The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) is simulation software for modeling the human portion of risk. HUNTER is built on a framework that separately models individual, task, and environmental factors. As a dynamic human reliability analysis (HRA) method, it affords advantages over traditional static or worksheet-based HRA approaches, such as the ability to model what-if scenarios and produce outputs beyond human error probabilities. In this paper, a validation was conducted of HUNTER task time estimates vs. empirical data collected for a steam generator tube rupture scenario. The HUNTER model produced using GOMS-HRA task-level primitives yielded timing results very closely aligned to empirical reference data for base and complex conditions, suggesting the tool holds promise for HRA requiring time estimation.

Dependence analysis refers to a method for adjusting the failure probability for a human action by considering the impact of the preceding human action in human reliability analysis (HRA). Most of the existing dependence analysis methods have been developed based on the approach suggested in the Technique for Human Error-Rate Prediction (THERP) HRA method. However, the THERP-based approaches may have limitations. For instance, it inevitably presents challenges regarding both the subjectivity of expert evaluation as well as experience using dependence for resource-intensive and time-consuming complex analyses. In addition, the THERP-based quantification approaches rarely explain the adjustment of human error probabilities (HEPs), since the quantification processes do not concretely account for the context at the moment when the action is present. For these reasons, the authors’ previous studies have conceptually suggested a dynamic approach to evaluating dependencies by extending the existing performance shaping factor (PSF) concept used for quantifying HEPs in HRA. The conventional PSF modeling methods are limited to separate modeling for each HFE. However, this study suggests PSFs can affect a set of human actions or a multiple HFE rather than a single HFE. With this assumption, the authors simplified the dependency analysis process to efficiently evaluate dependencies and increase the quantification speed with more explicable backgrounds. In this paper, the authors mainly focus on dependency quantification over time. This study utilizes the eight PSFs suggested in the Standardized Plant Analysis Risk-HRA (SPAR-H) method. The mathematical models and logical algorithms are investigated through literature. Integrating the effects of PSFs to generate HEPs with dependency effects is also investigated. Then, the applicability of the dynamic method is investigated based on a design-based accident scenario, steam generator tube rupture. Lastly, insights from this approach are discussed in the paper.

The importance of safety culture has been emphasized to achieve a high level of safety. In this light, a systematic method to more properly deal with safety culture is necessary. Here, a decision-making tool that can apply a graded approach to the analysis of safety culture is proposed, called the F-D matrix, which determines the frequency and the difficulty of safety culture attributes recently defined by the IAEA. A hierarchical model of difficulty contributors was developed as a scoring standard, and its elements were weighted via expert evaluation using the analytic hierarchy process. The frequency of the attributes was derived by analyzing reported events from nuclear power plants in the Republic of Korea. Period-by-period comparisons with the F-D matrix can show trends in the change of the maturity level of an organization's safety culture and help to evaluate the effectiveness of previously implemented measures. In the evaluating the difficulty of the attributes in the recently developed harmonized safety culture model, the difficulties of Trending, Benchmarking, Resilience, and Documentation and Procedures were found to be relatively high, while the difficulties of Conflicts are Resolved, Ownership, Collaboration, and Respect is Evident were found to be relatively low. A case study was conducted with an analysis period of 10 years to attempt to reflect the many changes in safety culture that have been made following the Fukushima accident in March 2011. As a result of comparing two periods following the Fukushima accident, the overall frequency decreased by about 40%, providing evidence for the effects of the various improvements and measures taken following the increased emphasis on safety culture. The proposed F-D matrix provides a new analytical perspective and enables an in-depth analysis of safety culture.

The importance of human and organizational factors in the nuclear industry has been emphasized, and along these lines there have been various improvements to human–system interfaces (HSIs). Operator support systems are one of the key HSIs that can contribute to reduce human errors. The system proposed in this work, called the concealed intelligent assistant (CIA), is designed as an operator support system using artificial intelligence algorithms. The CIA system validates operator actions and notifies operators of human errors to prevent adverse effects on plant integrity. The operation validation system consists of two-step filtering, namely through a procedural compliance check module and a comparison of safety impact evaluation module. Both modules employ a prediction algorithm with deep neural networks, where the former predicts the results of operators’ decision-making, and the latter implements plant parameter prediction strategies. The CIA system minimizes any increase in the cognitive burden on operators because it provides additional information only when a mistake is actually expected to adversely affect the integrity of the power plant.

Recovery human action is defined as the action that prevents deviant conditions from producing unwanted effects. Analyzing recovery actions has been a critical part in human reliability analysis (HRA). However, there are a couple of limitations to treat recovery actions only depending on the current HRA methods. Representatively, the existing recovery analysis does not specifically consider recovery actions as are occurred in actual nuclear power plants (NPPs). The overall goal of this study aims to develop a novel recovery analysis method to account for human action recoveries in context of scenarios as well as complement the limitations of existing recovery analysis. In this paper, the recovery analysis in current HRA methods and their challenges are introduced. A strategy to achieve the goal is introduced with a modified recovery definition. Then, how we have researched the approach will be introduced in the paper.

