Nuclear Power Plant Risk Research Articles

Computationally efficient Complete-Sampling-based fault tree analyses for seismic probabilistic safety assessment of nuclear facilities

The seismic risk of nuclear power plants (NPPs) can be assessed through seismic probabilistic safety assessment (PSA) based on fault tree analyses (FT). The conventional FT-based approach has limitations for handling the correlated relation of basic events and utilizing the sampling-based FT analysis for addressing such issues demands considerable computational cost. In particular, accurate consideration of the partial dependences between the failure probabilities of NPP components in seismic hazard events can further increase the computational cost of risk quantification. Therefore, this paper presents an improved complete-sampling-based seismic PSA risk quantification method. The novelty of this method is that it increases the efficiency of the risk quantification by introducing sequential sampling and truncation to the conventional fault tree analysis. In addition, as sub-concepts, we eliminate unrequired sampling extraction steps via integrated multivariate response distribution-based sampling for components and continual evaluation of the convergence of system failure probabilities. On three considered examples, including an actual NPP, the proposed method requires 91%–96% fewer samples than the conventional method, without sacrificing accuracy. These results indicate the efficiency of the proposed method, which is expected to become a useful tool in the future.

Environmental risk assessment and management of nuclear power plants based on big data analysis

In traditional research, monitoring data and samples are limited, and it is difficult to achieve ideal results in real-time monitoring and rapid response to environmental risks. By leveraging extensive environmental data gathered from nuclear power plants, the research employed machine learning methodologies for accurate feature selection and extraction of environmental parameters. An efficient environmental risk assessment model was successfully established by using a random forest algorithm. The 95% confidence interval for the area under the curve value spanned from 0.6894 to 0.9292. This provided a more dynamic and effective means for assessing and managing the environmental risks of nuclear power plants.

A time-reliability correlation for estimating the diagnosis error probability of a nuclear power plant with up-to-date Human–Machine interfaces

Despite being developed more than four decades ago based on expert judgment, the THERP time-reliability correlation (TRC) remains widely employed for calculating diagnosis human error probabilities in human reliability analysis for nuclear power plant risk assessment. However, with numerous advancements in nuclear plant equipment and operations, as well as the emergence of plants featuring advanced interfaces, there's a growing need to validate the THERP TRC. The objective of this study is to establish a TRC for the diagnosis human error probability in a modern reference nuclear power plant equipped with up-to-date human–machine interfaces and compare it with the median of the THERP TRC. To achieve this goal, we devised a method to gather event diagnosis times from a simulator and developed procedures to derive diagnosis TRCs using this data. Our findings indicate that while the median of the THERP TRC offers a conservative diagnosis human error probability for up to 25 min, it becomes overly optimistic beyond this threshold.

Marine ecological restoration strategies of coastal nuclear power plants in China

Low-carbon and green development model is the choice for responding to climate change. Under the dilemma between ever-growing energy demand and the associated environmental issues, nuclear power is one of the low-carbon and high-efficient energy options which compared to wind power and solar power. It is the only stable power source that can replace fossil energy on a large scale at present and even in the future. This paper briefly describes the operational risks of coastal nuclear power plants (CNPPs) and the upgrade process of nuclear reactors, systematically analyzes and summarizes the ecological environment problems, risk problems and ecological restoration and protection strategies in the construction and operation of CNPPs, including coastal wetland ecological environment, marine ecological environment, and the risk prevention measures for cooling water source. The results show that ecosystem restoration by human does not make up for the weakening of the ecological carrying capacity and ecosystem away from its baseline, nonetheless, the restoration of important wetland habitats is conducive to the improvement of the functions of key marine ecosystems and the maintenance of marine biodiversity.

Comparison of event tree/fault tree and convolution approaches in calculating station blackout risk in a nuclear power plant

Station blackout (SBO) risk is one of the most significant contributors to nuclear power plant risk. In this paper, the sequence probability formulas derived by the convolution approach are compared with those derived by the conventional event tree/fault tree (ET/FT) approach for the SBO situation in which emergency diesel generators fail to start. The comparison identifies what makes the ET/FT approach more conservative and raises the issue regarding the mission time of a turbine-driven auxiliary feedwater pump (TDP), which suggests a possible modeling improvement in the ET/FT approach. Monte Carlo simulations with up-to-date component reliability data validate the convolution approach. The sequence probability of an alternative alternating current diesel generator (AAC DG) failing to start and the TDP failing to operate owing to battery depletion contributes most to the SBO risk. The probability overestimation of the scenario in which the AAC DG fails to run and the TDP fails to operate owing to battery depletion contributes most to the SBO risk overestimation determined by the ET/FT approach. The modification of the TDP mission time renders the sequence probabilities determined by the ET/FT approach more consistent with those determined by the convolution approach.

