the internal events PSA;
• Ensure consistency between deterministic and probabilistic safety assessments.

Fire simulations were performed in the past using COMPBRN III simulation tool, supported by experimental results available then. This fire simulation code is outdated and not considered sufficiently qualified for NPP fire modeling applications.

Although certain modeling principles from the previous Fire PSA study remain valid, the dominant compartment types have changed in the new study. This change is attributed to the revised compartment definitions, more comprehensive impact assessment, and the latest fire ignition frequency data. Consequently, it was necessary to analyze a new set of scenarios.

3. Methodological Framework

The KKL Fire PSA was conducted in accordance with NUREG/CR-6850 [4,5] and was notably enhanced by incorporating the most recent KKL technical databases, as well as state-of-the-art methodologies and guidance on fire modeling [8,9,10,14,15,16,17].

The study encompassed a comprehensive evaluation of fire-induced cable impacts, revision of fire impact files, treatment of multiple spurious operations, and a pragmatic approach to multi-compartment analysis. Additionally, the application of sophisticated fire modeling tools, such as CFAST and FDS, enabled precise fire simulations.

Figure 1 illustrates the methodological framework, synthesizing theme-specific international guidelines into a refined analytical process.

Figure 1. Application of international guidelines in KKL Fire PSA.

Within this study, the broadly used fire modeling methodology was specifically tailored to the unique characteristics and requirements of the KKL. This customized approach was developed to provide a robust, reproducible, and detailed evaluation of fire scenarios at the plant, ensuring traceability and consistency across all analyses. By focusing on the specific needs of KKL, the methodology aligns with advanced analytical practices and complies with regulatory standards.

The fire scenario modeling process in this study follows the structured approach outlined in NUREG-1934 [8], with necessary modifications to account for KKL-specific characteristics, such as unique fire locations, propagation mechanisms, and ignition source properties. The key milestones of the fire scenario modeling process are outlined as follows:

• Defining fire modeling goals and objectives (time to damage, extent of fire propagation, hot gas layer (HGL) temperature, performance criteria of fire barriers, verification of spatial separation, etc.).
• Modeling of rooms, including an approach for room transformation (for complex geometries).
• Treatment of ventilation conditions (modeling of air leakages through doors, and use of a pseudo-wall vent for under-ventilated fires).
• Evaluation of fire modeling predictive parameters → selection of an appropriate fire modeling tool.
• Modeling of fire loads (ignition source and targets).
• Treatment of cable failures using empirical and thermally-induced electrical failure (THIEF) models.
• Evaluation of fire-generated conditions in the modeled enclosure.
• Sensitivity analysis to support the use of results (on fire parameters, geometric assumptions, damage criteria assumptions, etc.).
• Analysis and application of fire simulation results.

Following the quantitative screening stage, fire compartments identified as risk-significant were subjected to a comprehensive refinement phase. During this phase, the potential consequences of fire events were analyzed at a scenario level. Detailed plant walkdowns were conducted to verify the spatial information of ignition sources and targets, which are critical for modeling the fire scenarios. Additionally, an exhaustive review of plant documentation was conducted—including piping and instrumentation diagrams (P&IDs), equipment layout plans, cable routing diagrams, and heating, ventilation, and air conditioning (HVAC) system configuration, to ensure the accuracy and completeness of input data. The main objective was to determine the extent of fire damage from selected ignition sources, in terms of equipment impacted and their corresponding time to damage.

Fire growth and propagation dynamics were analyzed using CFAST and FDS over a two-hour simulation period. These simulations aimed to achieve a high-fidelity representation of fire-induced damage conditions within critical compartments. To ensure methodological rigor, a dedicated and plant-specific framework was developed, ensuring quality, consistency, and traceability throughout the fire modeling process. In total, 19 fire scenarios were analyzed in the following PSA-relevant buildings: drywell, containment, reactor auxiliary building, special emergency heat removal building, and operations building.

