Genetic Algorithm-based control of the wake of a bluff body

This work reports on the application of a Genetic Algorithm (GA)-based approach to control the wake of a bluff body. The control is achieved through the actuation of four air jets placed along the edges of the model's base. The dependence of the population size on the convergence of the genetic code was assessed, evidencing an increase of the number of elements in the population needed to learning more complex tasks. In this case, a sum of two sine waves is considered, where frequency and amplitude of each of the two sine waves are optimised. It is demonstrated that the GA converges to a control law yielding values of the drag reduction up to 11.2% with respect to the natural case. The cost function has been defined as to minimise the drag coefficient, without accounting for the energy spent in the actuation. The proper orthogonal decomposition (POD) applied to the fluctuating pressure signals highlights the most relevant features of the wake. The results show that in the natural case nearly 80% of the modal energy is associated with the first mode. Conversely, the forced case features a more evenly distributed energy content across the POD modes. The analysis of the first two modes reveals that for both cases the wake is governed by the shedding phenomenon. Furthermore, the analysis reveals that regardless of the actuation conditions, the top-down shedding represents the most significant phenomenon for the wake dynamics.