A dynamic probabilistic risk assessment platform for nuclear power plants under single and concurrent transients

ABSTRACT Dynamic probabilistic risk assessment (DPRA) of nuclear power plants (NPPs) has become one of the most critical research areas, especially in the aftermath of the 2011 Fukushima Daiichi nuclear accident. Uncertainty in NPP behavior is key when considering its safety under different operating conditions. Such uncertainty typically results from operation parameters, system conditions, and modeling assumptions. This study integrates the system dynamics (SD) modelling approach with an uncertainty analysis method to quantify the dynamic probabilistic risk in NPPs. To demonstrate the approach’s applicability, the average fuel temperature is used to estimate the probability of reactor core damage under different transients, representing perturbations in reactivity and steam valve coefficient. A Monte Carlo simulation is employed to investigate the effect of uncertainties associated with the different model parameters. A global sensitivity analysis demonstrates that the total delayed neutron fraction, the heat transfer coefficient from fuel to coolant, the coolant temperature coefficient of reactivity, and the fuel temperature coefficient of reactivity are the primary controllers of the plant response variability under the transients considered. In summary, the integration of SD modelling and uncertainty analysis presents an effective DPRA approach that overcomes the limitations of static counterparts while minimizing the computational resources required.

Discrete-Time Bayesian Networks Applied to Flexible Coping Strategies of Nuclear Power Plant Systems

The Fukushima Daiichi accident prompted the nuclear community to find a new solution to reduce the risk in nuclear power plants (NPPs) due to beyond-design-basis external events (BDBEEs). An implementation guide for diverse and flexible coping strategies (FLEX) has been presented by the Nuclear Energy Institute to manage the challenges of BDBEEs and enhance reactor safety. Due to the uniqueness of the FLEX systems, these systems can potentially carry dependencies among components not commonly modeled in NPPs. In this study, we investigate the effectiveness and applicability of both Bayesian networks (BNs) and discrete-time Bayesian networks in the reliability analysis of FLEX equipment. The study compares BNs with two other reliability assessment methods: fault tree and Markov chain. These methods are also shown to be capable of mapping into BNs to perform a reliability analysis of FLEX systems. A neutral dependency algorithm is used to simplify the conditional probability tables and reduce the complexity of the BNs. The results indicate that BNs are not only a powerful method for modeling FLEX strategies but are also effective techniques for inclusion of the dynamics of FLEX equipment in probabilistic risk analysis.

The Analysis of the World's Most Dangerous Nuclear Plant from the Standpoint of the Most Air-Polluted City in Europe

As a prominent item on the agenda of Turkish-Armenian relations from time to time, the Metsamor Nuclear Power Plant, located in the town of Metsamor in Armenia, has become an even more critical issue after the earthquakes in and around the region. The closest settlement to this nuclear power plant in the region is Iğdır. The people of Iğdır, who have current concerns about the power plant, have a different point of view towards the power plant. This study has examined the working principles and risks of nuclear power plants -which are usually seen as a clean energy source- and tried to clarify the misconceptions. Iğdır Province was the most air-polluted city in Europe in 2021. This situation has also caused the local people to establish a relationship between air pollution and the Metsamor Nuclear Power Plant. In particular, cancer cases caused by air pollution have been attributed to Metsamor, which does not contribute to air pollution; and this power plant has been seen as a cause of air pollution. This study explains primary reason for this misconception is the relevant public institutions' insufficient information to society despite the deputies' parliamentary questions submitted to the Grand National Assembly of Turkey.

Exhaustive simulation approach for severe accident risk in nuclear power plants: OPR-1000 full-power internal events

The estimation of severe accident risks in nuclear power plants (NPPs) is significant to confirm the safety level of NPPs and to reinforce their vulnerable points. Severe accident risks can be quantified by Level 2 probabilistic safety assessment (PSA), which traditionally applies a grouping feature to handle the tremendous numbers of accident scenarios. Accordingly, risk information is likely to be lost in the process of grouping similar scenarios and in the treatment of many scenarios with one representative scenario. To obtain more comprehensive risk information including source term behaviors and plant responses during severe accidents, this study suggests an exhaustive simulation approach with new software that helps to automatically generate a large number of input data for an accident simulation code, and performs an application study using PSA models from OPR-1000 full-power internal events. Only a three-day run time was required to simulate all the severe accident scenarios, totaling 690 scenarios, using a commercial computer. The application study revealed that the conventional grouping approach can either underestimate or overestimate overall NPP risk depending on the selection of the representative scenario.