Scenarios were defined based on the type of ignition source and the room considered, such that the scenario represents the bounding fire-generated conditions for a given ignition source type in that room. For example, in remote shutdown areas, fire from the control panel, logic cabinet, high-energy arcing fault (HEAF) cabinet, non-HEAF switchgear cabinet, and transformer was modeled to cover all the main ignition source types. Likewise, in cable canals and cable shafts, fire development from cables was modeled. The scenario definition facilitates the use of fire modeling results for a given ignition source type (e.g., cabinet or panel) in symmetrical rooms and rooms with the same ignition source types. Seventeen scenarios were modeled using CFAST, and two complex scenarios originating from non-rectilinear geometry rooms were modeled using FDS.

Figure 2 illustrates the type of fire scenarios analyzed within the study.

Figure 2. Types of fire scenarios analyzed for KKL.

The subsequent sections present the overarching principles and methodological steps followed in the KKL fire modeling framework. This includes a detailed discussion of the foundational concepts, assumptions, and analytical approaches that underpin the development and application of the fire modeling methodology. By addressing these generic aspects, the aim is to provide a thorough understanding of the systematic processes and considerations that guide the modeling efforts.

3.1. Step 1: Definition of Fire Modeling Goals and Objectives

The primary goals of the fire modeling process are to evaluate the potential consequences of fire scenarios, including the number of affected pieces of equipment, and to validate the spatial separation credited in defining specific fire compartments (e.g., drywell and containment). To achieve these goals, the following objectives were defined:

• Determine the time to damage for targets (e.g., equipment, cables) located near ignition sources and within the scenario boundary.
• Assess the extent of fire propagation and evaluate fire-generated conditions in the room, assuming a well-ventilated primary fire.
• Estimate the temperature of hot gas layer in adjacent rooms due to fire and/or smoke propagation from the main room where the fire scenario is postulated.
• Evaluate the maximum temperature and heat flux on fire doors and qualitatively assess their effectiveness in containing the fire.
• Evaluate the maximum smoke concentration, visibility, and habitability conditions for specific locations such as the main control room.

3.2. Step 2: Fire Scenario Characterization

This step involves a detailed characterization of all elements required to model a fire scenario. Key elements include the following:

• Room Details: Geometry, dimensions, and spatial layout of the enclosure.
• Ignition Source and Targets: Location and orientation of the ignition source, secondary combustibles, and targets within the enclosure.
• Ventilation Conditions: Airflow and oxygen availability in the room.
• Fire Protection Features: Credited fire protection measures, such as fire doors and suppression systems.
• Fire Parameters: Heat release rate (HRR), fire growth rate, calorific potential (CALPOT), and thermo-physical properties of materials.
• Damage Criteria: Damage thresholds for equipment and cable damage.

Focused plant walkdowns were conducted to provide scenario-specific inputs, such as selection of ignition sources, choice of fire location, spatial information, orientation of combustibles, and the nature of secondary fire loads.

3.3. Step 3: Room Modeling

For each fire scenario, all rooms connected to the primary fire room are identified and modeled. Connections through walls, ceilings, floors, vents, or doors are included if they influence fire-generated conditions. The modeling of adjacent rooms depends on the specific objectives of the scenario.

3.3.1. Fire Door Modeling

KKL fire doors are categorized into three resistance classes: EI30 (non-fire-rated), EI60 (60-min fire resistance), and EI90 (90-min fire resistance). Doors are initially modeled in a closed configuration, with air leakage through door clearances typically represented by an opening area equal to 2% of the total door height. For fire-rated doors, this opening area is increased by an additional 8% if the temperature on the exposed side reaches 400 °C, reflecting potential deformation possibilities at elevated temperatures. Non-fire-rated doors are conservatively assumed to open completely when the temperature on the exposed side reaches 100 °C, based on expert judgment.

An example of doors modeled for a fire scenario in the relay room is presented in Figure 3.

Figure 3. Doors modeled in a relay room.

3.3.2. Vent Modeling

Ceiling and floor vents, if present, are modeled to reduce room temperature and pressure and to provide oxygen for combustion. For instance, in CFAST, a single ceiling or floor vent may be modeled at the center of the room by calculating the net opening area of all ceiling and floor vents in the room, or individual ceiling or floor vents may be modeled with their respective opening areas if their positions and opening areas are known. If the ceiling or floor vent supports the routing of cable trays, cables, or any other structural obstructions, the net opening area is calculated based on engineering judgment.

