PSAM 16 - Session M13 Overview

PSAM 16 Conference Session M13 Overview

Session Chair: Jeffrey Julius (jjulius@jensenhughes.com)

Paper 1 MI240

Lead Author: Michelle Kichline Co-author(s): Jing Xing, Jing.Xing@nrc.gov James Chang, James.Chang@nrc.gov

Dependency Analysis Using the Integrated Human Event Analysis System Human Reliability Analysis Methodology
Dependency in the context of human reliability analysis (HRA) refers to the impact of success or failure of a human action on performance of subsequent human actions. Existing dependency models assess the level of dependency between two consecutive human failure events (HFEs) based on the coupling factors or commonalities that exist for both HFEs. The U.S. Nuclear Regulatory Commission (NRC) developed a new dependency model that is informed by behavioral and cognitive science and expands on existing dependency models by identifying the specific cognitive failure modes (CFMs), performance influencing factors (PIFs), and PIF attributes that are impacted by dependency. This new dependency model identifies and evaluates how failure of the first human action affects the context of subsequent human actions. The NRC presents the model in NUREG-2198, “The General Methodology of an Integrated Human Event Analysis System (IDHEAS-G),” issued May 2021. The NRC staff developed IDHEAS-G as a new HRA methodology for agency use. IDHEAS-G is a new general HRA methodology that can be used to develop application-specific HRA methods. The NRC used the IDHEAS-G framework to develop a new method for performing HRA for nuclear power plant probabilistic risk assessments, which is documented in Research Information Letter 2020-02, “Integrated Human Event Analysis System for Event and Condition Assessment (IDHEAS-ECA),” issued February 2020. Research Information Letter (RIL) 2021-14, “Integrated Human Event Analysis System Dependency Analysis Guidance (IDHEAS-DEP)” documents how to perform a dependency analysis using the dependency model that was developed as part of IDHEAS-G to be used with and the IDHEAS-ECA HRA method. This paper summarizes how to perform dependency analysis using IDHEAS-DEP.

Dependency Analysis Using the Integrated Human Event Analysis System Human Reliability Analysis Methodology

Dependency in the context of human reliability analysis (HRA) refers to the impact of success or failure of a human action on performance of subsequent human actions. Existing dependency models assess the level of dependency between two consecutive human failure events (HFEs) based on the coupling factors or commonalities that exist for both HFEs. The U.S. Nuclear Regulatory Commission (NRC) developed a new dependency model that is informed by behavioral and cognitive science and expands on existing dependency models by identifying the specific cognitive failure modes (CFMs), performance influencing factors (PIFs), and PIF attributes that are impacted by dependency. This new dependency model identifies and evaluates how failure of the first human action affects the context of subsequent human actions. The NRC presents the model in NUREG-2198, “The General Methodology of an Integrated Human Event Analysis System (IDHEAS-G),” issued May 2021. The NRC staff developed IDHEAS-G as a new HRA methodology for agency use. IDHEAS-G is a new general HRA methodology that can be used to develop application-specific HRA methods. The NRC used the IDHEAS-G framework to develop a new method for performing HRA for nuclear power plant probabilistic risk assessments, which is documented in Research Information Letter 2020-02, “Integrated Human Event Analysis System for Event and Condition Assessment (IDHEAS-ECA),” issued February 2020. Research Information Letter (RIL) 2021-14, “Integrated Human Event Analysis System Dependency Analysis Guidance (IDHEAS-DEP)” documents how to perform a dependency analysis using the dependency model that was developed as part of IDHEAS-G to be used with and the IDHEAS-ECA HRA method. This paper summarizes how to perform dependency analysis using IDHEAS-DEP.

Paper MI240 | Download the paper file. | Download the presentation PowerPoint file.

Name: Michelle Kichline (Michelle.Kichline@nrc.gov)

Bio: Michelle Kichline is a Senior Reliability and Risk Analyst in the Division of Risk Assessment in the NRC’s Office of Nuclear Reactor Regulation. Ms. Kichline joined the agency in 2010. Prior to joining the NRC, Michelle began her career as a system engineer at boiling water reactor and later earned a Senior Reactor Operator’s license. She holds a Bachelor of Science degree in Chemical Engineering from the Pennsylvannia State University and a Bachelor of Arts degree in Physics from Lycoming College.

