Abstract
One application for the recently implemented collision-history based sensitivity calculation capabilities of Serpent is the propagation of nuclear data (cross section) uncertainties to the results of Serpent. Of special interest are the nuclear data related uncertainties of the group and time constants generated by Serpent.
In this work, a number of group and time constants that can be generated by Serpent are examined to identify how the sensitivity calculation capabilities could be used or extended to calculate the sensitivities that are needed to apply the Sandwich rule of deterministic first-order uncertainty propagation.
For some group and time constants the required sensitivities could be calculated either directly with the existing implementation or with small additional changes that were implemented into Serpent in this work. For other group and time constants that require additional work, the required implementations were identified and listed.
Serpent was also extended to read in multi-group covariance (nuclear data uncertainty) libraries and calculate the nuclear data uncertainties of any sensitivity responses internally during the simulation so that the nuclear data uncertainty for these responses can be output to be analyzed by the user. The output allows users to identify the total nuclear data uncertainty of a specific response as well as the contributions from the covariances (uncertainties) in the nuclear data of different reaction modes of different nuclides.
Several smaller features were also implemented to pave the way for automating the uncertainty propagation whenever group and time constants are generated by the code.
While some results of the new capabilities were compared to previous results obtained with CASMO-4E using an older covariance data library, in the future a sampling based uncertainty propagation methodology needs to be applied for the verification of the implemented routines.
In this work, a number of group and time constants that can be generated by Serpent are examined to identify how the sensitivity calculation capabilities could be used or extended to calculate the sensitivities that are needed to apply the Sandwich rule of deterministic first-order uncertainty propagation.
For some group and time constants the required sensitivities could be calculated either directly with the existing implementation or with small additional changes that were implemented into Serpent in this work. For other group and time constants that require additional work, the required implementations were identified and listed.
Serpent was also extended to read in multi-group covariance (nuclear data uncertainty) libraries and calculate the nuclear data uncertainties of any sensitivity responses internally during the simulation so that the nuclear data uncertainty for these responses can be output to be analyzed by the user. The output allows users to identify the total nuclear data uncertainty of a specific response as well as the contributions from the covariances (uncertainties) in the nuclear data of different reaction modes of different nuclides.
Several smaller features were also implemented to pave the way for automating the uncertainty propagation whenever group and time constants are generated by the code.
While some results of the new capabilities were compared to previous results obtained with CASMO-4E using an older covariance data library, in the future a sampling based uncertainty propagation methodology needs to be applied for the verification of the implemented routines.
| Original language | English |
|---|---|
| Publisher | VTT Technical Research Centre of Finland |
| Publication status | Published - 9 Oct 2018 |
| MoE publication type | D4 Published development or research report or study |
Publication series
| Series | VTT Research Report |
|---|---|
| Number | VTT-R-04681-18 |