Guidelines for reliability analysis of digital systems in PSA context: Phase 2 Status Report

Stefan Authén, Johan Gustafsson, Jan-Erik Holmberg

Research output: Book/ReportReport

Abstract

The OECD/NEA CSNI Working Group on Risk Assessment (WGRisk) has set up a task group called DIGREL to develop a taxonomy of failure modes of digital components for the purposes of probabilistic safety assessment (PSA). A parallel Nordic activity carried out a pre-study where a comparison of Nordic experiences and a literature review were performed. The study showed a wide range of approaches and solutions to the challenges given by digital I&C. In 2011, a proposal for the failure modes taxonomy was defined. This is based on a set of requirements agreed on the purpose of the taxonomy. The following levels of details can be distinguished from the hardware point of view: (1) the entire system, (2) a division, (3) processing units (and cabinets), (4) modules, i.e. subcomponents of processing units and (5) generic components, i.e. subcomponents of modules. Module level seems to be the most appropriate from the PSA modelling point of view. The software failure modes taxonomy is still an open issue. An existing simplified PSA model has been complemented with fault tree models for a four-redundant distributed protection system in order to study and demonstrate the effect of design features and modelling approaches. The example shows that even rather simple I&C design leads to rather complex model despite of the fact that many things have been simplified and only a few protection signals are considered. One lesson from the example is that the Alpha factor model should be used to model common cause failures instead of the Beta factor model. Two options were developed to the comparison of different fail-safe principles. The role of detectable and undetectable failure modes with respect to the failed versus spurious actuations can be clearly seen in the results, showing the importance to model these features in PSA.
Original languageEnglish
Place of PublicationRoskilde, Denmark
Number of pages32
Publication statusPublished - 2012
MoE publication typeD4 Published development or research report or study

Publication series

SeriesNKS Nordic nuclear safety research
NumberNKS-261

Fingerprint

Reliability analysis
Taxonomies
Failure modes
Processing
Risk assessment
Hardware

Keywords

  • Digital I&C system
  • probabilistic safety assessment
  • reliability
  • nuclear power plant safety

Cite this

Authén, S., Gustafsson, J., & Holmberg, J-E. (2012). Guidelines for reliability analysis of digital systems in PSA context: Phase 2 Status Report. Roskilde, Denmark. NKS Nordic nuclear safety research, No. NKS-261
Authén, Stefan ; Gustafsson, Johan ; Holmberg, Jan-Erik. / Guidelines for reliability analysis of digital systems in PSA context : Phase 2 Status Report. Roskilde, Denmark, 2012. 32 p. (NKS Nordic nuclear safety research; No. NKS-261).
@book{ec53e9103ef741b28ca76b3961ec0fc8,
title = "Guidelines for reliability analysis of digital systems in PSA context: Phase 2 Status Report",
abstract = "The OECD/NEA CSNI Working Group on Risk Assessment (WGRisk) has set up a task group called DIGREL to develop a taxonomy of failure modes of digital components for the purposes of probabilistic safety assessment (PSA). A parallel Nordic activity carried out a pre-study where a comparison of Nordic experiences and a literature review were performed. The study showed a wide range of approaches and solutions to the challenges given by digital I&C. In 2011, a proposal for the failure modes taxonomy was defined. This is based on a set of requirements agreed on the purpose of the taxonomy. The following levels of details can be distinguished from the hardware point of view: (1) the entire system, (2) a division, (3) processing units (and cabinets), (4) modules, i.e. subcomponents of processing units and (5) generic components, i.e. subcomponents of modules. Module level seems to be the most appropriate from the PSA modelling point of view. The software failure modes taxonomy is still an open issue. An existing simplified PSA model has been complemented with fault tree models for a four-redundant distributed protection system in order to study and demonstrate the effect of design features and modelling approaches. The example shows that even rather simple I&C design leads to rather complex model despite of the fact that many things have been simplified and only a few protection signals are considered. One lesson from the example is that the Alpha factor model should be used to model common cause failures instead of the Beta factor model. Two options were developed to the comparison of different fail-safe principles. The role of detectable and undetectable failure modes with respect to the failed versus spurious actuations can be clearly seen in the results, showing the importance to model these features in PSA.",
keywords = "Digital I&C system, probabilistic safety assessment, reliability, nuclear power plant safety",
author = "Stefan Auth{\'e}n and Johan Gustafsson and Jan-Erik Holmberg",
note = "Project code: 73577",
year = "2012",
language = "English",
isbn = "978-87-7893-333-1",
series = "NKS Nordic nuclear safety research",
number = "NKS-261",

}

Authén, S, Gustafsson, J & Holmberg, J-E 2012, Guidelines for reliability analysis of digital systems in PSA context: Phase 2 Status Report. NKS Nordic nuclear safety research, no. NKS-261, Roskilde, Denmark.