Country: USA
Company: U.S. NRC
Job Title: Senior Reliability and Risk Analyst

Paper 2 NO51

Lead Author: Kohei Nonose Co-author(s): Yukihiro Kirimoto, kirimoto@criepi.denken.or.jp Yuko Hirotsu, hirotsu@criepi.denken.or.jp Kuhide Sasou, sasou@criepi.denken.or.jp

Models and Knowhow for Human Reliability Analysis on Portable Equipment
The use of portable equipment such as mobile diesel pumps and power generators may be necessary to respond to severe accidents at nuclear power plants; however, development and application of human reliability analysis (HRA) methods for portable equipment is scarce in the world. Our past study described models and knowhow for HRA on portable equipment we developed and their application example in tsunami probabilistic risk assessment. It showed definitions of types of steps, examples of application rules of the table of estimated human error probability (HEP) in THERP method for the on-site operation/work, and task timeline diagram developed for organizing actors, locations, and time information (e.g., time required for executing a task). The present study has improved them and prepared for additional models and knowhow for HRA on portable equipment, as shown in the followings: 1) example of application rule of Cause-Based Decision Tree Method for the emergency operations facility, 2) re-definition of types of steps, 3) example of application rule of the table of estimated HEP in the THERP method for the on-site operation/work for equipment not covered in the THERP method (especially for error recovery to error of commission), 4) evaluation method for repetitive work such as refueling to portable equipment in long-term response, and 5) evaluation method for transmission/receipt of directions and report. This study will be useful for utilities to conduct HRA on portable equipment in actual plants.

Models and Knowhow for Human Reliability Analysis on Portable Equipment

The use of portable equipment such as mobile diesel pumps and power generators may be necessary to respond to severe accidents at nuclear power plants; however, development and application of human reliability analysis (HRA) methods for portable equipment is scarce in the world. Our past study described models and knowhow for HRA on portable equipment we developed and their application example in tsunami probabilistic risk assessment. It showed definitions of types of steps, examples of application rules of the table of estimated human error probability (HEP) in THERP method for the on-site operation/work, and task timeline diagram developed for organizing actors, locations, and time information (e.g., time required for executing a task). The present study has improved them and prepared for additional models and knowhow for HRA on portable equipment, as shown in the followings: 1) example of application rule of Cause-Based Decision Tree Method for the emergency operations facility, 2) re-definition of types of steps, 3) example of application rule of the table of estimated HEP in the THERP method for the on-site operation/work for equipment not covered in the THERP method (especially for error recovery to error of commission), 4) evaluation method for repetitive work such as refueling to portable equipment in long-term response, and 5) evaluation method for transmission/receipt of directions and report. This study will be useful for utilities to conduct HRA on portable equipment in actual plants.

Paper NO51 | Download the paper file. |

Name: Kohei Nonose (nonose@criepi.denken.or.jp)

Bio: Kohei Nonose is a research scientist in the Nuclear Risk Research Center (NRRC), the Central Research Institute of Electric Power Industry (CRIEPI), where his research focuses on human reliability analysis at nuclear power plants. He received his PhD from the University of Tokyo and subsequently worked as an assistant professor there, while conducting joint research with Imperial College London and ETH Zurich from 2012 to 2014 on measuring air traffic controllers' team situational awareness. After that, he worked as a research scientist at the Human Factors Research Center of CRIEPI. He is currently working at the same institute's NRRC. He is a member of the Atomic Energy Society of Japan (AESJ).

Country: JPN
Company: Central Research Institute of Electric Power Industry
Job Title: Research Scientist

Paper 3 PA192

Lead Author: Vincent Paglioni Co-author(s): Torrey Mortenson (torrey.mortenson@inl.gov) Katrina M. Groth (kgroth@umd.edu)

The human failure event: what is it and what should it be?
The human failure event (HFE) is typically the intersection point between the probabilistic risk assessment (PRA) of a complex engineering system (i.e., nuclear power plant (NPP)) and a separate human reliability analysis (HRA) methodology. In that lens, the HFE represents the total accounting of human error in the safety assessment of NPPs. The HFE is a pivotal aspect of any PRA. Identifying and quantifying the human error probability (HEP), which is the probability associated with an HFE, has typically been the focus of HRA methods in nuclear power. The HFE has typically been the intersection point between the parent PRA and its embedded HRA method. However, “HFE” has not been rigorously defined for either HRA or PRA, and so the entire field of risk analysis lacks a formal definition of what constitutes an HFE . HFE is, of course, a failure of some sort – but at what level of abstraction should the HFE be defined? Is the HFE simply the result of any failed task, or should it represent something larger in scope than a single task or even a set of tasks can define? In this paper, we discuss the need for rigorous definition as a subset of a much broader necessity in HRA for clarity around key terminology. We propose a framework for identifying and defining HFEs at a high level of abstraction. Recognizing that task failures are not the exceedingly rare phenomenon that the notion of a failure event might suggest, we propose that the idea of an HFE must be something more complicated than a single task failure.

The human failure event: what is it and what should it be?