Development of a leading simulator/trailing simulator methodology as part of an integrated safety-security analysis for nuclear power plants

Nuclear power plant (NPP) risk assessment is broadly separated into disciplines of nuclear safety, security, and safeguards. Different analysis methods and computer models have been constructed to analyze each of these as separate disciplines. However, due to the complexity of NPP systems, there are risks that can span all these disciplines and require consideration of safety-security (2S) interactions which allows a more complete understanding of the relationship among these risks. A novel leading simulator/trailing simulator (LS/TS) method is introduced to integrate multiple generic safety and security computer models into a single, holistic 2S analysis. A case study is performed using this novel method to determine its effectiveness. The case study shows that the LS/TS method avoided introducing errors in simulation, compared to the same scenario performed without the LS/TS method. A second case study is then used to illustrate an integrated 2S analysis which shows that different levels of damage to vital equipment from sabotage at a NPP can affect accident evolution by several hours.

Probabilistic deep learning model as a tool for supporting the fast simulation of a thermal–hydraulic code

Following the Fukushima Daiichi accident, enhancing the safety of nuclear power plants has become the priority mission for the future of nuclear energy. Probabilistic safety assessment (PSA) is a well-known technique to quantify the anticipated risk of nuclear power plants according to the accident scenario. One approach to strengthening nuclear safety is reducing the uncertainty in PSA by analyzing a wide spectrum of accident scenarios. In doing this, massive simulations of thermal–hydraulic (TH) dynamics are required by running TH code. As the computation time for such large-scale simulation is a heavy burden, it is necessary to develop a fast simulation model, but related research has recently begun. For doing this, conditional autoencoder was firstly introduced, however, prediction accuracy can be further improved by exploiting temporal characteristics of simulation data. In this paper, we formalize deep learning-based fast simulation model of TH code, and propose a novel deep learning model, namely ensemble quantile recurrent neural network (eQRNN). By leveraging bi-directional long short-term memory, concatenative positional encoding, quantile regression, and model ensemble, the proposed eQRNN can provide a more accurate prediction on the simulation results and uncertain boundaries of its prediction. Compared to the base RNN model, the proposed eQRNN shows 39% and 28% lower error overall in terms of mean absolute percentage error (MAPE) and mean squared error (MSE). Finally, eQRNN achieved 75%, 79% lower MAPE and MSE than the existing conditional autoencoder-based model.

Estimating the frequency of cyber threats to nuclear power plants based on operating experience analysis

The necessity and importance of cyber risk assessments at nuclear power plants is increasingly being recognized in recent times. However, the cyber risk assessment methods developed thus far have focused on engineering evaluations and expert judgments. Moreover, there is no available probabilistic database of cyber threats for nuclear power plants. In this study the frequencies of cyber threats were estimated based on operating experience analysis. The operating experiences are grouped by their characteristics to suggest the list of cyber threats, and the frequency of each cyber threat is calculated by two-stage Bayesian updates. This research is significant because it is the first study on a probabilistic approach to estimate cyber threats. Furthermore, this research offers the advantage of being able to update the threat frequencies with additional data accumulation. We expect that this approach can also be applied to probabilistic safety assessment, which is the most widely used method for nuclear power plant risk assessment.

Comparisons of human reliability data between analog and digital environments

It is obvious that the management of operational risk for nuclear power plants (NPPs) is critical for their sustainability. In this regard, it should be stressed that the establishment of digital main control rooms (MCRs) in NPPs is gradually increasing these days, which are distinguished from analog MCRs equipped with conventional devices. Since the operational conditions of human operators working in digital MCRs are entirely different from those of analog MCRs, human reliability data should be clarified from the viewpoint of the raw data to be included in the related probabilistic safety assessment (PSA). In this study, two sets of human reliability data collected from both analog and digital MCRs were compared with respect to 21 basic task types that are observable during the performance of proceduralized tasks (e.g., emergency operating procedures). As a result, it seems that human error probabilities (HEPs) of the digital MCR are lower than those of the analog MCR overall. However, since certain digital MCR HEPs are higher for several particular task types, uncertainty in PSA could be enlarged if human reliability data secured from an analog MCR are directly used to conduct the PSA of a digital MCR. This paper is an extended version of the paper submitted to the ESREL 2020-PSAM15 conference.

