Being able to quantify mechanical vibrations is of key importance for the safety of nuclear power plants, as they are able to induce damage. In this work, numerical simulations are used to compute water flow and vibration in a densely packed bundle of 7 rods, mimicking an experimental setup. This flow configuration is chosen to resemble the coolant flow through a nuclear reactor core. Because of the wall proximity, a considerable velocity difference between the narrow gaps and the subchannels exists, with an inflection point in the velocity profile. This yields an unstable situation, and large vortices are continuously created through a mechanism similar to the Kelvin-Helmholtz instability. The vortex streets in between the rods are associated with a fluctuating pressure field, causing vibrations of the rods. The experimental setup contains 7 steel cylinders, encased in a hexagonal duct. The central rod contains a section where the steel is replaced by a water-filled silicone tube, clamped at both extremes to the steel rod, and the vibrations of this section are examined. The numerical approach consists of coupled fluid-structure interaction (FSI) simulations, with the flow being modelled using computational fluid dynamics (CFD) and the structure using computational solid mechanics (CSM). The available experimental data consist of Laser Doppler Anemometry (LDA) measurements and high-speed camera footage of the wall movement of the silicone rod. Equivalent data is collected from the numerical simulations. The simulations are repeated for different flow rates. The frequency spectrum of the coherent structures, and the frequency and amplitude of the wall movement are compared for each operating point, as well as their trend as a function of the flow rate. The dominant frequencies found in the simulation results were similar to the experimental results, although slightly higher. They also showed a linear trend, just like the experiments. A larger mismatch was present for the structural response, the frequencies found using the FSI model being more than twice as high.