BibTeX

@inproceedings{GrothPSAM2010,
title = {A Performance Shaping Factors Causal Model for Nuclear Power Plant Human Reliability Analysis.},
author = {Katrina Groth and Ali Mosleh},
year = {2010},
date = {2010-06-01},
booktitle = {Proceedings of the International Conference on Probabilistic Safety Assessment and Management (PSAM 10)},
address = {Seattle, WA},
abstract = {Many current Human Reliability Analysis (HRA) methods calculate human error probability (HEP) based on the state of various PSFs. There is no standard set of PSFs used in HRA, rather each method uses a unique set of PSFs, with varying degrees of interdependency among the PSFs. In calculating HEPs, interdependency is generally ignored or addressed through varying parameters in linear or loglinear formulas. These dependencies could be more accurately represented by a causal model of PSF relationships.
This paper introduces a methodology to develop a data-informed Bayesian Belief Network (BBN) that can be used to refine HEP prediction by reducing overlap among PSFs. The BBN framework was selected because it has the ability to incorporate available data and supplement it with expert judgment. The methodology allows the initial models to be updated as additional data becomes available. This paper presents a draft model based on currently available data from human error events in nuclear power plants. The resulting network model of interdependent PSFs is intended to replace linear calculations for HEPs.},
keywords = {Bayesian Networks, causal models, human error, Human Reliability Analysis (HRA), nuclear power, Performance Shaping Factors (PSFs)},
pubstate = {published},
tppubtype = {inproceedings}
}

Abstract

Many current Human Reliability Analysis (HRA) methods calculate human error probability (HEP) based on the state of various PSFs. There is no standard set of PSFs used in HRA, rather each method uses a unique set of PSFs, with varying degrees of interdependency among the PSFs. In calculating HEPs, interdependency is generally ignored or addressed through varying parameters in linear or loglinear formulas. These dependencies could be more accurately represented by a causal model of PSF relationships.
This paper introduces a methodology to develop a data-informed Bayesian Belief Network (BBN) that can be used to refine HEP prediction by reducing overlap among PSFs. The BBN framework was selected because it has the ability to incorporate available data and supplement it with expert judgment. The methodology allows the initial models to be updated as additional data becomes available. This paper presents a draft model based on currently available data from human error events in nuclear power plants. The resulting network model of interdependent PSFs is intended to replace linear calculations for HEPs.