1. Introduction
Nuclear power generation in South Korea began in 1978 with the commercial operation of the Gori 1 reactor, and currently there are a total of 26 nuclear power plants in operation nationwide. As of 2022, nuclear power accounts for 29.6% of total electricity production and plays an important role in power supply. Nuclear power plants have the advantages of high economic efficiency and low carbon emissions, but ensuring safety and continuous management are essential[1]. Recently, improving the safety of aging nuclear power plants and developing next-generation nuclear power technology have emerged as major issues[2].
In South Korea, digital twin technology is being applied to improve the safety of nuclear power plants, but until now, development has focused mainly on monitoring and measurement in terms of power plant equipment management. However, in order to strengthen the fire safety of aging nuclear power plants, a dedicated fire protection system platform based on digital twins is currently under development[3-5]. A digital twin refers to a virtual replica of an actual physical system created by modeling it in a digital environment, enabling more efficient maintenance, operation, and control in the digital space. Recently, various studies integrating digital twin technology for efficient management of the entire lifecycle of nuclear power plants have been actively conducted overseas as well[6-9].
The fire protection system at nuclear power plants is built based on relevant standards such as fire safety performance and technical standards, Nuclear Safety Commission notices, and electric power industry technical standards. Each system, including detection, notification, suppression, and safety shutdown, is installed independently and operated based on a passive management system[10,11]. The system configuration has structural limitations, including insufficient interconnectivity between components, human-centric inspection and maintenance, and limitations in real-time response. In particular, when system failures or malfunctions occur, the manual management system-centric approach results in highly inefficient management and control. Such inefficiency is particularly unsuitable for nuclear power plants, which are high-risk environments in the event of a fire. Therefore, the development of fire protection system specialized nuclear power plants with digital twins is necessary.
In a digital twin-based fire protection system management and control platform environment, it is necessary to obtain fire characteristic information on combustible materials related to nuclear power plants in order to automatically predict fires without human intervention and quickly activate fire protection systems through an automatic management system. The most commonly mentioned combustible material in nuclear power plant fire compartments is cable insulation material[12], and approximately 41% of fires that occurred in U.S. nuclear power plants from 1995 to 2007 were reported to have been caused by electrical factors[13]. The NUREG reports provided by the U.S. NRC present fire characteristic physical quantities based on criteria such as cable insulation material, cabinet volume, quantity, and openings; however, these values are set overly conservatively, and the application of this information is unclear, making it difficult to appropriately assess fire risks at domestic nuclear power plants. Therefore, to incorporate fire characteristic physical quantities suitable for domestic nuclear power plants into a digital twin-based platform, it is necessary to establish a Korean-type cable and cabinet database.
A hybrid fire detection system combining air-sampling detectors was implemented in the fire protection system. A water mist fire suppression system was introduced to minimize thermal shock to electrical and mechanical systems and emergency power equipment inside the nuclear power plant. Cabinet fire characteristic information was constructed by considering the cabinet's volume, type of combustible materials, and opening locations, while cable fire characteristic information was developed based on the cable's safety rating, shielding type, and intended use. Utilizing these research outcomes, this study aims to enhance the mirroring stage of the digital twin developed for the automatic management system of domestic nuclear power plant fire protection systems to the modeling and simulation stage.
2. Material and Methods
The generally accepted definition of a digital twin is a virtual replica of a physical entity that enables prediction, judgment, and control through various simulations and analyses based on data collected from the actual environment. The application of digital twin technology can vary greatly depending on its purpose. The scope of digital twins extends from simple mirroring to a wide range of applications, including prediction, verification, and interpretation through simulation, as well as control functions. Digital twin technology is expected to evolve from connection-based twins, which connect the physical and virtual worlds, to intelligent twins, and ultimately to autonomous twins. The key functions of digital twins according to their development stages are shown in Figure 1.
As described in Figure 1, even if the technical level of digital twins is defined, it does not necessarily develop sequentially through each stage. Therefore, it is important to select an appropriate reference model according to the purpose of using digital twin technology. The level of digital twin technology established in this study is mirroring, and we aim to develop it into the modeling and simulation stage. The digital twin established for this study is judged to be at the mirroring level. To enhance it to the modeling and simulation level, which is the minimum requirement for introducing an automated management system in nuclear power plants, real-time data integration with actual objects and verification/analysis/optimization using simulation models in a virtual environment must be performed.
