### Abstract

Compared to nominal summary statistics, a better method to show compliance with regulatory criteria is to determine the distribution of the assembly mean, taking into account the effects of manufacturing tolerances and uncertainty of the power history. The upper tolerance limit is straightforward to calculate with order statistics and with suitable parameters it is accepted by many regulatory authorities. In this work, the calculation of tolerance limits has been applied to mean fission gas release in complete fuel assemblies. A calculation system for such analyses has been developed at VTT, and by calculating each rod multiple times it can be used to determine the tolerance limits for complete nuclear reactor cores. In statistical analyses, typically a large number of simulations must be run. This increases the probability of failure of the simulation program due to non convergence of the calculation. This results in missing data, and this could prevent the use of order statistics which are based on the sorting of observations. With missing data, the real order of the observations is unknown. This work suggests a suitable data treatment method to overcome this problem.

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

Article number | 110438 |

Journal | Nuclear Engineering and Design |

Volume | 358 |

Early online date | 21 Nov 2019 |

DOIs | |

Publication status | E-pub ahead of print - 21 Nov 2019 |

MoE publication type | A1 Journal article-refereed |

### Fingerprint

### Cite this

}

**Determination of tolerance limits for fuel assembly fission gas release summary statistics.** / Loukusa, Henri (Corresponding Author); Tulkki, Ville.

Research output: Contribution to journal › Article › Scientific › peer-review

TY - JOUR

T1 - Determination of tolerance limits for fuel assembly fission gas release summary statistics

AU - Loukusa, Henri

AU - Tulkki, Ville

PY - 2019/11/21

Y1 - 2019/11/21

N2 - Instant release fraction is used in spent fuel final disposal studies to estimate the quickly released inventory from a spent fuel final disposal capsule. The instant release fraction has been correlated with fission gas release, which can be readily calculated with fuel performance codes. In spent fuel final disposal, individual rod fission gas releases are not important, as the fuel is treated in complete capsules containing multiple assemblies. Therefore assembly summary statistics are more suitable for this purpose. Summary statistics such as mean or maximum are useful in this application. However, the nominal values for the assembly mean fission gas release may be non conservative estimates, whereas the assembly maximum fission gas release is a highly conservative assumption. This results from the “cutoff log-normal” distribution of fission gas release in an assembly and in a more general level, the reactor core. The tail of the distribution contains a small amount of rods with a high fission gas release, whereas a large number of rods typically have low fission gas releases.Compared to nominal summary statistics, a better method to show compliance with regulatory criteria is to determine the distribution of the assembly mean, taking into account the effects of manufacturing tolerances and uncertainty of the power history. The upper tolerance limit is straightforward to calculate with order statistics and with suitable parameters it is accepted by many regulatory authorities. In this work, the calculation of tolerance limits has been applied to mean fission gas release in complete fuel assemblies. A calculation system for such analyses has been developed at VTT, and by calculating each rod multiple times it can be used to determine the tolerance limits for complete nuclear reactor cores. In statistical analyses, typically a large number of simulations must be run. This increases the probability of failure of the simulation program due to non convergence of the calculation. This results in missing data, and this could prevent the use of order statistics which are based on the sorting of observations. With missing data, the real order of the observations is unknown. This work suggests a suitable data treatment method to overcome this problem.

AB - Instant release fraction is used in spent fuel final disposal studies to estimate the quickly released inventory from a spent fuel final disposal capsule. The instant release fraction has been correlated with fission gas release, which can be readily calculated with fuel performance codes. In spent fuel final disposal, individual rod fission gas releases are not important, as the fuel is treated in complete capsules containing multiple assemblies. Therefore assembly summary statistics are more suitable for this purpose. Summary statistics such as mean or maximum are useful in this application. However, the nominal values for the assembly mean fission gas release may be non conservative estimates, whereas the assembly maximum fission gas release is a highly conservative assumption. This results from the “cutoff log-normal” distribution of fission gas release in an assembly and in a more general level, the reactor core. The tail of the distribution contains a small amount of rods with a high fission gas release, whereas a large number of rods typically have low fission gas releases.Compared to nominal summary statistics, a better method to show compliance with regulatory criteria is to determine the distribution of the assembly mean, taking into account the effects of manufacturing tolerances and uncertainty of the power history. The upper tolerance limit is straightforward to calculate with order statistics and with suitable parameters it is accepted by many regulatory authorities. In this work, the calculation of tolerance limits has been applied to mean fission gas release in complete fuel assemblies. A calculation system for such analyses has been developed at VTT, and by calculating each rod multiple times it can be used to determine the tolerance limits for complete nuclear reactor cores. In statistical analyses, typically a large number of simulations must be run. This increases the probability of failure of the simulation program due to non convergence of the calculation. This results in missing data, and this could prevent the use of order statistics which are based on the sorting of observations. With missing data, the real order of the observations is unknown. This work suggests a suitable data treatment method to overcome this problem.

UR - http://www.scopus.com/inward/record.url?scp=85075203289&partnerID=8YFLogxK

U2 - 10.1016/j.nucengdes.2019.110438

DO - 10.1016/j.nucengdes.2019.110438

M3 - Article

VL - 358

JO - Nuclear Engineering and Design

JF - Nuclear Engineering and Design

SN - 0029-5493

M1 - 110438

ER -