Application of Modal Decomposition Techniques for Reduced-Order Modelling of Once-Through Steam Generator

Abstract A fast simulation model that provides accurate estimation for thermal hydraulic parameters of the Once-through Steam Generator (OTSG) is of great significance for nuclear safety and economics, and can greatly reduce the computational cost of high-fidelity models for performing multi-query tasks. The Dynamic Mode Decomposition (DMD) and Proper Orthogonal Decomposition (POD) are commonly used non-invasive model reduction techniques and are data-driven which means the methods generate low-dimensional models as approximate representations of Full Order Model (FOM) rely only on data, thus have more advantages in reduced-order modeling. While the applications of these methods are rarely involving the OTSG. This paper aims to analyze the applicability of DMD and POD for OTSG reduced order modelling. The simulation data of main parameters calculated via RELAP5 are collected as snapshots. The influence of snapshots and modes truncation on the estimation performance of ROMs is analyzed. The results show that the POD model outperforms the DMD model when the same number of modes are used to build the ROM, although the energy ranking method is introduced to extract the main modes to improve the accuracy of the DMD model.