2.1 Development of digital-twin for fire protection system in nuclear power plants
The process of constructing a digital twin and processing data for the automatic control of a nuclear power plant's fire protection system is described in Figure 2. First, physical entities that exist in the actual space are selected. Next, the requirements and performance goals for the digital twin to be implemented in the virtual space are defined, and then a system for data collection and processing is designed. Data is collected and processed, and then visualized through 2D or 3D modeling of the digital twin in the virtual space based on the collected data. Simulation models and algorithms suitable for analysis purposes are developed to perform verification, analysis, and optimization in the virtual space. Subsequently, all data deemed abnormal through the monitoring process is transmitted to a cloud- based storage space, and in the event of an abnormal condition, abnormal data is transmitted to the actual object to activate the physical object.
In this study, prior to applying digital twin technology to the fire protection system of domestic nuclear power plants, a scaled-down reference model incorporating a water-based fire suppression system and an air-sampling detection system was created as part of a preliminary technical study. Information on the actual reference model is presented in Figure 3. Item (a) in Figure 3 shows photographs and numerical information of the installation space, while item (b) shows the layout of the water-based fire suppression system installed in that space. Item (c) presents photographs of the currently constructed reference model. Information on the actual objects to be implemented as digital twins in the virtual space of the reference model configured in Figure 3 is summarized in Table 1.
3D modeling was performed on the reference model of the actual space constructed in Figure 3. In the case of the digital twin platform currently under development, 3D modeling using a game engine requires a large amount of memory, resulting in very long response times and waiting times, which makes it difficult to apply to fire protection systems that require rapid management and control. Therefore, considering the industrial application characteristics, an image scanning method with a smaller file size was utilized while maintaining the same level of 3D modeling quality. A performance comparison between the game engine used for visualization and the image scanning method is presented in Table 2. In this study, the platform was constructed using 3D modeling and digital twin creation through multi-angle 2D image capture for digital twin construction.
Using the image model of the image scanning method shown in Table 2, 3D modeling was performed for the reference model described in Figure 3. The results of the 3D modeling of this reference model are shown in Figure 4. As shown in Table 1, Figure 4 shows that the main objects installed in the actual reference model were implemented as 3D models through image scanning.
2.2 Development of fire protection system in nuclear power plants linked digital-twin
This section describes the fire protection system that is linked to the digital twin platform described earlier. In this study, the fire protection system can be divided into two main components: the fire detection and notification system and the fire suppression system.
2.2.1 Development of fire detection and alarm system in nuclear power plants
In this study, the fire detection system linked to the digital twin can be explained as shown in Figure 5. Existing nuclear power plants are equipped with detection systems that can detect fires by recognizing various factors such as flames, smoke, and heat. To add an additional fire detection and alarm system to the existing equipment, an air-sampling fire detection system was developed. The air-sampling detection system was designed with an optimized detection path based on simulations of airflow within the detector. Additionally, based on the results of fire simulations (Temperature, CO, and CO2 concentrations) conducted using fire scenarios presented in the NUREG report, the system was developed to achieve optimal performance by reducing false alarms through the application of fire detection criteria and AND algorithms. Furthermore, when installed in hazardous areas, the system incorporates radiation shielding functionality to prevent malfunctions or failures caused by radiation.
2.2.2 Development of fire suppression system in nuclear power plants
In this study, the fire suppression system linked to the digital twin consists of a head, nozzle, and suppression valve for automatic control, as shown in Figure 6. It is believed that there are various rooms within a nuclear power plant where this system can be applied, such as the main control room, cable spreading room, and emergency diesel generator room. Performance evaluation indicators for water mist fire suppression systems include droplet diameter, spray density, momentum, dedicated piping design, momentum, and spray devices, among others. In this study, the water mist fire suppression system for the relevant room was designed to overcome these issues through the following technical characteristics.
When installing a micro-spray fire suppression system inside a nuclear power plant, stainless steel pipes were used to ensure durability even under high pressure. The water mist fire suppression system meets the NFPA 750 standard of 400 μm or more in droplet diameter and is designed to have a droplet diameter of 300 μm or less to ensure rapid cooling and penetration into surfaces. The appropriate discharge pressure for this system complies with the NFPA 750 standard for low pressure (1.2 MPa or less), and it is designed to enable mist formation and fire suppression even under low flow and low pressure conditions.
2.3 Fire characteristic information of cabinet and cable used nuclear power plants
In the data processing process of the digital twin shown in Figure 2, it is necessary to establish criteria for automatically determining abnormal data such as fires through an automatic management system and automatically activating the fire suppression system at the simulation and algorithm stage. Among these, it is necessary to construct and verify fire characteristic information for combustible materials, which account for the largest proportion of the interior of nuclear power plants. In this study, we constructed a database of fire characteristic information centered on cables and cabinets, which are the most widely distributed combustible materials in nuclear power plants.
