PSAM 16 - Session W03 Overview

PSAM 16 Conference Session W03 Overview

Session Chair: Andreas Bye (andreas.bye@ife.no)

Paper 1 RO173

Lead Author: Ronald Boring Co-author(s): Thomas Ulrich thomas.ulrich@inl.gov Jooyoung Park jooyoung.park@inl.gov Jeeyea Ahn jeeyea.ahn@inl.gov Yunyeong Heo yunyeong.heo@inl.gov

The HUNTER Dynamic Human Reliability Analysis Tool: Overview of the Software Framework for Modeling Digital Human Twins
The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) was previously developed as a simplified test case for dynamic human reliability analysis (HRA). HUNTER1 paired a dynamicized version of the SPAR-H HRA method that autocalculated the effects of performance shaping factors (PSFs), an implementation of the GOMS-HRA method to compute time for modeled human tasks, and an interface between the RAVEN modeling environment and RELAP5 thermo-hydraulics code. In this manner, a simple implementation of a virtual operator was coupled to a virtual plant model. In order to mature this framework, HUNTER2 has been initiated. HUNTER2 seeks to scale the earlier proof-of-concept demonstration into a software toolkit that can be deployed to support industry needs for dynamic HRA. One challenge that has existed with many dynamic HRA approaches is the need for bespoke software implementations for each modeled scenario. To avoid this, we are creating a scalable framework that can interface with different modeling tools as modules. The crucial elements of HUNTER2 are: (1) a task-modeling module; (2) a dynamic, autocalculated performance shaping factor module; (3) a decision-making module; (4) a scheduler module; and (5) a plant model. The scalable nature of HUNTER2 resides in its ability to switch modules through a codec that communicates with each module. For example, the current implementation of HUNTER2 continues developing the dynamic SPAR-H module begun with HUNTER1, but it is also possible to run a scenario without PSFs or with a different representation of PSFs. Similarly, the decision-making module can take many different forms, ranging from simple probabilistic decision paths to a more complex cognitive modeling architecture akin to artificial intelligence. Likewise, the world model may switch between reduced order plant models like the Rancor microworld simulator, to fully featured thermo-hydraulic modeling in RELAP5, to balance-of-plant modeling using EMRALD. The HUNTER2 architecture allows incorporation of dummy values, simplified modules, or rich models—allowing scalability of the interactions according to the modeling fidelity required. This paper discusses the development of the HUNTER2 framework and its implementation as a scalable and usable dynamic HRA software tool.

The HUNTER Dynamic Human Reliability Analysis Tool: Overview of the Software Framework for Modeling Digital Human Twins

The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) was previously developed as a simplified test case for dynamic human reliability analysis (HRA). HUNTER1 paired a dynamicized version of the SPAR-H HRA method that autocalculated the effects of performance shaping factors (PSFs), an implementation of the GOMS-HRA method to compute time for modeled human tasks, and an interface between the RAVEN modeling environment and RELAP5 thermo-hydraulics code. In this manner, a simple implementation of a virtual operator was coupled to a virtual plant model. In order to mature this framework, HUNTER2 has been initiated. HUNTER2 seeks to scale the earlier proof-of-concept demonstration into a software toolkit that can be deployed to support industry needs for dynamic HRA. One challenge that has existed with many dynamic HRA approaches is the need for bespoke software implementations for each modeled scenario. To avoid this, we are creating a scalable framework that can interface with different modeling tools as modules. The crucial elements of HUNTER2 are: (1) a task-modeling module; (2) a dynamic, autocalculated performance shaping factor module; (3) a decision-making module; (4) a scheduler module; and (5) a plant model. The scalable nature of HUNTER2 resides in its ability to switch modules through a codec that communicates with each module. For example, the current implementation of HUNTER2 continues developing the dynamic SPAR-H module begun with HUNTER1, but it is also possible to run a scenario without PSFs or with a different representation of PSFs. Similarly, the decision-making module can take many different forms, ranging from simple probabilistic decision paths to a more complex cognitive modeling architecture akin to artificial intelligence. Likewise, the world model may switch between reduced order plant models like the Rancor microworld simulator, to fully featured thermo-hydraulic modeling in RELAP5, to balance-of-plant modeling using EMRALD. The HUNTER2 architecture allows incorporation of dummy values, simplified modules, or rich models—allowing scalability of the interactions according to the modeling fidelity required. This paper discusses the development of the HUNTER2 framework and its implementation as a scalable and usable dynamic HRA software tool.