3.3.3. Room Geometry Transformation

Since fire modeling tools such as CFAST and FDS work best with rectilinear geometries, non-rectangular rooms are transformed into equivalent rectangular geometries. The transformation follows the predefined principles outlined below:

• Preservation of room height—to account for hot gas and smoke filling and to conserve the plume and ceiling jet correlations of the fire modeling tool.
• Preservation of room volume—to account for smoke filling and ventilation conditions. Typically, the length and/or width of the room are adjusted to maintain the overall aspect ratios close to the original aspect ratios and stay within the validation range of CFAST and FDS.
• The position of surrounding equipment and cables relative to the ignition source and surfaces is maintained to capture the radiation effects of fire.
• Preservation of total surface area of the boundaries—to conserve the heat transfer through walls, floor, and ceiling (depending on the location of the target of interest).
• Consideration of the limitations of the fire modeling tool—with respect to aspect ratio validation limits.

3.4. Step 4: Treatment of Ventilation Conditions in the Room

Ventilation conditions in the room (i.e., available oxygen) are critical for sustaining combustion of the fire loads in the scenario. They are expressed using the global equivalence ratio (denoted as Φ), which relates the energy release rate of the fire (HRR) to the energy release rate that can be supported by the mass flow rate of oxygen into the room. When the equivalence ratio is equal to 1, the exact amount of oxygen required for complete combustion is available in the enclosure. When the ratio is >1, the environment is fuel-rich, and the fire is considered to be under-ventilated. The reverse is true when the ratio is <1, signifying a well-ventilated environment, The calculation of equivalence ratio considers the opening area of wall vents and doors in the room.

Equivalence ratio is calculated as follows:

Φ = Q/(ΔH_O2 ∙ m_O2)

(1)

where

Q = heat release rate of the fire (kW);

ΔH_O2 = H heat of combustion for oxygen (kJ/kg);

m_O2 = mass flow rate of oxygen into the room (kg/s).

Modeling under-ventilated fires is challenging because the HRR profile for the fire is specified by the analyst and is not automatically adjusted by the fire models according to the conditions present in the room (which change dynamically based on fire behavior). To achieve the fire modeling objectives, the room must be well-ventilated to allow complete combustion of the primary fire. This implies that unless an additional oxygen supply is available in the room, the fire does not burn according to the defined HRR profile. In such cases, a pseudo-wall vent is provided in the room to ventilate the primary fire.

The positioning of the pseudo-wall vent depends on the specific room configuration, the location of the fire (ignition source), and the targets.

An example of a pseudo-wall vent created in a cable room is presented in Figure 4.

Figure 4. Pseudo-wall vent in a cable room.

Positioning of the wall vent plays an important role in FDS; however, it is less significant in CFAST models.

3.5. Step 5: Fire Modeling Tool Selection

Fire models are categorized into three types based on their simplicity and the degree to which spatial and temporal issues can be solved: algebraic and empirical models (FDTs), two-zone models (CFAST), and CFD models (FDS). The choice of the fire modeling tool depends on various factors, including:

• Predictive capabilities of the tool;
• Limitations of a particular fire modeling tool;
• Scenario objectives;
• Complexity of the room geometry.

The ability of a fire modeling tool to predict the fire-generated conditions of a scenario close to a realistic case is measured based on how well the non-dimensional fire parameters fit within the validation ranges. The important non-dimensional fire parameters and their relevance to fire scenario modeling are calculated and verified based on the limits specified in NUREG-1824 [10]. These parameters include room aspect ratio, global equivalence ratio, fire Froude number (the fire Froude number is defined as the ratio of inertial to buoyancy-induced velocities, see Table 3-3 of NUREG-1824 [10]. Its value must be within the validation range of the CFAST and FDS tools), the flame height to ceiling height ratio, and the distance from the target to the fire. These parameters were evaluated for each scenario to determine the appropriate modeling tool. Suitable modifications were made if the ranges exceeded the validation limits of the tool. For instance, the lower and upper limits of the fire Froude number (Q*) for CFAST are 0.1 and 9.1, and for FDS, they are 0.1 and 63. This largely depends on the HRR of the fire. If the value is found to violate the validation range for a given scenario, the burning area of the primary fire is modified to be within this range. The modified Q* is then used to calculate the flame height and plume temperature.

