A PCA compression method for reactor core transient multiphysics simulation. (October 2020)

• Record Type: Journal Article
• Authors: Cherezov, Alexey; Jang, Jaerim; Lee, Deokjung
• Is Part Of: Progress in nuclear energy. Volume 128(2020)
• Publication Date: 2020-10
• Subjects: Principal component analysis -- Rod ejection accident -- Data compression -- Reactor core multiphysics -- OPR-1000 -- MPCORE; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 333.7924
• DOI: 10.1016/j.pnucene.2020.103441
• Languages: English
• ISSNs: 0149-1970

Abstract

Abstract: The paper deals with the adaptation and application of the data compression algorithms to reactor core multiphysics analysis. Large capacity makes the storage, transfer and processing of multiphysics data cumbersome and expensive. The general purpose lossy compression methods are not able to guarantee the accuracy of compressed information with respect to the uniform norm which is important for nuclear reactor safety analysis. Therefore we developed another algorithm that reduces the data capacity in up to two folds and at the same time preserves the peak fluctuations of multiphysics parameters. The new method decomposes multiphysics data into strong and weak correlation parts. The strong correlation part is approximated by the principal components, while the weak correlation part is saved in memory using a sparse matrix format. The compression ratio of this approach has been tested to the series of the rod ejected accidents in reactor core OPR-1000 at the different fuel burnups. Graphical abstract: Image 1 Highlights: 3D pin-by-pin multiphysics simulation produces a large amount of data. General purpose lossy compression methods do not guarantee the accuracy with respect to the uniform norm. Compression algorithm based on PCA and sparse matrix format preserves multiphysics parameter's fluctuations. Compression ratio of the new method has been verified to the rod ejection accident simulation test.