Paper RO173 | Download the paper file. | Download the presentation pdf file.

Name: Ronald Boring (ronald.boring@inl.gov)

Bio: Dr. Ron Boring is a Distinguished Human Factors Scientist and Department Manager at Idaho National Laboratory (INL). He has led control room modernization and human risk efforts for a variety of national and international partners. He was the founder of the Human Systems Simulation Laboratory at INL and led development of prototyping tools such as the Advanced Nuclear Interface Modeling Environment (ANIME) and human factors evaluation methods like the Guideline for Operational Nuclear Usability and Knowledge Elicitation (GONUKE) to support control room development at U.S. utilities. He has developed the Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) method, which is used extensively for risk modeling. Dr. Boring has a Ph.D. in Cognitive Science from Carleton University. He was a Fulbright Academic Scholar to the University of Heidelberg, Germany, and currently holds the honorary title of Fellow of the Human Factors and Ergonomics Society.

Country: USA
Company: Idaho National Laboratory
Job Title: Manager

Paper 2 TH196

Lead Author: Tom Ulrich Co-author(s): Ronald L Boring, ronald.boring@inl.gov Jeeyea Ahn, jeeyea.ahn@inl.gov Yunyeong Heo, yunyeong.heo@inl.gov Jooyoung, Park jooyoung.park@inl.gov Roger Lew, rogerlew@uidaho.edu

The HUNTER Dynamic Human Reliability Analysis Tool: Procedurally Driven Operator Simulation
This paper describes the software implementation of the HUNTER dynamic human reliability analysis (HRA) framework. The HUNTER software tool is a dynamic HRA simulation driven by an operator model defined through existing operating procedures. The software attempts to create a simplified approach to dynamic HRA that does not rely on complex cognitive modeling and therefore provide a more accessible tool for analysts. Traditional HRA models human errors by building a static model of the operators’ activities surrounding the predefined human failure event. Retroactive analysis entails gathering information around a known event, modeling the event, and then extrapolating the failure to other aspects of the system in prospective manner to determine human error opportunities in other parts of the system that could be impacted by the type of modelled event. Historically, this is a largely manual task performed by an analyst or team of analysts and, as a result, traditional HRA often suffers from a level of variability across analyses. Dynamic HRA provides an opportunity to more objectively build models in which the analyst defines the system and constraints. The simulation then uses Monte Carlo simulation to identify human failure events, their probabilities, and the time course of the events. To achieve this, the analyst must define the system and the operators’ activities. Nuclear control room operating procedures are highly prescriptive of operator activities and can serve as a model of potential operator actions. At its core, the HUNTER approach relies on a virtual operator model based on the execution of procedures with dynamically defined contexts representing the HRA aspects of the simulation. The simulation supports configuring both static and dynamic SPAR-H performance shaping factors, based on the nature of each of the individual factors. The likelihood of failure is therefore changing over the course of the simulation based on the operator and the plant state. In addition to the operator context, the human reliability module provides a suite of goals, operators, methods, and selection rules (GOMS) primitives that represent basic units of operator activity. Each primitive has a predefined, SHERPA based human error probability and empirically derived time distributions collected from full-scope simulator experiments. Each procedure step is assigned a primitive, which defines the time taken to execute the step and the likelihood of success to execute the step. Procedures are closed loop representations of plant diagnoses and actions, meaning success achieves the desired goal and failure results in deviating and sometimes looping back on the same set of procedures, which therefore defines a closed simulation environment suitable for handling a broad set of simulation outcomes. As the operator model is executing procedures, a coupled plant model, such as RELAP5, runs in tandem to provide updated plant states based on the scenario’s natural progress due to a given induced transient and any changes due to operator actions. The execution of activities results in the desired plant state or failure to execute appropriate activities, resulting in exceeding a safe operating envelope, which is coded as a failure. The procedure implementation and each of the models described briefly above are conveyed in greater detail in the paper.

