Abstract
Sub-channel analysis is a major approach for the thermal-hydraulic design and safety evaluation of nuclear reactors. To improve the analysis transient prediction, we employ data assimilation to integrate the various time-scattered measurements and information from the program into a coherent framework. In this work, a sub-channel analysis data assimilation scheme has been developed, which is based on the Ensemble Kalman Filter. The Versatile Reactor Sub-channel Analyze Program (VERSA) is used as the model operator and the observation values and simulation conditions are retrieved from the PSBT benchmark. The system is able to learn from experiment data and improve the estimations of the void fraction. The effects of different assimilation parameters, assimilation frequency, and covariance inflation on the data assimilation results are evaluated. In the best-case scenario, we improve RSME and MAE by 43.09% and 44.55%, respectively. At most, the assimilation efficiency reaches 67.61%.
