Impact of Simulated Tube Length in FSI Predictions of the GOKSTAD Experiment

Abstract The European Go-Viking project aims to develop a deep understanding of flow-induced vibration (FIV) phenomena in cross-flow steam generator tube bundles used in operating pressurized water reactors. This understanding will be used to prevent mechanical damage and premature failure of the steam generators. To support this objective, the GOKSTAD experiment, consisting of a 7x7 cross-flow inline tube bundle (pitch-to-diameter ratio of 1.44), will be used to collect high-resolution data on flow-induced vibrations for comparison to high and low fidelity modeling efforts. The high to low fidelity modeling includes RANS and medium- to high-resolution large eddy simulations (LES) computational fluid dynamics (CFD) modeling to gather both field and integral data comparable to experimental measurements. Given the computational cost of the highest fidelity simulations, smaller unit cell computational domains of the GOKSTAD experiment were created to reduce the simulation costs while maintaining accuracy. The unit cell consists of 7 rows with 3 columns of tubes and periodic conditions in the streamwise and spanwise directions. In this reported study, the sensitivity of the simulation results to the spanwise direction parallel to the tubes was investigated. The spanwise length of the domain was varied for lengths of one, two and three diameters in size. An average bulk mass flow rate of 15 m3/hr (corresponding to a gap Reynolds number of 6650) was used for each case. The reported data was velocity signals, average streamwise velocity and RMS velocity collected at different planes within the bundle and compared. The results from this study provide guidance for future high-resolution simulations domain size requirements in the spanwise direction.