PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 3, 2026Nuclear Engineering and Technology0 citationsOpen Access

Development of quick neutron spectrum reconstruction module based on POD for burnup calculation of fast reactor

View Full Paper
NANaoto AizawaTWTomoaki WATANABEGCGo CHIBA

Key Points

  • Neutron spectrum reconstruction achieved with a few percent accuracy, reducing computational burden.
  • The module reproduces neutron spectra above several hundred eV with up to six POD basis vectors.
  • Application of reduced-order models predicts 1-group cross sections with accuracies of a few percent.
  • Integration with SWAT-X shows major actinoid compositions within approximately 6% of comparison results.

Abstract

A quick neutron spectrum reconstruction module based on proper orthogonal decomposition (POD) and a regression model for fast reactors is developed for the general-purpose burnup calculation code SWAT-X. This module enables burnup calculations to be performed quickly for arbitrary conditions based on a few parameters, eliminating the need for computationally intensive calculation codes. Reduced-order and regression models are constructed by applying the POD technique to the snapshot data of the multi-group neutron fluxes calculated using randomly-sampled parameter sets for a single hexagonal pin-cell of a sodium-cooled fast reactor. The neutron spectrum is reconstructed using a linear combination of POD basis vectors with the coefficients obtained from the regression model. The applicability of a pin-cell model, the accuracy of the reconstruction module itself, and its integration with SWAT-X are verified by comparing with the calculation results from the reactor physics code system CBZ. The comparison of nuclide inventories between pin-cell and whole core models shows an agreement within 5% for major actinoids and fission products with relatively large inventories. The developed module reproduces the neutron spectrum above several hundred eV with an accuracy of a few percent using up to six POD basis vectors. It also predicts the 1-group cross sections with an accuracy of a few percent for all parameter sets when the reduced-order model is used with eight POD basis vectors and third-degree polynomial regression. Furthermore, SWAT-X implementing the developed module reproduces major actinoid compositions within approximately 6% of the CBZ results.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Aizawa et al. (2026) studied this question.

synapsesocial.com/papers/69a7605dc6e9836116a2d0a9https://doi.org/10.1016/j.net.2026.104176
Ask AI
Helpful
Bookmark
Share
View Full Paper