3.6. Step 6. Modeling of Fire Loads

Accurate prediction of fire-generated conditions requires detailed information about fire loads, including their location, dimensions, orientation, and fire characteristics. The spatial and dimensional details are obtained from the scenario-specific walkdown checklists supported by disposition plans and photographs taken during walkdowns.

For each fire scenario, the fire loads in the room are categorized into ‘ignition source’ and ‘targets’. Typically, targets are the equipment or cables that result in initiating events or affect mitigation functions modeled in the PSA, or they can be secondary combustibles that aid in fire spread. For the KKL detailed fire modeling, the primary fire load is referred to as the ignition source, and all other fire loads in the model are referred to as targets. If the damage criterion for a target is met in the simulation, it will, in turn, be treated as an ignition source (i.e., secondary fire).

Figure 5 shows an illustration of primary fire and secondary fires in a relay room scenario.

Figure 5. Primary fire and secondary fires in a relay room scenario.

Some key assumptions regarding damage criteria include the following:

• For motors, switchgears, solenoids, etc., the fire vulnerability is limited to the vulnerability of the cable supporting the component.
• Electronic components based on integrated circuit-type devices can fail at lower damage thresholds as observed in fire experiments; the damage thresholds are based on Appendix H of NUREG/CR-6850 [5].
• Electromechanical devices such as relays, switches, and circuit breakers are qualified to higher damage thresholds. Since detailed guidance is not available, their damage thresholds are conservatively assumed to be the same as those of solid-state control components.

Passive fire protection features were credited for cables (if available), to account for the severity of fire propagation from cables and delay the damage and ignition of cables to a certain extent, depending on the type of cables. Based on Appendix Q of NUREG/CR-6850 [5], a time delay of 20 min is considered for cable trays with solid bottoms and thermoset cables, and 4 min for thermoplastic cables. Additionally, thermally-induced electrical failure (THIEF) and empirical models were applied to predict the electrical and functional failure of cables, accounting for the time delay of heating from the outer insulation or jacket to the inner conductor material.

Examples of HRR selection and fire location selection are presented in Figure 6, Figure 7 and Figure 8.

Figure 6. HRR selection for cabinet fires.

Figure 7. Selection of fire location in cable rooms (Orange dot represents the transient fire (ignition source) and grey dots represent vertical cable trays, denoted as CT1 to CT5).

Figure 8. Pyrosim render showing the selection of fire location in containment (sector 2 indicated within the yellow boundary).

3.7. Step 7: Modeling Assumptions

Some key global assumptions used for fire scenario modeling are listed below:

• Deflagration and detonation of fire loads, and fire-induced structural failures, are not considered in the scope.
• Cables within conduits are considered as potential damage targets but not as ignition sources, as they cannot contribute to fire growth.
• The incipient and decay phases of the fire growth stages are not considered in the modeling of fires. The decay time and its associated CALPOT are accounted for in the steady burning phase.
• Pyrolysis rates are independent of the irradiation level or gas temperature, except for cable trays, where flame spread models are used to propagate fire along the cable trays depending on the type of cables.

Scenario-specific assumptions, while critical to the analysis, are not discussed here for brevity.

3.8. Step 8: Sensitivity Analysis

Sensitivity analyses are conducted for each fire scenario to evaluate the impact of modeling assumptions and input parameters. This step is particularly important when fire-generated conditions appear overly conservative or conflict with plant design features. The results of these analyses are used to refine the model and ensure alignment with the physical and operational characteristics of the plant.

The scope of sensitivity analyses depends on the characteristics of the specific fire scenario. Among various parameters, the following have the most influence on fire-generated conditions:

• Choice of HRR and CALPOT values;
• Ventilation conditions;
• Location of fire source and targets;
• Zone of influence (ZOI) for different material choices;
• Insulation properties of cables (thermoset vs. thermoplastic);
• Presence of structural elements or obstructions in the path of the fire source and targets.

