A clustering genetic algorithm for actuator optimization in flow control

Active flow control can provide a leap in the performance of engineering configurations. Although a number of sensor and actuator configurations have been proposed the task of identifying optimal parameters for control devices is based on engineering intuition usually gathered from uncontrolled flow experiments. We propose a clustering genetic algorithm that adaptively identifies critical points in the controlled flow field and adjusts the actuator parameters through an evolutionary process. We demonstrate the capabilities of the algorithm for the fundamental prototypical problem of an actively controlled circular cylinder. The flow is controlled using surface-mounted vortex generators; the actuators used are belts mounted on the cylinder surface, that modify the tangential velocity on the cylinder surface, and jet actuators, that modify the normal velocity component on the surface. The proposed genetic algorithm performs the optimization of the actuators parameters, yielding up to 50% drag reduction. At the same time the genetic algorithm performs a sensitivity analysis of the optima it finds, thus allowing a deeper understanding of the underlying physics and also an estimation of which actuator would be easier to implement in a real experiment