The HUNTER Dynamic Human Reliability Analysis Tool: Procedurally Driven Operator Simulation

This paper describes the software implementation of the HUNTER dynamic human reliability analysis (HRA) framework. The HUNTER software tool is a dynamic HRA simulation driven by an operator model defined through existing operating procedures. The software attempts to create a simplified approach to dynamic HRA that does not rely on complex cognitive modeling and therefore provide a more accessible tool for analysts. Traditional HRA models human errors by building a static model of the operators’ activities surrounding the predefined human failure event. Retroactive analysis entails gathering information around a known event, modeling the event, and then extrapolating the failure to other aspects of the system in prospective manner to determine human error opportunities in other parts of the system that could be impacted by the type of modelled event. Historically, this is a largely manual task performed by an analyst or team of analysts and, as a result, traditional HRA often suffers from a level of variability across analyses. Dynamic HRA provides an opportunity to more objectively build models in which the analyst defines the system and constraints. The simulation then uses Monte Carlo simulation to identify human failure events, their probabilities, and the time course of the events. To achieve this, the analyst must define the system and the operators’ activities. Nuclear control room operating procedures are highly prescriptive of operator activities and can serve as a model of potential operator actions. At its core, the HUNTER approach relies on a virtual operator model based on the execution of procedures with dynamically defined contexts representing the HRA aspects of the simulation. The simulation supports configuring both static and dynamic SPAR-H performance shaping factors, based on the nature of each of the individual factors. The likelihood of failure is therefore changing over the course of the simulation based on the operator and the plant state. In addition to the operator context, the human reliability module provides a suite of goals, operators, methods, and selection rules (GOMS) primitives that represent basic units of operator activity. Each primitive has a predefined, SHERPA based human error probability and empirically derived time distributions collected from full-scope simulator experiments. Each procedure step is assigned a primitive, which defines the time taken to execute the step and the likelihood of success to execute the step. Procedures are closed loop representations of plant diagnoses and actions, meaning success achieves the desired goal and failure results in deviating and sometimes looping back on the same set of procedures, which therefore defines a closed simulation environment suitable for handling a broad set of simulation outcomes. As the operator model is executing procedures, a coupled plant model, such as RELAP5, runs in tandem to provide updated plant states based on the scenario’s natural progress due to a given induced transient and any changes due to operator actions. The execution of activities results in the desired plant state or failure to execute appropriate activities, resulting in exceeding a safe operating envelope, which is coded as a failure. The procedure implementation and each of the models described briefly above are conveyed in greater detail in the paper.

Paper TH196 | Download the paper file. | Download the presentation PowerPoint file.

Name: Tom Ulrich (thomas.ulrich@inl.gov)

Bio: Dr. Ulrich is a human factors and reliability research scientist at the Idaho National Laboratory. He has led and participated in several full-scope, full-scale simulator studies using the Human Systems Simulation Laboratory (HSSL) to investigate a range of nuclear control room topics. Dr. Ulrich possesses expertise in human performance assessment methodology with an emphasis on situation awareness and attention assessment via eye-tracking techniques. He is an expert in simulation and interface prototyping and has helped develop the advanced computerized operator support system (COSS), the Rancor Microworld, the ATRC digital control board, and numerous digital Turbine Control Systems. Dr. Ulrich’s active research includes dynamic human reliability analysis methodology development for nuclear power plant FLEX activities and development of prognostic human-machine interfaces (HMIs) for existing main control rooms and advanced reactor HMIs. Dr. Ulrich has also led an interdisciplinary res

