Nuclear Energy for New Europe 2009

(This is a HTML version of the abstract for information only. It can differ from the original, submitted by the author(s). Special characters, formulas and figures are not properly reproduced here. Please, contact the author(s) or refer to the printed Book of Abstracts for the correct version.)

No: 304

Conference: Nuclear Energy for New Europe 2009

Title: Comparison of Two HRA Methods Based on Nuclear Power Plant Simulator Training

Theme: Probabilistic Safety Assessment

Author(s): Gueorgui Petkov, Emil Stefanov

Contact : Gueorgui Petkov

E-mail: gipetkov@npp.bg

Address: Technical University of Sofia Electrical Power Dept.
1797 Sofia

Country: Bulgaria

In the PSA models of accident sequences, the designed response of operators and managers leads to a safe system state when a sequence of hardware systems and barriers operate successfully and the operators perform a set of required actions at the appropriate points in scenarios. Unsafe system states are reached through a combination of hardware failures and/or omissions of required actions. Consequently, the Human Reliability Analysis (HRA) methods are focused basically on the omission of the required actions at selected points of scenarios – Error Of Omission (EOO). However, under higher stress of hardware and barrier materials and humans, the operators and managers may often erroneously perform an action that will aggravate the scenario at any point of its evolution – Error Of Commission (EOC). The identification of the potentially significant hazards, EOCs and EOOs is the central problem of their treatment in the PSA. Other issues that should be addressed are modelling and quantification of the EOCs and EOOs in the accident conditions.
Context dependent HRA models, such as the Holistic Decision Tree (HDT) and Performance Evaluation of Teamwork (PET) methods, are the so-called second-generation HRA techniques.
The HDT model has been used for a number of PSA studies. The approach to the determination of Influence Factors, Quality Values, Anchor Values and verification of Human Error Probabilities (HEPs) has varied. Making detailed observations of operator responses is a very intensive task, especially if the observers record their notes manually. That is why the HDT method includes Data Collection System (DCS) – CREDIT. Use has been made of various domain experts, HRA experts and simulator results. The approaches have varied because of the available tools, experts and time/money.
The PET method reflects promising prospects for dealing with dynamic aspects of human performance. It distinguishes between three basic concepts determining the reliability of human performance: violation, cognitive erroneous actions (as EOCs) and executive erroneous actions (as EOOs). Based on these concepts a “second-by-second” macroscopic quantification of contexts of individual cognition and team communication processes is made. Naturally and technologically recognized context factors and conditions are taken into account. The results of contexts quantification are used in reliability models of cognition (Rasmussen’s Step-Ladder Model), communication and execution processes. The HEP calculations could be based on the thermal-hydraulic analyses, archives of training exercises, audio-visual system performance observation and operators’/instructors’ interpretation and justification. The Simulators Department of Kozloduy NPP has elaborated a methodology based on the PET method for quantitative assessment of operators’ team performance. The model data could be obtained from training programs, operation instructions and emergency procedures. This methodology could be applied for training evaluation on different scale simulators – full-scope, multifunctional and computer-based.
The proposed paper presents a comparison of the three HRA techniques for calculation of post-accident HEPs in the PSA, namely: HDT; and PET. Different event training scenarios based on standard (benchmark) PSA model assumptions are designed and their detailed boundary conditions are described. The HEPs of post-accident main control room critical operator’s actions are calculated by means of each investigated technique. For each inconsistency identified, a counter measure and recommendations are proposed.

No:	304
Conference:	Nuclear Energy for New Europe 2009
Title:	Comparison of Two HRA Methods Based on Nuclear Power Plant Simulator Training
Theme:	Probabilistic Safety Assessment
Author(s):	Gueorgui Petkov, Emil Stefanov
Contact :	Gueorgui Petkov
E-mail:	gipetkov@npp.bg
Address:	Technical University of Sofia Electrical Power Dept. 1797 Sofia
Country:	Bulgaria

In the PSA models of accident sequences, the designed response of operators and managers leads to a safe system state when a sequence of hardware systems and barriers operate successfully and the operators perform a set of required actions at the appropriate points in scenarios. Unsafe system states are reached through a combination of hardware failures and/or omissions of required actions. Consequently, the Human Reliability Analysis (HRA) methods are focused basically on the omission of the required actions at selected points of scenarios – Error Of Omission (EOO). However, under higher stress of hardware and barrier materials and humans, the operators and managers may often erroneously perform an action that will aggravate the scenario at any point of its evolution – Error Of Commission (EOC). The identification of the potentially significant hazards, EOCs and EOOs is the central problem of their treatment in the PSA. Other issues that should be addressed are modelling and quantification of the EOCs and EOOs in the accident conditions. Context dependent HRA models, such as the Holistic Decision Tree (HDT) and Performance Evaluation of Teamwork (PET) methods, are the so-called second-generation HRA techniques. The HDT model has been used for a number of PSA studies. The approach to the determination of Influence Factors, Quality Values, Anchor Values and verification of Human Error Probabilities (HEPs) has varied. Making detailed observations of operator responses is a very intensive task, especially if the observers record their notes manually. That is why the HDT method includes Data Collection System (DCS) – CREDIT. Use has been made of various domain experts, HRA experts and simulator results. The approaches have varied because of the available tools, experts and time/money. The PET method reflects promising prospects for dealing with dynamic aspects of human performance. It distinguishes between three basic concepts determining the reliability of human performance: violation, cognitive erroneous actions (as EOCs) and executive erroneous actions (as EOOs). Based on these concepts a “second-by-second” macroscopic quantification of contexts of individual cognition and team communication processes is made. Naturally and technologically recognized context factors and conditions are taken into account. The results of contexts quantification are used in reliability models of cognition (Rasmussen’s Step-Ladder Model), communication and execution processes. The HEP calculations could be based on the thermal-hydraulic analyses, archives of training exercises, audio-visual system performance observation and operators’/instructors’ interpretation and justification. The Simulators Department of Kozloduy NPP has elaborated a methodology based on the PET method for quantitative assessment of operators’ team performance. The model data could be obtained from training programs, operation instructions and emergency procedures. This methodology could be applied for training evaluation on different scale simulators – full-scope, multifunctional and computer-based. The proposed paper presents a comparison of the three HRA techniques for calculation of post-accident HEPs in the PSA, namely: HDT; and PET. Different event training scenarios based on standard (benchmark) PSA model assumptions are designed and their detailed boundary conditions are described. The HEPs of post-accident main control room critical operator’s actions are calculated by means of each investigated technique. For each inconsistency identified, a counter measure and recommendations are proposed.