Guidelines for reliability analysis of digital systems in PSA context : Phase 2 Status Report. / Authén, Stefan; Gustafsson, Johan; Holmberg, Jan-Erik.

Roskilde, Denmark, 2012. 32 p. (NKS Nordic nuclear safety research; No. NKS-261).

Research output: Book/ReportReport

TY - BOOK

T1 - Guidelines for reliability analysis of digital systems in PSA context

T2 - Phase 2 Status Report

AU - Authén, Stefan

AU - Gustafsson, Johan

AU - Holmberg, Jan-Erik

N1 - Project code: 73577

PY - 2012

Y1 - 2012

N2 - The OECD/NEA CSNI Working Group on Risk Assessment (WGRisk) has set up a task group called DIGREL to develop a taxonomy of failure modes of digital components for the purposes of probabilistic safety assessment (PSA). A parallel Nordic activity carried out a pre-study where a comparison of Nordic experiences and a literature review were performed. The study showed a wide range of approaches and solutions to the challenges given by digital I&C. In 2011, a proposal for the failure modes taxonomy was defined. This is based on a set of requirements agreed on the purpose of the taxonomy. The following levels of details can be distinguished from the hardware point of view: (1) the entire system, (2) a division, (3) processing units (and cabinets), (4) modules, i.e. subcomponents of processing units and (5) generic components, i.e. subcomponents of modules. Module level seems to be the most appropriate from the PSA modelling point of view. The software failure modes taxonomy is still an open issue. An existing simplified PSA model has been complemented with fault tree models for a four-redundant distributed protection system in order to study and demonstrate the effect of design features and modelling approaches. The example shows that even rather simple I&C design leads to rather complex model despite of the fact that many things have been simplified and only a few protection signals are considered. One lesson from the example is that the Alpha factor model should be used to model common cause failures instead of the Beta factor model. Two options were developed to the comparison of different fail-safe principles. The role of detectable and undetectable failure modes with respect to the failed versus spurious actuations can be clearly seen in the results, showing the importance to model these features in PSA.

AB - The OECD/NEA CSNI Working Group on Risk Assessment (WGRisk) has set up a task group called DIGREL to develop a taxonomy of failure modes of digital components for the purposes of probabilistic safety assessment (PSA). A parallel Nordic activity carried out a pre-study where a comparison of Nordic experiences and a literature review were performed. The study showed a wide range of approaches and solutions to the challenges given by digital I&C. In 2011, a proposal for the failure modes taxonomy was defined. This is based on a set of requirements agreed on the purpose of the taxonomy. The following levels of details can be distinguished from the hardware point of view: (1) the entire system, (2) a division, (3) processing units (and cabinets), (4) modules, i.e. subcomponents of processing units and (5) generic components, i.e. subcomponents of modules. Module level seems to be the most appropriate from the PSA modelling point of view. The software failure modes taxonomy is still an open issue. An existing simplified PSA model has been complemented with fault tree models for a four-redundant distributed protection system in order to study and demonstrate the effect of design features and modelling approaches. The example shows that even rather simple I&C design leads to rather complex model despite of the fact that many things have been simplified and only a few protection signals are considered. One lesson from the example is that the Alpha factor model should be used to model common cause failures instead of the Beta factor model. Two options were developed to the comparison of different fail-safe principles. The role of detectable and undetectable failure modes with respect to the failed versus spurious actuations can be clearly seen in the results, showing the importance to model these features in PSA.

KW - Digital I&C system

KW - probabilistic safety assessment

KW - reliability

KW - nuclear power plant safety

M3 - Report

SN - 978-87-7893-333-1

T3 - NKS Nordic nuclear safety research

BT - Guidelines for reliability analysis of digital systems in PSA context

CY - Roskilde, Denmark

ER -

Authén S, Gustafsson J, Holmberg J-E. Guidelines for reliability analysis of digital systems in PSA context: Phase 2 Status Report. Roskilde, Denmark, 2012. 32 p. (NKS Nordic nuclear safety research; No. NKS-261).