Country: USA
Company: Idaho National Laboratory
Job Title: Human Factors and Reliability Research Scientist

Paper 3 YY168

Lead Author: Yunyeong Heo Co-author(s): Thomas A. Ulrich, thomas.ulrich@inl.gov Ronald L. Boring, ronald.boring@inl.gov Jooyoung Park, Jooyoung.Park@inl.gov Jeeyea Ahn, jeeya@unist.ac.kr

The HUNTER Dynamic Human Reliability Analysis Tool: Coupling an External Plant Code
The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) is a framework to support dynamic human reliability analysis (HRA) with the aim to develop standalone software to perform the dynamic HRA calculations. Within the HRA, human actions in nuclear power plants (NPPs) are predicated by plant states, and human actions influence the plant. In other words, plant operations are necessarily recursive, and it becomes challenging to model complex human-plant interactions. Consequently, we have linked two software simulations that complement those shortcomings. RELAP5-3D—the Reactor Excursion and Leak Analysis Program (RELAP; Aumiller, Tomlinson, and Bauer 2001) is the foundational thermal-hydraulic software used to model nuclear systems. Using RELAP5-3D, we have simulated the plant operations proceeding according to procedures developed to address emergent situations in NPPs. Plant operations include various actions such as the operator checking plant parameters, as well as actions that are continuously performed over time until a specific parameter reaches certain criteria. That means that HUNTER and RELAP5-3D exchange information with each other and should be carried out simultaneously over time. To simulate plant operations, which represent the actual operator checks of plant parameters and corresponding manual control actions, changes in plant status are identified through simulation and performed according to the criteria and order of the procedure. Thus, the goal of coupling HUNTER with RELAP5-3D is to facilitate synchronous coupling, where human and plant models provide iterative feedback loops that drive the course of actions. The advantage of coupling with RELAP5-3D to serve as the external environment module in HUNTER is the ability to customize the plant model and streamline for particular model applications. In this paper, we will address the key features of the coupling and the coupling structure built to perform the feedback loops.

The HUNTER Dynamic Human Reliability Analysis Tool: Coupling an External Plant Code

The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) is a framework to support dynamic human reliability analysis (HRA) with the aim to develop standalone software to perform the dynamic HRA calculations. Within the HRA, human actions in nuclear power plants (NPPs) are predicated by plant states, and human actions influence the plant. In other words, plant operations are necessarily recursive, and it becomes challenging to model complex human-plant interactions. Consequently, we have linked two software simulations that complement those shortcomings. RELAP5-3D—the Reactor Excursion and Leak Analysis Program (RELAP; Aumiller, Tomlinson, and Bauer 2001) is the foundational thermal-hydraulic software used to model nuclear systems. Using RELAP5-3D, we have simulated the plant operations proceeding according to procedures developed to address emergent situations in NPPs. Plant operations include various actions such as the operator checking plant parameters, as well as actions that are continuously performed over time until a specific parameter reaches certain criteria. That means that HUNTER and RELAP5-3D exchange information with each other and should be carried out simultaneously over time. To simulate plant operations, which represent the actual operator checks of plant parameters and corresponding manual control actions, changes in plant status are identified through simulation and performed according to the criteria and order of the procedure. Thus, the goal of coupling HUNTER with RELAP5-3D is to facilitate synchronous coupling, where human and plant models provide iterative feedback loops that drive the course of actions. The advantage of coupling with RELAP5-3D to serve as the external environment module in HUNTER is the ability to customize the plant model and streamline for particular model applications. In this paper, we will address the key features of the coupling and the coupling structure built to perform the feedback loops.

