Reduced order models for fluid-structure interaction phenomena

Abstract

With the advent of active flow control devices for regulating the structural responses of systems involving fluid-structure interaction phenomena, there is a growing need of efficient models that can be used to control the system. The first step is then to be able to model the system in an efficient way based on reduced-order models. This is needed so that accurate predictions of the system evolution could be performed in a fast manner, ideally in real time.

The reduced-order model devised in this study proved to be an effective and efficient modeling method for fluid-structure interaction phenomena and it shown its applicability in very different kind of scenarios.

The model is tested on two cases: a cylinder suspended in a low Reynolds number flow that includes the lock-in region and an airfoil subjected to plunge oscillations in a high Reynolds number regime. For each case, in addition to training profile we also present validation profiles that are used to determine the performance of the reduced-order model.

This thesis proposes a novel reduced-order and efficient model for fluid-structure interaction phenomena. The model structure employed is such that it is generic for different fluid-structure interaction problems. Based on this structure, the model is first built for a given fluid-structure interaction problem based on a database generated through high-fidelity numerical simulations while it can subsequently be used to predict the structural response over a wide set of flow conditions for the fluid-structure interaction problem at hand.

However, existing reduced-order models of fluid-structure interaction phenomena that provide closed-form solutions are applicable to only a limited set of scenarios while for real applications high-fidelity experiments or numerical simulations are required, which are unsuitable as efficient or reduced-order models.