Fluid-structure interaction simulation of a wire-wrapped tube array using overset grids

Summary

Axial flows in tube bundle geometries are omnipresent in nuclear power plants, including fuel assemblies and heat exchangers. The tubes are often long and slender which makes them susceptible to flow-induced vibrations. Current reactor research investigates the use of wires helicoidally wound around each rod, to preserve their mutual distance. This poses new challenges for numerical simulations. The wire-wrapped geometry is complex, among other things due to small gaps, making the meshing process complicated and cumbersome. This research explores the use of overset grids for such wire-wrapped tube bundles. This meshing method allows grids to overlap and the flow solution is interpolated between the different grids, providing more freedom for meshing. In this case the approach consists of making one grid for a single tube with a wire, termed the component mesh, and placing this grid in a so-called background mesh that takes into account other geometrical features (e.g. the bundle duct). In the case of bundles, the same tube mesh can be repeated several times without additional meshing effort, regardless of the bundle size. The approach is verified using 4 cases found in literature, using the same component mesh either one or multiple times for each case, thus reducing the meshing effort to a minimum thanks to the freedom and re-usability overset offers. The first two cases involve a fluid-structure interaction simulation of a single wire-wrapped tube in a cylindrical domain, the first one simulating a steady deformation and the second one a free vibration of the tube. Good agreement with results in literature was found. The latter two cases are Computational Fluid Dynamics simulations of a 7-pin and 19-pin bundle in a hexagonal duct, and excellent agreement with literature results was obtained. The overset approach was proven beneficial for simulating wire-wrapped bundle geometries: with largely reduced effort the same predictive capabilities can be obtained and potentially even extended.

Abstract

Full Paper