The human failure event (HFE) is typically the intersection point between the probabilistic risk assessment (PRA) of a complex engineering system (i.e., nuclear power plant (NPP)) and a separate human reliability analysis (HRA) methodology. In that lens, the HFE represents the total accounting of human error in the safety assessment of NPPs. The HFE is a pivotal aspect of any PRA. Identifying and quantifying the human error probability (HEP), which is the probability associated with an HFE, has typically been the focus of HRA methods in nuclear power. The HFE has typically been the intersection point between the parent PRA and its embedded HRA method. However, “HFE” has not been rigorously defined for either HRA or PRA, and so the entire field of risk analysis lacks a formal definition of what constitutes an HFE . HFE is, of course, a failure of some sort – but at what level of abstraction should the HFE be defined? Is the HFE simply the result of any failed task, or should it represent something larger in scope than a single task or even a set of tasks can define? In this paper, we discuss the need for rigorous definition as a subset of a much broader necessity in HRA for clarity around key terminology. We propose a framework for identifying and defining HFEs at a high level of abstraction. Recognizing that task failures are not the exceedingly rare phenomenon that the notion of a failure event might suggest, we propose that the idea of an HFE must be something more complicated than a single task failure.

Paper PA192 | Download the paper file. | Download the presentation PowerPoint file.

Name: Vincent Paglioni (paglioni@umd.edu)

Bio: Vincent is a Ph.D. candidate in Reliability Engineering at the University of Maryland. His research, advised by Dr. Katrina Groth, focuses on the conceptualization, modeling and quantification of dependency in human reliability analysis. Vincent's research is examining dependency through a basis in causality and uses causal Bayesian Networks as the architecture for modeling dependency relationships between HRA variables.

Country: USA
Company: University of Maryland, College Park
Job Title: Graduate Research Assistant

Paper 4 YO82

Lead Author: Yochan Kim Co-author(s): Sun Yeong Choi (sychoi@kaeri.re.kr) Jinkyun Park (kshpjk@kaeri.re.kr) Jaewhan Kim (jhkim4@kaeri.re.kr)

Statistical evidence of minimum human error probability for an emergency event from simulation records
Human reliability analysis estimates error probabilities of human operators under given contextual conditions for predicting a probabilistic risk of complex systems. The human error probabilities have been often determined based on the limited operating data and simplified cognitive models; hence, there has been a recognition in the field of HRA that it is necessary to appropriately assign a conservative value to a very low error probability in consideration of various uncertainty factors. For example, EPRI [2010] suggested assigning different minimum values such as 1.0E-04, 1.0E-05, and 1.0E-06 according to the contextual factors by referring to the typical hardware failure probability. Whaley et al. [2011] recommended using 1.0E-5 as a lower bound of HEP with consideration of usefulness in cut-set investigations. However, no objective evidence supporting the reasonableness of the minimum value was not presented. This paper attempts to generate statistical information to determine the minimum error probability bound based on the human error data from the simulation records. The data is generated by observing the licensed crew behaviors in full-scope simulators. This paper provides a basis for the minimum human error probability by estimating the occurrence frequency of human error of a type that has not been analyzed much so far. In this study, we estimated the frequency of equipment operation not prescribed in the procedure during the process of performing tasks according to the procedure. As a result, considering the possibility of recovery, it was predicted that the operator could make a significant human error in the power plant with a probability of 1.0E-05. Even though additional data analysis and verification are essential, this study showed that a piece of statistical evidence for the minimum human error probability can be derived by predicting the human error probability of an unknown type.

Statistical evidence of minimum human error probability for an emergency event from simulation records

Human reliability analysis estimates error probabilities of human operators under given contextual conditions for predicting a probabilistic risk of complex systems. The human error probabilities have been often determined based on the limited operating data and simplified cognitive models; hence, there has been a recognition in the field of HRA that it is necessary to appropriately assign a conservative value to a very low error probability in consideration of various uncertainty factors. For example, EPRI [2010] suggested assigning different minimum values such as 1.0E-04, 1.0E-05, and 1.0E-06 according to the contextual factors by referring to the typical hardware failure probability. Whaley et al. [2011] recommended using 1.0E-5 as a lower bound of HEP with consideration of usefulness in cut-set investigations. However, no objective evidence supporting the reasonableness of the minimum value was not presented. This paper attempts to generate statistical information to determine the minimum error probability bound based on the human error data from the simulation records. The data is generated by observing the licensed crew behaviors in full-scope simulators. This paper provides a basis for the minimum human error probability by estimating the occurrence frequency of human error of a type that has not been analyzed much so far. In this study, we estimated the frequency of equipment operation not prescribed in the procedure during the process of performing tasks according to the procedure. As a result, considering the possibility of recovery, it was predicted that the operator could make a significant human error in the power plant with a probability of 1.0E-05. Even though additional data analysis and verification are essential, this study showed that a piece of statistical evidence for the minimum human error probability can be derived by predicting the human error probability of an unknown type.

Paper YO82 | Download the paper file. | Download the presentation pdf file.

Name: Yochan Kim (yochankim@kaeri.re.kr)

Bio: Yochan Kim is working for Korea Atomic Energy Research Institute as a senior researcher. He received a Ph.D. degree in industrial and systems engineering from the Korea Advanced Institute and Science and Technology, Daejeon, Korea, in 2011. The main interests include Human Reliability Analysis, data science, and human-computer interaction.

Country: KOR
Company: Korea Atomic Energy Research Institute
Job Title: Senior Researcher