PSAM 16 - Paper Overview

PSAM 16 Conference Paper Overview

Welcome to the PSAM 16 Conference paper and speaker overview page.

Lead Author: Anders Riber Marklund Co-author(s): Juan Franco Canepa, franco.canepa@vysusgroup.com Anders Riber Marklund, anders.ribermarklund@vysusgroup.com Sergey Galushin, sergey.galushin@vysusgroup.com Anders Olsson, anders.olsson@vysusgroup.com

Evaluation of the source term prediction capability of RASTEP against MAAP v5.04 calculations
RASTEP (Rapid Source Term Prediction) is a software tool and methodology, developed by Vysus Group in cooperation with the Swedish Radiation Safety Authority (SSM) aiming to provide decision support in nuclear emergency situations related to NPP severe accidents. The software is based on Bayesian Belief Networks (BBNs) which allows to relate variables that can take different states given the observations input by the user. These variables are interconnected throughout the network, using conditional probabilities and logic relations, to capture an online diagnosis of the plant. The prediction of the source term in a severe accident of a NPP (with sufficient time ahead) is important for any emergency response team. One of the added values of RASTEP is to provide a real time likelihood assessment over a set of possible source terms, based on observed plant conditions, in line with IAEA guidance. This trait is not only important for the early diagnosis of incidents or accidents, but it is crucial when the accident is already developed to predict which source term is the most probable, supporting decision-making to mitigate the off-site consequences of the event. This report analyzes the ability of RASTEP to predict with sufficient time ahead the source term in a generic PWR. Four transients were studied, namely Small Break LOCA, Large Break LOCA, Interfacing System LOCA and Station Black Out. For each transient, the RASTEP software was fed with data extracted from integral response code results of the same transient, using MAAP v5.04 calculations on a similar reactor model. For each of the four transients, the time of onset of radionuclide release was compared with the time at which the correct source term is detected as the most probable by the RASTEP model. The results of this work show that for each of the four transients, RASTEP predicts the source term ahead of time, even when there is a certain level of information that is lost due to defective sensors or information missed by the user. This evaluation scheme also provides an attractive possibility to quantify the quality assurance process for RASTEP models.

Evaluation of the source term prediction capability of RASTEP against MAAP v5.04 calculations

RASTEP (Rapid Source Term Prediction) is a software tool and methodology, developed by Vysus Group in cooperation with the Swedish Radiation Safety Authority (SSM) aiming to provide decision support in nuclear emergency situations related to NPP severe accidents. The software is based on Bayesian Belief Networks (BBNs) which allows to relate variables that can take different states given the observations input by the user. These variables are interconnected throughout the network, using conditional probabilities and logic relations, to capture an online diagnosis of the plant. The prediction of the source term in a severe accident of a NPP (with sufficient time ahead) is important for any emergency response team. One of the added values of RASTEP is to provide a real time likelihood assessment over a set of possible source terms, based on observed plant conditions, in line with IAEA guidance. This trait is not only important for the early diagnosis of incidents or accidents, but it is crucial when the accident is already developed to predict which source term is the most probable, supporting decision-making to mitigate the off-site consequences of the event. This report analyzes the ability of RASTEP to predict with sufficient time ahead the source term in a generic PWR. Four transients were studied, namely Small Break LOCA, Large Break LOCA, Interfacing System LOCA and Station Black Out. For each transient, the RASTEP software was fed with data extracted from integral response code results of the same transient, using MAAP v5.04 calculations on a similar reactor model. For each of the four transients, the time of onset of radionuclide release was compared with the time at which the correct source term is detected as the most probable by the RASTEP model. The results of this work show that for each of the four transients, RASTEP predicts the source term ahead of time, even when there is a certain level of information that is lost due to defective sensors or information missed by the user. This evaluation scheme also provides an attractive possibility to quantify the quality assurance process for RASTEP models.

PSAM 16 Conference Paper Overview

Welcome to the PSAM 16 Conference paper and speaker overview page.

Paper AN138 Preview

Author and Presentation Info

Download paper AN138.

Download the presentation PowerPoint file.