Paper YY168 | Download the paper file. |

Name: Yunyeong Heo (yyheo0207@unist.ac.kr)

Bio: Nuclear Engineering Combined M.S. and Ph.D. course

Country: KOR
Company: Ulsan National Institute of Science and Technology
Job Title: Graduate student

Paper 4 JO69

Lead Author: Jooyoung Park Co-author(s): Ronald Boring, Ronald.Boring@inl.gov Thomas Ulrich, Thomas.Ulrich@inl.gov Jeeyea Ahn, Jeeyea.Ahn@inl.gov Yunyeong Heo, Yunyeong.Heo@inl.gov

The HUNTER Dynamic Human Reliability Analysis Tool: Development of a Module for Performance Shaping Factors
The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) is a framework to support dynamic human reliability analysis (HRA) in communication with a variety of methods and tools. The HRA research team at Idaho National Laboratory has developed the HUNTER framework to meet industry HRA needs by the support of the Risk-Informed System Analysis (RISA) pathway of the U.S. Department of Energy’s Light Water Reactor Sustainability (LWRS) Program. The existing HUNTER (i.e., HUNTER 1.0) has conceptually proposed how the dynamic HRA could be performed using given information from thermal-hydraulics codes, cognitive models, HRA methods, and procedures, while the current HUNTER project (i.e., HUNTER 2.0) aims to systemically design the HUNTER modules and their functions, then develop a standalone HUNTER software to implement the dynamic HRA calculation. In this paper, how we have developed one of the HUNTER modules, the performance shaping factor (PSF) module, is introduced. The PSF refers to any factor that influences human performance such as workload or complexity. It has been used for highlighting human errors and adjusting the error probabilities in the existing HRA. In this paper, we mainly discuss how the PSF module supports the dynamic HRA calculation within the HUNTER software. We consider the eight PSFs suggested in the Standardized Plant Analysis Risk-HRA (SPAR-H) method, which is the representative HRA method widely used in the nuclear field. We also design the PSF module composed of the two functions: 1) the PSF qualification function that automatically or manually evaluates a PSF level, and 2) the PSF quantification function that dynamically or statically determines the PSF multiplier values and integrate them to adjust human error probabilities (HEPs). How each function works with the SPAR-H PSFs and how the PSFs adjust the HEPs and the time required for operators are investigated through literature.

The HUNTER Dynamic Human Reliability Analysis Tool: Development of a Module for Performance Shaping Factors

The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) is a framework to support dynamic human reliability analysis (HRA) in communication with a variety of methods and tools. The HRA research team at Idaho National Laboratory has developed the HUNTER framework to meet industry HRA needs by the support of the Risk-Informed System Analysis (RISA) pathway of the U.S. Department of Energy’s Light Water Reactor Sustainability (LWRS) Program. The existing HUNTER (i.e., HUNTER 1.0) has conceptually proposed how the dynamic HRA could be performed using given information from thermal-hydraulics codes, cognitive models, HRA methods, and procedures, while the current HUNTER project (i.e., HUNTER 2.0) aims to systemically design the HUNTER modules and their functions, then develop a standalone HUNTER software to implement the dynamic HRA calculation. In this paper, how we have developed one of the HUNTER modules, the performance shaping factor (PSF) module, is introduced. The PSF refers to any factor that influences human performance such as workload or complexity. It has been used for highlighting human errors and adjusting the error probabilities in the existing HRA. In this paper, we mainly discuss how the PSF module supports the dynamic HRA calculation within the HUNTER software. We consider the eight PSFs suggested in the Standardized Plant Analysis Risk-HRA (SPAR-H) method, which is the representative HRA method widely used in the nuclear field. We also design the PSF module composed of the two functions: 1) the PSF qualification function that automatically or manually evaluates a PSF level, and 2) the PSF quantification function that dynamically or statically determines the PSF multiplier values and integrate them to adjust human error probabilities (HEPs). How each function works with the SPAR-H PSFs and how the PSFs adjust the HEPs and the time required for operators are investigated through literature.

Paper JO69 | Download the paper file. |

Name: Jooyoung Park (jooyoung.park@inl.gov)

Bio:

Country: USA
Company: Idaho National Laboratory
Job Title: Scientist