The impact of these parameters on fire-generated conditions can be either optimistic or conservative, depending on the scenario specificities.

3.9. Step 9: Quality Assurance and Documentation

A robust two-stage interactive review process was followed for every fire scenario model. The fire models were developed by an independent consulting organization (RELSAFE) and reviewed by KKL PSA team, with plant personnel participating in walkdowns and providing essential information required for fire modeling. The overall fire modeling methodology, individual fire scenario models, results, and documentation were peer-reviewed and quality-assured by an independent fire modeling expert with vast experience in NPP applications and other safety-critical industries.

A standardized reporting structure was adopted for documenting the modeling of fire scenarios. Each fire scenario report includes the following elements:

• Fire scenario definition and geometry;
• Model elements and inputs;
• Characterization of ignition sources, secondary combustibles, and targets;
• Evaluation of predictive parameters and simulation settings;
• Analysis of fire-generated conditions, such as surface temperatures, heat flux, and gas layer temperatures.

CFAST/FDS input files, list of equipment and cables, walkdown photographs, temperature and heat flux measurements on components and cables were included as supporting attachments.

4. Considered Fire Scenarios

Table 1 presents a selection of fire scenarios analyzed as part of a detailed fire modeling task in KKL Fire PSA, together with key outcomes. Scenario numbers, compartment identifiers, compartment fire protection features, and fire scenario characterization aspects are not discussed here for brevity.

Table 1. Fire scenarios considered in KKL Fire PSA.

