Computational fluid dynamics of a large-scale steam generator using data assimilation empowered turbulence modeling

Data assimilation (DA) integrating limited experimental data and computational fluid dynamics is applied to improve the prediction accuracy of flow behavior in a large-scale steam generator (SG) system. The ensemble Kalman filter (EnKF) is used as the DA technique, and the Reynolds-averaged Navier–Stokes (RANS) modeling serves as the prediction framework. Two configurations—wet stator motor pump and canned motor pump—are tested at three different flow rates. The model constants, derived from the EnKF-based DA approach in our previous work, [Li et al., Ann. Nucl. Energy (unpublished) (2024)] are employed for verification. The DA model shows remarkable improvements and better predictions in jet penetration and flow separation than the default model. Results demonstrate that the optimized constants are transferable across different flow rates and configurations. For both reactor coolant pumps (RCPs), the DA-optimized model consistently reproduced the jet array, turbulent separation bubble size, and the inlet and outlet profiles of the RCPs in agreement with experimental data. These improvements arise from the collapse of the velocity distribution and pressure loss at varying flow rates, indicating a transition of flow in the outlet chamber and sudden expansion region into the second self-modeling zone. These improvements highlight the potential of EnKF-based DA for enhancing flow predictions in various SG system configurations, paving the way for more reliable applications in engineering design and operation.