Numerical Prediction of Flow Induced Vibrations in Nuclear Reactors through Pressure Fluctuation Modelling

Abstract

Flow induced vibrations (FIV) is a known problem in nuclear reactors. The interaction between the turbulent flow of the coolants and the fuel rod bundles generates vibrations which are undesirable. These vibrations can be a major cause of fatigue failures, stress corrosion cracking and fretting wear of materials, which lead to stand-still costs. Analytical models for FIV are not feasible for complex geometries. This is especially true, when a strong interaction between fluid and structure forces is involved. The dynamic nature and a strong coupling between the two phase generates a need to study FIV by the use of numerical approaches. To take into account the interaction between the fluid and the structure one has to couple the solvers for both the phases. Coupling is achieved by mapping the forces and displacements between the interfaces of fluid and structure domains; also known as the Partitioned coupling. Recent works have made it possible to reduce the coupling errors and make them more efficient through the use of better algorithms. Due to the wide range of the scales involved, vibrations due to turbulent fluctuations of the flow are particularly challenging to solve. A common approach to this problem is to use high fidelity solvers for fluids such as DNS or LES which are computationally expensive for long and complex domains. Attempts have been made for use of cheaper URANS models to capture the modes of the resulting structural vibration by providing an initial perturbation to the structure. However, URANS models compute the time averaged velocity and pressures. The models do not calculate the effect of pressure fluctuations due to turbulent flow fluctuations on the structure. As a result, URANS does not provide information on the amplitudes of the vibrations of the structure due to turbulent fluctuations. In the current study, the primary objective is to reduce computational costs, and to still capture the complete dynamic behavior of the structure through the use of URANS models. This is performed by the modeling of pressure fluctuations from the turbulent kinetic energy calculated from the URANS models. The modelled pressure fluctuations (p0) are added to the mean pressure (p), which are then mapped over to the fluid structure interface to generate these vibrations. The method is applied to several test cases related to the field of nuclear energy production and results compared to DNS results and experimental results. The results obtained show that the method is capable of simulating FIV caused due to random pressure fluctuations in a turbulent flow. Form the test cases it was concluded that the method has an advantage of lower cost than LES/DNS, but still has the potential of modelling the amplitude of vibrations by turbulent pressure fluctuations.