Data assimilation of turbulent flow in a large-scale steam generator: Part I- Iterative ensemble-Kalman filter-based reconstruction

Abstract

This research focuses on reproducing the global turbulent mean flow within a large-scale steam generator (SG) system using an iterative Ensemble Kalman Filter (EnKF)-based data assimilation (DA). A compressed directional loss model is introduced to reduce computational costs while considering volume flow rate redistribution across the U-shaped arrays. Results demonstrate that the DA approach improves predictions, showing better agreement with experimental data by widening the jet core, enhancing jet array penetration, and reducing turbulent separation bubble size. The inlet velocity profile at the reactor coolant pump (RCP) entrance is also accurately represented. The extensibility of the optimized model is validated at the RCP outlet. The DA model more accurately captures fluid dynamics, including acceleration, deceleration, and vertical movement in the sudden expansion region, leading to better estimations of total pressure loss. These improvements open up possibilities of DA approach for real engineering applications in both design and operation.