Safety culture refers to how an organizational culture prioritizes and values safety. Since the safety culture factors consist of a multi-layered organic structure such as employers, managers, and executives, vulnerable factors related to safety culture may differ depending on the organization. Weaknesses in safety culture have proven to contribute significantly to incidents and accidents across different industries, including the nuclear industry [1,2]. This is why it is important for people associated with organizations to understand and adhere to the characteristics of an effective safety culture. To this end, nuclear and nuclear-related organizations have in place the sets of safety culture guidelines. For example, there is a safety culture model developed by Korea Institute of Nuclear Safety (KINS) to inspect the safety culture in Korea [3]. However, the guidelines from various institutions were similar in intent but different in structure. This can create unnecessary complexity and uncertainty in understanding expectations and implementing programs to enhance safety culture. To address these challenges, International Atomic Energy Agency (IAEA) published the Harmonized safety culture model (HSCM) by discussing existing safety culture frameworks to harmonize different sets of guidelines from institutions including the World Association of Nuclear Operators (WANO), and the Institute of Nuclear Power Operations (INPO) as well as regulatory agencies. This model provides a description of the traits and attributes that are present in organizations with an effective culture for safety [4]. These factors, as representing the values and basic assumptions in organizations, can be used to measure the level of safety culture. Therefore, this study aims to develop a database that derived safety culture factors using HSCM for incidents occurred during seventeen years (1994 – 2020) due to lack of safety culture according to the Operational Performance Information System for nuclear power plant (OPIS) of the KINS. Thereafter, it is expected that the database developed in this study can be used to evaluate the different’ vulnerable factors according to the organizations.

There is a growing recognition that organizational factors such as safety culture significantly contribute to the safe operation of nuclear facilities. It was found that safety culture influences human/organizational safety performance, as organizations with higher maturity safety culture have lower accident rates [1-3]. By the necessity of this, various definitions of the concept of safety culture have been proposed according to different viewpoints [4-10]. In the nuclear field, the IAEA has defined safety culture as “That assembly of characteristics and attitudes in organizations and individuals which establishes that, as an overriding priority, nuclear plant safety issues receive the attention warranted by their significance” [6]. The purpose of addressing safety culture is to develop safety outcomes to achieve a higher level of safety. To effectively promote safety culture, an in-depth analysis of safety culture is required, such as to analyze how certain aspects or elements of safety culture (hereinafter referred to as safety culture elements) affect certain aspects of safety, what characteristics of safety culture elements have, and how these characteristics affect safety performance. Characteristics for each safety culture element can be analyzed as a deductive method. In this study, the concept of degree of difficulty is introduced as a generic term for the extent and difficulty of efforts made to meet safety culture principles. By introducing the concept of degree of difficulty for each element of safety culture, it is possible to analyze the safety culture from a decomposable point of view. For example, it is possible to determine whether a safety culture element that is frequently an issue is because it is difficult to comply with related principles. And it can help to establish an effective and appropriate level of response strategy according to the analysis result. In addition, safety culture represents the culture of the organization from a safety point of view, each organization has its own unique characteristics. In other words, the viewpoint on safety culture may differ depending on the characteristics or role of the organization. For example, regulatory agencies and operating agencies may have different views on a safety culture principle. To achieve high nuclear safety at the national level, communication between various organizations such as operating agencies and regulatory agencies is essential. However, if each organization has a different view on which of the various elements constituting the safety culture is an obstacle to achieving a high level of safety culture, it will be difficult to gather consensus in establishing appropriate strategies. It is required to understand at what point their views differ on safety culture. The introduced method can reveal the safety culture factors in which the differences in viewpoints are large and will help to promote mutual understanding between organizations.

Operating procedures are strictly followed in nuclear power plant operation. However, under a highly stressful condition such as emergency operation, human error probability can increase, with operators making mistakes in complying with the complex operating procedures. This paper proposes a procedure compliance check (PCC) system to monitor operator action and detect procedural deviation. If an operator action does not match the related procedural instruction, the PCC system notifies the operator in order to help them to recognize the mistake. A procedural logic process is constructed by referring to colored Petri nets. In situations requiring complex decisions, the PCC system employs a deep learning algorithm to predict operator judgement. The system was tested with data from a compact nuclear simulator, and demonstrated its potential to detect procedural noncompliance.

Nuclear Safety Culture is “that assembly of characteristics and attitudes in organizations and individuals which establishes that, as an overriding priority, nuclear power plant safety issues receive the attention warranted by their significance.” This term first appeared in the report of the IAEA after the Chernobyl accident in 1986. With this accident, the importance of safety culture has been paid attention and international consultations have been sought to promote nuclear safety culture.

Human operators always have the possibility to commit human errors, and in safety-critical infrastructures such as a nuclear power plant, human error could cause serious consequences. Since nuclear plant operations involve highly complex and mentally taxing activities, especially in emergency situations, it is important to detect human errors to maintain plant safety. This work proposes a method to predict the future trends of important plant parameters to determine whether a performed action is an error or not. To achieve this prediction, a recursive strategy is adopted that employs an artificial neural network as its prediction model. Two artificial neural networks were selected and compared: multilayer perceptron and long short-term memory (LSTM). Model training was accomplished using emergency operation data from a nuclear power plant simulator. From the comparison results, it was observed that the future trends of plant parameters were quite accurately predicted through the LSTM model. It is expected that the plant parameter prediction function proposed in this work can give useful information for detecting and recovering human errors.

Nowadays, automation has been generalized with artificial intelligences in many areas. In nuclear power plants, some features which have simple logics in nuclear power plants such as reactor trip and engineered safety features (ESFs) actuation have been automated, whereas, other components have not been automated yet, so human operators are still necessary to control the reactor in emergency or abnormal situations. However, there exists a risk of human errors since human operators are involved in nuclear power operations. That is because, human error may contribute to the risk of severe accidents. To reduce those human errors, moreover, to draw to extend the portion of automation in nuclear power plants, a framework which automatically detects Unsafe Acts (UAs) which are occurred in advanced main control rooms of nuclear power plants has been introduced. Human operators are supposed to operate nuclear power plants by following operating procedures. However, in real operational situation, they violate operating procedures sometimes to achieve the goal (to keep the plant integrity) based on their own experiences and their know-hows. Critical safety functions (CSFs) can disentangle whether an operator's action will adversely affect plant integrity. Thus, the UA autodetection system considers both procedure violation and CSFs violation to find out errors made by human operator.