Data-based hybrid reduced order modeling for vortex-induced nonlinear fluid–structure interaction at low Reynolds numbers

• We introduce a novel hybrid reduced-order model.
• We decompose fluid force into a frequency and amplitude/magnitude.
• Model is trained based on data that can come from numerical simulations or physical experiments.
• We model a non-linear structure interaction problem.

Fluid–structure interaction (FSI) phenomena are of significant importance in several engineering fields. Recently developed active flow control devices regulate the FSI in order to control the dynamic response of the structure that is involved. As a first step to use active control, computationally efficient reduced-order models are required. The reduced-order models must be able to predict the nonlinear structural dynamic response given an incoming flow condition. This paper presents a computationally efficient method for the construction of a hybrid reduced-order model for FSI problems based on data obtained through high-fidelity numerical simulations. The model splits the force and the structural dynamic response into two separate blocks and uses model reduction techniques to account for the flow field information. The current model is tested on a vibrating rigid cylinder submerged in a flow at low Reynolds number regime.