Theoretical approach for uncertainty quantification in probabilistic safety assessment using sum of lognormal random variables

Probabilistic safety assessment is widely used to quantify the risks of nuclear power plants and their uncertainties. When the lognormal distribution describes the uncertainties of basic events, the uncertainty of the top event in a fault tree is approximated with the sum of lognormal random variables after minimal cutsets are obtained, and rare-event approximation is applied. As handling complicated analytic expressions for the sum of lognormal random variables is challenging, several approximation methods, especially Monte Carlo simulation, are widely used in practice for uncertainty analysis. In this study, a theoretical approach for analyzing the sum of lognormal random variables using an efficient numerical integration method is proposed for uncertainty analysis in probability safety assessments. The change of variables from correlated random variables with a complicated region of integration to independent random variables with a unit hypercube region of integration is applied to obtain an efficient numerical integration. The theoretical advantages of the proposed method over other approximation methods are shown through a benchmark problem. The proposed method provides an accurate and efficient approach to calculate the uncertainty of the top event in probabilistic safety assessment when the uncertainties of basic events are described with lognormal random variables.

Research and Application of Standardized Technology for Physical Characteristics Risk Monitoring and Assessment of Nuclear Power Plant

Environmental risks generally exist in various production activities. Based on existing research foundation, in order to solve the related problems in the field of environmental risk of nuclear power plants, this paper conducts in-depth research and discussion on the monitoring and assessment of environmental risks of nuclear power plants. This paper proposes a set of standardized technical framework, based on this framework, an application platform is designed and implemented and applied to the environmental risk monitoring, assessment, and management of nuclear power plant.

Systematic approach and mathematical development for conditional core damage probabilities under station blackout of a nuclear power plant

Station blackout (SBO), a total loss of alternating current (AC) power, is considered to contribute significantly to nuclear power plant risk. The analysis of SBO risk is complicated due to time-dependent interactions among events. Simplifying assumptions that are currently used extensively in probabilistic safety assessment may lead to an overly conservative estimation of SBO risk, which may distort the overall plant risk profile and the associated accident management strategy. To calculate more accurate SBO risk, this study first identified SBO sequences and the associated time-dependent interactions by differentiating fail-to-start and fail-to-run failures of alternative AC power source and turbine-driven pumps. Mathematical formulas were developed to calculate the conditional core damage probabilities corresponding to the identified SBO sequences. These mathematical formulas were verified via a Monte Carlo simulation. This study is expected to contribute to the development of an integrated framework for SBO risk analysis that includes all possible SBO sequences, related events, and their time-dependent interactions.

Study on Probabilistic Safety Goals for Multimodule High-Temperature Gas-Cooled Reactor Based on Chinese Societal Risks

Nuclear safety goal is the basic standard for limiting the operational risks of nuclear power plants. The statistics of societal risks are the basis for nuclear safety goals. Core damage frequency (CDF) and large early release frequency (LERF) are typical probabilistic safety goals that are used in the regulation of water-cooled reactors currently. In fact, Chinese current probabilistic safety goals refer to the Nuclear Regulatory Commission (NRC) and the International Atomic Energy Agency (IAEA), and they are not based on Chinese societal risks. And the CDF and LERF proposed for water reactor are not suitable for high-temperature gas-cooled reactors (HTGR), because the design of HTGR is very different from that of water reactor. And current nuclear safety goals are established for single reactor rather than unit or site. Therefore, in this paper, the development of the safety goal of NRC was investigated firstly; then, the societal risks in China were investigated in order to establish the correlation between the probabilistic safety goal of multimodule HTGR and Chinese societal risks. In the end, some other matters about multireactor site were discussed in detail.

Probabilistic Non-Suppression Model of Electrical Fire in Nuclear Power Plant Using Maximum Likelihood Estimation Method

Electrical fire is the most significant fire-induced risk in nuclear power plants. In this study, we developed a probabilistic non-suppression model of electric fire by estimating the relevant parameters using the Maximum Likelihood Estimation (MLE) method to secure and improve the realism in Fire PRA in application to the Risk Informed Fire Protection Program. Based on the results of fitting 32 continuous probability distributions into fire incident data, 3-Parameter Weibull probability model was selected as the best fitted and adequate model and was compared with the existing exponential probability model. The comparison shows that the level of model adequacy is improved with decreases in BIC by 17.7%, in RSS by 24.4%, and in MSE by 58.2%. The 3-Parameter Weibull probability model drawn in this study is expected to contribute to the enhancement of the Fire PRA realism for supporting risk-based decision by reflecting real fire event experiences.