2.3.1 Fire characteristic information of cabinet used nuclear power plants
An overview of this study is presented in Figure 7 below. In order to establish fire characteristic information for cabinets used in nuclear power plants, the characteristics of the cabinets were first analyzed based on relevant literature. Considering these various factors, experimental conditions were set based on the purpose, volume, amount of combustible material, opening conditions, and applicable cables of the cabinets. The heat release rate was measured using a calorimeter applying the oxygen consumption method.
The results of measuring fire characteristics showed that the amount of combustible material and the volume of the cabinet did not have a significant impact on the analysis results, while the open conditions were found to have a significant impact. To interpret these results from a computational fluid dynamics perspective, simulation analysis was performed using a fire dynamics simulator (FDS).
A cabinet identical to the experimental conditions was simulated, and an analysis of the peak heat release rate (HRR) was performed based on the open conditions. While it appears that FDS can accurately predict the peak HRR under all simulation conditions, there are differences in the predicted values depending on the ventilation conditions. It is expected that a precise solution will be established through subsequent research.
2.3.2 Fire characteristic information of cable used nuclear power plants
An overview of this study is presented in Figure 8 below. In order to construct a database of fire characteristics for cables used in nuclear power plants, we selected cables currently in use, taking into consideration various variables (certification, application, shielding type, etc.). The thermal decomposition characteristics database was constructed using thermal gravimetric analysis, and the combustion characteristics database was constructed using a cone calorimeter in accordance with ISO 5660-1 standards.
The activation energy for the pyrolysis characteristic database was determined using the Kissinger and Ozawa- Flynn-Wall (OFW) methods based on thermogravimetric analysis (TGA) results. For the combustion characteristic database, the activation energy was derived using the Janssens method, which predicts temperature based on ignition time. To interpret these results from a computational fluid dynamics perspective, simulation analysis was performed using the Fire Dynamics Simulator (FDS).
The experimental environment was simulated, and sensitivity analysis of input parameters for the Simple model and Pyrolysis model was performed. It appears that accurate prediction is difficult due to the inherent limitations of FDS, such as reduced prediction performance for multi-material combustible materials and prediction performance that is dominantly influenced by reaction heat. It is expected that accurate solutions will be established through future follow-up research.
2.4 Advanced strategies of digital-twin specialized fire protection system in nuclear power plants
In order to establish an automatic management system for the fire protection systems in nuclear power plants, a digital twin at the mirroring level, an integrated fire protection system, and fire characteristic information of cabinets and cables used to identify abnormal conditions were developed. To enhance the mirroring-level digital twin to the modeling and simulation stage, measures have been established to integrate fire detection and notification systems, fire suppression systems, and fire characteristic information for cabinets/cables. The contribution of these measures from a fire lifecycle perspective is detailed in Figure 9. Figure 9 is a simple concept, and it is expected that subsequent research to enhance the mirroring-level digital twin to the modeling and simulation level will more accurately derive the contribution methods and data processing methods for each achievement.
3. Results & Discussion
The contents of this study can be summarized as follows:
1) To establish an automatic management system for fire protection in nuclear power plants, a mirroring-level digital twin was developed by constructing a scaled-down reference model and generating its 3D representation.
2) The newly developed fire detection and alarm system for nuclear power plants was designed to achieve optimal performance through simulations of airflow and fire scenarios. By applying a hybrid approach that includes video detection technology, false alarms that could occur in air-sampling detection systems were minimized, and a dedicated fire detection and alarm program compatible with the digital twin platform was implemented. The new fire suppression system was developed as a water mist system to simultaneously induce suffocation and cooling effects and minimize thermal shock for electrical fires, which are the most common type of fire in nuclear power plants. This system is designed to generate micro-water droplets of 400 μm or less in accordance with NFPA standards, and is designed to enable automatic control interconnection with the digital twin platform, including a pressure switch and solenoid valve.
3) A cabinet and cable fire characteristic database was constructed to support fire risk assessments in nuclear power plants by incorporating critical fire parameters. The database was validated and refined using Fire Dynamics Simulator (FDS), a widely adopted fire modeling tool.
This study is confined to the conceptual development of a digital twin-based automated management system for fire protection in nuclear power plants. It demonstrates the feasibility of advancing to the modeling and simulation stage by utilizing newly developed fire protection systems and cabinet/cable fire characteristic data. However, further research is needed to develop detailed implementation strategies and practical applications.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
