Linear feedback control for form-drag reduction on bluff bodies with a blunt trailing edge

The work described in this thesis is a computational investigation applying linear feedback control to reduce form-drag on bluff bodies with a blunt trailing edge. For such bodies, a large portion of the aerodynamic drag is associated with an unsteady separated region or wake downstream of the body. The development of tractable feedback strategies to control unsteady wakes promises strong benefits, both in terms of industrial applications and for furthering our understanding of the flow mechanisms at play. For this purpose, large-eddy simulations are carried out where a linear feedback controller targets an increase in the mean pressure force on the rear (base) of the body. The flows over two distinct geometries are examined: a backward-facing step and a bluff body with a rounded leading edge, often referred to as a D-shaped body. The control is effected by zero-net-mass-flux slot jets, responding to sensors located on the body base. Open-loop characterization provides information on the effects of actuation and some physical insight into the relation between the base pressure and wake dynamics. System identification is used to obtain a low-order model of the flow's response to actuation that can be used for control. The control strategy is based on the premise that reducing the fluctuations in the near-wake will cause an increase in the mean base pressure, hence a reduction in form-drag. The controllers are designed with classical frequency-domain methods, using a sensitivity transfer function to attenuate the size of the pressure force fluctuations. The influence of parameters such as the Reynolds number and the location and type of actuators is studied. For all cases, low-order linear feedback controllers successfully reduce the pressure force fluctuations and achieve sensible drag reductions. They do so with higher efficiency than the open-loop forcing considered. Uncertainties in the model and flow conditions can be to some extent mitigated by the robustness of the controller. The results support the conjecture linking the fluctuating and mean base pressure, although it is observed that further work is needed before such an approach can be used for optimization