Scenario Description	Results and Conclusions
Fire from a control panel in the emergency control room	➢ Fire from the ignition source panel damages and ignites the immediately adjacent panels and a cable tray. ➢ Cabinets from the adjacent banks or cable trays are not damaged. ➢ The surface temperature on the adjacent panels is 65 °C, which prevents manual operator actions. ➢ The maximum HGL temperature attained in the room is 48 °C. The fire doors experience a maximum temperature of 33 °C and remain intact for the entire duration of the fire simulation.
Fire from a switchgear cabinet in the emergency control room (non-HEAF).	➢ Six adjacent cabinets in the same row of the cabinet bank are damaged and ignited. ➢ The adjacent cabinet banks and overhead cable trays are not damaged. ➢ The maximum HGL temperature attained in the room is 63 °C. The fire doors experience a maximum temperature of 38 °C and remain intact for the entire duration of the fire simulation.
Fire from a HEAF cabinet in the safety division switchgear room	➢ The energetic fire damages two immediately adjacent cabinets on either side of the ignition source located within the ZOI of 0.9 m. With the combined fires from ignition source and secondary combustibles, two additional cabinets are damaged at t = 7 min due to high incident heat flux. Further fire propagation or damage to cabinets is not possible due to oxygen depletion. ➢ The maximum HGL temperature attained in the switchgear room is 137 °C. The maximum surface temperature attained on fire doors is 52 °C and 44 °C, so the fire doors remain intact for the entire duration of the fire simulation. ➢ With an extended ZOI of 2.7 m (sensitivity case assuming cabinet internals containing aluminum-based components), all cabinets in the same row are damaged. In addition, two cabinets in the adjacent row, separated by ~1.3 m distance, are also damaged. The results are conservative predictions because of the postulated ZOI.
Fire from open cables in cable trays in the room adjacent to the emergency control room	➢ A self-ignited cable fire originating in the center of the trays spreads along the length of the cable tray in ~22 min and burns a total of 15 m. Smoke filling is accelerated in the corridor and connected rooms due to the simultaneous ignition of multiple cable tray segments in a short period of time. ➢ The maximum HGL temperatures attained in the room, cable tray corridor, open cable area, and adjacent emergency control room are 275 °C, 333 °C, 126 °C, and 84 °C, respectively. The fire door experiences a maximum temperature of 66 °C and remains intact for the entire duration of the fire simulation.
Fire from cables in the cable shaft area	➢ The transient fire ignites the bottom-most target point in all five cable trays in less than 5 min in both simulations due to conservative fire location and closeness of the modeled target points in the narrow cable shaft room. With thermoplastic insulation, all cable trays are damaged at 3 min (slightly earlier than thermoset insulation). ➢ The fire propagates along the cable tray with a spread rate of 11 mm/s (thermoset) or 14 mm/s (thermoplastic) and ignites the next cable tray target points. Due to severe under-ventilation in the cable shaft area, the fire stops at the second target point in each cable tray and does not spread to other cable target points located at higher elevations. This corresponds to a cable length of 2 m in the shaft. ➢ Considering thermoset insulation, the maximum hot gas temperature in the main and adjacent cable shaft areas is 183 °C and 80 °C, respectively. With thermoplastic insulation, the hot gas temperature in the main and adjacent cable shaft areas is 202 °C and 76 °C, respectively.
Fire from a control cabinet in Decontic room (ABB’s modular process automation and control systems)	➢ The primary fire from the cabinet damages two immediately adjacent cabinets due to a high incident heat flux exceeding 6 kW/m2, and the overhead cable trays due to high surface temperature exceeding 205 °C. ➢ All overhead cables are ignited in ~10 min due to the minimal vertical spacing between the cable trays in the stack, as these trays are in the flame region of the primary fire. Horizontal fire spread in these trays is limited to a maximum of 2 m on either side of the ignited segment due to oxygen depletion in the room. ➢ Adjacent cabinet banks and other cable trays in the room are not damaged. ➢ The maximum HGL temperature in the Decontic room is 126 °C. The fire doors experience maximum surface temperatures of 49 °C, 129 °C, and 48 °C, respectively, so they remain intact for the entire duration of the fire simulation.
Fire from a control panel in the main control room (MCR)	➢ Fire from the ignition source damages the immediately adjacent panels. Panels in adjacent groups and cabinets at the back are not damaged. ➢ The maximum hot gas layer temperature in the main control room is 60 °C, and that in the adjoining room is 38 °C. The maximum surface temperature on fire doors is 51.6 °C and 51.5 °C, respectively. ➢ The hot gases and smoke accumulate readily in the MCR given a fire from the control panel. The smoke layer descends below 1.8 m from the floor elevation after 11 min, and the optical density surpasses 3 m−1 at ~8 min. It is conservatively assumed that these conditions may force control room abandonment after 11 min.
Fire from lubrication oil associated with reactor recirculation system (RRS) pump train A	➢ Fire originating from the drywell sector 2 damages only the components and cable trays in the same sector. Damage to components and cable trays in horizontal and vertical directions is not possible, even for the large fire assumed in this scenario. The maximum temperature recorded in adjacent sectors at higher elevation is 200 °C. The layout and arrangement of components and cables in different sectors of the drywell prevent fire spread to adjacent sectors. ➢ All components and cables in the drywell are qualified to operate under elevated ambient conditions. All safety division cables are either routed in metallic conduits or in enclosed cable trays, and are made of non-burnable insulation material. Flexible armored shielding is used at connection points. This prevents the ignition of secondary combustibles. Some cable trays in the flame or fire plume region are damaged due to high incident flux or surface temperatures. There are various structural elements surrounding the RRS pumps, which are not included in the model and, in reality, may shield the targets from flame radiation. ➢ The entire drywell volume is filled with smoke in 120 s after fire ignition. Gas temperatures in the fire region may reach 200 °C during the first 7 min; following burnout of the spill, the average temperature in the drywell stabilizes to ~80 °C. The simulation results indicate that conditions during the first 30 min of the scenario are not tenable for manual intervention in the fire scene and may remain so for a longer period.
Fire from a horizontal cable tray stack in containment sector 2	➢ Fire originating from sector 2 damages only the cable trays in the same sector that are closer to the ignition source. The wall temperature and incident heat flux on cable trays that are not in the vicinity of fire in sector 2 are below the damage thresholds. ➢ A maximum flame spread of 22.2 m is possible in 1 h from the time of ignition. Horizontal propagation to adjacent sectors can be prevented if the fire is suppressed within the first hour (industry experience shows that most of the fire events in containment are suppressed in 1 h). Vertical propagation to higher elevations is not possible even when multiple cable trays burn in the simulation. This is mainly because the hot gas temperature in the containment does not exceed the damage threshold values for cables. ➢ Most of the safety division cables are either routed in metallic conduits or in enclosed cable trays and are made of non-burnable insulation material, which prevents the ignition of secondary combustibles (similar to the observation in the drywell fire scenario). ➢ The entire containment volume is filled with smoke in 17 min from the onset of fire. The maximum average gas temperature across the domain and the boundaries of adjacent sectors is less than 50 °C. ➢ Based on the simulation results, it is inferred that manual suppression of fire within the first 30 min is deemed not possible due to reduced visibility conditions.