Electrical fire simulation in control room of an AGN reactor

Fire protection is one of important issues to ensure safety and reduce risks of nuclear power plants (NPPs). While robust programs to shut down commercial reactors in any fires have been successfully maintained, the concept and associated regulatory requirements are constantly changing or strengthening by lessons learned from operating experiences and information all over the world. As part of this context, it is necessary not only to establish specific fire hazard assessment methods reflecting the characteristics of research reactors and educational reactors but also to make decisions based on advancement encompassing numerical analyses and experiments. The objectives of this study are to address fire simulation in the control room of an educational reactor and to discuss integrity of digital console in charge of main operation as well as analysis results through comparison. Three electrical fire scenarios were postulated and twenty-four thermal analyses were carried out taking into account two turbulence models, two cable materials and two ventilation conditions. Twelve supplementary thermal analyses and six subsequent structural analyses were also conducted for further examination on the temperature and heat flux of cable and von Mises stress of digital console, respectively. As consequences, effects of each parameter were quantified in detail and future applicability was briefly discussed. On the whole, higher profiles were obtained when Deardorff turbulence model was employed or polyvinyl chloride material and larger ventilation condition were considered. All the maximum values considered in this study met the allowable criteria so that safety action seems available by sustained integrity of the cable linked to digital console within operators’ reaction time of 300 s.

A Comparison of Response to Potential Risk of Nuclear Power Plants in Major Countries: Focusing on Whistleblowing System

2011년 후쿠시마 사고 이후 한국에서도 원전의 안정성에 대한 우려가 높아지고 있다. 실제로 2016년 경주와 2017년 포항에서 발생했던 지진으로 인해 자연재해 관련 사고의 가능성도 높아진 상황이다. 게다가 2012년 고리 원전의 정전 사고은폐뿐만 아니라 2013년까지 계속된 위조부품 논란으로 인해 잠재적 위험요인의 통제가 중요한 사회적 이슈로 등장하게 되었다. 이에 본 논문은 원자력 관련 부정·부패를 잠재적 위험요인이라는 관점에서 주요국의 대응 체계를 비교하였다. 구체적으로는 한국을 포함한 원전 보유국의 원자력 안전 관련 내부고발제에 대해 살펴볼 수 있었다. 연구 결과, 일본은 내부고발제를 운영하고는 있지만 집행 실적 등이 제대로 공개되지 않고 있었다. 반면에 유럽의 경우에는 영국과 프랑스처럼 국가별로 자국의 상황에 맞게 다양한 방식으로 내부고발제를 운영되는 것으로 확인되었다. 그렇지만 이들 원전 보유국 가운데 내부고발제가 체계적으로 운영되고 있을 뿐만 아니라, 정보 공개가 가장 잘 이루어지고 있는 나라는 미국이었다. 반면에 한국에서는 독립 규제기관으로서 원자력안전위원회가 설립된 것도 후쿠시마 사고 이후였기 때문에, 역사와 경험이 짧은 한계를 지니고 있었다. 게다가 미국과 비교했을 때, 내부고발 관련 현황, 처리 실적, 주요 사건에 대한 보고, 정보 공개 등에서 상당한 보완이 필요한 상황이다. 그리고 유관 부처인 노동부 및 사법부와의 협력체계 구축도 미흡한 수준이었다. 결론적으로는 한국에서도 이러한 잠재적 위험요인을 해소할 수 있는 내부고발 관련 제도 및 문화의 정착이 필요하다는 제안이 이루어질 수 있었다.Since the accident of Fukushima in 2011, Koreans have had concerns on safety related to nuclear power plants. In fact, the probability of severe accidents in Korea has increased owing to two earthquakes of Gyeongju in 2016 and Pohang in 2017. Moreover, there has been a controversy about potential risk after the scandals of the incident concealment in 2012 and counterfeit components in 2013. Therefore, this study attempts to compare the responses to potential risk of nuclear power plants in major countries. Particularly, it focuses on whistleblowing system. As a result of comparison, although Japan introduced whistleblowing system, the performance and related statistics are not properly released. In Europe, the countries such as England and France have developed their own whistleblowing system. Among them, the United States is most exemplary because its whistleblowing system is well organized and information is actively disclosed. On the contrary, Korea has short history and insufficient experience related whistleblowing. In addition, it needs to release performance, statistics, and details on severe incidents. And cooperation with other departments is necessary either. In conclusion, it is suggested that Korea should strengthen whistleblowing system and cultural settlement to solve these potential risk of nuclear power plants.