An assessment of CFD-scale fluid–structure interaction simulations through comprehensive experimental data in cross-flow.

Experimental tests accessing both fluid and structure behaviors are mandatory for a consistent and comprehensive assessment of fluid–structure interaction (FSI) numerical simulations. In this paper, recently published results of an experimental configuration of two in-line cantilever cylinders subjected to water cross-flow were considered: instantaneous fluid velocities measurements are available on several positions inside the test section, together with the cylinder vibrations. FSI simulations were performed by coupling Ansys Fluent (for the fluid domain) with Ansys Mechanical (for the solid domain). URANS simulations and Scale-Adaptive Simulations (SAS) were employed as CFD simulation approach. The structure displacements were taken into account through an Arbitrary Lagrangian–Eulerian approach. The fluid–structure coupling was 2-way explicit. Simulations were performed for two different water mass flow rates. For the highest one, vortex-induced resonance was observed experimentally. The numerical results show consistent agreement in terms of shedding frequency and velocity spectra behind the cylinders. The calculated vibration response is overall consistent, despite some underestimations, for the cylinders not featuring vortex-induced resonance; nevertheless, the experimentally observed vortex-induced resonance could not be reproduced by the numerical simulations. • Existing experimental results on FIV employed to assess CFD FSI simulations. • URANS and hybrid URANS/LES CFD approaches were tested: the latter proved superior. • Good agreement in terms of flow velocity spectra and mean profiles. • Coupled fluid-structure simulations were run to calculate the cylinder vibration. • Better agreement with exp. measurements for the second cylinder than for the first one. • A resonance was observed experimentally for the first cylinder at a given flow rate. • The resonance could not be predicted numerically. [ABSTRACT FROM AUTHOR]