This is NKS NKS-R NKS-B News Seminars NKS Reports Phantom Library

You are here: Homepage NKS Reports View document

List all reports List all NKS-R reports List all NKS-B reports Search Reports  
NKS Programme Area:
Research Area:NKS R and B
Report Number:NKS-293
Report Title:Using Bayesian Belief Network (BBN) Modelling for Rapid Source Term Prediction – Final Report
Activity Acronym:RASTEP
Authors:Michael Knochenhauer, Vidar Hedtjärn Swaling, Francesco Di Dedda, Frida Hansson, Stina Sjökvist, Klas Sunnegård,
Abstract:The project presented in this report deals with a number of complex issues related to the development of a tool for rapid source term prediction (RASTEP), based on a plant model represented as a Bayesian belief network (BBN) and a source term module which is used for assigning relevant source terms to BBN end states. Thus, RASTEP uses a BBN to model severe accident progression in a nuclear power plant in combination with pre-calculated source terms (i.e., amount, composition, timing, and release path of released radio-nuclides). The output is a set of possible source terms with associated probabilities. One major issue has been associated with the integration of probabilistic and deterministic analyses are addressed, dealing with the challenge of making the source term determination flexible enough to give reliable and valid output throughout the accident scenario. The potential for connecting RASTEP to a fast running source term prediction code has been explored, as well as alternative ways of improving the deterministic connections of the tool. As part of the investigation, a comparison of two deterministic severe accident analysis codes has been performed. A second important task has been to develop a general method where experts' beliefs can be included in a systematic way when defining the conditional probability tables (CPTs) in the BBN. The proposed method includes expert judgement in a systematic way when defining the CPTs of a BBN. Using this iterative method results in a reliable BBN even though expert judgements, with their associated uncertainties, have been used. It also simplifies verification and validation of the considerable amounts of quantitative data included in a BBN.
Keywords:BBN, Bayesian Belief Network, Severe Accidents, Source Terms, Level 2 PSA, CPT, Conditional Probability Tables
Publication date:31 Oct 2013
ISBN:ISBN 978-87-7893-369-0
Number of downloads:2025
Download:pdf NKS-293.pdf
Contact NKS   NKS Sekretariatet
Boks 49
DK-4000 Roskilde
  Telephone +45 46 77 40 41
E-mail: nks@nks.org 
 

Address for visitors
Directions and map

Privacy Policy

Website last modified: 13 March 2019