An example of a drywell lube oil fire scenario was presented at the PSAM&ASRAM17 Conference in 2024 in Sendai, Japan [18]. Refer to the conference proceedings for more details.

Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13 show the rendering of fire-generated conditions in various scenarios.

Figure 9. Iso-surface view showing 205 °C temperature contour at the cable tray elevations +9.8 m to +10.8 m in the cabinet room.

Figure 10. Temperature and smoke concentration profile for MCR panel scenario.

Figure 11. HGL profile in switchgear rooms for HEAF scenario.

Figure 12. Layout of the containment FDS model (after geometry transformation) with ignition source in sector 2.

Figure 13. Gas temperature profile in containment at 2160 s (temperature profile ranges from 30 °C to 95 °C).

5. Insights and Applications

The comprehensive Fire PSA conducted at KKL provided critical insights into fire behavior and its implications for plant safety. The integration of advanced fire modeling tools, such as CFAST and FDS, combined with plant-specific data and international best practices, has yielded a high-fidelity representation of fire-induced conditions. These insights are crucial for both operational safety and regulatory compliance. Key findings from the study include the following:

• Improved Understanding of Fire Propagation Dynamics: The detailed simulations revealed that fire propagation is highly dependent on the choice of fire parameters (HRR, CALPOT), spatial configuration of ignition sources, targets, and ventilation conditions. For example, the drywell lube oil fire scenario demonstrated limited horizontal and vertical propagation due to spatial separation and the use of fire-resistant materials.
• Validation of Safety Features: The study confirmed the effectiveness of passive fire protection measures, such as fire doors and cable tray enclosures, in mitigating fire spread and delaying damage to critical components. This validation supports the continued use of these features as part of the plant’s fire safety strategy.
• Enhancement of Risk-Informed Decision-Making: The results of Fire PSA have been integrated into the plant’s overall risk management framework, enabling more informed decisions regarding maintenance, operational strategies, housekeeping of the temporary and transient fire loads, and emergency response planning.
• Support for Regulatory Compliance: The methodology and findings align with Swiss regulatory requirements (ENSI-A05) and international standards, thereby enhancing regulatory acceptance and demonstrating KKL’s commitment to maintaining the highest safety standards.
• Application to Other Safety Studies: The fire modeling results have been utilized to refine deterministic fire analysis and to support risk-informed applications, including the optimization of fire protection systems and emergency response procedures.

These insights underscore the importance of a tailored, plant-specific approach to fire risk assessment in nuclear power plants. By leveraging advanced tools and methodologies, KKL Fire PSA has aimed to set a benchmark for future studies in the field.

6. Future Work and Trends

The following key areas offer valuable opportunities for further research and development to strengthen fire safety and improve the realism of fire risk assessment for nuclear power plants:

