### Abstract

Original language | English |
---|---|

Place of Publication | Olkiluoto, Eurajoki |

Publisher | Posiva |

Number of pages | 60 |

Publication status | Published - 2011 |

MoE publication type | D4 Published development or research report or study |

### Publication series

Series | Working Report |
---|---|

Number | 2011-36 |

### Fingerprint

### Keywords

- Encapsulation and final disposal
- spent nuclear fuel
- defect
- NDT
- welding
- probability
- Bayes

### Cite this

*Analysis of probability of defects in the disposal canisters*. Olkiluoto, Eurajoki: Posiva . Working Report, No. 2011-36

}

*Analysis of probability of defects in the disposal canisters*. Working Report, no. 2011-36, Posiva , Olkiluoto, Eurajoki.

**Analysis of probability of defects in the disposal canisters.** / Holmberg, Jan-Erik; Kuusela, Pirkko.

Research output: Book/Report › Report

TY - BOOK

T1 - Analysis of probability of defects in the disposal canisters

AU - Holmberg, Jan-Erik

AU - Kuusela, Pirkko

N1 - Project code: 74108

PY - 2011

Y1 - 2011

N2 - This report presents a probability model for the reliability of the spent nuclear waste final disposal canister. Reliability means here that the welding of the canister lid has no critical defects from the long-term safety point of view. From the reliability point of view, both the reliability of the welding process (that no critical defects will be born) and the non-destructive testing (NDT) process (all critical defects will be detected) are equally important. In the probability model, critical defects in a weld were simplified into a few types. Also the possibility of human errors in the NDT process was taken into account in a simple manner. At this moment there is very little representative data to determine the reliability of welding and also the data on NDT is not well suited for the needs of this study. Therefore calculations presented here are based on expert judgements and on several assumptions that have not been verified yet. The Bayesian probability model shows the importance of the uncertainty in the estimation of the reliability parameters. The effect of uncertainty is that the probability distribution of the number of defective canisters becomes flat for larger numbers of canisters compared to the binomial probability distribution in case of known parameter values. In order to reduce the uncertainty, more information is needed from both the reliability of the welding and NDT processes. It would also be important to analyse the role of human factors in these processes since their role is not reflected in typical test data which is used to estimate “normal process variation”. The reported model should be seen as a tool to quantify the roles of different methods and procedures in the weld inspection process.

AB - This report presents a probability model for the reliability of the spent nuclear waste final disposal canister. Reliability means here that the welding of the canister lid has no critical defects from the long-term safety point of view. From the reliability point of view, both the reliability of the welding process (that no critical defects will be born) and the non-destructive testing (NDT) process (all critical defects will be detected) are equally important. In the probability model, critical defects in a weld were simplified into a few types. Also the possibility of human errors in the NDT process was taken into account in a simple manner. At this moment there is very little representative data to determine the reliability of welding and also the data on NDT is not well suited for the needs of this study. Therefore calculations presented here are based on expert judgements and on several assumptions that have not been verified yet. The Bayesian probability model shows the importance of the uncertainty in the estimation of the reliability parameters. The effect of uncertainty is that the probability distribution of the number of defective canisters becomes flat for larger numbers of canisters compared to the binomial probability distribution in case of known parameter values. In order to reduce the uncertainty, more information is needed from both the reliability of the welding and NDT processes. It would also be important to analyse the role of human factors in these processes since their role is not reflected in typical test data which is used to estimate “normal process variation”. The reported model should be seen as a tool to quantify the roles of different methods and procedures in the weld inspection process.

KW - Encapsulation and final disposal

KW - spent nuclear fuel

KW - defect

KW - NDT

KW - welding

KW - probability

KW - Bayes

M3 - Report

T3 - Working Report

BT - Analysis of probability of defects in the disposal canisters

PB - Posiva

CY - Olkiluoto, Eurajoki

ER -