• Advancing and Refining Fire PSA Methods: Building on the solid technical foundation of NUREG/CR-6850 [4,5] and its associated guidelines [7,8,9,10,11,12,13,14,15,16,17], there are promising opportunities for further enhancement, particularly in areas such as multi-compartment analysis, treatment of I&C cable failure modes, multiple-spurious operations, and evaluation of fire-induced failure of exposed steel structures. Targeted research and development in these topics can continue to increase the robustness and reliability of Fire PSA.
• Implementing Component-Based Fire Ignition Frequency: The industry-wide implementation of Fire PSAs in recent decades has resulted in the availability of detailed fixed ignition source component counts, making it possible to calculate fire ignition frequencies directly at the component level rather than relying on the traditional plantwide apportionment method, based on the number of components per bin per plant. With the component-based methodology for fire ignition frequency calculation recently developed, the next critical step is its systematic implementation across the industry. This methodological shift will significantly enhance the accuracy of fire risk assessments, allow for more precise quantification of risk, and provide a flexible framework for efficiently updating fire risk models as plant configurations change.
• Reducing Conservatisms in Fire-Induced Cable Failures: Modeling of circuit failures continues to be an area of conservatism in most Fire PSAs. As a result, current Fire PSAs often assume worst-case failure modes for these cables, depending on how they are represented in PSA. Developing practical guidance and more differentiated models—similar to those used for power cables—offers the potential for more accurate and less pessimistic risk assessments.
• Expanding Fire Modeling Capabilities: Future studies could explore the use of hybrid modeling approaches that combine zone models (e.g., CFAST) and computational fluid dynamics (CFD) models (e.g., FDS) to achieve a more efficient and comprehensive understanding of fire dynamics in complex plant geometries.
• Enhancing Realism in Fire Simulations: Expanding current guidelines to include practical approaches for modeling under-ventilated fires, battery fires, cable shaft fires, and MCR abandonment scenarios can further improve model accuracy. Compiling the Fire PSA implementation experience from various licensees and supplementing existing methodologies with insights from practical applications would be valuable. Additional areas of interest include guidance on crediting in-cabinet fire detection or early-warning systems, and accounting for the dynamic effects of fire suppression systems on fire development, rather than relying on a binary assumption (i.e., suppression activated or not). Addressing these aspects would significantly enhance the realism of fire modeling.
• Developing more Realistic Damage Criteria: Unlike cables, the current damage criteria for various fixed equipment are based on conservative assumptions and empirical data. Further experimental research can help to establish more realistic damage criteria, particularly for modern materials and components used in nuclear power plants, such as digital I&C components.
• Assessment of Emerging Fire Hazards: As nuclear power plants adopt new technologies and materials, proactively evaluating the associated fire risks—such as those related to increased use of digital instrumentation and control systems—will support forward-looking fire safety strategies.
• Enhancing Collaboration and Knowledge Sharing: Ongoing collaboration between international regulatory bodies, research institutions, and industry stakeholders is crucial for the development of standardized methodologies and the sharing of best practices in fire risk assessment.

By addressing future-oriented topics, ongoing research can build on the foundation established by the KKL Fire PSA, contributing to the continuous improvement of fire safety practices within the nuclear industry.

7. Conclusions

The bespoke fire modeling methodology developed for KKL underscores the critical importance of integrating plant-specific data, state-of-the-art tools, and internationally recognized guidelines to achieve realistic, efficient, and plant-specific fire risk assessments. This tailored approach has ensured consistency, quality, and traceability throughout the study, yielding valuable insights into fire behavior and its potential impact on plant safety. By aligning closely with KKL’s unique configuration and operational conditions, the methodology not only enhances the safety and reliability of the plant but also ensures compliance with regulatory standards and contributes to the continuous improvement of fire safety practices.

The KKL-specific fire modeling methodology effectively utilized advanced methods and research frameworks, including NUREG and EPRI guidelines, combined with pragmatic assumptions and detailed plant-specific solutions. Key aspects of the methodology—such as the characterization of fire parameters (e.g., heat release rate, calorific potential, fire duration), scenario-specific modeling considerations (e.g., spatial details of equipment, cables, cable trays, and structures; fire location; and ventilation conditions), and the selection of appropriate modeling tools (e.g., based on predictive parameters, scenario objectives, and computational efficiency)—were meticulously tailored to reflect the unique design and operational features of KKL. This alignment ensured realistic fire conditions for each scenario, which were subsequently used to refine the scenario consequences in the Fire PSA study.

The implementation of this methodology within KKL’s internal Fire PSA framework has delivered significant cost and safety benefits. It has optimized the time and resources required for fire simulations. Moreover, the simulation results have proven to be a realistic and valuable resource for other safety studies and risk-informed applications, further supporting the plant’s commitment to maintaining the highest standards of safety and operational excellence.

In conclusion, the KKL fire modeling methodology represents a robust, efficient, and forward-looking approach to fire risk assessment in nuclear power plants. It serves as a model for the development of tailored methodologies that prioritize plant-specific needs while ensuring compliance with global safety standards and promoting continuous improvement in fire safety management.